The superovulated environment, independent of embryo vitrification, results in low birthweight in a mouse model

Abstract

Epidemiological studies suggest that babies born following in vitro fertilization (IVF) and fresh embryo transfer are of lower birthweight than babies born following frozen embryo transfer, although the mechanism responsible for this phenotype is not known. We developed a novel mouse model that isolates the independent effects of embryo freezing and the superovulated environment, which cannot be performed in humans. We transferred blastocysts that had been vitrified and warmed, mixed with with fresh blastocysts, into individual pseudopregnant recipients produced by either natural mating or mating following injection with equine chorionic gonadotropin and human chorionic gonadotropin and hCG (superovulation). We found that superovulation of the recipient dams led to significantly lower fetal weight at term while blastocyst vitrification had no significant effect on fetal weight (1.43 ± 0.24 g fresh-natural, 1.30 ± 0.28 g vitrified-natural vs. 1.09 ± 0.20 fresh-superovulated, 0.93 ± 0.23 g vitrified-superovulated, P < 0.0001). Doppler ultrasound revealed increased median umbilical artery resistance in the placentae of near-term dams exposed to superovulation compared to naturally mated dams (0.927 vs 0.904, P = 0.02). Additionally, placental microvascular density was lower in superovulated compared to naturally mated dams (1.24 × 10⁻³ vessel/micron vs 1.46 × 10⁻³ vessels/micron, P = 0.046). Gene expression profiling suggested alterations in fetal genes involved in glucorticoid regulation. These results suggest a potential mechanism for altered birthweight following superovulation in our model and may have implications for human IVF.

Summary Sentence

Mouse pups born to recipients exposed to eCG and hCG prior to implantation are of a lower birthweight and have altered placenta vasculature, regardless of whether the pups arose from blastocysts that had been vitrified-warmed or transferred fresh.

Introduction

Assisted reproductive technologies (ART), and specifically in vitro fertilization (IVF), have resulted in 5 million births worldwide [1]. In the USA, it is estimated that 1.6% of all births are conceived with ART [2]. However, the effect of ART procedures on the long-term health of the resulting offspring is still unclear. While the overwhelming majority of babies born following IVF are healthy, epidemiological studies suggest clear differences between babies born following IVF compared to natural conception, including higher rates of small-for-gestational age (SGA) babies and higher rates of preeclampsia and other disorders of placentation in pregnancies resulting from ART [3, 4]. While some component of this may be due to underlying infertility, evidence from both human studies and animal models suggests an inherent effect of the ART procedures, including superovulation (the administration of high doses of gonadotropins to promote the development of multiple follicles), embryo culture, and embryo manipulation [5–10].

Specifically, numerous epidemiological studies reveal a higher rate of SGA and low-birthweight (LBW) babies following fresh embryo transfer (during so called “fresh cycles”) compared to frozen/warmed embryo transfer cycles (during “frozen cycles”) [11–14]. The studies suggest that the hormonal environment following embryo transfer may (at least partially) be responsible for the observed differences, as fresh embryos are transferred into the uterus directly following superovulation, with multiple corpora lutea producing high levels of hormones and other growth factors. By contrast, in a frozen/warmed cycle, embryos are transferred into a hormonally prepared uterus that more closely resembles the environment during a natural cycle. However, the role of the freeze/thaw process on neonatal outcomes is extremely difficult to isolate in these human studies.

One large study of national data from the Society for Assisted Reproductive Technology performed a secondary analysis of fresh and frozen cycles in women using donor oocytes to isolate the effect of the maternal environment. In these cycles, fresh or frozen embryos were transferred into similarly hormonally prepared uteri. The study found no difference in incidence of LBW among babies born from donor-oocyte embryos transferred fresh or frozen/warmed in which uterine preparation was similar between the two groups; in contrast, there was a higher incidence of LBW babies in fresh compared to frozen/warmed cycles among women using autologous oocytes, in which uterine preparation differed between the fresh and frozen cycles [12]. These data suggest that the environment, not the freeze/thaw process, may be responsible for the observed differences in rates of SGA, although many other factors, including immunological factors, differ between donor and autologous cycles. Observational studies have suggested higher rates of SGA and preeclampsia in patients with higher estradiol levels (a measure of response to superovulation), again suggesting that the hormonal environment following superovulation may affect neonatal outcomes [15–17]. However, other studies suggest higher rates of large-for-gestational age (LGA) babies following frozen/warmed embryo transfer cycles, suggesting that there may be an effect of the freezing process on fetal weight [18–20].

