Eight-week-old male C57BL/6 mice were purchased from Guangxi Medical University (Naning, China) and resided in a standard laboratory environment with room temperature of 22 ± 1 °C and 12/12 h light/dark cycle (26). Drinking water is sterilized and accessible.

Figure 1 shows the experimental study design. We randomly selected 18 mice into three groups: a control group (C group, n = 6), a high-fat model group (M group, n = 6) and a betaine treated group (B group, n = 6). The mice in the C group were fed with a standard diet (10% kcal fat). The M and B groups were fed a high-fat diet (45% kcal fat) for 17 weeks (10, 27, 28). Betaine was dissolved in water at a concentration of 2% w/v (29). Betaine was purchased from Beijing Solarbio Science & Technology Co., Ltd., Beijing, China. At the end of the 25th week, the mice were anesthetized for liver collection. The liver samples were frozen in liquid nitrogen and stored at −80°C until analysis. Besides, two mice closest to the average body weight were selected for high-throughput sequencing in group M and group B, respectively.

HepG2 cells were purchased from National Collection of Authenticated Cell Cultures (Beijing, China) and cultivated in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum at 5% CO₂ and 37°C. When there was an 80% confluence, the cells were seeded into 24-well plates. After 48 h, HepG2 cells were divided into two groups: cells in the OAPA group were cultured with oleic acid (OA, 200 mM) and palmitic acid (PA, 100 mM); cells in the OAPA + Betaine group were cultured with oleic acid, palmitic acid, and betaine (2 mM).

Liver and cell TG content was detected by a TG assay kit (Applygen Technology, Inc., Beijing, China) in accordance with the manufacturer's instructions. Ten microliters of liver frozen sections were stained with Oil Red O (Servicebio, Wuhan, China). The slides were viewed at 200× and 400× magnification.

Total RNA was isolated from M and B groups using TRIzol® RNA extraction reagent (Ambion, Austin, Texas, USA) (30). The mRNA sequencing libraries were established following the TruSeq^TM RNA Sample Preparation Guide (Illumina, San Diego, California, USA) and sequenced with an Illumina HiSeq-2000 for 100 cycles. High quality reads that passed the Illumina quality filters were kept for sequence analysis.

RPFs extraction and sequencing were performed by a commercial service company (Chi-Biotech, Shenzhen, China) based on a previous study (31), with minor modification. A total of 90 mg of liver tissue was pre-treated with 100 mg/ml cycloheximide for 15 min, then washed three times by pre-chilled PBS prior to the addition of 2 ml of cell lysis buffer to T-75 flasks. Cell debris was removed by centrifugation at 16,200 × g for 10 min at 4°C. Supernatants were transferred into 1.5 ml pre-cooled tubes and 2 μl of Ribolock RNase Inhibitor (40 U/μl, Fermentas) was added to each tube. RNase I (10 U/μl, Fermentas) was then added at 0.2 μl per tube, followed by incubation for 15 min at 37°C, and the reaction terminated with 1% SDS. The digested samples were layered on the surface of 15 ml of sucrose buffer (30% sucrose). The ribosomes were pelleted by ultracentrifugation at 185,000 × g for 5 h at 4°C. RNA was then extracted using the Trizol method and ribosomal RNA (rRNA) was removed with a Ribo-Zero rRNA Removal Kit (Mouse) (Epicenter). The fragments with the insertion size of ~28 nt were separated and purified from a gel. This fragment was sequenced by Illumina HiSeq-2000 sequencer for 50 cycles.

For both mRNA and RPF sequencing data sets, clean data were mapped to the reference genome (mm10) through the FANSe2 algorithm (32). The mRNA and RPFs in each sample were normalized by reads per kilo base per million reads (RPKM) (33). Differentially expressed genes (DEGs) were identified with the edgeR package (34). TE was calculated by RPKM _{ribosome−profiling}/RPKM _RNA−seq. Differential TE genes (DTEGs) were calculated by t-test. Principal component analysis (PCA), Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis, and gene set enrichment analysis (GSEA) were performed using the OmicShare tools, a free online platform for data analysis (http://www.omicshare.com/tools). Protein-protein interaction (PPI) network analysis of DTEGs was conducted using the online STRING website (https://string-db.org/).

