Potential conflict of interest: Nothing to report.
Liver development is regulated by various extracellular molecules such as cytokines and cell surface proteins. Although several such regulators have been identified, additional molecules are likely to be involved in liver development. To identify such molecules, we employed the signal sequence trap (SST) method to screen cDNAs encoding a secreted or membrane protein from fetal liver and obtained a number of clones. Among them, we found that T cell immunoglobulin and mucin domain 2 (Tim2) was expressed specifically on immature hepatocytes in the fetal liver. Tim2 has been shown to regulate immune responses, but its role in liver development had not been studied. We have examined the possible role of Tim2 in hepatocyte differentiation. At first, we prepared a soluble Tim2 fusion protein consisting of its extracellular domain and the Fc domain of human IgG (Tim2-hFc) and found that it bound to fetal and adult hepatocytes, suggesting that there are Tim2-binding molecules on hepatocytes. Second, Tim2-hFc inhibited the differentiation of hepatocytes in fetal liver primary culture, i.e., the expression of mature hepatic enzymes and accumulation of glycogen were severely reduced. Third, Tim2-hFc also inhibited proliferation of fetal hepatocytes. Fourth, down-regulation of Tim2 expression by small interfering RNA (siRNA) enhanced the expression of liver differentiation marker genes. Conclusion: It is strongly suggested that Tim2 is involved in the differentiation of fetal hepatocytes. (HEPATOLOGY 2007;45:1240–1249.)
The liver is a central metabolic organ in adults, whereas it is a major hematopoietic tissue in the fetus.1 The liver primordium is formed from the foregut endoderm and invades the mesenchyme of the septum transversum to give rise to the liver bud.2 The immature hepatocytes composing the liver bud are characterized by the expression of alpha-fetoprotein (AFP) and albumin (Alb). At mid-gestation to late-gestation, the fetal liver functions as the major hematopoietic organ but lacks most of the metabolic functions of the adult liver. Fetal liver acquires such metabolic functions through a series of maturation steps that accompany a decrease in hematopoietic activity and an increase in the expression of various genes involved in liver functions such as tyrosine aminotransferase (TAT), carbamoyl phosphate synthetase (CPS), and glucose-6-phosphatase (G6Pase). Terminal differentiation takes place after birth, and the fully matured liver expresses adult liver-specific enzymes such as tryptophan oxygenase (TO).
Several factors involved in hepatic development have been identified. Fibroblast growth factors (FGFs) and bone morphogenetic protein (BMP) play important roles in the formation of the liver bud,3, 4 and hepatocyte growth factor (HGF) is known to be essential for the proliferation of hepatocytes at the mid-stage of liver development.5 TNFα was reported to play a negative role in liver development.6 We previously showed that oncostatin M (OSM), an interleukin-6 family cytokine, stimulates the maturation of fetal hepatocytes in vitro.7 In the primary culture of mouse liver at mid-gestation, OSM promotes morphological and functional maturation of fetal hepatocytes as evidenced by the expression of various liver metabolic enzymes, accumulation of glycogen, clearance of ammonia, accumulation of lipids, and formation of junction structures,8 indicating that this culture system mimics liver development from mid-stage to late-stage. Although OSM is a potent factor that promotes differentiation of hepatocytes,9 the liver develops normally in mice deficient for OSM or the OSM receptor,10, 11 indicating that additional factors are involved in hepatic development. In fact, we have shown that hepatocytes partially differentiated in high-density cultures of fetal liver cells without OSM,12 and suggested that some membrane-bound proteins play a role in the differentiation of hepatocytes. Identification of the extracellular molecules in hepatocytes would aid understanding of the mechanisms of liver development. To this end, we have been attempting to clone the genes encoding a signal sequence by the SST method using a retroviral expression system13 and have already identified a type I transmembrane protein with EGF repeats, Delta-like (Dlk), also called Pref-1, as being a cell surface molecule expressed in immature fetal hepatocytes.14 Although the function of Dlk in liver development remains to be investigated, it provides an excellent means to isolate immature hepatocytes. By extending this screening, we found that T cell immunoglobulin and mucin domain 2 (Tim2) was highly expressed in fetal liver.
