Early life vitamin D depletion and mechanical loading determine methylation changes in the RUNX2, RXRA, and osterix promoters in mice

Early life vitamin D exposure is linked to later skeletal health with maternal vitamin D status in pregnancy associated with neonatal bone mass. The MAVIDOS study has demonstrated that vitamin D supplementation leads to reduced RXRA DNA methylation. Mice exposed to early life vitamin D deficiency have reduced bone mass and bone accrual in response to mechanical loading. Using the tibiae of these mice, we have examined the effect of diet and mechanical loading on the DNA methylation of promoters of genetic loci important for bone growth and development and their association with bone strength. Mechanical loading of mouse tibiae leads to a reduction of RXRA DNA methylation. Early life vitamin D deficiency is associated with altered methylation of osterix and Runx2 in these bones. Tibia strength was also demonstrated to be associated with a change in DNA methylation status in CpGs of the vitamin D receptor (VDR), ostrix, and RXRA genes. We have shown for the first time that mechanical loading of bone and early life vitamin D deficiency leads to changes in the epigenome of this tissue in key genes in the vitamin D and osteoblast differentiation pathway.


Background
Early life exposures have been hypothesised to lead to long-term changes in adult health [1]. Suboptimal early life exposures including maternal nutrition can lead to long-term adverse health effects [2], with poor intrauterine growth, a predictor of later osteoporosis and fracture risk [3,4]. There are both observational and interventional studies that show the effects of altered vitamin D exposure during pregnancy on bone mass accrual [4][5][6][7][8][9], although the mechanisms that are in play are incompletely understood.
One mechanism by which early life environment has been suggested to influence later phenotype is through the altered epigenetic regulation of genes. Epigenetic mechanisms induce heritable changes in gene expression without a change in the gene sequence. Epigenetic mechanisms include DNA methylation which can affect the accessibility of the DNA to the transcriptional machinery [10]. In general, hypermethylation of the promoter region is associated with gene silencing and hypomethylation with gene activity. There is Krstic et al. Genes & Nutrition (2022) 17:7 now substantial evidence from animal models that the early life environment can affect DNA methylation, with changes for instance in maternal diet associated with changes in DNA methylation and expression of key metabolic genes and the metabolic processes they control in the offspring [11]. Maternal undernutrition, overnutrition, low-protein intake, and micronutrient status during pregnancy have all been shown to influence the distribution of methyl marks in offspring [12][13][14][15][16][17]. Similar observations have been observed in humans, with famine exposure in utero linked to altered DNA methylation in the adult offspring [18,19], suggesting that early life environment can induce persistent epigenetic changes in the offspring.
To date, very few studies have examined the effect of vitamin D supplementation on DNA methylation as a potential mechanism by which vitamin D supplementation may affect bone accrual. Altered RXRA methylation has previously been shown to be predictive of bone health in later childhood in humans [20]. Altered methylation at ten previously identified CpG methylation sites in the RXRA locus has been observed in umbilical cord tissue stored at birth from the MAVIDOS study (multicentre, double-blind randomised, placebo-controlled trial of vitamin D 1000 IU/day from 14-week gestation to delivery) in the maternal supplementation group. This study found a positive effect of maternal vitamin D supplementation during pregnancy on neonatal bone mass for winter births (when background 25(OH)D-vitamin D concentrations are lowest). This suggests vitamin D may affect bone development accrual through the epigenetic of regulation of RXRA and related pathways [21,22]. Differential associations between RXRA methylation at birth and offspring bone mass at birth and in later childhood suggest the possibility that other environmental stimuli which affect bone outcomes, for example, mechanical loading, might also modify the epigenetic regulation of RXRA and genes related to bone development.
We have also assessed the response to mechanical stimulation using whole-body vibration (WBV) of a subgroup of the MAVIDOS cohort children. We found that those children whose mothers received vitamin D supplementation had a greater response to WBV as assessed by a rise in the circulating bone formation marker P1NP compared to the children whose mothers received placebo [23].
We previously created a murine model in which pregnant dams were either vitamin D deficient or replete, and their offspring moved to a vitamin D-replete diet at weaning. Tibias of the offspring were mechanically loaded, and offspring of vitamin D-deplete mice showed a lower bone mass in the non-loaded limb in the younger, skeletally immature, age group and reduced bone mass accrual in response to loading in both the growing and mature skeleton [24].
Taken together, both the human and preclinical data suggest that epigenetic processes may mediate the effect of early life environment during the in utero and young postnatal period whilst there is still plasticity in the genome. The accrual of bone mass during skeletal development is clearly complex and multifactorial. We have focused our attention on two key regulators of osteoblastic differentiation -Osterix and Runx2 -as well as the critical components on the vitamin D pathway mediating the response to administered vitamin D, namely the vitamin D receptor (VDR) and retinoic acid receptor (RXRA). Together, they comprise the machinery that transduces the nuclear signalling of vitamin D. Runx2 has been shown to induce differentiation of MSCs into immature osteoblasts, whereas Osterix is involved in the differentiation of pre-osteoblasts into mature osteoblasts and osteocytes [25].
We therefore hypothesised that we would be able to demonstrate epigenetic alterations in key elements of the vitamin D metabolism pathway -RXRA and VDR -and in genes regulating osteoblastic differentiation -Runx2 and Osterix [12,26] as a consequence of early life vitamin D depletion in our preclinical model and in response to loading, and that those changes would be associated with altered bone biomechanical properties.
Within this manuscript, we utilise data from our previously published work examining the effect of early life vitamin D deficiency on the bones of these mice. The data for bone strength (stiffness and ultimate force), as measured by 3-point bending on the harvested postmortem bones, is newly evaluated within linear regression analysis of the epigenetic data newly presented here. This was possible as the same bones analysed for 3-point bending were used for the DNA extraction and subsequent pyrosequencing [24]. Here, we present our analyses of the effect of diet and loading on the methylation status of the promoters in genes crucial in the vitamin D pathway and osteoblast differentiation in mouse tibias subjected to mechanical loading and later post-mortem biomechanical testing.

