Deciphering signature of selection affecting beef quality traits in Angus cattle

Abstract Artificial selection towards a desired phenotype/ trait has modified the genomes of livestock dramatically that generated breeds that greatly differ in morphology, produc- tion and environmental adaptation traits. Angus cattle are among the famous cattle breeds developed for superior beef quality. This paper aimed at exploring genomic regions under selection in Angus cattle that are associated with meat quality traits and other associated phenotypes. The whole genome of 10 Angus cattle was compared with 11 Han- woo (A-H) and 9 Jersey (A-J) cattle breeds using a cross- population composite likelihood ratio (XP-CLR) statistical method. The top 1% of the empirical distribution was taken as significant and annotated using UMD3.1. As a result, 255 and 210 genes were revealed under selection from A–H and A–J comparisons, respectively. The WebGestalt gene ontology analysis resulted in sixteen (A–H) and five (A–J) significantly enriched KEGG pathways. Several pathways associated with meat quality traits (insulin signaling, type II diabetes mellitus pathway, focal adhesion pathway, and ECM-receptor interaction), and feeding efficiency (olfactory transduction, tight junction, and metabolic pathways) were enriched. Genes affecting beef quality traits (e.g., FABP3, FTO, DGAT2, ACS, ACAA2, CPE, TNNI1), stature and body size (e.g., PLAG1, LYN, CHCHD7, RPS20), fertility and dystocia (e.g., ESR1, RPS20, PPP2R1A, GHRL, PLAG1),feeding efficiency (e.g., PIK3CD, DNAJC28, DNAJC3, GHRL, PLAG1), coat color (e.g., MC1-R) and genetic dis- orders (e.g., ITGB6, PLAG1) were found to be under positive selection in Angus cattle. The study identified genes and pathways that are related to meat quality traits and other phenotypes of Angus cattle. The findings in this study, after validation using additional or independent dataset, will pro- vide useful information for the study of Angus cattle in par- ticular and beef cattle in general.

Intensive artificial selection within and between breeds of livestock made the development of specialized breeds that are able to produce the intended amount and quality of prod- uct a reality. Artificial selection concentrates the genetics of certain individuals that causes differences in the specific patterns of change in allele frequencies, diversity, and haplo- type structure that in turn differentiate breeds under selection from others. In beef cattle, such kind of differential selection have resulted in several breeds of high growth rate, superior beef quality, and higher feed efficiency (Albertí et al. 2008). Apart from its positive impact in improving production and productivity of commercial traits, intensive artificial selection towards a particular trait has caused in several genetic disorders in several beef and dairy cattle breeds (Whitlock et al. 2008). Genetic disorders result in high mor- tality and reduced reproduction and productivity of herds, and greatly impact the profitability of the farm (Ciepłoch et al. 2017). This is because, genes causing genetic disor- ders are linked with those genes affecting economic traits of interest. Ciepłoch et al. (2017), reviewed several genetic disorders of beef cattle that are potentially caused by human artificial selection. In Angus cattle populations, dwarfism and fawn calf syndrome (Whitlock et al. 2008), arthrogrypo- sis multiplex, neuropathic hydrocephalus, and osteopetrosis (Whitlock 2010) have been reported in Australia and the US. Cardiomyopathy is a genetic disorder that has been reported in Holstein cattle (Guziewicz et al. 2007).

Angus cattle (Aberdeen Angus) are known cattle breeds developed around the early nineteenth century in Northeast Scotland. During the twentieth century, breeders made enor- mous changes in the growth, stature and body composition of American Angus cattle through selection (Arthur et al. 2001; McClure et al. 2010). The breed is characterized by its high muscularity, higher growth rate (ADG), wide pelvis and medium height and high level of beef fat (Albertí et al. 2008). Angus cattle are early finishing with high growth rate, eye muscle and yield (Chambaz et al. 2003). They are natu- rally polled and predominantly black or red in color (http:// www.thecattl that of natural selection, artificial selection towards a particular trait is expected to leave a distinctive signature on the genome that can be traced using genomic and bioin- formatics methods. Identification of signature of selection is used to pinpoint the adaptive events that have generated the enormous phenotypic variation observed between cattle breeds and has a biotechnological relevance (Utsunomiya et al. 2014). Recently, several methods have been developed and applied to scan for footprints of selection in several spe- cies and breeds of animals. The Bovine HapMap Consor- tium (2009), used iHS, FST and CLR methods to identify ongoing selection due to domestication, breed formation,and ongoing selection intended to enhance performance and productivity. In this study, we used the cross-population composite likelihood ratio (XP-CLR) test (Chen et al. 2010), a population differentiation method, to explore the genetic background contributing to the superior meat quality char- acteristics and associated genetic disorders in Angus cattle.

