Involvement of hepatocyte growth factor/scatter factor and met receptor signaling in hair follicle...

Involvement of hepatocyte growth factor/scatter factorand Met receptor signaling in hair folliclemorphogenesis and cycling

GERD LINDNER,*,1 ANDREAS MENRAD,† ERMANNO GHERARDI,‡ GLENN MERLINO,§

PIA WELKER,* BORI HANDJISKI,* BIRGIT ROLOFF,* AND RALF PAUS¶,2

*Department of Dermatology, Charite, Humboldt-University, Berlin, Germany; †ResearchLaboratories of Schering AG, Berlin, Germany; ‡Growth Factors Groups, Department of Oncology,MRC Centre, Cambridge, U.K.; and §Laboratory of Molecular Biology, National Cancer Institute,Bethesda, Maryland 20892-4255, USA; and ¶Department of Dermatology, University HospitalEppendorf, University of Hamburg, D-20246, Hamburg, Germany

ABSTRACT HGF/SF and its receptor (Met) areprincipal mediators of mesenchymal–epithelial in-teractions in several different systems and haverecently been implicated in the control of hair folli-cle (HF) growth. We have studied their expressionpatterns during HF morphogenesis and cycling inC57BL/6 mice, whereas functional hair growth ef-fects of HGF/SF were assessed in vivo by analysis oftransgenic mice and in skin organ culture. In normalmouse skin, follicular expression of HGF/SF andMet was strikingly localized: HGF/SF was found onlyin the HF mesenchyme (dermal papilla fibroblasts)and Met in the neighboring hair bulb keratinocytes.Both HGF/SF and Met expression peaked duringthe initial phases of HF morphogenesis, the stageof active hair growth (early and mid anagen),and during the apoptosis-driven HF regression (cata-gen). Met1 cells in the regressing epithelial strandappeared to be protected from undergoing apopto-sis. Compared to wild-type controls, transgenic miceoverexpressing HGF/SF under the control of theMT-1 promoter had twice as many developing HFand displayed accelerated HF development on post-natal day 3. They also showed significant catagenretardation on P17. In organ culture and in vivo,HGF/SF i.c. resulted in a significant catagen retar-dation. These results demonstrate an important roleof HGF/SF and Met in murine hair growth controland suggest that Met-mediated signaling might beexploited for therapeutic manipulation of humanhair growth disorders.—Lindner, G., Menrad, A.,Gherardi, E., Merlino, G., Welker, P., Handjiski, B.,Roloff, B., Paus, R. Involvement of hepatocytegrowth factor/scatter factor and Met receptor signal-ing in hair follicle morphogenesis and cycling.FASEB J. 14, 319–332 (2000)

Key Words: apoptosis z hair growth z HGF/SF z keratinocytesz murine hepatocyte growth factor receptor

Hepatocyte growth factor/scatter factor(HGF/SF) and its receptor, the product of theproto-oncogene Met (hepatocyte growth factor re-ceptor), have surfaced as versatile modulators of cellproliferation, migration, differentiation, and survivalduring mesenchymal–epithelial interactions in awide range of mammalian tissue interaction systemsin health and disease (for a review, see refs 1–3.

HGF/SF is a pleiotropic factor isolated indepen-dently as a growth-promoting agent for liver cells (4)and as a fibroblast-derived protein leading to disso-ciation and motility of epithelial cells (scatter factor)(5, 6). The factor is synthesized as a 728 amino acidprecursor that is processed by different proteases—e.g., HGF activator, tPA, or urokinase—to generatethe mature growth factor consisting of a disulfide-linked 69 kDa a chain and a 34 kDa b chain (1, 2).

Physiologically, HGF/SF has been described as amesenchymally derived, paracrine regulator of epi-thelial cells expressing the Met receptor. However,autocrine activation of HGF/SF promotes tumorgen-esis as demonstrated by the fact that transgenic (TG)mice overexpressing HGF/SF develop a remarkablybroad array of histologically distinct tumors of bothmesenchymal and epithelial origin (7).

Recently, HGF/SF has been implicated in hairgrowth control, a prototypical epithelial–mesenchy-mal interaction system. One intriguing feature of thehair follicle (HF) is that after a prolonged period ofgrowth (anagen), it spontaneously enters into aphase of rapid, apoptosis-driven, organ involution(catagen) until the follicle reenters anagen via aninterspersed resting phase (telogen) (8–10). Thisprocess is controlled by a stringent cross talk be-

1 This publication represents part of the Ph.D. thesis of G.L.2 Correspondence: Department of Dermatology, University

Hospital Eppendorf, University of Hamburg, Martinistr. 52,D-20246, Hamburg, Germany. E-mail: [email protected]

3190892-6638/00/0014-0319/$02.25 © FASEB

tween mesenchymally derived dermal papilla (DP)fibroblasts and epithelial keratinocytes. No othermammalian organ rhythmically undergoes, duringthe entire life span of the organism, such dramatic,but physiological, apoptosis and a rapid reconstruc-tion of an entire organ by remodeling the architec-ture of the follicle epithelium (8, 10, 11).

Several families of signaling molecules have re-cently been implicated in the control of HF develop-ment. After formation of the hair bud prior to birthin sonic hedgehog (Shh) knockout mice, HF devel-opment appears to be arrested due to a disruptedepithelial/mesenchymal communication (12). InTG mice overexpressing constitutively active b-cate-nin, an effector of intercellular adhesion signaling,de novo HF development occurs (13). Yet the exactrole in physiological HF induction remains to beclarified. However, HGF/SF stimulation increasesthe synthetic rate of b-catenin (14) and leads tonuclear accumulation of b-catenin (15) in epithelialcells in vitro.

Hair follicle morphogenesis and the periodicalgrowth and regression of murine HF criticallydepend on precisely organized mesenchymal– epi-thelial signaling (8, 10, 16, 17). In this context,HGF/SF and its receptor (Met) are of significanceto hair biologists, since this signaling system hasbeen shown to stimulate growth and motility ofkeratinocytes (18 –20), to promote melanogenesisof human melanocytes (19), is involved in thecontrol of angiogenesis in vivo (21–24), and hasthe ability to dissociate epithelial cells by modulat-ing the expression of cell adhesion molecules suchas cadherins (3, 25, 26).

Cultured dermal papilla cells of the HF weredescribed to express HGF/SF, which stimulatesgrowth of human HF in vitro (27). In addition,HGF/SF has been reported to stimulate the growthof mouse vibrissae follicles in vitro (20). A dose-dependent increase of DNA and protein synthesis aswell as an elongation of hair shafts could be mea-sured by adding human-derived HGF/SF to theculture medium (20). The same group reported thathuman recombinant HGF/SF can delay murine HFregression and promote the growth of pelage HF invivo (28), suggesting that exogenous HGF/SF canmanipulate HF cycling in mice.

We, therefore have explored the question ofwhether endogenous HGF/SF and Met are involvedin the regulation of hair follicle morphogenesis andcycling (growth, regression, resting) using theC57BL/6 mouse model (9, 29–35). The immunohis-tological expression patterns of HGF/SF and Metwere characterized during all stages of HF morpho-genesis and cycling. This was complemented by ananalysis of HGF/SF and Met gene expression duringsynchronized HF cycling (semiquantitative reverse

transcriptase-polymerase chain reaction, or RT-PCR).Furthermore, recombinant murine HGF/SF was ad-ministered to murine skin organ culture and in vivo(34, 35) in order to test its effects on spontaneous HFregression (catagen development). Finally, the charac-teristics of HF morphogenesis and cycling were com-pared by quantitative histomorphometry (12, 32, 35)between wild-type (WT) and TG mice overexpressingHGF/SF (7, 36). The results of these experimentsdemonstrate a major role for HGF/SF and Met inmurine HF morphogenesis and cycling.

MATERIALS AND METHODS

Animal models and tissue collection

Six- to 8-wk-old female C57BL/6 mice in the telogen stage ofthe hair cycle, weighing 15–20 g, were purchased fromCharles River (Sulzfeld, Germany). The animals were housedin community cages at the animal facilities of the VirchowHospital (Berlin) with 12 h light periods, and were fed waterand mouse chow ad libitum. The growth phase of the haircycle (anagen) was induced in the back skin of mice with allfollicles in the resting phase of the hair cycle (telogen; asjudged from their homogeneously pink back skin color) byapplication of a melted wax/rosin mixture (1:1) and undergeneral anesthesia as described previously (31, 32). After17–19 days, these depilation-induced anagen follicles enterspontaneously into catagen, as can be appreciated from theconversion of their skin color from black to gray, and finallyto pink hair shaft plucking. All stages of the anagen/catagen/telogen transformation of the murine hair cycle were studied(classification according to ref 29).

For analysis of HF morphogenesis, embryonic and neona-tal skin was harvested at various time points of estimatedgestational age (embryonic days E12.5, 16.5, 17.5, and 18.5;E0.5 5 the morning on which a vaginal plug is found) orpostnatal age (postnatal days P0–P19) (16). Most non-tylotrich pelage HF in murine back skin develop perinatally;even 1 day after birth (P1), all stages of HF morphogenesiscan still be found in mouse skin (33, 37). Stages of HFmorphogenesis were assessed by morphological criteria, asdescribed in detail elsewhere (38).

