Transcriptome and translatome analyses reveal the regulatory role of betaine in high fat diet (HFD)-induced hepatic steatosis


 Non-alcoholic fatty liver disease (NAFLD) is a common disease with a multitude of complications. Increasing evidence shows that the dietary supplement with betaine, a natural chemical molecule, can effectively reduce the fat accumulation in the liver. Translational regulation is considered to play a vital role in gene expression, but whether betaine functions through the regulation of gene translational level is still unclear. To this end, RNC-seq (ribosome-nascent chain complex bound mRNA sequencing) and RNA-seq co-analyses were performed to identify betaine target genes by using the liver samples from high-fat diet + betaine treated and high-fat diet treated mice. The results showed that betaine does play a lipid-lowering role by regulating the expression of gene translation levels; some NAFLD- and lipid metabolism- associated genes were differentially expressed at translational level, for example. And the mRNA translation ratio (TR) of gene significantly increased after betaine treatment. Besides, it is found that the regulation of some genes at transcriptional level is opposite to that at translational level, which indicates that transcriptional regulation and translational regulation may be independent from each other. Finally, we identified several candidate genes, especially Gpc1 , which may mediate the lipid-lowering effect of betaine in the liver. To sum up, this study depicted the molecular portrait of mice liver with or without betaine treatment from the angel of translatome and transcriptome, giving insights into the molecular mechanism of betaine-mediated lipid-lowering effect and also providing new clues for understanding and prevention of NAFLD.


Introduction
Increasing proof demonstrates that excessive fat/energy intake is related to the risk of developing nonalcoholic fatty liver disease (NAFLD) [1], which is a common form of chronic liver disease [2]. Approximately one-fourth of adults in the world are suffering from NAFLD, which is an increasing threat to human health [3]. This disease is a stress-associated hepatic injury and metabolic disorder featured by diffuse hepatic steatosis and triglyceride accumulation [4,5]. Some vital factors, such as lipid deposition, oxidative stress and inflammatory factors are considered to have significant effects on the occurrence and progression of NAFLD [6], which is linked to obesity [7], insulin resistance (IR) [8], diabetes [9], cardiovascular disease (CVD) [10], nonalcoholic steatohepatitis (NASH) [11], liver cirrhosis [12] and liver cancer [13]. At present, there is no consensus regarding the prevention and treatment of NAFLD, except healthy diet and reasonable exercise, combined with drug adjuvant therapy [14]. Thus, it is urgent to identify potential therapeutic targets, novel drugs and treatment. Betaine (N,N,N-trimethylglycine, glycine betaine) exists in various common foods，including beets, spinach, chinese wolfberry, wheat, shrimp and shellfish [15][16][17]. Betaine has been widely used for decades due to its anti-inflammatory, methyl donor, and osmotic pressure regulator effects [18] and has been proven to effectively improve NAFLD [19][20][21][22][23][24]. As reported, betaine prevented high fat diet-induced NAFLD by regulating the FGF10/AMPK signaling pathway in ApoE -/mice [19]. Supplementing betaine after high fat diet-induced NAFLD will make the downstream pathways, which are involved in insulin signal transduction, gluconeogenesis and glycogen synthesis, function normally and improve insulin resistance and steatosis [20]. Betaine, as a methyl donor, alleviates fatty liver induced by corticosterone through epigenetic modification [21].
Betaine attenuates hepatic steatosis by reducing methylation of the MTTP promoter and elevating genomic methylation in mice fed a high fat diet [22].
Betaine has significantly increased mitochondrial content and improved liver lipid metabolism in oleic and palmitic acids induced HepG2 cell [23].
Betaine can improve mitochondrial function in the liver of mouse model with NAFLD caused by methionine and choline, alleviate steatosis, and increase the number of autophagosomes in the mouse's liver [24].
In the process of biological genetic information transmission, translational regulation accounts for more than half of all regulation, which is the most important regulation mode in cells [25,26]. There is insufficient correlation between mRNA abundance and protein abundance, which indicates that there is a certain error in taking gene transcription abundance as protein abundance [27,28]. Therefore, it is not enough to evaluate the change of protein level through the difference of transcription level. Similarly, for proteome, some low-abundance proteins are not well identified by mass spectrometry (MS), which may lead to some low-abundance proteins with important biological functions not being detected [29]. Consequently, proteomics also has some shortcomings. The purpose of translatome is to study the process of protein production from mRNA translation, which can provide vital information for the translational regulation [30]. Ribosome nascent-chain complex (RNC) bound mRNA (RNC-mRNA) full length sequencing ( RNC-seq ) is a recently developed method to gaining a genome-wide view of the translational process. The emergence of translatome makes up for the deficiency of transcriptome and proteome, and provides a new perspective for analyzing biological issues. However, at present, the research on translatome is less than that on genomics, transcriptomics and proteomics [31]. Therefore, it is necessary to study translatome.
In this study, the combined analysis of translatome and transcriptome was used to explain betaine alleviate hepatic steatosis induced by high fat diet.
Mice were induced into NAFLD model after 17 weeks of high fat diet.
Regarding mice, the use of a high fat diet is the preferable model, since the phenotype of NAFLD developed by this model resembles human NAFLD most accurately [32]. Three groups of mice were included. Mice fed with normal fat diet are NFD group; those fed with high fat diet are NAFLD animal models, and this group is called HFD group; those fed with high fat diet and drinking water added with betaine are HB group. Two sequencing methods, RNA-seq and RNC-seq, were used to explore the differences and associations between transcriptional and translational levels corresponding to this phenotype to further screen out some known target genes and new functional genes. Therefore, this study aims to provide a new direction for the treatment of NFALD through the joint analysis of translatome and transcriptome.

