Figures
Abstract
N6-methyladenosine (m6A) is a ubiquitous reversible epigenetic RNA modification that plays an important role in the regulation of post-transcriptional protein coding gene expression. Liver is a vital organ and plays a major role in metabolism with numerous functions. Information concerning the dynamic patterns of mRNA m6A methylation during postnatal development of liver has been long overdue and elucidation of this information will benefit for further deciphering a multitude of functional outcomes of mRNA m6A methylation. Here, we profile transcriptome-wide m6A in porcine liver at three developmental stages: newborn (0 day), suckling (21 days) and adult (2 years). About 33% of transcribed genes were modified by m6A, with 1.33 to 1.42 m6A peaks per modified gene. m6A was distributed predominantly around stop codons. The consensus motif sequence RRm6ACH was observed in 78.90% of m6A peaks. A negative correlation (average Pearson’s r = -0.45, P < 10−16) was found between levels of m6A methylation and gene expression. Functional enrichment analysis of genes consistently modified by m6A methylation at all three stages showed genes relevant to important functions, including regulation of growth and development, regulation of metabolic processes and protein catabolic processes. Genes with higher m6A methylation and lower expression levels at any particular stage were associated with the biological processes required for or unique to that stage. We suggest that differential m6A methylation may be important for the regulation of nutrient metabolism in porcine liver.
Citation: He S, Wang H, Liu R, He M, Che T, Jin L, et al. (2017) mRNA N6-methyladenosine methylation of postnatal liver development in pig. PLoS ONE 12(3): e0173421. https://doi.org/10.1371/journal.pone.0173421
Editor: Yu Xue, Huazhong University of Science and Technology, CHINA
Received: November 1, 2016; Accepted: February 10, 2017; Published: March 7, 2017
Copyright: © 2017 He et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The data discussed in this publication are available from the GEO database (accession number GSE87327).
Funding: This work was supported by grants from the National High Technology Research and Development Program of China (863 Program) (2013AA102502), Sichuan Provincial Department of Science and Technology Program (2015JQO023), the National Natural Science Foundation of China (31530073, 31522055, 31472081 and 31401073), the National Program for Support of Top-notch Young Professionals, the Program for Innovative Research Team of Sichuan Province (2015TD0012), and the Fund of Fok Ying-Tung Education Foundation (141117). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Over 100 types of chemical modification to RNA have been described [1], most of which are formed by specific enzymatic modification of the primary RNA transcript during the tRNA complex maturation process [2, 3]. N6-methyladenosine (m6A) is one of the most prevalent modifications of eukaryotic mRNAs [4] with conserved topology across yeast [5], Arabidopsis thaliana [6, 7], Drosophila [8], mouse and human [9, 10]. Most of the m6A sites share a similar consensus m6A motif, RRm6ACH, where R represents a purine and H represents a non-guanine base [9]. It has been estimated that over one-third of genes in mouse and human transcriptomes are m6A methylated [9], and this figure rises to over 70% in Arabidopsis [7]. Transcriptome-wide analysis of m6A in mouse and human shows m6A sites preferentially appearing at distinct landmarks, around stop codons and within long internal exons [9, 10]. In Arabidopsis, m6A sites are also located immediately following transcription start sites (TSS) [6], which is thought to be the main difference between m6A patterns in plants and animals.
Recent evidences show that m6A methylation is involved in vast aspects of RNA metabolism in mammals [11–13], and has diverse characteristics in cells (HEK293T and embryonic stem cells) and tissues (brain and liver) [9, 10, 14]. Genes with m6A methylation mainly enriched in biologically important pathways, such as regulation of gene expression, differentiation and metabolism [6, 7, 9]. Nonetheless, m6A methylation profiles and functions in tissue at postnatal developmental stages have rarely been investigated.
The liver is a vital organ) with a wide range of functions, including nutrient metabolism, detoxification, protein synthesis, and the production of biochemicals necessary for digestion [15]. To investigate the developmental methylation changes of m6A in liver, we generated transcriptome-wide m6A methylation maps at three postnatal developmental stages that have distinct diets: newborn piglets (0 day old) receiving nutrients through sow placenta, suckling piglets (21 days old) fed with breast milk of the mother sow and adult multiparous sows (2 years old) fed with balanced artificial diet. We characterized the developmentally transcriptome-wide m6A distribution patterns and analyzed the relationship between gene expression and m6A modification. We also identified extensively m6A-modified genes which may contribute to the differences of potential biological functions among three developmental stages with the distinct nutrient conditions. These results provide a resource for identifying adenosine methylation modified mRNAs in liver and extended our knowledge of the role of m6A in development and growth of mammalian organs.
Materials & methods
Animals and tissue collection
Three female pigs (Rongchang pig, a Chinese indigenous breed) for each of three postnatal developmental stages (i.e., newborn piglets, suckling piglets at 21 days old, and adults at 2 years old) raised on a Rongchang pig elite reservation farm in Chongqing were used in this study. They have different sources of diets: newborn piglets receiving nutrients through the sow placenta, suckling piglets fed with breast milk of the mother sow and two-year-old adults fed with balanced artificial diets. Animals were humanely killed to ameliorate suffering by intravenous injection with 2% pentobarbital sodium (25mg/Kg). Liver from each of the animal was separated rapidly from each carcass, and immediately frozen in liquid nitrogen and stored at -80°C until use. All experimental procedures and sample collection in this study were approved by the Institutional Animal Care and Use Committee (IACUC) of Sichuan Agricultural University, under permit No. DKY-B20141401.
RNA preparation
High quality RNA from liver samples was isolated using Trizol according to the manufacturer’s instructions (Invitrogen). NanoDrop spectrophotometer (Thermo Scientific) was used to measure the concentration of RNA, and the integrity were tested by Agilent 2100 bioanalyzer. For isolation of poly(A) RNA, total RNA was subjected to two rounds of purification using oligo(dT)-coupled magnetic beads according to the manufacturer’s instructions (Ambion). Then, mRNA concentration was measured by Qubit 2.0 (Invitrogen) and the integrity was tested by agarose gel electrophoresis and Agilent 2100 bioanalyzer.
