Choline and Folic Acid in Diets Consumed during Pregnancy Interact to Program Food Intake and Metabolic Regulation of Male Wistar Rat Offspring

ABSTRACT

Background

North American women consume high folic acid (FA), but most are not meeting the adequate intakes for choline. High-FA gestational diets induce an obesogenic phenotype in rat offspring. It is unclear if imbalances between FA and other methyl-nutrients (i.e., choline) account for these effects.

Objective

This study investigated the interaction of choline and FA in gestational diets on food intake, body weight, one-carbon metabolism, and hypothalamic gene expression in male Wistar rat offspring.

Methods

Pregnant Wistar rats were fed an AIN-93G diet with recommended choline and FA [RCRF; 1-fold, control] or high (5-fold) FA with choline at 0.5-fold [low choline and high folic acid (LCHF)], 1-fold [recommended choline and high folic acid (RCHF)], or 2.5-fold [high choline and high folic acid (HCHF)]. Male offspring were weaned to an RCRF diet for 20 wk. Food intake, weight gain, plasma energy-regulatory hormones, brain and plasma one-carbon metabolites, and RNA sequencing (RNA-seq) in pup hypothalamuses were assessed.

Results

Adult offspring from LCHF and RCHF, but not HCHF, gestational diets had 10% higher food intake and weight gain than controls (P < 0.01). HCHF newborn pups had lower plasma insulin and leptin compared with LCHF and RCHF pups (P < 0.05), respectively. Pup brain choline (P < 0.05) and betaine (P < 0.01) were 22–33% higher in HCHF pups compared with LCHF pups; methionine was ∼23% lower after all high FA diets compared with RCRF (P < 0.01). LCHF adult offspring had lower brain choline (P < 0.05) than all groups and lower plasma 5-methyltetrahydrofolate (P < 0.05) than RCRF and RCHF groups. HCHF adult offspring had lower plasma cystathionine (P < 0.05) than LCHF adult offspring and lower homocysteine (P < 0.01) than RCHF and RCRF adult offspring. RNA-seq identified 144 differentially expressed genes in the hypothalamus of HCHF newborns compared with controls.

Conclusions

Increased choline in gestational diets modified the programming effects of high FA on long-term food intake regulation, plasma energy-regulatory hormones, one-carbon metabolism, and hypothalamic gene expression in male Wistar rat offspring, emphasizing a need for more attention to the choline and FA balance in maternal diets.

Introduction

Concurrent with the rising consumption of vitamin supplements in the past few decades has been an increase in the prevalence of obesity and diabetes in the population (1, 2). Specifically, mandatory folic acid (FA) fortification policies and the widespread additions to food and beverages have significantly reduced FA inadequacies in the Canadian population (3). However, increased supplement use has led to excessive intakes of FA, by which >1 in 10 North American women of child-bearing age take FA supplements in doses that exceeded the Tolerable Upper Intake Level(1 mg/d) (4).

We were the first to show that high (5- to 10-fold) maternal FA intake during pregnancy programs the development of obesity in Wistar rat offspring (5–8), associated with lasting changes in the expression of hypothalamic food intake regulatory genes acquired through DNA methylation (5). However, it remains unclear whether excesses of FA alone or imbalances between FA and other methyl-donor nutrients involved in one-carbon metabolism determined the outcome. Although nutrient–nutrient interaction between methyl donors has been previously explored, especially pertaining to micronutrient deficiencies (9, 10), the possible perturbation in the one-carbon metabolic pathway when nutrients are consumed in excess remains poorly documented. Like FA, choline plays a critical role in the production of S-adenosylmethionine (SAM), a key modulator of DNA methylation (11, 12). However, in contrast to FA, choline is absent from most common prenatal supplements in North America (13).

Choline is an essential nutrient for fetal development of brain regions that regulate memory and cognitive function (14). We have recently shown that choline is a contributor to the programming of hypothalamic food intake regulatory systems. A high (2.5-fold) choline diet fed during pregnancy was associated with higher expression of the orexigenic neuropeptide-Y neurons at birth, consistent with higher cumulative food intake and body-weight gain postweaning in the offspring compared with those born to mothers fed the recommended amounts of choline and FA (15). The objective of this study was to evaluate the interaction between choline and FA in maternal diets on the one-carbon cycle, hormonal regulation of food intake, body-weight gain, visceral adiposity, and bone health in the offspring at birth and later during adulthood. We hypothesized that dietary choline interacts with FA during pregnancy to program the long-term phenotype of the offspring.

