Abstract
Regulation of adipose tissue formation by adipogenic-regulatory proteins has long been a topic of interest given the ever-increasing health concerns of obesity and type 2 diabetes in the general population. Differentiation of precursor cells into adipocytes involves a complex network of cofactors that facilitate the functions of transcriptional regulators from the CCATT/enhancer binding protein, and the peroxisome proliferator-activated receptor (PPAR) families. Many of these cofactors are enzymes that modulate the structure of chromatin by altering histone-DNA contacts in an ATP-dependent manner or by posttranslationally modifying the histone proteins. Here we report that inhibition of protein arginine methyltransferase 5 (Prmt5) expression in multiple cell culture models for adipogenesis prevented the activation of adipogenic genes. In contrast, overexpression of Prmt5 enhanced adipogenic gene expression and differentiation. Chromatin immunoprecipitation experiments indicated that Prmt5 binds to and dimethylates histones at adipogenic promoters. Furthermore, the presence of Prmt5 promoted the binding of ATP-dependent chromatin-remodeling enzymes and was required for the binding of PPARγ2 at PPARγ2-regulated promoters. The data indicate that Prmt5 acts as a coactivator for the activation of adipogenic gene expression and promotes adipogenic differentiation.
Differentiation of adipocytes is a complex process involving numerous transcriptional regulatory proteins. The adipogenic regulatory protein peroxisome proliferator-activated receptor (PPAR)γ is considered the most critical of the regulators because its absence precludes formation of all adipose tissue (1, 2). However, the CCATT/enhancer binding protein (C/EBP)α is also a critical regulator of white adipose tissue formation (3), and there is considerable evidence for cross-talk and cooperativity between these proteins in driving the expression of downstream target genes that are expressed during adipocyte differentiation (4, 5). The C/EBPβ and -δ proteins contribute to adipogenesis by driving the expression of C/EBPα and PPARγ (6, 7) and likely contribute to late stages of differentiation as well (e.g. insulin-sensitive glucose uptake) during adipocyte differentiation, although this role is not well defined (8, 9). Additional positive regulators include sterol-regulatory element-binding protein (SREBP) (10, 11), Krox20 (12), signal transducer and activator of transcription 5 (13–15), and members of the Klf family (16–18). The combined actions of these numerous transcription factors result in a complex transcriptional cascade that causes adipocyte differentiation and tissue formation (reviewed in Refs. 19 and 20).
The chromatin environment in which the genes encoding the adipogenic transcription factors and their target genes exist adds an additional layer of complexity to the regulation of adipogenic gene expression. Histone modifications and histone-modifying enzymes associated with silent or repressed genes must be removed to facilitate gene activation during adipogenesis, and there is considerable evidence for multiple mechanisms to alleviate gene repression caused by different members of the histone deacetylase family (21, 22). Similarly, lysine methylation marks associated with gene repression are replaced with histone lysine methylation marks associated with gene activation during the gene activation process via the opposing activities of histone lysine methylases and demethylases (23, 24). Histone H3 and H4 acetylation is also associated with the activation of adipogenic gene expression (25), as is the binding and chromatin-remodeling activity of the SWI/SNF family of ATP- dependent chromatin-remodeling enzymes (26–28). The evidence supports the idea of a dynamic interplay between adipogenic transcription factors and cofactors that modify and/or remodel chromatin structure at the onset of and during adipogenic differentiation.
Characterization of histone acetylation and lysine methylation in relation to gene regulation has been extensive. In contrast, characterization of arginine methylation and the enzymes that mediate this histone modification has been more limited. The protein arginine methyltransferase (Prmt) family of enzymes is divided into two classes. Type I Prmts mono- and asymmetrically dimethylate arginine residues on substrate proteins, whereas type II Prmt mono- and symmetrically dimethylate arginine residues on their substrate proteins (reviewed in Refs 29 and 30). Of particular interest to us are the Prmt5 (type II) and Prmt4/Carm1 (type I) family members. Both Prmt5 and Prmt4 function as coregulators for a variety of transcriptional activators and repressors (reviewed in Refs. 29, 31, and 32). Intriguingly, both Prmt5 and Prmt4/Carm1 can be isolated from different cell types in complex with the SWI/SNF ATP-dependent chromatin-remodeling enzymes, suggesting a possible functional relationship between these histone-modifying and nucleosome-remodeling enzymes (33–35). Both SWI/SNF enzymes and Prmt4/Carm1 have been shown to be required for adipocyte differentiation (26–28, 36, 37). However, the possibility of a role for Prmt5 in adipogenesis and its relationship to SWI/SNF enzyme function during adipogenesis has not been explored.
Here we demonstrate using multiple experimental systems that knockdown of Prmt5 inhibits adipogenic differentiation and adipogenic gene expression and that overexpression in one system accelerates adipogenic differentiation. We demonstrate that Prmt5 binds to regulatory sequences of the PPARγ2 gene, as well as to PPARγ2 target gene promoters, and is responsible for the presence of the arginine-dimethylated histones at these genomic locations. The function of Prmt5 at these promoters is manifested in its requirement for the binding of the SWI/SNF chromatin-remodeling enzyme to adipogenic gene promoters and for the binding of PPARγ2 to PPARγ2 target genes.
Results
Prmt5 levels are relatively constant during the differentiation of 3T3-L1 preadipocytes and C3H10T1/2 cells into adipocytes
3T3-L1 and C3H10T1/2 cells are commonly used to model adipogenesis in tissue culture because both cell types will differentiate into adipocytes upon reaching confluence and subsequent exposure to a differentiation cocktail consisting of dexamethasone, insulin, and 3-isobutyl-1-methylxanthine (IBMX) (38, 39). We first examined protein levels of Prmt5 in 3T3-L1 differentiation model. Prmt5 protein levels were unchanged from the onset of differentiation (time 0) through 48 h after addition of differentiation cocktail. Levels increased modestly between 48 and 72 h after cocktail addition and remained at that level through day 6. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) levels were monitored as a control (Fig. 1A). In contrast, differentiating C3H10T1/2 cells showed no change in levels between d 0 and d 1 of differentiation, a modest increase in Prmt5 levels between d 2 and d 4 after differentiation, and a return to the initial Prmt5 levels at d 5 to d 6 after differentiation (Fig. 1B). We conclude that differentiation does not induce a significant change in Prmt5 protein levels in either system.
