Expression profiles of the MXD3 gene and association of sequence variants with growth traits in Xianan and Qinchuan cattle

Abstract Max dimerization protein 3 (MXD3) belongs to the MYC superfamily of basic helix‐loop‐helix leucine zipper transcription factors, and MXD3‐MAX heterodimers can bind to promoters of target genes to modulate their expression. The aim of this study was to determine the MXD3 mRNA expression levels in various cattle tissues comprising heart, liver, spleen, lung, kidney, Longissimus dorsi muscle and subcutaneous fat in Chinese Qinchuan and Xianan cattle breeds. The RT‐qPCR data showed that the MXD3 gene was variably expressed between all tissues and at levels that were significantly different between two breeds (p < .05). We used the polymerase chain reaction‐restriction fragment length polymorphism (PCR‐RFLP) method to investigate the possible association between single‐nucleotide polymorphisms (SNP) within the MXD3 gene and five different growth traits in cattle. We found two intronic SNPs (g.2694 C>T and g.3801 T>C) and one SNP in 3′untranslated region (3′UTR) (g.6263 G>A) of MXD3 gene. Association analysis revealed strong associations between pairwise and triple SNP combinations and the growth traits. Based on these results, we suggest that MXD3 polymorphisms could be useful as molecular markers in the Chinese beef cattle breeding program.


