Overexpression of placenta-specific noncanonical imprinted genes causes placental enlargement in intersubspecific hybrid mice

Abstract

Placental enlargement in somatic cell nuclear transfer–derived mice is attributed to biallelic expression of noncanonical (H3K27me3-dependent) imprinted genes owing to loss of imprinting (LOI). Here, we investigated whether a similar mechanism underlies placental enlargement in intersubspecific hybrids between BDF1 (Mus musculus domesticus) and HMI (M. m. castaneus) mice. Quantitative and allelic expression analyses revealed gene-specific LOI in (BDF1 × HMI)F1 placentas: Jade1 (Phf17) and Slc38a4 showed LOI in all placentas regardless of expression levels, whereas Gab1 and Sfmbt2 exhibited LOI only when expression levels were elevated. Notably, Jade1 and Slc38a4 also showed biallelic expression at lower levels in normal-sized (BDF1 × JF1 [M. m. molossinus])F1 placentas. Maternal knockout of Jade1, Slc38a4, Sfmbt2, or the Sfmbt2 miRNA cluster restored monoallelic expression and significantly reduced the weight of (BDF1 × HMI)F1 placentas, indicating that these genes were collectively responsible for placental enlargement in intersubspecific hybrid placentas. Transcriptomic analysis revealed that LOI of noncanonical imprinted genes occurred after implantation. These findings suggest that placental enlargement in (BDF1 × HMI)F1 hybrids is driven by overexpression of multiple noncanonical imprinted genes, resulting from LOI after implantation and additional hybrid-specific, yet unidentified, upregulation mechanisms.

In intersubspecific hybrid mice, placental enlargement is associated with the overexpression of multiple noncanonical imprinted genes resulting from loss of imprinting. Using maternal KO lines, we identified the genes responsible for overgrowth.

Graphical Abstract

Graphical Abstract.

Introduction

The placenta is an essential organ for embryonic development and fetal growth. During embryonic development in mice, the trophectoderm of the blastocyst at embryonic day (E) 3.5 gives rise to the extra-embryonic ectoderm and ectoplacental cone by E5.5 and subsequently develops into the early placenta, which is composed of the labyrinthine and the spongiotrophoblast layers by E9.5 [1]. Normal placentation requires proper gene expression, including both canonical and noncanonical (placenta-specific) imprinted genes [2]. Among the placental abnormalities observed in mice, placental enlargement from mid to late gestation occurs following somatic cell nuclear transfer (SCNT) [3–5]. Recent gene-knockout experiments revealed that this SCNT-associated placental enlargement was caused by the loss of placenta-specific noncanonical imprinting, which depends on maternal histone 3 lysine 27 trimethylation (H3K27me3) [2]. The noncanonical imprinted genes responsible for the placental enlargement include a large miRNA cluster within Sfmbt2 on chromosome 2 (C2MC) [6], Jade1 (Phf17), Gab1, and Sfmbt2 [7], or Slc38a4 alone at late gestation [8], all of which exhibit biallelic expressions due to loss of imprinting (LOI). The fact that loss of noncanonical imprinting causes placental enlargement was confirmed by the analysis of embryonic ectoderm development (EED) knockout embryos, which lack the H3K27me3 imprinting mark [9, 10]. Furthermore, maternal knockout analyses of each noncanonical imprinted gene identified C2MC, Slc38a4, and Gm32885 as genes responsible for placental enlargement [9].

Placental enlargement also occurs in mice through interspecific or intersubspecific crosses between laboratory and wild-derived mice [11–14], or the loss of specific genes such as Esx1 [15] and Plac1 [16]. Interspecific crossbreeding studies using laboratory mice and Mus spretus mice have shown that the placental size was determined by multiple loci clustered on the X chromosome that may act synergistically [11, 12]. Intersubspecific hybrid mice have been generated by crossing laboratory mice with wild-derived mice such as Mus musculus musculus, M. m. castaneus, M. m. molossinus, and M. m. domesticus [17, 18]. Although all of these intersubspecific crosses produce normal-appearing, fertile offspring, only the combination with HMI mice (M. m. castaneus) results in enlarged placentas, which can be further exacerbated by intracytoplasmic sperm injection (ICSI) [13]. They exhibit the expansion of the spongiotrophoblast layer and the interdigitating boundary between the labyrinthine and spongiotrophoblast layers. However, the mechanisms underlying placental enlargement in BDF1 (M. m. domesticus) × HMI hybrid mice remain unclear. In the present study, we investigated the etiology of placental enlargement in (BDF1 × HMI)F1 hybrid mice, focusing on H3K27me3-dependent noncanonical imprinted genes such as Jade1, Gab1, Sfmbt2, Slc38a4, Smoc1, C2MC, and Xist.