Prmt5 levels as a function of time of 3T3-L1 cell differentiation (A) and C3H10T1/2 cell differentiation (B). Whole-cell extracts were prepared from the samples indicated and used for Western blots. GAPDH levels are presented as a control.
Prmt5 is required for adipogenesis and adipogenic gene expression in 3T3-L1 and C3H10T1/2 cells stimulated to differentiate
To address whether Prmt5 is required for adipogenic differentiation, we manipulated Prmt5 expression by small interfering RNA (siRNA) transfection in 3T3-L1 preadipocytes. Subconfluent 3T3-L1 preadipocytes were transfected with a control or a pool of Prmt5-specific siRNA and allowed to become confluent. The cells were exposed 48 h after confluence to media containing the adipogenic differentiation cocktail consisting of dexamethasone, insulin, and IBMX (d 0) and analyzed 7 d later. 3T3-L1 preadipocytes differentiated when untreated or when transfected with a scrambled sequence control siRNA. The frequency of differentiation, as measured by Oil Red O staining, in Prmt5 siRNA-transfected cells was significantly decreased (Fig. 2A). Similar results were obtained when three of the four siRNA molecules comprising the siRNA pool were tested individually (Supplemental Fig. 1 published on The Endocrine Society's Journals Online web site at http://mend.endojournals.org), which further validates the specificity of the observed phenotype. Western analysis of each sample indicated that the Prmt5 siRNA-transfected cells contained significantly reduced levels of Prmt5 protein (Fig. 2B). mRNA levels of representative adipogenic genes were analyzed by real-time PCR after 7 d of differentiation. Gene expression was quantified as specific mRNA level relative to the level of cyclophilin mRNA in each sample, and the transcript level in the scrambled sequence control siRNA-transfected cells was normalized to 1. Levels of PPARγ2 and C/EBPα were severely compromised in 3T3-L1 cells transfected with siRNA to Prmt5 (Fig. 2C). Additionally, mRNA levels of resistin, adiponectin, and leptin, which are adipose-derived hormones that are markers of differentiation (40, 41), were reduced in the Prmt5 siRNA-treated cells but not in the control siRNA-treated cells (Fig. 2C). Similar results were obtained for adipocyte protein 2 (aP2), also called fatty acid binding protein 4 (Fabp4), which is a carrier protein for fatty acids that is expressed in both adipose and macrophages (42). The down-regulation of PPARγ2 expression in Prmt5 siRNA-treated cells was confirmed by Western blot analysis (Fig. 2D). Collectively, the data indicate that Prmt5 promotes adipogenic gene expression and adipogenic differentiation.
Prmt5 is required for 3T3-L1 preadipocyte differentiation. 3T3-L1 cells were untreated or transfected with either scrambled siRNA or siRNA against Prmt5 and grown to 2-d postconfluence. Samples were harvested either before or 7 d after induction of adipogenesis and stained for Oil Red O uptake (panel A), probed for protein levels of Prmt5 and the control p85 subunit of phosphatidylinositol-3 kinase by Western blot (panel B), analyzed for mRNA levels from the genes encoding the indicated adipogenic proteins (panel C), and probed for protein levels of PPARγ2 and control GAPDH (panel D). The data in panels A, B, and D are representative of three independent experiments. The data in panel C represents the average of three independent experiments ± sd.
We also explored the role of Prmt5 in the differentiation of the C3H10T1/2 mesenchymal stem cell line. These cells can be differentiated along multiple lineages depending on the signaling pathways activated by different differentiation cocktails (43, 44). Adipogenic differentiation can be obtained when confluent C3H10T1/2 cells are exposed to the standard adipogenic differentiation cocktail, and differentiation efficiency can be enhanced by the addition of a PPARγ ligand such as the thiazolidinedione, troglitazone (45). C3H10T1/2 cells were infected with an empty retroviral vector, with a retroviral vector encoding a Prmt5 antisense construct previously used to generate stable fibroblast cell lines that exhibited reduced levels of Prmt5 (33, 34), or with a retroviral vector encoding a FLAG-tagged Prmt5 protein that permitted us to assess the effects of Prmt5 overexpression in this system. Infected cells were drug selected and permitted to reach confluence. Examination of Oil Red O staining at d 3 after differentiation in Prmt5 overexpressing cells in the presence of troglitazone, a PPARγ ligand, indicates more intense Oil Red O staining than in the vector-infected control, suggesting that Prmt5 overexpression enhances the rate and/or efficiency of differentiation (Fig. 3A). In contrast, cells expressing the Prmt5 antisense construct showed almost no staining, supporting the data in Fig. 1 that indicate a requirement for Prmt5 (Fig. 3A). Cells differentiated for 6 d were also analyzed by Oil Red O staining. As expected, vector-infected cells treated with differentiation cocktail showed a modest number of cells containing lipid droplets whereas vector-infected cells treated with differentiation cocktail plus troglitazone showed an increased number of differentiating cells that contained larger, more brightly stained lipid droplets (Fig. 3B, top row). Infection with a Prmt5 antisense construct resulted in a reduced number of lipid droplet containing cells compared with the vector-infected controls, both in the absence and presence of troglitazone (Fig. 3B, middle row). In contrast, cells overexpressing Prmt5 showed an increased number of lipid droplet-containing cells, again both in the absence and presence of troglitazone (Fig. 3B, bottom row). The data indicate that Prmt5 modulates the efficiency of C3H10T1/2 cell differentiation. Western blot analysis verified that Prmt5 levels were altered in these cells (Fig. 3C); however, unlike in 3T3-L1 cells, only modest changes in Prmt5 level could be achieved. Analysis of adipogenic gene expression upon reduction or overexpression of Prmt5 was consistent with the differentiation results. Western blots showed that Prmt5 knockdown reduced the accumulation of PPARγ2 protein whereas overexpression increased PPARγ2 protein levels (Fig. 3D). Analysis of PPARγ2 mRNA correlated with the changes in protein expression (Fig. 4A). C/EBPα levels were not nearly as dynamic, but did show a down-regulation with loss of Prmt5 (Fig. 4B), suggesting perhaps that C/EBPα is not nearly as sensitive to changes in Prmt5 as PPARγ2. The downstream target genes also showed patterns of regulation that indicate they are regulated by Prmt5. Specifically, adipogenic gene expression was reduced in cells expressing the Prmt5 antisense construct and was enhanced in cells overexpressing Prmt5 (Fig. 4, C–F), although for some genes, the presence of the PPARγ ligand negated the effect of Prmt5 overexpression. Nevertheless, these data further support the conclusion that Prmt5 is a positive regulator of adipogenesis.
