SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Spermatogenic immunoglobulin superfamily (SgIGSF) is an intercellular adhesion molecule of the nectin-like family. While screening its tissue distribution, we found that it was expressed in fetal liver but not adult liver. In the present study, we examined which cells in developing and regenerating liver express SgIGSF via immunohistochemistry and Western blot analysis. In developing mouse liver, SgIGSF expression was transiently upregulated at perinatal ages and was restricted to the lateral membrane of biliary epithelial cells (BECs). In regenerating rat livers from the 2-acetylaminofluorene/partial hepatectomy model, SgIGSF was detected exclusively in oval cells that aligned in ductal and trabecular patterns by the second week posthepatectomy. In human livers, fetal and newborn bile ducts and cirrhotic bile ductules were clearly positive for SgIGSF, whereas disease-free adult bile ducts were negative. To investigate the role of SgIGSF in bile duct/ductule formation, we used an in vitro model in which rat hepatocyte aggregates embedded in collagen gels containing insulin and epidermal growth factor extend epithelial sheets and processes in the first week and form ductules within a month. The process and ductular cells were continuously positive for SgIGSF and cytokeratin 19, a BEC marker. When the aggregate culture was started in the presence of a function-blocking anti-SgIGSF antibody, the number of epithelial processes per aggregate was reduced by 80%. Conclusion: We propose that SgIGSF is a novel and functional BEC adhesion molecule that is expressed for a limited time during active bile duct/ductule formation. (HEPATOLOGY 2007;45:684–694.)

Spermatogenic immunoglobulin superfamily (SgIGSF),1 alternatively referred to as nectin-like molecule-2 (Necl-2),2 tumor suppressor in lung cancer-1,3 or synaptic cell adhesion molecule,4 is an intercellular adhesion molecule of the Necl family, which consists of four other members: Necl-1, -3, -4, and -5.2 Structurally, Necl family members have three immunoglobulin-like motifs in the extracellular domain, and an intracellular domain that lacks the ability to directly bind afadin, an F-actin–binding protein.1, 2 SgIGSF is expressed by a variety of cells including epithelial cells such as lung alveolar cells5 and pancreatic secretory cells,6 as well as nonepithelial cells such as neurons,4 spermatogonia,7 and mast cells.8 SgIGSF can interact homophilically or heterophilically depending on the cell types expressing SgIGSF and the binding partners available on adjacent cells. SgIGSF binds homophilically among neurons4 and between mast cells and neurons,9 and heterophilically between mast cells and fibroblasts8 and between spermatogonia and Sertoli cells.10 Recently, one of the heterophilic binding partners was identified as the Class I–restricted T cell–associated protein.11, 12

Although the physiologic roles of SgIGSF have been demonstrated in several cell types, including neurons and mast cells, primarily using cultured cells,4, 8 there are only a few reports that describe its function during ontogeny.5, 13 In preliminary work, we examined SgIGSF protein distribution in mouse neonates using immunohistochemistry, and found that signals specific for SgIGSF were detected in the liver. The immunoreactive cells appear to be biliary epithelial cells (BECs), suggesting a role for SgIGSF in the development of intrahepatic bile ducts (IHBDs).

During mouse embryogenesis, the primary liver bud appears as a cord of hepatoblasts by embryonic day 9.5 (E9.5).14, 15 The first event of differentiation of IHBDs is appearance of biliary precursor cells, a subset of hepatoblasts expressing high levels of biliary-specific cytokeratins, at E13.5-14.5.14 These precursor cells form a continuous single-layered ring, called a ductal plate, around the portal mesenchyme at E15.5, and the next day the ductal plate forms bilayers which will give rise to bile ducts in the portal mesenchyme at birth.14 Recent studies have identified several factors involved in these processes. The one-cut protein hepatocyte nuclear factor-6 is a transcription factor that is expressed in hepatoblasts and regulates the initiation of their differentiation to a BEC phenotype.16 The Notch receptor-mediated signaling pathway is important for IHBD formation, because mutations in the human Jagged-1, a Notch receptor ligand, are responsible for Alagille syndrome, which is characterized by a paucity of IHBDs.17–19 Cell–matrix and cell–cell interactions are also likely to influence IHBD development. For example, in the portal mesenchyme, extracellular matrix components change dramatically during hepatogenesis, coinciding with the onset of BEC differentiation,20 and BECs express neural cell adhesion molecule (NCAM) on the cell membrane in developing, but not mature, stages.21, 22 However, little is known about adhesion molecules other than NCAM, which may be expressed on BECs, or about their roles in IHBD development.

