Introduction
Although the genome remains relatively stable during the lifetime of an organism, independent of the environmental influences, the epigenome is modulated and highly susceptible to environmental stimuli, primarily in the initial stages of mammalian development, specifically, gametogenesis and early embryogenesis. Reference Grandjean, Barouki and Bellinger1,Reference Lillycrop and Burdge2 This modulation occurs because during this window of development, widely distributed epigenetic reprogramming involving DNA methylation and histone modifications must occur to produce a viable oocyte that can be fertilized to generate a good quality embryo. Reference Reik3 Therefore, studies investigating the influence of environmental stimuli, such as nutrition and in vitro manipulations, on the epigenetic profiles of gametes and embryos are essential for improving the quality of the embryos and health of the progeny in adulthood.
In vitro embryo production (IVP), the nutrition quality and physical status of oocyte donors, in vitro manipulation of oocytes and embryos, and in vitro culture conditions, especially the culture medium composition, can strongly influence embryo quality. Reference Watson, De Sousa and Caveney4–Reference Kadokawa, Tameoka, Uchiza, Kimura and Yonai6 In vivo endocrine and metabolic changes that can be induced by the nutritional conditions of oocyte donors are followed by alterations in follicular fluid composition, which may impair oocyte development and quality, as the oocyte is highly susceptible to any disturbance in its microenvironment. Reference Leroy, Opsomer, Van Soom, Goovaerts and Bols7 Taken together, all these aspects, either in vivo or in vitro, may specifically alter DNA methylation patterns and, in turn, impair initial embryo development. Reference Mann, Lee and Doherty8,Reference Young, Fernandes and McEvoy9 Moreover, aberrant establishment of the epigenetic profile during initial embryo development may result in a persistently abnormal epigenetic profile throughout development, thereby compromising the health of the fetus and progeny in adulthood. Reference Grandjean, Barouki and Bellinger1,Reference Sinclair, Allegrucci and Singh10
Both posttranslational histone modifications and DNA methylation patterns are crucial to the regulation of gene expression; S-adenosylmethionine (SAM), which is produced from methionine, is the substrate that provides methyl groups for the establishment of histone and DNA methylation patterns. Reference Jaenisch and Bird11 Methionine is a sulfur (S)-containing amino acid that can be supplied through the diet, Reference Acosta, Denicol and Tribulo12 as well as through homocysteine remethylation, in which a methyl group is transferred from 5-methyltetrahydrofolate (5-methyl-THF) or from betaine, which is a by-product of choline. Reference Olthof, van Vliet, Boelsma and Verhoef13,Reference Saw, Yuan and Ong14 In addition, B vitamins, including vitamins B2, B6, and B12, are cofactors in the homocysteine remethylation process. Reference Saw, Yuan and Ong14 Ruminants usually do not consume any dietary source of vitamin B12 and depend entirely on the ruminal microbiota for the synthesis of this vitamin, which is dependent on continuous cobalt (Co) supplementation in the diet. Reference Bechdel, Honeywell, Dutcher and Kunutsen15 Therefore, the supply of SAM as a methyl donor for DNA methylation depends on the availability of all these nutrients, which are involved in one-carbon metabolism. The major function of this cycle is ensuring that cells always have an adequate supply of SAM, even when the ingestion of methyl group donors, such as methionine, betaine, or choline, is low. Reference Inoue-Choi, Nelson and Robien16 Mammalian oocytes and embryos have been reported to express several important enzymes related to one-carbon metabolism. Reference Ikeda, Namekawa, Sugimoto and Kume17,Reference Kwong, Adamiak, Gwynn, Singh and Sinclair18 This finding suggests that oocytes and preimplantation embryos can independently utilize and metabolize nutrients from one-carbon metabolism, such as methionine, choline, betaine, folate, and B vitamins.
Therefore, considering the relevance of the one-carbon cycle as a methyl donor in the epigenetic machinery, we hypothesized that manipulating specific compounds associated with this cycle in the oocyte donor diet may change the epigenetic reprogramming of the progeny, thereby helping to elucidate how maternal nutrition influences the epigenome of the progeny and their health later in life. Reference Lillycrop and Burdge2,Reference Johnson, Javurek and Painter19
Accordingly, we evaluated the effects of different levels of S and Co in the diet of the oocyte donor Bos taurus indicus Nellore heifers during the entire pre- and periconceptional periods on oocyte and embryo production and on the DNA methylome of F1 embryos. Moreover, we evaluated candidate differentially methylated regions (DMRs) identified in embryos in the blood and sperm cells of their counterparts in adult animals.
Materials and methods
Animals and experimental diets
The experiment was approved by the Ethics Committee on Animal Use (CEUA-Protocol no. 98/2010), School of Veterinary Medicine and Animal Science of the Universidade Estadual Paulista “Júlio de Mesquita Filho.”
In this study, we used Nellore (B. taurus indicus) heifers (n = 20) aged approximately 30 months, with an average weight of 395.36 ± 33.3 kg and body condition scores (scale 1–5) between 3 and 4. The animals were fed in pens with widths of 80 cm per animal and allowed free access to water. We offered a diet based on sugarcane, urea, S, and mineral premixes. The animals were randomly allocated into two experimental groups of 10 animals each.