Prmt5 regulates C3H10T1/2 cell differentiation into adipocytes. C3H10T1/2 cells were infected with the retroviral vector or with retroviruses expressing Prmt5 or a Prmt5 antisense vector. Two-day postconfluent C3H10T1/2 cells were differentiated with FCS or with insulin, dexamethasone, and IBMX (IDM), or with IDM plus troglitazone (IDM+T). A and B, At 3 d or 6 d after induction, the extent of adipogenesis was assessed by staining for lipid accumulation with Oil Red O. C, Western blot analysis for Prmt5 and the control GAPDH protein levels from 2-d postconfluent cells. D, Western blot assessing protein levels of PPARγ2 and the control p85 subunit of phosphatidylinositol-3-kinase at 7 d after differentiation. The data are representative of three independent experiments.
Prmt5 regulates gene expression in C3H10T1/2 cells differentiating into adipocytes. mRNA were extracted from infected C3H10T1/2 cells that were stimulated with FCS, IDM, or IDM+T. mRNA levels for C/EBPα (A), PPARγ2 (B), adiponectin (C), resistin (D), leptin (E), and aP2 (F) were normalized to the level of cyclophilin B mRNA in the sample and expressed relative to vector-infected cells differentiated with FCS alone, which was assigned a value of 1. The data represent the average of three independent experiments each run in triplicate ± sd. IDM, Insulin, dexamethasone and IBMX; IDM + T, IDM + troglitazone.
Prmt5 is required for the induction of adipogenic regulators and downstream target genes
To better understand the relationship between Prmt5 and the adipogenic-regulatory proteins driving differentiation, we examined Prmt5 function in fibroblast cells induced to differentiate by the introduction of C/EBPα or PPARγ2. Previously, NIH3T3 immortalized fibroblasts were used to create clonal cell lines stably expressing a Prmt5 antisense vector. These clones show reduced levels of Prmt5 mRNA and protein and have been previously characterized (33, 34, 46–48). To ascertain whether Prmt5 contributed to adipogenic conversion of fibroblast-based cell lines, independently derived clones 12 and 15 (c12 and c15), along with control NIH3T3 cells, were infected with retroviruses encoding the adipogenic regulators C/EBPα or PPARγ2 or with the empty viral vector. After drug selection, cells were permitted to become confluent for 48 h and were treated with a standard adipogenic differentiation cocktail consisting of dexamethasone, insulin, IBMX, and the PPARγ ligand, troglitazone. mRNA levels of C/EBPα, PPARγ2, and representative adipogenic genes were analyzed by real-time PCR after 6 d of differentiation.
We first considered the expression of C/EBPα and PPARγ2. Specific gene expression in each sample was normalized to the level of cyclophilin mRNA, as done previously. Because there is essentially no C/EBPα or PPARγ2 mRNA in NIH3T3 cells, we normalized the expression of the ectopically expressed C/EBPα or PPARγ2 in NIH3T3 cells to 1. No difference in relative levels of ectopically expressed C/EBPα or PPARγ2 mRNA in each sample was observed (Fig. 5, A and B; compare darkly shaded bars). Introduction of C/EBPα in the parental NIH3T3 cells induced expression of the endogenous C/EBPα and PPARγ2 genes (Fig. 5, A and B), in agreement with previous findings (28, 49). However, in the Prmt5 antisense cells, introduction of C/EBPα failed to induce either endogenous C/EBPα or PPARγ2 (Fig. 5, A and B; compare lightly shaded bars). Introduction of PPARγ2 induced the endogenous PPARγ2 gene, but not the C/EBPα gene in the parental NIH3T3 cells, as shown previously (28, 49, 50), but in the Prmt5 antisense cells, exogenous PPARγ2 did not induce the endogenous PPARγ2 gene. Thus induction of the endogenous C/EBPα gene by ectopic C/EBPα is Prmt5 dependent, and induction of the endogenous PPARγ2 gene by either C/EBPα or PPARγ2 is Prmt5 dependent. We also note that the PPARγ1 isoform was not expressed in any of the samples evaluated (data not shown).
Prmt5 is required for differentiation specific gene expression in C/EBPα- and PPARγ2-differentiated fibroblasts. NIH3T3 and c12 and c15 Prmt5 antisense cells were infected with an empty retrovirus or viruses expressing C/EBPα or PPARγ2 and differentiated in the presence of adipogenic cocktail for 7 d. mRNA was isolated and levels of C/EBPα (A), PPARγ2 (B), adiponectin (C), resistin (D), leptin (E), and aP2 (F) were quantified and normalized to the expression of the cyclophilin B gene. Fold changes are relative to vector-infected NIH3T3 cells, which was set at 1. Results are the average of three independent experiments each performed in triplicate ± sd.
