Clinical data collection

Demographic and laboratory data were collected, including age, gender, white blood cell (WBC), neutrophil count, platelet count (PLT), TBIL, aspartate transaminase (AST), alanine transaminase (ALT), albumin (ALB), creatinine, blood urea nitrogen, PT, INR, PTA, hypersensitive C-reactive protein (hs-CRP), procalcitonin (PCT), lactate (LAC), and erythrocyte sedimentation rate. ACLF risk scores were calculated. All laboratory data were collected from the most recent examinations before surgery.

Collection of liver tissue homogenate and enzyme-linked immunosorbent assay

To collect tissue homogenate, surgically excised liver tissues were freshly harvested and cut into small pieces after PBS (0.01 M, pH 7.4) flushing. The 0.1 g liver tissues were transferred into a glass homogenizer with 0.9 mL of pre-cooled PBS. After the liver tissues were thoroughly ground to homogenization, the supernatant was collected and filtered through a 0.22-μm filter to remove any impurities before detection.

Following the manufacturers’ instructions (Adsbio, Jiangsu, China), ELISA kits were used to detect the concentrations of growth differentiation factor 15 (GDF15; 2305H16), fibroblast growth factor 21 (FGF21; 2305H22), interleukin-1β (IL-1β; 2305H19), tumor necrosis factor-α (TNF-α; 2305H12), interleukin-10 (IL-10; 2305H25), transforming growth factor β (TGF-β; 2305H33), and chemokine C-C motif ligand 2 (CCL2; 2305H36) in liver tissue homogenate.

Histopathology

Surgically excised liver tissues were kept in 4% paraformaldehyde. Fixed liver tissues were embedded in paraffin, cut into 3 μm sections, and stained with H&E. The samples were finally imaged and scored under light microscopy. The histopathological scores were assessed based on the “Guideline for diagnosis and treatment of liver failure (2018).” The ACLF pathological score was given according to the extent of liver tissue necrosis. Normal liver tissue structure received a score of 0; spotty necrosis, fusion necrosis, and bridging necrosis received a score of 2; submassive hepatic necrosis affecting approximately 1/2-2/3 of hepatic parenchyma received a score of 3; and massive hepatic necrosis affecting more than 2/3 of the hepatic parenchyma received a score of 4.

Immunohistochemical staining

Immunohistochemical staining of liver sections was performed. Several 3-μm sections were prepared from paraffin-embedded blocks of tissue. Sections were dewaxed, rehydrated by fractionated alcohol series, and heated in a microwave oven for 5 min in 10 mm sodium citrate (pH 6.0) to recover antigens. After antigen retrieval, paraffin sections were placed in 3% H₂O₂ and blocked for 10 min to eliminate endogenous peroxidase activity. Then, the samples were blocked and incubated with 10% goat serum for 30 min. The rabbit anti-human HIF-1α (1:200, Bioss, Beijing, China) was added as the primary antibody and incubation was done overnight at 4 °C, in a humidified box. Then, the secondary antibody was added dropwise before incubation at 37°C for 30 min. DAB was added as the substrate to enhance color development. When the color change was observed, the staining solution was immediately washed off with tap water. Hematoxylin was counterstained for 3 min, differentiated with 1% hydrochloric acid and alcohol, and rinsed with tap water for 10 min. Gradient alcohol dehydration was performed. Xylene was transparent, and the film was mounted with a neutral gum. The number of HIF-1α positive cells was quantified, and images were captured under a microscope. The Image J software (NIH Image, Bethesda, MD, United States) was used for analysis.

