Disruption of Folate Metabolism Causes Poor Alignment and Spacing of Mouse Conceptuses for Multiple Generations

Abnormal uptake or metabolism of folate increases risk of human pregnancy complications, though the mechanism is unclear. Here, we explore how defective folate metabolism influences early development by analysing mice with the hypomorphic Mtrr gt mutation. MTRR is necessary for methyl group utilisation from folate metabolism, and the Mtrr gt allele disrupts this process. We show that the spectrum of phenotypes previously observed in Mtrr gt/gt conceptuses at embryonic day (E) 10.5 is apparent from E8.5 including developmental delay, congenital malformations, and placental phenotypes. Notably, we report misalignment of some Mtrr gt conceptuses within their implantation sites from E6.5. The degree of misorientation occurs across a continuum, with the most severe form visible upon gross dissection. Additionally, some Mtrr gt/gt conceptuses display twinning. Therefore, we implicate folate metabolism in blastocyst orientation and spacing at implantation. Skewed growth likely influences embryo development since developmental delay and heart malformations (but not defects in neural tube closure or trophoblast differentiation) associate with severe misalignment of Mtrr gt/gt conceptuses. Typically, the uterus is thought to guide conceptus orientation. To investigate a uterine effect of the Mtrr gt allele, we manipulate the maternal Mtrr genotype. Misaligned conceptuses were observed in litters of Mtrr +/+ , Mtrr +/gt , and Mtrr gt/gt mothers. While progesterone and/or BMP2 signalling might be disrupted, normal decidual morphology, patterning, and blood perfusion are evident at E6.5 regardless of conceptus orientation. These observations argue against a post-implantation uterine defect as a cause of conceptus misalignment. Since litters of Mtrr +/+ mothers display conceptus misalignment, a grandparental effect is explored. Multigenerational phenotype inheritance is characteristic of the Mtrr gt model, though the mechanism remains unclear. Genetic pedigree analysis reveals that severe conceptus skewing associates with the Mtrr genotype of either maternal grandparent. Moreover, the presence of conceptus skewing after embryo transfer into a control uterus indicates that misalignment is independent of the peri- and/or post-implantation uterus and instead is likely attributed to an embryonic mechanism that is epigenetically inherited. Overall, our data indicates that abnormal folate metabolism influences conceptus orientation over multiple generations with implications for subsequent development. This study casts light on the complex role of folate metabolism during development beyond a direct maternal effect.


INTRODUCTION
Maternal folate deficiency in humans is associated with increased risk of congenital malformations (MRC Vitamin Study Research Group, 1991) and pregnancy complications, such as placental abruption, hemorrhage, and preeclampsia (Wen et al., 2008). These complications potentially lead to increased risk of spontaneous abortion, miscarriage, or pre-term birth (Bukowski et al., 2009). Yet, how folate metabolism functions in pregnancy establishment is not well understood. Folate, a vitamin that underpins one-carbon metabolism, is required for thymidine synthesis (Stover, 2004) and transfer of one-carbon methyl groups for cellular methylation reactions (Ducker and Rabinowitz, 2017). Therefore, folate is required for rapid cell division and epigenetic regulation of gene expression, which are both important for feto-placental development.
At implantation, human and mouse blastocysts embed into the uterine wall to improve fetal access to maternal resources. Mouse litters consist of 5-20 embryos that must be optimally spaced and oriented within the uterus for ideal growth. Spacing mechanisms are not well understood but likely involve mechanical forces generated by the uterus and cross talk between blastocysts and the uterine epithelium (Chen et al., 2013;Flores et al., 2020). The orientation of the developing mouse conceptus aligns with the antimesometrial-mesometrial axis of the decidual swelling, such that the placenta of each conceptus forms on the mesometrial side of the uterus (Smith, 1985). For normal alignment to occur, mouse blastocysts attach to the antimesometrial uterine epithelium via the mural trophectoderm (opposes the inner cell mass (ICM)) by embryonic day (E) 4.5 and invade into the decidualizing endometrium (Smith, 1985;Cross et al., 1994). In contrast, human blastocysts attach to the uterus via the polar trophectoderm (cells that contact the ICM) (Hertig et al., 1956). In mouse, the polar trophectoderm forms the ectoplacental cone (EPC) and extra-embryonic ectoderm, which together consist of progenitor populations of the mature placenta (Simmons and Cross, 2005;Watson and Cross, 2005). The base of the EPC abuts the chorion and the tip of the EPC extends upwards through the mesometrial decidua. The embryo forms below the chorion on the antimesometrial side of the implantation site. The maternal and embryonic factors required to establish normal conceptus orientation are not well studied, though it is likely determined by blastocyst orientation at implantation (Kirby et al., 1967;Red-Horse et al., 2004;Chen et al., 2013) and/or signals from the maternal decidua during post-implantation development (Alexander et al., 1996;Daikoku et al., 2011;Winterhager et al., 2013;Cha et al., 2014;Zhang et al., 2014). Furthermore, the extent to which deviation from normal conceptus alignment influences embryo and placenta development is not well understood.
Whether folate metabolism plays a role in implantation, spacing, and/or orientation remains unclear. Several mouse models of defective folate uptake, transport or metabolism suggest involvement of folate metabolism in early pregnancy. For instance, dietary folate deficiency in mice is associated with sub-fertility, increased embryonic lethality attributed to defects in decidualization (Gao et al., 2012;Geng et al., 2015), and low weight placentas with shallow trophoblast invasion (Pickell et al., 2009). During its metabolism, folate is converted to 5methyltetrahydrofolate by methylenetetrahydrofolate reductase (MTHFR). Methionine synthase (MTR) then catalyzes the transfer of the methyl group from 5-methyltetrahydrofolate to homocysteine to produce methionine (Shane and Stokstad, 1985). Methionine is the precursor of S-adenosylmethionine, the methyl donor of all methylation reactions (Ducker and Rabinowitz, 2017). Importantly, methionine synthase reductase (MTRR) ensures progression of both folate and methionine cycles by continued activation of MTR through reductive methylation of its vitamin B 12 co-factor (Yamada et al., 2006). Mutations in human MTHFR, MTR, and MTRR genes are linked to pregnancy complications (Ren and Wang, 2006;Furness et al., 2008;Kim et al., 2013;Wang et al., 2015), yet the mechanism of one-carbon metabolism in blastocyst spacing and conceptus orientation requires further study. Mutations in mouse genes associated with folate uptake or metabolism display gross placental phenotypes (e.g., Slc19a1 −/− , Mthfr −/− and Mtrr gt/gt ) or are yetto-be assessed for defects beyond the embryo proper (e.g., Folbp1 −/− , Folbp2 −/− , Slc46a1 −/− , and Mtr −/− ) (Piedrahita et al., 1999;Swanson et al., 2001;Elmore et al., 2007;Gelineau-van Waes et al., 2008;Pickell et al., 2009;Salojin et al., 2011;Padmanabhan et al., 2013). Nevertheless, many of these mutations cause embryonic lethality by E10.5, a telltale sign of severe defects in placentation (Watson and Cross, 2005).
We previously reported the Mtrr gt knockdown mutation generated through a gene-trap (gt) insertion into the mouse Mtrr locus (Elmore et al., 2007;Padmanabhan et al., 2013). The Mtrr gt allele sufficiently disrupts one-carbon metabolism by significantly reducing MTR activity (Elmore et al., 2007). Indeed, Mtrr gt/gt mice exhibit characteristics similar to clinical indicators of human folate deficiency (Krishnaswamy and Madhavan Nair, 2001) including plasma hyperhomocysteinaemia, macrocytic anemia in adulthood, and a wide spectrum of developmental phenotypes at E10.5 including failure of neural tube closure (Elmore et al., 2007;Padmanabhan et al., 2013;Padmanabhan et al., 2018). Therefore, the Mtrr gt mutation is a robust model to assess the effects of abnormal folate metabolism. Besides neural tube closure defects, other developmental phenotypes observed in Mtrr gt/gt conceptuses at E10.5 are growth restriction, developmental delay, cardiac malformations, and placenta phenotypes (Padmanabhan et al., 2013). While the molecular mechanism and developmental timing of these phenotypes is currently unclear, alterations in genome-wide DNA methylation associated with gene misexpression are apparent in Mtrr gt/gt conceptuses at E10.5 (Padmanabhan et al., 2013;Bertozzi et al., 2021;Blake et al., 2021). This result implicates an epigenetic mechanism.
Interestingly, the Mtrr gt mouse line also displays transgenerational epigenetic inheritance of developmental phenotypes (Padmanabhan et al., 2013). Transgenerational epigenetic inheritance is a non-conventional mode of inheritance caused by environmental stressors and/or metabolic disruption present only in the F0 generation that results in phenotypes at least up to the F3 generation (Blake and Watson, 2016). The mechanism of mammalian transgenerational epigenetic inheritance is not well understood. It is hypothesized that an epigenetic factor(s) is inherited via the germline in a manner that is independent of DNA mutation (Blake and Watson, 2016), yet it is difficult to tease apart the effects of the uterine environment from those intrinsic to the germline. In the case of the Mtrr gt mouse line, an Mtrr gt allele in either maternal grandparent (i.e., the F0 generation) is sufficient to disrupt germ cell DNA methylation  and cause growth defects and congenital malformations at E10.5 in the wildtype grandprogeny, in some cases up to the F4 wildtype generation (Padmanabhan et al., 2013). This phenomenon occurs when the F1 generation and all subsequent generations are wildtype for the Mtrr gene and display normal folate metabolism (Padmanabhan et al., 2013;Padmanabhan et al., 2018). While robust genetic pedigree analysis and embryo transfer experiments showed that congenital malformations occurred independent of the uterine environment (Padmanabhan et al., 2013), the specific mechanism of transgenerational epigenetic inheritance is yet-to-be determined in this and other models.
The overall study aim is to understand how defective folate metabolism due to the Mtrr gt mutation influences early development and the uterine-and embryo-specific effects involved. Notably, we observe a spectrum of conceptus misalignment from E6.5. The most severe form of this phenotype is visible at gross dissection as "eccentric placenta development" and associates with specific embryonic defects at midgestation. This finding has implications for the origin of some congenital malformations associated with folate deficiency. Misalignment occurred alongside twinning, which links folate metabolism to blastocyst orientation and spacing at implantation. To better understand the mechanisms behind conceptus misalignment, we investigate a maternal effect of the Mtrr gt allele. Data collected from histological analyses, highly controlled genetic pedigrees, and embryo transfer experiments reveal that severe conceptus misalignment occurs independent of the peri-and/or post-implantation uterus. This result differs from other mouse models displaying skewed growth. Instead, our data supports an embryo-specific defect as a cause of conceptus misorientation that is epigenetically inherited from either Mtrr +/gt maternal grandparent. Overall, we propose that defective folate metabolism disrupts blastocyst orientation and spacing over multiple generations, with implications for subsequent feto-placental development and a mechanism beyond a direct maternal effect.

