Mesenchymal Stromal Cells as Supportive Cells for Hepatocytes

Abstract

Hepatocytes and hematopoietic stem cells (HSCs) appear to share many of the same requirements for their survival, functionality, and proliferation. This may be due to a shared location during fetal development. Moreover, hepatocytes and HSCs are unable to function, or even survive, without stromal cell support. Bone marrow–derived mesenchymal stromal cells (MSCs) support the proliferation and functionality, not only of HSCs, but also of hepatocytes. Although knowledge of the mechanisms underlying HSCs' support is far more advanced than for hepatocytes, data suggest that many agents important for HSCs also maintain the normal hepatocyte phenotype in vitro. Thus, it is possible that new techniques for the maintenance and expansion of HSCs may also be useful for hepatocytes. Bone marrow–derived MSCs are easily cultured and expanded in vitro, and some data suggest that they are immunoregulatory as well as relatively nonimmunogenic. These observations suggest that allogeneic MSCs may be useful not only in supporting hepatocyte growth and proliferation but also in modulating immune responses such as stellate cell activation.

Introduction

All organs in the body comprise a supportive component, the stroma, and a functional component, the parenchyma, and a symbiotic^1,2 relationship between the two is increasingly recognized.^3,4,5 In the liver, hepatocytes are supported by the hepatic vascular system and the biliary tree, in addition to the connective tissue septa. Cells from each of these structures have been shown to influence hepatocytes in vitro, either through cytokine and/or matrix secretion^3,6,7,8,9 or by direct cell contact communication.^3,10 Similarly, hematopoiesis, which starts in the yolk sac, fetal liver, thymus, and spleen, but becomes exclusive to the bone marrow compartment, also depends on a stromal component at all of these locations.¹¹ In the bone marrow, the stroma is similarly composed of marrow vasculature and connective, or reticular tissue. The cells that elaborate the latter, like all connective tissue lineages, are derived from a mesenchymal stem cell. However, in the absence of definitive experimental demonstration of the existence of a mesenchymal stem cell, The International Society for Cellular Therapy has referred to these cells as bone marrow mesenchymal stromal cells (BM-MSCs), and described them as culture-adherent; positive for CD105, CD73, and CD90; negative for hematopoietic-specific surface molecules; and which can at least differentiate in vitro into bone, cartilage, and fat cells.¹² Hematopoietic stem cells (HSCs) are unable to either survive, proliferate, or develop their full differentiation potential in the absence of MSCs or a supportive cell population.^13,14 In rare cases, the adult liver is still able to support HSCs and their progenitors, and thus has the potential to provide HSC requirements.^15,16 This capability to provide stromal support to HSCs poses an interesting question: If cells from the liver can support HSCs, can cells from the bone marrow support hepatocytes?

Hepatocytes are cells that, like HSCs, are unable to function effectively without appropriate environmental cues. When hepatocytes are implanted in vivo, they survive extrahepatically in the spleen and thymus; two sites where fetal hematopoiesis occurs.^17,18 In vitro, hepatocytes in basal medium lose their functionality and die in a matter of days.^1,9,19 Supplements such as dexamethasone, ascorbic acid, nicotinamide, epidermal growth factor are able to maintain these cells for longer periods with a slower loss of their functionality.^20,21 Other attempts to improve in vitro cultures have been reported by also coculturing hepatocytes with a variety of other cells types in an effort to emulate the original in vivo environment (reviewed by Bhatia et al.³). The cell types used have included endothelial and epithelial cells, nonparenchymal liver cells, such as liver epithelial cells or stellate cells, fibroblasts, and BM-MSCs.^{1,3,22,23,24,25,26,27,28,29,30,31,32} Of these, BM-MSCs present important advantages: they are easily expanded in vitro, and some data suggest that they are immunoprivileged, immunomodulatory.^33,34,35 Thus, they may serve an additional role in hepatocyte transplantation by diminishing the host allogeneic response. Furthermore, given that BM-MSCs can regulate stellate cell activation^36,37 they may also induce regression of cirrhosis.

