The Landscape of Genomic Imprinting at the Porcine SGCE/PEG10 Locus from Methylome and Transcriptome of Parthenogenetic Embryos

In mammals, imprinted genes often exist in the form of clusters in specific chromosome regions. However, in pigs, genomic imprinting of a relatively few genes and clusters has been identified, and genes within or adjacent to putative imprinted clusters need to be investigated including those at the SGCE/PEG10 locus. The objective of this study was to, using porcine parthenogenetic embryos, investigate imprinting status of genes within the genomic region spans between the COL1A2 and ASB4 genes in chromosome 9. Whole-genome bisulfite sequencing (WGBS) and RNA sequencing (RNA-seq) were conducted with normal and parthenogenetic embryos, and methylome and transcriptome were analyzed. As a result, differentially methylated regions (DMRs) between the embryos were identified, and parental allele-specific expressions of the SGCE and PEG10 genes were verified. The pig imprinted interval was limited between SGCE and PEG10, since both the COL1A2 and CASD1 genes at the centromere-proximal region and the genes between PPP1R9A and ASB4 toward the telomere were non-imprinted and biallelically expressed. Consequently, our combining analyses of methylome, transcriptome, and informative polymorphisms revealed the boundary of imprinting cluster at the SGCE/PEG10 locus in pig chromosome 9 and consolidated the landscape of genomic imprinting in pigs.

In mammals, imprinted genes often exist in the form of clusters in specific chromosome regions (Lewis and Reik 2006). Therefore, detailed comparative characterization of large imprinted clusters is important to understand epigenetic regulation of genomic imprinting. In pigs, progress has been made in identifying imprinted genes (Li et al. 2008;Bischoff et al. 2009;Oczkowicz et al. 2012;Tian 2014;Ahn et al. 2020); however, still a relatively few genes and clusters have been identified and genes within or adjacent to putative imprinted clusters need to be investigated, including those at the SGCE/PEG10 locus.
In the mouse, a large cluster of imprinted genes within a 1-Mb region between Col1a2 and Asb4 in proximal (centromeric) chromosome 6 (6A1) was previously reported (Ono et al. 2003). Of nine murine genes belonging to the above interval (Col1a2, Casd1, Sgce, Peg10, Ppp1r9a, Pon1, Pon3, Pon2, and Asb4), two genes, Sgce and Peg10, are paternally expressed imprinted genes (Piras et al. 2000;Ono et al. 2003). This paternal expression appears critical in mammalian development as evidenced by early embryonic lethality of mice with maternal duplication of the proximal chromosome 6 cluster and with a deficiency of Peg10 (Ono et al. 2006). Three genes (Ppp1r9a, Pon3, and Pon2) from the same cluster were maternally expressed in the embryonic stages with some exceptions, and the Asb4 gene showed a strong maternal expression (Mizuno et al. 2002;Ono et al. 2003). Expressions of Col1a2 and Casd1 were biallelic, and Pon1 expression was low in the mouse embryo and undetectable in human-derived cell lines (Mizuno et al. 2002;Okita et al. 2003;Ono et al. 2003). Although the way that the epigenetic germline mark affects the expression of genes in cis remains to be determined, a single cis-acting imprinting control region (ICR) which may establish and maintain imprinting throughout the Col1a2-Asb4 cluster was identified at the SGCE/ PEG10 differentially methylated region (DMR) (Ono et al. 2003;Monk et al. 2008). This DMR/ICR acquired allele-specific DNA methylation in the maternal germline. In humans, on chromosome 7 (7q21.3), paternal expressions of SGCE and PEG10 was conserved (Ono et al. 2001;Müller et al. 2002;Grabowski et al. 2003). Human PPP1R9A gene was found to be imprinted and maternally expressed mainly in skeletal muscle tissues (Nakabayashi et al. 2004). In ovine fetuses, a putative imprinted interval was reported to be extended to between CASD1 (maternally expressed) and PON3 (paternally expressed), while the paternal expression of SGCE and PEG10 and maternal expression of PPP1R9A were conserved (Duan et al. 2018).
