Hepatic miR-149-5p upregulation fosters steatosis, inflammation and fibrosis development in mice and in human liver organoids

Background & Aims The incidence of metabolic dysfunction-associated steatotic liver disease (MASLD) is increasing worldwide. Alterations of hepatic microRNA (miRNA) expression/activity significantly contribute to the development and progression of MASLD. Genetic polymorphisms of miR-149 are associated with an increased susceptibility to MASLD development in humans. Aberrant expression of miR-149 was also associated with metabolic alterations in several organs, but the impact of hepatic miR-149-5p deregulation in MASLD remains poorly characterized. Methods MiR-149-5p was downregulated in the livers of mice by in vivo transduction with hepatotropic adeno-associated virus 8 harboring short-hairpin RNAs (shRNAs) specific for miR-149-5p (shmiR149) or scrambled shRNAs (shCTL). MASLD was then induced with a methionine/choline-deficient (MCD, n = 7 per group) diet or a fructose/palmitate/cholesterol-enriched (FPC, n = 8-12 per group, per protocol) diet. The impact of miR-149-5p modulation on MASLD development was assessed in vivo and in vitro using multi-lineage 3D human liver organoids (HLOs) and Huh7 cells. Results MiR-149-5p expression was strongly upregulated in mouse livers from different models of MASLD (2-4-fold increase in ob/ob, db/db mice, high-fat and FPC-fed mice). In vivo downregulation of miR-149-5p led to an amelioration of diet-induced hepatic steatosis, inflammation/fibrosis, and to increased whole-body fatty acid consumption. In HLOs, miR-149-5p overexpression promoted lipid accumulation, inflammation and fibrosis. In vitro analyses of human Huh7 cells overexpressing miR-149-5p indicated that glycolysis and intracellular lipid accumulation was promoted, while mitochondrial respiration was impaired. Translatomic analyses highlighted deregulation of multiple potential miR-149-5p targets in hepatocytes involved in MASLD development. Conclusions MiR-149-5p upregulation contributes to MASLD development by affecting multiple metabolic/inflammatory/fibrotic pathways in hepatocytes. Our results further demonstrate that HLOs are a relevant 3D in vitro model to investigate hepatic steatosis and inflammation/fibrosis development. Impact and implications: Our research shows compelling evidence that miR-149-5p plays a pivotal role in the development and progression of MASLD. By employing in vivo and innovative in vitro models using multi-lineage human liver organoids, we demonstrate that miR-149-5p upregulation significantly impacts hepatocyte energy metabolism, exacerbating hepatic steatosis and inflammation/fibrosis by modulating a wide network of target genes. These findings not only shed light on the intricate miR-149-5p-dependent molecular mechanisms underlying MASLD, but also underscore the importance of human liver organoids as valuable 3D in vitro models for studying the disease's pathogenesis.


Animal housing
Mice were adapted to the animal facility of University of Geneva for three weeks and kept in ventilated cages with 2 to 5 animals per cage.During the period of adaptation, mice had access to standard chow diet (SAFE-150 diet, SAFE, Augy, France) and water ad libitum.Animals were maintained in cages with appropriate enrichment (disposable house and nesting material) and with a 12h light/dark cycle at 23°C.

Diets and experimental procedures
Mice fed with a high sugar/high fat diet (HFD) Two month old C57BL/6J mice (Charles Rivers Laboratory) were submitted to four different isocaloric high sugar/high-fat diet (HFD: 45% kcal from fat, 17% kcal from sucrose) or a matched Control Diet (CD: 10% kcal from fat, 17% kcal from sucrose) for 16 weeks (n=5 per group).The different HFD were a Western Diet (WD: made with lard), an omega-3 enriched HFD (O-3D -similar to the WD but with 25% of the total fat mass replaced by omega-3 fish fatty acids) and a trans-hydrogenated fatty acid enriched HFD (THD -similar to the WD but with 23%-26% of the total fat mass replaced by trans-hydrogenated monounsaturated fatty acids).Detailed information on the different HFD diets is described on Table S1.

Mice fed with a methionine/choline-deficient diet (MCD)
Ten weeks old mice injected with AAV8-shCTL or AAV8-shmiR-149 (n=7 per group) were allowed to recover for 10 days and then were fed with a methionine/choline deficient diet (MCD; E15653-94, ssniff, Germany) for 19 days.During this feeding period, mice were weighted every two days.At the end of the experiment, mice were decapitated following isoflurane anesthesia and liver and blood samples collected for further analyses.Detailed information on the MCD diets is described on Table S1.

