Copy Number Variation of the CADM2 Gene and Its Association with Growth Traits in Yak

Simple Summary Cell adhesion molecule 2 (CADM2), also known as synaptic cell adhesion molecule 2 (SYNCAM2), is the mediator of synaptic signals enriched in the brain. Overlaps between copy number variation (CNV) regions in CADM2 and quantitative trait loci (QTL) related to body weight have been clarified in a previous study. In this study, two loci were amplified in the CADM2 gene (CNV1: 235,915 bp, exon 1 and partial intron 1; CNV2: 60,430 bp, intron 9) to explore the relationship between CNV types in the CADM2 gene and growth traits in 350 Ashidan yaks. Association analysis illustrated that no significant effect was found on growth traits in CNV1. However, the CNV2 mutation had a significant effect on body weight at the sixth month (p < 0.05). Individuals with the gain-type copy number variation CNV2 were significantly superior to those with loss- or normal-type in terms of body weight (p < 0.05). In summary, this study confirmed that CADM2-CNVs affect growth traits in yaks, and may be candidate genes for successful yak breeding and genetics projects. Abstract Copy number variation (CNV) is currently accepted as a common source of genetic variation. It is reported that CNVs may influence the resistance to disease and complex economic traits, such as residual feed intake, muscle formation, and fat deposition in livestock. Cell adhesion molecule 2 (CADM2) is expressed widely in the brain and adipose tissue and can regulate body weight through the central nervous system. Growth traits are important economic traits for animal selection. In this study, we aimed to explore the effect of CADM2 gene copy number variants on yak growth traits. Here, two CNVs in the CADM2 gene were investigated using the quantitative polymerase chain reaction (qPCR), and the association of the CNVs with growth traits in yak was analyzed using statistical methods by SPSS software. Differences were considered significant if the p value was < 0.05. Statistical analysis indicated significant association of CADM2-CNV2 with the body weight of the Chinese Ashidan yak. A significant effect of CNV2 (p < 0.05) was found on body weight at 6 months. In CNV2, the gain-type copy number variation exhibited greater performance than the other variants, with greater body weight observed at 6 months (p < 0.05). To the best of our knowledge, this is the first attempt to investigate the function of CADM2-CNVs and their association with growth traits in animals. This may be a useful candidate marker in marker-assisted selection of yaks.


Introduction
With large duplications and deletions in gene copy numbers detected in humans and other primates, it is not unreasonable to expect that copy number variation could be a significant source of genetic variation in other animals [1]. In previous studies, many molecular markers have been assessed for improving economic traits in animal breeding, including restriction fragment length

Sample Collection and Growth Traits
Three hundred and fifty unrelated female Ashidan yaks that had been raised with the same levels of nutrition were randomly selected from the Datong yak farm (Qinghai Province, China). None of the animals involved in this study were unhealthy. Phenotypic data (i.e., body weight, withers height, body length, chest girth) were measured in the selected Ashidan yaks at three stages of growth (i.e., at 6 months, 1 year and 18 months of age) for subsequent analysis. Phenotypic measurements were strictly in line with standard methods proposed by Gilbert et al. [34].

Genomic DNA Extraction and Copy Number Identification
Genomic DNA was extracted using an EasyPure Blood Genomic DNA kit (TransGen Biotech Co., Ltd., Beijing, China) and its concentration and quality measured using a Nanodrop 2000 spectrophotometer (ThermoFisher Scientific Inc., Waltham, MA, USA). Quantified DNA samples were stored at −20 • C.
The copy number variation regions CADM2-CNV1 and CADM2-CNV2 are located at 33,143,512-33,379,427 bp and 33,569,316-33,629,746 bp from the reference genome sequence NC_037328.1, respectively ( Figure 1). Animals 2019, 9,1008 3 of 11 LIHPS-CAAS-2017-115). Blood sample collection and body measurements were performed in strict accordance with the guide for the Care and Use of Laboratory Animals.

