Introduction
Acute-on-chronic liver failure (ACLF) is a severe clinical syndrome characterized by rapid deterioration of liver function in patients with pre-existing chronic liver disease following acute insults.1 Although definitions and diagnostic criteria for ACLF vary across regions, consensus highlights its high mortality.2,3 Globally, approximately 35% of hospitalized patients with decompensated cirrhosis develop ACLF, with an approximately 60% 90-day survival rate.4 Current treatment strategies primarily focus on organ support and management of complications. Although artificial liver support systems and liver transplantation have improved prognosis for some patients, high plasma consumption, elevated costs, increased risks of infection and bleeding, donor organ shortages, and postoperative complications remain substantial barriers.5 There is an urgent need to identify effective prognostic biomarkers and explore novel therapeutic strategies to improve outcomes in patients with ACLF.
Emerging evidence indicates that systemic inflammation and immune dysregulation are key drivers of ACLF development and progression. Macrophages have been shown to drive inflammatory progression in various models of liver injury, including ACLF.6,7 Macrophages exhibit remarkable heterogeneity and plasticity, and they can polarize into pro-inflammatory M1 or anti-inflammatory/tissue-repairing M2 phenotypes in response to distinct microenvironmental cues.8 Numerous studies have demonstrated that modulating macrophage polarization to attenuate its pro-inflammatory functions represents a promising therapeutic strategy for alleviating hepatic inflammation.9,10 Signal transducer and activator of transcription 1 (STAT1) is a key transcription factor regulating M1 polarization and is primarily activated through phosphorylation in response to interferon-gamma (IFN-γ) and other inflammatory stimuli. Once activated, STAT1 translocates to the nucleus and promotes transcription of pro-inflammatory genes such as iNOS, IL-6, and TNF-α, thereby driving macrophages toward the M1 phenotype.11 During the early stages of ACLF, both M1 and M2 macrophage subsets are activated; however, heightened pro-inflammatory activity of M1 macrophages leads to excessive production of inflammatory cytokines, contributing to hepatic injury and multiple organ failure.12
PDK4, a member of the pyruvate dehydrogenase kinase family, is widely expressed in various tissues, including the heart, liver, and skeletal muscle. It plays a pivotal role in cellular metabolism by phosphorylating specific serine residues on the pyruvate dehydrogenase E1 alpha subunit (PDHE1α), thereby inhibiting its activity. This prevents conversion of pyruvate to acetyl-CoA, limits entry of glucose into mitochondrial oxidative metabolism, and promotes glycolysis.13 Previous studies have primarily examined the role of PDK4 in metabolic diseases such as cancer, diabetes, and atherosclerosis.14,15 More recently, evidence indicates that PDK4 can also modulate inflammatory responses by regulating immune cell metabolism.16 However, the role of PDK4 in the pathogenesis of ACLF remains largely unknown.
This study investigates the function and underlying mechanisms of PDK4 in ACLF. Through multi-omics analysis, we found that PDK4 was significantly upregulated in hepatic macrophages and peripheral blood mononuclear cells (PBMCs) from patients with ACLF, and its expression levels were closely associated with disease severity and short-term mortality. Further experiments demonstrated that PDK4 promoted macrophage polarization toward the M1 phenotype, whereas PDK4 inhibition effectively alleviated liver inflammation and tissue injury in an ACLF mouse model. Mechanistically, PDK4 facilitated M1 polarization by enhancing STAT1 phosphorylation. Collectively, these findings reveal a critical role for PDK4 in regulating macrophage polarization and exacerbating liver injury in ACLF, and identify PDK4 as a potential therapeutic target for ACLF treatment.
Methods
Public transcriptomic data acquisition and analysis
The Gene Expression Omnibus (GEO; https://www.ncbi.nlm.nih.gov/geo/ ) datasets GSE168048 and GSE142255, and the Sequence Read Archive (SRA) BioProject PRJNA913603 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA913603 ) were used as data sources in this study.
The GSE168048 dataset contains transcriptomic data from PBMCs of 16 patients with ACLF, including eight non-survivors and eight survivors based on 28-day outcomes. The GSE142255 dataset comprises whole-blood transcriptomic data from 17 ACLF patients and seven healthy controls (HCs). The SRA BioProject PRJNA913603 provides single-cell RNA sequencing (scRNA-seq) data derived from liver tissues of five ACLF patients and three HCs.
