Comparative Analysis of the Human Proteome Profile in visceral Adipose Tissue and Liver Tissue in Individuals with Obesity With and Without MASLD and MASH

1. Introduction

The harmful impact of the excessive deposition of visceral adipose tissue (VAT) on risk of metabolic disease, type-2 diabetes (T2DM) and cardiovascular disease has been recognized for decades [1,2]Adipose tissue has major metabolic as well as endocrine functions. Adipose tissue derived adipocytokines and fat-derived metabolites such as free fatty acids affect the physiology of other organs [3,4,5] but VAT in particular is associated with metabolic disease risk. In clinical practice, waist circumference, which correlates with truncal VAT mass, is central to identifying patients at increased risk of metabolic comorbidities including metabolic associated steatotic liver disease (MASLD) and metabolic associated steatohepatitis (MASH)[6,7,8] An increased volume of visceral fat mass has a negative effect on the adipose tissue physiology and function as the subsequent increase in metabolic and oxidative load induce peripheral insulin resistance and act as stressors which alter the overall homeostasis of the VAT (ref). These alterations are exerted, in part, through changes in secretion of adipokines and proinflammatory cytokines with subsequent creation of an inflammatory and metabolic dysfunctional milieu with resultant adipocyte death and fibrosis deposition within the adipose tissue with infiltration and proliferation of macrophages contributing to obesity-associated adipose tissue inflammation [9,10,11,12,13]

In MASLD, triglycerides accumulate within the hepatocytes and cause liver steatosis; the condition may progress to MASH which is characterized by necroinflammation of the liver tissue and formation of liver fibrosis [14]. In MASLD, the changes that occur in VAT and liver tissue seem to share some pathophysiological characteristics. Mesenteric- and omental VAT are venously drained to the liver via the portal vein. According to the portal vein theory, free fatty acids adipocytokines and inflammatory cytokines derived from the VAT access the liver and could contribute to the development of liver steatosis and perhaps impact the progression of MASLD to MASH and liver fibrosis[15] .

Mass spectrometry (MS)-based proteomic analyses offer fast and large-scale analysis of the proteomes and can identify alterations in protein abundance in relation to disease. Therefore, MS-based proteomics could serve as a tool for identification of proteins that are up- or downregulated in patients with MASLD and MASH. The concomitant assessment of VAT and liver tissue allows for a detailed evaluation of the changes that occur in patients with mild, moderate or severe MASLD. Such analyses could provide important information about the underlying pathophysiology of MASLD.

As for MASLD, liver tissue proteomics has been successfully applied with focus on MASH associated hepatocarcinogenesis [16]and MASH fibrosis [17]. Additionally, our group has previously identified plasma biomarker candidates related to MASLD [18]. Consequently, it is of interest to perform detailed investigations of VAT in relation to MASLD.

In this cross-sectional study, we performed tissue proteomics analysis of paired omental VAT and liver tissue biopsies in 70 patients with severe obesity (SWO) without and with different stages of MASLD, as well as in nine control study subjects without obesity. We aimed to investigate a possible overlap of proteins that are simultaneously significantly up-or downregulated in both liver tissue and VAT and identify a potential MASH-specific proteome signature in both liver and VAT.

2. Materials and Methods

Cohort and Study Investigations

We collected omental VAT and liver tissue from 70 SWO undergoing laparoscopic bariatric surgery and control study subjects without obesity (CON) undergoing planned laparoscopic cholecystectomy at Copenhagen University Hospital Hvidovre between December 2016 and October 2019. Tissue samples were collected immediately after induction of anesthesia and trocar placement, before the actual surgical procedure.

After sampling, the VAT was trimmed and cut into smaller pieces of 50-100 mg and immediately snap frozen in liquid nitrogen before storage in a -80-degree freezer until analysis. Liver tissue was similarly trimmed, capsule removed, and was cut into pieces of 50 mg, snap frozen and stored as described above. In addition, part of the liver biopsy underwent fixation in paraformaldehyde for later paraffin embedment and histological analysis.

The study was conducted according to the Declaration of Helsinki and was approved by the Regional Ethics Committee of Capital Region, Denmark (H-16030784 and H-16020782). All study subjects gave oral and written consent.

Liver histology and Grouping of Study Subjects

Liver histology was assessed by three pathologists. MASLD activity score the (referred to as the NAFLD activity score, NAS) was used to group patients with obesity as

1)

No Liver steatosis (LS-): no liver steatosis present in liver biopsies

2)

Liver steatosis present (LS+): Liver steatosis present but without MASH (NAS score < 5)

3)

MASH: NAS ≥ 5 with points from all subcategories (steatosis, inflammation, and ballooning).

