Syndecan-l, a Cell-Surface Proteoglycan, Changes in Size and Abundance when Keratinocytes Stratify

Ralph D. Sanderson, Michael T. Hinkes, and Merton Bernfield Department of Pediatrics, Stanford University School of Medicine, Stanford, California, U.S.A.

In epidermis, keratinocytes in the basal cell layer differen­ evaluated before and after calcium-induced stratification and tiate, lose their attachment to the underlying extracellular differentiation. Cells growing as an unstratified monolayer matrix, and form extensive intercellular adhesions as they express a higher molecular mass form of syndecan-1 than do stratify. The alterations in cell-matrix and cell-cell adhesion stratified cells (modal relative mass of 160 kD versus required for keratinocyte stratification result from changes in 110 kD). This structural difference is due to larger and more the expression of numerous adhesion molecules. Syndecan-1, heparan sulfate chains on syndecan-1 from monolayer cells. a member of a family of cell-surface proteoglycans, is known In addition, the amount of cell-surface syndecan-1 changes to bind cells to interstitial matrix. Syndecan-1 localizes to with stratification; stratified cultures show approximately specific layers of mouse epidermal keratinocytes; its expres­ 2.5 times more syndecan-1 per cell than do unstratified cul­ sion is modest in the basal layer, heavy in the suprabasal tures, but do not significantly change the level of syndecan- layers, but absent from the most superficial, terminally dif­ 1-specific mRNA. Thus, the structure and amount of syn­ ferentiated layers. This layer-specific difference suggests that decan-1 may be regulated to meet the changing adhesive syndecan-1 expression changes with keratinocyte differen­ requirements of stratifying keratinocytes. ] Invest Dermatol tiation. To assess this hypothesis, syndecan-1 expression was 99:390-396, 1992 .

he outer protective barrier of the skin, the epidermis, tirely clear, but is probably the result of cha.nges in many molecules is a stratified epithelium composed of layers of kerati­ that act in concert. For example, upon keratinocyte differentiation, nocytes at increasing stages of differentiation. The quantitative and/or qualitative changes occur in desmosomes [4- mitotically active basal layer gives rise to differentiat­ 7], E- and P-cadherin [8-10], integrins [11-14], cell surface affinity ing cells that migrate upward through the spinous and for lectins [15-18] and proteoglycan content [19,20]. granularT layers, become keratinized in the cornified layers, and are Syndecan-l, a member of a family of integral membrane proteo­ eventually sloughed from the skin as dead squames. This upward glycans, behaves as a matrix receptor and binder of growth factors migration requires that cells destined to leave the basal layer break [21]. Syndecan-l localizes to the basolateral surface of simple epith­ contact with the underlying extracellular matrix and form new elia and binds these cells to extracellular matrix components includ­ associations with suprabasal cells, presumably through specific cell­ ing fibronectin [22], interstitial collagens [23]' and thrombospondin cell contacts. Compared to basal cells, differentiating keratinocytes [24]. When cross-linked at the cell surface, syndecan-1 associates are less adherent to tissue-culture plastic [1], types I and IV collagen with the actin cytoskeleton [25] and NMuMG cells made deficient [2], and fibronectin and basement membranes [3]. This reduced in syndecan-l by transfection with antisense eDNA become fusi­ cell-matrix adhesion suggests that keratinocyte stratification is ac­ form in shape indicating that syndecan-l is required for the mainte­ companied by loss or modification of matrix receptors. Moreover, nance of epithelial cell morphology [26]. aggregation studies show that a mixture of differentiating keratino­ Syndecan-1 is also expressed on stratified epithelia where it local­ cytes and basal cells will sort themselves out in homotypic fashion, !zes over t~e entir~ cell surface. Syndecan-l from stratified epithelia signifying preferential adhesion between cells at similar stages of IS ~mal~er 111 relative molecular mass than syndecan-l from simple. differentiation [4]. Therefore, changes in cell-matrix and cell-cell epithelia (model values of92 vs 160 kD) due to a reduction in size adhesivity occur during keratinocyte differentiation and stratifica­ and number of glycosaminoglycan chains [27]. Although the func­ tion. tion of the smaller form of syndecan-1 that surrounds stratified The basis for these changes in keratinocyte adhesivity is not en- epithelia is unknown, its pattern of expression suggests it may medi­ ate cell-cell adhesion. In addition, syndecan-1 and the heparin­ binding growth factor, bFGF, have a similar pattern of distribution Manuscript received March 11, 1992; accepted for publication May 28, in the epidermis [28,29], suggesting that syndecan-l heparan sulfate 1992. on these cells may mediate bFGF signal transduction, as does hepa­ Part of these data have appeared in abstract form (J Cell Bioi 107:803a, ran sulfate on other cell types [30,31]. 1988). In the present study, we examined syndecan-1 expression on kera­ Michael Hinkes's and Merton Bernfield's current address: Joint Program tinocytes that are maintained as a monolayer and then induced to in Neonatology, Harvard Medical School, Boston, MA 02115. Reprint requests to: Dr. Ralph D. Sanderson, Department of Pathology, stratify and differentiate by increasing the Ca++ concentration in University of Arkansas for Medical Sciences, 4301 West Markham Street, the culture medium [5]. We find that syndecan-1 becomes smaller Little Rock, AR 72205. in size but increases in amount at the cell surface with keratinocyte Abbreviations: stratification. These changes in syndecan-1 may be one mechanism bFGF: basic fibroblast growth factor used by differentiating cells to meet their changing adhesive and NMuMG: normal murine mammary gland growth requirements.

0022-202X/92/S05.00 Copyright © 1992 by The Society for Investigative Dermatology, Inc.