MSCs Support Hepatocyte Function

Corlu et al. showed that MSCs are able to preserve hepatocyte morphology longer than hepatocyte monocultures.^10,38 They also demonstrated that the membrane-associated liver-regulating protein (LRP) is in part responsible for this support; and when antibodies against this protein are added daily to the cocultures, the effect is lost.¹⁰ LRP is expressed by BM-MSCs, hepatocytes, liver epithelial cells, stellate cells, Kupffer cells, and similar cell types such as those in the thymus and spleen.³⁹ In cocultures of rat hepatocytes, and rat liver epithelial cells, they showed that LRP is essential for maintaining the mature hepatocyte phenotype and that cell contact is required for this effect.⁴⁰ Interestingly, this protein appears to be implicated in HSC support: when LRP activity of BM-MSCs and rat liver epithelial cells is inhibited, HSC supportive capacity is significantly diminished.⁴¹

The capacity of BM-MSCs to support hepatocytes has been further investigated both in vivo and in vitro. In vitro, in addition to many functional characteristics, the morphology of hepatocytes is maintained. Isoda et al. showed that hepatocyte/BM-MSC cocultures were able to keep the albumin and ammonia metabolic capacity at higher levels than in controls.²⁴ This appeared to be independent of cell–cell contact. Both conditioned medium and transwell coculture showed either the same or better maintenance than observed with contact cocultures. They concluded that interleukin-6 (IL-6) was one of the factors involved, as urea production was significantly improved when compared with untreated monocultures, or cultures with antibodies against IL-6. This factor has a cytoprotective effect, which has been demonstrated both in vivo and in vitro.^42,43 In the presence of harmful agents, hepatocytes exposed to IL-6 are able to maintain higher levels of albumin and urea secretion and also exhibit a better capacity to metabolize drugs.⁴² IL-6 has previously been described to have effects on liver regeneration; it can induce proliferation but in some cases can cause growth arrest.^44,45 This may be explained by the findings of Sun et al. who showed that IL-6 alone induces quiescence in hepatocytes, but in the presence of nonparenchymal cells, the effect is the opposite. IL-6 stimulates nonparenchymal cells to produce hepatocyte growth factor, which is a strong hepatic mitogen and one way by which it induces hepatocyte proliferation.⁴⁵

Mizuguchi et al. showed additionally that cell contact was important for hepatocyte proliferation while the differentiated state was maintained.¹ They showed maintenance of the expression of hepatic-specific genes and proteins. One of the important proteins maintained was tryptophan 2,3-dioxygenase, which is downregulated significantly upon hepatocyte dedifferentiation. C/EBPα and C/EBPβ are also present. C/EBPα is a transcription factor that induces the adult hepatocyte phenotype and promotes quiescence.^1,46 Conversely, C/EBPβ is crucial for normal hepatocyte proliferation and response to growth factors.^47,48 It appears that BM-MSC/hepatocyte coculture allows for a balanced coexpression of these two transcription factors, showing proliferation without the loss of the mature phenotype. Hepatocytes in coculture remain polarized and able to form bile canaliculi, characteristics of adult hepatocytes.⁴⁹

Additionally, Gu et al. showed that hepatocyte/BM-MSC cell proportions, soluble factors, and secreted extracellular matrix components were involved in hepatocyte maintenance.⁵⁰ Unlike others, Gu et al. used porcine BM-MSCs and hepatocytes, a species which could be potentially used for bioartificial liver systems. They demonstrated that the proportion of 2:1 (hepatocyte/BM-MSC) enabled hepatocytes to exhibit greater albumin and urea synthesis. The investigators also demonstrated that when BM-MSC matrix secretion is impaired using small-interfering RNAs, the effect of BM-MSCs on hepatocytes was diminished; albumin and urea production decreased compared to regular contact cocultures. Despite the effect of extracellular matrix proteins on the interaction, it was mentioned that in coculture without cell contact, the levels of both albumin and urea syntheses are higher than in monoculture, thus confirming the role of soluble factors on hepatocyte maintenance.

In other configurations, Takeda et al. cocultured BM-MSCs and hepatocytes in hydroxyapatite scaffolds.²³ They observed higher albumin production levels in vitro. In addition, scaffolds, with hepatocytes and BM-MSC, implanted into analbuminemic rats and cirrhotic mice increased serum albumin levels significantly more than did a scaffold with hepatocytes alone. Furthermore, they also detected higher levels of IL-6 in the cirrhotic mice implanted with scaffolds with cocultures. This higher level of IL-6 in serum was suggested by Takeda et al. as one of the factors that could improve the liver function in these models. One aspect that may have been of significance, using BM-MSCs as stroma, is the fact that BM-MSCs are strong angiogenic inducers through the secretion of vascular endothelial growth factor.^51,52 This alone would allow for higher hepatocyte survival in the scaffold and also a higher systemic effect due to the ability of hepatocytes to interact with the host's bloodstream.

