Membrane-anchored Heparin-binding Epidermal Growth Factor-like Growth Factor Acts as a Tumor Survival Factor in a Hepatoma Cell Line*

Heparin-binding epidermal growth factor (EGF)-like growth factor (HB-EGF), which belongs to the EGF family, is produced as a membrane-anchored form (pro-HB-EGF) and later processed to a soluble form (sHB-EGF). It is known that high expression of pro-HB-EGF occurs in hepatoma tissues, although its biological meaning remains unknown. We established two types of hepatoma cell lines (AH66tc), which stably produce pro-HB-EGF and sHB-EGF, respectively. While sHB-EGF-producing cells (sHB-AH) showed rapid growth, pro-HB-EGF-producing cells (pHB-AH) showed markedly suppressed cell growth as compared with the parental cells. Transforming growth factor β or serum-starved conditions induced apoptosis of mock and sHB-AH as well as the parental cells, but not of pHB-AH. The resistance to apoptosis upon serum-starved treatment was correlated with an increase in the rate of the G1 phase in the cell cycle due to up-regulation of the cyclin-dependent kinase inhibitor p21. The mechanism underlying this resistance of pHB-AH to apoptosis was thought to be related to the prolonged half-life of the EGF receptor followed by continuous phosphorylation of the tyrosine residues. These observations demonstrate a unique function of pro-HB-EGF that is not observed for the mature form and show that pro-HB-EGF may act as a tumor survival factor in hepatoma cells.

Heparin-binding EGF 1 -like growth factor (HB-EGF), which belongs to the EGF family, is synthesized as a pre-pro-form of 208 amino acids in length and is expressed as a pro-form integrated into the plasma membrane and then processed to a soluble form of 76 -87 amino acid residues through proteolytic cleavage (1,2). It was originally identified in the conditioned medium of a human histiotic lymphoma cell line, U937 (1), and gene or protein expression of HB-EGF was detected in various rat tissues (3), smooth muscle cells and macrophages of human artheriosclerotic plaques (4), and parietal cells of fundic glands and gastrin cells of pyloric glands in the human gastric mucosa (5). It stimulates the growth of a variety of cells in an autocrine or paracrine manner. Membrane-anchored HB-EGF (pro-HB-EGF), which is also known as a diphtheria toxin receptor (6), stimulates adjacent cell growth through cell-cell contact, i.e. in what is designated as a juxtacrine manner (7), and has been thought to contribute to cell growth like the soluble HB-EGF (sHB-EGF). High expression of HB-EGF was observed in human hepatocellular carcinoma tissues, but not in metastatic liver cancers or normal hepatocytes, suggesting that it is involved in the development or progression of hepatomas in an autocrine manner (8). During hepatocarcinogenesis in a rodent model, its expression was dramatically induced at the cancer (9). However, the polyclonal antibody used in those studies specifically recognizes the pro-form of HB-EGF, not the soluble form. There is no evidence that pro-HB-EGF is cleaved into the soluble form in hepatoma tissues. Even if the pro-form is processed to the soluble form, the mechanism remains unknown.
Both pro-HB-EGF and sHB-EGF bind to the EGF receptor (EGF-R) like other EGF family members. Signal transduction occurs after clustering of the receptors followed by phosphorylation of the tyrosine residues, which results in the triggering of cell proliferation (10). There are some reports on the clinical significance of EGF-R in breast (11) or ovarian cancer (12). These authors reported that different regulatory mechanisms for EGF-R expression exist in tumor cells compared with normal cells and that the enhanced EGF-R expression was associated with aggressive behavior of tumors. In contrast, the growth of some tumor cells with high levels of EGF-R was found, paradoxically, to be inhibited upon exposure to exogenous EGF even if phosphorylation occurred (13,14). Although both forms of HB-EGF bind to EGF-R, nothing has been reported about differences in the signal transduction following binding to EGF-R including its tyrosine phosphorylation.
