Genome-wide Association Study Reveals the IBSP Locus Affect Ear Size in Cattle

Ear size is a classical model for hot climate adaptation following evolution, however, the genetic basis of the traits associated with the ear size remains to be elucidated. In the current study, we performed GWAS on 158 cattle individuals to to explain the genetic mechanism of ear size. The results suggested signicant association of IBSP locus with ear size. A missense mutation (Threonine-250 → Isoleucine) on the seventh exon of IBSP was observed, which occurred at a quite conserved site and changed the three-dimensional (3D) structure simulations. In addition to GWAS, 14 cattle breeds were screened for the selection signals associating with the ear size using F st and SweepD. The selective sweep analysis also suggested that IBSP was under positive selection amongst 4 breeds with relatively large ear size. The allele distribution of this mutation was validated among 394 samples from 21 worldwide cattle breeds, which strongly implied the origin of the A allele mutation to be from the Bos taurus. ndings


Background
Ear is an auditory organ capable of converting sound waves into nerve impulses (1). The central nervous system in the brain then translates these impulses into a spoken language, birds chirping or the cow's moo, and so on. Besides auditory perception, ear serve other vital functions as well, like heat radiation (2), sense of balance (3)(4)(5) and external signals in the mood (6). Ear consist of three sections: the external, middle, and the inner ear (7). The external ear includes an external auditory canal (8), eardrum and a visible pinna that resides outside of the head, and it can collect sound and directs it down the ear canal.
Over the course of time, mammals adapted to their respective environmental conditions and evolved an extraordinary variety of ear shapes and sizes. For example, animals bearing high temperature and hot environment, evolved larger ears to combat the heat waves, such as African elephants. African elephants adjust their body temperature through their large ears, providing shade to reduce the loss of water in the body (9). Furthermore, a general observation includes larger ear sizes in indicine cattle harboring in the harsh hot environment, such as Brahman and Burmese. On the contrary, most animals inhabitant in the tropical or colder regions tend to have smaller ear sizes, such as the Arctic fox. The smaller size of ear can help the mammals to reduce heat dissipation and maintain their body temperature. Likewise, the cattle adapted to the cold environments such as Yanbian, Yakut, and Tibetan cattle, all share smaller ear areas trait.
Size and orthotropism are important conformational characteristics of ears with huge diversity in different species. With the development of whole-genome resequencing technology, genome-wide association study (GWAS) has recently been applied to reveal SNPs associated with complex ear traits in livestock such as pig, dog and sheep. Previous studies have shown that signi cant quantitative trait loci (QTL) for pig ear size were located on Sus scrofa chromosome (SSCs) 5 and 7 (10,11). A missense mutation (G32E) in the PPARD gene on SSC7 was considered to be responsible for the differences in porcine ear size (12,13). QTL ne mapping on SSC5 revealed that LEMD3 was one of the most important candidates for porcine ear size (14). Moreover, GWAS analysis revealed that DCC, SOX5, and PTPRD were potential candidate genes for ear size in sheep (15). Furthermore, genome fragments containing (WIF1 and HMGA2) also appeared to control ear size in both pigs and dogs (16). However, there is no current report on the ear size of cattle, although the available resources of cattle are very rich worldwide.
Cattle is not only important its meat and milk production but also an important large-animal model for human disease. Among different cattle breeds, ear size is rich in diversity, Bos indicus, such as Brahman, with unusually large and lop ears. In contrast, Bos taurus have small and erect ears. According to this phenomenon, ear size may be regarded as an important characteristic distinguishing Bos taurus and Bos indicus. China has a wide geographical latitude span, with taurine and indicine cattle breeds distributed from north to south, and it is a natural cattle breed resource bank for studying ear size. Yunling cattle is a typical hybrid of Brahman bull × Murray Grey bull/Yunnan indigenous cross cow, which is an ideal model for studying the genetic basis of complex traits, such as ear size trait. In the current study, GWAS was performed to screen out potential candidate genes, which were further systematically analyzed to identify the associated loci and the candidate mutation in the particular gene or region. Our ndings provide new insights into the genetic basis of ear size of cattle and provide important molecular markers for the identi cation of cattle germplasm resources.

