Abstract

A growing body of evidence from in vitro studies indicates that gap junction proteins connexins may have a tumor-suppressor function. Our previous double transfection experiments on HeLa cells have shown that a dominant-negative mutant V139 M of connexin32 (Cx32) can abolish gap junctional intercellular communication (GJIC). To examine whether the same dominant-negative mutant of Cx32 inhibits GJIC between hepatocytes in vivo and thus modulates cell proliferation and susceptibility to hepatocarcinogenesis, we created transgenic mice with the mutant Cx32 gene driven by a liver-specific albumin promoter. These mice developed normally both before and after birth, and GJIC in their liver was diminished, as expected. No increase in incidence of spontaneous tumors of any site was observed in the transgenic mice. Rather unexpectedly, cell proliferation during liver regeneration after partial hepatectomy was retarded by 24 h in the transgenic mice compared with the wild-type mice. In contrast, the transgenic male mice were more susceptible to diethylnitrosamine-induced hepatocarcinogenesis, developing more liver tumors with shorter latency. These results show that GJIC can coordinate cell growth both positively and negatively in vivo , supporting the idea that GJIC is essential for maintenance of homeostasis.

Introduction

The gap junction is a specific type of intercellular junction formed between two adjacent cells and functions as an intercellular channel through which small water-soluble molecules ( M r < 1000) passively pass directly from the inside of one cell into neighboring cells. Gap junctions are thus considered to act as a key element in the maintenance of tissue homeostasis ( 1 ). A gap junction channel consists of two juxtaposed hemichannels (connexons), each provided by one of two adjacent cells. Each connexon is composed of six molecules of the membrane-spanning connexin proteins, encoded by a family of genes consisting of at least 20 members in mammals. The pattern of connexin species expressed differs between tissues and changes during differentiation or development ( 2 ). One cell usually expresses two or more types of connexin protein; for example, both connexin (Cx)26 and Cx32 are expressed in hepatocytes. The general nature of the molecules that can pass is common to all types of gap junction, but it has been observed that each gap junction may have a relatively specific permeability ( 3 , 4 ).

Impaired gap junctional intercellular communication (GJIC) causes various abnormalities, leading to several diseases in humans. Many mutations in connexin genes have been reported, such as those in Cx26 in non-syndromic sensorineural deafness ( 5 ) and palmoplantar keratoderma ( 6 ), Cx31 in erythrokeratoderma variabilis ( 7 ) and hearing impairment ( 8 ), Cx32 in X-linked Charcot-Marie-Tooth disease ( 9 ) and Cx50 in zonular pulverulent cataract ( 10 ). Most of these mutations are thought to be of a loss-of-function type and some of them are suggested to have dominant-negative characteristics ( 11 ). Our laboratory has reported previously that several mutant forms of Cx26, Cx32 and Cx43 inhibit GJIC mediated by their wild-type counterparts in a dominant-negative manner when the mutant and wild-type forms are co-expressed in vitro ( 1215 ).

Cancer cells usually display uncontrolled growth and are often GJIC-deficient. While many studies have revealed that enforced expression of connexin in malignant cells restores GJIC and suppresses tumorigenicity ( 16 , 17 ), few have addressed the role of gap junctions in growth control by using in vivo systems.

Recent progress in genetic engineering has enabled us to analyze in vivo the function of specific genes by creation of transgenic and/or knockout mice ( 18 ). Several lines of mice deficient in different connexin genes are currently available and have been used to enrich our knowledge of the biological roles of connexins. However, connexin-deficient mice are not always ideal for studying the role of connexins in carcinogenesis, because the systemic effects of gene disruption often alter the normal function of the organ of interest and may complicate the analysis of carcinogenesis involving local events such as somatic mutations ( 19 ). Moreover, mice lacking the Cx26 and Cx43 genes, both of which are expressed in many tissues, die before and soon after birth, respectively ( 20 , 21 ). Therefore, we decided to generate transgenic mouse lines in which a dominant-negative mutant Cx32, driven by an albumin promoter, was expressed exclusively in the liver, in order to examine the effect of reduced GJIC on liver carcinogenesis in vivo in the absence of systemic effects caused by disruption of the Cx32 gene, that is expressed widely in the body. In this study, we have established such transgenic mouse lines and have characterized them in terms of cell growth and carcinogenesis in the liver.

Materials and methods

Mice

An Xho I– Xba I fragment isolated from Cx32 V139 M/pBluescriptII SK + phagemid ( 14 ) containing cDNA for mutant Cx32 V139 M (valine to methionine at codon 139) was inserted into the corresponding sites of pGEMAlbSVPA plasmid ( 22 ) such that Cx32 V139 M would be driven by the albumin promoter. This construct was digested with Apa I and Mlu I to generate a transgene fragment, which was then transferred into F 2 hybrid zygotes between F 1 hybrids (C57BL/6 J × CBA/J) under a contract with the Pasteur Institute (Paris, France). Founder mice and their offspring were screened by Southern blot and PCR for genotype analysis. Two lines (A and B) of mice expressing the transgene in the liver were finally selected. These lines A and B were then crossed with C57BL/6 J mice three times and four times, respectively.

For genotype analysis, 35 cycles of PCR were carried out using the forward primer (5′-GGAAGCCCTATAATGAGACC-3′), recognizing the albumin promoter, and the reverse primer (5′-AAGACTTCTCATCACCCCAC-3′), recognizing the coding region of the Cx32 V139 M gene, at 94°C for 30 s, 55°C for 1 min and 72°C for 1 min. Genomic DNA was isolated from the tail.

Allele-specific RT–PCR

Total RNA was extracted from liver with Trizol Reagent (Life Technologies, Paisley, UK) following the manufacturer's instructions. Extracted RNA was treated with 0.01 U/µl DNase I (Promega, Madison, WI) to eliminate traces of genomic DNA. Five nanograms of purified total RNA was transcribed to cDNA primed by random hexaoligonucleotide with MuLV reverse transcriptase (Promega), then subjected to allele-specific PCR to detect transgene-derived cDNA. Twenty-five cycles of PCR were performed using the forward primer (5′-TGCCAGGGAGGTGTGATATC-3′) and the reverse primer (5′-AAGACTTCTCATCACCCCAC-3′) at 94°C for 30 s, 55°C for 1 min and 72°C for 30 s.

