Journal of Science 111, 2529-2537 (1998) 2529 Printed in Great Britain © The Company of Biologists Limited 1998 JCS3800

Desmoglein 3 anchors telogen hair in the follicle

Peter J. Koch, M. G. Mahoney, George Cotsarelis, Kyle Rothenberger, Robert M. Lavker and John R. Stanley* Department of Dermatology, University of Pennsylvania School of Medicine, 211 CRB, 415 Curie Blvd, Philadelphia, PA 19104, USA *Author for correspondence (e-mail: [email protected])

Accepted 6 July; published on WWW 13 August 1998

SUMMARY

Little is known about the function of in the (acantholysis) between the cells surrounding the telogen normal structure and function of hair. Therefore, it was club and the basal layer of the outer root sheath . surprising that mice without 3 (the autoantigen Electron microscopy revealed ‘half-desmosomes’ at the in vulgaris) not only developed mucous plasma membranes of acantholytic cells. Similar membrane and skin lesions like pemphigus patients, but acantholytic histology and ultrastructural findings have also developed hair loss. Analysis of this phenotype been previously reported in skin and mucous membrane indicated that hair was normal through the first growth lesions of DSG3−/− mice and patients. phase (‘follicular neogenesis’). Around day 20, however, Immunoperoxidase staining with an antibody raised when the hair follicles entered the resting phase of the hair against mouse desmoglein 3 showed intense staining on the growth cycle (telogen), mice with a targeted disruption of cell surface of keratinocytes surrounding the telogen hair the desmoglein 3 gene (DSG3−/−) lost hair in a wave-like club in normal mice. Similar staining was seen in human pattern from the head to the tail. Hair then regrew and was telogen hair with an anti-human desmoglein 3 antibody. lost again in the same pattern with the next synchronous Finally, a scalp biopsy from a pemphigus vulgaris patient hair cycle. In adults, hair was lost in patches. Gentle hair showed empty telogen hair follicles. These data pulls with adhesive tape showed that anagen (growing) demonstrate that desmoglein 3 is not only critical for cell hairs were firmly anchored in DSG3−/− mice, but telogen adhesion in the deep stratified squamous epithelium, but hairs came out in clumps compared to that of DSG3+/− and also for anchoring the telogen hair to the outer root sheath +/+ littermates in which telogen hairs were firmly of the follicle and underscore the importance of anchored. Histology of bald skin areas in DSG3−/− mice desmosomes in maintaining the normal structure and showed cystic telogen hair follicles without hair shafts. function of hair. Histology of hair follicles in early telogen, just before clinical hair loss occurred, showed loss of Key words: Desmoglein, Pemphigus, , Hair

INTRODUCTION There are two major types of pemphigus, pemphigus vulgaris (PV) and pemphigus foliaceus (PF) (reviewed by Desmoglein 3 (Dsg3) is a desmosomal transmembrane Stanley, 1993). In both patients develop glycoprotein that belongs to the superfamily of cell against ; in PV against Dsg3 and in adhesion receptors (Amagai et al., 1991; Koch and Franke, PF against desmoglein 1 (Dsg1). It has been shown 1994). It is expressed and assembled into desmosomes in convincingly that these desmoglein-specific antibodies are certain strata of stratified squamous epithelia (Franke et al., pathogenic, that is they bind to the cell surface of desmoglein- 1994; Karpati et al., 1993; Schwarz et al., 1990; Arnemann et expressing keratinocytes in vivo and induce loss of cell al., 1993; Schafer et al., 1994; Schmidt et al., 1994). In skin, adhesion leading to the formation of blisters (Amagai et al., Dsg3 is expressed in keratinocytes of the basal and immediate 1992, 1994, 1995). PF blisters occur in the superficial suprabasal cell layers (e.g. Amagai et al., 1996). epidermis where Dsg1 is expressed, wheras PV blisters occur Even though desmosomes are multiprotein complexes thought in the deep epidermis and stratified squamous epithelia of to be important in maintaining cell-to-cell adhesion, there have mucous membranes where Dsg3 is expressed. This histologic been few in vivo models that confirm their actual function. One finding in PV is called suprabasilar acantholysis. exception has been the study of the pathophysiology of the Evidence that antibodies in PV interfere directly with an -mediated pemphigus diseases, which has provided adhesive function of Dsg3 in desmosomes was obtained from strong evidence for the importance of desmogleins in providing analyzing mice that we recently generated with a targeted cell adhesion between keratinocytes. disruption of the DSG3 gene (DSG3−/−) (Koch et al., 1997). 2530 P. J. Koch and others

