Functional and morphological comparison of three primary liver cell types cultured in the AMC bioartificial liver


  • Paul P.C. Poyck,

    1. Department of Surgery (Surgical Laboratory), Academic Medical Center, University of Amsterdam, Amsterdam, the Netherlands
    2. AMC Liver Center, Academic Medical Center, University of Amsterdam, Amsterdam, the Netherlands
    3. Liver Transplantation Unit, A. Cardarelli Hospital, Naples, Italy
    4. Centro Biotechnologie, A. Cardarelli Hospital, Naples, Italy
    5. Department of Gastroenterology and Hepatology, Academic Medical Center, University of Amsterdam, Amsterdam, the Netherlands
    Search for more papers by this author
  • Ruurdtje Hoekstra,

    1. Department of Surgery (Surgical Laboratory), Academic Medical Center, University of Amsterdam, Amsterdam, the Netherlands
    2. AMC Liver Center, Academic Medical Center, University of Amsterdam, Amsterdam, the Netherlands
    Search for more papers by this author
  • Albert C.W.A. van Wijk,

    1. Department of Surgery (Surgical Laboratory), Academic Medical Center, University of Amsterdam, Amsterdam, the Netherlands
    Search for more papers by this author
  • Chiara Attanasio,

    1. Liver Transplantation Unit, A. Cardarelli Hospital, Naples, Italy
    2. Centro Biotechnologie, A. Cardarelli Hospital, Naples, Italy
    Search for more papers by this author
  • Fulvio Calise,

    1. Liver Transplantation Unit, A. Cardarelli Hospital, Naples, Italy
    2. Centro Biotechnologie, A. Cardarelli Hospital, Naples, Italy
    Search for more papers by this author
  • Robert A.F.M. Chamuleau,

    1. Department of Gastroenterology and Hepatology, Academic Medical Center, University of Amsterdam, Amsterdam, the Netherlands
    Search for more papers by this author
  • Thomas M. van Gulik

    Corresponding author
    1. Department of Surgery (Surgical Laboratory), Academic Medical Center, University of Amsterdam, Amsterdam, the Netherlands
    • Dept. of Surgery (Surgical Laboratory), Academic Medical Center, Meibergdreef 9, 1105 AZ Amsterdam, The Netherlands
    Search for more papers by this author
    • Telephone: 0031-20-5665570; FAX: 0031-20-6976621


The selection of a cell type for bioartificial liver (BAL) systems for the treatment of patients with acute liver failure is in part determined by issues concerning patient safety and cell availability. Consequently, mature porcine hepatocytes (MPHs) have been widely applied in BAL systems. The success of clinical BAL application systems is, however, largely dependent on the functionality and stability of hepatocytes. Therefore, we compared herein the general metabolic and functional activities of MPHs with mature human hepatocytes (MHHs) in the Academic Medical Center (AMC)-BAL during a 7-day culture period. We also tested fetal human hepatocytes (FHHs), since their proliferation capacity is higher than MHHs and their function is increased compared to human liver cell lines. The results showed large differences between the 3 cell types. MHHs eliminated 2-fold more ammonia and produced 3-fold more urea than MPHs, whereas FHHs produced ammonia. Lidocaine elimination of FHHs was 3.5-fold higher than MPHs and 6.6-fold higher than of MHHs. Albumin production was not different between the 3 cell types. MPHs and FHHs became increasingly glycolytic, whereas MHHs remained metabolically stable during the whole culture period. MHHs and MPHs formed tissue-like structures inside the AMC-BAL. In conclusion, we propose that FHHs can be considered as a suitable cell type for pharmacological studies inside a bioreactor. However, we conclude that MHHs are the preferred cell source for loading a BAL device for clinical use, because of their high ammonia eliminating capacity and metabolic stability. MPHs should be considered as the best alternative cell source for BAL application, although their phenotypic instability urges application within 1 or 2 days after loading. Liver Transpl 13:589–598, 2007. © 2007 AASLD.

Chronic liver disease is responsible for over 1.4 million disability adjusted life years annually1 and ranks in the United States among the top 7 disease-related causes of death between the age of 25 and 64 yr.2 For end-stage liver failure, orthotopic liver transplantation remains the current treatment of choice. However, patients suffering from acute or acute-on-chronic liver failure may benefit from temporary extracorporeal artificial liver support used to tide them over until transplantation or to allow regeneration of their own liver to occur. Bioartificial liver (BAL) support systems are promising, as these systems rely on detoxification and synthetic function of liver cells.3 Over the last 2 decades, many BAL systems have been devised, of which only 8 systems have found application in a clinical setting.4

Many liver cell types have been used for preclinical and clinical application of BAL devices. The choice for the optimal cell source for BAL support devices is, however, still a matter of debate as the ideal set of criteria cannot be combined in one single cell type.4–10 Particularly, the availability, degree of liver specific function, and safety aspects are ongoing issues.