Apoptosis is regulated through complicated mechanisms, and rescue from apoptosis is related to the immortality of certain cells followed by malignant transformation (15). Many growth factors and cytokines have been reported to be involved in apoptosis in a positive or negative manner. Some hepatoma cells undergo apoptosis on treatment with transforming growth factor ␤ (TGF-␤) or anti-cancer drugs (16,17). These reports suggested that TGF-␤ regulated the proliferation of hepatoma cells as well as hepatocytes during liver regeneration (18), sometimes involving apoptosis. Prevention from apoptosis can be related to tumor development, including promotion of neo-* This work was supported in part by Grants-in-Aid for Cancer Research and Scientific Research on Priority Areas from the Ministry of Education, Science and Culture of Japan (to S. H. and N. T.) and a grant from the Sagawa Cancer Research Foundation (to S. H.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ The first two authors contributed equally to this work.
ʈ To whom all correspondence should be addressed. Tel.: 81-6-879-3420; Fax: 81-6-879-3429. 1 The abbreviations used are: EGF, epidermal growth factor; HB-EGF, heparin-binding EGF-like growth factor; EGF-R, EGF receptor; TGF-␤, transforming growth factor ␤; sHB-EGF, a soluble form of HB-EGF; pro-HB-EGF, a membrane-anchored form of HB-EGF; sHB-AH, sHB-EGF-producing AH66tc cell(s); pHB-AH, pro-HB-EGF-producing AH66tc cell(s); FCS, fetal calf serum; PBS, phosphate-buffered saline; TUNEL, terminal deoxytransferase-mediated dUTP-biotin nick end labeling. plastic progression, tumor growth, and resistance to cytotoxic anti-cancer agents (19). Some oncogenes, and tumor suppressor genes, such as bcl2 and p53, are involved in the prevention of apoptosis. While HB-EGF has been thought to be linked to hepatocarcinogenesis, there is no direct evidence of this. In this study, we transfected two forms of the HB-EGF gene, to produce pro-and soluble forms, into a rat hepatoma cell line and investigated their biological functions with emphasis on the relationships to apoptosis. We found that sHB-EGF stimulated the growth of hepatomas and pro-HB-EGF suppressed it through up-regulation of EGF-R and that hepatomas expressing high levels of pro-HB-EGF were resistant to apoptosis induced under several conditions. Possible mechanisms for these functions of pro-HB-EGF are proposed.

MATERIALS AND METHODS
Cell Culture-The rat hepatoma cell line, AH66tc (20), was donated by the Japanese Cancer Research Resources Bank and cultured in RPMI 1640 medium (Nikken BioMedical Laboratory, Kyoto, Japan) containing 100 g/ml of kanamycin (Sigma), 50 units/ml of penicillin (Banyu Corp., Tokyo, Japan) and 10% fetal calf serum (FCS). Cell numbers were determined on a hemocytometer under a microscope after harvesting with trypsin.
Establishment of Soluble HB-EGF or pro-HB-EGF Gene-transfected AH66tc Cells-An expression vector of sHB-EGF was constructed as follows. A point mutation to yield a stop codon (P 148 -stop) was introduced into the transmembrane domain of mouse HB-EGF cDNA (21) by site-directed mutagenesis with an in vitro mutagenesis kit (Takara Shuzo, Kyoto, Japan), using 5Ј-GATTCTCCACTTATAGAGTCAG-3Ј as a primer, and the mutated HB-EGF cDNA was inserted into pRc/CMV (Invitrogen). This method has been described in detail elsewhere (22). The inserted sHB-EGF gene was directly sequenced, and then the site of mutation was confirmed. An expression vector of pro-HB-EGF was constructed by inserting the full sequence of mouse HB-EGF cDNA into pRc/CMV. Each vector was transfected into AH66tc cells by the electroporation method, and selection was performed as to G418 (Sigma) resistance. Ten clones each of pro-HB-EGF cDNA-, sHB-EGF cDNA-, and vector alone-transfected AH66tc cells (mock) were obtained, and three clones were independently selected in each case.