In this study, the imprinted cluster containing the SGCE/PEG10 locus was determined by comparing parthenogenetic (PA) and normal control (CN) porcine embryos using whole-genome bisulfite sequencing (WGBS) and RNA sequencing (RNA-seq). In addition, to confirm the expression pattern of PPP1R9A, a single nucleotide polymorphism-based assay was performed. We report that the orthologous pig imprinted interval was between SGCE and PEG10, since both the centromere-proximal genes (COL1A2 and CASD1) and the telomeric genes between PPP1R9A and ASB4 were non-imprinted and biallelically expressed.

Ethics statement
All the treatments and experiments related to pigs were performed in accordance with study protocols and standard operating procedures that were approved by the Institutional Animal Care and Use Committee of the National Institute of Animal Science, Rural Development Administration (RDA) of Korea (approval number NIAS2015-670).
In vitro maturation of porcine oocytes and production of parthenogenetic embryos Pig oocytes were collected and in vitro maturation (IVM) was performed as described in our previous reports (Kwon et al. 2017;Ahn et al. 2020). In brief, from a local slaughterhouse (Nonghyup Moguchon, Gimje, Korea), ovaries from Landrace x Yorkshire x Duroc (LYD) pigs were obtained, and then immediately transported to our laboratory in saline solution supplemented with penicillin and maintained at 30-35°in a thermos. After collection of cumulus-oocyte complexes (COCs), they were washed in Tyrode's lactate-Hepes containing 0.1% (w/v) polyvinyl alcohol. Oocytes surrounded by several layers of cumulus cells were collected and washed three times in Grand Island,NY,USA) which is supplemented with 0.1% polyvinyl alcohol (w/v), 3.05 mM D-glucose, 0.91 mM sodium pyruvate, 0.57 mM cysteine, 0.5 mg/ml luteinizing hormone, 0.5 mg/ml follicle stimulating hormone, 10 ng/ml epidermal growth factor, 10% porcine follicular fluid (pFF), 75 mg/ml penicillin G, and 50 mg/ml streptomycin. For IVM, in each well of five four-well dishes (Nunc, Roskilde, Denmark) containing 500 mL of maturation medium, 50 COCs were placed and the oocytes were matured for 40-42 h at 38.5°in an incubator containing 5% CO 2 . Cumulus cells were then removed, and oocytes having the first polar body were selected and activated. In a fusion chamber with 250 mm diameter wire electrodes (BLS, Budapest, Hungary) covered with 0.3 M mannitol solution containing 0.1 mM MgSO 4 , 1.0 mM CaCl 2 , and 0.5 mM Hepes, selected oocytes were placed and fusion was achieved by two DC pulses (1 sec interval) of 1.2 kV/cm for 30 ms using an LF101 Electro Cell Fusion Generator (Nepa Gene Co., Ltd., Chiba, Japan). After electric stimulation of oocytes followed by 2 h of stabilization period, 200 parthenogenetic embryos were placed into oviducts of each of the two LYD surrogate gilts aged 12 months at onset of estrus (i.e., a total of 400 parthenogenetic embryos for two surrogates) to produce developed parthenogenetic embryos.
Collection of both fertilized control (CN) embryos and parthenogenetic (PA) embryos To produce fertilized control (CN) embryos, during the natural heat period at the onset of estrus (day 0) two LYD gilts were naturally mated with boars twice with a 6 h interval. Both 12 CN embryos and six PA embryos were recovered at day 21 from the gilts and surrogates, respectively, that had been confirmed pregnant by an ultrasound examination. Regarding the recovery, it was considered that some imprinted genes in pigs (e.g., H19, IGF2, XIST and PEG10) does not exhibit monoallelic expression until the blastocyst stage (Park et al. 2011), and porcine parthenogenetic embryos undergo morphological changes at approximately day 30 of gestation (Bischoff et al. 2010) and ceased fetal development with degenerated organs having a high rate of apoptosis of cells at day 35 of gestation (Moss et al. 2020). After the gilts and surrogates were killed, their reproductive tracts were sectioned and placenta were isolated from the uterus. Then, embryos separated from the surrounding placenta were placed on a piece of cleaning tissue in order to dry liquid on the surface of the embryos. Morphologically intact embryos with comparable sizes were selected for further experiments (lengths of 2.2, 2.1, and 2.1 cm for CN and 2.0, 2.0, and 1.6 cm for PA), and stored in liquid nitrogen until further use.
