Association of embryonic inositol status with susceptibility to neural tube defects, metabolite profile, and maternal inositol intake

Maternal nutrition contributes to gene–environment interactions that influence susceptibility to common congenital anomalies such as neural tube defects (NTDs). Supplemental myo‐inositol (MI) can prevent NTDs in some mouse models and shows potential for prevention of human NTDs. We investigated effects of maternal MI intake on embryonic MI status and metabolism in curly tail mice, which are genetically predisposed to NTDs that are inositol‐responsive but folic acid resistant. Dietary MI deficiency caused diminished MI in maternal plasma and embryos, showing that de novo synthesis is insufficient to maintain MI levels in either adult or embryonic mice. Under normal maternal dietary conditions, curly tail embryos that developed cranial NTDs had significantly lower MI content than unaffected embryos, revealing an association between diminished MI status and failure of cranial neurulation. Expression of inositol‐3‐phosphate synthase 1, required for inositol biosynthesis, was less abundant in the cranial neural tube than at other axial levels. Supplemental MI or d‐chiro‐inositol (DCI) have previously been found to prevent NTDs in curly tail embryos. Here, we investigated the metabolic effects of MI and DCI treatments by mass spectrometry‐based metabolome analysis. Among inositol‐responsive metabolites, we noted a disproportionate effect on nucleotides, especially purines. We also found altered proportions of 5‐methyltetrahydrolate and tetrahydrofolate in MI‐treated embryos suggesting altered folate metabolism. Treatment with nucleotides or the one‐carbon donor formate has also been found to prevent NTDs in curly tail embryos. Together, these findings suggest that the protective effect of inositol may be mediated through the enhanced supply of nucleotides during neural tube closure.


| INTRODUCTION
Neural tube defects (NTDs), such as anencephaly and spina bifida, result from incomplete closure of the neural tube during embryonic development.They are among the commonest birth defects, affecting 0.5-2 per 1000 pregnancies worldwide with higher rates in some countries.Identification of the biochemical and cellular causes of NTDs has proven challenging owing to their complex multigenic etiology and the contribution of environmental factors, including nutrition, teratogen exposure, and maternal diabetes. 1For example, potential risk variants have been identified in a number of genes, while suboptimal maternal intake of folate or vitamin B 12 are independent risk factors for NTDs. 2,3Conversely, maternal supplementation with folic acid can prevent a proportion of, but not all, NTDs. 4,5e focused on another nutrient that may modulate NTD risk, inositol (1,2,3,4,5,6-hexahydroxycyclohexane), a simple six-carbon sugar alcohol sometimes referred to as vitamin B 8 or B h . 6,7Among nine possible stereoisomers, myo-inositol (MI) is the predominant naturally occurring form.MI plays a wide range of cellular functions, including as a precursor in the synthesis of phosphoinositides and inositol phosphates that mediate various signaling pathways. 6,8 requirement for MI in neural tube closure was revealed by mouse whole embryo culture studies in which culture media was reconstituted after dialysis, to allow individual omission of specific small molecules.While the removal of several vitamins led to impaired growth or developmental progression, only the omission of MI resulted in the development of cranial NTDs. 9In parallel with potential risk imposed by insufficient MI, a beneficial effect of supplemental MI in the prevention of NTDs has been demonstrated in a range of contexts in mice, including NTDs induced by hyperglycemia and maternal diabetes. 10,11he interaction of MI status and genetic susceptibility to NTDs has been investigated in mice, most extensively in the curly tail (ct) mouse.This strain exhibits a low frequency of cranial NTDs and partially penetrant spinal NTDs, that are not preventable by folic acid.3][14] DCI had a notably greater protective effect than MI, being effective at lower doses and with a greater magnitude of effect at equivalent doses. 13,14hether MI and DCI act to prevent NTDs via the same or distinct mechanisms is not known.
6][17] Although supplemental folic acid does not prevent NTDs, the ct genetic background (with or without the Grhl3 ct allele) confers susceptibility to maternal folate deficiency, which leads to an increased rate of cranial NTDs (unlike in non-mutant strains). 18Notably, these cranial NTDs induced by maternal folate deficiency are also preventable by MI, but whether MI corrects a defect related to folate insufficiency or acts through an independent mechanism is not known.
Evidence for a protective effect of MI in experimental models, including folic acid-resistant NTDs, led to the evaluation of maternal MI supplementation as a potential preventive therapy for human NTDs.These studies aimed to test the concept that MI, used in combination with folic acid, could prevent more NTDs than possible with folic acid alone.Non-randomized studies and a pilot randomized, double-blind clinical trial in high-risk pregnancies reported no recurrent NTDs, suggesting a potential protective effect and need for further investigation. 19,20lthough several lines of evidence indicate that maternal MI status influences risk of NTDs in the offspring, there is still a gap in our knowledge regarding the relationship between maternal MI intake and synthesis, with embryonic MI status.It appears likely that maternal circulating MI plays a key role in determining availability to the embryo at neurulation stages.Maternal MI is obtained from diet and catabolized in the kidneys by MI oxygenase, generating glucuronate for excretion. 21Maternal MI status may also be influenced by de novo synthesis from glucose mediated by sequential action of hexokinase (HK), inositol-3-phosphate synthase 1 (ISYNA1), and inositol monophosphatase (IMPA1 or 2). 8he potential for a genetic component in the regulation of circulating MI was investigated in a genome-wide association study in healthy young adults.This study revealed a nominally significant association of plasma MI with SLC5A11, encoding a sodium-coupled inositol transporter (SMIT2), and suggestive association with three other genes. 22However, lack of association of genome-wide significance suggested that plasma MI is strongly influenced by non-genetic factors such as diet.

