Potential conflict of interest: Nothing to report.
The culture of primary hepatocytes as spheroids creates an efficient three-dimensional tissue construct for hepatic studies in vitro. Spheroids possess structural polarity and functional bile canaliculi with normal differentiated function. Thus, hepatocyte spheroids have been proposed as the cell source in a variety of diagnostic, discovery, and therapeutic applications, such as a bioartificial liver. Using a novel rocking technique to induce spheroid formation, kinetics of spheroid formation, cell-cell adhesion, gene expression, and biochemical activities of rat hepatocyte spheroids were tested over 14 days of culture. Evidence was provided that the formation of spheroids occurred faster and with fewer nonadherent hepatocytes in rocked suspension culture compared to a traditional rotational system. Hepatocyte spheroids in rocked culture showed stable expression of more than 80% of 242 liver-related genes including those of albumin synthesis, urea cycle, phase I and II metabolic enzymes, and clotting factors. Biochemical activity of rocked spheroid hepatocytes was superior to monolayer culture of hepatocytes on tissue culture plastic and collagen. Conclusion: Spheroid formation by rocker technique was more rapid and more efficient than by rotational technique. Rocker-formed spheroids appear suitable for application in a bioartificial liver or as an in vitro liver tissue construct. (HEPATOLOGY 2009.)
Novel systems are needed to facilitate short-term and long-term culture of hepatocytes for diagnostic, discovery, and therapeutic applications.1 Traditional monolayer culture of primary hepatocytes on tissue culture plastic is problematic and has been associated with a rapid loss of differentiated function.2 Although single-layer and double-layer surfaces of collagen or other biomatrix materials are associated with improved differentiated functions in vitro,3 biological surfaces can pose manufacturing hurdles and do not support high cell density culture of primary hepatocytes (exceeding 1 × 107 cell/mL). In contrast, spheroids, which are nonadherent multicell aggregates of greater than 40 μm diameter, provide a three-dimensional tissue construct which form spontaneously and allow suspension culture of primary hepatocytes at high cell density under oxygenated bioreactor conditions.4 Spheroid formation allows recapitulation of the cuboidal geometry of primary hepatocytes with relatively stable long-term differentiated function.5, 6 Reports of structural polarity and bile canaliculi formation by primary rat hepatocytes in spheroid aggregates provide further evidence that hepatic spheroids mimic the hepatocellular microanatomy of the liver.7
Original observations of tissue-like aggregate formation from isolated cells was reported by Moscona in 1961, using fetal liver cells and a rotational technique.8 The descriptive term “spheroid” was coined years later by Landry in 1985 when multicellular aggregates were formed from isolated rat hepatocytes after 3-5 days of culture on nonadherent plastic surfaces.9 Later, spheroids were observed forming from primary hepatocytes after 24 hours of rotational culture with 56% of the hepatocyte inoculum forming multicellular aggregates of greater than 40 μm in diameter.10 More recently, we reported preliminary observations that hepatocytes formed spheroids when suspended in rocked (oscillatory) motion.11 Rocking promoted mixing, oxygenation, and increased frequency of collisions between freshly isolated hepatocytes, which in turn accelerated their aggregation into clusters and the formation of spheroids.
Our preliminary report of spheroid formation under rocked conditions was limited to a 5-day culture interval using porcine hepatocytes.11 The goals of this study were to compare the spheroid formation kinetics of rat hepatocytes in rocked versus rotational culture and to characterize the gene expression and biochemical activity of rocker-formed rat hepatocyte spheroids over extended (14 days) duration. Spheroid formation was measured by direct observation and automated Coulter counter. In order to expand the evaluation of gene expression by hepatocyte spheroids, we used a custom microarray consisting of 242 liver-related genes in a wide array of cellular functions: cell adhesion, cell respiration, phase I and II metabolism, ureagenesis, glycolysis, stress response, and assorted transcription factors. Functionality of spheroids was compared to monolayer controls (tissue culture plastic [TCP]; collagen monolayer [CM]) using clinically relevant liver-specific biochemical assays: albumin production, ureagenesis, and cytochrome P450 activities. All studies were conducted using rat hepatocytes inoculated at an intermediate cell density (1 × 106 cell/mL) to allow comparisons of monolayer and spheroid geometries formed by static, rocked, or rotational technique. Our results indicate that the rocked technique is a simple yet rapid and highly efficient means of forming three-dimensional, suspension-stable, hepatic tissue constructs.