Ethics
All experiments were performed in accordance with UK Home Office regulations under the Animals (Scientific Procedures) Act 1986 Amendment Regulations 2012 and underwent review by the University of Cambridge Animal Welfare and Ethical Review Body.

Mice
The Mtrr gt mouse line was generated as previously described (Padmanabhan et al., 2013). Briefly, a gene-trap (gt) vector was inserted into intron 9 of the Mtrr gene. Upon germline transmission, the Mtrr gt allele was backcrossed to C57Bl/6J mice for eight generations (Padmanabhan et al., 2013). C57Bl/ 6J mice (The Jackson Laboratory), which are wildtype for the Mtrr gene, were used as controls and were bred in house separately from the Mtrr gt mouse line. All mice were housed in a temperature-and humidity-controlled environment with a 12-h light/dark cycle, and fed a standard chow (Rodent No. 3 chow; Special Diet Services, Essex, United Kingdom) ad libitum from weaning (includes folic acid at 2.99 mg/kg of diet, methionine at 0.37%, choline at 1422.4 mg/kg of diet, and vitamin B 12 at 19.2 μg/kg of diet). Mtrr gt/gt conceptuses were derived from Mtrr gt/gt intercrosses unless otherwise stated. For assessment of maternal effect, Mtrr +/+ , Mtrr +/gt and Mtrr gt/gt female mice derived from Mtrr +/gt intercrosses were mated with C57Bl/6J males, and the resulting litters were dissected at E6.5. For assessment of maternal grandparental effect, F0 Mtrr +/gt male or female mice were crossed with a C57Bl/6J mate. Their F1 Mtrr +/+ daughters were mated with C57Bl/6J males to generate F2 Mtrr +/+ progeny for dissection at E10.5 or for littering out. F2 Mtrr +/+ females were mated with C57Bl/6J males to generate F3 Mtrr +/+ progeny for dissection at E10.5 or littering out. F3 Mtrr +/+ females were mated with C57Bl/6J males to generate F4 Mtrr +/+ conceptuses for dissection at E10.5.

Genotyping
DNA samples were obtained from ear tissue or yolk sac for polymerase chain reaction (PCR) genotyping of the Mtrr gt allele or sex as reported previously (Padmanabhan et al., 2013;Tunster, 2017).

Dissections and Phenotyping
Noon of the day that the vaginal plug was detected was considered E0.5. Pregnant female mice were euthanized via cervical dislocation. Dissections were performed in cold 1x phosphate buffered saline (PBS). Each conceptus was individually scored for phenotypes (see below). Whole implantation sites at E6.5, or placentas and embryos separately at E8.5, E10.5, E14.5, and E18.5 Frontiers in Cell and Developmental Biology | www.frontiersin.org December 2021 | Volume 9 | Article 723978 were weighed, photographed, and either snap frozen in liquid nitrogen for molecular analysis or fixed in 10% natural-buffered formalin (Cat. No. HT501128; Sigma-Aldrich) at 4°C overnight and paraffin embedded for sectioning using standard protocols. A rigorous phenotyping regime was performed as previously described (Padmanabhan et al., 2013). Briefly, conceptuses were defined as "phenotypically normal" if they displayed 6-12 somite pairs at E8.5 and 30-40 somite pairs at E10.5 (according to e-Mouse Atlas Project; http://www.emouseatlas.org) and displayed no growth defect or gross abnormalities. For analysis at E10.5 only, an embryo was growth restricted if its crown-rump length was less than two standard deviations from the control C57Bl/6J mean crown-rump length. Growth restricted embryos also displayed normal somite pair counts and no other phenotype. At E8.5 and E10.5, developmentally delayed conceptuses exhibited <6 somite pairs or <30 somite pairs, respectively, but were otherwise normal for the developmental stage indicated by the somite pair counts. Severely affected conceptuses displayed one or more abnormality with appreciation for the developmental stage (e.g., somite pair count), as some of these embryos were also developmentally delayed. Abnormalities included: 1) delay neural tube closure whereby the length of the neural plate that adhered to form a tube was shorter than expected for the somite stage at E8.5, or failure of neural tube closure within the cranial or spinal cord region at E10.5, 2) heart defects (E10.5 only) including heart looping defect, pericardial edema, or hemorrhage, 3) placenta defects including failure of chorioallantoic attachment or eccentrically located placentas, or 4) "twinning" whereby two or more conceptuses shared a single implantation site. An implantation site that consisted of a decidual swelling with an ill-defined mass of fetal-derived cells was categorized as a resorption and considered dead. Conceptuses were allocated a phenotypic category based on their most severe phenotype and were only counted once.

Embryo Transfer
Embryo transfer experiments were performed and analyzed as previously described in detail (Padmanabhan et al., 2013). Briefly, using M2 media (Cat. No. M7167; Sigma-Aldrich), preimplantation wildtype embryos were flushed at E3.25 from the oviducts and uteri of Mtrr +/+ females (derived from one Mtrr +/gt parent) mated with C57Bl/6J males. Donor females were not superovulated. After a brief embryo culture period to prepare the recipient females, embryos were surgically transferred into the uteri of E2.5 pseudopregnant [C57Bl/6J × DBA/2] F1 hybrid (B6D2F1) females (Charles River). Control C57Bl/6J embryos were similarly transferred. Donor and recipient females were 7-10 weeks old. Embryos were transferred regardless of appearance and litters were never pooled. Transferred embryos were dissected at E10.5, the timing of which corresponded to the staging of the recipient female (Ueda et al., 2003), and scored for phenotypes as above.

