Small Molecule-Induced Complement Factor D (Adipsin) Promotes Lipid Accumulation and Adipocyte Differentiation

Adipocytes are differentiated by various transcriptional cascades integrated on the master regulator, Pparγ. To discover new genes involved in adipocyte differentiation, preadipocytes were treated with three newly identified pro-adipogenic small molecules and GW7845 (a Pparγ agonist) for 24 hours and transcriptional profiling was analyzed. Four genes, Peroxisome proliferator-activated receptor γ (Pparγ), human complement factor D homolog (Cfd), Chemokine (C-C motif) ligand 9 (Ccl9), and GIPC PDZ Domain Containing Family Member 2 (Gipc2) were induced by at least two different small molecules but not by GW7845. Cfd and Ccl9 expressions were specific to adipocytes and they were altered in obese mice. Small hairpin RNA (shRNA) mediated knockdown of Cfd in preadipocytes inhibited lipid accumulation and expression of adipocyte markers during adipocyte differentiation. Overexpression of Cfd promoted adipocyte differentiation, increased C3a production, and led to induction of C3a receptor (C3aR) target gene expression. Similarly, treatments with C3a or C3aR agonist (C4494) also promoted adipogenesis. C3aR knockdown suppressed adipogenesis and impaired the pro-adipogenic effects of Cfd, further suggesting the necessity for C3aR signaling in Cfd-mediated pro-adipogenic axis. Together, these data show the action of Cfd in adipogenesis and underscore the application of small molecules to identify genes in adipocytes.


Introduction
Adipocytes, energy reservoirs in vertebrate organisms, have the machinery required for lipolysis, glucose uptake, and triglyceride synthesis [1,2]. Adipocytes are also an endocrine organ that secretes fatty acids and adipokines to regulate systemic metabolism upon various stimuli including hormones, feeding, and exercise [3]. Numerous studies have identified regulators including Pparγ in adipocytes. Pparγ is necessary and sufficient for adipogenesis [4]. Pparγ upstream regulators including CCAAT/enhancer binding protein (C/ebp), Early B cell factor, Interferon regulatory factor, Kruppel -like factor, early growth response, Inhibitor of DNA binding, Zinc finger proteins, and its downstream target genes have shown for their roles in lipid synthesis and adipocyte differentiation [5,6,7]. As Pparγ would be regulated by various integrated pathways, identification of a broader window of Pparγ upstream and its target genes linked to adipogenesis would provide further understanding of Pparγ and adipocyte biology.
Complement generated mainly by hepatocytes functions within innate immune defense [8,9]. A recent study showed that the adipose tissue produces complement component 3 (C3), complement factor B (Cfb), and complement factor D (Cfd, also called adipsin), factors required for the production of C3a [10]. Most of the other complement factors are also secreted by the adipose tissue, suggesting non-immunogenic roles of complement in adipose tissues. Indeed, C3a is recognized by C3aR, which triggers triglyceride synthesis and its levels are a risk factor for developing diabetes [11,12,13,14]. Acylation stimulating protein (ASP, C3adesArg), a C3 cleavage product by exopeptidase activity, binds to C5L2 and also promotes triglyceride synthesis [15]. Thus, complement factors including Cfd in immune cells and adipocytes may be the mediators for interweaving energy metabolism and immune responses.
We have previously identified various pro-adipogenic small molecules. Harmine was identified as a Pparγ inducer that promoted adipocyte differentiation of preadipocytes and increased insulin sensitivity without any signs of deleterious effects associated with Pparγ activation in diabetic mice [16]. Similarly, another small molecule phenamil has been shown to induce tribbles-like 3, resulting in stimulation of Bone morphogenetic protein signaling and Pparγ induction [17,18]. These studies indicate that small molecule Pparγ inducers may be a useful strategy to delay the onset of obesity-related metabolic diseases.
In this study, a previously developed cell-based high throughput screening assay was performed and it led to the identification of new pro-adipogenic small molecules. We chose and utilized three small molecules as tools to identify adipogenic genes. Transcriptional profiling assays of these selected small molecules revealed potential new adipogenic regulators including Pparγ, Cfd, Ccl9, and Gipc2. Among these, Cfd was originally described as a adipocyte-specific gene, but its role in adipocyte differentiation remains unknown [19]. Hence, we further focused on Cfd to investigate its role in adipogenesis. We showed that Cfd induced by adipogenic small molecules exerts pro-adipogenic effects and its actions can be mediated by activating C3a-C3aR pathways. These studies provide evidence for previously unknown roles of Cfd in adipocyte differentiation and further highlight the utility of pro-adipogenic small molecules in identification of the genes involved in adipocyte biology.

