The role of bovine causal genes underlying dairy traits in Spanish Churra sheep

Transcription

1 doi: /j x The role of bovine causal genes underlying dairy traits in Spanish Churra sheep M. García-Fernández*, B. Gutiérrez-Gil*, J. P. Sánchez, J. A. Morán, E. García-Gámez, L. Álvarez and J. J. Arranz Departamento de Producción Animal, Facultad de Veterinaria, Universidad de León, León, Spain Summary In dairy cattle, quantitative trait nucleotides (QTNs) underlying quantitative trait loci (QTL) for milk production traits have been identified in bovine DGAT1, GHR and ABCG2 genes. The SPP1 gene has also been proposed to be a regulator of lactation. In sheep, QTL underlying milk production traits have been reported only recently, and no proven QTN has been identified. Taking into account the close phylogenetic relationship between sheep and cattle, this study examined the possible effects of the aforementioned genes on sheep milk production traits. We first studied the genetic variability of the DGAT1, GHR, ABCG2 and SPP1 genes in 15 rams of the Spanish Churra dairy sheep breed. Second, we performed an association analysis between SNPs identified in these genes and three milk production traits recorded in a commercial population of Churra sheep. This analysis revealed only three significant associations at the nominal level (P-value <0.05) involving allelic variants of the ABCG2 gene, whereas no significant association was found for the DGAT1, GHR and SPP1 genes. When the Bonferroni correction was applied to take into account the multiple tests performed, none of the associations identified at the nominal level remained significant. Nevertheless, taking into account the high level of false-negative findings that can arise when applying the stringent Bonferroni correction, we think that our results provide a valuable primary assessment of strong candidate genes for milk traits in sheep. Keywords ABCG2, candidate gene, dairy sheep, DGAT1, growth hormone receptor. Introduction In recent years, genetic research in livestock has focused on mapping and characterizing genes and markers that control production and quality traits, with the aim of considering this molecular information for a more effective genetic selection. In dairy cattle, several studies have identified segregating quantitative trait loci (QTL) for milk production traits ( However, detection at this level is insufficient to efficiently incorporate QTL information into breeding programmes. Address for correspondence J.-J. Arranz, Departamento de Producción Animal, Facultad de Veterinaria, Universidad de León, León, Spain. jjarrs@unileon.es *These two authors have equally participated in this work. Accepted for publication 15 September 2010 Instead, identification of the specific causal polymorphism (quantitative trait nucleotide, QTN) or of markers in complete linkage disequilibrium with the QTN is required to provide a useful tool for selection. To date, few QTNs have been identified in ruminant species, although new genomic tools, e.g. the SNP-array platforms and the availability of the bovine genome sequence ( Bos_taurus/Info/Index), have increased the number of reported putative QTNs. The first QTN identified in dairy cattle was responsible for a QTL influencing milk fat and milk protein percentages that was previously identified in the centromeric region of bovine chromosome 14 (BTA14) (Grisart et al. 2002). The QTN involved a lysine to alanine substitution (p.lys232ala) at the gene encoding acyl CoA:diacylglycerol acyltransferase 1 (DGAT1) and explained approximately 50% of the genetic variation in milk fat percentage. Another QTL with a major effect on milk yield and milk composition was identified on BTA20. Using a fine-mapping strategy, Blott et al. (2003) suggested that a 415

2 416 García-Fernández et al. phenylalanine to tyrosine substitution (p.phe279tyr) at the bovine growth hormone receptor (GHR) gene was the direct cause of the effects previously detected. In other cases, conflicting results were initially reported in relation to a given genetic effect detected in dairy cattle populations (De Koning 2006). Hence, in the central region of BTA6, two different genes were initially proposed to harbour the corresponding causal mutation for the same QTL influencing milk yield and milk composition: the gene encoding breast cancer resistance protein (ABCG2) (Cohen-Zinder et al. 2005), and the osteopontin (SPP1) gene (Schnabel et al. 2005). Subsequent studies showed that the missense mutation p.tyr581ser (tyrosine to serine substitution) identified in the ABCG2 