Adiponectin, resistin, and leptin gene expression levels were highly induced in C/EBPα- and in PPARγ2-differentiated parental NIH3T3 cells (Fig. 5, C–E) but were not induced at all in the Prmt5 antisense cells. Each of the genes assessed requires PPARγ2 for activation during differentiation. Thus, in the C/EBPα-differentiated Prmt5 antisense cells, the observed results are entirely consistent with the inability of C/EBPα to induce the PPARγ2 regulator. However, in the PPARγ2-differentiated Prmt5 antisense cells, PPARγ2 is exogenously expressed. Thus the failure to activate adiponectin, leptin, and resistin in these cells is not an indirect consequence of the cells being deficient for PPARγ2. Instead, the results indicate that Prmt5 is required to promote activation of these PPARγ2 target genes.
Expression of aP2 was also examined. As with the other genes, C/EBPα- or PPARγ2- driven differentiation caused a large induction of aP2 expression that was not observed in the Prmt5 antisense cell lines (Fig. 5F). However, some aP2 mRNA was detected in the Prmt5 antisense cell lines in the vector-infected cells, but expression was not further induced by C/EBPα- or PPARγ2-mediated differentiation. aP2 is regulated by both PPARγ2 and by the SREBP1 (11). SREBP1 activates PPARγ protein through the production of endogenous ligand (51). Assessment of the SREBP1c isoform in Prmt5 antisense cells indicated that mRNA levels were elevated approximately 5-fold (data not shown). This result suggested an explanation for the expression of aP2 in the absence of C/EBPα or PPARγ2 and also raised the possibility that Prmt5 normally represses SREBP1 expression. However, neither 3T3-L1 cells treated with siRNA against Prmt5 nor C3H10T1/2 cells expressing the Prmt5 antisense vector showed an increase in SREBP1a or c isoform expression (data not shown). Furthermore, reduction of Prmt5 levels in AML12 mouse hepatocytes, WI38 and TIG lung lipofibroblasts, C2C12 myoblasts, or C2C12 myotubes also failed to alter SREBP1 mRNA levels, despite Prmt5 levels being reduced 2.5- to 10-fold (data not shown). We concluded that the potential derepression of SREBP1 expression upon Prmt5 knockdown is not a general phenomenon and did not investigate any further. In contrast, the lack of aP2 induction in differentiated Prmt5 antisense cells is consistent with the results for the other adipogenic genes assayed and supports the idea that Prmt5 is required for adipogenic gene activation during differentiation.
Prmt5 binds to and methylates target histones and promotes binding of ATP-dependent chromatin-remodeling enzymes at adipogenic genes
Because it is known that PPARγ2 is required for adipogenic gene expression (2, 52) and adipose-derived hormone gene expression was compromised even in cells expressing exogenous PPARγ2, we asked whether Prmt5 directly affects PPARγ2 target gene expression. To assess this, we performed chromatin immunoprecipitation experiments on C/EBPα-expressing parental cells exposed to complete differentiation cocktail for 6 d. Under these conditions, both C/EBPα and PPARγ2 are expressed (Fig. 5, A and B). Analysis of the adiponectin and resistin promoters in vector-infected and in C/EBPα-expressing NIH3T3 cells indicated that Prmt5 binding was highly induced in the C/EBPα-expressing cells (Fig. 6, A and B). A documented histone substrate for Prmt5 is H3R8 (33). Prmt5 binding correlated with significantly increased levels of dimethylated (diMe) H3R8. (Fig. 6, A and B). Brg1 is an ATPase that is a catalytic subunit for mammalian SWI/SNF chromatin-remodeling enzymes (53–55). Prior work has demonstrated that SWI/SNF enzymes are essential for adipogenic differentiation in culture (26–28). As expected, Brg1 binding was similarly induced at the adiponectin and resistin promoters in the C/EBPα-expressing cells (Fig. 6, A and B). As a control, the coding sequence of the constitutively active EF1α gene was amplified. No change in binding of any of these factors or modified histones was observed at this locus (Fig. 6C). Thus, C/EBPα-induced differentiation results in chromatin structural changes at adipogenic genes that include Prmt5 binding, arginine dimethylation of histone residues that are Prmt5 substrates, and binding of the SWI/SNF chromatin-remodeling enzyme. Analysis of Prmt5 antisense cells revealed that Prmt5 binding did not occur, in keeping with the reduction of Prmt5 expression, that arginine dimethylation of H3R8 did not occur, and that Brg1 binding was abrogated at these gene promoters (Fig. 6, A and B). The results also indicate that differentiation-induced dimethylation of H3R8 at these loci is Prmt5 dependent, as is Brg1 binding.
Prmt5, diMe-H3R8, and Brg1 bind to the promoters of genes activated during C/EBPα-mediated differentiation. Chromatin immunoprecipitation experiments were performed on C/EBPα- or vector-infected NIH3T3, c12, and c15 cells differentiated in the presence of the adipogenic cocktail for 7 d. DNA immunoprecipitated by antibodies against Prmt5, diMe-H3R8, and Brg1 was amplified and quantified by real-time PCR for the adiponectin promoter (A), resistin promoter (B), EF1α coding sequence (C), and aP2 promoter (D). Data are presented as fold enrichment over the amount of DNA amplified after immunoprecipitation by the control IgG. Results are the average of three independent experiments each analyzed in triplicate ± sd.
We also analyzed the aP2 promoter for interactions with Prmt5, diMe-H3R8, and Brg1. As observed for the adiponectin and resistin promoters, binding of Prmt5 and enrichment of diMe-H3R8 was abrogated in the Prmt5 antisense cells after C/EBPα induced differentiation (Fig. 6D), which reinforces the idea that dimethylation of this histone arginine residue is mediated by Prmt5. However, Brg1 binding to the aP2 promoter was largely unaffected in the Prmt5 antisense cells. As previously stated, aP2 expression is also controlled by SREBP1 (11). In other cell types, SREBP1 has been shown to target Brg1-based SWI/SNF enzymes as part of the transcriptional control of SREBP1 target genes (56). The data suggest that differentiation-mediated induction of Brg1 binding at the aP2 promoter in the absence of Prmt5 is due to targeting by SREBP1 and that Brg1 binding at the aP2 promoter is not sufficient to promote maximal levels of aP2 expression in the absence of Prmt5.