Sample Collection and Growth Traits
Three hundred and fifty unrelated female Ashidan yaks that had been raised with the same levels of nutrition were randomly selected from the Datong yak farm (Qinghai Province, China). None of the animals involved in this study were unhealthy. Phenotypic data (i.e., body weight, withers height, body length, chest girth) were measured in the selected Ashidan yaks at three stages of growth (i.e., at 6 months, 1 year and 18 months of age) for subsequent analysis. Phenotypic measurements were strictly in line with standard methods proposed by Gilbert et al. [34].

Genomic DNA Extraction and Copy Number Identification
Genomic DNA was extracted using an EasyPure Blood Genomic DNA kit (TransGen Biotech Co., Ltd., Beijing, China) and its concentration and quality measured using a Nanodrop 2000 spectrophotometer (ThermoFisher Scientific Inc., Waltham, MA, USA). Quantified DNA samples were stored at −20 °C.

Primer Design
The National Center for Biotechnology Information (NCBI) primer-BLAST web tool was utilized to design the primers for the analysis of CADM2-CNVs (Table 1). A polymerase chain reaction (PCR) was utilized to verify the quality and optimum temperature of primers. Products were analyzed on 1% agarose gel electrophoresis.

Primer Design
The National Center for Biotechnology Information (NCBI) primer-BLAST web tool was utilized to design the primers for the analysis of CADM2-CNVs (Table 1). A polymerase chain reaction (PCR) was utilized to verify the quality and optimum temperature of primers. Products were analyzed on 1% agarose gel electrophoresis.

Quantitative PCR Analysis
A LightCyler ® 96 quantitative polymerase chain reaction (qPCR) system (Roche, Basel, Switzerland) with SYBR Green dye was used in this study because of its convenience and accuracy. The bovine basic transcription factor 3 (BTF3) gene was chosen as the diploid internal reference gene [35]. The total volume of the reaction mixture was 20 µL, comprising 40 ng genomic DNA with 10 µL of SYBR Premix Ex Taq II (TaKaRa, Dalian, China), 2 µL of amplification primers and 7 µL of ddH 2 O. The q-PCR procedure involved a 30 s preincubation at 95 • C then a two-step amplification with 45 cycles of 95 • C for 5 s followed by 62 • C for 30 s. After completion of the cycling, the reaction was terminated by incubation at 95 • C for 5 s then 65 • C for 60 s then 95 • C continuously. All experiments were conducted in triplicates, and the results represented as mean values ± standard deviation (SD).

Statistical Analysis
Copy number values of the yak CADM2 gene were determined using the formula: 2 × 2 −∆∆ct [36]. Cycle threshold (Ct) values for the target gene were normalized using the formula: ∆Ct = Ct target − Ct re f erence , with ∆Ct of the test samples normalized using the formula: ∆∆Ct = ∆Ct test − ∆Ct control . As the precondition of the analysis of variance, the homogeneity of variance test and the normality test of each trait in different copy number variation (CNV) regions were done by Bartlett test and Shapiro-Wilk test before association analysis. One-way analysis of variance (ANOVA) and a nonparametric test were used to analyze the association of CADM2 gene copy number variant types with growth traits in the Ashidan yak. The following model was used to analyze the CNV effects on growth traits: Y ij = µ + CNV ij + e ij , where i represents the ith region, j the jth CNV type, Y ij the observed growth trait, µ the overall mean of each trait, with CNV ij representing the effect of ith CNV region and jth CNV type of the CADM2 gene. e ij is a random error. Assessment of differences between means was performed by one-way ANOVA with a post hoc least significant digit (LSD) multiple comparison test. Differences in the distribution of CNV types for different CNV regions were analyzed using a chi-square test in terms of the frequency of CNV types [37]. Based on the results of association analysis, a Kendall Tau correlation test was carried out. Other than these, the combination and interaction effect of CNV1 and CNV2 also detected by two-way analysis of variance (ANOVA).
All the statistical analyses were carried out using IBM SPSS Statistics software (Version 23, IBM, New York, NY, USA). Significance level was set up at p < 0.05 in this experiment.

Quantitative PCR and the Distribution of CADM2-CNVs
Specific primers for the two CNVs are displayed in Table 1. Cycle threshold (Ct) values were used subsequently for quantitative analysis. The amplification curves and melt peaks ( Figure 2) were used to determine primer specificity. In addition, the distribution frequency of different CNV types in CNV1 and CNV2 were measured ( Figure 3 and Table 2).

