Genome-Wide Association Study Provides Insights Into the Genetic Architecture of Bone Size and Mass in Chickens

Genome

Genome-wide association study provides insights into the genetic architecture of bone size and mass in chickens

Journal: Genome

Manuscript ID gen-2019-0022.R2

Manuscript Type: Article

Date Submitted by the 20-Nov-2019 Author:

Complete List of Authors: GUO, Jun; Jiangsu Institute of Poultry Science, layer breeding Qu, Liang; Jiangsu Institute of Poultry Science DOU, Taocun; Jiangsu Institute of Poultry Science, layer breeding Shen, Manman; Jiangsu Institute of Poultry Science Hu, Yuping;Draft Jiangsu Institute of Poultry Science Ma, Meng; Jiangsu Institute of Poultry Science WANG, Kehua; Jiangsu Institute of Poultry Science, layer breeding

Bone length, Dominance effect, Genome-wide association study, Keyword: Heritability, Linear mixed model

Is the invited manuscript for consideration in a Special Not applicable (regular submission) Issue? :

https://mc06.manuscriptcentral.com/genome-pubs Page 1 of 29 Genome

1 Genome-wide association study provides insights into the genetic

2 architecture of bone size and mass in chickens

4 Jun Guo, Liang Qu, Tao-Cun Dou, Man-Man Shen, Yu-Ping Hu, Meng Ma and Ke-Hua Wang*

6 Jiangsu Institute of Poultry Science, Key Laboratory for Poultry Genetics and Breeding of Jiangsu

7 province, Yangzhou, Jiangsu, 225125, China

9 *Corresponding author:Kehua Wang

10 Jiangsu Institute of Poultry Science at Yangzhou, China.

11 Mailing address: Draft

12 No. 58 Cangjie Road, 225125, Yangzhou , China

13 Tel: +86(514) 85599012

14 Fax: +86(514)85599035

15 Phone number:13805276606

16 Email:[email protected]

https://mc06.manuscriptcentral.com/genome-pubs Genome Page 2 of 29

19 Abstract: Bone size is an important trait for chickens due to its association with osteoporosis in

20 layers and meat production in broilers. Here, we employed high density genotyping platforms to

21 detect candidate genes for bone traits. Estimates of the narrow heritabilities ranged from 0.37 ±

22 0.04 for shank length to 0.59 ± 0.04 for tibia length. The dominance heritability was 0.12±0.04

23 for shank length. Using a linear mixed model approach, we identified a promising locus within

24 NCAPG on chromosome 4, which was associated with tibia length and mass, femur length and

25 area and shank length. In addition, three other loci were associated with bone size or mass at a

26 Bonferroni-corrected genome-wide significance threshold of 1%. One region on chicken

27 chromosome 1 between 168.38 and 171.82 Mb, harboredHTR2A, LPAR6, CAB39L, and TRPC4. 28 A second region that accounted for 2.2%Draft of the phenotypic variance was located around WNT9A 29 on chromosome 2, where allele substitution was predicted to be associated with tibia length. Four

30 candidate genes identified on chromosome 27 comprising SPOP, NGFR, GIP, and HOXB3 were

31 associated with tibia length and mass, femur length and area, and shank length. Genome

32 partitioning analysis indicated that the variance explained by each chromosome was proportional to

33 its length.

34 Keywords: Bone length; Dominance effect; Genome-wide association study; Heritability; Linear

35 mixed model

https://mc06.manuscriptcentral.com/genome-pubs Page 3 of 29 Genome

37 Introduction

38 Bone growth is of importance to poultry production as skeletal problems are associated with

39 economic losses and welfare issue(Bradshaw et al. 2002; Kapell et al. 2012). Long bone

40 distortions are a common disease in broiler production, although the causes of these deformities

41 are multifactorial diseases. In Europe, about 44 billion broiler chickens with leg disorders are

42 sacrificed annually (Turner et al. 2003). Osteoporosis leads to the unbalanced bone resorption and

43 the loss of structural bone, and it is a major health problem in layers. Osteoporosis accounts for

44 20% to 35% of all mortalities during the egg laying cycle in caged hens (Anderson 2002;

45 Whitehead and Fleming 2000). Poultry selection breeding systems have traditionally focused on

46 the improvement of economic traits, but the selection on welfare traits such as bone-related traits

47 has been observed in recent years (Kapell et al. 2012; Whitehead 2004).

48 In general, bone length and mass are regardedDraft as important parameters for evaluating bone growth

49 in chickens and other species (Tsudzuki et al. 2007; Gao et al. 2010). Numerous factors can affect

50 bone length and mass, and most probably have genetic origins. González-Cerón et al. (2015)

51 reported that the heritabilities of tibia length and mass in a broiler populationwere0.54 ± 0.07 and

52 0.31 ± 0.06, respectively. Ragognetti et al. (2015) found that the heritabilities of tibia length and

53 mass in a F2 populationwere0.23 ± 0.08 and 0.23 ± 0.07, respectively. Abdellatif (1989) reported

54 that the heritability of shank length was 0.58 ± 0.28. Tsudzuki et al. (2007) estimated that shank

55 length had a high heritability of 0.37 in a crossbred population. Bone length and mass are highly

56 genetically determined, but the genetic architecture that underlies these traits is still poorly

57 defined.

58 The aim of the present study was to elucidate the genetic architecture of bone traits in the chicken

59 using a genome-wide association study (GWAS) approach. To do this we used an F2 population

60 obtained from White Leghorn and Dongxiang Blue-shelled chickens, and identified single

61 nucleotide polymorphisms (SNPs) related to bone growth. We investigated whether the bone

https://mc06.manuscriptcentral.com/genome-pubs Genome Page 4 of 29

62 traits were determined by common variants and estimated the genetic parameters for the bone

63 traits.

65 Material and methods

66 Ethics statement

67 All procedures involving animals were in compliance with the guidelines for the care and use of

68 experimental animals established by the Ministry of Agriculture of China. The ethics committee

69 of Jiangsu Poultry Science Institute specifically approved the present study.