gene by Cohen-Zinder et al. (2005) was the genuine causal mutation of interest (Olsen et al. 2007; Seroussi 2009). Nevertheless, expression of the SPP1 gene has been found to influence the expression of milk protein genes, which suggests a regulatory role for the gene product of SPP1 in lactation (Sheehy et al. 2009). In sheep, published reports of QTL underlying milk production traits are quite recent (Gutiérrez-Gil et al. 2009; Raadsma et al. 2009) compared to those for cattle. Causal mutations have been identified in this species with respect to other traits such as prolificacy and muscle conformation (Davis 2004; Clop et al. 2006). However, to the best of our knowledge, no proven QTN has been identified for milk production traits in this dairy species to date. Taking into account the close phylogenetic relationship between sheep and cattle, we aimed to test whether the genes influencing the phenotypic variation of dairy traits in cattle have similar effects in sheep. With this purpose, we first studied the genetic variability of the DGAT1, GHR, ABCG2 and SPP1 genes in 15 rams of Spanish Churra sheep, a highly dairy-specialized ovine breed. Second, we performed an association analysis between some of the polymorphisms identified in these genes and three milk production traits recorded in a commercial population of Churra sheep. Materials and methods Animals To evaluate the genetic variability in the DGAT1, GHR, ABCG2 and SPP1 genes in a commercial population of Churra sheep, DNA samples were taken from 15 unrelated Spanish Churra rams. Some of the polymorphisms identified were genotyped across daughters of the sequenced rams, involving a half-sib population of 799 ewes distributed among 16 different flocks. The average family size was around 53, ranging from 29 (family 3) to 131 (family 5) daughters per sire. All animals belonged to the selection nucleus of the National Association of Churra Breeders (ANCHE) and were bred by artificial insemination. Phenotypic records Phenotypic records analysed in this study were three milk production traits considered in the current selection scheme for Churra dairy sheep: test day milk yield, test day protein percentage and test day fat percentage. Data for these traits were obtained from the records routinely collected in the official milk recording process. A total of milk yield records were considered, having an average of ml, while only 8393 records for fat and protein percentage were available, with averages of 6.61% and 5.57%, respectively. Sequencing analysis and SNP genotyping Several primer pairs were designed to amplify different regions of the four genes under study, with special attention given to the coding regions, based on the corresponding reference sequences obtained from GenBank (GenBank accession numbers EU for DGAT1 and AM for GRH) or the bovine genome sequence btau_4.0 (Ensembl identifier ENSBTAG for ABCG2 and ENSBTAT for SPP1). See Table S1 for further details about the primer sequences, expected sizes and annealing temperatures for the amplified fragments as well as the gene regions covered by the different amplicons. All of the amplicons were sequenced in the 15 Churra rams to identify polymorphisms. After PCR amplification, the different fragments were purified by ExoSAP-IT. Dideoxy sequencing reactions were later performed in both directions for each fragment with the BIGDYE TERMINATOR V3.1 Cycle Sequencing Kit using the same primers as used for fragment amplification. Sequencing reactions were purified using the CleanSEQ Ò reagent set and analysed on a 3130XL DNA Analyser. For the four genes studied, the proportions of the total gene sequence and the total gene coding region that were analysed by sequencing analysis are indicated in Table 1. Nucleotide diversity (p) was estimated from the average pairwise differences per site between sequences as described by Tajima (1983). For each gene under assessment, several SNPs were selected for an association analysis with milk production traits. The selection criteria were (i) a high level of variation shown among the 15 Churra rams initially analysed, with at least three sires segregating for the SNPs selected to be genotyped and (ii) the SNP position in the gene sequence, with the aim of reaching an even distribution of markers along the coding regions of the different genes. A set of 14 SNPs (3 for DGAT1, 2 for GHR, 6 for ABCG2 and 3 for SPP1), indicated in bold font in Table S2, were selected for genotyping across the population of 799 Churra ewes considered in this study and were subsequently analysed in relation to milk production traits. These SNPs were

