The cumulative effect of assisted reproduction procedures on placental development and epigenetic perturbations in a mouse model

Abstract

Assisted reproductive technologies (ART) are associated with several complications including low birth weight, abnormal placentation and increased risk for rare imprinting disorders. Indeed, experimental studies demonstrate ART procedures independent of existing infertility induce epigenetic perturbations in the embryo and extraembryonic tissues. To test the hypothesis that these epigenetic perturbations persist and result in adverse outcomes at term, we assessed placental morphology and methylation profiles in E18.5 mouse concepti generated by in vitro fertilization (IVF) in two different genetic backgrounds. We also examined embryo transfer (ET) and superovulation procedures to ascertain if they contribute to developmental and epigenetic effects. Increased placental weight and reduced fetal-to-placental weight ratio were observed in all ART groups when compared with naturally conceived controls, demonstrating that non-surgical embryo transfer alone can impact placental development. Furthermore, superovulation further induced overgrowth of the placental junctional zone. Embryo transfer and superovulation defects were limited to these morphological changes, as we did not observe any differences in epigenetic profiles. IVF placentae, however, displayed hypomethylation of imprinting control regions of select imprinted genes and a global reduction in DNA methylation levels. Although we did not detect significant differences in DNA methylation in fetal brain or liver samples, rare IVF concepti displayed very low methylation and abnormal gene expression from the normally repressed allele. Our findings suggest that individual ART procedures cumulatively increase placental morphological abnormalities and epigenetic perturbations, potentially causing adverse neonatal and long-term health outcomes in offspring.

Introduction

Before the advent of successful in vitro fertilization (IVF) in 1978, couples experiencing infertility had little recourse for producing biological children. Today, fertility treatments and Assisted Reproductive Technologies (ART), including IVF, have helped many couples overcome their infertility, and account for >1% of births in the United States (1), and >4% of births in some European countries (2). Unfortunately, ART pregnancies are associated with a number of maternal and fetal health risks including stillbirth, preterm birth, low birth weight, abnormal placentation and other pregnancy complications (3–8). ART-conceived offspring are also at increased risk for congenital abnormalities and rare imprinting disorders, specifically Beckwith-Wiedemann, Russell-Silver and Angelman syndromes (9–13). The interpretations of human data are limited, however, because it is difficult to separate the iatrogenic effects of ART procedures from the patients' underlying infertility diagnosis as well as differences in maternal age, body mass index and environmental exposures.

Animal models are invaluable for investigating the effects induced by ART procedures. Experimental studies in mice demonstrate that ART procedures can perturb embryo development, decrease fetal weight and increase placental weight (14–17). With respect to imprinted genes, numerous studies link ART procedures to aberrant expression of imprinted genes, with the most severe effects observed in extraembryonic tissues (18–22). ART procedures increase the expression of the imprinted genes H19, Igf2, Kcnq1ot1, Cdkn1c, Peg3 and Ascl2 around the time of placental formation (19–22). Imprinted genes, which are expressed in a parent-of-origin specific manner and are epigenetically regulated, have important roles in growth and development, including placentation (23). It is unclear, however, if the ART-induced changes in placental imprinted gene expression result in abnormal placental development or adverse fetal outcomes at term.

In this study, we determined the effect of IVF on fetal and placental epigenetic profiles at term (E18.5) using a mouse ART model and investigated its impact on placental development. Because the abnormal phenotypes induced by ART are not observed similarly in all previous studies, there is still confusion over which procedures contribute to the specific morphological and epigenetic abnormalities. For example, most studies that report effects due to embryo culture also employ superovulation in tandem. Any abnormalities specifically induced by superovulation or embryo transfer (ET) procedures have not been addressed as few studies have included naturally conceived concepti controls for comparison. Similar to the transvaginal ET procedure used for IVF patients, non-surgical embryo transfer (NSET) in mice, which we currently employ, is an effective technique that eliminates stress caused by standard surgical procedures on recipient female mice (24,25). Nevertheless, NSET can cause adverse effects—even clinically-relevant low ejection speeds (<0.1 m/s) of transfer increase apoptosis and decrease the total number of cells in preimplantation embryos (26,27). To date, the effect of ET on epigenetic profiles and placental development relative to naturally conceived controls is unknown. Accordingly, we assessed the effect of NSET with and without superovulation.

