Intermediate basal cells of the prostate: In vitro and in vivo characterization

The Prostate 55:206 ^218 (2003)

Intermediate Basal Cells of the Prostate: InVitroand InVivoCharacterization

Levi A. Garraway,1 Douglas Lin,1 Sabina Signoretti,1,2 David Waltregny,3

James Dilks,1 Nandita Bhattacharya,1 and Massimo Loda1,2*1DepartmentofMedical Oncology,Dana Farber Cancer Institute,HarvardMedical School, Boston,Massachusetts

2Departmentof Pathology,BrighamandWomen’sHospital,HarvardMedical School, Boston,Massachusetts3Metastasis Research Laboratory,UniversityHospital,Universityof Liege,Belgium

BACKGROUND. Progenitor cells within the prostate basal layer may play important rolesin differentiation and carcinogenesis; however, prostate stem cell populations remainuncharacterized.METHODS. Immunohistochemical and immunoblot analyses were used to characterizeprostate epithelial cells (PrEC), a commercially available prostate basal cell isolate.RESULTS. Proliferating PrECs exhibited immunophenotypic characteristics most consistentwith basal cells, but during senescence PrECs up-regulated androgen receptor (AR) mRNA,p27, and low-molecular-weight cytokeratin (LMWCK) expression, suggestive of partialdifferentiation. PrECs also stained strongly for involucrin, which marked a subset of inter-mediate prostate basal cells invivo. Basal hyperplasia consistingof involucrin-positive cellswasprevalent in prostate tissue from androgen-ablated patients, and formed epithelial clustersflanked by involucrin-negative basal and luminalmonolayers. Cultivation of PrECs onmatrigeltogetherwith androgen-treated stromal conditionedmedia resulted in dense aggregates,with aperipheral rim of basal-like cells expressing p63 and basal cytokeratins.CONCLUSIONS. PrEC represents an epithelial population whose basal characteristics aremodified in response to matrigel, stromal factors, and senescence, consistent with a transientamplifying population. These cells may derive from a previously unrecognized, involucrin-positive subset present in vivo. Prostate 55: 206–218, 2003. # 2003 Wiley-Liss, Inc.

KEY WORDS: prostate; PrEC; stem cells; p63; differentiation; basal cells; involucrin

INTRODUCTION

Normal prostate epithelium contains three majorcell types in two morphologically defined compart-ments. The basal compartment consists of both basaland neuroendocrine cells, whereas the luminal com-partmentharbors secretory cells. Secretory cells expressthe androgen receptor (AR) [1] and prostate-specificantigen (PSA); they are highly differentiated, exhibita low mitotic index [2], and require continuous androgen exposure for survival. In contrast, basal cellspossess a relatively high mitotic index [3] express lowlevels of AR protein [1,4], and exhibit androgen in-dependence, as evidenced by their capacity to survivefollowing androgen withdrawal [5] and proliferateupon re-exposure to androgens [6,7]. These properties

Levi A. Garraway and Douglas Lin contributed equally to this work.

Grant sponsor: CaPCURE and Barr-Weaver Awards; Grant sponsor:National Cancer Institute; Grant number: R01-CA81755; Grantsponsor: Novartis (to ML); Grant sponsor: US Department ofDefense (to ML); Grant number: DAMD 17-98-8574; Grant sponsor:Massachusetts Department of Public Health (to LAG); Grantsponsor: Hershey Prostate Cancer/Survivor Walk Award; Grantsponsor: Department of Defense (to SS); Grant number: DAMD17-01-1-0051.

*Correspondence to: Massimo Loda, Dana Farber Cancer InstituteD740, 44 Binney Street, Boston, MA 02115.E-mail: [email protected] 15 October 2002; Accepted 18 December 2002DOI 10.1002/pros.10244

� 2003 Wiley-Liss, Inc.

have led many investigators to postulate that the basalcompartment includes a pluripotent stem cell popula-tion capable of generating terminally differentiatedsecretory and neuroendocrine cells [8–10]. Identifica-tion and characterization of these progenitor cellswith-in the prostate remains an area of active investigation.

Although the ‘stem cell model’ for prostate dif-ferentiation has gained widespread acceptance, con-troversy exists concerning the exact mechanismsinvolved, particularly with regard to the compositionand function of the basal layer. Accordingly, threedistinct models for prostate organization may beenvisioned, as outlined in Figure 1. Models A and Binvoke the presence within the basal compartment of aslowly dividing, androgen-independent progenitorsub-population which gives rise to neuroendocrineand secretory cells [9]. The latter may arise eitherdirectly or via an intermediate, more rapidly pro-liferating secondary population. These transient-amplifying cells (TA cells), are postulated to residebetween the basal and luminal layers, and to lack thecell cycle inhibitor p27 [11]. By the conventional model(Fig. 1A), most of the basal layer constitutes a reservoirof progenitor cells, capable of differentiating into sec-retory cells under appropriate environmental stimuli.An alternativemodel (Fig. 1B) offers the possibility thatstem cells give rise to distinct, terminally differentiatedbasal, and secretory cell populations. This hypothesisimplies that most basal cells, though capable of pro-liferation, are themselves irreversibly differentiatedand, therefore, unable to generate the secretory layer. Athird model (Fig. 1C) postulates discrete lineages forthe two layers, such that progenitor subtypes unique

to both basal and secretory cells repopulate each com-partment separately. By this model, TA cells, if theyexist at all, are confined to a single lineage, and neitherlayer gives rise to the other.

Evidence supporting each model exists in the litera-ture. In the developing and adult prostate, cells withimmunohistochemical profiles intermediate betweenbasal and secretory epithelia have been described[3,12–14]. Moreover, basal cells generally retain theability to undergo metaplasia involving squamousdifferentiation, particularly following androgen with-drawal [15,16], estrogen addition [17], or radiation [18].In aggregate, these data lend credence to the idea thatprostate basal cells possess considerable plasticity withregard to cell fate determination, andmay function as aprogenitor reservoir for the luminal compartment. Onthe other hand, studies in animal models suggest theexistence of sub-populations in both epithelial layersthat resist castration-induced apoptosis and pro-liferate following androgen repletion [5,19,20]. Thesedata resemble observations from other tissue types,such as the lung, where distinct progenitor cells maygive rise to separate basal and secretory lineages[21]. Thus, the nature and extent of prostate epithelialsub-populations capable of regenerating the luminalcompartment remains unresolved.

