Bcl-2 Delays and Alters Hepatic Carcinogenesis Induced by Transforming Growth Factor α¹

Tumor development and progression depend on the disruption of normal tissue homeostatic mechanisms that shift the balance between cell proliferation and cell death toward cell accumulation. There is considerable evidence suggesting that the modulation of the equilibrium between hepatocyte proliferation and cell death is a key event in the development of HCC³(1, 2, 3). However, the molecular basis for HCC development remains undefined, and it is not known whether apoptosis plays a causative role in liver carcinogenesis.

Apoptosis occurs naturally in the liver, albeit at a very low rate(4, 5). Normal hepatocytes are sensitive to apoptosis induced by Fas agonist antibodies and by TNF given in conjunction with a transcriptional inhibitor (6, 7). Both Bcl-2 and Bcl-X_L are expressed in the liver and may repress apoptosis through a common pathway (8). Bcl-2 is normally expressed in bile duct cells but not in hepatocytes; however, its de novo expression, both in cultured cell lines and in transgenic mice, protects these cells against apoptosis (9, 10). The exact role of apoptosis in the development of HCC has not been fully elucidated. In studies of rat liver carcinogenesis,Grasl-Kraupp et al. (11) demonstrated that although the overall rate of hepatocyte proliferation increased during tumor development, so did the rate of apoptosis. This, and other works (3, 12), demonstrated that proliferation and apoptosis are linked and that the ratio between the two, rather than the absolute levels of proliferation and apoptosis, is important in liver tumorigenesis.

The development of HCC has been studied extensively in transgenic mice overexpressing TGF-α either as a single gene (13, 14, 15, 16, 17, 18) or in double transgenic mice that overexpress both TGF-α and c-myc(19, 20). We have examined the pathogenesis of HCC in transgenic mice that overexpress TGF-α under the control of the MT1 promoter. In this model, 75–80% of male mice develop HCC after 12 months of age (13, 14, 15). In these animals, hepatocyte proliferation is 3–4-fold higher than that in nontransgenic mice during the first 2 months of life, and the liver:body weight ratio is elevated. As the animals age, hepatocyte proliferation remains high,but the liver:body weight ratio becomes similar to that of WT animals in which hepatocytes are essentially quiescent. The increase in hepatocyte proliferation in TGF-α transgenic mice appears to be compensated for by the increased rate of hepatocyte turnover that occurs in these animals (16). Thus, liver mass homeostasis is maintained for many months, presumably until a loss of the balance between cell proliferation and apoptosis leads to tumor development.

Because cell proliferation and apoptosis appear to be linked during HCC development, an analysis of the role of apoptosis in hepatocarcinogenesis requires a system in which apoptosis can be regulated independently of cell proliferation. To determine whether prevention of apoptosis would alter tumor development in mouse livers,we generated double transgenic mice that overexpress TGF-α and Bcl-2. Whereas both transgenes are driven by the MT1 promoter, the TGF-αgene construct is “leaky” (that is, it has a relatively high level of expression in the absence of an inducer), whereas Bcl-2 expression depends on induction by Zn or Cd. This feature provided the opportunity to differentially regulate Bcl-2 in animals expressing TGF-α. We postulated that the antiapoptotic effect of Bcl-2 would cause liver enlargement and accelerated tumorigenesis in TGF-α/Bcl-2 double transgenic mice. Instead, we found that tumor development in these animals was delayed and decreased in comparison with TGF-α single transgenic mice or TGF-α/Bcl-2 double transgenic mice in which Bcl-2 expression was not induced. Coexpression of TGF-α and Bcl-2 also had a striking effect on tumor cell proliferation, size, and morphology.