The phenotype of these mice strikingly resembled that of PV with 10 column volumes each of buffer 1 (10 mM Tris-HCl, pH 7.5), patients. These DSG3−/− mice developed oral and vaginal buffer 2 (10 mM Tris-HCl, pH 7.5/0.5 M NaCl) and buffer 3 (100 mM mucous membrane lesions with the typical histology of PV. At glycine/10 mM Tris-HCl/0.5 M NaCl, pH 2.5). After equilibration of sites of trauma, gross and histologic PV-like lesions of the skin the column in buffer 1, the heat inactivated serum (20 minutes, 56°C), were seen as well. These findings showed that a loss of Dsg3 diluted 1:4 in buffer 1, was loaded onto the affinity matrix. The loading step was repeated three more times. The affinity matrix was function is sufficient to cause a PV-like phenotype, thus then washed with 20 volumes each of buffers 1 and 2. Antibodies were supporting the idea that PV autoantibodies directly interfere eluted with 10 column volumes of buffer 3. Neutralization of with Dsg3-dependent cell-cell adhesion. approximately 18 ml of eluate with 2 ml of 1 M Tris-HCl, pH 8, was Surprisingly, around day 20 after birth, DSG3−/− mice followed by dialysis against phosphate-buffered saline (PBS). displayed another prominent phenotype, hair loss. Because Antibodies were then concentrated with Centriprep Concentrators little is known about the role of desmosomes and their (Amicon, Beverly, MA). Bovine serum albumin (BSA) was added components in the structure and function of hair, we analyzed (final concentration 0.1%) and the affinity-purified antibodies were the balding phenotype of these knockout mice to learn how stored at −20oC. Dsg3 may function in the hair follicle. Western blotting In this study we demonstrate defective anchorage of telogen Whole skin extracts from newborn mice were prepared by pulverizing hairs to the folliclar epithelium due to splitting of desmosomes skin in liquid nitrogen then incubating in 2× Laemmli buffer (Bio-Rad between the keratinocytes surrounding the hair club and the Laboratories) for 10 minutes at 100°C. Immunoblotting was basal layer of the outer root sheath (ORS). This same area of performed as previously described (Koch et al., 1997). The following the hair follicle, in both mice and humans, shows intense primary antibodies were used: mouse monoclonal anti-desmoglein staining for Dsg3. These data are the first to show a structural antibody (DG3.10, Biodesign, Kennebunk, ME; specific for Dsg1 and function for a desmosomal cadherin in the hair follicle. Dsg2) (Koch et al., 1990, 1991) and rabbit anti-Dsg3 (immunoaffinity-purified, see above). Immunohistochemistry MATERIALS AND METHODS Formalin-fixed, paraffin-embedded tissue was sectioned (4 µm thick) and tissue slides were subsequently deparaffinized by incubations in Mice xylene and in ethanol (2× 5 minutes in each solvent). The tissue was The DSG3−/− mice were obtained from matings of DSG3+/− then microwaved in ‘target unmasking fluid’ (TUF; Signet, Dedham, heterozygotes (Koch et al., 1997). MA) for 4.5 minutes at 900 W. After several PBS washes, the tissue was incubated in 0.1% trypsin/PBS for 10 minutes at 41°C and then Generation of a mouse Dsg3 fusion again washed in PBS. After an incubation of 7 minutes in 3% Total RNA was isolated from the skin of the back of a two day old hydrogen peroxide solution and washing with 20% ethanol, tissues DSG3+/+ mouse with the RNAzol B reagent (Tel-Test Inc., were incubated for 5 minutes in blocking buffer (1% normal goat Friendswood, TX). Subsequently, RNA was converted into single- serum/1% BSA/PBS). The first antibody was diluted in blocking stranded cDNA using the ‘Superscript Preamplification System’ buffer and incubated with the tissues for 2 hours at room temperature (Gibco-BRL, Grand Island, NY). A cDNA fragment of 317 base pairs or overnight at 4°C. After three washes in PBS (5 minutes each), encoding the amino acid sequence of the extracellular domain 5 (EC5) biotinylated goat anti-rabbit IgG or biotinylated goat anti-mouse IgG of mouse Dsg3 (for domain designation see Amagai et al., 1991) was (Vector Laboratories, Burlingame, CA; diluted 1:200 in blocking amplified by PCR using primers MDSG3F (5′- buffer) was added and the sections were incubated for 1 hour. After GCGGGATCCACTTCCTCACCTTCTGTG) and MDSG3R (5′- three PBS washes (5 minutes each), the sections were incubated with GCGAAGCTTCCCATATGTGTTAGGCTC; cloning sites are ‘Vectastain’ (Vector Laboratories) for 30 minutes and then washed underlined). In the PCR, various amounts of cDNA were used as a with PBS (3× 5 minutes). Antibody binding was detected with ‘Stable template in the presence of 20 pmol of each primer, 200 µM dNTP, DAB’ (Research , Huntsville, AL). The DAB reactions were 2 U AmpliTaq and 1× PCR buffer (enzyme and buffer supplied by stopped by rinsing the slides several times in water. After Perkin Elmer, Norwalk, CT) in a total volume of 100 µl. The counterstaining with hematoxylin (Gill’s Formulation #3, 1:10 diluted amplification conditions were: 1 minute 94°C, 2 minutes 54°C and 2 in water; Fisher Scientific, Pittsburgh, PA), sections were dehydrated minutes 72°C followed by 30 cycles with 45 seconds 94°C, 1 minute in ethanol and, finally, coverslips were applied with Elvanol. The 54°C, 1 minute 72°C and a final incubation for 10 minutes at 72°C. following primary antibodies were used: rabbit anti-mouse Dsg3 (3.9 The PCR product was purified with the ‘QIAquick PCR µg/ml, immunoaffinity-purified, see above) and mouse monoclonal Purification Kit’ (Qiagen, Santa Clarita, CA), digested with the anti-human Dsg3 antibody 5G11 (generous gift of Dr Margaret appropriate restriction enzymes, and then cloned into the BamHI and Wheelock, University of Toledo, OH). HindIII sites of the bacterial expression vector pQE30 (Qiagen) following standard procedures. A fusion protein of approximately 12 Histology and electron microscopy kDa consisting of the mouse Dsg3 EC5 domain and six amino- Histology and electron microscopy were performed as described terminal histidine residues was expressed in JM109 (Lavker et al., 1991). (Promega, Madison, WI) and purified using Ni-NTA Agarose (Qiagen) essentially as described (Holzinger et al., 1996). RESULTS Antibody production and affinity purification of antibodies Two rabbits were immunized with the mouse Dsg3 EC5 fusion protein Analysis of timing and pattern of clinical hair loss in (Tanaka et al., 1990). DSG3 knockout mice suggests loss of telogen hair For the affinity purification of Dsg3-specific antibodies, the Dsg3 EC5 fusion protein was coupled to ‘Affi-Gel 10 Gel’ (Bio-Rad In both DSG3−/− mice and their littermates, hair appeared at Laboratories, Hercules, CA) following the protocol for aqueous the same time after birth and at day 15 the coats were coupling provided by Bio-Rad. The affinity matrix was then washed indistinguishable, indicating that the initial hair growth Desmoglein 3 in telogen hair 2531

