Five-week-old male C57BL/6 mice were purchased from Guangxi Medical University (Guangxi, China) and used to establish the NAFLD model. MRPL44 transgenic mice were administered adeno-associated virus serotype 8 (AAV8), AAV8-CMV-MRPL44-ZsGreen via tail vein injection, while the control group received AAV8-CMV-ZsGreen. All animals were maintained under controlled conditions (23 ± 2 °C, 12 h light/dark cycle). At 8 weeks, mice were randomized into four groups and fed either a normal diet (ND) or high-fat diet (HFD) for 12 weeks, following our previously established protocol (16): (1) ND group: fed a ND and injected with AAV8-CMV-ZsGreen; (2) ND-MRPL44 group: fed a ND and injected with AAV8-CMV-MRPL44-ZsGreen; (3) HFD group: fed a HFD and injected with AAV8-CMV-ZsGreen; and (4) HFD-MRPL44 group: fed a HFD and injected with AAV8-CMV-MRPL44-ZsGreen. All animal protocols were approved by the Animal Ethics Committee of Guangxi University (GXU-2021-160).

Lean mass and fat mass of mice were analysed using a Niumag nuclear magnetic resonance (NMR) analyser (QMR23-060H-I, Niuma, Suzhou, China). Whole-body scans were performed with micro-CT (X-ray Safety Report Skyscan 1278, Bruker, Germany). Metabolic parameters including oxygen consumption, CO₂ production, energy expenditure, respiratory exchange ratio, and activity were assessed in metabolic cages (17).

HepG2 cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, Beijing, China) containing 1% penicillin-streptomycin, and 10% fetal bovine serum (BI, Guangzhou, China) at 37 °C and 5% CO₂. DMEM contains 4.5 g/L glucose, L-glutamine and sodium pyruvate. Cells were seeded in 24-well plates at a ratio of approximately 1–2 × 10⁵ cells/mL per well, and treated with oleic acid/palmitic acid (OA/PA, 2:1) for 24 h. The final concentration of OA was 200 μM and that of PA was 100 μM.

Human MRPL44 (NM_022915.5) was cloned into the pcDNA3.1(−) vector. The mt-Keima plasmid was purchased from HonorGene (HonorGene, Changsha, China). siRNA library was purchased from Guangzhou RuiBo (Guangzhou RiboBio Co., Ltd., Guangzhou, China), and each siRNA contained three sequences. Negative control was purchased from Sangon Biotech (Sangon Biotech Co., Ltd., Shanghai, China). Three siMRPL44 sequences and negative control sequences have been listed in Supplementary Table S1. Transient transfections of plasmids or siRNAs were conducted with Hieff Trans^® Liposomal Transfection Reagent (Yeasen Biotechnology, Shanghai, China) following manufacturer conditions.

The Triglyceride Assay Kit and Glycerol Assay Kit (Nanjing Jiancheng Bioengineering Institute, Nanjing, China) were used to assess the glycerol and triglyceride concentration. Fatty acid uptake assay was performed using the Fatty Acid Uptake Assay Kit. Lipase (LPS) activity was measured using a commercial kit (Lipase Assay Kit, Nanjing Jiancheng Bioengineering Institute, Nanjing, China). ATP content was measured using a commercial kit (ATP Assay Kit, Beyotime Biotechnology, Shanghai, China). All studies were then carried out in accordance with the manufacturer’s guidelines while quantifying protein concentrations. For each experiment, the BCA Protein Quantification Assay Kit (Beyotime Biotechnology, Shanghai, China) was used to measure the protein concentrations. The intracellular lipid deposition distribution was observed by oil red O staining, and the Oil Red O working solution was prepared in a ratio of Oil Red O storage solution to water of 3:2. Images were obtained under an inverted microscope (IX53; Olympus Corporation, Tokyo, Japan).

RNA was extracted with Trizol and the RNA concentration were determined using a microplate reader. To generate cDNA, 1 μg of RNA was reverse transcribed using the M-MLV enzyme and the random primer OLIGODT18 according to manufacturer’s instructions. For gene expression analysis, we performed quantitative real-time polymerase chain reaction (qPCR) of cDNA per sample using the GenStar commercial kit (2× RealStar Green Fast Mixture; Takara Bio, Japan). The expression of mRNA was normalized to expression of β-actin. The primer sequences were listed in Supplementary Table S2. Each experiment was repeated at least three times.

