If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Institute of Clinical Chemistry, Inselspital, Bern University Hospital, Bern, SwitzerlandDepartment of Visceral Surgery and Medicine, Inselspital, Bern University Hospital, University of Bern, Bern, SwitzerlandDepartment for BioMedical Research, Visceral Surgery and Medicine, University of Bern, Bern, Switzerland
# Current affiliation: Centre des Maladie Digestives Lausanne, Switzerland
Jean-Francois Dufour
Footnotes
# Current affiliation: Centre des Maladie Digestives Lausanne, Switzerland
Affiliations
Department of Visceral Surgery and Medicine, Inselspital, Bern University Hospital, University of Bern, Bern, SwitzerlandDepartment for BioMedical Research, Visceral Surgery and Medicine, University of Bern, Bern, Switzerland
We characterized the hepatic lipid distribution across liver zonation using DESI-MSI.
•
Fatty acids were predominantly detected in the periportal region while Triacylglycerols were detected mainly in the pericentral region.
•
Although phospholipids were distributed in the periportal and pericentral zones, phosphatidylinositols were located predominantly in the midzone.
•
De novo triacylglycerols biosynthesis appeared to be the most influenced pathway across the three zones.
•
AsSessing zone-specific hepatic lipid metabolism may lead to a better understanding of lipid homeostasis during disease progression.
Abstract
Background
&Aims: Lipid metabolism plays an important role in liver pathophysiology. The liver lobule asymmetrically distributes oxygen and nutrition, resulting in heterogeneous metabolic functions. Periportal and pericentral hepatocytes have different metabolic functions, which lead to generating liver zonation. We developed spatial metabolic imaging using Desorption Electrospray Ionization mass spectrometry (DESI-MSI) to investigate lipid distribution across liver zonation with high reproducibility and accuracy.
Methods
Fresh frozen livers from healthy mice with control diet were analyzed using DESI-MSI. Imaging was performed at 50 μm × 50 μm pixel size. Regions of interest (ROIs) were manually created by co-registering with histological data to determine the spatial hepatic lipids across liver zonation. The ROIs were confirmed by double immunofluorescence. Mass list of specific ROIs was automatically created, univariate and multivariate statistical analysis were performed to identify statistically significant lipids across liver zonation.
Results
A wide range of lipid species was identified, including fatty acids, phospholipids, triacylglycerols, diacylglycerols, ceramides, and sphingolipids. We characterized hepatic lipid signatures in 3 different liver zones (periportal, pericentral, and midzone) and validated the reproducibility of our method for measuring a wide range of lipids. Fatty acids were predominantly detected in the periportal region, while phospholipids were distributed in both periportal and pericentral zones. Interestingly, phosphatidylinositols, PI(36:2), PI(36:3), PI(36:4), PI(38:5), and PI(40:6) were located predominantly in the midzone (zone 2). Triacylglycerols and diacylglycerols were detected mainly in the pericentral region. De novo triacylglycerols biosynthesis appeared to be the most influenced pathway across the three zones.
Conclusions
The ability to accurately assess zone-specific hepatic lipid distribution in the liver could lead to a better understanding of lipid metabolism during the progression of liver disease.
Lay summary
Zone-specific hepatic lipid metabolism could play an important role in lipid homeostasis during disease progression. Herein, we defined the zone-specific references of hepatic lipid species in the three liver zones using molecular imaging. The De novo triacylglycerols biosynthesis was highlighted as the most influenced pathway across the three zones.
EC is an employee of Waters Corporation. The remaining authors have no conflict of interest in relation to the contents of the manuscript.
Authors’ contributions
MoM, JFD developed the concept and designed the study. PS, MaM developed and performed experiments. EC, AP and UR supported the method development. PS and MoM analyzed and interpreted the data. ST performed pathway analysis. PS and MoM drafted the original manuscript. All authors critically revised, contributed and approved the final version of the manuscript.
Financial support
JFD and MoM received funding from Waters Corporation. MoM received funding from Swiss National Science Foundation (SNSF) (grant no. 190686 and 213352).
Data availability statement
The authors confirm that the data supporting the findings of this study are available within the article and supplementary material. Due to the confidentiality of data, other than those is only available on request to the corresponding author (MoM).