Northern Blot Analysis-Total RNA was extracted from AH66tc transfectants according to the method reported by Chomczynski and Sacchi (23). 20 g of RNA was electrophoresed on a 1% agarose gel containing 2.2 M formaldehyde and then transferred onto a Zeta-probe membrane (Bio-Rad) by capillary action (24). After hybridization with 32 P-labeled mouse HB-EGF cDNA (21) at 42°C in the hybridization buffer (24), the membrane was washed twice for 30 min at 55°C with 30 mM sodium citrate, 300 mM NaCl, and 0.1% sodium dodecyl sulfate and once for 30 min at 55°C with 3 mM sodium citrate, 30 mM NaCl, and 0.1% sodium dodecyl sulfate and then exposed to x-ray film (Eastman Kodak Co.) with an intensifying screen for 2 days. The probe used for p21 mRNA analysis comprised about 300 bases of rat p21 cDNA synthesized by polymerase chain reaction using the sense primer CTCT-TCGGTCCCGTGGACAGT and the antisense primer CACCAGAGTG-CAAGACAGCGA, which were based on human and mouse p21 cDNA, respectively (25). The probes used for other p27 and p15 mRNA analyses were kindly provided by Dr. M. Nakanishi (Department of Geriatric Research, National Institute for Longevity Sciences). Northern blot analysis of p21, p27, and p15 was performed with the same procedure of HB-EGF mRNA detection.
Measurement of Soluble HB-EGF and Pro-HB-EGF-The biological activity of sHB-EGF and pro-HB-EGF was assayed as reported previously (7). For assay of the sHB-EGF activity, 25 l of condition medium derived from each AH66tc transfectant in a confluent state was added to EP170.7 cells, which are mouse leukemia cells whose growth depends on the EGF-R ligand, at 2 ϫ 10 4 cells/well in a volume of 200 l in 96-well plates. After 36 h of incubation, 1 Ci of [ 3 H]thymidine was added to each well. After a 4-h incubation with [ 3 H]thymidine, incorporation into DNA of EP170.7 cells was measured with a ␤-plate system (Pharmacia Biotech Inc.). To measure the juxtacrine activity of pro-HB-EGF, 2 ϫ 10 5 cells of each AH66tc transfectant were fixed with 5% buffered formaldehyde in 24-well plates and then overlaid with 1 ϫ 10 5 EP170.7 cells for 48 h. After a 4-h incubation with [ 3 H]thymidine, DNA synthesis by EP170.7 cells collected from the medium was measured by the same protocol as reported previously (7).
Western Blot Analysis on EGF-R and Its Phosphorylation-Each transfectant was lysed with Nonidet P-40 buffer (10 mM Tris-HCl, pH 7.8, 1% Nonidet P-40, 0.15 M NaCl, 10 g/ml aprotinin, and 10 M p-amidinophenyl methanesulfonyl fluoride hydrochloride) for 30 min. After centrifugation at 15,000 ϫ g for 10 min at 4°C, 10 g of proteins in the supernatant were subjected to 6% SDS-polyacrylamide gel electrophoresis and then transferred onto a nitrocellulose membrane using TE70 and TE77 Semiphor semidry transfer units (Hoefar Scientific Instruments, San Francisco, CA). After blocking in 3% bovine serum albumin/PBS overnight, the membrane was incubated with a 1:1000 dilution of anti-EGF-R sheep IgG. Upstate Biotechnology, Inc., Lake Placid, NY) as the first antibody for 4 h at room temperature. After washing with Tris-buffered saline containing 0.5% Tween 20 for 30 min, it was incubated with a 1:2000 dilution of horseradish peroxidaseconjugated rabbit anti-sheep IgG (DAKO, Kyoto, Japan) as the second antibody for 1 h. The filter was then washed with Tris-buffered saline containing 0.5% Tween 20 for 30 min and developed with a Western blot detection kit, ECL (Amersham Corp.). For the detection of phosphorylated tyrosine residues, AH66tc transfectants under several conditions were lysed with Nonidet P-40 buffer containing a phosphatase inhibitor (400 M sodium orthovanadate, 10 mM pyrophosphate, and 10 mM iodoacetate) for 30 min at 4°C, and PY20, a mouse monoclonal antiphosphotyrosine (Transduction Laboratories, KY), and peroxidase-conjugated anti-mouse IgG (DAKO, Kyoto, Japan) were used as the first and second antibodies, respectively. Western blotting was performed according to the method reported by Fan et al. (26) with a small modification. Each band was measured by a densitometer. Stability of Cell Surface EGF-R-3 ϫ 10 5 cells of sHB-AH and pHB-AH were plated on 6-cm dishes and then incubated for 12 h. Cell surface proteins on the cells were biotinylated with 0.1 mg/ml of sulfo-NHS-biotin (Pierce) for 20 min on ice as described previously (27) and then cultured in 6-cm dishes with the medium in which each transfectant was cultured. The cells were lysed with Nonidet P-40 buffer after 0, 1, 2, and 4 h. The cell lysates obtained with the Nonidet P-40 buffer described above were subjected to immunoprecipitation using anti-EGF-R. After SDS gel electrophoresis followed by blotting onto a nitrocellulose membrane, the filter was incubated with avidin-peroxidase conjugates, an ABC kit (VECTOR Laboratories Inc.), and developed with an ECL kit (Amersham).
