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
Corresponding author. , Novo Nordisk Research Centre Oxford, Novo Nordisk Ltd, Innovation Building, Old Road Campus, Roosevelt Drive, OX3 7FZ, Oxford. Telephone: +44 7979927835.
Oxford Centre for Diabetes, Endocrinology and Metabolism, Radcliffe Department of Medicine, University of Oxford, Churchill Hospital, Oxford, UKNational Institute for Health Research Oxford Biomedical Research Centre, Oxford University Hospital Trusts, UK
Corresponding author. , Novo Nordisk Research Centre Oxford, Novo Nordisk Ltd, Innovation Building, Old Road Campus, Roosevelt Drive, OX3 7FZ, Oxford. Telephone: +44 7979927835.
Genetically predicted increases in MTARC1 mRNA associate with poor liver health.
•
MARC1 knockdown decreases lipid accumulation in primary human hepatocytes.
•
Hepatocyte specific knockdown of mARC1 improves steatosis in a murine NASH model.
Abstract
Background and Aims
Non-alcoholic fatty liver disease (NAFLD) has a prevalence of ∼25% worldwide, with significant public health consequences, yet few effective treatments. Human genetics can help elucidate novel biology and identify targets of new therapeutics. Genetic variants in mitochondrial amidoxime reducing component 1 (MTARC1) have been associated with NAFLD and liver-related mortality, however, its pathophysiological role and the cell type(s) mediating these effects remain unclear. We aimed to investigate how MTARC1 exerts its effects on NAFLD by integrating human genetics with in vitro and in vivo studies of mARC1 knockdown.
Methods
Analyses including multi-trait colocalization and mendelian randomization were used to assess the genetic associations of MTARC1. In addition, we established an in vitro long-term primary human hepatocyte model with metabolic readouts and used the Gubra GAN-diet NASH mouse model treated with hepatocyte specific GalNAc-siRNA to understand the in vivo impacts of MTARC1.
Results
We show that genetic variants within the MTARC1 locus are associated with liver enzymes, liver fat, plasma lipids and body composition and these associations are due to the same causal variant (p.A165T, rs2642438G>A), suggesting a shared mechanism. We demonstrated that increased MTARC1 mRNA had an adverse effect on these traits using Mendelian Randomization, implying therapeutic inhibition of mARC1 could be beneficial. In vitro mARC1 knockdown decreased lipid accumulation and increased triglyceride secretion and in vivo GalNAc-siRNA mediated knockdown of mARC1 lowered hepatic, but increased plasma triglycerides. We found alterations in pathways regulating lipid metabolism and decreased secretion of 3-hydroxybutyrate upon mARC1 knockdown in vitro and in vivo.
Conclusions
Collectively, our findings from human genetics, and in vitro and in vivo hepatocyte-specific mARC1 knockdown support the potential efficacy of hepatocyte-specific targeting of mARC1 for treatment of NAFLD.
Lay Summary
A genetic mutation that causes an amino acid substitution in MTARC1 has previously been shown to reduce the risk of developing non-alcoholic fatty liver disease. Here we show that genetic mutations that increase the messenger RNA levels of MTARC1 lead to poor liver health, including the accumulation of fat. Experimentally, reducing mARC1 function in liver cells was found to reduce lipid content in primary human cells in a dish and in intact mice.
Hypothesis Prioritisation in multi-trait Colocalization
IV
Instrumental variables
siMTARC1
MTARC1 targeting siRNA
MR
mendelian randomization
MTARC1
mitochondrial amidoxime reducing component 1
GalNAc
N-Acetylgalactosamine
NAFLD
Non-alcoholic fatty liver disease
NASH
non-alcoholic steatohepatitis
ns
non-significant
siNT
non-targeting siRNA
PPAR
peroxisome proliferator-activated receptor
PBS
phosphate-buffered saline
PHH
primary human hepatocyte
siDGAT2
siRNA targeting DGAT2
TC
total cholesterol
TG
triglycerides
UKB
UK Biobank
VLDL
very low density lipoproteins
Conflicts of interest
A provisional patent application directed to the subject matter disclosed in this manuscript has been filed by Novo Nordisk A/S. L.C.L, L.C, L.S.H, R.R.K, C.M, J.N, C.E.D, S.T.H, R.P, I.S, E.J.L, T.N.D, A.C, S.H, B.G, M.E.G, E.T.M, A.H.A.E, W.G.H, K.C, J.F, J.M.M.H, B.A and M.A.R are Novo Nordisk A/S or Novo Nordisk Ltd employees. E.W and N.W.P. are former Dicerna employees. S.B.R. is a former employee of Dicerna Pharmaceuticals Inc. And presently employed by the Novartis Institutes for BioMedical Research (NIBR), Cambridge, MA, USA. W.L. is a former employee of Dicerna Pharmaceuticals and presently employed with Biogen Inc, Cambridge, MA, USA. L.H. is a scientific consultant for Novo Nordisk A/S. The remaining authors declare no competing interests.