Formation of bile duct–like structures takes place even in the fully differentiated liver under several conditions. When a partial hepatectomy (PH) is performed in adult rodents treated with 2-acetylaminofluorene (AAF), a chemical carcinogen that inhibits hepatocyte proliferation, characteristic oval-shaped cells appear in the periportal areas, form tortuous and branching ductules, and then gradually penetrate deep into the liver lobule.23–25 In diseased and injured livers of humans, an increase in bile duct–like structures occurs as a proliferative response known as a ductular reaction.21, 26 The typical ductular reaction is proliferation of small ducts lined by columnar epithelial cells, and is often seen in acute cholestasis.21, 26 An atypical ductular reaction is characterized by an anastomsing network of ductules lined by flattened cells, which is often seen at the edges of the expanded portal areas in chronic cholestatic diseases such as cirrhosis.21, 26

We previously established a culture system in which mature hepatocytes are allowed to aggregate on plastic dishes and are then cultured in collagen gels containing epidermal growth factor and insulin.27 After being cultured for more than 3 weeks, some hepatocytes form branching ductular structures encircled by a basement membrane. The ductular cells were positive for BEC markers such as cytokeratins (CKs) 7, 19, and 20, but were negative for liver stem cell markers such as alpha-fetoprotein and delta-like, and the expression of hepatocyte markers such as albumin gradually decreased.28 In addition, activation of the Jagged–Notch signaling pathways was evident before ductules were formed.28 Thus, our three-dimensional cultures are an in vitro model for bile ductule formation.

In the present study, we examined the expression profile of SgIGSF in liver development and regeneration, and determined the cell types that express SgIGSF. We further examined whether the adhesion molecule might play a role in ductular formation using our in vitro culture model. SgIGSF appears to be a novel BEC adhesion molecule that was expressed for a limited time during active duct/ductule formation.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Antibodies.

Two anti-SgIGSF antibodies were used: a rabbit polyclonal antibody against the C-terminal peptide and a function-blocking chicken monoclonal antibody (9D2) directed against the recombinant ectodomain of SgIGSF.9 Negative controls for 9D2 were normal chicken immunoglobulin (U04; Sigma Chemical Company, St. Louis, MO) and a mouse monoclonal antibody (TLD3A12; BD Pharmingen, San Jose, CA) that blocks the activity of rat platelet endothelial cell adhesion molecule-1 (PECAM-1).29 Anti-CK19 mouse monoclonal antibodies were purchased from Novacastra (Newcastle, UK) and Amersham Biosciences (Buckinghamshire, UK), and used for detection of human and rodent CK19. Other primary antibodies used in this study were anti-NCAM (12F11; BD Pharmingen), anti–β-actin (AC-74; Sigma Chemical Company), and GAPDH (goat polyclonal; Santa Cruz Biotechnology, Santa Cruz, CA). Peroxidase- and fluorophore-conjugated secondary antibodies were purchased from Amersham and Jackson ImmunoResearch (West Grove, PA), respectively.

Rodent and Human Tissue.

C56BL/6 mice and Fischer 344 rats were purchased from Japan SLC (Hamamatsu, Japan). Cirrhotic liver tissues were obtained from hepatocellular carcinoma patients who had suffered from viral hepatitis and underwent a partial hepatectomy at Kobe University Hospital from 2000 to 2002. The patients were not treated with chemotherapy or irradiation before surgery. Human perinatal livers were obtained from autopsy cases at Kobe University Hospital in 2005 from infants that died of abortion due to placental dysfunction and of cardiac abnormalities.

AAF/PH Model.