The neck region of murine back skin was harvestedparallel to the vertebral line to obtain longitudinal sectionsthrough the hair follicles (38, 39) The skin was deep frozenin liquid nitrogen, covered with embedding medium, andprocessed for immunohistochemistry and TUNEL staining(TdT-mediated dUTP-digoxigenin nick end-labeling) asdescribed below.

Transgenic mice

Inactivation of the HGF/SF gene (40) and that of met (41)cause embryonic lethality in mice between E12.5 and E16.5,around the time when HF formation is initiated. This rendersconventional knockout mice unsuitable for studies of HFmorphogenesis and cycling. Therefore, the skin of HGF/SFoverexpressing TG mice was analyzed. These TG mice weregenerated by inserting a mouse HGF/SF cDNA under thecontrol of the mouse metallothionein (MT-I) promoter (42),thus driving strong HGF/SF expression within the skin (7).TG mice express a characteristic 2.4 kb RNA in virtually alladult tissues at a level between 3- to 50-fold higher than the

320 Vol. 14 February 2000 LINDNER ET AL.The FASEB Journal

major 6 kb endogenous HGF/SF transcript (43). Mice werehoused in the animal facilities of the National Cancer Insti-tute, Bethesda, Md. Neonatal skin fragments from three tofive different TG and age-matched WT mice as controls(postnatal days 3 and 17) were washed repeatedly in phos-phate-buffered saline (PBS) buffer, fixed in 4% paraformal-dehyde, and embedded in paraffin for routine histology(hematoxilin/eosin staining) and quantitative histomor-phometry (32, 35, 38).

RT-PCR

Steady-state levels of skin HGF/SF and Met gene expressionwere studied by semiquantitative RT-PCR (35, 44, 45). TotalRNA was isolated from full-thickness back skin samples (ho-mogenized in liquid nitrogen), using a single-step guanidinethiocyanate-phenol-chlorophorm method (RNeasy total RNAkit, Quiagen, Hilden, Germany). Skin samples included thesubcutaneous (s.c.) skeletal muscle layer (panniculus carno-sus). cDNA was synthesized by reverse transcription of 3 mgtotal RNA, using a cDNA synthesis kit (Invitrogen, San Diego,Calif.) (in the case of HGF/SF, the amount of cDNA used forRT-PCR was 10-fold higher). The following sets of oligonu-cleotide primers (TIP Molbiol, Berlin, Germany) were used toamplify specific c-DNA—b-actin: 59-GAA AAC GCA GCT CAGTAA CAG TCC G and 59-TAA AAC GCA GCT CAG TAA CAGTCC G-39 (45); HGF/SF: 59-TT GGC CAT GAA TTT GACCTC-39 59-AC ATC AGT CTC ATT CAC AGC-39 (46); andMet: 59-GAA TGT CGT CCT ACA CGG CC-39 and 59-CAGGGG CAT TTC CAT GTA GG-39 (46).

Amplification was performed using Taq polymerase (LifeTechnologies, Inc., Grand Island, N.Y.) over 30 cycles, usingan automated thermal cycler (Perkin Elmer Cetus, Norwalk,Conn.). Each cycle consisted of the following steps: denatur-ating at 94°C (1 min), annealing at 60°C (45 s), and extensionat 72°C (45 s). PCR products were analyzed by gel electro-phoresis (2% agarose) (44, 45). For semiquantitative RT-PCR,linear correlation of signal intensity for b-actin was foundbetween 24 and 27 cycles, and for the other markers, between30 and 35 cycles using computer-assisted video scanner den-sitometry (ScanPack 2.0; Biometra, Gottingen, Germany).This allowed relative comparison of the signals from differentsamples.

Photographs and graphical presentations represent theresults of three independent experiments derived from threedifferent animals.

Immunohistochemistry

For simultaneous immunodetection of HGF/SF and Metwithin murine skin, cryosections were processed by combin-ing standardized immunofluorescence (IF) labeling methods(9). Cryostat sections (10 mm) were fixed in acetone (10 minat 220°C) and preincubated with 10% donkey normal serum,followed by an incubation with the primary sheep anti-mouseHGF/SF antibody overnight at 4°C obtained after repeatedimmunizations with purified recombinant HGF/SF. Sectionswere then incubated with FITC-conjugated donkey anti-sheepsecondary antibody (1:200, Jackson Laboratories, West Grove,Pa.) for 30 min at room temperature.

In the second part of this double-staining protocol, sec-tions were blocked with 10% goat normal serum in PBS for 10min at room temperature and incubated overnight withrabbit anti-Met antibody (Santa Cruz Biotechnology, SantaCruz, Calif.) at 4°C. The secondary antibody (goat anti-rabbitTRITC-conjugated F(ab)2 fragments, 1:200, Jackson Labora-tories) was incubated 30 min at room temperature.

Nuclei were covisualized by HOECHST 33342 (Sigma,

Deisenhofen, Germany). Each step (except after blocking)was interspersed by washing with PBS. Finally, slides weremounted with VectorShield (Vector Laboratories, Burlin-game, Calif.). Positive staining for HGF/SF and Met wasidentified by green and red fluorescence, respectively. As anegative control for Met, neutralizing peptide (1:50 in PBS;Santa Cruz Biotechnologies) was preincubated with the anti-Met antibody for 1 h at room temperature.

The distribution and intensity of HGF/SF- and Met-IR(Met-immunoreactivity/immunoreactive) cells of at least 50different HF per mouse (n55), were determined with afluorescence microscope (Zeiss, Jena, Germany) at magnifi-cations from 100 to 4003. All specific and reproducibleIR-patterns were recorded and summarized in computer-generated schematic representations (Designer 4.0, Micro-grafx) designed to reflect the key features of murine hairfollicle anatomy and their changes during hair follicle devel-opment and cycling as accurately as possible (33, 38). Pho-tomicrographs were processed using a digital image analysissystem (ISIS Metasystems, Altusheim, Germany).

Combined immunolabeling with TUNEL/Hoechst 33242staining

To evaluate apoptotic cells in colocalization with the IRpattern of the Met receptor or Ki67, a key marker forproliferating cells, we used a previously described combinedTUNEL/Hoechst 33342/antibody triple-staining method (9,47). In brief, 10 mm cryostat sections of C57BL/6 back skinwere freshly prepared and fixed in formalin, postfixed inethanol/acetic acid, and incubated with digoxigenin-dUTP inthe presence of TdT, followed by incubation with eitherrabbit anti-Met (Santa Cruz Biotechnology) or rabbit anti-Ki67 (Dianova, Hamburg, Germany) antiserum, respectively.Subsequently, TUNEL-positive cells were visualized by anti-digoxigenin FITC-conjugated F(ab)2 fragments, whereas Met-or Ki67-IR was detected by a TRITC-labeled goat-anti-rabbitantibody (Jackson Laboratories, West Grove, Pa.). Counter-staining with HOECHST 33342 dye (10 mg ml-1 in PBS,Sigma, Deisenhofen, Germany) was performed by a subse-quent incubation step. Finally, sections were mounted usingVectaShield (Vector Laboratories). Positive TUNEL controlswere run, as described (48), by comparison with tissuesections from the thymus of young mice, which display a highdegree of spontaneous thymocyte apoptosis. Negative con-trols for the TUNEL staining were made by omitting terminaldesoxynucleotidyl transferase (TdT), according to the man-ufacturer’s protocol. Sections were then examined under aZeiss Axioscope microscope (Jena, Germany), using the ap-propriate excitation-emission filter systems for studying thefluorescence, induced by Hoechst 33342, FITC, or TRITC.Photodocumentation was done with the help of a digitalimage analysis system (ISIS Metasystems)

In vivo HGF/SF injection

Recombinant HGF/SF was produced in the NS0 mousemyeloma line transfected with a full length mouse HGF/SFcDNA and purified from culture supernatants with a combi-nation of heparin-Sepharose6CL and Mono S column chro-matography (Amersham Pharmacia; M. Sharpe and E. Gher-ardi, unpublished results).

To check the effect of HGF/SF on spontaneous catagendevelopment (cf. refs 35, 47), the dorsal back skin of 14-day-old female C57BL/6 young mice (n55) with all HF in thefinal stage of HF morphogenesis was injected intradermallywith recombinant HGF/SF 8 mg/20 g body weight in 100 mlPBS (0.1% bovine serum albumin, or BSA) with consecutive

321HGF/SF AND MET IN MURINE HAIR GROWTH CONTROL

injections on days 15 and 16. The animals were killed at day17 p.d. (postdepilation: day after anagen induction by depi-lation). Note that around P17, murine pelage HF synchro-nously enter into HF cycling by entry into the first spontane-ous catagen (16, 33, 38). In addition, 17-day-old mice (n55)with all HF in the anagen/catagen transformation wereinjected intradermally with 1 mg/20 g body weight recombi-nant HGF/SF in 100 ml PBS (0.1% BSA) with a consecutiveinjection at day 18 p.d. The mice were killed on day 19 p.d. Allcontrol mice were injected with PBS (0.1% BSA) alone.

In both experiments, the neck region around the injectionsite of the dorsal skin was harvested and processed forsubsequent TUNEL/Hoechst 33342/Ki67 staining as de-scribed above.

