E~~RI~E~TAL
CELL
FXXEARCH
202,190-198
(l$K%)
Regulation of Urokinase Plasminogen Activator Localization in Keratinocytes by Calcium Ion and E-Cadherin PAMWLA J. JENS~N*,~ AND MARGARET J. ~VHEE~OCK~ *Department
of Dermatology, and tL?epartment
Univwsity of Eobgy,
of Pennsylvania, University
Philadelphiu, Pennsylvania Toledo, Ohio 43606
19104-6142;
of Toledo,
Cell-cell adhesion in keratinocyte culture can be modulated by changing the calcium ion concentration of the medium. In low (30 PM) calcium, keratinocytes proliferate as a monolayer that does not exhibit organized intercellular junctions. Elevation of the calcium concentration to 1.0 mM rapidly induces the formation of two distinct types of intercellular junctions: the adherens junction and the desmosome [l-4]. Organization of these junctions precedes stratification of the cells into a multilayered structure resembling an epithelial sheet. The mechanism by which calcium ion induces such dramatic morphological and morphogenetic changes in
keratinocyte culture is poorly understood. However, there is evidence that E-cadherin, a calcium-dependent, homotypic adhesion molecule [5], may play a regulatory role in the organization of interceIlular junctions in the keratinocyte as well as in other cell types [6-+X]. Further experiments have indicated that E-cadherin is also required for normal stratification of keratinocytes [6]. The development of a stratified epithelium from a keratinocyte monolayer requires atremendous reorganization of cell-cell and cell-substratum adhesion. New adhesion structures clearly are formed, but adhesion must also be disrupted in a regulated manner as the keratinocytes vertically migrate to form several layers. One possible mechanism for disruption of adhesion is proteolysis of the molecules that directly or indirectly mediate adhesion. Urokinase plasminogen activator (uPA)’ is a proteinase that has been implicated in pericellular proteolysis and matrix degradation during cell migration (for reviews see Refs. [g-12]). In most instances, uPA is thought to function by conversion of plasminogen to plasmin, a proteinase with the ability to cleave many proteins; however, uPA can also cleave fibronectin directly [13] and the possibility of additional substrates cannot be ruled out. In fibroblasts, HT 1080 fibrosarcoma cells, and rhabdomyosarcoma cells, uPA is localized to the focal contacts, suggesting that this enzyme may function to release cells from the substratum during migration [1416]. Urokinase PA on a membrane receptor has also been detected by immunoelectron microscopy at some regions of close cell-cell contact in cultures of fibrosarcoma and choriocarcinoma cells [15, 171. To begin to consider the hypothesis that uPA is a regulator of adhesion in the keratinocyte, we have examined with immunocytochemistry the distribution of uPA in cultured human keratinocytes during calciuminduced organization of intercellular junctions.
’ To whom reprint requests should be addressed at University of Pennsylvania, Department of Dermatology, 242 Clinical Research Building, PhilacIelphia, PA 19104-6142. Fax: (215) 573-2033.
’ Abbreviations used: PI-pbospbolipase specific phospholipase C; UPA, urokinase
In keratinocyte culture, the cellular distribution of many adhesion markers and the organization of intercellular junctions are controlled by the calcium ion eoncentration of the medium. We show in the present study that urokinase plasminogen activator (uPA) loealization in the human keratinoeyte is similarly dependent upon calcium concentration. At 30 pM calcium, uPA is present throughout the cell, often with a perinuclear concentration. Upon calcium elevation to 1.0 m&Z, uPA is concentrated along the cell-cell borders, where it colocalizes (at the light microscope level) with E-cadherin. Blocking antibody to E-cadherin delays the calcium-induced redistribution of uPA, in a manner very similar to the previously observed delay in redistribution of several adhesion-related markers, including vinculin, desmoplakin, and /Xl integrin. These data suggest a link between the redistribution of uPA to the cell-cell borders and the calcium-induced organization of intercelhxlar junctions in the human keratinocyte. The presence of UPA along the intercellular borders suggests that this enzyme may be involved in regulation of epidermal adhesion through proteolysis. CQ 1992 Academic
Press,
ho.
INTRODUCTION
ao14-4827/92 $mm Copyright 0 1902 by Academic Press, Inc. All rights of reproduction in any form reserved.
190
C, phosphatidylinositolplasminogen activator.
KERATINO~YTE
MATERIALS
AND
UROKINAS~
METHODS
Cell culture. Human neonatal foreskin keratinocytes were propagated in MCDB 153 medium with 30 PM CaCle and the following additives: bovine pituitary extract, insulin, epidermal growth factor, hydorcortisone, and high amino acids, as previously described [18201. This is referred to as “complete medium.” Cultures were used between the first and third passage. MCDB 153 medium was purchased from Irvine (Santa Ana, CA), Sigma (St. Louis, MO), and Clonetics (KGM with 30 p&f calcium). Epidermal growth factor was from Collaborative Research @edford, MA); insulin and hydrocortisone were from Sigma Chemical Co. Bovine pituitaries were from Pelfreez Biological (Rogers, AK). Antibodieslimmunocytochxmistry. The following primary antihuman UPA antibodies were used: rabbit anti-uPA IgG that recognizes 55- and 33-kDa enzymes (21); mouse monoclonal anti-uPA IgGl (clone 12 in ref. [22]), obtained from Monozyme (Virum, Denmark), that recognizes 55- but not 33-kDa enzymes; mouse monoclonal anti-uPA IgGl (No. 394 from American Diagnostica, Inc., Greenwich, CT) that recognizes 55 and 33-kDa enzymes; and a mouse monoclonal antibody (232) prepared as part of this study using as antigen partially purified urinary uPA (WinKinase, SterlingWinthrop, Rensselaer, NY, through the National Heart, Lung, and Blood Institute). The ZB2 antibody was shown to react with 55kDa UPA on immunoblots and to immunoprecipitate 55-kDa UPA but not 33-kDa UPA. It was purified from ascites using an ImmunoPure IgG purification kit (Pierce, Rockford, IL). The following irrelevant control antibodies were used: normal rabbit IgG and two irrelevant mouse monoclonal IgGl antibodies prepared against hepatitis and chlamydia antigens (kindly donated by Dr. Francee Boches, Baxter, Miami, FL). Rat monocional IgG and rabbit polyclonal antiserum against E-cadherin have been described [23]. Murine monoclonal anti-human transferrin receptor IgGl was purchased from Cbemicon International, Inc., (Temecula, CA). For imm~nostaining, keratinocytes were grown in complete medium in 12well plates until nearly or just confluent. In the indicated wells, cultures were fed with complete medium containing 1.0 rnM CaClx prior to staining; control wells were fed at the same time with normal complete medium containing 30 pM CaC&. Cultures were fixed with 1.0% paraformaldehyde for 10 min at room temperature, and then they were incubated for 60 min at room temperature with 0.1% saponin in PBS containing 10% goat serum (when the secondary antibody was made in goat), horse serum (when the secondary antibody was made in horse), or rabbit serum (when the secondary antibody was made in rabbit). Primary antibodies (at 4 fig/ml for rabbit IgG and 10 pg/ml for mouse IgG) were added in PBS with 10% serum and 0.1% saponin, and incubation was continued for 1 h. After extensive washing in PBS with 0.05% saponin, wells were incubated with secondary biotinyiated antibody (anti-rabbit IgG made in goat, anti-mouse IgG made in horse, or anti-rat IgG made in rabbit, all purchased from Vector Laboratories, Burlingame, CA). Finally, avidin-biotin-alkaline phosphatase and Substrate Kit 3 were applied according to specifications of the manufacturer (Vector Laboratories). For immunofluorescenee, similar procedures were used except that keratinocytes were plated on glass coverslips. In the double-labeling experiments, cells were incubated with a mixture of mouse monoclonai (clone 12) anti-uPA and rabbit anti-E-cadberin primary antibodies. The secondary reagent was a mixture of fluorescein-labeled antimouse IgG and rhodamine-labeled anti-rabbit IgG, both purchased from Organon Technica-Capped {Westchester, PA) and used at 1:25 dilution in PBS containing 10% goat serum. The coverslips were observed with a Zeiss Axiophot microscope equipped with epifluorescence. Blocking with anti-E-cadkerin antibodies. Purified IgG was prepared by protein A affinity chromatography of a previously described blocking rabbit antiserum against the external fragment of human