Financial Support statement
This work was supported by Novo Nordisk. S.R.N. has a Novo Nordisk Postdoctoral Fellowship run in partnership with the University of Oxford and the British Heart Foundation (Fellowship FS/15/56/31645 and FS/SBSRF/21/31013 to L.H).
Author contributions
L.C.L., M.A.R., L.S.H., R.P., E.J.L., T.N.D., I.S., B.G., A.C., M.E.G., A.H.A.E., C.E.D., S.T.H., J.N., S.R.N., E.W., S.B.R. and W.L. performed experiments and procedures. L.C.L., M.A.R., C.E.D., N.W.P., B.A., S.T.H., K.C., J.F., J.M.M.H. and W.G.H. Contributed to concept and design of the study. L.C.L., M.A.R., E.J.L., T.N.D., I.S., C.E.D., R.R.K., E.M.T., L.C., C.M., S.H., N.W.P., B.A., S.T.H., J.F., J.N., S.R.N., E.W., S.B.R. and W.L. contributed to formal analysis and data curation. L.C.L., M.A.R., C.E.D., R.R.K., L.C., C.M., J.M.M.H., B.A., S.T.H. and S.R.N. were involved in original draft preparation. M.A.R., B.A., J.M.M.H., L.H., J.F. and K.C. supervised the study. All authors reviewed and approved the final manuscript.
Data availability statement
The authors confirm that the data supporting the findings of this study are available within the article and supplementary information. The transcriptomic and metabolomic datasets are available in GEO (GSEXXX) and MetaboLights (MTBLS6909),
Non-alcoholic fatty liver disease (NAFLD) encompasses a spectrum of histologically categorised pathologies including simple steatosis, non-alcoholic steatohepatitis (NASH) and cirrhosis.
. Elucidating the impact of genetic variation on disease onset and progression has identified novel therapeutic targets for this major unmet medical need.
. Genome-wide association studies (GWAS) identified and validated a common missense variant (p.A165T, rs2642438 G>A) in mitochondrial amidoxime reducing component 1 (MTARC1) that protects against all-cause cirrhosis.
Vujkovic M, Ramdas S, Lorenz KM, Guo X, Darlay R, Cordell HJ, et al. A genome-wide association study for nonalcoholic fatty liver disease identifies novel genetic loci and trait-relevant candidate genes in the Million Veteran Program. 2021.
. A low-frequency coding variant (p.M187 K, rs17850677 T>A) and a rare stop codon variant (p.R200Ter, rs139321832 C>T) in MTARC1 are associated with lower blood cholesterol and protection from cirrhosis
, suggesting mARC1 inhibition may have therapeutic potential for the treatment of liver disease. A slight increase in plasma triglycerides (TG) was observed in carriers of the minor A allele, however, cardiometabolic disease-based association studies of this variant in MTARC1
does not raise any concerns for mARC1 modulation. Effects of mARC1 beyond cardiometabolic disease has not been well captured and trait-based phenome-wide association studies may inform mARC1 biology and safety profile.
The molecular function of mARC1 within NAFLD pathophysiology is unclear. mARC1 is a molybdenum-containing enzyme anchored to the outer mitochondrial membrane
The N-reductive system composed of mitochondrial amidoxime reducing component (mARC), cytochrome b5 (CYB5B) and cytochrome b5 reductase (CYB5R) is regulated by fasting and high fat diet in mice.
, little is known regarding the metabolic consequences of altering mARC1 levels. Depletion of mARC2, a mARC1 paralogue, decreases formation of glycerolipids in adipocytes and mARC2 knockout mice are protected from diet-induced obesity and associated metabolic disturbances.
. As mARC1 is highly expressed in adipocytes, the sufficiency of targeting mARC1 in hepatocytes to confer the genetically predicted protective metabolic effects is unknown.
Here, we compile in silico, in vitro and in vivo work to establish that hepatocyte MTARC1 is causally associated with NAFLD. We identified novel genetic associations with body composition and liver enzymes; demonstrate a shared genetic aetiology of NAFLD-related traits at the MTARC1 locus and establish a causal association of genetically predicted MTARC1 expression with these traits. mARC1 knockdown in a long-term primary human hepatocyte (PHH) system showed decreased lipid accumulation. In a murine model of NASH, we found hepatocyte-specific mARC1 depletion decreases steatosis and markers of fibrosis. Together this work provides the first experimental evidence that inhibition of hepatocyte mARC1 is sufficient to reduce hepatic steatosis.