In this manuscript, we used previously processed whole genome sequencing data where detailed sample informa- tion and resequencing procedures can be found (Kim et al. 2017). DNA samples of Angus and Jersey cattle breeds were obtained from the Institutional Animal Care and Use Com- mittee of the National Institute of Animal Science, Korea. For Hanwoo cattle, blood samples were collected from Hanwoo Improvement Center of the National Agricultural Cooperative Federation, and DNA was extracted using a G-DEXTMIIb Genomic DNA Extraction Kit (iNtRoN Biotechnology, Seoul, Republic of Korea) according to the manufacturer’s protocol. The DNA was checked for its qual- ity and inserts of ~ 300 bp was generated from a randomly sheared 3 μg of genomic DNA. The fragments of sheared DNA were end-repaired, A-tailed, adaptor ligated, and amplified using a TruSeq DNA Sample Prep. Kit (Illumina, San Diego, CA, USA). Paired-end sequencing was con- ducted using the Illumina HiSeq2000 platform with TruSeq SBS Kit v3-HS (Illumina) .Sequence reads were mapped against the reference bovine genome (UMD 3.1) using Bowtie2 (Langmead and Salzberg 2012). The overall alignment rate of reads to the reference sequence was 98.84% with an average read depth of 10.8×, and the reads covered 98.56% of the reference UMD3.1 genome (Kim et al. 2017).Open-source software packages were used for down- stream processing and variant calling.

Picard (https://broa- filtered potential PCR dupli- cates, and SAMtools (Bindea et al. 2009) created index and bam files. Genome analysis toolkit 3.1 (GATK) (McKenna et al. 2010) was used to perform local realignment of reads. The “UnifiedGenotyper” and “SelectVariants” arguments of GATK was used to call candidate SNPs. In order to fil- ter variants and avoid possible false positives, the “Vari- antFiltration” argument of the same software was adopted with the following options: (1) SNPs with a phred-scaled quality score of less than 30 were filtered; (2) SNPs with MQ0 (mapping quality zero; total count across all sam- ples of mapping quality zero reads) > 4 and quality depth (unfiltered depth of non-reference samples; low scores are indicative of false positives and artifacts) < 5 were filtered; and (3) SNPs with FS (Phred-scaled P value using Fisher’s exact test) > 200 were filtered since FS represents variation on either the forward or the reverse strand, which is indica- tive of false positive calls. For the haplotype information on each chromosome, BEAGLE (Browning and Browning 2007) was used to infer the haplotype phase and impute missing alleles for the entire set of cattle populations simul- taneously. Sequences used for this study are available from GenBank with the Bio project accession number of Angus (PRJNA318087), Jersey (PRJNA318089), and Hanwoo (PRJNA210523).