For immunofluorescence staining, HepG2 were washed with PBS and fixed in 4% paraformaldehyde (Solarbio, Beijing, China). After permeabilization in 0.5% Triton X-100 and blocked with 5% BSA (Solarbio, Beijing, China). Cells were incubated with primary antibodies at 4 °C overnight. Subsequently, the samples were washed with PBS and incubated secondary antibody with fluorescent label. The primary antibody used in immunofluorescence was: rabbit anti-LC3 (T55992S, Abmart, 1:500). Reactive Oxygen Species Detection Kit from Beyotime Biotechnology to observe and analyze intracellular ROS production. Images were captured using an inverted fluorescence microscope (IX53; Olympus Corporation, Tokyo, Japan), and the fluorescence values of ROS were quantified using a microplate reader (Infinity M200 PRO, Tecan, Switzerland).

Commercial chemical assays purchased from Beyotime Biotechnology were used to determine the content of glutathione peroxidase (GPx), superoxide dismutase (SOD), and malondialdehyde (MDA). The obtained supernatant was used to measure changes in these markers of oxidative stress following the manufacturer’s instructions.

MitoTracker^® Green probe (Yeasen Biotechnology, Shanghai, China) was used for mitochondrial staining. DNA was isolated from HepG2 cells using PCR lysis buffer (18), and the relative content of mitochondrial DNA and nuclear DNA was determined by quantitative PCR. The mitochondrial membrane potential assay kit with JC-1 (Beyotime Biotechnology, Shanghai, China) was used to detect mitochondrial membrane potential. Mitophagy levels were evaluated based on the colocalization of mitochondria and lysosomes, as well as the expression of the mt-Keima plasmid. LysoTracker Red DND-99 (Yeasen Biotechnology, Shanghai, China) was used to label intracellular lysosomes with red fluorescence.

Oxygen consumption rate (OCR) was measured in HepG2 cells using a Seahorse XF24 Metabolic Extracellular Flux Analyzer (Seahorse Biosciences, North Billerica, MA, United States). Oligomycin (1 μM), Trifluoromethoxy carbonylcyanide phenylhydrazone (FCCP, 1 μM), and rotenone/antimycin A (ROT/AA, 0.5 μM) were sequentially injected. Values were normalized against protein concentration.

The amino acid sequences of candidate proteins MRPL44 (Q9H9J2) and BNIP3 (Q12983) were obtained from the Universal Protein Resource (Uniprot) database, and analyzed using AlphaFold3 for structure prediction (19). Protein–protein docking was performed with the HDOCK server, and structural models were visualized in PyMOL (20, 21). For experimental validation, HepG2 cell lysates were subjected to co-immunoprecipitation using MRPL44 antibody or IgG control, followed by pulldown with Protein A/G beads. The precipitated proteins were analyzed by SDS-PAGE. In addition, protein expression levels of MRPL44, BNIP3, LC3, and OXPHOS components were assessed by Western blotting. The primary antibodies used were as follows: rabbit anti-MRPL44 (16394-1-AP, Proteintech, 1:1000), mouse anti-Total OXPHOS Rodent WB antibody Cocktail (ab110413, Abcam, 1:2000), mouse anti-Membrane Integrity WB antibody Cocktail (ab110414, Abcam, 1:2000), rabbit anti-LC3 (T55992S, Abmart, 1:1000), mouse anti-BNIP3 (Cat No. 68091-1, Proteintech, 1:5000), rabbit anti-β-actin (AF5012, Beyotime Biotechnology, 1:1000).

All experiments included at least three biological replicates. Statistical analyses between different groups were performed using the Student’s t test or one-way ANOVA (GraphPad Software). The data are presented as the mean ± SEM and the level of significance is indicated as follows: ^*p < 0.05 and ^**p < 0.01.

To identify putative genes linked to lipid metabolism, we purchased siRNA libraries and performed screening. Each target gene was represented by three siRNA sequences, which were pooled and co-transfected into HepG2 cells. After OA/PA treatment, triglyceride (TG) levels were measured and normalized to the negative control group (siNC). The data presented in Figure 1A, suggest that MRPL44 may be involved in lipid metabolism. The mRNA level of MRPL44 was examined in OA/PA-treated HepG2 cells. We found that OA/PA treatment significantly downregulated MRPL44 expression by 28% (p = 0.0075; Figure 1B). Subsequently, we designed three different siMRPL44 constructs to inhibit MRPL44, and their inhibition efficiencies in cells are shown in Figure 1C. Among them, siMRPL44-3 achieved the strongest knockdown effect, reducing MRPL44 expression by 55% (p = 0.0023). The results showed that siMRPL44 significantly increased the intracellular TG content of HepG2 cells (Figure 1D), with no significant effect on extracellular TG content (Figure 1E) or intracellular total cholesterol (TC) content (Figure 1F).