Ethics statement
According to the guidelines from the Council of China and the requirements of animal welfare, the relationship between researchers and cattle were handled very kindly. According to the recommendation of the Regulations for the Administration of Affairs Concerning Experimental Animals of China, the Institutional Animal Care and Use Committee of Northwest A&F University approved all animal experiments.
Sample collection and genome sequencing Individuals used for ear size trait comprised of 158 cattles (39 Brahman and 119 Yunling cattle). Yunling cattle is a typical hybrid cattle breed, bred by Yunnan Academy of Grassland and Animal Science. The ear tissue and blood samples were collected from adult females. The ear-level photos (with standard tick labels) and body size data of corresponding individuals were also collected in this study. The standard phenol-chloroform protocol was used to extract the genomic DNA from the ear (17). The sequence data used in this paper was obtained from published papers where detailed information about sampling and sequencing was available (18).

Measurement and calculation of ear size
In this study, we used the "living pixel method" to collect the ear shape photo with a standard scale of 475 cattles (39 Brahman, 119 Yunling, 65 Simmental 52 Burmese, 100 Wenshan, and 100 Dabieshan cattle; Table S1). The head of the cow was restrained and scale labelled left ear was attened for the picture.
Digital images were taken with a high -pixel camera in a horizontal position. Later the ear size of each animal was measured using Photoshop CS6 (Adobe, American). Firstly, the graph of the ear range and the graph of 1 cm 2 in the scale were obtained through the "magnetic lasso" tool, and the pixel numbers corresponding to the above two were obtained through the "histogram" tool. The size of the ear area could be obtained through the "measuring record" tool and the pixel ratio column.

Reads mapping and SNP calling
Default parameters were used to map clean reads to the cattle reference assembly GCF_002263795.1 using BWA-MEM (19). Duplicate reads were ltered using the "REMOVE_DUPLICATES = true" option of Picard tools. The "HaplotypeCaller", "GenotypeGVCFs" and "SelectVariants" argument of Genome analysis toolkit 3.8 (GATK) (20) were used for calling the raw SNPs. The average alignment rate and coverage were 99.54% and 5.61×, respectively. The argument "VariantFiltration" of the same software was applied to all raw SNPs with following options: QD < 2, FS > 60, MQRankSum < -12.5, ReadPosRankSum < -8.0 and mean sequence depth (for all individuals) > 1/3 × and < 3x. In addition, the haplotype-phase inference and missing alleles imputation were produced using Beagle (21) to carry out GWAS further. Based on ~ 41 M autosomal SNPs, we estimated the eigenvectors using smartPCA of EIGENSOFT v5.0 package (22) to adjust population structure in GWAS. The principal component 1 (PC1) based on the genotype matrix separated Brahman cattle from Yunling cattle, which is in accordance to the previous studies (18).

GWAS analysis
Based on the 158 sequenced genomes (Table S2), a total of 13,057,965 SNPs (MAF > 0.10; missing rate > 0.1) were used in GWAS for ear size trait. The primary association analysis was carried out using a genome-wide e cient mixed-model association (GEMMA) software package (23). The mixed linear model assumed the following model: where, y is a vector of phenotypes; α is a vector of xed effects representing marker effects; β is a vector of xed effects representing non-marker effects, and µ is a vector of unknown random effect. X, S, and K represent the incidence matrices relating α, β, and µ, respectively, and ε represents a vector of random residual effects. The top three PCs and feeding regimes were de ned as the S matrix. The kinship matrix calculated from nucleotide polymorphism was de ned as the K matrix.
The secondary association analysis used multiple linear regression model using PLINK (24), combining body height, cross high, head length, head width and rst PC as covariates to perform genome-wide association analysis on the ear size phenotype. To estimate the correction required for multiple testing, the number of linkage disequilibrium (LD)-pruned SNPs (750367) were de ned as the effective number of independent SNPs, was calculated using PLINK (--indep-pairwise 50 5 0.2). Therefore, the signi cance and suggestive threshold was de ned as approximately 5 × 10 − 8 (0.05/750,367) and 1 × 10 − 6 (1/750,367), respectively. In fact, the thresholds were widely used by numerous studies.

