Associations between SNPs in candidate bovine imprinted genes and performance traits
The recent availability of whole genome sequences has highlighted the wealth of DNA sequence variation contained within mammalian genomes, the vast majority of which exists as SNPs. The abundance of these genetic polymorphisms coupled with their ease of detection (via DNA sequencing) and ease-of-genotyping has resulted in their adoption as the marker of choice for genotype-phenotype association analyses in livestock genetic studies . Indeed, the recent advent of high-throughput SNP genotyping platforms for livestock, such as the Illumina BovineSNP50 assay , has provided animal geneticists with vast quantities of data for association studies performed at a genome-wide level. However, a perceived drawback of such genome-wide association (GWA) studies is the detection of false-positive associations between a SNP and a trait-of-interest which can confound studies, particularly when an associated SNP occurs in a gene or region of the genome displaying no obvious biological connection to the trait . The detection and removal of spurious genotype-phenotype associations in GWA studies requires stringent statistical analysis involving the use of multiple-testing corrections; however, these can significantly reduce the number of associations reported in a study . Furthermore, it is becoming increasingly recognised that correcting for multiple tests using conventional methods can be too restrictive in genotype-phenotype association studies resulting in SNPs displaying true associations being overlooked [54–56].
A commonly used method to circumvent the detection of spurious genotype-phenotype associations is the adoption of candidate gene strategies whereby SNPs are pre-selected for association analyses based on their location within or proximal to genes/loci known to have a molecular role in regulating a phenotype of interest [55, 57–59]. Candidate gene approaches are also expected to reduce the number of false-negative genotype-phenotype associations (i.e. true associations that are erroneously rejected after rigorous statistical testing) that can also be generated in GWA studies [60, 61]. Consequently, in the present study, we have adopted a candidate gene approach by analysing genotype-phenotype associations between SNPs in a panel of eight putatively imprinted bovine genes, one of which (PEG3) has been previously shown to be subject to genetic imprinting in cattle. The remaining seven genes have been shown to be imprinted in at least one other mammalian species and therefore may be imprinted in cattle based on the appreciable conservation of imprinting between orthologs from different species .
Mammalian imprinted genes have been shown to play a pivotal role in mediating growth and development. This suggests that imprinted genes may serve as candidate loci harbouring potentially important DNA sequence polymorphisms contributing to heritable variation in livestock performance traits--a hypothesis that is supported by a number of recent genotype-phenotype association studies performed in domestic livestock populations [8, 21, 22, 24, 26, 27, 62]. In this study, significant phenotypic associations (P ≤ 0.05) were detected between SNPs located proximal to or within six of the eight candidate bovine imprinted genes analysed--CALCR, GRB10, PEG3, RASGRF1, ZIM2, and ZNF215--and range of cattle performance traits; significant associations (P ≤ 0.05) were not observed between performance traits and SNPs within the PHLDA2 and TSPAN32 genes, although one SNP within the bovine TSPAN32 gene showed a tendency to be associated (P ≤ 0.10) with a number of the performance traits assessed.
It should be stated that in this study we applied a Bonferroni correction  in an attempt to minimise the incidence of false-positive associations. However, none of the adjusted genotype-phenotype association P-values were significant at the P ≤ 0.05 level following this correction. Despite this, we believe that the uncorrected P-values ≤ 0.05 for the genotype-phenotype associations reported in this candidate gene study are supported by the molecular biological functions of the candidate bovine imprinted genes analysed in this study.
For example, the CALCR gene encodes the calcitonin hormone receptor protein--a seven-transmembrane receptor located on the surface of osteoclasts to which calcitonin binds activating adenylate cyclase leading to the inhibition of osteoclastic bone resorption . Previous studies have shown that SNPs in the porcine CALCR gene (whose imprinting status has yet to be defined, although preferential maternal expression of this gene has been reported in mouse brain tissue ) are associated with osteological development and growth performance in pigs [66, 67]. Notably, no significant associations were observed between CALCR SNP genotypes and the more direct measures of animal growth in this study (i.e. body depth, chest width, rump width, rump angle and animal stature). However, the associations between both bovine CALCR SNPs analysed and angularity and body condition (both of which are measures of subcutaneous fat levels in live animals) as detected here does suggest that the CALCR locus encompasses or is located proximal to a QTL that contributes to inter-animal differences in bovine body conformation traits, especially those related to fat deposition.
GRB10 (or maternally expressed gene 1 [MEG1]) encodes an adapter protein which is known to interact with certain tyrosine kinase receptors, such as insulin receptors and insulin-like growth factor receptors , and acts to restrict foetal and placental growth during mammalian development . This gene displays preferential maternal expression in the majority of mouse tissues examined to-date, with bi-allelic expression of the human GRB10 ortholog in corresponding human tissues and preferential paternal expression in human and mouse brain tissue . Furthermore, perturbations of the imprinting status/gene dosage of GRB10, whereby the maternal copy of the GRB10 gene has been duplicated, has been shown to result in severe pre- and post-growth retardation in mice . In this study, SNP genotype associations were observed between the bovine ortholog of this gene and angularity, body conditioning score and rump angle--traits related to animal development and growth. Based on these observations in cattle, it is possible that mutations in the GRB10 gene sequence alter the ability of the GRB10 protein in restricting foetal growth and development hence leading inter-individual differences in growth.
