Fibroblast growth factor 2 and transforming growth factor β1 interactions in human liver myofibroblasts

GASTROENTEROLOGY1995;109:1986-1996

Fibroblast Growth Factor 2 and Transforming Growth Factor Interactions in Human Liver Myofibroblasts JEAN ROSENBAUM, SYLVIE BLAZEJEWSKI, ANNE-MARIE PRI~AUX, ARIANE MALLAT, DANIEL DHUMEAUX, and PHILIPPE MAVIER INSERMUnit~ 99, H~pital Henri Mondor, Cr6teil, France

Background & Aims: During liver fibrogenesis, myofibroblastic liver cells proliferate and synthesize components of fibrosis. Fibroblast growth factor 2 (FGF-2) is expressed in vivo in myofibroblastic liver cells (MFLCs) during fibrogenesis, and exogenous FGF-2 is mitogenic for MFLCs. The aim of this study was to study the expression and role of endogenous FGF-2 in cultured human MFLCs. Methods: FGF-2 and FGF-2 receptors were studied using immunoblotting. All RNA studies used ribonuclease protection. Growth of MFLCs was studied using [aH]thymidine incorporation and direct cell counting. Results: MFLCs expressed FGF-2 and its receptors FGF receptor 1 and FGF receptor 2. An antibody to FGF-2 blocked the mitogenic effect of transforming growth factor ~1 (TGF-J~I) for MFLCs but not TGF-j~l-induced increase in cellular fibronectin messenger RNA (mRNA). TGF-]]I increased levels of FGF-2 and FGF receptor mRNAs in MFLCs. We have previously shown that TGF-~I also increased platelet-derived growth factor (PDGF) A chain mRNA in these cells and that anti-PDGF antibody blunted the mitogenic effect of TGF-I~I. The present results show that anti-FGF-2 and anti-PDGF-AA are not additive and that FGF-2 and PDGF-AA are not sequentially induced by TGF-I~I. Conc/usions: FGF-2 mediates the mitogenic but not the profibrogenic effect of TGF-~I for human MFLCs, and autocrine FGF-2 and PDGF-A interact in the mediation of the mitogenic effect of TGF-J]I.

'epatic fibrosis is the major complication of chronic .liver diseases such as chronic viral hepatitis or alcoholic liver disease. It is characterized by excessive deposition of extracellular matrix components in the liver parenchyma, leading to disturbances in sinusoidal blood flow and in uptake and secretion of soluble molecules by hepatocytes. ~ It is now clear from many experimental and human studies that Ito cells (also called lipocytes, fat-storing cells, stellate cells, or perisinusoidal cells) are responsible for deposition of fibrosis in the liver. 1 In normal liver, Ito cells are located in the space of Disse, surrounding sinusoidal endothelial ceils. They are the major site of vitamin A storage in the body. 2 In the course of fibrogenesis, Ito ceils undergo profound phenotypical

H

changes; they progressively lose their vitamin A droplets, acquire an important network of microfilaments and a well-developed rough endoplasmic reticulum, and express the specific smooth muscle isoform of CZ-actin. Together, these changes correspond to the myofibroblastic, or "activated," phenotype of Ito cells. 3'4 During fibrogenesis, these myofibroblastic liver ceils (MFLCs) proliferate and can be found in large numbers in the fibrous septa where they synthesize fibrosis components) The mechanisms responsible for the proliferation and increased matrix synthesis by MFLCs have been studied in detail using mostly cultured rat and less frequently human MFLCs. Several studies have focused on the role of polypeptide growth factors secreted by neighboring Kupffer cells 6'7 and, most interestingly, by MFLCs themselves.* We have thus shown that transforming growth factor ~1 (TGF-~I) was mitogenic for human MFLCs and that this effect was mediated in part by the secretion of platelet-derived growth factor (PDGF) A chain by human MFLCs themselves. 9a° We and other investigators have also shown that TGF-~I is a potent inducer of extracellular matrix production by human Ito cells and MFLCs. II,12 Fibroblast growth factor 2 (FGF-2), also known as basic fibroblast growth factor, is a prototype of the ninemember FGF family. 13 All FGFs interact with their target cells through high-affinity, transmembrane receptors. i4 There are at least four types of receptors with intrinsic tyrosine kinase activity, derived from separate genes and numbered FGF-R1 through FGF-R4. FGFR1,15 FGF-R2, *6 and, controversially, FGF-R4,17'i8 bind Abbreviations used in this paper: CHAPS,3-[(3-cholamidopropyl)dimethyl-ammonio]-l-propanesulfonate;DMEM, Dulbecco'smodified Eagle medium;EGF,epidermalgrowthfactor; FGF,fibroblastgrowth factor; FGF-R,fibroblastgrowthfactor receptor;G3PDH,glyceraidehyde-3-phosphatedehydrogenase;MFLC, myofibroblasticliver cell; MTT, 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazoiiumbromide; PDGF, platelet-derivedgrowth factor; PMSF, phenylmethylsulfonyl fluoride; RNase, ribonuclease;SDS-PAGE,sodiumdodecylsulfatepolyacrylamidegel electrophoresis;TGF,transforminggrowthfactor; tRNA, transferRNA. © 1995 by the AmericanGastroenterologicalAssociation 0016-5085/95/53.00

December 1995

FGF-2 with high affinity. To exert its biological effects, FGF-2 also needs low-affinity b i n d i n g sites that are cellassociated heparan sulfate proteoglycans. 19 W e previously found that the expression of FGF-2 was increased in the liver of rats treated with carbon tetrachloride. FGF-2 expression was spatially associated with fibrous septa and was localized in MFLCs. 2° Preliminary results also show FGF-2 association with fibrous septa in chronic liver diseases in humans. 21 Because FGF2 is mitogenic for rat and h u m a n MFLCs, 9'22 these data suggest that MFLC-derived FGF-2 could act in an autocrine way to sustain MFLC proliferation in vivo. Studies dealing with the effect of FGF-2 on the synthesis of extracellular matrix components by various cell types have yielded contradictory results. 23 26 The aim of the present study was to examine the expression of FGF-2 a n d of its receptors in cultured h u m a n MFLCs. W e were especially interested in the possible involvement of FGF2 in the mediation of the mitogenic and profibrogenic effects of T G F - ~ I and in the interactions between FGF2 and P D G F - A A .

