Novel Nucleotide Variations, Haplotypes Structure and Associations with Growth Related Traits of Goat AT Motif-Binding Factor (ATBF1) Gene

The AT motif-binding factor (ATBF1) not only interacts with protein inhibitor of activated signal transducer and activator of transcription 3 (STAT3) (PIAS3) to suppress STAT3 signaling regulating embryo early development and cell differentiation, but is required for early activation of the pituitary specific transcription factor 1 (Pit1) gene (also known as POU1F1) critically affecting mammalian growth and development. The goal of this study was to detect novel nucleotide variations and haplotypes structure of the ATBF1 gene, as well as to test their associations with growth-related traits in goats. Herein, a total of seven novel single nucleotide polymorphisms (SNPs) (SNP 1-7) within this gene were found in two well-known Chinese native goat breeds. Haplotypes structure analysis demonstrated that there were four haplotypes in Hainan black goat while seventeen haplotypes in Xinong Saanen dairy goat, and both breeds only shared one haplotype (hap1). Association testing revealed that the SNP2, SNP5, SNP6, and SNP7 loci were also found to significantly associate with growth-related traits in goats, respectively. Moreover, one diplotype in Xinong Saanen dairy goats significantly linked to growth related traits. These preliminary findings not only would extend the spectrum of genetic variations of the goat ATBF1 gene, but also would contribute to implementing marker-assisted selection in genetics and breeding in goats.

To date, few studies about the nucleotide variations of goat ATBF1 gene and its effects on growth traits have been reported. To improve understanding of goat ATBF1 gene, this work firstly explored the novel nucleotide variations, haplotypes structure of goat ATBF1 gene, and analyzed its associations with growth related traits. These findings would not only extend the spectrum of genetic variations of the goat ATBF1 gene, but also would contribute to implementing MAS in genetics and breeding in goats.

Animals and data collection
In this study, a total of 707 goats from two well-known Chinese native goat breeds (Hainan Black goats [HNBG] = 284; Xinong Saanen dairy goats [XNSN] n = 423) were used. All selected individuals were healthy and unrelated. The HNBG goats were 2 to 3 years old and reared in native breeding farms, in Zanzhou County, Hainan province, China. All XNSN individuals were 2 to 6 years old, among which 21.3%, 50.8%, 8.9%, 12.7%, and 6.3% were 2 years old, 3 years old, 4 years old, 5 years old, and 6 years old, respectively. The XNSN goats were reared on Chinese native dairy goat breeding farm in Qianyang County, Shaanxi Province, China .

DNA isolation and DNA pool construction
Extraction of DNA samples from ear tissues and blood leukocytes (Sambrock et al., 2001;Green et al., 2012) were diluted to working concentration (50 ng/μL) according to our previous report . A total of 50 DNA samples from two breeds were randomly selected to construct DNA pools, which were used as templates for polymerase chain reaction (PCR) amplification to explore SNPs of ATBF1 gene.