Genomic DNA isolation and whole-genome bisulfite sequencing (WGBS) analysis From the whole collected embryos of CN and PA (n = 3 for each), genomic DNA was isolated and fragmented. For optimization of bisulfite conversion of genomic DNA, Accel-NGS Methyl-Seq DNA Library Kit (Swift Biosciences, Inc. Ann Arbor, MI, USA) was used according to the manufacturer`s instructions. PCR was conducted with adapter primers and Diastar EF-Taq DNA polymerase (Solgent, Daejeon, Korea) to amplify the bisulfite-treated DNA, under the following thermal conditions: 3 min at 95°followed by 35 cycles of 30 sec at 95°, 30 sec at 60°, and 30 sec at 72°, and a final extension for 5 min at 72°. After bead-based clean-up, the PCR products were sequenced using HiSeqX sequencer operated by Macrogen (Seoul, Korea). For each of the WGBS libraries, the number of raw reads were approximately 848M (CN1), 864M (CN2), 868M (CN3), 842M (PA1), 859M (PA2), and 851M (PA3). After trimming adapter sequences and filtering out reads shorter than 20bp using Trim Galore (v0.4.5), cleaned reads for each library [approximately 847M (CN1), 862M (CN2), 866M (CN3), 840M (PA1), 857M (PA2), and 849M (PA3)] were retrieved (Table S1). After mapping to the reference genome (Sscrofa11.1/susScr11) with BSMAP (v2.87), the clean mapped reads were sorted and indexed. Using 'methylatio.py' script in BSMAP, extraction of the methylation ratio of every single cytosine from the mapping results was conducted. DMRs were identified by the program metilene (v0.2-8) (Juhling et al. 2016) using the methylation ratios at all CpGs. The criteria for DMR were a presence of more than 10 CpGs, a genomic distance of less than 300 bp between CpGs, and a mean methylation difference of more than 0.2 between groups. DMRs with false discovery rate (FDR) , 0.05 were obtained as significant DMRs (Table S2). To visualize methylation ratios and significant DMRs throughout the genomic coordinates of the imprinted cluster, the R/Bioconductor package Gviz (v1.28.3) (Hahne and Ivanek 2016) was used.
RNA extraction, cDNA library construction, and RNA sequencing (RNA-seq) Total RNA from whole CN embryos (n = 3) and whole PA embryos (n = 3) samples was isolated with TRIzol reagent (Sigma-Aldrich, USA) following the manufacturer's instructions. DNase I was treated to the RNA samples to avoid genomic DNA contamination. The RNA samples were electrophoresed in 1.2% agarose gels to evaluate the integrity of RNA, which was then confirmed by more than 2 of 28S/18S rRNA ratio and more than 7 of RNA integrity number (RIN) using an Agilent 2100 BioAnalyzer. Using the ratios of A260/A280 and A260/A230 (1.8-2.0), the concentrations of RNA were assessed. One ug of total RNA was used to construct cDNA libraries with the TruSeq RNA Sample Prep Kit v.2 (Illumina, San Diego, CA, USA). The final cDNA library was produced using the protocol consisted of polyA-selected RNA extraction, RNA fragmentation, random hexamer primed reverse transcription and amplification. Quantification of the cDNA libraries was done by quantitative Real-Time PCR (qPCR) according to the qPCR Quantification Protocol Guide, and qualification of the libraries was assessed using the Agilent 2100 Bioanalyzer. The Illumina HiSeq2500 RNA-seq platform was used to sequence the library products (100nt paired-end). The number of raw reads generated for each RNA-seq library were approximately 77.  (Table S1). From the Assembly database at the National Center for Biotechnology Information (NCBI) (http:// www.ncbi.nlm.nih.gov/assembly/GCF_000003025.6), the reference genome sequence of Sus scrofa (Sscrofa11.1/susScr11) and annotation files were downloaded. Then, using HISAT2 (v2.1.0) (Kim et al. 2015) with default parameter settings, the clean sequencing reads were aligned to the reference genome, except for the parameter-dta regarding tailored transcriptome assembly. BAM format containing RNA-seq alignments were produced and sorted by SAMtools (v1.9), and then BAM file-derived read coverage (or depth) was plotted and visualized using the R/Bioconductor package Gviz (v1.28.3) (Hahne and Ivanek 2016).

