The Emerging Role of Zfp217 in Adipogenesis

Zinc finger protein 217 (Zfp217), a member of the krüppel-type zinc finger protein family, plays diverse roles in cell differentiation and development of mammals. Despite extensive research on the functions of Zfp217 in cancer, pluripotency and reprogramming, its physiological roles in adipogenesis remain unknown. Our previous RNA sequencing data suggest the involvement of Zfp217 in adipogenesis. In this study, the potential function of Zfp217 in adipogenesis was investigated through bioinformatics analysis and a series of experiments. The expression of Zfp217 was found to be gradually upregulated during the adipogenic differentiation in C3H10T1/2 cells, which was consistent with that of the adipogenic marker gene Pparg2. Furthermore, there was a positive, significant relationship between Zfp217 expression and adipocyte differentiation. It was also observed that Zfp217 could not only trigger proliferative defect in C3H10T1/2 cells, but also interact with Ezh2 and suppress the downstream target genes of Ezh2. Besides, three microRNAs (miR-503-5p, miR-135a-5p and miR-19a-3p) which target Zfp217 were found to suppress the process of adipogenesis. This is the first report showing that Zfp217 has the capacity to regulate adipogenesis.


Introduction
Obesity, a major global public health problem, is a complex disease that involves interactions between environmental and genetic factors [1]. The number of obesity patients keeps increasing year by year [statistical data by World Health Organization (WHO) official website], and related research has also experienced an explosive growth since 2011 ( Figure S1). Hence, it will be of great importance to further study the adipocytic gene and explore the mechanism of obesity for therapeutic strategies.
Excess body adiposity is largely attributable to adipocyte hypertrophy and hyperplasia [2]. The hyperplasia is the result of excessive stem cells or preadipocytes differentiating into mature adipocytes (a process called adipogenesis). Adipogenesis is a complicated process that usually consists of six stages: mesenchymal precursor, committed preadipocyte, growth-arrested preadipocyte, mitotic clonal expansion, terminal differentiation, and mature adipocyte [2]. The adipogenic event involves a cascade of transcription factors and signaling pathways, with peroxisome proliferator-activated receptor g (Pparg) and CCAAT/enhancer-binding proteins (Cebps) considered as the crucial determinants of adipogenic fate [3].
In the past decades, despite reports about many important signaling pathways and certain essential transcriptional factors involved in adipogenesis, many adipogenic factors still remain to be excavated. Recent studies have revealed that a small but increasing number of zinc finger proteins could act as key transcriptional regulators in adipogenesis [4]. The zinc finger protein family is extremely abundant in higher eukaryotes and can function as sequence-specific DNA-binding factors in a variety of life processes. Previous studies discovered that Shn-2, ZNF395, Zfp423, Zfp467, Zfp36L1, BCL6 and Zfp521 have a pivotal role in adipogenic commitment [4][5][6]. ZNF638, SLUG, Egr2, FBI-1, MCP-1, Hzf, MORC2, A20, Repin1, Zfp407, BCL11B and YY1 are positive regulators of preadipocyte differentiation [4,[7][8][9][10][11][12][13][14], while GATA2, GATA3, Egr1, Snail and PLZF are reported as negative regulators of the transition from preadipocyte to adipocyte [4,15]. In addition, the majority of the zinc-finger protein family members of Krüpple-like factors are reported to promote or suppress preadipocyte differentiation [4,16]. In our previous work, differentially expressed genes (DEGs) were identified by RNA sequencing during the in vitro adipocytic differentiation of porcine subcutaneous stromal vascular cells (ASVC) on Days 0, 2 and 4, and Zfp217 was found significantly upregulated at Day 2 during the adipogenesis [17]. It suggested that Zfp217 was probably implicated in regulating cell adipogenic event.

The Potential Adipogenesis Role of Zfp217 Based on Gene Expression Omnibus (GEO) Datasets
To investigate the potential roles of Zfp217 in adipogenesis and obesity, we scanned differential expression genes associated with adipogenesis in GEO datasets, and found that the expression of Zfp217 was highly correlated with marker genes during the adipogenesis process of model cells ( Figure 1A-C). As shown in Figure 1A, the expression of Zfp217 and Pparg was nearly consistent after adipogenic induction. Generally, obesity triggers multiple metabolic syndromes and is considered the most important predictor of diabetes mellitus. Obesity individuals show higher free fatty acid levels which stimulate insulin resistance, and then develop into diabetes mellitus [38]. Here, the expression of Zfp217 fluctuated with that of Pparg in three kinds of individuals classified as having diabetes mellitus (DM), impaired glucose tolerance (IGT), and normal glucose tolerance (NGT) ( Figure 1D), and this implied that the potential effect of Zfp217 on obesity may be connected with Pparg. In addition, it found that Zfp217 and Pparg showed a similar trend in Ezh2 flox/flox preadipocytes when infected with retroviruses expressing Cre or vector alone ( Figure 1E). The Wnt signaling genes are also changed with Ezh2 as in previous research ( Figure 1E).

