Characterization of Hair Follicle Bulge in Human Fetal Skin: The Human Fetal Bulge Is a Pool of Undifferentiated Keratinocytes

Characterization of Hair Follicle Bulge in Human Fetal Skin: The Human Fetal Bulge Is a Pool of XJndifferentiated Keratinocytes ]Vlasashi Akiyama,*t Beverly A. Dale,t$§ Tung-Tien Sun,** and Karen A. Holbrook*,t4 Departments of *Biological Structure, tMedicine (Dermatology), tOral Biology and §Periodontics, University of Washington School o f JVledicine, Seattle, Washington; '"Epithelial Biology Unit, Ronald O. Perelman Department of Dermatology and Department of Phamiacology, Kaplan Comprehensive Cancer Center, New York University School of Medicine, New York, New York; and t^VJniversity of Florida, Gainesville, Florida, U.S.A.

I t has been suggested that the bulge of the hair follicle contains a pool of follicular stem cells that m a y serve as a target site of graft-versus-host disease a n d as a source of cells with carcinogenic potential. X h e bulge is prominent in the developing follicle although it is a subtle swelling in the adult follicle. In t h i s paper, we studied the bulge in human fetal skin specimens. Ultrastructurally, the bulge cells, especially the interior cells, have abundant free ribosomes a n d glycogen particles, but almost no cytoplasmic organelles indicative of differentiation. Immunostaining with several specific anti-keratin antibodies demonstrated that the bulge cells express keratins of b o t h stratified and simple epithelia. Melanocytes and

Merkel cells, defined by immunohistochemical and ultrastructural criteria, are seen among bulge cells. Laser confocal microscopy revealed that primitive smooth muscle cells attached directly to the bulge initially at the mid-bulbous hair peg, the stage when the bulge is most prominent. K-laminin and type VII collagen are strongly expressed in the dermoepidermal junction of the bulge and between the matrix area of the bulb and the dermal papilla. Thus, the bulge ofhuman hair follicle is not only an attachment site for arrector pili muscle, but also a pool of keratinocytes that are relatively undiiferentiated. Key

T

btilge cells, relatively little is known about the bulge ofhuman hair folhcles. In the adult follicle, the btilge is an inconspicuous, subtle swelling but it is prominent in the bulbous hair peg of the mid-gestational fettis. In view of the potential importance of the bulge in human follicular and epidermal diseases, we studied the strticture of the bulge area of htiman fetal follicles tising election microscopy, immunohistochemistry, and confocal laser scanning microscopy. In these studies, we paid special attention to the basement membrane zone ofhuman fetal skin because it is believed to play an important role in regulating the growth and differentiation of keratinocytes [7-11]. O u r results clearly indicate that the bulge area contains several subpopulations of cells with different states of differentiation, as defined by their ultrastructural features and their keratin patterns. Tbese data have implications on the localization of follicular stem cells and on the interactions between bulge and arrector pili muscle.

he bulge of the hair follicle is a collection of keratinocytes that is contiguous with the outer root sheath (ORS) [I]. It is located below the opening of the sebaceous duct, and serves as the attachment site of arrector pili muscle. The bulge also marks the distalm o s t extent of the permanent portion of the follicle that persists in all phases of the hair cycle [2]. Although the bulge has been known for more than 100 years, until recently it was rarely mentioned in t h e literature and its function and significance have been obscure. Using tritiated thymidine-labeling, Cotsarelis et al [3] showed t h a t the bulge area of mouse hair follicles contains a population of slow-cycling cells that could be induced to proliferate by 0-tetraclecanoylpliorbol-13-acetate. Based on these findings, the authors suggested that the bulge is the site of follicular stem cells [3]. This finding has provided new explanations for several phenoinena: why t l i e lower follicle was "di.spensable," why the matrix keratinocytes (as the "transient amplifying cells") can only grow for a limited period of time | 3 | , and why damage to the bulge cells plays such an important role in graft-versus-host disease [4 — 6]. Although much h a s been learned about the kinetics and morphology of mouse

words: stem celllkeratinldemwepiderinal jtiiiction/atrector pili muscle. J Invest Dermatol 103:844-850, 1995

MATERIALS AND METHODS Tissue Human fetal skin was ohtained from abortuses of 50—160 d estimated gestational age (EGA). Fetal age was estimated hy crown-rump length, foot length, and menstrual age. Skiu specimens were taken from limbs, trunk, and scalp and processed for iminunohistology, routine histology, and electron microscopy.

Manuscript received March 25, 1995; final revision received July 17, 1995; accepted for puhlication August 3, 1995. Reprint requests to; Dr. Masashi Akiyama, Department of Dermatology, ICcio University School of Medicine, 35 Shinanomachi, Shinjuku, Tokyo 1 60, Japan. Abbreviations; cv-SMA, a-smooth muscle actin; EGA, estimated gestational age; RER, rough endoplasniic reticulum.

Antibodies The anti-type IV collagen mouse monoclonal antibody (MoAh) was purchased from Cliemicon International (Teniecula, CA). The anti-type VII collagen MoAh, NP-185, recognizes a domain near the amino-terminus of type VII collagen. The anti—K-laminin MoAh BM 165 reacts witli the 165-kD chain of K-laminin |12|. Both the anti-typo VII

0022-202X/95/S09.50 • SSDK1022-202X(95)00384-3 • Copyright (O 1995 hy The Society for Investigative Dermatology, Inc. 844

HAIR FOLLICLE BULGE IN HUMAN FOTAL SKIN

V O L . lfl.S, NO. 6 DECEMBER

Table I.