Introduction
The liver architecture is built from thousands of hexagonal structures called hepatic lobules, composed of a central vein in the middle and hexagon corner of portal tracts. The hepatic lobule displays a gradient of oxygen and nutrient levels across the hepatocyte along the sinusoid from the periportal area to the pericentral site, resulting in asymmetrically distributed metabolic functions and generating a pattern known as liver zonation.
Liver zone 1 is located in the periportal region, continued with the intermediate zone in the middle of the lobule called zone 2 and the pericentral region referred as zone 3 (Figure 1). Different liver zonation exhibits different metabolic and biochemical pathways, including xenobiotic, amino acid, carbohydrate, and lipid metabolism. Although the differences in metabolic zonation have been reported previously, our understanding of zone-specific lipid metabolism in the liver is limited. The heterogeneity function of fatty acid metabolism in hepatocytes isolated from different liver zones has been observed previously. The fatty acid oxidation predominantly occurs in the periportal hepatocytes, while the pericentral hepatocytes showed higher rates of lipogenesis.
In addition, zone-specific proteomics data from isolated hepatocytes revealed a higher rate of fatty acid uptake and fatty acid synthesis in periportal hepatocytes.
However, few studies have investigated and characterized the zone-specific hepatic lipid signature.
Figure 1The histology guideline from H&E and double immunofluorescence staining for classification of liver zonation and ROIs drawing by co-registration with ion image. (a) H&E staining on the liver tissue. (b) The 10 pixels of each region were used to quantify the amount of lipids. (c) Double-immunofluorescence staining with liver zone-specific markers; GS-6: pericentral hepatocytes (blue), E-Cad: periportal hepatocytes (red), and DAPI (yellow). (d) The hepatocytes-stained in the periportal, pericentral, and midzone. Zone 1 (Z1): Periportal zone, Zone 2 (Z2): Midzone, Zone 3 (Z3): Pericentral zone, CV: Central vein, PV: Portal vein, BD: Bile duct, and HA: Hepatic artery
Several mass spectrometry (MS)-based lipidomic studies reported a wide range of lipid species in the liver using various techniques, including direct infusion (DIMS),
Lipid fatty acid profile analyses in liver and serum in rats with nonalcoholic steatohepatitis using improved gas chromatography-mass spectrometry methodology.
Although these techniques are well-established and allow good coverage of lipid species in the liver homogenate, they lack the resolution to differentiate the zonation of lipid metabolism in the liver. MS-based spatial metabolic imaging could provide a unique opportunity to assess zone-specific lipid metabolism in the liver. Recent studies have successfully characterized lipid signatures in the liver using MS-based imaging, including Secondary Ion Mass Spectrometry (SIMS)
Mapping the triglyceride distribution in NAFLD human liver by MALDI imaging mass spectrometry reveals molecular differences in micro and macro steatosis.
techniques. Even though both SIMS and MALDI are widely used for spatial metabolomics studies, both require sample preparation and ion generation in the high vacuum region of mass spectrometry.
The matrix choice applied for MALDI preparation is a crucial step influencing ion production; thus, matrix heterogeneity can affect the spatial integrity of the tissue surface.
A mass spectrometry imaging technique called desorption electrospray ionization (DESI) is an ambient ionization technique that analyses the spatial distribution and localization of metabolites directly from tissue samples with simple sample preparation and without damaging tissue or cellular morphology,
making it possible to perform further histology analysis. The potential of DESI-MSI for assessing lipid metabolism in different physiological conditions such as breast cancer,
has been demonstrated. DESI-MSI analysis in liver adenocarcinoma showed that unsaturated fatty acids containing phospholipids are predominant in the non-tumor region while sphingomyelin is elevated in the tumor region.
Here we aim to develop spatial DESI-MSI imaging to characterize hepatic lipid metabolism across the liver zonation. Understanding the role of zone-specific hepatic lipid metabolism in the normal liver is crucial to understand the metabolic changes during the disease stage.
We believe this technique provides the opportunity to differentiate lipid metabolism in different liver zonation with high accuracy.
Materials and methods
Liver tissue collection
Five livers were collected from 16 weeks old healthy male C57Bl6/J mice (Charles River, Freiburg, Germany). The eight weeks mice were acclimatized to the housing facility for one week. They were housed under controlled temperature (22 ± 2 °C) and 12 h light-dark cycles. They were fed a standard diet (Catalogue number 5053, LabDiet, USA). The health, body weight, and food intake were monitored weekly.
