1. Developmental Biology
  2. Stem Cells and Regenerative Medicine
Download icon

Differential expression of Lutheran/BCAM regulates biliary tissue remodeling in ductular reaction during liver regeneration

  1. Yasushi Miura
  2. Satoshi Matsui
  3. Naoko Miyata
  4. Kenichi Harada
  5. Yamato Kikkawa
  6. Masaki Ohmuraya
  7. Kimi Araki
  8. Shinya Tsurusaki
  9. Hitoshi Okochi
  10. Nobuhito Goda
  11. Atsushi Miyajima
  12. Minoru Tanaka  Is a corresponding author
  1. National Center for Global Health and Medicine, Japan
  2. Waseda University, Japan
  3. The University of Tokyo, Japan
  4. Kanazawa University Graduate School of Medicine, Japan
  5. Tokyo University of Pharmacy and Life Sciences, Japan
  6. Hyogo College of Medicine, Japan
  7. Kumamoto University, Japan
Research Article
  • Cited 0
  • Views 729
  • Annotations
Cite this article as: eLife 2018;7:e36572 doi: 10.7554/eLife.36572

Abstract

Under chronic or severe liver injury, liver progenitor cells (LPCs) of biliary origin are known to expand and contribute to the regeneration of hepatocytes and cholangiocytes. This regeneration process is called ductular reaction (DR), which is accompanied by dynamic remodeling of biliary tissue. Although the DR shows apparently distinct mode of biliary extension depending on the type of liver injury, the key regulatory mechanism remains poorly understood. Here, we show that Lutheran (Lu)/Basal cell adhesion molecule (BCAM) regulates the morphogenesis of DR depending on liver disease models. Lu+ and Lu- biliary cells isolated from injured liver exhibit opposite phenotypes in cell motility and duct formation capacities in vitro. By overexpression of Lu, Lu- biliary cells acquire the phenotype of Lu+ biliary cells. Lu-deficient mice showed severe defects in DR. Our findings reveal a critical role of Lu in the control of phenotypic heterogeneity of DR in distinct liver disease models.

https://doi.org/10.7554/eLife.36572.001

eLife digest

Bile is a green to yellow liquid that the body uses to break down and digest fatty molecules. The substance is produced by the liver, and then it is collected and transported to the small bowel by a series of tubes known as the bile duct.

When the liver is damaged, the ‘biliary’ cells that line the duct orchestrate the repair of the organ. In fact, the duct often reorganizes itself differently depending on the type of disease the liver is experiencing. For example, the biliary cells can form thin tube-like structures that deeply invade liver tissues, or they can grow into several robust pipes near the existing bile duct. However, it remains largely unknown which protein – or proteins – drive these different types of remodeling.

Miura et al. find that, in mice, the biliary cells which invade an injured liver have a large amount of a protein called Lutheran at their surface, but that the cells that form robust ducts do not. This protein helps a cell attach to its surroundings. In addition, the biliary cells can adopt different types of repairing behaviors depending on the amount of Lutheran in their environment.

Further experiments show that it is difficult for genetically modified mice without the protein to reshape their bile duct after liver injury. Finally, Miura et al. also detect Lutheran in the remodeling livers of patients with liver disease. Taken together, these results suggest that Lutheran plays an important role in tailoring the repairing roles of the biliary cells to a particular disease. The next step would be to clarify how different liver conditions coordinate the amount of Lutheran in biliary cells to create the right type of remodeling.

https://doi.org/10.7554/eLife.36572.002

Introduction

The liver is known to possess high capacity for regeneration upon injury. In acutely injured or surgically resected livers, regeneration is usually achieved by proliferation and hypertrophy of residual hepatocytes (Fausto and Campbell, 2003; Miyaoka et al., 2012). By contrast, under chronic or severe liver injury that impairs the proliferation of hepatocytes, liver progenitor cell (LPC) has been postulated to contribute to liver regeneration by differentiating into hepatocytes and biliary epithelial cells (BECs), also known as cholangiocytes (Thorgeirsson, 1996; Fausto, 2004; Miyajima et al., 2014). This response is known as ductular reaction (DR), in which LPC/biliary cell with BEC marker expression proliferates from the portal areas of injured livers, forming pseudo-ductular structures. DRs are frequently observed in human chronic liver diseases and rodent models including fatty liver disease and cholangiopathy (Shafritz and Dabeva, 2002; Roskams et al., 2003; Gouw et al., 2011; Wood et al., 2014). In zebrafish models, biliary cells have been reported to contribute to regenerating hepatocytes after substantial loss of hepatocytes (Choi et al., 2014). In mouse models, accumulating evidence by in vitro assay or transplantation experiments of biliary cells supports the presence of potential LPC with clonogenicity and bi-lineage differentiation capacity in the biliary compartment (Suzuki et al., 2008a; Okabe et al., 2009; Dorrell et al., 2011; Lu et al., 2015). In addition, a recent study using in vivo genetic lineage tracing experiment demonstrated that biliary cells can regenerate hepatocytes as facultative LPC under impaired hepatocyte regeneration in mice (Raven et al., 2017). Thus, DR is considered as a process of liver regeneration in chronically injured liver. In fact, genetically manipulated mice with defects in DR have been reported to show impaired recovery from chronic liver injury (Ishikawa et al., 2012; Takase et al., 2013; Shin et al., 2015).

To explore the nature of LPC in DR, several mouse injury models have been developed previously. In particular, two dietary models using 3,5-Diethoxycarbonyl-1,4-dihydrocollidine supplemented (DDC) diet and choline-deficient, ethionine-supplemented (CDE) diet have been extensively utilized to characterize LPC in mice for many years (Preisegger et al., 1999; Akhurst et al., 2001). Although both models induce massive DR, the pathological features resulting from these two methods are quite distinct; CDE-induced injury is thought to be a mouse model of non-alcoholic fatty liver disease with extensive hepatic damage (Knight et al., 2005; Aharoni-Simon et al., 2011), while DDC-induced injury is considered as a model of chronic cholangiopathy with portal biliary damage and severe cholestasis (Fickert et al., 2007). Considering the different pathological features of these two models, DR is assumed to be regulated depending on the severity and type of liver injury. In fact, it has been reported that morphological and functional heterogeneity of ductal cells is evident in DRs originating from many different pathological conditions in human patients and rodent models (Sell, 1998; Alvaro et al., 2007; Priester et al., 2010; Kaneko et al., 2015). Similar to such previous observations, CDE- and DDC-induced DR exhibited quite distinct morphology; the former extends outwards away from the portal vein, showing primitive ductules with spindle-like shape, while the latter remains around the portal vein, forming obvious bile duct-like structures. However, very little is known about the molecular mechanisms accounting for the phenotypic difference of DR among liver disease models.

In the present study, we identified Lutheran blood group glycoprotein (Lu) as a crucial molecule to control the morphological heterogeneity of DR. Lu, also known as Basal Cell Adhesion Molecule (BCAM) or CD239 is a member of the immunoglobulin superfamily, which is composed of five Ig-like domains on the extracellular site, a single transmembrane domain and a short C-terminal cytoplasmic tail (Parsons et al., 1995). Lu is known as a laminin receptor, which has been studied in context of sickle cell disease (Udani et al., 1998; El Nemer et al., 1999), and is capable of binding to laminin-511/521 via laminin alpha5 (Lama5) chain (Parsons et al., 2001). It has been reported that Lu/BCAM is widely expressed in cells and tissues including hematopoietic cells, placenta and kidney, and is developmentally regulated in human liver (Parsons et al., 1995). However, the expression profile of Lu in mouse LPCs remains to be investigated. Here, we show that Lu is a robust marker to discriminate between CDE-induced and DDC-induced DR; Lu is highly expressed in proliferating biliary cells in the CDE model, but downregulated in the DDC model. By using fluorescence activated cell sorting (FACS), we have isolated Lu-positive and Lu-negative biliary cells from injured liver. By comparison of both biliary cells in scratch assay and cyst formation assay, we revealed a role for Lu in regulating the morphological heterogeneity of DR. Further analysis using Lu-deficient mice demonstrated that the extension of biliary tree was significantly suppressed during DR in the CDE-induced liver injury. Collectively, our findings demonstrate that Lu functionally regulates the remodeling of biliary tissue during liver regeneration, and provide new insights into the heterogeneity of LPCs among liver disease models.

Results

Lutheran is differentially expressed in CDE- and DDC-induced DR

We have previously identified Epithelial cell adhesion molecule (EpCAM) as a marker for murine LPC/BEC (Okabe et al., 2009). To investigate the difference in DRs between the CDE and DDC models, immunohistochemical analysis of EpCAM was performed. Both protocols of feeding CDE and DDC diet for 3 weeks induced robust DR accompanied by biliary cell expansion. However, the appearance of propagating ducts exhibited strikingly distinct features between two models; CDE-induced biliary cells displayed primitive capillary-like morphology, invading into the parenchymal area extensively, while DDC-induced biliary cells exhibited remarkable duct-like structure, remaining around the periportal area (Figure 1A). Similarly, immunostaining of CK19, another LPC marker, showed a pattern similar to that of EpCAM (Figure 1—figure supplement 1). Because most known LPC markers are uniformly expressed in both types of DR, these molecules may not account for the heterogeneity of LPC. Therefore, we examined the expression profile of Lu in both injury models, because we have identified Lu as a marker for hepatoblasts, a fetal type LPC during liver development. Co-staining of liver sections using anti-EpCAM and anti-Lu antibodies revealed that Lu was detected in extending biliary cells of the CDE-fed liver, whereas we could not find such signals in biliary cells of DDC-fed liver except intense signal in EpCAM- ducts (Figure 1B). As reported by the previous paper that Lu is stained in hepatic arteries and portal vein of adult human liver (Parsons et al., 1995), co-staining of Lu and platelet endothelial cell adhesion molecule (PECAM), an endothelial marker, in the DDC-fed liver revealed that the EpCAM- duct with strong fluorescence is hepatic artery (Figure 1—figure supplement 2). Next, to compare the expression level of Lu in biliary cells, we performed flow cytometric (FCM) analysis using anti-EpCAM and anti-Lu antibodies. In untreated normal liver, FCM and immunohistochemical analyses demonstrated that most EpCAM+ biliary cells showed moderate expression of Lu (Figure 1C and Figure 1—figure supplement 3). By contrast, approximately 35% of EpCAM+ cells in CDE-fed liver showed high expression of Lu, whereas nearly half of EpCAM+ cells in DDC-fed liver exhibited low or negative expression of Lu (Figure 1C). These opposing expression profiles of Lu in the CDE and DDC models led us to hypothesize that Lu might regulate the morphological heterogeneity of expanding biliary cells between these distinct injury models.

