Eur. J. Biochcm. 199, 337-345 (1991) FEBS 1991

r]

001429569100448A

Enhancement of vitronectin expression in human HepG2 hepatoma cells by transforming growth factor-/3l Katri KOLI ’, Jouko LOHI’, Aarno HAUTANEN’ and Jorma KESKI-OJA’.3 Departments of Virology, o f 2 Internal Medicine and of (Received January 4/March 8, 1991)

-

Dermatology and Vcnereology, University of Helsinki, Finland

EJB 91 0033

Liver cells are considered the principal source of plasma vitronectin. The human hepatoma cell line HepG2 produces vitronectin into its culture medium. In the current work we have analyzed the regulation of vitronectin by transforming growth factor-81 (TGFPI) in this hepatoma cell line by Northern hybridization, polypeptide and immunoprecipitation analyses and compared the response to another TGFP-regulated gene, plasminogen activator inhibitor (PAI-I). Rabbit antibodies raised against human plasma-derived vitronectin were used in immunodetection. Polypeptide and immunoprecipitation analyses of the medium and cells, as well as immunoblotting analysis of the cells and their extracellular matrices, indicated enhanced TGFPl -induced production and extracellular deposition of vitronectin. Accordingly, TGFPl enhanced the expression of vitronectin mRNA at picomolar concentrations (2 - 20 ng/ml) as shown by Northern hybridization analysis. Comparison of the temporal TGFB induction profiles of vitronectin and PAI-1 mRNAs showed that vitronectin was induced more slowly but the vitronectin mRNAs persisted longer. In addition, platelet-derived and epidermal growth factors had an effect on vitronectin expression, but it was of lower magnitude. TGF8l enhanced the expression of PAI-1 but, unlike previous reports, epidermal growth factor did not have any notable effect on PAI-1 in these cells. The results indicate that TGF8l is an efficient regulator of the production of vitronectin by HepG2 cells and that PAI-1 and vitronectin are not coordinately regulated. In addition, with affinity purified antibodies to vitronectin receptor, we observed strong enhancement of the M subunit of the receptor in response to TGF8l. These effects of TGFP are probably involved in various processes of the liver where matrix induction and controlled pericellular proteolysis is needed, as in tissue repair.

Vitronectin, also known as serum spreading factor or S-protein, is a multifunctional protein present in human plasma at concentrations of 200 - 500 yg/ml [l]. Vitronectin was identified as an adhesive protein which promotes the attachment and spreading of different cells in vitro. The only known synthesis site of vitronectin is the liver [2-41. Vitronectin is also present in insoluble form in tissues and these deposits of the protein probably accumulate from the circulation [5]. Vitronectin consists of a mixture of intact and proteolytically nicked 75-kDa polypeptides; the nicked polypeptide is held together by intramolecular disulfide bonds, and it separates into 65-kDa and 10-kDa polypeptides after chemical reduction [6]. This adhesion protein can promote the attachment and spreading of a variety of normal and neoplastic cells. The activity is mediated through an Arg-Gly-Asp adhesion sequence [7]. Specific cell surface receptors for vitronectin recognizing this tripeptide have been identified in endothelial, osteogenic, and fibroblastic cell types [8] (for review see [9]). Correspondence to J. Keski-Oja, Department of Virology, University of Helsinki, Haartmaninkatu 3 , SF-00290 Helsinki, Finland Abbreviations. u-PA, urokinase-type plasminogen activator; t-PA, tissue-type plasminogen activator; PAI, plasminogen activator inhibitor; TGFD, transforming growth factor-p; EGF, epidermal growth factor; PDGF, platelet-derived growth factor; bFGF, basic fibroblast growth factor; ConA, concanavalin A.

Vitronectin receptor (a,P3) belongs to the cytoadhesin family of integrins which share a common P3 subunit. Recently the M, subunit has been shown to associate also with other 8 subunits [lo, 111. The vitronectin receptor of hepatoma cells is reportedly of the a,p5 type [12]. Receptors for other members of the family of adhesive proteins including fibronectin, fibrinogen and von Willebrand factor also recognize the ArgGly-Asp sequence (see [I 31). The receptor - ligand interaction is very specific as shown with fibronectin and its receptor [14]. Vitronectin also interacts with collagen [ 151, glycosaminoglycans including heparin and binds to glass [6].The protein may have a regulatory function in the blood coagulation system, since it interferes with the inhibition of thrombin and coagulation factor X, by antithrombin 111 [16, 171. Also, in the complement system, vitronectin appears to be a regulatory protein; it is known to inhibit the cytolytic membrane attack complex [18]. Vitronectin, like other extracellular matrix components such as laminin and fibronectin, is involved in events requiring re-establishment of extracellular interactions including cell migration, tissue remodelling, and tumor invasion and metastasis. These events also involve controlled and targeted proteolysis where growth modulatory peptides have regulatory roles (see [19]). The wide spectrum serine protease plasmin is an important enzyme mediating the degradation of extracellular matrix. Urokinase-type plasminogen activator (u-PA) and tissue-type plasminogen activator (t-PA) initiate the pro-