The mouse model has been a useful tool for studying the effects of ART on pregnancy outcomes [7]. Our laboratory has previously demonstrated that the superovulated environment alone is sufficient to produce a LBW phenotype in mice, and has further suggested that the mechanism may be related to placental development [21]. In this study, we have used a novel experimental design and male mice transgenic for green florescent protein (GFP) to “tag” embryos and transfer fresh and vitrified/warmed embryos together into pseudopregnant recipients. This study design allows us to isolate the effects of the superovulated environment and the embryo cryopreservation process on fetal and placental weight in a mouse model and remove the possibility that differences seen are due to subtle difference in the maternal environment. In addition, the model has revealed changes to the placental vasculature that may explain the observed effects on the fetuses.

Materials and methods

Embryo collection and culture

All experiments and procedures were approved by the University of Pennsylvania Institutional Animal Care and Use Committee. Adult mice were obtained and housed in a temperature-controlled environment with a 12 h dark/12 h light cycle and fed feed and water ad libitum. Six-week-old female CF1 mice (Envigo, Indianapolis, IN) were superovulated with intraperitoneal injections of 5 IU equine chorionic gonadotropin (eCG) (PMSG, EMD Millipore, Billerica, MA) followed by 5 IU of human chorionic gonadotropin (hCG) (Sigma-Aldrich, St. Louis, MO) 48 h later. Females were mated with transgenic males heterozygous for GFP (C57BL/6-Tg[CAG-EGFP]1Osb/J, Jackson Laboratory, Bar Harbor, ME). Zygotes with two pronuclei (PN) were collected approximately 22 h post-hCG. Embryos were collected in HEPES-buffered Whitten medium [22] and treated with hyaluronidase (1 mg/ml, Sigma-Aldrich) to disperse the cumulus cells. After cumulus cell removal, the embryos were washed through a series of drops of culture medium and placed in groups of 20 embryos in 50 μl drops of K⁺ simplex optimized medium with amino acids (KSOM+AA) (Specialty Media, EMD Millipore) under mineral oil. The embryos were then cultured in a humidified atmosphere at 37°C with 5% CO₂ and 5% O₂.

Embryo vitrification

After 4 days of culture, embryos that had progressed to the expanded blastocyst stage were grouped by GFP status. A portion of the embryos were vitrified and stored for up to 6 months prior to transfer; the others were transferred fresh. Blastocysts were placed in groups of 6 in 50 μl equilibration solution (Irvine Scientific, Santa Ana, CA) for 6 min followed by 30 s in vitrification solution (Irvine Scientific). Embryos were then loaded onto a vitrification device (Cryolock, Biotech, Alpharetta, GA; donated) and immediately submerged in liquid nitrogen. The device was capped and transferred to a liquid nitrogen tank (Taylor-Wharton, Minnetonka, MN) for storage. Prior to embryo transfer, blastocysts were warmed by transferring the vitrification device immediately to a 250 μl drop of thawing solution (Irvine Scientific) for 1 min followed by 4 min each in 50 μl drops of dilution solution and two drops of washing solution (Irvine Scientific). Blastocysts were then placed in KSOM+AA.