To demonstrate the concentration and gradient-depen-dent catagen retardation in HGF/SF-treated skin, four con-secutive microscopic fields of a single representative skinsection were taken and merged using a computer-aideddigital image analysis system (ISIS Metasystems).

Skin organ culture

Four millimeter punch biopsies were prepared under sterileconditions from adolescent C57BL/6 mouse back skin withall HF in the late anagen VI or early catagen stage of theinduced hair cycle (i.e., day 17 after depilation; ref 29),following previously described protocols (31, 34, 35, 47, 49)with some modifications. For each experimental group, eightto ten randomized skin punches derived from the back skinof three different mice were placed (dermis down) on gelatinsponges (Gelfoam, Upjohn Co., Kalamazoo, Mich.) in 35 mmpetri dishes, containing 5 ml Williams E supplemented with50 mg/ml L-glutamine, 1% antibiotic/antimycotic mixture(Life Technologies), 0.1% hydrocortisone, and 0.25% insu-lin. After addition of 10 ng/ml recombinant mouse HGF/SF,organ cultures were incubated at the air-liquid interphase for48 h at 37°C in 5% CO2 and 100% humidity, with one changeof medium and the appropriate amount of growth factor after24 h. At the end of incubation, all skin fragments were washedrepeatedly in PBS, fixed in 4% paraformaldehyde, and em-bedded in paraffin for routine histology and histomorphom-etry.

Quantitative histomorphometry and statistical analysis

IR patterns were scrutinized by studying at least 50 differ-ent HF per mouse and five mice were assessed per stage.For each stage of HF morphogenesis and cycling, the majorIR patterns were recorded in previously prepared, comput-er-generated schematic representations of murine HF mor-phogenesis and cycling, which allow a standardized, easilyreproducible, and systematic comparison of different fol-licular IR patterns (33, 38).

The number of hair follicles per unit length of epidermis(HF ostia) was calculated in paraffin sections of HGF/SFtransgenic skin (n53) at P3 and compared to that of age-matched wild-type controls (n53). The percentage of hairfollicles in different stages of morphogenesis was assessed anddefined on the basis of accepted morphological criteria (16,38). At least 60 longitudinally cut HF sections in .50 micro-scopic fields, derived from three HGF/SF TG animals, wereanalyzed and compared to that of $3 age-matched wild-typemice.

For counting the number of HF in HGF/SF-transgenicand age-matched wild-type mice, horizontally sectionedtissue were screened using established morphological cri-teria (16, 38, 51).

At least 100 measurements from 4 sections per animal were

performed and compared to wild-type skin, using digitalimage analysis system (Zeiss KS400, Jena, Germany). Sectionswere analyzed at 3200 or 3400 magnification, and meansand se were calculated from pooled data. Differences werejudged as significant if the P value was ,0.05, as determinedby the independent Student’s t test for unpaired samples.

Since differences in HF morphogenesis and cycling areassociated with distinct changes in the overall skin thickness(12, 31, 38, 50, 51), skin thickness was compared between TGand WT skin by assessing the distance between the stratumcorneum and the distal border of the s.c. muscle layer(panniculus carnosus). In total, 40–50 such measurementswere performed in a corresponding number of differentmicroscopic fields derived from three animals per mutantand wild-type group. All sections were analyzed at 320 and3200 magnification, and means and se were calculated frompooled data. Differences were judged as significant if the Pvalue was lower than 0.05 as determined by the independentStudent’s t test for unpaired samples.

RESULTS

Met gene expression in normal mouse skin is haircycle dependent

As a first phenomenological indicator for a possibleinvolvement of HGF/SF and Met signaling in hairgrowth control, HGF/SF and Met gene transcriptionin full-thickness adolescent mouse skin were charac-terized by semiquantitative RT-PCR analysis duringthe induced murine hair cycle (44, 45). AlthoughMet transcripts were not detectable in mouse skinwith all HF in the resting stage of the hair cycle(telogen; Fig. 1A, B, day 0), RNA extracts fromfull-thickness mouse skin showed two peaks of steady-state levels of Met transcripts during early (Fig. 1A, B,day 1) and late anagen (Fig. 1A, B, day 12) as well asduring the anagen-catagen transformation (Fig. 1A,B, day 17). Only low Met mRNA steady-state levelswere seen in mid-anagen skin (Fig. 1A, B, day 8). Lowbut detectable steady-state levels of HGF/SF-mRNAwere seen during early and the late anagen (Fig. 1A,B, days 3 and 12) (note that 10-fold more concen-trated cDNA had to be used for the RT-PCR reactioncompared to Met and b-actin), indicating thatHGF/SF mRNA is transcribed at comparatively lowlevels in murine skin.

The follicular expression of HGF/SF and Met isstage specific and localized to the mesenchyme(HGF/SF) and epithelium (Met)

Next we analyzed the IR patterns for the correspond-ing proteins by IF. Hair follicle morphogenesis wascharacterized by a developmentally regulated HGF/SF-IR pattern that was restricted exclusively to mes-enchyme-derived dermal papilla fibroblasts and theirprecursor cells whereas Met distribution was seen inthe neighboring follicle epithelium.

During all stages of hair follicle morphogenesis,


most epidermal keratinocytes (both in the basal andsuprabasal layers) were Met-IR (Fig. 2A, B, red fluo-rescence; Fig. 3 stage 0–8, red). During the visibleonset of HF morphogenesis (stage 1), the invaginat-ing keratinocytes that later form the new hair plugwere all Met-IR (Fig. 2A, asterisk; Fig. 3A). Stage 2 ischaracterized by formation of the hair peg, whichwas also homogeneously positive for Met (Fig. 2B,arrowhead; Fig. 3B). During stages 3 and 4, Met-IRdeclined in the outer cell layers of the distal hair pegbut remained noticeable in the developing proximalinner root sheath (IRS) (Fig. 2C, arrowhead; Fig.3C). Clustered dermal fibroblasts in close vicinity tothe proximal hair peg showed neither Met- norHGF/SF-IR during the early stages of HF morpho-genesis (stages 1 to 3) (Fig. 2A, B; Fig. 3A, B). Duringstage 4, DP fibroblasts were the first cells to becomehomogeneously HGF/SF-IR (Fig. 2C, green fluores-cence, arrow; Fig. 3C, green). DP fibroblasts of the

developing HF remained HGF/SF-IR until the onsetof catagen, by which stage the HF entered the firsthair cycle (Fig. 2D–F, arrowheads; Fig. 3D–F). Duringstage 5, the distal portion of the HF epithelium,including the gradually formed bulge/isthmus re-gion, showed prominent Met-IR (Fig. 3D), whereasthe hair matrix became Met negative (Fig. 2D; Fig.3D).

Whereas Met-IR declined from the proximal andcentral ORS and proximal IRS during the subse-quent stages 6–8 (Fig. 2E; Fig. 3E), DP fibroblastsremained strongly HGF/SF-IR (Fig. 2E, arrowhead;Fig. 3E).

With entry into the first adult hair cycle, which isinitiated by HF regression (10, 38), catagen I and IIwere associated with an increase of Met-IR in theregressing hair bulb, particularly in the proximal IRSand ORS (Fig. 2F, arrows; Fig. 3F ). HGF/SF-IRdramatically declined from the shrinking DP (Fig.2F, arrowhead) and was absent within the DP duringthe subsequent catagen stages (Fig. 2G; Fig. 3G).Throughout catagen, the central IRS and ORS wereMet negative whereas distal ORS keratinocytes, in-cluding the bulge/isthmus region, displayed Met-IRduring catagen stages I-VII (Fig. 3F–H ). The keratin-ocytes of the regressing hair matrix directly abovethe DP became Met negative (Fig. 2G, arrowhead;Fig. 3G), whereas the residual proximal IRS and ORSof the regressing HF epithelium remained Met-IR foras long as they existed during catagen (Fig. 2G,arrow; Fig. 3G). In catagen VI and VII, Met wasprominently expressed in the secondary hair germand the epithelial strand (Fig. 2H, arrowheads; Fig.3H ), where massive apoptosis occurred spontane-ously (see Fig. 2M, arrow) (9).