338 teolytic cascade through catalyzing the conversion of protease zymogen plasminogen to active plasmin. These plasminogen activators therefore play a key role in the regulation of extracellular proteolytic events. Several of their inhibitors (PAI) have been identified, like the endothelial cell-type PA1 (PAI-l), the placental-type PA1 (PAI-2), the urinary PA1 (PAI-3), and the protease nexins. In addition, a 66-kDa urokinase-sensitive protein has been found in the extracellular matrix of cultured fibroblasts [20]. PAI-1 appears to be the most efficient inhibitor of u-PA and t-PA. Vitronectin and PAI-1 have been found to associate with each other in vitro and in cell culture [21], and vitronectin appears to act as a binding protein for the PAI-1 [22]. The interaction between these two proteins does not interfere with the ability of vitronectin to promote the adhesion and spreading of cells [23]. The binding of PAI-1 to vitronectin is known to stabilize and activate PAI-1 [24]. Transforming growth factor-@(TGFP) is a growth modulator that stimulates the growth of some fibroblastic rodent cells but acts as a strong growth inhibitor for cells of epithelial origin [25]. TGFP is a very potent regulator of PAI-1 production in a number of cell lines and it is also known to affect the proteolytic balance of the cells by directly affecting the expression of plasminogen activators (see [19]). The growth inhibitory effects of TGFP may be related to increased production of the matrix components and decreased proteolysis of the extracellular matrix [25]. In addition to its effects on the extracellular matrix, TGFP regulates the expression of the integrins, cellular receptors for matrix proteins like fibronectin and vitronectin [8, 261. In the current work we have used the human hepatoma cell line HepG2 as a model to analyze T G F j regulation of vitronectin production and extracellular proteolysis in cultured cells. We have compared the relationships between TGFPl regulation of vitronectin and PAI-1 by mRNA, immunoprecipitation and polypeptide analyses. Enhanced extracellular deposition of both proteins was observed by fluorography and immunoblotting. Our results indicate that T G F j affects the production of both vitronectin and PAI-1 in divergent ways in cultured liver epithelial cells. Enhanced production and extracellular deposition of vitronectin by TGFP may participate in TGFj-induced growth inhibition and augment the formation of the extracellular matrix of liver cells under various physiological and pathological conditions.

EXPERIMENTAL PROCEDURES Materials

Porcine-platelet-derived TGFPl and basic fibroblast growth factor (bFGF) were purchased from R&D Systems (Minneapolis, MN), epidermal growth factor (EGF) from Collaborative Research (Bedford, MA), and platelet derived growth factor (PDGF) from Bethesda Research Laboratories (Gathersburg, MD). Vitronectin and rabbit and monoclonal antibodies raised against vitronectin were a kind gift of Erkki Ruoslahti (La Jolla Cancer Research Foundation, La Jolla, CA). Antivitronectin receptor antibodies to human placental vitronectin receptor were prepared in rabbits as described [14]. The crude receptor antiserum was first passed through human-plasma-protein - Sepharose and fibronectin-receptor - Sepharose columns followed by vitronectin-receptor Sepharose column.

Cell cultures and treatment with growth,factors

Human hepatoblastoma (hepatocellular carcinoma, ATCC HB 8065) cells (HepG2) were cultivated in medium 199 containing 10% heat-inactivated fetal calf serum (Gibco, Paisley, UK), 100 IU/ml penicillin, and 50 pg/ml streptomycin. The cultures were incubated at 37°C in a humidified 5% COz atmosphere. Confluent cell layers were washed with serum-free medium 199 and then incubated under serum-free conditions for 8-24 h. The medium was then replaced by fresh medium 199, growth factors were added, and the cultures were incubated for an additional 1-36 h. All experiments were carried out under serum-free conditions. Radioactive labeling and polypeptide analyses

Confluent serum-starved cultures of HepG2 cells were labeled with [35S]methionine (50 pCi/ml, > 1000 Ci/mmol. Amersham, UK) in the presence of TGFPl (0.2 - 20 ng/ml) or EGF (2-20 ng/ml) at 3 7 T for 18 h. The media were collected and clarified by centrifugation. To isolate medium glycoproteins, especially PAI-1, 1-ml aliquots of the media were absorbed with 50 p1 50% (by vol.) suspension of concanavalin A linked to Sepharose (ConA - Sepharose; Pharmacia LKB Biotechnology Inc., Uppsala, Sweden) in phosphate-buffered saline (NaC1/Pi = 10 mM sodium phosphate, 170 mM sodium chloride, pH 7.4). Gelatin - Sepharose was used to isolate gelatin-binding proteins. The Sepharose particles were washed three times with 1 ml NaCI/P, and the bound proteins were dissolved in Laemmli’s gel sample buffer containing 10% 2-mercaptoethanol. Polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate (SDS/ PAGE) was performed according to Laemmli 1271using vertical discontinous sodium dodecyl sulfate/polyacrylamide gel slabs. The radioactive molecular mass markers used were myosin (200 kDa), phosphorylase b (92.5 kDa), bovine serum albumin (69 kDa), ovalbumin (46 kDa), carbonic anhydrase (30 kDa), and lysozyme (14.3 kDa; Amersham Corp.). Immunoprecipitation

Immunoprecipitation from the culture medium was carried out using mouse monoclonal antibodies to PAI-1 (Americian Diagnostica, Inc., New York, NY) or purified IgG fraction of rabbit anti-vitronectin antiserum; 1-ml aliquots of the media were incubated with the antibodies for 18 h at 4°C followed by absorption with 50 p1 protein-A- Sepharose (50% by vol. in NaCl/Pi) in an end-over rotary shaker for 1 h. The Sepharose particles were collected by centrifugation and washed three times with NaC1/Pi. The immunocomplexes were eluted from protein-A - Sepharose by Laemmli gel sample buffer and analyzed by 5% SDS/PAGE under reducing conditions. In blocking experiments purified vitronectin (10 pg/ ml) was added with a pretested dilution of the antibodies. Preparation of the extracellular matrices

After collection of the medium, the cultures were rinsed with NaCl/P,. Subsequently, the cultures were extracted three times for 5-min periods each with 10 mM Tris/HCl pH 8.0, containing 0.5% sodium deoxycholate and 1 mM phenylmethylsulfonyl fluoride in an ice bath [28]. The substrdtumbound proteins were extracted with Laemmli sample buffer and analyzed by SDSjPAGE and fluorogrdphy.

Immunohlotting analysis HepG2 cells were extracted with Laemmli's gel sample buffer after a 48-h incubation with TGFfil (0.2 -20 ng/ml) or EGF (2 - 20 ngjml) under serum-free conditions and subjected to SDSjPAGE under reducing conditions. Proteins were then transferred electrophoretically onto nitrocellulose membranes for 2 h at 480 mA. Membranes were saturated with 5% milk in NaCl/Pi/Triton X-100 (0.5%) and incubated with antibodies in 0.05 M Tris-HC1 pH 7.4 containing 0.05 M EDTA, 0.15 M NaC1, 0.05% Tween 20. After several washes with the same buffer, the bound antibodies were detected using peroxidase-conjugated anti-immunoglobulins, purchased from Dakopatts (Capenhagen, Denmark) or '2SI-labeled protein A (2 - 10 pCi/pg) purchased form DuPont-New England Nuclear (Hertfordshire, UK).