Embryo transfer

Blastocysts were transferred into pseudopregnant CF1 female mice [plus or minus superovulation (5 IU eCG and 5 IU hCG)] generated by mating to vasectomized C57BL/6J males (Jackson Laboratory). The presence of a copulatory plug confirmed mating. Ten blastocysts, five fresh and five vitrified/warmed, were nonsurgically transferred into a single horn of each pseudopregnant female on postcoital day 3.5 using the Nonsurgical Embryo Transfer Device (ParaTechs, Lexington, KY) as per the manufacturer's protocol. GFP status was used to designate the embryos as fresh or vitrified/warmed. For example, five GFP-positive vitrified/warmed embryos were transferred together with five GFP-negative fresh embryos into a single recipient. Conversely, five GFP-negative vitrified/warmed embryos were transferred together with five GFP-positive fresh embryos in a separate recipient. Four experimental groups were generated: (1) fresh blastocysts—natural recipient (Fresh-Nat), (2) vitrified blastocyst—natural recipient (Vit-Nat), (3) fresh blastocysts—superovulated recipient (Fresh-SO), (4) vitrified blastocyst—superovulated recipient (Vit-SO) (Figure 1).

Figure 1.

Experimental design: female mice were superovulated and mated to males heterozygous for GFP. Zygote embryos were obtained the following day and cultured to blastocysts, at which stage half of the embryos were vitrified. Pseudopregnant females were generated through either natural mating or superovulation and mating to vasectomized males. Ten blastocysts (five fresh and five vitrified/warmed) were transferred into each recipient, and GFP status was used to “label” the embryos. For example, five GFP-positive vitrified/warmed embryos were transferred along with five GFP-negative fresh embryos into a single recipient, while five GFP-negative vitrified/warmed embryos were transferred along with five GFP-positive fresh embryos into another recipient. A total of 84 transfers were performed. Pregnant mice were killed at E18.5 and four experimental groups were generated.

Fetal evaluation and tissue collection

Pregnant females were killed 15 days following embryo transfer (day E18.5) via asphyxiation and cervical dislocation. Implantation sites, including fetus and placenta, were carefully dissected from the uterine horn, and each was analyzed for GFP status using a fluorescent filter to determine origination from a fresh or vitrified/warmed blastocyst. The fetus and placenta were dissected from each implantation site and fetal and placental weights were obtained. The placenta was bisected through the midplacental plane, with half placed in 10% phosphate-buffered formalin for histological analysis and the other half snap-frozen and stored at –80°C. Fetal liver and tail tissue were snap-frozen and stored at –80°C for molecular analysis.

Naturally mated control group

Six-week-old CF1 female mice were mated to transgenic males heterozygous for GFP (C57BL/6-Tg[CAG-EGFP]1Osb/J, Jackson Laboratory); a copulatory plug postcoitum confirmed mating. Pregnant females were killed at E18.5 and the implantation sites dissected, weighed, and stored as above.

DNA/RNA extraction and complementary DNA preparation

Genomic DNA was extracted from frozen placenta, liver, and tail using the QiAmp DNA Micro Kit (Qiagen, Germantown, MD) according to the manufacturer's recommendation. RNA extraction and isolation were performed using the RNAeasy Micro Kit (Qiagen) according to the manufacturer's instructions. RNA concentration and quality was determined with a NanoDrop spectrophotometer (Thermo Fischer Scientific, Waltham, MA). Synthesis of complementary DNA (cDNA) was performed using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Waltham, MA) using 2 μg of RNA.

Sex determination

Sex determination of the fetus was performed using isolated tail DNA. PCR was performed using primers for the Sry gene: Sry forward primer: 5^΄- TTG TCT AGA GAG CAT GGA GGG CCA TGT CAA, and reverse primer: 5^΄-CCA CTC CTC TGT GAC ACT TTA GCC CTC CGA. The reactions were performed in 25 μl for 26 cycles (95°C for 30 s, 60°C for 1 min, 72°C for 30 s).

Histological analysis

Placentas were fixed overnight at 4°C, dehydrated, and embedded in a paraffin block. Care was taken to orient the half-placenta vertically so that cross sections of the placenta could be obtained. Serial tissue sections of 4 μm thickness were cut and mounted on glass slides by the Abramson Cancer Center Histology Core (University of Pennsylvania, Philadelphia, PA). Tissues were deparaffinized using citrate buffer, pH 6.0 (EMD Millipore) and stained with hematoxylin and eosin. Slides were digitized by the histology core of the Children's Hospital of Philadelphia (Philadelphia, PA). Images were viewed and measurements taken using Spectrum software (Leica Biosystems, Buffalo Grove, IL); the area of each zone was outlined manually and the surface area was calculated. Measurements were made through the midsagittal plane of the placenta and performed as previously described [23]. For the microvessel analysis, tissues were deparaffinized using citrate buffer and stained with monoclonal antibody to PLVAP (MECA-32) (Bio-Rad, Raleigh, NC). Quantification of vessel density was performed using Spectrum and ImageJ software [24].