Mice were weighed, anesthetized with pentobarbital (100 mg/kg, i.p.), and euthanized in the afternoon. The liver was collected, snap-frozen, and stored at -80 °C. All the experiments were conducted according to the regulations of the Bern Animal Welfare Committee, Canton of Bern, Switzerland (BE42/19).
Tissue preparation
The frozen liver was sectioned at 10 μm thickness under -21 °C using the HYRAX C60 Cryostat machine (Zeiss, Jena, Germany) after mounting with a drop of distilled water. The tissue slices were thaw-mounted on glass slides (Thermo Scientific™, MA, USA) and stored at -80 °C until DESI-MSI analysis.
The slides were dried for 15 min at room temperature prior to DESI-MSI analysis. The optical image was created for each slide using CanoScan LiDE 210 scanner (Canon, Tokyo, Japan) and the slide was placed on DESI two‐dimensional moving stage holder.
DESI-MSI analysis setup and parameters
Spatial lipid imaging data were performed using a 2D Omni spray stage (Prosolia, USA) coupled with Xevo™ G2-XS QTof (Waters Corporation, UK). DESI-MSI imaging was acquired in positive and negative ionization mode over the mass range of m/z 100 to 1,200. The imaging experiments were performed at 50 μm2 pixel size. The High-Performance DESI sprayer incidence angle was set up at 75°. The distance between the sprayer and tissue surface was 2 mm and between the sprayer and inlet was 1 mm. The charged spray solvent was made of 98% methanol (Biosolve Chimie, Dieuze, France) and 2% MilliQ water (Merck Millipore, Massachusetts, United States). The solvent was sprayed at a flow rate of 2.0 μL/min. The sampling cone voltage was set up at 120 and 110 V for positive and negative ionization, respectively. The nebulizing gas (Nitrogen) was set to 8.5 PSI. The collection angle of the heated transfer line (HTL) inlet was 10° with a distance of 0.5 mm from the tissue surface.
Histology analysis
The classification of liver zonation was performed directly on the liver tissue used for DESI-MSI analysis by an experienced pathologist. The tissue was stained using standard hematoxylin and eosin (H&E) staining protocol. The stained tissue sections were scanned by a Panoramic 250 Flash II slide scanner with a 20x objective (3DHISTECH Ltd., Budapest, Hungary) to generate histology scans.
Immunofluorescence staining
Frozen liver slides were fixed for 5 minutes with 4% paraformaldehyde and washed with Bond™ wash solution. The heating antigen retrieval was applied using EDTA/Tris and blocked with Opal™ blocking buffer. They were stained by glutamine synthetase (1:10,000, Sigma-Aldrich, catalogue G2781), E-cadherin (1:200, Santa Cruz Biotechnology, Inc., catalogue sc-7870) antibodies, and DAPI. Then washed and mounted with ProLong™ Gold Antifade reagent. The immunofluorescence-stained slides were scanned by a Panoramic 250 Flash II slide scanner with a 20x objective (3DHISTECH Ltd., Budapest, Hungary).
Data and imaging analysis
MassLynx™ Software V4.2 (Waters Corporation, UK) was used for data acquisition and spectrum preview. DESI ion images were visualized using the High Definition Imaging (HDI™) software V1.6 (Waters Corporation, UK). The DESI-MSI data were imported into LipostarMSI V1.1.0b28 (Molecular Horizon srl, Italy) to perform data pre-processing. Briefly, pre-processing of the DESI-MSI data included peak alignment, profile smoothing, baseline correction, and peak picking.
LipostarMSI: Comprehensive, Vendor-Neutral Software for Visualization, Data Analysis, and Automated Molecular Identification in Mass Spectrometry Imaging.
The data set was normalized by total ion count (TIC). Tentative identification of lipids based on the accurate mass was accomplished by searching against LIPID MAPS databases (https://www. lipidmaps.org), then lipid identification was confirmed by DESI-MS/MS analysis.