We used SNPhylo pipeline (Lee et al. 2014b), using autoso- mal SNPs, to construct a phylogenetic tree to understand the relationship between breeds. SNPhylo uses the SNPRelate (Zheng et al. 2012) package to check and filter for qual- ity of SNPs applying minor allele frequency (MAF) and missing rate threshold, and make use of linkage disequilib- rium; MUSCLE (Edgar 2004) for multiple sequence align- ment; and PHYLIP package (Plotree and Plotgram 1989) to determine the phylogenetic tree by a maximum likeli- hood method. In the SNPhylo pipeline, we used the options of 1000 bootstrapping samples, 29 autosomes, and MAF threshold of 0.05. A total of 14,049 SNPs were selected and used for the phylogenetic tree construction. We visualized the phylogenomic tree using FigTree ( uk/software/figtree/).In addition, analysis of population structure was per- formed in a Bayesian model-based analysis using STRUC- TURE software (Evanno et al. 2005) to identify groups of individuals corresponding to the uppermost hierarchical levels. STRUCTURE assumes a model in which there are K populations (clusters), which contribute to the genotype of each individual characterized by a set of allele frequencies at each marker locus. The software applies a Markov Chain Monte Carlo (MCMC) estimation of allele frequencies in each of the K populations and the degree of admixture for each animal. The number of clusters were inferred using the options of length of Burnin Period of 2000, the number of MCMC Reps after Burnin period of 200,000 and MAF of 0.05. We used PLINK (Purcell et al. 2007) to generate input files used by STRUCTURE using -thin option that retained 4008 loci. We also calculated a population differentiation index (FST), with 50 kb windows of 5 kb steps, between the populations considered using VCFtools to understand the genetic distance between them.

We performed an XP-CLR statistical test (Chen et al. 2010), which implements composite likelihood methods for detecting selective sweep genomic regions differentially selected between two populations. It is a multi-locus slid- ing window based test that detects recent selective sweep regions based on allele frequency differentiation between two populations. Here, we carried out pairwise compari- sons of the genomes of 10 Angus with 11 Hanwoo (A–H) and 9 Jersey (A–J) cattle breeds using the XP-CLR software package (Chen et al. 2010). Angus cattle are a specialized beef breed native to Aberdeenshire in Scotland (Arthur et al. 2001; McClure et al. 2010). The Hanwoo cattle are a result of interbreeding between taurine and zebu cattle and its his- tory as a draft animal dates back at least 5000 years, and since recently, it has been intensively selected for high beef quality (Chung 2014). Jersey cattle are a small dairy cattle known for their better milk quality (The Bovine HapMap Consortium 2009). Since these populations have different demographic and selection histories, comparing Angus cat- tle with Hanwoo and Jersey breeds could help uncover the signature of selection related to meat quality traits and other specific phenotypes of Angus cattle.

In order to calculate XP-CLR values, we followed a previously used procedure where non-overlapping sliding windows of 50 kb and a maximum number of 600 SNPs within each window were used (Kim et al. 2017). Accord- ing to the software, a weighted composite likelihood ratio (CLR) scheme was adopted in estimating XP-CLR—the pairwise correlation coefficient (r2) of SNPs from the refer- ence population are used to give weights. When the cor- relation coefficients are greater than 0.95, CLR scores for the two SNPs are down-weighted. Finaly, the top 1% (0.01) of the empirical distributions were designated as candidate sweeps and genes that span the window regions were defined as candidate genes (Kim et al. 2017). Significant genomic regions identified from both comparisons were annotated based on UMD 3.1.WEB-based GEne SeT AnaLysis Toolkit (WebGestalt), which integrates data from centrally and publicly curated databases as well as computational analyses (Wang et al. 2013b), was used for gene enrichment analysis. Gene sets identified from each of the comparisons were submitted separately to identify significantly enriched gene ontology biological process (GO-BP) terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. We employed a hypergeometric statistical method and Bonferroni multiple adjustment using a Homo sapiens genome as a reference set.
