The desmosomal cadherin desmocollin (Dsc)1 is expressed in upper epidermis where strong adhesion is required. To investigate its role in vivo, we have genetically engineered mice with a targeted disruption in the Dsc1 gene. Soon after birth, null mice exhibit flaky skin and a striking punctate epidermal barrier defect. The epidermis is fragile, and acantholysis in the granular layer generates localized lesions, compromising skin barrier function. Neutrophils accumulate in the lesions and further degrade the tissue, causing sloughing (flaking) of lesional epidermis, but rapid wound healing prevents the formation of overt lesions. Null epidermis is hyperproliferative and overexpresses keratins 6 and 16, indicating abnormal differentiation. From 6 wk, null mice develop ulcerating lesions resembling chronic dermatitis. We speculate that ulceration occurs after acantholysis in the fragile epidermis because environmental insults are more stringent and wound healing is less rapid than in neonatal mice. This dermatitis is accompanied by localized hair loss associated with formation of utriculi and dermal cysts, denoting hair follicle degeneration. Possible resemblance of the lesions to human blistering diseases is discussed. These results show that Dsc1 is required for strong adhesion and barrier maintenance in epidermis and contributes to epidermal differentiation.
The desmosomal cadherins, desmocollin (Dsc)* and desmoglein (Dsg), mediate adhesion in desmosomes, one of the principal types of intercellular junction in epithelia and cardiac muscle (Chitaev and Troyanovsky, 1997; Marcozzi et al., 1998; Tselepis et al., 1998). Each constitutes a cadherin subfamily that comprises three genetic isoforms (Dsc1, 2, and 3 and Dsg1, 2, and 3) (Buxton et al., 1993). Dscs and Dsgs show tissue-specific expression: Dsc2 and Dsg2 are ubiquitous in desmosome-containing tissues, whereas Dscs 1 and 3 and Dsgs 1 and 3 are restricted to stratified epithelia. In stratified epithelia such as epidermis, the desmosomal cadherins also show differential expression with the “1” isoforms expressed in suprabasal differentiated layers and the “2” and “3” isoforms in more basal layers (Garrod et al., 1996). Dsg1 and 3 and Dsc1 and 3 expression in epidermis is graded reciprocally; the “3” isoforms decrease exponentially from the basal layer as the “1” isoforms increase (Shimizu et al., 1995; North et al., 1996). This implies that their expression patterns may be linked, a possibility also suggested by the adjacent chromosomal locations of desmosomal cadherin genes (Hunt et al., 1999). Where they occur together, Dsc1 and Dsc3 are mixed in the same desmosomes. Thus, the Dsc1:Dsc3 ratio increases with stratification, indicating turnover during epidermal differentiation and showing that adhesion at different levels in the epidermis involves different combinations of desmosomal cadherins (North et al., 1996).
Desmosomal adhesion plays an important role in the maintenance of tissue architecture as indicated by the study of human disease and null mutations in mice (Hashimoto et al., 1997; Amagai, 1999; Mahoney et al., 1999; Green and Gaudry, 2000). The loss of expression of desmosomal components in some human carcinomas and the ability of desmosomal expression to block cellular invasion suggests that desmosomes may also have a tumor suppressor function (Garrod, 1995; Shinohara et al., 1998; Tselepis et al., 1998).
A pressing question is whether the differential expression of desmosomal cadherins indicates specific functions for the distinct isoforms. Their graded expression in stratified epithelia implies a possible role in epithelial differentiation. Stratification itself is clearly not dependent on the presence of multiple desmosomal cadherin isoforms, since cornea possesses only Dsc2 and Dsg2 (Messent et al., 2000). In other epithelia, Dsc3 and Dsg3 are associated positionally with cell proliferation and the early stages of differentiation, whereas Dsc1 and Dsg1 are associated with terminal differentiation and keratinization. Thus, the “1” isoforms may provide strong adhesion to resist abrasion in keratinized epithelia, and the “3” isoforms may provide weaker adhesion required for cell proliferation and motility (Legan et al., 1994). Desmosomal cadherins also may be involved in differentiative signaling. Desmosomes participate in both inside-out and outside-in signal transduction (Osada et al., 1997; Wallis et al., 2000). Furthermore, the Dsc and Dsg cytoplasmic domains bind the armadillo proteins plakoglobin (PG) and plakophilin, and PG signaling in the epidermis has been demonstrated, since overexpression affects hair growth (Charpentier et al., 2000). The nuclear location of plakophilins suggests a possible gene regulation role (Hatzfeld, 1999). Thus, either a structural/morphogenetic role and/or a direct or indirect differentiative signaling role for desmosomal cadherins may be envisaged. Mice expressing dominant negative Dsg3 exhibit abnormalities of epidermal proliferation and differentiation (Allen et al., 1996), although Dsg3−/− mice show no obvious abnormalities of epidermal differentiation (Koch et al., 1997).