Regions of interest (ROIs) were manually created by co-registration with histological data to determine spatial hepatic lipids using HDI software. The liver zonation was classified into zone 1: periportal zone, zone 2: midzone, and zone 3: pericentral zone, as shown in Figures 1a and 1c. Three different ROIs were selected in each liver zonation. Each region consisted of 10 pixels used to quantify the amount of lipids in each ROI (Figure 1b). The mass list of specific ROIs was automatically created, and statistical analysis was performed to compare lipids between liver zonation. Lipid levels were measured in three ROIs. Thus, for each zone, 30 pixels were considered to assess lipid intensity in each zone. For the analysis, the medians of lipid intensity in each of the three ROIs were considered. For statistical analysis, a repeated-measure Analysis of Covariance (ANCOVA) was performed for each lipid with zonation (portal, midzone, or central) as fixed effect and mouse identifier as random effect. The chi-square p-value of the vein localisation term was reported. To account for multiple-testing, Benjamini-Hochberg False Discovery Rate (FDR) was estimated. All the statistical analysis were analysed by R version 4.0.2.
Pathway analysis
We performed pathway analysis on the list of 117 significantly different lipids across the three ROIs. We used the mouse metabolic pathway set from the PathBank database,
Over-representation analysis (ORA) was also carried out for additional insight into the pathways’ importance. We used the same pathway definitions and the hypergeometric test to calculate p-values. The background metabolite set for ORA was defined as the totality of the participating compounds in all pathways. All calculations were made using R version 4.0.2.
Results
Optimization of liver sample preparation for lipid detection
The most common liver tissue used for histological analysis is the formalin-fixed, paraffin-embedded (FFPE) tissue. Initially, we tested DESI analysis using FFPE tissues with and without the deparaffinization method. None of the lipids were reliably detectable in FFPE liver tissues; therefore, the fresh frozen tissue was used for our DESI-MSI analysis. The frozen liver tissue was embedded with the cryosection embedding material, optimal cutting temperature (OCT) compound, and gelatin, which caused ion suppression especially in positive mode. Therefore, we decided to proceed without embedding material as a series of sections at 5 and 10 μm; the 5 μm sections were used as histology guidelines. The 10 μm sections were utilized for DESI analysis. We assessed the effect of tissue storage without embedding material on liver morphology over 7 months, and there was no significant difference during the storage period at -80 °C, as shown in Fig. S1. DESI-MSI imaging was acquired in positive and negative ionization mode. Several DESI parameters, including spray solvent composition and spray angle were optimized. The capillary voltage was adjusted every 0.05 kV from 0.5 to 0.75 kV in both positive and negative mode. The optimal voltage was at 0.6 kV, generating high intensity and signal stability in the liver tissue. The HTL voltage was optimized at 11 V (369 °C) for positive and 13 V (494 °C) for negative ionization. To reduce run time without compromising data quality, the stage velocity was increased from 50 μm/s to 200 μm/s, increasing the scan rate from 1 Hz to 4 Hz under the spatial resolution of 50 μm. The chemical and physical properties of the spray solvent system affect the range of molecular detection by DESI-MSI.
Desorption electrospray ionization (DESI) mass spectrometry and tandem mass spectrometry (MS/MS) of phospholipids and sphingolipids: ionization, adduct formation, and fragmentation.
However, in our observation, DESI-MSI with methanol and water did not detect important lipid species relevant to liver metabolism, such as diacylglycerols (DGs), triacylglycerols (TGs), and cholesterol esters. A couple of studies reported that the modification of the methanol and water spray solvent with salts such as ammonium acetate
enhances the ability to detect TGs in biological samples using DESI-MSI analysis. We observed that modification of organic spray solvent by adding salts affects the sprayer capacity; thus, we immersed fresh-frozen mouse liver slides in 25 mM ammonium acetate for 20 seconds prior to DESI-MSI analysis. The immersion in ammonium acetate improved triglycerides detection (m/z ≥ 840) in positive mode significantly. All TGs were detected as [M+Na]+, and [M+NH4]+ adducts, whereas [M+H]+ and [M+K]+ adducts were low abundant.
Determination of hepatic lipid signature using DESI-MSI
Our method detected 10,546 features in positive and negative ionization modes. All mass to charge ratio (m/z, a ratio of an ion mass to its ion charge), which led to isotope pattern matching, were identified by searching against the LIPID MAPS database, and then they were confirmed by fragmentation pattern using DESI-MS/MS analysis. We successfully identified 269 lipids from the liver tissue. An example of MS/MS fragmentation using DESI-MSI is shown in Figure 2. The m/z 758.6 and m/z 747.5 were confirmed to be protonated adduct [M+H]+ of PC(16:0_18:2) and deprotonated [M-H]- of PA(18:0_22:6), respectively. The [M+H]+ and [M+K]+ are the most commonly reported adducts in positive mode and [M-H]- adducts in negative mode.
lipidomic analysis. The DESI-MSI method detected a broad spectrum of lipid species in the liver, including fatty acids (FAs), lysophospholipids (LPLs), phosphatidic acids (PAs), phosphatidylcholines (PCs), phosphatidylethanolamines (PEs), phosphatidylglycerols (PGs), phosphatidylinositols (PIs), phosphatidylserines (PSs), DGs, TGs, ceramides (Cers), and sphingolipids (SLs). The ion images in Fig. S2 show that the most abundant PC detected in the liver is PC(34:2) (m/z 758.6) agrees with the previous detected by MALDI.
