Journal of Dermatological Science (2005) 39, 147—154

www.intl.elsevierhealth.com/journals/jods

Hair cycle-specific expression of versican inhuman hair follicles

Tsutomu Soma *, Masahiro Tajima, Jiro Kishimoto

Shiseido Life Science Research Center, 2-12-1 Fukuura, Kanazawa-ku, Yokohama 236-8643, Japan

Received 25 November 2004; received in revised form 22 March 2005; accepted 28 March 2005

KEYWORDSAndrogenic alopecia;Bulge;Cytokeratin15;Proteoglycans;Stem cells

Summary

Background: Versican, a large chondroitin sulfate proteoglycan molecule, is impli-cated in the induction of hair morphogenesis, the initiation of hair regeneration, andthe maintenance of hair growth in mouse species. In contrast, in human hair follicles,the distribution and the roles of versican remains obscure.Objectives: To elucidate the implication of versican in normal human hair growth.Methods: Versican expression was examined by in situ hybridization (mRNA) andimmunohistochemistry (protein).Results: The results clearly showed specific versican gene expression in the dermalpapilla of anagen, which apparently decreased in the dermal papilla of catagen hairfollicles. No specific signal was detectable in telogen hair follicles. Consistent with ISHresults, versican immunoreactivity was extended over the dermal papilla of anagenhair follicles, and again, this staining diminished in the catagen phase of human hairfollicles. Interestingly, versican proteins were deposited outside K15-positive epithe-lial cells in the bulge throughout the hair cycle. Versican immunoreactivity in thedermal papilla was almost lost in vellus-like hair follicles affected by male patternbaldness.Conclusion: Specific expression of versican in the anagen hair follicles suggests itsimportance to maintain the normal growing phase of human as well as mouse.# 2005 Japanese Society for Investigative Dermatology. Elsevier Ireland Ltd. All rightsreserved.

1. Introduction

Mammalian hair follicles have a unique cyclical re-growth stage during their lifetime. Accordingly,evidence emphasizes the importance of epithe-

* Corresponding author. Tel.: +81 45 7887291;fax: +81 45 7887277.

E-mail address: tsutomu.souma@to.shiseido.co.jp (T. Soma).

0923-1811/$30.00 # 2005 Japanese Society for Investigative Dermadoi:10.1016/j.jdermsci.2005.03.010

lial—mesenchymal interaction for hair cyclingbetween follicular stem cells and their specificmesenchymal dermal papilla (DP) cells. Comparedwith the recent progress of follicular stem cellresearch, the characterization of DP cells is virtuallyunknown, and one reason is the lack of a markermolecule, which shows relative specificity in itsgene or protein expression for DP cells. One classof such candidate molecules is proteoglycans.

tology. Elsevier Ireland Ltd. All rights reserved.

148 T. Soma et al.

Proteoglycans, a large family of glycosylated pro-teins, have covalently linked several sulfated glyco-saminoglycans, such as chondroitin sulfate, derm-atan sulfate, and heparan sulfate. The biologicalfunctions of proteoglycans involve development, celladhesion, cell migration and differentiation to mole-cular signalling as well as a large number of watermolecules [1,2]. Clinical patients with abnormalmucopolysaccharide metabolism (Hurler’s syn-drome) have very thick hair and a faster rate of hairgrowth during childhood [3]. Immunohistochemicalstudies with several proteoglycan-specific antibodiesdemonstrated the accumulation of glycosaminogly-can in the DP, particularly for chondroitin sulphateproteoglycan [4].

Versican is a chondroitin sulphate proteoglycaninvolved in matrix assembly and structure, and celladhesion [5]. It is abundant in the dermis duringhuman fetal skin development compared to adultskin. A recent transgenic study using a versican pro-moter implicated its expression in bothmesenchymalcondensation and hair induction during hair cycling[6]. Versican immunoreactivity is localized in the DPof active hair follicles (anagen) in mice and rats, buthas not been elucidated in human species [7—9].Here, we examined the onset of versican geneexpression, and the deposition of versican proteindurung human hair cycle to evaluate the implicationof versican in normal human hair growth.

