Sperm DNA methylation defects in a new mouse model of the 5,10-methylenetetrahydrofolate reductase 677C>T variant and correction with moderate dose folic acid supplementation

Abstract

5,10-Methylenetetrahydrofolate reductase (MTHFR) is an enzyme that plays a key role in providing methyl groups for DNA methylation, including during spermatogenesis. A common genetic variant in humans (MTHFR 677C>T) results in reduced enzyme activity and has been linked to various disorders, including male infertility. A new animal model has been created by reproducing the human equivalent of the polymorphism in mice using CRISPR/Cas9. Biochemical parameters in the Mthfr 677TT mice recapitulate alterations found in MTHFR 677TT men. Our aims were to characterize the sperm DNA methylome of the Mthfr 677CC and TT mice on a control diet (2 mg folic acid/kg diet) and assess the effects of folic acid supplementation (10 mg/kg diet) on the sperm DNA methylome. Body and reproductive organ weights, testicular sperm counts, and histology were examined. DNA methylation in sperm was assessed using bisulfite pyrosequencing and whole-genome bisulfite sequencing (WGBS). Reproductive parameters and locus-specific imprinted gene methylation were unaffected by genotype or diet. Using WGBS, sperm from 677TT mice had 360 differentially methylated tiles as compared to 677CC mice, predominantly hypomethylation (60% of tiles). Folic acid supplementation mostly caused hypermethylation in sperm of males of both genotypes and was found to partially correct the DNA methylation alterations in sperm associated with the TT genotype. The new mouse model will be useful in understanding the role of MTHFR deficiency in male fertility and in designing folate supplementation regimens for the clinic.

Introduction

DNA methylation occurs when a methyl group is added to the fifth carbon of cytosine in DNA. This well-studied epigenetic modification occurs most commonly in the context of CpG dinucleotides, of which there are ∼32.3 million in the human genome (Greenberg and Bourc'his, 2019; Gershman et al., 2022). DNA methylation patterns are established by DNA methyltransferase (DNMT) enzymes (Okano et al., 1998) and can be modified in response to environmental factors. DNMTs catalyze the reaction in which methyl groups from S-adenosylmethionine (SAM) are transferred to cytosines.

The production of SAM involves one-carbon metabolism, a series of interlinking pathways that include the methionine and folate metabolic cycles (Clare et al., 2019); the latter pathway utilizes folate, a B vitamin known to be important for numerous biochemical reactions. Folate must be acquired through dietary intake, as it cannot be synthesized by humans. Folic acid is a stable synthetic form of folate used in supplements. Once absorbed in the body, folic acid is converted to derivatives by several enzymes in the folate cycle. One key regulatory protein in folate metabolism is 5,10-methylenetetrahydrofolate reductase (MTHFR), which generates the folate derivative 5-methyltetrahydrofolate (5-MTHF). 5-MTHF is shuttled to the methionine cycle and used to convert homocysteine into methionine, which can then be used to produce SAM. As the universal methyl donor, SAM is used for methylation reactions such as DNA methylation (Goll and Bestor, 2005; Ducker and Rabinowitz, 2017). Dietary and/or genetic disruptions of enzymes involved in one-carbon metabolism can negatively impact folate metabolism and the generation of SAM.

In humans, the common 677C>T genetic variant of the MTHFR gene results in an alanine to valine substitution at amino acid 222 (A222V). This substitution alters the structure of the enzyme rendering it thermolabile, and reducing MTHFR-specific activity in homozygous 677TT genotype individuals to 40–50% of the wildtype 677CC genotype values when measured at 37°C (Kang et al., 1991; Rozen, 1997). A meta-analysis has found that the global frequencies of the T allele and the 677TT genotype are around 24 and 7.7%, respectively, with the highest prevalence being found in the Caucasian population (Yadav et al., 2017). A more recent study examining African-American, Caucasian, and Hispanic populations in the USA found a 30% T allele and 10% 677TT genotype frequency, with Hispanics having the highest prevalence (Graydon et al., 2019).

Individuals with the 677TT genotype are predisposed to various disorders, including male infertility in some populations (Gong et al., 2015; Liew and Gupta, 2015). Given that this genotype does not always result in male infertility, it has been proposed that reduced dietary folate availability or genetic background effect, in combination with the 677TT genotype, likely leads to male infertility. The 677TT genotype is associated with lower folate status, higher concentrations of circulating homocysteine, and reduced lymphocyte DNA methylation (Frosst et al., 1995; Friso et al., 2002). To compensate for their lower levels of MTHFR activity, 677TT individuals are often given folate supplements to help stabilize the enzymatic activity and improve the function of this thermolabile protein. However, the ideal dose and form of folate for supplementing those with 677TT genotype is unknown.

To better understand the physiological impact of MTHFR deficiency caused by the 677TT genotype, an animal model was created consisting of a knockout mouse, with a targeted inactivation of the Mthfr gene (Chen et al., 2001). As the haploinsufficient mice (Mthfr^+/–) had roughly 50% of the wildtype MTHFR activity, albeit not the thermolability, they were used in initial studies to emulate individuals with the MTHFR 677TT human genotype (Kang et al., 1991). In early reproductive studies in BALB/c and C57BL/6 background Mthfr^+/– male mice, we reported normal sperm counts, testis histology, and fertility as compared to wildtype littermates (Kelly et al., 2005; Chan et al., 2010). In contrast, in Mthfr^–/– males, with a complete absence of MTHFR activity, strain-specific alterations in testicular histology, sperm count, and fertility were found, revealing an important role of genetic background.

Recently, a new mouse model has been developed using clustered regularly interspaced short palindromic repeats/CRISPR-associated protein 9 (CRISPR/Cas9) to introduce a point mutation into the mouse genome (Reagan et al., 2022). In these mice, the polymorphism altered 806C>T in the Mthfr gene, leading to the A262V amino acid substitution; this alteration is equivalent to the human 677C>T polymorphism and corresponding A222V mutation in humans. This mouse is referred to as the Mthfr 677C>T model, reflecting the human nomenclature. This new model better reflects the biochemical parameters found in humans, including thermolability and enzymatic activity of the MTHFR protein. In addition, metabolic changes in TT mice, such as increased plasma homocysteine levels, were similar to human clinical data (Reagan et al., 2022). Therefore, this new Mthfr 677C>T mouse model provides an opportunity to better understand how MTHFR deficiency, emulating that of 677TT men, affects the sperm DNA methylome as well as the impact of folic acid supplementation. Our objectives were to characterize basic reproductive parameters and the sperm DNA methylome in the Mthfr 677C>T variant mouse model by comparing CC and TT mice. Additionally, we studied the effects of diets supplemented with moderate doses of folic acid on sperm DNA methylation of mice of the two genotypes.