We report that ART procedures performed under optimal conditions rarely affect imprinted genes in the fetus, whereas, placental development, is severely impacted. Non-surgical embryo transfer with or without superovulation induces placentomegaly, while the addition of superovulation results in further abnormal morphological phenotypes. Furthermore, robust epigenetic perturbations are observed in term IVF placentae. Taken together, our findings implicate ART procedures in contributing to abnormal placentation, further increasing the risk for poor pregnancy outcomes.

Results

Fetal and placental weights of term concepti from two different mouse crosses

We generated term concepti using two mouse crosses, C7xB and CFxB (see Materials and Methods), to evaluate the effects of ART procedures on fetal and placental development in mice. In total, we collected 18 natural C7xB concepti from 4 litters and 20 IVF C7xB concepti from 10 litters (Supplementary Material, Table S1). Although the litter sizes varied between the groups, the naturally conceived controls allowed comparing phenotypic differences between concepti that were not exposed to any ex vivo manipulation associated with ART and concepti that were exposed to all procedures utilized in a standard IVF protocol. At E18.5, C7xB IVF fetuses were significantly smaller and the placentae were significantly larger compared with naturally conceived controls (Fig. 1A and B; Supplementary Material, Table S1). Moreover, the fetal to placental (F:P) weight ratio was significantly decreased in C7xB IVF concepti (Fig. 1C; Supplementary Material, Table S1).

Figure 1.

Fetal and placental weights of E18.5 C7xB and CFxB concepti. Naturally conceived and IVF-derived C7xB concepti were assessed for fetal weight (A), placental weight (B) and F:P ratios (C). Naturally conceived, ET, SET and IVF-derived CFxB concepti were assessed for fetal weight (D), placental weight (E) and F:P ratios (F). Each dot represents an individual fetus or placenta collected from a single conceptus. Statistical significance was determined using the mean. P-values with brackets are indicated where statistical differences were detected (P < 0.05).

In addition to naturally conceived and IVF-derived CFxB offspring, we also assessed the effects of different manipulations utilized during ART procedures by generating in vivo fertilized concepti that were only exposed to the ET procedure or superovulation and embryo transfer (SET). We collected 16 natural CFxB concepti from 3 litters, 16 ET CFxB concepti from 6 litters, 15 SET CFxB concepti from 7 litters and 19 IVF CFxB concepti from 11 litters (Supplementary Material, Table S1), and compared all groups to one another for all parameters assessed in this study. There was no significant difference in fetal weight for any of the CFxB fetuses (Fig. 1D; Supplementary Material, Table S1). In contrast, the CFxB ET, SET and IVF placentae were significantly larger (Fig. 1E; Supplementary Material, Table S1) compared with natural controls, and the IVF placentae were significantly larger compared with ET samples (Fig. 1E; Supplementary Material, Table S1). In addition, the F:P weight ratio of ET, SET and IVF placentae was significantly decreased compared with natural controls (Fig. 1F; Supplementary Material, Table S1). We also assessed whether fetal and placental weight alterations were specific to male or female concepti, but no sex-specific differences were detected in either genetic background (Supplementary Material, Table S2). Thus, whereas the relative fetal weight of IVF offspring was dependent on genetic background, the overgrowth phenotype of the IVF-derived placentae was conserved between different genetic backgrounds. These results also demonstrate that the use of ET alone can induce placentomegaly in term mouse concepti, and the combination of SET can increase the severity of the overgrowth phenotype.

Histological analysis of placental tissues

Because IVF-derived placental tissues were significantly larger in both mouse crosses, we sought to determine whether this overgrowth phenotype was associated with abnormal morphology. To this end, we assessed gross morphology of placental tissues by measuring the area of the junctional and labyrinth zones and calculating a junctional to labyrinth zone ratio using H&E-stained cross-sections. Notably, C7xB IVF placentae exhibited a significantly increased junctional to labyrinth zone ratio indicating abnormal morphology (Figs 2 and 3A). An expansion of the junctional zone was detected in C7xB IVF placentae by measuring the relative expression of the established junctional zone markers Tpbpa and Prl8a8 (Fig. 3B). Because the junctional zone contains several different cell types, we also wanted to assess whether specific cell types were undergoing preferential expansion, or if all cell types in the junctional zone were increased in placental tissues. We observed no significant difference in the expression of a glycogen trophoblast cell marker (Pcdh12), but increased expression of a marker for canal and spiral artery trophoblast giant cells (TGC; Prl2c2) was detected in IVF-derived placentae (Fig. 3B).