Sample Preparation for MS Analysis

Frozen liver and VAT biopsies were homogenized on a Precellys24 homogenizer (Bertin Technologies, France) in 300 µl SDC reduction and alkylation buffer (PreOmics GmbH, Martinsried, Germany) containing 1x Roche phosphatase inhibitor with ceramic beads (2.8 and 1.4mm zirconium oxide beads, Precellys). Approximately 15 mg of liver and 50 mg of VAT were processed as the starting material. Homogenates were incubated at 95 °C for 10 min (1200 rpm) and subsequently transferred to a new 1.5 ml Eppendorf tube, from which 45 µl of homogenate was further transferred to a 96-well plate and sonicated for 5 min using the Covaris Adaptive Focused Acoustics (AFA) sonication system (Covaris, USA). Protein content was determined by Tryptophan assay and an aliquot of 50 µg was digested overnight with LysC and trypsin in a 1:50 ratio (μg of enzyme to μg of protein) at 37 °C (1700 rpm). Peptides were acidified to a final concentration of 0.1% trifluoroacetic acid (TFA). Approximately 20 µg of peptides were loaded on Stage-Tips, washed with isopropanol/1% TFA (200μl) and then 0.2% TFA (200μl). Peptides were eluted with 60μl of elution buffer (80% acetonitrile/1% ammonia) and dried at 60°C using a SpeedVac centrifuge (Eppendorf, Concentrator plus). Dried peptides were dissolved and sonicated in 5% acetonitrile/0.1% TFA and concentration measured using Nanodrop. Peptide mixtures were stored at -80 °C until further analysis. Quality control samples of pooled tissue homogenate was included for measuring the workflow variation.

LC-MS/MS Analysis (Liquid Chromatography-Mass Spectroscopy/Mass Spectrometry Analysis)

Samples were measured using LC-MS instrumentation consisting of an EASY-nLC 1200 system (Thermo Fisher Scientific, San Jose, CA) interfaced on-line with a Q Exactive HF-X Orbitrap (Thermo Fisher Scientific, Bremen, Germany). Purified peptides were separated on 42.5 cm HPLC-columns (ID: 75 µm; in-house packed into the tip with ReproSil-Pur C18-AQ 1.9 µm resin (Dr. Maisch GmbH)). Approximately 500 ng of peptides were injected for each LC-MS/MS analysis. Peptides were loaded in buffer A (0.1% formic acid) and eluted with a linear 82 min gradient of 3-23% of buffer B (0.1% formic acid, 80% (v/v) acetonitrile), followed by an 8 min increase to 40% of buffer B. The gradients then increased to 98% of buffer B within 6 min, which was kept for 4 min. Flow rates were kept at 350 nl/min. Re-equilibration was done for 4 μl of 0.1% buffer A at a pressure of 700 bar. Column temperature was kept at 60 °C.

For liver tissue samples, MS spectra were acquired using data-independent acquisition (DIA) mode, enabled by MaxQuant.Live [19] in which the scan protocol was defined. Each acquisition cycle consisted of a survey scan at resolution of 60,000 with an automatic gain control target (AGC) of 3e6, maximum injection time of 100 ms, followed by 66 DIA cycles at resolution of 15,000 with an AGC of 3e6/22ms IT at range 300-1,650 m/z). HCD fragmentation was set to normalized collision energy of 27%. In all scans, PhiSDM [20]was enabled with 100 iterations, spectra type was set to centroid. For adipose tissue samples, the DIA-MS method consisted of an MS1 scan from 350-1,650 m/z range (AGC target of 3e6, maximum injection time of 60 ms) at a resolution of 60,000 and 32 DIA segments (Dataset EV1 , AGC target of 3 x106, maximum injection time of 45 ms). The acquisition of samples was randomized to avoid bias.

Quantification and Statistical Analysis: Raw Data Processing and Analysis

All raw files were analyzed with Spectronaut Pulsar X (version 13.10) with default settings except that “Quantification” data filtering parameter “Q-value” was set to “complete”. DIA hybrid spectra were searched against in-house generated libraries for human liver and adipose tissue using the same LC setup searching against human FASTA Uniprot database (version 201801, containing 93k entries). Proteins with more than 30% of missing values were discarded. Samples with less than 2,500 proteins quantified were also discarded. Normalization based on median protein intensity of each sample was performed. Remaining missing values of the dataset were imputed by drawing random samples from a normal distribution (down-shifted mean by 1.8 standard deviation (SD) and scaled SD (0.3) relative to that of proteome abundance distribution, with which we performed the statistical analysis. Statistical and bioinformatics analysis were performed with the Perseus software (version 1.6.5.0) and Python software. Specifically, one-way ANOVA was performed using the Python open-source statistical package ‘pingouin’ (https://pingouin-stats.org/) corrected for multiple hypothesis testing by Benjamini-Hochberg at 5% FDR (false discovery rate), followed by Tukey’s HSD (honestly significant difference) test. Pair-wise correlation between VAT and liver tissue protein levels was performed with the pingouin.pairwise_corr module with a significance level of Benjamini-Hochberg corrected FDR at 5%.