Materials and methods

Mice and diets

The Mthfr 677C>T mouse strain, on the C57BL/6J background (Jackson Laboratory, Bar Harbor, ME, USA), has previously been studied by our group (Reagan et al., 2022). To produce experimental animals, mice heterozygous for Mthfr 677C>T were intercrossed to produce litter-matched male and female Mthfr CC, Mthfr CT, and Mthfr TT mice as previously described in Reagan et al. (2022).

The experimental design (Fig. 1A) consists of four groups of male C57BL/6J mice: Mthfr TT mice and wildtype mice as control (Mthfr CC) on different diets. The diets consisted of a control diet (CD, 2 mg folic acid/kg diet) and moderate dose folic acid-supplemented diet (FASD, 5-fold supplemented: 10 mg folic acid/kg diet) (Fig. 1A). Mice were placed on their respective diets at 4 weeks of age and maintained on them for 4 months. The treatment duration allowed the exposed mice to complete approximately two rounds of spermatogenesis and sperm maturation. The amino acid-defined diets (Inotiv, West Lafayette, IN, USA) follow AIN93 recommendations (Reeves, 1997) and have been previously used by our laboratories (Christensen et al., 2016, 2018). Succinylsulfathiazole was added to inhibit folate production by intestinal bacteria. Following 4 months on the different diets, mice were euthanized, and tissues and organs were collected (see below). Experiments and procedures were carried out in conformity with the Canadian Council on Animal Care guidelines and the study was approved by the Animal Care Committee of the Research Institute of the McGill University Health Centre (RI-MUHC). Mice were housed under a 12-h light:12-h dark cycle in a temperature and humidity-controlled environment at the RI-MUHC-specific pathogen-free animal facility with food and water ad libitum.

Figure 1.

Experimental design and general health of mice unaffected by genotype or diet. (A) Male Mthfr CC and TT mice were fed control diets (CD, 2 mg/kg/diet) or folic acid-supplemented diets (FASD, 10/mg/kg/diet) starting at 4 weeks of age. After 4 months, mice were euthanized and (B) body weights and (C–E) reproductive organ weights were compared for the different genotype/diet groups. Values shown as mean ± SEM. One-way ANOVA with multiple comparisons were performed for (B–E). CC, wildtype genotype; TT, homozygous Mthfr 677C>T polymorphism.

Tissue collection and sperm DNA isolation

Mice were euthanized by being anesthetized with isoflurane, followed by CO₂ asphyxiation and exsanguination by cardiac puncture. Next, emptied seminal vesicles, paired testes, and epididymides were removed, trimmed free of fat, and weighed. The right testis was frozen in liquid nitrogen, and the left testis was immersed in Bouin’s fixative (Sigma-Aldrich, Oakville, ON, Canada) for at least 24 h, washed and placed in 70% EtOH. Mature spermatozoa from the paired cauda epididymides were collected as previously described (Chan et al., 2012). Briefly, cauda epididymides were excised, cut in PBS, and agitated on a shaker for 10 mins at room temperature. The supernatant was filtered through a 40-μm cell strainer and kept on ice. This step was repeated, and the filtered supernatant was combined with the previous one. The sperm suspension was centrifuged (1800g for 5 min at 4°C) to pellet cells then washed with a hypotonic solution (0.45% NaCl) a minimum of three times. Samples were spun down in-between washes (1800g for 5 min at 4°C). After the last spin, supernatant was removed, conserving as much as possible the sperm pellets, which were immediately frozen at –80°C until use.

Sperm DNA extraction

Sperm were lysed in a buffer containing a final concentration of 150 mM Tris, 10 mM EDTA, 40 mM dithiothreitol (Invitrogen, Burlington, ON, Canada), 2 mg/ml proteinase K (Invitrogen, Burlington, ON, Canada), and 0.1% sarkosyl detergent (Sigma-Aldrich, Oakville, ON, Canada) and were incubated overnight at 37°C. Following lysis, DNA was then extracted using the QIAamp^® DNA Mini kit (Qiagen, Toronto, ON, Canada) according to the manufacturer’s protocols.

Genotyping

Genotype confirmation was performed on the mouse sperm DNA samples (n = 6/group). Primers were used to amplify the region containing the Mthfr 677C>T polymorphism (Supplementary Table S1) by PCR. The Sal1 restriction enzyme recognizes the sequence “GTCGAC.” Amplicons containing the T-allele are digested, creating two bands of 351 and 242 bp; C-alleles appear as a single band of 593 bp. An enzyme mix was prepared using 1 μl of Sal1 (Invitrogen, Burlington, ON, Canada), 2.5 μl of buffer, and 16.5 μl water. Next, 20 μl of the enzyme mix were added to 5 μl of each individual PCR product and incubated at 37°C overnight to ensure proper enzyme digestion. PCR amplicons were analyzed by running on a 2% agarose gel electrophoresis and visualized by ethidium bromide.

Sperm counts

Frozen right testes were weighed and homogenized (Polytron; Brinkmann Instruments, Riverview, FL, USA) for 2× 15 s in 5 ml homogenizing solution on ice (0.9% NaCl, 0.1% thimerosal, and 0.5% Tween-20). Elongated spermatozoal heads, which are resistant to homogenization, were counted using a hemocytometer, as described by Kelly et al. (2005). Counts for four hemocytometric chambers were averaged and the number of spermatozoal heads per testis was calculated.