K E Y W O R D S
inositol, mass spectrometry, metabolome, mice, neural tube defects, nucleotides The presence of higher MI concentration in coelomic and amniotic fluid than in maternal serum during human pregnancy suggests that there is fetal MI production and/ or active transport into the embryonic compartment. 21,23ransfer of MI from maternal circulation occurs via the yolk sac at the initial stages of neurulation in mouse and humans.This may be mediated by sodium-coupled transporters encoded by SLC5A3, SLC5A11 (SMIT1 and SMIT2), or a proton-coupled transporter (HMIT) encoded by SLC2A13, each of which are expressed in human yolk sac. 21he extent to which MI synthesis occurs in the embryo is not known, nor whether this can compensate for varying maternal MI availability.In the current study, we investigated the effect of maternal dietary MI status and supplementation on embryonic inositol content, metabolite status, and frequency of NTDs in mice, using the MIsensitive curly tail strain.

| Mouse strains and maintenance
Mice were inbred C57Bl/6J and random-bred curly tail (ct) strains.The ct mice have been maintained as a closed colony, homozygous for the Grhl3 hypomorphic allele, for more than 25 years. 24Mice were kept at 22°C on a controlled diurnal cycle of 12 h light and 12 h dark.Food and water were freely available.Mice were transferred to control (0.19% w/w myo-inositol) or inositol-deficient (no added myo-inositol) diet (TestDiet 5W8E and 5W8E minus inositol) at 4-6 weeks of age (Figure S1).Females were maintained on either diet for a minimum of 2 weeks prior to overnight mating to generate experimental litters, and during pregnancy.Oral MI supplementation was achieved by addition of MI (5 mg/mL) in drinking water from E0.5.Intraperitoneal injections of 400 mg/kg of MI or DCI (Sigma) in phosphate-buffered saline (PBS) were given at 24 h (E9.5) and 4 h (E10.5)prior to embryo collection.All litters were included in the analysis.

| Whole embryo culture
Embryos were explanted from the uterus at E9.5, leaving the yolk sac and ectoplacental cone intact, and cultured for 24 h in rat serum that had been immediately centrifuged and heat-inactivated upon collection. 25MI (50 μg/mL), L690330 (1 μM) (Tocris Bioscience), or PBS (vehicle control) were added to cultures after 30 min as 1% (v/v) additions.Embryos were allocated to the treatment group by sequential addition in order of the size of the conceptus to avoid inadvertent stage differences at the start of culture.Cultures were gassed with 20% O 2 , 5% CO 2 , 75% N 2 at E9.5 (at the start of culture, and after addition of reagents) and 40% O 2 , 5% CO 2 , 55% N 2 at E10.5.At the end of culture, yolk sac circulation was assessed to confirm viability of cultures, number of somites was recorded and crown-rump length and posterior neuropore length were measured using an eyepiece graticule. 26Analysis was performed blind to the treatment group.

| Whole mount in situ hybridization
Embryos were fixed overnight in 4% paraformaldehyde in diethylpyrocarbonate (DEPC)-treated PBS, dehydrated through a methanol series, and stored in 100% methanol at −20°C.To generate probes, cDNA fragments of Impa 1 (545 bp) and Isyna1 (565 bp) were cloned into pGEM-T using specific primer pairs (5′GTTCTCCAGCCGACTTGGTA and 5′GCAGTGGATTCCCATCTCAT for Impa1; 5′-CA A CGACCTGGTGTTTGATG and 5′GATGA GG AA G T C CACCAGGA for Isyna1).4 | Quantification of inositol by gas chromatography-mass spectrometry Embryos were rinsed in PBS and frozen immediately on dry ice and stored at −80°C. E0.5 embryos used for analysis were at the 28-34 somite stage and biochemical analysis was blind to the treatment group.Embryos were thawed on ice and, following addition of 300 μL PBS, embryos were homogenized by sonication using a hand-held sonicator, and an aliquot was taken for protein assay.Samples were centrifuged for 1 min at 12 000 g, and 10 μL solution was removed for protein quantification by QuBit protein assay (1 μL per assay).A 600 μL aliquot of acetonitrile was added to each tube, incubated for 2 min at room temperature, and centrifuged for 15 min at 13 000 rpm.The supernatants were transferred to fresh tubes, dried via lyophilization, and stored at −20°C.Derivatization of samples and GC-MS were performed as described previously.22 GC was performed using a Rxi-5Sil MS fused silica (5% diphenyl/95% dimethylpolysiloxane), 30 m × 0.25 mm I.D, 0.25 μm film thickness column (RESTEK), which enables separation of inositol enantiomers (myo, d-chiro, allo, muco, and scyllo) on the basis of retention time.Selected ion monitoring (SIM) was used for analysis of inositol enantiomers (M/Z 372.5-373.5)and the myoinositol-d 6 (378.5-379.5)internal standard.Peak areas corresponding to MI, DCI, and MI-d 6 , were determined by peak integration.