The generation of TGF-α transgenic mice of the MT42 line has been reported previously (13). Heterozygous human Bcl-2 transgenic mice with the Bcl-2 transgene under the control of the mouse MT1 promoter on a C57B6C3H background were provided by Dr. Stanley Korsmeyer (Dana Farber Cancer Institute, Harvard Medical School,Boston, MA 02115). Homozygous TGF-α transgenic mice were crossed with WT CD1 mice to generate TGF-α heterozygotes. Female heterozygous TGF-α transgenic mice were mated to male heterozygous Bcl-2 mice to generate the WT, TGF-α, and Bcl-2 single transgenic and TGF-α/Bcl-2 double transgenic mice, all in a hybrid (CD1 × C57/B6C3H) background. Genotypes were determined by PCR amplification of genomic DNA obtained from mouse tails. To fully induce transgene expression, mice were maintained on 25 mmZnSO₄ drinking water (zinc water) from the time of weaning (3 weeks of age). In addition to the mice that were maintained continuously on zinc water, one group received water without zinc addition for the duration of the study, and another group received zinc water from 8.5 to 12.5 months. Between 25 and 30 F₁ male mice from each genotype were used for the long-term tumor studies. All animals were maintained and cared for in accordance with NIH guidelines for animal care.

Livers were examined, and part of the tissue was fixed either in neutral buffered formalin for 24 h for routine histology, IHC, and ISH or in methacarne (70% methanol, 20% chloroform and 10% acetic acid) for 1 h and transferred to 100% methanol for 24 h for BrdUrd IHC. After fixation, tissue was embedded in paraffin. For routine histological analysis, 5-μm sections were cut from paraffin-embedded blocks and stained with H&E.

For Bcl-2 IHC, sections were deparaffinized, rehydrated, and treated for 30 min with 0.3% H₂O₂to inactivate endogenous peroxidase activity. Sections were microwaved for 10 min in 10 mm sodium citrate buffer (pH 6.0) before incubation in primary antibody (mouse antihuman Bcl-2; clone N-100;Santa Cruz Biotechnology, Santa Cruz, CA) diluted1:500 in PBS containing 5% horse serum. Staining was detected using the Vectastain elite ABC kit (Vector Laboratories, Burlingame, CA). BrdUrd IHC was performed as described above with the following modifications: trypsin digestion was used in place of microwave retrieval, followed by incubation in 2.5 m HCl for 10 min at 37°C. Mouse anti-BrdUrd diluted 1:40 was used as the primary antibody (Dako,Carpinteria, CA).

ISH was performed as described by Alpers et al.(21), except that slides were exposed in the dark for 5 days. A 600-bp ³⁵S-UTP-labeled TGF-α antisense probe was transcribed from the same DNA template used for the Northern blot analysis (see below). A TGF-α sense probe was used as a negative control.

Total liver RNA was isolated from snap-frozen tissue using a RNA purification kit (RNAeasy mini kit; Qiagen, Chatsworth, CA). Total RNA(10 μg) was separated on a 1.1% agarose/formaldehyde gel and transferred to a nylon filter by capillary transfer. Membranes were prehybridized for 1 h in NorthernMax Prehyb/hyb buffer (Ambion,Austin, TX) and then hybridized overnight at 65°C with a[³²P]UTP-labeled TGF-α antisense riboprobe diluted in NorthernMax buffer. After hybridization with the TGF-αriboprobe, the membranes were rehybridized with either a[³²P]UTP-labeled β-actin (Fig. 3) or cyclophilin (Fig. 7) riboprobe to verify loading. The TGF-αriboprobe was synthesized off the T7 promoter from a PCR-generated DNA fragment containing the T7 promoter ligated to a 600-bp TGF-α cDNA fragment (Lig n Scribe; Ambion). The β-actin and cyclophilin riboprobes were synthesized off the T7 promoters of the pTRI-β-actin and pTRI-cyclophilin templates (Ambion).

Total liver protein was isolated from snap-frozen tissue by homogenization in radioimmunoprecipitation assay buffer (1.0% NP40,0.5% sodium deoxycholate, and 0.1% SDS in PBS) containing 1 mm DTT, 0.5 mmp-aminoethylbenzenesulfonyl fluoride, 2 μg/ml aprotinin, 2μg/ml pepstatin, 2 μg/ml leupeptin, and 10 μg/ml soybean trypsin inhibitor. Protein concentrations were determined using the Bradford protein assay (Bio-Rad, Hercules, CA). Total protein (25 μg) was separated on a 10% PAGE gel, transferred to nylon membranes, and blocked in 5% milk containing 0.1% Tween 20 before incubation with antibody (rabbit anti-Bcl-2; clone N19; Santa Cruz Biotechnology). Enhanced chemiluminescence (Santa Cruz Biotechnology) was used for detection.

