3D Culture System for Liver Tissue Mimicking Hepatic Plates for Improvement of Human Hepatocyte (C3A) Function and Polarity.

In vitro 3D hepatocyte culture constitutes a core aspect of liver tissue engineering. However, conventional 3D cultures are unable to maintain hepatocyte polarity, functional phenotype, or viability. Here, we employed microfluidic chip technology combined with natural alginate hydrogels to construct 3D liver tissues mimicking hepatic plates. We comprehensively evaluated cultured hepatocyte viability, function, and polarity. Transcriptome sequencing was used to analyze changes in hepatocyte polarity pathways. The data indicate that, as culture duration increases, the viability, function, polarity, mRNA expression, and ultrastructure of the hepatic plate mimetic 3D hepatocytes are enhanced. Furthermore, hepatic plate mimetic 3D cultures can promote changes in the bile secretion pathway via effector mechanisms associated with nuclear receptors, bile uptake, and efflux transporters. This study provides a scientific basis and strong evidence for the physiological structures of bionic livers prepared using 3D cultures. The systems and cultured liver tissues described here may serve as a better in vitro 3D culture platform and basic unit for varied applications, including drug development, hepatocyte polarity research, bioartificial liver bioreactor design, and tissue and organ construction for liver tissue engineering or cholestatic liver injury.

1. Introduction

In China, there is a high prevalence of hepatitis and other types of liver disease, leading to a high rate of liver failure. Approximately 500,000 new cases of liver failure are diagnosed annually with a mortality rate of 80% [1]. The main clinical presentations of liver failure include coagulation defects, jaundice, hepatic encephalopathy, and ascites. Though liver transplantation remains the ideal treatment, insufficient numbers of available viable organs are a limitation, highlighting the need for alternative treatments.

One proposed area for alternative treatment of liver failure is tissue engineering. Currently, the focus is to establish a 3D spatial composite of cells and biomaterials to generate living tissues that can be used for morphological, structural, and functional reconstruction of diseased tissues, thereby achieving permanent replacement [2]. In vitro 3D hepatocyte culture is the core of liver tissue engineering and involves factors such as cells, scaffolds, and three-dimensional topology. The liver itself is an extremely good hepatocyte culture reactor.

Recent studies using in vitro 3D hepatocyte culture include systems based on microcarriers [3], hollow fibers [4-6], decellularized liver scaffolds [7-9], spheroidal aggregates [10-13], cell sheets [14-16], microfluidic chips [17-20], and 3D bioprinted cultures [21-24]. Despite the advantages of increased hepatocyte viability compared to 2D cultures, such systems share a common shortcoming. They only achieve hepatocyte growth in a 3D environment that cannot completely recapitulate the fine 3D topological structure of hepatic plates. In particular, the utilized cells do not conform to the actual coculture system inside the liver and, thus, are not comparable to in vivo hepatocytes. In vivo hepatocyte structural and functional polarity includes three functional regions: the apical membrane (bile canaliculi), lateral membrane (hepatocyte face), and basal membrane (sinusoidal surface). The unique functions of the liver are dependent on hepatocyte polarity [25], which marks mature hepatocytes. The theory of hepatocyte polarity is based on studies of in vitro Madin-Darby canine kidney cells, in vivo zebrafish vasculature, fruit fly tracheal epithelial cells, and nematode excretory cells [26-28]. Results from the limited number of hepatocyte-based studies using conventional 3D culture do not mimic the physiological structure of the liver, including cellular polarity, and they do not establish suitable 3D controls. Therefore, improved models are needed for hepatocyte polarity-related studies.

In this study, we employed microfluidic technology to achieve a flexible combination of functional units, cell growth microenvironment construction, and easy operation. A microfluidic chip that better matches the human hepatic plate structure was designed using natural alginate hydrogels as a cellular scaffold. Human hepatocytes and endothelial cells were used to construct microliver tissues that mimic hepatic plate structures. Hepatocyte viability, function, and polarity were systematically compared. Transcriptome sequencing analysis was used to examine in-depth changes in hepatocyte polarity pathways. An improved 3D culture platform for in vitro 3D hepatocyte culture and polarity research based on these data would provide a microliver unit for tissue-engineered liver construction and allow bioartificial liver bioreactor design, drug metabolism evaluation, and liver pathophysiology research.

2. Materials and Methods

2.1. Microfluidic Chip Production

Validation of previous studies [17] confirmed that soft lithography using polydimethylsiloxane was the optimal process to prepare the chips. Figure 1(a) shows the design plan. Chips were manufactured at the Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences. A 1 cm stainless steel adapter (internal diameter: 1.2 mm, external diameter: 1.8 mm) (Suzhou Wenhao Microfluidic Technology Co., Ltd., China) was inserted into the chip injection port. The tips of seven disposable intravenous infusion needles (Jiangxi Lule Medical Equipment Group Co., Ltd., China) were cut and connected to the stainless steel adapter. The interface was sealed and fixed using WACKER gels (Wacker Chemie AG, Munich, Germany).

