Functional Analysis of Human Hepatocytes Isolated From Chimeric Mouse Liver.

Chimeric mice with humanized liver were first established by transplanting primary human hepatocytes (PHHs) isolated from a Japanese 27-year-old donor into complementary DNA-urokinase-type plasminogen activator/severe combined immunodeficiency mice. The PHHs from the Japanese donor increased more than 100-fold in the mouse liver, and human hepatocytes purified from the chimeric mouse liver (hcPHs) were successfully transplanted into second-passaged mice. These PHHs and hcPHs can produce human albumin and preserve many liver-specific enzyme genes, which are important for liver function. Interestingly, hepatitis B virus can be infected with these chimeric mice; hepatitis B viral DNA and hepatitis B surface antigen levels were detectable. In conclusion, hcPHs can be an ideal cell source for analysis of human hepatocytes.

Hepatitis C virus dynamics and cellular gene expression in uPA-SCID chimeric mice with humanized livers during intravenous silibinin monotherapy.

Legalon SIL (SIL) is a chemically hydrophilized version of silibinin, an extract of milk thistle (Silybum marianum) seeds that has exhibited hepatoprotective and antiviral effectiveness against hepatitis C virus (HCV) in patients leading to viral clearance in combination with ribavirin. To elucidate the incompletely understood mode of action of SIL against HCV, mathematical modelling of HCV kinetics and human hepatocyte gene expression studies were performed in uPA-SCID-chimeric mice with humanized livers. Chronically HCV-infected mice (n = 15) were treated for 14 days with daily intravenous SIL at 469, 265 or 61.5 mg/kg. Serum HCV and human albumin (hAlb) were measured frequently, and liver HCV RNA was analysed at days 3 and 14. Microarray analysis of human hepatocyte gene expression was performed at days 0, 3 and 14 of treatment. While hAlb remained constant, a biphasic viral decline in serum was observed consisting of a rapid 1st phase followed by a second slower phase (or plateau with the two lower SIL dosings). SIL effectiveness in blocking viral production was similar among dosing groups (median ε = 77%). However, the rate of HCV-infected hepatocyte decline, δ, was dose-dependent. Intracellular HCV RNA levels correlated (r = 0.66, P = 0.01) with serum HCV RNA. Pathway analysis revealed increased anti-inflammatory and antiproliferative gene expression in human hepatocytes in SIL-treated mice. The results suggest that SIL could lead to a continuous second-phase viral decline, that is potentially viral clearance, in the absence of adaptive immune response along with increased anti-inflammatory and antiproliferative gene expression in human hepatocytes.

Induction of indoleamine 2,3-dioxygenase in livers following hepatectomy prolongs survival of allogeneic hepatocytes after transplantation.

OBJECTIVES: Indoleamine 2,3-dioxygenase (IDO), which catalyzes the breakdown of tryptophan into kyneurenine, has immunologic significance for the induction of maternal tolerance and liver allograft tolerance by inhibiting T-cell activation. In the present study, we compared survival of syngeneic or allogeneic hepatocytes in livers with or without hepatectomy. Subsequently, we investigated gene expression and localization of IDO in the recipient liver. METHODS: DA and Fisher 344 rats were used in the following experimental groups: group 1, DA hepatocytes transplanted into hepatectomized Fisher 344 rats; group 2, Fisher 344 hepatocytes transplanted into hepatectomized Fisher 344 rats; group 3, DA hepatocytes transplanted into nonhepatectomized Fisher 344 rats; and group 4, Fisher 344 hepatocytes transplanted into nonhepatectomized Fisher 344 rats. After transplantation, the surviving cells were evaluated on day 5. The IDO signal of the recipient liver was detected by reverse transcriptase polymerase chain reaction (RT-PCR) and immunohistochemistry. RESULTS: In the hepatectomized groups subjected to allogeneic or syngeneic hepatocyte transplantation, the number of surviving hepatocytes was greater than in the nonhepatectomized group after transplantation. The IDO signals (RT-PCR) in the hepatectomized groups were stronger than those in the nonhepatectomized groups. Immunohistochemistry demonstrated that the IDO signal is located in liver antigen-presenting cells, such as Kupffer cells or dendritic cells, and not expressed in hepatocytes. CONCLUSIONS: Our results demonstrated that IDO is induced in antigen-presenting cells of hepatectomized livers by which subsequently transplanted cells may be protected from rejection by inhibiting indirect or direct recognition of donor antigen and further T-cell activation.