Characterization of pluripotent prostate cell typesholds important implications for carcinogenesis andtumor progression. Most prostate tumors secrete PSAand display cytokeratin profiles similar to secretorycells [22]; therefore, to a first approximation thesecancers appear to arise from the transformation of lum-inal cells. However, if basal cells do in fact constitute

Fig. 1. Models forprostateepithelialdifferentiation.A:Theprostatebasalmonolayeriscomposedof stemcellsandtransient-amplifyingcells(TAcells; shownhereas adistinctlayer forclarity),whichare capable ofdifferentiatinginto secretoryepithelia (andneuroendocrine,NE, cells)under appropriate environment stimuli.B: Alternatively, stem cells give rise to distinct, terminally differentiatedbasal, neuroendocrine, andsecretorycellpopulations, thelatterviaTAcells.C:Athirdmodelpostulatesdiscretelineages for the twolayers, such thatprogenitor subtypesunique to bothbasal and secretorycells repopulate each compartment separately.By thismodel,TAcells, if theyexist at all, are confined to asinglelineage,andneither layergivesrise to theother.

Intermediate Basal Cells of the Prostate 207

luminal progenitors as outlined above, prostate neo-plasms can in principle arise from any of a varietyof precursors, including less differentiated secretorysub-populations or basal cell intermediates. In sup-port of this hypothesis, many advanced prostatetumors exhibit characteristics similar to basal cells ofthe normal prostate, such as androgen independent-growth and expression of genes normally restricted tothe basal compartment [8,23–26]. Therefore, androgen-independent prostate tumors must either derive fromless mature secretory precursors or undergo de-differentiation such that certain basal-specific genesare re-expressed as an adaptive response to androgenwithdrawal.

Experimental interrogation of the severalmodels forprostate differentiation and carcinogenesis wouldbenefit substantially from an in vitro basal cell system.Most in vitro studies of human prostate epitheliarequire direct access to surgical or biopsy specimens,which limits widespread accessibility and reproduci-bility. Commercial preparation of primary prostateepithelial cells, termed PrEC, has become available,which may facilitate in vitro studies by avoiding theneed for independent access to clinical samples. Thesecells grow in an androgen-independent fashion andhave been shownpreviously to possess several featuresof basal epithelia. We have characterized these cellsby immunohistochemical analyses and followingexposure to environmental factors that may influencedifferentiation in vivo. Our results show that PrECconstitutes a basal cell sub-population that may dif-ferentiate partially during senescence, or followingexposure to stromal factors and reconstituted basementmembrane. PrECs lack AR protein; this may precludethe acquisition of definitive luminal properties. How-ever, these cells also express involucrin, a marker ofsquamous epithelia [27], and as such they may re-present a newly recognized, seemingly hyperplasticsubtype distinct from the native basal layer in vivo.Culture of PrEC populations in vitro and in the absenceof androgen may, therefore, select a previouslyuncharacterized basal derivative useful for additionalstudies of prostate differentiation, hyperplasia, andneoplastic transformation.

MATERIALSANDMETHODS

Cell Culture and Prostatectomy Specimens

Human PrEC and human prostate stromal cells(PrSC)were obtained fromClonetics (SanDiego,CA) atearly passage (P2 or P3), and cultured in the prostateepithelial cell basal medium (PrEBM) or stromal cellbasal medium (SCBM) supplied by the company.PrEBM was supplemented with human epidermalgrowth factor, triiodothyronine, transferrin, epinephr-

ine, gentamicin sulfate, amphotericin B, bovine pitui-tary extract, bovine insulin, hydrocortisone, andretinoic acid additives provided by the manufacturer.Similarly, SCBM was supplemented with hFGF, insu-lin, fetal bovine serum (FBS), and antibiotic. PrEC andPrSC cellswere allowed to reach 80% confluence beforethey were sub-cultured. Immunohistochemistry wasperformed on cells from proliferating (P4 or P5) andsenescent PrEC (P9-P12; see below). Individual cellisolates obtained fromClonetics varied as towhen theyentered senescence, as noted; however, in separateexperiments the growth plateau was documented bydetailed growth curve analyses (not shown). All ex-periments were performed on at least two separatePrEC isolates, with similar results. LNCaP and DU-145prostate cancer cell lines (ATCC, Rockville, MD) weregrown in RPMI 1640 and DMEM, respectively, with10% FBS. Prostate and urothelial specimens wereretrieved from the files of the departments of Pathologyof the Brigham and Women’s Hospital and the BethIsrael Deaconess Medical Center (Boston, MA) andutilized for immunohistochemical experiments.

Immunohistochemistry

Immunostainingwas performed in tissue specimensand paraffin-embedded cell lines with the followingantibodies: 4A4, which recognizes all six p63 iso-types [29] (1:50 dilution); 34bE12, which binds high-molecular-weight cytokeratins (HMWCK) 5, 14, 1, and10 (1:50 dilution); anti-CK18 (1:100 dilution), 35bH11,which recognized CK8 (1:100 dilution); anti-p27(1:200 dilution), anti-AR (1:50 dilution), anti-involucrin(1:100 dilution), anti-loricrin (1:300 dilution), and anti-PSA (1:50 dilution). Section (5 mm)were depariffinized,rehydrated, and heated in 10 mM citrate buffer, pH 6.0(BioGenex, San Ramon, CA) in a 750 W microwaveoven for 15 min. Slides were cooled at room tempera-ture for 30 min. The antibodies were then applied atroom temperature for 2 hr in an automated stainer(Optimax Plus 2.0 bc, BioGenex). Detection steps wereperformedutilizing theMultiLink-HRPkit (BioGenex).Peroxidase activity was localized with 3,3-diamino-benzidine (DAB) or 3,3-diaminobenzidine-nickel chl-oride (DAB-NC). Standardized development timeperiods allowed accurate comparison of all samples.The sections were counterstained with hematoxylin,rehydrated andmounted formicroscopic examination.Substitution of the primary antibody with the phos-phate buffered saline (PBS) served as negative control.

Stromal Cell ConditionedMedium (SCM)andDifferentiation Experiments

SCMwas prepared by growing PrSC cells in 150 cm2

tissue culture flasks (Becton Dickinson Labware,

208 Garrawayet al.

Franklin Lakes, NJ) until they reached 100% conflu-ence. Themediawas then changed to SCBMcontaining10 nM dihidrotestosterone (DHT, Sigma ChemicalCompany, St. Louis, MO). After 48 hr, the resultingconditioned media were harvested, centrifuged (400gfor 10min), and filteredwith a 0.2 mm (very low proteinbinding) polyethersulfone filter (Corning Incorpo-rated, Corning,NY).A solution containing 30%stromalconditionedmedia in PREGMwas immediately addedto PrEC cells; the remaining conditioned media werestored at 48C for a maximum of 6 days.

For the differentiation experiments, PrEC wereplated onto untreated 6-well plates (Falcon-BectonDickinson, Lincoln Park, NJ) at a density of 106 cells/well, or onto 6-well plates containing reconstitutedbasement membrane inserts (matrigel, CollaborativeResearch, Bedford, MA). The next day (day 0), themedia were changed and the cells were cultured foran additional 6 days, with fresh media added every2 days. In other experiments, retinoic acid was omittedfrom PrEBM, or replaced with varying concentrationsof all-trans retinoic acid (ATRA, Sigma ChemicalCompany).