Then following with PC(36:4) (m/z 782.6), and PC(38:6) (m/z 806.6). We detected FA(18:2) and linoleic acid as the most abundant FA in the liver, which was significantly expressed in the periportal zone. At the same time, PI(38:4) (m/z 885.6) was the most abundant phospholipid, significantly higher in the pericentral zone.
Figure 2DESI-MS/MS spectrum and ion images of healthy mouse liver obtained from the m/z 758.6 (a) and the m/z 747.5 (b). The lipid identification using DESI-MS/MS was confirmed as PC(16:0_18:2) and PA(18:0_22:6), representatively. The yellow color represents the highest intensity of the lipids.
The reproducibility of the method was assessed using intraday and interday analysis. The results are shown in Table S1. Most detected lipids had intraday and interday deviation lower than 20%, according to the FDA guidance for bioanalytical method validation and study sample analysis. Only the low abundant lipids showed high variation (CV%=20-37%). In order to assess the reproducibility of the detected lipids, ten pixels were selected for each region to assess the analytical variability. Table S2 shows the reproducibility of DESI-MSI in detecting lipids in different zones.
Hepatic lipid distribution across liver zonation
Spatial metabolic imaging sheds light on the distribution of hepatic lipids across the liver zonation. Statistically significant expressed lipids across liver zonation are illustrated in Figure 3 and Table S3. Debois et al. detected FA(16:1) outside of steatotic vesicles, FA(18:1) in the intermediate localization, and FA(18:2) in lipid droplets.
However, the localization of these FAs and the others in the liver lobule has not been previously described. In our study, polyunsaturated fatty acids (PUFAs); FA(18:2), FA(20:2), FA(20:3), arachidonic acid (FA(20:4)), FA(22:4), and DHA (FA(22:6)) were significantly higher in the periportal region (Figure 3 & Figure 4a). Interestingly, the phospholipids containing arachidonic acid PS(22:4_20:4), PE(16:0_20:4), PE(16:1_20:4), and PE(18:1_20:4) were also significantly higher in the periportal region compared to other zones (Figure 4b). This pattern was also observed among the phospholipids containing DHA, such as PS(20:0_22:6), PE(16:0_22:6), PE(18:0_22:6), PA(18:0_22:6), and PA(18:1_22:6) which were expressed predominantly in the periportal region as shown in Figure 4b. PCs were the predominant lipid species in healthy mouse liver. Interestingly, we observed that the distribution of PCs in different zone was highly related to their FA composition. As shown in Fig. S3a, PC(32:1), PC(34:1), PC(34:2), PC(34:3), PC(34:4), PC(36:3), PC(36:6), PC(37:6) and PC(38:6) were strongly located in the periportal region whereas PC(35:2), PC(36:2), PC(36:4), PC(37:1), PC(37:2), PC(37:3), PC(38:0), PC(39:5), and PC(40:4) were significantly more expressed in the pericentral region (Fig. S3b). Heterogeneous distribution was also observed in PIs (Figure 5); most PIs, such as PI(38:3), PI(38:4), PI(39:4), PI(42:8), and PI(42:9), were detected mainly in the pericentral region whereas, PI(34:2), PI(36:2), PI(36:3), PI(36:4), PI(38:5), PI(38:6), PI(40:5) and PI(40:6) were mainly distributed in the midzone (zone 2). As shown in Fig. S4, LPLs such as LPA(18:0), LPE(16:0), and LPC(18:1) were concentrated in periportal areas, whereas LPC(18:0) was localized in pericentral areas. PGs did not show a significant difference in distribution across the zonation. DGs and TGs exhibited similar patterns; as shown in Figure 6, both lipid classes were predominantly present in the pericentral region. Furthermore, we detected Cer and several SLs across liver zonation (Fig. S5). Cer(37:0) was significantly higher in the pericentral region. Although most of the SLs were mainly expressed in the pericentral region, SM(36:0), SM(38:0), and SM(38:2) were significantly higher in the periportal region.