The feed was offered to the animals twice a day and was chemically similar for both groups, except for the Co and S contents. The diet primarily consisted of sugarcane, urea, and mineral premix (SAL GRAMA NUTRIÇÃO ANIMAL®) (Table 1). The chemical composition of the mineral premix was verified through an analysis performed before the initiation of the experiment. The chemical composition of sugarcane was based on three analyses performed throughout the experiment and aimed to adjust the diets only for the maintenance of the animals (Table 2). Sugarcane was chosen as the forage due to its low content of S, Co, and protein (Table 2). Heifers were randomly assigned to one of two dietary treatments: (1) control with sugarcane (ad libitum) + urea/S (9:1) (1%) and mineral premix and (2) low S/Co with sugarcane (ad libitum) + urea (1%) and mineral premix without any source of S and Co (Tables 1 and 2). The diets were administered over 6 months. Every day prior to providing the feed, any remaining food was removed from the troughs and weighed to calculate the daily consumption. The animals were weighed every 15 days with 8 h of fasting to calculate the average daily gain (ADG) and for blood collection. At the end of the experiment, all heifers (n = 20) were inseminated with the same bull that was used in the IVP and were subsequently placed in a brachiaria grass pasture with free access to water and mineral salt. Among the progeny of those heifers, we collected blood and semen from seven bulls (three animals from heifers of the control group and four from heifers of the low-S/Co group) to confirm the methylome data. Semen was collected by electroejaculation, and samples were stored in liquid nitrogen (−196 ºC) until DNA isolation was performed. The scheme of the experimental design is shown in Fig. 1.
Table 1. Consumption of the experimental diets
a Mineral premix without any source of sulfur and cobalt.
b Chemical composition of mineral premix is shown in Table 2.
Table 2. Chemical composition of sugarcane and mineral premix
SD, standard deviation.
1 Mineral premix (GRAMA®). Basic composition of mineral premix: dicalcium phosphate, sodium chloride, calcium iodate, ventilated sulfur, zinc oxide, copper oxide, cobalt carbonate, and sodium selenite.
Fig. 1. Scheme of the experimental design. Heifers (n = 10 per treatment) were randomly assigned to one of two dietary treatments, control and low S/Co (S – sulfr; Co – cobalt). The diets were administered over 6 months. The animals were weighed every 15 days with 8 h of fasting to calculate the average daily gain (ADG) and for blood collection. Ovum pickup (OPU) (n = 7/animal) was performed weekly, beginning 3 months after the experimental diets were first offered. After OPU has finished, all heifers (n = 20) were inseminated (AI) with the same bull that was used in the in vitro embryo production (IVP). D0 – Day 0; D7 – Day 7; D14 – Day 14; D21 – Day 21; D28 – Day 28; D35 – Day 35; and D42 – Day 42.
Blood biochemical analyses
Folic acid, vitamin B12, homocysteine, insulin, glucose, and insulin growth factor 1 (IGF1) were measured in the blood plasma of all heifers (n = 20) every 15 days (n = 11 assays per animal). All animals were fasted for 8 h before blood sample collection. Analyses were performed by the Instituto Sabin, Brasília-DF, Brazil. Glucose was measured using Advia 2400 equipment following the hexokinase methodology (Siemens, Berlin, Germany). Folic acid and basal insulin were measured using ADVIA Centaur equipment according to a chemiluminescence-based methodology (Siemens, Berlin, Germany). Vitamin B12 was measured using ADVIA Centaur equipment according to the Centaur methodology (Siemens, Berlin, Germany). The levels of IGF1 and homocysteine were measured via chemiluminescence analysis using IMMULITE XPi 2000 equipment (Siemens, Berlin, Germany).
Oocyte retrieval by ovum pickup (OPU)
OPU (n = 7/animal) was performed weekly, beginning 3 months after the experimental diets were first offered. For OPU, we used an ultrasonographic scanner (Aloka SSD 500® Japan) coupled to a probe with a microconvex sector transducer, a 7.5-MHz model UST-9125-7.5 (Aloka®, Japan), a WTA® transvaginal guide (Brazil), and a VMAR5100 Cook® vacuum pump (Australia) coupled to a system with a WTA® 18 G needle (0.9 x 70) (Brazil). Prior to follicular aspiration, caudal epidural anesthesia using 3–5 mL of 2% lidocaine (Pearson®; Eurofarma, Brazil) was administered. The oocytes were aspirated in phosphate-buffered saline medium with 5% bovine fetal serum (Sigma-Aldrich® St. Louis, MO) and 1 μL/mL sodium heparin (Liquemin® iv; Roche, Switzerland) at a constant temperature of 39°C. After aspiration, the cumulus–oocyte complexes (COCs) were sent to the laboratory to be selected Reference Caixeta, Ripamonte, Franco, Junior and Dode20 and subjected to IVP.
IVP
All media used for IVP were supplied by GENEAL Genetics and Animal Biotechnology S/A, Uberaba-MG, Brazil. COC grades I, II, and III, at 15–30 per treatment (control and low S/Co), were transferred to 150 μL drops of maturation medium, coated with silicone oil, and incubated for 22 h at 39°C in 5% CO2. After maturation, the oocytes were washed and transferred to 150 µL of fertilization medium drops. Semen with known in vitro fertility from a bull that is routinely used in our laboratory was thawed at 37°C for 30 s in a water bath, and the sperm cells were selected via centrifugation on discontinuous Percoll gradient of 45–90%. Reference Machado, Carvalho and Filho21 Sperm and COCs were coincubated for 18 h at 39°C in 5% CO2. The day of in vitro insemination was considered day zero (D0). Following coincubation, potential zygotes derived from the fertilized oocytes were washed and transferred to drops containing 150 μL of culture medium. The zygotes were cultured at 39°C in 5% CO2. The embryos were evaluated on D2, D7, and D8 to calculate their cleavage, blastocyst, and hatching rates, respectively.
