Deoxyribonucleic acid methylation

Deoxyribonucleic acid methylation is the most stable and most studied epigenetic modification. In mammals, DNA methylation takes place predominantly on cytosine residues of CpG dinucleotides. Cytosine methylation results from the transfer of a methyl group from S-adenosylmethionine (SAM, the universal methyl group donor) to the carbon-5 position of the cytosine ring. The family of enzymes responsible for DNA methylation are DNA methyltransferases (DNMTs). DMNT1 contributes to maintenance of DNA methylation during DNA replication and cell mitosis, ensuring heritability of the methylation status. DNMT1 preferentially binds to hemi-methylated DNA and copies the DNA methylation of the parental strand to a newly replicated strand. The de novo DNMTs are DNMT3a, DNMT3b and the co-factor DNMT3L. DNMT3a and DNMT3b methylate DNA during embryogenesis and cell differentiation as well as gametogenesis and in response to environmental challenges. DNMT3L lacks a catalytic domain and is proposed to play the role of co-factor, enabling the de novo methylation function of DNAMT3a and DNMT3b.

The genome-wide distribution of DNA methylation is not homogenous and does not have the same effects on gene transcription regardless of its occurrence in promoters, enhancers, gene bodies, CpG rich-regions (namely CpG islands (CGIs)), flanking CGIs or in CpG-poor regions. From the whole genome DNA methylation analysis, it is currently accepted that CGIs often overlap with promoter regions and are unmethylated. When they become methylated, they robustly repress transcription as notably seen in X-chromosome inactivation and genomic imprinting. CG-poor promoters display tissue-specific methylation patterns and there is an inverse correlation between methylation at these promoters and transcription. On the contrary, intragenic methylation positively correlates with transcription. Intergenic methylation was often associated with silencing of repetitive DNA and transposable elements and is involved in the maintenance of genome stability (Jones, Reference Jones2012).

Post-translational modifications of Histone tails

Histone proteins are structural elements involved in the nucleosome formation and DNA package into chromatin. Histones can be submitted to many post-translational modifications (PTMs) affecting mainly the N-terminal tails of core histones, including acetylation (lysine), methylation (lysine and arginine), phosphorylation (serine and threonine), ADP ribosylation, sumoylation (lysine), ubiquitylation (lysine), butyrylation, citrullination, crotonylation, formylation, proline isomerization and propionylation (Prakash and Fournier, Reference Prakash and Fournier2018). PTMs are dynamic and reversible processes. The writing, reading and erasing of these groups are insured by key protein families that control epigenetic signalling. The ‘Histone Code’ hypothesis, whereby distinct histone modifications, on one or more tails, act sequentially or in combination to form a ‘histone code’ that is read by other proteins to bring about distinct downstream events, has been proposed (Strahl and Allis, Reference Strahl and Allis2000). Thus, the presence of these different chemical groups in histone tails greatly influences the histone–DNA interactions and defines chromatin structure affecting gene expression.

Within the PTMs, histone acetylation and histone methylation are the most studied mechanisms in relation with the nutritional status of individuals. Histone acetylation is regulated by the balanced action of histone acetyltransferase and histone deacetylases, largely studied for their potential therapeutic use as anti-cancerous agents. Acetylation of histone tails loosens the histone–DNA interactions, thus enabling the DNA access to transcription machinery leading to gene expression, whereas deacetylation of histone tails strengthens the interactions of histones with DNA and is generally associated with gene repression. Histone methylation levels are also dynamically regulated, such as acetylation of histones, by histone methyltransferases and histone demethylases. Lysine residues (K4, 9, 27, 36 and 79 for H3; K20 for H4) can be mono-, di- or tri-methylated, whereas argine residues can become only mono-methylated. As for DNA methylation, SAM is required as a methyl donor for histone methylation. The effect of this PTM depends on both the modified residue and the extent of methylation: methylation of histone H3 on lysine 4 and 36 is related to an open chromatin structure and thus, an active transcriptional state, whereas H3K9 and H3K27 methylation recruits proteins that compact chromatin, which generally results in gene silencing.