Vitamin D status
At weaning, all mice, irrespective of early life dietary exposure, were weaned onto the control vitamin D-containing diet. All animals were vitamin D replete when mechanically loaded, and we have previously reported this data [24].
The effect of early life vitamin D deficiency on the DNA methylation of Runx2, Osterix, VDR, and RXRA, genes known to play essential roles in bone formation, was assessed. Pyrosequencing primers were designed across the proximal promoter regions of these genes (Fig. 4).

Mechanical loading alters DNA methylation status in mice tibiae
DNA methylation was assessed in response to loading irrespective of prenatal and early life vitamin D depletion/repletion. Mice had lower levels of DNA methylation at the RXRA CpG1 site in the non-loaded as opposed to loaded limbs (p = 0.0373; difference −2.47% 95% CI −4.79, −0.16; non-loaded n = 11/group; loaded = 12/ group) (Fig. 1a). There were no significant differences in methylation at individual CpGs upstream of the VDR, Osterix, or Runx2 promoters between the groups.
The mice were analysed with respect to early life vitamin D deficiency or sufficiency. We found that mechanical loading was associated with lower levels of DNA methylation at the RXRA B CpG 6 site in the non-loaded limb compared to the loaded limb in those exposed to early life deficiency (p = 0.026, mean difference = −4.7%, 95% CI: −8.80, −0.69; n = 6/group) (Fig. 1b). There were no significant differences in methylation status in the other genes examined in response to mechanical loading.

Early life vitamin D deficiency is associated with altered offspring DNA methylation in mouse tibiae
DNA methylation changes in response to prenatal and early life vitamin D depletion irrespective of loading were first investigated. Methylation of Osterix CpG 11 was reduced in the tibiae of all offspring mice exposed to an early life diet deficient in vitamin D compared to the tibiae of offspring mice exposed only to the control diet (p = 0.0184, mean difference = 1.16%, 95% CI 0.22, 2.10; n = 12/group) (Fig. 2a). There were no significant differences in methylation at individual CpGs upstream of the RXRA, VDR, or Runx2 promoters between the diet groups. We next assessed epigenetic changes in response to prenatal and early life vitamin D depletion in the absence of loading. Methylation of Osterix CpG 11 was reduced in the non-loaded tibiae of offspring mice exposed to a diet deficient in vitamin D compared to the non-loaded tibiae of offspring mice exposed to the control diet (p = 0.031, mean difference = 1.50, 95% CI 0.12, 2.96; n = 6/ group) (Fig. 2b). There were no significant differences in methylation at individual CpGs upstream of the RXRA, VDR, or Runx2 promoters between the diet groups.
We then assessed epigenetic changes in response to prenatal and early life vitamin D depletion in the presence of loading. Methylation of Runx2 CpG 24 was reduced in the loaded tibiae of offspring mice exposed to a diet deficient in vitamin D compared to the loaded tibiae of offspring mice exposed to the control diet (p = 0.036, mean difference = 3.4%, 95% CI 0.27, 6.61; n = 6/ group) (Fig. 2c). There were no significant differences in methylation at individual CpGs upstream of the RXRA, VDR, or Osterix promoters between the treatment groups.