During the resting stage of the adolescent haircycle (telogen), no Met- or HGF/SF-IR were seen inmurine hair follicles (Fig. 2I; Fig. 3I ) (except in thecells of the arrector pili muscle (Fig. 2I, arrowhead),which were Met-IR during the entire hair cycle (Fig.3F–L). Transient amplifying cells in the follicle epi-thelium just above the DP (52) were the first tobecome Met-IR in early anagen I (Fig. 2J, arrowhead;Fig. 3J ). Keratinocytes of the developing IRS andouter root sheath (ORS) began to show substantialMet-IR in anagen II and III (Fig. 2K, arrow; Fig. 3K ).Anagen III was the first stage of the adolescent haircycle where HGF/SF-IR was again seen within the DP(Fig. 2K, arrowhead; Fig. 3K ). Note that, morpho-logically, anagen III resembles to some extent stage 4of HF development (compare Fig. 2D, K and Fig. 3D,K ) since anagen development recapitulates in partHF morphogenesis (10, 38, 53, 54). HGF/SF re-mained consistently expressed in the DP duringanagen III-catagen II (Fig. 2F, K, L, arrowheads; Fig.3F, K, L). The proximal and central IRS and thecentral and distal ORS were Met-IR in anagen IV

Figure 1. Semiquantitative RT-PCR of HGF/SF- and MetmRNA during the depilation-induced hair cycle. A) mRNAsteady-state levels during the induced hair cycle. At definedtime points total RNA from full-thickness skin of C57BL/6mice was extracted and reverse transcribed. A semiquantita-tive RT-PCR was performed, using primers specific for Met,b-actin and HGF/SF (10-fold more concentrated cDNA wasused for the RT-PCR reaction compared to Met and b-actin).Representative results from one of three experiments: days 0and 25, resting stage (telogen); days 1–12, active hair growth(anagen); days 17–19, apoptosis-driven HF regression (cata-gen). B) Densitometric analysis of RT-PCR signals specific forHGF/SF (black bars) and Met (dashed bars) (means 6 ofn53). X axis: days after anagen induction by depilation. Thecorresponding hair cycle stages are indicated at the bottom.Asterisks indicate significant differences between defined HFstages of the murine hair cycle, Student’s t test, *P,0.05.


Figure 2. HGF/SF and Met immunoreactivity (IR) during normal murine hair follicle morphogenesis and cycling. Cryostatsections (10 mm) of mouse skin were double-stained by HGF/SF (green fluorescence) and Met (red fluorescence) antisera.Counterstaining of cell nuclei was done by Hoechst 33342 (blue fluorescence) (A–L). To evaluate apoptotic cells incolocalization with the IR pattern of the Met-receptor (M) or Ki67, a key marker for proliferating cells (N), we used a previously


(Fig. 3, anagen IV) and anagen V, whereas Met-IRdisappeared from the proximal hair matrix in lateanagen VI (Fig. 2L; Fig. 3L). Central IRS and ORSwere prominently Met-IR (Fig. 2L; arrows; Fig. 3L)and the distal ORS including the bulge/isthmusregion remained Met positive (Fig. 3L). The IRpatterns for Met and HGF/SF during the subsequentcatagen phase of the adolescent hair cycle wereexactly as described above for the first catagendevelopment in infantile mice (not shown).

Since HGF/SF expression seen by immunohisto-chemistry was restricted to the relatively smallpopulation of dermal papilla fibroblasts, this mayexplain why only a very low HGF/SF transcriptnumber was detected by RT-PCR (Fig. 1A, B). Wefailed to demarcate convincing in situ hybridiza-tion (ISH) signals for HGF/SF mRNA in thismouse model despite rigorous attempts and theuse of different in situ probes and ISH techniquesfollowing previously successful ISH protocols forthe murine system (46, 55).

Expression of Met-IR and TUNEL positivityappeared to be mutually exclusive

Based on observations that the cutaneous steady-state levels of Met transcripts and the correspondingprotein expression were up-regulated during theanagen-catagen transformation (Figs. 1–3) and thatMet-IR was expressed particularly strongly in selectedcompartments of the regressing hair follicle epithe-

lium during catagen (Fig. 2H, M, arrowheads; Fig.3H ), we next asked whether or not keratinocytesundergoing catagen-associated apoptosis in the re-gressing epithelial strand and the secondary hairgerm of catagen HF coexpress Met-IR. Using atechnique for double immunovisualization ofTUNEL-positive and Met-IR cells in the regressingHF, we demonstrated that TUNEL1 (i.e., apoptoticcells) did not display Met-IR (Fig. 2M, arrow). Thissuggested that Met expression awards HF keratino-cytes some degree of protection from apoptosis andthat administration of HGF/SF could inhibit apop-tosis-driven catagen (see below).

Injection of HGF/SF in vivo retards catagendevelopment

To probe this concept, recombinant mouseHGF/SF was injected intradermally into the backskin of either 14- or 17-day-old mice, whose backskin HF were all in anagen VI, a few days beforethe spontaneous onset of catagen (P14) or thatwere about to undergo catagen development(P17) (8, 10, 33, 38).

As shown in Fig. 2N and Fig. 4, recombinantHGF/SF indeed significantly retarded spontaneouscatagen development in C57BL/6 mice during boththe early and late stages of HF regression. HGF/SF-injected mice (first injection on day 14 p.d.) showeda gradually delayed catagen development from thearea of injection to the periphery (Fig. 2N ), as

described combined TUNEL/Hoechst 33342/antibody triple staining method (9, 47). The images represent IR pattern derivedfrom analyzing .50 longitudinally sectioned follicles from the lower back of 5 harvested C57BL/6 mice per time point.Abbreviations: APM, arrector pili muscle; DP, dermal papilla; TAC, transient amplifying cells; E, epidermis; (T)ES, (trailing)epithelial strand; HM, hair matrix; IRS and ORS, inner and outer root sheath; SG, sebaceous gland; SHG, secondary hair germ.Scale bars are 50 mm (except those for panels E, I, K, L, which are 100 mm, and panel N, which is 200 mm). A) Startingmorphogenesis stage 1, invaginating keratinocytes estimated to form the new hair plug became Met-IR (red fluorescence,asterisk). Epidermal keratinocytes were Met-IR during all stages of the HF morphogenesis. B) Stage 2 is characterized byformation of the hair peg, which was positively stained for Met (arrowhead), whereas during stage 3 and later stages Met-IRdeclined in the cell layers of the outer hair peg, but remained stained in the inner most region of the peg. C) DP fibroblasts ofstage 4 follicles were the first cells to become HGF/SF-IR (arrow); they increased in intensity during stage 5 and the developingproximal IRS became Met-IR (arrowhead). D) During stage 5, DP fibroblasts became increasingly HGF/SF-IR (arrow) and thecentral portion of the HF remained Met-IR (arrow). E) Met-IR further declined from the proximal IRS and ORS (arrowhead),whereas the distal portion remained constantly stained. F) While entering the first hair cycle, HGF/SF-IR of DP fibroblastsdeclined dramatically in catagen II (arrowhead) and totally disappeared from catagen III-anagen II. Met-IR increased in theproximal IRS and ORS (arrows) in catagen II. G) IRS keratinocytes directly above the DP showed no Met-IR, whereas residuaryproximal and central IRS/ORS prominently expressed Met during catagen V. H) The trailing epithelial strand (TES)(arrowhead) and the secondary hair germ (SHG) were Met highlighted during catagen VI–VIII. I) During telogen, almost noMet-IR (except the muscle arrector pili, arrowhead) and no HGF/SF-IR were seen in murine HF. J) Transient amplifying cells(TAC) in the follicle epithelium just above the DP were the first to become Met-IR in very early anagen (arrowhead). K) Duringthe adult hair cycle, DP fibroblast for the first time became HGF/SF-IR during anagen III (arrowhead), and remained staineduntil catagen II. Met-IR was increased in the developing proximal IRS (arrow). L) Anagen VI HF showed an intensivelyHGF/SF-IR DP (arrowhead) and Met-IR was prominently expressed in the central IRS (arrow) and ORS. M) TUNEL1 cellsshowed no colocalization with Met-IR (arrow). The secondary hair germ displayed strong Met-IR (arrowhead). N) TheTUNEL/Hoechst 33342/Ki67-stained representative histology section from a 17-day-old mouse shows a gradually retardedcatagen development toward the injection site of recombinant HGF/SF (open arrowhead), with HF ranging from late anagenVI with isolated TUNEL1 cells (green fluorescence, arrowheads) and clusters of Ki67-IR keratinocytes of the proximal HF (pinkfluorescence, arrows) to catagen VIII HF with an assemblage of TUNEL1 cells within the trailing epithelial strand (arrowheads)and the lack of Ki67-IR cells in the proximal HF epithelium. The image was derived from four consecutive photomicrographsof the same representative skin section and was recomposed with the help of a digital imaging system.


demonstrated by TUNEL/Hoechst 33342/Ki67 tri-ple staining. On day 17, close to the injection site(open arrowhead), late anagen VI HF that wereabout to undergo the anagen-catagen transforma-tion displayed almost no TUNEL1 cells, but insteadshowed dense clusters of Ki671 keratinocytes in theproximal follicle epithelium (Fig. 2N, arrows). Alongthe HGF/SF gradient (i.e., in the periphery of thesite of growth factor injection), most HF alreadydisplayed the signs of far progressed catagen devel-opment, as demonstrated by a dramatically increasedamount of TUNEL1 cells in the regressing epithelialstrand and secondary hair germ (Fig. 2N, arrow-heads) (9), along with a lack of cells proliferating inthe proximal HF (Fig. 2N, arrows). Complementaryto this highly suggestive, qualitative observation, thequantitative assessment of HF stages around theHGF/SF injection site confirmed the presence ofsignificantly higher numbers of HF in earlier catagenstages compared to vehicle-treated control mice,which displayed an increased amount of HF in verylate catagen and telogen (Fig. 4A).