Figure 2.

Representative cross-sections of E18.5 C7xB and CFxB placentae. Placental cross-sections from natural, ET, SET and IVF groups were stained with H&E. Black lines outline the junctional zone. Scale bar represents 1000 µm.

Figure 3.

Quantification of histological cross-sections and expression analysis of junctional zone markers in C7xB and CFxB placental tissues. Stereologic analysis of the junctional zone area to labyrinth zone area ratio in C7xB (A) and CFxB (C) placentae. Expression of cell-type specific markers in the junctional zone for glycogen and spongiotrophoblast cells (Tpbpa), spongiotrophoblast cells (Prl8a8), glycogen cells (Pcdh12) as well as canal and spiral-associated TGCs (Prl2c2) using real-time PCR in C7xB (B) and CFxB (D) placentae. Values are displayed as mean ± SEM. a, P < 0.05 compared with natural controls using the variance ratio test. b, P < 0.05 compared with ET using the variance ratio test. Values are displayed as mean ± SEM. JZ, junctional zone. Lab, labyrinth zone.

In CFxB concepti, a significant increase in the junctional to labyrinth zone ratio was observed in the CFxB SET and IVF placentae (Figs 2 and 3C), whereas the CFxB ET placentae had morphology that was comparable to naturally conceived controls (Fig. 3C). Similar to C7xB IVF placentae, the CFxB SET and IVF placental tissues exhibited increased expression of junctional zone markers (Fig. 3D) as well as a marker for canal and spiral-associated TGCs (Fig. 3D). The CFxB ET placentae, however, only had a statistical difference in expression for one junctional zone marker (Tpbpa; Fig. 3D). Taken together, these results demonstrate that IVF placentae have altered placental morphology as well as increased expansion of spongiotrophoblast cells and canal and spiral-associated TGCs in the junctional zone. In addition, the data further support that the combination of SET exacerbates adverse placental phenotypes when compared with the use of ET alone.

Expression and DNA methylation profiles of glucose and amino acid transporters in placental tissues

A critical function of the placenta is to supply the growing fetus with a sufficient amount of nutrients to facilitate normal development throughout gestation. Expression levels of transporter genes in a placenta are sensitive to a variety of environmental perturbations (28), including IVF (16,29). Therefore, we analyzed expression of two glucose transporters, Glut1 and Glut3, and two system A amino acid transport family members, Snat2 and Snat4, in E18.5 placental tissues from both mouse crosses. For the C7xB IVF placentae, only Snat4 expression was significantly increased compared with natural controls (Fig. 4A). However, for the CFxB placentae expression of Glut3 was significantly different in both SET and IVF, and Snat4 expression was significantly different in IVF placentae compared with natural controls (Fig. 4B).

Figure 4.

Expression and DNA methylation profiles of glucose and amino acid transporters in C7xB and CFxB placentae. Relative expression levels of glucose transporters, Glut1 and Glut3, and amino acid transporters, Snat2 and Snat4, in C7xB (A) and CFxB (B) placental tissues. Bisulfite pyrosequencing was used to measure DNA methylation at the Glut3 promoter in C7xB (C) and CFxB (D) placental tissues. Bisulfite pyrosequencing was used to measure DNA methylation at the Snat4 DMR in C7xB (E) and CFxB (F) placental tissues. Each dot represents the average methylation from all of the CpG sites analyzed in an individual sample. Statistical significance was determined using variance for both expression and methylation analyses. P-values with brackets are indicated where statistical differences were detected (P < 0.05).