(‘follicular neogenesis’) in DSG3−/− mice was clinically areas due to loss of telogen hair before anagen hair has a normal (Fig. 1A). As hairs are synchronized in anagen in this chance to replace it. Conversely, premature loss of telogen hair, initial hair growth, this finding suggested that anagen hairs as in these mutant mice, will result in bald areas. might be normal in DSG3 knockout mice. However, at around One week after the hair loss was first visible, most of the day 20 after birth, DSG3−/− mice started losing their coat. This head and at least part of the back were bald (Fig. 1B,C). This hair loss occurred whether the animals were housed finding clearly indicated that all truncal hair types were individually or in groups. While most animals showed signs of affected by the DSG3−/− mutation. (For a description of the hair loss 23 to 25 days after birth, some mutants lost hair as different types of mouse hairs see (Dry, 1926; Sundberg and early as 16-18 days after birth. The hair loss started on the Hogan, 1994).) One to two weeks later, hair regrowth was forehead and proceeded towards the back of the animals (Fig. visible on the head. Individual variations of this pattern were 1B,C). Previous analysis of the hair cycle in mice has shown observed. For example, hair regrowth on the head was that the early hair cycles are synchronized along the dorsal sometimes seen before total hair loss occurred on the back (Fig. body axis, in that hair goes from the first anagen into telogen 2D). Three to four weeks after the initial hair loss, most at about 18-20 days and that this transition proceeds from the animals showed a normal coat (data not shown). head to the back (Dry, 1926; Chase, 1954). The timing and The DSG3 knockout mice went through at least two pattern of hair loss in DSG3−/− mice, therefore, suggested complete cycles of hair loss and regrowth in a dorsal to ventral some defect in the telogen hair anchorage. Although the pattern. In older animals, hair was lost and regrown in a mosiac DSG3−/− mice were also losing hair on the ventral site, this (patchy) pattern, i.e. the head to tail pattern was lost (Fig. 1E). process did not seem to proceed along the body axis, perhaps This pattern of hair loss correlates with the finding that in older because the head to tail synchronization of hair growth is more normal mice hair cycles from anagen to telogen in random apparent dorsally. patches (Dry, 1926; Chase, 1954). Thus, these bald patches in Most telogen hairs of normal mice are initially retained and the DSG3−/− mice are consistent with premature loss of hair remain firmly anchored when new hair growth (anagen) is in those patches in telogen before the anagen hair has replaced initiated (for a detailed discussion see Dry, 1926). It is therefore it. not unusual to see telogen club hairs retained in a follicle To confirm loose anchorage of telogen hair in these knockout together with a growing anagen hair (see example in Fig. 2B). mice we devised a ‘hair pull’ test. A small (approx. 1 cm) piece This retention and anchorage of telogen hair prevents bald skin of adhesive tape was placed on the hair of the back, just distal