Tim2 is a type 1 transmembrane protein with an immunoglobulin domain and a mucin domain in the extracellular region, and has a putative tyrosine-phosphorylation site in the intracellular region.15 Tim2 belongs to the Tim gene family, which consists of 8 members that includes putative family members Tim5, Tim6, Tim7, and Tim8 in the mouse. Tim1 is also known as a hepatitis A virus receptor in humans16 and was shown to induce T cell activation and to inhibit the development of peripheral tolerance.17 Tim1 was shown to inhibit proliferation of T cells.18 Tim2 is highly homologous to Tim1 in the mouse and the following observations have been reported. Tim2 was expressed on activated T cells and bound its ligand, semaphorin 4A (Sema4A), which was expressed on activated macrophages, B cells and dendritic cells.19, 20 Tim2 was expressed in B cells and in adult liver and kidney, and bound H-ferritin.21 Proliferation of activated T cells and production of Th2 cytokines were increased in Tim2-deficient mice.22 Low-level expression of Tim2 was observed in biliary atresia model mice,23 and Tim2 expression was significantly reduced by the loss of hepatocyte nuclear factor 4α (HNF4α) in hepatocytes.24 However, the expression and function of Tim2 in hepatic development have never been studied. Here, we show that Tim2 is expressed in both fetal and adult hepatocytes, and that it regulates the differentiation of hepatocytes.
AFP, alpha-fetoprotein; Alb, albumin; CPS, carbamoyl phosphate synthetase; Dlk, Delta-like; G6Pase, glucose-6-phosphatase; NPC, nonparenchymal cell; OSM, oncostatin M; PEPCK, phosphoenolpyruvate carboxykinase; Sema4A, semaphorin 4A; siRNA, small interfering RNA; SST, signal sequence trap; TAT, tyrosine aminotransferase; Tim2, T cell immunoglobulin and mucin domain 2.
Materials and Methods
Mice and Cells.
C57BL/6 mice (Nihon SLC, Hamamatsu, Japan) were used for all experiments. All animals received human care in accordance with guidelines prepared by the University of Tokyo. Ba/F3 cells were cultured in RPMI-1640 medium (Invitrogen, Carlsbad, CA) containing 10% fetal bovine serum (FBS) (Equitech-Bio, Kerrville, TX), 50 μg/ml gentamycin, 50 μM 2-mercaptoethanol, and 2 ng/ml interleukin-3. BOSC23 cells and COS7 cells were cultured in Dulbecco's modified Eagle's medium (Sigma-Aldrich, St. Louis, MO, or Invitrogen) containing 10% FBS and 50 μg/ml gentamycin.
Preparation of Liver Cells and Fetal Liver Culture.
Fetal liver cells were prepared and cultured according to the method of Kamiya et al.7 The isolation of Dlk+ cells from fetal liver has been described.14 A single-cell suspension of adult liver cells was prepared as reported.25
Signal Sequence Trap.
The SST method used to screen the fetal liver cDNA library was described elsewhere.14 For screening of the clones, genomic DNA was extracted from Ba/F3 clones and used as a template for PCR. Amplified DNA was transferred to a positively charged nylon membrane after electrophoresis, and hybridized with DIG-labeled probes for Alb, AFP, plasminogen, prosaposin, α1-microglobulin, and Dlk. Membranes were then incubated with alkaline phosphatase–conjugated anti-DIG antibodies (Roche Diagnostics, Indianapolis, IL) at room temperature. The signal was developed with CDPstar (Roche Diagnostics).
