1st Page

Journal of Animal Science - Animal Genetics

Genomewide association analysis of growth traits in Charolais beef cattle1


This article in JAS

  1. Vol. 94 No. 11, p. 4570-4582
    Received: Feb 02, 2016
    Accepted: May 18, 2016
    Published: October 27, 2016

    2 Corresponding author(s):

  1. F. J. Jahuey-Martínez*,
  2. G. M. Parra-Bracamonte 2*,
  3. A. M. Sifuentes-Rincón*,
  4. J. C. Martínez-González,
  5. C. Gondro,
  6. C. A. García-Pérez* and
  7. L. A. López-Bustamante§
  1. * Centro de Biotecnología Genómica-Instituto Politécnico Nacional, Reynosa, Tamaulipas, México, 88710
     Universidad Autónoma de Tamaulipas-Facultad de Ingeniería y Ciencias, Victoria, Tamaulipas, México, 87749
     The Centre for Genetic Analyses and Applications, University of New England, Armidale, NSW, Australia, 2351
    § Charolais Herd-Book of Mexico-RON B Charolais Ranch, Hermosillo, Sonora, México


The objective of this study was to perform a genomewide association study (GWAS) for growth traits in Charolais beef cattle and to identify SNP markers and genes associated with these traits. Our study included 855 animals genotyped using 76,883 SNP from the GeneSeek Genomic Profiler Bovine HD panel. The examined phenotypic data included birth, weaning, and yearling weights as well as pre- and postweaning ADG. After quality control, 68,337 SNP and 823 animals were retained in the analysis. The association analysis was performed using the principal components method via the egscore function of the GenABEL version 1.8-0 package in the R environment. Eighteen SNP located in 13 BTA were associated with growth traits (P < 5 × 10−5). The most important genes in these genomic regions were TRAF6 (tumor necrosis factor receptor-associated factor 6), CDH11 (cadherin 11, type 2, OB-cadherin), KLF7 (Kruppel-like factor 7), MIR181A-1 (microRNA 181A-1), and PRCP (prolylcarboxypeptidase [angiotensinase C]), due to their relationships with perinatal and postnatal survival, bone growth, cell adhesion, regulation of adipogenesis, and appetite. In conclusion, this study is the first to describe a GWAS conducted in beef cattle in Mexico and represents a basis for further and future research. This study detected new QTL associated with growth traits and identified 5 positional and functional candidate genes that are potentially involved in variations of the analyzed traits. Future analyses of these regions could help to identify useful markers for marker-assisted selection and will contribute to the knowledge of the genetic basis of growth in cattle and be a foundation for genomic predictions in Mexican Charolais cattle.


Animal growth is an important trait in beef cattle production systems and is the main objective of the genetic improvement of beef cattle in Mexico. Growth characteristics such as animal BW from birth to adulthood are excellent indicators for assessing the efficiency of farms engaged in meat production. For example, birth weight is an important indicator of animal viability and is also used as a selection criterion to improve calving ease. Weaning weight is a good indicator of production in cow–calf systems, and yearling weight is a trait that is typically correlated with final and carcass weights.

The use of thousands of SNP markers in genomewide association studies (GWAS) has allowed the identification of genomic regions that influence traits of productive interest in livestock species (Zhang et al., 2012). Several studies have identified and confirmed many QTL for growth traits in beef cattle (Snelling et al., 2010; Lu et al., 2013; Buzanskas et al., 2014), which, in turn, have served as the basis for the search for the nucleotides responsible for the phenotypic variation (Takasuga, 2016). Genomewide association studies have also indicated that growth traits are mainly controlled by multiple QTL with very small effects. Therefore, it is important to continue the investigation of the genetic architecture of growth traits to achieve the discovery of new QTL that may help to explain the variation in these traits.

Charolais cattle are one of the most important breeds for meat production in Mexico due to their growth, muscularity, production efficiency, and maternal ability, among other traits, and their selection, management, and genetic improvement could potentially be improved by the use of high-throughput genotyping technologies. Therefore, the objective of this study was to perform GWAS using genomewide SNP markers in registered Charolais beef cattle to identify QTL for growth traits and define genes as potential candidates for further studies.


Approval from the ethical committee for animal care and use was not necessary because the samples used in this study consisted of hair follicles and commercialized semen straws.

Population and Phenotypic Data

The Charolais cattle population used in this study included 712 females and 143 males registered in the database of the Charolais-Charbray Herdbook de México A.C. The cattle were born between 1999 and 2013 and represented the progeny of 142 sires and 593 cows. This population came from herds located in the northeast and northwest of Mexico and was raised under extensive production systems. Phenotypic data were provided by the breeding association and included records of BW at birth (BWT; kg), 205-d adjusted weaning weight (BW at weaning [WWT]; kg), 365-d adjusted yearling weight (yearling BW [YWT]; kg), preweaning ADG (AGW; kg), and postweaning ADG (PWG; kg). Table 1 shows the descriptive statistics for each trait.

View Full Table | Close Full ViewTable 1.

Descriptive statistics for growth traits1 of Charolais beef cattle

Trait n n(QC)2 Mean SD Minimum Maximum
BWT, kg 855 823 40.488 6.712 20 62
WWT, kg 840 809 227.554 39.571 103.7 377.8
AGW, kg 840 809 0.912 0.186 0.291 1.603
YWT, kg 418 409 356.693 58.137 211.2 582.2
PWG, kg 418 409 0.807 0.284 0.499 1.469
1AGW = preweaning ADG; BWT = BW at birth; PWG = postweaning ADG; WWT = BW at weaning; YWT = yearling BW.
2n(QC) = n after quality control.