All results were analyzed by GraphPad Prism 8 and expressed as the mean ± standard error. The unpaired two-tailed T-test (for two groups) and one-way ANOVA (for multiple groups) were used to identify the significance of difference. If P < 0.05, the difference was considered statistically significant.

From week 9, the body weight of mice in the M and B groups was significantly higher than that in the C group, indicating successful establishment of the high-fat animal model. There was no significant difference in body weight between the M and B groups (Figure 2A). In terms of liver weight, epididymal fat/body weight ratios, and feces TG levels, significantly higher were found in the M group compared with the C group, with no significant difference between the B and M groups (Figures 2B–D). Compared with the C group, the liver TG levels in the M group increased approximately 1.7 times. There was a significant reduction in hepatic TG levels in the B group compared with the M group (Figure 2E). The regulatory effect of betaine on TG accumulation in cultured HepG2 cells was assessed by adding oleic acid (200 μM, OA) and palmitic acid (100 μM, PA) to DMEM medium to simulate a high lipid environment. Compared with the OAPA group, intracellular TG levels for the OAPA + Betaine group were significantly reduced (Supplementary Figure 1). Moreover, a large number of liver lipid droplets were found in the M group by Oil Red O staining, while the number of lipid droplets was significantly decreased in the B group (Figure 2F). These results demonstrate betaine to reduce high fat diet-induced hepatic fat accumulation in mice.

To evaluate betaine regulation of hepatic lipid metabolism, we compared ribosomal profiles of the M and B group livers by Ribo-seq and RNA-seq. At the transcriptional level, 15,847 and 16,215 genes were identified in the M and B groups, respectively (RPKM > 0). At the translation level, 14,308 and 11,063 genes were identified in the M and B groups, respectively (RPKM > 0). The gene expression levels for both the transcriptome and the translatome were similar with normal distribution, log2 RPKM = 3 (Supplementary Figures 2A–D). The identification and quantification information for the transcriptome and translatome are shown in Supplementary Tables 1, 2. PCA of RNA-seq and ribosome profiling showed that biological replicates for the M and B groups to be highly related (Supplementary Figures 2E,F). Supplementary Figure 2G demonstrates the RPFs length distribution peaked at 31 nt in both M and B groups, which was similar to that reported by other species (21–24). The pattern of RPF distribution ratios was similar for the M and B groups (Supplementary Figure 2H). The mRNAs protein-coding sequences (CDS) contained the majority of RPFs in the M and B group, with an average distribution ratio of 59.06 and 64.68%, respectively. The 3′UTR distribution ratio was 37.10 and 29.29%, respectively, for the M and B groups. The 5′UTR distribution ratio was 3.82 and 6.02% for the M and B groups for the RPF distributions, respectively. These data demonstrate the reproducibility of our analysis.

Based on RNA-seq data, transcriptional gene expression was compared between the M and B groups of mice. In total, 15,300 genes were expressed in both the M and B groups, which were highly related (R² = 0.9089) (Figures 3A,B). Nine hundred twenty-three up-regulated and 807 down-regulated DEGs of transcriptome were identified based on |log₂ fold change| > 1 and P < 0.001 (Figure 3C). The heat map exhibited 1,730 DEGs based on gene expression level (Figure 3D). In order to further understand the betaine mechanisms of action in hepatic steatosis, GSEA was performed. By GSEA, several positive correlations were identified including; fatty acid beta oxidation, lipid oxidation, and fatty acid catabolic processes (Figure 3E). Supplementary Figure 3 presents the gene set and expression levels of the above up-regulated pathways. These results demonstrated betaine to relieve hepatic steatosis by up-regulation of fatty acid beta oxidation, lipid oxidation, and fatty acid catabolic processes.

