Reproductive outcomes from maternal loss of Nlrp2 are not improved by IVF or embryo transfer consistent with oocyte-specific defect.

Abstract

Nlrp2 encodes a protein of the oocyte subcortical maternal complex (SCMC), required for embryo development. We previously showed that loss of maternal Nlrp2 in mice causes subfertility, smaller litters with birth defects and growth abnormalities in offspring, indicating that Nlrp2 is a maternal effect gene and that all embryos from Nlrp2-deficient females that were cultured in vitro arrested before the blastocysts stage. Here, we used time-lapse microscopy to examine the development of cultured embryos from superovulated Nlrp2-deficient and wild-type mice after in vivo and in vitro fertilization. Embryos from Nlrp2-deficient females had similar abnormal cleavage and fragmentation and arrested by blastocyst stage, irrespective of fertilization mode. This indicates that in vitro fertilization does not further perturb or improve the development of cultured embryos. We also transferred embryos from superovulated Nlrp2-deficient and wild-type females to wild-type recipients to investigate if the abnormal reproductive outcomes of Nlrp2-deficient females are primarily driven by oocyte dysfunction, or if a suboptimal intra-uterine milieu is a necessary factor. Pregnancies with transferred embryos from Nlrp2-deficient females produced smaller litters, stillbirths, and offspring with birth defects and growth abnormalities. This indicates that the reproductive phenotype is oocyte-specific and is not rescued by development in a wild-type uterus. We further found abnormal DNA methylation at two maternally imprinted loci in the kidney of surviving young adult offspring, confirming persistent DNA methylation disturbances in surviving offspring. These findings have implications for fertility treatments for women with mutations in NLRP2 and other genes encoding SCMC proteins.

Summary sentence.

Female mice with loss of Nlrp2 have poor reproductive outcomes that are not rescued by in vitro fertilization or transfer of embryos to the uterus of a wild-type female, supporting that this maternal effect mutation causes an oocyte-specific defect.

Introduction

About 10–15% of couples experience infertility, the inability to conceive after 6–12 months [1]. Known causes of infertility can be categorized as defects of oocyte or sperm production, abnormalities of the reproductive tract that affect sperm transport, abnormalities of implantation, and a broad category of immunological factors [1]. However, 30% of infertility remains unexplained even after standard work-up [2] and is currently treated empirically with variable success rates [2–4]. Thus, identifying new underlying causes of unexplained infertility is crucial to advance treatment options [5–7]. A subset of women clinically diagnosed with unexplained infertility do not have defects in gamete production, fertilization or embryo implantation, but their embryos fail to cleave and develop into blastocysts resulting in loss of pregnancy before it is clinically recognized [3, 7, 8]. Some of these infertile women have inactivating mutations in maternal effect genes that produce mRNAs and proteins which are stored in the developing oocyte and are essential before the zygotic genome is activated during maternal to zygotic transition (MZT) [9–11]. During MZT, maternal transcripts and proteins are degraded and the embryo begins to transcribe RNA from the zygotic (embryonic) genome [12, 13]. A subset of maternal effect genes encode proteins of the subcortical maternal complex (SCMC), a cytoplasmic protein complex localized under the oocyte cortex that persists in the periphery of the preimplantation embryo. The SCMC, initially identified in mice and later in humans, plays a critical role in preimplantation embryo development [11, 14–19]. SCMC proteins known to date are abundant in early embryos and include transducin-like enhancer of split 6 (TLE6), KH domain containing 3 (KHDC3), protein-arginine deiminase type 6 (PADI6), and oocyte-expressed members of the NLR family of proteins, NLRP2, NLRP5, and in humans, NLRP7 [11, 14, 16–24].

Loss-of-function mutations in mouse and human genes encoding these SCMC proteins result in a variety of reproductive phenotypes including subfertility/infertility, recurrent pregnancy loss, a subset of which present as familial recurrent hydatidiform moles in humans, and imprinting disorders in the offspring [10, 19, 25–35]. Other adverse reproductive outcomes, observed primarily in animal models, include stillborn and liveborn offspring with birth defects or growth abnormalities [19, 36]. For some of these genes, pregnancies or offspring with these adverse outcomes have loss of DNA methylation at multiple differentially methylated regions (DMRs) that control the mono-allelic expression of maternally imprinted genes. This supports an incompletely characterized role of at least a subset of SCMC proteins in establishment or maintenance of DNA methylation at maternally imprinted loci [11, 18, 19]. Maternal autosomal recessive loss-of-function mutations of NLRP7 and KHDC3L cause imprinting disorders and recurrent familial hydatidiform molar pregnancy. Maternal loss of NLRP7 has been associated with infertility, recurrent pregnancy loss, including recurrent molar pregnancy, and offspring with multi-locus imprinting defects, who clinically present with Beckwith-Wiedemann syndrome [25–27, 29, 34, 36–43].

We previously characterized mice with loss of Nlrp2 [24], considered the ancestral homolog of the human NLRP7 and NLRP2 genes, which are thought to have arisen through a genomic duplication that is not present in rodents [11, 18, 44, 45]. We found that female mice with loss of function of Nlrp2 have a range of reproductive phenotypes. This includes severely reduced fertility, with fewer and smaller litters that include stillborn and liveborn mice with craniofacial and limb abnormalities, and altered growth, but also normal developing offspring [24]. In vitro cultured embryos retrieved from superovulated Nlrp2-null females after mating with wild-type males resulted in predominant embryo arrest at the blastocyst stage, suggesting a more severe phenotype with in vitro embryonic development. In contrast, embryos derived from wild-type or heterozygous Nlrp2-null females develop normally regardless of the embryonic Nlrp2 genotype [24].

