SEARCH

SEARCH BY CITATION

Keywords:

  • cell–cell interactions;
  • endothelial cells;
  • hepatoblasts;
  • immunomagnetic separation;
  • sinusoids;
  • VEGF

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Author contributions
  10. References
  11. Supporting Information

Previous studies have shown that various cell–cell interactions between hepatoblasts and nonparenchymal cells, including sinusoidal endothelial cells and stellate cells, are indispensable for the development of fetal murine hepatic architecture. The present study was undertaken to determine the effects of hepatoblasts on the sinusoidal structural formation using a culture system of fetal mouse livers. Primitive sinusoidal structures extensively developed in fetal livers, and were composed of LYVE-1- and PECAM-1-positive endothelial cells, desmin-positive stellate cells and F4/80-positive macrophages. When fetal liver cells at 12.5 days of gestation were cultured in vitro, hepatoblasts spread on glass slides and gave rise to hepatocytes on day 5. Desmin-positive stellate cells also spread on the glass slides. PECAM-1-positive endothelial cells became slender and developed into anastomosing capillary networks. When fetal liver cells were cultured without hepatoblasts, which were excluded by an immunomagnetic method using anti-E-cadherin antibodies, endothelial cells had impaired growth and capillary formation. These results demonstrated that capillary formation of endothelial cells was induced by the presence of hepatoblasts. VEGF and the conditioned medium containing humoral factors produced by hepatoblasts/hepatocytes did not induce capillary formation of endothelial cells in cultures of nonparenchymal cells, although they significantly increased the number of endothelial cells on the glass slides. The presence of hepatoblasts also significantly stimulated expression of CD32b mRNA, which is a sinusoidal endothelial marker. Hepatoblasts may work as a positive stimulator of sinusoid morphogenesis and maturation in liver development, in which a signal other than VEGF may play a decisive role, together with VEGF.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Author contributions
  10. References
  11. Supporting Information

The liver is a well vascularized organ in which the blood from the portal vein and hepatic artery flows into the hepatic vein via abundant sinusoid channels. During embryogenesis, the hepatic vascular system develops with liver primordium formation; hepatic cords of hepatoblasts extending from the liver diverticulum invade the subjacent septum transversum mesenchyme, omphalomesenteric veins and posterior cardinal veins (Du Bois, 1963; Severn, 1972; Matsumoto et al. 2001; Nonaka et al. 2007; Sugiyama et al. 2010), which leads to the formation of an immature liver having primitive sinusoid structures (Shiojiri & Sugiyama, 2004). Several markers for sinusoidal endothelial cells, including LYVE-1 and stabilin2, which are respectively an endocytotic receptor for hyaluronan and a fasciclin-domain-containing hyaluronan receptor that binds to a variety of molecules, have recently been described (Banerji et al. 1999; Nonaka et al. 2004; Hansen et al. 2005). Stabilin2- and PECAM-1-positive endothelial cells are detectable along the liver diverticulum and hepatic cords, and expression of these markers persists during fetal liver development (Matsumoto et al. 2001; Nonaka et al. 2007; Sugiyama et al. 2010). LYVE-1 expression in hepatic sinusoidal endothelial cells is detectable a little later than that of PECAM-1 and stabilin2 (Nonaka et al. 2007). Morphological maturation of sinusoidal endothelial cells such as their characteristic fenestrated cytoplasm commences at mid-gestational stages and the SE-1 antigen, their specific molecular marker, is expressed at these stages of rat liver development (Enzan et al. 1997; Morita et al. 1998; Yoshida et al. 2007). The SE-1 antigen has recently been demonstrated to be CD32b, which is a low affinity Fcγ receptor (March et al. 2009).

Molecular mechanisms underlying cell–cell interactions involved in the growth and maturation of hepatoblasts/hepatocytes have been extensively studied, and several molecules such as tumor necrosis factor (TNF)α, oncostatin M and Wnt9a, which are produced by nonparenchymal cells, mediate the cell–cell interactions (Kamiya et al. 1999; Kamiya & Gonzalez, 2004; Matsumoto et al. 2008; Lemaigre, 2009). However, the importance of cell–cell interactions in hepatic vascular development has been hardly noted. VEGF signaling is essential for hepatic vascular development because the inactivation of the gene for Flk-1, one of the VEGF receptors, results in agenesis of vascular development and anomaly of liver primordium formation (Matsumoto et al. 2001). VEGF, which may be provided by hepatoblasts/hepatocytes and stellate cells, activates the Flk-1 receptor of neighboring sinusoidal endothelial cells to induce their growth and morphogenesis in the early stages of liver development (Sugiyama et al. 2010). However, it remains to be experimentally studied whether fetal hepatoblasts/hepatocytes or stellate cells play a pivotal role in sinusoid development. Cocultivation of adult sinusoidal endothelial cells with primary hepatocytes or with a hepatocellular model of hepatocytes and fibroblasts can prolong their characteristic phenotype, in which paracrine signals such as VEGF and extracellular matrices may be involved (DeLeve et al. 2006; Hwa et al. 2007; March et al. 2009).

