This article’s related datasets include COVID-19 and ALF from the Gene Expression Omnibus (GEO) database. The COVID-19 dataset (GSE180226) comprises 20 COVID-19 lung samples and three normal lung samples, and the ALF dataset (GSE38941) contains 17 HBV-infected ALF liver samples and ten healthy controls.

The “limma” package in RStudio software (version 4.2.2) was used to identify DEGs with standard | Log2Fold Change | > 1 and | adj. p-value | < 0.05 for the COVID-19 dataset (GSE180226). Meanwhile, DEGs with | Log2Fold Change | > 1 and | adj. p-value | < 0.05 for the ALF dataset (GSE38941) were also identified. We obtained COVID-19 and ALF common DEGs using the “Venn” package in R software.

The “clusterProfiler” package in R software (version 4.2.2) was used to perform GO and KEGG enrichment analysis. Other packages, including “dplyr,” “org.Hs.e.g.,.db,” “circlize,” “RColorBrewer,” “ggplot2,” “enrichplot,” “ggpubr,” and “ComplexHeatmap” were used for data annotation and visualization. p-value <0.05 was set as the cutoff criterion for common DEGs.

PPI networks were established by Search Tool for the Retrieval of Interacting Genes (STRING) database based on the combined score >0.9 and visualized by Cytoscape 3.9.1 (version 4.2.2) to reveal the interactions among proteins of common DEGs.

The expression of the common hub genes in GSE120652 for ALF and GSE139602 for ACLF was detected. GSE120652 contains three acetaminophen-induced ALF samples and three healthy samples, and GSE134431 contains eight ACLF samples and six healthy samples. Wilcoxon test was used to compare the two datasets.

The cryopreserved human hepatocytes were plated on collagen type I-coated (Rat-tail collagen type I, Gibco) plate at a density of 1 × 10⁵/cm² in a modified liver expansion medium containing William’s E Medium supplemented with 2% B27 (without VA), 1% PS (All from Invitrogen), 50 ng/mL EGF, 20 ng/mL HGF (Both from Peprotech), 200 μM 2-phospho-L-ascorbic acid (pVc), 3 μM CHIR99021, 5 μM SB431542, 5 μM Lysophosphatidic acid (LPA), 0.5 μM Sphingosine-1-phosphate (S1P) (All from MCE), and 1% fetal bovine serum (FBS) (Gibco, US). After seeding for 7–10 days, expanded cells were passaged at a ratio of 1:2 after dissociation with Accutase (eBioscience). The expansion medium was changed every day. Hepatocyte donor information is shown in Table 1.

The Direct-zol RNA Kits (ZYMO Research, R2052) were used to isolate total RNA from liver tissues and hepatocytes, followed by reverse transcription using the TransScript First-Strand cDNA Synthesis SuperMix (TransGen, AT311). The KAPA SYBR^® FAST qPCR Kit (KAPA Biosystems, KK4601) was used to perform relative gene expression on a CFX384 Touch Real-Time PCR Detection System (Bio-Rad). The housekeeping genes (RRN18S) were used as an internal reference to analyze RT-qPCR results using the 2^(−ΔΔCt) method. All the primer sequences for RT-qPCR are provided in Table 2.

The animal experiments were conducted under a protocol (2019-X15-03) that the ethics committee approved by the Laboratory Animals Center of Chinese PLA General Hospital (Beijing). Twenty-one male C57BL/6J mice (weight 30 g, aged 8–10 weeks) were purchased from Sipeifu (Beijing, China) and randomly divided into three groups to obtain liver tissue for molecular and histological studies (n = 7/group): mice with corn oil only without CCl4 injury (NC), mice with corn oil and CCl4 injury (CCl4), and mice with Apcin and CCl4 injury (Apcin). Before the CCl4 injection, the Apcin mice group was intraperitoneally injected with Apcin (30 mg/kg) dissolved in corn oil. Eight hours later, the CCl4 and Apcin mice were given CCl4 2.9 mL/kg dissolved in corn oil (Solarbio, C7030) by intraperitoneal injection. The mice in the NC group were injected with corn oil as a control. After 24 h of CCl4 treatment, all mice were sacrificed to obtain serum and liver tissues. Another 21 mice were randomly divided into three groups (n = 7/groups) to obtain the survival rate of ALF, as described above. Seven days after injection, all mice were sacrificed, and the survival rate can be further analyzed.