Analysis of differential gene expression
The htseq-count script (Anders et al. 2015) was used to count aligned RNA-seq reads using each BAM file and an annotation file as inputs, and a count matrix was generated. Using the R/Bioconductor package DESeq2 (v1.26.0) (Love et al. 2014), read counts from individual fetal samples were normalized for sequencing depth and differential expression between CN embryos and PA embryos was calculated. To obtain significant differentially expressed genes (DEGs), the combined criteria of FDR , 0.05 and the absolute log2-fold change . 1 was used, where a fold change is defined as the expression in samples of PA divided by the expression in samples of CN.
Polymorphism-based assays Potential single nucleotide polymorphism (SNP) sites on the last exon of the porcine PPP1R9A gene was obtained from the Ensembl Genome Browser (http://www.ensembl.org). The gDNAs of boar (Korean native pig, KNP), sow (KNP), and offspring (neonates, 2-day-old) were isolated from ear tissue. RNA was extracted from six tissues (heart, kidney, liver, lung, muscle, and spleen) from neonates (2-day-old) and reverse-transcribed to cDNA. A primer set for amplification of a genomic DNA (gDNA) fragment containing those potential SNPs (316 bp) was designed on the last exon: forward, 59-AGGCTCTTGGGATGACATCAT-39 and reverse, 59-ACGGTTCC-CCAG TCTGCCAAAG -39. To amplify a corresponding cDNA fragment (350 bp) while avoiding genomic DNA contamination, a forward primer (59-ACTGGAGAGCAACTCCTGCAGT-39) was designed on the exon coming just before the last exon and was coupled with the above reverse primer on the last exon. The conditions for PCR amplification were 95°for 2 min, appropriate cycles with linear amplification ranges of 95°for 20 s, 60°(gDNA) or 54°(cDNA) for 40 s, 72°for 30 s, with an additional extension step at 72°for 5 min. The amplified gDNA fragment was sequenced to identify heterozygous offspring with potential SNPs. The informative SNP sequences in gDNA of offspring were compared with sequences in amplified cDNA of offspring to determine maternal and/or paternal expression. Those SNP sequences of gDNA in the boars and sows were compared with above sequences to determine parentof-origin-specific gene expression.

RESULTS
Differentially methylated regions were only located at the SGCE/PEG10 locus within the COL1A2-ASB4 region in pigs Nine genes (from COL1A2 to ASB4) in pig chromosome 9 were mapped to an approximate 1-Mb region between 74.17 and 75.20 Mb according to the genome sequence of Sus scrofa (Sscrofa11.1/ susScr11). The order of those genes in this locus is conserved in the human chromosome 7 (7q21.3) and mouse chromosome 6 (6 A1). In the mouse, DNA methylation in this region was examined in four CpG islands located in the promoter regions of the Casd1, Sgce-Peg10, Ppp1r9a, and Pon2 genes, and bisulfite sequencing analyses between reciprocal F1 hybrids revealed that only the promoter region of Sgce-Peg10 is differentially methylated in a parent-of-origin-specific manner (maternally hypermethylated) and the other three regions were non-methylated (Ono et al. 2003).