To approach the question of the function of Dsc isoforms, we have performed targeted ablation of the mouse Dsc1 gene. This results in a complex phenotype, showing epidermal fragility together with defects of epidermal barrier and differentiation. The epidermis of neonatal mice shows lesions, which resemble those found in IgA pemphigus, and older mice develop chronic dermatitis. These results demonstrate that Dsc1 contributes substantially to epidermal adhesion and function.
Characterization of the Dsc1 gene
Two λ clones (λC1 and λC4) and a PAC clone (364c17) containing Dsc1 sequence were obtained from library screens. From these, the mouse Dsc1 gene was characterized in detail. The gene spans ∼32 kb and consists of 17 exons (Fig. 1). Exon 10 is the largest (260 bp) and exon 16, which is alternatively spliced to generate the “a” and “b” forms of the protein (Collins et al., 1991), is the smallest (46 bp). Single exons do not correspond to known structural elements such as the extracellular domain internal repeats or the transmembrane or cytoplasmic domains (Fig. 1). All intron-exon borders conform to 5′ and 3′ splice site consensus sequences (Table I). The organization of mouse Dsc1 shows a striking resemblance to that of other Dsc and classical cadherin genes (Greenwood et al., 1997; Whittock et al., 2000) with exon boundaries highly conserved (Fig. 1).
Generation of Dsc1-null mice by homologous recombination
The Dsc1 gene was disrupted in embryonic stem (ES) cells using the targeting vector shown in Fig. 2 A. On homologous recombination, the vector replaces the translation initiation codon (ATG) and the coding sequence from exon 1 with a neo-cassette. 98 bp at the 3′ end of exon 1 (including 63 bp encoding the start of the Dsc1 signal peptide) and 2.2 kb of intron 1 are replaced by vector DNA.
Four ES cell clones containing a disrupted Dsc1 allele were identified, and one was used to generate chimeric mice that transmitted the mutated allele to their progeny. Mice heterozygous for the targeted mutation were identified by Southern blotting; wild-type and mutant alleles produced fragments of 20- and 6-kb, respectively (Fig. 2 B). Additional Southern blotting with a probe located 3′ to the homologous sequences in the targeting vector confirmed the presence of the mutated allele in heterozygous mice, and a neo-derived probe was used to show the presence of only one copy of the neo gene (unpublished data). Heterozygous mice, which could not be distinguished from their wild-type littermates, were crossed and their progeny genotyped. Litter sizes were normal. Of the pups, 26% were wild-type (+/+), 56% were heterozygous (+/−), and 18% were homozygous for the disrupted allele (−/−) (n = 308). Thus, the number of homozygous-null offspring was slightly lower than expected. Dsc1 is first expressed at embryonic day (E)13.5 in the outer layers of epidermis (King et al., 1996; Chidgey et al., 1997). To determine whether absence of Dsc1 results in some embryonic mortality, we genotyped E17 embryos from heterozygous crosses. Of these, 28% were wild-type, 43% were heterozygous, and 29% were −/− (n = 65). This is a normal Mendelian ratio. These data suggest that some of the Dsc1−/− neonates are eaten by the mothers at birth, although no direct evidence for this was found. Dsc1−/− mice were fertile: pups from Dsc1−/− × Dsc1−/− crosses were reared and weaned successfully.