Effects of Immobilized HB-EGF on AH66tc Cells-250 ng of human recombinant HB-EGF (28) dissolved in 250 l of 0.2 M borate buffer, pH 9.6, was added to each well (16-mm diameter) of a 24-well plate (125 l/cm 2 of 0.01% HB-EGF, 0.2 M borate buffer, pH 9.5). The plate was incubated for 24 h at 4°C. The HB-EGF solution was subjected to measurement of the immobilization efficiency of HB-EGF. The immobilization efficiency of HB-EGF was calculated as the average of quadruplicate measurements of soluble HB-EGF activity in each of four independent wells. As judged on this measurement, 50 ng of soluble HB-EGF was immobilized. The wells were then washed five times with PBS. 5 ϫ 10 3 parental AH66tc cells were plated with 500 l of DMEM, 2% FCS in each well. For estimation of the soluble HB-EGF activity of AH66tc cells, cells were plated with 500 l of DMEM, 2% FCS containing an appropriate amount of HB-EGF in wells pretreated with 0.2 M borate buffer, pH 9.6, alone. The cells were incubated for 3 days at 37°C, trypsinized, and then counted with a Coulter counter (Coulter Corp., Hialeah, FL). The level of tyrosine phosphorylation on EGF-R was determined as follows: 6-cm dishes were pretreated with 125 l/cm 2 of 0.2 M borate buffer, pH 9.5, with or without 0.01% HB-EGF for 24 h at 4°C. 1 ϫ 10 6 AH66tc cells harvested with PBS, 0.5 mM EDTA were plated with 1 ml of DMEM, 0.1% bovine serum albumin or 1 ml of DMEM, 0.1% bovine serum albumin, 5 ng of HB-EGF on an HB-EGFimmobilized or nonimmobilized dish, respectively. After 12.5 min of incubation at 37°C, the cells were immediately harvested by pipetting and chilled in ice water. The chilled cells were immunoprecipitated with anti-EGF-R IgG, followed by Western blotting for detection of phosphotyrosine residues on EGF-R as described above.
Analysis of Cell Cycle Distribution-AH66tc transfectants in a subconfluent state and a confluent state were harvested with PBS containing 0.2% EDTA and fixed with 70% ethanol. The cells were treated with 1 mg/ml of RNase (Sigma) for 20 min and then stained with 100 g/ml of propidium iodide (Sigma). After filtration of the cells through 50 -70-m pore size nylon meshes, cell cycle distribution was analyzed with a FACScan flow cytometer (Becton Dickinson, San Jose, CA).
Apoptosis on Treatment with TGF-␤ and Serum-starved Conditions-TGF-␤ treatment was as follows. 1 ϫ 10 5 AH66tc cells (mock) were plated in six-well plates with RPMI medium containing 10% FCS. After a 12-h incubation, the medium was changed to fresh medium, and the cells were incubated with 5 ng/ml TGF-␤ for the indicated times. sHB-AH and pHB-AH were treated with 5 ng/ml of TGF-␤ for 24 h. The DNA of each AH66tc transfectant was purified by proteinase K treatment, followed with phenol/chloroform. The DNAs were electrophoresed on a 0.9% agarose gel containing ethidium bromide. Apoptosis in situ was detected by the terminal deoxytransferase-mediated dUTP-biotin nick end labeling (TUNEL) method. Briefly, sHB-AH and pHB-AH were cultured in an eight-well chamber slide (Lab-Tek, Nunc). After 5 ng/ml of TGF-␤ treatment for 24 h, the cells were fixed with 3% paraformaldehyde for 20 min. The slide was rinsed three times with distilled water, submerged in terminal deoxytransferase buffer (Takara Shuzo, Shiga, Japan), and incubated in biotinated dUTP and terminal deoxytransferase at 37°C for 90 min. Condensed DNA was visualized with a fluorescein isothiocyanate-labeled secondary antibody and was photographed through a fluorescence microscope. Detailed procedures were given in the manufacturer's protocol of the in situ apoptosis detection kit (Takara Shuzo). Serum starvation was as follows. AH66tc transfectants were grown in medium containing 10% FCS until a confluent state was reached, and then the medium was changed to serumfree medium. Photographs were taken at 1, 7, and 14 days after the change to serum-free medium.