Immunoblotting

For estimation of Cx26 and Cx32, total protein was extracted from the liver of transgenic and wild-type mice. The tissue was homogenized in a sample buffer containing 60 mM Tris–HCl, pH 6.8, 2% sodium dodecyl sulfate (SDS), 12% glycerol, 0.1 M dithiothreitol and 1 mM phenylmethylsulfonyl fluoride. Protein was electrophoresed in 12% SDS polyacrylamide gel and transferred onto polyvinylidene difluoride membrane (Amersham, Little Chalfont, UK). A monoclonal antibody (6-3G11, raised against the C-terminal region of Cx32, diluted 1:1000) ( 23 ) or polyclonal antibody (raised against a synthetic peptide from the cytoplasmic loop of Cx32, diluted 1:5000) ( 24 ) was used to detect Cx32. A polyclonal anti-Cx26 antibody, kindly provided by Drs A.Kuraoka and Y.Shibata (Fukuoka, Japan), was diluted at 1:1000. After incubation with a goat anti-rabbit or anti-mouse antibody conjugated with peroxidase, the reaction was revealed either with a solution containing diaminobenzidine (Sigma, St Louis, MO) and nickel sulfate in the presence of hydrogen peroxide or with an enhanced chemiluminescence (ECL) western blotting analysis system (Amersham). The ECL component was applied following the manufacturer's instructions. Light emission was detected by a short exposure to a light-sensitive autoradiography film (Hyperfilm ECL, Amersham).

Immunostaining of connexin

Fresh liver slices from transgenic or wild-type mice were embedded in 7% gelatin (Sigma), frozen and sectioned in a cryostat (Reichert-Jung, Germany). The 6 μm slices were air-dried, fixed in 10% formalin in PBS for 5 min, washed and incubated overnight at 4°C with mono- or polyclonal antibodies against Cx32 (the same as for immunoblotting) or polyclonal antibody against Cx26. A goat anti-rabbit or mouse immunoglobulin antibody conjugated with peroxidase (Sigma) was applied. The reaction was revealed with a solution containing diaminobenzidine, nickel chloride and hydrogen peroxide, and further enhanced by silver development ( 25 ). For immunofluorescence, we followed the previously described procedure by using the above antibodies against Cx32 ( 26 ) and observed the fluorescence signals under a fluorescence Nikon E-800 microscope (Tokyo, Japan).

Evaluation of the GJIC capacity in liver

To evaluate GJIC capacity, an ex vivo Lucifer yellow (LY) dye transfer assay was performed using fresh liver slices from transgenic or wild-type mice, as described previously ( 27 ). In brief, 12 injections of LY CH (Sigma) were performed with a hand-made glass needle into a fresh thin (1 mm) slice of mice liver using an Eppendorf microinjector Model 5242 (Hamburg, Germany). Injected slices were quickly washed in PBS at room temperature, embedded in 7% gelatin (Sigma), frozen and then cut into 6 µm cryosections. Each spot of LY dye spread on serial sections was photographed under a fluorescence Olympus Vannox T microscope (Tokyo, Japan). The areas of LY dye spreading were measured using the NIH image analysis system.

Partial hepatectomy and evaluation of S-phase hepatocytes

To induce liver regeneration, the left liver lobe was surgically removed from the liver of transgenic and wild-type male mice, under anesthesia with ether according to the technique of Higgins and Anderson ( 28 ) modified for mice. Five mice each of both genotypes were used per time point.

To evaluate the number of S-phase hepatocytes, groups of transgenic and wild-type male mice received an i.p. injection of 10 mg/kg body wt of bromodeoxyuridine (BrdU) (Sigma) 0, 24, 48, 72, 96 or 120 h after surgery. After 1 h of exposure to BrdU, the mice were killed and representative liver slices were fixed in methacarn (60% methanol, 30% chloroform and 10% acetic acid) for 8–12 h for BrdU staining, or embedded in 7% gelatin solution in PBS and frozen for subsequent staining of connexins. Sections were stained for BrdU using anti-BrdU monoclonal antibodies (diluted 1:1000, Sigma) and the avidin–biotin–peroxidase system (Dako, CA), revealed by diaminobenzidine. About 3000 hepatocytes were counted in random fields from at least three lobes of each liver, and the percentage of positive nuclei was calculated.

Chemical mitogen-induced proliferation in the liver

1,4-Bis[2-(3,5-dichloropyridyloxy)]benzene (TCPOBOP), a phenobarbital-like inducer of microsomal monooxygenase activity ( 29 ), was used as a mitogen. TCPOBOP synthesized as described ( 30 ) was kindly provided by Mrs Croizy (Paris, France). Male mice aged ∼15 weeks received a single dose of 5.8 mg/kg body wt TCPOBOP dissolved in corn oil by gavage. Groups of transgenic and wild-type mice were killed at 0, 24, 48 and 72 h after administration of the mitogen. One hour before killing, each mouse was treated with BrdU as described above. Mice and livers were weighed, and representative liver slices were processed for BrdU immunostaining.

Hepatocarcinogenesis

A single dose of 0.9% diethylnitrosamine (DEN) (Sigma) in saline solution was injected intraperitoneally (5 µg/g body wt) to groups of 15-day-old males and females of both transgenic and wild-type mice using the protocol described by Vesselinovitch and Mihailovich ( 31 ). The animals were genotyped at 6–8 weeks of age, and divided into four groups. Animals from each group were killed 15, 25, 35 and 49 weeks after DEN treatment. At the time of killing, body and liver weight were recorded for each mouse. Livers were examined for the presence of gross lesions. The lobes were separated and the number and size of superficial lesions were recorded. Internal lesions were counted and measured after preparation of 3-mm sections. Representative slices from standard positions in the liver lobes and those from large liver lesions were collected and fixed in methacarn for histopathological examination. Liver lesions were classified histopathologically according to Harada et al . ( 32 ).

Statistical analysis

For comparison of means of transgenic and wild-type mouse liver parameters, the Student's t test with a 95% confidence interval was applied. For comparison of the multiplicity data in the hepatocarcinogenesis study, a two-way analysis of variance (ANOVA), considering time and genotype as the sources of variation, was performed.

Results

Expression of mutant Cx32 V139 M transgene

We have generated two independent lines of transgenic mice. To verify liver-specific expression of the introduced mutant Cx32 V139 M gene, allele-specific RT–PCR was performed. While one line (designated as line A) expressed mRNA from the transgene only in the liver as expected ( Figure 1A ), the other (designated as line B) showed strong leaky expression in kidney and brain in addition to the liver ( Figure 1B ). Both lines of mice were used for most of the experiments and no difference was found between them. In this paper, only the data from line A are presented.

Fig. 1.