Mice lacking Dsc1 show a neonatal phenotype involving skin, hair, eyes, and growth
Using primers P1 and P2 specific for exons 1 and 5, respectively (Fig. 2 A), full-length Dsc1 mRNA was absent from the epidermis of Dsc1−/− mice by reverse transcriptase (RT)-PCR (Fig. 3 A). To confirm the absence of a 5′-truncated message, RT-PCR was performed using primers P3 and P4 specific for exons 4 and 7, respectively (Fig. 2 A). No evidence for a truncated message was found (Fig. 3 A). Primer P3 hybridizes to DNA downstream of that encoding the beginning of the mature protein. Hence, it is extremely unlikely that a truncated message smaller than that which would be detected by primer pair P3-P4 would produce a functional protein. Western blotting with a polyclonal antibody raised against a substantial portion of the extracellular domain of Dsc1 (North et al., 1996) showed absence of the protein from Dsc1−/− epidermis (Fig. 3 B). No evidence for the production of a truncated protein was found. Western blotting also showed no clear differences in expression of other desmosomal components (Dsc2, Dsc3, Dsg1+2, PG, desmoplakin [DP]) (Fig. 3 B). Immunofluorescence demonstrated absence of Dsc1 from Dsc1−/− epidermis and the inner root sheaths of hair follicles (Fig. 3, C and D). Other desmosomal components were unchanged (DP shown) (Fig. 3, E and F).
At birth, homozygous mice were generally indistinguishable from their littermates, but 10% were born with one or both eyes open (Fig. 4, A and B). Failure of eyelid fusion led to corneal damage and inflammation and corneal opacity or micropthalmia later in life. From day 2, null mice could be distinguished from their normal (i.e., Dsc1+/+ and Dsc1+/−) littermates because null skin was covered in white flakes or scales (Fig. 4, C and D).
Although born at the same size, null mice usually grew less rapidly (Fig. 4 D), and by weaning (21 d) they were ∼30% smaller than littermates (Fig. 4 E; see Fig. 9 A). Slower growth was not due to an obvious feeding defect. The guts of null mice had milk in their stomachs and a normal appearance throughout their length, indicating that the food was processed normally. Dsc1 is expressed in mouse tongue, hard palate, esophagus, and forestomach (King et al., 1996, 1997), but histological examination of these tissues from null mice revealed no abnormalities of the epithelia. Inspection of the oral cavities of null pups revealed no blistering of epithelia. Thus, Dsc1−/− mice did not show the type of oral blistering defects responsible for feeding defects and small size in Dsg3−/− mice (Koch et al., 1997). Of null animals, 10% remained runted, but others caught up after weaning so that by 3 mo there was no significant difference in weight between them and normal mice. Thus, although growth was slower adult size was not generally affected.
Dsc1 is expressed in the Hassal bodies of human thymus (Nuber et al., 1996). To test for an effect on thymus, we examined it histologically and counted numbers of peripheral blood T cells labeled with antibodies to CD4 and CD8 by flow cytometry. No differences were found.
Mice lacking Dsc1 show abnormal epidermal proliferation and differentiation
Histology showed that the epidermis of Dsc1−/− mice was considerably thicker than that of littermates. Normal epidermis consisted of two to three cell layers below the cornified layer, but that of null mice had one, two, or three additional layers, and thickening was found at all body sites. At 2 d, thickening was principally in the spinous layer (Fig. 3, E and F, and Fig. 5, A and B), although localized hyperkeratosis and parakeratosis were also seen (Fig. 5 C). The epidermal cells often but not always appeared larger in the thickened null epidermis (Fig. 5, A and B, compared with Fig. 3, E and F). To determine whether epidermal thickening was due to increased cell proliferation, we examined the skin from null and normal mice using an antibody against Ki67. Proliferating cells were infrequent (3% of all epidermal nuclei) in normal mouse skin and found exclusively in the basal cell layer (Fig. 5 D). By contrast, in null mouse epidermis keratinocytes with nuclear Ki67 were extremely abundant (33% of all epidermal nuclei) and observed frequently in suprabasal layers (Fig. 5 E).