Gene expression: microarray

RNA from placenta and fetal liver was submitted to the Molecular Profiling Facility of University of Pennsylvania for GeneChip labeling and hybridization. RNA was converted to cDNA, amplified, and hybridized to Affymetrix Mouse Gene 2.0 ST arrays (Affymetrix, Santa Clara, CA) as per the manufacturer's protocols as described in the Ovation Pico WTA system v2 user guide (NuGEN, San Carlos, CA) and the Affymetrix GeneChip Expression Analysis Technical Manual. Arrays were scanned and processed using Affymetrix Command Console software yielding probe intensity files for each sample. Probe intensity files were normalized using robust multi-array averages (Partek Genomics Suite v6.6, Partek, St. Louis, MO) yielding log₂-normalized intensities for each transcript ID in each sample. Statistical analysis was performed using SAM, Statistical Analysis of Microarrays (Stanford University, Palo Alto, CA) [25] to determine significant differences in gene expression in the four groups based on two variables (maternal environment and embryo vitrification); significance was determined by fold-change and the false discovery rate for multiple testing. Ingenuity pathway analysis (Ingenuity, Redwood City, CA) was performed on the normalized intensities to identify functional networks and pathways that differentiated the groups.

Gene expression: RT-qPCR

Real-time quantitative PCR (RT-qPCR) was performed with the ABI Prism 7900HT Sequence Detection System (Applied Biosystems). Each reaction was performed in 10 μl with 50 ng of cDNA template. Each sample was run in triplicate, and the reactions were carried out for 40 cycles. The following TaqMan probes were used (Applied Biosystems): Hand1 (Mm00433931_m1), Prl2c2 (Mm04208104_gH), Pcdh12 (Mm00450488_m1), Prl8a8 (Mm00452401_m1), and Tek (Mm00443243_m1). Gene expression was normalized to reference gene H2a (Mm00501974_s1) and relative gene expression compared to the reference group (Fresh-Nat) was calculated using the ΔΔCT method. Fold change is expressed as mean ± standard deviation (SD).

Doppler analysis

A group of pregnant mice at E17.5 was assessed using high-frequency microultrasound as described by Hernandez-Andrade et al. [26]. Briefly, anesthesia was induced with 4%–5% isoflurane and maintained with 1%–2% isoflurane. Mice were secured to a heating pad with tape, and hair was removed from the abdomen with a chemical hair remover (Nair, Church & Dwight Corp., Ewing, NJ). Mice were placed on a heated field, and temperature, cardiac and respiratory rates were monitored during the entire ultrasound procedure. The ultrasound probe was immobilized with a mechanical holder. The uterine arteries were identified lateral to the bladder on the side of the implantation sites. Doppler readings were obtained as close to a 0° angle of insonation possible. Each fetus was evaluated with a high-frequency linear 40 MHz ultrasound probe by a single experienced ultrasound technician blinded to treatment group. Doppler recordings were performed on each fetal umbilical artery and vein at the segment of the umbilical cord immediately following cord exit from the fetal abdomen. Uterine artery Doppler recordings were performed on both uterine horns. Three to six recordings were made for each vessel and averaged. Peak systolic velocity (PSV) and end diastolic velocity (EDV) were calculated. Resistance index was calculated as (PSV−EDV)/PSV. After the procedure, mice were placed under a warming lamp for recovery prior to being placed back in their cages. All mice were killed on day E18.5 as described above.