Non-coding RNA

Only about 20% of gene transcription across the mammalian genome is associated with protein-coding genes indicating that ~80% of transcripts are non-coding (Nolte-‘t Hoen et al., Reference Nolte-‘t Hoen, Van Rooij, Bushell, Zhang, Dashwood, James, Harris and Baltimore2015). Non-coding RNAs include the small RNAs (sncRNAs; <200 nucleotides) consisting mainly of microRNAs (miRNAs, 19–24 nucleotides long), short interfering RNAs, Piwi-interacting RNAs, small nucleolar RNAs and the long non-coding RNAs (lcRNAs).

The most studied family of sncRNAs is the miRNAs, defined as endogenous, single-strand, non-coding sncRNAs that act at the post-transcriptional level by binding target complementary sequences of messenger RNAs. Their precursors are large primary-miRNAs generated by Pol-II mediated transcription that are subsequently processed in the nucleus to long premature miRNAs and subsequently cleaved by the DICER enzyme (RNAse type III enzyme) in the cytoplasm. After introduction in the RNA-induced silencing complex, the complementary mRNA sequences are targeted and either degraded or their translation inhibited. MiRNAs can remain in the cell (intra-cellular miRNAs) but also be transferred to the extra-cellular fluid compartments either free, in exosomes, in apoptotic bodies, or in association with high-density lipoproteins or bound to RNA-binding proteins (Rome, Reference Rome2015). These extra-cellular miRNAs are disseminated and target other organs (Nolte-‘t Hoen et al., Reference Nolte-‘t Hoen, Van Rooij, Bushell, Zhang, Dashwood, James, Harris and Baltimore2015).

The miRNAs are well conserved among species (Chen and Rajewsky, Reference Chen and Rajewsky2007). They regulate gene expression through multiple pathways. Most miRNAs have several targets (Ross and Davis, Reference Ross and Davis2014) and may also interact with other RNAs that are not their direct target to modulate gene expression (Pasquinelli, Reference Pasquinelli2012). Some miRNAs have been shown to be involved in global DNA hypomethylation as they target DNMTs in 3ʹuntranslated regions of genes. For example, in cattle, miRNA-152 was shown to target DNMT1 (Wang et al., Reference Wang, Bian, Wang, Li, Wang, Li and Gao2014).

Extra-cellular miRNAs vary depending on the individual’s nutritional status as a result of interactions with the gut flora, with specific variation of the same miRNAs in the stools and in the blood of individuals in response to specific nutriments, but also as a response to specific properties of nutriments, such as anti-inflammatory activity (Rome, Reference Rome2015). Studies also indicate that miRNAs contained in nutriments may also cross the intestinal barrier and be transported to target organs such as the liver or the pancreas, where they can induce metabolic responses (Nolte-‘t Hoen et al., Reference Nolte-‘t Hoen, Van Rooij, Bushell, Zhang, Dashwood, James, Harris and Baltimore2015). Thus, the quantity and the composition of miRNA containing extra-cellular vesicles correlate well with many physiological and pathological conditions (Delage and Dashwood, Reference Delage and Dashwood2008) and they are obvious candidates as biomarkers and/or to modulate phenotypes in domestic species (Rome, Reference Rome2015).

Long non-coding RNAs (>200 nucleotides) are very diverse and poorly conserved among species (Ma et al., Reference Ma, Bajic and Zhang2013). They are also largely involved in transcriptional regulation. The most studied lncRNAs are involved in biological process such as X chromosome inactivation in females and genome imprinting (Borensztein et al., Reference Borensztein, Syx, Ancelin, Diabangouaya, Picard, Liu, Liang, Vassilev, Galupa, Servant, Barillot, Surani, Chen and Heard2017). The molecular mechanisms whereby these lncRNAs act are very complex and specific of each locus; generally, the production of lncRNA drives a local transcriptional silencing (Ma et al., Reference Ma, Bajic and Zhang2013).