Primary hepatocytes, either allogeneic or xenogeneic, have an excellent function and are therefore favorite cell sources for BAL application. However, mature porcine hepatocytes (MPHs) are not attractive for clinical application because of risks related to xenotransplantation.11 Mature human hepatocytes (MHHs), on the other hand, are scarce, as their sources are limited to either discarded donor livers or small parts obtained during liver resections.12, 13 Human hepatic cell lines have the advantage of an infinite proliferation capacity and can potentially serve as a stable cell source. However, the lack of sufficient liver-specific functions limits their use for clinical BAL application. Alternatively, fetal human hepatocytes (FHHs) can be considered as an attractive human liver cell source. These cells have 2 advantages: 1) a much higher proliferation capacity than MHHs; and 2) increased liver-specific functions, in some respects higher than human liver cell lines,14 showing even further increase with aging of the fetus.15

So far, the 2 most frequently used cell sources for BAL application, i.e., MHHs and MPHs, have not been analyzed in a clinically-applied BAL system in a single study. Furthermore, FHHs have not been investigated in the setting of BAL using a 3-dimensional perfused culture system and subsequently compared with MHHs and MPHs. The aim of this study, therefore, was to compare hepatic functionality and metabolism of 3 types of primary hepatocytes in a for-laboratory-testing, downscaled version of the Academic Medical Center (AMC)-BAL.16–19 The 3 cell types used were as follows: 1) MPHs, as this cell type has been widely used for preclinical and clinical BAL application; 2) MHHs isolated from liver resection material, as this cell type displays highest physiological function and therefore is considered to be the gold standard; and 3) FHHs isolated from fetuses at the highest, yet acceptable age (23-24 weeks). Function of these 3 cell types was assessed for up to 7 days and the general tissue architecture in full transverse sections of the laboratory-scale AMC-BAL bioreactor was examined after termination of the culture period.


AMC, Academic Medical Center; BAL, bioartificial liver; MPH, mature porcine hepatocytes; MHH, mature human hepatocytes; HHCL, human hepatic cell lines; FHH, fetal human hepatocytes; LDH, lactate dehydrogenase; CYP, cytochrome p450.


All procedures were conducted in accordance with the institutional guidelines, either by the Animal Ethical Committee or Medical Ethical Committee of the Academic Medical Center and the Cardarelli Hospital. Human material was collected after written consent and used when serological tests were negative for human immunodeficiency virus, hepatitis B virus, and hepatitis C virus.

Isolation Procedures of Primary Hepatocytes

MPH isolation

MPHs were isolated from livers of female pigs (20-24 kg) by a 2-step collagenase perfusion technique according to a modified protocol of Seglen20 as previously described.21 MPHs were suspended in ice-cold William's E culture medium, consisting of William's E medium supplemented with 10% (vol/vol) heat-inactivated fetal bovine serum (BioWhittaker, Verviers, Belgium), 2 mmol/L glutamine, 1 μmol/L dexamethasone, 20 mU/mL insulin, 2 mmol/L ornithine, 100 μg/mL streptomycin, 100 U/mL penicillin, and 0.25 μg/mL fungizone. Total yield of isolated hepatocytes was estimated by determination of the cell pellet volume after 3 centrifugations at 50g for 3 minutes each.21 Viability was determined by trypan blue exclusion test using a Bürker Bright line cytometer (Optik Labor, Bad. Homburg, Germany) (Table 1).

Table 1. Specification of Used Liver Material, Isolation Outcomes, and Amount of Hepatocytes Loaded in the Bioreactor
SGenderAge (yr)EtiologyTT (hour)GA (weeks)Weight (gm)*Yield (109 vH)Viability (%)Loaded (109 vH)
  • Abbreviations: S, hepatocyte source; M, male; F, female; y, years; TT, transport time of the liver; GA, gestational age; hr, hour; vH, viable hepatocytes; MHHs, mature human hepatocytes; FHHs, human fetal hepatocytes; MPHs, mature porcine hepatocytes; nd, not determined.

