Introduction

Sheep are one of the main livestock resources around the world, consuming the produce meat, milk, wool, and fur [1]. Fat-tailed sheep make up a quarter of the world's sheep population. In China, there are more than 98 indigenous sheep breeds, 80% of which are fat-tailed sheep. Fat tails are also food for humans. However, with the increase in people's life standards, mutton consumption has been increasing, and the utilization rate of fat in sheep is lower. The intensity of fat deposition in the sheep tail is higher than that in the rest of the body [2]. Conversely, excessive fat deposition affects the feed conversion rate of sheep, which increases the cost of farmers' breeding. In production, a large amount of fat tail is directly discarded, forming a considerable amount of waste. Reducing fat deposition can produce leaner meat. It is important for the sheep industry for pastures to increase economic profits [3]. Consequently, this provides a better understanding of the molecular mechanism of lipogenesis in sheep breeding and control of carcass fat [4].

In recent years, several genomic approach studies have been conducted to reveal the important genes of the fat tail phenotype in different sheep breeds [5]. BMP2 and PDGFD are likely potential genes related to fat deposition in the tails of sheep [1. These data were preliminarily analysed. The sample correlation was performed with the Pearson correlation coefficients based on gene expression levels, and these coefficients were reflected in the form of a heatmap (Fig. 1).

Table 1 Descriptive statistics of sequence quality and map** rate from Hu sheep and Tibetan sheep
Full size table
Fig. 1
figure 1

a Heatmap of gene expression of adipose tissue in the Hu sheep (H1, H2 and H3) and Tibetan sheep (Z1, Z2 and Z3) tails; b PCA of gene expression of adipose tissue in the Hu sheep (H) and Tibetan sheep tails (Z)

Full size image

Differential expressed genes analysis

RNA-Seq analysis of the two sheep breeds showed a total of 2108 genes detected using the reference genome. The genes with |FRKM|≥ 1.5 and FDR ≤ 0.01 were identified as DEGs, resulting in 1247 upregulated and 861 downregulated DEGs in the Hu sheep breeds. The top 8 differentially expressed genes involved in fat deposition were S100A8, CLDN, TBX15 and HGF, which were upregulated in Hu sheep, while SELENBP1, MSC, EIF4EBP4 and GRB10 were downregulated in Hu sheep (Fig. 2, Table S1).

Fig. 2
figure 2

a The upregulated and downregulated genes in adipose tissue of Hu sheep tail. b RT–qPCR validation of 8 randomly selected genes identified by RNA-Seq analysis

Full size image

Eight DEGs were selected for RT–qPCR to validate our RNA-seq data. Among these DEGs, the trends of gene expression changes were consistent with the RNA-seq results (Fig. 2b).

GO and KEGG enrichment analyses

GO and KEGG enrichment analyses were performed to obtain the biological relationships of the DEGs. The results of GO terms showed a nominal significance of Q-value < 0.01 in that 134 GO terms were enriched in the three groups (Fig. 3a). These included lipids metabolic process (GO:0,006,629), fatty acid metabolic process (GO:0,006,631) and cellular lipid metabolic process (GO:0,044,255). KEGG also identified 13 significant pathways (Q-value < 0.01), including regulation of lipolysis in adipocytes (Q-value = 2.04E-04), carbon metabolism (Q-value = 1.13E-04), thermogenesis (Q-value = 1.35E-04) and glycine, serine, and threonine metabolism (Q-value = 8.35E-04) (Fig. 3b). These pathways were involved in fat metabolism or energy metabolism. ClueGO functional analysis of these potential DEGs also constructed a plausible pathway network for fat deposition in sheep [15]. A total of 219 terms were enriched. ClueGO analysis showed that most DEGs were regulated by fatty acid derivative metabolic processes (P value Bonferroni = 0.0096), regulation of plasma lipoprotein particle levels (P value Bonferroni = 0.0249) and lipid oxidation (P value Bonferroni = 0.0294) (Fig. 3c). Furthermore, some pathways associated with cold stimulation were also enriched, such as regulation of cold-induced thermogenesis (P value Bonferroni = 1.8618E-06), cold-induced thermogenesis (P value Bonferroni = 1.8618E-06) and temperature homeostasis (P value Bonferroni = 6.3048E-06).