Figure 1 with 4 supplements see all
Phenotypic difference of DR and biliary cells between the CDE model and DDC model.

(A) Immunohistochemical analysis of CDE-fed and DDC-fed mouse liver sections for EpCAM. (B) Co-localization of EpCAM and Lu in CDE-fed and DDC-fed mouse liver sections. (C) Comparison of Lu expression level in EpCAM+ cells among normal diet-fed, CDE-fed and DDC-fed mouse livers by flow cytometric analysis. PV: portal vein. Scale bar: 100 μm.

https://doi.org/10.7554/eLife.36572.003

Isolation of biliary cells with a distinct level of Lu from chronically injured liver

To uncover the role of Lu in regulating biliary cell morphology, we isolated Luhigh and Lu-/low biliary cell fractions from CDE-injured livers by FACS, and then cultured each population. Although Lu is highly expressed in endothelial cells, we confirmed that a contamination of such cells in EpCAM+ gating is highly unlikely as shown in Figure 2—figure supplement 1. We refer to these cultured biliary cells as Lu+ BC and Lu- BC in the following studies. Interestingly, similar to the morphological characteristics of DR in vivo, Lu+ BC displayed spindle-like shape while Lu- BC showed epithelial-like morphology even after several passages (Figure 2A). Fluorescent staining using Phalloidin demonstrated that the actin fibers and pseudopods are formed in the atypical cell body of Lu+ BC while F-actin accumulates in the periphery of round-shaped cell body of Lu- BC (Figure 2B). By contrast, when freshly isolated EpCAM+ cholangiocytes from untreated normal liver were cultured on the dish, the attached cells exhibited a mixture of round and indefinite morphology (Figure 2—figure supplement 2). Furthermore, FCM analysis of cultured Lu+ BC and Lu- BC revealed that the expression profile of Lu was maintained in each cell; Luhigh-derived biliary cells continued to express Lu at a high level, while Lu-/low-derived biliary cells hardly expressed Lu (Figure 2D). By contrast, several stem cell markers such as CD24 and CD44 were expressed similarly between Lu+ BC and Lu- BC, suggesting that these cells are closely related, but distinct homogeneous populations.

Figure 2 with 2 supplements see all
Culture of Lu+and Lu- BC isolated from injured liver.

(A) Representative images of Lu+ BC and Lu- BC by bright field microscopy. (B) Immunocytochemistry for F-actin in cultured Lu+ BC and Lu- BC. Arrowhead indicates pseudopod. (C) Flow cytometric analysis of Lu, Sca1, CD24, CD44, and CD71 expression levels in Lu+ BC and Lu- BC. The cultured Lu+ BC and Lu- BC were used for analysis after 6 passages. Scale bar: 100 μm.

https://doi.org/10.7554/eLife.36572.008

Comparison of cell motility and cyst forming capacity between Lu+ BC and Lu- BC

Considering that biliary cell with high expression of Lu in the CDE model exhibits an invasive phenotype in vivo, Lu+ BC is expected to have a higher capacity for cell motility than Lu- BC. To compare the capacity for cell motility between Lu+ BC and Lu- BC, we performed an in vitro scratch assay, by which the moving distance was evaluated after creating a scratch on the dish. As expected, Lu+ BC showed significantly higher motility capacity than Lu- BC (Figure 3A), suggesting that there is a causal link between Lu expression level and cell motility.

Evaluation of Lu- BC and Lu+ BC characteristics by scratch assay and cyst formation assay.

(A) Scratch assay using Lu- BC or Lu+ BC. Representative images of day 0, day 1, and day 2 after scratch are shown. Quantitative data of cell moving distance at day 1 are plotted in a graph with mean and standard deviation. n = 4 biological replicates. Dotted line indicates cell front of scratched gap. (B) Schematic diagram of three-dimensional culture and cyst generation. (C) Bright field microscopic image of three-dimensionally cultured Lu+ BC and Lu- BC at culture day 4 or day 6. Arrows and arrowheads point to cysts and cell aggregates, respectively. The details of formed cysts are shown below. The cell cluster devoid of luminal structure was regarded as ‘Cell aggregate’. Scale bar: 100 μm.

https://doi.org/10.7554/eLife.36572.011

On the other hand, biliary cells with low or no expression of Lu in the DDC model exhibited an obvious luminal structure around the portal vein in vivo, suggesting that Lu- BC has higher capacity for duct formation than Lu+ BC. As previously reported, LPCs are able to form cyst-like luminal structures with biliary epithelial polarity in a three-dimensional (3D) organoid culture (Tanimizu et al., 2007). We thereby compared the cyst-forming capacity between Lu- BC and Lu+ BC in the 3D culture system (Figure 3B). After 6 days of culture, Lu- BC formed a large number of cystic structures, while Lu+ BC formed only small cell aggregates, demonstrating that Lu- BC has higher duct-forming capacity than Lu+ BC (Figure 3C). Thus, Lu+ BC and Lu- BC showed distinct features of cell motility and cyst formation in vitro. These results suggest that Lu may play a crucial role in the cell kinetics of biliary cell.

Lu- BC acquires the characteristics of Lu+ BC by overexpression of Lu

To investigate whether Lu expression is responsible for the characteristics of biliary cells, cDNA encoding either mouse Lu (mLu) and green fluorescent protein (GFP) or only GFP as a control was transduced into Lu- BC by retroviral vector. The mLu-transduced Lu- BC (Lu- BC-mLu) showed morphological change into a spindle-like shape, while GFP-transduced control (Lu- BC-GFP) retained a round shape (Figure 4A). In the scratch assay, Lu- BC-mLu showed higher mobility than Lu- BC-GFP (Figure 4B). In contrast, the cyst formation assay demonstrated the reduced cyst-forming capacity of Lu- BC-mLu, resulting in the formation of numerous small aggregates (Figure 4C). These results strongly suggested that Lu- BC acquired the Lu+ BC-like phenotype after Lu expression, and that Lu might endow biliary cells with characteristics of spindle shape morphology and enhanced motility.

Lu regulates the motility and cyst formation capacities of biliary cells.

(A) Schematic description for generating Lu- BC-mLu or Lu- BC-GFP by using retrovirus vector. Flow cytometric analyses show expression level of Lu and GFP. Bright field microscopic images show morphological change in Lu- BC-mLu. (B) Scratch assay using Lu- BC-GFP or Lu- BC-mLu. Representative images of day 0, day 1, and day 2 after scratch are shown. Quantitative data of cell moving distance at day 1 are plotted in a graph with mean and standard deviation. n = 5 biological replicates. Dotted line indicates cell front of scratched gap. (C) Bright field microscopic image of three-dimensionally cultured Lu- BC-GFP or Lu- BC-mLu at culture day 6. Arrows and arrowheads point to cysts and cell aggregates, respectively. The details of formed cysts are shown below. The cell cluster devoid of luminal structure was regarded as ‘Cell aggregate’. Scale bar: 100 μm.

https://doi.org/10.7554/eLife.36572.013

Constitutive association of Lama5 with DR in chronically injured livers

Although the differential expression profile of Lu is likely to be relevant to the phenotypic heterogeneity of DR, the molecular mechanism by Lu remains unclear. Because Lu is known to be a receptor for Lama5 and bind to Laminin-511 and -521, we next investigated the expression of Lama5 in CDE- and DDC-injured livers. Intriguingly, double staining of EpCAM and Lama5 revealed that most expanding biliary cells are fully surrounded by Lama5 in both liver injury models (Figure 5A). Considering the accumulation of Lama5 in the vicinity of biliary cells, it is plausible that Lama5 may be secreted from biliary cell itself rather than the environmental niche cell. Indeed, the expression of Lama5 mRNA was verified in both EpCAM+ biliary cells isolated from CDE- and DDC-injured livers (Figure 5B), implying the involvement of Laminins in Lu-driven regulation. While Lu is capable of binding to Laminin-511/521 via Lama5, these laminins are also known as a ligand for Integrinα3β1/α6β1 (Kikkawa et al., 2007). It has been reported that Lu binds to Lama5 competitively with Integrinα3β1/α6β1 and promotes tumor cell migration by modulating Integrin-mediated cell attachment to Laminin-511 protein (Kikkawa et al., 2013). Taking these evidences into account, Lu may regulate the morphogenesis of DR via an Integrin-mediated manner. Given that Lu plays a role in the competitive inhibition against Laminin-511/521 and Integrinα3β1/6β1 axis in biliary cell as shown in Figure 5—figure supplement 1, high expression of Lu would be reproduced by inhibition of integrinβ1 (Itgb1) signaling. To address this possibility, we first investigated the expression of Integrinα3 (Itga3), Integrinα6 (Itga6) and Itgb1 in Lu- BC and Lu+ BC. As shown in Figure 5—figure supplement 2, all integrin components were expressed in Lu- BC and Lu+ BC, indicating that Lu- BC and Lu+ BC are potentially competent to cell signaling via Integrinα3β1/α6β1-Laminin-511/521 axis. We next examined the effect of neutralizing antibody against Itgb1 on the motility and duct formation capacity of Lu- BC in vitro. Although the inhibition of Itgb1 signaling did not affect the expression of Lu (Figure 5—figure supplement 3), it dramatically changed Lu- BC to Lu+ BC-like phenotype in both scratch assay and cyst formation assay (Figure 5C and D). Conversely, we investigated the effect of Itgb1 activation on Lu+ BC. Because TS2/16 antibody has been reported to activate Itgb1 signaling (Rozo et al., 2016), we added it to the 3D culture of Lu+ BC. As a result, Lu+ BC acquired cyst formation capacity by the activation of Itgb1 (Figure 5—figure supplement 4). These data strongly suggested that Lu regulates the characteristic of DR by modulating the Itgb1 signaling.