70 Study design

71 The F2 population was generated from a reciprocal cross between an indigenous breed (Dongxiang Blue-shelled

72 chickens) and commercial layer(White Leghorn). Dongxiang Blue-shelled chickens were introduced from 73 Jiangxi province into an experimental farm Draft at Jiangsu Institute of Poultry Science in 1998. White Leghorn 74 chickens were kindly provided by Shanghai Poultry Breeding Company. Six White Leghorn cocks were mated

75 with 133 Dongxiang hens and six Dongxiang cocks were mated with 80 White Leghorn hens, and the F1

76 generations comprised of 1,029 birds (White Leghorn cocks × Dongxiang hens)and 552 birds (Dongxiang cocks

77 × White Leghorn hens).The F2 generations were produced by the F1 individuals within the respective crosses to

78 obtain 1,856 cockerels and 1,893 pullets. All of the F2 birds were measured to determine their bone traits. The

79 laying mash contained 16.16% crude protein, 10.64 MJ·kg−1 metabolizable energy, 3.4% calcium, and 0.52%

80 total phosphorus.

81 Trait measurement

82 The shank length was quantified using a measuring tape at 508 days old as the distance from the

83 hock joint to the tarsometatarsus. The F2 chickens were then sacrificed to obtain other bone

84 measurements. The right femur and tibiae were dissected from the carcass. Muscle and connective

85 tissue were carefully removed from the bone with a scalpel. The tibia and femur lengths were

86 measured with Vernier calipers. The femur area was determined by dual energy X-ray

87 absorptiometry using the Discovery DXA system (Hologic, Inc. Bedford, MA, USA).

88 Genotyping 4

https://mc06.manuscriptcentral.com/genome-pubs Page 5 of 29 Genome

89 DNA was extracted from the whole blood according to a standard protocol(Moore and Dowhan

90 2002). Genotyping analyses of 1,534 samples from the F2 generation were conducted using an

91 Affymetrix Axiom 600K Chicken Genotyping Array (Affymetrix, Inc., Santa Clara, CA, USA).

92 More details of the genotyping array were described by Kranis et al. (2013). The quality of the

93 array data was evaluated with Affymetrix Power Tools and PLINK software (Purcell et al. 2007).

94 SNPs were excluded if they had a minor-allele frequency (MAF) <1%. SNPs that deviated from

95 the Hardy–Weinberg equilibrium (P value < 1e−6) were removed. SNPs on the sex chromosome

96 were removed. Samples with call rates <95% were removed. Phasing analyses were performed

97 with Beagle software (version 4.0) (Browning and Browning 2007). Finally, 435,867 autosomal

98 SNPs and 1,512 samples passed the quality control procedure.

99 Association analysis

100 Before the association test, an independentDraft SNP set was established using PLINK with a window

101 size of 25 SNPs, a step of five SNPs, and an r2 threshold of 0.2. The principal components (PCs)

102 were then obtained using the linkage equilibrium SNPs. The top five PCs were assigned as

103 covariates in the linear mixed model. In the present study, the effective number of independent

104 tests was 59,308, and thus, the genome-wide suggestive and significant P-values were 1.69e−5 and

105 8.43e−7, respectively. We searched for the candidate genes closest to the associated SNPs in

106 GeneCards (http://www.genecards.org/), Ensembl (http://asia.ensembl.org, version 89 ), and

107 NCBI (http://www.ncbi.nlm.nih.gov) database. The positions of interesting SNPs were obtained

108 from Ensembl version 89 and NCBI Gallus_gallus-5.0.

109 Conditional analysis was conducted to examine the potential associated SNPs that might be

110 masked by a strong signal. Briefly, the initial screen involved testing with the strongest SNP

111 covariate. Association analysis conditioning was then implemented based on the selected SNP(s)

112 to iteratively search for the top SNPs one by one using a stepwise model selection procedure until

113 no SNP had a conditional P-value that passed the significance level.

https://mc06.manuscriptcentral.com/genome-pubs Genome Page 6 of 29

114 A standard linear mixed model implemented in GEMMA was used to correct for the stratification

115 of the study population (Zhou and Stephens 2012). The statistical model used for the association

116 test can be represented by the matrix expression:

117 y  W  x  u  

118 where y is the vector of phenotypic values for all F2 chickens, W is the incidence matrix of fixed

119 effects including including a column of ones; αis the vector of the corresponding coefficients

120 including the intercept, x is the vector of marker genotypes, β is the corresponding effect size of

2 2 121 the marker; u is the vector of random effects with u~(0,GRM a ), where  a is the genetic

122 variance and GRM represents the genomic relationship matrix, where, GRM was calculated with

123 the GCTA package and used to replace A matrices (Yang et al. 2011); and ε is the vector of the 124 residual variances with ε~N (0,I 2 ), Draft where  2 is the residual variance component. The

125 statistical model employed was as described above for single-marker GWAS. The Wald test

126 statistic was used as a standard to select SNPs associated with bone traits. The genomic inflation

127 factor was calculated using the R package GenABEL (Aulchenko et al. 2007). Manhattan plots

128 were generated with R as described by Yi et al.(2015). Quantile–quantile (Q-Q) plots were used

129 to analyze the extent to which the observed distribution of the test statistic followed the expected

130 distribution. Q-Q plots were obtained with the gap package in R(Zhao 2007).

131 Heritability

132 In order to investigate the roles of additive and dominant genetic contributions to bone traits, we

133 applied a restricted maximum likelihood method to estimate the heritability based on the genomic

134 relationship matrix as implemented in the GCTA package. The model could be also extended to

135 estimate the dominant genetic variance (Zhu et al. 2015). Genetic correlations were obtained with

136 a bivariate animal model. Phenotypic correlations were obtained using WOMBAT software

137 (Meyer 2007). We estimated the genetic variance explained by chromosomes and individual

138 SNPs with the additive genetic GRM. The mixed linear model was used to estimate the genetic 6

https://mc06.manuscriptcentral.com/genome-pubs Page 7 of 29 Genome

139 variance for each chromosome. We partitioned the chicken genome into 28 autosomes and two

140 linkage groups. GCTA allowed the relationships among individuals to be replaced by the GRM

141 and the eigenvectors obtained by principal component analysis (PCA) were embedded as

142 covariates to correct for population stratification. In this analysis, the first five PCs were included

143 as covariates in order to adjust the structure of the stratified population and the cryptic

144 relatedness.