In the present study, using an immunomagnetic method, E-cadherin-positive hepatoblasts were excluded from fetal liver cell cultures to examine their role in vascular development. The exclusion of hepatoblasts resulted in poor growth and gene expression of hepatic constituent cells, including sinusoidal endothelial cells. Capillary formation was severely impaired in the absence of hepatoblasts. Their effects could not be restored by the addition of a conditioned medium prepared from fetal liver cell cultures or VEGF.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Author contributions
  10. References
  11. Supporting Information

Animals

C3H/HeSlc mice (SLC, Hamamatsu, Japan) were used. Animals were kept under a controlled light/dark cycle and fed a standard chow diet and water ad libitum. They were mated during the night, and noon of the day a vaginal plug was found was considered 0.5 days of gestation (E0.5). Mouse fetuses at E12.5 were used for cell culture studies. E12.5, E14.5, E17.5, neonatal (P0; postnatal 0 day), P14 and adult livers were used for PECAM-1, LYVE-1, desmin and F4/80 immunohistochemistry and histology. All animal experiments were carried out in compliance with the ‘Guide for Care and Use of Laboratory Animals’ of Shizuoka University.

Exclusion of E-cadherin-positive hepatoblasts

Exclusion of hepatoblasts from fetal liver cell suspensions was performed using an immunomagnetic method according to the manufacturer’s instructions. Immunomagnetic beads (Dynal, A. S., Oslo, Norway) coupled with sheep anti-rat IgG antibodies, were incubated with 0.1% gelatin/20 mm Tris/150 mm NaCl/10 mm CaCl2 (TBS) to suppress their nonspecific binding to cells (Murphy et al. 1992; Nitou et al. 2002). After washing in 1% bovine serum albumin (BSA)/TBS, the beads were incubated with a rat anti-mouse E-cadherin antibody [ECCD1 (1/1000 dilution); Takara Bio Inc., Otsu, Japan] at room temperature for 30 min, washed twice with 10% fetal bovine serum (FBS) (Sigma Chemical Co., St. Louis, MO, USA)/DM-160, and then used for cell exclusion experiments.

Livers were dissected out from E12.5 mouse fetuses and then diced. Liver tissues were treated with 1000 U mL−1 dispase (Godo Shusei Co. Ltd., Tokyo, Japan) dissolved in 10% FBS/Dulbecco’s modified Eagle’s medium (DMEM; Gibco, Grand Island, NY, USA) for 60 min at 37 °C (Nitou et al. 2002; Sugiyama et al. 2007, 2010). After gentle pipetting with a Pasteur pipette, undigested tissues were removed from the cell suspension by filtration using a nylon mesh filter (132-μm pore size) (Nihon Rikagaku Kikai Co. Ltd., Tokyo, Japan). Cells were collected by centrifugation at 50 g for 10 min and the resultant cellular pellet was washed twice with DM-160 (Kyokuto Seiyaku Co. Ltd., Tokyo, Japan; Takaoka & Katsuta, 1971, 1975) containing 10% FBS and 0.01% deoxyribonuclease I (Worthington Biochem. Corp., Freehold, NJ, USA). Cells were resuspended in 10% FBS/DM-160 (106 cells mL−1) after washing. The viability was more than 95% by the trypan blue exclusion test. The cell suspension was incubated with the anti-E-cadherin antibody-coupled beads (10-fold the cell number) for 30 min at 4 °C (Nitou et al. 2002). As a control, cell suspensions were incubated with immunomagnetic beads that were pretreated with 1% BSA/TBS in place of the anti-E-cadherin antibody. Using a magnetic particle concentrator (MPC), E-cadherin-positive cells were excluded from the cell suspension. The E-cadherin-positive cell-depleted fraction (nonparenchymal fraction) was washed twice with 10% FBS/DM-160 by centrifugation, and resuspended in 10% FBS/DM-160 containing 10−7 m dexamethasone, 20 μg mL−1 heparin and antibiotics. This cell suspension was adjusted to the original volume before mixing with immunomagnetic beads. The bead fraction, in which hepatoblasts were contained, was again incubated with dispase for 60 min at 37 °C to detach the beads from separated hepatoblasts (Nitou et al. 2002). Following the reconcentration of cell-free beads with the MPC, the resultant cell suspension was centrifuged at 50 g for 10 min. The cellular pellet was washed and resuspended in 10%FBS/DM160 containing dexamethasone, heparin and antibiotics (hepatoblast fraction).