Expression of mutant Cx32 V139 M and Cx26. (A and B) Allele-specific RT–PCR for mutant Cx32 V139 M in transgenic (tg) mice from line A ( A ) and line B ( B ). While the mouse from line A expresses transgene-derived mRNA only in the liver, that from line B shows a leaky expression in other organs. ( C ) Immunoblotting of Cx32 in the liver of the offspring crossed between line A and Cx32-deficient line. Mutant protein is expressed in the liver of the mouse having no wild-type Cx32 but mutant Cx32 V139 M (lane Cx32 tg + ). ( D ) Immunoblotting of Cx26 in the livers of both transgenic and wild-type mice.

Fig. 1.

Expression of mutant Cx32 V139 M and Cx26. (A and B) Allele-specific RT–PCR for mutant Cx32 V139 M in transgenic (tg) mice from line A ( A ) and line B ( B ). While the mouse from line A expresses transgene-derived mRNA only in the liver, that from line B shows a leaky expression in other organs. ( C ) Immunoblotting of Cx32 in the liver of the offspring crossed between line A and Cx32-deficient line. Mutant protein is expressed in the liver of the mouse having no wild-type Cx32 but mutant Cx32 V139 M (lane Cx32 tg + ). ( D ) Immunoblotting of Cx26 in the livers of both transgenic and wild-type mice.

As no antibody specific to mutant V139 M Cx32 is available, we could not confirm that the expressed mRNA was properly translated to the corresponding protein in the liver of line A. To resolve this problem, we crossed line A mice with Cx32-deficient mice ( 33 ) so as to obtain mice in which only the mutant Cx32 V139 M was expressed in the liver, without any trace of endogenous wild-type Cx32. As shown in Figure 1C , Cx32 V139 M protein was detected in the liver of such a hybrid mouse, indicating that mRNA from the Cx32 V139 M construct was indeed translatable in hepatocytes.

The transgenic mice manifest no abnormal phenotype

Both lines of transgenic mice (A and B) were bred and developed normally. No gross or histological differences were observed between transgenic and wild-type mouse livers.

It has been reported that levels of Cx26 protein are greatly reduced in the liver of Cx32-deficient mice without any decrease in its mRNA ( 33 ). However, our transgenic mice showed a level of Cx26 expression in the liver similar to that of the wild-type counterpart ( Figure 1D ) and Cx26 protein was detected immunohistochemically as plaques at a cell–cell contact area of hepatocytes in mice of both genotypes ( Figure 2A and B ).

Fig. 2.

Immunohistochemistry of Cx26 ( A and B ) and Cx32 ( C and D ) in the livers of transgenic (A and C) and wild-type (B and D) mice. Scale bar: 15 µm.

Fig. 2.

Immunohistochemistry of Cx26 ( A and B ) and Cx32 ( C and D ) in the livers of transgenic (A and C) and wild-type (B and D) mice. Scale bar: 15 µm.

We and others reported previously that the mutant Cx32 V139 M per se could not be integrated into the plasma membrane ( 14 , 34 ). Furthermore, when the mutant Cx32 V139 M gene was transfected into wild-type transfected HeLa cells, the numbers of gap junction plaques decreased dependently on the level of expression of the mutant. We therefore suggested that the mutant Cx32 V139 M might form non-functional heteromeric complexes with the wild-type Cx32 that might be defective for localization into the plasma membrane of cells ( 14 ).

Our transgenic mouse livers, however, showed a similar pattern of Cx32 in plaques to that in the livers of the wild-type siblings ( Figure 2C and D ). Interestingly, while no significant signal was detected in the cytoplasm or nucleus in our transgenic mouse hepatocytes co-expressing both the mutant Cx32 V139 M and wild-type ( Figure 3E ), the mutant was localized in the nucleus in the absence of the wild-type, as revealed by immunofluorescence, in the liver of the hybrid mouse between Cx32-deficient and our transgenic mice ( Figure 3A ). These results suggest that, while the mutant Cx32 V139 M per se is incapable of forming gap junctions, it may be able to participate in gap junction plaques in hepatocytes in vivo only in the presence of the wild-type Cx32, whether or not connexons containing the mutant Cx32 are functional.

Fig. 3.

Immunofluorescence of Cx32 ( A , C and E ) and nuclear counter-staining with propidium iodide ( B , D and F ) in the livers of hybrid (Cx32 tg + mouse) (A and B), Cx32-deficient (C and D) and transgenic (E and F) mice. Arrow heads indicate the mutant Cx32 V139 M proteins localized in the nuclei. Scale bar: 15 µm.

Fig. 3.

Immunofluorescence of Cx32 ( A , C and E ) and nuclear counter-staining with propidium iodide ( B , D and F ) in the livers of hybrid (Cx32 tg + mouse) (A and B), Cx32-deficient (C and D) and transgenic (E and F) mice. Arrow heads indicate the mutant Cx32 V139 M proteins localized in the nuclei. Scale bar: 15 µm.

GJIC capacity is reduced in transgenic mouse liver

In HeLa cells, the mutant Cx32 V139 M almost shut off cell–cell communication exerted by the wild-type in a dominant-negative manner, as revealed by the LY dye-coupling assay ( 14 ). To examine whether GJIC capacity is reduced in transgenic mouse liver by a dominant-negative effect of the mutant, we measured the cell coupling capacity in the livers of six transgenic mice (line A) and wild-type siblings by LY microinjection ( Figure 4A ), as described in Materials and methods. As shown in Figure 4B , GJIC was reduced by 30% in the liver of transgenic mice, confirming that the mutant Cx32 V139 M acts in vivo as a dominant-negative inhibitor of wild-type Cx32. This inhibitory effect of the mutant Cx32 was lower than expected, probably because the modestly expressed mutant protein ( Figure 3A ) may have failed to abolish completely the function of the wild-type. Moreover, it is probable that Cx26 co-expressed with Cx32 in hepatocytes also contributed to such a residual GJIC.

Fig. 4.

GJIC capacity as measured by the areas of LY spots after ex vivo dye-transfer assay in fresh livers of wild-type and transgenic mice. ( A ) Representative LY spots. Scale bar: 200 µm. ( B ) GJIC capacity in fresh livers of wild-type and transgenic mice. Six each of the wild-type and transgenic mice were subjected to the assay. Ten slices per mouse were selected randomly and microinjected with LY into 12 sites per slice. *P < 0.05.

Fig. 4.

GJIC capacity as measured by the areas of LY spots after ex vivo dye-transfer assay in fresh livers of wild-type and transgenic mice. ( A ) Representative LY spots. Scale bar: 200 µm. ( B ) GJIC capacity in fresh livers of wild-type and transgenic mice. Six each of the wild-type and transgenic mice were subjected to the assay. Ten slices per mouse were selected randomly and microinjected with LY into 12 sites per slice. *P < 0.05.