Materials and methods
CTAT table available as a Supplementary Document.
Genetic analyses
All analyses were performed in R v3.6.3 unless otherwise stated. Full details are provided in the Supplementary Materials and Methods and Tables S1-3.
Rare variant association tests in the UK biobank (UKB)
We tested the association of rare variants in the region of MTARC1 for association with the blood biomarkers (alanine aminotransferase (ALT), aspartate transaminase (AST) and alkaline phosphatase (ALP)) in 414,763 European ancestry participants in the UKB with whole exome sequencing.
The Nile Red ratio was calculated as the neutral lipid fluorescence (540-15 nm/600-20 nm) divided by the phospholipid fluorescence (540-15 nm/640-20 nm).
. To normalise data across experiments non-targeting siRNA (siNT) values were set to 0% for 0 μM FFA mix and 100% for 800 μM FFA mix treatment, using the following equation:
High Resolution Mass Spectrometry Improves Data Quantity and Quality as Compared to Unit Mass Resolution Mass Spectrometry in High-Throughput Profiling Metabolomics.
Journal of Postgenomics Drug & Biomarker Development.2014; 4
Two in vivo studies were carried out. Male C57BL/6JRj mice (n=10-14 or n=10-16), 5 weeks old were fed the GAN diet (40 % fat, 22 % fructose and 2 % cholesterol, D09100310, Research Diets, USA) for 44 or 36 weeks respectively. Mice were dosed weekly subcutaneously for the last 8 weeks with 3 mg/kg N-Acetylgalactosamine (GalNAc)-conjugated siRNA targeting murine Mtarc1 (GalNAc-siMtarc1). Data were compared to a NASH vehicle control group. All animal experiments were conducted in accordance with Novo Nordisk and Gubra bioethical guidelines, which are fully compliant to internationally accepted principles for the care and use of laboratory animals. The described experiments were covered by personal licenses for Jacob Jelsing (licenses #2013-15-2934-00784 and #2017-15-0201-01215) issued by the Danish Committee for animal research. See support-information-section" title="https://aasldpubs.onlinelibrary.wiley.com/doi/full/10.1002/hep.32063 - support-information-section">Supplementary Materials and Methods for details and Table S6. Mouse plasma metabolites were analysed by MS-Omics using a Thermo Scientific Vanquish LC coupled to Thermo Q Exactive HF MS. An electrospray ionisation interface was used as the ionisation source in negative and positive ionisation mode. The UPLC was performed using a slightly modified version of a previously described protocol.
Statistical analysis of in vitro and in vivo experiments
GraphPad Prism 9.0.1 was used for statistical analysis of in vitro experiments. For lipid accumulation and apo B in PHH two-way ANOVAs for siRNA and FFA mix condition with Tukey’s post-hoc testing within FFA mix condition was used. For PHH experiments performed in a single FFA mix condition a paired two-sided T-test was performed. Unpaired two-sided t-tests were used to evaluate all in vivo results. Significance is shown as: *P-value<0.033, **P-value <0.002, ***P value ≤0.001. No adjustment of P-values to account for multiple comparisons has been made.
Results
Metabolic trait associations with MTARC1 are due to the same genetic variants
Despite data showing associations at the MTARC1 locus with ALT, AST, ALP, plasma LDL, total cholesterol (TC), apo A and apo B,
, it is unclear whether these associations are due to the same shared or distinct genetic causal variants and mechanisms. To address this, the HyPrColoc
method was used (Supplementary Methods). The genetic associations of liver enzymes (ALT, AST, ALP, SHBG), circulating lipids (apo A, HDL cholesterol, apo B, LDL, cholesterol), liver fat accumulation, and body composition traits (whole body fat free mass, whole body water mass, trunk fat free mass, basal metabolic rate) identified at the MTARC1 locus were due to the same genetic variant i.e. colocalised with a posterior probability of 0.80 (Fig. 1A; Fig. S1). The missense variant (p.A165T, rs2642438 G>A) explained 100% of the posterior probability, suggesting it could be the causal variant (Fig. 1B). Thus, these data provide support for a shared genetic mechanism underlying the associations of these metabolic traits at the MTARC1 locus.
Figure 1Statistical colocalization of cardio-metabolic traits at the MTARC1 locus (A) Regional association plots at the MTARC1 gene locus, showing the shared genetic associations with liver enzymes (ALT & AST), lipids (total cholesterol) and liver fat accumulation. Details about the statistical analysis and source of the data are given in the Methods section. Color key indicates the correlation, r2, with the respective lead variants in the GWAS (B) Plot showing the colocalization posterior probabilities explained by each of the genetic variants at the extended MTARC1 locus region in the colocalization analysis. Table shows the parameters used for HyPrColoc analysis and results from HyPrColoc.