C/EBPα-mediated adipogenic differentiation of NIH3T3 cells results in the expression of PPARγ2. However, PPARγ2 expression was blocked in the C/EBPα-expressing Prmt5 antisense cells (Fig. 5B). We therefore reexamined factor binding at the regulatory sequences controlling the expression of the adiponectin, resistin, and aP2 genes in control and Prmt5 antisense cells ectopically expressing PPARγ2. As expected, the association of Prmt5 and diMe-H3R8 at each gene promoter was induced in PPARγ2-differentiated cells, and the binding of Prmt5 and diMe-H3R8 was blocked in the Prmt5 antisense cells (Fig. 7, A–C). The presence of PPARγ2 did not facilitate the binding of Brg1 at the adiponectin and resistin promoters (Fig. 7, A and B), and Brg1 binding at the aP2 promoter remained Prmt5 independent (Fig. 7C), as observed in the previous experiment. In the NIH3T3 control cells, PPARγ2 was bound to all three promoters (Fig. 7, A–C), as would be expected. However, PPARγ2 binding was not observed at any of the promoters in the Prmt5 antisense cells, suggesting that Prmt5 may be required for PPARγ2 binding to these promoters (Fig. 7, A–C). As a control for binding specificity, no factor binding was observed at the constitutively expressed EF1α gene (Fig. 7D).
Prmt5, diMe-H3R8, Brg1, and PPARγ2 bind to the promoters of genes activated during PPARγ2-mediated differentiation. Chromatin immunoprecipitation experiments were performed on PPARγ2- or vector-infected NIH3T3 and c15 cells differentiated in the presence of the adipogenic cocktail for 7 d. DNA immunoprecipitated by antibodies against Prmt5, diMe-H3R8, Brg1, or PPARγ2 was amplified and quantified by real-time PCR for the adiponectin promoter (A), resistin promoter (B), aP2 promoter (C), or EF1α coding sequence (D). Data are presented as fold enrichment over the amount of DNA amplified after immunoprecipitation by the control IgG. Results are the average of three independent experiments each analyzed in triplicate ± sd, except for the Brg1 chromatin immunoprecipitation, which are the results of two independent experiments, each analyzed in triplicate ± sd.
The expression data also indicated that activation of the PPARγ2 gene itself was dependent on Prmt5 (Fig. 5). Assessment of PPARγ2 promoter sequences at sequences previously shown to bind C/EBP factors, c-fos, and Brg1 (28, 57, 58) repeatedly failed to identify the presence of Prmt5 and diMe-H3R8, although Brg1 was present (Fig. 8, A and B), consistent with earlier work (28). A recent report, however, identified a region approximately 10 kbp upstream of the PPARγ2 transcription start site that was identified as an enhancer-like region based on the localization of acetylated H3K9 and diMe-H3K4 (59, 60). Analysis of this sequence revealed binding of Prmt5, diMe-H3R8, and Brg1, all in a Prmt5-dependent manner (Fig. 8, C and D). The data indicate that Prmt5 is required for the same histone modification and chromatin-remodeling enzyme binding at the PPARγ2 enhancer-like region as occurs at the promoters of the characterized PPARγ2 target genes. The significance of Prmt5, diMe-H3R8, and Brg1 binding to enhancers vs. promoter sequences is not understood; nevertheless, the expression of PPARγ2 correlates with the binding of Prmt5 and diMe-H3R8 at PPARγ2-regulatory sequences.
Differential binding of Prmt5, diMe-H3R8, and Brg1 to the promoter and upstream enhancer-like sequence at the PPARγ2 locus. Chromatin immunoprecipitation experiments were performed on C/EBPα, PPARγ2, or vector-infected NIH3T3, c12, and c15 cells differentiated in the presence of the adipogenic cocktail for 7 d. DNA immunoprecipitated by antibodies against Prmt5, diMe-H3R8, or Brg1 was amplified and quantified by real-time PCR for the PPARγ2 promoter (A and B) or sequences approximately 10 kb upstream of the PPARγ2 promoter (C and D) previously implicated as having enhancer-like properties (60). Data are presented as fold enrichment over the amount of DNA amplified after immunoprecipitation by the control IgG. Results are the average of three independent experiments each analyzed in triplicate ± sd.
Prmt5 binds to sequences upstream of adipogenic genes in retroperitoneal adipose
ChIP experiments for Prmt5, diMe-H3R8, and Brg1 were repeated in primary retroperitoneal adipose tissue isolated from 2 wk old mice. Robust binding to the PPARγ2 target genes adiponectin, resistin and aP2 was observed (Fig. 9A–C), while no binding was observed at the EF1α locus (Fig. 9D). As observed in tissue culture, Brg1, but not Prmt5 or diMe-H3R8, was present at the PPARγ2 proximal promoter (Fig. 9E), while both enzymes and the modified histone could be observed at the sequences located 10kB upstream of the PPARγ2 transcription start site (Fig. 9F). These results demonstrate that Prmt5, Brg1 and diMe-H3R8 exist at adipogenic sequences in vivo and highlight the physiological relevance of the binding experiments done in tissue culture cells.
Prmt5, diMe-H3R8, and Brg1 interactions with adipogenic sequences documented on tissue culture cells also occur in retroperitoneal adipose isolated from 2-wk-old mice. Chromatin immunoprecipitation experiments were performed as described in Materials and Methods. DNA immunoprecipitated by antibodies against Prmt5, diMe-H3R8, or Brg1 was amplified and quantified by real-time PCR for the adiponectin promoter (A), resistin promoter (B), aP2 promoter (C), EF1α coding sequence (D), PPARγ2 promoter (E), and sequences approximately 10 kb upstream of the PPARγ2 promoter (F) previously implicated as having enhancer-like properties (60). Skeletal muscle was used as a negative control. Data are presented as fold enrichment over the amount of DNA amplified after immunoprecipitation by the control IgG. Results are the average of three chromatin immunoprecipitation experiments performed on independent tissue isolates, each analyzed in triplicate ± sd.