In Situ Hybridization
In situ hybridisation probes were generated with gene-specific primer sets (Supplementary Table S1) that contained T7 RNA polymerase promoter sequence (TAATACGACTCACTATAG GG) attached to each reverse primer and T3 RNA polymerase promoter sequence (AATTAACCCTCACTAAAGGG) attached to each forward primer (Outhwaite et al., 2019). Probe synthesis of digoxigenin (DIG)-labelled anti-sense and sense probes was performed via PCR using DIG RNA labelling Mix (Cat. No. 11277073910; Roche, Welwyn Garden City, United Kingdom) according to the manufacturer's instructions. The in situ hybridization protocol was performed as previously described (Simmons et al., 2008) with the following alteration: paraffinembedded tissue sections were used and thus, an initial dewaxing/ rehydration step was added.

Statistics
Statistical analyses were performed using GraphPad Prism 8 software (La Jolla, CA, United States). Parametric data was analyzed by unpaired t test or one-way ANOVA. Nonparametric data was analyzed by Mann Whitney test. Correlations were analyzed by linear regression or Fisher's exact test. p < 0.05 was considered significant.

Software and Imaging
Primer sequences were designed using the NCBI primer design tool (Ye et al., 2012). Conceptuses were dissected using a Zeiss SteReo Discovery V8 microscope and photographed with an AxioCam MRc5 camera and AxioVision 4.7.2 software (Carl Zeiss). Tissue sections were scanned using a NanoZoomer (Hamamatsu Photonics, United Kingdom) and viewed using NDP Scan 2.7 software (Hamamatsu Photonics). Images were analyzed using ImageJ software (NIH, Bethesda, MD, United States). Graphs were generated using Graphpad Prism 8 software.

Mtrr gt/gt Conceptuses Exhibit
Developmental Phenotypes at E8.5 To better understand the effects of the Mtrr gt/gt mutation on early mouse development, we first determined where MTRR enzyme functions in the conceptus by performing immunohistochemistry on histological sections of whole C57Bl/6J implantation sites at E8.5. MTRR protein was widely expressed in wildtype embryonic and extraembryonic cell types including the yolk sac, amnion, allantois, and trophoblast progenitors (i.e., the chorion and EPC) at E8.5 ( Figures 1A-C). This expression pattern was consistent with other folate metabolic enzymes and transporters at E8.5 (Cherukad et al., 2012). While expressed throughout the mesometrial and lateral decidua ( Figure 1A), the antimesometrial decidua displayed striped MTRR expression in the primary and secondary decidual zones at E8.5 ( Figure 1C) suggesting regionalized expression in the decidua. Overall, the broad expression of MTRR protein in wildtype conceptuses at E8.5 suggested a potential mechanistic role of folate metabolism in early fetoplacental development and decidualization. Mtrr gt/gt embryos and placentas at midgestation display a substantial decrease in wildtype Mtrr transcript levels (6-24% of control levels) as determined by RT-qPCR (Padmanabhan et al., 2013), emphasizing the knockdown effect of the Mtrr gt allele. Next, to explore the extent of developmental disruption caused by the Mtrr gt/gt mutation before E10.5, rigorous phenotype assessment of Mtrr gt/gt conceptuses was performed at E8.5. Mtrr gt/gt conceptuses derived from Mtrr gt/gt intercrosses were scored for gross abnormalities (N 50 conceptuses from six litters). C57Bl/6J conceptuses, which are genetically wildtype for Mtrr and bred separately from the Mtrr gt mouse line, were used as controls (N 43 conceptuses from five litters). This is because the Mtrr gt allele was backcrossed into the C57Bl/6J mouse strain and Mtrr +/+ conceptuses derived from Mtrr +/gt intercrosses display multigenerational inheritance of developmental phenotypes (Padmanabhan et al., 2013). Similar to E10.5 (Padmanabhan et al., 2013), Mtrr gt/gt litter sizes and resorption rates were consistent with controls at E8.5 ( Figure 1D; Table 1). Allocation to a phenotypic category depended upon the most severe phenotype exhibited by each conceptus (see methods). The majority of C57Bl/6J conceptuses at E8.5 were phenotypically normal (79.1% of conceptuses), displaying 6-12 somite pairs and no other obvious abnormalities (Figures 1D,E; Table 1). Nearly all other C57Bl/6J conceptuses were resorbed (18.6% of conceptuses; Figure 1D; Table 1), appearing as decidual swellings with an ill-defined mass of fetal-derived cells. In contrast, only 42.0% of Mtrr gt/gt conceptuses at E8.5 were phenotypically normal upon gross inspection (Figures 1D,F; Table 1; p 0.009). The remaining Mtrr gt/gt conceptuses were either resorbed (8.0% of conceptuses; Figure 1D;  Table 1). Defects that were considered severe were delay of neural tube closure (10% of conceptuses; Table 1) whereby the length of the neural plate that formed a tube was shorter than expected based on somite pair count at E8.5 ( Figure 1H) and might lead to failure of neural tube closure by E10.5 ( Figures 1I-L), and/or placenta phenotypes whereby chorioallantoic attachment was absent (4% of conceptuses) or the placenta was eccentrically located (12% of conceptuses; Figures 1P,Q; Table 1). In contrast, C57Bl/6J conceptuses at E8.5 infrequently displayed developmental delay (2.3% of conceptuses; Table 1) and severe abnormalities were undetected (Figures 1D,I,O; Table 1). The phenotypes in Mtrr gt/gt conceptuses at E8.5 were similar to those observed at E10.5 ( Figures 1J,K), though gross heart defects including hemorrhage, enlargement, and pericardial edema observed at E10.5 ( Figures 1L-N) (Padmanabhan et al., 2013) were not observed at E8.5. A more detailed assessment of the heart at E8.5 is required in the future. Overall, some developmental phenotypes caused by the Frontiers in Cell and Developmental Biology | www.frontiersin.org December 2021 | Volume 9 | Article 723978 Frontiers in Cell and Developmental Biology | www.frontiersin.org December 2021 | Volume 9 | Article 723978 6 Mtrr gt mutation likely originate prior to or at E8.5 (e.g., neural tube closure defects and placenta defects) while others do not become apparent until after E8.5 (e.g., heart defects).

A Low Frequency of Twinning Occurs in Mtrr gt/gt Conceptuses Throughout Gestation
During the phenotypic assessment of Mtrr gt/gt conceptuses, we also observed a low frequency of dizygotic twinning that was apparent from E6.5 to E18.5 (0.7-2.3% of implantation sites, N 118-308 sites (stage specific)) ( Figures 2A,B). Twinning occurred when at least two conceptuses occupied a single decidual swelling. This phenotype was never detected in C57Bl/6J controls (N 314 conceptuses). The occurrence of male and female embryos together in a single implantation site indicated dizygosity and suggested that twinning in the Mtrr gt mouse line likely resulted from poor blastocyst spacing at implantation. However, similar to controls, Mtrr gt/gt implantation sites were generally evenly spaced along the length of Mtrr gt/gt uteri ( Figure 2C) rather than clustered together as is observed when there is a major uterine defect (Lu et al., 2013;Flores et al., 2020). Also, no more than one twinned implantation site was usually observed per Mtrr gt/gt litter (this study; Padmanabhan et al., 2013) indicating incomplete phenotype penetrance. Therefore, a uterine-specific defect was unlikely. Instead, we hypothesise that twinning caused by the Mtrr gt/gt mutation reflected an embryo-specific defect.