Quantitative histomorphometry of mice injectedat day 17 of the first adult hair cycle also revealed aretardation of catagen development in HGF/SF-treated mice. On day 19, the majority of HF invehicle-treated mice were in catagen VIII or telogenstage (Fig. 4B, black bars), whereas skin sections ofHGF/SF-treated mice showed a significantly re-tarded catagen development, as evident from thepredominance of HF in late catagen (catagen stagesVII and VIII) (Fig. 4B, gray bars). This accelerationof catagen development by one stage correspondedto a significant, morphologically easily recognizabledifference in skin thickness between HGF/SF-treated (442.3128.6 mm) and control skin

proximal IRS. Catagen II: While entering the first hair cycle,HGF/SF-IR of DP fibroblasts declined in catagen II anddisappeared from catagen III-anagen II. Met-IR increased inthe proximal IRS and in catagen II. Catagen V: Met-IRbecomes negative in the lower ORS keratinocytes of theregressing hair matrix above the DP, but remained positive inthe proximal IRS and ORS. DP fibroblasts displayed noHGF/SF-IR. Catagen VII: In catagen VII, Met is prominentlyexpressed in the trailing epithelial strand (TES) and thesecondary hair germ (SHG). Telogen: During telogen, almostno Met- (except the muscle arrector pili) or HGF/SF immu-noreactivity (IR) was seen in murine HF. Anagen I: Transientamplifying cells in the follicle epithelium just above the DPwere again the first to become Met-IR in very early anagen.Anagen III: Keratinocytes of the developing IRS, began toshow substantial Met-IR in anagen II and III. DP fibroblastbecame HGF/SF-IR again during anagen III–VI. Anagen IV:The proximal and central IRS and the central and distal ORSwere Met-IR in anagen IV and V. HGF/SF-IR remainedstrongly expressed during anagen development. Anagen VI:Met-IR disappeared from the proximal hair matrix in lateanagen VI. The central IRS and ORS were prominentlyMet-IR and the distal ORS including the bulge/isthmusregion remained Met positive.

Figure 3. Schematic representation of summarized immuno-reactivity patterns of HGF/SF and Met during selected stagesof murine hair follicle morphogenesis and subsequent hairfollicle cycling. Cell populations with Met-IR expression aredepicted as red regions, those with HGF/SF-IR are shown ingreen. Key stages of murine hair follicle morphogenesis andcycling are indicated according to (56) with modifications(8). The summary scheme was derived from analyzing .50longitudinally sectioned follicles from the lower back of 5harvested C57BL/6 mice per time point. Abbreviations: APM,arrector pili muscle; BM, basal membrane; DP, dermal pa-pilla; TES, trailing epithelial strand; IRS and ORS, inner andouter root sheath; SG, sebaceous gland; SHG, secondary hairgerm; TES, trailing epithelial strand. The arrow placed be-tween stage 8 and catagen II indicates the initiation of HFcycling by entry into catagen. Stage 0: During morphogenesis,epidermal keratinocytes were Met-IR and remained Metpositive during subsequent stages (stages 1–8, red). Stage 1:Starting morphogenesis stage 1, invaginating keratinocytes,estimated to form the new hair plug, became Met-IR. Stage 2:Stage 2 is characterized by formation of the hair peg, whichwas positively stained for Met. Stage 4: During stage 4 andfurther stages, Met-IR declined in the distal cell layers of theouter hair peg, but remained stained in the proximal regionof the peg. Dermal papilla (DP) fibroblasts of stage 4 follicles,were the first cells to become HGF/SF-IR (green). Stage 5:Within the DP, HGF/SF-IR increased intensity during stage 5and the proximal ORS became Met negative during stage 5and after developmental stages (stages 5–8), whereas thedistal portion of the HF, including the forming bulge/isthmus region remained Met-IR. Stage 8: DP remainedstrongly HGF/SF positive, whereas Met-IR declined from the


(294.7134.3 mm, P .0.05). Murine skin thickness isstrictly coupled to synchronized HF cycling andanagen VI skin is considerably thicker than catagen

skin, which again is thicker than telogen skin (31, 50,56), which indicated that the catagen-telogen transi-tion in vehicle-treated mice had almost been com-pleted at this time point, further attesting to thepotent catagen-inhibitory activity of HGF/SF in vivo.

HGF/SF overexpression increases the number ofdeveloping HF and accelerates HF morphogenesis

Given the highly differential, developmentally con-trolled expression of HGF/SF during HF morpho-genesis (Figs. 2 and 3), we next explored whetherHGF/SF overexpression affects HF development.Neonatal HGF/SF transgenic mice were comparedwith age-matched WT mice controls for the numberof HF that had developed as well as for the speed ofHF morphogenesis. In normal mice, almost all HFwere in the early to mid stages of HF morphogenesis(stages 4 to 6) around postnatal day 3 (P3). InHGF/SF overexpressing mice, most HF were insteadalready in advanced stages of HF morphogenesis(stages 7–8) (Fig. 5A, C), whereas practically all HFof wild-type mice were in earlier stages (5 and 6) atthis time point (Fig. 5B, C). This relative accelerationof HF morphogenesis in HGF/SF transgenic micewas independently confirmed by the fact that, theskin was substantially thicker in HGF/SF-overex-pressing mice (735.2161.2 mm) (P,0.05) at P3compared to wild-type mice (423.3152.1 mm), as iseasily visible in the representative photomicrographsshown in Fig. 5A, B. Since murine skin thickness isstrictly coupled to synchronized HF cycling and skinthickness is dramatically increasing during sequen-tially morphogenic stages (38), this strongly supportsthe concept that HGF/SF overexpression acceleratesHF morphogenesis.

Moreover, a significantly increased total number(twofold) of hair bulbs was counted at P3 in trans-genic mice (Fig. 5D; Fig. 5F, white bar) compared toWT controls (Fig. 5E; Fig. 5F, black bar). In addition,as an independent, even more reliable parameter,the amount of HF that had developed by P3 was alsoassessed by counting the number of detectable HFostia per millimeter epidermal length (16, 38, 51) soas to avoid the problem of counting individual HFmore than once. This supplementary quantitativehistomorphometrical assessment also revealed ahighly significant increase (P,0.001) in the numberof HF ostia in HGF/SF overexpressing mice (Fig. 5G,white bar) compared to age-matched wild-type mice(Fig. 5G, black bar). These data underscore thepotent HF morphogenic activity of HGF/SF.

HGF/SF overexpression retards catagendevelopment

After the completion of HF morphogenesis (which isoften mislabeled as the ‘first hair cycle’), the hair

Figure 4. Recombinant HGF/SF is capable to retard catagendevelopment in vivo. Cryosections (10 mm) from C57BL/6 mice,treated with i.c. recombinant HGF/SF or vehicle at differentstages of the depilation-indudced hair cycle, were processed forTUNEL/Hoechst 33342/Ki67 triple fluorescent staining in or-der to demonstrate the capability of HGF/SF to retard catagendevelopment. Quantitative histomorphometry was used as well-defined criteria to determine the different stages of catagenprogression. The percentage of HF in defined catagen stageswas evaluated according to well-defined morphological criteriaon day 17 after depilation. The dynamic flow chart shows a shiftof retarded catagen development in HGF/SF treated mice withmost HF in catagen VII (A, hashed bar), whereas in control micethe majority of HF display catagen VIII (A, black bar). Micetreated with recombinant HGF/SF at progressive catagen stages(injected at days 17 and 18, harvested at day 19 p.d.), display asignificantly enhanced number of HF in catagen VII and VIII (B,gray bars), compared to vehicle-treated mice, which show mostHF already in catagen VIII or telogen (B, black bars). A, B)Asterisks indicate significant differences between defined cata-gen stages in HGF/SF injected vs. control mice, Student’s t test,*P,0.05).


Figure 5. HGF/SF overexpression in neonatal mice (P3) displays twice asmuch follicles and is associated with accelerated hair follicle morphogen-esis. For HF counting and hair cycle staging, skin paraffin sections ofHGF/SF overexpressing and corresponding age-matched wild-type micewere processed for hematoxilin/eosin staining. Percentage of HF indefined stages of HF morphogenesis was evaluated by quantitative histo-morphometry according to well-defined morphological criteria (16, 54) onday 3 (P3) (A–C) and the absolute number of HF unit was counted in HFbulbs (D–F) and epidermal ostiae/mm epidermis units (G) of strictlyhorizontally cut skin section as the most reliable criteria to avoid multiplecounting of the same follicle. The representative pictures demonstrate theprogression of hair follicle development in HGF/SF overexpressing mice,where the majority of stage 7–8 HF deeply reach into the broad subcutis(A), whereas wild-type mice only displayed HF in stages 5–6,which arelocated at the dermis-subcutis border (B). The graphical representationsummarizes the percentage of HF stages per 10 microscopic fields (MF),which peaked at stage 8 in HGF/SF overexpressing mice (C, open circles)

and at stage 6 in age-matched wild-type mice (C, black squares). During neonatal morphogenesis at P3, HGF/SF transgenic miceshowed a significant (twofold) increase in the number of hair follicles counted as hair bulbs (D) compared to age-matched,wild-type mice (E) summarized in panel F. A graphical representation of the epidermal ostiae/mm epidermis showed a highlysignificant increase in HF of HGF/SF overexpressing mice (F, open bar) compared to age-matched wild-type mice (F, black bar).***P,0.001. Scale bars: 100 mm in panels A and B, and 50 mm in panels D, E.


follicle begins its life-long cycle of regression,resting, and growth by spontaneous entry into thefirst catagen stage (10), which occurs around P17(8, 10, 38).