Materials and methods

Animal care and ethics

All experiments using mice were approved by the Animal Experimentation Committee at the RIKEN Tsukuba Institute (T2024-EP004) and were performed in accordance with the committee’s guidelines. Mice were maintained with free access to food and water, controlled lighting conditions (daily light from 07:00 to 21:00), and a specific pathogen-free condition.

Mice

C57BL/6 N (B6), DBA/2, and (B6 × DBA/2) F1 (BDF1) strain mice (7–10 weeks old) were purchased from Japan SLC (Shizuoka, Japan). ICR mice were purchased from CLEA Japan (Tokyo, Japan). The wild-derived mouse strains JF1 (M. m. molossinus, RBRC00639) [19] and HMI (M. m. castaneus, RBRC00657) [20] (Figure 1A) were provided by RIKEN BioResource Research Center (BRC). Gab1 (RBRC00440) [21] and Xist knockout (KO) (RBRC01260) [22, 23] mouse lines were provided by RIKEN BRC. Jade1 [9], Sfmbt2 [6], Slc38a4 [24], Smoc1 [9], and C2MC [25] KO lines were generated in-house. To generate intersubspecies maternal KO (mKO) embryos/placentas, mKO B6 female mice were mated with DBA/2 male mice to produce mKO BDF1 female mice, and their collected oocytes were inseminated with HMI spermatozoa. All retrieved fetuses were genotyped by PCR.

Figure 1.

Placental weight and histology in (BDF1 × JF1)F1 and (BDF1 × HMI)F1 hybrid mice. (A) Part of the evolutionary phylogenetic tree of subspecies mouse strains modified from Guénet and Bonhomme, 2003 [18]. Myr, million years ago. (B) PAS-stained images of hybrid placentas. LB, labyrinthine layer; ST, spongiotrophoblast layer; D, decidua. (C) The weights of hybrid placentas. Each gray dot represents the placenta of one fetus. Each bar indicates the mean ± SE. The numbers in the bars denote the number of samples. P was calculated by the Welch t-test. ^***P < 0.001. (D) Area of the labyrinthine and spongiotrophoblast layers in hybrid placentas.

Preparation of mature oocytes

B6, BDF1, and mKO BDF1 female mice were superovulated by injecting 7.5 IU of pregnant mare serum gonadotropin (PMSG, ZENOAQ, Fukushima, Japan); after 48 h, 7.5 IU of human chorionic gonadotropin (hCG, Aska-Pharmaceutical, Tokyo, Japan) was injected. Cumulus–oocyte complexes (COCs) were collected from the oviducts at 15–17 h after hCG injection and used for in vitro fertilization (IVF) or ICSI.

Collection and cryopreservation of epididymal spermatozoa

Epididymides were collected from JF1 and HMI male mice (8–12 weeks old). Mature spermatozoa were frozen using a slight modification of the method developed by Nakagata and Takeshima [26, 27]. The sperm cryopreservation solution (R18S3) contained 18% raffinose (Difco, NJ, USA) and 3% skim milk (Difco). Fat and blood were carefully removed from the epididymides using fine forceps and scissors on filter paper. Approximately 10 incisions were made in each cauda epididymis using fine scissors in 50–80 μL of R18S3 solution. The resulting sperm suspension was then divided into 4 μL aliquots. Each aliquot was placed on top of a 0.25 mL mini-sized plastic straw (Cassou straw; IMV Technologies, L’Aigle, France) and cooled in LN2 vapor for 10 min before being submerged in LN2. The mini-sized frozen straws, containing an average of 4.4 × 10⁵ (JF1) and 2.7 × 10⁵ (HMI) spermatozoa, were stored in a liquid nitrogen tank until use for ICSI or IVF.

In vitro fertilization

Cumulus–oocyte complexes obtained from superovulated BDF1 females were placed in 80 μL droplets of human tubal fluid (HTF) medium [28] containing 1.25 mM reduced glutathione (GSH). Frozen JF1 and HMI spermatozoa were thawed and preincubated for 50 min in 100 μL droplets of HTF containing 0.1 mg/mL polyvinylalcohol (PVA) and 0.4 mM methyl-β-cyclodextrin (MBCD). An aliquot (20 μL) of the sperm suspension was added to the HTF drops containing the COCs to start insemination. After incubation for 3–4 h, morphologically normal fertilized embryos were washed and cultured in CZB medium [29]. All procedures were performed at 37°C under 5% CO₂ in air. The embryos were cultured until they reached morula (at 72 h post-insemination [hpi]) and blastocyst (96 hpi).

