Prenatal correction of IGF2 to rescue the growth phenotypes in mouse models of Beckwith-Wiedemann and Silver-Russell syndromes.

Beckwith-Wiedemann syndrome (BWS) and Silver-Russell syndrome (SRS) are imprinting disorders manifesting as aberrant fetal growth and severe postnatal-growth-related complications. Based on the insulator model, one-third of BWS cases and two-thirds of SRS cases are consistent with misexpression of insulin-like growth factor 2 (IGF2), an important facilitator of fetal growth. We propose that the IGF2-dependent BWS and SRS cases can be identified by prenatal diagnosis and can be prevented by prenatal intervention targeting IGF2. We test this hypothesis using our mouse models of IGF2-dependent BWS and SRS. We find that genetically normalizing IGF2 levels in a double rescue experiment corrects the fetal overgrowth phenotype in the BWS model and the growth retardation in the SRS model. In addition, we pharmacologically rescue the BWS growth phenotype by reducing IGF2 signaling during late gestation. This animal study encourages clinical investigations to target IGF2 for prenatal diagnosis and prenatal prevention in human BWS and SRS.

INTRODUCTION

Beckwith-Wiedemann syndrome (BWS; OMIM 130650) and Silver-Russell syndrome (SRS; OMIM 180860) are imprinting disorders (Monk et al., 2019) that display fetal overgrowth and fetal growth retardation, respectively. Most BWS symptoms (Brioude et al., 2018) directly or indirectly depend on aberrant growth regulation during gestation: large birth weight; enlarged placenta, tongue, kidney, and liver; and childhood cancers associated with overgrowth (Brioude et al., 2019), such as Wilms tumor of the kidney. A total of 68% of BWS children undergo at least one surgery related to BWS, including tongue or kidney reduction or cancer treatment (McNeil et al., 2002; Naujokat et al., 2019; Style et al., 2018). SRS symptoms (Wakeling et al., 2017) that depend on aberrant growth regulation include small for gestational age (SGA), relative macrocephaly at birth, severe feeding difficulties, and low body mass index.

The opposite growth anomalies of BWS and SRS correspond to the reciprocal genetic and epigenetic (DNA methylation) defects of the same genetic region (Eggermann, 2009). Approximately 80% of BWS and 60% of SRS cases have a molecular defect at chromosome 11p15.5 (Kalish et al., 2014) that hosts two imprinted domains. In domain 1, the paternally expressed insulin-like growth factor 2 (IGF2) gene, regulated by the imprinting control region 1 (ICR1), promotes growth, and in domain 2, the maternally expressed cyclin-dependent kinase inhibitor 1c (CDKN1C), regulated by the imprinting control region 2 (ICR2), inhibits growth. The phenotypes of BWS and SRS can be effectively modeled in the mouse (Chang and Bartolomei, 2020). Cdkn1c mutant mice do not exhibit neonatal overgrowth (Tunster et al., 2011), whereas models with Igf2 misregulation manifest in fetal and neonatal BWS-like overgrowth or SRS-like growth retardation (Singh et al., 2012).

IGF2 encodes the fetal growth factor IGF2. Its RNA is expressed in the chromosome inherited from the father in mice and humans (DeChiara et al., 1991; Giannoukakis et al., 1993). Our knowledge regarding mechanisms that regulate the paternal allele-specific expression of IGF2 is based on genetic studies that focused on the orthologous mouse Igf2-H19 imprinted domain, as summarized in Figure 1A (Singh et al., 2012). IGF2 regulates the supply of nutrients by the placenta and also the demand for nutrients by the fetus (Constância et al., 2002). H19 is a non-coding RNA expressed from the maternally inherited chromosome (Bartolomei et al., 1991; Ferguson-Smith et al., 1993). The H19 RNA releases miR-675 at late gestation to suppress the growth of the placenta just before birth (Keniry et al., 2012). Igf2 and H19 are co-regulated in organs of endoderm and mesoderm origin by their shared enhancers (Leighton et al., 1995b). The genomic imprint of this domain is inherited from the father and consists of CpG methylation along the ICR1 (Thorvaldsen et al., 1998). In the soma, ICR1 functions as a DNA-methylation-sensitive insulator (Figure 1A). We and others reported that four CTCF molecules bind in the unmethylated maternal ICR1 allele to insulate the Igf2 gene from its enhancers, but in the paternal allele, CTCF binding is prevented by DNA methylation, allowing Igf2 transcription (Bell and Felsenfeld, 2000; Hark et al., 2000; Kaffer et al., 2000; Kanduri et al., 2000; Szabó et al., 2000).

Figure 1.

Mouse models for IGF2-dependent BWS and SRS

(A) The structure of the H19/Igf2 domain is depicted in the normal +/+ mice. Imprinted genes expressed in the maternal and paternal chromosomes are shown as red and blue rectangles, respectively. CTCF insulator protein (yellow oval) binds the unmethylated maternal ICR1 allele, insulating the enhancers (green box) from the Igf2 promoter. DNA methylation (black lollipops) in the paternal ICR1 prevents CTCF binding and thus allows Igf2 promoter activation by the distant enhancers.

(B) BWS mouse model: CTCF binding is abolished by point mutations (x) in the maternal ICR1, resulting in biallelic Igf2 expression.

(C) SRS mouse model: the paternal ICR1 is replaced by the mutant chicken beta globin insulator (CHB). The CHB is unmethylated, thus binds CTCF, and results in biallelic enhancer insulation and reduced Igf2 expression.

(D) Rescue model: the compound heterozygote mice are expected to exhibit normalized ICR1 function and Igf2 expression. To generate offspring with four genotypes, including the rescued one, +/CTCFm females were crossed with CHB/+ males.

(E) Representative weanling siblings are shown from the rescue cross, one genotype each at 21 days postpartum (dpp). Three phenotypes and four genotypes were recovered.

(F) Littermate newborn pups are shown from the rescue cross, one genotype each.

(G) Uterus-mate fetuses are shown from the rescue cross at 18.5 dpc.

See also Figure S1.

BWS molecular defects at the ICR1 can be genetic or epigenetic (Brioude et al., 2018). The insulator model (Figure 1A) predicts that in each of these BWS classes (Figures S1A–S1C; Azzi et al., 2014; Bachmann et al., 2017; Brioude et al., 2018; Netchine et al., 2013; Riccio et al., 2009) IGF2 is transcribed from two chromosomes instead of one, resulting in an elevated dose of IGF2 protein, which causes the overgrowth symptoms. In mouse genetic models, an increased dose of Igf2 recapitulated the overgrowth phenotypes of BWS. These models include the Igf2 transgenic mice (Eggenschwiler et al., 1997; Sun et al., 1997), mice with paternal duplication of distal chromosome 7 (PatDup.dist7) (Rentsendorj et al., 2010), and mice that carry mutation in the CTCF binding sites in the maternal ICR1 allele (Szabó et al., 2004). Elevated and biallelic IGF2 RNA expression was found in the tongue and fibroblast of human BWS patients (Weksberg et al., 1993; Reik et al., 1995), supporting the insulator model in humans.

