Tissue-Specific and Minor Inter-Individual Variation in Imprinting of IGF2R Is a Common Feature of Bos taurus Concepti and Not Correlated with Fetal Weight

The insulin-like growth factor 2 receptor (IGF2R) is essential for prenatal growth regulation and shows gene dosage effects on fetal weight that can be affected by in-vitro embryo culture. Imprinted maternal expression of murine Igf2r is well documented for all fetal tissues excluding brain, but polymorphic imprinting and biallelic expression were reported for IGF2R in human. These differences have been attributed to evolutionary changes correlated with specific reproductive strategies. However, data from species suitable for testing this hypothesis are lacking. The domestic cow (Bos taurus) carries a single conceptus with a similar gestation length as human. We identified 12 heterozygous concepti informative for imprinting studies among 68 Bos taurus fetuses at Day 80 of gestation (28% term) and found predominantly maternal IGF2R expression in all fetal tissues but brain, which escapes imprinting. Inter-individual variation in allelic expression bias, i.e. expression of the repressed paternal allele relative to the maternal allele, ranged from 4.6−8.9% in heart, 4.3−10.2% in kidney, 6.1−11.2% in liver, 4.6−15.8% in lung and 3.2−12.2% in skeletal muscle. Allelic bias for mesodermal tissues (heart, skeletal muscle) differed significantly (P<0.05) from endodermal tissues (liver, lung). The placenta showed partial imprinting with allelic bias of 22.9−34.7% and differed significantly (P<0.001) from all other tissues. Four informative fetuses were generated by in-vitro fertilization (IVF) with embryo culture and two individuals displayed fetal overgrowth. However, there was no evidence for changes in imprinting or DNA methylation after IVF, or correlations between allelic bias and fetal weight. In conclusion, imprinting of Bos taurus IGF2R is similar to mouse except in placenta, which could indicate an effect of reproductive strategy. Common minor inter-individual variation in allelic bias and absence of imprinting abnormalities in IVF fetuses suggest changes in IGF2R expression in overgrown fetuses could be modulated through other mechanisms than changes in imprinting.


Introduction
The multifunctional mannose 6 phosphate/insulin-like growth factor 2 receptor (M6P/IGF2R), hereafter referred to as IGF2R, mediates endocytosis and subsequent clearance or activation of a variety of ligands involved in the regulation of cell growth and motility, including insulin-like growth factor 2 and transforming growth factor b [1,2].
The fundamental role of Igf2r in prenatal growth regulation [18][19][20] suggests that quantitative variation in imprinting could affect phenotype via gene dosage effects [21]. Indirect evidence for such an effect was obtained in the sheep model where fetal overgrowth induced by embryo culture was associated with hypomethylation at a CpG site in an intronic sequence element (differentially methylated region 2, DMR2) implicated in IGF2R imprinting and down regulated IGF2R expression [22]. However, the imprinting status of IGF2R was not determined in this study [22].
Species differences in imprinting, in particular in the placenta, appear to be linked to differences in reproductive strategies, e.g. litter size, gestational length, maturity of newborns and lifetime reproductive capability [23]. The inconsistent data on IGF2R imprinting in human have been interpreted as evidence for a polymorphic trait [15,17], where the observed minority of imprinted or partially imprinted specimens could signal an evolutionary transition to biallelic expression in the population [17]. It was further hypothesized that differences in imprinting of Igf2r/IGF2R between mouse and human could be a consequence of different reproductive strategies, including competition between multiple fetuses and the shorter gestation period requiring a more efficient placenta in mouse [17,23]. However, comparative data from other species suitable for testing this hypothesis are lacking.
The domestic cow (Bos taurus) has a similar gestation length as human (,280 days), carries a single conceptus with comparable maturity at birth, has a similar lifetime reproductive capability, and shows a conserved IGF2R gene structure with high sequence homology to human [24]. These features make the cow an excellent model for comparisons with human and mouse, but current Bos taurus IGF2R imprinting data is limited to two fetal liver samples of unspecified age [12]. In the present study, we determined the tissue-specific imprinting status of IGF2R in first trimester Bos taurus concepti generated in-vivo or in-vitro. We quantified variation in allele-specific expression bias, i.e. expression of the repressed paternal allele relative to the predominantly expressed maternal allele, determined methylation levels in DMR2, and analyzed relationships between expression bias and fetal phenotype in fetuses with or without overgrowth after in-vitro fertilization (IVF) procedures with embryo culture. Our data show that in all fetal tissues but brain, Bos taurus IGF2R is imprinted and predominantly expressed from the maternal allele as in mouse. However, in contrast to mouse, we detected partial imprinting in placenta that could be related to differences in reproductive strategy between cow and mouse [23]. Contrary to expectations, imprinting and DNA methylation were not affected by IVF with in-vitro embryo culture and there was no correlation between minor variation in inter-individual allele-specific expression bias and fetal weight.