Intracytoplasmic sperm injection

Cumulus–oocyte complexes collected from superovulated B6, BDF1, or mKO BDF1 females were treated with 0.1% hyaluronidase (Merck Millipore, Darmstadt, Germany) in CZB medium to remove cumulus cells. The oocytes were placed in fresh CZB medium and incubated at 37°C under 5% CO₂ in air until use for ICSI. The straws containing frozen spermatozoa were removed from liquid nitrogen and retrieved into Hepes-CZB medium. Dispersed spermatozoa were transferred into Hepes-CZB containing 10% polyvinylpyrrolidone droplets on an injection plastic dish and mixed well using a glass capillary. The oocytes (8–10 oocytes per set) were transferred into a Hepes-CZB drop without bovine serum albumin on the injection dish. ICSI was performed on a cooled stage using a piezo-driven micropipette, as described [30, 31]. A head of the single spermatozoon was separated from the tail by applying a few piezoelectric pulses to the head-tail junction, and then, the isolated sperm head was injected into an oocyte. All injection procedures were performed within 2–3 h of oocyte collection. Injected oocytes were kept at room temperature for 15 min for oocyte membrane repair. Only surviving oocytes were cultured in CZB at 37°C under 5% CO₂ in air.

Embryo transfer and cesarean section

Pseudo-pregnant female ICR mice were obtained by mating them with vasectomized ICR males. The presence of a vaginal plug was designated as embryonic day (E) 0.5. Two-cell embryos were transferred into the oviducts of E0.5 pseudo-pregnant ICR females (8–12 weeks old) under anesthesia induced by an intraperitoneal injection of a mixed anesthetic solution (0.3 mg/kg of medetomidine, 4.0 mg/kg of midazolam, and 5.0 mg/kg of butorphanol). On E11.5 and E19.5, the fetuses were collected via cesarean section. Fetal and placental weights were then recorded. To identify the genotypes of the placentas, each fetal tail and placenta was linked by number. The tails were frozen at −20°C until genotyping. The placentas were cut in two: one half was fixed in 10% formalin neutral buffer solution (Muto Pure Chemicals, Tokyo, Japan) until histological examination, and the other was frozen at −80°C until gene expression and RNA-seq analyses.

Genotyping

Genomic DNA was isolated from tails using a Mouse Tail DNA Extraction Kit (101 Bio, CA, USA). PCR was performed using TaKaRa Ex Taq Hot Start Version (Takara Bio, Shiga, Japan). The primers used are described in Supplementary Table S1.

Placental histology

The fixed placentas were embedded in paraffin blocks, which were cut into 4 μm-thick sections. Periodic acid Schiff (PAS) staining was performed using sections. The areas of the labyrinthine and spongiotrophoblast layers of the stained placentas were measured using Image J (1.54p, National Institute of Health).

Allelic expression analysis with Sanger sequencing

Total RNAs were extracted from frozen placentas using TRIzol reagent (Thermo Fisher Scientific, Waltham, MA, USA) and RNeasy Mini kits (QIAGEN, Hilden, Germany). Complementary DNA (cDNA) was synthesized from mRNA using SuperScript IV first-strand synthesis system (Thermo Fisher Scientific). cDNA samples were amplified by PCR using primers targeting the DNA sequences listed in Supplementary Table S1, and subsequently sequenced by Sanger sequencing using the same PCR primers. Allelic expression was determined by single-nucleotide polymorphisms (SNPs) between B6 (M. m. domesticus; maternal allele) and JF1 (M. m. molossinus; paternal allele) or HMI (M. m. castaneus; paternal allele) strains. The SNP positions and allelic bases of each gene are listed in Supplementary Table S2.

Quantitative RT-PCR

cDNA samples were the same as those used in the allelic expression analysis. mRNA expression levels were quantified using a QuantStudio 7 flex system with the PowerUp SYBR Green Master Mix. The Ct values were normalized against those of beta-actin. The relative quantifications of the (BDF1 × JF1) and (BDF1 × HMI)F1 hybrid placentas were calculated against the expression levels of (BDF1 × BDF1)F1 placentas. The primer sequences are shown in the Supplementary Table S1.