Animal and diet
18 Five-week-old C57BL /6 similar weight male mice used in the experiment were purchased from Changsha Tianqin Biotechnology Co., Ltd., Changsha, China, and were acclimated with normal fat diet for 2 weeks. All mice were housed in a controlled environment with a temperature of 24±2℃ and a 12-hour light/dark cycle, with free access to food and water [33]. At week 8, the mice were randomly divided into 3 groups (n=6 in each group), balanced to achieve a similar average body weight. The NFD (control) group was fed with a normal fat diet (11.7% kcal fat). The HFD (NAFLD) group was fed a high fat diet (45% kcal fat; MD12032; Jiangsu Medicience Ltd., Jiangsu, China). The HB (NAFLD + Betaine) group was fed a high fat diet with betaine in drinking water (2% w/v; Beijing Solarbio Science & Technology Co., Ltd., Beijing, China) [19,34]. The mice were sampled in they were 25

Liver sample collection
After feeding the mice to the age of 25 weeks, euthanasia was conducted on them via cervical dislocation after anesthesia. Liver tissues were immediately separated, frozen in liquid nitrogen, and stored at -80℃ for later use.

Triglyceride measurement and Oil Red O staining
Triglyceride (TG) levels in liver samples or cells were measured with an analytical kit (Applygen Technology, Inc., Beijing, China), as directed by the manufacturer. The TG content was normalized to protein concentration that was determined by a BCA protein quantitative analytical kit (P0009; Beyotime Institute of Biotechnology, Shanghai, China).
Liver tissues were prepared into frozen sections, which were stained with Oil Red O (G1016; Wuhan Servicebio Technology Co., Ltd., Wuhan, Hubei, China) for 10 min, and then counter-stained with hematoxylin for 2 min. The slides were viewed at 200× magnification.

RNA extraction
Total RNA was isolated by TRIzol RNA extraction reagent (Ambion, Inc., Austin, TX, USA) under the manufacturer's instructions. Total RNA samples were prepared from two independent treatment groups. Equal amount of total RNA from each preparation was pooled respectively, for subsequent library construction and RNA-seq.

