INTRODUCTION

The outer surface of the body is covered by epithelial tissuesand structures that develop from the embryonic ectoderm. Theepidermis, a stratified squamous epithelium, forms a protectivebarrier while the major adnexal structures of the skin, the hairand the sweat glands, provide the important function ofregulating body temperature. Much has been learned about thebiology of the epidermis and of the hair, including the kineticsof their renewal, factors that can modulate their growth, theirpattern of cell division under normal and wound responseconditions, and the major structural proteins responsible fortheir differentiated functions. The epidermis and the hair havebeen amenable to study in detail because these structures areso large and accessible, easy to examine biochemically andhistologically, and because these tissues continually renew. Inthe case of the epidermis, the keratinocyte, which is the celltype that forms this tissue, can be grown in culture, permitting

cell biologic, biochemical, and molecular genetic analysis ofthe mechanisms regulating its growth and differentiation.

Much less is known about the regulation of growth, tissuemorphogenesis, and differentiation-related gene expression ofthe sweat glands. There are two types of sweat glands, eccrineand apocrine. Several million eccrine sweat glands having anestimated aggregate mass of about 100 grams are distributedover the body surface (reviewed by Sato, 1993). These glandsare comprised of a coiled tube, lying at the border of the dermisand subcutaneous fat, connected to the surface by a duct thatpenetrates the epidermis to permit expulsion of the gland’ssecretory products. The sweat glands can lower bodytemperature by producing a film of water on the surface of theskin that yields evaporative cooling. The secretory portion ofthe gland, at the distal end of the coil, transports an isotonicplasma filtrate into its lumen. A region of the gland fartheralong the coil, as it connects to the duct, contains absorptivecells which selectively remove NaCl such that the sweat is

1925Journal of Cell Science 112, 1925-1936 (1999)Printed in Great Britain © The Company of Biologists Limited 1999JCS4638

We have characterized precisely the cytokeratin expressionpattern of sweat gland myoepithelial cells and haveidentified conditions for propagating this cell type andmodulating its differentiation in culture. Rare, unstratifiedepithelioid colonies were identified in cultures initiatedfrom several specimens of full-thickness human skin. Thesecells divided rapidly in medium containing serum,epidermal growth factor (EGF), and hydrocortisone, andmaintained a closely packed, epithelioid morphology whenco-cultured with 3T3 feeder cells. Immunocytochemicaland immunoblot analysis disclosed that the cells differedfrom keratinocytes in that they were E-cadherin-negative,vimentin-positive, and expressed an unusual set ofcytokeratins, K5, K7, K14, and K17. When subculturedwithout feeder cells, they converted reversibly to a spindlemorphology and ceased K5 and K14 expression. Underthese conditions, EGF deprivation induced flattening,growth arrest, and expression of α-smooth muscle actin (α-sma). Coexpression of keratins and α-sma is a hallmark of

myoepithelial cells, a constituent of secretory glands.Immunostaining of skin sections revealed that only sweatgland myoepithelial cells expressed the same pattern ofkeratins and α-sma and lack of E-cadherin as the cell typewe had cultured. Interestingly, our immunocytochemicalanalysis of ndk, a skin-derived cell line of uncertainidentity, suggests that this line is of myoepithelial origin.Earlier immunohistochemical studies by others had foundmyoepithelial cells to be K7-negative. We tested five K7-specific antibodies that can recognize this protein inwestern blots and in the assembled keratin filaments ofmesothelial cells. Three of these antibodies did notrecognize the K7 present in myoepithelial cell filaments orin HeLa cell filaments, indicating that some K7 epitopes aremasked when K7 pairs with K17 instead of with its usualkeratin filament partner, K19.

Key words: Cultured cell, Eccrine, Apocrine, Keratin, Actin

SUMMARY

Human sweat gland myoepithelial cells express a unique set of cytokeratins

and reveal the potential for alternative epithelial and mesenchymal

differentiation states in culture

Margarete Schön1,*, Jennifer Benwood1, Therese O’Connell-Willstaedt2 and James G. Rheinwald1,2,‡

1Division of Dermatology/Department of Medicine, Brigham and Women’s Hospital, and 2Division of Cell Growth and Regulation,Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA 02115, USA*Present address: Department of Dermatology, Heinrich-Heine University, Moorenstrasse 5, 40225 Düsseldorf, Germany‡Author for correspondence (e-mail: jrheinwald@rics.bwh.harvard.edu)

Accepted 9 April; published on WWW 26 May 1999

1926

hypotonic when it reaches the surface. Apocrine sweat glandsform at different times and from a different cell lineage thanthe eccrine glands. Apocrine glands are formed as part of thepilosebaceous unit in several restricted regions, primarily inaxillary and pubic skin, and they first become functionallyactive at puberty. They secrete sweat and other products (Cohn,1994) into the pilosebaceous duct next to the hair shaft, ratherthan directly onto the epidermal surface. Unlike the epidermisand the hair follicle, eccrine and apocrine glands do notregularly renew themselves via cell division and terminaldifferentiation.

Histologic and immunohistochemical studies have revealedthat eccrine glands begin forming at 13 weeks of gestation fromclusters of cells that form buds in the embryonic epidermis andmigrate downward as cords into the dermis. Initially, cells inthe gland anlagen appear rather homogenous and express acombination of keratins and vimentin consistent with theiridentity as pluripotent progenitor cells which, as the glandmatures, become restricted to one of three types ofdifferentiation: duct, secretory, or myoepithelial (Holbrook andWolff, 1993; Moll and Moll, 1992). Myoepithelial cells, a celltype defined by its expression of keratin-type intermediatefilaments and α-smooth muscle actin (Franke et al., 1980;Norberg et al., 1992), surround and interdigitate with thesecretory coil cells and are thought to provide tensile supportagainst distention as the gland fills with sweat. These cells mayalso contract slightly in response to neurotransmitters, therebyaiding sweat expulsion (Sato et al., 1979).

Small, short-term cultures of ductal and secretory epithelialcells can be prepared from outgrowths of eccrine glandsisolated by digesting skin specimens with dispase andcollagenase and microdissecting the adnexal structures (seeCollie et al., 1985; Lee et al., 1986; Jones et al., 1988; Yokozekiet al., 1990; Bell and Quinton, 1991). The culture of sweatgland myoepithelial cells has not yet been described, however.Scientists at a local biotechnology company who were seriallyculturing epidermal keratinocytes from skin biopsies to preparegrafts for burn patients (see O’Connor et al., 1981; Leigh et al.,1991) brought to our attention the occasional presence in thesecultures of rare epithelial cell colonies of morphology distinctfrom that of keratinocytes. We report here the characterizationof these as sweat gland myoepithelial cells and theidentification of conditions promoting their rapid division inserial culture. We have found that this cell type expresses aunique pattern of cytokeratin proteins, including coexpressionof the unusual K7/K17 cytokeratin pair, and that modifyingculture conditions induces a reversible transition betweenepithelioid and fibroblastoid morphology and pattern ofcytoskeletal protein expression.

MATERIALS AND METHODS

CellsFull-thickness skin biopsies of undamaged skin obtained from burnpatients (typically from the axilla) were processed by the productionlaboratory at Biosurface Technology, Inc. (Cambridge, MA) as thefirst step toward expanding the keratinocyte population in culture togenerate epidermal autografts (O’Connor et al., 1981; Leigh et al.,1991). The skin biopsies were minced into small fragments and stirredin 0.25% trypsin at 37°C. Released cells were recovered bycentrifugation, resuspended in FAD medium, and co-cultivated with

3T3 feeder cells for ~7 days until cultures were nearly confluent.These primary cultures then were suspended with trypsin/EDTA andcryopreserved. Ampules were then thawed and plated under differentconditions to expand the keratinocyte and myoepithelial cellpopulations.

