Mouse primary hepatocytes (MPH) were obtained from 4-week-old C57 mice using a previously described type I collagenase liver perfusion protocol (19,25-30). 293-Derived 293pTP cells were previously described (25,31). The human colon cancer HCT116 cell line were obtained from the American Type Culture Collection. The cells were maintained in complete Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% FBS (Lonsa Science SRL), 100 U penicillin and 100 μg streptomycin at 37°C in 5% CO₂. Levofloxacin, azithromycin and paracetamol were purchased from Sigma-Aldrich; Merck KGaA. Unless indicated otherwise, all other reagents were purchased from Thermo Fisher Scientific, Inc. or Sigma-Aldrich; Merck KGaA.

The use and care of animals in the present study was approved by the Research and Experimental Animal Use Ethics Committee of Chongqing Medical University (Chongqing, China; permit no. SCXK(YU)20070001). All animal experiments were performed in accordance with US National Institutes of Health Guide for the Care and Use of Laboratory Animals (32). The 4-week-old C57BL/6 male mice were obtained from the Animal Resource Center of Chongqing Medical University.

Anesthesia was performed using intraperitoneal injection of 3% sodium pentobarbital at a dose of 50 mg/kg. Harvesting of mouse liver tissue was performed according to a modified liver perfusion protocol (33-37). Specifically, following anesthesia of the mice, following aseptic techniques, an incision was made in upper-middle abdomen across the abdominal and chest cavities to expose the liver and heart. Following the blockade of the right heart circulation, perfusion of ~15 ml cold sterile PBS was rapidly performed from the left ventricular until the liver turned pale. Concomitantly, the rapid intra-cardiac perfusion led to acute cardiac arrest and mortality of the mice, which was additionally confirmed by cervical dislocation. Mice in the control non-perfusion group were sacrificed with CO₂, followed by cervical dislocation.

The liver was resected and rinsed in PBS twice in 10-cm cell culture dishes, and then cut into small tissue pieces with ophthalmic scissors, followed by passing the minced liver tissue through 80-mesh (sift80) or 200-mesh (sift200) cell strainer/filter. The recovered microsphere tissue pieces were washed with sterile PBS by low speed centrifugation [500 × g for 5 min at room temperature (RT)], and immediately used for the in vitro culture assays as described subsequently.

To establish the LMTC model, the recovered liver microsphere tissue pieces were cultured in 24-well plates with various concentrations of FBS and/or bovine calf serum (BCS; Sijiqing; Zhejiang Tianhang Biotechnology Co., Ltd.) containing 100 U penicillin and 100 μg streptomycin at 37°C in 5% CO₂. Medium was changed daily for the first 3 days, and then changed every other day.

Recombinant adenoviruses were generated by using the AdEasy technology and amplification as described previously (38-40). The Recombinant adenovirus Ad-BMP9 was previously characterized (41-45). Ad-BMP9 also co-expresses enhanced GFP. An analogous adenovirus expressing GFP (Ad-GFP) was used as a mock virus control. For all adenoviral infections, polybrene (4-8 μg/ml) was added to potentiate infection efficiency as described previously (46).

The preparations for BMP9 conditioned medium were performed as previously described (45). Briefly, subconfluent HCT116 cells were infected with the optimal titer (MOI =50) of Ad-BMP9. At 24 h after infection, the culture was changed to serum-free Opti-MEM media (Thermo Fisher Scientific, Inc.). The media were collected every 12 h for 4 times consecutively. The pooled BMP9 conditioned medium was centrifuged at 500 × g for 10 min at RT to remove any cell debris, aliquoted and stored at −80°C. The control conditioned medium was also prepared in the same fashion from Ad-GFP infected HCT116 cells (45,47).

The recovered liver tissue was rinsed with PBS, fixed with 4% paraformaldehyde for 30 min at RT, and subjected to H&E staining as described previously (48-50). The fixed tissue was also stained with Hoechst33258 (10 μg/ml Hoechst 33258 in PBS) for 5 min at RT, examined and recorded under a fluorescence microscope (magnification, ×200).