Whole-genome bisulfite sequencing (WGBS) and identification of candidate DMRs
WGBS was performed at a facility at the Roy J. Carver Biotechnology Center, University of Illinois, as follows. One pool of grade I D8 expanded blastocysts was used for each treatment (n = 68 embryos for the control group and n = 41 embryos for the low-S/Co group) to collect the genomic DNA for WGBS. All heifers by group contributed proportionally to the embryos in each pool. As we used a single biological replicate, it was necessary to employ specific methods that can identify DMRs through two-group comparisons without biological replicates. Reference Juhling, Kretzmer, Bernhart, Otto, Stadler and Hoffmann22,Reference Wu, Xu and Feng23 The WGBS experiment served as an initial screening for identifying potential genomic regions as candidate DMRs for characterization (next subsection). Genomic DNA from the embryos was isolated using the QIAamp DNA Micro kit (Qiagen®) following the manufacturer’s instructions. The shotgun DNA libraries were prepared using the Library Construction kit from Kapa Biosystems® with one modification: following adaptor ligation, the libraries were treated with the EZ DNA Methylation Lightning kit (Zymo Research®). Bisulfite-treated libraries were amplified with Kapa HiFi Uracil+ DNA polymerase. The libraries were pooled in equimolar concentrations, quantified through quantitative polymerase chain reaction (PCR), and sequenced on two lanes for 166 cycles from each end of the fragments on an Illumina HiSeq2500. For quality control, the Trim Galore! (http://www.bioinformatics.babraham.ac.uk/projects/trim_galore/) program was utilized. PCR duplications and nonconverted reads were removed using MarkDuplicates.jar from the PICARD toolkit (“Picard Toolkit.” 2019. Broad Institute, GitHub Repository. http://broadinstitute.github.io/picard/; Broad Institute). The Bismark package Reference Krueger and Andrews24 was used to map bisulfite-treated sequencing reads to a B. taurus reference genome (UMD3.1), and methylation calls were performed. For the two libraries, the total number of reads was 375,817,934, and the mapping is summarized in Table 3. Non-CpG methylation was 0.6% for each sample, indicating high sodium bisulfite conversion efficiency (>99%). Genomic features were categorized as exon, intron, intragenic, and intergenic regions according to the information available in the UMD3.1 reference. The Bis-single nucleotide polymorphism (SNP) method was employed for SNP calling. Reference Liu, Siegmund, Laird and Berman25 Homemade scripts were developed to convert the output from Bis-SNPs for use in the next step, which was the detection of differential DNA methylation. To annotate differentially methylated CpGs and DMRs, the Metilene method was employed. Reference Juhling, Kretzmer, Bernhart, Otto, Stadler and Hoffmann22 Metilene was utilized in this analysis because it provides several parameter adjustments, and this technique is one of the few methods developed to date for annotating DMRs without biological replicates. The following parameters were employed to run Metilene: a minimum of 10 CpG distant at a maximum of 50 nucleotides from one other and P ≤ 0.05. The annotation of the Gene ID was made using an in-house script and the UMD3 annotation file. The predicted proteins from the Ensembl79_UMD3.1 genome were annotated with PFAM_33.1 version. A correspondence table between PFAM and gene ontology (GO) terms (PFAM2GO – version date: 2018/09/08) was employed to infer GO terms in each bovine protein. The hypergeometric test within the FUNC program (PMID: 17284313) was utilized to determine enriched GO terms (FDR < 0.05) in the significantly differentially methylated genes. The GO term fold-enrichment values were calculated as the ratio between the observed (methylated) and expected (genome) gene frequencies.
Table 3. Summary of whole genome bisulfite sequencing
(*) Coverage = (read count * read length)/total genome size.
Control.
93,269,406 * 2 * 165/2,660,922,743 = 11.56.
Low S/Co.