RNC-mRNA extraction
Two biological replicates from each group were selected to perform RNC-mRNA extraction. RNC isolation was performed in accordance with a previously reported procedure [31] with some modifications. A total of 90 mg of liver tissues were pre-treated with 100mg/ml cycloheximide for

Sequence analysis
For both mRNA and RNC-mRNA sequencing data sets, high quality reads were mapped to RefSeq mRNA reference sequence (mm10 25-Mar-2017) through the FANSe2 algorithm [35] with the parameters -L80 -E5 -I0 -S14 -B1 -U0. Reads that mapped to alternative splice variants of one gene were merged.
The mRNA and RNC-mRNA in each sample were normalized by RPKM [36] (reads per kilo base per million reads), while relative abundance between two groups was normalized by the edgeR package [37].
Differentially expressed mRNA and RNC-mRNA were identified via the edgeR package with a |fold-change| cutoff of ≥ 2 and FDR cutoff of < 0.01； |fold-change| cutoff of ≥ 2 and FDR cutoff of < 0.05. The TR of a gene of one sample was calculated as previously described: the quotient of the RNC-mRNA RPKM and the mRNA RPKM. Differentially regulated TR was calculated by a t-test with a |fold-change| cutoff of ≥ 2 and p-value cutoff of < 0.05. GO and KEGG were conducted by the Metascape software for differentially expressed genes.

Cell culture and transient transfection of siRNA
HepG 2 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% of fetal bovine serum and 1% of antibiotic under an atmosphere of 5% of CO 2 at 37 °C [33]. When there was an 80% confluence, the cells were seeded into 24-or 6-well plates. After GGCATGGACTGTGGTCATGAG.

Statistical analysis
Results of experiments were presented as mean ± the standard deviation (SD). The Student's t-test was employed to determine the significance of the difference between two groups, and the one-way analysis of variance (ANOVA)for more than two groups. P-value < 0.05 was considered significant. Figure 1 shows the experimental design of the present study.

Betaine significantly improved hepatic steatosis in HFD NAFLD mice
After a feeding period of 17 weeks, compared with NFD group, the liver TG level of mice in HFD group increased by twice, which indicates the successful establishment of NAFLD model induced by high fat diet.
However, compared with the HFD group, the TG level in the liver of the HB group significantly decreased ( Figure 2A). In addition, this result was confirmed by oil red O staining of liver sections ( Figure 2B). This indicates that betaine does alleviate fat deposition induced by high-fat diet.

Overview of RNA-seq and RNC-seq in HFD and HB group mice livers
To investigate the gene expression regulation associated with phenotypes at the translatome and transcriptome levels, RNA-seq and RNC-seq were performed on HFD group and HB group mice liver. The identification and quantification information on transcriptome and translatome were given in Table S1−2. Principal component analysis shows that there is a high correlation between mRNA/RNC-mRNA abundance among the two biological repeats of each group, which indicates the reliable reproducibility of our analysis. More importantly, the difference obtained by RNC-seq is much larger than the difference obtained by RNA-seq between HFD group and HB group, which demonstrates that the translatome can reflect the difference between treatments better than the transcriptome ( Figure 2C). About 16,000 genes were identified by RNAseq and about 10,000 genes were identified by RNC-seq in HFD and HB groups ( Figure 2D). When RNA-seq is compared with RNC-seq dataset, the majority of genes overlap ( Figure 2D), which implies that most mRNAs have entered the translation process. The abundance of RNC-mRNAs in HFD group and HB group have a high correlation with mRNA (R 2 = 0.5918 and 0.4751, p < 0.001; Figure 2E), which suggests that the transcriptome and translatome can be combined for analysis. For single transcription level or single translation level, the gene expression abundance of HFD group is highly correlated with that of HB group (R 2 = 0.8235 and 0.7271, p < 0.001; Figure S1A). The abundance distributions of RNC-mRNAs in HFD group and HB group are approximately lognormal ( Figure 2F).
Similar distributions are observed when it comes to the abundance distributions of mRNAs (Figures S1B).