The Association of Zfp217 Expression with Adipogenesis and Obesity
In order to explore the relationship between Zfp217 expression and adipogenesis, we investigated the expression of Zfp217 in non-differentiated C3H10T1/2 cells and during their differentiation into adipocytes. During adipogenesis, the mRNA level of Zfp217 increased steadily, with an obvious peak at Day 3 (by 17-fold compared to 0 h; p < 0.01), followed by a decrease at Day 4, and then a significant increase during differentiation up to Day 8 and Day 10 (both by 43-fold compared to 0 h and p < 0.01) (Figure 2A). At the protein level, the expression of Zfp217 was highly correlated with the extension of the induction time, and the protein expression pattern was similar to that of the adipogenic marker gene Pparg2 ( Figure 2B). Furthermore, we separated five kinds of adipose tissues from high fat diet (HFD) and normal chow diet (NCD) fed mice based on Zfp217's potential role in obesity, and tested the mRNA level of Zfp217 and Pparg2. When compared with the NCD group, the mRNA level of Zfp217 and Pparg2 for the HFD group showed a similar trend in the visceral adipose tissues (which included epididymal, inguinal and perirenal adipose) ( Figure 2C), but presented a reverse trend in the brown adipose and subcutaneous adipose ( Figure 2C).
Meanwhile, in order to investigate the expression of Zfp217 in cell models with different adipogenic ability, and the variation of Zfp217 expression during the early stage of adipogenic induction, we examined the Zfp217 expression levels in NIH3T3, C3H10T1/2 and 3T3-L1 at non-differentiation and 6 h after differentiation. Notably, the mRNA level of Zfp217 was increased in all three cells at 6 h after differentiation ( Figure 2D). We also investigated the mRNA level of early pro-adipogenic marker genes. Interestingly, both the expression patterns of Zfp217 in the three cell models and at 6 h post-differentiation are similar to those of early pro-adipogenic marker genes (Cebpb, KLF4, Ebf1 and ZNF395). Furthermore, we separated five kinds of adipose tissues from high fat diet (HFD) and normal chow diet (NCD) fed mice based on Zfp217's potential role in obesity, and tested the mRNA level of Zfp217 and Pparg2. When compared with the NCD group, the mRNA level of Zfp217 and Pparg2 for the HFD group showed a similar trend in the visceral adipose tissues (which included epididymal, inguinal and perirenal adipose) ( Figure 2C), but presented a reverse trend in the brown adipose and subcutaneous adipose ( Figure 2C).
Meanwhile, in order to investigate the expression of Zfp217 in cell models with different adipogenic ability, and the variation of Zfp217 expression during the early stage of adipogenic induction, we examined the Zfp217 expression levels in NIH3T3, C3H10T1/2 and 3T3-L1 at nondifferentiation and 6 h after differentiation. Notably, the mRNA level of Zfp217 was increased in all three cells at 6 h after differentiation ( Figure 2D). We also investigated the mRNA level of early proadipogenic marker genes. Interestingly, both the expression patterns of Zfp217 in the three cell models and at 6 h post-differentiation are similar to those of early pro-adipogenic marker genes (Cebpb, KLF4, Ebf1 and ZNF395).

Zfp217 Regulates Adipogenesis of C3H10T1/2 and 3T3-L1
In order to examine the effects of Zfp217 on adipogenesis, siRNA-mediated loss-of-function and overexpression-mediated gain-of-function approaches were used to study the effect of Zfp217 on the adipogenic differentiation of C3H10T1/2 and 3T3-L1. The mRNA level of Zfp217 was significantly changed at Day 2 after transfection ( Figure S2). Adipocyte differentiation was evaluated by Oil Red O staining and a triglyceride assay. As shown in Figure 3A, Zfp217 knockdown resulted in an obvious decrease in the number of lipid droplets formed, both in C3H10T1/2 and 3T3-L1. Consistent with this observation, triglyceride contents were also significantly attenuated due to lack of Zfp217 in C3H10T1/2 and 3T3-L1 ( Figure 3B). Meanwhile, we also observed that Zfp217 overexpression led to an induction of white adipocyte phenotype lipid droplet formation, a slight increase of the triglyceride content in 3T3-L1, and a remarkable increase in C3H10T1/2. These results imply that Zfp217 is required for adipogenesis of the two types of adipogenic cell models.