92,527,690 * 2 * 165/2,660,922,743 = 11.47
Characterization of candidate differentially methylated genes using bisulfite PCR (BS-PCR) in adult progeny
Among the candidate DMRs that were identified in D8 expanded blastocysts using WGBS, we first preselected 32 candidate DMRs (Table 4) that are physically linked to genes related to epigenetic mechanisms and reproduction, due to the research interest of our laboratory. Subsequently, we selected two DMRs located in the DNMT1 and DNMT3B genes (in boldface in Table 4) to validate the WGBS using BS-PCR. As biological samples, we used genomic DNA from the blood and sperm cells of seven bulls (three progenies from heifers of the control group and four from the low-S/Co group). Genomic DNA was isolated from the blood using the DNeasy Blood and Tissue kit (Qiagen) according to the manufacturer’s instructions. Genomic DNA from the sperm was isolated using a salting-out procedure Reference Carvalho, Michalczechen-Lacerda and Sartori26 with an additional step in which 5 mM Dithiothreitol (DTT) was added. Subsequently, genomic DNA (500 ng) was treated with sodium bisulfite using the EZ DNA Methylation-Lightning kit (Zymo Research, Irvine, CA, USA) according to the manufacturer’s instructions. Sodium bisulfite-treated DNA was stored at −80°C until PCR amplification was performed. Primers for amplifying the CpG islands of the DNMT1 and DNMT3B genes were designed using MethPrimer Reference Li and Dahiya27 and Bisulfite Primer Seeker software (http://www.zymoresearch.com/tools/bisulfite-primer-seeker). Primer sequences, GenBank access number, number of CpG sites, amplicon size, and annealing temperature are described in Table 5. The total volume of the reaction mixtures was 20 μL, comprising 1× Taq buffer, 1.5 mM MgCl2, 0.4 mM dNTPs, 1 U Platinum™ Taq polymerase (Invitrogen, CA, USA), 0.5 μM of each primer (forward and reverse), and 2 μL of bisulfite-treated DNA. PCR was performed with an initial denaturing step at 94°C for 3 min followed by 29 cycles at 94°C for 40 s, annealing temperature for 1 min, and 72°C for 1 min with a final extension at 72°C for 15 min. The annealing temperature for each gene is specified in Table 5. The amplicons were purified on an agarose gel using the Wizard® SV Gel and PCR Clean-Up System (Promega, Madison, WI, USA) according to the manufacturer’s instructions. The purified amplicons were cloned into the TOPO TA Cloning® vector (Invitrogen, CA, USA) and transferred into DH5α cells using a heat shock procedure. Plasmidial DNA was isolated using Pure Yield Plasmid Miniprep (Promega), and individual clones were sequenced using BigDye® cycle sequencing chemistry and an ABI3100 automated sequencer. Electropherogram quality was analyzed using Chromas®, and methylation patterns were processed using the QUantification tool for Methylation Analysis (QUMA, http://quma.cdb.riken.jp/top/index.html). Reference Kumaki, Oda and Okano28 The DNA sequences were compared with GenBank reference sequences (GenBank ID listed in Table 5), and only sequences that originated from clones with ≥ 95% identity and ≥ 97% cytosine conversion were utilized in the analysis.
Table 4. Differentially methylated regions between control and low-S/Co groups for genes related to epigenetic machinery, spermatogenesis, and one-carbon cycle metabolism in bovine embryos
Differentially methylated regions in boldface were used to validate the WGBS using BS-PCR.
Table 5. Primers for methylation analysis
F, forward; R, reverse; bp, base pair.
Statistical analysis
For the blood biochemical parameters, to evaluate the effects of time, diet, and the interaction of time and diet on the response variables, a factorial design was employed considering two factors, namely, time and diet. A normal distribution was assigned, and a covariance analysis model, with weight serving as a covariate, was used. A Poisson distribution was assigned to the number of aspirated follicles and the number of total oocytes. We employed a generalized linear model to evaluate the effect of diet on these responses. The fit using the Poisson distribution did not adjust all the variability. Thus, we used the quasi-likelihood method for estimation of the heterogeneity factor. We also evaluated the effects of time, diet, and the interaction of time and diet on the number of total oocytes. For the embryo production data, a natural choice was the binomial distribution. The fit using the binomial distribution did not adjust all the variability. The heterogeneity present was treated using the quasi-likelihood method (quasi-binomial). All analyses were performed using the R Program “Development Core Team” (R Foundation for Statistical Computing, Vienna, Austria). 29 ADG, daily food consumption, and BS-PCR data were compared using the t test and the Mann–Whitney test for data showing normal distribution and data not showing normal distribution, respectively. The data were compared using the Prophet program Version 5.0 (BBN Technologies System; NIH, Bethesda, MD, USA, 1996). Analysis of the WGBS data is described in detail above under the heading “Whole-genome bisulfite sequencing (WGBS) and identification of candidate DMRs.” Global methylation data were compared between groups and genomic features using two-tailed Fisher’s exact test and two-tailed paired t test. The data were compared using GraphPad Prism software (GraphPad Software, La Jolla, California, USA; www.graphpad.com) and the R program “Development Core Team” (R Foundation for Statistical Computing, Vienna, Austria). 29
Results
ADG and daily food consumption
Body weight measurements were performed 11 times in each heifer every 15 days. Every time we weighed the heifers, they were subjected to 8 h of fasting to calculate the ADG. No significant differences (Student’s t test, P ≥ 0.05) were found in ADG (control: 0.311 ± 0.104 kg/d; low S/Co: 0.207 ± 0.208 kg/d) or average daily food consumption (Table 1) between the control and low-S/Co groups.
Plasma biochemical profile
Analysis of covariance showed no significant interaction of diet and time for any plasma metabolites between treatments, including homocysteine (P = 0.126), folic acid (P = 0.99), B12 (P = 0.541), IGF-I (P = 0.95), insulin (P = 1.00), and glucose (P = 0.256). Time significantly affected the folic acid (P < 0.001), vitamin B12 (P < 0.001), IGF-I (P < 0.001), insulin (P < 0.001), and glucose (P < 0.001) concentrations. The profile of the low-S/Co group was significantly different from that of the control group with respect to the following metabolites: homocysteine (P < 0.001), folic acid (P = 0.011), vitamin B12 (P < 0.001), IGF1 (P = 0.034), and glucose (P = 0.003). The low-S/Co group exhibited higher levels of homocysteine, folic acid, and glucose but lower levels of vitamin B12 and IGF1 than did the control group (Fig. 2 and Table 6).
Fig. 2. Distribution of plasmatic levels of the biochemical metabolites measured in the control and low-S/Co groups over time. Time represents blood sample collection every 15 days with animals with 8 h of fasting. No significant interactions of diet and time were observed for any plasma metabolites. Time significantly affected the folic acid, B12, IGF1, insulin, and glucose concentrations (P < 0.05). The low-S/Co group exhibited higher levels of homocysteine, folic acid, and glucose but lower levels of vitamin B12 and IGF1 than did the control group (P < 0.05). Data are represented by means ± SD. S, sulfur; Co, cobalt; IGF1, insulin growth factor 1.
