Maintenance of epithelial cell shape and polarity determines many vital cell functions, including the appropriate response to external stimuli. Murine hepatocytes cultured in a three-dimensional Matrigel matrix formed highly polarized organoids characterized by specific localization of an ERM (ezrin/radixin/moesin) protein, radixin, at microvillus-lined membrane domains. These apical domains surrounded a lumen and were bordered by tight junctions. The hepatocyte organoids were functional as judged by the high level of albumin secretion and accumulation of bilirubin. Stimulation of the Fas/CD95 death receptor, which is highly hepatotoxic in vivo, was a strong inducer of apoptosis in the polarized organoids. This was in sharp contrast to the monolayer hepatocyte cultures, which were protected from death by exacerbated NF-κB signalling following engagement of the death receptors. Thus, hepatocytes in polarized, functional organoids modulate an intracellular signal transduction pathway, allowing the recapitulation of their physiological response to an apoptotic stimulus.

Programmed cell death, a process of eliminating excess, defective or potentially dangerous cells, is essential during embryonic development and in adult homeostasis, as well as in controlling such major threats to the integrity of metazoans as viral infections and oncogenic transformation (Green and Evan, 2002). Apoptosis, which appears to be the most common form of programmed cell death in the animal kingdom, is subject to intricate controls, ultimately targeting the activation of a family of cysteine proteases, the caspases (Salvesen and Abrams, 2004). The sophistication of the control mechanisms, interwoven into the major signalling networks of the cell, renders the final outcome of apoptotic signalling (i.e. survival or death) strongly dependent on the physiological context of the cell.

One important aspect of an epithelial cell's physiology is its apico-basolateral polarization. Even though many epithelial cells adopt a partially polarized morphology in a monolayer culture, the establishment and maintenance of a strongly polarized state requires appropriate adhesive substrata and cell-cell contacts, and the correct geometrical orientation (Chen et al., 1997; Knust and Bossinger, 2002; O'Brien et al., 2002). Epithelial cells cultured in a physiologically relevant three-dimensional (3D) matrix form polarized organoids, with the basal cell surfaces in contact with the matrix and the apical membrane domains, delimited by tight junctions between adjacent cells, pointing into a lumen. For example, thyroid cells embedded in collagen organize into follicles (Chambard et al., 1981), Madin-Darby canine kidney (MDCK) cells form cysts and tubules in both collagen and Matrigel [a 3D matrix derived from the Enggelbreth-Holm-Swarm mouse sarcoma (Zegers et al., 2003)], and mammary epithelium grown in Matrigel forms hollow acini composed of polarized cells that retain many characteristics of mammary tissue in vivo (Weaver et al., 1997), including the resistance to a range of apoptotic stimuli (Weaver et al., 2002).

Hepatocytes are epithelial cells of complex polarity, characterized by the presence of several distinct apical membrane domains. In vivo, the apical surfaces of adjacent cells form microvillus-lined lumens called bile canaliculi. Primary hepatocytes can be cultured in vitro over extensive periods of time. Although they rapidly dedifferentiate in standard monolayer cultures (for a review, see LeCluyse et al., 1996), several culture conditions have been described that allow long-term maintenance of the differentiated phenotype (Dunn et al., 1989; Lazaro et al., 2003; Michalopoulos et al., 2001; Moghe et al., 1996; Rojkind et al., 1980). For example, cells in collagen gels remain cuboidal and polarized, and maintain the production of hepatocyte-specific markers (Dunn et al., 1989; LeCluyse et al., 1996; Moghe et al., 1996; Richert et al., 2002). Matrigel, a tumour-derived laminin-rich basal-membrane extracellular matrix, provides an especially appropriate environment for hepatocyte culture, as judged by the cells' morphology and metabolism (LeCluyse et al., 1996; Semler and Moghe, 2001; Semler et al., 2000). Interestingly, in addition to the chemical composition of the matrix, the physical properties of the gels (especially the pore size and the pliability) profoundly influence both the spatial organization and the differentiation of the cells (Moghe et al., 1996; Ranucci et al., 2000; Semler and Moghe, 2001).

It is widely accepted that the physiological context of a cell is a strong determinant of the outcome of an apoptotic stimulation (Danial and Korsmeyer, 2004). Although the influence of culture conditions on hepatocyte viability and function has been extensively studied (Allen et al., 2001; Block et al., 1996; Dunn et al., 1989; Khalil et al., 2001; LeCluyse et al., 1996; Michalopoulos and Pitot, 1975; Richert et al., 2002; Rojkind et al., 1980; Semler et al., 2000), little is known about the effect of hepatocyte morphology and physiology on the modulation of apoptosis. One well-described pathway of apoptosis induction originates from the engagement of death receptors, such as tumour necrosis factor receptor (TNFR) or Fas (CD95), expressed on the surface of many cell types (Debatin and Krammer, 2004). Upon binding of their respective ligands, the receptors recruit multiprotein complexes, in which the initiator procaspase 8 undergoes activation mediated by proteolytic cleavages. Recruitment of other proteins and activation of additional signalling pathways control this first step of apoptotic signalling. In particular, NF-κB activation, frequently associated with death receptor stimulation, constitutes a potent restraint of the cell's apoptotic response (Baud and Karin, 2001). The importance of NF-κB signalling in the liver is exemplified by the phenotype of mice invalidated for the p65/RelA subunit of NF-κB: the animals die at embryonic day 15-16 of liver destruction (Beg et al., 1995), owing to TNFR1-dependent apoptosis (Rosenfeld et al., 2000).

Liver cells are exquisitely sensitive to Fas stimulation in vivo: fulminant hepatitis is a frequently lethal consequence of massive hepatocyte apoptosis following Fas activation (Ryo et al., 2000; Song et al., 2003). In mice, intravenous injection of an agonistic anti-Fas antibody kills the animals through liver destruction and haemorrhage (Ogasawara et al., 1993). However, in monolayer cultures of both immortalized and primary hepatocytes, strong pro-survival signalling accompanies death-receptor signal transduction, rendering the cells resistant to Fas- or TNFR-mediated apoptosis in the absence of NF-κB inhibition.

In the present study, we have used mhAT3F, a differentiated murine hepatocyte cell line (Antoine et al., 1992; Levrat et al., 1993), to assess the possible links between the cells' polarity and function with their sensitivity to apoptosis induced by stimulation of death receptors. Contrary to the reported resistance to apoptosis of polarized mammary epithelium (Weaver et al., 2002), strong polarization of hepatocytes in a three-dimensional culture diminished NF-κB activation and restored the cells' in vivo sensitivity to the stimulation of the Fas pathway.

