Connexins are subunits of gap junction channels, which allow direct transfer of ions, secondary messenger molecules, and other metabolites between contacting cells. Gap junctions are believed to be involved in tissue homeostasis, embryonic development, and control of cell proliferation. Several studies have shown that cell damage signals are transmitted through gap junctions when cells are irradiated or when cells bearing the herpes simplex virus-thymidine kinase (HSV-TK) gene are treated with ganciclovir. We established 2 lines of transgenic rats with a dominant-negative mutant of connexin 32 gene under control of the albumin promoter. In the livers of transgenic rats, membrane localization of normal endogenous connexin 32 protein is disturbed, and gap junction capacity measured by scrape dye-transfer assay in vivo is markedly decreased when compared with wild-type rats. The present investigation concerned susceptibility to the liver-toxic substances D-galactosamine and carbon tetrachloride. These toxicants induced massive liver cell death and elevated serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels in the wild-type rats; however, much fewer liver cells were damaged and serum enzyme elevation was much lower in the transgenic rats. In conclusion, gap junctional intercellular communication (GJIC) plays an important role in toxic effects of chemicals; damage or death signals may pass through gap junctions in the rat liver in vivo. (HEPATOLOGY 2004;40:205–210.)
Gap junctional channels of contact cells allow intercellular exchange of small molecules, such as ions, second messengers, and cellular metabolites.1–4 It is believed that gap junctional intercellular communication (GJIC) plays important roles in tissue homeostasis,3 embryonic development,5 and carcinogenesis.6, 7 “Metabolic cooperation” involves the transfer of a lethal metabolite formed in a cell containing the normal hypoxanthine-guanine phosphoribosyltransferase gene to a cell containing a mutant for this gene.8 Many reports have shown that some of the bystander effects of herpes simplex virus-thymidine kinase (HSV-TK) gene therapies for cancer cells are dependent on GJIC.9–14 Because the therapeutic gene HSV-TK cannot be easily introduced into the entire cell population of a tumor, successful eradication of tumors depends on this bystander effect. GJIC is directly involved in the transfer of the toxic metabolites of ganciclovir, which pass directly from HSV-TK-expressing cells to surrounding cells that do not express it. Similar transmission of damage signals through the gap junction has been reported from irradiated to nonirradiated cells.15, 16 However, involvement of toxic chemicals in in vivo cell damage remains unclear.
To study mechanisms of liver damage by chemicals, rat models are often used.17–19 In this study, transgenic rats carrying a dominant-negative mutant of connexin 32 (Cx32), the major gap junction protein in liver,20 were investigated for susceptibility to D-galactosamine and carbon tetrachloride in comparison to wild-type littermates to determine whether the loss of GJIC impacts toxicity.
The rat Cx32 complementary DNA, under control of the albumin gene promoter,21, 22 and an enhancer cloned in the pGEM-7 vector were gifts from Dr. H. Yamasaki, Kwansei Gakuin University. Thirty-six nucleotides corresponding to amino acids 113 to 124 in the intracytoplasmic loop of Cx32 (5′-GAC CCC CTT CAC CTG GAA GAG GTA AAG AGG CAC AAG-3′) were deleted using site-direct mutagenesis technique (ExSite PCR-Based Site Directed Mutagenesis Kit, Stratagene, La Jolla, CA). To tag the transgene, 18 nucleotides for 6xHis (5′-CAT CAT CAC CAT CAC CAT-3′) were inserted just before the stop codon (TGA) using the same kit. The transgene was removed from the vector using ApaI and MluI and purified using QIAGEN columns (QIAGEN, Tokyo, Japan).
Production and Screening of the Transgenic Rats.
Generation of transgenic rats was performed by SLC Inc. (Shizuoka, Japan), using eggs from Sprague-Dawley rats. DNA isolation from rat tails, and the polymerase chain reaction (PCR)-based screening assay was performed as previously described.23, 24 Sequences of the PCR primers were 5′-AAC GTG GCG CAG GTG GTG TA-3′ and 5′-ATG GTG ATG GTG ATG ATG GC-3′, which are located on the coding region and the 6xHis tag for the transgene. Southern blotting analysis for Cx32 was also performed using the SmaI fragment of Cx32 cDNA after digestion of genomic DNA with BamHI. Heterologous transgenic males for the studies were routinely obtained by mating heterologous transgenic males and wild-type Sprague-Dawley females (SLC Inc.).