191
LOCALIZATION
E-cadberin [23]. Normal rabbit IgG was purchased (Organon Technica-Cappel). Anti-E-cadherin IgG and normal rabbit IgG were diluted to 20 pg/ml in complete MCDB 153 with 30 FM calcium. After an overnight incubation with antibodies, the calcium was raised to 1.0 m&f and incubation was continued for 4-24 h. ExtractionlELISA. After aspiration of the conditioned media, keratinocyte cultures in 12-well plates were scraped and extracted for 1 h at 4’C in 2 h!f KSCN plus 0.1% Triton X-100 (0.5 ml/2 wells); the lysates were then dialyzed overnight against 0.1 &f Tris-HCl plus 1 mM EDTA, pH 8.1 1241. ELISA for uPA was performed as described previously 1201 with a few modifications. Flat 96-well microtitration plates (Dynatech Laboratories, Chantilly, VA) were coated overnight with 2 @g/well of goat anti-human uPA (No. 389, American Diagnostica) in 0.05 M sodium carbonate, pH 9.6. Between all incubations, washing was carried out five times in PBS with 0.01% Tween-80 and 1 mg/ml BSA. Standards or samples (100 ~1) that had been diluted in PBS with 0.1% Tween 80, 5 mA4 EDTA, and 1 mg/ml BSA were added and incubation was continued for 6 h. Rabbit anti-uPA IgG (0.4 ag/welh Ref. [21]) was added for overnight incubation. Peroxidase-labeled goat anti-rabbit IgG (0.025 pg/welh Kirkegaard and Perry Laboratories, Gaithersburg, MD) was then added for 1 h and finally 2,2’also-~(3-ethylbenzthiazo~e sulfonate) peroxidase substrate, diluted as per directions of the manufacturer (Kirkegaard and Perry), was added. Color development was followed with a Titertek Multi” skan (Flow Laboratories, McLean, VA). The reaction was stopped with 1% SDS (50 &‘well). Single-chain UPA (a generous gift of Dr” D. Moir, Collaborative Research, Bedford, MA) was used as standard the assay was sensitive from 0.05 to 5.0 rig/ml. Acid and phospholipase C treatments. For acid treatment, cultures were incubated for 5 min at room temperature with 400 pi/well of isotonic acidic buffer (50 m&f glycine-HCl, 0.1 M NaCl, pH 3.0), neutralized with 100 PI/well of 0.5 M Hepes, 0.1 &f NaCl, pH 7.5, and washed in MCDB base medium; this treatment has previously been shown to strip ligand from uPA receptors of keratinocytes and other cell types without toxicity [20, 251. For phospholipase C treatment [26], cultures were incubated in phosphatidylinositol-specific phospholipase C (PI-phospholipase C) (0.36 U/mh No. 1143-069 from Boehringer-Mannheim Biochemica, Indianapolis, IN) for 10 min at 37’C and then washed in MCDB base medium prior to immunocytochemical staining.
KESXJLTS
Immunocytochemical localization studies of keratinocytes at 30 piW calcium revealed uPA distributed throughout the cell, often with a perinuclear concentration; the intensity of staining varied greatly from cell to cell (Fig. 1A). Upon elevation of the calcium concentration to 1.0 mM, the keratinocyte cultures showed a dramatic redistribution of uPA to the cell-cell borders. Within 1 h after calcium elevation of nearly confluent cultures, a few areas showed uPA along the cell-cell borders (Fig. lB, arrow). By 3 h nearly all fields showed cell-cell border stain for uPA (Fig. 1C); however, the regions of the culture in which cell-cell contact was lacking did not show preferential uPA localization at the cell periphery (Fig. lC, arrowhead). The intensity of uPA staining at the cell-cell borders increased after 7 and 24 h of incubation in high-calcium medium. Vari-
192
JENSEN
AND
WI-IEELOCK
FIG. 1. Localization of &‘A with time after calcium elevation. In one well of a confluent keratinocyte cuhure the calcium was maintained at 30 PM (A); in other wells, the calcium concentration was elevated to 1.0 mMfor 1 h (B), 3 h (C), 7 h (D and F): or 23 h (E). All wells were then processed for immunocytochemical staining with monoclonal (clone 12) anti-uPA antibody (A-E) or with irrelevant control anti-hepatitis monoclonal antibody (F). Staining was visualized with an alkaline phosphatase/blue substrate detection method. Staining for uI’A at the cell-cell borders was not detectable in cultures maintained in low calcium (A); but within 1 h after elevating the calcium, ceil-cell border stain was observed in limited areas (B, arrow). After 3 h in high calcium, cell-cell border stain for UPA was observed in most fields, but not in the hmited areas in which cell-cell contact had not as yet been made (C, arrowhead). Ceil-cell border stain for uPA became more widespread and increased in intensity with time after calcium elevation (C-E). Bar = 20 Fm.
KERATINOCYTE
UROKINASE
LOCALIZATION
1%
FiG. 4. Effect of low pH and PI-phospholipase C treatment on uPA localization. The calcium concentmtion was elevated. to 1.0 m&” ia each well of a nearly confiuent keratinocyte culture and incubation was continued for 2 h. ID some wells (B> E, and G) the celis were then treated with isotonic buffer at pH 3.0, washed, and treated with PI-phospholipase C. All we& were processed for ~mmu~ocyto~~emistry using the following primary antibodies: (A, B, F, and G): rabbit anti-uPA IgG; (Cl: control normal rabbit IgG; (D and El: rat anti-E-cadberin IgG. Bar = 4O~rn (A-E) orI0 pm (F a&G).
ability of staining intensity from field to field was consistently noted (l?ig. XEJ. Identical results were obtained with several antibodies that recognized different determinants of the UPA molecule. Specifically, the same pattern of staining was
observed with the following antibo ies: a polyclonal rabbit antibody &at re~og~~~~ 55. ad 33-kDa uPA [21]; a ~~~Q~l~~a~ antibody (2 2) that recognized ESkDa, but not 33 kDa UPA; a mono&ml antibody (clone 12; Ref. [22]) that recognized the a~~~o-t~~~i~~~ fmg-
194
JENSEN
TABLE Urokinase
PA
Levels in Keratinocytes and High Calcium
at Low
3 6 22
Calcium concentration 30 pM 1.0 mM 30 /.LM 1.0 mM 30 /.LM 1.0 mM
plasminogen
activator
Cell-associated (rig/well) 7.2 8.9 12.0 12.3 15.3 14.1
* * 2 ? zk 2
1.4 0.9 0.6 1.2 0.3 2.7
Secreted (rig/well) 4.6 4.3 9.7 8.2 61.8 57,9
Note, Cultures were grown to near confluence in complete with 30 pM calcium. At Time 0, the medium was aspirated were fed with complete MCDB containing 30 pM or 1.0 mM After 3, 6, or 22 h of further incubation, conditioned media lysates were harvested from replicate wells. Urokinase PA was measured by ELISA. Each value is the average ?Y range cate samples.
?I zk 2 2 k 2
WHEELOCK
Control experiments were performed with antibody to the transferrin receptor. The distribution of this receptor was consistently diffuse and was not dependent upon calcium concentration (Figs. 2F and 2G). Even at regions of close cell-cell contact observed after overnight incubation in high-calcium medium (Fig. 2G), there was no concentration of transferrin receptor.