Table 6. Effects of the low-S/Co diet on metabolites of one-carbon cycle, IGF1, insulin, and glucose levels in the peripheral circulation
a,b Different letters indicate different means at a 5% significance level.
Oocyte retrieval and embryo production
The diets significantly affected the total numbers of oocytes. No effects of time or the interaction of diet and time were observed on the total numbers of oocytes. The low-S/Co group produced fewer oocytes than the control group (ANOVA, P = 0.0438) (Fig. 3 and Table 7). The diets did not influence any of the embryo production parameters (Table 8). Heifers from the low-S/Co group produced an average of 8.66 D8 embryos by OPU, whereas heifers from the control produced an average of 17.33 D8 embryos by OPU.
Fig. 3. Number of oocytes recovered by ovum pickup (OPU). Each OPU represents the average number of oocytes recovered from 10 heifers of each treatment, control and low S/Co. No effects of time or the interaction of diet and time were observed on the total number of oocytes. The low-S/Co group produced fewer oocytes than the control group (P < 0.05). Data are represented by means ± SD.
Table 7. Effect of the diets on the total number of oocytes per animal recovered by ovum pickup
a,b Different letters indicate different means at a 5% significance level.
Table 8. Percentage of blastocysts produced in vitro on day 7 (D7) and day 8 (D8) of development with respect to the number of oocytes, and D8 hatched blastocyst percentage with respect to the number of D8 blastocysts
Identification of candidate DMRs in embryos
We characterized the global methylation profile of F1 embryos using WGBS.
The mean CpG methylation values were 23.42% and 26.21% for the control and low-S/Co embryo groups, respectively. Independent of the treatments, the mean value for global CpG methylation of the embryos was 24.19%. Figure 4a presents the mean CpG methylation values for each chromosome. Although both groups exhibited a similar methylation profile for all chromosomes, the low-S/Co group exhibited a globally higher level of methylation than the control (Fisher’s exact test, P = 4.32E-27). We additionally evaluated the CpG methylation percentage for different genomic regions (exon, intron, and intergenic regions); independent of the treatment, the intergenic regions exhibited less methylation than did the other regions (Student’s t test, P < 2E-16) (Fig. 4b). Finally, we identified potential candidate DMRs distributed according to different genomic features (Fig. 4c). Overall, we identified 2320 candidate DMRs between the low-S/Co and control groups (Fig. 4d and Supplementary Table S1). For all genomic features analyzed, the low-S/Co group showed higher methylation for most candidate DMRs compared to the control group [1322 (56.98%) and 998 (43.02%), respectively] (Fig. 4d). In addition, the exons exhibited fewer DMRs between the control and low-S/Co groups [388; ∼16.7% (Fig. 4c)]. Table 4 presents the candidate DMRs identified between the groups for genes with functions related to the epigenetic machinery, spermatogenesis, and one-carbon cycle metabolism. With respect to GO analysis, no specific terms or pathways were enriched among the 2320 candidate differentially methylated genes (Fig. 5).
Fig. 4. Whole genome bisulfite sequencing (WGBS) of preimplantation bovine embryos. A - Percentage of global CpG methylation by chromosome generated by WGBS of embryos produced using oocytes aspirated from Nellore heifers that were offered diets with different levels of sulfur and cobalt. Control – control group; low S/Co – diet without any source of sulfur and cobalt. B – Percentage of methylated cytosines (CpGs) by different genomic features (intron, exon, intergenic, and intragenic genomic regions). C – Distribution, in percentage, of the differentially methylated regions (DMRs) between the low S/Co and control groups by different genomic features (exon, intron, and intergenic regions). D – Numbers and percentages of DMRs that were more methylated in each group by each genomic feature.
Fig. 5. Gene ontology annotation of significantly differentially methylated genes. Bars represent the number of genes annotated in each GO category. The orange dotted line represents the fold enrichment of each term. Values above 1 suggest a possible term enrichment, but no statistical support was found. Red bars = biological process. Blue bars = molecular function. Green bars = cellular component.
Characterization of candidate DMRs in adult progeny by BS-PCR
We also evaluated the DNA methylation patterns of candidate DMRs in blood and sperm cells of adult animals. To characterize the WGBS data in adult progeny, we selected two DMRs identified in the DNMT1 and DNMT3B genes (Fig. 6b and Table 4). Overall, we sequenced 406 clones to characterize the DNA methylation patterns of those two DMRs in the blood and sperm cells of the seven progeny bulls from heifers of both groups. The DNA methylation profile of each animal and tissue is presented in Fig. 6c and 6d, indicating a hypermethylated pattern in both tissues and genes for all animals. WGBS for embryos exhibited 10% and 77% methylation in the control and low-S/Co groups for DNMT1, respectively, and 46.66% and 90% methylation in the control and low-S/Co groups for DNMT3B, respectively (Fig. 6a and Table 4). Therefore, the blood cells and sperm of the adult progeny exhibited a similar result to the embryos from the low-S/Co group and a hypermethylated pattern compared to the embryos in the control group for both genes.