Rare and common exonic variants in MTARC1 associate with liver phenotypes
To systemically elucidate the impact of MTARC1 exonic variants on liver enzymes, single variant association analysis of 270 exonic MTARC1 variants with ALT, AST ALP, circulating lipids and body composition traits were performed. We replicated the association of the G allele of the common missense variant (p.A165T, rs2642438 G>A) with increased ALT, ASP and AST as well as with circulating lipids (Table S7). Further, we identified a novel association of a rare missense variant, rs12023067 A allele, with increased ALT (Table S7). Another rare missense variant, rs144056103, was suggestively associated with increased AST [P<5x10-5] (Table S7). To improve statistical power to identify rare variant associations, we performed gene-based aggregate tests of rare MTARC1 variants with liver enzymes. No significant (P<1x10-5) effects were obtained for ALP, ALT, AST, circulating lipids or body composition traits (Table S8).
MTARC1 gene expression associates with metabolic traits
Genetic variants associated with MTARC1 gene expression (i.e. expression quantitative trait loci (eQTLs)) were used to test whether genetically determined MTARC1 gene expression is associated with metabolic conditions. Publicly available eQTL data for MTARC1 in 114 datasets were used and up to three distinct eQTLs (distinct genetic variants, r2<0.1, associated with MTARC1 expression with P-value ≤5 × 10-8) were identified, in 16 unique tissues or cells including muscle, left ventricle-heart, stomach, transverse colon and rectum (Methods;Table S3A). Importantly, most of the identified instrumental variables (IV) are not in linkage disequilibrium with the common missense variant (p.A165T, rs2642438 G>A) and show no association with MTARC2 gene expression (Tables S3A&B). Analysis of these IVs showed that for every genetically-predicted standard deviation increase in MTARC1 gene expression levels in muscle, there is an associated increase in levels of ALT, AST, ALP, apo A, apo B, TC and liver fat accumulation, but lower levels of basal metabolic rate and whole body fat-free mass (Table 1, Fig. S2). Similarly, genetically-predicted increases in MTARC1 gene expression level in muscle were nominally associated with increased fibrosis and cirrhosis of liver (Table S9, Fig S2). The analyses of MTARC1 gene expression was performed in other tissues where eQTLs were available, and consistent results were obtained in left ventricle and rectum to those obtained in muscle (Table S9). These results were robust to sensitivity analyses with a range of different MR approaches (Methods). Although causal estimates from MR-Egger were less significant than those derived from the MR-IVW approach, we obtained consistent effect directions of MTARC1 gene expression on liver phenotypes (Table S9). Taken together, our analyses suggest that genetically determined increases in MTARC1 gene expression are associated with adverse effects on liver enzymes, liver fat and circulating lipids.
Table 1Mendelian randomization estimates for the effect of genetically determined gene expression level of MTARC1 on the metabolic traits. Effect estimates and P-values are provided for the inverse-variance weighting (IVW) method. P-values are shown from the Cochran’s Q-test for heterogeneity. Full details of the results from the different MR analyses, including details of data sources and number of cases, are reported in Tables S1, S2, S3a, S3b & S9.
Exposure
Outcome
MR causal estimate (IVW)
Heterogeneity
Beta (se)
pval
P-value
MTARC1 gene expression (Muscle)
Alkaline phosphatase
0.0841 (0.0061)
2.96E-43
0.8102
Alanine aminotransferase
0.0527 (0.0054)
2.08E-22
0.3087
Aspartate aminotransferase
0.036 (0.0057)
2.03E-10
0.0011
Direct bilirubin
0.0495 (0.0063)
6.37E-15
0.374
Total bilirubin
0.0778 (0.006)
6.48E-38
0.0285
SHBG
0.0306 (0.0057)
7.80E-08
0.2329
Apolipoprotein A
0.0387 (0.0059)
6.45E-11
0.9309
Apolipoprotein B
0.0445 (0.0059)
4.00E-14
0.1508
Cholesterol
0.0469 (0.0057)
1.80E-16
0.9365
LDL direct
0.045 (0.0057)
1.90E-15
0.9305
Urate
-0.0409 (0.0095)
1.55E-05
0.3518
Phosphate
0.0254 (0.0059)
1.41E-05
0.9591
Liver fat accumulation
0.0648 (0.018)
4.58E-04
0.0655
Whole body fat-free mass
-0.021 (0.0039)
6.11E-08
0.0266
Whole body water mass
-0.021 (0.0039)
5.84E-08
0.0294
Basal metabolic rate
-0.0193 (0.0041)
1.97E-06
0.0897
Trunk fat-free mass
-0.0209 (0.0039)
8.93E-08
0.0574
1MTARC1 gene expression in muscle is instrumented by 3 SNPs (cis-eQTL).
