Organ maintenance of human sebaceous glands: in vitro effects of 13-c/s

retinoic acid and testosterone

JOHN RIDDEN1'*, D. FERGUSON2 and TERENCE KEALEY1

'Nit/field Department of Clinical Biochemistry, Oxford University, John Radcliffe Hospital, Headington, Oxford 0X3 9DU, UK^'Department of Electron Microscopy, John Radcliffe Hospital, Headington, Oxford 0X3 9DU, UK

•Present address: Department of Clinical Biochemistry, University of Cambridge, Addenbrookes Hospital, Hills Road, Cambridge CB2 2QR,UK

Summary

Human sebaceous glands were isolated by shear-ing, and maintained for 7 days either on definedmedium, on medium supplemented with 3//M-tes-tosterone or on medium supplemented with both3 fiM-testosterone and l/iM-13-cis retinoic acid.

Freshly isolated glands retained their in vivomorphology. On maintenance, the glands retainedtheir freshly isolated rates of cell division, but thesebocytes showed increased keratinization andthere was multilayering of the peripheral undiffer-entiated cells. However, glands maintained in thepresence of lftM-13-cis retinoic acid showed verylittle luminal keratinization and only a small de-gree of multilayering.

On autoradiography, freshly isolated glandsretained their in vivo pattern of [mef/iy/-3H]thymi-dine incorporation. Similar patterns 'were seenwhen glands were maintained for 7 days with orwithout testosterone. However, in the presence ofboth testosterone and 13-cis retinoic acid there wasonly slight graining.

Following 7 days maintenance the rate of lipo-genesis fell significantly. This was partiallyreversed by testosterone, but further inhibited by13-cis retinoic acid. The patterns of lipids that aresynthesised after a week's maintenance are very

similar to those seen in freshly isolated glands,except that the squalene: cholesterol ratio is revers-ibly regulated by 3 /UM-testosterone and ifiM-reti-noic acid.

Protein synthesis was maintained at the samerates as for freshly isolated glands under all con-ditions of maintenance. Whereas DNA syntheticrates were maintained in the presence of testoster-one, they were significantly inhibited by 13-cisretinoic acid.

Glandular wet weights were retained under allconditions of maintenance, except that they weresignificantly reduced by 13-cis retinoic acid.

This study shows that human sebocytes continueto divide on organ maintenance, but that they do notdifferentiate fully. However, this provides the firstdemonstration that 13-cis retinoic acid acts on hu-man sebaceous glands directly, reducing the rate ofcell division and the rate of lipogenesis, whichshows that the maintained human sebaceous glandmight provide a useful model for studying the effectof 13-cis retinoic acid on human sebocytes.

Key words: sebaceous gland, retinoic acid, organmaintenance.

Introduction

The importance of the retinoids in regulating the growthand differentiation of vertebrate cells has been increas-ingly recognised. Clinically, retinoids are useful in thetreatment of a number of diseases of differentiationincluding ichthyosis, psoriasis and acne vulgaris (Bollag,1983). Etretinate is useful in the treatment of dermallymphoma, while 13-czs retinoic acid may be anti-onco-genic (Kraemer et al. 1988). In vitro, retinoids willinfluence the growth and differentiation of many cell lines

Journal of Cell Science 95, 125-136 (1990)Printed in Great Britain (6) The Company of Biologists Limited 1990

(Sporn et al. 1984). Moreover, vitamin A has beenrecognised to be a morphogen, following the observationthat a gradient of vitamin A regulates the embryonicgrowth and development of chick limb buds (Thaller andEichele, 1987). At least three mammalian retinoic acidreceptors have recently been cloned (Petkovich et al.1987; Giguere et al. 1987; Benbrooke^a/. 1988; Zelentefal. 1989), and they have been characterized as membersof the steroid/thyroxine family of nuclear receptors.These act as transducible transcription-enhancer factors(Green and Chambon, 1986). However, only one of the

125

genes whose transcription retinoic acid enhances has beencharacterised (LaRosa and Gudas, 1988a,b) and itsbiological significance remains obscure.

To help determine the mode of action of retinoids inhuman tissues, we have isolated human sebaceous glandsas an in vitro model. These are the dermal glands thatsecrete sebum onto the skin; and it is these glands andtheir follicular ducts that are affected in acne vulgaris.

Sebaceous glands secrete sebum by holocrine secretion(Plewig and Christophers, 1974). Sebocytes originate atthe periphery of the sebaceous gland lobule, and as theygrow they accumulate lipid and move centrally. Eventu-ally, after some 7-28 days, (Epstein and Epstein, 1966;Plewig and Christophers, 1974) the cell dissolves, liberat-ing sebum into the sebaceous duct, which directs it to thesurface of the skin.

Clinically, Yi-cis retinoic acid is an extremely effectivetreatment for acne vulgaris (Peck et al. 1979). In vivostudies show that Yi-cis retinoic acid reduces sebumsecretion to about 10% of pre-treatment levels (Jones etal. 1983). Histologically, it can be shown that this isassociated with a marked decrease in sebaceous gland size(Peeked al. 1982; Gomez, 1982).

The human sebaceous gland, therefore, should providean excellent model for the study of Yi-cis retinoic acidaction. To characterise this, we have isolated the glands,maintained them for up to 14 days, and determined the invitro effect of 13-m retinoic acid on their light- andelectron-microscopic appearance, their keratin immuno-histochemistry, their wet weights, their protein andDNA contents, their rate of DNA and protein synthesis,their autoradiographic response to [mei/i3'/-3H]thymi-dine, and their rates and patterns of lipid synthesis.

Sebaceous glands develop at puberty under the influ-ence of testosterone (Pochi et al. 1977). The mammaliantestosterone receptor is believed to be of the same familyas the retinoid (Green and Chambon, 1986). However,the mode of action of testosterone in the regulation of cellgrowth is not known. To characterise the value ofmaintained isolated sebaceous glands as a model for thestudy of testosterone action, its in vitro effects on thesebaceous gland parameters described above were alsodetermined.

Materials and methods

MaterialsTissue culture plastics were obtained from Becton Dickson,Between Towns Road, Cowley, Oxford 0X4 3LY England.Williams E media, Earle's balanced salt solution, phosphate-buffered saline (PBS), bovine foetal calf serum, L-glutamine,penicillin, streptomycin and Fungizone were from Gibco,Trident House, PO Box 35, Renfew Rd, Paisley PA3 LEF,Scotland. Nitrocellulose membrane filters were from WhatmanLtd, Maidstone, England. All other tissue culture reagents andchemicals were from Sigma Chemical Co Ltd, Fancy Rd,Poole, Dorset, UK, or BDH Ltd, Broom Rd, Poole, Dorset,UK, and of the highest purity available. Radiochemicals werefrom Amersham International, White Lion Rd, AmershamHP7 9L4. Rabbit anti-keratin wide-spectrum polyclonal anti-body and Dako System 40 avidin-biotin labelling system were

from Dako Ltd, 22, The Arcade, The Octagon, HighWycombe, Bucks HP11 2HT.

Methods

Isolation of viable human sebaceous glands. Samples ofnormal midline chest skin (5 mm X 60 mm) were obtained frommale patients aged between 20 and 75 undergoing cardiacsurgery at the John Radcliffe Hospital, Headington, Oxford orthe Bristol Royal Infirmary, Bristol, UK. Ethical committeepermission has been granted for this technique. Glands wereisolated by shearing (Kealey et al. 1986).

