Hepatic Polarization Accelerated by Mechanical Compaction Involves HNF4α Activation

There remain few data about the role of homeostatic compaction in hepatic polarization. A previous study has found that mechanical compaction can accelerate hepatocyte polarization; however, the cellular mechanism underlying the effect is mostly unclear. Hepatocyte nuclear factor 4 alpha (HNF4α) is crucial for hepatic polarization in liver morphogenesis. Therefore, we sought to identify any possible involvement of HNF4α in the process of hepatocyte polarization accelerated by mechanical compaction. We first verified in the nonhepatic cell model HEK-293T, and the hepatic cell model primary hepatocytes that the mechanical compaction on cell aggregates simulated by using transient centrifugation can directly activate the expression of HNF4α promoters. Moreover, data using primary hepatocytes showed that the HNF4α expression is positively associated with the levels of compaction force: 2.1-folds higher at the mRNA level and 2.1-folds higher at the protein level for 500 g vs. 0 g. Furthermore, activated HNF4α expression is associated with the enhanced biliary canalicular formation and the increased production of albumin and urea. Pretreatment with Latrunculin B, an inhibitor of F-actin, and SHE78-7, an inhibitor of E-cadherin, which both interrupt the pathway of mechanical transduction, partially but significantly reduced the HNF4α expression and production of albumin and urea. In conclusion, HNF4α can be actively involved in the hepatic polarization in the context of environmental mechanical compaction.


Introduction
Hepatic polarity is important for the generation and maintenance of physiological function in hepatocytes [1]. It also contributes to diverse pathological processes, such as hepatitis B virus cellular intrusion, intrahepatic cholestasis, and steatosis [2,3]. However, the mechanism underlying the maintenance of polarity in hepatocytes in mature liver tissues is far from clear. A previous study has found that mechanical compaction accelerated hepatocyte repolarization and bile canaliculus (BC) formation by imposing extensive intercellular contact and adhesion [4]. Mechanical compaction accel-erated hepatocyte repolarization without impairing the hepatocytes' remodeling and functional capabilities. However, cellular mechanism underlying this interesting effect is yet to be understood.
Recently, it became evident that mechanical forces applied by the pericellular environment or generated inside cells function as important upstream signals that trigger the establishment of cell and tissue polarity [5][6][7]. For instance, for the Drosophila wing, disc planar cell polarity proteins coordinately reorient in line with stress patterns induced by hinge contraction [8], and for endothelial cells, shear stress regulates the forward and the reverse planar cell polarity of the vascular endothelium through upregulation of glycogen synthase kinase-3 in vivo and in vitro [9]. To respond to various mechanical forces, cells may sense and translate mechanical forces into a biochemical signal by cellular elements such as the actin cytoskeleton and cadherinmediated cell adherent junctions [10].
Hepatocyte nuclear factor 4 alpha (HNF4α), an orphan member of the nuclear receptor superfamily (NR2A1), is physiologically expressed at a high level and binds to the promoters of 12% of genes in the adult liver [11]. A role for HNF4α in hepatic polarity was suggested because its deficiency in the embryonic mouse liver resulted in an abnormal tissue architecture and a lack of cell polarity [12]. Its overexpression induces cell polarity in F9 embryonic carcinoma [13] and H5 hepatoma [14]. Although the effects of the mechanical force on the establishment and maintenance of cell polarization have been generally recognized [5][6][7], whether there could be an involvement of HNF4α in the accelerated hepatic polarization by mechanical compaction still lacks supporting evidence. Therefore, in the present study, we sought to investigate (1) whether mechanical compaction can directly enhance HNF4α expression and (2) whether the cytoskeleton as well as cadherin adherent junctions is involved in the process.

