Genome-Wide Association Study for Fatty Acid Composition in American Angus Cattle

Simple Summary Almost everybody depends on livestock for various reasons directly or indirectly. Consequently, improving livestock production means improving human life. Meat plays important role in human life, as it is good source of protein and energy. Meat composition depends on breed’s genetics and environmental factors. Fatty acids (FA) play important role in human diet and health. FA add flavor and taste to meat. Fatty acid composition of meat is a complex polygenic trait that is controlled by genetics and environmental factors. Therefore, the objective of the present study was to identify genomic regions associated with FA composition in American Angus. Thirty-six different genomic regions were identified associated with variation in at least one FA. The genomic regions associated with more than one FA and high genetic variance, harbor good candidate genes (e.g., FABP2, FASN, FADS2, FADS3 and SCD). The identified makers could be used to select for altered FA profile and help to increase the understanding of the genetic basis of FA composition. Furthermore, findings from the present study could help to devise effective breeding plans and selection strategies for the improvement of beef FA profile. Abstract Livestock is an important commodity playing a major role in the global economy. Red meat plays an important role in human life, as it is a good source of animal protein and energy. The fatty acid content of beef has been shown to impact the eating experience and nutritional value of beef. Therefore, this study aimed to identify genomic regions which can account for genetic variation in meat fatty acid content. Genotypes imputed to the Illumina BovineHD 770K BeadChip were used in this study. Thirty-six 1-Mb genomic regions with a posterior probability of inclusion (PPI) greater than 0.90 were identified to be associated with variation in the content of at least one fatty acid. The genomic regions (1Mb) which were associated with more than one fatty acid trait with high genetic variance and harbored good candidate genes were on Chromosome (Chr) 6 (fatty acid binding protein 2), Chr 19 (thyroid hormone receptor alpha, fatty acid synthase), Chr 26 (stearoyl-CoA desaturase), and Chr 29 (thyroid hormone responsive, fatty acid desaturase 2, and fatty acid desaturase 3). Further studies are required to identify the causal variants within the identified genomic regions. Findings from the present study will help to increase understanding of the variation in fatty acid content of beef and help to enhance selection for beef with improved fatty acid composition.


Introduction
Beef has a high nutritional value, and it is a rich source of minerals, vitamins, and protein. The consumer is becoming more concerned about their health and more conscious about the quality of the meat that they consume. Consumers have been told that beef consumption is associated with some serious health issues, such as heart diseases and obesity [1]. However, recent findings have shown that the long-standing belief that beef is associated with cardiovascular disease is incorrect [2][3][4][5][6][7]. Furthermore, fat is a very important constituent of the human daily diet; it provides energy and also contains essential fatty acids, and adds flavor to food [8]. The fatty acids present in animal tissues can be separated into phospholipid and triacylglyceride fractions [9]. Fatty acid composition and fat content of the beef are associated with the taste, flavor, and sensory properties of the meat [10]. It has been reported that fatty acid composition varies across different breeds and feeding regimes [10,11].
The mechanism that control fatty acid composition of meat is a complex process that is regulated by genetics and environmental factors. There have been several studies published that evaluated the extent to which genetics controlled variation in fatty acid composition in Santa Gertrudis, Brahman, Hereford, Nellore, and Black Angus cattle breeds [12][13][14]. Identification of genomic markers and regions associated with beef fatty acids could be used to select for an improved fatty acid profile and to alter the saturated to mono-and polyunsaturated fatty acid ratios. The objective of the present study was to identify genomic regions associated with fatty acid composition in American Angus cattle.

Animal Selection
The purebred American Angus cattle used in the current study were reared according to standard animal care procedures, approved by the Iowa State University Animal Care and Use Committee. All the research animals were raised on Iowa State University research demonstration farms.

Sampling and DNA Isolation
A total of 2177 American Black Angus calves sired by 134 sires were used in this study. Blood samples were collected from the jugular vein. DNA samples were collected as previously described by Garmyn et al. [15]. DNA was stored at -20 degrees Celsius until further processing.

Genotype Data
Animals were originally genotyped with either the BovineSNP50 BeadChip (Illumina, San Diego, CA, USA) or the BovineHD BeadChip (Illumina, San Diego, CA, USA) by Neogen GeneSeek Operations (Lincoln, NE, USA). Animals genotyped with the BovineSNP50 BeadChip were imputed to the BovineHD BeadChip SNP density using FImpute [16] and SNPipeline package (Hailin Su, https://github.com/cbkmephisto/SNPipeline (accessed on 27 February 2017)) by using 820 Angus individuals originally genotyped on the BovineHD BeadChip. These 820 individuals included animals from the ISU herd and external animals. A filter of 0.05 minor allele frequency was applied and all markers with missing information were excluded. After filtering, a total of 199,431 markers were excluded from analysis, leaving a total of 574,662 markers for data analyses. Genome coordinates are relative to the Bovine UMD 3.1 genome assembly.