Discussion
It is becoming increasingly appreciated that arginine methyltransferases play a vital role in the regulation of cellular gene expression (61–63). In this study, we show that Prmt5 is required for adipogenesis through activation of PPARγ2 and PPARγ2 target genes. In three cell models for adipogenesis, 3T3-L1 preadipocytes, C3H10T1/2 mesenchymal stem cells, and NIH3T3 fibroblasts expressing adipogenic-regulatory proteins, down-regulation of Prmt5 was associated with deficiencies in PPARγ2 expression as well as in the expression of the PPARγ2 target genes encoding aP2 and the adipose-derived hormone factors, adiponectin, leptin, and resistin. In addition, the overexpression of Prmt5 in C3H10T1/2 cells in the presence of adipose differentiation stimuli resulted in accelerated lipid accumulation. The knockdown of Prmt5 prevented adipogenesis even in the presence of ectopic C/EBPα or PPARγ2 in NIH3T3 cells. This result indicated that the absence of PPARγ2 target gene expression was not solely due to the absence of PPARγ2 and that Prmt5 is necessary for both PPARγ2 expression and expression of PPARγ2 target genes.
Chromatin immunoprecipitation showed that Prmt5 binding and Prmt5-dependent H3R8 methylation were induced at the adiponectin, aP2, and resistin promoters under standard differentiation conditions in tissue culture cells, and these proteins and modifications were also present at these loci in primary mouse adipose tissue. Prmt5 binding and methylation at proximal promoter sequences of target genes is consistent with results previously reported by us and others (46, 64–66). These data strongly suggest that Prmt5 functions directly at the regulatory sequences of these adipogenic genes and similarly implicates symmetric dimethylation of histones in the activation process. Although we found no evidence for direct binding of Prmt5 to the PPARγ2 promoter, Prmt5 binding and Prmt5-dependent H3R8 methylation were detected at a region approximately 10 kb upstream of the PPARγ2 start site, in a previously characterized region that undergoes dynamic histone acetylation and lysine methylation changes during the course of adipogenesis (59, 60). It is not clear why Prmt5 and Prmt5-directed histone methylation occurs at this upstream element as opposed to the proximal promoter region. Future efforts will further investigate this novel result.
Along with Prmt5 and diMe-H3R8, the chromatin-remodeling enzyme Brg1, the ATPase of the SWI/SNF chromatin remodeling complex that is required for activation of adipogenic gene activation during adipogenesis (27, 28), was present at adipogenic hormone-regulatory sequences after the onset of differentiation. In the absence of Prmt5, Brg1 binding was diminished. These results indicate that Prmt5 plays an essential role in promoting adipogenic differentiation by facilitating the binding of ATP-dependent chromatin-remodeling enzymes required for adipogenic gene expression. This is intriguing because it suggests that the histone methyltransferase may precede ATP-dependent remodeling at the regulatory sequences of the gene being activated. We also observed that Brg1 binding was independent of Prmt5 at the aP2 promoter. However, aP2 is also a target for the SREBP1 activator, which has previously been shown to interact with SWI/SNF in other cell types (56). We therefore suggest that targeting of Brg1 can occur via multiple mechanisms during adipogenic gene activation, although we note that Prmt5-independent SREBP1 targeting of Brg1 to all potential SREBP1 target genes is likely not universal, because there are reports that SREBP1 can activate the resistin and the adiponectin promoters (67, 68) yet Brg1 targeting to these promoters was not observed when Prmt5 levels were reduced (Figs. 6 and 7). Regardless, it is clear that Prmt5-independent mechanisms that promote Brg1 binding to target genes are not sufficient to permit adipogenesis in culture, or, in this specific case, activation of the aP2 gene. The data reinforce the importance of Prmt5 function and also suggest that Prmt5 has additional functions beyond facilitating ATP-dependent chromatin remodeling.
Our previous findings that Prmt5 promotes myoblast determination protein (MyoD)-induced muscle differentiation, which is also a Brg1-dependent process (46, 47), led us to speculate that Prmt5 may be serving a similar role in adipose. However, the work presented here demonstrates some significant differences from the results observed in skeletal muscle differentiation. Although Prmt5 is required for activation of the myogenin gene at early times of muscle gene expression during MyoD-induced differentiation (46, 47), it was not required for myogenic late gene expression mediated by the myogenin protein itself (47). In contrast, Prmt5 is required for the activation of both PPARγ2 expression and the expression of PPARγ2 target genes, suggesting that Prmt5 may play a broader role in adipogenic gene expression than during myogenic gene expression.
Nevertheless, a key similarity between the requirement for Prmt5 in the skeletal muscle and in the adipose differentiation models examined to date is that many Prmt5-dependent genes in both systems require the presence of Prmt5 at target gene loci in order for the Brg1 chromatin remodeling enzyme to bind to these promoter sequences. The similar findings in both differentiation systems suggest that facilitation of ATP-dependent remodeling enzyme function may be a common step in the activation of tissue-specific genes. The exact mechanism by which a histone methyltransferase facilitates the ATP-dependent chromatin-remodeling enzyme interaction with the chromatin remains unknown. One possibility is that arginine-methylated histones are better substrates for Brg1-based chromatin remodeling. Alternatively, Prmt5-mediated arginine methylation may facilitate one or more additional histone modifications, which may then lead to chromatin-remodeling enzyme interaction and function. A more speculative version of this model would suggest that arginine methylation by different Prmts might independently or cooperatively facilitate Brg1 binding and chromatin remodeling. The Carm1/Prmt4 type I arginine methyltransferase asymmetrically dimethylates H3R17 and H3R26 (69) and has been shown to be essential for brown adipose tissue formation and for activation of PPARγ2 target genes (36). In skeletal muscle and in skeletal muscle models for differentiation, Carm1/Prmt4 binds to myogenic genes expressed at late times of differentiation and in culture models is necessary for the activation of these genes via targeting of the Brg1-based SWI/SNF chromatin-remodeling enzyme (47). Thus there is precedence for the function of different Prmt enzymes converging on a common step in gene activation. Finally, it remains possible that the Prmt5 function derives from a scaffolding effect that cooperates with, or is independent of, its enzymatic function.