Fig. 3
figure 3

a GO enrichment for differentially expressed genes in adipose tissue of Hu sheep tail and Tibetan sheep; b KEGG enrichment for differentially expressed genes in adipose tissue of Hu sheep tail and Tibetan sheep; c Mechanism of fat deposition signalling pathway in differentially expressed genes; d The top 40 hub differentially expressed genes identified in PPI networks

Full size image

To further determine the functional relationship to understand the DEGs, a PPI network formed interactions between upregulated and downregulated DEGs. According to node degree, we identified hub genes among these important DEGs. For better visualization, we reconstructed the interactors of the top 40 DEGs (Fig. 3d) using the cytohubba plug-in. GPI, ACACA, ACSS2, TTN, and FASN were upregulated in Hu sheep, and WAS, LPIN1, GFAP, and FN1 were downregulated in Hu sheep.

The effect of BMP2 on adipogenesis induction

Based on our previous studies, we applied selection signal analysis to identify selection signals in sheep with different tail types [12]. These two methods found 43 candidate genes that may potentially be related to fat tail development, including BMP2, HOXA11 and PPP1CC, which may play important roles in fat tail formation. Among these genes, BMP2 is also strongly selected in the largest region by hapFLK [12]. Fat-tailed fixation is caused by a selective sweep near the retrotransposition hotspot on chromosome 13, and this diversity affects BMP2 expression [14, 16]. We performed western blotting of BMP2 in the tail fat of Hu sheep and Tibetan sheep, and the results showed that BMP2 was highly expressed in Hu sheep (Fig. 4a). This was therefore selected for BMP2 overexpression in preadipocytes. BMP2 overexpression lentivirus transfected into preadipocytes (Fig. 4b), and after induced differentiation, BMP2 overexpression also increased the mRNA levels of PPAR-γ and LOX when compared with NC at 0 d (P value < 0.05) (Fig. 4c, d). There was no difference at 1, 3 and 5 days. Sheep preadipocytes on day 5 were stained with oil red O. Many small lipid droplets were stained red, and lipid rings were visible. The number of lipid drops in the BMP2 overexpression group was greater than that in the NC group (Fig. 4e), which showed that fat deposition in the sheep tail was increased by BMP2 overexpression.

Fig. 4
figure 4

a The regulation of the protein level of BMP2 in adipose tissue of Hu sheep tail (H1, H2 and H3) and Tibetan sheep tail (Z1, Z2 and Z3). We cut 30-70 kDa of protein to transferred onto a PVDF membrane. b Relative expression of BMP2 in BMP2 over-expression and NC. c Relative expression of PPAR-γ in BMP2 over-expression and NC. d Relative expression of LOX in BMP2 over-expression and NC. e Oil red O staining when BMP2 was overexpressed in sheep preadipocytes

Full size image

Discussion

Sheep tail fat deposition has a complex genetic regulation mechanism, which is determined by the environment and genes. Fat-tails are used to save energy in food plentiful seasons. During cold winters and in harsh environments, it provides the necessary energy to help sheep subsist [3]. Tail adipose tissue is an important tissue site for fat deposition in fat tail sheep. Research on the tail cutting of Mongolian sheep and Lanzhou large-tailed sheep found that the fat originally deposited in the tail was blocked by the tail fat metabolic pathway, and part of the fat was transferred and deposited in the rest of the body, mainly subcutaneous fat and intra-abdominal fat [17]. Therefore, the study of the fat tail is also an important material in the study of fat metabolism. Hu sheep are a short fat-tailed sheep breed and are also a good model to study fat deposition in the tail. Previous studies have mainly focused on the differential gene expression patterns or molecular genetic mechanisms among different sheep [1, 2, 8, 28]. VLDLR is involved in triglyceride metabolism [29] and is associated with fat deposition in pigs [50]. Then, raw reads were trimmed with Trimmomatic to remove adapter sequences and poly-N and low-quality sequence reads to obtain clean reads [2, 51]. Clean reads were saved in fastq format. Clean reads were processed using HISAT v2.1.0 [52], and the clean reads were aligned to Ovis_aries (Oar_v3.1). Moreover, using Bowtie v2.2.5, we built an index of the reference genome [53], and the level of expression was calculated using RSEM [54]. In addition, principal component analysis (PCA) was performed using SARTools software [55]. According to the gene expression patterns, the samples were clustered, and the degree of similarity/difference between the gene expression profiles was detected [2].