Primary Hepatocyte Isolation and Maintenance.
Adult male wild-type C57BL/6 mice (6-8 weeks old) were purchased from the Laboratory Animal Center of Guangdong Province. Primary hepatocytes were isolated from C57BL/6 mice via collagenase perfusion as described previously [15]. Cells were cultured in William's E medium (Sigma-Aldrich, St. Louis, MO) supplemented with 1 mg/mL bovine serum albumin (Mp Biomedicals, California, USA), 5 ng/mL epidermal growth factor (Prospec, Saint Louis, USA), 0.2 units/mL insulin (Sigma-Aldrich), 100 nmol/L dexamethasone (Sigma-Aldrich), and penicillin/streptomycin (Sigma-Aldrich). Prior to seeding, the cell culture substrates were coated with 0.012 mg/mL acidic collagen (1 : 417 dilution of PureCol® collagen (Advanced BioMatrix, USA) in 6 mM acetic acid). The cells were then seeded onto the collagen-coated membrane and maintained in a 5% CO 2 atmosphere at 37°C. Culture medium was changed every day. All animal experiments were previewed and approved by the Animal Care and Use Committee of Southern Medical University.

Luciferase Reporter
Assay. Cells cultured in 6-well plates were transfected with Gaussia Luciferase (GLuc) and Secreted Alkaline Phosphatase (SEAP) promoter reporter clones containing the HNF4α promoter 1 (P1, Cat. no. HPRM16317-PG04) or P2 (Cat. no. HPRM20338-PG04) gene sequences (GeneCopoeia, Germantown, MD) by using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA). Culture medium was collected at 24 h after transfection. GLuc and SEAP activities were measured using the Secrete-Pair Dual Luciferase Assay Kit (GeneCopoeia) according to the manufacturer's instructions. Data were normalized against SEAP. A GLuc and SEAP promoter reporter clone with the glyceraldehyde 3-phosphate dehydrogenase (GAPDH) gene sequence insertion was used as a positive control, and a wild-type GLuc and SEAP promoter reporter clone was used as the negative control group (NEG).

Mechanical
Compaction on the Cell Seeding. We used instant centrifugation to simulate the mechanical compaction on cell seeding, as centrifugation can generate the compaction force between the cells, and more importantly, it can quantify the force. The schematic diagram of mechanical compaction on cells is shown in Figure 1. In brief, singlecell suspension was seeded into V-bottom 96-well plates and centrifuged for 5 min at 0 × g, 500 × g, 1000 × g, and 1500 × g. Cells were maintained in 5% CO 2 at 37°C and then were cultured. The culture medium was collected, and the GLuc and SEAP activities of HNF4α P1 and P2 were assessed at 0 h, 3 h, 6 h, 12 h, and 24 h after centrifugation. The total RNA and total protein in cells were extracted to measure the mRNA and the protein expression levels of HNF4α. Cells centrifuged at 0 × g were used as the control. Since 500 × g had the most significant effect, it was used in subsequent experiments.

Disruption of the Polymerization of Cellular F-Actin.
Cells were cultured with 1000 and 2000 nM Latrunculin B (LatB; Cat. no. ab144291, Abcam, Cambridge, England) for 2 h to disrupt the polymerization of cellular F-actin [16]. The morphology of F-actin polymerization was observed after specifically stained F-actin with Alexa Fluor 488 phalloidin (Cat. no. A12379, 1 : 20, Invitrogen) by using a Leica DM2500 fluorescence microscope with a 20x objective. Since 2000 nM LatB had the better effects, it was used in the mechanical compaction and its subsequent experiments.

Disruption of the Establishment of Cadherin Adherent
Junction. Cells were treated with 7.5 and 15 μg/mL SHE78-7 (Cat. no. 135700, Invitrogen) for 24 h to disrupt the establishment of cadherin adherent junction [17]. The E-cadherin adherent cell-cell contacts were observed by using an inverted-phase contrast microscope with a 20x objective. Since 15 μg/mL SHE78-7 had the better effects, it was used in the mechanical compaction and its subsequent experiments.