2. Materials and methods

2.1. Tissues, cells, and antibodies

Human tissue specimens from balding (three males,mean age 35 years) and non-balding (four males, twofemale, mean age 36 years) scalp-skin were obtainedfrom plastic surgery with the informed consent ofdonors. DP cells were isolated from scalp tissues bymicro-dissection, followed by in vitro cultivation.Mouse monoclonal anti-versican antibody (clone 2-B-1) purchased from Seikagaku Corp. (Tokyo, Japan)wasappliedat10 mg/mlforboth immunofluorescenceand immunoperoxidase staining [10].Chickpolyclonalanti-cytokeratin 15 (K15) (provided by Covance

Table 1 PCR primers for the amplification of versican isof

Primer Sequence Positiona

F1 gctgcaaaagagtgtgaaaa 55408—55427F2 tggtgaagaaacaaccagtg 65284—65303F3 ctcatgttcctcccactacc 65331—65350R1 agtggtaacgagatgcttcc 80518—80499R2 tgggcaaagtatttgtagca 96685—96666a GenBank accession number AC026696.

Research Products Inc., Berkeley, CA) was used as amarker of the bulge region at 100-fold dilution.

2.2. RT-PCR

DP cellswere cultured in Dubecco’smodifiedminimalessential medium (Invitrogen, Carlsbad, CA) supple-mented with 10% FBS. Total RNAs were extractedfrom the cultured DP cells with ISOGEN solution(Nippon Gene, Toyama, Japan), followed by thesynthesis offirst-strandcDNAsusingoligo(dT)primersand Superscript II (Invitorgen, Carlsbad, CA). cDNAfragments of the human versican N-terminal-con-served domain (position 502—1160 in GenBank acces-sion number X15998) were amplified by RT-PCR usingtotal RNA from the human DP cells described above.Specific cDNA fragmentsof four versicanvariants (V0,V1, V2, and V3 isoforms) were amplified by RT-PCRusing the primers listed in Table 1. All amplificationswere performed for 35 cycles using the followingconditions: 94 8C for 30 s, 58 8C for 30 s, and 72 8Cfor 1 min with High Fidelity PCR System (Roche Diag-nostics, Indianapolis, IN). For the synthesis of anti-sense RNA probes, the recognition sequence of T7polymerase was added to the anti-sense primers ofeach variant, which were applied by RT-PCR.

2.3. In situ hybridisation

For in situ hybridization (ISH), scalp tissue pieceswere fixed in phosphate-buffered formalin (pH 7.2)for more than 1 week, then embedded in paraffinwax. Digoxigenin (DIG)-labeled anti-sense RNAprobes were prepared with an in vitro transcriptionkit (Roche Diagnostics) using versican cDNA frag-ments containing the recognition sequence of T7polymerase amplified by RT-PCR. DIG-labeled ISHwas performed on 6 mm sections using the VentanaDiscovery HX system (Ventana Japan, Yokohama,Japan) as described previously [11].

2.4. Immunofluorescence

For immunofluorescence staining, human scalp-skinpieces were fixed with 4% paraformaldehyde at 4 8Cfor 12—16 h, and then dehydrated and embedded in

orms

Isoform Primer pair Size (bp)

V0 F2 and R1 538V1 F1 and R1 501V2 F3 and R2 529V3 F3 and R2 529

versican in human hair follicles 149

Fig. 1 RT-PCR analysis of versican isoforms in culturedDP cells. Total RNA of cultured DP cells was applied to RT-PCR analysis using isoform-specific primer pairs (Table).Lane 1, sizemarker (100 bp ladder); lane 2, V0; lane 3, V1;lane 4, V2; lane 5, V3; lane 6, GAPDH.

paraffin wax. Tissue sections of 6 mm were boiledtwice in 10 mM citrate-buffer (pH 6.0) for 5 minusing a microwave oven for antigen relevance. Ver-sican immunoreactivity was visualized with Alexa594-labeled anti-mouse IgG antibody (Invitorgen).Chick anti-human keratin 15 (K15) polyclonal anti-bodies were used for double labeling with versicanimmunofluorescence visualized by FITC-labeledanti-mouse IgG antibody (Nakaraitecqure, Tokyo,Japan). K15 immunoreactivity was monitored withAlexa 594-labeled anti-chick IgG antibody (Invitor-gen). All sections were mounted with a vector shield(Vector Labs, Burlingame, CA) containing 40,6-dia-midino-2-phenylindole (DAPI).