Reagents

Laminin-rich basal membrane from Enggelbreth-Holm-Swarm mouse sarcoma (Matrigel), MatriSperse, type-1 collagen, Falcon cell-culture inserts and anti-β-catenin and anti-γ-catenin antibodies were from BD Biosciences. Anti-ZO-1 antibody was a kind gift from T. Flemming, University of Cambridge, Cambridge, UK, anti-mouse albumin was from Bethyl Laboratories. Polyclonal anti-radixin antibody was prepared against human recombinant radixin, as previously described for ezrin (Andreoli et al., 1994), secondary antibodies and TRITC-labelled phalloidin were from Sigma. Agonistic anti-Fas antibody (Jo2) was from Pharmingen, polyclonal anti-caspase-3 antibody was from CST, anti-rat CD2 antibody was from Cedarlane, TNFα was from Peprotech and cycloheximide, actinomycin D, dexamethasone, triiodothyronine and insulin were from Sigma-Aldrich. Anti-NF-κBp65 antibody was a kind gift from A. Israël (Institut Pasteur, Paris, France) and M. Karin (UCSD, La Jolla, CA), anti-IκBα and anti-IKKβ antibodies were from M. Karin. Plasmid encoding rat CD2 antigen, lacking the intracytoplasmic domain, was a gift from C. Norbury (University of Oxford, UK). The (Igκ)3-conaluc NF-κB reporter construct contains firefly luciferase cloned downstream from three consensus NF-κB binding sites (Munoz et al., 1994). The IκBAA expression vector encodes a mutant form of IκBα with S32A and S36A substitutions. These mutations prevent IκBα phosphorylation by IKK and thus render it nondegradable by the ubiquitin-dependent proteasome pathway (DiDonato et al., 1996). Both plasmids were a kind gift of A. Israël.

Cell culture

MhAT3F cells were cultured in DMEM supplemented with 5% foetal calf serum, 5 μg ml–1 insulin, 1 μM dexamethasone, 1 μM triiodothyronine, 250 ng ml–1 fungizone, 20 μg ml–1 streptomycin and 25 U ml–1 penicillin.

For the 3D Matrigel culture, monolayer grown cells were harvested by trypsinization, taken up in the complete medium and separated into single-cell suspension by several passages through a Pasteur pipette. 50 μl complete medium containing 100,000 cells was added to 250 μl Matrigel, gently mixed, deposited into a Falcon cell culture insert and incubated at 37°C until Matrigel solidified. The cells were then covered with complete medium and grown at 37°C under 5% CO2 for up to 8 days with the change of medium every 2 days. For statistical analysis of organoid formation, 40-70 structures in randomly chosen fields were counted in two independent experiments.

For the collagen sandwich cultures, 100 μl collagen (2.5 mg ml–1) in complete medium was left to solidify in a culture insert at 37°C for 30 minutes. The culture was seeded with 8000 cells in 500 μl complete medium for 2 hours and covered with a second layer of 100 μl collagen and the complete medium. The cells were cultured for up to 8 days with medium change every 2 days.

Sample preparation and immunofluorescence

Floating and attached cells from monolayer cultures were harvested, pooled and deposited on slides by cytospin centrifugation. Matrigel or collagen cultures were included in Cryomatrix (Shandon), snap frozen in liquid N2 and kept at –80°C. For analysis, 20 μm sections were deposited on slides, fixed in 4% paraformaldehyde for 10 minutes, permeabilized for 2 minutes in 0.1% Triton X-100, rinsed in PBS and blocked with 10% foetal calf serum. Alternatively, caspase-3 staining was performed on whole collagen inserts, in which case the fixation period was 15 minutes. Samples were incubated with primary antibodies followed by the appropriate fluorochrome-labelled secondary antibodies (both for 1 hour at room temperature). Control sections were stained with secondary antibodies only. Actin was visualized by TRITC-phalloidin labelling. Nuclei were counterstained with Hoechst 33258 (2 μM). The mounting media were Mowiol (Calbiochem) for the two-dimensional (2D) samples or glycerol/PBS (9:1) containing p-phenylenediamine for the Matrigel and collagen sections.

Confocal microscopy

Immunofluorescence confocal images were acquired using a Zeiss LSM 510 inverted laser-scanning confocal microscope equipped with an external argon laser. Images were captured at 0.5 μm intervals using a Zeiss fluor 63× objective, deconvoluted with Huygens Pro and reconstructed in 3D with Imaris. Individual confocal stacks are shown, unless otherwise indicated.

Transmission electron microscopy and histology

Matrigel cultures were fixed in situ with 2.5% glutaraldehyde in 0.1 M phosphate buffer pH 7.3 for 1 hour, washed for 15 minutes in the same buffer with 6.84% sucrose, post-fixed for 1 hour in 2% osmium tetroxide, dehydrated in a graded ethanol series followed by propylene oxide treatment for 3D culture and another ethanol-dehydration step for 2D samples, and embedded in Epon 812. Semi-thin sections were stained with toluidine blue and ultra-thin sections were contrasted with uranyl acetate and lead citrate, and observed with a Jeol 1200X transmission electron microscope.

Histochemistry

Hall's bilirubin staining was performed according to a standard protocol (Sheehan and Hrapchak, 1980).

Apoptosis assay

Cells were incubated with Jo2 antibody (1 μg ml–1) or TNFα (100 ng ml–1) for 22 hours, cycloheximide (1 μg ml–1) or actinomycin D (2 μg ml–1) were added, as indicated. Apoptotic cells were identified by immunofluorescence staining of activated caspase 3, the total number of cells was assessed by nuclear staining with the Hoechst 33258 dye. For quantification, at least 400 cells were counted in randomly chosen fields. The exact number of apoptotic cells in 3D organoids being more difficult to estimate, apoptosis in 3D cultures was quantified by counting caspase-3-positive organoids rather than cells. At least 80 organoids were analysed for each Matrigel section. Typically, only a few cells were caspase-3 positive in untreated organoids in the course of their maturation, whereas, in the course of the response to the stimulation of the death receptors, a caspase-3-positive organoid was composed mostly of cells with activated caspase 3.

NF-κB reporter assay

Subconfluent monolayer cells were co-transfected with the NF-κB-dependent luciferase reporter (Igκ3-conaluc) (Munoz et al., 1994) or a control plasmid lacking the NF-κB binding sites, and a constitutive Renilla luciferase construct. 18 hours later, cells were collected, split in two and either plated as a monolayer culture or embedded in Matrigel. After 24 hours or 8 days, respectively, for the 2D and 3D cultures, the cells were treated with Jo2 antibody (1 μg ml–1) or TNFα (100 ng ml–1) for 7 hours and harvested by scraping and by MatriSperse treatment, respectively, for 2D and 3D cultures. The cells were lysed and luciferase activities were assayed using the Promega dual luciferase kit according to the manufacturer's instructions. In all cases, and specifically in the samples derived from the 3D culture, unstimulated activities of both Renilla and firefly luciferases (measured in lysates of transfected cells without any death-receptor stimulation) were at least 30 times higher than the low non-specific background values of untransfected samples. The unstimulated firefly luciferase activity, normalized to Renilla luciferase, was arbitrarily set as 1 and the activation of NF-κB signalling was presented as a fold increase over this value.

Western-blot analysis

Matrigel or collagen inserts were recovered, rinsed in PBS and digested at 37°C in the presence of protease inhibitors, with Matrisperse for 1 hour or collagenase H for 30 minutes. The cells were centrifuged, lysed in 1× SDS sample buffer, analysed by PAGE and blotted onto nitrocellulose membranes. Secretion of albumin was allowed to proceed for 24 hours in serum-free medium and an aliquot of medium was analysed. Cell-free matrices or cultures of murine fibroblasts were used as a control for detection of bovine albumin; both gave rise to a negligible signal.