Organ maintenance of sebaceous glands. Sebaceous glandswere first maintained overnight at 37°C, on nitrocellulosefilters, pore size 0.45 ^m (Fell, 1953; Trowell, 1959), floatingon 5 ml Williams E medium supplemented with 10 % foetal calfserum, 2 mM-L-glutamine, 100 units ml~ penicillin and strepto-mycin and 2.5[igm\~l Fungizone and buffered in humidified95 % air: 5 % CO2. Glands to be maintained for a further 7 dayswere floated on Williams E medium containing the following:2 mM-L-glutamine, 100 units ml~ penicillin and streptomycinand 2.5/Jgml~l Fungizone, lO/ugml"1 insulin, lO/Ugml"1

transferrin, lOngml" hydrocortisone, lOngmF 1 epidermalgrowth factor, lOngrnl" sodium selenite, 3 nM-triiodothyro-nine, trace elements (Gibco), lOngrnl" prostaglandin Ei and10 fig ml~ bovine pituitary extract and buffered in a humidified5 % CO2 atmosphere. The medium was supplemented whereappropriate with 1 /jM-13-cis retinoic acid and/or 3 ̂ M-testoster-one; 1 [iM-13-cis retinoic acid was chosen, because of the dose-response curve (see Fig. 5, below); 3 jttM-testosterone waschosen because the dose-response curve (see Fig. 6, below)showed it gave a maximal response. 13-C/s retinoic acid wasdissolved in dimethylsulphoxide (DMSO), the final concen-tration of which did not exceed 0.02% (v/v), and testosteronewas dissolved in ethanol, the final concentration of which didnot exceed 0.002% (v/v). Where 13-cw retinoic acid was absentfrom control experiments, DMSO was added to 0.02% (v/v)and where testosterone was absent, ethanol was added to0.002% (v/v).

[U-I4C] sodium acetate and [U-l4C]glucose incorporation.Five glands were incubated in 300,1(1 of bicarbonated-bufferedsaline ( N a \ 144mM; K + , 5.9rnM; Mg2+, 1.2mM; Ca2+,1.25 HIM; Cl", 125mM; SO4~, 1.2mM; PO4~, 1.2mM;HCO3~, 25 mM; pH7.4) (Krebs and Henseleit, 1932), contain-ing either 2mM-[U-14C]glucose at 1.66^Ciml~' (sp. act.833/iCimmol"1) or 2mM-[U-14C]sodium acetate at1.66juCiml~ (sp. act. SSSjuCimmol"1), in humidifiedair: CO2 (95 %: 5 %) at 37°C. After 3 h the glands were washedin four changes of PBS. Control experiments showed that thefurther release of radioactivity after the fourth wash wasnegligible. The lipids were extracted by homogenization, usinga glass/glass homogeniser, into chloroform: methanol: 0.88%potassium chloride: water at a ratio of 2:2:1:0.8 (by vol.)(Bligh and Dyer, 1959). The chloroform fraction was filteredthrough glass wool and collected in a 20 ml glass scintillationvial. The extract was dried under a stream of nitrogen and 10 mlof scintillant added. The radioactivity was determined by liquidscintillation spectrometry using a Tricarb system 4000^-counter.

Control experiments showed that the rate of lipogenesis from[U-14C]sodium acetate was linear over 24 h and that therecovery of exogenous [14C]cholesterol and glycerol tri-[14C]palmitate was over 85 %.

Lipid class identification by thin-layer chromatography(TLC). The dried sebaceous gland lipid extract was redissolvedin 300 jul of chloroform: methanol: 0.88% potassium chloride(5:5: 1, by vol.), and spotted onto a 20 cmX20 cm, 250 /Am silicagel chromatography plate. The plate was then developed in four

126 J. Ridden et al.

directions by four different solvents (Cooper et al. 1974).Solvent 1 (light petroleum 40-60°C:diethyl ether: acetic acid(SO: 50:1, by vol.), solvent 2 (light petroleum: benzene (70: 30,v/v), solvent 3 (light petroleum 40-60°C). To separate thepolar lipids, the bottom 3 cm of the plate was cut off anddeveloped in solvent 4 (chloroform: methanol: acetic acid: water(25:15:4:2, by vol.). Each test plate was run simultaneouslywith a lipid standard plate.

The plates were then autoradiographed using Kodak X-omatAR X-ray film, and the radioactive spots removed into scintil-lation vials and 10 ml of scintillant was added. The radioactivitywas determined by liquid scintillation spectrometry using aTricarb system 4000 /3-counter.

Rates ofDNA and protein synthesis. A dual assay system wasemployed, whereby five to eight sebaceous glands were incu-bated in 500 ,ul Williams E medium containing 500,UM-[U-14C]leucine (sp. act. 2mCi^mol~'), 3^M-[wje//y)/-3H]thymi-dine (sp. act. 1.33 ,uCi nmol"1) and equilibrated at 37°C inhumid 5 % CO2:95 % air for 3 h. The glands were removed andwashed in three changes of PBS containing 3 ^M-thymidine todisplace non-covalently bound [w7e//ry/-3H]thymidine. Hom-ogenization took place in a glass/glass homogeniser in ice-cold100 mM-K/EDTA, pH 12.4, and the homogenate left at 4°C for30min to release and dissolve DNA and protein (West et al.1985); the debris was then removed by centrifugation at10000g for 3min. The macromolecules were precipitated bythe addition of ice-cold perchloric acid (10%, v/v) to thesupernatant, which was left for at least 2h. The precipitate wascollected by being passed, under vacuum, through WhatmanGF/C filters previously washed with lOmM-thymidine toreduce any non-specific binding of [/we//;;y/-3H]thymidine to thefilter. The filters were then washed with 10 ml of ice-cold 10%(w/v) trichloroacetic acid (TCA) and 5 ml of 5 % (w/v) TCAfollowed by 1 ml of ethanol: diethyl ether (1:1, v/v). The filterswere dried at 60°C in an oven, and the radioactivity wasdetermined by dual-counting liquid scintillation spectrometryusing a Tricarb system 4000 /3-counter.

Control experiments showed that the rate of both leucine andthymidine incorporation by glands into TCA-precipitable ma-terial was linear over 18 h.

DNA and protein assay. Prior to the precipitation of theproteins and DNA by the addition of 10% perchloric acid(v/v), two 100-jUl samples were removed, 100 (x\ for total proteindetermination using Coomassie Blue (Bradford, 1976) and100 ji\ for total DNA determination using diaminobenzoic acid(DABA) (Fiszer-Szafarj et al. 1981).

Autoradiography. Five to eight sebaceous glands were incu-bated in 500^il Williams E medium containing 3 [iM-[methyl-3H]thymidine (sp. act. 13.3^Cinmon') equilibrated at 37°C,in humid 5 % CO2: 95 % air for 6 h. The glands were washed infour, 1-ml changes of PBS containing 3 ,UM-thymidine, toremove any non-specific binding. The glands were blotted andfixed overnight in PBS containing 4.25 % glutaraldehyde. Serial2,um semi-thin sections were taken using a microtome and every10th section was prepared for autoradiography. Exposure wasfor 1 month. The scetions were then stained with heamotoxylinand eosin.