In mammals, both the PEG3 and ZIM2 genes form an imprinted gene cluster, a feature common to many imprinted genes . The PEG3 gene cluster is located on chromosomes 7 and 19 in mouse and humans, respectively, and consists of at least five differentially-imprinted genes, although analysis of this domain in human, mouse and cow has revealed some species-specific gene rearrangements . The paternally expressed PEG3 gene encodes a Krüppel-type zinc finger protein that may play a role in transcriptional regulation [72–74]. Also, the murine ortholog of this gene, Peg3, has been shown to be critical in cellular and behavioural functions including cellular proliferation, apoptosis and nurturing behaviour [40, 75]. The role of the maternally expressed ZIM2 gene is less well understood, but it has been shown to share at least seven upstream exons and a transcriptional start site with PEG3 in humans, suggesting some similarities for the function of the PEG3 and ZIM2 gene products . Two of the seven SNPs within the bovine PEG3 gene cluster were associated with animal stature, while one of these two SNPs was also associated with angularity, thus supporting a role in growth for this imprinted domain. In addition, three PEG3 domain SNPs were associated with perinatal mortality (with an additional two PEG3 domain SNPs displaying a tendency to be associated with this trait), while one PEG3 SNP was associated with gestation length, suggesting that the bovine PEG3 imprinted genes cluster underlies QTL for calf performance and fertility. Interestingly, aberrant methylation of the PEG3 gene (resulting in altered expression) has been observed in cases involving stillbirths and aborted foetuses in humans [76, 77] and aborted cloned bovine embryos , suggesting that this gene has an important role in embryo and foetal viability and survival.
One bovine RASGRF1 SNP was analysed in this study and it displayed associations with milk protein percentage and was the only analysed SNP to be associated with somatic cell count. RASFGR1 encodes the Ras protein-specific guanine nucleotide releasing factor 1 protein, which has been shown to play a role in signal transduction and growth and development in mice . Previous analyses performed by us identified this gene as being associated with growth traits in performance-tested Limousin cattle . Although no associations between the single analysed RASGRF1 SNP with growth were observed in the current study, the data presented here suggest that this gene may play a role in animal health as indicated by the association with somatic cell score--an often cited indicator of resistance to clinical and subclinical mastitis [80, 81]. It is unclear how RASGRF1 associates with resistance/susceptibility to mastitis; however, previous work has shown that expression of RASGRF1 affects the function of the growth hormone-insulin-like growth factor 1 (GH-IGF-1) axis , which can modulate the inflammatory response to mastitis .
Finally, we detected associations with a number of growth-related traits and the bovine ZNF215 gene, which encodes an alternatively spliced zinc-finger DNA binding protein that is localised in the nucleus. Moreover, the ZNF215 protein has been shown to contain both a Krüppel-associated (KRAB) box and SCAN box (i.e. SRE-ZBP; CT-fin51; AW-1; Number 18) amino-acid structural domains found Krüppel-like C2H2 zinc finger DNA binding proteins, both of which act to repress transcription [83–85]. In humans, ZNF215 is preferentially expressed from the maternally inherited allele and maps to an imprinted gene cluster on human chromosome (HSA) 11p15.5, the genomic region associated with Beckwith-Wiedemann syndrome (BWS)--a genetic disorder characterised by a range of growth abnormalities, including gigantism . It has been proposed that genetic rearrangements disrupt the normal functioning of the genes located within the imprinting domain on HSA11p15.5 (including ZNF215) resulting in the manifestation of the BWS phenotype; however, to-date, no functional ZNF215 mutations in BWS patients have been reported . In the current study, two SNPs within the bovine ZNF215 ortholog were analysed for associations with performance traits. Both SNPs displayed associations with animal stature and angularity while one ZNF215 SNP (rs42575474) was associated with milk protein percentage, culled cow and progeny carcass weight, body depth and rump width. These data suggest that DNA sequence variation within the bovine imprinting domain orthologous to HSA11p15.5 located on BTA15 may also harbour important quantitative trait nucleotides (QTNs) that similarly influence animal growth. Indeed, it is possible that mutations in the ZNF215 gene may alter the binding affinity of the ZNF215 protein to DNA sequences and hence alter the expression of other genes involved in animal growth and developmental pathways.
With the exception of the CALCR rs42940189 SNP (a non-synonymous mutation resulting in the substitution of an asparagine amino acid to an aspartic amino acid, both of which are small polar amino acid residues, at amino acid position 116 of the CALCR protein), the ORF gene model location of the remaining 16 SNPs analysed in this study (i.e. two upstream, five intronic, five synonymous coding and four non-coding 3'UTR SNPs) does not immediately suggest that these polymorphisms are functional. However, previous studies have shown that non-coding SNPs can have a regulatory function by altering the efficiency of DNA binding proteins that modulate gene expression. For example, a single G-to-A substitution within a non-coding regulatory region of the 3rd intron of the maternally imprinted porcine IGF2 gene has been shown to be the causal mutation for a QTL influencing muscle mass and fat deposition in pigs. It is postulated that the 'A' allele at this locus prevents the binding of a transcriptional repressor protein to the IGF2 gene sequence; hence individuals inheriting a sire-derived 'A' allele at this SNP display increased muscle mass and reduced fat content due to over-expression of paternally-derived IGF2 mRNA [21, 87].
3'UTR sequences of protein-coding mRNA transcripts have been shown to have an important function in regulating post-transcriptional process, such as the transportation of mRNA from the nucleus to cytoplasm, mRNA stability and the efficiency of protein translation [88, 89]. This has led some authors to suggest that 3'UTR sequences harbour potentially important DNA sequence variants influencing phenotypes in mammals . This assertion further supported by genetic data from livestock whereby 3'UTR SNPs have been shown to be associated with dairy performance traits in cattle [91, 92]. However, while it is tempting to speculate that the non-coding SNPs displaying associations with performance traits in the current study are causal it is more likely that these SNPs are associated (through LD) with causal regulatory mutations (or set of mutations) located proximal to, or within, the genetic loci studied that have not yet been identified.