Pure myoepithelial cell lines were isolated from cultures initiatedfrom three skin biopsies: BRSO from the axilla of a 13-year-old (yo)male, CYHI from the axilla of a 28 yo female, and DOLA from theaxilla of a 19 yo male. Mid-lifespan cultures of CYHI and BRSO werekaryotyped by the Dana-Farber Cancer Institute’s clinicalcytogenetics laboratory, using trypsin-Giemsa banding, and werefound to be diploid female and male, respectively. Myoepithelial cellswere also identified in, but were not isolated as pure populations from,cultures initiated from three other skin biopsies: JAKE from theabdomen of a 31 yo male, BABU from the axilla of a 28 yo female,and C5-SkH(my) from the scalp of a >40 yo individual whose genderwas not recorded. The properties of myoepithelial cells in culture werecompared with those of the normal newborn foreskin keratinocytestrain N (Rheinwald and Beckett, 1980), the normal human foreskinfibroblast strain S1-F, the human adult peritoneal mesothelial cellstrain LP-9 (Wu et al., 1982), and the human skin-derived cell linendk (Adams and Watt, 1988).

Culture conditionsHuman myoepithelial cells and keratinocytes were routinely culturedin ‘FAD medium’, consisting of DME/F12 (Sigma) (1:1 v/v) + 5%newborn calf serum (Hyclone) + 5 µg/ml insulin, 0.4 µg/mlhydrocortisone (HC), 10 ng/ml epidermal growth factor (EGF),1.8×10−4 M adenine, 10−10 M cholera toxin, 2×10−11 Mtriiodothryonine, and penicillin/streptomycin (Allen-Hoffmann andRheinwald, 1984; Rheinwald, 1989). For use as feeder cells, themouse fibroblast cell line 3T3J2 was grown in DME + 10% calf serummedium, either lethally irradiated with 5,000 R gamma-irradiationfrom a cobalt source or treated for 2.5 hours with 3 µg/ml mitomycinC (Sigma), and plated at ~5×104 cells/cm2 (Rheinwald and Green,1975). Alternatively, myoepithelial cells were cultured in the absenceof feeder cells in DME/F12 medium supplemented with 10% serum,hydrocortisone, and EGF. For some experiments, keratinocytes andmyoepithelial cells were grown in Gibco keratinocyte serum-freemedium (ker-sfm) (Life Technologies, Inc., Gaithersburg, MD) (Pirisiet al., 1987) supplemented with 30 µg/ml bovine pituitary extract, 0.1ng/ml EGF, penicillin/streptomycin, and additional CaCl2 to bring thetotal [Ca2+] to 0.4 mM (Schön and Rheinwald, 1996). Ndk cells(Adams and Watt, 1988) at 8th passage (kindly provided by F. Watt.,ICRF, Lincoln’s Inn Fields, London) were cultured in FAD medium.HeLa cells were cultured in DME/F12 medium supplemented with10% calf serum. Cells were subcultured by incubation with 0.1%trypsin/0.02% EDTA in PBS and were cryopreserved in liquidnitrogen as suspensions in DME/F12 medium supplemented with 10%calf serum and 10% dimethylsulfoxide.

To compare growth in different medium formulations and in thepresence or absence of specific hormones, growth factors, or growthinhibitors, cells were plated at 2×103 cells per 9 cm2 well in 6-wellplates in the desired medium, refed every 2 to 4 days, and eithersuspended to count or else fixed and stained with Methylene Blue 8to 10 days after plating. When used, TGF-β1 (R&D Systems) wasadded to a concentration of 1 ng/ml (see Rollins et al., 1989) andphorbol myristyl acetate (PMA) (Sigma) to a concentration of 10−8

M.Organotypic cultures were prepared as described by Parenteau et

al. (1991) and by Schön and Rheinwald (1996). Briefly, 6.9×104

human fibroblasts (newborn foreskin-derived, strain B038) weresuspended in 3 ml of bovine collagen type I (0.7 mg/ml)(Organogenesis, Inc., Canton, MA). This suspension then was cast ontop of 1 ml of an acellular collagen gel in six-well tissue culture trayinserts, the bottoms of which were polycarbonate filters of 3 µm poresize (Organogenesis, Inc.). In some experiments, myoepithelial cells

M. Schön and others

1927Sweat gland myoepithelial cell differentiation

or myoepithelial cells plus fibroblasts were embedded in the collagengels. The cells were allowed to contract the collagen gels for 4 daysat 37°C. Keratinocytes and/or myoepithelial cells then were seededonto the gels at a density of 2×105 per cm2 of gel surface area andwere cultured submerged for 4 days in DME/F12 medium (3:1 v:v)supplemented with 0.3% calf serum, 5 µg/ml insulin, 0.4 µg/mlhydrocortisone, 2×10−11 M triiodothyronine, 5 µg/ml transferrin, 10−4

M ethanolamine, 10−4 M phosphoethanolamine, 5.3×10−8 Mselenious acid, and 1.8×10−4 M adenine. The cultures were then raisedto the air/liquid interface and cultured for 10 more days, after whichthey were embedded in OCT compound for cryosectioning.

Two-dimensional gel electrophoresisUsing methods described by Wu et al. (1982), subconfluent cultureswere metabolically labeled with 100 mCi [35S]methionine for 4 hoursand the Triton/high salt-insoluble, cytoskeletal fraction was isolated.Equal amounts of labeled protein were resolved by non-equilibriumpH gradient electrophoresis (NEpHGE) in the first dimension and bySDS-polyacrylamide gel electrophoresis (SDS-PAGE) in the seconddimension. Labeled proteins were detected by autoradiography usingKodak X-OMAT film.

Antibodies The following mouse monoclonal antibodies were used forimmunoperoxidase staining and western blotting: keratin 5-specificantibody AE14 (Lynch et al., 1986) was kindly provided by T.-T. Sun.Keratin 7-specific antibody OVTL 12/30 (van Niekerk et al., 1991)was purchased from Biodesign International (Kennebunk, ME).Keratin 7-specific antibody Ks7.18 (Bartek et al., 1991), keratin 18-specific antibody CK2 (Debus et al., 1982), and vimentin-specificantibody V9 (Osborn et al., 1984) were purchased from Boehringer-Mannheim Corp. (Indianapolis, IN). Keratin 7-specific monoclonalantibodies RCK105 (Ramaekers et al., 1983) and CK7 (Tölle et al.,1985) were kindly provided by F. Ramaekers and M. Osborn,respectively. Keratin 7-specific antibody LDS-68 (Southgate et al.,1987), keratin 14-specific antibody CKB1 (Caselitz et al., 1986), andsmooth muscle α-actin-specific antibody 1A4 (Skalli et al., 1986)were purchased from Sigma (St Louis, MO). Keratin 17-specificantibody E3 (Troyanovsky et al., 1989) was kindly provided by S. M.Troyanovsky. Keratin 19-specific antibody Ks19.1 (A53-B/A2)(Karsten et al., 1985) was purchased from ICN Biomedicals, Inc. E-cadherin-specific antibody E4.6 (Cepek et al., 1994) was kindlyprovided by M. B. Brenner.

Immunocytochemical and immunohistochemical stainingCultures grown in plastic tissue culture plates or multiwell trays wererinsed briefly in water and fixed for 15 minutes to 4 weeks in cold (−20°C) methanol. They then were air-dried for 15 minutes andincubated for 30 minutes each, separated by 5 minute PBS rinses, with1% normal goat serum, primary mouse monoclonal antibody atpredetermined optimal dilutions, biotinylated goat anti-mouse IgGsecondary antibody, and avidin-biotin peroxidase complex(immunocytochemical detection reagents purchased from VectorLaboratories, Burlingame, CA). The cultures were incubated for afinal 5 minutes with the peroxidase substrate 3-amino-9-ethyl-carbazole, the enzymatic reaction stopped by rinsing and incubatingfor 10 minutes with 10% formalin in PBS. The cultures were thenmounted with a glass coverslip and aqueous mounting medium formicroscopic analysis and photography.