  • *

    Weight refers to the liver resection sample of MHHs and to the whole fetal liver for FHHs.

 1M78Metastasis colon carcinoma461.1870.65
 2F40Bile duct papilloma231.5931.0
 5F36Hepatocellular adenoma451.9971.0
 1nd Bilateral polycystic kidneys + multiple brain cysts192320.40.19940.10
 2nd Tetralogy of Fallot142420.30.20810.14
 3nd Mitral valve atresia172426.50.17530.15
F nd14.5 ± 3.797.3 ± 1.41.0 ± 0.0

MHH isolation

Human liver tissue was collected from patients who underwent partial hepatectomy for various reasons (Table 1). Hepatocytes were isolated by a modified 2-step collagenase perfusion technique as described by Hoekstra et al.22 The cells were suspended in ice-cold William's E medium (as MPH isolation), and their quantity and viability were determined by trypan blue exclusion test, using a Bürker Bright line cytometer (Optik Labor) (Table 1).

FHH isolation

Human fetal livers were obtained from abortions of 23-24-week-old fetuses suffering from congenital disorders irreconcilable with life (Table 1). One fetus died in utero 30 minutes before the partus, whereas 2 fetuses died shortly after parturition. After hepatectomy, fetal livers were preserved on Celsior at 4°C and transported under sterile conditions from Naples (Italy) to Amsterdam (The Netherlands). Fetal livers were washed with Hank's Balanced Salt Solution (HBSS; BioWhittaker) to remove Celsior, and weighed. All extrahepatic tissue was dissected. Fetal livers were then cut into small pieces and subsequently digested in 50 mL 0.03% collagenase-P (Roche, Mannheim, Germany)-HBSS solution for 15 minutes at 37°C. The first cell suspension was sieved through a surgical gauze, whereas the residual fetal tissue was subjected to a second digestion by collagenase for 15 minutes. After sieving a second time, both cell suspensions were combined and centrifuged at 50g for 3 minutes at 4°C and collected in 25-mL ice-cold DMEM culture medium (Dulbecco's modified Eagle's medium; BioWhittaker) containing 10% heat-inactivated fetal bovine serum (HI-FBS; BioWhittaker), 2 mmol/L L-glutamine (BioWhittaker), 1 μmol/L dexamethasone (Centrafarm, Etten-Leur, The Netherlands), 10 μg/mL insulin, 5.5 μg/mL transferrin, 6.7 ng/mL selenium-X (ITS mix; Invitrogen, Grand Island, NY), 100 U/mL penicillin, and 100 μg/mL streptomycin (penicillin/streptomycin mix, BioWhittaker). The quantity of cells and their viability was determined by trypan blue exclusion test using a Bürker Bright line cytometer (Optik Labor) (Table 1).


We used the laboratory-scale AMC-BAL bioreactor, which is a 10× downscaled bioreactor of the second generation AMC-BAL with an internal volume of 55 mL.21 The general configuration of the bioreactor, as described in detail by Flendrig et al.,16, 17 consists of a polycarbonate housing containing a 3-dimensional nonwoven hydrophilic polyester matrix circularly wound around a polycarbonate core. Between the matrix layers, hydrophobic polypropylene gas capillaries are situated in a parallel fashion in which the ends are embedded in polyurethane resin and fitted with gas inlet and outlet caps.

Hepatocyte Culture

Hepatocytes were injected under sterile conditions into the bioreactor through 3 different loading ports using a 60-mL syringe containing a final volume of 50 mL culture medium for all cell types. Bioreactors were then placed in an incubator at 37°C and oxygenated with sterile 95% air and 5% CO2 at a flow rate of 150 mL/minutes during the whole experiment. To ensure optimal cell attachment and an even cell distribution, bioreactors charged with MHHs and MPHs were rotated 340° (back and forth) on the longitudinal plane and through the gravitational field at 1 revolution/minutes for 2 hours, whereas bioreactors with FHHs were rotated for 4 hours. After this attachment period, dead and unattached cells were removed by flushing 100 mL of fresh William's E culture medium through the bioreactor at 15 mL/minutes. The bioreactors were then continuously perfused overnight with 150 mL recirculating William's E culture medium at 15 mL/minute.