Imprinting of IGF2R in Fetal Tissues and Placenta
Sequencing of bovine IGF2R fragments orthologous to human IGF2R exon 48 revealed a conserved microsatellite with the Bos taurus alleles (TG) 9/10 ( Figure S1). Genotyping of microsatellite alleles in 68 fetuses recovered on Day 80 post insemination (28% term) and of parental samples identified 12 heterozygous concepti with unequivocal maternal and paternal alleles ( Figure 1A). Quantitative microsatellite analyses showed that IGF2R is not imprinted in bovine fetal brain but predominantly expressed from the maternal allele in all other fetal tissues and the placenta. Minor inter-individual variation in allele-specific expression bias, i.e. expression level of the repressed paternal allele, was observed in all tissues. Measured relative to the maternal allele, observed paternal expression ranged from 4.628.9% in heart, 4.3210.2% in kidney, 6.1211.2% in liver, 4.6215.8% in lung, 3.2212.2% in muscle and 22.9234.7% in placenta ( Figure 1B, C).
Low levels of paternal Igf2r expression were previously detected in transgenic and crossbred mouse fetuses [5,6], but could be criticized on grounds of potential Igf2r transgene [5] and subspecies cross effects [25]. Measurements of paternal IGF2R expression in bovine fetal tissues clearly confirm the 5-10% of paternal expression estimated in mouse [5]. Furthermore, our data are consistent with the observation in mouse embryonic stem cells that establishment of imprinted gene expression increases transcription from the maternal allele up to 10-fold while transcription from the paternal allele remains constant [26].
Interestingly, our quantitative analysis of IGF2R imprinting revealed a significant tissue-effect on allele-specific expression bias (ANOVA P,0.001). Pair wise comparisons of least square means showed significant differences (t-tests P,0.05) between tissues originating from different cell lineages ( Figure 1C). Tissues derived from mesoderm (heart, skeletal muscle) showed higher relative expression from the paternal allele than tissues derived from endoderm (liver, lung). Bisulphite pyrosequencing of CpG dinucleotides in DMR2 ( Figure 2A) revealed significant DNA methylation differences between mesodermal (skeletal muscle) and endodermal (liver) tissue at 20 out of 21 surveyed sites (Paired samples t-tests P,0.05-P,0.001) ( Figure 2B). Together with the biallelic expression found in ectoderm derived brain, this cell lineage effect suggested that in cow, similar to mouse [5,6] and most likely sheep [27], imprinted expression of IGF2R in the embryo proper is not established until gastrulation.
Expression from the paternal IGF2R allele was 3-4 fold higher in placenta as compared with fetal tissues originating from the embryo proper (P,0.001) and can be classified as partial imprinting ( Figure 1B, C). Since placenta and embryo proper derive from the two earliest lineages of the embryo, trophectoderm and inner cell mass, respectively, establishment of a trophectoderm-specific imprinted allelic expression bias prior to establishment of imprinted expression in the embryo proper (see above) could explain these differences. Alternatively, the high proportion of expression from the paternal IGF2R allele could indicate a transition from imprinted maternal expression in the first trimester of gestation to biallelic expression in later developmental stages. Temporary imprinting with intermediate states was reported for mouse and human Slc22a3/SLC22A3, which is in the Igf2r/IGF2R imprinted cluster with some shared control elements and mechanisms [8,17,28]. Future analyses of samples from more advanced developmental stages should help clarify this aspect.
Unlike the hemochorial placenta of mouse and human, the synepitheliochorial placenta in bovine presents a strong barrier that prevents mixing of fetal and maternal bloods and tissue [29]. Recent concerns about contamination of fetal placenta with maternal tissue [28,30] nevertheless prompted us to test our sampling procedure for fetal placenta (cotyledon) for potential maternal contamination. We compared high copy number mitochondrial DNA (mtDNA) haplotypes of fetal placenta and maternal tissue from pregnancies with discordant mtDNA haplotypes after embryo transfer [31] and found no evidence for contamination with maternal tissue (not shown). Future studies need to address the possibility that different cell types in bovine fetal placenta [32] could have a different imprinting status and present as partially imprinted when analyzed in one sample.
The imprinted allelic expression bias in bovine placenta is clearly reminiscent of the partial imprinting reported for some fetal and placental samples in human [15,16]. It has been speculated that inconsistent data on imprinting of IGF2R in human might stem from difficulties in interpreting exon 48 microsatellite data due to potential PCR artifacts [33]. However, a more recent report of polymorphic IGF2R imprinting in human term placentae was based on SNP sequencing [17] and thus excluded microsatellite artifacts. Furthermore, in the same study SLC22A2, which is located in the IGF2R imprinted cluster, showed polymorphic imprinting concordant with IGF2R imprinting status [17]. In the cow fetuses of the present study, the paternal microsatellite allele was always the longer allele ( Figure 1A and B). Since Taq DNA polymerase slippage in TG/CA microsatellites favors repeat contraction over expansion by a factor of .14 [34], the bulk of any artifacts, if present, would be shorter than the original allele. Accordingly, we were able to detect a contracted fragment peak, but only when loading very large amounts of sample unsuitable for allele quantification, and found no evidence for repeat expansion ( Figure S2). Thus, the paternal allele cannot be confounded with PCR artifacts in our analysis of paternal contributions to IGF2R expression.

