Polar overdominance – a putative molecular mechanism and the new examples in mammals

Polar overdominance is closely related to parental imprinting a phenomenon where only one (maternal or paternal) allele is expressed in the offspring. In contrast to gametic imprinting, polar overdominance is not very well known and only few examples of this phenomenon have been reported in mammals. One of them is the callipyge phenotype in sheep, which appears in the offspring only when the mutated allele comes from the father and the wild allele comes from the mother. In review, latest concepts on the molecular mechanism underlying the callipyge phenotype are presented and the new examples in mammals are discussed.


INTRODUCTION
Polar overdominance is closely related to parental imprinting -a phenomenon where only one -maternal or paternal allele is expressed in the offspring.Gametic imprinting was fi rst discovered in the murine Igf2 gene (De Chiara et al., 1991) and since then has been studied intensively.To date approximately 80 imprinted genes are known (Moore et al., 2008) in various mammals.One of the most signifi cant studies connected with gametic imprinting was performed by Kono et al. (2006), who used mutant mice with a 13-kilobase deletion in the H19 gene as non-growing oocyte donors and generated a viable parthenogenetic mouse by ap-propriate expression of the Igf2 and H19 genes.This experiment has proved that gametic imprinting was a barrier to developing a parthenote.Another important fi nding was the localization of a causative mutation in the IGF2 gene, affecting muscularity of pigs (Van Leare et al., 2003).This mutation can be effectively used in the pig industry to improve meatiness of pigs by marker assisted selection.

CALLIPYGE LOCUS -THE FIRST EXAMPLE OF POLAR OVERDOMI-NANCE IN MAMMALS
In contrast to gametic imprinting, polar overdominance is not very well known and only few examples of this phenomenon have been reported in mammals.It was fi rst noticed in a ram called Solid Gold.This sheep was famous for extreme muscularity, especially pronounced in the hindquarters.Surprisingly, Solid Gold passed its unique features only to approximately 15% of its offspring.This suggested that the callipyge phenotype was inherited in a non-Mendelian manner.After a series of crossings it turned out that the callipyge phenotype appears in the offspring only when the mutated allele comes from the father and the wild allele comes from the mother (C pat /N mat genotype) (Cockett et al., 1996).Animals with other genotypes (C pat /C mat , N pat /N mat , N pat /C mat ) did not exhibit the characteristic phenotype.
The CLPG locus has been localized in the telomeric region of chromosome 18, within a cluster of imprinted genes.After sequencing, a causative mutation has been described (Freking et al., 2002).Single Nucleotide Polymorphism (SNP CLPG ) located in an intergenic region, between DLK1 and MEG3, was unique to Solid Gold and its offspring.Since then, substantial efforts have been undertaken to explain the molecular mechanism by which SNP CLPG causes the callipyge phenotype.The causality of SNP CLPG was further supported by the demonstration that Solid Gold was mosaic for this SNP (Smit et al., 2003).

PUTATIVE MOLECULAR MECHANISM UNDERLYING POLAR OVER-DOMI-NANCE IN THE CALLIPYGE LOCUS IN SHEEP.
The region adjacent to SNP CLPG is expressed as a 547-bp RNA transcript containing an open reading frame (ORF) predicting 123 amino acids in which SNP CLPG would alter a serine codon to a proline, but this ORF is not conserved in humans and mice and the likelihood that it produces the corresponding peptide is unknown (Freking et al., 2002).It was supposed that SNP CLPG somehow alters expression of adjacent genes.There are several imprinted genes in the neighborhood of SNP CLPG and all of them could take a part in developing the callipyge phenotype.To date, OCZKOWICZ M. paternally expressed DLK1 (Drosophila Like Homologue 1) is considered as the main candidate gene.DLK1 codes for signal protein, which regulates differentiation of various cell lines, muscle and adipose tissue.Its expression decreases during differentiation of adipose tissue and thus it is supposed to keep preadipocytes in undifferentiated state.
In this region, there are also three maternally expressed genes which do not code for proteins (MEG3/GTL2, antiPEG11 and MEG8) and paternally expressed PEG11/Rtl1, which has an open reading frame and hypothetically codes for proteins, but this has not been identifi ed so far (Cockett et al., 2005).
The fi rst possible explanation of the molecular mechanism underlying polar overdominance at the callipyge locus emerged after examination of the expression of paternally expressed DLK1, PEG11, and maternally expressed MEG3 and MEG8 genes in the four possible SNP CLPG genotypes (Charlier et al., 2001a).DLK1 and PEG11 transcripts were very abundant in skeletal muscle of individuals that have the CLPG mutation on their paternal chromosome (genotypes C mat /C pat , N mat /C pat ), whereas in skeletal muscle of N mat /N pat , C mat /N pat individuals, they were present at very low concentration or absent.Analogous results were obtained for MEG3 and MEG8.Expression of these genes was higher in skeletal muscles of individuals that have the CLPG mutation on their maternal chromosome.Moreover, CLPG mutation did not alter imprinting status of analysed genes.The authors concluded that CLPG mutation enhanced expression level of DLK1, MEG3, PEG11 and MEG8 genes in cis, probably by modifying the activity of a common regulatory element.CLPG phenotype results from a DLK1 and/or PEG11 overexpression in skeletal muscle with simultaneous underexpression of MEG3 and MEG8.The model explaining why callipyge homozygotes do not exhibit the callipyge phenotype was proposed: in C mat /C pat individuals overexpression of MEG8 and MEG3 or other maternally expressed genes interfere in trans with DLK1 and PEG11 and inhibit expression of DLK1 and PEG11 (Charlier et al., 2001a;Georges et al., 2003; Figure 1).The authors proposed that this interference could operate at the transcriptional level (competition for transcription factors), on the stability of the DLK1/PEG11 mRNAs or even at the protein level.However, new data indicate microRNA interference as a very probable factor (Seitz et al., 2004a;Royo et al., 2008).