Hepatocyte Function Tests

Each test sequence consisted of an oxygen consumption test followed by a function test, as described previously.21 Briefly, oxygen consumption was determined by measuring the decrease in oxygen tension during the first 15 minutes after closure of the oxygen supply to the bioreactor. A function test was performed by flushing the bioreactor with 100 mL of test medium composed of William's E culture medium supplemented with 500 μg/mL lidocaine, 2 mmol/L L-lactate with either 5 mmol/L NH4Cl for MPH and MHH bioreactors or 0.5 mmol/L NH4Cl for FHH bioreactors, followed by recirculation of 100 mL test medium for 2 hours at 37°C. Samples were taken at 30, 60, 90, and 120 minutes, respectively, and subsequently analyzed for concentrations of ammonia, urea, lidocaine, porcine albumin, glucose, and lactate as well as activities of aspartate aminotransferase and lactate dehydrogenase (LDH), as described previously.21 Human albumin was determined by enzyme-linked immunosorbent assay using goat-anti-human serum albumin antibody (ab8940, 1:100; Abcam, Cambridge, United Kingdom) and horseradish peroxidase conjugated rabbit-anti-human serum albumin (ab7394, 1:5,000; Abcam). Ammonia and lidocaine elimination, urea and albumin production capacity, aspartate aminotransferase and LDH release, glucose and lactate consumption and/or production rates were determined by calculating the changes in concentration in test medium per hour per billion cells loaded in the bioreactor.

MPH bioreactors (n = 15) were tested at day 1 (n = 15), 3 (n = 11), 5 (n = 10), and 7 (n = 3). MHH (n = 5) and FHH (n = 3) bioreactors were all tested at day 1, 3, 5, and 7, respectively.

Bioreactor Tissue Preparation for Histological Analysis

Bioreactors were fixed in 10% formalin after the final function test and stored at 4°C. Complete transverse 8 μm sections of the laboratory-scale bioreactor (ØID 22 mm) were obtained after embedding the whole bioreactor in paraffin as follows. Fixed bioreactors were manually flushed with an increasing ethanol series, i.e., 100 mL per flush at 37°C; 1× 50% (vol/vol), 1× 70% (vol/vol), and 4× with 100% ethanol (vol/vol) (Merck). After dehydration, bioreactors were flushed 3 times with 100 mL xylene (Merck, Darmstadt, Germany) at 37°C and followed by a single flush of 100 mL paraffin (Variwax; Klinipath, Duiven, The Netherlands) at 62°C. After paraffin embedding, bioreactor housings were removed by milling, leaving an intact roll of the paraffinized matrix and capillaries. This roll was then transversely sliced into 1-cm discs, of which each disc was inserted in stainless steel cylinders (ØID 22 mm, 1 cm wide) with openings at both sides. Each cylinder was then placed in a 25-mL beaker containing prewarmed paraffin and incubated for 12 hours under vacuum at 62°C to remove all air. Finally, and after cooling, each bioreactor disc was embedded in a mold and made ready for sectioning.

All sections were stained with hematoxylin and eosin to evaluate tissue architecture and organization.

Statistical Analysis

Statistical analysis was performed by using SPSS 12.0.1 for Windows software (SPSS; Chicago, IL). Results are reported as means ± standard error. Repeated measurement analysis of variance rank tests were used to compare the 3 cell types over the 7-day culture period, and to compare differences per day within 1 cell type. Significance was reached if P < 0.05. Prism version 4.0 (GraphPad Prism, San Diego, CA) was used for graphical presentation of the data.


Cell Sources

In total, 23 bioreactors were loaded with 3 types of primary hepatocytes as follows: 1) 15 bioreactors with 1.0 × 109 viable MPHs; 2) 5 bioreactors with 0.65 × 109 to 1.0 × 109 viable MHHs (average 0.90 ± 0.12 × 109); and 3) 3 bioreactors with 0.10 × 109 to 0.15 × 109 viable FHHs (average 0.13 ± 0.02 × 109). The amount of loaded hepatocytes varied, depending on their availability. Table 1 shows a specification of the original liver material that was used in this study, including the results of the isolation procedures.