Bulk RNA-seq data (GSE168048, GSE142255) were analyzed using GEO2R to identify differentially expressed genes (DEGs) between ACLF patients and HCs, or between 28-day non-survivors and survivors (|log2FC| > 1, P < 0.05). scRNA-seq data from SRA (PRJNA913603) were processed using Celescope v2.0.7 (Singleron) for demultiplexing, UMI counting, and alignment to the GRCh38 reference genome. Quality control and integration were performed in Seurat (v5.0.1),17 retaining cells with ≥200 detected genes, UMI counts below the 95th percentile, and mitochondrial content ≤ 15%. After normalization and variable gene selection, reciprocal PCA and FindIntegrationAnchors were used for data integration. PCA and clustering (resolution = 0.8) were conducted using RunPCA and FindClusters, respectively. Cell types were annotated using canonical markers. Macrophage-specific DEGs were identified with FindMarkers (adjusted P < 0.05, |log2FC| > 0.25). Gene Ontology enrichment analysis was conducted using WebGestalt,18 and significant pathways (FDR < 0.05) were visualized via bar plots of normalized enrichment scores.
Patient samples
PBMCs were collected from 22 patients with ACLF at the time of their initial hospital admission, prior to any medical intervention, and from six HCs between June 2023 and December 2024 at the Third Hospital of Hebei Medical University. In addition, liver tissue samples were collected from three patients with ACLF (resected during liver transplantation) and three control donors (normal liver tissues obtained during liver transplantation procedures). All ACLF patients received standardized medical management, including but not limited to etiological therapy, hepatoprotective and transaminase-lowering agents, nutritional support, and management of associated complications such as ascites, hepatic encephalopathy, and coagulopathy. The enrollment criteria for ACLF patients were defined according to the Guidelines for the Diagnosis and Treatment of Liver Failure (2018 Edition, China), which included the following key components: (I) acute deterioration occurring on the basis of pre-existing chronic liver disease; (II) serum total bilirubin ≥ 10 times the upper limit of normal or a daily increase > 17.1 µmol/L; (III) prothrombin activity ≤ 40% or international normalized ratio ≥ 1.5. The exclusion criteria included: (I) age < 18 years or >80 years; (II) pregnancy; (III) concomitant hepatocellular carcinoma or other malignancies; (IV) severe extrahepatic diseases; (V) incomplete clinical data. This study protocol was approved by the Ethics Committee of the Third Hospital of Hebei Medical University (Ethical Approval Number: K2024-052-1). All research procedures were conducted in accordance with the Declaration of Helsinki. Informed consent was obtained from all participants.
Animal model and treatments
Male BALB/c mice (four weeks old) were purchased from Huafukang Bioscience (Beijing, China) and housed under SPF conditions at 22°C with 50% humidity and a 12-h light/dark cycle.
All animal experiments were approved by the Animal Ethics Committee of the Third Hospital of Hebei Medical University (Approval Number: Z2024-036-2).
An ACLF model was induced in mice by combined treatment with carbon tetrachloride (CCl4), D-galactosamine (D-GalN), and lipopolysaccharide (LPS).19 After one week of acclimatization, mice were randomly divided into three groups (n = 6 per group).Control group: received vehicle only. ACLF group: chronic liver injury was induced by intraperitoneal injection of 20% CCl4 (5 mL/kg), twice weekly for 12 weeks. Acute exacerbation was triggered three days after the last dose by LPS (100 µg/kg, Sigma, USA) and D-GalN (0.5 g/kg, TargetMol). PDK4-IN group: same treatment as the ACLF group, plus oral administration of PDK4-IN-1 hydrochloride (PDK4-IN, 50 mg/kg, TargetMol, Shanghai, China) 8 h before LPS/D-GalN.20 Six hours after acute stimulation, mice were anesthetized. Blood was collected via orbital puncture, and peritoneal macrophages were harvested by PBS lavage. Additionally, liver tissues were fixed in formaldehyde and embedded in paraffin for subsequent histopathological examination.
Cell culture and treatment
The human monocytic cell line THP-1, the murine macrophage cell line RAW264.7, and the human hepatocyte cell line LO2 were purchased from Pricells Biotechnology (Wuhan, China). THP-1 and LO2 cells were cultured in RPMI-1640 medium (Gibco, NY, USA) supplemented with 10% fetal bovine serum, while RAW264.7 cells were maintained in high-glucose DMEM (Gibco, NY, USA) with 10% fetal bovine serum. All cells were incubated at 37°C in a humidified atmosphere with 5% CO2. THP-1 cells were differentiated into THP-1 macrophages using phorbol 12-myristate 13-acetate (PMA, MCE, NJ, USA). THP-1 macrophages were polarized toward the M1 phenotype by stimulation with LPS (100 ng/mL, Sigma, MO, USA) and IFN-γ (20 ng/mL, Novoprotein, Suzhou, China) for 24 h, or into the M2 phenotype by treatment with IL-4 and IL-13 (both 20 ng/mL, MCE, NJ, USA) for 24 h.21
RAW264.7 cells were polarized into M1 macrophages by stimulation with 1 µg/mL LPS for 24 h,22 or into M2 macrophages with 20 ng/mL IL-4 (MCE, NJ, USA) for 24 h.23
For PDK4 inhibition experiments, cells were pretreated with PDK4-IN at different concentrations (0.625, 2.5, and 5 µM) for 6 h, followed by LPS/IFN-γ stimulation to induce M1 polarization. For overexpression experiments, THP-1 macrophages were transfected with PDK4 overexpression plasmids (OE-PDK4) or empty vectors (OE-EV) using PolyJet™ transfection reagent (SignaGen Laboratories, MD, USA). After 24 h of transfection, cells were stimulated with LPS and IFN-γ to induce M1 polarization.
