Bioreactors are the essential component of bioartificial livers (BALs) or Liver Assist Devices (LADs) that have been developed for clinical use in order to keep patients with acute liver failure alive until their liver regenerates or they receive a transplant. In order to be a successful cell type employed in such a device the cells need to exhibit a high and broad level of hepatic functionality. In this respect, HepaRG™ represents a very promising cell for such use since they have high phase 1 and 2 activities and transporter functions and can be produced in large quantities.
Although HepaRG™ cells are maximally differentiated for drug metabolizing functions by DMSO, other hepatic functions including ammonia elimination, urea production, galactose elimination and expression of the urea cycle enzyme, carbamoylphosphate synthetase, as well as glutamine synthetase, albumin and transferrin, are markedly higher when DMSO is removed from the culture (Hoekstra et al., Int. J. Biochem. Cell Biol., 2011). This finding is consistent with our current research, which has revealed that HepaRG™ cells are heterogeneous in their phenotype and are a mixture of periportal and centrilobular hepatocytes - the ratio of each phenotype may be affected by the DMSO concentration. Periportal hepatocytes, which contain essential urea cycle enzymes, preferentially detoxify ammonium by metabolising it to urea; whereas, centrilobular hepatocytes, containing glutamine synthetase, preferentially incorporate ammonium to form glutamine.
HepaRG™ cells have been applied to the Academic Medical Center bioartificial liver (AMC-BAL) (Figure 1A). The resulting HepaRG-AMC-BALs, cultured in the absence of DMSO, demonstrated good ammonia and lactate elimination, as well as apolipoprotein A-1 production at rates comparable to primary human hepatocytes, whilst retaining good levels of CYP3A4 (Nibourg et al., PLOSone, 2012). A 3D architecture with extensive cell-cell contact is achieved around the capillaries and matrix fibres present in the BAL (Figure 1B and C).
The inclusion of carbamoylglutamate in the medium increases the urea-cycle activity further by activating carbamoylphosphate synthetase. This is particularly important when developing BALs, which have the essential function of removing highly toxic endogenous metabolites such as ammonium. For example, HepaRG-AMC-BALs have been shown to increase the survival time of rats with induced acute liver failure (Nibourg et al., PLOSone, 2012). Moreover, the progression of hepatic encephalopathy and kidney failure caused by ammonia accumulation was attenuated by HepaRG-AMC-BAL treatment.
Darnelle et al. (DMD, 2011) investigated the application of HepaRG™ cells to a scaled-down bioreactor (2ml volume, Figure 2A). The arrangement of the cells was similar to that in-vivo such that Pgp and MRP2 proteins located at the apical membranes of hepatocytes forming bile canaliculi-like structures (Figure 2B and C). Low levels of intracellular LDH and AST enzyme leakage and a constant secretion of albumin over the incubation confirmed low cell damage in the bioreactor. While CYP1A1/2, CYP2C9, and CYP3A4 activities were stable over weeks, CYP2B6 increased over this time (Table 1). Moreover, CYP3A4 activities could be inhibited by ketoconazole and CYP3A4 and CYP2B6 activities could be induced by rifampicin, the extent of which were similar to those observed in-vivo (Table 1). This bioreactor model will enable long-term, sequential kinetic DDI studies to be performed using the same cell culture system.
Figure 2. Dynamic 3D bioreactor.
(A) Analytical scale 2 ml bioreactor; (B) CK19 stained biliary canaliculi (green); (C) phase-contrast image of cell aggregates between capillaries.
Adapted from Darnelle et al., DMD, 2011.
“HepaRG™ cells will be suitable for BAL application.”
Hoekstra et al., Int J Artif Organs, 2012.