The mRNA was chemically fragmented approximately 150-nucleotide-long using NEBNext® Magnesium RNA Fragmentation Module (NEB #E6150S) according to the manufacturer's protocol. Standard ethanol precipitation was performed to precipitate the fragmented RNA. Size distribution of the RNA fragments was evaluated by agarose gel electrophoresis.
RNA immunoprecipitation (RIP)
About 5 μg fragmented mRNA was subjected to immunoprecipitation, according to the reported MeRIP method [16]. Briefly, fragmented RNA was incubated for 2 h at 4°C with 10 μg m6A antibody (Synaptic Systems Cat. No.202003, diluted to 0.5 mg/ ml) in 1,000 μl RIP buffer (50 mM Tris-HCl, 750 mM NaCl and 0.5% Igepal CA-630), supplemented with 2 mM RVC (Sigma) and 200 U RNasin (Promega). To reduce nonspecific binding, protein-A beads were pre-blocked in 1,000μl RIP buffer with 0.5 mg/ml BSA for 2h at 4°C. The pre-blocked protein-A beads were then incubated with above mixture for another 2 h at 4°C. The beads were vigorously washed using 1,000 μl RIP buffer three to four times. Discard the RIP buffer. Add 300 μl dilution buffer (10 mM Tris-HCl pH 7.5) into the bead tube and incubate at 50°C for 90 min. Eluted RNA was precipitated by ethanol-NaAc solution and glycogen (Life Technologies) overnight at -80°C. The eluted RNA was treated with RNasin (Promega) according to the manufacturer’s instructions.
Library preparation and high-throughput sequencing
NEBNext® Ultra™ RNA Library Prep Kit for Illumina® (E7530L, New England Biolabs) was used to construct the libraries from immunoprecipitated RNA and input RNA. In case that adaptor-ligated DNA population was evident in the library, size selection was performed using AMPure XP Beads according to the manufacturer’s instructions. Successful preparation of each library before proceeding to massively parallel deep sequencing was confirmed by Agilent 2100 Bioanalyzer and RT-PCR. Paired-end sequencing using standard 150 nucleotides read size was done with Illumina HiSeq 4000 sequencing platform. Raw sequencing data was processed by the Illumina base-calling pipeline.
Data analysis
Adaptor and low-quality bases were trimmed with Skewer (version: 0.1.126) [17], and the clean reads were aligned to the reference pig genome (Sscrofa10.2) downloaded from Ensemble (www.ensembl.org) with BWA MEM (version:0.7.12) [16]. Duplicated reads were marked with Samblaster (version: 0.1.22) [18]. Only the uniquely mapped (MAPQ ≥ 13) and non-duplicated alignments were used for peak calling. MACS2 (version: 2.1.0.20150420) [19] was employed to perform peak calling with a threshold of q 0.05.
Peaks with ≥ 50% length overlap in at least two biological replicates were defined as high-confidence peaks and used for further analysis. The 101 nucleotides centered on the summits detected by MACS2 were used for detection of the consensus m6A motif by DREME (version: 4.10.2) [20]. Motif central enrichment was performed by CentriMo (version: 4.10.2) [21] with 301 nucleotides centered on the summits. To compare the positional distribution of the motif in the peaks, the top three RRm6ACH motifs and one false positive sequence are shown.
Fragments with no less than 50% partial overlap with peaks or with no less than 50% overlap with peak bases were counted with bedtools (version: 2.25.0) [22] and normalized to the total as fragments per million (FPM). The immunoprecipitation FPM was divided by the input FPM to calculate the signal enrichment of the peaks. Differential methylation was determined by Student’s t test (P < 0.05) in two stages. Higher methylation peaks of one stage was defined as the peaks with log2-transformed fold changes of peak enrichment > 0 or < 0, compared to the other two stages (Student’s t test, P < 0.05), plus peaks uniquely found in one stage. Gene expression was calculated by featureCounts (version:1.5.0-p3) [23] using input unique alignment reads. Differential expression was determined by DESeq2 (version: 1.12.4) with P < 0.05 [24]. Higher expression genes were defined as genes with FPKM (reads per kilo base of exon model per million mapped reads) log2-transformed fold changes > 0 or < 0, compared to other two stages (Student’s t test, P < 0.05). GO analysis was performed using DAVID (Database for Annotation, Visualization and Integrated 16 Discovery, version: 6.8) web server (https://david-d.ncifcrf.gov/) [25].
Results
MeRIP-seq summary
We performed a transcriptome-wide survey of m6A methylation in porcine liver at three developmental stages using the MeRIP-Seq technique [9, 16]. A total of 18 libraries consisting of three replicates of input and MeRIP samples from the three stages were sequenced (S1 Table). An average of 9.33 giga base-pair (Gb) of high-quality data for each MeRIP library and 7.67 Gb for each input library were generated. After removing reads aligned to multiple positions of the pig genome and duplicated reads derived from PCR artifacts, an average of 6.70 Gb for each MeRIP library and 5.86 Gb for each input library uniquely aligned to reference pig genome (Sscrofa 10.2). Reads of paired input and MeRIP libraries were used to identify peaks. For all three replicates, 32,661 distinct narrow m6A peaks from newborn, 25,921 from suckling and 28,848 from adult stages were successfully detected in liver transcriptomes, which harbor an average of 13,578 transcribed genes (S1 Table).