Total RNA was extracted from tissues or cultured cells with Trizol reagent (Invitrogen) and mRNA was prepared using FastTrack (Invitrogen) or an Oligotex-MAG mRNA purification kit (Takara Bio, Shiga, Japan). After electrophoresis, RNA was transferred to a positively charged nylon membrane, hybridized with a DIG-labeled probe, and analyzed as described above. The full coding sequence of Tim2 cDNA and also a 39-base oligonucleotide specific to Tim2 (5′-GGAGCTTCCCCCAAAAAAGTGGTCGAACGGACCAGATGT-3′) were used for the detection of Tim2 mRNA.
Tim2 rapid amplification of cDNA ends (RACE) assay was performed on cDNA from mouse liver at embryonic day 18.5 (E18.5), using the Marathon cDNA Amplification Kit (BD Biosciences, Franklin Lakes, NJ) according to manufacturer instructions. The primer used for 3′RACE of Tim2 was 5′-CACCGTCTTGGACCAGAAAG-3′. For 5′RACE of Tim2, 5′-GACCTGGTCTTCACATCTGGTCCGTTCG-3′ was used.
Production of Recombinant Tim2-hFc Fusion Protein and Biotinylation.
The cDNA fragment encoding the extracellular domain of Tim2 was prepared from a fetal liver cDNA library, using a pair of oligonucleotide primers containing a sense sequence including an XhoI site (5′-TTCTCGAGCCTCCGATAGGCAGAG-3′) and an antisense sequence including a BamHI site (5′-AAGGATCCGGCTTCTGTGGAGGGATTAC-3′). The resulting XhoI-BamHI fragment was inserted into the CDM8 human IgG Fc cassette to generate Tim2-hFc proteins. The plasmid DNA (6-11 μg) was transfected into COS7 cells by using FuGENE6 Transfection Reagent (Roche Diagnostics) in a 10-cm dish. Tim2-hFc protein was purified from the culture medium at 3 or 4 days after transfection, by using a protein G column (MAb Trap kit; GE Healthcare Bio-Sciences, Piscataway, NJ). About 10-30 μg of Tim2-hFc protein was obtained from 1 plate. Purified protein was biotinylated by using an ECL Protein Biotinylation Module (GE Healthcare) according to manufacturer instructions.
Aliquots of 106 cells were incubated with biotinylated anti-Tim2 rat monoclonal antibodies (Ab) on ice for 30 minutes following the blocking of the Fc receptor with anti-FcR antibodies for 15 minutes. Cells were washed with cold PBS and stained for 15 minutes with allophycocyanin (APC)-conjugated streptavidin (SA-APC) (BD Biosciences). They were then washed and analyzed with a FACSCalibur (BD Biosciences).
For Tim2-binding assays, cells were incubated with biotinylated Tim2-hFc protein on ice for 1 hour with anti-FcR antibodies. Cells were washed with cold PBS and stained for 15 minutes with SA-APC for flow cytometric analysis.
In order to estimate the level of glycogen in hepatocytes, cultured fetal hepatocytes were harvested at day 6 with 0.2 M glycine buffer followed by the addition of chloroform. The aqueous phase containing glycogen was separated by centrifugation, and glycogen was precipitated by addition of ethanol. Glycogen dissolved with distilled water was measured on a spectrophotometer at 490 nm wavelength after being mixed with 5% phenol and sulfuric acid.
Small Interfering RNA.
A small interfering RNA (siRNA) oligonucleotide containing a 19-base-pair sequence of mouse Tim2 was synthesized by Hokkaido System Science (Hokkaido, Japan) and cloned into the vector pSIREN-RetroQ (BD Biosciences) according to manufacturer instructions. The sequences we used for Tim2 were: No. 1, 5′-TCGTTCCTATGTGTTGGGG-3′; No. 2, 5′-GATCCACACATGTACCAAC-3′; No. 3, 5′-CCAGAGTCT-CTACCTCTAC-3′; and No. 4; 5′-GACACTCCTTACCCTGAAG-3′. The control oligonucleotide included in the kit was used as a negative control. About 16-24 hours before lipofection, fetal liver cells were plated on gelatin-coated dishes. The siRNA vectors were transfected with Lipofectamine and Plus Reagent with OPTI-MEM (Invitrogen). Three hours after lipofection, the medium was changed to fetal liver culture medium. At 24 hours after lipofection, the transfected cells were selected by using 5 μg/ml of puromycin for 24-28 hours. Three days after lipofection, the expression of Tim2 and HPRT was examined by quantitative PCR, and the protein level of Tim2 was confirmed by flow cytometric analysis using anti-Tim2 Ab.