FUNCTIONAL STUDIES OF GENES LOCATED IN THE CALLIPYGE REGION
Similarly as in sheep, in the human and mouse genome a cluster of imprinted genes including DLK1, MEG3/GTL2, PEG11/RTL1, antiPEG11 and MEG8 is present.Human DLK1 has been mapped to chromosome 14q32 that is syntenic to mouse distal chromosome 12 and sheep chromosome 18 (Gubina et al., 1999).Moreover, it was proved that in humans and mice, DLK1 is paternally expressed whereas MEG3 is maternally expressed (Kobayashi et al., 2000;Wylie et al., 2000).Studies on mouse uniparental disomy for chromosome 12 (UPD12) revealed that both types of UPD (maternal and paternal) result in embryos that are non-viable.Paternal UPD12 conceptuses died in late gestation and exhibited placentomegaly, had costal cartilage defects and hypo-ossifi cation of mesoderm-derived bones.On the other hand, mUPD12 conceptuses survive to term, but die perinatally.Maternal UPD12 conceptuses are growth retarded and both types of UPD12 exhibit skeletal muscle maturation defects (Georgiades et al., 2000).In humans, maternal UPD14 results in scoliosis, hypotonicity, early puberty, obesity and blepharophimosis.
In order to determine the role of Dlk1 in mouse mUPD12 and human mUPD 14 phenotype, Dlk1-null mice were generated.Dlk1-null mice displayed growth retardation, eyelid and skeletal abnormalities, hypercholesterolaemia, hyperlipidaemia, enlarged fatty liver and accelerated obesity.Results of the experiment indicated that Dlk1 is the gene responsible for most of the imprinting-related phenotypes of mouse mUPD12 and syntenic human mUPD14; however, some of the phenotypes observed in human mUPD14 like hypotonicity and early puberty were not apparent in Dlk1-null mice and may be caused by misexpression of another imprinted gene, possibly Peg11.In addition, a role of Dlk1 gene in maintaining homeostasis of lipid metabolism was demonstrated as Dlk1-null mice exhibited hypercholesterolaemia and hyperlipidaemia.Furthermore, to determine if heterozygotes with either maternal or paternal inheritance of Dlk1 knockout allele (m -/+ or +/p -) have phenotypes similar to those of wild-type mice or null mice respectively, male or female heterozygotes (+/-) were mated with wild-type mice.Western blot analysis demonstrated that Dlk1 protein was expressed exclusively from the paternal allele.Moreover, (+/p -) animals exhibited the same growth retardation and increase in fat pad weight as Dlk1-null mice.In contrast, (m -/p+) have the same growth rate and fat pad weight as normal littermates (Moon et al., 2002).