1
Urokinase Time U4
AND
0.1 0.2 0.2 0.1 7.2 2.1
MCDB and cells calcium. and cell content of dupli-
ment of uPA; and a monoclonal antibody that detected a determinant common to 33- and 55kDa uPA. Despite the dramatic change in the localization of uPA when calcium was elevated, calcium did not alter the total cell-associated or secreted uPA in the cultures within the time frame of these experiments (Table 1). The redistribution of uPA observed upon elevation of the calcium concentration was very similar to the calcium-induced redistribution patterns of several molecules involved in cell-cell adhesion. Vinculin and a-actinin (markers of adherens junctions), desmoplakin (a marker of desmosomes), /?l integrin, P-cadherin, and E-cadherin all were shown to accumulate along the cellcell borders upon elevation of the calcium concentration in keratinocyte culture [l, 4, 6, 27-291. We have previously shown that the redistribution of E-cadherin was extremely rapid within U-20 min after calcium elevation E-cadherin was detected at some cellcell borders, and all fields showed cell-cell border staining by 1 h [6]. As documented in Fig. 1, the redistribution of uPA was noticeably slower, taking several hours for nearly all fields to show intercellular localization. These findings indicate that uPA redistributed to the cell-cell borders in concert with the organization of intercellular junctions, but with a time course consistently slower than the redistribution of E-cadherin. Colocalization experiments (Figs. 2B, 2C and 2D, 2E) demonstrated that uPA and E-cadherin had identical distributions at the light microscope level along the cell-cell borders in keratinocyte cultures at 1.0 rnM calcium. E-Cadherin was found almost exclusively at cellcell borders, while uPA was found not only at cell-cell borders but also in the perinuclear region of at least some cells.
Effect of Blocking E-Cadherin Function upon uPA Redistribution The calcium-induced redistribution of uPA to the cell-cell borders and its colocalization with E-cadherin suggested a link between UPA localization and the formation of intercellular junctions between keratinocytes. A regulatory role for E-cadherin in the organization of keratinocyte intercellular junctions has been proposed from studies that showed a delay in the redistribution of P-cadherin, /31 integrin, vinculin, and desmoplakin to the cell-cell borders upon calcium elevation in the presence of blocking antibodies to E-cadherin [6]. The redistribution of these markers was delayed but not completely prevented; redistribution to the cell-cell borders of each marker did occur after longer (i.e., overnight) incubation in high calcium in the continued presence of anti-E-cadherin antibody. To determine if blocking antibodies to E-cadherin had a similar inhibitory effect on uPA redistribution, cultures were preincubated with rabbit anti-E-cadherin IgG [23]; calcium was then elevated for various periods of time prior to immunofluorescent staining for uPA. Comparison with control cultures demonstrated that anti-E-cadherin IgG delayed the redistribution of uPA. After 4 h in high calcium, control cultures revealed cellcell border stain for uPA in most fields (Fig. 3C), as described above; however, cultures preincubated with antibody to E-cadherin showed almost no cell-cell border stain for uPA at this time (Fig. 3D). When cultures were incubated for 24 h in high calcium, uPA redistributed to the cell-cell borders even in the continued presence of anti-E-cadherin IgG (Fig. 3F), and the staining pattern resembled that of the controls. These results on uPA localization in the presence of anti-E-cadherin antibody are very similar to those previously observed for several adhesion-related markers [6]. The explanation for the eventual recovery of redistribution ability in the continued presence of anti-E-cadherin antibody is unknown, but we have shown that it was not the result of recovery of E-cadherin function [6]. Receptor-Bound uPA Many cell types, including the keratinocyte [20, 301, bear a specific plasma membrane receptor for uPA. Recent studies have demonstrated that the uPA receptor is linked to the cell surface by a glycosyl- phosphatidylino-
‘FIG+ 2. CoIocaiization of E-cadberin and UPA in keratinocytes. (A-E): Nearly confluent keratinocytes on glass covers&x were incabate& in compkte medium containing 1.Q m&f calcium for 12 h+ The celIs were tbea processed for ~mrn~~o~~ore~e~ce with a mixture of murine anti-GA and rabbit ~ti-E-cadherin antibodies. (I3 and D): The loc~izat~o~ of~-ca~e~~. (C and E): The corr~s~oad~~~ IocaGzatio~ of uP-A in the same respective fields. (A): The phase contrast view of I3 and C. (l?-H): ~erat~noc”~es i~~bated in 30 fl calcium (S’} or in 1.0 mM calcium (G and HI for &e final 12 h of cuIture were processed for immu~o~~oresce~ce with antibody to transferFin receptor. (HI: The p&se contrast view of G. TransferriB receptor was never concentrated at the cell-ceII borders, even along areas of cIose ceII-cell contact at high calcium (G and l-X> arrowheads). Bar = 15 pm (A-C) or 6 ,um (D-H).
196
JENSEN
AND
WHEELOCK
FIG. 3. Antibodies against E-cadherin delayed the calcium-induced redistribution of uPA, Nearly confluent keratinocyte cultures in complete medium with 30 WM calcium were preincubated with normal rabbit IgG (A, C, and E) or rabbit anti-E-cadherin IgG (B, D, and F). Without changing the medium, the calcium concentration in some wells was then increased to 1.0 mM for varying lengths of time prior to immunofluorescent staining with monoclonal anti-uPA IgG (2B2). (A and B): Cells at 30 &f calcium. (C and D): Cells after 4 h in 1.0 mM calcium. (E and F): Cells after 24 h in 1.0 mM calcium. Bar = 20 pm.
with PI-phospholipase C; the cultures were then stained for uPA. Nearly all of the uPA staining at the cell-cell borders was lost in cultures so treated (compare Fig. 4A vs 4B and Figs. 4F vs 4G). This result strongly suggests that uPA at the cell-cell borders is on the cell surface and receptor-bound. In several experiments, cultures were exposed either to isotonic low-pH buffer alone or to PI-phospholipase C alone prior to uPA staining. The diminution of uPA staining intensity following either treatment on its own was extremely variable from experiment to experiment
and from field to field. In contrast, in each of three experiments in which cells were exposed both to low pH and PI-phospholipase C, nearly all cell-cell border stain for uPA was lost upon analysis by either immunofluorescence or immunoalkaline phosphatase detection methods. As a control for possible nonspecific effects of the low pH and PI-phospholipase, staining for E-cadherin was also performed on treated cells. No consistent difference was observed in the pattern of staining for E-cadherin in treated compared to control cultures, although
KERATINOCYTE
LJROKINASE
in some fields of the treated cultures the staining was a little less intense or less distinct at the cell-cell borders (Figs. 4D and 4E)* DISCUSSION
Rapid translocation of adhesion molecules to the cell-cell borders as well as structural organization of both adherens junctions and desmosomes are well documented in keratinocyte culture after calcium elevation [l-4,6]. In the present study, we have shown that localization of the proteolytic enzyme uPA in the keratinocyte is also controlled by calcium concentration. Urokinase PA distribution in the keratinocyte has several features in common with the distribution of a number of adhesion molecules including E-cadherin, P-cadherin, /Xl integrin, vinculin, and desmoplakin: (i) Each of these molecules is redistributed to the cell-cell borders upon elevation of the calcium concentration to 1.0 mM. (ii) The time course for redistribution of all molecules is rather rapid, although E-cadherin redistribution consistently precedes tbat of uPA. (iii) Inhibitory antibodies to E-cadherin delay by several hours the calcium-induced redistribution of each molecule. (iv) With time, redistribution of botb uPA and the adhesion-related molecules eventually occurs even in the continued presence of anti-E-cadherin antibodies [6]. These findings suggest that uPA localization is coordinated with intercellular junction formation in the keratinocyte. Our previous data as well as those of others indicate that E-cadherin plays a regulatory role in the early stages of intercellular junction organization [6-81. Comparison of the time course for uPA redistribution (Pig- 1) and E-cadherin redistribution [6] reveals that the latter is more rapid, consistent with its regulatory role. Cadherins are thought to be linked to the cytoskeleton through a group of proteins termed catenins; studies with anti-cadherin antibodies have shown that the catenins coimmunoprecipitate with cadherins [31-371. In the keratinocyte, we have previously shown that the profile of proteins that coimmunoprecipitate with Ecadherin is identical in cells maintained in 30 pM calcium or switched to 1 mM calcium for 12 h [36], in agreement with observations from other cell types [36]. Urokinase PA is not found among the proteins that coimmunoprecipitate (Wheelock? unpublished observations)Y indicating that uPA is not tightly associated with E-cadherin Urokinase PA is synthesized and secreted in many cell types [38], including the keratinocyte [39], as a 55a single-chain precursor molecule. Little if any conversion of the single-chain precursor to the two-chain enzyme is found in keratinocyte culture [39]. Like many other cell types [llS 38], the human keratinocyte [20,30] bears a specific, ceI1 surface uPA receptor that recently
LOCALIZATION
197
has been shown to be linked to the plasma membrane by a glycosyl-phosphatidylinositol lipid an&or [26]. There is evidence that at least some proteins so anchored have enhanced mobility in the plasma membrane (for review see Ref. [4O]). Under our culture con tions? the keratinocyte uPA receptor is occupied with endogenous uPA [20]. In the present study, uPA stain at the cell-cell borders is lost in cultures treated face-acting reagents (isotonic 1 phospholipase). These agents ba tively to dissociate ligand from receptor and to release glycosyl-pbosphatidylinositol-linked proteins, including the uPA receptor [20, 25, 26]. These data strongly suggest that the uPA concentrated along the keratinocyte-keratinocyte borders in response to calcium elevation is on the outside of the ceil a is r~~e~to~-bound. Previous studies have &own A is not rapidly internalized receptor9 but rather appears to be concentrated at the cell surface [41], awaiting activation for pericellular proteolysis when appropriate. The mechanism of receptor-bound uP-4 activation is likely to involve cell-bound plasmin [42]. In summary, we suggest that receptor-bound uPA behaves similarly to several adhesion-related molecules in that its cellular distribution is controlled by calcium concentration and can be modulated by E-cadherin. When concentrated along the ceil-cell borders of’ keratinocytes that exhibit intercellular junctions uPA is poised to participate in the cornroIled disruption of cell-cell adhesion structures that likely accompanies epidermal remodeling or morphogenesis. This work was supported by Grants CA 44464 and AR-39674 sub 06 from the National Institutes of Health to M.J.W. and I’.J.J.? respectively and by Grant DIR-8804006 from the National Science Foundation to M.J.W. The authors thank Dr. Norman Schechter for critical reading of the manuscript. Tile technical assistance of Mr. Kelby Frame in propagation of keratinocyte cultures is acknowledged with thanks.