Zfp217 Positively Regulates Adipogenesis by Suppressing Cell Proliferation and Interacting with Ezh2
Given that Zfp217 regulates adipogenesis with its loss-of-function and gain-of-function in C3H10T1/2, we further investigated the potential mechanism of Zfp217 in C3H10T1/2. 5-Ethynyl-2′deoxyuridine (EdU) assays were performed on cells transiently transfected separately with Zfp217 siRNA or pcDNA3.1-Zfp217 and its control. Our data indicated that Zfp217 overexpression blocked DNA synthesis in C3H10T1/2, whereas Zfp217 knockdown significantly increased EdU incorporation (p < 0.05) ( Figure 4A,B). It was also observed that Zfp217 could increase the amount of C3H10T1/2 cells in G0/G1-phase ( Figure S3). That is, Zfp217 arrested C3H10T1/2 cells at the G0/G1 phase, and Zfp217 overexpression blocked DNA synthesis, which may be mediated by changes in the cell cycle.
According to previous reports and our bioinformatics analysis, we speculated that Zfp217 may participate in adipogenesis by combining with Ezh2, which suppresses Wnts. In order to test this hypothesis, the open reading frame (ORF) of full-length mouse Zfp217 was cloned into the pACT expression vectors to generate VP16-Zfp217 chimeric proteins. Meanwhile, the ORF of full-length mouse Ezh2 was cloned into the pBIND expression vectors to generate GAL4-Ezh2 chimeric proteins. Next, the interaction of Zfp217 and Ezh2 was tested. A robust interaction requires that the mammalian two-hybrid assay signal in pACT-Zfp217/pBIND-Ezh2 should be significantly above background signals derived from both vectors individually ( Figure 4D). We did find that the group of pACT-

Zfp217 Regulates Adipogenesis of C3H10T1/2 and 3T3-L1
In order to examine the effects of Zfp217 on adipogenesis, siRNA-mediated loss-of-function and overexpression-mediated gain-of-function approaches were used to study the effect of Zfp217 on the adipogenic differentiation of C3H10T1/2 and 3T3-L1. The mRNA level of Zfp217 was significantly changed at Day 2 after transfection ( Figure S2). Adipocyte differentiation was evaluated by Oil Red O staining and a triglyceride assay. As shown in Figure 3A, Zfp217 knockdown resulted in an obvious decrease in the number of lipid droplets formed, both in C3H10T1/2 and 3T3-L1. Consistent with this observation, triglyceride contents were also significantly attenuated due to lack of Zfp217 in C3H10T1/2 and 3T3-L1 ( Figure 3B). Meanwhile, we also observed that Zfp217 overexpression led to an induction of white adipocyte phenotype lipid droplet formation, a slight increase of the triglyceride content in 3T3-L1, and a remarkable increase in C3H10T1/2. These results imply that Zfp217 is required for adipogenesis of the two types of adipogenic cell models.