Cell culture

Cell suspensions of 70 μL, from which E-cadherin-positive cells were excluded, and control cell suspensions (remixed cell suspensions of the nonparenchymal fraction and hepatoblast fraction and liver cell suspensions without the immunomagnetic separation steps) were cultured for 5 days on the glass area of Teflon-coated slides (AR Brown Co. Ltd., Tokyo, Japan), which were coated with type I collagen (20 μg cm−2), at 37 °C in a water-saturated atmosphere containing 5% CO2. The medium was changed on days 1 and 3. In some experiments, VEGF (20–200 ng mL−1; Genzyme-Techne Corp., Boston, MA, USA) or a conditioned medium prepared from E12.5 liver cell cultures between days 3 and 4 was added to the culture medium (at the concentration of 50%), which was specified in the data. After 5 days, cultured cells were fixed in cold acetone (−20 °C) for 10 min for hematoxylin and eosin (H–E) staining and immunohistochemistry.

Immunohistochemistry

Liver tissues for PECAM-1, LYVE-1, desmin and F4/80 immunohistochemistry were frozen in liquid nitrogen. Frozen sections were cut at 8 μm thickness and fixed in cold acetone (−20 °C) for 10 min.

Hydrated sections and cultured cells were incubated for 1 h at room temperature with the primary antibodies listed in Supporting Information Table S1. After thorough washing with PBS, sections were incubated with a Cy3- or fluorescein-labeled donkey anti-rabbit or rat IgG antibody (Jackson ImmunoResearch Lab., West Grove, PA, USA) (1/500 dilution for the Cy3-labeled antibody and 1/50 dilution for the fluorescein-labeled antibody) for 1 h at room temperature, washed again, and mounted in buffered glycerol containing p-phenylenediamine (Johnson & de C Nogueira Araujo, 1981). Double immunofluorescent analyses were carried out using these secondary antibodies. Control incubations were carried out in PBS containing 1% BSA in place of the primary antibodies. Sections were also stained with H–E.

RT-PCR

Total RNA was extracted from developing livers and cell culture samples on days 1, 3 and 5 using Isogen (Nippon Gene, Tokyo, Japan). Complementary DNA was synthesized from total RNA (0.5 μg) in 10 μL of reaction mixture containing 2.5 μm oligo dT primer, 0.25 mm dNTP, 2 U μL−1 RNase inhibitor, and 10 U μL−1 PrimeScript® Reverse Transcriptase (Takara Bio Inc.), according to the manufacturer’s instructions.

The PCR reaction was conducted in 10 μL of the reaction mixture, using Ex-Taq DNA polymerase (0.025 U μL−1; Takara Bio Inc.). Primers listed in Supporting Information Table S2 were used at 0.5 μm. After various dilutions of template cDNA, we optimized the concentration for each primer. In these concentrations, amplification by PCR did not reach a plateau and could be used for semiquantitative analysis. PCR cycles were as follows: initial denaturation at 95 °C for 5 min, followed by 25–35 cycles at 94 °C for 30 s, at 60 °C for 30 s, at 72 °C for 1 min, and final extension at 72 °C for 7 min. PCR products were separated by 2% agarose gel electrophoresis. After band data were obtained, the signal intensity was quantified with imagej software (National Institutes of Health, Bethesda, MD, USA).

