Dysregulation of lipid metabolism in the pseudolobule promotes region-specific autophagy in hepatitis B liver cirrhosis

Background:

Chronic hepatitis B (CHB) infection leads to liver cirrhosis (LC), the end stage of liver fibrosis. The precise diagnosis and effective therapy for hepatitis B cirrhosis are still lacking. It is highly necessary to elucidate the metabolic alteration, especially the spatial distribution of metabolites, in LC progression.

Methods:

In this study, LC-MS/MS together with an airflow-assisted ionization mass spectrometry imaging system was applied to analyze and compare the metabolites’ spatial distribution in healthy control (HC) and hepatitis B LC tissue samples. The liver samples were further divided into several subregions in HC and LC groups based on the anatomical characteristics and clinical features.

Results:

Both the LC-MS/MS and mass spectrometry imaging results indicated separated metabolite clusters between the HC and LC groups. The differential metabolites were mainly concentrated in lipid-like molecules and amino acids. The phosphatidylcholines (PCs), lysoPCs, several fatty acids, and amino acids reduced expression in the LC group with region specific. Acyl-CoA thioesterase 2 and choline/ethanolamine phosphotransferase 1, which regulate PC and fatty acid metabolism, were significantly decreased in the pseudolobule. Meanwhile, the increased expression of LC3B and p62 in the pseudolobule indicated the upregulation of autophagy.

Conclusions:

Hepatitis B LC induced region-specific autophagy by increasing the expression of LC3B and p62 in the pseudolobule and by dysregulation of unsaturated fatty acids, amino acids, and PC metabolism. The mass spectrometry imaging system provided additional metabolites’ spatial information, which can promote biomarker screening technology and support the exploration of novel mechanisms in LC.

INTRODUCTION

Hepatitis B infection is one of the most common reasons for chronic liver diseases worldwide. During the long-term infection, the liver suffers from continuous injury and forms fibrosis (FB), which further becomes cirrhosis, the end stage of FB.¹ Chronic hepatitis B (CHB) infection leads to liver cirrhosis (LC) and HCC, affecting over 250 million patients and resulting in >1 million deaths every year worldwide.² In addition, LC negatively affects the patient’s quality of life and also increases the cost of treatment and the economic burden on the family.³ Therefore, the effective treatment of HBV-related LC can not only improve the quality of life and extend the life span of patients but also reduce spending on therapy, which can relieve the economic burden on the family and the government.⁴ However, an efficient therapy for HBV-related LC is still lacking, and its disease progression and mechanism have not been elucidated yet. Thus, finding novel molecular and pivotal altering metabolic pathways or metabolites is of high importance, which can further promote the development of LC treatment.

The liver plays a key role in producing metabolites on a large scale, including a variety of fatty acids, plenty of amino acids and their derivates, and also small molecules.⁵ Based on the anatomical structure, the functional working and repeating units of the liver are hepatic lobules (HLs) with a hexagonal shape and a diameter of around 1 mm in humans.⁶ In each HL, portal nodes consist of a portal vein (PV), hepatic artery, and bile duct (BD). The blood enters the portal nodes and drains into the central vein (CV) through sinusoidal blood vessels. The bile acids are secreted by the hepatocytes and transported outward in the opposite direction to the blood flow, through the bile canaliculi to the BD. Taken together, the difference in blood and bile acids’ transport directions leads to the formation of a unique spatial microenvironment inside the liver and supports the maintenance functions of the liver.⁷ Nevertheless, on the contrary, in LC conditions, the HL becomes fibrogenic, and with disease progression, it eventually becomes a pseudolobule (PL) with cirrhosis. During the disease progression, the portal nodes and CV are squeezed and lose their original shape and functions; this condition is termed a spatial microenvironment alteration. Thereby, unveiling the heterogeneity in LC would be another direction to explore from the incidence of HBV infection to the end stage of HBV-related LC.

It is important to understand the severity of LC and also the diverse ranges of potential outcomes, which are critical for predicting the therapy outcomes and personalized treatment projects. The heterogeneity and pathogenesis of HBV-related LC have been closely linked to a complicated gene-environment variation, metabolic disorders, and the alteration of the spatial liver environment.⁸ Herewith, to better understand the cirrhosis progression, it is necessary to uncover the relationship between portal hypertension and related circulatory changes to the metabolic pathways and gene expression alteration, for instance, to explore the role of these changes in therapeutic strategies and clinical outcomes.⁹ Metabolomics, as one of the quantitative measurement approaches toward multiple parametric metabolic shifts and responses in living systems, has been widely agreed as an efficient study strategy for the pathophysiological changes associated with different disease stages or injury conditions.^10,11 The rapidly developed metabolomics method has become a robust and powerful technique in clinical research, which involves screening biomarkers, characteristic phenotypes, and stages of the disease, and further discovery mechanism pathways.¹² Characterizing the spatial alteration of metabolites and their ruched pathway would reveal the heterogeneity of LC and provide a perspective viewpoint in HBV-related LC metabolic enzyme discovery.

