Graphical Abstract
INTRODUCTION
Metabolic dysfunction-associated steatotic liver disease (MASLD), formerly known as nonalcoholic fatty liver disease (NAFLD) [1], affects a staggering one-quarter of the global population, posing a significant health burden [2-4]. MASLD comprises a spectrum of conditions, ranging from isolated hepatic steatosis, potentially progressing to metabolic dysfunction-associated steatohepatitis (MASH), which is marked by inflammation and fibrotic scaring of the liver, which can ultimately lead to cirrhosis and hepatocellular carcinoma. Lifestyle modification and emerging pharmacological approaches offer promising treatment for MASLD. However, given the heterogeneous nature and complexity of MASLD the disease still pose a significant clinical challenge. For example, bariatric surgery is effective for treating MASLD in individuals with obesity [5]. However, there are many people with MASLD who are not overweight, and thus this option is not suitable for treatment at the population level. Recently, resmetirom, a thyroid hormone beta receptor agonist, has been approved by the US Food and Drug Administration to treat MASH and fibrosis. However, only 1 in 4 study participants responded with a meaningful reduction in fibrosis [6], making the identification of novel therapeutics still of paramount importance.
MASLD has a strong genetic component, with a common variant in the PNPLA3 gene exhibiting the most robust effect size. Other common variants in GCKR, TM6SF2 and MBOAT7 also contribute to MASLD risk [7-9]. A genome-wide association study (GWAS) for an all-cause liver cirrhosis phenotype identified a common missense variant (rs2642438) in the mitochondrial amidoxime-reducing component 1 (MARC1) gene as an additional genetic determinant of MASLD [10]. The MARC1 rs2642438 minor allele robustly associates with lower liver triglycerides, reduced transaminase levels, and protection against cirrhosis and hepatocellular carcinoma [10]. Interestingly, the protective effect of the MARC1 variant extends to alcoholic liver disease [11].
The MARC1 rs2642438 results in an amino acid substitution from alanine (A) to threonine (T) at position 165 of the protein (p.A165T), leading to reduced protein stability and higher proteasomal degradation [12-14]. Altogether, this these data suggest that the downregulation in the levels of hepatic MARC1 may potentially be beneficial in reducing the risk of liver steatosis.
However, the molecular mechanisms underlying MARC1-related protection against MASLD remain inconclusive. It has been shown that MARC1 p.A165 downregulation does not affect de novo lipogenesis or fatty acid oxidation in primary human hepatocytes (PHHs), despite a decrease in 3-hydroxybutyrate (3HB), a by-product of β-oxidation [15]. In contrast [16], overexpression of the protective, but not of the risk allele, of MARC1 resulted in a higher oxygen consumption rate in hepatoma cells, as measured by Seahorse assay, suggesting a gain of function of the protective mutant protein. However, MARC1-downregulation in primary human adipocytes resulted in an upregulation of CPT1A, the gene coding a key β-oxidation enzyme [16], suggesting the variant is a loss of function.
Given these contradictory observations we sought to elucidate the specific mechanisms underlying the MARC1-related protective effect against MASLD in PHHs. Here, we present findings that shed light on the complex role of MARC1 in regulating hepatic lipid homeostasis and cellular stress responses.
MATERIALS AND METHODS
Cell culture
Cryopreserved PHHs were obtained from the commercial vendor, BioIVT. They were seeded with CP media and maintained with HI media (BioIVT). Detailed methods are available in the supplementary information.
siRNA transfection
Cells were transiently transfected for 48 hours, with scramble siRNA 30 nM (AM4611; ThermoFisher Scientific, Cleveland, OH, USA) or MARC1 siRNA (mix of s34872, s34873, s34874; concentration 30 nM each; Thermo Fisher Scientific) with Lipofectamine 3000, as per the manufacturer’s instructions. PHH maintenance medium (HI medium, BioIVT) was used for transfection and maintenance in PHH experiments.
RNA-sequencing analysis
Total RNA was extracted using miRNeasy Tissue/cells Advanced Mini Kit (QiaGen Inc., Valencia, CA, USA). Reads were aligned to GRCh38 reference genome by STAR53 (v2.7.10a) and gene-level read counts were quantified by RSEM52 (v1.3.3) software against the Ensembl (release 107). The DESeq2 R package (v.1.38.3) was used to perform a differential expression analysis. The resulting P-values were adjusted using Benjamini and Hochberg’s approach for controlling the false discovery rate (FDR).