Quantitative PCR Analysis
A LightCyler ® 96 quantitative polymerase chain reaction (qPCR) system (Roche, Basel, Switzerland) with SYBR Green dye was used in this study because of its convenience and accuracy. The bovine basic transcription factor 3 (BTF3) gene was chosen as the diploid internal reference gene [35]. The total volume of the reaction mixture was 20 μL, comprising 40 ng genomic DNA with 10 μL of SYBR Premix Ex Taq II (TaKaRa, Dalian, China), 2 μL of amplification primers and 7 μL of ddH2O. The q-PCR procedure involved a 30 s preincubation at 95 °C then a two-step amplification with 45 cycles of 95 °C for 5 s followed by 62 °C for 30 s. After completion of the cycling, the reaction was terminated by incubation at 95 °C for 5 s then 65 °C for 60 s then 95 °C continuously. All experiments were conducted in triplicates, and the results represented as mean values ± standard deviation (SD).

Statistical Analysis
Copy number values of the yak CADM2 gene were determined using the formula: 2 2 ∆∆ [36]. Cycle threshold (Ct) values for the target gene were normalized using the formula: ∆ , with ∆ of the test samples normalized using the formula: ∆∆ ∆ ∆ . As the precondition of the analysis of variance, the homogeneity of variance test and the normality test of each trait in different copy number variation (CNV) regions were done by Bartlett test and Shapiro-Wilk test before association analysis. One-way analysis of variance (ANOVA) and a nonparametric test were used to analyze the association of CADM2 gene copy number variant types with growth traits in the Ashidan yak. The following model was used to analyze the CNV effects on growth traits: , where i represents the ith region, j the jth CNV type, the observed growth trait, the overall mean of each trait, with representing the effect of ith CNV region and jth CNV type of the CADM2 gene.
is a random error. Assessment of differences between means was performed by one-way ANOVA with a post hoc least significant digit (LSD) multiple comparison test. Differences in the distribution of CNV types for different CNV regions were analyzed using a chi-square test in terms of the frequency of CNV types [37]. Based on the results of association analysis, a Kendall Tau correlation test was carried out. Other than these, the combination and interaction effect of CNV1 and CNV2 also detected by two-way analysis of variance (ANOVA).
All the statistical analyses were carried out using IBM SPSS Statistics software (Version 23, IBM, New York, NY, USA). Significance level was set up at p < 0.05 in this experiment.

Quantitative PCR and the Distribution of CADM2-CNVs
Specific primers for the two CNVs are displayed in Table 1. Cycle threshold (Ct) values were used subsequently for quantitative analysis. The amplification curves and melt peaks ( Figure 2) were used to determine primer specificity. In addition, the distribution frequency of different CNV types in CNV1 and CNV2 were measured ( Figure 3 and Table 2).
The majority of individual animals had the loss-type variant in CNV1, the proportion exceeding 50%. However, in CNV2, the largest number of individuals had the normal variant, followed by the loss-type. The gain-type variant was similarly distributed within CNV1 and CNV2. The chi-square test indicated a significant difference in the distribution of CNV types between CNV1 and CNV2 (p < 0.01), suggesting that the CNV type is specific for different loci (Table 3).      Chi-square values (χ ) for differences between CNVs. ** Represent statistically significant differences at a level of p < 0.01.