Epigenetic effects of nutrients: importance of one-carbon metabolism

The above-mentioned feeding practices may induce epigenetic modifications that could subsequently affect parental gametes, embryo development and phenotype and performance of offspring. Indeed, epigenetic mechanisms elicited by nutritional factors are numerous (Delage and Dashwood, Reference Delage and Dashwood2008; Canani et al., Reference Canani, Costanzo, Leone, Bedogni, Brambilla, Cianfarani, Nobili, Pietrobelli and Agostoni2011; Ross and Davis, Reference Ross and Davis2014) (Table 1). Factors in herbivore and particularly ruminant nutrition that could possibly induce epigenetic modifications are discussed below.

Table 1 Nutritional factors relevant to herbivore nutrition and their epigenetic roles in animal and human research, updated from Delage and Dashwood (Reference Delage and Dashwood2008); Canani et al. (Reference Canani, Costanzo, Leone, Bedogni, Brambilla, Cianfarani, Nobili, Pietrobelli and Agostoni2011)

Of particular interest is methionine as a precursor of SAM. Indeed, SAM is not only the primary methyl donor for DNA methylation, but also for protein and histone methylation. S-adenosylmethionine is a species generated in the cyclical cellular process called one-carbon metabolism. One-carbon metabolism is catalysed by several enzymes in the presence of dietary micronutrients, including folate, choline, betaine and co-factors such as B vitamins (B₁₂, B₆ and B₂). Thus, any imbalance in these nutritional factors may possibly affect epigenetic marks.

Gamete quality

Two phases of whole epigenome ‘reprograming’, and especially DNA methylome reprograming, have been identified during the life of an individual. The first one takes place during gametogenesis and the second one during the early embryo development. According to data obtained in mouse models, highly methylated primordial germ cells are progressively demethylated during their proliferation and migration phases towards the still undifferentiated foetal gonads (Smallwood and Kelsey, Reference Smallwood and Kelsey2012). At the time of sex determination, PGCs are thus poorly methylated in the foetal gonads. Then, de novo methylation occurs but at different times for male and female germ cells. In male germ cells, de novo DNA methylation occurs in foetal prospermatogonia arrested in mitosis. At birth, male germ cells methylome has already been re-established and will be maintained during the individual’s life by methylation maintenance activity during the proliferation phase (Smallwood and Kelsey, Reference Smallwood and Kelsey2012). Conversely, de novo methylation in female germ cells only occurs after birth during the concomitant phases of oocyte and follicular growth, when germ cells do not proliferate and thus does not require maintenance activity. Beyond these differences, the final distribution of methylated CpGs in the genome differs between oocyte and sperm (Stewart et al., Reference Stewart, Veselovska and Kelsey2016). Among the differences, regions involved in the regulation of imprinted genes are differentially methylated during oogenesis and spermatogenesis.

Histone modifications are also very dynamic during spermatogenesis and are associated with incorporation of highly sperm-specific histone variants (Ly et al., Reference Ly, Chan and Trasler2015). Finally, the majority of histones are removed and replaced by protamines. The remaining histones and other chromatin proteins are located in structurally and transcriptionally relevant positions in the genome and carry diverse PTM relevant to the control of embryonic gene expression (Meyer et al., Reference Meyer, Rosenkrantz, Carbone and Chavez2017).

The oocyte nucleus is also a carrier of epigenetic information and recent evidence revealed that large domains labelled with specific histone modifications during oogenesis are inherited by the early embryo’s genome and may be responsible for early embryonic gene expression (Dahl et al., Reference Dahl, Jung, Aanes, Greggains, Manaf, Lerdrup, Li, Kuan, Li, Lee, Preissl, Jermstad, Haugen, Suganthan, Bjoras, Hansen, Dalen, Fedorcsak, Ren and Klungland2016).