Similarly, in pigs, the promoter regions of the CASD1, SGCE-PEG10, PPP1R9A, and PON2 genes contained CpG islands, and only the areas containing the promoter regions of SGCE and PEG10 displayed DMR (Figure 1). DNA methylation of each of the three parthenotes and three controls was exhibited, and there were two CpG islands: one located in the 1 st exons and intergenic region of SGCE and PEG10 and the other located in the 2 nd exon of PEG10 ( Figure 1). The genomic DNA around the latter CpG island was hypermethylated in all CN and PA embryos indicating non-DMR; whereas, hypomethylation of CN was found at the 1 st exons of SGCE and PEG10, intergenic region, and introns where GC content is high (Figure 1). As a result, maternally hypermethylated DMR in PA was identified.
Promoter regions of genes in the COL1A2-ASB4 region mostly showed non-methylation A close view of promoter regions of seven genes other than SGCE and PEG10 was obtained. The promoter regions close to transcription start sites and first exons of the CASD1, PPP1R9A, and PON2 genes enclosed CpG islands and high GC contents. These promoter regions were almost non-methylated in both CN and PA as shown by a regional hypomethylation (Figure 2). It suggests that these regions are biallelically active and the CASD1, PPP1R9A and PON2 genes are not imprinted. The promoter region of the COL1A2 gene also showed a similar hypomethylation, suggesting active and non-imprinting status of this gene (Figure 2). In addition, the promoter region of the PON1 gene contained a relatively narrow range of hypomethylation and the promoter regions of the PON3 and ASB4 genes did not possess hypomethylated areas in both CN and PA, suggesting a relatively inactive status of these promoters and non-imprinting of these genes (Figure 2) Allele-specific expression in the porcine COL1A2-ASB4 region showed a partly conserved pattern Expression levels of each gene in the porcine COL1A2-ASB4 region were investigated with RNA-seq analysis. The SGCE and PEG10 genes were expressed only in CN, but not in PA (Figure 3). Because CN has one paternal and one maternal allele and PA has two maternal alleles, the sole expression in CN, but not in PA, could be derived from expression in the paternal allele ('paternal expression') which is consistent with previous reports (Bischoff et al. 2009; Park et al. Figure 1 High-throughput DNA methylation profiling of the SGCE/PEG10 locus in porcine embryos determined by WGBS. GeneRegionTrack represents an approximate 1.03-Mb region between the COL1A2 and ASB4 genes on porcine chromosome 9 (chr9: 74,174,484 -75,203,450). Nine genes are indicated by brown boxes for protein-coding transcripts (short boxes: noncoding region; tall boxes: protein-coding region), and directions are marked by black horizontal arrows. From I track to PA-CN track: I, CpG islands; GC%, GC content; PA, mean methylation ratio of parthenotes (pink histogram lines); CN, mean methylation ratio of controls (blue histogram lines); and PA-CN, mean methylation ratio of parthenotes subtracted by that of controls (green histogram lines). In PA-CN track, hypermethylated DMR in PA (FDR , 0.05) is overlaid with red histogram lines. Yellow highlighted area, an area containing the promoter regions of SGCE and PEG10 and two CpG islands that is zoomed in. In the bottom plot, PA1 -3, methylation ratios of three different parthenotes; CN1 -3, methylation ratios of three different controls; PA, mean ratio; CN, mean ratio; PA-CN, PA subtracted by CN; and DMR, differentially methylated region (FDR , 0.05). 2011). Among the genes between PPP1R9A and ASB4 toward the telomere, PON2 was expressed biallelically as described previously (Zhou et al. 2011) and expression of ASB4 was also biallelic ( Figure  3). Expression of PON1 was low, but appeared to be biallelic, and PON3 expression was not detectable. Regarding expression of PPP1R9A, previous studies showed that it is expressed biallelically in tissues from 2-month piglets, but maternally in placental tissues (Bischoff et al. 2009;Zhang et al. 2011), and it appeared to be biallelic (Figure 3). In addition, expressions of both the COL1A2 and CASD1 genes at the centromere-proximal region were biallelic. Taken together, compared with humans and mice, only SGCE and PEG10 genes showed a conserved paternal allele-specific expression pattern.