Liver regeneration after partial hepatectomy is delayed in transgenic mice

In the quiescent liver of Cx32-deficient mice, the number of S-phase hepatocytes is significantly increased, as measured by bromodeoxyuridine (BrdU) incorporation, compared with wild-type liver, although no gain of liver weight was seen ( 35 ). In contrast, when our transgenic mice and wild-type counterparts were examined by BrdU labeling in the quiescent liver, they both presented a similar percentage of S-phase hepatocytes (data not shown).

We next subjected the animals to partial hepatectomy (resection of the left lobe) to observe the effect of reduced GJIC on hepatocyte proliferation under growth pressure. In the wild-type mouse liver, the hepatocyte BrdU labeling index rose gradually, reaching a 50-fold increase at 48 h after surgery ( Figure 5A and B ). On the other hand, the transgenic mouse liver showed a peak of DNA synthesis 72 h after surgery ( Figure 5A ). Despite a delay in proliferative response to partial hepatectomy, the transgenic mice were able to complete the process of liver regeneration.

Fig. 5.

DNA synthesis and expression patterns of hepatic connexins after partial hepatectomy. ( A ) BrdU labeling index in regenerating liver in transgenic and wild-type mice after partial hepatectomy. Five mice each of both genotypes were used per time point. No error bar is indicated when the SD is too small to show. *P < 0.05. **P < 0.05. ( B ) Immunostaining of BrdU 48 h after partial hepatectomy. ( C ) Immunoblotting of Cx26, Cx32 and albumin after partial hepatectomy.

Fig. 5.

DNA synthesis and expression patterns of hepatic connexins after partial hepatectomy. ( A ) BrdU labeling index in regenerating liver in transgenic and wild-type mice after partial hepatectomy. Five mice each of both genotypes were used per time point. No error bar is indicated when the SD is too small to show. *P < 0.05. **P < 0.05. ( B ) Immunostaining of BrdU 48 h after partial hepatectomy. ( C ) Immunoblotting of Cx26, Cx32 and albumin after partial hepatectomy.

It is well known that expression of both Cx26 and Cx32 in a rat liver is abolished immediately after partial hepatectomy and restored gradually as the regeneration progresses ( 36 ). In the wild-type mice, unlike rats, both Cx26 and Cx32 continued to be expressed only with a slight decline ( Figure 5C ), which was much less drastic than seen in the rat liver. The transgenic mice also showed a similar expression pattern of Cx26 protein to the wild-type, indicating that the mutant Cx32 V139 M could not modify the expression of Cx26. As shown in Figure 5C , the partially hepatectomized liver maintained similar levels of albumin, suggesting that the transgene remained activated during liver regeneration. As the mutant Cx32 V139 M is not distinguished from the wild-type by the antibody, the expression of Cx32 looks higher in the transgenic mice than the wild-type counterparts ( Figure 5C ).

Transgenic and wild-type livers respond to the mitogen TCPOBOP in a similar manner

The above results suggested that GJIC should play a role in the regulation of mitogenic signals triggered by partial hepatectomy. Is GJIC involved in regulation of the signals induced by direct mitogen? In order to estimate whether the delay in proliferative response after a mitogenic stimulus described above is specific to compensatory regeneration or is a common phenomenon associated with cell growth in the liver, we also analyzed the dynamics of proliferative response in the liver of transgenic mice versus wild-type mice after treatment with the strong primary liver mitogen TCPOBOP ( 29 ).

A single administration of TCPOBOP induced a progressive gain of liver weight in both transgenic and wild-type mice by 30% during the following 72 h in a similar manner ( Figure 6A ). No significant difference in BrdU labeling index between the transgenic and wild-type mice was found at any time point ( Figure 6B ). We thus conclude that the delayed G 1 –S transition observed under conditions of decreased GJIC ( Figures 4 and 5A ) is specific to compensatory regeneration induced by partial hepatectomy.

Fig. 6.

Relative liver weight ( A ) and BrdU labeling index ( B ) of proliferating liver induced by TCPOBOP in transgenic and wild-type mice. No error bar is indicated when the SD is too small to show.

Fig. 6.

Relative liver weight ( A ) and BrdU labeling index ( B ) of proliferating liver induced by TCPOBOP in transgenic and wild-type mice. No error bar is indicated when the SD is too small to show.

Transgenic mice with mutated Cx32 are not prone to spontaneous liver tumors but have increased susceptibility to DEN-induced hepatocarcinogenesis

To verify whether expression of mutated Cx32 predisposes mice to hepatocarcinogenesis, we first estimated the incidence of spontaneous liver lesions in transgenic mice. Twelve male and thirteen female transgenic mice and five male and six female wild-type mice, aged 13–22 months, were necropsied and their livers were examined grossly and histopathologically. Although a few pre-neoplastic lesions were found in the liver of two transgenic male mice, no liver tumors were found in any of the examined mice.

We next examined whether expression of the transgene increased the susceptibility of mice to chemical hepatocarcinogenesis. Fifteen-day-old transgenic and wild-type mice were given a single i.p. injection of DEN and then kept for 15, 25, 35 or 49 weeks, when groups of them were killed and necropsied and examined for the presence of liver lesions.

At 15 weeks after DEN treatment, no gross lesion was found in the liver of either type of mouse, although a few mice showed microscopically detectable pre-neoplastic liver lesions (data not shown). Later, at 25, 35 and 49 weeks of experiment, both transgenic and wild-type mice developed liver tumors. Furthermore, the multiplicity of liver tumors (total number of gross lesions/number of mice carrying gross lesions) was higher in transgenic mice of both sexes compared with their wild-type counterparts ( Table I ). In addition, at 35 and 49 weeks after DEN treatment, liver lesions ( Figure 7 ) in the transgenic mice were larger than in the wild types ( Table I ).

Fig. 7.

Liver lesions developed 49 weeks after treatment with DEN. ( A ) Clear cell focus. ( B ) Hepatocellular adenoma. ( C ) Hepatocellular carcinoma. Scale bar: 50 µm.

Fig. 7.

Liver lesions developed 49 weeks after treatment with DEN. ( A ) Clear cell focus. ( B ) Hepatocellular adenoma. ( C ) Hepatocellular carcinoma. Scale bar: 50 µm.

Table I.