RNA-seq library preparation using embryos

In vitro fertilization–derived intersubspecific hybrid embryos were collected at the morula (72 hpi) and blastocyst (96 hpi) stages and washed three times in phosphate-buffered saline (PBS) containing 0.05% bovine serum albumin (BSA) before snap-freezing in PCR tubes (eight embryos per tube). Total RNAs were purified from frozen embryos, reverse transcribed, and amplified using SMART-Seq v4 kits (Takara Bio). Sequencing libraries were prepared with the SMART-Seq Library Prep Kit (Takara Bio) and Unique Dual Index Kit (Takara Bio). After purification of the amplified libraries using AMPure XP beads (Beckman Coulter, Brea, CA, USA), the quantity and quality of sequence libraries were determined by using Quantus Fluorometer (Promega, Madison, WI, USA) and TapeStation (Agilent Technologies, Santa Clara, CA, USA). The libraries were pooled and sequenced on an Illumina NovaSeq X as 150 bp paired-end reads.

RNA-seq library preparation of placentas

In E11.5 placentas, the decidua from the maternal tissue was removed before RNA extraction. Total RNAs were extracted from frozen placentas with TRIzol reagent (Thermo Fisher Scientific) and RNeasy Mini kits (QIAGEN), and then libraries were constructed using the TruSeq stranded mRNA Library Prep Kit (Illumina, San Diego, CA, USA), according to the manufacturer’s instructions. The libraries were sequenced on an Illumina NovaSeq6000 as 100 bp paired-end reads.

RNA-seq data analysis for allelic quantification

Raw sequencing reads were trimmed and filtered using fastp (v0.20.1) [32] to remove adapter sequences and low-quality reads. The high-quality reads were mapped with HISAT2 (v2.2.1) [33] to an N-masked version of the mouse reference genome (mm10) for JF1 or HMI SNPs. The mapped reads were separated into two alleles (C57BL/6 and JF1 or HMI) using SNPSplit (v0.5.0) [34] followed by gene reads counting by using featureCounts (v2.0.6) [35]. The mapping data were visualized using the Integrated Genome Browser (IGV, v2.16.2; https://software.broadinstitute.org/software/igv/). RNA quantification was performed using allele-specific read counts obtained from SNP split analysis. Read counts aligned to each allele were used to calculate allelic expression ratios.

Western blotting

E19.5 intersubspecific hybrid placental tissues were homogenized in Laemmli Sample Buffer (62.5 mM Tris–HCl, pH 6.8, 10% glycerol, 1% lithium dodecyl sulfate, 2.5% β-mercaptoethanol, and 0.005% bromophenol blue) and heated at 95°C for 5 min. The supernatants were separated using 12% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene difluoride (PVDF) membranes using a Trans Blot-Turbo transfer system (Bio-Rad, Hercules, CA, USA). Membranes were blocked with EzBlock Chemi solution (ATTO, Tokyo, Japan) for 1 h at room temperature, incubated with primary antibodies in PBS containing 0.05% Tween20 (PBS-T; MP Biomedicals, Irvine, CA, USA) and 10% EzBlock Chemi solution at 4°C overnight. After washing twice (10 min each) with PBS-T, the membranes were incubated with secondary antibodies in PBS-T containing 10% EzBlock Chemi solution for 1 h at room temperature. After washing twice with PBS-T, the protein bands were visualized using Amersham ECL Prime Western Blotting Detection Reagent (Cytiva, Marlborough, MA, USA) and scanned using a C-DiGit Blot Scanner (LI-COR Biotechnology, Lincoln, NE, USA) and Image Studio Digits software (version 5.2, LI-COR Biotechnology). The following antibodies were used: mouse anti-SLC38A4 (Santa Cruz, Dallas, TX, USA, 1:2500), goat anti-ACTIN (Santa Cruz, 1:10 000), horseradish peroxidase (HRP)–conjugated donkey anti-mouse IgG (Millipore Sigma, Burlington, MA, USA, 1:50 000), and HRP-conjugated donkey anti-goat IgG (1:50 000; Millipore Sigma). Band intensities were measured using ImageJ (v1.54p, NIH).

Statistical analysis

All quantitative data are represented as means ± SE. Statistical analyses were performed in R (v4.3.2). Normality of the data was assessed using the Shapiro–Wilk test. Because the data followed a normal distribution, significant differences (P < 0.05) between groups were determined using the Welch t-test.