| INTRODUC TI ON
The application of molecular genetics in animal husbandry has many important advantages (Khodabakhshzadeh et al., 2016;Zamani, Akhondi, & Mohammadabadi, 2015). One such significant advantage is the genotyping of individuals for specific genetic loci known to be associated with phenotypic traits of relevance to cattle farming and breeding (Khodabakhshzadeh et al., 2016;Mohammadreza Esfandyarpoo, & Mousapour, 2017). Often all the genes that affect a polygenic trait are not precisely known, although a number of candidate genes with major effects have been recognized . In the candidate gene approach, the process of identifying such genes responsible for a polygenic trait variation includes the selection of candidate genes based on the relationship between physiological or biochemical processes involved in the expression of the phenotype, and subsequent testing of the selected genes as putative quantitative trait loci (QTL) (Mousavizadeh et al., 2009;Ruzina et al., 2010). The bovine genome is densely covered by single-nucleotide polymorphisms (SNP) markers, which facilitates the search for genes with significant effects on quantitative trait variation (Javanmard et al., 2008). Integrated approaches in terms of management and genetic improvement are of crucial importance for enhancement of production (Mohammadabadi & Sattayimokhtari, 2013;Soufy et al., 2009). Furthermore, economical and biological efficiency of production enterprises generally improves by increasing productivity and reproductive performance of animals (Mohammadabadi & Sattayimokhtari, 2013;Taghi Vajed Ebrahimi, Mohammadabadi, & Esmailizadeh, 2017;Zamani et al., 2015). Growth phenotypes such as body weight and size of cattle are often used as selection criteria because of their association with meat production . Complex traits are usually highly polygenic; a recent meta-analysis for cattle stature identified 163 genome-wide significant loci (Bouwman et al., 2018). Cattle selection programmes based on the molecular genetic information is a powerful and effective strategy to enhance economic quantitative traits. Thus, marker-assisted selection (MAS) at the DNA level significantly increases selection accuracy of purposeful phenotypes and shortens the generation intervals (Bouquet & Juga, 2013). As one of the largest beef producer in the world, China began directional selection in beef cattle for melioration of meat performance after 1980s to satisfy an increasing consumption of beef (Waldron, Jimin, Huijie, Xiaoxia, & Tre, 2013). However, Chinese beef production still relies on imports from abroad as well as on collaborations between breeders to improve Chinese beef production (Waldron et al., 2013). The need to maintain and improve local genetic resources has been recognized as a priority, at the world level. For example, biodiversity studies depicting a deep picture of the genetic variability of the available sheep breeds provide favourable opportunities for both genetic conservation programmes as well as for enhancing production efficiency by means of controlled and well-designed crossbreeding systems exploiting breed diversities, heterosis and breed complementarity (Taghi Vajed Ebrahimi et al., 2017). Genetic diversity in indigenous breeds is a major concern considering the necessity of preserving what may be a precious and irreplaceable richness, regarding future productive demands (Khodabakhshzadeh et al., 2016). Conservation should be based on a deep knowledge of the genetic resources of the specific breed (Mohammadreza et al., 2017;Zamani et al., 2015). Therefore, it is important to characterize genetically indigenous breeds. Genes affecting polygenic traits and characterizing milk or meat production are difficult to identify (Shojaei et al., 2011;Soufy et al., 2009). The maintenance of genetic diversity in livestock species requires the adequate implementation of conservation priorities and sustainable management programmes, which should be based on comprehensive information regarding the structure of the populations, including sources of genetic variability among and within breeds (Mousavizadeh et al., 2009;Ruzina et al., 2010). Genetic diversity is an essential component for population survival, evolution, genetic improvement and adaptation to changing environmental conditions (Taghi Vajed Ebrahimi et al., 2017), and molecular methods based on molecular markers, such as RAPD, RFLP and microsatellites, are useful tools to study the underlying genetics Mousavizadeh et al., 2009;Taghi Vajed Ebrahimi et al., 2017).
The basic helix-loop-helix leucine zipper transcriptional regulators, which belongs to the MYC-MAX-MAD network, are central players in the control of cell-cycle progression, proliferation, apoptosis and transformation (Grandori, Cowley, James, & Eisenman, 2000). The MAX-MYC heterodimer is a sequence-specific transcriptional activator, whereas the MAX-MAD complex acts as a sequence-specific transcriptional repressor (Lüscher & Vervoorts, 2012). The Max dimerization protein 3 (MXD3), member of MAD family, has been considered a potential target for therapeutic treatment of brain and central nervous system cancers due to its role in cellular proliferation and tumorigenesis (Barisone et al., 2012). MXD3 expression is significantly upregulated in visceral adipose tissues in human obese adults as well as in a zebrafish model of diet-induced obesity in which downregulation of MXD3 expression suppressed the formation of visceral adiposity (Shimada et al., 2014). In addition, the expression of MXD3 was three-fold reduced in adult skeletal muscle tissues compared with the fetal period in Qinchuan (QC) cattle. MXD3 gene was also enriched in the Gene Ontology (GO) terms of DNA binding, protein binding, negative regulation of transcription, DNA-dependent and protein dimerization activity (Li et al., 2017). Therefore, the objective of this study was to identify SNPs within MXD3 and then to analyze their associations with the growth traits of Chinese cattle. We studied two different breeds, the QC breed, which is one of the most important Chinese beef cattle breeds, yet exhibiting worse growth performance and carcass traits than imported European cattle breeds (Xie, Meng, Cui, & Ren, 2012;Xie, Meng, Ren, Shi, & Zhou, 2012), as well as the Xianan (XN) cattle which is a crossbreed between Charolais cattle and Nanyang cattle.
To the best our knowledge, our study provided the first association analysis of MXD3 sequence variations with growth traits and expression profiles of MXD3 in seven tissues under the best normalized reference genes in cattle, which may improve the understanding of the molecular basis and the application of MAS for beef cattle breeding in China.

| Cattle and data collection
A total of 499 samples of QC cattle (2-year-old, n = 141; 3.5-yearold, n = 181) and XN cattle (n = 177, 2-year-old) were collected from two farms: the farm of Fufeng country in Shaanxi and the farm of Nanyang city in Henan provinces of China, respectively. All of the individuals were females. Five growth traits for association analysis were recorded at the age of 2 years old, females with no pregnancy both in QC (n = 141) and XN (n = 170) cattle following Gilbert's method (Gilbert, Bailey, & Shannon, 1993).
The growth traits were body weight (BW, kg), body length (BL, cm), body height (BH, cm), chest circumference (ChC, cm) and hip cross height (HCH, cm). The genomic DNA from each cattle was purified from 2% heparin-treated jugular blood samples and diluted to 50 ng/µl and subsequently stored at −20°C for further usage following the standard procedures (Sambrook & Russell 2001).