Identi cation of candidate genes in the GWAS associated loci
We used the following strategy to narrow down our ndings. Firstly, PLINK (24) was used to estimate candidate regions by pairwise linkage disequilibrium correlation (r 2 > 0.6) between SNPs related to ear size characteristics and the count of suggestive SNPs < 3 were used as false positives and removed.
Secondly, the SNPs (leading SNPs) with the P wald values < 1 × 10 − 6 were characterized as the candidate SNPs, and functional annotation of suggestive associated SNPs was carried out according to the Bos taurus reference genome in package ANNOVAR version (25).

Ear size variation traits analyses
The ear size among different cattle individuals and breeds were obviously different. As a general observation, the ear size of Bos indicus in Indo-Pak region was rather bigger than the Bos taurus in the world, while hybrid cattle breeds depicted diversi ed ear size (Fig. 1a). According to this observational phenomenon, the ear size data was collected and calculated from 475 adult female cattle include 6 representative breeds(i.e. Burmese, Brahman, Simmental, Yunling, Wenshan and Dabieshan cattle) using pixel method (29 Table 1). The ear size distribution of all individuals showed a unimodal distribution according the histogram and density plot (Fig. 1b). And it can be well distinguished in three distinct intervals: big ear (> 230 cm 2 ), middle ear (< 230 cm 2 & >148 cm 2 ) and small ear (< 148 cm 2 ) with 25% and 75% quantiles of the all individuals' boxplot analysis (Fig. 1c). Boxplot and F statistic was used to analyze the variance in the ear size among six cattle breeds. The results showed that Bos indicus depicted largest ear size, while Bos taurus represented the smallest ear size, while the hybrid (Yunling) and the Chinese indicine (Wenshan and Dabieshan cattle) ranged in the middle (Fig. 1c). Moreover, the F statistic results also suggested that the ear size among cattle breeds were signi cantly different, supporting the above statement.   (Table S3). However, most of the signi cant potential SNPs were located on BTA 6 (36.79 ~ 38.90 Mb) (Fig. 2b). The signi cant locus (P = 1.29 × 10 − 8 ) was located in LOC112447053, and near MEPE and IBSP. (Fig. 2c). The Quantile-Quantile plot (QQ-plot) in Fig. 2d showed the observed and expected P-values of the GWAS for ear size. The dashed line represents the distribution of the SNPs under the null hypothesis that there is no association of SNPs with the trait of interest. The strong deviation of the observed from the expected P-values for QQ-plots indicate that there were more SNPs signi cantly associated with all of the ear size trait than would be expected by chance.
Since, body size traits and ear size data could have a shared underlying genetic basis, the correlation was measured and analyzed for each trait in Figure S1. Pearson correlation coe cient was used to assess the linear relationships of ear size and body size traits. The result showed that the body height, cross high, head length and head width had a weak-to-moderate positive correlation with the ear size trait. The multiple linear regression model was analyzed using these related traits as covariates for secondary GWAS analysis using PLINK (24). The autosomal SNP scan for ear size revealed associated markers as showed in Figure S2. A total of 293 SNPs were observed on the potential region and most of them were found on BTA6. The most signi cant SNP (P = 5.74 × 10 − 11 ) was located on the intron of IBSP. The two GWAS results strongly suggested that IBSP had a strong correlation with ear size and might be a key candidate gene in uencing cattle ear size.
Genetic differentiation of the mutation in IBSP among different ear size groups To clarify whether ear size was selected in different cattle breeds, pairwise xation index (F st )(30) was used to measure the genetic differentiation between big (Brahman, Burmese) and small ear group  . 3a). To further investigate the differentiation of these mutation loci across diverse cattle breeds, 14 cattle breeds were used for nucleotide diversity (θπ) and SweepD analysis ( Figure S3, S4) (31,32). The results showed that the SweepD value on these loci were observed to be near zero in Bos taurus (Angus, Kazakh, Hanwoo, Hereford, Holstein, Mishima, Simental, and Yanbian cattle) and over 200 in South Asian indicine (Brahman, Burmese), which means that IBSP may be selected in indicine breeds (Fig. 3b). According to the functional annotation, a missense mutation was found on the seventh exon of IBSP (Threonine-250→Isoleucine, T250I) (Fig. 3c). To further evaluate the functional impact of the variants, we aligned the mutant IBSP protein with its ortholog proteins in bovidae (Fig. 3d) and other diverse vertebrates ( Figure S5). The comparison reveals that T250I was a quite conserved amino acid mutation, which is invariant among all the other animals we examined except Bos Taurus. We calculated the linkage disequilibrium (LD) values of the SNPs, which were shared in the region of IBSP. We observed strong linkage in this region ( Figure S6). At the same time, prediction based 3D structure of protein showed that the mutation site of IBSP can change its protein structure (Fig. 3e). All those results imply that T250I is most likely causal mutation for the IBSP sweep in cattle ear size.
Allele frequency of IBSP mutation among different cattle breeds indicated the origin In our study, we genotyped the DNA sequences of IBSP with 394 cattle representing Bos indicus, Bos taurus and hybrid cattle to investigate the genotype frequency of the mutation loci across diverse cattle breeds. Two genotypes (A and G) of the IBSP mutation loci were found (Fig. 4a, 4b).  (Fig. 4b). The allele distribution indicates that the A allele of IBSP may have originated in Bos taurus. Due to the in ltration of South Asian indicine, the Chinese indicine caused the reduce in the frequency of A allele. We also found the G allele in African taurine at low frequency, which may be due to the large introduction of South Asian indicine after the African rinderpest in 1890 (34).