Since HGF/SF-overexpressing TG mice showed asignificant acceleration of follicle development duringneonatal morphogenesis on P3, it was most intriguingthat the HF in transgenic skin sections from P17showed signs of a significant retarded catagen develop-ment compared to age-matched wild-type littermates(Fig. 6A–C), even though their morphogenesis likelywas completed earlier than in control skin (Fig. 5A–C).As documented by representative photomicrographsand quantitative histomorphometry, the majority of HFin HGF/SF TG mice at P17 were still in late anagen VIto early catagen (Fig. 6A, C, open circles), whereas themajority of WT HF already were in late stages ofcatagen development (Fig. 6B, C, black squares) at thistime.

Recombinant HGF/SF retards catagendevelopment in skin organ culture

To address the question of whether the observedalterations in HF cycling in HGF/SF transgenic andHGF/SF-injected mice might be connected to con-

sequences of systemic or other secondary effects ofHGF/SF, such as an altered innervation or vascularsupply, rather than reflecting a direct role ofHGF/SF in hair follicle control, HGF/SF was addedto organ-cultured murine skin (47, 49). For thispurpose, biopsies were taken from normal C57BL/6mouse skin 17 days after anagen induction by depi-lation (31, 32) so that skin contained homogeneous,well-defined HF populations about to undergo spon-taneous, apoptosis-driven regression (9, 29, 32) andwere devoid of functional innervation or vascularsupply. These skin fragments were cultured at theair-liquid interphase on gelatin gels for 48 h in thepresence or absence of recombinant HGF/SF.

In line with the results of the in vivo studies delin-eated above, quantitative histomorphometric analysisrevealed that HGF/SF-supplemented skin biopsiesshowed a significant retardation of catagen develop-ment in skin organ culture. The majority of HF in skinbiopsies that had been cultured in the presence of 10ng/ml of recombinant HGF/SF displayed catagenstage VI (Fig. 7, open circles), whereas most HF invehicle-treated control skin were already in catagen VII,VIII, or telogen (Fig. 7, black squares). Thus, HGF/SFinhibits catagen development by a local, intracutane-ous effect on HF cycling .

Figure 6. HGF/SF overexpression in neonatal mice displays retarded catagen development on P17. For HF counting and haircycle staging, skin paraffin sections of HGF/SF overexpressing and corresponding age-matched wild-type mice were processedfor hematoxilin/eosin staining. Percentage of HF in defined stages of HF morphogenesis was evaluated by quantitativehistomorphometry according to well-defined morphological criteria (16, 54) on day 17 (P17) after birth (A–C). **P,0.01. Scalebar: 100 mm. The representative pictures demonstrate the retardation of hair follicle catagen development in HGF/SFoverexpressing mice, where all stage 8 and early catagen I HF deeply reached into the broad subcutis (A), whereas wild-type micedisplayed already regressing HF in late catagen stages of the first adolescent hair cycle. Note the already shrunken skin thickness(B). The graphical representation summarizes the percentage of HF stages, which peaked at stage 8/early catagen I in HGF/SFoverexpressing mice (C, open circles) and late catagen VI in age-matched wild-type mice (C, black squares).


DISCUSSION

HF morphogenesis results in the development of aunique, hair shaft-producing mini-organ, which arisesfrom tightly choreographed epithelial–mesenchymalinteractions and the recruitment and coordination ofstringently regulated programs of keratinocyte differ-entiation, proliferation, and apoptosis (63). Every stepin the extensive tissue remodeling events that consti-tute HF morphogenesis and the subsequent HF cyclingappears to be controlled by changes in the localbalance of stimulating and suppressing activities ofdifferent growth factors and their corresponding re-ceptors (8, 10, 17, 47, 57, 58). Here we provide evi-dence that HGF/SF and Met, an essential growthfactor/receptor pair for the development and functionof a variety of mammalian organs (1–3, 40, 41, 43, 59),is one such factor.

We report the first comprehensive expression mapof HGF/SF and Met during murine hair follicle mor-phogenesis and cycle. The described follicular expres-sion patterns of Met and HGF/SF demonstrate astriking developmental regulation and hair cycle de-pendence, and suggest that under physiological condi-tions, Met expression is restricted to the HF epitheliumwhereas its ligand is produced by the HF mesenchyme.This designates the HF an exquisite model system forfurther dissection of the role of HGF/SF in epithelial–mesenchymal interactions in general. The transient

and strikingly localized mesenchymal appearance ofHGF/SF in the DP (during morphogenesis and duringanagen III to catagen II of the hair cycle), along withthe substantial epithelial Met expression in the vicinityof HGF/SF-expressing DP fibroblasts, provided sugges-tive phenomenological evidence for an involvement ofthis growth factor-signaling system in hair growth con-trol, both during morphogenesis and cyclic remodel-ing of this mini-organ.

HGF/SF and its receptor (Met) are involved in thecontrol of angiogenesis in vivo (21–24). Since angio-genesis is closely coupled to HF anagen development(60), this might be an additional argument for thesignificance of HGF/SF and Met to hair biology.

Extracellular proteolytic cleavage by urokinase isrequired for activation of HGF/SF, and nexin-1 as wellas PA inhibitor type 1 is a potential inhibitor ofurokinase (61). Nexin-1 is also an inhibitor of throm-bin and t-PA, and has been reported to be an impor-tant activity-limiting factor HGF/SF (1, 61). Nexin-1 isexpressed exclusively in rat DP fibroblasts during ana-gen, whereas it is absent during catagen (62). Thiscorresponds well to the observation that HGF/SF-IR ismost prominent during anagen (see Fig. 3), and sug-gests that an antiproteolytic system (nexin-1) operatesin the HF, thus regulating HGF/SF activity.

Anagen induction likely requires multiple intercon-nected signals that initiate Met expression to achievecompetency for HGF/SF binding. This might be ac-complished by key regulatory factors implicated in thecontrol of HF formation during morphogenesis such asShh, b-catenin, Bmp family members, and their antag-onists (12, 13, 16, 58). The strong morphogenic poten-tial of the HGF/SF/Met signaling system within mu-rine skin is supported by the finding that HGF/SF-overexpressing transgenic mice displayed a twofoldincrease in the number of HF and a significantlyaccelerated HF morphogenesis.

In addition, we show that HGF/SF overexpressionretards HF regression in vivo and that HGF/SF iscapable of reproducing this effect in denervated skinorgan culture and in vivo, suggesting that HGF/SF/Met signaling rank among the elusive molecular con-trols of catagen (8, 10, 63). DP-derived HGF/SF may beimportant for supporting epithelial cell growth and forsuppressing follicle keratinocyte apoptosis during ana-gen development and maintenance. Since no Met-IRwas seen in TUNEL1 cells, while the level of Mettranscripts was highest during early catagen, selectedMet-expressing cells in the regressing epithelial strandmay be rescued from apoptosis during catagen bysuccessfully competing for the declining level ofHGF/SF produced by DP cells (Fig. 2). Thus, theHGF/SF/Met signaling system may contribute to con-trol of apoptosis-driven (9) catagen development bythe induction of keratinocyte apoptosis via HGF/SFdeprivation, as has been described for several other

Figure 7. HGF/SF retards catagen development in murine skinorgan culture. Punch biopsies, taken from C57BL/6 mouseback skin at day 17 of the depilation-induced hair cycle (i.e., withall HF about to enter into the anagen VI-catagen transforma-tion), were incubated at the air-liquid interphase during 48 h inthe presence of 10 ng/ml of recombinant HGF/SF, and per-centage of HF at the distinct catagen stages was quantitativelyevaluated as described above. Compared to vehicle controls(black squares), a significant retardation of catagen develop-ment was seen in skin fragments, incubated with HGF/SF (opencircles), demonstrated in the percentage of the catagen III-V,catagen VI, and catagen VIII HF (mean6se. n5 8–10 biopsiesper group: Student’s t test, asterisks indicate significant differ-ences to the control, *P,0.05).


model systems (64–66). HGF/SF has been reported tobe a potent cell survival factor that suppresses epithelialcell apoptosis (67–69). BAG-1, a functional bindingpartner of the apoptosis inhibitory protein bcl-2, canassociate with Met, thereby linking the HGF/SF/Metsignaling pathway with the anti-apoptotic machinery(67, 70).

It is unclear whether HGF/SF is also involved inanagen induction. However, we recently noted thatcyclosporin A, a potent trigger for hair growthinduction (30, 71), causes a premature expression ofHGF/SF in dermal papilla fibroblasts (G. Lindnerand R. Paus, unpublished results). Therefore, cyclo-sporin A may induce anagen, at least in part, by theup-regulation of intrinsic HGF/SF, thus triggeringpremature keratinocyte proliferation and differenti-ation in the surrounding HF epithelium.

In conclusion, our study implicates the HGF/SFand Met system in hair growth control and indicatesthat Met agonists and antagonists should be consid-ered as novel agents for therapeutic hair growthcontrol (cf. refs. 8, 63).

The excellent technical assistance of R. Pliet, E. Hagen, E.Andermarcher, and R. Sharpe is most gratefully acknowl-edged. This study was supported in part by grants from DFG(Pa 345/8–2) and Pierre Fabre to R.P. and from the DAAD toG.L. Work in E.G.’s laboratory is supported by the U.K.Medical Research Council.