Protein Extraction andWesternBlotAnalysis

Cells grown on matrigel were recovered withMatriSperseTMCell Release Solution (BectonDickinsonLabware) according to the manufacturer’s directions.Cell extractswere prepared by addition of 3–5 volumesof lysis buffer (50 mM Tris-base, pH 7.5, 250 mMNaCl,0.1% Triton X-100, 1 mM EDTA, 50 mM NaF, 1 mMNa3VO4, 1 mM DTT) to a cell pellet with the followingprotease inhibitors: 0.1 mM PMSF, 1 mg/ml leupeptin,10 mg/ml soybean trypsin inhibitor, L1 chloro-3-(4-tosylamido)-4 phenyl-2-butanon (TPCK), 10mg/ml L1chloro-3-(4-tosylamido)-7-amido-2-heptanon-hydro-chloride (TLCK), 1 mg/ml aprotinin. Protein lysateswere placed on ice for 30 min, vortexed every 10 min,and cleared by centrifugation at 12,000g for 20 minat 48C. The supernatants were retrieved and frozenat �808C until they were used for immunoblot as-says. Total protein concentrations were determinedwith the Bio-Rad protein assay system (Bio-Rad Lab-oratories, Hercules, CA); BSAwas used as a calibrationstandard.

Protein lysate (50 mg) from each sample was elec-trophoresed in 12 and 6% SDS–polyacrylamide gelsand transferred to nitrocellulose membranes (Protran,Schleicher & Schuell, Keene, NH) for immunode-tection. After the transfer was complete, the mem-branes were stained with Ponceau S (Sigma ChemicalCompany) to verify equal sample loading and transfer(data not shown). For thematrigel experiments, protein

lysateswere resolved on SDS–polyacrylamide gels andsubjected to silver staining to verify equal proteinsample loading (datanot shown). Themembraneswereblocked with 5% non-fat dry milk in Tris-bufferedsaline (20 mM Tris-base, pH 7.6, 150 mM NaCl)containing 0.1% Tween-20 (TBS-T), and immuno-blotted as described previously [57]. The followingantibodies were used: anti-p27 (0.1 mg/ml, Trans-duction Laboratories, Inc., Lexington, KY), anti-AR(PG-21, 1.0 mg/ml, Upstate Biotechnology, Inc., LakePlacid, NY), 34bE12 (1:10,000, Dako, Carpinteria, CA),anti-CK18 (1:10,000, Sigma Chemical Company), anti-PSA (1:500, BioGenex), anti-involucrin (1:10,000, SigmaChemical Company), and anti-p63 (4A4, 1:500; kindlyprovided by FrankMcKeon, Harvard Medical School).After washing with TBS-T, the membranes wereincubated with horseradish peroxidase (HRP)-conju-gated secondary antibodies (Bio-RadLaboratories) anddeveloped using an enhanced chemiluminescencedetection system (ECL detection kit; Amersham Corp.,Arlington Heights, IL).

ReverseTranscription (RT) and PolymeraseChain Reaction(PCR)

Total RNA was extracted from PrEC, LNCaP, andDU-145 cells with the RNeasy mini kit (Qiagen,Chatsworth, CA), according to the manufacturer’sdirections. For cDNA synthesis, 1 mg of total RNAwas subjected to RT in a 20 ml reaction mixture con-taining the following: 250 mM of each dNTP, 20 U ofRNase inhibitor, 50 U of MuLV RT, 2.5 mM randomhexamers, and 1� PCR buffer (1.5 mM MgCl2; allreagentswere purchased from PEApplied Biosystems,Foster City, CA). The reactionmixturewas incubated at428C for 45 min and denatured at 998C for 5 min. Foreach sample, a control reaction lacking RT was alsoincluded. PCR reactions employed 5 ml of the RT re-action in a 50 ml reaction mixture containing thefollowing: 250 mM of each dNTP, 1� PCR buffer(1.5 mM MgCl2), 5 U of Ampli-Taq DNA polymerase(Perkin-Elmer Corp., Norwalk, CT), and 320 nM offorward and reverse AR primers (sense: 50-GAAGC-CATTGAGCCAGGTGT-30, antisense: 50-TCGTCCAC-GTGTAAGTTGCG-30). These oligonucleotide primersare located on exons 4 and 5 of the humanAR gene andare expected to amplify an AR cDNA fragment of163 bp in size [58]. Touch-down PCR was performedfor 32 cycles of the reaction. The first ten cyclesconsisted of denaturation at 928C for 40 sec, annealingat 62–558C for 40 sec, and elongation at 748C for 90 sec.The remaining 22 cycles consisted of 928C for 40 sec,548C for 40 sec, and 748C for 90 sec. Efficiency of theRT step and RNA quantity were controlled in eachsample by PCR amplification of the b2-microglobulin

Intermediate Basal Cells of the Prostate 209

(b2-mic) gene. Intron spanning oligonucleotide pri-mers that yielded a 135-bp b2-mic cDNA product [59]were used at 500 nM under the same RT and PCRconditions as the AR gene (sense: 50-GAAAAAGAT-GAGTATGCCTG-30, antisense: 50-TTCACTCAATC-CAAATGCGG-30). The resultant PCR products wereelectrophoresed in a 2% agarose gel and stained withethidium bromide.

RESULTS

Immunohistochemical Characteristicsof Proliferating PrEC

Proliferating PrECs were subjected to immunohis-tochemical analysis using antibodies against severalcommon prostate epithelial markers. The results werecompared to immunostaining profiles of in vivoprostate and basal/transitional epithelia, as shown inFigure 2A and Table I. The 34bE12 antibody, a widelyused marker for prostate basal cytokeratins [28], stain-ed PrECs intensely (Fig. 2A,iv). 34bE12 binds a panelthat includes HMWCK 5,14,1, and 10. Consistent withprevious observations, 34bE12 stained the prostatebasal layer in vivo (Fig. 2A,i), but basal transitionalepithelia were negative (Table I). Most PrEC nuclei alsostained positive for p63, a p53 homologue implicated instem/TA cell regeneration and expressed exclusivelyin the basal layer ofmanyepithelial types, including theprostate and urothelium (Fig. 2A,iii,vi) [29,30]. Incontrast, antibodies against AR and PSA demonstratedno immunoreactivity in PrECs (Fig. 2A,v); basal cells invivo were weakly AR-positive (Fig. 2A,ii) and PSA-negative (not shown).