M a t e r i a l s and M e t h o d s Culture of Human MFLCs MFLCs were obtained from explants of human liver obtained during hepatectomy as previously described, m For this study, 5 isolates were used. They originated from histologically normal liver fragments taken at the periphery of hepatocellular adenoma (n = 1), focal nodular hyperplasia (n = 1), hepatic metastasis in = 2), or hepatocellular carcinoma (n = 1). Tissues were obtained and handled in accordance with French ethical regulations. The ceils used in this study were at passages 2 - 8 . Ceils were routinely characterized by immunostaining for desmin and smooth muscle o~-actin. Exhaustive characterization of the cell isolates has been reported previously, lo

Measurement of MFLC Proliferation DNA synthesis. MFLCs were seeded at a density of 5000/weU in 96-well plates (Costar, Cambridge, MA) in Dulbecco's modified Eagle medium (DMEM) (GIBCO-BRL; Life Technologies, Cergy-Pontoise, France) containing 5 % fetal calf serum and 5% human serum, blood group AB (DMEM 5/5), and allowed to grow to confluence. Quiescence was induced by a 2 - 3 - d a y incubation in serum-free Waymouth's medium (GIBCO-BRL). D N A synthesis was induced by the addition of recombinant human growth factors previously shown to be mitogenic for MFLCs. 9'1° The growth factors used were TGF~1, PDGF-AA or PDGF-BB, and epidermal growth factor (EGF) (GIBCO-BRL). In some experiments, the monoclonal antibody to FGF-2, DG2 (a generous gift ofT. Reilly, DuPontMerck Pharmaceutical Co., Wilmington, DE), 27 or an isotypematched mouse monoclonal immunoglobulin (Ig) G, MOPC 21 (Sigma, L'Isle d'Abeau, France), were included in the experi-

FGF-2 IN HUMAN LIVER MYOFIBROBLASTS 1987

mental media. DG2 antibody has been extensively characterized. It impairs binding of FGF-2 to its cell surface receptors and blocks its mitogenic ef{ect, er The ceils were then incubated for 2 days with [methyt-3H]thymidine (25 Ci/mmol, 0.5 ~tCi/ well; Amersham, Les Ulis, France) present for the last 40 hours. Trichloroacetic acid-precipitable radioactivity was then measured as described) ° Three or six wells were used for each experimental point. Cell growth. MFLCs were seeded at a density of 105/ dish in 35-mm dishes. On the following day, the medium was replaced by Waymouth's medium supplemented with 0.5% human serum, which was growth factor-depleted by heparinSepharose chromatography l° and containing test substances. Fresh medium and additives were renewed at day 3. After 6 days, the cells were detached with trypsin-ethylenediaminetetraacetic acid (EDTA) and counted in a hemocytometer. In experiments examining the effect of anti-PDGF-AA and the combined effect of anti-PDGF-AA and anti-FGF-2, cell growth was assessed using a colorimetric assay. Briefly, cells were seeded at low density in 96-well plates. Agonists were added at day + 1 and renewed at day +4. Reduction of (3[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide) (MTT) was measured at day +7, as previously described, e* The anti-PDGF-AA antibody used in these experiments was from R&D Systems (Oxon, England). Normal goat IgG (Sigma) was used as control.

Detection of FGF-2 and FGF-R1 in Cultured MFLCs by Immunoblotting Detection of FGF-2. To prepare samples for immunoblotting, monolayers were placed on ice, washed twice with cold phosphate-buffered saline (PBS), and scraped in PBS containing dithiothreitol (0.2 mmol/L), EDTA (1 mmol/L), and the protease inhibitors leupeptin (1 JIg/mL), aprotinin (1 btg/ mL), pepstatin A (1 btg/mL), and phenylmethylsutfonyl fluoride (PMSF) (1 mmol/L). The cells were collected by centrifugation, and the pellet was resuspended in 20 mmol/L Tris-HC1, pH 7.4, containing 1 mol/L NaC1, 0.1% 3-[(3cholamidopropyl ) dimethylammonio ] - 1 - propanesulfonate (CHAPS), and the previous additives. The ceils were lyzed by two freeze-thaw cycles followed by sonication. The lysate was centrifuged for 15 minutes at 12,000g, and the pellet was discarded. The NaC1 concentration of the supernatant was adjusted to 0.5 mol/L in 20 mmol/L Tris-HCl, pH 7.4, before adding heparin-Sepharose beads (Pharmacia, St. Quentin en Yvelines, France) pre-equilibrated in 20 mmol/L Tris-HC1 plus 0.5 mol/L NaC1 with dithiothreitol, EDTA, CHAPS, and PMSF at the above-mentioned concentrations. The mixture was agitated on a rotary shaker for 2 - 4 hours at 4°C. The beads were then sedimented in a benchtop centrifuge and washed four times with the equilibration buffer. The bound proteins were eluted by boiling for 3 minutes in Laemmli's reducing buffer. Proteins were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE, 12% acrylamide) and transferred to nitrocellulose (Hybond C "Extra"; Amersham) using a semidry system (Semi-Phor; Hoefer

1988

ROSENBAUM ET AL.

GASTROENTEROLOGY Vol. 109, No. 6

Table 1. Construction of RNA Probes Fragment ends (complementary RNA size) mRNA

Clone

(nucleotides)

Encoded region

Source

FGF-2 FGF-R1 FGF-R2 FGF-R4 PDGF-A chain G3PDH

PRF11A h2 TK14 pilE6-2 ia PDGF-A-D1 HHCMC32

Nco1-465/BamH1-872 (407) Bg/ll-914/Kpn1-1196 (282) Bglll-2099/ EcoRI-2291 (192) Mlu1-1173/ Sma1-1445 (272) Sa11-715/Stu1-972 (257) Hincll-84/ Hindlll-256 (172)

65% of coding region Transmembrane and juxtamembrane domains Cytoplasmic kinase domain Transmembrane domain Coding region NH2 part of the protein

Prats et al. 3° Johnson et al. 3± Houssaint et al. 32 Vainikka et al. 18 Betsholtz et al. 33 ATCC

ATCC, American Type Culture Collection (Rockville, MD).

Scientific Instruments, San Francisco, CA). Nonspecific sites were blocked by an overnight incubation at 4°C in buffer A (10 mmol/L Tris-HC1, p H 8, 0.17 mol/L NaC1, 2 mmoll L CaC12, 5% skimmed milk, and 0.2% Nonidet P40). The monoclonal anti-FGF-2 antibody DG22v was used at a concentration of 1 btg/mL in buffer B (10 mmol/L Tris-HC1, pH 8, 0.17 mol/L NaC1, 2 mmol/L CaC12, 0.5 % skimmed milk, and 2% Nonidet P40). Bound antibody was detected as described previously m with the Vectastain ABC-AP kit (Vector, Burlingame, CA). Negative controls included the use of the irrelevant isotype-matched IgG MOPC 21 instead of DG2 and preabsorption of DG2 with heparin-Sepharose-bound recombinant human FGF2 (Upstate Biotechnologies Inc., Lake Placid, NY). Detection of FGF-R1. All steps were performed at 4°C. MFLCs were washed twice with PBS and scraped in 1 mmol/L NaHCO 3 containing 5 mmol/L EDTA and the same protease inhibitors as for FGF-2 extraction. They were homogenized with eight strokes in a Dounce homogenizer (Blaessing Glass Spec Co., Rochester, NY) and then fi_trther disrupted using a Polytron (Kinematica AG, Littau-Luzern, Switzerland). The homogenate was centrifuged at 30,000g for 20 minutes. The pellet was redissolved in a 20 mmol/L HEPES buffer, pH 7.4, containing 10% glycerol and 2% Triton X-100 (vol/vol). An aliquot was diluted 1:2 with a twofold concentrated sample buffer and run on 7.0% SDS-PAGE. Blotting was the same as for FGF-2 except that 0.1% SDS was included in the transfer buffer. Monoclonal anti-FGF-R1 antibody (Upstate Biotechnologies Inc.) was used at a concentration of 5 ].tg/mL. Negative controls included omission of the primary antibody or its replacement with MOPC 21. The remainder of the procedure was as described for FGF2 detection.

