Sumoylation of methionine adenosyltransferase alpha 1 promotes mitochondrial dysfunction in alcohol-associated liver disease

Abstract

Background and aims: Methionine adenosyltransferase alpha1 (MATα1) is responsible for the biosynthesis of S-adenosylmethionine in normal liver. Alcohol consumption enhances MATα1 interaction with peptidyl-prolyl cis-trans isomerase NIMA-interacting 1 (PIN1), which blocks MATα1 mitochondrial targeting, resulting in lower mitochondrial MATα1 content and mitochondrial dysfunction in alcohol-associated liver disease (ALD) in part through upregulation of cytochrome P450 2E1. Conversely, alcohol intake enhances SUMOylation, which enhances cytochrome P450 2E1 expression. MATα1 has potential SUMOylation sites, but whether MATα1 is regulated by SUMOylation in ALD is unknown. Here, we investigated if MATα1 is regulated by SUMOylation and, if so, how it impacts mitochondrial function in ALD.

Approach and results: Proteomics profiling revealed hyper-SUMOylation of MATα1, and prediction software identified lysine 48 (K48) as the potential SUMOylation site in mice (K47 in humans). Experiments with primary hepatocytes, mouse, and human livers revealed that SUMOylation of MAT1α by SUMO2 depleted mitochondrial MATα1. Furthermore, mutation of MATα1 K48 prevented ethanol-induced mitochondrial membrane depolarization, MATα1 depletion, and triglyceride accumulation. Additionally, CRISPR/CRISPR associated protein 9 gene editing of MATα1 at K48 hindered ethanol-induced MATα1-PIN1 interaction, degradation, and phosphorylation of MATα1 in vitro. In vivo, CRISPR/CRISPR associated protein 9 MATα1 K48 gene-edited mice were protected from ethanol-induced fat accumulation, liver injury, MATα1-PIN1 interaction, mitochondrial MATα1 depletion, mitochondrial dysfunction, and low S-adenosylmethionine levels.

Conclusions: Taken together, our findings demonstrate an essential role for SUMOylation of MATα1 K48 for interaction with PIN1 in ALD. Preventing MATα1 K48 SUMOylation may represent a potential treatment strategy for ALD.

FIGURE 1.. Alcohol induces MATα1 SUMOylation in human and mouse liver.

(A) NIAAA pair-fed and ethanol-fed mouse livers were used to enrich SUMOylated proteins using VIVAbind SUMO columns as described in methods. Intensity of the extracted precursor isotopic envelope (M, M + 1, M + 2) of a representative MATα1 peptide YLDEDTVYHLQPSGR (SUMOylated proteins) and FVIGGPQGDAGVTGR (Un-SUMOylated proteins). All matched the theoretical isotopic distribution. (B) IP analysis of NIAAA liver lysates using anti-SUMO1 or anti-SUMO2/3 antibody followed by western blot analysis against MATα1. Values represent mean ± SEM from control (n=4) and ethanol-fed (n=4) mouse liver. *p<0.05 and **p<0.04 vs. control. (C) NIAAA liver paraffin embedded sections and (D) human tissue arrays were stained with MATα1 and SUMO2/3 followed by secondary antibody-PLA probe conjugates, then detected using red duolink detection reagent kit. A representative area is shown at 400× magnification. Each tissue was quantified by densitometry using ImageJ and represented as the proximity ligation/fluorescence count per cell. Mean ± SE from control (n=3) and ethanol-fed (n=6) mouse liver; normal (n=6) and AS (n=18) tissues. **p<0.01 vs. control, †p<0.02 vs. normal liver. Abbreviations: PLA, proximity ligation assay; AS, alcoholic steatosis

FIGURE 2.. Silencing SUMO2 prevents MATα1 depletion caused by ethanol treatment.