To investigate the effect of PDK4 on LO2 hepatocyte injury through the regulation of macrophage polarization, THP-1 macrophages were divided into three groups: Control, LPS/IFN-γ (L/I), and L/I + PDK4-IN. Following polarization induction, the culture medium was replaced with RPMI-1640 complete medium, and the cells were further incubated for 24 h. The supernatants were then collected, centrifuged to remove cellular debris, and mixed with complete RPMI-1640 medium at a 1:1 ratio for subsequent culture of LO2 cells.24
RNA extraction and quantitative real-time PCR (RT-qPCR)
Total RNA was extracted using TRIzol reagent (Invitrogen, CA, USA). Complementary DNA was synthesized using reverse transcription kits (Yeasen, Shanghai, China). RT-qPCR was performed using SYBR Green on an ABI 2000 system (Applied Biosystems, CA, USA). Each 20 µL reaction contained 10 µL 2× SYBR Green Master Mix, 0.5 µL forward primer, 0.5 µL reverse primer, and 9 µL diluted complementary DNA template. Relative gene expression was calculated using the 2−ΔΔCt method, with target gene expression levels normalized to the housekeeping gene β-actin and presented as fold change relative to the control group.25 Primer sequences are listed in Table 1.
Table 1Sequences of the primers used for quantitative real-time PCR
| Gene | Forward primer (5′–3′) | Reverse primer (5′–3′) |
|---|
| Human | | |
| PDK4 | GGAGCATTTCTCGCGCTACA | ACAGGCAATTCTTGTCGCAAA |
| CD86 | CTGCTCATCTATACACGGTTACC | GGAAACGTCGTACAGTTCTGTG |
| iNOS | TTCAGTATCACAACCTCAGCAAG | TGGACCTGCAAGTTAAAATCCC |
| IL-6 | ACTCACCTCTTCAGAACGAATTG | CCATCTTTGGAAGGTTCAGGTTG |
| TNF-α | CCTCTCTCTAATCAGCCCTCTG | GAGGACCTGGGAGTAGATGAG |
| CD206 | TCCTTGTGGGATTGTCCTGC | AAGCCGCTGTCTCTGTCTTC |
| β-actin | CATGTACGTTGCTATCCAGGC | CTCCTTAATGTCACGCACGAT |
| Mouse | | |
| PDK4 | AGGGAGGTCGAGCTGTTCTC | GGAGTGTTCACTAAGCGGTCA |
| CD86 | TGTTTCCGTGGAGACGCAAG | TTGAGCCTTTGTAAATGGGCA |
| iNOS | ACATCGACCCGTCCACAGTAT | CAGAGGGGTAGGCTTGTCTC |
| IL-6 | TAGTCCTTCCTACCCCAATTTCC | TTGGTCCTTAGCCACTCCTTC |
| TNF-α | AATGGCCTCCCTCTCATCAGTT | CCACTTGGTGGTTTGCTACGA |
| CD206 | GTTCACCTGGAGTGATGGTTCTC | AGGACATGCCAGGGTCACCTTT |
| β-actin | GGCTGTATTCCCCTCCATCG | CCAGTTGGTAACAATGCCATGT |
Western blotting
Cells or tissues were lysed in RIPA buffer. Proteins were separated by SDS-PAGE and transferred to nitrocellulose membranes, followed by blocking with 5% non-fat milk. Membranes were incubated with primary antibodies against PDK4 (Proteintech, Wuhan, China), CD86 (Abmart, Wuhan, China), CD206 (Huabio, Hangzhou, China), iNOS (Huabio, Hangzhou, China), STAT1 (Huabio, Hangzhou, China), p-STAT1 (Abmart, Shanghai, China), p-PDHE1α (Huabio, Hangzhou, China), and β-tubulin (Abmart, Shanghai, China), followed by HRP-conjugated secondary antibodies (Abmart, Shanghai, China). Protein bands were visualized using enhanced chemiluminescence reagents.26
Flow cytometry
Flow cytometry was employed to assess the polarization status of THP-1 macrophages. In brief, cells were incubated with fluorophore-conjugated anti-CD86 antibody (M1 marker) in the dark for 20 m. After thorough washing, the cells were resuspended in buffer and analyzed using a CytoFLEX flow cytometer (Beckman Coulter, CA, USA) to determine the distribution of different macrophage phenotypes.27
Apoptosis of LO2 cells was also evaluated by flow cytometry. Briefly, LO2 cells were co-cultured with macrophage-conditioned medium for 24 h. After harvesting, Annexin V-FITC and propidium iodide were sequentially added, followed by incubation in the dark for 20 m prior to flow cytometric analysis.28
Cell Counting Kit-8 (CCK-8) assay