1.
O’Keefe, E. J., Briggaman, 3iol. 105, 807.
2.
Hennings, H., Michael, D., Cheng, C., Steinert, I?~, Holbrook, and Yuspa, S. H. (1980) Cell 1 Hennings, H., and Holbrook, K. (1983) Exn. Ceil Res. 143, Watt, F. M., Mattey, D. L., and Garrod, D. I% (1984) J. Cell 99, 2211. Takeichi, M. (1988) &velopmerzt 102,639. Wheelock, M. J., and Jensen, I’. J. (1992) J. Cell BioZ. 117, Gumbiner, B., Stevenson, I%, and Grimaldi? A. (1988) J. 3iol. 107, 1575. Jongen, W. M. F., Fitzgerald, D. J., Asamoto? M., I’iccoli,
3. 4. 5. 6. I. 8.
R. A., and Herman,
B. (1587)
J. CelZ K., 127. I?ioi.
415. Cell C.>
198
JENSEN Slaga, T. J., Gras, D., Takeichi, CellBiol. 114, 545. Dane, K., Behrendt, Molecular Biology New York.
M., and Yamasaki,
N., Ronne E., Ellis, of the Cardiovascular
H. (1991)
Vaheri, A., Stephens, Tapiovaara, H. (1990)
11.
Vassalli, J.-D., uest. &3,1067.
12
Pollanen, Res. 57,
13.
Gold, L. L, Schwimmer, R., and Quigley, J. P. (1989) 262,529. Hebert, C. A., and Baker, J. B. (1988) J. Cell BioZ.
14.
R. W., Salonen, E.-M., PoIlanen, Cell Diff. Deu. 32, 255. A.-P.,
J., Stephens, 273.
J., Hedman, J. CelL Biot.
and Belin,
R. W., and Vaheri,
K., Nielsen, 106,8’7.
J.
V., and Blasi, F. (1990) System, p. 173, Liss,
10.
Sappino,
AND
D. (1391) A. (1991)
b. S., Dane,
J., and
J. %a.
In-
Adu. Cancer Biochem.
106,
J.
Pollanen, A. (1988)
16.
Pollanen, J., Vaheri, A., Tapiovaara, H,, Riley, E., Bertram, Woodrow, G., and Stephens, R. W. (1990) Proc. Natl. Acad. USA S7,2230.
K., and Vaheri, K., Sci
17.
Jensen, P.
18. 19.
Boyce, S. T., and Ham, R. G. (1983) J 1rxxsr. DerrrzutoZ. 81,33s. Shipley, G. D., and Pittelkow, M. R. (1987) Arch. Dermatol. 123, 154la.
20.
McNeill,
21.
Morioka, S., Jensen, Res. 161, 364.
22.
Grendahl-Hansen, J., Ralfkiaer, E., Nielsen, P., Frentz, G., and Dane: K. (1987) J. Zrtvest.
23.
Wbeelock, M. J., Buck, C!. A., Bechtol, K. B., and Damsky, C. H. (1987) J. Cell. Biachem. 34, 187. Jensen, P, J., John, M., and Baird, J. (1990) Exp. Cell Res. 187, 162.
Andreasen,
24.
Received Revised
March version
H., and Jensen,
P. J. (1990)
P. J., and Lazarus,
30, 1992 received June
Cell Reg.
J., and
1, 843.
G. S. (1985)
25.
26. 27.
28. 29. 30.
1241.
15.
H., Christensen, E. I., Ebbesen, P., Gliemann, P. A. (1990) Ceu Eeg. 1,1043.
W~ELOCK
31. 32. 33. 34. 35. 36. 37. 38.
Exp.
Cell 39.
1, 1992
L. S,, Kristensen, Dermatol. 68,28.
40. 41.
42.
Stoppelli, M. P., Tacchetti, C., Cubellis, M. V., Corti, A., Hearing, V. J., Cassani, G., Appeila, E., and Blasi, F. (1986) Cell 46, 675. Ploug, M., Renne, E., Behrendt, N., Jensen, A. L., Blasi, F*, and Dane, K. (1991) J. Bio!. Chem. 266,1926. Carter, W. G., Wayner, E. A., Bouchard, T. S., and Kawr, P. (1990) J. Cell Bio!. 110,1387. Larjava, H., Pehonen, J., Akiyama, S. K., Yamada, S. S., Gralnick, H. R., Uitto, J., and Yamada, K. M. (1990) J. Ceil Viol. 110, 803. Marchisio, P. C., Bondanza, S., Cremona, O., Caneedda, R., and Re Luca, M. (1991) J Cell BioL 112, 761. Del Rosso, M., Fibbi, G., Ihi, G., Grappone, C., Pucci, M., Cal” dini, R., Magnelli, L., Fiminai, M., Lotti, T., and Panconesi, E. (1990) J. Znvest. Dermatol. 94, 310. Kemler, R., and Ozawa, M. (1989) Bioessays 11,88. McCrea, P., Turck, C. W., and Gumbiner, B. (1991) Science 254,1359. Nagafuchi, A., Takeichi, M., and Tsukita, S. (1991) Cell 65,849. Peyrieras, N., Louvard, Aead. Sci. USA 62,8067.
D.,
and
Jacob,
F. (1985)
Prac.
?.ruti.
Wheelock, M. J. (1990) Z&p. Cell Res. 191,186. Wheelock, M. J., and Knudsen, K. A. (1991) 1n Viuo 5,505. Wheelock, M. J., and Knudsen, K. A. (1991) Differentiatzbn 46, 35. Blasi, F., Vassalli, J.-D., and Dane, K. (1987) J. Cell Biol. 104, 801. Hashimoto, K., Prystowsky, J. H., Baird, J., Lazarus, G. S., and Jensen, P. J. (1988) J. Znvest. Dermatol. 90,823. Low, M. G. (1989) Biochim. Z3iophys. Acta 988,427. Stoppehi, M. P., Corti, A., Soffientini, A., &assani, G., Blasi, F., and Assoian, R. K. (1985) Proc. Natl. Acad. Sei. USA 62,4939. Ellis, V., Scully, M. F., and Kakkar, V. V. (1989) J. Biol. Chem. 264,2185.
