Healthy male BALB/c mice (Charles River, Boston, MA, United States), aged 12–30 weeks, were used. Mice were maintained in controlled temperature, humidity and a light-dark cycle from 07.00 a.m. to 07.00 p.m. and allowed food and water ad libitum. All experimental protocols and animal experimentation ethics were carried out in accordance with the approved guidelines of the Vrije Universiteit Brussel (VUB, Belgium) and according to European Guidelines for the Care and Use of Laboratory Animals. All animal experimentation protocols were approved by the Ethical Committee of Animal Experimentation of the Vrije Universiteit Brussel (VUB, Belgium) (LA 123 02 12, projects 18-212-1; 19-212-1).

Human liver tissue was obtained from hepatectomies performed by the Department of Surgery at the University Hospital of Brussels (UZ Brussel). After surgical resection, liver tissue samples from the specimen in an area unaffected by and distant from the lesion for which surgical resection was intented for. The liver tissue was immediately put in ice cold IGL-1 solution (Institute Georges Lopez). The ethical approval Reference 2015/278; B.U.N. 143201525406 was obtained from the ethical committee of the UZ Brussel and was in accordance with the Declaration of Helsinki. All participants signed their informed consent prior to the donation of liver tissue. An overview of human patient characteristics is given in Table 1.

Mice were anesthetized using an i.p. injection with 100 μl Dolethal^® (Vetoquinol). After anesthesia of the mouse, the median liver lobe was isolated and immediately put on ice cold IGL-1 solution. Liver tissue was sliced using a Leica VT1200S vibrating blade microtome (Leica Biosystems; speed: 0.1 mm/s, amplitude: 2 mm, step size: 250 μm), while being submerged in Williams’ Medium E medium (Gibco), supplemented with or without 2.5 mM VPA (Sigma-Aldrich). Next, the 250 μm thick sections were subsequently punched with a disposable 3 mm biopsy puncher with plunger (Kai Medical). We analyzed PCLS with different diameters (1, 3, or 8 mm) for liver culture stability. The RNA yield of discs with 1 mm diameter was not sufficient for robust and reproducible quantitative real-time polymerase chain reaction (qPCR) analysis. 3 mm diameter discs, compared to 8 mm diameter discs, resulted in a better maintenance of mRNA levels of hepatocyte markers for at least 3 days (Supplementary Figure 1). Using a P1000 pipet with cut tip, PCLS were transferred into a 24-well plate, pre-filled with cold supplemented WME medium. After slicing and punching was finished, PCLS were refreshed with pre-heated WME medium (Gibco), supplemented with 1% Ultraglutamine I (Lonza) and 1% Penicillin-Streptomycin (Life Technologies) and cultured on an orbital shaker (Infors Celltron) at a speed of 80 rpm. PCLS were cultured at 37°C, and refreshed 2–4 h after production and every 24 h.

For exposures to compounds, PCLS were cultured in 2.5 mM VPA until day 3 of culture. From day 3 to day 5 of culture, VPA concentration was reduced to 1 mM and PCLS were simultaneously exposed to compounds or corresponding solvent control. Compounds concentrations used: 10 ng/mL TGF-β1 (PreproTech), 1 mM acetaminophen (APAP) (Sigma-Aldrich), 10 μM SB-525334 (Alk5 inhibitor) (Sigma-Aldrich), 20 mM N-acetyl-L-cysteine (NAC) (Sigma-Aldrich). On day 5 of culture (after a 48 h exposure), PCLS and culture medium were collected for analysis.

For viability analysis, PCLS culture medium was collected and centrifuged (8 min, 640 g). Supernatant was transferred to a new tube and diluted 1/5 in lactate dehydrogenase (LDH) storage buffer (200 mM Tris–HCl, 10% glycerol, 1% BSA) and stored at −80°C until analysis. LDH levels were measured using the LDH-Glo Cytotoxicity Assay (Promega) according to manufacturer’s instructions.