The AAF/PH model has been described in detail previously.30 Briefly, AAF (Tokyo Chemical Industry, Tokyo, Japan) was administered daily to rats via gavage at 15 mg/kg body weight for 4 days. On the next day, the rats underwent a standard PH, removing two thirds of the liver, and were then administered AAF daily at the same dosage for 5 days. Animals were sacrificed on days 0, 4, 7, 9, 13, and 20 after PH. All experiments with rats were approved by the Animal Care Committee of the Hyogo College of Medicine and were performed in accordance with the criteria outlined in the Guide for the Care and Use of Laboratory Animalsprepared by the National Academy of Sciences.

In Vitro Bile Ductule Formation Model and Evaluation of Process Formation.

Procedures for in vitro bile ductule formation have been described in detail previously.28, 29 Briefly, hepatocytes were isolated from 6- to 10-week-old male Fischer 344 rats via collagenase perfusion and low-speed centrifugation and were cultured on positively charged plastic dishes (Primaria; Becton-Dickinson Labware, Franklin Lakes, NJ) for 5 days in serum-free Williams' E medium containing 10 ng/ml epidermal growth factor (Roche Diagnostics, Mannheim, Germany) and 10−7 M insulin (Sigma) during which time they formed spheroidal aggregates. Subsequently, the aggregates were embedded in type I collagen gel matrices (Nitta Gelatin, Osaka, Japan) and cultured in Williams' E medium containing 10% fetal bovine serum, 10 ng/ml epidermal growth factor and 10−7 M insulin. The medium was changed three times a week. In some experiments, both the media and matrices were supplemented with various concentrations of one of three antibodies—9D2, U04, or TLD3A12—at the time the three-dimensional cultures were initiated. Neutralizing antibody experiments were also performed in medium containing 5% conditioned medium from nonparenchymal rat liver cells (NPC-CM), which promotes process extension27 (unpublished data).

Three-dimensional cultures were observed under a phase-contrast microscope, and cellular processes sprouting from the aggregates were classified according to previously defined criteria.27 On the indicated days, the number of cellular processes from individual aggregates was counted. The mean values were calculated by counting the processes from approximately 50 aggregates for each treatment group. Experiments were repeated three times independently, and the data are expressed as the mean ± SE. Statistical differences were evaluated via ANOVA using Stat View software (Abacus Concepts Inc., Cary, NC). A P value less than 0.05 was considered statistically significant.

Western Blot Analysis.

Liver tissues were frozen in liquid nitrogen, crushed, and vigorously vortexed in buffer containing 50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1% Triton X-100, and protease inhibitor cocktail (Sigma Chemical Company). Three-dimensional cultures were homogenized in buffer containing 10 mM Tris-HCl (pH 7.5), 158 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, and protease inhibitors. Insoluble components including collagen fibers were removed by centrifugation. The resulting lysates were separated on 10% SDS-polyacrylamide gels, transferred to Immobilon (Millipore, Bedford, MA), and analyzed via Western blotting as described previously.9 Some blots were stained with Ponceau S (Sigma Chemical Company).

Enzymatic Digestion of N-Linked Glycosylation and Polysialylation.

Protein samples of liver tissues were incubated with PNGase F or neuraminidase (New England Biolabs, Beverly, MA) according to the manufacturer's instructions. Briefly, 100 μl of liver or brain tissue that had been frozen in liquid nitrogen and crushed were vigorously vortexed in a 5× volume of the denaturing buffer supplied with the PNGase F kit and incubated at 100°C for 10 minutes. One fiftieth of the sample was incubated at 37°C for 1 hour in the presence of PNGase F (500 U) or neuraminidase (100 U). The samples were then separated on SDS-polyacrylamide gels.

Immunohistochemistry and Immunofluorescence.