The IHC and IF staining of the liver microsphere tissue were performed as described previously (51-53). The recovered liver tissue was fixed with 4% paraformaldehyde for 30 min at RT, followed by antigen retrieval and immunostaining with anti-proliferating cell nuclear antigen (PCNA; 1:100-1:200; cat. no. 13110; Cell Signaling Technology, Inc.) or anti-Myc proto-oncogene protein (c-Myc; 1:100-1:200; cat. no. ab39688; Abcam) antibodies. Control rabbit IgG (1:200; cat. no. 011-000-003; Jackson ImmunoResearch Laboratories, Inc.) was used as a negative control.

The ICG uptake/release assay was performed as described previously (30). Briefly, recovered liver tissue and cells were washed with PBS and incubated with ICG (1 mg/ml in complete DMEM) at 37°C for 30 min, followed by two washes with PBS. For ICG release detection, the ICG-containing medium was replaced with 2% BCS DMEM, and the tissue and cells were incubated for an additional 3 h at 37°C in a cell culture incubator. ICG uptake/release was observed and recorded under a bright field microscope (magnification, ×200).

PAS staining was performed as described previously (19,27,28,30). Briefly, recovered liver tissue was fixed with 4% paraformaldehyde for 30 min at RT, followed by staining with 0.5% periodic acid solution for 5 min at RT. Following rinsing in distilled water for 3 min, the tissues were incubated in the Schiff's reagent for 15 min, followed by thorough rinsing with tap water. Cell staining was recorded under a bright field microscope (magnification, ×200).

Total RNA from both tissue and cells was isolated by using TRIzol reagent (CoWin Biosciences) according to the manufacturer's protocol. The perfused mouse liver tissues from normal 6-week-old C57BL/6 mice (n=5 males/group) and drug or PBS-treated 6-week-old C57BL/6 mice (n=5 males/group) were dissected out and homogenized in the TRIzol reagent. The recovered liver tissue and cells were lysed in TRIzol reagent. Total RNA was extracted and subjected to reverse transcription reactions with hexamer and M-MuLV reverse transcriptase (New England Biolabs, Inc.). The cDNA products were additionally diluted and used as PCR templates. The gene-specific PCR primers (Table SI) were designed using Primer3 Plus (http://www.bioinformatics.nl/cgi-bin/primer-3plus/primer3plus.cgi). TqPCR was performed using SYBR Green-based TqPCR analysis on a CFX-Connectunit system (Bio-Rad Laboratories, Inc.), as described previously (54). TqPCR reactions were performed in triplicate. GAPDH was used as the reference gene. Quantification of gene expression was performed using the 2^−ΔΔCq method (55).

Cell proliferation was assessed using the Premix WST-1 Cell Proliferation Assay System (Clontech Laboratories, Inc.), as described previously (56-59). The MPH seeded in 96-well plates at 6,000 cells/well were treated with levofloxacin (at 1, 5 and 25 μM), azithromycin (at 25, 125 and 625 nM), paracetamol (at 20, 100 and 500 μM) or DMSO for 24, 48 or 72 h. The Premix WST-1 Reagent was added to each well, followed by incubation at 37°C for 60 min and reading at 440 nm using the EL800 microplate reader (BioTek Instruments, Inc.). Each assay was performed in triplicate.

All quantitative experiments were performed in triplicate and/or repeated 3 times. Data are presented as mean ± standard deviation. Significant differences between groups were determined using a one-way analysis of variance followed by a Least Significant Difference post hoc test. P<0.05 was considered to indicate a statistically significant difference.