Before filtering for 70 % (meaning that a given protein must be expressed/detectable in 70% of the samples within the specific tissue type) we detected 3200 protein groups (17570 peptides) in VAT and 5151 protein groups in liver tissue (35965 peptides).

The following databases were used to characterize function, cellular compartment and tissue distribution/expression of the significant proteins: Uniprot.org, proteinatlas.org, metabolicatlas.org, genecards.org and omim.org.

Illustrations were made in Perseus software and with use of biorender.com.

See Figure 1 for schematic overview of the study workflow.

3. Results

Table 1 depicts the clinical characteristics of the three MASLD groups (LS- n= 34, LS+ n=20, MASH n=16) and controls (CON n= 9). The number of patients with T2DM in each group was three (LS-), seven (LS+) and nine (MASH), respectively. Insulin resistance evaluated by HOMA-IR was progressively worsened from CON to MASH of about 4-fold. In the MASLD groups the BMI ranged from 41.8 - 44.6 kg/m² compared with 24.4 kg/m² the CON group.

3.1.1. Liver Tissue

In liver tissue we detected 32 differentially expressed proteins (DEPs) with significant up-or downregulation between the four groups (Table 2 and Figure 2). Of these, 18 were upregulated, and 14 were downregulated in the MASLD groups when compared to CON.

Six of the 14 downregulated proteins and three of the 18 upregulated proteins were ‘liver specific’ indicating an impaired synthesis of proteins in the liver (annotations were retrieved from the Human Protein Atlas [ref: https://pubmed.ncbi.nlm.nih.gov/21752111/]).

The DEPs grouped primarily in three predominant significance patterns: 1) Significant up-or downregulation in the MASLD groups (12 proteins) vs. CON with otherwise no significance among the four groups 2) Significant up- or downregulation between LS+ or MASH vs. CON and/or LS-. (ten proteins) 3) Significant up-or downregulation between LS+ and MASH combined vs. CON and/or LS- (seven proteins).

Six proteins were significantly altered in the MASH group compared with LS+, LS- and CON. One protein was upregulated (mevalonate kinase (MVK)) and five proteins were downregulated (hydroxysteroid 17-beta dehydrogenase 11 (HSD17B11), cyclin-dependent kinase inhibitor 1B (CDKN1B), death associated protein (DAP), methionine adenosyltransferase 1A (MAT1A)) and VAMP associated protein A (VAPA). Four proteins were significantly altered in the LS+ group compared with remaining groups; One was up-regulated (archain 1 (ARCN1)) and three were downregulated (cytochrome P450 family 3 subfamily A member 4 (CYP3AB), sideroflexin-2 (SFXN2), and heat shock protein family A (Hsp70) member 1B (HSPA1B).

3.1.2. Differentially Expressed Upregulated Liver Proteins (Table 2 and Figure 2)

Amongst the 18 upregulated proteins, 12 proteins showed the highest intensities in the MASH group and six proteins showed the highest intensities in the LS+ group. None of the proteins expressed highest intensity in LS- group.

When we sorted the upregulated proteins according to their main biological functions, the ”metabolism” group (proteins associated with primary cellular metabolic functions) comprised the highest number of proteins (eight proteins, IDH2, GPD1, OLA1, MVK, HPGD, PLIN1, PLIN2). As previously mentioned, MVK was the only upregulated DEP in MASH.

The perilipins (perilipin 1 (PLIN1) and perilipin 2 (PLIN2)) coat lipid droplets and are otherwise known to be highly expressed in adipocytes [21]. Out of the 32 hepatic DEPs PLIN1 and PLIN2 were the proteins with highest fold changes (around 3), but whereas PLIN1 was 3-fold higher in LS+ and MASH compared to CON (q-value = 0.006) for PLIN2 this 3-fold increase was observed between LS+/MASH and LS-, with no statistical significance when compared to CON.

Other liver proteins with significant up-regulation grouped in ‘intracellular transport proteins’ (four proteins) and proteins related to the cytoskeleton, intracellular signal transduction and extracellular matrix formation (five proteins). In the former group we found apolipoprotein 3 (APOL3), a lipid transporter which showed a 1-2.5-fold increase in the MASLD groups. Two proteins from the latter group, moesin (MSN) and radixin (RDX) were significantly upregulated in all three MASLD groups. Together with the protein ‘ezrin’, MSN and RDX are recognized as the ‘ezrin/radixin/Moesin (ERM) family’ [22]. Lastly, tumor protein p53 inducible protein 3 (TP53I3) was highly upregulated protein with tripled intensity in MASH and LS+ and doubled intensity in LS- when compared to CON.

3.1.3. Differentially Expressed Downregulated Liver Proteins (Table 2 and Figure 2)

Amongst the downregulated DEPs, 50% had the lowest intensities in MASH and 50% had the lowest intensities in LS+. None of the downregulated proteins had lowest intensities in LS-.