For immunohistochemistry, paraformaldehyde-fixed liver tissues and acetic acid/ethanol (1:99)–fixed three-dimensional cultures were embedded in paraffin, cut into 4-μm sections, and processed as described previously.5 For immunofluorescence double-staining, paraffin sections were deparaffinized and autoclaved, and frozen sections were fixed in acetic acid/ethanol (1:99). Blocking and antibody reaction buffers were the same as those for immunohistochemistry. Sections were incubated firstly with a mixture of anti-CK19 and anti-SgIGSF (rabbit polyclonal) antibodies, and secondly with a mixture of Cy2-conjugated anti-mouse IgG and Cy3-conjugated anti-rabbit IgG antibodies. Sections were examined via epifluorescence microscopy (BX50; Olympus, Tokyo, Japan). Cy2 and Cy3 images were recorded using a CCD camera (DP70; Olympus) and were merged using DP Controller software (Olympus).

Results

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Expression of SgIGSF in Mouse Livers at Various Ages.

Western blot analysis was performed on mouse liver tissues at various ages using antibodies against the C-terminus (rabbit polyclonal) and the ectodomain (9D2 monoclonal) of SgIGSF and yielded identical results (Fig. 1A). In the perinatal period, SgIGSF was detected as two bands (Fig. 1A, left). Before and after this period, SgIGSF was detected as a single band and the expression levels were significantly lower. When liver extracts were incubated with PNGase F, SgIGSF migrated as an approximately 60-kd band at all ages examined (Fig. 1A, right). After treatment with neuraminidase, all the immunoreactive bands were reduced by approximately 10 kd and perinatal SgIGSF still migrated as a doublet (Fig. 1B, left), suggesting that SgIGSF contained almost comparable amounts of polysialic acid at all ages. Similar treatment of positive control brain extracts caused desialylation of NCAM proteins recognizable as a conversion from approximately 230 kd to approximately 150 kd (Fig. 1B, right), as reported previously.31 Thus, SgIGSF proteins appear to receive developmental stage-specific glycosylation in the liver.

thumbnail image

Figure 1. Western blot analysis of SgIGSF in mouse liver tissues of various ages. Protein samples were prepared from liver and brain tissues of mice of the indicated days of ages. After incubation with either PNGase F or neuraminidase (+), or without these enzymes (−), samples were electrophoresed on a 10% SDS-polyacrylamide gel and blotted with anti-SgIGSF antibodies against the C-terminus (C), ectodomain (N), or anti-NCAM antibody. After stripping, blots were reprobed with an anti–β-actin antibody to indicate the total amount of protein per lane. E, embryonic days.

Download figure to PowerPoint

No SgIGSF signal was detectable via immunohistochemistry in any parenchymal cells of the embryonic liver from E11.5 to E14.5 (data not shown). In E16.5 liver, SgIGSF-positive parenchymal cells emerged around the portal veins, and formed ductal structures at birth (Fig. 2A,B). SgIGSF signals were preferentially localized to the lateral membranes of ductal cells. Immunofluorescence double-staining revealed that these cells were also positive for CK19, indicating that they were BECs. Representative results are shown in Fig. 2D–K. BECs were SgIGSF-positive from the time they formed a single-layer ductal plate, and continued to stain for SgIGSF while they moved into bilayers and partly began to form ducts (Fig. 2D–G). At the next stages, lateral membrane staining with the SgIGSF antibody was prominent in distorted, developing bile ducts, whereas circular, well-formed bile ducts often displayed lower reactivity (Fig. 2H–K). The intensity of SgIGSF signals in BECs decreased drastically with age after birth, and positive cells were rarely detectable at weaning or later (Fig. 2C).

thumbnail image

Figure 2. Immunohistochemical staining of SgIGSF in mouse liver tissues. Sections of the liver at (A) embryonic day 16.5 and at (B) 1 day and (C) 4 weeks after birth were incubated with the anti-SgIGSF (rabbit polyclonal) antibody and stained with aminoethylcarbazole. The nuclei were counterstained with hematoxylin. (A,B) Boxed regions are enlarged to show staining of BEC membranes. SgIGSF-positive and -negative ducts are indicated by (B) arrowheads and (C) arrows, respectively. (D-K) Immunofluorescence double-staining of (D, E) embryonic day 16.5, (F,G) embryonic day 17.5, and (H-K) newborn livers. Frozen sections were incubated with anti-SgIGSF (rabbit polyclonal) and anti-CK19 antibodies, and stained with (D,F,H,J) Cy3 and (E, G, I, K) Cy2, respectively. (E,G,I,K) Boxed regions of the Cy3 images are merged with the Cy2 images and enlarged at the upper right corners. (Original magnification: ×400.) PV, portal vein.