The CCK-8 assay was conducted to detect changes in LO2 cell viability. LO2 cells were co-cultured with macrophage-conditioned medium for 24 h. Then, 10 µL CCK-8 reagent was added. After incubation for 4 h, the OD value was detected at 450 nm with a microplate reader, with blank culture medium as the control. Finally, cell viability was calculated, and results were expressed as percentages.28
Cellular immunofluorescence staining
Cells were seeded on coverslips, fixed with 4% paraformaldehyde for 15 m, permeabilized with 0.1% Triton X-100 for 10 m, and blocked with 5% BSA for 1 h. Cells were incubated overnight at 4°C with primary antibodies against CD68 and CD86 (Proteintech), followed by fluorophore-conjugated secondary antibodies. Nuclei were counterstained with DAPI. Images were captured using an Axioscope 5 fluorescence microscope (ZEISS, Germany), and fluorescence intensity was quantified using ImageJ software.22
Multiplex immunofluorescence immunohistochemistry
Liver tissues were fixed in 10% paraformaldehyde, embedded in paraffin, and sectioned. After deparaffinization and antigen retrieval, sections were blocked with 5% BSA and incubated with primary antibodies against PDK4 (Proteintech), CD68 (Abcam, Cambridge, UK), and CD86 (Abcam), followed by corresponding fluorophore-conjugated secondary antibodies. Nuclei were stained with DAPI. Images were captured with an Axioscope 5 fluorescence microscope (ZEISS).29
Enzyme-linked immunosorbent assay (ELISA)
Concentrations of IL-6 and TNF-α in culture supernatants and mouse serum were measured using human (Jonlnbio, Shanghai, China) and mouse (Thermo Fisher, MA, USA) ELISA kits, respectively.
Biochemical assays
Serum levels of alanine aminotransferase, aspartate aminotransferase, and total bilirubin were measured using an automatic biochemical analyzer (Rayto, Shenzhen, China) to evaluate liver function.
Histology and immunohistochemistry
Liver tissues were fixed in 10% formalin, embedded in paraffin, and sectioned for hematoxylin and eosin staining. For immunohistochemistry, sections were stained with antibodies against CD86 (Proteintech), followed by HRP-conjugated secondary antibodies and DAB development to assess the M1 polarization level.30
RNA-seq and gene set enrichment analysis (GSEA)
To investigate the mechanism by which PDK4 regulates macrophage polarization, two treatment conditions were established: macrophages stimulated with LPS/IFN-γ (control group) and macrophages pretreated with PDK4-IN (PDK4-IN-1 hydrochloride, TargetMol) for 6 h prior to LPS/IFN-γ stimulation (PDK4-IN group). Total RNA was extracted, and libraries were constructed for paired-end sequencing on the DNBSEQ platform. Raw sequencing reads were quality-controlled using SOAPnuke and aligned to the human genome (GRCh38) using HISAT2 (v2.2.1). Gene counting was performed using featureCounts. Differential expression analysis was conducted using DESeq2 (v1.40.2) with a significance threshold of Q value ≤ 0.05.31 GSEA (gsea-3.0) was applied to rank genes based on log2 fold change (log2FC) and identify enriched pathways.32 The gene set c2.cp.kegg.v6.2.symbols.gmt from the MSigDB (v6.2) database was used as a reference.33
Statistical analysis
All data are presented as mean ± standard deviation. Differences between two groups were assessed using unpaired Student’s t-test, while comparisons among multiple groups were evaluated by one-way ANOVA followed by Tukey’s post hoc test. Correlations were analyzed using Pearson’s correlation coefficient. A P-value < 0.05 was considered statistically significant. Statistical analyses were performed using GraphPad Prism 10 (GraphPad, San Diego, CA, USA).