The complex genetic and epigenetic molecular defects found in SRS at the ICR1 (Figures S1D and S1E; Azzi et al., 2014; Begemann et al., 2015; Eggermann et al., 2006; Gicquel et al., 2005; Peñaherrera et al., 2010; Wakeling et al., 2017; Liu et al., 2017; Yamoto et al., 2017) are consistent with reduced IGF2 expression and fetal growth retardation. Mice that carry an Igf2 null mutation are 60% the size of normal siblings (DeChiara et al., 1990). Very small size also characterizes the embryos carrying maternal duplication of distal chromosome 7 MatDup.dist7 (Han et al., 2010) and fetuses that are engineered to exhibit biallelic ICR1 insulation (Lee et al., 2010; Szabó et al., 2002; Engel et al., 2004), due to greatly reduced Igf2 levels. Reduced IGF2 RNA levels in human SRS fibroblasts (Gicquel et al., 2005) and epimutation positive fetuses and placentas (Yamazawa et al., 2008) support the insulator model in humans. Reduced IGF2 protein levels in SRS children (Begemann et al., 2015; Binder et al., 2017) suggest the involvement of IGF2 misregulation in SRS.

The goal of the current work was to address three clinical challenges: (1) IGF2, a known facilitator of fetal growth, has not been targeted in clinical therapy of BWS or SRS; (2) BWS and SRS children are normally diagnosed after birth, when most of the symptoms have already developed and IGF2 levels can no longer be corrected efficiently; and (3) the management of these diseases does not include prenatal intervention. To address these challenges, we used mouse models to derive proof-of principle evidence showing that IGF2-based diagnosis and therapy at the fetal stage could prevent the overgrowth symptoms of BWS and the growth retardation of SRS.

RESULTS

Animal models of BWS and SRS

We previously generated gene-targeted mouse lines to model the molecular features of BWS and SRS cases that are related to the ICR1 (Lee et al., 2010; Szabó et al., 2002, 2004). The BWS model (CTCFm/+, the mother’s genotype is always written first) was created by introducing point mutations in the four CTCF binding sites (CTCFm allele) in the maternal ICR1 allele (Szabó et al., 2004). Igf2 is biallelically expressed in the absence of enhancer insulation (Figure 1B), resulting in fetal overgrowth. The SRS model (+/CHB) (Lee et al., 2010) was generated by replacing the methylated, paternal H19 ICR1 with tandem copies of the unmethylated chicken β-globin insulator region (CHB), which contains two functional CTCF binding sites and insulates Igf2 in the paternal chromosome (Lee et al., 2010; Szabó et al., 2002; Figure 1C). Biallelic ICR1 insulation results in undetectable Igf2 and in turn causes fetal growth retardation.

A double genetic rescue corrects growth anomalies of the BWS and SRS models

We predicted that a paternally inherited CHB allele rescues the overgrowth phenotype of the maternally inherited CTCFm allele. At the same time, a maternally inherited CTCFm allele should rescue the growth retardation of a paternally inherited CHB allele. We generated a rescue mouse model, CTCFm/CHB (Figure 1D), which is a compound heterozygote of the mutant alleles from the BWS model (CTCFm/+) and SRS model (+/CHB). We crossed +/CTCFm mothers with CHB/+ fathers to obtain offspring of four genotypes (Figures 1A–1D). We examined the resulting weanlings at 21 days postpartum (dpp), the newborns at 1 dpp, and fetuses at 18.5 days postcoitum (dpc) (Figures 1E–1G). We found three phenotypes and four genotypes: large (CTCFm/+), small (+/CHB), and normal (+/+ or CTCFm/CHB). The mice of the rescue CTCFm/CHB genotype were indistinguishable from the wild-type siblings except by genotyping.

The weight of the CTCFm/+ and +/CHB weanlings was significantly larger and smaller than that of their +/+ siblings, respectively (Figure 2A). Importantly, the weight of the rescued CTCFm/CHB weanlings was statistically indistinguishable from the +/+ samples, indicating that the growth anomalies of both the BWS and SRS models were eliminated in the compound heterozygotes. We plotted the weight of the pups from birth until weaning age (Figure 2B). The CTCFm/+ pups were larger than their siblings and the +/CHB pups were smaller during postnatal growth. However, the weight of CTCFm/CHB pups remained indistinguishable from the +/+ pups during the first 3 weeks of life.

Figure 2.

Genetic rescue corrects the growth phenotypes of BWS and SRS models

(A) Boxplot shows the weight distribution of pups on the day of weaning, +/+ (n = 12), CTCFm/+ (n = 14), +/CHB (n = 7), and CTCFm/CHB (n = 16). The data were collected from eight litters of +/CTCFm females crossed with CHB/+ males.

(B) Postnatal weight gain of individual pups from two litters is plotted by genotype from birth to weaning at 21 dpp.

(C and D) Boxplots depict the distribution of fetus (C) and placenta (D) weights for early (13.5 dpc) and late (18.5 dpc) fetal stages. The numbers of fetuses and placentas collected from multiple litters were as follows: +/+ (n = 11), CTCFm/+ (n = 11), +/CHB (n = 17), and CTCFm/CHB (n = 16) at 13.5 dpc; and +/+ (n = 10), CTCFm/+ (n = 16), +/CHB (n = 6), and CTCFm/CHB (n = 6) at 18.5 dpc. The significant differences between mutant and +/+ samples are indicated.

See also Figure S2.

The weights of CTCFm/+ fetuses and placentas were significantly larger than those of their +/+ siblings at 18.5 dpc, whereas the +/CHB weights were significantly smaller than the weights of their +/+ siblings at 18.5 dpc (Figures 2C and 2D), which in agreement with our earlier work (Lee et al., 2010; Szabó et al., 2004). The size difference of +/CHB fetuses and placentas was already significantly different from the +/+ samples at 13.5 dpc. Importantly, the weights of the rescued CTCFm/CHB fetuses (Figure 2C) and placentas (Figure 2D) were statistically indistinguishable from those of +/+ samples at 13.5 dpc and at 18.5 dpc. Preliminary evidence suggested that the volume of the amniotic fluid may also differ in the CTCFm/+ versus +/+ and +/CHB/ versus +/+ genotypes but not between the CTCFm/CHB and +/+ genotypes.

To corroborate the results of this genetic rescue experiment, we generated CTCFm/CHB pups by crossing CTCFm/CTCFm females with CHB/+ males. The weights of the resulting CTCFm/CHB fetuses and placentas were significantly smaller than those of the CTCFm/+ fetuses at 18.5 dpc (Figure S2A), and the body weights remained smaller during postnatal growth (Figure S2B).

These results collectively support the hypothesis that the growth anomalies of the IGF2-dependent BWS model (at least the fetal overgrowth, placentomegaly, and macrosomia) and the growth anomalies of the IGF2-dependent SRS model (fetal growth retardation and microsomia) can be prevented by genetic rescue. It is important to note that the growth anomalies of both the BWS and SRS models were normalized in the compound heterozygotes already before birth and remained normalized after birth.

