Genome-Wide Identication of Cyclin Dependent Kinase (CDK) Family Genes Inuencing Adipocyte Differentiation in Cattle

Background: The cyclin dependent kinases (CDKs) are protein kinases regulating important cellular processes such as cell cycle and transcription. A variety of studies have shown that many CDK genes also played a critical role during adipogenic differentiation. However, there is a lack of systematic research on the CDK gene family regulating bovine adipocyte differentiation.Therefore, this study aimed to characterize CDK family genes in bovine and study the expression pattern during adipocyte differentiation. Results: We performed a genome-wide analysis and identied 25, 25, 22, 21, 22, 24, 22 and 24 CDK genes in Bos taurus, Bos indicus, Hybrid-Bos taurus, Hybrid bos indicus, Bos grunniens, Bos mutus, Bison and Bubalus bubalis, respectively. All the CDK genes classied into 8 subfamilies through phylogenetic analysis. Chromosome localization displayed 25 bovine CDK genes distributed on 16 chromosomes. Collinearity analysis revealed that CDK family genes of Bos taurus were extensively homologous with Bos indicus, Hybrid-Bos taurus, Hybrid bos indicus, Bos grunniens and Bubalus bubalis. Tanscriptome analysis showed that several of the CDK family genes had relatively high expression levels in preadipocytes compared with differentiated adipocytes, which is generally similar to qPCR, indicating that it could have a signicant function in the growth of the emerging lipid droplets. Conclusion: We performed a comprehensive analysis for the CDK family genes including identication, phylogenetic classication, structural characterization, chromosomal distribution, collinearity analysis and expression prole analysis by tanscriptome sequencing and qPCR. The results provide a basis for further study to determine the roles of CDK family genes in regulating adipocyte differentiation, which is benecial for beef quality improvement. This study conducted a comprehensive genome-wide analysis of CDK family genes in bovidae. A total of 185 CDK genes were identied and grouped into eight distinct clades. Collinearity analysis revealed that CDK family genes were homologous between cattle and other species in bovinae. The expression analysis and functional prediction indicated that CDKs may play an signicant and complicated role in regulating bovine adipocyte differentiation. The results provided an essential reference for further studies of CDK family genes in the regulation of adipocyte differentiation in cattle. against those of other ve bovine species above. The collinearity analysis between Bos taurus and other ve species for orthologous genes was conducted using MCScanX toolkit[46]. The results of collinearity analysis and orthologous CDKs were visualized by TBtools[45]. of

CDKs are a large family of serine/threonine protein kinases that were rst discovered in the regulation of cell cycle, and they had diverse functions in various of biological processes in eukaryotes including mRNA processing, regulation of transcription [8][9][10][11]. Recently, they have been shown to regulate adipocyte differentiation and lipid droplet formation by phosphorylating a series of associated transcription factors or adipocyte-speci c genes. CDK6, a member of CDK family genes, was targeted by miR107 to inhibit Notch and its downstream gene Hes1, thereby inhibiting glucose uptake and triglyceride synthesis in adipocytes [12]. MAPK and CDK2/cyclinA sequentially activated C/EBPβ by maintaining the phosphorylated state of Thr188 during the progression of mitotic clonal expansion (MCE) and adipocytes terminal differentiation [13]. Insulin activated the CCND3-CDK4 complex which in turn phosphorylated Ser388 of the insulin receptor IRS2 to maintain the active status of the insulin signaling pathway in adipocyte, eventually promoting de novo lipid synthesis [14]. In addition, CDK4 could phosphorylate Rb to release E2F leading to preadipocyte proliferation as well as phosphorylate PPARγ to regulate the terminal differentiation of adipocytes [15]. CDK5 could reduce the insulin sensitivity of adipocytes by phosphorylating Ser273 of PPARγ, and inhibition the phosphorylation of Ser273 would promote browning and thermogenesis of white adipose tissue [16]. CDK7 complex could inactivate PPARγ through the phosphorylation of PPARγ-S112 to inhibit adipogenesis [17]. CDK8 promotes the ubiquitination and degradation of SREBP-1c by phosphorylating its serine residues, thus inhibiting the adipogenesis [18]. These ndings inspired our curiosity to explore the effects of CDK family genes on bovine adipocytes differentiation.
However, the expression patterns and regulatory mechanisms of CDKs in bovine adipocytes have not been systematically studied and elucidated.
Therefore, the present study aimed to detect CDK family genes in the bovine genome, and then perform a detailed analysis of the classi cation, physicochemical properties, phylogenetic analysis, structural features, and functional analysis. Furthermore, the expression pattern analysis by transcriptom and qPCR veri cation was performed in order to identify essential members of CDK family that affect adipogenic differentiation. Our study provided a deep insight into CDKs that in uence adipogenic differentiation, which is essential for future study in improving IMF in the process of bovine breeding.