These data support that the reproductive phenotypes of females that lack Nlrp2 result from functional defects in the oocytes and highlighted two important questions that warrant investigation. First, the more severe phenotypes of in vitro cultured embryos retrieved from superovulated Nlrp2-null females after spontaneous mating, compared to those of embryos and offspring from natural pregnancies, prompts the question if in vitro fertilization (IVF) of superovulated oocytes would independently affect embryo development [6]. To address this, we performed in vitro fertilization and embryo culture of superovulated oocytes from Nlrp2-null females and compared their development to those of recovered in vivo fertilized embryos. Second, our experiments thus far cannot exclude a role for altered uterine receptivity to implantation and support of the post-implantation embryo development in Nrlp2-deficient females. This could be because NLRP proteins are known to have functions in innate immunity and apoptosis [45–47]. To test this, we transferred in vivo fertilized embryos from females lacking Nlrp2 expression to the uterus of wild-type mice and compared their development to that of transferred embryos from wild-type females. The results of these experiments should increase our understanding of the mechanisms causing adverse reproductive outcomes and help with counseling and selection of appropriate fertility treatment options for women with mutations in these genes.

Materials and Methods

Experimental animals

All experiments were approved by the Baylor College of Medicine Institutional Animal Care and Use Committee (protocol AN-2035). Animal facilities were accredited by the Association for Assessment and Accreditation for Laboratory Animal Care International (AAALAC). The Nlrp2-deficient mouse model (Nlrp2^tm1a/tm1a) was generated on a C57BL6/J background and genotyped as previously described [29]. To generate wild-type Nlrp^+/+ and null Nlrp2^tm1a/tm1a offspring, Nlrp2^+/tm1a females of at least 6 weeks of age were mated with Nlrp2^+/tm1a males of at least 6–8 weeks of age. Pups were weaned on postnatal day (PND) 21 and female and male pups were separated and housed with up to 4 animals per cage. All animals were provided standard water, feed (Pico Lab, LabDiet), and nesting material ad libitum. After determining their genotypes, Nlrp2^+/+and Nlrp2^tm1a/tm1a males and females were selected for all further experiments.

Mouse Superovulation

Four to five-week-old female Nlrp2^+/+ (wild-type) and Nlrlp2^tm1a/tm1a (null) mice were injected intraperitoneally with 5 IU of pregnant mare serum gonadotropin (PMSG; Biovendor, North Carolina, Cat# 50-893-505 or Prospecbio, East Brunswick, New Jersey, Cat# HOR-272). After 47–48 h, the mice were injected intraperitoneally with 5 IU of human chorionic gonadotropin (hCG; EMD Millipore, Massachusetts, Cat# 3672221000IU or Fisher Scientific, Hampton, New Hampshire Cat# ICN19859110).

In vitro fertilization

Sperm pre-incubation media (TYH and 0.75 mM methyl-β-cyclodextrin or MCBD) was equilibrated overnight at 37°C in 5% CO_2. For each IVF experiment, fresh sperm was collected from the cauda epididymides of two 8-week to 8-month old fertile Nlrp2^+/+ males and pooled before incubation in capacitation medium for 10–20 minutes. Superovulated females were humanely euthanized 20 h after hCG administration using isoflurane and cervical dislocation. The oviducts were dissected and placed in warmed high calcium human tubal fluid (HTF) medium (Irvine Scientific, Santa Ana, California, Cat# 90126). Gentle traction was applied to the ampulla to create a micro-perforation and release the cumulus-oocyte-complexes (COCs). The COCs were rinsed in the HTF medium until all supporting cells and debris were removed. The capacitated sperm suspension was then added to the HTF medium containing the oocytes and co-cultured for 4 h at 37°C with 5% CO₂ and 5% O₂. Next, embryos were washed in equilibrated SAGE 1-step media (Origio, Denmark, Cat# AR67010010) before the embryo culture.

In vivo fertilization

For the in vivo fertilization group used for comparison to the IVF group, superovulated female mice were paired 1:1 on the day of hCG administration with a Nlrp2^+/+ who was between 8 weeks to 8 months old and had proven fertility. Approximately 20 h later, female mice with copulatory plugs were humanely euthanized using isoflurane and cervical dislocation. Oviduct dissection and recovery of COCs containing fertilized oocytes were performed as described further down by the IVF group via EmbryoScope time-lapse microscopy system.

Embryo Culture

Embryo culture and morphokinetic assessment were performed in the same manner for in vitro and in vivo fertilized embryos. EmbryoSlide culture dishes (Origio, Charlottesville VA, Cat# MXL3–135) were filled with the SAGE 1-step media, overlaid with mineral oil (Vitrolife, Sweden, Cat# 10029), and equilibrated overnight at 37°C in 5% CO₂ and 5% O₂. In vivo fertilized embryos were transferred (two per well) to the EmbryoSlide culture dishes approximately 1 h after oviduct dissection and in vitro fertilized embryos were similarly transferred after washing and equilibration for 1 h in SAGE 1-step media. The EmbryoScope time-lapse microscopy system captured images every 10 minutes for the duration of culture. Videos (Supplemental video 1 & 2) were compressed to 3–5 h/sec Fertilization was confirmed by the presence of two pronuclei or a second polar body. We assessed embryo morphokinetics as follows: fading of the two pronuclei (PNf), division to two cells (t2), division to three cells (t3), and so on; until division to more than eight cells, compaction to morula (tM), start of the blastocoel cavity formation (tEB), blastocyst cavity filling 50% of the embryo (tB), expansion of the blastocyst within the zona pellucida (tExpB), and the hatching blastocyst (tHB). Morphokinetic parameters were assessed by three independent examiners. The researchers were blinded to the genotype and source of the embryos analyzed.