Fig. 1. Different stages of hair loss in DSG3−/− mice. (A) 15 day old DSG3−/− mouse with a normal coat. (B) About one week after the initial hair loss, the head and the neck are bald (31 day old mouse). (C,D) Two to three weeks after the initial hair loss, a new wave of hair regrowth is visible (C, 36 day old mouse; D, 37 day old mouse). The dark backskin in C indicates that the hair follicles are in anagen. Short agouti hair are seen on the back of the mouse in C as well. Note the pronounced hair growth on the forehead of D. As exemplified by C and D, individual variations in the timing and extent of the initial hairloss and the regrowth of hair were observed. (E) Older mice (3 month old mouse) lose and regrow hair in patches. (F,G) By applying adhesive tape to the back followed by a gentle pull in the direction of hair growth, lose anchoring of hair in mutant mice can easily be demonstrated (F, 23 day old DSG3−/− mouse; G, 20 day old DSG3+/+ control littermate; both mice were in telogen as confirmed by histology). (H) The hair on the adhesive tape shown in F is telogen club hair. Note that the hair club (cl) is surrounded by at least one cell layer derived from the club sheath. 2532 P. J. Koch and others to the neck, and peeled off in the direction of hair growth (i.e. of the hair growth cycle (Fig. 2B). However, in early telogen, front to back). Few or no hairs were found on tape from either hair follicles of the DSG3−/− mice showed separation between DSG3−/− animals or their DSG3+/− and +/+ littermates in the the cells surrounding the telogen hair club and the ORS, i.e. first anagen hair growth cycle (13 day old mice), showing that within the ‘club sheath’ (Dry, 1926) (Fig. 2A,C,F). This anagen hairs were well anchored. In contrast, starting on day histology was reminiscent of the PV-like lesions seen in 18, when hair follicles on the dorsum had just gone into telogen mucous membranes and traumatized skin of DSG3−/− mice, (as confirmed by histology), clumps of hair came out in where clefts were initially formed between the basal DSG3−/− mice but few, if any, hairs came out in their +/− and keratinocyte cell layer and the first suprabasal cell layer (Koch +/+ littermates (Fig. 1F,G). This result in normal mice is et al., 1997). These clefts in the telogen hair follicles were expected because, as discussed above, mice must retain telogen never observed in control mice (Fig. 2E). Histology of bald hairs until anagen hairs replace them to prevent bald areas from skin areas revealed empty, dilated telogen hair follicles, forming. Microscopic examination of the hair pulled from consistent with the loss of the telogen hairs. However, the ORS DSG3−/− mice showed telogen hairs with characteristic clubs epithelium was preserved (Fig. 2D). The histopathology (Fig. 1H). described above was also observed in older mice that underwent phases of hair loss. Histology and ultrastructure of hair loss in DSG3−/− To determine whether abnormal desmosome function was mice show loss of telogen hairs due to an adhesion associated with the loss of telogen hairs in DSG3 knockout defect around the hair clubs mice, we performed electron microscopy of hair follicles in We next evaluated the skin histology of DSG3−/− and control early telogen. Cells from the base and roof of lesions within mice (DSG3+/−, DSG3+/+) at various stages of the first hair the club sheath showed separation at their plasma membranes cycle. We did not observe any consistent differences between (Fig. 3A). In some cases, ‘half-desmosomes’ with attached knockout and control mouse hair follicles in the anagen phase cytokeratin filaments were seen within the cytoplasm,

Fig. 2. Histology from DSG3−/− (A,C,D,F) and DSG3+/+ (B,E) mice. (A) 23 day old mutant mouse with hair loss (the adhesive tape shown in Fig. 1F was derived from this mouse). Hairs are in the resting phase of the hair growth cycle (telogen). Numerous lesions are visible in the lower hair follicles (one example indicated by bracket). (B) DSG3+/+ littermate (age matched). Note that the control animal was already in anagen. (C) Telogen hair follicle of a 22 day old DSG3−/− mouse. Only the lower portion of the telogen follicles (see bracket) shows cleft formation within the club sheath. (D) Cysts (cy) in a bald area of a 24 day old DSG3−/− mouse. These structures represent the final stage of hairloss in DSG3−/− mice. (E) Telogen follicle of a normal control mouse with epithelial cell layers of the club sheath (bracket) surrounding the hair club (cl). (F) Higher magnification of the follicle shown in C. Note the cleft formation between the two epithelial cell layers. Bars, 50 µm. Desmoglein 3 in telogen hair 2533

Fig. 3. Follicular defect in DSG3−/− mice is between the basal and suprabasal keratinocytes of the outer root sheath portion of the telogen hair follicle. Ultrastructure of the lowermost portion of a follicle during the telogen phase of the hair growth cycle (A) showing a marked separation (double arrowheads) between the basal and the innermost keratinocytes of the outer root sheath (ORS). Note the close apposition of these innermost cells to the club hair (cl). ap, arrector pili muscle. High magnification of portions of the innermost keratinocytes (B,C) showing profiles of internalized ‘half- desmosomes’ (arrows) with attached filaments (arrowheads). suggestive of internalization after the desmosomes split (Fig. Taken together these ultrastructural findings are consistent with 3B,C). As reported previously, these half-desmosomes were a defect in cell-cell adhesion within the telogen club sheath. also found in suprabasilar lesions in mucous membranes of Finally, sequential skin histology of DSG3−/− and control DSG3−/− mice (Koch et al., 1997). We also noted a reduction littermate skin indicated that the initiation of the second anagen in the number of desmosomes that were formed between the (anagen II), which usually occurred at around day 22 to 23 after epithelial cell layers surrounding the DSG3−/− telogen hair birth in the skin of the back of control mice, was delayed in clubs when compared to wild-type controls (Fig. 4). This DSG3−/− mice (Table 1, compare Fig. 2A and B). In some observation is consistent with the previous observation that the animals, anagen was delayed up to 8 days. basal layer of DSG3−/− mucous membranes may have fewer desmosomes (Koch et al., 1997). In addition to a reduction in Localization of Dsg3 in mouse telogen hair follicles desmosomes, the cell surface of epithelial cells surrounding the Because Dsg3 seemed to have an important function in the DSG3−/− telogen club hair was organized into numerous telogen hair follicle (especially the club), we wanted to extensions that protruded into the enlarged determine its localization. As no antibody was available that intercellular spaces. Telogen follicles from control mice was specific for mouse Dsg3, we raised rabbit antisera against consisted of round to cuboidal keratinocytes, closely apposed a His-fusion protein containing the mouse Dsg3 fifth to each other with a normal complement of desmosomes. extracellular domain (EC5) (Amagai et al., 1991), an area that 2534 P. J. Koch and others

