Potential conflict of interest: Nothing to report.
It is well-accepted that hepatic stellate cells (HSCs) can develop into myofibroblast-like cells that synthesize extracellular matrix proteins and contribute to liver fibrosis. Recently, molecular markers of stem/progenitor cells were discovered in HSCs of rats. Moreover, the cells displayed the capacity to differentiate and to participate in liver regeneration. In addition, stellate cells possess signaling pathways important for maintenance of stemness and cell differentiation such as hedgehog and β-catenin–dependent Wnt signaling. All these properties are congruently found in stem/progenitor cells. Stem cells require a special microenvironment, the so-called stem cell niche, to maintain their characteristics. Thus, we investigated if the space of Disse, where stellate cells reside in the liver innervated by the sympathetic nervous system and surrounded by sinusoidal endothelial cells and parenchymal cells, exhibits similarities with known stem cell niches. The present study describes the niche of stellate cells within the liver of rats that is composed of sinusoidal endothelial cells, which release stromal cell–derived factor-1 to attract stellate cells via the cysteine-X-cysteine receptor 4, basal lamina proteins (laminin and collagen type IV), and parenchymal cells, which synthesize β-catenin–dependent Wnt ligands and Jagged1. Conclusion: The space of Disse shows analogies to typical stem cell niches comprising of basal lamina components, sympathetic innervation, and adjacent cells that constitute a milieu by paracrine factors and direct physical interactions to retain HSCs at this site and to influence their cellular fate. The space of Disse serves as a niche of stellate cells, which is a novel function of this unique organ structure. (HEPATOLOGY 2009.)
Stellate cells reside within the perisinusoidal space of Disse in the liver that is lined by parenchymal cells and fenestrated sinusoidal endothelial cells (SECs). Hepatic stellate cells (HSCs) display characteristic lipid droplets that contain vitamin A mainly as retinyl palmitate.1 After liver injury, quiescent HSCs become activated, lose their vitamin A stores, and develop into contractile myofibroblast-like cells, which secrete extracellular matrix proteins. It is widely accepted that these HSC-derived myofibroblast-like cells contribute to liver fibrosis.2 Controversies exist about the embryonic origin of HSCs, because they express proteins of all three germ layers.3 Bone marrow stem cells are able to generate cells of all germ layers, and it was suggested that the bone marrow might be a source of stellate cells in adult vertebrate liver.4, 5 This indicates a mesodermal origin of stellate cells, which was confirmed by a very recent study.6 In this study, submesothelial cells positive for activated leukocyte cell adhesion molecule (ALCAM) were shown to store lipid droplets after retinol treatment and to acquire the phenotype of HSCs.6 ALCAM is a cell-surface immunoglobulin superfamily member expressed by primitive hematopoietic stem cells and associated with embryonic hematopoiesis as well as vasculogenesis.7 Interestingly, stellate cells of rat liver display characteristics of stem/progenitor cells. They express CD133 (cluster designation 133) and Oct4 (octamer 4), which are known from pluripotent stem cells, and can differentiate into hepatocyte-like and endothelial-like cells in vitro. This points to a role for HSCs in liver regeneration.8 A recent study provided additional hints that HSCs are directly involved in restoration of liver mass also in vivo.9 Moreover, HSCs possess signaling pathways required for stem cell functions such as hedgehog and β-catenin–dependent Wnt signaling.10, 11 Similar to hematopoietic stem cells, quiescent HSCs are resistant against CD95-mediated apoptosis and proliferate in response to CD95 ligand.12–15 Thus, there is growing evidence that stellate cells are stem/progenitor cells of the vertebrate liver.
