distribution of desmosomal proteins in f9 embryonal carcinoma

Journal of Cell Science 103, 69-80 (1992)Printed in Great Britain © The Company of Biologists Limited 1992

69

Distribution of desmosomal proteins in F9 embryonal carcinoma cells and

epithelial cell derivatives

KATRINA T. TREVOR* and LARA S. STEBEN

Center for Molecular Biology, Wayne State University, 5047 Gullen Mall, Detroit, MI 48202, USA

'Author for correspondence

Summary

In diverse epithelia, cytoskeletal keratin intermediatefilaments (IFs) associate with the cytoplasmic face ofintercelluar junctional desmosomes. The processesunderlying desmosome formation and keratin IF inter-actions remain unclear. We have examined F9 embry-onal carcinoma (EC) cell differentiation as a model forembryonic development of epithelial surface desmo-somes. As determined by immunofluorescence mi-croscopy and biochemical protein techniques, F9 ECcells, which lack surface desmosomes and keratin IFs,express the desmosomal proteins desmoplakins I and II(DP 0 1 ) , desmoglein I (DG I) and plakoglobin (PK). DPl/ll are present at low levels and are relatively soluble inbuffer containing Triton X-100. Immunofluorescencelocalizes DP l/ll to the juxtanuclear, centrosomal region.Species of DG I and PK are detected in both the TritonX-100-soluble and -insoluble protein fractions. DG Iappears dispersed throughout the cell while PK residesat cell-cell boundaries. In epithelial cell cultures inducedby retinoic acid (R A) treatment, each of the desmosomalproteins is organized into punctate desmosome-likestructures with the appearance of simple epithelial

K8/K18 IFs. The steady-state levels of DP 0 1 and PKincrease with a partitioning of the majority of thedesmosomal components into the insoluble fraction. Inepithelial cells which lack distinct surface desmosomes,an intracellular association of keratin bundles with DP0 1 is observed, suggesting that keratin filaments mayfacilitate the translocation of these desmosomal com-ponents to the cell surface. Parietal endoderm-like cells,derived by treatment with RA and dibutyryl cAMP, areanalogous to F9 EC cells in that the cells expressdesmosomal components and do not display surfacedesmosomes. Moreover, K8 and K18 do not formdistinct filaments, and the protein and RNA levels of K8are low relative to epithelial cells induced by RA alone.The F9 system appears to be a relevant model for studiesof desmosome assembly and the potential interactions ofdesmosomal proteins and keratin IFs in embryonicepithelial cell types.

Key words: desmosomes, differentiation, embryonalcarcinoma, keratins.

Introduction

The establishment of epithelial cell architecture isfundamental to embryo development and tissue mor-phogenesis and maintenance. Intercellular junctionaldesmosomes and cytoskeletal keratin intermediatefilaments (IFs) are characteristic structural componentsof vertebrate epithelial cells (for review see Franke etal. 1987). Desmosomes consist of at least eight differentproteins that are organized into two distinct plasmamembrane domains: a membrane domain, involved inintercellular adhesion, and a cytoplasmic plaque, whichserves as an anchorage site for keratin filaments(Steinberg et al. 1987; Franke et al. 1987; Green andJones, 1990; Garrod et al. 1990). At the ultrastructurallevel, keratin filaments radiate from the nuclear regionto the cell surface with attachment to the desmosome

cytoplasmic face (for review see Goldman et al. 1990;Skalli and Goldman, 1991).

Several major desmosomal proteins and glyco-proteins have been identified and localized by the use ofspecific antibodies. The adhesive, intercellular mem-brane domain of the desmosome is composed ofglycoproteins, including desmoglein I (DG I; Mr,165,000), the closely related desmocollins I and II (Mr,130,000 and 115,000) and a Mr 22,000 component. BothDG I and desmocollin have been identified as membersof the cadherin family of cellular adhesion molecules(Koch et al. 1990; Nilles et al. 1991; Mechanic et al.1991). The cytoplasmic plaque consists of nonglycosyl-ated proteins termed desmoplakins (DPs): DP I (MT,250,000), DP II (220,000), DP III or plakoglobin (PK;Mr, 83,000) and DP IV (Mr, 78,000). DP I and II arerelated and appear to be derived from a single gene

70 K. T. Trevor and L. S. Steben

(Franke et al. 1987; Green et al. 1988; Luisa et al.,1992). In addition to these desmosomal proteins, minorcomponents have been located in the desmosomes ofdiverse epithelia (Franke et al. 1987; Garrod et al.1990).

The primary method used to study the biochemicaland morphological assembly and disassembly of desmo-somes is to lower the external levels of calcium inepithelial cultures. Low calcium reduces intercellularadhesion and inhibits desmosome formation whilesurface desmosome assembly is reinitiated upon returnto normal calcium levels (Green and Jones, 1990;Garrod et al. 1990). Conflicting reports exist concerningthe regulated assembly and disassembly of the desmo-some and the potential role of keratin IFs in desmo-some formation (Bologna et al. 1986; Duden andFranke, 1988; Jones and Goldman, 1985; Pasdar andNelson, 1988a,b; Pasdar et al. 1991). Differing resultsmay arise from the particular epithelial cell type usedand its specific response to calcium levels.

Differentiation of murine F9 embryonal carcinoma(EC) cells provides a unique in vitro model to studyprocesses of early embryonic epithelial cell biogenesis(for review see Hogan et al. 1983). The earliest stages ofmammalian development involve a complex sequenceof changes leading to the formation of extraembryonicepithelial cell types (Fleming and Johnson, 1988; Hoganet al. 1983). F9 EC cells, which undergo littlespontaneous differentiation in vivo or in vitro, can beinduced into distinct populations of extraembryonicendoderm by exposure to retinoic acid (RA) or RA anddibutyryl cAMP (Strickland and Mahdavi, 1978; Strick-land et al. 1980; Hogan et al. 1983; Grover andAdamson, 1986).

Recently, we and others have used molecular geneticapproaches to explore the interactions of desmosomalproteins and keratin IFs as well as the cellular functionof keratin filaments. Our previous studies demonstrateda striking co-localization of DP I and II with collapsedK8/K18 filaments in RA-induced F9 cells expressing adominant negative mutant of mouse K18 (Trevor,1990). In addition, perturbation of K8/K18 IFs com-promised the stable formation of visceral endodermepithelium in differentiating F9 cell aggregates (Trevor,1990). Similarly, a structural disruption of IFs is seen inepithelial cells expressing mutant domains of DP Iwhich co-align with cytoplasmic IFs (Stappenbeck andGreen, 1992), and biologically defective skin develop-ment is observed in mice expressing a dominantnegative mutant of the epidermal keratin K14 (Vassaret al. 1991). It is interesting that embryonic stem (ES)cells carrying targeted, inactivated alleles of keratin K8form visceral endoderm which appears to assemblesurface desmosomes in the absence of simple epithelialK8/K18 IFs (Baribault and Oshima, 1991). Theseresults raise questions concerning the potential role ofkeratin IFs in desmosome assembly and the structuralrequirements of keratin IFs for epithelial integrity.