Gross liver lesions in Cx32 transgenic and wild-type mice subjected to hepatocarcinogenesis with DEN as the initiating agent

Time after DEN Group No. of mice with lesions/no. of mice examined Incidence (%) Multiplicity (mean ± SD) Total number of gross lesions in the group  Number (%) of gross lesions in the group according to size
 
     

 

 

 

 

 
<1 mm
 
<3 mm to ≥1 mm
 
<10 mm to ≥3 mm
 
<20 mm to ≥10 mm
 
<30 mm to ≥20 mm
 
≥30 mm
 
25 weeks           
    Females-WT 8/10 80 1.6 ± 0.52 13 4 (31) 9 (69) 0 (0) 
    Females-TG 8/12 67 2.1 ± 1.1 17 10 (59) 7 (41) 0 (0) 
    Males-WT 6/7 86 3.5 ± 2.6 21 12 (57) 9 (43) 0 (0) 
    Males-TG 12/12 100 4.8 ± 2.1 57 32 (56) 25 (44) 0 (0) 
35 weeks           
    Females-WT 4/4 100 1.8 ± 1.0 4 (57) 3 (43) 0 (0) 
    Females-TG 8/10 80 2.9 ± 1.9 23 10 (43) 13 (57) 0 (0) 
    Males-WT 8/8 100 12.5 ± 8.1 100 63 (63) 31 (31) 6 (6) 
    Males-TG 11/11 100 16.3 ± 11.7 179 62 (35) 98 (55) 16 (89) 3 (1.7) 
49 weeks           
    Females-WT 10/11 91 7.6 ± 6.6 71 42 (59) 28 (39) 1 (1.4) 
    Females-TG 5/6 83 14.2 ± 22.3 75 58 (77) 15 (20) 2 (2.7) 
    Males-WT 14/14 100  18.2 ± 11.4 a 255 110 (43) 98 (38) 41 (16) 4 (1.6) 1 (0.39) 1 (0.39) 
    Males-TG 12/12 100  33.4 ± 23.7 a 401 145 (36) 163 (41) 84 (21) 5 (1.3) 3 (0.75) 3 (0.25) 
Time after DEN Group No. of mice with lesions/no. of mice examined Incidence (%) Multiplicity (mean ± SD) Total number of gross lesions in the group  Number (%) of gross lesions in the group according to size
 
     

 

 

 

 

 
<1 mm
 
<3 mm to ≥1 mm
 
<10 mm to ≥3 mm
 
<20 mm to ≥10 mm
 
<30 mm to ≥20 mm
 
≥30 mm
 
25 weeks           
    Females-WT 8/10 80 1.6 ± 0.52 13 4 (31) 9 (69) 0 (0) 
    Females-TG 8/12 67 2.1 ± 1.1 17 10 (59) 7 (41) 0 (0) 
    Males-WT 6/7 86 3.5 ± 2.6 21 12 (57) 9 (43) 0 (0) 
    Males-TG 12/12 100 4.8 ± 2.1 57 32 (56) 25 (44) 0 (0) 
35 weeks           
    Females-WT 4/4 100 1.8 ± 1.0 4 (57) 3 (43) 0 (0) 
    Females-TG 8/10 80 2.9 ± 1.9 23 10 (43) 13 (57) 0 (0) 
    Males-WT 8/8 100 12.5 ± 8.1 100 63 (63) 31 (31) 6 (6) 
    Males-TG 11/11 100 16.3 ± 11.7 179 62 (35) 98 (55) 16 (89) 3 (1.7) 
49 weeks           
    Females-WT 10/11 91 7.6 ± 6.6 71 42 (59) 28 (39) 1 (1.4) 
    Females-TG 5/6 83 14.2 ± 22.3 75 58 (77) 15 (20) 2 (2.7) 
    Males-WT 14/14 100  18.2 ± 11.4 a 255 110 (43) 98 (38) 41 (16) 4 (1.6) 1 (0.39) 1 (0.39) 
    Males-TG 12/12 100  33.4 ± 23.7 a 401 145 (36) 163 (41) 84 (21) 5 (1.3) 3 (0.75) 3 (0.25) 
a

Significantly different by the Student's t test ( P < 0.05).

WT, wild-type; TG, transgenic.

Histological examination of liver slices sampled from standard positions revealed that the transgenic male mice had more hepatocellular carcinomas than the wild-type males. Moreover, at the end of the study (49 weeks), the incidence of histologically diagnosed hepatocellular carcinomas was 42% in the transgenic male mice compared with 14% in the wild-type counterparts ( Table II ).

Table II.

Histological characteristics of liver lesions in transgenic or wild-type mice after treatment with diethylnitrosamine (DEN)

Group  25 weeks after DEN No. (%) of mice with lesions/no. of mice examined
 
   35 weeks after DEN No. (%) of mice with lesions/no. of mice examined
 
   49 weeks after DEN No. (%) of mice with lesions/no. of mice examined
 
  

 
F
 
HA
 
HCC
 
F
 
HA
 
HCC
 
F
 
HA
 
HCC
 
Fem-WT 3/10 (30) 0/10 0/10 2/4 (50) 0/4 0/4 12/12 (100) 1/12 (8.3) 1/12 (8.3) 
Fem-TG 5/12 (42) 1/12 (8.3) 0/12 7/9 (78) 0/12 0/12 4/5 (80) 0/5 1/5 (20) 
Male-WT 4/7 (57) 1/7 (14) 0/7 8/8 (100) 3/8 (38) 0/8 14/14 (100) 8/14 (57) 2/14 (14) 
Male-TG 11/12 (92) 2/12 (17) 0/12 11/11 (100) 7/11 (64) 2/11 (18) 12/12 (100) 9/12 (75) 5/12 (42) 
Group  25 weeks after DEN No. (%) of mice with lesions/no. of mice examined
 
   35 weeks after DEN No. (%) of mice with lesions/no. of mice examined
 
   49 weeks after DEN No. (%) of mice with lesions/no. of mice examined
 
  

 
F
 
HA
 
HCC
 
F
 
HA
 
HCC
 
F
 
HA
 
HCC
 
Fem-WT 3/10 (30) 0/10 0/10 2/4 (50) 0/4 0/4 12/12 (100) 1/12 (8.3) 1/12 (8.3) 
Fem-TG 5/12 (42) 1/12 (8.3) 0/12 7/9 (78) 0/12 0/12 4/5 (80) 0/5 1/5 (20) 
Male-WT 4/7 (57) 1/7 (14) 0/7 8/8 (100) 3/8 (38) 0/8 14/14 (100) 8/14 (57) 2/14 (14) 
Male-TG 11/12 (92) 2/12 (17) 0/12 11/11 (100) 7/11 (64) 2/11 (18) 12/12 (100) 9/12 (75) 5/12 (42) 

F, pre-neoplastic focus; HA, hepatocellular adenoma; HCC, hepatocellular carcinoma; WT, wild-type; TG, transgenic.