Discussion
The external ear of cattle is mainly composed of cartilage and thin layers of skin. The function of cartilage in the external ear is a stent which determine the size and the shape of the ear (35). In the current study, IBSP as the main candidate gene in uencing the ear size of cattle were screened out with two GWAS methods. In addition to the functional analysis, this gene is potentially the biological candidate genes for ear size characteristics. Integrin Binding Sialoprotein (IBSP) had been mentioned many times in numerous studies of cartilage and chondrocytes. The main IBSP-knockout mice showed a shorter body, reduced chondrocyte proliferation, and impaired cartilage absorption. At the same time, the lack of IBSP can alter bone growth, the formation and mineralization of primary bone (36). The protein encoded by IBSP is the main structural protein in the bone matrix, and nearly 12% non-collagenous proteins are from IBSP in human osteoblasts, hypertrophic chondrocytes and osteoclasts (37)(38)(39)(40). IBSP is mainly related to tissue development and cell growth, which may relate to its mechanism in promoting the growth and proliferation in bovine ear chondrocytes.
We discovered and validated one missense mutation at IBSP in Bos taurus, of which T250I occurred in a well-de ned domain and quite conserved site ( Figure S5). In nearly 70 vertebrate species, this mutation was conserved, except for Bos Taurus. We also found this missense mutation in IBSP altering the 3D structure of the expressed protein, which may change the function of the protein. At the same time, the statistical analysis of IBSP missense genotype among different cattle breeds showed that GG genotype was mainly present in Bos indicus with large ear size, AA genotype was mainly present in Bos taurus which had small ear size, and three genotypes were present in hybrid cattle which was living in south of China and with complex ear size. Through the mutation frequency of IBSP missense variant among different cattle breeds, we found that the lowest frequency of A allele was in South Asian indicine, the frequency was near or equal to one in East Asian taurine, European taurine and Eurasian taurine. Moreover, the mutations of the A allele was mainly found in cattle with small ear size living in cold environments, while the relatively large ear size of cattle living in tropical or subtropical regions had almost no mutation frequency. We speculated that the missense mutation of IBSP may be the main factor that changes the cattle ear size, and speculate that the mutations of the A allele in IBSP may originate from Bos taurus. The difference in ear size may be due to the variable environmental conditions in which they are located. The larger ear area can help Bos indicus to dissipate heat temperature, and the small ear size may be more helpful for Bos taurus to keep their temperature in the cold environment. The temperature was not high or low in southern China, so hybrid cattle has the diverse ear size phenotype.

Conclusions
In conclusion, the IBSP loci related to cattle ear size was screened out by GWAS, and the missense mutation in IBSP (T250I) was speculated to be the candidate mutation affecting ear size among cattle breeds. These ndings not only have important theoretical signi cance for the exploration of major genes of ear area traits but also provide important molecular markers for the identi cation of cattle germplasm resources.
Abbreviations None Declarations