Overall, current metabolomics studies of HBV-related LC mainly focus on blood and urine samples, which provide bulk information on metabolites within the whole-body circulation but not on the specific characteristics of liver tissue lesions. Missing spatial region information leads to difficulty in identifying the key factors that affect LC progression, elucidating the special microenvironment of LC lesions, and revealing precise mechanisms. Mass spectrometry imaging (MSI), as a modern technique, has shown robust power in several fields, such as in spatial cancer metabolite alteration study and in situ drug molecule distribution and dynamics.^13,14 This way, alterations of small molecules can be directly correlated to anatomical features. Thus, MSI may facilitate the development of personalized medicine, faster diagnosis, and further understanding of the heterogeneity of the liver. In this study, we use an airflow-assisted ionization mass spectrometry imaging (AFAI-MSI) system to evaluate the metabolites’ spatial distribution in healthy and HBV-related LC tissues and reveal the alteration of metabolites involved in LC disease progression.

METHODS

Chemicals and reagents

HPLC grade formic acid was obtained from Merck (Muskegon), and acetonitrile was purchased from Thermo Fisher. Other reagents, unless otherwise specified, including cryo-gel, 0.5% eosin solution, and hematoxylin, were purchased from Sigma-Aldrich. Acyl-CoA thioesterase 2 (15633-1-AP), choline/ethanolamine phosphotransferase 1 (20496-1-AP), and p62 (18420-1-AP) primary antibodies were obtained from Proteintech. LC3 (83506) primary antibody was purchased from Cell Signalling Technology.

Sample collection and preparation

In total, 7 healthy liver samples were obtained from patients who suffered abdominal trauma or a car accident, together with 15 LC samples that were obtained after surgery at the Shenzhen People’s Hospital and The Third People’s Hospital of Shenzhen from January 2021 to May 2022. The study was conducted in accordance with the Declarations of Helsinki and Istanbul, and the ethic was approved by the Ethics Committees of the Shenzhen People’s Hospital (LL-KY-2022083) and The Third People's Hospital of Shenzhen (G2022027). Informed consent was obtained from all participants. The samples for MSI were embedded in the optimal cutting temperature compound and cut to 10 µm thickness slides with a Leica CM1950 cryostat microtome and stored at −80°C until analysis. For the metabolomics study, samples were homogenized and extracted with an isotope-labeled internal standard mixture extraction solution (acetonitrile:methanol = 1:1).

Tissue spatial region group selection

For further spatial metabolomics analysis, the tissue spatial regions were selected based on anatomical units. In the healthy control (HC) group, 4 subgroups (HL, PV-HC, CV-HC, and BD-HC), were selected for spatial analysis. Moreover, for the LC group, PL was replaced with HL, FB was the boundary area around PL, and the relative normal (RN) was the region close to the healthy condition, plus the CV-LC and BD-LC. Besides, in the LC group, based on the higher Child-Pugh score and bile siltation, the 4 samples were further separated into cirrhosis higher and lower groups (LC-H and LC-L).

AFAI-MSI-QE

All of the MSI assays were carried out with a custom-made AFAI ion source equipped with a Q-Exactive mass spectrometer (Thermo Scientific). The spray solvent was a mixture of acetonitrile and water (8:2, vol/vol) plus 0.1% formic acid in the positive mode at the flow rate of 5 μL/min. Nitrogen gas was sprayed at 45 L/min, and the capillary temperature was 350°C. The MSI analysis was based on scanning the surface of the tissue at a scan rate of 200 μm/s in the x-direction together with a vertical step of 100 μm in the y-direction, with a total scan area of 1 cm × 1 cm.

LC-MS/MS

LC-MS/MS analysis was performed using a UHPLC system (Vanquish; Thermo Fisher Scientific) with a Waters ACQUITY UPLC BEH Amide column (2.1 mm×100 mm, 1.8 μm) coupled to an Orbitrap Exploris 120 mass spectrometer (Orbitrap MS; Thermo). The chromatography was operated in the gradient mode with an injection volume of 2 μL and the flow rate at 0.5 mL/min. Solvent A consisted of 25 mmol/L ammonium acetate and 25 mmol/L ammonia in water, and solvent B was acetonitrile. The gradient was started with 95% of solvent B for 0.5 minutes, then decreased linearly from 95% to 65% B over 0.5–7 minutes and from 65% to 40% B over 7–8 minutes, held at 40% B from 8 to 9 minutes, and then reversed back to 95% B over 9–12 minutes. The column oven temperature was 30°C, and the autosampler temperature was 4°C. The MS/MS spectra were collected with Orbitrap Exploris 120 mass spectrometer based on the information-dependent acquisition mode using the Xcalibur 4.4 acquisition software (Xcalibur; Thermo). The electrospray ionization source conditions were set as follows: sheath gas flow rate 50 Arb, Aux gas flow rate 15 Arb, capillary temperature 320°C, full MS resolution 60,000, MS/MS resolution 15,000, collision energy 10/30/60 in the NCE mode, and spray voltage 3.8 kV (positive) or −3.4 kV (negative).