EFFECTS OF CLPG GENOTYPE ON EXPRESSION OF ADJACENT GENES
Since the origin of the model explaining callipyge phenotype (Figure 1), many studies concerning expression of genes within the callipyge cluster have been conducted.Generally, results of these studies are in agreement with the proposed model and indicate new possible factors contributing to the callipyge phenotype.Bidwell et al. (2004) observed a signifi cant effect of the callipyge genotype on DLK1 expression in muscles that undergo hyperthrophy (Gluteus medius) in 8-week-old lambs.Expression level of this gene was 6-fold higher than in normal animals.In muscle that does not undergo hypertrophy (Supraspinatus) DLK1 transcript abundance was the same in four analysed genotypes.Moreover, Murphy (2005) reported that DLK1 expression was elevated prenatally, but in adult animals expression decreased threefold.Strikingly, in N mat /C pat animals elevated levels of DLK1 transcript were maintained.This observation was limited only to hypertrophy responsive muscles.These results were confi rmed by Perkins et al. (2006), who observed 16-fold and 5-fold higher expression of DLK1 in N mat /C pat vs N mat /N pat in 2-and 8-week-old lambs, respectively, in hypertrothy responsive muscle.Similarly to previous results, no differences were observed in Supraspinatus.
Contribution of DLK1 gene to developing the callipyge phenotype was analysed at the protein level as well.Davis et al. (2004) showed by immunohistochemistry that DLK1 protein is present only in Longissimus dorsi of N mat /C pat animals.In muscles of C mat /C pat animals there was no DLK1 protein despite the presence of DLK1 mRNA.However, White et al. (2008) detected DLK1 protein, albeit at a lower level in the C mat /C pat and N mat /C pat animals.The authors suggested that dif-ferences refl ect the sensitivities of the methods used.Nevertheless, observations greatly supported the hypothesis about an important role of DLK1 in developing the callipyge phenotype.
Contribution of PEG11 to developing the callipyge phenotype is more elusive.The PEG11 gene contains an intronless open reading frame of 1333 amino acids (Charlier et al., 2001b); however, to date the PEG11 protein has not been identifi ed.PEG11 mRNA abundance in hypertrophy responsive muscle (Semimembranosus) was 33-fold higher in N mat /C pat , compared with other genotypes at 2 weeks of age.Moreover, in another hypertrophy responsive muscle -Gluteus medius of N mat /C pat animals expression of PEG11 was 14-fold higher than expression of anti-PEG11-transcript transcribed from the plus strand (Charlier et al., 2001b), whereas in the animals of the other three genotypes the situation was just the opposite -antiPEG11 transcripts were expressed at higher levels than PEG11 (Bidwell et al., 2004).Perkins et al. (2006) demonstrated that PEG11 expression was elevated in N mat /C pat 2-week-old lambs in hypertrophy responsive and non-hypertrophy responsive muscles.Compared to prenatal time point PEG11 expression increased in both muscles by seven-to eight-fold in callipyge animals by 2 weeks of age, but declined in the other three genotypes.Several miRNAs processed from an-tiPEG11 have been detected (Davis et al., 2005).The authors demonstrated also the in vivo trans-inhibition of the paternally expressed PEG11 by microRNA processed from maternally expressed antiPEG11.
Expression of maternally expressed genes (MEG3 and MEG8) was studied by Murphy et al. (2005) and Perkins et al. (2006).In C mat /N pat and C mat /C pat animals they observed higher expression of MEG3 and MEG8 genes than in N mat /N pat , N mat /C pat sheep, but the observations were not restricted to hypertrophy responsive muscles.In humans and mice, microRNA processed from MEG8 has recently been identifi ed (Cavaille et al., 2002;Seitz et al., 2004b).However, further studies are needed to understand the molecular mechanism by which maternally expressed genes affect the callipyge phenotype.