Additionally, ClueGo, a Cytoscape plug-in, was used to integrate GO BP terms and KEGG pathways in order to functionally organize into GO/pathway term networks (Bindea et al. 2009). Finally, we used the SNPEff variant annotation and effect prediction tool (Cingolani et al. 2012) to annotate and predict the effects of genetic variants (such as amino acid changes) for the genes considered as candi- date genes. We have drawn the Manhattan plot of the −log10 transformed XP-CLR P values from both comparisons using R software. The gene names and descriptions used are based on genecards (

Result and discussion
DNA samples collected from 10 Angus, 11 Hanwoo, and 9 Jersey cattle breeds sequenced to ~ 11× genome coverage each was used for the study. After standard data prepara- tion and re-sequencing procedures, an average alignment rate of 98.84% covering 98.56% of the taurine reference genome (UMD 3.1) was obtained. Using several methods and software potential PCR duplicates and false positive calls were filtered, and finally, a total of ~ 37 million SNPs were obtained and used for further analysis.We constructed a non-rooted Phylogenetic tree of sample populations using SNPhylo (Lee et al. 2014b). As a result, individual animals within breeds clustered together sepa- rately from individuals of other breeds (Fig. 1a). Similarly,to understand the admixture level of sample populations, STRUCTURE (Evanno et al. 2005) was used to construct structure at 2 and 3 population assumptions (Fig. 1b). At ancestral populations (K) of 2, Angus and Hanwoo breeds clustered together separately from Jersey whereas, at K = 3, all the three breeds became different even though Angus showed some sort of admixture with Hanwoo and Jersey. The Weir and Cockerham weighted FST estimates between A–H (0.12977), A–J (0.19003), and H–J (0.20374), confirmed the distance of Jersey from both of the breeds. His- torically, three of the breeds are originated separately and developed in different areas for different purposes. The Angus (Arthur et al. 2001; McClure et al. 2010) and Hanwoo (Chung 2014) are beef breeds whereas Jersey (The Bovine HapMap Consortium 2009) are dairy cattle.In order to infer the positive selection signature of gene regions that are related to the phenotypes of Angus cattle, we compared the genomes of Angus cattle with Hanwoo (A–H) and Jersey (A–J) cattle breeds using XP-CLR statis- tics following previous procedures (Kim et al. 2017). XP- CLR compares allele frequency differentiation between two populations (Chen et al. 2010). The Manhattan plot of the −log10 transformed XP-CLR score P values of both com- parisons are presented in Fig. 2. Then, by annotating the top 1% outlier regions of the empirical distribution, 255 and value, and x-axis shows chromosomal positions. The horizontal dot- ted lines represent the 1% outlier regions in both of the comparisons 210 genes were identified from A–H and A–J comparisons, respectively (Fig. 1c; Additional file 1: Table S1 and S2).

To further understand the biological functions of the genes identified from both comparisons, we submitted the gene lists to WebGestalt gene ontology analysis tool (Wang et al. 2013b). Here, we used the KEGG pathway and GO-BP terms analysis. As a result, sixteen and five KEGG pathways were significantly enriched (adj P value < 0.05) from A–H and A–J gene lists, respectively (Fig. 3; Additional file 2: Table S3). However, no intersecting GO-BP terms were found significant (adjP value > 0.05) from the comparisons. In this study, we described those pathways and genes that affect the phenotypes of Angus cattle based on literature and their biological functions (Table 1).
Pathways and genes related to meat quality traits Meat quality is a complex and multi-factorial trait affected by genetic and non-genetic factors. Genes that are involved in different biological and cellular functions including mus- cle growth, glycolysis, adipogenesis, muscle contraction, stress reaction, proteolysis, and apoptosis influence meat quality traits such as intramuscular fat (IMF), tenderness and drip loss (Ladeira et al. 2016). In this study, significantly enriched pathways affecting meat quality traits include insulin signaling (A–H), type II diabetes mellitus pathway (A–H), focal adhesion pathway (both comparisons) and ECM-receptor interaction (A–H) (Fig. 3). These pathways affect the amount, distribution, and composition of fat in meat which is a determinant factor for its quality (Ladeira et al. 2016). Insulin signaling and type II diabetes mellitus pathways are related to adipogenesis and IMF deposition (Cui et al. 2012a; Ladeira et al. 2016).