Fatty Acid Profile
For fatty acid profile analysis, animals were slaughtered at commercial slaughtering facilities. All the slaughtering procedures were carried out by trained personnel. Carcass data collection, tissue sampling, and fatty acid profile analysis were carried out. Fatty acid composition was analyzed as previously reported [17].
For each fatty acid, phenotypic observations were recorded on a fat percentage basis to estimate the marker effect. In this study, 56 different fatty acid traits were included.

Statistical Analysis
Imputed genotype data were utilized to estimate the SNP effect associated with fatty acid composition. Statistical analysis was performed using the BayesB method for genomic prediction [18]. Data were analyzed via the following model: where y is the observable value for fatty acid, and X and Z are fixed and random effects, respectively. In this model, b is the fixed effect (age, sex, and population mean), u is the random effect marker, and e is the residual effect [18,19]. Fixed effect and covariates included: contemporary group, sex and hot carcass weight, longissimus muscle area at 12th rib, subcutaneous fat thickness at 12th rib and chemically extracted fat. All the analyses were performed using GenSel software [20].
A chain of 50,000 iterations with the first 5000 as burn-in was used, and the parameter pi (π) was set at 0.99906 (99.9%; approximately 540 SNP markers with a non-zero effect), while genetic and residual variances for each trait were estimated using BayesC (initial variances set as half the total phenotypic variance) before being used in BayesB [21,22]. The posterior probability of inclusion and correlation between QTL and trait were calculated, as described by [20].

Fatty Acid Data Statistics
Summary statistics for the studied fatty acids traits are given in Table 1.

Trait
σ 2 e , g × 10 −10 σ 2 g , g × 10 −10 h 2 All other traits (excluding those described above) appeared to have low heritability values. For these traits, the amount of phenotypic variance explained by the markers was low. The lowest heritability estimated was on fatty acid ratio C18:0/C16:0 (0.005). Previously reported heritability values for fatty acid traits were lower than in the present study. In American Angus, using 50k SNP chip data, the highest heritability value (0.57) was reported for saturated fatty acid C14:0 [18], while in Nellore cattle, the highest reported value was 0.24 for C17:0 and C18:3-n6 [12]. Another study in Canada on beef cattle showed higher h 2 values of 0.57 and 0.59 for a saturated fatty acid (C17:0) and monounsaturated fatty acids (C14:1 and C18:1) [23]. A previous study showed that analysis of fatty acid content on a fat percentage basis was able to explain a greater proportion of phenotypic variance by SNP markers, as compared to using fatty acid content on a beef basis [18].