Zfp217 Positively Regulates Adipogenesis by Suppressing Cell Proliferation and Interacting with Ezh2
Given that Zfp217 regulates adipogenesis with its loss-of-function and gain-of-function in C3H10T1/2, we further investigated the potential mechanism of Zfp217 in C3H10T1/2. 5-Ethynyl-2 -deoxyuridine (EdU) assays were performed on cells transiently transfected separately with Zfp217 siRNA or pcDNA3.1-Zfp217 and its control. Our data indicated that Zfp217 overexpression blocked DNA synthesis in C3H10T1/2, whereas Zfp217 knockdown significantly increased EdU incorporation (p < 0.05) ( Figure 4A,B). It was also observed that Zfp217 could increase the amount of C3H10T1/2 cells in G0/G1-phase ( Figure S3). That is, Zfp217 arrested C3H10T1/2 cells at the G0/G1 phase, and Zfp217 overexpression blocked DNA synthesis, which may be mediated by changes in the cell cycle.
According to previous reports and our bioinformatics analysis, we speculated that Zfp217 may participate in adipogenesis by combining with Ezh2, which suppresses Wnts. In order to test this hypothesis, the open reading frame (ORF) of full-length mouse Zfp217 was cloned into the pACT expression vectors to generate VP16-Zfp217 chimeric proteins. Meanwhile, the ORF of full-length mouse Ezh2 was cloned into the pBIND expression vectors to generate GAL4-Ezh2 chimeric proteins. Next, the interaction of Zfp217 and Ezh2 was tested. A robust interaction requires that the mammalian two-hybrid assay signal in pACT-Zfp217/pBIND-Ezh2 should be significantly above background signals derived from both vectors individually ( Figure 4D). We did find that the group of pACT-Zfp217/pBIND-Ezh2 had a significant interaction signal comparable to control group ( Figure 4C). Therefore, we can conclude that Zfp217 also interacts with Ezh2 in the mouse background.
Additionally, in order to confirm whether Zfp217 represses Wnt signaling genes to facilitate adipogenesis, we checked the expression changes of Wnts (Wnt6 and Wnt10b) and two major adipogenic marker genes (Cebpa and Pparg2) after treatment with Zfp217. It was observed that Zfp217 overexpression could notably suppress the mRNA level of Wnt6 at Day 4 post-adipogenic induction, and that of Wnt10b at Day 2 ( Figure 4E), but the mRNA level of the two Wnt signaling genes was slightly decreased in the other time points. Furthermore, at Day 4 of post-adipogenic induction, the mRNA levels of Cebpa and Pparg2 were significantly changed, with variation in Zfp217 expression ( Figure 4F). Hence, it can be concluded that Zfp217 could positively regulate adipogenesis by suppressing cell proliferation and interacting with Ezh2, which represses Wnt signaling genes. Zfp217/pBIND-Ezh2 had a significant interaction signal comparable to control group ( Figure 4C). Therefore, we can conclude that Zfp217 also interacts with Ezh2 in the mouse background. Additionally, in order to confirm whether Zfp217 represses Wnt signaling genes to facilitate adipogenesis, we checked the expression changes of Wnts (Wnt6 and Wnt10b) and two major adipogenic marker genes (Cebpa and Pparg2) after treatment with Zfp217. It was observed that Zfp217 overexpression could notably suppress the mRNA level of Wnt6 at Day 4 post-adipogenic induction, and that of Wnt10b at Day 2 ( Figure 4E), but the mRNA level of the two Wnt signaling genes was slightly decreased in the other time points. Furthermore, at Day 4 of post-adipogenic induction, the mRNA levels of Cebpa and Pparg2 were significantly changed, with variation in Zfp217 expression ( Figure 4F). Hence, it can be concluded that Zfp217 could positively regulate adipogenesis by suppressing cell proliferation and interacting with Ezh2, which represses Wnt signaling genes.   . Zfp217 suppresses cell proliferation and interacts with Ezh2, which represses Wnt signaling genes, to facilitate adipogenesis. (A) DNA synthesis of C3H10T1/2 cells was measured by 5-ethynyl-2′-deoxyuridine (EdU) incorporation assay after the indicated transfection. The EdU staining (red dots) represents the population of newborn cells. Hoechst 33342 staining (blue dots) was used to label cell nuclei. Scale bar indicates 100 μm; (B) EdU incorporation quantitative analysis. Five microscopic fields were randomly selected. EdU positive cells (%) were quantified by Image J, which was calculated by the formula: EdU positive cell (100%) = red dots/(red dots + blue dots) × 100%. All data were obtained from three reproducible experiments; (C) In the mammalian two-hybrid system, the Zfp217-Ezh2 interaction is confirmed by significantly higher luciferase activity in cells transfected with pACT-Zfp217 and pBIND-Ezh2, compared to cells transfected with one expression vector and one empty vector or two empty vectors. Results were displayed as firefly luciferase activity normalized to renilla luciferase activity and represented as means ± SEM (n = 4); (D) Schematic representation of the Mammalian Two-Hybrid System. The pG5luc Vector contains five GAL4 binding sites upstream of a minimal TATA box, which in turn is upstream of the firefly luciferase gene. The interaction between the two test proteins, expressed as GAL4-X and VP16-Y fusion constructs. An interaction between proteins X and Y brings the VP16 and GAL4 domains into close proximity, and results in an increase in luciferase expression over the negative controls; (E) The mRNA levels of Wnt6 and Wnt10b at non-induction, two days and four days post-adipogenic induction after treatment with pcDNA3.1-Zfp217 or empty pcDNA3.1 vector (n = 3); and (F) The mRNA levels of Cebpa and Pparg2 at four days post-adipogenic induction after treatment with pcDNA3.1-Zfp217 or empty pcDNA3.1 vector/Zfp217 siRNA or siRNA control (n = 3). Reverse transcriptase-quantitative polymerase chain reaction (qRT-PCR) data were standardized with β-actin and represented as means ± SEM. * p < 0.05, ** p < 0.01 versus control. Five microscopic fields were randomly selected. EdU positive cells (%) were quantified by Image J, which was calculated by the formula: EdU positive cell (100%) = red dots/(red dots + blue dots) × 100%. All data were obtained from three reproducible experiments; (C) In the mammalian two-hybrid system, the Zfp217-Ezh2 interaction is confirmed by significantly higher luciferase activity in cells transfected with pACT-Zfp217 and pBIND-Ezh2, compared to cells transfected with one expression vector and one empty vector or two empty vectors. Results were displayed as firefly luciferase activity normalized to renilla luciferase activity and represented as means ± SEM (n = 4); (D) Schematic representation of the Mammalian Two-Hybrid System. The pG5luc Vector contains five GAL4 binding sites upstream of a minimal TATA box, which in turn is upstream of the firefly luciferase gene. The interaction between the two test proteins, expressed as GAL4-X and VP16-Y fusion constructs. An interaction between proteins X and Y brings the VP16 and GAL4 domains into close proximity, and results in an increase in luciferase expression over the negative controls; (E) The mRNA levels of Wnt6 and Wnt10b at non-induction, two days and four days post-adipogenic induction after treatment with pcDNA3.1-Zfp217 or empty pcDNA3.1 vector (n = 3); and (F) The mRNA levels of Cebpa and Pparg2 at four days post-adipogenic induction after treatment with pcDNA3.1-Zfp217 or empty pcDNA3.1 vector/Zfp217 siRNA or siRNA control (n = 3). Reverse transcriptase-quantitative polymerase chain reaction (qRT-PCR) data were standardized with β-actin and represented as means ± SEM. * p < 0.05, ** p < 0.01 versus control.
To verify whether those miRNAs affected the Zfp217 expression level, we transiently transfected C3H10T1/2 cells with miRNAs mimics and investigated Zfp217 mRNA level at 24 h after transfection and protein level at 48 h after transfection. The results of reverse transcriptase-quantitative polymerase chain reaction (qRT-PCR) showed that over-expression of miRNAs (miR-1a-3p, miR-503-5p, miR-135a-5p and miR-19a-3p) dramatically suppressed the mRNA levels of Zfp217 (p < 0.01) ( Figure 5F), which was similar to western blot results ( Figure 5E). Although miR-1a-3p could restrain Zfp217 expression level, it could not bind the 3 UTRs of Zfp217 by "seed region". On the contrary, miR-26a-5p could bind the 3 UTRs of Zfp217, but it did not affect the Zfp217 expression level.
To verify whether those miRNAs affected the Zfp217 expression level, we transiently transfected C3H10T1/2 cells with miRNAs mimics and investigated Zfp217 mRNA level at 24 h after transfection and protein level at 48 h after transfection. The results of reverse transcriptase-quantitative polymerase chain reaction (qRT-PCR) showed that over-expression of miRNAs (miR-1a-3p, miR-503-5p, miR-135a-5p and miR-19a-3p) dramatically suppressed the mRNA levels of Zfp217 (p < 0.01) ( Figure 5F), which was similar to western blot results ( Figure 5E). Although miR-1a-3p could restrain Zfp217 expression level, it could not bind the 3′ UTRs of Zfp217 by "seed region". On the contrary, miR-26a-5p could bind the 3′ UTRs of Zfp217, but it did not affect the Zfp217 expression level.  The brilliant blue ring represents the predicted miRNAs in human (Homo sapiens, hsa). The pink ring represents the predicted miRNAs in mouse (Mus musculus, mmu). The gray blue ring represents the predicted miRNAs in both human and mouse. Venn diagram was output to statistical result; (B) Conservation of the miRNA binding site in the Zfp217-3′ unstranslated (UTR) region. The miRNA seed match region is highlighted in red; (C,D) Two pmirGLO vector constructs, containing either the Zfp217-3′ UTR-wt or the Zfp217-3′ UTR-mut with corresponding miRNA seed region, were transfected into BHK-21 cells either alone or in combination with negative control or each miRNA mimics. Renilla luciferase activity was used to normalize firefly luciferase activity. Data represents means ± SEM (n = 4); and (E,F) The expression levels of Zfp217 was measured by Western blot and qRT-PCR for different treated groups as indicated. The qRT-PCR data represents means ± SEM (n = 3). * p < 0.05, ** p < 0.01 versus control.
We also transiently transfected 3T3-L1 cells and C3H10T1/2 cells with the predicted miRNA mimics or negative control before adipogenic-induction to elucidate whether those miRNAs indeed The brilliant blue ring represents the predicted miRNAs in human (Homo sapiens, hsa). The pink ring represents the predicted miRNAs in mouse (Mus musculus, mmu). The gray blue ring represents the predicted miRNAs in both human and mouse. Venn diagram was output to statistical result; (B) Conservation of the miRNA binding site in the Zfp217-3 unstranslated (UTR) region. The miRNA seed match region is highlighted in red; (C,D) Two pmirGLO vector constructs, containing either the Zfp217-3 UTR-wt or the Zfp217-3 UTR-mut with corresponding miRNA seed region, were transfected into BHK-21 cells either alone or in combination with negative control or each miRNA mimics. Renilla luciferase activity was used to normalize firefly luciferase activity. Data represents means ± SEM (n = 4); and (E,F) The expression levels of Zfp217 was measured by Western blot and qRT-PCR for different treated groups as indicated. The qRT-PCR data represents means ± SEM (n = 3). * p < 0.05, ** p < 0.01 versus control.