Here we report the expression and subcellulardistribution of DP I and II, DG I and PK in F9 EC cellsand differentiated derivatives. Although there is vari-

ance in the levels and localization of these proteins, theproteins are detected in the distinct F9 epithelial celltypes, as well as undifferentiated F9 EC cells. In RA-induced epithelial cells, the desmosomal proteinsappear to assemble into discrete, surface desmosomestructures. Moreover, there is an obvious intracellularassociation between DP I and II and keratin IFs indifferentiating cells which lack assembled desmosomes.This result suggests a role for keratin filaments in theformation of the desmosome. As shown in this study,the F9 system offers a unique, relevant model forstudies of desmosome formation during the morpho-genesis of extraembryonic epithelial cell types.

Materials and methods

Cell culture and differentiationThe F9.22 EC cell line has been described previously (Trevorand Oshima, 1985). Cells were cultivated in Dulbecco'smodified Eagle's medium supplemented with pyruvate (110mg/1), glutamine (0.04%), 10% fetal bovine serum, andgentamycin sulfate at 30 jUg/ml.

To induce epithelial differentiation, cells were plated at adensity of 6 x 103 cells/cm2 and RA (Sigma Chemical Co., St.Louis, MO) was added to a final concentration of 1 /AA. Fordifferentiation to the PE phenotype, F9 EC cells weresimilarly plated and grown in the presence of 1 pM RA and0.1 mM dbcAMP (Sigma). Cells were fed on day 3 with drug-containing medium and analyzed on day 5 of induction.

Immunofluorescent labeling and antiseraImmunofluorescent staining was performed on methanol-fixed cells as previously described (Oshima, 1981). Apolyclonal rabbit antiserum recognizing DP I and n (DP I/1I)and a second polyclonal rabbit antiserum specific for DG Iwere provided by M. Pasdar and W. J. Nelson. Both antiserahave been extensively characterized (Pasdar and Nelson,1988a,b; Pasdar and Nelson, 1989). Troma I, a rat monoclonalantiserum to the mouse keratin K8, has been previouslydescribed (Trevor, 1990; Brulet et al. 1980). To verify thestaining patterns of DP F/ll in RA-induced cells, a mousemonoclonal antibody directed against DP I and II was used(Boehringer Mannheim Corp., Indianapolis, IN). A mono-clonal mouse antibody against plakoglobin (PG5.1) and apolyclonal guinea pig antibody recognizing DG 1 wereobtained from P. Cowin (Cowin et al. 1986; Cowin andGarrod, 1983). Mouse monoclonal antisera against vimentinwas purchased from Sigma Chemical Co., St. Louis, MO.Primary antibodies were visualized by using the appropriatefluorescently labeled secondary antibodies: FITC- or rhoda-mine-conjugated goat antiserum to rabbit IgG, rhodamine-conjugated goat antiserum to mouse IgG, and FITC-conjugated goat antiserum to rat IgG (Sigma). Controlsincluded incubation of fixed cells with only secondaryantibodies and verification that secondary antibodies werespecies specific.

Wheat germ agglutinin was utilized to label the Golgicomplex (Singh et al. 1988). Cells were preincubated with 2^g/ml of unlabeled wheat germ agglutinin (Sigma) for 7minutes at 4°C before fixation in cold methanol and wereprocessed as for immunofluorescence staining using FITC-labeled wheat germ agglutinin (Sigma).

Cell fractionationSeparation of cellular proteins into soluble and insoluble

Distribution of desmosomal proteins 71

(cytoskeleton) fractions was performed using the procedureof Triton X-100/high-salt extraction (Fey et al. 1984; Pasdarand Nelson, 1988a). Cell monolayers were rinsed twice withcold phosphate-buffered saline, gently scraped from the dishand pelleted at 2,000 g for 3 min at 4°C. The cells were thenextracted in buffer containing 50 mM NaCl, 300 mM sucrose,10 mM PIPES (pH 6.8), 3 mM MgCI2, 0.5% v/v Triton X-100,1.2 mM phenylmethylsulfonyl fluoride (PMSF), 0.02 mg/mlDNase I and 0.1 mg/ml RNase for 20 min at 4°C. A solution of2.5 M (NH4)2SO4 was added to a final concentration of 250mM and incubation was continued for another 5 min. Sampleswere centrifuged at 20,000 g for 15 min to yield a supernatant(soluble fraction) and pellet (insoluble, cytoskeletal fraction).The cytoskeletal pellet was resuspended in a buffer containing1% w/v SDS, 5 mM EDTA, and 10 mM Tris-HCl, pH 7.5.

Immunoprecipitation and immunoblottingThe procedures used for total cell lysate preparation,immunoprecipitation and gel electrophoresis were describedpreviously (Oshima, 1982). DP I and II proteins weredetected by immunoprecipitation of metabolically labeledproteins using the polyclonal rabbit antiserum against DP Iand II. Cell monolayers were labeled for 3 h in mediumdeficient in methionine that was supplemented with 35S-methionine (150 /iCi/ml) and 10% fetal bovine serum. Foranalysis of DP I and II in total cell lysates, the proteins wereimmunoprecipitated from 5 X 106 cts/min of [35S]methionine-labeled lysate. To determine the distribution of DP I and IIproteins in the soluble and insoluble cell protein fractions, theproteins were immunoprecipitated from 5 x 106 cts/min ofsoluble, labeled protein and 5 x 105 cts/min of the insolublecell fraction. As in previous reports (Cervera et al. 1981),approximately 15% of the cellular proteins was present in thepelleted, cytoskeletal protein pool as determined by radioac-tive counting of incorporated [35S]methionine in each cellfraction.

For immunoblotting, protein concentrations were deter-mined by the BCA method (Pierce, Rockford, IL). Ten figand 80 ng of the insoluble and soluble protein fractions,respectively, were separated on SDS-polyacrylamide gels(PAGE) and transferred electrophoretically to nitrocellulosefilters. The filters were processed for protein analysis using theAmersham ECL Western blotting detection system (Amer-sham Inc., Amersham, UK). Primary antibody binding wasdetected by using the appropriate horseradish peroxidase-conjugated secondary antibody (Cappel Research Products,Durham, NC) followed by chemiluminescent fluorography.Each experiment was performed a total of three times. Therelative amounts of respective proteins were determined fromresulting fluorograms by laser scanning image analysis using aBio Image Laser Scanner (Bio Image, Ann Arbor, MI).

RNA analysisTotal cellular RNA was isolated by the acid, single stepprocedure described by Chomczynski and Sacchi (1987). ForNorthern analysis, 20 /ig/lane of total RNA was denatured informaldehyde, electrophoresed through a 1.2% formalde-hyde/agarose gel, and transferred to nitrocellulose (Sambrooket al. 1989). fiP-labeled DNA probes were generated fromcDNA fragment inserts by the random priming method(Feinberg and Vogelstein, 1984), employing the MultiprimeDNA labeling system (Amersham, Amersham, UK). Filterswere hybridized with 6 x 10* cts/min per ml of the respectiveprobes and washed using standard conditions (Sambrook etal. 1989). The cDNA clones encoding human beta actin(pHBA-1), laminin Bl (pPE386) and mouse K8 (Endo A;

clone alpha 2) have been previously described (Ponte et al.1984; Kurkinen et al. 1983; Vasseur et al. 1985).