Statistical analysis

The primary outcome of the study was fetal weight: an a priori power calculation determined that 17 fetuses per group would provide 80% power to detect a 15% difference in fetal weight with an alpha of 0.05. Two-way ANOVA followed by Sidak multiple comparisons testing was used to evaluate between-group differences when all four experimental groups were compared. Multiple linear regression models were used to control for confounding factors, including litter size. Student t-test and Mann-Whitney U-tests were used to compare differences between groups in parametric and nonparametric data, respectively, when comparing two groups. A Fisher exact test was used for categorical data. A P-value < 0.05 was considered significant. Statistical analyses were performed with GraphPad Prism version 6 (San Diego, CA) and STATA version 13 (StatCorp, College Station, TX).

Results

Transfer and pregnancy outcomes

Our unique study design allowed us to independently determine the effects of both embryo vitrification and the superovulated peri-implantation environment on fetal growth with fresh and vitrified/warmed embryos transferred together into pseudopregnant recipients (either naturally mated to a vasectomized male or superovulated prior to mating) (Figure 1). We transferred a total of 840 embryos (420 fresh, 420 vitrified/warmed) into 84 recipient mice: 33 naturally mated pseudopregnant recipients and 51 superovulated and mated pseudopregnant recipients. Pregnant mice were killed at E18.5, and the fetuses and placentas were evaluated. A total of 17 natural and 15 superovulated litters were obtained for pregnancy rates of 51.5% and 29.4%, respectively (P = 0.07) (Table 1). We obtained 50 fetuses from the natural recipients and 46 fetuses from the superovulated recipients. Additionally, 60 fetuses were obtained from eight litters from control mice (naturally cycling mice mated to GFP-heterozygous males). Mean litter size was 2.9 (range 1–6) in the natural and 3.0 (range 1–6) in the superovulated recipients and did not differ significantly between the two groups (P = 0.83). Mean litter size was larger in the naturally mated control group (7.5, range 1–15, P = 0.05). Due to differences in litter size, comparisons between the control group and the experimental groups are of limited direct value; results are presented for reference purposes in Supplemental Table S1. The overall implantation rate per embryo transferred did not differ between the fresh and vitrified/warmed groups (34.4% fresh vs 25.0% vitrified, P = 0.20).

Table 1.

Results from embryo transfers.

	Natural recipients	Superovulated recipients	P-value
Transfers	33	51	N/A
Pregnancies (%)	17 (51.5%)	15 (29.4%)	0.07
Mean litter size (range)	2.9 (1–6)	3.0 (1–6)	0.83
Number of fetuses	50 (29 fresh 21 vitrified/warmed)	46 (27 fresh 19 vitrified/warmed)	N/A

Fetal and placental weight

In order to determine whether the effects of the superovulated environment and/or vitrification in our mouse model replicate what is seen in human epidemiologic studies, we examined fetal and placental weight near term (E18.5). Mean fetal and placental weights for each group are presented in Figure 2. Mean fetal weight was significantly lower in in the superovulated compared to natural environment (1.43 ± 0.24 g Fresh-Nat, 1.30 ± 0.28 g Vit-Nat, vs 1.09 ± 0.20 Fresh-SO, 0.93 ± 0.23 g Vit-SO) (Figure 2A). Two-way ANOVA demonstrated a significant effect of the superovulated environment on fetal weight [F (1,78) = 47.0, P < 0.0001]; there was no significant interaction between superovulation and vitrification. Sidak multiple comparison testing following two-way ANOVA demonstrated significant differences in fetal weight between the natural and superovulated environment in fetuses arising from both fresh and vitrified embryos (P ≤ 0.03). However, weight did not differ in fetuses arising from fresh or vitrified/warmed embryos in either the natural or superovulated environment (P ≥ 0.2). Multiple linear regression, performed to assess the effect of uterine environment (natural or superovulated) and embryo vitrification on fetal weight while controlling for litter size, showed a significant effect of uterine environment on fetal weight that was independent of litter size. Similar results were found when fetal weight was assessed in males (n = 35) and females (n = 40) separately (Figure 2B). There were no significant differences in placental weight between the four experimental groups (Figure 2C). These results demonstrate that the superovulated environment, but not embryo vitrification, results in growth restriction in the fetus, without a significant change in placental weight.

Figure 2.