Hepatocyte-Specific Functions

We determined the ammonia and lidocaine elimination as well as urea and albumin production to assess the liver-specific function of MPHs, MHHs, and FHHs in the bioreactor (Fig. 1). Mature hepatocytes from both pigs and humans eliminated ammonia with a mean over 7 days of 89.4 ± 6.5 μmol/hour/109 cells for MPHs and 185.6 ± 19.7 μmol/hour/109 cells for MHHs, indicating a 2-fold higher ammonia elimination capacity of MHHs (P < 0.05) (Fig. 1A). In contrast, FHHs increasingly produced ammonia with a mean of 63.9 ± 26.0 μmol/hour/109 cells over 7 days (P < 0.05 MPHs and MHHs). Despite this incapacity for ammonia elimination, FHHs produced urea at a rate comparable with that of MPHs, i.e., 37.6 ± 10.2 μmol/hour/109 cells and 33.0 ± 8.8 μmol/hour/109 cells over 7 days, respectively (Fig. 1B). MHHs showed a 3- to 4-fold higher urea production, i.e., 124.0 ± 26.7 μmol/hour/109 cells, as compared to MPHs and FHHs (P < 0.05). Urea production of MPHs was significantly higher at day 1 as compared to day 5 and 7, and likewise, MHHs produced significantly less urea at day 7 as compared to day 1 and 3. If ammonia elimination is only determined by urea synthesis, 2 moles of ammonia are converted into 1 mole of urea. However, the ammonia-urea ratio was rather variable: it increased in time for MPHs as well as MHHs. This tendency, although not significant, was stronger for MPHs as compared to MHHs with a mean ratio of 1.8 ± 0.5 at day 1 and 5.5 ± 2.7 at day 7 for MPHs and for MHHs, a mean ratio of 1.6 ± 0.4 at day 1 and 2.7 ± 1.4 at day 7.

Figure 1.

Hepatocyte-specific functions of MPH (lined bars), MHH (black bars), and FHH (white bars) cultured in AMC-BAL bioreactors over a 7-day period. (A) Ammonia elimination, (B) urea production, and (C) lidocaine elimination, expressed in μmol/hour/109 viable hepatocytes (mean ± standard error), and (D) albumin production expressed in μg/hour/109 viable hepatocytes (mean ± standard error). Significance between cell types: *P < 0.05 over 7 days; ns, not significant. Significance within 1 cell type: ≈P < 0.05 vs. day 1; ∞P < 0.05 vs. day 3; #P < 0.05 vs. days 5 and 7; $P < 0.05 vs. day 1 and 3; &P < 0.05 vs. days 3, 5, and 7.

Lidocaine elimination, as a parameter for cytochrome p450 3A4/1A2 (human) and 3A29 (pig) activity, was significantly different for all 3 cell types over the 7-day culture period (Fig. 1C). Lidocaine elimination of MPHs was 1.9-fold higher than of MHHs. Lidocaine elimination of FHHs was 3.5-fold higher than of MPHs and 6.6-fold higher than of MHHs, with an average over 7 days of 199.3 ± 40.4 μmol/hour/109 cells for FHHs, 57.5 ± 7.8 μmol/hour/109 cells for MPHs, and 30.1 ± 5.7 μmol/hour/109 cells for MHHs. Lidocaine elimination of MPHs at day 1 was significantly higher than at days 3, 5, and 7, whereas no daily differences were observed for MHHs and FHHs. No changes in lidocaine concentration were observed during a function test in an empty bioreactor.

Albumin production was not significantly different for all 3 cell types over the 7-day culture period (Fig. 1D). A tendency toward a higher albumin production for FHH > MHH > MPH was observed, with an average over 7 days of 126.8 ± 80.5 μg/hour/109 cells, 109.9 ± 36.6 μg/hour/109 cells, and 79.5 ± 26.9 μg/hour/109 cells, respectively. MHH and MPH tended to increase over time, whereas FHH remained stable during the culture period.

These results show that most liver-specific functions differ considerably between MPHs, MHHs, and FHHs when cultured inside the laboratory-scale AMC-BAL.

General Metabolic Function and Activity

We determined the production or consumption of lactate and glucose as parameters for carbohydrate metabolism (Fig. 2). After an initial phase of glucose production, all cell types increasingly consumed glucose and produced lactate (Fig. 2A and B). However, distinct differences between all 3 cell types were observed in the overall conversion rates of glucose and lactate (P < 0.05). The overall conversion rate of glucose, calculated by the slope over 7 days and expressed in μmol/hour/109 cells per day, was 69.3 for MPHs (r2 = 0.996), 13.0 for MHHs (r2 = 0.996), and 102.9 for FHHs (r2 = 0.81). The same tendency was observed for the overall conversion rate of lactate: 115.6 for MPHs (r2 = 0.99), 6.4 for MHHs (r2 = 0.46), and 234.8 for FHHs (r2 = 0.86). These observations indicate that MHHs were less glycolytic inside the bioreactor than MPHs and FHHs.

Figure 2.

