The remarkable regenerative capacity of the liver has been appealing to human imagination since centuries. In Greek mythology, the Greek gods exploited the regenerative capacity of the liver when they punished Prometheus by having an eagle eat his liver, which grew back overnight. Nevertheless, the cellular and molecular mechanisms that underlay liver regeneration remain poorly understood.
The liver plays a key role in many life-supporting metabolic functions (e.g., bile production, glucose metabolism and glycogen storage, vitamin storage, metabolization of drugs and other xenobiotics, production of albumin and coagulation factors, etc.). Therefore, failure of restoring basic hepatic functions can have devastating consequences and can become life threatening. Interestingly, and relatively unique to the liver, the first line of defense against liver damage is that mature hepatocytes, that are chiefly in G0
, re-enter the cell cycle to replace the lost hepatocytes, or undergo hypertrophy. However, if hepatocytes are damaged too extensively to participate in the liver repair process, liver regeneration occurs via differentiation of (facultative) hepatocyte progenitor/stem cells. The origin of these progenitors remains not fully understood, and likely includes liver progenitor cells, cholangiocytes, and intermediary hepatocytes (recently reviewed in [1
]). However, if liver damage persists, or gives rise to liver fibrosis and cirrhosis, none of the above mechanisms can repair lost hepatocytes, giving rise to liver failure. In such cases, orthotopic liver transplantation (OLT) is currently accepted as the gold-standard treatment. OLT is also the treatment of choice for certain inborn liver-based metabolic diseases (reviewed in [2
]). Even though OLT can provide a cure, long waiting lists related to shortage of suitable donors, the morbidity associated with this therapy, the high cost of the treatment, and the need for life-long immunosuppressive treatment are some of the reasons why alternative therapies are sought after.
One such alternative for OLT is transplantation of individual cells. Cells already used clinically are primary human hepatocytes, isolated from cadaveric livers, to regenerate the liver tissue. Other cells, in clinical trials or tested in preclinical studies, include hepatocyte progenitors or expanded liver derived cells, but also cells from tissues other than liver, like mesenchymal stromal cells, and transdifferentiated or dedifferentiated cells that are then redifferentiated to cells with hepatocyte features. Mechanisms of action of these latter cells include direct regeneration, but also fusion with existing hepatocytes and—as a result of this—in vivo transdifferentiation into hepatocytes [3
], or support of regeneration via secretion of paracrine factors [5
]. In this review, we will focus chiefly on which (stem) cell populations are candidate cells for engraftment and repopulation of the liver, based on preclinical and clinical studies. The definitions we will be using to describe the different effects exerted by transplanted cell populations on the liver are detailed in Table 1
3. Grafting of Gold-Standard Control, Freshly Isolated Primary Human Hepatocytes
In 1969, Berry and Friend described a collagenase-based perfusion protocol which enabled them to efficiently isolate primary hepatocytes from rat livers [28
]. As this discovery made it possible to isolate viable single cells from hepatic tissue, it laid the foundations for the subsequent development of hepatocyte transplantations. Almost one decade later, Groth et al. were the first to perform an actual transplantation of isolated primary rat hepatocytes into the liver of Gunn rats [29
Over the years, several adaptations of the collagenase perfusion method led to the application of this protocol for the isolation of primary human hepatocytes (PHHs) [30
], whose repopulation potential has been extensively studied in rodents. Transgene-induced liver damage mouse models have been used to evaluate the repopulation of PHHs in mouse liver. In uPA-SCID mice, hepatocyte loss is relatively mild, which is seen from the relative high survival rates of non-chimeric animals, but is protracted, allowing efficient repopulation by human hepatocytes [26
]. Meuleman et al. obtained stable engraftment of human hepatocytes that were successfully infected with HCV for at least 4 months with RI ranging between 25% and >90%, depending on the source of PHHs [31
]. Nevertheless, humanized uPA-SCID mice are not as healthy as their littermates transplanted with murine hepatocytes, suggesting that human grafts do not necessarily affect the overall long-term health of the recipient animals in a positive manner [31
]. Similarly, Tateno et al. reported RIs up to 96% in uPA-SCID mice treated with anti-asialo GM1 antibodies and Futhan 6 months after transplantation [26
]. Omission of anti-asialo GM1 antibodies and Futhan caused significant mouse mortality within 2 months even if the RI was >50%. Tateno et al. also successfully transplanted PHHs in hemizygous cDNA-uPA-SCID mice, achieving RIs similar to those observed in uPA-SCID mice for at least 6 months.
