Pien Tze Huang (PTH) powder was provided by Zhangzhou Pien Tze Huang Pharmaceutical Co., Ltd. (China National Medical Products Administration approval number: Z35020242). Carbon tetrachloride (CCl₄, purity ≥ 99.5 %, Cat. No. C110623) and corn oil (Cat. No. C116025) were purchased from Shanghai Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). TRIzol reagent (Cat. No. 15596026) was obtained from Thermo Fisher Scientific (USA). Hydrogen peroxide (H₂O₂, 3 % solution for microbiology; Cat. No. H755825) was purchased from Shanghai Aladdin Biochemical Technology Co., Ltd. Assay kits for alanine aminotransferase (ALT, Cat. No. C009-2-1), aspartate aminotransferase (AST, Cat. No. C010-2-1), and alkaline phosphatase (ALP, Cat. No. A059-2-2) were all purchased from Nanjing Jiancheng Bioengineering Institute (Nanjing, China).

Quality control for PTH powder was performed according to the standards of the Chinese Pharmacopoeia (2025 edition), and the presence of its main characteristic components (such as notoginsenosides and ginsenosides) was confirmed by reference to previous HPLC-MS analysis results (Zhu et al., 2021[55]). For animal experiments, the powder was dissolved in distilled water to prepare a 50 mg/mL stock solution for subsequent use. Carbon tetrachloride (CCl₄) was mixed with corn oil at a ratio of 1:9 (v/v) to prepare a 10 % (v/v) CCl₄ working solution. For cellular experiments, the powder was dissolved in sterile PBS to prepare a 30 mg/mL stock solution. The solution was heated in a 75 °C water bath for 30 min, followed by ultrasonication for 30 min. After centrifugation at 3000 rpm for 10 min, the supernatant was filtered sequentially through a 70 μm cell strainer and a 0.22 μm syringe filter to ensure sterility. The stock solution was aliquoted and stored at −20 °C until use.

Eighteen male C57BL/6 mice, aged 6 weeks and of SPF grade, were purchased from Shanghai Jiesijie Laboratory Animal Co., Ltd. (Shanghai, China). The animals were housed under controlled conditions at 21-22 °C with 50-55 % humidity and a 12/12 h light-dark cycle, with free access to food and water. All animal experimental procedures were approved by the Institutional Animal Care and Use Committee of Shanghai Jiao Tong University (Approval No. IACUC-2017-0033) and conducted in accordance with the ARRIVE guidelines.

The 18 mice were randomly divided into three groups (n = 6): Control group, Cirrhosis group, and Cirrhosis-PTH treatment group (Cirrhosis-PTH). Mice in the Cirrhosis and Cirrhosis-PTH groups received intraperitoneal injections of 10 % CCl₄ (10 µL/g body weight) twice weekly for 12 consecutive weeks to induce the cirrhosis model; mice in the Control group received intraperitoneal injections of an equal volume of corn oil. Concurrently, mice in the Cirrhosis-PTH group received daily intragastric administration of PTH (0.25 mg/g body weight), while mice in the Control and Cirrhosis groups received an equal volume of distilled water by gavage. All treatments were continued for 12 weeks. After the final administration, mice were anesthetized with isoflurane and euthanized by cervical dislocation. Blood samples were collected from the orbital sinus to obtain serum, and liver tissue samples were either preserved in 10 % formalin or stored at -80 °C for subsequent analysis.

Serum levels of ALT, AST, and ALP were measured using an automated biochemical analyzer. The liver index (liver weight/body weight × 100 %) was calculated.

Liver tissues were fixed in 4 % paraformaldehyde, embedded in paraffin, and sectioned into 4 μm slices. Hematoxylin-eosin (HE) staining, Masson's trichrome staining, and Sirius Red staining were performed respectively. Stained sections were observed and photographed under a light microscope.

Total RNA was extracted from liver tissues using TRIzol reagent. RNA purity was assessed by A260/A280 ratio, with all samples showing ratios > 1.9. After ribosomal RNA depletion, two types of sequencing libraries were constructed: lncRNA, circRNA, and mRNA sequencing libraries were prepared using the NEBNext RNA Library Prep Kit, while miRNA sequencing libraries were prepared using the VAHTS Small RNA Library Prep Kit. The lncRNA/circRNA/mRNA libraries were sequenced on the Illumina HiSeq 4000 platform with 150 bp paired-end reads, and the miRNA libraries were sequenced on the Illumina NovaSeq platform with 50 bp paired-end reads.