General metabolic function and activity of MPH (lined bars), MHH (black bars), and FHH (white bars) over 7 days in the AMC-BAL bioreactors. (A) Glucose consumption, (B) lactate production, and (C) oxygen consumption. Expressed in μmol/hour/109 viable hepatocytes (mean ± standard error). Significance between cell types: *P < 0.05 over 7 days. Significance within 1 cell type: (A) MPH all days, not day 5 vs. day 7; MHH and FHH all days, not day 1 vs. day 3, and day 5 vs. day 7; (B) MHH not significant between all days; MPH all days, not day 5 vs day 7; FHH all days, not day 1 vs. day 3; (C) **P < 0.05 vs. day 3.

Furthermore, we used oxygen consumption as a parameter for general metabolic activity of the hepatocytes inside the bioreactor (Fig. 2C). All cell types increasingly consumed oxygen throughout the culture period and no significant differences were found between the 3 cell types over a period of 7 days. However, oxygen consumption of MPHs increased with 4.9 μmol/hour/109 cells per day (r2 = 0.93) and, accordingly, differed significantly per day. In MHHs, a nonsignificant daily increase in oxygen consumption of 1.4 μmol/hour/109 cells per day (r2 = 0.73) was found, and in FHHs, the daily increase in oxygen consumption was 4.4 μmol/hour/109 per day (r2 = 0.47) of which day 7 was significantly higher than day 3.

Cellular Integrity

Aspartate aminotransferase and LDH releases were determined as a parameter for hepatocellular damage (Fig. 3).23 Aspartate aminotransferase release did not differ significantly between the 3 cell types over 7 days (Fig. 3A). However, within each cell type, aspartate aminotransferase release at day 1 was significantly higher when compared to days 3, 5, and 7, respectively, and is related to the washout of unattached cells and debris after charging the bioreactor at day 0. This trend, i.e., high at day 1 and low at days 3, 5, and 7, was also observed for the LDH release in MHHs, but not in MPHs and FHHs (P < 0.05) (Fig 3B). MPHs showed a significant increase of LDH release from day 3 to 7. LDH release from FHHs remained constant at a relatively high mean level of 2.7 ± 0.7 U/hour/109 cells over 7 days.

Figure 3.

Aspartate aminotransferase and LDH release of MPH (lined bars), MHH (black bars), and FHH (white bars) over 7 days in the AMC-BAL bioreactors. (A) AST release, (B) LDH release. (A-B) are expressed in U/hour/109 viable hepatocytes (mean ± standard error). Between cell types: *P < 0.05 over 7 days; ns, not significant. Significance within 1 cell type: #P < 0.05 days 3, 5, and 7; $P < 0.05 vs. days 1 and 7.

Histological Analysis

Transverse sections of MPH, MHH, and FHH bioreactors are shown in Figure 4. The analyzed MPH and MHH bioreactors were loaded with 1.0 × 109 cells, whereas the FHH bioreactor was loaded with 0.15 × 109 cells. The sections were derived from the same position of the bioreactors; i.e., 1 cm from the central loading port, in an area with a high density of cells. In the MPH and MHH sections (Fig. 4A-F), tissue-like structures were observed around gas capillaries, attached to and in between the nonwoven matrix layers of the bioreactor. FHHs were predominantly located inside the matrix layer (Fig. 4H and I). Single cells or small cell aggregates were mainly observed inside the matrix layers of bioreactors loaded with all 3 cell types. In general, tissue-like structures were preferentially located next to or in close vicinity of gas capillaries, or in areas through which culture medium primarily flows; i.e., between the matrix layers. Cells forming these tissue-like structures consisted of cells showing hepatocyte morphology. At the edges of these structures, directed toward the flow stream of the medium in the extracapillary/matrix area, flattened cells were aligned. Inside large tissue-like cell clusters, viability was low, as pycnotic or anuclear cells were observed. These findings were more pronounced in MPH than in MHH bioreactors, but were not observed in FHH bioreactors. In general, a layer of approximately 150 μm of viable cells could be observed for the tissue-like cell structures surrounding capillaries and of 70 μm for structures directly facing the flow stream of the culture medium. At high magnification, deposition of extracellular matrix was observed for all 3 cell types. Multiple transverse sections at different bioreactor positions revealed a more even longitudinal cell distribution of MHHs and FHHs throughout the bioreactors as compared to MPHs, which were mainly located at the 3 sites where cell loading had taken place (data not shown).

Figure 4.