Establishment and Characterization of Two Types of HB-EGF Gene-transfected AH66tc
Cells-Expression of HB-EGF mRNA was detected in neither the parental AH66tc cells nor mock transfectants that had been transfected with pRc/CMV alone. In contrast, 3 clones of sHB-AH (sHB-AH1, -2, and -3) and 3 clones of pHB-AH (pHB-AH1, -2, and -3), which were randomly selected from among 10 clones, showed the expression levels of HB-EGF mRNA in the following orders: sHB-AH3 Ͼ sHB-AH1 Ͼ sHB-AH2, and pHB-AH2 Ͼ pHB-AH3 Ͼ pHB-AH1 (Fig. 1). To determine their functional expression, the soluble or juxtacrine growth factor activity of each clone was measured (Fig. 2). Although the conditioned medium produced by sHB-AH stimulated DNA synthesis of EP170.7 cells, such an effect was not seen in the conditioned medium of pHB-AH or mock ( Fig. 2A), indicating that a significant amount of sHB-EGF was secreted by sHB-AH but not detected in the conditioned medium of pHB-AH or mock cells. In contrast, pHB-AH stimulated DNA synthesis of EP170.7 cells in a juxtacrine manner (7), but sHB-AH and mock did not (Fig. 2B), indicating that pHB-AH expressed the functional pro-HB-EGF on their cell surface.
Cell Growth of HB-EGF Gene-transfected AH66tc-sHB-AH showed dramatically a rapid growth compared with the other two groups, pHB-AH and mock cells (Fig. 3), suggesting that HB-EGF stimulated the growth of AH66tc cells. Surprisingly, each type of pHB-AH showed much slower growth than mock cells. After 72 h of observation, the difference was prominent. Since pHB-AH also exhibit a juxtacrine activity, the mechanism of growth suppression of pHB-AH must be studied further.
Expression of EGF-R and Its Phosphorylation-HB-EGF is known to bind with EGF-R, followed by signal transduction.
The suppression of cell growth by pro-HB-EGF described above suggested that some modification occurred in EGF-R. To investigate this, we examined the expression of EGF-R by Western blotting (Fig. 4A). Protein expression of EGF-R was about 2-3 times increased in pHB-AH in comparison with the expression level in sHB-AH and mock. The data were almost the same under 70% confluent and 100% confluent conditions (data not shown). After EGF-R binds its ligands, signal transduction occurs with clustering of the receptor, followed by phosphorylation of the tyrosine residues, and then the receptor is internalized into the cytoplasm (10). The levels of phosphotyrosine residues on EGF-R were equally high in both sHB-AH and pHB-AH but low in mock (Fig. 4B). Since the protein level of total EGF-R was higher in pHB-AH than in sHB-AH, the net phosphorylation on EGF-R in sHB-AH and pHB-AH was almost the same. When exogenous HB-EGF was added to mock, a 6-fold increase in phosphorylation on EGF-R as compared with the untreated control was observed. In the case of sHB-AH and pHB-AH, this increase was 0.97 and 1.60-fold, respec- After 48 h, DNA synthesis by EP170.7 cells was assayed as described under "Materials and Methods." Control, addition of medium alone. Data represent means Ϯ S.D. for three experiments. B, formalin-fixed AH66tc cells were cocultured with EP170.7 cells for 48 h, and then DNA synthesis by EP170.7 cells was assayed as described under "Materials and Methods." The details of the procedure were as reported previously (7). Data represent means for three experiments, and the bars indicate S.D.