Figure 5 with 4 supplements see all
Itgb1 signaling is critical for regulating the phenotype of biliary cells.

(A) Expression analysis for Lama5 in injured liver. Co-staining of EpCAM and Lama5 was performed in liver sections of CDE-fed mouse and DDC-fed mouse. (B) Evaluation of Lama5 gene expression in EpCAM+ cells isolated from CDE-fed and DDC-fed mouse livers by quantitative RT-PCR. Data are plotted in a graph with mean and standard deviation. n = 4 biological replicates. n.s.: not significance. (C) Scratch assay using Lu- BC in the presence or absence of Itgb1 neutralizing antibody. Hamster IgM was used as control. Representative images of day 0, day 1, and day 2 after scratch are shown. Dotted line indicates cell front of scratched gap. Quantitative data of cell moving distance at day 1 are plotted in a graph with mean and standard deviation. n = 4 biological replicates. (D) Bright field microscopic image of three-dimensionally cultured Lu- BC-GFP at culture day 12 in the presence of Itgb1 neutralizing antibody or control IgM. Arrows and arrowheads point to cysts and cell aggregates, respectively. The details of formed cysts are shown below. The cell cluster devoid of luminal structure was regarded as ‘Cell aggregate’. Scale bar: 100 μm.

https://doi.org/10.7554/eLife.36572.015

Bcam KO mice show a key role for Lu in DR in CDE-induced liver injury

To verify the role of Lu/Bcam in DR in vivo, we generated Bcam knockout (KO) mice, in which the 11 bp deletion within a signal sequence-coding region of exon1 resulted in the frame shift of the Bcam gene (Figure 6A and Figure 6—figure supplement 1). The Bcam KO mice were healthy, showing no obvious developmental abnormality as reported previously (Rahuel et al., 2008). To compare the phenotype of DR between Wild-type (WT) and Bcam KO mice, mice were fed a CDE or DDC diet for 3 weeks and then their livers were analyzed by immunohistochemistry for EpCAM. When mice were fed the CDE diet, the proliferation of biliary cells occurred in both WT and Bcam KO mouse livers (Figure 6B). The loss of Lu protein in KO mouse was confirmed by immunostaining for Lu (Figure 6C). To our surprise, Lu-deficient biliary cells failed to spread outwards from the portal vein while WT biliary cells extended into the parenchymal area. The migrating distance of biliary cells from the portal vein was significantly shorter in Bcam KO mice than WT mice (Figure 6D).

Figure 6 with 1 supplement see all
Bcam KO mice show drastic phenotype change in DR.

(A) Schematic diagram of experiments using Bcam KO mouse. (B) Co-staining of EpCAM and Ki67 in WT and Bcam KO mouse liver sections in the CDE model. The ratio of Ki67+ cell per EpCAM+ cell is plotted in a graph with mean and standard deviation. n = 4 biological replicates. n.s.: not significance. (C) Co-staining of EpCAM and Lu in CDE-fed WT and Bcam KO mouse liver sections. (D) Quantitative analysis for the distance from portal vein to distal biliary cells in the CDE and DDC models. Data are plotted in a graph with mean and standard deviation. Statistical significance among groups is determined using one-way ANOVA. n = 5 biological replicates. (E) Immunostaining of EpCAM in WT and Bcam KO mouse liver sections in the DDC model. (F) Magnified immunohistochemical image of DR in CDE-fed and DDC-fed Bcam KO mouse liver. (G) Morphological image of EpCAM+ cells sorted from WT and Bcam KO mouse fed a CDE diet for 3 weeks. The cells were directly stained with Phalloidin to visualize F-actin 72 hr after plating on culture dish. Arrowhead points to pseudopod. Scale bar: 100 μm.

https://doi.org/10.7554/eLife.36572.024

By contrast, in the DDC model, the migrating distance of biliary cells from the portal vein showed no significant difference between WT and Bcam KO mice (Figure 6D and E). However, a slight hyperplasia of duct-like structures was observed in Bcam KO mice in both DDC and CDE models (Figure 6F). Moreover, freshly isolated EpCAM+ biliary cells from WT CDE-fed mouse livers showed spindle-like shape and pseudopods on the dish, while those from Bcam KO mice predominantly exhibited rounder morphology resembling Lu- BC (Figure 6G). These results suggested that Lu plays a critical role in the definition of morphological heterogeneity of DR in vivo.

CD239 is expressed in DR in human liver disease

To investigate the expression of Lu/BCAM (CD239) in human liver disease, we performed immunostaining of CD239 for a few resected samples obtained from the surgery to remove liver cancer. Intriguingly, CD239 was stained in DR in patients of chronic liver disease including non-alcoholic steatohepatitis, hepatitis B virus and hepatitis C virus (Figure 7), suggesting that the expression profile of Lu is conserved in humans.

Expression of CD239 in human liver disease.

Immunohistochemical images for Lu/BCAM (CD239) in human liver disease are shown. Cirrhotic liver sections obtained from the patients of non-alcoholic steatohepatitis (A), hepatitis B virus (B) and hepatitis C virus (C) were stained. DRs are denoted with arrows.

https://doi.org/10.7554/eLife.36572.027

Discussion

DR is often observed in various situations of chronic liver injury or submassive liver cell loss. A number of anatomical and histological analyses of human and rodent liver tissues have supported the notion that DR represents the expansion of LPC for supplying transit-amplifying cells to replenish the damaged hepatic cells. However, the origin and the role of LPC in liver regeneration is still under intensive debate. Recent lineage tracing experiments in mice have revealed that the hepatocytic differentiation of LPCs derived from the biliary compartment is negligible in DDC-induced liver injury (Malato et al., 2011; Español-Suñer et al., 2012; Tarlow et al., 2014; Rodrigo-Torres et al., 2014). This is reasonable because DDC-injury is a model of chronic cholangiopathy, which requires replenishment of cholangiocytes or bile ducts to be recovered. In fact, it has been reported that mice with impaired DR causes severe jaundice in the DDC model (Takase et al., 2013). Consistently, we observed many bile ducts with an obvious luminal structure in the DR of DDC-fed liver. Therefore, the downregulation of Lu in LPC may represent a process of cholangiocytic differentiation and reinforcement of duct formation for bile excretion in the DDC model.

In contrast to the DDC-model, it is still controversial whether LPC may contribute to hepatocyte regeneration in other chronic liver injury models. The contribution of LPC derived from biliary component to hepatocyte regeneration has been reported in the CDE model using two different lineage tracing approaches based on BEC marker genes, Osteopontin (Spp1) and Hnf1β (Español-Suñer et al., 2012; Rodrigo-Torres et al., 2014), although to a much lesser extent. The valid but low contribution of LPC of biliary origin to hepatocytic differentiation may be explained by robust proliferation of mature hepatocytes, because hepatocyte-mediated regeneration is not inhibited in the CDE model. This notion is strongly supported by a more recent report from Forbes’s group suggesting that a combination of CDE-injury and inhibition of hepatocyte proliferation causes physiologically significant and higher contribution of biliary cells to hepatic regeneration (Raven et al., 2017). Therefore, Lu-mediated high motility of LPC in the CDE model may contribute to the rapid delivery of hepatic progenitor cells to the damaged parenchymal area far from the portal vein.

In the DR, environmental factors play crucial roles in the regulation of LPC proliferation and differentiation. Several immune cell-derived cytokines such as TNF-related WEAK inducer of apoptosis (TWEAK), interleukin-6 and interleukin-22 have been shown to be pro-mitotic for LPCs (Jakubowski et al., 2005; Yeoh et al., 2007; Feng et al., 2012). Cell signaling pathways including Wnt, Notch, HGF and EGF are reportedly responsible for fate decisions of LPCs (Boulter et al., 2012; Fiorotto et al., 2013; Kitade et al., 2013). Although the regulatory mechanism of Lu expression in LPCs is unclear, these signaling pathways may be worthy of investigating.

Extracellular matrix (ECM) is also known to serve a niche for LPC regulation in chronic liver injury as well as liver development. Of note, laminin is an important component of LPC niche affecting cell fate decision. It has been reported that the escape of LPCs from the laminin basement favors their hepatocytic differentiation (Español-Suñer et al., 2012; Paku et al., 2004), while laminin aids maintenance of LPC and biliary cell phenotype (Lorenzini et al., 2010; Boulter et al., 2013). Consistently, it has been reported that Lama5 KO mice show defects in bile duct formation during liver development and that Itgb1 signaling is required for cholangiocytic differentiation from hepatoblast, a fetal-type LPC (Tanimizu et al., 2012). It is therefore highly probable that Itgb1 signaling in the context of laminin/integrin axis is crucial for biliary regeneration and duct formation from LPC. Our in vitro data also supported the idea that Itgb1 signaling modulated by Lu expression may govern DR, depending on liver injury type.