Extrinsic tibia strength is associated with changes in DNA methylation status
There was a positive association between tibial stiffness and DNA methylation of VDR CpG 4 and CpG 5 in all the bones examined irrespective of diet and loading (b = 0.315, r 2 = 0.776, p = 0.0008, r 2 = 0.746, b = 0.261, p = 0.031, respectively; n = 24).
The percentage change in DNA methylation status in RXRA A CpG 4 and Osterix CpG 11 in response to loading was negatively associated with the percentage change in stiffness following mechanical loading irrespective of diet (b = −2.03, p = 0.02; b = 0.04, p = 0.0088) (Fig. 2).
The percentage change in DNA methylation status of Osterix CpG 17 was negatively associated with the percentage change in ultimate force following mechanical loading irrespective of diet (b = −0.5, p = 0.019).

Discussion
We have previously shown in this group of mice that early life vitamin D deficiency led to lower bone mass and reduced bone mass accrual in response to mechanical loading [24]. We are now able to show that as well as the phenotype being altered; the epigenome is altered in the skeleton of the adult offspring exposed to early life vitamin D deficiency and, separately, in response to mechanical stimulation. This suggests that the early life environment can have long-term consequences for the skeleton that persist into adulthood.
We found alterations in DNA methylation in the promoters of four genes whose functions are important for skeletal development and bone accrual. Early life dietary deficiency of vitamin D was associated with altered methylation in Osterix (non-loaded bone) and Runx2 (loaded bone). Thus, we have shown, for the first time, that mechanical loading of bone can alter DNA methylation; this was seen in RXRA A CpG 1 irrespective of early life vitamin D depletion and in RXRA B CpG 6 in mice who had had early life depletion.
Methylation status in VDR CpG 4 and 5 was associated with tibial stiffness irrespective of diet or loading; this association was maintained for non-loaded limbs irrespective of dietary exposure, where VDR CpG 1 methylation status was also associated with tibial stiffness. Change in percentage methylation status in response to loading was associated with change in stiffness for two sites: RXRA A CpG 4 and Osterix CpG 11. The only association with change in ultimate force in response to loading was the change in methylation status of Osterix CpG 17.
DNA methylation was measured upstream of the transcriptional start site of four genes: RXRA, VDR, Runx2, and Osterix. RXRA and VDR play an important role within the vitamin D pathway; their heterodimerisation is key to mediating their downstream transcriptional regulatory activity through targeting of specific vitamin D response elements in the promoter regions of many genes. The vitamin D receptor is widely distributed within different tissues, and effects on a broad range of biological processes have been linked to current and prior vitamin D status [27]. Runx2 and Osterix are transcription factors which play critical roles in osteoblast differentiation. Runx2 is a primary transcription factor and one of the earliest markers of osteoblastogenesis; it lies upstream of Osterix in the osteoblast differentiation pathway [28].
A prenatal vitamin D-deficient diet, either irrespective of loading or in the absence of loading, was associated with lower Osterix CpG 11 methylation compared to the control diet. The Osterix CpGs of interest are located 369 bp upstream of the Osterix transcription start site (TSS) within the promoter region. The functional importance of these CpGs has not been investigated previously; however, methylation in promoter regions is usually associated with gene repression [29]. Reduced Osterix promoter methylation in mice born to vitamin D-deficient mothers might thus be associated with increased Osterix expression which would be expected to promote osteoblastic differentiation. Early life vitamin D deficiency did not alter trabecular parameters in non-loaded bones of these mice at age 18 weeks, but there was an increase in the ratio of mineralisation surface to tibial bone surface at the medial and lateral areas of the tibia and increased cortical porosity compared to the control mice suggesting increased bone turnover at those sites [24]. This could be part of a compensatory mechanism in response to low maternal vitamin D during pregnancy to maintain skeletal homeostasis.
In the loaded limbs, mice exposed to a prenatal vitamin D-deficient diet had lower Runx2 methylation at the CpG 24 site compared to control mice. The Runx2 CpGs are located over 2-kbp upstream of the Runx2 TSS, and Ensembl maps this region as part of the Runx2 promoter. Wakitani et al. found that there was an inverse association between DNA methylation of Runx2 CpG-2101 and Runx2 gene expression [30]. This CpG site is 40 bp away from our CpGs of interest, so it is likely that similar associations would be shown. This would suggest that lower methylation is likely to be associated with increased Runx2 gene expression.