Light microscopy. Glands were first fixed in PBS containing4.25% glutaraldehyde for 24 h. The glands were then dehy-drated by processing through an ethanol series of increasingconcentration, followed by impregnation with chloroform, andembedded in paraffin wax. The glands were serially cut into 4-/lm sections, followed by staining in heamotoxylin and eosin.

lmmunohistochemistry. Glands were fixed, dehydrated andcut as described above. The slides were dewaxed in xylene andre-hydrated by passing through an ethanol series of decreasingconcentration, from 100% to water. The glands were then

treated with 0 .1% trypsin for 30min to expose epitopes asdescribed by Culling et al. (1980). Endogenous peroxidaseactivity was blocked by the addition of methanolic hydrogenperoxide (Escribano et al. 1987) and the sections were thenlabelled with the polyclonal antibody, which recognises keratinsubunits 60, 58, 56, 52, 51 and 48(xlO3)Mr (Steinert, 1975;Franke et al. 1978) and stained as described by Lloyd et al.(1985). Control experiments were carried out using non-immune rabbit serum in the place of anti-keratin antibody.

Electron microscopy. For electron microscopy, glands werefixed in 4.25 % glutaraldehyde as described above, followed by1 % osmium tetroxide in the same buffer. The glands weredehydrated, then impregnated with propylene oxide and em-bedded in Emix (Emscope) resin, and the blocks were polymer-ised overnight at 60cC. Ultrathin sections were taken using aReichert Ultracut and collected on copper grids, then stainedwith uranyl acetate and lead citrate. The sections were viewedusing a Philips EM301 transmission electron microscope atapproximately 60 kV.

Results

Light microscopy

Fig. 1 illustrates the histological differences observed insebaceous glands under different conditions. Fig. 1Ashows a freshly isolated gland showing both a large lobeof typical sebaceous morphology, and the sebaceous duct(D). The three distinct sebocyte cell types can be seen:(1) peripheral undifferentiated cells (UDC); (2) partiallydifferentiated sebocytes (PDC) containing a small num-ber of lipid vacuoles (L); and (3) the central maturedifferentiated sebocyte (MC) containing many lipidvacuoles, which deform the nucleus.

Fig. 2A shows a typical immunohistochemical sectionof a freshly isolated gland labelled with an antibody tokeratin subunits 60, 58, 56, 52, 51 and 48(XlO3)Mr, thebrown stain being where the keratin antibody has bound.Heavy staining of the sebaceous duct can be seen withkeratinous material filling the duct. Mature sebocytes arestained but, interestingly, neither PDC nor UC. Thesurrounding collagen is also stained but this is an artifactthat appears in negative control sections stained withnon-immune rabbit sera (Fig. 2B).

Fig. IB illustrates a sebaceous gland after 7 daysmaintenance in the absence of both testosterone and13-czs retinoic acid. The architecture has changed, and itnow shows a multilayered peripheral poorly partiallydifferentiated sebocyte (PPDC) population of betweenfive and eight layers. We refer to such cells as poorlypartially differentiated sebocytes because they occupy thespace of partially differentiated sebocytes, but they showonly very occasional lipid droplets (L). Mature sebocytesare seen centrally. Surprisingly, the luminal surfaceshows considerable keratinization (K), which is not seenin freshly isolated glands. This is confirmed as keratin byFig. 2C. This figure also shows that the PPDC unlike thePDC and UC of freshly isolated glands, reacted with theantibody to 60, 58, 56, 52, 51 and 48(xlO3)Mr keratinsubunits.

The follicular duct has undergone two changes(Fig. IB). First, much greater luminal keratinizationthan is seen in freshly isolated ducts has developed.

Sebaceous gland organ maintenance 127

udc

1Appdc

•;-%w

me v < ^ l '

B~Tppdc

Fig. 1. The light-microscopic appearance of fresh glands and glands maintained for 7 days under different conditions.Sebaceous glands were prepared for light microscopy as described in Materials and methods. These fresh glands (X50) showtypical sebaceous gland morphology, with well-defined three cell types: undifferentiated cells (udc) partially differentiated cells(pdc) and mature sebocytes (me). The follicular duct has a number of layers of cells and a small degree of superficialkeratinization. This micrograph is representative of « = S subjects, 5 sebaceous glands per subject. B. These glands (X50) weremaintained for 7 days on defined medium. The architecture has changed and a multilayered poorly partially differentiatedsebocyte population (ppdc) can be seen, as can considerable luminal keratinization (k). A small population of mature sebocytes(me) can be seen centrally. 1, lipid vacuole. The follicular duct (d) of a gland shows two major changes from fresh; greaterluminal keratinization (k) has occurred and follicular duct cells have formed deeper layers. This micrograph is representative of«=5 subjects, 5 sebaceous glands per subject. C. Illustrated is a gland (X50) maintained for 7 days in the presence of 3,UM-testosterone. There is no marked difference from that of 7 days' maintenance in the absence of testosterone. The multilayeredappearance of the lobe and the luminal keratinization of the follicular duct are not markedly different from that seen in B. Thismicrograph is representative of « = 5 subjects, S sebaceous glands per subject. D. Illustrated is a sebaceous gland (X108)maintained for 7 days in the presence of both 3 fJM-testosterone and 1 flM-\3-cis retinoic acid. The architecture of the fresh glandhas been maintained to a large degree. There is no multilayering of the lobe and the luminal keratinization of the follicular ductis no greater than that of fresh glands. This micrograph is representative of »=5 subjects, 5 sebaceous glands per subject.

Second, follicular duct cells form deeper multilayers thanare seen in freshly isolated material. In the absence ofsaggital sections taken a fixed distance from the sebaceousduct entry into the follicular duct, this latter findingcannot be rigorously quantified. But repeated randomsections invariably confirmed this subjective judgement.

Fig. 1C illustrates a sebaceous gland maintained for 7days in medium supplemented with 3 ,UM-testosterone. It

shows that testosterone does not markedly alter thearchitecture that develops over a week's maintenance.There is a multilayered, poorly differentiated sebocytepopulation of between five and eight layers, and luminalkeratinization has developed. The follicular duct appearsto have keratinized to the same degree as follicular ductsmaintained in the absence of testosterone, which ismarkedly greater than in freshly isolated glands. Immuno-

128 jf. Ridden et ah

Fig. 2. The immunohistochemistry of fresh sebaceous glandsand glands maintained for 7 days under different conditions,using a polyclonal keratin antibody to keratin subunits 60, 58,56, 52, 51 and 48(XlO3)Mr. A. Illustrated is a typical sectionof a freshly isolated gland stained with a polyclonal antibodyto keratin subunits 60, 58, 56, 52, 51 and 48(XlO3)Mr. Thebrown stain is where the antibody has bound. Heavy stainingis seen around the duct (d) which is filled with keratinousmaterial. In the acini there is a clear boundary between theme that stain deeply for keratin and the non-staining area ofpdc and uc. This section is representative of n = 5 subjects; 5sebaceous glands per subject. B. Illustrated is a typicalsection of'a freshly isolated gland stained with non-immunerabbit sera, which acts as a control. No non-specific stainingcan be seen in the gland, but there is staining of the collagensurrounding the gland, which is seen when the sections arestained with both specific antibody and non-immune serum.This section is representative of w=S subjects; 5 sebaceousglands per subject. C. Illustrated is a typical section of agland maintained for 7 days on defined medium in thepresence of testosterone and stained with a polyclonalantibody to keratin subunits 60, 58, 56, 52, 51 and48(XlO )Mr. The brown stain is where the antibody hasbound. The pattern of staining has changed from that offresh gland. We now see heavy staining of both ms and theperipheral ppde. This section is representative of n = 5subjects; 5 sebaceous glands per subject. D. Illustrated is atypical section of a sebaceous gland maintained on definedmedium in the presence of both testosterone and \3-cisretinoic acid and stained with a polyclonal antibody to keratinsubunits 60, 58, 56, 52, 51 and 48(XlO3)A?r. The brownstain is where the antibody has bound. The pattern ofstaining has changed from that of fresh gland. We now havestaining of both me and pdc. This section is representative of« = 5 subjects, 5 sebaceous glands per subject.