Liver tissue is rich in blood cells and lipofuscin, pyridine (NADPH) and flavin coenzymes, which are common causes of high auto-fluorescence. To minimize the cytotoxicity and auto-fluorescence caused by blood cell disintegration, left ventricular perfusion was performed to remove intrahepatic blood in the anesthetized mice (Fig. 1A-a). The liver tissue was prepared in different sizes by sifting through sift80 or sift200 filter strainers, and the microsphere tissue preparations were cultured in different concentrations of FBS or BCS-containing DMEM (Fig. 1A-b). The perfused liver tissue culture exhibited significantly decreased auto-fluorescence at 24 and 72 h after culturing, compared with culture established with tissue from the non-perfusion group (Fig. 1A-c).

When the liver microsphere tissue sifted with the sift80 strainer was cultured with 0, 2 or 10% FBS or BCS DMEM medium, followed by medium change every day for the first 3 days, and then every other day for the rest of the study period, it was identified that the microsphere tissue was in the healthiest state in 2% BCS medium (Fig. 1B-a).

It was also identified that non-sifted liver tissue pieces were mostly suspended in 2% BCS medium, and the sift80 tissue was mostly attached to the bottom of the culture plates and grew healthily for up to 14 days in 2% BCS medium (Fig. 1B-b). Notably, the sift200 microsphere tissue was demonstrated to grow somewhere in between that of the non-sifted tissue and sift80 microsphere tissue at day 7, but significantly deteriorated at day 14 (Fig. 1B-b). It is noteworthy that, although the LMTC culture could be passaged 1 to 2 times, the percentage of viable cells that survived the passages was rather low, at <10% (data not shown). Collectively, these results indicated that the optimal LMTC culturing conditions should include preparation of the mouse liver tissue with 80-mesh sift strainer and culturing of the sift80 microsphere tissue in 2% BCS/DMEM medium with daily medium change for the first 72 h, and then once every 2 days thereafter.

The long-term survival of the LMTC model was additionally analyzed by culturing sift80 microspheres in 2% BCS/DMEM for up to 14 days. H&E staining of the sift80 microsphere tissue in 2% BCS/DMEM was performed and it was identified that the cellularity was preserved at the 3 time-points examined (days 3, 7 and 14), although the microsphere tissue scaffolds became loose and disintegrated to a certain extent (Fig. 2A-a). Accordingly, the Hoechst 33258 staining assay indicated that the majority of the cell nuclei were readily visible without any apparent fragmentation (Fig. 2A-b).

It was also identified that the sift80 microsphere tissue exhibited strong positive staining for PCNA at days 3 and 7, with a decreased level of staining at day 14 (Fig. 2A-c, top row). Similarly, high levels of c-Myc expression were observed at days 3 and 7, although the levels had significantly decreased by day 14 (Fig. 2A-c, bottom row). These IHC staining results were further confirmed by IF staining. As liver tissue exhibits auto-fluorescence, Sudan Black B was employed to effectively suppress background fluorescence (Fig. 2B-a). IF staining with PCNA and c-Myc antibodies indicated that the sift80 microspheres markedly expressed both PCNA and c-Myc at day 7, but that the levels were rather weak at day 14. Taken together, these results demonstrate that the sift80 microspheres may be healthily maintained for at least 7 days, and potentially for up to 14 days.

To determine whether the cultured microsphere tissue exhibited normal liver functions, ICG uptake/release and Periodic Acid-Schiff (PAS) staining of glycogen storage assays were performed, which are liver-specific functions (19,28,60). It was identified that the sift80 microsphere tissue cultured at days 3 and 7, and to a lesser extent at day 14, was able to uptake ICG effectively (Fig. 3A-a). Conversely, it was demonstrated that the ICG was effectively released from the cultured sift80 microsphere tissue at the 3 time points examined (Fig. 3A-b). Furthermore, marked PAS staining of the cultured sift80 microsphere tissues indicated that the hepatocytes were viable and healthy, particularly at days 3 and 7 (Fig. 3B). These results suggested that the sift80 microsphere tissue may retain normal hepatic functions under the in vitro culture conditions for up to 14 days.