For immunohistochemical staining, tissue samples or organotypiccultures were embedded in OCT compound (Miles Inc., Elkhart, IN)and snap-frozen in liquid nitrogen. Five µm cryostat sections wereacetone-fixed and stained by the ABC-immunoperoxidase method(reagents from Vector Labs) as described above.

Western blotting Equal amounts (10 µg/lane) of Triton/high salt-insoluble protein (the

‘cytoskeletal fraction’) of cultured cells, prepared as described by Wuet al. (1982), were fractionated by SDS-PAGE and electrotransferredto nitrocellulose filter paper. These blots were first incubated for 1hour in a blocking buffer consisting of 50 mM Tris, 200 mM NaCl,0.05% Tween, and 10% nonfat powdered milk (pH 7.5), and then ina PBS solution of primary antibody for 30 minutes at roomtemperature. A peroxidase-conjugated, goat anti-mouse secondaryantibody (Promega, Madison, WI) then was applied and antigen-antibody complexes detected by chemiluminescence using the ECLkit from Amersham, Inc.

Fluorescence-activated cell scanning (FACS) Cells were detached from culture dishes using 0.1% trypsin/0.02%EDTA in PBS, fixed with 1% paraformaldehyde in PBS,permeabilized with 0.3% saponin (Sigma), and rinsed in PBS. Thecells then were incubated in suspension with saturating amounts ofprimary antibody or with isotype-matched, non-immune mouse IgG,followed by FITC-conjugated goat anti-mouse IgG. Finally, cells wererinsed twice in PBS and analyzed in a FACScan apparatus (Becton-Dickinson) using Cell Quest Software (BD ImmunocytometrySystems).

RESULTS

Isolation of morphologically distinctive epithelialcell lines from some human skin-derivedkeratinocyte culturesRare colonies of morphology different from that of typicalepidermal keratinocyte colonies (Fig. 1a,c,d) were apparent insome early passage cultures initiated from cells disaggregatedby trypsin from full-thickness human skin biopsies and platedin the 3T3 feeder layer system (Rheinwald and Green, 1975;Rheinwald, 1989). These ‘variant’ colonies were tightlypacked but unstratified and intercellular spaces appeared widerand more refractile when viewed with phase contrast optics.Immunocytochemical staining disclosed that, unlikekeratinocytes, the variant cells did not express E-cadherin (Fig.1b) or P-cadherin (data not shown).

Pure populations of this morphologically variant cell typewere isolated from early passage mixed (predominantlykeratinocyte) skin cell cultures from three different donors. Inone case, CYHI, keratinocytes became nonproliferative by theend of the third passage, much earlier than usual, therebypermitting the variant cells to prevail and take over thepopulation. From two other donors, BRSO and DOLA,colonies of variant morphology were identified in third passagecultures, isolated with cloning cylinders, and passagedsubsequently as pure populations. The three variant cell lines,named CYHI, BRSO, and DOLA for the initials of theirdonors, were indistinguishable from one anothermorphologically. Their distinctive, non-keratinocyte colonymorphology (Fig. 1c) remained constant during serial passagewith 3T3 fibroblast feeder cells in FAD medium until theysenesced after 45-60 population doublings.

BRSO, DOLA, and CYHI cells, unlike epidermalkeratinocytes, grew just as rapidly (Td~1 day) in FAD mediumwithout fibroblast feeder cells, even from very low densityplatings (e.g. <100 cells/cm2). Under these conditions,however, the cells adopted a very different morphology,quickly losing contact with neighboring cells in growingcolonies and becoming spindly or stellate (Fig. 1e), a‘fibroblastoid’ morphology which, nevertheless, was easily

1928

distinguishable from that of the longer and more slenderhuman dermal fibroblast. Karyotype analysis of BRSO andCYHI disclosed that they were diploid, consistent with theconclusion that they are a normal constituent cell of skin andnot an abnormal keratinocyte variant that had arisen duringgrowth in culture.

The pattern of cytoskeletal protein expression bythe ‘variant’ cell type in culture is consistent withtheir identity as myoepithelial cellsSeeking to identify this novel skin cell type, we used 2-D gelelectrophoresis to analyze the intermediate filament proteinsthe cells express in culture (Fig. 2a-c). DOLA (and BRSO, notshown) expressed keratins K5, K6 (minor), K7, K14, and K16or K17 (the latter two keratins are not resolved in this system)as well as high levels of vimentin. Western blot analysis withkeratin subunit-specific monoclonal antibodies confirmed theexpression by DOLA and BRSO cells of K5, K7, K17, andvimentin (Fig. 3d) and also of K14 (data not shown). Theexpression of K7 and high levels of vimentin by DOLAcells (Fig. 2a) distinguished these cells from epidermal

keratinocytes (Fig. 2b). The expression by DOLA cells of K7in the absence of any other simple epithelial keratindistinguished them from mesothelial cells (Fig. 2c) as well asfrom all other cell types characterized to date (see Wu et al.,1982; Moll et al., 1982). Indeed, coexpression of K7 and K17is very unusual, K19 being the apparent partner for K7 inkeratin filament assembly in all normal epithelial cell typesstudied to date (Moll et al., 1982).

When near-confluent DOLA or BRSO cultures growing inthe absence of 3T3 feeder cells were refed with mediumlacking EGF, the cells soon stopped dividing and adopted amore flattened morphology, some developing prominentcytoplasmic stress fibers. Many of these flat, growth-arrestedcells were found to express smooth muscle α-actin (Figs 1f,2f), reminiscent of the expression of this actin isoform bygrowth-arrested myoepithelial cells cultured from mammarygland (Peterson and van Deurs, 1988; O’Hare et al., 1991).These authors had also reported that mammary glandmyoepithelial cells are dependent on EGF for growth inculture, consistent with the possibility that the cells we hadcultured from skin were myoepithelial cells, which are a

M. Schön and others

Fig. 1. Identification of a rare,morphologically distinct epithelialcell type growing in epidermalkeratinocyte cultures initiated fromfull thickness biopsies of humanskin. (a) Third passage culture ofBABU skin cells growing in FADmedium + 3T3 feeder cells, showingan epidermal keratinocyte colony atleft merging with an unstratifiedepithelial cell colony at right,indicated by arrowhead. The largecells in the upper right corner are3T3 feeder cells. (b) Another regionof the same culture as in aimmunostained for E-cadherin,showing an E-cadherin-negativeepithelial cell colony at right,indicated by arrowhead, bordered byE-cadherin-positive keratinocytes.(c) 6th passage (~35 cumulativepopulation doublings) culture ofDOLA that had been seriallysubcultured in FAD medium + 3T3feeder cells, showing long-termretention of the morphologycharacteristic of DOLA, BRSO, andCYHI cells when grown under theseconditions. (d) Second passageculture of C5-SkH skin cellsgrowing in the same conditions as c,showing a pure keratinocyte colonyof typical morphology bordered by3T3 feeder cells. (e) 8th passageculture of BRSO (~45 cumulativepopulation doublings) that had beengrown for three passages in FADmedium without 3T3 feeder cells,showing the characteristic spindly and stellate morphology that BRSO, DOLA, and CYHI cells adopt when cultured under these conditions.(f) 6th passage DOLA cells immunostained for α-smooth muscle actin after having been cultured for 7 days in DME/F12 mediumsupplemented with serum and HC but without EGF, showing α-sma expression by a majority of the cells, ranging from low to high levels(arrowhead points to one of the several completely negative cells in the field). Bar, 200 µm, for all panels.

1929Sweat gland myoepithelial cell differentiation

constituent of the secretory coils of eccrine and apocrine sweatglands (reviewed by Sato, 1993; Holbrook and Wolff, 1993).