Results

Placental histology of intersubspecific hybrid mice

First, we analyzed the histological features of (BDF1 × HMI)F1 placentas by comparing them with those of (BDF1 × JF1)F1 placentas, which are also intersubspecific but do not exhibit placental enlargement (Figure 1A and B). The average weight of (BDF1 × HMI)F1 placentas was significantly higher than that of (BDF1 × JF1)F1 placentas (0.27 ± 0.02 g vs. 0.12 ± 0.005 g, P = 8.107e-08, Figure 1C). The fetal side of the mouse placenta is mainly composed of two layers: the labyrinthine layer and the spongiotrophoblast layer. Measurement of the average area of each layer revealed that both the labyrinthine and spongiotrophoblast layers contributed to the enlargement of (BDF1 × HMI)F1 placentas (Figure 1B and D).

Expression analysis of placenta-specific, noncanonical imprinted genes in hybrid placentas

Because LOI of placenta-specific noncanonical imprinted genes occurs in SCNT placentas and their biallelic expression contributes to placental enlargement, we quantified the expression levels of Jade1, Gab1, Sfmbt2, Slc38a4, and Smoc1 in (BDF1 × HMI)F1 and control (BDF1 × JF1)F1 placentas with Quantitative RT-PCR. The average expression levels of Jade1, Gab1, Sfmbt2, and Slc38a4 in (BDF1 × HMI)F1 placentas were significantly higher than those of (BDF1 × JF1)F1 placentas (P < 0.01 [fold-change 1.3] in Sfmbt2; P < 0.001 [1.6, 1.5, and 4.4] in Jade1, Gab1 and Slc38a4; Figure 2A). Increased SLC38A4 expression was also confirmed using western blotting (P = 0.06 [1.5], Supplementary Figure S1). We next examined putative allelic expression patterns of noncanonical imprinted genes in hybrid placentas using SNPs between laboratory (BDF1) and wild-derived (HMI and JF1) mice with Sanger sequencing. Jade1 and Slc38a4 showed biallelic expression in all samples in both (BDF1 × HMI)F1 and (BDF1 × JF1)F1 placentas, regardless of expression levels (Figure 2A). Gab1 and Sfmbt2 showed normal paternal expression in many hybrid placentas, but they occasionally displayed biallelic expression specifically in (BDF1 × HMI)F1 placentas, particularly when their expression levels were high (Figure 2A). Smoc1 showed stochastic disruptions in the allelic expression pattern, with a tendency for biallelic expression to be associated with higher expression in both BDF1 × HMI)F1 and (BDF1 × JF1)F1 placentas (Figure 2A). Furthermore, we validated and quantified these results precisely by performing allelic expression analysis using RNA-seq data from both hybrid placentas. This analysis also confirmed the gene-specific allelic expression patterns observed in the Sanger sequencing results (Figure 2B and Supplementary Figure S2A and B).

Figure 2.

Expression levels and allelic expression patterns of Jade1, Gab1, Sfmbt2, Slc38a4, and Smoc1 in hybrid placentas. (A) Comparisons of mRNA expression levels of the five genes between (BDF1 × JF1)F1 and (BDF1 × HMI)F1 placentas (left). Gray dots denote a placenta from one fetus. Error bars show the mean ± SE. The numbers in the graph represent the number of samples. P was calculated by the Welch t-test. ^**P < 0.01, ^***P < 0.001. The expression levels and allelic expression patterns of the five genes in each sample (right). The black and red asterisks denote putative monoallelic and biallelic expressions, respectively. Allelic expression patterns were identified using single-nucleotide polymorphisms (SNPs). (B) Allele-specific read counts of Jade1, Gab1, Sfmbt2, Slc38a4, and Smoc1 in hybrid placentas. Read counts were determined by RNA-seq analysis. Red and blue bars indicate maternal (Mat) and paternal (Pat) allele–derived reads, respectively. Error bars show the mean ± SE. The numbers in the graph represent the number of samples. The read count for each replicate is shown in Supplementary Figure S2A and B.