390 VOL. 99, NO.4 OCTOBER 1992 SYNDECAN-1 EXPRESSION CHANGES ON KERATINOCYTES 391

MATERIALS AND METHODS into ice-cold buffer immediately upon excision from the mouse and extracted as described above for cells. Cell Culture Primary mouse keratinocytes were prepared and cultured by the method of Yuspa et al. [32]. Briefly, skins were Purification of Syndecan-1 Extracts in buffer A were diluted removed from newborn BALB/c mice (1-2 days old), trypsinized with 50 mM TBS, pH 7.4 to a final concentration of 0.6 M urea, for 6-12 h at 4· C and the dermis mechanically separated from the 0.1% Triton X-I00. Monoclonal 281-2 bound to Sepharose 4B epidermis. Epidermal cells were dispersed by stirring and plated at beads (281-2 beads) was added to diluted extracts and gently rocked 4.2 X 106 cells per 60 mm culture dish in Delbecco's modification overnight at 4· C. The following day, the beads were placed in a of Eagle's Minimal Essential Medium prepared without calcium polyprep disposable column (Bio-Rad, Richmond, CA) and washed chloride (Whittaker Bioproducts, Walkersville, MD) and contain­ with 20 mM acetic acid (pH 3.1) to remove nonspecifically bound ing 8% chelex-treated fetal calf serum, 100 U /ml penicillin, 100 molecules. Syndecan-1 was isolated from conditioned media follow­ J1.gj rnl streptomycin and 0.07 mM calcium chloride. The following ing centrifugation and addition of I M Tris (pH 7.4) to a final con­ day, unattached cells were removed and fresh media containing centration of 20 mM. 281 -2 beads were introduced, incubated and 0 .05 mM Ca++ was introduced. Under these conditions, cells reach washed as above. For removal of glycosaminoglycan chains, synde­ confluence as a monolayer in 3 - 5 days and can be maintained in this can-l bound to 281-2 beads was incubated with 0.1 unit/ml of proliferative stage even at high cell density. To induce stratification heparitin sulfate lyase (heparitinase; Seikagaku, Rockville, MD) or of keratinocytes, the calcium chloride concentration in the medium 0.05 unit/ml chondroitin ABC lyase (ABCase; Seikagaku, Rock­ 35 was increased to 1.2 mM. For radiolabeling, 100 ,uCi/ml H2 S04 ville, MD) using previously described conditions [36]. Following (leN Biomedicals, Costa Mesa, CA) were added to the medium for digestions, beads were washed extensively with dH20. For prepara­ 48 h. tion of samples for SDS-PAGE gels, syndecan-l was eluted from beads by boiling in SDS-PAGE sample buffer [2% SDS (wt/vol), bn.:rnunolocalization Immunolocalizations were with MoAb 5% glycerol (vol/vol), 0.025% bromophenol blue, 140 mM Tris 281-2, a rat anti-mouse IgG2a specific for an epitope on the extra­ and 60 mM boric acid, pH 8.0]. cellular domain of the syndecan-l core protein [33]. Tissue was fixed in 3.7% paraformaldehyde, embedded in paraplast. sectioned, SDS-PAGE and Western Blotting Syndecan-1 was fraction­ deparaffinized in xylene and re-hydrated in a graded s e~ ie.s of alco­ ated on a 3.8-20% SDS polyacrylamide gel containing urea and hols, including incubation in absolute methanol conta1l1lng 0.3% boric acid, as described by Koda et al. [23], and transferred by the method of Towbin et al. [37] to Gene-Trans (Plasco, Woburn, H 2 0 2 to block endogenous peroxidase activity. Sections were blocked with 190 ,ug/ml rabbit IgG followed by incubation over­ MA), a cationic nylon membrane. Membranes were then blocked night at 4· C with 3.8 ,ug/mI281-2 or as a control, Mel-14 [34] , a rat with blotto [38] and probed overnight at 4 0 C with 125I-labeled anti-mouse monoclonal IgG2a specific for a lymphocyte homing antibody 281-2, washed and exposed to Kodak X-OMAT AR film receptor. Sections were washed six times in PBS and incubated with at -80·C. 5 J1.g/ml biotinylated rabbit anti-rat IgG (no cross reaction. with Determination of Glycosaminoglycan Size Radiosulfate-la­ mouse IgG; Vector Laboratories. Burlingame, CA) for 30 mmutes beled syndecan-l bound to 281-2 beads was treated with nitrous at room temperature. After washing, sections were incubated for 30 acid to remove heparan sulfate or chondroitinase to remove chon­ minutes at room temperature with an avidinhorseradish peroxidase droitin sulfate [36], and the digested glycosaminoglycan was reagent prepared from a Vectastain Elite ABC kit (Vec.tor ~aborato­ washed from the beads and collected. Glycosaminoglycan remain­ ries). Reaction product was generated with 0.05% dlaml11obenzl­ ing on syndecan-1 was released with the addition of 0.1 M sodium dine tetrahydrochloride containing 0.015% H 0 in PBS. Staining 2 2 hydroxide in 1 M potassium borohydride for 24 hat 37"C. Glyco­ of sections was enhanced with 0.5% CUS04 in 0.9% N aCI for 5 saminoglycan was fractionated on Sepharose CL-6B (Pharmacia) in minutes. Following mounting with covers lips, photographs were a buffer of 2% (wt/vol) SDS, 0.15 M NaCI, 50 mM sodium acetate taken with a Zeiss Photomicroscope II equipped with a blue HOY A (pH 5.0) and 0.02% (wt/vol) sodium azide. Fractions were mixed 49 rnm filter and using Kodak Tech Pan film. For immunofluores­ with AquaMix (ICN Biomedicals, Costa Mesa, CA) scintillation cence, primary keratinocytes (3 X 106 cells/35 mm dish) were mixture and radioactivity was quantitated by an LKB 1219 liquid added to preformed collagen gels [35] cast at a final collagen con­ scintillation counter. Size of the glycosaminoglycan chains was de­ centration of 1 mg/ml (rat tail. type I collagen). The cells and gel termined by comparison with a molecular size calibration curve were submerged in culture media containing 1.2 mM calcium and, generated