Proteomic sample preparation
Details for the experiment are provided in supplementary information. Samples were processed using a modified SP3 method [17]. The protein abundance ratio was transformed into the log2 ratio and a t-test was performed. Using the enrichment analysis from FerrDB [18] database for ferroptosis, a list of proteins for each driver, regulator or suppressor of ferroptosis was obtained and the heatmap of the proteins was plotted from our proteomics data using the fold change values. The odds ratio (OR) of ferroptosis-related proteins in the list of all significantly different levels of proteins after MARC1 downregulation is shown.
Oil-Red-O (ORO) staining
Forty-eight hours after transfection, cells were subjected to ORO staining (Sigma-Aldrich, St. Louis, MO, USA) and counterstained using DAPI (Sigma-Aldrich) as previously performed [19]. Images were captured at 40X magnification with NIS-Element 5.30.04 software (Nikon Instruments, Inc., Melville, NY, USA) and analyzed using an in-house macro in ImageJ (v.1.52h, NIH). ORO-stained neutral lipid content was normalized to the number of DAPI stained nuclei.
β-oxidation assay
The β-oxidation rate was assessed as previously described [19,20]. Cells were incubated with HI medium supplemented with 8.3 µCi/mL [3H]palmitate (PerkinElmer, Denton, TX, USA) and 110 µM PA (Avanti Polar Lipids, Alabaster, AL, USA)) and 1% BSA for 2 hours. Cell media were collected and centrifuged at 12,000 rpm for 5 minutes after adding 50 μL of 20% BSA and 27 μL of 70% perchloric acid. The supernatant was collected, and the radioactivity of the water-soluble fraction was quantified by liquid scintillation counting (Beckman Coulter, Brea, CA, USA). Data were normalized for the number of cells.
Western blotting
For Western blot analysis in cell lysate, total proteins were obtained using M-PER® containing a complete protease inhibitor cocktail (Sigma-Aldrich). For ApoB100 detection, cell media were collected and subjected to ultracentrifugation at a potassium bromide (KBr) density of 1.063 g/mL to separate TG-rich-lipoproteins (TRL) fraction. Proteins were then separated by 10% (for ATG7), 12% (for MARC1) or 6% (for ApoB100) SDS-PAGE and transferred to nitrocellulose membranes (Amersham; Cytiva, Buckinghamshire, UK).
Statistical analysis
Data are presented as mean±standard deviation. P-values were calculated using a non-parametric Mann-Whitney test unless specified otherwise. Statistical analysis was performed using GraphPad Prism version 9 (San Diego, CA, USA). All the reported P-values are two-sided. The P-values<0.05 were considered statistically significant.
The association between MARC1 rs2642438 and rank inverse transformed 3-hydoxybutyrate in Europeans from the UK Biobank (n=239,075) was performed using linear regression analysis in REGENIE adjusting for age, sex, age×sex, age2 and age2×sex, body mass index, first 10 genomic principal components and array batch [21,22].
Detailed method for all the other experiments and descriptions are available in the supplementary information.
RESULTS
Downregulation of the MARC1 p.A165 risk allele reduces hepatocyte neutral lipid content
To test the effect of MARC1 ablation on intracellular triglyceride homeostasis, we downregulated MARC1 by siRNA in PHHs carrying genes encoding for the wild-type (p.A165) risk or the (p.T165) protective protein. We found that a 70% reduction in MARC1 gene expression (Fig. 1C) resulted in lower intracellular neutral lipid content, as measured by Oil-Red-O staining and its quantification in hepatocytes homozygote for the risk allele (Fig. 1A) but not in the protective allele (Fig. 1B). Next, we repeated the same experiment in four different cell lines homozygous for the risk allele; namely, HepG2, HuH7, Hep3B2 and HepaRG with virtually identical results (Supplementary Fig. 1). We also repeated the same experiment in another PHH donor that was homozygous for the risk allele and observed similar results (Supplementary Fig. 2A). Next, to test the effect of the p.A165T substitution at the protein level, we measured the intracellular MARC1 protein levels by Western blotting in PHHs homozygous for either the risk or the protective allele (Fig. 1D). We detected a 50% reduction in protein levels in PHHs homozygous for the protective compared to the risk allele. These findings align with previous studies demonstrating accelerated degradation of the protein encoded by the protective allele [12].
To further understand the molecular mechanisms underlying the protection conferred by the downregulation of MARC1, we focused on PHHs from donors homozygous for the wild-type (risk) allele.