Imprinting of IGF2R after in vitro Embryo Production
Four of the informative fetuses investigated in the present study were generated by in-vitro fertilization (IVF) procedures that involved embryo culture to the blastocyst stage [34]. Two of these IVF fetuses displayed fetal overgrowth (96.0 g and 105.8 g) and two showed normal weights (71.2 g and 89.8 g) based on comparisons with the eight fetuses conceived in vivo (mean 78.0 g, SD 4.49 g; heaviest fetus 86.8 g). Contrary to expectations from IGF2R expression and DMR2 DNA methylation data in overgrown sheep fetuses [22], which could be interpreted as evidence for changes in genomic imprinting [35], allelic expression ratios and DMR2 methylation levels of IGF2R in overgrown IVF, normal weight IVF and in-vivo conceived cow fetuses were similar ( Figure 3A-D). Furthermore, we detected no significant correlations (P.0.10) between minor inter-individual variation in allelic expression bias and fetal weight. This suggested that changes in IGF2R expression in overgrown fetuses after in-vitro embryo culture [22] could be modulated through other mechanisms than changes in imprinting, perhaps through a canonical, transcription factor driven mechanism [21]. The first trimester fetuses used in the present study were in an early stage of fetal overgrowth and we cannot exclude the possibility that a more severe phenotype later in gestation could be correlated with measurable differences in imprinting. However, our findings are consistent with the limited available data on the role of imprinted genes in developmental plasticity which have so far generally failed to provide evidence for substantial relaxation of imprinting to explain altered expression of imprinted genes [21].
In conclusion, our data showed imprinted expression of IGF2R in cow fetuses with the exception of brain as in mouse. The partial imprinting observed in cow placenta could signal an effect of reproductive strategy as proposed by MONK et al. [17] and explored in detail by Frost and Moore [23] to explain differences in IGF2R imprinting between mouse and human. Tissue-specific quantitative variation in imprinting is a common feature of IGF2R expression in normal fetal development. The absence of imprinting abnormalities after IVF with embryo culture and lack of correlations between minor inter-individual variation in allelic bias and fetal weight suggest that changes in IGF2R expression in overgrown fetuses may be modulated through other mechanisms than changes in imprinting.

Fetal Tissue Samples
All experiments involving animals were performed in accordance with the relevant guidelines for the care and use of animals and with approval by the responsible animal welfare authority, the Regierung von Oberbayern. No animals were specifically generated for this study. Samples were obtained from fetuses generated by artificial insemination after estrus cycle synchronization and by in vitro fertilization procedures and embryo transfer as described previously [36]. Fetuses generated by in-vitro fertilization (IVF) with embryo culture were classified as overgrown when their weight exceeded the mean of control fetuses conceived in-vivo by more than four standard deviations. Brain samples were from the upper-left hemisphere of the telencephalon, liver samples from the Lobus hepatis sinister, and skeletal muscle samples from the Musculus gluteus maximus. Lung samples were obtained from the tip of the lungs and heart muscle was collected from the apex of the heart. The kidney sample consisted of one half of a bisected kidney. Cotyledon (Placenta fetalis) was sampled from placentomes close to the fetus. All samples were collected on ice and stored in RNAlater (Ambion, Austin, USA) at 280uC after 24 h incubation at 4uC.