with chondroitin sulfate standards [39]. after 72 hours, the medium was removed from the cultures and ice-cold Tissue-Tek (Miles Scientific, Naperville, IL) was added to RNA Isolation and Northern Blotting RNA isolation and cells growing on collagen gels. After incubation for 30 minutes on Northern blots were prepared as previously described [40]. Briefly, ice, the gel was frozen with dry ice, removed from the culture dish poly (A) RNA was extracted from cultured cells with 4 M guani­ and 6,um cryostat sections prepared. Following blocking with 10 dine isothiocyanate in 5 mM sodium citrate, pH 7.0,0.1 M 2-mer­ J1.gjml normal rat IgG, sections were incubated with biotinylated captoethanol, 0.5% N-lauryl sarcosine. Following purification by MoAb 281-2 (2,ug/ml) for 30 minutes, washed in PBS and incu­ CsCI density centrifugation, phenol/chloroform extraction and bated with avidin/FITC (5 ,ug/ml) for 30 minutes. Following oligo (dT)-cellulose chromatography, RNA was separated by elec­ washing in PBS, sections were mounted in 90% glycerol, 10% PBS, trophoresis in a 1.2% agarose-formaldehyde gel. RNA was blotted and viewed on a Zeiss Photomicroscope II equipped with epifluore­ to GeneScreen (New England Nuclear, Boston, MA) and rrobed scence. Photographs were taken using Kodak Tri-X film exposed at with a RNA probe prepared by the in vitro transcription 0 the 5' ASA 800. EcoRI-SacI fragment of syndecan-l clone PM-4. Extraction of Cells and Tissues Cell layers were washed three Quantitation of Syndecan-1 Culture medium conditioned by times in ice-cold 10 mM TBS, pH 7.4, incubated with TBS con­ stratified and unstratified keratinocytes was collected and centri­ taining 0.5 mM EDTA and the cells dislodged from the dish with a fuged to remove cell debris. Cell layers were washed 3 - 4 times cell scraper. Following washing of cells two times with ice-cold with ice-cold PBS, incubated with ice-cold PBS containing 0.5 mM TBS, 0.5 mM EDTA, cell pellets were extracted with 6 M urea, 1 % EDTA, and the cells harvested from the dish with a cell scraper. To Triton X-I00, 10 mM Tris-HCI (pH 7.4), 5 mM benzamidine, remove cell surface syndecan-l, the cells were pelleted by centrifu­ 5 ruM N-ethylmaleimide, 1 mM PMSF, 0.5 mM EDTA, 5,ug/ml gation, resuspended in ice-cold PBS with 0.5 mM EDTA and ex­ pepstatin A and 0.15 M NaCI (buffer A). Extracts were vortexed posed to 20 ,ug/ml trypsin (Sigma, St. Louis, MO; 2X crystalized, periodically during a 4 h incubation on ice then centrifuged at Type III) for 5 minutes on ice with gentle agitation. Soybean trypsin 21,000 X g for 15 minutes at 4· C and the supernatants analyzed. inhibitor was then added to a final concentration of 100 ,ug/ml, cells For extraction of syndecan-l from newborn skin, tissue was placed were pelleted by centrifugation, and the supernatant containing 392 SANDERSON ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

the substratum, becomes intense in the spinous layers and dimin­ ishes in the granular layers before disappearing completely in corni­ fied layers. No cells in the dermis stain. This pattern of syndecan-l expression is identical with that seen in other stratified epithelial tissues, such as the tongue and vagina [28]. The attenuation of staining for syndecan-l where basal cells contact the substratum may be due to masking of the 281-2 epitope because staining of murine skin with a polyclonal antisera to syndecan-l indicates that syndecan-l is present at this site [43]. Stratified keratinocytes growing on collagen gels in culture are three to four layers thick. Cross sections of these cultures reveal that syndecan-l surrounds the stratified cells but is absent from the apical surface of the most superficial cell layer (Fig lB), a pattern similar to that seen in vivo.

Syndecan-l Size Decreases with Keratioocyte Stratifica­ tion To establish whether the molecular mass of syndecan-l changes with keratinocyte stratification, we compared syndecan-l extracted from monolayer and stratified cells. Radiosulfate was added to cultures maintained as mono layers (0.05 mM Ca++) and at this time, some monolayer cultures received media containing

1.2 mM Ca++ to induce stratification. After 48 hours, the 35S04-la- beled syndecan-l from monolayer and stratified cells was extracted, immunoisolated with MoAb 281-2 and its relative mass compared by SDS-PAGE. Syndecan-l migrates on PAGE as a smear, corre­ sponding to a range of molecular masses. Syndecan-l from mono­ layer cells is larger (model size approximately 160 kD) than that from stratified cells (approximately 110 kD; Fig 2). A similar dif­ ference in relative mass is seen with syndecan-l isolated from media Figure 1. Syndecan-1 localization in newborn skin and cultured stratified conditioned by monolayer and stratified keratinocytes. Syndecan-l keratinocytes. (A) Section of paraffin-embedded newborn mouse skin immunoisolated from newborn skin is identical in relative mass to stained with antibody 281-2. Immunoperoxidase stain surrounds cells of the basal and spinous.layers, but is weak or absent in the region of contact with the basement membrane (arrows). Cells of the most superficial regions of the granular layer and cells within the cornified layer do not stain (arrowhead). (B) Immunofluorescence staining for syndecan-1 on a frozen section of cultured stratified keratinocytes growing on a type I collagen gel. The cell layer is 3 - 4 cells thick. Stain surrounds the individual cells, but is absent from the apical surface of cells in the most superficial layer (arrowheads). Bar, 10 Jim.