To further investigate the role of MRPL44 in lipid metabolism, we overexpressed MRPL44 in HepG2 cells. Transfection of MRPL44 significantly upregulated MRPL44 expression compared with the pcDNA3.1(−) group. mRNA levels increased by approximately 75-fold (p < 0.001; Figure 2A), and protein levels were elevated by about 2-fold (p < 0.001; Supplementary Figure S1A), as confirmed in Figure 2B. Oil Red O staining demonstrated that MRPL44 overexpression reduced intracellular lipid accumulation (Figure 2C). Quantification revealed a 20% decrease in neutral lipid staining compared with the pcDNA3.1(−) group (p = 0.0073; Figure 2D). The intracellular TG and TC determination results showed that MRPL44 significantly reduced intracellular TG levels by 29% (p < 0.001, Figure 2E), with no significant effect on extracellular TG levels (Figure 2F). Moreover, MRPL44 reduced the intracellular TC levels (Figure 2G).

To explore how MRPL44 is involved in lipid deposition, we measured lipase content and fatty acid uptake capacity. MRPL44 markedly increased lipase content (Figure 2H) and significantly reduced the uptake of fatty acid (Figures 2I,J). Meanwhile, we found that MRPL44 significantly decreased the mRNA expression levels of genes associated with fatty acid synthesis (FASN, SREBP1C) while significantly increased the levels of mRNA expression of genes associated with fatty acid hydrolysis (LCAT, CPT1, LIPC, HSL) (Figure 2K). Based on these findings, we speculate that MRPL44 might inhibit lipid accumulation by encouraging fatty acid hydrolysis.

Lipid metabolism and mitochondrial function are tightly related. To investigate the impact of MRPL44 on mitochondrial function, we performed several assays. The results of the mitochondrial fluorescent probe assay indicated that MRPL44 transfection significantly increased mitochondrial activity (Figures 3A,B), and elevated mtDNA levels by approximately 30% (p = 0.0126, Figure 3C). Quantitative analysis of genes associated with mitochondrial fusion, fission revealed that MRPL44 upregulated mRNA expression of the fission -related gene FIS1 by 0.32-fold (p = 0.0419), and the fusion-related genes MFN2 by 0.68-fold (p < 0.001), OPA1 by 0.21-fold (p = 0.014) (Figure 3D). To further explore the effects of MRPL44 overexpression on mitochondrial function, we conducted a mitochondrial stress test (Figure 3E). The results demonstrated that MRPL44 enhanced basal cellular respiratory capacity, mainly accompanied by increased ATP production by 20.8% (p < 0.001, Figure 3F). Additionally, the maximal respiratory capacity in the MRPL44 group was slightly higher than in the pcDNA3.1 (−) group (Figure 3G), though the difference was modest. Notably, MRPL44 significantly elevated non-mitochondrial respiration by 0.28-fold (p < 0.001; Figure 3H), improved the coupling efficiency of the electron transport chain (Figure 3I). Consistent with these findings, intracellular ATP measurements revealed a 1.76-fold increase in ATP levels in cells overexpressing MRPL44 cells (p = 0.0015; Figure 3J). Moreover, we observed a 0.28-fold upregulation of PGC-1α mRNA, a key regulator of mitochondrial biogenesis and respiration, in MRPL44 group (p = 0.01; Figure 3K). We then examined the levels of mitochondrial membrane-related proteins and oxidative phosphorylation-related proteins. MRPL44 modestly upregulated the expression of several mitochondrial membrane proteins and proteins involved in oxidative phosphorylation (Figures 3L,M; Supplementary Figures S1B,C). Therefore, we propose that MRPL44 aids in increasing mitochondrial mass and maintaining mitochondrial function.