Figure 3Circular dendrogram of the statistically significant lipids predominantly express in three liver zones; Zone 1: Periportal zone (red), Zone 2: Midzone (yellow), Zone 3: Pericentral zone (blue) includes FA: fatty acid, LPL: lysophospholipid, PA: phosphatidic acid, PC: phosphatidylcholine, PE: phosphatidylethanolamine, PI: phosphatidylinositol, PS: phosphatidylserine, DG: diacylglycerol, TG: triacylglycerol, Cer: ceramide, and SL: sphingolipid. AA: arachidonic acid and DHA: docosahexaenoic acid
Figure 4Fatty acids and phospholipids containing n-6 PUFA arachidonic acid and n-3 PUFA DHA distribution were observed in the healthy liver. Statistical analysis was performed to compare FAs distributed across three liver zones (Periportal zone (red), Midzone (yellow), and Pericentral zone (blue)); the Violin boxplot showed FA(20:2), FA(20:3), FA(20:4), FA(22:4), and FA(22:6) were detected predominantly in the Portal tracts (a). DESI-MSI ion images of arachidonic acid FA(20:4), DHA FA(22:6), phospholipids containing arachidonic acid, and phospholipids containing DHA distribute across liver zonation and highly express in the periportal zone (b). The yellow color represents the highest intensity of the lipids. CV: Central vein, and PT: Portal tracts
Figure 5Spatial distribution of phosphatidylinositols obtained from healthy mice showed the localization of PI(38:3), PI(38:4), PI(39:4), PI(42:8), and PI(42:9) in the pericentral zone of liver lobule whereas PI(34:2). PI(36:2), PI(36:3), PI(36:4), PI(38:5), and PI(40:6) were mainly located in Midzone. The yellow color represents the highest intensity of the lipids. CV: Central vein, and PT: Portal tracts
Figure 6DESI-MSI ion images of DG(36:2), DG(36:3), DG(38:2), DG(38:4), DG(38:5), TG(50:4), TG(50:5), TG(51:1), and TG(52:2) showed the similar distribution patterns of diacylglycerols and triacylglycerols in healthy mice liver. The distribution of these lipids was predominantly detected in the pericentral zone. The yellow color represents the highest intensity of the lipids. CV: Central vein, and PT: Portal tracts
We performed double-immunofluorescence staining with liver-specific markers to validate the zonation pattern of the liver. We used glutamine synthetase (GS-6), which is selectively expressed in pericentral hepatocytes, and E-cadherin (E-Cad) for periportal hepatocytes as well as DAPI to stain the hepatocytes nucleus.
Figure 7 shows the determination of liver zonation using a) H&E staining, b) double-immunofluorescence, and c) zone-specific lipid distribution using DESI-MSI.
Figure 7Comparison of the liver zonation pattern using liver histology, immunofluorescence and molecular imaging with DESI. (a) H&E staining of liver tissue after DESI-MSI analysis. (b) Double-immunofluorescence staining shows the proto-central axis along with the liver zonation; GS-6: pericentral hepatocytes (blue), E-Cad: periportal hepatocytes (red), and DAPI (yellow). (c) Red, green, and blue overlay ion image from DESI-MSI; blue: PI(38:3) located in the pericentral area, green: PI(36:3) mainly expressed in the midzone, and red: PE(38:6) predominantly presented in the periportal region. CV: Central vein, MZ: Midzone, and PV: Portal vein
Identification of key metabolic pathways in hepatic zonation
We performed pathway analysis to investigate the alteration in lipid metabolism across liver zonation. The top ranked pathways with their respective dominant ROI expression profiles are displayed in Figure 8. The analysis revealed that the de novo triacylglycerols biosynthesis is the most influenced pathway across the three regions, followed by phospholipid, cardiolipin, and sphingolipid biosyntheses. Other implicated pathways include the metabolism of inositol and phosphatidylinositol, arachidonic, alpha-linolenic, and linoleic acids, as well as leukotriene biosynthesis.