Discussion
Despite its participation in a variety of important life processes due to its special structure, the role of the zinc finger protein family in adipogenesis is poorly understood. In our previous study, we found that a differentially expressed zinc finger protein gene Zfp217 was significantly upregulated at Day 2 during adipogenesis [17]. However, most of the previous studies are mainly focused on histone modification, pluripotency and oncogenesis [25][26][27]40,41], and its role in adipogenesis has not been reported. In this study, to test whether Zfp217 plays a role in adipogenesis, we retrieved GEO datasets and performed a bioinformatics analysis of the downloaded datasets. We found that the expression of Zfp217 is not only correlated with adipogenic marker genes, but also fluctuated with that of Pparg in three diabetes mellitus-related clinical statuses. Although the expression of Pparg is not obvious regularity in three clinical-status as the complex factors in vivo, the role of Pparg on obesity and diabetes mellitus is well known [42,43]. Therefore, we suspect that Zfp217 may have a potential effect on obesity and its role may be under the control of Pparg. Furthermore, bioinformatics analysis and literature review suggest that Zfp217 may affect adipogenesis by combining with Ezh2. Hence, a series of experiments were performed to test this hypothesis.
Previous studies have reported that Pparg acts as a transcriptional activator and has a central role in adipocyte differentiation, adipocyte metabolism and obesity [44]. Pparg is essential for both white and brown adipocytes differentiation under different conditions [42,45]. Ucp1 is a marker gene of brown adipocyte [46]. The ectopic expression of Ucp1 can increase the population of brown adipocytes [47], and Ucp1 can be also present in white adipose tissue [48]. Interestingly, in two types of white adipocytes (human SBGS pre-adipocyte and 3T3-L1 cells), Zfp217 and Pparg are classified together. In brown preadipocytes, Zfp217 and Ucp1 classified more closely. This may imply that the relationship between Zfp217 and Ucp1 is closer than Zfp217 and Pparg in brown preadipocytes. It suggests that Zfp217 may have different regulatory mechanisms in different type of adipocytes. The Together, the results showed that Zfp217 is directly targeted by miR-503-5p, miR-135a-5p and miR-19a-3p which impair adipocyte differentiation. This is a further evidence of Zfp217 has the effect on adipogenesis.