Genotyping and Quality Control

The animals were genotyped using 76,883 SNP markers from the GeneSeek Genomic Profiler Bovine HD panel (GGPHD, Neogen Corp., Lincoln, NE). Before association analysis, the quality of the genotypic data was verified using the SNPQC program (Gondro et al., 2014). The genotypes were considered successful if they presented a GenCall value greater than 0.50, and all SNP with lower values were discarded (n = 2,629). Those SNP that were monomorphic (n = 363), presented call rates of less than 90% (n = 5,368) or minor allele frequencies <0.01 (n = 1,162), or deviated from Hardy–Weinberg equilibrium according to Fisher’s exact test and exhibited P-values > 1 × 10−15 (n = 723) were also eliminated. In addition, SNP with unknown coordinates in the assembly of the bovine genome UMD version 3.1 (Zimin et al., 2009; n = 99) and SNP that were not located on autosomal chromosomes (n = 2,112) were discarded.

Samples were also eliminated if they exhibited call rates of less than 80% (n = 14) or levels of heterozygosity above 3 SD (n = 26), taking into account that the mean and SD of the observed heterozygosity were 0.385 and 0.018, respectively. Finally, a Pearson correlation was computed between each pair of samples according to their genotype information, obtaining an average of r = 0.57 and minimum and maximum values of 0.037 and 0.86, respectively. A maximum value of r = 0.98 for detecting potentially duplicate samples (n = 0) was also considered. A total of 68,337 SNP and 823 samples passed the quality control procedure and were retained for further analysis. Quality control and subsequent analyses were performed in the R environment (R Development Core Team, 2008).

Population Structure and Association Analysis

Population structure was expected in the sample because the cattle were from 2 different regions in northern Mexico. This expectation was supported by the fact that the tested herds presented different selection objectives and origins of the imported genetic material (i.e., semen and sires). To investigate the structure of the population, a genomic relationships matrix was first calculated using information on genotypes according to VanRaden (2008), and a singular value decomposition and principal components (PC) analysis was then performed. The preliminary analysis indicated that 28.6% of the variance in the data was explained by the first 2 PC. Therefore, we decided to perform a genomewide association analysis using the PC method proposed by Price et al. (2006). For this analysis, the egscore function from the GenABEL package (Aulchenko et al., 2007) was used. To account for population stratification, this function uses the genomic kinship matrix to derive axes of genetic variation (PC) and then both the phenotypes and genotypes are adjusted onto these axes. For each trait, a linear model was fitted including the first 2 PC as covariates. For the analysis of BWT, the model also included the contemporary group (CG) and the linear and quadratic effects of the age of the dam. The CG included herd, sex, year, and season of birth. The statistical model used to adjust the other traits included only the contemporary group as well as the PC as covariates; age of dam was excluded here because it was not a significant factor in previous analysis. Finally, the association between corrected genotypes and phenotypes was assessed via correlation, and P-values were obtained by calculating the square of the correlation multiplied by N − K − 1, in which N was the number of genotyped individuals and K was the number of PC. Minimum allele frequencies, allele substitution effect, and percentage of phenotypic variance explained by the SNP were estimated. Single nucleotide polymorphisms with P-values < 5 × 10−5 were considered significant and associated with growth traits. The proportion of phenotypic variance explained by the SNP was estimated by dividing the χ2 value for a degree of freedom by the number of individuals used for the analysis of each SNP marker followed by multiplication by 100. All described analyses and estimations were performed using GenABEL package (Aulchenko et al., 2007).

Analysis of Genomic Regions with Significant SNP

The closest genes to significant markers and those located within a 250-kb window on both sides of the SNP location were identified. The list of genes was obtained using the snp2gene.LD function from the Postgwas package (Hiersche et al., 2014). The distance between SNP and genes was calculated as the difference between the marker position and the beginning or end of the gene according to the coordinates from the assembly of the bovine genome UMD version 3.1. Gene functions were investigated in the UniProt database (UniProt Consortium, 2014). Annotations from humans or mice were used when there was no information on the particular genes in cattle. Genes were considered functional and positional candidates if they were biologically related to the trait under study and supported by experimental evidence in the literature. Finally, we determined whether significant SNP mapped against QTL previously associated with growth-related traits such as BW, carcass, and reproduction traits were deposited in the cattle AnimalQTLdb (Hu et al., 2013). For this purpose, we used SNP positions according to the Btau4.6 (Elsik et al., 2015) genome sequence because many of the previously reported QTL had no well-defined positions in the assembly of the bovine genome UMD version 3.1.


A total of 68,337 SNP from the GGPHD panel were evaluated for associations with growth traits of Charolais cattle. On average, 2,356 SNP markers were evaluated in each BTA. Bos taurus chromosomes 1 and 25 exhibited the highest and lowest numbers of SNP, respectively. The average distance between adjacent SNP was 36,298 bp, and the minimum (75 bp) and maximum distances (3,085,904 bp) between adjacent SNP were found on BTA 2 and 6, respectively.

The effect of population structure was removed from the analysis in this study, which can be observed in the quantile–quantile plot for each GWAS (Fig. 1). According to the significance threshold considered (P < 5 × 10−5), 18 SNP were associated with growth traits (Table 2). These markers were distributed on BTA 1, 2, 4, 6, 8, 9, 11, 15, 16, 18, 21, 24, and 29. Figure 2 shows the Manhattan plots in which the −log10 transformations of the P-values are plotted for each of the GWAS. Genes and QTL previously associated with growth-related traits are shown in Table 3. Tables 4 and 5 show complete descriptions including the identifier number and exact location of each gene as well as any previously reported QTL located in the genomic regions identified in this study.

Figure 1.
Figure 1.

Quantile–quantile (QQ) plots for the genomewide association study of growth traits in Charolais beef cattle. The straight line in the QQ plots indicates the distribution of SNP markers under the null hypothesis, and the skew at the edge indicates that these markers are more strongly associated with the traits than would be expected by chance. BWT = BW at birth; WWT = BW at weaning; AGW = preweaning ADG; YWT = yearling BW; PWG = postweaning ADG.


View Full Table | Close Full ViewTable 2.