2.5. Immunoperoxidase staining

For immunoperoxidase staining, tissue sectionswere digested with proteinase K at 20 mg/ml for30 min at room temperature (Nakaraitecque). Theanti-mouse staining kit (HISTFINE, Nichirei, Tokyo,Japan) was used according to the manufacturer’sinstructions. Sections were developed with TrueBlue (Kirkegaard and Perry Labs, Gaithersburg,MD) followed by counterstaining with contrast red(Kirkegaard and Perry Labs).

3. Results

3.1. Predominant expression of versicantranscripts in the DP of anagen hair follicles

Human cultured DP cells mainly expressed both V0and V1 isoforms (Fig. 1) at RT-PCR level. To comparethe expression level of versican isoforms in humanhair follicles in vivo, we prepared 6 mm adjacentsections of human scalp-skin. Using ISH analysis on

Fig. 2 In situ hybridization analysis of versican isoforms in thstained with the specific anti-sense RNA probe for each versversican transcripts. (A) Endogenous mRNA expression of V0bottom of anagen hair follicles. Melanin granules were observExpression level of V1 isoform was lower compared to V2 andalmost equal to V3 isoform in the DP cells (D) [scale bars, 50

the adjacent sections, we revealed that human ana-gen hair follicles in vivo showed a predominant geneexpression of V2 and V3 isoforms (Fig. 2C and D) incontrast to in vitro (seeFig. 1). Theexpression level ofthe V1 isoform (Fig. 2B) was equal to or lower than V2and V3. Messenger RNAs of the V0 isoform (Fig. 2A)were not detectable, even in anagen hair follicles.cRNA probes for the N-terminal domain, sharedamong four isoforms,which should detect all versicanisoforms, were applied to evaluate the versican geneexpression during human hair cycles. The highestlevel of expression for versican mRNA was detectedin the DP cells of human hair follicles in the anagenphase (Fig. 3A). Bulb matrix cells also expressedsignificant versican mRNA in anagen hair follicles(large arrows in Fig. 3A). Versican gene expressionin the DP cells was dramatically reduced in the cata-gen phase, butwas not completely lost (Fig. 3B). Cellsin the upper portion of DP were still positive forversican mRNA expression (small arrows in Fig. 3B).Versican transcripts were not detectable in the DP oftelogen hair follicles (Fig. 3C). We could not detectversican transcripts in either the epidermis or thedermis of adult scalp-skin tissues (data not shown).

e DP cells of anagen hair follicles. Adjacent sections wereican isoform. Blue staining indicated a positive signal forisoform was not detected in the onion-shape DP at theed as dark-brown colors in the hair shaft above the DP. (B)V3 isoforms in the DP cells. (C) Signals of V2 isoform weremm].

150 T. Soma et al.

Fig. 3 In situ hybridization analysis of versican during the human hair cycle. Human scalp-skin sections were stainedwith the anti-sense RNA probe for the conserved domain of four versican isoforms. (A) In anagen hair follicles, the DP andbulb matrix (large arrows) showed positive signals for versican mRNA. (B) Very weak staining (small arrows) was observedin the upper portion of the DP in early-mid catagen hair follicles, which were identified by the constriction of bulb matrixcompared to anagen hair follicles. (C) The positive signal was barely seen in telogen hair follicles, which weredistinguished by the fully keratinized club (CL) above the condensed DP [scale bars, 50 mm].