To determine whether betaine plays a lipid-lowering role at the translational level, the translational expression abundances were assessed. There were 10,813 overlapping genes in the M and B groups, which were highly related (R² = 0.7961) (Supplementary Figures 4A,B). A volcano plot was used to screen translatome DEGs by |log₂ fold change| > 1 and P < 0.05. Among those, 254 genes were highly expressed in the B group and 320 genes were highly expressed in the M group (Figure 4A). A heatmap visualized the expression of 574 DEGs (Figure 4B). GO and KEGG enrichment analyses were used to investigate DEGs function. Figure 4C identified the GO biological process terms, classified by –log10 (P-value). There was significant enrichment in organic acid metabolic process, oxoacid metabolic process, and carboxylic acid metabolic process. KEGG analysis identified significant enrichment pathways including non-alcoholic fatty liver disease, oxidative phosphorylation, and insulin resistance (Figure 4D). As shown in Supplementary Figure 4C, the KEGG network diagram identified 17 DEGs that Ribo-seq found enriched in the NAFLD pathway. Therefore, betaine can regulate hepatic steatosis at the translational level.

According to both transcriptome and ribosome profiling data, transcriptional, and translational differences were detected between the M and B groups. There were 923 up-regulated and 807 down-regulated DEGs in the transcriptome, and 254 up-regulated and 320 down-regulated DEGs in the translatome (Figure 5A). Among the DEGs, coordinately DEGs at both the transcriptional and translational levels contained 28 up-regulated and 23 down-regulated DEGs (Figure 5B). Figure 5C shown a quadrant diagram of fold changes in transcriptional and translational differences between the B and M groups. |log₂(fold change of RPKM) |>1 and P < 0.001 for the transcriptome and P < 0.05 for the translatome divided the change pattern into nine categories. Class E was comprised of 79.96% of all genes and represented no significant change in transcriptional or translational levels, whereas 0.35 and 0.29% of genes located in class C and G, respectively, had gene changes congruent at the transcriptional and translational levels. Another 19.4% of genes were situated in the six discordantly regulated groups (classes A, B, D, F, H, and I). Supplementary Table 3 lists gene information for the different classes. These results indicated that betaine exerts different regulatory effects at the transcriptional and translational levels.

KEGG enrichment analysis was carried out for classes B, D, F, and H (Figure 5D). Genes in classes B and H were differentially expressed at the translational level without significant difference in transcription. Among them, translationally up-regulated DEGs were significantly associated with pathways of oxidative phosphorylation, thermogenesis, and NAFLD (class B), while translationally down-regulated genes were also observed to be enriched in NAFLD (class H). On the contrary, genes in classes D and F were differentially expressed at transcriptional level without remarkable change at the translational level. In class D, genes with down-regulated expression in transcription were significantly enriched for spliceosome, ribosome biogenesis in eukaryotes, and the mRNA surveillance pathway. There were some transcriptionally up-regulated DEGs (class F), participated in metabolic pathways including, fatty acid degradation, steroid biosynthesis, and biosynthesis of unsaturated fatty acids. By comparison of transcriptome and translatome, we found betaine to more profoundly regulate liver lipid metabolism at the translation level.

TE is a vital translation indicatrix, reflecting the efficiency of mRNA entry into translation. TE is linked to the expression abundance of transcription and translation. Supplementary Figure 5A identified 58 up-regulated DTEGs and 102 down-regulated DTEGs (|log2 fold change of TE| > 2 and P < 0.05), demonstrating that gene TE is important for the effects of betaine on hepatic lipid metabolism. Supplementary Figure 5B showed the value of log₂TE to be mainly distributed between −2 and 3. Functional protein association network analysis was conducted and the interactions of the 160 DTEG coding proteins analyzed to identify target genes of betaine. Results identified 139 nodes and 101 edges that were interconnected, and four DTEGs (IDI1, CYP51A1, TM7SF2, and APOA4) that were enriched in three pathways including lipid biosynthetic processes, regulation of lipid biosynthetic processes, and cholesterol biosynthetic processes (Figure 6A). Compared to the M group, the TEs of IDI1, CYP51A1, TM7SF2, and APOA4 were decreased in the B group (Figure 6B). These results suggest that betaine relieves liver lipid metabolism disorders, possibly through inhibition of gene TE involved in lipid biosynthetic processes, regulation of lipid biosynthetic processes, and cholesterol biosynthetic processes.