Parameters and statistics of SNP associated with growth traits of Charolais beef cattle1

Trait SNP ID2 BTA UMD3.1,3 bp Btau4.6,4 bp Allele MAF5 β,5 kg SE Percentage Var5 P-value
BWT rs42512403 15 68,021,790 66,969,167 G/A 0.44 −0.160 0.034 2.6 0.0000359
WWT rs43378829 4 30,834,682 31,656,994 C/A 0.46 −0.753 0.156 2.8 0.0000064
rs109462000 11 72,437,674 74,486,805 A/G 0.26 1.293 0.280 2.6 0.0000159
rs109028958 18 31,925,240 30,754,012 G/A 0.36 0.915 0.205 2.4 0.0000305
rs42339359 29 12,381,867 12,854,197 A/G 0.16 −1.695 0.389 2.3 0.0000442
YWT rs133575890 6 68,061,364 68,907,863 G/A 0.40 1.643 0.339 5.7 0.0000014
rs110545556 2 95,861,435 99,648,226 A/G 0.29 2.131 0.469 5.0 0.0000062
rs110875592 6 68,059,441 68,909,786 A/C 0.43 1.445 0.319 5.0 0.0000065
rs134262111 1 3,515,994 3,339,804 A/G 0.06 5.767 1.356 4.4 0.0000233
rs109532225 15 42,922,452 41,199,707 G/A 0.22 2.536 0.605 4.4 0.0000308
rs109419103 21 56,601,433 56,469,386 G/A 0.21 2.600 0.624 4.2 0.0000340
rs43538101 8 24,510,080 26,019,429 C/A 0.37 1.608 0.386 4.2 0.0000348
rs109300035 6 68,410,488 69,261,925 G/A 0.42 −1.411 0.341 4.1 0.0000380
rs135029160 6 68,027,231 68,871,374 A/G 0.45 1.317 0.319 4.1 0.0000414
AGW rs43378829 4 30,834,682 31,656,994 C/A 0.46 −0.003 0.000 2.6 0.0000176
rs109462000 11 72,437,674 74,486,805 A/G 0.26 0.005 0.001 2.5 0.0000261
PWG rs41566285 24 34,013,385 34,397,620 A/G 0.34 −0.008 0.001 5.2 0.0000058
rs41569509 9 9,372,653 8,968,100 A/C 0.41 −0.007 0.001 5.4 0.0000066
rs42219323 16 80,103,005 76,347,639 C/A 0.12 0.016 0.001 4.5 0.0000280
rs41585202 24 47,181,172 48,010,063 A/G 0.40 0.006 0.001 4.3 0.0000355
1AGW = preweaning ADG; BWT = BW at birth; PWG = postweaning ADG; WWT = BW at weaning; YWT = yearling BW.
2ID = identification.
3UMD version 3.1 (Zimin et al., 2009).
5MAF = minimum allele frequency; β = allele substitution effect; Var = phenotypic variance explained by the SNP.
Figure 2.
Figure 2.

Manhattan plots of the P-values for the genomewide association study of growth traits in Charolais beef cattle. The horizontal line indicates the significance threshold for significant associations (P < 5 × 10−5). BWT = BW at birth; WWT = BW at weaning; YWT = yearling BW; AGW = preweaning ADG; PWG = postweaning ADG.


View Full Table | Close Full ViewTable 3.

Genes and previously reported QTL1 located near SNP associated with growth traits of Charolais beef cattle

Trait_SNP ID2_BTA_Mb Genes in ±250 kb QTL Reference
BWT_rs42512403_15_68.0 TRAF6, RAG1, RAG2, C15H11orf74- SNP CALEASE Sahana et al. (2011)
ADG Peters et al. (2012)
WWT, AGW_rs43378829_4_30.8 DNAH11- SNP CDCA7L- RAPGEF5 LMA and MARBL Mizoshita et al. (2004)
CALEASE Yokouchi et al. (2009)
YRLHT, MHT, MARBL, WWT, WWTMM, and YWT Thomasen et al. (2008) and McClure et al. (2010)
WWT, AGW_rs109462000_11_72.4 IFT172, KRTCAP3, NRBP1, PPM1G, ZNF513, SNX17, EIF2B4, GTF3C2, MPV17, UCN, TRIM54, DNAJC5G, SLC30A3, CAD, SLC5A6- SNP- TCF23, PRR30, PREB, ABHD1, CGREF1, KHK, EMILIN1, AGBL5, TMEM214, MAPRE3, DPYSL5 MARBL McClure et al. (2010)
WWT_rs109028958_18_31.9 CDH11- SNP SB Seidenspinner et al. (2009)
RFI Sherman et al. (2009)
WWT McClure et al. (2010)
WWT_rs42339359_29_12.3 DLG2- SNP- CCDC90B, ANKRD42, PCF11 CWT Kim et al. (2003)
CWT and TEND Casas et al. (2000, 2003)
RFI Nkrumah et al. (2007)
MHT, LMA, FATTH, CWT, and WWTMM McClure et al. (2010)
YWT_ rs109300035_6_68.4 CORIN, NFXL1, CNGA1, NIPAL1- SNP TXK- TEC
YWT_rs110545556_2_95.8 CPO- SNP KLF7- MIR2355
YWT_rs134262111_1_3.51 SNP- MIR2284I, MIR2284X
YWT_rs109532225_15_42.9 MRVI1, LYVE1, RNF141, ADM- SNP- SBF2 MHT, CALEASE, WWT, and WWTMM McClure et al. (2010)
YWT_rs109419103_21_56.6 GPR68- SNP- CCDC88C, SMEK1
YWT_rs43538101_8_24.5 - SNP SLC24A2- SB Kühn et al. (2003)
YRLHT, MHT, BWT, and CWT McClure et al. (2010)
PWG_rs41566285_24_34.0 RBBP8- SNP CALEASE, WWT, and YWT McClure et al. (2010)
PWG_rs41569509_9_9.3 SNP- COL9A1 MARBL Imai et al. (2007)
RFI Sherman et al. (2009)
CALEASE, MARBL, YWT, and WWT McClure et al. (2010)
PWG_rs42219323_16_80.1 MIR181A-1- SNP
PWG_rs41585202_24_47.1 PIAS2- SNP- KATNAL2, HDHD2, IER3IP1 FCR Sherman et al. (2009)
MARBL McClure et al. (2010)
1AGW = preweaning ADG; BWT = BW at birth; CALEASE = calving ease; CWT = carcass weight; FATTH = fat thickness at the 12th rib; FCR = feed conversion ratio; LMA = LM area; MARBL = marbling score; MHT = height at maturity; MWT = BW at maturity; PWG = postweaning ADG; RFI = residual feed intake; SB = stillbirth; TEND = tenderness score; WWT = BW at weaning; WWTMM = weaning weight-maternal milk; YWT = yearling BW; YRLHT = yearling height.
2ID = identification.