Figure 1

The overall experimental process. (a) Chip design and description: the main reaction channel of the chip measures 400 μm (W4) by 180 μm (D1) by 50 mm (L10). The inlet channels (A-G) are 1700 μm circular openings. The chip surface pores are connected to their respective inlet channels. The pore diameter φ is approximately 1200 μm. Other dimensions: L1 = 9967.12 μm, L2 = 8293.91 μm, L3 = 8467.12 μm, L4 = 328.87 μm, L5 = 8577.26 μm, L6 = 8532.61 μm, L7 = 300 μm, L8 = 8567.12 μm, L9 = 8467.12 μm, L11 = 70 mm, W2 = 100 μm, W3 = 200 μm, W5 = W6 = 12.3 mm, W7 = 25 mm, A1 = 30°, A2 = 60°, A3 = 90°, D2 = 2200 μm, D3 = 1000 μm, D4 = 3380 μm, and D5 = 2200 μm. (b) Three-dimensional culture formats used in this study: hepatic plate mimetic three-dimensional culture (i) and mixed three-dimensional culture (ii). (c) Hepatic plate-mimetic 3D coculture system schematic: laminar flow of the four solutions in the chip reaction channel (i) and overall experimental process (ii). (d) The mixed 3D coculture experimental process. (e) Images of the sandwich coculture method layers: bottom collagen coating (i), inoculated cells (ii), and upper collagen coating (iii) (all error bars: 100 μm).

2.2. Cell Culture, Hydrogel Solution Preparation, and Experimental Group Design

C3A human hepatocytes (CRL-10741, ATCC, USA) and EA.hy926 human endothelial cells (CRL-2922, ATCC) were inoculated into culture flasks with Dulbecco's modified Eagle medium (DMEM) basal media (Gibco Inc., Gaithersburg, MD, USA) containing 10% fetal bovine serum (FBS) (Gibco). Cells were cultured in a 5% carbon dioxide cell culture incubator at 37°C. When cells reached the logarithmic growth phase, they were trypsinized and mixed with alginate solution. The concentrations of the C3A and EA.hy926 cells in the alginate solution were 3 × 107 and 1 × 107 cells/mL, respectively. Table 1 shows the alginate hydrogel solution composition.

Table 1

Preparation of the alginate hydrogel solution.

Solution components	Concentration	Solution components	Concentration
Preparation of the alginate hydrogel solution
Alginate	0.7% g/mL	BSA	1% g/mL
NaCl	0.9% g/mL	HEPES	10 mmol/L
Buffer solution preparation
Dextran	10% g/mL	HEPES	10 mmol/L
NaCl	0.9% g/mL
Gelation solution preparation
Dextran	10% g/mL	NaCl	0.72% g/mL
BaCl2	20 mmol/L	HEPES	10 mmol/L

BSA: bovine serum albumin; HEPES: N-2-hydroxyethylpiperazine-N-2-ethane sulfonic acid.

Design of the experimental and control groups was as follows. The cell plating ratio was C3A : EA.hy926 (3 : 1), with triplicate wells for each group. Experimental group: hepatic plate-mimetic three-dimensional coculture (H) (Figure 1(b)(i)). Control group 1: mixed 3D coculture (M) (Figure 1(b)(ii)). Control group 2: sandwich coculture (S). Control group 3: mixed 2D coculture/plate coculture (P).

2.3. Preparation of Experimental Samples

2.3.1. Experimental Group (Hepatic Plate-Mimetic 3D Coculture)

C3A and EA.hy926 cells were resuspended in a sodium alginate solution (3 × 107 and 1 × 107 cells/mL, respectively). The buffer solution (10 μL/min), C3A cell suspension in sodium alginate (20 μL/min), EA.hy926 cell suspension in sodium alginate (10 μL/min), and gelation solution (66.7 μL/min) were successively injected into the microfluidic chip with a microinjection pump. Laminar flow conditions occurred when these four liquids converged at the main channel (Figure 1(c), i). The solutions moved through the chip as follows (from medial to lateral): hepatocyte solution, endothelial cell solution, buffer solution, and gelatin solution. Based on these principles, a microfluidics chip was used to form alginate hydrogel wires that encapsulated the hepatic plate structure. After the hydrogel was formed, it was introduced into a culture plate containing DMEM and 10% FBS culture medium (Figure 1(c), ii). To identify optimal flow rate conditions and examine whether cells formed a hepatic plate-like arrangement, we used different dyes (green, yellow, red, and blue dyes) to simulate laminar flow. We then identified optimal protocols and flow rates for introducing the four solutions into the microfluidic chip. Then, live cells were collected for dual fluorescent labeling using Calcein-AM (Dojindo Laboratories, Japan) to label C3A cells and Mito-Red (Dojindo Laboratories, Japan) to label EA.hy926 cells. Hepatic plate structure and hepatocyte/endothelial cell arrangement were observed using a Zeiss LSM880 laser confocal microscope with Airyscan (Carl Zeiss, Germany).

2.3.2. Control Group 1 (Mixed 3D Coculture)

C3A and EA.hy926 cells were mixed 3 : 1 and resuspended in sodium alginate solution at a concentration of 4 × 107 mixed cells/mL. The alginate solution containing mixed cells was injected into the chip through the most central channel. Cell-free sodium alginate solution was injected into the channel that was originally injected with sodium alginate solution and EA.hy926 cells (Figure 1(d)). The remaining steps were the same as the hepatic plate-mimetic 3D coculture group. Similarly, we collected live cells for dual fluorescent labeling using Calcein-AM to label C3A cells and Mito-Red to label EA.hy926 cells. The disorderly arrangement of hepatocytes and endothelial cells inside the hydrogel was observed using Zeiss LSM880 laser confocal microscope.

2.3.3. Control Group 2 (Sandwich Coculture)

(1) Collagen Coating

Rat tail type I collagen (2 mL at 5 mg/mL) was added to centrifuge tubes in an ice bath, and uniformly mixed with 6.9 mL sterile ultrapure water and 120 μL 0.1 mol/L NaOH. Subsequently, 1 mL 10x PBS was added and mixed to obtain the final 1 mg/mL collagen solution. The prepared collagen solution was immediately added to the culture plates and incubated at 37°C for 2 h to allow the collagen to solidify. Figure 1(e), i shows the collagen-coated bottom layer.