Statistics

All data are expressed as the mean ± SD. Statistical significance was determined using the Student t-test (Microsoft excel).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Author contributions
  10. References
  11. Supporting Information

Expression of endothelial markers during liver development

Throughout liver development, vascular structures, including portal veins, central veins and sinusoidal structures, were well developed even in immature fetal livers (Fig. 1A,B). In E12.5 livers, LYVE-1 expression was detectable in all endothelial cells of sinusoids and central veins (Fig. 1A). Endothelial cells of portal veins were negative for LYVE-1 immunohistochemistry throughout liver development (Fig. 1B–F). The LYVE-1-positive endothelial cells of the sinusoids were located very close to desmin-positive stellate cells and F4/80-positive macrophages throughout liver development (Fig. 2A–L). Hepatoblasts and hepatocytes resided near sinusoidal structures (inset in Fig. 1A). LYVE-1 signals were strong in sinusoidal endothelial cells at E17.5 and P0, but endothelial cells of central veins were weakly positive (Fig. 1C,D). At P14 and adult stages, endothelial cells of central veins became negative for LYVE-1, and only sinusoidal endothelial cells in mid-zonal regions of the lobule were positive (Fig. 1E,F).

image

Figure 1.  Development of vascular system and expression of LYVE-1 (red) during liver development. (A) E12.5 liver. (B) E14.5 liver. (C) E17.5 liver. (D) P0 liver. (E) P14 liver. (F) Adult liver. Cytokeratin immunostaining (green) and DAPI staining of nuclei (blue) are also shown. LYVE-1-positive primitive sinusoids (arrows) are well developed at E12.5, E14.5, E17.5 and P0 (A–D). Cytokeratin-positive hepatoblasts reside close to LYVE-1-positive sinusoidal endothelial cells (inset in A). Although endothelial cells of portal veins (PV) are negative for LYVE-1 expression, those of central veins (CV) are weakly positive in fetal and neonatal livers (B–D). At P14, LYVE-1 expression is heterogeneous and is detectable in sinusoidal endothelial cells in mid-zonal regions of liver lobules (E). Also in adult liver, LYVE-1 expression of sinusoidal endothelial cells is limited (F). Arrowhead indicates LYVE-1-positive endothelial cells of a lymphatic vessel. DP, ductal plates. Scale bars: 100 μm.

Download figure to PowerPoint

image

Figure 2.  Double immunofluorescent analyses of LYVE-1 (red) and desmin (green) (A–F), and F4/80 (red) and desmin (green) (G–L) expression during liver development. (A,B,G,H) E12.5 liver. (C,D,I,J) E17.5 liver. (E,F,K,L) Adult liver. (B,D,F,H,J,L) Magnified pictures of rectangles in (A,C,E,G,I,K), respectively. LYVE-1-positive sinusoidal endothelial cells are located close to desmin-positive stellate cells during development (A–F). F4/80-positive macrophages are also located close to desmin-positive stellate cells during development (G–L). CV, central vein; PV, portal vein. Scale bars: 50 μm.

Download figure to PowerPoint

The PECAM-1 signals were seen in all vascular endothelial cells, megakaryocytes and some hemopoietic cells in fetal mouse livers (Fig. 3A,B). The immunostaining was strong in sinusoidal endothelial cells of E12.5 and E14.5 livers, but those in E17.5 livers became weaker (Fig. 3B). In P0 livers, PECAM-1 expression was strong in endothelial cells of both portal and hepatic veins but became significantly low in sinusoidal endothelial cells (Fig. 3C). In adult livers, sinusoidal endothelial cells did not express or only very weakly expressed PECAM-1, and endothelial cells of portal veins, central veins and hepatic arteries strongly expressed it (Fig. 3D).

image

Figure 3.  PECAM-1 expression in endothelial cells during liver development. (A) E12.5 liver. (B) E17.5 liver. (C) P0 liver. (D) Adult liver. Endothelial cells of primitive sinusoids (arrows) at E12.5 are PECAM-1-positive (A). Megakaryocytes (asterisk) are also PECAM-1-positive. Sinusoid structures (arrows) are elongated at E17.5 in section (B). In P0 liver, PECAM-1 signals are lower in sinusoidal endothelial cells (C). Although the expression of PECAM-1 is perfectly downregulated in sinusoids (arrows) in the adult stage, endothelial cells of the hepatic artery and biliary vessels are strongly PECAM-1-positive in addition to those of portal and central veins (D). BD, bile duct; CV, central vein; HA, hepatic artery; PV, portal vein; V, portal vein or central vein. Scale bars: 100 μm.

Download figure to PowerPoint

RT-PCR analysis demonstrated that expression of Pecam1 and Lyve1, which was high in fetal livers, decreased in adult livers (Figs 4A,B and S1). CD32b expression, a marker for mature sinusoidal endothelial cells, increased during liver development (Figs 4D and S1). Stabilin2 expression was also higher in adult livers than in fetal livers (Figs 4C and S1).

image

Figure 4.  Expression of endothelial cell markers during liver development. (A–D) Semiquantitative RT-PCR analyses of Pecam1, Lyve1, Stab2 and CD32b mRNA expression in E12.5, E17.5, P0 and adult livers. The numbers in parentheses denote those of samples examined. Pecam1 and Lyve1 expression decreased in postnatal livers (A,B). CD32b and Stab2 expression increased during liver development (C,D).