View Full Table | Close Full ViewTable 4.

Genes close to the SNP associated with growth traits of Charolais beef cattle

SNP_BTA Gene in ±250 kb Gene ID1 Distance,2 kb Description
rs42512403_15 TRAF6 539124 U 232.3 TNF (tumor necrosis factor) receptor-associated factor 6
RAG1 506302 U 191.4 Recombination activating gene 1
RAG2 782387 U 172.2 Recombination activating gene 2
C15H11orf74 614192 U 110.6 Chromosome 15 open reading frame, human C11orf74
rs43378829_4 DNAH11 497208 U 35.0 Dynein, axonemal, heavy chain 11
CDCA7L 514631 Cover Cell division cycle-associated 7-like
RAPGEF5 519566 D 182.3 Rap guanine nucleotide exchange factor (GEF) 5
rs109462000_11 IFT172 100848219 U 214.7 Intraflagellar transport 172
KRTCAP3 508550 U 212.9 Keratinocyte-associated protein 3
NRBP1 532919 U 200.3 Nuclear receptor-binding protein 1
PPM1G 286880 U 164.5 Protein phosphatase, Mg2+/Mn2+ dependent, 1G
ZNF513 100138621 U 160.5 Zinc finger protein 513
SNX17 529972 U 154.7 Sorting nexin 17
EIF2B4 521926 U 149.9 Eukaryotic translation initiation factor 2B, subunit 4 delta, 67 kDa
GTF3C2 782752 U 125.6 General transcription factor IIIC, polypeptide 2, beta 110 kDa
MPV17 505763 U 113.1 MpV17 mitochondrial inner membrane protein
UCN 518336 U 110.7 Urocortin
TRIM54 535320 U 89.1 Tripartite motif containing 54
DNAJC5G 616608 U 84.1 DnaJ (Hsp40) homolog, subfamily C, member 5 gamma
SLC30A3 512803 U 64.4 Solute carrier family 30 (Zinc transporter), member 3
CAD 504261 U 32.9 Carbamoyl-phosphate synthetase 2, aspartate transcarbamylase, and dihydroorotase
SLC5A6 516021 U 12.2 Solute carrier family 5 (Sodium/multivitamin and iodide cotransporter),
TCF23 616841 D 32.6 Transcription factor 23
PRR30 782932 D 53.1 Proline-rich 30
PREB 525256 D 55.5 Prolactin regulatory element-binding
ABHD1 510774 D 59.6 Abhydrolase domain containing 1
CGREF1 507586 D 83.8 Cell growth regulator with EF-hand domain 1
KHK 614868 D 88.0 Ketohexokinase (fructokinase)
EMILIN1 540451 D 101.6 Elastin microfibril interfacer 1
AGBL5 538585 D 118.1 ATP/GTP-binding protein-like 5
TMEM214 514683 D 144.6 Transmembrane protein 214
MAPRE3 528839 D 160.4 Microtubule-associated protein, RP/EB family, member 3
DPYSL5 100126171 D 233.7 Dihydropyrimidinase-like 5
rs109028958_18 CDH11 785475 U 873.5 Cadherin 11, type 2, OB-cadherin (osteoblast)
rs42339359_29 DLG2 505383 U 57.7 Disks large homolog 2
CCDC90B 506046 D 78.3 Coiled-coil domain containing 90B
ANKRD42 522004 D 110.9 Ankyrin repeat domain 42
PCF11 510604 D 173.9 PCF11 cleavage and polyadenylation factor subunit
ATP10D 536495 U 138.1 ATPase, class V, type 10D
CORIN 527107 Cover Corin serine peptidase
NFXL1 784953 D 179.9 Nuclear transcription factor, X-box-binding-like 1
rs109300035_6 CORIN 527107 U 233.8 Corin serine peptidase
NFXL1 784953 U 110.8 Nuclear transcription factor, X-box-binding-like 1
CNGA1 281700 U 56.2 Cyclic nucleotide gated channel alpha 1
NIPAL1 539757 U 18.9 NIPA-like domain containing 1
TXK 504782 Cover TXK tyrosine kinase
TEC 504733 D 58.4 Tec protein tyrosine kinase
rs110545556_2 CPO 508767 U 160.5 Carboxypeptidase O
KLF7 537747 Cover Kruppel-like factor 7 (ubiquitous)
MIR2355 100313333 D 26.7 MicroRNA mir-2355
rs134262111_1 MIR2284I 100313114 D 64.2 MicroRNA mir-2284i
MIR2284X 100526413 D 91.7 MicroRNA mir-2284x
rs109532225_15 MRVI1 281918 U 247.8 Murine retrovirus integration site 1 homolog
LYVE1 404179 U 229.9 Lymphatic vessel endothelial hyaluronan receptor 1
RNF141 539455 U 168.8 Ring finger protein 141
ADM 280713 U 9.1 Adrenomedullin
SBF2 510498 D 48.5 SET-binding factor 2
rs109419103_21 GPR68 281799 U 10.1 G protein-coupled receptor 68
CCDC88C 515039 D 28.3 Coiled-coil domain containing 88C
SMEK1 407220 D 199.4 SMEK homolog 1, suppressor of mek1 (dictyostelium)
rs43538101_8 SLC24A2 525618 Cover Solute carrier family 24 (sodium/potassium/calcium exchanger), member 2
rs41566285_24 RBBP8 512977 U 65.1 Retinoblastoma-binding protein 8
rs41569509_9 COL9A1 282195 D 144.7 Collagen, type IX, alpha 1
rs42219323_16 MIR181A-1 100313401 U 417.0 MicroRNA mir-181a-1
rs41585202_24 PIAS2 533403 U 183.6 Protein inhibitor of activated STAT, 2
KATNAL2 514354 D 4.2 Katanin p60 subunit A-like 2
HDHD2 505403 D 64.9 Haloacid dehalogenase-like hydrolase domain containing 2
IER3IP1 768079 D 123.7 Immediate early response 3 interacting protein 1
1ID = identification.
2D = downstream; U = upstream.