Quantitative Real-Time PCR Analysis.
Total RNA (1 μg) and random hexamer primers were used to synthesize cDNA using the First-strand cDNA synthesis kit (GE Healthcare) or SuperScript III First-Strand Synthesis System (Invitrogen). The cDNA was mixed with SYBR Green Realtime PCR Master Mix (Toyobo, Osaka, Japan) and the primers (400 nM), then was analyzed on a LightCycler (Roche Diagnostics). The samples were denatured at 95°C for 30 seconds, and then subjected to 40 cycles of denaturation at 95°C for 10 seconds, annealing at 56°C for 10 seconds, and extension at 72°C for 10 seconds. The ratio of target gene mRNA to housekeeping gene (HPRT) mRNA was determined. The primers used for Tim2 were the same sequences used by Chen et al.21 Primers for phosphoenolpyruvate carboxykinase (PEPCK) were 5′-TGGAGGAGATCGACAGGTAT-3′ and 5′-CAGCTAACGTGAAGAACTGG-3′ and for G6Pase were 5′-AACCCATTGTGAGGCCAGAGG-3′ and 5′-TACTCATTACACTAGTTGGTC-3′.
Fetal liver cultured cells were harvested with PS buffer (50 mM Hepes, 100 mM NaF, 10 mM NaPPi-10H2O, 4 mM EDTA, 2 mM Na3VO4, 2 mM Na2MoO4-2H2O) to which was added NP-40 and protease inhibitor cocktail (Roche). After being loaded on 10% SDS-PAGE, proteins were transferred to a polyvinyl difluoride membrane. Following blocking, the membrane was incubated with primary antibody (Ab) [anti-phospho-STAT3 Ab, anti-STAT3 Ab (Cell Signaling Technology, Danvers, MA) and anti-actin Ab (Santa Cruz, Santa Cruz, CA)]. The signal was developed with Lumi-Light (Roche) after incubation with HRP-conjugated secondary Abs (GE Healthcare).
Mouse E14.5 fetal liver was fixed with 4% paraformaldehyde (PFA) at 4°C overnight, soaked in 25% sucrose, and frozen-sectioned after embedding in O.C.T. Compound (Sakura Finetechnical, Tokyo, Japan). Sections were stained with anti-Tim2 rat Ab and anti-Alb goat Ab (Bethyl Laboratories Inc., Montgomery, TX). Signals were visualized by anti-rat IgG-Cy3 (Sigma) and anti-goat IgG Alexa Fluor 488 (Invitrogen). Active caspase 3 staining was performed with anti-active caspase-3 rabbit Ab (Promega) and visualized by anti-rabbit IgG-Alexa Fluor 546 (Invitrogen). Nuclei were stained with DAPI. Apoptosis was induced by actinomycin D and TNFα according to the method of Pierce et al.26 Ki67 staining was performed with anti-Ki67 mouse IgG (BD Biosciences) and visualized by anti-mouse IgG-Alexa Fluor 546 (Invitrogen).
Dual luciferase assay was performed with the Luciferase Assay System (Promega, Madison, WI) and luminometer TD20/20 (Turner Designs, Sunnyvale, CA). Transfection efficiency was normalized by Renilla Luciferase.
Isolation of cDNA Fragments Encoding a Signal Sequence from Fetal Liver.