Mice fed with a fructose/palmitate/cholesterol/trans-fat-enriched diet (FPC diet)
Ten weeks old mice injected with AAV8-shCTL or AAV8-shmiR-149 were allowed to recover for 10 days and then were fed with a Fructose/Palmitate/Cholesterol/Trans-Fat-enriched diet (FPC, TD.19142, Envigo, USA) for 10 weeks (short FPC, n=8-12 per group) or 24 weeks (Long FPC, n=10-12 per group).For the 24 weeks protocol, mice received a second injection of adenoviruses (1x10 11 GC/mouse) at 8 weeks of diet in order to ensure hepatic knockdown of miR-149.The FPC diet was previously reported to induce liver steatosis after 8-10 weeks and liver steatosis, fibrosis and inflammation after 16-24 weeks (1).During the feeding periods, mice were weighted each week and glycemia was measured at 9 a.m. using a Glucometer (AccuCheck -Roche) on blood collected from the tail vein after 10 (n=16-18 per group) and 23 (n=9-11 per group) weeks of diet.Glucose and pyruvate tolerance tests were performed on mice at 7/14 weeks of diet and 8/22 weeks of diet, respectively.The same blood samples (fasted: n=8-12 per group at 7 weeks; fed: n=7-10 per group at 10 weeks, n=8-9 per group at 22 weeks) were used to measure insulinemia by ELISA (Mercodia Ultrasensitive Mouse Insulin ELISA -10-1249-01).Detailed information on the FPC diets is described on Table S1.

Glucose and Pyruvate Tolerance Tests
For the glucose tolerance test, mice were fasted 6 hours prior the intraperitoneal injection of 2 g/kg of glucose (n=10-12 per group at 7 weeks, n=9-11 per group at 14 weeks).For the pyruvate tolerance test, mice were fasted 18 hours prior the intraperitoneal injection of 2 g/kg of pyruvate (Sigma)(n=8-11 per group at 8 weeks, n=9-11 per group at 22 weeks).Blood glucose levels were measured at 0, 15, 30, 60, 90 and 120 minutes post-injection of glucose/pyruvate using a Glucometer (AccuCheck -Roche) in blood samples collected from the tail vein.

Metabolic cages and EchoMRI
Metabolic phenotyping of FPC fed mice (23 weeks, n=6-7 per group) was performed for 7 days using metabolic cages (LabMaster) after 2 days of adaptation prior to recording calorimetric parameters (O2 consumption, CO2 production, respiratory exchange ratio and energy expenditure), food and water intake and locomotor activity.During this procedure, mice were isolated in metabolic cages.Body composition was measured using a positron emission tomographic whole-body composition analyzer (EchoMRI-700, Houston, Texas, USA).

Plasma analyses
Blood recovered during sacrifice following decapitation was centrifuged at 5000 rpm (Centrifuge) for 10 minutes.Plasma was collected and glucose, aspartate/alaninaminotransferases (ASAT/ALAT), cholesterol and triglycerides levels were analyzed using Cobas 8000 system (Roche, Switzerland).

Isolation of primary hepatocyte
Primary mouse hepatocytes were isolated from LPTENKO mice and control littermates as previously described (2, 3).
At day 17, Matrigel drops were disrupted with gentle pipetting to release the HLOs and they were kept in suspension in complete HCM medium supplemented with 10% Matrigel (Gibco) until the end of the experiment (day 21, day 24 or day 28).

Ethical authorization to use ESC HS420 cells was provided by the Geneva Health Head
Office (authorization number R-FP-S-2-0028) and performed following the Swiss guidelines on Research involving embryonic stem cells.

Synthetic Oligonucleotide transfections
Huh7 cells were transfected 24 hours after seeding using Interferin (Polyplus transfection, Illkirch, France) and Optimem (Gibco), following manufacturer's instructions.HLOs were transfected using Lipofectamine (ThermoFisher Scientific) and Optimem (Gibco), following manufacturer's instructions.HLOs were kept in DMEM (1g/L glucose, Gibco) supplemented with 1% PS for 24 hours and then the medium was changed to HCM supplemented as described in 2.2.2. until the end of the experiment.Huh7 cells and HLOs were transfected with miRIDIAN microRNA miR-149-5p mimic (Mimic 149, Horizon Discovery, UK) or miRIDIAN microRNA Mimic Negative Control #1 coupled to a fluorophore or not (CTRL Mimic/ CTRL Mimic FL, Horizon Discovery, UK) at concentrations of 10 nM and 25 nM for Huh7 cells and HLOs respectively.

