Reappearance and long-term maintenance of connexin32 in proliferated adult rat hepatocytes: use of serum-free L-15 medium supplemented with EGF and DMSO

Gap junctions are intercellular membrane channels that link neighboring cells and that mediate reciprocal exchanges of small molecules and ions, including second messengers such as cAMP, inositol triphosphate and Ca²⁺, between adjacent cells in contact (Sáez et al., 1986, 1989a; Spray et al., 1986, 1988). Gap junctional intercellular communication (GJIC) is thought to play a crucial role in the maintenance of homeostasis, morphogenesis, cell differentiation and growth control in multicellular organisms (Loewenstein, 1979; Bennett et al., 1991; Lang et al., 1991; Mesnil and Yamasaki, 1993). Gap junctions are composed of proteins called ‘connexins’, which have different molecular masses in different tissues (Haefliger et al., 1992). In rat hepatocytes, two homologous connexin molecules have been identified (Kumar and Gilula, 1986; Nicholson et al., 1987): connexin32 (Cx32) as a major component and connexin26 (Cx26) as a minor component.

In primary cultures of rat hepatocytes, the maintenance and induction of gap junctions have been attempted by many researchers, but only a few laboratories have reported success in the short-term maintenance of Cx32 (Sáez et al., 1986; Spray et al., 1987; Traub et al., 1989; Musat et al., 1993). Spray et al. (1987) showed that extracellular matrix components such as proteoglycans and glycosaminoglycans can induce the synthesis and expression of gap junctions. Recently, Musat et al. (1993) reported that Matrigel in culture medium is effective in maintaining the immunoreactivity to Cx32 for one week. A coculture system (adult rat hepatocytes and Balb/c 3T3 cells) was also shown to be useful for maintaining gap junctions (Mesnil et al., 1993). However, to our knowledge, the long-term maintenance and the reinduction of gap junctional proteins in primary cultures of pure hepatocytes have not been reported.

Dimethyl sulfoxide (DMSO) is well known to induce differentiation in many cell lines (Higgins and Borenfreund, 1980; Higgins et al., 1983). For the purpose of maintaining the dif-ferentiated functions, 2% DMSO is also used in primary cultures of rat hepatocytes (Baribault and Marceau, 1986; Isom et al., 1985, 1987; McGowan, 1988; Kost and Michalopoulos, 1991). Isom et al. (1985) first reported that adult rat hepatocytes in a chemically defined medium supplemented with epidermal growth factor (EGF) and 2% DMSO survived much longer and synthesized albumin much better than cells cultured in standard serum-free medium. Recently, we established a culture system in which, by adding 2% DMSO to the culture medium after the hepatocytes proliferate, the cells are able to recover differentiated functions such as albumin and transferrin secretion, and glucose-6-phosphatase activity (Mitaka et al., 1993). In the present study, we show that Cx32 can reappear in primary cultures of rat hepatocytes and that the gap junctional structures can be well maintained together with extensive GJIC for more than one month.