The genetic rescue corrects abnormal organ sizes in the BWS and SRS models

The overgrowth of different organs poses major problems in BWS children, and managing these complications is the major part of medical care in BWS patients after birth (Brioude et al., 2018). Congenital heart disease, including enlarged heart is found in 13%–20% of BWS cases, but the involvement of fetal IGF2 overexpression has not been recognized. Macroglossia is a common symptom in BWS children that often requires surgical reduction of the tongue. ICR1 gain of methylation (GOM) and paternal uniparental disomy (UPD), which both cause IGF2 overexpression, are risk factors for childhood tumors, such as in Wilms tumor of the kidney and hepatoblastoma of the liver (Brioude et al., 2019). We tested whether the aberrant organ sizes can be corrected in the BWS and SRS models by genetically normalizing the ICR1 function.

We crossed +/CTCFm mothers with CHB/+ fathers and assessed the distribution of organ weight in the weaned pups. The brain weight was normal in the CTCFm/+ pups but was significantly smaller in the +/CHB pups than in +/+ pups (Figures 3A and 3B). The heart, kidney, liver, lung, and tongue organs were significantly larger in the CTCFm/+ than in the +/+ pups, whereas these organs were significantly smaller in the +/CHB than in the +/+ pups (Figures 3A and 3B). However, the weight of each organ of the CTCFm/CHB pups was indistinguishable from the +/+ pups (Figures 3A and 3B), except for the tongue that was slightly larger in the CTCFm/CHB pups.

Figure 3.

Organ weights are corrected in the compound heterozygote pups

(A) Boxplot shows the weight distribution of selected organs (or pairs of organs for lungs and kidneys) on the day of weaning. The liver values were plotted on a larger scale as shown to the right. The numbers of fetuses and placentas collected from multiple litters were +/+ (n = 11), CTCFm/+ (n = 11), +/CHB (n = 17), and CTCFm/CHB (n = 16) at 13.5 dpc; and +/+ (n = 10), CTCFm/+ (n = 16), +/CHB (n = 6), and CTCFm/CHB (n = 6) at 18.5 dpc. The significant differences between the different mutant genotypes compared to +/+ samples are marked with the p values.

(B) One litter of pups and some of their organs, namely, tongue (T), brain (B), kidney (K), and heart (H), are shown at weaning according to the genotypes indicated at the top.

(C) Boxplot shows the relative organ/body weights for the brain and the tongue. Other organs did not exhibit significant differences between mutant and +/+ genotypes.

See also Figure S3.

We found that most of the internal organs were proportional to the body weight in both models, that is the relative organ/body weights of CTCFm/+ and +/CHB pups did not differ significantly from the +/+ pups, except for the brain and the tongue (Figure 3C; Figure S3). The brain/body ratio was slightly smaller in the CTCFm/+ pups and it was larger in the +/CHB pups than in the +/+ pups. This result is consistent with the relative microcephaly reported earlier of BWS (Lendvai et al., 1968) and relative macrocephaly of SRS (Wakeling et al., 2017) children. Relative microcephaly is no longer included in the description of BWS (Brioude et al., 2018), but because the BWS model had this phenotype, it was interesting to follow up on it. The relative brain weight of the CTCFm/CHB pups was indistinguishable from the +/+ pups (Figure 3C). The tongue/body ratio was significantly larger in CTCFm/+ than in +/+ pups. However, it was not significantly different between the CTCFm/CHB and +/+ pups. These results collectively demonstrate that the generalized organ weight anomalies found in the heart, kidney, liver, and lung of the BWS and SRS models can be corrected genetically by normalizing ICR1 insulation. In addition, the relative microcephaly and relative macrocephaly growth phenotypes of the IGF2-dependent BWS and SRS models, as well as the relative macroglossia of the BWS model, can also be corrected.

The double genetic rescue normalizes IGF2 levels at the fetal stage

We examined the underlying molecular mechanism of growth normalization in the compound heterozygous mice. Because both the CTCFm and CHB mutations misregulate the imprinted Igf2 and H19 genes, we expected that the misexpressions of these genes are the major cause of the body and organ size aberrations in the mutant genotypes. We collected 11 organs at 18.5 dpc and measured Igf2 and H19 transcript levels by using quantitative reverse-transcription PCR (Figures 4A and S4A) and quantified parental allele-specific transcription using multiplex allelotyping assays (Figures 4B and S4B). We found that the organ sizes at 21 dpp generally corresponded to the Igf2 transcript levels at 18.5 dpc. The heart, kidney, liver, lung, and tongue organs were large, small, and normal-sized, respectively, in the CTCFm/+, +/CHB, and CTCFm/CHB weanlings (Figure 3A), and these organs exhibited 2-fold higher, greatly reduced, and normal Igf2 RNA expression levels, respectively, at 18.5 dpc (Figure 4A). The liver was an exception, as its size was normalized in the rescue model at 21 dpp, but Igf2 transcript levels were not normalized at 18.5 dpc (Figure 4A). Postnatal liver weight is, therefore, partially independent of fetal Igf2 levels. The +/+ organs expressed Igf2 from the paternal allele except the brain, which expressed biallelic Igf2 (Figure 4B), as expected (DeChiara et al., 1991). In agreement with the insulator model, elevated Igf2 transcript levels in the CTCFm/+ fetus organs (Figure 4A) corresponded to biallelic Igf2 transcription (Figure 4B). The brain of +/CHB pups was slightly but significantly smaller than normal (Figure 3A) and corresponded to reduced Igf2 expression in the fetal brain (Figure 4A). The brain sizes were normal in the CTCFm/+ and CTCFm/CHB genotypes (Figure 3A), and Igf2 expression levels were normal (Figure 4A). Relative microcephaly in the BWS model (Figure 3C) was consistent with the loss of imprinting of Igf2 in the body only but not in the brain, where biallelic Igf2 expression was unchanged (Figure 4B).

Figure 4.

Igf2 RNA and IGF2 protein levels are normalized in the genetically rescued fetuses

(A) Relative Igf2/Gapdh transcript levels are depicted in the major organs in the 18.5 dpc fetuses. Uterus-mate fetuses are shown, with one female and one male in each of the four genotypes as indicated by the color code. The average values of the biological duplicates are plotted with the standard deviation.

(B) Allele-specific expression of Igf2 transcription in the major organs of the BWS model and normal 18.5-dpc fetuses. Uterus-mate fetuses are shown, with one female and one male in each genotype. The genotypes are shown to the left. The numbers refer to the percent maternal allele in the total expression determined using multiplex allelotyping assays. Single nucleotide polymorphisms between the 129S1 versus JF1/MsJ inbred mouse lines, inherited from the mothers and fathers, respectively, were used for quantifying parental alleles in the Igf2 transcript. The colors depict paternal-allele-specific bias in blue and biallelic expression in yellow.