Masking of some K7 epitopes in myoepithelial cellkeratin filaments, correlated with K7/K17coexpressionArguing against a myoepithelial identity, however, were theresults of a comprehensive immunohistochemical analysis of

keratins expressed during human sweat gland development(Moll and Moll, 1992), which had concluded that secretory coilcells express K7, but not K17, and that myoepithelial cellsexpress K17, but not K7. That study employed the K7-specificantibody CK7 (Tölle et al., 1985), an antibody which had beenfound previously not to immunostain mammary glandmyoepithelial cells (Fischer et al., 1987). We sought to resolvethis question by comparing the immunocytochemical stainingcharacteristics of five independently derived, K7-specificmonoclonal antibodies on DOLA and BRSO cells and on twoother cell lines that express K7, the primary human mesothelialcell line LP-9 and HeLa cells. As shown in Fig. 3 andsummarized in Table 1, all five antibodies immunostained thekeratin filaments of mesothelial cells and four out of four ofthis set that were tested by western blotting all recognizedelectrophoretically-separated K7 protein, both from DOLAand from LP-9. Three of the five antibodies, however, Ks7.18(Bartek et al., 1991), CK7 (Tölle et al., 1985), and LDS-68

Fig. 2. Distinctive pattern of structural proteins expressed by thenovel skin-derived cell type. (a-c) Cytoskeletal extracts from DOLAcells (a), epidermal keratinocyte strain N (b), and mesothelial cellstrain LP-9 (c) were analyzed by two-dimensional gelelectrophoresis, with non-equilibrium pH gradient electrophoreticseparation in the horizontal dimension (more basic proteins left,more acidic proteins right) followed by SDS-polyacrylamide gelelectrophoretic separation in the vertical dimension (larger proteinstop, smaller proteins bottom). Asterisks identify the position ofkeratin subunits that were not completely dissociated from theirfilament partners during the initial isoelectric focusing step (see Wuet al., 1982; Moll et al., 1982; Franke et al., 1983). (Keratins K16 andK17 are not resolved in this system, so the presence of K17 must bedetermined by immunologic methods (see d). (d-f) Western blotdetection of structural proteins expressed (the protein specificallyrecognized by the respective antibody in each extract is shown atright). (d) Lane 1: BRSO; lane 2: DOLA; lane 3: epidermalkeratinocyte strain N; lane 4: mesothelial cell strain LP-9. Theantibody used to detect K7 in this experiment was RCK105.(e) Lanes 1 and 2: DOLA; lanes 3 and 4: epidermal keratinocytestrain N. The cells in lanes 1 and 3 were cultured in FAD mediumwith 3T3 feeder cells, and the cells in lanes 2 and 4 were cultured inGibco ker-sfm medium. (f) Lane 1: cultured bovine aortic smoothmuscle cells; lane 2: EGF-deprived culture of BRSO; lane 3: EGF-deprived culture of DOLA; lane 4: epidermal keratinocyte strain N.For each antibody the relative electrophoretic mobility of the mainimmunostained band, compared with co-electrophoresed molecularmass standards, was as expected for the respective protein.

Fig. 3. Keratin K7 in the filaments of intact DOLA cells is notrecognized by some K7-specific antibodies that recognize this keratinin the filaments of mesothelial cells. (a,b) DOLA cells; (c,d) LP-9mesothelial cells. (a and c) Immunostained with K7-specific antibodyOVTL 12/30. (b and d) Immunostained with K7-specific antibodyKs7. 18. (e) Western blot of LP-9 (lane 1) and DOLA (lane 2) proteinextracts, showing that the Ks7. 18 antibody recognizes K7 in theSDS-PAGE fractionated extracts of both cell lines. (Asterisk in eindicates a lower molecular mass polypeptide recognized by theantibody in the LP-9 extract, which may be a K7 proteolyticdegradation product.) Bar, 200 µm, for a-d.

1930

(Southgate et al., 1987), did not immunostain DOLA cells (Fig.3), BRSO cells, or HeLa cells (data not shown). The HeLa cellline is unusual in that it expresses K7, K8, K17, K18, and onlytrace amounts of K19 (Moll et al., 1983). As summarized inTable 1, the results we observed are consistent with theconclusion that the epitope(s) on K7 recognized by many K7-specific monoclonal antibodies is masked in cells in which K7pairs with K17 in filament formation instead of with its morecommon partner, K19.

Identification of K7+/K17+ myoepithelial cells insweat glands of human skin in vivoWe used a K7-specific antibody, OVTL 12/30 (van Niekerk etal., 1991), that could recognize this keratin in the filaments ofDOLA cells to stain cryosections of human scalp skin. Asshown in Fig. 4 and summarized in Table 2, of all the cell typesand structures in skin, only the myoepithelial cell componentof sweat gland secretory coils exhibited the same pattern ofstructural protein expression as DOLA and BRSO cells inculture, namely, K5/K14, K7/K17, vimentin, α-smooth muscleactin, and an absence of E-cadherin. The cells we have culturedare therefore almost certainly normal sweat glandmyoepithelial cells.

M. Schön and others

Table 1. Masking of epitopes recognized by some keratinK7-specific antibodies in keratin filaments of myoepithelial

cells, correlated with expression of K17 as a potentialpairing partner

Mesothelial Myoepithelialcell cell HeLa

Immunostained by K7 antibody:

RCK 105 + + +OVTL 12/30 + + +CK7 + − −KS7.18 + − −LDS-68 + − −

Keratins expressedby this cell type:

K5 − + −K7 + + +K8 + − +K14 − + −K17 − + +K18 + − +K19 + − −, ±

Fig. 4. The pattern of structural protein expression of eccrine sweat gland myoepithelial cells detected immunohistochemically in vivo isidentical to that of DOLA and BRSO cells in culture. Cryostat sections of human scalp skin were immunostained for the following proteins: (aand c) K7, (b and d) K17, (e) K14, (f) vimentin, (g) α-smooth muscle actin, and (h) E-cadherin. (a and b) S indicates a sebaceous gland, Hindicates a hair follicle, and E indicates an eccrine sweat gland. (d-f) Asterisks indicate the luminal, secretory cell layers of eccrine glandsecretory coils. (d,f,g,h) Open arrowheads indicate the ductal portions of eccrine glands. (h) The arrows indicate the outer, unstainedmyoepithelial cell layer of the eccrine gland secretory coil. Bars: 100 µm (a and b); 50 µm (c-h).

1931Sweat gland myoepithelial cell differentiation

Reversible modulation of myoepithelial cellmorphology and keratin expression under differentconditions of cultureThe epithelioid-to-fibroblastoid transition of myoepithelialcells when cultured in the absence of 3T3 feeder cells,described above, proved to be accompanied by marked changesin keratin protein expression (Fig. 5). In the ‘fibroblastoid’state, K5 and K14 content decreased in most cells toundetectable levels, K7 increased, and K17 remained about thesame. We found that this altered morphology and keratinexpression pattern was rapidly reversible; when DOLA orBRSO cells that had been serially passaged for ten doublingsor more in the absence of 3T3 feeder cells were then replatedwith feeder cells, they converted back to an epithelioidmorphology within several days and re-expressed K5 and K14,while reducing their level of K7. This ‘switching’ of keratinexpression was quantitated by FACScan analysis (Fig. 6). Wedo not yet know the mechanism by which the presence of 3T3fibroblast feeder cells exerts this marked phenotypic change inmyoepithelial cells. Preliminary experiments disclosed that co-

cultivation with irradiated human dermal fibroblasts alsoinduced conversion of morphologically fibroblastoidmyoepithelial cells back to an epithelioid phenotype, butfeeding morphologically fibroblastoid myoepithelial cells withmedium conditioned by a confluent culture of 3T3 cells did notinduce epithelioid conversion. Interestingly, DOLA cellstransduced to express the HPV16 E6 and E7 oncoproteins didnot arrest growth or induce expression of α-smooth muscleactin in the absence of EGF and also did not convert toepithelioid morphology when cultured with 3T3 feeder cells(data not shown).