Methods

Study design

The experimental procedure was approved by the University of Toronto Animal Care Committee (protocol no. 20011893). First-time 2-d-pregnant Wistar rats (170–250 g; Charles River, Montreal, Canada) were individually housed in ventilated plastic cages with a 12:12-h light-dark cycle (lights on at 0700 at 22° ± 1°C) and access to food and water ad libitum. Dams were randomized on arrival to 4 dietary treatment groups (n = 12/group). The control [recommended choline and recommended FA (RCRF)] group was fed the standard AIN-93G diet with 2.5 g/kg choline bitartrate and 2 mg/kg FA (D10012; Research Diets). The high (5-fold) FA groups were fed either low choline [0.5-fold, low choline and high FA (LCHF)], recommended choline [1-fold, recommended choline and high FA (RCHF)], or high choline [2.5-fold, high choline and high FA (HCHF)] only during pregnancy. At birth, each litter was sexed and culled to 6 pups per dam. Animals were killed by rapid decapitation and brain and blood were collected from 2 male pups per dam. During the 3-wk lactation period, all dams were switched to the RCRF diet. At weaning, 1 male pup per dam was maintained on the RCRF diet for 20 wk postweaning. Food intake was measured weekly throughout the entire experimental period for offspring and dams. Cumulative food intake of the offspring postweaning was calculated as the sum of total weekly food intake in grams. Body weight of dams was measured on arrival, at birth, and weekly throughout lactation and at birth and weekly thereafter from weaning to 20 wk postweaning in offspring. Offspring were killed by rapid decapitation for blood and tissue collection at 20 wk postweaning following a 6- to 8-h daytime feed deprivation (food removed between 07:30 and 09:30). Visceral fat pads (retroperitoneal, mesenteric, epididymal) were weighed, and visceral adiposity index was calculated from the total fat-pad mass (grams) as percentage of final body weight.

Diet composition

All diets were isocaloric with similar macronutrient, vitamin, and mineral composition but only differing in choline and FA content (Supplemental Table 1). The amount of choline and FA added to the gestational diets was calculated based on the FDA-specified coefficient for the conversion of doses from humans to rodents (16). The 5-fold FA dose during pregnancy is based on our previous studies showing it is a safe and nontoxic dose that results in an obesogenic phenotype in the male offspring postweaning. The AIN-93G diet contains 1 g of free choline/kg (4000 kcal) or the equivalent of 250 mg/1000 kcal in women. The low- or 0.5-fold choline diet is lower than recommended but not deficient, consistent with the Canadian population (≤300 mg/d) and is sufficient to prevent the development of fatty liver in rodents (17). The high-choline rat diet contains 2.5 g of free choline or the equivalent to 625 mg/1000 kcal. This dose is equivalent to ∼1.1g of choline consumed by a woman consuming 1800 kcal and is lower than the Tolerable Upper Limit (3.5 g/d). For further confirmation of the choline and FA content and stability in the gestational diets, the HCHF gestational diet was analyzed by LC-MS/MS at Medallion Labs (Minneapolis, MN, USA) after 36 mo of storage at −80°. The results of this analysis showed that the choline and FA content reflected our original design and is highly stable (total choline = 3 g/kg diet and FA = 0.009 g/kg diet).

Femur length and bone mineral density

After termination, 10 rat carcasses per group were imaged using the Bruker Xtreme in-vivo X-ray imaging system (Bruker). Images of the right femur were acquired using X-ray source energy of 45 keV, binning set at 1 × 1, F stop at 5.6, exposure time was 15 s, and the X-ray filter was 0.8 mm. All images were processed for bone length using a line tool in the manual ROI window placed at each end of the femur. Bone mineral density was processed using the Bruker Molecular Imaging software version MI 5.3.2 with the Bruker Bone Density Software Module as previously described (18).

Plasma insulin and leptin concentrations

Plasma insulin (catalog no. 80-INSRT-E01; ALPCO) and leptin (catalog no. EZRL-83K; EMD Millipore) were measured in offspring at birth and 20 wk postweaning using commercially available ELISA kits according to the manufacturers’ instructions.

Brain and plasma choline, folate, and one-carbon metabolite concentrations

Whole brains collected from 1 pup per dam (n = 5–6/group) at birth and a half of adult offspring brains (cut sagittally) at 20 wk postweaning (n = 5–6/group) were snap frozen then pulverized in liquid nitrogen to measure the concentrations of choline, betaine, and methionine by LC-MS/MS at Cornell University, as described in references (19) and (20). SAM and S-adenosylhomocysteine (SAH) were quantified to assess the methylation potential in the whole brain at birth using LC-MS/MS at Cornell University, as described in references (21) and (22). Plasma samples from adult offspring (20 wk postweaning; n = 6–8/group) were shipped to our collaborator at the Baylor Scott & White Research Institute for analysis of one-carbon metabolites. Plasma concentrations of 5-methyltetrahydrofolate (5-MTHF) was measured by LC –electrospray tandem MS (23). Plasma concentrations of choline, betaine, methionine, total homocysteine, total cysteine, and cystathionine were also measured, as described in reference (24); SAM and SAH were measured as described elsewhere (25).

Hypothalamic RNA extraction

To explore the effects of the gestational choline and FA diets on hypothalamic gene expression, 3 groups were selected based on significant changes in our primary outcome measures (postweaning food intake and body-weight gain). The 3 groups selected were the control (RCRF), RCHF, and HCHF groups. At birth, the whole hypothalamus was dissected from frozen brains collected from 5 pups/group. At 20 wk postweaning, the arcuate nucleus of the hypothalamus (ARC; bregma A 2.9–2.3 mm) was isolated from frozen brains collected from 6 offspring/group. The location of the ARC was confirmed using the fornix and third ventricles as landmarks (26, 27). The samples were homogenized using a TissueRuptor homogenizer (Qiagen Tech). RNA from the homogenized tissue was isolated using Trizol reagent (Invitrogen) and chloroform extraction according to the manufacturer's protocol and quantified by NanoDropTM2000Spectrophotometer (Thermo Scientific, Inc.).