RNA Preparation and Ribonuclease Protection Analysis of the Messenger RNAs for FGF-2, FGF-R1, FGF-R2, FGF-R4, and PDGF-A Chain MFLCs were seeded at a density of 106/dish in 10-cm dishes and allowed to adhere for 3 - 4 hours. They were then made quiescent by a 60-hour incubation in serum-free Waymouth's medium. Agonists were then added in fresh medium, and total RNA was extracted at the indicated time points according to the technique of Chomczynski and Sacchi 29 and

quantified by the measurement of its absorbance at 260 nm. Ribonuclease (RNase) protection assays were performed as pre• • 12 vlously described. Briefly, RNAs were hybridized in solution 32 to [ P]uridine triphosphate-labeled antisense riboprobes generated by in vitro transcription with the Gemini Riboprobe System (Promega, Madison, WI) in the presence of [ C ~ - 3 2 p ] uridine triphosphate. The probes, all of human origin, are shown in Table 1.18'3°-33Hybridization was performed at 50°C in a final volume of 30 btL with 1 0 - 2 0 btg of RNA and a 1 2 X 106 cpm probe. Yeast transfer RNA (tRNA) (Sigma) was used as a negative control. After overnight incubation, samples were incubated for 60 minutes at 37°C with 40 ~tg/mL RNase A (Boehringer-Mannheim, Meylan, France) and 2 btg/mL RNase T1 (Sigma). This was followed by addition of 0.5% SDS and 125 btg/mL proteinase K (final concentrations) and further incubation for 30 minutes at 37°C. Undigested hybrids were extracted with phenol-chloroform, ethanol-precipitated, and analyzed on denaturing polyacrylamide-urea gels. Dried gels were exposed for autoradiography using Hyperfilm (Amersham). In some experiments, the RPA II kit for RNase protection (Ambion, Austin, TX) was used according to manufacturer's instructions. Results were quantified by excising the band from the gel and counting the radioactivity in a scintillation counter. They were normalized for RNA loading using a glyceraldehyde-3-phosphate dehydrogenase (G3PDH) probe in the hybridization cocktail.

Analysis of Cellular Fibronectin Messenger RNA Abundance Cells were seeded at 2 × 105/35-mm dish in DMEM 5/5. After a few hours, the medium was replaced by serumfree Waymouth's medium, and the ceils were incubated for 24 hours. TGF-~I was added with or without anti-FGF-2 antibody or control IgG. The cells were incubated for 48 hours with these various additives, and RNA was prepared and analyzed by RNase protection with probes for fibronectin and G3PDH. Construction of the fibronectin probe, which detects both the plasma and cellular isoform, is described elsewhere. 12

Statistical Analysis Results are presented as mean -+- 1SE. Differences between groups were analyzed by one-way analysis of variance

December 1995

(ANOVA) using the Statview II 1.03A software (Abacus Concepts Inc., Berkeley, CA).

FGF-2 IN HUMAN LIVER MYOFIBROBLASTS 1989

A

A

B

C

2

3

Results Detection of FGF-2 and Its Messenger RNA in Cultured MFLCs Synthesis of FGF-2 by MFLCs was shown at the protein and messenger R N A (mRNA) level. Immunoblot analysis of MFLC extracts with a monoclonal antibody to FGF-2 showed four bands characteristic of human FGF-2 migrating at 18, 21, 22, and 24 kilodaltons (Figure 1A). These bands were not present when an irrelevant antibody was used instead of the anti-FGF-2 antibody or when the antibody was preincubated with an excess of FGF-2. Analysis of FGF-2 m R N A using a sensitive liquid hybridization technique showed the expected RNase-resistant fragment of 407 nucleotides (Figure 1B). No hybridization signal was found when tRNA was used instead of MFLC RNA.

69.0

-

45.0

-

29.0 24.0

-

20.1

-

14.2

-

Detection of FGF-2 Receptors in Cultured MFLCs Expression by MFLCs of the FGF-2 high-affinity receptor, FGF-R1, was shown by immunoblotting with a monoclonal antibody specific for the extracellular domain of FGF-RI. This antibody detected a single 100ll0-kilodalton band in MFLC membrane extracts (Figure 2A). m R N A analysis by RNase protection showed that MFLCs expressed transcripts for FGF-R1, as shown by the presence of a protected fragment of 282 nucleotides. This fragment was absent when tRNA was used for hybridization instead of MFLC RNA (Figure 2B). Furthermore, MFLCs also expressed transcripts for another FGF-2 high-affinity receptor, FGF-R2, as shown by the presence on autoradiograms of a specifically protected 192-nucleotide fragment (Figure 2B). No transcripts for FGF-R4 were found (data not shown).

B

1

Effect of Anti-FGF-2 Antibody on MFLC Growth In these experiments, we evaluated the possible autocrine involvement of the FGF-2/FGF-2 receptor sysFigure 1. MFLCs express FGF-2. (A) FGF-2 in MFLCs shown by immunoblotting. MFLC lysates were semipurified on heparin-Sepharose and analyzed by SDS-PAGE and immunoblotting using a monoclonal antiFGF-2 antibody (see Materials and Methods). MFLC extracts were loaded in lanes A-C; lane D contains recombinant human FGF-2 (1 ng). Lanes A and D were incubated with anti-FGF-2 antibody, lane B with an irrelevant antibody, and lane C with the anti-FGF-2 antibody preincubated with an excess of FGF-2. Molecular weight markers are indicated in kilodaltons. (B) FGF-2 mRNA in MFLCs shown by RNase protection. Total RNA extracted from MFLCs was analyzed by RNase protection with a probe specific for FGF-2. Lane 1, intact probe; lane 2, tRNA; and lane 3, MFLC RNA. The specifically protected fragment is outlined with an arrow.

~ ~i~i!i~ili~ii~il :~i ¸¸¸ i~

i:

i

D

1990

ROSENBAUM ET AL.