Microscopy of 3 primary liver cell types cultured for 7 days in the AMC-BAL (hematoxylin and eosin staining): MPH (A-C), MHH (D,E), and FHH (G-I). (A,D,G) Full transverse sections of the laboratory scale bioreactor showing homogeneous distribution of large tissue-like structures for MPHs and MHHs (bar 2.5 mm). (B,E,H) These tissue-like structures are preferentially located next to or in close vicinity of gas capillaries, or in areas through which culture medium primarily flows (indicated by asterisks), i.e., between the matrix layers (bar 250 μm; black triangles point toward gas capillaries). At higher magnification, some pycnotic or anuclear cells are present, whereas the majority of cells are viable. Flattened cells with endothelial morphology are present at the edges of the tissue-like structures (bar 50 μm; open triangles point toward fibers of the matrix).


This study compared 3 types of freshly isolated primary hepatocytes cultured inside a bioreactor consisting of a 3-dimensional perfused culture system; i.e., FHHs, MHHs, and MPHs. Although MHHs and MPHs are frequently used cell types for BAL-systems, no comparison has been made between these 2 primary cell types, nor with FHH, in a 3-dimensional BAL. In the field of pharmacology and toxicology, hepatocytes of various species and maturation are used for in vitro predictive models of newly developed drugs.24 Unfortunately, these studies mainly use hepatocytes in a 2-dimensional configuration without medium flow. Direct extrapolation to BAL devices is therefore difficult.

The results in the current study show large differences in liver-specific and metabolic functions between MPH, MHH, and FHH. However, oxygen consumption, as a measure of cell volume and general metabolic activity, did not differ between the 3 cell types during the whole culture period (Fig. 1C).25 This suggests that the measured differences in liver-specific and metabolic functions are not attributable to the initial state of the cells (either fresh or Celsior preserved), the isolation method used (central or manual perfusion), or inadequate calculation of the number of cells charged in the bioreactors. Moreover, the high reproducibility of functional results within the groups shows that the variation in viability of the isolates, particularly of the MHHs and FHHs, is acceptable. The observed differences may be explained by species differences, although this cannot hold true for the comparison between FHHs and MHHs.

The values for ammonia elimination and urea production of MPH are in accordance with previous studies from our laboratory (Fig. 1A and B).19, 21, 26 The higher values observed for MHH should be attributed to a higher and more efficient urea cycle and glutamine synthetase activity. We point out, however, that urea can be produced by conversion of arginine into ornithine by arginase I although the urea cycle may not be intact, and that ammonia can be eliminated exclusively by glutamine synthetase. As a result, the elimination of ammonia into urea does not necessarily follow the molar ratio of 2, as was observed in this study for MPH and MHH. FHH, however, produced ammonia. Production of ammonia is a phenomenon of hepatocytes observed in different culture systems of human hepatic cell lines (HepZ, HepG2, C3A, HuH-7, and PCL/PRF/5)27–29 and for mature rat, human, and porcine hepatocytes under different culture conditions.28, 30 Recently, we also observed ammonia production in primary FHHs cultured in monolayer.14 In this study, ureagenesis was observed in FHHs at a level comparable to that of MPHs, as has been shown previously.31 The different relations between ammonia and urea metabolism, i.e., production of ammonia and urea by FHHs and ammonia elimination and urea production by MHHs and MPHs, may be explained by the ontogenic activity of the kidney-type and liver-type glutaminases in relation to the urea cycle.32, 33 The kidney-type glutaminase is expressed most in the fetal liver, and not in the postnatal liver. As a consequence, fetal liver releases free ammonia from glutamine catabolism by the kidney-type glutaminase, independent of the production of urea from ammonia via the urea cycle. In mature liver, however, only liver-type glutaminase is expressed, which produces ammonia that is directly used for the synthesis of urea, since glutaminase-derived ammonia is directly channeled to carbamoyl-phosphate synthetase and does not escape the mitochondria.34, 35 This explanation for the production of ammonia by FHH should be considered as speculative, since we did not investigate the kidney, as well as liver-type glutaminase at the protein level in this study.