histochemical experiments with antibody to keratin sub-units, 60, 58, 56, 52, 51 and 48(x 103)Mr (not illustrated)and control immunohistochemical experiments with non-immune rabbit sera (not illustrated) confirm this askeratin.

Fig. ID illustrates a sebaceous gland and follicularduct that have been maintained in the presence of both3/iM-testosterone and 1 fiM-13-cis retinoic acid. It isshown at a higher magnification than in Fig. 1A-Cbecause the gland has shrunk (see Table 3, below).Fig. ID shows that the addition' of 13-czs retinoic acideffects two significant change: (1) the architecture of thefreshly isolated sebaceous lobule has been maintained, inthat there is a histological progression from peripheralundifferentiated cells to partially differentiated cells andto central mature sebocytes; and (2) there is no luminalkeratinization.

Fig. 2D shows, however, that the PDC and UC that infreshly isolated glands did not react with the antibody tokeratin subunits, 60, 58, 56, 52, 51 and 48(X 103)yWr, nowdo so.

Although not illustrated in Fig. ID, other sectionsshow that the sebaceous duct is keratinized, but to nogreater extent than that of freshly isolated glands.

Au to radiographySebocytes. Fig. 3 illustrates an autoradiograph of fresh

glands (A) and glands maintained for 7 days underdifferent conditions (B, C and D).

Fresh glands (Fig. 3A) show heavy graining at theperiphery of the sebaceous acini, indicating that theundifferentiated sebocytes are actively incorporatingtritiated thymidine in vitro, which is the same pattern ofincorporation as is seen in vivo (Epstein and Epstein,1966; Plewigand Christophers, 1974). Partially differen-tiated or mature sebocytes did not incorporate [methyl-3H]thymidine in vitro, which is also the same pattern ofincorporation as is seen in vivo (Epstein and Epstein,1966; Plewig and Christophers, 1974).

Glands maintained for 7 days (Fig. 3B) still showheavy graining at the periphery, which indicates that theundifferentiated cell population is the only dividing cellpopulation. This same pattern is also observed whenglands are maintained in the presence of testosterone(Fig. 3C).

Glands maintained in the presence of both 3 fiu-testosterone and 1 fiM-13-cis retinoic acid (Fig. 3D) alsoshow graining but to a much lesser degree. No staining isseen centrally and only a few grains appear at theperiphery. This finding indicates that peripheral cells arestill the only dividing cells and that there has beenconsiderable inhibition of DNA synthesis by 13-cisretinoic acid.

Sebaceous duct cells. Approximately one fifth of allautoradiographic sections cut through sebaceous duct. Itwas noted that sebaceous duct in freshly isolated glandsshowed an average of three cells per duct that incorpor-ated tritiated thymidine (a total of 10 ducts seen; notillustrated). This confirms the findings of Epstein andEpstein (1966) that sebaceous duct cells in vivo incorpor-ate [we//2;y/-3H]thymidine. Sebaceous ducts in glandsmaintained for a week with and without 3 |UM-testosteronedemonstrated an average of 1.95 and 2 cells per duct thatincorporated [methyl- H]thymidine (20 and 14 ductsseen, respectively). Sebaceous ducts in glands main-tained for a week in 3/iM-testosterone and 1 [XM-13-cisretinoic acid demonstrated only 0.375 cell per duct thatincorporated tritiated thymidine (8 ducts seen).

Fibroblasts. Human sebaceous glands isolated byshearing contain at least three different cell types;sebocytes, duct cells and fibroblasts. Fibroblasts in vitrodid not incorporate [methyl-3H]thymidine, and thiswould not have been expected as the glands are main-tained on serum-free medium. Control experiments haveshown that primary fibroblast outgrowth on standardplastic Petri dishes or Primaria cell culture flasks does notoccur in our hands from human sebaceous glands in thismedium.

Electron microscopyFig. 4 shows transmission electron micrographs of iso-lated sebaceous glands after maintenance under differentconditions.

Fig. 4A illustrates a section of a freshly isolated gland.Two to three layers of peripheral undifferentiated cells(UDC) can be seen, as can the more central partiallydifferentiated cells (PDC), with their large lipid-contain-

Sebaceous gland organ maintenance 129

ing vacuoles. The UDC are loosely associated and veryfew desmosomes can be seen.

Fig. 4B shows the periphery of a sebaceous glandmaintained for 7 days. The smooth endoplasmic reticu-lum has acquired a vascular appearance. A partiallydifferentiated sebocyte can be seen with a high content oflipid droplets. These partially differentiated cells can beseen by light microscopy, randomly spaced throughout

the multilayered poorly partially differentiated cells(PPDC). A high degree of cell association of peripheralcells can be seen from the lack of intercellular spaces andthe high proportion of desmosomes (DE), unlike theloosely packed appearance of fresh peripheral cells.

Fig. 4C shows the periphery of a sebaceous glandmaintained for 7 days in the presence of 3 /iM-testoster-one. The peripheral cells have acquired a large number of

130 Jf. Ridden et al.

Fig. 3. The autoradiographic pattern of [methyl-3H]thymidine incorporation by fresh sebaceous glands andglands maintained for 7 days under different conditions.A. Illustrated is an autoradiograph of a freshly isolatedhuman sebaceous gland (X70) that has been exposed to[we//iy/-3H]thymidine and then autoradiographed, asdescribed in Materials and methods. The in vivo pattern ofthymidine incorporation was retained. This autoradiograph isrepresentative of n=5 subjects, 5 sebaceous glands persubject. B. Illustrated is an autoradiograph of a sebaceousgland (XlOO) maintained for 7 days in defined medium, thenexposed to [methyl-3H]thymidine. The in vivo pattern ofincorporation was retained. This autoradiograph isrepresentative of w=5 subjects, 5 sebaceous gland per subject.C. Illustrated is an autoradiograph of a sebaceous gland(X70) maintained for 7 days in defined mediumsupplemented with 3 ;UM-testosterone, followed by exposure to[wei/jy/-3H]thymidine. The in vivo pattern of thymidineincorporation was retained. This autoradiograph isrepresentative of w = 5 subjects, 5 sebaceous glands persubject. D. Illustrated is an autoradiograph of a sebaceousgland (X12S) maintained for 7 days in defined mediumsupplemented with both 3 jUM-testosterone and 1 [iM-13-cisretinoic acid following exposure to [methyl-3H]thymidine.Fewer grains (g) can be seen. However, what grains there areappear peripherally. Thus although the incorporation of[me</f>>/-3H]thymidine is less, the in vivo pattern ofincorporation has been maintained to a large degree. Thisautoradiograph is representative of «=S subjects, 5 sebaceousglands per subject.

keratin filaments (K). A greater degree of cell associationhas occurred as can be seen from the lack of intercellularspaces and the high proportion of desmosomes (DE).The greater cell association and increased keratin contentindicate that the peripheral cells have lost normal UDCand PDC morphology.