Mtrr gt/gt Mutation Causes Skewed Orientation of the Entire Conceptus
Given the high frequency of gross placental phenotypes identified in Mtrr gt/gt conceptuses at E8.5 and E10.5 (Padmanabhan et al., 2013), the placenta was assessed further. No placental phenotypes were detected in C57Bl/6J control conceptuses at E8.5 or E10.5 (Table 1; Figures 1O,R) (Padmanabhan et al., 2013). However, Mtrr gt/gt conceptuses showed two distinct gross placental phenotypes. The first defect was failure of the allantois to contact and attach to the chorion (E8.5: 4.0% of conceptuses ( Table 1); E10.5: 0.3% of conceptuses (Padmanabhan et al., 2013)), which prevents subsequent placental formation and causes embryonic lethality (Cross et al., 2003). The cause of this chorioallantoic attachment defect was unclear and it was not considered further here. The second phenotype was an eccentric placenta whereby the region of allantoic attachment appeared askew within the implantation site (E8.5: 12.0% of conceptuses (  (Padmanabhan et al., 2013)). Declining phenotypic frequencies during gestation suggested that the phenotype normalized or some Mtrr gt/gt conceptuses with these placental defects underwent lethality between E8.5 and E10.5.
Data displayed as mean incidence per litter ± se. Percentage of all conceptuses assessed in brackets. a p < 0.01. b p < 0.05. c Conceptuses that displayed severe abnormalities were often developmentally delayed. These conceptuses were only counted once and were attributed to the most severe phenotype category. Eccentric placenta location, no chorioallantoic attachment, and neural tube closure defects were subcategories of severe abnormalities.
Frontiers in Cell and Developmental Biology | www.frontiersin.org December 2021 | Volume 9 | Article 723978 the allantois was attached in a normal manner in the center of the chorion. However, a proportion of Mtrr gt/gt placentas displayed trophoblast compartments (including the chorion and EPC) that were noticeably askew within their decidual swellings ( Figures  3A-C). The degree of skewed growth was determined by measuring the angle between the midline of the decidual swelling and the midline of the chorion or EPC ( Figures  3A-C Figures  3D,E). Given the spectrum of phenotypic severity, we propose that the eccentrically-located placenta phenotype identified during gross dissection likely denotes the extreme end of the skewed conceptus orientation spectrum. Moreover, less severe conceptus skewing was likely unappreciated during gross dissection.
Embryonic Phenotypes Associate With Skewed Orientation in Mtrr gt/gt Conceptuses at E10.5 To determine whether embryonic phenotypes were associated with conceptus misalignment, we assessed our phenotyping data from Mtrr gt/gt conceptus dissections at E10.5 (Padmanabhan et al., 2013;Padmanabhan et al., 2017) for overlap between severe conceptus skewing (denoted as "eccentric placenta development" during gross dissection in this study and "eccentric chorioallantoic attachment" in a previous study (Padmanabhan et al., 2013)) and embryo growth defects or congenital malformations. Refer to our other published work for a full assessment of the individual phenotypic outcomes in Mtrr gt/gt conceptuses at E10.5 (Padmanabhan et al., 2013;Padmanabhan et al., 2017). We selected E10.5 for the time point of assessment due to the size of this data set (>215 Mtrr gt/gt conceptuses at E10.5) and because the detrimental effects of misaligned orientation on development might be more apparent at this stage. A caveat of this analysis  Separately, we observed that conceptuses with severe skewing showed an association with heart malformations including pericardial edema or reversed heart looping (9/18 of misaligned Mtrr gt/gt embryos had a heart abnormality; Fisher's exact test, p < 0.0001; Figures 3I,K,L). Yet, none of the 18 Mtrr gt/gt embryos that failed to close their neural tube in either the cranial region ( Figure 3L) or spinal cord region by Frontiers in Cell and Developmental Biology | www.frontiersin.org December 2021 | Volume 9 | Article 723978 E10.5 (Padmanabhan et al., 2013) displayed severe skewing ( Figure 3J) indicating a lack of correlation between phenotypes. While a causal relationship is yet-to-be established, it is possible that skewed conceptus orientation might influence the rate of embryonic development and/or heart formation.

Skewed Orientation Does Not Alter
Trophoblast Differentiation of Mtrr gt/gt Conceptuses at E8.5 To explore whether skewed conceptus orientation altered the development of the trophoblast population, we performed in situ hybridization to detect the expression of key chorion and EPC trophoblast marker genes (Simmons et al., 2008). Histological sections of C57Bl/6J control (N 6 placentas) and Mtrr gt/gt placentas (N 14 placentas) were assessed at E8.5. Mtrr gt/gt placentas were subdivided into those with aligned and skewed orientation. Trophoblast marker gene expression of Tpbpa (EPC marker), Hand1 (sinusoidal TGC progenitor and P-TGCs), Syna (syncytiotrophoblast-I progenitors) and Gcm1 (syncytiotrophoblast-II progenitors) was present in all Mtrr gt/gt placentas assessed, regardless of the degree of skewing (Figures 3A-C, 4A-L). These data indicated that the trophoblast populations were present and patterned in Mtrr gt/gt placentas, and thus their formation was not greatly affected by skewed conceptus growth. However, Prl7b1 mRNA expression (invasive trophoblast cell marker) was reduced in Mtrr gt/gt placentas with skewed orientation. The few Prl7b1+ cells that were present were located in the lateral decidua rather than in the mesometrial decidua ( Figures 4M-O). Mislocalization of invasive trophoblast cells might have significant implications for uterine spiral artery remodelling and placental function, and should be explored in future studies.
Mtrr gt/gt Conceptus Misalignment Is Not Associated With Decidual Blood Sinus Area at E8.5 To better understand the cause of conceptus misalignment, we investigated a potential maternal effect of the Mtrr gt allele. Another model of skewed EPC growth exhibits dilated Frontiers in Cell and Developmental Biology | www.frontiersin.org December 2021 | Volume 9 | Article 723978 decidual blood sinuses (Winterhager et al., 2013). This association suggested that defects in the perfusion or structure of the maternal decidua might lead to skewed orientation of conceptus growth. Therefore, we measured the combined blood sinus area in the lateral and mesometrial decidual regions in centrally located transverse histological sections of C57Bl/6J and Mtrr gt/gt implantation sites at E8.5 (N 6-12 placentas/group). Notably, the Mtrr gt/gt conceptuses were derived from an Mtrr gt/gt mother and father. Both aligned and skewed Mtrr gt/gt conceptuses were separately considered. The average proportion of the decidua attributed to blood sinuses was similar in C57Bl/6J, Mtrr gt/gt aligned, and Mtrr gt/gt skewed placentas (p 0.259; Figure 4, Figures 5A-D). Furthermore, there was no correlation between decidual blood sinus area and chorion angle (p > 0.30; Supplementary Figure S2A) or EPC angle (p > 0.14; Figure 5E) in either Mtrr gt/gt group assessed compared to controls. Next, we examined whether left-right asymmetry in blood sinus area was evident and potentially guided the direction of conceptus misalignment. To do this, the ratio of right:left decidua blood sinus area was calculated. Graph showing average area of decidual blood sinuses (mean ± s.d.) per total decidua area assessed in histological sections of C57Bl/6J (grey bar) and Mtrr gt/gt (black and blue bars) placentas at E8.5. Aligned (Al) and skewed (Sk) placentas are represented (N 6-12 placentas/group, with at least three sections assessed per placenta). One-way ANOVA, p 0.2586. (E) Linear regression analyses of decidua blood sinus area in relation to EPC angle for C57Bl/6J and Mtrr gt/gt conceptuses. Aligned and skewed Mtrr gt/gt placentas were considered separately (N 6-12 placentas/group). (F) Data depicting the ratio of decidua blood sinus area on the right and left sides of implantation sites from C57Bl/6J (C57) and Mtrr gt/gt (gt/gt) conceptuses at E8.5. Data represented as mean ± sd. Aligned (Al) and skewed (Sk) placentas were considered separately (N 6-12 placentas/group). Dashed red line indicates a ratio of 1. One-way ANOVA, p 0.6050. (G) Linear regression analysis to determine whether conceptus skewing occurred towards or away from the greatest decidual blood sinus area. The EPC angle and the ratio of right:left decidua blood sinus area were compared in C57Bl/6J and Mtrr gt/gt conceptuses at E8.5 (N 6-12 conceptuses/group). Aligned (grey or black dots) and skewed (blue dots) conceptuses are shown. Dashed red line, right:left blood sinus area ratio of 1. Dashed grey line, line of best fit. See also Supplementary  Figures 2A Like the controls, decidual blood sinus area was similar on the right and left sides of Mtrr gt/gt implantation sites, and was independent of whether conceptus skewing occurred (p 0.605; Figure 5F). For each implantation site, the angle and direction of conceptus skewing was then compared to right-left decidual blood sinus ratio. Using histological sections of individual implantation sites, the chorion or EPC angle was classified as positive if it was skewed to the right and negative if it was skewed to the left. This value was plotted against the ratio of right:left decidual blood sinus area for that site ( Figure 5G and Supplementary Figure S2B). In this model, a line-of-best-fit sloping downwards towards the right would indicate that the chorion/EPC angled away from the greater blood sinus area, and a slope upwards towards the right would indicate that the chorion/EPC angled towards the greater blood sinus area. In C57Bl/6J controls, chorion and EPC skewing was absent regardless of whether the blood sinus area was greater on the right or left ( Figure 5G and Supplementary Figure S2B). Similarly in Mtrr gt/gt conceptuses, there was no statistically significant relationship between blood sinus size and the direction of chorion or EPC growth (p > 0.37; Figure 5G and Supplementary Figure S2B). These data suggested that the volume of maternal blood perfused into the implantation site was unlikely related to the orientation of Mtrr gt/gt conceptuses at E8.5.