Results

Immuno-localization of DP I and II proteins in F9EC cells and differentiated derivativesAs described previously (Hogan et al. 1983), F9 ECcells undergo dramatic morphological and biochemicalchanges when grown in the presence of inducing agents.Undifferentiated F9 EC cells form dense, tightlypacked colonies of small rounded cells (Fig. 1A). Aftera 5 day exposure to RA, approximately 60-70% of thecells assume a flattened, epithelioid morphology withdistinct, opposing cell boundaries (Fig. IB). The cellsare similar to primitive extraembryonic endoderm andpossess numerous markers including simple epithelialkeratin IFs composed of K8 and K18 (Trevor andOshima, 1985; Oshima et al., 1988). Growth in thepresence of RA and dbcAMP results in dispersed,rounded cells with long filopodia (Fig. 1C), whichresemble primary cultures of embryonic parietal endo-derm (PE) (Strickland et al. 1980; Hogan et al. 1983).Less than 20% of induced PE cells show a moreflattened, epithelial morphology.

Indirect immunofluorescent staining using an anti-serum which recognizes DP I and II (DP l/ll) localizedthese proteins to a perinuclear region in both undiffer-entiated F9 EC cells and the rounded PE cells (Fig.1D,F). In RA-treated cultures, areas of flattened cellsdisplayed a punctate alignment of DP I and II at cell-cell boundaries (Fig. IE), characteristic of surfacedesmosomes.

The juxtanuclear localization of DP l/ll in F9 EC cellsand PE cells suggested that the proteins might reside inthe vicinity of the centrosome. As Golgi stacks areknown to be grouped near the centrosomal region(Bershadsky and Vasiliev, 1988), cells were stainedsimultaneously with DP l/ll antibodies followed byrhodamine-conjugated secondary antibodies and FITC-labeled wheat germ agglutinin, which detects the Golgicomplex (Singh et al. 1988). Double fluorescencedetection revealed very similar perinuclear stainingpatterns (Fig. 2), suggesting that DP l/ll are present inthis subcellular region.

Solubility properties of DP I and DP IIInvestigations of the subcellular state of surface desmo-somal components in epithelial cell cultures havedemonstrated that desmosomal proteins, including DPI and II, are notoriously insoluble in Triton X-100/high-salt buffer (Jones and Goldman, 1985; Bologna et al.1986; Kapprell et al. 1987; Pasdar and Nelson, 1988a).The solubilities of DP I and II are enhanced whenepithelial desmosomes are disrupted by decreasing theexternal level of calcium (Jones and Goldman, 1985;Duden and Franke, 1988; Pasdar and Nelson, 1988a).Prior to analyzing the subcellular distributions of DP Iand II in the F9 cell types, the steady-state levels of DP Iand II were examined by immunoprecipitation of total,


Fig. 1. Morphological features of F9 cell types and immunolocalization of DP I/I I. (A) Morphology of undifferentiated F9cells; (B) RA-derived epithelial cells; (C) PE cells after treatment with RA and dbcAMP. (D) Indirect immuno-localizationof DP I/I I in F9 EC cells; (E) RA-induced epithelial cells; (F) PE cells. Cells were methanol-fixed and preincubated with apolyclonal rabbit antibody against DP i/n followed by rhodamine-labeled goat anti-rabbit IgG. Note that areas of cellsshown in (A-C) are different from those shown in (D-F). Differentiated cell cultures were analyzed after 5 days growth inthe presence of inducing agents. Bar, 10 /an.

labeled cellular proteins (Fig. 3A). Densitometricscanning of the autoradiogram indicated that DP I andII were present at approximately an 8-fold lower level inundifferentiated F9 cells relative to RA-derived epi-thelial cells. In PE cells, the amounts of DP I and IIwere about 50% less than detected in RA-treated cells.Analysis of the solubilities of the proteins, as character-ized by their extraction in Triton X-100/high-salt buffer,showed that in both undifferentiated F9 cells and PEcells, the majority of DP I and II existed in the solubleprotein fraction (Fig. 3B). Moreover, smaller proteinbands were consistently observed in the soluble fractionof the PE cells suggestive of a degradation ormodification of soluble DP I and/or II proteins in thiscell type. In RA-induced cells, an enrichment of DP Iand II was detected in the pelleted, cytoskeletal fractionwith approximately 70% of the proteins partitioned intothis fraction.

Detection of DG I and PK in F9 EC cells anddifferentiated derivativesBecause DP I and IT were detected in undifferentiatedF9 cells, we investigated the nature of two otherdesmosomal components, DG I and PK, in F9 EC cellsand the differentiated cell types. The transmembraneglycoprotein DG I is found in all desmosomes and hasbeen implicated in the adhesive mechanism of surfacedesmosomes (Franke et al. 1987; Garrod et al. 1990).Although originally described as a desmosomal com-ponent of the cytoplasmic plaque, PK is detected notonly in desmosomal junctions but also in other types ofadhering intercellular junctions, including intermediatejunctions (Cowin et al. 1986; Kapprell et al. 1987).

Immuno-staining of F9 EC cells localized PK to cell-cell boundaries in a pattern outlining the periphery ofthe cells (Fig. 4), indicative of its presence in intercellu-lar adhering junctions (Cowin et al. 1986). In RA-


UD R R/D

Fig. 2. Localization of DP I/II in F9 EC and PE cell types.Double immunofluorescence of F9 EC cells (A and B); PEcells (C and D). Cultures were preincubated with wheatgerm agglutinin for 7 min at 4CC to saturate the plasmamembrane binding sites, fixed, and then incubated with arabbit polyclonal DP I/II antibody followed by rhodamine-labeled anti-rabbit IgG (A and C) and FITC-labeled wheatgerm agglutinin (B and D). Bar, 10 /im.

induced epithelial cells, a faint linear staining along cellboundaries was discerned, underlying discrete spots.Characteristic of epithelial cell types, the distribution ofPK in punctate, surface spots most likely corresponds toits association with desmosomes while the more linearpattern of staining is representative of its distribution inother types of intercellular adhering junctions (Cowinet al. 1986). In PE cells, which lack intimate cell-cellcontacts, a very faint staining for PK was observed inthe cell body.

Indirect immunofluorescence of DG I showed adiffuse, grainy staining in undifferentiated F9 cellswhile RA-induced epithelial cells displayed a punctate,surface pattern typical of desmosomes formation withsome intracellular staining (Fig. 4). In nearby epithelialcells, which appeared to lack a distinct surface localiz-ation for DG I, a scattered, granular staining wasobserved with more intense grains in the vicinity of thenucleus. As for F9 EC cells, cells differentiated to thePE cell phenotype displayed a scattered pattern ofstaining over the cell body and faint fluorescent grainsextending into the long cytoplasmic filopodia.