Fig. 4D shows the periphery of a sebaceous glandmaintained in medium supplemented with both 3 /UM-testosterone and 1 (iM-13-cis retinoic acid. The undiffer-entiated cells at the periphery are flatter than fresh.However, the peripheral UDC are still two layers deep,as is generally seen in fresh gland. Both partially differen-tiated and mature sebocytes are present. Thus 13-cisretinoic acid seems to be maintaining the ultrastructure ofthe sebocytes.

LipogenesisWe initially compared 2mM-[U-14C]glucose and 2mM-[U-14C]acetate as substrates for in vitro lipogenesis byisolated sebaceous glands. The incorporation rate from[U-14C]glucose was 212±45pmol per gland h~ (mean±S.E.M., n=5 subjects). This compares well with ourprevious reports of 39.7±3.7 (Kealey et al. 1986) and114.8±22.3 (Cassidy e« a/. 1986). However, when2m\i-[U-14C]acetate was used as substrate the rate of incorpor-ation was 569.6±66.4 pmol per gland h~ (mean±S.E.M.,«=12 subjects), which indicates that acetate may be apreferable substrate for lipogenesis in the human seb-aceous gland. Middleton et al. (1988) have reported a rateof 1003 ± 141 pmol per gland h"1 for lipogenesis using8 mM-acetate in human sebaceous glands isolated byshearing, but they studied facial glands while we studied

thoracic glands, and facial glands may secrete moresebum than thoracic glands.

Control experiments (w = 3) demonstrated that the rateof lipid synthesis is similar at both 2mM- and 4 mM-acetate; 2mM-[U-14C]sodium acetate was used in allsubsequent lipogenesis experiments. The rate at whichlipids were synthesized was linear over a 12-h period; theincubation period adopted routinely was 3 h.

Fig. 5 shows the effect of 13-cis retinoic acid concen-tration on the rate of lipogenesis after 7 days' mainten-ance. The effect was dose-dependent and effective atconcentrations as low as 10 nM. 13-cz's retinoic acid was,therefore, routinely used at a concentration of 1 fiM, as nofurther significant reduction in lipid production wasobserved at IO^M. 13-cis retinoic acid is known to becytotoxic at concentrations of 100 (iM and above (Kubiluset al. 1981). The serum concentration of 13-cis retinoicacid during acne therapy is between 1 (iM and 100 nM(Colburn et al. 1985).

Fig. 6 shows the effect of testosterone concentration onthe rate of lipogenesis after 7 days' maintenance. Testos-terone stimulates lipogenesis significantly at concen-trations as low as 1 fM and it would seem that thisstimulation was maximal as there is no further significantincrease in lipogenesis at greater testosterone concen-trations. We used 3 |UM-testosterone to be assured ofmaximal stimulation in our model. The serum concen-tration of testosterone in normal adult males variesbetween 11.3 and 34.5 nM (Sutton et al. 1973; Wall et al.1973), of which approximately between 1 and 2 % is free.This physiological concentration of testosterone of ap-proximately 0.1 nM is two orders of magnitude higherthan the concentration of 1 fM, which we have found togive a near maximum response, in vitro. This discrep-ancy might be accounted for by the in vivo circulation ofoestrogen, which counteracts the effect of testosterone onsebaceous glands (Strauss et al. 1962), and so mayrequire the higher in vivo testosterone concentration.

Fig. 7 shows that, on organ maintenance, there is aprogressive fall in lipogenesis over 14 days. Fig. 6 alsoshows that 3 ,UM-testosterone significantly reduced therate of loss of lipogenesis over 14 days, while 1 fXM-13-cisretinoic acid significantly accelerated it. Control exper-iments showed that this retinoic acid inhibition of lipidsynthesis was to a large extent recoverable. Glandsmaintained for 14 days in the presence of both 1 |UM-13-mretinoic acid and 3 ^tM-testosterone decreased the rate oflipid synthesis from 596.6+56.6 to 59.4±20 pmol perglandh"1 (mean±S.E.M.; n=4 subjects). However, par-allel experiments in which glands were maintained in thepresence of \3-cis retinoic acid and testosterone for 7days, followed by a further 7 days in the presence oftestosterone only, gave a rate of 146.8±31.9 (mean±S.E.M.; «=4 subjects). This is significantly higher(P^O.01) as determined by Student's f-test.

Table 1 shows that the pattern of lipogenesis after 7days' maintenance is very similar to that seen in freshlyisolated glands. The only important differences occur inthe rates of cholesterol and squalene synthesis. Whenglands are maintained in the absence of either testoster-one or 13-cis retinoic acid, the percentage of total lipid

Sebaceous gland organ maintenance 131

-r > ppdc

synthesized that is accounted for by squalene falls from28.9% to 3.2%, while the percentage that is attributableto cholesterol rises from 1.5 % to 42.3 %. These changesare reversed by maintenance with 3 /iM-testosterone,while this reversal is itself largely reversed by the furtheraddition of lftM-13-m retinoic acid to the testosterone-containing medium.

As these changes correlate well with the rate of totallipogenesis, and as lipogenesis is NADPH-dependent,and as the conversion of squalene to cholesterol is alsoNADPH-dependent, these findings can be explainedthus: when total lipogenesis is high, NADPH is preferen-tially oxidised by that lipogenesis, leaving very little as acoenzyme for the conversion of squalene into cholesterol.

132 jf. Ridden et al.

Fig. 4. The electron micrographic appearance of freshsebaceous glands and glands maintained for 7 days underdifferent conditions. A. Illustrated is a section of a freshlyisolated gland (X2200). Two to three layers of peripheralundifferentiated cells (udc) can be seen, as can the morecentral differentiating cells (pdc); 1, lipid vacuole. Thiselectron micrograph is representative of n = 5 subjects, 5sebaceous glands per subject. B. Illustrated is the peripheryof a sebaceous gland maintained for 7 days (X2200). Theoutermost peripheral undifferentiated cells (udc) haveacquired a high degree of cell-cell contact as judged by thelack of intercellular spaces and the large number ofdesmosomes and show a degree of disorganisation unseen infresh glands. Poorly partially differentiated cells can be seenmore centrally with their elaborate network of endoplasmicreticulum (er). This electron micrograph is representative of« = 5 subjects, 5 sebaceous glands per subject. C. Illustratedis the periphery of a sebaceous gland maintained for 7 days inthe presence of 3^M-testosterone (X4000). It has a verysimilar appearance to that of glands maintained in the absenceof testosterone, the peripheral cells have acquired a highdegree of cell-cell contact as judged by the lack ofintercellular spaces and the large number of desmosomes (de)and there is a high degree of disorganisation. This electronmicrograph is representative of w = 5 subjects, 5 sebaceousglands per subject. D. Illustrated is the periphery of asebaceous gland maintained in the presence of both 3 fiM-testosterone and 1/UM-13-CM retinoic acid (X2200). Theundifferentiated cells at the periphery (udc) show a moreflattened appearance than the udc of fresh glands. Bothpartially differentiated and mature sebocytes are present.This electron micrograph is representative of w = 5 subjects, 5sebaceous glands per subject.