SeaHorse analyses
Twenty-four hours after transfection of miRIDIAN microRNAs, Huh7 cells were reseeded in a 96-well Seahorse Agilent Plate at 25'000 cells per well.Twenty-four hours post-reseeding, different Seahorse XF metabolic assays were performed in a Seahorse XFe96 Analyzer according to the manufacturer's recommendations -MitoStress, GlycoRate and Substrate Oxidation Test.At the end of each assay, cells were fixed with 4% paraformaldehyde for 15 minutes at room temperature, stained with Hoechst (1 µg/mL, 33342, ThermoFisher Scientific) for 10 minutes and scanned on Cytation 5.
Following image acquisition, cell number was counted using the Gen5 software (BioTek) and used to normalize Seahorse assays results using Wave 2.4.0.software (Agilent Technologies).

Substrate Oxidation Stress Test
Seahorse XF Substrate Oxidation Stress (Long Chain Fatty Acid 103672-100 kit) was performed as described by the manufacturer, using the following drug concentrations: Etomoxir (E) -4 µM, Oligo -1.5 µM, FCCP -2 µM and Rot/AA -0.5 µM.Conditions were etomoxir was not injected are identified as medium only (M).

Insulin stimulation
Cells were seeded and transfected as described in 1.2.2 and in 1.2.4.Cells were grown until 70% confluence, starved (FBS-free medium) for 6 hours and stimulated with insulin (Mixtard, 10 -7 M) for 15 minutes or not.Following stimulation, cells were flash frozen using liquid nitrogen.

Morphological assessment of lipid droplets accumulation in hepatocytes (steatosis)
Huh7 cells and HLOs were cultured and transfected as described in 1.2.2 -1.2.4.

Establishment of inflammation/ fibrosis in human liver organoids
HLOs were cultured as described in 1.2.2.At day 21 HLOs were exposed to a mixture of cytokines (TNFα, TGFβ, IL-6 and IL-1β, 10ng/mL each) for 3 days or to FA-enriched medium for 7 days (as described in 1.2.8.).Following incubation with cytokines/FAs, HLOs were collected and expression of inflammatory/fibrotic markers was assessed through RT-qPCR.

Determination of the cellular mitochondrial mass
Huh7 cells were cultured and transfected as described in 1.2.2 and in 1.2.4.To observe mitochondrial morphology and quantify mitochondrial mass, cells were stained at 72-hours post-transfection with MitoTracker™ Red CMXRos (200 nM, M7512, ThermoScientific) for 20 minutes at 37°C.Subsequently, cells were fixed with 4% paraformaldehyde for 15 minutes at room temperature and counter-stained with Hoechst (1µg/mL, 33342, ThermoFisher Scientific) for 10 minutes.Coverslips were mounted using anti-fading agent (DAKO, S3023, Agilent) and imaged with Axiocam Fluo (Zeiss).In parallel, mitochondrial DNA was purified using the QiAmp ® DNA Micro kit (56304, Qiagen).Then, amplification of nuclear and mitochondrial DNA was performed and the mitochondrial/nuclear DNA ratio was calculated as previously described (5).

Polysome fractionation
Huh7 cells were cultured and transfected as described in 1. Fractionated ribosomes were monitored and collected using Density Gradient Fractionation System (ISCO).

RNA extraction and Real-time qPCR
RNA extraction from ribosomal fractions, flash frozen cells, HLOs and mouse tissues was performed using Trizol (Ambion, Thermo Scientific, USA) according to manufacter's instructions.RNA concentration was measured using NanoDrop (Thermo Scientific).Prior to real-time qPCR, reverse transcription was performed using the High-Capacity cDNA Reverse Transcription kit (Applied Biosystems™).For miRNAs expression, the reverse transcription was performed as previously described (6).qPCR was performed using the PowerUp™ SYBR™ Green Master Mix for Real-Time PCR and the QuantStudio 5 Real-time PCR System and data analysis software (Applied Biosystems™), according to manufacturer's specifications.Primer sequences used are described in Table S2.Gene expression was quantified using the ΔΔCT method.