RNA sequencing in hypothalamus of pups at birth

Whole-transcriptome analysis was conducted using RNA sequencing at The Centre for Applied Genomics at the Hospital for Sick Children, Toronto, Ontario, Canada. Differential gene expression in the whole hypothalamus of newborn pups was compared between the RCRF vs RCHF and RCRF vs HCHF groups. Whole-transcriptome analysis involved a poly-A selection of mRNA. Paired-end 2 × 125-bp sequencing was performed with Illumina HiSeq 2500 instrument (Illumina) at a sequencing depth of ≈20 million paired reads per sample. Sequence data were demultiplexed and base calls were converted to FASTQ format with bcl2fastq2 v.2.20 (Illumina). The sequence read quality was assessed using FastQC v.0.11.5 (Babraham Bioinformatics) (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Adapter trimming and removal of lower quality ends was performed using Trim Galore v.0.5.0 (Babraham Bioinformatics) (http://www.bioinformatics.babraham.ac.uk/projects/trim_galore/). Trimmed reads were screened for presence of rRNA and mtRNA sequences using FastQ-Screen v.0.10.0 (Babraham Bioinformatics) (http://www.bioinformatics.babraham.ac.uk/projects/fastq_screen/). RSeQC package v.2.6.2 (http://rseqc.sourceforge.net/) was used to confirm strand-specific RNA-Seq library construction and to assess positional read duplication and read distribution across exonic, intronic, and intergenic regions. The quality of the trimmed reads was re-assessed with FastQC. STAR aligner v.2.6.0c (https://github.com/alexdobin/STAR, https://academic.oup.com/bioinformatics/article/29/1/15/272537) was used to align the trimmed reads to Rnor_6.0 genome downloaded from Ensembl database version 98.6 using Ensembl gene models. The STAR alignments were processed to extract raw read counts for genes using htseq-count v.0.6.1p2 (HTSeq; http://www-huber.embl.de/users/anders/HTSeq/doc/overview.html). Two-condition differential gene expression was performed with DESeq2 (https://bioconductor.org/packages/release/bioc/html/DESeq2.html) v.1.26.0s, using R v.3.6.1. Initial minimal filtering of 10 reads per gene for all samples was applied to the datasets. More strict filtering to increase power was automatically applied via independent filtering on the mean of normalized counts within the DESeq results function. Multiple testing correction of the P values was performed by the Benjamini-Hochberg method.

Functional enrichment analysis of differentially expressed genes

The biological functions of the upregulated and downregulated genes were investigated using the enrichment annotation modules [Kyoto Encyclopedia of Genes and Genomes Pathway (KEGG) and Gene Ontology (GO) Biological Processes] in Metascape (http://metascape.org). Terms with a P value <0.05 with number of enriched genes >3 were considered to be significantly associated with the genes.

Validation of select hypothalamic differentially expressed genes from RNA-sequencing experiment by qRT-PCR

RNA extracted from whole hypothalamus from pups at birth was used to validate differentially expressed genes (DEGs) known to play a role in hypothalamic energy regulation. cDNA was synthesized using the High Capacity cDNA Archive Kit (Applied Biosystems) on the TProfessional Standard Gradient 96 thermocycler (Biometra). Real-time qRT-PCR was performed on the ABI PRISM 7900 Sequence Detection System (Applied Biosystems). An equal volume of cDNA (1 μL/sample) was added to Power SYBR Green Master Mix (Applied Biosystems) with gene-specific primers (Supplemental Table 2). The cycle conditions were 50°C for 2 min, 95°C for 2 min, 40 cycles at 95°C for 15 s, and 60°C for 1 min. Quantification was performed using β2-microglobulin (β2M) as a suitable reference gene selected based on a preliminary screening for the lowest variation. Results are expressed as fold-change by the 2−^{Δ Δ} CT (cycle threshold) method (28).

Statistical analysis

SAS statistical software (version 9.4; SAS Institute, Inc.) was used for all analyses. All datasets were tested for normality and found to be normally distributed. ANOVA by the PROC MIXED procedure followed by Tukey's post hoc test was used to determine the treatment effects on body-weight gain over time with gestational diet and time as the main factors and a diet × time interaction term. Cumulative food intake, body weight, and plasma and tissue measures were compared between groups using 1-factor ANOVA followed by Tukey's post hoc test. Pearson correlation analysis was performed on brain choline, betaine, and methionine concentrations at birth. Relative mRNA expression by qRT-PCR was compared using ANOVA followed by Dunnett's test to compare RCHF vs RCRF and HCHF vs RCRF. On all occasions, 1 pup per litter (dam) from each dam in a group was used for analyses. Statistical significance was declared at P ≤ 0.05. All values are expressed as mean ± SEMs.

Results

Food intake and body weight of dams during pregnancy and lactation

Food intake in dams during pregnancy and lactation was affected by time (P < 0.0001), but there was no effect of the gestational diet (P > 0.05) and no diet × time interaction effect (P-interaction > 0.05; Supplemental Figure 1a). No differences in body weights were observed in dams at arrival and birth (P < 0.05; Supplemental Figure 1b).