FastQC (Brown et al., 2017[4]) (v0.11.9) was used to assess raw sequencing quality. For mRNA and lncRNA data, adapter sequences, poly-N sequences, and low-quality reads were removed using in-house Perl scripts to obtain clean reads. Clean reads were aligned to the mouse reference genome (GRCm38) using hisat2 (Kim et al., 2019[24]) (v2.2.1), and transcript assembly and quantification were performed using stringtie (Pertea et al., 2015[33]) (v2.1.5) to obtain the FPKM expression matrix. For miRNA data, Cutadapt (Kechin et al., 2017[23]) (v4.2) was used to remove adapters and sequences shorter than 10 nt, and the processed sequences were aligned to the mouse reference genome (GRCm38) using Bowtie (Langmead and Salzberg, 2012[27]) (v1.3.1) and miRDeep2 (Friedländer et al., 2012[16]) for miRNA quantification. For circRNA data, Trimgalore (Sun, 2020[39]) and Fastp (Chen et al., 2018[8]) were used to remove adapters, poly-N sequences, and low-quality reads, Bowtie2 (Langdon, 2015[26]) was used to remove rRNA and E. coli contamination sequences, and CIRI2 (Li et al., 2021[29]) was used for circRNA identification and quantification. We confirm that all RNA-seq libraries were prepared in a single experiment using the same batch of reagents, pooled, and then loaded across lanes of the same flow cell for sequencing. No batch effect was observed, and therefore batch effect correction was not required.

Differential expression analysis of mRNA, lncRNA, miRNA, and circRNA was performed using the DESeq2 (Varet et al., 2016[42]) (v1.40.1) R package. Significance thresholds were set as follows: for mRNA and lncRNA, |log₂FC| > 1 and adjusted P-value < 0.01; for miRNA, |log₂FC| > 1 and P-value < 0.05; for circRNA, |log₂FC| > 0.25 and P-value < 0.05. Data visualization and heatmap generation were performed using the ggplot2 (Cao et al., 2023[5]) (v3.4.2) and pheatmap (Gaujoux and Seoighe, 2010[17]) R packages. Principal component analysis (PCA) and hierarchical clustering analysis were performed based on the mRNA expression profiles of all samples.

Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses of differentially expressed mRNAs were performed using the clusterProfiler (Wu et al., 2021[49]) (v4.4.4) R package. The significance threshold for enrichment was set at P-value < 0.05.

Drug similarity analysis was performed based on the Connectivity Map (CMap) database. Mouse gene symbols of differentially expressed genes in the Cirrhosis-PTH vs. Cirrhosis comparison were converted to human homologous gene symbols and divided into up-regulated and down-regulated gene sets. Due to the input limitations of the CLUE query tool (Lamb, 2007[25]), all up-regulated genes and the top 150 down-regulated genes ranked by |log₂FC| value were selected for query. The PTH-induced gene expression profile was compared with drug-induced gene expression profiles in the CMap database, and connectivity scores (ranging from-100 to 100) were calculated, with higher scores indicating stronger similarity. The analysis results from human liver-derived cell lines (HEPG2, PHH, and HUH7) were specifically focused on.

To investigate whether non-coding RNAs participate in the regulatory effects of PTH on cirrhosis through the ceRNA mechanism, we integrated database prediction and differential expression analysis strategies. First, Miranda, Targetscan, and RNAhybrid (Enright et al., 2003[15]; Shi et al., 2017[37]; Zhang et al., 2018[52]) were used to predict potential target miRNAs of DEcircRNAs, as well as potential target mRNAs of DEmiRNAs and DElncRNAs. However, no target miRNAs associated with DElncRNAs were identified in this study. Subsequently, the predicted interaction pairs were integrated with the differential expression analysis results, retaining only those pairs in which all molecules were differentially expressed in the corresponding comparison groups. Based on this, condition-specific ceRNA networks were constructed for the Cirrhosis vs. Control and Cirrhosis-PTH vs. Cirrhosis comparisons, respectively.

We used the protein abundance and gene expression data from 32 human tissues published in the study by Jiang et al. (Jiang et al., 2020[21]) to query the protein abundance and tissue-specific expression of candidate biomarker genes in different human tissues. The Therapeutic Target Database (TTD, https://db.idrblab.net/ttd/) was then searched to identify whether candidate biomarker genes associated with cirrhosis have been previously reported (Chen et al., 2002[9]).