Adult stem cells maintain their characteristics throughout their lifetime in a special microenvironment, the so-called stem cell niche, a concept that was originally evolved for mammalian hematopoietic stem cells.16 The niche protects stem cells and fulfills certain qualities. For instance, differentiated cells adjacent to stem cells constitute a milieu by soluble factors and direct physical interactions to promote the undifferentiated state of stem cells and to retain them within the niche, often situated on a basal lamina.17 One of the soluble factors released by non–stem cell neighbors such as endothelial cells, osteoblasts, and other stromal cells of the bone marrow is stromal cell–derived factor-1 (SDF1).18, 19 The chemokine cysteine-X-cysteine receptor 4 (CXCR4) is a G protein–coupled seven-span transmembrane protein which binds SDF1 exclusively.20, 21 The SDF-1/CXCR4 axis plays an essential role in directing hematopoietic stem cells along a SDF1 gradient to their final niche in the bone marrow during ontogenesis,22 but also in their mobilization into the peripheral blood.23 Stem cell niches are further innervated by the sympathetic nervous system, which is involved in regulation of hematopoietic stem cell recruitment.24 The recruitment of stem cells from the bone marrow is at least in part mediated by local down-regulation of SDF1 that facilitates the release of stem cells into the blood stream. Therefore, the interaction of SDF1 and CXCR4 is an essential process of stem cell niches. In addition, Wnt signaling via β-catenin and Notch signaling are also crucial elements of stem cell niches. These signaling pathways maintain properties of stem/progenitor cells such as self-renewal by slow and symmetric cell divisions without differentiation.25, 26 Stem cells within their niche are normally quiescent and stay in G0 of the cell cycle. In the normal liver, stellate cells are also in G0 as indicated by the absence of Ki-67 synthesis of freshly isolated HSCs of rats.11 This quiescent stage of rat HSCs is characterized by a scanty cytoplasm containing lipid droplets with retinoids, the presence of glial fibrillary acidic protein (GFAP), and an undetectable α-smooth muscle actin (α-SMA) expression. The present study was carried out to investigate if elements of known stem cell niches are required to preserve the quiescent stage of stellate cells within adult rat liver. The findings suggest that the space of Disse serves as a niche for HSCs and displays analogies to stem cell niches.
Stellate cells, SECs, Kupffer cells, and parenchymal cells used in this study were isolated from the liver of male Wistar rats essentially as described earlier.1, 27, 28 The rats were obtained from the local animal facility of the university, and all animals received care according to the German Animal Welfare Act. Cell culture techniques and the procedures of semiquantitative and quantitative reverse transcription polymerase chain reaction (RT-PCR) as well as western blot and immunofluorescence staining techniques were carried out as described.11 Technical details and descriptions of migration assays are given in the Supporting Information.
HSCs Express CXCR4 and Notch Receptors.
Isolated HSCs of rats lost their typical lipid droplets and developed into myofibroblast-like cells during culture (Fig. 1A-C). Freshly isolated stellate cells and HSC-derived myofibroblast-like cells expressed CXCR4 as investigated through analysis of messenger RNA (mRNA) (Supporting Fig. 2) and protein level (Fig. 1D-F). Although the expression of the HSC marker GFAP decreased (Fig. 1G), the CXCR4 synthesis increased during their culture-dependent development into myofibroblast-like cells. Western blot analysis of HSC membrane/cytoskeleton fractions displayed elevated CXCR4 levels during culture time (Fig. 1H). Among liver cell types tested, only the SECs synthesized SDF1 (Fig. 1I). The mRNA of SDF1 was also abundant in freshly isolated HSCs (Supporting Fig. 2), but a typical protein band of SDF1 between 8 and 14 kDa as observed in SECs was not detected (Fig. 1I). The different liver cell types were also analyzed for their synthesis of the Notch receptor ligand Jagged1 (Jag1) that typically signifies stem cell neighbors in the stem cell niche. Although the mRNA of Jag1 and Jag2 were found (Supporting Fig. 2), quiescent stellate cells displayed no protein synthesis of these Notch ligands. Jag1 appeared in protein lysates of parenchymal cells and Kupffer cells as investigated by western blot analysis (Fig. 1J). The protein Jag1 was only found in cultured HSCs that developed into myofibroblast-like cells (not shown). The Notch receptors (Notch1, Notch2, and Notch4) were present at the mRNA level in freshly isolated HSCs (Supporting Fig. 2). The Notch target genes hairy and enhancer of split 1 (Hes1) as well as hairy/enhancer of split-related with YRPW motif-like (Heyl) were expressed in freshly isolated HSCs, demonstrating active Notch signaling. Notch1 was further detected by western blot analysis at approximately 120 kDa in HSCs that were investigated after 1 day of culture. The synthesis of Notch1 decreased during culture and development of HSCs into myofibroblast-like cells (Supporting Fig. 3). The decline of Notch1 expression followed reduced nuclear β-catenin levels of cultured HSCs,11 suggesting that β-catenin–dependent Wnt signaling was involved in the regulation of Notch1 synthesis. This was further confirmed with the small synthetic molecule TWS119 (GSK-3β-inhibitor 3-[[6-(3-aminophenyl) 1H-pyrrolo [2,3-d] pyrimidin-4-yl]oxy]-phenol) known to mimic β-catenin–dependent Wnt signaling by inhibiting the glycogen synthase kinase-3β.11 Low concentrations of TWS119 (0.1 μM) increased the Notch1 protein levels by 82% ± 24% (P < 0.01; n = 5) in cultured HSCs compared to untreated control cells (Supporting Fig. 3).