Albumin and Urea Secretion
Assays. Culture media were collected at 0 h, 3 h, 6 h, 12 h, and 24 h from each well and stored at -20°C until the hepatic function assay was performed. The albumin and the urea production were quantified using a fully automatic biochemistry analyzer (Olympus Au400; Tokyo, Japan). 2.9. Western Blot Analysis. The expression level of HNF4α protein was measured by western blotting as previously described [18]. The following antibodies were used: primary antibodies HNF4α (Cat. no. ab201460, 1 : 1000, Abcam) and GAPDH (Cat. no. KC-5G5, 1 : 8000, KangChen Bio-tech, Shanghai, China) and an HRP-conjugated secondary antibody (Cat. no. bs-0295G-HRP, 1 : 8000, Boster, Wuhan, China). Endogenous GAPDH was used as a loading control. The X-ray film of the internal reference control GAPDH and the target protein HNF4α band was photographed using a gel image scanning analysis system. The integrated optical density value of each band was measured by using the grayscale analysis software Image-Pro Plus 6.0. The optical density ratio of the target protein/internal reference protein is the relative expression of the target protein.

RNA Extraction and Real-Time Polymerase Chain
Reaction (RT-PCR) Analysis. Total RNAs from cell cultures were lysed using TRIzol® Reagent (Takara, Dalian, China), and genomic DNA was removed using the gDNA Eraser (Takara). RNA was reversely transcribed using RT Primer Mix and PrimeScript RT Enzyme Mix I (Takara). Expression analysis was performed by SYBR Green-based RT-PCR using SYBR® Premix Ex Taq II (Tli RNaseH Plus, Takara) and the 7500 RT-PCR Systems (Foster City, CA). Primers used for genes of interest were purchased from GeneCopoeia. Results were calculated using Ct values and normalized against endogenous GAPDH mRNA levels.

Generation of Transient Cell Lines with Enhanced
HNF4α. Nowadays, there is no relevant literature report on which the HNF4α isoform regulates the polar phenotype of liver epithelial cells. We searched for transcripts of HNF4α in the National Center for Biotechnology Information gene library and found that the transcript NM-000457.5 which encodes HNF4α2 is the longest in the all isoform transcripts, which means the transcript NM-000457.5 carries and expresses the most information of HNF4α compared with other isoform transcripts. Moreover, the transcript NM-000457.5 covers 71-86% of the complete sequence of the different HNF4α isoforms and contains the C, D, and E domains of HNF4α. In addition, in terms of the transcriptional function of the HNF4α isoforms, the α1 and α2 isoforms were the strongest regulators of gene expression [19]. Therefore, we choose to overexpress and silence the HNF4α2 isoform in the HepG2 cells to further explore the function of HNF4α in the polar phenotype of liver epithelial cells.
To establish cell lines with enhanced HNF4α, the recombinant adenovirus containing HNF4α2 gene sequences (Ad-HNF4α; Cat. no. HH20150327LJM-Ap01, Hanbio Biotechnology, Shanghai, China) or a negative control adenovirus (Ad-Mock, Hanbio Biotechnology) was used to infect HepG2 cells. The HNF4α expression was determined at the mRNA and protein levels.

Generation of Stable Cell Lines with Suppressed HNF4α.
To establish cell lines with suppressed HNF4α, the recombinant lentivirus containing oligonucleotides encoding a short hairpin RNA targeting HNF4α2 (LV-HNF4α; Cat. no. HH20150316LJM-Lv01, Hanbio Biotechnology) or a negative control lentivirus (LV-Mock, Hanbio Biotechnology) was used to infect HepG2 cells with polybrene (Cat. no. H9268, Sigma-Aldrich). After puromycin (Cat. no. A11138, Invitrogen) selection for 2 weeks, the stable clones were 3 BioMed Research International obtained. The HNF4α expression was determined at the mRNA and protein levels.