(C) IGF2 peptide levels are normalized in the rescued genotype. Western blot results are shown for the IGF2 signaling peptide (IGF2s) in the amniotic fluid and for the IGF2 protein in the amnion, yolk sac, and kidney organs of eight uterus-mate embryos, with one female and one male in each genotype, at 13.5 dpc. The four genotypes are marked at the top. The numbers above the images represent the relative average IGF2 protein values in the mutant compared with the +/+ samples after quantifying the film image by densitometry. Relative IGF2 levels were calculated between IGF2/GAPDH values except for the amniotic fluid. In the absence of cellular protein content, relative values were calculated based on IGF2 concentration per equal volumes of amniotic fluid in each sample.

For additional imprinted genes please see Figures S4, S5, and S6.

H19 transcript levels showed reciprocal patterns to Igf2 transcript levels (Figure S4). For example, the CTCFm/+, +/CHB, and CTCFm/CHB organs had greatly reduced, 2-fold higher, and normal H19 expression levels, respectively, compared to +/+ organs (Figure S4A). H19 was maternally expressed in every +/+ fetal organ but was biallelically expressed in each organ of the +/CHB fetuses (Figure S4B). H19 has no effect on body growth (Leighton et al., 1995a), whereas it slows the growth of the placenta immediately before birth (Keniry et al., 2012). A reduced or doubled level of H19 (Figure S4A) may have contributed to the increased and reduced placenta size in the CTCFm/+ and +/CHB fetuses at 18.5 dpc (Figure 2D).

H19 and Igf2 genes belong to an imprinted gene network (IGN), and H19 long non-coding RNA (lncRNA) is a regulator of a large node of this IGN (Gabory et al., 2009; Monnier et al., 2013). To test whether the ICR1 affects allele-specific transcription beyond the H19/Igf2 domain, we examined additional imprinted genes along distal chromosome 7 (Figure S5) and elsewhere (Figure S6). We found that, except for H19, Igf2, and Igf2as, the allele-specific expression of imprinted genes was undisturbed in the insulator mutants and in the compound heterozygotes. Igf2as, which is paternally expressed in the mouse (Moore et al., 1997), exhibited paternal expression in the +/+ organs except in the brain, where it was biallelic, and also showed biallelic expression in CTCFm/+ organs, similar to Igf2 (Figure S5). These results (Figures S5 and S6) collectively suggest that the effects of the CTCFm or CHB mutations on allele-specific transcription are confined to the H19/Igf2/Igf2as imprinted domain along chromosome 7. This is reassuring for future gene therapy approaches that may target IGF2 or the ICR1 for in utero treatment of BWS and SRS cases.

We found, using western blotting, that the secreted IGF2 (IGF2s) peptide level was elevated 4-fold in the amniotic fluid of the BWS model and was close to undetectable in the SRS model at 13.5 dpc (Figure 4C). Importantly, the level of the IGF2 peptide was corrected to ~1-fold in the amniotic fluid of the compound heterozygous rescued fetuses. The level of the IGF2 precursor was also normalized in the amnion and the kidney. In summary, the BWS and SRS growth phenotypes were rescued in the CTCFm/CHB mice, which displayed corrected Igf2 RNA and IGF2 protein levels (Figures 4A and 4C).

The IGF2-dependent BWS and SRS models are molecularly very similar to human BWS and SRS anomalies that predict IGF2 misexpressions (Figure S1); therefore, such human cases can be predicted to respond to the normalization of IGF2 levels or IGF2 signaling. Even though IGF2 has long been considered the culprit in both BWS and SRS, IGF2 is not targeted in therapy. In human children, BWS and SRS are normally diagnosed and managed only after birth. Next, we present evidence in the mouse that the fetal stage is optimally suited for diagnosis and intervention in IGF2-dependent BWS and SRS because IGF2 activity is specific to the fetal stage.

The fetal stage presents a window of opportunity for diagnosing IGF2-dependent BWS and SRS growth aberrations

We measured Igf2 transcript levels in the organs of the 18.5-dpc fetuses using quantitative reverse-transcription PCR (Figure 5A). We found that Igf2 levels were much higher in the extraembryonic organs (placenta, yolk sac, and the amnion), than in the organs of the fetus body (liver, kidney, tongue, brain, heart, lung, muscle, and stomach). Among fetal organs, the liver exhibited the highest expression of Igf2, which plummeted after birth following a brief peak at 7 dpp and remained low during adulthood (Figure 5B). Igf2 did not exhibit the brief postnatal surge in the kidney and brain. Postnatal downregulation of Igf2 mRNA was also reported in the heart, muscle tongue, and lung (Weber et al., 2001; Constância et al., 2000; Lui et al., 2008). This expression pattern suggests that the IGF2 protein exerts its major growth-promoting effect during fetal development while the extraembryonic organs persist. It also implies that any intervention or diagnosis that targets IGF2 will be successful prenatally but not much longer after birth.

Figure 5.

The fetal stage is best suited for intervention in IGF2-dependent BWS

(A) Igf2 level is the highest in the extraembryonic organs of the fetus. Relative Igf2/Gapdh transcript levels are shown as detected by qRT-PCR in the extraembryonic and embryonic organs at 18.5 dpc using six biological replicates. The average values of replicate samples are plotted with the standard deviation.

(B) Igf2 expression plummets soon after birth. Relative Igf2/Gapdh transcript levels are shown in the liver, kidney, and brain at 18.5 dpc, on postpartum day 3 (n = 2), day 7 (n = 2), day 14 (n = 3), day 21 (n = 4), 3 months (n = 1), and 8 months (n = 4); and in the kidney and brain on days 3, 7, and 14 and at 3 months (ns are as above).

(C and D) Abnormal placenta weight indicates subsequent BWS-like fetus weight anomaly. Growth curves in the BWS model. Fetus (C) and placenta (D) weights are depicted as boxplots for early (13.5 dpc), mid (16.5 dpc), and late (18.5 dpc) fetal stages. The significance of differences between the CTCFm/+ and +/+ genotypes (p values, non-paired two-sided Student’s t test) is marked. The numbers of fetuses and placentas were +/+ (n = 11) and CTCFm/+ (n = 11) at 13.5 dpc; +/+ (n = 24) and CTCFm/+ (n = 12) at 16.5 dpc; and +/+ (n = 20) and CTCFm/+ (n = 19) at 18.5 dpc. (E and F) IGF2 peptide level is diagnostic of BWS in the amniotic fluid. Western blots.

(E) The short, secreted IGF2 peptide (IGF2s) is detected in the amniotic fluid, and the full-length IGF2 protein is found in the organs at 13.5 dpc. GAPDH protein served as loading control.

(F) IGF2 levels are elevated in the amniotic fluid in the BWS model at 16.5 dpc. Uterus-mate samples, one female and one male, are shown from each genotype. The level of IGF2s and IGF2 protein was quantified in the amniotic fluid and in the amnion organ at 16.5 dpc by using densitometry of the film image. The numbers above the images represent relative average values of the CTCFm/+ compared with those of the +/+ samples.