Effects of SenescenceWithin PrEC

PrECs obtained at early passage (P2 or P3; seeMaterials and Methods) typically reached a growthplateau (senescence) between P9 and P12, as definedby mitotic arrest for >2 weeks. Whereas proliferat-ing PrEC did not express p27 protein (Fig. 2B,iii),senescent PrEC cells stained positive in vitro for thecell cycle inhibitor p27, as did luminal epithelia in vivo(Fig. 2B,v,i, respectively). Following growth plateau,PrECs also demonstrated strong immunoreactivitywith 35bH11, an antibody recognizing low-molecular-weight CK8. Both 35bH11 and a CK18-specific anti-body stained the prostate secretory layer and super-ficial transitional epithelia in vivo, as expected (Table Iand Fig. 2B,ii); basal layers were mostly (>95%)negative. Although proliferating PrECs focally stainedpositive with 35bH11 (Fig. 2B,iv), this staining wasmuch more pronounced at senescence (Fig. 2B,vi).Senescent cells continued to express p63 and HMWCK(not shown).

Isolation and propagation of PrEC involves cultiva-tion in the absence of androgen,whichmay cause rapidloss of AR expression [31,32]. In fact, we could notdetect AR protein by immunoblotting or immuno-fluorescence in any PrEC sub-population (Figs. 2 and 4and data not shown). Moreover, despite a variety ofexperimentalmanipulations, wewere unable to induceAR protein expression within PrEC, including culturein the presence of androgens (not shown). To examinewhether AR mRNA was present in these cells, totalRNA was recovered from proliferating and senescentPrEC and subjected to semi-quantitative RT-PCR usingprimers complementary to sequences within exons 4and 5 of the AR gene. The PCR product correspond-ing to AR mRNA was detected as a strong band insenescent PrEC but was barely visible in proliferat-ing PrEC (Fig. 2C). Abundant AR product was amp-lified from LNCaP, an androgen-sensitive prostatecancer line known to express high levels of AR proteinand undetectable, as expected, in PC-3 or DU145,two androgen-independent prostate cancer lines.Thus, PrEC entering senescence displayed indepen-dent molecular phenotypes consistent with partial dif-ferentation, including expression of p27, LMWCKs,and AR mRNA. However, AR protein was uniformlyabsent.

Involucrin ExpressionWithin PrECandNormal Prostate

Hormonal and growth factor alterationsmay inducehyperplasia or squamous metaplasia within the pros-tate basal layer [15,33,34]. To investigate this possibi-lity, PrECs and prostate specimens were stained withan antibody against involucrin, a protein associatedwith squamous differentiation/metaplasia [27,35].Formalin-fixed, paraffin-embedded blocks of prolifer-ating PrECs stained strongly for involucrin in vitro(Fig. 3,iii) while in human prostate tissue, involucrinstaining characterized numerous clusters of epithelialcells (Fig. 3,ii). However, both PrECs and prostateepithelia in vivo stained negative for loricrin, anothermarker for squamous differentiation [36] (Table I).Moreover, the involucrin-positive cells appearedhyperplastic and basal by morphology; the latter wasconfirmed by their nuclear positivity for p63 (Fig. 3,i)[37]. The involucrin-positive areas were invariablyflanked in vivo by normal appearing basal and luminalmonolayers, which were involucrin-negative (Fig. 3,i).An epithelial compartment flanked by basal and sec-retory cells has been described previously byDeMarzoet al. [11] as ‘‘transiently amplifying’’ because of theabsence of p27 expression. Interestingly, involucrinpositive hyperplastic regions stained negative for p27in immunohistochemical step sections (Fig. 3,ii,iv).

210 Garrawayet al.

However, p27-negative and involucrin-negative in-termediate basal cells were also identified. Bothinvolucrin-positive and -negative intermediate cellswere expanded significantly in prostate specimensderived from patients treated for at least 3months withtotal androgen ablation in a neoadjuvant setting [38](data not shown). PrEC, therefore, resembled a distinctbasal cell subset present in vivo, amplified under

conditions of androgen ablation, and with character-istics suggestive of hyperplasia.

Effects ofMatrigel and Stromal Factorson PrECDifferentiation

Proliferating PrEC populations were then sub-jected to other conditions likely to influence their

Fig. 2. Immunohistochemicalcharacteristicsofprostateepithelialcells (PrEC),andeffectsof senescence.A:ProliferatingPrECas formalin-fixed, paraffin embedded cell block, and prostate tissue samples were subjected to immunohistochemical analysis, as described in Materialsand Methods. Immunostaining was performed using antibodies against HMWCK (i, v) AR (ii, v), and p63 (iii, vi).B: Proliferating (iii, iv) andsenescent (v, vi) PrEC as formalin-fixed, paraffin embedded cell block were subjected to immunostaining along with prostate tissue samplesas above.Antibodies againstp27 (i, iii, v) andLMWCK(CK8; ii, iv, vi)wereused.C:TotalRNAwasextracted fromsenescentandproliferatingPrEC and subjected to semi-quantitativepolymerase chainreaction (PCR)withprimers that spannedexons 4 and 5 of theARgene.The PCRproductcorrespondingtoARmRNAwasdetectedinsenescentPrECandinLNCaPbutwasmarkedlydiminishedinproliferatingPrEC,DU-145,and PC-3 cells. A control reaction lacking reverse transcriptase (RT) was also included. Efficiency and RNA quantity were controlledineachsamplebyPCRamplificationof theb2-microglobulin(b2-mic)gene.BothARandb2-micPCRamplificationsoccurredinthelinearphaseof thereaction.

Intermediate Basal Cells of the Prostate 211

differentiation state in vivo [39], and examined for ex-pression of the cytokeratins and other marker proteinsdescribed above. Cells were cultivated in the pre-sence or absence of reconstituted basement mem-brane (matrigel). In addition, androgen-treated SCM(see Materials and Methods) was used to test thehypothesis that androgens might stimulate basal celldifferentiation via an intermediate elaborated bystromal cells [40–42]. After 6 days onmatrigel,Westernblot analysis was performed to determine the status ofcytokeratins, p63, p27, AR, and involucrin.

Matrigel caused down-regulation of bothHMWCKsand LMWCKs (Fig. 4). In contrast, p27 expression wasinduced onmatrigel, consistentwith cell growth arrest.Matrigel also caused mild accumulation of involucrin,which was highly expressed in PrEC as describedabove (Fig. 4). Expression of p63 was unaffected re-lative to untreated PrEC. Addition of androgen-treatedSCM to cells grown in the absence of matrigel alsocaused a down-regulation of HMWCK; however,expression of low-molecular-weight CK18 was mini-mally affected by SCM. SCM did not affect cytokeratin

TABLE I. ImmunohistochemistryonHumanProstate,Urothelium*, andCultured Prostate Epithelial Cells

Antibody SpecificityBasal

prostateLuminalprostate

Basalurothelium*

Superficialurothelium

PrEC(proliferating)

PrEC(senescent)

34bE12 CK5,14 þ � � � þ þa-CK18 CK18 � þ � þ þ nd35bH11 CK8 � þ � þ þ þþa-p27 p27 protein � þþ � þ � þa-AR AR protein � þþ nd nd � �a-Involucrin Involucrin a � þ þ þ þa-Loricrin Loricrin � � nd nd � nda-PSA PSA protein � þþ � � � �4A4 p63 protein þ � þ � þ þ

*The epithelium lining the prostate urethra is termed transitional epithelium; urothelium is similar and exists elsewhere in thegenitourinary tract (see Discussion).aExpressed in areas of basal hyperplasia. PSA, prostate-specific antigen; PrEC, prostate epithelial cells; nd, not done.