To identify cell surface proteins and secreted proteins in fetal hepatic cells, we applied the SST method to E14.5 mouse liver. Because fetal liver is a major hematopoietic organ, the SST library was constructed from E14.5 hepatic cells deprived of CD45+ and TER119+ hematopoietic cells. The library was transfected in Ba/F3 cells, and the resultant 640 positive clones were analyzed by amplifying the cDNA fragment integrated into the genome by PCR. We previously performed the SST screening and found that fetal liver cells preferentially expressed Alb, AFP, plasminogen, prosaposin, α1-microglobulin and Dlk.14 Therefore to exclude such abundant clones, we employed Southern blotting using the probes for those genes. By this negative selection, we eventually identified 108 different cDNA clones (Supplementary Table 1). Half of the cDNA clones encoded a secreted protein or hypothetical secreted protein. One-fourth of the clones encoded transmembrane protein or hypothetical transmembrane protein, and the rest of them encoded intracellular membrane protein. Among transmembrane proteins, we focused on the ones whose function in fetal liver had never been studied. Because northern blot analysis revealed that Tim2 was highly expressed in the liver (Fig. 1A), we further analyzed Tim2.
Expression Profile of Tim2.
Although it was shown that Tim2 is expressed in activated T cells,19, 20, 22 B cells, and adult liver and kidney,21 its expression in other cell types had not been known. Northern blot analysis of various adult tissues revealed that Tim2 was highly expressed in the liver and was weakly detectable in kidney, whereas it was undetectable in other tissues (Fig. 1A). Although it was previously reported that Tim2 is expressed in T cells and B cells, Tim2 mRNA was not detected in spleen. This may be due to very low expression of Tim2 in unstimulated T cells, though Tim2 is expressed in activated T cells and B cells.21 In E14.5 mouse fetus, northern blot analysis showed that Tim2 was strongly expressed in the liver and weakly expressed in the kidney (Fig. 1B). In the liver, 2 forms of Tim2 RNA were expressed from E11.5 to adulthood (Fig. 1C). Whereas the expression of Tim2 mRNA of high molecular weight increased with liver development, it was almost absent in adult liver. On the other hand, the lower molecular weight mRNA species appeared at a slightly later stage and the expression continued into adulthood. To identify these 2 forms of Tim2, we performed RACE assay using E18.5 mouse liver cDNA and found 2 alternatively spliced mRNA species with the same coding sequence of Tim2 in the database (GenBank), i.e., BC025096 and NM_134249. It suggests that 2 distinct mRNA species will correspond to these spliced variants with no difference in amino acid sequence.
Tim2 Is Expressed on Immature and Mature Hepatocytes.
The liver is composed of parenchymal cells (i.e., hepatocytes) and nonparenchymal cells (NPCs) such as endothelial cells, stellate cells, cholangiocytes, and various kinds of blood cells. To identify the cell types that expressed Tim2, we performed a flow cytometric analysis using anti-mouse Tim2 rat monoclonal Ab. The anti-Tim2 Ab was generated by immunizing rat with a recombinant soluble form of Tim2 fused with human IgG Fc fragment produced as described.19 We confirmed this anti-Tim2 Ab recognized the Tim2 ectodomain by flow cytometric analysis using Ba/F3 cells transfected with Tim2 cDNA (Supplementary Fig. 1). We previously reported that most immature fetal hepatocytes express Dlk/Pref-1.14 Flow cytometric analysis revealed that almost all Dlk+ fetal liver cells expressed Tim2 as early as E11.5 (Fig. 2A). To identify the cells expressing Tim2 in adult liver, we fractionated adult liver cells into parenchymal cells and NPCs by low-speed centrifugation. The hepatocyte fraction was negative for CD45 (data not shown), indicating that hematopoietic cells were excluded from this fraction. Flow cytometric analysis revealed that all adult hepatocytes expressed Tim2 (Fig. 2B). Immunohistochemical analysis of E14.5 fetal liver with anti-Tim2 Ab and anti-Alb Ab also showed that Tim2 was expressed in fetal hepatocytes (Fig. 2C). Strong staining signal was observed at the junctions of cells, suggesting that Tim2 may be involved in the interaction between heptaocytes.