Results
scRNA-seq reveals M1-like polarization predisposition in early-stage ACLF liver macrophages
We systematically analyzed publicly available scRNA-seq data (PRJNA913603) from liver biopsies of five patients with ACLF and three HCs. UMAP visualization identified multiple non-parenchymal cell populations, including macrophages, monocytes, T cells, B cells, mast cells, neutrophils, fibroblasts, endothelial cells, epithelial cells, blood cells, and cycling cells. Notably, distinct clustering patterns were observed between ACLF and HC samples (Fig. 1A and B).
DEG analysis of hepatic macrophages revealed 569 upregulated and 710 downregulated genes in patients with ACLF compared with HCs (Fig. 1C). Gene Ontology enrichment analysis of the upregulated genes showed significant activation of immune-related pathways, including type I interferon signaling, IFN-γ response, and antigen processing and presentation (Fig. 1D). Taken together, these findings suggest that liver macrophages in early-stage ACLF may exhibit a predisposition toward pro-inflammatory, M1-like polarization.
Elevated PDK4 expression in PBMCs and hepatic tissues correlates with adverse clinical outcomes in ACLF patients
To identify genes associated with ACLF pathogenesis and prognosis, we analyzed multiple transcriptomic datasets. In the whole-blood dataset GSE142255, 164 genes were upregulated and 224 were downregulated in patients with ACLF compared with HCs (Fig. 2A). In the PBMC dataset GSE168048, 837 genes were upregulated and 631 were downregulated when comparing 28-day non-survivors with survivors (Fig. 2B). To further screen candidate genes in macrophages, we applied more stringent criteria (absolute log2 fold change > 1, P < 0.05) to scRNA-seq data, identifying 139 upregulated and 389 downregulated genes (Fig. 2C). Integrating the DEGs from GSE142255, GSE168048, and macrophage single-cell data identified two overlapping candidate genes: PDK4 and THBS1 (Fig. 2D and E). Both genes were significantly upregulated in PBMCs from non-survivors compared with survivors (P = 0.0055 and P = 0.0281, respectively) (Fig. 2F and G). Notably, receiver operating characteristic analysis showed that PDK4 had higher predictive performance for 28-day mortality (AUC = 0.88) than THBS1 (AUC = 0.78) (Fig. 2H). Moreover, upregulation of PDK4 in macrophages from patients with ACLF was more pronounced than that of THBS1 (PDK4: log2FC = 1.36; THBS1: log2FC = 1.10). Therefore, PDK4 was selected for further investigation.
In an independent validation cohort composed of newly collected clinical PBMC samples, RT-qPCR analysis further confirmed that PDK4 mRNA expression was significantly upregulated in PBMCs from patients with ACLF compared with HCs (P = 0.0028) (Fig. 2I). Further analysis showed that its expression level was significantly positively correlated with MELD scores (R2 = 0.7932, P < 0.0001) and COSSH-ACLF II scores (R2 = 0.6057, P < 0.0001) (Fig. 2J). Consistently, immunohistochemical and multiplex fluorescence staining of liver tissues revealed a marked increase in CD68+CD86+ M1-like macrophage infiltration in patients with ACLF. Notably, PDK4 expression was prominently upregulated in these M1-like macrophages (Fig. 2K). Collectively, these results suggest that PDK4 is a key gene involved in ACLF pathogenesis and progression, potentially acting by modulating macrophage phenotypes.
PDK4 is highly expressed in M1 macrophages under ACLF conditions both in vivo and in vitro
To investigate its role in macrophage immune function during ACLF, we first assessed PDK4 expression in in vivo mouse models of ACLF. RT-qPCR analysis showed that PDK4 expression was significantly increased in PBMCs and liver tissues from ACLF mice compared with HCs (P = 0.0015 and P < 0.0001, respectively) (Fig. 3A). In peritoneal macrophages isolated from ACLF mice, expression of CD86, a classical marker of M1 polarization, was significantly elevated, and PDK4 expression was also markedly upregulated in these macrophages (P = 0.0005 and P = 0.0002, respectively) (Fig. 3B).
We next examined whether PDK4 expression was induced during M1 macrophage polarization in vitro. Stimulation of THP-1 macrophages with LPS/IFN-γ, as well as LPS treatment of RAW264.7 cells, produced a time-dependent increase in CD86 expression, confirming successful M1 polarization. Concurrently, PDK4 expression progressively increased over time under both conditions (Fig. 3C–F). Subsequently, we observed increased phosphorylation of the PDK4 downstream substrate PDHE1α (Supplementary Fig. 1A and B), suggesting enhanced PDK4 enzymatic activity.