Download figure to PowerPoint

SgIGSF Expression During Rat Liver Regeneration After PH.

SgIGSF expression was examined in regenerating livers of the rat AAF/PH model. We confirmed histologically that oval cells started to emerge on day 3 post-PH and formed ductules actively in the second week (data not shown). Western blot analysis consistently showed that expression levels of CK19, a marker for oval cells as well as BECs,32 increased progressively after PH and reached a maximum on day 13 (Fig. 3). SgIGSF protein levels also increased through the second week and peaked on day 9 (Fig. 3). In addition, SgIGSF slightly increased in apparent mass in the second week, probably due to additional glycosylation, because the mass was reduced to approximately 60 kd at all time points after PNGase F treatment (data not shown). In the regenerating liver, the major source of SgIGSF expression appeared to be oval cells, because SgIGSF-specific signals were distributed in the periportal areas and the surrounding parenchyma on day 9, and these positive cells were smaller than hepatocytes and possessed a round nucleus (Fig. 4A,B). This was confirmed via immunofluorescence double-staining, which revealed that almost all SgIGSF-positive cells were also positive for CK19 (Fig. 4C,D). On day 13, periportal ductular oval cells and midlobular trabecular oval cells were still CK19-positive but were often negative for SgIGSF (Fig. 4E–H). These observations are consistent with the results of the Western blot analysis showing that SgIGSF levels declined by day 13, while CK19 levels remained high (Fig. 3). These results may reflect a decrease in the number of newly synthesized oval cell ductules and an increase in the number of oval cells committed to hepatocytic differentiation after day 9.

thumbnail image

Figure 3. Western blot analysis of CK19 and SgIGSF in regenerating liver after PH. Rat liver lysates of the AFF/PH model were prepared on the indicated days after PH, electrophoresed on a 10% SDS-polyacrylamide gel, and blotted with anti-CK19 or anti-SgIGSF (rabbit polyclonal) antibodies. After stripping, the blot was reprobed with an anti–β-actin antibody to indicate the total amount of protein per lane.

Download figure to PowerPoint

thumbnail image

Figure 4. Expression of SgIGSF is restricted to oval cells in the AFF/PH model. (A) Sections of paraformaldehyde-fixed liver tissues from day 9 after PH were incubated with anti-SgIGSF and stained with aminoethylcarbazole. The nuclei were counterstained with hematoxylin. (B) Enlargment of the boxed region in (A). Arrowheads indicate SgIGSF-positive ductular oval cells. (C-H) Frozen sections of livers from (C,D) day 9 and (E-H) day 13 after PH were double-stained with anti-SgIGSF and anti-CK19 antibodies and were visualized with (C,E,G) Cy3 and (D,F,H) Cy2, respectively. Arrowheads in (C) and (D) indicate SgIGSF/CK19-double positive ductular oval cells. Arrowheads in (E–H) indicate CK19-positive/SgIGSF-negative oval cells. Arrows in panels G and H indicate the direction of central veins. (Original magnification: ×200 [panel A], ×400 [panels B-H].) CV, central vein; PV, portal vein.

Download figure to PowerPoint

SgIGSF Expression in Human Liver Development and Regeneration.

We examined SgIGSF expression in developing and regenerating livers of humans via immunofluorescence double-staining. In late–gestational stage fetuses (Fig. 5A), the anti-SgIGSF antibody clearly stained the lateral membranes of CK19-positive epithelial cells lining the immature bile ducts (Fig. 5C,D). Sections of newborn livers yielded similar results, whereas there were few parenchymal cells positive for SgIGSF in disease-free adult livers (data not shown).