145 Data availability

146 The authors state that all data necessary for confirming the conclusions presented in the article are

147 represented fully within the article.

148 Results

149 We collected bone growth trait data from the F2 population. The F2 chickens were slaughtered at

150 a mean age of 508 (±1) days. The coefficientDraft of variation varied from 4.1% to 14.1%. GWAS

151 analyses and genetic estimation were performed for shank length, tibia length and mass, and

152 femur length and area.

153 Genome-wide association studies

154 The region-specific and conditional analysis results for the bone traits are localized on four

155 chromosomes (Table 1). The inflation factors for P-values varied from 1.02 (for shank length) to

156 1.12 (for femur area), thereby suggesting that the PCs and covariates used in the analysis

157 efficiently decreased the impact of population stratification. The results represented as a

158 Manhattan plot based on the long bone length showed that the significant SNPs were mapped to

159 chromosomes 1, 2, 4 and 27 (Figure 1). The Q-Q plot shown in Figure 1 (right) provides a

160 graphical representation of the GWAS analysis results on bone length. The curve of the observed

161 P-value clearly deviated from the null distribution, which suggests that the observed P-values at

162 the tail of the distribution were smaller than those expected by chance and they probably reflected

163 genuine associations. The determined genetic associations are summarized (Table 1).

https://mc06.manuscriptcentral.com/genome-pubs Genome Page 8 of 29

164 Single-marker association analysis showed that four loci reached the genome-wide significance

165 level (P=8.43×10−7). The most significant SNP related to femur length was located at 3.42 Mb on

166 chromosome 27. The most significant signal for tibia and shank length was found on rs14491030,

167 which resulted in a missense mutation. In addition to the regions on chromosome 4 and

168 chromosome 27, two loci were detected that had associations with long bone length, which were

169 located in a region close to 169 Mb on chromosome 1 and 2.79 Mb on chromosome 2. These

170 genetic loci included SNPs located in or neighboring HTR2A (shank), TRPC4 (tibia), WNT9A

171 (tibia), PPARGC1A (femur), LCORL (tibia), NCAPG (femur, shank, tibia), SPOP (tibia),

172 TNFRSF16/NGFR (femur, tibia), GIP (femur), and HOXB3 (shank). The Venn diagram in Figure

173 2 presents the numbers of shared SNPs associated with femur, shank, and tibia length.

174 GWAS identified three regions that had significant association with femur area and tibia mass,

175 and the Q-Q plot for these two traits supportedDraft the results shown in the Manhattan plot (Figure 3).

176 The three regions were located on chromosomes 1, 4, and 27. The peak genome-wide significant

177 SNP effects for femur area and tibia mass were found on chromosome 4, where most of the

178 significant associated loci spanned the region from 74.3 Mb and 77.9 Mb. The distribution of

179 significant SNPs related to tibia mass is shown in Figure 3.

180 To identify any signals that might be masked by the strong signals, we performed conditional

181 single-marker association analyses based on 139 SNPs with P-value was lower than 8.43e-7 for

182 tibia mass, where 132 dropped out and seven were retained as genome-wide significant (Table 1).

183 In addition, 91 out of 95 SNPs associated with femur area dropped out in after stepwise

184 conditional analysis. A SNP at 169.2 Mb on chromosome 1 was identified as a significant marker

185 for tibia mass and femur area, and it was located in the intron of the CAB39L gene. Moreover, no

186 significant association was found on chromosome 1 after conditional analysis of two additional

187 variants. No additional signals on chromosomes 2, 4, and 27 were detected in our conditional

188 analyses.

https://mc06.manuscriptcentral.com/genome-pubs Page 9 of 29 Genome

189 Genetic parameters

190 The additive genetic heritabilities estimated from the genomic relationship matrix were moderate

191 to high, where they ranged from 0.46 ± 0.04 for shank length to 0.66 ± 0.04 for tibia length (Table

192 2). The phenotypic correlation coefficients were lower than the corresponding genetic

193 correlations. Significant positive dominance variances were found in the bone traits (Table 3).

194 The heritability estimates were reduced after including dominant effects in the genetic model.

195 Partitioning of genomic variants by chromosomes

196 To assess the distribution of the genetic components on femur area, we investigated the

197 phenotypic variance explained by each chromosome. As shown in Figure 4, the variance

198 explained by each chromosome was proportional to its length (adjusted R2=0.67). These results

199 are consistent with the infinitesimal modelDraft theory, i.e., common variants throughout the genome

200 accounted for the total variances. In total, 28 autosomes and two linkage groups accounted for

201 67.13% of the variance in femur area. These results did not differ from the univariate heritability

202 estimates (Table 1). The contributions of particular chromosomes differed from each other. Two

203 chromosomes, chromosome 1 and 4, explained more than 5.00% of the variance in femur area.

204 Discussion

205 In the present study, we found that bone size and mass in F2 chickens had a strong genetic

206 component with moderate heritability. We identified variants associated with bone size and mass

207 traits that mapped to the well-known LCORL-NCAPG locus and to three regions on

208 chromosomes 1, 2, and 27. Moreover, genetic evaluations suggested that common SNPs were an

209 important contributors to the variations in bone size and mass. These findings could be useful for

210 understanding the genetic architecture of bone traits and they may provide molecular markers for

211 practical breeding applications.

https://mc06.manuscriptcentral.com/genome-pubs Genome Page 10 of 29

212 The most convincing possible association identified in the present study was between the bone

213 traits and SNPs in the LCORL-NCAPG region on chromosome 1. Previous studies reported that

214 this region contains genetic elements that regulate body size in horses(Metzger et al. 2013;

215 Staiger et al. 2016; Tetens et al. 2013), cattle (Karim et al. 2011; Pausch et al. 2011; Setoguchi et

216 al. 2011), sheep (Matika et al. 2016; Wang et al. 2015) and humans (Soranzo et al. 2009;

217 Visscher et al. 2010; Yang et al. 2010). In cattle, a missense mutation in the NCAPG gene can

218 potentially led to a difference in body frame size (Setoguchi et al. 2011). In horses, indels in

219 the 3′-untranslated region sequence of the LCORL gene is associated with skeletal variation

220 (Staiger et al. 2016). In chicken, the LCORL/NCAPG genes were found to have associations with

221 body weight (Gu et al. 2011; Liu et al. 2013), daily feed consumption (Wolc et al. 2013) and egg

222 weight (Yi et al. 2015). Comparing the quantitative trait loci (QTL) locations for similar traits in

223 chickens may help to identify candidate Draftgenes associated with bone traits. Over 10 independent

224 study groups have reported QTL linked with bodyweight, tibia weight, carcass weight, and shank

225 length QTL in this region (www.animalgenome.org/cgi-bin/QTLdb/GG/index).