Fig. 4. Comparison of similar regions of telogen follicles from DSG3+/+ (A) and DSG3−/− (B) mice reveal abnormalities in cellular organization. Numerous desmosomes (arrowheads) are observed between the basal and suprabasal keratinocytes of the outer root sheath from DSG3+/+ mice (A). Follicular epithelium in this region of the hair follicle consists of two cell layers immediately adjacent to the club hair (cl). Keratinocytes that comprise the outer root sheath of uninvolved telogen follicles from a DSG3−/− mouse (B) are characterized by wide intercellular spaces (*), a marked reduction in desmosomes (arrowheads) and a cell surface organized into microvillus folds. The highly interdigitating boundary between the innermost keratinocytes cells and the club hair (cl) suggests that a tight complex is still present between these two regions of the telogen follicle. d, dermis; nu, nucleus. shows little homology among desmogleins. The antisera was Dsg3 in human telogen hair follicles and loss of hair affinity purified on the recombinant fusion protein that was used in pemphigus vulgaris for immunization (see Materials and Methods). Specificity of Immunoperoxidase staining of human telogen hair follicles this antibody was demonstrated on immunoblotting of newborn with a monoclonal antibody against human Dsg3 showed mouse skin extracts. The antibodies recognized the 130 kDa similar staining to that in the mouse, described above (Fig. 6A). Dsg3 from DSG3+/+ but not DSG3−/− skin, whereas they did The keratinocytes surrounding the hair club were intensely not recognize Dsg1 or Dsg2 from either skin extract (Fig. 5A). stained. The lateral margin of the basal cells of the ORS was The anti-Dsg3 antibody showed intense immunoperoxidase less intensely stained, however, their apices showed a strong staining of the cell surface of keratinocytes surrounding the reaction. mouse telogen hair club (Fig. 5B). The basal layer of the ORS Because PV is a blistering mediated by at the hair club showed cell surface staining, most intensely autoantibodies against Dsg3 in which the mucous membrane between the apex of the cells and the cells surrounding the club and skin phenotype is similar to DSG3−/− mice, we speculated (exactly where the split occurs, arrow in Fig. 5B). There was that there might be telogen hair loss in these patients. As less intense staining between the lateral margins of the ORS opposed to mice in which essentially all hair in a patch are basal cells (arrowhead in Fig. 5B) and no staining at the synchronized in the hair cycle, adult human hair growth is basement membrane. DSG3−/− mice showed no staining at all asynchronous with only a small fraction of the total scalp hair (data not shown). These data correlate the localization and follicles in telogen. Therefore, telogen hair loss may not be an function of Dsg3 in the telogen hair follicle. obvious clinical phenotype in pemphigus patients. In addition Desmoglein 3 in telogen hair 2535

Fig. 5. Immunoblot and immunohistochemical detection of Dsg3. (A) Western blots demonstrating the specificity of the immunopurified rabbit anti-Dsg3 antibodies. Skin extracts from newborn DSG3−/− (−/−) and DSG3+/+ (+/+) mice were probed with mouse monoclonal antibody DG3.10 (αDsg1+2) or rabbit polyclonal anti-Dsg3 antibodies (αDsg3). The rabbit antibodies specifically detect a protein of 130 kDa in DSG3+/+ skin which is absent in the DSG3−/− skin. DG3.10 recognizes of approx. 150 kDa (Dsg1 and Dsg2; Schmelz et al., 1986, Koch et al., 1990, 1991) in both extracts. (B) Immunohistochemical detection of Dsg3 in a telogen hair follicle of a DSG3+/+ mouse. DSG3−/− hair follicles do not stain with this antibody (data not shown). The position of the club (cl) and basement membrane (bm, see broken line) are indicated. Note that Dsg3 is absent from the basement membrane but clearly present at cell-cell borders (arrowhead). The Dsg3 staining is particularly pronounced in the cell layer proximal to the club (arrow). Bar, 50 µm. there are not many scalp biopsies from pemphigus patients. epidermis down the entire ORS of mostly empty dilated hair Nevertheless, we did find histological evidence for telogen hair follicles that contained only small residual unanchored hair loss in one patient. In this case acantholysis extended from the shafts and necrotic detached cells (Fig. 6B). Similar findings of acantholysis extending down the ORS of hair follicles have been reported, although hair loss has not been commented upon (Ackerman, 1978). Therefore, even though hair loss is not a major phenotype in pemphigus patients, there are abnormalities in the hair follicles, consistent with a role for Dsg3 in the structure and function of human hair.