Expression of the Transgene.
Total RNA was extracted with ISOGEN (Nippon Gene, Tokyo Japan), followed by DNase treatment (Clontech, Palo Alto, CA). Reverse transcriptase-PCRs (RT-PCR) were performed using an RNA LA PCR Kit (AMV) (Version 2.1; Takara, Otsu, Japan) with the primers 5′-AGA GCG TGT GGG GTG ATG AGA AGT C-3′ and 5′-TAG CCC GGG TAG AGC AGA TAG AA-3′, which amplify a Cx32 fragment containing the deleted part of the transgene. PCR products were separated in acrylamide gels (GenePhor DNA Separation System; Amersham Pharmacia Biotech, Buckinghamshire, England) and silver-stained.
Immunohistochemical Staining of Cx32.
Immediately upon sacrifice, livers from the transgenic and wild-type rats were excised and slices 4 to 5 mm thick were cut with a razor blade and immersed in isopentane precooled to approximately −130°C in a liquid nitrogen bath. They were stored at −75°C in a deep freezer until use. Frozen sections were cut at 5 μm and fixed in cold acetone. A polyclonal antibody against Cx32 (provided by V. Krutovskikh, International Agency for Research on Cancer, Lyon, France), which recognizes the deleted part of the transgene, and a monoclonal antibody against connexin 26 (Cx26) (Zymed Laboratories, Inc., South San Francisco, CA), biotin- conjugated anti-rabbit or anti-mouse immunoglobulin G, and FITC-labeled streptavidin (Vector Laboratories Inc., Burlingame, CA) were used to visualize the endogenous Cx32 and Cx26 proteins under fluorescence microscopy (Olympus AX-70, Tokyo, Japan).
Western Blotting for Cx32.
The liver tissues were lysed in radio immunoprecipitation assay buffer (1% Triton X-100, 0.1% SDS, 0.5% sodium deoxycholate, 0.1 mol NaCl, 10 mmol sodium phosphate[pH 7.4]) with a protease inhibitor cocktail (Roche Diagnostics, Basel, Switzerland); protein concentrations were determined by the Bradford method using protein assay kits (Bio-Rad Laboratories, Hercules, CA). Ten μg samples were mixed with SDS sample buffer (50 mmol Tris-HCl(pH 6.8), 100 mmol dithiothreitol, 2% SDS, 0.1% bromophenol blue, 10% glycerol), separated by 12% SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes (Amersham Bioscience, Piscataway, NJ). Membranes were incubated with the antibody in PBST (phosphate-buffered saline [PBS], 0.1% Tween 20) containing 5% nonfat dry milk overnight at 4°C. This was followed by incubation with secondary antibodies conjugated with peroxidase (Amersham Bioscience, Tokyo, Japan) at a dilution of 1:2000 in PBST containing 5% nonfat dry milk. The bands were visualized on x-ray films using an ECL™-plus detection kit (Amersham Bioscience).
Measurement of GJIC Capacity in Liver In Vivo.
This was performed with minor modifications according to the method published previously.25 Briefly, 5 mm-thick liver slices were cut and 3 to 4 incisions (1 mm depth) were made with a blade, followed by the dropping of 0.05% Lucifer Yellow in PBS on the liver slices. After 3 minutes, the slices were washed in PBS 3 times, then embedded and frozen in the Tissue-Tek® Optmal cutting temperature (OCT) compound 4583 (Sakura Finetechnical Co. Ltd., Tokyo, Japan). Frozen sections 6 μm thick were made and photographed by fluorescence microscope. Spread of the dye was measured using an image analyzer (IPAP-WIN, Sumika Technos Co., Osaka, Japan).