SDF1-Dependent Migration of HSCs.
When HSCs were cultured under serum-free conditions in a Boyden chamber (Fig. 2A), freshly isolated (1 day) as well as activated HSCs (7 and 14 days) were able to migrate in response to 50 ng/mL SDF1α or SDF1β (Fig. 2B-D). The well-known effect of fetal bovine serum (FBS) on HSC migration was used as a positive control in this experiment. The migratory response to SDF1α/β exceeded that under the influence of FBS at all three time points. In addition, SDF1β exhibited the greatest effect, and myofibroblast-like cells after 7 days of culture had apparently the highest migration capability. As shown above, SECs were identified as an important source of SDF1 in the normal rat liver (Fig. 1I). Therefore, the migration of HSCs was investigated in Boyden chambers in the absence or presence of SECs under serum-free conditions (Fig. 3A,B). SECs strongly attracted HSCs, and this effect was completely antagonized through addition of 0.5 μg/mL SDF1 antibodies (Fig. 3B).
Liver Parenchymal Cells Affect HSCs.
Freshly isolated HSCs were cocultured with liver parenchymal cells and the two cell types were separated by a membrane that allowed permeation of soluble factors only. When HSCs were cultured alone (Fig. 4A1) their cell size increased (Fig. 4A2) and myofibroblast-like cells with elevated α-SMA expression developed within 7 days of culture as determined by immunofluorescence staining (Fig. 4A3). Under this culture condition, the HSCs lost their typical GFAP staining (Fig. 4A4) and displayed frequently two or more nuclei per cell (Fig. 4A3, 4A4). Cocultures with parenchymal cells for 7 days (Fig. 4B1) preserved the morphology of freshly isolated stellate cells with lipids (Fig. 4B2) and retinoids that emitted fluorescent light after ultraviolet light excitation (not shown). Compared to HSCs grown in monoculture, the α-SMA synthesis remained low (Fig. 4B3), the staining pattern of GFAP was close to that of freshly isolated stellate cells (Fig. 4B4), and polynucleated cells were rare under coculture conditions (Fig. 4B3, 4B4). Western blot analysis showed that α-SMA was elevated by 56% ± 11% (P < 0.001; n = 5) and the GFAP content was lowered by 81% ± 30% (P < 0.05; n = 5) in HSCs maintained in monocultures compared to those cocultured with parenchymal cells (Fig. 4C). Also the synthesis of the matrix metalloproteinase 13 (MMP13), which typically increased during development of HSCs into myofibroblast-like cells (not shown), was elevated by 57% ± 5% (P < 0.001; n = 4) in HSCs under monoculture conditions compared to the coculture system. In cocultures, the Notch1 expression of HSCs was 150% ± 58% (P < 0.05; n = 4) higher than in HSCs of monocultures. Moreover, Ki-67 was increased by 69% ± 14% (P < 0.01; n = 5) in HSCs grown in the monoculture system, suggestive for elevated cell proliferation. Apparently, parenchymal cells released soluble factors that support quiescence of HSCs. As shown recently, β-catenin–dependent Wnt signaling is involved in keeping HSCs quiescent.11 Therefore, the nuclear β-catenin levels were analyzed by western blot and found to be increased by 184% ± 58% (P < 0.05; n = 5) in HSCs cocultured with parenchymal cells (Fig. 4D). In the presence of parenchymal cells, the expression of the Wnt signaling target gene Myc in HSCs was 183% ± 36% (P < 0.01; n = 4) higher than in HSCs of monocultures as analyzed by quantitative RT-PCR after 24 hours of culture. Also, the mRNA amounts of CD133 and Oct4 in HSCs of cocultures were 280% ± 190% (P < 0.05; n = 4) and 1236% ± 620% (P < 0.05; n = 5) above the expression levels of HSC monocultures (Supporting Fig. 4). The protein synthesis of CD133 decreased during culture of HSCs and their development into myofibroblast-like cells, but remained elevated during coculture with parenchymal cells. The CD133 protein synthesis was 46% ± 13% (P < 0.01; n = 5) higher in HSCs of cocultures than in HSCs maintained in monocultures (Fig. 4C; Supporting Fig. 5).