Immunoblotting identified PK in both the soluble andpelleted, insoluble fractions (Fig. 5). Protein bandssmaller than PK (Mr 83,000) were also detected whichprobably represent degradation products. Accountingfor these smaller species, the amount of PK in the

DP I/II -

B

DP I/II I

UD R

-205

-116

R/D

PS P S PS

-205

Fig. 3. Solubility properties of DP I and II.(A) Autoradiogram of DP L and II immunoprecipitatedfrom total cellular protein. Cell lysates (5 x 106 cts/min)were immunoprecipitated with DP I/II antiserum.(B) Autoradiogram of DP I/II immunoprecipitated fromsoluble (S) and insoluble (P) fractions of F9 cells. Proteinswere immunoprecipitated from 5 x 106 cts/min of thelabeled, soluble fraction and 5 x 105 cts/min of theinsoluble fraction and resolved by SDS-PAGE (10% gel).Labeled proteins detected by fluorography.Undifferentiated F9 cells (UD); RA-derived epithelial cells(R); PE cells (R/D)- Relative molecular mass markers:myosin (205 x 103), beta-galactosidase (116.5 X 103). DP Ihas an apparent molecular mass of approximately 250 x103 and DP II of about 220 X 103.

pelleted, insoluble fraction of F9 EC cells was onlyslightly higher than that partitioned into the solublepool. In RA-derived epithelial cells, the relative level ofinsoluble PK was increased with approximately twice asmuch PK partitioning into this fraction. Estimations ofthe total amount of PK in each cell type indicated thatRA epithelial cells possessed a 3-fold higher level of PKrelative to F9 EC cells. Barely detectable amounts ofPK were identified in the fractionated protein pools ofPE cells, confirming the observed faint immuno-staining of this cell type.

Analysis of DG I (Mr 165,000) revealed that each celltype had similar levels of the protein present in theinsoluble fraction (Fig. 5). In addition, a distinct lowerMr protein estimated at 125,000 was detected in thesoluble protein pools of both undifferentiated F9 cellsand PE cells. This species was virtually undetectable inRA epithelial cells.

Intracellular association of DP i/ll and keratinsIn RA-treated cultures, approximately 50% of theepithelial cells displayed punctate staining for DP I/II,DG I and PK at opposing cell-cell membranes,indicative of desmosome formation. The remainingcells of epithelial morphology displayed a linear surfacestaining for PK (not shown), as might be expectedbecause PK localized to cell boundaries in undifferen-tiated F9 cells. As mentioned, DG I was distributed in a


Fig. 4. Immuno-localization of PK and DG I in F9 EC cells and differentiated derivatives. (A-C) Immuno-staining for PK;(D-F) localization of DG I. Undifferentiated F9 cells (A and D); RA-derived epithelial cells (B and E); PE cells (C andF). Cells were processed for indirect immunofluorescence using a mouse monoclonal antibody against PK or a polyclonalrabbit antibody recognizing DG I followed by the appropriate rhodamine-labeled secondary antibodies. Bar, 10 /un.

highly diffuse, granular pattern. Double immunofluor-escence of DP 0 1 and keratin IFs revealed some cellswith filamentous-like speckles of DP 0 1 decoratingkeratin filaments while other cells displayed distinct,larger spots of DP 0 1 along the filaments (Fig. 6A-D).This intracellular association was most apparent in cellslacking a punctate surface staining for DP l/ll. Asshown in Fig. 6 (E and F), little if any DP l/ll stainingco-aligned with keratin IFs in those cells displaying DPl/ll at the cell surface. Moreover, similar patterns of DP0 1 distribution were observed when only DP 0 1staining was performed using this particular antibody aswell as when employing a monoclonal antibody recog-nizing DP 0 1 (data not shown).

Unlike epithelial cells derived by RA treatment, therounded PE cells did not possess distinct keratinfilaments as determined by immuno-staining using anantibody recognizing K8 (Fig. 7A). Some cells com-

pletely lacked keratin staining while others displayed aweak co-localization of aggregated keratin material tothe perinuclear region. Double immunofluorescenceshowed that this staining coincided with that of DP 0 1(Fig. 7A,B). Similar staining was observed whenemploying an antibody recognizing K18 (data notshown). Although extended keratin filaments appearedto be absent in PE cells, immunofluorescent staining forvimentin revealed elongated vimentin IFs traversing thecytoplasmic filopodia (Fig. 7C).

Expression of keratin K8 in PE cells and RA-inducedepithelial cellsWestern analysis of K8 confirmed that the amount ofkeratin protein present in PE cultures was low relativeto RA-induced cells (Fig. 8A). Densitometric scanningof immunoblots revealed that PE cells possessed a 5-


UD

P S

R

P S

R/D

P S

consistently about 30% less than that observed in RNAfrom RA-treated cells.

PK — - -

DG I — — _

Fig. 5. Solubility properties of PK and DG I in F9 EC cellsand differentiated cell types. Undifferentiated F9 cells(UD); RA-differentiated epithelial cells (R); PE cells(R/D). Following cell fractionation, 10 /ig of the soluble(S) protein pool and 80 /ig of the pelleted, insoluble (P)protein fraction were separated by SDS-PAGE (12.5%gel), transferred to nitrocellulose and immunoblotted. PKwas detected using a monoclonal mouse antibody againstPK; DG I was detected employing a monoclonal guinea pigantibody recognizing DG 1. Protein bands were developedby incubating the filters with appropriate horseradishperoxidase-labeled secondary antibodies followed bychemiluminescent fluorography. Filled circles indicatemigration of relative molecular mass standards. For the PKimmunoblot, the upper circle indicates beta-galactosidase(116.5 x 103 Mr); and lower, bovine serum albumin (77 x103 Mr). For the DG 1 immunoblot, the upper circledesignates the migration of myosin (205 x 103 Mr); andlower, beta-galactosidase (116.5 x 103 MT).

fold lower level of K8 protein than that detected in RA-induced epithelial cells. As approximately 10-20% ofthe cells assumed a more flattened, epithelioid mor-phology possessing K8/K18 IFs, a portion of thedetected K8 protein is accounted for by these cells. InRA-treated cells, K8 partitioned into the insoluble,cytoskeletal fraction as expected for a filamentouskeratin protein. PE cells also exhibited a segregation ofK8 to the insoluble pool, indicating that the aggregatednature of K8 and K18 proteins observed by immuno-staining did not influence the solubility properties of theproteins.

Northern analysis demonstrated that the amount ofK8 protein synthesized in PE cells correlated with alower level of encoding K8 RNA (Fig. 8B). The amountof steady-state K8 RNA was about 4-fold less than thatdetected in RNA isolated from RA epithelial cells.Characteristic of the PE phenotype, the steady-statelevel of RNA encoding the extracellular matrix proteinlaminin Bl was dramatically increased in PE cells. Asdescribed by others, high amounts of laminin Bl RNAparallel the increased amount of laminin Bl proteinsynthesized upon F9 EC cell differentiation in thepresence of RA and dbcAMP (Durkin et al. 1986;Cooper et al. 1983). F9 EC cells express only low levelsof laminin Bl while there is a more modest increaseafter RA treatment. In undifferentiated F9 cell RNA,the amount of actin RNA was slightly lower than thatdetected after RA induction, similar to the differencesobserved in studies of the nuclear transcription rates ofactin RNA in these two cell types (Oshima et al. 1988).In PE cells the amount of steady-state actin RNA was

Discussion

We have determined that undifferentiated F9 cells,which lack surface desmosomes and keratin IFs, possesspre-existing pools of at least four known constituents ofthe desmosome: DP I, DP II, PK and DG I. Like F9 ECcells, cells of the PE phenotype do not assemble surfacedesmosomes, although the cells express desmosomalcomponents. Upon RA-induced epithelial cell differen-tiation, the proteins assemble at cell-cell boundaries asdiscrete structures, characteristic of desmosomes.