The concept of a 3D culture bioreactor has been adapted to develop a high-throughput screening assay involving the use of stirred bioreactors (spinner vessels) to identify chronic effects of compounds (Leite et al., Toxicol. Sci., 2012). The spinner vessels contain multiple spheroids that can either be used for on-line sampling or aliquoted into multi-well formats. Such in-vitro models need to exhibit good metabolic activities, a polar arrangement of the hepatocytes resembling the in-vivo architecture and longevity. Leite et al. showed that HepaRG™ cells formed spheroids with tissue-like arrangements of cells (Figure 3) that exhibited stable levels of CYP1A2, CYP2B6, CYP2C9 and CYP3A4 activities over 3 weeks, which were all responsive to prototypical inducers over this time. In addition, CYP3A4 and UGT activities were demonstrated up to 7 weeks in culture, the latter being significantly higher than UGT activities present in 2D monolayer cultures.
“HepaRG™ cells cultured in 3D spinner-bioreactors are an attractive tool for toxicological studies, showing a liver-like performance and demonstrating a practical applicability for toxicodynamic approaches.”
Leite et al., Toxicol. Sci., 2012
The “hanging drop” method is a simple but elegant method of producing hepatocyte spheroids of a precise and standardised size for high-throughput assays (see Insphero website for more details on the method for GravityPLUS plates). HepaRG™ cell form 2000-cell spheroids of 200 µm diameter after 3-6 days using this technique and the resulting 3D microtissues, maintained over 3 weeks, resemble that of the liver, including the polarised location of bile canaliculi (Figure 4) (Gunness et al., Tox Sci., 2013). Unlike the HepaRG™ spheroids reported by Leite et al. (Toxicol. Sci., 2012), in which both cholangiocytes and hepatocytes were present, the hanging drop spheroids were composed almost completely of hepatocytes. Albumin secretion by HepaRG™ spheroids was consistently higher (15-30 pg/day/cell) than that in 2D HepaRG™ cell cultures (5-10 pg/day/cell). Likewise, the secretion of urea, glucose, lactate and pyruvate were generally much higher in 3D HepaRG™ spheroids than 2D cultures, especially at the later timepoints (17-21 days). The cytotoxicity of acetaminophen (the EC50 of which was comparable across two labs) was higher in HepaRG™ spheroids than in 2D cultures, an observation that was attributed to the significantly higher CYP2E1 activities in 3D, resulting in greater production of the toxic intermediate, NAPQI.
Recently, Rebelo et al. (Arch. Toxicol., 2014) investigated the structure and metabolic function of HepaRG™ cells microencapsulated in alginate and cultured in DMSO-free medium. The spheroids resembled those in Figure 3 in terms of a polarised architecture and functional MRP2. Importantly, CYP3A4 and CYP1A2 activities in these spheroids were not compromised by the absence of DMSO and were comparable to those in differentiated 2D cultures.
The use of metabolically stable HepaRG™ spheroids will allow for repeat dose long-term toxicity studies, and in combination with non-parenchymal cells such as Kupffer cells, will help to elucidate the toxicity of compounds not detected in conventional 2D short-term primary hepatocytes cultures.
Biochip microfluidic models: More complex in-vitro hepatic models have been developed including co-cultures and fluidic systems that allow medium to flow over the cells. These models take into account that the liver is composed of more than one cell type; moreover, the microfluidic conditions recreate the in-vivo flow of blood and architecture of the cells, thus increasing their functionality. Another use of this model is to investigate whether the toxicity of a compound in an organ e.g. the kidney or brain, may be the result of its bioactivation to a toxic metabolite in the liver. An example of this was demonstrated by Choucha-Snouber et al. (Biotechnol Bioeng., 2013), who used HepaRG™ and MDCK cells in a liver-kidney co-culture model in a microfluidic biochip (Figure 5). They demonstrated that the nephrotoxicity of the anticancer agent, ifosfamide, was due to its biotransformation via hepatic pathways involving CYP3A4/5 and CYP2B6 to acrolein. When MDCK cells alone were incubated with ifosfamide, it was non-toxic but it became markedly toxic to these cells when co-cultured with HepaRG™ cells.
Figure 5. Photo of the perfusion system (A) and the details of the liver-Kidney microfluidic bioreactor (B).