To validate the above expression patterns, DEGs were screened. Among the nine genes in the porcine COL1A2-ASB4 region, the SGCE and PEG10 genes were differentially expressed and showed significantly higher expression in CN indicating their paternal allelespecific expression (Table 1). In contrast, the expression of PPP1R9A tended to be higher in PA, but it was statistically nonsignificant indicating its expression is biallelic. To confirm the expression pattern of PPP1R9A, a SNP-based assay was performed as presented below (Figure 4). The rest of detected genes, the COL1A2, CASD1, PON1, PON2, and ASB4 genes, did not show a difference between CN and PA indicating their biallelic expression and expression of the PON3 gene was not detected (Table 1).

SNP-based analysis revealed biallelic expression of PPP1R9A
Among SNPs in the last exon of the porcine PPP1R9A gene, a SNP rs329892462 (G/A) was heterozygous (informative) in our gDNA samples from offspring ( Figure 4). The A allele in gDNA of offspring was derived from either sow or boar. Subsequently, cDNA from six tissues including heart, kidney, liver, lung, muscle and spleen of offspring were sequenced to investigate whether or not allele-specific expression occurs. In all tested tissues of both offspring from two different combinations of boar and sow, allele-specific patterns were not shown and biallelic expressions with an approximate A:G ratio of 1:1 were detected (Figure 4).
Comparative approach highlighted diversity and universality of CALCR and ASB4 imprinted cluster Imprinted expression of the genes located between CALCR and ASB4 was compared between mice and humans (Table 2). In general, expression patterns in the porcine embryos within the syntenic region were closer to the expression in humans than that of mice, but in the pig the PPP1R9A gene showed biallelic expression in all tested tissues Figure 2 Promoter regions of the seven genes other than SGCE and PEG10 and CpG islands in the COL1A2-ASB4 region and methylation profiles. GeneRegionTrack represents regions containing a transcription start site and the first exon of the seven genes. From I track to PA-CN track: I, CpG islands; GC%, GC content; PA1 -3, methylation ratios of three different parthenotes; CN1 -3, methylation ratios of three different controls; PA, mean methylation ratio of parthenotes; CN, mean methylation ratio of controls; and PA-CN, mean methylation ratio of parthenotes subtracted by that of controls.
[embryonic and 2-day-old in the current study, and 2-month-old as reported in (Zhang et al. 2011)]. This biallelic expression of porcine PPP1R9A was different from the mouse and human which showed maternal expression in embryonic or fetal skeletal muscle tissues (Nakabayashi et al. 2004;Monk et al. 2008) and ovine fetuses that showed maternal expression in the kidney (Duan et al. 2018). Our polymorphism-based analysis revealed biallelic expression of PPP1R9A in the muscle and kidney in 2-day-old neonates ( Figure 4). In addition, based on this study and literature, the imprinting pattern was illustrated by displaying differences and similarities among various species: selected mammals (pig, human, mouse, and sheep), the marsupial tammar wallaby with imprinting of PEG10 but not SGCE (Suzuki et al. 2007), and the avian (chicken) deficient in imprinting (Frésard et al. 2014) (Figure 5).