Body weight and plasma hormones of pups at birth

The choline and FA content of gestational diets did not affect body weight of male pups at birth (RCRF, 7.04 ± 0.13 g; LCHF, 6.93 ± 0.12 g; RCHF, 6.91 ± 0.14 g; HCHF, 6.73 ± 0.22 g; P > 0.05). Diet fed during pregnancy affected plasma insulin and leptin concentrations in pups at birth. The HCHF pups had ∼55% lower plasma insulin concentrations compared with the LCHF group (P < 0.05) (Figure 1A). HCHF pups also had 39% lower plasma leptin concentrations than RCHF pups (P < 0.05) (Figure 1B).

FIGURE 1.

Plasma insulin (A) and leptin (B) concentrations in male newborn pups from Wistar rat dams fed recommended choline and FA (1-fold; RCRF, control) or high (5-fold) FA with choline at 0.5-fold (LCHF), 1-fold (RCHF), or 2.5-fold (HCHF) during pregnancy. Values are means ± SEMs, n = 10–12/group. Labeled means without a common letter differ, P ≤ 0.05 (ANOVA and Tukey's test). FA, folic acid; HCHF, high choline and high folic acid; LCHF, low choline and high folic acid; RCHF, recommended choline and high folic acid; RCRF, recommended choline and recommended folic acid.

Cumulative food intake, body-weight gain, bone health, and visceral adiposity of male rat offspring at 20 wk postweaning

Cumulative food intake after 20 wk postweaning was ∼10% (P < 0.05) higher in the RCHF and the LCHF (P = 0.07) offspring compared with RCRF and HCHF pups, which were not different from each other (Figure 2A). RCHF and LCHF offspring also had higher body-weight gain over the 20-wk postweaning period compared with the RCRF group, which was not different from the HCHF offspring, and this effect was significantly dependent on the factor of time (time effect, P < 0.0001; diet effect, P < 0.01; diet × time, P-interaction < 0.0001) (Figure 2B). Thus, we have stratified body-weight gain according to the stage of life of the rat offspring (29). During the peri-adolescent stage (week 2–6 postweaning), the LCHF and RCHF offspring had 11% and 7% higher body-weight gain compared with the controls, respectively (P < 0.05) (Figure 2C). When the offspring reached adulthood starting at 7 wk and up to 20 wk postweaning, only the RCHF group had 17% higher weight gain compared with the controls, whereas the LCHF group was not different from controls (P < 0.05) (Figure 2D). In both life stages, weight gain in the HCHF offspring was not significantly different from all 3 groups (Figure 2C, D). At 20 wk postweaning, the RCHF offspring had greater final body weight compared with controls (P < 0.05) but were not different from the LCHF and HCHF offspring, which were not different from each other or the controls (Table 1). Plasma leptin and insulin concentrations, visceral adiposity, femur length, and bone mineral density were not different between all groups (Table 1).

FIGURE 2.

Cumulative food intake (A) and body-weight gain (B) from 1 to 20 wk postweaning of male offspring from Wistar rat dams fed recommended choline and FA (1-fold; RCRF, control) or high (5-fold) FA with choline at 0.5-fold (LCHF), 1-fold (RCHF), or 2.5-fold (HCHF) during pregnancy. Body-weight gain of male Wistar rat offspring during the peri-adolescence stage (weeks 2–6 postweaning) (C) and the young-adulthood stage (weeks 7–20 postweaning) (D). Values are means ± SEMs, n = 9–11/group. Labeled means without a common letter differ, P ≤ 0.05 (ANOVA and Tukey's test). FA, folic acid; HCHF, high choline and high folic acid; LCHF, low choline and high folic acid; RCHF, recommended choline and high folic acid; RCRF, recommended choline and recommended folic acid.

TABLE 1.

Final body weight, visceral adiposity, bone mineral density, and plasma hormones of adult male offspring born from Wistar rat dams fed recommended choline and FA (1-fold; RCRF, control) or high (5-fold) FA with choline at 0.5-fold (LCHF), 1-fold (RCHF), or 2.5-fold (HCHF) during pregnancy¹

	RCRF	LCHF	RCHF	HCHF	P
Final body weight, g	769 ± 19.8b	833 ± 15.9a,b	847 ± 19.9a	822 ± 17.4a,b	0.04
Visceral adiposity, % fbw	8.18 ± 0.40	7.46 ± 0.46	8.51 ± 0.28	8.06 ± 0.37	0.25
Epididymal fat, % fbw	3.11 ± 0.19	2.88 ± 0.15	3.12 ± 0.08	3.07 ± 0.21	0.64
Mesenteric fat, % fbw	0.83 ± 0.05	0.73 ± 0.05	0.77 ± 0.06	0.79 ± 0.09	0.75
Retroperitoneal fat, % fbw	4.24 ± 0.28	3.84 ± 0.27	4.63 ± 0.26	4.19 ± 0.21	0.20
Liver weight, % fbw	2.64 ± 0.09	2.64 ± 0.07	2.58 ± 0.09	2.67 ± 0.06	0.84
Femur length, cm	3.57 ± 0.07	3.49 ± 0.09	3.46 ± 0.09	3.41 ± 0.08	0.97
Bone density, g/cm3	2.17 ± 0.19	2.23 ± 0.20	2.26 ± 0.09	2.31 ± 0.08	0.72
Insulin, ng/mL	3.26 ± 0.15	3.63 ± 0.32	3.88 ± 0.44	3.68 ± 0.28	0.63
Leptin, ng/mL	41.52 ± 3.13	33.0 ± 2.96	42.4 ± 4.01	35.24 ± 7.09	0.36

¹Values are means ± SEMs, n = 9–11. Labeled means without a common letter differ, P ≤ 0.05 (ANOVA and Tukey's test). FA, folic acid; HCHF, high choline and high folic acid; LCHF, low choline and high folic acid; RCHF, recommended choline and high folic acid; RCRF, recommended choline and recommended folic acid; % fbw, % of final body weight.