Behavior of myoepithelial cells cultured withkeratinocytes and fibroblasts in Type I collagenorganotypic cultures In light of the remarkable morphologic plasticity andmodulation and switching of keratin expression patterns underdifferent conditions of conventional culture, we attempted tostudy their differentiation potential in three-dimensional,‘organotypic’ culture under conditions that would permit them

Table 2. Immunohistochemical detection of filament and junctional proteins in the epithelial structures of human skinα-s.m.

Structure K5 K7 K8 K14 K17 K19 Vim actin E-cad

Epidermis + − − + − − − − +Hair follicle + + − + + + − − +Sebaceous gland + + − + + − − − +Eccrine duct + − − + − + − − +Eccrine secretory − + + − − + − − +Eccrine myoepithelium + + − + + − + + −

Fig. 5. The feeder fibroblast-modulated morphologic transition of myoepithelial cells is accompanied by alterations in keratin expression.DOLA cells were cultured in FAD medium for eight days with (a-e) or without (f-j) 3T3 feeder cells, such that they adopted an epithelioid orfibroblastoid morphology, respectively. They were then immunostained for the indicated proteins: (a,f) keratin K5; (b,g) keratin K14; (c,h)keratin K7 (antibody OVTL 12/30); (d,i) keratin K17; (e,j) vimentin. Bar, 100 µm, for all panels.

1932

to interact with a Type I collagen gel matrix and with otherskin cell types. Myoepithelial cells were co-cultured inorganotypic culture with either epidermal keratinocytes,dermal fibroblasts, or both (Fig. 7). When plated on the surfaceof a Type I collagen gel that contained dermal fibroblasts andcultured at the air-liquid interface, myoepithelial cells formeda single cell layer, very different from the stratified,differentiated epidermal tissue structure formed bykeratinocytes (Fig. 7a,b). When embedded in a collagen gelwith dermal fibroblasts, myoepithelial cells remained as singlecells or formed small aggregates aligned along collagen fibersbut did not form complex structures, such as ducts, discernibleat the level of light microscopy. When plated as a mixed cellsuspension with keratinocytes on the surface of organotypiccultures, myoepithelial cells tended to sort into clusters,preferentially aggregating with one another on the collagen

layer. They also formed close cell-cell attachments withsurrounding keratinocytes such that they moved upward withthem to the level of the granular layer when the keratinocytesformed a stratified, keratinized epithelium (Fig. 7c). However,there was no evidence of pore formation similar to theacrosyringium formed by eccrine sweat gland duct cells as theypenetrate the epidermis in vivo. Myoepithelial cells embeddedalone within a collagen gel adopted a spindle shapedmorphology identifiable by phase microscopy and contractedthe gel, similar to the behavior of dermal fibroblasts (data notshown).

Identification of proliferative myoepithelial cells asrare constituents in cultures of other skin biopsies We sought to identify and culture myoepithelial cells fromeight additional full-thickness skin biopsies in order to estimate

M. Schön and others

Fig. 6. Quantitative assessment ofmodulated keratin expression bycultured myoepithelial cells by FACSanalysis. DOLA cells were cultured asdescribed in Fig. 5 and their expressionof intermediate filament proteins wasassessed by FACS (middle and bottomrows). Epidermal keratinocytes (strainN) were analyzed for comparison (toprow). The numbers in each panelindicate the mean fluorescence intensityof the peak.

Fig. 7. Myoepithelial cells in organotypic culturepreferentially aggregate but do not stratify or formobvious glandular structures. DOLA cells were culturedat the air-liquid interface either on top of a Type Icollagen gel or within the gel, either alone or togetherwith epidermal keratinocytes or with dermal fibroblasts inrelative proportions indicated below each panel of thefigure. Cryostat sections of the cultures wereimmunostained for keratin 7 to identify myoepithelialcells (a and c) or for keratin 17 to identify bothkeratinocytes and myoepithelial cells (b). Arrows at leftand right indicate the surface epithelial/collagen gelinterface. Bar, 100 µm, for all panels.

1933Sweat gland myoepithelial cell differentiation

the population density of proliferative myoepithelial cells inskin and to attempt to generate additional primary lines of thiscell type. Rare colonies of myoepithelial cells were identifiedin second and third passage cultures initiated from three ofthese biopsies, JAKE, BABU, and C5-SkH (see Fig. 1a-c). Atthe time of their first detection, colonies of myoepithelial cellswere present in these three cultures at only about 1/1000-1/10,000 the frequency of keratinocyte colonies and at onlyabout 1/10-1/100 the frequency of fibroblast colonies. We wereable to confirm the identity of myoepithelial cells byimmunostaining cultures enriched in these cells, generated byisolating regions of cultures containing colonies ofcharacteristic myoepithelial morphology. However, we wereunable to obtain pure myoepithelial cell cultures free fromkeratinocyte and/or fibroblast contamination before senescencelimited their growth.

We sought to identify selective conditions for obtaining purepopulations of myoepithelial cells from mixed skin cellcultures by comparing the growth rates of myoepithelial cellswith those of epidermal keratinocytes and dermal fibroblasts invarious medium formulations and in the presence of severalkeratinocyte growth inhibitors. BRSO and DOLA grew withsimilar doubling times in a variety of culture mediumformulations, including DME/F12 and M199, supplementedwith calf serum, EGF, and HC, and also in Gibco ker-sfmsupplemented with EGF, HC, and either bovine pituitaryextract or calf serum. Similar to keratinocytes and unlikefibroblasts, BRSO and DOLA cells were strongly inhibited byTGF-β and by PMA (data not shown). We have not yetidentified conditions that are simultaneously permissive for

myoepithelial cell growth and selective against bothkeratinocytes and fibroblasts.

The ndk cell line exhibits characteristics ofmyoepithelial cellsThe origin and identity of the ‘ndk’ line, a human skin-derivedcell line with unusual properties (Adams and Watt, 1988), hasremained unresolved. Ndk was isolated from newborn foreskinas a population of morphologically distinctive cells that grewduring serial passage in the 3T3 feeder layer system andpossessed a longer, albeit finite, replicative lifespan than thekeratinocytes that had predominated in the first severalpassages. The ndk cells were reported to have a slightkaryotypic abnormality, to express keratins similar byelectrophoretic analysis to those of epidermal keratinocytes,and to be unable to form cornified envelopes. As shown in Fig.8, our immunocytochemical analysis disclosed that ndk cellsexpress α-sma when deprived of EGF and they express K17and K7, the latter recognizable in intact cells by the OVTL12/30 but not by the Ks7.18 antibody. These results suggest thatndk is of myoepithelial origin and is not a non-differentiating,mutational variant that arose from a normal epidermalkeratinocyte of the donor. The ndk cells exhibited somedifferences from our BRSO and DOLA myoepithelial celllines, perhaps as a result of their chromosome abnormality orbecause they were near replicative senescence when weexamined them. They did not convert to an epithelioidmorphology when plated with 3T3 cells, a small fraction of thecells in the population expressed K19, and a similar fractionwere immunostained by the Ks7.18 antibody. Whether the

Fig. 8. Ndk cells express structural proteinsconsistent with a myoepithelial origin. Ndk cells at8th passage were cultured in multiwell plates for 8days in FAD medium. A phase contrast view of thecells is shown in a. Cells were immunostained forthe following proteins: (b) keratin K17; (c) keratinK7 (antibody OVTL 12/30); (d) keratin K7(antibody Ks7. 18); (e) keratin K19; (f) α-smoothmuscle actin (the cells in this well had been refedwith medium lacking EGF for the final five daysbefore fixation). Bar, 200 µm, for all panels.