Non-invasive method to detect induction of CYP3A4 in chimeric mice with a humanized liver.

Chimeric mice with a humanized liver have been previously established by the transplantation of human hepatocytes to urokinase-type plasminogen activator/severe combined immunodeficiency mice. A non-invasive method to detect the induction of cytochrome P450 (CYP) 3A4 was evaluated in chimeric mice with a humanized liver. Dexamethasone (DEX) was used as a probe drug to detect induction; and rifampicin was used as a model drug to induce CYP3A4. Before and after rifampicin treatment (50 mg kg(-1), intraperitoneal injection once a day for 4 days) in the chimeric mice, DEX was subcutaneously injected and the urinary excretion of 6beta-hydroxydexamethason (6betaOHD) and DEX was determined. The metabolic ratio (6betaOHD/DEX) significantly increased after rifampicin treatment. Livers from the control and rifampicin-treated chimeric mice were stained immunohistolochemically with antibodies against CYP3A4 and CYP3A5. CYP3A4 and CYP3A5 were detected in the area of humanized liver, but staining was intense for CYP3A4 and very weak for CYP3A5. Only the staining of CYP3A4 was increased after rifampicin treatment. Formation of 6betaOHD by human liver microsomes was higher than that formed by mouse liver microsomes. Metabolite formation was catalysed by both CYP3A4 and CYP3A5 and the intrinsic clearance (V(max)/K(m)) by CYP3A4 was found to be 50-fold higher than that of CYP3A5. The results of the present study indicate that estimation of the changes of the urinary metabolic ratio (6betaOHD/DEX) in the chimeric mice with a humanized liver is a very useful tool for detecting the induction of CYP3A4 by a non-invasive method.

Human cord blood cells transplanted into chronically damaged liver exhibit similar characteristics to functional hepatocytes.

Human umbilical cord blood (CB) cells have many advantages as a source for stem cell transplantation because of immaturity and availability. It has been reported that CB cells transplanted into an injured liver displayed hepatocyte-like phenotypes. However, there have been few studies to characterize CB-derived hepatocyte-like cells (HLCs). In this study, CB cells were transplanted into mice with 2 types of liver damage: transient and chronic damage. We analyzed the expression of hepatic differentiation markers in CB-derived HLCs. In the liver of NOD/SCID mice with transient damage, CB-derived HLCs were detected infrequently at 3 weeks after transplantation. In contrast, in the liver of SCID mice damaged chronically by a urokinase-type plasminogen activator transgene under the control of albumin promotor/enhancer (ALB-uPA/SCID mice), more human HLCs colonized the host liver compared with hosts with transiently damaged livers. The CB-derived HLCs in both the transiently and the chronically damaged liver expressed a few markers of human hepatocytes, whereas the transcripts related to mature hepatic functions, including cytochrome P450s, were detected only in the ALB-uPA/SCID mice. These data indicated that CB cells were able to display a similar phenotype to functional hepatocytes in the recipient liver with chronic damage. CB cells may represent a transplantable source for chronic decompensated liver disease.

In vivo induction of human cytochrome P450 3A4 by rifabutin in chimeric mice with humanized liver.

The induction of human cytochrome P450 enzymes (CYPs) often poses a serious problem in clinical practice. The induction of CYP3A leads to a decrease in the pharmacological potency of drugs, since many drugs are substrates of CYP3A. The present study examined the in vivo induction potency of human CYP3A in chimeric mice with humanized liver, recently established in Japan, by a specific inducer of human CYP3A enzyme activity in this experimental condition, rifabutin, which is an analogue of rifampicin. The chimeric mice were treated intraperitoneally daily for 4 days with rifabutin (50 mg kg(-1) day(-1)). The mRNA, protein and enzyme activity in liver of the chimeric mice were measured by reverse-transcriptase polymerase chain reaction, Western blot analysis and high-performance liquid chromatography, respectively. In the chimeric mice, the human CYP3A4 mRNA expression, CYP3A4 protein content, testosterone 6ss-hydroxylase activity and dexamethasone 6-hydroxylase activity were increased 7.4-, 3.0-, 2.4- and 1.9-fold, respectively, by treatment with rifabutin. The mRNA expression of other human CYPs, transporters and nuclear receptors was not significantly changed by rifabutin. On the other hand, rifabutin was demonstrated not to increase the murine Cyp3a enzyme activities in the control mice. It was demonstrated that human CYP3A4 expressed in the chimeric mice with humanized liver was induced by rifabutin, suggesting that human CYP3A4 in the chimeric mice had induction potency. This chimeric mouse model may be a useful animal model to estimate and predict the in vivo induction of CYPs in human.