| Analysis of folates by ultra-performance liquid chromatography-tandem mass spectrometry
Samples were homogenized by sonication in buffer containing 20 mM ammonia acetate, 0.1% ascorbic acid, 0.1% citric acid, and 100 mM dithiothreitol at pH 7. Protein was removed by precipitation by addition of two volumes of acetonitrile and centrifugation (12 000 g at 4°C).Supernatants were transferred, lyophilized, stored at 80°C, and re-suspended in dH 2 O before analysis.Folate analysis was carried out by ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) as described previously. 28,29Metabolites were resolved by reversed-phase UPLC (Acquity UPLC BEH C18 column, Waters Corporation, UK).Solvents for UPLC were as follows: Buffer A, 5% methanol, 95% Milli-Q water, and 5 mM dimethylhexylamine at pH 8.0; Buffer B, 100% methanol, and 5 mM dimethylhexylamine.The column was equilibrated with 95% Buffer A: 5% Buffer B. The sample injection volume was 25 mL.The UPLC protocol consisted of 95% Buffer A: 5% Buffer B for 1 min, followed by a gradient of 5%-60% Buffer B over 9 min and then 100% Buffer B for 6 min before re-equilibration for 4 min.Metabolites were eluted at flow rate of 500 nL/ min, with a wash step with 100% Buffer B at flow rate of 600 nL/min.UPLC was coupled to a XEVO-TQS mass spectrometer (Waters Corporation) operating in negative-ion mode using the following settings: capillary 2.5 kV, source temperature 150°C, desolvation temperature 600°C, cone gas flow rate 150 L/h, and desolvation gas flow rate 1200 L/h.Folates were measured by multiple reaction monitoring with optimized cone voltage and collision energy for precursor and product ions. 28,30

| Metabolite analysis by ultra-performance liquid chromatographytandem mass spectrometry
Mass spectrometry-based metabolite analysis was performed by Metabolon (Morrisville, NC, USA).Briefly, samples were extracted at a constant mass-to-volume ratio and prepared using a MicroLab STAR system (Hamilton Company) with proteins removed by methanol precipitation.Sample extracts were dried and reconstituted in solvent for analysis by ultra-performance liquid chromatography coupled to tandem mass spectrometry (UPLC-MS/MS) using ACQUITY UPLC (Waters) and Q-Exactive high resolution/accurate mass spectrometer interfaced with a heated electrospray ionization (HESI-II) source and Orbitrap mass analyzer (Thermo Scientific).Methods comprised (i) reverse phase LC using a C18 column with conditions optimized for negative-ion mode electrospray ionization (ESI); (ii) LC using a HILIC column (Waters UPLC BEH Amide 2.1 × 150 mm, 1.7 μm) with negative-ion mode ESI; (iii) and (iv) using positive ion mode ESI following elution from a C18 column (Waters UPLC BEH C18-2.1 × 100 mm, 1.7 μm) optimized for (iii) hydrophilic compounds using gradient elution from using water and methanol, containing 0.05% perfluoropentanoic acid (PFPA) and 0.1% formic acid (FA), or (iv) more hydrophobic compounds, using gradient elution with methanol, acetonitrile, water, 0.05% PFPA and 0.01% FA.
Raw data were extracted, peak-identified, and QC processed using Metabolon's hardware and software.Compounds were identified by comparison to library entries of authenticated standards based on retention time/ index (RI), mass-to-charge ratio (m/z), and chromatographic data (including MS/MS spectral data).Peaks were quantified using area-under-the-curve.The overall process variability for the embryo dataset was 7% (based on median relative standard deviation for all endogenous metabolites).

| Statistical analysis
Pairwise comparisons were performed by student t-test (for normal distribution of data).Multiple group comparisons were performed by one-way ANOVA with Holm-Sidak method for post hoc comparisons (using SigmaStat, Systat Software Inc.).

| RESULTS
To examine the effect of maternal inositol intake on embryonic inositol status, we established models of maternal myo-inositol (MI) deficiency and supplementation.Female mice were maintained on defined diets containing MI at a concentration found in typical mouse chow (standard diet) or without MI (inositol-deficient).An additional group was maintained on the standard diet and supplemented with MI (5 mg/mL in drinking water) during pregnancy.

