Thermogenic (beige and brown) adipocytes protect animals against obesity and metabolic disease. However, little is known about the mechanisms that commit stem cells toward different adipocyte lineages. We show here that p107 is a master regulator of adipocyte lineage fates, its suppression required for commitment of stem cells to the brown-type fate. p107 is strictly expressed in the stem cell compartment of white adipose tissue depots and completely absent in brown adipose tissue. Remarkably, p107-deficient stem cells uniformly give rise to brown-type adipocytes in vitro and in vivo. Furthermore, brown fat programming of mesenchymal stem cells by PRDM-BF1-RIZ1 homologous domain containing 16 (Prdm16) was associated with a dramatic reduction of p107 levels. Indeed, Prdm16 directly suppressed p107 transcription via promoter binding. Notably, the sustained expression of p107 blocked the ability of Prdm16 to induce brown fat genes. These findings demonstrate that p107 expression in stem cells commits cells to the white versus brown adipose lineage. Stem Cells2014;32:1323–1336
The ever-increasing prevalence of weight gain and obesity and their comorbidities of cardiovascular disease, type II diabetes, and cancers represent a significant health problem . Obesity is a complex disease whereby environmental and genetic components collectively result in a positive energy balance. Functionally and morphologically different types of adipocytes play a crucial role in energy hemostasis by influencing glycemic and lipid control . Different adipocyte lineages are localized within distinct anatomical regions throughout the body. In mammals, white adipocytes are distributed within major reservoirs located intra-abdominal (visceral) and in subcutaneous white adipose tissue (WAT) depots [3, 4]. Brown adipocytes that are found specifically between the scapulae in rodents and in the neck region of humans make up the classic brown adipose tissue (BAT) depot [4-6]. Another brown-type adipocyte, beige or brown in white (brite), is found dispersed within WAT depots where their development is enhanced by cold adaptation and β3 adrenergic stimulation [7-9]. In humans, beige adipocytes are found in the cervical, supraclavicular, axillary, and paravertebral anatomical locations . Despite their nearly identical phenotypes, beige and brown adipocytes have different developmental origins [10, 11].
Brown-type adipocytes (both brown and beige) are mainly involved in metabolizing energy and heat production [12, 13]. In brown-type fat, oxidative phosphorylation is uncoupled such that the process of respiration is inefficient allowing for significant heat production (nonshivering thermogenesis) through consumption of calories without ATP synthesis. This process is facilitated by the upregulation of thermogenic promoting factors including uncoupling protein 1 (Ucp1) .
Studies demonstrate that brown-type adipocytes have antiobesity potential and can improve glycemic control. An original report estimated that as little as 50 g of BAT is sufficient to metabolize 20% of the daily energy needs of an individual if maximally stimulated . In mice, genetic ablation of BAT causes a propensity toward obesity, diabetes, and hyperlipidemia [15, 16]. Moreover, increasing brown adipocyte activity using transgenic mice overexpressing Ucp1 produced a lean phenotype with reductions in fat stores and triglyceride levels [17, 18]. Whereas transplanted human or mouse brown adipocytes increased lipid and glucose tolerance and insulin sensitivity in mice [19, 20].
In humans, recent data suggest a role for brown-type adipocytes in metabolism since their amount was inversely correlated with body mass index (BMI) [21, 22]. This was confirmed in a study that found activated brown-type adipocytes in most human subjects, except for the most obese . The degree of activation correlated inversely with BMI and percentage of body fat, and directly with resting energy expenditure.
Thus, encouraging stem cells to differentiate into brown-type adipocytes would represent a novel approach to reduce positive energy balance and help combat metabolic diseases . Evidence suggests that the three adipose lineages, white, brown, and beige, arise from committed progenitors known as preadipocytes. These were predominately derived from the lineage commitment of resident mesenchymal stem cells (MSCs), associated with the vasculature, of mesodermal and neuroectodermal origin [25-30]. MSCs and their precursors including the lineage committed preadipocytes are found within the stromal vascular fraction (SVF) of adipose tissue. Indeed, stem and progenitor cells within the SVF of human adipose tissue represent as much as 3% of the total number of cells . Much of our current understanding of the transcriptional basis of adipocyte differentiation is based on studies using lineage committed cell lines . However, very little is known regarding the transcriptional signals that commit MSCs to the white, brown, or beige lineages.