Evaluation of mRNA expression of human drug-metabolizing enzymes and transporters in chimeric mouse with humanized liver.

The hepatic mRNA expression of human drug-metabolizing enzymes and transporters in chimeric mise with almost-completely humanized liver (replacement index: 71-89%) was investigated. The mRNAs of 58 human phase I enzymes, 26 human phase II enzymes, 23 human transporters, and five mouse Cyps were measured in the chimeric mice with humanized liver generated using hepatocytes from a Japanese donor. The mRNA expression of 52 human phase I enzymes, which includes 20 human CYPs, 26 human phase II enzymes and 21 human transporters was ascertained in the chimeric mouse liver. Among them, the expression of the target mRNAs vital for liver function such as the metabolism and secretion of endogenous compounds appeared to be maintained. The central value for the expression ratio in all target genes in chimeric mouse liver to the donor liver was 0.46, which was lower than the substitution rate of chimeric mouse liver by donor liver. The ratio of mouse Cyp mRNA expression of chimeric mouse liver to that of control mouse liver was 0.19 or less, except for that of Cyp2b10. There were good correlations between the mRNA expression levels of human hepatic albumin gene, the values of the rate of replacement of mouse liver by human liver, and the human blood albumin concentration in the chimeric mice. The chimeric mice with humanized liver may be a useful tool for the evaluation of drug-drug interactions such as the inhibition and induction of drug-metabolizing enzymes and transporters.

Expression of pleiotrophin in hepatic nonparenchymal cells and preneoplastic nodules in carbon tetrachloride-induced fibrotic rat liver.

Pleiotrophin (PTN) is a heparin-binding protein, which induces growth, angiogenesis, differentiation, and transformation of cells. The aim of this study was to examine the role of PTN in liver fibrogenesis. Rats were treated with carbon tetrachloride (CCl4) for 3-9 weeks to induce liver fibrosis. The sirius-red staining of these liver tissue sections clearly showed the development of fibrosis and glutathione S-transferase placental type-positive preneoplastic nodules emerged at 7 weeks of the treatment. PTN expression was investigated in fibrotic liver tissues at the mRNA level using a real-time reverse transcription polymerase chain reaction and at the protein level by immunohistochemistry. Quantity of PTN mRNA increased 5-fold in fibrotic liver tissues at 7 weeks of CCl4-treatment over the control values. Immunohistochemistry localized PTN protein on hepatic nonparenchymal cells, mostly stellate cells and some of Kupffer cells, and the preneoplastic nodules in fibrotic liver tissues. PTN mRNA expression is significantly upregulated in the CCl4-induced chronic rat fibrotic liver tissues. We suggest that PTN might be involved in fibrogenesis and preneoplastic changes of liver.

[Characterization of a highly proliferative population of adult hepatocytes and its use for tissue engineering].

It is well known among cell biologists that normal hepatocytes of adult mammals are difficult to replicate repeatedly in vitro, irrespective of the fact that these cells can grow well clonally in vivo. We developed a culture medium (HCGM) wherein the normal hepatocytes of adult Fischer rats replicate repeatedly and form clonal colonies. This growth requires the presence of hepatic stellate cells (HSCs). Fractionation and separation of hepatocytes by a combination of centrifugation and cell sorting revealed the presence of a highly proliferative population of hepatocytes in the adult liver, called small-sized hepatocytes (SHs-R3). SHs-R3 showed a 3- to 4-fold higher growth potential than large-sized hepatocytes (SHs-R2) both in vitro and in vivo. The in vivo growth potential was estimated using the retrorsine-dipeptidylpeptidase IV (DPPIV)--rat model in which DPPIV-positive SHs-R3 were transplanted into the liver of retrorsine-treated DPPIV-negative mutant rats which were then subjected to two-thirds partial hepatectomy. The results of our studies suggest a close relationship between SHs-R3 and small hepatocyte-like progenitor cells, as reported by Gordon et al. We showed that the liver of adult humans also contains a highly proliferative population of hepatocytes, which possibly corresponds to SHs-R3. Research is now being undertaken to utilize this population of human hepatocytes to develop an artificial liver.