A

1

GASTROENTEROLOGYVol. 109, No. 6

2

3

- 200

-116

- 96 - 66

- 45

B

1

2

3 O

i~

~

1 -91

2

3


~

li ¸¸

FGFR1

FGFR2

Figure 2. Expression of FGF-2 receptors by MFLCs. (A) FGF-R1 in MFLCs shown by immunoblotting. An MFLC membrane-enriched fraction was prepared and analyzed using SDS-PAGE and immunoblotting (see Materials and Methods). FGF-R1 was detected with a monoclonal antibody specific for the extracellular domain of the protein. Lane 1, anti-FGF-R1 antibody; lane 2, irrelevant primary antibody; and lane 3, no primary antibody. Molecular weight markers are indicated in kilodaltons. (B) FGF-2 receptor mRNAs in MFLCs shown by RNase protection. Specific probes for either FGF-R1 (left) or FGF-R2 (right) were used. Design of both panels is identical: lane 1, intact probe; lane 2, tRNA; and lane 3, MFLC RNA. The specifically protected fragments are outlined with arrowheads.

tern on MFLC growth. This was performed by measuring the DNA synthesis of MFLC grown in serum-free medium and stimulated by various growth factors in the presence of an anti-FGF-2 antibody. As shown in Figure 3A, TGF-~I stimulated DNA synthesis over a wide range of concentrations. A plateau was attained at 0.5-

1 ng/mL (20-40 pmol/L) of TGF-~I, and the median effective concentration was 0.19 + 0.03 ng/mL (7.6 + 1.2 pmol/L). The effect of TGF-~I could be completely blocked by preincubation of the growth factor with a monoclonal antibody to TGF-~ (data not shown), showing that it was not caused by a contaminant but by TGF~1 itself. When increasing concentrations of anti-FGF-2 antibody were added to MFLCs stimulated with 0.5 ng/ mL TGF-~I, a dose-dependent inhibition of the mitogenic effect of TGF-~I was observed (Figure 3B). This inhibition reached 84% + 10% at a dose of 10 /.tg/mL of antibody. Half-maximal inhibition was already obtained at an antibody concentration of 1 ~glmL. Control IgG had no significant effect on the mitogenic potency of TGF~1. Neither anti-FGF-2 nor control IgG had any significant effect in the absence of TGF-~I. The decrease in [methyl-3H]thymidine incorporation was not artifactual because the actual growth of cells was similarly affected by anti-FGF-2 antibody over a period of 6 days (Figure 3C). The same results were obtained when cell growth was assessed using MTT reduction (data not shown). The effect of anti-FGF-2 antibody could not be explained by a cytotoxic effect; microscopic examination of the cultures failed to show evidence of cell death, and measurement of lactate dehydrogenase release did not show any differences between antibody-treated or control cells (data not shown). The specificity of the anti-FGF-2 antibody with respect t o the mitogenic effect of TGF-~I was tested by adding this antibody to MFLC-stirntflated by other mitogenic growth factors. As shown in Figure 3D, anti-FGF-2 antibody did not significantly modify [methyl-3H]thymidine incorporation in response to PDGF-AA, PDGF-BB, or EGF. Several lines of evidence argued against a direct interaction of the anti-FGF-2 antibody with TGF-~I ; the antiFGF-2 antibody did not block the inhibitory effect of TGF~1 in a standard CCL64 growth inhibition assay34 (Figure 4), and it did not label TGF-~I on Western blots (data not shown). These data collectively suggest that the mirogenic effect of TGF-~I for human MFLCs is indirect and mediated by endogenous, MFLC-derived FGF-2. In agreement with such an indirect effect is the analysis of the kinetics of DNA synthesis after addition of growth factor to quiescent MFLCs, showing that the bulk of DNA synthesis stimulated by TGF-~I was delayed compared with that induced by PDGF-AA or PDGF-BB (Table 2).

Effect of Anti-FGF-2 Antibody on Gene Expression of Cellular Fibronectin by Cultured MFLCs As a marker for the profibrogenic effect of TGF1, we chose cellular fibronectin, a molecule that is abun-

December 1995

FGF-2 iN HUMAN LIVER MYOFIBROBLASTS

A

1991

l

4

3 Q 2

0

0.01

0.~1

1

0

10

5

TGFfll (ng/ml)

C

IgG (gg/ml)

160

D

12o

0

g r.~

10

100 /"-

i

50, ~

40

og ,

Figure 3. FGF2 specifically mediates the mitogenic effect of TGF-~I for MFLCs. (A) Effect of recombinant human TGF-~I on [methyi-3H]thymidine incorporation by MFLCs. Confluent, quiescent cells were incubated for 48 hours with TGF-131at the indicated concentrations. [Methyl-3H]thymidine was present during the last 40 hours. The results are presented as the fold increase in counts per minute in cells treated with TGF-~I vs. those without TGF-~I and are the mean (+ISE) of 6 - 2 3 experiments conducted in sextuplicate. (B) The mitogenic effect of TGF-~I for MFLCs is inhibited by an anti-FGF-2 antibody. The experimental design was the same as in A. TGF-~I was used at a concentration of 0.5 ng/mL. A, Control antibody; @, anti-FGF-2 antibody. The results are expressed as the percentage of values obtained with TGF-~I alone and are calculated as follows: % = A - B/C - B, where A is TGF-~I + IgG, B is no additives; and C is TGF-~I alone. These results are the mean (+_ISE) of 6 experiments conducted in sextuplicate. Mean (+ISE) stimulation by TGF-~I was 3.4 + 0.3 in these experiments. *P < 0.05, Student's t test after ANOVA. Results with IgGs alone are not presented for the sake of clarity but did not significantly differ from control values at every concentration tested. (C) Anti-FGF-2 antibody inhibits TGF-131-induced MFLC growth. MFLCs were grown for 6 days in Waymouth's medium with 0.5% serum and the indicated additives. TGF-~I was used at a concentration of 1 ng/mL and the antibodies at 10 ~g/mL. The graph shows one representative experiment conducted in triplicate wells of 4 similar experiments. (D) Anti-FGF-2 antibody fails to inhibit the mitogenic effects of PDGF-AA, PDGF-BB, and EGF. The experimental design was the same as in A. PDGF-AA and PDGF-BB were used at concentrations of 20 ng/mL and EGF at 1 ng/mL. The results are expressed as in B and are the mean (+ISE) of 8 - 9 experiments conducted in sextuplicate. Mean (_+ISE) stimulation by PDGF-AA, PDGF-BB, and EGF alone was 5.2 _+ 1.0, 5.9 _+ 1.5, and 1.6 + 0.3, respectively. The differences were not significant by ANOVA.

dant in liver fibrosis and whose mRNA is highly upregulated by TGF-~I in a variety of ceils, including rat 35 and human Ito cells.12 Addition of either control or antiFGF-2 antibody had no effect on the basal expression of cellular fibronectin mRNA by MFLCs (Figure 5). As previously shown, TGF-~I increased cellular fibronectin mRNA expression by MFLCs. Anti-FGF-2 antibody did not alter significantly this effect compared with control IgG (Figure 5). Furthermore, exogenous human recombi-

nant FGF-2 (10 ng/mL) did not increase cellular fibronectin mRNA expression by MFLCs between 2 and 24 hours (data not shown).