REFERENCES

1. Birchmeier, C., and Gherardi, E. (1998) Developmental roles ofHGF/SF and its receptor, the c-Met tyrosine kinase. Trends CellBiol. 8, 404–410

2. Trusolino, L., Pugliese, L., and Comoglio, P. M. (1998) Inter-actions between scatter factors and their receptors: hints fortherapeutic applications. FASEB J. 12, 1267–1280

3. Jiang, W. G., and Hiscox, S. (1997) Hepatocyte growth factor/scatter factor, a cytokine playing multiple and converse roles.Histol. Histopathol. 12, 537–555

4. Nakamura, T., Nawa, K., Ichihara, A., Kaise, N., and Nishino, T.(1987) Purification and subunit structure of hepatocyte growthfactor from rat platelets. FEBS Lett. 224, 311–316

5. Stoker, M., Gherardi, E., Perryman, M., and Gray, J. (1987)Scatter factor is a fibroblast-derived modulator of epithelial cellmobility. Nature (London) 327, 239–242

6. Gherardi, E., Gray, J., Stoker, M., Perryman, M., and Furlong, R.(1989) Purification of scatter factor, a fibroblast-derived basicprotein that modulates epithelial interactions and movement.Proc. Natl. Acad. Sci. USA 86, 5844–5848

7. Takayama, H., LaRochelle, W. J., Sharp, R., Otsuka, T., Kriebel,P., Anver, M., Aaronson, S. A., and Merlino, G. (1997) Diversetumorigenesis associated with aberrant development in miceoverexpressing hepatocyte growth factor/scatter factor. Proc.Natl. Acad. Sci. USA 94, 701–706

8. Paus, R. (1996) Control of the hair cycle and hair diseases ascycling disorders. Curr. Opin. Dermatol. 3, 248–258

9. Lindner, G., Botchkarev, V. A., Botchkareva, N. V., Ling, G., vander Veen, C., and Paus, R. (1997) Analysis of apoptosis duringhair follicle regression (catagen). Am. J. Pathol. 151, 1601–1617

10. Paus, R., Muller-Rover, S., and Botchkarev, V. (1999) Chrono-biology of the hair follicle: hunting the hair cycle clock. J. Invest.Dermatol. Symp. Proc. In press

11. Jahoda, C. A., and Reynolds, A. J. (1996) Dermal epidermalinteractions: adult follicle derived cell populations and hairgrowth. Dermatol. Clin. 14, 573–584

12. St-Jacques, B., Dassule, H. R., Karavanova, I., Botchkarev, V. A.,Li, J., Danielian, P. S., McMahon, J. A., Lewis, P.M., Paus, R., andMcMahon, A. P. (1998) Sonic hedgehog signaling is essentialfor hair development. Curr. Biol. 8, 1058–1068

13. Gat, U., DasGupta, R., Degenstein, L., and Fuchs, E. (1998)De Novo hair follicle morphogenesis and hair tumors inmice expressing a truncated beta-catenin in skin. Cell 95,605– 614

14. Balkovetz, D. F., Pollack, A. L., and Mostov, K. E. (1997)Hepatocyte growth factor alters the polarity of Madin-Darbycanine kidney cell monolayers. J. Biol. Chem. 272, 3471–3477

15. Papkoff, J., and Aikawa, M. (1998) WNT-1 and HGF regulateGSK3 beta activity and beta-catenin signaling in mammaryepithelial cells. Biochem. Biophys. Res. Commun. 247, 851–858

16. Philpott, M. F., and Paus, R. (1998). Principles of hair folliclemorphogenesis. In Molecular Biology of Skin Appendages Morpho-genesis (Chuong,C. M., ed) pp. 75–103, Landes Bioscience,Austin, Texas

17. Stenn, K. S., Combates, N. J., Eilertsen, K. J., Gordon, J. S.,Pardinas, J. R., Paromoo, S., and Prouty, S. M. (1996) Hairfollicle growth control. Dermatol. Clin. 14, 543–557

18. McCawley, L. J., O’Brien, P., and Hudson, L. G. (1998) Epider-mal growth factor (EGF)- and scatter factor/hepatocyte growthfactor (SF/HGF) -mediated keratinocyte migration is coinci-dent with induction of matrix metalloproteinase (MMP)-9.J. Cell. Physiol. 176, 255–265

19. Matsumoto, K., and Nakamura, T. (1991) Hepatocyte growthfactor: molecular structure and implications for a central role inliver regeneration. J Gastroenterol. Hepatol. 6, 509–519

20. Jindo, T., Tsuboi, R., Imai, R., Takamori, K., Rubin, J. S., andOgawa, H. (1994) Hepatocyte growth factor/scatter factor stim-ulates hair growth of mouse vibrissae in organ culture. J. Invest.Dermatol. 103, 306–309

21. Rosen, E. M., Grant, D. S., Kleinman, H. K., Goldberg, I. D.,Bhargava, M. M., Nickoloff, B. J., Kinsella, J. L., and Polverini, P.(1993) Scatter factor (hepatocyte growth factor) is a potentangiogenesis factor in vivo. Symp. Soc. Exp. Biol. 47, 227–234

22. Schmidt, N. O., Westphal, M., Hagel, C., Ergun, S., Stavrou, D.,Rosen, E. M., and Lamszus, K. (1999) Levels of vascular endo-thelial growth factor, hepatocyte growth factor/scatter factorand basic fibroblast growth factor in human gliomas and theirrelation to angiogenesis. Int. J. Cancer 84, 10–18

23. Montesano, R., Soriano, J. V., Malinda, K. M., Ponce, M. L.,Bafico, A., Kleinman, H. K., Bottaro, D. P., and Aaronson, S. A.(1998) Differential effects of hepatocyte growth factor isoformson epithelial and endothelial tubulogenesis. Cell Growth Differ. 9,355–365

24. Jiang, W. G., and Harding, K. G. (1998) Enhancement of woundtissue expansion and angiogenesis by matrix-embedded fibro-blast (dermagraft), a role of hepatocyte growth factor/scatterfactor. Int. J. Mol. Med. 2, 203–210

25. Birchmeier, C., Bladt, F., and Yamaai, T. (1997) The functionsof HGF/SF and its receptor, the c-Met tyrosine kinase, inmammalian development. CIBA Found. Symp. 212, 169–177(discussion: pp. 177–182)

26. Peluso, J. J. (1997) Putative mechanism through which N-cadherin-mediated cell contact maintains calcium homeostasisand thereby prevents ovarian cells from undergoing apoptosis.Biochem. Pharmacol. 54, 847–853

27. Shimaoka, S., Tsuboi, R., Jindo, T., Imai, R., Takamori, K.,Rubin, J. S., and Ogawa, H. (1995) Hepatocyte growth factor/scatter factor expressed in follicular papilla cells stimulateshuman hair growth in vitro. J. Cell. Physiol. 165, 333–338

28. Jindo, T., Tsuboi, R., Takamori, K., and Ogawa, H. (1998) Localinjection of hepatocyte growth factor/scatter factor (HGF/SF)alters cyclic growth of murine hair follicles. J. Invest. Dermatol.110, 338–342

29. Straile, W. Z., Chase, H. B., and Arsenault, C. (1961) Growthand differentiation of hair follicles between activity and ques-cence. J. Exp. Zool. 148, 205–222

30. Paus, R., Stenn, K. S., and Link, R. E. (1989) The induction ofanagen hair growth in telogen mouse skin by cyclosporine Aadministration. Lab. Invest. 60, 365–369

31. Paus, R., Stenn, K. S., and Link, R. E. (1990) Telogen skincontains an inhibitor of hair growth. Br. J. Dermatol. 122,777–784


32. Paus, R., Handjiski, B., Czarnetzki, B. M., and Eichmuller, S.(1994) A murine model for inducing and manipulating hairfollicle regression (catagen): effects of dexamethasone andcyclosporin A. J. Invest. Dermatol. 103, 143–147

33. Paus, R., Foitzik, K., Welker, P., Bulfone-Paus, S., and Eichmul-ler, S. (1997) Transforming growth factor-beta receptor type Iand type II expression during murine hair follicle developmentand cycling. J. Invest. Dermatol. 109, 518–526

34. Paus, R., Luftl, M., and Czarnetzki, B. M. (1994) Nerve growthfactor modulates keratinocyte proliferation in murine skinorgan culture. Br. J. Dermatol. 130, 174–180

35. Botchkarev, V. A., Botchkareva, N. V., Welker, P., Metz, M.,Lewin, G. R., Subramaniam, A., Bulfone-Paus, S., Hagen, E.,Braun, A., and Lommatzsch, M. (1999) A new role for neuro-trophins: involvement of brain-derived neurotrophic factor andneurotrophin-4 in hair cycle control. FASEB J. 13, 395–410

36. Jakubczak, J. L., LaRochelle, W. J., and Merlino, G. (1998) NK1,a natural splice variant of hepatocyte growth factor/scatterfactor, is a partial agonist in vivo. Mol. Cell. Biol. 18, 1275–1283