Immunohistochemistry

The tissue section slides were stained using hematoxylin and eosin (HE staining) based on the previous study.¹³ The images were captured using a Fenghuang microscope at ×40. The study was conducted in accordance with the Declarations of Helsinki and Istanbul.

Immunofluorescence

The tissue section slides were fixed for 20 minutes with 4% paraformaldehyde at room temperature and permeabilized with 0.5% Triton for 5 minutes. After PBS washing, the slides were blocked with 3% BSA (Sigma) for 1 hour. The primary antibodies (CST; Proteintech) were applied for a certain concentration and incubated at 4°C overnight. The slides were washed, and secondary antibodies (Invitrogen) added and incubated for 1 hour in the dark. Finally, the slides were mounted and left overnight in the dark. The images were captured with a Zeiss LSM900 confocal microscope and quantified with Fiji ImageJ software (1.53C).

Data analysis

The raw data were converted to the mzXML format using ProteoWizard and processed with an in-house program, which was developed using R and was based on XCMS, for peak detection, extraction, alignment, and integration. The metabolite annotation was applied together with HMDB, MONA, METLIN, and an in-house MS2 database (BiotreeDB). The cutoff for annotation was set at 0.3. The differential metabolites were selected based on p < 0.05 (Student t test) and an orthogonal partial least-squares discriminant analysis (OPLS-DA) VIP (variable importance in projection) >1.

For MSI, the raw data files were converted into .cdf format and applied to homemade imaging software (MassImager) for further data analysis and image reconstruction based on the previous report.¹⁵ The region of interest was extracted accurately by matching with the HE staining image and developing 2-dimensional data matrixes. The metabolites detected by AFAI-MSI were annotated using the pySM pipeline and an in-house SmetDB database (Lumingbio). The evaluation and identification of endogenous metabolites in the region of interest were done using OPLS-DA and partial least-squares discriminant analysis. The differential metabolites were obtained based on p < 0.05 (Student t test and 1-way ANOVA Dunnett test) and VIP >1.¹⁶ Statistical analysis was done using GraphPad Prism 9.0. The results were shown as mean±SD.

RESULTS

Characteristics and stage of LC

The demographic and clinical indices for LC-MS/MS and AFAI-MSI studies are summarized in Supplemental Tables S1 and S2 (http://links.lww.com/HC9/A406), respectively. All of the participants had HbsAg-positive results, which indicated HBV infection. The clinical test results showed similar trends for all patients, such as decreased total protein (TP), albumin (ALB), prealbumin (PA), cholinesterase (CHE), and antithrombin III (AT III). Combining these with the ultrasound results indicated that all patients suffered from HBV-related LC. In addition, the Child-Pugh and Model of End-stage Liver Disease scores were calculated to demonstrate the LC condition and reflected cirrhosis progression in AFAI-MSI patients. Based on the higher level of Child-Pugh (level C) and Model of End-stage Liver Disease scores together with HE staining, the 4 patients were further divided into 2 groups: cirrhosis higher (LC-H) with higher scores and bile siltation, and cirrhosis lower (LC-L) with relatively lower scores and without bile siltation. These groups were used to evaluate the alteration of metabolites with the progression of cirrhosis.

Metabolite alteration detected by LC-MS/MS in LC

In total, 15 LC and 4 HC tissue samples were used for LC-MS/MS analysis. The OPLS-DA results showed that these 2 groups of samples had distinct features (Figure 1B); the differential m/z values were more concentrated in downregulated metabolites (1344), and only 161 were upregulated (Figure 1C). All of the identified metabolites were attributed to 11 categories; the top 3 were organic acids and derivatives (42.86%), lipids and lipids-like molecules (21.43%), and organoheterocyclic compounds (11.11%), respectively (Figure 1D). The metabolites that had the top Z-scores were amino acids, glycerophospholipids, and organonitrogen compounds; most compounds had higher expression in the LC group except phosphatidylcholine (PC) (22:1(13Z)/14:0) (Figure 1E, G). The Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis showed that the top ruched pathways were arginine biosynthesis, glycerophospholipid metabolism, and metabolic pathways (Figure 1F). The metabolomic study showed that in LC metabolic dysregulation existed, and the study mainly focused on lipids and amino acid metabolism. Thereby, looking into the spatial details of metabolic disorder would provide the spatial dynamic changes of metabolic pathways.