Testicular histology

Fixed left testes were dehydrated through a graded series of ethanol into xylene and embedded in paraffin. Tissue blocks were bisected laterally and embedded in paraffin, cut into 5-μm thick serial transverse cross sections, and every sixth section was stained with hematoxylin and eosin. Four to eight sections from each animal were examined under an upright fluorescence Leica DM6000 B microscope (Wetzlar, Germany) and imaged using a digital camera with Leica Application Suite with DMControl for Windows software. Normal tubules were identified as having all germ cell types in the majority of the seminiferous epithelium, but could contain some mild alterations (e.g. small and few vacuoles). Additionally, complete spermatogenesis was determined by assessing whether all 12 stages of the seminiferous epithelium were present across the sections analyzed. Abnormal tubules, as described in Karahan et al. (2021), were identified as possessing at least one of the following criteria: asymmetric distribution of germ cells within the tubule, germ cells present in a part of the tubule cross-section but not on the other part; tubule with early germ cells but missing spermatids; tubule with spermatozoa and spermatids but no apparent early germ cells; and Sertoli cell-only phenotype, i.e. tubule has only Sertoli cells, no noticeable germ cells within the tubule. The number of abnormal tubules in 100 seminiferous tubules examined for each mouse was used to calculate the proportion of abnormal tubules.

Bisulfite pyrosequencing for imprinted gene methylation

Sperm DNA samples were screened for possible somatic cell contamination through bisulfite pyrosequencing of the imprinting control regions (ICRs) for paternally methylated (H19) and maternally methylated (small nuclear ribonucleoprotein polypeptide N (Snrpn) and paternally expressed gene 1/mesoderm-specific transcript (Peg1/Mest)) imprinted genes. Assessment of ICR methylation also provided an indication of the impact of the Mthfr genotype and folic acid diets on these genes, as they are known to be critical for offspring development.

For all pyrosequencing assays, 500 ng of sperm DNA underwent bisulfite conversion with the EpiTect^® bisulfite kit (Qiagen, Toronto, ON, Canada), according to the manufacturer’s protocol. Bisulfite PCR was conducted using primers and pyrosequencing was performed as described in Dejeux et al. (2009). Briefly, regions of interest were PCR-amplified with one of the primers being biotinylated. Capture of the biotinylated strand was performed with streptavidin-coated sepharose beads and washed using the PyroMark^® Q24 Vacuum Workstation (Qiagen, Toronto, ON, Canada). A sequencing primer was annealed to the isolated captured template strand and the pyrosequencing reaction was conducted using the PyroMark^® Q24 kit (Qiagen, Toronto, ON, Canada), as per the manufacturer’s protocol. Primer sequences utilized for PCR amplification and sequencing were acquired from previously published studies and can be found in Supplementary Table S1 (Susiarjo et al., 2013; de Waal et al., 2014). For somatic cell contamination screening, sperm DNA samples were considered to be contaminated if methylation across all analyzed CpGs in imprinted genes deviated from the expected high levels of methylation for H19 (>90%) and low levels for Snrpn and Peg1/Mest (<10%). All studied samples met the criteria for purity and thus, no sperm DNA samples were excluded.

Genome-wide DNA methylation analyses

Whole-genome bisulfite sequencing (WGBS) was performed at the McGill Genome Centre on the folic acid-supplemented and control samples (n = 6/group, 24 total). Between 2 and 3 μg of sperm DNA per sample were sent to ensure the best quality of library. The WGBS library was created using the KAPA High Throughput Library Preparation kit (Roche/KAPA Biosystems, Oakville, ON, Canada), as previously described (Chan et al., 2019). Briefly, sperm DNA was spiked with 0.1% (w/w) unmethylated λ and pUC19 DNA (Promega, Madison, WI, USA). DNA was sonicated and fragment sizes of 300–400 bp were confirmed on a Bioanalyzer DNA 1000 LabChip (Agilent, Mississauga, ON, Canada). Following fragmentation, DNA-end repair of double-stranded DNA breaks, 3'-end adenylation, adaptor ligation, and clean-up steps were performed according to KAPA Biosystems’s protocol. The resulting bisulfite DNA was quantified with OliGreen (Life Technology/Thermo Fisher, Toronto, ON, Canada) and amplified with 9–12 PCR cycles using the KAPA HiFi HotStart Uracil + DNA Polymerase kit (Roche/KAPA Biosystems, Oakville, ON, Canada), according to suggested protocols. The final WGBS library was purified using Agencout AMPure Beads (Beckman Coulter, Brea, CA, USA), validated on Bioanalyzer High Sensitivity DNA LabChip kits (Agilent, Mississauga, ON, Canada), and quantified by PicoGreen (Thermo Fisher, Toronto, ON, Canada). The samples were sequenced over two lanes on the Illumina NovaSeq6000 S4 v1.5 (San Diego, CA, USA), PE150, 12 samples per lane. Sequencing data were processed with the GenPipes Methyl-Seq pipeline (Bourgey et al., 2019). This pipeline includes protocols to align reads to the reference genome, methylation calling, removal of duplicate reads, and generate metric files.

Downstream analyses

ICR locus coordinates

The location of 24 ICRs was obtained for imprinted genes from a previously published study (Strogantsev et al., 2015). The list included four ICRs of paternally methylated imprinted genes, Igf2/H19, Dlk1-Meg3, Rasgrf1, and zinc finger DBF-type containing 2 (Zdbf2), and 20 ICRs of maternally methylated imprinted genes, Gpr1, Nnat/Peg5, Mcts2/H13, Gnas_ProXL, Gnas_Ex1A, Fkbp6, Peg10, Mest, Nap1l5, Snrpn, Peg3, Inpp5f_v2, Cdh15, Zac1/Plagl1, KvDMR1, Grb10, Zrsr1, Peg13, Igf2r, and Impact. WGBS data were then intersected, using Bedtools software (Quinlan and Hall, 2010), with the mm10 coordinates to retrieve specific ICR methylation data. The resulting list was used to calculate individual and mean DNA methylation values for each different group of samples.

Differentially methylated tiles

For WGBS data, DNA methylation analysis was performed using methylKit software (Akalin et al., 2012). Here, methylKit was used for the identification of differentially methylated tiles (DMTs, 100 bp tiles) between two groups, where ≥10× CpG coverage in all samples and a difference of at least 5% methylation is required. A Benjamini–Hochberg false discovery-based method for P-value correction and passing the q-value threshold (q = 0.01) were used.

Genomic annotations

DMTs were annotated with HOMER software version 4.9.1 (Heinz et al., 2010). Hypermethylated and hypomethylated DMTs were annotated separately.