DISCUSSION

The data in this report demonstrate the importance of Dsg3, a desmosomal cadherin, in the structure and function of hair. On the dorsum of the mouse, hair follicles in any particular area cycle in a synchronous fashion and enter telogen at approximately the same time (Dry, 1926; Chase, 1954). It is therefore not surprising that the telogen defect in our DSG3−/− mice resulted in a dramatic loss of hair in a large area of the coat following a defined pattern. In the absence of Dsg3, as soon as hair follicles enter telogen, a defect in adhesion between the cells surrounding the

Table 1. The hair growth cycle on back skin (shoulder region) of DSG3−/− and control mice Age (days) Number of animals Hair cycle Fig. 6. Immunohistochemical detection of Dsg3 in a human telogen 13 2mt, 2ct1 Anagen hair follicle and hair histopathology of a pemphigus vulgaris patient. 18* 1mt, 3ct2 Telogen, catagen, anagen (A) Immunohistochemical detection of Dsg3 in the club sheath 20* 2mt, 2ct1 Telogen 22* 1mt, 2ct1 Telogen (bracket) of a human telogen hair follicle. The position of the 3 basement membrane (bm) is indicated (broken line). (B) Scalp 23±1* 2mt, 2ct mt: telogen; ct: early anagen 24±1* 3mt, 2ct4 mt: telogen; ct: early anagen biopsy from a PV patient showing loss of suprabasilar epidermal cell 25±1* 2mt, 2ct1 mt: telogen; ct: early anagen layers. Only basal keratinocytes (bk), which are attached to the 31±1* 1mt, 1ct5 mt: early anagen; ct: anagen basement membrane, are left. The lesion extends into the two hair follicles (hf) shown. One cell layer of outer root sheath-derived *Cell-cell separation in the club sheath of telogen hair follicles in DSG3−/− keratinocytes lines the dilated hair follicles. Arrows point towards mice. 1One DSG3+/+, one DSG3+/−; 2two DSG3+/+, one DSG3+/−; 3two residual unanchored hair shafts. Asterisk demarcates necrotic DSG3+/+; 4two DSG3+/−; 5one DSG3+/−. mt, mutant (DSG3−/−); ct, control detached outer root sheath cells. Bars, 50 µm. (DSG3+/−, DSG3+/+). 2536 P. J. Koch and others mouse hair club and the basal cells of the ORS, i.e. within the formation between the outer and inner root sheath cell layers. club sheath, causes loss of anchorage of the club hair to its These findings are fundamentally different from the ones follicle. Electron microscopy of the affected area demonstrates reported in this paper because the pathology in DSG3−/− mice ‘half-desmosomes’ and widened intercellular spaces, is seen both at a different stage of the hair cycle and in a suggesting that although desmosomes form, their adhesive different area of the follicle. We did not find any gross or function is abnormal. Furthermore, the number of desmosomes histological indication for lesions in the anagen hair of within the telogen club sheaths of DSG3−/− mice is reduced, DSG3 −/− mice. If there is indeed a difference with respect to contributing to the fragility of this tissue. We have previously the pathology in balPAS and DSG3−/− mice, this difference reported analagous histologic and ultrastructual findings in could be attributed either to the as yet undefined nature of the studies of the acantholysis that occurs in the stratified balPAS mutation or the genetic background of the two mouse squamous epithelium of epidermis, oral and vaginal mucous lines. Our mice were on a segregating Sv129 and C57Bl/6 membrane of these DSG3 knockout mice (Koch et al., 1997). background, whereas balPAS was congenic to the 129/SvPas Consistent with these findings suggesting a role for Dsg3 in inbred strain. anchoring the telogen hair club to the ORS, we demonstrate The finding of lesions within mouse hair follicles due to the intense immunoperoxidase staining for Dsg3 in normal mice absence of Dsg3 prompted us to see whether similar lesions in this area, analagous to the localization of Dsg3 in the basal were present in PV patients. First we showed that, as in mice, and immediate suprabasal layers of epidermis, where intense staining for Dsg3 occurs in the cells surrounding the suprabasilar acantholysis occurs in DSG3−/− mice. telogen hair club. Based on the hypothesis that PV results from In spite of this dramatic telogen hair loss in DSG3−/− mice, a loss of Dsg3 function caused by Dsg3-specific the first anagen hair cycle is normal in both growth and autoantibodies, one might expect to find hair pathology in anchorage, demonstrating that the absence of Dsg3 is, in patients. However, in contrast to the synchronized hair growth principle, compatible with hair growth and that anagen hair has seen in mice, in adult humans hair follicles cycle independently alternative or additional anchorage mechanisms