It remains undetermined whether high expression of Lu in biliary cells is an essential step for hepatocytic differentiation. Alternatively, the up-regulation of Lu may provide LPC with a cue to escape from ECM-rich periportal area, as evidenced by Bcam KO phenotype in the CDE model. We showed that the downregulation of Lu in ductular reactive cells facilitated the cystogenic phenotype. However, the loss of Lu would be also expected for the hepatocytic differentiation from LPC because mature hepatocytes do not express Lu. In line with the idea, intriguingly, the expression of Lu seemed to be downregulated in EpCAM+ cells at the tip of DR in the CDE model (Figure 1B). The discrepancy may be explained by the microenvironment surrounding LPCs; the periportal region is rich in ECM including laminin whereas the parenchymal region is poor in it. Taking such microenvironment into consideration, the loss of Lu in parenchymal area may be a sign of hepatocytic differentiation by the escape of LPC from laminin deposition. Further expression analysis of Lu and Laminin with a combination of lineage tracing experiment will uncover the mechanism underlying LPC-mediated liver regeneration depending on the microenvironment.

In addition to the signaling molecules and ECM, several non-parenchymal cells have been shown to be involved in the microenvironment for LPC. Intriguingly, Hul et al. reported that prolonged Kupffer cell (KC) depletion did not influence the proliferation of LPCs but reduced their invasive behavior in the CDE model (Van Hul et al., 2011). The LPCs of KC-depleted mice exhibit phenotype resembling cells of biliary lineage with rounder morphology. More interestingly, the LPCs remain closer to the portal area, in places delineating a pseudo-lumen. This phenotypic change of LPCs by KC depletion closely resembles that of Bcam KO mice in the CDE diet. On the other hand, Boulter et al. reported that hepatic macrophage played a role in promoting LPC specification to hepatocytes by expressing Wnt3a in the CDE model (Boulter et al., 2012). These evidences suggest that hepatic macrophages may be a key regulator of Lu expression in a process of LPC specification. Further studies using a lineage tracing experiment in Bcam KO mice under various types of chronic liver injuries will provide a clue to better understanding the molecular mechanisms underlying the phenotypic heterogeneity, as well as fate determination of LPCs during liver regeneration.

In conclusion, the present study demonstrates that the expression profile of Lu in biliary cells dramatically changes during DR depending on the type of liver injury, which in turn dictates the morphological characteristics of biliary cells such as cell motility and duct formation (Figure 8). This molecular mechanism would be expected to be conserved in human liver diseases. Thus, Lu is a novel marker for classification of DR and an interesting functional molecule for investigating the nature of LPCs. Our findings will provide new insights into the significance of biliary cell heterogeneity in liver regeneration.

Graphical abstract of this study.
https://doi.org/10.7554/eLife.36572.028

Materials and methods

Key resources table
Reagent typeDesignationSource or referenceIdentifiersAdditional information
Antibodyanti-EpCAM (rat monoclonal)PMID: 19429791(1:100–500)
Antibodyanti-EpCAM (rabbit polyclonal)AbcamAbcam:ab71916; RRID:AB_1603782(1:400)
Antibodyanti-Lutheran (rat monoclonal)this paper(1:500)
Antibodyanti-Sca1 (rat monoclonal)BioLegendBioLegend:108107; RRID:AB_313344(1:400)
Antibodyanti-CD24 (rat monoclonal)Miltenyi BiotecMB:130-102-731; RRID:AB_2656573(1:400)
Antibodyanti-CD71 (rat monoclonal)Miltenyi BiotecMB:130-109-632; RRID:AB_2659126(1:400)
Antibodyanti-CD44 (rat monoclonal)Miltenyi BiotecMB:130-110-117; RRID:AB_2658152(1:400)
Antibodyanti-FcR (rat monoclonal)BioLegendBioLegend:101320; RRID:AB_1574975
and BioLegend:101302; RRID:AB_312801
(1:100)
Antibodyanti-CK19 (rabbit polyclonal)PMID:12665558(1:1000)
Antibodyanti-Lama5 (rabbit polyclonal)PMID:9151674(1:200)
Antibodyanti-Ki67 (rat monoclonal)ThermoFisherThermoFisher:14-5698-80;
RRID:AB_10853185
(1:200)
Antibodyanti-PECAM (rat monoclonal)BD BiosciencesBD:553373; RRID:AB_394819(1:100)
Antibodyanti-CD31 (rabbit polyclonal)NovusNB:NB100-2284; RRID:AB_10002513(1:100)
Antibodyanti-CD239 (rabbit monoclonal)AbcamAbcam:2994–1; RRID:AB_2065309(1:100)
Antibodyanti-CD29 (hamster monoclonal)BD BiosciencesBD:555002; RRID:AB_3956365 μg/mL for 3D culture
Antibodyanti-CD29 (mouse monoclonal)BioLegendBiolegend:303010; RRID:AB_31432650 μg/mL for 3D culture
AntibodyHamster IgM, λ1 isotype
(hamster monoclonal)
BD BiosciencesBD:553957; RRID:AB_4796395 μg/mL for 3D culture
AntibodyMouse IgG1, κsotype
(mouse monoclonal)
BioLegendBioLegend:40140450 μg/mL for 3D culture
Cell line
(Homo sapiens)
Plat-E (Platinum-E)PMID: 10871756RRID:CVCL_B488Cell line established in
T. Kitamura lab

Animal models

C57BL/6J mice were purchased from Clea-Japan, Inc. (Tokyo, Japan). All animals were maintained in a standard Specific-Pathogen-Free (SPF) room at the institutional animal facility. All animal experiments were performed according to institutional guidelines and approved by the Animal Care and Use committee of the Institute of Molecular and Cellular Biosciences, The University of Tokyo (approval numbers 2501, 2501–1, 2609, 2706, and 3004), Kumatomo University (approval number A27-092), Hyogo College of Medicine (approval number 16–043, 16–046), and National Center for Global Health and Medicine Research Institute (approval numbers 15080, 16023, 17086 and 18069). To induce liver injury, a diet containing 0.1% DDC (Clea-Japan Inc. Tokyo, Japan) or the choline-deficient, ethionine-supplemented diet (MP Biomedicals, CA, USA) was fed to 6-week-old mice for 3 weeks.

Study approval for human samples

The study using human samples was approved by the Kanazawa University Ethics Committee (approval number 305–4), and all of the analyzed samples are derived from patients who provided informed written consent for the use of their tissue samples in research.

Antibodies

The information about antibodies used for FACS and immunohistochemistry is described in Key resources table. The rat anti-EpCAM monoclonal antibody was generated as described previously (Okabe et al., 2009). The rabbit anti-CK19 polyclonal antibody was generated as described previously (Tanimizu et al., 2003). The rat anti-Lutheran monoclonal antibody used in this study was generated by immunization of a rat with mouse fetal hepatic cells as described previously (Suzuki et al., 2008b), and biotinylated for FACS using ECL Protein Biotinylation Module (GE Healthcare UK Ltd, UK). The specific reactivity against mouse Lu was validated by flow cytometric (FCM) analysis of Ba/F3 cells transfected with Bcam cDNA by a retroviral vector, pMxs/IRES-GFP (Kitamura et al., 2003) (Figure 1—figure supplement 4). The anti-Lutheran monoclonal antibody (D295-3) and anti-EpCAM monoclonal antibody (D269-3) are commercially available from MBL International Corporation, MA, USA.

Liver perfusion and cell sorting

Cells were isolated from murine livers as described previously (Okabe et al., 2009). Briefly, liver cells were dissociated by perfusion of collagenase solution. Non-parenchymal cells (NPCs) were prepared by removal of hepatocytes with repeated centrifugation at 100 g for 2 min. Then, NPCs were incubated with anti-FcR antibody for blocking non-specific binding, followed by with fluorescein isothiocyanate (FITC)-conjugated anti-EpCAM monoclonal antibody for 30 min on ice. After incubation with anti-FITC microbeads (1:10–100 dilution, Miltenyi Biotec, Bergisch Gladbach, Germany), EpCAM+ cells were enriched by autoMACS pro (Miltenyi Biotec). After MACS, cells were incubated with biotin-conjugated anti-Lutheran monoclonal antibody for 30 min on ice. After wash, cells were incubated with allophycocyanin (APC)-conjugated streptavidin (1:100–500 dilution, BD bioscience, NJ, USA) for 20 min on ice and analyzed or purified by fluorescence-activated cell sorting (FACS) using Moflo XDP (Beckman-Coulter, CA, USA) and BD FACSCanto II (BD bioscience). Dead cells were excluded by propidium iodide (Sigma-Aldrich, MO, USA) staining.

Cell culture

Sorted EpCAM+ cells from CDE-fed mouse liver were cultured with modified William’s-E medium as previously reported (Okabe et al., 2009). For in vitro assay, the cells expanded at 3 to 6 passages were used. For cytoskeleton staining, Alexa Fluor 488 Phalloidin (1:500 dilution, Thermo Fisher Scientific, MA, USA) was used.

Scratch assay

The cells were seeded in a 24-well plate at a confluency of 90–95% the day before scratch assay. After 12 to 24 hr of culture, the medium was removed and fresh medium with 10 μg/mL Mitomycin C (Wako Pure Chemical Industries) was added to fully confluent cells to inhibit further proliferation. The treated cells were incubated continuously for 150 min at 37°C. After incubation, the cell layer was scratched crosswise with a micropipette tip. Each well was washed twice with PBS to prevent detached cells from re-adhering. After creating the scratch, the cells were continuously cultured without cytokines. The moving distance was calculated by subtracting the half of gap length at Day 1 from that at Day 0. For experiments of antibody administration, Hamster anti-rat CD29 (555002, BD Pharmingen) or Hamster IgM (553957, BD Pharmingen) was added in the culture at a final concentration of 1 μg/mL.