--

ECF

TGI'IJ

c 200

-

69

-

46

-

92.5

20

2

339

0.2

20

2

(ng/nll)

t FN

f

PAI-1

30-

Caseinolysis assays The radial caseinolysis assay used for the determination and quantification of the plasminogen activator activity was carried out as described [29]. The caseinolysis gels contdn plasminogen (KabiVitrum, Stockholm, Sweden) and casein in 1.2% agarose (FMC BioProducts, Rockland, ME). Plasminogen, when activated by plasminogen activator present in the medium sample, degrades casein and forms a clear disc of caseinolysis in the gel during the sample diffusion, proportional to the activity of the sample and time of diffusion. Human urokinase was used as a standard, and plasminogen activator activity is expressed in international units (IUjml). R N A isolation and Northern hybridization analyses Confluent cultures of cells were incubated under serumfree conditions for 24 h and exposed to growth factors (TGFP, EGF, bFGF, PDGF) and/or cycloheximide (1.5 pg/ml) or actinomycin D (10 pg/ml) as indicated. Poly(A)-rich mRNA was prepared using one cycle of oligo(dT)-cellulose chromatography [30]. RNA was quantified by absorbance measurement at 260 nm. For Northern analysis 5 - 10 pg RNA was fractionated on 1.2% agarose gels containing 2.2 M formaldehyde. Gels were treated with 3 M sodium chloride, 0.3 M sodium citrate, pH 7.0 for 45 min at room temperature and RNA was transferred to nitrocellulose membranes. Prehybridization and hybridization were performed at 42 "C in 50% deionized formamide, Denhardt solution, 0.75 M sodium chloride, 5 % SDS, l .25 M sodium phosphate, 0.005 M EDTA, pH 7.5. Membranes were hybridized to cDNA probes specific for human vitronectin [4] and bovine PAI-1 [31]. A cDNA probe specific for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as a control. The cDNA probes were labeled with 32Pusing the nick translation DNA labeling system (Amersham, UK). Membranes were washed several times at 65 "C in 0.1 5 M NaCl, 0.015 M sodium citrate, 0.1YOSDS. The relative amounts of radioactivity of the bands were estimated from fluorograms by laser scanning densitometry where indicated. RESULTS Polypeptide and immunoprecipitation analysis of secreted proteins To analyze growth-factor-induced alterations in the polypeptides secreted by HepG2 cells, confluent cultures were

Fig. 1 . ConA - Sepharose-binding proteins of HepG2-conditioned medium. Confluent cultures of cells were labeled with [35S]methioninein the presence of growth factors under serum-free conditions for 18 h as described in Methods. Aliquots of the medium were absorbed with ConA - Sepharose and the bound proteins, dissolved in Laemmli sample buffer, were analyzed by SDS/PAGE. Fluorograms of [35S]methionine-Labeled proteins are shown. The migration of fibronectin (FN) and PAI-1 are indicated on the right and molecular mass markers (in kDa) are shown on the left. The concentrations of growth factors are indicated on the figure. C: untreated control

exposed to TGFPl (0.2-20 ng/ml) or EGF (2-20 ng/ml) and labeled with [3 'Slmethionine under serum-free conditions. Changes of polypeptides were analyzed by SDSjPAGE followed by fluorography. In accordance with studies on other cells the analysis of the ConA-binding proteins of the culture medium indicated that the levels of PAI-1 were notably elevated in samples from TGF-P treated cells but not from EGFtreated cells at 18 h after the onset of the incubation (Fig. 1). The intensity of the fibronectin polypeptide (220 kDa) was enhanced in TGFP but not notably in EGF-treated cells. The enhancement of fibronectin by TGFP was seen more clearly in fluorograms of gelatin-binding proteins (gel not shown). Fibronectin was the only detectable gelatin-binding protein these cells secreted. Immunoprecipitation analysis of PAI-1 was in accordance with the results of the ConA-binding assay indicating enhanced synthesis of PAI-1 (Fig. 2A). As seen in both analyses (Figs 1, 2A) the TGFPl induction in PAI-1 production was strong and dose-dependent. In the immunoprecipitation analysis we found radiolabeled 65-kDa polypeptides that coprecipitated nonspecifically. Although vitronectin is known to act as a binding protein for the PAI-1 [21], possible interaction between PAI-1 and vitronectin was not investigated from the immunoprecipitates. However, vitronectin-like polypeptides were not observed in the ConA-binding proteins (Fig. 1). Immunoprecipitation analysis of the medium from TGFPl -treated HepG2 cells using polyclonal anti-vitronectin antibodies showed that the accumulation to the medium of a 65-kDa protein was enhanced in a dose-dependent manner while the effect of EGF was barely notable (Fig. 2B). The amount of the intact 75-kDa protein was only moderately increased, suggesting continuous processing to the lower-molecular-mass form. The relative amounts of the 75-kDa and 65-kDa polypeptides varied between different experiments (Fig. 2C). Their dose-dependent increase in TGFP-treated

340 A

TGFB

c

B

EGF

TGFB EGF 2 0.2' '20 2 ' (ngiml)

c '20

'20 2 0.2' '20 2 ' (ng/ml)