Slices were snap frozen in liquid nitrogen and stored at −80°C until RNA extraction. Total RNA was extracted with TRIzol Reagent (Thermofisher) according to the manufacturer’s manual. Samples were homogenized by crushing the tissue with 5 mm steel beads (Qiagen) with a Retsch MM 400 laboratory mill at 30 Hz for 2 min. Afterward, RNA was reverse transcribed into cDNA using the MLV Reverse Transcriptase (Promega). For qPCR, GoTaq qPCR Master Mix with BRYTE green (Promega) was used. qPCR was done using the Quantstudio3 real-time PCR system (Thermofisher). Gene-specific primers (Table 2) were produced by Integrated DNA Technologies (IDT Leuven). For analysis according to the Delta–Delta threshold (Ct) method, each Ct value was normalized against the respective reference genes (for mouse: the mean of Gapdh and Gtf2b; for human: the mean of GAPDH and YWHAZ). The best reference genes were selected out of six genes using GeNORM, performed in R environment. Each data point corresponds to one PCLS.

PCLS culture medium was collected and centrifuged (8 min, 640 g) and stored at −80°C until analysis. For the analysis, 500 μl of medium was used. miRNA was extracted using the Nucleospin^® miRNA Plasma kit (Macherey-Nagel) according to the manufacturer’s protocol. Caenorhabditis elegans miRNA-39 (cel-miR-39) (Qiagen) was spiked into the lysate before extraction and served as an external processing control.

Snap frozen PCLS were dissolved in RIPA lysis buffer (1% NP-40, 0.1% sodium-deoxycholate, 0.04% SDS, 50 mM Tris–HCl, 10 mM NaF, 30 mM NaCl, and 0.4mM EDTA), supplemented with complete protease-inhibitors (Roche Diagnostics) and PhosSTOP phosphatase inhibitors (Roche Diagnostics). Lysates were sonicated twice (15 s, 50% amplitude) in a 4°C water bath using a digital sonifier (Branson). Protein concentrations were quantified using the Micro BCA™ Protein assay kit (Thermo Fisher Scientific) according to manufacturer’s instructions. Thirty micrograms of total protein was used for western blotting. Protein expression was assessed using antibodies against β-actin (1:5,000, Sigma-Aldrich), PDGFRβ (1:1,000, Abcam), Albumin (1:1,000, Novus Biologicals), and αSMA (1:5,000, Vendor, Lifespan Technologies). Protein bands were visualized with an ImageQuant™ LAS 4000 (GE Healthcare Life Sciences). Protein bands were quantified using ImageJ (NIH).

PCLS culture medium was centrifuged (8 min, 640 g). Supernatant was transferred to a new tube and stored at −80°C until use. Secreted Albumin or Collagen 6 protein levels were measured with a commercially available ELISA kit (MyBiosource, Albumin: MBS2516177, Collagen 6: MBS2704928). Protocol was executed according to the manufacturer’s instructions. Absorbance values were obtained with an iMark™ microplate absorbance reader (Bio-Rad).

PCLS were fixed for 10 min in formalin and stored in PBS at 4°C until embedding. Paraffin embedded PCLS were sliced in 4 μm sections. Next, sections were deparaffinized and rehydrated. After rehydration, sections were incubated for 3 min in hematoxylin (Carl Roth), followed by a rinse with tap water and acidified water. Next, sections were rinsed in running tap water for 10 min, followed by a 3 min incubation in eosin (Sigma-Aldrich) staining. After the staining, the sections were shortly rinsed with water and subsequently dehydrated through graded washes of ethanol and water. Sections were mounted in DPX mounting medium.

After RNA extraction, RNA was processed for RNASeq by NovaSeq SP 100 × 6 bp single-end sequencing. This resulted in, on average, 7 M reads per sample. Quality control and trimming was performed using FastQC and STAR was used for mapping of the reads to the reference genome (Mus_musculus_GRCm38.p6) (19). The python package StringTie was used for assembly of genes and transcripts (20). After generation of raw counts, DESeq2 in R was used to normalize counts, determine differentially expressed genes and perform principal component analysis (PCA) (21). The Database for Annotation, Visualization, and Integrated Discovery (DAVID) v6.8 was used to perform Gene Ontology (GO) analysis (22). GSEA software v4.1.0 was used to perform gene set enrichment analysis (GSEA) (23). For GSEA analyzes, gene set permutation type and a false discovery rate (FDR) statistic threshold of 0.1 was chosen to determine significantly enriched gene sets. Bulk RNAseq data of day 3 PCLS cultures has been deposited in the GEO public data base under accession number GSE194128.

All schemes were created with Biorender.com using an Academic License.