released syndecan-l analyzed by a previously described radioimmu­ kD noassay [41]. Briefly, a wet cationic membrane (Gene-Trans) was placed into an immunodot apparatus (V & P Scientific, San Diego, CA) and samples of culture medium and trypsin-released cell surface material from stratified and unstratified cells were loaded onto the 200- membrane using mild vacuum. Also loaded were known amounts of syndecan-l added to either culture medium unconditioned by cells (culture medium standards) or to PBS containing 0.5 mM EDTA, 20 p,g/ ml trypsin and 100 p,g/ ml soybean trypsin inhibitor (trypsin­ released standards). After loading, the membrane was placed in 0.025% glutaraldehyde to fix protein to the membrane. The mem­ 92.5- brane was then rinsed briefly with distilled H 20, blocked with blotto, washed once with TBS pH 7.4 containing 1 % FCS and 0.3% Tween 20 (buffer B), and then incubated overnight at 4 0 C in buffer B containing 125I-labeled MoAb 281-2. The membrane was 69- washed six times in buffer B to remove unbound antibody, dried, and the dots cut out and counted in a gamma counter (Hewlett Packard, Palo Alto, CA). Quantity of DNA in the cell pellets re­ 46- maining after trypsinization was determined by the method of Ka­ puscinski and Skoczylas [42].

'--__~, 'L. __.....J RESULTS Keratinocytes in the Epidermis and in Culture Have a Simi­ cells media lar Pattern ofSyndecan-l Expression To determine whether keratinocytes stratifying in vitro provide a useful system to study Figure 2. Syndecan-1 decreases in size upon keratinocyte stratification. Radiolabeled syndecan-1 was extracted from cells growing as monolayers or changes in syndecan-l expression, we compared syndecan-l locali­ 48 h after calcium induced stratification. Monolayer and stratified cultures zation in cross sections of newborn mouse skin and stratified cul­ were of the same age. Syndecan-1 from cell extracts and culture media was tures of mouse keratinocytes. In the epidermis, syndecan-l isolated by 281-2 affinity purification followed by SDS-PAGE. For compari­ surrounds the surfaces of basal, spinous and lower granular layers son, syndecan-1 from newborn skin was extracted and purified in an identi­ (Fig lA and [28,43]). Staining is attenuated where basal cells contact cal manner and visualized by Western blotting using 1251_281_2. VOL. 99, NO.4 OCTOBER 1992 SYNDECAN·1 EXPRESSION CHANGES ON KERATINOCYTES 393 tha t of syndecan-1 extracted from keratinocytes stratifying in cul­ kD ture (Fig 2). 50 25 15 syndecan-t Core Protein Size Does Not Change with Kera­ tinocyte Stratification The relative mass of the syndecan-1 core 20 protein from monolayer and stratified keratinocytes was evaluated on Western blots. Following treatment with enzymes that remove '-9 both heparan sulfate and chondroitin sulfate chains, the syndecan-1 .? core proteins from monolayer and stratified cells were identical in 16 ~ relative mass (approximately 69 kD; Fig 3). Bands migrating at 0.. higher molecular mass (approximately 140 kD) are seen following () the enzyme digestions, possibly representing core protein aggre­ 0'<:3' 12 gates. en lO Stratified cells, but not monolayer cells, express intact syndecan-1 M that migrates at the size of glycosaminoglycan-free core protein (Fig I- Z 3; lanes 1 and 4). The majority of syndecan-1 from both monolayer UJ 8 and stratified cells appears to contain solely heparan sulfate because () II: enzymatic removal of heparan sulfate only yields core protein UJ bands. However, studies with 35S04-labeled syndecan-1 indicate 0.. that a small portion of the molecules from both cultures also contain 4 chondroitin sulfate; no molecules were detected that contained solely chondroitin sulfate (data not shown). GlycosaminogJycan Composition of Syndecan-l Changes 0 with Keratinocyte Stratification Syndecan-1 bound to im­ rnunobeads was treated with either heparitinase or chondroitinase 0 0.2 0.4 0.6 0.8 1.0 and the residual glycosaminoglycan chains removed with alkali. Analysis on CL-6B Sepharose columns shows that the heparan sul- Kav Figure 4. Syndecan-1 heparan sulfate chains are larger on monolayer than on stratified keratinocytes. Radiolabeled syndecan-l extracted from mono­ layer and stratified cells was prepared as in Fig 2 and treated with chondroi­ Q) Q) tinase to remove chondroitin sulfate. The heparan sulfate released from the II) II) core protein with alkali/potassium borohydride was then fractionated on CU CU Sepharose CL-6B in 2% SDS buffer. Heparan sulfate from monolayer kerati­ U U c m c m nocytes has a Kav of 0.43 and from stratified keratinocytes a Kav of 0.50. CU cu (,,) < (,,) < Q) + Q) + 'C 'C Q) Q) Q) Q) C C II) II) II) II) fate chains derived from monolayer keratinocytes behave as larger >- >- CU CU II) CU CU II) chains than those from stratified keratinocytes (modal size 34 vs . >- ~ >-~ (,,) -(,,) en en 21 kD; Fig 4). Chondroitin sulfate represents only 5 -10% of the -CU en en CU xx total radiosulfate-labeled glycosaminoglycan and the chains are c X X c - + + - + + identical in size (8 kD) on syndecan-1 from monolayer and strati­ fied keratinocytes (data not shown; cf. [271). Thus, assuming a simi­ kD lar extent of sulfation, the difference in syndecan-1 size between ~nonolayer and st~atified cells is at least partially due to a difference m the Slze of thelr heparan sulfate chains. Amount of Syndecan-l at the Cell Surface Increases with 200- Keratinocyte Stratification Syndecan-l localization in the epi­ dermis indicates that it is more abundant on suprabasal cells than on basal cells (Fig lA). Therefore, the amount of syndecan-l at the cell surface or shed into the culture medium was compared in mono­ layer and stratified cultures. Keratinocytes were grown to con­ 92.5- fluence as monolayers in 0.05 mM Ca++ (3 to 5 d) and then induced to stratify for 2 d by raising the Ca++ to 1.2 mM. Medium was then changed for all cultures and the cells incubated an additional 24 h, 69- after which syndecan-1 on the cell surface or shed into the culture medium was quantitated by radioimmunoassay. After stratification, the amount of syndecan-l per cell increases approximately 2.5-fold 46- at the cell surface and in the culture medium (Table I). Abundance ofSyndecan-l mRNA Does Not Change Signifi­