Summary of Immunohistochemical Staining with Anti-Keratin Antihodies in Fetal Bulge" Bulge

Kerarin Anribody AEl

AE2 AE3

Molecular Weight (kD)

Catalog'' Number

40 48 50 56.5 56.5 67 52 54 56

19 If, 14 111 10 1 8 7 6

• 58

59 64 65-67 AE13 AEl 4 34 /3 E12 35 J3 HI 1 Anti-40kn C K 18.04 C K 20.3

845

Inner

Iiitcrfollicular Epidermis

Outermost

ORS

Basal Layer Cells

Intermediate Layer Cells

Periderm

.S

4 3 1,2

44

46 51 58 58 56 52 40 45 46

5 6 8 19 18 20

, moderate reaction; +, weakly po.sitive; —, negative. " + + +, stroiif^ly positive; '' Moll etrt/[21]. ^' Basal cells only. ''So,)me cells weakly positive, ells). ' A ftfew individual positive cells (presiiniahly Morkcl cell

collagen antibody and tbe anti-K-laminin antibody were kind gifts from Dr. Robert E. Burgeson, Harvard Medical School, Boston. Tbe mouse monoclonal HMB.-45 antibody nsed in tbe present study was made against a huinan metastatic lesion of pigmented melanoma in an axillary lymph node. It reacts with fetal and neonatal melanocytes as well as melanoma cells f13|. MoAbs AEl, AE2, AE3, AE13, AE14, 34 fi E12, and 3.S p Mil to buman keratins were prepared as described elsewhere |14-18|, A monospecific polyclonal antibody against 40-kD keratin [19,2()| was a generous gift from D r . James lUleinwald, Harvard Medical School, Boston. Molecular weights and catalog numbers |211 of tbe specific keratins recognized by the anti-keratin antibodies are summarized in Table I. Mouse anti-cytokeratin 18 antibody (Inno MAbCK 18.04) (IBL Researcb Products Corp., Cambridge, MA) and anti-cytokeratin 20 antibody (Inno MAbCK 20.3) (IBL Research Products Corp.) were nsed for the detection of Merkel cells [22-24]. Monse monoclonal anti-HLA-DRa antibody (Dako Corp., Santa Barbara, CA), anti-CD la antibody (Dako Corp.), and anti-a-smooth muscle actin (u-SMA) antibody (Sigma Chemical Co., St. Louis, MO) were employed to detect Langerbans cells and smooth nmscle cells, respectively. Immunohistochcmistry Skin specimens were fixed in Canloy's fixative for die immunobistocheniical labeling witb anti-keratin antibodies. Atter fixation, specimens were rinsed in absolute etbanol, rehydrated to 70% ethanol, and processed for paraffin embedding. For the labeling with HMB-45, anti-HLA-DRa antibody, anti-keratin 18 antibody, anti-keratin 20 antibody, and anti-«-SMA antibody, skin specimens were qnick-trozen, cut in tbe cryostat (Tissue-Tek), and fixed in 100% acetone for 8 min. All 5-/xm sections were stained by the avidin-biotin-peroxidase complex metbod |2.S|. Antibody dilutions were 1/1000 to I/.SOOO for the monoclonal anti-keiatin antibodies and control ascites fluid. Antibody dilution for anti-40-kD keratin antibody was 1/1000 and tbe dilntions for anti-keratin 18 antibody and anti-keratin 20 antibody were 1/20. Tbe dilutions for the other antibodies were 1/200. Sections were incnbated hi normal horse (goat [changes for polyclonal antibodies are given in parentheses]) serum and endogenous peroxidase activity was blocked in 0.01% bydrogen pero.xide solution, then incubated in MoAb (polyclonal antibody) for 30 min, biotin-conjugated horse anti-monse IgG and IgM (goat anti-rabbit IgG) for 30 mill, and then aviciin mixed with biotin-eonjugated pero.xidase tor 45 min (Vector Laboratories, Inc., Burlingame, CA). Sections were rinsed extensively with phosphate-buffered saline between incubations. Color was developed by inctibation in freshly prepared substrate solution containing .SO m M Tris-HCI, pH 7.6, 3,3'-dianiinobenzidine-HCl (0.0.S mg/ml), and