Figure 8Top-ranked pathways across the three regions of interest. Colors indicate the region(s) in which each lipid family was predominantly detected. Multiple colors indicate that lipid classes were expressed in multiple zones. FA: fatty acid, LPL: lysophospholipid, PA: phosphatidic acid, PC: phosphatidylcholine, PE: phosphatidylethanolamine, PG: phosphatidylinositol, PI: phosphatidylinositol, PS: phosphatidylserine, DG: diacylglycerol, TG: triacylglycerol, Cer: ceramide, SL: sphingolipid, SM: sphingomyelin, Sph: sphingosine, G3P: glycerol-3-phosphate, CDP-DG: CDP- diacylglycerol, PGP: phosphatidylglycerophosphate, and CL: cardiolipin (red - periportal zone, yellow - midzone, blue - pericentral zone, white – not detected)
However, our understanding of zone-specific lipid metabolism is limited. Mass spectrometry imaging with minimum sample preparation, such as DESI, would facilitate investigating zone-specific lipid metabolism without impacting the integrity of liver tissue. Moreover, label-free and matrix-free molecular imaging of tissue histology is beneficial for precise assessment of lipid metabolism across liver zonation, especially in clinical settings, as it is simple and reproducible compared to other imaging techniques. Thus, we developed and validated a robust and reproducible method of spatial metabolic imaging using Desorption Electrospray Ionization mass spectrometry to assess hepatic lipid zonation. Our approach allows characterizing a wide range of lipid species across the liver zonation with high reproducibility and accuracy. Immersion of prepared tissue in 25 mM ammonium acetate prior to DESI-MSI analysis improved the detection of TGs due to the formation of ammonium adduct, which was in agreement with the previous report.
We identified 269 lipids across the liver zonation that could be monitored reproducibly. Hepatic lipid alteration has been shown to play a crucial role in the pathogenesis of liver diseases. Several studies reported lipid levels in liver homogenates and circulation, such as serum and plasma, using various MS techniques.
Lipid fatty acid profile analyses in liver and serum in rats with nonalcoholic steatohepatitis using improved gas chromatography-mass spectrometry methodology.
However, these analyses do not capture the zonation of lipid metabolism in the liver. The liver zonation may have a vital role in liver diseases; thus, assessing zone-specific hepatic lipid metabolism will improve our understanding of the pathogenesis of the liver. A couple of MS-based imaging studies using SIMS and MALDI has been described the zone-specific lipids in the liver. Debois et al.
exhibited the zonation of lipids, including FAs and DGs, in healthy liver using SIMS. However, arachidonic acid (FA(20:4)), phospholipids, and TGs, could not be detected by SIMS. PCs and PEs zonation has been reported in the normal obese liver by Wattacheril et al.,
Mapping the triglyceride distribution in NAFLD human liver by MALDI imaging mass spectrometry reveals molecular differences in micro and macro steatosis.
using MALDI. However, the spatial distribution of other lipid species, such as LPLs, PAs, PSs, PIs, Cers, and SLs across liver zonation has not been investigated previously. Therefore, we investigated the localization of 269 identified lipids, which could be measured reliably and accurately with DESI-MSI. One hundred seventeen lipids (Figure 3) were significantly different across the three zones. We observed that FAs, LPLs, PAs, PCs, PEs, PSs, PIs, DGs, TGs, Cers, and SLs expressed heterogeneously across liver zonation while PGs were distributed homogeneously. It is worth mentioning that despite our efforts to detect cholesterol and cholesterol esters, we were not able to detect them reliably using DESI-MSI. This is, at least in part, due to the low affinity and low acidity of cholesterol impact on the ionization by DESI.
Further investigations are required to optimize this technique for cholesterol and cholesterol esters detection.
We observed a gradient increase in FAs from the periportal to pericentral regions. This is expected as the blood flow from the portal vein and diffuses along the sinusoid to the central vein,
result in different FAs uptake from the periportal to the pericentral area. This is in agreement with the previous findings indicating higher FAs uptake rate as well as higher FAs synthesis in periportal hepatocytes.
reported a higher capacity of FAs esterification into cellular lipids and VLDL in the pericentral hepatocytes, which supports the low intensity of FAs detected in the pericentral area. Schleicher et al. used mathematical multicompartment model of hepatic FA metabolism coupled with blood flow simulations and demonstrated that the lower oxygen supply in the pericentral zone enhanced the pericentral TG accumulation, whereas the higher rate of FA oxidation in the periportal dominated by higher oxygen supplying.