(A) Fetal weight at E18.5 in the four experimental groups. Fetal weight was significantly smaller in the superovulated compared to natural recipients (*P ≤ 0.0001), but did not differ significantly in fetuses that arose from fresh or vitrified/warmed embryos (n = 28 Fresh-Nat, 19 Vit-Nat, 27 Fresh-SO, 19 Vit-SO). (B) Including fetal sex as a factor did not change the effect of the superovulated environment or vitrified embryos on fetal weight; no significant interaction was noted (**P ≤ 0.003 compared to natural groups, *P = 0.03 compared to natural vitrified/warmed). (C) There were no significant differences in placental weight between the groups. Data are presented as mean ± SEM.

Structural placental analysis

Although there was no difference in placental weight between groups, based on our previous findings, we hypothesized that there may be structural or functional changes in the placenta contributing to the differences seen in fetal weight. We therefore performed assessments of placental structure and function. Histologically, we analyzed the ratio of the labyrinth zone to the junctional zone as a measure of placental structure (Figure 3A). There were no observed differences in the labyrinth-to-junctional zone ratio between any of the four experimental groups (Figure 3B). An additional assessment of placental structure was performed with RT-qPCR for markers of specific cell types in the placenta. Similar to the findings on histological analysis, there were no differences in expression of genes representing specific placental cell types between the four experimental groups (data not shown). These data suggest that neither superovulation nor embryo vitrification affect trophoblast differentiation and placental architecture.

Figure 3.

(A) The areas of the labyrinth zone (L) and junctional zone (J) were computed and the labyrinth-to-junctional zone ratio (L: J ratio) was calculated. (B) There were no differences in the L: J ratio between the four experimental groups (n = 22 Fresh-Nat, 18 Vit-Nat, 24 Fresh-SO, 16 Vit-SO). Data are presented as mean ± SEM.

Placental vascular assessment

Our human data suggest that superovulation leads to changes in factors, such as members of the VEGF family, critical to fetal vasculogenesis. Therefore, we performed a functional analysis of placental blood flow via high-frequency Doppler ultrasound in superovulated and natural recipients, to assess whether the hormonal environment affects fetal blood flow. Umbilical artery resistance index was calculated for placentas on the day prior to tissue dissection (E17.5) in superovulated (n = 31 placentas from 11 litters) and natural recipients (n = 16 placentas from 6 litters). We observed increased median umbilical artery resistance index in the superovulated compared to natural recipients (0.927 vs 0.904, P = 0.02) (Figure 4A, B). Due to the nature of the Doppler assessment, we were unable to differentiate fresh from vitrified/warmed placentas in assessing Doppler flow. There was no significant difference in uterine artery resistance between the natural recipients and the superovulated recipients. We further assessed placental vascular structure by calculating placental microvascular density in placentas at E18.5 from superovulated (n = 15 placentas) and natural (n = 14 placentas) recipients, arising from both fresh and vitrified/warmed embryos. Microvessel density was lower in the superovulated compared to natural-recipient placentas (1.24 × 10⁻³ vessel/micron vs 1.46 × 10⁻³ vessels/micron, P = 0.046), which is consistent with the observed increased vascular resistance in superovulated placentas (Figure 4C, D). These data demonstrate that the hormonal environment in utero affected blood flow to the fetus, which may be due to changes in placental vascular development.

Figure 4.

(A, B) Median umbilical artery resistance index, as measured by Doppler ultrasound, was increased in superovulated (n = 31 placentas from 11 litters) compared to natural placentas (n = 16 placentas from 6 litters), (**P = 0.02). (C, D) Microvessel density was lower in superovulated (n = 15) compared to natural (n = 14) placentas at E18.5, *P = 0.046. Data are presented as mean ± SEM.