General features of m6A methylation
After merging the three replicates, we identified a total of 11,022, 8,727 and 9,860 distinct peaks in newborn, suckling and adult stages, respectively. On average, over 80% of the identified peaks were consistently detected in at least two biological replicates of each stage (S1A Fig) which confirmed the high reproducibility of MeRIP-seq among biological replicates (S1B Fig). We detected 8,855 high-confidence m6A peaks in newborn, 7,350 in suckling, and 7,961 in adult stages (S2 Table). We used the recurrent peaks as high-confidence m6A peaks for subsequent analysis (S3 Table). Consequently, 5,848 peaks were consistently detected within all three stages (Fig 1A). On average, 74.24% of the identified high-confidence peaks overlapped with intragenic regions, representing transcripts of 4,676, 4,103 and 4,339 genes in the newborn, suckling and adult developmental stages, respectively (S2 Table). For all three stages, 3,481 genes were consistently modified by m6A (Fig 1B and S4 Table). Our results showed that about one-third of expressed genes were modified by m6A, with 35.09%, 30.40%, and 32.61% of expressed genes (FPKM > 0.1) detected in the newborn, suckling and adult stages, respectively (S2 Table). The results also showed that the liver transcriptome contains about 1.33 to 1.42 m6A peaks per m6A modified genes (S2 Table).
(A) Venn diagram showing the overlap of m6A peaks in newborn (8,855), suckling (7,350) and adult (7,961). There are 5,848 common peaks among the three stages, which with ≥ 50% length overlap between stages. (B) Venn diagram showing the overlap of m6A modified genes. Respectively, 4,676 genes in newborn, 4,103 in suckling and 4,339 in adult were m6A methylated. For all three stages, 3,481 genes were consistently modified. (C) Proportion of genes containing variant numbers of m6A peaks. Majority of modified genes (74.60%) contain one or two m6A peaks, while the rest contains more. (D) Sequence logo representing the most common consensus motif (RRm6ACH) in the m6A peaks. The consensus sequence was detected by DREME (version: 4.10.2), using the 101 nucleotides centered on the summits of called original narrow peaks.
The number of m6A modified sites varied from 1 to 14 among individual genes, while 74.60% of modified genes harbor only one or two m6A peaks. The remaining genes (25.40%) contain three or more peaks (Fig 1C), which is much higher than that previously reported for human brain (16.70%) [10].
The classic consensus sequence, RRm6ACH, where R represents a purine and H represents a non-guanine base [26, 27] was found in most (78.90%) of the detected narrow peaks (S5 Table). The consensus sequence (Fig 1D) observed in the current study indicated conserved m6A methylation among different species. The most frequent two motifs were GGm6ACC (21.99%) and GGm6ACT (21.80%) (Fig 1D). We further performed a motif central enrichment analysis using CentriMo (version: 4.10.2) [21], and found motifs detected in peaks were mainly enriched around the summits of called narrow peaks and at the center of merged peaks (S2 Fig).
Topological pattern of m6A methylation
To understand the topological pattern of m6A methylation in liver transcriptome, we investigated the distribution profiles of m6A peaks. Consistent with previous studies in mouse and human [6, 9], we observed a significant enrichment of m6A peaks around the start and stop codons of transcripts in all three stages (Fig 2A). To further confirm the preferential localization of m6A along transcripts, we categorized m6A peaks into five non-overlapping segments: TSS (200 nucleotides downstream of the TSS), 5' untranslated region (UTR), coding sequence (CDS), stop codon segment (a 400 nucleotide window centered on the stop codon) and 3' UTR. m6A peaks were most abundant in CDS (40.75% to 41.97%) and stop codon segments (26.03% to 28.67%), followed by the 3' UTR, TSS and then 5' UTR segments (Fig 2B). After segment normalization according to relative fraction of the transcriptome occupied by each segment, we observed that the 5' UTR and stop codon were the most enriched segments (Fig 2C).
(A) Distribution of summits of m6A peaks along transcripts. Each transcript was divided into three parts: -2Kb, CDS, +2Kb. Each part was divided into 100 bins, and the percentage of m6A summits of each bin was determined. Moving averages (4 bins) of summit percentage of newborn (red), suckling (green) and adult (blue) are shown. (B) Graphical representation of frequency of m6A peaks in five non-overlapping segments of three stages (TSS: 200 nucleotides downstream of the TSS, stop codon: a 400 nucleotide window centered on the stop codon). m6A peaks were most abundant in CDS and stop codon segments. (C) Top, relative enrichment of m6A peaks across transcript segments, normalized by the relative fraction that each segment occupies in the transcriptome. Bottom, schematic of the five segments. 5' UTR and stop codon were the most enriched segments after normalization.
Relationship between m6A methylation and gene expression
We next determined whether gene expression in porcine liver is correlated with the presence of m6A modification by plotting the fraction of genes with m6A peaks in each of the segments as a function of expression level (Fig 3A). Each segment showed a similar non-monotonic pattern as previously exhibited in a mouse and human study [9]. The non-monotonic pattern showed that most m6A modified genes were expressed at moderate levels (Fig 3A).
(A) Fraction of genes with m6A peaks in each of the segments as a function of expression level. Most of the modified genes were expressed at moderate levels. Genes expressed at the two extremes were less methylated. (B) Plot of m6A peak enrichment and mRNA abundance in the three stages. Obvious negative correlation between m6A peak enrichment and modified mRNA abundance was found (Pearson’s r = -0.47 to -0.42, P < 10−16). Lines represent the linear trend for the obtained values.