(2) Cell Inoculation. After incubation, 0.5 mL of DMEM basal media was added and preequilibrated for 30 min. C3A cells and EA.hy926 cells were digested using trypsin and mixed at a ratio of 3 : 1 prior to resuspension in complete culture medium. The cells were then inoculated into culture plates. The culture plates were carefully placed into the incubator and incubated for 1 h to allow complete cell adhesion (Figure 1(e), ii).

(3) Upper Collagen Layer. Excess culture medium was aspirated, and the prepared collagen solution was layered above the cultured cells. The plates were incubated for 1 h at 37°C to allow the collagen to solidify (Figure 1(e), iii). Then, the culture plates were removed, and excess collagen solution was aspirated. Finally, complete culture medium was added, and the plates were incubated at 37°C.

2.3.4. Control Group 3 (Mixed Two-Dimensional Coculture/Plate Culture)

Cultured C3A cells and EA.hy926 were digested using trypsin, resuspended in complete culture medium, and inoculated into culture plates.

2.4. Histomorphological Observations, Plotting of Growth Curves, Hepatocyte Viability, Function, Gene Expression Levels, and Other Dynamic Tests

2.4.1. Dynamic Observations of Changes in the Morphological Structure of Hepatocytes

An inverted microscope (CK-2, Olympus, Japan) was used to observe the growth of dynamic hepatocyte morphological structures from the experimental and control groups at experimental days 0, 1, 3, 5, 7, 9, 11, and 13.

2.4.2. Plotting of Hepatocyte Growth Curves

Samples from each group were collected on experimental days 1, 3, 5, 7, 9, 11, and 13. To remove the encapsulating hydrogel from the hepatic plate-mimetic and mixed 3D groups, the hydrogel was immersed in PBS containing alginate lyase (1 U/mL) for 1 min. After the cells contained inside the hydrogel were released, the cells were trypsinized and counted. Type I collagenase solution (200 U/mL) was added to the sandwich culture group to digest the rat tail collagen for 30 min. Cells were then digested with trypsin and counted. Routine trypsin digestion and counting was performed for the plate culture group. Samples from the four groups were collected at various time points. Cell counts were used to plot the cell growth curve, which was used to standardize cellular function and viability data and to ensure comparability.

2.4.3. Hepatocyte Viability Dynamic Testing via CCK-8

CCK-8 reagent (Dojindo Laboratories, Japan) (10% of culture medium volumes) was added to the culture plates and incubated for 2.5 h at 37°C. Samples with colorimetric changes were transferred into new plates. A microplate reader (Multiskan GO, Thermo Scientific, USA) was used to measure the absorbance at 450 nm. The sampling time points were experimental days 1, 3, 5, 7, 9, 11, and 13.

2.4.4. Dynamic Testing of Hepatocyte Function

Table 2 shows test markers for hepatocyte function, which were all measured using ELISA kits (). Details of the detection methods are found in the operating instructions for the various ELISA kits. The sampling time points were experimental days 1, 3, 5, 7, 9, 11, and 13.

Table 2

Table of liver function categories.

Liver function category	Marker
Detoxification	CYP3A4, CYP2D6, CYP1A2, DIA, GST
Ammonia metabolism and urea synthesis	NH3, urea
Glucose metabolism	GCK
Protein metabolism	ALB, TF, AAT, APOA1
Hormone metabolism	PXR/NR1I2, TT
Synthesis of coagulation factors	F2, F5, F7

DIA: diazepam; GST: glutathione-S-transferase; NH3: ammonia; GCK: glucokinase; ALB: albumin; TF: transferrin; AAT: α1-antitrypsin; APOA1: apolipoprotein A1; PXR/NR1I2: pregnane X receptor; F2: coagulation factor II; F5: coagulation factor V; F7: coagulation factor VII.

2.4.5. Dynamic Testing of Gene Expression Levels

Hepatocyte functional gene expression was measured using quantitative polymerase chain reaction (qPCR). The sampling time points were experimental days 1, 3, 5, 7, 9, 11, and 13. Cell processing was performed using the same methods as growth curve experiments. Total RNA was extracted using TRIzol reagent and QIAGEN RNA extraction kit, according to the manufacturer's specifications. For qPCR analysis, RNA was reverse transcribed to cDNA using HiScript II Q RT SuperMix (R223-01, Vazyme, USA). Real-time PCR was performed using LightCycler® 480 II Real-time PCR Instrument (Roche, Swiss), with QuantiFast® SYBR® Green PCR Master Mix (Qiagen, Germany). Reactions were incubated at 95°C for 5 min, followed by 40 cycles of 95°C for 10 s and 60°C for 30 s. Each sample was run in triplicate for analysis. shows the designed primers. β-Actin was used as a housekeeping gene control.

2.5. Dynamic Testing of Hepatocyte Polarity

Table 3 shows hepatocyte polarity markers. Gene expression, protein levels, and ultrastructural changes were evaluated by qPCR, immunofluorescence staining, and transmission electron microscopy, respectively. The sampling time points were experimental days 1, 7, and 13. TRIzol reagent and Qiagen RNA extraction kits were used for RNA extraction. shows the designed primers. F-actin was used as a housekeeping gene control. Immunofluorescence staining was carried out as follows: 4% paraformaldehyde fixation, paraffin embedding, sectioning into 5 μm sections, overnight incubation with primary antibodies at 4°C, and incubation with secondary antibodies at room temperature for 1 h. Nuclei were stained with 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI). shows detailed antibody information. Electron microscopy (H-7500, Hitachi, Japan) samples were processed as follows: 2.5% glutaraldehyde fixation, ultrathin sectioning, observation, and photography.

Table 3

Hepatocyte polarity markers.