Histones are not the only Prmt5 substrates. Prmt5 can dimethylate components of the polymerase II-associated transcription machinery (70, 71), as well as the p53 transcription factor (72) and the methylated DNA binding protein 2 (73, 74). Other Prmt also have been shown to methylate transcription factors and transcriptional coregulatory proteins (reviewed in Ref. 62). Of particular interest is the finding that Prmt1, a type I arginine methyltransferase distinct from Carm1/Prmt4, can methylate receptor-interacting protein140 (75), a nuclear receptor coregulator (reviewed in Ref. 76). In addition to functioning as a corepressor molecule in the regulation of lipid and glucose metabolism in mature adipose, recent findings suggest that receptor-interacting protein 140 also works in the cytoplasm to regulate glucose transporter 4 trafficking and lipolysis (76–78).
Because both muscle and adipose are derived from the mesenchymal lineage, Prmt5 may be necessary for further differentiation and for differentiation-dependent gene expression changes in mesenchymal precursor cells. Whether Prmt5 also controls the differentiation of additional lineages derived from the mesenchymal stem cells (MSC) is unclear and under investigation. However, the ability of Brg1 to modulate MSC cell cycle and to promote aspects of adipocyte differentiation in MSC in the absence of adipogenic-signaling molecules (79) suggests that a role for Prmt5 is likely. In addition, recent work indicates a requirement for Prmt5 in glial cell differentiation (80), suggesting a broader role in tissue development. Intriguingly however, the function of Prmt5 as a transcriptional coregulator can be context dependent. Prmt5 acts as a corepressor of gene expression involved in growth control, which is consistent with a role in promoting terminal cell differentiation (33, 34, 65, 81, 82). In addition, Prmt5 has recently been shown to be required to maintain pluripotency in embryonic stem (ES) cells, suggesting that the enzyme suppresses differentiation in the ES cell context (83). This evokes comparison to the Brg1 chromatin-remodeling enzyme, a fraction of which can be found associated with Prmt5 (33, 34). Brg1 is required for the differentiation of many different cell lineages (84, 85) yet also has been shown to bind to the promoters of genes associated with pluripotency in ES cells, whereas loss of Brg1 in ES cells results in the loss of the ES cell phenotypes and down-regulation of pluripotency markers (85–88). The pluripotency requirement for Brg1 is reported to be coupled to a specific conformation of the SWI/SNF enzyme that is comprised of selected Brg1-associated subunits (85–87), that presumably is different than the makeup of Brg1 enzymes associated with promoting cell differentiation. How Prmt5 functions in both an activation and repression capacity in differentiation is unclear. As has been indicated for Brg1, association with different subsets of interacting proteins, perhaps including Brg1, could mediate different functions. In this scenario, Prmt5 could function as a repressor in the pluripotent state, but once differentiation has been initiated, it could promote lineage specification. Alternatively, the potential diversity of substrates could make Prmt5 function entirely dependent upon its immediate local environment; its intrinsic ability to symmetrically dimethylate diverse substrate proteins could be used for multiple purposes that encompass not only transcriptional activation and repression but also other cellular processes. This may likely be the case, given that additional Prmt5 substrates include spliceosomal proteins SmB/B, SmD1, SmD3, and Lsm4 (89, 90), the ribosomal protein S10 (91), isoforms of fibroblast growth factor 2 (92), and nucleolin (93). Thus Prmt5 may be a multifunctional cofactor in a diverse array of cellular processes. Considerable additional work will be required to distinguish between the multiple potential mechanisms for Prmt5 function in different gene-regulation events.
Materials and Methods
Cell culture
3T3-L1 cells, NIH3T3 cells, and Prmt5 antisense construct-expressing cell lines (33) were maintained in DMEM supplemented with 10% calf serum in 5% CO2. C3H10T1/2 mesenchymal stem cells were maintained in 10% fetal calf serum (FCS). For adipogenic differentiation of 3T3-L1 cells, 2-d postconfluent cells were differentiated using a standard adipogenic cocktail (1 μg/ml insulin, 0.25 μg/ml dexamethasone, 0.5 mm IBMX with 10% FCS). After 48 h, cells were maintained in medium containing 1 μg/ml insulin until harvest. C3H10T1/2 cells were also differentiated using the same above cocktail either with or without troglitazone (Calbiochem, La Jolla, CA) at 10 μm. Prmt5 antisense cell lines were maintained in 2.5 μg/ml puromycin. Cells were grown to 50% confluence and then differentiated into the adipogenic lineage by ectopic expression of vectors containing PPARγ2 or C/EBPα along with a blasticidin resistance marker. After blasticidin selection, cells were cultured to 2-d postconfluence and exposed to the differentiation cocktail (1 μg/ml insulin, 0.25 μg/ml dexamethasone, 0.5 mm IBMX, 10 μm troglitazone, in 10% FCS). After 48 h, cells were maintained in medium containing 1 μg/ml insulin until harvest.
Oil Red O staining
The differentiating cells were fixed with 10% phosphate-buffered formalin for 1 h. The cells were washed with PBS and 60% isopropanol (94). The cells were then stained with a working solution of 60% Oil Red O (60:40 stain-water) for 1 h and washed repeatedly with water to remove excess Oil Red O.
RNA isolation and analysis
Total RNA was isolated from samples using Trizol reagent (Invitrogen, Carlsbad, CA). Total RNA (1 μg) from each sample was used to prepare cDNA using Superscript III reverse transcriptase (Invitrogen). Quantitative RT-PCR was performed by monitoring in real time the increase in fluorescence of the SYBR Green dye using the Quantitect SYBR Green PCR kit (QIAGEN) or GoTaq SYBR Green kit (Promega Corp., Madison, WI) on the ABI StepOne Plus Sequence Detection System (Applied Biosystems, Foster City, CA). Relative amounts of each mRNA were determined using the Comparative Ct method (95) and normalized to the relative levels of cyclophilin B expression. Primer sequences are listed in Supplemental Table 1.