In most of the hepatocyte/BM-MSC coculture literature, it could be argued that BM-MSCs have hepatic potential that can be induced both by growth factors and by coculture, either by direct cell–cell contact or separated by a semipermeable membrane.^{26,53,54,55,56} This can result in misinterpretation of data when testing the capacity of BM-MSC to support hepatocytes, as higher levels of albumin and urea can be generated through cells differentiating from the heterogeneous BM-MSC population, rather than from the hepatocytes alone. Nevertheless, Lange et al., showed using GFP-labeled BM-MSCs, that hepatocytes maintained their viability in such cocultures in contrast to hepatocyte monocultures.²⁶ In addition, BM-MSC hepatic differentiation likely accounts for only a small percentage of the population, as conditioned medium is able to induce changes similar to those of transwell cocultures. Taken together, these data suggest that any differentiation that occurs may not significantly change the outcome of hepatocyte/BM-MSC coculture experiments.^24,57

Furthermore, BM-MSC-conditioned medium not only maintains hepatocyte functionality, improves hepatocyte survival and proliferation in vitro^24,57 but can also have pronounced effects in vivo, providing significant rescue from fulminant hepatocyte failure^57,58 In a rat model, where fulminant hepatocyte failure was induced with D-galactosamine, the animals were treated with conditioned medium or saline. Treatment with conditioned medium resulted in reduced apoptotic hepatocellular death and a lower inflammatory response effects that translated into a higher survival rate in the animals.^57,58 These data suggest that BM-MSCs may provide a means to synthesize active biotherapeutic agents rather than being used directly in conventional cell therapy.

Putative Mechanisms for MSC–Hepatocyte Interactions

BM-MSCs appear to provide a number of cues for hepatocyte growth and development (Figure 1). They secrete cytokines important in hematopoiesis that also mediate hepatocyte proliferation and differentiation. Hepatocyte growth factor enhances hematopoiesis by acting synergistically with other factors^59,60,61 and is a highly potent mitogen of hepatocytes, induces hepatocyte maturation, and has a cytoprotective effect.^61,62,63 Stem cell factor, involved in hematopoietic progenitor cell survival, renewal, and differentiation^44,64 also stimulates hepatocyte proliferation.^44,64 Stem cell factor appears to act downstream of tumor necrosis factor-α (TNF-α) and IL-6 hepatocyte proliferative pathways, as both TNF-α and IL-6 proliferative effects are, in part, due to stem cell factor.^44,64 Liver nonparenchymal cell IL-6 secretion is stimulated by TNF-α, also secreted by BM-MSCs.⁶⁵ Upon liver damage, TNF-α serum levels increase, thereby activating IL-6 expression.⁶⁶ TNF-α also has direct effects on hepatocyte proliferation and matrix secretion,⁶⁷ and also affects HSCs by increasing their proliferation and differentiation rate in the presence of IL-3 (ref. 68). Epidermal growth factor is a potent mitogen of hepatocytes, produced by BM-MSCs at variable levels.^63,67,69 Transforming growth factor-β, an antimitogenic agent, is secreted by BM-MSCs.⁶³ However, as these cells adapt to different environments and change their cytokine secretion rate,⁶⁹ it is possible that the net effects, BM-MSCs have over hepatocytes, are mitogenic. TNF-α confers resistance to HSCs against transforming growth factor-β, which also has antiproliferative effects on HSCs,⁷⁰ an event that could also happen with hepatocytes, diminishing any antiproliferative effect arising from transforming growth factor-β.^69,71

Figure 1.

Putative mechanisms by which MSCs support hepatocytes, as described in the text. Reciprocal mechanisms are not illustrated since they are beyond the scope of this article.

Another environmental cue that significantly affects hepatocyte functionality is the extracellular matrix.^6,7,8,50 BM-MSCs in coculture actively synthesize collagen type-I, which can embed hepatocytes to provide an effect similar to a collagen sandwich monoculture.^7,8,9,49 Collagen gel sandwich cultures of hepatocytes help to preserve hepatocyte function for weeks as opposed to standard monocultures where functionality is lost within the first days of culture.^7,8,9 Other proteins that have effects on hepatocyte functionality and can be secreted by BM-MSCs include laminin and fibronectin.^6,50,72,73 Of note, laminin is mainly expressed in fetal liver while fibronectin is produced in adult liver.⁶ Other possible important extracellular components secreted by BM-MSCs include dermatan and chondroitin sulfate–containing proteoglycans. They have been associated with the preservation of cell communication junctions and maintenance of normal levels of hepatocyte-specific genes. ^74,75,76,77