CELL
FXXEARCH
202,190-198
(l$K%)
Regulation of Urokinase Plasminogen Activator Localization in Keratinocytes by Calcium Ion and E-Cadherin PAMWLA J. JENS~N*,~ AND MARGARET J. ~VHEE~OCK~ *Department
of Dermatology, and tL?epartment
Univwsity of Eobgy,
of Pennsylvania, University
Philadelphiu, Pennsylvania Toledo, Ohio 43606
19104-6142;
of Toledo,
Cell-cell adhesion in keratinocyte culture can be modulated by changing the calcium ion concentration of the medium. In low (30 PM) calcium, keratinocytes proliferate as a monolayer that does not exhibit organized intercellular junctions. Elevation of the calcium concentration to 1.0 mM rapidly induces the formation of two distinct types of intercellular junctions: the adherens junction and the desmosome [l-4]. Organization of these junctions precedes stratification of the cells into a multilayered structure resembling an epithelial sheet. The mechanism by which calcium ion induces such dramatic morphological and morphogenetic changes in
keratinocyte culture is poorly understood. However, there is evidence that E-cadherin, a calcium-dependent, homotypic adhesion molecule [5], may play a regulatory role in the organization of interceIlular junctions in the keratinocyte as well as in other cell types [6-+X]. Further experiments have indicated that E-cadherin is also required for normal stratification of keratinocytes [6]. The development of a stratified epithelium from a keratinocyte monolayer requires atremendous reorganization of cell-cell and cell-substratum adhesion. New adhesion structures clearly are formed, but adhesion must also be disrupted in a regulated manner as the keratinocytes vertically migrate to form several layers. One possible mechanism for disruption of adhesion is proteolysis of the molecules that directly or indirectly mediate adhesion. Urokinase plasminogen activator (uPA)’ is a proteinase that has been implicated in pericellular proteolysis and matrix degradation during cell migration (for reviews see Refs. [g-12]). In most instances, uPA is thought to function by conversion of plasminogen to plasmin, a proteinase with the ability to cleave many proteins; however, uPA can also cleave fibronectin directly [13] and the possibility of additional substrates cannot be ruled out. In fibroblasts, HT 1080 fibrosarcoma cells, and rhabdomyosarcoma cells, uPA is localized to the focal contacts, suggesting that this enzyme may function to release cells from the substratum during migration [1416]. Urokinase PA on a membrane receptor has also been detected by immunoelectron microscopy at some regions of close cell-cell contact in cultures of fibrosarcoma and choriocarcinoma cells [15, 171. To begin to consider the hypothesis that uPA is a regulator of adhesion in the keratinocyte, we have examined with immunocytochemistry the distribution of uPA in cultured human keratinocytes during calciuminduced organization of intercellular junctions.
’ To whom reprint requests should be addressed at University of Pennsylvania, Department of Dermatology, 242 Clinical Research Building, PhilacIelphia, PA 19104-6142. Fax: (215) 573-2033.
’ Abbreviations used: PI-pbospbolipase specific phospholipase C; UPA, urokinase
In keratinocyte culture, the cellular distribution of many adhesion markers and the organization of intercellular junctions are controlled by the calcium ion eoncentration of the medium. We show in the present study that urokinase plasminogen activator (uPA) loealization in the human keratinoeyte is similarly dependent upon calcium concentration. At 30 pM calcium, uPA is present throughout the cell, often with a perinuclear concentration. Upon calcium elevation to 1.0 m&Z, uPA is concentrated along the cell-cell borders, where it colocalizes (at the light microscope level) with E-cadherin. Blocking antibody to E-cadherin delays the calcium-induced redistribution of uPA, in a manner very similar to the previously observed delay in redistribution of several adhesion-related markers, including vinculin, desmoplakin, and /Xl integrin. These data suggest a link between the redistribution of uPA to the cell-cell borders and the calcium-induced organization of intercelhxlar junctions in the human keratinocyte. The presence of UPA along the intercellular borders suggests that this enzyme may be involved in regulation of epidermal adhesion through proteolysis. CQ 1992 Academic
Press,
ho.
INTRODUCTION
ao14-4827/92 $mm Copyright 0 1902 by Academic Press, Inc. All rights of reproduction in any form reserved.
190
C, phosphatidylinositolplasminogen activator.
KERATINO~YTE
MATERIALS
AND
UROKINAS~
METHODS
Cell culture. Human neonatal foreskin keratinocytes were propagated in MCDB 153 medium with 30 PM CaCle and the following additives: bovine pituitary extract, insulin, epidermal growth factor, hydorcortisone, and high amino acids, as previously described [18201. This is referred to as “complete medium.” Cultures were used between the first and third passage. MCDB 153 medium was purchased from Irvine (Santa Ana, CA), Sigma (St. Louis, MO), and Clonetics (KGM with 30 p&f calcium). Epidermal growth factor was from Collaborative Research @edford, MA); insulin and hydrocortisone were from Sigma Chemical Co. Bovine pituitaries were from Pelfreez Biological (Rogers, AK). Antibodieslimmunocytochxmistry. The following primary antihuman UPA antibodies were used: rabbit anti-uPA IgG that recognizes 55- and 33-kDa enzymes (21); mouse monoclonal anti-uPA IgGl (clone 12 in ref. [22]), obtained from Monozyme (Virum, Denmark), that recognizes 55- but not 33-kDa enzymes; mouse monoclonal anti-uPA IgGl (No. 394 from American Diagnostica, Inc., Greenwich, CT) that recognizes 55 and 33-kDa enzymes; and a mouse monoclonal antibody (232) prepared as part of this study using as antigen partially purified urinary uPA (WinKinase, SterlingWinthrop, Rensselaer, NY, through the National Heart, Lung, and Blood Institute). The ZB2 antibody was shown to react with 55kDa UPA on immunoblots and to immunoprecipitate 55-kDa UPA but not 33-kDa UPA. It was purified from ascites using an ImmunoPure IgG purification kit (Pierce, Rockford, IL). The following irrelevant control antibodies were used: normal rabbit IgG and two irrelevant mouse monoclonal IgGl antibodies prepared against hepatitis and chlamydia antigens (kindly donated by Dr. Francee Boches, Baxter, Miami, FL). Rat monocional IgG and rabbit polyclonal antiserum against E-cadherin have been described [23]. Murine monoclonal anti-human transferrin receptor IgGl was purchased from Cbemicon International, Inc., (Temecula, CA). For imm~nostaining, keratinocytes were grown in complete medium in 12well plates until nearly or just confluent. In the indicated wells, cultures were fed with complete medium containing 1.0 rnM CaClx prior to staining; control wells were fed at the same time with normal complete medium containing 30 pM CaC&. Cultures were fixed with 1.0% paraformaldehyde for 10 min at room temperature, and then they were incubated for 60 min at room temperature with 0.1% saponin in PBS containing 10% goat serum (when the secondary antibody was made in goat), horse serum (when the secondary antibody was made in horse), or rabbit serum (when the secondary antibody was made in rabbit). Primary antibodies (at 4 fig/ml for rabbit IgG and 10 pg/ml for mouse IgG) were added in PBS with 10% serum and 0.1% saponin, and incubation was continued for 1 h. After extensive washing in PBS with 0.05% saponin, wells were incubated with secondary biotinyiated antibody (anti-rabbit IgG made in goat, anti-mouse IgG made in horse, or anti-rat IgG made in rabbit, all purchased from Vector Laboratories, Burlingame, CA). Finally, avidin-biotin-alkaline phosphatase and Substrate Kit 3 were applied according to specifications of the manufacturer (Vector Laboratories). For immunofluorescenee, similar procedures were used except that keratinocytes were plated on glass coverslips. In the double-labeling experiments, cells were incubated with a mixture of mouse monoclonai (clone 12) anti-uPA and rabbit anti-E-cadberin primary antibodies. The secondary reagent was a mixture of fluorescein-labeled antimouse IgG and rhodamine-labeled anti-rabbit IgG, both purchased from Organon Technica-Capped {Westchester, PA) and used at 1:25 dilution in PBS containing 10% goat serum. The coverslips were observed with a Zeiss Axiophot microscope equipped with epifluorescence. Blocking with anti-E-cadkerin antibodies. Purified IgG was prepared by protein A affinity chromatography of a previously described blocking rabbit antiserum against the external fragment of human