Male Sprague-Dawley rats (Shizuoka Laboratory Animal Center, Hamamatsu, Japan) weighing about 300-400 g were used to isolate hepatocytes by the two-step liver perfusion method of Seglen (1976) with some modification (Mitaka et al., 1991a). Briefly, the liver was perfused in situ through the portal vein with 150 ml of Ca²⁺,Mg²⁺-free Hanks’ balanced salt solution (HBSS) supplemented with 0.5 mM EGTA (Sigma Chemical Co., St Louis, MO), 0.5 mg/l insulin (Sigma) and antibiotics. After the initial brief perfusion, the liver was perfused with 200 ml HBSS containing 40 mg of collagenase (Yakulto Co., Tokyo, Japan) for 15 minutes. The isolated cells were purified by Percoll iso-density centrifugation (Kreamer et al., 1986). Viability of the cells by the Trypan Blue exclusion test was more than 90% in these experiments. The cells were suspended in L-15 medium with 0.2% bovine serum albumin (BSA; Seikagaku Kogyo Co., Tokyo, Japan), 20 mM HEPES (Dojindo, Kumamoto, Japan), 0.5 mg/l insulin (Sigma), 10⁻⁷ M dexamethasone (Sigma), 1 g/l galactose (Sigma), 30 mg/l proline (Sigma), and antibiotics. The hepatocytes were plated on 35 mm or 60 mm culture dishes (Corning Glass Works, Corning, NY), which were coated with rat tail collagen (500 μg dried tendon/ml of 0.1% acetic acid) (Michalopoulos and Pitot, 1975), and placed in a 100% air incubator at 37°C. Two to three hours after plating, the medium was changed to L-15 medium supplemented with 0.2% BSA, 20 mM HEPES, 0.5 mg/l insulin, 10⁻⁷ M dexamethasone, 1 g/l galactose, 30 mg/l proline, 20 mM NaHCO₃, 5 mg/l transferrin (Wako Pure Chemical Inc., Osaka, Japan), 0.2 mg/l CuSO₄·5H₂O, 0.5 mg/l FeSO₄·4H₂O, 0.75 mg/l ZnSO₄·7H₂O, 0.05 mg/l MnSO₄, 5 μg/l Na₂SeO₃, 10 ng/ml EGF (Collaborative Res. Inc., Lexington, MA), and antibiotics. The cells were then placed in a humidified, 5% CO₂:95% air incubator at 37°C. The medium was replaced with fresh medium every other day, and 2% DMSO (Aldrich Chemical Co. Inc., Milwaukee, WI) was added after 96 hours of culture (Mitaka et al., 1993).

The cells grown on coated glass coverslips (Biocoat; Becton Dickinson Labware, MA) were fixed with acetone for 30 minutes at −20°C. After rinsing with phosphate-buffered saline (PBS), the coverslips were incubated at room temperature (RT) for 1 hour with a rabbit anti-rat Cx32 (J-peptide) polyclonal antibody (Sakamoto et al., 1992) and the cells were then incubated with fluorescein isothio-cyanate (FITC)-conjugated anti-rabbit IgG (DAKO, Copenhagen, Denmark) at RT for 1 hour. Some coverslips were used for double staining for Cx32 and 5-bromo-2′-deoxyuridine (BrdU; Sigma). Forty μM BrdU was added to the culture medium 24 hours before the fixation. The polyclonal anti-Cx32 and monoclonal anti-BrdU (DAKO) antibodies were used as a primary antibody. The secondary antibodies used were FITC-conjugated anti-rabbit IgG for Cx32 staining and FITC-conjugated anti-mouse IgG (DAKO) for BrdU staining. All samples were examined with a Nikon Fx epifluorescence photomicroscope (Nikon, Tokyo, Japan).

The dishes were washed with PBS twice and 1 ml of the buffer (1 mM NaHCO₃ and 2 mg/l leupeptin (Sigma)) was added to 60 mm dishes. The cells were scraped and collected in Eppendorf tubes and then sonicated for 30 seconds. The sonicates were centrifuged at 4,500 g for 10 minutes. The final pellets were resuspended in the same buffer. The protein concentration of the samples was determined using a protein assay kit (Bio-Rad, Richmond, CA). Twenty μg of protein of each sample per lane was applied and separated by electrophoresis in 12.5% SDS-polyacrylamide gel (Daiichi Pure Chemicals Co., Tokyo, Japan). After electrophoretic transfer to a nitrocellulose membrane (Bio-Rad) using semi-dry blotting for 6 hours (0.65 mA/cm²), the membrane was stained with Ponceau S (Sigma) and photographed. Thereafter, the membrane was saturated overnight at 4°C with a blocking buffer (25 mM Tris-HCl, pH 8.0, 125 mM NaCl, 0.1% Tween-20, 4% skim milk) and was incubated with a mouse monoclonal anti-rat Cx32 (amino acid residues 95-125) antibody (Goodenough et al., 1988) at RT for 2 hours. The membrane was stained by the avidin-biotin complex method (Vectastain ABC kit, Vector Laboratories, Burlingame, CA) and 3,3′-diaminobenzidine (DAB) was used as a substrate.