| Dietary MI content influences maternal plasma MI
We first tested whether lack of dietary MI affects circulating MI in pregnant mice at mid-gestation (E10.5-12.5)using gas chromatography-mass spectrometry (GC-MS).Compared with the standard diet, maintenance on MIdeficient diet caused a significant reduction in maternal plasma MI concentration in females of C57Bl/6J (nonmutant) and curly tail (ct) strains (Figure 1A).Plasma MI did not differ between strains when comparing the same diet, suggesting that ct does not exhibit a genetically determined deficit in circulating MI or altered response to MI deficiency.Supplemental MI in drinking water led to a significant increase in plasma MI compared with standard diet in both strains (Figure 1A).

| Effect of maternal dietary MI on embryonic development and MI content
We examined the effect of maternal diet and plasma MI status on embryonic development and MI content at E10.5.MI status did not affect overall developmental progression, as determined by number of somites at E10.5, in either C57Bl/6 or ct embryos (Table S1).Similarly, average embryo growth, as indicated by crown-rump length, was not affected by maternal MI status in C57Bl/6 embryos, but there was a small increase in crown-rump of ct embryos among litters from MI-supplemented dams (Table S1).
Next, we asked whether the diet-induced alteration in circulating maternal MI led to altered embryonic MI content at E10.5, using quantification by GC-MS (normalized to protein content).In both C57Bl/6 and ct strains, embryos that developed under conditions of maternal MI deficiency had significantly diminished mean tissue MI content compared with embryos in other groups (Figure 1B).The MI content of embryos developing under MIsupplemented conditions did not differ from the standard diet, despite the significant increase in maternal plasma MI on the supplemented diet.This suggests that there is either a limit on embryonic uptake or that surplus MI is metabolized to downstream metabolites in the embryo.No strain differences between C57Bl/6 and ct were observed among embryos developing under the same dietary conditions (Figure 1B).

| MI content in relation to neural tube closure phenotype in ct embryos
Average MI content in the ct strain showed a greater inter-embryo range of values in the standard diet and MI-supplemented groups than in the MI-deficient group (Figure 1B).Given that the ct strain displays partially penetrant cranial and spinal NTDs, we therefore carried out sub-group analysis to investigate whether MI content differed with neural tube phenotype.Cranial neural tube closure is normally complete by E9.5, so cranial NTDs were scored at E10.5.Open cranial neural folds (the forerunner of exencephaly and subsequent anencephaly) were present in 5%-7% of embryos collected.Spina bifida (spinal NTDs) results from failure closure of the posterior neuropore (PNP) and, as a marker for ensuing spina bifida, we therefore selected embryos with a severely enlarged PNP (0.6 mm or greater length) at E10.5.Definitive spina bifida was detectable at E12.5.
All ct embryos from the MI-deficient group had similarly low MI content irrespective of NTD phenotype (Figure 2), suggesting that diminished MI content, at least to the extent induced by maternal diet, is insufficient to cause NTDs in every ct embryo.However, within the standard diet group there was marked variation in MI content among the NTD sub-groups.Notably, ct embryos with cranial NTDs exhibited a lower MI content than unaffected ct embryos, with intermediate values for embryos with enlarged PNP (Figure 2).A similar trend was observed among ct embryos of MI-supplemented dams.Together, these findings suggest that diminished MI content is associated with an increased risk of cranial NTDs.
Among ct embryos that developed under standard or MI-supplemented maternal conditions (n = 38), six of eight embryos with cranial NTDs corresponded to those embryos with the lowest MI content and all lie in the lower half of values.As a combined group, ct embryos with cranial or spinal NTDs corresponded to all the embryos in the lowest quartile of MI content and represented 17 of 19 in the lowest half of values (8/8 with cranial NTDs and 9/16 with spinal NTDs 9/16 are in the lowest half of values).Hence, ct embryos with NTDs were 14-fold more likely to be represented in the lower half of MI content values than in the upper half (OR 14.6, 95% confidence 2.6-82, p < .005).
Our analysis of embryos at E10.5 and E12.5 showed that de novo synthesis of MI in the embryo is not sufficient to compensate for maternal deficiency.Yet, except for NTDs, overall gross development appeared normal at E10.5.It is not known whether the sensitivity of cranial neurulation to low MI content is associated with an increased MI requirement in the cranial region, or limited synthesis capacity in this region compared with other parts of the embryo.Transcripts of the Isyna1 and Impa1 genes, which encode the enzymes that mediate inositol synthesis (Figure 3A), are detected at neurulation stages. 31To examine potential tissue specificity of expression, we used whole mount in situ hybridization for Isyna1 and Impa 1 (Figure 3).Isyna1 expression was detected in the closing neural folds and neuroepithelium of the spinal cord at E9.5 (Figure 3C-E) but was less abundant or absent in the cranial neural tube (Figure 3B), suggesting that endogenous MI synthesis capability may be lower in the cranial than spinal region.IMPA1 is required both in MI synthesis and for recycling of inositol phosphates (Figure 3A), and we observed widespread expression of Impa1 at E9.5 (Figure 3F), as reported at E10.5, 31 including in the cranial and spinal neuroepithelium (Figure 3F-I).