Insulin stimulates the expression of genes that encode lipogenic enzymes in the adipose tis- sue (Ladeira et al. 2016). Focal adhesions are related to cell junction that connects the cytoskeleton of a cell to the ECM (Li et al. 2010). It is related to lipid metabolism and influences IMF deposition (Cui et al. 2012a). Genes in this pathway (ACTB, COL5A2, MYL9, PARVA, PIK3CD, and
TLN2) are involved in muscle development and influences tenderness and texture of meat (Li et al. 2010; Cui et al. 2012a; Lee et al. 2013). ECM-receptors have an important role in adipogenesis and meat tenderness (Li et al. 2010). It has been found enriched from genes differentially expressed from fat depots of omental, subcutaneous and intramuscu- lar fat (Hausman et al. 2009; Cui et al. 2012a; Lee et al. 2013). Genes involved in ECM-receptor interaction were previously found significantly up-regulated in subcutane- ous fat and intramuscular fat (Lee et al. 2013). The positive selection of genes related to meat quality traits in Angus cattle as compared to Hanwoo cattle might be because of the differences in demographic and selection histories that differentiate the allele frequency spectrum between them (Utsunomiya et al. 2014).Regulation of eicosanoid secretion was significantly enriched (P < 0.05) in the ClueGo network of A–J gene list (Fig. 4b). Eicosanoids are signaling molecules derived from enzymatic or non-enzymatic oxidation of essential fatty acids like arachidonic acid (Madsen et al. 2005). Fatty acids influence adipogenesis as precursors for the eicosanoids, as well as regulators of transcription. PLA2R1, involved in this network, is a fatty acid transporter that induces cell prolif- eration and serves as a receptor of a phospholipase for the production of lipid mediators. It has been found associated with fat deposition, body weight and egg production perfor- mance in chicken (Gheyas et al. 2015). Besides pathways, FABP3 and TNNI1genes were identi- fied in A–H comparison in relation to meat quality traits (Table 1). FABP3 is a protein-coding gene that plays a role in intracellular transport of long-chain fatty acids and their acyl-CoA esters. Polymorphisms in FABP3 gene has been found associated with intramuscular fat and fatty acid com- position in porcine meat (Puig-Oliveras et al. 2016), cyto- solic fatty acid and lipid binding (Berton et al. 2016), and beef ribeye area and ribeye area to hot carcass weight ratio(Blecha et al. 2015). TNNI1, a gene expressed in the slow- twitch skeletal muscle fibers, is a constituent protein of the troponin complex located on the thin filaments of striated muscle to which its expression affects meat quality traits through its effect on muscle fiber composition (Yang et al. 2010). Polymorphisms in the gene region is associated with intramuscular fat, marbling score and pork color in Large- White Meishan pigs (Yang et al. 2010), drip loss and com- pression force in Mong Cai pigs (Ngu and Nhan 2012), and pH24 and drip loss of Longissimus Dorsi muscle in pork (Pierzchala et al. 2014).In the A-J comparison, ACS, ACAA2, FTO, DGAT2, and CPE genes were identified in relation to meat quality (Table 1). Acetyl-CoA (ACS, ACAA2) genes are essential for de novo fatty acid synthesis (Ladeira et al. 2016). These genes play a role in the activation of long-chain fatty acids for the synthesis of cellular lipids and degradation via beta- oxidation. ACS/ACSL1, called Long-Chain-Fatty-Acid- CoA Ligase 1, has been found differentially expressed and upregulated for saturated fatty acids (SFA—palmitic, stearic, oleic fatty acids) in the Longissimus Dorsi muscle of high- fat Nellore cattle (Berton et al. 2016). ACAA2, acetyl-CoA acyltransferase 2, is expressed in the subcutaneous fat tissue of beef cattle involved in adipogenesis (Wang et al. 2013a). Screening for non-synonymous mutations representing puta- tive functional variants, four (one known—rs211177037; and three new—27:14225708, 27:14245782, 27:14252901) and two new (24:49919351, 24:49933799) missense vari- ants were identified in ACS/ACSL1 and ACAA2 gene regions, respectively (Additional file 2: Table S4). FTO is a nuclear protein of the AlkB related non-haem iron and 2-oxoglutarate-dependent oxygenase superfamily that has a role in the regulation of global metabolic rate, energy expenditure and homeostasis, body size