400

•a 360c

320

a. 2 8 0

c

1 240oa.8 200cB2 160

U 120

I...; .....i ......I i i , . i .

10- 10" 10"6 10"Log dose testosterone (M)

Fig. 6. The effect of increasing the concentration oftestosterone in the maintenance medium on the rate oflipogenesis of isolated human sebaceous glands maintained for7 days. w = 3 subjects, 3 glands from each subject per timepoint±S.E.M.. Glands were maintained for 7 days, and theirrate of lipogenesis determined as described in Materials andmethods. ** indicates that the rate of lipogenesis wassignificantly different (P=£0.05) from glands maintained inthe absence of testosterone, as determined by Student's /-test.

•i"" 300n

| 200-a.co

Oe-8 100-c

0 J

10" 10" 10" 10"Concentration of 13-m retinoic acid (M)

Fig. 5. The effect of increasing the concentration of 13-CMretinoic acid in the maintenance medium on the rate oflipogenesis of isolated human sebaceous glands maintained for7 days. w=4 subjects, 3 glands from each subject per timepoint±S.E.M. The sebaceous glands were maintained asdescribed in Materials and methods. The rate of lipogenesiswas related to the dose of 13-CM retinoic acid wtih asignificance of P^O.01 using the Kendall ranked correlationtest (w=5 subjects).

3 ;(M-testosteroneNo additions3 /(M-testosterone +1 fiM-13-cis retinoicacid

n—•—r4 6

Days maintained

Fig. 7. The rate of lipogenesis as a percentage of the controlvalue, by isolated human sebaceous glands maintained forvarious lengths of time under different conditions. «=14subjects. Glands were maintained as described in Materialsand methods. * indicates a signficant difference from controlglands (day 0) (*/>s£0.01 and **Ps=0.001). f indicates adifference from glands maintained in the absence oftestosterone at the same point ( tP^0.05 and f t ^=0.01).| indicates that there is a significant difference betweenglands maintained under a given condition (fP^O.05,1HJ/>s£0.01 and fifflP^O.OOl). All P values were determinedby Student's /-test («=5 subjects).

Sebaceous gland organ maintenance 133

Table 1. [U-C] acetate uptake into lipids by the sebaceous gland

Total lipids

Squalene

Wax esters+cholesterolesters

Triglycerides

Free fatty acids

1,3-Diglyceride

1,2-Diglyceride

Cholesterol

Monoglyceride

Lysolecithin

Sphingomyelin

Phosphatidyl choline

Phosphatidyl serine+inositol+ethanolamine

Phosphatidic acid

Cardiolipin

Freshly isolated

569.2+60.3

163.7±4.8(28.9%)

36.6+1.5(6.47 %)

129.2+1.2(22.8%)

7.2+0.1(1.24%)

7.0+0.05(0.5 %)

10.1+0.05(1.8%)

8.5+0.05(1.5%)

13.1 + 1.2(2.3%)

131.9+12.1(23.3%)

20.4+2.2(3.6%)

12.5 + 1.1(2.2%)

19.8+0.2(3.5%)

7.8±0.06(1.4%)

9.6±0.02(1.7%)

No additions

94.2±21.2***

3.0±0.69***(3.2%)

3.0±0.8(0.4%)

36.4±8.4*(38.8%)

0.8±0.2(0.9%)

0.3±0.1(0.3%)

0.4±0.1(0.4%)

39.812.0***(42.3%)

0.05±0.01(0.05 %)

5.4+1.2(5.7%)

1.8+0.2(1.9%)

0.310.07(0.2%)

0.910.2(0.9%)

0.610.2(0.6%)

3.910.5(4.2%)

Conditions of incubation

3 jUM-testosterone

266.3+48.2***ftt62.3111.2****

(23.3%)

0.2+0.03(0.1%)

107.6119.3*(40.3%)

32.2+5.8****tttt(12.1%)

0.510.1(0.2%)

0.8+0.1(0.3%)

1.110.2**tttt(0.4%)

0.5+0.1(0.2%)

29.3+5.2(11.0%)

1.0+0.2(0.4%)

0.410.07(0.7%)

26.9+4.8(10.1%)

1.010.2(0.4%)

2.910.5(1.1%)

1 /iM-13-c/i retinoic acid+3 ^M-testosterone

49.3121.3***11

8.513.5(14.4%)

0.4+0.1(0.6%)

3.8+1.5***+tH(6.4%)

1.2+0.5H(2.0%)

0.6+0.2(1.0%)

0.910.4(1.5%)

27.3 + 11.1***11(46.1 %)

0.9+0.3(1.5%)

10.914.4(18.4%)

1.710.6(2.8%)

0.4+0.07(0.7%)

0.05+0.02(0.1%)

1.7+0.6(2.8%)

1.010.4(1.8%)

Values are pmol per gland h ' (meanis.E.M., n=S subjects).Glands were maintained as described in Materials and methods. * Significance as determined by Student's /-test, relative to freshly isolated

glands (*/>«0.05, •*P=S0.01 and ***/>=S0.001). f Significance relative to glands maintained for 7 days with no additions (tP^O.05, +t^=S0.00 5 and t f f t ^ 0 - 0 0 1 ) - H A significant difference with respect to 3 ^M-testosterone (1[Ps=0.01).

Table 2. [U-14C]leucine (x\—l2 subjects) and /"methyl-3H]thymidine (n=12 subjects) uptake into the protein andDNA of the sebaceous gland, under various conditions of maintenance (mean±S.E.M.)

Condition of maintenanceThymidine incorporation(fmolmg"1 h~') (4 days)

Thymidine incorporation.-it,-'(fmolmg~'h~') (7 days)

[U-'4C]leucine incorporation(pmolmg-1 h~') (7 days)

Control (fresh)No additions3 /JM-testosterone3 //M-testosterone+1 /4M-1

retinoic acid

148.5 + 11.7127.5114.9147.919.3110.4+13.9*

173.68+45.61159.01+49.58112.38+39.8

28.4+8.l**f

170.3115.2175.9+20.3158.1112.9

144.65 + 14.7

Glands were maintained and assayed as described in Materials and methods. * A significant difference with regard to fresh glands (*PS0.05,**P^0.001). f Significance between glands maintained 4 and 7 days of the same condition (+PS0.005). As determined by Student's Mest.

When lipogenesis is inhibited, either by maintenance intestosterone-free medium or by 13-cis retinoic acid, thisprocess is reversed.

DNA and protein synthesis

Table 2 shows that the rates of DNA synthesis are

unchanged over 4 or 7 days' maintenance when comparedwith those of freshly isolated glands. Furthermore,maintenance in medium supplemented with 3 |UM-testos-terone does not affect the rate. However, supplemen-tation of the medium with 1 jj.M-13-cis retinoic acid effectsa statistically significant fall in the rate of DNA synthesisafter 4 days' maintenance, which by 7 days has fallen to

134 Jf. Ridden et al.

Table 3. Total DNA, protein and wet weight of glandsmaintained under different conditions (n=12 subjects;

mean + S.E.M.)