At the age of 8 weeks, male transgenic and wild-type rats received single intraperitoneal injections of 300 mg/kg body weight of D-galactosamine or 0.5 mL/kg body weight of 50% carbon tetrachloride solution in corn oil. Control rats received the vehicle. After 24 hours, serum was collected for measurement of alanine aminotransferase (ALT) and aspartate aminotransferase (AST), and rats were killed under ether anesthesia. Livers were immediately removed, weighed, and cut into slices of 2 to 3-mm thickness, fixed in buffered formalin, and then processed for embedding in paraffin for histological evaluation after hematoxylin-eosin staining.
Measurement of ALT and AST.
Serum contents of ALT and AST were measured using a Hitachi-Biochemical Automatic Analyzer 7070 (Hitachi Ltd., Tokyo, Japan).
Four male and 1 female founder rats were obtained. Four animals transmitted the transgene to the next generation, and 2 expressed messenger RNA (mRNA) for the transgene in liver. Immunohistochemical staining of endogenous Cx32 revealed that livers from males of one line lacked positive spots of gap junction plaques except in limited areas (Fig. 1), and livers of another line demonstrated a patchy distribution of hepatocytes with gap junction spots. Therefore, male transgenic rats from the line that almost lacked Cx32 immunostaining in the liver were used for the experiments. For unknown reasons, female transgenic rats were found to express very little mRNA of the transgene regardless of a high copy number in the genomic DNA; these rats also demonstrated a similar Cx32 immunostaining pattern to that of the wild type.
Southern blotting analysis of genomic DNA from tails of the second generation of the transgenic founder rat revealed 2 sublines: 1 with high copy numbers of the transgene (≈50 copies) and the other with only approximately 5 copies. These transgenic rats were designated Cx32ΔTg-high and Cx32ΔTg-low, respectively.
RT-PCRs using primers flanking the Cx32 deleted lesion were performed to amplify both endogenous and transgenic Cx32 mRNA. This analysis revealed that livers of Cx32ΔTg-high expressed the transgene more abundantly than those of Cx32ΔTg-low, in which the expression level of the transgene was equivalent to endogenous Cx32 (Fig. 2).
The scrape dye-loading method in vivo revealed that livers of Cx32ΔTg-high and Cx32ΔTg-low had significantly reduced GJIC capacity. In wild-type rat livers, the dye spread quite well, but livers of both transgenic rats had limited capacity for communication, resulting in little dye spread (Fig. 3).
Western blotting for Cx32 detected an approximately 26-kd band in livers of both wild-type and transgenic rats in which normal gap junction spots were not visible by immunostaining using the same antibody (Fig. 4).
On treatment with D-galactosamine or carbon tetrachloride, ALT and AST were elevated dramatically, but the levels were much lower in the transgenic rats compared to the wild-type rats (Table 1).
Table 1. Serum AST and ALT Levels After Treatments of D-Galactosamine and Carbon Tetrachloride (CCl4) in Wild-type and Transgenic Rats
No. of Rats
Statistical significance (compared to wild-type group) determined by Fisher protected least significant difference test.
Histopathological examination revealed that, after treatment with D-galactosamine, massive necrosis was observed in the areas around the central veins of the liver in the wild-type rats, but only single cell necrosis was evident in the transgenic rats. In the wild-type rats receiving carbon tetrachloride, liver cell necrosis and ballooning were apparent; in the transgenic livers, no obvious necrosis was observed and fewer ballooning cells were apparent (Fig. 5).
The present study provided strong evidence that liver toxicity in the rat is strongly dependent on GJIC. Thus, with both D-galactosamine and carbon tetrachloride, the levels of liver damage were much lower in our transgenic rats than in their wild-type counterparts.
Cx32 is the main gap junction protein in hepatocytes,20 and a study using Cx32 knockout mice indicated that gap junctions are involved in intercellular propagation not only of nervous but also of hormonal signals from periportal to perivenous hepatocytes.26 The Cx32-deficient mice have high incidences of spontaneous and chemically induced liver tumors.27 While high sensitivity to diethylnitrosamine induced hepatocarcinogenesis of these mice was established,28–30 their response to hepatotoxic substances is not clear. Recently, transgenic mice using a similar construct were reported.31 These mice carried the dominant-negative mutant Cx32 V139M gene. Their GJIC capacity was reduced in the liver, and they showed high susceptibility to chemical hepatocarcinogenesis.31 In human cases, mutations affecting the Cx32 gene are associated with the X-linked form of the hereditary peripheral neuropathy Charcot-Marie-Tooth disease (CMTX).32 Cx32-deficient mice develop a late-onset progressive peripheral neuropathy with abnormalities comparable to those associated with CMTX, thus providing proof of the critical role of Cx32 in the maintenance of peripheral nerve myelin and of an animal model for CMTX.33 However, abnormality of liver function or high susceptibility to liver tumor formation in patients have not been reported so far. Therefore, roles of gap junctions in liver are still largely unknown.