We found that the difference in fetus weights between CTCFm/+ samples and +/+ samples greatly increased between 16.5 dpc and 18.5 dpc (Figure 5C). The placenta weights were already 50% higher in CTCFm/+ samples than in the +/+ samples at 16.5 dpc (Figure 5D). These results show that the aberrant placenta growth in the IGF2-dependent BWS model at 16.5 dpc is indicative of the subsequent fetus overgrowth that occurs between 16.5 and 18.5 dpc. Thus, the late fetal stage represents a window of time when the IGF2-dependent BWS overgrowth may be intercepted.

IGF2 in the amniotic fluid is diagnostic of the SRS and BWS models

We performed western blotting using protein lysates from organs of a 13.5-dpc fetus and found that the full-length IGF2 protein, at ~18 kilodaltons (kDa), was clearly detectable in the amnion, yolk sac, and kidney; whereas the secreted IGF2 peptide (~8 kDa) was abundant only in the amniotic fluid (Figure 5E). Next, we explored the possibility of using IGF2 peptide level in the amniotic fluid as a fetal diagnostic marker in IGF2-dependent BWS and SRS. We found that the amniotic fluid had extremely low IGF2 peptide concentration in the SRS model (+/CHB) at 13.5 dpc (Figure 4C), at the time when the placenta weights indicated the SRS phenotype (Figure 2D). IGF2 peptide level was 4 times higher in the CTCFm/+ than in the +/+ amniotic fluid at 16.5 dpc (Figure 5F), when the placenta weights indicated the BWS phenotype (Figure 5D). These results show that IGF2 peptide levels in the amniotic fluid can be used as diagnostic targets in the SRS and BWS models at 13.5 and 16.5 dpc, respectively.

Pharmacological rescue of the BWS overgrowth phenotype

The results of the genetic rescue experiment were consistent with the idea that correcting IGF2 levels normalizes the growth anomalies in the BWS model. We next explored the possibility of a pharmacological intervention in BWS. IGF2 signaling is mediated by the IGF1 receptor (IGF1R) (Baker et al., 1993; Bergman et al., 2013; Gicquel and Le Bouc, 2006). Picropodophyllin (PPP) is a drug approved by the Food and Drug Administration (FDA) to block IGF2 signaling. It is known to inhibit IGF1R kinase activity (Girnita et al., 2004). We treated CTCFm/+ and +/+ fetuses in utero with 20 mg/kg PPP for 2 days (starting on 16.5 dpc) or 5 days (starting on 13.5 dpc) and measured the weight of the fetuses and placentas at 18.5 dpc. We found a statistically significant reduction in the weight of PPP-treated CTCFm/+ fetuses (Figure 6A) compared with untreated CTCFm/+ controls after 2 days or 5 days of PPP treatment. The weight of CTCFm/+ fetuses was indistinguishable from that of +/+ control fetuses after 5 days of PPP treatment, suggesting that fetus weight was corrected in those mutants by PPP. The weight of +/+ fetuses was also significantly reduced after 2 days or 5 days of PPP but to a lesser extent than the CTCFm/+ fetuses. The weight of CTCFm/+ and +/+ placentas (Figure 6B) did not change significantly after 2 or 5 days of PPP, suggesting that contrary to what we found in the fetus, blocking the IGF1R does not reduce IGF2 activity in the placenta.

Figure 6.

Prenatal pharmacological rescue of the BWS model

(A and B) Prenatal treatment by the IGF1R inhibitor picropodophyllin (PPP) antagonizes the IGF2 overdose and prevents fetal overgrowth in the BWS model fetuses. Boxplots show fetus (A) and placenta (B) weights from PPP-treated and control mothers. PPP was administered orally to CTCFm/+ pregnant dams at 20-mg/kg dose for 2 days, starting at 16.5 dpc CTCFm/+ (n = 15) and +/+ (n = 14), 4 litters, or for 5 days, starting at 13.5 dpc CTCFm/+ (n = 19) and +/+ (n = 16), 4 litters. Control fetuses CTCFm/+ (n = 12) and +/+ (n = 12) from 3 litters were untreated. Adjusted p values are shown to indicate the significant differences of treated versus control samples of the same genotype.

(C–E) Prenatal PPP prevents the overgrowth phenotypes of the weanling pups in the BWS model. Boxplots show the weight distribution of the total body (C), selected organ weights (D), and relative organ/body weights (E) on the day of weaning at 21 dpp. The significant differences between treated CTCFm/+ and control +/+ samples and between treated CTCTm/CHB and control +/+ are marked above horizontal bars. The significance values between treated and control samples of the same genotype are indicated above the boxes.

(F) PPP acts by blocking IGF1R signaling. Western blot of 18.5-dpc kidney samples. The protein lysates of two kidney samples, one female and one male, from each genotype and from each treatment group are displayed. Pan and phosphorylated IGF1R, AKT, and p44/42 MAPK(ERK1/2) antibodies were used. Fetus weights are provided at the bottom in the top panel. GAPDH was the loading control. The signals were quantified from the images, and averages of the duplicates are displayed as ratios of the altered conditions compared with the +/+ untreated samples.

See also Figure S7.

We also addressed the postnatal growth effects of PPP. The body weight of PPP-treated CTCFm/+ pups was significantly smaller than that of the untreated CTCFm/+ pups at the weaning age (Figure 6C) and was indistinguishable from the untreated +/+ pups, consistent with a rescued growth phenotype. Surprisingly, the weight of PPP-treated +/+ pups was not significantly smaller than the untreated +/+ pups (Figure 6C), suggesting that they can catch up in growth after the initial reduction at the fetal stage (Figure 6A). We then examined the organ weight at 21 dpp (Figure 6D). The PPP-treated CTCFm/+ heart was indistinguishable from the heart of the control +/+ pups, showing a complete rescue. The kidney and lung organs of the PPP-treated CTCFm/+ pups were significantly smaller than those of the CTCFm/+ control pups, although they were still slightly larger than those of the control +/+ pups, consistent with a partially rescued growth phenotype in those organs. The tongues of the PPP-treated and control CTCFm/+ pups were not significantly different. However, the difference between CTCFm/+ and +/+ tongues were highly significant (p = 0.0002), but the difference between PPP-treated CTCFm/+ and control +/+ tongues were hardly significant (p = 0.048), consistent with a partially rescued growth phenotype in the tongue as well (Figure 6D). PPP treatment did not result in reduced liver weight at weaning, and this may be related to the brief postnatal peak of Igf2 transcription observed specifically in the liver (Figure 5B) or the independence of liver from IGF1R signaling. Because PPP affected the body weight but not the liver weight, the relative liver/body weights of CTCFm/+ and +/+ pups have increased after PPP (Figure 6E). The head/body (relative microcephaly) and tongue/body (relative macroglossia) phenotypes were corrected by the PPP intervention (Figure 6E). Importantly, none of the organs of PPP-treated +/+ pups changed significantly compared with the untreated controls. These results collectively suggest that most of the BWS-like overgrowth phenotypes can be intercepted by prenatal pharmacological intervention.