191
LOCALIZATION
E-cadberin [23]. Normal rabbit IgG was purchased (Organon Technica-Cappel). Anti-E-cadherin IgG and normal rabbit IgG were diluted to 20 pg/ml in complete MCDB 153 with 30 FM calcium. After an overnight incubation with antibodies, the calcium was raised to 1.0 m&f and incubation was continued for 4-24 h. ExtractionlELISA. After aspiration of the conditioned media, keratinocyte cultures in 12-well plates were scraped and extracted for 1 h at 4’C in 2 h!f KSCN plus 0.1% Triton X-100 (0.5 ml/2 wells); the lysates were then dialyzed overnight against 0.1 &f Tris-HCl plus 1 mM EDTA, pH 8.1 1241. ELISA for uPA was performed as described previously 1201 with a few modifications. Flat 96-well microtitration plates (Dynatech Laboratories, Chantilly, VA) were coated overnight with 2 @g/well of goat anti-human uPA (No. 389, American Diagnostica) in 0.05 M sodium carbonate, pH 9.6. Between all incubations, washing was carried out five times in PBS with 0.01% Tween-80 and 1 mg/ml BSA. Standards or samples (100 ~1) that had been diluted in PBS with 0.1% Tween 80, 5 mA4 EDTA, and 1 mg/ml BSA were added and incubation was continued for 6 h. Rabbit anti-uPA IgG (0.4 ag/welh Ref. [21]) was added for overnight incubation. Peroxidase-labeled goat anti-rabbit IgG (0.025 pg/welh Kirkegaard and Perry Laboratories, Gaithersburg, MD) was then added for 1 h and finally 2,2’also-~(3-ethylbenzthiazo~e sulfonate) peroxidase substrate, diluted as per directions of the manufacturer (Kirkegaard and Perry), was added. Color development was followed with a Titertek Multi” skan (Flow Laboratories, McLean, VA). The reaction was stopped with 1% SDS (50 &‘well). Single-chain UPA (a generous gift of Dr” D. Moir, Collaborative Research, Bedford, MA) was used as standard the assay was sensitive from 0.05 to 5.0 rig/ml. Acid and phospholipase C treatments. For acid treatment, cultures were incubated for 5 min at room temperature with 400 pi/well of isotonic acidic buffer (50 m&f glycine-HCl, 0.1 M NaCl, pH 3.0), neutralized with 100 PI/well of 0.5 M Hepes, 0.1 &f NaCl, pH 7.5, and washed in MCDB base medium; this treatment has previously been shown to strip ligand from uPA receptors of keratinocytes and other cell types without toxicity [20, 251. For phospholipase C treatment [26], cultures were incubated in phosphatidylinositol-specific phospholipase C (PI-phospholipase C) (0.36 U/mh No. 1143-069 from Boehringer-Mannheim Biochemica, Indianapolis, IN) for 10 min at 37’C and then washed in MCDB base medium prior to immunocytochemical staining.
KESXJLTS
Immunocytochemical localization studies of keratinocytes at 30 piW calcium revealed uPA distributed throughout the cell, often with a perinuclear concentration; the intensity of staining varied greatly from cell to cell (Fig. 1A). Upon elevation of the calcium concentration to 1.0 mM, the keratinocyte cultures showed a dramatic redistribution of uPA to the cell-cell borders. Within 1 h after calcium elevation of nearly confluent cultures, a few areas showed uPA along the cell-cell borders (Fig. lB, arrow). By 3 h nearly all fields showed cell-cell border stain for uPA (Fig. 1C); however, the regions of the culture in which cell-cell contact was lacking did not show preferential uPA localization at the cell periphery (Fig. lC, arrowhead). The intensity of uPA staining at the cell-cell borders increased after 7 and 24 h of incubation in high-calcium medium. Vari-
192
JENSEN
AND
WI-IEELOCK
FIG. 1. Localization of &‘A with time after calcium elevation. In one well of a confluent keratinocyte cuhure the calcium was maintained at 30 PM (A); in other wells, the calcium concentration was elevated to 1.0 mMfor 1 h (B), 3 h (C), 7 h (D and F): or 23 h (E). All wells were then processed for immunocytochemical staining with monoclonal (clone 12) anti-uPA antibody (A-E) or with irrelevant control anti-hepatitis monoclonal antibody (F). Staining was visualized with an alkaline phosphatase/blue substrate detection method. Staining for uI’A at the cell-cell borders was not detectable in cultures maintained in low calcium (A); but within 1 h after elevating the calcium, ceil-cell border stain was observed in limited areas (B, arrow). After 3 h in high calcium, cell-cell border stain for UPA was observed in most fields, but not in the hmited areas in which cell-cell contact had not as yet been made (C, arrowhead). Ceil-cell border stain for uPA became more widespread and increased in intensity with time after calcium elevation (C-E). Bar = 20 Fm.
KERATINOCYTE
UROKINASE
LOCALIZATION
1%
FiG. 4. Effect of low pH and PI-phospholipase C treatment on uPA localization. The calcium concentmtion was elevated. to 1.0 m&” ia each well of a nearly confiuent keratinocyte culture and incubation was continued for 2 h. ID some wells (B> E, and G) the celis were then treated with isotonic buffer at pH 3.0, washed, and treated with PI-phospholipase C. All we& were processed for ~mmu~ocyto~~emistry using the following primary antibodies: (A, B, F, and G): rabbit anti-uPA IgG; (Cl: control normal rabbit IgG; (D and El: rat anti-E-cadberin IgG. Bar = 4O~rn (A-E) orI0 pm (F a&G).
ability of staining intensity from field to field was consistently noted (l?ig. XEJ. Identical results were obtained with several antibodies that recognized different determinants of the UPA molecule. Specifically, the same pattern of staining was
observed with the following antibo ies: a polyclonal rabbit antibody &at re~og~~~~ 55. ad 33-kDa uPA [21]; a ~~~Q~l~~a~ antibody (2 2) that recognized ESkDa, but not 33 kDa UPA; a mono&ml antibody (clone 12; Ref. [22]) that recognized the a~~~o-t~~~i~~~ fmg-
194
JENSEN
TABLE Urokinase
PA
Levels in Keratinocytes and High Calcium
at Low
3 6 22
Calcium concentration 30 pM 1.0 mM 30 /.LM 1.0 mM 30 /.LM 1.0 mM
plasminogen
activator
Cell-associated (rig/well) 7.2 8.9 12.0 12.3 15.3 14.1
* * 2 ? zk 2
1.4 0.9 0.6 1.2 0.3 2.7
Secreted (rig/well) 4.6 4.3 9.7 8.2 61.8 57,9
Note, Cultures were grown to near confluence in complete with 30 pM calcium. At Time 0, the medium was aspirated were fed with complete MCDB containing 30 pM or 1.0 mM After 3, 6, or 22 h of further incubation, conditioned media lysates were harvested from replicate wells. Urokinase PA was measured by ELISA. Each value is the average ?Y range cate samples.
?I zk 2 2 k 2
WHEELOCK
Control experiments were performed with antibody to the transferrin receptor. The distribution of this receptor was consistently diffuse and was not dependent upon calcium concentration (Figs. 2F and 2G). Even at regions of close cell-cell contact observed after overnight incubation in high-calcium medium (Fig. 2G), there was no concentration of transferrin receptor.