Taken from Choucha-Snouber et al., 2010. Read full publication
LiverChip™ Perfusion Plate: HepaRG™ cells have also been applied to the perfused model, LiverChip™. The LiverChip™ itself is a scaffold in which hepatocytes form a 3D architecture mimicking the capillary bed structure of the liver sinusoid. More information on this model can be found on liverchip.com
Invitrocue: HepaRG™ cells have been successfully applied to HepatoCue multi-well plate formats. This technology involves the conjugation of varying ratios of RGD peptide and galactose (GAL) bioligands to increase cell attachment and performance. More information on this technology can be found on invitrocue.com
“Among the tested co-culture model, the HepaRG™ cells appear more suitable than HepG2/C3a to simulate the liver behaviour.” Choucha-Snouber et al. (Biotechnol Bioeng. 2013)
HepaRG™ cells are progenitor cells with a bipotent differentiation capability and therefore make a suitable cell type with which to repopulate livers. The resulting humanized-liver mice can be used to investigate drug metabolism and toxicity, as well as hepatitis virus infections. Early studies by Cerec et al. (Hepatology, 2007) showed that SCID/bg mouse livers could be repopulated with HepaRG™ cells. Two months after transplantation, the mice were shown to produce the human a1 antitrypsin protein, detected in serum using ELISA. In addition, the presence of human cells in the liver was confirmed by histological staining of human serum albumin in liver sections taken from mice (Figure 6).
Recent research by Higuchi et al. (Xenobiotica, 2014) in which partially differentiated HepaRG™ cells (after 7- and 21-day differentiation in 2D) were transplanted into TK-NOG mice showed that 12 weeks after transplantation, they had fully differentiated into mature hepatocytes or biliary cells within the recipient livers (Figure 7). In contrast to differentiated cells, proliferative HepaRG™ cells differentiated only towards biliary-like cells, which expressed cytokeratin-19 but no human albumin. Immunohistological staining showed the hepatocyte-like cells stored glycogen and expressed human albumin, CYP3A4 and the multidrug resistance-associated protein 2 transporter. An additional advantage of HepaRG™ cells is they also showed negligibly low tumorigenicity (i.e. they are unlikely to become cancerous in the mice) and therefore would not affect the generation of humanized-liver mice.
“HepaRG™ is a promising cell source for the steady generation of humanized-liver mice.”
Higuchi et al., 2014.
As demonstrated by Cerec et al. (Hepatology, 2007), HepaRG™ cells are bipotent progenitor cells that express the main markers of in-vivo hepatic progenitors (see HepaRG™ Features). If HepaRG™ hepatocyte-like cells are plated at low density, they transdifferentiate into both hepatocytes and biliary cells. This characteristic has been investigated further to determine how HepaRG™ cells can be used to generate other cell types, such as pancreatic cells and cholangiocytes. An example of the latter has been reported by Dianat et al. (Hepatology, 2014), who showed that these progenitor cells can be differentiated into cholangiocytes using growth hormone, epidermal growth factor, IL-6, followed by taurocholate. The resulting cholangiocytes displayed adult-specific markers (cytokeratin 7 and osteopontin, SOX9 and HNF-6) and functions (such as calcium release in response to hormonal stimulation). This plasticity displayed by HepaRG™ cells opens up a vast area of research not only into the generation of different cell types but also in generating additional banks of the HepaRG™ cells themselves.
Figure 1. HepaRG-AMC-BAL (A); HA staining of transverse sections with the gas capillaries (red arrows) (B) and structure of the polyester matrix fibres with HepaRG™ cells (C).
Bars: 200 µm (A) and 20 µm (B).
Adapted from Hoekstra et al., Int. J. Biochem. Cell Biol., 2011 (A) and Nibourg et al., PLOSone, 2012 (B and C).
Table 1. Basal CYP activities in HepaRG™ cells in analytical scale bioreactor over time and the effects of ketoconazole and rifampicin.
Adapted from Darnelle et al., DMD, 2011.