Genome-Wide Association Study
A total of 56 fatty acid traits including saturated, monounsaturated, polyunsaturated, and fatty acid groups were used for genome-wide association studies. The identified genomic regions (1-Mb windows) that showed high genetic variance and posterior probability of inclusion (PPI) greater than 90% for having non-zero genetic variance or above are presented in Table 3.    The highest estimated genetic variance explained by a single SNP window was 88.09% for the fatty acid C22:1, while the window with the lowest estimated genetic variance, 1.72%, was for fatty acid C18:1t12. Regarding the SNP window which explained the highest level of genetic variance (30_72), there was a potential candidate gene, phosphatidylinositol specific phospholipase C X domain containing 1 (PLCXD1), on the pseudo-autosomal region (PAR). This gene is X-linked in ruminants. PLCXD gene products are phosphodiesterases involved in the regulation of cytosolic calcium and have protein kinase activity [24,25].
This study identified a total of 36 different 1-Mb SNP windows that were associated with fatty acid content of skeletal muscle. Three windows (19_51, 26_21, and 29_18) appeared to be associated with most beef fatty acid traits. These three windows were previously reported in the same Black Angus population using the Bovine SNP50 BeadChip [18]. Many of the 1-Mb SNP windows were associated with more than one fatty acid trait. . This region was also previously reported to be associated with important fatty acid traits [18,26]. This SNP window contains a good candidate gene-fatty acid synthase (FASN). This gene has been reported to be involved in beef fatty acid composition [27]. It has also been reported to be associated with adipose composition, milk fatty acid composition, and milk fat content in many different breeds of cattle. These reports indicated that this gene has a pivotal role and is an important candidate gene for fatty acid composition [26,[28][29][30][31].
Similar to the window at 51 Mb on chromosome 19, there were additional SNP windows that were associated with more than ten fatty acid traits, including a window at 21 Mb on Chromosome 26, which was associated with 13 FA traits, and a window at 18 Mb on Chromosome 29, which was associated with 11 FA traits. These two SNP windows were also previously reported to be associated with various FA traits [18,26]. These two regions harbor good candidate genes for fatty acid composition, including stearoyl-CoA desaturase (SCD) and thyroid hormone responsive (THRSP). Previous studies have reported that SCD is associated with meat fat composition and milk fat composition [28][29][30][31]. Thyroid hormone responsive and stearoyl-CoA desaturase is involved in fatty acid synthesis [32]. Variants in the THRSP gene have been shown to be associated with the synthesis of beef fatty acids, which are expected to have a direct impact on beef quality [18,33]. It has also been reported that both SCD and THRSP genes are involved in lipid metabolism in cattle [34]. The 1-Mb SNP windows that harbor good candidate genes for fatty acid synthesis and fat regulation are shown in Table 4.
In the present GWAS, some new genomic windows were identified which were not previously reported by any study. These new SNP windows are associated with different fatty acid traits and harbor good candidate genes for fatty acid composition. On chromosome 16, a 1-Mb SNP window at 4 Mb was associated with SFA, MUFA, MUFA/SFA, UFA/SFA, and C16:1, C18:1/C16:0, C18:0. This window contains good candidate genes-6-phosphofructo-2-kinase/fructose-2,6-biphosphatase 2 (PFKFB2) and peptidase M20 domain-containing 1 (PM20D1). The PFKFB2 gene has a role in degradation and synthesis of fructose-2,6-biphosphate [26]. It has been previously reported that a QTL spanning this region is related to fat thickness at the 12th rib in American Angus [35]. Peptidase M20 domain-containing 1 (PM20D1) is an enzyme that synthesizes N-acyl amino acids (NAAs). NAAs are bioactive lipids composed of fatty acyl chains. PM20D1 regulates the condensation and hydrolysis of N-acyl amino acids from free amino acids and fatty acids [36][37][38]. Fatty acid binding protein 2 (FABP2) may have a role in lipogenesis and adipose tissue weight variability [39]. This gene has not been identified in previous GWAS as having a significant association with fatty acid composition [40]. Bardet-Biedl syndrome 4 (BBS4) is involved in the secretion and expression of Follistatin-like 1 (FSTL1), which is associated with adipogenesis. BBS4 also plays a role in fatty acid profile, lipolysis, and fat accumulation [41,42]. Acetyl-CoA acyltransferase 2 (ACAA2) codes for an enzyme from the thiolase family. This enzyme is involved in elongation and degradation of fatty acids. It has been associated with milk yield and fat yield in dairy sheep [43]. Fatty acid desaturase 2 (FADS2) and fatty acid desaturase 3 (FADS3) belong to the fatty acid desaturase family. This family of genes creates a cis double bond in FA chains at specific sites and is associated with desaturation of fatty acids and blood phospholipids [44,45]. Oxysterol binding protein like 5 (OSBPL5) is a lipid transporter, chiefly linked with the exchange of phosphatidylserine with phosphatidylinositol 4-phosphate. It has a role in maintaining cholesterol balance [46,47].
These genes have not been previously identified by any GWAS as being related to fatty acid content and fat regulation. In the present GWAS, we did not identify any SNP windows which contain some genes (LXR, LXRA, SREBP1, PPARG, ACSL1, LEP, ACACA, FABP4, and SL1TRK6) previously shown to be associated with FA composition and variation in beef cattle [27,[48][49][50][51][52][53]. This may indicate that genetic control of fatty acid content varies greatly from breed to breed.

Conclusions
Genome-wide association studies can provide insight into understanding the mechanisms underlying fatty acid composition. Furthermore, genomic selection methodology can be used to select for, and to alter, fatty acid content. This study utilized imputed BovineHD BeadChip (770k) genotypes along with skeletal muscle fatty acid content phenotypic data to identify 36 1-Mb SNP windows that had a PPI > 0.90. Some of these SNP windows have been previously reported, including 19_51, 26_21, and 29_18. In addition, some new genomic regions that had not been previously reported to be associated with fatty acid content were identified: 6_7, 19_42, and 29_42. Fatty acid composition and deposition are complex polygenic traits having low to moderate heritability. The genomic regions identified in the present study and associated potential candidate genes for FA composition could help increase understanding of the genetic basis of FA composition in beef cattle (Black Angus). This study could also help to devise sensible breeding plans and selection strategies based on identified genomic regions for the improvement of beef fatty acid profile.  Data Availability Statement: Data will be available upon request.

Conflicts of Interest:
The authors declare no conflict of interest.