Animals were provided ad libitum access to water and standard chow. All animal procedures were performed under the guidelines set forth by the Mayo Foundation Animal Care and Use Committee and are in accordance with those set forth by the National Institutes of Health. All compounds were obtained from Sigma (St. Louis, MO) unless noted otherwise.
Rat Hepatocyte Isolation.
Hepatocytes were isolated from 250-300 g male Sprague Dawley rats (Harlan, Indianapolis, IN) by a two-step perfusion method as previously described.12 All harvests yielded hepatocytes with viability exceeding 90% by trypan-blue dye exclusion.
Spheroid and Monolayer Culture.
Freshly isolated hepatocytes were suspended in culture medium [Williams-E medium with 10% fetal bovine serum (Mediatech, Herndon, VA), 10 U/mL penicillin G, 100 μg/mL streptomycin, 10 μg/mL insulin, 5.5 μg/mL transferrin, 5 ng/mL sodium selenite] at 1 × 106 cells/mL. Suspended hepatocytes were inoculated into four different culture vessels: spheroid rocker dishes (10 cm × 8 cm × 2 cm); spheroid rotational flasks (6 cm diameter); monolayer plates of tissue culture plastic (60 mm; Corning, Corning, NY); or monolayer plates of type I collagen-coated plastic (60 mm Biocoat; BD Labware, Franklin Lakes, NJ). Cell density (1 × 106cells/mL) and plating density (1.8 × 105cells/cm2) were standardized by inoculating cell suspensions of 20 mL per spheroid vessel and 5 mL per monolayer plate. Spheroid dishes were rocked continuously at 0.25 Hz whereas spheroid flasks were rotated on an orbital spinner. Monolayer plates were placed on a stable rack. All cultures were incubated at 37°C and 5% CO2.
Medium was first changed at 24 hours and subsequently at weekly intervals. Dexamethasone (100 ng/mL) was added on day 1 and weekly thereafter. One day prior to sampling, cultures received 10 μM β-naphthoflavone (βNF), 8 μM diazepam, and 5% vol/vol heavy deuterium-enriched ammonia gas (2.23 mM; Cambridge Isotope Laboratories Inc., Andover, MA) to test the hepatic clearance ability of these compounds. βNF was omitted from noninduced cultures. Samples were stored at −80°C.
Morphometrics of Hepatocyte Spheroid Formation.
Rocked and rotational cultures were sampled (1 mL) from 4-48 hours. Spheroid diameter, spheroid number, and free hepatocyte number were measured by phase contrast microscopy (Axiocam; Carl Zeiss Inc., Thornwood, NY) and by automated technique using a Multisizer3 (560 μm aperture; Beckman Coulter, Fullerton, CA).
Hepatocytes were fixed in 4% paraformaldehyde, rinsed with phosphate-buffered saline, and prehybridized with 0.1% Triton and 1% goat serum. Cells were incubated in mouse anti-rat E-cadherin antibody (1:500, #610181; BD Biosciences, San Jose, CA) and secondary antibody: fluorescein-5-isothiocyanate–labeled goat anti-mouse immunoglobulin G (1:1000; Molecular Probes, Invitrogen, Carlsbad, CA). Immunofluorescence was imaged by confocal microscopy (LSM510; Carl Zeiss Inc.).
Spheroids were fixed in Trumps solution, sectioned, and imaged by transmission electron micrscopy in the Microscopy Core Facility (Mayo Clinic, Rochester, MN).
Spheroid samples (300 μL) were resuspended in a 50 μL mixture of acridine orange and ethidium bromide stain (FluoroQuench; One Lambda, Canoga Park, CA). Cells were imaged using fluorescent microscopy using fluorescein isothiocyanate filters and AttoArc HBO 100 W excitation. Fluorescence was quantified using ImageProPlus6.0 (Media Cybernetics, Silver Spring, MD).
Calcium Influence on Spheroid Formation.
Hepatocytes were cultured in medium with or without ethylene glycol tetraacetic acid (EGTA) supplementation, equimolar (31.4 mM) to calcium concentration. Hepatocytes were viewed by phase contrast microscopy after 48 hours of rocking. EGTA was also added after spheroid formation (48 hours) to assess the influence of calcium depletion on formed spheroids.