1934

Ks7.18+ cells were the same cells that were K19+ in the ndkculture was not determined, but it seems likely that at leastsome K7 in the K19+ cells paired with K19 instead of K17 toform filaments, exposing the epitope recognized by the Ks7.18antibody.

DISCUSSION

Having begun with the objective of determining the identify ofa novel cell type that occasionally appears in human epidermalkeratinocyte cultures, we have identified cell type-specificmarkers and have determined permissive culture conditions forsweat gland myoepithelial cells. The in vitro growthrequirements of skin myoepithelial cells, most notably theirdependence upon EGF and HC in addition to mitogens presentin serum or pituitary extract, are consistent with the results ofearlier studies of human mammary gland myoepithelial cellsin short-term or mixed cell cultures (Peterson and van Deurs,1988; O’Hare et al., 1991) and are similar to those previouslyidentified for two simple (as distinguished from stratified)epithelial cell types, mesothelial cells and kidney tubuleepithelial cells (Connell and Rheinwald, 1983; Rheinwald andO’Connell, 1985).

Our study discovered the expression of cytokeratin K7 bymyoepithelial cells, which had not been detected previously(Moll and Moll, 1992) using an antibody that recognizes thiskeratin in other cell types, including the secretory coil cells ofthe sweat gland. Comparing the keratin expression patterns andimmunocytochemical staining characteristics of culturedmyoepithelial cells, mesothelial cells, and HeLa cells led us tothe conclusion that when cells express K7 and K17 but notK19, K7 pairs with K17 in filament formation and results inmasking of epitopes recognized by several commonly used K7-specific monoclonal antibodies. K7/K17 coexpression is veryunusual. Immunohistologic detection of K7/K17 pairing incells, accomplished by using an appropriate set of K7 and K17antibodies as we have reported here, could be useddiagnostically to aid in identifying normal and neoplastic cellsof myoepithelial origin in skin and other tissues.

The extraordinary degree of morphologic anddifferentiative plasticity of the myoepithelial cell in culturewas unexpected, considering the limited ability of sweatglands to regenerate in developmentally mature skin. Thecells responded to the presence of serum, when in the absenceof fibroblast feeder cells, with a pronounced elongation to aspindle shape accompanied by an altered pattern of keratinexpression. We do not know whether there is a preciseanalogue of this situation in vivo, although structuraldisruption and exposure to serum factors during the severaldays following deep incisional or abrasive skin woundingcould elicit in myoepithelial cells some of the changes weobserved in culture. The plasticity and considerableproliferative potential of myoepithelial cells is consistentwith a potential role for this cell type in skin wound healingin addition to its structural function in the intact sweat gland.Myoepithelial cells may contract and remodel collagen fibersin granulation and scar tissue, a possibility supported by thefact that this cell type can contract a Type I collagen gel inculture.

The ability to switch between an epithelial and mesenchymal

cell morphology and pattern of protein expression occurs aspart of the morphogenesis of many tissues during developmentbut it is a very uncommon property among somatic cell typesin the adult. Mesothelial cells and renal cortical tubuleepithelial cells are two other cell types that can undergo asimilar epithelial-mesenchymal conversion in culture and,presumably, under certain conditions in vivo (Connell andRheinwald, 1983; LaRocca and Rheinwald, 1984; Rheinwaldand O’Connell, 1985). We found that myoepithelial cells stablytransfected to express the HPV16 E6 and E7 proteins wereunable to arrest growth and express α-sma when cultured inthe absence of EGF and also were unable to convert to anepithelial phenotype when co-cultured with 3T3 feeder cells.This indicates that the ability to modulate expression of bothof these features of myoepithelial cell differentiation requiresnormal p53 and/or pRB dependent cell cycle regulation. Themyoepithelial cell’s ‘epithelial/mesenchymal’ switchingdiffers from that of the mesothelial and kidney tubule cell inthat the myoepithelial cell continues to express two keratins,K7 and K17, in its morphologically fibroblastoid state, whereasmesothelial cells and kidney tubule epithelial cells becomecompletely keratin-negative.

The ndk cell line, isolated as a morphologic variant from anepidermal keratinocyte culture by Adams and Watt (1988), hasproved to express distinctive myoepithelial markers, K7/K17and α-sma, suggesting that this cell line is not a differentiationdefective keratinocyte mutant, as had been proposed initially.The late passage ndk cells we examined exhibited slightdifferences from our three diploid primary myoepithelial celllines, however. The ndk cells did not reacquire an epithelioidmorphology when cultured with 3T3 feeder cells and a smallproportion of the cells expressed K19. This may be a result ofthe chromosome 1 abnormality identified in this cell line(Adams and Watt, 1988).

Both the apocrine and eccrine sweat glands of human skincontain myoepithelial cells. The cells we cultured were derivedfrom full-thickness skin specimens, most of which were fromaxilla, so our myoepithelial cell lines could be of either eccrineor apocrine origin. We were unable to obtain specimens ofhuman skin from which the relatively rare apocrine glandscould be identified in sections for immunohistochemicalanalysis. Ultrastructural examination has identifiedmyoepithelial cells in the secretory coil of apocrine glands aswell as eccrine glands (for example, see Saga and Takahashi,1992). Although the developmental formation of these twotypes of sweat glands is different, they form at different timesfrom separate cell lineages in the primitive epidermis, thefunction of myoepithelial cells in all glands that contain themappears to be the same. Myoepithelial cells from eccrine,apocrine, and even mammary, salivary, and lacrymal glandswould, therefore, be expected to have very similar patterns ofgene expression and regulation. Possible differences amongthem might include differential responsiveness to reproductivehormones and neurotransmitters which regulate the function ofsome of these glands and may affect the myoepithelial cellsdirectly.

The stringent growth requirement of myoepithelial cells forEGF and HC, even in the presence of high concentrations ofserum, is also a characteristic of human mesothelial cells andkidney tubule epithelial cells (Connell and Rheinwald, 1983;Rheinwald and O’Connell, 1985). However, in contrast to the

M. Schön and others

1935Sweat gland myoepithelial cell differentiation

latter cell types, myoepithelial cells are inhibited by TGF-β andby PMA, resembling the keratinocyte in this respect. Furtherresearch on the biology of this cell type will be aided greatlyby identifying a culture medium formulation that promotesmyoepithelial cell proliferation while completely orsubstantially preventing keratinocyte and fibroblastproliferation, thereby permitting routine initiation of primarymyoepithelial cell lines from full-thickness skin biopsies. Rareproliferative myoepithelial cells in skin biopsies couldpotentially be selected out from the large population ofepidermal keratinocytes by culturing in serum-supplementedGibco ker-sfm medium without 3T3 feeder cells. However, anydermal fibroblasts also present in such cultures can eventuallyoutgrow the myoepithelial cells in this and all other mediumformulations we have tested to date. Enriched myoepithelialcell populations have been obtained from mammary gland cellsuspensions by antibody-based cell sorting, exploiting theexpression of the CALLA antigen by mammary glandmyoepithelial cells in vivo (O’Hare et al., 1991; Gomm et al.,1995). Although myoepithelial cells are present at severalorders of magnitude lower proportion of the total cellpopulation in skin than in the mammary gland, such a strategymay aid the isolation of pure sweat gland myoepithelial cellpopulations.