2Represents metabolic traits SD change per SD increase in MTARC1 gene expression in Muscle.
3Heterogeneity P-value was assessed using Cochran’s Q-test.
mARC1 knockdown decreases lipid accumulation and apo B secretion in PHH
To assess the impact of MTARC1 loss of function in hepatocytes an siRNA protocol was established to knockdown MTARC1 gene expression in PHH. PHH transfected with MTARC1 targeting siRNA (siMTARC1) maintained reduced MTARC1 mRNA levels compared to siNT for 17 days (Fig. 2A). Consistent with the 12 day predicted protein half-life of mARC1 in PHH,
, decreases in mARC1 protein were observed 10 days following siMTARC1 treatment but became more pronounced on day 17 (Fig. 2B, Fig. S3A). Protein and mRNA expression levels of mARC2 were unaffected by siMTARC1 (Fig. S3B, C). The effect of siMTARC1 and siRNA targeting DGAT2 (siDGAT2) on lipid accumulation was determined in the presence or absence of 800 μM FFA (Fig. 2C). The siRNAs effectively reduced expression (56-86%) of their target mRNA (DGAT2 and MTARC1) in 25 individual experiments utilising seven PHH donors (Table S10). Both siMTARC1 and siDGAT2 significantly reduced lipid accumulation in both conditions (Fig. 2D, E). Analysis of conditioned media from 16 experiments demonstrated that both siMTARC1 and siDGAT2 significantly decreased apo B secretion (Fig. 2F, Table S11). To assess the effect of mARC1 on lipid dynamics, siRNA-treated PHH were incubated with 13C-labelled glucose and fructose or 2H-labelled palmitate and oleate for the measurement of de novo lipogenesis (DNL) and fatty acid (FA) oxidation, respectively. mARC1 knockdown had no effect on DNL or FA oxidation (Fig. S4A, B). To determine if mARC1 knockdown alters the import and export of lipids, FFA and TG concentrations were quantified in conditioned media from the above experiments. The majority of the 800 μM FFA was taken up by the PHH and this was unaffected by siMTARC1 (Fig. S4C). In contrast, siMTARC1 significantly increased the concentration of TG in conditioned media (Fig. 2G).
Figure 2mARC1 knockdown decreases lipid accumulation and apo B secretion in primary human hepatocytes (PHH) (A) Validation of mRNA and (B) protein levels following MTARC1 siRNA knockdown in PHH over 17 days (C) Workflow for culturing of PHH, siRNA transfection and endpoint collections (D) Lipid accumulation data following siRNA mediated knockdown of MTARC1 and DGAT2 relative to non-targeting (NT) siRNA, n=25 independent experiments (E) SiRNA knockdown of MTARC1 and DGAT2 reduces lipid accumulation with and without FFA mix loading in PHH. Fluorescence images of nuclei (blue) and lipid droplets (hot) from siRNA knockdown PHH treated with FFA mix. Scale bar: 50 μm (F) Reduced apo B expression observed following knockdown with MTARC1 siRNA relative to NT siRNA, n=16 independent experiments (G) Increased media TG levels observed following knockdown with MTARC1 siRNA relative to NT siRNA, n=8 independent experiments. Data presented as mean + 95% CI, *P value<0.033, **P value<0.002, ***P value <0.001 (D & F: Two-way ANOVA with Tukey’s post-hoc testing within FFA mix conditions, G: Paired t-test). Fold Change (FC) Relative quantification (RQ).
Hepatocyte-specific mARC1 knockdown reverses steatosis and decreases markers of fibrosis in the gubra DIO-NASH mouse model
To determine the impact of hepatocyte-specific knockdown of Mtarc1 on hepatic steatosis, inflammation and fibrosis in vivo, a DIO-NASH mouse model was utilised. Briefly, C57BL/6JRj mice fed a high fat, high fructose, high cholesterol diet for 44 weeks were randomised based on biopsy results after 36 weeks and treated weekly with GalNAc-siMtarc1 or phosphate-buffered saline (PBS) for an additional 8 weeks. GalNAc-siMtarc1 effectively decreased Mtarc1 mRNA and slightly, but significantly, also reduced Mtarc2 mRNA (Fig. 3A, Fig. S5A). GalNAc-siMtarc1 decreased the liver to body weight ratio without altering body weight (Fig. 3B, Fig. S5B). GalNAc-siMtarc1 decreased histologically determined steatosis (Fig. 3C, Table 2). Also, intrahepatic concentrations of TG and TC determined by extraction were decreased by GalNAc-siMtarc1 treatment, while plasma TGs increased and plasma TC were reduced (Fig. 3D-G). Plasma liver enzymes (Fig. S5C-D) and metrics of hepatic inflammation were unchanged by mARC1 knockdown (Fig. S5E-H, Table 2). GalNAc-siMtarc1 did not alter total fibrosis, as assessed by sirius red staining, nor levels of collagen 1 but reduced hepatic α-SMA staining (Fig. 4A-C, Table 2). In addition, expression of several genes involved in fibrogenesis were downregulated by GalNAc-siMtarc1 (Fig. 4D). Mild hepatocyte ballooning was observed in the GalNAc-siMtarc1 treated group (Table 2) but as glycogen accumulation can confound the quantification of hepatocyte ballooning,
, measurement of hepatic glycogen was included in a second animal experiment. The majority of the effects of mARC1 knockdown on hepatic phenotypes were reproduced in the second study, including mild hepatocyte ballooning (Fig. S6A-F) and hepatic glycogen was significantly elevated in the GalNAc-siMtarc1 group (Fig. S6G).