Extraction of DNA and RNA.
Samples of spheroids (5 mL) and hepatocytes scraped from plates were pelleted and aliquoted for DNA or RNA analysis. DNA quantification was performed using the DNeasy kit (Qiagen, Valencia, CA). Total RNA was extracted immediately and stored at −80°C in RNAlater (Qiagen, Valencia, CA) until quantification.
Custom Microarray Analysis.
A custom microarray of 242 liver-related genes with rat-specific oligonucleotide sequences was created and analyzed by the Advanced Genomic Technology Center (Mayo Clinic, Rochester, MN). Microarray analysis was performed on spheroid samples obtained on days 2, 7, and 14 and compared to fresh hepatocytes (day 0). Each sample not induced with βNF was compared to the baseline (day 0) sample, and the βNF-induced samples were compared to noninduced samples from the same day. Ten comparisons were performed for each gene sequence. Gene expression was reported as the log2 fold-change of signal intensity between compared samples.
Albumin Gene Expression and Production of Rat Albumin.
Rat albumin messenger RNA copy number was measured by a standardized quantitative competitive reverse transcription polymerase chain reaction (QC-RT-PCR) with capillary electrophoresis detection in the Mayo Immune Monitoring Lab Core Services as previously described.13 Concentration of rat albumin in culture media was determined by enzyme-linked immunosorbent assay (ELISA) kit (Bethyl Laboratories, Montgomery, TX) using a sheep anti-rat albumin antibody.
qPCR Analyses of Hepatocyte Gene Expression.
Expression of Cdh1, Nags, Ass, Asl, Otc, Arg1, Cps1, Cyp1a1, Cyo1a2, Aldh3a1, Nqo1, Gsta2, Chst10, Cyp2c, Cyp2a1, and Hmox1 genes was determined using real-time RT-PCR (qPCR). Briefly, 8 or 16 ng total RNA and 50 ng of primers were added to a 10 μL reaction containing Quantifast SYBR green RT-PCR master mix (Qiagen). The qPCR reactions were performed in 384-well Clear Optical reaction plates (Applied Biosystems), and detection was determined using an ABI 7900 HT Fast Real-Time PCR system (Applied Biosystems, Foster City, CA). Cycle threshold (CT) values were determined using Applied Biosystems SDS 2.0 software. Fold-changes in gene expression were calculating using the ΔΔCT method. Results were based on RNA obtained from four independent experiments.
Quantification of Ammonia and Isotopes of Urea.
Rates of ureagenesis were determined by the conversion of heavy ammonia (15ND3) to native and deuterium-enriched urea. Isotopes of urea were quantified with capillary gas chromatography/mass spectrometry as previously reported.14
Cytochrome P450 1A (CYP1A) activity was assessed by the conversion of 7-ethoxyresorufin to resorufin as previously described.15 Resorufin was detected at excitation and emission wavelengths of 530 and 590 nm, respectively, using a FL600 microplate fluorescence reader (Bio-Tek Instruments Inc., Winooski, VT).
CYP3A and CYP2C11 Activity.
CYP3A and CYP2C11 activities were determined from media samples supplemented with diazepam as previously described.11 Data acquisition was completed by Chrom Perfect software (Denville, NJ).
Two-way analysis of variance and Student t-test were used to assess statistical significance of differences in biochemical activity and gene expression between culture conditions and time points. A Mann-Whitney rank sum test was used to compare diazepam metabolism values. A nonparametric Spearman correlation was used to compare albumin QC-RT-PCR and albumin protein data. Results were expressed as mean values ± standard error of the mean.
Gene expression was compared using log2 transforms of the median signal intensity. Data was normalized by a semiparametric approach to correct for intensity-dependent effects.16 We used SAS Version 9 statistical software (SAS Institute Inc., Cary, NC) for these analyses.
Total Hepatocyte Number, Hepatocyte Viability, and Incorporation into Spheroids.