To confirm the mechanism of how PPP treatment rescues BWS overgrowth, we tested the effects of PPP on IGF1R signaling in the treated fetuses (Figure 6F). IGF2 binds to IGF1R and induces its autophosphorylation, which activates the Ras-mitogen-activated protein kinase (MAPK) and PI 3-kinase signaling pathways (Hakuno and Takahashi, 2018). In semi-quantitative western blots, we found an increased level of phospho-IGF1R in the CTCFm/+ compared with that of the +/+ fetal kidneys at 18.5 dpc. The level of phospho-IGF1R was reduced in the PPP-treated CTCFm/+ samples compared with that of the control CTCFm/+ samples. In addition, we also found a reduction in phospho-MAPK (ERK1/2) and phospho-AKT in the PPP-treated samples compared with the control samples. These findings support the proposal that CTCFm/+ overgrowth in the kidney is mediated by the IGF2-IGF1R axis and that PPP corrects the overgrowth by interfering with IGF2 signaling.

We did not find any adverse effects of PPP treatment on the mothers. The animals were monitored daily, and no behavioral or physical abnormalities were observed during and after the treatment. All of the PPP-treated mothers nursed and groomed the pups until the weaning day. The weight gain of PPP-treated CTCFm/+ pregnant females was similar to that of the control mothers (Figure S7A), and they became pregnant again within the next 2 months. The number of fetuses was similar for the first (PPP-treated) and second (untreated) pregnancies of these females (Figure S7B). In summary, we demonstrated that the IGF2-dependent BWS fetal overgrowth phenotypes can be prevented by prenatal pharmacological intervention.

DISCUSSION

In this work, we presented evidence by using our IGF2-dependent BWS and SRS mouse models in support of IGF2-based in utero diagnosis and intervention strategies. We showed that (1) the prenatal intervention through genetically correcting IGF2 is successful in these models at the late stage of gestation, (2) a proof-of principle pharmacological method is available for normalizing IGF2 activity and fetal growth in the BWS model, (3) the placenta size anomalies are indicative of subsequent fetus growth anomalies in BWS and SRS, and (4) the secreted IGF2 peptide level in the amniotic fluid is a diagnostic biomarker for prenatal intervention. These results should encourage human studies in which BWS and SRS fetuses that exhibit IGF2 misexpressions are identified directly by measuring IGF2 peptide levels in the amniotic fluid and the IGF2 signaling output is adjusted in utero to normalize growth before birth.

What is realistic in prenatal management, in particular with respect to the window size between prenatal detection of SRS/BWS and successful prenatal treatment? The fetal stage is the optimal time for IGF2-based intervention because of the timing and the location of IGF2 expression. We found that the extraembryonic organs have the highest Igf2 RNA and IGF2 protein levels, and these organs are only available at the fetal stage. We suggest that apart from the role of IGF2 in the placenta, the secreted IGF2 peptide supplied into the fluids surrounding the fetus by the amnion and yolk sac also has an important growth-promoting effect on the fetus. The majority of fetus growth takes place in the third trimester. Therefore, the last 3 months should be the most effective time window for intervention, considering that 5 days of PPP treatment was successful for correcting BWS overgrowth in the mouse.

Importantly, we suggest that the cases that would respond to IGF2 therapy could be identified by directly measuring IGF2 peptide levels. We found diagnostic 4-fold of normal IGF2 peptide levels in the amniotic fluid of the BWS model at 16.5 pc and undetectable IGF2 in the SRS model at 13.5 dpc, before the fetuses showed abnormal growth (Figures 4 and 5). Based on these results in the mouse, we propose that human prenatal diagnosis could be done successfully by measuring IGF2 in the amniotic fluid. IGF2 was detected in human amniotic fluid and exocoelomic fluid by different methods (Blahovec et al., 2001; Chard et al., 1994; Merimee et al., 1984; Nonoshita et al., 1994). The exact IGF2 levels are contradictory in these studies due to the differences in the methodologies used for measuring IGF2 (Blahovec et al., 2001; Chard et al., 1994; Miell et al., 1997; Nonoshita et al., 1994), collecting amniotic fluid (Jauniaux and Gulbis, 2000), and statistics (Verhaeghe et al., 1999). These measurements will need to be revisited using uniform assays and should include verification by western blot.

Who should be screened for IGF2 level in utero? We suggest that every familial BWS and SRS pregnancy case should be screened for IGF2 peptide level. Other cases should be screened if indicated by anatomical and/or molecular defects characteristic of BWS and SRS that include ICR1 (Figure S1). Quantifying IGF2 levels in the amniotic fluid will reveal whether IGF2 levels are abnormal and also how much adjustment is necessary. We showed in the mouse that placenta size is an early indication of fetus size abnormalities in the BWS and SRS models. In a retrospective human study that included 12 BWS pregnancies, fetal ultrasound detected exomphalos, macroglossia, visceromegaly, polyhydramnios, and placentomegaly in 67%, 50%, 83%, 58%, and 83% of the cases, respectively, (Kagan et al., 2015). The median gestational age at the time of examination was 22.6 (range, 19.0–29.7) weeks. Aberrant IGF2 levels should be detected at these times according to the corresponding time points during mouse gestation (Figures 4 and 5). Children born by assisted reproductive technologies have a 4- to 10-fold increased frequency of BWS and SRS (Hiura et al., 2012; Mussa et al., 2017; Hattori et al., 2019), and a fraction of these children have a loss of methylation at ICR1. They may also benefit from IGF2 screening. We are optimistic that IGF2 detection will clearly identify the cases that can be helped by adjusting IGF2 signaling in utero.

Which cases would respond to in utero adjustment of IGF2 signaling? Based on the evidence in the mouse rescue models including the current work, we propose that IGF2-based intervention would be successful in each case that displays aberrant IGF2 levels, even though these may have very different molecular features (Figure S1). However, cases with normal IGF2 levels would not respond to IGF2 therapy even if they displayed specific molecular diagnostic features of BWS or SRS disorders, for example those positive for ICR2 epigenotype only. We predict that BWS cases with IGF2 overdose would respond to reducing the IGF2 signaling output and SRS cases with severely reduced IGF2 would benefit from IGF2 enhancing therapy in utero. We speculate that other imprinted disorders may respond to in utero IGF2 treatment if they displayed characteristic fetal growth anomalies and also diagnostic IGF2 levels. One possibility is Temple syndrome (TS14), which has overlapping symptoms with SRS, and TS14 fibroblasts with molecular abnormalities at 14q32.2 displayed low levels of IGF2 RNA expression (Abi Habib et al., 2019).