DISCUSSION
In this study, we report a comprehensive assessment of DNA methylation and allelic expression in the large chromosomal interval between COL1A2 and ASB4 in pigs. Our generation of parthenogenetic porcine embryos followed by strategies that combine whole-genome DNA methylation, transcriptome, and SNP-based allelic expression enabled the large-scale imprinting profiling on a genomic level. The methylome and transcriptome were constructed independently for  three individuals in each control and parthenogenetic group to reduce genetic variability, and SNP analysis further evaluated allelic expression of the PPP1R9A gene in neonatal pigs. This experimental design was appropriate to detect, not only paternal expression which showed an all-or-none fashion (i.e., SGCE and PEG10 expressions in only control embryos having a paternal allele), but also allelic expression which was verified by the ensuing SNP analysis, followed by determining the boundary of imprinting cluster. Genomic imprinting operates in mammals via several gene silencing mechanisms: e.g., i) direct silencing of a promoters where the non-transcribed allele is heavily methylated, ii) inhibiting long-range communication between promoters and enhancers by binding of an insulator on the unmethylated allele and forming chromatin loops, iii) repressing a long noncoding RNA (ncRNA) transcript by methylation only on one of the parental alleles and silencing the other allele by the non-repressed long ncRNA. iv) isoform-dependent silencing originated from alternative promoter usage where different isoforms are silenced or active depending on the influence of DNA methylation (Barlow and Bartolomei 2014;Ahn et al. 2020). Regarding the third mechanism, unlike other known imprinted clusters, there was no evidence of a long ncRNA influenced by the porcine SGCE/PEG10 DMR as reported in the mouse and human (Monk et al. 2008). Also, alternative promoter usage in the fourth mechanism was not likely to occur in this genomic interval because there was no evidence of expression of isoforms other than the displayed main transcripts according to our screening of transcriptome data using Integrative Genomics Viewer (IGV). Within the porcine COL1A2-ASB4 region, maternal alleles of SGCE/PEG10 locus appears inactive through direct silencing of promoters by maternal hypermethylation as described in the first mechanism, resulting in paternal expression. In addition, this maternal hypermethylation (and an insulator binding on unmethylated allele) could be involved in inhibition of long-range communication between a promoter and an enhancer as described in the second mechanism. However, genes other than SGCE and PEG10 including PPP1R9A did not show an allele-specific expression pattern, but a biallelic expression, and therefore the insulator-mediated mechanism was not the mode of action in the SGCE/PEG10 imprinting cluster in pigs. A previous study showed, using semiquantitative RT-PCR, expression of porcine PPP1R9A was greater in the parthenote than the control placental sample (a parthenote: control ratio of 1.7), suggesting maternal expression, but similar in the brain, fibroblast, and liver (Bischoff et al. 2009). Another study revealed that  expression of PPP1R9A is biallelic in multiple tissues of 2-month-old piglets, except skeletal muscle, fat, and testis in which expression of PPP1R9A was not detectable (Zhang et al. 2011). To confirm whether or not allele-specific expression of PPP1R9A occurs in neonatal pigs, polymorphism-based assays were conducted for both parents and offspring and it revealed a biallelic expression of PPP1R9A which is different from its maternal expression in the mouse, human, and sheep (Ono et al. 2003;Nakabayashi et al. 2004;Duan et al. 2018). The telomeric (distal) boundary of the large imprinted cluster organized around Sgce/Peg10 locus in the mouse proximal region toward the centromere of chromosome 6 (6A1) reaches the Asb4 gene, based on previous reports on maternally expression of Ppp1r9a, Pon3, Pon2, and Asb4 in mice (Piras et al. 2000;Mizuno et al. 2002;Ono et al. 2003;Nakabayashi et al. 2004). However, the boundary spans between CASD1 and PON3 in sheep (Duan et al. 2018), while in the human it spans between SGCE and PPP1R9A (Ono et al. 2001;Grabowski et al. 2003;Okita et al. 2003;Monk et al. 2008). In addition, there are reports about monoallelic expression of mouse Calcr and Tfpi2 and human CALCR (in brain) and TFPI2 toward the centromeric (proximal) region, but the expression of those genes was restricted to brain and placenta, respectively Monk et al. 2008). Our porcine parthenogenetic embryos showed an imprinting boundary between SGCE and PEG10 which is narrower than the human one, and expression of CALCR and TFPI2 was not monoallelic in this study ( Figure S1). As illustrated, genomic imprinting of SGCE/PEG10 is conserved among mammals, but not in the marsupial and avian ( Figure 5). The tammar wallaby showed a distinct imprinting pattern with a DMR located only at the promoter region of PEG10, but not within the promoter region of SGCE (Suzuki et al. 2007), while the chicken has been found to be deficient in genomic imprinting (Shin et al. 2010;Renfree et al. 2013;Frésard et al. 2014). Regarding the role of imprinting, both the murine Sgce and Peg10 genes are expressed in the embryonic tissues and their paternal expression is essential for embryonic development, while maternal uniparental inheritance of the genomic region containing Sgce and Peg10 is associated with embryonic lethality (Piras et al. 2000;Ono et al. 2006). As shown in Peg10 knockout mice, deletion of Peg10 gave rise to growth retardation and an early embryonic lethal phenotype (Ono et al. 2006). It has been reported that polymorphisms in PEG10 genes are associated with carcass traits and fat deposition in pigs (Qiao et al. 2016), and SGCE and PEG10 are candidate genes for piglet birth weight (Zhang et al. 2014). Taken together, genomic imprinting in the SGCE-PEG10 cluster may play an important role in growth and development of pigs.