One-carbon metabolite concentrations in brains of male offspring at birth and 20 wk postweaning

At birth, pups born to dams maintained on the HCHF diet had 22% higher free choline concentrations compared with LCHF pups (P < 0.05; Table 2) but not RCRF or RCHF pups. Additionally, HCHF pups had 33% higher brain betaine concentrations compared with LCHF pups (P < 0.01) but not RCRF or RCHF pups (Table 2). Pup brain choline and betaine concentrations were also significantly positively correlated (P < 0.05; Supplemental Figure 2a). Brain methionine concentrations were, on average, reduced by 23% in all high FA groups (LCHF, RCHF, and HCHF) compared with the control (RCRF) (P < 0.01; Table 2). Pup brain methionine was not significantly correlated with brain choline or betaine concentrations (P-value > 0.05; Supplemental Figure 2b, c). There were no differences in brain concentrations of SAM, SAH, or the SAM to SAH ratio between the groups (P > 0.05; Table 2). At 20 wk postweaning, offspring born to dams fed the LCHF diet had the lowest free choline concentration compared with all groups (P < 0.001; Table 2). However, there were no significant differences in brain betaine, methionine, SAM, SAH, or SAM:SAH between groups (P > 0.05; Table 2).

TABLE 2.

Whole-brain one-carbon metabolites of male rat offspring at birth and 20 wk postweaning born from Wistar rat dams fed recommended choline and FA (1-fold; RCRF, control) or high (5-fold) FA with choline at 0.5-fold (LCHF), 1-fold (RCHF), or 2.5-fold (HCHF)during pregnancy¹

	nmol/g of tissue
	RCRF	LCHF	RCHF	HCHF	P
Birth
Choline	151 ± 7.04a,b	138 ± 5.58b	151 ± 8.59a,b	175 ± 11.6a	0.019
Betaine	19.9 ± 1.06a,b	16.3 ± 0.63b	20.1 ± 1.03a,b	24.6 ± 1.74a	0.002
Methionine	86.9 ± 4.65a	59.8 ± 3.18b	73.2 ± 3.15b	67.7 ± 3.3b	0.001
SAM	44.5 ± 2.48	45.9 ± 1.82	42.4 ± 1.82	42.5 ± 1.83	0.54
SAH	2.74 ± 0.22	2.56 ± 0.17	2.59 ± 0.13	2.47 ± 0.06	0.66
SAM:SAH	16.7 ± 1.47	18.2 ± 0.93	16.7 ± 1.39	17.2 ± 0.77	0.76
Postweaning
Choline	1180 ± 47.4a	323 ± 24.8b	1130 ± 66.1a	1120 ± 87.9a	0.0001
Betaine	15.5 ± 1.42	16.5 ± 0.91	16.6 ± 0.34	17.5 ± 0.86	0.63
Methionine	59.1 ± 2.31	61.1 ± 1.67	59.6 ± 1.03	58.7 ± 1.81	0.44
SAM	18.5 ± 0.94	22.7 ± 1.76	19.5 ± 1.08	22.5 ± 2.11	0.23
SAH	3.11 ± 0.21	2.81 ± 0.16	2.98 ± 0.17	2.92 ± 0.19	0.74
SAM:SAH	6.13 ± 0.71	8.27 ± 0.90	6.64 ± 0.48	7.68 ± 0.48	0.13

¹Values are means ± SEMs, n = 5–6. Labeled means without a common letter differ, P ≤ 0.05 (ANOVA and Tukey's test). FA, folic acid; HCHF, high choline and high folic acid; LCHF, low choline and high folic acid; RCHF, recommended choline and high folic acid; RCRF, recommended choline and recommended folic acid; SAH, S-adenosylhomocysteine; SAM, S-adenosylmethionine.

Choline, folate, and one-carbon metabolite concentrations in plasma of male offspring at 20 wk postweaning

At 20 wk postweaning, HCHF offspring had higher plasma methionine concentrations compared with RCRF offspring (P < 0.05) but not RCHF and LCHF offspring. HCHF offspring had lower total homocysteine concentrations compared with the RCRF and RCHF groups but were not significantly different from LCHF. HCHF offspring also had lower cystathionine than LCHF offspring (P < 0.05) but were not significantly different from RCRF and RCHF groups (Table 3). Plasma concentrations of 5-MTHF were the lowest in the LCHF offspring compared with the RCRF and the RCHF groups (P < 0.01) but not different from the HCHF group (P = 0.07). Plasma concentrations of free choline, betaine, SAM, SAH, and total cysteine were not different between groups during adulthood (P > 0.05; Table 3).

TABLE 3.