FIGURE 1.

The metabolites detected by LC-MS/MS in HC and LC tissue samples. (A) The workflow of LC-MS/MS and AFAI-MSI. (B) OPLS-DA comparison of HC and LC groups’ data based on LC-MS/MS. (C) The volcano plot of differential metabolites. (D) Summary of a superclass and a class of detected metabolites. (E) Z-score of differential metabolites. (F) KEGG of enrichment pathways of differential metabolites. (G) The expression of key metabolites. Abbreviations: AFAI-MSI, airflow-assisted ionization mass spectrometry imaging; HC, healthy control; KEGG, Kyoto Encyclopedia of Genes and Genomes; LC, liver cirrhosis; LC-MS/MS, Liquid Chromatography Tandem Mass Spectrometry; OPLS-DA, orthogonal partial least-squares discriminant analysis; PC, phosphatidylcholine; PLS-DA, partial least-squares discriminant analysis; VIP, variable importance in projection.

Metabolite alteration in LC

Postoperative LC and HC tissue sections were divided into several histologic groups based on anatomical units, and the general workflow is shown in Figure 1A. The HE staining images were captured to support the anatomical information, combining both MSI and HE images, which delivers both the anatomical characters and the metabolite signatures’ spatial resolution of MSI in positive (Figure 2A) and negative (Supplemental Figure S1a, http://links.lww.com/HC9/A344) ionization modes. Based on each pixel point of MSI, the OPLS-DA was used to analyze the discrepancy between the 2 selected regions. For the whole HC and LC sections, the results indicated that the metabolites between HC and LC were significantly different and separated (Figure 2 for the positive mode and Supplemental Figure S1b, http://links.lww.com/HC9/A344 for the negative mode). Moreover, in the LC group, the metabolites showed obvious 2 clusters for LC-L and LC-H groups. This further confirmed that, during the progression of the disease, the change of metabolites occurred and was significant. The total number of metabolites identified in the HC and LC groups based on positive and negative ionization modes was 365, 186, 391, and 220, respectively (Figure 2C, Supplemental Figure S1c, http://links.lww.com/HC9/A344). All of the identified metabolites were separated into 12 categories, which were similar to the LC-MS/MS results: mainly lipids and lipid-like molecules, organic acids and derivatives, organoheterocyclic compounds, and organic oxygen compounds. Of all the identified metabolites, more were detected in the positive ionization mode compared with the negative mode (Figure 2D). While comparing the metabolites in these 2 groups, both HC and LC groups were found to have similar metabolites, and 359 and 185 compounds were the same in positive and negative modes, respectively. Only 31 and 36 metabolites were unique in the LC group in positive and negative ionization modes, respectively. To eliminate the effect of the dependency of heteroscedasticity based on the ion signal intensity, the peaks of compounds from all pixel points were selected and subjected to log transformation and then OPLS-DA analysis. The differential metabolites were selected based on VIP >1 in the OPLS-DA analysis. The differential metabolites showed distinctly opposite expression panels. Most of the metabolites in the HC groups showed an increased expression, while the same compounds in the LC group showed decreased expression (Figure 2E, Supplemental Figure S1f, http://links.lww.com/HC9/A344). In addition, the correlation analysis showed that proline betaine had a negative correlation with all other metabolites in the positive mode (Figure 2F). Moreover, in the negative ionization mode, most of the compounds were positive and closely correlated with other metabolites (Supplemental Figure S1g, http://links.lww.com/HC9/A344).

FIGURE 2.

Total metabolites detected by AFAI-MSI in HC and LC tissue samples. (A) HE and MSI diagrams of HC and LC sample sections in positive ionization mode. (B) OPLS-DA comparison of AFAI-MSI data based on positive ionization mode. (C) Total metabolites detected by the AFAI-MSI method in HC and LC groups. (D) Metabolites’ difference between positive and negative ionization modes in HC and LC samples. (E) The differential metabolite expression of HC and LC groups. (F) The correlation of differential metabolites. Abbreviations: AFAI, airflow-assisted ionization; HC, healthy control; HE, hematoxylin and eosin; LC, liver cirrhosis; MSI, mass spectrometry imaging; OPLS-DA, orthogonal partial least-squares discriminant analysis.