Gene expression analysis
Microarray was performed as described previously [17,18]. Total RNAs from 24 hour pro-adipogenic small molecule treated F442A cells were extracted using TRizol reagent (Invitrogen, Carlsbad, CA) and further purified using RNAeasy columns (QIAGEN, Chatsworth,CA). cDNA preparation and hybridization to Mouse-6 expression Beadchip were performed by the UCLA core facility.
Male 7 week olds C57BL/6 mice were purchased from Central Lab Animal Inc. (Seoul, Korea). After 1 week of adaption, the mice were divided into two groups. The mice in one group were fed a normal diet (ND, n = 8) and the other group (n = 8 per group) were fed a high-fat diet containing 60% calories from fats (HFD; D12492, Research Diets Inc., New Brunswick, NJ, USA) and individually housed in a temperature-controlled room with a 12-hour light/dark cycle for 8 weeks. The mice were euthanized and adipose tissues were dissected to isolate RNAs. Homogenized adipose tissues with TRIzol were centrifuged to remove lipid layer. The isolated total RNAs were reversely transcribed and the synthesized complementary DNA was used for realtime PCR analysis. Mice were euthanized by carbon dioxide inhalation and immediately exsanguinated by cardiac puncture. Animal studies were carried out in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Committee on the Ethics of Animal Experiments of the Sungkyunkwan University (Permit Number: SKKUIACUC-20150037).

Knockdown studies
The scrambled control and Cfd-specific shRNA sequences were synthesized by Genolution Pharmaceuticals, Inc. (Seoul, Korea). To generate stable cells, shRNAs were cloned into pRe-troSuper and Phoenix E packaging system was used as described [23]. The sense sequences of two independent Cfd-specific and C3aR-specific shRNAs were as follows: Cfd sh #1: 5 0 -GCU AAUGGUACCUCUUUCAUU -3 0 , Cfd sh #2: 5 0 -CAUUGAUGACACAUUCU CUUU -3 0 . For C3aR silencing, siRNAs were transfected for 12 hours using Lipofectamine RNAiMAX (Invitrogen). A concentration of 30 nM siRNAs was transfected into 80% confluent 3T3-L1 cells. The cells were replenished with fresh medium after 12 hours and 48 hours later the cells were differentiated into adipocytes. Transfection was repeated at least three times. The sense sequence of the control, nonspecific scrambled RNA was 5 0 CCUCGUGCCGUUCCAUCAG GUAGUU -3 0 . The sense sequences of two independent C3aR-specific siRNAs (shRNAs) were as follows: Overexpression studies HEK293T cells were transfected for 48 hours with pBabe-puro or pBabe-puro-Cfd and retroviral packaging vectors [21]. Polybrene (8 ng/ml, Sigma) was further added and viral supernatants were harvested and filtered. C3H10T1/2 or 3T3-L1 cells were exposed to viruses for 24 hours and then stable cells were selected by puromycin treatment for 2 weeks and differentiated into adipocytes.

Statistical analysis
Data are presented as the mean ± SEM. Statistical significance differences were determined using one-way ANOVA and the Student-Newman-Keuls test. Differences in gene expression and lipid accumulation in two groups were analyzed using a two-tailed unpaired Student's ttest. Statistical significance was defined as P < 0.05.