Total RNA was extracted from the cells, using the single-step thiocyanate-phenol-chloroform extraction method (Chomczynski and Sacchi, 1987) as modified by Xie and Rothblum (1991). For the electrophoresis, 10 μg of total RNAs was loaded on a 1% agarose gel containing 0.5 mg/l ethidium bromide. Gels were capillary-blotted in 20 (1) standard saline citrate (SSC) onto a nylon membrane (Hybond-N; Amersham Corp., Buckinghamshire, UK) and fixed with ultraviolet light. For detection of Cx32 mRNA, digoxigenin (DIG)-labeled RNA probes were prepared from rat cDNAs (Paul, 1986), using an RNA labeling kit (Boehringer Mannheim, Mannheim, Germany). Hybridization, washing and chemiluminescence detection were carried out following the DIG luminescence protocol (Höltke et al., 1992). Scanning densitometry was performed using a Macintosh Quadra 800 computer (Apple Computer, Cupertino, CA) and an EPSON GT-6000 scanner (Seiko Epson, Suwa, Japan). The signals were quantified by the NIH Image 1.52 Densimetric Analysis Program (Wayne Rasband, NIH, Bethesda, MD) (Masters et al., 1992).

The preembedding labeling method was used. The cells were fixed with 1% paraformaldehyde in 3% sucrose/0.1 M PBS (pH 7.4) at 4°C for 2 hours and were permeabilized with 0.2% Triton X-100. The cells were treated at 4°C for 30 minutes with a blocking buffer (4% skim milk in 0.1 M PBS) and were then incubated with a monoclonal antirat Cx32 antibody overnight at 4°C. After washing three times with 3% sucrose/0.1 M PBS, the cells were stained using the ABC method. Thereafter, the cells were refixed in 2.5% glutaraldehyde/0.1 M cacodylate buffer (pH 7.3) overnight at 4°C, postfixed in 2% osmium tetroxide in the buffer, dehydrated with graded ethanol, embedded in situ in Epon 812, and ultrathin sections were cut in a Sorvall Ultra-microtome MT-5000. The sections were stained with uranyl acetate followed by lead citrate, and examined at 60 kV with a JEM transmission electron microscope (JEOL, Tokyo, Japan).

For measuring GJIC, we used the scrape loading/dye transfer method (EL-Fouly et al., 1987) with some modification. Hepatocytes on 35 mm dishes were rinsed several times with PBS. Two or three lines were made around the center of the dish using a surgical blade and 2 ml of 0.05% Lucifer Yellow CH (LY; Sigma) in PBS was added to the dishes. LY is a small molecule (457 Da) which can freely move through gap junctions from loaded cells to neighboring ones. Three minutes after the dye treatment, the cells were rinsed several times with PBS to remove excess dye. We also used rhodamine dextran (10 kDa; Sigma), which is known not to move through gap junctions, as a control dye. We immediately observed the intensity of LY transfer with an Olympus inverted microscope equipped appropriately (Olympus, Tokyo, Japan) and photographed it.

We have shown that primary cultured hepatocytes can proliferate in modified L-15 medium supplemented with 10 ng/ml EGF and 20 mM NaHCO₃ in a 5% CO₂:95% air incubator and are maintained by the use of 2% DMSO (Mitaka et al., 1991b, 1993). The morphology of the cells is shown in Fig. 1. The cells exhibited a few mitoses until day 2 after plating (Fig. 1a) and many mitoses were observed after day 3 of culture. The number of cells increased (Fig. 1b) and almost doubled on day 5. Without DMSO, the cells began to die after day 6 and the number of cells gradually decreased. However, cells treated with 2% DMSO from 96 hours continued to proliferate until day 6 and the number of cells was maintained for more than 2 weeks (Fig. 1c,d,e and f). Although some hepatocytes began to die 2 to 3 weeks after plating, fairly large numbers of cells remained for a long time. We could observe them for as long as 70 days. After 2% DMSO addition, many hepatocytes gradually became larger and we could observe granular particles and some phagosomes in the cytoplasm of relatively large cells (Fig. 1f). Futhermore, the borders of cells became clear with time in culture and the cells looked thicker.