Global gene expression

To more globally assess differences between pregnancies arising in superovulated and natural environments, and from fresh and vitrified/warmed blastocysts, we performed global gene expression in extraembryonic tissue (placenta) and fetal tissue (liver) from E18.5 fetuses using the Affimetrix GeneChip Mouse Gene ST 2.0 Array (n = 4 per subgroup, 16 total). There were no significant differences in gene expression observed in placental or fetal liver tissue between fresh and vitrified/warmed embryos. When conceptuses from superovulated and natural recipients were compared, we observed minimal changes in placental gene expression; however, there were significant changes observed in fetal liver. There was differential expression observed in 101 genes (false discovery rate <20%, fold-change ≥1.5), with 93 genes downregulated and 8 upregulated in superovulated compared to natural environment (Supplemental Table S2). Ingenuity pathway analysis revealed enrichment of genes involved in catabolic processes and enrichment of multiple cytochrome p-450 genes. RT-qPCR validation confirmed the observed gene expression changes in several of the genes including Hsd17b6, Cyp23b7, Cyp2j5, and Tat (Supplemental Figure S1).

Discussion

In this study, we utilized a novel study design with a transgenic mouse model and embryo transfer experiments to isolate the putative effects of the superovulated environment and blastocyst cryopreservation on fetal and placental weight. Our results indicate that the superovulated environment post-embryo transfer results in a significant decrease in fetal, but not placental, weight near term, while the cryopreservation process has no significant effect on fetal or placental weight. Additionally, we have demonstrated significant differences in both umbilical artery blood flow and placental microvascular density in term fetuses following superovulation. Together, these results indicate that the superovulated environment, but not the cryopreservation process, significantly alters fetal growth and affects placental vasculogenesis, which may be responsible for the impaired fetal growth.

Our laboratory has previously found that the superovulated environment post-embryo transfer is sufficient to result in lower fetal weight in a mouse model [21]. In that study, blastocysts were collected from naturally mated mice (in the absence of gonadotropin administration) and transferred into natural or superovulated pseudopregnant recipients. The mean decrease in fetal weight (1.33 g natural vs 0.99 g superovulated) was similar to that observed in the current study (1.43 g fresh-natural, 1.30 g vitrified-natural, vs 1.09 g fresh-superovulated, 0.93 g vitrified-superovulated). However, the prior study also found a decrease in placental weight, which was not observed in the current study. The major differences between these studies were (1) the use of embryos conceived following natural conception in the prior study compared to embryos resulting from superovulated mice in the current study (2) the collection of embryos at the blastocyst stage in the previous study compared to collection of embryos at the zygote (2PN) stage with subsequent in vitro culture to the blastocyst stage in this study. These findings suggest that pre-implantation embryo exposures play an independent role in placental development. Prior studies have demonstrated that both superovulation and embryo culture can affect blastocyst development and placental DNA methylation in mice [27–31]. However, the fact that the primary outcome measure, fetal weight, was decreased in both studies confirms that superovulation has an effect on fetal growth that is independent of the effects on the gamete and pre-implantation embryo.

The mechanism responsible for the LBW following fresh IVF is poorly understood. In this study, the novel findings of changes in umbilical artery resistance and microvascular density in term placentas following transfer into a superovulated environment suggest that changes in placental vasculogenesis may be responsible for the clinically observed differences in perinatal outcomes in fresh and frozen IVF embryo transfers. One possible mechanism for the role of superovulation in altering placental vasculogenesis involves the vascular endothelial growth factor (Vegf) pathway. Vegf production by the corpora lutea is increased following superovulation, and, if sufficient levels are produced, can result in capillary dysfunction that mimics human ovarian hyperstimulation syndrome in a rat model [32]. Vegf is also expressed by the mouse endometrium and is important for uterine decidual angiogenesis and implantation [33, 34]. Prior research in the mouse has shown that overexpression of endometrial Vegf induces placental production of soluble Fms-like tyrosine kinase-1 (sFlt-1) that, by antagonizing Vegf, leads to placental vascular dysfunction and a phenotype similar to that seen in human preeclampsia [35]. The superovulated environment has been shown to alter corpora lutea hormonal production that, in turn, alters endometrial gene expression in mouse endometrium [36]. However, the effect of superovulation on endometrial expression of Vegf and its receptors remains to be elucidated. We are currently performing additional studies to establish the causality of the effect of superovulation on placental vascular development focusing on the effect of factors produced by the ovary following superovulation on endometrial expression of Vegf and other factors. It should be noted, as well, that the superovulation protocol utilized in our mouse model differs from that clinically used in human IVF, both in terms of formulations (eCG compared to recombinant FSH), length of time, and dose. Further research on the effects of specific gonadotropins should be undertaken before any findings are translated to clinical care.