To investigate the effects of the null mutation on epidermal differentiation, we examined the expression of differentiation markers by immunofluorescence. No differences between normal and null epidermis from 2-d-old animals were observed in the distributions of the basal markers P-cadherin (Fig. 6, A and B), β4-integrin (Fig. 6, D and E), and keratin 5 (unpublished data). Early markers of terminal differentiation (keratin [K]1, involucrin) and a marker for late epidermal differentiation (filaggrin) were normal in 2-d-old null mice (unpublished data). However, a striking difference in the hyperproliferation markers keratins 6 and 16 was found. In normal 2-d epidermis, expression of K6 was restricted mostly to hair follicles (Fig. 6 G), although localized interfollicular epidermal staining was found occasionally. In null epidermis, K6 was present in both hair follicles and interfollicular epidermis (Fig. 6 H). Such K6 expression was observed at all body sites but was patchy locally: in some places expression was present throughout interfollicular epidermis but was absent from others. In normal epidermis, K16 was expressed in hair follicles, and superficial staining was present in suprabasal interfollicular epidermis (Fig. 6 J). In null animals, expression of the protein was upregulated and detected in all cell layers of interfollicular epidermis (Fig. 6 K).
In later life, null mice were susceptible to skin ulceration (see below). In lesioned epidermis from adult null mice, basal layer markers were expressed in all cell layers (Fig. 6, C and F). There were no differences in the distributions of K1 and filaggrin. Both K6 (Fig. 6 I) and K16 (Fig. 6 L) were found throughout lesioned epidermis.
Cornified envelope formation is an important aspect of epidermal terminal differentiation. Because a Dsc, Dsc3, has been reported as a component of cornified envelopes (Robinson et al., 1997) and because Dsc1−/− mice showed abnormal epidermal differentiation, we compared cornified envelopes from the epidermis of null and wild-type mice. No difference in appearance was detectable.
Dsc1 deficiency produces epidermal fragility with localized loss of barrier activity
Two observations from neonatal mice suggested that the epidermis of Dsc1−/− mice is more fragile than that of littermates. First, histology revealed splits in the epidermis. Such acantholysis, which was absent from normal epidermis, occurred within the granular layer between the granular and spinous layers (Fig. 7 A) or between the cornified and granular layers. Second, when we attempted to isolate the epidermis from 2-d-old mice we found that null epidermis could only be removed in small pieces, whereas normal epidermis remained in sheets (Fig. 7, B and C).
We determined the effects of eliminating Dsc1 on desmosome ultrastructure in the epidermis. No abnormalities in the structure or number of desmosomes were found, including those at the level of the upper spinous and granular layers where Dsc1 expression is strongest (Fig. 7, D and E). Similarly no changes in desmosomes in the inner root sheaths of hair follicles were apparent (unpublished data). By EM, epidermal splits showed an absence of cell lysis and desmosomes, indicating that splitting occurred because of weakened adhesion (Fig. 7 F).
We speculated that flaky skin and epidermal fragility could lead to defects in the skin barrier. To examine this, barrier activity was tested by dye penetration assays (Hardman et al., 1998; Marshall et al., 2000). E17 embryos showed no staining so the barrier was complete before birth. Also, 2-d-old normal mice stayed uniformly white, indicating complete barrier formation (unpublished data; Hardman et al., 1998). However, 2-d-old null mice showed numerous dark spots, indicating localized loss of barrier function (Fig. 8 A). These lesions were not left-right symmetric, indicating that environmental factors contributed to their formation.
Histology of stained spots showed that formation and repair of the lesions could be viewed as sequential events. Early lesions appear in the subcorneal region of epidermis as localized accumulations of neutrophils (Fig. 8, B and C). Such lesions bear a striking resemblance to those found in the subcorneal pustular dermatosis form of IgA pemphigus (Fig. 8 B compared with Fig. 1 c in Hashimoto et al., 1997). Next, the epidermis is degraded by inflammatory cells and partially detaches from the underlying matrix (Fig. 8, D and E), allowing penetration of dye (Fig. 8 A). Concurrently, epithelial cell migration into the wound site is initiated (Fig. 8 E). Finally, when wound healing is complete the damaged epidermis detaches from the underlying tissue, leaving behind an intact newly formed barrier (Fig. 8 F).
To quantify loss of barrier function, body skin from 2-d-old animals was used for determination of permeability constants (Kp) to water and mannitol, a marker for paracellular permeability (Fig. 8 G). Values for Kp were significantly higher (p < 0.05) for null animals. The null skin Kp for mannitol was twice that for normal skin, and permeability to water was increased 1.6-fold (Fig. 8 G). These results indicate that absence of Dsc1 causes a defect in the skin barrier.