| Detection and identification of the variants within MXD3 gene
Using the bovine genome sequences in GenBank (Accession No. AC_000164), seven pairs of primer were designed by Primer v5.0 software (PREMIER Biosoft International) (Table S1). Genomic DNA pools of 40 randomly selected individuals were mixed gently from the two breeds, amplified and sequenced to identify the polymorphisms within MXD3. After comparing the MXD3 sequence to National Center for Biotechnology Information (NCBI), three mutations within bovine MXD3 were found. Accordingly, pairs of primers were designed to genotype the mutations based on restriction fragment length polymorphism (PCR-RFLP) method using restriction enzyme Taq I, Hha I and Pvu II (Thermo Fisher Scientific) (Table S2).

| RNA isolation and qRT-PCR analysis
Seven tissues (heart, liver, spleen, lung, kidney, longissimus dorsi muscle and subcutaneous fat) were collected from three QC and three XN cattle when they were slaughtered at the cattle plant.
After washing with phosphate buffered saline (PBS), samples were put immediately into liquid nitrogen and stored at −70°C for the subsequent steps; each tissues had three replicates. Next, Trizol reagent (Takara Co., Ltd) was used to extract the RNA. The RNA purity and concentration were determined using a NanoDrop 2000 spectrophotometer and we randomly chose three samples to check the purified RNA by 0.8% agarose gel electrophoresis ( Figure S1).
Next, the StarScript II One-step RT-PCR Kit (Takara Co., Ltd) with 2 µg RNA as the template was used for cDNA synthesis. Accurate normalization is a prerequisite for analysing target gene expression under various experimental conditions and samples. The RefFinder software (https://www.heart cure.com.au/reffi nder/?type=refer ence) (Xie, Xiao, Chen, Xu, & Zhang, 2012) can conveniently and efficiently assess four software applications termed geNorm (Excelbased) (Vandesompele et al., 2002), BestKeeper (Excel-based) (Pfaffl, Tichopad, Prgomet, & Neuvians, 2004) and NormFinder (Excel-based) (Andersen, Jensen, & Ørntoft, 2004). The standardization algorithms termed deltaCt method (Silver, Best, Jiang, & Thein, 2006) was also assessed in RefFinder. Herein, we employed RefFinder to assess the numbers and expression stability of reference genes. GAPDH (glyceraldehyde-3-phosphate dehydrogenase), ACTB (beta actin), RPL19 (ribosomal protein 19) and EMD (emerin) were selected as the most appropriate reference genes and the relevant primers used in qRT-PCR were shown in Table S1. The melt curve was added automatically to verify amplification efficiency and no template control reactions (Bio-Rad). The 0.005, 0.05, 0.5, 5 and 50 ng cDNA were used to test the amplification efficiency. The correlation coefficient (r 2 ) was between 0.9312 and 0.9992, whereas the slope was −1.922 to −2.925 ( Figure S2). The gene expression levels were calculated based on geNorm and 2 −ΔΔCT method and we performed two-tailed Student's t test to compare the differential expressions in various tissues from two adult cattle breeds.
Multiple comparisons were performed for SNPs and their interactions using least significance difference (LSD) method based on R package agricolae (version 1.3-1).

| Identification and genetic characteristic of genetic variants within MXD3 gene
The bovine MXD3 is located at position ~40.1 Mb on Bos taurus autosome (BTA) 7 and consists of 6 introns and 7 exons. In this study, we identified three SNPs, i.e. g.2694 C>T (dbSNP Accession No.
In general, the predominant genotypes were heterozygous with frequencies above 0.500 except for the homozygous CC genotype at position g.2694 C>T, which was the most frequent in XN cattle. The frequencies of the C alleles of the two intron-SNPs and the C and G allele in 3′UTR-SNP were greater than 0.5 in both breeds. According

| LD and haplotype information analysis
Next, we calculated the LD parameters r 2 and D′ for all pairs of the three SNP loci (Figure 2

| Association analysis between single SNP and growth traits
An association analysis was performed to provide novel information  (Table 3).