Lipid peroxidation causes oxidative damage in mitochondria, mitochondrial damage lowers the membrane potential of the mitochondria, and accumulates ROS. We found that MRPL44 restored the OA/PA-induced decline in mitochondrial membrane potential, as reflected by a 19% increase compared with the OA/PA group (p = 0.046; Figures 4A,B). MRPL44 effectively reduced intracellular ROS levels by 13% (p = 0.008; Figures 4C,D), decreased the content of the peroxidation product MDA by 13.7% (p = 0.012; Figure 4E). MRPL44 downregulated the mRNA expression of the proinflammatory cytokines TNF-α by 35% (p = 0.018) and IL-1 by 72% (p = 0.022; Figure 4F). Subsequently, MRPL44 significantly enhanced the activity of the antioxidant enzyme SOD by 24.6% (p = 0.013; Figure 4G), whereas no significant change was observed in GPx activity (Figure 4H). These results suggest that MRPL44 may exert protective effects against oxidative stress predominantly via the SOD pathway.

Mitophagy plays a critical role in maintaining intermediary metabolism by serving as one of the most effective mechanisms for clearing damaged mitochondria. To investigate the impact of MRPL44 on mitophagy, we first measured the expression of autophagy-related genes. MRPL44 significantly reduced LC3 I mRNA levels while markedly increasing Atg5 mRNA levels (Supplementary Figure S2A). We then used the autophagy inhibitor chloroquine (CQ, 50 μM) to further explore the effects of MRPL44 on autophagy. Western blot analysis of autophagy-related proteins revealed that MRPL44 transfection resulted in increased LC3 II protein expression, decreased LC3 I protein expression compared to the control group. Following CQ treatment, the accumulation of LC3 II protein was further enhanced (Figure 5A; Supplementary Figure S2B). Similarly, immunofluorescence staining of LC3 showed increased fluorescence intensity of LC3 protein after MRPL44 transfection, with a further increase after CQ treatment (Figure 5B; Supplementary Figure S2C).

Mitophagy is closely associated with high levels of lysosomal activity. Using fluorescence probes to label mitochondria and lysosomes, we observed that the MRPL44 group exhibited a greater number of autolysosomes compared to the pcDNA3.1 (−) group (Figure 5C). To assess MRPL44 impact on mitophagy, we used the mt-Keima plasmid. MRPL44 overexpression exhibited a stronger red fluorescence signal compared to the pcDNA3.1(−) group, indicating that mitochondria were in an acidic environment and showing higher mitophagy levels (Figure 5D). These results suggest that MRPL44 enhances mitophagy by promoting the clearance of damaged mitochondria.

BNIP3, as a mitophagy receptor, plays a crucial role in maintaining mitochondrial homeostasis. We found that overexpression of MRPL44 significantly upregulated BNIP3 expression, with mRNA levels increased by 48% (p = 0.045; Figure 5E), and protein expression consistently elevated as shown in Supplementary Figures S2D,E. Inhibition of BNIP3 expression impaired autophagosome formation, leading to a reduction in LC3-II protein levels (Figure 5F; Supplementary Figure S2F). When siBNIP3 was co-transfected with MRPL44, the reduction in LC3-II expression caused by siBNIP3 was rescued (Figure 5G; Supplementary Figures S2G, S3A,B). Mitochondrial-lysosomal staining and co-localization analysis revealed that the siBNIP3 + pcDNA3.1(−) treatment group exhibited weaker mitochondrial and lysosomal fluorescence signals, whereas the siNC + MRPL44 group showed significantly enhanced co-localization between mitochondria and lysosomes. The siBNIP3 + MRPL44 treatment group partially restored this co-localization, suggesting that MRPL44 exerts a compensatory effect in the absence of BNIP3 (Supplementary Figure S3C). Furthermore, compared to the siNC group, siBNIP3 significantly increased intracellular TG levels in HepG2 cells. Co-transfection with MRPL44 alleviated the TG accumulation induced by siBNIP3 (Supplementary Figures S3E,F). These findings suggest that MRPL44 may influence BNIP3-mediated LC3 protein dynamics, thereby affecting mitophagy in HepG2 cells.