Embryo transfer

Superovulation was first induced in young Nlrp2^+/+ and Nlrp2^tm1a/tm1a females as described above. Nlrp2^+/+ females were then mated with Nlrp2^tm1a/tm1a males, producing heterozygous Nlrp2^M+/P-/Z+ embryos in which all the zygotic (Z+) Nlrp2 was maternally contributed (M⁺), but none was paternally contributed Nlrp2 (P⁻). Reciprocally, Nlrp2^tm1a/tm1a females were mated with Nlrp2^+/+ males to produce heterozygous Nlrp2^M-/P+/Z+ embryos, in which all the zygotic (Z+) Nlrp2 was paternally contributed (P+), but that lacked maternally contributed Nlrp2 (M−). Thus, all embryos were heterozygous and only differed by which parent contributed the single functional Nlrp2 allele. The morning after mating (day E0.5), female mice were checked for the presence of vaginal plugs, euthanized and dissected for isolation of the female reproductive tract, and embryo and COCs collection. Dissected reproductive tracts containing Nlrp2^M+/P−/Z+ and Nlrp2^M−/P+/Z+ embryos from Nlrp2^+/+ and Nlrp2^tm1a/tm1a females respectively, were kept in separate pools. COCs were flushed from the oviducts and inspected for the presence of fertilized oocytes, which were the only ones used for embryo transfer. Around 10–19 embryos/female recipient were surgically transferred by experienced personnel at the BCM Genetically Engineered Mouse Core (GEM Core) into wild-type females of the Institute of Cancer Research (ICR) strain, keeping the Nlrp2^M+/P−/Z+ genotypes separate from the Nlrp2^M−/P+/Z+ genotypes, as they cannot be differentiated molecularly. ICR females were chosen because of their good uterine receptivity and optimal reproductive outcomes with embryo transfers. Before embryo transfer, ICR females were mated with vasectomized ICR males to create a pseudopregnant uterine environment. For embryo transfer, pseudopregnant females at 0.5 days post coitum (dpc) were anesthetized with an intraperitoneal (IP) injection of 80 mg/kg body weight ketamine and 16 mg/kg xylazine. The ovary and ovarian fat pad were visualized through a small longitudinal incision that was made between the rib cage and the hip was then made. Subsequently, a small incision was made in the abdominal wall next to the fat pad above the ovary through which the edge of the infundibulum of the oviduct was immobilized, and the transfer pipette was inserted into the opening of the ampulla to transfer the embryos. The reproductive tract was then gently placed back into the abdominal cavity and the abdominal wall and skin were closed. Female recipients were singly housed and monitored daily until embryonic day E17.5, and at that time monitoring was increased to up to three times daily. Gestational length and the exact timing of deliveries were recorded for each female recipient.

We collected data on the mean number of ovulated oocytes, fertilization rate of female donors, percentage of female recipients that produced litters, number of offspring per litter, live births, stillbirths, neonatal or postnatal deaths, any abnormal offspring phenotypes including gross morphological and developmental abnormalities. Liveborn pups were monitored daily until PND21 for weight, survival and any changes in the observed congenital abnormalities. Any stillborn, neonatally and postnatally demised pups were collected for tissue and organ dissections.

Skeletal Staining

Whole-mount skeletal preparations were performed on a subset of demised and euthanized offspring. After euthanizing and dissecting the pups, skin, muscle and all internal organs were removed. The carcasses were fixed overnight in 95% ethanol on a rotator at room temperature. They were then stained overnight with Alcian blue dye (Sigma-Aldrich, St. Louis MO, Cat# A3157) and placed again in 95% ethanol for at least 3 h, after which they were transferred into a 2% Potassium Hydroxide (KOH) solution and incubated for 24 h. They were then stained overnight in alizarin red (Sigma-Aldrich, St. Louis MO, Cat# A5533) after which the skeleton was cleared in 1% KOH in 20% glycerol for at least 2 days. Skeletons were then stored in a 1:1 mixture of glycerol and 95% ethanol and imaged.

Bisulfite sequencing

We isolated five tissues (spleen, kidney, liver, brain and testis/ovaries) from six-week-old males and females each for the Nlrp2^M+/P−/Z+ and Nlrp2^M−/P+/Z+ genotypes (N=5 per group). There were no significant difference in weight and size of any of the tissues from male and female animals and no gross morphological abnormalities were observed. We isolated genomic DNA from five tissues (spleen, kidney, liver, brain and testis/ovaries) of six week-old males and females each for the Nlrp2^M+/P−/Z+ and Nlrp2^M−/P+/Z+ genotypes (N=5 per group) using the DNeasy Blood and Tissue Kit (Qiagen, Cat # 69506). Before DNA isolation all the tissues were in weighed and checked for any gross abnormalities; none were observed. Two-hundred ng of DNA was bisulfite-converted using the EZ DNA Methylation-Direct Kit (Zymo Research, Cat # D5020), from which we amplified the DMRs of Zac1, Mest and Impact using our previously described primers and protocol [24]. PCR products were cloned into the PCR4 TOPO TA cloning vector (Life Technologies, Cat # K4575–02). Sixteen colonies for each tissue and locus were selected following blue-white screening, cultured and sequenced using the M13 reverse primer (GeneWiz, New Jersey). We used the QUantification tool for Methylation Analysis (QUMA) [48] to align FASTA sequences with the genomic reference sequence, quantify bisulfite conversion efficiency, calculate percentage of methylated versus unmethylated CGs, and export to files as lollipop diagrams.

Statistical Analysis

Variables were reported using descriptive statistics. Continuous variables were checked for normality using the Kolmogorov-Smirnov test as well as by standard histograms and were reported as mean ± SD. Categorical variables were reported as n (%). Statistical significance for comparisons was evaluated using the Student’s t-Test for continuous variables and the Chi-Square, Fisher’s Exact Test for categorical variables or ANOVA as appropriate. Survival curves for offspring were generated using Kaplan-Meier Analysis. Differences in survival functions between the groups of the study were tested using the Logrank test (Mantel-Cox). For morphokinetic parameter analysis mixed model regression was used.

Statistical analyses were performed using IBM SPSS Statistics (Armonk, New York, Version 23.0.0.0.) and graphs were created using Graph Pad Prism 6 (La Jolla, California, version 6.01).

Results

In vivo and in vitro fertilization produce similar numbers of fertilized oocytes.