0.8 mL serum was incubated for 30 min and then centrifuged at 3,000 rpm for 5 min to get the supernatant to collect the mouse blood sample. The automatic chemistry analyzer was used to measure the aspartate aminotransferase (AST) and alanine aminotransferase (ALT), with each biochemical parameter being repeated four times.

Hepatocytes on the glass bottom plate (Cellvis, P24-1.5H-N) were fixed with 4% paraformaldehyde in PBS at room temperature for 20 min. Then the cells permeabilized in 0.25% Triton X-100 and blocked with 2.5% normal donkey serum (Jackson ImmunoResearch Laboratories, 017-000-001) in PBS. Next, the hepatocytes were incubated with primary antibodies at room temperature for 1 h, including Anti-ALB (ab243894, Abcame, 1:200), Anti-CDC20 (sc-5296, Abcame, 1:200), and Anti-CK19(ab181557, Abcame, 1:250). Hepatocytes were incubated with the secondary antibodies, including DyLight^® 650 Donkey anti-rabbit (ab96894, Abcame, 1:200) and DyLight^® 650 Donkey anti-mouse (ab96878, Abcame, 1:200) for 1 h at room temperature. Finally, the samples were stained with 4′,6-diamidino-2-phenylindole (DAPI) (Invitrogen, D1306) for 5 min and observed and photographed under an inverted fluorescent microscope using Nikon Imaging System at the same exposure.

After paraffin-embedded, the liver tissue was cut into sections of 4 μm thickness. Then the slides immunohistochemistry for the TUNEL staining was performed using the TUNEL Assay Kit-HRP-DAB (Abcam, ab206386). The H&E staining was performed using a regular protocol. All the results were randomly taken at the same exposure.

Graph Pad 8.0 was used for statistical analysis. All the data were the means ± standard deviation (SD). Differences with p < 0.05 was considered statistically significant. Student’s t-test and Wilcox were used to analyze differences between two or multiple groups. Kaplan-Meier analysis was used for the overall survival.

To explore the common changes of epithelial cells under acute injury to find potential targets for injury recovery and tissue regeneration, we downloaded the transcriptome data of COVID-19 and ALF from the GEO database. We identified 6,592 and 2,191 DEGs by comparing them with healthy data (Figures 1A,B). Further, the volcano plots demonstrated DEGs of ALF and COVID-19, respectively (Figures 1C,D). One hundred ninety-two co-upregulated genes and 227 co-downregulated genes were shown in Venn diagrams between ALF and COVID-19 datasets (Figures 1E, F and Supplementary Material S1). These results indicated common mechanisms and genetic expression trends among different tissues under acute injury.

GO and KEGG enrichment analyses were used further to analyze the role and function of common DEGs. GO enrichment analysis included three categories: biological process (BP), cell composition (CC), and molecular function (MF). The common DEGs were mainly enriched in amino acid metabolic processes or metabolic processes of other substances (BP), the complex of collagen trimers, endoplasmic reticulum lumen, and fibrillar collagen trimer (CC), and extracellular matrix reconstruction and antioxidant stress response (MF) (Figure 2A). The KEGG enrichment revealed that the common DEGs mainly enriched in protein digestion and absorption, complement and coagulation cascades, and drug metabolism pathways (Figure 2B).