Fig. 3. InvolucrinexpressionwithinPrECandnormalprostate,andunderconditionsofandrogenablation.Normalprostate tissuefromradi-calprostatectomy (i, ii, iv,v)was stained forp63 (i), involucrin (ii), p27 (iv), andloricrin (v). Involucrin andloricrin immmunohistochemistryonformalin-fixed,paraffinembeddedcellblockofproliferatingPrECare showninpanelsiii andvi, respectively.

212 Garrawayet al.

expression significantly in matrigel-treated cells. Inter-estingly, SCM induced down-regulation of p63 inPrECs cultured on both plastic and matrigel. ARremained absent under all conditions tested, and thelevels of p27 and involucrin were largely unaffected bySCM.

Cultivation on matrigel induced dense cellularclustering and markedly decreased proliferation, asreported previously for PrEC [39] (Fig. 5,ii). The aggre-gates formed by PrEC on matrigel and in the presenceof androgen-treated SCM were embedded in paraffin,

sectioned, and subjected to more detailed histochem-ical analysis. As shown in Figure 5,iii, cells at theperiphery of these clusters appearedflattened,whereascells near the center were larger, more rounded, andshowed signs of vacuolization. Peripheral cell nucleistained positive for p63, while this protein was gen-erally down-regulated in the large, round central cells(Fig. 5,iv). Similarly, although HMWCK were globallydown-regulated by Western blot analysis, cells at theperiphery of the matrigel-induced clusters maintain-ed significant 34bE12 immunoreactivity while centralcells were negative (Fig. 5,v). Involucrin exhibitedabundant and homogeneous expression throughoutthe structures (Fig. 5,vi). Consistent with the Westernanalysis, low-molecular-weight CK8 and CK18 werevirtually undetectable in both cell types (data notshown). Themorphological alterations observed on re-constituted basement membrane in the presence ofstromal components were, therefore, associated withcellular changes suggestive of partial differentiation.

Effect of Retinoic Acid

As noted above, the presence of involucrin withinPrEC raised thepossibility that these cellswere actuallycommitted to a squamous differentiation program, asobserved in urothelia and other epithelial lines follow-ing prolonged in vitro cultivation [43,44]. The effectsof SCM and/or matrigel might, therefore, reflect squ-amous metaplasia instead of luminal changes. To testthis possibility, PrEC was cultured in the presence orabsence of androgen-treated SCM and increasing con-centrations of ATRA, which blocks or reverses squa-mous differentiation in a variety of epithelial systems[45–48]. ATRA caused an accumulation of p63 andmaintained high levels of HMWCKs in PrEC culturedunder standard conditions (Fig. 6). This was consistentwith persistence of a basal phenotype, in accordancewith previous observations [49,50]. However, additionof SCM induced down-regulation of both basal cellmarkers despite high concentrations of ATRA. In-volucrin expression remained abundant despite thepresence of ATRA (Figs. 3, 4, and 5; see Materials andMethods). Thus, although involucrin expression with-in PrECs suggested squamous differentiation, theseresults suggested that SCM by itself triggered a differ-entiation pathway distinct from squamous metaplasia.

DISCUSSION

The identification of cells giving rise to the majorepithelial lineages within the mature prostate willallow greater insight into mechanisms of differentia-tion and carcinogenesis relevant to this organ. Theprostate basal layer is postulated to harbor a progenitorcell population, and this study characterizes PrEC, a

Fig. 4. Effectsofmatrigelandstromalfactorsontheexpressionofdifferentiation markers within PrEC. PrECs were seeded ontountreated 6-well plates or onto 6-well plates containing reconsti-tutedbasementmembrane insets (matrigel), and treatedwith 30%androgen-stimulated stromal cell-conditionedmedia (SCM). After6 days,Westernblot analysiswasperformed to determine thepro-tein levels of p63, cytokeratins, p27, AR, and involucrin. LNCaP,HeLa, andkeratinocyte cellswereincludedascontrols.

Intermediate Basal Cells of the Prostate 213

commercially available basal cell isolate, with regardsto its immunophenotype and response to environ-mental factors affecting cell differentiation.

Our results agree with previous observations that invivo, prostate basal epithelia stain positive for the34bE12 antibody, indicating HMWCK expression [51].These cells also express p63, a protein critical to themaintenance of progenitor cells in several epithelialtypes, including prostate [37]. AR and LMWCK areexpressed at low levels in basal cells, if at all. Progenitorcells within the basal layer are believed to differentiateinto luminal epithelia, which lack HMWCK and p63,

but express LMWCK, AR, and PSA at high levels.Although some reports suggest that luminal cells pos-sess a degree of mitotic capacity [6,19,20], they alsoexpress abundantp27protein, implying substantial cellcycle inhibition.

In vitro, we found that proliferating PrECs exhibitcertain immunohistochemical properties consistentwith a basal lineage: they express HMWCK and p63,but lack AR and PSA. However, they also express lowlevels of secretory cytokeratins, such as CK8 andCK18.While some prostate basal subtypes may co-expressHMWCK and LMWCK [52], in our hands levels of

Fig. 5. Immunohistochemical analysis ofPrECculturedwithmatrigel andandrogen-treatedSCM.Following treatment,PrECwere fixedonmatrigelwith10%formalinovernight,embeddedinparaffin,sectionedandsubjectedtoimmunostaining,asdescribedinMaterialsandMethods.i:InvertedphasemicroscopyofPrECculturedonplasticor(ii)matrigel;(iii)hematoxylinandeosin(H&E)stainingofPrECgrowninmatrigelandin SCM; (iv) PrEC (matrigel, SCM-treated) immunohistochemical staining for p63, (v) high-molecular-weightcytokeratin (HMWCK), and (vi)involucrin.

214 Garrawayet al.

LMWCKalso accumulate, alongwith p27, when PrECsreach senescence. However, unlike most prostateepithelial monolayers in vivo, these cells stain stronglyfor involucrin, a squamous differentiation marker. Inthis study, we also show that a subset of prostateintermediate basal cells in vivo express involucrin.