A plot of m6A peak enrichment level versus mRNA abundance revealed a negative correlation between enrichment and gene expression in all three stages (Pearson’s r = -0.47 to -0.42, P < 10−16) (Fig 3B), which is consistent with previous reports [10, 14]. A more detailed, distribution pattern-based analysis also showed negative correlations between m6A peak enrichment and gene expression for all five segments (S3 Fig). Among them, stop codon (average Pearson’s r = -0.50, P < 10−16) and CDS (average Pearson’s r = -0.47, P < 10−16) segments showed the highest negative correlation, and relatively lower correlation rates were found for 3' UTR (average Pearson’s r = -0.42, P < 10−16), 5' UTR (average Pearson’s r = -0.33, P < 10−7) and TSS (average Pearson’s r = -0.28, P < 10−6) segments (S3 Fig).
m6A modified genes involve in biologically important pathways
To assess the function of m6A in porcine liver, 3,481 genes that were consistently modified by m6A methylation in all three stages were subjected to gene functional enrichment analysis using DAVID tool (version: 6.8). As a result, these genes were significantly (P < 0.05, Benjamini-Hochberg corrected) enriched in a variety of cellular functions relevant to “RNA metabolic process” (951 genes), “regulation of transcription” (745 genes), “regulation of signal transduction” (581 genes) and “biosynthetic process” (1045 genes) (S6 Table). In addition, some genes were specifically involved in cell differentiation- and liver development- related categories, such as “regulation of cell differentiation” (333 genes), “hepaticobiliary system development” (42 genes) and “liver development” (40 genes) (Fig 4).
Some genes consistently modified by m6A methylation in all three stages were specifically involved in cell differentiation- and liver development- related categories. Different colors represent P values, and sizes represent gene numbers. X-axis represents fold enrichment. Detailed functional enrichment analysis results of all consistently modified genes, are available in S6 Table.
Differential m6A modification and gene expression
An average of 34.94% of m6A modified genes showed differential methylation, and an average of 17.19% of expressed genes were differentially expressed between two stages (Table 1). A paired analysis of differential methylation and differential expression showed that liver of newborn pigs had the largest m6A differential methylation ratio (average 25.58%) and the smallest differential expression ratio (average 4.76%) among the three stages, while liver of the suckling pigs showed an opposite extreme (Table 1). This result further confirmed the negative correlation between m6A methylation and gene expression.
Analysis of differentially m6A modified genes showed that 748 genes (16.00%) in newborn, 275 (6.70%) in suckling, and 280 (6.45%) in adult stages were more highly enriched for m6A peaks compared with the other two stages (S4 Fig). These genes exhibited much higher negative correlations between gene expression and m6A peak enrichment (average Pearson’s r = -0.56, P < 10−16) compared with correlations between overall peak enrichment and gene expression (S5 Fig). Based on gene functional enrichment analysis, genes showing higher m6A methylation at the newborn stage were significantly (P < 0.05) enriched to bile acid secretion and nutrients metabolic processes, such as “bile acid biosynthetic process” (5 genes), “oligosaccharide metabolic process” (5 genes) and “cholesterol metabolic process” (7 genes) GO categories, “alanine, aspartate and glutamate metabolism” (5 genes), “biosynthesis of unsaturated fatty acids” (4 genes) and “glycine, serine and threonine metabolism” (5 genes) pathways (Fig 5A). Genes with higher m6A methylation at the suckling stage were related to “glycosaminoglycan metabolic process” (6 genes), “UDP-N-acetylgalactosamine metabolic process” (2 genes) GO categories and the “oocyte meiosis” (5 genes) pathway (Fig 5B). Genes with higher m6A methylation at the adult stage were related to “fatty acid transport” (3 genes), “circadian rhythm” (9 genes) and “lysine degradation” (4 genes) categories or pathways (Fig 5C).
GO terms of genes showing a higher enrichment of m6A methylation in newborn (A), suckling (B) and adult (C). Different colors represent P values, and sizes represent gene numbers. X-axis represents fold enrichment.
Having shown a negative correlation between m6A methylation and gene expression, we next explored functions of genes with relatively high m6A methylation levels and low levels of expression. We found that 31 genes in the newborn stage and three in the adult stage simultaneously showed higher m6A methylation and lower expression compared with the other two stages (S7 Table). The genes in the newborn stage were potentially expressed for organic acid biosynthetic and metabolic processes, such as “oxoacid metabolic process”, “folic acid-containing compound metabolic process” and “glycine, serine and threonine metabolism” (Table 2). Typically, GATM shows the highest methylation at the newborn stage and the lowest at the adult stage (Fig 6), which mainly involves in amino acid and organic acid metabolism. In contrast, the lowest expression of GATM was at the newborn stage and highest at the adult stage. Only three genes (EDEM2, MAFK, UBALD2) were detected with higher m6A methylation and lower expression at the adult stage. These genes involved in “mannosyl-oligosaccharide 1,2-alpha-mannosidase activity” (EDEM2) and “transcription factor activity” (MAFK) (Table 2).
Opposite trends of the m6A methylation level (left panel) and gene expression level (right panel) of GATM are shown. Gene expression level is presented by the accumulation of input reads.
Discussion
This study reports comprehensive transcriptome-wide patterns of m6A in porcine liver, based on a previously reported MeRIP-seq method [9, 16]. We report abundant m6A sites in the porcine liver transcriptome with a density of 1.33–1.42 site per gene, which is comparable with that obtained in mouse liver (~1.34 m6A sites per coding gene). We profiled features and patterns of m6A, including the extent of m6A gene modification, m6A distribution in transcripts, and occurrence of the consensus m6A methylation motif. All showed high concordance with m6A characteristics of previous reports [9, 10], indicating conserved RNA adenosine methylation between pig and other species.
Most of the m6A modified genes were expressed at a medium level, and a negative correlation was found between m6A peak enrichment and gene expression (Pearson’s r = -0.47 to -0.42, P < 10−16). m6A is a chemical mark associated with transcript turnover. m6A-marked transcripts have a significantly shorter RNA half-life and increased rates of mRNA decay compared to non-m6A-marked transcripts [14]. Meanwhile, a high level of m6A methylation may endow transcripts expressed at low levels with high RNA stability, or provide a signal for reader protein binding [9, 44]. In addition, we noted that over 25% of m6A-marked genes harbor three or more m6A sites, which may increase RNA stability or probability targeted by m6A readers. These results indicate that m6A methylation mediates a level of post-transcriptional regulation of gene expression.