Expression analysis of other candidate genes responsible for placental enlargement in hybrid placentas

In addition to placenta-specific noncanonical imprinted genes, several other genes are associated with placental enlargement. Esx1 is downregulated in the early developmental stages of interspecific placental enlargement [36]. Plac1 expression is dysregulated in enlarged placentas of cloned and interspecific hybrid mice [37, 38]. Therefore, we quantified the expression levels of Esx1 and Plac1 to determine whether they were involved in the enlargement of intersubspecific placentas. The expression levels of Esx1 in hybrid placentas at early and late gestation were not significantly different between the (B6 × JF1)F1 and (B6 × HMI)F1 placentas (E11.5: P = 0.51; E19.5: P = 0.74). On the contrary, although Plac1 expression levels in hybrid placentas at E11.5 were not significantly different between (B6 × JF1)F1 and (B6 × HMI)F1 placentas (P = 0.46), the levels in (B6 × HMI)F1 placentas at late gestation at E19.5 were significantly increased compared to those of (B6 × JF1)F1 (P = 0.04). These results suggest that Plac1 might have increased size, as reported in a previous Plac1 overexpression experiment that caused placental enlargement [16] (Supplementary Figure S3).

Identification of genes responsible for placental enlargement in hybrid placentas by maternal allele knockout experiments

To determine which noncanonical imprinted genes are responsible for placental enlargement in (BDF1 × HMI)F1 mice, we produced genetically identical (BDF1 × HMI)F1 placentas carrying a mKO of each noncanonical imprinted gene. If the gene was responsible for placental enlargement, its mKO would ameliorate the placental abnormality (Figure 3A). We prepared mKO placentas for the five coding imprinted genes and two noncoding imprinted genes (C2MC and Xist). Our KO design allowed us to distinguish the effect of the host Sfmbt2 gene and its miRNA cluster, C2MC, on placental enlargement [6]. We found that the average weights of Jade1 mKO, Sfmbt2 mKO, C2MC mKO, and Slc38a4 mKO placentas were significantly reduced (P < 0.05 in Sfmbt2 [0.216 ± 0.01 g], P < 0.01 in C2MC [0.21 ± 0.01 g], P < 0.001 in Jade1 [0.159 ± 0.01 g] and Slc38a4 [0.175 ± 0.01 g]) compared to those of wild-type (WT) placentas (0.258 ± 0.01 g) from the same littermates (Figure 3B and C). In contrast, there were no significant differences in placental weights of Gab1 mKO (P = 0.4 [0.239 ± 0.02 g]), Smoc1 mKO (P = 0.2 [0.234 ± 0.01 g]) and Xist mKO (P = 0.7 [0.271 ± 0.03 g]) placentas (Figure 3B and C). We also confirmed that the expression levels of each knockout gene in Jade1 mKO, Gab1 mKO, Sfmbt2 mKO, Slc38a4 mKO, and Smoc1 mKO placentas were significantly decreased compared to those of WT placentas (P < 0.05 in Sfmbt2 mKO and Smoc1 mKO; P < 0.01 in Jade1 mKO; P < 0.001 in Gab1 mKO and Slc38a4 mKO, Supplementary Figure S4). Consistently, Jade1 mKO, Sfmbt2 mKO, C2MC mKO, and Slc38a4 mKO placentas had significantly reduced labyrinthine and/or spongiotrophoblast layers compared to those of WT placentas (Figure 3D). These findings indicate that Jade1, Sfmbt2, C2MC, and Slc38a4 were responsible for placental enlargement in (BDF1 × HMI)F1 mice. Unexpectedly, fetal weights in Jade1 mKO, Sfmbt2 mKO, C2MC mKO, and Slc38a4 mKO mice remained unchanged compared to WT mice despite reductions in placental weights (Supplementary Figure S5), indicating that fetal overgrowth in (BDF1 × HMI)F1 hybrids was not directly caused by placental hyperplasia resulting from LOI.

Figure 3.

Placental enlargement was significantly improved in the Jade1 mKO, Sfmbt2 mKO, Slc38a4 mKO, and C2MC mKO mice. (A) Scheme showing the experimental process to produce mKO placentas. GeneX includes Jade1, Gab1, Sfmbt2, Slc38a4, Smoc1, C2MC, and Xist. HMI sperm were injected into Jade1^Δm/+, Gab1^Δm/+, Sfmbt2^Δm/+, Slc38a4^Δm/+, Smoc1^Δm/+, C2MC^Δm/+, and Xist^Δm/+ BDF1 oocytes using ICSI. Next, two-cell embryos cultured overnight were transferred into pseudo-pregnant recipient female mice (embryo transfer, ET). On day 19.5, fetuses and placentas were retrieved with a cesarean section. The genotype of each placenta was determined by PCR using fetal tail DNA. Average weight of placentas was compared between WT and mKO. (B) PAS-stained images of placentas. Scale bar = 1 mm. LB, labyrinthine layer; ST, spongiotrophoblast layer. (C) Placental weights of WT (BDF1 × HMI)F1, Jade1, Gab1, Sfmbt2, Slc38a4, Smoc1, C2MC, Xist mKO, and (BDF1 × JF1)F1. Gray dots denote a placenta from one fetus. Error bars show the mean ± SE. The numbers in the bars represent the number of samples. P was calculated using the Welch t-test. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001. (D) Areas of the labyrinthine (LB) and spongiotrophoblast (ST) layers in hybrid placentas. P was calculated using the Welch t-test. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001.