One month before treatment, young adult Bcl-2 transgenic mice (C57B6C3H background) and WT littermates were placed on zinc water to induce the Bcl-2 transgene. Animals received an i.p. injection of 10 μg of mouse anti-Fas antibody (clone Jo2; PharMingen, San Diego, CA) and were killed 3 h after the injection. Livers were resected, fixed in 10% buffered formalin, and embedded in paraffin. Sections were stained with H&E, and the number of apoptotic nuclei (determined by morphology)per ×400 high-power field was counted.

For proliferation studies, animals received an i.p. implant of Alzet osmotic 3-day pumps (model 1003D; Alza Corp. USA) that released a solution of 15 mg/ml BrdUrd (Boehringer Mannheim, Indianapolis, IN) at a rate of 1.0 μl/h. At the end of day 3, the mice were killed, the livers were removed, and sections were prepared for BrdUrd IHC. BrdUrd incorporation was measured by counting the number of positive nuclei labeled with a mouse monoclonal anti-BrdUrd antibody (1:40 dilution;Dako) and detected with Vector Laboratories ABC elite kit (as described above). A section of the duodenum was included on each slide to confirm the delivery of the BrdUrd. A minimum of three animals/group were counted, and a total of 3000 cells/animal were counted.

Statistical analyses were performed using the GraphPad Prism version 2.0 program (GraphPad Software, Inc. San Diego, CA). Student’s t test analyses were used to calculate probability(P) values. Results are presented as mean ± SE. For all tests, P < 0.05 was accepted as significant.

The basal level of TGF-α expression in the liver of TGF-αtransgenic mice is high, although it can be increased by the addition of zinc sulfate to drinking water (zinc water). By using Western blot analysis, we determined the extent to which the expression of the Bcl-2 transgene could be modulated by placing Bcl-2 transgenic mice on zinc water. Expression levels of the human Bcl-2 transgene in mice maintained on either normal drinking water or zinc water for 30 days were examined. There was no expression of Bcl-2 in the livers of WT littermates of Bcl-2 transgenic mice, and little, if any, Bcl-2 transgene expression was detectable in Bcl-2 transgenic mice drinking normal water. In contrast, expression was elevated approximately 10-fold in the Bcl-2 transgenic mice that received zinc water (Fig. 1 A).

To verify the ability of the Bcl-2 transgene to inhibit hepatocyte apoptosis, both Bcl-2 transgenic and WT mice were treated with anti-Fas agonist antibody. Both groups of mice were given zinc water for 1 week before treatment to induce Bcl-2 transgene expression. Mice were injected with 10 μg of anti-Fas antibody (clone Jo2) and killed 3 h later (Fig. 1 B). Treatment with the Jo2 antibody induced massive apoptosis in WT animals, such that more than 30 apoptotic nuclei/×400 field were found. In contrast, Bcl-2 transgenic mice had, on average, only 2 apoptotic nuclei/×400 field. The protection against Fas-induced hepatocyte cell death in vivoin Bcl-2 transgenic mice was similar to the results reported by Lacronique et al. (9).

To examine the effects of Bcl-2 expression on TGF-α-mediated hepatocarcinogenesis, we generated double transgenic mice capable of expressing both TGF-α and Bcl-2. This was accomplished by mating heterozygous TGF-α female mice of the MT42 line (CD1 background) to heterozygous Bcl-2 males (C57B6/C3H background) to produce TGF-α/Bcl-2 double transgenic mice, TGF-α and Bcl-2 single transgenic mice, and WT littermates. A significant characteristic of TGF-α transgenic mice of the MT42 line on a CD1 background is the highly elevated hepatocyte proliferation evident in young animals. Although the proliferation diminishes with time, it remains elevated up to the point of tumor development (16). Because the newly generated lines of TGF-α single transgenic mice and TGF-α/Bcl-2 double transgenic mice have a different genetic background from the MT42 mice that were previously studied, it was necessary to determine the levels of hepatocyte proliferation in animals with the hybrid background and to ascertain whether Bcl-2 expression would alter the TGF-α proliferative effect.