Size-dependent in vivo growth potential of adult rat hepatocytes.

The present study was performed to determine whether hepatocytes show a size-dependent growth in vivo using as a growth assay system, a retrorsine/partial hepatectomy model of dipeptidyl dipeptidase IV-deficient (DPPIV(-)) mutant Fischer rats. Nearly pure populations of small hepatocytes (SHs) and parenchymal hepatocytes (PHs) were prepared from DPPIV(+) rats. The same number of these SHs and PHs was transplanted into the liver of retrorsine-treated and two-thirds partial hepatectomized DPPIV(-) rats. At 21 days after transplantation, colonies derived from donor hepatocytes were detected as DPPIV(+) cells by enzyme histochemistry. SHs were approximately three times more proliferative than PHs (673 +/- 25 cells/colony versus 226 +/- 10 cells/colony, mean +/- SE). SHs were subfractionated by a fluorescence-activated cell sorter into SH-R2s and SH-R3s. SH-R3s showed a lower extent of granularity and autofluorescence, and a smaller size than SH-R2s that showed characteristics similar to PHs. The growth potential of SH-R3s assayed as above was approximately three times higher than that of SH-R2s (1,101 +/- 46 cells/colony versus 341 +/- 13 cells). These results indicate that the in vivo growth potential of hepatocytes is heterogeneous and is correlated with their size, and the extent of their granularity and autofluorescence.

Proteome analysis of rat hepatic stellate cells.

Proteome analysis was performed on cellular and secreted proteins of normal (quiescent) and activated rat hepatic stellate cells. The stellate cells were activated either in vitro by cultivating quiescent stellate cells for 9 days or in vivo by injecting rats with carbon tetrachloride for 8 weeks. A total of 43 proteins/polypeptides were identified, which altered their expression levels when the cells were activated in vivo and/or in vitro. Twenty-seven of them showed similar changes in vivo and in vitro, including up-regulated proteins such as calcyclin, calgizzarin, and galectin-1 as well as down-regulated proteins such as liver carboxylesterase 10 and serine protease inhibitor 3. Sixteen of them showed different expression levels between in vivo and in vitro activated stellate cells. These results were reproducibly obtained in 3 independent experiments. The up-regulation of calcyclin, calgizzarin, and galectin-1, as well as the down-regulation of liver carboxylesterase 10 were directly confirmed in fibrotic liver tissues. Northern blots confirmed up-regulation of the messenger RNAs (mRNAs) of calcyclin, calgizzarin, and galectin-1 in activated stellate cells, indicating that these changes were controlled at the mRNA level. In addition a list compiling over 150 stellate cell proteins is presented. The data presented here thus provide a significant new protein-level insight into the activation of hepatic stellate cells, a key event in liver fibrogenesis.

Heterogeneity of growth potential of adult rat hepatocytes in vitro.

Nearly pure populations of small hepatocytes (SHs), parenchymal hepatocytes (PHs), and nonparenchymal cells (NPCs) were prepared from the adult rat, and cocultures of hepatocytes and NPCs were reconstituted from them first to obtain the direct evidence that NPCs promote the growth of hepatocytes and second to compare the growth potential between SHs and PHs. SHs and PHs underwent multiple divisions when cocultured with NPCs, whereas neither SHs nor PHs formed colonies at 10 days when cultured alone. Stellate cells in the NPCs were shown to be responsible for this growth promotion. SHs showed a higher growth capacity than PHs. To clearly show the relationship between the growth potential and the size of hepatocytes, SHs and PHs were further fractionated by a fluorescence-activated cell sorter, because the size distribution of SHs and PHs was half overlapped. SHs produced 2 cell populations, SH-R2 and SH-R3. The former showed a greater extent of granularity and autofluorescence than the latter. In contrast, PHs produced only 1 population (PH-R2), which corresponded to the SH-R2. The size of hepatocytes of SH-R3 was smaller (17.1 +/- 0.2 microm) than those of SH-R2 (22.6 +/- 0.5 microm) and PH-R2 (24.1 +/- 0.1 microm) and there was not a significant overlap in the size distribution between the 2 groups. The hepatocytes of SH-R3 were highly replicative and 4 or 5 times higher in their growth potential than those of SH-R2 and PH-R2. We concluded that the growth potential of hepatocytes is heterogeneous and is correlated with their size and the extent of their granularity and autofluorescence.