Plasma one-carbon metabolites of male offspring at 20 wk postweaning born from Wistar rat dams fed recommended choline and FA (1-fold; RCRF, control) or high (5-fold) FA with choline at 0.5-fold (LCHF), 1-fold (RCHF), or 2.5-fold (HCHF) during pregnancy¹

	Postweaning
	RCRF	LCHF	RCHF	HCHF	P
Choline, μM	13.5 ± 0.90	13.6 ± 1.04	12.5 ± 0.45	14.3 ± 0.36	0.45
Betaine, μM	109 ± 8.87	105 ± 5.33	91.2 ± 3.83	101 ± 3.28	0.22
Methionine, nM	76.8 ± 1.44b	82.3 ± 2.75a,b	79.1 ± 1.73a,b	84.2 ± 1.21a	0.04
SAM, nM	214 ± 6.54	239 ± 11.8	213 ± 6.12	210 ± 5.39	0.06
SAH, nM	68.3 ± 6.32	71.3 ± 7.85	60.1 ± 2.42	65.4 ± 3.45	0.57
SAM:SAH	3.21 ± 0.30	3.70 ± 0.24	3.53 ± 0.12	3.52 ± 0.07	0.39
Total homocysteine, μM	7.31 ± 0.48a	5.77 ± 0.68a,b	6.80 ± 0.25a	4.23 ± 0.26b	0.002
Total cysteine, μM	255 ± 7.45	239 ± 4.59	254 ± 5.85	251 ± 1.39	0.21
Cystathionine, nM	932 ± 83.2b	1220 ± 69.5a	1030 ± 62.8a,b	911 ± 54.7b	0.019
5-Methyltetrahydrofolate, nM	178 ± 4.42a	136 ± 9.94b	168 ± 7.26a	164 ± 7.89a,b	0.006

¹Values are means ± SEMs, n = 6–8. Labeled means without a common letter differ, P ≤ 0.05 (ANOVA and Tukey's test). FA, folic acid; HCHF, high choline and high folic acid; LCHF, low choline and high folic acid; RCHF, recommended choline and high folic acid; RCRF, recommended choline and recommended folic acid; SAH, S-adenosylhomocysteine; SAM, S-adenosylmethionine.

Hypothalamic DEGs in male offspring at birth and 20 wk postweaning

To examine the effects of the gestational diets on hypothalamic gene expression, whole-transcriptome analysis was performed in the hypothalamus from pups born from RCHF, HCHF, and control (RCRF) dams (Supplemental Data Sheets 1 and 2). These groups were selected based on the observed significant differences in cumulative food intake and weight gain, our primary measures. Differential gene expression was only observed in the whole hypothalamus of newborn pups from dams fed the HCHF diet compared with the controls. At a cutoff of an adjusted P value <0.25, a total of 144 genes were differentially expressed, of which 38 were downregulated and 106 were upregulated (Supplemental Data Sheet 3). Functional term enrichment analysis showed several downregulated genes may be involved in GO biological processes including the negative regulation of cell projection organization (GO:0031345), monocarboxylic acid metabolic processes (GO:0032787), and pyruvate metabolic processes (GO:0006090), as well as 2 KEGG pathways related to Alcoholism (ko05034) and Axon Guidance (ko04360). The upregulated genes were enriched in 10 biological processes, of which the top 4 were related to spliceosome complex assembly (GO:0000245), DNA damage regulation (GO:0000077), positive regulation of DNA binding (GO:0043388), and response to oxidative stress (GO:0006979) (Supplemental Figure 3 and Supplemental Table 3). Furthermore, several KEGG pathways were enriched such as choline metabolism in cancer (ko05231) and mitogen-activated protein kinase (MAPK) signaling (ko04010) (Supplemental Figure 4 and Supplemental Table 4). Two genes previously identified to play a role in energy regulation were found to be downregulated: Klotho (Kl; −0.77 log-fold change, adjusted P = 0.11) and corticotrophin-releasing hormone (Crh; −0.79 log-fold change, adjusted P = 0.23). These genes were further validated by qRT-PCR, by which Kl and Crh mRNA expression was ∼0.5-fold lower in the HCHF compared with the RCRF groups (Figure 3A). At 20 wk postweaning, no changes in mRNA expression of the selected genes were observed in the ARC between groups (Figure 3B).

FIGURE 3.

Relative mRNA expression of DEGs identified to play a role in energy regulation in the hypothalamus at birth (A) and the ARC at adulthood (20 wk postweaning) (B) of male offspring born from Wistar rat dams fed recommended choline and FA (1-fold; RCRF, control) or high (5-fold) FA with choline at 1-fold (RCHF) or 2.5-fold (HCHF) during pregnancy. Values are means ± SEMs. n = 4–6/group *Different from control, P ≤ 0.05 (ANOVA and Dunnett's test). ARC, arcuate nucleus of the hypothalamus, Crh, corticotropin-releasing hormone; DEG, differentially expressed gene; FA, folic acid; HCHF, high choline and high folic acid; Kl, Klotho; RCHF, recommended choline and high folic acid; RCRF, recommended choline and recommended folic acid.

Discussion

The results of this study support our hypothesis that choline and FA interact in maternal diets to program the phenotype of offspring. By increasing choline in the gestational diet, the effect of high-FA diets on increased food intake was mitigated in the male offspring during adulthood. These results not only highlight the role of gestational choline as a contributor to the long-term programming of food intake regulation of the offspring but also demonstrate that this effect is dependent on its interaction with FA in the maternal diet.