We measured hepatocyte proliferation in 4-week-old male TGF-α/Bcl-2 double transgenic mice, TGF-α and Bcl-2 single transgenic mice, and WT controls (Fig. 2,A) that had been maintained on zinc water from the time of weaning at 3 weeks of age. As measured by 3-day BrdUrd incorporation,hepatocyte proliferation was approximately 4-fold higher in TGF-α and TGF-α/Bcl-2 transgenic mice compared with Bcl-2 transgenic and WT mice (Fig. 2 A). The percentage of BrdUrd-positive cells in TGF-α single transgenic mice (18%), was similar to that in TGF-α/Bcl-2 double transgenic mice (17%), demonstrating that Bcl-2 expression did not affect the ability of TGF-α to enhance hepatocyte proliferation in 4-week-old mice. Although hepatocyte proliferation was increased in the TGF-α transgenic mice at 4 weeks of age, the labeling index was approximately half of that observed previously with the MT42 line (16). The amount of TGF-α mRNA expressed in the animals used for the present study was similar to that of MT42 mice (data not shown), indicating that the genetic background of the mice can influence the extent of hepatocyte proliferation independent of the amount of TGF-α being expressed.

To determine whether the increase in hepatocyte proliferation seen at 4 weeks of age had an effect on organ size, the liver:body weight ratio was examined. In both the TGF-α single transgenic and TGF-α/Bcl-2 double transgenic mice, this ratio was significantly increased compared with the ratio in WT and Bcl-2 single transgenic mice (Fig. 2,B). The expression of Bcl-2 in the TGF-α/Bcl-2 double transgenic and Bcl-2 single transgenic mice was verified by Western blot analysis (Fig. 2 C). Bcl-2 was detected only in Bcl-2 single transgenic mice or in TGF-α/Bcl-2 double transgenic mice. Thus, at 4 weeks of age, mice expressing either TGF-α alone or TGF-α and Bcl-2 had elevated levels of hepatocyte proliferation and increased liver:body weight ratios.

Previous work using TGF-α transgenic mice of the MT42 line (CD1 background) showed that there was 10–12% hepatocyte proliferation from 4 months of age to 8 months of age (16). In contrast,in the hybrid background mice used in the present study, the rate of hepatocyte proliferation was negligible in 4–8-month-old TGF-α or Bcl-2 single transgenic mice as well as in TGF-α/Bcl-2 double transgenic mice. In all of these lines, hepatocyte proliferation did not differ from that of WT mice. Similarly, the liver:body weight ratios in 8–10-month-old mice were similar in WT, single transgenic,and double transgenic mice (Fig. 3A). To determine whether TGF-α continued to be expressed in TGF-α single transgenic and TGF-α/Bcl-2 double transgenic mice, RNA was obtained from 8–10-month-old mice and analyzed by Northern blotting. The data shown in Fig. 3,B demonstrate that TGF-αwas expressed at very high levels in both TGF-α single transgenic and TGF-α/Bcl-2 double transgenic mice, indicating that coexpression of Bcl-2 (as determined by Western blot analysis, Fig. 3 C) with TGF-α did not affect the expression of the TGF-α transgene.