The breeding strategies and timelines for in vitro and in vivo fertilization experiments are illustrated in (Figure 1A). We recovered 229 COCs from ten Nlrp2^+/+ females (Average/mouse: 22.9 [Range: 6–44]) and 188 COCs from nine Nlrp2^tm1a/tm1a females (Average/mouse: 20.9 [Range: 4–41]) (p=0.76), indicating no difference between maternal genotypes in the number of COCs after in vivo superovulation and fertilization (Figure 1B). We recovered 305 COCs from ten Nlrp2^+/+ females (Average/mouse: 30.5 [Range: 9–70]) and 198 COCs from eight Nlrp2^tm1a/tm1a females (Average/mouse: 24.8 [Range: 8–48]) (p=0.49), indicating that there was also no difference between maternal genotypes in the number of COCs recovered for in vitro fertilization (Figure 1C). We microscopically confirmed fertilization as the presence of zygotes with two pronuclei or two polar bodies and calculated fertilization rates for each female as the number of zygotes divided by the number of oocytes multiplied by 100. There was no difference in the in vivo fertilization rates of oocytes from Nlrp2^+/+ females (50.6% [Range 0–100%]) compared to those from Nlrp2^tm1a/tm1a females (43.4% [Range: 3–86%]) (p=0.86) (Figure 1D). There was also no difference in the in vitro fertilization rates of oocytes from Nlrp2^+/+ females (63.9% [Range: 37.9–100%]) compared to those from Nlrp2^tm1a/tm1a females (72.8% [Range: 29.6–100%]) (p=0.86) (Figure 1E). Thus, oocyte quality and fertilization potential were not affected by maternal loss of Nlrp2 in either in vitro or in vivo fertilization of the superovulated oocytes.

Figure 1: Oocyte or zygote recovery and fertilization rate between maternal genotypes with in vivo and in vitro fertilization.

(A) Experimental outline and breeding strategies for in vivo and in vitro fertilization. (B, C) The number of recovered cumulus oocyte complexes (COCs) was not different between maternal genotypes after in vivo (B) or in vitro fertilization (C). (D, E) The fertilization rate was not different between maternal genotypes after in vivo (D) or in vitro fertilization (E). Each wild-type Nlrp2^+/+ mouse is depicted as a circle and each Nlrp2^tm1a/tm1a null mouse is depicted as a square.

Cultured preimplantation embryos from Nlrp2^tm1a/tm1a dams had higher rate of embryo arrest and abnormal morphology after in vivo and in vitro fertilization.

We cultured in vivo and in vitro fertilized embryos in the EmbryoScope time-lapse microscopy system for up to 150 h. We calculated embryo arrest as the percentage of embryos from confirmed fertilized oocytes that stopped dividing or disintegrated at the 2–4 cell, 5–8 cell, morula, blastocoel cavitation and hatching blastocyst stages. Cultured embryos from Nlrp2^tm1a/tm1a dams had lower rates of normal appearing preimplantation development, compared to embryos from Nlrp2^+/+ dams after in vivo fertilization (Figure 2A, Figure 3) and after in vitro fertilization (Figure 2B, Figure 4), Furthermore, cultured embryos from Nlrp2^tm1a/tm1a dams had no difference in survival to the blastocyst stage between modes of fertilization (Figure 2C).

Figure 2: Survival and morphokinetic analysis of in vitro developing embryos.

(A-C) Survival of in vitro cultured embryos showed lower survival at each developmental stage for embryos from Nlrp2^tm1a/tm1a females compared to those from Nlrp2^+/+ females, after in vivo (A) and in vitro (B) fertilization with no difference between the two modes of fertilization (C). (D-F) Graphs of morphokinetic parameters by EmbryoScope time-lapse microscopy shows that for those embryos that continued to divide, comparing embryos from Nlrp2^tm1a/tm1a females to those from Nlrp2^+/+ females, after in vivo (D) and in vitro (E) fertilization with no difference between the two modes of fertilization (F). The list of assessed parameters with p-values is provided in accompanying Table 1. Embryos from Nlrp2^tm1a/tm1a are represented as squares and those from Nlrp2^+/+ females as circles. X-axis shows the stages of cellular division (X-axis); Y-axis is the time in hours (h).

Figure 3: In vivo fertilized embryos of superovulated Nlrp2^tm1a/tm1a females have defective development during in vitro embryo culture.

(A-G) Each bar graph represents the percentage of morphologically typical embryos at the indicated stage remaining from the originally cultured 84 zygotes recovered from Nlrp2^+/+ females (white bar) or 64 zygotes recovered from Nlrp2^tm1a/tm1a females (grey bar) with p-values or each stage noted above each graph. The images to the right of graphs A-G are representative still images at the indicated stage from time lapse embryoscopy videos of in vitro cultured embryos; top image (+/+) are embryos from Nlrp2^+/+ females and bottom image (tm1a/tm1a) are embryos from Nlrp2^tm1a/tm1a females. Panels represent following stages: (A) 1-cell, (B) 2–4 cell, (C) 5–8 cell, (D) morula (note abnormal morphology of morulas in in tm1a/tm1a), (E) blastocoel cavity (in tm1a/tm1a, one embryo did not form a blastocoel and the arrow points to an abnormal blastocoel), (F) blastocyst, (G) hatching blastocyst. No blastocysts or hatching blastocysts were found in tm1a/tm1a. (H) this graph represents the % of morphologically atypical embryos found at the blastocyst stage.

Figure 4: In vitro fertilized embryos of superovulated Nlrp2^tm1a/tm1a females have defective development during in vitro embryo culture.

(A-G) Each bar graph represents the percentage of morphologically typical embryos at the indicated stage remaining from the originally in vitro fertilized 184 zygotes from Nlrp2^+/+ females (white bar) or 137 in vitro fertilized zygotes recovered from Nlrp2^tm1a/tm1a females (grey bar) with p-values or each stage noted above each graph. The images to the right of graphs A-G are representative still images at the indicated stage from time lapse embryoscopy videos of in vitro cultured embryos from Nlrp2^tm1a/tm1a females (tm1a/tm1a). Panels represent following stages: (A) 1-cell, (B) 2–4 cell, (C) 5–8 cell, (D) morula, (E) blastocoel cavity (one embryo did not form a blastocoel and the arrow points to an abnormal blastocoel), (F) blastocyst, (G) hatching blastocyst. No blastocysts or hatching blastocysts were found in tm1a/tm1a. (H) this graph represents the % of morphologically atypical embryos found at the blastocyst stage.