PHHs can also successfully repopulate the liver of FRG mice. In the original report, 1 × 106
PHHs were transplanted in FRG mice wherein the uPA
gene was overexpressed using adenoviral vectors [17
]. This yielded RIs in ±1/3rd of recipients of >1% and RIs of >30% in 16% of recipients where human albumin levels of >1 mg/mL were found, and this after only 1 round of NTBC re-administration. Subsequent studies demonstrated similarly high levels of repopulation even without uPA adenoviral vector infusion, when 5 × 106
PHHs were grafted [35
]. Moreover, human hepatocytes, harvested from primary transplanted FRG mice can be highly successfully re-transplanted in secondary FRG mice [36
], effectively expanding human hepatocytes in mice in vivo. This has enabled the routine use of mice with a humanized liver (Yecuris FRG and FRGN mice) for studies related to viral infections, gene therapy, drug metabolization, and disease modeling [36
In addition, human hepatocyte repopulation in other transgene-induced liver damage mouse models, such as TK-NOG and AFC8 mice, was also demonstrated, even if RIs appear lower in these mice than uPA-SCID or FRG mice. Hasegawa et al. observed repopulation of ±43% in TK-NOG mice [22
], while Washburn et al. demonstrated a RI of ±15% in AFC8 mice [24
] (of note, cells grafted in AFC8 mice were not PHHs but hepatocyte progenitor cells).
Both freshly isolated as well as cryopreserved PHHs have been used to graft in mice [26
]. However, cryopreserved PHHs often have a poorer cell viability compared to freshly isolated PHHs, and their functionality and plating ability are significantly poorer compared to freshly isolated PHHs [37
]. Nevertheless, successful transplantations have been performed with cryopreserved PHHs and it should be noted that the use of cryopreserved PHHs is more user-friendly in a preclinical setting [26
The observation that human hepatocytes could repopulate mouse livers led to the first transplantation of primary hepatocytes in 10 patients in 1992 [40
]. Since then, several reports have been published describing PHH transplantation in patients with a number of different liver diseases (reviewed in [41
]). Despite these promising (pre)clinical results obtained with PHHs, the bottleneck in the whole process persists, namely, the shortage of suitable donor material. In preclinical mice models, serial transplantation can be used to overcome this problem. As described above, isolation of donor hepatocytes from a primary host and subsequent transplantation of these hepatocytes into a secondary host enables the expansion of fresh PHHs from the same source and the generation of a large number of transplanted animals from a single donor [17
]. However, this is not clinically applicable. Therefore, researchers have been investigating strategies for expanding the limited pool of PHHs or generating de novo hepatocytes from other cell sources suitable for downstream applications (Figure 1
Finally, it is of importance to note that mature (adult liver-derived) hepatocytes are the true gold-standard cells for liver repopulation. It is, indeed, generally accepted that fetal hepatocytes have a far less engraftment/repopulation potential than their mature counterparts [42
], likely because they are less capable to home and subsequently expand in the adult cell niche of the host liver. Therefore, in the search for alternatives for PHHs, it will likely be necessary to use/create cells with functional characteristics that approximate those of fully mature hepatocytes.