0.01 "Ml bydrogen peroxide at room temperature for 10 min. Couuterstaining witb hematoxylin was performed wben it was necessary to identify tbe bulge area. Immunofluorcsccnt Labeling Whole mount samples of skin or 40jum—tbick sections of skin cut in the cryostat were used. Skin specimens, quick-frozen and defrosted before use, were cut into sqnare pieces (1 mm") or 41)-jUin-tbick sections, fixed in 100% cold acetone for 8 min, incnbated in normal horse and goat sernm and tllen incnbated in primary antibody solution (1:100) for 3 h at 37°C, and Hnorescein isothiocyanate-conjugated horse anti-mouse IgG and IgM (Vector Laboratories, Inc.) for 40 min at room temperature. Tissue pieces or sections were washed %vith pbospbateburtered saline extensively between incubations. Counterstaining was done by incubating the samples in 0.1% propidiuni iodide (to demonstrate nuclei) (Sigma Chemical Co.) solution for 40 seconds for tissne pieces and 10 seconds for cryosections. Stained samples were placed dermal side up on a glass slide and covered witb a coverslip in Vectasbield (Vector Laboratories, Inc.) mounting mediuni. Confocal Microscopy Laser scanning confocal microscopy was performed with a Nikon epiHuorescence microscope equipped with an Odyssey system (Noran Instrnnients, Middleton, Wisconsin) coupled to an Image-1 image analysis system (Universal Imaging, Westchester, PA). Tbe images of sections throtigh the bulge were acbieved by taking the images in tbe xy plane (parallel to substratnm) at fi.xed steps of 1, 2, or 5 jLtm in the z direction. Image processing was done by setting the gain and black level on both the computer and photo-multiplier to provide an optimal signal. Data recorded from 256 video frames were summed and the resultant images were subjected to fnrtber contrast enhancement. A three-dimensional analysis was performed by a Personal bis graphics workstation (Silicon Grapiiics, Monntain View, CA). Electron Microscopy Tissue samples were fixed for 2-4 honis in onehalf strengdi Kaniovsky's fi.xative [Karnovsky MJ: J Cell Biol 27:137AI38A, 1965 (abstr)|, then washed in bnffc'r and postfixed in 1% OsO^ in distilled water for an additional lionr. Tbey were then flooded with V%> aqueous uranyl acetate for 1 h to stain ctt hloc. Dehydration was carried ont in a graded series of alcoliols, transferred into propylene oxide, and embedded in Epon |26]. Sections of 1 fim tbickness were cut for light microscopy and stained by tbe metbod of Ricbardson et al [27]. Tbin sections were cut at 70 nm thickness, stained witb iiranyl acetate and lead

846

Tim JCIUKNAL (ir INVnSTlGATlVE DEKMATOLOGY

AKIYAMA ET AL

citrate [28], and viewed in microscope.

JEOL 1200 EXII transmission electron RESULTS

The Human Hair Follicle Bulge Is Prominent in the Bulbous Hair Peg of the Mid-Gestational Fetus Around 80 d EGA, collectiotis of basal epidermal cells begin to bud into the dennis to form bair germs at sites where mesencbynial cells bave aggregated. Tbe bair germ then elotigates into a cord of keratinocytes, called the hair peg, that grows downward into the dermis. The hair peg differentiates into the bulbous bair peg stage at 1 6 to 1 8 weeks EGA. At tbis time a bulb is fortned at the terminus of the follicle, and two bulges emerge from tbe posterior surface of tbe folhcle. The superior bvtlge is the primordium of the sebaceous gland and the inferior swelling is the true "bulge" (Fig 1), First ohserved as a small collection of cells protruditig sligbtly frotn the follicle (Fig Id), the bulge later forms a prominent hemispherical protrusion during tbe mid-bulbous hair peg stage (Fig 1ft), In tbe more differentiated lanugo hair follicle of 19 weeks EGA or older, the bulge becomes smaller, but it remains distinct from tbe rest of the ORS (Fig lc). hi the newborn, the bulge becomes an even less conspicuotts swelling of the ORS, Htiman hair follicles at the btilbous hair peg stage provide particularly convenient material for studying the formatioti and ultrastructural cbaracteristics of bulge cells. The Human Bulge Cells, Both the Interior And the Outermost Cells, Show UndifFerentiated Morphologic Features Using confocal laser scanning microscopy, we examined bulbous hair pegs stained with propidiuni iodide; this technique revealed an outermost layer of bulge cells surrounding a group of smaller, interior cells. A three-dimensional view of the b'ulge reconstructed from the confocal images showed that the bulge wraps all arotnid the hair follicle. This image was confirmed by examining multiple cross-sections through tbe follicle in the region of the bulge. Conventional light microscopy showed that the bulge cotisists of densely packed cells of relatively uniform size. The liticlei of cells located in the interior portion of the bulge (the "interior" cells) were round or oval-shaped and pale, and most had a single nucleolus. Their cytoplasm was less dense than that of other "regular" ORS cells. No tnitoses were observed in these bulge cells, consistent with their "slow-cycling" nature as reported in mice by Cotsarelis et al [3], In the matrix cells of the hair bulb, mitotic figures were frequently seen, Ultrastructurally, the interior