Discussion
Despite its participation in a variety of important life processes due to its special structure, the role of the zinc finger protein family in adipogenesis is poorly understood. In our previous study, we found that a differentially expressed zinc finger protein gene Zfp217 was significantly upregulated at Day 2 during adipogenesis [17]. However, most of the previous studies are mainly focused on histone modification, pluripotency and oncogenesis [25][26][27]40,41], and its role in adipogenesis has not been reported. In this study, to test whether Zfp217 plays a role in adipogenesis, we retrieved GEO datasets and performed a bioinformatics analysis of the downloaded datasets. We found that the expression of Zfp217 is not only correlated with adipogenic marker genes, but also fluctuated with that of Pparg in three diabetes mellitus-related clinical statuses. Although the expression of Pparg is not obvious regularity in three clinical-status as the complex factors in vivo, the role of Pparg on obesity and diabetes mellitus is well known [42,43]. Therefore, we suspect that Zfp217 may have a potential effect on obesity and its role may be under the control of Pparg. Furthermore, bioinformatics analysis and literature review suggest that Zfp217 may affect adipogenesis by combining with Ezh2. Hence, a series of experiments were performed to test this hypothesis.
Previous studies have reported that Pparg acts as a transcriptional activator and has a central role in adipocyte differentiation, adipocyte metabolism and obesity [44]. Pparg is essential for both white and brown adipocytes differentiation under different conditions [42,45]. Ucp1 is a marker gene of brown adipocyte [46]. The ectopic expression of Ucp1 can increase the population of brown adipocytes [47], and Ucp1 can be also present in white adipose tissue [48]. Interestingly, in two types of white adipocytes (human SBGS pre-adipocyte and 3T3-L1 cells), Zfp217 and Pparg are classified together. In brown preadipocytes, Zfp217 and Ucp1 classified more closely. This may imply that the relationship between Zfp217 and Ucp1 is closer than Zfp217 and Pparg in brown preadipocytes. It suggests that Zfp217 may have different regulatory mechanisms in different type of adipocytes. The specific mechanism of Zfp217 in brown preadipocytes needs more detailed experiments to be verified. In this study, multiple mouse adipose tissues and cell lines were first used to analyze the expression patterns of Zfp217 in the adipogenic process or the obesity model. With the extension of adipogenic induction time, Zfp217 expression showed a steady increase, which was very similar to that of key adipogenic marker gene Pparg2. However, for the five kinds of different adipose tissues from HFD/NCD mice, the expression levels of Zfp217 and Pparg2 in brown adipose tissues were not consistent as those in visceral adipose tissues or white adipocytes, and this may have been due to many influencing factors in vivo, or other unknown regulatory mechanisms. Visceral adipocytes are crucial contributors to the burden of obesity and are more harmful to human health [49]. The mRNA level of Zfp217 and Pparg2 only had a similar pattern in cell adipogenesis and visceral fat, it may suggest that the role of Zfp217 on white adipogenesis and obesity is closely related to Pparg. What's more, Pparg has two transcription factor binding sites at the 5 UTR of Zfp217 sequence predicted using JASPAR datasets (Table S4), it implies that Zfp217 may be the downstream gene of Pparg. The expression of Pparg2 was not always increased in HFD group, suggesting that there was a protective mechanism against diet-induced weight gain. The protective mechanism may be induced by inflammation factors, DNA methylation or neuromodulation (as epididymal and perirenal tissues are richly endowed with nodose ganglions) [50][51][52]. Zfp217 structure and properties were in accordance with early key pro-adipogenic genes. For the three cell models with different adipogenic ability before or after induction, the expression pattern of Zfp217 was similar to that of early key pro-adipogenic genes. Meanwhile, the protein structure of KLF4 and ZNF395 also had zinc finger protein motifs. Furthermore, all four early key pro-adipogenic genes could be linked to cancer [29,[53][54][55]. Therefore, the results of expression pattern analysis further supported the results of bioinformatics analysis.
In addition, "gain-of-function" and "loss-of-function" experiments further confirmed that Zfp217 promoted adipogenesis in both C3H10T1/2 cells and 3T3-L1 cells. Consistent with this finding, Zfp217 significantly positively regulated pro-adipogenic core transcription factors Cebpa and Pparg2. It seems that the increase of the Zfp217 expression obviously decreased EdU incorporation, and vice versa. Since the cell cycle distribution is changed by Zfp217 overexpression, this phenomenon is a result of the inhibition of mitotic clonal expansion. Namely, Zfp217 could arrest C3H10T1/2 cells at G0/G1 phase and blocked DNA synthesis, and it turns out that Zfp217 overexpression decreased EdU incorporation. This suggests that Zfp217 regulates adipogenesis, probably by suppressing cell proliferation. In order to understand the potential molecular mechanism of Zfp217 in adipogenesis, we surveyed the reports concerning Zfp217 or adipogenesis in the past decades, and found that Zfp217 was bound to Ezh2 and participated in histone modification in MCF7, which is a breast cancer cell line [40]. Besides, Ezh2 represses Wnt signaling genes to facilitate adipogenesis as well [56]. Concordant with our previous speculation, Zfp217 was found to interact with Ezh2 according to a mammalian two-hybrid assay. Furthermore, Zfp217 overexpression was found to restrain the mRNA level of Wnt6 and Wnt10b, the downstream target genes of Ezh2 during the adipogenic differentiation process [56], both of which act as a negative molecular switch in governing adipogenesis [57]. Combining GEO analysis, we speculate that Zfp217 could cooperate with Ezh2 in facilitating adipogenesis.
We also found that three miRNAs (miR-503-5p, miR-135a-5p and miR-19a-3p) directly suppressed the expression levels of Zfp217 by bonding its 3 UTRs. Consistent with our findings, previous studies have also found that those miRNAs are related to adipogenesis. miR-503-5p inhibits adipogenesis by targeting bmpr1a in C3H10T1/2 cells [58], miR-135a-5p suppresses adipogenes by activating the canonical Wnt/β-catenin signaling in 3T3-L1 cells [59], and miR-19a-3p acts as a serum miRNAs biomarker for pancreatic cancer-associated new-onset diabetes mellitus [60] and it is also involved in C3H10T1/2 chondrogenic differentiation [61]. Based on our results and related reports, these three miRNAs impair adipocyte differentiation, and Zfp217 is directly targeted by them; this in turn confirmed that Zfp217 has an effect on adipogenesis.
Although our research remains to be advanced and improved upon, and although the role of Zfp217 on brown adipogenesis is still indistinct, we have verified that Zfp217 has a positive effect on white adipogenesis, through bioinformatics, literature mining and a series experiments. To the best of our knowledge, this is the first report showing that Zfp217 promotes white adipogenesis in both C3H10T1/2 cells and 3T3-L1 cells, and that Zfp217 may cooperate with Ezh2 in facilitating its adipogenesis or triggering proliferative defects in C3H10T1/2 cells. Also three miRNAs which impair adipocyte differentiation are proposed to directly target Zfp217 in the mouse background ( Figure 6C).