Figure 3. 3D organization of the HepaRG™ spheroids.
Pgp efflux transporters (red), gathered near the accumulation of the green actin staining, indicating the formation of biliary canaliculum–like structures (white arrow).
Adapted from Leite et al., Toxicol. Sci., 2012.
Figure 4. Structure of HepaRG™ spheroids after 20 days in culture.
(A) HE staining; (B) MRP-2 activity, assessed using CMFDA assay
Adapted from Gunness et al., Toxicol. Sci, 2013.
“3D organotypic HepaRG™ cultures permitted more accurate assessment of acute toxicity of tested compounds, which is similar to in vivo hepatotoxicity. These 3D systems, in addition, will be of great value in assessing chronic toxicity because they are functional for long periods.”
Gunness et al., Toxicol. Sci, 2013.
Micropatterning is a recently developed culture technique involving the attachment of one or more cells in pockets, or “micropatterns”, in an organised arrangement. The method used by Mercey et al. (Biomatrials, 2010) to culture HepaRG™ cells involved seeding the cells into 6 mm deep fibronectin-coated micropatterns arranged on an agarose matrix. Under these conditions, purified HepaRG™ hepatocytes were able to rebuild a biliary pole and express high levels of CYP3A4. The differentiated status of HepaRG™ hepatocytes makes them ideal for use in high-throughput screening of toxic effects of compounds in metabolically functional cells.
This is especially relevant for the genotoxicity comet assay, the analysis of which requires the nuclei to be suspended in agarose gel and an electrical current passed across them to separate out the DNA fragments (thus forming the DNA comet tails). HepaRG™ cells cultured in micropatterns allow for bioactivation of progenotoxins in a relevant target cell and subsequent processing for comets. The processing the cells for comets, involving detachment of the cells using trypsin, is time-consuming but, importantly, it also causes higher background DNA damage that, if too high, can invalidate the assay.
Liver-intestinal models: Co-culture models have been used to help elucidate cross-talk between tissues occurring in-vivo. For example, Rossi et al. (TIV. 2012) used a Caco-2-HepaRG™ co-culture trans-well model to investigate retinoid transport across the intestine and metabolism in both the intestine and hepatocytes. Caco-2 cell monolayers in the insert formed a barrier between the compound added to the apical compartment and the metabolically functional HepaRG™ cells cultured on the bottom of the basolateral compartment. HepaRG™ cells were shown to exhibit retinol-dependent secretion of Retinol Binding Protein 4 (a carrier protein secreted by hepatocytes to maintain physiological concentrations of retinol and transport to other organs) and up-regulate the mRNA expression of the retinoic-metabolising enzyme, CYP26A1, in response to retinoid treatment.
Primary hepatocyte-HepaRG™ models: In addition to their use in co-cultures of cells from different organs and of different cells within the liver, HepaRG™ cells have also been used in co-cultures to increase the lifespan of primary hepatocytes (Dembélé et al., Protocol Exchange, 2014 (doi:10.1038/protex.2014.003)). The life span of monkey primary hepatocytes was increased from 10-14 days to longer than a month when co-cultured with HepaRG™ cells. The resulting model was then used to study infection by parasites.
Stellate cells-HepaRG™ models: Basu et al. (Apoptosis, 2006) used trans-well culture format to co-culture HepaRG™ and activated humans stellate cells (HSCs) to determine mechanisms involved in the apoptosis of HSCs. Cells were cultured alone or as a co-culture in which HepaRG™ cells were placed in the insert above the HSCs on the bottom of the well or vice versa – this allowed them to confirm that a soluble mediator secreted from immortalized human hepatocytes played an important role in HSC growth regulation.
Figure 6. Staining of transplanted (A) and untransplanted (B) SCID, upa+/- mouse liver for human serum albumin.
Adapted from Cerec et al., Hepatology, 2007.
Figure 7. Immunohistological analysis of humanised livers from TK-NOG mice transplanted with HepaRG™ cells.
Taken from Higuchi et al. (Xenobiotica, 2014)