The repair and regeneration capacity of skin adnexalstructures and the regulatory mechanisms involved are subjectsof great interest, in part for the potential applicability of thisknowledge to developing improved transplantation andpharmacologic strategies for restoring skin damaged by burns.The epidermis can be restored permanently on large areas offull thickness wounds by applying sheets of culturedautologous epidermal keratinocytes to residual dermal orgranulation tissue (O’Connor et al., 1981; Leigh et al., 1991)but regenerating the very complex structures of thepilosebaceous unit and the eccrine sweat gland presents a muchgreater challenge (reviewed by Martin, 1997). The patterns anddensity of these adnexae in the normal adult skin aredetermined and the structures themselves formed during mid-gestation development. Pluripotent progenitor cells, which inthe embryonic epidermis migrate downward, replicate, andultimately differentiate into the three cell types of the matureeccrine sweat gland, are unlikely to be present in adult skin.After a partial thickness wound, the eccrine sweat duct canreestablish its connection through the regenerated epidermis tothe skin surface. Duct cells, therefore, must retain bothreplicative potential and the ability to form their appropriatethree dimensional structure, presumably using the deeper,undamaged part of the duct both as a source of stem cells andas a template. In full thickness wounds, however, both the ductand the secretory coil of the sweat gland are destroyed.Although the surface of such a wound may becomereepithelialized by keratinocytes migrating from the peripheryof the wound and differentiating to reestablish the epidermis,sweat glands do not regenerate.

Cells of the sweat gland secretory coil normally undergovery little turnover and replacement (for example, seeMorimoto and Saga, 1995). The type and extent of damage anddisruption from which this structure can recover by repairingitself and reestablishing connection with remnants of the ductis unknown, however, and the proliferative potential of themyoepithelial cells is consistent with regeneration potential.

Our initial attempt to assess the histogenic potential ofmyoepithelial cells in vitro, using a Type I collagen matrixorganotypic culture system with dermal fibroblasts andepidermal keratinocytes, was inconclusive. Culturingmyoepithelial cells in the presence of basement membranecollagens and laminins, sweat gland duct and secretory cells,and soluble morphogenetic proteins such members of the BMPfamily would provide a more comprehensive in vitroassessment of the histogenic potential of this cell type. Thedevelopment of methods for identifying and culturingmyoepithelial cells which we have reported here, combinedwith methods for initiating cultures of eccrine duct andsecretory cells (Collie et al., 1985; Lee et al., 1986; Jones etal., 1988; Pedersen, 1989) provides a starting point forinvestigations aimed at understanding the regenerativepotential of the sweat gland and at transplanting cultured cellsto restore these structures in damaged skin.

We thank J. Sigler, O. Kehinde, S. Schaffer, A. Schermer, and R.Tubo of Biosurface Technology, Inc. for providing cryopreservedampules of early passage cells cultured from human skin biopsies andfor providing several pure or enriched cell populations of cells whichwe subsequently characterized as myoepithelial cells. We thank Dr R.Tantravahi for karyotyping the CYHI and BRSO cell lines, Dr N.Parenteau for providing materials for and advice on preparingorganotypic cultures, and Dr F. Watt for providing the ndk cell line.We also thank Drs T.-T. Sun, S. M. Troyanovsky, M. Osborn, and F.Ramaeckers for providing keratin antibodies. This work wassupported by Skin Disease Research Center grant # P30 AR42689from the N.I.H. to T.S. Kupper, by grants from BiosurfaceTechnology, Inc. and from Organogenesis, Inc. to J.G.R., and by apostdoctoral fellowship from the Deutsche Forschungsgemeinschaftto M.S. Some of the data reported here were presented at the 1995meeting of the Society for Investigative Dermatology and werepublished in abstract form (O’Connell-Willstaedt, T., Schön, M. andRheinwald, J. G. (1995) J. Invest. Dermatol. 104, 641).

REFERENCES

Adams, J. C. and Watt, F. M. (1988). An unusual strain of humankeratinocytes which do not stratify or undergo terminal differentiation inculture. J. Cell Biol. 107, 1927-1938.

Allen-Hoffmann, B. L. and Rheinwald, J. G. (1984). Polycyclic aromatichydrocarbon mutagenesis of human epidermal keratinocytes in culture.Proc. Nat. Acad. Sci. USA 81, 7802-7806.

Bartek, J., Vojtesek, B., Staskova, Z., Bartkova, J., Kerekes, Z., Rejthat,A. and Kovarik, J. (1991). A series of 14 new monoclonal antibodies tokeratins: characterization and value in diagnostic histopathology. J. Pathol.164, 215-224.

Bell, C. L. and Quinton, P. M. (1991). Effects of media buffer systems ongrowth and electrophysiologic characteristics of cultured sweat duct cells.In Vitro Cell Dev. Biol. 27, 47-54.

Caselitz, J., Walther, B., Wustrow, J., Seifert, G., Weber, K. and Osborn,M. (1986). A monoclonal antibody that detects myoepithelial cells inexocrine glands, basal cells in other epithelia and basal and suprabasal cellsin certain hyperplastic tissues. Virchows Arch A Pathol. Anat. Histopathol.409, 725-738.

Cepek, K. L., Shaw, S. K., Parker, C. M., Russell, G. J., Morrow, J. S.,Rimm, D. L. and Brenner, M. B. (1994). Adhesion between epithelial cellsand T lymphocytes mediated by E-cadherin and the αEβ7 integrin. Nature372, 190-193.

Cohn, B. A. (1994). In search of human skin pheromones. Arch Dermatol.130, 1048-1051.

Collie, G., Buchwald, M., Harper, P. and Riordan, J. R. (1985). Culture ofsweat gland epithelial cells from normal individuals and patients with cysticfibrosis. In Vitro Cell Dev Biol. 21, 597-602.

Connell, N. D. and Rheinwald, J. G. (1983). Regulation of the cytoskeleton

1936

in mesothelial cells: reversible loss of keratin and increase in vimentinduring rapid growth in culture. Cell 34, 245-253.

Debus, E., Weber, K. and Osborn, M. (1982). Monoclonal cytokeratinantibodies that distinguish simple from stratified squamous epithelia:Characterisation on human tissues. EMBO J. 1, 1641-1647.

Fischer, H. P., Altmannsberger, M., Weber, K. and Osborn, M. (1987).Keratin polypeptides in malignant epithelial liver tumors. Differentialdiagnostic and histogenic aspects. Am. J. Pathol. 127, 530-537.

Franke, W., Schmid, E., Reudenstein, C., Appelhans, B., Osborn, M.,Weber, K. and Keenan, T. W. (1980). Intermediate-sized filaments ofprekeratin type in myoepithelial cells. J. Cell Biol. 84, 633-654.

Franke, W. W., Schiller, D. L., Hatzfield, M. and Winter, S. (1983). Proteincomplexes of intermediate-sized filaments: melting of cytokeratincomplexes in urea reveals different polypeptide separation characteristics.Proc. Nat. Acad. Sci. USA 80, 7113-7117.

Gomm, J. J., Browne, P. J., Coope, R. C., Liu, Q. Y., Buluwela, L. andCoombes, R. C. (1995). Isolation of pure populations of epithelial andmyoepithelial cells from the normal human mammary gland usingimmunomagnetic separation with dynabeads. Anal. Biochem. 226, 91-99.

Holbrook, K. A. and Wolff, K. (1993). The structure and development ofskin. In Dermatology in General Medicine (ed. T. B. Fitzpatrick, A. Z. Eisen,K. Wolff, I. M. Freedberg and K. F. Austen), pp. 97-145. McGraw-Hill, NewYork.

Jones, C. J., Bell, C. L. and Quinton, P. M. (1988). Different physiologicalsignatures of sweat gland secretory and duct cells in culture. Am. J. Physiol.255, C102-C111.

Karsten, U., Papsdorf, G., Roloff, G., Stolley, P., Abel, H., Walther, I. andWeiss, H. (1985). Monoclonal anti-cytokeratin antibody from a hybridomaclone generated by electrofusion. Eur. J. Cancer Clin. Oncol. 21, 733-740.

LaRocca, P. J. and Rheinwald, J. G. (1984). Coexpression of simpleepithelial keratins and vimentin by human mesothelium and mesotheliomain vivo and in culture. Cancer Res. 44, 2991-2999.

Lee, C. M., Carpenter, F., Coaker, T. and Kealey, T. (1986). The primaryculture of epithelia from the secretory coil and collecting duct of normalhuman and cystic fibrotic eccrine sweat glands. J. Cell Sci. 83, 103-118.