bulge cells bad abundant free ribosomes and glycogeti particles in tbeir cytoplastn, but almost no cytoplasmic organelles indicative ot differentiation, e.g., rough endoplasmic reticulum (R£R) and intermediate filaments (Fig 2). Mitochondria were rarely seen in the interior cells. Small bundles of fine filaments were observed in the peripheral cytoplasm of the cell, but bundles of keratin filaments typical of keratinocytes were not found. Matrix cells of tbe bulb region, like the bulge cells, had almost no cytoplasmic organelles indicative of differentiation althotigh mitochondria were seen more frequetitly in the matrix cells than in the btilge cells. A larger atnount of keratin btmdles and other cytoplasmic orgatielles were seen in the basal cells of the interfoUicular epidermis than in the bulge cells. Completely fortned desmosonies were evident in the bulge cells although the number was smaller when compared witb other ORS and interfoUicular epidermal cells. The outermost bulge cells (tbat are in contact with the basement membrane) were slightly larger than interior cells and cohtmnar-shaped in radial order. They had cytoplastnic vacuoles, sparse RER, small bundles of intermediate filaments, and an elongated, oval nucleus. The cytoplasmic otganelles were located primarily at the peripheiy ot the cell adjacent to the basement membrane side. The cytoplasm tn this region contained a microfilanientous network. The Basement Membrane Zone Is Almost Completely Formed in the Bulge Region The basement membrane surrounding all of tbe follicle was contiiiuotis. Immature heniidesniosomes joined btilge cells witb basement membrane in the same manner as seen in the ORS, but aiicboring fibrils were not apparent in the bulge area. In the interfollictilar epidermis, dermoepidermal junction was formed cotnpletely. Antibodies to type IV collagen stained uniformly the basement membrane of the epidermis and the entire follicle (Fig 3ii). The K-laminin immunoreactivity was strikingly different, however, in that although it was positive in the basement membrane of the interfoUicular epidermis and the upper hair follicle including the bulge area, but diminished at the lower margin of the bulge and completely negative in the hair bulb. K-laniiniti immtinoreactivity was again positive in the basement membrane between the tnatrix and the dermal papilla (Fig 3ft). Type VII collagen, like K-laininin, was present in the interfoUicular epidermis and the tipper hair folhcle including the btilge area, bttt reduced below the bulge (Fig 3c). The basement membrane between the matrix and the dermal papilla was also positive with type VII collagen antibody as with anti-K-laminin.

HAIR FOLLICLE BULGE IN HUMAN FETAL SKIN

VOL. 105, NO. 6 DECEMBER

847

F i g u r e 2. The bulge cell shows undifferentiated niorphologic features (125 d EGA), rr) An electron micrograph ofthe "interior" bulge cells sbows t h e bulge consists of densely packed, uniform cells that are undifferentiated in morphology, h) An "hiterior" bulge cell bas only free libosomes and glycogen particles in the cytoplasm. Bars, a, 2 /Mm; h, i /am.

Melanocytes And Merkel Cells Are Seen in the Bulge Melanocytes were detected imnumoliistochemically in the human fetal bnlge tising the HiVlB-45 antibody. Neither Langerhans cells (as defined hy HLA-DR and CDla) nor keratin-1 8-positive cells w e r e detected in the bnlge. In the peripheral part ofthe bulge ofthe 18—20 weeks EGA skin, anti-keratin 20 antihody detected positively labeled cells. Ultrastructurally, these cells contained specific granviles identifying them as Merkel cells. Bulge Cells Contain Keratins of Both Stratified And Simple Epithelia The results of keratin expression studies are summarized in Tables I and II. All the bulge cells were strongly irnmuno-positive when reacted with the AEl antibody that detects keratins 19, 16, 14, and 10 (Fig 4a). Cells of the ORS and

sebaceous gland were also positive with AEl. No staining was seen in the hulge with the AE2 antibody that reacts with keratins 1 and 10 (Fig 46). The bulge cells were intensely positive with AE3 antibody, which recognizes all basic keratitis (Fig 4c). AE13 antibody, which reacts with type I hair keratins, stained only the cortex and cuticle of the hair shaft; the hulge cells were negative (Fig 4d). Both 34 j3 E12 antibody for keratins 5 and 6, and AE14 which recognizes keratin 5, stained all the bulge cells positively; the outermost cells reacted with the greatest intetisity (Figs 4eJ). The 35 p H l l antibody detected keratin 8 and immunostained both the interior and the outermost cells of the bulge (Fig 4(j). The anti-4()-kD keratin (keratin 19) antibody selectively stained the bulge cells; the ORS and sebaceous gland were entirely negative

'\ f

r

t

J

V

F i g u r e 3. Major basement membrane components are expressed in the bulge region (108 d EGA), a) Type IV collagen is positive in the basement mei-nhrane zone of the entire follicle including the bulge {mivivlicads). It) K-laminin is positive in the bulge area (anmvlieails). The staining is rednced jnst helow the bulge. <) Type VII collagen immunoreactivity is seen in the bulge area {airowUcaiU), but diminished below it. (Tbe dennal papilla is not included in tbis section.) havs, ci, h, c, 80 /xm.

848

THE ICIUKNAL OF INVESTIGATIVE DERMATOLOGY

AKIYAMA ET AL

Tahle II.

Keratin Expression of Human Fetal Bulge Cells " Bulge

Keratin Catalog Number'' (Antibody) 1 10 5 14 8 19

(AE2) (AE2) (AEl 4, 34 (i El 2) (AEl) (35 /3 H l l ) (Anti-40kD)

Molecular Weight - (kD)

Inner

Outermost

ORS (hasal cells)

InterfoUicular Basal Cells

67

56.5 58

50 52 40

' + + , strongly positive; +, positive; - , negative. ' Moll et al |211.