View Full Table | Close Full ViewTable 5.

Previously reported QTL1 found near the SNP associated with growth traits of Charolais beef cattle

Trait_SNP ID2_BTA_Mb QTL QTL ID QTL in Btau4.6,3 bp QTL reference
BWT_rs42512403_15_68.0 ADG 22798 66,162,109–75,437,676 Peters et al. (2012)
CALEASE 15192-4 66,823,500–67,865,000 Sahana et al. (2011)
WWT, AGW_rs43378829_4_30.8 MARBL 10013 8,553,221–50,937,899 Yokouchi et al. (2009)
YRLHT 10709 10,994,551–32,226,979 McClure et al. (2010)
MARBL 10707 10,994,551–32,226,979 McClure et al. (2010)
MHT 10712 10,994,551–32,226,979 McClure et al. (2010)
WWT 10708 10,994,551–35,086,352 McClure et al. (2010)
WWTMM 10711 19,135,469–32,226,979 McClure et al. (2010)
YWT 10710 19,135,469–32,226,979 McClure et al. (2010)
CALEASE 4655 28,268,762–35,086,352 Thomasen et al. (2008)
MARBL 1729 28,268,762–38,039,020 Mizoshita et al. (2004)
LMA 1728 28,268,762–38,039,020 Mizoshita et al. (2004)
WWT, AGW_rs109462000_11_72.4 MARBL 10909 62,118,964–82,382,031 McClure et al. (2010)
WWT_rs109028958_18_31.9 SB 11362 2,437,515–33,011,652 Seidenspinner et al. (2009)
RFI 5294 17,918,086–34,957,653 Sherman et al. (2009)
RFI 5293 17,918,086–35,720,845 Sherman et al. (2009)
WWT 11063 22,926,481–36,500,992 McClure et al. (2010)
WWT_rs42339359_29_12.3 CWT 1316 2,819,208–27,838,509 Kim et al. (2003)
CWT 1344 5,310,955–36,108,874 Casas et al. (2003)
MWT 11296 6,513,888–26,822,859 McClure et al. (2010)
TEND 1373 7,471,912–37,390,549 Casas et al. (2000)
RFI 4403-4 7,760,040–28,758,488 Nkrumah et al. (2007)
FATTH 11291 11,626,483–17,915,367 McClure et al. (2010)
MHT 11290 11,626,483–17,915,367 McClure et al. (2010)
LMA 11289 11,626,483–17,915,367 McClure et al. (2010)
MWT 11293 11,626,483–26,822,859 McClure et al. (2010)
CWT 11292 11,626,483–26,822,859 McClure et al. (2010)
WWTMM 11294 11,626,483–26,822,859 McClure et al. (2010)
YWT_ rs109300035_6_68.4
YWT_rs109532225_15_42.9 MHT 11001 29,452,266–50,231,208 McClure et al. (2010)
CALEASE 11000 36,441,038–51,210,066 McClure et al. (2010)
WWT 11002 40,819,540–51,210,066 McClure et al. (2010)
WWTMM 11003 40,819,540–51,210,066 McClure et al. (2010)
YWT_rs43538101_8_24.5 BWT 10825 15,196,699–29,543,811 McClure et al. (2010)
CWT 10826 15,196,699–29,543,811 McClure et al. (2010)
MHT 10828 24,157,132–29,543,811 McClure et al. (2010)
YRLHT 10830 24,157,132–29,543,811 McClure et al. (2010)
SB 11443 24,157,132–48,094,361 Kühn et al. (2003)
PWG_rs41566285_24_34.0 YWT 11197 30,004,563–41,590,595 McClure et al. (2010)
CALEASE 11196 30,004,563–41,590,595 McClure et al. (2010)
WWT 11198 30,004,563–41,590,595 McClure et al. (2010)
PWG_rs41569509_9_9.3 RFI 5279 5,803,533–12,440,486 Sherman et al. (2009)
WWT 10852 8,888,721–18,194,890 McClure et al. (2010)
YWT 10851 8,888,721–18,194,890 McClure et al. (2010)
MARBL 10850 8,888,721–18,194,890 McClure et al. (2010)
CALEASE 10853 8,888,721–18,194,890 McClure et al. (2010)
MARBL 4507 8,888,721–25,471,420 Imai et al. (2007)
PWG_rs41585202_24_47.1 FCR 5308 25,435,547–51,266,517 Sherman et al. (2009)
MARBL 11200 41,590,595–50,955,296 McClure et al. (2010)
1AGW = preweaning ADG; BWT = BW at birth; CALEASE = calving ease; CWT = carcass weight; FATTH = fat thickness at the 12th rib; FCR = feed conversion ratio; LMA = LM area; MARBL = marbling score; MHT = height at maturity; MWT = BW at maturity; PWG = postweaning ADG; RFI = residual feed intake; SB = stillbirth; TEND = tenderness score; WWT = BW at weaning; WWTMM = weaning weight-maternal milk; YWT = yearling BW; YRLHT = yearling height.
2ID = identification.
3Btau4.6 = bovine genome sequence assembly Btau4.6 (Elsik et al., 2015).