In contrast, female transgenic and wild-type mice developed much fewer pre-neoplastic or neoplastic lesions than the males and there was no significant difference between the two genotypes in terms of incidence of liver tumors ( Tables I and II ).

The ANOVA of the multiplicity data, considering genotype and time (weeks) as sources of variation, indicated that both contributed significantly to the observed variance.

Discussion

In this study, to explore the role of Cx32 in growth control of hepatocytes in vivo , we generated transgenic mouse lines in which the mutant Cx32 V139 M gene, driven by the albumin promoter, is expressed only in the liver. This approach avoids systemic effects caused by gene disruption, as in gene knockout models. We have reported previously that the introduced mutant abolishes GJIC exerted by wild-type Cx32 in HeLa cells in a dominant-negative manner, probably due to formation of non-functional heteromeric connexons composed of both wild-type and mutant Cx32 proteins.

Although, as expected, GJIC capacity was reduced in the liver of the transgenic mice ( Figure 4B ), the extent of reduction was much more modest than that seen in HeLa cells ( 14 ). Initially, we expected that this mutant would interfere with both Cx32 and Cx26 co-expressed in a hepatocyte because these two types of connexin can form heteromers in vitro ( 37 ). However, our results suggest rather that Cx26 was not affected by the mutant Cx32 V139 M in vivo ( Figure 2A and B ) and that the hepatocytes of the transgenic mice can maintain a moderate level of GJIC due to Cx26 and residual Cx32 that have escaped from the dominant-negative effect of the mutant Cx32. The dominant-negative effect of the V139 M mutant of Cx32 may be much less efficient in vivo than in HeLa cells in vitro .

The most remarkable finding in this study was that reduced GJIC in the transgenic liver delayed regeneration after partial hepatectomy ( Figure 5A ). Although the mechanism of triggering of compensatory regeneration of the liver remains controversial and largely unknown, it has been proposed that mitogenic stimuli for hepatocytes appear as humoral factors in the blood shortly after surgery and initiate a chain of events necessary for proliferation of hepatocytes ( 38 ). Such factors exert their effects first in periportal hepatocytes, and spread to other zones in the hepatic acinus in a synchronized manner. Correspondingly, hepatocytes start to proliferate in periportal areas, followed by pericentral areas ( 39 ). Thus, such spatial and time-dependent changes during the triggering of liver regeneration may be involving signal transmission at extra-, intra- and intercellular levels. It has indeed been suggested that gap junctions play a role in liver regeneration by spreading second messengers that appear immediately after partial hepatectomy ( 40 ). Reduced GJIC in the liver of the transgenic mice could be responsible for inefficient diffusion of such factors, resulting in the delayed entry of hepatocytes to S-phase.

Our findings with transgenic mice after partial hepatectomy are specific to compensatory regeneration, as treatment with TCPOBOP induced a mitogenic response and primary hyperplasia in the liver of both the transgenic and wild-type mice in a similar manner. TCPOBOP is a liver tumor promoter and categorized as a peroxisome proliferator, which can directly stimulate individual hepatocytes. Thus, it appears that the events modulated by TCPOBOP are downstream to those regulated by gap junctions.

Another important finding in this study was an increased susceptibility of transgenic mice to chemical hepatocarcinogenesis. The transgenic mice displayed not only a higher multiplicity of liver lesions but also a higher incidence of both hepatocellular adenomas and hepatocellular carcinomas, suggesting that reduced GJIC in the transgenic liver facilitated a promotion and progression process during carcinogenesis. This observation is quite consistent with the established idea that cells having no or low GJIC can escape from growth control and gain a growth advantage for clonal expansion ( 17 ).

Our results suggest that GJIC acts positively on cell proliferation during liver regeneration and negatively on tumor development and progression. We have no clear explanation for this apparent contradiction. However, the long-standing idea that gap junctions function to maintain homeostasis of the cellular society suggests that gap junctions transmit certain molecules that are essential for maintaining normal liver volume. Thus, in response to the lack of liver mass after partial hepatectomy, gap junctions play a role in promoting cell growth for regeneration, and once the liver has recovered its original size, the gap junctions may switch their role to normalization of excessive cell growth.

Similarly to our transgenic mice, Cx32-deficient mice manifest higher susceptibility to chemical hepatocarcinogenesis ( 41 , 42 ). However, unlike our transgenic mice, they display excessive glycogen storage in the liver and develop neuropathy, similar to X-linked Charcot-Marie-Tooth disease, due to impaired conductance of sympathetic nerves and atrophy of peripheral nerves, respectively. In addition, mice lacking Cx32 show increased incidence of spontaneous liver tumors and elevated BrdU incorporation even in the quiescent liver. This notable difference in proliferative status of the liver between Cx32 knockout mice and our transgenic mice could be due partly to a significantly reduced amount of Cx26 protein expressed in the liver of Cx32-deficient mice but not in our transgenic mice ( Figures 1D and 2A and B ). Cell growth regulation and tumor-suppressive potency of Cx26 have been shown both in vitro and in vivo ( 43 , 44 ). These results and our own observations reported here are consistent with the idea that impaired cell coupling disturbs the capacity of a tissue to maintain homeostasis, particularly in response to external cell growth stimuli. Considering that connexins are essential in maintaining homeostasis of various tissues, it is not surprising that connexin gene deletion in mice and connexin gene mutations in humans result in various diseases.

Although numerous studies have provided evidence for growth and tumor suppression by connexins in vitro , our experiments in this study in vivo rather suggest that connexin does not act simply as a growth suppressor. What we and others have so far observed is that the presence of functional gap junctions nearly always leads cells towards necessary and sufficient growth to maintain proper function of tissues. As the cell lines often used for in vitro studies are immortalized or even already transformed, it is probable that the function of gap junctions in such cells is to bring the elevated levels of growth down to normal. We now think that gap junctions play a role not only in growth suppression but also in coordinated cell growth, i.e. both positively and negatively to achieve a normal growth rate.

5
To whom correspondence should be addressed Email: yasu@med.akita-u.ac.jp

We are very grateful to Ms Martine Croizy (Institut Curie de Recherche, Orsay, France) for providing us with TCPOBOP; to Prof. Klaus Willecke (Institut für Genetik, Universität Bonn, Germany) for providing us with their Cx32-deficient mice; to Ms Colette Piccoli, Nicole Martel, Dominique Galendo, Nicole Lyandrat and Mireille Laval (International Agency for Research on Cancer, Lyon, France) for technical assistance, and to Dr John Cheney for editing the manuscript. M.L.Z.D. was recipient of a fellowship (97-03207-0) from the Fundação de Amparo à Pesquisa do Estado de São Paulo, São Paulo, SP, Brazil. This work was partly supported by a grant (R01-CA40534) from the US National Institutes of Health (to H.Y.).