Figure 3Hepatocyte-specific Mtarc1 knockdown reverses steatosis in a DIO-NASH mouse model.C57BL/6JRj mice fed a high fat, high fructose, high cholesterol diet for 44 weeks were randomized based on biopsy results and treated weekly with a GalNAc-conjugated siRNA targeting murine Mtarc1 (GalNAc-siMtarc1) or PBS for 8 weeks (A) mRNA expression for Mtarc1(B) % Liver weight of body weight (BW) (C) Liver triglycerides (TG) as mg/g of liver (D) Liver total cholesterol (TC) as mg/g of liver (E) Plasma TG (mmol/L) (F) Plasma TC (mmol/L). Data is presented as mean + 95% CI n=10-14, *P value<0.033, ***P value <0.001 (unpaired T-test).
Figure 4Hepatocyte-specific Mtarc1 knockdown decreases markers of fibrosis in a DIO-NASH mouse model. C57BL/6JRj mice fed a high fat, high fructose, high cholesterol diet for 44 weeks were randomized based on biopsy results and treated weekly with a GalNAc-conjugated siRNA targeting murine Mtarc1 (GalNAc-siMtarc1) or PBS for 8 weeks (A) Fibrosis as PSR, % area fraction (B) Col1 as % area fraction (C) a-SMA as % area fraction (D) mRNA expression for fibrosis associated genes; Timp1, TgfB, Col1a1, Mmp2 and Mmp9. Data is presented as mean + 95% CI n=10-14, non-significant (ns), *P value<0.033, **P value<0.002, ***P value <0.001 (unpaired T-test).
mARC1 knockdown results in transcriptional changes related to lipid metabolism
To determine the global transcriptional impact of MTARC1 loss of function, RNA-sequencing on PHH treated with siMTARC1 or siNT (Methods) was performed. In addition to MTARC1, siMTARC1 altered 17 mRNAs (Fig. 5A, Table S12) enriched in processes and pathways including response to lipid and lipid transport (Table S13). Gene Set Enrichment Analysis (GSEA) identified alterations in lipid and FA metabolic processes and the peroxisome proliferator-activated receptor (PPAR) signalling pathway (Fig. S7A&B, Table S14). RNA-sequencing of the liver of the GalNAc-siMtarc1 treated DIO-NASH mice revealed over 250 differentially expressed protein coding genes (Methods, Table S15) compared to vehicle treated mice. Although there was no overlap with regulated mRNAs in PHH (except Mtarc1), 11 of the 13 mouse homologues of genes dysregulated in the PHH trended in the same direction; significantly more than expected at random (P-value (binomial) = 0.022; Fig. 5A). Also, in agreement with the human transcriptome profiling, effects of GalNAc-siMtarc1 on genes involved in the PPAR signalling and lipid metabolism were observed (Table S16-17).
Figure 5Effect of MTARC1 siRNA in Primary Human Hepatocytes (PHH) and mouse evaluated by omics approaches (A) RNA-seq data showing fold change of significantly dysregulated mRNAs and (B) metabolomic data showing significantly dysregulated metabolites in PHH (red) and mouse tissue from bulk liver prep (blue). FDR≥0.05 (empty points) or FDR<0.05 (solid points). 3-Hydroxybutyrate (blue star) was significantly reduced in both PHH and mouse samples (C) Protein-interaction network analysis of significant metabolites revealed significant enrichment in the Kennedy pathway due primarily to dysregulation of phosphocholine, phosphoethanolomine, and serine (red star).