Birth Weight

The rs42512403 SNP was the only marker associated with BWT and was located at 68.0 Mb of BTA 15. This SNP presented an allelic substitution effect of −0.160 kg, explaining 2.6% of the phenotypic variance of BWT. Genes located closer to this SNP included C15H11orf74 (open reading frame of human C11orf74 chromosome 15), RAG2 and RAG1 (recombination-activating gene 2 and recombination-activating gene 1, respectively), and TRAF6 (TNF receptor-associated factor 6). The most important gene identified in this region was TRAF6. This gene is located 232.3 kb upstream of the rs42512403 SNP and belongs to the tumor necrosis factor receptor (TNF-R)-associated factor (TRAF) family. Functions related to this gene include the formation and differentiation of functional osteoclasts and the proliferation, apoptosis, and invasion of osteosarcoma cells (Meng et al., 2012). Several studies have highlighted its importance in peri- and postnatal survival in mice. Lomaga et al. (1999) found that when a mutation was induced in the TRAF6 gene, mice homozygous for the mutation were phenotypically normal at birth but died prematurely showing osteopetrosis, and the few surviving animals exhibited a 20 to 30% lower body mass and were shorter in length. This evidence suggests that TRAF6 could participate in the early stages of development in cattle and could, therefore, influence BWT. Quantitative trait loci located in this region have been associated with calving ease in Holstein cattle (Sahana et al., 2011) and with ADG in Brahman cattle (Peters et al., 2012).

Weaning Weight and Preweaning ADG

Four SNP markers were associated with WWT, and 2 of these markers were also associated with AGW, showing effects in the same direction. The average phenotypic variation explained by each marker for both WWT and AGW was 2.5%. One of these markers was rs43378829, located at 30.8 Mb of BTA 4, which showed allelic substitution effects of −0.753 kg on WWT and −0.003 kg on AGW. This SNP was found in the CDCA7L (Type 7 protein associated with cell division cycle) gene. Other genes located in this region were DNAH11 (dynein axonemal heavy chain 11) and RAPGEF5 (rap guanine nucleotide exchange factor 5). Although no evidence was found regarding the roles of these genes in growth-related processes, several QTL previously associated with growth characteristics have been identified in this region. For example, McClure et al. (2010) found QTL associated with WWT, YWT, height, and other features. Additionally, Thomasen et al. (2008) identified a QTL associated with calving ease. Others reported the QTL in this region have also been associated with marbling and LM area (Mizoshita et al., 2004; Yokouchi et al., 2009; McClure et al., 2010).

The second SNP associated with both characteristics, rs109462000, was found in a gene-dense region of BTA 11. This SNP exhibited allelic substitution effects of 1.293 and 0.005 kg on WWT and AGW, respectively. An analysis of this region in the UMD version 3.1 bovine genome assembly revealed the presence of 26 genes, extending from IFT172 (intraflagellar transport protein 172) to DPYSL5 (dihydropyrimidinase type 5). Among these 26 genes, only TCF23 (transcription factor 23) and CGREF1 (cell growth regulator with EF-hand domain 1) presented interesting features. According to the literature, TCF23 could participate in the inhibition of myogenesis, that is, the process of the determination and formation of muscle tissue cells. Regarding CGREF1, this gene is responsible for cell adhesion in a calcium-dependent manner and is capable of inhibiting growth in various cell lines. Furthermore, CGREF1 has been identified as one of the genes that is differentially expressed in meat samples of high and low quality (Bernard et al., 2007). Bernard et al. (2007) analyzed samples of the longissimus thoracis muscle of Charolais cattle and assessed the expression profiles of 5,418 genes. CGREF1 was among the 23 genes that were found to be overexpressed when meat samples exhibited higher tenderness, juiciness, and flavor. Coincidentally, the only QTL reported in this region has been associated with the marbling score (McClure et al., 2010).

Two other SNP that were exclusively associated with WWT were located at 31.9 and 12.3 Mb of BTA 18 and 29, respectively. On BTA 18, rs109028958 SNP presented an allelic substitution effect of 0.915 kg, and the closest gene to this marker was CDH11 (cadherin 11, type 2, OB-cadherin [osteoblast]), located 873.5 kb upstream. This gene encodes a type of calcium-dependent glycoprotein that belongs to the family of cadherins involved in cell adhesion processes that have been associated with osteogenesis (Marie et al., 2014). Studies in mice have suggested the importance of cadherins (mainly CDH11 and CDH2) in postnatal bone development and bone mass maintenance (Di Benedetto et al., 2010; Farber et al., 2011). In this region, McClure et al. (2010) also identified a QTL associated with WWT, and Sherman et al. (2009) and Seidenspinner et al. (2009) reported QTL associated with residual feed intake and stillbirth, respectively.

On BTA 29, the rs42339359 SNP showed an allelic substitution effect of −1.695 kg and was located near the genes DLG2 (discs, large homolog 2), CCDC90B (coiled-coil domain containing 90B), ANKRD42 (ankyrin repeat domain-containing protein 42), and PCF11 (cleavage and polyadenylation factor subunit). None of the genes identified in this genomic region presented functional information related to growth. However, an important gene in this region that has been associated with appetite regulation is located just 438.2 kb downstream of the rs42339359 SNP. This gene encodes a carboxypeptidase enzyme, PRCP (prolylcarboxypeptidase [angiotensinase C]), which is responsible for the degradation and inactivation of α-melanocyte stimulating hormone (α-MSH), an important regulator of energy metabolism (Wallingford et al., 2009; Diano, 2011). In vivo experiments conducted by Wallingford et al. (2009) demonstrated that the inhibition of its enzymatic activity causes decreases in food intake and weight in mice. Similar results were found by Jeong et al. (2012), who reported reductions in BW gain and fat, hepatic steatosis, and improved glucose metabolism when mice were fed a high-fat diet. Therefore, PRCP has been proposed as a genetic marker for controlling weight gain and obesity (Shariat-Madar et al., 2010). Previously reported QTL in this region have been associated with carcass weight (Casas et al., 2000, 2003; Kim et al., 2003; McClure et al., 2010) and residual feed intake (Nkrumah et al., 2007), among other characteristics (McClure et al., 2010). It is likely that the association found in BTA 29 may be due to the action of the PCRP gene and its possible influences on weight gain and postnatal growth in cattle.