226 Our GWAS demonstrated that one SNP located within the WNT9A gene was significantly

227 associated with tibia length. The WNT9A is a key regulator of Indian hedgehog (Ihh), which plays

228 an important role in the proliferation and function of chondrocytes during skeletogenesis (Später

229 et al. 2006)). Chondrocyte maturation is crucial for bone size because premature or delayed

230 maturation leads to shortened bones (Spagnoli et al. 2007). The WNT9A is considered to be

231 critical for determining where the joints will be formed, and it was shown that Wnt/β-catenin

232 signaling is necessary and sufficient for inducing the early steps of synovial joint formation (Guo

233 et al. 2004).

234 The SNP rs316174212 located in the first intron of the TRPC4 gene was significantly associated

235 with tibia length and mass, and femur area. TRPC4 is a channel protein that modulates calcium

https://mc06.manuscriptcentral.com/genome-pubs Page 11 of 29 Genome

236 entry and it is known to be involved in cell proliferation. The preservation of bone size and mass

237 relies on adequate proliferation, the secretion of matrix proteins, and the bone metabolism

238 balance. Osteoblast proliferation can be trigged by calcium influx through the activation of

239 TRPCs (Abed et al. 2009; Labelle et al. 2007). Therefore, variants in the TRPC4 gene might

240 affect the activity of osteoblasts. Further experiments are necessary to explore the roles of the

241 TRPC4 gene.

242 The variants of the NGFR gene were predicted to be causally linked with bone length, femur area,

243 and tibia mass in this population. NGFR encodes the receptor of nerve growth factor (NGF) and it

244 belongs to the Tumor Necrosis Factor (TNF) superfamily. A previous study of osteoblastic cells

245 demonstrated that the expression levels of NGF are harmonious with bone homeostasis (Togari et 246 al. 2000). NGF enhanced bone regenerationDraft in a fracture model and stimulated the differentiation 247 of osteoblastic cells (Grills et al. 1997; Mogi et al. 2000). In our previous study, we identified the

248 TNF superfamily member RANK as a candidate gene related to the bone mineral content (Guo et

249 al. 2017). It is reasonable to assume that the effects of NGF on bone traits are probably mediated

250 via alterations in its receptor.

251 Estimates of genetic parameters based on the genomic relationship matrix have recently been

252 developed to correct the pedigree error and to improve the accuracy of heritability (Hayes et al.

253 2009; Legarra et al. 2009). Most of the traits considered in this study (except for shank length)

254 were such difficult-to-measure traits and it is necessary to accumulate records for several

255 generations. Estimates from a shallow pedigree depth may underestimate the actual relationship

256 between two individuals, thereby leading to erroneous parameters. We dissected the variance

257 components using a mixed linear model coupled with a genomic relationship matrix. The

258 heritability of shank length was similar to estimates obtained in previous studies based on a

259 pedigree matrix (González-Cerón et al. 2015; Tsudzuki et al. 2007). The other traits have rarely

https://mc06.manuscriptcentral.com/genome-pubs Genome Page 12 of 29

260 been estimated using pedigree-based approaches. The introduction of dominance genetic effects

261 into animal models was imaportant for animal breeding. First, it provided more accurate estimated

262 breeding values according to Toro and Varona (2010). Second, it provided useful information

263 regarding mate allocation. Thus, dominance models have been used in genetic evaluations or

264 selection for other species (Mielenz et al. 2006; Serenius et al. 2006; Nagy et al. 2014). The

265 positive genetic correlations suggest that any selection based on shank length would have positive

266 effects on the tibia and femur dimensions.

267

268 In the present study, we used a resource population crossbred from a local breed and White

269 Leghorn: (i) to identify candidate genes involved in bone growth and development, and (ii) to

270 estimate genetic parameters associated withDraft bone traits. The long-term goal of our study is to gain

271 a better understanding of the genetic architecture of long bone growth in chickens. The bone traits

272 measured in this population were polygenic traits because they were under the control of a large

273 number of SNPs, and each SNP was weakly associated with bone size. Our analysis supports a

274 dominant genetic effect on bone traits.

275 Acknowledgements

276 The authors thank the Ministry of Agriculture of the People’s Republic of China (grant No. CARS-40-K01), the

277 Agriculture Committee of Jiangsu Province (grant No. PZCZ201729).

278 279 References:

280 Abdellatif, M. A. 1989. Genetic study on Dandarawy chickens. II. Heritability of live and carcass

281 measurements. Genet. Sel. Evol.,21(2):199-203. doi: 10.1186/1297-9686-21-2-199.

282 Abed, E., Labelle, D., Martineau, C., Loghin, A., and Moreau, R. 2009. Expression of transient

283 receptor potential (TRP) channels in human and murine osteoblast-like cells. Mol. Membr.

284 Biol.,26(3):146-158. doi: 10.1080/09687680802612721. PMID: 1911514. 12

https://mc06.manuscriptcentral.com/genome-pubs Page 13 of 29 Genome

285 Anderson, K. (Yeaer) Final Report of the 34th North Carolina Layer Performance and

286 Management Test. NC State University, Cooperative Extension Service, Raleigh, pp. 15-20.

287 Aulchenko, Y. S., Ripke, S., Isaacs, A., and van Duijn, C. M. 2007. GenABEL: an R library for

288 genome-wide association analysis. Bioinformatics,23(10):1294-1296. doi:

289 10.1093/bioinformatics/btm108. PMID: 17384015.