To additionally determine whether the in vitro culture would affect hepatic gene expression pattern, the expression profiles of a panel of eight hepatocyte-specific genes were compared (19,27,28,30,61-63) in the cultured microspheres and that of the freshly-isolated mouse liver tissue and the MPH. Using TqPCR analysis, it was identified that cytochrome P450 family 1 subfamily Amember 2, cytochrome P450 family 3 subfamily Amember 4, microsomal glutathione S-transferase 1 and albumin (Alb), and to a lesser extent cytochrome P450 2A5 and Maob, were highly expressed in the sift80 microsphere tissue cultured at days 3, 7 and 14, and in the freshly-prepared mouse liver tissue and in the MPH, although their expression in the sift80 microsphere tissue samples decreased over time (Fig. 3C). Conversely, the expression levels of α fetoprotein and MET proto-oncogene, receptor tyrosine kinase were low or undetectable in all samples (Fig. 3C). The hepatic gene expression in the microsphere tissue samples appeared to decrease over time, particularly at day 14, which may be indicative of lower viability of the microsphere tissue after culture for 2 weeks. Collectively, these results demonstrated that the LMTC samples exhibited a normal hepatic gene expression profile under the in vitro culture conditions for at least 7 days, and potentially up to 14 days, although the 2-week microsphere culture exhibited apparent morphological changes with signs of tissue degradation and tissue debris (data not shown). Therefore, considering the morphological changes and the diminished hepatic functions and gene expression profile of the 2-week microsphere culture, the LMTC microspheres' viability may be limited to no longer than 2 weeks under the culture conditions of the present study.

Whether the LMTC model could be used to study the cell-based signaling pathways that may be involved in the effective delivery of transgenes into the microsphere tissue or exposure to secreted growth factors was examined. Whether the mouse liver microsphere tissue could be effectively transduced with recombinant adenoviruses, which represent one of the most effective gene delivery approaches, was first investigated. Considering the fact that the liver tissue may exhibit different proliferative potential at different ages, sift80 microsphere tissues from newborn, 14-day-old and 28-day-old mouse liver tissue were prepared, and then infected with the same titer of adenovirus Ad-GFP. Only sparsely infected GFP⁺ hepatocytes were detected in the sift80 microsphere tissue prepared from newborn mice at days 3 and 7, while GFP⁺ cells were very few or undetectable in the microsphere tissue from 14-day-old and 28-day-old mouse liver tissue; the Ad-GFP-infected MPH cells were used for the control (Fig. 4A). In fact, the adenovirus-mediated transgene delivery to whole mouse liver tissue was analyzed in vitro at 0, 3, 7, 10 and 14 days after birth, and only sparsely distributed, focal Ad-GFP infection was observed (data not shown). These results suggest that mouse liver tissues, whilst viable in culture, may have limited intrinsic proliferative potential for transgene expression in vitro.

Whether liver microsphere tissues were responsive to signal molecule stimulation was then examined. It has been previously demonstrated that BMP9 is a highly expressed secreted protein in fetal mouse liver tissues, and that it exerts pleiotropic effects on stem cell proliferation and differentiation (41-43,64-68). The BMP9-containing conditioned medium was first prepared by infecting HCT116 cells with Ad-BMP9, as described previously (45,47,69), while the Ad-GFP-infected HCT116 cells were used to prepare for the control conditioned medium (Fig. S1). A high level of BMP9 expression in the Ad-BMP9-infected HCT116, but not in the Ad-GFP infected cells, was confirmed by qPCR (Fig. S1). The BMP9-conditioned medium and GFP control medium were used to stimulate the freshly prepared sift80 microsphere tissue (Fig. 4B). At 3, 7 and 14 days post-BMP9 stimulation, total RNA was isolated and TqPCR analysis of BMP9 downstream target genes was performed as described previously (65,70-73). It was identified that the BMP9 downstream early responsive genes, Smad6 and Smad7, were significantly upregulated at day 3, while other target genes including inhibitor of DNA binding 1, HLH protein, inhibitor of DNA binding 2, connective tissue growth factor and cellular communication network factor 1 were significantly upregulated at days 3 and 7. However, the expression of the majority of the target genes significantly decreased to basal level at day 14 of culturing. Taken together, these results suggested that mouse liver microsphere tissues, whilst not serving as an ideal recipient for transgene delivery, may be used as an effective in vitro 3D model system for cytokine or growth factor-based cell signaling studies.