Because Glut3 (a non-imprinted gene) and Snat4 (an imprinted gene) can be epigenetically regulated by DNA methylation (30,31), we assayed DNA methylation at these loci using bisulfite pyrosequencing. DNA methylation for Glut3 was measured at a CpG island 1000 bp upstream of the transcriptional start site. Consistent with the lack of a significant expression difference, there was no change in Glut3 methylation for C7xB IVF placentae (Fig. 4C). In contrast, a significant increase of DNA methylation in CFxB IVF placentae (Fig. 4D) correlated with a reduced expression of Glut3 in these tissues (Fig. 4B). A significant reduction of DNA methylation at the Snat4 differentially methylated region (DMR) was observed in C7xB and CFxB IVF placentae (Fig. 4E and F), which corresponds to the increased level of expression of this locus in both mouse crosses (Fig. 4A and B). No changes in DNA methylation were detected in CFxB ET and SET placentae at any of the transporter genes (Fig. 4D and F). These results demonstrate that IVF can induce epigenetic changes at crucial transporter genes that result in aberrant expression profiles in placental tissues, but use of ET and superovulation does not promote such epigenetic changes at these loci. Notably, our data also show that ART can promote epigenetic changes at imprinted as well as non-imprinted genes indicating that epigenetic abnormalities observed in placenta are not specific to imprinted genes in IVF-derived mice.

DNA methylation analysis of fetal and placental tissues

We previously described a loss of imprinted gene expression and DNA methylation in mid-gestation placentae (22). Because it was unclear whether the aberrant imprinting patterns would persist or undergo correction by elimination of the cells exhibiting loss of imprinting we assayed imprinting in term placenta. We also analyzed DNA methylation in fetal brain and liver tissue because IVF offspring have exhibited behavioral and physiological abnormalities (32,33). DNA methylation was measured at imprinting control regions (ICRs) for one paternally methylated ICR (H19/Igf2), and three maternally methylated ICRs (Snrpn, Peg3 and Kcnq1ot1) using bisulfite pyrosequencing. A significant reduction in methylation was detected at the H19/Igf2 ICR compared with controls in C7xB brains (Fig. 5A), while DNA methylation levels were comparable to natural controls for Snrpn, Peg3 and Kcnq1ot1 ICRs (Fig. 5B–D). Similarly, a significant reduction in methylation was detected at the H19/Igf2 ICR in C7xB livers (Fig. 5E), while DNA methylation levels were comparable to natural controls for Snrpn, Peg3 and Kcnq1ot1 ICRs (Fig. 5F–H). Notably, we observed reduced DNA methylation at the H19/Igf2 ICR in the same three IVF-derived fetuses for both liver and brain tissues (Fig. 5A and E) indicating that the imprinting defect occurred early in development prior to lineage specification and was maintained throughout gestation. Although reduced methylation was only detected at the H19/Igf2 ICR in fetal tissues, placentae from IVF C7xB fetuses showed significant changes in methylation for all four ICRs compared with natural controls (Fig. 5I–L).

Figure 5.

DNA methylation profiles of brain, liver and placental tissues from C7xB concepti. Bisulfite pyrosequencing was used to measure DNA methylation in fetal brain at the H19/Igf2 ICR (A), Snrpn ICR (B), Peg3 ICR (C) and Kcnq1ot1 ICR (D). Bisulfite pyrosequencing was used to measure DNA methylation in fetal liver at the H19/Igf2 ICR (E), Snrpn ICR (F), Peg3 ICR (G) and Kcnq1ot1 ICR (H). Bisulfite pyrosequencing was used to measure DNA methylation in term placentae at the H19/Igf2 ICR (I), Snrpn ICR (J), Peg3 ICR (K) and Kcnq1ot1 ICR (L). Gray dots indicate the IVF-derived concepti that exhibited reduced methylation in liver, brain and placental tissues at the H19/Igf2 ICR. Crosshair = individual C7xB placenta with low methylation levels associated with biallelic expression in Figure 6. Statistical significance was determined using variance. P-values with brackets are indicated where statistical differences were detected (P < 0.05).