200

-

92.5

-

92.5

69

-

69

-

46

-

46

-

30

-

30

-

c FN

PAI-1

200

-

C

c FN

VN

1

200

-

92.5

-

69

-

46

-

2

3

4

-

FN

VN

Fig. 2. Itnmunoprecipitation analysis of vitronectin and PAI-1 from the culture medium of HepG2 cells treated with TGFB and EGF. Confluent cultures of cells were labeled with [35S]methionine in the presence of growth factors under serum-free conditions for 18 h as described in Methods. The radiolabeled proteins of the media were immunoprecipitated with a monoclonal antibody to PAI-1 (A) or purified IgG fraction of the anti-vitronectin antiserum (B). The specificity control (C) shows immunoprecipitation of the radiolabeled proteins of the medium with purified IgG fraction of rabbit anti-vitronectin antiserum (lane I), monoclonal anti-vitronectin antibodies (lane 2), or with monoclonal antivitronectin antibodies in the presence of purified vitronectin (10 pg) (lane 3). Lane 4 shows the absorption of the medium samples with proteinA - Sepharose which was comparable to normal rabbit IgG control. The samples were analyzed by SDSjPAGE followed by fluorography. The migration of fibronectin (FN), vitronectin (VN) and PAI-1 are indicated on the right and molecular mass markers (in kDa) are shown on the left. The concentrations of growth factors are indicated on the figure. C : untreated control

cells was, however, clearly seen. Monoclonal anti-vitronectin antibodies precipitated same bands seen using polyclonal antibodies and the precipitation of both 75-kDa and 65-kDa polypeptides was inhibited by purified vitronectin (Fig. 2C). Fibronectin (220 kDa), which binds to protein-A - Sepharose, was seen in all immunoprecipitation analyses as a nonspecific protein (Fig. 2 C). Precipitation with non-immune rabbit serum gave results comparable to protein-A - Sepharose binding proteins. These results indicate that the synthesis of vitronectin polypeptides is specifically enhanced by TGFPl . Coprecipitation of 48-kDa proteins was not observed. Extracellular deposition of vitronectin and PAI-1 are enhanced by TGFP TGFB modulates the expression of major extracellular matrix proteins, fibronectin and collagen in various types of normal and malignant cells [32]. This rapid and specific increase of protein synthesis is followed by increased incorporation of fibronectin and collagen into the matrix. In addition, the deposition of PAI-1 to the matrix is enhanced [33]. It was therefore of interest t o analyze whether TGFP stimulation also affects the deposition of vitronectin to the extracellular matrix. To observe the TGFP-induced alterations in extracellular proteins, HepG2 cells were labeled with [35S]methionine as described in Methods. The cultures were extracted with sodium deoxycholate and the matrices were dissolved in Laemmli's sample buffer. The samples were subjected to SDS/ PAGE and polypeptide alterations were detected from fluorograms of the gels. Analysis of the matrix preparations indicated that, in accordance with the secreted polypeptides, the extracellular deposition of PAI-1 (48 kDa) was enhanced in TGFPl-treated cells (Fig. 3). The TGFP-induced deposition of PAI-1 to the extracellular matrix was strong and dosedependent. The increase of PAI-1 in the matrix was also seen in immunoblotting analysis (not shown). In EGF-treated cul-

TGFR

lo.nz zoo

-

92.5

-

0.2

z

EGF

20'

'2

20

' c

(ngirnl)

69

-

4-

46

-

4-

30

-

4-

PAI-I

Fig. 3. Effect of TGFB on extracellular matrix proteins. Confluent cultures of cells were labeled with [35S]methioninein the presence or absence of growth factors for 18 h under serum-free conditions as described in Methods. Extracellular matrices were prepared from the cultures using the sodium deoxycholate extraction procedure. The culture dish was finally extracted with Laemmli's gel sample buffer and the samples were analyzed by SDSjPAGE followed by fluorography. The migration of the vitronectin polypeptides (two arrows) and PAI-1 are indicated on the right and molecular mass markers (in kDa) are shown on the left. The concentrations of growth Factors are indicated on the top of the figure. C : untreated control

tures the amount of PAI-1 protein was unaffected. Cultured HepG2 cells deposited only negligible amounts of fibronectin into their extracellular matrix as judged by the fluorograms. Unexpectedly, TGFPl had no major effect on the accumulation of fibronectin to the matrix in these cells (Fig. 3). In addition, two major radiolabeled proteins of 65 kDa (a doublet) and 75 kDa (a minor band in TGFP-treated cells) were

A LOO

-

C '20

TGFR

z

B

EGF

0.2' '20

z

C 200

-TGFB

(ngirnl)

0.2

z

20

z

(ng/mO

C

-

-

200

69

-

t t V N69

-

46

-

46

-

92.5

92.5

TGFP

EGF

20

341

2

20

(ng/ml)

c

VN

69 -

92.5

Fig. 4. Immunoblotting analysis of cell-associated vitronectin in HepG2 cultures. Confluent cultures of cells were treated with TGFP or E G F (concentrations shown on the figure) under serum-free conditions for 48 h as described in Methods. The extracts were analyzed by SDS/ PAGE followed by immunoblotting with purified IgG fraction of anti-vitronectin antiserum. (A) Total cell extracts; (B) extracellular matrices prepared using the sodium deoxycholate extraction procedure. The concentrations of growth factors are indicated on the figure. C: untreated control. Immunoreactive vitronectin and molecular mass markers (in kDa) are indicated

observed in the fluorograms (see arrows in Fig. 3). The molecular masses of the proteins suggested that they may be vitronectin. The molecular forms of extracellular vitronectin molecules was then assessed by immunoblotting analysis. Immunoblotting analysis was carried out of total cells and isolated matrices. Analysis of total cell extracts using polyclonal anti-vitronectin antibodies showed enhancement of three major polypeptides of 65 - 75 kDa in TGFB-treated cultures (Fig. 4A). To analyze the proportion of these retained in the extracellular matrix (substratum-bound proteins), we analyzed the matrix preparations (Fig. 3) by immunoblotting. It was found that the 75-kDa form was preferentially retained in the matrix (Fig. 4B). In addition, it was noted that E G F had a negligible effect on vitronectin accumulation to the matrix. Effect of TGFP on vitronectin receptor synthesis

Since TGFB regulates the expression of receptors for matrix proteins 1261, we wanted to analyze whether the expression of the vitronectin receptor was affected in these cells. To investigate effects of TGFP on vitronectin receptor synthesis, HepG2 cells, treated with TGFPl (2-20 ng/ml) or untreated, were extracted with Laemmli's sample buffer and analyzed by immunoblotting (see Methods) using polyclonal affinity purified placental vitronectin receptor ( c I , ~ ~anti) bodies. Bound antibodies were detected using '251-labeled protein A and fluorography. TGFP-treated cells showed strong enhancement of a polypeptide corresponding to the CI subunit (1 60 kDa) of vitronectin receptor compared to untreated cells (Fig. 5). The p subunit of the receptor (100 kDa), which is aberrant (P5) in hepatoma cells, was not detected in the fluorogram. The activity of TGFPI-induced PAL1