Tim2 Binds to Fetal and Adult Hepatocytes.
It was reported that Tim2 interacts with Sema4A,19 and Chen et al. recently reported that Tim2 binds H-ferritin.21 To assess the interaction of Tim2 with fetal liver cells and evaluate the effect on hepatic development, we constructed a soluble form of Tim2 fusion protein consisting of the extracellular domain of Tim2 and the Fc domain of human IgG (Tim2-hFc). Tim2-hFc protein was prepared by an expression system using COS7 cells and purified from the culture medium with a protein G column. The purity and the dimerization of Tim2-hFc protein were evaluated by electrophoresis under reducing and nonreducing conditions (Fig. 3A). To investigate the binding of Tim2-hFc to fetal liver cells, the total cell population was incubated with either biotinylated Tim2-hFc protein or biotinylated control human IgG, followed by fluorescein isothiocyanate (FITC)-conjugated anti-Dlk Ab and SA-APC. Flow cytometric analysis revealed that Tim2-hFc bound to almost all Dlk+ cells, indicating that Tim2 could bind to fetal hepatocytes (Fig. 3B). When adult liver parenchymal cells were investigated by flow cytometry, Tim2-hFc also bound to adult hepatocytes (Fig. 3C). These results indicate that Tim2-binding protein is expressed on Dlk+ fetal hepatocytes as well as mature adult hepatocytes.
Suppression of Fetal Liver Maturation and Proliferation by Tim2-hFc.
We previously established a primary culture system of fetal hepatocytes, in which OSM strongly induces the functional as well as morphological maturation of hepatocytes.7 Taking advantage of this system for monitoring hepatic development, we examined the function of Tim2 on the differentiation of hepatocytes. Because FACS analysis showed that Tim2-hFc interacted with fetal hepatocytes, it was expected that the addition of Tim2-hFc could have any effects on hepatocyte differentiation. In fact, the addition of Tim2-hFc protein to fetal liver cultures disturbed typical morphology of mature hepatocytes, which was induced in the culture without Tim2-hFc (data not shown). Because this culture system contained not only hepatocytes but also NPCs, there was a possibility that Tim2-hFc protein affected the differentiation of hepatocytes indirectly via NPCs. To exclude this possibility, we isolated only immature Dlk+ hepatocytes from fetal liver by sorting with anti-Dlk Ab, and reexamined the effect of Tim2-hFc. It also inhibited the morphological maturation of hepatocytes (Fig. 4A) and the expression of TAT and CPS (Fig. 4B). We previously showed that differentiated hepatocytes accumulated a large amount of polysaccharide in vitro.7 Quantitative analysis of glycogen in cultured Dlk+ cells demonstrated that the accumulation of glycogen was also severely reduced by Tim2-hFc (Fig. 4C). These results indicate that Tim2-hFc directly binds to immature hepatocytes to inhibit many aspects of differentiation in vitro. We previously reported that STAT3 plays a major role for the OSM-induced differentiation in fetal liver culture.27 To determine whether Tim2-hFc interfered with OSM signaling, tyrosine phosphorylation of STAT3 was investigated by western blotting. Tim2 did not affect the phosphorylation state of STAT3, suggesting that Tim2-hFc inhibited fetal liver differentiation independently of the STAT3 pathway (Fig. 4D).
To assess whether Tim2-hFc affected cell growth, we examined the number of cultured Dlk+ fetal hepatocytes. Compared with control IgG, the cell number was significantly reduced in the presence of Tim2-hFc after 4 days of culture (Fig. 5A). To determine the reason for the reduction of cell number, we examined cell viability. Immunostaining with anti-active caspase 3 Ab (Fig. 5B) showed that Tim2-hFc did not activate caspase 3, suggesting that it did not induce apoptosis. We then investigated cell proliferation. The number of Ki67-positive cells, i.e., proliferating cells, was also reduced after 3 days of culture (Fig. 5C). These results indicate that the binding of Tim2-hFc to fetal hepatocytes suppresses cell proliferation without affecting apoptosis.