The largest groups of proteins (eight proteins; GLYCKTK, GRHDR, GCHFR, MAT1A, ACHY, HSD17B2, HSD17B11, CYP3A4) belonged to the ‘metabolism’ group. Glycerate kinase (GLYCKTK) which is involved in the catabolism of serine and metabolism of fructose was significantly downregulated in all three MASLD groups. Cytochrome P450 family 3 subfamily A member 4 (CYP3A4) was significantly downregulated in LS+.

3.1.4. Visceral Adipose Tissue

In VAT we found 59 DEPs between the four groups (Table 3 and Figure 3). In VAT 42 proteins (71%) were upregulated and 17 proteins (29%) were significantly downregulated. Moreover, we observed considerable differences in protein intensities with up to 8.5-fold higher or lower intensities between groups (primarily CON vs. one or more of the MASLD groups) for specific proteins. Significant differences for a given DEP between the groups were predominantly observed between the MASLD groups collectively and CON (34 upregulated and nine downregulated) followed by significance between MASH/LS+ and LS-/CON (four upregulated and two downregulated) with the rest of the proteins (four upregulated and five downregulated) showing other significance patterns.

Only a single protein, the mitochondrial sulfide:quinone oxidoreductase (SQOR), was exclusively downregulated in MASH. LS+ had significant upregulation of the serine/threonine-protein phosphatase 2A activator (PTBA), a chaperone protein governing DNA repair and protein folding.

3.1.5. Differentially Expressed Upregulated VAT Proteins (Table 3 and Figure 3)

Upregulated proteins were grouped according to main function and are depicted in Table 3. Nine of the upregulated proteins were involved in lipid metabolism. Notably, four of these (hydroxysteroid dehydrogenase (AKR1C1), steroid 5β-reductase (AKR1D1), aldose reductase AKR1B1, hydroxysteroid dehydrogenase (AKR1C3)) belonged to the aldo-keto reductase family.

We found upregulation of the complement system through complement C1 subcomponents C1r and C1q. They assemble with subcomponent C1s to form the C1 complex which is the first component in the serum complement system and activator of the classical (antigen-antibody) component pathway.

Other proteins with significant upregulation belonged to intracellular signaling, cytoskeletal and extracellular matrix proteins and proteins related to cell/organelle quality including several chaperones (heat shock protein beta-6 (HSPB6), crystallin alpha B (CRYAB), parkinsonism associated deglycase (PARK7) and coactosin-like protein (COTL1). Of these, HSPB6 also functions as a stress sensor [23]

Notably, myoglobin (MB), a protein found primarily in muscle with excellent oxygen binding and reservoir capacities showed prominent upregulation with 4-6 times higher expression in MASLD groups than CON. Lastly, several proteins aiding angiogenesis (e.g., tyrosin-3 monooxygenase (YWHAZ), RTN4 and HSPB6) were upregulated in MASLD groups.

3.1.6. Differentially Expressed Downregulated VAT Proteins (Table 3 and Figure 3)

The 17 downregulated proteins were scattered over several groups.

Five proteins were mitochondrial proteins and included the ATP synthase subunit alpha (ATP5A1), which is one of the core components of the ATP synthase complex in the electron transport chain.

In addition, the mitochondrial 2-oxoisovalerate dehydrogenase subunit alpha (BCKHDA) was also significantly downregulated in LS+ and in MASH by 2-fold compared with CON. BCKHDA is the alpha subunit of the decarboxylase component of the branched chain dehydrogenase (BCKD) complex that catalyzes the second and irreversible step in the catabolism of branched chain amino acids valine, leucine and isoleucine.

3.1.7. Associations Between VAT and Liver Tissue in MASLD: Correlation Data (Table 4).

24 proteins correlated significantly between VAT and liver tissue (Table 4). Of these, 22 correlated positively and two correlated negatively. The protein with the highest correlation coefficient was Hemoglobin subunit gamma-1 (HBG1) which is the gamma chain of fetal hemoglobin (α2ϒ2) (pearson r = 0.71). Interestingly, angiotensinogen (AGT), the precursor of angiotensin was the protein with the second strongest correlation coefficient.

Table 4. The 24 significant FDR adjusted proteins by correlation analysis between VAT and liver tissue.