Our results suggest successful strategies for normalizing IGF2 signaling in utero. IGF2 affects growth by signaling through IGF1R. A BWS case was recently reported with complex genetic abnormalities, paternal uniparental disomy of chromosome 11 (pat(11)dup), and 15q terminal deletion, resulting in two paternal IGF2 copies and also IGF1R haploinsufficiency (Giabicani et al., 2019). The patient exhibited growth retardation after birth but not before birth. This finding is in agreement with the explanation that, as in the mouse, in humans the relative IGF2:IGF1R dose determines growth rate at the fetal stage but not postnatally. We showed that a specific inhibitor of IGF1R kinase activity, PPP (Girnita et al., 2004), reduced IGF2 signaling and normalized the growth in PPP-treated CTCFm/+ fetuses. Their body and internal organs were normal at the time of weaning, suggesting that the BWS-like overgrowth phenotype can be prevented pharmacologically by adjusting IGF2 signaling in utero. PPP affected the CTCFm/+ fetuses to a greater extent than the +/+ fetuses, perhaps because of the increased capacity of the larger CTCFm/+ placenta for transmitting PPP to the fetus. This result is reassuring for human intervention because twinning is relatively more common in BWS than in the general population (Weksberg et al., 2002; Cohen et al., 2019). Other drugs could also be considered, for example a number of different IGF1R targeting drugs that are FDA approved and/or are currently undergoing clinical trials for treating various diseases including cancer and thyroid eye disease (Dolgin, 2020). We have not rescued the SRS model’s growth restriction phenotype pharmacologically, but such treatments are feasible based on large animal studies. IGF1 delivered into the amniotic fluid of intrauterine growth restriction model lambs increased the birth weight (Eremia et al., 2007; Spiroski et al., 2018). Similarly, fetal intra-amniotic IGF2 treatment would benefit human SRS cases that have severely reduced levels of IGF2. It is important to mention that the IGF2:IGF1 ratio is 12:1 in the human amniotic fluid (Merimee et al., 1984; Nonoshita et al., 1994).

We expect that as a consequence of correcting fetal growth and birth weight, other growth-dependent features can also be ameliorated or prevented by correcting IGF2 levels prenatally. We expect that the rescued newborns would continue to grow normally after birth, as we observed in the mouse rescue experiments. Among BWS symptoms, macroglossia would likely also respond to PPP based on our finding of the partially reduced tongue in weanlings that have undergone PPP treatment in utero. On the other hand, based on our results in the placenta (Figure 6B), we do not expect that placentomegaly could be prevented by PPP. This is consistent with genetic evidence suggesting that IGF1R is not responsible for IGF2 signaling in the placenta (Baker et al., 1993). This is perhaps less of a problem because the placenta is disposed at birth. The association between overgrowth and increased tumor risk in BWS is well documented (Brioude et al., 2019). Wilms tumor and hepatoblastoma display molecular defects consistent with IGF2 overexpression, such as loss of imprinting in the 11p15 region (Honda et al., 2008; Montagna et al., 1994; Rainier et al., 1995; Taniguchi et al., 1995). Overexpression of IGF2 was also detected in those tumors (Akmal et al., 1995; Scott et al., 1985). The tumor risk in BWS is the highest in patients with molecular defects that predict IGF2 overexpression (28% in patients with ICR1 GOM and 16% in patients with upd(11)pat) but low in patients with other molecular defects (2.6% in patients with ICR2 loss of methylation) (Maas et al., 2016; Mussa et al., 2016). We cannot directly address Wilms tumor and hepatoblastoma because these human complications are not observed in the mouse BWS model. Our results (Figure 6D) predict that kidney size will be corrected in the child after correcting IGF2 signaling activity in the fetus, and this may result in a reduced Wilms tumor incidence. On the other hand, liver size may not be prevented by in utero PPP treatment (Figure 6D).

Pharmacological intervention for IGF2 could be continued postnatally in BWS children and may provide an additional benefit of reducing liver size considering that IGF2 has the longest postnatal presence in the liver. The major IGF2 activity takes place before birth in human children as in the mouse (Constância et al., 2000; Weber et al., 2001; Lui et al., 2008; also see Figure 5), but the IGF2 mRNA level has a brief postnatal peak at 2 months of age in humans, after which it declines to approximately one-tenth of the peak level (Li et al., 1996). The liver-specific promoter 1 (P1), specifically found in humans, is proposed as a major source of IGF2 after birth (de Pagter-Holthuizen et al., 1988). We expect that an improvement in the BWS and SRS growth symptoms could be achieved by continuing IGF2-based therapies in the first year of life. However, if the correction takes place only after birth, these children will be born with a birth defect.

In summary, our results provide a proof-of-principle for clinical studies that will establish future prenatal diagnosis and treatment options of IGF2-dependent human BWS and SRS.

STAR★METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Piroska E. Szabó (piroska.szabo@vai.org).

Materials availability

This study did not generate new unique reagents.

Data and code availability

The published article includes all datasets generated or analyzed during this study.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Mouse models

All animal experiments were performed according to the National Institutes of Health Guide for the Care and Use of Laboratory animals, with Institutional Care and Use Committee-approved protocols at Van Andel Institute (VAI). We previously generated the BWS and SRS models using gene targeting in 129SI/ImJ ES cells. The Pgkneo selection cassette was removed by crossing the mutant mice with the Hprt-Cre transgenic mouse line (Tang et al., 2002). The BWS model was created by introducing point mutations in the four CTCF binding sites (CTCFm) in the ICR1 allele (Szabó et al., 2004). To generate the SRS model we replaced the ICR1 with tandem copies of the unmethylated chicken β-globin insulator region (CHB), which contains two functional CTCF binding sites but due to mutagenesis it lacks the boundary elements of the chicken insulator (Lee et al., 2010). Genotyping primers are listed in Table S1. For the present study the CHB mutation was backcrossed ten times into JF1/MsJ inbred strain from The Jackson Laboratory (JAX 003720) to saturate the genome with SNPs, resulting in the CHB.JF1 sub line. Female and male uterus-mate fetuses were used as biological replicates in the experiments that compare RNA levels, allelic expression, IGF2 protein levels or IGF2 signaling between the genotypes. Female and male embryos were combined from multiple pregnancies in experiments to plot growth between genotypes and between treated and control mice. Litter numbers, fetus, newborn, and weanling numbers are provided in the figure legends.

METHOD DETAILS

RNA isolation and quantification

We purified total RNA using RNA-Bee reagent (Tel-Test). To measure individual transcript levels, we prepared cDNA from 400 ng total RNA for qPCR assays as we did earlier (Lee et al., 2010). Contaminating DNA was removed with the DNA-free DNA Removal Kit (Ambion). Reverse transcription was performed using equal amount of RNA using the Superscript III Random Primer Synthesis kit (Invitrogen). The qRT-PCR primer sequences are listed in Table S2.

Sequenom allelotyping

To measure the portion of each parental allele in the total transcript levels we used a multiplex Sequenom (now Agena Bioscience) allelotyping assay as we did earlier (Iqbal et al., 2015). These assays are based on single nucleotide polymorphisms (SNPs) that distinguish between the inbred JF1/MsJ (JF1) and 129S1 (129) mouse strains. Each “unextended” primer (UEP) abuts a SNP in a target transcript, and the incorporating nucleotides differ in molecular mass between the parental alleles. The abundance of the extended UEP is quantified by mass spectrometry. Amplified cDNA samples were spotted onto a 384 SpectroCHIP Array (Agena Bioscience). Automated spectra acquisition was performed in a MassArrayCompact mass spectrometer (Sequenom) using the Spectroacquire program (Sequenom) and was analyzed by MassArray Typer v3.4. RNA-mixing standards were routinely run to verify linear response in measured versus input allele-specific transcription: for example, total RNA from JF1 and 129 embryos was mixed indifferent percent ratios (0:100, 10:90, 30:70, 50:50,70:30, 90:10, and 100:0) before cDNA preparation and Sequenom allelotyping. A 50:50 RNA mix was used for RNA skew correction. The percentage of transcription of each allele in the total expression was calculated at each given SNP. DNA samples were run from each embryo to verify the heterozygous genotype at each SNP. The unextended extend primer (UEP) and PCR primer sequences are listed in Table S3.