Yearling Weight

Nine SNP markers were associated with YWT. Three of these markers, including rs133575890, rs135029160, and rs110875592, were located at 68.0 Mb of BTA 6. These SNP markers were in moderate linkage disequilibrium (average r2 = 0.48) and showed an average allele substitution effect of 1.468 kg. These markers were located within intronic regions of the CORIN (corin serine peptidase) gene. Other genes within this genomic region were ATP10D (ATPase, class V, type 10D) and NFXL1 (nuclear transcription factor, X-box-binding-like 1). The rs109300035 SNP was another marker located on BTA 6, but at 68.4 Mb. Unlike the SNP markers found in the CORIN gene, this SNP presented an allelic substitution effect of 1.411 kg on YW and was located within the TXK (TXK tyrosine kinase) gene. The genes NIPAL1 (NIPA-like domain containing 1), CNGA1 (cyclic nucleotide gated channel α 1), and TEC (Tec protein tyrosine kinase) were located in this region of BTA 6, in addition to NFXL1 and CORIN.

The functional information on these 2 genes is not directly related to growth processes. The CORIN gene encodes a serine-type endopeptidase and is involved in atrial natriuretic peptide hormone processing and blood pressure regulation. Thus far, this gene has been linked to only coat color in a population of Nelore-Angus cattle (Hanna et al., 2014). Regarding the other genes identified in these genomic regions of BTA 6, none of the genes showed functional relationships with growth characteristics. Moreover, according to searches in the AnimalQTLdb database (Hu et al., 2013), there were no previously reported QTL in the 2 identified regions of BTA 6.

At 95.8 Mb of BTA 2, the rs110545556 SNP exhibited an allelic substitution effect of 2.131 kg and was located within the KLF7 (Kruppel-like Factor 7 [ubiquitous]) gene. Other genes located on this region were MIR2355 (microRNA 2355) and CPO (carboxypeptidase O). KLF7 is a transcription factor belonging to the family of Kruppel-like factors. Similar to other members of this family, KLF7 is considered an important regulator of energy metabolism. Overexpression of this gene inhibits preadipocyte differentiation (Wu and Wang, 2013) and, therefore, has negative effects on adipogenesis (i.e., the process of adipocyte development) and fat metabolism. Zhang et al. (2013) associated the level of KLF7 mRNA expression with the abdominal fat content in divergently selected chickens and effectively observed that KLF7 overexpression inhibited the differentiation of preadipocytes but also increased their proliferation. More importantly, Ma et al. (2011) identified 2 SNP (A42075G and T42025C) located in intron 2 that were associated with growth traits in Nanyang cattle. Specifically, they found an association between the SNP A42075G with ADG at 6 mo of age and with YWT.

At 3.5 Mb of BTA 1, the rs134262111 SNP showed the largest effect on YWT, with an allele substitution effect of 5.767 kg. Even when the number of observations for this marker was low (minimum allele frequency = 0.06), this SNP was of interest due to its possible pleiotropic effect observed when lists of SNP (with a P-value < 0.005) were compared across traits. With a P-value of 0.001, this SNP, as well as 20 other SNP markers (including rs133575890 and rs110875592 of BTA 6), showed a suggestive association signal with WWT. This marker is located near microRNA MIR2284I and MIR2284X. MicroRNA are considered to be the general regulators of gene expression, and these molecules are known to control many biological processes, including cell differentiation, animal development, and metabolism, via the posttranscriptional regulation of transcription factors and other genes. Recent findings have shown that microRNA play important roles in the regulation of adipogenesis (Peng et al., 2014), although only certain microRNA have been identified as being differentially expressed in the adipose tissue of beef cattle (Jin et al., 2010). As in BTA 6, there were no previously reported QTL in this genomic region.

At 42.9 Mb of BTA 15, the rs109532225 SNP presented an allelic substitution effect of 2.536 kg and was located near the ADM (adrenomedullin), RNF141 (ring finger protein 141), LYVE1 (lymphatic vessel endothelial hyaluronan receptor 1), MRVI1 (murine retrovirus integration site 1 homolog), and SBF2 (SET-binding factor 2) genes. A recent study by Widmann et al. (2013) reported a possible influence of the MRVI1 gene in divergent growth of cattle. It is noteworthy that this previous study was based on the construction of a gene interaction network from GWAS results related to growth characteristics. According to their results, Widmann et al. (2013) identified MRVI1 as one of the genes with the greatest numbers of interactions with the NCAPG (non-SMC condensin I complex, subunit G) gene. The latter gene, located in the region from 38.57 to 38.1 Mb of BTA 6, is one of the most cited genes potentially influencing bovine growth (Snelling et al., 2010; Saatchi et al., 2014). In the model proposed by Widmann et al. (2013), MRVI1 and NCAPG could influence body growth through their effects on the nitric oxide–arginine pathway.

At 56.6 Mb of BTA 21, the rs109419103 SNP exhibited an allelic substitution effect of 2.6 kg. Genes located in this region were GPR68 (G protein coupled receptor 68), CCDC88C (coiled-coil domain containing 88C), and SMEK1 (SMEK homolog 1, suppressor of mek1 [dictyostelium]). However, despite the known functions of these genes, none of them demonstrated any direct relationship with growth. According to searches in the AnimalQTLdb database (Hu et al., 2013), there were no previously reported QTL in this genomic region of BTA 21.

At 24.5 Mb of BTA 8, the rs43538101 SNP exhibited an allelic substitution effect of 1.608 kg and was located within the SLC24A2 (solute carrier family 24 [sodium/potassium/calcium exchanger], member 2) gene. No relevant information was found regarding the function of SLC24A2. Quantitative trait loci previously associated with height, birth and carcass weight (McClure et al., 2010), and stillbirth (Kühn et al., 2003) have been reported in this genomic region.