In undifferentiated F9 cells, DP l/ll appear to exist inthe juxtanuclear, centrosomal region of the cell asrelatively soluble components. Although DP l/ll immu-nolocalization coincides with the Golgi complex, asdetermined by wheat germ agglutinin labeling, a moredetailed microscopic examination is required to deter-mine their exact subcellular location. Also, recentanalysis of P19 EC cells has demonstrated low levels ofsoluble DP l/ll localized near the nucleus (Schmidt etal. 1992). The steady-state amounts of DP I and IIdetected in undifferentiated F9 cells are 8-fold less thanthat observed in RA-induced epithelial cells. Perhaps,in F9 EC cells there is a lower expression of the singlegene which is thought to encode both DP I and II(Green et al. 1988; Green et al. 1990). Gene expressionmay be increased following RA exposure. Alterna-tively, the proteins are more rapidly degraded whensoluble, as demonstrated in studies of desmosomeassembly in MDCK epithelial cells exposed to lowcalcium medium (Pasdar and Nelson, 1988a).

In contrast to the more soluble states of DP I and II inF9 EC cells, the major glycosylated form of DG I isdetected in the insoluble protein fraction. The steady-state level is comparable to that in RA-inducedepithelial cells which are undergoing desmosomeformation, suggesting that the protein is relativelystable in the undifferentiated cells. In addition, asoluble species of about 125,000 Mr is detected, which islarger than the estimated Mr of 107,703 of unglycosy-lated human DG I. (Nilles et al. 1991). This proteinspecies may be a form of DG I which has yet to undergocomplete glycolytic processing or may represent adegradation product of the fully processed DG I.Previous investigations of MDCK cells have indicatedthat DG I is glycosylated in the Golgi complex andrapidly titrated into an insoluble pool with subsequenttransport via microtubule "tracks" to the plasmamembrane (Pasdar and Nelson, 1989a; Pasdar et al.1991). In undifferentiated F9 cells, DG I appearsdispersed in a granular pattern rather than entirelylocalized to the Golgi region or distributed in afilamentous pattern as might be expected for anassociation with microtubules. Further biochemical andcytological analyses will reveal aspects of DG Iprocessing and its subcellular location(s).

In F9 EC cells, PK occurs at cell-cell boundaries,


Fig. 6. Lntracellular association of DP 0 1 and keratin IFs in RA-induced epithelial cells. Cells were fixed and co-stainedwith a rabbit monoclonal against DP 0 1 (A, C, E) and a monoclonal rat antibody (Troma I) recognizing K8 (B, D, F). Aand B, C and D, E and F represent different patterns of DP 0 1 staining seen relative to keratin IFs. Primary antibodieswere visualized using the appropriate rhodamine- and FITC-labeled antisera as secondary antibodies. Bar, 5 ^m.

indicating an association with intercellular junctions.PK exists not only in epithelial cell desmosomes but inother types of intercellular adhering junctions, includ-ing intermediate junctions, where it is associated withmicrofilaments, vinculin and alpha-actinin (Cowin et al.1986; Kapprell et al. 1987). A slightly higher portion ofPK partitions into the insoluble protein pool relative tothe soluble fraction. Identification of PK in both thesoluble and insoluble protein fractions might beexpected, as these different forms have been observed

in both desmosome and non-desmosome-bearing cellsand tissues (Cowin et al. 1986; Kapprell et al. 1987).

Analogous to undifferentiated F9 cells, PE cellsexhibit similar immunolocalization characteristics andbiochemical distributions of both DP I/II and DG I.However, unlike F9 EC cells, the level of PK is barelydetectable, as might be expected for a cell type whichlacks intimate cell-cell contacts. Either PK is rapidlydegraded or the level of PK gene expression is relativelylow. Moreover, immunofluorescence reveals that F9-


P S P S P S

Fig. 7. Immuno-localization of keratin protein and DP I/IIin PE cells. (A and B) Double indirectimmunofluorescence for keratin protein and DP I/O,respectively. Keratin was detected using a rat monoclonalantibody (Troma \) against K8. DP /U was localizedemploying a polyclonal rabbit antiserum against DP I/II.Appropriate FITC- and rhodamine-labeled antisera wereused as secondary antibodies. (C) Detection of vimentinfilaments using a primary monoclonal mouse antibodyfollowed by rhodamine-labeled goat anti-mouse secondaryantiserum. Bar, 10 ^m.

derived PE cells do not possess extensive keratin IFs, assuggested previously (Boiler and Kemler, 1983). Whenpresent, aggregated keratin proteins co-localize withDP 0 1 to the juxtanuclear region while vimentin existsas elongated filamentous structures. As compared toRA-induced epithelial cells, PE cells express loweramounts of keratin proteins, which correspond toreduced levels of keratin RNA.

In contrast to PE cells derived in vitro, PE cells of themouse embryo appear to possess both K8/K18 andvimentin IFs (Lane et al. 1983; Lehtonen et al. 1983).This discrepancy in IF content between PE cells in vivoand those differentiated in vitro may relate to therepresented developmental stage. Embryonic PE cellswere examined in later post-implantation embryos,which possess a well-established parietal endoderm. F9-derived PE cells possibly represent an earlier, pre-implantation stage of PE differentiation. The featuresof PE cells derived in vitro are highly reminiscent ofthat observed for fibroblastoid cells derived from thethe conversion of epithelial cells: intercellular junc-tions, including desmosomes, are absent; DP I and IIare more soluble; keratin expression is decreased; andvimentin IFs are predominant (Boyer et al. 1989). Inaddition, this cell conversion is dependent on proteinphosphorylation, which is similarly implicated in the F9

mK8

BUD R R/D

mK8

LN B1

Actin

Fig. 8. Expression of keratin K8 in F9 cells.(A) Immunoblot of mouse K8 (mK8). Ten ,ug and 80 ^g ofthe pelleted, insoluble (P) and soluble (S) proteinfractions, respectively, were separated by SDS-PAGE(12.5% gel), transferred to nitrocellulose andimmunoblotted using the rat monoclonal Troma 1antiserum. Protein bands were developed using ahorseradish peroxidase-labeled anti-rat antiserum followedby chemiluminescent fluorography. Undifferentiated F9cells (UD); RA-derived epithelial cells (R); PE cells(R/D). Bars indicate relative molecular mass standards:upper bar, bovine serum albumin (77 x 103); lower bar,ovalbumin (46.5 x 103). Lower bands detected in the RA-treated sample most likely represent degradation, whichcan occur during protein isolation. (B) Northern analysis.Following isolation of total cellular RNA, 20 jig/lane wasdenatured in formaldehyde, electrophoresed through a1.2% formaldehyde/agarose gel, and transferred tonitrocellulose. RNA species were detected by hybridizationwith the respective -^P-labeled DNA probes followed byfluorography. mK8, mouse K8 RNA; LN Bl, laminin BlRNA.

system. Generation of the PE phenotype requiresdbcAMP or other compounds which elevate cAMPlevels in combination with RA (Strickland et al. 1980).It is of significance that the levels of the mRNAsencoding retinoic acid receptors (RARs) are greatlyreduced in the presence of RA and cAMP analogsrelative to RA treatment alone (Hu and Gudas, 1990).A modulation of the effects of RA on gene expressionby cAMP analogs may account for the reducedexpression of keratins detected in PE cells.