~------~I I~ ______~ cantly with Keratinocyte Stratification Syndecan-l mRNA monolayer stratified levels in monolayer and stratified cultures were measured to deter­ mine if the change in amount and size of syndecan-1 after stratifica­ tion correlates with a change in syndecan-l-specific mRNA. Equal Figure J. Syndecan-l core protein size is identical on monolayer and strati­ amounts of poly(A) RNA from monolayer and stratified cultures fied cells. Western blot of syndecan-l before and after removal of glycosa­ nllnoglycan chains is shown. Syndecan-l was extracted, purified, and sub­ (48 h) were examined by Northern blot with an RNA probe specific jected to SDS-PAGE following no treatment (intact syndecan-l), treatment for syndecall-l mRNA. The resulting autoradiogram (Fig 5) reveals W'ith heparitinase (HS Lyase) or with both heparitinase and chondroitinase syndecan-1 mRN A is present in similar relative abundance in both (ABCase). Following transfer to a cationic nylon membrane, syndecan-l was cultures (1.2-fold increase with stratification). Thus, the changes in detected with 12sI_labeled 281-2, which recognizes the core protein. syndecan-l seen after stratification are apparently controlled by 394 SANDERSON ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Table I. Quantitation of Syndecan-1 from Monolayer and area increases upon stratification [44,45). However, an increase in Stratified Cultures' syndecan-1 amount is consistent with the reported increase in prote­ oglycan synthesis during epidermal differentiation [19] despite an Syndecan-1 (ng/Ilg DNA) overall decrease of 60% in non-specific protein synthesis [5]. This Experiment 1 Experiment 2 increase in syndecan-1 amount is not due to increased retention of cell surface syndecan-l because syndecan-l shed into the culture Monolayer medium increases to the same extent. Nor is the increase due to Cell surface 1.8 1.6 retention of shed syndecan-1 between cell layers. The shed proteo­ Conditioned medium 4.6 4.9 glycan lacks the transmembrane and cytoplasmic domains and mi­ Stratified grates at 56 kD [41], whereas the core protein from stratified layers Cell surface 3.9 4.0 Conditioned medium 12.5 13.2 migrates as an intact core protein at 69 kD [27) and Fig 3, even when the cell layers are extracted without prior washing (data not • Syndecan-l from monolayer and stratified keratinocytes growing in culture was shown). purified and the amount of syndecan-l per Ilg of DNA was determined as described in The functional significance of changes in syndecan-1 molecular Materials atld Methods. structure and abundance is not well understood. However, recent studies comparing NMuMG (epithelial) and MPC-ll (B lymph­ oid) syndecan-l demonstrate that the 160 kD form of syndecan-1 post-transcriptional mechanisms. The two species of syndecan-1 isolated from NMuMG cells has greater than a four-fold higher mRNA detected at 2.6 and 3.4 kb are consistent with previously affinity for type I collagen than does the 85 kD form of syndecan-1 reported results [40]. isolated from MPC-ll cells [46]. Similarly, a high molecular weight form of syndecan-1 from embryonic tooth mesenchyme DISCUSSION binds to tenascin better than syndecan-1 from NMuMG cells [47]. Keratinocyte cohesiveness changes during passage from basal to These data are consistent with previous studies which indicate that cornified layers. We now find that the structure and amount of cell surface syndecan-1 are altered during keratinocyte stratification in vitro. The results suggest that syndecan-l is regulated at both the molecular and cellular levels to meet the changing adhesive require­ ments of keratinocytes during differentiation and stratification. a: In this study, the alterations in syndecan-1 found in monolayer w a >­ w and stratified cultures are probably less extensive than the actual e:{ u. differences between basal and supra basal keratinocytes in the epider­ ..J l­ mis. Stratified cultures retain a layer of undifferentiated basal cells o e:{ while monolayer cultures contain some differentiating cells that are Z a: prevented from stratifying by the low Ca++ [1]. Therefore, because o l­ (/) the monolayer and stratified cultures used here contain mixed popu­ :E lations of basal (undifferentiated) and suprabasal (differentiating) cells, the observed in vitro changes in syndecan-l structure and amount likely underestimate those occurring in vivo. Relative to monolayer cells, syndecan-l from stratified keratino­ cytes is partly free of glycosaminoglycan chains and is lower in relative mass (approximately 110versus 160 kD) with smaller hepa­ ran sulfate chains (approximately 21 versus 34 kD) on the same size core protein (approximately 69 kD). However, differences in hepa­ ran sulfate chain size alone cannot account for the difference in relative mass; syndecan-l from stratified cells must