3.2. Specific accumulation of versicanprotein in the DP during the anagen phaseand in the bulge during the hair cycle

During the human hair cycle, versican immunofluor-escence was intense in the DP of anagen hair folli-cles (Fig. 4A). This immunofluorescence observed inthe DP was diminished in the catagen phase(Fig. 4B), consistent with the gene expression pat-tern of versican described above (see Fig. 3B). Inter-estingly, the DP of telogen hair follicles showed faintimmunofluorescence (Fig. 4C) although mRNA wasnot detectable during this phase (see Fig. 2C).Strong versican immunofluorescence was alsoobserved in the connective tissue sheath surround-

Fig. 4 Immunofluorescence staining of versican during theanalyzed by immunofluorescence staining in human hair follicrepresented positive versican immunofluorescence, and bluestained in the lower portion of anagen hair follicles. (B) In lcolumn (EC), the DP was almost negative for versican immunoversican immunofluorescence staining in the telogen phase. Tshowed intense immunofluorescence for versican in telogenfollicles (B) [scale bars, 50 mm].

ing the bulge region in anagen (data not shown),catagen (Fig. 4B), and telogen (Fig. 4C), but not thelower bulb of anagen hair follicles (Fig. 4A).

We performed double-immunofluorescence label-ing combined with anti-K15 antibody and anti-versi-can antibody to confirm versican deposition aroundthe bulge region of hair follicles (Fig. 5). Versicanimmunofluorescence (green colors) was detected inthe connective tissue sheath along with K15-positive(red colors) follicular epithelial cells below thesebaceous gland in anagen hair follicles (Fig. 5A).In K15-positive cells, intense K15 immunofluores-cence was restricted to the outer side adjacent tothe connective tissue sheath (Fig. 5B; higher magni-fication around K15-positive cells).

human hair cycle. Localization of versican proteins wasles in anagen, catagen, and telogen phases. Red stainingstaining indicated nuclei. (A) Only the DP was intenselyate catagen hair follicles characterized by the epithelialfluorescence staining. (C) The DP was weakly positive forhe connective tissue sheath surrounding the bulge regionhair follicles as well as anagen (see Fig. 5 and catagen

versican in human hair follicles 151

Fig. 5 Double immunofluorescence labeling of the bulge region with versican and K15 in anagen hair follicles. The sitesof versican immunofluorescence were compared with the sites of K15 immunofluorescence around the bulge region inhuman anagen hair follicles. (A) Strong versican immunofluorescence (green) in the connective tissue sheath waslocalized outside of K15 (red)-positive cells. Blue staining indicated nuclei [scale bar, 50 mm]. (B) Higher magnificationaround K15-positive cells. Intense K15 immunofluorescence (red) was observed in the outermost portion of K15-positiveORS cells [scale bar, 20 mm].

3.3. Lower versican level in vellus-like hairfollicles of male pattern baldness

To investigate whether versican expression isrelated to male pattern baldness, we examinedboth the gene and protein expression level ofversican in vellus-like hair follicles observed inmale pattern baldness. Versican gene expressionis almost lost in the DP of vellus-like hair follicles inthe anagen phase (Fig. 6A). Versican immunoreac-

tivity was significantly lower in the DP of vellus-likehair follicles (Fig. 6B) compared to terminal hairfollicles (Fig. 4A) although the dermal compo-nents surrounding the vellus-like hair folliclesappeared positive for versican immnostaining.The apparent down-regulation of versican expres-sion was confirmed by more than 15 vellus-like hairfollicles in balding scalp-skin tissues derived fromthree individuals (at least five vellus per eachindividual).

152 T. Soma et al.

Fig. 6 Analysis of versican expression in vellus-like hair follicles affected by androgenic alopecia. (A) Versicantranscripts (blue) were barely detected in the DP (white broken outline) of vellus-like hair follicles, which weredistinguished from normal hair follicles by the miniature DP (B) Versican immunoreactivity (blue green) was detected inthe fibrous dermal tissues surrounding vellus-like hair follicles but not almost lost in the DP (black broken outline). Redstaining indicated nuclei [scale bars, 50 mm].

4. Discussion

In this study, we showed the anagen-specific expres-sion of versican in the DP during human hair cycle,similar to mouse species. In addition, versicanexpression was diminished in the DP of vellus-likehair follicles affected by androgenic alopecia. Theseresults show that versican is required for normalhuman hair growth.