DNA methylation profiles were also measured at the same ICRs using CFxB fetal and placental tissues. Natural and IVF CFxB liver and brain tissues had normal methylation profiles for all ICRs analyzed (Supplementary Material, Fig. S1A–H). Placentae from IVF CFxB fetuses had significant changes in methylation at the H19/Igf2, Peg3 and Kcnq1ot1 ICRs (Supplementary Material, Fig. S1I, K and L), but normal methylation at the Snrpn ICR (Supplementary Material, Fig. S1J). Importantly, CFxB ET and SET placentae exhibited normal DNA methylation profiles at the H19/Igf2 and Peg3 ICRs (Supplementary Material, Fig. S2) indicating that use of ET and superovulation does not induce abnormal methylation at imprinted genes in placental tissues. Thus, the methylation abnormalities at imprinted genes in mid-gestation IVF placentae appear to be maintained throughout gestation, and are not detected in ET or SET samples suggesting that these manipulations do not contribute to altered DNA methylation profiles in placental tissues.

Because human studies have shown that a subset of IVF offspring exhibit abnormal methylation at multiple genes in placental tissues (Carmen Sapienza, personal communication), we performed an outlier analysis to determine whether IVF-derived murine placentae are affected in a similar manner. Our data showed that a subset of IVF placental tissues from both mouse crosses has irregular methylation at multiple genes (Supplementary Material, Fig. S3 and Table S3). Intriguingly, the placental tissue with highest number of methylation abnormalities in the C7xB cross also exhibited abnormal DNA methylation in fetal tissues suggesting that placental DNA methylation profiles may be informative for epigenetic programing in the fetal tissues (Supplementary Material, Fig. S3 and Table S3).

To assess whether the observed epigenetic defects were locus-specific or the result of impaired DNA methylation machinery, we analyzed global DNA methylation levels at repetitive elements using luminometric methylation assays (LUMA). We found that DNA methylation levels were unchanged in brain and liver tissues, but global DNA methylation levels were significantly decreased in C7xB and CFxB placentae (Supplementary Material, Fig. S4). These results indicate that systemic impairment of endogenous DNA methylation machinery contributes to the abnormal methylation observed in IVF placental tissues.

Allele-specific expression of imprinted genes in fetal and placental tissues

To determine whether abnormal DNA methylation levels at ICRs correlated with aberrant expression of imprinted genes, we used strain-specific polymorphisms in C7xB concepti to measure allele-specific expression of H19, Igf2, Cdkn1c, Peg3, Snrpn and Kcnq1ot1. Allele-specific expression of H19, Igf2 and Cdkn1c was measured in liver, and Peg3, Snrpn and Kcnq1ot1 were assayed in brain because these genes are highly expressed in these tissues. A significant increase in biallelic expression of H19 was observed in IVF-derived liver tissues compared with natural controls, and the three samples that exhibited reduced DNA methylation at the H19/Igf2 ICR in Figure 5 also displayed biallelic H19 expression (Fig. 6A). Conversely, monoallelic expression of Igf2 and Cdkn1c was maintained in liver tissues (Supplementary Material, Fig. S5A and B). Significant biallelic expression of Peg3, Kcnq1ot1 and Snrpn were also observed in IVF-derived brain tissues (Fig. 6B and C; Supplementary Material, Fig. S5C). Allele-specific expression was analyzed for all six imprinted genes in natural and IVF placental tissues. A significant increase in biallelic expression was detected for H19, Peg3 and Kcnq1ot1 in IVF placentae (Fig. 6D–F), and no differences in expression were detected for Igf2, Cdkn1c and Snrpn (Supplementary Material, Fig. S5D–F). Moreover, there was one C7xB IVF placenta with reduced methylation at all four ICRs (Fig. 5I–L, individual data point denoted by crosshair) as well as biallelic expression for H19, Peg3, Kcnq1ot1 and Snrpn (Fig. 6D–F and S5F). Collectively, these results suggest that ART can induce biallelic expression of imprinted genes in both fetal and placental tissues from fully developed concepti, but epigenetic defects occur at a much higher frequency in IVF-derived term placentae.

Figure 6.

Allele-specific expression levels of H19, Peg3 and Kcnq1ot1 in fetal and placental tissues from C7xB concepti. Allele-specific expression of H19 was assessed in liver (A) and placental (D) tissues. Allele-specific expression of Peg3 and Kcnq1ot1 was assessed in brain (B and C) and placental (E and F) tissues. Each dot represents the proportion of expression from the normally repressed allele. Gray dots indicate the samples that exhibited reduced methylation in Figure 5. Crosshair = individual C7xB placenta with high biallelic expression associated with low DNA methylation. Statistical significance was determined using variance. P-values with brackets are indicated where statistical differences were detected (P < 0.05).