| The protective effect of MI on spinal neural tube closure depends in IMPase activity
IMPA1 is expressed at neurulation stages (Figure 3), and we previously found that the ability of MI to normalize PNP closure in ct embryos abrogated by co-administration of lithium chloride, which suppresses inositol phosphate metabolism by inhibition of IMPase activity. 12However, lithium can also inhibit other enzymes such as GSK3β, thereby acting as an agonist of the canonical Wnt pathway.We therefore tested the effect of a specific IMPase inhibitor, L690330, on PNP closure in whole embryo culture.After culture for 24 h from E9.5-10.5, the PNP length of ct embryos treated with MI was significantly smaller than among vehicle controls (Figure 4A), as previously reported. 12In contrast, the mean PNP length of embryos treated with both MI and L690330 did not differ from controls (Figure 4A), confirming that the protective effect of MI on PNP closure is dependent on IMPase activity.

| Metabolome analysis in embryos exposed to maternal inositol supplementation
The effect of supplemental MI on embryonic metabolism was further investigated, using a dosing strategy that was previously found to prevent spinal NTDs in ct embryos. 12,14e confirmed that MI treatment by intraperitoneal injection (400 mg/kg/day at E9 and E10) led to a significant reduction in PNP length in embryos with 30-31 somites at E10.5, showing a protective effect as previously observed in cultured ct embryos 12 (Figure 4B).
In parallel with MI treatment, an additional group of ct mice were treated with the same dose of d-chiro-inositol (DCI), which has also been found to prevent spinal NTDs in ct embryos. 12,14GC-MS analysis of blood, collected 4 h after treatment, confirmed that MI or DCI injection led to significantly increased concentration in maternal plasma of MI or DCI respectively (Table S2).Neither MI nor DCI treatment led to a significant change in the concentration of the other enantiomer.
Having confirmed that MI or DCI injection leads to elevated maternal plasma concentration of the corresponding inositol enantiomer, we analyzed the effect on the abundance of metabolites in embryos at E10.5 (n = 5-6 pools of two embryos per group).The developmental stage and size of analyzed embryos did not differ with the treatment group (Table S3), and embryos with cranial NTDs were excluded from analysis to reduce inter-embryonic variability in MI content in the control (standard diet) group.
Metabolites were analyzed using a platform comprising ultra-performance liquid chromatography coupled to tandem mass spectrometry (UPLC-MS/MS).We validated this approach by analysis of MI content in the groups of MI-supplemented and control (standard diet) ct embryos, as well as embryos from the MI-deficient dietary group.Compared with the control diet, administration of MI, but not DCI, led to a significant increase in MI content, while maternal MI deficiency led to lower embryonic MI content, as predicted by maternal treatment and plasma inositol concentrations (Figure 5A).DCI was detected in embryos following maternal DCI administration, but in the other treatment groups DCI was at or below the limit of detection (Figure 5B).For example, raw MS data showed 4-8 million counts in the DCI treatment group, compared with approximately 50 000 counts in 2 out of 17 embryos in other groups and no detection in the remainder.We glucose abundance and found significantly lower abundance in embryos supplemented with MI (Figure 5C).
In conclusion, neither MI nor DCI treatment led to a significant change in the abundance of the other enantiomer (i.e., MI did not affect DCI or vice versa) in either maternal plasma or embryonic tissue.While maternal glucose was not affected by MI or DCI treatment, embryonic glucose fell after MI treatment, possibly due to glucose sharing a transporter with MI and its use in MI synthesis. 21