Effects through the paternal germline

Epidemiological data in humans demonstrate hereditary transmission of metabolic traits though the paternal germ line. For example, children of males whose mothers underwent to severe undernutrition while pregnant during the Amsterdam famine in the winter of 1944–1945 had increased adiposity compared with controls born before or conceived after the famine (Veenendaal et al., Reference Veenendaal, Painter, de Rooij, Bossuyt, van der Post, Gluckman, Hanson and Roseboom2013). Such paternal transmission of metabolic traits has also been demonstrated in rodent models, with many nutritional, metabolic or stress-linked conditions in males associated with offspring phenotypic and epigenetic modifications (Carone et al., Reference Carone, Fauquier, Habib, Shea, Hart, Li, Bock, Li, Gu, Zamore, Meissner, Weng, Hofmann, Friedman and Rando2010). Modifications in sperm include changes in DNA methylation (Lambrot et al., Reference Lambrot, Xu, Saint-Phar, Chountalos, Cohen, Paquet, Suderman, Hallett and Kimmins2013; Martinez et al., Reference Martinez, Pentinat, Ribo, Daviaud, Bloks, Cebria, Villalmanzo, Kalko, Ramon-Krauel, Diaz, Plosch, Tost and Jimenez-Chillaron2014), histone composition (Terashima et al., Reference Terashima, Barbour, Ren, Yu, Han and Muegge2015) but also in sperm miRNA. In an elegant experiment in rodents, it was shown that the offspring of nutritionally induced obese, diabetic male mice also developed obesity. The injection of miRNA isolated from the sperm of these obese males into one cell embryos that were subsequently transferred to control females induced the same phenotype, highlighting the important role of sperm miRNA (Grandjean et al., Reference Grandjean, Fourre, De Abreu, Derieppe, Remy and Rassoulzadegan2015).

In domestic animals, epigenetic modifications in sperm were associated with fertility, age or environment (Rahman et al., Reference Rahman, Kamal, Rijsselaere, Vandaele, Shamsuddin and Van Soom2014; Kutchy et al., Reference Kutchy, Menezes, Chiappetta, Tan, Wills, Kaya, Topper, Moura, Perkins and Memili2018; Lambert et al., Reference Lambert, Blondin, Vigneault, Labrecque, Dufort and Sirard2018). More recently, semen from high and low fertility bulls was characterized for its DNA methylation signatures and used to produce embryos by in vitro fertilization. Despite similar morphology and development to the blastocyst stage, preimplantation embryos derived from high and low fertility bulls displayed significant transcriptomic differences (Kropp et al., Reference Kropp, Carrillo, Namous, Daniels, Salih, Song and Khatib2017). To our knowledge, no studies have been performed so far to determine the effect of nutrition on epigenetic maturity of male gametes in domesticated animals and its consequence on subsequent offspring.

Embryo quality

The second genome-wide epigenetic reprograming takes place during early embryonic development. It starts at fertilization, when both parental genomes meet in the same oocyte inherited cytoplasm. It is first marked by the replacement of sperm protamines by histones (in contrast to what happens during spermiogenesis), together with an overall DNA demethylation accompanied by important histone modifications. DNA demethylation affects both paternal and maternal genomes at different times, and by partially different mechanisms (Canovas and Ross, Reference Canovas and Ross2016). In the mouse embryo, the average DNA methylation reaches its lowest level at the blastocyst stage. DNA methylation then considerably increases in the epiblast and to a lesser extent in the extraembryonic lineages (visceral endoderm and trophectoderm), which correlates with a sharp increase in the expression of de novo methylase enzymes in the epiblast (Zhang et al., Reference Zhang, Xiang, Yin, Du, Peng, Wang, Fidalgo, Xia, Li, Zhao, Zhang, Ma, Xu, Wang, Li and Xie2018). Histones are also subject to massive reprograming during early development with a tendency to the removal of transcription-repressive marks and their replacement by rather permissive ones. These modifications correlate with and are necessary to the transcriptional activation of the newly formed embryonic genome.