A total of 32 TGF-α, 26 TGF-α/Bcl-2, and 19 Bcl-2 transgenic mice and 33 WT mice were examined for tumor development between 8 and 15 months of age. Livers of Bcl-2 transgenic mice showed no significant histological alterations and were similar to those of WT mice throughout the duration of the study. All TGF-α/Bcl-2 double transgenic mice killed at 8–10 months of age showed multiple foci of basophilic hepatocytes containing dysplastic cells with abnormal nuclear morphology (Fig. 4, B and D), and one of the animals had a small,clearly demarcated adenoma. Unlike the TGF-α/Bcl-2 double transgenics, livers of the TGF-α single transgenic mice showed no basophilic foci (Fig. 4, A and C). However, one of the nine mice killed at this time had a well-developed HCC with a diameter of approximately 0.5 cm. Thus, at 8–10 months of age, the livers of the TGF-α/Bcl-2 double transgenic mice contained potentially preneoplastic lesions detected by histological analysis. Although a TGF-α transgenic mouse had developed a HCC detectable at gross examination, the morphology of the liver of TGF-α transgenic mice at this age was normal. The findings were entirely different in animals killed at later times. In animals examined at either 11–12 months or 13–15 months of age, TGF-α single transgenic mice had a much higher incidence of HCC than double transgenic mice (Fig. 5, A and B).

At 11–12 months, six of seven TGF-α transgenic mice had HCC, whereas no tumors were detected grossly or histologically in TGF-α/Bcl-2 double transgenic mice. At 13–15 months, more than 80%of TGF-α transgenic mice had HCC, whereas only 54% of the double transgenics had tumors. In addition, of the 20 TGF-α single transgenic mice that did develop liver tumors, 16 had tumors that were larger than 1 cm in diameter. In contrast, of the seven TGF-α/Bcl-2 double transgenic mice that developed liver tumors, only one had a tumor diameter larger than 1 cm. Thus, coexpression of TGF-α and Bcl-2 in the double transgenic mice decreased the incidence of HCC,delayed tumor development, and inhibited tumor growth.

We evaluated both mitotic and apoptotic activity in tumors from livers of TGF-α single transgenic mice and TGF-α/Bcl-2 double transgenic mice (Fig. 5 C) by determining the number of mitotic and apoptotic bodies per 1000 cells in the tumors. The overall mitotic activity in the tumors from TGF-α single transgenic mice was nearly three times higher than that of tumors from TGF-α/Bcl-2 double transgenic mice. Apoptotic bodies were also observed more frequently(3.3 versus 0.7 cells/1000 cells counted) in tumors from the TGF-α transgenic mice than in tumors from TGF-α/Bcl-2 double transgenic mice. Overall, in mice maintained on zinc water from the time of weaning, tumors from TGF-α/Bcl-2 double transgenic mice had less mitotic and apoptotic activity than tumors from TGF-α single transgenic mice.

In addition to the differences in tumor incidence and mitotic activity observed between the TGF-α single transgenic and TGF-α/Bcl-2 double transgenic mice, Bcl-2 had profound effects on the morphology of TGF-α-induced liver tumors. Tumors that developed in TGF-α single transgenic mice were well-differentiated HCCs that displayed either solid or trabecular patterns (Fig. 6, A and D). The morphology of these tumors was similar to our previous observations (Ref. 13; Fig. 6, A and D). In contrast, tumors that developed in TGF-α/Bcl-2 double transgenic mice consisted of very large single or binucleated hepatocytes that formed long, 1-cell-thick villiform structures separated by blood-filled spaces and cavities. The hepatocytes in these villi were often individually lined by endothelial cells (Fig. 6, B and E). Although variation occurred from animal to animal, the liver tumors that did arise in the TGF-α/Bcl-2 double transgenic mice were hemorrhagic and cystic, with abundant endothelial cells in some areas of the tumors.

We also studied tumor development and tumor morphology in a group of TGF-α/Bcl-2 double transgenic mice that were not maintained on zinc water. Because Bcl-2 expression is minimal or undetectable in these animals (see below), they served as an excellent control for the analysis of tumorigenesis in the TGF-α/Bcl-2 double transgenic mice described above, in which Bcl-2 was induced by maintaining the animals on zinc water throughout their lifetime. Tumors that developed in double transgenic mice in which Bcl-2 was not induced (Fig. 6, C and F) were solid or trabecular HCC, similar to those detected in TGF-α single transgenic mice. No animal in this group developed villous tumors with vascular spaces as observed in TGF-α/Bcl-2 double transgenic mice in which Bcl-2 was expressed. These findings provide direct evidence that Bcl-2 expression modified the morphology of liver tumors induced by TGF-α.