Several lines of evidence indicate the importance of a balanced intake of choline with FA in gestational diets. First, consistent with our previous reports (5, 7, 30), the high maternal FA diet resulted in increased cumulative food intake and body-weight gain in male offspring. Second, adding choline to the high-FA diets modulated these effects, as evidenced by food intake and weight gain similar to those born to dams fed the control RCRF diet and led to changes in hypothalamic regulatory genes at birth. Third, the imbalance between FA and choline in the gestational diets also altered the metabolic phenotype of the offspring at birth. Fourth, the interaction of choline with FA was evidenced by its effect on the one-carbon cycle, at birth and in the adult offspring.

Consistent with our previous studies (5, 30), we have shown that high (5-fold) FA intake during pregnancy remains to be the main contributor to the long-term programming of weight gain in male Wistar rat offspring, by which RCHF offspring had significantly higher body weight compared with the controls over the 20-wk postweaning period. Moreover, this study shows for the first time that varying choline amounts may modulate the programming effect of high-FA gestational diets in an age-dependent manner. This is evidenced by the higher weight gain in the LCHF offspring during peri-adolescence that decreases during adulthood when compared with the controls. In contrast, offspring born to dams fed the balanced HCHF gestational diet were comparable to the controls, by which they had lower cumulative food intake than the RCHF and LCHF offspring. However, there was no significant difference in weight gain between the HCHF offspring compared with all 3 groups, suggesting that the addition of choline to a high-FA diet mitigated some but not all the metabolic programming effects of high FA intake alone. This study also contrasts our previous report showing that a high (2.5-fold) choline gestational diet in the presence of recommended FA content leads to higher food intake and body weight in the adult offspring (15). These findings suggest that the effect of maternal choline intake is dependent on the content of other nutrients in the maternal diet. The protective effect of maternal choline supplementation on the development of obesity in the offspring has been previously described (31–33). In these reports, female mice were maintained on a 4.5-fold choline-supplemented diet for 4 wk prior to mating and throughout gestation and both the mothers and their offspring were maintained on a high-fat diet. This shows that the in utero programming effects of choline may vary in response to the dose, period of intake, as well as the macro- and micronutrient composition of both the maternal and postweaning diet. Since the rat offspring in our current study were all maintained on a normal-fat diet for 20 wk, an obesogenic phenotype was not induced. It is possible that exposing the offspring to an obesogenic environment postweaning may help exacerbate the effects of the gestational diets on adiposity and other biomarkers of obesity. We have previously reported that a high gestational methyl vitamin diet (10-fold the recommendation for FA, vitamin B-6, and vitamin B-12) was associated with higher intrabdominal adiposity in offspring weaned to a high-fat (60%) diet for 8 wk compared with those born from dams fed a control diet (6). Thus, the interaction effect of the gestational diet with the postweaning diet as well as the time of measurement of body weight and adiposity may contribute to the observed outcome.

At birth, pups born to dams fed either RCHF or LCHF diets had higher plasma concentrations of leptin and insulin compared with those born to the HCHF-fed dams. Although feed-deprived plasma leptin and insulin concentrations in the adult offspring were not affected by maternal diet, the effect of the high-FA gestational diets on food intake and body-weight gain persisted later in life. Although the exact mechanism remains unclear, both leptin and insulin are known potent neurotrophic factors at birth that regulate hypothalamic neural development, which is fully developed by 4 wk after birth (34). Thus, dysregulation in hormone signaling at birth may have led to changes in hypothalamic neural development that associate with later-life hyperphagia and increased weight gain during adulthood (35).

The balance between FA and choline in the gestational diets affected brain choline and one-carbon metabolism in offspring both at birth and during adulthood. In newborn pup brains, both free choline and betaine concentrations were higher in those from the HCHF diet than from the LCHF diet but were not different from the control diet. The observation that brain choline reflects dietary choline intakes during gestation is consistent with a previous report of a direct relation between maternal dietary choline intake and fetal brain choline metabolite concentrations (36). Additionally, all high-FA gestational diets led to reduced pup brain methionine concentrations compared with the control. Methionine is an essential nutrient for fetal brain development both as a methyl donor, derived from folate and betaine in the one-carbon cycle, and as an amino acid required for protein synthesis. However, there were no differences in measures of methylation potential between groups in pup brain concentrations of SAM, SAH, or SAM:SAH, warranting further investigation of the impact of these diets on DNA methylation in the long-term effects of the maternal diets. During adulthood, offspring born to dams fed the imbalanced LCHF diet had the lowest brain free choline concentrations compared with offspring from dams fed the recommended or supplemented choline diets. These findings suggest that the low in utero availability of choline to the fetus may have imprinted on brain choline homeostasis and metabolism later in life. A previous study has reported that maternal choline supplementation exerts lasting effects on offspring choline metabolism, including upregulation of the hepatic phosphatidylethanolamine N-methyltransferase (PEMT) pathway and enhanced provision of choline and PEMT-derived phosphatidylcholine to the brain (37). Therefore, it is possible that the LCHF diet during pregnancy may be associated with inadequate provision of choline to the brain later in life, leaving uncertain the implications of this imbalance on long-term brain development and cognitive function.