1
Urokinase Time U4
AND
0.1 0.2 0.2 0.1 7.2 2.1
MCDB and cells calcium. and cell content of dupli-
ment of uPA; and a monoclonal antibody that detected a determinant common to 33- and 55kDa uPA. Despite the dramatic change in the localization of uPA when calcium was elevated, calcium did not alter the total cell-associated or secreted uPA in the cultures within the time frame of these experiments (Table 1). The redistribution of uPA observed upon elevation of the calcium concentration was very similar to the calcium-induced redistribution patterns of several molecules involved in cell-cell adhesion. Vinculin and a-actinin (markers of adherens junctions), desmoplakin (a marker of desmosomes), /?l integrin, P-cadherin, and E-cadherin all were shown to accumulate along the cellcell borders upon elevation of the calcium concentration in keratinocyte culture [l, 4, 6, 27-291. We have previously shown that the redistribution of E-cadherin was extremely rapid within U-20 min after calcium elevation E-cadherin was detected at some cellcell borders, and all fields showed cell-cell border staining by 1 h [6]. As documented in Fig. 1, the redistribution of uPA was noticeably slower, taking several hours for nearly all fields to show intercellular localization. These findings indicate that uPA redistributed to the cell-cell borders in concert with the organization of intercellular junctions, but with a time course consistently slower than the redistribution of E-cadherin. Colocalization experiments (Figs. 2B, 2C and 2D, 2E) demonstrated that uPA and E-cadherin had identical distributions at the light microscope level along the cell-cell borders in keratinocyte cultures at 1.0 rnM calcium. E-Cadherin was found almost exclusively at cellcell borders, while uPA was found not only at cell-cell borders but also in the perinuclear region of at least some cells.
Effect of Blocking E-Cadherin Function upon uPA Redistribution The calcium-induced redistribution of uPA to the cell-cell borders and its colocalization with E-cadherin suggested a link between UPA localization and the formation of intercellular junctions between keratinocytes. A regulatory role for E-cadherin in the organization of keratinocyte intercellular junctions has been proposed from studies that showed a delay in the redistribution of P-cadherin, /31 integrin, vinculin, and desmoplakin to the cell-cell borders upon calcium elevation in the presence of blocking antibodies to E-cadherin [6]. The redistribution of these markers was delayed but not completely prevented; redistribution to the cell-cell borders of each marker did occur after longer (i.e., overnight) incubation in high calcium in the continued presence of anti-E-cadherin antibody. To determine if blocking antibodies to E-cadherin had a similar inhibitory effect on uPA redistribution, cultures were preincubated with rabbit anti-E-cadherin IgG [23]; calcium was then elevated for various periods of time prior to immunofluorescent staining for uPA. Comparison with control cultures demonstrated that anti-E-cadherin IgG delayed the redistribution of uPA. After 4 h in high calcium, control cultures revealed cellcell border stain for uPA in most fields (Fig. 3C), as described above; however, cultures preincubated with antibody to E-cadherin showed almost no cell-cell border stain for uPA at this time (Fig. 3D). When cultures were incubated for 24 h in high calcium, uPA redistributed to the cell-cell borders even in the continued presence of anti-E-cadherin IgG (Fig. 3F), and the staining pattern resembled that of the controls. These results on uPA localization in the presence of anti-E-cadherin antibody are very similar to those previously observed for several adhesion-related markers [6]. The explanation for the eventual recovery of redistribution ability in the continued presence of anti-E-cadherin antibody is unknown, but we have shown that it was not the result of recovery of E-cadherin function [6]. Receptor-Bound uPA Many cell types, including the keratinocyte [20, 301, bear a specific plasma membrane receptor for uPA. Recent studies have demonstrated that the uPA receptor is linked to the cell surface by a glycosyl- phosphatidylino-
‘FIG+ 2. CoIocaiization of E-cadberin and UPA in keratinocytes. (A-E): Nearly confluent keratinocytes on glass covers&x were incabate& in compkte medium containing 1.Q m&f calcium for 12 h+ The celIs were tbea processed for ~mrn~~o~~ore~e~ce with a mixture of murine anti-GA and rabbit ~ti-E-cadherin antibodies. (I3 and D): The loc~izat~o~ of~-ca~e~~. (C and E): The corr~s~oad~~~ IocaGzatio~ of uP-A in the same respective fields. (A): The phase contrast view of I3 and C. (l?-H): ~erat~noc”~es i~~bated in 30 fl calcium (S’} or in 1.0 mM calcium (G and HI for &e final 12 h of cuIture were processed for immu~o~~oresce~ce with antibody to transferFin receptor. (HI: The p&se contrast view of G. TransferriB receptor was never concentrated at the cell-ceII borders, even along areas of cIose ceII-cell contact at high calcium (G and l-X> arrowheads). Bar = 15 pm (A-C) or 6 ,um (D-H).
196
JENSEN
AND
WHEELOCK
FIG. 3. Antibodies against E-cadherin delayed the calcium-induced redistribution of uPA, Nearly confluent keratinocyte cultures in complete medium with 30 WM calcium were preincubated with normal rabbit IgG (A, C, and E) or rabbit anti-E-cadherin IgG (B, D, and F). Without changing the medium, the calcium concentration in some wells was then increased to 1.0 mM for varying lengths of time prior to immunofluorescent staining with monoclonal anti-uPA IgG (2B2). (A and B): Cells at 30 &f calcium. (C and D): Cells after 4 h in 1.0 mM calcium. (E and F): Cells after 24 h in 1.0 mM calcium. Bar = 20 pm.
with PI-phospholipase C; the cultures were then stained for uPA. Nearly all of the uPA staining at the cell-cell borders was lost in cultures so treated (compare Fig. 4A vs 4B and Figs. 4F vs 4G). This result strongly suggests that uPA at the cell-cell borders is on the cell surface and receptor-bound. In several experiments, cultures were exposed either to isotonic low-pH buffer alone or to PI-phospholipase C alone prior to uPA staining. The diminution of uPA staining intensity following either treatment on its own was extremely variable from experiment to experiment
and from field to field. In contrast, in each of three experiments in which cells were exposed both to low pH and PI-phospholipase C, nearly all cell-cell border stain for uPA was lost upon analysis by either immunofluorescence or immunoalkaline phosphatase detection methods. As a control for possible nonspecific effects of the low pH and PI-phospholipase, staining for E-cadherin was also performed on treated cells. No consistent difference was observed in the pattern of staining for E-cadherin in treated compared to control cultures, although
KERATINOCYTE
LJROKINASE
in some fields of the treated cultures the staining was a little less intense or less distinct at the cell-cell borders (Figs. 4D and 4E)* DISCUSSION
Rapid translocation of adhesion molecules to the cell-cell borders as well as structural organization of both adherens junctions and desmosomes are well documented in keratinocyte culture after calcium elevation [l-4,6]. In the present study, we have shown that localization of the proteolytic enzyme uPA in the keratinocyte is also controlled by calcium concentration. Urokinase PA distribution in the keratinocyte has several features in common with the distribution of a number of adhesion molecules including E-cadherin, P-cadherin, /Xl integrin, vinculin, and desmoplakin: (i) Each of these molecules is redistributed to the cell-cell borders upon elevation of the calcium concentration to 1.0 mM. (ii) The time course for redistribution of all molecules is rather rapid, although E-cadherin redistribution consistently precedes tbat of uPA. (iii) Inhibitory antibodies to E-cadherin delay by several hours the calcium-induced redistribution of each molecule. (iv) With time, redistribution of botb uPA and the adhesion-related molecules eventually occurs even in the continued presence of anti-E-cadherin antibodies [6]. These findings suggest that uPA localization is coordinated with intercellular junction formation in the keratinocyte. Our previous data as well as those of others indicate that E-cadherin plays a regulatory role in the early stages of intercellular junction organization [6-81. Comparison of the time course for uPA redistribution (Pig- 1) and E-cadherin redistribution [6] reveals that the latter is more rapid, consistent with its regulatory role. Cadherins are thought to be linked to the cytoskeleton through a group of proteins termed catenins; studies with anti-cadherin antibodies have shown that the catenins coimmunoprecipitate with cadherins [31-371. In the keratinocyte, we have previously shown that the profile of proteins that coimmunoprecipitate with Ecadherin is identical in cells maintained in 30 pM calcium or switched to 1 mM calcium for 12 h [36], in agreement with observations from other cell types [36]. Urokinase PA is not found among the proteins that coimmunoprecipitate (Wheelock? unpublished observations)Y indicating that uPA is not tightly associated with E-cadherin Urokinase PA is synthesized and secreted in many cell types [38], including the keratinocyte [39], as a 55a single-chain precursor molecule. Little if any conversion of the single-chain precursor to the two-chain enzyme is found in keratinocyte culture [39]. Like many other cell types [llS 38], the human keratinocyte [20,30] bears a specific, ceI1 surface uPA receptor that recently
LOCALIZATION
197
has been shown to be linked to the plasma membrane by a glycosyl-phosphatidylinositol lipid an&or [26]. There is evidence that at least some proteins so anchored have enhanced mobility in the plasma membrane (for review see Ref. [4O]). Under our culture con tions? the keratinocyte uPA receptor is occupied with endogenous uPA [20]. In the present study, uPA stain at the cell-cell borders is lost in cultures treated face-acting reagents (isotonic 1 phospholipase). These agents ba tively to dissociate ligand from receptor and to release glycosyl-pbosphatidylinositol-linked proteins, including the uPA receptor [20, 25, 26]. These data strongly suggest that the uPA concentrated along the keratinocyte-keratinocyte borders in response to calcium elevation is on the outside of the ceil a is r~~e~to~-bound. Previous studies have &own A is not rapidly internalized receptor9 but rather appears to be concentrated at the cell surface [41], awaiting activation for pericellular proteolysis when appropriate. The mechanism of receptor-bound uP-4 activation is likely to involve cell-bound plasmin [42]. In summary, we suggest that receptor-bound uPA behaves similarly to several adhesion-related molecules in that its cellular distribution is controlled by calcium concentration and can be modulated by E-cadherin. When concentrated along the ceil-cell borders of’ keratinocytes that exhibit intercellular junctions uPA is poised to participate in the cornroIled disruption of cell-cell adhesion structures that likely accompanies epidermal remodeling or morphogenesis. This work was supported by Grants CA 44464 and AR-39674 sub 06 from the National Institutes of Health to M.J.W. and I’.J.J.? respectively and by Grant DIR-8804006 from the National Science Foundation to M.J.W. The authors thank Dr. Norman Schechter for critical reading of the manuscript. Tile technical assistance of Mr. Kelby Frame in propagation of keratinocyte cultures is acknowledged with thanks.