Downregulation of the MARC1 p.A165 risk allele promotes fatty acid catabolism via increased β-oxidation
In hepatocytes, lipid accumulation can be due to perturbation of three major pathways: a) β-oxidation; b) secretion of triglyceride-rich ApoB100-containing lipoproteins; or c) de novo lipogenesis. Thus, we examined the effect of MARC1 downregulation on these 3 major pathways. We found that downregulation of MARC1 p.A165 (risk) resulted in higher β-oxidation as measured by radioactive tracers in PHHs (Fig. 2A; Supplementary Fig. 2B). We did not see differences in a) ApoB100 secretion as measured by Western blotting, b) triglycerides concentrations in the culture media as measured by mass spectrometry analysis (MS) (Fig. 2B; Supplementary Fig. 2E–H) or c) in intracellular de novo triglyceride biosynthesis media as by radiolabelled glycerol incorporation into triglycerides (Fig. 2C; Supplementary Fig. 2C). In contrast, downregulation of MARC1 p.T165 (protective) did not have any differences in these pathways (Supplementary Fig. 3).
Next, to examine the impact of high fatty acid stress on lipid metabolism, we incubated PHHs homozygous for either the risk or protective allele with a 300 µM fatty acid mixture (oleic acid and palmitic acid, 1:1 ratio; for 48 hours). Similar to the previous experiment (Fig. 2A), we found that MARC1 p.A165 (risk) downregulation resulted in an increased β-oxidation, nearly doubling the rate (Fig. 2D). In contrast, β-oxidation levels remained unchanged in the cells expressing MARC1 p.T165 (protective) (Fig. 2D). We did not detect differences in ApoB100 protein level in the triglyceride-rich lipoprotein, as quantified by Western blot, and in triglyceride levels as measured by MS analysis regardless of the MARC1 genotype (Fig. 2E). These data suggest that MARC1 p.A165 downregulation reduces intracellular neutral lipids content primarily by increasing fatty acid breakdown via elevated β-oxidation.
To translate our results to humans, we examined the 3HB plasma levels, a proxy of β-oxidation, produced primarily in the liver [23,24], in carriers of the MARC1 rs2642438 variant (encoding for the p.A165T substitution) in 239,075 individuals from the UK Biobank (Fig. 2F). The data was adjusted for different clinical conditions that can potentially affect β-oxidation. The minor allele encoding for the mutant p. T165 protein was associated with higher levels of 3-HB demonstrating that inactivation of MARC1 results in higher β-oxidation.
MARC1 p.A165 knockout affects intracellular lipid composition
To investigate lipid quantity and composition in cells with a stable ablation of MARC1, we used CRISPR/Cas9-mediated stable KO HepG2 cells (homozygous for the risk allele). Raman imaging was utilized to analyze lipids based on unambiguous lipidic Raman markers, like the intensity ratio between CH2 vibrations (2,850 and 2,880 cm-1) and CH3 vibrations (2,930 cm-1) [25,26]. Total lipid content was calculated for each cell as the ratio of lipid-assigned areas to total cell area (Fig. 3A, 3B, Supplementary Figs. 4, 5), and we observed lower total lipid content in MARC1 KO cells.
Based on the spectral characteristics there were three levels of saturations as shown by different colors superimposed on optical images of the cells (Fig. 3A); namely, saturated fatty acids (SFAs), monounsaturated fatty acids (MUFAs) and di-unsaturated fatty acids (DUFAs) (Fig. 3C). An area-based quantification of these lipid species (Fig. 3D) showed that in MARC1 KO cells there were lower levels of SFAs-MUFAs and MUFAs, with slightly increased level of DUFAs.
Next, we performed basis analysis (BA) including a spectral base with calibration curves for unsaturated and saturated cholesterol esters (CEs) (Supplementary Fig. 6). We generated fluorescence-like pseudo-color images of cells (Fig. 3E) and quantified the results (Fig. 3F), which showed lower saturated CEs content in MARC1 KO cells with no changes in unsaturated CEs compared to mock cells.