Gross Decidual Morphology Is Unchanged by Maternal Mtrr gt Allele
Other studies showed that skewed conceptus orientation might result from abnormal decidual remodeling and signaling (Alexander et al., 1996;Zhang et al., 2014). Therefore, we assessed the effects of the maternal Mtrr gt mutation on decidualization. First, to understand the degree of Mtrr knockdown caused by the Mtrr gt allele in the decidua, wildtype Mtrr transcript levels were determined via RT-qPCR in whole implantation sites at E6.5 as a proxy for decidual cells (N 5-7 whole implantation sites/group). Implantation sites at E6.5 consist largely of decidualized stroma with a relatively low contribution by fetal-derived cells. This experiment was important because the extent of wildtype Mtrr gene knockdown in Mtrr +/gt and Mtrr gt/gt cells is tissue specific (Elmore et al., 2007;Padmanabhan et al., 2013;Sowton et al., 2020). As expected, implantation sites from Mtrr +/+ mothers displayed a similar level of wildtype Mtrr transcripts as C57Bl/6J controls (98.0% of control levels; Figure 6D). Furthermore, this level was reduced to 62.0 and 6.0% of controls in implantation sites from Mtrr +/gt and Mtrr gt/gt mothers, respectively (p < 0.0001; Figure 6D), and was within the range observed in other Mtrr +/gt and Mtrr gt/gt tissues (Elmore et al., 2007;Padmanabhan et al., 2013;Sowton et al., 2020). Total Mtrr transcript levels (includes Mtrr + and Mtrr gt RNA) were similar across all maternal genotypes (p 0.4941; Figure 6D) indicating that transcriptional compensation for wildtype Mtrr deficiency did not occur.
Next, gross decidual morphology was assessed in histological sections of implantation sites at E6.5. The extent of decidualization was unaffected by a maternal Mtrr gt allele at E6.5 since the average implantation site weight did not differ between control and Mtrr +/+ , Mtrr +/gt , or Mtrr gt/gt mothers ( Figure 6E; p 0.114). Furthermore, implantation site weight was similar between aligned and skewed conceptuses regardless of genotype ( Figure 6E; p 0.393), and there was no apparent Frontiers in Cell and Developmental Biology | www.frontiersin.org December 2021 | Volume 9 | Article 723978 FIGURE 7 | Maternal and grandparental Mtrr gt allele might affect molecular signaling in mouse decidua at E6.5. (A) RT-qPCR analysis of decidualization gene markers in whole implantation sites at E6.5 derived from C57Bl/6J, Mtrr +/+ , Mtrr +/gt or Mtrr gt/gt mothers (N 5-7 sites/group). Data presented as mean ± s.d., relative to C57Bl/6J controls (normalized to 1). One-way ANOVA with Tukey's multiple comparison test, *p < 0.05, **p < 0.01, ***p < 0.001. (B) Progesterone receptor (PGR) immunostaining in implantation sites at E6.5 derived from C57Bl/6J mothers and aligned and skewed conceptuses from Mtrr gt/gt mothers. N 3 sites/genotype were assessed. (C) COX2 immunostaining (brown) in implantation sites at E6.5 derived from C57Bl/6J, Mtrr +/+ , Mtrr +/gt , or Mtrr gt/gt mothers and C57Bl/6J fathers. N 2-5 sites/maternal genotype were assessed. DNA, blue. Both aligned and skewed conceptuses are shown. In panels (B,C), the degree of skewing is indicated by green dotted line that bisects the conceptus compared to black dotted line that bisects the decidual swelling. (D) No primary antibody control in a skewed Mtrr gt/gt implantation (Continued ) correlation between implantation site weight and degree of conceptus skewing (Supplementary Figures S1A,B). No litter size effect on implantation site weight at E6.5 was observed (Supplementary Figure S1C). Typically, an implantation site is ellipsoid with the mesometrial-antimesometrial axis (length) showing a longer dimension than the lateral axis (width). A rounder implantation site might indicate defective decidualization (Alexander et al., 1996). Centrally located transverse sections of implantation sites from C57Bl/6J mothers and from Mtrr +/+ , Mtrr +/gt , or Mtrr gt/gt mothers with aligned or skewed conceptuses (N 3-8 implantation sites/ group) were measured along both axes. No significant differences in decidua length:width ratios were observed in implantation sites from any maternal genotypic group compared to controls regardless of skewing (p 0.137; Figure 6F). Furthermore, there was no association between degree of conceptus skewing and implantation site dimensions (p > 0.39; Figure 6G). When lateral decidual cell density was quantified by nuclear counts in implantation sites containing skewed conceptuses derived from Mtrr +/+ , Mtrr +/gt , or Mtrr gt/gt mothers (N 4-9 implantation sites/group), there was no significant difference compared to C57Bl/6J conceptuses (p 0.170; Supplementary Figures 2C,D). Altogether, gross decidual morphology was unaffected by a maternal Mtrr gt allele.

Signaling Pathways in Decidua Are Potentially Disrupted by the Mtrr gt Mutation
Next, we explored the effects of maternal Mtrr gt allele on the molecular regulation of decidualization. Molecular markers of decidualization were assessed by RT-qPCR in whole implantation sites at E6.5 from C57Bl/6J, Mtrr +/+ , Mtrr +/gt , and Mtrr gt/gt mothers (N 5-7 sites/group). Progesterone receptor (PGR) (Lydon et al., 1995) and its downstream effectors NR2F2 (Wetendorf and DeMayo, 2012) and BMP2 (Lee et al., 2007) are indispensable for decidualization. While Pgr and Nr2f2 mRNA expression was significantly reduced in the implantation sites from Mtrr +/gt and Mtrr gt/gt mothers compared to C57Bl/6J and Mtrr +/+ mothers (p < 0.0012; Figure 7A), transcript levels of downstream targets of PGR signalling including Hand2 and Hoxa10 (Lim et al., 1999;Okada et al., 2014) were within the normal range (p 0.4132 and p 0.0796, respectively; Figure 7A). The degree of conceptus skewing was unknown in the samples assessed since each implantation site was snap frozen immediately after dissection, whereas analysis of conceptus alignment at this stage was determined in histological sections. However, immunohistochemistry suggested subtly lower PGR protein levels in both Mtrr gt/gt aligned and skewed conceptuses compared to C57Bl/6J controls ( Figure 7B) though this decrease in protein expression might be insufficient to alter downstream effectors. Future studies should include protein quantification of PGR and downstream targets. Notably, Bmp2 transcripts were significantly decreased in implantation sites from all three Mtrr maternal genotypes relative to controls (p 0.0001; Figure 7A). Low Bmp2 transcript levels in Mtrr +/+ conceptuses derived from Mtrr +/+ mothers that associated with normal Pgr transcripts ( Figure 7A) and normal decidualization (Supplementary Figures 2C,D) suggested that potential BMP2 signalling defects might occur independent of progesterone signalling and decidualization defects. Further experiments are required to elucidate the specific role of BMP2 signalling in the establishment of conceptus orientation.