Methionine is required for normal developmental of the bovine embryo. Using ethionine, a compound that blocks the metabolism of methionine in the One-carbon cycle, it was shown that methionine was necessary for development from the morula to the blastocyst stage (Ikeda et al., Reference Ikeda, Sugimoto and Kume2012). Addition of SAM to the culture media restored development to the blastocyst stage. Subsequently, the effects of maternal supplementation on gene expression at the blastocyst stage were evaluated in embryos recovered from cows supplemented with different levels of methionine (1.89% v. 2.43% of metabolizable protein) from calving to embryo collection (Penagaricano et al., Reference Penagaricano, Souza, Carvalho, Driver, Gambra, Kropp, Hackbart, Luchini, Shaver, Wiltbank and Khatib2013). The number of recovered embryos and their quality were not affected (Wiltbank et al., Reference Wiltbank, Garcia-Guerra, Carvalho, Hackbart, Bender, Souza, Toledo, Baez, Surjus and Sartori2014) but RNA sequencing analysis of resultant high quality embryos revealed that the small increase in methionine supplementation induced important down-regulation of genes involved in embryo development and immune responses. One of the two upregulated genes reported coded for apolipoprotein L, 3 like, which is involved in transport and metabolism of lipids and cholesterol. The authors hypothesized that as the concentrations of methionine, histidine and lysine are rate limiting in the uterine fluids of pregnant cows at the time of embryonic elongation (Hugentobler et al., Reference Hugentobler, Diskin, Leese, Humpherson, Watson, Sreenan and Morris2007), these effects may be beneficial. Subsequent work showed in a limited number of animals that methionine supplementation before and 30 days after calving reduced blastocyst methylation, whereas methionine supplementation at the time of follicular growth and embryo production did not affect methylation but increased lipid contents of blastocysts (Acosta et al., Reference Acosta, Denicol, Tribulo, Rivelli, Skenandore, Zhou, Luchini, Correa, Hansen and Cardoso2016).

Placental function

In domestic herbivores, the placental structure is classified as epitheliochorial, that is, the chorion (foetal placenta) faces the maternal uterine endometrium. This contrasts with the hemochorial placenta of rodents and primates, where the chorion is directly in contact with maternal blood. Nutrient transport across the placenta relies on a wide variety of mechanisms, involving in particular passive and active transporters. Transplacental transfers are regulated by maternal nutrition: the placenta acts as a sensor to nutrient availability through signalling pathways such as the mammalian target of rapamycin (mTOR) complex that regulates AA-transport or the peroxisome proliferator-activated receptor γ (PPARγ) involved in the regulation of lipid pathways. Moreover, the placenta is an endocrine organ that produces growth factors such as the maternally imprinted IGF2 and a large variety of hormones, including steroid hormones.

Maternal nutrition with nutriments that affect epigenetic marks may induce direct effects on placental function, acting on transporters, placental sensors or growth factors and hormones or affect foetal development through the action of active compounds on the foetus.

Indeed, pregnant ewe supplementation with protected methionine in the last month of gestation produced lambs that tended to be heavier than non-treated controls (Liu et al., Reference Liu, Lei, Hancock, Scanlan, Broomfield, Currie and Thompson2016). Similarly, in dairy cows, treatment with protected methionine in the last month of gestation resulted in increased plasma methionine concentrations (by 29%), maternal food intake, maternal plasma insulin concentrations and calf birth weight. These incremental changes were associated with upregulation of placental genes involved in neutral AA and glucose transport together with increased gene and protein expression of mTOR (Batistel et al., Reference Batistel, Alharthi, Wang, Parys, Pan, Cardoso and Loor2017). In these example, DNA methylation was not explored and it is thus difficult to conclude whether effects were mediated by epigenetic mechanisms.

Maternal diet and long-term effect: foetal and post-natal development and health

In sheep, maternal nutrition (corn diet v. hay) in the second half of gestation significantly reduced the expression of imprinted (H19, MEG8, PEG1, DLK1 and IGF2R) and DNMT genes in the muscles of late pregnancy foetuses (Lan et al., Reference Lan, Cretney, Kropp, Khateeb, Berg, Penagaricano, Magness, Radunz and Khatib2013). The authors suggested that the higher gene expression in the hay group could be related to a reduced availability of dietary methionine in hay. Indeed, higher methylation was observed for the paternally imprinted IGF2R in the muscle of offspring from corn-fed dams.