| Association analysis between combined SNPs and growth traits
To improve the reliability of the association results, we also conducted pairwise SNPs analyses (

| Detection of expression levels of MXD3 in seven different tissues
To assess and compare the expression level of MXD3 in different tissues, we first determine which reference genes would be most suitable for normalization. As shown in Figure 3a, geNorm ranked RPL19 and EMD as the best combination for gene expression normalization in various cattle tissues. Delta CT, BestKeeper and Normfinder suggested a single reference gene EMD (Figure 3a and Figure S3). Therefore, we used two approaches to normalize the MXD3 expression level based on our results: (a) A geometric mean of the two most stably expressed genes (RPL19 and EMD) combined as suggested by Vandesompele et al. (2002), (b) Using a single best gene EMD as suggested by the 2 −ΔΔCT method. The data shown in Figure 3 demonstrate that the MXD3 expression levels as measured by the two different methods were highly consistent. Furthermore, we observed that the MXD3 gene showed the highest expression level in spleen, followed by muscle and fat in QC cattle, whereas, in XN cattle, the highest expression was found in the liver followed by heart and muscle (Figure 3b-e). It is also noteworthy that the MXD3 was differently expressed in the seven tissues between two breeds based on t test (p < .001 or p < .05) (Figure 3f).

| D ISCUSS I ON
Cattle have the ability to convert low-quality forage into highquality food such as milk and beef, which are important sources of human nutrition. However, Chinese indigenous cattle with inferior productivity indexes cannot satisfy the increasing demand for beef production in China. Over recent years, researchers have investigated the genetic basis of many complex phenotypes and many polymorphisms in various functional genes have been reported associated with cattle economic important traits, suggesting they may be useful as markers for the genetic improvement of cattle (Bouwman et al., 2018). In this study, we investigated the TA B L E 4 Association analysis of pair-wise SNPs with growth traits in the analysed sample   Daz, 2014;Ngo et al., 2019;Satake et al., 2014Satake et al., , 2016 We identified two intronic SNPs (g.2694 C>T and g.3801 T>C) and one SNP (g.6263 G>A) in the 3′UTR of MXD3, which we suggest could be used to promote the genetic improvement of Chinese cattle. Thus, a major result of this study is the observations of strong associations between pairwise and triple SNP combinations and the performance traits measured in the two breeds. Another noticeable example of non-coding SNPs with phenotypic impact is a mutation 4251 nt (C>T) in intron 1 of the growth hormone-releasing hormone (GHRH) gene, which was significantly associated with body weight in QC cattle . Non-coding SNPs may have effects on mRNA metabolism by transcriptional enhancement or repression, or by influencing nucleosome-positioning elements in the gene or via an effect on the assembly of spliceosome components (Le Hir, Nott, & Moore, 2003). Interestingly, the TargetScan algorithm predicts that the 3′UTR-SNP (g.6263 G>A) in MXD3 may decrease its interaction with bta-miR-22-3p. This agrees well with the previous observation that bta-miR-22-3p is associated with cattle development based on genome-wide profiles on muscle tissue between fetal and adult QC cattle (Agarwal, Bell, Nam, & Bartel, 2015;Huang et al., 2014;Sun et al., 2015). However, further studies are needed to confirm the interaction between miR-22-3p and 3′-SNP in MXD3 and to better understand how this affects complex phenotypes.
The detection of gene variants associated with economically important traits has provided insights into the genetic architecture of complex traits and diseases in cattle (Bouwman et al., 2018;Suravajhala, Kogelman, & Kadarmideen, 2016). Recently, researchers have paid increased attention to the integration of muti-omic data, which will provide a system-level understanding of the biology of complex traits and be instrumental in improving sustainable breeding of productive and healthy animals (Suravajhala et al., 2016).
Thus, future use of present and emerging omics-technologies will be applied to collaboration with breeders and farmers.

ACK N OWLED G EM ENTS
This study was supported by the National Natural Science

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest.

AUTH O R CO NTR I B UTI O N
DH performed the experiments, analysed the data, and DH and BT drafted the manuscript. JSB and XW improved the manuscript. SJP, XYL and YZH developed and collected the samples. HC and XW conceived and designed the study, reviewed the study and supervised the research.

E TH I C A L A PPROVA L
All animal procedures were carried out according to protocols approved by the College of Animal Science and Technology, Northwest A&F University, China.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.