Decidual COX2 Expression Is Normal in Implantation Sites With Skewed Orientation
When the gene encoding for COX2 protein is knocked out, defects in implantation and decidualization result (Lim et al., 1999). COX2 is important for prostaglandin synthesis and is initially expressed in the decidua adjacent to the implanting embryo at E4.5 followed by restricted expression in the mesometrial decidua by E5.5 and at E7.5 (Chakraborty et al., 1996). Aberrant COX2 expression was also observed in other models of conceptus misorientation (Daikoku et al., 2011;Cha et al., 2014;Zhang et al., 2014). Therefore, we investigated COX2 protein expression in the decidua at E6.5 from Mtrr +/+ , Mtrr +/gt , and Mtrr gt/gt mothers (mated to C57Bl/6J males), and at E8.5 from Mtrr gt/gt mothers (mated to Mtrr gt/gt males), particularly in association with aligned and skewed conceptus orientation. Within centrally located transverse histological sections, COX2 immunostaining in C57Bl/6J controls was restricted to the mesometrial decidua above the tip of the EPC, in line with the antimesometrial-mesometrial axis ( Figures 7C,E). A similar pattern was observed in implantation sites from Mtrr +/+ , Mtrr +/gt and Mtrr gt/gt mothers, regardless of the extent of conceptus skewing ( Figure 7C). Strikingly, the most skewed Mtrr gt/gt conceptus at E8.5 showed normal mesometrial decidual patterning of COX2 even though the EPC was more laterally located ( Figure 7E). These data suggested that general decidual patterning was established and that COX2 expression did not dictate conceptus orientation in the Mtrr gt model.
Frontiers in Cell and Developmental Biology | www.frontiersin.org December 2021 | Volume 9 | Article 723978 Mtrr gt mouse line is known for its multigenerational effects on development specifically through the maternal grandparental lineage (Padmanabhan et al., 2013;Blake et al., 2021). To this end, we reanalyzed our published phenotype data set at E10.5 whereby several genetic pedigrees were established to rigorously assess the phenotypic effects of a maternal grandparental Mtrr gt allele on the subsequent wildtype generations (Padmanabhan et al., 2013;Padmanabhan et al., 2017). Additional litters were included where possible. The frequency of the phenotype previously referred to as "eccentric chorioallantoic attachment" (Padmanabhan et al., 2013) and re-characterised in this study as severe conceptus skewing was used as an indicator of conceptus misalignment. While assessing severe misalignment alone does not give a full picture of the extent of skewing in these pedigrees, the number of conceptuses assessed (N 86-308 conceptuses/pedigree) allowed for statistical analysis. The breeding scheme was as follows (Figure 8): An F0 Mtrr +/gt female or male mouse was crossed with a control C57Bl/6J mate. Their wildtype adult daughters (i.e., the F1 generation) were selected for mating with a C57Bl/6J male and the wildtype grandprogeny (i.e., the F2 generation) were dissected and assessed at E10.5. Continuing a similar mating scheme allowed us to also assess the wildtype F3 and F4 generations. C57Bl/6J crosses were analyzed as a control.
Overall, a low frequency of severe conceptus skewing was observed in the F2 Mtrr +/+ progeny of the Mtrr +/gt maternal grandmother (MGM) pedigree (5.1% of conceptuses, p < 0.01; Figure 8C) and the Mtrr +/gt maternal grandfather (MGF) pedigree (2.0% of conceptuses, p < 0.05; Figure 8H). This phenotype was not observed in the controls ( Figure 8A). Notably, when the Mtrr genotype of the F1 female was altered (e.g., F1 Mtrr +/+ or F1 Mtrr +/gt females), there was no impact on the frequency of severe skewing in the F2 Mtrr +/+ progeny in either pedigree (MGM: Figure 8C versus Figure 8D; MGF: Figure 8H versus Figure 8I). This result indicated that the Mtrr gt genotype in the maternal grandparent was most important for conceptus misalignment, an observation consistent with other phenotypes (Padmanabhan et al., 2013).
There was a negligible paternal effect of the Mtrr gt allele on conceptus misalignment since this phenotype was rare in the direct offspring of an Mtrr +/gt male (0.7% of conceptuses, p 0.183 compared to controls; Figures 8A,G). Since Mtrr +/gt males were derived from Mtrr +/gt parents ( Figure 8G), this result also emphasized that a paternal grandparental effect was unlikely. Importantly, an Mtrr gt allele in a male was sufficient to cause conceptus misalignment in the F2 Mtrr +/+ grandprogeny via his F1 Mtrr +/+ daughters ( Figure 8H), and highlighted the potential importance of the F1 uterine environment (see below). Since Mtrr gt/gt conceptuses at E10.5 derived from Mtrr gt/gt intercrosses showed a similar frequency of skewing (5.8% of conceptuses, p < 0.001) as F2 Mtrr +/+ conceptuses from the MGM pedigree ( Figure 8B versus Figure 8C), the mechanism was likely due to Mtrr deficiency in one (or both) maternal grandparent. Notably, the frequency of skewing was greater in F2 Mtrr +/+ progeny when derived from an Mtrr +/gt MGM compared to an Mtrr +/gt MGF ( Figure 8C versus Figure 8H) suggesting that the causative mechanism might be different when initiated via the F0 oocyte versus the F0 sperm, as was previously hypothesized . In further support of this hypothesis, we observed that severe conceptus misalignment occurred until at least until the F4 wildtype generation when derived from an F0 Mtrr +/gt female ( Figures 8E,F) yet only until the F3 wildtype generation when derived from an F0 Mtrr +/gt male ( Figures  8J,K). Phenotypic inheritance to the F3 generation and beyond indicates transgenerational epigenetic inheritance as a mechanism for conceptus misalignment in both pedigrees (Padmanabhan et al., 2013;Blake and Watson, 2016). Of note, the resolution of a transgenerationally-inherited phenotype (e.g., conceptus misalignment in the MGF pedigree) has not been previously reported in the Mtrr gt model. Overall, the effects of abnormal folate metabolism caused by the Mtrr gt mutation persisted for several wildtype generations with detrimental effects on conceptus alignment.

Severe Conceptus Misalignment in the Mtrr gt Model Might Be Independent of the Uterus
Lastly, to determine whether the uterine environment influenced conceptus alignment, a published embryo transfer experiment was reanalyzed (Padmanabhan et al., 2013). In this case, preimplantation F2 Mtrr +/+ embryos were collected at E3.25 from the oviducts and uteri of F1 Mtrr +/+ females (MGM or MGF pedigree) and transferred into the normal uteri of B6D2F1 pseudopregnant females ( Figures 8M,N). C57Bl/6J embryos were transferred as a control ( Figure 8L). The transferred embryos were allowed to implant and conceptuses were dissected at E10.5 for phenotypic analysis (Padmanabhan et al., 2013). Our hypothesis was if severe conceptus misalignment was absent at E10.5 after embryo transfer, then an abnormal uterine environment in the original F1 Mtrr +/+ donor female was the cause of this phenotype. However, if severe misalignment was observed in the transferred conceptuses, then the defect occurred independent of the peri-or post-implantation uterine environment. We reported that some of the developmental phenotypes observed in F2 Mtrr +/+ conceptuses persisted after embryo transfer including congenital malformations (mechanism independent of F1 uterine environment) but not fetal growth restriction or developmental delay, which were rescued (mechanism dependent on F1 uterine environment) (Padmanabhan et al., 2013).
Here, our analysis revealed that severe conceptus misalignment persisted after embryo transfer suggesting that this phenotype occurred independent of the F1 Mtrr +/+ uterine environment ( Figures 8L-N) (Padmanabhan et al., 2013). The percentage of transferred F2 Mtrr +/+ conceptuses that were affected was similar to the non-transferred F2 Mtrr +/+ conceptuses ( Figures 8M,N versus Figures 8C,H). While the phenotypic frequency did not reach statistical significance, this was likely due to the low number of transferred embryos analyzed (N 36-60 embryos from six to nine litters; Figures 8M,N). None of the transferred C57Bl/6J control conceptuses exhibited severe skewing ( Figure 8L) indicating that the embryo transfer protocol itself was not responsible for this phenotype. Overall, we propose that conceptus misalignment in the Mtrr gt model occurred independent of a peri-and/or post-implantation uterine environment effect. Alternatively, a defect intrinsic to the embryo is likely.