Effect of TGF-131 on mRNA Expression of FGF-R1 and FGF-R2 by MFLCs We attempted to define the mechanisms underlying the mediation of the mitogenic effect of TGF-~I by FGF-2 by studying the expression of FGF-2 and its recep-

1992

ROSENBAUM ET AL.

GASTROENTEROLOGY Vol. 109, No. 6

100

3

.T

2 > o 1

f

° v,,,~

o

0

.

0

.

.

.

.

.

.

=

10

100

0

TGFt I (pg/ml) Figure 4. Anti-FGF-2 antibody does not block the growth inhibitory effect of TGF-J31for CCL64 cells. CCL64 cells were seeded at a density of 50,O00/well in 24-well plates in DMEM containing 0.3% fetal calf serum. After 4 hours, TGF-~I, which was preincubated for 1 hour at 37°C with either 10 pg/mL anti-FGF-2 antibody (l~), 10 ~Lg/mL MOPC 21 (~), or plain medium (ll) was added to the cells. [Methyl-3H]thymidine incorporation into DNA was measured during a &hour pulse 1 6 - 2 0 hours after adding TGF-~I. Results are the mean of triplicate wells from I representative experiment of 3.

tors after MFLC exposure to T G F - ~ I . RNase protection analysis showed that TGF-[31 increased FGF-2 m R N A steady-state level in MFLCs as early as 2 hours after stimulation. The peak level was observed at 6 - 8 hours with a decrease at 24 hours (Figure 6A). As shown in Figure 6B, the increase in FGF-2 m R N A steady-state level was dose dependent with respect to T G F - ~ I . In most experiments, an ~2-fold induction was obtained, although a maximum stimulation of 13.5-fold was observed in the experiment shown in Figure 6B (data normalized for RNA loading using G 3 P D H gene expression). The mRNAs for FGF-R1 and FGF-R2 were also increased by TGF-~I (Figure 7). A maximal induction of

Table 2. Kinetics of DNA Synthesis in MFLCs After S t i m u l a t i o n With TGF-~I or PDGF Labeling period

TGF-131 (1 ng/mL) PDGF-AA (10 ng/mL) PDGF-BB (10 ng/mL)

8-30 h

30-48 h

1:8 + 0.2 2.8 + 0.3 2.7 + 0.3

4.6 ± 0.9 2.7 + 0.4 2.1 ___0.3

NOTE. Confluent quiescent MFLCs were stimulated with TGF-131,PDGFAA, or PDGF-BB and labeled with [3H]thymidine during the indicated times. Results are expressed as fold DNA synthesis over baseline and the mean (_+ISE) of 4 experiments.

1

0

TGFt~I (ng/ml) Figure 5. FGF-2 does not mediate the TGF-131-induced increase in cellular fibronectin mRNA in MFLCs. Effect of anti-FGF-2 antibody on cellular fibronectin mRNA level. Cells were incubated for 48 hours with or without TGF-~I in the presence or absence of anti-FGF-2 (DG2) or control antibodies (10 #g/mL). Cellular fibronectin mRNA level was normalized using G3PDH message and expressed as fold increase over basal values. The results are the mean ( + I S E ) of 5 experiments. FI, No antibody; [], DG2; [], control IgG.

1.6-fold was observed at 8 hours for FGF-R1 m R N A and persisted until 24 hours. For FGF-R2 mRNA, the induction was 1.7-fold at 8 hours and increased to 3.2fold at 24 hours.

A

~

2

3

4

5

6

S

1

2

3

4

S

6

w

,!

Figure 6. TGFI31 increases the steady-state levels of FGF2 mRNA in MFLCs. (A) Kinetics of FGF-2 mRNA induction by TGF-131 in MFLCs. MFLCs were incubated with 1 ng/mL TGF-131. At various times, RNA was extracted and analyzed by RNase protection with a FGF-2 mRNA probe. Lane 1, intact probe; lanes 2-6, MFLC RNA extracted at O, 2, 4, 8, and 24 hours after TGF-I31 stimulation, respectively. (B) Doseresponse analysis of the induction of FGF-2 mRNA by TGF-I~I in MFLCs. MFLCs were incubated with TGF-]31 for 6 hours. RNAs were analyzed by RNase protection with an FGF-2 mRNA probe. Lane 1, intact probe; lane 2, tRNA; Lanes 3-6, MFLC RNA extracted from cells treated with O, 0.1, 0.5, and 1 ng/mL TGF-131, respectively. The apparent lack of visibility of FGF-2 transcripts in the absence of TGF-~I, compared with Figure 1B, is caused by a shorter exposure time in Figure 6A and B.

December 1995

1

FGF-2 IN HUMAN LIVER MYOFIBROBLASTS 1993

2

3 i

4

Table 3. Lack of Additivity of FGF-2 and PDGF-AA Antibodies on MFLC Proliferation

5 i

% of TGF[31-induced growth

Additive

~1-

FGFR1

Anti-FGF-2 antibody Control mouse IgG Anti-PDGF-AA antibody Control goat IgG Anti-FGF-2 + anti-PDGF-AA Control IgG

45,1 103.0 64.9 122.0 53.2 109,3

_+ 13.6 + 1.7 _+ 17,9 _+ 25,6 +_ 5.5 _+ 1.1

NOTE. Cells were seeded at low density in 96-well plates. Agonists were added at day +1 and renewed at day +4, Reduction of MTT was measured at day +7 as described in Materials and Methods. TGF-~I was used at a concentration of 1 ng/mL, anti-FGF-2 at 10 pg/mL, and anti-PDGF-AA at 50 ~g/mL. Results are the mean (+ISE) of 3 experiments,

~1-

FGFR2

Figure 7, TGF-~I increases the steady-state levels of both FGF-R1 and FGF-R2 mRNAs in MFLCs. MFLCs were either unstimulated (lane 3) or incubated with TGF-131 (1 ng/mL) for 8 hours (lane 4) or 24 hours (lane 5). RNA was extracted and analyzed by RNase protection with probes specific for FGF-R1 (top) or FGF-R2 (bottom). Unhybridized probe was run on lane 1, and the probe was hybridized with tRNA on lane 2,

FGF-2 and its receptors. Expression of FGF-2 is evidenced by the detection of its m R N A and the associated protein. MFLC-derived FGF-2 shows four bands on Western blot that are consistent with the multiple translation initiation sites on human FGF-2 mRNA. 3°'36 mRNAs for both high-affinity receptors FGF-R1 and FGF-R2 were found using RNase protection analysis. The presence of the FGF-R1 protein was also shown using Western blotting of crude membrane preparations. The size of the detected protein (110 kilodaltons) is compatible with one of the multiple possible splice variants