Studies in mice have shown that Runx2 overexpression can promote osteoclast differentiation through RANKL (expressed by both osteoblasts and osteocytes), and that loss of Runx2 expression leads to reduced/ absent RANKL likely due to maturational arrest of osteoblasts, with normal levels of OPG. Therefore, a decrease in Runx2 methylation could be linked to an increase in Runx2 expression, with increased osteoclast differentiation through osteoblast-expressed RANKL. Conversely, Runx2 overexpression also inhibits osteoblastic differentiation [31,32]. Reduced osteoblast differentiation would likely reduce bone formation; coupled with increased osteoclastic resorption, this might contribute to our previous findings of lower bone mass and reduced bone accrual in response to mechanical loading.
Irrespective of prenatal diet, loading induced epigenetic changes at the RXRA A CpG1 site, with reduced methylation in the non-loaded as opposed to loaded limbs. For mice exposed to early life vitamin D deficiency, mechanical loading was linked to higher RXRA B CpG 6 methylation compared to the non-loaded limb. These RXRA CpGs are located 390-bp upstream of the RXRA TSS, within the promoter region. In the mice exposed to prenatal vitamin D deficiency, mechanical loading significantly improved trabecular and cortical parameters (but to a lower extent than in those who were vitamin D replete), but there was a decrease in the ratio of mineralisation surface to bone surface at the lateral part of the tibia [24]. RXRA has been shown to play a role in osteoclast differentiation, and a study of the loss of RXRA function in haematopoietic cells resulted in giant nonresorbing osteoclasts and increased bone mass [33]. In the MAVIDOS trial, we have shown that maternal cholecalciferol supplementation during pregnancy is associated with lower RXRA CpG methylation in umbilical cord tissue compared to cord tissue from placebo-supplemented mothers [21]. Moreover, studies within the Southampton Women's Study cohort have shown that there is an inverse association between RXRA CpG methylation and bone measures in childhood [20]. Our findings suggest that RXRA CpG methylation in relation to vitamin D depletion may exert its effect on bone parameters through reduced bone formation.
By comparing DNA methylation in loaded and nonloaded bones, we have shown for the first time that mechanical tibial loading alters DNA methylation (of RXRA) from the same bone tissue. The mechanostat describes the mechanism by which structural changes in bone architecture occur in response to mechanical loading. We demonstrate here that DNA methylation plays a role in this mechanism by either initiating a change in expression or consolidating a change in expression mediated by a change in histone modification. Moreover, as DNA methylation is a relatively stable mark, this also suggests a mechanism by which bone loading in early life may induce persistent effects at later time points through the life course.
We have previously demonstrated in these mice that the values for stiffness and ultimate force (measure of extrinsic bone strength) of their tibiae were similar. That mechanical loading did not, however, increase bone stiffness in the early life-deficient mice unlike the control group. The force required to break the bone (ultimate force) in both groups of mice increased in response to mechanical loads but by a lesser amount in the early life-deficient mice [24]. DNA methylation of VDR in non-loaded limbs is positively associated with tibial stiffness and negatively associated in loaded limbs. The VDR CpGs are located 26 bp from the VDR TSS, within the promoter, and therefore are likely to lead to a decrease in VDR expression. The ability of epigenome to change in response to the external stimulus of mechanical loading in adult life is a novel finding.
This study focused on CpGs within the proximal promoters; it would be interesting to look across the whole promoter region and upstream enhancer regions. A further potential study limitation is in CpGs of interest, we are seeing relatively low percentage changes in methylation (< 5%), and it is unknown what such differences mean. We presume this to reflect the number of cells expressing the gene in a tissue and so may affect function. Genomic DNA isolated from whole blood of individuals who were in utero during the Dutch Hunger Winter compared to unexposed same-sex siblings has shown hypomethylation of the differentially methylated regions of the imprinted IGF2 gene with a change of 5.2% [18]. The adult offspring of mothers in the Motherwell cohort -pregnant women encouraged to eat 0.45 kg red meat daily and low carbohydrate foods -was recruited to examine their epigenome. Low levels of methylation change were identified at similar levels to our study [34]. A further limitation is that we have no cell type correction data available. Tibial bone consists of many cell types, and methylation patterns are known to be tissue and cell type specific. A fourth limitation is that the mice were loaded twice, during their pubertal growth between weeks 8 and 10 and during adult growth in weeks 16-18. Ideally, DNA methylation would have been measured in mice that were only loaded in adulthood. Furthermore, no other studies have looked at the effects of two periods of loading.