The kinetics of spheroid formation was compared under rocked and rotational conditions as summarized in Fig. 1, upper panel. Measurements of spheroid size, spheroid number, hepatocyte number, and percent incorporation of hepatocytes into spheroids were obtained at the time of inoculation through 48 hours. These measurements were made in triplicate. Aggregation of hepatocytes was observed under both rocked and rotational conditions by 4 hours; however, hepatocyte aggregates were larger and more numerous in rocked cultures at all time points (Fig. 1, lower panel). Well-formed spheroids were observed throughout the rocked spheroid cultures at 12 hours. By 24 hours of rocking, 85% of inoculated hepatocytes had incorporated into well-formed spheroids of greater than 40 μm diameter (Table 1). At 48 hours, only 13% of cell particles measured less than 40 μm diameter by Coulter counter (Fig. 2). The majority of rocker-formed aggregates (84%) ranged in diameter of 75-200 μm (Fig. 3; day 2). In contrast, only 58% of hepatocytes incorporated into spheroids under rotational conditions at 24 hours. Most unattached hepatocytes appeared dead. Rotational-formed spheroids were less spherical at earlier time points.
Table 1. Spheroid Formation After 24 Hours: Rotational Versus Rocked Method
Rotational (60 rpm)
Rocked (0.25 Hz)
P < 0.001 versus rotational culture (based on 1000 measurements per group, repeated in triplicate).
Spheroid number and cell number remained relatively stable after day 2 during the first week of rocked culture. In contrast, average spheroid size decreased from day 2 (125-150 μm) to day 7 (75-100 μm) consistent with formation of mature spheroids during the first week of rocked culture (Fig. 3). Spheroid formation provided a protective benefit, because more than 95% of hepatocytes present in spheroids were viable at each time point. A decline in spheroid number and cell number was observed during the second week of culture. Approximately 30% of hepatocytes inoculated into rocked dishes remained in spheroids on day 14. In contrast, less than 10% of hepatocytes inoculated into rotational flasks remained in spheroids on day 14. Therefore, spheroids formed by rocker technique were used in subsequent studies.
Expression of Liver-Related Genes in Spheroids by Custom Microarray.
We profiled gene expression of rat hepatocytes over 14 days of rocked spheroid culture. A custom microarray of 242 liver-related genes was developed for this task. We observed that the average expression of 85% (206) of the genes remained stable, defined as less than two-fold increase or 50% decrease relative to baseline, over 14 days in culture (Fig. 4). Expression increased two-fold or greater in 11 (5%) genes and decreased 50% or more in 25 (10%) genes. A complete listing of custom microarray data can be found at http://mayoresearch.mayo.edu/mayo/research/nyberg_lab/.
Hepatocyte Cell Adhesion.
The expression of several cell adhesion genes including E-cadherin (Cdh1), intracellular adhesion molecule 1 (Icam1), and fibronectin (Fn1) remained stable in rocked spheroid culture (Fig. 5). Expression of N-cadherin (Cdh2), selectin (Sele), laminin gamma 1 (Lamc1), and six other integrin-related genes (Itga1, Itgae, Itgam, Itgb1, Itgb3, and Itgb4) also remained stable. The pattern of expression of Cdh1 was validated by qPCR.
Because of the known importance of E-cadherin as an adhesive protein between epithelial cells, we studied the role of E-cadherin during spheroid formation by rocked technique. Cell samples were examined by immunohistochemical staining during initial hours of spheroid formation in culture medium and culture medium supplemented with EGTA to remove free calcium. We observed rapid spheroid formation and strong staining of E-cadherin on the surface of cells formed in the presence of free calcium (Fig. 6A -F). However, spheroids failed to form in medium supplemented with EGTA (Fig. 6G). Furthermore, formed spheroids quickly fragmented after addition of EGTA on day 2 (Fig. 6H,I). Most unattached hepatocytes appeared dead after 24 hours in calcium-free conditions, thus confirming the importance of calcium-dependent adhesion in the formation and viability of liver tissue constructs such as spheroids. These studies suggest an essential role for E-cadherin in the prevention of cell death after separation of hepatocytes from their extracellular matrix. This form of cell death, termed anoikis, has been observed after separation of other epithelial cell types from their surrounding matrix.17
Gene Expression and Albumin Production.