Download figure to PowerPoint

Capillary formation in fetal liver cell cultures

When E12.5 liver cells were cultured in vitro, hepatoblasts formed spherical aggregates on day 1, gradually spread on the glass slides and gave rise to large hepatocytes (Fig. S2A–F), expressing carbamoylphosphate synthase I (CPSI; urea cycle enzyme) on day 5 (Fig. S3A). Endothelial cells formed small aggregates on day 1, and capillary networks after day 3, which were not lined by desmin-positive stellate cells (Fig. 5A,C,E,F). The capillary networks were observed only on hepatocyte sheets (Fig. S4). Desmin-positive stellate cells spread on the glass slides like hepatoblasts/hepatocytes (Fig. 5B,D–F). Macrophages expressing F4/80 antigen were also detectable in these cultures, the distribution of which appeared to be random (Supporting Information Fig. S3B).

image

Figure 5.  Development of capillary networks in cultures of E12.5 liver cells. (A,B) Day 1. (C–F) Day 5. (F) Magnified picture of the rectangle in (E). Small cell aggregates of PECAM-1-positive endothelial cells observed on day 1 (A) form capillary networks on the glass slide on day 5 (C). (A, inset) PECAM-1-positive endothelial cells (red) intermingling with hepatoblasts (green; cytokeratin). (C, inset) Endothelial cells forming capillary networks are LYVE-1-positive. Desmin-positive stellate cells form aggregates (arrowhead) with hepatoblasts on day 1 (B) but spread extensively on the glass slide on day 5 (D). (B, inset) Desmin-positive cells (green; arrow) are incorporated into aggregates of hepatoblasts (red; E-cadherin). Desmin-positive stellate cells do not align with PECAM-1-positive endothelial cells, which form capillary networks (E,F). Scale bars:100 μm (A–E).

Download figure to PowerPoint

RT-PCR analyses showed that CD32b expression and Actb expression, which indicated the growth of fetal liver cells, increased with culture time (Fig. 6D,E and Supporting Information S5). Pecam1 expression decreased, whereas Lyve1 expression did not show any remarkable change during culture (Figs 6A,B and S5). Stabilin2 expression decreased on day 5 (Figs 6C and S5).

image

Figure 6.  Expression of endothelial cell markers in cultures of E12.5 liver cells. (A–E) Semiquantitative RT-PCR analyses of Pecam1, Lyve1, Stab2, CD32b and Actb mRNA expression on days 1, 3 and 5, respectively. The numbers in parentheses denote those of samples examined. Although Pecam1 expression decreases with culture, Lyve1 expression does not show any significant change (A,B). Stab2 expression decreases on day 5 (C). CD32b expression increases with culture (D). Increased Actb expression is also seen (E).

Download figure to PowerPoint

Exclusion of E-cadherin-positive hepatoblasts from E12.5 liver cell suspensions

E12.5 liver cell suspensions contained single cells and cellular aggregates, which were roughly thought to be nonparenchymal cells and hepatoblasts, respectively (Supporting Information Fig. S6A). When E-cadherin-positive hepatoblasts were excluded from E12.5 liver cell suspensions by the immunomagnetic method, no cellular aggregate in nonparenchymal cell fractions was observed with phase-contrast microscopy (Fig. S6C). Hepatoblast fractions contained only cellular aggregates of hepatoblasts (Fig. S6B). RT-PCR analysis of E-cadherin mRNA indicated that hepatoblasts were excluded from the nonparenchymal fractions (Figs 7 and S6D). In hepatoblast fractions, there was little contamination by endothelial cells expressing Pecam1, but contamination by stellate cells was often observed in the RT-PCR analysis of Pecam1 and desmin mRNA expression (Figs 7 and S6D).

image

Figure 7.  RT-PCR analyses of cell-type-specific markers in each fraction separated by the immunomagnetic bead method from E12.5 liver cell suspensions. Hepatoblast fraction expresses Cdh1 but not Pecam1. Desmin expression in the hepatoblast fraction appears to be stronger than that in the nonparenchymal cell fraction. The numbers in parentheses denote cycles of each PCR.