STRING was used to construct the PPI network to explore gene interactions further and establish a hub gene network, in which 81 nodes and 250 edges were included (Figure 3A). CytoHubba was used to identify the top 15 genes in Cytoscape based on the degree method scores in the PPI network, including BUB1B, CDC20, ASPM, NUSAP1, CEP55, UBE2C, KIF11, RRM2, PBK, PTTG1, NCAPG, TPX2, KIF4A, NUF2, and CENPF (Figures 3B,C). We believe these targets may reflect the change of epithelial cells in the state of acute injury. Validation was performed for ALF and ACLF from GEO, and the results showed that these hub genes were upregulated in ALF compared to normal tissues, although without significant differences (Figures 4A, B). However, the trend is not evident in ACLF (Figures 4A, B). These results indicate that chronic injury weakens the regenerative ability of epithelial cells, consistent with current theoretical studies, suggesting the role of hub genes in acute injury from the side.

To further explore hub genes’ biological effects on tissue injury, GO and KEGG analyses were conducted again on the 15 hub genes. Interestingly, the enrichment status of hub genes was apparent. In GO enrichment, the BP category, CC category, and MF category all showed functional enrichment related to cell proliferation, mitosis, and spindle formation. KEGG enrichment analysis demonstrated pathways associated with cell proliferation and viral infection (e.g., Cell cycle, Human T−cell leukemia virus one infection, and Oocyte meiosis) (Figures 5A, B). These results indicate that tissue regeneration is the most apparent phenomenon under the stress of acute failure.

In the in vitro hepatocyte regeneration model, serum and cytokines simulating the injury environment were used to induce hepatocyte proliferation (Figure 6A). Hepatocytes entered the progenitor cell state with the upregulation of hub genes such as CDC20, CENPF, CEP55, KIF4A, KIF11, NCAPG, NUSAP1, PTTG1, TPX2, and UBE2C (Figure 6B, Supplementary Material S2A). Compared to the primary hepatocytes, progenitor markers such as EPCAM, AFP, and CK19 expressed much higher in liver progenitors while liver function genes including cytochrome P450 (CYP) enzymes and urea metabolism genes downregulated (Figures 6C, D, Supplementary Material S2A). Immunofluorescence staining confirmed the changes in cellular identity and function (Figures 6E, F). These results demonstrated the relationship between hub genes and liver regeneration. The in vivo experiment schematic is represented in Figure 7A. Similarly, in the CCL4-induced ALF model to monitor liver injury and regeneration, we also observed significantly upregulated hub genes in the ALF liver than healthy control (Figure 7C, Supplementary Material S2B). In addition, progenitor genes were upregulated, while CYP enzymes and other functional genes were down (Figures 7D, E, Supplementary Material S2B). These results reflect the consistent changes in response to stress stimulation and following regeneration patterns in liver cells.

During ALF, extensive hepatocytes undergo dedifferentiation, and proliferation may influence liver function. Therefore, we wonder if maintaining the identity of hepatocytes and moderate inhibition of proliferation will contribute to the survival of ALF. CDC20 inhibitor Apcin has been reported to inhibit cancer cell proliferation by competitively inhibiting anaphase-promoting complex/Cyclosome (APC/C)-dependent ubiquitination. It was found that Apcin pretreatment can attenuate liver failure and improve the survival rate in a mouse model (Figure 7B). Liver damage and inflammatory response were relieved by Tunnel and H&E staining (Figure 7F). Serum of mice was collected for biochemical detection, including alanine aminotransferase (ALT) and aspartate transaminase AST (Figure 7G). These data illustrate the positive effect of Apcin on ALF, which may also help with the treatment of COVID-19.

In conclusion, these results indicate that regardless of COVID-19, ALF, or other acute failures, the epithelium entered the cell proliferation stage induced by injury, and this regeneration contributed to repairing and recovery in the long run. However, short-term loss of function may aggravate the general condition. Therefore, proper regulation of the balance between cell proliferation and function maintenance may be beneficial for patients to recover from acute failure. The hub genes identified in this paper and their corresponding targets, such as CDC20 inhibitor-Apcin, may be a new possible approach for ALF and COVID-19 treatment.