Fluorescence immunocytochemistry was carried out to examine the Cx32 immunoreactivity of hepatocytes at 3 hours, 8 hours, 24 hours, 48 hours, 72 hours, 96 hours, day 6, day 8, day 10, day 12, day 14, day 18 and day 28 after plating. We used two kinds of Cx32 antibodies, rabbit polyclonal (Sakamoto et al., 1992) and mouse monoclonal (Goodenough et al., 1988). The reactivity of these antibodies to Cx32 was almost the same in this experiment. In the hepatocytes cultured in medium without DMSO (from 3 hours to day 14), Cx32-positive macular spots were observed between adjacent hepa-tocytes from 3 hours to 72 hours (Fig. 2a,b and c). After 96 hours, Cx32-positive spots were rarely observed (Fig. 2d). On the other hand, in the hepatocytes treated with 2% DMSO from 96 hours (from 96 hours to day 28), Cx32-positive spots partly reappeared between adjacent cells from 2 days after DMSO treatment (day 6) and the size of the stained spots gradually increased with time of DMSO treatment (Fig. 2e). By about 6 days after DMSO addition, most hepatocytes possessed Cx32-positive spots (Fig. 2f). With time in culture with DMSO, large immunoreactive structures were often found at the borders of adjacent cells and thin lines partly surrounding hepatocytes were also seen. In addition, small dots were observed in the cytoplasm of the hepatocytes (Fig. 2g). Eventually at day 28, Cx32-positive spots could be observed (Fig. 2h).

To elucidate the relationship between cell growth and the formation of gap junctions in hepatocytes, we performed double-staining for BrdU and Cx32. As previously described, in hepatocytes cultured without DMSO, more than 50% of the cells were positively stained with BrdU in the periods of 48-72 hours and 72-96 hours (Mitaka et al., 1991b). At 96 hours after plating, many hepatocytes that had incorporated BrdU into their nuclei were seen but Cx32-positive spots were virtually absent (Fig. 3a). On the other hand, in the hepatocytes treated with 2% DMSO from 96 hours, the number of BrdU-positive cells dramatically decreased. At 4 days after 2% DMSO treatment (day 8), only a few BrdU-positive cells were observed although many Cx32-positive spots were seen between most adjacent hepatocytes (Fig. 3b).

Fig. 4 shows the changes in the amount of Cx32 protein in hepatocytes cultured in medium with (+) or without (−) 2% DMSO. The upper lanes were reacted with the monoclonal anti-Cx32 antibody. The lower lanes, stained with Ponceau S, are shown to confirm the presence of equal amounts of protein. Without DMSO, the amount of the Cx32 protein gradually decreased by 96 hours and was then maintained at a low level. In the hepatocytes with 2% DMSO from 96 hours, the amount of protein increased and was well maintained until day 12.

Fig. 5 shows the changes in transcript expression of Cx32 mRNA in hepatocytes cultured in medium with (+) or without (−) 2% DMSO. Without DMSO treatment (from 3 hours to day 14 after plating), the expression of Cx32 mRNA gradually decreased until 96 hours, although a slight increase in the expression was observed at 48 hours, and then remained at a very low level. In the hepatocytes treated with 2% DMSO from 96 hours (from day 6 to day 28), the expression of Cx32 mRNA increased 10 days after DMSO treatment (day 14) and was then maintained at a level as high as that for hepatocytes at 3 hours after plating.