Protein name	Gene name	q-value	Pearsons r	Location in cell	Tissue specificity	Main function(s)
Hemoglobin subunit gamma-1	HBG1	0.0000	0.717	C	Placenta	FetalHG, Oxb, OxT
Immunoglobulin heavy constant alpha 2	IGHA2	0.0000	0.630	PM, Secreted	Low	ImR
Angiotensinogen	AGT	0.0000	0.624	Secreted	Liver	RAAS, BPReg
Ribosyldihydronicotinamide dehydrogenase [quinone]	NQO2	0.0000	0.603	C	Low	DeTox, OxStress protector
Glutathione S-transferase theta-1	GSTT1	0.0001	0.595	C	Breast	GluthB, GluthM
NAD(P)HX epimerase	NAXE	0.0001	0.586	C, N	Low	NAD(P)HXrep
Immunoglobulin heavy constant mu	IGHM	0.0002	0.557	PM, Secreted	Low	ImR
Complement C4-A	C4A	0.0005	0.554	Secreted	Liver	CP, ImR(+)
Copine-1	CPNE1	0.0006	0.535	N, NM	Low	TNFaSig, TF
Enoyl-CoA hydratase domain-containing protein 3, mitochondrial	ECHDC3	0.0006	0.532	M	Liver, muscle	FAM, LiM
Complement C4-B	C4B	0.0010	0.521	Secreted	Liver	CP, ImR(+)
Afamin	AFM	0.0025	0.5	ECM, Se	Liver	PT, FAB
N-acetylneuraminate lyase	NPL	0.0452	0.492	VE, PM	Blood	CM
Epsin-1	EPN1	0.0041	0.491	PM, C	Low	Endocytosis
Acyl-coenzyme A thioesterase 1	ACOT1	0.0064	0.476	M	Liver	FAM, AcoAM
Phosphoglucomutase-1	PGM1	0.0068	0.473	C	Muscle	CM, GluM
Glycogenin-2	GYG2	0.0290	0.44	C, N	Adipose tissue, brain, breast	GlycB
Tyrosine-protein kinase CSK	CSK	0.0313	0.433	C, V	Lymphoid tissue	Reg, Imm
Heat shock protein HSP 90-alpha	HSP90AA1	0.0360	0.428	C	Vagina	Cha, HVI, SR
Interferon-induced protein with tetratricopeptide repeats 1	IFIT1	0.0367	0.426	C	Low	HVI, Imm
Histidine-rich glycoprotein	HRG	0.0455	0.415	Secreted	Liver	Angio(+), BC, Chem
Tetratricopeptide repeat protein 38	TTC38	0.0498	0.411	C, secreted	Liver, intestine	Unknown
60S ribosomal protein L38	RPL38	0.0453	-0.416	ER, C	Low	RibP, Translation
RAB14 member RAS oncogene family	RAB14	0.0450	-0.419	Er, C	Low	RibP, Translation
Oxb; oxygen binding, OxT; Oxygen transport, ImR; Immune response, RAAS; Renin-angiotensiogen-aldosterone-system, BPReg; Blood pressure regulator, DeTox; Detoxification process, OxStress; Oxidative stress, GluthB; Gluthation biosynthesis, GluthM; Gluthation metabolism, NAD(P)HXrep; Repairs NAD(P)H hydrates (NAD(P)HX), CP; Complement pathway, ImR; Immuneresponse, TNFaSig; TNF-alpha signalling,TF; transcription factor, FAM; Fatty acid metabolism, LiM; Lipid metabolism, PT; Protein transport, FAB; Fatty acid binding,AcoAM; Acetyl CoA metabolism, CM; Carbohydrate metabolism, GlycB; Glycogen breakdown, Imm; Adaptive immunity, Cha; Chaperone, HVI; Host-virus interaction Angio+; positive regulator of angiogenesis, BC; Blood coagulation, Chem; Chemotaxis, RibP; Ribosomal protein

Table 4. The 24 significant FDR adjusted proteins by correlation analysis between VAT and liver tissue.