290 Bradshaw, R., Kirkden, R., and Broom, D. 2002. A review of the aetiology and pathology of leg

291 weakness in broilers in relation to welfare. Avian Biol Res,13(2):45-103. doi:

292 10.3184/147020602783698421

293 Browning, S. R., and Browning, B. L. 2007. Rapid and Accurate Haplotype Phasing and

294 Missing-Data Inference for Whole-Genome Association Studies By Use of Localized

295 Haplotype Clustering. Am. J. Hum. Genet.,81(5):1084-1097. doi: 10.1086/521987. PMID:

296 17924348. Draft

297 Gao, Y., Du, Z.Q., Feng, C.G., Deng, X.M., Li, N., Da, Y. and Hu, X.X., 2010. Identification of

298 quantitative trait loci for shank length and growth at different development stages in

299 chicken. Anim. Genet. 41(1):101-104. doi: 10.1111/j.1365-2052.2009.01962.x. PMID:

300 19917046.

301 González-Cerón, F., Rekaya, R., and Aggrey, S. E. 2015. Genetic analysis of bone quality traits

302 and growth in a random mating broiler population. Poult. Sci.,94(5):883-889. doi:

303 10.3382/ps/pev056.PMID: 25784765.

304 Grills, B. L., Schuijers, J. A., and Ward, A. R. 1997. Topical application of nerve growth factor

305 improves fracture healing in rats. J. Orthop. Res.,15(2):235-242. doi:

306 10.1002/jor.1100150212.PMID: 9167626.

307 Gu, X., Feng, C., Ma, L., Song, C., Wang, Y., Da, Y., Li, H., Chen, K., Ye, S., Ge, C., Hu, X.,

308 and Li, N. 2011. Genome-wide association study of body weight in chicken F2 resource

309 population. PLoS One,6(7):e21872. doi: 10.1371/journal.pone.0021872.PMID: 21779344.

https://mc06.manuscriptcentral.com/genome-pubs Genome Page 14 of 29

310 Guo, J., Sun, C., Qu, L., Shen, M., Dou, T., Ma, M., Wang, K., and Yang, N. 2017. Genetic

311 architecture of bone quality variation in layer chickens revealed by a genome-wide

312 association study. Sci. Rep.,7:e45317. doi: 10.1038/srep45317. PMID: 28383518.

313 Guo, X., Day, T. F., Jiang, X., Garrett-Beal, L., Topol, L., and Yang, Y. 2004. Wnt/β-catenin

314 signaling is sufficient and necessary for synovial joint formation. Genes

315 Dev.,18(19):2404-2417. doi: 10.1101/gad.1230704. PMID: 15371327.

316 Hayes, B. J., Visscher, P. M., and Goddard, M. E. 2009. Increased accuracy of artificial selection

317 by using the realized relationship matrix. Genet Res,91(2):47-60. doi:

318 10.1017/S0016672308009981. PMID: 19220931.

319 Kapell, D., Hill, W., Neeteson, A.-M., McAdam, J., Koerhuis, A., and Avendano, S. 2012.

320 Twenty-five years of selection for improved leg health in purebred broiler lines and

321 underlying genetic parameters. Poult.Draft Sci.,91(12):3032-3043. doi: 10.3382/ps.2012-02578.

322 PMID: 23155010.

323 Karim, L., Takeda, H., Lin, L., Druet, T., Arias, J. A., Baurain, D., Cambisano, N., Davis, S. R.,

324 Farnir, F., and Grisart, B. 2011. Variants modulating the expression of a chromosome

325 domain encompassing PLAG1 influence bovine stature. Nat. Genet.,43(5):405-413. doi:

326 10.1038/ng.814. PMID: 21516082.

327 Kranis, A., Gheyas, A. A., Boschiero, C., Turner, F., Yu, L., Smith, S., Talbot, R., Pirani, A.,

328 Brew, F., Kaiser, P., Hocking, P. M., Fife, M., Salmon, N., Fulton, J., Strom, T. M.,

329 Haberer, G., Weigend, S., Preisinger, R., Gholami, M., Qanbari, S., Simianer, H., Watson,

330 K. A., Woolliams, J. A., and Burt, D. W. 2013. Development of a high density 600K SNP

331 genotyping array for chicken. BMC Genomics,14:59. doi: 10.1186/1471-2164-14-59.

332 PMID: 23356797.

333 Labelle, D., Jumarie, C., and Moreau, R. 2007. Capacitative calcium entry and proliferation of

334 human osteoblast-like MG-63 cells. Cell Prolif.,40(6):866-884. doi:

https://mc06.manuscriptcentral.com/genome-pubs Page 15 of 29 Genome

335 10.1111/j.1365-2184.2007.00477.x. PMID: 18021176.

336 Legarra, A., Aguilar, I., and Misztal, I. 2009. A relationship matrix including full pedigree and

337 genomic information. J. Dairy Sci.,92(9):4656-4663. doi: 10.3168/jds.2009-2061. PMID:

338 19700729.

339 Liu, R., Sun, Y., Zhao, G., Wang, F., Wu, D., Zheng, M., Chen, J., Zhang, L., Hu, Y., and Wen, J.

340 2013. Genome-wide association study identifies Loci and candidate genes for body

341 composition and meat quality traits in Beijing-You chickens. PLoS One,8(4):e61172. doi:

342 10.1371/journal.pone.0061172.PMID: 23637794.

343 Matika, O., Riggio, V., Anselme-Moizan, M., Law, A. S., Pong-Wong, R., Archibald, A. L., and

344 Bishop, S. C. 2016. Genome-wide association reveals QTL for growth, bone and in vivo

345 carcass traits as assessed by computed tomography in Scottish Blackface lambs. Genet. Sel.

346 Evol.,48:11. doi: 10.1186/s12711-016-0191-3.Draft PMID: 26856324.

347 Metzger, J., Schrimpf, R., Philipp, U., and Distl, O. 2013. Expression levels of LCORL are

348 associated with body size in horses. PLoS One,8(2):e56497. doi:

349 10.1371/journal.pone.0056497. PMID: 23418579.

350 Meyer, K. 2007. WOMBAT—A tool for mixed model analyses in quantitative genetics by

351 restricted maximum likelihood (REML). J Zhejiang Univ Sci B,8(11):815-821. doi:

352 10.1631/jzus.2007.B0815. PMID: 17973343.