Two interpretations may be envisioned that accountfor the immunohistochemical profiles observed withinproliferating PrEC: (1) PrEC contains prostate basalcells, or basal transitional epithelia, that have under-gone squamous metaplasia, or (2) these cells representa novel, intermediate hyperplastic rather than meta-plastic, involucrin-positive basal cell subset withinthe prostate. Several lines of evidence suggest thatmetaplasia is less likely. First, whereas growth arrest isfrequently a hallmark of squamous differentiation [53],both PrECs in vitro and involucrin-positive prostatebasal cells in vivo appear hyperplastic, morphologi-cally, and histochemically (they are p27-negative).Second, involucrin-positive PrECs stain negative forother squamous differentiation markers (e.g., loricrin)both in vitro and in vivo. Additionally, although priorstudies have demonstrated that the process of squa-mous metaplasia may be reversed or inhibited by

retinoic acid in many cell types, including genitour-inary epithelial cells [43], we found no effect of ATRAon HMWCK expression within PrEC, despite the factthat p63 expressionwas up-regulated by retinoids. Thehistochemical profile of PrEC may, therefore, be in-trinsic rather than metaplastic, despite in vitro culti-vation in the absence of androgen.

In vivo, involucrin-positive cells express p63 butlack p27, and they are almost always flanked bynormal-appearing, involucrin-negative basal, and sec-retory monolayers. These cells may, therefore, repre-sent a ‘transiently amplifying’ population postulatedpreviously within prostate epithelial layers. If so, theseresults are most consistent with either model A or B inFigure 1 (see Introduction). Although we cannot de-lineate the mechanisms for prostate differentiationwith certainty based on this study alone, our resultssuggest that PrEC represents the in vitro counterpart ofa previously uncharacterized subset, and exhibitsfeatures consistent with a hyperplastic or proliferatingintermediate basal population.

Cultivation of PrECs on reconstituted basementmembrane and in the presence of stromal componentsallowed expression of several features characteristicof at least partial differentiation toward a secretoryphenotype [39,54]. When PrECs were cultured onmatrigel alone, cell proliferation diminished sub-stantially and complex clustering occurred, consistentwith previous observations [39]; these changes wereassociated with enhanced p27 expression and down-regulation of cytokeratin. Addition of androgen-treated SCM resulted in down-regulation of p63, aphenomenon expected to occur during luminal differ-entiation [29,37]. Cellular clusters formed under theseconditions contained a peripheral rim of flat, elongatedcells that stained positive for p63, and high-molecular-weight CK. These flanked a morphologically distinctcentral cell type, which tended to lack both p63 andcytokeratins.

It is tempting to speculate that the changes describ-ed above represent partial differentiation towards aluminal phenotype within PrEC. However, severalconsiderations temper this interpretation. One issueconcerns the CK profile of matrigel-treated PrEC.LMWCK was virtually absent following cultivationon matrigel despite the addition of stromal compo-nents. Aberrant CK expression in primary prostateepithelia has been observed previously following ex-posure to matrigel [39]; this may reflect differences inits composition relative to native prostate basementmembrane. Also, in other epithelial systems, expres-sion of CK18 has been shown to decrease at high celldensities [55]; thus, it is possible that the tight cellularclustering observed on matrigel adversely impacts CKexpression within PrEC. Alternatively, the expression

Fig. 6. Effectof retinoic acid onmarker expressionwithin PrEC.PrEC were seeded onto 6-well plates and treated with 0, 3, and300 nM all-trans retinoic acid (ATRA) in the presence or absenceof 30% androgen-stimulated SCM. After 6 days,Western blot anal-ysiswasperformedto examine the statusofp63,HMWCK,p27,andAR.Prostate cancer cell lines,DU-145 andLNCaPwere included ascontrols.

Intermediate Basal Cells of the Prostate 215

of LMWCK may require physical separation of thedifferentiating cell from the basement membrane, afunction provided by the basal layer in vivo. Thismight occur in conjunction with formation of alumen, or some other physiological proxy for apicalpolarity. Additional experiments investigating factorsthat regulate CK expression should prove informativein this regard.

The persistent lack ofAR andPSA expressionwithinPrEC presents another confounding issue with regardsto differentiation. In accordance with standard meth-ods for isolation of proliferating prostate epithelia,PrECs were cultured in the absence of serum andandrogen. Conditions of androgen deprivation in vitrohave been shown previously to correlate with loss ofAR expression [56], so it is not surprising that PrECpopulations were uniformly AR-negative. Androgenwithdrawal in vivo induces massive secretory cellapoptosiswithin the prostate luminal compartment [5],while the basal layer survives and may undergohypertrophy or squamous metaplasia, as described inIntroduction [16]. It is intriguing, then, that prostatespecimens from patients undergoing androgen abla-tion therapymay demonstrate an increased abundanceof involucrin-positive basal cell clusters. Our observa-tions suggest that androgen withdrawal may select foroutgrowth of a prostate basal subtype lacking AR butexpressing involucrin.

In the prostate, establishment and maintenance ofthe secretory layer is critically dependent on androgenexposure. Presumably, this involves AR-dependentsignaling in both the epithelium and stroma [9].Commercial preparations of PrEC have already under-gone two or three passages in the absence of androgen(see Materials and Methods); in our hands, directaddition of androgen to PrEC cultures had no effect ondifferentiation or AR protein expression (not shown).We suspect, then, that even if PrECs contain progeni-tor cells, the absence of AR precludes full luminaldifferentiation in vitro. This hypothesis could be vali-dated by introduction of recombinant AR into PrEC,followed by androgen exposure and evaluation forluminal changes (e.g., PSA or prostate acid phospha-tase secretion). Indeed, preliminary data suggests thatectopic expression ofARwithin these cells by retroviraltransduction results in PSA expression and otherluminal changes following DHT addition (W. Hahn,personal communication). Additional experimentsalong these lines will expand understanding of thepluripotent potential of prostate basal cells.

In conclusion, we have shown that PrEC exhibitsseveral immunophenotypic features consistent withprostate basal epithelia, and may differentiate par-tially toward a secretory phenotype in the presence ofstromal factors, reconstituted basement membrane,

and senescence. These cells resemble a novel, inter-mediate basal cell subset present in vivo, which maysignify a transiently amplifying compartment. Theabsence of androgen during initial PrEC cultivationin vitro (or the lack of functional AR) likely favorsthe expansion of these cells and precludes the com-pletion of the secretory differentiation pathway. Addi-tional studies of this cell population in vitro mayimprove our understanding of the cellular processesthat direct differentiation and neoplastic transforma-tion in the prostate.

ACKNOWLEDGMENTS

We thankMiles Brown andWilliamHahn for criticalreview of this manuscript. We also thank WilliamSellers for helpful discussions.

REFERENCES

1. Prins GS, Birch L, Greene GL. Androgen receptor localizationin different cell types of the adult rat prostate. Endocrinology1991;129:3187–3199.

2. McNeal JE, Haillot O, Yemoto C. Cell proliferation in dysplasiaof the prostate: Analysis by PCNA immunostaining. Prostate1995;27:258–268.