Immunohistochemistry

The resected left lobe of liver was embedded into OCT compound (Sakura Finetek Japan), and frozen by liquid nitrogen. The frozen block was cut into 8 μm slices by Microtome Cryostat HM 525 (Thermo Fisher Scientific). Fixation was performed by using 4% paraformaldehyde (Wako Pure Chemical Industries, Osaka, Japan) or cold acetone (Wako Pure Chemical Industries). For blocking buffer, 5–10% skim milk (BD bioscience) or 3% FBS (Thermo Fisher Scientific) was used. Primary antibodies used for immunohistochemistry were rabbit anti-CK19 polyclonal antibody, rat anti-EpCAM monoclonal antibody, rabbit anti-EpCAM polyclonal antibody, rabbit polyclonal anti-Laminin α5 antibody (a kind gift from Dr. Jeffrey H. Miner), rabbit anti-PECAM polyclonal antibody, rat anti-Ki67 monoclonal antibody and rat anti-Lutheran monoclonal antibody. The information about antibodies is described in Key resources table. All images were captured using KEYENCE BZ-X710:BZ-X Viewer, Zeiss Axio observer z1: AxioCamHR3 or Olympus FV3000. The ratio of Ki67+ cell per EpCAM+ cell was calculated using Hybrid Cell Count function in the Dual Signal Extraction mode of BZ-X Analyzer. An average value of three random images per mouse was treated as a representative value for the mouse. Quantification of the distance of biliary cell cluster/cell from the center of the portal vein was performed using a previously reported method (Best et al., 2016). Briefly, the distance from the center of the portal vein to the most distal EpCAM-stained cell was measured, and then the mean diameter of the portal vein was subtracted from this value to eliminate the influence of the size of the portal vein. An average value of six to fourteen random images of portal region per mouse was treated as a representative value for the mouse.

Three-dimensional culture

For three-dimensional culture, Cellmatrix Type I-A (Nitta Gelatin, Osaka, Japan) and Matrigel with Growth Factor Reduced (Corning, MA, USA) were used for gel components. Chilled cellmatrix and Matrigel were mixed at 1:9 ratio and used to coat the surface of the culture dish. After solidification of the coating layer by incubating at 37°C, the mixture of cell suspension and gel at 1:1 ratio was added. After 30 min of incubation at 37°C in 5% CO2 chamber, culture medium was loaded on top of the double-layered gel. The top layer of medium was changed twice a week. All images were captured using DS-Fi2-L3 (Nikon Corporation, Tokyo, Japan) under a phase-contrast microscope (Nikon ECLIPSE TS100, Nikon Corporation) after 6–12 days of culture. For the experiments using neutralizing and activating antibodies against Itgb1, Hamster anti-rat CD29 (555002, BD Biosciences) and TS2/16 (303010, Biolegend) were added in the culture at a final concentration of 5 μg/mL and 50 μg/mL, respectively. For each control, Hamster IgM, λ1 isotype control (553957, BD Pharmingen) or Mouse IgG1, κ isotype control (401404, Biolegend) were used. For quantification of the size and formation efficiency of cyst, 50 cells were cultured in individual wells of a 96-well plate. After 6 days of culture, the image was captured by a phase-contrast microscope. All visible cell clusters were counted according to the diameter of lumen. The cell cluster devoid of luminal structure was counted as ‘Cell aggregate’.

Establishment of Lu-expressing Lu- BC by retroviral vector

For overexpression of mouse Lu (mLu) in Lu- BC, Bcam cDNA was amplified with two primers 3’- CTCGAGTCACTGCCGCCACTGCAG −5’ and 3’- GTCGACTTACATTCCCTGGAGGAAG −5’ by RT-PCR and inserted into the EcoRI and XhoI restriction enzyme sites of pMxs-IG plasmid vector (kindly provided by Dr. Kitamura) was used. For the production of retrovirus, pMxs-mLu-IG or pMxs-IG was transfected into the packaging cell line Platinum-E (Morita et al., 2000) by using lipofectamine 2000 (Invitrogen). The culture supernatant was centrifuged at 6000 g at 4°C overnight to recover virus particles. The precipitated virus particles were dissolved in culture medium and used to infect Lu- BC. After 16–24 hr, the culture media was replaced with fresh media and the culture was continued overnight. The cells expressing both mLu and GFP or only GFP were sorted by FACS and named as Lu- BC-mLu or Lu- BC-GFP, respectively.

RNA extraction and reverse transcription

RNA extraction was performed using ISOSPIN Cell and Tissue RNA (NIPPON GENE, Toyama, Japan) according to the manufacturer’s instruction. For tissue homogenization, FastPrep-24 (MP Biomedicals) was used. Reverse transcription from RNA to cDNA was performed by PrimeScript RT Master Mix (Takara-bio, Shiga, Japan).

Quantitative RT-PCR

Quantitative RT-PCR was performed using LightCycler480 (Roche, Basel, Switzerland) with the Universal Probe Library system. The sequence of used primers is (5’ to 3’) EpCAM-Forward: AGAATACTGTCATTTGCTCCAAACT, EpCAM-Reverse: GTTCTGGATCGCCCCTTC, Lama5-Forward: GGCCTGGAGTACAATGAGGT, Lama5-Reverse: CACATAGGCCACATGGAACA, ITGB1-Forward: TCAACATGGAGAACAAGACCA, ITGB1-Reverse: CCAACCACAGCTCAATCTCA, ITGA3-Forward: TCAACATGGAGAACAAGACCA, ITGA3-Reverse: CCAACCACAGCTCAATCTCA, ITGA6-Forward: GCGGCTACTTTCACTAAGGACT, and ITGA6-Reverse: TTCTTTTGTTCTACACGGACGA.

Generation of Bcam/Lu Deficient Mice

 pT7-sgRNA and pT7-hCas9 plasmid were kindly provided from Dr. Ikawa (Osaka University, Japan) (Mashiko et al., 2013). After digestion with EcoRI, Bcam mRNA synthesis was performed using an in vitro RNA transcription kit (mMESSAGE mMACHINE T7 Ultra Kit, Thermo Fisher Scientific), according to the manufacturer’s instructions. A pair of oligos targeting Bcam gene was annealed and inserted into the BbsI site of the pT7-sgRNA vector. The sequences of the oligos were as follows: Bcam/Lu (5’- AAC CCC CTG ACG CCC GCG CA −3’), which is located at exon 1 of Bcam/Lu gene. After digestion with XbaI, gRNAs were synthesized using the MEGAshortscript Kit (Thermo Fisher Scientific). The precipitated RNA was dissolved in Opti-MEM I (Thermo Fisher Scientific) at 0.4 μg/μL. C57BL/6N female mice (Clea-Japan Inc.) were used in this study. IVF was performed according to the Center for Animal Resources and Development’s (at Kumamoto University, Japan) protocol (http://card.medic.kumamoto-u.ac.jp/card/english/sigen/manual/onlinemanual.html). Electroporated embryos were cultured in KSOM medium, and transferred the next day to the oviducts of pseudo-pregnant females on the day of vaginal plug detection. Genome Editor electroporator and LF501PT1-10 platinum plate electrode (BEX Co.Ltd., Tokyo, Japan) were used for electroporation. 50 embryos prepared were subjected to electroporation. The collected embryos cultured in KSOM medium were placed in the electrode gap filled with 5 μl of Opti-MEM I containing sgRNA and hCas9 mRNA. The electroporation conditions were 25V, five times. The eggs were then cultured in KSOM medium at 37°C and 5% CO2 in an incubator until the two-cell stage.

Human tissue and immunohistochemistry

Three individual surgical specimens of cirrhotic liver were obtained from the patients with hepatocellular carcinoma. The deparaffinized and rehydrated sections were microwaved in EDTA buffer (pH 9.0) for 20 min in a microwave oven. Following endogenous peroxidase blocking, these sections were incubated at 4˚C overnight with rabbit anti-CD239 monoclonal antibody against human Lutheran/BCAM (1:100 dilution, Epitomics, CA, USA) and then at RT for 1 hr with goat anti-rabbit immunoglobulins conjugated to peroxidase labeled-dextran polymer (K4003, Envision, Dako, Tokyo, Japan). After benzidine reaction, sections were lightly counterstained with hematoxylin.

Statistical analysis

Statistical analyses and the determination of p value were performed using GraphPad Prism software. Statistical significance between two groups was evaluated using Mann-Whitney U test and considered for p<0.05. For comparison of four groups, one-way analysis of variance (ANOVA) was applied, and once F-test was significant, multiple comparisons between each group were conducted by Tukey’s multiple comparisons. Statistical significance was set at two-tailed p values < 0.05. Values derived from at least four biological replicates were plotted in a graph with mean and standard deviation. The exact number of biological samples was described in each figure legend and source data. There was no exclusion of outliers in all experiments. Group allocation was performed without any bias. A statistical method of sample size calculation was not used during study design.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
    Atypical ductular proliferation and its inhibition by transforming growth factor beta1 in the 3,5-diethoxycarbonyl-1,4-dihydrocollidine mouse model for chronic alcoholic liver disease
    1. KH Preisegger
    2. VM Factor
    3. A Fuchsbichler
    4. C Stumptner
    5. H Denk
    6. SS Thorgeirsson
    (1999)
    Laboratory Investigation; a Journal of Technical Methods and Pathology 79:103–109.
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57

Decision letter

  1. Marianne E Bronner
    Senior Editor; California Institute of Technology, United States
  2. Frédéric Lemaigre
    Reviewing Editor; Université catholique de Louvain, Belgium

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "Differential expression of Lutheran regulates biliary tissue remodeling in ductular reaction during liver regeneration" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Marianne Bronner as the Senior Editor. The following individual involved in review of your submission has agreed to reveal his identity: Holger Willenbring (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

The manuscript by Miura et al. investigates the ductular reaction, an extension of the biliary system found in most types of chronic liver injury. It has previously been shown that the morphology of ductular reaction depends on the type of liver injury, for example, preferential injury to cholangiocytes or hepatocytes. However, the mechanism underlying the different modes and morphologies of ductular reactions is not known. The authors identify Lutheran (Lu), a laminin receptor, as a regulator of ductular morphology, using mouse models in which ductular reactions are induced by feeding a diet containing CDE or DDC, which injures the hepatocytes and cholangiocytes, respectively. They show that Lu expression is high in ductular reactive cells following CDE diet and low following DDC diet, and that Lu+ cells are more motile but less efficient in duct formation than Lu- biliary cells, using in vitro assays and new Lu knockout mice. The authors provide evidence that Lu may function by competing with integrins for binding to laminin alpha5.