Primary mouse hepatocytes were transfected with SUMO2 and SUMO3 siRNAs for 48 hours and treated with 100 mM ethanol for the last 24 hours. (A) SUMO2 and SUMO3 mRNA levels were measure by RT-PCR. Values represent mean ± SEM from n=3 experiments. *p<0.04 vs. Sc; †p<0.02 vs. EtOH. (B) Total and mitochondrial proteins were used to immunoprecipitate SUMO2/3 and MATα1, respectively, to examine their interactions with each other and TOM20 by western blotting. Actin and COXIV were used as housekeeping. Data represent mean ± SEM from n=3 experiments. *p<0.02 vs. Sc; †p<0.01 vs. EtOH for total fraction. *p<0.04 vs. Sc; †p<0.04 vs. EtOH for mitochondria fraction. (C) MATα1 and COXIV immunofluorescence with DAPI as counterstaining. Abbreviations: Sc, scramble control; EtOH, ethanol

FIGURE 3.. Blocking MATα1 SUMOylation at lysine 48 protects against ethanol-induced mitochondrial MATα1 depletion.

Primary mouse hepatocytes were transfected with His EV, MATα1 WT and MUT as indicated in the methods for 48 hours and treated with 100 mM EtOH for the last 24 hours as indicated. (A) RNA was extracted from cells and expression of Mat1a was analyzed by RT-PCR. Data represent mean ± SEM from n=3 experiments. *p<0.02 vs. EV. (B) Total and mitochondria proteins were used to analyze MATα1 protein levels by western blotting. Actin and COXIV were used as housekeeping. Values represent mean ± SEM from n=3 experiments. *p<0.04 vs. WT control. (C) MATα1 and COXIV protein abundancies were visualized by immunofluorescence with DAPI as counterstaining. Abbreviations: MATα1 WT, MATα1 wild type; MUT, MATα1 lysine 48 mutated to arginine

FIGURE 4.. Crosstalk between MATα1 SUMOylation and PIN1 interaction.

Primary mouse hepatocytes were transfected with the His overexpression vector of MATα1 WT, MUT or PIN1 siRNA for 48 hours and treated with 100 mM for the last 24 hours. (A) Total and mitochondria proteins were extracted and immunoprecipitated with the anti-histidine antibody to measure the complex formation of MATα1 with SUMO2/3 and PIN1 in the total proteins and with TOM20 in the mitochondrial fraction by western blotting. Actin and COXIV were used as housekeeping. Data represent mean ± SEM from n=3 experiments. *p<0.03 vs. WT control for total. *p<0.05 vs. WT control for mitochondria. (B) Total proteins were immunoprecipitated with anti-SUMO2/3 antibody to measure MATα1 SUMOylation by western blotting after PIN1 silencing. Results are shown as mean ± SEM, n=3 experiments. *p<0.04 vs. Sc; †p<0.01 vs. EtOH for silencing. (C) IP analysis of serine phosphorylation of MATα1 by western blotting. Data represent mean ± SEM, n=3 experiments. *p<0.01 vs. WT control. Abbreviations: His, histidine-tagged; MATα1 WT, MATα1 wild type; MUT, MATα1 K48R mutant; Sc, scramble control; EtOH, ethanol

FIGURE 5.. Blocking MATα1 SUMOylation prevents alcohol-induced triglycerides accumulation and mitochondrial membrane depolarization.

EV, WT and MUT vectors were overexpressed in primary mouse hepatocytes for 48 hours in co-treatment with 100 mM EtOH for the last 24 hours. (A) Cells were incubated with Nile Red to stain the lipid droplets and measure the level of triglycerides. Data represent mean ± SEM from n=4 experiments. *p<0.04 vs. EV control; †p<0.03 vs. EV+EtOH. (B) Analysis of mitochondria membrane polarization by JC-1 staining as indicated in methods. Results are shown as mean ± SEM, n=3 experiments. *p<0.04 vs. EV control; †p<0.01 vs. EV+EtOH. Abbreviations: EV, Empty vector; WT, MATα1 wild type vector; MUT, MATα1 K48R mutant vector, EtOH, ethanol

FIGURE 6.. CRISPR-directed editing of MATα1 SUMOylation-site protects against alcohol-induced mitochondrial injury by increasing MATα1 mitochondrial content.