Discussion

In this study, we addressed two unresolved questions concerning ART: Do in vitro manipulations result in epigenetic perturbations in term fetal and placental tissues and is placental development affected? Utilizing a mouse model, we simultaneously compared term concepti from (1) natural conception and gestation; (2) naturally conceived blastocysts that were transferred to pseudopregnant recipients (ET); (3) blastocysts conceived in vivo after superovulation that were transferred to pseudopregnant recipients (SET) and (4) IVF procedures, which include superovulation, IVF and embryo culture to the blastocyst stage prior to ET. Our findings demonstrate that even a minimal in vitro manipulation such as NSET can impact placental development. Importantly, as the number of manipulations increases, the morphological and molecular phenotype of the placenta becomes more severe (Fig. 7).

Figure 7.

Summary of ART procedures and their associated placental phenotypes. Fertilization and development to the blastocyst stage occurred in vivo for all groups except IVF. For IVF, fertilization and embryo culture to the blastocyst stage were performed under optimized media and oxygen conditions (23). For ET, blastocysts were non-surgically transferred to pseudopregnant recipient female mice.

The adverse effect of ART on fetal and placental outcomes has been the subject of previous studies (14–16,18–22,29,34–40). In contrast to our study, the great majority of these previous studies utilized a single control group that is comparable to our SET group, and only a few include natural or ET control groups (14,21,22,38). To our knowledge, this study is the first to evaluate the effect of NSET with and without superovulation on placental development and epigenetic profiles of both the placenta and its associated fetus. Results from the natural and ET groups strongly suggest that IVF outcomes are more severe than previously suspected.

Embryo transfer in the absence of hormonal manipulation induces placentomegaly, demonstrating that placental weight is particularly sensitive to in vitro manipulations. The brief handling of embryos required for transfer and/or the subtle differences in the receptive uterine environment of the pseudopregnant recipient is enough to significantly increase placental weight by 50% when compared with placentae following natural conception (Fig. 1 and Supplementary Material, Table S1). Interestingly, the size of both the junctional zone and labyrinth is increased, differing from the junctional zone overgrowth phenotype observed when superovulation is added (Fig. 3). This overgrowth is not the result of a change to a single cell type, but rather several junctional zone cell types (Fig. 3). It is clear from histological analyses that ART procedures significantly alter placental development, but the physiological impact of these changes remains to be elucidated. While structural changes can indicate dysfunction, they can also be the result of compensatory changes that do not significantly affect overall physiological function.

Transgenic models of several imprinted genes (Cdknc1, H19, Igf2r and Phlda2) exhibit similar junctional zone phenotypes (41–48), and it is tempting to suggest that all the morphological changes observed in the placenta are the direct result of alterations in the expression of these imprinted genes. However, our results provide at least three reasons why placental morphological abnormalities cannot simply be the result of abnormal imprinted gene expression. First, DNA methylation was normal for all the ICRs we assessed in ET and SET concepti, but placentomegaly and junctional zone overgrowth are still observed, strongly suggesting morphological changes occur independently of changes to imprinted genes. Second, our results along with previous work (18–20,22), support that IVF and subsequent embryo culture result in the hypomethylation of ICRs, which is associated with an aberrant expression from the repressed allele, rather than reduced expression that would be predicted to phenocopy the placental abnormalities observed in knockout models. Finally, epigenetic perturbations were also observed at a non-imprinted gene and reduced genome-wide methylation was detected by LUMA, suggesting that epigenetic gene regulation was more widely affected by procedures employed in IVF.

Embryo transfer with or without superovulation induces abnormal morphological changes in the absence of drastic epigenetic changes, but perturbations in ICR methylation and imprinted gene expression are observed in term IVF placentae. Prior to this study, it was known that superovulation, IVF and embryo culture cause hypomethylation of ICRs in the trophectoderm at the blastocyst stage and these changes are still apparent by the time of placenta formation at mid-gestation (18–22,49). However, the epigenetic effects of superovulation are not detected at term (38). The placenta is renown for its remarkable plasticity by comparison with other organs; it is capable of responding to changes caused by genetic disorders and environmental stressors, likely through epigenetic mechanisms, including DNA methylation (50). These observations left open the possibility that epigenetic perturbations observed with IVF could be resolved during placental development. Our findings that epigenetic changes induced by IVF persist in term placental tissues strongly exclude this possibility.