References

1.
Loewenstein,W.R. (
1981
) Junctional intercellular communication: the cell-to-cell membrane channel.
Physiol. Rev.
  ,
61
,
829
–913.
2.
Bruzzone,R., White,T.W. and Paul,D.L. (
1996
) Connections with connexins: the molecular basis of direct intercellular signaling.
Eur. J. Biochem.
  ,
238
,
1
–27.
3.
Goldberg,G.S., Lampe,P.D. and Nicholson,B.J. (
1999
) Selective transfer of endogenous metabolites through gap junctions composed of different connexins.
Nature Cell Biol.
  ,
1
,
457
–459.
4.
Elfgang,C., Eckert,R., Lichtenberg-Frate,H., Butterweck,A., Traub,O., Klein,R.A., Hulser,D.F. and Willecke,K. (
1995
) Specific permeability and selective formation of gap junction channels in connexin-transfected HeLa cells.
J. Cell Biol.
  ,
129
,
805
–817.
5.
Kelsell,D.P., Dunlop,J., Stevens,H.P., Lench,N.J., Liang,J.N., Parry,G., Mueller,R.F. and Leigh,I.M. (
1997
) Connexin 26 mutations in hereditary non-syndromic sensorineural deafness.
Nature
  ,
387
,
80
–83.
6.
Richard,G., White,T.W., Smith,L.E., Bailey,R.A., Compton,J.G., Paul,D.L. and Bale,S.J. (
1998
) Functional defects of Cx26 resulting from a heterozygous missense mutation in a family with dominant deaf-mutism and palmoplantar keratoderma.
Hum. Genet.
  ,
103
,
393
–399.
7.
Richard,G., Smith,L.E., Bailey,R.A., Itin,P., Hohl,D., Epstein,E.H.Jr, DiGiovanna,J.J., Compton,J.G. and Bale,S.J. (
1998
) Mutations in the human connexin gene GJB3 cause erythrokeratodermia variabilis.
Nature Genet.
  ,
20
,
366
–369.
8.
Xia,J.H., Liu,C.Y., Tang,B.S. et al . (
1998
) Mutations in the gene encoding gap junction protein beta-3 associated with autosomal dominant hearing impairment.
Nature Genet.
  ,
20
,
370
–373.
9.
Bergoffen,J., Scherer,S.S., Wang,S., Oronzi Scott,M., Bone,L.J., Paul,D.L., Chen,K., Lensch,M.W., Chance,P.F. and Fischbeck,K.H. (
1993
) Connexin mutations in X-linked Charcot-Marie-Tooth disease.
Science
  ,
262
,
2039
–2042.
10.
Shiels,A., Mackay,D., Ionides,A., Berry,V., Moore,A. and Bhattacharya,S. (
1998
) A missense mutation in the human connexin50 gene (GJA8) underlies autosomal dominant ‘zonular pulverulent’ cataract, on chromosome 1q.
Am. J. Hum. Genet.
  ,
62
,
526
–532.
11.
White,T.W., Deans,M.R., Kelsell,D.P. and Paul,D.L. (
1998
) Connexin mutations in deafness.
Nature
  ,
394
,
630
–631.
12.
Duflot-Dancer,A., Mesnil,M. and Yamasaki,H. (
1997
) Dominant-negative abrogation of connexin-mediated cell growth control by mutant connexin genes.
Oncogene
  ,
15
,
2151
–2158.
13.
Krutovskikh,V.A., Yamasaki,H., Tsuda,H. and Asamoto,M. (
1998
) Inhibition of intrinsic gap-junction intercellular communication and enhancement of tumorigenicity of the rat bladder carcinoma cell line BC31 by a dominant-negative connexin 43 mutant.
Mol. Carcinogen.
  ,
23
,
254
–261.
14.
Omori,Y., Mesnil,M. and Yamasaki,H. (
1996
) Connexin32 mutations from X-linked Charcot-Marie-Tooth disease patients: functional defects and dominant negative effects.
Mol. Biol. Cell
  ,
7
,
907
–916.
15.
Omori,Y. and Yamasaki,H. (
1998
) Mutated connexin43 proteins inhibit rat glioma cell growth suppression mediated by wild-type connexin43 in a dominant-negative manner.
Int. J. Cancer
  ,
78
,
446
–453.
16.
Trosko,J.E. and Ruch,R.J. (
1998
) Cell-cell communication in carcinogenesis.
Front. Biosci.
  ,
3
,
d208
–d236.
17.
Yamasaki,H. and Naus,C.C.G. (
1996
) Role of connexin genes in growth control.
Carcinogenesis
  ,
17
,
1199
–1213.
18.
Dagli,M.L.Z., Ward,J.M., Merlino,G., Enomoto,A. and Maronpot,R.R. (
2000
) Hepatic pathology in genetically engineered mice. In Ward,J.M., Mahler,J.F., Maronpot,R.R. and Sundberg,J.P. (eds), Pathology of Genetically Engineered Mice . Iowa State University Press, Ames, IA, pp. 317–335.
19.
Fausto,N. (
1999
) Lessons from genetically engineered animal models. V. Knocking out genes to study liver regeneration: present and future.
Am. J. Physiol.
  ,
277
,
G917
–921.
20.
Gabriel,H.D., Jung,D., Butzler,C., Temme,A., Traub,O., Winterhager,E. and Willecke,K. (
1998
) Transplacental uptake of glucose is decreased in embryonic lethal connexin26-deficient mice.
J. Cell Biol.
  ,
140
,
1453
–1461.
21.
Reaume,A.G., de Sousa,P.A., Kulkarni,S., Langille,B.L., Zhu,D., Davies,T.C., Juneja,S.C., Kidder,G.M. and Rossant,J. (
1995
) Cardiac malformation in neonatal mice lacking connexin43.
Science
  ,
267
,
1831
–1834.
22.
Shiota,G., Rhoads,D.B., Wang,T.C., Nakamura,T. and Schmidt,E.V. (
1992
) Hepatocyte growth factor inhibits growth of hepatocellular carcinoma cells.
Proc. Natl Acad. Sci. USA
  ,
89
,
373
–377.
23.
Takeda,A., Kanoh,M., Shimazu,T. and Takeuchi,N. (
1988
) Monoclonal antibodies recognizing different epitopes of the 27-kDa gap junction protein from rat liver.
J. Biochem.
  ,
104
,
901
–907.
24.
Krutovskikh,V., Mazzoleni,G., Mironov,N., Omori,Y., Aguelon,A.-M., Mesnil,M., Berger,F., Partensky,C. and Yamasaki,H. (