tively. The effects of exogenous HB-EGF were not prominent in these cells, because EGF-R was already phosphorylated through the autocrine mechanisms of HB-EGF or the juxtacrine mechanisms of pro-HB-EGF produced by themselves. When phosphorylation on EGF-R was reexamined at 30 min after changing to fresh medium, the level of phosphorylation was markedly decreased in both sHB-AH and pHB-AH but not so changed in mock. The levels of decreases were significantly higher in sHB-AH than pHB-AH (80 -90% versus 60 -70%). These results suggested that EGF-R on sHB-AH was phosphorylated by HB-EGF in an autocrine manner and that phosphorylation of EGF-R on pHB-AH was decreased by unidentified mechanisms. This decrease may be caused by a cleavage of pro-HB-EGF due to a stress of changing the medium to a fresh one, which is similar to our previous observations (27). Although the signal via EGF-R was transmitted in pHB-AH, cell growth was suppressed. This inhibitory effect of pro-HB-EGF is controversial and may be involved with the increased EGF-R.
To determine the details of the mechanism for up-regulation of EGF-R, the stability of EGF-R was investigated. Since EGF-R was internalized after ligand binding within a minute or so, levels of biotinylated EGF-R in sHB-AH were gradually consumed in an autocrine manner and undetectable at 4 h (Fig.  4C). In contrast, the level of biotinylated EGF-R in pHB-AH did not change for more than 4 h, suggesting that internalization of EGF-R followed by binding with pro-HB-EGF was extremely delayed.
Growth Suppression in the Pro-HB-EGF Model-Although transfection of the pro-HB-EGF gene suppressed the growth of AH66tc cells with up-regulation of EGF-R, secondary effects including up-or down-regulation of other molecules such as adhesion molecules must be ruled out. To show the direct effect of pro-HB-EGF, we immobilized HB-EGF on a plate and then observed the growth of the parental AH66tc cells (Fig. 5). Although the addition of exogenous HB-EGF stimulated the growth of AH66tc cells, immobilized HB-EGF slightly suppressed it (p Ͻ 0.03) compared with control. Tyrosine phosphorylation of EGF-R was observed upon both the addition of soluble HB-EGF and the addition of immobilized HB-EGF, but it was not observed in a control. These observations suggest that growth suppression of pHB-AH is a direct effect of pro-HB-EGF itself. The same phenomenon was observed for human hepatoma cell lines HepG2 and Hep3B (data not shown).
Apoptosis Induced by Several Conditions-While HB-EGF has been thought to promote cell growth of hepatomas, specific roles of pro-HB-EGF in vivo have been suggested. Each AH66tc transfectant was treated with TGF-␤ or under serum-starved conditions. Treatment with TGF-␤ for 24 -48 h induced apoptosis with a DNA ladder formation of AH66tc cells (Fig. 6A).
This was observed for sHB-AH but was completely inhibited in pHB-AH (Fig. 6B). While the DNA ladder formation seemed unclear in sHB-AH treated with TGF-␤, some apoptotic cells were observed in sHB-AH by a TUNEL method, but not in pHB-AH (Fig. 6C). When these cells were treated with an anti-cancer drug, adriamycin or cisplatin, pHB-AH were also more resistant than the other two types (data not shown). Furthermore, when three types of AH66tc cells were cultured under serum-starved conditions for a long time, many mocks and sHB-AH were dead after 1 week, but approximately 80% of the pHB-AH were still alive after 2 weeks (Fig. 7). When pHB-AH were cultured in a serum-starved medium containing 5 ng/ml of exogenous HB-EGF, the number of surviving cells was decreased after 1 week, and the expression of EGF-R was slightly decreased (data not shown). As mentioned above, overexpression of pro-HB-EGF enhanced the resistance against any  4. Western blot analysis of EGF-R and tyrosine phosphorylation. A, 10 g of cell lysates from AH66tc transfectants was applied to a 6% SDS-acrylamide gel and then analyzed by Western blotting. B, cellular proteins were extracted from AH66tc transfectants with Nonidet P-40 buffer containing a phosphatase inhibitor. 10 g of proteins was electrophoresed on a 6% SDS acrylamide gel, and then Western blot analysis was performed using anti-phosphotyrosine. Lanes 1, 4, and 7 indicate untreated controls; lanes 2, 5, and 8 indicate samples of cells at 30 min after changing the medium to fresh one; and lanes 3, 6, and 9 indicate samples of cells treated with 5 ng/ml of exogenous HB-EGF for 15 min. C, cell surface proteins from 3 ϫ 10 5 cultured AH66tc cells were biotinylated, and then the cells were cultured in the medium in which they had been cultured previously. Cell lysates were immunoprecipitated with anti EGF-R, and then Western blotting was performed. The bands of biotinylated EGF-R were visualized with a biotin-avidin system and ECL development. The amounts of pHB-AH samples were half of sHB-AH, because basal expression levels were different as shown in Fig. 4A. All procedures are described in detail under "Materials and Methods." We presented the results for mock 1, sHB-AH3, and pHB-AH3 cells. The results for other clones were almost identical to those described here. stress that we tested here.