Primers design and DNA sequencing
The 5' UTR, exons, introns and 3' UTR regions of the goat ATBF1 gene were amplified from the constructed DNA pools. Fourteen pairs of primers were designed to amplify the goat ATBF1 gene using Primer Premier Software (version 5.0) based on the sheep ATBF1 gene sequence (GenBank Accession No. NC_019471) as the goat was not available (Table 1). PCR reactions were performed in 25 μL volume containing 50 ng genomic DNA, 0.5 μM of each primer, 1× Buffer (including 1.5 mM MgCl 2 , 200 μM dNTPs and 0.625 units of Taq DNA polymerase [MBI, Vilnius, Lithuania]). The Touch-Down PCR protocol was as follows: denatured at 95°C for 5 min, followed by 35 cycles of 94°C for 30 s, 68°C to 51°C for 30 s, and 72°C for 2 min, finally extended at 72°C for 10 min. Then to sequence accurately, the products were sequenced only when they had a single objective band of each pair of primers.
In order to detect these SNPs, the PCR-restriction fragment length polymorphism (RFLP) and PCRamplification-created restriction site (ACRS)-RFLP were carried out. i) For the NC_019471:g.25504 G>A (SNP1) locus, the endonuclease Thermus aquaticus YT-1 (TaqI) (TCGA) was used to genotype the SNP of g.25504 G, not g.25504 A. ii) For the NC_019471: g.25748 G>A (SNP2) locus, created restriction endonuclease Moraxella species (Msp1) site (CCGG) was formed when the forward primer actual nucleotide "T" was induced to "C" at NC_019471: g.25746 locus. Thus the Msp1 could recognize the SNP of g.25748 G with induced point mutation g.25746 C, not with g.25746 T. iii) For the NC_019471: g.26902 A>G (SNP3) locus, new restriction endonuclease Haemophilus influenzae Rf (HinfI) site (GANTC) was established by changing the reverse primer actual nucleotide "A" to "T" at NC_019471: g.26905 locus. Then the SNP of g.26902 G with induced point mutation g.26905 T could be genotyped by HinfI PCR -ACRS-RFLP, rather than g.26905 A. iv) For the NC_019471: g.32001 C>G (SNP4) locus, the endonuclease Bacillus megaterium T110 (AvaI) site (CYCGRG) was used to genotype the allele of g. 32001 G, not the g. 32001 C. v) Since the NC_019471: g.32029 A>G (SNP5) also could not be genotyped by the natural restriction or economic restriction endonuclease, the other reverse primer was designed to form new restriction endonuclease Streptomyces achromogenes (ScaII) (CCGCGG) point. The actual nucleotide "A" was induced into "G" at the NC_019471: g.32031, so the Streptomyces achromogenes (ScaII) could genotype the SNP of g.32029 G with induced point mutation g.32031G, not with g.32031 A. vi) For the NC_019471: g.163442 C>G (SNP6) locus, the endonuclease pancreatic secretory trypsin inhibitor (PstI) (CTGCAG) was used to genotype the SNP of g. 163442 G, not g. 163442 C. vii) For the NC_019471: g.163517A>G (SNP7) locus, the endonuclease MspI (CCGG) was used to genotype the SNP of g. g.163517 G, not g. g.163517 A.
The associations of the genetic variations and growthrelated traits were calculated according to the general linear model by the SPSS software (version 18.0) (International Business Machines [IBM] Corporation, New York, USA) for Windows. Statistical testing was carried on the records of growth traits of HNBG and XNSN goats. The mixed statistical of the linear model analysis, not including the effects of farm, sex, season of birth (spring versus fall), age of dam and sire, which had no significant effects on the variation of traits in the mammal populations (Lan et al., 2007;Zhao et al., 2013). Therefore, the statistical linear model was: Y ijk = μ+A i +G j +e ijk, , where Y ijk is the observation of the body measurement traits, μ is the overall mean of each trait, A i is the fixed effect of age, G j is the fixed effect of genotype or combined genotype, and e ijk is the random residual error (He et al., 2014;Wang et al., 2014). Thus the fixed effect of genotypes and age was a major source of variation and the p-value for the difference between the least squares means was less than 0.05. Diplotypes of combined haplotypes of SNPs with growth traits correlation analysis were carried out to explore the possible interactions between the SNPs. The model was similar to above model analysis, except that the interaction between two SNPs was treated as a fixed effect.

Novel nucleotide variations within goat ATBF1 gene
After DNA sequencing and alignment analysis, seven SNPs loci were firstly found, namely, SNP1-7 (Figure 1). The SNP1-TaqI locus (25504 G>A) was located at exon 2 and mutated from G to A, resulting in a missense mutation, CGA (372 R) to CAA (372 Q), which could be genotyped by the TaqI PCR-RFP method (Figure 2a). The SNP2-MspI locus (25748 G>A) was located at exon 2 and mutated from G to A, resulting in a synonymous change, TCG (453 Ser) to TCA (453 Ser), which could be genotyped by the MspI PCR-ACRS-RFP method (Figure 2b). The SNP3-HinfI locus (26902 A>G) was located at exon3 and mutated from A to G, resulting in a missense change, AAA (453 K) to TCA (453 E), which could be genotyped by the HinfI PCR-ACRS-RFP method (Figure 2c).The SNP4-AvaI locus (32001 C>G) was located at intron 3 and mutated from C to G, which could be genotyped by the AvaI PCR-RFP method (Figure 2d). The SNP5-ScaII locus (32029 A>G) was located at intron 3 and mutated from A to G, which could be genotyped by the ScaII PCR-ACRS-RFP method ( Figure  2e). The SNP6-PstI locus (163442 C>G) was located at exon 8 and mutated from C to G, which could be genotyped by the PstI PCR-RFP method (Figure 2f). The SNP7-MspI locus (163517A>G) was located at intron 8 and mutated from A to G, which could be genotyped by the MspI PCR-RFP method (Figure 2g).

Frequencies of genotypes and alleles within goat ATBF1 gene
Statistics analysis showed that the frequencies of genotypes and main alleles are different at different SNP loci in two goat breeds (Table 2). For example, only one genotype of SNP4-AvaI, SNP5-SacII, and SNP6-PstI was found in HNBG, but three genotypes were found in XNSN dairy goat. The frequencies of two alleles of each SNP locus in XNSN dairy goat, SNP4-AvaI and SNP5-SacII loci were approximately same except the SNP6-PstI locus. As shown in Table 2  diversity except those that had only one kind of genotype and most SNPs loci were at HWE except SNP2-MspI and SNP5-SacII loci in XNSN dairy goat and SNP7-MspI locus in HNBG.