Structural polarity
Basal membrane domain (liver sinusoidal face):NTCP (sodium-taurocholate cotransporting polypeptide): used to extract bile salts from liver sinusoids.
Lateral membrane domain (hepatocyte face):ZO-1: tight junction formation;occludin: tight junction formation;CD147/CE9 (cluster of differentiation 147/extracellular matrix metalloproteinase inducer): induces matrix metalloproteinase secretion.
Apical domain region (bile canaliculi face):MRP2/ABCC2 (multidrug resistance-associated protein): promotes bile canaliculi formation and structural integrity;DPPIV (dipeptidyl peptidase-IV): bile canaliculi-specific protein and regulation of extracellular bioactive peptide concentration.
Ultrastructure: cell-cell connections, bile canaliculi network, polarized Golgi apparatus

2.6. Transcriptome Sequencing Analysis and Validation

Total RNA was extracted using TRIzol (Invitrogen, USA) and Qiagen RNA extraction kits. RNA purity and quantification were evaluated using a NanoDrop ND-2000 spectrophotometer (Thermo Scientific). RNA integrity was assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies, USA). Samples with an RNA integrity number (RIN) ≥ 7 were further analyzed. Sequencing libraries were constructed using a TruSeq Stranded mRNA LT Sample Prep Kit (Illumina, USA). Libraries were sequenced on an Illumina HiSeq X Ten platform with 150 bp paired-end reads. Raw fastq files were processed using the NGS QC Toolkit (version 2.3.3, parameters: IlluQC_PRLL.pl N 5 -l 70 -s 20; TrimmingReads.pl -q 20; AmbiguityFiltering.pl -t5 -n 35). Adapter, poly-N, and low-quality reads were filtered from the raw data. Sequencing reads were then mapped to the human genome (GRCh38.p7) using hisat2 (version 2.0.5, parameters: --rna-strandness RF --fr). Fragments per kilobase million (FPKM) was calculated using cufflinks (version 2.2.1, parameters: -u --max-bundle-frags 2000000 --library-type fr-firststrand). Read pileups were quantified by htseq-count (version 0.6.0, parameters: -f bam -s reverse -r name). Differentially expressed genes (DEGs) were identified using the DESeq R package functions (estimateSizeFactors and nbinomTest). P values < 0.05 and fold change > 1.5 (or fold change < 0.667) were set as the differential expression threshold. Gene Ontology (GO) enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment of DEGs were analyzed using the hypergeometric distribution method in R. We performed Gene Ontology and KEGG pathway enrichment analysis using The Database for Annotation, Visualization and Integrated Discovery (DAVID) version 6.7 (https://david-d.ncifcrf.gov/home.jsp). Transcriptome sequencing data from the hepatic plate-mimetic 3D culture group and the mixed 3D culture group from experimental days 1, 7, and 13 were compared. Samples were screened based on KEGG-enriched pathways with significant differences to identify changes in hepatocyte polarity pathways. Relevant DEG expression was validated by qPCR ( shows the designed primers).

2.7. Statistical Methods

All data are expressed as mean ± standard deviation of at least three replicates. All statistical analyses were performed using GraphPad Prism 5 (La Jolla, CA, USA).

3. Results

3.1. Production of Microfluidic Chips, Synthesis of Hepatic Plate-Mimetic 3D Liver Tissue Hydrogels, and Validation of Hepatic Plate Mimetic Arrangement

Different liquids were successively microinjected into the microfluidic chip (Figures 2(a)–2(c)) to form a hydrogel-encapsulated hepatic plate structure. The hydrogel micelles were approximately 100 μm in diameter (Figures 2(d) and 2(e)). The microfluidic chip has an efficient hydrogel micelle production speed, with 5 mm hydrogel micelles containing 6 × 105 hepatocytes formed every minute, continuously. Laminar flow simulations with different color dyes (Figure 2(f)), actual liquids (Figure 2(g)), and real-time dual fluorescent labeling of live cells (Figure 2(h)) demonstrated the feasibility of the hepatic plate-mimetic 3D coculture protocol. The tested flow rate conditions all displayed laminar flow. In the hydrogel-encapsulated hepatic plate structures prepared using this experimental protocol, C3A cells in the center 2–3 rows were surrounded by EA.hy926 cells, showing a hepatic plate structure arrangement. The hydrogel formation speed at the chip outlet was approximately 5 m/min, with a deposition rate of 6 × 105 hepatocytes/minute.

Figure 2

Production of microfluidic chips, synthesis of hepatic plate-mimetic 3D liver tissue hydrogels, and validation hepatic plate mimetic arrangement. Actual chip (a), Microchannels inside the chip as imaged with a CK-2 microscope (Olympus, Japan) (b), chip after connecting to the tubing (c), external appearance of hydrogel micelles (d), microscopic observations of a hydrogel micelle (e), simulation of laminar flow distribution (f), laminar flow distribution of actual liquids in the chip (g), and real-time dual fluorescent labeling of live cells (h), showing the hepatic plate-mimetic 3D culture group (left) and the mixed 3D culture group (right, green fluorescent Calcein-AM-labeled C3A cells, red fluorescent Mito-Red-labeled EA.hy926 cells). All scale bars represent 100 μm.