Overall, the data are well presented, the figures are of very good quality and the findings are of importance, not only for the liver field. However, a number of deficiencies need to be addressed to improve the manuscript.

Essential revisions:

1) The expression of Lu needs to be better characterized.

- Lu is stated to be expressed in hepatic artery in Figure 1B but a co-staining needs to be provided to support this statement. Also, the authors should show the expression of Lu in control untreated liver by means of a EpCAM/Lu costaining.

- In the CDE model, some of the cells of the ductular reaction are Lu+ (high), others not. What is the topography of the Lu+ cells ? Lu+ cells indeed appear to be closer to the portal area and Lu- deeper in the parenchyma. Quantification may confirm this.

- A large population of ductular reactive cells have normal Lu expression in CDE-treated mice, and half of the ductular reactive cells in DDC-treated animals do not show detectable Lu as compared to controls. Can the authors speculate on the reason for heterogeneous expression of Lu in the ductular reactive cells in CDE- and DDC-treated mice?

2) The authors propose that Lu and Integrin alpha3 beta1/alpha6 beta1 compete for binding to laminin alpha5 (Lama5), because inhibition of Integrin beta1 in cultured Lu- biliary cells mimics the effects of Lu overexpression. The concept of competition between Integrin and Lu should be strengthened and additional controls are required:

- Can the authors overexpress the alpha3 (or alpha6) and beta1 integrin subunits in cultured Lu+ cells and verify if the cells are converted to a cystogenic phenotype?

- Is the expression of Integrin alpha3 beta1/alpha6 beta1 similar in vitro in Lu

and Lu+ biliary cells?

- The authors should provide information on the expression of Lu in Lu- cells incubated with Integrin beta1 neutralising antibodies.

3) The expression of Lama5 raises several questions.

- Lama5 is detected in vivo around ductular reactive cells in both CDE (Lu high) and DDC (Lu low) models, yet Figure 5B suggests that Lu- biliary cells express more Lama5 than Lu+ biliary cells. Is the difference shown in Figure 5B statistically significant?

- Macrophages play a major role in laminin deposition during liver disease, including in DDC-treated livers. The authors should compare the mRNA expression of Lama5 also in this cell type.

- In the CDE model, laminin appears to predominantly co-localise with ductular reactive cells into the parenchyma, while the Lu (high) population is restricted to a more proximal periportal zone (see comment #1). As per the authors' theory, when the cells lose Lu, laminin binding to integrin should be facilitated. Hence the cells should re-adopt a tubular, pseudo-ductular conformation, which is not the case. What is the authors' explanation?

- Stuart Forbes' group demonstrated that anchorage of liver progenitor cells to a laminin sheet maintains the biliary phenotype of the cells while loss of laminin favors hepatocyte differentiation. Also, in vivo, laminin degradation favors the differentiation of liver progenitor cells in the CDE model (PMID:22922013). Can the authors explain how they reconcile their data with the literature?

4) Lu is proposed to control cell motility and cyst formation. The experiments in Figures 3 to 5 require additional controls:

- In the DDC model, ductular reactions expands as bile duct like shape in vivo. Do the Lu- biliary cells in vitro have the same shape of a control cholangiocytes? In other words, control cholangiocytes should be added as control in Figure 2.

- An additional biliary marker like EpCAM in Figure 4A would confirm that the observed changes are due to Lu expression and not to contamination with other cell types. Were these cells freshly isolated or passaged?

- Concluding that Lu+ cell have higher motility than Lu-/low cells requires excluding higher proliferation in Figure 3A.

- Cyst size and cyst formation efficiency should be quantified in Figures3C, 4C and 5D.

5) A number of statements need to be clarified and additional information on the methods should be provided:

- In the subsection “Isolation of biliary cells with a distinct level of Lu from chronically injured liver”, the authors state that Lu (high)-derived biliary cells continued to express Lu at high levels. Information on the number of passages in Figure 2C should be given.

- The statement "proliferating biliary cell with high expression of Lu in the CDE model" requires information about proliferation markers in Lu+ and Lu-/low cells.

- Since Lu is also expressed in arterial branches, the authors have to describe the FACS gating strategy used to sort the biliary cells and to exclude contamination by arterial endothelial cells. Similarly, it is shown in Figure 6E that hepatocytes in the CDE model express EpCAM. How are those hepatocytes excluded from sorting ? Are those hepatocytes also Lu (high)?

6) Lutheran knockout livers treated with CDE do only show very low numbers of EpCAM+/HNF4+ cells as compared to wild-type CDE-treated livers. The authors conclude from this experiment that high expression of Lutheran "might be a process of hepatocytic differentiation of liver progenitor cells". Such conclusion is not sufficiently supported by the data, since co-expression of EpCAM and HNF4 does not provide solid evidence for hepatocyte differentiation of progenitors to hepatocytes. The reviewers suggest to remove Figure 6E which is quite preliminary.

7) It is not clear what the authors refer to by "slight dilation of duct-like structures was observed in Lu KO mice (Figure 6D and F)". Better images are required to support this statement.

8) It would considerably expand the impact of the paper if some information is provided about the expression of Lu in ductular reaction in human liver disease.

https://doi.org/10.7554/eLife.36572.032

Author response

Essential revisions:

1) The expression of Lu needs to be better characterized.

- Lu is stated to be expressed in hepatic artery in Figure 1B but a co-staining needs to be provided to support this statement. Also, the authors should show the expression of Lu in control untreated liver by means of a EpCAM/Lu costaining.

Because Lu has been reported to be stained in hepatic arteries intensely and in portal vein of adult human liver (Parsons et al., 1995), we supposed that the intense signal of Lu which is EpCAM-negative corresponds to mouse hepatic arteries. To support the idea, we have performed a co-staining of Lu and PECAM (CD31), an endothelial cell marker. As expected, Lu was stained in PECAM+ endothelial cells including hepatic artery and portal vein. Notably, Lu was detected in hepatic artery most intensely, indicating that the Lu+EpCAM- vessel around the portal vein is hepatic artery. We have added the immunohistochemical data in Figure 1—figure supplement 2. In addition, we performed a EpCAM/Lu costaining in control untreated liver and added the data in Figure 1—figure supplement 3 according to the reviewer’s comment. Basically, the aim of Figure 1B was to compare the expression level of Lu in ductular reaction between CDE and DDC models. Therefore, the photographic sensitivity was adjusted to the most intense fluorescence of hepatic artery without saturation, and these images were acquired at the same time of exposure on the same day. In those immunohistochemical settings, not only low/negative but also middle level of Lu in EpCAM+ cell shown by Figure 1C was not visible in the DDC-liver, whereas the high level of Lu in ductular reaction was detected in the CDE-liver. Thus, the undetected signal of Lu in this experimental setting does not necessarily mean no expression of Lu. Consistently, as shown in Figure1—figure supplement 3, a weak signal of Lu was detected in EpCAM+ bile duct by longer exposure while the Lu signal was saturated in hepatic artery or intensely detected in portal vein.

- In the CDE model, some of the cells of the ductular reaction are Lu+ (high), others not. What is the topography of the Lu+ cells ? Lu+ cells indeed appear to be closer to the portal area and Lu- deeper in the parenchyma. Quantification may confirm this.

As mentioned by the reviewer, several EpCAM+ cells apart from the portal vein seem to be lacking in Lu signal. If it was interpreted favorably, this image may reflect the hepatocytic differentiation process of EpCAM+ cells at the tip of ductules. It is consistent with the lack of Lu expression in mature hepatocyte. Although the idea is very attractive, the quantification of EpCAM+Lu- cells is difficult for some reasons mentioned below. 1) The invisible signal of Lu does not necessarily mean no Lu expression as mentioned above. Because the signal of Lu is faint even in the bile duct originally expressing Lu (Figure1—figure supplement 3), it is difficult to set the threshold of Lu-negative cell by immunohistochemical analysis. 2) EpCAM+Lu- cells seem to appear as a part of ductular reaction, i.e. the tip of extending ductules. Because the tip in ductular reaction is diverse and not always apart from the portal vein in the injured liver, the quantification based on the distance from the center of portal vein may bring a misleading result. From these reasons, we think that such quantification base on immunohistochemistry should be carefully treated in terms of topography. However, the reviewer’s comment is implicative, and the expression of Lu seems to be “downregulated” apart from the portal vein. Therefore, we have added such a description about the possible topography of the Lu+ cells without quantification in the Discussion section (fifth paragraph). We greatly appreciate the reviewer’s suggestions.

- A large population of ductular reactive cells have normal Lu expression in CDE-treated mice, and half of the ductular reactive cells in DDC-treated animals do not show detectable Lu as compared to controls. Can the authors speculate on the reason for heterogeneous expression of Lu in the ductular reactive cells in CDE- and DDC-treated mice?

Although it is unclear whether the reviewer inquires after the mechanism of transcriptional regulation of Lu or the origin of ductular reactive cells, we speculated the reasons as follows. As shown in the FACS data on normal diet (Figure 1C), most of EpCAM+ cells with middle expression of Lu must be normal bile duct around the portal vein. Therefore, we speculate that approximately a half of EpCAM+ cells with middle expression of Lu in the CDE- and DDC-injured liver may correspond to the original bile ducts, whereas the Lu+ (high) or Lu- cells may be derived from expandable progenitor cells. However, we have no evidence about the speculation at present, because there is no method to specifically label only ductal cells with characteristic of liver progenitor. Further study using a novel lineage tracing experiment will be required to prove the idea. Regarding the transcriptional mechanism, Lu has been originally identified as an antigen of human erythropoietic lineage, and there are consensus CACC-binding sites, a GATA-1 binding site and multiple E-boxes in the Lu promoter. Although the implication of KLF1 transcription factor in Lu expression has been previously reported in red blood cell disease, the transcriptional regulation of Lu in the other tissues remains largely unknown. We speculate that some kinds of factors implicated in each pathology, e.g. damage associated molecular patterns (DAMPs) released from damaged hepatocytes or bile acids may regulate the expression of Lu. However, we have no evidence about the opposite regulatory mechanisms of Lu expression between CDE- and DDC-injuries at present.