At least four isoforms of the versican gene exist asvariants by alternative splicing [12,13]. Interest-ingly, two isoforms (V2 and V3) are dominant in invivo hair follicles in contrast to cultured DP cells,which harbor V0 and V1 isoforms as their majorforms. The tissue-restricted expression pattern ofversican variants was also observed in other humanadult tissues [14]. The elevation and/or diminutionof the definite variant forms have been shown duringthe developmental stage of the neural crest [15]. Inaddition, overexpression of the V3 variant, whichhas no binding domain of chondroitin sulfate, altersarterial smooth muscle cell adhesion, migration,and proliferation in vitro [14,16]. A distinct expres-sion pattern between in vivo and in vitro may pro-vide valuable information about the hair inductiveability of human DP cells. In mice, the same V3variant was also dominantly expressed in DP cells,

implicating the improvement of hair induction withthis non-chondroitin sulfate proteoglycan. It shouldbe examined using in vivo grafting assays whetherthe overexpression of a definite versican isoform inhuman cultured DP cells can restore their hairinductive ability.

The tissue-restricted distribution of versican var-iants was also confirmed at their protein level usingseveral anti-versican antibodies, whose epitope hadbeen already mapped [14]. As clone 2-B-1 used inthis study is known to bind specifically to the C-terminal globular domain common to all four iso-forms of human versican [10], the immunoreactivitydescribed here showed the entire immunoreactivityfor versican. In mouse and rat hair follicles, versicanimmunoreactivity is very intense and specific in theanagen phase of the mature hair cycle. In this study,human hair follicles also showed specific accumula-tion of versican proteins in the DP in the anagenphase. The apparent diminution of versican immu-noreactivity was observed in catagen hair folliclesidentical to the mouse species [6]. Since the pre-dominant gene expression of versican continueduntil just before catagen entry (Fig. 3B), continuousversican gene expression throughout the anagenphase seems necessary for the maintenance of nor-mal hair growth. In the dermis, no versican tran-

versican in human hair follicles 153

script was detected despite abundant versican IR.These observations are consistent with the expres-sion pattern of the transgene in the dorsal skin ofversican transgenic mice [6], and DP-derived versi-can may serve as source of interfollicular immunor-eactivity of versican.

Recent developments have shown that the stemcells of the follicular epithelium exist in the bulgeregion below the sebaceous gland. K15 is preferen-tially expressed in bulge cells, and its promotercould target hair follicle bulge cells [17,18]. Surpris-ingly, versican protein was deposited outside of K15-positive epithelial cells in the bulge area duringhuman hair cycle. Since versican protein was slightlydeposited in the DP of telogen hair follicles withoutits gene expression, versican protein derived fromthe bulge area may be incorporated into the DPbefore the next anagen entry. Versican depositionaround the bulge also correlated with a previousreport showing the relationship between versicanand innervation [19], which demonstrated that ver-sican was associated with nerve fibres during thedevelopment of rat vibrissa hair follicles.

Although proteoglycans are involved in hair mor-phogenesis and hair cycling [4,6], the correlationbetween hair follicle-related disorders and proteo-glycans, including versican, remains unclear. Forexample, the abnormal accumulation of glycosami-noglycans leads to thick hairwith a faster growth rateduring childhood in Mucopolysaccharidoses (Hurler’ssyndrome) [3]. Here, we first showed the low expres-sion of versican in vellus-like hair follicles affected byandrogenic alopecia (Fig. 6). This lower expression ofversican in androgenic alopecia is considerably regu-lated at its transcriptional level since mRNA is notdetectable in the dermal papilla of vellus-like hairfollicles in the anagen stage (Fig. 6A). Although theminimal promoter region of the human versican genewas reported by Naso et al. [20], no functional bind-ing site of the androgen receptor has been identifiedin this promoter sequence. Some potent androgenresponsiveelementsmaybe located in the regulatorysequence of the versican gene.

Acknowledgment

The authors would like to thank Dr Tetsuo Ezaki forhis cooperation in obtaining materials.

References

[1] Jones FS, Jones PL. The tenascin family of ECM glycopro-teins: structure, function, and regulation during embryonic

development and tissue remodelling. Dev Dyn 2000;218:235—59.

[2] Perrimon N, Bernfield M. Specificities of heparan sulphateproteoglycans in developmental processes. Nature2000;404:725—8.

[3] Muenzer J. Mucopolysaccharidoses. Adv Pediatr 1986;33:269—302.