Expression of the transgenes was examined in tumors and adjacent tissue using Northern blot analysis and ISH for TGF-α and Western blot and IHC for Bcl-2. Compared with WT mice, expression of TGF-αmRNA was higher in the tumors and adjacent tissue of both TGF-α/Bcl-2 double transgenic mice and TGF-α single transgene mice (Fig. 7,A). Although there was considerable variation among the animals, TGF-α mRNA expression was generally higher in TGF-α single transgenic mice than in double transgenic mice. Bcl-2 expression, as determined by Western blot analysis, was relatively uniform in the tumors and surrounding tissue of TGF-α/Bcl-2 double transgenic mice(Fig. 7,B). Expression of Bcl-2 was weak or undetectable in double transgenic mice that did not receive zinc water to induce this transgene (Fig. 7,B). IHC analysis of Bcl-2 expression was consistent with the pattern of expression detected in Western blots,that is, there was intense Bcl-2 staining in both tumor and adjacent tissue (Fig. 8, A–C). Detection of TGF-α by ISH revealed that TGF-αwas more strongly expressed in HCC of TGF-α single transgenic mice than in the liver tumors that developed in TGF-α/Bcl-2 double transgenic mice.

The liver tumors that developed in TGF-α/Bcl-2 double transgenic mice maintained on zinc water were smaller in size and less mitotically active than those of TGF-α single transgenic mice. A possible explanation for this finding is that Bcl-2 may inhibit tumor progression. To investigate this hypothesis, exposure to zinc water in a group of mice was delayed until they reached 8.5 months of age. Because Bcl-2 expression in uninduced mice is very low or undetectable,Bcl-2 expression in this group of mice is considered to have started at 8.5 months. Four months later, these animals were killed and examined for tumor development. Grossly detectable HCCs developed in 50% (two of four) of TGF-α single transgenic mice that served as controls for these experiments but were absent in all nine of the TGF-α/Bcl-2 double transgenic mice examined (Fig. 9). Histological analysis of the livers of double transgenic mice revealed no obvious abnormality, with the exception of a small nonneoplastic proliferative lesion detected in an otherwise normal liver section in one animal (data not shown). Therefore, late induction of the Bcl-2 transgene reduced the tumorigenic effect of TGF-α.

During carcinogenesis in rat liver, the rates of both cell proliferation and apoptosis increase in hyperplastic nodules(22). If the rates of proliferation and apoptosis could be independently modulated, it would be expected that increased proliferation combined with decreased cell death would cause rapid hepatic enlargement and tumor formation. To test the hypothesis that inhibition of apoptosis enhances tumor growth, we crossed TGF-αtransgenic mice with Bcl-2 transgenic mice that are protected against Fas-mediated hepatocyte apoptosis. Overexpression of TGF-αin the livers of transgenic mice leads to the development of adenomas and HCCs in more than 85% of these animals after 12 months of age(13, 15). Contrary to the expectations, tumor development was inhibited in the double transgenic mice that overexpressed both TGF-α, a potent hepatocyte mitogen, and Bcl-2. Expression of the Bcl-2 transgene was inducible by adding zinc to the drinking water. Using the inducible system for Bcl-2 expression, it was possible to establish that inhibition of tumor growth in TGF-α/Bcl-2 double transgenic mice was a consequence of Bcl-2 expression. Tumors that eventually developed in TGF-α/Bcl-2 double transgenic mice were smaller in size, had lower mitogenic and apoptotic activity than those in TGF-α single transgenic mice, and had an unusual morphology.