Functional Analysis of HepG2
Polarity. Briefly, cells were washed with fresh medium, incubated with medium containing 20 μg/mL fluorescein diacetate (Cat. no. F7378, FDA, Sigma-Aldrich) for 40 min at 37°C, and viewed using a Carl Zeiss LSM880 confocal microscope with a 60x oil objective within 5 min. The mean areas of FDA localization were identified and quantified five nonoverlapping fields of view using Image-Pro Plus 6.0 software.
2.14. Statistical Analysis. All data are expressed as means ± standard deviation (means ± S:D:) of three independent experiments. GraphPad Prism 5.01 and Adobe Illustrator CS6 were used to generate plots and perform statistical analyses. Data were analyzed using either one-way analysis of variance (Dunnett's multiple comparison test) or twoway analysis of variance (Bonferroni posttests). A level of P < 0:05 was considered statistically significant.

Mechanical Compaction Activates HNF4α Expression.
The initiation of HNF4α gene expression in liver cells mainly occurs by the activation of HNF4α P1 or P2 [20]. The renal epithelial HEK-293T cells were first treated to avoid the effect of endogenous HNF4α. As shown in Figure 2(a), the relative luciferase activity of HEK-293T cells transfected with GLuc and SEAP promoter reporter clones containing the HNF4α P1 gene sequences was 5.9-folds higher (P < 0:001) compared with that of the NEG. The relative luciferase activity of cells transfected with clones containing the HNF4α P2 gene sequences was 6.2-folds higher (P < 0:001). In the positive control cells transfected with the clones containing GAPDH gene sequences, the relative luciferase activity was 14.8-folds higher (P < 0:001).
We applied the same treatment to primary hepatocytes, and the relative luciferase activities of HNF4α P1 and P2 were similar to those in the HEK-293T cells. Compared with the control group (0 × g), the HNF4α P1 or P2 relative luciferase activities in primary hepatocytes stimulated by 500 × g, 1000 × g, or 1500 × g were 1.3-1.9-folds higher at 3 h (P < 0:05) and increased to 1.5-2.9-folds higher at 24 h (P < 0:001, Moreover, compared with the control group (0 × g), the HNF4α mRNA and protein expression levels in primary hepatocytes stimulated by centrifugation at 500 × g were equal to those in the control cells at 3 h and increased to 2.1-(P < 0:001) and 2.1-(P < 0:001) folds higher at 24 h (Figures 3(a)-3(c)), respectively.
Collectively, mechanical compaction can effectively activate HNF4α expression. Since 500 × g had the most significant effect, it was used in the subsequent experiments.

Mechanical Compaction Accelerates BC Formation and
Enhances Primary Hepatocyte Functions. Here, we used BC formation to exemplify the formative process of functional structures in the primary hepatocyte. Primary hepatocytes were transfected with clones containing the HNF4α P1 or P2 gene sequences, seeded into V-bottom 96-well plates, simulated by 500 × g, and placed on static culture for 24 h. Then, the ultrastructures of the primary hepatocytes were observed by TEM. As shown in Figure 3(d), we found that the BC cavities, microvilli, and tight junctions were present in the primary hepatocytes stimulated by 500 × g and absent in the control group cells stimulated by 0 × g.

Disruption of F-Actin Polymerization Partially Suppresses
HNF4α Expression Stimulated by Mechanical Compaction. In the control primary hepatocytes, the distribution of F-actin showed a typical network of microfilament bundles (stress fibers). In the primary hepatocytes treated with 1000 nM LatB, the contraction of F-actin microfilament bundles was not complete and some stress fibers disappeared. In the primary hepatocytes treated with 2000 nM LatB, contraction of F-actin microfilament bundles was complete and most of the stress fibers disappeared (Figure 5(a)). Consistent with previous results [23], LatB at 2000 nM had a better suppressive effect on F-actin polymerization and was used in the subsequent experiments. 4 BioMed Research International In the primary hepatocytes treated with 2000 nM LatB and stimulated by 500 × g, the relative luciferase activities of HNF4α P1 were 19.2% (P = 0:089) lower than those in the cells stimulated by 500 × g and untreated with LatB at 3 h and 20.9% (P < 0:05) lower at 24 h ( Figure 5(b)). Similarly, the relative luciferase activities of HNF4α P2 were 11.8% (P = 0:093) lower at 3 h and 18.5% (P < 0:05) lower at 24 h ( Figure 5(c)).