1994
) Altered homologous and heterologous gap-junctional intercellular communication in primary human liver tumors associated with aberrant protein localization but not gene mutation of connexin 32.
Int. J. Cancer
  ,
56
,
87
–94.
25.
Merchenthaler,I., Stankovics,J. and Gallyas,F. (
1989
) A highly sensitive one-step method for silver intensification of the nickel-diaminobenzidine endproduct of peroxidase reaction.
J. Histochem. Cytochem.
  ,
37
,
1563
–1565.
26.
Momiyama,M., Omori,Y., Ishizaki,Y., Nishikawa,Y., Tokairin,T., Ogawa,J. and Enomoto,K. (
2003
) Connexin26-mediated gap junctional communication reverses the malignant phenotype of MCF-7 breast cancer cells.
Cancer Sci.
  ,
94
,
501
–507.
27.
Krutovskikh,V.A., Oyamada,M. and Yamasaki,H. (
1991
) Sequential changes of gap-junctional intercellular communications during multistage rat liver carcinogenesis: direct measurement of communication in vivo .
Carcinogenesis
  ,
12
,
1701
–1706.
28.
Higgins,G.M. and Anderson,R.M. (
1931
) Experimental pathology of the liver. Restoration of the liver of the white rat following partial surgical removal.
Arch. Path.
  ,
12
,
186
–202.
29.
Manenti,G., Dragani,T.A. and Della Porta,G. (
1987
) Effects of phenobarbital and 1,4-bis[2-(3,5-dichloropyridyloxy)]benzene on differentiated functions in mouse liver.
Chem. Biol. Interact.
  ,
64
,
83
–92.
30.
Poland,A., Mak,I., Glover,E., Boatman,R.J., Ebetino,F.H. and Kende,A.S. (
1980
) 1,4-Bis[2-(3,5-dichloropyridyloxy)]benzene, a potent phenobarbital-like inducer of microsomal monooxygenase activity.
Mol. Pharmacol.
  ,
18
,
571
–580.
31.
Vesselinovitch,S.D. and Mihailovich,N. (
1983
) Kinetics of diethylnitrosamine hepatocarcinogenesis in the infant mouse.
Cancer Res.
  ,
43
,
4253
–4259.
32.
Harada,T., Enomoto,A., Boorman,G.A. and Maronpot,R.R. (
1999
) Liver and gallbladder. In Maronpot,R., Boorman,G.A. and Gaul,B. (eds), Pathology of the Mouse . Cache River Press, Vienna, IL, pp. 119–183.
33.
Nelles,E., Butzler,C., Jung,D. et al . (
1996
) Defective propagation of signals generated by sympathetic nerve stimulation in the liver of connexin32-deficient mice.
Proc. Natl Acad. Sci. USA
  ,
93
,
9565
–9570.
34.
Deschenes,S.M., Walcott,J.L., Wexler,T.L., Scherer,S.S. and Fischbeck,K.H. (
1997
) Altered trafficking of mutant connexin32.
J. Neurosci.
  ,
17
,
9077
–9084.
35.
Temme,A., Buchmann,A., Gabriel,H.D., Nelles,E., Schwarz,M. and Willecke,K. (
1997
) High incidence of spontaneous and chemically induced liver tumors in mice deficient for connexin32.
Curr. Biol.
  ,
7
,
713
–716.
36.
Kren,B.T., Kumar,N.M., Wang,S.Q., Gilula,N.B. and Steer,C.J. (
1993
) Differential regulation of multiple gap junction transcripts and proteins during rat liver regeneration.
J. Cell Biol.
  ,
123
,
707
–718.
37.
Stauffer,K.A. (
1995
) The gap junction proteins β 1 -connexin (connexin-32) and β 2 -connexin (connexin-26) can form heteromeric hemichannels.
J. Biol. Chem.
  ,
270
,
6768
–6772.
38.
Michalopoulos,G.K. and DeFrances,M.C. (
1997
) Liver regeneration.
Science
  ,
276
,
60
–66.
39.
Baratta,B., Rizzoli,R., Galliani,I., Vitale,M., Rizzi,E., Matteucci,A., Galanzi,A., Zamai,L. and Mazzotti,G. (
1996
) Early events of liver regeneration in rats: a multiparametric analysis.
Histochem. Cell Biol.
  ,
105
,
61
–69.
40.
Neveu,M.J., Hully,J.R., Babcock,K.L., Vaughan,J., Hertzberg,E.L., Nicholson,B.J., Paul,D.L. and Pitot,H.C. (
1995
) Proliferation-associated differences in the spatial and temporal expression of gap junction genes in rat liver.
Hepatology
  ,
22
,
202
–212.
41.
Moennikes,O., Buchmann,A., Ott,T., Willecke,K. and Schwarz,M. (
1999
) The effect of connexin32 null mutation on hepatocarcinogenesis in different mouse strains.
Carcinogenesis
  ,
20
,
1379
–1382.
42.
Evert,M., Ott,T., Temme,A., Willecke,K. and Dombrowski,F. (
2002
) Morphology and morphometric investigation of hepatocellular preneoplastic lesions and neoplasms in connexin32-deficient mice.
Carcinogenesis
  ,
23
,
697
–703.
43.
Mesnil,M., Krutovskikh,V., Piccoli,C., Elfgang,C., Traub,O., Willecke,K. and Yamasaki,H. (
1995
) Negative growth control of HeLa cells by connexin genes: connexin species specificity.
Cancer Res.
  ,
55
,
629
–639.
44.
Lee,S.W., Tomasetto,C. and Sager,R. (
1991
) Positive selection of candidate tumor-suppressor genes by subtractive hybridization.
Proc. Natl Acad. Sci. USA
  ,
88
,
2825
–2829.

Author notes

1Unit of Multistage Carcinogenesis, International Agency for Research on Cancer, 150, cours Albert-Thomas, 69372 Lyon Cedex 08, France

Present addresses: 2Department of Pathology, Faculty of Veterinary Medicine and Zootechny, University of São Paulo, CEP 05508-900, S. Paulo-SP, Brazil, 3School of Science and Technology, Kwansei Gakuin University, Sanda 669-1337, Japan and 4Department of Pathology, Akita University School of Medicine, Akita, 010-8543, Japan