The fatty acid metabolism, tricarboxylic acid cycle (TCA), and oxidative phosphorylation pathways are upregulated after MARC1 downregulation
To gain insight into the intracellular pathways modified by MARC1 downregulation, we performed transcriptomic analyses of PHHs homozygous for the risk allele. Among 13,787 identified transcripts, 65 were significant at P<0.05 after adjusting for multiple testing. Among these, 19 genes were downregulated and 46 were upregulated. The top 30 differentially expressed genes (DEGs) are represented in Figure 4A, 4B. Furthermore, we performed pathway enrichment analysis of all the significant DEGs and found the result showed that ‘xenobiotic metabolism’ was downregulated, whereas ‘fatty acid metabolism’ was upregulated after MARC1 downregulation (Supplementary Fig. 7). To further confirm the results, we performed proteomic analyses (n=3) by using LC-MS3 in cells transfected with either scramble siRNA or MARC1 siRNA. Among the 3,900 proteins identified, 367 were differentially expressed (P-value<0.05). Among these, 185 proteins were lower and 182 proteins were higher. The top 30 differentially expressed proteins and MARC1 are plotted in a heatmap (Fig. 4C). Among these top 30 proteins, the genes for 5 of them (ALCAM, IL6ST, GOLGA4, STEAP3 and SCAMP1) were also differentially expressed in the same direction in the transcriptomic analyses (Fig. 4A). Notably, consistent with the transcriptomic results, we observed upregulation of the ‘fatty acid metabolism’ pathway after MARC1 downregulation, as found in pathway enrichment analysis (Supplementary Fig. 8). Additionally, ‘oxidative phosphorylation’, ‘ROS pathway’, and ‘TCA cycle pathway’ were upregulated after MARC1 downregulation. Upregulation of all these pathways is consistent with higher mitochondrial activity favouring fatty acid breakdown and activation of downstream metabolic pathways.
Collectively, our results from radioactive tracer studies, UKBB human genetics data, transcriptomics and proteomics in PHHs support the notion that MARC1 downregulation results in higher fatty acids utilization via elevated β-oxidation.
Ferroptosis suppressor proteins are upregulated after MARC1 downregulation
In our proteomic analyses, we observed that ‘heme metabolism’ and associated pathways were enriched in MARC1 downregulated cells (Supplementary Fig. 8). Ferroptosis has been shown to be an important pathway in MASLD susceptibility [27,28]. Therefore, we next performed a targeted analysis where we examined the status of ferroptosis (suppressor, regulator or driver of ferroptosis) by examining the 185 downregulated and 182 upregulated proteins. Notably, we found that only the ferroptosis suppressor proteins were upregulated in MARC1 downregulated cells (Fig. 4D–E), whereas drivers of ferroptosis did not change.
MARC1 downregulation reduces reactive oxygen species levels
To gain insights into how MARC1 downregulation may be beneficial in reducing the risk of MASLD, we examined cellular stress responses in PHHs. MARC1 downregulation resulted in lower levels of ROS (Fig. 5A), suggesting an improvement in the cellular redox state.
Next, to test lipid peroxidation levels, we incubated PHHs with a polyunsaturated fatty acid (PUFA) mixture of 200 µM linoleic acid (LA) and 200 µM arachidonic acid (AA) for 48 hours. Due to their high degree of unsaturation, PUFAs are more susceptible to free radical attack, leading to lipid peroxidation. Incubating cells with PUFAs essentially creates a controlled environment rich in substrates prone to lipid peroxidation. We did not observe differences in lipid peroxidation levels as measured by 4-hydroxynonenal (4HNE) staining after MARC1 downregulation (Fig. 5B). Additionally, we did not observe any alterations in lipophagy, a cellular process responsible for the degradation of lipids. This was measured by changes in the protein levels of autophagy-related gene 7 (ATG7) by Western blotting, or in other markers as quantified by our proteomics analysis (Fig. 5C, 5D).
DISCUSSION
The main finding of this work is that the MARC1 rs2642438 minor allele encoding the p.T165 variant results in lower hepatocyte triglyceride content by increasing β-oxidation in PHHs. Moreover, the downregulation of MARC1 p.A165 results in a more favourable phenotype by reducing ferroptosis and ROS levels.
We started by examining PHHs from human donors homozygous for either the risk (p.A165) or the protective (p.T165) allele of MARC1, incubated with different fatty acid concentrations. Downregulation of MARC1 reduced neutral lipid content in hepatocytes homozygous for the wild-type (and risk) allele (p.A165) but not in those homozygous for the protective allele (p.T165). Moreover, when we compared protein levels between MARC1 p.A165 and p.T165, we observed ~50% reduction of MARC1 protein levels in hepatocytes with the protective (p.T165) allele. This is consistent with our previous observation that the MARC1 rs2642438 variant is a loss of function and that the protein levels are lower due to increased proteasomal degradation [12].