Functional analysis and PPI network construction

Phyper function in R performed gene set enrichment analysis in up- or downregulated genes. The significant enrichment of Gene Ontology (GO) functional terms with DEGs that showed differential expression was analysed (P value ≤ 0.01) [56]. Furthermore, the significantly DEGs were also enriched in the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (P value ≤ 0.05) [57]. Among these DEGs, protein–protein interactions (PPIs) were considered in the STRING database (https://cn.string-db.org/) [58].

Isolation of preadipocytes from adipose tissue of Hu sheep and transfection of BMP2 overexpression

The BMP2 overexpression lentivirus was constructed as described by Lu [14]. An empty lentivirus vector was used as the negative control (NC) [14]. Preadipocytes were isolated from Hu sheep tail fat (70-day-old foetus). Primary preadipocytes were cultured in vitro by collagenase digestion. The cells were incubated in complete medium at 37 ℃ for two days, and the cells were almost all adherent to the wall. Cells were cultured in a plate with 1000 µl medium (5% FBS, 1% PS) in 6-well plates, and the titration of lentiviruses was MOI = 100 of the final construct together with BMP2 overexpression in triplicate for 24 h. Then, the cells were transferred to new complete medium (10% FBS, 1% PS). When the cells showed contact inhibition, the induction differentiation medium (complete medium + 0.5 mM isobutylmethylxanthine + 10 mg/mL insulin + 1 µM dexamethasone) was changed for 2 days. The final cells were cultured in maintenance differentiation medium (complete medium + 10 mg/mL insulin) for 2 days [59]. Cells were set as the first day when cultured with differentiation medium. RNA was extracted from cells with BMP2 overexpression and NC at various times (0, 1, 3 and 5 days).

Oil red O staining

Oil red O dye (Solarbio, China) and distilled water at a ratio of 3:2 was used to filter the mixture. Cells differentiated for 5 days were washed twice with PBS and fixed with 4% paraformaldehyde for 20 min. Then, oil red O dye was added, and the cells were incubated for 15 min, washed with distilled water 2–3 times and observed and photographed under a microscope.

RT–qPCR and western blot

Reverse transcription was performed according to the instructions of the Primer Script II 1st strand kit (Takara, China). The primers were designed by Primer 5.0 (Table S2). β-actin was used as the reference gene. Three biological replicates and triplicate technical replicates were obtained. The reaction and calculation were described by ** [60].

Fat tissues were extracted from tails of Hu sheep (H1, H2 and H3) and Tibetan (Z1, Z2 and Z3) and 1 ml RIPA lysis buffer and 1 mM PMSF (Beyotime, Shanghai, China) were added to obtain the total proteins; thus, the protein concentrations were measured with the BCA method (Beyotime, China). Proteins were separated on 10% SDS–PAGE and cut 30–70 kDa of protein to transferred onto a PVDF membrane (Millipore, USA). The membrane was sealed with quick sealing fluid (Lablead, China) and washed with TBST (Solarbio, China) three times. Proteins were detected with rabbit monoclonal anti-β-tubulin (50 kDa, Proteintech, USA) and rabbit monoclonal BMP2 (44 kDa, Proteintech, USA). The reaction band was developed by using enhanced chemiluminescence (Epizyme, China) and images of the PVDF membrane were recorded with a JP-K600 imaging system (JiaPeng, China).