Differentiation of Hepatocytes Is Enhanced by Down-Regulation of Tim2 Expression.
To uncover the role of Tim2 in the differentiation, we knocked down Tim2 expression in the fetal liver culture system using 4 different siRNAs and examined their effect on differentiation. Introduction of Tim2 siRNA into the fetal liver culture by transient transfection successfully reduced the Tim2 expression level as shown by quantitative real-time PCR (Fig. 6A). Interestingly, northern blotting showed that all these siRNAs knocked down Tim2 expression and enhanced the expression of TAT and CPS, which are liver differentiation markers, although the effect was variable depending on the siRNA (Fig. 6B). We selected siRNA No. 3 for further analysis, because it exhibited the strongest effect on the expression of Tim2. The reduction of Tim2 protein level by siRNA No. 3 was also confirmed by flow cytometry using anti-Tim2 Ab (Fig. 6C). In addition to TAT and CPS, expression of PEPCK and G6Pase was increased by siRNA No. 3 as shown by quantitative real-time PCR (Fig. 6D). The results suggest that Tim2 expression may negatively modulate the expression of liver differentiation genes and that Tim2 binding protein might cancel the negative signal of Tim2. These results indicate a possibility that binding between Tim2 and its receptor/ligand affects liver differentiation.
In this study, we employed the SST method to isolate cDNAs encoding a signal sequence from E14.5 fetal liver cells. As demonstrated here as well as in the literature, SST is a powerful means to isolate cDNAs coding for a secreted or membrane-bound protein, and we isolated a number of such cDNAs from fetal liver. In this report, we describe a possible role for one of them, Tim2, in the development of the liver. Tim2 is a part of the Tim family which has 8 members in the mouse. Tim2 exhibits the highest homology to Tim1, and in humans, Tim1 is known as a receptor for the hepatitis A virus, but its function in liver remains unclear. Curiously, there is no human Tim member similar to Tim2, which is highly homologous to human Tim1. Moreover, Tim1 is barely detectable in mouse hepatocytes (our unpublished data and Chen et al.21). Although there is no evidence that mouse Tim2 binds hepatitis A virus, it may be a counterpart of human Tim1. Tim2 was previously shown to play an important role in immune responses and it was recently shown to be a receptor for H-ferritin. However, the role of Tim2 in liver development had not been studied until now.
We show that Tim2 is highly expressed not only in adult hepatocytes, as reported, but also in Dlk+ immature fetal hepatocytes. Dlk+ hepatocytes in fetal liver differentiate into both hepatocytes and biliary epithelial cells in vitro,14 indicating that Dlk+ cells are bipotential hepatoblasts. The differentiation of fetal hepatocytes into mature hepatocytes is induced in vitro by cytokines such as OSM. The differentiation of hepatocytes can also be strongly induced at high cell density, suggesting that cell-cell contacts contribute to differentiation.12 Nakamura et al. also reported that hepatic membrane molecule(s) modulated cell growth and liver function via cell-cell contacts.28 Because Tim2 is expressed on hepatocytes and Tim2-hFc binds to hepatocytes in fetal and adult liver, Tim2 appears to mediate cell-cell contacts between hepatocytes, which may affect the differentiation of hepatocytes. Actually, immunohistochemistry using anti-Tim2 Ab showed that strong Tim2 signal was observed between cells, suggesting that Tim2 may be localized between hepatocytes (Fig. 2C).