DISCUSSION
We have shown here that abnormal folate metabolism due to the Mtrr gt mutation in mice influences conceptus orientation and spacing within the uterus. These phenotypes have implications for subsequent embryo and placenta development. Our data highlight the complexity of phenotype establishment associated with abnormal folate metabolism, with the mechanism acting beyond a direct maternal effect. We propose that conceptus misalignment and twinning are caused by an embryo-specific defect rather than a defect in the peri-/post-implantation uterine environment. We also implicate the Mtrr gt genotype of either maternal grandparent as the initial effector of conceptus misalignment, which persists at least until the wildtype great grandprogeny generation and exemplifies a transgenerationally inherited phenotype. In general, the mechanism of transgenerational epigenetic inheritance is not well understood and will take additional studies to elucidate. Given the evidence presented here, conceptus misalignment can negatively impact embryo and placenta development. The degree of skewing occurred over a continuum, though whether minor skewing is harmful to development is currently unclear as we were only able to assess the consequences of severe skewing. Even though normal trophoblast lineage patterning (Simmons et al., 2008) occurred in severely skewed Mtrr gt/gt conceptuses at E8.5, an improperly oriented placenta would likely have insufficient access to maternal blood for normal feto-placental development as supported by the mislocalisation of spiral arteryremodeling Prl7b1 + trophoblast cells. Severe conceptus skewing at E10.5 was also associated with embryonic developmental delay and heart malformations. Others have proposed the placentaheart axis whereby that primary placental phenotype can result in secondary cardiovascular abnormalities (Barak et al., 1999;Hemberger and Cross, 2001;Perez-Garcia et al., 2018), and vice versa (Outhwaite et al., 2019). Whether this phenomenon occurs in skewed conceptuses is unclear and requires more careful consideration of heart development in the Mtrr gt model from earlier stages and investigation by conditional mutation or tetraploid aggregation experiments that isolate the consequences of heart-only or placenta-only knockdown of Mtrr. Notably, neural tube defects, which are associated with maternal folate deficiency in humans (MRC Vitamin Study Research Group, 1991) and occur in the Mtrr gt/gt embryos at E10.5 with incomplete penetrance (Padmanabhan et al., 2013), did not correlate with conceptus misalignment in our study suggesting a different causative mechanism.
Dizygotic twinning is a rare event in mice (De Clercq et al., 2019) with few reports of this phenomenon in the literature.
While the cause is currently unknown, we hypothesize that dizygotic twinning is independent of the number of ovulations and results from poor blastocyst spacing prior to implantation. Thereafter, twinned embryos with overlapping development share one decidual swelling and display placental fusion as gestation progresses, similar to observations in other studies (De Clercq et al., 2019;Lopez-Tello and Sferruzzi-Perri, 2021). Blastocyst spacing in the mouse uterus is not well understood, yet it is a highly regulated event requiring mechanical processes in the uterus (e.g., muscular contractions and ciliary movement) and molecular signaling (e.g., LIF, ephrin A) between the uterus and blastocyst (Chen et al., 2013;Shi et al., 2014;Flores et al., 2020;Fujiwara et al., 2020). Manipulation of one uterine parameter typically leads to irregular spacing and clustering of an entire litter (Lu et al., 2013;Shi et al., 2014;Flores et al., 2020). For example, excessive intrauterine fluid results in crowding of implantation sites likely due to the disruption of physical and chemical interactions between the uterine epithelium and blastocysts (Lu et al., 2013). However, twinning is infrequently observed in this model (Lu et al., 2013). Since the frequency of twinning within and between Mtrr gt/gt litters was low and the general spacing of Mtrr gt/gt litters was otherwise normal, we propose an embryo-specific defect rather than a uterinespecific defect. Embryo-specific factors that affect uterine spacing are largely unknown (Chen et al., 2013). It is hypothesized that signals from a blastocyst (e.g., estrogen, prostaglandin, ephrin A) might initiate or maintain uterine events/molecular profiles required for optimal blastocyst spacing and implantation (Chen et al., 2013;Fujiwara et al., 2020). However, further experiments are required to better understand whether inequitable disruption of signaling among embryos of one litter can lead to poor spacing at a localized level (e.g., twinning). Additionally, it is unclear whether blastocysts secrete/receive inhibitory signals to/from neighboring blastocysts to maintain optimal uterine distance for fetal growth. Notably, other models of conceptus misalignment display extensive embryo spacing defects (Daikoku et al., 2011;Cha et al., 2014), yet it is unclear whether disrupted blastocyst signaling and/or poor responsiveness of the blastocyst to maternal signals is a common mechanism between twinning and orientation phenotypes.
The establishment of conceptus orientation is highly prescribed and depends on mesometrial-anti-mesometrial orientation in the uterus, ICM orientation within the blastocyst, and directional growth of the conceptus within the decidua (Chen et al., 2013). While the cellular and molecular events involved in embryo positioning largely remain unknown (Chen et al., 2013), factors intrinsic to the uterus are best studied. For instance, aberrant formation of uterine crypts misorient the blastocyst at implantation (Daikoku et al., 2011;Cha et al., 2014;Zhang et al., 2014), decidualization defects associate with misalignment of conceptuses post-implantation (Alexander et al., 1996;Zhang et al., 2014), and disorganization of the decidual vasculature network and increased blood perfusion misdirects trophoblast invasion resulting in skewed growth (Winterhager et al., 2013). The establishment of signaling gradients in decidua (e.g., WNT5) (Daikoku et al., 2011;Cha Frontiers Goad et al., 2017) and the responsiveness of the conceptus to these signals are also important for conceptus alignment. We observed low Bmp2 mRNA expression in implantation sites of litters known to display conceptus skewing. However, a causative relationship was not established and it is unclear whether low Bmp2 transcripts translate into low BMP2 protein levels or signaling activity. Further experiments are required to better understand the significance of this finding in the context of conceptus misalignment. While the uterine-specific mechanisms indicated above are yet-to-be explored in detail in the Mtrr gt model, evidence presented here argues against a uterine defect as the main cause of conceptus misalignment. For instance, normal morphology and patterning of Mtrr gt/gt decidua was observed, and skewed F2 Mtrr +/+ conceptuses were present after embryo transfer into a control uterus. As performed, the embryo transfer experiment ruled out an effect of the peri-and/or postimplantation stage uterus on conceptus alignment. However, based on the timing of the transfer into the recipient uterus, the F2 Mtrr +/+ embryos might be exposed to abnormal factors in the oviductal/uterine fluid of the donor F1 Mtrr +/+ females that might cause incorrect orientation at and after implantation. These factors are currently unknown but might be hormonal (Chen et al., 2013) and the effects would need to be long lasting and/or irreversible. To more definitively eliminate a uterine effect as a causative factor in embryo misalignment, additional embryo transfer experiments should be performed: 1) The transfer of C57Bl/6J embryos (at E3.25) into an F1 Mtrr +/+ uterus or Mtrr gt/gt uterus. Normal conceptus alignment would indicate no involvement of the peri-and/or post-implantation uterine environment.
2) The transfer of F2 Mtrr +/+ embryos at an earlier developmental stage (e.g., two-cell stage) or after in vitro fertilization into a control uterus. In this case, the effects of the oviductal fluid on conceptus alignment would be taken into account.