We found that genes involved in hepatic cell differentiation and liver development were consistently modified by m6A methylation in the three postnatal developmental stages. Previous studies have analogous results. For instance, m6A methylation in embryonic stem cells regulates core pluripotency factors involved in development and differentiation [14]. Depletion of m6A levels results in embryonic stem cells advancing from self-renewal toward differentiation to specific lineages [14, 45, 46]. In Arabidopsis, a reduction of m6A can impair early embryonic development at the globular stage [47]. Recent studies of the differential methylation of m6A in three different organs of Arabidopsis revealed that m6A is an important contributor to organ differentiation [7]. Therefore, m6A may be an important conserved regulator of cell differentiation and development in both animals and plants.
The liver plays an important role in metabolism, with numerous functions including the regulation of glycogen storage, plasma protein synthesis, decomposition of red blood cells, hormone production, and detoxification [15]. In this study, we found that at any one stage, genes with higher levels of m6A methylation had metabolic functions required for or specific to this stage. The newborn liver mainly metabolizes nutrients obtained through the placenta from the sow and synthesizes macromolecules that are necessary for rapid growth [41–43]. Functional enrichment analysis of genes with higher m6A methylation at this stage produced terms such as “metabolic pathways”, “biosynthesis of unsaturated fatty acids”, “bile acid biosynthesis process” and “oligosaccharide metabolic process”. In the suckling stage, the liver mainly metabolizes nutrients from breast milk, which is enriched in carbohydrates and protein, including total solids, fat, lactose, total protein, total whey protein and individual immunoglobulin classes [48]. Interestingly, genes in this stage with higher m6A methylation were enriched for the terms “UDP-N-acetylgalactosamine metabolic process”, “glycosaminoglycan metabolic process” and other monosaccharide or glycoprotein metabolic processes. In the adult stage, the liver has a fully developed hepaticobiliary system compared with the other two stages [49]. Nutrients at this stage are mainly derived from a nutritionally balanced artificial diet [50], which contains mainly starch, protein, fat, and some additives, such as lysine and threonine [51, 52]. Genes with higher m6A methylation at this stage were mainly related to metabolic activities, such as “lysine degradation”, “regulation of catalytic activity”, “fatty acid transport” and “positive regulation of metabolic process”, which correspond to adult metabolic functions of the liver. In addition, terms such as “rhythmic process” and “circadian rhythm” were also enriched for highly m6A-modified genes in the adult stage, indicating circadian clock control of liver metabolic functions [53]. Therefore, highly m6A-methylated genes at a particular stage showed enriched terms that were consistent with liver function at this stage.
Genes with higher levels of m6A methylation and lower levels of expression also showed close association with metabolic activities in each stage. In newborn liver, genes were mainly involved in metabolic processes of organic acids, such as carboxylic acid, oxoacid and dicarboxylic acid, indicating that m6A methylation plays roles in balancing post-transcription expression levels of genes involved in central metabolic processes. For example, GATM and GNMT had higher levels of m6A methylation and lower levels of expression in newborn and are involved in amino acid metabolism [29] and S-adenosylmethionine (SAM) synthesis [54]. GATM encodes the first and rate-limiting product in creatine biosynthesis, and creatine and its phosphorylated form play essential roles in energy metabolism [55]. GNMT is a potential tumor suppressor that is commonly inactivated in human hepatoma. GNMT affects transmethylation kinetics and SAM synthesis by limiting homocysteine remethylation fluxes [54]. Cytochrome P450 family 1 subfamily A member 2 (CYP1A2) encodes a member of the cytochrome P450 superfamily of enzymes involved in the metabolism of a variety of compounds, including steroids, fatty acids, and xenobiotics [31]. m6A methylation of CYP1A2 was less and expression higher in suckling and adult stages, compared with the newborn. Expression of CYP1A2 appears to be induced by dietary constituents [56]. In suckling and adult stages, expression of CYP1A2 may be induced by various dietary constituents [56], especially the artificial diet in the adult stage, which contains a high content of polycyclic aromatic hydrocarbons (PAHs) [57, 58]. CYP1A2 mainly catalyzes conversion of PAHs to more polar and water-soluble metabolites, and the resultant metabolites are readily excreted from the body [59]. In the newborn stage, CYP1A2 expression is relatively low, as it is not induced by such dietary constituents. Higher m6A methylation may endow the relatively fewer transcripts with enough stability to perform their function. In the adult stage, the highly m6A modified and lowly expressed gene, EDEM2, mainly initiates ER-associated glycoprotein degradation by catalyzing the first mannose trimming step [60], which may play important roles in metabolism of glycoprotein of deriving from artificial diet.
In conclusion, we characterized diverse patterns of m6A in genes expressed in the porcine liver and showed that these genes act as important regulators in three developmental stages. The high negative correlation between levels of m6A methylation and modified gene expression suggests that adenosine methylation plays important biological roles in negative regulation of post-transcriptional gene expression. The m6A modified genes were mainly involved in regulation of differentiation and development of the porcine liver. As growing conditions and diets change in the three developmental stages, the liver undergoes different stimuli and nutrient levels, which may influence the differential expression of the transcriptome and differential m6A methylation of the epitranscriptome.
Supporting information
S1 Fig. High reproducibility of MeRIP-seq.
(A) Distribution of peaks in three biological replicates across three stages. On average, 80% of peaks were shared by at least two replicates. (B) Heat map of Pearson’s correlation of read count of transcripts in both immunoprecipitation (MeRIP) and input data across nine samples.
https://doi.org/10.1371/journal.pone.0173421.s001
(TIF)
S2 Fig.