3 Bovine causal genes in dairy sheep 417 Table 1 Summary of the variability of the DGAT1, GHR, ABCG2 and SPP1 genes in the 15 Churra rams analysed in this study. Total number of SNPs identified (SNPs within the gene coding region) Gene region analysed (bp) Gene size (bp) (Genbank reference sequence) Coding region (bp) Gene sequence analysed (%) Gene coding region analysed (%) Nucleotide diversity (p) Gene DGAT1 3 (2) 2732 Sheep 8676 (EU178818) GHR 13 (6) 3509 Cow (ENSBTAG ) ABCG2 30 (9) 7121 Cow (ENSBTAG ) SPP1 7 (2) 3020 Cow 8199 (ENSBTAG ) genotyped by KBioScience using their novel fluorescencebased competitive allele-specific PCR (KASPar) assay (for details, see Five of the 14 genotyped SNP markers were located within a gene coding region, with none of them causing an amino acid change. For all the genotyped SNPs, Hardy Weinberg equilibrium was tested considering the whole population genotyped using the Fisher exact test. This test served as a genotype quality control and to test for population stratification within the data set. Following standard guidelines for SNP association studies, those SNPs showing a minor allele frequency (MAF) <0.1 were not considered for further study (DGAT1_g.7255C>A and DGAT1_g.7486C>T). Based on this, a total of 12 SNPs (1 for DGAT1, 2 for GHR, 6 for ABCG2 and 3 for SPP1) were subjected to an association analysis with milk production traits. Association analysis To investigate the effects of the genotyped polymorphisms on milk production traits, the following mixed model was fitted individually for each trait and SNP locus of interest: y ijklm ¼ HTD i þ WIM j þ L k þ s l þ p m þ b SNP m þ e ijklm Where y ijklm is the dependent variable analysed (test day records for milk yield, milk fat percentage and milk protein percentage for the mth ewe). HTD i is the effect of the ith level of Herd-Test day combination (979 levels), WIM j is the effect of the jth class of week in milk (25 levels), L k is the effect of the kth level of lactation order (six levels, records from lactations >5 were all grouped together), s l is the random effect of the lth ram, sire of the mth ewe, p m is the random effect of the ewe m, b refers to the allele substitution effect for the particular SNP studied at this time and SNP m is referring to a covariate accounting for the number of copies of a particular allele that the animal m is carrying at the studied locus (e.g. 1, GG; 2, AG; or 3, AA). Variance components assumed for the random effects in the model were those used in the routine genetic evaluation of the breed for these traits, and they correspond to the following heritabilities: 0.13, 0.20 and 0.08, for milk yield, protein percentage and fat percentage, respectively. The corresponding repeatability values were 0.35, 0.30 and Given these genetic parameters, significance of the SNP effect was tested using an F test, after solving the previously described mixed model by a least squares procedure. The test was repeated for each of the studied loci and traits; solving mixed model equations and hypothesis testing were both conducted using R (R Development Core Team, 2008). To prevent false-positive findings, the Bonferroni correction for multiple testing was performed. By a principal component (PC) analysis, the number of independent traits was estimated as 3. Subsequently, for each set of genes within a chromosome, a PC analysis showed that the number of loci needed to explain 95% of the variance of the genotypes was 5 for chromosome 6 (OAR6), which includes ABCG2 and SPP1. For the other two chromosomes where the GHR and DGAT1 genes map (OAR16 and OAR9), the number of SNPs required was the same as those analysed (2 and 1, respectively). Finally, the number of independent tests was computed as the product of independent traits times the sum of independent SNPs across chromosomes. In our case, the number of independent tests was 24. Thus, for a false-positive rate of 0.05, our significance threshold must be set at Results The results of the variability identified in the studied genes are summarized in Table 1. The polymorphisms identified in each gene are described (SNP identification, location in the gene, amino acid change) in Table S2 according to the corresponding reference sequence. For the 14 SNPs genotyped across our resource population, the MAF observed ranged from 0.03 to 0.47 (Table S2). Based on the Fisher exact test, genotype frequencies at all loci were consistent with Hardy Weinberg equilibrium in the Churra population under study (data not shown).