Statistical analysis

Results are expressed as means ± s.d. Each assay, performed in triplicate, was repeated at least twice. Statistical significance was analysed by Fisher's test: ****, P≤0.0001; ***, P≤0.001; *, P≤0.05.

Hepatocytes form organoids in a Matrigel-based 3D culture

The mhAT3F hepatocyte cell line, established from a mouse with liver-targeted production of SV40 large T antigen, retains a pattern of gene expression resembling that of normal hepatocytes (Levrat et al., 1993). The cells grow in monolayer cultures as flattened clusters (Fig. 1a). When seeded as single-cell suspension in three-dimensional basal membrane (Matrigel), they continued to proliferate for several days, forming compact organoids of 20-30 cells (Fig. 1b), although considerably larger structures could also be found. Once the organoids were formed, the proliferation stopped, as judged by time-lapse microscopy analysis (data not shown). Most structures (80%) at day 4 were filled with cells (Fig. 1c), whereas the fully formed organoids, which represented about 60% of structures at day 8, were composed of cells lining either one or several intercellular lumens (Fig. 1d-f). The lumen arose, at least in part, from apoptotic death of cells lacking contact with the Matrigel. Indeed, in the course of the organoid formation, cells in the interior, but not those of the exterior layer, contained activated caspase 3, a hallmark of apoptosis (Fig. 1g). Caspase-3 activation was seen in 45% of organoids at day 4, and the labelling was even stronger at day 5 (60% of structures), whereas the process was completed at day 8 and only 8% of the organoids were labelled with the anti-activated caspase-3 antibody. Caspase-3 activation was indeed associated with apoptosis in this context, as witnessed by the presence of apoptotic bodies inside some of the lumens (Fig. 1e).

Organoids contain hepatocytes that are cuboid, highly polarized and functional

The epithelial mhAT3F cells produced abundant, membrane-localized E-cadherin and β- and γ-catenin, in both confluent monolayers and 3D organoids (see Fig. S1 in supplementary material), as well as cortically distributed F-actin (Fig. 2A,B). We searched for three hallmarks of defined apical structures – localized distribution of radixin [a member of the ERM family whose association with bile canalicular membranes is essential for normal liver function (Kikuchi et al., 2002)], localized presence of microvilli and the presence of tight junctions separating the microvillus-lined apical domains from the basolateral membrane. In the monolayer-grown mhAT3F cells, radixin was strongly produced only by a subset of cells and localized in a punctate pattern on the surface facing the culture medium. In the organoids, all cells stained positive for this marker, whose localization was specific to the surfaces bordering the lumens (Fig. 2A). Similarly, ZO-1 was detected both in 2D and 3D cultures (Fig. 2B), suggesting that the tight junctions could be formed under both conditions. As expected, the labelling of monolayer-grown cells was predominantly basolateral and some, but not all, cells stained positive. In the organoids, the labelling was particularly intense around the lumen, indicative of a well-formed tight-junction belt.

Fig. 1.

MhAT3F hepatocytes form organoids in Matrigel. Phase-contrast microscopy of monolayer grown cells (a) and organoids formed after 8 days of culture in Matrigel (b). After 4 days of culture, most of the organoids were filled with cells (c), whereas, at day 8, most were hollow (d) as visualized by Hoechst 33258 nuclear staining (blue). Multiple lumens are visualized on semi-thin sections of day 8 organoids (e,f). Apoptotic bodies (ab) and debris can be seen within some lumens (L) (e), whereas others contained no dense material (f). n, nucleus. Before lumen formation, cells in the interior of the organoids contain activated caspase 3 (green) (g). Bars, 20 μm (a-d,g), 3.5 μm (e,f).

Fig. 1.

MhAT3F hepatocytes form organoids in Matrigel. Phase-contrast microscopy of monolayer grown cells (a) and organoids formed after 8 days of culture in Matrigel (b). After 4 days of culture, most of the organoids were filled with cells (c), whereas, at day 8, most were hollow (d) as visualized by Hoechst 33258 nuclear staining (blue). Multiple lumens are visualized on semi-thin sections of day 8 organoids (e,f). Apoptotic bodies (ab) and debris can be seen within some lumens (L) (e), whereas others contained no dense material (f). n, nucleus. Before lumen formation, cells in the interior of the organoids contain activated caspase 3 (green) (g). Bars, 20 μm (a-d,g), 3.5 μm (e,f).

Transmission-electron-microscopy analysis confirmed that both monolayer cells and organoids contained polarized cells with visible tight junctions (Fig. 2Ca,b, monolayer, Fig. 2Cc,d, 3D). However, we estimate that only about 50% of monolayer-grown hepatocytes contained tight junctions and distinct apical, microvillus-lined, membrane domains. These extended apical domains facing the medium contained short, sparse microvilli, whereas abundant, well-formed microvilli lined the intercellular lumens of 3D organoids (Fig. 2C). Furthermore, the cells in 3D structures and in monolayers had drastically different shapes: a cuboid geometry was observed in the 3D culture, whereas the monolayer cells were flat (estimated height/width ratio=0.90±0.16 and 0.32±0.06, respectively). Thus, mhAT3F cells were flattened and weakly polarized in monolayer cultures and acquired a cuboid shape and strong polarization when arranged into 3D organoids in the Matrigel.

Collagen-sandwich culture is an alternative support compatible with hepatocyte polarization (Dunn et al., 1989; LeCluyse et al., 1996; Moghe et al., 1996). Under our experimental conditions of low-density seeding, mhAT3F cells adopted two distinct morphological organizations in collagen: spheroid organoids similar to those formed in the Matrigel and extended sheets of cells (Fig. 2D). Whereas the distribution of radixin was very similar in organoids formed in either 3D matrix (Fig. 2A,E), the cells in the extended configuration showed more uniform labelling.

In order to ascertain whether the cells grown under different conditions retained functional characteristics of differentiated hepatocytes, we first measured their capacity to secrete albumin (Fig. 3A). Monolayer-grown cells synthesized and secreted albumin that was detectable in the conditioned medium. The secretion was slightly increased in immature (day 4) organoids and strongly increased both in the mature Matrigel-grown organoids and in collagen-grown cells. Bilirubin accumulation is another stringent test for hepatocyte function. In contrast to the cells grown as a monolayer (Fig. 3Ba), the organoids produced histologically detectable bilirubin (Fig. 3Bb), as judged by a green colour after Hall's histological staining (Sheehan and Hrapchak, 1980). Cholestatic liver section, characterized by excess bile accumulation, served as a positive control in this experiment (Fig. 3Bc).

Thus, the morphology, polarization and function of the mhAT3F cells were drastically improved when the cells were cultured embedded in a three-dimensional matrix, with the most dramatic effects observed upon their assembly into compact organoids.

Fig. 2.