At the transcriptional level, the coregulator PRDM-BF1-RIZ1 homologous domain containing 16 (Prdm16) specifies the brown adipocyte fate [11, 33]. Expression of Prdm16 can stimulate a profound brown adipose selective gene expression pattern and in immortalized noncommitted murine embryonic fibroblasts (MEFs) forces a brown versus white adipose phenotype [33, 34]. Moreover, Prdm16 can act as a molecular switch determining the brown adipocyte over skeletal muscle lineage and determine the prothermogenic program of subcutaneous WAT in mice [11, 35].
The retinoblastoma susceptibility protein (Rb) family of transcriptional corepressors has also been implicated in adipocyte development. Much evidence suggests that the archetype family member Rb functions during adipogenesis and is involved in the maintenance of the terminal differentiated state [36-39]. p107, another Rb family member, has also been implicated in the development of adipose tissue. It is highly regulated during adipocyte differentiation of the committed preadipocyte cell line 3T3-L1 by increasing dramatically after day 1 and disappearing by day 7 [40, 41]. The importance of p107 to adipogenesis is highlighted by the inhibition of adipocyte differentiation during its knockdown in 3T3-L1 cells . However, genetically deleted p107 (p107 KO) MEFs, which are an enriched source of noncommitted MSCs, can readily undergo adipocyte differentiation, with an increased potential over controls [37, 43]. Developmentally, p107 KO mice displayed a pronounced beige adipocyte phenotype with significant upregulation of Ucp1 suggesting that p107 regulates stem cell commitment to determine adipocyte lineage fates . In this report, we now show that suppression of p107 expression in a stem cell autonomous manner is required for commitment of MSCs to the brown-type adipocyte fate. We found that suppression of p107 expression is required for commitment of MSCs to the brown-type adipocyte fate. Moreover, brown fat programming of MSCs by Prdm16 reduces p107 levels in stem cells; this reduction is required for the action of Prdm16 in brown fat. In summary, these findings demonstrate that p107 downregulation is necessary for lineage commitment of the brown over white adipocytes.
Materials and Methods
Mice and Dissections
All animal experiments were performed according to the guidelines approved by the Animal Care Committee of York University, Toronto, Canada. Experiments were performed on adult (8–16-week old) mice of the Balb/c genetic background, including those with a targeted deletion of p107 . To stimulate brown type cell formation, some mice were injected at 1 mg/kg intraperitoneal, daily for 7 days with β3-adrenergic agonist, CL-316,243 (Sigma, St. Louis, MO, www.sigmaaldrich.com) diluted in saline. Various adipose depots were dissected for analysis.
Cell Types and Tissue Culture
MEFs representing an enriched source of primary MSCs were isolated from embryos of a heterozygous p107 KO mating pair at 14.5 days postcoitum. For isolation of the SVF, specific adipose depots were dissected, minced, and digested with 1 mg/ml collagenase I (Sigma) in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) at 37°C for 45 minutes. After, the cell digested slurry was passed through a 100 µm cell strainer and cells were pelleted by centrifugation at 250g for 5 minutes and used immediately for analysis or cultured for adipogenic differentiation. C3H10T1/2 cells are a MSC line obtained from the American Type Culture Collection.
All cells were grown in DMEM 10% FBS and 1% penicillin/streptomycin. MEFs or C3H10T1/2 cells were differentiated to adipocytes after reaching confluence for 48 hours (time 0) as before . The serum-free regimen for adipocyte differentiation used DMEM/F-12 (1:1) media instead of DMEM with 10% FBS. After 7 days of differentiation, some cultures were stained with oil red O for lipid.