Download figure to PowerPoint

Culture and capillary formation of nonparenchymal cells without hepatoblasts

Nonparenchymal cells of E12.5 livers, including PECAM-1-positive endothelial cells, desmin-positive stellate cells and F4/80-positive cells, were poorly adhesive to the glass slides, which were collagen-coated, when hepatoblasts were excluded by the immunomagnetic method (Fig. 8A,G). In the control culture (remixed culture), in which hepatoblasts and nonparenchymal fractions were recombined and cultured, capillary networks were observed, as seen in the crude cultures of fetal liver cells (Fig. 8E–H). The capillary formation always occurred on hepatocyte sheets. When the concentration of hepatoblasts was reduced to 1/3 or 1/10 in remixed cultures, capillary formation was reduced (Supporting Information Table S3).

image

Figure 8.  Cultures of E12.5 liver cells from which E-cadherin-positive cells are excluded (A–D) and control cultures (E,F) on day 3. (A–C, E, F) PECAM-1 immunohistochemistry for nonparenchymal cell cultures. (D) Double immunohistochemistry for PECAM-1 (red) and desmin (green) in nonparenchymal cell cultures. The presence of the conditioned medium and VEGF (20 ng mL–1) in the culture medium increases the number of endothelial cells (arrows) on the glass slides (B,C) compared with the culture without these factors (A). Endothelial cells have a round morphology. Desmin-positive stellate cells decrease in cultures of nonparenchymal cell fractions (D) compared with the crude cultures shown in Fig. 5. Endothelial cells in control cultures (remixed and crude cultures) aggregate and became slender on the glass slides (E, F; arrowheads). Blue, DAPI staining. Scale bars: 50 μm. (G) The numbers of PECAM-1-positive endothelial cells in cultures of nonparenchymal cell fractions and controls, the medium of which contained the conditioned medium and VEGF (20 ng mL−1). (H) Capillary formation of endothelial cells in cultures of nonparenchymal fractions and controls. In the absence of hepatoblasts, endothelial cells do not form capillary networks.

Download figure to PowerPoint

The addition of the conditioned medium to the culture medium for nonparenchymal cells, which were prepared from conventional cultures of fetal liver cells and may have contained various humoral factors produced by hepatoblasts/hepatocytes, had some positive effects on the adhesion and survival of endothelial cells on the glass slides (Fig. 8B–D). However, it failed to induce nonparenchymal cells to form capillary structures without hepatoblasts (Fig. 8B–D). VEGF (20 ng mL−1) also did not induce capillary formation by nonparenchymal cells without hepatoblasts (Fig. 8C,D). VEGF at higher concentrations in the culture medium stimulated the adhesion and growth of endothelial cells, depending on its concentration, but capillary formation was still deficient in these cultures (Suppporting Information Fig. S7A,C). When nonparenchymal cells were cultured at higher cell concentrations (three times higher than the ordinary concentration), capillary formation did not occur (Figs 9A,B and S7B, C). Endothelial cells became slender in some cases but did not develop into capillary networks as seen in remixed and crude cultures.

image

Figure 9.  Effects of cell concentration on capillary formation in cultures of E12.5 liver cells depleted of hepatoblasts. The culture medium contained VEGF (20 ng mL−1) and the conditioned medium. (A) PECAM-1-immunohistochemistry for nonparenchymal cell cultures at the ordinary cell concentration. (B) PECAM-1-immunohistochemistry for nonparenchymal cell cultures at a threefold higher cell concentration. PECAM-1-positive endothelial cells do not spread on the glass slides, and are round (A,B; arrows). Scale bars: 50 μm.

Download figure to PowerPoint

RT-PCR analysis showed that expression of Pecam1 and CD32b was very low in cultures of nonparenchymal cells without hepatoblasts (Fig. 10 and Supporting Information S8). The ratio of CD32b to Pecam1 expression, which indicates CD32b expression in cells expressing Pecam1 and the maturation level of sinusoidal endothelial cells, was remarkably low in culture of nonparenchymal cell fractions, compared with remixed and crude cultures. The ratio of CD32b to Actb expression was also low in cultures of nonparenchymal cells (data not shown). Desmin expression was also impaired in the absence of hepatoblasts (Fig. S8). These data indicated that capillary morphogenesis and maturation of sinusoidal endothelial cells took place as a result of intimate cellular interactions between hepatoblasts and endothelial cells.