Protein name	Gene name	q-value	Pearsons r	Location in cell	Tissue specificity	Main function(s)
Hemoglobin subunit gamma-1	HBG1	0.0000	0.717	C	Placenta	FetalHG, Oxb, OxT
Immunoglobulin heavy constant alpha 2	IGHA2	0.0000	0.630	PM, Secreted	Low	ImR
Angiotensinogen	AGT	0.0000	0.624	Secreted	Liver	RAAS, BPReg
Ribosyldihydronicotinamide dehydrogenase [quinone]	NQO2	0.0000	0.603	C	Low	DeTox, OxStress protector
Glutathione S-transferase theta-1	GSTT1	0.0001	0.595	C	Breast	GluthB, GluthM
NAD(P)HX epimerase	NAXE	0.0001	0.586	C, N	Low	NAD(P)HXrep
Immunoglobulin heavy constant mu	IGHM	0.0002	0.557	PM, Secreted	Low	ImR
Complement C4-A	C4A	0.0005	0.554	Secreted	Liver	CP, ImR(+)
Copine-1	CPNE1	0.0006	0.535	N, NM	Low	TNFaSig, TF
Enoyl-CoA hydratase domain-containing protein 3, mitochondrial	ECHDC3	0.0006	0.532	M	Liver, muscle	FAM, LiM
Complement C4-B	C4B	0.0010	0.521	Secreted	Liver	CP, ImR(+)
Afamin	AFM	0.0025	0.5	ECM, Se	Liver	PT, FAB
N-acetylneuraminate lyase	NPL	0.0452	0.492	VE, PM	Blood	CM
Epsin-1	EPN1	0.0041	0.491	PM, C	Low	Endocytosis
Acyl-coenzyme A thioesterase 1	ACOT1	0.0064	0.476	M	Liver	FAM, AcoAM
Phosphoglucomutase-1	PGM1	0.0068	0.473	C	Muscle	CM, GluM
Glycogenin-2	GYG2	0.0290	0.44	C, N	Adipose tissue, brain, breast	GlycB
Tyrosine-protein kinase CSK	CSK	0.0313	0.433	C, V	Lymphoid tissue	Reg, Imm
Heat shock protein HSP 90-alpha	HSP90AA1	0.0360	0.428	C	Vagina	Cha, HVI, SR
Interferon-induced protein with tetratricopeptide repeats 1	IFIT1	0.0367	0.426	C	Low	HVI, Imm
Histidine-rich glycoprotein	HRG	0.0455	0.415	Secreted	Liver	Angio(+), BC, Chem
Tetratricopeptide repeat protein 38	TTC38	0.0498	0.411	C, secreted	Liver, intestine	Unknown
60S ribosomal protein L38	RPL38	0.0453	-0.416	ER, C	Low	RibP, Translation
RAB14 member RAS oncogene family	RAB14	0.0450	-0.419	Er, C	Low	RibP, Translation
Oxb; oxygen binding, OxT; Oxygen transport, ImR; Immune response, RAAS; Renin-angiotensiogen-aldosterone-system, BPReg; Blood pressure regulator, DeTox; Detoxification process, OxStress; Oxidative stress, GluthB; Gluthation biosynthesis, GluthM; Gluthation metabolism, NAD(P)HXrep; Repairs NAD(P)H hydrates (NAD(P)HX), CP; Complement pathway, ImR; Immuneresponse, TNFaSig; TNF-alpha signalling,TF; transcription factor, FAM; Fatty acid metabolism, LiM; Lipid metabolism, PT; Protein transport, FAB; Fatty acid binding,AcoAM; Acetyl CoA metabolism, CM; Carbohydrate metabolism, GlycB; Glycogen breakdown, Imm; Adaptive immunity, Cha; Chaperone, HVI; Host-virus interaction Angio+; positive regulator of angiogenesis, BC; Blood coagulation, Chem; Chemotaxis, RibP; Ribosomal protein

Four of the ten proteins with highest correlation coefficients were related to the immune system; two were immunoglobulins (immunoglobulin heavy constant alpha 2 (IGHA2) and immunoglobulin heavy constant mu (IGHM)) and two belonged to the complement cascade, namely complement C4-A (C4A) and complement C4-B (C4B). The latter two are the components of the complement component C4, a central component of the classically activated pathway in the complement cascade.

Ribosyldihydronicotinamide dehydrogenase [quinone] (NQO2), glutathione S-transferase theta-1 (GSTT1) and NAD(P)HX epimerase (NAXE), proteins with the third, fourth fifth highest correlation coefficients are all involved in oxidative defense, detoxification and/or repair mechanisms.

4. Discussion

No Overlapping DEPs Between VAT and Liver Tissue in Subjects with Obesity, MASLD and MASH

This paper describes the largest human study with evaluation of the associations between the proteome in VAT and liver tissue to date.

Understanding the metabolic and functional disturbances in VAT and liver tissue is central to the understanding of the underlying pathophysiology of MASLD as it may also provide information on the metabolic disease risk in general. Considerable clinical overlap exists between the MASLD conditions (liver steatosis vs. MASH) but possibly with distinct patterns for different disease states. Protein patterns could here serve as phenotype identifiers and guide us towards biomarker discovery in MASLD. In that context it is important to map the proteomic signatures in VAT in relation to MASLD, as future biomarkers could reflect VAT dysfunction and in addition may be able to predict the progression to MASH, liver fibrosis and liver cirrhosis.

Where the untargeted proteomics approach in tissue has previously been applied in human studies of other metabolic diseases, T2DM especially [24,25] MASLD remains relatively unexplored and there are no human studies for specific comparison for VAT and liver tissue proteomics in the context of severity of MASLD.

We hypothesized that we would find specific proteins that were simultaneously and significantly up-or downregulated in the two tissues but we found no such proteins.

Parallel to the lack of overlapping proteins, we also did not find evidence of the presence of a VAT specific proteome that reflected the degree of MASLD severity and in general we found very little difference in the protein expression patterns between the MASLD groups. For example, we did not find inflammatory proteins in VAT to be generally more upregulated in MASH patients. In fact, we found merely a single VAT protein which was significantly upregulated in MASH only, namely SQOR. SQQR catalyzes the primary step in the metabolism of hydrogen sulfide (H₂S) within the mitochondria[26]. H₂S is a gasotransmitter, which is toxic in high concentrations and at low concentrations acts cytoprotective and is involved in many different biological functions including anti-inflammatory and proinflammatory abilities[27]. However, from this data we are not able to conclude, whether SQOR is upregulated due to high H₂S amount in MASH.