Hepatocytes in the organoids are strongly polarized. MhAT3F cells were grown either in a monolayer or included in Matrigel or a collagen sandwich. The cells had an epithelial phenotype, visualized by the cortical actin distribution (phalloidin-TRITC, red), both in monolayer (2D) and in the organoids (3D). (A) Radixin (fluorescein-isothiocyanate labelled, green), a marker of the apical membrane domains in hepatocytes, showed a strong apical labelling of a subset of monolayer cells, easily visible in 3D-reconstructed (3D reconst.) image and a z-axis sections (below). It decorated the lumen-exposed membranes of all cells in 3D organoids. Nuclei are stained with Hoechst 33258 (blue). 3D-reconstructed images are shown, except for overlay panels, which show single confocal stacks. (B) ZO-1 (fluorescein-isothiocyanate labelled, green), a tight-junction marker, had a heterogeneous distribution in monolayer cells but was strongly localized around the lumen in the organoids. (C) Transmission electron microscopy of monolayer (a,b) and organoids (c,d) allows visualization of microvilli (mv) and tight junctions (tj). Multiple small (a) or larger (b) lumens (IL) lined with microvilli can be seen in the organoids. n, nucleus. (D) Phase-contrast microscopy of cells in the collagen-sandwich culture. Two types of structures can be seen: spheroid organoids (arrows) and extended sheets of cells. (E) Immunofluorescence analysis of collagen-sandwich cultures. Radixin labelling (green) is punctate in extended cells (top) and strongly localized to the interior of the organoids (bottom). Bars, 20 μm (A,B,E), 1 μm (C), 40 μm (D).

Fig. 2.

Hepatocytes in the organoids are strongly polarized. MhAT3F cells were grown either in a monolayer or included in Matrigel or a collagen sandwich. The cells had an epithelial phenotype, visualized by the cortical actin distribution (phalloidin-TRITC, red), both in monolayer (2D) and in the organoids (3D). (A) Radixin (fluorescein-isothiocyanate labelled, green), a marker of the apical membrane domains in hepatocytes, showed a strong apical labelling of a subset of monolayer cells, easily visible in 3D-reconstructed (3D reconst.) image and a z-axis sections (below). It decorated the lumen-exposed membranes of all cells in 3D organoids. Nuclei are stained with Hoechst 33258 (blue). 3D-reconstructed images are shown, except for overlay panels, which show single confocal stacks. (B) ZO-1 (fluorescein-isothiocyanate labelled, green), a tight-junction marker, had a heterogeneous distribution in monolayer cells but was strongly localized around the lumen in the organoids. (C) Transmission electron microscopy of monolayer (a,b) and organoids (c,d) allows visualization of microvilli (mv) and tight junctions (tj). Multiple small (a) or larger (b) lumens (IL) lined with microvilli can be seen in the organoids. n, nucleus. (D) Phase-contrast microscopy of cells in the collagen-sandwich culture. Two types of structures can be seen: spheroid organoids (arrows) and extended sheets of cells. (E) Immunofluorescence analysis of collagen-sandwich cultures. Radixin labelling (green) is punctate in extended cells (top) and strongly localized to the interior of the organoids (bottom). Bars, 20 μm (A,B,E), 1 μm (C), 40 μm (D).

Hepatocyte organoids are sensitive to Fas stimulation

Intravenous injection of an agonistic anti-Fas antibody (Jo2) is strongly hepatotoxic, leading to massive apoptotic death of hepatocytes and liver haemorrhage (Ogasawara et al., 1993). By contrast, both primary and immortalized murine hepatocytes in monolayer culture survived treatment with even very high doses of the same antibody (Fig. 4A and data not shown), even though their apoptotic signal transduction pathway was fully functional, as revealed by the efficient induction of apoptosis following Fas stimulation in the presence of an inhibitor of protein synthesis (Fig. 4A-C). In order to account for the different sensitivities of hepatocytes to Fas stimulation in vivo and in vitro, we assessed the effects of polarization of mhAT3F on their response to death-receptor stimulation. Treatment of 3D organoids by Jo2 resulted in massive apoptosis, as shown both by morphological changes (Fig. 4A) and by activation of caspase 3, detected in over 80% of organoids (Fig. 4B,D). The same treatment performed in parallel on cells in a monolayer culture gave rise to no apoptotic response above the low background (Fig. 4B,C).

Fig. 3.

mhAT3F organoids are functional. (A) Western-blot analysis of albumin secretion by mhAT3F cells grown under the indicated conditions. Albumin accumulation was allowed to proceed for 24 hours and the amount of medium assayed was normalized to the number of cells present in the culture, estimated by their GAPDH content. A representative blot is shown, the numbers represent mean fold increase compared with monolayer cultures, as calculated after scanning of three western blots. (B) Monolayer grown mhAT3F show no detectable bilirubin (a), whereas the green Hall's staining of the whole organoids (b) indicates the accumulation of bilirubin. A section of a cholestatic human liver (c) is shown as a positive control. Bars. 20 μm.

Fig. 3.

mhAT3F organoids are functional. (A) Western-blot analysis of albumin secretion by mhAT3F cells grown under the indicated conditions. Albumin accumulation was allowed to proceed for 24 hours and the amount of medium assayed was normalized to the number of cells present in the culture, estimated by their GAPDH content. A representative blot is shown, the numbers represent mean fold increase compared with monolayer cultures, as calculated after scanning of three western blots. (B) Monolayer grown mhAT3F show no detectable bilirubin (a), whereas the green Hall's staining of the whole organoids (b) indicates the accumulation of bilirubin. A section of a cholestatic human liver (c) is shown as a positive control. Bars. 20 μm.

These data suggested that the maintenance of cell shape and polarization might lead to an increased sensitivity to apoptosis. To settle this point unambiguously, we took advantage of the fact that the mhAT3F cells adopted two distinct configurations in a collagen sandwich – spherical organoids and extended sheets of cells (Fig. 2D). Upon Fas stimulation of the collagen-grown cells, only the spherical organoids underwent apoptosis, whereas the same treatment performed in the presence of cycloheximide killed all the cells, irrespective of their 3D-organisation (Fig. 4G). To gain a more quantitative insight into this effect, we counted all structures present in a culture insert. Before treatment, intact spherical organoids represented 39% of all structures (n=117) but, after incubation with Jo2, they dropped to 7% (n=128).

One major difference between subconfluent monolayers versus cells organized into mature 3D organoids is the cells' proliferative status. However, quiescence was not responsible for the increased sensitivity to Fas stimulation, because the fully confluent mhAT3F monolayer cultures, characterized by a drastically reduced BrdU labelling index (not shown), did not die upon treatment with Jo2 (see Fig. S2 in supplementary material). Thus, cell density and proliferative status did not account for the apoptotic response of the 3D organoids.

Interestingly, organoids in the Matrigel were not significantly more sensitive to the stimulation of a related death-receptor pathway. As shown in Fig. 4F, treatment of 3D cultures with TNFα did not give rise to significant caspase-3 activation. However, the full apoptotic response to TNFα was obtained by simultaneously blocking macromolecular synthesis in both 2D and 3D cultures (Fig. 4E,F).