Metabolites’ spatial distribution in HC and LC subregions

The advantage of the AFAI-MSI study is that it can provide specific region information when combined with HE staining results. Therefore, it is worthwhile to explore the subregion metabolites’ shift in HC and LC groups to reveal the exact metabolites and relative pathway changes during LC progression. Figure 3A and Supplemental Figure S2a (http://links.lww.com/HC9/A345) show the MSI of different subregions in the HC group in both positive and negative ionization modes, respectively, while pointing out differences in metabolite expression. The metabolites were shared within the subregions without any significant difference (Figure 3B–D for the positive mode and Supplemental Figure S2b–d, http://links.lww.com/HC9/A345 for the negative mode); only some metabolites showed higher Z-scores in certain subregions, such as palmitic acid and anabasine in the HL region (Figure 3E for the positive mode and Supplemental Figure S2e, http://links.lww.com/HC9/A345 for the negative mode). Moreover, with these key metabolites, the main ruched pathways were choline metabolism in cancer, glycerophospholipid metabolism, beta-alanine metabolism, and arginine and proline metabolism based on KEGG analysis (Figure 3F, supplemental Figure S2f, http://links.lww.com/HC9/A345). Compared with the whole LC and HC section results, within the HC subregion, the metabolite alteration was not as obvious as in the whole section comparison. Unlike the inconspicuous changing of metabolites among subregions in the HC group, in the LC group, the metabolites showed clearly different expressions. Among the 6 subregions, >262 metabolites were shared within the LC sample’s subregion, at least 36 and 29 unique compounds were identified in the RN group, and the highest numbers of specific metabolites found in the PL group were 89 and 63 for positive and negative modes, respectively (Figure 4A, Supplemental Figure S3a, http://links.lww.com/HC9/A346). In addition, unlike in the HC group, the changing expression of metabolites was found within the subregions in LC. Histamine and dihydrogenistein showed a significant increase, while allantoic acid and melezitose decreased in PV and PL regions (Figure 4B, Supplemental Figure S3b, http://links.lww.com/HC9/A346). Moreover, the metabolites’ alteration within the subregions shared similar trends. Like glycerophosphocholine and propionylcamitine had the highest expression level in RN regions might indicate healthy condition metabolic level yet decreased in PL and FB regions (Figure 4C, Supplemental Figure S3c, http://links.lww.com/HC9/A346). These results reflected that among the subregions of LC, the metabolites had distinct alteration ways. For the KEGG pathway analysis, the main pathways involved in the LC subregions were not similar to those of the HC group. The top 5 ruched pathways were choline metabolism in cancer, retrograde endocannabinoid signaling, Fc epsilon RI signaling pathway, regulation of lipolysis in adipocytes, and glycerophospholipid metabolism (Figure 4D, Supplemental Figure S3d, http://links.lww.com/HC9/A346). The pathway activity of the top ruched pathway shown in PL and CV regions was the highest. On the contrary, BD and PV had relatively lower pathway activities (Figure 4E). The possible reason for the difference in pathway activities is that the PL region would be affected mostly by HBV, and the structure is shifted from a healthy HL to that of a high collagen-contained and FB false lobule. During this progression, the metabolic pathways would be changed the most. Moreover, the metabolites that participated in glycerophospholipid metabolism and α-linoleic acid metabolism, such as arachidonic acid and PC (14:0/20:2(11Z,14Z)), were significantly decreased (Figure 4F). Taken together, within the anatomical structure of the false lobule, the key metabolites and pathways showed spatial alteration; these would provide additional therapy targets specific to the disease region.

FIGURE 3.

The alteration of metabolites’ spatial distribution in subregions of the HC group. (A) HE and MSI images of different subregions in HC samples based on the positive ionization mode. (B) Positive ionization mode of PLS-DA analysis for HC subregions. (C) Metabolite variation among different HC subregions. (D) Heatmap of differential metabolites based on VIP > 1. (E) Z-score of differential metabolites. (F) Alteration of metabolic pathways among subregions of HC. Abbreviations: BD, bile duct; CV, central vein; HC, healthy control; HE, hematoxylin and eosin; HL, hepatic lobule; MSI, mass spectrometry imaging; PLS-DA, partial least-squares discriminant analysis; PV, portal vein; VIP, variable importance in projection.

FIGURE 4.

The metabolites’ spatial distribution in subregions of LC samples. (A) Metabolite variation among different subregions of LC. (B) Comparison of differential metabolites in LC subregions. (C) The gradient expression of differential metabolites in subregions. (D) Alteration of metabolic pathways among LC subregions. (E) Pathway activity pattern in subregions. (F) The network of key metabolic pathways and related metabolites. Abbreviations: BD, bile duct; CV, central vein; LC, liver cirrhosis; PL, pseudolobule; PV, portal vein; RN, relative normal.