Based on the protein expression patterns we generally observed the tendency of CON and LS- group together and LS+ and MASH group together or that the significance of a given protein (e.g., AKR1B1, SLC25A20, RTN4, BCKDHA and ECI1) was found between LS+ and MASH vs. CON but not LS- vs. CON. Yet, the log2 intensities for the given protein were often still numerically much higher or lower in LS- compared with CON and thus may indicate a distinct biological difference – probably obesity - between these groups despite the absence of histological liver steatosis in both groups. The observed expression patterns in VAT therefore appear to reflect the metabolic deterioration that is associated with central obesity. The severity of MASLD is linked to the degree of metabolic dysregulation[28]. Our analyses are unable to determine if the observed differences reflect common risk factors or if there is a direct link between metabolic dysfunction in VAT and MASLD severity. Of note, among the included participants, those with MASH and LS+ were more likely to have T2DM, they had a higher HOMA-IR, and had higher levels of plasma liver enzymes. This was, however, not clearly reflected in the VAT DEPs.

Changes in the VAT Proteome in Obesity, T2DM and MASLD

No previous studies have investigated the proteome profile in human VAT in relation to MASLD but several have explored the human obese VAT proteome in relation to T2DM [29,30,31], in metabolically healthy vs. unhealthy subjects with obesity [32] as well as differences between SAT and VAT from individuals with obesity [33]. Several of the other VAT upregulated DEPs found in our study have been identified in other human obesity/T2DM studies e.g., annexin (ANXA5)[29,37] , liver carboxyl esterase 1 (CES1)[37] (48), C1QC [37](48) and myoglobin (MB)[38] . Moesin (MSN), which we found upregulated in the liver has similarly been found upregulated in VAT and SAT in subjects with obesity[29] . In addition, we found upregulation of C1QC, which is the c-chain of the C1qof human complement subcomponent C1q was significantly upregulated in all three MASLD groups. Additionally, we found upregulation of the perilipins PLIN1 and PLIN2 in liver tissue but not VAT. PLIN1 is believed to function as a lipid droplet protector that modulates the action of the hormone sensitive lipase in adipose tissue and thus helps regulate lipid metabolism. In its absence leanness is promoted on the expense of insulin resistance in PLIN1 knock-out mice[34]. PLIN1 and especially PLIN2 have previously been associated with the development of MASLD in rodent models and humans [35,36].

We also note the relative ‘overrepresentation’ of downregulated DEPs related to mitochondrial function and metabolism – e.g., BCKDHA, ATP5A1, PDHX, AK3 and SQOR, the latter being the single VAT DEP significant in MASH only. Prominent downregulation of mitochondrial proteins and proteins related to the respiratory machinery in obesity has recently been recognized in a diet-induced obesity mouse model [39], but systematic human data are lacking. The significant downregulation of BCKDHA is interesting as plasma levels of branched chain amino acids (leucine, isoleucine and valine) have been reported to be increased in insulin resistant individuals with obesity. In addition, adipose tissue is recognized as an important site for BCAA catabolism[40] . Other studies have also reported significant downregulation in obese adipose tissue of the enzymes responsible for BCAA catabolism [41], including the branched chain ketoacid dehydrogenase complex (BCKDHC) of which the BCKDHA is a subunit and that catalyzes the irreversible catabolic step in BCAA breakdown.

Correlation Analyses Pinpoint Inflammatory and Detoxification Proteins

In the correlation analyses between VAT and liver, the top 10 positive correlations were dominated by inflammatory proteins and proteins involved in oxidative defense mechanisms and detoxification processes. This could point towards simultaneously upregulation and overlapping pathophysiology in the two tissues. It could be argued that the high abundance of secretory proteins (Table 4) amongst the proteins with high correlation scores represent ‘contamination’ of plasma and blood vessels in the two tissues. However, we would then have expected to see high correlation scores of other plasma proteins with very high abundance in plasma -e.g., albumin or hemoglobin subunit A (normal hemoglobin). But this was not the case. Rather, the finding of IGHA2, IGHM, C4A and C4B is probably reflective of synergy in intrahepatic and intra-adipose tissue antibody production as a systemic response to regulation of immune homeostasis and inflammation.

The protein with the highest correlation score was fetal hemoglobin gamma chain (HBG1). Under normal physiological conditions (except pregnancy) fetal hemoglobin only exists in very limited amounts in adults comprising <0.6 % of total hemoglobin[43]. The synthesis of fetal hemoglobin, which has higher oxygen affinity than hemoglobin is confined to a population of erythrocytes termed the F-cells[43].. The presence of HBG1 and the high correlation score of 0.717 between the two tissues can only be speculative as there is very limited data on adult fetal hemoglobin in conditions other than β-thalassemia and sickle cell anemia. Also, we have not investigated the specific intensities of HBG1 between MASLD groups and CON in VAT and liver. We can only speculate as to whether severe obesity and/or metabolic disease perhaps induce an increase in F-cell erythrocytes that are capable of producing hemoglobin gamma-chains? Further research into this finding is warranted.