As an alternative to a defective uterine environment, a mechanistic defect intrinsic to the embryo might cause conceptus misalignment in the Mtrr gt model. Under normal circumstances, mouse blastocysts attach to the uterine epithelium via adhesion molecules (e.g., integrins, L-selectins) expressed on the mural trophectoderm (Cross et al., 1994;Red-Horse et al., 2004) so that the embryonicabembryonic axis is parallel to the mesometrialantimesometrial axis (Smith, 1985). How the blastocyst attaches to the uterine epithelium to establish a correct orientation is not well understood. One hypothesis is that the blastocyst might attach in any orientation and then roll along the uterine epithelium into the correct orientation (Kirby et al., 1967;Red-Horse et al., 2004). Otherwise, it was proposed that ICM cells might mobilise within the blastocyst to properly align in the implantation site after trophectoderm adhesion to the uterine epithelium (Kirby et al., 1967;Meilhac et al., 2009;Morris et al., 2010;Chen et al., 2013). Therefore, critical adhesion molecules on the outer membrane of the mural trophectoderm, the inner membrane of the trophectoderm, or the ICM of Mtrr blastocysts should be examined for regional expansion or misexpression as a cause for skewed growth. The occurrence of normal litter sizes from Mtrr gt/gt intercrosses at midgestation (this study; Padmanabhan et al., 2013) suggests that implantation of Mtrr gt/gt blastocysts is achieved regardless of this hypothetical adhesion defect. Alternatively, normal blastocyst orientation at implantation might be followed by unresponsiveness of some Mtrr conceptuses to cytokine signals from the decidua (e.g., WNTs, FGFs, BMPs) that guide mesometrial-antimesometrial alignment (Red-Horse et al., 2004;Daikoku et al., 2011;Cha et al., 2014;Goad et al., 2017). Further studies are required to elucidate the specific embryocentric mechanisms and molecular pathways involved in conceptus orientation, and how they are influenced by abnormal folate metabolism. Since MTRR enzyme is required for the transmission of methyl groups for methylation reactions in the cell (Shane and Stokstad, 1985), it is possible that epigenetic regulation of crucial genes required for conceptus orientation is disrupted in skewed Mtrr gt/gt conceptuses. Candidate genomic regions will be revealed by future whole methylome and transcriptome analyses in early embryos and uterine tissue.
It is unclear whether the phenotypes that underlie genetic disruption of abnormal folate metabolism (e.g., the Mtrr gt allele) are similar to maternal dietary folate deficiency, and what mechanistic parallels can be drawn between these two models. Although Mtrr gt/gt mice recapitulate developmental phenotypes (e.g., neural tube defects) and anemia observed in humans with folate deficiency (Padmanabhan et al., 2013;Ducker and Rabinowitz, 2017;Padmanabhan et al., 2018), diet-induced folate-deficient mice do not unless they are challenged with genetic insufficiency or viral infection (Heid et al., 1992;Koury and Horne, 1994;Burgoon et al., 2002;Burren et al., 2008). Uterine structure at implantation and the frequency of blastocyst implantation are likely unaffected by pre-conceptual dietary folate deficiency in female mice (Gao et al., 2012), though post-implantation decidualization and deciduoma formation might be inhibited (Geng et al., 2015). Decidualization was normal in Mtrr gt/gt mice at E6.5 (this study). Conceptus misalignment and twinning are yet-to-be reported in studies of folate deficiency or in other genetic mouse models where folate uptake or metabolism is disrupted leading to embryonic defects or lethality in early to mid-gestation (Piedrahita et al., 1999;Swanson et al., 2001;Gelineau-van Waes et al., 2008;Pickell et al., 2009;Salojin et al., 2011). Therefore, it is currently unclear whether these phenotypes are specific to the Mtrr gt model or a more general disruption of folate uptake/metabolism. Additional experiments should be performed to determine whether dietary supplementation of Mtrr gt/gt mice prevents the developmental phenotypes and/or conceptus misalignment observed.
Importantly, severe conceptus skewing at E10.5 in the Mtrr gt model exhibits a pattern of transgenerational inheritance. This nonconventional mode of inheritance is typically caused by exposure to an environmental stressor or metabolic disruption only in the initiating F0 generation and occurs independent of DNA basesequence mutations (Blake and Watson, 2016). The Mtrr gt mouse line is a known model of transgenerational epigenetic inheritance as several embryonic phenotypes are inherited by the wildtype Frontiers in Cell and Developmental Biology | www.frontiersin.org December 2021 | Volume 9 | Article 723978 grandprogeny and great grandprogeny when either maternal grandparent is a carrier for the Mtrr gt mutation (Padmanabhan et al., 2013;Blake et al., 2021). While the mechanism of epigenetic inheritance of phenotypes is unknown in the Mtrr gt model and other mammalian multigenerational models, it is particularly difficult to explore when the phenomenon occurs via the maternal lineage. This is because defects in the maternal physiology and uterine environment are difficult to tease apart from abnormal germ cell factors (Watson and Rakoczy, 2016). For this reason, we utilized embryo transfer experiments to eliminate potential mechanistic effects of the peri/post-implantation uterine environment. While an epigenetic mechanism is the most likely cause of this transgenerational effect, folate metabolism is required for cellular methylation and thymidine synthesis (Ducker and Rabinowitz, 2017). Therefore, disruption of folate metabolism potentially implicates both epigenetic and genetic instability. Notably, thymidine synthesis is likely unaffected by the Mtrr gt mutation since de novo mutations, which might transpire from low thymidine levels resulting in uracil misincorporation into DNA (Blount et al., 1997), occur at a similar frequency in C57Bl/6J control and Mtrr gt/gt mice . Therefore, we conclude that the wide range of phenotypes observed over multiple generations in the Mtrr gt model is not the result of genetic instability. Alternatively, considerable alteration in patterns of DNA methylation was identified in adult germ cells and somatic tissues of the Mtrr gt model, with some methylation changes associated with gene misexpression (Padmanabhan et al., 2013;Bertozzi et al., 2021;Blake et al., 2021). The range of developmental phenotypes observed in the Mtrr gt model (this study, Padmanabhan et al., 2013) suggests that this epigenetic instability occurs stochastically in each affected individual, ultimately leading to disruption of different genetic pathways and different phenotypes (Blake and Watson, 2016). This proposed stochasticity of inter-individual epigenetic changes might explain incomplete penetrance of the conceptus misorientation (and other phenotypes).
How abnormal epigenetic patterns are inherited is not well understood since they must circumvent reprogramming that occurs between generations or be reconstructed in the next generation (Jablonka, 2013;Blake and Watson, 2016). Altered patterns of DNA methylation and histone modifications, or misexpression of small non-coding RNA in germ cells are proposed candidates of epigenetic inheritance (Chen et al., 2016;Sharma et al., 2016;Blake et al., 2021;Lismer et al., 2021) and are currently being explored in the Mtrr gt model and other models of transgenerational epigenetic inheritance. While there is no evidence of abnormal mitochondrial function in Mtrr gt/gt livers (Sowton et al., 2020) or of small nucleotide polymorphisms in genes important for mitochondrial function in Mtrr gt/gt embryos with congenital malformations , oocyte mitochondria from F0 Mtrr +/gt or F1 Mtrr +/+ females are yet-to-be assessed for defects. Notably, we discerned a greater frequency and later resolution of conceptus misalignment in wildtype conceptuses derived from an Mtrr +/gt maternal grandmother versus an Mtrr +/gt maternal grandfather. This observation might hold clues for future mechanistic exploration including the identification of different initiating factors in an F0 oocyte versus an F0 spermatozoon.
The association between folate metabolism and conceptus alignment has implications for early pregnancy loss associated with folate deficiency in humans, and for culture conditions involved in assisted reproductive technologies. Our study reinforces the importance of folate in healthy pregnancies, yet emphasizes that the role of folate metabolism during development is more complex than a direct uterine effect. Preconceptual supplementation of folic acid should be considered. Since conceptus skewing occurred several wildtype generations after the initial metabolic disruption, it is possible that unexplained pregnancy loss in humans might be associated with folate deficiency in the grandparental generation. This conclusion emphasizes the importance of folate fortification programs, which might take more than one generation to show the full effects.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.

ETHICS STATEMENT
The animal study was reviewed and approved by the UK Home Office under the Animals (Scientific Procedures) Act 1986 Amendment Regulations 2012 and underwent review by the University of Cambridge Animal Welfare and Ethical Review Body.

AUTHOR CONTRIBUTIONS
EW: conceptualization, experimental design, and funding acquisition. KM, JR, and EW: dissections, phenotyping, and histology preparation. KM: RNA isolation and RT-qPCR. AW, KM, JR, and XT: histological staining, immunostaining and in situ hybridization. AW, KM, and EW: data analysis. AW and EW: data interpretation and manuscript writing. AW, EW, and KM: manuscript editing. All authors contributed to the article and approved the submitted version.

FUNDING
This study was funded by the Lister Research Prize from the Lister Institute for Preventative Medicine (to EW). KM was supported by a Newnham College (Cambridge) studentship and an A.G. Leventis award. JR was supported by a Newton International Fellowship.

ACKNOWLEDGMENTS
We thank Simon Tunster for technical advice, and Claire Senner for critical discussion of the manuscript.