Morphological analysis at each stage, cleavage arrest and disintegration survey of in vivo fertilized embryos (Figure 3) and in vitro fertilized embryos (Figure 4), revealed two phases with the greatest arrest of development of embryos from Nlrp2^tm1a/tm1a females; at the 2–4-cell stage and after the morula stage. We also performed morphokinetic analysis to measure the rate by which embryos reach various stages over time. In vivo fertilized embryos of Nlrp2^tm1a/tm1a females developed slower than those of their wild-type counterparts at the morula (tM), early blastocoel (tEB), 50% expanded blastocoel (tB), and expanded blastocoel (tExpB) stages, while in vitro fertilized embryos of Nlrp2^tm1a/tm1a females developed slower than wild-type counterparts starting at the 2-cell stage (Figure 2D–F; Table 1). These combined data show that cultured embryos from superovulated oocytes of Nlrp2^tm1a/tm1a females arrest early, develop slower or disintegrate before the blastocyst stage, after both in vivo and in vitro fertilization.

Table 1:

Comparison of morphokinetic parameters obtained by Embryoscope time-lapse microscopy imaging between embryos from Nlrp2^tm1a/tm1a and Nlrp2^+/+females cultured after in vivo and in vitro fertilization. P-values indicated.

Morphokinetic Parameters	In vivo P-value	In vitro P-value
tPNf: Fade of two pronuclei	0.3485	0.428
t2: 2 Cell	0.0657	0.017
t3: 3 Cell	0.5589	0.046
t4: 4 Cell	0.5651	<0.0001
t5: 5 Cell	0.6277	0.0009
t6: 6 Cell	0.8221	0.0022
t7: 7 Cell	0.6062	0.0012
t8: 8 Cell	0.4077	0.0035
tM: Morula	0.011	<0.0001
tEB: Early Blastocyst Cavity	<0.0001	<0.0001
tB: 50% Blastocyst Cavity Expansion	<0.0001	<0.0001
texpB: Expanded Blastocyst Cavity	<0.0001	<0.0001
tHB: Hatching Blastocyst	0.894	0.0003

Transfer of Nlrp2^M−/P+/Z+ embryos to the uterus of wild-type females does not rescue poor pregnancy outcomes

We performed eight rounds of superovulation on 42 Nlrp2^+/+ donor females, mated with Nlrp2^tm1a/tm1a males to produce heterozygous Nlrp2^M+/P-/Z+ embryos and on 47 Nlrp2^tm1a/tm1a females of similar ages mated with Nlrp2^+/+ males to produce Nlrp2^M−/P+/Z+ embryos. Both maternal genotypes were represented in each superovulation round (Figure 5A). There was no difference between Nlrp2^tm1a/tm1a and Nlrp2^+/+ donors in the number of copulatory plugs and evidence of superovulation when reproductive tracts were dissected. The average number of recovered COCs per mouse was 32 for Nlrp2^+/+ donors and 29 for Nlrp2^tm1a/tm1a donors (Supplementary Table 1). Fertilization rate, calculated as the number of fertilized oocytes over the total number of COCs flushed from the oviducts, was not different between maternal genotypes: Nlrp2^+/+ donors produced 384 embryos out of 1359 flushed COCs (28%) and Nlrp2^tm1a/tm1a donors produced 357 embryos from 1337 flushed COCs (26.5%) (Chi-Square Test, P = 0.3), (Supplementary Table 1). Thus, as expected [29], ovarian response to ovulation induction and fertilization was not affected by maternal loss of Nlrp2.

Figure 5: Embryo transfer experimental strategy and pregnancy outcomes.

(A) Schematic overview of breeding scheme. Nlrp2^tm1a/tm1a females are bred with Nlrp2^+/+ males to create Nlrp2^M-/P+/Z+ embryos (top panel) and Nlrp2^+/+ females are mated with Nlrp2^tm1a/tm1a males produce Nlrp2^M+/P-/Z+ embryos (bottom panel) for transfer to wild-type ICR females (Wt ICR). (B) Litter sizes (Y-axis) at postnatal day 0 (PND0; left two bars) and remaining litter sizes at postnatal day 21 (PND21; right two bars) of ICR recipients of Nlrp2^M+/P−/Z+ (blue) and Nlrp2^M−/P+/Z+ (red) embryos, excluding recipients that did not achieve pregnancy. The average litter size +/− standard deviation is given below each bar and the N above each bar indicates the total number of offspring for respective stage and genotype; p-values indicate the significance of difference between genotype. (C) Stillbirth rates in Nlrp2^M+/P−/Z+ and Nlrp2^M−/P+/Z+ offspring. (D) Livebirth rates of Nlrp2^M+/P-/Z+ and Nlrp2^M-/P+/Z+ offspring, excluding recipients that did not achieve a pregnancy. (E) Kaplan-Meyer survival curves of Nlrp2^M+P−/Z+ and Nlrp2^M−/P+/Z+ offspring from PND0 to PND21. (Y axis is cumulative survival; X axis is time in days). Survival function is significantly different between the two groups (Log-rank Test; P<0.001). (F) Offspring weights in grams (g) at PND5, PND10, and PND21. Total number of offspring weighed, and p-values are given above each graph (Nlrp2^M+/P-/Z+ are in blue and Nlrp2^M−/P+/Z+ are in red).