Alteration of metabolites in cirrhosis subregions

The metabolic enzymes are pivotal factors for biological network systems and are considered as key modulators to mediate complicated metabolic reactions and compound activity, which are always recognized as potential targets for new drug discovery. To demonstrate the metabolites changing along with cirrhosis progression, time-series analysis was applied to explore metabolite expression trends among HL, RN, FB, and PL subregions. In total, 28 clusters were identified. Among all the clusters, 5 clusters (14, 17 in positive mode, and 1, 8, and 10 in negative mode) of the metabolites presented a progressively increasing tendency. On the contrary, 5 clusters (8, 20 in positive mode, and 9, 13, and 17 in negative mode) showed gradually decreasing trends (Figure 5A, supplemental Figure S4a, http://links.lww.com/HC9/A347). While overlapping the metabolites of these 10 clusters with differential metabolites, 32 compounds were found to be the same, and the metabolic pathways mainly focused on were glycerophospholipid metabolism and biosynthesis of unsaturated fatty acids. While the uric acid in cluster 14 showed a gradually increasing expression, the docosahexaenoic acid in cluster 9 gradually reduced with spatial distribution (Figure 5C, supplemental Figure S4c, d, http://links.lww.com/HC9/A347). The ruched pathways had more similar panels in FB and PL compared with HL (Figure 5B, Supplemental Figure S4b, http://links.lww.com/HC9/A347). The top ruched pathways were biosynthesis of unsaturated fatty acids, fatty acid biosynthesis, and α-linoleic acid in the negative ion mode (Figure 5B), and choline metabolism in cancer and glycerophospholipid metabolism in the positive ion mode (Supplemental Figure S4b, http://links.lww.com/HC9/A347). The different panels of pathways in RN, FB, and PL can be attributed to the variation in the subregion’s disease progression and the RN region being closer to the healthy condition compared with the FB and PL areas. Expression of acyl-CoA thioesterase 2 and choline/ethanolamine phosphotransferase 1 was significantly decreased in PL, which indicated the dysregulation of fatty acid biosynthesis (Figure 5D, E). Taken together, comparing the metabolites in HL, RN, FB, and PL subregions, the clear patterns of rising or reducing expression of certain metabolites would correlate with LC progression.

FIGURE 5.

Comparison of metabolite alteration between HC and LC groups. (A) Time-series analysis of key metabolites’ expression in disease-related regions based on the negative ionization mode. (B) Alteration of metabolic pathways of HBV infection progression in LC regions (blue background: HL vs. RN, green background: HL vs. fibrosis, and red background: HL vs. PL). (C) Key metabolites’ spatial expression in LC regions based on the negative ionization mode. (D) CEPT1 and ACOT2 expression levels in HC and LC groups. (E) The quantification of CEPT1 and ACOT2 foci in 2 groups. *p < 0.05, and ****p < 0.0001. Abbreviations: ACOT2, acyl-CoA thioesterase 2; CEPT1, choline/ethanolamine phosphotransferase 1; HC, healthy control; HL, hepatic lobule; LC, liver cirrhosis; PL, pseudolobule; RN, relative normal.

Metabolite changing promoted LC progression

After analyzing the metabolites in cirrhosis subregions, we further asked whether, within the PL region, the metabolite alteration pattern would be different. For the PL region, we further divided each false lobule into 3 annuluses, named from inside to the boundary as the inside region, the mid-region, and the peripheral region, respectively, and compared the differential metabolites within these 3 regions. We found that most differential metabolites showed similar expression levels from inside to peripheral, and only some metabolites had gradient expression levels, such as histamine and PE (20:3(8Z,11Z,14Z)/20:3(8Z,11Z,14Z)) (Figure 6A, B, Supplemental Figure S5a, b, http://links.lww.com/HC9/A348). For instance, we divided the LC group into lower and higher cirrhosis conditions to illustrate the association between disease progression and metabolic pathway alteration. Partial least-squares discriminant analysis was done on the PL-L, PL-H, and HL groups; the results showed that the metabolites were quite separated from the PL-L, PL-H, and HL-HC groups (Supplemental Figure S5c, http://links.lww.com/HC9/A348). This indicated that, with the cirrhosis progressing, the metabolites gradually changed at different disease stages. The downregulated metabolites in the LC group were mainly lipids and lipid-like molecules, organic oxygen, and nitrogen compounds, while the compounds with increased expression were more concentrated in organic acids and derivatives in positive ionization mode (Supplemental Figure S5d, http://links.lww.com/HC9/A348). In contrast, in the negative ionization mode, most metabolites were lipids and lipid-like molecules with significantly reduced expression (Supplemental Figure S5e, http://links.lww.com/HC9/A348). Since subregions of the LC group showed different patterns of metabolite alteration, we further used time-series analysis to identify the compounds in the same regions of HL-HC, PL-L, and PL-H that had increased or decreased tendency to select the key molecules that continued changing via cirrhosis progression. The metabolites were separated into 29 clusters; among these, only compounds in clusters 9 and 27 were in the positive mode and those in clusters 6, 14, and 23 were upregulated with cirrhosis progression. However, the decreased metabolites were present in more clusters in both positive and negative ionization modes, and in clusters 8, 14 and clusters 5, 26 in positive and negative modes, respectively, as shown in Figure 6C and Supplemental Figure S5f (http://links.lww.com/HC9/A348). The example metabolites of each cluster are shown in Figure 6D with quantification. Similar to cirrhosis subregions, the metabolic pathways involved in cirrhosis disease progression were glycerophospholipid, arginine, and proline metabolism in the positive mode and biosynthesis of unsaturated fatty acids, α-linolenic, and linolenic acid metabolism in the negative mode (Supplemental Figure S5g, http://links.lww.com/HC9/A348). Moreover, the expression of LC3B and p62 was increased in the LC group (Figure 6E–G). This result illustrates that in hepatitis B LC, the autophagy level is increased, which is related to dysregulation of the phospholipid metabolism. Further studies are needed to demonstrate the detailed mechanism.