Immunofluorescence staining

For double immunofluorescence staining, paraffin-embedded tissues were cut into 3.5 mm sections. Each section was deparaffinized and blocked with EDTA (pH 8.0) antigen repair solution (1:49, Aifang, Changsha, China) at 96°C for 20 min. Then, the sections were incubated overnight at 4°C with the following primary antibodies: Mouse anti-human CD68 (1:200, Aifang, Changsha, China), mouse anti-human Ki67 (1:200, Boster, Wuhan, China), rabbit anti-human HLA-DR (1:300, Abcam, Cambridge, United Kingdom), rabbit anti-human CD206 (1:200, Aifang, Changsha, China), and mouse anti-human CD163 (1:150, Aifang, Changsha, China). The samples were then incubated with FITC tag goat-anti-rabbit IgG (1:500), FITC tag goat-anti-mouse IgG (1:500), cy3 tag goat-anti-rabbit IgG (1:500), and cy3 tag goat-anti-mouse IgG (1:500) for 1 h. The sections were slightly dried and incubated with the DAPI staining solution (Solarbio, Beijing, China) at room temperature and in the dark, for 10 min. After DAPI staining for 5 min, the sections were washed with PBS three times. Each of the three washing procedureswas done for 5 min. The washed sections were then sealed with an anti-fluorescence attenuated sealing solution that contained DAPI. Fluorescence images were scanned using a digital pathology scanner (KF-FL-020, Zhejiang, China) and images were captured. The results were analyzed using the Image J software (NIH Image, Bethesda, MD, United States).

TEM

The liver tissue was dissected into 1-mm pieces and fixed in 4% paraformaldehyde and 0.1 M sodium cacodylate buffer (pH 7.2) containing 2% glutaraldehyde, at 4°C overnight. After three times of washing in buffer, the samples were post-fixed in 2% osmium tetroxide and 1% uranyl acetate for 2 h, rinsed in water, dehydrated in ascending ethanol series and 100% acetone, before it was infiltrated and embedded in eponate. Ultrathin sections were cut using a Leica EM UC7. The sections were exposed to the primary stain (5% aqueous uranyl acetate). They were then exposed to the secondary stain (lead citrate) and visualized using a 120 kv TEM HT7800 (Hitachi, Japan). Five random fields of view were imaged per group to quantify the size and number of mitochondria. Mitochondria were identified based on their morphology. Mitochondrial length, width, and cross-sectional area were measured using Image Pro-Plus 6.0 software (Media Cybernetics, Silver Spring, MD, United States). Mitochondrial count analysis was performed at the original magnification of 7000 .

Targeted metabolomics

Ultra-high performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) system (Waters XEVO TQ-S Micro, Waters Corporation) was used for targeted metabolomics analysis of 16 organic acids in liver samples. Briefly, 17 liver samples from patients with HBV-ACLF and 12 liver samples from HDs were collected and stored in an Eppendorf Safelock microcentrifuge tube. The centrifuge tube was placed in a low-temperature centrifuge and centrifuged at 12000 rpm at 4 °C for 5 min. Then, 50 µL of supernatant were taken and 50 µL of propionic acid isotope internal standard (IS; 5 μg/mL) and 50 µL of 3-nitrophenylhydrazine (250 mmol/L, prepared with 50% methanol/aqueous solution) were added to it. Thereafter, 50 µL of 1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide (150 mmol/L, prepared with 75% methanol/aqueous solution (containing 7.5% pyridine)) was added and placed in a shock mixer at 30°C for 30 min. The supernatant was extracted by adding 50 µL of 2, 6-di-tert-butyl-p-cresol methanol solution (2 mg/mL) and 250 µL of 75% methanol solution at 12000 rpm for 5 min, and then used for LC-MS analysis.

Chromatographic separation was performed using Waters ACQUITY UPLC I-CLASS ultra-high performance liquid chromatography with a Waters UPLC BEH C18 column (2.1 mm × 100 mm I.D., 1.7 µm; Waters Corp., Milford, MA, United States) at a column temperature of 40°C. The mobile phase was composed of 0.1% formic acid aqueous solution (A). The methanol: Isopropyl alcohol ratio was 8:2 (B). The flow rate was 0.30 mL/min, and the injection volume was 5 μL.

Multireaction monitoring of 16 organic acids (IS) was performed using the Waters XEVO TQ-S Micro series four-bar mass spectrometry system. The optimal Mass Spectrometric parameters were as follows: Ion source voltage was 3.0 kV, temperature was 150°C; desolvation temperature was 450°C, desolvation gas flow rate was 1000 L/h; cone-hole gas flow rate was 10 L/h.