The analyze whether the TGFPl-induced PAI-1 was biologically active, we measured its effects on the proteolytic activity of the medium. Confluent cultures of the cells were

Fig. 5. Analysis of vitronectin receptor expression. Confluent cultures ofcells were treated with TGFPl (concentrations shown on the figure) under serum-free conditions for 24 h. Cells extracted with Laemmli's gel sample buffer were analyzed by SDSjPAGE followed by immunoblotting with affinity purified antibodies to isolated placental vitronectin receptor. Bound antibodies were detected using '''Ilabeled protein A and fluorography. The relevant portion of the gel is shown. Molecular mass markers (in kDa) are shown on the left and the concentrations of TGFB are indicated on the top of the figure. C : untreated control. Note that only the 160-kDa c(, subunit (arrow) was recognized by the antibodies

U-PA

[Uirnl]

TGFD

+

TGFD

-

U-PA

3

1.5

0.3

only

Fig. 6. Demonstrations qf PAI-1 activity in HepG2-conditioned medium. Confluent cultures of cells were incubated in serum-free medium for 18 h. After the incubation, new serum-free medium was changed and thc cells were incubated for 48 h in the presence or absence of TGFP (2 ngiml). Medium from TGFj-treated (TGFP +) and untreated cultures (TGFP-) were collected and analyzed by caseinolysis-in-agar assays. Urokinase (u-PA) standard dilutions were added to each samplc as in the lowest control line. Note that medium from untreated cultures also contained PA1 activity

washed twice with serum-free medium and incubated under serum-free conditions as shown in the legend to Fig. 6. Analysis by caseinolysis-in-agar assays indicated that the serum-free conditioned culture medium of HepG2 cells did not exhibit detectable caseinolytic activity in the presence or absence of plasminogen, suggesting that the total proteolytic activity of the cultures was low. After addition of purified u-PA to the wells of the caseinolysis plates (Fig. 6), it was found that the secreted PAI-1 was biologically active and inhibited the activity of u-PA in a dose-dependent manner. Addition of SDS (final concentration of 0.3%) to the caseinolysis wells did not decrease the activity, suggesting that the majority of PA1 activity was due to the SDS-insensitive PAI-1. Similar results were obtained using purified t-PA (not shown, see Fig. 7). The activity was not notably increased, suggesting that possible complexing with vitronectin did not decrease the

342 Relative value of mRNA

A Medium

-

tPA

10

50

-

10

50 nM

kDa 69

-

f VN

46

-

t PA!-1

B Matrix

0

1 0

-L

C

20 30 (TGFO 1 0 ngiml)

10 Time (h)

40

TGFB

Iinlc

Fig. 7. Effects of t-PA on TGFP-induced cell-associated PAI-1 and vitronectin. Confluent cultures of cells were labeled with [35S]methionine for 18 h in the presence or absence of TGFP (5 ngiml). The serum-free medium was changed and the cells were treated with increasing concentrations (shown on the figure) of t-PA for 3 h. The total proteolytic activity of the culture was determined by caseinolysis assays (A). Pericellular matrices were prepared and alterations of vitronectin and PAI-1 were observed in fluorography of SDSjPAGE (B). The migration of the immunoreactive 75-kDa vitronectin (VN) polypeptide (compare to Fig. 4B) and PAI-1 are indicated on the right and molecular mass markers are shown on the left. C: untreated control cells

activity of PAI-1. Purified vitronectin (3 pg) was also tested for PA1 activity in the caseinolysis assay with negative results (data not shown). The effects of t-PA on TGFP-induced cell-associated PAI-1 and vitronectin was analyzed as follows. Confluent cultures of cells were labeled with [35S]methioninefor 18 h in the presence or absence of TGFP. The serum-free medium was changed and the cells were treated with increasing concentrations (10 - 50 nM) of t-PA for 3 h. At the termination of the incubation the medium was collected, and the total proteolytic activity of the culture medium was determined by caseinolysis assays (Fig. 7). Pericellular matrices were then prepared and alterations of vitronectin and PAI-1 were observed in fluorography of SDS/PAGE. It was found that total proteolytic activity was not markedly inhibited in control cultures when compared to t-PA standard (not shown). In TGFP-treated cultues t-PA activity was markedly inhibited (Fig. 7A). Estimation of t-PA activity indicated a decrease of about 95% at the highest t-PA concentration used. Analysis of the fluorogram of the respective matrices prepared after termination of the t-PA treatment showed that the amount of PAI-1 was clearly reduced by t-PA while the amount of vitronectin was not notably affected (Fig. 7 B). The identity of the 75-kDa vitronectin was analyzed by immunoblotting (not shown, see Fig. 4B). The result of vitronectin immunoblotting of the matrices was in agreement with the fluorogram of the 75-kDa region. The release of TGFP-induced PAI-1 from the matrix was dependent on the amount of t-PA. The results also suggest that interaction of t-PA with matrix-associated PAI-I does not have any notable effect on the amount of matrix-associated vitronectin. Effect of TGFPI on the expression of the vitronectin and PAI-I genes

To study the temporal effects of TGFPl on vitronectin and PAI-1 gene expression at the mRNA level, confluent cultures of HepG2 cells were treated with TGFPl under serum-free conditions (see Methods). Cytoplasmic poly(A)rich mRNA was isolated and Northern hybridization analyses

(11)