(Fig 4/i). Matrix cells of the bnlb region were weakly immunopositive witb AE3 antibody. No staining was seen in tbe matrix cells witb tbe other anti-keratin antibodies. These data indicate that bnlge cells contain keratins of both stratified and simple epithelia, and that their keratin composition is distingnishable histocheniically from that of trne ORS cells. Arrector Pili Muscle Is Attached to the Bulge Forming Hemidesmosome-Like Structures After Twenty Weeks EGA After 17 weeks EGA, cells containing a-SMA were seen assembled into streak-like strnctures in snperficial dermis, directed toward the bnlge area ofthe follicles. These cells were most likely primitive arrector pili smooth muscle cells. Conventional light microscopy revealed that these spindle cells, with dark cytoplasm and pale vacuoles aronnd oval or round nuclei, were attached to

each other. From 18 to 20 weeks EGA, the a-SMA-positive smooth muscle cells were observed near and around the bulge. Whole mount skin samples at 18 weeks EGA examined with confocal microscopy showed that the a-SMA-positive cells do tiot contact the bulge btit terminate in the dermis in the vicinity ofthe bulge. After 20 weeks EGA, however, poles of a-SMA-positive, spindle cells were seen attached directly to the bulge surface. Electron microscopy confirmed the identification ofthe smooth muscle cells at 18 weeks EGA, on the basis of extensive RER and abundant niyoftlanients in the cytoplasm and pinocytotic vesicles at the periphery of the cells. The nuclei were round or oval and located centrally in the cytoplasm. The basement membrane of the follicle was continuous atid unaltered at the site where the arrector pili muscle cells attached to

^rf*. i

J

i'i

f Figure 4. The bulge is hnmuno-reactive with AEl, AE3, AE14, 34 /3 E12, 35 fi H l l (anti-K8), and anti-40kD (anti-K19), but non-reactive with AE2 (anti-Kl, KIO) and AE13 (anti-hair keratin) (17-22 weeks EGA), n) AEl, /)) AE2, f) AE3,
VOL.

105, NO. 6 DECEMBEH. 1995

Figure 5. The basement membrane and hemidesinosome-like structures are seen in the primary attachment site ofthe arrector pili muscle (APM) to the bulge (140 d EGA). The basement membrane (^arrou'heaAs) is continnously observed in the attachment site. Bnlge cells h a v e hemidesmost)me-like strtictures t^ciiioivs'). Bar, 0.5 jLtm.

t h e surfiice of the bulge at 20 weeks EGA (Fig 5). Hetiiidesmosomc-like structures were observed in the bulge cell at the attachr n e n t site. A small amount of filamentous material anchoring these structures was seen in the cytoplasm of arrector pili muscle cells at t h e attachment site. DISCUSSION T h e Bulge as a Site of Follicular Stem Cells Cotsarelis ct al [3] proposed that the bulge cells are the stein cells ofthe hair follicle in mice. This hypothesis is entirely consistent with the previous w o r k of Montagna and Chase [29], who found that a complete hair b u l b can regenerate frotn cells in the ORS when the hair matrix of t h e bulb is destroyed by X-irradiation. Based on this observation, Montagna and Parakkal |3()| suggested that the germinative source o f t h e hair follicle resides in the ORS and not in the bulb. Oliver [31-33] and Ibrahitii and Wright [34] showed that the lower half of rat (vibrissa) hair follicle could be surgically removed, atid regeneration of a new hair bulb could be itiduced by the implantation of a n e w dermal papilla. Similarly, surgical retiioval of human axillary h a i r follicles up to a level "near the sebaceous gland" did not prevent the regeneration of new hair follicles [35]. These observat i o n s support the idea that keratinocytes of the lower follicle are dispensable, i.e., stem cells are not in the hulh. Although it has heen suggested that human follicles without a bulge can regenerate [35], this was based on a relatively gross surgical shaving experitiient. •Whether this surgery resulted in the precise excision ofthe bulge vk^ithout removing the sebaceous gland remains to be validated [36]. Robins and Breathnach [37[ showed that the fine structure of h u m a n fetal follicular bulge is sitnilar to that of fetal hair germ. Our d a t a revealed that bulge cells in the hutnan fetus have only a small amount of keratin filaments, but contain a large number of free ribosomes. These relatively undifferentiated features are shared hy adult mouse bulge cells [3] and the non-serrated basal cells ofthe d e e p rete ridges of monkey palm epidermis [38,39], and are suggestive that these cells represent the stem cells. Althotigh the matrix cells of the hulh region had the similar undifferentiated features of ultrastructure, mitotic figures were seen frequently in the matrix area and the matrix cells had a larger number of mitochondria in their cytoplasm. Thus, the matrix cells ofthe bulh seemed to represent the transiently amplifying cells. Recent studies have indicated that the e.xpression of specific keratins defines different types or pathways of epithelial differentiation [40]. Moreover, it has been shown that specific acidic and basic keratins frecjuently form the so-called keratin "pairs," as defined by frec]uent co-expression [41]. Our results indicate that the bulge cells express the basal keratins K5 and K14, as well as two simple epithelial keratins, K8 and K19. Using MoAbs, it has been