In RA-induced epithelial cell cultures, DP 0 1 , DG Iand PK assemble into desmosomes, on the basis of theircharacteristic punctate, surface staining. Relative to F9EC cells, the levels of DP 0 1 and PK increase, and themajority of the proteins are transferred to the insolubleprotein fraction, indicating of desmosome biogenesis.The majority of DG I appears as a fully processedprotein with only low levels of the soluble MT 125,000species observed. In differentiating cells lacking dis-crete, surface desmosomal structures, dual immunoflu-orescence reveals a distinct co-alignment of DP 0 1 with


K8/K18 filaments. Individual cells display faint specklesof DP l/ll or distinct bead-like dots which appear todecorate keratin IFs. As RA-induced differentiation isnot a fully synchronous phenomenon, one interpret-ation is that these patterns of DP I/II staining representstages of DP I/II assembly prior to their appearance atthe cytoplasmic membrane. Further transport of DP0 1 to the cell surface may be facilitated by translo-cation along keratin IFs.

Studies of developing embryos and cultured epi-thelial cells have resulted in opposing views concerningdesmosome assembly and keratin IF interactions. Oneopinion is that the desmosome serves as an organizingcenter for IFs following desmosome assembly (Over-ton, 1962; Dembitzer et al. 1980; Jackson et al. 1981;Bologna et al. 1986; Duden and Franke, 1988). Otherresults suggest that keratin IFs may play a regulatoryrole in desmosome assembly (Jones and Goldman,1985; Jones and Grelling, 1989; Pasdar and Nelson,1988a,b; Pasdar et al. 1991). Recent investigations ofMDCK epithelial cell cultures maintained in lowcalcium-containing medium and switched to a normalcalcium level have demonstrated that a soluble complexof DP I and II is recruited to the insoluble cell fractionin association with keratin IFs upon initiation ofdesmosome assembly (Pasdar and Nelson 1988a,b;Pasdar et al. 1991). Moreover, a colocalization of DPl/ll with disrupted K8/K18 IFs is observed in RA-induced F9 cells expressing a dominant negative mutantof K18 (Trevor, 1990). Transient expression of mutantdomain regions of DP I in epithelial cells demonstratesa coalignment of the proteins along keratin IFs(Stappenbeck and Green, 1992). This result supports aprevious notion that DP l/ll might interact with IFs, onthe basis of a common periodicity of charged residues(Green et al. 1990). In this report, we demonstrate theoccurrence of a spatial interaction of DP l/ll withK8/K18 filaments during F9 epithelial cell differen-tiation without direct manipulation of normal cellularphysiology or cytoarchitecture.

A recent study on desmosome biogenesis in pre-implantation mouse embryos has demonstrated thatsurface desmosome formation first appears in thetrophectoderm epithelium (Fleming et al. 1991) withtheir subsequent occurrence in the later-stage visceralendoderm surrounding the egg cylinder (Hogan et al.1983). As in F9 EC cells, a linear pattern of PKmembrane association is seen in the 16 to 32-cell stagemorula (Fleming et al. 1991). K8 and K18 proteinsynthesis and filament polymerization precedes theappearance of surface desmosomes, occurring at the 4-to 8-cell stage of mouse development (Oshima et al.1983; Chisholm and Houliston, 1987). Taken together,these reports are provocative when considering a rolefor keratin filaments in the facilitation of desmosomeassembly.

Of significance, embryonic stem (ES) cells, whichlack keratin IFs due to targeted inactivation of both K8alleles, show an apparently normal differentiation tovisceral endoderm, including the plasma membraneappearance of DP l/U in desmosome-like structures

(Baribault and Oshima, 1991). Yet, dominant negativemutants of keratin proteins have been shown to preventvisceral endoderm formation in F9 cells (Trevor, 1990)or result in defective skin epithelium as in the humangenetic disorder epidermolysis bullosa simplex (Vassaret al. 1991; Coulombe et al. 1991). These two latterresults suggested that keratins have an importantstructural function. Considering that visceral endodermcan form in the absence of keratin IFs, the effects ofmutant keratin proteins may reflect secondary struc-tural alterations arising from the induced collapse ofkeratin filaments. In F9 epithelial cells, one suchperturbation appears be a physical pulling of associatedDP l/ll with their subsequent localization to K8/K18aggregates (Trevor, 1990).

Although differentiated K8(-) ES cells lack keratinIFs, the K8 partners, K18 and K19, are detected at lowlevels and are observed to associate with surface DP l/llin a subpopulation of the cells. How these keratinproteins arrive at the plasma membrane is unclear.Detailed comparison of DP l/ll localization duringdifferentiation of normal and K8(-) ES cells will helpto clarify the potential role of keratin IFs in DP l/llassembly and translocation. Such analysis may revealdifferences in desmosome assembly and/or structure.At present, there is no evidence supporting a direct rolefor IFs in intracellular transport, as is well known forcytoskeletal microtubules (Vale, 1987; Kreis, 1990).However, a mechanism for a kinesin-dependent inter-action of vimentin IFs with microtubules and transportvesicles has been proposed (Gyoeva and Gelfand,1991).

As explored in this study, the F9 differentiationsystem provides an opportune model for addressingnumerous questions regarding desmosome assemblyunder conditions which mimic the normal differen-tiation of early embryonic epithelial cell types. Atpresent, we are in the process of evaluating theexpression and distribution of other known desmo-somal components. Further investigations will elucidateour understanding of the establishment and mainten-ance of simple epithelial cell, as well as tissue,architecture.

We thank M. Pasdar, J. Nelson, and P. Cowin for theirgenerous gifts of antisera; R. G. Oshima, J. Couvreur and M.Kurpakus for their discussions of this manuscript; and R.Logan for technical assistance. This work was supported inpart by grants to K.T. Trevor from the National Institute ofHealth (GM-44582-01) and the March of Dimes (5-811). K.T.Trevor is a recipient of a Basil O'Connor Starter Award fromthe March of Dimes.

References

Baribault, H. and Oshima, R. G. (1991). Polarized and functionalepithelia can form after the targeted inactivation of both mousekeratin 8 alleles. J. Cell Biol. 115, 1675-1684.

Bcrshadsky, A. D. and Vasiliev, J. M. (1988). In Cytoskdeton (ed. P.Siekevitz). New York and London: Plenum Press.

Boiler, K. and Kemler, R. (1983). In vitro differentiation ofembryonal carcinoma cells characterized by monoclonal antibodies


against embryonic cell markers. In Teralocarcinoma Stem Cells.Cold Spring Harbor Conferences on Cell Proliferation, vol. 10 (ed.L. M. Silver, G. R. Martin, S. Strickland), pp. 39-49. New York:Cold Spring Harbor Laboratory Press.