8 FGF-2-PDGF-AA

Interactions

W e have previously shown that T G F - ~ I induced PDGF-A chain m R N A and protein in human MFLCs and that an antibody to PDGF blocked D N A synthesis and growth of MFLCs. t° It was thus of interest to determine whether the effects of a n t i - F G F - 2 and anti-PDGF antibodies were additive. As shown in Table 3, although either antibody used alone inhibited the mitogenic effect of TGF-~I, their combination was not additive. This result suggested that the FGF-2 and the PDGF-AA pathways were somehow linked. W e examined the possibility that FGF-2 or PDGFAA, being both induced by TGF-[31, could in sequence increase the expression of one another. However, compared with T G F - ~ I , FGF-2 did not significantly increase the expression of PDGF-A chain m R N A (Figure 8), and PDGF-AA had only a minor effect on FGF-2 m R N A expression (data not shown).

Discussion This study shows that MFLCs possess all the components of an FGF-2 autocrine loop, i.e., expression of

~

4

0

2

4

6

8

Time (hours) Figure 8. Compared kinetics of induction of PDGF-A chain by TGFI~I and FGF-2. In these experiments, RNA samples were simultaneously hybridized with PDGF-A chain and G3PDH probes. Individual bands were excised and counted, and the results were normalized using the G3PDH signal. TGF-131 (11) was used at a concentration of 1 ng/ mL and FGF-2 ( 0 ) at 10 ng/mL, The results are the mean of 2 - 3 experiments.

1994

ROSENBAUM ET AL.

of FGF-R1. 37 Additionally, the synthesis of heparan sulfate by human MFLCs, which provides low-affinity binding sites for FGF-2, has been shown using specific enzymatic digestion of endogenously [35S]sulfate-labeled glycosaminoglycans (A. Gressner and J. Rosenbaum, unpublished results) and by immunocytochemical studies. 38 Finally, our data provide evidence that this FGF-2 autocrine loop is activated by TGF-~I because exposure of human MFLCs to TGF-~I leads to an increase in the mRNA levels of FGF-2 itself and of its high-affinity receptors FGF-R1 and FGF-R2. A role for MFLC-derived FGF-2 in the growth of these ceils was shown because addition of a blocking antibody to FGF-2 inhibited TGF-~l-mediated growth stimulation of human MFLCs. The mediation by FGF-2 seems to be specific of TGF-~I because anti-FGF-2 antibody did not alter the growth-stimulating effects of PDGFAA, PDGF-BB, or EGF. To our knowledge, our results thus show for the first time the involvement of a FGF2 autocrine loop in the mediation of a biological effect ofTGF-~I. However, FGF-2 has been shown to mediate the mitogenic effect of other agonists such as thrombin 39'4° or the combination of interferon alfa and interleukin 2. 41 As for TGF-[~I, these molecules do not act on tyrosine kinase-coupled receptors. It is thus possible that the FGF-2/FGF-2 receptor system provides the necessary tyrosine kinase for the mitogenic effect of these agonists. We also show that TGF-~I increases FGF-2 receptor mRNA levels in human MFLCs. Up-regulation of FGF2 receptors by TGF-~I has not been widely reported. In one study, TGF-~I increased FGF-2 binding sites in dermal fibroblasts, suggesting that this effect was responsible for the potentiation of the mitogenic effect of exogenous FGF-2 by TGF-~1.42 The induction of FGF-2 receptors by TGF-~I could explain the disparity between the mitogenic responses to TGF-~I and exogenous FGF2 in human MFLCs. *° The minor mitogenic effect of exogenous FGF-2 that we have reported 1° could be related to a low abundance of FGF-2 receptors, and, in the case of TGF-~I stimulation, coinduction of FGF-2 and FGF-R1 could amplify the mitogenic response to endogenous FGF-2. We previously reported that MFLC-derived PDGFAA also plays a role in the mitogenic effect of TGF-~I on MFLCs) ° In these experiments, TGF-~I increased the expression of PDGF-A chain mRNA and PDGFAA secretion by MFLCs, and TGF-~l-stimulated MFLC growth was partly blocked by an antibody to PDGF. In the present study, we show that the combination of antiFGF-2 and anti-PDGF-AA antibodies is not additive with regard to the mitogenic effect of TGF-~I. These combined data suggest a complex interaction between

GASTROENTEROLOGYVol. 109, No. 6

TGF-~I, PDGF-AA, and FGF-2, with FGF-2 and PDGF-AA being obligatory comitogens for human MFLCs. It has been shown that PDGF can increase FGF2 mRNA levels43 and that FGF-2 can mediate PDGFAA-induced migration of smooth muscle cells. 44 However, we could not show a significant induction of FGF-2 mRNA by PDGF-AA in our cellular model. In addition, PDGF-A chain mRNA level was not increased by FGF2. These data thus rule out a possible sequential induction of FGF-2 by PDGF or of PDGF by FGF-2, which would explain the comitogenic effect of these two growth factors. The interaction between the FGF-2 and PDGFAA pathways may be related to a transmodulation of the receptor of one of these growth factors by the other one. Indeed, such a mechanism has been shown for FGF-2, which up-regulates the g-type PDGF receptor in vascular smooth muscle cells. 45 We are currently investigating this issue in our laboratory together with a possible upregulation of FGF-2 receptors by PDGF-AA. Finally, we examined the role of endogenous FGF-2 in the mediation of the profibrogenic effect of TGF-~I. As a marker for this effect, we chose cellular fibronectin mRNA, a transcript that is highly regulated by TGF~1 in a variety of cells, including rat 35 and human Ito cells. 12 Cellular fibronectin mRNA level was increased by TGF-~I, but addition ofanti-FGF-2 did not significantly modify this effect. These results are in keeping with the lack of effect of exogenous FGF-2 on cellular fibronectin mRNA level and suggest that, in contrast to its mitogenic effect, the profibrogenic effect of TGF-~I on human MFLCs does not require FGF-2. Uncoupling between the effects of TGF-~I on cell growth and extracellular matrix production has already been reported in several cell types.46-48 It has been suggested that these two responses could involve different transduction pathways at the receptor level. 49'5° In conclusion, we show that FGF-2 acts as an endogenous mediator of the mitogenic effect of TGF-~I for human MFLCs but is not involved in its profibrogenic effect. Our results also suggest a complex interaction between FGF-2 and PDGF-AA in the mediation of the mitogenic effect of TGF-~I for human MFLCs. The relevance of FGF-2 in liver fibrogenesis now needs to be assessed in in vivo experiments to determine its possible interest as a target in the therapy of liver fibrosis. References 1. Friedman SL. The cellular basis of hepatic fibrosis. Mechanisms and treatment strategies. N Engl J Med 1993;328:1828-1835. 2. Hendriks HFJ, Elhanany E, Brouwer A, De Leeuw A, Knook DL. Uptake and processing of [3H]retinoids in rat liver studied by electron microscopic autoradiography. Hepatology 1988; 8 : 2 7 6 285.