MAIR FOLLICLE BULGE IN HUMAN FETAL SKIN

849

shown that K19-positive cells are located in the bulge ofthe human hair follicle [42-46], K8 and K19 are generally thought to be markers of simple epithelia. They are present in the basal cells in the interfoUicular epidermis early in development, and disappear at birth as keratinization takes place [42,47]. Bulge cells lack the keratinization-related keratins K1 and Kl 0. The bulge cells, interfoUicular developing basal cells, and the sebaceotis gland share the similar pattern of keratin expression although the sehaceous gland lacks the keratins of simple epithelia. The bulge cells, like the hasal cells of the developing epidermis [42], express keratins of both stratified and simple epithelia. These ultrastructural and keratin expression data clearly indicate that the bulge of human fetal follicles contains a population of relatively primitive keratinocytes. These cells are distinct from other ORS cells, and their cellular features are suggestive of their being the follicular stem cells. Reynolds and Jahoda [48] reported that rat pelage follicle dennal papilla cells induce follicle neogenesis in non-folhcular skin. This finding suggested that the only cells required to induce complete follicle neogenesis in non-follictilar skin are those from the dermal papilla. The dermal papilla cells may play the similar crucial role in inducing follicle neogenesis in human fetal development. In this context, the relation between the bulge cells and the dennal papilla cells is thought to be an interesting, potential area of future investigation. Several skin tumors are thought to originate from the hair follicle [49-51]. Thus it would he of great interest to know whether the bulge is a target for neoplastic transformation, hi this regard, it is interesting to note that a close relationship seems to exist hetween the bulge and basal cell epithelioma [1,52]. The present morphologic and immunohistologic studies revealed that the hulge ofthe fetal follicle contained Merkel cells. Recently, Merkel cells were reported to be abundant in the bulge and suggested to play an inductive and growth-stimulative role for the perifollicular nerve plexus and the developing arrector pili muscle [53,54]. The Merkel cell-nerve complex may have an influence on the development of the bulge. Dermal-Epidermal Junctional Components of the Bulge Recent results indicate that basement menihrane components perform, in addition to structural organization, a variety of biologic functions including the modulation of cell metabolism and differentiation [55]. The hasenient membrane of the interfoUicular epidermis and hair folHcles, for example, is known to play a role in the morphogenesis, differentiation, and proliferation of the keratinocytes [7-11]. In the present study, we have shown that the hasement membrane ofthe bulge area is ultrastructurally similar to that of the interfoUicular epidermis, although the anchoring fibrils are apparently immature. Of great interest is the finding that K-laminin and type VII collagen are present in the hasement membrane of the bulge and hetween matrix cells and the dermal papilla, but are undetectable in the basement lnemhrane zone ofthe ORS located between the bulge and the matrix ofthe hulh. These two portions, the bulge and the bulh matrix adjacent to the dermal papilla, where K-laminin and type VII collagen are expressed strongly, are thought to be the major sites of cell proliferation in the foUicle, i.e., the bulge presumably represents a stem cell pool and the matrix has a population of rapidly proliferating (transient amplifying) cells [3]. Considering the fact that basement membrane components are known to modulate keratinocyte proliferation [7-11,55], the heterogeneity in the hasement membrane zone is certainly interesting and worthy of further investigation. Attachment of Arrector Pili Muscle to the Bulge Although it is well known that the hulge is the attachment site ofthe arrector pili muscle, little was known about when and how the attachment actually occtirs during development. We have shown here that individual, primitive smooth muscle ceUs are first formed in the derinis above the hulge at the bulbous hair peg stage, and that tbey do not make direct contact with the bulge at the mid-bulbous hair peg stage when the bulge is most prominent. In addition, in the

850

THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

AKIYAMA ET AL

early attachment sites of tbe arrector pili muscle and the bulge in t h e late bulbous hair peg stage, we can see only a small area of cell-cell contact via basement membrane. In this contact area, hemidesmosome-like structures, i.e., dense plaque and microfibrils associated witb it, were observed, Tbese structures appear to function as tbe adbesive apparatus between tbe arrector pili muscle a n d tbe bulge. The unique interactions between the bulge and tbe arrector pili muscle could play an important role in maintaiiiing tbe phenotype of tbe bulge cells.

24.

25.

26. 27. 28. 29.

IVe tttank Dr, Lyntie T, Stnitttfor ttetpfut discussiotts and Robert A, Uttdenmod for the preparatiott of pttoto,i;rapits, Tttis worti was supported itl part by gtctttts (HD-i7664, AR-21557) frottt tttc Natiottat Ittstittites of Heattb, Dr, Akiyntna ti'as a rccipiettt of a postdoctornt Jelloit'sttip of ttie Vettara Memorial Fouttdatioti,

30. 31. 32. 33.