2) The authors propose that Lu and Integrin alpha3 beta1/alpha6 beta1 compete for binding to laminin alpha5 (Lama5), because inhibition of Integrin beta1 in cultured Lu- biliary cells mimics the effects of Lu overexpression. The concept of competition between Integrin and Lu should be strengthened and additional controls are required:

- Can the authors overexpress the alpha3 (or alpha6) and beta1 integrin subunits in cultured Lu+ cells and verify if the cells are converted to a cystogenic phenotype?

We thank the reviewer for the suggestion to strengthen the concept of competition between integrin and Lu. However, the suggested experiment may be difficult. We think that integrin signals are basically complicated due to many factors, e.g. a variety of combination patterns of alpha and beta subunits, distinct affinity to various ECM, and complex intracellular signaling including Fak and p130Cas pathways. Because we cannot control each expression level of two distinct transfected genes, the balance between alpha and beta subunits are supposed to be out of order in this experiment. In addition, it is unclear whether the transfected integrins can work well, because we could not find such experiments in the previous papers. Alternatively, it may be straightforward to show that the activation of beta1 integrin changes the phenotype of Lu+ cells into a cystogenic one. Therefore, we performed the addition of agonistic anit-beta1 integrin antibody to Lu+ cells in the 3D culture. As expected, Lu+ cells were converted to a cystogenic phenotype. We have added the new data in Figure 5—figure supplement 4.

- Is the expression of Integrin alpha3 beta1/alpha6 beta1 similar in vitro in Lu- and Lu+ biliary cells?

We have compared the transcriptional level of Integrin alpha3 beta1/alpha6 beta1 between Lu- and Lu+ biliary cells by real-time RT-PCR. As a result, both biliary cells expressed Integrin alpha3 beta1/alpha6 beta1 similarly suggesting that these cells are potentially competent to cell signaling via beta1 integrin-laminin axis. We have added the new data in Figure 5—figure supplement 2.

- The authors should provide information on the expression of Lu in Lu- cells incubated with Integrin beta1 neutralising antibodies.

According to the reviewer’s comment, we investigated the expression of Lu by real-time RT-PCR. As a result, there was no change of Lu expression after the incubation with neutralizing antibody. We have added the data in Figure 5—figure supplement 3.

3) The expression of Lama5 raises several questions.

- Lama5 is detected in vivo around ductular reactive cells in both CDE (Lu high) and DDC (Lu low) models, yet Figure 5B suggests that Lu- biliary cells express more Lama5 than Lu+ biliary cells. Is the difference shown in Figure 5B statistically significant?

Although the information on statistical analysis in Figure 5B was included in a source file “Figure 5B_Mann-Whitney.xlsx” at 1st submission, we forgot to mention it in the main figure. Therefore, we have specified “n.s.” in the revised Figure 5B. The scattering value of Lama5 expression in DDC samples may be due to the quality of freshly isolated EpCAM+ cells because DDC model causes cholangiopathy resulting in the damage or activation of EpCAM+ biliary cells. Considering the expression of Lama5 at a protein level in Figure 5A, we think that there is no apparent difference of Lama5 expression in most of ductular reactive cells between CDE and DDC models.

- Macrophages play a major role in laminin deposition during liver disease, including in DDC-treated livers. The authors should compare the mRNA expression of Lama5 also in this cell type.

According to the reviewer’s comment, we isolated macrophages from injured livers as CD45+Ly6G-F4/80+CD11b+ fraction by FACS, and examined mRNA expression of Lama5 in those cells by real-time RT-PCR. Compared to the EpCAM+ biliary cells, the expression level of Lama5 was marginal or not detected in each macrophage isolated from CDE- and DDC-injured livers as shown in Author response image 1. Although macrophages play a major role in ECM deposition during liver disease, we concluded that macrophages per se are not a producer of Lama5, but rather an activator of ECM-producing cells as reported in our paper (PMID:28779552, Matsuda et al. Hepatology 2018).

- In the CDE model, laminin appears to predominantly co-localise with ductular reactive cells into the parenchyma, while the Lu (high) population is restricted to a more proximal periportal zone (see comment #1). As per the authors' theory, when the cells lose Lu, laminin binding to integrin should be facilitated. Hence the cells should re-adopt a tubular, pseudo-ductular conformation, which is not the case. What is the authors' explanation?

We think the discrepancy may be explained by a different microenvironment surrounding the EpCAM+ cells between in vivo and in vitro. It is known that the extracellular matrix such as laminins and collagens are rich around the portal vein in vivo, while it is relatively poor in the parenchymal area. By using the 3D culture system that is ECM-rich condition including collagens and Matrigel, we evaluated the “cystogenic capacity” of Lu+ and Lu- cells in vitro. However, the amount and composition of ECM would be quite distinct between at the deeper area in the parenchyma and the periportal zone in vivo. Therefore, the microenvironment in the parenchyma would be unsuitable for solid tubular formation. We think that the liberation of biliary cells from ECM-rich periportal zone into ECM-poor parenchymal area is a most important role of Lu as evidenced by Lu KO phenotype. As mentioned above, we think that the EpCAM+ cells losing Lu at the tip of ductular reaction may be differentiating into hepatocytic cells. If the expression of Lama5 is also downregulated in those cells, the integrin signal via Laminin-511 may be attenuated, which is not the case with Lu- BC in ECM-rich periportal area of the DDC model. In either case, the morphological phenotype of ductular reaction would be controlled by not only Lu expression but also the microenvironment in vivo. We have discussed about the importance of microenvironment in the Discussion section (fifth paragraph). We believe that Lu orchestrates the complicated integrin signaling depending on the microenvironment surrounding ductular reactive cells.

- Stuart Forbes' group demonstrated that anchorage of liver progenitor cells to a laminin sheet maintains the biliary phenotype of the cells while loss of laminin favors hepatocyte differentiation. Also, in vivo, laminin degradation favors the differentiation of liver progenitor cells in the CDE model (PMID:22922013). Can the authors explain how they reconcile their data with the literature?

We basically agree with the idea, and our paper is consistent with the previous works referred to by the reviewer. As previously reported by us (Kaneko et al., 2015), ductular reactive cells substantially function as bile duct to drain bile. As shown in this paper, extending ductular reactive cells are basically surrounded by Lama5, suggesting that laminin sheet is required to maintain the biliary phenotype. By contrast, Lu seems to be downregulated in EpCAM+ cells at the tip of ductular reaction. If such EpCAM+ cells are differentiating into hepatocytic cells, the downregulated Lu may contribute to the decrease of laminin deposition surrounding those cells. We have added the speculation in the Discussion section (fifth paragraph).

4) Lu is proposed to control cell motility and cyst formation. The experiments in Figures 3 to 5 require additional controls:

- In the DDC model, ductular reactions expands as bile duct like shape in vivo. Do the Lu- biliary cells in vitro have the same shape of a control cholangiocytes? In other words, control cholangiocytes should be added as control in Figure 2.

We are not sure what cells the reviewer considers “control cholangiocytes” in this comment. If the reviewer considers EpCAM+ cells isolated from untreated normal liver as control cholangiocytes, we do not think it is a suitable control for comparison with Lu+ and Lu- BC. In this study, Lu+ and Lu- BCs were proliferative and used for every assay after at least 3 passages. On the other hand, as previously reported (Okabe et al., 2009), approximately 0.5% of EpCAM+ cells (50 cells per 10,000 cells) in normal liver can survive and proliferate to form a colony in this culture. We demonstrated that the colony forming cells include the cell having LPC-like properties. In other words, most of EpCAM+ cells i.e. mature cholangiocytes will be dead or excluded at a relatively early stage under this culture condition. Because the passaged EpCAM+ cells in the same manner do not represent normal cholangiocytes, such cells will be inappropriate for control cholangiocytes. Alternatively, we investigated the shape of EpCAM+ cells isolated from untreated normal liver after 7 days of primary culture. Although most of cholangiocytes were dead during the culture, the attached and survived cells showed a mixture of round and indefinite shape. Although we are not sure whether this data would be informative in this study, we have added the data in Figure 2—figure supplement 2.

- An additional biliary marker like EpCAM in Figure 4A would confirm that the observed changes are due to Lu expression and not to contamination with other cell types. Were these cells freshly isolated or passaged?

At first, the cells expanded at 3 to 6 passages were used for all in vitro assays as described in Materials and methods section (subsection “Cell Culture”). In the case of Figure 4A, we used the cells after 3 passages for transduction of Lu cDNA or GFP cDNA. We have added the FACS data about the EpCAM expression of the parental cells in Figure 4A, indicating that there is no contamination of other cell types.

- Concluding that Lu+ cell have higher motility than Lu-/low cells requires excluding higher proliferation in Figure 3A.

For scratch assay, we treated the tested cells with Mitomycin C to stop the proliferation as described in Materials and methods section (subsection “Scratch Assay”). Therefore, cell proliferation does not affect cell motility in this assay.

- Cyst size and cyst formation efficiency should be quantified in Figures3C, 4C and 5D.

We evaluated the cyst size and cyst formation efficiency in 3D assays. We have added the new data on quantification regarding Figures 3C, 4C, 5D and newly added Figure5—figure supplement 4.

5) A number of statements need to be clarified and additional information on the methods should be provided:

- In the subsection “Isolation of biliary cells with a distinct level of Lu from chronically injured liver”, the authors state that Lu (high)-derived biliary cells continued to express Lu at high levels. Information on the number of passages in Figure 2C should be given.