Porcine hepatocytes are considered a relatively good experimental human model for cytochrome p450 (CYP) 3A drug metabolism. Caution, however, is required for direct extrapolation of the activity of CYP isoforms from pigs to humans, because the activity differs per substrate.36–39 In this study, lidocaine elimination of MPHs, mediated by the CYP3A29 isoform, was 1.9-fold higher than of MHHs (Fig. 1C). Lidocaine elimination of FHHs, on the other hand, was 6.6 times higher than MHH. The fetal isoform of human CYP3A4 is CYP3A7 and this accounts for up to 50% of total fetal hepatic CYP and nearly all CYP3A enzymes.40 Metabolic activity of CYP3A7 is significantly lower than CYP3A4, depending on the concentration and metabolite used.41 Interestingly, however, total CYP3A7 content of FHH of 94-168 days gestational age is approximately 310 pmol/mg microsomal protein, whereas total CYP3A4 content in MHH is approximately 60 pmol/mg microsomal protein.40 This can explain the higher lidocaine elimination of FHH observed in this study. We emphasize at this point, however, that no data are available on the effect of a 3-dimensional high cell density perfused culture system on CYP3A4/1A2, 3A29, and 3A7 activity of MPH, MHH, and FHH, respectively, using lidocaine.

In contrast to the urea production and ammonia and lidocaine elimination, albumin production was not significantly different between all 3 types of hepatocytes. This was also observed by Jasmund et al.42 in a 2-dimensional sandwich culture system with MPH and MHH under different culture conditions. A stable or an increased albumin production of MPH and MHH during the first 7 days of culture has been shown in monolayer,42–44 as well as several bioreactor systems.21, 45–48 FHH of 23-24-week-olds were able to produce albumin at a rate of MPH and MHH. This was also observed for FHH cultured in monolayer, secreting albumin at 4-21 μg/mL/1.8-2.5 × 106 FFH,49–51 which resembles the albumin production by MPH and MHH in monolayer.43

The increase in glucose consumption, lactate production, oxygen consumption (Fig. 2), and LDH release (Fig. 3B) of MPH are in accordance with previous studies.19, 21, 26 These observations may imply an increase in anaerobic glycolysis in combination with mitochondrial uncoupling of the oxidative phosphorylation.52, 53 An increase in lactate production, LDH release, and oxygen consumption can also, at least in part, be a reflection of cell growth.25 In parallel, nonparenchymal liver cells produce more lactate and grow faster than parenchymal liver cells as observed in monolayer culture.54 Fetal livers contain more nonparenchymal cells than mature livers because of their hematopoietic function. Therefore, the higher glucose consumption, lactate production, oxygen consumption rate, and LDH release of FHHs can either be explained by a higher content of nonparenchymal cells or a higher in vitro proliferation capacity of FHH itself. Dedifferentiation does not seem to play a major role, since the hepatic functions remain relatively stable. MHHs, on the other hand, are more stable in the general metabolic functions during the 7-day culture than MPHs and FHHs. This can be explained by the composition of human and porcine liver cell isolates. Since MPH isolates contain more cellular aggregates and form spheroids easier than MHH isolates,47, 55, 56 high dense cellular but also anoxic areas can be formed, as was observed by histological examination. In contrast, MHH isolates exist of mainly single cells, which were distributed evenly throughout the bioreactor.

In summary, at present MPH and MHH are generally considered as the most acceptable cell sources for BAL devices. FHH have, to date, not been demonstrated to be a relevant cell source for BAL application. The higher proliferation capacity of FHHs as compared to MHHs and MPHs, and the higher liver-specific function compared to human hepatic cell lines have encouraged us to test this cell type in the AMC-BAL. We observed large differences between the 3 cell types when investigated in the 3-dimensional perfused AMC-BAL system. Our results indicate that FHHs, despite their higher proliferative capacity and CYP 3A activity, are not an ideal cell type for BAL application, since FHHs produce ammonia and display high metabolic instability. FHH in a bioreactor may, on the other hand, be considered as a favorable cell type for pharmacological and toxicological studies.57, 58 MHHs outperformed MPHs and FHHs in their ammonia detoxification, urea production, and metabolic stability. From this study, we conclude that MHHs are the preferred cell source for clinical use in a BAL. On the other hand, MPHs can be regarded as the second best cell type because of their unrestricted availability. Since bioreactors loaded with MPH are less stable than MHH bioreactors, these bioreactors should be preferably used within 1 or 2 days after loading in the clinical setting.


We thank G. Huyzer and A. Maas of the Surgical Laboratory for their assistance in porcine liver harvest surgery. We thank Dr. Fabio Sirimarco (Department of Gynaecology and Obstetrics) and Dr. Giuseppe Nazzaro (Department of Obstetric Pathology) for their help in collecting the fetal livers, and Dr. Antonio Faiella and Daniele Morelli for their help in the logistics concerning the fetal livers.