Similarly, in Angus-cross beef cattle, a high-starch, corn-based maternal diet was shown to increase calf birthweight compared with calves born to dams fed an isocaloric grass hay based diet (Radunz et al., Reference Radunz, Fluharty, Relling, Felix, Shoup, Zerby and Loerch2012). A 2-year old offspring also had modified glucose and insulin metabolism as well as lower marbling scores and reduced intramuscular lipid contents. These were associated with increased expression of imprinted genes in muscle, together with increased expression of DNMT3a (Wang et al., Reference Wang, Lan, Radunz and Khatib2015).

Although these studies seem to indicate that DNA methylation is a major mechanism explaining the observed effects, a combination of different epigenetic mechanisms is likely to occur. Indeed, supplementation of beef cows with grain mix in the last third of gestation was shown to increase feed intake and reduce basal plasma insulin in offspring with no significant effects on marbling score and only moderate modification in gene expression in Longissimus muscle (Moisa et al., Reference Moisa, Shike, Shoup, Rodriguez-Zas and Loor2015). Nevertheless, downregulation of the anti-adipogenic miRNA-34a was observed in the muscle of the offspring of moderately supplemented cows. Furthermore, an interaction between maternal diet and weaning time was observed for miRNAs involved in the regulation of adipogenesis and insulin resistance (Moisa et al., Reference Moisa, Shike, Shoup and Loor2016).

Generally, dietary supplementation is used in the peripartum period in order to prevent or limit the negative energy balance usually observed in high-yielding dairy cows. As previously described, methionine participates to the 1-C metabolism and to synthesis of SAM. A rumen-protected methionine supplementation to the diet given to Holstein cows throughout the peripartum period was shown to upregulate PPARα gene expression and its target genes (ANGPTL4, FGF21 and PCK1) in liver (Osorio et al., Reference Osorio, Jacometo, Zhou, Luchini, Cardoso and Loor2016) associated with improved lipid metabolism and immune function (Li et al., Reference Li, Batistel, Osorio, Drackley, Luchini and Loor2016). Moreover, methionine supplementation induced a decrease in global hepatic DNA methylation and an increase in the methylation of CpG in PPARα promoter region (Osorio et al., Reference Osorio, Jacometo, Zhou, Luchini, Cardoso and Loor2016). This was the first study to demonstrate an epigenetic effects of methionine supplementation on adult tissues in ruminants. Nevertheless, the mechanistic connections between global DNA and region-specific PPARα methylation with PPARα gene expression and functional outcomes in liver remain to be elucidated. Moreover, follow up of the resultant offspring will be also beneficial to determine possible multigenerational effects and the associated epigenetic mechanisms.

Diets rich in folic acid or/and vitamin B₁₂ have complex effect on lactation performance and dry matter intake, with an increase in plasma glucose and a decrease in hepatic lipids. In sheep, restricting the supply of vitamin B₁₂ and folate during the periconceptional period led to widespread epigenetic alterations to liver DNA methylation in offspring and modified adult health-related phenotypes (Sinclair et al., Reference Sinclair, Allegrucci, Singh, Gardner, Sebastian, Bispham, Thurston, Huntley, Rees, Maloney, Lea, Craigon, McEvoy and Young2007). Thus, in dairy cows, it would also be interesting to evaluate the effects of maternal folate/vitamin B₁₂ supplementation diet during the periconceptional period on offspring health. Taking into account that around conception, embryo development competes for nutrients with mammary gland in dairy cows, it was demonstrated that females born to mothers that were lactating while pregnant produced less milk, lived shorter and were metabolically less efficient than females whose foetal life developed in the absence of maternal lactation (Gonzalez-Recio et al., Reference Gonzalez-Recio, Ugarte and Bach2012). Although these multigenerationnal effects are relatively low, they should not be ignored and a vitamin-supplemented diet could be a positive nutritional strategy to be used.