and fat accu- mulation, and control of adipocyte differentiation into fat cells. It is found highly expressed in fat tissue contributing to fattier phenotype in pigs (Tempfli et al. 2016), associated with fatness-related traits such as intramuscular fat deposi- tion and back fat thickness in beef cattle (Wei et al. 2011) and associated with marbling score in Hanwoo cattle (Chung 2014). In this gene region, one novel (14:22243331) and one known (rs381025074) missense variants were identified (Additional file 2:Table S4).DGAT2 encodes an enzyme which catalyzes the synthe- sis of triglycerides. It has been found associated with the accumulation of SFA in the intramuscular tissue of Nel- lore cattle with extreme values of fatty acid (Berton et al. 2016), and Korean beef cattle (Jeong et al. 2012). A new missense variant (15:55971894) was identified in this gene. CPE is an enzyme involved in the production of neuroen- docrine peptide hormones and neuropeptides including insulin, vasopressin, and oxytocin. Mutation in this gene is associated with obesity and infertility in mice (Naggert et al. 1995), marbling score and breeding value of back fat thickness in Hanwoo beef cattle (Shin and Chung 2007), and found upregulated for SFA in the Longissimus Dorsi muscle of high-fat Nellore cattle (Berton et al. 2016). Two novel (17:680132, 17:692409) and one known (rs210567645) mis- sense variants were identified in this gene region. DNAJ genes (DNAJC28, DNAJC3) have an anti-apoptotic role that is important for meat tenderness (O’brien et al. 2014). The expression of a gene family (DNAJA1—not iden- tified here) has been reported to characterize 60% of the variations in meat tenderness of Charolais cattle (Bernard et al. 2007). MYL9 is a gene related to muscle biology and accretion. O’brien et al. (2014), reported the positive selec- tion of this gene in Angus cattle. In the gene regions, five (DNAJC3), five (DNAJC28) missense variants were identi- fied (Additional file 2: Table S4).The positive selection of these putative genes and path- ways enriched might contribute to the superior beef quality of Angus cattle. Angus cattle are known beef breed devel- oped for its quality, higher growth rate and feed conversion efficiency (Arthur et al. 2001; McClure et al. 2010).Body size in cattle, as measured by body weight, is an important trait in meat production. It is influenced by many genes of smaller effect size (Kemper et al. 2012). In relation to this, several genes including PLAG1 (A–H), RPS20 (A–J), LYN, and CHCHD7 (both) were identified in this study (Table 1). PLAG1 encodes a zinc finger protein whose acti- vation results in up-regulation of genes that control growth leading to cell proliferation. Its association with height and body weight variation in several cattle breeds (Littlejohn et al. 2012; Utsunomiya et al. 2013; Takasuga 2016; Fink et al. 2017), and carcass weight in Japanese Wagyu cattle (Nishimura et al. 2012) was previously reported. LYN is a protein kinase gene that regulates cell proliferation, sur- vival, differentiation, migration, adhesion, degranulation, and cytokine release. RPS20 also is a ribosomal gene that catalyzes protein synthesis. Both, LYN and RPS20 genes have been found associated with body weight, stature, and pre-weaning average daily gain in Nellore cattle (Utsu- nomiya et al. 2013; Fink et al. 2017). CHCHD7 is a eukary- otic protein consisting of two pairs of cysteines that form two disulfide bonds stabilizing a coiled coil–helix–coiled coil–helix (CHCH) fold (Cavallaro 2010). It affects carcass weight in Japanese Wagyu cattle (Nishimura et al. 2012), and height in Jersey and Holstein cattle (Utsunomiya et al. 2013; Fink et al. 2017). Searching for non-synonymous mutations, we identified one (14:25056041), and two (14:24880380, 14:24881091) novel missense variants in the CHCHD7 and LYN gene regions, respectively (Additional file 2: Table S4). Domestication and artificial selection for increased meat production have changed the physical and morphological structure of domesticated animals. Moreover, Angus cattle have been intensively artificially selected for higher beef production and quality; this resulted Angus cat- tle being among the larger breeds with higher growth rate (Arthur et al. 2001; Chambaz et al. 2003; McClure et al. 2010; Kemper et al. 2012). Selection