Data was analyzed using Graphpad Prism 8 (Graphpad, Palo Alto) statistical software. As our data is considered being non-parametric, quantitative variables are expressed as boxplots (median – min to max). In case of two conditions, a Mann–Whitney test was performed, or in case of more than two conditions, Kruskal–Wallis test with Dunn’s post hoc test were performed. For each graph, n = x indicates how many biological repeats were performed, with one biological repeat being one different mouse/human liver.

PCLS cultures have frequently been used for compound metabolization and toxicity testing. Several studies have evaluated the anti-fibrotic effects of drugs and direct activation of HSCs by TGFβ or/and PDGF-BB can be easily induced (16). However, the evaluation of drug-induced liver fibrosis and HSC activation has not been well documented yet (10). We set out to evaluate whether it is possible to induce HSC activation in traditional PCLS cultures by exposing them to acetaminophen (APAP), shown to transiently induce HSC activation after an acute exposure (24), and fibrosis when chronically administered to mice (25). PCLS with 3 mm diameter and 250 μm thickness were obtained by slicing mouse liver tissue with a Leica VT1200S vibrating blade microtome. Subsequently, liver discs were punched with a 3 mm core biopsy puncher and cultured for 5 days in regular WME medium (Figure 1A). Figure 1B shows a freshly sliced 3 mm disc imaged using light microscopy and Figure 1C a 4 μm hematoxylin-eosin (H&E) section of the disc showing the intact liver architecture of the disc. The production and culture of PCLS induces a cut surface, which results in cell death. When PCLS were cultured for 5 days in regular WME medium, high levels of LDH, representing cell death, were detected on day 1 of culture which strongly reduced from day 2 onward (Figure 1D). This hepatocyte damage eventually results in a fibrotic response over culture time as demonstrated by the significant increase in mRNA levels of genes associated with HSC activation (Acta2, Col5a2, and Col1a1) on day 5 of culture (Figure 1E). Exposure of PCLS to 1 mM acetaminophen (APAP) during the first 48 h of culture (Figure 1F) demonstrated that an (additional) hepatocyte-damage and subsequently induced pro-fibrotic response could not be established (Figures 1G,H). This suggests that in standard PCLS conditions, the injury associated with the production of the slices is most likely too high and additional damage by drugs, and subsequent fibrosis, is not easily achieved.

VPA, a histone deacetylase class I inhibitor, is known to inhibit HSC transdifferentiation and activation (26). We hypothesized that if we could inhibit the first wave of HSC activation, which is induced during the initial damage to hepatocytes as a result of the slicing and punching procedure, we would perhaps be able to induce drug-induced liver damage and fibrosis at a later time point of the cultures. To investigate this, PCLS were cultured for 5 days with regular WME or medium supplemented with 2.5 mM VPA (Figure 2A), the dose that can inhibit culture-induced activation of primary mouse HSCs (26). The addition of 2.5 mM VPA to the culture medium had beneficial effects on the procedure-induced fibrosis progression, as HSC activation genes were not significantly upregulated after 5 days of culture in the presence of VPA, in contrast to the regular standard conditions (Figure 2B). This tendency was also observed on protein level, as PDGFRβ protein expression did not increase over time in the presence of VPA (Figure 2D). Hepatocyte-specific gene expression was not different on day 5 when comparing both conditions (Figure 2C). However, albumin protein expression remained slightly higher over time in the presence of VPA (Figure 2D). Moreover, H&E staining shows a better preservation of cell nuclei and liver histology and architecture on day 5 when PCLS were cultured in the presence of 2.5 mM VPA (Figure 2E).

To get more insight into the underlying mechanisms of the beneficial effects of VPA for the cultures, RNA sequencing (RNA-seq) of PCLS cultured for 3 days in the absence (CTR) or presence (VPA) of 2.5 mM VPA was performed. PCA plot shows a division of CTR and VPA treated slices over PC1, which accounts for 34% of variance (Figure 2F). Moreover, the fibrogenic aHSC gene signature recently identified by De Smet et al. (9), is significantly enriched in standard PCLS culture conditions when compared to VPA supplemented conditions (Figure 2G). Differential gene expression analysis showed that 48 genes were differentially upregulated and 130 genes were differentially downregulated (such as Aebp1, Col1a1, and Col3a1) in the presence of VPA compared to control conditions (data not shown). Further downstream analysis demonstrates that VPA downregulated genes are associated with the GO category “ECM organization” and “collagen fibril organization” and the KEGG pathway “ECM-receptor interaction” (Figure 2H). This data demonstrates that VPA extends the life span of PCLS cultures up to 5 days as tissue integrity is better preserved and the procedure-induced fibrogenic aHSC transcriptional program is strongly repressed.