also contain fewer heparan sulfate chains. We cannot rule out the possibility that kb the reduction in heparan sulfate chain size is due in part to increased heparitinase activity in stratified cultures. However, if the differ­ ence in heparan sulfate size was due to heparitinase, it is likely that -3.4 heparan sulfate from stratified cultures would have a broader size distribution than that of heparan sulfate from monolayer cultures. - 2.6 However, data in Fig 4 indicate that although the size of heparan sulfate from stratified and monolayer cells differs, the heparan sul­ fate from stratified cells does not exhibit a broader distribution in size than that of heparan sulfate from monolayer cultures. These results extend our previous finding that syndecan-1 from different tissue . types exhibits a molecular polymorphism which correlates with tissue organization [27]. That is, syndecan-1 from simple epithelia (approximately 160 kD) is larger in size and has a greater number of heparan sulfate and chondroitin sulfate chains than that from stratified epithelia (approximately 92 kD). By mani­ pu lating keratinocyte morphology in vitro we have now tested whether syndecan-1 polymorphism is primarily due to tissue type or to cellular organization. Our results establish that in epidermal kera­ tinocytes, changes in syndecan-1 structure are determined by cellu­ Figure 5. The size and relative abundance of syndecan-l mRNA is similar in monolayer and stratified keratinocytes. Poly (A) RNA was isolated from lar organization and/or differentiation. monolayer and stratified cells, and equal amounts were loaded on a 1.2% On a per cell basis, stratified keratinocytes contain up to two and a agarose-formaldehyde gel. The RNA was transferred and hybridized as half times the amount of cell surface syndecan-1 than monolayer described in Materials mId Methods. Scanning densitometry indicates that the keratinocytes (Table I), although the density of syndecan-1 on mRNA detected from stratified cells is 1.2 times more abundant than that monolayer and stratified cells may be similar because cell surface from monolayer cells. VOL. 99, NO.4 OCTOBER 1992 SYNDECAN-l EXPRESSION CHANGES ON KERATINOCYTES 395 differences in glycosaminoglycan chain size and/or comrosition 3. Toda K, Grinnell F: Activation of human keratinocyte fibronectin can affect the interactions of proteoglycans with a variety 0 ligands receptor function in relation to other ligand-receptor interactions. J [48,49J. Invest Dermatol 88:412 - 417, 1987 In addition to the changes in syndecan-l structure and abundance 4. Watt FM: Selective migration of terminally differentiating cells from described in the present work, similar changes occur with cell trans­ the basal layer of cultured human epidermis. J Cell BioI 98:16-21, forma tion. Transformed NMuMG cells express a smaller form of 1984 syndecan-1, pile up in culture and are highly invasive when injected 5. Hennings H, Michael D, Cheng C, Steinert P, Holbrook K, Yuspa SH: into athymic nude mice, while their non-transformed counterparts Calcium regulation of growth and differentiation of mouse epider­ express the larger 160 kD form of syndecan-1, grow as monolayers mal cells in culture. Cell 19:245 - 254, 1980 in culture and are non-invasive [50,51]. Loss or reduction in synde­ 6. Parrish EP, Garrod DR, Mattey DL, Hand L, Steart PV, Weller RO: can-1 expression occurs following hormone-induced transforma­ Mouse antisera specific for desmosomal adhesion molecules of su­ tion of simple epithelia [52] and in skin tumors induced by ultravio­ prabasal skin cells, meninges and meningioma. Proc Nat! Acad Sci USA 83:2657 - 2661, 1986 let irra diation [53]. Thus, in addition to mediating adhesive changes during normal cell differentiation, changes in syndecan-l structure 7. Skerrow CJ, Clelland DG, Skerrow D: Changes to desmosomal anti­ gens and lectin-binding sites during differentiation in normal and/or amount at the cell surface may be important in tumor human epidermis: a quantitative ultrastructural study. J Cell Sci gro"Wth and metastasis. 92:667 - 677, 1989 The changes in syndecan-1 expression seen during keratinocyte 8. Damsky CH, RichaJ, Solter 0, Knusden K, Buck CA: Identification stratification may signify a change in syndecan-1 adhesive function. and purification of a cell surface glycoprotein mediating intercellu­ The larger, heparan sulfate-rich syndecan-l molecules found on the lar adhesion in embryonic and adult tissue. Cell 34:455-466, 1983 monolayer cells may be involved in cell-matrix interactions, as seen 9. Nose A, Takeichi M: A novel cadherin cell ad hesion mol ecule: its in simple epithelia such as mammary epithelial cells. On the other expression patterns associated with implantation and organogenesis hand, the more abundant, smaller, heparan sulfate-poor syndecan-1 of mouse embryos. J Cell BioI 103:2649-2658, 1986 found on the stratified cells may be involved predominantly in 10. Hirai Y, Nose A, Kobayash S, Takeichi M: Expression and role of cell-cell adhesion. Interestingly, removal of syndecan-1 glycosa­ E-Cadherin and P-cadherin adhesion molecules in embryonic histo­ minoglycan chains results in what appears to be core protein aggre­ genesis. 1. Skin Morphogenesis. Dev 105:263 -270,1989 gates (see Fig 3). Recent studies suggest syndecan-2 from the rat 11. Adams JC, Watt FM: Changes in keratinocyte ad hesion during termi­ liver may also form aggregates following the removal of glycosa­ nal differentiation: reduction in fibrone ctin binding precedes aSp! minoglycan chains [54]. Finally, loss of syndecan-1 on terminally integrin loss from the cell surface. Cell 63:425 - 435, 1990 differentiated cells may permit desquamation. 