We then histologically localized placental cells expressing the five noncanonical imprinted genes in C57BL/6 term placentas (E18.5) using the Spatial Transcript Omics DataBase (STOmics DB; https://db.cngb.org/stomics/stamp/) [39]. Jade1 and Slc38a4 were expressed in both the labyrinthine (syncytiotrophoblast cell I) and spongiotrophoblast (spongiotrophoblast cell) layers, whereas Sfmbt2 was predominantly expressed in the spongiotrophoblast layer (Figure 4). These spatial patterns of gene expression were consistent with the results of the mKO placental analysis above.

Figure 4.

Expression localization of the five imprinted genes in E18.5 placenta of C57BL/6 mouse. Localization of cells expressing Jade1, Gab1, Sfmbt2, Slc38a4, and Smoc1 in mouse placenta. These data were obtained using the Spatial Transcript Omics DataBase (STOmics DB: https://db.cngb.org/stomics/) [39]. GC, glycogen cell; SpT, spongiotrophoblast cell; SynT, syncytiotrophoblast cell.

Loss of noncanonical imprinting in intersubspecific hybrid embryos/placentas occurs after implantation

After implantation, maternal H3K27 imprinting marks are replaced by DNA methylation marks, which ensure the paternal expression of noncanonical imprinted genes during placentation. Therefore, we examined whether loss of noncanonical imprinting in (BDF1 × HMI)F1 placentas was inherited from preimplantation development or arose de novo after implantation. We performed RNA-seq and allelic quantification of intersubspecies hybrid embryos, including morulae, blastocysts, and E11.5 placentas, using SNPsplit. Noncanonical imprinted genes exhibited a normal paternal expression at the morula and blastocyst stages in (BDF1 × HMI)F1 embryos as well as (BDF1 × JFI)F1 embryos (Figure 5A and B, Supplementary Figure S6). In contrast, at E11.5 placentas, we confirmed that noncanonical imprinted genes, including Jade1, Gab1, Slc38a4, and Smoc1, showed biallelic expression or reduced paternal bias (increasing maternal allelic expression) (Figure 5A and B). Indeed, the paternal allelic expression ratios for maternal alleles of these genes in E11.5 placentas were significantly reduced compared with those in preimplantation embryos (BDF1 or B6 × JF1: P < 0.01 in Jade1 and Smoc1, P < 0.001 in Gab1 and Slc38a4; BDF1 or B6 × HMI: P < 0.01 in Jade1, P < 0.001 in Slc38a4, Figure 5C). These findings suggest that loss of noncanonical imprinting occurred after implantation.

Figure 5.

Dynamics of H3K27me3 imprinting during the embryonic developmental process of hybrid pre-implantation embryos and E11.5 placentas. (A) Genome browser view of H3K27me3-dependent noncanonical imprinted genes in (BDF1 [B6] × JF1)F1 and (BDF1 [B6] × HMI)F1 hybrid morulae, blastocysts, and E11.5 placentas. In this figure, we used IVF-derived embryos (eight embryos per batch) instead of ICSI-derived embryos; placental enlargement also occurred in IVF-derived placentas (Supplementary Figure S6). Mat: maternal, Pat: paternal. RefSeq gene tracks were derived from IGV. (B) Heat map showing parental allele-specific gene expression of H3K27me3-dependent imprinted genes. (C) Allelic expression ratios (pat/mat) of Jade1, Gab1, Sfmbt2, Slc38a4, and Smoc1 in intersubspecific hybrid preimplantation embryos (morula and blastocyst) and E11.5 placentas. Pat, paternal allelic expression; Bi, bi-allelic expression. P was calculated by the Welch t-test. ^**P < 0.01, ^***P < 0.001.