[4] Westgate GE, Messenger AG, Watson LP, Gibson WT. Distri-bution of proteoglycans during the hair growth cycle inhuman skin. J Invest Dermatol 1991;96:191—5.

[5] Shinomura T, Nishida Y, Ito K, Kimata K. cDNA cloning of PG-M, a large chondroitin sulfate proteoglycan expressed duringchondrogenesis in chick limb buds. Alternative splicedmulti-forms of PG-M and their relationships to versican. J BiolChem 1993;268:14461—9.

[6] Kishimoto J, Ehama R, Wu L, Jiang S, Jiang N, BurgesonRE. Selective activation of the versican promoter byepithelial—mesenchymal interactions during hair folli-cle development. Proc Natl Acad Sci USA 1999;96:7336—41.

[7] du Cros DL, LeBaron RG, Couchman JR. Association of versi-can with dermal matrices and its potential role in hairfollicle development and cycling. J Invest Dermatol1995;105:426—31.

[8] Harris SJ, Jahoda CA. A correlation between versican andneurofilament expression patterns during the developmentand adult cycling of rat vibrissa follicles. Mech Dev2001;101:227—31.

[9] Sorrell JM, Carrino DA, Baber MA, Caplan AI. Versican inhuman fetal skin development. Anat Embryol 1999;199:45—56.

[10] Isogai Z, Shinomura T, Yamakawa N, Takeuchi J, Tsuji T,Heinegard D, et al. 2-B-1 antigen characteristicallyexpressed on extracellular matrices of human malignanttumors is a large chondroitin sulfate proteoglycan, PG-M/versican. Cancer Res 1996;56:3902—8.

[11] Nitta H, Kishimoto J, Grogan TM. Application of automatedmRNA in situ hybridization for formalin-fixed, paraffin-embedded mouse skin sections: effects of heat and enzymepretreatment on mRNA signal detection. Appl Immunohis-tochem Mol Morphol 2003;11:183—7.

[12] Dours-Zimmermann MT, Zimmermann DR. A novel glycosa-minoglycan attachment domain identified in two alternativesplice variants of human versican. J Biol Chem1994;269:32992—8.

[13] Zako M, Shinomura T, Ujita M, Ito K, Kimata K. Expression ofPG-M(V3), an alternatively spliced form of PG-M without achondroitin sulfate attachment in region in mouse andhuman tissues. J Biol Chem 1995;270:3914—8.

[14] Cattaruzza S, Schiappacassi M, Ljungberg-Rose A, SpessottoP, Perissinotto D, Morgelin M, et al. Distribution of PG-M/versican variants in human tissues and de novo expression ofisoform V3 upon endothelial cell activation, migration,and neoangiogenesis in vitro. J Biol Chem 2002;277:47626—35.

[15] Perissinotto D, Iacopetti P, Bellina I, Doliana R, Colombatti A,Pettway Z, et al. Avian neural crest cell migration isdiversely regulated by the two major hyaluronan-bindingproteoglycans PG-M/versican and aggrecan. Development2000;127:2823—42.

[16] Lemire JM, Merrilees MJ, Braun KR, Wight TN. Overexpres-sion of the V3 variant of versican alters arterial smoothmuscle cell adhesion, migration, and proliferation in vitro. JCell Physiol 2002;190:38—45.

[17] Liu Y, Lyle S, Yang Z, Cotsarelis GS. Keratin 15 promotertargets putative epithelial stem cells in the hair folliclebulge. J Invest Dermatol 2003;121:963—8.

154 T. Soma et al.

[18] Lyle S, Christofidou-Solomidou M, Liu Y, Elder DE, Albelda S,Cotsarelis G. Human hair follicle bulge cells are biochemi-cally distinct and possess an epithelial stem cell phenotype.J Invest Dermatol Symp Proc 1999;4:296—301.

[19] Harris SJ, Jahoda CA. A correlation between versican andneurofilament expression patterns during the development

and adult cycling of rat vibrissa follicles. Mech Dev2001;101:227—31.

[20] Naso MF, Zimmermann DR, Iozzo RV. Characterization of thecomplete genomic structure of the human versican gene andfunctional analysis of its promoter. J Biol Chem 1994;269:32999—3008.