0

1

6

12

24

36

kb -3.2 PAI-1 VN GAPDH

-2.3

- 1.5 - 1.3

Fig. 8. Time dependence of TGFP regulation of mRNAs f o r vitronectin andPA1-1. Confluent cultures of HepG2 cells were treated with TGFB (10 ng/ml) under serum-free conditions (see Methods). At times shown the cultures were lysed, and poly(A)-rich mRNA was isolated and analyzed by Northern hybridization using glyceraldehyde-3phosphate dehydrogenase (GAPDH), vitronectin (VN), and PAI-1 probes successively. The radioactivity was estimated from the fluorogram by laser scanning densitometry. For PAI-1, values of both mRNAs were summarized. The values of the graph are expressed as units relative to GAPDH (lowest lane in the fluorogram). All exposures are from the same filter

were carried out using vitronectin and PAI-1 cDNA probes. A probe specific for glyceraldehyde-3-phosphatedehydrogenase was used to control the relative amounts of mRNA. A TGFP/ time dependency study of the mRNAs showed that TGFP rapidly enhanced the expression of the PAI-1 gene with peak levels resulting between 6-12 h. This increase of PAI-1 mRNA was followed by a gradual decrease (Fig. 8). Unlike PAI-1, the mRNA for vitronectin was continuously increasing and after 36 h of incubation the relative mRNA level was still high (Fig. 8, chart). The effects of basic fibroblast growth factor (bFGF, 10 ngiml), platelet-derived growth factor (PDGF, 1 Ujml) and EGF (see below) on vitronectin mRNA synthesis were also evaluated as controls. The EGF concentrations in our experiments were the same as used by Lucore et al. [34]. EGF and PDGF increased the vitronectin mRNA levels but the effect was of lower magnitude than the enhancement in TGFP-treated cells (Fig. 9). The effect of PDGF was not further characterized. No effect in response to bFGF was observed after 5 h of incubation (data not shown). The doses of PDGF and bFGF used are known to be effective in growth stimulation of other cells. Analysis of different concentrations of PDGF (0.03 -3 Ujml) and bFGF (0.1 - 10 ng/ml) did not reveal major changes in mRNA levels. To determine whether TGFP-mediated regulation of PAI-1 and vitronectin gene expression required protein synthesis, HepG2 cells were treated with cycloheximide (1.5 pg/ ml) in the presence or absence of growth factors. Cycloheximide itself slightly enhanced the expression of vitronectin and PAI-1 genes and the inhibition of protein synthesis did not prevent the effect of TGFPl (Fig. 9A), indicating that protein synthesis was not needed for TGFD regulation of the vitronectin and PAI-1 genes. Actinomycin D (10 pg/ml), whichinhibits RNA synthesis, blocked the TGFPl induction of vitronectin mRNA (Fig. 9B), suggesting that the

343 A

X

.c

x

U

2

-

c

Y

- 3.2

PAI-1

- 2.2 VN

- 1.5

GAPDH

- 1.3

B

n z m

+

VN GAPDH

-1.5 - 1.3

Fig. 9. hyfect of cycloheximide und uciinomycin D on vitronecfin mRNAs. Confluent cultures of HepG2 cells were incubated under serum-free conditions for 24 h and then exposed to growth factors and drugs for 5 h (see Methods). (A) Treatment with cycloheximide (chx; 1.5 pg/ml) in the presence of T G F j (2 ng/ml) or EGF (2 ngiml); (B) treatment with T G F j ( 5 ng/ml) in the presence or absence of actinomycin D (act. D ; 10 pg/ml). In this experiment PDGF (1 IU/ ml) was used as a control (ctrl). After the incubation the cells were lysed, and poly(A)-rich mRNA was isolated and analyzed by Northern hybridization using glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and vitronectin (VN) probes as indicated on the figure. The sizes of the mRNAs are shown on the right

enhancement was due to continuous RNA synthesis rather than stabilization of the mRNA. Effects of E G F on vitronectin and PAI-I EGF has been reported to affect the expression of PAI-1 in HepG2 cells [34]. We therefore included EGF in our experiments as a control. We observed slight alterations at the mRNA levels for vitronectin but not for PAI-1 (Fig. 9A). Polypeptide analyses of the extracellular matrices and medium using immunoprecipitation and ConA-binding protein analyses for PAI-1 (Figs 1-3) further indicated that the effect of EGF on PAI-1 is negligible. The slight enhancement of EGF on vitronectin deposition into the matrix, as demonstrated by immunoblotting analyses of total cells and in the matrix preparations, is in agreement with the result of mRNA analysis (Fig. 4). This suggests that EGF or similar growth factors may regulate the expression of vitronectin in cultured liver cells. However, EGF did not affect the production of PAI-1.

DISCUSSION TGFBs appear to have important functions in the regulation of the formation and degradation of the extracellular matrices of cultured cells. TGFP elevates the production of fibronectin and procollagen resulting in enhanced formation of the matrix [32]. The role of TGFPl in hepatic fibrosis,