HB group mice livers
In order to identify the crucial functional genes related to phenotype, transcriptional and translational differentially expressed genes (DEGs) were screened by edge R method, respectively, with thresholds of |log2 Another example shows the opposite situation, in which the mRNA level of Upf3b has decreased and the RNC-mRNA level has increased ( Figure   4F). These results imply that transcriptional regulation and translational regulation play relatively independent roles in gene expression regulation.

HB-feeding Generates Increased TR
Translation ratio (TR) indicates the proportion of mRNA entering the translation process, which is also called translation initiation efficiency. TR value is the quotient of RNC-mRNA abundance and mRNA abundance.
The mean mRNA abundance of HB group is slightly lower than that of HFD group (log2 mean mRNA RPKM of HFD = 1.368 and log2 mean mRNA RPKM of HB=0.9136, p < 0.0001; Figure 5A), the mean RNC-mRNA abundance in HB group is slightly higher than that in HFD group (log2 mean RNC-mRNA RPKM of HFD = 3.103 and log2 mean mRNA RPKM of HB =3.285, p < 0.0001; Figure 5A). Average TR of HB group is slightly higher than that of HFD group (log2 average TR of HFD=-0.1127, log2 average TR of HB = 0.1975, p<0.0001; Figure 5B). Similarly, for TR, the ratio of genes with higher TRs (log2 TR ≥ 1) in HB group is higher than that in HFD group ( Figure 5C). In addition, compared with HFD group, HB group has 767 differentially up-regulated TRs genes (DU-TR; log2 fold change>1, p<0.05), and 296 differentially down-regulated TRs genes (DD-TR; log2 fold change<-1, p<0.05). The genes corresponding to differentially up-regulated TR far exceed those differentially downregulated ( Figure 5D). All these results show that the translation ratio of HB group has increased.
Next, GO and KEGG enrichment analysis of differentially regulated TRs genes was conducted, and GO results show that it is related to translation, ribosome, ribosomal subunit and rRNA binding (classic Fisher <0.01; Figure S2A); KEGG results show that it is related to Ribosome and RNA transport (p < 0.05; Figure S2B). These results suggest that betaine may be involved in the regulation of lipid metabolism by affecting the mRNA translation process.

The Relationship Between TR and mRNA/RNC-mRNA Abundance
Effort was made to explore whether the abundance of mRNA and RNC-mRNA and the length of mRNA would affect TR. After statistical analysis, it is found that TR is weakly correlated with mRNA or RNC-mRNA abundance in HFD group and HB group (respectively,R 2 =0.1537 and 0.0357, Figure S2C; R 2 =0.1976 and 0.1431, Figure S2D； both p<0.001).
TR has no high correlation with mRNA length in either HFD group or HB group (R 2 =0.0046 and 0.0007 respectively; Figure S2E). In the same way, fold changes of TRs (HB/HFD) are still not highly correlated with mRNA length (R 2 =0.0104；; Figure S2F). This indicates that the abundance of mRNA and RNC-mRNA and the length of mRNA have little contribution to translation regulation.  Figure   5D). In the genes with non-DRTRs, there is a Kozak sequence with high probability near the start codon ( Figure 5E). Common sequences are common among genes with DU-and DD-TRs, especially those with DU-TRs ( Figure 5E). However, DD-TRs have some variations. For example, the base at -9 changed from "G" to "U"; the base at -7 changed from "C" to "U"; and the base at -6 changed from "G" to "A". In other words, it is due to the effects of betaine on translation level regulation that there are more differentially up-regulated TR genes than differentially downregulated TR genes ( Figure 5D), and that the average TR value of HB is higher than HFD group ( Figure 5B). More specifically, nucleotides at positions -9, -7 and -6 in Kozak sequence region attached to start codon change from "GC" to "AU" (Figure 5E), which may affect the process of mRNA translation.