3.2. Morphological Observations, Plotting of Cell Growth Curves, and Dynamic Testing of Hepatocyte Viability

As shown in Figure 3(a), in the hepatic plate-mimetic group (Group H) and the mixed 3D group (Group M), cocultured cells were proliferated on days 3–5 and gradually formed a microliver cord (micro-liver tissue), which was encapsulated by alginate hydrogels. After that, microliver morphology was maintained. In the mixed 2D/plate group (Group P), the cocultured cells proliferated rapidly. At approximately 3–5 days, the plates were confluent, with overlapping cell layers. Gradually, the cells became wrinkled, cell-cell gaps increased, and cells started to detach. Large numbers of floating dead cells and cell debris appeared in the cell culture medium. Compared with Group P, the sandwich group (Group S) showed reduced cell proliferation. The cells reached confluency on days 5–7, with regular morphology. Subsequently, cell proliferation was reduced by contact inhibition. Only small numbers of cells were detached by the end of the observation period.

Figure 3

Morphological observations, cell growth curves, and hepatocyte viability. (a) Morphological observations in each group during liver tissue generation (all scale bars: 100 μm). (b) Cell growth curves and dynamic testing of hepatocyte viability: cell growth curves (i) and viability trends of the four groups (ii). The data are presented as the mean ± SD. Two-way ANOVA: H vs. P: ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001; H vs. S: ▲P < 0.05, ▲▲P < 0.01, ▲▲▲P < 0.001; H vs. M: ■P < 0.05, ■■P < 0.01, ■■■P < 0.001. H: hepatic plate-mimetic group; M: mixed 3D group; S: sandwich group; P: mixed 2D/plate group.

From the cellular growth curves (Figure 3(b), i), cellular growth in the four groups was generally consistent with cell morphology; i.e., the hydrogel-encapsulated cells in Groups H and M exhibited lower proliferation trends. The overall cell proliferation from Group H was lower than that of Group M. Cells in Group P proliferated fastest among the four groups. Peak proliferation occurred on day 9. Subsequently, there was a trend towards a decrease in cell number. Due to the constraints of the rat tail collagen matrix, cell proliferation in Group S was slower than that of Group P, but faster than those of Groups H and M.

Cell viability of the four groups except Group P initially increased, followed by a decrease (Figure 3(b), ii, ). Groups H and M showed peak cell viability on day 7. Group S reached its maximum viability on day 9. The overall cell viability for Group P continuously decreased. Intergroup comparisons indicated that Group H cell viability was significantly higher than that of Group M at all tested time points, except days 5 and 7 (Figure 3(b), ii, ). Group H showed significantly better cell viability values than Group P apart from day 1, and Group S except at days 9 and 11 (Figure 3(b), ii, ).

Overall, the order of cell and tissue morphology maintenance was H> M> S> P and that of cell growth speed and apoptosis was P> S> M> H. Based on these results, cell viability of Group H was higher than that of all other tested groups.

3.3. Liver Function of Tissues Generated Using the Various Platforms

To evaluate hepatic plate-mimetic 3D cultured tissue function, we tested clinically relevant liver functions, such as detoxification, ammonia metabolism and urea synthesis, bilirubin metabolism, glucose metabolism, protein metabolism, hormone metabolism, and synthesis of coagulation factors (as shown in Table 2) using ELISA kits (). As shown in Figure 4 and , the abundance of the functional proteins (CYP3A4, CYP2D6, CYP1A2, GST, DIA metabolism, ammonia metabolism, urea, TT metabolism, PXR, ALB, AAT, APOA1, TF, GCK, F2, F5, and F7) in the liver demonstrate the advantages of hepatic plate-mimetic 3D culture and mixed 3D culture (Groups H and M). As culture duration increased, most liver function markers also increased. Conversely, the abundance of these markers in Groups S and P either remained unchanged or continuously decreased. Among these markers, ALB, AAT, TF synthesis, CYP1A2, CYP2D6, GST, DIA metabolism, ammonia metabolism, and TT metabolism were significantly higher in Group H than those in other groups. There were no significant differences or slightly poorer abundance of the other functional proteins (CYP3A4, urea, PXR, APOA1, GCK, F2, F5, and F7) in Group H compared with Group M, but these were all generally higher than levels in Groups S and P. Notably, the liver function gene expression levels (Figure 5 and ) corresponded to protein levels as culture duration increased. Among these genes, CYP2D6, CYP1A2, CPS1, F2, F5, and F7 expression levels in Group H were significantly higher than those in the other groups. CYP3A4 expression in Group H became gradually higher than in Groups S and P as culture duration increased but showed varying increase and decrease when compared to gene expression in Group M. In Groups H and M, GSTA1 and PXR expression showed continuous, nonsignificant increases. By day 13, GSTA1 was significantly higher in Group H than that in Groups S and P, whereas PXR was significantly higher in Group H on days 7 and 13 than those in Groups S and P.

Figure 4

Dynamic testing of hepatocyte function: comparison of CYP1A2 synthesis (a), CYP2A6 synthesis (b), CYP3A4 synthesis (c), glutathione-S-transferase (GST) synthesis (d), diazepam (DIA) synthesis (e); ammonia (NH3) metabolism (f); urea synthesis (g), testosterone (TT) metabolism (h), and pregnane X receptor (PXR) synthesis (i), albumin (ALB) synthesis (j), α1-antitrypsin (AAT) synthesis (k), apolipoprotein A1 (APOA1) synthesis (l), transferrin (TF) synthesis (m), glucokinase (GCK) synthesis (n), Factor II (F2) synthesis (o), Factor V (F5) synthesis (P), and Factor VII (F7) synthesis (q). Comparison of the data are presented as mean ± SD. Two-way ANOVA: H vs. P: ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001; H vs. S: ▲P < 0.05, ▲▲P < 0.01, ▲▲▲P < 0.001; H vs. M: ■P < 0.05, ■■P < 0.01, ■■■P < 0.001. H: hepatic plate-mimetic group; M: mixed 3D group; S: sandwich group; P: mixed 2D/plate group.