Bologna, M., Allen, R. and Dulbecco, R. (1986). Organization ofcytokeratin bundles by desmosomes in rat mammary cells. J. CellBiol. 102, 560-567.

Boyer, B., Tucker, G. C , Valles, A. M., Franke, W. W. and Thiery,J. P. (1989). Rearrangements of desmosomal and cytoskeletalproteins during the transition from epithelial to fibroblastoidorganization in cultured rat bladder carcinoma cells. J. Cell Biol.109, 1495-1509.

Brulet, P., Babinet, C , Kemler, R. and Jacob, F. (1980). Monoclonalantibodies against trophectoderm-specific markers during mouseblastocyst formation. Proc. Nat. Acad. Sci. USA 77, 4113-4117.

Cervera, M., Dreyfuss, G. and Penman, S. (1981). Messenger RNA istranslated when associated with the cytoskeletal framework innormal and VSV-infectcd HeLa cells. Cell 23, 113-120.

ChLsholm, J. C. and Houliston, E. (1987). Cytokeratin filamentassembly in the preimplantation mouse embryo. Development 101,565-582.

Chomczynskl, P. and Sacchi, N. (1987). Single step method of RNAisolation by acid guanidinium thiocyanate-phenol-chloroformextraction. Anal. Biochem. 162, 156-159.

Cooper, A. R., Taylor, A. and Hogan, B. L. M. (1983). Changes in therate of laminin and entactin synthesis in F9 embryonal carcinomacells treated with retinoic acid and cyclic AMP. Develop. Biol. 99,510-516.

Coulombe, P. A., Hutton, M. E., Letal, A., Hebert, A., Paller, A. S.and Fuchs, E. (1991). Point mutations in human keratin 14 genes ofcpidermolysis bullosa simplex patients: Genetic and functionalanalyses. Cell 66, 1301-1311.

Cowin, P. and Garrod, D. R. (1983). Antibodies to epithelialdesmosomes show wide tissue and species cross-reactivity. Nature302, 148-150.

Cowin, P., Kapprell, H.-P., Franke, W. W., Tamkun, J. and Hynes,R. O. (1986). Plakoglobin: A protein common to different kinds ofintercellular adhering junctions. Cell 46, 1063-1073.

Dembitzer, H. M., Herz, F., Schermer, A., Wolley, R. C. and Koss,L. G. (1980). Desmosomes development, an in vitro model. J. CellBiol. 85, 695-702.

Duden, R. and Franke, W. W. (1988). Organization of desmosomalplaque proteins in cells growing at low calcium concentrations. J.Cell Biol. 107, 1049-1063.

Durkin, M. E., Phillips, S. L. and Chung, A. E. (1986). Control oflaminin synthesis during differentiation of F9 embryonal carcinomacells. Differentiation 32, 260-266.

Feinberg, A. P. and Vogelstein, B. (1984). A technique forradiolabelling DNA restriction endonuclease fragments to highspecific activity. Addendum. Anal. Biochem. 137, 266-267.

Fey, E. G., Wan, K. M. and Penman, S. (1984). Epithelialcytoskeletal framework and nuclear matrix-intermediate filamentsscaffold: Three-dimensional organization and protein composition.J. Cell Biol. 98, 1973-1984.

Fleming, T. P., Garrod, D. R. and Elsmore, A. J. (1991). Desmosomebiogenesis in the mouse preimplantation embryo. Development112, 527-539.

Fleming, T. P. and Johnson, M. H. (1988). From egg to epithelium.Anna. Rev. Cell Biol. 4, 459-485.

Franke, W. W., Cowin, P., Schmelz, M. and Kapprell, H.-P. (1987).The desmosomal plaque and the cytoskeleton. In JunctionalComplexes of Epithelial Cells (ed. G. Bock and S. Clark), pp. 26-44. Chichester, Sussex: John Wiley and Sons.

Garrod, D. R., Parrish, E. P., Mattey, D. L., Marston, J. E.,Measures, H. R. and Vilela, M. J. (1990). Desmosomes. InMorphoregulatory Molecules (ed. G. M. Edelman, B. A.Cunningham and J. P. Thiery), pp. 315-339. The NeurosciencesInstitute Publications Series: John Wiley and Sons, Chichester,Sussex.

Goldman, R. D., ZackrofT, R. V. and Steiner, P. M. (1990).Intermediate filaments: An overview. In Cellular and MolecularBiology of Intermediate Filaments (ed. R. D. Goldman and P. M.Steinert), pp. 3-17. New York: Plenum Press.

Green, K. J., Goldman, R. D. and Chisholm, R. L. (1988). Isolation

of cDNAs encoding desmosomal plaque proteins: Evidence thatbovine desmoplakins I and II are derived from two mRNAs and asingle gene. Proc. Nat. Acad. Sci. USA 85, 2613-2617.

Green, K. J. and Jones, J. C. R. (1990). Interaction of intermediatefilaments with the cell surface. In Cellular and Molecular Biology ofIntermediate Filaments (ed. R. D. Goldman and P. M. Steinert),pp. 147-171. New York: Plenum Press.

Green, K. J., Parry, D. A. D., Steinert, P. M., Virata, M. L. A.,Wagner, R. M., Angst, B. D. and Nilles, L. A. (1990). Structure ofthe human desmoplakins: Implications for functions in thedesmosomal plaque. J. Biol. Chem. 265, 2603-2612.

Grover, A. and Adamson, E. D. (1986). Evidence for the existence ofan early common biochemical pathway in the differentiation of F9cells into visceral or parietal endoderm: Modulation by cyclicAMP. Develop. Biol. 114, 492-503.

Gyoeva, F. K. and Gelfand, V. I. (1991). Coalignment of vimentinintermediate filaments with microtubules depends on kinesin.Nature 353, 445-448.

Hogan, B. L. M., Barlow, D. P. and Tilly, R. (1983). F9teratocarcinoma cells as a model for the differentiation of parietaland visceral endoderm in the mouse embryo. Cancer Surveys 2.115-140.

Hu, L. and Gudas, L. J. (1990). Cyclic AMP analogs and retinoic acidinfluence the expression of retinoic acid receptor alpha, beta, andgamma mRNAs in F9 teratocarcinoma cells. Mol. Cell. Biol. 10,391-396.

Jackson, B. W., Grund, C , Schmid, E., Burki, K., Franke, W. W.and lllmensee, K. (1981). Formation of cytoskeletal elementsduring mouse embryogenesis. Differentiation 17, 161-179.

Jones, J. C. R. and Goldman, R. D. (1985). Intermediate filamentsand the initiation of desmosome assembly. J. Cell Biol. 101, 505-517.

Jones, J. C. R. and GreUing, K. A. (1989). Distribution ofdesmoplakin in normal and cultured human keratinocytes and inbasalcell carcinoma cells. Cell Motil. Cytoskel. 13, 181-194.

Kapprell, H.-P., Cowin, P. and Franke, W. W. (1987). Biochemicalcharacterization of the soluble form of the junctional plaqueprotein, plakoglobin, from different cell types. Eur. J. Biochem.166, 505-517.