Data acquisition and processing were conducted using the MassLynx 4.1 software (Waters Corp., Milford, MA, United States), and the concentration of each metabolite was calculated based on the standard curve. The calculated concentrations of organic acids were imported into the SIMCA software (v. 14.1, Umetrics, Sweden) for multivariate analyses, including orthogonal partial least squares-discriminant analysis (OPLS-DA) and replacement test. Finally, differential metabolites were used for pathway enrichment analysis.

Liver histopathology

Under the light microscope, the following observations were made: the structure of the liver tissue of some HDs was complete and clear; the structure of liver lobules was clear and normal; the volume of hepatocytes was uniform; the nuclei were in the middle; and the cytoplasm was red-stained. The hepatocytes were arranged radially around the central vein, and there was no sign of necrosis, steatosis, fibroplasia, and inflammatory infiltration. The liver of the others showed diffuse fatty lesions, disordered lobular structure, deranged hepatic cords, ballooned or swollen hepatocytes, loose cytoplasm, diffuse lipid droplets of different sizes in the cytoplasm, and a small number of punctate necrosis. Pathologic assessment of the liver in patients with ACLF showed destruction of the normal structure as well as collapse or incomplete collapse of reticular scaffolds. The liver tissue showed massive or submassive necrosis, along with abundant inflammatory cell infiltration. The percentage of normal hepatocytes significantly decreased. Hepatic sinusoids were significantly dilated, congested, and even hemorrhagic. Circular or triangular portal areas were observed around necrosis. Massive bile ducts, in the form of tubes or branches, were observed around the edge of the necrotic zone and the confluence area. Cholestasis and bile plugs were seen in some of the lumens. The hepatic lobules were structurally disorganized, with massive extracellular matrix hyperplasia. Residual hepatocytes were wrapped and segmented into nodules of varying sizes by fibrous septae. Hepatocytes in the residual nodules showed various degrees of hepatocyte ballooning. Some hepatocytes were eosinophilic, and apoptotic bodies and binucleated hepatocytes were visible (Figure ​(Figure1A).1A). Histopathological scoring showed that hepatocyte necrosis markedly increased in the liver of patients with ACLF compared with the HDs, and the difference was statistically significant (Figure ​(Figure1B1B).

Open in a separate window

Figure 1

Liver histopathology in healthy donors and patients with acute-on-chronic liver failure. A: HE staining of the liver tissue from healthy donors (n = 10) and patients with acute-on-chronic liver failure (n = 12); B: Histopathological Scoring. Data were compared using Mann-Whitney test. ACLF: Acute-on-chronic liver failure; HD: Healthy donor.

Upregulated hypoxia markers in patients with ACLF

The transcription factor HIF-1α mediates the adaptive response to hypoxia. In normoxic conditions, HIF-1α is hydroxylated and rapidly degraded. Hypoxia upregulates HIF-1α and promotes its translocation to the nucleus to form a complex with HIF-1β. This promotes the transcription of genes that are essential for hypoxic adaptation[6]. The expression of key mediators of hypoxia in the liver samples was examined through immunohistochemistry to assess whether patients with ACLF suffered from anoxia whammy. The results from immunohistochemical staining demonstrated that the expression of HIF-1α was elevated in the cytoplasm and nucleus of hepatocytes in patients with ACLF compared with the HDs. This showed significant hypoxic injury in patients with ACLF (Figure ​(Figure22).

Open in a separate window

Figure 2

Expression of hypoxia-inducible factor-1α increased in the liver tissues of patients with acute-on-chronic liver failure. A: Immunohistochemical staining of hypoxia-inducible factor-1α (HIF-1α); B: Semi-quantitative analysis of HIF-1α expression. Data were compared using unpaired student’s t-test. ACLF: Acute-on-chronic liver failure; HD: Healthy donor; HIF-1α: Hypoxia-inducible factor-1α.