3. Bonkhoff H, Stein U, Remberger K. Multidirectional differentia-tion in the normal, hyperplastic, and neoplastic human prostate.Simultaneous demonstration of cell-specific epithelial markers.Hum Pathol 1994;25:42–46.

4. Bonkhoff H, Remberger K. Widespread distribution of nuclearandrogen receptors in the basal cell layer of the normal andhyperplastic human prostate. Virchows Archiv—A, Path AnatHistopathol 1993;422:35–38.

5. Evans GS, Chandler JA. Cell proliferation studies in the ratprostate: II. The effects of castration and androgen-inducedregenerationuponbasal andsecretory cell proliferation. Prostate1987;11:339–351.

6. Sinha AA, Bentley MD. The relationship of epithelial cell typesin the ventral prostate glands of castrated mice treated withtestosterone. Anat Rec 1984;208:533–544.

7. Mahapokai W, Xue Y, van Garderen E, van Sluijs FJ, Mol JA,Schalken JA. Cell kinetics and differentiation after hormonal-induced prostatic hyperplasia in the dog. Prostate 2000;44:40–48.

8. Bonkhoff H. Role of the basal cells in premalignant changes ofthe human prostate: A stem cell concept for the development ofprostate cancer. Eur Urol 1996;30:201–205.

9. Foster CS, Ke Y. Stem cells in prostatic epithelia. Int J Exp Pathol1997;78:311–329.

10. De Marzo AM, Nelson WG, Meeker AK, Coffey DS. Stem cellfeatures of benign and malignant prostate epithelial cells. J Urol1998;160:2381–2392.

11. De Marzo AM, Meeker AK, Epstein JI, Coffey DS. Prostate stemcell compartments: Expressionof the cell cycle inhibitor p27Kip1in normal, hyperplastic, and neoplastic cells. Am J Pathol 1998;153:911–919.

12. Bonkhoff H, Stein U, Remberger K. The proliferative function ofbasal cells in the normal and hyperplastic human prostate.Prostate 1994;24:114–118.

216 Garrawayet al.

13. Verhagen AP, Ramaekers FC, Aalders TW, Schaafsma HE,Debruyne FM, Schalken JA. Colocalization of basal and luminalcell-type cytokeratins in human prostate cancer. Cancer Res1992;52:6182–6187.

14. Devaraj LT, Bostwick DG. Atypical basal cell hyperplasia ofthe prostate. Immunophenotypic profile and proposed clas-sification of basal cell proliferations. Am J Surg Pathol 1993;17:645–659.

15. Nevalainen MT, Harkonen PL, Valve EM, Ping W, Nurmi M,Martikainen PM. Hormone regulation of human prostate inorgan culture. Cancer Res 1993;53:5199–5207.

16. Guinan P, Didomenico D, Brown J, Shaw M, Sharifi R, Ray V,Shott S, Rubenstein M. The effect of androgen deprivation onmalignant and benign prostate tissue. Med Oncol 1997;14:145–152.

17. Nevalainen MT, Valve EM, Makela SI, Blauer M, Tuohimaa PJ,Harkonen PL. Estrogen and prolactin regulation of rat dorsaland lateral prostate in organ culture. Endocrinology 1991;129:612–622.

18. Gaudin PB, Zelefsky MJ, Leibel SA, Fuks Z, Reuter VE.Histopathologic effects of three-dimensional conformal externalbeam radiation therapy on benign and malignant prostatetissues. Am J Surg Pathol 1999;23:1021–1031.

19. English HF, Santen RJ, Isaacs JT. Response of glandular versusbasal rat ventral prostatic epithelial cells to androgen with-drawal and replacement. Prostate 1987;11:229–242.

20. Mirosevich J, Bentel JM, Zeps N, Redmond SL, D’Antuono MF,Dawkins HJ. Androgen receptor expression of proliferatingbasal and luminal cells in adult murine ventral prostate.J Endocrinol 1999;162:341–350.

21. Engelhardt J, Schlossberg H, Yankaskas J, Dudus L. Progenitorcells of the adult human airway involved in submucosal glanddevelopment. Development 1995;121:2031–2046.

22. OkadaH,TsuburaA,OkamuraA, SenzakiH,NakaY,KomatzY,Morii S. Keratin profiles in normal/hyperplastic prostatesand prostate carcinoma. Virchows Archiv—A, Path AnatHistopathol 1992;421:157–161.

23. Liu AY, Corey E, Bladou F, Lange PH, Vessella RL. Prostatic celllineage markers: Emergence of BCL2þ cells of human prostatecancer xenograft JuCaP 23 following castration. Int J Cancer1996;65:85–89.

24. Byrne RL, Horne CH, Robinson MC, Autzen P, Apakama I,Bishop RI, Neal DE, Hamdy FC. The expression of waf-1, p53,and bcl-2 in prostatic adenocarcinoma. Br J Urol 1997;79:190–195.

25. Reiter RE, Gu Z, Watabe T, Thomas G, Szigeti K, Davis E,Wahl M, Nisitani S, Yamashiro J, Le Beau MM, Loda M, WitteON. Prostate stem cell antigen: A cell surface marker over-expressed in prostate cancer. Proc Natl Acad Sci USA 1998;95:1735–1740.

26. Magi-Galluzzi C, Loda M. Molecular events in the early phasesof prostate carcinogenesis. Eur Urol 1996;30:167–176.

27. Walts AE, Said JW, Siegel MB, Banks-Schlegel S. Involucrin, amarker of squamous and urothelial differentiation. An immu-nohistochemical study on its distribution in normal andneoplastic tissues. J Pathol 1985;145:329–340.

28. van de Voorde W, Baldewijns M, Lauweryns J. Florid basalcell hyperplasia of the prostate. Histopathology 1994;24:341–348.

29. Yang A, Kaghad M, Wang Y, Gillett E, Fleming MD, Dotsch V,Andrews NC, Caput D, McKeon F. p63, a p53 homolog at3q27-29, encodes multiple products with transactivating,

death-inducing, and dominant-negative activities. Mol Cell1998;2:305–316.

30. YangA, Schweitzer R, SunD,KaghadM,WalkerN, BronsonRT,Tabin C, SharpeA, Caput D, CrumC,McKeon F. p63 is essentialfor regenerative proliferation in limb, craniofacial, and epithelialdevelopment. Nature 1999;398:714–718.

31. Prins GS. Differential regulation of androgen receptors inthe separate rat prostate lobes: Androgen independent ex-pression in the lateral lobe. J Steroid Biochem 1989;33:319–326.

32. Ruizeveld de Winter JA, van Weerden WM, Faber PW,van Steenbrugge GJ, Trapman J, Brinkmann AO, van derKwast TH. Regulation of androgen receptor expression inthe human heterotransplantable prostate carcinoma PC-82.Endocrinology 1992;131:3045–3050.