Fig. 6. Representation of DNMT1 and DNMT3B gene structures, GC content, CpG island prediction, and DNA methylation profiles. (A) Percentage of CpG methylation of DNMT1 and DNMT3B genes on embryos obtained by whole genome bisulfite sequencing. (B) The green bars represent the input sequences; below, blue lines represent introns, blue arrows represent exons and red arrows represent the PCR primer positions. For each gene, DNMT1 and DNMT3B, the GC percentage, CpG island positions, and DMR genomic coordinates are shown. The figure was generated by using Geneious v2020.0.5 (Biomatters, New Zealand). (C) and (D) DNA methylation profiles of DNMT1 (10 CpGs) and DNMT3B genes (13 CpGs), respectively, in blood cells and sperm. Each line represents an individual DNA clone, and each circle represents a CpG dinucleotide. Black circles represent methylated cytosines, and white circles represent unmethylated cytosines. The percentage of methylation of each animal (1515, 1521, 1509, 1513, 1520, 1529, and 1530) is represented by the mean ± standard deviation of the mean.
For both genes, the blood cells exhibited similar DNA methylation for both groups (Fig. 6c and 6d). In contrast, for DNMT1, the low-S/Co group exhibited a slightly higher methylation than the control in the sperm (Mann–Whitney test, P = 0.0356) (Fig. 6c), thereby corroborating the WGBS results, which showed that the control exhibited less methylation in the blastocysts than the low-S/Co group (Table 4).
In addition, we compared the methylation status for each CpG. We identified two specific CpGs (3 and 6) and one CpG (6) that were differentially methylated between the control and low-S/Co groups in the blood cells and sperm cells, respectively, for DNMT1 (Fig. 6c). No individual CpG sites were observed to be differentially methylated for DNMT3B (Fig. 6d).
Discussion
During the prenatal period, epigenetic reprogramming can be susceptible to environmental factors, such as maternal nutrition. Reference Grandjean, Barouki and Bellinger1,Reference Acosta, Denicol and Tribulo12,Reference Laanpere, Altmae, Stavreus-Evers, Nilsson, Yngve and Salumets30,Reference Young, Rees, Sinclair and Langley-Evans31 SAM, which is produced by the one-carbon cycle, is the substrate that methylates genomic DNA and histones. Reference Yi, Melnyk, Pogribna, Pogribny, Hine and James32 Therefore, determining the role played by specific compounds of this cycle in embryo and fetal epigenetic programming is essential to elucidate the effects of nutrition in this window of development.
Mammalian folliculogenesis and oogenesis are constant and continuous processes in the ovaries that begin during fetal life with primordial follicle formation and end with ovulation. Reference van den Hurk and Zhao33 However, the entire duration of these processes in cattle is still widely discussed in the literature. Reference Lussier, Matton and Dufour34–Reference Webb, Garnsworthy, Gong and Armstrong36 When a primordial follicle is recruited for growth, the enclosed oocyte immediately starts its molecular activity, initiating its transcriptional and epigenetic programming. In this study, we subjected the animals to experimental diets for 3 months before initiating OPU, and the diets were administered until the OPU cycles were completed. This strategy ensured that all the aspirated oocytes were under the effect of the diets during complete period of oogenesis.
We first sought to evaluate the influence of low levels of S and Co in the diet on the blood plasma levels of certain metabolites related to the methionine cycle. We observed significant differences between the low-S/Co and control groups with respect to all metabolites, except insulin (Fig. 2 and Table 6). We suggest that the synthesis of S amino acids and B vitamins by ruminal microbiota could be decreased in the low-S/Co group, thereby impairing the one-carbon cycle. In addition to the lower levels of S and Co in this diet, a substantial amount of nitrogen in the diet was supplied by urea, a nonprotein source of nitrogen, which could contribute to decreased S amino acid synthesis, considering that in this situation, the supply of an extra source of S is essential. We observed that the low-S/Co diet decreased vitamin B12 levels and increased homocysteine and folic acid levels (Fig. 2 and Table 6). Our results are supported by reports from the literature, which demonstrated that vitamin B12 deficiency decreased methionine synthesis through 5-methyltetrahydrofolate-homocysteine methyltransferase (MTR), thereby inducing hyperhomocysteinemia and folate accumulation as 5-methyl-THF, Reference Scott37,Reference Stangl, Schwarz, Muller and Kirchgessner38 which is the most common form detected in bovine plasma. 39 Thus, the low-S/Co diet may lead to an accumulation of homocysteine because of the lack of vitamin B12, a cofactor of the MTR enzyme, which leads to decreased activity of the remethylation pathway of homocysteine, which uses 5-methyl-THF as a substrate. Instead, this diet may induce the use of an alternative to the transsulfuration or choline/betaine pathway through the BHMT enzyme. However, in the 1980s, pioneering studies in ruminants demonstrated that the folate pathway, which uses 5-methyl-THF as the substrate, is more important than the choline/betaine pathway in these animals, Reference Xue and Snoswell40,Reference Xue and Snoswell41 possibly explaining the accumulation of homocysteine in the low-S/Co group.