Figure 5

Quantification of hepatocyte functional gene expression levels: comparison of gene expression associated with liver detoxification and metabolism (a–f), ammonia metabolism and clearance (g, h), protein metabolism (i–l), and coagulation function (m–o). The data are presented as mean ± SD. Two-way ANOVA: comparison between H and other groups: ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001. H: hepatic plate-mimetic group; M: mixed 3D group; S: sandwich group; P: mixed 2D/plate group.

3.4. Testing of Hepatocyte Polarity

3.4.1. Comparison of Gene Expression Levels of Hepatocyte Structural Polarity Proteins

As shown in Figure 6(a) and , as culture duration increased, Occludin, DPPIV, and NTCP expression in Group H showed statistically higher levels than all other groups at days 7 and 13. CD147 expression in Group H also exhibited the significantly highest levels, with the exception of day 13, where expression in Group H was slightly lower than that of Group P. The overall expression trend of the ZO-1 gene expression in Group H was similar to that of MRP2. Expression levels of both genes were higher in Group H than those of all other groups at all time points except for day 13 compared with Group M. The differences of ZO-1 expression were statistically significant on days 7 and 13.

Figure 6

Comparison of hepatocyte polarity. (a) Comparison of gene expression levels of hepatocyte structural polarity proteins. The data are presented as the mean ± SD. Two-way ANOVA: comparison between H and other groups: ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001. (b) Immunofluorescence staining of hepatocyte polarity proteins: staining of occludin (occludin: green; cell nucleus: blue) (i), staining of ZO-1 (ZO-1: green; cell nucleus: blue) (ii), staining of MRP2 (MRP2: red; cell nucleus: blue) (iii); staining of CD147 (CD147: red; cell nucleus: blue) (iv). All scale bars: 10 μm. (c) Comparison of cultured hepatocyte ultrastructure: red rectangles represent magnified sites of every group at various time points, with the corresponding magnified images to the right of the original image. Red letters represent the following: H: hepatocyte; EC: endothelial cell; A1: alginate hydrogel; C: collagen; N: cell nucleus; BC: bile canaliculi; TJ: tight junction; Mi: microvilli; G: Golgi bodies; P: phagosome; AP: autophagosome. Scale bar: the scale bars for the low-magnification images of Groups H and M on days 1, 7, and 13 represent 2 μm. The scale bar for high-magnification images is 500 nm, except for the right image in M13, where the scale bar is 2 μm. The scale bars for the left images for Groups S and P on day 1 are 5 μm. The scale bars for the right images for Groups S and P on day 1 are 2 μm. The scale bar for panel S7 is 2 μm. The scale bars for P7, S13, and P13 are 5 μm. H: hepatic plate-mimetic group; M: mixed 3D group; S: sandwich group; P: mixed 2D/plate group.

3.4.2. Comparison of Immunofluorescence Staining of Hepatocyte Structural Polarity Proteins

To determine if gene expression quantification reflected protein abundance in the hepatic plates, we analyzed protein expression of the genes analyzed in Figure 6(a). Figure 6(b) shows that occludin, ZO-1, MRP2, and CD147 staining in Group H was much more obvious and continuous than those in other groups on days 7 and 13 as the culture duration increased.

3.4.3. Comparison of Hepatocyte Structural Polarity Ultrastructure by Electron Microscopy

Bile canaliculi, which constitute microchannels formed from local membrane invaginations between adjacent cells, are the representative of polarized hepatocytes. Figure 6(c) shows that Group H formed more mature bile canaliculi as culture duration increased, which appeared as tight connections and microvilli protrusions into the lumen. In addition, Golgi bodies that displayed polar distribution (located between the bile canaliculi and cell nuclei) were present. Although bile canaliculi were also formed in Group M, they were less mature. Moreover, numerous phagocytotic vesicles were formed on day 13, showing poor cell health. Cells from Groups S and P did not have intact cell structures because of apoptosis and necrosis which occurred by this time point.

3.5. Transcriptome Sequencing and Analysis

To determine why the hepatic plate-mimetic 3D cultures afford better bionic function, we performed transcriptome sequencing of Groups H and M at different time points. Figure 7(a) depicts DEGs, and shows the corresponding pathway enrichment results.

Figure 7

Transcriptome sequencing analysis outcomes. (a) Volcano plots of differentially expressed genes with an absolute fold change > 1.5 (or fold change < 0.667) at different time points in the hepatic plate-mimetic 3D culture group (H) and the mixed 3D culture group (M). Gray and blue colors denote genes without significant differences. Red and green indicate significant upregulated and downregulated genes, respectively. (b) KEGG enrichment bubble charts of the top twenty differentially upregulated genes. Pathways with >1.5 enrichment at different time points in the hepatic plate-mimetic 3D culture group (H) and the mixed 3D culture group (M). H1/M1, Day 1; H7/M7, Day 7; H13/M13, Day 13. (c) KEGG enrichment of the bile secretion pathway for differentially upregulated genes between the hepatic plate-mimetic 3D culture group (H) and the mixed 3D culture group (M) on day 7 with >1.5 enrichment. Red denotes a significant difference; yellow denotes a nonsignificant difference but enrichment >1.5; orange denotes a nonsignificant difference but enrichment > 1.3 or ≤1.5; purple denotes a nonsignificant difference but enrichment > 1 or ≤1.3. BA: bile acids; PL: phospholipids. (d) Differentially expressed genes associated with bile secretion selected from sequencing results and KEGG pathway analysis were analyzed by qPCR. The data represent mean ± SD. Two-way ANOVA: comparison between H and M: ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001. H: hepatic plate-mimetic group; M: mixed 3D group. (e) qPCR validation and correlation of expressed genes associated with bile secretion after sequencing and KEGG pathway analysis.