Western blot

We obtained amniotic fluid from 13.5 dpc and 16.5 dpc embryos by puncturing the visceral yolk sac and amnion and drawing about 50 μl liquid using a 19-gauge hypodermic needle attached to a 1 mL syringe. The maximum volume of amniotic fluid we harvested depended on the genotype: 50, 75, 75, and 110 μl was drawn from +/CHB, +/+, CTCFm/CHB, and CTCFm/+ genotypes at 13.5 dpc. We removed the cell content by centrifugation (in an Eppendorf centrifuge, 13000 rpm, 2 min at room temperature). We added equal volume of protein sample buffer (62.5 mM Tris, pH 6.8; 20% glycerol; 2% SDS; 5% 2-mercapto-ethanol; 0.05% Bromophenol Blue) according to the volume of the supernatant we collected. We boiled the sample for 10 minutes, and kept it frozen until electrophoresis. We also collected whole amnion in 50 μl protein sample buffer, whole yolk sac in 200 μl of protein sample buffer and one kidney in 50 μl of protein sample buffer. For detecting IGF2 at 13.5 dpc we used 3 μl of amniotic fluid, 2 μl of amnion, 0.5 μl of yolk sac and 10 μl of kidney protein sample per well for electrophoresis in a 4%–20% gradient protein gel. After blotting, the membranes were incubated with an antibody against IGF2 (ABclonal Cat No A2086 1:1000). The signal was visualized by chemiluminescence and exposure to X-ray film. The images were scanned on a ChemiDoc XRS+ imaging system using Image Lab 5.2.1 (Bio-Rad) software and were quantified using the Fiji software.

For detecting IGF1R and its downstream targets we collected fetal kidney samples at 18.5 dpc. One kidney was homogenized in 200 μL of RIPA buffer supplemented with proteinase inhibitor and phosphatase inhibitors. The total lysate was left on ice for 30 minutes and sonicated for 5 minutes. After centrifugation for 15 minutes, the supernatant was transferred to a clean tube and the protein concentration was measured by spectroscopy at 280 nm. 10 μg protein was boiled with protein sample buffer and electrophoresed on a 4%–20% gradient SDS-polyacrylamide gel and blotted onto a PVDF membrane. Blots were first reacted with the anti-phosphorylated receptor antibody, then stripped and reacted with the pan-receptor antibody, then stripped again and reacted with the loading control anti-GAPDH antibody. The antibodies used for this experiment were IGF1 receptor (#3027), Phospho-IGF1 receptor (Tyr1135)(DA7A8) (#3918), AKT (C67E7) (#4691), Phospho-AKT (Ser473) (D9E)(#4060), p44/42 MAPK(Erk1/2)(137F5) (#4695), Phospho-p44/42 MAPK (Erk1/2)(Thr202/Tyr204)(D13.14.4E) (#4370). All six antibodies were from Cell Signaling Technology. The GAPDH antibody was from ABclonal (#AC035).

Treatment of mice with PPP

CTCFm/+ females were mated with +/+ males. They were orally dosed daily with 20 mg/kg PPP (AXL1717, APExBIO Cat No A3209) for two days starting at 16.5 dpc, or for five days, starting at 13.5 dpc using cheerio cereal (General Mills) as carrier. In a pilot experiment we used 40 mg/kg PPP for five days and observed a great reduction in the weight of fetuses, even below the untreated normal fetuses. Fetuses and placentas were collected on 18.5 dpc. To control for litter size and for a potential in utero competition between genotypes, we excluded from the analysis the small litters (less than 6 fetuses) and the litters that yielded only one genotype.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical Methods

The normality of distribution within each group of samples was accessed using a Shapiro-Wilk test. If the P-value of the Shapiro-Wilk test was p > 0.05, then the data within the group was determined as normally distributed; whereas a p < 0.05 indicated that the data follow a non-normal distribution. For determining the significance between two normally distributed groups, a non-paired two-sided Student’s t test was used to calculate P-values. When one or both groups followed non-normal distribution, a non-paired two-sided Wilcox rank sum test was performed to calculate P-values. Adjusted P-values between two or more normally distributed groups were determined by one-way ANOVA followed by a Tukey post hoc test. For multiple comparisons with at least one non-normally distributed group, Kruskal-Wallis was followed by Dunn’s multiple comparison test to obtain Holm-Bonferroni adjusted P -values. Boxplots and statistics were generated using R v3.6.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
IGF1 receptor	Cell Signaling Technology	Cat#3027; RRID:AB_2122378
Phospho-IGF1 receptor (Tyr1135)(DA7A8)	Cell Signaling Technology	Cat#3918; RRID:AB_10548764
AKT (C67E7)	Cell Signaling Technology	Cat#4691; RRID:AB_915783
Phospho-AKT (Ser473)(D9E)	Cell Signaling Technology	Cat#4060; RRID:AB_2315049
p44/42 MAPK(Erk1/2)(137F5)	Cell Signaling Technology	Cat#4695; RRID:AB_390779
Phospho-p44/42 MAPK (Erk1/2)(Thr202/Tyr204)(D13.14.4E)	Cell Signaling Technology	Cat#4370; RRID:AB_2315112
GAPDH antibody	ABclonal	Cat#AC035; RRID:AB_2769863
IGF2 antibody	ABclonal	Cat#A2086; RRID:AB_2764106
Chemicals, peptides, and recombinant proteins
PPP (AXL1717)	APExBIO	Cat#A3209
Critical commercial assays
RNA-Bee reagent	Tel-Test	Cat#CS-104B
Superscript III Random Primer Synthesis kit (Invitrogen™)	ThermoFisher Scientific	Cat#18080051
DNA-free™ DNA Removal Kit (Ambion™)	ThermoFisher Scientific	Cat#AM1906
384 SpectroCHIP Array	Agena Bioscience	Cat#10411
Experimental models: organisms/strains
Mouse: Rr27tm1Pes (CTCFm)	Szabó et al., 2004	RRID:MGI:3046755
Mouse: Rr27tm4Pes (CHB)	Lee et al., 2010	N/A
JF1/MsJ inbred strain	The Jackson Laboratory	JAX: 003720; RRID:IMSR_JAX:003720
Oligonucleotides
Primers for genotyping, see Table S1	This paper and Lee et al., 2010	N/A
Primers for RT-PCR, see Table S2	Lee et al., 2010	N/A
Primers for Sequenom allelotyping, see Table S3	This paper	N/A
Software and algorithms
SpectroAcquire	Agena (formerly Sequenom)	N/A
MassArray Typer v3.4	Agena (formerly Sequenom)	N/A