353 Mielenz, N., Noor, R.R., Schüler, L., 2006. Estimation of additive and non-additive genetic

354 variances of body weight, egg weight and egg production for quails (Coturnix coturnix

355 japonica) with an animal model analysis. Archives Animal Breeding 49:300-307. doi:

356 10.5194/aab-49-300-2006.

357 Mogi, M., Kondo, A., Kinpara, K., and Togari, A. 2000. Anti-apoptotic action of nerve growth

358 factor in mouse osteoblastic cell line. Life Sci.,67(10):1197-1206. doi:

359 10.1016/S0024-3205(00)00705-0. PMID: 10954053.

https://mc06.manuscriptcentral.com/genome-pubs Genome Page 16 of 29

360 Moore, D., and Dowhan, D. 2002.Purification and concentration of DNA from aqueous solutions.

361 John Wiley & Sons, New Jersey.

362 Nagy, I., Farkas, J., Curik, I., Gorjanc, G., Gyovai, P., Szendrő, Z., 2014. Estimation of additive

363 and dominance variance for litter size components in rabbits. Czech J. Anim. Sci

364 59(4):182-189.doi: 10.17221/7342-CJAS.

365 Pausch, H., Flisikowski, K., Jung, S., Emmerling, R., Edel, C., Götz, K.-U., and Fries, R. 2011.

366 Genome-wide association study identifies two major loci affecting calving ease and

367 growth-related traits in cattle. Genetics,187(1):289-297. doi: 10.1534/genetics.110.124057.

368 PMID: 21059885.

369 Purcell, S., Neale, B., Todd-Brown, K., Thomas, L., Ferreira, M. A., Bender, D., Maller, J., Sklar,

370 P., De Bakker, P. I., and Daly, M. J. 2007. PLINK: a tool set for whole-genome association

371 and population-based linkage Draft analyses. Am. J. Hum. Genet.,81(3):559-575. doi:

372 10.1086/519795.PMID: 17701901.

373 Ragognetti, B. N. N., Stafuzza, N. B., Silva, T. B. R., Chud, T. C. S., Grupioni, N. V., Cruz, V. A.

374 R., Peixoto, J. O., Nones, K., Ledur, M. C., and Munari, D. P. 2015. Genetic parameters and

375 mapping quantitative trait loci associated with tibia traits in broilers. Gen. Mol.

376 Res.,14(4):17544-17554. doi: 10.4238/2015.December.21.27. PMID: 26782399.

377 Serenius, T., Stalder, K., Puonti, M., 2006. Impact of dominance effects on sow longevity. J.

378 Anim. Breed. Genet. 123(6):355-361. doi: 10.1111/j.1439-0388.2006.00614.x. PMID:

379 17177689.

380 Setoguchi, K., Watanabe, T., Weikard, R., Albrecht, E., Kühn, C., Kinoshita, A., Sugimoto, Y.,

381 and Takasuga, A. 2011. The SNP c.1326T>G in the non-SMC condensin I complex, subunit

382 G (NCAPG) gene encoding a p.Ile442Met variant is associated with an increase in body

383 frame size at puberty in cattle. Anim. Genet.,42(6):650-655. doi:

384 10.1111/j.1365-2052.2011.02196.x.PMID: 22035007.

https://mc06.manuscriptcentral.com/genome-pubs Page 17 of 29 Genome

385 Soranzo, N., Rivadeneira, F., Chinappen-Horsley, U., Malkina, I., Richards, J. B., Hammond, N.,

386 Stolk, L., Nica, A., Inouye, M., and Hofman, A. 2009. Meta-analysis of genome-wide scans

387 for human adult stature identifies novel Loci and associations with measures of skeletal

388 frame size. PLoS Genet.,5(4):e1000445. doi: 10.1371/journal.pgen.1000445. PMID:

389 19343178.

390 Später, D., Hill, T. P., O'Sullivan, R. J., Gruber, M., Conner, D. A., and Hartmann, C. 2006.

391 Wnt9a signaling is required for joint integrity and regulation of Ihh during chondrogenesis.

392 Development,133(15):3039-3049. doi: 10.1242/dev.02471. PMID: 16818445.

393 Spagnoli, A., O'Rear, L., Chandler, R. L., Granero-Molto, F., Mortlock, D. P., Gorska, A. E.,

394 Weis, J. A., Longobardi, L., Chytil, A., and Shimer, K. 2007. TGF-β signaling is essential

395 for joint morphogenesis. J. Cell Biol.,177(6):1105-1117. doi: 10.1083/jcb.200611031.

396 PMID: 17576802. Draft

397 Staiger, E. A., Al Abri, M. A., Pflug, K. M., Kalla, S. E., Ainsworth, D. M., Miller, D., Raudsepp,

398 T., Sutter, N. B., and Brooks, S. A. 2016. Skeletal variation in Tennessee Walking Horses

399 maps to the LCORL/NCAPG gene region. Physiol. Genomics,48(5):325-335. doi:

400 10.1152/physiolgenomics.00100.2015. PMID: 26931356.

401 Tetens, J., Widmann, P., Kühn, C., and Thaller, G. 2013. A genome-wide association study

402 indicates LCORL/NCAPG as a candidate locus for withers height in German Warmblood

403 horses. Anim. Genet.,44(4):467-471. doi: 10.1111/age.12031.PMID: 23418885.

404 Togari, A., Mogi, M., Arai, M., Yamamoto, S., and Koshihara, Y. 2000. Expression of mRNA for

405 axon guidance molecules, such as semaphorin-III, netrins and neurotrophins, in human

406 osteoblasts and osteoclasts. Brain Res.,878(1-2):204-209. doi:

407 10.1016/S0006-8993(00)02700-1. PMID: 10996153.

408 Toro, M. A., and L. Varona. 2010. A note on mate allocation for dominance handling in genomic

409 selection. Genet. Sel. Evol. 42:33. doi: 10.1186/1297-9686-42-33. PMID: 20699012.

https://mc06.manuscriptcentral.com/genome-pubs Genome Page 18 of 29

410 Tsudzuki, M., Onitsuka, S., Akiyama, R., Iwamizu, M., Goto, N., Nishibori, M., Takahashi, H.,

411 and Ishikawa, A. 2007. Identification of quantitative trait loci affecting shank length, body

412 weight and carcass weight from the Japanese cockfighting chicken breed, Oh-Shamo

413 (Japanese Large Game). Cytogenet. Genome Res.,117(1-4):288-295. doi:

414 10.1159/000103190. PMID: 17675870.