The cells after 6 passages were used for FACS analysis. We have added the information about the passage number in the legend of newly assigned Figure 2D.

- The statement "proliferating biliary cell with high expression of Lu in the CDE model" requires information about proliferation markers in Lu+ and Lu-/low cells.

Because Ki67+ cells were observed in the ductular reactive cells in the CDE model as shown in Figure 6B, we have described "proliferating biliary cell” unintentionally. We have no intention of claiming the correlation between cell proliferation and Lu expression at all. In fact, as shown in Figure 6B, the ratio of Ki67+ cells per EpCAM+ cells was equivalent between WT and Lu KO mice, strongly suggesting that Lu expression is not involved in the regulation of cell proliferation. Because this sentence may be misleading for now, we have deleted “proliferating” in the text.

- Since Lu is also expressed in arterial branches, the authors have to describe the FACS gating strategy used to sort the biliary cells and to exclude contamination by arterial endothelial cells. Similarly, it is shown in Figure 6E that hepatocytes in the CDE model express EpCAM. How are those hepatocytes excluded from sorting ? Are those hepatocytes also Lu (high)?

According to the reviewer’s advice, we have added the FACS gating strategy in Figure 2—figure supplement 1. For FACS, we used only non-parenchymal cell (NPC) fraction, of which the hepatocyte fraction was removed by repeated centrifugation at 100g for 2 minutes. Although the information is described in the reference (Okabe et al., 2009), we have added it in the Materials and methods section (subsection “Liver Perfusion and Cell Sorting”). In addition, we have added FACS data of triple staining with anti-EpCAM, anti-Lu and anti-PECAM antibodies in order to clarify no contamination of endothelial cells in the purified EpCAM+ cell fraction (Figure 2—figure supplement 1). These data demonstrated that a contamination of endothelial cells was highly unlikely. In addition, our culture condition is not suitable for passage of endothelial cells because it does not contain VEGF.

Regarding the Lu expression in EpCAM+ hepatocyte-like cells, a co-staining of the liver tissue was technically difficult because Lu staining required acetone fixation, which spoiled HNF4α staining at the same time. Because we have removed Figure 6E as described below, we did not refer to the expression of Lu in EpCAM+ hepatocytic cells in the text.

6) Lutheran knockout livers treated with CDE do only show very low numbers of EpCAM+/HNF4+ cells as compared to wild-type CDE-treated livers. The authors conclude from this experiment that high expression of Lutheran "might be a process of hepatocytic differentiation of liver progenitor cells". Such conclusion is not sufficiently supported by the data, since co-expression of EpCAM and HNF4 does not provide solid evidence for hepatocyte differentiation of progenitors to hepatocytes. The reviewers suggest to remove Figure 6E which is quite preliminary.

We are grateful for the reviewers’ suggestions. According to the comments, we have removed Figure 6E. We would like to focus on the morphological features of ductular reaction rather than cell fate in this paper.

7) It is not clear what the authors refer to by "slight dilation of duct-like structures was observed in Lu KO mice (Figure 6D and F)". Better images are required to support this statement.

In the use of phrase, we have mistaken “hyperplasia” for “dilation”. We have revised it and added magnified images to show the hyperplasia of duct-like structures in Figure 6F.

8) It would considerably expand the impact of the paper if some information is provided about the expression of Lu in ductular reaction in human liver disease.

With the cooperation of Prof. Kenichi Harada who is an expert on the pathology of liver diseases at Kanazawa University, we investigated a few cirrhotic liver samples obtained from the surgery of HCC. As expected, Lu was stained in ductular reaction in patients of liver diseases including NASH, HBV and HCV. We added new data in Figure 7.

https://doi.org/10.7554/eLife.36572.033

Article and author information

Author details

  1. Yasushi Miura

    1. Department of Regenerative Medicine, Research Institute, National Center for Global Health and Medicine, Tokyo, Japan
    2. Department of Life Science and Medical Bioscience, Graduate School of Advanced Science and Engineering, Waseda University, Tokyo, Japan
    Contribution
    Conceptualization, Resources, Data curation, Formal analysis, Validation, Investigation, Visualization, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared
  2. Satoshi Matsui

    1. Department of Regenerative Medicine, Research Institute, National Center for Global Health and Medicine, Tokyo, Japan
    2. Laboratory of Cell Growth and Differentiation, Institute of Molecular and Cellular Biosciences, The University of Tokyo, Tokyo, Japan
    Contribution
    Data curation, Formal analysis, Validation, Investigation, Visualization, Writing—review and editing
    Competing interests
    No competing interests declared
  3. Naoko Miyata

    Department of Regenerative Medicine, Research Institute, National Center for Global Health and Medicine, Tokyo, Japan
    Contribution
    Formal analysis, Investigation, Data acquisition by FACS
    Competing interests
    No competing interests declared
  4. Kenichi Harada

    Department of Human Pathology, Kanazawa University Graduate School of Medicine, Kanazawa, Japan
    Contribution
    Resources, Data curation, Formal analysis, Validation, Investigation, Visualization, Writing—review and editing
    Competing interests
    No competing interests declared
  5. Yamato Kikkawa

    Department of Clinical Biochemistry, Tokyo University of Pharmacy and Life Sciences, Tokyo, Japan
    Contribution
    Conceptualization, Resources, Writing—review and editing
    Competing interests
    No competing interests declared
  6. Masaki Ohmuraya

    Department of Genetics, Hyogo College of Medicine, Hyogo, Japan
    Contribution
    Resources, Funding acquisition, Visualization, Writing—original draft
    Competing interests
    No competing interests declared
  7. Kimi Araki

    Institute of Resource Development and Analysis, Kumamoto University, Kumamoto, Japan
    Contribution
    Resources
    Competing interests
    No competing interests declared
  8. Shinya Tsurusaki

    1. Department of Regenerative Medicine, Research Institute, National Center for Global Health and Medicine, Tokyo, Japan
    2. Laboratory of Stem Cell Regulation, Institute of Molecular and Cellular Biosciences, The University of Tokyo, Tokyo, Japan
    Contribution
    Data curation, Investigation, Visualization
    Competing interests
    No competing interests declared
  9. Hitoshi Okochi

    Department of Regenerative Medicine, Research Institute, National Center for Global Health and Medicine, Tokyo, Japan
    Contribution
    Supervision, Project administration, Writing—review and editing
    Competing interests
    No competing interests declared
  10. Nobuhito Goda

    Department of Life Science and Medical Bioscience, Graduate School of Advanced Science and Engineering, Waseda University, Tokyo, Japan
    Contribution
    Conceptualization, Supervision, Writing—review and editing
    Competing interests
    No competing interests declared
  11. Atsushi Miyajima

    Laboratory of Cell Growth and Differentiation, Institute of Molecular and Cellular Biosciences, The University of Tokyo, Tokyo, Japan
    Contribution
    Conceptualization, Resources, Supervision, Funding acquisition, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared
  12. Minoru Tanaka

    1. Department of Regenerative Medicine, Research Institute, National Center for Global Health and Medicine, Tokyo, Japan
    2. Laboratory of Stem Cell Regulation, Institute of Molecular and Cellular Biosciences, The University of Tokyo, Tokyo, Japan
    Contribution
    Conceptualization, Resources, Supervision, Funding acquisition, Investigation, Visualization, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    m-tanaka@ri.ncgm.go.jp
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2500-7973

Funding

Japan Society for the Promotion of Science (26110007)

  • Minoru Tanaka

Japan Society for the Promotion of Science (26253023)

  • Atsushi Miyajima

Japan Society for the Promotion of Science (26110001)

  • Masaki Ohmuraya

Japan Agency for Medical Research and Development (JP17be0304201)

  • Minoru Tanaka

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We would like to thank Dr. Jeffrey H Miner (Washington University, St Louis, MO) for supplying anti-Laminin alpha5, and Dr. Toshio Kitamura (The University of Tokyo, Tokyo, Japan) for providing a retroviral vector, pMxs/IRES-GFP. We also thank CY Kok for her editorial assistance, Y Kamiya for mouse and technical assistance, C Koga for cell sorting, the University of Tokyo IMCB Olympus Bioimaging Center (TOBIC) for acquiring image data, and the members of the Miyajima and Okochi laboratory for their helpful discussion and suggestions. This work was supported, in part, by Grants-in-Aid for Scientific Research on Innovative Areas (26110007 to MT and 26110001 to MO) and Scientific Research A (26253023 to AM) from the Japan Society for the Promotion of Science, Japan, and by Research Program (JP17be0304201 to MT) from Japan Agency for Medical Research and Development (AMED).

Ethics

Human subjects: The study using human samples was approved by the Kanazawa University Ethics Committee (approval number 305-4), and all of the analyzed samples are derived from patients who provided informed written consent for the use of their tissue samples in research.

Animal experimentation: All animal experiments were performed according to institutional guidelines and approved by the Animal Care and Use committee of the Institute of Molecular and Cellular Biosciences, The University of Tokyo (approval numbers 2501, 2501-1, 2609, 2706 and 3004), Kumatomo University (approval number A27-092), Hyogo College of Medicine (approval number 16-043, 16-046), and National Center for Global Health and Medicine Research Institute (approval number 15080, 16023, 17086 and 18069). Every effort was made to minimize animal suffering and to reduce the number of animals used.

Senior Editor

  1. Marianne E Bronner, California Institute of Technology, United States

Reviewing Editor

  1. Frédéric Lemaigre, Université catholique de Louvain, Belgium

Publication history

  1. Received: March 11, 2018
  2. Accepted: July 28, 2018
  3. Accepted Manuscript published: July 30, 2018 (version 1)
  4. Version of Record published: August 23, 2018 (version 2)

Copyright

© 2018, Miura et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 729
    Page views
  • 149
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)