Custom gene microarray, QC-RT-PCR, and ELISA were used to assess gene expression and production of albumin under rocked spheroid and monolayer culture conditions. By custom microarray analysis, albumin gene expression in spheroid cultures remained at 86% of baseline on day 2 and declined to 57% on day 14. This pattern of albumin gene expression was confirmed by QC-RT-PCR (Fig. 7A). Copy number of the albumin gene was significantly higher in spheroid culture compared to TCP or CM at all time points (P < 0.05). Production of albumin protein was also significantly higher in spheroid culture compared to monolayer control cultures at all time points (Fig. 7B; P < 0.05). A correlation was suggested between gene copy number and production rate of albumin by nonparametric comparison (Spearman correlation coefficient = 0.56; P = 0.09).
Six genes of the urea cycle were included on the custom array (Fig. 8A): n-acetylglutamate synthase (Nags), arginino succynate synthase (Ass), arginino succynate lyase (Asl), arginase 1 (Arg1), carbamoyl-phosphate synthetase 1 (Cps1), and ornithine transcarbamolase (Otc). Cps1 and Otc encode for proteins which catalyze the initial steps of the urea cycle within the inner membrane of the mitochondria. Nags encodes for an inner mitochondrial protein. Arg, Ass, and Asl express cytoplasmic proteins of the urea cycle. The expression of three urea cycle genes (Ass, Asl, and Nags) remained stable during 14 days of rocked spheroid culture; Arg1 was up-regulated to 250% at day 2 and declined gradually to 89% of baseline on day 14; Cps1 and Otc showed a mean reduction of 55% and 36% of baseline, respectively. Expression of these six genes was also measured by qPCR. Patterns of expression of all six genes were similar by microarray and qPCR analysis. However, greater reductions in expression of Cps1 and Otc were determined by qPCR (8-fold and 25-fold decrease on day 14, respectively).
Rocker-formed spheroids demonstrated significant levels of ureagenesis at all time points based on conversion of heavy ammonia to heavy urea (Fig. 8B). Total ureagenesis was greater in rocked spheroid cultures than monolayer control conditions at all time points. The ratio of heavy urea to total urea produced by spheroid hepatocytes was also greater than monolayer controls at all time points. The decline in ureagenesis over time paralleled the decline in expression of Cps1 and Otc.
Phase I Metabolism.
Thirty-nine genes involved in phase I metabolism were assessed by microarray in rocked spheroid culture. These genes showed the greatest variability in expression of any group tested (Fig. 4). Average expression of 19 genes (49%) remained stable. One phase I gene (Cyp1a1 [Fig. 9]) was up-regulated over time (without addition of βNF) by both custom microarray and qPCR analysis, whereas the gene expression of Cyp1a2 declined; 18 other phase I genes were down-regulated less than 50% of baseline, including three isoforms of Cyp3a, five isoforms of Cyp2c, and four isoforms of Cyp2d, and Cyp2e1. Cyp1a2, Cyp2a1, Cyp2a2, and Cyp2c11 were also down-regulated by qPCR.
Biochemical activities of multiple cytochrome P450 enzymes were determined by addition of test substrates to hepatocyte cultures. The 7-ethoxyresorufin conversion to resorufin (CYP1A1, CYP1A2) was highest in spheroids compared to monolayer conditions on all days (P < 0.05, Fig. 9). Collective activity of CYP3A and CYP2C11 was determined by biotransformation of diazepam to oxazepam. Oxazepam concentrations were significantly higher in spheroids (4.35 ± 0.98 μM) compared to both monolayer culture conditions after day 2 (TCP: 1.42 ± 0.35 μM; CM: 1.21 ± 0.34 μM; P < 0.001) on day 1. Production of oxazepam peaked in spheroid culture on day 7. This trend paralleled the expression of Cyp3a genes determined by microarray.
Phase II Metabolism/Conjugation.
Expression of 14 of 16 (88%) phase II metabolism genes, including six sulfotransferases and all five uridine diphosphate glucuronosyl transferases (Udpgt), remained stable during 14 days of spheroid culture. Sulfotransferase 1B (Sult1b) was up-regulated whereas sulfotransferase-estrogen-preferring (Ste) was down-regulated.
Gene Induction by βNF.