mARC1 regulates the kennedy pathway and 3-hydroxybutyrate
To gain insight into the metabolic impact of mARC1 knockdown, metabolomic analysis on conditioned media from PHH treated with either siMTARC1 or siNT was performed. Knockdown of mARC1 affected the abundance of 47 metabolites including increases in O-phosphoethanolamine, phosphocholine, and a decrease in 3-hydroxybutyrate (Fig. 5B; Table S18). A network analysis of putative interacting proteins of the metabolites significantly altered by siMTARC1 identified enrichments in the Kennedy pathway from sphingolipids (WP3933, False Discovery Rate (FDR)=0.01) and One-carbon metabolism and related pathways (WP3940, FDR=0.04) (Fig. 5C). To test if hepatocyte-specific Mtarc1 was sufficient to impact metabolite concentration in the circulatory system of a whole organism, metabolites in plasma from mice treated with GalNAc-siMtarc1 were profiled. Thirteen of the 47 MTARC1 affected metabolites identified in our PHH samples were also identified in mice (Methods, Table S19). Of these, only 3-hydroxybutyrate, was altered in GalNAc-siMtarc1 animals (Fig. 5B, Table S20).
Discussion
Using a combination of genetic approaches and experimental manipulations a causal relationship between mARC1 expression and liver enzymes, plasma lipids and liver fat has been demonstrated and reductions in hepatocyte mARC1 expression decrease hepatic lipid content. Previous genetic findings have been extended
by establishing novel associations of rare MTARC1 variants with liver enzymes, a shared aetiology of traits at the MTARC1 locus and a relationship between increased MTARC1 mRNA and worsening of liver enzymes, apo B and body composition. mARC1 knockdown in PHH decreases lipid content and apo B secretion. In a murine model of NASH, hepatocyte-specific mARC1 knockdown improved hepatic steatosis and decreased markers of fibrosis. Thus, while genetic variation in MTARC1 is causally associated with a range of metabolic traits, hepatocyte-specific decreases in MTARC1 mRNA are sufficient to recapitulate alterations in hepatic and serum lipids.
A common MTARC1 missense variant (p.A165T, rs2642438 G>A) has been associated with all-cause cirrhosis, NAFLD, liver enzymes and serum lipids.
. Here, HyPrColoc was applied to show that rs2642438 is the likely causal variant across these traits demonstrating a shared genetic mechanism. In addition to the common missense variant (p.A165T, rs2642438 G>A), we identified novel missense MTARC1 variants associated with liver enzymes. However, as the impact of missense variants in MTARC1 on protein levels, localisation and activity is unclear, the directional effect of MTARC1 can only be inferred from the rare loss-of-function variant encoding a premature stop codon (p.R200Ter, rs139321832:C:T).
Letter to the editor: The clinically relevant MTARC1 p.Ala165Thr variant impacts neither the fold nor active site architecture of the human mARC1 protein.
. Using MR, we demonstrated that increased MTARC1 mRNA was associated with adverse effects on metabolic traits. Although the tissues with available eQTLs (muscle, left heart ventricle and rectum) may not contribute to the associated phenotypes, the concordant effect direction indicates a consistent impact of the gene variants on gene expression across tissues. Our work reports a novel association of the MTARC1 locus with body composition traits suggesting an additional metabolic benefit of decreased mARC1 expression and underscores how genetic variants can have diverse systemic effects likely driven by multiple organs.
The cell type(s) that mediate the effect of genetic variation in MTARC1 on the various metabolic traits has important therapeutic implications. Beneficial effects of hepatocyte-specific MTARC1 inhibition could be therapeutically tractable using GalNAc-siRNA technology. While PHH are the gold standard model for metabolic modelling, their long-term utility is limited by rapid dedifferentiation in vitro. Utilizing the recently described 5C protocol for long-term PHH culture
we have shown that siRNA knockdown is effective and persistent in this model and lipid accumulation can be stimulated by exposure to a FFA mix and inhibited by knocking down DGAT2, which catalyses the last step of TG synthesis. SiMTARC1 decreased lipid accumulation in both the absence and presence of the FFA mixture. The decreased lipid accumulation was not associated with alterations in FA uptake, as assessed by the remaining concentration observed in conditioned media, or oxidation of FA or DNL from glucose. GalNAc-siMtarc1 treatment of Gubra DIO-NASH mice decreased hepatic TG content. PBS was used as a vehicle control rather than a GalNAc-NT siRNA due to difficulties in assessing proper internalisation without an mRNA knockdown readout and the potential of off-target effects that may confound results. While a contribution of the GalNAc moiety to the effects of GalNAc-siMtarc1 treatment cannot be excluded, this concern is mitigated by concordant effects of siMTARC1 across model systems with independent siRNA delivery methods. Knockdown of hepatocyte mARC1 decreased neutral lipid content in PHH and mice suggesting that hepatocyte MTARC1 mediates the genetic associations with NAFLD. This has major therapeutic implications given the clinical utility of GalNAc conjugation to effectively deliver RNAi-based therapies to hepatocytes.