In hepatocytes 6 days after DMSO treatment (day 10), many high-density structures were observed between adjacent cells (Fig. 6a). Enlargement of the structures revealed that the DAB-positive substances decorated the gap junctions (Fig. 6b). The anti-rat Cx32 antibodies actually reacted to the gap junctional proteins. The ultrastructural visualization of gap junctions by immunolabeling showed that the number and the length of the immunoreactive structures increased with the time of DMSO treatment (data not shown).

As shown in Fig. 7, hepatocytes treated with DMSO possessed the ability to communicate with many cells. In hepatocytes 72 or 96 hours after plating, the spread of LY from cutting lines was 2 to 3 cells thick (Fig. 7a). However, 2 days after DMSO treatment (day 6) the dye reached the 5th or 6th cell from the line (Fig. 7b). Furthermore, in cells treated with DMSO for 10 days (day 14), LY spread widely through a layer of more than 10 cells (Fig. 7c). GJIC among the hepatocytes with DMSO might well be related to the number of Cx32-immunoreactive structures.

In the present study, we demonstrated that proliferated rat hepatocytes could reexpress and maintain gap junctional protein Cx32 in modified L-15 medium supplemented with EGF and 2% DMSO. Many researchers have attempted to maintain differentiated functions and to induce gap junctions in primary cultured hepatocytes by changing culture conditions and by using various substances such as nicotinamide (Inoue et al., 1989; Mitaka et al., 1991a), phenobarbital (Miyazaki et al., 1985), sodium butyrate (Staecker et al., 1988), extracellular matrix (Enat et al., 1984; Spray et al., 1987; Mustat et al., 1993), and by spheroid formation (Koide et al., 1990; Yuasa et al., 1993). When glycosaminoglycan, proteoglycan and Matrigel were added to the medium, Cx32 could be maintained for several days (Spray et al., 1987; Mustat et al., 1993).

cAMP was also reported to delay the disappearance of gap junctions (Sáez et al., 1989b). However, these experiments were carried out using cells which usually died within two weeks, and their gap junctions disappeared within one week. Although Mesnil et al. (1993) recently demonstrated that rat hepatocytes cocultured with Balb/3T3 cells maintained GJIC for more than two weeks, no one has reported the long-term maintenance or new synthesis of Cx32 proteins in primary cultures of pure hepatocytes. We also tried to maintain not only the differentiated functions but also the immunoreactivity to Cx32 of rat hepatocytes by the use of many different culture conditions. However, Cx32-positive cells disappeared within 7 days when we maintained the differentiated cells from the first culture day (data not shown).

The disappearance and reappearance of gap junctions in vivo have been observed in regenerating liver (Yee and Revel, 1978; Traub et al., 1983; Miyashita et al., 1991; Kren et al., 1993). During regeneration after partial hepatectomy, the immunoreactivity to, and the mRNA of, Cx32 rapidly decreased and both recovered their normal levels of expression within 96 hours after the operation (Kren et al., 1993). Dermietzel et al. (1987) reported the relationship between the loss of gap junctions and cell proliferation both in the liver after partial hepatectomy and in cultured hepatocytes. They showed that immunoreactivity to Cx32 was inversely correlated to BrdU labeling. When the hepatocytes ceased proliferating, Cx32 expression was rapidly recovered in the liver after surgery. Thus, we tried to use proliferating hepatocytes to induce Cx32 in primary cultures. As previously reported (Mitaka et al., 1991b), primary hepatocytes went through multiple cell divisions in modified L-15 medium supplemented with EGF in a CO₂ incubator. In addition, 2% DMSO inhibited their proliferation and induced the differentiated functions when the agent was added after the cells proliferated (Mitaka et al., 1993). By the use of these culture conditions, Cx32 could be reexpressed and maintained for a long time. Moreover, this reappearance of Cx32 accompanied the induction of mRNA and the synthesis of new proteins. The cells showed 80% recovery of the amount of mRNA compared to the level of isolated hepatocytes (Fig. 5), and treatment with cycloheximide resulted in inhibition of the reappearance of Cx32 immunoreactivity (data not shown). The synthesis of the proteins continued for more than a month as long as the hepa-tocytes were maintained. Furthermore, it is of importance that the newly formed gap junctional structures had the potential to communicate between cells, and the ability to transfer the dye to distant cells seemed to correlate with the amounts of immunoreactive Cx32. These results suggest that the reappearance of Cx32 in primary cultures of rat hepatocytes requires its transient disappearance, caused by the proliferation of the cells.