33. Juniewicz PE, Lemp BM, Barbolt TA, LaBrie TK, Batzold FH,Reel JR. Dose-dependent hormonal induction of benign pro-static hyperplasia (BPH) in castrated dogs. Prostate 1989;14:341–352.

34. Sinowatz F, Amselgruber W, Plendl J, Kolle S, Neumuller C,Boos G. Effects of hormones on the prostate in adult and agingmen and animals. Microsc Res Tech 1995;30:282–292.

35. Serra V, Lara C, Ramirez AA, Marzo MC, Valcuende F,CastellsA, Bonilla-Musoles F. Precocious appearance ofmarkersof squamous differentiation in metaplastic cells of human endo-cervix. Arch Gynecol Obstet 1989;246:233–242.

36. Hohl D, Ruf Olano B, de Viragh PA, Huber M, Detrisac CJ,Schnyder UW, Roop DR. Expression patterns of loricrin invarious species and tissues. Differentiation 1993;54:25–34.

37. Signoretti S, Waltregny D, Dilks J, Issac B, Lin D, Garraway L,Yang A, McKeon F, Loda M. p63 is a prostate basal cell markerand is required for prostate development. Am J Pathol 2000;157:1769–1775.

38. Signoretti S,MontironiR,Manola J,AltimariA,TamC,BubleyG,Balk S, Thomas G, Kaplan I, Hlatky L, Hahnfeldt P, Kantoff P,Loda M. Her-2-neu expression and progression toward andro-gen independence in human prostate cancer. J Natl Cancer Inst2000;92:1918–1925.

39. Fong CJ, Sherwood ER, Sutkowski DM. Reconstituted basementmembrane promotes morphological and functional differentia-tion of primary human prostate epithelial cells. Prostate 1991;19:221–235.

40. Cunha GR, Foster B, Thomson A, Sugimura Y, Tanji N, Tsuji M,Terada N, Finch PW, Donjacour AA. Growth factors asmediators of androgen action during the development of themale urogenital tract. World J Urol 1995;13:264–276.

41. Chung LW, Davies R. Prostate epithelial differentiationis dictated by its surrounding stroma. Mol Biol Rep 1996;23:13–19.

42. Byrne RL, Leung H, Neal DE. Peptide growth factors in theprostate as mediators of stromal epithelial interaction. Br J Urol1996;77:627–633.

43. Southgate J, Hutton KA, Thomas DF, Trejdosiewicz LK. Normalhuman urothelial cells in vitro: Proliferation and induction ofstratification. Lab Invest 1994;71:583–594.

44. Turyk ME, Golub TR, Wood NB, Hawkins JL, Wilbanks GD.Growth and characterization of epithelial cells from normalhuman uterine ectocervix and endocervix. In Vitro Cell Dev Biol1989;25:544–556.

45. Niles RM, Loewy BP, Brown K. The effect of retinoic acid ongrowth and proto-oncogene expression in hamster trachealepithelial cells. Am J Resp Cell Mol Biol 1990;2:365–371.

Intermediate Basal Cells of the Prostate 217

46. Lotan R. Squamous cell differentiation markers in normal,premalignant, and malignant epithelium: Effects of retinoids.J Cell Biochem Suppl 1993;17F:167–174.

47. Fujimoto W, Marvin KW, George MD, Celli G, Darwiche N,De Luca LM, Jetten AM. Expression of cornifin in squamousdifferentiating epithelial tissues, including psoriatic, and reti-noic acid-treated skin. J Invest Dermatol 1993;101:268–274.

48. Lee H, Dawson M, Walsh G, Nesbitt J, Eckert R, Fuchs E,HongW, Lotan R, Kurie J. Retinoic acid receptor- and retinoid Xreceptor-selective retinoids activate signaling pathways thatconverge on AP-1 and inhibit squamous differentiation inhuman bronchial epithelial cells. Cell Growth Differ 1996;7:997–1004.

49. Bamberger C, Pollet D, Schmale H. Retinoic acid inhibitsdownregulation of DeltaNp63alpha expression during terminaldifferentiation of human primary keratinocytes. J InvestDermatol 2002;118:25–34.

50. Shalinsky DR, Bischoff ED, Gregory ML, Gottardis MM,Hayes JS, Lamph WW, Heyman RA, Shirley MA, Cooke TA,Davies PJ. Retinoid-induced suppression of squamous cell dif-ferentiation in human oral squamous cell carcinoma xenografts(line 1483) in athymic nudemice. Cancer Res 1995;55:3183–3191.

51. Robinson EJ, Neal DE, Collins AT. Basal cells are progenitors ofluminal cells in primary cultures of differentiating humanprostatic epithelium. Prostate 1998;37:149–160.

52. YangY,Hao J, Liu X,Dalkin B,Nagle RB.Differential expressionof cytokeratin mRNA and protein in normal prostate, prostaticintraepithelial neoplasia, and invasive carcinoma. Am J Pathol1997;150:693–704.

53. Jetten AM, Harvat BL. Epidermal differentiation and squamousmetaplasia: Fromstem cell to cell death. J Dermatol 1997;24:711–725.

54. Liu AY, True LD, LaTray L, Nelson PS, Ellis WJ, Vessella RL,Lange PH, Hood L, van den Engh G. Cell–cell interaction inprostate gene regulation and cytodifferentiation. ProcNatl AcadSci USA 1997;94:10705–10710.

55. Pantschenko AG, Woodcock-Mitchell J, Bushmich SL, Yang TJ.Establishment and characterization of a caprine mammaryepithelial cell line (CMEC). In Vitro Cell Dev Biol Anim 2000;36:26–37.

56. Rundlett SE, Gordon DA, Miesfeld RL. Characterization of apanel of rat ventral prostate epithelial cell lines immortalized inthe presence or absence of androgens. Exp Cell Res 1992;203:214–221.

57. Pagano M, Tam SW, Theodoras AM, Beer-Romero P, Del Sal G,Chau V, Yew PR, Draetta GF, Rolfe M. Role of the ubiquitin-proteasome pathway in regulating abundance of the cyclin-dependent kinase inhibitor p27. Science 1995;269:682–685.

58. Nirde P, Georget V, Terouanne B, Galifer RB, Belon C, Sultan C.Quantitation of androgen receptor messenger RNA fromgenital skin fibroblasts by reverse transcription-competitivepolymerase chain reaction. J Steroid Biochem Mol Biol 1998;66:35– 43.

59. Kubota T, Koshizuka K, Williamson EA, Asou H, Said JW,Holden S, Miyoshi I, Koeffler HP. Ligand for peroxisomeproliferator-activated receptor gamma (troglitazone) has potentantitumor effect against humanprostate cancer both in vitro andin vivo. Cancer Res 1998;58:3344–3352.

218 Garrawayet al.