Altered mitochondrial ultrastructure in the liver of patients with ACLF

TEM images with different magnifications are shown in Figure ​Figure3A.3A. In HDs, mitochondrial morphology, cristae, and matrix were intact and orderly arranged. Compared to normal mitochondrial conformation in HDs, mitochondrial ultrastructure was severely disorganized in the liver of patients with ACLF. Some mitochondria were severely swollen and transformed into vacuolated structures with abnormal mitochondrial morphology. The mitochondrial outer membrane was mostly incomplete, while the cristae were sparse, disorganized, or absent. The density of the mitochondrial matrix was lower and filled with an electron-dense material in patients with ACLF, suggesting extensive mitochondrial degeneration. A small number of autophagosomes and autophagic lysosomes were seen in the liver of patients with ACLF (Figure ​(Figure3A).3A). Compared with HDs, the liver of patients had an increased number of mitochondria in each microscopic field. The length, width, and area of the mitochondria also increased (Figure ​(Figure3B3B and ​andC),C), as part of the repair response after liver injury.

Open in a separate window

Figure 3

Mitochondrial morphology and function in the liver of patients with acute-on-chronic liver failure. A: Representative electron microscopy images of hepatic mitochondria from 3 patients with acute-on-chronic liver failure (ACLF) and 3 healthy donors (HDs). Orange arrows indicate autophagosome; the blue arrow indicates autophagic lysosome; B: Length, width, and area of each mitochondrion; C: Number of mitochondria at original magnification of 7000 ; D: Growth differentiation factor 15 levels in liver homogenates of 12 HDs and 17 patients with ACLF; E: Fibroblast growth factor 21 levels in liver homogenates. Data with normal distribution were compared using unpaired student’s t-test, and the Mann-Whitney test was used for non-normal data. ACLF: Acute-on-chronic liver failure; HD: Healthy donor; GDF15: Growth differentiation factor 15; FGF21: Fibroblast growth factor 21.

Mitochondrial dysfunction in patients with ACLF

TEM showed severe ultrastructural damage of mitochondria in the livers of patients with ACLF. GDF15 and FGF21 Levels in tissue homogenates were examined to reveal mitochondrial dysfunction. GDF15 and FGF21 are circulating markers for mitochondrial disorders and are widely used in diagnosis of mitochondrial diseases[7-9] and other serious diseases, such as liver failure[4,10,11], heart failure[12,13], and sepsis[14,15]. Their expression levels positively correlated with disease severity. The expression of GDF15 in liver homogenates of patients with ACLF increased compared to the HD group (Figure ​(Figure3D).3D). This revealed mitochondrial dysfunction in patients with ACLF. Nevertheless, FGF21 Levels were significantly lower in patients with ACLF compared to the HDs (Figure ​(Figure3E).3E). In contrast, plasma FGF21 levels were higher in patients with ACLF in some studies[6,11]. Some studies reported that plasma levels of FGF21 were not associated with the severity of ACLF[11].

Metabolic profiles of liver in patients with ACLF

Mitochondria drive metabolic adaption to environments. This study investigated whether mitochondrial dysfunction in patients with ACLF was accompanied by metabolic change. Therefore, the levels of organic acid metabolites in tissues of HDs and ACLF patients were determined through targeted metabolomics. A total of 16 different types of organic acids were quantified. The concentrations of the 16 organic acids were imported into the SIMCA software for OPLS-DA. The results are shown in the OPLS-DA scatter diagram (Figure ​(Figure4A).4A). The OPLS-DA model effectively differentiated between HD and ACLF samples, suggesting that there were differences in total organic acid metabolites between HDs and patients with ACLF, with lower inter-sample variability among patients with ACLF. The 200 permutation tests of our data demonstrated no overfitting in the OPLS-DA model [Q2 = (0.0, −0.474); Figure ​Figure4B].4B]. The hierarchical cluster analysis showed that the 16 organic acids could differentiate between ACLF patients and HDs (Figure ​(Figure4C).4C). Volcano plots were used to identify fumarate, methylmalonic acid, succinate, α-ketoglutaric acid, LAC, phenylacetic acid, and ethylmalonic acid as differential metabolites (Figure ​(Figure4D4D).