Central enrichment of consensus RRm6ACH motif sequences around m6A peak summits (A) and peak center of merged m6A peaks (B). Top three consensus RRm6ACH motif sequences (GGACT/A/C) and one false positive sequence (GCAGC) were discovered by DREME, using the 101 nucleotides centered on the summits of called original narrow peaks. Motif central enrichment was performed by CentriMo (version: 4.10.2) with 301 nucleotides centered on the summits or peak center of merged m6A peaks. Each curve shows the density (averaged over bins of 40 bp width) of the best strong site (score ≥ 5 bits) for the named motif at each position in the m6A peak regions (301 bp). The legend shows the motif, its central enrichment P-value, the width of the most enriched central region (w), and the number of peaks (n out of 70,131 summits in (A), n out of 8,379 merged peaks in (B)) that contain a motif site. This similar tendency of central enrichment of RRm6ACH motifs suggested that we used a reliable merging process to deal with peaks in multiple biological replicates and groups.
https://doi.org/10.1371/journal.pone.0173421.s002
(TIF)
S3 Fig. Plot of m6A peak enrichment and mRNA abundance in five non-overlapping segments.
Higher negative correlation rates were found in stop codon (average Pearson’s r = -0.50, P < 10−16) and CDS (average Pearson’s r = -0.47, P < 10−16) peaks compared with UTR and TSS peaks. Lines represent the linear trend for the obtained values.
https://doi.org/10.1371/journal.pone.0173421.s003
(TIF)
S4 Fig.
Venn diagram of paired comparison of genes with higher methylation among stages (A), paired comparison of genes with lower expression among stages (B), and overlap of genes with higher methylation and lower expression in each stage (C). “N” represents newborn, “S” represents suckling and “A” represents adult. “>” represents higher methylation between stages and “<” indicates lower gene expression.
https://doi.org/10.1371/journal.pone.0173421.s004
(TIF)
S5 Fig. Plot of m6A enrichment and mRNA abundance of genes with higher m6A methylation compared with each of the other two stages.
A higher negative correlation rate was found in all three stages (average Pearson’s r = -0.56, P < 10−16).
https://doi.org/10.1371/journal.pone.0173421.s005
(TIF)
S1 Table. Summary of sequenced and mapped data of the MeRIP-Seq and input RNA-seq samples.
https://doi.org/10.1371/journal.pone.0173421.s006
(DOCX)
S2 Table. Number of expressed genes, merged peaks and proportion of m6A modified transcripts in each group.
https://doi.org/10.1371/journal.pone.0173421.s007
(DOCX)
S3 Table. Lists of m6A peaks in three stages.
https://doi.org/10.1371/journal.pone.0173421.s008
(XLSX)
S4 Table. Persistently m6A modified genes in the three developmental stages.
https://doi.org/10.1371/journal.pone.0173421.s009
(XLSX)
S5 Table. Diverse patterns of m6A motif sequences (RRm6ACH).
https://doi.org/10.1371/journal.pone.0173421.s010
(DOCX)
S6 Table. Functional categories of consistently modified genes in porcine liver of the three developmental stages.
https://doi.org/10.1371/journal.pone.0173421.s011
(XLSX)
S7 Table. Genes showing higher m6A peak enrichment and lower expression levels than the other two stages.
https://doi.org/10.1371/journal.pone.0173421.s012
(XLSX)
Author Contributions
- Conceptualization: SH ML DL HW.
- Formal analysis: SH LJ TC ST YL HL.
- Investigation: SH MH RL DL.
- Resources: SH HW LD ZJ.
- Writing – original draft: SH.
- Writing – review & editing: ML XL DL ZJ.
References
- 1. Machnicka MA, Milanowska K, Osman Oglou O, Purta E, Kurkowska M, Olchowik A, et al. MODOMICS: a database of RNA modification pathways—2013 update. Nucleic Acids Res. 2013;41(Database issue):D262–267. pmid:23118484
- 2. Boschi-Muller S, Motorin Y. Chemistry enters nucleic acids biology: enzymatic mechanisms of RNA modification. Biochemistry (Mosc). 2013;78(13):1392–404.
- 3. Urban A, Behm-Ansmant I, Branlant C, Motorin Y. RNA sequence and two-dimensional structure features required for efficient substrate modification by the Saccharomyces cerevisiae RNA:{Psi}-synthase Pus7p. J Biol Chem. 2009;284(9):5845–5858. pmid:19114708
- 4. Gilbert WV, Bell TA, Schaening C. Messenger RNA modifications: form, distribution, and function. Science. 2016;352(6292):1408–1412. pmid:27313037
- 5. Bodi Z, Bottley A, Archer N, May ST, Fray RG. Yeast m6A methylated mRNAs are enriched on translating ribosomes during meiosis, and under rapamycin treatment. PLoS ONE. 2015;10(7):e0132090. pmid:26186436
- 6. Luo GZ, MacQueen A, Zheng G, Duan H, Dore LC, Lu Z, et al. Unique features of the m6A methylome in Arabidopsis thaliana. Nat Commun. 2014;5:5630. pmid:25430002
- 7. Wan Y, Tang K, Zhang D, Xie S, Zhu X, Wang Z, et al. Transcriptome-wide high-throughput deep m(6)A-seq reveals unique differential m(6)A methylation patterns between three organs in Arabidopsis thaliana. Genome Biol. 2015;16:272. pmid:26667818
- 8. Levis R, Penman S. 5'-terminal structures of poly(A)+ cytoplasmic messenger RNA and of poly(A)+ and poly(A)- heterogeneous nuclear RNA of cells of the dipteran Drosophila melanogaster. J Mol Biol. 1978;120(4):487–515. pmid:418182
- 9. Dominissini D, Moshitch-Moshkovitz S, Schwartz S, Salmon-Divon M, Ungar L, Osenberg S, et al. Topology of the human and mouse m6A RNA methylomes revealed by m6A-seq. Nature. 2012;485(7397):201–206. pmid:22575960
- 10. Meyer KD, Saletore Y, Zumbo P, Elemento O, Mason CE, Jaffrey SR. Comprehensive analysis of mRNA methylation reveals enrichment in 3' UTRs and near stop codons. Cell. 2012;149(7):1635–1646. pmid:22608085
- 11. Maity A, Das B. N6-methyladenosine modification in mRNA: machinery, function and implications for health and diseases. The FEBS J. 2015.