Results
Identi cation of the members in the CDK family  The length of amino acid sequences of 25 cattle CDK proteins ranged from 292 (CDK5) to 1512 (CDK13), and their molecular weight (Mw) was 33288.47-164717.14 Da, which correlated well with the protein length. The isoelectric points (pI) of most CDK family proteins was higher than 8.0, which containing more basic amino acids than acidic amino acids, except for 2 neutral proteins(CDK5 and CDK16), whose pI are 7.57 and 7.23, respectively, and 5 acidic proteins (CDK4, CDK6, CDK11B, CDK15 and CDK20), whose pI is between 5.34 and 6.68. Moreover, we detected all the 25 CDK proteins contained the Serine/Threonine Kinase conserved domain (Additional le 3).

Structural features of bovine CDK family members
To explore the structural characteristics of bovine CDK proteins and genes, the conserved motifs and gene structures were projected based on their phylogenetic relationships (Fig. 1). Results showed the CDKs of cattle initially categorized into three main subfamily according to the evolutionary clades. Among 25 bovine CDK family genes, the rst subfamily contains 6 members including CDKL1, CDKL2, CDKL3, CDKL4, CDKL5 and CDK20. The second subfamily possesses CDK10 and CDK11B, and the other members belongs to the third subfamily. Six conserved domains(Motif 1, 3, 5, 6, 7, and 9), containing 29, 21, 21, 21, 21, and 21 amino acids respectively, were shared among all the CDK family proteins (Additional le 4). As a small branch in the third subfamily, CDK16, CDK17 and CDK18, have all of the ten motifs. CDK4, CDK15 and CDK20 all consists of eight motifs, while CDK4 lacks of Motif 4 and Motif 10, CDK15 is short for Motif 2 and Motif 10, and CDK20 is without Motif 4 and Motif 10. The rest CDK proteins comprise nine motifs lacking of CDK10, which indicates they all have the same conserved patterns.
The items of introns, coding sequences (CDS) and untranslated region (UTR) were various among CDK family genes, for instance, the gene length CDKs ranged from 3599nt (CDK4) to 678562nt (CDK14), which is mainly due to the variation in intron. The number of CDS varied from 7 to 17 and the length and layout of 3'UTR and 5'UTR were also various in the noncoding areas. Although CDS, introns and UTRs varied greatly, analysis discovered that CDK family members in the same evolutionary branch tend to show similar gene structures and semblable conserved patterns in motifs.

Phylogenetic relationship of CDK proteins in different organisms
To assess evolutionary relationships of CDK proteins between cattle and other organisms, we conducted a phylogenetic analysis of animals in bovinae (Bos taurus, Bos indicus, Bos grunniens, Hybrid-Bos Indicus, Hybrid-Bos taurus, Bos mutus, Bison bison bison, and Bubalus bubalis). Besides, CDK proteins in Homo sapiens and Mus musculus were also included for they have been studied extensively as two model organisms. Accordingly, 236 amino acid sequences from 10 organisms were aligned to generate nonrooted Neighbor-Joining (NJ) tree Chromosomal distribution and collinearity analysis of CDK genes CDK family genes were mapped on the chromosomes of six bovinae species (Fig. 3). 25 represents the synteny of genes between two species, while 'N' means not and '-' means lacking of the gene.
The expression analysis of CDKs in different tissue The expression pattern of genes could provide important references for their function. To explore the expression pattern of the CDK gene family during adipogenic differentiation, we investigated the relative expression level in 163 samples of 60 tissue types including heart, liver, spleen, lung, kidney, muscle, fat, etc. The results showed that CDKs displayed differential expression patterns in diverse tissues (Fig. 5a), which could be classi ed into 5 groups (A to E). As a marker gene for adipocyte differentiation, PPARγ had a high expression in Group B including omental fat, intramuscular fat, subcutaneous fat and mammary gland fat, indicating that the results is reliable.
The 25 CDKs could be grouped into 4 categories according to their expression patterns and they all expressed in 60 tissues, suggesting that they may play a broad regulatory role in life activities. Group (CDK4, CDK9 and CDK11B) showed the highest expression levels, followed by Group (CDK3, CDK5, CDK7, CDK8, CDK10, CDK18, and CDK20) and Group (CDK1, CDK2, CDK6, CDK12, CDK13, CDK14, CDK16, and CDK17). Group comprised the rest members of CDKs, whose expression level was the lowest. Further analysis of the ve different fat tissues revealed that CDK9 was highly expressed in all the fat tissues and its expression pattern was similar to PPARγ (Fig. 5b).