In our study, the low-S/Co group exhibited higher glucose and lower IGF1 concentrations than the control group (Fig. 2 and Table 6). IGF1 is a mitogenic factor Reference Levine, Feng, Mak, You and Jin42 that participates in glucose regulation through homeostasis. The increased glucose level may have contributed to a decrease in IGF1 secretion, thereby preventing further increases in glucose and causing an even greater decrease in plasmatic IGF1 levels. Studies with mice Reference Wang, Yang and Wang43 showed that gluconeogenesis was induced in response to an increase in homocysteine, which contributed to insulin resistance. A study in sheep that evaluated the progeny of animals subjected to a folate- and B12-deficient diet demonstrated similar results with respect to insulin resistance. Reference Sinclair, Allegrucci and Singh10 Reports from the literature additionally support the association between nutrition and ovarian function, and the glucose–insulin–IGF system may be one of the major mechanisms underlying this association. Reference Armstrong, McEvoy and Baxter44,Reference Scaramuzzi, Campbell and Downing45 Most circulating IGF1 is derived from liver synthesis and regulated by growth hormone, insulin, and nutritional intake. Reference Yakar, Setser and Zhao46 IGFs are observed to be important in the early stages of folliculogenesis, and their deregulation may cause severe impairment in preantral and antral follicular development. Reference Webb, Garnsworthy, Gong and Armstrong36 In the present study, we found that animals from the low-S/Co group produced fewer oocytes and showed lower levels of IGF1 than the control group (Figs. 2 and 3 and Tables 6 and 7), which may be explained by the importance of the IGF system for ovarian function, as mentioned above. Considering our hypothesis that the synthesis of S amino acids and B vitamins by ruminal microbiota could be decreased in the low-S/Co group, we suggest that this decrease is also a cause for the lower levels of IGF1 observed in this group. We also investigated the effect of time on the total number of oocytes, and no significant effect was found (Fig. 3). This result may be observed because OPUs started 3 months after the animals were receiving the experimental diets. Figure 3 shows that since the first OPUs, heifers from the control group were already producing more oocytes. Despite the significant effect of the diet on the total number of oocytes (Table 7), the treatments produced no effects on the embryo production rate (Table 8). However, whether this diet could affect the quality of the produced embryos has not been determined to date. The pregnancy rate is one of the best parameters to test embryo quality; however, it was not possible to evaluate this aspect in this study. Instead, based on our initial hypothesis, we performed preliminary molecular analyses in F1 embryos and in the progeny in adulthood. Considering the relevance of nutrition for normal epigenetic reprogramming in the initial development and its influence on the health of the progeny in childhood and adulthood, Reference Grandjean, Barouki and Bellinger1 we investigated whether the manipulation of specific compounds of the one-carbon cycle in the diet offered to the oocyte donors during the pre- and periconceptional periods could alter the genome-wide DNA methylation patterns of F1 in vitro-produced embryos and, if so, whether these altered methylation patterns are maintained in the tissues in adult animals.
Despite the low availability of S and Co in the diet, F1 embryos produced from the low-S/Co group showed higher levels of global methylation than did the F1 embryos from the control: 26.21% and 23.42%, respectively. This result is in contrast to the observation of Acosta et al. (2016) who showed that embryos from cows supplemented with methionine were less heavily methylated. Reference Acosta, Denicol and Tribulo12 However, Acosta et al. measured global methylation through immunofluorescence, and we evaluated embryo methylomes through sequencing at single-base-pair resolution. In contrast, Mattocks et al. (2017) observed increased hepatic global methylation when a methionine-restricted diet was administered to mice and suggested that this result could be observed because of a protective effect of the methionine-restricted diet on hepatic DNA hypomethylation. Reference Mattocks, Mentch and Shneyder47 Although we had not measured the levels of methionine, it is possible that animals from the low S/Co group had less methionine due to the low availability of S in their diet. In this regard, as Mattocks et al. (2017) suggested, our results may be the consequence of a protective effect on the oocytes in the low-S/Co group, in which the DNA methylation reprogramming of the oocytes was altered, consequently inducing a protective effect against the loss of DNA methylation in the embryos. Animals from the low-S/Co group exhibited differences with respect to all blood biochemical parameters, except insulin, compared to the control animals (Fig. 2 and Table 6). We hypothesize that the disturbance in the plasma metabolite profile of the one-carbon cycle compounds in the low-S/Co animals may serve as an indicator of changes in metabolism which, in turn, would interfere with the DNA methylome Reference Yi, Melnyk, Pogribna, Pogribny, Hine and James32 of F1 embryos. Irrespective of the groups, we observed that the average global CpG methylation of the embryos was 24.19%, which was very similar to that in other studies, including a study indicating that human embryos show 25.7% global methylation. Reference Okae, Chiba and Hiura48 These results are in keeping with the findings of other studies, which showed that embryonic cells exhibit less DNA methylation than spermatozoa Reference Okae, Chiba and Hiura48,Reference MacDonald and Mann49 or adult somatic tissues, Reference Zhou, Xu and Bickhart50 considering that embryonic cells are in the initial process of de novo DNA methylation. Reference Reik, Dean and Walter51 In any case, global methylation data must be interpreted with caution because even if global methylation patterns have increased or decreased, specific regions of the genome can behave in the opposite manner. Therefore, we believe that global methylation values alone are not informative and relevant.
We additionally evaluated the CpG methylation percentage according to individual chromosomes. Embryos from both groups showed similar methylation profiles; however, the low-S/Co group exhibited a globally higher level of methylation than the control when the data from all chromosomes were examined (Fig. 4a). This result suggests that the low-S/Co diet affected the methylation profile of the embryos in a nonspecific manner in the genome.
We observed that for all genomic regions, the embryos produced from the low-S/Co group exhibited a higher CpG methylation percentage than did the embryos from the control (Fig. 4b). However, the intergenic regions showed less methylation than the intragenic regions (gene body). This result is in keeping with the findings of other studies, including those performed in mice Reference Shirane, Toh and Kobayashi52 and chickens, Reference Li, Li and Wang53 that showed gene bodies with higher methylation.