Western Blot
Protein extraction from flash frozen cells and mouse tissues was performed using RIPA buffer (50 mM Tris-HCl, pH 6.8, 100 mM DTT, 2%SDS, 0.1% bromophenol blue, 10% glycerol).Tissues were homogenized using TissueLyser.Protein lysates were centrifuged at 12000 g for 10 minutes and the supernatant collected.Protein concentration was determined using BCA protein assay kit (Pierce Biotechnology).5-10 µg of protein was charged in 5-20% gradient sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels and transferred to nitrocellulose membranes (RPN303D, Amersham, Switzerland).Membranes were blocked with polyvinyl alcohol and incubated overnight with primary antibody at 4°C.Membranes were washed with 0.1% TBS-Tween and incubated with secondary antibodies at room temperature for 1 hour.Membranes were then incubated with ECL Prime Substrate (RPN22232, Amersham, Switzerland) for 1 minute and revealed using the PXI/PXI Touch system (Syngene, Synoptics group, UK) or Fusion instrument (Vilber, France).Between each step, membranes were washed with 0.1% TBS-Tween.Signal was quantified using GeneSys (Syngene, Synoptics group, UK) or ImageJ™ software.Antibodies used are described in Table S2.

Histology, Immunohistochemistry and immunofluorescence
Mouse tissues were fixed overnight with 4% paraformaldehyde (PFA) while HLOs were fixed for 2 hours in 4% PFA and then transferred to PBS for dehydration and paraffinembedding.Tissue and HLOs specimens were cut into 5 µm sections prior to staining or immunohistochemistry/immunofluorescence.For morphological analysis, tissue sections from 3 samples collected from different lobes of the explanted liver were stained with hematoxylin and eosin while for fibrosis analysis, sections were stained with Sirius Red.For immunohistochemistry and immunofluorescence, tissue/HLOs sections were deparaffinized, rehydrated and heated in citrate buffer or treated with Proteinase K for antigen retrieval.Subsequently, sections were permeabilized with 0.3% Triton X-100 in TBS for 15 minutes, blocked with 10% goat serum (ab138478, Abcam) in 1% BSA-TBS for 2h at room temperature and incubated overnight at 4°C with primary antibody diluted in 1% BSA-TBS.Endogenous peroxidase was blocked by incubation with 0.3% H2O2 solution for 15 minutes at room temperature, prior to incubation with secondary antibody at room temperature for 1 hour.Between each step, slides were washed in 0.025% Triton X-100 in TBS.Signal was revealed following an incubation with DAB Substrate kit (ab64238, Abcam, UK) or with secondary antibodies conjugated with a fluorophore and counterstained with hematoxylin for 3 minutes or with Hoechst (1 µg/mL, 33342, ThermoFisher Scientific) for 10 minutes.
Following staining or immunohistochemistry/immunofluorescence, slides were mounted and scanned in AxioScan Z.1 (Zeiss) or imaged in Axiocam Fluo (Zeiss) for subsequent analysis.The slides from the different experimental groups were then analyzed blindly by two different researchers.Antibodies and concentrations used are listed in Table S2.

Image analysis
Batched analysis and quantification of lipid droplet number/size per µm 2 of tissue were For HLO image analysis, 3D confocal stacks, acquired with multiple wavelengths, were automatically processed using a dedicated framework developed in Matlab R2023a (The MathWorks).Specifically, Nikon nd2 files were accessed through the Bio-formats package (7).Organoid localization was achieved by the maximum voxel-wise of all available channels following intensity normalization.Subsequently, alternating sequential filters and Otsu's method (8) for global image segmentation were employed.
Nuclei, labeled with DAPI and/or HNF4, and lipid droplets were segmented following anisotropic diffusion filtering using Cellpose 2.0 (6) in 3D, with its pre-trained models for "nuclei" or "cyto".The resulting two sets of nuclei were combined to identify hepatocyte nuclei.Lipid droplets were assigned to specific cell nuclei based on the shortest Euclidean distance centroid to centroid.Finally, the results were quantified in terms of organoid volume, hepatocyte cell-proportions in the organoids, total lipid droplet volume, and mean fluorescent intensities in HNF4-positive or -negative nuclei.