In this study, we have established a transgenic rat model with a reduced GJIC capacity, using a dominant-negative mutant of Cx32 under control of the albumin promoter. The transgene features deletion of 12 amino acids of the internal loop of Cx32. The reasons for choosing this mutant are that similar deletion of the gene was found in the CMTX patient34 and corresponding deletion of the connexin 43 gene showed a strong dominant-negative effect in rat-bladder carcinoma cells.35
Immunohistochemical and Western blotting analysis of endogenous Cx32, using a rabbit polyclonal antibody against the region deleted in the dominant-negative mutant Cx32, revealed abrogation of normal gap junctional plaque formation in cell membranes of hepatocytes of the transgenic rats. The polyclonal antibody we used in this study recognizes amino acids 98 to 124 in Cx32. Because these are deleted in the transgene (missing amino acids 113-124), the transgene (mutated exogenous Cx32) could not be detected by this antibody. Despite the lack of Cx32 spots, a band for endogenous Cx32 was detected by Western blotting in the transgenic rat livers, indicating that normal Cx32 exists in the cells but is not able to form gap junctions. In Cx32 immnofluorescent staining, detection was only possible when gap junctions made plaques on the plasma membrane. When connexons are diffusely present on membrane or in cytoplasma, they may not be visible. Localization of another major gap junction protein in liver, Cx2636 was also disrupted in livers of transgenic rats (Fig. 6), indicating close interaction between Cx26 and Cx32 proteins in liver.37, 38 Similar Cx26 abnormality has been observed in Cx32-deficient mouse livers.26
The present results are in line with findings for HSV-TK gene therapy, in which cancer cell death is caused by bystander effects.9–14 It has been shown that tumors regress when tumor cells are transfected with thymidine kinase gene from herpes simplex virus by retroviral vectors and then exposed to ganciclovir, a nucleoside analogue. Ganciclovir is phosphorylated by HSV-TK and incorporated into DNA, causing a DNA replication stop and cell death. In series of these experiments, it was found that cells with HSV-TK were able to cause cell death of neighboring cells without this enzyme. This bystander effect appears due to transfer of phosphorylated ganciclovir molecules from HSV-TK+ to HSV-TK− cells through gap junctions. Bystander effects through gap junctions may also be involved in cell damage by irradiation.15, 16 In the present study, we clearly showed that gap junctions also contribute to toxic effects.
Carbon tetrachloride requires enzymatic bioactivation by cytochrome P-450 (CYP)2E1 for its biochemical effects.19, 39 We have investigated CYP2E1 mRNA expression level in livers of wild-type and transgenic rats but have found no significant difference (data not shown). D-Galactosamine is a hepatotoxin whose toxicity has been related to a depletion of cellular uridine triphosphate (UTP) and other uridine nucleotides induced by this agent.17, 40 Mechanisms of the toxic effects and morphological changes of damaged livers by these 2 compounds are completely different. However, the transgenic rats are resistant to both, indicating that damage signals pass through gap junctions of hepatocytes independent of their source. It is of interest in this context that rat hepatocyte nodules have no or reduced connexin expression and GJIC capacity.41, 42 These nodules are resistant to the necrogenic effects of D-galactosamine.43
In summary, the present study clearly shows that GJIC plays important roles in toxic effects of chemicals and that damage or death signals can pass through gap junctions in rat liver in vivo. More importantly, these findings should stimulate research on the “stem cell” or “dedifferentiation” theories of hepatocellular carcinogenesis.
The authors thank Dr. Malcolm A. Moore for his kind linguistic advice during preparation of the manuscript.