415 Turner, J., Garces, L., and Smith, W. 2003. The welfare of broiler chickens in the European

416 Union. A report by Compassion in World Farming trust distributed in association with The

417 European Coalition for Farm Animals.

418 Visscher, P. M., McEVOY, B., and Yang, J. 2010. From Galton to GWAS: quantitative genetics

419 of human height. Genet Res,92(5-6):371-379. doi: 10.1017/S0016672310000571. PMID:

420 21429269.

421 Wang, H., Zhang, L., Cao, J., Wu, M., DraftMa, X., Liu, Z., Liu, R., Zhao, F., Wei, C., and Du, L.

422 2015. Genome-Wide Specific Selection in Three Domestic Sheep Breeds. PLoS

423 One,10(6):e0128688. doi: 10.1371/journal.pone.0128688. PMID: 26083354.

424 Whitehead, C., and Fleming, R. 2000. Osteoporosis in cage layers. Poult. Sci.,79(7):1033-1041.

425 doi: 10.1093/ps/79.7.1033. PMID: 10901207.

426 Whitehead, C. C. 2004. Overview of bone biology in the egg-laying hen. Poult.

427 Sci.,83(2):193-199. doi: 10.1093/ps/83.2.193. PMID: 14979569.

428 Wolc, A., Arango, J., Jankowski, T., Settar, P., Fulton, J. E., O’Sullivan, N. P., Fernando, R.,

429 Garrick, D. J., and Dekkers, J. C. M. 2013. Pedigree and genomic analyses of feed

430 consumption and residual feed intake in laying hens. Poult. Sci.,92(9):2270-2275. doi:

431 10.3382/ps.2013-03085. PMID: 23960108.

432 Yang, J., Benyamin, B., McEvoy, B. P., Gordon, S., Henders, A. K., Nyholt, D. R., Madden, P.

433 A., Heath, A. C., Martin, N. G., and Montgomery, G. W. 2010. Common SNPs explain a

434 large proportion of the heritability for human height. Nat. Genet.,42(7):565-569. doi:

https://mc06.manuscriptcentral.com/genome-pubs Page 19 of 29 Genome

435 10.1038/ng.608. PMID: 20562875.

436 Yi, G., Shen, M., Yuan, J., Sun, C., Duan, Z., Qu, L., Dou, T., Ma, M., Lu, J., Guo, J., Chen, S.,

437 Qu, L., Wang, K., and Yang, N. 2015. Genome-wide association study dissects genetic

438 architecture underlying longitudinal egg weights in chickens. BMC Genomics,16:746. doi:

439 10.1186/s12864-015-1945-y. PMID: 26438435.

440 Zhao, J. H. 2007.A Genetic Analysis Package with R. J. Stat. Softw.,23(8):1-18.

441 Zhou, X., and Stephens, M. 2012. Genome-wide efficient mixed-model analysis for association

442 studies. Nat. Genet.,44(7):821-824. doi: 10.1038/ng.2310. PMID: 22706312.

443 Zhu, Z., Bakshi, A., Vinkhuyzen, AnnaA. E., Hemani, G., Lee, SangH., Nolte, IljaM.,

444 vanVliet-Ostaptchouk, JanaV., Snieder, H., Esko, T., Milani, L., Mägi, R., Metspalu, A.,

445 Hill, WilliamG., Weir, BruceS., Goddard, MichaelE., Visscher, PeterM., and Yang, J. 2015.

446 Dominance Genetic Variation ContributesDraft Little to the Missing Heritability for Human

447 Complex Traits. Am. J. Hum. Genet.,96(3):377-385. doi: 10.1016/j.ajhg.2015.01.001.

448 PMID: 25683123.

449

https://mc06.manuscriptcentral.com/genome-pubs Genome Page 20 of 29

450 Table 1 Loci associated with bone traits in F2 chickens

nearest/candidate Proportion Length of block Supporting SNP Chr position Distance (kb) MAF beta wald Trait gene phenotypic variance (kb) SNPs

rs312726815 1 168378181 HTR2A upstream 16 kb 0.40(C/A) 0.04(0.01) 0.039 1.31E-07 1541 4 Shank length

//-0.10(0.02) //0.046 //1.84E-08 //4418 //138 // Tibia mass

rs313207223 1 168801990 LPAR6 upstream 2 kb 0.48(G/A) -0.07(0.01) 0.049 5.07E-07 4454 94 Femur area

// -0.11(0.02) //0.054 //5.51E-08 //4418 //138 // Tibia mass

rs313913043 1 169162192 CAB39L intron variant 0.35(A/G) //0.06(0.01) //0.048 //2.41E-07 //285 //4 // Femur area

// 0.11(0.02) //0.045 //1.62E-10 //1395 //7 // Tibia mass

rs316174212 1 171824791 TRPC4 intron variant 0.41(T/C) -0.08(0.01) 0.060 1.88E-09 4454 94 Femur area Draft//-0.48(0.09) //0.018 //5.84E-08 //3917 //33 // Tibia length // -0.11(0.02) //0.057 //1.78E-09 //4418 //138 // Tibia mass

rs313359226 2 2788128 WNT9A downstream 124 kb 0.05(A/G) -0.77(0.13) 0.022 1.42E-08 539 3 Tibia length

rs14490329 4 74354802 PPARGC1A intron variant 0.09(G/A) 0.37(0.07) 0.019 6.01E-07 2735 8 Femur length

// 0.15(0.02) //0.026 //5.30E-10 //4048 //59 // Tibia mass

rs314487178 4 76400165 LCORL intron 0.23(T/C) 0.45(0.08) 0.026 2.50E-08 3780 56 Tibia length

// 0.10(0.02) //0.033 //3.38E-09 //4048 //59 // Tibia mass

rs14491030 4 76458342 NCAPG missense 0.06(G/A) 0.16(0.02) 0.051 7.76E-19 2623 81 Femur area