Disruption of E-Cadherin-Mediated Cell Adherent
Junctions Partially Suppresses HNF4α Expression Stimulated by Mechanical Compaction. In the control primary hepatocytes, the cadherin adherent junction showed a typical morphology. In the primary hepatocytes treated with 7.5 μg/mL SHE78-7, the cadherin adherent junctions were lost in some areas. In the primary hepatocytes treated with 15 μg/mL SHE78-7, cadherin adherent junctions were lost entirely in most areas (Figure 6(a)). As previously reported [24], SHE78-7 caused a concentration-related disruption in cadherin adherent junction. SHE78-7 at 15 μg/mL had a better suppressive effect on cadherin adherent junctions and was used in the subsequent experiments.
Compared with the control cells stimulated by 500 × g and untreated with SHE78-7, the relative luciferase activities of HNF4α P1 in primary hepatocytes exposed to 15 μg/mL Moreover, in the primary hepatocytes treated with 15 μg/mL SHE78-7 and stimulated by 500 × g, the protein expression level of HNF4α was 10.3% (P = 0:087) lower than that of the control cells untreated with LatB and stimulated by 500 × g at 3 h and 22.3% (P < 0:01) lower than that at 24 h (Figures 6(d) and 6(e)). The mRNA expression level of HNF4α was 10.8% (P = 0:058) lower at 3 h and 29.8% (P < 0:001) lower at 24 h (Figure 6(f)).

Overexpression of HNF4α Accelerates BC-Like Lumina
Formation and Maturation in HepG2 Cells. To investigate whether the enhanced or suppressed expression of HNF4α can be involved in hepatic polarization, the HNF4α expression in HepG2 cells was manipulated by infection with Ad-HNF4α or LV-HNF4α and Ad-Mock or LV-Mock as the blank control, of which the effects were verified at both the mRNA and protein levels. The detailed results are shown in supporting Figure S2.
The excretion of fluorescent bile acid analogues or organic ions, such as FDA, is a common method to explore functional polarity in cultured cells [25][26][27]. In the HepG2 cells with HNF4α overexpression, FDA secretion began at 72 h postseeding and FDA filled BC-like lumina between cells (Figure 8(a)). The areas of secreted FDA were 0:25 ± 0:03 mm 2 at 72 h postseeding and increased to 3:04 ± 0:13 mm 2 at 120 h postseeding (P < 0:001, Figure 8(b)); the metabolism and excretion of FDA both increased in a time-dependent manner. In the HepG2 cells with the HNF4α expression suppressed, we did not observe the formation of BC-like lumina even at 120 h postseeding. In the  The relative luciferase activities of HNF4α P1 and P2 and (d-f) the mRNA and protein levels of HNF4α in primary hepatocytes exposed to LatB and stimulated by 500 × g were lower than those in cells stimulated by 500 × g and untreated with LatB but still higher than those in cells just stimulated by 0 × g (two-way ANOVA analysis and Bonferroni posttests, & LatB treated vs. 0 g, # LatB treated vs. 500 g, n = 3, * P < 0:05, * * P < 0:01, * * * P < 0:001).  . (b, c) The relative luciferase activities of HNF4α P1 and P2 and (d, f) the mRNA and protein levels of HNF4α in primary hepatocytes exposed to SHE78-7 and stimulated by 500 × g were lower than those in cells stimulated by 500 × g and untreated with SHE78-7 but still higher than those in cells just stimulated by 0 × g (two-way ANOVA analysis and Bonferroni posttests,  Figure 7: Disruption of F-actin polymerization and E-cadherin-mediated cell adherent junctions partially weakens primary hepatocyte functions triggered by mechanical compaction. (a-h) The levels of albumin and urea production in primary hepatocytes exposed to LatB or SHE78-7 and stimulated by 500 × g were lower than those in cells stimulated by 500 × g and untreated with LatB or SHE78-7 but still higher than those in cells just stimulated by 0 × g (two-way ANOVA analysis and Bonferroni post-tests, & LatB or SHE78-7-treated vs. 0 g, # LatB or SHE78-7-treated vs. 500 g, n = 3, * P < 0:05, * * P < 0:01, * * * P < 0:001).