REFERENCES Madsen A: Studies on tlie "bulge" (Wulst) in superficial basal cell epithelioinas. Atcb Dertttatol 89:698-708, 1964 2 . I'inkus H: Embryology of hair. In: Montagna W, Ellis RA (eds.). T/ic Iii(}io,{;y of Hair Groiftit, Academic Press, New York, New York, 19.S8, pp \-32 3 . Cotsarelis G, Sun T-T, Lavker RM: Label-retaining cells reside in the bulge area of pilosebaceous unit: implications for follicular stem cells, hair cycle, and skin carcinogenesis. Ceil 61 :l.'i29-13.17, 199tl 4 . Murphy GF, Lavker RM, Whitaker U, Korngold R: Cytotoxic foUiculitis in GVHD. J Cittait I'athol 18:309.-314, 1990 5 . Sale GE, Beauchamp M: ParafoUicular hair bulge in human GVHD: a stem cell-rich primary target. Botte Manoir Tritttspimtt 11:223-225, 1993 6 . Sale GE, Beauchamp MD, Akiyama M: Parafollicular bulges, but not hair bulb keratinocytes, are attacked in graft-versus..host disease of human skin. Bottc Marrott' Trctttsplattt 14:411-413, 1994 7 . Westgate GE, Shaw DA, Harrap GJ, Couchman JR: Immunohistochemical localization of basement membrane components during hair follicle morpho.. genesis. J Ittvcst Dertttatol 82:2.^9-264, 1984 8. Fine j - D , Smith LT, Holbrook KA, Katz SI: The appearance of four basement membrane zone anrigens in developing human fetal skin. J httn-st Dertttatoi 83:66-69, 1984 9. Couchman JR, Gibson WT: Expression of basement membrane components through morphological changes in the hair growth cycle. Deti liiol 1t)8:290298, 1985 1 0 . Guo M, Grinnell F: Basement membrane and human epidermal differentiation in vitro. J Jf/i'M/ Dertttatoi 93:372-378, 1989 1 1 . Messenger AG, Elliott K, Temple A, Randall VA: Expression of basement membrane proteins aud interstitial collagens in dermal papillae of human hair follicles. J htvest Denttatot 96:93-97, 1991 1 2. Rousselle P, Lunstrum GP, Keene DR, Burgeson RE: Kalinin: an epitheliumspecific basement membrane adhesion molecule that is a component of anchoring filaments. J Ceii Biol 114:567-576, 1991 1 3 . Gown AM, Vogei AM, Hoak D, Gough F, McNutt MA: Monoclonal antibodies specific for melanocytic tumors distinguish subpopulations of melanocytes. Atn J Patitoi 123:19.1-203, 1986 14. Gown AM, Vogel AM: Monoclonal antibodies to intermediate filament proteins of htiman cells: unique and cross-reacting antibodies._/ Ceil Bioi 95:414-424, 1982 1 i, Tseng SCG, Jarvinen M, Nelson WG, Huang HW, Woodcock-Mitchell J, Sun T-T: Correlation of specific keratins with different types of epithelial differentiation: monoclonal antibody studies. Ceii 30:361-372, 1982 1 6. Woodcock-Mitchell J, Eichner R, Nelson WG, Sun T-T: Immunolocalization of keratin polypeptides in human epidermis usiug monoclonal antibodies. J Celt iffii( 95:580-588, 1982 17. Dhouailly D, Xu C, Manahe M, Schermer A, Sun T-T: Expression of hair-related keratins in a soft epithelium: subpopulations of human and mouse dorsal tongue keratinocytes express keratin markers for hair-, skin- and esophagealtypes of differentiation. Exp Ceil lies 181:141-158, 1989 18. Moll R, Dhouailly D, Sun T-T: Expression of keratin 5 as a distinctive feature of epithelial and biphasic mesotheliomas: an immunohiitochemical study using monoclonal antibody AE14. Virchows Ardt B Cell I'atttoi 58:129-145, 1989 19. Wu Y, Rheinwald JG: A new small (40Kd) keratin filament protein made in some cultured human squamous cell carcinomas. Cell 25:627-63.5, 1981 20. Wn Y, Parker L, Binder N, Beckett M, SinardJ, Griffiths C, RheinwaldJG: The mesothelial keratins: a new family of cytoskeletal proteins identified in cultured mesothehal cells and nonkeratinizing epithelioma. Cell 31:693-703, 1982 21. Moll R, Franke "WW, Schiller DL, Geiger B, Kreplcr R: The catalog of human cytokeratins: patterns of expression in normal epithelia, tumors and cultured cells. Ce/; 31:11-24, 1982 22. Moll R, Moll I, Franke WW: Identification of Merkel cells in human skin by specific cytokeratin antibodies: changes of cell density and distribution in fetal and adult plantar epidermis. Dijferetttiatiott 28:136-154, 1984 23. Moll R, Schiller DL, Franke WW: Identification of protein IT ofthe intestinal cytoskeleton as a novel type 1 cytokeratin with unusual properties and expression patterns. J Celt Bioi 111:567-580, 1990

34.

1.

35.

36.

37. 38. 39. 40.

41.