Next, we wanted to use the VPA-supplemented conditions to test for drug-induced liver injury and fibrosis induction. However, we noticed that in VPA supplemented PCLS cultures induction of HSC activation with TGFβ at day 3 was not very robust, and that complete wash-out of VPA from day 3 until day 5 induced HSC activation genes (Supplementary Figure 2A). However, reduction of the VPA concentrations to 1 mM from day 3 to day 5, did not result in a significant induction of HSC activation (Supplementary Figure 2B). We used this modified setup, further referred to as PCLS^V, to test direct HSC activation by TGFβ during 48 h starting at day 3 (Figure 3A). Exposure of these PCLS^V cultures to TGFβ at day 3 clearly induced HSC activation as evidenced by an increase in mRNA levels of HSC activation markers Col5a2, Acta2, and Lox (Figure 3B). TGFβ-induced HSC activation could be inhibited by co-exposing PCLS^V with SB-525334 (Alk5i), a selective inhibitor of TGFβRI (27). As this experiment demonstrated that the HSCs were still functional at that time point, i.e., able to respond to a direct stimulus, we next investigated whether HSC activation (liver fibrosis) could be induced to the same extent by damaging the hepatocytes with APAP (28). Exposure to APAP induced hepatotoxicity, as APAP administration slightly reduced albumin secretion in the culture medium (Figure 3C), and resulted in a clear increase in LDH leakage (Figure 3D).

To confirm whether this increased LDH leakage was due to the cytotoxic effect of APAP, PCLS^V cultures were treated with NAC, which is currently still the only antidote used in the clinic for APAP-induced liver injury (29). NAC is known to block APAP-induced hepatotoxicity, as it acts as a substitute of glutathione, when administered within 8 h after overdose in patients (30). However, studies have shown that NAC treatment in mice is not effective beyond 4 h after APAP administration, as murine have an accelerated metabolism (31). Therefore, NAC was administered simultaneously with APAP to the PCLS^V cultures at day 3. Indeed, the co-administration of APAP with NAC prevented the decreased secretion of albumin (Figure 3C) and decreased the LDH leakage (Figure 3D). The APAP- induced hepatocyte damage resulted in a pro-fibrotic response as demonstrated by an increase in Col1a1, Col5a2, and Lox expression after APAP administration, while this effect was blocked when NAC was co-administered (Figure 3E). The same tendencies were observed at the protein level, as PDFRβ expression in the slices (Figure 3F) and collagen 6 secretion in the culture medium (Figure 3G) were increased in APAP exposed PCLS^V, while this was not observed in the presence of NAC. These results show that in the PCLS^V cultures, the hepatotoxic effect of APAP results in a pro-fibrotic response of the HSCs, which can be inhibited through NAC administration.

Lastly, we investigated whether our findings obtained in mouse PCLS^V cultures could be extrapolated to human PCLS cultures. Therefore, human PCLS^V cultures were established and tested for their fibrotic response at day 3 following TGFβ- or APAP exposure (Figure 4A). Human PCLS^V cultures exposed to TGFβ show increased expression of HSC activation markers on mRNA (Figure 4B) and protein level (Figure 4C). This effect could be very efficiently blocked by the inhibitor SB-525334 (Alk5i).

Human PCLS^V exposed to increasing concentrations of APAP did however not result in an increase of LDH leakage (Figure 4E). However, a dose-dependent increase of mir122-5p, a miRNA that can be used as a surrogate of hepatocyte damage (32), was detected in human PCLS^V culture medium (Figure 4D). Unfortunately, although there was a clear trend for the induction of COL1A1 and LOXL2 mRNA levels, due to the large variations in the extend of activation, these increases were not significant (Figure 4F). These large variations in HSC activation are most likely due to the great difference in health status of our donor livers, as a strong correlation was found between LDH leakage and mRNA HSC activation levels on day 5 of culture (Figure 4G).