To examine transepidermal water loss (TEWL), measurements were performed on 25 wild-type and 29 null pups. These gave values of 5.2 mg/h and 6.3 mg/h, respectively. Thus, TEWL for null mice was ∼1.2-fold greater than for wild-type controls (p = 0.02).
Dsc1-null mice develop alopecia and chronic dermatitis
The coats of adult null mice had a slightly untidy ruffled appearance (Fig. 9 A). Their pelage hairs were indistinguishable from those of normal littermates with similar distributions of awls, auchenes, zig-zags, and guard hairs, and whiskers appeared normal. In later life (usually manifesting at between 1 and 2 mo), >90% of null mice suffered hair loss and skin ulceration (Fig. 9, B and C). Hair loss was most common ventrally (Fig. 9 B), but some individuals had bald patches on the back, and two mice showed more general hair loss. Ulceration was most common around the muzzle (Fig. 9 C) but was also found elsewhere on the head, ventrally, or on the back. Muzzle lesions appeared between 6 wk and 6 mo, and affected mice were killed. Mice that were caged alone developed the same phenotype.
To investigate the mechanism of hair loss, we examined hair follicles at various stages of the hair cycle. No striking abnormalities were found at 5 (first anagen), 18 (catagen), 21 (telogen), and 28 d (second anagen). Tape stripping of the coats detached no more hairs from null than from normal mice, indicating that hairs were firmly anchored in null follicles.
In older animals, severe epidermal and follicular hyperplasia was observed in ulcerated areas of skin (Fig. 9 C) and occasionally in uninvolved skin. In hyperplastic epidermis, substantial thickening of the stratum corneum (hyperkeratosis) was observed (Fig. 9, D and E). Normally differentiated hair follicles were absent and replaced by large comedo-like malformations (utriculi) and cysts (Fig. 9, D and F). Such structures characteristically result from disintegration of hair follicles, suggesting that the observed alopecia is a result of hyperplasia and loss of normal hair follicle morphology. In some ulcerated regions, the epidermis was lost completely (Fig. 9 E). Severe inflammation and tissue necrosis were also observed. The most striking examples of focal overgrowth of keratinocytes were found in both apparently normal and ulcerated whisker pads, which showed a marked papillomatous nature (Fig. 9, G compared with H) with hyperkeratosis and parakeratosis. No evidence of malignant conversion was found. Dsc1-null mice were diagnosed as suffering from chronic dermatitis.
Deletion of Dsc1 has two consequences for epidermis. First, weakened adhesion gives rise to epidermal fragility and acantholysis. In early life, this produces a striking localized loss of barrier function and skin flaking. In older mice, substantial ulceration occurs. Second, epidermal differentiation is abnormal, manifested as epidermal hyperproliferation and thickening, an upregulation of keratin 6 and 16 expression, and localized hyperkeratosis and parakeratosis.
Absence of Dsc1 did not result in any alteration in the structure of desmosomes in the upper epidermis, but acantholysis indicates that they are more weakly adhesive than in normal mice, presumably relying on other Dscs for their structure and adhesion. However, there was also no apparent alteration in the expression of other desmosomal components, a result similar to that found in Dsg3−/− mice (Koch et al., 1997). Thus, although the expression of the glycoprotein genes has been postulated to be linked (North et al., 1996; King et al., 1997) two examples of gene deletion have not resulted in compensatory upregulation of other family members.
The acantholysis in the upper epidermis in Dsc1−/− resembles that seen in the human autoimmune blistering disease pemphigus foliceus (PF) (Mahoney et al., 1999), reflecting the similar epidermal expression patterns of Dsc1 and the PF antigen Dsg1. In PF, acantholysis results from binding of IgG autoantibody to the extracellular domain of Dsg1. There is no evidence for the involvement of antibodies to Dsc1 in PF. Subsequently, the localized acantholytic lesions in Dsc1−/− mice become infiltrated with polymorphonuclear cells (neutrophils). The lesions are reminiscent of those seen in the human subcorneal form of IgA pemphigus in which IgA autoantibodies to Dsc1 bind to the peripheries of epidermal keratinocytes (Hashimoto et al., 1997). No evidence for the pathogenicity of these antibodies has yet emerged. Whether the resemblance between the lesions is coincidental or has significance for the etiology of IgA pemphigus awaits clarification.