We transferred the 384 Nlrp2^M+/P−/Z+ embryos to 23 ICR recipients and the 357 Nlrp2^M−/P+/Z+ embryos to another 24 ICR recipients and closely monitored their reproductive outcomes. Four of the 23 recipients with Nlrp2^M+/P−/Z+ embryos and four of the 24 recipients with Nlrp2^M−/P+/Z+ embryos did not become pregnant, which was not significantly different between both groups (P>0.9). Considering that success in achieving pregnancy is one of the outcomes, we needed to evaluate for the Nlrp2^tm1a/tm1a maternal effect mutation. We, therefore, analyzed the pregnancy outcomes both with and without the inclusion of female recipients that did not become pregnant. Ninety-six pups were born alive from transferred Nlrp2^M+/P-/Z+ embryos and 74 pups were born from transferred Nlrp2^M−/P+/Z+ embryos. When females who did not get pregnant are excluded (Figure 5B), the average litter size at PND0 per recipient is 5.1±1.9 for Nlrp2^M+/P-/Z+ offspring and 3.7±2.1 for Nlrp2^M−/P+/Z+ offspring, which is significantly different (P=0.04; Figure 5B). When the 4 recipients in each group that did not get pregnant are included, the average litter size at PND0 per recipient was 4.2±2.6 for Nlrp2^M+/P−/Z+ offspring and 3.1±2.3 for Nlrp2^M-/P+/Z+, which was not significantly different (p=0.14; Supplementary Figure 1A). There were more stillborn Nlrp2^M−/P+/Z+ pups (N=5, 7% stillbirth rate) than stillborn Nlrp2^M+/P−/Z+ pups (N=0, 0% stillbirth rate) (P = 0.01; Figure 5C). We also calculated live birth rate as the number of liveborn pups over the number of transferred embryos. When we excluded the eight female recipients (4 in each group) for which embryo transfer did not result in a pregnancy (0 pups born), live birth rates were 30.0% for Nlrp2^M+/P−/Z+ pups and 23.5% for Nlrp2^M-/P+/Z+ pups (Chi-Square Test, P = 0.06, Figure 5D). When including female recipients that did not become pregnant after surgical embryo transfer, these rates were 25.0% for Nlrp2^M+/P−/Z+ pups and 19.5% for Nlrp2^M−/P+/Z+ pups (Chi-Square Test, P=0.06) (Supplementary figure 1B). Both comparisons approached, but did not reach, significance.

Nlrp2^M−/P+/Z+ embryos have lower postnatal survival and growth.

All offspring were followed until weaning on PND21 at which time 94 of the 96 Nlrp2^M+/P−/Z+ (98.0%) had survived. The two that did not survive until PND21 were runted and euthanasia was recommended by the veterinarian. In contrast, only 58 of the 74 Nlrp2^M−/P+/Z+ offspring (78.4%) were alive at PND21 (5 were stillborn and 11 died before PND21), which was significantly different (Figure 5B). Excluding the females that did not have pregnancies, the average number of surviving offspring per recipient at PND21 was lower for Nlrp2^M−/P+/Z+ offspring (2.9±2.5) than for Nlrp2^M+/P−/Z+ offspring (5.0±2.2) (P=0.009; Figure 5B). This difference persisted when females that did not have pregnancies are excluded: 2.4±2.5 for Nlrp2^M-/P+/Z+ and 4.1±2.7 for Nlrp2^M+/P−/Z+ offspring (P=0.03; Supplementary Figure 1A). Survival over time was significantly lower for Nlrp2^M-/P+/Z+ offspring by Kaplan-Meier statistics (Logrank Test P<0.001; Figure 5E).

We weighed the surviving offspring on PND5, PND10 and at the time of weaning on PND21. The average weight of Nlrp2^M−/P+/Z+ offspring at all timepoints was lower compared to that of Nlrp2^M+/P-/Z+ offspring and most surviving Nlrp2^M−/P+/Z+ offspring appeared growth restricted at PND21 (Supplementary Table 2 and Figure 5F).

Nlrp2^M−/P+/Z+ mice have congenital anomalies that are not rescued by embryo transfer to wild-type females.

Because we previously observed congenital anomalies in naturally conceived embryos and stillborn Nlrp2^M-/P+/Z+ offspring [29], we examined liveborn and stillborn offspring immediately after birth for presence of externally visible morphological differences and congenital anomalies. Liveborns were then monitored 1–2 times per day for the first 5 days of life to document any new findings and early postnatal deaths. Seven of the 74 Nlrp2^M−/P+/Z+ newborn pups (9.5%) had at least one congenital anomaly, while Nlrp2^M+/P−/Z+ newborn offspring had no externally visible anomalies (Fisher’s Exact Test, P=0.01, Figure 6A). Of the 7 offspring with anomalies, 4 were males and 2 were females, and we were unable to determine gender for one. Abnormalities were variable and included abnormal craniofacial features (shortened and asymmetric snout, abnormally-shaped skull, small eyes, and apparent micrognathia) and asymmetric abnormalities of the extremities (apparent shortening and contractures of hindlimbs and/or forelimbs) (Figure 6B, C). Some of the Nlrp2^M-/P+/Z+ offspring with these abnormalities were found dead on PND1, others died on subsequent days, and some were severely growth restricted (Figure 6C). We examined milk spots on the newborns to evaluate if the growth restriction was related to inability to feed and found that all Nlrp2^M+/P−/Z+ and 96% of Nlrp2^M−/P+/Z+ liveborn pups had a normal milk spot, which was not different between the groups (Fisher’s Exact Test, P = 0.08).

Figure 6: Morphological and skeletal abnormalities in Nlrp2^M-/P+/Z+ offspring.

(A) Graph depicts congenital anomaly rate observed in newborn Nlrp2^M+/P-/Z+ and Nlrp2^M-/P+/Z+ offspring. (B) Images of stillborn pups showing craniofacial (arrowheads) and limb abnormalities (arrows) in Nlrp2^M−/P+/Z+ offspring. (C) Comparisons of skeletal staining between the Nlrp2^M+/P−/Z+ and Nlrp2^M-/P+/Z+ offspring at PND21 (posterior view). The Nlrp2^M−/P+/Z+ mouse is severely growth restricted with shortened long bones. In all images Nlrp2^M+/P−/Z+ are on the left and Nlrp2^M−/P+/Z+ are on the right. (Additional skeletal staining is shown in supplementary Figure 2)

Because the external anomalies primarily involved craniofacial structures, limbs and overall growth, we did whole-mount Alcian blue and alizarin red skeletal staining on a few offspring to look for skeletal abnormalities. We examined a Nlrp2^M−/P+/Z+ pup that was very small at birth, had poor interval growth and died on PND21, a growth-restricted 16-day old Nlrp2^M−/P+/Z+ pup, and a normal appearing Nlrp2^M+/P−/Z+ pup that was euthanized at PND21 for comparison. Skeletal findings in the two Nlrp2^M−/P+/Z+ mice included short long bones, absent patella, and absent or incompletely joined secondary ossification centers (Figure 6D). While there was no change in the number of vertebrae, both Nlrp2^M−/P+/Z+ offspring had unfused vertebral processes and six sternal bones instead of five (Supplementary Figure 2). While these observations require further confirmation on more animals, they suggest that Nlrp2^M−/P+/Z+ mice have delayed bone development.