10
BioMed Research International control HepG2 cells, FDA secretion began at 96 h postseeding, and the areas of secreted FDA were lower than those in cells with HNF4α overexpression.

Discussion
Our data suggest that HNF4α can be actively involved in the hepatic polarization in the context of environmental mechanical compaction, which is associated with the actin cytoskeleton and cadherin-mediated cell adherent junctions. In our study, centrifugation was used to accumulate cells with an increased packing density and contact before static culture. Cell seeding is an important step in tissue engineering, and various cell seeding methods have been studied nowadays [28][29][30][31]. External force seeding is one of these methods, such as centrifugation [32,33] and filtration seeding [34]. Centrifugation seeding is demonstrated to be superior to static and spinner flask methodologies in terms of seeding efficiency and cellular distribution within scaffolds [35,36]. Using centrifugation seeding, a high seeding efficiency of up to 90% can be achieved in a short seeding time of 3-5 min for 1000-1800 rpm. Centrifugation of cells imposes an external force, and this force within a range of <2000 rpm, which we normally perform in mammalian cell culture, does not cause damage to cells [36]. Therefore, here, we used centrifugation to perform the mechanical force on cells. Moreover, when the same treatment to primary hepatocytes is applied, we found that the relative luciferase activities of HNF4α P1 and P2 induced by centrifugation were similar to those of the HEK-293T cells. The results indicated that the direct activation of HNF4α P1 and P2 was independent from cell type, and using centrifugation to perform mechanical force on cells is an effective method.
A previous study has found that mechanical compaction can accelerate hepatocyte polarization [4]; however, the cellular mechanism underlying the effect is mostly unclear. In the present study, HNF4α expression was significantly elevated with mechanical compaction clearly verified by luciferase reporter assay in both nonhepatic and hepatic cell models and then by tests at the mRNA and protein levels, therefore supporting the argument that mechanical compaction can directly activate HNF4α expression. HNF4α has been showed to be crucial for hepatic polarization in liver morphogenesis [12,13,37]; therefore, we went on determining whether HNF4α is involved in the accelerated hepatic polarization. We established HepG2 cell lines with both enhanced and suppressed HNF4α expression and found that the enhanced expression of HNF4α can accelerate HepG2 cell polarization, while the suppressed HNF4α expression inhibited HepG2 cell polarization. These findings indicated that, although there can be several pathways underlying the accelerated hepatic polarization by mechanical compaction, it is probable that mechanical compaction directly promotes the HNF4α expression which then affiliates the preparation of polarization-related pathways inside the hepatocyte. Meanwhile, mechanical compaction also produced the enhanced contact between the cells, which together with the intracellular preparation accelerates the polarization formation between the hepatocytes. Nevertheless, the accelerated polarization can be physiological as we showed that the presence of the functional structure of polarity (BC) and the level of typical hepatic production function (albumin and urea) were consistent with polarity acceleration.
In addition, we also found that the cell-cell contacts stimulated by mechanical compaction had positive effects on the expression of HNF4α. When the F-actin polymerization or cadherin-mediated adherent junctions were disrupted, which are two essential molecules of the cell-cell contact formation, the activation of HNF4α by mechanical compaction was partially and significantly suppressed. This was possibly due to the interrupted transduction of mechanical stimuli to promote HNF4α expression or the blockage of F-actin and E-cadherin which could directly suppress the expression of HNF4α or its related upstream regulatory molecules. We also found that suppressed activation of HNF4α was associated with the weakened polarization of primary hepatocytes and their production function. Thus, all these demonstrate that the mechanical compaction, promotion of HNF4α expression, and accelerated hepatic polarity are indeed closely associated.