The expression of canonical Wnt target genes by HSCs grown in coculture implied the presence of Wnt ligands. The synthesis of Wnt ligands such as Wnt7a/b and Wnt10b by HSCs, parenchymal cells, SECs, and Kupffer cells were subsequently investigated, and parenchymal cells were found to be a major source of these canonical Wnt ligands (Fig. 4E). The expression of Wnt7a/b and Wnt10b by hepatocytes was also confirmed by RT-PCR (Supporting Fig. 2). Increased nuclear β-catenin levels of HSCs indicated that parenchymal cells released Wnt ligands. However, the influence of hepatocytes on HSCs differed with respect to their stage of activation. In HSCs cultured for 7 days without parenchymal cells, an expression of liver markers such as α-fetoprotein, albumin, hepatocyte nuclear factor 1α (HNF1α), HNF 4α, HNF6, and multidrug resistance protein 2 (MRP2) was normally undetectable (Supporting Fig. 6A,C); this was also the case in coculture experiments, where HSCs preserve some characteristics of their quiescent stage (Supporting Fig. 6B,C). When HSCs, which developed into myofibroblast-like cells during 7 days of culture, were cocultured for additional 7 days with parenchymal cells, the expression of hepatocyte markers was induced (Supporting Fig. 6D-F). Obviously, the effects of parenchymal cells largely depend on the stage of HSC activation or development.
The Stellate Cell Niche Architecture.
The basal lamina proteins laminin and collagen type IV are able to promote the quiescent stage of stellate cells as described by earlier studies.29–31 The colocalization of these two components with the stellate cell marker GFAP was investigated by immunofluorescence staining in the normal rat liver. Stellate cells were found embedded between laminin (Fig. 5A,B) and collagen type IV (Fig. 5C,D) in liver sinusoids. The membrane-bound ligand of Notch signaling Jag1 was detected in the cell membrane of parenchymal cells, and HSCs that synthesize Notch1 (Supporting Figs. 2 and 3) were in close contact to these Jag1-presenting hepatocytes (Fig. 5E,F).
Similar to hematopoietic stem cells of the bone marrow, HSCs synthesize CXCR4 and its expression is sustained in HSC-derived myofibroblast-like cells. Quiescent HSCs as well as myofibroblast-like cells were able to migrate in response to SDF1α/β, demonstrating a functional CXCR4. SDF1 was found to be synthesized by SECs, and they were able to attract HSCs in an SDF1-dependent manner. Endothelial cells of the bone marrow are already known as a source of SDF1, which is involved in homing and retention of bone marrow stem cells.18 In analogy to this, the attraction of HSCs by SECs via SDF1 could be a mechanism to retain stellate cells in their niche, the space of Disse (Fig. 6). The HSC migration requires elevated CXCR4 synthesis and local inactivation of SDF1 by MMP.32 The migration of HSCs was shown to be dependent on MMP2 and MMP9.33 Lowered SDF1 synthesis of SECs that probably support HSC migration can be experimentally induced through basic fibroblast growth factor (bFGF). A decrease of SDF1 mRNA levels in freshly isolated SECs by 91% ± 5% (P < 0.001; n = 3) compared to untreated controls was induced by 50 ng/mL bFGF within 24 hours (Supporting Fig. 7). However, elevated SDF1 expression, as observed during early liver regeneration, might be also a mechanism to retain HSCs in the liver and to attract additional stem cell compartments of the body. The mobilization of hematopoietic stem cells from the bone marrow can be induced by granulocyte-colony stimulating factor (G-CSF) that triggers the release of proteases to degrade adhesion molecules and SDF1. Noradrenergic signals of the sympathetic nervous system are thought to be critical mediators of G-CSF–induced mobilization of stem cells.24 Nerve endings are found in the liver close to HSCs, and HSCs in situ respond to perivascular nerve stimulation with a release of the osmolyte myoinositol and exhibit Ca2+ signals in response to phenylephrin.34–36 Stimulation with noradrenaline, the neurotransmitter of the peripheral sympathetic nervous system, results in a rapid release of prostaglandin F2α and prostaglandin D2 by freshly isolated HSCs. These prostaglandins activate the glycogenolysis in the neighboring parenchymal cells.37 These findings indicate that the sympathetic nervous system may also be relevant for the HSC niche (Fig. 6).