Mature hepatocytes, isolated from livers that are not suitable as donors for whole organ transplantations remain the gold standard source of cells for liver repopulation. However, as they are a scarce commodity, researchers are actively looking for alternative cell sources for cell-based liver transplantations. If these alternative cell populations could be generated from the patient him/herself, they may also circumvent the need for immunosuppressive treatment. In this review, we highlight that currently, none of the described alternative cell populations have the same ability to engraft and repopulate the liver to a similar level as PHHs, although some of them (MSCs and AECs) have been proven to be successful in alleviating symptoms of certain liver diseases, even without repopulating the host liver. Although, engraftment of the alternative cell sources (e.g., 2D cultured hepatocytes, LPCs, liver organoids, and (trans)differentiated HLCs) can be observed in the mouse livers, they fail to repopulate the injured livers to the same extent as PHHs (see Table 2
). This is not surprising, as fully mature PHHs have a significantly higher propensity to engraft and repopulate the liver, compared with for instance fetal hepatocytes [42
]. When PHHs are expanded in vitro, they dedifferentiate (i.e., they undergo epithelial-to-mesenchymal transition and rapidly loose mature hepatic functions within 24–72 hours after culturing) and hence acquire an immature more fetal-like phenotype [151
]. This is also true for all the alternative cell populations tested until now.
We believe that the failure of cells, other than mature PHHs, to robustly engraft and repopulate the host liver are two-fold (Figure 2
): (1) immature or dedifferentiated hepatocytes cannot home efficiently into the liver microenvironment, and (2) inability of the engrafted cells to survive and proliferate in the host liver. These conclusions follow from the following experimental observations:
First, in strong contrast with PHHs, immature cells (be it fetal hepatocytes, dedifferentiated and expanded PHHs, LPCs, transdifferentiated “hepatocytes”, or PSC-HLCs) may not sufficiently recognize signals for invading and engrafting into the adult hepatic tissues when administered via the blood circulation. This can explain why engraftment efficiency drastically increases when these cells are directly introduced as a sheet of cells in the liver parenchyma [141
This problem will only be solved once we understand the cellular and molecular processes underlying the poor homing and engraftment potential of the immature cells. It is believed that hepatocytes, when transplanted via the circulation, become entrapped in the liver sinusoids, through which they need to migrate to reach the liver parenchyma. When trapped in the sinusoid, this causes transient portal hypertension, and due to a transient ischemia-reperfusion injury, Kupffer cells become activated, induce endothelial activation and enhance vascular permeability. Hepatocytes then attach to the endothelium and migrate through the fenestrae to home into the parenchyma. Although this process is not very well understood, it is believed that interactions between grafted hepatocytes, Kupffer cells, liver sinusoidal endothelial cells, stellate cells, and the surrounding extracellular matrix all play a role in this homing and initial engraftment process [153
]. The process is relatively inefficient, as even for the gold-standard adult liver derived PHHs, it is believed that only a small fraction of hepatocytes can home into the parenchyma. Some studies have been done to track whether transplanted cells engraft into the liver using, for instance, superparamagnetic iron oxide nanoparticles (SPIO) labeled hepatocytes. One drawback of this approach is that activated Kupffer cells and macrophages in and around the sinusoid lead to very fast scavenging of dying SPIO labeled hepatocytes, making interpretation by magnetic resonance imaging (MRI) of the presence of hepatocytes difficult [156
]. As an alternative, Indium-111-labeled cells could be used, as has been done clinically [157
]. Therefore, to devise methods that would enhance the homing and engraftment of alternative sources of hepatocytes, these processes will need to be better studied.