Chromatin Immunoprecipitation assays
Cultured cells were cross-linked in 1% formaldehyde, quenched with 0.125 m glycine, and lysed in buffer containing 1% sodium dodecyl sulfate (SDS), 10 mm EDTA, 50 mm Tris-HCl (pH 8.1) containing protease inhibitors (1 mm phenylmethylsulfonyl fluoride, 1 μg/ml aprotinin, 1 μg/ml pepstatin A). Samples were incubated on ice, and the DNA was sheared by sonication in a Bioruptor (Diagenode) to obtain an average length of 500 bp. A total of 100 μg of sonicated DNA was diluted 10-fold in immunoprecipitation buffer [0.01% SDS, 1.1% Triton X-100, 1.2 mm EDTA, 16.7 mm Tris (pH 8.1), 167 mm NaCl] containing protease inhibitors and precleared with a 50% slurry of protein A beads (Amersham Pharmacia Biotech, Piscataway, NJ) at 4 C for at least 1 h. Cleared lysates were incubated with polyclonal rabbit antisera against Prmt5 (34), diMe-H3R8 (33), Brg1 (96), or normal rabbit IgG (Millipore Corp., Bedford, MA) or a commercial antibody for PPARγ2 (sc-7273; Santa Cruz Biotechnology, Inc., Santa Cruz, CA) at 4 C overnight. Protein A beads were added to precipitate immune complexes from the cell lysates and incubated for 1 h at 4 C. Beads were collected by centrifugation and were washed sequentially for 5 min at 4 C with wash 1 [0.1% SDS, 1% Triton X-100, 2 mm EDTA, 20 mm Tris-HCl (pH 8.1), 150 mm NaCl], wash 2 (wash 1 containing 500 mm NaCl), and wash 3 [0.25 m LiCl, 1% Nonidet P-40, 1% sodium deoxycholate, 1 mm EDTA, 10 mm Tris-HCl (pH 8.1)], and finally were washed twice with Tris-EDTA (pH 8.0). Immune complexes were eluted from the beads with 1% SDS in Tris-EDTA (pH 8.0), and protein-DNA cross-links were reversed by adding 200 mm NaCl and heating at 65 C overnight. After treatment with proteinase K, the samples were purified with the QIAquick PCR purification kit (QIAGEN). Analysis of immunoprecipitated DNA was performed by quantitative PCR (QPCR) using a Quantitect SYBR Green PCR kit (QIAGEN) or GoTaq SYBR Green kit (Promega Corp., Madison, WI). Primer sequences are listed in Supplemental Table 1.
Retroperitoneal adipose tissue and skeletal muscle from the upper hind limbs were isolated from 2-wk-old wild-type mice on a mixed 129SV/J-C57BL/6 background, minced, and treated with type II collagenase (Sigma Chemical Co., St. Louis, MO) in PBS supplemented with 1 mm CaCl2 for 1 h with agitation at 37 C. Cells were centrifuged through a 40-μm cell strainer to remove the stromal vascular fraction, and the remaining mature adipocyte population was homogenized and prepared for chromatin immunoprecipitation as described above.
Protein analysis
Day 0 and d 6 C3H10T1/2 and d 0 and 7 3T3-L1 cells in 10-cm plates were scraped into 1 ml of PBS. The cell pellets were snap frozen in liquid N2 and stored at −80 C. The pellets were resuspended in 200 μl of lysis buffer containing 50 mm Tris-HCl (pH 7.5), 150 mm NaCl, 0.5% Nonidet P-40, and 20% glycerol, along with Sigma protease inhibitor cocktail. After sonication, the protein amounts were quantified using a Bradford assay. Protein (50 μg) was loaded on an SDS-PAGE gel and transferred to a nitrocellulose membrane. Primary antibodies were Prmt5 (Santa Cruz; sc-22132) and GAPDH (Sigma; G9295), PPARγ2 P0744 (Sigma), and p85 phosphatidylinositol-3-kinase (Millipore). Bands were detected by secondary conjugated horseradish peroxidase antibodies and enhanced chemiluminescence detection.
3T3-L1 siRNA transfection
Before transfection, the 3T3-L1 cells were grown in antibiotic-free medium. Cells were transfected at 60–70% confluence using Lipofectamine 2000 (Invitrogen) with 1 μm of siRNA (scrambled sequence control, a smart pool of Prmt5 siRNA, or individual siRNA from the Smartpool) for 6 h. Cells were induced to differentiate 48 h after transfection. The cells were harvested for protein, mRNA, and Oil Red O staining 7 d later. Cells transfected only with Lipofectamine alone (no siRNA) served as controls. siRNA sequences are listed in Supplemental Table 1.
Acknowledgments
We thank C. Mallappa (Department of Cell Biology, University of Massachusetts Medical School) for technical assistance and advice. We thank R. Mudhasani (Department of Cell Biology, University of Massachusetts Medical School) for assistance and advice in isolating mature mouse adipocytes.
This work was supported by National Institutes of Health Grants R01DK084278 (to A.N.I. and S.S.), by R21DK079239 (to A.N.I.) and by F32DK082263 (to S.E.L.). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the National Institutes of Health. A.N.I. is a member of the University of Massachusetts Diabetes Endocrinology Research Center, which is supported by National Institute of Health (NIH) Grant 5P30DK32520.
Disclosure Summary: The authors have nothing to declare.
Abbreviations
- aP2
Adipocyte protein 2
- C/EBP
CCATT/enhancer binding protein
- diMe-H3R8
dimethylated H3R8
- ES
embryonic stem
- FCS
fetal calf serum
- GAPDH
glyceraidehyde-3-phosphate dehydrogenase
- IBMX
3-isobutyl-1-methylxanthine
- MSC
mesenchymal stem cells
- PPAR
peroxisome proliferator-activated receptor
- Prmt
protein arginine methyltransferase
- SDS
sodium dodecyl sulfate
- siRNA
small interfering RNA
- SREBP
sterol-regulatory element-binding protein.