Open in a separate window

Figure 4

Metabolic profiles of the liver of patients with acute-on-chronic liver failure. A: Orthogonal partial least squares-discriminant analysis scatter plot; B: Permutation tests; C: Hierarchical clustergram of organic acid metabolites; D: Volcano plot comparing organic acid metabolites between patients with acute-on-chronic liver failure (n = 17) and healthy donors (n = 12); E: Quantitative analysis of differentially expressed organic acid metabolites; F: Kyoto Encylopedia of Genes and Genomes pathway enrichment analysis. Data with normal distribution were compared using unpaired student’s t-test, and the Mann-Whitney test was used for non-normal data. ACLF: Acute-on-chronic liver failure; HD: Healthy donor.

Statistical analysis of data from targeted metabolomics revealed seven differential organic acids. Compared with HDs, fumarate and methylmalonic acid levels decreased in patients with ACLF, whereas succinate, α-ketoglutaric acid, LAC, phenylacetic acid, and ethylmalonic acid levels increased in these patients (Figure ​(Figure4E).4E). The LAC /pyruvate ratio, a surrogate marker for the cytosolic NADH/NAD⁺ ratio and a hallmark of the redox state[16] was also calculated. The LAC /pyruvate ratio was significantly elevated in patients with ACLF compared to HDs (Figure ​(Figure4E),4E), suggesting enhanced anaerobic glycolysis. The liver in patients with ACLF patients used the extra-mitochondrial pathways to generate ATP. The seven differential organic acids were imported into the MetaboAnalyst website (https://www.metaboanalyst.ca/) for metabolic pathway enrichment analysis. The pathway enrichment analysis identified six pathways (P < 0.05) enriched in ACLF (Figure ​(Figure4F).4F). These pathways play key roles in glucose metabolism, fatty acid metabolism, and amino acid metabolism.

Widespread macrophages activation in the liver of patients with ACLF

Mitochondrial dysfunction affected liver metabolism in patients with ACLF. Double immunofluorescence staining with anti-ki67 and anti-CD68 antibodies was used to label proliferating macrophages and determine whether metabolic transformation affected hepatic macrophages (Figure ​(Figure5A).5A). There was no statistically significant difference in cell proliferation (Ki67⁺) between HDs and patients with ACLF (Figure ​(Figure5B).5B). However, there was significant macrophage (CD68⁺) infiltration in the liver of ACLF patients (Figure ​(Figure5C).5C). Double immunofluorescence staining with Ki67 and CD68 showed no significant differences in the number of proliferating macrophages between the groups (Figure ​(Figure5D),5D), suggesting that the infiltrating macrophages originated from peripheral circulating monocytes rather than liver-resident macrophages (i.e., Kupffer cells). CCL2, a key chemokine that regulates monocyte/macrophage migration and infiltration, was overexpressed in the liver of patients with ACLF (Figure ​(Figure5E),5E), suggesting an increased capacity of circulating monocytes for hepatic infiltration in patients with ACLF.

Open in a separate window

Figure 5

Macrophages were widely activated in the liver of patients with acute-on-chronic liver failure. A: Representative microscopic images of the double immunofluorescence staining of Ki67 and CD68 in the liver (400 fold). The red fluorescence signal represented Ki67; the green fluorescence signal represented CD68; and the blue fluorescence signal represented DAPI; B: Ki67 immunofluorescence intensity in the liver of healthy donors (n = 10) and patients with acute-on-chronic liver failure (n = 12); C: CD68 immunofluorescence intensity in the liver; D: The mean number of Ki67⁺ CD68⁺-positive cells in five × 200 fields; E: Chemokine C-C motif ligand 2 levels in liver homogenates. Data with normal distribution were compared using unpaired student’s t-test, and the Mann-Whitney test was used for non-normal data. ACLF: Acute-on-chronic liver failure; HD: Healthy donor.