thumbnail image

Figure 5. Expression of SgIGSF in human livers. Liver tissues were removed from (A,C,D) a fetus at 24 weeks' gestation and (B, E-J) a male adult with cirrhosis. Sections of paraformaldehyde-fixed tissues were double-stained with anti-SgIGSF and anti-CK19 antibodies and were visualized with (C,E,G,I) Cy3 and (D,F,H,J) Cy2, respectively. (A,B) After sections were examined via epifluorescence microscopy, they were stained with hematoxylin-eosin. Panels G through J are high-power images of the left and right boxed regions in panels E and F, respectively. The insets of panels C, G, and F are Cy3-Cy2 merged images of the boxed regions in panels C and D, G and H, and I and J, respectively. (Original magnification: ×400 [panels A, C, D, G-J]; ×200 [panels B, E, F].) HC, hepatic cord; IA, interlobular artery; PM, portal mesenchyme.

Download figure to PowerPoint

In cirrhotic livers (Fig. 5B), the SgIGSF antibody exclusively stained atypical ductules at the edge of the fibrous septum (Fig. 5E,G). These ductules were also positive for CK19 (Fig. 5F,H). In contrast, a somewhat larger duct near the center of the fibrous septum expressed high levels of CK19 but markedly reduced levels of SgIGSF (Fig. 5I,J). Because this duct accompanied an artery, it was regarded as a physiologically mature IHBD, not an actively proliferating ductule.

SgIGSF Expression and Its Role in an In Vitro Ductule Formation Model.

The above experiments consistently suggested that SgIGSF might play a role in bile duct/ductule formation. We examined this hypothesis using an in vitro model in which mature hepatocytes are allowed to aggregate on plastic dishes (on-dish culture) and aggregates are then cultured in collagen gels (in-gel culture).27, 28 Western blot analysis revealed that primary hepatocytes did not express either SgIGSF or CK19 at the time of culture initiation. After 3 days, SgIGSF expression was detectable in hepatocyte aggregates in the on-dish cultures, and its level increased slightly during the first week of in-gel culture (Fig. 6A). In contrast, CK19 expression was not detectable until the aggregates were transferred to the in-gel cultures, and its level increased in the second week of in-gel culture (Fig. 6A). GAPDH levels increased in long term in-gel cultures, while a set of bands cross-reactive to the SgIGSF antibody were present at constant intensity throughout the culture, indicating, together with Ponceau staining, equal loading of the protein lysates into each lane of the gel (Fig. 6A). However, Western blot analysis was not suitable to compare expression levels in cultures incubated for 3 weeks or longer, because these cultures contained a considerable proportion of activated stellate cells.

thumbnail image

Figure 6. Expression of SgIGSF and its role in in vitro ductule formation. In this model, hepatocytes were cultured on positively charged dishes (on-dish culture) to form spheroidal aggregates, which were subsequently cultured in collagen gels (in-gel culture) to extend epithelial processes and form ductules. (A) Expression levels of SgIGSF and CK19 in the cultures were examined via Western blot analysis. Protein lysates were prepared from the on-dish and in-gel cultures at various times, electrophoresed on a 10% SDS-polyacrylamide gel, and blotted with the anti-SgIGSF (rabbit polyclonal; uppermost panel) or anti-CK19 (second panel) antibodies. After stripping, the blot was probed again with the anti-GAPDH antibody (third panel). The fourth panel shows a set of bands which are recognized by the anti-SgIGSF antibody, and the fifth a part of the membrane staining with Ponceau S. (B) Immunohistochemical staining of SgIGSF and CK19 in the aggregates, (a-c) epithelial cell extensions, and (d-g) ductules. The in-gel cultures (a, b, 3-day; c, 7-day; d-g; 31-day) were fixed with acetic acid-ethanol, embedded in paraffin, and sectioned. Some sections were incubated with (a, c, d) anti-SgIGSF or (b) anti-CK19 antibodies and stained with aminoethylcarbazole. The nuclei were counterstained with hematoxylin. Another section was double-stained with anti-SgIGSF and anti-CK19 antibodies and was visualized with (e) Cy3 and (f) Cy2, respectively. Both images were merged in panel g. Arrowheads in panels b and c indicate positive staining in the cytoplasm and on the cell membrane, respectively. (Original magnification ×400.) (C) To examine whether an SgIGSF neutralizing antibody (9D2) inhibits process formation from the aggregates, in-gel cultures were initiated in the presence of U04, (a, c) a control antibody, or (b, d) 9D2 at a concentration of 10 μg/ml. In some cases (c, d), the culture media were supplemented additionally with NPC-CM at a concentration of 5%. After 2 days, the cultures were observed under a phase-contrast microscope. Representative images are shown. In the presence of 9D2, the aggregates sprouted only a small number of short processes accompanying the appearance of small globoid debris (b, d; arrowheads). (D) To quantify the inhibitory effect, in-gel cultures were started in the presence of (a) U04, (a) 9D2, or (b) TLD3A12, a PECAM-1 blocking antibody, at various concentrations as indicated. Some cultures were supplemented with NPC-CM at a concentration of 5% (+NPC-CM). After 2 days, the number of spiny processes sprouting per aggregate was averaged from the observation of approximately 50 aggregates for each treatment. The mean values of three independent experiments were plotted in panels a and b, with bars indicating the SE. *P < 0.01, **P <0.001 via ANOVA when compared with values in the presence of U04 (10 μg/ml).