To investigate the structural basis underlying the interaction between MRPL44 and BNIP3, we utilized AlphaFold3 to predict the structure of the protein complex. The predicted model is presented in Figure 5H (the protein nucleic acid complex structure were predicted by Hefei Keiing Biotechnoloay Co., Ltd.), with per-residue confidence levels indicated by the predicted local distance difference test (pLDDT) scores. The predicted template modeling score (pTM) and the interface pTM score (ipTM) were 0.54 and 0.10, respectively. The sum of these scores (ipTM + pTM > 0.5) suggests a potential interaction between the two proteins. Additionally, the predicted aligned error (PAE) map demonstrated high intra-domain positional accuracy, further supporting the reliability of the individual structural predictions. Subsequent protein–protein docking analysis revealed that the two proteins, MRPL44 (cyan) and BNIP3 (purple) form a stable docking interface (Figure 5I). Structural displays from different angles indicate that the two have tight spatial matching and potential interaction regions on the surface, suggesting that MRPL44 and BNIP3 may have direct physical binding, providing a structural basis for their regulation of mitophagy. Magnified views of the docking model highlighted several key intermolecular contacts, residues such as ALA188, TYR182, and PHE157 of BNIP3 form hydrogen bonds and electrostatic interactions with GLU163, LYS153, and GLU183 of MRPL44 (Supplementary Figure S3D). Furthermore, co-immunoprecipitation assays confirmed the physical association between MRPL44 and BNIP3 (Figure 5J). These results suggest that MRPL44 may activate the mitochondrial autophagy process through the BNIP3-LC3 axis.

Given the effects of MRPL44 on lipid metabolism, we employed mice with liver-specific overexpression of MRPL44 to examine its role in vivo. A schematic diagram of mouse model rearing is shown in Figure 6A. The mRNA and protein levels of MRPL44 were detected in the liver, confirming high upregulation of MRPL44 (Figures 6B,C; Supplementary Figure S1D). The liver morphology of the four groups of mice is shown in Figure 6D. Data analysis revealed that liver-specific overexpression of MRPL44 had no significant effect on body weight, fat distribution, fat content, and fat mass in HFD-fed mice (Supplementary Figures S4A–G). Metabolic cage analysis showed that compared to the HFD group, MRPL44 overexpression did not result in significant changes in the respiratory exchange ratio (Figure 6E), but it did increase oxygen consumption (Figure 6F), carbon dioxide production (Figure 6G), and energy expenditure (Figures 6H,I).

We then examined liver lipid deposition in the four groups. Liver images showed that MRPL44 overexpression significantly reduced liver weight in HFD-fed mice (Figures 7A,B). MRPL44 also significantly decreased TG (Figure 7C) and TC levels (Figure 7D). However, liver TC levels were increased in ND-MRPL44 mice compared to the ND group. This increase may be attributed to the lower hepatic metabolic load under ND conditions, where MRPL44 overexpression likely promotes cholesterol storage to support metabolic demands. H&E and Oil Red O staining of liver sections demonstrated a marked reduction in lipid deposition in the MRPL44 group (Figure 7E). Subsequently, we quantified the expression of genes related to lipid metabolism and inflammation in the liver of mice. The results indicated that MRPL44 significantly downregulated the mRNA expression of genes involved in fatty acid synthesis and inflammatory factors. Specifically, ACC was downregulated by 44% (p = 0.009), SREBP1C by 43% (p = 0.018), SCD1 by 66% (p = 0.032), TNF-α by 66% (p = 0.027), and IL-1β by 69% (p < 0.001). Conversely, MRPL44 significantly upregulated the mRNA expression of LCAT by 60% (p < 0.001), LIPC by 52% (p = 0.001), and CPT1 by 51% (p = 0.016) (Figure 7F; Supplementary Figure S1E). In addition, we assessed the expression of genes related to mitochondrial fission, fusion and autophagy. The findings showed that MRPL44 markedly increased the mRNA expression of OPA1 by 0.53-fold (p = 0.01), FIS1 by 0.77-fold (p = 0.02) and DRP1 by 0.63-fold (p = 0.003) (Figure 7G). Moreover, MRPL44 upregulated BNIP3 mRNA levels by 57% (p = 0.013), downregulated LC3A by 41% (p = 0.028), and upregulated LC3B by 42% (p = 0.016) (Figure 7H). Consistent with these changes, protein levels of BNIP3 and LC3 in the liver of MRPL44-overexpressing mice were significantly elevated (Figure 7I; Supplementary Figure S1F).

In summary, these results suggest that MRPL44 mitigates HFD-induced hepatic lipid deposition and improves MASLD.

This study has several limitations. First, we used of only male C57BL/6 mice may limited the generalizability of the findings, as numerous studies have reported sex differences in NAFLD susceptibility and mitochondrial quality control pathways. Future studies should include both sexes to better understand these differences. Second, we used only β-actin as the reference gene for qPCR, which is commonly used in NAFLD research. However, the MIQE guidelines recommend validating 2–3 reference genes to ensure accuracy. This limitation will be addressed in future experiments. Lastly, the small sample size in some experiments (three biological replicates per group) may limit statistical power. Future studies should expand the sample size to enhance statistical significance.