4 418 García-Fernández et al. The association analysis performed for the 12 candidate SNPs genotyped that showed MAF > 0.1, and the three milk production traits considered in this study revealed four SNP-trait combinations with an associated nominal P-value <0.05 (Table 2). All these associations involved polymorphisms identified in the ABCG2 gene. Hence, from the nine SNPs analysed for ABCG2, the SNPs located in Intron 3 (ABCG2_c G>A), Exon VI (ABCG2_c.711G>A) and Exon 9 (ABCG2_c.1128C>T) influenced at the nominal level milk fat percentage. When the Bonferroni correction was applied to take into account the multiple tests performed, none of these associations identified at the nominal level remained significant, as none reached the corrected threshold (P-value <0.002). The estimated effects for the nominal associations identified for ABCG2 were of similar magnitude (about SD units), with the least frequent allele for SNP ABCG2_c.1128C>T (Exon 9) increasing the milk fat content and the least frequent alleles for ABCG2_c G>A (Intron 3) and ABCG2_c.711G>A (Exon 4) decreasing the milk fat content. Additionally, marker ABCG2_c.711G>A (Exon 4) also showed a nominal P-value <0.05 in relation to milk yield, with the least frequent allele increasing the milk yield by approximately 0.08 SD units (Table 2). None of the SNPs analysed for DGAT1, GHR and SPP1 genes showed a nominal P-value <0.05. Discussion Many QTL influencing a myriad of phenotypic traits have been mapped in the broadly distributed Holstein dairy cattle population. However, a limited number of genes, including DGAT1 (Grisart et al. 2002), GHR (Blott et al. 2003) and ABCG2 (Olsen et al. 2007), have accumulated compelling evidence to be considered genes harbouring a causal mutation or QTN. In dairy sheep, only a few recent studies have reported QTL in local breeds or crossed populations (Gutiérrez-Gil et al. 2009; Raadsma et al. 2009), and no QTN underlying milk traits has been proposed in this small ruminant species to date. Although major advances are currently taking place in the field of sheep genomics, significant limitations have thus far hindered identification of QTNs in dairy sheep (e.g. limited number of described markers, incomplete sheep genome sequence, small size of populations with routinely recorded phenotypic data, etc.). Building on the advances previously reported in dairy cattle, this study looked for possible relationships between milk production traits and allelic variants in the aforementioned causal genes in a commercial population of Spanish Churra dairy sheep. In addition, despite the fact that SPP1 was ruled out in favour of ABCG2 as the causal gene for a dairy QTL located in BTA6 (Olsen et al. 2007; Seroussi 2009), we also included this gene in our analysis because in vitro functional studies have shown that the product of SPP1 plays a significant role in the modulation of milk protein gene expression (Sheehy et al. 2009). We initially estimated the genetic variability of the four analysed genes in 15 rams of the selection nucleus of the Churra sheep breed. A total of 53 SNPs were identified, with ABCG2 and GHR showing the highest SNP frequency in the analysed sequences and DGAT1 showing the lowest variability. To test the possible association of the candidate genes considered in this work with dairy production traits measured in a commercial population of 799 Churra ewes, 12 of the identified SNPs were analysed in relation to three milk production traits following a repeatability animal model. Although random regression models fit better for test day measurements because of a better consideration of the covariance between records, this kind of model remains undeveloped in dairy sheep. In addition, in studies in goats, very little advantage in terms of breeding value predictions has been achieved using random regression models over the repeatability animal model (Menéndez-Buxadera et al. 2010). Furthermore, for the purpose of this study (gene association analysis) by considering sire and permanent effects, the most relevant correlations between records (because of relationships and repeated records within an animal) are considered. The