Human HepG2 cells, obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China), were cultured in Dulbecco's Modified Eagle Medium (DMEM, Gibco, USA) supplemented with 10 % fetal bovine serum (FBS, Gibco, USA), 100 U/mL penicillin and 100 µg/mL streptomycin (Gibco, USA) at 37 °C in a 5 % CO₂ atmosphere. The culture medium was routinely replaced, and cells were either subcultured or used in experiments when they reached 80-90 % confluence. Cells were treated with different concentrations of PTH (0.05, 0.1, and 0.2 mg/mL) for 24 h based on a previous study (Liu et al., 2026[30]).

HepG2 cells (2 × 10³ cells/mL) were seeded in 96-well plates and treated with PTH (0.05, 0.1, and 0.2 mg/mL) for 24 h, followed by exposure to a gradient of H₂O₂ concentrations (200-1000 μM) for 2 h. Control cells were treated with vehicle alone or PTH alone without H₂O₂ exposure. Cell viability was assessed by incubating the cells with 10 μL of CCK-8 solution (Beyotime, Shanghai, China) for 1 h following the manufacturer's protocol, and the absorbance was measured at 450 nm using a microplate reader.

The hsa-miR-122-5p inhibitor was purchased from MedChemExpress (Shanghai, China). HepG2 cells were seeded in 6-well plates and transfected with 100 nM of the inhibitor using Lipofectamine 3000 (Thermo Fisher Scientific) according to the manufacturer's instructions. Approximately 6 h after transfection, the medium was replaced with fresh complete medium containing serum and antibiotics. The cells were then incubated for an additional 48 h before subsequent analysis. The transfected cells were divided into the following four groups: a control group without any treatment, a negative control inhibitor group (NC inhibitor), an miR-122-5p inhibitor group, and an miR-122-5p inhibitor plus PTH group.

Total RNA (including miRNA) was extracted from HepG2 cells using the SPARKeasy Improved Tissue/Cell RNA Kit (SparkJade, Shanghai, China; Cat. No. AC0202). For mRNA detection, reverse transcription was performed using the BeyoRT™ III First-Strand cDNA Synthesis Mix (Beyotime, Shanghai, China; Cat. No. D7182). For miRNA detection, reverse transcription was carried out using the miRNA 1st Strand cDNA Synthesis Kit (by stem-loop) (Vazyme, Nanjing, China; Cat. No. MR101-02). qPCR was performed using the BIO-RAD CFX96 Real-Time PCR Detection System with TruQuant SYBR Green (Thermo Fisher Scientific, USA; Cat. No. FB05001010). The primer sequences are listed in Supplementary Table 1. Gene expression levels were normalized to GAPDH (for mRNA) and U6 (for miRNA), and relative expression was calculated using the 2^-ΔΔCt method.

All data are presented as mean ± standard error of the mean (SEM). Comparisons among multiple groups were performed using one-way analysis of variance (ANOVA) followed by Tukey's post hoc test. A P-value < 0.05 was considered statistically significant. Statistical analyses were performed using R software (v4.1.0).

Compared with the Control group, mice in the Cirrhosis group exhibited significantly elevated serum levels of ALT and AST, indicating marked hepatocellular damage and inflammatory response. PTH treatment significantly reduced both ALT and AST levels (Figure 2A, B(Fig. 2)). Furthermore, serum levels of ALP were markedly increased in cirrhotic mice and were significantly reduced following PTH treatment (Figure 2C(Fig. 2)). Histological analysis further supported these conclusions. HE staining demonstrated that PTH treatment significantly attenuated the severe cirrhotic morphology and cellular disorganization observed in cirrhotic mice. Masson staining and Sirius Red staining revealed extensive collagen fiber deposition and pseudolobule formation in the livers of cirrhotic mice, both of which were significantly reduced following PTH treatment (Figure 2D-F(Fig. 2)). Finally, compared with Control mice, the liver index was significantly decreased in the Cirrhosis group, reflecting liver atrophy and functional impairment. PTH intervention significantly improved this, indicating partial recovery of liver mass (Figure 2G(Fig. 2)). Taken together, these results show that PTH treatment ameliorates CCl₄-induced liver injury and cirrhosis in mice.