In support of this, another study by Bulutoglu et al. indicated that the lowest oxygen supply, as in the pericentral region, drove the highest TGs accumulation at the lowest free FA supplementation.
Mapping the triglyceride distribution in NAFLD human liver by MALDI imaging mass spectrometry reveals molecular differences in micro and macro steatosis.
we demonstrated that TGs were predominantly localized in the pericentral region. This may be explained by the fact that the pericentral area has a lower oxygen supply and generates a lower rate of FA oxidation.
Once the FA oxidation is saturated, to avoid cytotoxicity, the excess FAs are esterified to TGs and stored as lipid droplets or secreted in the form of VLDL. Glycerophospholipids are a major lipid component of cellular membranes.
The reliable assessment of PIs is challenging due to their low abundance in biological system, as well as analytical challenges such as the high polarity of the phosphate group and acidic nature of phosphates. However, our DESI-MSI method successfully detected the PIs in deprotonated form. Most notably, we discovered that PIs, which are involved in cellular signaling, cell apoptosis,
reported that zone 2 hepatocytes might be a vital zone for liver repopulation. During homeostasis and regeneration, zone 2 is an essential source of new hepatocytes, driven by the insulin-like growth factor binding protein 2–mechanistic target of rapamycin–cyclin D1 (IGFBP2-mTOR-CCND1) axis that might regulate by phosphatidylinositol 3-kinase–mTOR signaling.
In addition, we demonstrated that the zonation pattern of specific lipids across the liver lobule corresponds to microscopic histology as performed by double immunofluorescence staining (Figure 7).
Finally, we performed pathway analysis to explore the metabolic pathways associated with liver zonation (Figure 8). TG biosynthesis via glycerol-3-phosphate (G3P), appeared to be the most zone-influenced pathway. Halpern et al. and Ben-Moshe et al. showed that around 50% of liver genes exhibited zone-specific expression in the liver lobule, however, they only reported acylglycerol-3-phosphate acyltransferases (Agpat2 and Agpat3) protein expressed significantly in the pericentral zone.
In addition, phospholipid and cardiolipin biosynthesis were very highly ranked in our pathway analysis. This is potentially due to the involvement of FAs, PA, and LPA as the major precursors in cardiolipin synthesis and remodelling.
The main goal of the current study was to develop a spatial molecular imaging approach in order to elucidate lipid metabolism across liver zonation and explore the hepatic lipid signatures in healthy liver. The simple sample tissue preparation for DESI-MSI analysis (label-free, matrix-free and without damaging tissue or cellular morphology) allows for performing co-registration ion images with histological data directly on the tissue used for the DESI-MSI analysis without further steps. This is beneficial for clinical studies where the availability of liver biopsies is limited. The ability to perform spatial liver imaging with high accuracy and precision will lead to a better understanding of zone-specific hepatic lipid metabolism, especially in chronic liver diseases such as NAFLD progression. Further research could explore the alteration of lipid distribution across the liver zonation in liver diseases such as NAFLD or investigate the zonation adaptation during the circadian rhythm.
Acknowledgements
We thank Dr. P. Kumar and Dr. M. St-Pierre from Department for BioMedical Research, Visceral Surgery and Medicine, University of Bern for their support.
Appendix A. Supplementary data
The following is/are the supplementary data to this article:
Lipid fatty acid profile analyses in liver and serum in rats with nonalcoholic steatohepatitis using improved gas chromatography-mass spectrometry methodology.
Mapping the triglyceride distribution in NAFLD human liver by MALDI imaging mass spectrometry reveals molecular differences in micro and macro steatosis.
LipostarMSI: Comprehensive, Vendor-Neutral Software for Visualization, Data Analysis, and Automated Molecular Identification in Mass Spectrometry Imaging.
Desorption electrospray ionization (DESI) mass spectrometry and tandem mass spectrometry (MS/MS) of phospholipids and sphingolipids: ionization, adduct formation, and fragmentation.
00056-3&id=ga1.jpg){kind=link}
00056-3&id=gr1.jpg){kind=link}
00056-3&id=gr2.jpg){kind=link}
00056-3&id=gr3.jpg){kind=link}
00056-3&id=gr4.jpg){kind=link}
00056-3&id=gr5.jpg){kind=link}
00056-3&id=gr6.jpg){kind=link}
00056-3&id=gr7.jpg){kind=link}
00056-3&id=gr8.jpg){kind=link}