Another purpose of the niche is the preservation of a slow-cycling, self-renewing, and undifferentiated state of stem cells. Preservation of an undifferentiated state of HSCs in the presence of parenchymal cells was indicated by the sustained CD133 and Oct4 expression. Although only a subset of HSCs can be selected by antibodies against CD133 and magnetic cell sorting,8 indicating variable antigen abundance, CD133 and Oct4 can be stained with antibodies on all adherent HSCs that survived cell isolation without cell sorting.11 Therefore, the expression of stem/progenitor cell markers is apparently a feature of all quiescent HSCs. Stem cells maintain their characteristics within their niches and are often found on a basal lamina. Examples are muscle-specific stem cells, so-called satellite cells that stay in close contact to the basement membrane of myofibers and undifferentiated spermatogonia on the basement membrane of the seminiferous epithelium of the testis, where the germ stem cell niche is thought to be located.38, 39 The interaction of stem cells with extracellular matrix proteins can suppress the onset of their terminal differentiation.40 In line with this, basal lamina proteins such as laminin and collagen type IV promote the quiescent stage of HSCs29–31 and, therefore, represent important elements of their niche (Fig. 6).
Several factors released or presented by non–stem cells are required for stem cell functions and are important for their niche. Signaling pathways such as Wnt, Notch, Janus kinase–signal transducer and activator of transcription (JAK-STAT), bone morphogenetic protein, Activin/transforming growth factorβ/Nodal, and hedgehog interact to maintain stemness or to influence the developmental fate of stem cells within the niche.40–42 In a previous study, the relevance of β-catenin–dependent Wnt signaling for sustaining quiescent HSCs was already demonstrated.11 In a search of the origin of canonical Wnt ligands among cells of the stellate cell niche, liver parenchymal cells were identified as an important source. Elevated nuclear amounts of β-catenin and increased expression of the Wnt target gene Myc in HSCs indicated that Wnt ligands from parenchymal cells affect stellate cells. Moreover, Notch1 synthesis was found to be controlled by β-catenin–dependent Wnt signaling, as suggested by treatment of HSCs with the glycogen synthase kinase-3β inhibitor TWS119, and was increased in HSCs cocultured with parenchymal cells. Elevated Notch1 levels are typical for freshly isolated HSCs. Therefore, Wnt signaling via β-catenin apparently promotes the quiescent stage of HSCs. Thus, Wnt ligands are most likely additional elements of the stellate cell niche (Fig. 6). How specific HSCs will respond to Wnts and other factors apparently depends on their stage of development and intrinsic genetic program. In contrast to freshly isolated HSCs, culture-activated HSCs expressed the hepatocyte markers α-fetoprotein, albumin, HNF1α, HNF6, and MRP2 in the presence of parenchymal cells, suggestive for changes of their genetic program during activation. Notch signaling is also known to guide cell fate decisions of stem cells. The expression of Notch1 by HSCs and the synthesis of its ligand Jag1 by parenchymal cells further support the novel concept of a stellate cell niche in the liver, because stem cell neighbors synthesize Jag1 to support self-renewal of stem cells via Notch1 within the niche.26 The relevance of this signaling pathway for HSCs is currently unknown; however, due to the typical distribution of Jag1 and Notch1, which is consistently found in stem cell niches, and the expression of Notch target genes (Hes1, Heyl) by HSC, Notch signaling was included in the scheme of the stellate cell niche (Fig. 6). Further studies are required to investigate if Notch1-mediated signaling is essential to maintain the quiescent stage of HSCs.
In conclusion, the space of Disse is a stellate cell niche within rat liver, a novel function of this unique organ structure. This niche is composed of basal lamina proteins and adjacent cells like parenchymal cells and SECs that secrete paracrine factors to maintain the characteristics of HSCs and to retain them within the space of Disse. Parenchymal liver cells control the cellular fate of HSCs depending on their stage of activation. It will be of utmost importance to investigate how signaling pathways such as Wnt and Notch interact to control the stellate cell niche, and if they are affected in disease and aging.
We are grateful to Claudia Rupprecht for her excellent technical assistance and to the German Research Foundation (Deutsche Forschungsgemeinschaft, SFB575, Experimental Hepatology) and the Research Commission of the Medical Faculty of the Heinrich-Heine-University, Düsseldorf, for their financial support.