Association Analysis
Association analysis indicated a significant correlation between copy number type and phenotypic measurements in the Ashidan yak population. Bartlett test showed the homogeneity of variance of each trait within each locus (Supplementary Table S1). And the Shapiro-Wilk test showed the normality of each trait in different CNV regions (Supplementary Table S2). One-way ANOVA indicated that the CNV2 mutation had a significant effect on body weight (p < 0.05) in 6th month yaks (Table 5). Multiple comparisons were conducted based on the ANOVA described above. Interestingly, the gain-type copy number variant of CADM2-CNV2 exhibited clear phenotypic superiority in comparison with the loss and normal types, with greater body weight values in 6th month Ashidan yaks (p < 0.05). Furthermore, our correlation test verified the significant correlation between CNV2 and body weight in 6th month (p < 0.01) ( Table 6).
No significant difference was observed in the traits at 12 and 18 months. Association analysis of the CADM2 gene CNVs with growth traits is presented in Table -5.
Moreover, a two-way analysis of variance (ANOVA) was also carried out to test the joint effect of CADM2-CNV1 and CADM2-CNV2. The results revealed a significant association in 18th month withers height both in CNV1 (p < 0.01) and CNV2(p < 0.05) ( Table 7). But no significant association was found in their combination.  The majority of individual animals had the loss-type variant in CNV1, the proportion exceeding 50%. However, in CNV2, the largest number of individuals had the normal variant, followed by the loss-type. The gain-type variant was similarly distributed within CNV1 and CNV2. The chi-square test indicated a significant difference in the distribution of CNV types between CNV1 and CNV2 (p < 0.01), suggesting that the CNV type is specific for different loci (Table 3). Chi-square values (χ 2 ) for differences between CNVs. ** Represent statistically significant differences at a level of p < 0.01.

Association Analysis
Association analysis indicated a significant correlation between copy number type and phenotypic measurements in the Ashidan yak population. Bartlett test showed the homogeneity of variance of each trait within each locus (Supplementary Table S1). And the Shapiro-Wilk test showed the normality of each trait in different CNV regions (Supplementary Table S2). One-way ANOVA indicated that the CNV2 mutation had a significant effect on body weight (p < 0.05) in 6th month yaks (Tables 4  and 5). Multiple comparisons were conducted based on the ANOVA described above. Interestingly, the gain-type copy number variant of CADM2-CNV2 exhibited clear phenotypic superiority in comparison with the loss and normal types, with greater body weight values in 6th month Ashidan yaks (p < 0.05). Furthermore, our correlation test verified the significant correlation between CNV2 and body weight in 6th month (p < 0.01) ( Table 6).  Association of CNVR30 of the CADM2 gene with yak growth traits. a,b,c Represent statistically significant differences at a level of p < 0.05; * p < 0.05. Table 6. Correlation test of body weight in 6 months and CNV2.
Kendall Tau correlation coefficient 0.261 Significance 0.007 ** ** Represent statistically significant differences at a level of p < 0.01.
No significant difference was observed in the traits at 12 and 18 months. Association analysis of the CADM2 gene CNVs with growth traits is presented in Tables 4 and 5.
Moreover, a two-way analysis of variance (ANOVA) was also carried out to test the joint effect of CADM2-CNV1 and CADM2-CNV2. The results revealed a significant association in 18th month withers height both in CNV1 (p < 0.01) and CNV2(p < 0.05) ( Table 7). But no significant association was found in their combination.