for high growth rate in cattle compromise repro- ductive performances and cause dystocia. Genes affecting fetal growth, the size of the pelvis of the dam, mature body weight, and other factors have been reviewed to affect dys- tocia (Zaborski et al. 2016). Therefore, genes controlling these factors cause dystocia. In this study, quite a few genes including ESR1, RPS20, PPP2R1A (A-J), GHRL, PLAG1 (A-H) and zinc finger proteins that directly or indirectly con- tribute to the factors associated with dystocia were identified under selection in Angus cattle (Table 1). ESR1 is a nuclear receptor family of transcription factors that mediate cellular signaling of estrogens which have effects on reproduction at various stages of development (Lazari et al. 2009). ESR1 is among the genes which influence calving difficulty (Zabor- ski et al. 2016). Genes influencing fetal growth including RPS20 (Zaborski et al. 2016), PPP2R1A (Cole et al. 2014), GHRL (Maltecca et al. 2011), and PLAG1 (Utsunomiya et al. 2013; Juma et al. 2016) affect dystocia. PPP2R1A is among the major Ser/Thr phosphatases involved in cell growth and signaling (Cole et al. 2014). GHRL, a ligand for growth hormone secretagogue receptor type 1, induces the release of growth hormone from the pituitary and reported to be linked to fetal growth (Maltecca et al. 2011). PLAG1 is associated with birth weight in Nellore cattle (Utsunomiya et al. 2013) and its association with calving ease has been recently reviewed by Takasuga (2016). Zinc finger proteins regulate bone and skeletal development in mammals (Zabor- ski et al. 2016), and might be associated with birth diffi- culty. Regardless of higher growth performances in Angus cattle, Archer et al. (1998) reported non-significant differ- ences of reproductive performances and incidence of dysto- cia between heifers selected for higher growth rate, control lines, and those selected for low growth rate. The relatively smaller birth weight but the fast growth rate of Angus cattle might explain this ( reproduction/articles/regulating-birth-weight-in-beef-cattle). Genes that are related to fertility (FSHR, CORIN) were also identified under selection. FSHR (A–J), a receptor for the follicle-stimulating hormone, is crucial for follicu- lar development and estradiol production in females, and regulation of sertoli cell function and spermatogenesis in males (Desai et al. 2013). Loss/gain of function mutation in this gene has been reported previously (Desai et al. 2013). CORIN (A–J) plays a role in female pregnancy by pro- moting trophoblast invasion and spiral artery remodeling in the uterus (Cui et al. 2012b; Soares et al. 2014). It is expressed in the pregnant mouse and human uterus to which its impaired expression is associated with preeclampsia, a major risk factor for placental abruption (Cui et al. 2012b; Nagashima et al. 2013). Sperm and spermatogenesis associ- ated genes (SPAG1, SPATA22, and SPATA6) were also iden tified and may affect reproduction. Mammary gland epithelium development was enriched in the ClueGo networks (Fig. 4b). Genes involved in the network (ESR1, PRLR) affect reproductive performances in livestock. PRLR, a receptor for prolactin, have a role in reproduction function through mammary gland develop- ment, lactation, and regulation of maternal behavior. It is associated with embryonic survival rate (Khatib et al. 2009). Prolactin is a pleiotropic hormone that affects many physi- ological functions including mammary gland development, lactogenesis, and fertility (Donato Jr and Frazão 2016).Feeding efficiency in beef cattle production is directly asso- ciated with the profitability of the farm (Do et al. 2014). In relation to this, KEGG pathways of olfactory transduction (A–H), tight junction (A–H), and metabolic pathways (both) were enriched (Fig. 3). Olfactory transduction pathways affect the perception of odor through olfactory receptors and biochemical signaling events which as a result influence food preference and food consumption (Do et al. 2014; Stafuzza et al. 2017). Olfactory transduction pathway has been found associated with residual feed intake (RFI) in pigs (Do et al. 2014) and cattle (Abo-Ismail et al. 2014; Zhao et al. 2015; Stafuzza et al. 2017). In a GWAS analysis of Angus cattle, tight junction and endocytosis pathways were enriched in relation