Cell Cycle and Expression of p21 mRNA-While EGF-R has a stimulatory effect on the proliferation of a wide variety of normal and cancer cells (10), the addition of EGF to tumor cells with high levels of EGF-R causes growth inhibition through prolonged induction of the cyclin-dependent kinase inhibitor, p21 (WAF1) (29). Overexpression of pro-HB-EGF induced both increases of EGF-R and suppression of cell growth. To determine whether or not the mechanism of growth suppression is linked to p21 induction, the cell cycle distribution and mRNA expression of cyclin-dependent kinase inhibitors p21, p27, and p15 was investigated among three types of AH66tc cells. When they were in a sparse state, there were no differences in the cell cycle distribution among the three groups of cells. In a confluent state, however, the G 1 :G 0 ratio in pHB-AH was extremely increased compared with in mock and sHB-AH (Fig. 8A). At this time, expression of p21 mRNA was a little increased under confluent conditions in both mock and sHB-AH and dramatically increased in pHB-AH (Fig. 8, B and C), but expression of p27 and p15 mRNAs was not changed in the confluency of each transfectant despite different expression levels of p15 mRNA among each clone. DISCUSSION The present study demonstrates that ectopically expressed pro-HB-EGF has a specific role in the EGF-R-mediated signal transduction pathway for negative cell proliferation, which is completely the reverse effect of sHB-EGF. Pro-HB-EGF is expressed in various tissues that are not thought to be involved in cell proliferation, such as heart and vascular endothelial cells, suggesting that its original function in vivo is not always promotion of cell growth. Since pro-HB-EGF binds to EGF-R as well as sHB-EGF, the difference in their biological effects described here is thought to be due to how their signals after binding to the receptor are sent to an intracellular system. While both forms of HB-EGF induced the phosphorylation of EGF-R (Fig. 4B), its phosphorylation state may be prolonged in the case of pHB-AH, since the half-life of EGF-R in pHB-AH was longer than that of sHB-AH.
pHB-AH showed a marked increase of EGF-R protein compared with sHB-AH and mock cells (Fig. 4A), which depended on prolongation of the half-life of EGF-R (Fig. 4D). When the EGF-R ligands bind to EGF-R, internalization of the receptor followed by signal transduction occurs (10). In the case of pro-HB-EGF, it was thought that this internalization was delayed or inhibited and that the phosphorylation of EGF-R was continuous. Prolonged stimulation to EGF-R by pro-HB-EGF A, extracts from AH66tc cells (mock 1) treated with TGF-␤ for the indicated times were analyzed for DNA fragmentation by electrophoresis on a 0.9% agarose gel containing ethidium bromide. B, sHB-AH3 and pHB-AH3 were treated with 5 ng/ml of TGF-␤ for 48 h. DNAs extracted from these cells were subjected to electrophoresis. PstI-digested DNA was used as a molecular weight marker (lane M). C, in situ apoptosis was examined by the TUNEL method. sHB-AH and pHB-AH were treated with 5 ng/ml of TGF-␤ for 24 h, and apoptotic cells were visualized by a fluorescein isothiocyanate-labeled secondary antibody. All procedures are described in detail under "Materials and Methods." We presented the results for mock 1, sHB-AH3, and pHB-AH3 cells. The results for other clones were almost the same as those described here.
up-regulated the expression of p21 mRNA but not of other cyclin-dependent kinase inhibitors, p15 and p27 mRNAs (Fig.  8). Recently, cell growth arrest and induction of p21 in tumor cells expressing high amounts of EGF-R have been reported to be mediated by signal transducers and activators of transcription 1 (STAT1) (30). Whether STAT1 is induced by pro-HB-EGF has not yet been investigated.