Glycogen, glucose and glucose-6-phosphate content assay
Frozen liver tissues (100 mg) from mice fed FPC diet for 24 weeks (n=9-10 per group) were lysed in 8 volumes of perchloric acid (6%).Lysates were centrifuged at 10'000 g for 15 min at 4°C.Supernatants were collected and neutralized with potassium carbonate (K2CO3, 3.2mM) until pH reached 6.5-8.5.Following another centrifugation at 10 000 g for 15 min at 4°C, glycogen content was measured with Keppler and Decker method, as previously described (7).Glycogen was partially hydrolyzed in NaOH (0.15 M) for 20 minutes at 100ºC and digested by α-amidoglucosidase for 1h at 45°C into glucose.Glucose was measured after the addition of hexokinase (0.7U/mL) and glucose-6-phosphate was measured after the addition of NADP + (0.9 mM) and glucose-6-phosphate dehydrogenase (0.7U/mL).NADPH production was detected at 340 nm.

Triglyceride assay
Intra-hepatic triglycerides were measured using the Triglyceride-Glo™ assay (Promega) according to manufacturer's instructions.The results obtained for the FPC diet were further validated using the <Folch= method as described in (9).

Microarray analysis
Livers from 4 months old LPTENKO (Pten lox/lox , AlbCre +/-) mice and wild-type littermates (Pten lox/lox , AlbCre -/-; n=3 per group) were used for miRNAs microarray analysis.RNA was extracted as mentioned in section 1.4.and 500 ng of total RNA was used for subsequent analyses.Microarray miRNA expression profiles of were performed in the Genomics platform at University of Geneva.miRNA expression profiles were obtained using the Affymetrix GeneChip ® miRNA 3.0 Array (Affymetrix).After quality control, data was normalized and summarized using the robust multichip analysis (Affymetrix Microarray Suite).Partek was used to determine ANOVA p-values and fold-changes.

Statistical analysis
For animal experimentation we have performed power analysis (G*Power software, v.
performed using QuPath software (version 0.4.3.) and CellPose package (version 2.0)(6) following pretrained automated tissue/lipid droplet detection.Sirius red positive areas per µm 2 of tissue were detected using QuPath's pixel classifier default settings following pretrained automated tissue detection.Relative BODIPY and MitoTracker™ Red CMXRos signal in Huh7 cells was measured with CellProfiler v4.2.1.and normalized to cell number.Single lipid droplets morphological aspects in Huh7 cells were also analyzed with CellProfiler (number, area, diameter).

Fig. S12 -
Fig. S12 -Overexpression of miR-149-5p in the hepatic cell line Huh7 leads to deregulation of genes involved in metabolic and inflammatory pathways.Gene ontology enrichment analysis of KEGG pathways with (A) downregulated and (B) upregulated genes identified in transcriptomic analyses of polysomal fractions from Huh7 cells transfected with synthetic oligonucleotides mimicking miR-149-5p or with scrambled mimics.Significantly deregulated genes were identified using the following thresholds -fold-change to control mimic = |1.5|and false discovery rate (FDR) < 0.05.

table 1 -
1 (t-test), effect size 1.1, power: 0.9) to calculate the minimum number of animals necessary to observe an expected effect of 40% reduction in regards to steatosis between shCTL and shmiR149 mice.This analysis allowed to determine a Composition of the different diets used.
minimum of 12 mice per group.Statistical analyses were performed using GraphPad Prism 8 Software (GraphPad Software, San Diego, CA, USA).Results are represented as mean ± standard deviation (SD).Outliers test was performed using the ROUT method (Q = 1%).Unpaired t-test with Welch's correction was performed to compare two groups.One-way ANOVA test with Holm-Sidak correction was applied to compare more than two groups.To evaluate the independence of categorical variables, chisquare test or Fisher's test was used.Results of statistical tests are represented in figure legends as follows: * p-value < 0.05, ** p-value < 0.01, *** p-value < 0.001, **** p-value < 0.0001.Supplementary methods

table 2 -
List of primers used and respective sequences.

table 3 -
List of antibodies used and respective application.

Table S1 -Literature screening performed using the 237 potential targets of miR-149 identified with translatomic analyses. Only studies related with MASLD/MASH or metabolic deregulations were considered.
with insulin resistance and body mass index in humans.Additionally, Inhbe gene expression increased in the livers of db/db mice.Downregulation of INHBE suppressed body weight gain due to decreased fat rather than lean mass.It also decreased the respiratory quotient and increased plasma total ketone bodies, suggesting enhanced whole-body fat utilization.

Table S2 -
List of studies reporting association between different pathologies and miR-149 single In bold it is highlighted reports where variants were linked with nonalcoholic fatty liver disease and hepatocellular carcinoma.