Discussion
Researchers have previously given considerable attention to the analysis of molecular markers associated with economic traits in a number of species. For example, genetic variation in the ovine PROP1 gene was identified in 670 New Zealand Romney sheep, which were related to traits such as tailing weight and growth rate [38]. One 43-bp indel polymorphism in HS6ST3 was associated with shank length, chest depth and other growth traits in 860 chickens [39]. SNPs within the coding region of the KDM4D gene were explored and linked with testis morphology traits of male pigs [40]. However, copy number variance gives rise to more substantial mutations and more extensive genetic effects than other molecular markers [41]. It has been established that the total number of nucleotides affected by copy number variants is larger than the total number of SNPs [42]. Recently, CNV maps in cattle [35], sheep [43] and pigs [44] have been constructed, indicating that structural polymorphisms within them constitute an important proportion of their genomes. Furthermore, CNVs have been associated with various growth traits, such as body length in cattle [45] and pigs [46] and pigmentation traits in goats [47] and horses [48], in addition to reduced immunological resistance to diseases [49]. Consistent with these results, it is widely believed that this form of genetic variation in livestock is of great importance to phenotypic traits.
Quantitative polymerase chain reaction (q-PCR) is widely used, especially for the measurement of copy number variation because of its accuracy and simplicity. To investigate the probable effects of CADM2-CNVs in the yak growth process, we conducted an association study between 12 phenotypic traits and different CNV types using a general linear model for the first time. Significant association was observed between 6th body weight and CNV2. The statistics demonstrate that weaning weights are significantly correlated with CNV2 type. Interestingly, this is in agreement with the conclusion of previous studies [28]. This phenomenon may be due to the function of the CADM2 gene, which plays an important role in maintaining the synaptic circuitry of the central nervous system [50], and can mediate leptin sensitivity, thermogenesis and energy metabolism [51]. Notably, the gain-types in CNV2 caught our attention due to the outstanding body weight in 6th month animals. Here, no significant association between CNV1 and CNV2 and the body weight in 12 month-olds was observed, while in cattle a QTL harboring CADM2 gene was thought to be associated with the yearling weight. This was most likely cause by the following points. First of all, a quantitative trait like body weight is regulated by multiple genes with small effect. Though a 350 yak sample size is an appropriate sample to examine the variance, the marker's effect is too insufficient to be detected. A larger population is needed to be studied in the future. Secondly, it is probably due to the yak's special physiological regulation mechanism due to its particularity harsh living conditions. Unlike the cattle industry, yak has to struggle to survive in a cold season lasting for six months after weaning, which is extremely cold along with severe forage shortage.
A 6 months hay period they will endure after weaning, which is different from cattle. This phenomenon may lead to a unique physiological regulation of growth in yaks. Here, we did not find any associations between CNV types and phenotype at 12 and 18 months.
To explain why this CNV has such an uncommon influence on phenotype, firstly the CADM2 gene had been reported to overlap a QTL region that is correlated with body weight (weaning and yearling) [28]. Secondly, body weight and energy metabolism are maintained by the central nervous system and previous studies have identified that CADM2 is expressed more abundantly in different regions of the brain than in any other tissue [52]. Furthermore, plenty of studies have reported that CADM2 can significantly improve body weight and decrease leptin sensitivity via the brain and adipose tissue [32,53]. Moreover, genome-wide association studies in humans have demonstrated clear evidence that the CADM2 gene is strongly associated with the body mass index through the encoding of membrane proteins that mediate synaptic assembly [31,52,54]. Because of the importance of both the CADM2 gene and copy number variance in animals, we propose that CADM2-CNVs can be used as indispensable molecular markers of genetic improvement of the yak. To the best of our knowledge, copy number variance of CADM2 has not been previously investigated in livestocks such as cattle, sheep, pigs, etc. Therefore, we speculate that the CADM2 gene might play a crucial role in growth traits in the yak.
Growth traits are an economically important factor in animal husbandry that can improve production performance by selective breeding [55]. Consistent with this research, numerous studies have illustrated the association between gene copy number type and growth traits. Shi et al. demonstrated that the CNV of the LEPR gene exhibited significant correlation with growth traits in Chinese cattle [56]. Jiang et al. clarified that CNV of the sheep SHE gene was greatly linked with economic traits, such as body length and chest width [57]. Lin et al. suggested that the CNV of SOX6 is responsible for muscle development in chickens [58]. We propose that these molecular markers in various livestock offer the theoretical promise of animal selection.
The yak is a specialized breed famous for its adaptation to life on the plateau and hypoxia. Distinctive regulatory mechanisms require exploration, particularly their growth mechanism. In the present study, the two CNVs of the CADM2 gene in the Ashidan yak have been characterized and validated. The results demonstrate the effect of the CADM2 CNVs on phenotypic traits in different CNV types. From this evidence, we have established that CADM2 CNVs can act as molecular markers in Ashidan yaks for early selection.

Conclusions
In conclusion, this study described the distribution of CADM2 gene copy number variant types in 350 Ashidan yaks. CADM2-CNV2 is significantly associated with body weight in 6 months in the Ashidan yak. Further studies, such as the mechanism of physiological regulation with the CADM2 gene in yaks need to be carried out. Also a larger population is required to validate the association of this CADM2 gene with growth traits in the yak.
Supplementary Materials: The following are available online at http://www.mdpi.com/2076-2615/9/12/1008/s1, Table S1: The homogeneity of variance test of each trait, Table S2: The test of normality of each trait in different CNV regions.