Membrane-anchored forms of growth factors, which mostly belong to the EGF family, show the same growth factor activity as their soluble forms (31). The activity is not as strong as that of their soluble forms, and cell-cell contact is necessary for their signal transduction (27). The specific activity of membraneanchored molecule is found in the c-Kit ligand, which is essential for hematopoietic cell proliferation and differentiation (32). Recently, Grell et al. (33) reported that membrane-anchored tumor necrosis factor was the prime activating ligand of the 80-kDa tumor necrosis factor receptor, which is the minor form of the receptor. These reports suggested the essential roles of membrane-anchored growth factors. The reasons for the different signal transduction between two forms of the c-Kit ligand are the prolonged activation and longer life span of the c-Kit protein (34). The mechanism of growth suppression by pro-HB-EGF is similar to the phenomenon in the case of the c-Kit ligand. An immobilized anti-Kit monoclonal antibody behaves like a membrane-anchored form of the c-Kit ligand rather than its soluble form (35). With a similar system, immobilized HB-EGF also inhibited the growth of AH66tc cells (Fig. 5).
What is the biological significance of overexpression of pro-HB-EGF in hepatomas? Hepatomas at the earliest stage do not always need rapid growth, but it is required for escape from various immune systems. Resistance against several factors may play a role in the early progression of hepatomas. A hepatoma overexpressing pro-HB-EGF showed strong resistance to several factors. Although the mechanism underlying resistance to TGF-␤-induced apoptosis remains unknown, the resistance to serum-starved treatment is thought to be due to G 1 arrest induced by up-regulation of p21 (Fig. 8). The same phenomenon of resistance to apoptosis was observed in myoblast differentiation with p21 induction (36). In some cases, cells under G 1 arrest undergo apoptosis (15). However, pHB-AH  1, 4, and 7), confluent conditions (lanes 2, 5, and 8), and 48 h after the attainment of confluent conditions (lanes 3, 6, and 9) were analyzed by Northern blot hybridization. Details were given under "Materials and Methods." A gel stained with ethidium bromide is shown for comparable amounts of RNAs. C, the density of each p21 mRNA band was measured with a densitometer. We present the results for mock 1, sHB-AH3, and pHB-AH3 cells. The results for other clones were almost identical to those described here.

FIG. 7.
Morphological changes in AH66tc transfectants cultured under serum-starved conditions. After the three types of AH66tc cells had been grown to a confluent state, the medium was changed to new medium containing no FCS. Photographs were taken at 1, 7, and 14 days after the medium change (original magnification, ϫ 100). We describe the results for mock 1, sHB-AH3, and pHB-AH3 cells. The results for other clones were almost identical to those described here.
showed slow growth with a high percentage of the G 1 phase in the cell cycle. Greenberg et al. (37) reported opposing effects of extracellular signal-regulated kinase and c-Jun N-terminal kinase mitogen-activated protein kinase on apoptosis, using rat pheochromocytoma cells, PC-12. Whereas signaling via EGF-R is related to activation of extracellular signal-regulated kinase or c-Jun N-terminal kinase, there were no differences of extracellular signal-regulated kinase and c-Jun N-terminal kinase activities between sHB-AH and pHB-AH (data not shown). When pro-HB-EGF is cleaved by some unidentified proteases, sHB-EGF stimulates the growth of hepatoma cells. A large hepatoma that shows rapid growth cannot be caught only by the immune system. It is very interesting to speculate on whether the unidentified protease is activated in the large hepatoma. High expression of EGF-R was found in a subline of a human hepatoma cell line, and the mechanism underlying up-regulation of EGF-R is dependent on the stability of the receptor on the surface membrane (38). These authors stated that overexpression of EGF-R might contribute to the greater neoplastic potential of the hepatoma. The present study demonstrates that pro-HB-EGF is one of the factors that induces high expression of EGF-R. This effect is a unique function of pro-HB-EGF that the soluble form does not have. It may be related to very early development of hepatomas as a cell survival factor.