December 1995

3. French SW, Miyamoto K, Wang K, Jui L, Briere L. Role of the Ito cell in liver parenchymal fibrosis in rats fed alcohol and a high fat-low protein diet. Am J Pathol 1988; 132:73-85. 4. Bachem MG, Riess U, Gressner AM. Liver fat storing cell proliferation is stimulated by epidermal growth factor/transforming growth factor alpha and inhibited by transforming growth factor beta. Biochem Biophys Res Commun 1989; 162:708-714. 5. Mak KM, Lieber CS. Lipocytes and transitional cells in alcoholic liver disease: a morphometric study. Hepatology 1988;8:10271033. 6. Friedman SL, Arthur MJP. Activation of cultured rat hepatic lipocytes by Kupffer cell-conditioned medium. Direct enhancement of matrix synthesis and stimulation of cell proliferation via induction of platelet-derived growth factor receptors. J Clin Invest 1989; 84:1780-1785. 7. Meyer DH, Bachem MG, Gressner AM. Modulation of hepatic lipocyte proteoglycan synthesis and proliferation by Kupffer cellderived transforming growth factors type 131 and type c~. Biochem Biophys Res Commun 1990; 171:1122-1129. 8. Bachem MG, Meyer D, Melchior R, Sell KM, Gressner AM. Activation of rat liver perisinusoidal lipocytes by transforming growth factors derived from myofibroblastlike cells. A potential mechanism of self perpetuation in liver fibrogenesis. J Clin Invest 1992;89:19-27. 9. Rosenbaum J, Mallat A, Pr6aux AM, Mavier P, Zafrani ES, Dhumeaux D. Effect of polypeptide growth factors on the growth of cultured human hepatic myofibroblast-like cells (transformed Ito cells). In: Wisse E, Knook DL, McCuskey RS, eds. Cells of the hepatic sinusoid. Volume 3. Rijswijk, The Netherlands: The Kupffer Cell Foundation, 1991:255-258. 10. Win KM, Charlotte F, Mallat A, Cherqui D, Martin N, Mavier P, Pr~aux AM, Dhumeaux D, Rosenbaum J. Mitogenic effect of transforming growth factor-131 on human Ito cells in culture: evidence for mediation by endogenous platelet-derived growth factor. Hepatology 1993; 18:137-145. 11. Casini A, Pinzani M, Milani S, Grappone C, Galli G, Jezequel AM, Schuppan D, Rotella CM, Surrenti C. Regulation of extracellular matrix synthesis by transforming growth factor 131 in human fatstoring cells. Gastroenterology 1993; 105:245-253. 12. Blazejewski S, Pr6aux AM, Mallat A, Mavier P, Dhumeaux D, Schuppan D, Rosenbaum J. Human myofibroblast-like cells obtained by outgrowth are representative of the fibrogenic cells in the liver. Hepatology 1995;22:788-797. 13. Coulier F, de Lapeyri~re O, Birnbaum D. Complexit6 de la famille des facteurs de croissance FGF: la preuve par 9. M~decine/ Sciences 1993; 10:1113-1115. 14. Givol D, Yayon A. Complexity of FGF receptors: genetic basis for structural diversity and functional specificity. FASEB J 1992;6:3362-3369. 15. Ornitz DM, Leder P. Ligand specificity and heparin dependence of fibroblast growth factor receptors 1 and 3. J Biol Chem 1992; 267:16305-16311. 16. Dionne CA, Crumley G, Bellot F, Kaplow JM, Searfoss G, Ruta M, Burgess WH, Jaye M, Schlessinger J. Cloning and expression of two distinct high-affinity receptors cross-reacting with acidic and basic fibroblast growth factors. EMBO J 1990;9:26852692. 17. Ran D, Reich R, Chedid M, Lengel C, Cohen OE, Chart AML, Neufeld G, Miki T, Tronick SR. Fibroblast growth factor receptor 4 is a high affinity receptor for both acidic and basic fibroblast growth factor but not for keratinocyte growth factor. J Biol Chem 1993; 268:5388-5394. 18. Vainikka S, Partanen J, Bellosta P, Coulier F, Basilica C, Jaye M, Alitalo K. Fibroblast growth factor receptor-4 shows novel features in genomic structure, ligand binding and signal transduction. EMBO J 1992;11:4273-4280. 19. Yayon A, Klagsbrun M, Esko JD, Leder P, Ornitz DM. Cell surface,