Our results demonstrate that the placenta is more sensitive to epigenetic perturbations than the fetus, a finding consistent with previous studies that examined earlier stages of development (18–20,22). Placental epigenetic effects are more prevalent, although three C7xB individuals exhibited very low DNA methylation at the H19/Igf2 ICR in both fetal brain and liver (Fig. 5). Brain and liver tissues are derived from the ectoderm and endoderm, respectively, indicating that alterations to the H19/Igf2 ICR likely occurred early in development, prior to or during epiblast differentiation (E4). We speculate that DNA methylation status of the H19/Igf2 ICR is affected as early as the differentiation of the trophectoderm (E3), as DNA methylation is also low at the H19/Igf2 ICR in the placentae of these three individuals. This is aligned with the consensus hypothesis that ART procedures disrupt maintenance of imprinted genes in preimplantation development (19,38). Although the reduction in C7xB fetal weight with IVF can be indirectly caused by placental dysfunction, we cannot totally exclude direct effects in the embryo.

The Barker Hypothesis, or the Developmental Origins of Health and Disease (DoHaD) Hypothesis postulates that environmental influences in utero can impact the long-term health of offspring. It has become clear that the prenatal period is a critical window of development. Phenotypes observed in this study, namely, low birth weight and abnormal placentation are certainly implicated in the etiology of cardiovascular and metabolic diseases (51), and further investigation into the long-term health effects of ART is warranted.

Materials and Methods

Animals

Two mouse crosses were used. B6(CAST7) (C7) mice possess chromosome 7 from Mus musculus castaneus on a C57BL/6J (B6) background. C7 females were mated with B6SJLF1 males (Jackson Laboratory) and the resulting F1 progeny (C7xB) carried strain-specific single nucleotide polymorphisms (SNPs) that were used to analyze allele-specific expression of imprinted genes on chromosome 7 (52). CF1 females (Harlan laboratories) were also crossed with B6SJLF1 males to generate progeny (CFxB), and served as surrogates for all blastocysts that underwent an ET (see below). The day on which a vaginal plug was observed was denoted as embryonic day (E) 0.5, and the pregnant dams were euthanized 18 days later to collect E18.5 naturally conceived fetuses and placental tissues. All animal work was conducted with the approval of the Institutional Animal Care and Use Committee at the University of Pennsylvania.

Generation of ET and SET concepti

We also generated concepti that were only exposed to ET and concepti that were exposed to both SET. CF1 females were naturally mated with B6SJLF1 males to generate ET and SET concepti. For ET concepti, in vivo fertilized blastocysts were flushed from both uterine horns at embryonic day (E) 3.5 using Whitten/HEPES buffer (53) and then transferred into potassium simplex optimized medium (KSOM) containing amino acids (54) for approximately 10 min. Blastocyst-stage embryos were subsequently transferred into 2.5-day or 3.5-day postcoitum pseudopregnant CF1 females. Embryo transfers were performed using the Non-Surgical Embryo Transfer Device (NSET; Paratechs), and a maximum of 10 embryos were transferred to each surrogate. For SET concepti, CF1 females were stimulated with gonadotropins as previously described (22), and in vivo fertilized blastocysts were flushed and transferred into pseudopregnant CF1 females using NSET. The day of ET was denoted as E3.5, and all recipient females were euthanized 15 days after transfer to obtain E18.5 embryos.

Generation of IVF concepti

In vitro fertilization was performed as previously described using sperm from B6SJLF1 males and eggs from C7 or CF1 females (22). All in vitro fertilized embryos were cultured for 3.5 days at 37°C in a reduced oxygen environment (5% CO₂, 5% O₂, 90% N₂) to obtain morula/blastocyst-stage embryos that were subsequently transferred into pseudopregnant CF1 females as described above. IVF-derived concepti were subsequently collected at E18.5.