Cell contact also plays an important role on hepatocyte maintenance. At low densities for instance, hepatocyte proliferation is enhanced.⁷⁸ Direct contact effects have been shown by Corlu et al. as an important aspect for the maintenance of hepatocyte morphology¹⁰ and by Mizuguchi et al. as a mediator of the polarized state and cell proliferation.^1,49 As mentioned previously, Corlu et al. identify LRP as a key player of cell–cell communication. However, cell–cell communication in this system is more complex. For example, connexin-43 correlates with the capacity of fat-storing cell clones in maintaining hepatocytes.⁷⁹ Connexin-43 is not only present in BM-MSCs but may be involved in inter-MSC communication.⁸⁰ It is conceivable that this gap-junction protein provides another means for BM-MSCs to communicate with hepatocytes.

It has been suggested that Jagged1-dependent Notch signaling is involved in BM-MSC/hepatocyte communication due to an increase of Jagged1 expression in BM-MSCs when cocultured in direct contact with hepatocytes.¹ Notch signaling is a conserved evolutionary mechanism initiated by a ligand–receptor interaction between neighboring cells.^81,82,83 It regulates a broad range of events during embryonic and postnatal development, including proliferation, apoptosis, border formation, and cell fate decisions.⁸² Notch signaling is involved in the development of vertebrate self-renewing organs, during tumorigenesis, in the inhibition of differentiation, in lineage specification at developmental branch points, and in the induction of differentiation.^{81,82,84,85,86} More specifically, the Jagged1–Notch interaction is implicated in cell fate decisions in hepatoblasts and promotes the formation of hematopoietic primitive precursor cell populations.^81,83,87,88 Furthermore, downregulation of Notch signaling promotes the development of the mature hepatocyte phenotype in hepatoblasts,⁸¹ suggesting that its upregulation might be related to a controlled dedifferentiation that leads to proliferation.

HSC and progenitors form cell–cell contacts with BM-MSCs in vivo.⁸³ In vitro, such contacts appear essential for the maintenance of long-term HSC cultures.⁸³ Stromal cells from both fetal liver and bone marrow express Jagged1;⁸⁷ in addition, hepatocytes express Notch1 both in vivo and in vitro;⁸⁹ and in the regenerating liver the levels of Jagged1 and Notch1 are upregulated.⁹⁰ This evidence further supports the possibility that BM-MSCs interact with hepatocytes by the Jagged1–Notch1 mechanism.

BM-MSCs are capable of differentiating into multiple lineages in vitro and in vivo to bone, cartilage, marrow stroma, adipose cells.⁷² It is unclear therefore whether BM-MSCs found in vivo or a type of differentiated BM-MSCs are involved in hepatocyte maintenance. Mizuguchi et al. suggest that BM-MSCs, when supporting hepatocytes, are mainly fibroblasts due to their high production of collagen type-I.⁴⁹ Given that BM-MSCs have the capability of differentiating into other cell types, it is possible that an adipocytic or preadipocytic cell provides the support required by hepatocytes, especially because of data implicating fat-storing cells in mediating this nurturing process.^30,79 It can, however, be argued that epithelial cells such as the rat liver epithelial cells are able to provide support to hepatocytes. It should be noted, however, that BM-MSCs are very similar to the pericyte population,⁹¹ hence the stellate cell may be the closest type anatomically to support hepatocytes, given that it is both lipocytic and pericytic.⁹²

Other Sources of MSCs

Recent studies show that MSCs share characteristics that are common regardless of the tissue source.^93,94,95 In conducting in vitro studies, it is worthwhile to explore alternative sources of MSCs as a potential support for hepatocytes. For example, adipose-derived mesenchymal progenitors are easily acquired in great quantities although like BM-MSCs, their harvesting is invasive.⁹⁶ Moreover, they do not support hematopoiesis as efficiently as do BM-MSCs.^97,98 In contrast, placenta and umbilical cord–derived MSCs are obtained noninvasively, support hematopoiesis^99,100,101 and data suggest that they have a similar nonimmunogenic and immunomodulatory phenotype to BM-MSCs.^98,99 Further studies are required with these easily accessible alternatives to assess their ability to support hepatocytes in vitro. Studies of the coculture of MSCs with hepatocytes have uncovered important mechanisms that mediate the maintenance, proliferation and differentiation of hepatocytes, and may eventually lead to their manufacture for clinical use.