We found that Tim2-hFc blocks the differentiation of hepatocytes induced by OSM in vitro as evidenced by morphology, reduced expression of various differentiation marker genes, and reduced accumulation of glycogen. Tim2-hFc also suppressed proliferation of fetal hepatocytes. As a result of growth inhibition by Tim2-hFc, cell-cell interaction between fetal hepatocytes might be inhibited and hence differentiation might be blocked. Alternatively, Tim2-hFc might block the binding of Tim2 to its receptor/ligand and inhibit signals contributing to hepatocyte differentiation. Because control hIgG Fc had no effect on hepatocyte differentiation, the inhibitory effect must be mediated by Tim2 extracellular domain, although the possibility is not precluded that the binding of Tim2-hFc to its receptor/ligand simultaneously interferes with the function of other cell surface molecules. Because the liver is a major hematopoietic organ at the fetal stage, there are numerous blood cells surrounding fetal hepatocytes. However, as liver development proceeds, fetal hepatocytes gradually come into contact with each other to form the hepatic cord. Thus, it is possible that the binding of Tim2 to its receptor/ligand on neighboring cells occurs infrequently at the early stage of liver development, but increases as liver develops. In our experiment, the knockdown of Tim2 expression by siRNA enhanced the differentiation of hepatocytes as shown by the increased expression of liver differentiation marker genes in primary cultures of fetal hepatocytes. Assuming that Tim2 delivers a negative signal for differentiation and the binding to its receptor/ligand attenuates such signaling, cell-cell contacts between hepatocytes via Tim2 would contribute to hepatic differentiation. Consistent with this idea, the recent report showed that transfection of Tim2 cDNA in a T cell line resulted in inhibition of transcriptional activity of NFAT and AP-1.29 We previously reported that OSM induced fetal liver maturation via the gp130-STAT3 pathway. We investigated the possibility that STAT3 may be involved in Tim2 signaling. However, siRNA-mediated Tim2 knockdown did not affect STAT3 activity by luciferase reporter assay, suggesting that Tim2 does not modulate STAT3 activity (data not shown).
Two proteins have been shown to bind Tim2: Sema4A and H-ferritin. Although Sema4A mRNA was detected in hepatocytes by reverse transcription PCR, anti-Sema4A Ab failed to bind hepatocytes significantly (data not shown). Therefore, it seems unlikely that Sema4A is the major ligand for Tim2 in hepatocytes. Ferritin is a polymeric protein composed of 24 subunits, namely heavy (H) and light (L) chains in any ratio.30 Ferritin is primarily present intracellularly and controls cellular iron homeostasis which is implicated in the generation of reactive oxygen species.31 On the other hand, evidence suggests that H-ferritin inhibits the proliferation of T cells and impairs the maturation of B cells.21 Moreover, it was recently demonstrated that Tim2 binds to and is required for the internalization of H-ferritin. Because ferritin is a large protein complex with multiple H and L subunits, it may bind multiple Tim2 proteins. Thus, Tim2 expressed on a hepatocyte may bind the H-ferritin bound to Tim2 on a neighboring cell, contributing to a cell-cell contact between hepatocytes. It is possible the internalized H-ferritin may affect the differentiation of hepatocytes. In fact, because ferritin mRNA was detected in the culture of Dlk+ fetal hepatocytes throughout culture days 1 to 6 (data not shown), the observed effects in the culture might depend on H-ferritin. Alternatively, it is also possible that there is a novel binding protein for Tim2 on hepatocytes which cancels the negative signal for differentiation induced by Tim2.
In summary, although the precise mechanism remains to be studied, the results described here strongly suggest that Tim2 is involved in the differentiation of hepatocytes. In addition, because Tim2 is expressed on activated T cells and hepatocytes express Tim2-binding molecules, it is also possible that T cells and hepatocytes interact with each other when activated T cells invade the liver parenchyma in an inflammatory reaction.
We thank K. Sekine (Meikai University, Japan), Y. Kishi (MBL, Japan), A. Kamiya (IMS, University of Tokyo) and S. Watanabe (IMS, University of Tokyo) for helpful discussion.