Discussion

In this study, we identified that overexpression of noncanonical imprinted genes as the cause of placental enlargement in (BDF1 × HMI)F1 mice. Of five genes examined, four genes, Jade1, Gab1, Sfmbt2, and Slc38a4, were highly expressed in (BDF1 × HMI)F1 compared to (BDF1 × JF1)F1 placentas. Furthermore, these noncanonical imprinted genes showed biallelic expression owing to LOI in a gene-specific manner. Consistent with this, mKO experiments revealed that Jade1, Slc38a4, Sfmbt2, and C2MC were involved in the placental enlargement. Unexpectedly, Jade1 and Slc38a4 were also biallelically expressed in normal-sized (BDF1 × JF1)F1 placentas, indicating that LOI alone is insufficient, and other unknown mechanisms specifically drive overexpression of noncanonical imprinted genes in (BDF1 × HMI)F1 placentas. Based on these results, the dysregulation of noncanonical imprinted genes in intersubspecific placentas can be categorized into the following categories: Jade1 and Slc38a4 showed LOI in all placentas regardless of expression levels, whereas Gab1 and Sfmbt2 exhibited LOI only when expression levels were elevated, and Smoc1 displayed sporadic LOI independent of expression levels.

Notably, such mechanisms have no effect on (BDF1 × JF1)F1 placentas of normal size. A recent study using intersubspecific embryonic stem cell lines (ESCs) from crosses between laboratory mice (B6) and wild-derived mice (MSM [M. m. molossinus], PWK [M. m. musculus], and HMI) suggested the involvement of putative cis- and trans-elements in modifications of the gene expression profiles, depending on the combination of the parental strains [40].

The histological characteristics of enlarged placentas in (BDF1 × HMI)F1 mice were similar to those of SCNT- and Eed-knockout placentas lacking the noncanonical imprinting mark of H3K27me3 [9]. Recent studies have revealed that SCNT-associated placental enlargement is caused by biallelic expression of C2MC, Slc38a4, Sfmbt2, Jade1, and Gab1, although the combination of responsible genes varied with the study [6–8]. Therefore, overexpression of C2MC, Jade1, and Slc38a4 commonly causes placental enlargement in SCNT and (BDF1 × HMI)F1 mice. However, SCNT-associated placental enlargement is largely attributable to the expansion of the spongiotrophoblast layer, in contrast to (BDF1 × HMI)F1 placental enlargement that involved both the trophoblast and labyrinthine layers. In SCNT-derived placentas, the labyrinthine layer shows severely impaired development, leading to irregularly arranged fetal capillaries [5–7, 41]. SCNT placentas likely possess mechanisms independent of noncanonical imprinting loss.

We showed that normal paternal expression of the noncanonical imprinted genes was maintained in preimplantation (BDF1 × HMI)F1 embryos but was lost by E11.5 placentas, suggesting the LOI occurs shortly after implantation. Chen et al. reported that noncanonical imprinting memory is maintained after implantation via conversion from H3K27me3 to DNA methylation marks by the de novo DNA methyltransferases [42]. It is important to note that at least Jade1 and Slc38a4 lost the imprinting memory in both JF1 and HMI hybrid placentas. Therefore, misregulated conversion of histone methylation to DNA methylation likely occurs more frequently in intersubspecific hybrids than previously assumed. Further molecular analysis may reveal the mechanisms underlying intersubspecific incompatibilities in imprinting regulation.

Intersubspecific hybrid mice (e.g., B6 × CAST/Ei) have been widely used as models for allele-specific gene expression analysis based on SNPs. For example, they have been employed in studies of DNA methylation [43], cis- and trans-regulatory elements [44], allelic expression [45], and chromatin accessibility [46]. In this study, RNA-seq allelic expression analysis revealed that H3K27me3-dependent, noncanonical imprinting existed in preimplantation embryos but was lost after implantation in intersubspecific hybrid placentas. Studies using intersubspecific hybrid placentas not only provide epigenetic insights into placental development but also reveal novel mechanisms of subspecies genomic incompatibility that contribute to the loss of noncanonical imprinting.

Author contributions

AO and KI conceived and designed the study. ST, AO, and KI wrote the manuscript. ST generated WT and mKO ICSI placentas and performed their analysis. ST prepared RNA-seq libraries from embryos. NW prepared RNA-seq libraries from E11.5 WT placentas. ST and SF conducted allelic expression analysis of RNA-seq data. AH maintained the wild-derived mouse strains and cryopreserved their sperm. KM, SM, and KI established the KO mouse lines.

Conflict of interest: The authors have declared that no conflict of interest exists.

Data availability

The data underlying this article are available in the DDBJ BioProject database (PRJDB38525).