Differential expression analysis revealed that in the Cirrhosis vs. Control comparison, 3,231 mRNAs were significantly differentially expressed. In the Cirrhosis-PTH vs. Cirrhosis comparison, 1,689 mRNAs were identified as differentially expressed (Supplementary Figure 1). PCA of mRNA expression profiles showed clear separation of the three groups, with the Cirrhosis-PTH group clustering closer to Control than to Cirrhosis (Figure 3A(Fig. 3)). Hierarchical clustering analysis confirmed this pattern, demonstrating that the gene expression signature of PTH-treated mice was more similar to controls (Figure 3B(Fig. 3)). To identify genes specifically reversed by PTH treatment, we intersected the differentially expressed genes (DEGs) from the two comparisons. A total of 1,518 genes were dysregulated in cirrhosis and significantly reversed by PTH treatment. The log₂ fold changes of these overlapping genes showed a strong negative correlation between the two comparisons (r = -0.925, P < 2.2e-16), indicating that PTH treatment was associated with transcriptional changes opposite to those observed in the disease condition.

To characterize the biological processes associated with cirrhosis and its treatment by PTH, we performed Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis on DEGs from both comparisons (see Figure 3C-F(Fig. 3)). In the Cirrhosis vs. Control comparison, GO analysis revealed enrichment of terms related to "protein binding" (GO:0005515), "response to stimulus" (GO:0050896), and "cell periphery" (GO:0071944). KEGG pathway analysis identified 25 significantly enriched pathways (P < 0.05), including Cytokine-cytokine receptor interaction (mmu04060), MAPK signaling pathway (mmu04010), NF-κB signaling pathway (mmu04064), and PI3K-Akt signaling pathway (mmu04151). In the Cirrhosis-PTH vs. Cirrhosis comparison, GO analysis showed enrichment of terms including "receptor ligand activity" (GO:0048018), "response to stimulus" (GO:0050896), and "extracellular region" (GO:0005576). KEGG analysis revealed significant enrichment of pathways including Cytokine-cytokine receptor interaction (mmu04060), IL-17 signaling pathway (mmu04657), MAPK signaling pathway (mmu04010), TNF signaling pathway (mmu04668), and NF-κB signaling pathway (mmu04064).

To further explore the clinical relevance of these pathway-level changes, we performed CMap analysis using the 1,518 PTH-responsive genes. The PTH-induced transcriptomic signature showed significant similarity to that of statins (simvastatin and atorvastatin) and beta-blockers (metoprolol and propranolol), all of which are drugs currently in clinical trials for cirrhosis (Supplementary Tables 2-4). This convergence between pathway enrichment and drug similarity predictions provides additional support for the potential therapeutic relevance of PTH in cirrhosis.

To explore potential regulatory interactions among differentially expressed coding and non-coding RNAs, we first analyzed the expression profiles of non-coding RNAs. In the Cirrhosis vs. Control comparison, 491 lncRNAs, 137 miRNAs, and 214 circRNAs were significantly differentially expressed (Supplementary Figure 2). In the Cirrhosis-PTH vs. Cirrhosis comparison, 218 lncRNAs, 61 miRNAs, and 145 circRNAs were identified as differentially expressed (Supplementary Figure 3). These differentially expressed non-coding RNAs, together with the previously identified DEmRNAs, provided the basis for subsequent ceRNA network construction.

We constructed condition-specific ceRNA networks by integrating DEcircRNAs, DEmiRNAs, and DEmRNAs. In the Cirrhosis vs. Control comparison, the ceRNA network comprised 130 DEcircRNAs, 29 DEmiRNAs, and 31 DEmRNAs (Supplementary Figure 4A, Supplementary Table 5). In the Cirrhosis-PTH vs. Cirrhosis comparison, the network comprised 52 DEcircRNAs, 10 DEmiRNAs, and 16 DEmRNAs (Supplementary Figure 4B, Supplementary Table 6).