Primary mouse hepatocytes were transfected with CRISPR reagents and Cas9 vector to cause CRISPR-directed HDR as described in methods. (A) WT or cells with Cas9 alone were used as controls. CRISPR editing at the MATα1 locus (upper panel) was confirmed by PCR (left panel) using primers that specifically detected the edited region and by NGS (center panel), while the efficiency of gene editing (%) was calculated by densitometry analysis of PCR amplicon bands (right panel). Data represent mean ± SEM, n=3 experiments. *p<0.001 vs. WT modified; †p<0.001 vs WT unmodified. (B) HDR or WT cells were assessed to analyze MATα1 total and mitochondrial abundancies by western blotting. Data represented mean ± SEM from n=3 experiments. *p<0.03 vs. WT control; †p<0.05 vs. HDR control for total proteins. **p<0.02 vs. WT control. (C) mROS and ATP levels. Data represent mean ± SEM, n=3 experiments. *p<0.04 vs. WT control and †p<0.05 vs. WT EtOH for mROS. **p<0.05 vs. WT control and §p<0.05 vs. WT EtOH for ATP. (D) Mitochondrial OCR. Results are shown as mean ± SEM, n=3 experiments. *p<0.03 vs. WT control; †p<0.03 vs. WT EtOH. Abbreviations: WT, wild-type; HDR, CRISPR-edited; mROS, mitochondria ROS; EtOH, ethanol; OCR, oxygen consumption rate

FIGURE 7.. In vivo CRISPR gene editing of MATα1 K48 attenuates ethanol-induced liver injury in the NIAAA model.

NIAAA pair-fed and ethanol-fed mouse livers were treated to mutate the lysine 48 to arginine residue (K48→R48) as indicated in methods. (A) Treatment protocol. (B) Histology and immunohistochemistry of mouse liver tissues. H&E (under 100X magnification), Cas9 immunofluorescence and Oil Red staining of liver sections. (C) Efficiency of CRISPR editing at the MATα1 K48 to R48 residue by NGS. Data represents mean ± SEM, n=4–5 control and n=5 ethanol-fed mice. *p<0.001 vs. control WT modified; †p<0.001 vs. control WT unmodified. (D) ALT, AST and TG levels were measured as indicated in methods. Data represents mean ± SEM, n=4–5 control and n=5 ethanol-fed mice. *p<0.02 vs. WT control, †p<0.05 vs. EtOH control for ALT; *p<0.05 vs. WT control, †p<0.05 vs. EtOH control for AST; *p<0.05 vs. WT control, †p<0.03 vs. EtOH control for triglycerides. Abbreviations: TG, triglycerides; WT, wild type; EtOH, ethanol

FIGURE 8.. In vivo CRISPR gene editing of MATα1 K48 protects against ethanol-induced MATα1 degradation and SAMe depletion.

MATα1 K48 was gene edited to R48 in the livers of NIAAA pair-fed and ethanol-fed mouse livers as indicated in the methods. (A) Total and mitochondrial lysates were used to immunoprecipitate MATα1 to analyze complex formation with PIN1 and measure MATα1 subcellular abundancies by western blotting. Data represented mean ± SE from n=4–5 mice per group. The lanes were run on the same gel but were non-contiguous. (B) Total proteins were immunoprecipitated with an anti-pSer antibody to check the serine phosphorylation status of MATα1 by western blotting. Normal mouse IgG served as a negative control. Data are normalized to actin and shown as mean ± SE from n=4–5 mice per group. *p<0.03, †p<0.04, ‡p<0.01 and **p<0.02 vs. WT control; §p<0.01 vs. WT EtOH. The lanes were run on the same gel but were non-contiguous. (C) SAMe levels were measured as described in methods. Data represent mean ± SEM from n=4–5 mice per group. *p<0.04 vs. WT control; †p<0.04 vs. WT EtOH. Abbreviations: SAMe, S-adenosylmethionine; pSer, phospho serine; WT, wild type; EtOH, ethanol