Collection of fetal and placental tissues

Fetuses and placentae were isolated by mechanical dissection of the conceptus and wet weights were recorded immediately after collection. Fetal liver and brain tissues were harvested, snap frozen in liquid nitrogen and stored at −80°C until further use. Placental tissues were bisected through the attachment site of the umbilical cord, and half of the placenta was fixed overnight using 10% phosphate-buffered formalin for histological analysis while the other half was snap frozen in liquid nitrogen and stored at −80°C for DNA and RNA isolation.

Histological analysis of placental tissues

After overnight fixation in 10% phosphate-buffered formalin, one half of the bisected placentae was dehydrated in ethanol and xylenes, embedded in paraffin wax and cut in 5 µM cross-sections. Two consecutive sections were stained with hematoxylin and eosin (H&E) and used for image analysis. Images were collected using an EVOS FL Auto Cell Imaging System and software (Life Technologies). For each image, the labyrinth and junctional zones were measured using ImageJ v1.4.5 (National Institute of Health) by a person who was unaware of the image's history. Both H&E-stained sections were analyzed for each sample, and measurements for each cross-section were averaged. Only full, unfragmented, cross-sections with distinct junctional and labyrinth zones were used for quantification. Sections with fragmentation artifacts from the dissection, embedding or sectioning process were excluded from histological analysis.

Isolation of DNA and RNA from fetal and placental tissues

To analyze DNA methylation and expression of multiple genes in the same tissue, DNA and RNA were isolated simultaneously from snap frozen fetal and placental tissues as previously described (22).

Gene expression analyses

First-strand synthesis was performed using 1 µg of total RNA extracted from fetal and placental tissues. RNA samples were treated with DNase (Promega) and converted to cDNA using Superscript III reverse transcriptase (Invitrogen) and random hexamers. To test for genomic DNA contamination, cDNA samples without reverse transcriptase were processed in parallel. Real-time quantitative PCR was performed on a 7900HT Fast Real-Time PCR Machine (Applied Biosystems) using the Power SYBR Green master mix (Applied Biosystems) and with 0.2 µM primer concentrations. For each primer set (Supplementary Material, Table S4), the reaction efficiency (E) was estimated using a standard curve and expression levels were quantified by measuring the cycle threshold (Ct) for each sample using the E(^−Ct) method (55). All samples were run in duplicate. Relative expression was calculated using the geometric mean of quantified expression from two different endogenous controls, Rplpo and Nono, which were chosen because of their stable expression levels across multiple samples.

The cDNA from C7xB fetal and placental tissues was used to assess allele-specific expression of multiple imprinted genes. Allele-specific expression of H19, Igf2, Peg3, Kcnq1ot1, Cdkn1c, Snrpn was measured using PCR-based assays and strain-specific SNPs as described previously (22,52).

Bisulfite pyrosequencing and LUMA

DNA methylation was measured at multiple loci using bisulfite pyrosequencing as previously described (22). Primers for all pyrosequencing assays can be found in Supplemental Data (Supplementary Material, Table S5). The LUMA was used to assess global methylation levels by measuring methylation at repetitive elements throughout the genome as previously described (22,56).

Statistical analysis

To assess the cumulative effects of ART statistical significance was determined using an unpaired student's t-test for embryo weights, placental weights, fetal:placental weight ratios, the histological analysis of the placentae and global methylation using LUMA. Because our previous studies have shown that only a subset of samples exhibit irregularities at individual loci (22), real-time PCR, bisulfite pyrosequencing and allele-specific expression data were analyzed using the variance ratio test to calculate an F statistic and determine whether there are statistically significant differences between groups of samples. Single embryos and placentae were used as statistical units. For the CFxB concepti, all four groups were compared with one another so the Bonferroni correction was utilized to account for the familywise error rate. To identify outliers in the bisulfite pyrosequencing data the ROUT method of regression was utilized in GraphPad Prism using a Q value of 10%. All statistical analyses were performed using GraphPad Prism version 6.0d. Samples were considered statistically significant if P < 0.05.

Supplementary Material

Supplementary Material is available at HMG online.

Funding

This work was funded by the National Institutes of Health P50 grant HD068157 (M.S.B.) as well as postdoctoral fellowships supported by the National Institutes of Health T32ES019851 (E.D.W.) and The Lalor Foundation (L.A.V.).