Complement factor D (Cfd) is induced by pro-adipogenic small molecules
Three new adipogenic small molecules were identified by a previously developed cell-based high throughput screening assay [16]. To investigate the transcriptional profiling, 3T3-F442A cells were treated with selected small molecules or a Pparγ agonist GW7845 for 24 hours and microarray analysis was performed ( Figure B in S1 File, Fig 1A and 1B). Direct Pparγ target genes were excluded by comparing with the list of genes induced by GW7845. Numerous genes specific to a single compound and overlapped genes induced by at least two compounds were identified (Fig 1C). We hypothesized that specific genes would be more related to the unique mechanism of each compound, whereas genes commonly induced by three small molecules could be convergent genes acting on promoting Pparγ expression and adipocyte differentiation. As we were interested in identifying new players in adipocyte differentiation and Pparγ inducers, we focused on the overlapped genes. Only four genes were regulated by at least two different small molecules. Pparγ was induced by all three compounds but not by GW7845 treatments for 24 hours, further verifying the approaches. Other candidate genes included Ccl9 induced by PT7 and PT26, Gipc2 induced by PT7 and PT24, and Cfd induced by PT24 and PT26 (Fig 1C and 1D). We next investigated the expression profiles of these genes during adipocyte differentiation of 3T3-L1 cells and in mice. Similar to the aP2 and Pparγ expression profiles, Cfd expression was increased during adipocyte differentiation. Ccl9 expression was promptly decreased at day 2 and then moderately increased during adipogenic processes (Fig 2A). Unlike expression patterns in 3T3-F442A cells, Gipc2 expression was not detectable in 3T3-L1 cells. In tissues, Cfd expression was adipose tissue selective with highest expression in epididymal fat (eWAT). Ccl9 was also highly expressed in adipose tissues. Gipc2 expression was high in kidney, lung, and intestine, but it was almost excluded from fat tissues (Fig 2B). The expression profiles of Ccl9 and Cfd in adipocytes are consistent with previous data [24,25]. Ccl9 expression was increased in eWAT and brown adipose tissues (BAT) from HFD fed obese mice compared to control lean mice. Cfd expression was defective in obese mice (Fig 2C and 2D). We focused on Cfd at this stage to further investigate its roles in adipocytes.

Complement factor D (Cfd) promotes adipocyte differentiation
To test whether Cfd can stimulate lipid accumulation, we stably expressed retrovirus harboring shRNAs targeting Cfd in C3H10T1/2 cells and induced to differentiate into adipocytes. As expected, Cfd expression was lower in Cfd-shRNA cells compared to control cells (Fig 3). Two independent shRNA-expressing cells showed suppressed expression of Pparγ and its target genes during adipocyte differentiation compared to control (scr) cells (Fig 3). To further confirm this, we stably overexpressed Cfd in 3T3-L1 cells using pBabe retroviral system. Cfd stably overexpressing retrovirus infected cells (pBp-Cfd) compared to empty vector expressing virus infected cells (pBp) showed increased lipid accumulation (Fig 4A). In addition, adipocyte markers were increased in Cfd overexpressing cells (Fig 4B). Consistent effects by Cfd overexpression in C3H10T1/2 cells were also observed (Figure C in S1 File). Taken together, loss and gain of function studies can assign the role of Cfd during adipocyte differentiation. Silencing of Cfd suppresses adipocyte differentiation. Stable C3H10T1/2 cells expressing two independent shRNAs against Cfd or scrambled shRNA were generated. Stable cells were differentiated into adipocytes for 8 days and expression of Cfd, Pparγ, C/ebpα, and aP2 was measured by real time PCR analysis. Data are presented as means +/-s.e.m. and are representative of two independent experiments. Each independent experiment was carried out in triplicate. Statistical significance differences were determined using one-way ANOVA and the Student-Newman-Keuls test. Different letter at each samples are significant (P <0.05)

Complement factor D induces C3aR activation
Cfd is a serine protease secreted by adipocytes and it cleaves factor B, resulting in increased C3a and C3b production [9]. C3a has also been shown to promote triglyceride synthesis in adipose tissue [26]. Based on these, we reasoned that induced levels of C3a by Cfd can be responsible for increases in lipid accumulation and adipocyte differentiation. To test this possibility, we measured C3a production in media from control (pBp) and Cfd overexpressing 3T3-L1 preadipocytes (pBp-Cfd). C3a production was three times higher in pBp-Cfd cells compared to control pBp cells (Fig 5A). These data show the possibility that increased C3a production, at least in part, can mediate the pro-adipogenic effects in Cfd expressing cells.