37. Vielkind, U., Sebzda, M. K., Gibson, I. R., and Hardy, M. H.(1995) Dynamics of Merkel cell patterns in developing hairfollicles in the dorsal skin of mice, demonstrated by a monoclo-nal antibody to mouse keratin 8. Acta Anat. 152, 93–109

38. Paus, R., Muller-Rover, S., van der Veen, C., Maurer, M.,Eichmuller, S., Ling, G., Hofmann, U., Foitzik, K., Mecklenburg,L., and Handjiski, B. (1999) A comprehensive guide for therecognition and classification of distinct stages of hair folliclemorphogenesis. J. Invest. Dermatol. 113, 523–532

39. Paus, R., Hofmann, U., Eichmuller, S., and Czarnetzki, B. M.(1994) Distribution and changing density of gamma-delta Tcells in murine skin during the induced hair cycle. Br. J.Dermatol. 130, 281–289

40. Schmidt, C., Bladt, F., Goedecke, S., Brinkmann, V., Zschiesche,W., Sharpe, M., Gherardi, E., and Birchmeier, C. (1995) Scatterfactor/hepatocyte growth factor is essential for liver develop-ment. Nature (London) 373, 699–702

41. Bladt, F., Riethmacher, D., Isenmann, S., Aguzzi, A., andBirchmeier, C. (1995) Essential role for the c-met receptor inthe migration of myogenic precursor cells into the limb bud.Nature (London) 376, 768–771

42. Jhappan, C., Stahle, C., Harkins, R. N., Fausto, N., Smith, G. H.,and Merlino, G. T. (1990) TGF alpha overexpression in trans-genic mice induces liver neoplasia and abnormal developmentof the mammary gland and pancreas. Cell 61, 1137–1146

43. Takayama, H., La Rochelle, W. J., Anver, M., Bockman, D. E.,and Merlin, G. (1996) Scatter factor/hepatocyte growth factoras a regulator of skeletal muscle and neural crest development.Proc. Natl. Acad. Sci. USA 93, 5866–5871

44. Petho-Schramm, A., Muller, H.-J., and Paus, R. (1996) FGF5 andthe murine hair cycle. Arch. Dermatol. Res. 288, 264–266

45. Welker, P., Foitzik, K., Bulfone-Paus, S., Henz, B. M., and Paus,R. (1997) Hair cycle-dependent changes in the gene expressionand protein content of transforming growth factor beta1 andbeta3 in murine skin. Arch. Dermatol. Res. 289, 554–557

46. Woolf, A. S., Kolatsi-Joannou, M., Hardman, P., Andermarcher,E., Moorby, C., Fine, L.G., Jat, P. S., Noble, M. D., and Gherardi,E. (1995) Roles of hepatocyte growth factor/scatter factor andthe met receptor in the early development of the metanephros.J. Cell Biol. 128, 171–184

47. Botchkarev, V. A., Welker, P., Albers, K. M., Botchkareva, N. V.,Metz, M., Lewin, G. R., Bulfone-Paus, S., Peters, E. M., Lindner,G., and Paus, R. (1998) A new role for neurotrophin-3: involve-ment in the regulation of hair follicle regression (catagen).Am. J. Pathol. 153, 785–799

48. Bulfone-Paus, S., Ungureanu, D., Pohl, T., Lindner, G., Paus, R.,Ruckert, R., Krause, H., and Kunzendorf, U. (1997) Interleukin-15protects from lethal apoptosis in vivo. Nature Med. 3, 1124–1128

49. Li, L., Paus, R., Slominski, A., and Hoffman, R. M. (1992) Skinhistoculture assay for studying the hair cycle. In Vitro Cell. Dev.Biol. 28, 695–698

50. Hansen, L. S., Coggle, J. E., Wells, J., and Charles, M. W. (1984)The influence of the hair cycle on the thickness of mouse skin.Anat. Rec. 210, 569–573

51. Botchkarev, V. A., Botchkarev, N. V., Albers, K. M., van derVeen, C., Lewin, G. R., and Paus, R. (1998) Neurotrophin-3involvement in the regulation of hair follicle morphogenesis.J. Invest. Dermatol. 111, 279–285

52. Cotsarelis, G., Sun, T. T., and Lavker, R. M. (1990) Label-retaining cells reside in the bulge area of pilosebaceous unit:implications for follicular stem cells, hair cycle, and skin carci-nogenesis. Cell 61, 1329–1337

53. Paus, R., Schilli, M. B., Handjiski, B., Menrad, A., Henz, B. M.,and Plonka, P. (1996) Topical calcitriol enhances normal hairregrowth but does not prevent chemotherapy-induced alopeciain mice. Cancer Res. 56, 4438–4443

54. Hardy, M. H. (1992) The secret life of the hair follicle. TrendsGenet. 8, 55–61

55. Sonnenberg, E., Meyer, D., Weidner, K. M., and Birchmeier, C.(1993) Scatter factor/hepatocyte growth factor and its receptor,the c-met tyrosine kinase, can mediate a signal exchangebetween mesenchyme and epithelia during mouse develop-ment. J. Cell Biol. 123, 223–235

56. Chase, H. B. (1954) Growth of the hair. Physiol. Rev. 34, 113–12657. Peus, D., and Pittelkow, M. R. (1996) Growth factors in hair

organ development and the hair growth cycle. Dermatol. Clin. 14,559–572

58. Oro, A. E., and Scott, M. P. (1998) Splitting hairs: dissectingroles of signaling systems in epidermal development. Cell 95,575–578

59. Olivero, M., Rizzo, M., Madeddu, R., Casadio, C., Pennacchietti,S., Nicotra, M. R., Prat, M., Maggi, G., Arena, N., and Natali,P. G. (1996) Overexpression and activation of hepatocytegrowth factor/scatter factor in human non-small-cell lung car-cinomas. Br. J. Cancer 74, 1862–1868

60. Stenn, K.S., Fernandez, L.A., and Tirrell, S.J. (1988) Theangiogenic properties of the rat vibrissa hair follicle associatewith the bulb. J. Invest. Dermatol. 90, 409–411

61. Naldini, L., Tamagnone, L., Vigna, E., Sachs, M., Hartmann, G.,Birchmeier, W., Daikuhara, Y., Tsubouchi, H., Blasi, F., andComoglio, P.M. (1992) Extracellular proteolytic cleavage byurokinase is required for activation of hepatocyte growth fac-tor/scatter factor. EMBO J. 11, 4825–4833

62. Yu, D. W., Yang, T., Sonoda, T., Gaffney, K., Jensen, P. J.,Dooley, T., Ledbetter, S., Freedberg, I. M., Lavker, R., and Sun,T. T. (1995) Message of nexin 1, a serine protease inhibitor, isaccumulated in the follicular papilla during anagen of the haircycle. J. Cell Sci. 108, 3867–3874

63. Paus, R., and Cotsarelis, G. (1999) Hair follicle biology in healthand disease. N. Engl. J. Med. 341, 491–497

64. Yo, Y., Morishita, R., Nakamura, S., Tomita, N., Yamamoto, K.,Moriguchi, A., Matsumoto, K., Nakamura, T., Higaki, J., andOgihara, T. (1998) Potential role of hepatocyte growth factor inthe maintenance of renal structure: anti-apoptotic action ofHGF on epithelial cells. Kidney Int. 54, 1128–1138

65. Conner, E. A., Wirth, P. J., Kiss, A., Santoni-Rugiu, E., andThorgeirsson, S. S. (1997) Growth inhibition and induction ofapoptosis by HGF in transformed rat liver epithelial cells.Biochem. Biophys. Res. Commun. 236, 396–401

66. Santoni-Rugiu, E., Preisegger, K. H., Kiss, A., Audolfsson, T.,Shiota, G., Schmidt, E. V., and Thorgeirsson, S. S. (1996)Inhibition of neoplastic development in the liver by hepatocytegrowth factor in a transgenic mouse model. Proc. Natl. Acad. Sci.USA 93, 9577–9582

67. Metcalfe, A., and Streuli, C. (1997) Epithelial apoptosis. Bioes-says 19, 711–720

68. Longati, P., Albero, D., and Comoglio, P. (1996) Hepatocytegrowth factor is a pleiotropic factor protecting epithelial cellsfrom apoptosis. Cell Death Differ. 3, 23–28

69. Frisch, S. M., and Francis, H. (1994) Disruption of epithelialcell-matrix interactions induces apoptosis. J. Cell Biol. 124,619–626

70. Bardelli, A., Longati, P., Albero, D., Goruppi, S., Schneider, C.,Ponzetto, C., and Comoglio, P. M. (1996) HGF receptor associ-ates with the anti-apoptotic protein BAG-1 and prevents celldeath. EMBO J. 15, 6205–6212

71. Maurer, M., Handjiski, B., and Paus, R. (1997) Hair growthmodulation by topical immunophilin ligands: induction ofanagen, inhibition of massive catagen development, and relativeprotection from chemotherapy induced alopecia. Am. J. Pathol.150, 1433–1441

Received for publication April 30, 1999.Accepted for publication October 6, 1999.


Involvement of hepatocyte growth factor/scatter factor and met receptor signaling in hair follicle...

Documents

Transcript of Involvement of hepatocyte growth factor/scatter factor and met receptor signaling in hair follicle...