compared to from each other and only about 5-20% of follicles on the scalp telogen hair. Although subsequent hair cycles may be delayed, are in telogen (Kligman, 1961; Courtois et al., 1996). Therefore the bald skin areas are still capable of growing new hair, thus any clinical phenotype in PV would be subtle if only telogen showing that the stem cells and follicular papillae, which are hair were affected. In one biopsy from a PV patient, we were required for hair growth, retain their functions (Cotsarelis et able to demonstrate loss of the hair from follicles. However, al., 1990). We did not detect any defects in anagen hairs in any because we had only one biopsy showing the appropriate hair subsequent hair cycles either. As we could identify follicles, more extensive studies will be necessary to determine acantholysis only in telogen hair, the defect seems limited to to what extent hair loss occurs in these patients and which part that part of the hair cycle and is not one of loss of anagen hair of the hair growth cycle is affected. causing the follicle to enter telogen. The results presented in this paper are a first step in starting The reason for the delay of the second anagen in DSG3−/− to dissect out the function of desmosomes and their mice is unclear. However, it has been reported that malnutrition components in the structure and function of mammalian hair. can delay or even prevent the initiation of anagen in mice (Chase, 1954) (see also the description of ‘badly grown’ mice We thank Dora Dogadkin (electron microscopy), Dorothy in (Dry, 1926)). DSG3−/− mice are runted, most likely due to Campbell (histology), William Witmer (photography) and Zhi Hong oral erosions that prevent sufficient food intake (Koch et al., Wang for excellent technical assistance. We thank Dr Margaret 1997). We therefore believe that malnutrition may cause the Wheelock, University of Toledo, OH, for providing us with antibody delay of anagen in these mice. Alternatively, it is possible that 5G11. Peter J. Koch and M G. Mahoney were supported by research fellowships from the Dermatology Foundation. This work has been the premature loss of telogen club hair creates a wound that supported by National Institutes of Health grant AR43776. contributes to this phenomenon. Montagutelli and colleagues recently described a mouse PAS mutation (bal ) with a hair phenotype similar to that of REFERENCES DSG3 knockout mice (Montagutelli et al., 1997). Although these authors did not determine the defect in these mice at the Ackerman, A. B. (1978). Histologic Diagnosis of Inflammatory Skin Diseases. molecular level, they did state that this mutation is coallelic Lea & Febiger, Philadelphia. with the balding mutation (balJ; Davisson et al., 1994; Amagai, M., Klaus-Kovtun, V. and Stanley, J. R. (1991). Autoantibodies Montagutelli et al., 1997; Sundberg, 1994). We have recently against a novel epithelial cadherin in pemphigus vulgaris, a disease of cell J adhesion. Cell 67, 869-877. shown that the latter mutation (bal ) represents a spontaneous Amagai, M., Karpati, S., Prussick, R., Klaus-Kovtun, V. and Stanley, J. R. DSG3−/− mutation (Koch et al., 1997). The predicted amino (1992). Autoantibodies against the amino-terminal cadherin-like binding acid sequence encoded by the mutated DSG3 gene of the balJ domain of pemphigus vulgaris antigen are pathogenic. J. Clin. Invest. 90, mouse lacks most of the cytoplasmic sequences of the wildtype 919-926. protein due to a nucleotide insertion in exon 14 that causes a Amagai, M., Hashimoto, T., Shimizu, N. and Nishikawa, T. (1994). Absorption of pathogenic autoantibodies by the extracellular domain of frame-shift and premature stop codon. Consequently, no Dsg3 pemphigus vulgaris antigen (Dsg3) produced by baculovirus. J. Clin. Invest. protein was detected on the cell surface of keratinoctes in mice 94, 59-67. homozygous for the balJ mutation (Koch et al., 1997). Due to Amagai, M., Hashimoto, T., Green, K. J., Shimizu, N. and Nishikawa, T. its coallelism with the balJ mutation, it can be assumed that the (1995). Antigen-specific immuoabsorption of pathogenic autoantibodies in balPAS mutation prevents the synthesis of a functional Dsg3. pemphigus foliaceus. J. Invest. Dermatol. 104, 895-901. PAS Amagai, M., Koch, P. J., Nishikawa, T. and Stanley, J. R. (1996). Pemphigus Montagutelli et al. (1997) attributed the hair loss of bal mice vulgaris antigen (Desmoglein 3) is localized in the lower epidermis, the site to a defect in the anagen hair follicle, in particular to cleft of blister formation in patients. J. Invest. Dermatol. 106, 351-355. Desmoglein 3 in telogen hair 2537