No significant effect of the cryopreservation process on fetal weight, gene expression, or placental function was observed in this study. Human studies have suggested improved neonatal outcomes in frozen compared to fresh embryo transfer; however, some have suggested an increased risk of LGA babies in frozen embryo transfer cycles [18–20]. This observational finding was not replicated in our study, as no increased fetal weight was observed following blastocyst cryopreservation/thaw. However, our study utilized embryos that were fertilized in vivo, which may differ from IVF embryos in the effects of vitrification. Prior research has suggested that blastocyst vitrification, the technique of replacement of the water contents of the embryo with cryoprotectants to prevent ice-crystal formation followed by rapid cooling in liquid nitrogen, may result in aberrant methylation of the H19/Igf2 locus in mouse fetuses [37]. Further research on the observed increase in LGA babies following frozen embryo transfer in humans is warranted. Specifically, further research should address the effect of vitrification on embryos created via IVF or intracytoplasmic sperm injection.

Our analysis of gene expression, via microarray with RT-qPCR validation, demonstrates no significant differences between fresh and vitrified/warmed embryos, in both fetal and placental tissue. This reassuring finding suggests that embryo vitrification is safe, especially in conjunction with our phenotypic findings, in which no difference between fresh or vitrified/warmed embryos were observed. However, there were significant changes in gene expression in the embryonic liver as a result of the superovulated environment. This finding is concerning, in light of both the phenotype of LBW as well as multiple studies demonstrating an association of LBW and long-term metabolic disease [38, 39]. Ingenuity pathway analysis revealed downregulation of genes in the cytochrome P-450 pathway, including the Cyp2j5, Cyp2C enzymes, and Tat. Cyp2j5, a cytochrome P450 enzyme abundant in mouse kidneys, has been shown to be important in sex-specific blood pressure regulation; specifically, female mice-deficient in Cyp2j5 have elevated blood pressure [40]. The expression of many cytochrome P450 genes is regulated by glucocorticoids [41, 42]; we hypothesize that the observed changes in gene expression may be an effect of a stress response as a result of growth restriction [43, 44]. This is supported by data suggesting babies exposed to excess glucocorticoid exposure (either through exogenous administration or as the result of maternal stress environment/growth restriction) are at higher risk for cardiovascular and metabolic disease later in life [43, 45]. Further research into alterations in these genes may be important for establishing the etiology of the decreased fetal weight observed following superovulation and potential long-term implications of this phenotype.

In summary, our study has confirmed an effect of the superovulated environment post-embryo transfer on fetal growth and has demonstrated changes in placental vascular function resulting from the superovulated environment. Additionally, our study demonstrated no significant effect of the cryopreservation process on fetal growth or gene expression. In light of recent observational studies noting an association between superovulation and placental disorders, including growth restriction and preeclampsia, in fresh IVF cycles in humans [17, 46], this study confirms that, independent of any effect on the oocyte or pre-implantation embryo, including the vitrification process, the maternal environment is responsible for the growth restriction seen during fresh IVF cycles. In addition, our study provides reassurance as to the safety of cryopreservation. In addition, by demonstrating changes in placental blood flow and vasculogenesis following superovulation, this study is one of the first to demonstrate a potential mechanism for the effect of superovulation on fetal growth. An understanding of the mechanism(s) responsible for these adverse neonatal outcomes is critical to altering current ART procedures and ensuring their safety.

Supplementary data

Supplementary data are available at BIOLRE online.

Supplemental Information contains Supplemental Experimental Procedures, two figures and one table are available online.

Supplemental Figure S1: Relative gene expression validation by q-PCR of genes differentially expressed in the fetal liver by microarray. *P < 0.05

Supplemental Table S1: Outcomes of naturally-mated controls compared to experimental groups.

Supplemental Table S2: Differentially expressed genes in fetal liver by microarray

Supplementary data

Supplementary data are available at BIOLRE online.