Changes in the Liver Proteome in Relation to Obesity and MASLD

In the liver tissue we found a couple of DEPs that were upregulated or downregulated in MASH and LS+ exclusively, but it is difficult to conclude on any specific patterns, as the DEPs were very heterogenous both in function and cellular compartment. However, we did observe a tendency of the upregulated proteins to be related to lipid and cholesterol metabolism (OLA1, ACLY, MVK, HPGD, APOL3, PLIN1 and PLIN2) and to cytoskeleton and ECM reorganization (COL18A1, RDX and MSN). Upregulation of ECM proteins in MASLD have been recognized in a study by Yuan and colleagues, who analyzed the liver tissue proteome from 12 patients with obesity classified as metabolically healthy but with obesity and 44 patients with obesity and MASLD according to liver histology[44] (53). By gene ontology analysis, they furthermore report significant downregulation of mitochondrial oxidative phosphorylation through downregulation of components of the complex I (NADH dehydrogenase complex) and complex IV (cytochrome c oxidase) in MASLD subjects. In comparison, we found significant downregulation of complex III (CYC1) but this downregulation was equal in all MASLD groups compared with CON.

We have previously investigated the plasma proteome in a study cohort comprising of 48 individuals with and without MASLD and liver cirrhosis and validated a promising panel of plasma proteins in a mouse model[18] (18). Among the promising proteins we found the polymeric immunoglobin receptor PIGR to be significantly upregulated in MASLD and to increase with increase in liver disease (obese without MASLD, T2DM with MASLD and patients with cirrhosis). PIGR is a transmembrane glycoprotein, a Fc receptor, that enables transcytosis of immunoglobulins from the basolateral to the apical surface of epithelial cells, thus mediating the secretion of IgA and IgM[45]. We did not confirm the findings of upregulated PIGR in the present data in neither liver nor adipose tissue. However, we did discover MAL proteolipid protein 2 (MAL2) to be roughly 8-fold increased in all three MASLD groups compared with CON. MAL2 was the protein showing the biggest difference in intensities between groups. This is interesting because MAL2 is an essential component of the transcytotic machinery[46,47]and has previously been implicated in PIGR mediated transcytosis where depletion of MAL2 blocked polymeric immunoglobulin receptor transcytosis in liver cells (the hep g2 cell line)[46] (55). Though data derived from cell lines should be evaluated with some caution the potential association between MAL2 in VAT and PIGR found in plasma in two different MASLD cohort could is still noteworthy. Adding to this notion, the proteins with the second and the sixth strongest correlations between VAT and liver tissue were the heavy chains of precisely IgA and IgM (IGHA2 and IGHM, respectively).

Strengths and Limitations

We wanted to explore potential overlapping DEPs in MASH between VAT and liver and to characterize the VAT proteome in MASH. The rationale was to assess common pathophysiological traits in the two tissues. However, this exploratory approach has limitations, that may hinder the true answer(s) to the hypotheses. By dividing the study cohort up into three MASLD groups based on histology and applying one-way ANOVA, we may have missed interesting and significant proteins. There are also some inherent drawbacks in proteomics studies in general. For instance, the generated data depend on the subjective threshold set for DEPs. In addition, the analysis discards most of the proteins and as a result focus on the significant DEPs which comprise only <0.1 % of the analyzed total proteome in the tissue. Furthermore, by filtering for 70% (meaning, that only proteins which are detectable in 70% of the samples enter the analysis) there is an inherent risk of filtering out proteins which were indeed significant in one group only. We may therefore have missed important proteins in one or more of the groups.Finally, the grouping of the study subjects according to the histopathological MASLD severity may pinpoint the weaknesses of the NAFLD activity score, as we observe very little difference in protein expression between the MASLD groups, particular between LS+ and MASH.

Proteomics studies and the generated data herein are in general difficult to compare due to significant heterogeneity in endpoints, grouping, analytical methodology (e.g., targeted vs untargeted approach) and data acquisition (use of FDR, considerations regarding significance). In addition, most studies have very low n, down to 4 per group[31], but typically n is around 10 per group. In comparison, our cohort consisted of 79 study subjects and the applied LC-MS analyses performed are of very high quality and robustness.

5. Conclusions

Altogether we found no overlap of proteins in VAT and liver tissue and neither in VAT nor in liver tissue did we find a MASH specific protein pattern, as DEPs, especially in VAT tended to group according to obesity and not in relation to MASLD severity. Consequently, from this exploratory study and the data analyses here applied the contribution from VAT in MASLD pathophysiology is ambiguous. However, several immunomodulating proteins correlated significantly between liver and VAT and could address shared pathophysiological characteristics but whether immunomodulation in the VAT affects the immunomodulation in the liver cannot be concluded from the present study.

As this dataset comprises the largest human proteome dataset to date in study subjects with severe obesity and MASLD this data can be used to generate and support hypotheses in MASLD pathophysiology and further explore disease traits.