NF-κB regulates the apoptotic response to death-receptor stimulation

Activation of NF-κB by the ligand-bound death receptor is a widely accepted paradigm of survival signalling (for a review, see Baud and Karin, 2001). Accordingly, we asked whether interfering with NF-κB activation was sufficient to promote Fas-mediated apoptosis in mhAT3F cells. Monolayer-grown cells were co-transfected with vectors encoding a truncated rat CD2 antigen and a mutant `super-repressor' IκB protein, whose production efficiently inhibits p65/p50 NF-κB activation (Munoz et al., 1994). After treatment with the Jo2 anti-Fas antibody, transfected cells were labelled with the anti-CD2 antibody and apoptotic cells were detected by staining of activated caspase 3. No caspase activation was seen in untransfected (CD2-negative) cells, whereas 25% of transfected cells responded to Fas stimulation by activating caspase 3 (Fig. 5A,B). Thus, NF-κB activation contributed to survival signalling in this model of Fas-induced apoptosis.

We reasoned that, in order to account for the different sensitivities to Fas stimulation, NF-κB should be activated differently following death-receptor engagement in 2D- and 3D-cultured cells. However, the levels of production of the p65 NF-κB subunit, as well as of IκBα and of IKKβ, two major regulators of NF-κB activity, were not altered by the 3D culture conditions (Fig. 5C and data not shown). In order to set up a reporter assay of the transcriptional activity of NF-κB following Fas stimulation, we verified that the limited number of cell divisions that occurred during the formation of an organoid (estimated at four to five for most of the structures) allowed the maintenance of detectable levels of a transiently transfected reporter gene (not shown). Next, we transfected cells with a luciferase-based NF-κB reporter plasmid and, 24 hours later, dispatched them into monolayer and 3D Matrigel cultures. Stimulation with TNFα or anti-Fas antibody was performed after an additional 24 hours in monolayers or 8 days in Matrigel. As shown in Fig. 5D, engagement of death receptors gave rise to a considerably stronger NF-κB activation in monolayer cultures than in the polarized organoids. Furthermore, treatment with TNFα was a stronger activator of NF-κB than stimulation of the Fas pathway. As a consequence, there was no detectable NF-κB activation over background in the polarized organoids of mhAT3F cells treated with the anti-Fas antibody.

Fig. 4.

MhAT3F organoids are sensitized to Fas-induced apoptosis. (A-F) Monolayers (2D) or 3D organoids (3D) were treated with the anti-Fas antibody (Jo2) in the presence or absence of cycloheximide, as indicated. (A) Phase-contrast microscopy. Arrows indicate organoids enlarged in inserts. (B) Pooled floating and adherent monolayer cells deposited on microscope slides by cytospin (top) or sections of Matrigel (bottom) were analysed by immunofluorescent staining of active caspase 3. Nuclei are shown in blue (Hoechst 33258), caspase 3 in green (fluorescein isothiocyanate). (C,D) Quantification of caspase-3-positive cells in monolayer cultures (C; 450 cells counted) and in 3D cultures (D; 40 organoids counted). (E) Monolayer cultures treated with TNFα with or without actinomycin D, as indicated, were collected and analysed as in C. (F) 3D organoids were treated with TNFα with or without actinomycin D and analysed as in D. (G) Cells grown in collagen sandwich were treated as indicated and analysed by phase-contrast microscopy (top) or by immunofluorescent staining of the entire collagen plug with antibody against active caspase 3 (green). Nuclei were stained with Hoechst 33258 (blue). Bars, 20 μm (A,B), 40 μm (G).

Fig. 4.

MhAT3F organoids are sensitized to Fas-induced apoptosis. (A-F) Monolayers (2D) or 3D organoids (3D) were treated with the anti-Fas antibody (Jo2) in the presence or absence of cycloheximide, as indicated. (A) Phase-contrast microscopy. Arrows indicate organoids enlarged in inserts. (B) Pooled floating and adherent monolayer cells deposited on microscope slides by cytospin (top) or sections of Matrigel (bottom) were analysed by immunofluorescent staining of active caspase 3. Nuclei are shown in blue (Hoechst 33258), caspase 3 in green (fluorescein isothiocyanate). (C,D) Quantification of caspase-3-positive cells in monolayer cultures (C; 450 cells counted) and in 3D cultures (D; 40 organoids counted). (E) Monolayer cultures treated with TNFα with or without actinomycin D, as indicated, were collected and analysed as in C. (F) 3D organoids were treated with TNFα with or without actinomycin D and analysed as in D. (G) Cells grown in collagen sandwich were treated as indicated and analysed by phase-contrast microscopy (top) or by immunofluorescent staining of the entire collagen plug with antibody against active caspase 3 (green). Nuclei were stained with Hoechst 33258 (blue). Bars, 20 μm (A,B), 40 μm (G).

Fig. 5.

NF-κB regulates Fas-induced apoptosis. (A) Monolayer-grown mhAT3F were co-transfected with vectors encoding a truncated rat CD2 and IκBAA and treated with the Jo2 anti-Fas antibody. Floating and adherent cells were pooled, deposited on a slide by cytospin and analysed (nuclei are blue, active caspase 3 is green and CD2 is red). (B) Quantification of cells containing active caspase 3 within the subpopulation of transfected cells. (C) Western blot of NF-κB p65 subunit in total cell extracts of cells grown in monolayer (2D) or as organoids in Matrigel (3D). Actin serves as a loading control. (D) Monolayer cultures of mhAT3F cells were co-transfected with NF-κB/firefly-luciferase and constitutive Renilla-luciferase reporter constructs. 18 hours later, cells were replated and grown either as monolayer cultures (open bars) or 3D organoids (hatched bars) and stimulated by the anti-Fas antibody Jo2 or TNFα, as indicated, for 7 hours. Results are shown as means±s.e.m. Bars, 20 μm.

Fig. 5.

NF-κB regulates Fas-induced apoptosis. (A) Monolayer-grown mhAT3F were co-transfected with vectors encoding a truncated rat CD2 and IκBAA and treated with the Jo2 anti-Fas antibody. Floating and adherent cells were pooled, deposited on a slide by cytospin and analysed (nuclei are blue, active caspase 3 is green and CD2 is red). (B) Quantification of cells containing active caspase 3 within the subpopulation of transfected cells. (C) Western blot of NF-κB p65 subunit in total cell extracts of cells grown in monolayer (2D) or as organoids in Matrigel (3D). Actin serves as a loading control. (D) Monolayer cultures of mhAT3F cells were co-transfected with NF-κB/firefly-luciferase and constitutive Renilla-luciferase reporter constructs. 18 hours later, cells were replated and grown either as monolayer cultures (open bars) or 3D organoids (hatched bars) and stimulated by the anti-Fas antibody Jo2 or TNFα, as indicated, for 7 hours. Results are shown as means±s.e.m. Bars, 20 μm.

3D mhAT3F organoids are polarized and functional

MhAT3F immortalized murine hepatocytes are well-differentiated cells as judged by the high-level production of liver-specific transactivators, including HNF1, HNF2, HNF3 and C/EBPα (Antoine et al., 1992) and of major liver-specific proteins such as albumin, L-type pyruvate kinase and aldolase B (Levrat et al., 1993). In the present work, we have extended these characterizations by showing that culture in an appropriate 3D environment led to a strong apico-basolateral polarization and the maintenance of liver-specific functions such as a high level of albumin secretion and accumulation of bilirubin. In agreement with previous reports on long-term culture of primary hepatocytes (LeCluyse et al., 1996; Moghe et al., 1996; Semler and Moghe, 2001; Semler et al., 2000), we found that the microenvironment of the Matrigel basal-membrane extract provided the best spatial and biochemical cues for mhAT3F polarization and function. It is noteworthy that Matrigel's chemical composition is similar both to the basal-membrane components of the liver in vivo (Stamatoglou and Hughes, 1994) and to the endogenous matrix deposited by aggregates of primary liver cells in suspension culture (Landry et al., 1985).