SauratJ-H, Didierjean L, Skalli O, Siegenthaler G, Gabbiani G: The intermediate filament proteins of normal epidermal rabbit Merkel cells are cytokeratins. J htvesi Dertttatoi 83:431-435, 1984 Hsu SM, Raine L, Fanger H: Use of avidin-biotin-peroxidase complex (ABC) in immunoperoxidase techniques: a comparison between ABC and unlabeled antibody (PAP) procedures. J Hisocitetti Cytocltettt 29:577-580, 1981 Luft JH: Improvements in epoxy resin embedding methods. J Biopltys Biodtcm C)*i( 9:409-414, 1961 Richardson KC, Jarrett L, Finke EH: Embedding in epoxy resins for ultrathin sectioning in electron microscopy. Staitt Tecitttol 35:313-323, 1960 Reynolds ES: The use of lead citrate at high pH as an electron opaque stain in electron microscopy. J Ceii Biol 17:208-212, 1963 Montagna W, Chase HB: Histology and cytochemistry of human skin: X-irradiation ofthe scalp. Am] Attat 99:415-445, 1956 Montagna W, Parakkal PF: The sttttctnre attd Jttttctiott of skitt, 3rii ed. Academi Press, New York, New York, 1974 Oliver RF: Whisker growth after removal ofthe dermal papilla and lengths of follicle in the hooded rat.J Etttbtyoi Exp Morjthoi 15:331-347, 1966 Oliver RF: Ectopic regeneration of whisker in the hooded rat from implanted length of vibrissa follicle wall J Etlthryol H,xp Morpitoi 17:27-34, 1967 Oliver RF: The experimental indnction of whisker growth in the hooded rat by implantation of dermal papillae. J Htttltr)'ol Uxp Moqtitot 18:43-51, 1967 Ibrahim L, Wright EA: A quantitative study of hair growth using mouse and rat vibrissal follicles. J Etttlttyol Exp Morpitoi 72:209-224, 1982 Inaba M, Anthony J, McKinstry C: Histologic study of the regeneration of axillary hair after removal with subcutaneous tissue shaver. J htt^est Dertitatol 72:224-231, 1979 Oliver RF, Jahoda CAB: The dermal papilla and maintenance of hair growth. In: Rogers GE, Reis PJ, Ward KA, Marshall RC (eds.). Tite Biolo);y of IVool md Hair, Chapman and Hall, London, 1988, pp 51-67 Rohins EJ, Breathnach AS: Fine structure of the human foetal hair follicle at hair-peg and early bulbous-peg stages of development. Jylimr 104:553-569, 1969 Lavker RM, Sun T-T: Heterogeneity in epidennai basal keratinocytes: morphological and functional correlations. Sciettce 215:1239-1241, 1982 Lavker RM, Sun T-T: Epidermal stem cells. J /iimf Dertttatol 81:121S-127S, 1983 Galvin S, Loomis C, Manabe M, Dhouailly D, Sun T-T: The major pathways of keratinocyte differentiation as defined hy keratin expression: an overview. Atit' Dertttatol 4:277-299, 1989 Sun T-T, Eichner R, Schermer A, Cooper D, Nelson WG, Weiss RA: Classification, expression, and possible mechanisms of evolution of maniniaUan epithelial keratins: a unifying model. In: Levine A, Topp W, V,^nde Woude G, Watson JD (eds.). Cattcer Cells, Voi, 1. Tite Tritttsforttted l'itettotype. Cold Spring

42.

43.

44.

45.

46.

47. 48.

49. 50. 51. 52.

53.

54.

55.

Harbor Laboratory', Cold Spritlg Harbor, New York, 1984, pp 169-176 Dale BA, Holbrook KA, Kimball JR, Hoff M, Sun T-T: Expression of epidermal keratins and filaggriu dnring liuman fetal skin development. J Cell liiol 101:1257-1269, 1985 Holbrook KA, Fisher C, Dale BA, Hartley R: Morphogenesis ofthe hair follicle during the ontogeny of human skin. In: Rogers GE, Reis PJ, Ward KA, Marshall RC (eds.). The Bioh^y of IVooi itttd Hair, Chapman and Hall, New York, New York, 1989, pp 1.5-35 Holbrook KA, Smith LT, Kaplan ED, Minami SA, Hebert GP, Underwood RA: Expression of morphogens during human follicle development itt t'ii'O and a model for studying follicle morphogenesis itt t'itro.J Ittvest Dertttatol 101:39S49S, 1993 Stasiak PC, Purkis PE, Leigh IM, Lane EB: Keratin 19: predicted atnino acid sequence and broad tissue distribution suggest it evolved from keratinocyte keratins. _; iiM'MI Dertttatoi 92:707-716, 1989 Narisawa Y, Hashimoto K, Kohda H: Immunohistochemical demonstration of keratin 19 expression in isolated human hair follicles. J Ittvest Denttatol 1(13:191-195, 1994 Sun T-T, Tseng SCG, Huang AJ-W, Cooper D, Schermer A, Lynch MH, Weiss R, Eichner R: Monoclonal antibody studies of mammalian epithelial keratins: a review. Atttt NY Acad Sci 455:3(17-329, 1985 Reynolds AJ, Jahoda CAB: Cultured dennal papilla cells induce follicle formation and hair growth by transdifferentiation of an adult epidermis. Developtttettt 115:587-593, 1992 Berenblum I, Haran-Ghera N, Trainin N: An experimental analysis of the hair cycle effect in mouse skin carcinogenesis. brj Cattcer 12:402-413, 1958 Ghadially FN: The role of the hair follicle in the origin and evolution of some cutaneous neoplasms of man and experimental animals. Cattcer 14:801-816, 1961 Argyris TS: Hair growth cycles and skin neoplasia. Natl Cattcer lust Motto^rapit 10:33-43, 1963 Leshin B, White WL: Folliculocentric hasaloid proliferation. The bulge (der Wulst) revisited. Atcit Dertttatol 126:900-906, 1990 (erratum appears in Atcit Dertttatoi 126:1598, 1990) Narisawa Y, Hashimoto K, Nakaniura Y, Kohda H: A high concentration of Merkel cells in the bnlge prior to the attachment ofthe arrector pili muscle and the formation of the perifollicular nerve plexus in human fetal skin. Arch Dertttatoi Res 285:261-268, 1993 Narisawa Y, Hashimoto K, Kohda H: Epithelial skirt and bulge of human facial vellus hair follicles and associated Merkel cell-nerve complex. Arch Dcrttiatol 7