On average, the phenotypic variance explained by these 9 aforementioned SNP was 4.6%; but excluding the largest effect found on BTA 1 (rs134262111; 5.767 kg), and the average allelic substitution effect for the SNP markers associated with YWT was 1.48 kg.

Postweaning ADG

Four SNP markers were associated with PWG, which explained 4.8% of the observed phenotypic variance on average. At 34.0 and 47.1 Mb of BTA 24, the rs41566285 and rs41585202 SNP showed allelic substitution effects of −0.008 and 0.006 kg, respectively. The closest genes to these SNP markers were RBBP8 (retinoblastoma-binding protein 8) for rs41566285 and KATNAL2 (katanin p60 subunit A-like 2), HDHD2 (haloacid dehalogenase-like hydrolase domain containing 2), IER3IP1 (immediate early response 3 interacting protein 1), and PIAS2 (protein inhibitor of activated STAT, 2) for rs41585202. Although it is unclear how RBBP8 might be related to growth, this gene has been reported as one of the loci associated with height in humans (Hirschhorn and Lettre, 2009). For the remainder of the genes, no reports were found regarding their influence on growth traits. Previously reported QTL found in these genomic regions have been associated with calving ease, WWT, YWT, marbling score (McClure et al., 2010), and feed conversion (Sherman et al., 2009).

The other SNP markers associated with PWG were rs41569509 and rs42219323, located at 9.3 and 80.1 Mb of BTA 9 and 16, respectively. On BTA 9, the rs41569509 SNP showed an effect of −0.007 kg and was located near the COL9A1 (collagen, type IX, α 1) gene. Some SNP (ss86352089 and ss86352090) located in COL9A1 have been associated with rib eye area (longissimus dorsi) in pigs (Fan et al., 2009). However, it is unclear how COL9A1 could functionally influence growth. In this genomic region, there have been previously reported QTL associated with residual feed intake (Sherman et al., 2009), BW (McClure et al., 2010), and marbling score (Imai et al., 2007).

On BTA 16, the only gene identified close to the rs42219323 SNP was MIR181A-1, located 417.0 kb upstream. As previously mentioned, microRNA are very important in the genetic regulation of various biological processes, including adipogenesis. According to Li et al. (2013), miR-181a is involved in adipogenesis via the regulation of tumor necrosis factor-α (TNF-α). Briefly, TNF-α is a multifunctional cytokine that inhibits adipocyte differentiation via the negative regulation of the transcription factors guiding adipogenesis (such as C/EBPα, CCAAT/enhancer binding protein alpha and PPARγ, peroxisome proliferator activated receptor gamma). When Li et al. (2013) overexpressed miR-181a in porcine adipocytes, they observed that the expression of TNF-α (a potential target of miR-181a) was inhibited. Moreover, lipid accumulation and an increase in the amount of triglycerides also occurred. Therefore, the increased expression levels of miR-181a favored adipogenesis. Li et al. (2013) also detected the overexpression of some genes associated with fatty acid synthesis, including PDE3B (phosphodiesterase 3B), LPL (lipoprotein lipase), PPARγ, GLTU1 (glucose transporter 1), GLUT4 (glucose transporter 4), adiponectin, FASN (fatty acid synthase), HSL (hormone-sensitive lipase), and ATGL (adipose triglyceride lipase). The AnimalQTLdb database (Hu et al., 2013) did not show the presence of previously reported QTL in this genomic region.

Other GWAS have reported associations of genomic regions (for example, on BTA 6 and 14) with several traits, including growth (Snelling et al., 2010; Saatchi et al., 2014). Although one of these genomic regions, located between 43.0 and 45.0 Mbp of BTA 6, showed a slight association signal (i.e., BWT), it was not sufficiently significant to reach the P-value threshold defined in this study. The discrepancy between our results and other GWAS can be explained by differences in the studied livestock populations (i.e., in the breed, size, and design of the population), applied statistical methods and models, marker density, and other factors. However, this study identified some genomic regions that had not been previously associated with growth (e.g., the QTL located on BTA 1, 2, 6, 9, and 21). Some of our results also overlapped with genomic regions previously associated with growth characteristics, as reported in the literature.

The phenotypic variance explained by the SNP identified in this study was very small (3.9% on average). In growth trait studies, it is expected that most SNP markers will explain only a small proportion of the observed phenotypic variance due to the polygenic control over such traits and because individual genes only slightly influence a phenotype. However, consideration of the set of SNP that are significantly associated with each trait may allow a greater proportion of phenotypic variance to be explained. For example, the 4 SNP associated with WWT were able to explain 10.1% of the variance in that trait.


The present GWAS identified 18 SNP associated with growth characteristics of Charolais cattle. Eight of these SNP were identified in intronic regions of the CDCA7L, CORIN, TXK, KLF7, and SLC24A2 genes. Although these associations can be considered to be moderately significant (compared with other GWAS), there is evidence that some of the genes are functionally related to growth through processes involving perinatal and postnatal survival, bone growth, cell adhesion, regulation of adipogenesis, and appetite. Therefore, we were able to define 5 candidate genes with potential associations with growth traits in Charolais beef cattle, including TRAF6, CDH11, KLF7, MIR181A-1, and PRCP. Subsequent studies examining these genomic regions could lead to the identification of polymorphisms with potential uses in marker-assisted selection, providing a deeper understanding of the genetic basis of growth traits in cattle.

This represents the first study to describe a GWAS conducted in beef cattle in Mexico. Further analysis using the present information as a basis would allow to conduct assessments on the ontogeny and specific search of causative mutations for live weight traits. Furthermore, the examination of particular and general genic effects would indicate the possibility to include genomic information into current genetic evaluations.




Be the first to comment.

Please log in to post a comment.
*Society members, certified professionals, and authors are permitted to comment.