Although the acantholytic lesions appear similar, those in young Dsc1−/− mice do not progress. In the human diseases, red encrusted lesions are characteristic of PF and red pustular blisters are characteristic of IgA pemphigus, yet the epidermis of the mice remain unblemished apart from the appearance of white flakes. We suggest that skin lesions in young mice are triggered by local acantholysis but do not progress because wound healing is rapid. Reepithelialization of wounds takes ∼50% longer in 6–8-wk-old mice than in neonatal mice (Whitby and Ferguson, 1991). This difference in the rate of wound healing can account for the difference in severity between the lesions of neonatal mice and older mice. The environmental insults experienced by older animals are more severe than in young mice during suckling. Repeated insults, especially scratching, give rise to lesions in the epidermis that cannot be healed as rapidly as in neonatal animals. Exacerbation of lesions leads to ulceration and chronic dermatitis, which in this case is a proliferative response of the epidermis to damage. Similar lesions have been found in Dsg3 −/− mice (Amagai et al., 2000).
Epidermal barrier function is compromised in neonatal Dsc1−/− mice. The defect appears to arise as a consequence of localized acantholysis and thus does not imply a direct role for Dsc1 in barrier formation, although we cannot completely exclude the possibility of a slight, more generalized defect. The effect is clearly mild and not life threatening. However, the slight reduction in epidermal barrier function and increase in water loss may cause stress to null mice and result in growth retardation. Area measurements of water and mannitol permeability and TEWL indicate a decrease in barrier function of less than twofold. A fourfold increase in TEWL is not lethal in mice transgenic for COOH-terminally truncated loricrin (Suga et al., 2000). By contrast, a targeted mutation in the keratin 10 gene results in an eightfold increase in TEWL and neonatal death (Jensen et al., 2000). Mice deficient in the cornified envelope proteins involucrin and loricrin exhibit no neonatal barrier defect, although the latter show delayed barrier formation before birth (Dijan et al., 2000; Koch et al., 2000). Therefore, these mice are affected less severely than Dsc1−/− mice.
Clear evidence for hyperproliferation and epidermal thickening was found in Dsc1−/− mice. Hyperproliferation may be a response to damage to the skin barrier. Underlying cells can sense defects in cells above and respond by proliferating in several experimental situations. Disruption of the barrier elicits DNA synthesis and epidermal hyperplasia (Proksch et al., 1991; Denda et al., 1998), and expression of the whn gene and NH2-terminally truncated Dsg3 in suprabasal layers of transgenic mouse epidermis leads to both hyperproliferation and an immune response (Allen et al., 1996; Prowse et al., 1999). However, the major barrier defect in Dsc1−/− mice is localized, whereas the hyperproliferation and epidermal thickening is not. Thus, it seems likely that alterations in the program of keratinocyte terminal differentiation occur in the absence of Dsc1. Two pieces of evidence support this view. First, proliferating keratinocytes were found in suprabasal layers of null epidermis, and second, parakeratosis was found, suggesting that the later stages of differentiation are delayed. Furthermore, there is evidence to suggest that expression of K6 and K16 in interfollicular epidermis is associated with alterations in differentiation rather than hyperproliferation (Schermer et al., 1989; Sellhayer et al., 1993; Porter et al., 1998b). However, the whole question of whether the effects we observe are a primary consequence of loss of Dsc1 or a secondary consequence of acantholysis and wound healing is a complex one that will form the subject of further investigation.
Epidermal hyperproliferation may suggest a resemblance to the common human skin disease psoriasis. However, the changes in differentiation and skin inflammation are less severe than in psoriasis, and other features, such as loss of the granular layer, formation of deep rete pegs, and dilation of dermal blood vessels, are absent.