We have shown that the Matrigel embedded mhAT3F cells organized into compact spheroids that matured, through apoptotic elimination of internally localized cells, into structures containing one, or several, intercellular lumens (Figs 1, 2). Large lumens are usually absent from 3D structures formed by primary hepatocytes in 3D Matrigel cultures (Moghe et al., 1996). A possible explanation for this discrepancy might be the difference in the mechanism of organoid formation: whereas the primary cells aggregate into organoids, the mhAT3F structures are formed by several cell divisions in the first days of culture.

Relatively large lumens and a series of adjacent small lumens could both be detected in mhAT3F organoids (Fig. 2C). The latter suggest a possible mechanism of lumen formation by the fusion of small apical domains, reminiscent of tubulogenesis in tracheal morphogenesis in Drosophila (Lubarsky and Krasnow, 2003). In addition, apoptosis clearly participated in the organoid lumen formation, as witnessed by the transient caspase-3 activation during the process of maturation and the presence of apoptotic bodies in some of the lumens (Fig. 1).

Lumen formation by apoptosis that occurs in 3D cultures of kidney (O'Brien et al., 2002) and mammary (Debnath et al., 2002; Muthuswamy et al., 2001) epithelium is believed to reflect the physiological death occurring during development and homeostasis of these tissues. Apoptosis as a mechanism for lumen formation is more surprising in hepatocyte organoids. Even though an acinar arrangement of partially depolarized hepatocytes has been described in vivo during organogenesis and regeneration, and in some cases of well-differentiated hepatocellular adenomas (Stamatoglou and Hughes, 1994), hepatocytes do not usually form large tubular structures but rather small canaliculi. However, in contrast to the 3D culture conditions, the extracellular matrix in a healthy liver is not deposited uniformly but lines the tracts of hepatocytes. The geometrical constraints imposed on cells striving to maintain contacts with both the matrix and the neighbouring cells, and to preserve a free apical surface are thus different in vivo and in 3D cell culture. This could be sufficient to account for an altered arrangement of cells in the organoids. In any case, the highly polarized localization of radixin, a marker of hepatocytic bile canaliculi (Kikuchi et al., 2002), argues in favour of strong cellular polarization of cells in the organoids.

It is noteworthy that, in contrast to the reported structures of mature MDCK or mammary epithelium organoids (O'Brien et al., 2002; Weaver et al., 1997), the hepatocyte organoids often contained several intercellular lumens, apparently bordered by as few as two or three cells (Fig. 2; see Fig. S3 in supplementary material). Such organization is reminiscent of a morphology of a genuine bile canaliculus in the liver.

Organoids undergo apoptosis upon Fas stimulation

Hepatocytes are a major physiological target of Fas/CD95 signalling in vivo in both mice and humans. Intravenous injection of an agonistic anti-Fas antibody kills the mouse through liver destruction owing to hepatocyte apoptosis and haemorrhage (Ogasawara et al., 1993), and animals with a homozygous invalidation of the Fas gene, in addition to immunological disorders, suffer from hepatomegaly, presumably caused by inefficient hepatocyte apoptosis (Adachi et al., 1995). Similarly, both acute hepatitis and fulminant hepatic failure in human patients appear to be related to strong pro-apoptotic Fas signalling (Ryo et al., 2000). By contrast, neither primary nor immortalized hepatocytes grown in traditional monolayer cultures are killed by Fas-receptor engagement unless survival signalling, which follows the death-receptor stimulation (Baud and Karin, 2001), is simultaneously inhibited.

This difference of behaviour could be due partly to the in vivo effect of Fas stimulation on the liver endothelial cells (Jodo et al., 2003). Our data show that, alternatively but not exclusively, the altered geometry of the cells in a monolayer culture can affect survival signalling and lead to resistance to Fas-induced apoptosis. Assembly into 3D organoids preserved the hepatocyte-specific morphology and function, and strongly influenced the cells' apoptotic responses. The observed sensitivity to Fas stimulation was due neither to the proliferation arrest (see Fig. S2 in supplementary material) nor to the simple contact with Matrigel, because the sensitization to Fas stimulation only occurred after 3 days of culture (i.e. was coincident with the organoid formation) (data not shown). Thus, sensitivity to Fas-induced apoptosis correlated with the establishment of strong cellular polarization and correct geometry. This conclusion is further strengthened by the results obtained in the collagen culture, where two types of cell organization were present simultaneously. Cells that formed compact organoids, similar to those observed in the Matrigel culture, underwent massive apoptosis, whereas those arranged into extended sheets survived Fas stimulation. Because these strikingly different behaviours co-existed in the same culture, they provide a strong argument for a causal relationship between cellular geometry and the apoptotic response.

Interestingly, in contrast to the mammary epithelium, in which 3D polar acini were resistant to a range of apoptotic stimuli, including death-receptor engagement (Weaver et al., 2002), maintenance of polarity in hepatocytes correlated with an increased sensitivity to Fas stimulation. This apparent discrepancy could be a reflection of differences in the physiology of the two cell types because, in contrast to many cell types, healthy hepatocytes are a major target of Fas-induced apoptosis in vivo (Ogasawara et al., 1993; Ryo et al., 2000; Song et al., 2003).

NF-κB regulates the apoptotic response to death-receptor stimulation

The apoptotic signal-transduction pathway leading from the engagement of death receptors to caspase-3 activation has been described in considerable detail (for a review, see Wallach et al., 1999). It involves the recruitment of DISC (death-inducing signalling complex) to the intracytoplasmic domain of the trimeric receptor, resulting in proteolytic activation of the initiator caspase 8. In hepatocytes, which are type II cells in respect to Fas signalling (Scaffidi et al., 1998), the resulting apoptotic signalling is strictly dependent on caspase-8-mediated cleavage of Bid, a pro-apoptotic member of the Bcl2 family (Disson et al., 2004; Yin et al., 1999).

Large numbers of cells are required for a purification of DISC. This is incompatible with 3D culture: as a result we could not analyse it biochemically in our experimental system. As a consequence, we cannot formally exclude the possibility that cellular polarization altered the production and/or the subcellular localization of the receptors and their associated proteins. However, the inhibition of macromolecular synthesis gave rise to an efficient activation of the apoptotic cascade that was indistinguishable in monolayer and 3D cultures. This finding argues against major differences in the composition or arrangement of the apoptotic machinery in cells cultured under the two experimental conditions.