Some evidence for a long-term programming effect of the gestational diets on the one-carbon cycle was also indicated in plasma one-carbon metabolites at 20 wk postweaning. Offspring born to dams fed the HCHF diet had lower concentrations of plasma total homocysteine and cystathionine, but not methionine, compared with the RCHF and LCHF groups, respectively. Cystathionine is derived from homocysteine in the trans-sulfuration pathway and both metabolites are known biomarkers for cardiovascular diseases and other metabolic disorders (38, 39). Overall, adult offspring born to mothers fed the LCHF diets averaged 20% higher concentrations of cystathionine than those fed the more balanced RCRF or HCHF diets. Additionally, the LCHF diet led to the lowest plasma 5-MTHF concentrations but was similar in all other groups with higher choline additions. The mechanisms underlying the lasting effect of the maternal diets on the one-carbon cycle in this study remain to be determined. However, a possible mechanism may be supported by previous studies showing that excessive FA intake in the presence of low choline in gestational diets inhibits methylenetetrahydrofolate reductase activity, the enzyme that catalyzes 5-MTHF synthesis, leading to delays in embryonic development and later-life metabolic disorders associated with less efficient utilization of folate in methylation reactions (40, 41).

High FA intakes affect in utero DNA methylation-dependent alterations in hypothalamic gene expression, and are associated with later-life food intake dysregulation and weight gain in the offspring (5). The RNA-seq analysis at birth provides additional evidence of a potential in utero programming effect of the gestational diets on hypothalamic food intake regulatory mechanisms. Differential expression of 144 genes was observed in the whole hypothalamus of newborn pups from dams fed the HCHF diet compared with the controls. Two target genes involved in regulation of food intake were validated, namely Kl and Crh. Kl mRNA expression was downregulated in the hypothalamus, consistent with recent evidence for its novel role in central food intake regulation in mice (42). Additionally, the maternal HCHF diet resulted in downregulation in Crh gene expression, a main contributor of cortisol production. Higher stress induces higher CRH-stimulated cortisol release, leading to increased food intake (43), and is associated with negative long-term health effects (44, 45). These results are consistent with a recent clinical trial showing that choline supplementation during pregnancy reduced CRH cord blood concentrations in newborns (22). In contrast to the effects at birth, mRNA expression of Kl and Crh in the ARC due to the maternal diets was not affected at 20 wk postweaning, adding further evidence that the primary effect of maternal diets on programming the phenotype of offspring may be due to in utero effects as well as plasticity of the brain postweaning. DEGs known to be involved in phospholipid and fatty acid metabolism [i.e., choline kinase B (46), phospholipase A2 group IVB, and carnitine palmitoyltransferase 1B (47)] and post-translational modification of histones [histone deacetylase (Hdac)-3 and -6] (48) were also found (Supplemental Table 5). Functional term enrichment analysis showed several upregulated and downregulated genes were involved in multiple biological processes and KEGG pathways related to axon guidance, cell signaling, DNA damage, and response to oxidative stress (see Supplemental Tables 3 and 4). These findings suggest that the effect of the gestational diets on adult offspring may also be related to other metabolic pathways and possibly epigenetic programming mechanisms that warrant future investigation.

A limitation of this study is that RNA-seq analysis was performed in the whole hypothalamus of the offspring at birth as a preliminary exploration of the effects of the gestational choline and FA diets on gene expression. This may have introduced greater variability, leading to the observed modest changes in the expression of genes that are specific to food intake and energy expenditure regulation. To further elucidate the mechanisms underlying the observed phenotypic changes in the offspring, further research is needed to assess the effect of the gestational diets on gene and protein expression within distinct hypothalamic nuclei. Another limitation of this study is that sex differences in response to the gestational diets were not explored. In addition, the application of these studies to human diets is uncertain. Although these observations arise from utilizing a rat model, the findings are consistent with clinical trials. Choline supplementation has been shown to benefit fetal brain development and functions and metabolic outcomes (49). In addition, a higher requirement for choline during pregnancy (50) has been found in mothers who have been maintained on prenatal supplements containing 2-fold the recommendation for FA.

In conclusion, increased choline in gestational diets modified the programming effects of high FA on long-term food intake regulation, plasma energy-regulatory hormones, one-carbon metabolism, and hypothalamic gene expression in male Wistar rat offspring, emphasizing a need for more attention to the choline and FA balance in diets of women of childbearing age.

Notes

This research was supported by the Canadian Institute of Health Research, Institute of Nutrition, Metabolism and Diabetes (CIHR-INMD; MOP-130286), and the Natural Sciences and Engineering Research Council of Canada (NSERC-RGPIN-2016-06639).

Author disclosures: The authors report no conflicts of interest.

Supplemental Tables 1–5, Supplemental Figures 1–4, and Supplemental Data Sheets 1–3 are available from the “Online Supporting Material” link in the online posting of the article and from the same link in the online table of contents at http://jn.nutrition.org.

Abbreviations used: ARC, arcuate nucleus of the hypothalamus; Crh, corticotrophin-releasing hormone; DEG, differentially expressed gene; FA, folic acid; GO, Gene Ontology; HCHF, high choline and high folic acid; KEGG, Kyoto Encyclopedia of Genes and Genomes Pathway; Kl, Klotho; LCHF, low choline and high folic acid; PEMT, phosphatidylethanolamine N-methyltransferase; RCHF, recommended choline and high folic acid; RCRF, recommended choline and recommended folic acid; RNQ-seq, RNA sequencing; SAH, S-adenosylhomocysteine; SAM, S-adenosylmethionine; 5-MTHF, 5-methyltetrahydrofolate.