image

Figure 10.  Impaired maturation of sinusoidal endothelial cells in cultures of E12.5 liver cells in the absence of hepatoblasts. The culture medium contained VEGF (20 ng mL−1) and the conditioned medium. Semiquantitative RT-PCR of CD32b mRNA expression on day 3. The numbers in parentheses denote those of samples examined. The expression of CD32b mRNA is lower in cultures of nonparenchymal cell fractions than in remixed and crude cultures.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Author contributions
  10. References
  11. Supporting Information

The present study demonstrated that characteristic hepatic venous channels, sinusoids, developed in the early stages of mouse liver development, consistent with previous papers showing that they start to develop with formation of the liver primordium (Matsumoto et al. 2001; Shiojiri & Sugiyama, 2004; Nonaka et al. 2007). Desmin-positive stellate cells and F4/80-positive macrophages form the sinusoids with endothelial cells throughout liver development, suggesting that these cell types may interact intimately with one another to form sinusoidal structures and establish their functions. Hepatoblasts/hepatocytes also resided close to the sinusoidal structures in fetal livers. During hepatic sinusoid development, Pecam1 and Lyve1 expression decreased in postnatal livers, whereas CD32b and Stabilin2 expression increased, which could represent valid molecular phenotypes allowing us to evaluate the maturation level of sinusoidal endothelial cells as reported in previous papers (Nonaka et al. 2007; Yoshida et al. 2007).

The present study demonstrated experimentally that the presence of hepatoblasts induced capillary formation and maturation of sinusoidal endothelial cells in primary cultures of fetal mouse liver cells, from which hepatoblasts were depleted in the experimental group by the immunomagnetic method. This result suggests that hepatoblasts may also induce morphogenesis and maturation of neighboring sinusoids during liver development.

Their action may involve humoral paracrine signals such as VEGF (March et al. 2009; Sugiyama et al. 2010). Hepatoblasts and hepatocytes have been shown immunohistochemically to be positive for VEGF, and may produce VEGF to stimulate sinusoidal development (Shimizu et al. 2001; Sugiyama et al. 2010). However, as shown in the present study, the addition of VEGF, even in high concentrations, could not replace the effect of hepatoblasts for capillary formation, although VEGF significantly increased the number of endothelial cells that survived on the glass slides. These results suggested that VEGF might stimulate the growth or survival of endothelial cells in our culture system but have a poor effect on capillary formation by itself. Other factors produced by hepatoblasts could be required for capillary formation in addition to VEGF. These factors may not be humoral because the conditioned medium, which may contain various factors produced by hepatoblasts/hepatocytes, had no effect on capillary formation, though it also stimulated the adhesion and spreading of endothelial cells on the glass slides. It is still possible that humoral factors, including Wnts, may act locally between hepatoblasts and endothelial cells (Klein et al. 2008; Ding et al. 2010). Furthermore, hepatoblasts can affect capillary formation by their juxtacrine mechanisms. The capillary formation always occurred on hepatocyte sheets in remixed and conventional cultures. It is also possible that extracellular matrices, which may be produced as a result of cellular interactions between hepatoblasts and nonparenchymal cells, are involved. March et al. (2009) have already shown that extracellular matrices are important for survival of adult sinusoidal endothelial cells with VEGF. Because the wells used for our cell cultures were coated with type I collagen, this extracellular component probably does not induce sinusoid morphogenesis in vivo.

The cultures of nonparenchymal cells at higher concentrations failed to significantly induce capillary formation by endothelial cells. They had a slender morphology in some cases but did not develop into capillary networks as seen in control cultures. Thus, the absence of capillary formation in cultures of nonparenchymal cell fractions may not be due to a shortage of the number of endothelial cells. This result is consistent with the data at higher VEGF concentrations, in which capillary formation was deficient, irrespective of the increase in the number of endothelial cells on the glass slides.

Desmin-positive stellate cells may be important in sinusoidal development because they are located close to sinusoidal structures and support them throughout liver development, as shown in previous papers and in the present study (Enzan et al. 1997; Shiojiri & Sugiyama, 2004; Lee et al. 2007). The importance of stellate cells in sinusoid structures and cellular interactions between sinusoidal endothelial cells and stellate cells has been indicated experimentally using the rodent partial hepatectomy model and coculture systems (Lee et al. 2007; DeLeve et al. 2008; Wirz et al. 2008). In our separation protocol with an immunomagnetic method, RT-PCR analysis of each cell type marker demonstrated that stellate cells were partially excluded from the cell suspensions of fetal liver cells with hepatoblasts. These cell populations are possibly decisive for capillary formation. However, increased nonparenchymal cell populations, including both stellate cells and endothelial cells, failed to form capillary structures. Thus, the presence of hepatoblasts may have a decisive role in capillary formation in our culture system. However, it is possible that stellate cells firmly adhering to hepatoblasts may work for capillary formation.