Using RNA sequencing and/or proteomics, defining the molecular make-up of for instance fetal and adult PHHs, different batches of PHHs with differing engraftment potential, or possibly even better, genetically identical, early and late passage ex vivo expanded PHHs (both in 2D and 3D) should also aid in deciphering why specific populations have a reduced engraftment and repopulation potential. The observation that some human and mouse HLCs may be more capable of engrafting the liver, when they are harvested from the primary grafted animal, and used to engraft into secondary mice, suggests that once cells have adapted to an adult in vivo liver microenvironment, engraftment capacities are boosted. Further expanding on the molecular studies that evaluated the re-isolated hepatocytes with single cell/single nuclei RNA sequencing studies in comparison with the initially grafted cells might therefore also provide insights in what molecular changes are occurring in the cells when placed in the in vivo liver environment, and might provide clues on how to enhance homing and engraftment [118
Except for generating cells that are better adapted to the in vivo hepatic environment, this environment itself has a big influence on the engraftment efficiency. A strong correlation exists between the level of liver damage and engraftment/repopulation efficiencies; however, inducing more extensive liver damage to enhance homing is not clinically translatable. Moreover, the more severe the injury, the greater the inflammatory response, which then may lead to excess scavenging by Kupffer cells/macrophages of the grafted cells. It is interesting to note that studies have demonstrated that co-transplantation of hepatocytes and MSCs increases the number of donor hepatocyte clusters in the liver, which was attributed to the fact that the cells home better to the liver [158
] by enhancing attachment of the hepatocytes. In addition, the strong immunomodulatory/anti-inflammatory effects of MSCs may aid in the initial survival of donor hepatocytes in the inflamed liver niche environment [159
]. Therefore, it might be interesting to test if co-culture or co-infusion of MSCs (and/or AECs), that themselves cannot repopulate the liver parenchyma, with alternative sources of hepatocytes can enhance the latter’s homing and initial survival.
The second important issue to address is how to enhance proliferation (and survival) of engrafted cells to obtain repopulation levels that are clinically relevant. Indeed, introducing an anti-apoptosis gene in HLCs improved their survival in vivo [142
], and providing chemical compounds that support proliferation also appeared to enhance repopulation of PSC-HLCs [20
]. Although interfering with cell death processes and enhancing cell proliferation hold intrinsic risks for enhancing tumor formation, we believe that devising strategies that would increase proliferation may be an interesting avenue to improve liver repopulation by immature hepatocytes, the more that some of these compounds have also been suggested to increase regeneration potential of endogenous hepatocytes [160
]. However, if the maturation state of the alternative cell sources could be improved, these measures may not be required. Furthermore, we also have to keep in mind that that the risk of tumorigenesis remains due to incomplete differentiation of the transdifferentiated hepatocytes or PSC-HLCs. Better differentiation protocols will therefore also be needed to overcome this problem.
We conclude that none of the alternative sources of “hepatocytes” for liver repopulation currently can robustly engraft and repopulate the liver. Comparing cells derived from the liver (2D expanded hepatocytes, liver progenitor cells, and organoids) with cells derived from tissues other than liver (transdifferentiated hepatocytes or PSC-derived hepatocytes), no very clear differences exist in their ability to home, engraft, and proliferate in the host liver; and some differences in repopulation ability reported appear to be mainly due to differences in animal models used, and the inclusion or not of additional immunomodulating, anti-apoptosis, and proliferation inducing modifications in the engraftment studies.
From a clinical perspective, the optimal cell source would be cells that can be obtained from easily accessible and autologous sources, that can be easily expanded to multiple billions of cells, and that do not hold the risk for tumor formation. iPSC-derived HLCs best fulfill the first two criteria, whereas expanded cells (in 2D or as organoids) likely hold less chance for tumor formation. Nevertheless, neither of these cell technologies are currently fully mature, and hence underperform significantly in regards to engraftment and repopulation compared with PHHs. However, progress over the last 5–10 years has been remarkable. Ongoing and future studies are now aimed at understanding how to create more mature progeny from in vitro hepatocyte sources. Studies should also identify means to improve homing/engraftment and subsequent proliferation and repopulation by the less mature cells we can currently generate, and/or to enhance the receptiveness of the liver microenvironment for hepatocyte engraftment. Such insights will then pave the way for starting to test regenerative strategies starting from cultured hepatocyte progeny.