Analysis of Gene Expression Omnibus Series (GSE) Data
All relevant GSE data were retrieved from GEO datasets ( Table S1). The probe ID was converted into an Entrez Gene ID according to the Gene Expression Omnibus Platform (GPL) file. Adipogenesis-unrelated GSE samples were excluded. Differentially expressed genes were identified using the R package limma, with an Adj.P.Val < 0.05, and |log 2 Fold Change| > 1 selected as the threshold for screening DEGs. The ggplot2 package was used for the visualization of the results.

Animals
Healthy male Kun Ming (KM) mice were housed in specific pathogen-free facilities on a 12/12 h light/dark cycle. Animals (four weeks of age) were randomized into chow diet (n = 6, researcher diets D12450B 10 Kcal %) and high-fat diet (n = 6, researcher diets D12492 60 Kcal %) groups. Animals were fed for ten weeks. Age-matched male littermates were used for all experiments. Five kinds of adipose tissues used in this study were carefully dissected from five depots [interscapular, epididymal, inguinal, perirenal white adipose tissue (WAT), and interscapular brown adipose tissue (BAT)], and immediately frozen in liquid nitrogen and stored at −80 • C for gene expression analysis. All experiments were performed in accordance with relevant guidelines and regulations. All experimental protocols were approved by the Ethics Committee of Huazhong Agricultural University with the permit number No. 30700571 for this study.

Cell Culture and Adipocyte Differentiation
3T3-L1 cells and C3H10T1/2 cells were purchased from the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China), and cultured in Dulbecco's modified Eagle's medium (DMEM, Gibco) with 10% fetal bovine serum (FBS, Gibco, Gaithersburg, MD, USA). The cells were induced to differentiation by addition of MDII cocktail (0.5 mM 3-isobutyl-1-methylxanthine, 1 µM dexamethasone, 5 µg/mL insulin and 100 µM indomethacin) in 10% FBS medium at 2 d after reaching confluence. Two days after induction, the cells were refed with 10% FBS medium with 5 µg/mL insulin for two days, then maintained in 10% FBS medium until cells were fully differentiated. NIH3T3 cells and BHK-21 cells were also cultivated in 10% FBS medium.

Oil Red O Staining
The fully differentiated cells were rinsed twice with PBS, and then fixed with 4% formalin for 30 min at room temperature. Then the cells were rinsed with PBS and stained with filtered Oil Red O (Sigma, Saint Louis, MO, USA) for 30 min, followed by washing twice in water. Stained cells were visualized by light microscopy (Nikon, Tokyo, Japan).