Growth potential and differentiation capacity of adult rat hepatocytes in vitro.

We have previously reported a medium that supports the continuous growth of hepatocytes without their losing replicative potential and differentiation capacity for an extended period. The medium contains four key substances in addition to fetal bovine serum, that is, epidermal growth factor, nicotinamide, ascorbic acid 2-phosphate, and dimethyl sulfoxide. When a nonparenchymal cell fraction containing small hepatocytes and nonparenchymal cells was cultured in this medium, small hepatocytes grew clonally and differentiated into cells expressing either mature hepatocyte marker proteins or biliary cell marker proteins. The growth potential of small hepatocytes was variable among the cells, the highest case being that of a single cell that produced a colony containing over 100 cells in 10 days. When a hepatocyte was allowed to divide for 105 days, it produced a colony of approximately 0.2 mm2, which contained approximately 1,700 hepatocytes, indicating that the cell divided more than 10 times. Thus, for the first time, we showed the presence of a small compartment of bipotent and highly replicative clonogenic hepatocytes in the rat adult liver in vitro.

A long-term culture of human hepatocytes which show a high growth potential and express their differentiated phenotypes.

The present study succeeded for the first time in cultivating for more than 2 months human normal hepatocytes which showed a high growth potential and expressed their differentiated phenotypes. Constituents of culture medium were critical for this culture, and the medium optimized for their growth contained fresh human serum, fetal bovine serum, Swiss 3T3-cell conditioned medium, L-ascorbic acid 2-phosphate, epidermal growth factor, nicotinamide, and dimethyl sulfoxide. Hepatocytes steadily replicated and formed colonies which continued to increase in size up to around 35 days. The number of hepatocytes in the most replicative colonies increased 17-fold during 31 days. Cells in colonies expressed normal differentiated hepatocytic phenotypes for as long as 35 days. These hepatocytes retained normal liver functions at least for 70 days such as to secrete albumin, and to metabolize lidocaine and D-galactose.

Pleiotrophin as a Swiss 3T3 cell-derived potent mitogen for adult rat hepatocytes.

Rat parenchymal liver cells were cultured in the presence of lethally treated Swiss 3T3 cells. This co-culture allowed hepatocytes to produce colonies containing more than 300 cells in 30 days. Hepatocytes in colonies appeared morphologically normal and some of them were suggested to have bipotental differentiation capacity. The initial growth stimulatory activity of the feeder cells was replaceable with their conditioned medium (CM). Biochemical analysis of an active principle in the 3T3 cell-CM identified pleiotrophin. Pleiotrophin purified from the 3T3 cell-CM, recombinant human pleiotrophin, chemically synthesized human pleiotrophin, and midkine promoted the growth of hepatocytes as well. Reverse transcription-polymerase chain reaction clearly showed that the synthesis of mRNA of pleiotrophin was stimulated in the regenerating liver induced by either partial hepatectomy or the treatment with d-galactosamine, strongly suggesting a biological significance of pleiotrophin in the proliferation of hepatocytes in vivo. From these results we concluded that pleiotrophin is a new potent growth factor for adult parenchymal hepatocytes. This study indicates the importance of mesenchymal stimulation for the growth of adult rat hepatocytes.

Growth and differentiation of adult rat hepatocytes regulated by the interaction between parenchymal and non-parenchymal liver cells.

We have devised a medium which supports the continuous growth of hepatocytes without losing their replicative potential and differentiation capacity for a longer period. The medium HCGM, contains four key substances in addition to foetal bovine serum. They are epidermal growth factor, nicotinamide, ascorbic acid 2-phosphate and dimethylsulphoxide. When a non-parenchymal cell fraction containing small hepatocytes and non-parenchymal cells was cultured in HCGM, small hepatocytes grew clonally and differentiated into cells expressing either mature hepatocyte marker proteins or biliary cell marker proteins. Thus, for the first time, we showed the presence of a small compartment of bipotent and highly replicative clonogenic hepatocytes in the rat adult liver. HCGM also supported the growth of stellate cells (Ito cells) which were in the original preparation, suggesting the important role of stellate cells for the successful cultivation of hepatocytes. Together, these results suggest that a microenvironment is produced as a result of cooperative interactions between hepatocytes and stellate cells: one which stimulates the growth and differentiation of clonogenic hepatocytes.