We further assessed PDK4 expression under M2-polarizing conditions. Treatment of THP-1 macrophages with IL-4 and IL-13, or stimulation of RAW264.7 cells with IL-4, successfully induced M2 polarization, as evidenced by increased CD206 expression (Supplementary Fig. 1C and D). Notably, PDK4 expression in M2-polarized macrophages did not differ significantly from unpolarized controls (P = 0.4572 and P = 0.3153, respectively) (Supplementary Fig. 1E and F).
These findings indicate that PDK4 upregulation is closely associated with M1 macrophage polarization during ACLF, consistently observed in both in vivo and in vitro models.
PDK4 overexpression promotes M1 polarization of macrophages in vitro
To elucidate the role of PDK4 in macrophage polarization, THP-1 macrophages were transfected with an OE-PDK4 or an OE-EV. Western blot analysis confirmed that PDK4 protein expression was significantly increased in OE-PDK4 THP-1 macrophages compared with OE-EV controls (P = 0.0003) (Fig. 4A).
Following transfection, OE-PDK4 and OE-EV THP-1 macrophages were stimulated with LPS/IFN-γ to induce M1 polarization. Western blot analysis showed that PDK4 overexpression markedly increased the M1 macrophage markers CD86 and iNOS relative to the OE-EV group (Fig. 4B). Flow cytometry further showed that the proportion of CD86+ cells increased from 33.4% in the OE-EV group to 54.8% in the OE-PDK4 group (P < 0.0001) (Fig. 4C). Immunofluorescence staining also revealed significantly higher CD86 fluorescence intensity in the OE-PDK4 group compared with controls (P = 0.0024) (Fig. 4D), indicating that PDK4 overexpression promotes M1 polarization. To assess the functional consequences of PDK4 overexpression, the expression and secretion of pro-inflammatory cytokines were evaluated. RT-qPCR analysis revealed that the mRNA levels of IL-6 and TNF-α were significantly upregulated in OE-PDK4 macrophages compared with controls upon LPS/IFN-γ stimulation (P = 0.0011 and P = 0.0008, respectively) (Fig. 4E). In line with these findings, ELISA demonstrated that the secretion levels of IL-6 and TNF-α were also markedly increased in the OE-PDK4 group (P < 0.0001 for each) (Fig. 4F).
Collectively, these findings indicate that PDK4 overexpression promotes M1 polarization of macrophages and enhances expression and secretion of the pro-inflammatory cytokines IL-6 and TNF-α in vitro.
Inhibition of PDK4 attenuates M1-like polarization of macrophages and alleviates LO2 cell injury
To explore the role of PDK4 inhibition in macrophage polarization, we first assessed the cytotoxicity of a PDK4-IN in THP-1 macrophages using a CCK-8 assay. PDK4-IN at concentrations below 10 µM had no significant effect on cell viability (P = 0.1508) (Fig. 5A). Based on these results, low (0.625 µM), medium (2.5 µM), and high (5 µM) concentrations were selected for subsequent experiments. Western blot analysis showed that PDK4-IN pretreatment reduced the M1 macrophage markers CD86 and iNOS in a dose-dependent manner after LPS and IFN-γ stimulation, with the most significant suppression at the high concentration (P < 0.0001 and P = 0.0001, respectively) (Fig. 5B, Supplementary Fig. 2A). To determine the minimal effective concentration for M1 polarization inhibition, we tested lower doses (0.312 µM and 0.156 µM). CD86 expression was unchanged at these concentrations (P = 0.9973 and P = 0.4466, respectively) (Supplementary Fig. 2B), indicating that 0.625 µM is the minimal effective concentration.
Flow cytometry further confirmed that the proportion of CD86+ macrophages was significantly decreased after PDK4-IN pretreatment (5 µM, P = 0.0005) (Fig. 5C). Consistently, immunofluorescence staining showed reduced CD86 in macrophages treated with PDK4-IN (5 µM), with a significant decrease in relative fluorescence intensity (P = 0.0011) (Fig. 5D). In addition, RT-qPCR analysis showed that PDK4-IN pretreatment downregulated IL-6 and TNF-α mRNA in a concentration-dependent manner, with the most significant suppression at the high concentration (P < 0.0001 for each) (Fig. 5E). ELISA confirmed that PDK4-IN pretreatment significantly reduced secretion of IL-6 and TNF-α in supernatants of stimulated macrophages (P = 0.0007 and P = 0.0008, respectively) (Fig. 5F).
Importantly, we found that conditioned media from PDK4-IN-pretreated macrophages reduced LO2 hepatocyte apoptosis (P < 0.0001) (Fig. 5G) and increased cell viability (P = 0.0012) (Fig. 5H) compared with media from macrophages treated with LPS/IFN-γ alone.