Conclusion
We have shown for the first time changes in the epigenome in adult mice that following mechanical loading of bone with differential methylation are present in the epigenome of key genes in the vitamin D and osteoblast differentiation pathways.
We have also demonstrated a series of changes in methylation in response to early life vitamin D depletion with and without the effects of mechanical loading. These give support to the hypothesis that early life events can have long-lasting consequences for skeletal health and further support our view that vitamin D supplementation during pregnancy and early life is a crucial factor that can be modified for the benefit of future generations.

Animals
The murine model has been described in detail previously (Fig. 3) [24]. Four-week-old C57BL/6 (Charles River, Kent, UK) female mice were purchased and housed at the University of Sheffield Biological Services Unit and had access to food and water ad libitum. All mice were housed in a single room with a 12-h light/dark cycle at 22 °C under UV-filtered fluorescent lighting.
Female mice were fed a vitamin D-supplemented diet (1000 units/kg) or vitamin D-free diet based on AIN93G (Research Diets Inc., USA) diet from 4 weeks of age. Mice were randomly allocated to either vitamin D replete (1000 units/kg) or deplete diet. At 10 weeks of age, female mice were mated with healthy control adult males; dams remained on their respective diets throughout gestation until pup weaning. Female offspring remained with their dams until weaning at day 22 whereupon they were weaned onto the vitamin D-replete diet (n = 24) (male pups not included in this protocol). All animals were supplemented with 2 mmol/l Ca 2+ in their drinking water.
This study was carried out as per regulations under the Animals (Scientific Procedures) Act 1986. The study was approved by the local committee of the University of Sheffield that oversees animal use within the institution. Loading experiments were performed under isoflurane anaesthesia. All efforts were made to minimise animal suffering.

Mechanical loading
A noninvasive method of tibial loading was used at ages 8 and 16 weeks as previously reported [24]. The peak load of 11 N was selected as this is known to induce an osteogenic response in female C57BL/6 mice [35][36][37]. A 10.5-N dynamic load was superimposed onto a 0.5 N preload at a rate of 160,000 N/s. Forty trapezoidal-waveform load cycles (0.2 s hold at 11 N) with a 10-s interval between each cycle were applied to each mouse's left tibia 3 times per week for 2 weeks.
Mice were euthanised on day 15 (where day 1 was the first of the loading cycle following the second episode of loading at age 16 weeks. Both tibias were dissected and frozen wrapped in PBS-soaked gauze at −20 °C. The contralateral non-loaded limb (right tibia) was treated as an internal control for loading (the functional adaptation of the bone is confined to the loaded limb) [38] Bone mechanical testing Bones previously stored at −20 °C in PBS-soaked gauze were placed in PBS at 4 °C overnight prior to testing at room temperature. Samples were subjected to threepoint bending analysis using a Bose ElectroForce 3200 mechanical testing machine with a 450 N load cell as previously described [24]. Briefly, rounded supports and press head were used with a constant span support distance of 6 mm for all tibias; bones were positioned with the press head contacting the midshaft. A preload of 0.5 N was applied to the bone before compressing at a constant force of 0.25 N/s until failure. From the force-displacement curve, stiffness (n/mm) was calculated from the linear elastic region and breaking force from the load at failure.

Genomic DNA extraction
Following biomechanical testing, bones were again stored frozen at −20 °C wrapped in PBS-soaked gauze. DNA was extracted from tibiae using the DNeasy Blood and Tissue Kit as per the manufacturer's instructions. DNA was stored at 4 °C for short-term use and at −20 °C for longer-term storage.
Pyrosequencing primers were designed to measure the DNA methylation status of CpGs sites within the promoter regions of RXRA, VDR, and Osterix, where sequence context permitted the efficient binding of primers. For Runx2, DNA methylation was measured across CpG sites −2484 to −2148 bp upstream of the Runx2 promoter, as this was a region previously identified to be differentially methylated [30].

Statistical analyses
Independent t-tests compared %DNA methylation between the non-loaded and loaded limbs of mice exposed to a control or prenatal vitamin D-deficient diet. To determine whether there were any associations between DNA methylation and tibial stiffness or ultimate force, linear regression analyses were adjusted for the treatment group and for mechanical loading.