Download figure to PowerPoint

Immunohistochemistry was performed on in-gel cultures. At day 3, SgIGSF was detected clearly between hepatocytes in the aggregates and weakly between epithelial cells extending from the aggregates (Fig. 6Ba). At day 7, when the aggregates were reduced in size as the processes elongated, SgIGSF signals were less clear in the aggregates, but more intense on the lateral membranes of epithelial cells extending from the aggregates (Fig. 6Bc). At 3 weeks, the position of the original aggregates was not clear because the epithelial cells within the gels were all forming ductular processes, in which SgIGSF signals were localized intensely on the lateral membranes (Fig. 6Bd, Be, Bg). In the following 2 months, ductular structures continued to develop and stain positively for SgIGSF (data not shown). As we reported previously,27, 28 CK19 signals were not detected in the early, day 3 aggregates, but were detected in the extended epithelial cell cultures (Fig. 6Bb) and thereafter in the processes and ductules (Fig. 6Bf, Bg).

We examined whether blocking SgIGSF influenced the growth of cellular processes. Hepatocyte aggregates were embedded in collagen gels containing various concentrations of neutralizing antibody against SgIGSF (9D2) or PECAM-1 (TLD3A12), or control chicken immunoglobulin Y (U04). Neither the control antibody (Fig. 6Ca) nor the PECAM-1 blocking antibody (data not shown) affected process formation. In contrast, the SgIGSF blocking antibody significantly suppressed process formation and resulted in the emergence of small globular debris surrounding the spheroidal aggregates (Fig. 6Cb), suggesting abortive process formation. Similar results were obtained in the presence of NPC-CM (Fig. 6Cc, Cd). The inhibitory effects of the SgIGSF blocking antibody were distinct for the first 3 days but became less clear thereafter (data not shown), probably because of antibody degradation and incomplete accessibility within the gel. We counted the number of spiny processes per aggregate after 2 days of in-gel cultures. On average, 3.7 processes were detectable per aggregate in the presence of control antibody, whereas the SgIGSF blocking antibody reduced the process number by 80% (0.8 processes per aggregate at 20 μg/ml) and the reductions were dose-dependent (Fig. 6Da). Similar inhibitory effects were detected in cultures supplemented with NPC-CM (Fig. 6Da). The PECAM-1 blocking antibody did not affect process growth (Fig. 6Db).

Discussion

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

In this study, we found the first evidence that, in mouse livers, BECs express high levels of SgIGSF during IHBD development. SgIGSF was detectable as early as the ductal plate stage, but levels were drastically decreased after birth. In rat livers of the AAF/PH model, SgIGSF was expressed strongly by oval cells which formed ductules and trabecules. In both developing and regenerating livers, the duration of SgIGSF upregulation was just a few days, during which time the BECs and oval cells were actively forming ducts and ductules. Analogous results were obtained in human livers. SgIGSF expression was clearly detectable in developing bile ducts and ductules of perinatal and cirrhotic livers, respectively, but was not found in physiologically mature IHBDs. These expression profiles suggest that SgIGSF may be involved in bile duct/ductule formation.