Leigh, I. M., McKay, I., Carver, N., Navsaria, H. and Green, C. (1991).Skin equivalents and cultured skin: from the petri dish to the patient. Wounds3, 141-148.

Lemullois, M., Rossignol, B. and Mauduit, P. (1996). Immunolocalizationof myoepithelial cells in isolated acini of rat exorbital lacrimal gland:cellular distribution of muscarinic receptors. Biol. Cell 86, 175-181.

Lynch, M. H., O'Guin, W. M., Hardy, C., Mak, L. and Sun, T.-T. (1986).Acidic and basic hair/nail (‘hard’) keratins: their co-localization in uppercortical and cuticle cells of the human hair follicle and their relationship to‘soft’ keratins. J. Cell Biol. 181, 2593-2606.

Martin, P. (1997). Wound healing–aiming for perfect skin regeneration.Science 276, 75-81.

Moll, I. and Moll, R. (1992). Changes of expression of intermediate filamentproteins during ontogenesis of eccrine sweat glands. J. Invest. Dermatol. 98,777-785.

Moll, R., Franke, W. W. and Schiller, D. L. (1982). The catalog of humancytokeratins: Patterns of expression in normal epithelia, tumors and culturedcells. Cell 31, 11-24.

Moll, R., Krepler, R. and Franke, W. W. (1983). Complex cytokeratinpolypeptide patterns observed in certain human carcinomas. Differentiation23, 256-269.

Morimoto, Y. and Saga, K. (1995). Proliferating cells in human eccrine andapocrine sweat glands. J. Histochem. Cytochem. 43, 1217-1221.

Norberg, L., Dardick, I., Leung, R., Burford-Mason, A. and Rippstein, P.(1992). Immunogold localization of actin and cytokeratin filaments inmyoepithelium of human parotid salivary gland. Ultrastruct Pathol. 16, 555-568.

O’Connor, N. E., Mulliken, J. B., Banks-Schlegel, S., Kehinde, O. andGreen, H. (1981). Grafting of burns with cultured epithelium prepared fromautologous epidermal cells. The Lancet 10, 75-78.

O’Hare, M. J., Ormerod, M. G., Monaghan, P., Lane, E. B. and Gusterson,B. A. (1991). Characterization in vitro of luminal and myoepithelial cellsisolated from the human mammary gland by cell sorting. Differentiation 46,209-221.

Osborn, M., Debus, E. and Weber, K. (1984). Monoclonal antibodies specificfor vimentin. Eur. J. Cell Biol. 34, 137-143.

Parenteau, N. L., Nolte, C. M., Bilbo, P., Rosenberg, M., Wilkins, L. M.,Johnson, E. W., Watson, S., Mason, V. S. and Bell, E. (1991). Epidermisgenerated in vitro: Practical considerations and applications. J. Cell.Biochem. 45, 245-251.

Pedersen, P. S. (1989). Human sweat duct cells in primary culture. Basicbioelectric properties of cultures derived from normals and patients withcystic fibrosis. In Vitro Cell Dev. Biol. 25, 342-352.

Peterson, O. W. and van Deurs, B. (1988). Growth factor control ofmyoepithelial-cell differentiation in cultures of human mammary gland.Differentiation 39, 197-215.

Pirisi, L., Yasumoto, S., Feller, M., Doniger, J. and DiPaolo, J. A. (1987).Transformation of human fibroblasts and keratinocytes with humanpapillomavirus type 16 DNA. J. Virol. 61, 1061-1066.

Ramaekers, F., Huysmans, A., Moesker, O., Kant, A., Jap, P., Herman, C.and Vooijs, P. (1983). Monoclonal antibody to keratin filaments, specificfor glandular epithelia and their tumors. Lab. Invest. 49, 353-361.

Rheinwald, J. G. and Green, H. (1975). Serial cultivation of strains of humanepidermal keratinocytes: The formation of keratinizing colonies from singlecells. Cell 6, 331-344.

Rheinwald, J. G. and Beckett, M. A. (1980). Defective terminaldifferentiation in culture as a consistent and selectable character ofmalignant human keratinocytes. Cell 22, 629-632.

Rheinwald, J. G. and O'Connell, T. M. (1985). Intermediate filamentproteins as distinguishing markers of cell type and differentiated state incultured human urinary tract epithelia. Ann. NY Acad. Sci. 455, 259-267.

Rheinwald, J. G. (1989). Methods for clonal growth and serial cultivation ofnormal human epidermal keratinocytes and mesothelial cells. In CellGrowth and Division: A Practical Approach (ed. R. Baserga), pp. 81-94.IRL Press, Oxford.

Rollins, B. J., O'Connell, T. M., Bennett, G., Burton, L. E., Stiles, C. D.and Rheinwald, J. G. (1989). Environment-dependent growth inhibition ofhuman epidermal keratinocytes by recombinant human transforming growthfactor-β. J. Cell. Physiol. 139, 455-462.

Saga, K. and Takahashi, M. (1992). Immunoelectron microscopiclocalization of epidermal growth factor in eccrine and apocrine sweatglands. J. Histochem. Cytochem. 40, 241-249.

Sato, K., Nishiyama, A. and Kobayashi, M. (1979). Mechanical propertiesand functions of the myoepithelium in the eccrine sweat gland. Am. J.Physiol. 237, C177-C184.

Sato, K. (1993). Biology of the eccrine sweat gland. In Dermatology inGeneral Medicine (ed. T. B. Fitzpatrick, A. Z. Eisen, K. Wolff, I. M.Freedberg and K. F. Austen), pp. 221-241. McGraw-Hill, New York.

Schön, M. and Rheinwald, J. G. (1996). A limited role for retinoic acid andretinoic acid receptors RARα and RARβ in regulating keratin 19 expressionand keratinization in oral and epidermal keratinocytes. J. Invest. Dermatol.107, 428-438.

Skalli, O., Ropraz, P., Treciak, A., Benzonana, G., Gillessen, D. andGabbiani, G. (1986). A monoclonal antibody against α-smooth muscleactin: a new probe for smooth muscle differentiation. J. Cell Biol. 103, 2787-2796.

Southgate, J., Williams, H. K., Trejdosiewicz, L. K. and Hodges, G. M.(1987). Primary culture of human oral epithelial cells. Growth requirementsand expression of differentiated characteristics. Lab. Invest. 56, 211-223.

Tölle, H.-G., Weber, K. and Osborn, M. (1985). Microinjection ofmonoclonal antibodies specific for one intermediate filament protein in cellscontaining multiple keratins allow insight into the composition of particular10 nm filaments. Eur. J. Cell Biol. 38, 234-244.

Troyanovsky, S. M., Guelstein, V. I., Tchipysheva, T. A., Krutovskikh, V.A. and Bannikov, G. A. (1989). Patterns of expression of keratin 17 inhuman epithelia: dependency on cell position. J. Cell Sci. 93, 419-426.

van Niekerk, C. C., Jap, P. H., Ramaekers, F. C., van de Molengraft, F.and Poels, L. G. (1991). Immunohistochemical demonstration of keratin7 in routinely fixed paraffin-embedded human tissues. J. Pathol. 165, 145-152.

Wu, Y.-J., Parker, L. M., Binder, N. E., Beckett, M. A., Sinard, J. H.,Griffith, C. T. and Rheinwald, J. G. (1982). The mesothelial keratins: Anew family of cytoskeletal proteins identified in cultured mesothelial cellsand nonkeratinizing epithelia. Cell 31, 693-703.

Yokozeki, H., Saga, K., Sato, F. and Sato, K. (1990). Pharmacologicalresponsiveness of dissociated native and cultured eccrine secretory coil cells.Am. J. Physiol. 258 (Reg Int Comp Physiol 27), R1355-R1362.

M. Schön and others