C3aR signaling stimulates adipocyte differentiation
To further directly investigate the role of C3a in adipocytes, we treated 3T3-L1 cells with C3 or C3a and induced to differentiate into adipocytes. Similar to the effects by Cfd, treatment with either C3 or C3a stimulated lipid accumulation (Fig 5B). Furthermore, a C3aR (C3a receptor) agonist C4494 also increased lipid accumulation and expression of adipocyte markers comparable to the levels in Cfd overexpressing 3T3-L1 cells (Fig 5C and 5D).
Since C3a can exert its effects through interaction with the receptor C3aR, we investigated the role of C3aR in adipocytes. To test the actions of C3aR in adipogenesis, we silenced C3aR using two independent shRNAs and assessed the effects on adipocyte differentiation. In C3aR deficient C3H10T1/2 cells, lipid accumulation was decreased compared to scrambled shRNA transfected control (scr) cells (Fig 6A). Expression of C3aR and adipocyte markers was Complement Factor D Promotes Adipogenesis consistently decreased in the C3aR deficient cells (Fig 6B-6D). Furthermore, C3aR expression was also necessary for adipocyte differentiation in 3T3-L1 cells ( Figure D in S1 File), exhibiting that the C3aR signaling promotes lipid accumulation and adipocyte differentiation.

C3aR is required for Cfd action in adipocytes
To test whether Cfd can promote adipocyte differentiation by acting on C3aR signaling, we overexpressed Cfd and concomitantly silenced C3aR expression in 3T3 preadipocytes. Interestingly, increased expression of Pparγ in Cfd overexpressing cells (pBp-Cfd) was significantly impaired in the C3aR-deleted cells (C3aR si#1 and #2). Induction of Pparγ target genes including C/ebpα, aP2, Cd36, and adiponectin (adipoQ) by Cfd overexpression was also blunted in C3aR-silenced cells (Fig 7A). These data suggest the requirement of C3aR signaling, at least in part, for Cfd-mediated pro-adipogenic effects.
Based on the possibility that C3aR signaling is crucial for the effects of Cfd, we treated pBp-Cfd and pBp cells with C3aR agonist (C4494) and compared the adipogenic potentials. C3aR- activated cells showed higher levels of lipid accumulation compared to DMSO-treated cells. C3aR activation in pBp-Cfd cells, however, failed to further induce lipid accumulation. Similarly, the expression levels of adipocyte markers (aP2, C/ebpα, and Cd36) were higher in C3aR activated pBp cells, but C3aR activation in pBp-Cfd cells did not further produce synergistic induction of adipocyte genes (Figure E in S1 File). These data show the functional importance of C3aR signaling in Cfd-mediated pro-adipogenic effects (Fig 7B).