FIGURE 6.

Comparison of metabolites’ alteration in spatial and disease progression dimensions. (A) Gradient distribution of metabolites in the PL region based on the positive ionization mode. Red, yellow, and green circles indicate PR, MR, and IR regions, respectively. (B) The quantification of key metabolites in PR, MR, and IR regions. (C) Time-series analysis of key metabolites’ expression along with disease progression. (D) The key metabolites’ spatial distribution and quantification from the time-series cluster. (E) LC3B and p62 expression in HC and LC groups. The yellow arrow points to the cell with autophagy. (F) The quantification of LC3 and p62 expression. (G) The mRNA expression level of LC3 and p62. *p < 0.05, **p < 0.01, and ****p < 0.0001. Abbreviations: HC, healthy control; IR, inside region; LC, liver cirrhosis; MR, mid-region; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PR, peripheral region.

DISCUSSION

Most studies on HBV-related cirrhosis were mainly focused on patients’ serum samples, and few studies have involved tissue samples or the tissue homogenate, which led to a lack of information on spatial alteration during disease progression and cannot reflect a practical microenvironment within the LC-specific region. The current study is designed to determine region-specific alteration in metabolite markers and link the expression of metabolites with the progression of cirrhosis, which is supplementary to understanding the biological and biochemical changes occurring in LC. In this study, AFAI-MSI was first applied to obtain metabolites’ spatial distribution in HBV-related LC and HC liver tissue. The multivariate statistical analysis combined with the HE staining image indicated differences in metabolites among the different regions, and the metabolic pathway of the selected metabolites was demonstrated to reveal potential markers for HBV-related LC progression. The total number of metabolites identified was >400; yet, in serum sample studies, the number of molecules ranged from dozens to a hundred. This means tissue samples could detect more metabolites than serum samples. Moreover, the categories of identified compounds were similar; most were lipid-like molecules, organic acids, and derivatives, such as various types of unsaturated fatty acids, PC, PE, and amino acids. In addition, the HE staining image overlaid with the MSI image supported the extraction of region-specific metabolite profiles, which suggested that certain metabolites’ ion intensity would indicate the spatial expression of compounds.

The liver has a strong ability to regenerate itself after suffering damage with dynamic processes,¹⁷ which is important for chronic liver diseases such as HBV infection, NAFLD, and HCC. It is well known that phospholipid molecules play key roles in liver lipid metabolism and regeneration progress due to they being the main components of the cell membrane. The dysregulation of lipid and lipid-like compounds had been demonstrated after partial hepatectomy during liver regeneration.¹⁸ Furthermore, the disordered accumulation of lipid and lipid molecules had been shown to be related to LC and regulated inflammation.¹⁹ In the current study, we have in total detected 16 PC metabolites in both HC and LC groups; among these, 12 PC metabolites together with 6 lysoPCs and choline molecules showed decreased expression in the LC group. The lysoPC mainly comes from the circulation produced by PC via phospholipase A2.^20,21 Since the expression level of PC was significantly reduced, the byproduct of lysoPC would also decline. While comparing within the subregions of the samples, the expression of PC and choline was found to be significantly downregulated in the PL group (Figure 7A). Park and colleagues used the MALDI MSI system to evaluate mouse livers infected with HBV. The results indeed showed alteration in expression of PC while being infected with HBV, which is similar to our results. However, the finding of no significant spatial distribution of PC in mouse liver samples was different from our finding.²² The reason for this difference might be the time and progress of the sample infected with HBV; in our study, all of the samples collected were of LC, which is the end stage of FB. In contrast, the mouse model injected with the virus stayed for a short time. Herein, the patients’ sample infection degree was higher than that of the mouse sample. In general, there are 2 essential pathways modulating PC biosynthesis: CDP-choline and phosphatidylethanolamine N-methyltransferase.²² The present study indicated that PCs, lysoPCs, and PE participated in multiple signaling pathways and were involved in the proinflammatory response, and that oxidative stress can be a novel lipid molecule target.²³ The alteration in expression of these molecules would further activate gene expression, leading to an immune response. Nevertheless, in HBV-related LC progression, the mechanism is still not revealed, and the spatial alteration of PC would be a potential target for screening new drugs and revealing the mechanism of liver disease progression.