PDK4 promotes macrophage M1 polarization via STAT1-dependent signaling
To investigate downstream mechanisms by which PDK4 regulates macrophage polarization, we performed transcriptome sequencing. The results showed that PDK4 inhibition significantly altered the gene expression profile of LPS/IFN-γ-stimulated THP-1 macrophages, identifying a total of 172 DEGs, including 46 upregulated and 126 downregulated genes (Q value < 0.05) (Fig. 6A). GSEA further showed that the JAK-STAT signaling pathway was significantly suppressed in the PDK4-IN group (NES = −1.8076, FDR q-value = 0.0313) (Fig. 6B).
Western blot analysis was performed to further validate the effect of PDK4 inhibition on STAT1 activation. Compared with control, LPS/IFN-γ stimulation markedly increased STAT1 phosphorylation, whereas PDK4-IN pretreatment significantly reduced this response (P = 0.0006) (Fig. 6C). Conversely, PDK4 overexpression further enhanced STAT1 phosphorylation after LPS/IFN-γ stimulation (P = 0.0013) (Fig. 6D).
To determine whether PDK4 promotes macrophage M1 polarization via STAT1, we examined expression of the M1 markers CD86 and iNOS. PDK4 overexpression significantly increased CD86 and iNOS protein levels, whereas treatment with the STAT1 inhibitor fludarabine effectively reversed these increases (P = 0.0110 and P = 0.0011, respectively) (Fig. 6E). These findings demonstrate that PDK4 facilitates macrophage M1-like polarization through increased STAT1 phosphorylation.
Inhibition of PDK4 attenuates hepatic injury in mice by suppressing macrophage M1 polarization
To investigate the role of PDK4 in the pathogenesis of ACLF, a mouse model was established by chronic administration of CCl4 combined with acute stimulation by LPS/D-GalN. A PDK4-IN was given orally 8 h before acute stimulation (Fig. 7A). Gross examination and hematoxylin-eosin staining showed that livers from ACLF mice exhibited extensive hemorrhage and necrosis with severely disrupted hepatic architecture. In contrast, PDK4-IN treatment markedly improved liver appearance and preserved tissue structure (Fig. 7B). Serum biochemical analysis showed that alanine aminotransferase, aspartate aminotransferase, and total bilirubin were significantly elevated in ACLF mice compared with controls, whereas PDK4-IN significantly reduced transaminases and bilirubin (P < 0.0001 for each) (Fig. 7C). Consistently, serum cytokine analysis indicated that TNF-α, IL-1β, and IL-6 were markedly increased in ACLF mice, and PDK4-IN significantly suppressed these cytokines (P < 0.0001, P = 0.0004, and P < 0.0001, respectively) (Fig. 7D).
Immunohistochemical analysis showed that the M1 macrophage marker CD86 was significantly upregulated in livers of ACLF mice compared with controls, whereas PDK4-IN treatment markedly reduced the area of positive staining for this marker (Fig. 7E). Consistently, RT-qPCR analysis showed that PDK4-IN treatment significantly reduced mRNA expression of both CD86 and iNOS in livers of ACLF mice (P = 0.0478 and P = 0.0011, respectively) (Fig. 7F). Similarly, Western blot analysis revealed that CD86 and iNOS protein levels were also significantly decreased (P = 0.0106 and P = 0.0025, respectively) (Fig. 7G). In peritoneal macrophages, PDK4-IN also downregulated mRNA expression of CD86, iNOS, and IL-6 (P = 0.0079, P < 0.0001, and P = 0.0012, respectively) (Fig. 7H). Collectively, these results suggest that PDK4 inhibition attenuates liver injury and systemic inflammation in ACLF mice, at least in part, by suppressing macrophage M1 polarization.
Discussion
ACLF is a life-threatening clinical syndrome characterized by systemic inflammation and immune dysregulation, with a markedly high mortality rate.34,35 Emerging evidence indicates that macrophage polarization plays a pivotal role in immune imbalance and disease progression during ACLF.36,37 In this study, we identified PDK4 as a crucial modulator of macrophage function that contributes to the development and progression of ACLF.