Activation of NF-κB by the ligand-bound death receptor constitutes a potent survival signal (for a review, see Baud and Karin, 2001). Although this has been well described for TNFR engagement in several cell types, its involvement in Fas signalling is less well documented. Our data confirm a previous report of the inhibition of NF-κB signalling in hepatocytes that conferred an increased sensitivity to Fas-induced apoptosis (Hatano et al., 2000). Moreover, we have shown that the NF-κB activation by death-receptor engagement in polarized hepatocytes was much reduced in comparison to unpolarized cells. The molecular mechanisms responsible for this difference are under investigation. It is neither due to altered production of NF-κB itself (Fig. 5C) nor its regulators IκBα and IKKβ (not shown). Furthermore, the decrease, or the lack of NF-κB activation following TNFR or Fas stimulation, respectively, was unlikely to be due to a lower accessibility of the receptors to the ligands in the 3D matrix, because the apoptotic response to the same treatments was actually increased under these conditions.

Taken together, our data support the idea that the maintenance of polarized cell morphology has a strong impact on the cellular interpretation of an apoptotic signal. Interestingly, our results argue against the notion that cell polarity is an universal anti-apoptotic feature of an epithelial cell (Weaver et al., 2002) but, rather, suggest that it allows to recapitulate important aspects of a cell physiology.

We are grateful to A. Israël for NF-κB reporter and IκB AA plasmids and for the anti-IκB antibody, H. Kamata for the NF-κB western blot, S. Tran for help with the immunofluorescence labelling, E. Antoine for BrdU labelling, and C. Marty-Double for help with thebilirubin assays. We thank all the members of the UH lab for comments and discussions and D. Fisher for carefully reading the manuscript. Supported by INSERM, CNRS and Association pour la Recherche contre le Cancer (ARC) grant 4769 (to UH). D.H. was in part supported by a fellowship from the Fondation de Treilles.