1.
O’Keefe, E. J., Briggaman, 3iol. 105, 807.
2.
Hennings, H., Michael, D., Cheng, C., Steinert, I?~, Holbrook, and Yuspa, S. H. (1980) Cell 1 Hennings, H., and Holbrook, K. (1983) Exn. Ceil Res. 143, Watt, F. M., Mattey, D. L., and Garrod, D. I% (1984) J. Cell 99, 2211. Takeichi, M. (1988) &velopmerzt 102,639. Wheelock, M. J., and Jensen, I’. J. (1992) J. Cell BioZ. 117, Gumbiner, B., Stevenson, I%, and Grimaldi? A. (1988) J. 3iol. 107, 1575. Jongen, W. M. F., Fitzgerald, D. J., Asamoto? M., I’iccoli,
3. 4. 5. 6. I. 8.
R. A., and Herman,
B. (1587)
J. CelZ K., 127. I?ioi.
415. Cell C.>
198
JENSEN Slaga, T. J., Gras, D., Takeichi, CellBiol. 114, 545. Dane, K., Behrendt, Molecular Biology New York.
M., and Yamasaki,
N., Ronne E., Ellis, of the Cardiovascular
H. (1991)
Vaheri, A., Stephens, Tapiovaara, H. (1990)
11.
Vassalli, J.-D., uest. &3,1067.
12
Pollanen, Res. 57,
13.
Gold, L. L, Schwimmer, R., and Quigley, J. P. (1989) 262,529. Hebert, C. A., and Baker, J. B. (1988) J. Cell BioZ.
14.
R. W., Salonen, E.-M., PoIlanen, Cell Diff. Deu. 32, 255. A.-P.,
J., Stephens, 273.
J., Hedman, J. CelL Biot.
and Belin,
R. W., and Vaheri,
K., Nielsen, 106,8’7.
J.
V., and Blasi, F. (1990) System, p. 173, Liss,
10.
Sappino,
AND
D. (1391) A. (1991)
b. S., Dane,
J., and
J. %a.
In-
Adu. Cancer Biochem.
106,
J.
Pollanen, A. (1988)
16.
Pollanen, J., Vaheri, A., Tapiovaara, H,, Riley, E., Bertram, Woodrow, G., and Stephens, R. W. (1990) Proc. Natl. Acad. USA S7,2230.
K., and Vaheri, K., Sci
17.
Jensen, P.
18. 19.
Boyce, S. T., and Ham, R. G. (1983) J 1rxxsr. DerrrzutoZ. 81,33s. Shipley, G. D., and Pittelkow, M. R. (1987) Arch. Dermatol. 123, 154la.
20.
McNeill,
21.
Morioka, S., Jensen, Res. 161, 364.
22.
Grendahl-Hansen, J., Ralfkiaer, E., Nielsen, P., Frentz, G., and Dane: K. (1987) J. Zrtvest.
23.
Wbeelock, M. J., Buck, C!. A., Bechtol, K. B., and Damsky, C. H. (1987) J. Cell. Biachem. 34, 187. Jensen, P, J., John, M., and Baird, J. (1990) Exp. Cell Res. 187, 162.
Andreasen,
24.
Received Revised
March version
H., and Jensen,
P. J. (1990)
P. J., and Lazarus,
30, 1992 received June
Cell Reg.
J., and
1, 843.
G. S. (1985)
25.
26. 27.
28. 29. 30.
1241.
15.
H., Christensen, E. I., Ebbesen, P., Gliemann, P. A. (1990) Ceu Eeg. 1,1043.
W~ELOCK
31. 32. 33. 34. 35. 36. 37. 38.
Exp.
Cell 39.
1, 1992
L. S,, Kristensen, Dermatol. 68,28.
40. 41.
42.
Stoppelli, M. P., Tacchetti, C., Cubellis, M. V., Corti, A., Hearing, V. J., Cassani, G., Appeila, E., and Blasi, F. (1986) Cell 46, 675. Ploug, M., Renne, E., Behrendt, N., Jensen, A. L., Blasi, F*, and Dane, K. (1991) J. Bio!. Chem. 266,1926. Carter, W. G., Wayner, E. A., Bouchard, T. S., and Kawr, P. (1990) J. Cell Bio!. 110,1387. Larjava, H., Pehonen, J., Akiyama, S. K., Yamada, S. S., Gralnick, H. R., Uitto, J., and Yamada, K. M. (1990) J. Ceil Viol. 110, 803. Marchisio, P. C., Bondanza, S., Cremona, O., Caneedda, R., and Re Luca, M. (1991) J Cell BioL 112, 761. Del Rosso, M., Fibbi, G., Ihi, G., Grappone, C., Pucci, M., Cal” dini, R., Magnelli, L., Fiminai, M., Lotti, T., and Panconesi, E. (1990) J. Znvest. Dermatol. 94, 310. Kemler, R., and Ozawa, M. (1989) Bioessays 11,88. McCrea, P., Turck, C. W., and Gumbiner, B. (1991) Science 254,1359. Nagafuchi, A., Takeichi, M., and Tsukita, S. (1991) Cell 65,849. Peyrieras, N., Louvard, Aead. Sci. USA 62,8067.
D.,
and
Jacob,
F. (1985)
Prac.
?.ruti.
Wheelock, M. J. (1990) Z&p. Cell Res. 191,186. Wheelock, M. J., and Knudsen, K. A. (1991) 1n Viuo 5,505. Wheelock, M. J., and Knudsen, K. A. (1991) Differentiatzbn 46, 35. Blasi, F., Vassalli, J.-D., and Dane, K. (1987) J. Cell Biol. 104, 801. Hashimoto, K., Prystowsky, J. H., Baird, J., Lazarus, G. S., and Jensen, P. J. (1988) J. Znvest. Dermatol. 90,823. Low, M. G. (1989) Biochim. Z3iophys. Acta 988,427. Stoppehi, M. P., Corti, A., Soffientini, A., &assani, G., Blasi, F., and Assoian, R. K. (1985) Proc. Natl. Acad. Sei. USA 62,4939. Ellis, V., Scully, M. F., and Kakkar, V. V. (1989) J. Biol. Chem. 264,2185.