Although the global CpG methylation levels of the embryos did not change markedly with respect to the diets (23.42% and 26.21% for the control and low-S/Co groups, respectively), we identified DMRs across the genome that were distributed among the different genomic features (Fig. 4c and Fig. 4d), which is in keeping with the findings described in other reports from the literature. Reference Masser, Hadad and Porter54 For all genomic features analyzed, that is, exons, introns, and intergenic regions, the low-S/Co group exhibited higher methylation for most DMRs compared to the control (Fig. 4d). In addition, the exons were less differentially methylated between the low-S/Co and the control groups (Fig. 4c and 4d). In contrast, intergenic regions, in addition to being less methylated (Fig. 4b), were more differentially methylated between the low-S/Co and control groups (Fig. 4c and 4d); this result was similar to that observed by Masser et al. Reference Masser, Hadad and Porter54 These results may suggest that the effects of environmental stimuli on DNA methylation patterns in the genome may depend on the genomic region. We identified DMRs related to a wide variety of different biological functions (Supplementary Table S1). We performed a GO analysis, and no specific terms or pathways were enriched among the 2320 candidate differentially methylated genes (Fig. 5). This result suggests that the possible disturbance of the methionine cycle (Table 6), which is the universal supplier of methyl groups in the cell, impairs the global genome methylation in a nonspecific manner. In Table 4, we present certain DMRs for genes involved in the epigenetic machinery, spermatogenesis, and one-carbon cycle metabolism, which are of special interest to our laboratory. Therefore, these DMRs are candidates to be investigated in the future studies focusing on the influence of nutrition on embryonic and/or fetal epigenetic reprogramming and fertility.
Despite the several differences that we observed in the DNA methylome of embryos between the two groups, our data require further characterization for two major reasons. First, we performed WGBS using only one biological sample. The second and primary reason is that blastocysts undergo extensive epigenetic reprogramming at this stage of embryo development, and it is possible that several DMRs that we identified in the embryos could be correctly reprogrammed later in development, with no inter- or transgenerational effects being detected. Reference Jenkins, James and Aston55 Therefore, WGBS was performed on the embryos, and we tested its results using tissues from the progeny in adulthood. We believe that it is more relevant to identify whether certain candidate DMRs in groups of embryos maintain differences in methylation patterns in progeny in adulthood. We selected two candidate DMRs for characterization using BS-PCR and performed a consistent experiment using seven progenies in adulthood and sequenced 406 plasmidial DNA clones from blood and sperm cells to confirm the DNA methylation profile of the two candidate DMRs. Although we characterize only two DMRs, we are providing a supplementary file that lists all candidate DMRs. The BS-PCR results showed that both candidate DMRs, DNMT1 and DNMT3B, showed a hypermethylated pattern in both tissues for all animals; this result is in contrast to the WGBS data for the embryos of the control group, which showed a hypomethylated pattern (Fig. 6 and Table 4). These results suggest that even blastocysts of both groups showed different patterns of methylation in DNMT1 and DNMT3B, throughout development, the initial embryo cells from both groups of embryos were reprogrammed similarly in both blood and sperm progenitor cells. Thus, our results suggest that DMRs identified in embryos have a high probability of losing their differentially methylated pattern, owing to the extensive epigenetic reprogramming that occurs during development. Reference Reik, Dean and Walter56 Therefore, we strongly believe that any embryo methylome data must be confirmed in adult individuals and in different tissues.
Finally, we emphasize that we made a considerable effort to evaluate the effects of specific compounds of the methionine cycle in the diet offered over the entire peri- and preconceptional periods on the DNA methylome of embryos and adult progeny, which was a highly expensive and time-consuming experiment. Notably, most studies in cattle only evaluate the final stages of oogenesis; and in this study, we evaluate the effects of nutrition on the entire period of oogenesis in which epigenetic reprogramming occurs in oocytes. Therefore, our results may improve our understanding of the impact of maternal diet on epigenetic reprogramming in cattle and may contribute to the development of specific diets that can be supplied to oocyte donors during specific phases of the reproductive cycle for IVP systems. Although considerable research remains regarding livestock, this information opens new avenues of study for investigations of the impact of altered epigenetic reprogramming in early life on the growth and development of the fetus and the health and production capacity of the offspring in adulthood. Importantly, as we evaluated the effect of maternal nutrition on the epigenetic reprogramming of in vitro embryos and in adult progeny, our results may help to establish a foundation for studies on the developmental origins of health and disease in humans.
Acknowledgments
The authors thank Enoque L. Silveira and Vladinis O. Miranda for their help in conducting the experiment. We would like to thank Editage (www.editage.com) and AJE (www.aje.com) for English-language editing.
Financial Support
We thank Sabin Laboratory - Center Research Support, Brasilia, Brazil for performing biochemical analyses; FAPESP (Process 2010/07971-3) for financial support for the first author; GENEAL Genetics and Animal Biotechnology S/A for supplying the IVP culture media; and Embrapa Genetic Resources and Biotechnology Center and FAP-DF for their support in this study. Maurício M Franco and Alexandre R Caetano are CNPq research fellows.
Author’ Contributions
ARF, RS, RGJ, DBA, RAF, MAND, and MMF, conceived and designed the project, performed the experiments, and analyzed and interpreted the data. ARF, LNV, RCT, MMCC, PG, ASM, NRK, IP, BDMS, JFWS, LOL, JP, and ARC performed the experiments. In addition, all authors made substantial contributions to the composition or critical revision of the manuscript and supplied important intellectual content. ARF, LNV, RCT, and MMF wrote the manuscript.
Conflict of Interest
The authors have no conflicts of interest to declare.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S2040174421000222