Gene ontology

Gene ontology (GO) term pathway enrichment was first performed on the list of genic DMTs only, using the Bioconductor R package topGO (version 2.42.0) (Alexa and Rahnenführer, 2023). All significant terms discovered (weighted Fisher P < 0.05) were imported to the R package Reduce + Visualize GO (rrvgo), which helps reduce GO terms by identifying redundant terms based on semantic similarity (Sayols, 2023). Terms were reduced using high stringency (threshold = 0.9) and clustered based in the parentTerm size by omitting the score parameter.

Statistical analyses

Data were analyzed using GraphPad Prism 9 (GraphPad Prism version 9.4.0 for Windows, GraphPad Software, San Diego, CA, USA) and are presented as mean ± SEM or absolute values, as indicated. DNA methylation at ICRs from bisulfite pyrosequencing and WGBS results were determined by averaging all individual CpGs within a given ICR, and then averaged across all animals within diet groups. Hyper- and hypomethylated DMTs were compared to a background set of tiles (i.e. all tested tiles from the methylKit analysis) and enrichment in genomic regions/repetitive elements was determined through Chi squared test with Yate’s correction. Other statistical calculations were made by one-way ANOVA followed my multiple comparison or unpaired Student’s t-test. Histograms of methylation differences and scatterplots of methylation values were generated using R (version 4.0.3) and RStudio version 1.4.1103 (RStudio Team, 2021). P ≤ 0.05 was considered significant for all analyses. Statistical analysis calculating percentage correction following FASD for individual DMTs was calculated using group mean values in Abbott’s formula [$1-\left(\frac{CC\_\mathrm{CD}\hspace{0.17em}-\hspace{0.17em}TT\_\mathrm{FASD}}{CC\_\mathrm{CD}\hspace{0.17em}-\hspace{0.17em}TT\_\mathrm{CD}}\right)]\times 100$ .

Results

Mthfr genotype and folic acid-supplemented diets did not alter body weights, reproductive organ weights, testicular sperm counts, or testicular histology

We initially assessed whether the Mthfr 677TT genotype and/or FASD affected the overall health or reproductive organ weights of male mice. Similar body and reproductive organ weights were observed in all groups (Fig. 1B–E). Reflecting the lack of alteration in testis weights, testicular sperm counts were unaffected by either genotype or diet (Fig. 2A). An examination of testicular histology demonstrated that >97% of seminiferous tubules were normal across all four groups, with all 12 stages of the seminiferous epithelium present (Fig. 2B and C); only a small percentage of tubules (2–3%) were found to be abnormal, with defects such as vacuolization, asymmetric distribution of germ cells, and Sertoli cell-only tubules found (data not shown). These results indicate that neither the Mthfr 677TT genotype nor folic acid supplementation affected the general or reproductive health of the animals.

Figure 2.

Mouse sperm counts and testicular histology were unaffected by genotype or diet. (A) Testicular sperm counts were performed from all groups. (B) The proportion of abnormal testicular tubules of all groups studied were evaluated from (C) 20× histological cross sections of the testes. Values in (A) are shown as mean ± SEM and one-way ANOVA with multiple comparisons were performed. Black scale bar located at the bottom right (C) represents 200 μm. CC, wildtype genotype; CD, control diet; FASD, folic acid-supplemented diet; TT, homozygous Mthfr 677C>T polymorphism.

Methylation of imprinted loci in sperm was unaffected by Mthfr genotype or folic acid supplementation

To determine the impact of the Mthfr 677C>T polymorphism alone and in combination with supplementation, we first investigated the methylation status of imprinted genes. A representative subset of samples (n = 6/group) was selected for sperm DNA isolation; this subset did not differ from the entire group. To assess the purity of the sperm samples, locus-specific bisulfite pyrosequencing was utilized to assess the germline ICRs of the paternally methylated H19 and the maternally methylated Snrpn and Peg1-imprinted genes. The assays were designed to analyze methylation status of six CpGs in H19 and five in Snrpn and Peg1. Initial analyses with bisulfite pyrosequencing of sperm DNA demonstrated that the methylation status of the three aforementioned imprinted loci was within the normal range in all groups, with high levels of methylation (>90%) of the paternally methylated and low levels (<10%) in the maternally methylated ICRs (Supplementary Fig. S1A).

Genome-wide methylation analyses were then performed on these same samples, where WGBS was conducted on control and folic acid-supplemented samples. Data from this approach were used to investigate in greater depth the same imprinted genes analyzed by bisulfite pyrosequencing, as well as other known ICRs. WGBS data with at least 10× coverage were able to provide information for a total of 24 ICRs, analyzing from 6 to 172 CpGs within a given ICR. In all groups, the methylation levels were within the expected ranges for paternally and maternally methylated imprinted genes (Supplementary Fig. S1B and C, respectively), with the exception of Zdbf2, which showed areas with varying levels of methylation, possibly non-imprinted differentially methylated regions. Overall, sperm DNA methylation at the ICRs of imprinted genes was not affected by genotype or folic acid supplementation.

The Mthfr TT genotype results in a predominant hypomethylation of sperm DNA

We used WGBS to determine the genome-wide impact of the Mthfr 677C>T polymorphism on the sperm DNA methylome. The effect of genotype was examined by comparing Mthfr TT versus CC mice on CD. There were 360 DMTs found when comparing the sperm of Mthfr TT versus Mthfr CC mice, with a predominance of hypomethylated tiles (219/360, or 60%) (Fig. 3A, left and Table 1). The greatest alterations in methylation were found mainly in areas of intermediate (20–80%) methylation (Fig. 3A, right and Supplementary Data File S1). While the majority of the methylation changes observed were between 5 and 10% in magnitude, larger magnitude changes (10–25%) were also seen, especially for hypomethylation (Supplementary Fig. S2A). Annotation of DMTs demonstrated that both hyper- and hypomethylated tiles were mostly located in intergenic (53–60% of tiles) and intronic areas (32–40%), while a smaller proportion was found in promoter-transcription start site regions (0.7–1.8%), as compared to a background set of all tested tiles (Fig. 3B). Statistical analyses showed that exons (4.2–5.0%) were significantly enriched owing to the genotype effect for both hyper- and hypomethylated DMTs (P < 0.0001).

Figure 3.