Adachi, M., Suematsu, S., Kondo, T., Ogasawara, J., Tanaka, T., Yoshida, N. and Nagata, S. (
1995
). Targeted mutation in the Fas gene causes hyperplasia in peripheral lymphoid organs and liver.
Nat. Genet.
11
,
294
-300.
Allen, J. W., Hassanein, T. and Bhatia, S. N. (
2001
). Advances in bioartificial liver devices.
Hepatology
34
,
447
-455.
Andreoli, C., Martin, M., Le Borgne, R., Reggio, H. and Mangeat, P. (
1994
). Ezrin has properties to self-associate at the plasma membrane.
J. Cell Sci.
107
,
2509
-2521.
Antoine, B., Levrat, F., Vallet, V., Berbar, T., Cartier, N., Dubois, N., Briand, P. and Kahn, A. (
1992
). Gene expression in hepatocyte-like lines established by targeted carcinogenesis in transgenic mice.
Exp. Cell Res.
200
,
175
-185.
Baud, V. and Karin, M. (
2001
). Signal transduction by tumor necrosis factor and its relatives.
Trends Cell Biol.
11
,
372
-377.
Beg, A. A., Sha, W. C., Bronson, R. T., Ghosh, S. and Baltimore, D. (
1995
). Embryonic lethality and liver degeneration in mice lacking the RelA component of NF-kappa B.
Nature
376
,
167
-170.
Block, G. D., Locker, J., Bowen, W. C., Petersen, B. E., Katyal, S., Strom, S. C., Riley, T., Howard, T. A. and Michalopoulos, G. K. (
1996
). Population expansion, clonal growth, and specific differentiation patterns in primary cultures of hepatocytes induced by HGF/SF, EGF and TGF alpha in a chemically defined (HGM) medium.
J. Cell Biol.
132
,
1133
-1149.
Chambard, M., Gabrion, J. and Mauchamp, J. (
1981
). Influence of collagen gel on the orientation of epithelial cell polarity: follicle formation from isolated thyroid cells and from preformed monolayers.
J. Cell Biol.
91
,
157
-166.
Chen, C. S., Mrksich, M., Huang, S., Whitesides, G. M. and Ingber, D. E. (
1997
). Geometric control of cell life and death.
Science
276
,
1425
-1428.
Danial, N. N. and Korsmeyer, S. J. (
2004
). Cell death: critical control points.
Cell
116
,
205
-219.
Debatin, K. M. and Krammer, P. H. (
2004
). Death receptors in chemotherapy and cancer.
Oncogene
23
,
2950
-2966.
Debnath, J., Mills, K. R., Collins, N. L., Reginato, M. J., Muthuswamy, S. K. and Brugge, J. S. (
2002
). The role of apoptosis in creating and maintaining luminal space within normal and oncogene-expressing mammary acini.
Cell
111
,
29
-40.
DiDonato, J., Mercurio, F., Rosette, C., Wu-Li, J., Suyang, H., Ghosh, S. and Karin, M. (
1996
). Mapping of the inducible IκB phosphorylation sites that signal its ubiquitination and degradation.
Mol. Cell Biol.
16
,
1295
-1304.
Disson, O., Haouzi, D., Desagher, S., Loesch, K., Hahne, M., Kremer, E. J., Jacquet, C., Lemon, S. M., Hibner, U. and Lerat, H. (
2004
). Impaired clearance of virus-infected hepatocytes in transgenic mice expressing the hepatitis C virus polyprotein.
Gastroenterology
126
,
859
-872.
Dunn, J. C., Yarmush, M. L., Koebe, H. G. and Tompkins, R. G. (
1989
). Hepatocyte function and extracellular matrix geometry: long-term culture in a sandwich configuration.
FASEB J.
3
,
174
-177.
Green, D. R. and Evan, G. I. (
2002
). A matter of life and death.
Cancer Cell
1
,
19
-30.
Hatano, E., Bradham, C. A., Stark, A., Iimuro, Y., Lemasters, J. J. and Brenner, D. A. (
2000
). The mitochondrial permeability transition augments Fas-induced apoptosis in mouse hepatocytes.
J. Biol. Chem.
275
,
11814
-11823.
Jodo, S., Kung, J. T., Xiao, S., Chan, D. V., Kobayashi, S., Tateno, M., Lafyatis, R. and Ju, S. T. (
2003
). Anti-CD95-induced lethality requires radioresistant Fcgamma RII+ cells. A novel mechanism for fulminant hepatic failure.
J. Biol. Chem.
278
,
7553
-7557.
Khalil, M., Shariat-Panahi, A., Tootle, R., Ryder, T., McCloskey, P., Roberts, E., Hodgson, H. and Selden, C. (
2001
). Human hepatocyte cell lines proliferating as cohesive spheroid colonies in alginate markedly upregulate both synthetic and detoxificatory liver function.
J. Hepatol.
34
,
68
-77.
Kikuchi, S., Hata, M., Fukumoto, K., Yamane, Y., Matsui, T., Tamura, A., Yonemura, S., Yamagishi, H., Keppler, D. and Tsukita, S. (
2002
). Radixin deficiency causes conjugated hyperbilirubinemia with loss of Mrp2 from bile canalicular membranes.
Nat. Genet.
31
,
320
-325.
Knust, E. and Bossinger, O. (
2002
). Composition and formation of intercellular junctions in epithelial cells.
Science
298
,
1955
-1959.
Landry, J., Bernier, D., Ouellet, C., Goyette, R. and Marceau, N. (
1985
). Spheroidal aggregate culture of rat liver cells: histotypic reorganization, biomatrix deposition, and maintenance of functional activities.
J. Cell Biol.
101
,
914
-923.
Lazaro, C. A., Croager, E. J., Mitchell, C., Campbell, J. S., Yu, C., Foraker, J., Rhim, J. A., Yeoh, G. C. and Fausto, N. (
2003
). Establishment, characterization, and long-term maintenance of cultures of human fetal hepatocytes.
Hepatology
38
,
1095
-1106.
LeCluyse, E., Bullock, P. and Parkinson, A. (
1996
). Strategies for restoration and maintenance of normal hepatic structure and function in long-term cultures of rat hepatocytes.
Adv. Drug Deliver. Rev.
22
,
133
-186.
Levrat, F., Vallet, V., Berbar, T., Miquerol, L., Kahn, A. and Antoine, B. (
1993
). Influence of the content in transcription factors on the phenotype of mouse hepatocyte-like cell lines (mhAT).
Exp. Cell Res.
209
,
307
-316.
Lubarsky, B. and Krasnow, M. A. (
2003
). Tube morphogenesis: making and shaping biological tubes.
Cell
112
,
19
-28.
Michalopoulos, G. and Pitot, H. C. (
1975
). Primary culture of parenchymal liver cells on collagen membranes. Morphological and biochemical observations.
Exp. Cell Res.
94
,
70
-78.
Michalopoulos, G. K., Bowen, W. C., Mule, K. and Stolz, D. B. (
2001
). Histological organization in hepatocyte organoid cultures.
Am. J. Pathol.
159
,
1877
-1887.
Moghe, P. V., Berthiaume, F., Ezzell, R. M., Toner, M., Tompkins, R. G. and Yarmush, M. L. (
1996
). Culture matrix configuration and composition in the maintenance of hepatocyte polarity and function.
Biomaterials
17
,
373
-385.
Munoz, E., Courtois, G., Veschambre, P., Jalinot, P. and Israel, A. (
1994
). Tax induces nuclear translocation of NF-kappa B through dissociation of cytoplasmic complexes containing p105 or p100 but does not induce degradation of I kappa B alpha/MAD3.
J. Virol.
68
,
8035
-8044.
Muthuswamy, S. K., Li, D., Lelievre, S., Bissell, M. J. and Brugge, J. S. (
2001
). ErbB2, but not ErbB1, reinitiates proliferation and induces luminal repopulation in epithelial acini.
Nat. Cell Biol.
3
,
785
-792.
O'Brien, L. E., Zegers, M. M. and Mostov, K. E. (
2002
). Opinion: building epithelial architecture: insights from three-dimensional culture models.
Nat. Rev. Mol. Cell Biol.
3
,
531
-537.
Ogasawara, J., Watanabe-Fukunaga, R., Adachi, M., Matsuzawa, A. and Kasugai, T. (
1993
). Lethal effect of the anti-Fas antibody in mice.
Nature
364
,
806
-809.
Ranucci, C. S., Kumar, A., Batra, S. P. and Moghe, P. V. (
2000
). Control of hepatocyte function on collagen foams: sizing matrix pores toward selective induction of 2-D and 3-D cellular morphogenesis.
Biomaterials
21
,
783
-793.
Richert, L., Binda, D., Hamilton, G., Viollon-Abadie, C., Alexandre, E., Bigot-Lasserre, D., Bars, R., Coassolo, P. and LeCluyse, E. (
2002
). Evaluation of the effect of culture configuration on morphology, survival time, antioxidant status and metabolic capacities of cultured rat hepatocytes.
Toxicol. In Vitro
16
,
89
-99.
Rojkind, M., Gatmaitan, Z., Mackensen, S., Giambrone, M. A., Ponce, P. and Reid, L. M. (
1980
). Connective tissue biomatrix: its isolation and utilization for long-term cultures of normal rat hepatocytes.
J. Cell Biol.
87
,
255
-263.
Rosenfeld, M. E., Prichard, L., Shiojiri, N. and Fausto, N. (
2000
). Prevention of hepatic apoptosis and embryonic lethality in RelA/TNFR-1 double knockout mice.
Am. J. Pathol.
156
,
997
-1007.
Ryo, K., Kamogawa, Y., Ikeda, I., Yamauchi, K., Yonehara, S., Nagata, S. and Hayashi, N. (
2000
). Significance of Fas antigen-mediated apoptosis in human fulminant hepatic failure.
Am. J. Gastroenterol.
95
,
2047
-2055.
Salvesen, G. S. and Abrams, J. M. (
2004
). Caspase activation – stepping on the gas or releasing the brakes? Lessons from humans and flies.
Oncogene
23
,
2774
-2784.
Scaffidi, C., Fulda, S., Srinivasan, A., Friesen, C., Li, F., Tomaselli, K. J., Debatin, K. M., Krammer, P. H. and Peter, M. E. (
1998
). Two CD95 (APO-1/Fas) signaling pathways.
EMBO J.
17
,
1675
-1687.
Semler, E. J. and Moghe, P. V. (
2001
). Engineering hepatocyte functional fate through growth factor dynamics: the role of cell morphologic priming.
Biotechnol. Bioeng.
75
,
510
-520.
Semler, E. J., Ranucci, C. S. and Moghe, P. V. (
2000
). Mechanochemical manipulation of hepatocyte aggregation can selectively induce or repress liver-specific function.
Biotechnol. Bioeng.
69
,
359
-369.
Sheehan, D. and Hrapchak, B. (
1980
).
Theory and Practice of Histotechnology
. Ohio: Battelle Press, Columbus, Ohio.
Song, E., Lee, S. K., Wang, J., Ince, N., Ouyang, N., Min, J., Chen, J., Shankar, P. and Lieberman, J. (
2003
). RNA interference targeting Fas protects mice from fulminant hepatitis.
Nat. Med.
9
,
347
-351.
Stamatoglou, S. C. and Hughes, R. C. (
1994
). Cell adhesion molecules in liver function and pattern formation.
FASEB J.
8
,
420
-427.
Wallach, D., Varfolomeev, E. E., Malinin, N. L., Goltsev, Y. V., Kovalenko, A. V. and Boldin, M. P. (
1999
). Tumor necrosis factor receptor and Fas signaling mechanisms.
Annu. Rev. Immunol.
17
,
331
-367.
Weaver, V. M., Petersen, O. W., Wang, F., Larabell, C. A., Briand, P., Damsky, C. and Bissell, M. J. (
1997
). Reversion of the malignant phenotype of human breast cells in three-dimensional culture and in vivo by integrin blocking antibodies.
J. Cell Biol.
137
,
231
-245.
Weaver, V. M., Lelievre, S., Lakins, J. N., Chrenek, M. A., Jones, J. C., Giancotti, F., Werb, Z. and Bissell, M. J. (
2002
). Beta4 integrin-dependent formation of polarized three-dimensional architecture confers resistance to apoptosis in normal and malignant mammary epithelium.
Cancer Cell
2
205
-216.
Yin, X. M., Wang, K., Gross, A., Zhao, Y., Zinkel, S., Klocke, B., Roth, K. A. and Korsmeyer, S. J. (
1999
). Bid-deficient mice are resistant to Fas-induced hepatocellular apoptosis.
Nature
400
,
886
-891.
Zegers, M. M., O'Brien, L. E., Yu, W., Datta, A. and Mostov, K. E. (
2003
). Epithelial polarity and tubulogenesis in vitro.
Trends Cell Biol.
13
,
169
-176.