Impaired macrophage polarization in the liver of patients with ACLF

Macrophages can acquire different phenotypes depending on environmental and immune signals. They are mainly classified into two major groups: (1) Classically activated macrophages endowed with pro-inflammatory and microbicidal functions and (2) alternatively activated macrophages with anti-inflammatory and tissue remodeling properties. The classically activated macrophages (CD68⁺ HLA-DR⁺) and alternatively activated macrophages (CD163⁺ CD206⁺) were labeled via double immunofluorescence staining (Figure ​(Figure6A6A and ​andB)B) to determine whether there was a phenotypic change in hepatic macrophages. The results showed that the abundance of CD68⁺ HLA-DR⁺ macrophages increased (Figure ​(Figure6C),6C), while the abundance of CD163⁺ CD206⁺ macrophages decreased in patients with ACLF compared to HDs (Figure ​(Figure6D).6D). Macrophages were polarized toward the classically activated phenotype (Figure ​(Figure6E).6E). The results from the cytokine assay were consistent with those from immunofluorescence. Compared with HDs, patients with ACLF exhibited elevated expression of IL-1β and TGF-β1 (Figure ​(Figure6F).6F). IL-1β is a CD68⁺ HLA-DR⁺ macrophage-derived pro-inflammatory cytokine and TGF-β1 is a potent pro-fibrotic factor. As a CD163⁺ CD206⁺ macrophage-derived anti-inflammatory cytokine, IL-10 was downregulated in patients with ACLF (Figure ​(Figure6F).6F). There was no significant difference in TNF-α levels between the groups (Figure ​(Figure6F6F).

Open in a separate window

Figure 6

Generalized macrophages activation showed a classically activated phenotype in the liver of patients with acute-on-chronic liver failure. A: Representative microscopic images of double immunofluorescence staining of CD68 and HLA-DR in the liver (400 fold). The red fluorescence signal represented CD68; the green fluorescence signal represented HLA-DR; and the blue fluorescence signal represented DAPI; B: Representative microscopic images of the double immunofluorescence staining of CD206 and CD163 in the liver (400 fold). The red fluorescence signal represented CD206; the green fluorescence signal represented CD163; and the blue fluorescence signal represented DAPI; C: The mean number of CD68⁺ HLA-DR⁺-positive cells in five × 200 fields; D: The mean number of CD206⁺ CD163⁺-positive cells in five × 200 fields; E: The ratio of CD68⁺ HLA-DR⁺/CD206⁺ CD163⁺-positive cells in healthy donors and patients with acute-on-chronic liver failure; F: The levels of macrophage-derived cytokines (interleukin-1β, tumor necrosis factor-α, interleukin-10, and transforming growth factor-β1) in liver homogenates. Data with normal distribution were compared using unpaired student’s t-test, and the Mann-Whitney test was used for non-normal data. ACLF: Acute-on-chronic liver failure; HD: Healthy donor; IL-1β: Interleukin-1β; IL-10: Interleukin-10; TNF-α: Tumor necrosis factor-α; TGF-β1: Transforming growth factor.

Research motivation

Currently, the mechanisms underlying immune and metabolic derangement in patients with advanced HBV-ACLF are unclear.

Research objectives

To identify the bioenergetic alterations in the liver of patients with HBV-ACLF causing hepatic immune dysregulation and metabolic disorders.

Research methods

This study evaluated the mitochondrial ultrastructure, metabolic characteristics, and immune microenvironment of the liver of the subjects.

Research results

There was extensive hepatocyte necrosis, immune cell infiltration, and ductular reaction in the liver of patients with ACLF. Hypoxia significantly changed the ultrastructure of mitochondria as evidenced by mitochondrial swelling and ridge destruction. Mitochondrial oxidative phosphorylation decreased, while anaerobic glycolysis was enhanced in patients with HBV-ACLF. Circulating monocyte-derived macrophages widely infiltrated the liver of patients with HBV-ACLF. Patients with ACLF had a high abundance of CD68⁺ HLA-DR⁺ macrophages and elevated levels of both interleukin-1β and transforming growth factor-β1 in their livers. The abundance of CD206⁺ CD163⁺ macrophages and expression of interleukin-10 decreased.

Research conclusions

Bioenergetic alteration driven by hypoxia and mitochondrial dysfunction affects hepatic immune and metabolic remodeling, leading to advanced HBV-ACLF.