As shown in Table 2, parent-of-origin-dependent expression is often a developmental stage-and tissue-specific. In the mouse, embryonic days of analyzed embryos were between E10 and E15.5 which corresponds to Carnegie Stage (CS) 11-22, and the d21 porcine embryos used in this study were at CS15 (Hikspoors et al. 2016). Human fetal tissues collected in a previous study were from 8-to 18-week-old fetuses (which are in or after CS23) that were obtained at the termination of pregnancies (Monk et al. 2008). Therefore, the developmental stage of murine and porcine embryos was earlier than the human fetuses which requires careful attention. In addition, although our study did not include in vitro matured oocytes undergone in vitro fertilization (IVF), a previous study using these IVF oocytes showed that more exposure to in vitro culture condition might produce an adverse effect on dynamics of DNA methylation and increase DNA methylation during embryonic development (Deshmukh et al. 2011). This can suggest that there is a possibility of increased DNA methylation during in vitro handling even though our PA oocytes were transferred into the oviduct instead of culturing in vitro after activation. However, the degree of increase could be low, because in our previous reports almost complete absence of DNA methylation in PA was found in Nesp DMR which is paternally hypermethylated (Ahn et al. 2020). Additionally, there were some embryonic losses due to abnormal shaping but, among well-developed d21 PA embryos, we selected PA embryos that can be compared with Figure 5 Simplified schematic representation of methylation and gene expression within the region spans between COL1A2 and ASB4 (or PDK4) genes in various species. Filled red lollipops indicate paternally hypermethylated DMRs. Paternal (blue), maternal (pink), weak maternal (light pink), biallelic (gray), tissue and/or stage-dependent maternal or biallelic (green) and low (blank with dotted gray perimeter) expressions are denoted. The direction of transcription is indicated with arrows. The genomic distance was scaled based on the NCBI gene database (http:// www.ncbi.nlm.nih.gov/gene), except the tammar which was based on Suzuki et al., 2007. control embryos based on intact morphology and sizes. Furthermore, the same chromosomal content was compared between control and PA embryos by genotyping the sex of control embryos and selecting female control embryos (i.e., control embryos with 36,XX were compared with PA embryos with a double of 18,X).
In conclusion, porcine parthenogenetic embryos were used to explore a putative imprinted genomic interval, and this study revealed that the porcine SGCE/PEG10 locus may not be clustered with a nearby gene locus. This cluster contains a single cis-acting DMR/ICR at the SGCE/PEG10 locus, and direct silencing of promoters by hypermethylation is the mechanism underlying genomic imprinting of SGCE/PEG10. Our findings provide a comprehensive overview of one of the important imprinted clusters of pig chromosome 9 that may be implicated in the regulation of pig growth and development. Additionally, our dataset and forthcoming publication will greatly facilitate analyses on other imprinted loci in the pig genome.