Nlrp2^M−/P+/Z+ offspring born after embryo transfer have altered DNA methylation at selected imprinted loci

We previously showed higher DNA methylation levels at the DMRs of the maternally imprinted loci, Zac1 and Impact in stillborn offspring and at the Zac1 DMR in E9.5 embryos of Nlrp2^tm1a/tm1a females [24]. In that study, there was also greater spread but no changes in levels of methylation at the Mest DMR in E9.5 embryos of Nlrp2^tm1a/tm1a females. We therefore focused our analysis of DNA methylation at these DMRs in spleen, kidney, liver, brain, and testis or ovaries of 6-week-old Nlrp2^M−/P+/Z+ and Nlrp2^M+/P-/Z+ offspring that were born after embryo transfer to recipient wild-type females. The imprinted DMRs were amplified from bisulfite-converted genomic DNA and amplicons were cloned into a TA-vector, after which 16 individual clones of each were sequenced. We first confirmed by QUMA analysis that bisulfite conversion efficiency was higher than 99% and calculated the percentage of methylated versus unmethylated CpGs. There was lower DNA methylation at the Zac1 DMR in the kidney of Nlrp2^M−/P+/Z+ males (P=0.04) (Figure 7A) and more DNA methylation at the Mest DMR in the kidney of Nlrp2^M−/P+/Z+ females (P= 0.0043) (Figure 7B). No other changes in DNA methylation levels were found in tissues of Nlrp2^M−/P+/Z+ or Nlrp2^M+/P−/Z+ offspring.

Figure 7: Altered DNA methylation at selected imprinted loci in Nlrp2^M-/P+/Z+ offspring born after embryo transfer to wild-type females.

Box plots of DNA (CpG) methylation levels (%) in kidney at the imprinted differentially methylated region (DMR) of Zac1 (A) and Mest (B) in 6-week-old female and male offspring. Methylation levels were lower at the Zac1 DMR in Nlrp2^M-/P+/Z+ males and higher at the Mest1 in offspring were lower in Nlrp2^M−/P+/Z+ females. White plots are from Nlrp2^M+/P-/Z+ offspring and gray plots are from Nlrp2^M−/P+/Z+ offspring.

Discussion

Nlrp2 encodes a protein (NLRP2) of the oocyte SCMC, which is an essential complex for preimplantation embryo development in mice and humans. The SCMC is assembled by proteins translated from maternally encoded RNAs [15, 17–19]. In addition to NLRP2, other SCMC proteins include NLRP5, NLRP7 and KHDC3L (in humans), OOEP, KHDC3 (in mice), and TLE6 [17–19]. The SCMC has a functional and regulatory role in the storage of maternal RNAs and proteins [49]. It is located in the cytoplasm under the cell membrane of the oocyte. It remains there in cells at the periphery of the preimplantation embryo but is absent in areas of cell-yo-cell contact. Without the SCMC there are disruptions in spindle formation, alignment of chromosomes, aneuploidy, and imprinting [19, 49]. Several studies have shown that if any protein member of the SCMC complex is absent it destabilizes the whole complex, resulting in developmental defects along with embryonic lethality in many cases [50–54]. Loss of function of these proteins causes infertility, subfertility and recurrent pregnancy loss [10, 19, 25–34]. Maternal loss-of-function mutations of NLRP7, NLRP2, NLRP5 and KHDC3L have been implicated in human multi-locus imprinting disorders that can present as recurrent hydatidiform molar pregnancy, or imprinting disorders like Beckwith-Wiedemann syndrome, in their children [25–27, 29, 34, 36, 38–43, 55]. These conditions have loss of CpG methylation at multiple maternally imprinted DMRs [25–27, 34, 37, 38].

The current study extends our previously reported findings of fewer and smaller litters, higher rates of stillbirth and birth defects in newborns, prenatal and postnatal growth abnormalities, and increased neonatal death with maternal loss of Nlrp2. In naturally conceived pregnancies, these abnormalities were seen by E9.5 when embryos were developmentally delayed with structural abnormalities and growth restriction. The embryonic phenotype was also more pronounced with in vitro culture with all embryos arresting at various timepoints before the blastocyst stage [24].

We therefore first examined the effect of fertilization conditions on in vitro embryo development. In the current study we showed that in vitro fertilization has the same negative influence on embryo development and survival during in vitro culture. We also used a one-step, rather than two-step embryo culture medium as in our prior study [24], showing that this change in composition of the culture medium also did not significantly influence embryo development, but more studies with different media compositions are needed to form generalizable conclusions. While the mechanisms underlying this effect of in vitro culture on embryo development are not yet known, these findings indicate that embryos from Nlrp2-deficient females are more sensitive to this environment. They also support that in vitro fertilization would not be a good therapeutic strategy for infertility or pregnancy loss with maternal effect mutations in NLRP2 gene, and possibly with other genes that encode proteins of the SCMC, since maternal loss of NLRP2 disrupts the SCMC [24]. This is consistent with failure of IVF as a treatment for infertile women with mutations in genes that encode SCMC proteins [31, 32, 43, 56]. IVF is also not successful for women with familial recurrent hydatidiform molar pregnancies [34, 57, 58].