12. Carter WG, Wayner EA, Bouchard TS, Kaur P: The role ofintegrins Changes in syndecan-1 expression may also modulate the expres­ a2p1 and a3pl in cell-cell and cell-substrate adhesion of human sion of other keratinocyte adhesion molecules. Mammary epithelial epidermal cells. J Cell Bioi 110: 1387 - 1404, 1990 cells made deficient in cell surface syndecan-1 by transfection with 13. De Luca M, Tamura RN, Kajiji S, Bondanza S, Rossino P, Cancedda R, an antisense eDNA construct become fusiform in morphology, lose Marchisio PC, Quaranta V: Polarized integrin mediates human ker­ expression of E-cadherin, and show a redistribution of fJ1 -integrins atinocyte adhesion to basal lamina. Proc Nat! Acad Sci USA [55]. If syndecan-1 acts similarly in epidermal keratinocytes, synde­ 87:6888 - 6892, 1990 can-l may be involved in the regulation ofE-cadherin and fJl-inte­ 14. Larjava H, PeltonenJ, Akiyama SK, Yamada SS, Gralnick HR, Uitto J, grin expression and could orchestrate their roles in differentiation Yamada KM: Novel function for P1 integrins in keratinocyte cell­ cell interactions. J Cell BioI 110:803-81 5, 1990 and stratification. In addition to its role in adhesion, syndecan-1 also binds to bFGF 15. Reano A, Faure M, Jacques Y, Reichert U, Schaefer H, Thivolet J: Lectins as markers of human epidermal cell differentiation. Differ­ [21,56,57] and likely binds to other members of the heparin-bind­ entiation 22:206-210,1982 ing growth factor family, such as keratinocyte growth factor [58]. 16. Nemanic MK, Whitehead JS, Elias PM: Alterations in membrane Because binding of these growth factors to heparan sulfate may be sugars during epidermal differentiation: visualization with lectins required for growth factor-mediated signal transduction [30,31 J, and role of glycosidases. J Histochem Cytochem 31:887 -897,1983 their action on keratinocytes could be temporally and spatially re­ 17. Watt FM: Involucrin and other markers of keratinocyte terminal dif­ stricted by the expression of syndecan-1. For instance, syndecan-l ferentiation. J Invest DermatoI81:100S-103S, 1983 may retain bFGF at the cell surface as the keratinocytes stratify and 18. Bell CM, Skerrow CJ: Factors affecting the binding of lectins to nor­ move away from the reservoir ofbFGF in the basement membrane. mal human skin. Br J Dermatol 111 :5 17 - 526, 1984 Also, the recent finding that syndecan-1 increases in abundance on 19. Lamberg SI, Yuspa SH, Hascall VC: Synthesis of hyaluronic acid is proliferating keratinocytes during wound healing suggests that syn­ decreased and synthesis of proteoglycans is increased when cultured decan-l may rarticipate in growth factor-mediated repair of the mouse epidermal cells differentiate. J Invest Dermatol86:659 - 667, epidermis [59 . 1986 In conclusion, syndecan-l is a well characterized cell-matrix re­ 20. Rahemtulla F, Moorer CM, Wille JJ Jr: Biosynthesis of proteoglycans ceptor in simple epithelia that changes in structure and quantity by proliferating and differentiating normal human keratinocytes when keratinocytes stratify. These changes imply that syndecan-l cultured in serum-free medium. J Cell PhysioI140:98-106, 1989 has an alternative function in the epidermis, perhaps as a cell-cell 21. Bernfield M, Kokenyesi R, Kato M, Hinkes MT, SpringJ, Gallo RL, adhesion and/or growth factor binding molecule. Thus, syndecan- Lose EJ: Biology of the syndecans: a fami ly of transmembrane hepa­ 1 could play a role in altering the cellular interactions that are ran sulfate proteoglycans. Annu Rev Cell Bioi (in press) characteristic of epidermal differentiation and stratification, 22. Saunders S, Bernfield M: Cell surface proteoglycan binds mouse mam­ mary epithelial cells to fibronectin and behaves as a receptor for interstitial matrix. J Cell Bioi 106:423 - 430, 1988 This study was supported by a gratJl frol1l th e A;thritis FouMdatiolJ (RDS) a/Jd by 23. Koda JE, Rapraeger A, Bernfield M: Heparan sulphate proteoglycans National 1t1Stillltes oj Health gralJts AR01816 (MTH) al1d HD06763 (MB). from mouse mammary epithelial cell s: cell surface proteoglycan as a receptor for interstitial collagens. J Bioi Chem 260:8157 -8162, 1985 REFERENCES 24. Sun X, Mosher DF, Rapraeger A: Heparan sulfate-mediated binding of epithelial cell surface proteoglycan to thrombospondin. J BioI 1. Watt FM, Green H: Stratification and terminal differentiation of cul­ Chern 264:2885-2889,1989 tured epidermal cell s. Nature (London) 295:434-436,1982 25. Rapraeger A, Jalkanen M, Bernfield M: Cell surface proteoglycan 2. Stanley JR, FoidartJ-M, Murray JC, Martin GR, Katz SI: The epider­ associates with the cy toskel eton at the baso lateral cell surface of mal cell which selectively adheres to a collagen substrate is the basal mouse mammary epithelial cells. J Cell Bioi 103:2683 - 2696,1986 cell. J Invest Dermatol 74:54-58,1980 26. Saunders S, N guyen H, Bernfield M: Mammary epithelial cells trans- 396 SA NDERSON ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