analysis performed here has revealed four associations with a nominal P-value <0.05 for SNPs identified in the ABCG2 gene. The three SNPs showing significant nominal associations in Churra sheep are located in Intron 3, Exon 6 and Exon 9 of the ABCG2 gene and show a high level of LD (with a pairwise r 2 estimated with Haploview Table 2 Trait SNP combinations showing a nominal P-value <0.05 in the association analysis performed in this study. Together with the nominal P-values obtained in the analysis, the magnitude of the allelic substitution effect and the corresponding standard errors expressed in units of the trait are indicated in brackets. The magnitudes of the estimated effects expressed in phenotypic standard units of the corresponding trait are also given in square brackets. ABCG2_c G>A Intron 3 ABCG2_c.711G>A Exon 6 ABCG2_c.1128C>T Exon 9 Milk protein percentage (g/100 g) Milk fat percentage (g/100 g) ()0.097 ± 0.036) [)0.07] 0.01 ()0.092 ± 0.035) [)0.066] Milk yield (ml/day) ( ± ) [0.079] (0.089 ± 0.036) [0.064]

5 Bovine causal genes in dairy sheep 419 (Barret et al. 2005) ranging from 0.97 to 1). These three SNPs influenced milk fat percentage, whereas that located in Exon 6 also showed certain influence on milk yield. In cattle, the p.tyr581ser mutation in the ABCG2 gene exerts its main effect on milk protein and milk fat content and affects milk yield to a lesser extent. Hence, our results may support a conserved influence of the ABCG2 gene on milk fat content in sheep as well as cattle, perhaps related to the cholesterol transport role that has been suggested for this gene by several authors (Cohen-Zinder et al. 2005; Sheehy et al. 2009). Although Duncan et al. (2007) described a significant association between a SNP in Intron 4 of ABCG2 and resistance to facial eczema disease, to the best of our knowledge, the present study is the first report to evaluate the effects of ABCG2 on sheep milk production traits. None of the nominal significant associations reported here was significant when the Bonferroni correction was applied. Because of this, as well as the limited size of the resource population analysed, we acknowledge the preliminary nature of these results and the need to confirm the significant associations identified here at the nominal level by independent studies before attempting their practical implementation in sheep breeding programmes. Furthermore, even if confirmed, functional studies would be required to understand the molecular and physiological mechanisms underlying such associations. Nevertheless, and taking into account the high level of false negative that can arise when applying the stringent Bonferroni correction, we think that our results provide a valuable primary assessment of strong candidate genes for milk traits in sheep and may pave the way towards targeted marker-assisted programmes that would improve the dairy sheep industry. We did not find any significant association between Churra milk production traits and three of the genes that are known to influence these traits in cattle, DGAT1, GHR and SPP1 (Grisart et al. 2002; Blott et al. 2003; Sheehy et al. 2009). This may suggest that despite the close phylogenetic relationship between cattle and sheep, the genetic architecture of milk production in these two ruminant species may involve different underlying factors. Alternatively, the natural and/or artificial selection processes that have occurred in these two species might have caused different functional mutations to become fixed. The differences observed between cattle and sheep with respect to functional allelic variants highlight the importance of current research projects focused on the sheep genome (The International Sheep Genomics Consortium in press). Progress in sheep genomics will ensure that future analyses will not be dependent on candidate genetic information derived from cattle genomic studies. Acknowledgements This work was supported by the Spanish Ministry of Science (Project AGL ) and a grant (Project GR43) for excellent research groups from the Castilla and León regional government (Junta de Castilla y León). M. García- Fernández and B. Gutiérrez-Gil