where the synthesis of several matrix proteins is increased, is under extensive investigation [35]. In addition to its effects on matrix formation, TGFPl modulates the extracellular proteolytic activity, which together may contribute to the growth inhibitory effects of TGFPl (see [19, 361). In addition to its direct effects on genes, TGFP affects the expression of a number of genes via the induction of activators of transcription, thejun oncogene family [37]. Vitronectin appears to act as a binding protein for PAI-1. PAI-1 is a constituent of the extracellular matrix of fibroblastic cells and its amount is enhanced by TGFP [33, 381. PAI-1 also associates to the extracellular matrix of endothelial cells [21]. Wun et al. [24] purified PAI-1 from HepG2-cellconditioned medium using immobilized anhydrourokinase. PAI-1 invariably copurified with vitronectin and they found evidence that vitronectin stabilizes and activates PAI-1. TGFPl is a major growth inhibitor for liver epithelial cells [39, 401. It may thus have a role in maintaining the normal quiescent state of hepatocytes in adult animals. TGFPl also inhibits the growth of hepatocytes removed from partially hepatectomized animals, [41] and enhanced TGFPl mRNA levels are present in nonparenchymal cells of the liver [42]. TGFPl can thus prevent uncontrolled hepatocyte proliferation during liver regeneration in vivo. TGFPl, if administered intravenously, is rapidly taken up by the liver and metabolized [43]. It was therefore of interest to analyze the response to TGFP at the subcellular level. In the present work we have characterized the role of TGFPl in the regulation of two major extracellular proteins, vitronectin and PAI-1, in the human hepatoma cell line HepG2 by polypeptide analysis, immunoblotting and Northern hybridization analysis. We find here that the synthesis and secretion of both vitronectin and PAI-1 are enhanced in HepG2 cells by picomolar concentrations of TGFPl . T G F j l increased the levels of mRNAs for vitronectin and PAI-1. The two different mRNAs of PAI-1 (3.2 and 2.3 kb) [31] were rapidly elevated by TGFPl, but the increase in the expression of the vitronectin gene was not as rapid as the PAI-1 gene. The vitronectin mRNAs showed steady increase and the relative mRNA level was still icnreasing after 36 h of incubation with TGFP1. At this time point the relative mRNA level was at least 10 times higher in TGFPl-treated cells. These results indicate that the mechanisms of mRNA elevation of the vitronectin and PAI-1 genes by TGFPl are different. In our Northern hybridization analyses using human fibroblasts and HT-1080 fibrosarcoma cells we did not find evidence for vitronectin mRNA expression (our unpublished results). Actinomycin D but not cycloheximide blocked the TGFPl induction of the vitronectin gene, which suggests that continous RNA synthesis but not protein synthesis, is needed for the enhancement of vitronectin mRNA. Cycloheximide, an inhibitor of protein synthesis, did not block the increase in PAI-1 mRNAs following TGFPl treatment but instead enhanced the effect. In human lung fibroblasts and carcinoma cells TGFP also rapidly elevates the PAI-1 mRNA levels, but in fibroblasts the induction is persistent while in carcinoma cells the levels of PAI-1 mRNA decline [44]. These results indicate that the PAI-1 mRNA response to TGFP is highly dependent on cell type. The increase in vitronectin and PAI-1 gene expression was followed by enhanced production and extracellular deposition of both proteins. Immunoblotting analysis for vitronectin of total cell extracts revealed immunoreactive polypeptides between 65-75 kDa, some of which may represent partially processed or degraded forms of the protein. Analysis of the substratum-bound proteins showed enhanced accumulation

344 of vitronectin into the matrix. Accordingly, immunoprecipitation analysis of the medium showed enhanced production of vitronectin polypeptides. These results indicate that TGFPl is an efficient regulator of the production of vitronectin by HepG2 cells. Immunoprecipitation with monoclonal PAI-1 antibodies and affinity binding (ConA) analyses of the proteins of the culture medium showed a dose-dependent increase in PAI-1 production. With caseinolysis-in-agar assays it was found that the secreted PAI-1 was biologically active. Polypeptide and immunoblotting analyses of the extracellular matrices indicated enhanced deposition of both vitronectin and PAI-1 in TGFP1-treated cells. In fibroblastic cells TGFP enhances the production and extracellular deposition of PAI-1 and the 66-kDa urokinasesensitive polypeptide [20, 331. It was therefore of interest to compare the relationships between vitronectin and the protease-sensitive extracellular fibroblast matrix protein of 66 kDa that is degraded to a 62-kDa form by urokinase and thrombin [20]. In our gel electrophoretic and immunoblotting analyses those proteins behave like distinct entities and no vitronectin immunoreactivity has been detected in isolated fibroblast matrices (our unpublished results). Evaluation of the effects of E G F on vitronectin and PAI-1 production revealed only minor alterations at the m R N A level for vitronectin, while immunoblotting and polypeptide analyses showed a negligible increase in vitronectin production and extracellular deposition. Unlike previous reports [34], no effects of E G F were observed on PAI-1 synthesis, extracellular deposition or at the mRNA level. This result may just reflect differences between HepG2 cell lines used in different laboratories. TGFP also affects the production of the receptors for the extracellular matrix components. In fibroblastic and osteogenic cells, the expression of vitronectin receptors is elevated and the synthesis and cell-surface expression of fibronectin receptors is also stimulated when the cells are exposed to TGFPl [8,45]. Accordingly, TGFP can efficiently regulate the expression of individual integrin subunits in various normal and malignant cells types [8, 261. As expected, we observed that the expression of the CI subunit of vitronectin receptor in HepG2 hepatoma cells was enhanced by TGFP. The antibodies raised against placental vitronectin receptor (a,P3) did not recognize the subunit of the hepatoma receptor. This is probably due to the association of the a, subunit in HepG2 hepatoma cells with the P5 subunit instead of /j3 [12]. Different x and P subunits can associate to form receptors with different adhesion properties which can influence cell migration and differentiation processes. By modulating integrin subunit expression, and thus integrin specifities, TGFP may modulate various cellular functions. The effect of TGFP on the vitronectin gene has various biological implications. The enhancement of vitronectin expression by TGFP supports the idea of TGFP as a major regulator of extracellular matrix formation and degradation. Interaction of vitronectin with PAI-1, which is also strongly regulated by this polypeptide growth factor, could be a n important way to stabilize PAI-1 and to regulate its activity at specific areas. Interactions of vitronectin with different macromolecules and its adhesive activity are likely to play a role in a number of pathophysiological events. The regulation of vitronectin gene by growth factors, especially TGFP, is a novel observation that suggests paracrine mechanisms as major effectors in the regulation of vitronectin synthesis in the liver.

We thank Ms Marja Valasjarvi for fine technical assistance. This work was supported by the Academy of Finland and the Finnish Cancer Foundation.