Expression analysis of CDKs in preadipocytes and differentiated adipocytes by RNA-seq
Transcriptome analysis of 25 CDKs in preadipocytes and differentiated adipocytes revealed that CDKs showed a up-regulation trend in preadipocytes compared with differentiated adipocytes except for CDK1, CDK3, CDK6, CDK19, CDKL1 and CDKL4 (Fig. 6). CDK7 displayed a signi cant high expression, whereas CDK1 showed a signi cant low expression in preadipocytes within the 95% con dence interval. And CDK4, CDK8, CDK9 and CDK14 all displayed a signi cant high expression in preadipocytes within the 99% con dence interval.

Expression analysis of CDKs during adipocyte differention by qPCR
To further explore the expression pattern of CDK family genes, preadipocytes collected from perirenal adipose tissue of premature calves were induced differentiation. The results of oil red O staining showed that lipid droplet accumulation was signi cantly increased in adipocytes induced for 10 days compared to preadipocytes (Additional le 5), indicating that the induction and differentiation was successful. And we conducted qPCR to detect the expression of CDKs at 0, 2, 4, 6 and 10 days during adipocytes differentiation (Fig. 7). Results suggested that CDKs showed a relatively high expression in preadipocytes and then decreased as differentiation process went on in addition to CDK1, CDK15, CDK18, CDKL3 and CDKL5. The three members, CDK1, CDKL3 and CDKL5, all had the highest expression on the second day of differentiation and the lowest expression points were on the 6, 8 and 6 day, respectively. The expression level of CDK15 and CDK18 increased with adipocyte differentiation and reached the peak on the fourth day, then decreased.

Discussion
Cattle is known as an important species for supplying meat. The IMF content directly affects the taste and avor of beef and it is of great scienti c signi cance to reveal the molecular regulation mechanism of IMF deposition for meat quality improvement. The CDK family genes encoding functional proteins have been well studied in the regulation of transcription, metabolism and cell differentiation [8][9][10]. However, investigation of CDKs in adipocyte differentiation, especially in bovidae, was limited. Since cattle and several species of bovidae were sequenced, the vast amount of genetic resources might serve as references for exploring the evolution and function of CDK gene family and advancing genome science in bovidae.

Structural features of bovine CDK family proteins and genes
The activity of proteins depend on their functional motifs and domains [19]. Six conserved amino acid sequences including Motif 1, Motif 3, Motif 5, Motif 6, Motif 7 and Motif 9 were well-kept among all CDK family members in cattle, indicating the high conservation in motif distribution of CDK family proteins. These highly conserved motifs usually locate at the active sites of the enzymes, which may play essential roles in maintaining the structure, binding to substrate and catalyzing [20,21]. CDK16, CDK17 and CDK18, a small branch in a subfamily, have all of the ten motifs, meaning some speci c functions may exist in the three members. It is speculated that What's more, the gene structural analysis showed that the distribution and number of CDSs, introns, and UTRs were various in CDKs. This divergence was mainly caused by the length and layout of introns and UTRs, while the gene coding sequences translated proteins were similar. In other words, the nucleic acid sequences of bovine CDK family members were less conversed compared with amino acid sequences. These results have suggested that the similarity of amino acid sequences, especially that in the conversed motifs, may play essential roles in keeping the kinase functions of CDK proteins.

Phylogenetic relationship of CDK family proteins
The phylogenetic analysis of CDK family proteins in ten species provided an in-deep insight for their evolutionary relationships [22]. The results revealed that CDK family proteins were classi ed into eight major clades and the same member from different species rst clustered in one branch, indicating that they were conserved in sequences among the 10 species. Clade was separated out initially while Clade , and clustered into a subfamily and the others clustered into another subfamily, manifesting that they have been evolved asymmetrically and the evolutionary relationship between this three subfamilies may be relatively far. Notably, Clade includes all the Cyclin Dependent Kinases Like proteins (CDKL1, CDKL2, CDKL3, CDKL4 and CDKL5), which is consistant with the study in human that divided the CDKs into CDK and CDKL [10]. As expected, members of the CDK proteins with a closer relationship tend to have a nearer evolutionary distance, which means that they may cluster together rst. For example, Bos Indicus CDK1 rst clustered with that of Bubalus bubalis, and then get together with Hybrid-Bos taurus, Bos taurus, Bos mutus, Hybrid-Bos Indicus, Bos grunniens, Bison, human and mouse in turn.