The influence of βNF on gene expression of spheroid hepatocytes was quantified by microarray data at each time point. Nine genes summarized in Table 2 were induced by βNF. The expression of most inducible genes remained high throughout the culture period. Six genes (Cyp1a1, Cyp1a2, Cyp2a, Aldh3a1, Nqo1, and Hmox) were also examined by qPCR, which confirmed their induction by βNF.
Table 2. Rat Genes Induced by β-Naphthoflavone (βNF) in Rocked Spheroid Conditions
Expression of 15 of 18 antioxidant genes (83%) remained stable during 14 days of culture. Two antioxidant genes, glutathione synthetase (Gss) and glutathione S-transferase Mu 1 (Gstm1), showed increased expression by an average greater than two-fold over 14 days. Glutathione reductase (Gsr), glutathione S-transferase theta 2 (Gstt2), and microsomal glutathione S-transferase 1 (Mgst1) showed a decline in expression.
Expression of five of seven (71%) clotting factor genes remained stable during 14 days of culture. Thrombin and factor X showed an average two-fold decrease from baseline over 14 days.
The most important observation from this study is that spheroid formation occurs rapidly and most efficiently by rocker technique. Rapid aggregation of hepatocytes into spheroids by rocked technique was associated with fewer dead cells and improved spheroid morphology compared to the traditional rotational technique. These observations are significant and predict that a higher dose of primary hepatocytes is possible in a spheroid reservoir configuration compared to prior designs of bioartificial livers.18, 19 Rocker-formed spheroids may also prove to be an alternative method for preparation and maintenance of hepatocytes prior to cellular transplantation.20 Transplantation of hepatocyte spheroids has been reported previously.21 Studies are planned to test whether rocker-formed spheroids are suitable for therapeutic purposes such as hepatocyte transplantation.
The influence of spheroid size on hepatocyte function was previously studied by Glicklis et al.,22 who predicted that the diffusion of oxygen would not be rate limiting and central hepatocytes would not become hypoxic if spheroid diameter was on the order of 100 μm. Our observations support those of Glicklis et al. in that we found optimal spheroid size to be approximately 100-150 μm based on viability and mitochondrial membrane potential of rocked spheroid hepatocytes (unpublished observations). We determined mitochondrial membrane potential of hepatocytes by tetramethylrhodamine ethyl ester perchlorate staining.23
The formation of spheroids by rocker technique draws focus on the anchorage dependency of epithelial cells such as hepatocytes and the importance of cell-cell contact and adhesion to extracellular matrix in maintaining differentiated phenotype. In turn, hepatocyte spheroids formed by rocker technique preserved differentiated function better than classic monolayer culture techniques. Alternative techniques exist for culture of hepatocytes and creation of in vitro liver tissue constructs, including collagen sandwich24 and hepatocyte coculture.6, 25 Comparisons to these more elaborate, labor-intensive techniques will be helpful in establishing the role of the rocked spheroid technique in experimental biology of liver tissue constructs.
Further optimization of the rocked hepatocyte culture system may also be required, because a progressive decline in cell number was observed over the 2-week study interval. The decline in both metabolic activity and protein production paralleled the decline in cell number, despite stable gene expression on a viable cell basis. The reason for cell loss from the spheroid system has not yet been determined. Some of the spheroid losses occurred during changes of media, while other losses occurred by adhesion of spheroids to the walls of the bioreactor and along the liquid-air interface where surface tension was greatest. Cell death between time points was another possibility because hepatocytes appeared viable in intact spheroids when examined by vital staining and transmission electron microscopy (data not shown) during the 14 day culture period.
In conclusion, we offer a thorough analysis of the biochemical functions and liver-specific gene expression of rat hepatocyte spheroids formed by rocker technique over 14 days in culture. More than 85% of hepatocytes were incorporated into spheroids by rocker technique after 24 hours compared to 58% by conventional rotational technique. Cell loss occurred in all culture systems studied; however, the rate of cell loss was lower and biochemical activity was significantly greater at all time points in rocked spheroid cultures compared to control conditions. Custom microarray studies confirmed stable and inducible expression of many liver-related genes in spheroids formed by rocked technique. Further optimization of this system for application as a bioartificial liver is warranted.
Special thanks to Nancy Borson, Trace Christensen, Chris Kolbert, Perry Loken, Vernadette Simon, and Jim Tarara for their expert technical assistance. Thanks to Diane Grill for statistical assistance.