Interestingly, the increase in SgIGSF levels during liver development and regeneration was accompanied by the transient appearance of isoforms with different molecular weights, which appeared to be the result of altered glycosylation, including polysialylation. Developmental stage-dependent alterations in glycosylation levels are also observed in SgIGSF proteins expressed on spermatogonia10 and lung alveolar cells.5 Embryonic and adult neurons contain highly and minimally sialylated NCAM, respectively, which endows the neurons with a great capacity for plasticity and stability of cell–cell interactions.33, 34 Hepatic SgIGSF contained polysialic acid, although the amount was much smaller than that of embryonic NCAM. Multiple functions conferred on SgIGSF by divergence in glycosylation levels may be important for duct/ductule development. However, the present results could not exclude the possibility that the doublet band detected in perinatal and post-PH livers represented new splice variants of SgIGSF, although RT-PCR failed to detect any isoforms reported previously35, 36 (Ito A, unpublished data). Further investigation is necessary to understand the mechanism and consequence of SgIGSF modification during liver development and regeneration.

In the in vitro ductule formation model, SgIGSF was localized to the lateral membranes of epithelial cells migrating from the hepatocytic aggregates and arranged in ductules, and was necessary for process extension from the aggregate. Masuda et al.,37 using a three-dimensional culture system of MDCK monolayer cysts, demonstrated recently that forced expression of SgIGSF in MDCK cells suppressed hepatocyte growth factor–induced epithelial-to-mesenchymal transitions. In MDCK cell cysts, SgIGSF was localized exclusively to the lateral membranes. The similar subcellular localizations of SgIGSF in MDCK cysts and bile duct–like structures developed in our culture model suggest that SgIGSF-mediated cell–cell interactions may contribute to induction and maintenance of epithelial alignment.

Necl-5 has been characterized previously as a rat hepatic oncofetal membrane glycoprotein.38, 39 Necl-5 is present at exceedingly low levels in adult liver hepatocytes, but is transiently upregulated during fetal liver development and during liver regeneration after PH and carbon tetrachloride administration.40 Expression profiles of Necl-5 in hepatocytes seemingly resemble those of SgIGSF in BECs. Hepatocytes begin to express Necl-5 immediately after isolation and maintain high levels of expression in culture for 7 days.38 Because our experiments showed that SgIGSF expression was detectable in hepatocyte aggregates at days 3 and 5, hepatocytes appear to have the potential to express both Necl-5 and SgIGSF after 3 days in culture. Previous work38 has indicated that Necl-5 is located exclusively on the cell membranes of 24-hour hepatocyte cultures. However, a closer appraisal of the data in this work shows that Necl-5 was also present in the cytoplasm after 3 and 7 days, suggesting that internalization of Necl-5 from the cell surface into the cytoplasm can occur in hepatocytes after extended periods of culture. Necl-5 internalization is known to result from its heterophilic binding with nectin-3,41 and our preliminary data showed that Necl-5 actually binds SgIGSF in dynamic adhesive interactions between Sertoli and spermatogenic cells (unpublished data). Thus, these data may suggest that heterophilic binding of both molecules might take place among cultured hepatocytes, and consequently lead to Necl-5 internalization. Although this hypothesis needs to be evaluated in future work, Necl-5 downregulation coupled with SgIGSF prevalence on the cell surface may trigger bile duct–like differentiation of hepatocytes in culture.

In conclusion, the present study identified SgIGSF as a novel BEC adhesion molecule and revealed that expression of SgIGSF was up-regulated during active duct/ductule formation, although there is a need for clarifying a functional role of SgIGSF in vivo. Further studies are needed to address whether SgIGSF binds homophilically or heterophilically between BECs and to clarify its functional significance in vivo. However, the present study suggests that SgIGSF-mediated cell–cell interactions may play a role in a ductular alignment of BECs, an event essential for liver development and regeneration.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

We thank A. Kawashima (Kobe University Graduate School of Medicine), M. Kakihana (Hyogo Medical School), and E. Kumagai (Akita University School of Medicine) for technical assistance.

References

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References