To identify key regulatory molecules underlying the anti-cirrhotic effect of PTH, we intersected the two condition-specific ceRNA networks and applied a stringent filter requiring reversed expression trends, specifically molecules that were dysregulated in cirrhosis and normalized by PTH treatment. This analysis identified a core regulatory module comprising 6 DEmiRNAs, 8 DEmRNAs, and 27 DEcircRNAs. The six core miRNAs were mmu-miR-122-5p, mmu-miR-155-5p, mmu-miR-199a-3p, mmu-miR-200c-3p, mmu-miR-223-3p, and mmu-miR-30c-5p (Figure 4(Fig. 4)). Among these, mmu-miR-122-5p exhibited the highest connectivity within the network, targeting two mRNAs (P4ha1 and Pkm) and interacting with 13 circRNAs. Notably, mmu-miR-122-5p is the most abundant and liver-specific miRNA (Bandiera et al., 2015[2]). The eight core mRNAs were P4ha1, Runx1, Inpp5d, Trp53inp1, Rrad, Pkm, Eva1a, and Jun. Functional enrichment analysis of these eight genes revealed significant enrichment of inositol-4,5-bisphosphate 5-phosphatase activity (GO:0030487), pyruvate kinase complex (GO:1902912), and the B cell receptor signaling pathway (KEGG: map04662). These genes were also associated with the activation of innate immune cells, including leukocytes and granulocytes, suggesting their involvement in immune regulation (Figure 5(Fig. 5)).

To assess the translational relevance of these findings, we investigated the distribution of human homologous genes of the above predicted DEmRNAs (P4ha1, Runx1, Inpp5d, Trp53inp1, Rrad, Pkm, Eva1a, and Jun) in human tissues based on the dataset provided by Jiang et al. We found that P4HA1, the human homolog of P4ha1, ranked second in protein abundance in the liver among all 32 tissues examined. Furthermore, the tissue specificity score of P4HA1 also ranked second among all 32 tissues (Figure 6(Fig. 6)). Except for P4HA1, the protein abundance of six genes (RUNX1, INPP5D, RRAD, PKM, EVA1A, and JUN) was at low levels in the liver (see Supplementary Figure 5). Consistent with this liver-enriched expression pattern, P4HA1 is a known target of Lufironil (also known as HOE 077), a small molecule inhibitor of prolyl 4-hydroxylase that has entered Phase II clinical trials for cirrhosis (TTD, Target ID: T61729; Drug ID: D01NCH).

Based on dose-response experiments, 0.1 mg/mL PTH and 400 μM H₂O₂ were selected for subsequent validation (Supplementary Figure 6). To directly test whether miR-122-5p regulates P4HA1, HepG2 cells were transfected with 100 nM miR-122-5p inhibitor or negative control. Transfection with miR-122-5p inhibitor significantly decreased miR-122-5p expression, confirming successful knockdown (Figure 7A(Fig. 7)). P4HA1 mRNA expression was significantly increased in the inhibitor group and was significantly decreased following PTH co-treatment (Figure 7B(Fig. 7)). In an oxidative stress-induced liver injury model, HepG2 cells were treated with H₂O₂. H₂O₂ treatment decreased miR-122-5p expression and increased P4HA1 expression (Figure 7C-D(Fig. 7)). PTH pretreatment significantly increased miR-122-5p expression compared to the H₂O₂ group (Figure 7C(Fig. 7)).

See also the supplementary data.

This research was supported by the Fundamental Research Funds for the Central Universities (YG2023LC13, YG2023QNA07), Natural Science Foundation of Fujian Province (2022J01530), National Nature Science Foundation of China (82003856, 82274025, 81773818, 81273596, 81503051), Sichuan Province science and technology plan (24NSFSC1283), Major projects of scientific and technological innovation 2030 (2021ZD0200801, 2021ZD0200800), National Key Research and Development Program of China (2022YFC2703200, 2016YFC0905000, 2016YFC1200200), Shanghai science and Technology Innovation Fund (21002411100, 20DZ2202000), Shanghai Municipal Science and Technology Major Project (2017SHZDZX01).

The manuscript was written by all authors. All authors have given approval of the final version of the manuscript. Conceptualization, writing - original draft preparation: SYQ and XYL; Conceptualization: ZXZ and ZLC; Methodology, investigation: XQZ, XLZ, HW, CH and JHZ; Data curation, formal analysis: YXL, FY, YTZ, YZW, QGX, RYC and WB; Data curation, formal analysis, MLY, ZKY and ZYC; Writing - review and editing, project administration, supervision: SYQ and ZXZ.

The RNA-seq (miRNA, circRNA and mRNA & lncRNA) datasets of 18 samples have been deposited in NCBI Gene Expression Omnibus (GEO) with accession ids GSE269011, GSE269012 and GSE269013.

The authors declare that they have no conflict of interest.

No artificial intelligence tools were used in the preparation, writing, editing, or analysis of this manuscript.