Previous studies have shown that PDK4 is upregulated in various inflammatory diseases, such as sarcoidosis and septic cardiomyopathy, and its expression correlates with disease severity.38,39 However, studies have also reported downregulated PDK4 expression in synovial cells from rheumatoid arthritis, suggesting that its function may be disease-specific.40 Our study revealed that high PDK4 expression in PBMCs was significantly associated with poor clinical outcomes in patients with ACLF, suggesting that PDK4 may serve as a biomarker reflecting the extent of inflammation and disease progression, with prognostic value. Notably, PDK4 is not only upregulated in various inflammatory diseases but may also actively contribute to inflammation and disease progression by modulating immune cell function. PDK4 influences multiple immune cell types, including macrophages and T cells.16,41 Specifically, PDK2 and PDK4 act as metabolic checkpoints of macrophage polarization. Genetic deletion or pharmacologic inhibition of these kinases significantly attenuated M1 polarization in high-fat diet–induced models of insulin resistance.42 Moreover, advanced glycation end products promoted M1 macrophage polarization by activating the HIF-1α/PDK4 signaling pathway.43 Consistently, our findings showed that PDK4 was markedly upregulated under M1-polarizing conditions, highlighting its critical role in macrophage functional programming. PDK4 overexpression led to a pronounced increase in M1-associated markers, including CD86 and iNOS, and enhanced secretion of key pro-inflammatory cytokines such as TNF-α and IL-6. Persistent elevation of these cytokines is closely linked to immune dysregulation, exacerbated hepatic injury, and multiorgan failure, and their circulating levels correlate positively with poor short-term prognosis, serving as predictors of early mortality.44–46 Conversely, pharmacologic inhibition of PDK4 significantly suppressed M1 polarization and reduced production of pro-inflammatory mediators, highlighting its essential role in sustaining the M1 phenotype and amplifying the inflammatory cascade.
Inhibition of macrophage M1 polarization has been recognized as an effective strategy to alleviate liver injury. Previous studies showed that mesenchymal stem cell therapy could alleviate liver inflammation and injury by suppressing M1 polarization.47 More recently, a traditional Chinese herbal formulation, Chishao-Fuzi, was reported to improve liver function, coagulation abnormalities, and histopathologic damage in ACLF rats through a similar mechanism.36 In our in vitro experiments, supernatants from conventional M1 macrophages significantly induced injury in LO2 hepatocytes, whereas supernatants from PDK4-inhibited M1 macrophages showed markedly reduced cytotoxicity, suggesting that PDK4 inhibition could attenuate M1-mediated hepatocellular injury. In vivo, treatment with a PDK4 inhibitor in ACLF mice reduced intrahepatic infiltration of M1 macrophages and ameliorated inflammatory responses and histopathologic alterations, further supporting the therapeutic potential of targeting PDK4 to inhibit M1 macrophage polarization in ACLF.
Previous studies have established that STAT1 activation is a key mechanism driving macrophage polarization toward the pro-inflammatory M1 phenotype.11,48 Our study found that PDK4 overexpression significantly promoted M1 polarization and enhanced STAT1 phosphorylation, whereas PDK4 inhibition concurrently reduced STAT1 phosphorylation and the M1 polarization capacity of macrophages. Although PDK4 was previously thought to drive M1 polarization primarily through enhanced glycolysis and metabolic reprogramming, our findings reveal that it also facilitates M1 polarization by activating the STAT1 signaling pathway, suggesting a novel regulatory mechanism and providing new insights into its immunomodulatory function.
Notably, the precise mechanism by which PDK4 regulates STAT1 remains to be defined. As a metabolism-associated enzyme, PDK4 promotes glycolysis, suppresses oxidative phosphorylation, and induces mitochondrial dysfunction. Based on current evidence, we hypothesize that PDK4 may indirectly activate STAT1 through enhanced glycolysis and the consequent burst of reactive oxygen species, a notion supported by multiple studies.49,50 Moreover, we cannot exclude the possibility that PDK4 directly modulates STAT1 activity via non-metabolic mechanisms, such as protein–protein interactions.
From a clinical translation perspective, targeting PDK4 holds substantial therapeutic potential. By specifically inhibiting M1 polarization and the ensuing cytokine storm, this strategy may markedly attenuate hepatocellular injury and systemic inflammation, as reflected by reduced transaminases, bilirubin, and inflammatory markers such as IL-6. Intervening in this pivotal pathway is also expected to mitigate the progression of multiple organ failure, thereby potentially improving short-term survival. Future studies should prioritize the development of liver-targeted nanoparticle delivery systems to enhance drug specificity and explore combination strategies with existing treatments, such as artificial liver support systems. Further validation of this pathway across ACLF models of diverse etiologies will facilitate clinical translation.
We acknowledge several limitations. First, given the multifaceted roles of PDK4 in metabolic and inflammatory regulation, its effects may not be confined to macrophages and could involve non-macrophage-mediated mechanisms that warrant investigation. Second, although this study focused on the correlation between PDK4 and STAT1 signaling, the causal molecular mechanisms remain incompletely defined and require further validation through in-depth experiments. Finally, although regulation of STAT1 signaling by PDK4 has not yet been validated in animal models, the significant immunomodulatory effects observed in the ACLF model suggest that PDK4 represents a promising therapeutic target in ACLF and merits rigorous exploration.