At day 7, Group H showed significant enrichment of genes associated with the bile secretion pathway (), which is unique to hepatocytes. This was not observed in Group M. Similarly, the KEGG enrichment results of differentially upregulated genes between Groups H and M also showed significant enrichment of transcripts associated with the bile secretion pathway on day 7 (Figure 7(b)).

Figure 7(c) shows a schematic of the bile secretion pathway identified from the enriched KEGG map. Based on this information, we further selected the bile secretion pathway-associated genes for farnesoid X receptor (FXR), small heterodimer partner receptor (SHP), retinoid X receptor alpha (RXRα), aquaporin 8 (AQP8), ABCG8, BSEP, MRP2, multidrug resistance protein 1 (MDR1), MDR3, OATs, organic solute transporter (OST-α), and SULT2A1 for qPCR validation. Validation by qPCR showed that the gene expression levels of FXR, SHP, RXRα, AQP8, MRP2, MDR1, MDR3, and OST-α were generally consistent with the sequencing results. Further, these two datasets were significantly correlated (Figures 7(d) and 7(e) and ).

4. Discussion

Liver failure is a complex disease which is associated with a high mortality rate. Though liver transplantation is the ideal treatment, numerous limitations, including scarcity of organs for transplantation, the high cost, and the lifelong immunosuppression, restrict its use in patients. Liver tissue engineering has emerged as a potential solution for liver failure diagnoses. In vitro 3D hepatocyte culture is the core of liver tissue engineering. The principle of 3D culture is to simulate the growth environment of hepatocytes inside the body. Therefore, it is extremely important to simulate the physiological structure and liver microenvironment to the greatest degree possible. This bionic principle has stimulated the development of various methodologies to optimize in vitro 3D hepatocyte culture models [3-24]. Among these technologies, microfluidic chip 3D cultures can accurately control the microenvironment for hepatocyte culture, achieve unique “chip organ” advantages, and facilitate the construction of microliver tissues. Research in this area has a marked potential to provide a scientific basis for organ construction.

Based on the above theories, we selected the microfluidic chip from existing research techniques as a technical basis for further development. We then designed microfluidic chips to achieve a flexible combination of various functional units at the microscale, accurate construction of a microenvironment for cell growth, and construction of a microliver to create a “chip organ.” The advantages of this method are twofold: simple, gentle operation conditions and a chip design that better matches the human hepatic plate structure. C3A is selected as the hepatocyte in the extracorporeal liver assist device (ELAD), which has proven to be effective in liver support and biocompatible in patients in clinical trials [29]. The EA.hy926 cell line has been widely used in coculture models related to liver tissue engineering [30, 31]. Additionally, the two cell lines are relatively easy to passage. Therefore, we used C3A and EA.hy926 cells for hepatic plate-like coculture, with natural alginate hydrogels encapsulating cells to act as cell growth scaffolds. Using these components, we successfully constructed human hepatic plate-mimetic microliver tissues. This microliver tissue was developed for clinical application. The conclusions drawn from experiments using these cells can directly guide clinical research. The constructed microliver plate tissues can maximally reconstruct the 3D hepatocyte topological structure. The cell growth scaffolds of these tissues consist of alginate hydrogels, which have a hydrophilic interior and good biocompatibility. Furthermore, the hydrogel itself is a porous matrix permeable to micromolecules, such as gases, glucose, and albumin, but impermeable to macromolecules, such as IgG. Thus, it exhibits some immunoisolation effects [32]. Considering that the diameter of the hydrogel is only 100 μm, there are no limitations regarding nutrient secretion, gas transport, or metabolite excretion between cells in the generated liver tissues and the culture medium.

As the core of in vitro 3D hepatocyte culture, the hepatocyte viability and functions attract the most attention. In this study, we found that the group order of cell and tissue morphology maintenance was H> M> S> P and that of cell growth speed and apoptosis was P> S> M> H. The cell viability of Group H was higher than those of other groups. Even though the various liver function markers showed differences at gene and protein levels, the conclusions obtained from comparison of the two evaluation systems were generally consistent. Three-dimensional cultures will gradually demonstrate the functional improvement and maintenance advantages of most liver function markers as the culture duration increases. In addition, the expression of some liver function genes under hepatic plate-mimetic 3D culture conditions will be significantly higher than those in mixed three-dimensional cultures, particularly in drug-metabolizing enzyme genes, ALB, and ammonia metabolism. These markers are the most clinically important for assessing liver function and are extremely important markers for assessing bioartificial livers and drug metabolism testing. These results also explained the possible differences in liver function comparisons, because hepatocyte proliferation and differentiation/functional status are mutually antagonistic under in vitro culture conditions. For example, after partial hepatectomy, hepatocytes enter the cell cycle, and genes associated with liver metabolism are temporarily inhibited until hepatocyte division is completed [33]. This indicates that hepatocytes first proliferate prior to adopting liver function under in vivo physiological conditions. Therefore, in vitro culture conditions should conform to this pattern.