Isom et al. (1985) first reported that supplementation of chemically defined medium with 2% DMSO enabled maintenance of differentiated hepatocytes in culture for extended periods. In the present and in previous experiments (Baribault and Marceau, 1986; Isom et al., 1987; McGowan, 1988; Kost and Michalopoulos, 1991), DMSO played an important role in the maintenance of cells; the agent seems to induce the differentiated functions and structures of mature hepatocytes. Although many laboratories have reported long-term culture of such differentiated hepatocytes, none of them can show the well expressed gap junctions which are thought to be one of the highly differentiated functions of hepatocytes. Therefore, DMSO itself dose not suffice for the continuous expression of gap junctions. In fact, when non-proliferating hepatocytes cultured in the absence of EGF were treated with 2% DMSO, Cx32 immunoreactivity was not observed in our experiments. Thus, it seems that there may be several steps in the differentiation of primary cultured hepatocytes. The ability to secrete albumin and transferrin is one of the basic functions of differentiated hepatocytes. Differentiated hepatocytes, whose functions are more similar to those in vivo, may need to have the ability for cell-cell communication through gap junctions as well as the basic functions. In primary cultured hepatocytes, to form functioning gap junctions the cells may need not only DMSO but also the ability to proliferate.

In the liver, both Cx32 and Cx26 are expressed in parenchymal cells, and Cx43 is found in perisinusoidal cells and in the cells of the liver capsule (Berthoud et al., 1992). In primary cultures of hepatocytes, both Cx32 and Cx26 were found and their expressions were reported to be correlated with the differentiated level of the cells. Cx43 was virtually undetected in well-differentiated hepatocytes (Stutenkemper et al., 1992). In this experiment Cx26 was immunocytochemically detected in cells treated with 2% DMSO and the expression of Cx26 mRNA was demonstrated by northern blot analysis (data not shown). However, the expression was very low and unstable compared to that of Cx32. Further studies are necessary for the explanation of the results and for the improvement of the culture conditions. On the other hand, Cx43 expression was reported in hepatic epithelial cells, which often appear in primary cultures of rat hepatocytes (Spray et al., 1991; Oh et al., 1993). In the present experiments an increase in the expression of Cx43 mRNA was observed with time in culture when the cells were cultured without 2% DMSO, whereas positive staining of Cx43 was seen immunocytochemically in nonparenchymal cells (data not shown).

The present study presents evidence that the combination of proliferated rat hepatocytes and 2% DMSO can induce and maintain the gap junctional protein Cx32 in primary cultures. Although the precise mechanism of the induction of Cx32 remains unknown, this culture system may be useful for studying the construction of the gap junctional structure, the intracellular pathways of the proteins, and the regulation of the synthesis and processing of the proteins. In addition, as hepatocytes have the ability to maintain extensive GJIC, together with hepatic functions, they may be useful in an in vitro test system for detecting liver-tumor promoters.

We thank Dr Masahito Oyamada, Ms Minako Kuwano, Ms Yohko Takahashi and Mr Hideki Itoh for their technical assistance. We also thank Mr Kim Barrymore for help with the manuscript. This work was supported by Grants-in-Aid from the Ministry of Education, Science and Culture, Japan, by a Grant-in-Aid from the Hokkaido Geriatric Research Institute, and a Grant-in-Aid from the Sapporo Medical University Foundation for Research Promotion.