Although all observations thus far are consistent with a primary defect in the oocyte as the cause of reproductive failure in Nlrp2-deficient mice, they do not exclude an additional contribution of a suboptimal uterine environment. This is important because human NLRP2, its homologue NLRP7, and other NLRPs are components of the inflammasome and have been implicated in regulation of innate immune defense system and apoptosis [11, 44, 45, 59–62]. Modulation of the maternal immune system at the maternal-fetal interface is essential for normal embryo implantation and pregnancy. It has not been directly demonstrated that NLRPs are involved in these processes, but one in vitro study showed a role in decidualization of endometrial stromal cells [63]. The inflammasome has also been implicated in the pathogenesis of endometriosis [64]. We conducted the embryo transfer experiments in this study in as an initial approach to investigate the role of the maternal uterine environment on the embryonic phenotype of females with loss of Nlrp2. The presence of abnormalities in the offspring of Nlrp2-deficient mice after they were transferred to a wild-type uterus, which were similar to those observed with spontaneous mating [24], provided evidence that the primary mechanism is an oocyte defect, since there was no evidence of phenotypic rescue after embryo transfer and development in a wild-type uterus. Future experiments, with transfer of wild-type embryos into the uterus of Nlrp2-deficient females, will determine whether their uterus can fully support the development of pregnancies from females who do not lack Nlrp2. This would provide experimental confirmation that IVF with donor oocytes is a valid therapeutic strategy for individuals with mutations in genes that encode SCMC proteins, an approach that has already been successful in a small number of reported patients [65, 66]. However, considering that loss-of-function mutations of maternal effect genes are rare in humans, such experiments in mice will allow us to better study whether there is still a higher risk for embryonic loss or other phenotypic abnormalities with donor oocyte pregnancies.

The observation of altered DNA methylation in embryos and offspring from females with loss of Nlrp2 [24] and the widespread loss of maternally acquired imprinted DNA methylation in pregnancies and offspring of human patients [25–27, 34, 37, 38], prompted us to examine if this is also the case in adult offspring after embryo transfer. While the methylation abnormalities were not pronounced and restricted to kidney, their presence confirmed that this epigenetic alteration persist into adulthood in surviving offspring which had not been previously shown for this mouse model. Future studies on embryos, and more severely affected surviving offspring are planned to examine if methylation disturbances correlate with the severity of the phenotype and if alterations at specific loci can explain the distinctly observed craniofacial, limb and growth abnormalities.

While donor oocyte pregnancy is an available reproductive solution for women with mutations in SCMC genes, it is not an ideal option for couples who desire to use autologous gametes for personal, religious or societal reasons prefer pregnancies with their own gametes. The current mouse model will allow us to further explore other therapeutic options to overcome the loss of maternal Nlrp2 in vivo. It has already been demonstrated in elegant in vitro experiments that introduction of the Nlrp2 mRNA in developing oocytes can be achieved, but that overexpression may cause increased blastocysts apoptosis, implying that levels of NLRP2 must be carefully titrated [67]. Although ethical ramifications of such genetic manipulation make this strategy not currently acceptable for humans, the current mouse model will allow us to explore other options for restoring Nlrp2 function. Approaches that could be examined include further manipulating the composition of the culture media with in vitro fertilization to supplement oocytes and embryos with the missing proteins or downstream factors, or other means of supplementing the oocyte pool with NLRP2 by treating the mice themselves. Further detailed study of the DNA methylation defects and the causal epigenetic mechanisms could open avenues for investigating epigenetic therapies.

In summary, our new data validate that mice with loss of Nlrp2 are a good model for studying human disorders caused by mutations in this and other maternal effect genes that encode proteins of the SCMC. They support that the defects are oocyte-specific and motivate future deeper exploration of the mechanisms that underly the reproductive failure caused by these mutations and potential new therapeutic strategies.

Supplementary Material

43032_2020_360_MOESM1_ESM

Supplemental Figure 1: Embryo transfer pregnancy outcomes including recipients that did not achieve pregnancy. (A) Litter sizes (Y-axis) at postnatal day 0 (PND0; left two bars) and remaining litter sizes at postnatal day 21 (PND21; right two bars) of ICR recipients of Nlrp2^M+/P−/Z+ (blue) and Nlrp2^M−/P+/Z+ (red) embryos, including recipients that did not achieve pregnancy. The average litter size +/− standard deviation is given below each bar and the N above each bar indicates the total number of offspring for respective stage and genotype; p-values indicate the significance of difference between genotype. (B) Livebirth rates of Nlrp2^M+/P−/Z+ and Nlrp2^M−/P+/Z+ offspring, including recipients that did not achieve a pregnancy.

43032_2020_360_MOESM2_ESM

Supplemental figure 2: Comparisons of skeletal staining between the Nlrp2^M+/P−/Z+ and Nlrp2^M−/P+/Z+ offspring. (A) Skeletal staining at PND21 of the same mice shown in Figure 6, but from a lateral view; Nlrp2^M+/P−/Z+ is on the left and Nlrp2^M−/P+/Z+ is on the right. The Nlrp2^M−/P+/Z+ mouse is severely growth restricted with shortened long bones. (B) Black arrows point to fusion between vertebral bones in the Nlrp2^M+/P−/Z+ mouse which is not seen in the Nlrp2^M−/P+/Z+ mouse (dashed arrows). (C) Black arrow in the Nlrp2^M+/P−/Z+ mouse points to the patella, which was not seen in the Nlrp2^M−/P+/Z+ mouse (dashed arrow). (D) Secondary ossification centers in the hind-paw of Nlrp2^M+/P−/Z+ offspring (black arrow) appears to underdeveloped in the Nlrp2^M−/P+/Z+ mouse (dashed arrow).

43032_2020_360_MOESM4_ESM

Supplemental video 1: Embryoscope time-lapse microscopy (Orignal 149.7 h, condensed to 28.76 sec) of in vivo fertilized embryos from Nlrp2^tm1a/tm1a mice cultured in vitro. Developmental arrest at two-cell stage (bottom embryo) and no obvious blastocoel cavity formation followed by fragmentation (top embryo) are shown.

43032_2020_360_MOESM5_ESM

Supplemental video 2: Embryoscope time-lapse microscopy (Orignal 161.8 h, condensed to 31.16 sec) of cultured in vitro fertilized embryos from Nlrp2^tm1a/tm1a mice. Morphologically atypical blastocoel cavities are formed in both embryos before fragmentation is observed.