| Effects of maternal MI status on embryonic metabolite profile
Having validated the utility of metabolome data in detecting changes in embryonic abundance of MI and DCI resulting from maternal intake, we analyzed approximately 550 biochemicals detected in embryonic tissue using the LC-MS/MS platform.Overall, compared with embryos that developed under standard conditions, we observed significantly altered abundance of 71 metabolites (including MI) in the MI-supplemented group (Figure 5D) and 153 metabolites, including DCI, in the DCI-supplemented groups (Figure 5E) compared with the standard diet group.In each case, the majority of inositol-responsive metabolites exhibited an increase in abundance (Figure 5D,E).
To focus on potential protective effects of MI and DCI in ct embryos, we filtered the list of altered metabolites to remove those which also differed in MI-deficient compared with control embryos (with the same direction of change).This analysis identified 36 metabolites associated with both maternal MI and DCI supplementation.An additional 12 metabolites were only affected by MI (including MI and glucose) and 77 (including DCI) were only significantly affected by DCI (Figure 5F, Tables S4-S6).While MI was significantly more abundant in MI but not DCI-treated embryos, inositol 1-phosphate showed a similar magnitude of increase in both treatment groups compared with controls, although this was only statistically significant in the MI-treatment group (Figure S3).
Among metabolites altered by MI and DCI treatment, we noted a disproportionate representation of nucleotides, with 28/53 of analyzed nucleotides and related metabolites exhibiting a change in embryonic abundance.Purines and pyrimidines were both altered in abundance following MI or DCI supplementation, but a greater number of purines were affected (12/17 nucleotides affected by MI and DCI; Figure 5).For example, the nucleotide precursors adenosine monophosphate, guanosine monophosphate, and thymidine monophosphate were significantly more abundant following MI or DCI treatment (Figure 6A-C).Adenoylsuccinate, the intermediate in AMP synthesis from IMP, exhibited among the greatest change in relative abundance in embryos exposed to MI or DCI treatment (Figure 6D).
Purine and thymidylate biosynthesis depend on folate one-carbon metabolism (FOCM), a network of reactions that transfers one-carbon units (primarily derived from serine and glycine) between folate molecules.To test whether MI supplementation affects FOCM, we analyzed the relative abundance of the six major folate species using targeted LC-MS/MS methodology. 28Compared with controls, ct embryos that were exposed to maternal MI supplementation exhibited a shift in folate profile, with a significant increase in the proportion of tetrahydrofolate (THF) and decrease in the proportion of 5-methyl THF (Figure 6E).
Although DCI is an inositol enantiomer that differs from MI only in the orientation of one hydroxyl group, it has been found to be more effective than MI in the prevention of NTDs in ct embryos. 14Interestingly, embryonic metabolite profile was altered to a greater extent by DCI than MI.For most metabolites that were significantly altered by both treatments, there was a non-significant trend toward a greater effect of DCI (23/36 metabolites).More strikingly, a greater number of metabolites, particularly amino acids and lipids were affected by DCI but not MI.Among sub-categories of lipids, DCI did not affect abundance of short or long chain fatty acids, whereas DCI-responsive metabolites included higher proportions of the analyzed metabolites in the sub-categories monohydroxy and dihydroxy fatty acids such as dihydroxybutyrate (6/7 analyzed metabolites), and the endocannabinoids, arachidonylataurine, oleoyltaurine, stearoyltaurine, and palmitoyltaurine (Table S6).DCI-responsive amino acids were enriched for metabolites related to histidine and lysine metabolism (Table S6).