Hair loss in Dsc1−/− mice differs from that in Dsg3-null animals. In the latter, hair loss occurs in wave-like patterns as the result of defective cell adhesion in the telogen phase of the hair cycle (Koch et al., 1998). Dsc1 is expressed in inner cell layers of the hair follicle (Chidgey et al., 1997; King et al., 1997), but despite extensive examination we found no abnormalities in timing of the hair cycle or in structure of hair follicles in Dsc1−/− mice of up to 4 wk of age. Hair loss occurred generally in older animals and was often associated with loss of normal hair follicle morphology and the presence of comedo-like structures (utriculi) and dermal cysts. Such structures characteristically result from hair follicle degeneration (Montagna et al., 1952; Mann, 1971; Panteleyev et al., 1998). Further work will be required to determine whether hair follicle degeneration is an indirect consequence of the hyperproliferative and inflammatory environment or whether Dsc1 has a role in the maintenance of normal follicle morphology.
In conclusion, the phenotype of Dsc1−/− mice demonstrates a role for this desmosomal glycoprotein in keratinocyte adhesion in the upper epidermis and in the normal function of the skin. Furthermore, our data suggest that Dsc1 contributes to the regulation of epidermal differentiation, and Dsc1−/− mice may prove useful in determining the role of desmosomes in skin disease.
Materials And Methods
Isolation and characterization of the mouse Dsc1 gene
Mouse 129SVJ λFixII (Stratagene) and PAC (RPCI21; UK HGMP Resource Centre) genomic libraries were screened with overlapping fragments of the mouse Dsc1 cDNA. Two unique nonoverlapping λ clones (λC1 and λC4) and one PAC clone (364c17) were characterized in detail. Intron sizes were determined by generating PCR products using synthetic oligonucleotide primers based on flanking exon sequences derived from the Dsc1 cDNA sequence (King et al., 1996). Exon-intron boundaries were determined by DNA sequencing and comparison with cDNA sequence.
Construction of a targeting vector and generation of Dsc1-null mice
To construct a targeting vector, a PvuII fragment from λC1, comprising 2.1 kb of DNA 5′ of the Dsc1 initiation codon, was cloned upstream of a neomycin resistance gene under the control of a phosphoglycerate kinase promoter (Fig. 2 A). A 5.2-kb NotI (from the λFixII vector) to HindIII fragment from λC4, containing exons 2–5 of Dsc1, was cloned downstream of the neo gene. Vector DNA was linearized and electroporated into ES cells. Neomycin-resistant clones were isolated and screened by Southern blot analysis using BamHI-digested genomic DNA and a probe (1 kb), which hybridized to DNA 5′ to the homologous sequences used in the targeting vector (Fig. 2 A). One recombinant ES cell clone was injected into C57Bl/6J blastocysts, and the resulting chimeric male offspring were mated with female C57Bl/6J mice. Tail DNA from agouti offspring was tested for the presence of the mutated allele by Southern blot hybridization. Heterozygous mice were intercrossed to obtain mice, which were homozygous for the Dsc1-null allele.
Newborn (2-d-old) mice were killed, and their skin was removed and immersed in 5 mM EDTA in PBS for 3 min at 50°C. The epidermis was peeled from the dermis, homogenized in Trizol reagent, and total RNA was prepared according to the manufacturer's instructions (Life Technologies). First strand cDNA synthesis was performed from RNA using a kit (Roche) and random primers. An aliquot (1 μl) of the first strand cDNA reaction was amplified using primers P1 (AGGGAGCACCTTCTCTAAGCAG) and P2 (AAGGCTCTTTGTCTACTCCTGG), specific for exons 1 and 5, respectively, and P3 (GGTCCATTTCCACAACACATT) and P4 (TGTACTTCAGACGAGTGTGTA), specific for exons 4 and 7, respectively. Glyceraldehyde-3-phosphate dehydrogenase primers (CLONTECH Laboratories, Inc.) were added to PCR reactions as a positive control.
Epidermis was isolated from 2-d-old mice as above and dissolved in sample buffer. Proteins were separated by SDS-polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membrane (Amersham Pharmacia Biotech). Primary antibodies were JCMC (1:10,000) against Dsc1 (North et al., 1996), JB1 (1:500) against Dsc2 (Hyam, J.L.M., personal communication), rabbit anti–mouse Dsc3 (1:100; unpublished data; Liu, K., and D. Marshall, personal communication), DG3.10 (1:100) against Dsg1+2 (Progen), mouse anti–human PG (1:5,000; Transduction Laboratories), and 11–5F (1:40) against DP (Parrish et al., 1987). Primary antibodies were detected using appropriate peroxidase-conjugated secondary antibodies and the ECL detection system (Amersham Pharmacia Biotech).