Collinearity analysis of CDKs in bovidae
Page 11/18 The family genes may distribute on different chromosomes or co-locate on the same chromosome, which are generally de ned as segmental duplication events in the former and tandem duplication events in the latter [23][24][25]. Chromosomal distributions of the CDK family genes showed that they located on 13 to 16  14 Mb in Chr 9 of Bos grunniens. The de ciency and discrepancies of CDK genes might be caused by the sequence variation and chromosome rearrangement in the process of evolution [29]. In addition, Chr16 of Bos taurus showed syntenic relationship with Chr5 and scaffold NW_020228957.1 of Bubalus bubalis (Fig. 4e), suggesting that NW_020228957.1, which hasn't been assembled yet, may be a part of Bubalus Chr5. Meanwhile, it was syntenic between CDK18 of Bos taurus Chr16 and that in scaffold NW_020228957.1 of Bubalus. In a word, the extensive homology provided rich perspectives for studying the function and evolution of CDK family genes in bovidae.
CDK genes affecting adipocyte differentiation CDK family proteins, as a kind of phosphorylases, could regulate adipocytes differentiation by phosphorylating a series of transcription-related factors or adipocyte-speci c genes [12][13][14][15][16]18]. To dissect the expression pattern of CDK family genes, we analyzed the expression values of the 25 members in 60 tissue types. As a results, CDK4, CDK9 and CDK11B showed the highest expression in four types of fat tissues (omental fat, intramuscular fat, subcutaneous fat and mammary gland fat) in relative with other tissues. So we suspected the three genes may play more important and wide-ranging roles in adipose tissue compared with other members of CDK family.
Previous studies have revealed that CDK4 could phosphorylate IRS2 and Rb to promote adipogenesis [14,15].
CDK9, a component of positive transcription elongation factor b (P-TEFb), could phosphorylate the C-terminal domain of RNA polymerase II and regulate the transcription of target genes by facilitating transcriptional elongation [30]. In 3T3-L1 cells, CDK9 increased the adipogenic potential by phosphorylating PPARγ directly and inducing its transcriptional activity [31]. CDK11B had similar expression patterns with CDK4 and CDK9 by tissue expression analysis, while it has not been reported to be involved in adipogenic differentiation by now. It would be valuable to further explore the function of CDK11B in the regulation of adipocytes differentiation.
Adipogenic differentiation is a complicated and well-organized process regulated by multiple genes. Analyzing the expression patterns of CDK family genes during adipocytes differentiation is the basis of exploring their functions. Results of transcriptome analysis and qPCR validation both revealed that CDK4, CDK7, CDK8 and CDK9 showed signi cant high expression in preadipocytes. It is speculated that the four members may have important functions in targeting newly generated lipid droplets. The expression of CDK1, CDKL3 and CDKL5 reached the highest in the second day, while CDK15 and CDK18 reached the peak in the fourth day indicating that they may play regulatory roles during adipocyte differentiation in turn. The functions of CDK1, CDK4, CDK7, CDK8 and CDK9 in adipocytes differentiation have been preliminarily studied and need to be further explored [14,15,17,18,[31][32][33][34][35]. Expression analysis revealed that CDK15, CDK18, CDKL3 and CDKL5, whose functions has not been studied, may also play signi cant roles. In addition, the expression trends of some members were inconsistent between RNA-seq and qPCR validation. For instance, there was no signi cant differences in the expression of CDK2, CDK6, CDK10, CDK11B, CDK12, CDK16, CDK17, CDK19 and CDKL1 by RNA-seq analysis, but they showed a signi cant down-regulation in qPCR detection. This discrepancy may caused by different sample sources. The samples for RNA-seq were separated from inguinal subcutaneous fat of two 1 year old male Qinchuan cattle, while samples for qPCR were from perirenal fat of a premature female Holstein calf. In summary, the functions of CDKs during adipocytes differentiation is complicated and need to be studied in depth and analyzed comprehensively.
The CDK family genes and the interacted genes constructed an integrative network by literature mining using Agilent Literature Search plug-in of Cytoscape (Additional le 7) [36]. For example, CDK7 could directly activate CDK9 to maintain the high expression of MDM4 and MDM2 [34,35]. MDM2 facilitates adipocyte differentiation through CRTC-mediated activation of STAT3 [37]. Overall, the results revealed that CDK family genes encoding the enzymes directly or indirectly interact with each other or some other genes, playing non-redundant roles, collectively regulating the life activity including cell cycle, adipocyte differentiation, lipid metabolism etc.