Hepatocytes secrete significant amounts of TGs within very low-density lipoproteins (VLDL). As the anti-steatotic effect observed in carriers of the A allele of rs2642438 is accompanied by elevated plasma TG, increased VLDL secretion has been proposed as a potential mechanism by which MTARC1 loss of function may reduce hepatic lipid content.
. Our findings that hepatocyte mARC1 knockdown increased supernatant and plasma TG in PHH and mice, respectively, support this hypothesis. Furthermore, the decreased secretion of apo B in PHH treated with siMTARC1parallels decreased apo B in carriers of the A allele of rs2642438.
. This suggests that loss of hepatocyte mARC1 function may decrease the number, but increase the size, of secreted VLDL particles leading to reduced apo B and increased plasma TG, respectively. Transcriptomic analysis demonstrated a decreased expression of genes involved in FA metabolic processes and PPAR signalling upon mARC1 knockdown in vitro and in vivo. These findings are consistent with the decreased concentration of 3-hydroxybutyrate in cell culture supernatant and mouse plasma following mARC1 knockdown. Observed alterations in the Kennedy pathway in siMTARC1 treated PHH are consistent with the observation that carriers of the A allele of rs2642438 have increased phosphatidylcholine in liver biopsies, which is essential for the secretion of VLDL.
. Together, these findings suggest that MTARC1 inhibition influences TG secretion, rather than lipid catabolism, to reduce hepatic TG. The molecular mechanisms by which mARC1 exerts its effects on TG metabolism, and whether changes in 3-hydroxybutyrate and the Kennedy pathway are causative, require further study.
While human genetics links MTARC1 to more advanced forms of NAFLD,
, hepatocyte-specific knockdown of mARC1 did not improve previously established fibrosis in a DIO-NASH model. This may be due to a contribution of non-hepatocyte mARC1 to the genetic signal, species differences or temporal considerations. The anti-steatotic effect of mARC1 inhibition may be sufficient to prevent, but not reverse, the development of fibrosis in agreement with a Mtarc1-induced reduction of α-SMA. Alternatively, a longer treatment period may be necessary to impact fibrosis in the DIO-NASH model. Future studies will be necessary to determine if mARC1 alters fibrosis, if such an effect is mediated via steatosis or direct mechanisms and when in the disease progression therapeutic intervention may be effective.
In summary, genetically predicted MTARC1 gene expression is associated with multiple metabolic traits and hepatocyte mARC1 likely mediates the association with hepatic and plasma lipids. Knockdown of mARC1 in PHH and mice decreases intrahepatocellular lipid content and increases TG secretion. This work supports the therapeutic utility of hepatocyte-specific targeting of mARC1 and further studies to elucidate the mechanisms by which mARC1 impacts NAFLD.
Acknowledgments
The authors would like to thank T. Monfeuga, L. Payne, H.D. Radic, H. Lykkegaard, M.B. Larson and M. Grønborg for their excellent technical support. P. S. Petersen and M. Feigh at Gubra Aps are acknowledged for study management and development of the biopsy-confirmed DIO NASH mouse model. M. Abrams, C. Rijnbrand, and M. McGrath for sourcing the work carried out by Dicerna Pharmaceuticals Inc. We gratefully acknowledge the UK Biobank and all the participants in UK Biobank for providing data application #53639. The graphical abstract was created with BioRender.com.
Appendix A. Supplementary data
The following is/are the supplementary data to this article:
Vujkovic M, Ramdas S, Lorenz KM, Guo X, Darlay R, Cordell HJ, et al. A genome-wide association study for nonalcoholic fatty liver disease identifies novel genetic loci and trait-relevant candidate genes in the Million Veteran Program. 2021.
The N-reductive system composed of mitochondrial amidoxime reducing component (mARC), cytochrome b5 (CYB5B) and cytochrome b5 reductase (CYB5R) is regulated by fasting and high fat diet in mice.
High Resolution Mass Spectrometry Improves Data Quantity and Quality as Compared to Unit Mass Resolution Mass Spectrometry in High-Throughput Profiling Metabolomics.
Journal of Postgenomics Drug & Biomarker Development.2014; 4
Letter to the editor: The clinically relevant MTARC1 p.Ala165Thr variant impacts neither the fold nor active site architecture of the human mARC1 protein.
00024-1&id=ga1.jpg){kind=link}
00024-1&id=gr1.jpg){kind=link}
00024-1&id=gr2.jpg){kind=link}
00024-1&id=gr3.jpg){kind=link}
00024-1&id=gr4.jpg){kind=link}
00024-1&id=gr5.jpg){kind=link}