As hepatocyte function is highly dependent on its polarity, changes in hepatocyte polarity are often observed in many liver diseases, such as hepatitis, liver cancer, and cholestasis. Changes in hepatocyte polarity are often accompanied by defects in the formation of bile canaliculi networks and cholestasis. The occurrence of tight junction protein mutations in hepatocytes leads to defects in bile canaliculi formation, disaggregation of epithelial cells, and disorders in liver structure [34]. The liver function results obtained in the present study indicate that the Group H cultures exhibited significant improvement in some important liver function assessment markers than the Group M cultures. Therefore, combined with the basic theory of hepatocyte polarization, the hepatic plate-mimetic 3D culture model appears more beneficial for the reconstruction and maintenance of hepatocyte polarity and can further promote the expression and maintenance of liver function. In turn, the gene expression levels of hepatocyte polarity indicated that as culture duration increased, the protein levels of occludin, DPPIV, NTCP, and CD147 in Group H were significantly higher than those in other groups. In addition, immunofluorescence staining showed that occludin, ZO-1, MRP2, and CD147 staining in Group H was much more obvious and continuous than those in other groups on days 7 and 13 as the culture duration increased. Consistent with these results, Group H demonstrated more mature ultrastructure as culture duration increased, including the presence of polar-distributed Golgi bodies. In comparison, Group M exhibited fewer and less mature features, whereas Group S and P cells were largely nonviable. These results all showed that the hepatic plate-mimetic 3D culture model is better than other culture systems at promoting hepatocyte polarity reconstruction and maintenance.

To further explore the potential mechanisms accounting for improved hepatocyte polarity reconstruction, we performed transcriptome sequencing. We found that the hepatic plate-mimetic 3D culture procedure could influence the hepatocyte-unique bile secretion pathway. The regulatory mechanisms of this pathway may involve the regulation of nuclear receptors, FXR, SHP, and RXRα, to regulate bile uptake. Bile promotes lipid digestion and absorption, absorption of lipid-soluble vitamins, and dissolution of cholesterol [35, 36]. In addition, the bile pathway is important for clearing toxins, carcinogens, drugs, and drug metabolites (exogenous substances) [37]. The bile pathway also constitutes an important secretory pathway for endocrine substances and endogenous metabolites, such as cholesterol, bilirubin, and hormones. Generally, once bile is synthesized by the liver, it will be secreted by the bile canaliculi located between hepatocytes, flowing through the hepatic ducts into the biliary system. Blockage of bile secretion leads to cholestasis. This causes toxic effects on hepatocytes, resulting in cholestatic liver disease. The clinical presentations of cholestatic liver disease include jaundice, pruritus, deep-colored urine, pale stools, and xanthelasma. Because our findings point to the bile secretion pathway as a potential mechanism for increased hepatocyte polarity, the use of 3D hepatocyte cultures for liver bioengineering has high clinical significance.

This study had some limitations that require further improvement and optimization, such as increasing the types of cocultured nonparenchymal cells. Furthermore, future experiments should make use of primary human hepatocytes for experiments. The extracellular matrix component, alginate hydrogel, could also be substituted with a liver decellularized Matrigel that could better mimic the liver. Further analyses should supplement and improve the fluorescence staining results, 3D reconstruction images of hepatocyte polarity proteins, and validation of bile secretion pathway regulatory genes. RNA interference would enable in-depth analysis of relevant mechanisms, and further optimization and scaling-up of liver lobule tissue mimetics based on the hepatic plate mimetic could be used to facilitate the practical application of hepatic plate mimetic tissues.

5. Conclusion

In summary, the hepatic plate-mimetic 3D culture system we developed exhibits multiple advantages compared to standard 3D mixed culture systems. Specifically, we identified that the hepatic plate-mimetic 3D culture could (1) form microhepatic plate structures and maintain better tissue morphology, resulting in better cell viability compared to the other tested groups; (2) exhibit significant advantages with regard to the expression of some drug-metabolizing enzymes, albumin, ammonia metabolism, and other clinically important liver function markers; (3) reconstruct and maintain hepatocyte polarity, which was observed in the expression of membrane proteins in the lateral and apical domain (bile canaliculi region) in hepatocytes, along with formation of more mature bile canaliculi; and (4) promote changes in the bile secretion pathway, which may occur through regulation of crucial genes such as those encoding FXR, SHP, RXRα, MRP2, MDR1, MDR3, NTCP, OST-α, and AQP8 (Table 4). However, the specific effector mechanisms involved require further experimental validation. These conclusions provide a scientific basis and strong evidence for the generation of physiological structures of bionic livers under 3D cultures. The 3D hepatocyte culture platform established herein may serve as a better in vitro hepatocyte culture platform and may improve the research and development of drug metabolism assays, mechanistic research on hepatocyte polarization, bioartificial liver bioreactor design, tissue and organ construction in liver tissue engineering, and research on cholestatic liver injury.

Table 4

Comparison between the four culture systems.

	P	S	M	H
Type of cells	C3A human hepatocytes and EA.hy926 human endothelial cells
Cell arrangement	Disorderly arrangement			Hepatic plate-like arrangement
Culture system	Mixed two-dimensional coculture/plate culture	Sandwich coculture	Mixed 3D coculture (conventional 3D culture)	Hepatic plate-mimetic 3D coculture
Cell and tissue morphology maintenance	H> M> S> P
Cell growth speed and apoptosis	P> S> M> H
Cell viability	Group H was higher than all the other groups.
Hepatocyte function and the gene expression levels	Group H exhibited significant advantages with regard to the expression of some drug-metabolizing enzymes, albumin, ammonia metabolism, and other clinically important liver function markers.
Gene expression levels of hepatocyte polarity proteins	As culture duration increased, the protein levels of occludin, DPPIV, NTCP, and CD147 in Group H were significantly higher than those in other groups.
Immunofluorescence staining of hepatocyte structural polarity proteins	Occludin, ZO-1, MRP2, and CD147 staining in Group H was much more obvious and continuous than those in other groups on days 7 and 13 as the culture duration increased.
Hepatocyte polarity ultrastructure	Group H promoted the formation of more mature bile canaliculi.
Transcriptome sequencing and analysis	Group H promoted changes in the bile secretion pathway.

H: hepatic plate-mimetic group; M: mixed 3D group; S: sandwich group; P: mixed 2D/plate group.