Relationship Between Translational Regulation and Transcriptional
Regulation TR value represents the ratio of mRNA entering translation initiation, then the differential expression of RNC-mRNA abundance represents the result of gene regulation at transcription level and translation initiation level.
Next, efforts were made to explore whether transcriptome regulation and translation initiation will affect the expression regulation of RNC-mRNA.
As shown in Figure 6A, the genes were divided into nine quadrants according to the relationship between the fold change of mRNA at the transcription level and the fold change of TR at the translation initiation level. It is noticed that in e quadrant (-1<log2 fold change of mRNA<1, -1 < log2 fold change of TR < 1), there is no differential regulation of transcriptome and TR, and only a few DE RNC-mRNAs fall in quadrant e.
In other words, the vast majority of DE RNC-mRNAs fall into the quadrant of transcriptome differential regulation or TR differential regulation. This indicates that transcription regulation and translation initiation work together to change the abundance of RNC-mRNAs.
As mentioned above, the correlation between TR and mRNA abundance in HB group and HFD group is rather weak ( Figure S2C). However, as shown in Figure

Selection and Verification of Potential Functional Genes
The RNA-seq and RNC-seq data of the NFD group obtained in our previous study were used to screen candidate functional genes [30].
According to the threshold | log2 fold change | > 1 and FDR < 0.01, 31 transcriptional DEGs and 106 translational DEGs respectively were selected in the comparison of HFD group and NFD group. As mentioned in Figure Figure   7A). For transcriptional DEGs, only 7 genes are shared by the different genes in HFD group compared with NFD group and HB group compared with HFD (right in Figure 7A). Furthermore, for HB group compared with HFD group and HFD group Compared with NFD group, the heat maps level [23]. The beneficial effects of betaine on NAFLD are associated with the reduction of hepatic oxidative stress, inflammation and apoptosis, and the enhancement of Akt/mTOR signaling and autophagy [42]. This study also proves that the TG level in the liver of mice in the HB group has significantly decreased compared with that in the HFD group (Figure 2A and B).
Betaine, as a natural methyl donor, has certain influence on the m 6 A methylation apparent modification of mRNA. It has been reported that m 6 A readers may cause different downstream effects by recognizing m 6 A apparent modification [43][44][45]. The binding of YTHDF2 protein to m 6 A site is related to shortening the half-life of mRNA, and can maintain the m 6 A modification of 5'UTR region, thus promoting nonclassical mRNA translation. Therefore, it is worthy to study the effect of betaine on mRNA translation. Translatome, as a new technology of omics research in recent years, may provide important information on many biological problems for further research. It has been reported that RNC-seq can detect lncRNA [46][47][48], circRNA [49][50][51] and pri-miRNA [52] which encode polypeptide.
These RNA molecules, considered as "non-coding" in the past, contain one or more small open reading frames (small ORFs), which can be translated into small peptides with less than 100 amino acids. Therefore, RNC-seq was conducted for the phenotype that betaine can alleviate hepatic steatosis.
In transcriptome and translatome analysis, it is found that translation level can reflect the difference between different treatments more significantly  Figure S2A). Furthermore, the relationship between the length of mRNA and TR was analyzed, and the results shows that the length of mRNA is not related to the translation ratio ( Figure S2C).

Conclusion
The results that include triglyceride measurement and oil red O staining demonstrated that betaine improved HFD-induced hepatic steatosis.
Through the integrative application of hepatic transcriptome and translatome, the pharmacological mechanism of betaine and a potential

Consent for publication
Not applicable

Availability of data and materials
All data generated or analysed during this study are included in this published article and its supplementary information files.          Figure 1 Overview of experimental strategy For HFD group and HB group, the relationships between two biological repeats obtained by RNA-seq and