are funded by the Spanish Ministry of Science (FPI and ÔJuan de la CiervaÕ programmes, respectively). E. García-Gámez is funded by the Spanish Ministry of Education (FPU programme). References Barret J.C., Fry B., Maller J. & Daly M.J. (2005) Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 21, Blott S., Kim J.J., Moisio S. et al. (2003) Molecular dissection of a quantitative trait locus: a phenylalanine-to-tyrosine substitution in the transmembrane domain of the bovine growth hormone receptor is associated with a major effect on milk yield and composition. Genetics 163, Clop A., Marcq F., Takeda H. et al. (2006) A mutation creating a potential illegitimate microrna target site in the myostatin gene affects muscularity in sheep. Nature Genetics 38, Cohen-Zinder M., Seroussi E., Larkin D.M. et al. (2005) Identification of a missense mutation in the bovine ABCG2 gene with a major effect on the QTL on chromosome 6 affecting milk yield and composition in Holstein cattle. Genome Research 15, Davis G.H. (2004) Fecundity genes in sheep. Animal Reproduction Science 82 83, De Koning D.J. (2006) Conflicting candidates for cattle QTLs. Trends in genetics 22, Duncan E.J., Dodds K.G., Henry H.M., Thompson M.P. & Phua S.H. (2007) Cloning, mapping and association studies of the ovine ABCG2 gene with facial eczema disease in sheep. Animal Genetics 38, Grisart B., Farnir F., Karim L., Mni M., Simon P., Taylor J.F., Vanmanshoven P., Wagenaar D., Womack J.E. & Georges M. (2002) Positional candidate cloning of a QTL in dairy cattle: identification of a missense mutation in the bovine DGAT1 gene with major effect on milk yield and composition. Genome Research 12, Gutiérrez-Gil B., El-Zarei M.F., Alvarez L., Bayón Y., de la Fuente L.F., San Primitivo F. & Arranz J.J. (2009) Quantitative trait loci underlying milk production traits in sheep. Animal Genetics 40, Menéndez-Buxadera A., Molina A., Arrebola F., Gil J.M. & Serradilla J.M. (2010) Random regression analysis of milk yield and milk composition in the first and second lactations of Murciano-Granadina goats. Journal of Dairy Science 93, Olsen H.G., Lien S., Gautier M., Nilsen H., Roseth A., Berg P.R., Sundsaasen K.K., Svendsen M. & Meuwissen T.H.E. (2007) Genetic support for a quantitative trait nucleotide in the ABCG2 gene affecting milk composition of dairy cattle. BMC Genetics 8, 32. R Development Core Team (2008) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria, Raadsma H.W., Jonas E., McGill D., Hobbs M., Lam M.K. & Thomson P.C. (2009) Mapping quantitative trait loci (QTL) in sheep. II. Meta-assembly and identification of novel QTL for milk production traits in sheep. Genetics Selection Evolution 41, 45.

6 420 García-Fernández et al. Schnabel R.D., Kim J.J., Ashwell M.S., Sonstegard T.S., Van Tassell C.P., Connor E.E. & Taylor J.F. (2005) Fine-mapping milk production quantitative trait loci on BTA6: analysis of the bovine osteopontin gene. Proceedings of the National Academy of Sciences of the United States of America 102, Seroussi E. (2009) The concordance test emerges as a powerful tool for identifying quantitative trait nucleotides: lessons from BTA6 milk yield QTL. Animal Genetics 40, Sheehy P.A., Riley L.G., Raadsma H.W., Williamson P. & Wynn P.C. (2009) A functional genomics approach to evaluate candidate genes located in a QTL interval for milk production traits on BTA6. Animal Genetics 40, Tajima F. (1983) Evolutionary relationship of DNA sequences in finite populations. Genetics 105, The International Sheep Genomics Consortium, Archibarld A.L., Cockett N.E., Dalrymple B.P., Faraut T., Kijas J.W., Maddox J.F., McEwan J.C., Oddy V.H., Raadsma H.W., Wade C., Wang J., Wang W. & Xun X. The sheep genome reference sequence: a work in progress. Animal Genetics 41, Supporting information Additional supporting information may be found in the online version of this article. Table S1 Details of primers and annealing temperatures used in this study. Table S2 List of SNPs identified in the DGAT1, GHR, ABCG2 and SPP1 genes in 15 rams of the Churra sheep breed. As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer-reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.