Genotype effect from whole-genome bisulfite sequencing, genomic distribution, repeat sequences, and gene ontology terms. (A) Summary of differentially methylated tiles (DMTs, left) and scatterplot showing methylation values of Mthfr TT and CC groups (right). (B) Genomic distribution of background, hyper-, and hypomethylated DMTs. (C) Distribution of repetitive elements with DMTs. (D) Gene ontology term analysis was performed on all genic hyper- and hypomethylated DMTs separately. All significant terms were reduced using high stringency (threshold = 0.9) using the R package Reduce + Visualize GO (rrvgo). The dotted line indicates the P < 0.05 threshold for significance for FDR. Statistical analyses performed for (B and C) were Chi squared with Yate’s correction tests; ****P < 0.0001. CC, wildtype genotype; CD, control diet; FDR, false discovery rate; hyper, hypermethylated; hypo, hypomethylated; LINE, long interspersed nuclear element; LTR, long terminal repeat; SINE, short interspersed nuclear element; TSS, transcription start site; TT, homozygous Mthfr 677C>T polymorphism; WGBS, whole-genome bisulfite sequencing.

Table 1.

Summary of the number of differentially methylated tiles discovered from each effect studied using whole-genome bisulfite sequencing.

Effect studied	Group comparison	Hypermethylated	Hypomethylated	Total*
Genotype effect	TT CD versus CC CD	141	219	360
FASD effect in Mthfr CC	CC FASD versus CC CD	137	108	245
FASD effect in Mthfr TT	TT FASD versus TT CD	194	131	325
Extreme effect	TT FASD versus CC CD	110	105	215

CC, wildtype genotype; CD, control diet; FASD, folic acid-supplemented diet; Mthfr, 5,10-methylenetetrahydrofolate reductase; TT, homozygous Mthfr 677C>T polymorphism.

^*Average total number of tiles tested = 471 000.

With a large proportion of DMTs found in intergenic regions, we examined whether repetitive elements were affected by the Mthfr genotype. Long terminal repeats (LTRs) elements were significantly increased amongst the hypermethylated DMTs (P = 0.0009, Fig. 3C). Evolutionarily young retrotransposable elements are known to escape reprogramming in germ cells and pre-implantation development. In our previous study examining the model with complete absence of MTHFR, these young transposable elements were found to be enriched in sperm of Mthfr^–/– mice compared to wildtype litter mates (Karahan et al., 2021). With the new Mthfr 677C>T mouse model, young retrotransposable elements were not significantly impacted by genotype, with only a small proportion affected (2–3%), similar to the background levels (Supplementary Fig. S2B).

GO enrichment analysis was performed on both hyper- and hypomethylated genic DMTs. Results revealed interesting terms for biological processes including synaptic membrane adhesion, brain development, locomotory exploration behavior, and neuron differentiation (Fig. 3D). Interestingly, some genic DMTs demonstrating the greatest changes in sperm DNA methylation linked to genotype have been implicated in neurological function and disorders. These include parathyroid hormone 2 receptor (Pth2r), which is involved in central nervous system regulation (Dettori et al., 2023), protocadherin beta 19 (Pcdhb19), a pseudogene related to protocadherin 19 (Pcdh19), and fibroblast growth factor 12 (Fgf12), which are both linked to epilepsy (Samanta, 2020; Seiffert et al., 2022). A full list of significant terms can be found in Supplementary Data File S2.

Folic acid supplementation results in more hypermethylation than hypomethylation of sperm DNA

Next, our analysis focused on the effect of folic acid supplementation on the sperm of both Mthfr CC and TT mice. When compared to the Mthfr CC mice, Mthfr TT mice fed the FASD had a larger number of DMTs in their sperm (325 versus 245 DMTs) (Table 1). For both genotypes, the FASD caused both hyper- and hypomethylation, with a predominance of hypermethylation (TT: 194 hypermethylated and 131 hypomethylated; CC: 137 hypermethylated and 108 hypomethylated) (Fig. 4A and Table 1). While many of the diet-associated sperm DNA methylation differences were <10% in magnitude, there were DMTs showing higher magnitude changes (10–25%), most markedly for the hypermethylated DMTs in the Mthfr TT group. Areas having intermediate levels of methylation showed the greatest alterations owing to the FASD (Supplementary Fig. S3A–D).

Figure 4.

Increased DNA methylation following folic acid supplementation in mice on both Mthfr CC and TT genotypes. (A) Summary of differentially methylated tiles (DMTs) from WGBS. (B) Genomic distribution from hyper- and hypomethylated DMTs compared to background, for the FASD effect on Mthfr CC (left) and TT (right) genotypes. (C) Proportion of hyper- and hypomethylated DMTs overlapping repetitive sequences. Gene ontology term analysis was performed on all genic DMTs for the FASD effect on (D) Mthfr CC and (E) TT genotypes, looking at hyper- and hypomethylated DMTs separately. All significant terms were reduced using high stringency (threshold = 0.9) using the R package Reduce + Visualize GO (rrvgo). The dotted line indicates the P < 0.05 threshold for significance for FDR. Statistical analyses performed in (B and C) were Chi squared with Yate’s correction tests; ****P < 0.0001, **P < 0.01. CC, wildtype genotype; CD, control diet; FASD, folic acid-supplemented diet; FDR, false discovery rate; hyper, hypermethylated; hypo, hypomethylated; LINE, long interspersed nuclear element; LTR, long terminal repeat; Mthfr, 5,10-methylenetetrahydrofolate reductase; SINE, short interspersed nuclear element; TSS, transcription start site; TT, homozygous Mthfr 677C>T polymorphism; WGBS, whole-genome bisulfite sequencing.

For both Mthfr CC and TT animals, alterations in sperm DNA methylation were annotated to mainly intergenic (60–68% of tiles) and intronic regions (27–32%) of the genome following the FASD (Fig. 4B). Statistical analyses of the diet effect showed enrichment for exonic regions (5.2–6.8%) in hypermethylated DMTs in both CC and TT genotypes, while these same regions were only enriched at hypomethylated DMTs of Mthfr CC mice (3.7%). A similar analysis was performed on repetitive sequences, with only the proportion of LTRs being enriched in the hypomethylated DMTs for the folic acid effect in Mthfr CC animals (P = 0.0055, Fig. 4C). Young retrotransposons were not over-represented, with ∼3% of DMTs overlapping such sequences (Supplementary Fig. S3E).