- 12. Yue Y, Liu J, He C. RNA N6-methyladenosine methylation in post-transcriptional gene expression regulation. Genes Dev. 2015;29(13):1343–1355. pmid:26159994
- 13. Liu N, Pan T. N6-methyladenosine-encoded epitranscriptomics. Nat Struct Mol Biol. 2016;23(2):98–102. pmid:26840897
- 14. Batista PJ, Molinie B, Wang J, Qu K, Zhang J, Li L, et al. m6A RNA modification controls cell fate transition in mammalian embryonic stem cells. Cell stem cell. 2014;15(6):707–719. pmid:25456834
- 15. Abshagen K, Kuhla A, Genz B, Vollmar B. Anatomy and physiology of the hepatic circulation. PanVascular Medicine. 2015:3607–3629.
- 16. Dominissini D, Moshitch-Moshkovitz S, Salmon-Divon M, Amariglio N, Rechavi G. Transcriptome-wide mapping of N6-methyladenosine by m6A-seq based on immunocapturing and massively parallel sequencing. Nat Protoc. 2013;8(1):176–189. pmid:23288318
- 17. Jiang H, Lei R, Ding SW, Zhu S. Skewer: a fast and accurate adapter trimmer for next-generation sequencing paired-end reads. BMC Bioinformatics. 2014;15:182. pmid:24925680
- 18. Faust GG, Hall IM. SAMBLASTER: fast duplicate marking and structural variant read extraction. Bioinformatics. 2014;30(17):2503–2505. pmid:24812344
- 19. Zhang Y, Liu T, Meyer CA, Eeckhoute J, Johnson DS, Bernstein BE, et al. Model-based analysis of ChIP-Seq (MACS). Genome Biol. 2008;9(9):R137. pmid:18798982
- 20. Bailey TL. DREME: motif discovery in transcription factor ChIP-seq data. Bioinformatics. 2011;27(12):1653–1659. pmid:21543442
- 21. Bailey TL, Machanick P. Inferring direct DNA binding from ChIP-seq. Nucleic Acids Res. 2012;40(17):610–621.
- 22. Quinlan AR. BEDTools: The swiss-army tool for genome feature analysis. Curr Protoc Bioinformatics. 2014;47(11.12.11):11.2.1–.2.34.
- 23. Liao Y, Smyth GK, Shi W. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics. 2014;30(7):923–930. pmid:24227677
- 24. Anders S, Huber W. Differential expression of RNA-Seq data at the gene level–the DESeq package. Embl. 2013.
- 25. Huang da W, Sherman BT, Lempicki RA. Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res. 2009;37(1):1–13. pmid:19033363
- 26. Csepany T, Lin A, Baldick CJ Jr., Beemon K. Sequence specificity of mRNA N6-adenosine methyltransferase. J Biol Chem. 1990;265(33):20117–20122. pmid:2173695
- 27. Wei CM, Gershowitz A, Moss B. 5'-Terminal and internal methylated nucleotide sequences in HeLa cell mRNA. Biochemistry. 1976;15(2):397–401. pmid:174715
- 28. Huang da W, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc. 2009;4(1):44–57. pmid:19131956
- 29. Humm A, Fritsche E, Steinbacher S, Huber R. Crystal structure and mechanism of human L-arginine:glycine amidinotransferase: a mitochondrial enzyme involved in creatine biosynthesis. EMBO J. 1997;16(12):3373–3385. pmid:9218780
- 30. Riedel TJ, Johnson LC, Knight J, Hantgan RR, Holmes RP, Lowther WT. Structural and biochemical studies of human 4-hydroxy-2-oxoglutarate aldolase: implications for hydroxyproline Metabolism in primary hyperoxaluria. PLoS ONE. 2011;6(10):1510–1520.
- 31. Nelson DR, Zeldin DC, Hoffman SM, Maltais LJ, Wain HM, Nebert DW. Comparison of cytochrome P450 (CYP) genes from the mouse and human genomes, including nomenclature recommendations for genes, pseudogenes and alternative-splice variants. Pharmacogenetics. 2004;14(1):1–18. pmid:15128046
- 32. Pemberton TA, Srivastava D, Sanyal N, Henzl MT, Becker DF, Tanner JJ. Structural studies of yeast Δ(1)-pyrroline-5-carboxylate dehydrogenase (ALDH4A1): active site flexibility and oligomeric state. Biochemistry. 2014;53(8):1350–1359. pmid:24502590
- 33. Wahl B, Reichmann D, Niks D, Krompholz N, Havemeyer A, Clement B, et al. Biochemical and spectroscopic characterization of the human mitochondrial amidoxime reducing components hmARC-1 and hmARC-2 suggests the existence of a new molybdenum enzyme family in eukaryotes. J Biol Chem. 2010;285(48):37847–37859. pmid:20861021
- 34. Antony A, Tang YS, Khan RA, Biju MP, Xiao X, Li QJ, et al. Translational upregulation of folate receptors is mediated by homocysteine via RNA-heterogeneous nuclear ribonucleoprotein E1 interactions. J Clin Invest. 2004;113(2):285–301. pmid:14722620
- 35. Lee SL, Wallace SL, Barone R, Blum L, Chase PH. Familial deficiency of two subunits of the first component of complement. C1r and C1s associated with a lupus erythematosus-like disease. Arthritis Rheum. 1978;21(8):958–967. pmid:737019
- 36. Zahid H, Miah L, Lau AM, Brochard L, Hati D, Bui TT, et al. Zinc-induced oligomerization of zinc alpha2 glycoprotein reveals multiple fatty acid-binding sites. Biochem J. 2016;473(1):43–54. pmid:26487699