FIGURE 7.

The alteration of metabolic pathways promotes LC progression. (A) The expression of PC in HC and LC groups with different subregions. (B) The metabolic pathways of differential lipid molecules involved in LC progression. (C) The dysregulation of an amino acid metabolic pathway in LC. *p < 0.05, and **p < 0.01. Abbreviations: HC, healthy control; HL, hepatic lobule; LC, liver cirrhosis; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PL, pseudolobule.

In addition, besides reducing PC quantity, the linoleic and α-linoleic acid metabolism pathways were also decreased in the LC group. Among these 2 pathways, several metabolites, including arachidonate, α-linoleic acid, and the relative derives, had downregulated (Figure 7B). It is well known that arachidonic acid is a polyunsaturated fatty acid in the presence of PC, which has been reported to be involved in the construction of membranes and characterized as a biomarker in the metabolism study of liver diseases.²⁴ The metabolomics study of patients’ serum samples also found a similar reduction of arachidonic acid. For instance, several unsaturated and saturated fatty acids declined in the LC group, such as eicosatrienoic acid, docosapentaenoic acid, docosahexaenoic acid, palmitic acid, and stearic acid (Figure 7B). The alterations of those unsaturated fatty acids have also been found in alcohol-associated cirrhosis, and, when combined with arachidonic acid, docosahexaenoic acid, and eicosapentaenoic acid, would support HCV, and diet and chemical-induced hepatic disorder.^25,26 Besides the downregulation of unsaturated fatty acids, several saturated fatty acids, such as palmitic and stearic acids, also showed declined expression. Unlike the unsaturated fatty acids, which can benefit hepatic diseases’ outcomes, the saturated fatty acids were insufficient to support disease outcomes.²⁷ Therefore, the results demonstrated that the selected unsaturated fatty acid compounds would offer antiviral activity and provide a novel target for therapy.

The amino acids and their derivatives are essential for maintaining the stability of cellular physiology, and always act as the key regulators in the signaling pathway.²⁸ The distinction between the amino acids of patients’ plasma samples of cirrhosis from England and the USA has pointed out the metabolic and chemical variation between stable and unstable cirrhosis.²⁹ The reduction of branched-chain amino acids, such as leucine, isoleucine, and valine, and the upregulation of other kinds of amino acids have been confirmed.³⁰ However, in the present study, the leucine and valine were not changed in the LC group. On the contrary, the arginine and d-proline declined in the LC group. Proline, as an essential amino acid, is in the subcellular microenvironment and has been demonstrated to participate in a variety of biology functions, such as apoptosis and autophagy; it has also been recognized as having important role in cancer metabolism. The MSI image indicated that d-proline was significantly decreased in the PL region compared with the HL in the HC group (Figure 7C). This result was consistent with a previous study in serum samples of HBV cirrhosis and HCC.³¹ The previous studies had indicated that arginine had a promising function in the liver via mediating the nitric oxide signaling pathway, which maintained vascular health conditions and impacted hepatic circulation.³² Besides, it also works on the urea cycle to facilitate fatty acid oxidation and ammonia detoxification.³³ The cause of blood ammonia balancing is critical for regular body conditions. Microcystin-leucine-arginine (MC-LR) can induce mouse liver FB by activating HSCs with upregulated fibrotic markers α-smooth muscle actin and collagen I.³⁴ Moreover, it also leads to epithelial-mesenchymal transition in mice lung and alveolar epithelial cells.³⁵ Thus, dysregulation of amino acids in LC is one of the essential factors to promote fibrogenesis.

In summary, the results showed that the metabolite species between HC and LC were similar, but their expression levels were different. There was a significant decrease of lipid compounds and some amino acids in the LC group, such as arginine, proline, lysoPCs, PCs, and several fatty acids. In addition, PC-like metabolites had altered specific spatial distribution and were obviously reduced in the PL region of the LC group. Moreover, the downregulation of these molecules was associated with cirrhosis progression. Taken together, the results indicated that, during the disease progression, the main alterations were in the expression level of lipid molecules and their spatial distribution. Understanding the metabolites’ spatial distribution and the regulation of genes can provide better knowledge on LC pathogenesis. Further studies are required to reveal the role of the altered metabolites in HBV-related LC progression.