GO enrichment analysis using genic DMTs from mice fed FASD revealed shared terms between both genotypes such as cell adhesion and axonogenesis. In addition, other nervous system-related terms, such as regulation of postsynaptic potential, synaptic transmission, and behavioral terms, were observed (Fig. 4D and E). The full list of significant GO terms can be found in Supplementary Data File S2. Taken together, these results suggest a role for MTHFR activity and folic acid status in neurological development and function.

FASD partially corrects changes associated with the TT genotype

The observation that genotype and FASD had opposite effects, in terms of the direction of change in the altered DMTs, led us to hypothesize that a moderate dose of folic acid might have beneficial effects on sperm DNA methylation in Mthfr TT mice. To examine this, we first compared the extreme condition: Mthfr CC mice on CD as compared to Mthfr TT mice on FASD. For the TT FASD versus CC CD comparison, as compared to the genotype comparison (TT CD versus CC CD), there was a decrease in both sperm hypermethylation (110 versus 141 DMTs) and hypomethylation (105 versus 219 DMTs) (Table 1 and Supplementary Fig. S4A). The results suggest that folate supplementation of the TT mice may be correcting the genotype effect. The magnitude of change, genomic/repetitive element distribution and GO terms found were similar to those for the genotype or diet effects (Supplementary Fig. S4B–D).

Next, we examined the genic DMT lists of the Mthfr genotype effect and the diet effect on Mthfr TT mice to see if common genes were affected. A total of seven genes were found to be altered in the opposite direction (Table 2): four genes hypomethylated owing to genotype were found to be hypermethylated due to diet effect (kinesin family member 6 (Kif6), solute carrier family 12 member 1 (Slc12a1), Pcdhb19, and von Willebrand factor A domain containing 3A (Vwa3a)), and vice versa for three genes (Epha3, Pcdh9, and Zfp608); similar to the genes mentioned previously for the genotype effect, several of the common genes here are implicated in the nervous system and, in addition, to cilia function in brain and fertility (Javier-Torrent et al., 2019; Patir et al., 2020; Chan et al., 2023; Takagishi et al., 2023). Interestingly, alterations in sperm DNA methylation of these genes, while in opposite directions, showed a similar magnitude of change.

Table 2.

Direction and magnitude of change from common genes affected by both Mthfr genotype and in Mthfr TT mice on a folic acid-supplemented diet.

Gene name	Gene description	Genotype effect		FASD diet effect on Mthfr TT
		Direction	Change in DNA meth.	Direction	Change in DNA meth.
Kif6	Kinesin family member 6	Hypo	–15.2	Hyper	13.0
Slc12a1	Solute carrier family 12, member 1		–8.5		5.7
Pcdhb19	Protocadherin beta 19		–21.6		21.2
Vwa3a	von Willebrand factor A domain containing 3A		–11.7		12.4
Epha3	Eph receptor A3	Hyper	5.5	Hypo	–6.4
Pcdh9	Protocadherin 9		12.4		–15.8
Zfp608	Zinc finger protein 608		15.8		–15.4

FASD, folic acid-supplemented diet; meth, methylation; Mthfr, 5,10-Methylenetetrahydrofolate reductase; TT, homozygous Mthfr 677C>T polymorphism.

To look beyond effects on individual genes, we proceeded to determine if methylation at DMTs altered by genotype broadly across the genome could be corrected with the moderate FASD. Of the 360 DMTs found due to genotype (141 hypermethylated and 219 hypomethylated), 133 tiles (49 hypermethylated and 84 hypomethylated) were sequenced in sufficient depth of coverage in the TT FASD group, allowing us to investigate whether there was correction of sperm DNA methylation defects. For each of the 133 commonly sequenced tiles, the mean values of all the samples from CC CD, TT CD, and TT FASD groups were plotted. DMTs due to genotype showing hypermethylation in TT CD demonstrated mainly a reduced level of methylation, with TT FASD returning close to CC CD methylation levels (red lines, Fig. 5A, left). Similarly, hypomethylated tiles increased their DNA methylation after supplementation with folic acid (blue lines, Fig. 5A, right). To determine if these changes were statistically significant, we used Abbott’s formula and Student’s t-test. For the hypermethylated tiles, 44 DMTs showed different levels of correction, while only five of them did not correct (blue lines in Fig. 5A, left). The average percentage of correction (38%) was highly significant in this group (Fig. 5B, left). The correction was greater in the hypomethylated group, where 82 DMTs displayed an average correction percentage of 59% with a P < 0.0001 (Fig. 5B, right); only two hypomethylated DMTs did not show correction (red lines in Fig. 5A, right). Similar to the distribution due to genotype or the FASD (Figs 3B and 4B), the majority of the corrected DMTs were mostly found in intergenic regions (48 and 60%), introns (45 and 32%), and exons (7 and 5%), for hyper- and hypomethylated DMTs, respectively (Fig. 5C and Supplementary Data File S3); only two hypomethylated DMTs were found in promoters of genes. These results demonstrated that a moderate dose of folic acid was able to partially correct DNA methylation alterations in sperm associated with the TT genotype.

Figure 5.

Mthfr genotype effects on hyper- and hypomethylated DMTs are corrected when supplemented with a moderate dose of folic acid. (A) Mean methylation values from CC CD, TT CD, and TT FASD were intersected with Mthfr genotype effect differentially methylated tiles (DMTs) (TT CD versus CC CD comparison); hypermethylated (left) and hypomethylated DMTs (right). Each dot represents the average of all six samples in the specified group. Red lines represent tiles that decreased methylation while blue lines represent tiles with increased methylation after supplementing the Mthfr TT genotype with folic acid. (B) Statistical analysis calculating the percentage correction following FASD for individual DMTs was determined using group means values in Abbott’s formula: [$1-\left(\frac{CC\_\mathrm{CD}\hspace{0.17em}-\hspace{0.17em}TT\_\mathrm{FASD}}{CC\_\mathrm{CD}\hspace{0.17em}-\hspace{0.17em}TT\_\mathrm{CD}}\right)]\times 100$. Means $\pm$SEM are shown. Unpaired Student’s t-test was used to compare each group; ****P < 0.0001. (C) Genomic distribution of DMTs corrected following FASD. CC, wildtype genotype; CD, control diet; FASD, folic acid-supplemented diet; Mthfr, 5,10-methylenetetrahydrofolate reductase; TT, homozygous Mthfr 677C>T polymorphism.