Conclusions
This study conducted a comprehensive genome-wide analysis of CDK family genes in bovidae. A total of 185 CDK genes were identi ed and grouped into eight distinct clades. Collinearity analysis revealed that CDK family genes were homologous between cattle and other species in bovinae. The expression analysis and functional prediction indicated that CDKs may play an signi cant and complicated role in regulating bovine adipocyte differentiation. The results provided an essential reference for further studies of CDK family genes in the regulation of adipocyte differentiation in cattle.

Structural features analysis
To further evaluate the structural diversity of cattle CDK genes and proteins, a phylogenetic Neighbor-Joining tree was constructed and the conserved motifs were detected in MEME 5.0 [44] and visualized in TBtools [45]. The minimum and maximum number of amino acids in each motif were 6 and 50. The motif number of each CDK protein was limited to 10. Also, coding sequences and corresponding genomic sequences of bovine CDKs were loaded into the TBtools to portray the numbers and positions of CDSs and introns graphically.

Chromosomal distribution and collinearity analysis
Positional information of predicted CDK genes of Bos taurus, Bos grunniens, Hybrid-Bos Indicus, Hybrid-Bos taurus, Bos indicus and Bubalus bubalis were extracted from the genomic sequence and annotation les and then were visualized in TBtools [45]. The identi ed CDKs of each species were mapping on chromosomes.
Comparisons between each two genomes were determined by all-against-all BLASTP searches (e-value = 10 − 5 ) using the proteome sequences of Bos taurus as queries against those of other ve bovine species above. The collinearity analysis between Bos taurus and other ve species for orthologous genes was conducted using MCScanX toolkit [46]. The results of collinearity analysis and orthologous CDKs were visualized by TBtools [45].
The RNA-Seq data of preadipocytes and differentiated adipocytes was downloaded from the National Center for Biotechnology Information (NCBI) Sequence Read Archive(SRR3056892, SRR3064490, SRR3064491, SRR3064492) [47] and transformed into fastaq format by Fastq-dump.The sequencing quality was checked using FastQC [48]. Quality control of raw sequence data, including removal of the adapter sequences and low-quality sequences were performed using the Trim_galore. Clean reads were then mapped to the Bos taurus genome(ARS-UCD1.2.101) using STAR. The RSEM and FeatureCounts was used to calculate the expression of transcripts. Data was normalized by calculating the RPKM for each gene. These results were used to analyze the expression of CDKs between preadipocytes and differentiated adipocytes in cattle. The RNA-Seq data of 163 bovine tissue samples were downloaded from Ruminant Genome Database (http://animal.nwsuaf.edu.cn/code/index.php/RGD) [49]. The SRR number and adjusted RPKM values of 163 tissue samples were provided in Additional le 8. The heatmap was performed in R software.
Isolation, culture and induction differentiation of bovine primary adipocytes Primary adipocyte was isolated and cultured from the perirenal adipose tissue of premature calf in Zerui ecological breeding farm. Type collagenase digestion method was used for the isolation and cultivation of calf preadipocytes. The method described by Huang et al. [50] was adopted in the induction of preadipocytes differentiation, and the method described by Wang et al. [51] was applied for oil red O staining.
Rna Extraction And Quantitative Rt-pcr (qrt-pcr) RNA extraction and quantitative RT-PCR (qRT-PCR) According to reference sequence from NCBI, quantitative primers of CDK family genes were designed used Primer Premier 5.0 software and the primer sequences were provided in Additional le 9. Total RNA were extracted at 0d, 2d, 4d, 6d and 10d during the differentiation of bovine preadipocytes by phenol-chloroform method using the TRIzol reagent (9109, Takara). RNA samples were measured for absorbance at 260 nm and 280 nm in the multifunctional full-wavelength Multiskan and the samples with an OD260/OD280 ratio between 1.8 and 2.0 was used in the subsequent experiment. Then, 1000 ng total RNA was reverse transcribed using random primers with Moloney murine leukemia virus reverse transcriptase (Takara Bio, Kyoto, Japan). Realtime PCR was carried out in a CFX96 Touch Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA) with SYBR Green Master Mix (Takara Bio, Kyoto, Japan).

Statistical Analysis
All qRT-PCR results were calculated using a 2 −∆∆Ct method. Three independent technical repetitions were processed for each test. Statistical signi cance was examined using Graphpad Prism 7.0 software.