Immunohistochemical demonstration of the gap junctional protein connexin 32 and proliferating cell nuclear antigen in glutathione S-transferase placental form-negative lesions of rat liver induced by diethylnitrosamine and clofibrate.

The distributions of a gap junctional protein, connexin 32 (cx 32), and proliferating cell nuclear antigen (PCNA) were examined immunohistochemically in glutathione S-transferase placental form (GST-P)-negative foci, induced in rat liver by initiation with diethylnitrosamine (DEN, 200 mg/kg) followed by promotion with clofibrate (1% in diet) in an in vivo medium-term assay system for hepatocarcinogenesis. The results were compared to those in GST-P-positive foci induced by DEN alone. The treatment with clofibrate caused the appearance of GST-P-negative foci, increased in size as compared to GST-P-positive foci in the same liver or induced by the DEN alone. The proportion of PCNA-positive hepatocytes in GST-P-negative foci was significantly higher than in the surrounding parenchyma, indicating increased cell proliferation. The numbers of cx 32-positive spots per hepatocyte in GST-P-negative foci were clearly decreased, reaching 65.4% at week 20 and 51.8% at week 30 of values for surrounding normal hepatocytes. In GST-P-positive foci induced by DEN, only a slight decrease (80%) was observed at week 8. These findings show that a positive association between the sustained inhibition of gap junctional intercellular communication and increased cell proliferation of GST-P-negative foci in Fischer-344 male rats induced with DEN and promoted with clofibrate.

Growth and differentiation in culture of clonogenic hepatocytes that express both phenotypes of hepatocytes and biliary epithelial cells.

A cell fraction containing small hepatocytes and nonparenchymal cells was isolated from the adult rat liver and was cultured in the presence of vitamin C. epidermal growth factor, nicotinamide, and dimethylsulfoxide. All of the small hepatocytes that had attached to a dish expressed hepatocytic phenotypes such as albumin, transferrin, and cytokeratin (CK)8 and CK18 but not biliary markers such as BD1, CK7, and CK19. Single hepatocytes started to proliferate and grew into colonies. The growth potential was variable among the cells, the highest case being that a single cell produced a colony containing over 100 cells in 10 days. The hepatocytes in the colony developed through a proliferation phase and then a differentiation phase. They showed very high bromodeoxyuridine labeling indexes during the first 7 days (proliferation phase), which gradually decreased thereafter. Phenotypic alterations took place at 7 to 10 days, and some hepatocytes started to express mature hepatocyte markers and biliary markers (differentiation phase). The presence of cells that coexpress albumin and biliary markers (CK7 and CK19) was demonstrated by double immunocytochemistry. In addition, cells were identified that ceased to express albumin and in turn were positive for CK19 or CK7. Therefore, the colony was considered to contain liver progenitor-like cells that can differentiate during culture into cells expressing phenotypes of mature hepatocytes or biliary epithelial cells.

Long-term cultivation of adult rat hepatocytes that undergo multiple cell divisions and express normal parenchymal phenotypes.

The present study succeeded in cultivating normal adult rat hepatocytes for at least 85 days without losing their replicative potential and differentiation capacity. Small pieces of hepatocyte aggregates (clusters) were prepared from the primary culture of hepatocytes and used as starting material for the growth experiment. Some of the hepatocytes started to proliferate at 3 days when the clusters were cultured in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 10 ng/ml epidermal growth factor, 10 mmol/L nicotinamide, 0.2 mmol/L L-ascorbic acid 2-phosphate, and 1% dimethylsulfoxide. Clusters continued to grow and formed colonies. All the cells covering colonies expressed normal hepatocyte-specific proteins. The number of albumin-expressing cells in the most replicative colonies increased sixfold during 32 days. Most of the cells were mononucleate and small in size and some of them expressed immature hepatocyte markers such as alpha-fetoprotein. Electron microscopy of cells in colonies revealed the presence of peroxisomes in the cytoplasm and desmosomes, tight junctions, and bile canaliculus-like structures between the cells. Depletion of one of the additives inhibited the growth of hepatocytes. The culture medium used also supported the growth of stellate cells (Ito cells) that had contaminated the original preparation in small numbers and seems to cooperatively stimulate a proliferative population of hepatocytes.