It is well known that hepatocytes are a type of specialized epithelial cell with unique hepatic polarity phenotype in vivo. Hepatic polarity is the important key factor for the generation and maintenance of physiological function in hepatocytes [2]. Much effort is taken to mimic the hepatic polarity morphology when culturing hepatocytes in vitro so as to maintain proper hepatic functions as long as possible. Specifically, for hepatocytes cultured in vitro, they entered a flattened morphology, remain nonpolarized, and rapidly lose their liver-specific functions [2]. Therefore, hepatocytes are considered as fragile cells, and it is important to explore how to tune the microenvironment properly in order to maintain or enhance hepatic differentiation and function. Under experimental conditions, hepatocytes have been found to alter their multicellular structure like BC to enhance the hepatic functions when they experience mechanical stimuli such as surface binding affinity, aggregate size, substrate rigidity, and mechanical compaction [4,38,39]. Moreover, it became evident that mechanical forces applied by the pericellular environment or generated inside cells function as important upstream signals that trigger the establishment of cell and tissue polarity [5][6][7]. In this study, centrifugation was used to perform mechanical force on hepatocytes. As a result of centrifugation-generated forces, hepatocytes undergoing compaction not only exhibited accelerated repolarization, an in vivo-like morphology, but also better maintained hepatic functions as compared to the same cells cultured without centrifugation. As a novel method to modulate cell compaction and intercellular interactions, using instant centrifugation to simulate the mechanical compaction on the cell is a promising approach to maintain cultured hepatocyte in vitro at an in vivo-like morphology without the need to apply any additional processing.
HNF4α is a transcription factor essential for expression of a large array of genes that define hepatocyte function; for example, genes define the synthetic, metabolic, and detoxifying functions of the hepatocyte [40]. Hepatocytes from the liver deficient in HNF4α were small, failed to make proper cell-cell contacts, and lacked normal BC [12]. In the study, we found that mechanical compaction directly promotes the HNF4α expression to accelerate hepatocyte polarization. Since HNF4α is a key important regulator of the physiological function of hepatocytes, this further illustrates that mechanical compaction is a promising approach to maintain the function of cultured hepatocytes in vitro.
As the main cell type in the liver, hepatocytes are responsible for detoxification, metabolism, bile and protein production, and so on [41]. In vitro cultured hepatocytes are widely used in regenerative medicine, tissue engineering, drug testing platforms, and biological mechanism studies. Now, we have established a model to maintain the function of cultured hepatocytes in vitro; in the future studies, we can explore the longest time of the model for maintaining hepatocyte function and the possibility of using the model for drug toxicology testing.

In Conclusion
Our data demonstrate the first direct evidence that mechanical compaction can accelerate hepatocyte polarization through activating HNF4α, and the HNF4α activation can be associated with cell-cell contacts modulated by the actin cytoskeleton and cadherin-mediated cell adherent junctions. Our study provides an interesting insight into the molecular pathway underlying the physiological hepatocyte polarization associated with the extracellular environmental mechanical force. Future investigation is needed to identify the other important pathways in addition to the HNF4α activation for maintaining the normal hepatic polarization by environmental mechanical force, which can be meaningful for the biological understanding as well as the liver medicine. In addition, we have established a model to maintain the function of cultured hepatocytes in vitro. In the future studies, we can explore the longest time of the model for maintaining hepatocyte function and the possibility of using the model for drug toxicology testing, tissue engineering, biological mechanism studies, and so on.

Data Availability
The data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
All authors declare no competing financial interests. Figure S1: mechanical compaction has no damage on the cell viability. Figure S2: HNF4α expression of HepG2 cells could be effectively modulated by Ad-or LV-HNF4α recombination vector. (Supplementary Materials)