Intracellular triglyceride homeostasis in hepatocytes is governed by three fundamental pathways; namely, β-oxidation, de novo lipogenesis and lipoprotein-mediated lipid secretion. β-oxidation breaks down fatty acids to produce viable energy, whereas de novo lipogenesis is the process whereby new fatty acids are synthesized, usually in the presence of a positive caloric balance [29], and lipoprotein-mediated lipid secretion is used to secrete lipid species from the hepatocytes.
Previous studies have reported contradictory results regarding the means by which MARC1 affects hepatocyte lipid homeostasis. It has been shown that MARC1 p.A165 downregulation does not affect de novo lipogenesis or fatty acid oxidation in PHHs, despite a decrease in 3-HB, indicating decreased β-oxidation [15]. In contrast [16], overexpression of the protective, but not of the risk allele, of MARC1, resulted in a higher oxygen consumption rate in hepatoma cells, as measured by Seahorse assay, suggesting a gain of function of the protective mutant protein. However, MARC1 downregulation in primary human adipocytes resulted in an upregulation of CPT1A, a key β-oxidation enzyme [16], suggesting the variant is a loss of function.
Here, we demonstrate that MARC1 (p.A165) downregulation reduces intracellular lipid content by increasing fatty acid utilization via elevated β-oxidation in PHHs. We were able to show this outcome by using 3 different methodologies: a) radioactive tracers, b) transcriptomics, and c) proteomics. Moreover, we demonstrate that MARC1 ablation modifies the quality of lipid species, as observed by Raman spectroscopy in MARC1 KO HepG2 cells. Indeed, MARC1 ablation resulted in lower levels of saturated and monounsaturated fatty acids and lower levels of cholesterol esterified with saturated fatty acids. The relevance of these qualitative changes for liver disease remains to be determined.
Finally, to test the translatability of our findings in humans, we examined differences in plasma levels of 3-HB in participants of the UK Biobank stratified by MARC1 rs2642438. 3HB is synthesized primarily in hepatocytes as a by-product of β-oxidation and secreted into the bloodstream as a ketone body and therefore serves as a proxy for the activity of hepatocyte β-oxidation. Consistent with our data in PHHs in vitro, circulating 3-HB levels were higher in carriers of the rs2642438 minor allele compared to wild-type individuals, indicating higher β-oxidation activity in these people.
Among the intracellular pathways, lipophagy is a specialized form of autophagy where lipid droplets are specifically targeted for lysosomal degradation [30]. However, we observed no difference in the lipophagy markers after MARC1 downregulation.
Ferroptosis plays a pivotal role in the progression of liver disease [31]. While ferroptosis can target dysfunctional cells, excessive cell death mediated by ferroptosis can lead to inflammation and worsening of MASLD. We observed that the ferroptosis suppressor proteins were upregulated in MARC1 downregulated PHHs in our proteomic analysis. This strongly indicates that MARC1 downregulation leads to reduced ferroptosis in PHHs.
As MARC1 is located in the mitochondrial outer membrane [12,14,32,33], and mitochondrial dysfunction leads to elevated cellular ROS levels [34], we measured the overall levels of ROS after MARC1 downregulation in PHHs. Interestingly, even though MARC1 downregulation leads to higher fatty acid degradation through β-oxidation, the ROS levels in the PHHs were reduced.
Under physiological conditions, ROS are produced during the β-oxidation process, especially in the case of incomplete shuttling of electrons in the electron transfer chain (ETC) [34,35]. However, in the presence of either an efficient mitochondrial antioxidant pathway or efficient ETC, cellular ROS levels are reduced [34,36]. Overall, the reduction in ROS is compatible with the beneficial effect of the MARC1 variant against MASLD. We should note that measurement of the oxygen consumption rate could have strengthened our omics-based findings that downstream pathways of β-oxidation and mitochondrial activity were upregulated after MARC1 downregulation. This remains a limitation of this study. However, our results are consistent with two other studies on MARC1 in different models that, showed increased levels of antioxidant pathways after MARC1 downregulation [16,37], although, we did not observe any change in the lipid-peroxidation level after MARC1 downregulation, as quantified by 4-HNE immunofluorescence in PHHs.
In conclusion, our results demonstrate that MARC1 inhibition in carriers of the risk protein results in lower hepatocyte lipids due to higher β-oxidation. Moreover, MARC1 downregulation upregulates beneficial pathways involved in cell survival. These data suggest that downregulation of MARC1 may be a potential strategy to reduce liver steatosis in humans.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Supplement1
Print