Triglyceride Assay
The fully differentiated cells were rinsed twice with PBS, collected in saline solution, and then sonicated for homogenization. The concentrations of triglyceride (TG) in the lysates of cells were measured with the commercial kits (Applygen, Beijing, China). The content of triglycerides was normalized to the content of protein (µmol/mg protein) using the bicinchoninic acid (BCA) assay kit (Thermo Scientific, Waltham, MA, USA).

Quantitative Real-Time RT-PCR (qRT-PCR) Analysis
Total RNA was obtained by using TRIzol (Invitrogen, Carlsbad, CA, USA). The cDNA was synthesized using RevertAid™ First Strand cDNA Synthesis Kit (Thermo Scientific) according to the manufacturer's protocol. The qRT-PCR was performed in triplicate using IQ SYBR green Supermix (Bio-Rad, Hercules, CA, USA) on CFX384 (Bio-Rad). Relative quantification was calculated by the 2 −∆∆Ct method and normalized by β-actin. Primers are listed in Table S2.

Western Blotting
Proteins were detected using antibodies against Zfp217 (sc-67223, Santa Cruz, Dallas, TX, USA), Pparg2 (sc-22020, Santa Cruz), and β-actin (sc-47778, Santa Cruz). Total cell lysate was extracted using RIPA lysis buffer on ice. Equal amounts of proteins were separated by SDS-PAGE electrophoresis. The protein bands were transferred to a polyvinylidene difluoride (PVDF) membrane (Millipore, Boston, MA, USA). The PVDF membrane was blocked with 5% skimmed milk and incubated with primary antibodies. The results were visualized using horseradish peroxidase-conjugated secondary antibodies (Santa Cruz) and enhanced chemiluminescence.

EdU Cell Proliferation Assay
The cell proliferation assay was carried out using EdU according to the manufacturer's manual (RiboBio, Guangzhou, China). Briefly, cells (1 × 10 4 cells/cm 2 ) were cultured in 96-well plates and added final-concentration 50 µM EdU of each well for 4 h at 37 • C. Then, the cells were fixed using 4% paraformaldehyde for 30 min at room temperature and permeabilized with 0.5% Triton X-100 for 10 min. Next, the cells were washed with PBS and incubated with EdU staining solution for 30 min. Finally, the cells were re-stained with Hoechst 33,342 (200 µL per well) for 30 min. The cells were imaged under a fluorescent microscope (Nikon).

Mammalian Two-Hybrid Assay
Protein-protein interactions were assayed using the CheckMate™ Mammalian Two-Hybrid System (Promega, Madison, WI, USA). Briefly, BHK-21 cells were plated at a density of 1 × 10 5 cells/cm 2 in 48-well plates, cultured for 24 h and then co-transfected with the mixture of pACT-and pBIND-derived constructs or control vector, and a pG5luc reporter vector. The cells were passively lysed at 48 h post transfection. Firefly and renilla luciferase activities were measured using the Dual-Luciferase ® Reporter Assay System (Promega) and an EnSpire Multimode Plate Reader with two automated injectors (PerkinElmer, Norwalk, CT, USA).

The Prediction and Screening of miRNAs
The miRNAs which targeting Zfp217 were predicted by miRNAWalk2.0 at the specific cut-off criteria (SUM ≥ 6, Targetscan = True) [39]. It extracted data from the overlap-miRNAs both in human and mouse. Conservative-overlap miRNAs were screened out from the overlap-miRNAs using targetscan. Base on the targeting principle of miRNAs, the GO annotation analysis of the conservative-overlap miRNAs was performed using their reverse prediction program targetgenes and performed by DAVID v6.8 [62].

Dual-Luciferase Reporter Assay
The 3 UTR fragments of Zfp217 were amplified from the mouse genome. The corresponding mutant fragments were amplified by overlap extension PCR. The primers and mutant sites were as shown in Table S3 and Figure S4. The PCR fragments were cloned into pmirGLO vectors (Promega). All vectors were confirmed by sequencing. BHK-21 cells were plated at a density of 1 × 10 5 cells/cm 2 in 48-well plates and then co-transfected with of the 3 UTR-pmirGLO vectors (wt or mut) and the miRNAs mimics (The sequences were listed in Table S2) or negative control (GenePharma, Shanghai, China) using Lipofectamine 2000 (Invitrogen). The dual-luciferase activity was measured utilizing the Dual-Glo Luciferase Assay System (Promega). Firefly luciferase activity was normalized to the corresponding Renilla luciferase activity. Experiments were performed in quadruplicate wells of a 48-well plate and repeated at least three times.

Statistical Analysis
Data values were presented as mean ± SEM. Statistical analyses were performed using two-tailed Student's t-tests. p < 0.05 was considered as statistically significant.

Conclusions
In this study, we reported a new adipogenic activator Zfp217 and proposed a new mechanism of adipogenesis. The data presented in this study not only provide an insight into the adipogenesis regulatory networks, but also suggest that Zfp217 may be an important therapeutic target in obesity.

Conflicts of Interest:
The authors declare no conflict of interest.