Our data also showed that CD32b expression was defective in cultures of nonparenchymal cell fractions in the absence of hepatoblasts. Hepatoblasts may have a positive effect on the maturation of sinusoidal endothelial cells both in vitro and in vivo. It is likely that capillary formation in vitro or sinusoid morphogenesis is coupled with endothelial maturation showing higher expression of CD32b.

It is noteworthy that endothelial cells of central veins transiently expressed LYVE-1 but lost that expression after birth. Sinusoidal endothelial cells expressed LYVE-1 throughout liver development, although the distribution of LYVE-1-positive sinusoidal endothelial cells was limited in postnatal livers. Endothelial cells of portal veins never expressed this molecule. Expression of connexin40 and connexin37 has been shown to be confined to endothelial cells of portal veins and hepatic arteries (Shiojiri et al. 2006). These differences in gene expression of the four vessels may be related to the cellular lineages during development.

The molecular mechanisms of sinusoid development, including elucidation of molecules that are produced by hepatoblasts/hepatocytes and induce capillary formation, remain to be resolved. Our culture system excluding one specific cell type in fetal liver cell cultures may be invaluable in the course of such studies to uncover these mechanisms.

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Author contributions
  10. References
  11. Supporting Information

During murine liver development, hepatoblasts may work as a positive stimulator of sinusoid morphogenesis and maturation, in which a signal other than VEGF may play a decisive role, together with VEGF.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Author contributions
  10. References
  11. Supporting Information

We thank Professor Emeritus Takeo Mizuno of the University of Tokyo and Prof. Nelson Fausto of the University of Washington for their interest in our study and encouragement, and Mr. Kim Barrymore for his help in preparing our manuscript. This work was supported in part by Grants-in-aid from the Ministry of Education, Culture, Sports, Science and Technology, the Japanese Government (#22570063).

Author contributions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Author contributions
  10. References
  11. Supporting Information

All authors (Y.T., S.Y., T.K. & N.S.) contributed to the design, acquisition of data, data analysis/interpretation, drafting of the manuscript, critical revision of the manuscript and approval of the article.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Author contributions
  10. References
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgements
  9. Author contributions
  10. References
  11. Supporting Information

Fig. S1. RT-PCR analyses of Pecam1, Lyve1, Stab2, CD32b and Actb mRNA expression in E12.5, E17.5, P0 and adult livers.

Fig. S2. Histogenesis of hepatic organoids in primary cultures of E12.5 liver cells.

Fig. S3. Differentiation of hepatocytes and macrophages in cultures of E12.5 liver cells on day 5.

Fig. S4. Capillary networks formed on a hepatocyte sheet in primary culture of E12.5 liver cells.

Fig. S5. RT-PCR analyses of Pecam1, Lyve1, Stab2, CD32b and Actb mRNA expression on days 1, 3 and 5.

Fig. S6. Exclusion of E-cadherin-positive hepatoblasts from E12.5 liver cell suspensions.

Fig. S7. Effects of VEGF and cell concentration on capillary formation in cultures of E12.5 liver cells depleted of hepatoblasts.

Fig. S8. RT-PCR analyses of Cadh1, Pecam1, CD32b, Des and Actb in cultures of nonparenchymal cell fractions of E12.5 liver cells on day 3.

Table S1 Primary antibodies used in immunohistochemistry.

Table S2 Primers used in RT-PCR analysis.

Table S3 Capillary formation and hepatoblast concentrations.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer-reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

FilenameFormatSizeDescription
joa1532_sm_FigS1.tif2092KSupporting info item
joa1532_sm_FigS2.tif3016KSupporting info item
joa1532_sm_FigS3.tif3448KSupporting info item
joa1532_sm_FigS4.tif2311KSupporting info item
joa1532_sm_FigS5.tif2409KSupporting info item
joa1532_sm_FigS6.tif6258KSupporting info item
joa1532_sm_FigS7.tif9606KSupporting info item
joa1532_sm_FigS8.tif2056KSupporting info item
joa1532_sm_TableS1.xlsx9KSupporting info item
joa1532_sm_TableS2.xlsx38KSupporting info item
joa1532_sm_TableS3.xlsx9KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.