Detection of Microsatellite Polymorphism and Genotyping
Genomic DNA was extracted from parental blood or semen by standard protocols and from fetal liver tissue with the E.Z.N.A. Tissue DNA Mini Kit II (PEQLAB Biotechnologie GmbH, Erlangen, Germany). PCR was performed on 50-100 ng of DNA with 0.4 mM of each specific oligonucleotide primer (59-6-FAM-CAGCTGGTGAAGTCAAAC-39 and 59-AGTCTGA-CATCTCGACTC-39) using 1X PCR buffer, 0.2 mM dNTPs and 1 U Taq DNA polymerase from Invitrogen (Karlsruhe, Germany) in a total volume of 25 ml and the following conditions: 4 min at 94uC, followed by 35 cycles of 30 sec at 94uC, 90 sec at 52uC, 90 sec at 72uC and a final extension of 30 min at 72uC. Genotypes were determined with an ABI 3100 Genetic Analyzer and GeneScan Genotyper version 3.7 NT software (Applied Biosystems, Darmstadt, Germany). Amplicons of detected alleles were cloned (TOPO TA Cloning Kit, Invitrogen) and sequenced (MWG Biotech, Ebersberg, Germany) to confirm sequence identity. The nucleotide sequences of two 437 bp fragments amplified with primers 59-GAGTCAGTAACAGCTGCAG -39 and 59-AATCCAAGTGCAAGGCTG -39 and representing different Bos taurus alleles were deposited in GenBank (AY752984 and AY752985). Additional sequencing with primers

RNA Isolation, Reverse Transcription and Imprinting Analysis
Tissue samples were homogenized in TRIZOL Reagent (Invitrogen) at 1 ml per 50 mg tissue following manufacturer's instructions. The RNA was checked for integrity by agarose gel electrophoresis and quantified by A260 measurement. Twenty-five mg DNase I treated RNA (Trizol TM protocol, Invitrogen) were reverse transcribed in a 25 mL reaction with 400 U of SuperScript III enzyme (Invitrogen), 50 pmol of custom oligo-dT primer (59-GAGATTTTTTTTTTTTTTTTTTTTVN-39), 10 mM DTT (Invitrogen), 0.4 mM dNTPs and 20 U of RNase inhibitor (Promega, Mannheim, Germany). We used 0.25 mL cDNA in PCR reactions as described above to quantify expression of maternal and paternal alleles using the ABI 3100 Genetic Analyzer and GeneScan Genotyper software. We controlled for the possibility of genomic DNA contamination by performing PCR assays on RNA samples that had not been reverse transcribed and all results were negative ( Figure S3).
Analysis of variance and t-tests were performed with SPSS 15. Differences were considered significant at P,0.05.

DNA Methylation Analysis
Analysis of DNA methylation at 21 CpG sites in IGF2R DMR2 was performed using a combination of methylation-sensitive high resolution melting (MS-HRM) and pyrosequencing as described [37]. In brief, genomic DNA was bisulfite-converted using the EZ DNA Methylation-Gold Kit (Zymo Research, Irvine, CA, USA). The region of interest (Figure 2A) was amplified using the EpiTect HRM PCR Kit (Qiagen, Hilden, Germany) in a 25 ml reaction volume at 56uC annealing temperature. Primers were insilico designed with the Pyromark Assay Design software version 2.0 (Qiagen) and tested for unbiased PCR-amplification utilizing artificially generated methylation standards [37]. Forward primer 59-GGGGGTGGTTAAGGTTAGA-39 and biotinylated reverse primer 59-BIO-AACCTTCTCAACACCTTAC-39 amplified a 287 bp segment of bovine IGF2R DMR2 [38]. The amplicon was purified using Wizard SV Gel and PCR Clean-Up System (Promega, Madison, WI, USA) and pyrosequenced on a Pyromark Q24 system (Qiagen) as described [39]. Sequencing primers for pyrosequencing were 59-GTAGGAGTTAGAGTAGAAAT-39 and 59-GAAAGATAAGATATAATAGAG-39.