As DILI remained the leading cause of acute liver failure and post-market drug withdrawals (5), whether the liver microspheres could be used to assess the hepatotoxicity induced by liver injury drugs was assessed. A total of 3 commonly-used hepatotoxic drugs, levofloxacin, azithromycin and paracetamol, were selected. Using a WST-1 cell proliferation assay, it was identified that the 3 drugs inhibited cell proliferation rates of the MPH in a dose- and time-dependent manner (Fig. 5A). Therefore, the following optimal doses were selected for subsequent in vitro studies: Levofloxacin at 5 μM (Fig. 5A-a), azithromycin at 125 nM (Fig. 5A-b) and paracetamol at 100 μM (Fig. 5A-c).

The effects of the 3 drugs on ICG uptake in the MPH and the freshly prepared sift80 microspheres were analyzed. When subconfluent MPH cells were treated with levofloxacin (5 μM), azithromycin (125 nM), paracetamol (100 μM) or DMSO control, the ICG uptake was remarkably inhibited by all 3 drugs at both 24 h and 72 h, compared with that of the DMSO control group (Fig. 5B-a). However, when the sift80 microsphere tissue was treated with levofloxacin (5 μM), azithromycin (125 nM), paracetamol (100 μM) or DMSO control, the ICG uptake was not significantly inhibited by the 3 drugs at 24 h, but was markedly inhibited at 72 h, compared with the DMSO control group (Fig. 5B-b). Therefore, these results demonstrate that the mouse liver microsphere tissue culture model may be used as an effective in vitro surrogate system to assess the hepatic functional abnormality caused by hepatotoxic drugs.

Whether the LMTC model was able to predict the hepatotoxicity that was closely correlated with the hepatotoxic effects obtained from in vivo animal studies was also examined. The mice were treated with the levofloxacin, azithromycin, paracetamol or DMSO control, and mouse liver tissue was collected at 24 h or 72 h after treatment. Concomitantly, the subconfluent MPH and freshly-prepared sift80 microsphere samples were treated with levofloxacin, azithromycin, paracetamol or DMSO. Total RNA was isolated from the drug-treated mouse liver tissues, the MPH and the sift80 microspheres for quantitative analysis of the expression of a panel of 5 hepatotoxicity-associated genes, including ATP binding cassette subfamily B member 11 (Abcb11), cytochrome P450 family 2 subfamily E member 1 (Cyp2e1), cytochrome P450 family 7 subfamily Amember 1, nuclear receptor subfamily 0 group B member 2 (Nr0b2) and nuclear receptor subfamily 1 group H member 4 (74-76). In the levofloxacin-treated mouse liver tissue, MPH and sift80 microsphere tissue groups, while the majority of the hepatotoxicity-associated genes (with the exception of Nr0b2 in the mouse liver group) were not significantly upregulated at 24 h post treatment, 2/5 genes (Abcb11 and Nr0b2) in the mouse liver group, all 5 genes in the MPH group, and 4/5 genes in the sift80 microsphere tissue group were significantly upregulated at 72 h after treatment (Fig. 6A). In the azithromycin treatment groups, while all 5 genes were repressed at 24 h, all 5 genes were highly upregulated in all three groups (with the exception of Cyp2e1 in the MPH group) at 72 h after treatment (Fig. 6B). Similar results were observed in the paracetamol treatment groups, and all 5 genes were highly upregulated in all three groups (with the exception of Cyp2e1 in the mouse liver group) at 72 h after treatment (Fig. 6C). Notably, for the 3 drugs examined, the magnitudes of gene expression upregulation were increased in the sift80 microsphere tissue groups compared with that in the mouse liver groups at the 72 h treatment time point, suggesting that the LMTC model may be more sensitive in predicting drug-associated hepatotoxicity.