Discussion

MTHFR deficiency is implicated in an increased risk of numerous human health disorders, including male infertility. A common MTHFR polymorphism in humans is the 677C>T variant, which results in a thermolabile enzyme in individuals with the 677TT genotype. Here, we used a new animal model with the mouse equivalent of the human 677C>T variant. These mice reproduce biochemical parameters, which are consistent with human clinical data, such as a thermolabile enzyme with reduced enzymatic activity in the liver and elevated plasma homocysteine levels (Reagan et al., 2022). The Mthfr 677C>T mice provided us with an opportunity to better understand the effect of the TT genotype and folate status on basic male reproductive parameters and sperm DNA methylation patterns in a mouse model more accurately representing 677TT men than previously used mouse models. While reproductive parameters, such as organ weights, testicular morphology, and sperm counts, were unaffected, sperm DNA methylation abnormalities were found in the Mthfr 677TT mice, with more decreased than increased methylation observed. Of potential clinical relevance, a moderate dose of supplemental folic acid partially corrected sperm DNA hypo- and hypermethylation abnormalities found in the Mthfr TT mice, returning methylation levels at several genic and intergenic regions back to Mthfr CC levels.

Genomic imprinting is critical for normal mammalian development. Disruptions of these important sequences can cause growth and developmental abnormalities and imprinting disorders such as Beckwith–Wiedemann syndrome, Silver–Russell syndrome, and transient neonatal diabetes. More serious disturbances of imprinted genes can lead to recurrent molar pregnancy, miscarriage, or infertility (Tomizawa and Sasaki, 2012). Since sex-specific DNA methylation patterns on imprinted genes are programmed in the germline, we examined whether the methylation of these sequences was affected by Mthfr genotype and/or folate status. Using locus-specific pyrosequencing and WGBS techniques, we showed that sperm DNA methylation levels of these imprinted genes were unaffected by genotype or diet. WGBS allowed us to obtain greater coverage of CpGs and precise levels of methylation across 24 ICRs. Importantly, the moderate dose of folic acid, while providing increased levels of methyl groups, did not lead to aberrant increased levels of methylation of maternally methylated imprinted genes in sperm.

Transposable elements make up a large portion of the genome: about 45% in humans (Lander et al., 2001) and 37.5% in mice (Mouse Genome Sequencing Consortium et al., 2002). Abnormal methylation at repeat sequences can result in the activation of transposable elements that have the capacity of disrupting genomic stability. In particular, evolutionarily young elements escape reprogramming windows in both germ cells and pre-implantation development, which may act as a mechanism for inter- and transgenerational inheritance (Gagnier et al., 2019). We have previously examined a mouse model with completely absent MTHFR and demonstrated that Mthfr^–/– males exhibited an enrichment of altered sperm DNA methylation at young transposable elements and a worsening reproductive phenotype across generations (Karahan et al., 2021). Interestingly, our results in the Mthfr 677C>T mouse model demonstrated that, while the LTR class of repetitive elements was enriched owing to genotype and diet, the same family of young transposable elements were not over-represented. The results suggest that, unlike the Mthfr^–/– mouse model (Karahan et al., 2021), hypomethylation of young retrotransposons may not contribute to epigenetic inheritance and worsening of reproductive phenotypes across generations in the single nucleotide polymorphism mouse model, but this will require further investigation.

Our study indicates that the general and reproductive health of the Mthfr 677C>T mice was not significantly affected by the TT genotype or the FASD. This is in line with results from a previous model used to emulate human MTHFR 677TT individuals, the haploinsufficient (Mthfr^+/–) mouse knockout (Chen et al., 2001). In terms of sperm DNA methylation, when comparing the TT versus CC genotypes on a CD, we observed hyper- and hypomethylation, with a predominance for loss of sperm DNA methylation. Interestingly, using reduced representation bisulfite sequencing (RRBS) in the Mthfr-haploinsufficient mouse, sperm DNA showed a predominance of hypermethylation in Mthfr^+/– compared to wildtype animals (Aarabi et al., 2018). Furthermore, while methylation at 360 DMTs was affected in the sperm of the Mthfr TT mice in the current study, altered methylation at over 1000 of tiles was affected in the sperm of the Mthfr^+/– mice in our previous study (Aarabi et al., 2018). The differences observed between the two mouse models may be explained by several factors. One possibility is the different techniques used. WGBS, used in the current study, assesses methylation at all ∼20 million CpG sites in the mouse genome, thus providing full coverage of the sperm DNA methylome. In contrast, RRBS interrogates CpG-dense regions, such as CpG islands, owing to its reliance on enzymatic digestion; areas with high CpG density are normally hypomethylated in the genome. Therefore, analysis with RRBS may be biased to demonstrate increases in DNA methylation. Another possibility is that different mouse backgrounds were used; BALB/c in the Aarabi (2018) study and C57BL/6J in the current study. In males, with complete loss of MTHFR protein (Mthfr^–/– mice), adverse reproductive effects were more marked in mice on a BALB/c versus a C57BL/6 background (Kelly et al., 2005; Chan et al., 2010). This explanation could be tested in future experiments by crossing the new Mthfr 677C>T model onto the BALB/c background. An additional possibility is that the new mouse model may more accurately reflect individuals with the MTHFR TT genotype; this could be tested in future experiments by carrying out WGBS on the sperm of these men.

Humans with the MTHFR 677C>T polymorphism are at increased risk for several neurological pathologies such as vascular dementia, Alzheimer’s disease, and cognitive decline (Rajagopalan et al., 2012; Sun et al., 2015; Rai, 2017). The initial paper using the new mouse model demonstrated morphological and vascular abnormalities in the brain of Mthfr TT mice, making it a new genetic model to help study mechanisms leading to such diseases. Our study is the first to examine DNA methylation in this mouse model. Interestingly, alterations in sperm DNA methylation caused by the Mthfr genotype affected genic regions enriched in GO terms for neurological processes such as brain development, neuron differentiation, and synaptic membrane adhesion. For example, the locus demonstrating the greatest loss of methylation owing to genotype was found in the Pth2r gene. This, and similar receptors within the family, have been shown to be expressed in many areas of the brain and proposed to have protective effects on neurodegeneration and mediate regulatory roles in the central nervous system (Dettori et al., 2023). Furthermore, altered methylation was found in Pcdhb19 (second most hypomethylated tile), a pseudo gene of Pcdh19, a cell adhesion molecule located at the synapse. Similarly, methylation was also affected in Fgf12 (most hypermethylated tile), a protein involved in modulating sodium channels involved in synaptic transmission. Both Pcdh19 and Fgf12 have been implicated in epilepsy (Samanta, 2020; Seiffert et al., 2022). Future studies will be needed to determine whether sperm DNA methylation abnormalities in Mthfr TT mice in nervous system-related genes have functional consequences for the next generation.