| DISCUSSION
We found that maternal inositol status directly affected circulating MI or DCI.This in turn led to corresponding changes in embryonic inositol content.Hence, our findings suggest that de novo MI is insufficient to fully compensate for dietary inositol deficiency in either the maternal or embryonic compartment.In contrast, loss of function of IMPA1 can cause partially penetrant lethality which can be rescued by maternal MI supplementation 32 suggesting a requirement for embryonic MI synthesis (even under standard maternal dietary conditions) and/or recycling of inositol phosphates (Figure 3A).
In whole embryo culture, removal of MI from the culture medium leads to cranial NTDs in non-mutant strains of mice and in the ct strain. 9In contrast, we found that MI deficiency induced by maternal dietary modification did not cause NTDs in a non-mutant (C57Bl/6) strain.Hence, it appears likely that residual MI supply from maternal sources (with/without embryonic de novo synthesis) is sufficient for neural tube closure in the absence of additional NTD risk factors, genetic or non-genetic.This is consistent with findings in Slc5a3 null mice in which a 77% reduction in MI content was not reported to cause NTDs, although post-natal phenotypes and lethality were observed. 33,34n the NTD-susceptible ct strain, we found a striking correlation between low embryonic MI content and cranial NTDs, suggesting that cranial neural tube closure is particularly sensitive to MI status.We previously found that diminished supply of MI leads to reduced incorporation of MI into inositol phosphoinositides and inositol phosphates, 12 and we speculate that diminished activity of inositol-dependent signaling may contribute to failure of cranial neurulation in genetically susceptible embryos.Consistent with this idea, a requirement for inositol phosphate and phosphoinositide metabolism in neural tube closure is also suggested by the presence of cranial NTDs among mouse embryos with loss of function of enzymes such as ITPK1 and PIPKγ. 35,36e speculate that in humans, as in mice, maternal MI status may influence embryonic MI status and hence susceptibility to NTDs in combination with other genetic and/or environmental risk factors (as in the multigenic ct mouse model).Plasma inositol concentrations are broadly comparable in mice and humans.For example, among mice maintained on the standard diet in this study, the mean plasma MI of approximately 69 μM was approximately double the mean value of 28 μM that we observed in a population of young healthy adults. 22In considering MI deficiency as a potential risk factor for NTDs, serum MI concentration has been determined among mothers and their infants with spina bifida.Although, a relatively small-scale study, there was a non-significant difference in mean values between NTDs and controls but the lowest decile of maternal MI values was significantly associated with having a child with spina bifida. 37It is not yet known whether these data are representative of MI status during early pregnancy or whether there is an association with cranial NTDs.
MI deficiency is associated with an increased risk of cranial NTDs in ct mice, and supplementary MI has been found to prevent NTDs. 12,14We previously proposed that integrity of the inositol phosphate cycle is required for the normalization of spinal neural tube closure in the ct model, and in the current study, we confirmed these findings using a specific inhibitor of inositol monophosphatases.In this inositol-responsive model, we further investigated the effect of MI or DCI on embryonic metabolism, which has not previously been investigated.Supplemented embryos exhibit increased abundance of purine and pyrimidine nucleosides and nucleotides, including precursors for DNA and ATP synthesis.Although this effect had not been predicted, several possible mechanisms could potentially explain the increased production of nucleotides in inositol-treated embryos.MI is interconverted with glucose 6-phosphate (Figure 3A), which is both a precursor in (i) the pentose phosphate pathway that generates the nucleotide precursor phosphoribosyl pyrophosphate (PRPP) and (ii) the pathway for the generation of serine which is a one-carbon donor via FOCM to purine and thymidylate biosynthesis.MI or DCI could also potentially promote nucleotide biosynthesis through PI-dependent modulation of FOCM or nucleotide biosynthesis pathways.For example, the phosphatidylinositol 3-kinase mediator Akt (activated by PIP 3 ) may promote purine biosynthesis, while PI3-K inhibitors suppress purine synthesis. 38Our finding of abrogation of the MI protective effect by IMPase inhibition is compatible with a need for flux through PI and inositol phosphate signaling.
In parallel with an increased nucleotide abundance, MI treatment led to altered folate profile with an increase in the proportion of THF and decrease in 5-methyl THF.An effect of MI on folate one-carbon metabolism (FOCM) had not previously been identified and this shift in pattern is similar to that observed among ct embryos following maternal supplementation with formate. 17Formate is a one-carbon donor to FOCM, which supports biosynthesis of purines and dTMP and, like MI, can prevent NTDs in ct embryos. 17Interestingly, direct supplementation of ct mice with combinations of thymidine and purine nucleotides has previously been found to prevent spinal and cranial NTDs. 39Inositol treatment and nucleotide supplementation have each been shown to correct the hindgut endoderm cell proliferation defect that is known to be the cause of spinal NTDs in ct embryos. 13,39Hence, we hypothesize that MI and DCI treatment act to support the production of nucleotides required for rapid cellular proliferation occurring in the embryo during neural tube closure.possible mode of action of inositol via nucleotide supply is equivalent to one of the proposed mechanisms by which supplemental folic may act to prevent NTDs.For example, FA supports thymidylate biosynthesis and suppresses uracil misincorporation into DNA, 40,41 and stimulation of cell cycle progression in the neuroepithelium has been observed in a FA-responsive mouse model of NTDs. 42The fact that FA does not prevent NTDs in the ct strain may reflect the dual requirement in FOCM for both the tetrahydrofolate backbone molecule (which can be generated by reduction of supplemental FA) and one-carbon supply (which can be substituted by supplemental formate).DCI has been found to have a greater protective effect than MI at the same dose and is effective at lower doses. 14Analysis of the metabolite profile of MI-and DCIsupplemented embryos revealed considerable overlap of their effect suggesting that protective mechanism(s) may be shared, for example, via enhanced supply of nucleotides.Interconversion of the two enantiomers can be mediated by an endogenous epimerase, 43,44 suggesting that DCI could act to enhance MI production or vice versa.However, the interconversion rate may be low in vivo, 45 and activity in the embryo has not been determined.The present study shows that DCI supplementation increases DCI but not MI abundance in maternal circulation or the embryo.Hence, increased effectiveness of DCI compared with MI appears unlikely to be mediated via enhanced delivery and conversion to MI.It is possible that DCI has a greater magnitude of effect on those metabolites that are affected by both MI and DCI, and we found evidence for a trend toward this.Alternatively, there may be a protective effect of one or more of the metabolites that are specifically affected by DCI, which include some nucleotides but also a number of amino acids and groups of lipids such as mono-and dihydroxy fatty acids.Other possibilities are that the lowering of glucose content in MI-treated embryos provides a limitation that is not incurred by DCI, or that DCI has effects on metabolites that were not present on the panel analyzed in this study.
In summary, the ct mouse provides a model system to investigate potential risk factors and preventive treatments for NTDs.Supplemental inositol can prevent NTDs in ct mice and we hypothesize that this is mediated, at least in part, through the enhanced production of nucleotides.Pilot clinical studies have suggested that there may also be a protective effect of maternal MI supplements in preventing the recurrence of human NTDs. 19,20If confirmed, this would potentially offer a means to enhance NTD prevention using inositol in combination with folic acid.

F I G U R E 1
Influence of dietary intake of myo-inositol on levels in maternal plasma and embryo.(A) Plasma myo-inositol (MI) concentration in pregnant mice of the C57Bl/6 and curly tail (ct) strains maintained with differing dietary MI status.Compared with the standard diet, plasma MI was significantly lower in mice maintained on an MI-deficient diet (*p < .05,**p < .005;ANOVA) and significantly elevated with oral MI supplementation ( # p < .05).(B) MI content (normalized to protein content) in embryos at E10.5 after exposure to differing maternal dietary MI states.In both strains, maternal MI deficiency led to lower MI content than both other dietary regimes (*p < .005;ANOVA), whereas oral MI supplementation did not significantly increase embryonic MI beyond values on the standard diet.There were no differences between strains maintained on the same diet.Embryo numbers on standard, MI-deficient, and MI-supplemented diets: 6, 5, 6 for C57Bl/6; 19, 21, 18 for ct embryos.