Condition ofmaintenance

Fresh3 jUM-testosteroneNo additions1 |UM-13 cis

retinoic acid +3 ^jM-testosterone

DNA content(Hg per gland)

2.19±0.2S2.46±0.272.03±0.211.45 ±0.18*

Protein content(jig per gland)

29.5±3.823.9±2.825.9±2.819.7±2.0*

Wet weight(mg)

3.1+0.232.8±0.252.4±0.18*

1.94±0.2«*

The glands were maintained, and their DNA and protein contentand wet weights were determined as described in Materials andmethods, with additions noted in the table.

* Level of significance with regard to fresh glands• 0 0 as determined by Student's Mest.

only 25 % of that of glands maintained in mediumsupplemented with 3 jUM-testosterone alone. Table 3shows that 13-m retinoic acid effects a significant fall inglandular DNA content following 7 days' maintenance.

Table 2 also shows that neither ljUM-13-m retinoicacid nor 3 /iM-testosterone affects the rate of proteinsynthesis in maintained glands, and that maintenanceitself does not alter the rate of protein synthesis whencompared with freshly isolated glands. However, Table 3shows that 13-m retinoic acid decreases total glandularprotein content after a week's maintenance, which mayreflect a retinoid effect on protein breakdown.

Table 3 shows that fresh glands and glands maintainedin the absence of \3-cis retinoic acid contain similarquantities of DNA and protein. However, glands main-tained in the presence of 13-m retinoic acid showsignificantly less DNA, which indicates that there arefewer cells present. This is compatible with the signifi-cantly lower wet weight.

Discussion

We have shown that isolated human sebaceous glands canbe maintained for up to 14 days with the retention ofviability as determined by light and electron microscopy,the rate of DNA and protein synthesis, the autoradio-graphic pattern of [we^>'/-3H]tnymidine incorporation,DNA and protein contents, wet weights and the rates andpatterns of lipogenesis. Furthermore, we show in vitroeffects of 13-cw retinoic acid and testosterone that indi-cate that the maintained isolated human sebaceous glandmay be a good model for exploring their molecular modesof action. Nevertheless, isolated human sebaceous glandsthat are maintained as we have described do deviate fromthe situation in vivo.

It appears that the rate of formation of new sebocytes isunchanged on maintenance, as is shown by the rates andautoradiographic patterns of [tnethyl- HJthymidine in-corporation, and as is also shown by the rates of [U-14C]leucine incorporation. However, it appears that thenew sebocytes do not differentiate fully, as is shown: (1)microscopically by the expression of squamous meta-

plasia, by the scarcity of lipid vacuoles and by thedevelopment of luminal keratinization; (2) biochemi-cally, by the progressive loss of lipogenesis; and (3) bythe development of a peripheral sebocyte keratin stainingpattern that is normally restricted to mature sebocytes.

However, it appears that those sebocytes that werealready committed to terminal differentiation at themoment that the glands were isolated, continue todifferentiate normally. This accounts for the microscopicappearance of maintained glands, which show bothperipheral, undifferentiated or poorly differentiated cellsand a small central population of fully differentiated cells.This also accounts for the rate of loss of lipogenesis onmaintenance. In vivo studies using injected [me thy I -3H]thymidine have shown that the human sebocyte life-span lies between 7 and 28 days (Plewig et al. 1974;Epstein and Epstein, 1966), and our in vitro finding thatlipogenesis has halved by 6 days is compatible with thesuggestion that in vitro lipogenesis is the product largelyof sebocytes that were committed to terminal differen-tiation on isolation.

Testosterone does not alter the rates of new cellformation in vitro as shown by the rates and patterns of[mef/ry/-3H]thymidine and [U-14C]leucine incorpor-ation. Nor does it markedly promote full differentiationas determined microscopically. However, it does helpmaintain biochemical differentiation as shown by itseffect in reducing the rate of loss of lipogenesis in vitro.

13-m retinoic acid, however, seems in vitro to mimicits in vivo effects. Thus 13-cis retinoic acid causes afourfold reduction in the rates of both DNA synthesisand lipogenesis, and it prevents the development ofluminal keratinization. These effects are compatible withthe in vivo clinical effects of 13-m retinoic acid (Peck etal. 1982). Furthermore, they explain the shrinkage ofsebaceous glands seen in vivo following therapy with13-m retinoic acid, because the size of a holocrinesecreting gland will depend, in part, on the rate of celldivision, and we have shown that 13-m retinoic aciddecreases DNA synthesis and the autoradiographic up-take of [me£/z;y/-3H]thymidine.

In vitro, 13-cis retinoic acid preserves sebaceous glandarchitecture better than does maintenance either in itsabsence, or in the presence of testosterone alone, and thiscan be attributed to its inhibition of new, undifferentiatedsebocyte growth. The rate of protein synthesis in vitroappears to be unchanged by a week's maintenance in thepresence of 13-m retinoic acid, but this can be attributedto the cells that were already committed to terminaldifferentiation on isolation.

In conclusion, we have shown that human sebaceousglands retain their viability on maintenance, and we haveshown that they retain the rates of cell division shownwhen freshly isolated. Although sebocytes that werecommitted to differentiation on isolation apparently pro-gress normally through their life cycle, newly synthesisedcells do not differentiate fully on maintenance, and this isnot reversed by testosterone. Nonetheless, we haveshown that the glands' in vitro response to 13-m retinoicacid is very similar to their clinical in vivo response. Thisprovides the first demonstration that 13-m retinoic acid

Sebaceous gland organ maintenance 135

acts on human sebaceous glands directly, and it providesa useful model for studying the effect of this agent onhuman sebocyte turnover.

We thank Mrs Ysanne Smart for the electron microscopy,Mrs Christine C. Ridden for her assistance and helpful dis-cussion, and the surgeons at the Bristol Royal Infirmary andJohn Radcliffe Hospital for supplying us with skin. We alsothank Miss B. Disbrey at the Department of Histochemistry forthe sectioning for immunohistochemistry.

This work was supported by the MRC. T.K. is a WellcomeSenior Clinical Research Fellow. We thank Hoffman La Rochefor the gift of \3-cis retinoic acid.

References

BENBROOK, D., LERNHARDT, E. AND PFAHL, M. (1988). A new

retinoic acid receptor identified from a hepatocellular carcinoma.Mature, Ijond. 333, 669-672.

BLIGH, E. G. AND DYER, W. J. (1959). A rapid method of total lipidextraction and purification. Can. J. Biochem. Physiol. 37, 911-917.

BOLLAG, W. (1983). Vitamin A a-retinoids: from nutrition topharmacotherapy in dermatology and oncology. Lancet i, 860-863.

BRADFORD, M. M. (1976). A rapid and sensitive method for thequantitation of microgram quantities of protein utilizing theprinciple of protein-dye binding. Analyt. Biochem. 72, 248-254.

CASSIDY, D. M., LEE, C. M., LAKER, M. F. AND KEALEY, T.

(1986). Lipogenesis in isolated human sebaceous glands. FEBSLett. 200, 173-178.

COLBURN, W. A., VANE, F. M., BUGGE, C. J., CARTER, D. E. AND

EHMANN, C. W. (1985). Pharmokinetics of 14C-isotretinoin inhealthy volunteers and volunteers with billiary t-tube drainage.DnigMetab. Dispos. 3, 327-332.