Haplotype structure and linkage disequilibrium analysis
Four haplotypes were found in HNBG while seventeen haplotypes in XNSN dairy goat (Table 3). Only 1 haplotype (hap 1) was simultaneously found in both breeds, but the frequency was low (8.5%). The frequency of the hap 4 (27.5%) was highest in HNBG, and the hap 13 (14.1%) was the highest in XNSN dairy goat. The LD of seven SNPs in two populations was analyzed. As shown in Table 4    G (0.737).

Relationships between the genetic variations and related-growth traits
The associations of the genetic variations with growth related traits except SNP1 and SNP3 loci were determined (Table 6). In the SNP2-MspI locus, the genotype of AG had demonstrated significantly superior HuWI traits than genotype GG in HNBG, while genotype GG was found to have significantly superior BL, ChC, and ChCI traits when compared with genotype AA, as well as genotype GG and AG had significantly superior BLI traits in XNSN dairy goat. The different genotypes of SNP5-ScaII locus had significantly associated with BW, demonstrating that the genotype AA and GG was superior to AG in XNSN dairy goat. The different genotypes of SNP6-PstI locus had significant associate with BL, which demonstrated that the genotype CC and GG was superior to CG in XNSN dairy goat. In SNP7-MspI locus, the different genotypes were found to be significantly associate with CaC and CaCI traits in HNBG and TI trait in XNSN dairy goat. For the locus, the genotype GG was superior in HNBG and genotype AA and AG in XNSN dairy goat.

Effects of the interaction of each two single nucleotide polymorphisms to growth traits
Though the r 2 values of HNBG between SNP6 and SNP7 were low, but at the same time, the D' values were  high (0.861), so we analyzed the effects of the interaction between SNP6 and SNP7 of HNBG with growth traits as well as between SNP4 and SNP5 (0.670), SNP4 and SNP6 (0.574), SNP4 and SNP7 (0.642), SNP6 and SNP7 (0.737) of XNSN. As shown in Table 7, the diplotypes of SNP6 and SNP7 were found to have significant effects on ChC (p = 0.025). The phenotype ChC trait of combined genotypes CC-AA, CC-AG, CC-GG, CG-AG, and GG-GG was greater than CG-GG in XNSN.

DISCUSSION
As a cancer suppressor gene, ATBF1 gene not only regulates cell proliferation and differentiation (Ninomiya et al., 2002;Ishii et al., 2003;Jung et al., 2011), but also interacts with PIAS3 to suppress STAT3 signaling way (Nishio et al., 2012;Jiang et al., 2014). Most importantly, ATBF1 is necessary for the Pit1 gene activation, indicating that ATBF1 could indirectly participate in the regulative roles of Pit1 gene, including regulating Wnt/beta-catenin pathway and POU1F1 pathway (Carvalho et al., 2006;Qi et al., 2008;Davis et al., 2010). All these functional experiments suggested that the ATBF1 gene would affect growth traits of livestock. Therefore, this work studied the relationship between the nucleotide variations of this gene and growth related traits in goats.  We found seven novel SNPs, of which two were missense mutations (SNP1 and SNP3), two were synonymous changes (SNP2 and SNP6) and three SNPs loci (SNP4, SNP5, and SNP7) were located at several introns. The missense mutation loci (SNP1 and SNP3) only had one kind of genotype of each locus, meaning that the mutation frequency was very low. The missense mutation with amino acid change could affect protein structure, resulting in loss of normal function, which might cause embryonic lethality. We detected haplotypes structure and found the common haplotype (hap1) had a relatively high frequency in two breeds, for the haplotype was present in the population for a long time. The haplotypes of highest frequencies in HNBG and XNSN dairy goat were different, probably caused by variety distinctiveness.
Association testing revealed that the SNP2, SNP5, SNP6 and SNP7 loci were also found to significantly associate with growth-related traits in goats. Among them, although SNP2 and SNP6 were synonymous mutations, they might affect transcriptional efficiency for codon preference and stability of mRNA (Chamary et al., 2005). Many studies have shown that no change of amino acid sequence could still affect gene performance, for example, two synonymous SNPs of bovine NUCB2 gene were significantly associated with growth traits (Li et al., 2010). Although SNP5 and SNP7 were intronic mutations, they also might affect alternatively spliced transcripts of mRNA or transcription factor binding, thus affecting phenotype. A famous example of intronic mutation was located at intron 3 of the porcine IGF2 gene. This mutation lead to a significant effect in skeletal muscle (Van et al., 2003). Besides, the combined genotypes of SNP6 and SNP7 in Xinong Saanen dairy goats was significantly linked to growth related traits. Therefore, this association data reflected that these nucleotide variations within ATBF1 gene produced significant effects on growth related traits,  suggesting that this gene can be used as a marker gene in improving goat growth traits. Briefly, seven novel SNPs mutations were firstly found, and four of them significantly affected goat growth related traits, which extends the known genetic variations spectrum of goat ATBF1 gene and is a benefit towards implementing MAS in genetics and breeding of goats.

CONFLICT OF INTEREST
We certify that there is no conflict of interest with any financial organization regarding the material discussed in the manuscript.