NEW POSSIBLE MOLECULAR FACTORS CONTRIBUTING TO CALLIPYGE PHENOTYPE
Another factor changing the expression of DLK1 and PEG11 was suggested by Murphy et al. (2006), who examined methylation status of the callipyge region and expression of CLPG1 transcript -noncoding RNA produced from the region covering SNP CLPG , described by Freking et al. (2002).Prenatally, CLPG1 transcript was expressed biallelically in Longissimus dorsi of sheep of all four genotypes, whereas after birth this expression was maintained only in animals carrying the mutation (C mat /C pat , N mat /C pat ).Additionally, presence of SNP CLPG changed local C mat /N pat CpG methylation in adult animals.In callipyge sheep, there was CpG hypomethylation of this region compared with normal sheep, presence of CLPG1 transcript and high expression of DLK1.The authors suggested that SNP clpg may alter the ability of regional chromatin to condense by inhibition of chromatin remodelling complex assembly, inhibition of DNA methylation or histone modifi cation and this affects expression of DLK1 and PEG11.
Studies on infl uence of SNP CLPG on epigenetic marking were continued by Takeda et al. (2006), who demonstrated that the mutation alters methylation status of DLK1-GTL2 intergenic region.Because SNP CLPG is able to change expression of a core cluster of neighboring genes it is thought to be located within the locus control region (LCR).Previous reports have shown that such LCRs are DNase-I hypersensitive (Li et al., 2002), therefore the authors decided to check the presence and effect of the CLPG mutation on DNase I Hypersensitive Sites (DHS).They detected at least three DHSs in N/N animals and at least two additional DHSs in C/C animals.CLPG allele exhibited higher sensitivity to DNase I in skeletal muscle but not in the liver.Moreover, they proved that SNP CLPG strongly enhances biallelic, long-range transcription throughout the DLK1 -GTL2 IG region.They confi rmed the results obtained by Murphy et al. (2006) and demonstrated that the effect of SNP CLPG is not limited to its vicinity but extends to the entire 90kb DLK1-GTL2 IG region.
Although primary effectors of a muscle hypertrophy phenotype are known, several attempts to identify new genes that may play a role in the expression of the callipyge phenotype have been undertaken.Vuocolo et al. (2006) used microarrays (bovine GeneChip) and identifi ed 159 differentially expressed genes in Longissimus dorsi of normal and callipyge sheep.Eight of them, including DLK1 and MEG3 were differentially expressed at birth and at 12 weeks of age.These were: HDAC9 -histone deacetylase 9, activating transcription factor-3 (ATF3), Ras protein dexamethasone-induced1 (RSAD 1), leucine-rich repeat-containing protein 29 (LRRC2), FOS transcript mapping to 3` UTR of the maternally expressed SLC22A3 gene.Several of them are transcription factors -positive or negative regulators of myogenesis.The authors proposed that overexpression of DLK1 results in inhibiton of the activity of Notch 1 -a known negative regulator of myogenesis and eventually alters expression of FOS, ATF 3 and HDAC 9 (Vuocolo et al., 2006).
A similar study was performed by Fleming-Wadell et al. (2007), who analysed gene expression in Longissimus dorsi of normal and callipyge sheep at 10, 20 and 30 days of age.After combining data from Affymetrix microarray and Real-Time PCR, they chose several differentially expressed genes (PFKM, PDE7A, LOC513822, DNTTIP1) to be the most likely candidates for direct involvement in DLK1 and/or PEG11-mediated muscle growth.

NEW EXAMPLES OF POLAR OVERDOMINANCE IN MAMMALS
The considerable effect of SNP CLPG on muscularity of sheep has made it a very interesting marker for marker assisted selection in other farm animals.However, SNP CLPG appeared to be a private allele, encountered exclusively in the calipyge fl ock (Smit et al., 2003).Nevertheless, Kim et al. (2004) identifi ed DLK1 polymorphism which is associated with growth, fatness and body composition in pigs.In addition, the authors proved a polar overdominant inheritance of this polymorphism.These results were confi rmed by QTL (Quantitative Trait Locus) analyses (Li et al., 2008).Additionally, paternal expression of DLK1 and maternal expression of MEG3 in porcine muscle and adipose tissues was shown (Li et al., 2008).
In humans a SNP located only 93bp from the polymorphism described by Kim (2004) in pigs has been identifi ed (Wermter et al., 2008).It was shown that this polymorphism is associated with child and adolescent obesity.The study was performed on 1025 trio families and analysis of the allelic transmission pattern indicated the existence of polar overdominance in this locus.The biological function of this silent polymorphism is not known.Sequencing of human DLK1-MEG3 region did not reveal additional polymorphism, so further molecular analyses are needed to identify the biological mechanism underlying polar overdominant effects of DLK1 gene in pigs and humans.
Recently, Wolf et al. (2008) performed a genome-wide scan for imprinted QTL affecting growth in mice.The results suggested that imprinting patterns may be more diverse than previously assumed.Not only new patterns of imprinting (bipolar dominance and polar underdominance) but also changes in the pattern of imprinting during development have been described.Bipolar dominance is the expression pattern where the two heterozygotes differ from each other but the two homozygotes have similar phenotypes.In polar underdominance expression pattern is the same as in polar overdominance, but one of the heterozygotes exhibits lower intensity of the trait than the other three genotypes.In Wolf's study six loci were paternally expressed and only one maternally expressed.Additionally, three loci showed bipolar expression pattern, one -polar underdominance and three -polar overdominance.What is more, expression pattern of some loci shifted during development, for example the Wti2.1 locus showed bipolar dominance early in the development and changed to maternal expression by week 7.The loci identifi ed by the authors are probably novel imprinted genes as no currently known imprinted genes are located within the confi dence intervals of appropriate chromosomes (Wolf et al., 2008).OCZKOWICZ M. CONCLUSIONS Knowledge about the mechanism underlying polar overdominance is continuously increasing.Expression studies revealed that DLK1 gene is probably the primary effector and that RNA interference plays a key role in developing the callipyge phenotype.New genes that may contribute to this unusual phenotype in a response to changes in DLK1 expression have been identifi ed and need further studies.Finally, examination of new examples and patterns of expression in mammals may shed a new light on the molecular mechanism underlying this phenomenon.

Figure 1 .
Figure 1.A model for the callipyge polar overdominance according to Georges et al. (2003); description in the text