REFERENCES 1. Shaffer. M. C.. Folev. T. P. & Barnes. D. W. (1984) J . Lab. Clin. Med.’103, 783-76;. 2. Barnes, D. W. & Reing, J. (1985) J . Cell. Physiol. 125, 20721 4. 3. Jenne, D. &Stanley, K . K. (1985) EMBO J . 4,3153-3157. 4. Suzuki, S., Oldberg, A., Hayman, E. G., Pierschbacher, M. D. & Ruoslahti, E. (1985) EMBO J . 4, 2519.-2524. 5. Hayman, E. G., Pierschbacher, M. D., Ohgren, Y. & Ruoslahti, E. (1983) Proc. Natl Acad. Sci. USA 80,4003-4007. 6. Ruoslahti, E., Suzuki, S., Hayman, E. G., Ill, C. & Pierschbacher, M. D. (1987) Methods Enzymol. 144,430-437. 7. Hayman, E. G., Pierschbacher, M. D. & Ruoslahti, E. (1985) J . Cell Biol. 100, 1948- 1954. 8. Ignotz, R. A,, Heino, J . & Massague, J. (1989) J . Biol. Chem. 264, 389- 392. 9. Ruoslahti, E. (1988) Annu. Rev. Biochem. 57, 375-413. 10. Cheresh, D. A,, Smith, J. W., Cooper, H. M. & Quaranta, V. (1989) Cell57, 59-69. 11. Bodary, S. C. & McLean, J. W. (1990) J . Biol. Chem. 265,59385941. 12. Ramaswamy, H. & Hemler, M. E. (1990) EMBO J . Y, 1561 1568. 13. Hynes, R. 0. (1987) Cell48, 549-554. 14. Hautanen, A,, Gailit, J., Mann, D. M. & Ruoslahti, E. (1989) J . Biol. Chem. 264,1437-1442. 15. Gebb, C., Hayman, E. G., Engvall, E. & Ruoslahti, E. (1986) J . Biol. Chem. 261, 16698-16703. 16. 111, C. R. & Ruoslahti, E. (1985) J . Biol. Chern. 260, 1561015615. 17. Preissner, K. T. & Miiller-Berghaus, G. (1987) J . B i d . Chem. 262, 12247 - 12 253. 18. Tschopp, J., Masson, D., Schafer, S., Peitsch, M. & Preissner, K. T. (1988) Biochemistry 27, 4103-4109. 19. Laiho, M. & Keski-Oja, J. (1989) Cancer Res. 49, 2533-2553. 20. Keski-Oja, J. & Vaheri, A. (1982) Biochim. Biophys. Acta 720, 141- 146. 21. Mimuro, J. & Loskutoff, D. J. (1989) J . Biol. Chem. 264, 936939. 22. Declerck, P. J., De Mol, M., Alessi, M.-C., Baudner, S., Pgques, E.-P., Preissner, K. T., Miiller-Berghaus, G. & Collen, D. (1988) J . Biol. Chem. 263,15454-15461. 23. Salonen, E.-M., Vaheri, A., Pollanen, J., Stephens, R., Andreasen, P., Mayer, M., Dana, K., Gailit, J. & Ruoslahti, E. (1989) J . Biol. Chem. 264, 6339-6343. 24 Wun, T.-C., Palmier, M. O., Siegel, N. R. & Smith, C. E. (1989) J . Biol. Chem. 264,1862-1868. 25 Moses, H. L., Tucker, R. F., Leof, E. B., Coffey, R . J. J., Halper, J. & Shipley, G. D. (1985) Cancer Cells (Cold Spring Harbor) 3, 65-71. 26 Heino, J., Ignotz, R. A,, Hemler, M. E., Crouse, C. & Massague, J. (1989) J . Biol. Chem. 264, 380-388. 27 Laemmli, U. K. (1970) Nature 227. 680-685. 28. Hedman, K., Kurkinen, M., Alitalo, K., Vaheri, A,, Johansson, S. & Hook, M. (1979) J . Cell Biol. 81, 83-91. 29. Saksela, 0. (1981) Anal. Biochem. I l l , 216-282. 30. Maniatis, T. E., Fritsch, E. F. & Sambrook, J. (1982) Molecular cloning, a laboratory manual, Cold Spring Harbor Laboratory, Cold Spring Harbor NY. 31. Ny, T., Sawdey, M., Lawrence, D., Millan, J. L. & Loskutoff, D. J. (1986) Proc. Nut1 Acad. Sci. USA 83, 6176-6780. 32. Ignotz, R. A. & Massague, J . (1986) J . Bid. Chem. 261, 43374345. 33. Laiho, M., Saksela, O., Andreasen, P. A. & Keski-Oja, J. (1986) J . Cell Biol. 103, 2403-2410. 34. Lucore, C. L., Fuji, S., Wun, T.-C., Sobel, B. E. & Billadello, J. J. (1988) J . Bid. Chem. 263, 15845-15848. ~I

345 35. Czaja, M. J., Weiner, F. R., Flanders, K. C., Giambrone, M.-A., Wind, R., Biempica, L. & Zern, M. A. (1989) J . Cell Biol. 108, 2477 - 2482. 36. Keski-Oja, J., Blasi, F., Leof, E. B. & Moses, H. L. (1988) J . Cell Biol. 106, 451 -459. 37. Pertovaara, L., Sistonen, L., Bos, T. J., Vogt, P. K., Keski-Oja, J. & Alitalo, K. (1989) Mol. Cell. B i d . Y, 1255- 1262. 38. Laiho, M., Saksela, 0 . & Keski-Oja, J. (1987) J . Bid. Chem. 262, 17467-11474. 39. McMahon, J. B., Richards, W. L. del Campo, A. A,, Song, M.-K. H. & Thorgeirsson, S. S. (1986) Cancer Res. 46,4665-4671. 40. Wollenberg, G. K., Semple, E., Quinn, B. A. & Hayes, M. A. (1987) Cancer Res. 47, 6595-6599.

41. Strain, A. J., Frazer, A,, Hill, D. J. & Milner, R. D. G. (1987) Biochem. Biophys. Res. Commun. 145, 436 -442. 42. Braun, L., Mead, J . E., Panzica, M., Mikumo, R., Bell, G . I . & Fausto, N. (1988) Proc. Nut1 Acad. Sci. USA 85, 15391543. 43. Coffey, R . J. J . , Kost, L. J., Lyons, R. M., Moses, H. L. & LaRusso, N. F. (1987) J . Clin. Invest. 80, 750-751. 44. Keski-Oja, J., Raghow, R., Sawdey, M., Loskutoff, D. J., Postlethwaite, A. E., Kang, A. H. & Moses, H. L. (1988) J . Biol. Chem. 263,3111-3115. 45. Ignotz, R. A. & Massaguk, J. (1987) Cell51, 189-197.