Histological and immunochemical analysis
For histological analysis, tissues were fixed routinely in formol saline, embedded in paraffin, and stained with hematoxylin and eosin. Cell proliferation was examined in paraffin sections by staining for Ki67 (Gerdes et al., 1984). Expression of desmosomal constituents and differentiation markers were examined in frozen sections. Tissue was embedded in cryoprotectant (OCT; Miles), frozen in liquid nitrogen, and cryosections fixed in ice-cold methanol (5 min). Primary antibodies were JCMC (1:500), PAT-C (1:10) against Dsc2 (Messent et al., 2000), DG3.10 (1:50), PG-11E4 (1:100) against PG (Zymed Laboratories), 11–5F (1:25), MK1 (1:2,500) against keratin 1 (Covance), MK6 (1:2,500) against keratin 6 (Covance), MK14 (1:20,000) against keratin 14 (Covance), RPmK16 (1:500) against keratin 16 (Porter et al., 1998a), rat anti–mouse β4 integrin (10 μg/ml; Pharmingen), PCD-1 (10 μg/ml) against P-cadherin (R & D Systems), rabbit antiinvolucrin (1:100; Covance), and rabbit antifilaggrin (1:1,000; Covance). Primary antibodies were detected with the appropriate FITC-conjugated secondary antibodies. Sections were examined using an Olympus BX49 microscope.
Fresh skin was dissected into small pieces (∼1 × 1 × 0.5 mm) and fixed by immersion in 2% formaldehyde (freshly made from paraformaldehyde powder) plus 2% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.3). Tissues were fixed for 2 h at room temperature and then washed four times with cacodylate buffer. They were then post-fixed with 1% osmium tetroxide for 2 h, dehydrated through an ethanol series, and embedded in Spurr's resin. Ultrathin sections were contrasted with uranyl acetate and lead citrate and examined on a Philips 400 electron microscope.
Analysis of skin barrier function
Dye penetration assays.
Passive diffusion of solutes.
Samples of mouse back skin were clamped in continuous flow-through perfusion chambers. 3H2O (5 μCi/ml) and [14C]mannitol (0.1 μCi/ml) were added to the donor compartment, and PBS (4 ml/h) was passed across the connective tissue side and collected hourly for up to 18 h. Samples were counted in a Beckman LS 6000 liquid scintillation counter, and a steady-state permeability constant (Kp) was determined in units of cm/min (Healy et al., 2000; Selvaratnam et al., 2001). Experiments were performed at 32°C.
This was determined using a Tewameter (Courage and Khazaka). After decapitation or halothane anesthesia, a probe (1 cm diameter) was applied to the back of 2-d-old animals, and measurements of 1-min duration were taken.
Cornified envelope preparation
Epidermis from 2-d-old mice (see above) was heated to 95°C for 10 min in extraction buffer (0.1 M Tris-HCl, pH 8.5, 2% SDS, 20 mM dithiothreitol, and 5 mM EDTA), and the cornified envelopes were collected by centrifugation at 5,000 g for 15 min. The extraction was repeated once, and the envelopes were resuspended and stored in extraction buffer.
We are grateful to Dr. S. Andrew, Dr. E. Bell, Professor C.M. Griffiths, Dr. M. Hardman, Professor T. Hashimoto, Mr. D. Marshall, Mr. G. Morrissey, Mrs. J. Morrissey, and Miss Anjana Patel for help and advice.
The work was supported by the Wellcome Trust, Medical Research Council, and Royal Society.
J. Hewitt's present address is Queen's Medical Centre, Genetics Division, Nottingham University, Nottingham NG7 2UH, UK.
A. North's present address is The Rockefeller University, 1230 York Ave., New York, NY 10021.
Abbreviations used in this paper: DP, desmoplakin; Dsc, desmocollin; Dsg, desmoglein; ES, embryonic stem; K, keratin; PF, pemphigus foliaceus; PG, plakoglobin; RT, reverse transcriptase; TEWL, transepidermal water loss.