Arnemann, J., Sullivan, K. H., Magee, A. I., King, I. A. and Buxton, R. S. L., Murphy, G. F., Whitaker-Menezes, D. and Stanley, J. R. (1997). (1993). Stratification-related expression of isoforms of the desmosomal Targeted disruption of the pemphigus vulgaris antigen (desmoglein 3) gene in human epidermis. J. Cell Sci. 104, 741-750. in mice causes loss of keratinocyte cell adhesion with a phenotype similar Chase, H. B. (1954). Growth of the hair. Physiol. Rev. 34, 113-126. to pemphigus vulgaris. J. Cell Biol. 137, 1091-1102. Cotsarelis, G., Sun, T. T. and Lavker, R. M. (1990). Label-retaining cells Lavker, R. M., Dong, G., Zheng, P. S. and Murphy, G. F. (1991). Hairless reside in the bulge area of pilosebaceous unit: implications for follicular micropig skin. A novel model for studies of cutaneous . Am. J. stem cells, hair cycle and skin carcinogenesis. Cell 61, 1329-1337. Pathol. 138, 687-697. Courtois, M., Loussouarn, G., Hourseau, S. and Grollier, J. F. (1996). Montagutelli, X., Lalouette, A., Boulouis, H. J., Guenet, J. L. and Periodicity in the growth and shedding of hair. Br. J. Dermatol. 134, 47-54. Sundberg, J. P. (1997). Vesicle formation and follicular root sheath Davisson, M. T., Cook, S. A., Johnson, K. R. and Eicher, E. M. (1994). separation in mice homozygous for deleterious alleles at the balding (bal) Balding: a new mutation on mouse chromosome 18 causing hair loss and locus. J. Invest. Dermatol. 109, 324-328. immunological defects. J. Hered. 85, 134-136. Schafer, S., Koch, P. J. and Franke, W. W. (1994). Identification of the Dry, F. W. (1926). The coat of the mouse (mus musculus). J. Genet. 16, 287- ubiquitous human desmoglein, Dsg2, and the expression catalogue of a 340. subfamily of desmosomal cadherins. Exp. Cell Res. 211, 391-399. Franke, W. W., Koch, P. J., Schafer, S., Heid, H. W., Troyanovsky, S. M., Schmelz, M., Duden, R., Cowin, P. and Franke, W. W. (1986). A constitutive Moll, I. and Moll, R. (1994). The desmosome and the syndesmos: cell transmembrane glycoprotein of Mr 165,000 (desmoglein) in epidermal and junctions in normal development and in malignancy. [Review] [62 refs]. non-epidermal desmosomes. II. Immunolocalization and microinjection Princess. Takamatsu. Symp. 24, 14-27. studies. Eur. J. Cell Biol. 42, 184-199. Holzinger, A., Phillips, K. S. and Weaver, T. E. (1996). Single-step Schmidt, A., Heid, H. W., Schafer, S., Nuber, U. A., Zimbelmann, R. and purification/solubilization of recombinant proteins: application to surfactant Franke, W. W. (1994). Desmosomes and cytoskeletal architecture in protein B. Biotechniques 20, 804-6, 808. epithelial differentiation: cell type-specific plaque components and Karpati, S., Amagai, M., Prussick, R., Cehrs, K. and Stanley, J. R. (1993). intermediate filament anchorage. Eur. J. Cell Biol. 65, 229-245. Pemphigus vulgaris antigen, a desmoglein type of cadherin, is localized Schwarz, M. A., Owaribe, K., Kartenbeck, J. and Franke, W. W. (1990). within keratinocyte desmosomes. J. Cell Biol. 122, 409-415. Desmosomes and : constitutive molecular components. Kligman, A. M. (1961). Pathologic dynamics of human hair loss. 1. Telogen Annu. Rev. Cell Biol. 6, 461-491. effluvium. Arch. Dermatol. 83, 175-198. Stanley, J. R. (1993). Cell adhesion molecules as targets of autoantibodies in Koch, P. J., Walsh, M. J., Schmelz, M., Goldschmidt, M. D., Zimbelmann, pemphigus and pemphigoid, bullous diseases due to defective epidermal cell R. and Franke, W. W. (1990). Identification of desmoglein, a constitutive adhesion. Adv. Immunol. 53, 291-325. desmosomal glycoprotein, as a member of the cadherin family of cell Sundberg, J. P. (1994). Handbook of Mouse Mutations with Skin and Hair adhesion molecules. Eur. J. Cell Biol. 53, 1-12. Abnormalities: Animal Models and Biomedical Tools (ed. J. P. Sundberg), Koch, P. J., Goldschmidt, M. D., Walsh, M. J., Zimbelmann, R. and pp. 187-191. CRC Press, Boca Raton. Franke, W. W. (1991). Complete amino acid sequence of the epidermal Sundberg, J. P. and Hogan, M. E. (1994). Handbook of Mouse Mutations desmoglein precursor polypeptide and identification of a second type of with Skin and Hair Abnormalities (ed. J. P. Sundberg), pp. 57-68. CRC, Ann desmoglein gene. Eur. J. Cell Biol. 55, 200-208. Arbor. Koch, P. J. and Franke, W. W. (1994). Desmosomal cadherins: another Tanaka, T., Korman, N. J., Shimizu, H., Eady, R. A. J., Klaus-Kovtun, V., growing multigene family of adhesion molecules. Curr. Opin. Cell Biol. 6, Cehrs, K. and Stanley, J. R. (1990). Production of rabbit antibodies against 682-687. carboxy-terminal epitopes encoded by bullous pemphigoid cDNA. J. Invest. Koch, P. J., Mahoney, M. G., Ishikawa, H., Pulkkinen, L., Uitto, J., Shultz, Dermatol. 94, 617-623.