- 37. Leong SR, Baxter RC, Camerato T, Dai J, Wood WI. Structure and functional expression of the acid-labile subunit of the insulin-like growth factor-binding protein complex. Mol Endocrinol. 1992;6(6):870–876. pmid:1379671
- 38. Takahashi Y, Morales FC, Kreimann EL, Georgescu MM. PTEN tumor suppressor associates with NHERF proteins to attenuate PDGF receptor signaling. EMBO J. 2006;25(4):910–920. pmid:16456542
- 39. Feranchak AP, Doctor RB, Troetsch M, Brookman K, Johnson SM, Fitz JG. Calcium-dependent regulation of secretion in biliary epithelial cells: the role of apamin-sensitive SK channels. Gastroenterology. 2004;127(3):903–913. pmid:15362045
- 40. Seno M, Futami J, Tsushima Y, Akutagawa K, Kosaka M, Tada H, et al. Molecular cloning and expression of human ribonuclease 4 cDNA. Biochim Biophys Acta. 1995;1261(3):424–426. pmid:7742370
- 41. Merregaert J, Van Langen J, Hansen U, Ponsaerts P, El Ghalbzouri A, Steenackers E, et al. Phospholipid scramblase 1 is secreted by a lipid raft-dependent pathway and interacts with the extracellular matrix protein 1 in the dermal epidermal junction zone of human skin. J Biol Chem. 2010;285(48):37823–37837. pmid:20870722
- 42. Mast SW, Diekman K, Karaveg K, Davis A, Sifers RN, Moremen KW. Human EDEM2, a novel homolog of family 47 glycosidases, is involved in ER-associated degradation of glycoproteins. Glycobiology. 2005;15(4):421–436. pmid:15537790
- 43. Toki T, Itoh J, Kitazawa J, Arai K, Hatakeyama K, Akasaka J, et al. Human small Maf proteins form heterodimers with CNC family transcription factors and recognize the NF-E2 motif. Oncogene. 1997;14(16):1901–1910. pmid:9150357
- 44. Niu Y, Zhao X, Wu YS, Li MM, Wang XJ, Yang YG. N6-methyl-adenosine (m6A) in RNA: an old modification with a novel epigenetic function. Genomics Proteomics Bioinformatics. 2013;11(1):8–17. pmid:23453015
- 45. Wang Y, Li Y, Toth JI, Petroski MD, Zhang Z, Zhao JC. N6-methyladenosine modification destabilizes developmental regulators in embryonic stem cells. Nat Cell Biol. 2014;16(2):191–198. pmid:24394384
- 46. Zhao BS, He C. Fate by RNA methylation: m6A steers stem cell pluripotency. Genome Biol. 2015;16:43. pmid:25723450
- 47. Zhong S, Li H, Bodi Z, Button J, Vespa L, Herzog M, et al. MTA is an Arabidopsis messenger RNA adenosine methylase and interacts with a homolog of a sex-specific splicing factor. Plant Cell. 2008;20(5):1278–1288. pmid:18505803
- 48. Andrada AD, Cortés CC. Changes in the composition of sows' milk between days 5 to 26 of lactation. Span J Agric Res. 2004(3):333–336.
- 49. Lima JP. Anatomy and physiology of the liver secretory apparatus. Arq Gastroenterol. 1980;17(3):149–160. pmid:7016087
- 50. Aherne FX, Williams IH. Nutrition for optimizing breeding herd performance. Vet Clin North Am Food Anim Pract. 1992;8(3):589–608. pmid:1446272
- 51. Dourmad JY, Etienne M. Dietary lysine and threonine requirements of the pregnant sow estimated by nitrogen balance. J Anim Sci. 2002;80(8):2144–2150. pmid:12211384
- 52.
Council N. Nutrient Requirements of Swine: Eleventh Revised Edition. Washington, DC: The National Academies Press; 2012.
- 53. Reinke H, Asher G. Circadian Clock Control of Liver Metabolic Functions. Gastroenterology. 2016;150(3):574–580. pmid:26657326
- 54. Wang YC, Tang FY, Chen SY, Chen YM, Chiang EP. Glycine-N methyltransferase expression in HepG2 cells is involved in methyl group homeostasis by regulating transmethylation kinetics and DNA methylation. J Nutr. 2011;141(5):777–782. pmid:21411609
- 55. Zhu Y, Evans MI. Estrogen modulates the expression of L-arginine:glycine amidinotransferase in chick liver. Mol Cell Biochem. 2001;221(1–2):139–145. pmid:11506177
- 56. Fontana RJ, Lown KS, Paine MF, Fortlage L, Santella RM, Felton JS, et al. Effects of a chargrilled meat diet on expression of CYP3A, CYP1A, and P-glycoprotein levels in healthy volunteers. Gastroenterology. 1999;117(1):89–98. pmid:10381914
- 57. Phillips DH. Polycyclic aromatic hydrocarbons in the diet. Mutat Res. 1999;443(1–2):139–147. pmid:10415437
- 58. Vakharia DD, Liu N, Pause R, Fasco M, Bessette E, Zhang QY, et al. Effect of metals on polycyclic aromatic hydrocarbon induction of CYP1A1 and CYP1A2 in human hepatocyte cultures. Toxicol Appl Pharmacol. 2001;170(2):93–103. pmid:11162773
- 59. Shimada T. Xenobiotic-metabolizing enzymes involved in activation and detoxification of carcinogenic polycyclic aromatic hydrocarbons. Drug Metab Pharmacokinet. 2006;21(4):257–276. pmid:16946553
- 60. Ninagawa S, Okada T, Sumitomo Y, Kamiya Y, Kato K, Horimoto S, et al. EDEM2 initiates mammalian glycoprotein ERAD by catalyzing the first mannose trimming step. J Cell Biol. 2014;206(3):347–356. pmid:25092655