to RFI and average feed intake, respectively (Rolf et al. 2012), that are measures of feeding efficiency. Olfac- tory receptor genes involved in olfactory transduction path- way (OR4F15, OR6C76, OR5D13, OR4N2) were previously identified under selection in African Sanga cattle in rela- tion to feeding intake (Taye et al. 2017). Additionally, genes associated with feeding efficiency including GHRL (Sher- man et al. 2008), PIK3CD, DNAJC28, DNAJC3 (Zhou et al. 2015), and PLAG1 (Fortes et al. 2013) were identified in this study. GHRL is a powerful appetite stimulant and known to play a major role in energy homeostasis. Recently, Sherman et al. (2008) identified a polymorphism in this gene region that is associated with RFI and FCR in beef cattle. Selec- tion efforts towards increased metabolic efficiency of cattle have resulted in decreased feed intake regardless of growth (Stockton 2003; Rolf et al. 2012). The positive selection of the aforementioned genes and pathways might contribute to the feeding efficiency of Angus cattle (Stockton 2003; Rolf et al. 2012). Angus cattle are characterized by their solid black or red coat color ( aberdeen-angus/). Genes associated with coat color (MC1- R, TUBB3—A–J) were identified in this study. These genes encode a protein that controls melanogenesis and are involved in biological functions of pigmentation and inflammation. Polymorphism in MC1-R gene region was previously reported to affect black/red color in Angus cat- tle (Klungland et al. 1995), Holstein-Frisian cattle (Zhao et al. 2015), and sheep (Våge et al. 1999). Searching for non-synonymous genetic variants in this gene region, we identified three (one known—rs109688013, and two novel −18:14758030, 18:14758197) missense variants (Additional file 2: Table S4).Fawn Calf Syndrome (FCS) is a heritable abnormality of skeletal development reported in Angus cattle. It is a non- lethal developmental genetic defect in calves (Whitlock 2010). FCS is related to Marfan Syndrome in humans, a malformation of connective tissue within the skeletal system ( tyly-fawn-calf-syndrome), that is caused by a mutation in fibrillin-1 gene (FBN1) (Jovanović et al. 2007)—not identi- fied here. Instead, an integrin gene (ITGB6—A–H) which acts as a receptor for FBN1 (Jovanović et al. 2007) was found under selection. A mutation in the integrin gene can affect the function of FBN1 and lead to FCS in Angus cattle. We have identified two known (rs110694377, rs136500299) and one new (2:36325642) missense variants in this gene region (Additional file 2: Table S4). In addition, hypertrophic cardiomyopathy (HCM), an inherited cardiac disorder, is significantly (P < 0.05) enriched in the KEGG pathways of A–H gene list (Fig. 3). Although it is not reported in Angus cattle, HCM is a notorious issue in Holstein cattle (Guziewicz et al. 2007). Genes involved in this pathway include ACTB, PRKAB1, ITGB6, and ITGA8. Gene families (ACTC1, ITGAV, and ITGA2) associated with cardiomyopathy were previously reported under selection in Holstein cattle (Lee et al. 2014a).Dwarfism is a genetic disorder characterized by systemic skeletal disorders, including shortness and deformity of limbs, head, and vertebrae, that is recognized in multiple breeds of cattle including Angus cattle (Whitlock et al. 2008). The PLAG1 gene that affects stature has been tested for a knockout experiment where the plag1 K.O. mice showed dwarfism and low fertility (Takasuga 2016). Conclusion In this study, the analysis of positive selection signature identified genes and pathways that contribute for the phe- notypes of Angus cattle including meat quality traits, stat- ure and body size, coat color, and genetic disorders to be under positive selection. Intensive artificial selection might be the probable selective pressure for the positive selection of genes affecting meat quality and body size. Genes that are related to genetic disorders might be because of their association with genes affecting other production traits. The findings in this study, if followed by a validation using addi- tional or independent dataset, can provide useful information for the study of beef cattle in general and Angus cattle in particular. These findings will contribute to the detection of functional candidate genes which have undergone positive selection in future PF-06424439 studies.