Small, randomized control studies showed that high doses of folic acid (5 mg daily intake; >10 times the recommended dietary allowance) result in improved sperm concentrations in infertile men (Wong et al., 2002; Ebisch et al., 2007). These early studies led physicians to propose the use of folic acid supplements for their infertile patients. A more recent study of folic acid supplements (5 mg/day for 6 months) in 30 men with idiopathic infertility did not show evidence of improvements in sperm counts but did show altered DNA methylation in sperm, with enhancement of DNA hypomethylation in MTHFR 677TT men (Aarabi et al., 2015); the latter results were confirmed using a more comprehensive and accurate DNA methylation analysis technique, sperm-specific customized capture sequencing (Chan et al., 2019). A recent randomized clinical trial of folic acid (5 mg/day) and zinc (30 mg/day) for 6 months did not show improvements in sperm parameters or effects on the sperm DNA methylome; however, this study used arrays that would not target intergenic regions and did not examine effects of the MTHFR 677TT genotype (Jenkins et al., 2022). A study using BALB/c Mthfr^+/– mice showed the same sperm hypomethylating effect of high-dose folic acid supplementation (10-fold supplementation) (Aarabi et al., 2018). Together, the human and mouse sperm DNA methylation studies suggested that folic acid supplements of 5 mg/day might be deleterious for some men, such as those with the MTHFR 677TT genotype. These studies led us to use lower doses of folic acid in the current study.

In contrast to the sperm DNA hypomethylating effect, we previously reported with high-dose folic acid supplements (10-fold supplementation), the moderate dose supplements used in the current study resulted in more sperm hypermethylation than hypomethylation for both Mthfr CC and TT genotypes. The moderate dose results fit with the postulated beneficial effects of folic acid to increase levels of methyl donors available for methylation reactions. Importantly, the moderate dose of folic acid supplementation was able to partially correct the genotype effects. Folic acid was particularly effective in correcting sperm DNA hypomethylation (82 DMTs) caused by the TT genotype. Interestingly, 44 DMTs were corrected with supplementation by decreasing sperm DNA methylation, and we postulate that FASD may affect other epigenetic marks, such as histone methylation, which may in turn cause hypomethylation of DNA (Lismer and Kimmins, 2023).

While many affected tiles were found in intergenic regions, some DMTs corrected with the addition of FASD were found in interesting genic regions. For example, a DMT in the promoter of dynein axonemal heavy chain 7 (Dnah7c) was found to be hypomethylated due to TT as compared to the CC genotype and demonstrated an increased methylation in TT mice fed the FASD. This gene is a component of the inner dynein arm of the axoneme of cilia/flagella and loss of function of this gene has been implicated in impaired spermatogenesis (Zhang et al., 2002; Gao et al., 2022). In addition, two other genes that were shown to have opposite effects when comparing genotype and diet effects, Vwa3a and Kif6, have been implicated in cilia motility and infertility (Patir et al., 2020; Chan et al., 2023; Takagishi et al., 2023). While in this study, we did not examine fertility in the TT genotype mice, in humans, a predisposition to male infertility has been reported in MTHFR 677TT individuals in some populations (Gong et al., 2015). Strain differences in male fertility have been observed in mice with complete absence of MTHFR (Mthfr^–/– mice) (Chan et al., 2010). It would thus be interesting in follow-up studies to examine the effect of the newly developed Mthfr 677C>T mice on different background strains.

Our findings also demonstrated that GO terms involved in nervous system and brain development were affected in the TT genotype. Interestingly, genes such as kalirin RhoGEF kinase (Kalrn) and Pcdhb19, mentioned previously, were corrected with the FASD, increasing the methylation in the first intron and first exon of these genes, respectively. These genes have been shown to be essential for synaptic function, and nervous system development and disorders (Emond et al., 2009; Parnell et al., 2021). Overall, correction of altered sperm DNA methylation was demonstrated to be statistically significant, suggesting that a moderate dose in individuals with the 677C>T polymorphism might partially help alleviate abnormal epigenetic patterns in sperm.

In conclusion, our initial characterization of the reproductive parameters and sperm DNA methylation in the recently developed Mthfr 677C>T mice (Reagan et al., 2022), opens the door to future experiments to investigate clinically relevant issues related to male infertility. This includes genetic strain effects and advanced paternal age as well as the impact of other types of folate supplementation (e.g. 5-methyltetrahydrofolate). As MTHFR is present at high levels in germ cells in the prenatal testis (Garner et al., 2013), the Mthfr 677C>T mice could also be useful in testing whether and what levels of gestational folate might be able to correct the DNA methylation abnormalities we found in adult sperm postnatally.

Supplementary data

Supplementary data are available at Molecular Human Reproduction online.

Data availability

All sequencing and methylation array data are deposited at the Gene Expression Omnibus (GEO) database accession number GSE240712.

Authors’ roles

J.T., R.R., K.E.C., D.C., and E.M.D.R. designed the study. E.M.D.R., D.C., and K.E.C. performed the experiments and/or analyses of the data. A.M.R. and G.R.H. generated the congenic C57BL/6J Mthfr 677C>T mouse. E.M.D.R., D.C., and J.T. wrote the manuscript. All authors revised the manuscript. All authors read and approved the final manuscript.

Funding

E.M.D.R. was the recipient of a FQRNT Team Grant (Réseau Québécois en Reproduction) Diversity Recruitment Scholarship and an RI-MUHC Studentship. This research was funded by the Canadian Institutes of Health Research (CIHR) to J.T. (FND-148425). This work was also supported by a grant from the CIHR and Genome Canada (CEE-151619) to J.T. and CIHR (PJT-173521) to R.R. J.T. is a Distinguished James McGill Professor. This research was enabled in part by support provided by Calcul Québec and the Digital Research Alliance of Canada.

Conflict of interest

The authors declare that they have no competing interests.