fected with antisense eDNA red uce cell surface sy ndecan-l and cells produce basement membrane and cell surface heparan sulphate become fibrob lastic in morphology (abstr).] Cell Biol l09:5a, 1989 proteoglycans containing distinct co re proteins. J Cell Bioi 27. Sanderson RD, Bernfield M: Molecular po lymorphism of a cell surface 106:953 - 962, 1988 proteoglycan: distinct structures on simple and stratified epithelia. 44. W att FM, Green H: Involucrin synthesis is correlated with cell size in Proc Nat! Acad Sci USA 85:9562 -9566, 1988 human epidermal cu ltures. ] Cell Bioi 90:738-742,1981 28. Hayashi K, Hayashi M,]alkanen M, Firestone]H, Trelstad RL, Bern­ 45. Barrandon Y, Green H: Cell size as a determinant of the clone-forming field M: Immunocytochemistry of cell surface heparan sulphate pro­ ability of human keratinocytes. Proc Nat! Acad Sci USA 82:5390- teoglycan in mouse tissues. A li ght and electron microscopic study.] 5394,1985 Histochem Cytochem 35:1079-1088,1987 46. Sanderson RD, Sneed TB, Young LA, Sullivan GL, Lander AD: Adhe­ 29. Gonzalez A-M, Buscaglia M, Ong M, Baird A: Distribution of basic sion of P lymphoid (MPC-l1) cell s to type I collagen is mediated fibroblast growth factor in the 18-day rat fetus: loca lization in the by the integral membrane proteoglycan, sy ndecan. ] Immunol basement membranes of diverse tissues.] Cell Bioi 110:753 - 765, 148:3902- 3911,1992 1990 47. Salmivirta M, Elenius K, Vaini o S, Hofer U, Chiquet-Ehrismann R, 30. Yayon A, K1agsbru n M, Esko JD, Leder P, Ornitz DM: Cell surface, Thesleff I, ]alkanen M: Syndecan from embryonic tooth mesen­ heparin-like molecules are required for binding of basic fibroblast chyme binds tenascin.] Bioi Chern 266:7733-7739,1991 growth factor to its high affinity receptor. Cell 64:841-848, 1991 48. Ruoslahti E, Engvall E: Complexing of fibronectin, glycosaminogly­ 31. Rapraeger AC, Krufka A, Olwin BB: Requirement of heparan sulfate cans and collagen. Biochim Biophys Acta 631:350-358,1980 for bFGF-mediated fibroblast growth and myoblast differentiation. 49. Klebe R], Mock PJ: Effect of glycosaminoglycans on fibronectin-me­ Science 252: 1705 - 1708, 1991 diated cell attachment. ] Cell Physiol 112:5 - 9, 1982 32. Yuspa SH, Hawley-Nelson P, Stanley]R, Hennings H: Epidermal cell 50. David G, Van der Schueren B, Bernfield M: Basal lamina formation by culture. Transplan Proc 12(suppll):114- 122, 1980 normal and transformed mouse mammary epithelial cells duplicated 33 . ] alkanen M, Nguyen H, Rapraeger A, Kurn N, Bernfield M: Heparan in vitro. ] Nat! Cancer Inst 67:719 - 728, 1981 sulfate proteoglycans from mouse mammary epithelial cell s: locali­ 51. Saunders S, ]alkanen M, Bernfield M: Cell surface proteoglycan (PG) zation on the cell surface with a monoclonal antibody. J Cell Bioi is altered in structure and reduced in amount following malignant 101:976 -984, 1985 transformation of mouse mammary epithelial cells (abstr) .. ] Cell 34. Gallatin MW, Weissman IL, Butcher EC: A cell surface molecule Bioi 105:132a, 1987 involved in organ-specific homing of lymphocytes. Nature 52. Leppa S, Harki'men P, ] alkanen M: Steroid-induced epithelial-fibro­ 304:30- 34, 1983 blastic conversion associated with syndecan suppression in S115 35. Schor SL: Cell proliferation and migration on co llagen substrata in mouse mammary tumor cell s. Cell Regul 2:1-11, 1991 vitro.] Cell Sci 41:159-175,1980 53. Inki P, Stenback F, Talve L,]alkanen M: Immunohistochemicallocali­ 36. Rapraeger A,]alkanen M, Endo E, Koda], Berfield M: The cell surface zation of syndecan in mouse skin tumors induced by UV irradiation. proteoglycan from mouse mammary epithelial cell s bears chondroi­ Am] PathoI139:1333-1340. 1991 tin sulfate and heparan sulfate glycosaminoglycans. ] Bioi Chern 54. Pierce A, Lyon M, Hampson IN, Cowling G], Gallagher ]T: Molecu­ 260: 11 046 - 11052, 1985 lar cloning of the major cel l surface hepara n sulfate proteoglycan 37. Towbin H, Staehelin T, Gordon]: Electrophoretic transfer of proteins from rat liver. ] Bioi Chern 267:3894-3900,1992 fro m polyacrylamide gels to nitrocellulose sheets: procedure and 55. Kato M, Bernfield M: Epithelial ce ll s made deficient in syndecan lose some applications. Proc Nat! Acad Sci USA 76:4350-4354,1979 cell surface expression of E-cadherin and PI integrins (abstr) .] Cell 38. Johnson DA, Gautsch ]W, Sportsman ]R, Elder ]H: Improved tech­ Bioi 111 :263a, 1990 nique utilizing nonfat dry milk for analysis of proteins and nucleic 56. Kiefer MC, Stephans ]C, Crawford K, Okino K, Barr PJ: Ligand-af­ acids transferred to nitrocellulose. Gene Ana l Tech 1:3 -8, 1984 fin.ity cloning and structure of a cell surface heparan sulfate proteo­ 39. Wasteson A: Properties of fractionated chondroitin sulphate from ox glycan that binds basic fibroblast growth factor. Proc Nat! Acad Sci nasa l septa. Biochem] 122:477-485, 1971 USA 87:6985 - 6989, 1990 40. Saunders S, ]alkanen M, O'Farrell S, Bernfield M: Molecular cloning 57. Bernfield M, Hooper KC: Possibl e regulation ofFGF activity by syn­ of syndecan, an integral membrane proteoglycan. ] Cell Bioi decan, an integral membrane heparan sulfate proteoglycan. Ann NY 108:1547 -1556, 1989 Acad Sci 638:182- 194, 1991 41. ]alkanen M, Rapraeger A, Saunders S, Bernfield M: Cell surface proteo­ 58. Rubin]S, Osada H, Finch PW, Taylor WG, RudikoffS, Aaronson SA:. glycan of mouse mammary epithelial cells is shed by cleavage of its Purification and characterization of a newly identified growth fac tor matrix-binding ectodomain fro m its membrane-associated domai n. specific for epithelial cells. Proc N at! Acad Sci USA 86:802-806, ] Cell Bioi 105:3087-3096, 1987 1989 42. Kapuscinski], Skoczylas B: Simple and rapid flu orimetric method for 59. Elenius K, Vainio S, Laato M, Sa lmivirta M, Thesleff I, ] alkanen M: DNA microassay. Anal Biochem 83:252-255, 1977 Induced expression of syndecan in hea ling wounds. ] Cell Bioi 43. ] alkanen M, Rapraeger A, Bernfield M: Mouse mammary epithelial 114:585 -595, 1991