The inhibition of hepatocarcinogenesis in double transgenic mice was particularly surprising because numerous foci of abnormal basophilic cells were detected in the livers of these animals sampled between 6 and 10 months of age. These foci were not detected in either TGF-α or Bcl-2 single transgenic mice, suggesting that they resulted from the combined expression of TGF-α and Bcl-2 in the double transgenic mice. Although basophilic foci observed during liver carcinogenesis are often considered to be preneoplastic lesions (23, 24), this was not the case in TGF-α/Bcl-2 double transgenic mice. The basophilic foci that developed in these animals neither persisted nor progressed into adenomas or carcinomas.

Inhibition of TGF-α tumorigenesis in TGF-α/Bcl-2 double transgenic mice could be the consequence of at least two types of effects: (a) inhibition of TGF-α expression by Bcl-2; or(b) a direct effect of Bcl-2 expression on cell proliferation. Analysis of TGF-α expression in TGF-α/Bcl-2 double transgenic mice did not reveal an obvious decrease of TGF-αexpression relative to that in TGF-α single transgenic mice. On the other hand, a study of the timing of DNA replication during liver regeneration in Bcl-2 single transgenic animals showed that the peak of DNA synthesis was delayed in these animals.⁴These observations lead us to conclude that Bcl-2 can inhibit cell cycle progression. If this is indeed the major mechanism by which Bcl-2 expression inhibits tumorigenesis in TGF-α/Bcl-2 double transgenic mice, it would be expected that Bcl-2 would have an effect on tumor progression rather than initiation. This appears to be the case, as demonstrated by experiments in which Bcl-2 expression initiated at 8.5 months of age proved to be sufficient to inhibit tumor development.

Similar to our results, recent studies using Bcl-2 transgenic mice have revealed that Bcl-2 expression can inhibit tumorigenesis. Murphy et al. (25) found that that Bcl-2 expression delayed mammary tumor development in dimethylbenz(a)anthracene-treated mice, and de La Coste et al. (26) demonstrated the ability of Bcl-2 to inhibit liver tumorigenesis in c-myc transgenic mice. These in vivo studies correlate with tissue culture studies in which overexpression of Bcl-2 was shown to delay cell cycle entry and decrease cell proliferation (27, 28, 29). These results provide an explanation for the tumor suppressor effects rather than the oncogenic effects of Bcl-2 in TGF-α/Bcl-2 double transgenic mice.

The antiproliferative effect of Bcl-2 combined with its antiapoptotic activity may explain some aspects of the atypical morphology of the small numbers of tumors that eventually developed in the TGF-α/Bcl-2 double transgenic mice. It is possible that the very large hepatocytes present in these tumors resulted from the inhibition of both apoptosis and replication of cells exposed to the constant TGF-α mitogenic stimulus. However, the antiapoptotic effects off Bcl-2 are unlikely to have contributed to the proliferation of endothelial cells and the formation of large vascular spaces observed in these tumors. It is of interest that TGF-α can have angiogenic effects (30). However, the tumors formed in TGF-α single transgenic mice were typical solid or trabecular HCCs without vascular features.

Our previous studies of liver tumorigenesis in TGF-α transgenic mice were done with the MT42 line in a CD1 background. In these animals, TGF-α enhanced hepatocyte proliferation at 1 month of age,and elevated hepatocyte replication persisted until tumors emerged. Because the TGF-α/Bcl-2 double transgenic mice had a CD1 × C57B6C3H background, we produced TGF-α single transgenic mice with this new genetic background to serve as controls for tumorigenesis experiments. TGF-α transgenic mice resulting from the CD1 × C57B6C3H cross had enhanced hepatocyte proliferation at 1–2 months of age, but no later. In these animals, tumors emerged in livers that were morphologically normal and had no increased hepatocyte proliferation. The mechanisms by which neoplasms developed in the absence of elevated hepatocyte proliferation preceding tumor formation in TGF-α transgenic mice remain to be established. One possibility is that the enhanced hepatocyte proliferation that occurs during the first 2 months of life in these animals is sufficient to cause genomic damage that results in tumorigenesis many months later.

We thank Colin Pritchard for technical assistance with the ISH,Maryland Rosenfeld and Jean Camblell for critical reading of the manuscript, and members of the Fausto laboratory for helpful discussions.