FGF-2 IN HUMAN LIVER MYOFIBROBLASTS

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

1995

heparinlike molecules are required for binding of basic fibroblast growth factor to its high-affinity receptor. Cell 1991;64:841848. Charlotte F, Win KM, Pr6aux AM, Mavier P, Dhumeaux D, Zafrani ES, Rosenbaum J. Immunolocalization of heparin-binding growth factors (HBGF) types 1 and 2 in rat liver. Selective hyperexpression of HBGF-2 in carbon tetrachloride-induced fibrosis. J Pathol 1993; 169:471-476. Rosenbaum J, Charlotte F, Pr6aux AM, Dhumeaux D, Zafrani ES, Mavier P. Overexpression of heparin-binding growth factor 2 (HBGF 2) in human cirrhotic liver. Evidence for a role of Ito cellderived HBGF 2 in liver fibregenesis. Hepatology 1992; 16:99A. Pinzani M, Gesualdo L, Sabbah GM, Abboud HE. Effects of platelet-derived growth factor and other polypeptide mitogens on DNA synthesis and growth of cultured rat liver fat-storing cells. J Clin Invest 1989;84:1786-1793. Tan EML, Rouda S, Greenbaum SS, Moore JH, Fox JW, Sollberg S. Acidic and basic fibroblast growth factors down-regulate collagen gene expression in keloid fibroblasts. Am J Pathol 1993;142: 463-470. Drago J, Nurcombe V, Pearse MJ, Murphy M, Bartlett PF. Basic fibroblast growth factor upregulates steady-state levels of laminin B1 and B2 chain mRNA in cultured neuroepithelial cells. Exp Cell Res 1991; 196:246-254. Hurley MM, Abreu C, Harrison JR, Lichtler AC, Raisz LG, Kream BE. Basic fibroblast growth factor inhibits type I collagen gene expression in osteoblastic MC3T3-E1 cells. J Biol Chem 1993; 268:5588-5593. Majors A, Erhart LA. Basic fibroblast growth factor in the extracellular matrix suppresses collagen synthesis and type III procollagen mRNA levels in arterial smooth muscle cell cultures. Arteriosclerosis Thrombosis 1993; 13:680-686. Reilly TM, Taylor DS, Herblin WF, Thoolen MJ, Chiu AT, Watson DW, Timmermans PBMWM. Monaclonal antibodies against basic fibroblast growth factor which inhibit its biological activity in vitro and in viva. Biochem Biophys Res Commun 1989;164:736743. Denizot F, Lang R. Rapid calorimetric assay for cell growth and survival. Modifications to the tetrazolium dye procedure giving improved sensitivity and reliability. J Immunol Methods 1986;89:271-277. Chomczynski P, Sacchi N. Single-step method of RNA isolation by acid guanidium thiocyanate-phenol extraction. Anal Biochem 1987; 162:156-159. Prats H, Kaghad M, Prats AC, Klagsbrun M, Lelias JM, Liauzun P, Chalon P, Tauber JP, Amalric F, Smith JA, Caput D. Highmolecular mass forms of basic fibroblast growth factor are initiated by alternative CUG codons. Prec Natl Acad Sci USA 1989; 86:1836-1840. Johnson DE, Lee P, Williams LT. Diverse forms of a receptor for acidic and basic fibroblast growth factors. Mol Cell Biol 1990; 10:4728-4736. Houssaint E, Blanquet PR, Champion-Arnaud P, Gesnel MC, Torriglia A, Courtois Y, Breathnach R. Related fibroblast growth factor receptor genes exist in the human genome. Proc Natl Acad Sci USA 1990;87:8180-8184. Betsholtz C, Johnsson A, Heldin CH, Westermark B, Lind P, Urdea MS, Eddy R, Shows TB, Philpot K, Mello AL, Knott TJ, Scott J. cDNA sequence and chromosomal localization of human plateletderived growth factor A-chain and its expression in tumour cell lines. Nature 1986;320:695-699. Martyr6 MC, Magdelenat H, Bryckaert MC, Laine-Bidren C, Calvo F. Interferon-y in viva reverses the increased platelet levels of platelet-derived growth factor and transforming growth factor-13 in patients with myelofibrosis with myeloid metaplasia. Br J Haematol 1991; 77:80-86. Ramadori G, Knittel T, Odenthal M, SchwSgler S, Neubauer K,

1996

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

ROSENBAUM ET AL.

Meyer Zum BOschenfelde KH. Synthesis of cellular fibronectin by rat liver fat-storing (Ito) cells: regulation by cytokines. Gastroenterology 1992; 103:1313-1321. Florkiewicz RZ, Sommer A, Human basic fibroblast growth factor gene encodes four polypeptides: three initiate translation from non-AUG codons. Proc Natl Acad Sci USA 1989;86:3978-3981. Shi E, Kan M, Xu J, Wang F, Hou J, McKeehan WL. Control of fibroblast growth factor receptor kinase signal transduction by heterodimerization of combinatorial splice variants. Mol Cell Biol 1993; 13:3907-3918. Roskams T, Rosenbaum J, David G, Desmet V. Integral membrane and matrix heparan sulfate proteoglycan expression in human Ito cells in culture (abstr). Hepatology 1994;20:391A. Weiss RH, Maduri M. The mitogenic effect ofthrombin in vascular smooth muscle cells is largely due to basic fibroblast growth factor. J Biol Chem 1993;268:5724-5727. Herbert JM, Dupuy E, Laplace MC, Zini JM, Bar Shavit R, Tobelem G. Thrombin induces endothelial cell growth via both a proteolytic and a non-proteolytic pathway. Biochem J 1994; 303:227-231. Cozzolino F, Torcia M, Lucibello M, Morbidelli L, Ziche M, Platt J, Fabiani S, Brett J, Stern D. Interferon-(z and interleukin 2 synergistically enhance basic fibroblast growth factor synthesis and induce release, promoting endothelial cell growth. J Clin Invest 1993; 91:2504-2512. Kikuchi K, Yamakage A, Smith EA, LeRoy EC, Trojanowska M. Differential modulation of bFGF receptors by TGF-I3 in adult skin, scleroderma skin, and newborn foreskin fibroblasts. J Invest Dermatol 1992;99:201-205. Goldsmith KT, Gammon RB, Garver Rh Modulation of bFGF in lung fibroblasts by TGF-~ and PDGF. Am J Physiol 1991; 261:L378-L385. Sato Y, Hamanaka R, Ono J, Kuwano M, Rifkin DB, Takaki R. The stimulatory effect of PDGFon vascular smooth muscle migration is mediated by the induction of endogenous basic FGF. Biochem Biophys Res Commun 1991;174:1260-1266. SchSIImann C, Grugel R, Tatje D, Hoppe J, Folkman J, Marm6 D,

GASTROENTEROLOGY VoI. IO9, No. 6

46.

47.

48.

49.

50.

Welch HA. Basic fibroblast growth factor modulates the mitogenic potency of the platelet-derived growth factor (PDGF) isoforms by specific upregulation of the PDGF (z receptor in vascular smooth muscle cells. J Biol Chem 1992;267:18032-18039. Rogers SL, Lynn Cutts J, Gegick PJ, McGuire PG, Rosenberger C, Krisinski S. Transforming growth factor-~l differentially regulates proliferation, morphology, and extracellular matrix expression by three neural crest-derived neuroblastoma cell lines. Exp Cell Res 1994;211:252-262. Franz~n P, Ichijo H, Miyazono K. Different signals mediate transforming growth factor-J31-induced growth inhibition and extracellular matrix production in prostatic carcinoma cells, Exp Cell Res 1993;207:1-7, Stampfer MR, Yaswen P, Alhadeff M, Hosoda J. TGF~ induction of extracellutar matrix associated proteins in normal and transformed human mammary epithelial cells in culture is independent of growth effects. J Cell Physiol 1993;155:210-221. Chen RH, Ebner R, Derynck R. Inactivation of the type II receptor reveals two receptor pathways for the diverse TGF-~ activities. Science 1993; 260:1335-1338. Derynck R. TGF-~-receptor--mediated signaling. Trends Biol Sci 1994; 19:548-553.

Received May 31, 1995. Accepted August 14, 1995. Address requests for reprints to: Jean Rosenbaum, M.D., Groupe de Recherches pour I'Etude du Foie, Universit6 de Bordeaux II, 146 rue L6o Saignat, 33076 Bordeaux, France. Fax: (33) 56-51-40-77. Supported in part by a grant from Institut Electricit6 Sant6. Dr. Rosenbaum's and Dr. Blazejewski's present affiliation is: Groupe de Recherches pour I'Etude du Foie, Universit6 de Bordeaux II, Bordeaux, France. The authors thank D. Cherqui for providing the surgical liver tissue samples, T. Reilly for supplying the anti-fibroblast growth factor 2 antibodies, and H. Prats, L. T. Williams, R. Breathnach, K. Alitalo, C. H. Heldin, and C. Melo for providing the complementary DNAs.