V497D//V497G // 1.03(0.12) //0.042 //7.56E-17 //3780 //56 // Tibia length

//0.45(0.08) //0.021 //2.26E-08 //2735 //8 //Femur length

// 0.09(0.01) //0.029 //1.12E-11 //1458 //14 // Shank length

// 0.19(0.03) //0.034 //7.94E-15 //4048 //59 // Tibia mass

rs313458995 27 3351243 SPOP upstream 12.99 kb 0.13(T/C) 0.48(0.09) 0.018 1.08E-07 608 20 Tibia length

rs316237751 27 3416927 TNFRSF upstream 6.3 kb 0.09(T/C) 0.10(0.02) 0.025 1.63E-09 412 5 Femur area

https://mc06.manuscriptcentral.com/genome-pubs Page 21 of 29 Genome

16(NGFR) //0.72(0.11) //0.030 // 4.39E-11 //608 //20 // Tibia length

//0.47(0.07) //0.035 // 1.20E-11 //608 //19 //Femur length

// 0.14(0.02) //0.027 //1.62E-09 //467 //9 // Tibia mass

rs315528383 27 3681806 GIP upstream 2 kb 0.17(A/G) 0.28(0.05) 0.019 1.17E-07 608 19 Femur length

rs313491013 27 3841884 HOXB3 intron variant 0.06(A/T) 0.06(0.01) 0.017 4.71E-07 5 2 Shank length

Draft

https://mc06.manuscriptcentral.com/genome-pubs Genome Page 22 of 29

452 453 Table 2Genetic parameters estimated with a genomic relationship matrix and additive

454 model

shank length tibia length tibia mass femur length femur area

shank length 0.46 ± 0.04 † 0.45 ± 0.03 0.33 ± 0.03 0.31 ± 0.03 0.37 ± 0.03

tibia length 0.82 ± 0.04 0.66 ± 0.04 0.66 ± 0.02 0.70 ± 0.02 0.73 ± 0.02

tibia mass 0.68 ± 0.05 0.80 ± 0.03 0.62 ± 0.04 0.28 ± 0.02 0.84 ± 0.01

femur length 0.70 ± 0.06 0.85 ± 0.03 0.83 ± 0.04 0.55 ± 0.04 0.59 ± 0.02

femur area 0.76 ± 0.05 0.86 ± 0.02 0.98 ± 0.01 0.82 ± 0.04 0.61 ± 0.04

455 †Data above the diagonal correspond to the phenotypic correlation coefficients; data

456 on the diagonal are the heritability estimated with an additive model; data below the 457 diagonal correspond to the geneticDraft correlation coefficients.

https://mc06.manuscriptcentral.com/genome-pubs Page 23 of 29 Genome

458 Table 3Variance components and heritabilities estimated with additive and dominant

459 models

2 2 2 2 2 2  a  d  e  p ha hd

shank length 0.05 ± 0.01 0.02 ± 0.01 0.07 ± 0.01 0.15 ± 0.01 0.37 ± 0.04 0.12 ± 0.04

tibia length 9.55 ± 0.99 0.83 ± 0.48 5.98 ± 0.65 16.35 ± 0.81 0.59 ± 0.04 0.05 ± 0.03

tibia mass 0.52 ± 0.06 0.03 ± 0.03 0.41 ± 0.04 0.93 ± 0.05 0.48 ± 0.04 0.03 ± 0.03

femur length 5.23 ± 0.65 0.46 ± 0.21 5.51 ± 0.51 11.20 ± 0.52 0.47 ± 0.04 0.04 ± 0.02

femur area 0.32 ± 0.03 0.01 ± 0.01 0.25 ± 0.02 0.58 ± 0.03 0.55 ± 0.04 0.02 ± 0.02

460

Draft

https://mc06.manuscriptcentral.com/genome-pubs Genome Page 24 of 29

461 Table S1 Summary of bone traits in the population

traits Sample size Minimum Maximum Mean SD

Shank length cm 1479 5.35 8.78 7.603 0.479

Tibia length mm 1433 71.51 120.51 107.339 4.359

Tibia mass g 1436 3.90 11.00 7.145 1.007

Femur length mm 1412 55.05 94.67 73.599 3.595

Femur area cm2 1435 6.07 12.88 9.078 0.807

462

Draft

https://mc06.manuscriptcentral.com/genome-pubs Page 25 of 29 Genome

463 Figure captions

464 Figure 1 Manhattan and quantile–quantile (QQ) plot of long bone length

465 Manhattan plot showing the association of tibia length with chromosomes (X-axis).

466 The p-values of the association test were transformed to –log10 (p-values). The

467 horizontal green and yellow lines indicate the whole‑genome significance (P-value =

468 8.43×10−7) and suggestive thresholds (P-value = 1.69×10−5), respectively. A QQ plot

469 depicting the distribution of expected P-values against observed P-values for each

470 single nucleotide polymorphism.

471 Figure 2 Venn diagram illustrating the overlap of single nucleotide polymorphisms

472 associated with bone length Draft

473 Figure 3 Manhattan and quantile–quantile (QQ) plot of femur area and tibia mass

474 Figure 4 Estimates of chromosome-wise heritability on femur area are drawn against

475 the chromosome length (x-axis)

476 The blue line represents heritability regressed on chromosome length. Grey area

477 around the blue line is the 95% confidence level interval for prediction from the

478 linear model. Chromosomes 1, 4, 6 and 27 fall outside the 95% confidence interval,

479 indicating that these chromosomes could explain more heritability than expected

480 according to chromosome length.

https://mc06.manuscriptcentral.com/genome-pubs Genome Page 26 of 29

Draft

Figure 1 Manhattan and quantile–quantile (QQ) plot of long bone length

https://mc06.manuscriptcentral.com/genome-pubs Page 27 of 29 Genome

Draft

Figure 2 Venn diagram illustrating the overlap of single nucleotide polymorphisms associated with bone length

https://mc06.manuscriptcentral.com/genome-pubs Genome Page 28 of 29

Draft

Figure 3 Manhattan and quantile–quantile (QQ) plot of femur area and tibia mass

https://mc06.manuscriptcentral.com/genome-pubs Page 29 of 29 Genome

Draft

Figure 4 Estimates of chromosome-wise heritability on femur area are drawn against the chromosome length (x-axis)

270x203mm (300 x 300 DPI)

https://mc06.manuscriptcentral.com/genome-pubs