COOPER, M. F., THODY, A. J. AND SHUSTER, S. (1974). Hormonal

regulation of cutanous lipogenesis: Effects of hypophosectomy,posterior-hypophysectomy and cr-melanocyte-stimulating hormonetreatment. Biochim. biophys. Ada 360, 193-204.

CULLING, F. C. et al. (1980). The effect of various fixatives andtrypsin digestion upon the staining of routine paraffin-embeddedsections by the peroxidase-antiperoxidase and immunofluorescencetechnique. .7. Histochem. 3, 10-16.

EPSTEIN, E. H. AND EPSTEIN, W. L. (1966). New cell formation inhuman sebaceous glands, j ' . invest. Derm. 46, 453-458.

ESCRIBANO, L. M. et al. (1987). Endogenous peroxidase activity inhuman cutaneous and adenoidal mast cells. J. Histochem.Cytochem. 35, 213-217, 1987.

FELL, H. B. (1953). Recent advances in organ culture. Sci. Prog.,Oxf. 162, 212-238.

FlSZER-SZAFARJ, B., SZAFARJ, D. AND DE MURILLO (1981). Ageneral, fast and sensitive micromethod for DNA determination.Analyt. Biochem. 110, 165-170.

FRANKE, W. W., WEBER, K., OSBORN, M., SCHMID, E. AND

FREUDENSTEIN, C. (1978). Antibody to prekeratin. Expl Cell Res.116, 429-445.

GIGUERE, V., ONG, E. S., SEGUI, P. AND EVANS, R. M. (1987).

Identification of a receptor for the morphogen retinoic acid.Nature, Loud. 330, 624-629.

GOMEZ, E. C. (1982). Action of isotretinoin and etretinate on thepilosebaceous unit. J'. Am. Acad. Derm. 6, 746-750.

GREEN, S. AND CHAMBON, P. (1986). A superfamily of potentiallyoncogenic hormone receptors. Nature, Land. 324, 615—617.

JONES, D. H., KING, K., MILLER, A. J. AND CUNLIFFE, W. J.

(1983). A dose-response study of 13-cis-retinoic acid in acnevulgaris. Br.J. Derm. 108, 333-345.

KEALEY, T., LEE, C. M., THODY, A. J. AND COAKER, T. (1986).

The isolation of human sebaceous glands and apocrine glandssweat glands by shearing. Br.J. Derm. 114, 181-188.

KRAEMER, K. H., D I GIOVANNA, J. J., MOSHELL, A. N., TARONE,

R. E. AND PECK, G. L. (1988). New Engl.J. Med. 318,1633-1637.

KREBS, H. A. AND HENSELEIT (1932). Untersuchungen uber dieHarnstoffbilding im Tierkorper. Hoppe-Sevler's Z. phvsiol. Chem.210, 143-148.

KUBILUS J., RAND, R. AND BADEN, H. P. (1981). Effects of retinoicacid and other retinoids on the growth and differentiation of 3T3supported human keratinocytes. In Vitm 9, 786-795.

LAROSA, G. J. AND GUDAS, L. J. (1988a). An early effect of retinoicacid: Cloning of an mRNA (Era-1) exhibiting rapid and proteinsynthesis-independent induction during teratocarcinoma stem celldifferentiation. Proc. imtn. Acad. Sci. U.S.A. 85, 329-333.

LAROSA, G. J. AND GUDAS, L. J. (1988&). Early retinoic acid-induced F9 teratocarcinoma stem cell gene ERA-1: Alternatesplicing creates transcripts for a homeobox-containing protein andone lacking the homeobox. Molec. cell. Biol. 8, 3906-3917.

LLOYD, R. V., SCHMIDT, K., MCCOY, J. P. AND WILSON, J. P.

(1985). A rapid immunostaining method utilizing preformedantibody-avidin-biotin-peroxidase complexes. Am. J. din. Path. 83,636-639.

MlDDLETON, B . , BlRDI, I . , HEFFRON, M . AND MARSDEN, J. R.

(1988). The substrate determines the rate and the pattern ofneutral lipid synthesized by isolated human sebaceous glands.FEBS Lett. 231, 59-61.

PECK, G. L., OLSEN, T., BUTKUS, D., PANDYA, M., ARNAUD-

BATTANDIER, J., GROSS, E. G., WINDHORST, D. B. AND CHERIPKO,

J. (1982). Isotretinoin versus placebo in treatment of cystic acne.J. Am. Acad. Demi. 6, 735-745.

PECK, G. L., OLSEN, T., YODER, F., STRAUSS, J. S., DOWNING, D.

T., PANDYA, M., BUTKUS, D. AND ARNAUD-BATTANDIER, J.

(1979). Prolonged remission of cystic and conglobate acne with 13-cis-retinoic acid. New Eng.J. Med. 300, 329-333.

PETKOVICH, M., BRAND, N. J., KRUST, A. AND CHAMBON, P. (1987).

A human retinoic acid receptor which belongs to the family ofnuclear receptors. Nature, Ijond. 330, 444-450.

PLEWIG, G. AND CHRISTOPHERS, E. (1974). Renewal rate of humansebaceous glands. Ada derm. Stockh. 54, 177-182.

POCHI, E., STRAUSS, J. S. AND DOWNING, D. T. (1977). Skin

surface lipid composition, acne prepubertal development andurinary excretion of testosterone and 17 ketosteroid in children, j ' .invest. Denn. 69, 485-489.

SPORN, M. B., ROBERTS, A. B. AND GOODMAN, D. S. (1984). The

Retinoids, vol. 1-2. Academic Press, Orlando, FL.STEINERT, P. M. (1975). The extraction and characterization of

bovine epidermal tt-keratin. Biochem. J. 149, 39-48.STRAUSS, J. S., KLIGMAN, A. M. AND POCHI, P. E. (1962). Effect of

androgens and estrogens on human sebaceous glands. J. invest.Denn. 39, 139-155.

SUTTON, J. R., COLEMAN, M. J., CASEY, J. AND LAZARUS, L.(1973). Androgen response during physical exercise. Br. Med. J. 1,520-522.

THALLER, C. AND EICHELE, G. (1987). Identification and spatialdistribution of retinoids in the developing chick limb bud. Nature,Loud. 327, 625-628.

TROWELL, O. A. (1959). The culture of mature organs in syntheticmedium. Expl Cell Res. 16, 118-147.

WALL, J. R., ZIMMET, P. Z., JARRET, R. J., BAILES, M. AND

RAMAGE, C. M. (1973). Fall in plasma testosterone levels innormal male subjects in response to an oral glucose load. Lancet i,967-968.

WEST, D. C , SATTER, A. AND KUMAR, S. (1985). A simplified in

situ solubilization procedure for the determination of DNA andcell number in tissue cultured mammalian cells. Analvt. Biochem.147, 289-295.

ZELENT, A., KRUST, A., PETKOVICH, M., KASTNER, P. AND

CHAMBON, P. (1989). Cloning of murine cvand /3 retinoic acidreceptors and a novel receptor y predominantly expressed in skin.Nature, Land. 339, 714-717.

{Received 16 September 1989 -Accepted 27 September 1989)

136 J. Ridden et al.