Discussion
Previously, bioactive small molecules were utilized to drag out critical players in cell biology [27]. MyoD in myogenesis and Pparγ in glucose metabolism were previously revealed by investigating the molecular actions of small molecules [28,29,30]. Similarly, pro-adipogenic small molecules can also be used as a tool to discover new players in biological processes. Toward this end, we utilized a previously developed cell-based high throughput screening assay to identify pro-adipogenic small molecules. Top 35 small molecules (top 0.02%) were initially selected to test their effects on promoting adipogenesis and Pparγ expression in 3T3-F442A cells. From these candidates, we further selected three most potent pro-adipogenic compounds (named PT7, PT24, and PT26). Molecular signatures of these selected Pparγ inducers were analyzed and these led to the identification of four genes (Pparγ, Cfd, Ccl9, and Gipc2). Although we investigated the previously unknown adipogenic roles of Cfd in the current study, it is also possible that other genes such as Ccl9 and Gipc2 can be potential candidates for Pparγ induction as shown in impaired adipocyte differentiation in Ccl9-silenced cells (Figure F in S1 File). As Pparγ upstream factors such as CDK5 are considered as new molecular targets for insulin resistance and other metabolic diseases, mechanism studies of these compounds may provide new molecular targets for obesity and obesity-related metabolic diseases [31,32].
Cfd is a differentiation dependent serine protease secreted by adipocytes and its expression is deficient in obese and diabetic disease mice models [33,34,35]. Furthermore, recent studies show that Cfd stimulates insulin secretion and maintains β cell function through C3a-C3aR signaling [36]. Cfd is also involved in triacylglycerol synthesis in human adipocytes and C3a levels are a risk factor for developing diabetes [11,12,13,14]. An association between familial C3 deficiency and partial lipodystrophy [37] also suggests the roles of complement system in pathological conditions associated with obesity and its related metabolic diseases [33,34]. In this study, we showed that Cfd promoted adipocyte differentiation and its effects can be mediated by stimulating C3aR signaling pathways. Therefore, it is possible that Cfd may function in both adipocytes and pancreatic β cells to serve as a critical player in obesity and diabetes [36]. Accordingly, it would be worth to further investigate the actions of Cfd in obesity and metabolic diseases in vivo.
Cfd secreted by adipocytes was reported as a Pparγ downstream target gene [38]. We show the role of Cfd in stimulating Pparγ expression and adipocyte conversion. First, we identified Cfd as a gene regulated by Pparγ inducers. Second, agonistic regulation of Pparγ by GW7845 treatment in preadipocytes failed to acutely induce Cfd expression. Third, the selected pro-adipogenic small molecules did not exhibit Pparγ agonistic activities. Treatment with any of these compounds in combination with Pparγ agonist GW7845 for 24 hours additively increased the expression of aP2 but not that of Pparγ itself, further suggesting that GW7845 and these small molecules act through different mechanisms (Figure G in S1 File). Finally, ectopic overexpression and knockdown experiments showed that Cfd is sufficient and necessary for induction of Pparγ during adipocyte differentiation. Based on these, it is likely that Cfd can induce Pparγ expression, and in turn, Pparγ drives the expression of Cfd in adipocytes. However, currently we do not know the molecular mechanism by which Cfd regulates Pparγ expression in adipocytes. Similarly, the upstream factors regulated by the pro-adipogenic compounds (PT24 and PT26) inducing Cfd expression has not been investigated. In addition, it will be interesting to investigate the pro-adipogenic effects of these compounds in human adipocytes in the future.
The roles of Cfd have been well recognized in immune cells that protect against infection. In the immune system, hepatic C3b, by spontaneous hydrolysis of C3, associates with Cfd (adipsin) to cleave factor B, resulting in generation of C3 convertase [9,10]. Subsequently, C3a and C3b production activates the alternative pathway and protects against pathogen-induced infection. The role of complement system is also conserved in adipocyte biology. The C3a-C3aR signaling stimulates triglyceride synthesis in adipocytes [12]. Complement factors including Cfd, C3, C3aR, and Cfb are also produced by adipose tissues suggesting the dual effects in adipocytes and immune cells [9,10]. In line with this, our data show the action of secreted Cfd in promoting lipid accumulation and adipocyte differentiation by acting on C3aR, further indicating that the complement factors exert dual actions in adipocyte and immune cells. Although, we cannot totally exclude the possible roles of intracellular Cfd in adipogenesis, it seems clear that the immune cells and adipocytes in adipose tissues control glucose metabolism and insulin sensitivity. Thus, proper regulations on complement systems in adipose tissue may provide alternative strategies against metabolic diseases.
In conclusion, we identified Cfd as a gene induced by pro-adipogenic small molecules and further showed the roles of Cfd in activating C3aR, inducing Pparγ expression, and adipocyte differentiation. Our study provides evidence for the unidentified roles of Cfd in adipocytes and also highlights the utility of pro-adipogenic small molecules as tools to dissect adipocyte biology.