F I G U R E 2
Embryo myo-inositol (MI) content at E10.5 among embryos subdivided by maternal diet and neural tube defect (NTD) phenotype.Embryonic MI content did not differ with NTD phenotype under maternal MI deficiency, but significantly varied by phenotype under standard dietary conditions (p < .01,ANOVA) or with supplemental MI (p < .05,ANOVA).MI content was significantly lower in embryos with failed cranial neural tube closure than in unaffected embryos (*p < .005,Holm-Sidak method).Embryos with an enlarged PNP exhibited a non-significant trend toward lower MI contents compared with unaffected littermates.Values for individual embryos are shown, together with mean ± SEM for each group.

F I G U R E 3
Expression of the inositol synthesis pathway in mouse embryos during neural tube closure.(A) Summary of the key steps in inositol (I) synthesis, leading to production of phosptatidylinositol (PI), highly phosphorylated phosphoinositides, and inositol phosphates (IP) that can be recycled to inositol.(B-I) Whole mouse in situ hybridization shows widespread expression of Isyna1 (B-E) and Impa1 (F-I).Expression of Isyna1 is detected in the neuroepithelium (ne) of the spinal neural tube (D) and open neural folds (E), but with lower or absent expression in the cranial neural tube (arrow in C).Impa1 expression is present in the neural tube at all axial levels (G, I).Dashed lines in B and F indicate planes of section in C-E and G-I respectively.

F I G U R E 4
The protective effect of supplemental myo-inositol (MI) on posterior neuropore (PNP) closure is dependent on IMPase activity.(A) PNP length of ct embryos at the 30-31 somite stage following embryo culture for 24 h from E9.5.MI supplementation of the culture media leads to a significant reduction in PNP length (*significantly different compared with control, p < .05),whereas this effect is abrogated by co-administration of L690330.Addition of L690330 alone has no effect on PNP length.(B) Maternal MI administration by intraperitoneal injection leads to a significant reduction in PNP length among ct embryos at the 30-31 somite stage (E10.5)compared with vehicle-treated controls (*p < .05).I G U R E 5 of metabolites in myo-inositol (MI)-and d-chiro-inositol (DCI)-treated embryos at E10.5.Metabolite abundance was determined by LC-MS/MS in embryos exposed to maternal inositol supplementation (n = 5-7 groups per condition, two embryos per group).Embryonic MI (A), DCI (B), and (C) glucose content were significantly affected by maternal MI deficiency (−MI), or administration of MI (+MI) or DCI (+DCI) (*significantly different from standard diet (Std), p < .05).(D, E) Altered abundance of metabolites in (D) MI-and (E) DCI-treated embryos shown by plot of p-value (log10) against relative abundance (compared with Std): the horizontal dotted line corresponds with p < .05 and vertical dotted line corresponds with mean value in control (standard diet) embryos.Colored dots correspond to metabolites that significantly differ from standard diet (p < .05).Note that DCI (139-fold increase compared with standard was omitted from graph).Example metabolites are labeled (2-AH, 2-aminoheptanoate; 5-av, 5-aminovalerate; BHBA, 3-hydroxybutyrate, DHB, 2S,3R-dihydroxybutyrate; meCys, Smethylcysteine; Trig, trigonelline).(F) Numbers of metabolites significantly altered by supplemental MI and/or DCI (total numbers and number after filtering to remove metabolites also affected by MI deficiency).(G) Analysis of inositol-responsive metabolites.Total number of metabolites per family (filtered as in F) is indicated, with proportion altered by MI and/or DCI indicated by colored segments.

F
I G U R Altered abundance of nucleotides and folates in embryos exposed to maternal myo-inositol (MI) supplementation.Among nucleotides and related metabolites affected by maternal MI supplementation the relative abundance of AMP (A), GMP (B), TMP (C) and adenoylsuccinate (D) were more abundant in embryos from MI or d-chiro-inositol (DCI)-treated litters compared with controls (*significantly different from embryos that developed under standard dietary (Std) conditions, p < .05).(E) Maternal MI treatment led to a shift in the embryonic folate profile, with increased abundance of THF and decreased abundance of 5-methyl THF (*significantly different from standard diet, p < .05;n = 5 embryos per group).DHF, dihydrofolate; THF, tetrahydrofolate; CH-THF, methyenyl-THF; 5mTHF, 5-methyl THF; CHO-THF, formyl-THF).
K.-Y.L., A.J.C., and N.D.E.G. conceived and designed the research; K.-Y.L., E.W., S.C.P.D., E.N., S.S., D.S., S.E., and N.D.E.G. conducted the research and analyzed the data; N.D.E.G., K.-Y.L., and A.J.C. wrote and edited the paper and all authors approved the draft.NG had primary responsibility for the final content.