Koch, P. J., Walsh, M. J., Schmelz, M., Goldschmidt, J. D.,Zimbelmann, R. and Franke, W. W. (1990). Identification ofdesmoglein, a constitutive desmosomal glycoprotein, as a memberof the cadherin family of cell adhesion molecules. Eur. J. Cell Biol.53, 1-12.

Kreis, T. E. (1990). Role of microtubules in the organization of theGolgi apparatus. Cell Motil. Cytoskel. 15, 67-70.

Kurklnen, M., Barlow, D. P., Helfman, D. M., Williams, J. G. andHogan, B. L. M. (1983). Isolation of cDNA clones for basal laminacomponents: Type IV collagen. Nucl. Acids Res. 11, 6199-6209.

Lane, E. B., Hogan, B. L. M., Kurkinen, M. and Carrels, J. I. (1983).Co-expression of vimentin and cytokeratins in parietal endodermcells of early mouse embryo. Nature 303, 701-704.

Lehtonen, E., Lehto, V.-P., Paasivuo, R. and Virtanen, I. (1983).Parietal and visceral endoderm differ in their expression ofintermediate filaments. EMBO J. 2, 1023-1028.

Luisa, M., Virata, A., Wagner, R. M., Parry, D. A. D. and Green, K.J. (1992). Molecular structure of the human desmoplakin I and IIamino terminus. Proc. Nat. Acad. Sci. USA 89, 544-548.

Mechanic, S., Raynor, K., Hill, J. E. and Cowin, P. (1991).Desmocollins form a distinct subset of the cadherin family of celladhesion molecules. Proc. Nat. Acad. Sci. USA 88, 4476-4480.

Nilles, L. A., Parry, D. A. D., Powers, E. E., Angst, B., Wagner, R.M. and Green, K. J. (1991). Structural analysis and expression ofhuman desmoglein: A cadherin-like component of thedesmosomes. J. Cell Sci. 99, 809-821.

Oshima, R. G. (1981). Identification and immunoprecipitation ofcytoskeletal proteins from murine extraembryonic endodermalcells. J. Biol. Chem. 256, 8124-8133.

Oshima, R. G. (1982). Developmental expression of murineextraembryonic endodermal cytoskeletal proteins. J. Biol. Chem.257, 3414-3421.

Oshima, R. G., Howe, W. E., Klier, F. G., Adamson, E. D. andShevinsky, L. H. (1983). Intermediate filaments protein synthesis inpreimplantation murine embryos. Develop. Biol. 99, 447-455.


Oshimu, R. G., Trevor, K., Shevinsky, L. H., Ryder, O. A. andCecena, G. (1988). Identification of the gene coding for the Endo Bmurine cytokeratin and its methylated, stable inactive state inmouse nonepithelial cells. Genes Develop. 2, 505-516.

Overton, J. (1962). Desmosome development in normal andreassociating cells in the early chick blastoderm. Develop. Biol. 4,532-548.

Pasdar, M., Krzeminski, K. A. and Nelson, W. J. (1991). Regulationof desmosome assembly in MDCK epithelial cells: Coordination ofmembrane core and cytoplasmic plaque domain assembly at theplasma membrane. J. Cell Biol. 113, 645-655.

Pasdar, M. and Nelson, W. J. (1988a). Kinetics of desmosomeassembly in Madin Darby canine kidney epithelial cells: Temporaland spatial regulation of desmoplakin organization andstabilization upon cell-cell contact. I. Biochemical analysis. J. CellBiol. 106, 677-685.

Pasdar, M. and Nelson, W. J. (1988b). Kinetics of desmosomeassembly in Madin Darby Canine Kidney epithelial cells: Temporaland spatial regulation of desmoplakin organization andstabilization upon cell-cell contact. II. Morphological analysis. J.Cell Biol. 106, 687-695.

Pasdar, M. and Nelson, W. J. (1989). Regulation of desmosomeassembly in epithelial cells: Kinetics of synthesis, transport, andstabilization of desmoglein I, a major protein of the membrane coredomain. J. Cell Biol. 109, 163-177.

Ponte, P., Ng, S.-Y., Engel, J., Gunning, P. and Kedes, L. (1984).Evolutionary conservation in the untranslated regions of actinmRNAs: DNA sequences of a human beta -actin cDNA. Nucl.Acids Res. 12, 1687-1696.

Sambrook, J., Fritsch, E. F. and Maniatis, T. (1989). In MolecularCloning. New York: Cold Spring Harbor Laboratory Press.

Schmidt, J. W., Brugge, J. S. and Nelson, W. J. (1992). pp60srctyrosine kinase modulates P19 embryonal carcinoma cell fate byinhibiting neuronal but not epithelial differentiation. J. Cell Biol.116, 1019-1033.

Singh, U. K., Maheshwari, R. K., Damewood IV, G. P., Stephensen,C. B., Oliver, C. and Freidmen, R. H. (1988). Interferon alters

intracellular transport of vesicular stomatitis virus glycoprotein. J.Biol. Reg. Homeost. Ag. 2, 53-63.

Skalli, O. and Goldman, R. D. (1991). Recent insights into theassembly, dynamics and function of intermediate filamentnetworks. Cell MotU. Cytoskel. 19, 67-79.

Stappenbeck, T. S. and Green, K. J. (1992). The desmoplakincarboxyl terminus coaligns with and specifically disruptsintermediate filament networks when expressed in cultured cells. J.Cell Biol. 116, 1197-1209.

Steinberg, M. S., Hisato, S., Giudice, G. J., Shida, M., Patel, N. H.and Blaschuk, O. W. (1987). On the molecular organization,diversity and functions of desmosomal proteins. In JunctionalComplexes of Epithelial Cells (ed. G.Bock and S. Clark), pp. 3-25.Chichester, Sussex: John Wiley and Sons.

Strickland, S. and Mahdavi, V. (1978). The induction ofdifferentiation of teratocarcinoma stem cells by retinoic acid. Cell15, 393-403.

Strickland, S., Smith, K. K. and Marotti, K. R. (1980). Hormonalinduction of differentiation in teratocarcinoma stem cells:Generation of parietal endoderm by retinoic acid and dibutyrylcAMP. Cell 21, 347-355.

Trevor, K. T. (1990). Disruption of keratin filaments in embryonicepithelial cell types. New Biol. 2, 1004-1014.

Trevor, K. and Oshima, R. G. (1985). Preimplantation mouseembryos and liver express the same type 1 keratin gene product. J.Biol. Chem. 260, 15,885-15,891.

Vale, R. D. (1987). Intracellular transport using microtubule-basedmotors. Anna. Rev. Cell Biol. 3, 347-378.

Vassar, R., Coulombe, P. A., Degenstein, L., Albers, K. and Fuchs,E. (1991). Mutant keratin expression in transgenic mice causesmarked abnormalities resembling a human genetic skin disease.Cell 64, 365-380.

Vasseur, M., Duprey, P., Brulet, P. and Jacob, F. (1985). One geneand one pseudogene for the cytokeratin endo A. Proc. Nat. Acad.Sci. USA 82, 1155-1159.

(Received 3 December 1991 - Accepted, in revised form,26 May 1992)

distribution of desmosomal proteins in f9 embryonal carcinoma

Documents