Virus Production and Cell Line Derivation
For virus production of shp107 and control shRNA, the retroviral plasmids shp107 (TRCN0000218550; Sigma) and control pLKO.1 shRNA (TRCN0000218550; Sigma) were used. For Prdm16, pMSCV, and p107 virus production, the viral vectors pMSCV-PRDM16, pMSCV, and pBrit-p107 were used, respectively. C3H10T1/2 cells were transduced with 8 µg/ml of polybrene. After 24 hours cells were lysed or washed and refed for cell line derivation. Cell lines were derived by selecting virally transduced cells each with 2 µg/ml puromycin to produce MSC-shp107, MSC-shctl, MSC-Prdm16, MSC-Ctl, and MSC-p107.
Western Blot Analysis
For Western blot analysis, cells or tissue were lysed in RIPA buffer (0.5% Nonidet P40, 0.1% sodium deoxycholate, 150 mM NaCl, 50 mM Tris-Cl, pH 7.5) containing protease inhibitors (Roche, Mississauga, Canada, www.rochecanada.com/portal/ca). Protein lysates were Western blotted according to standard protocol. Antibodies used were anti-Ucp1 (Millipore, Billerica, MA, www.millipore.com), C18 anti-p107 (Santa Cruz, Biotechnology, Dallas, TX, www.scbt.com), rabbit polyclonal Prdm16, M-2 anti-Flag (Sigma), and monoclonal anti-α-tubulin (Sigma).
1 × 107 p107 KO MEFs, control Wt MEFs, or control Wt MEFs transduced with p107 were resuspended in 200 µl of saline, and transplanted subcutaneously into the sternum region of wild-type BALB/c mouse hosts. Brown adipocyte production was stimulated by CL-316,243 (Sigma) diluted in saline injected at 1 mg/kg intraperitoneal, daily for 7 days beginning on the day of the transplant. Six to eight weeks post-transplantation newly formed adipose tissue pads derived from the transplanted cells were excised and processed for histological analysis.
For tissue culture RNA isolation, cells were lysed in RNA Lysis Buffer (FroggaBio, Toronto, Canada, www.froggabio.com) before RNA was purified using the Total RNA mini kit according to manufacturer's instructions (FroggaBio). For RNA isolation from adipose tissue, the TRIzol (Invitrogen, Carlsbad,CA, www.lifetechnologies.com) method was used according to the manufacturer's instruction. RNA (1 µg) was reverse transcribed into cDNA using the GeneAmp Kit (Applied Biosystems, Carlsbad, CA, www.lifetechnologies.com). Reverse transcribed RNA (25 ng) was used for PCR analysis. The qPCR assays were performed on the ABI 7500 Fast cycler using SYBR green PCR Master mix (Invitrogen). Primer sets (Supporting Information Table S1) and conditions for PCR were previously described elsewhere . Relative expression of cDNAs was determined after normalization with β-actin using the ΔΔCt method.
Images were obtained at room temperature using the Axiovert 40C (Carl Zeiss, Oberkochen, Germany, www.coporate.zeiss.com) microscope with ×10 0.25 LD A-Plan Ph1 Var1 or ×20 0.30 LD A-Plan Ph1 Var1 objectives. Digital images were captured using the Power Shot G5 camera with conversion lens adaptor LA-DC8B, 58–52 mm (Canon, Ota, Japan, www.canon.com) and were processed with Photoshop (Adobe, San Jose, CA, www.adobe.com).
C3H10T1/2 cells were transfected with increasing amounts of pMSCV-Prdm16 or pCMV E2F1, along with the 900 bp proximal p107 promoter fused to the luciferase gene (a kind gift from Dr. Julian Sage, Stanford University) . pMSCV and pcDNA3 were used as “filler” controls. Transfection efficiency was normalized to renilla luciferase activity by addition of PRL-SV40. Twenty four hours post-transfection, luciferase activity was determined using the Dual-Luciferase assay kit (Promega, Madison, WI, www.promega.ca).
Quantitative Chromatin Immunoprecipitation Assay
Quantitative chromatin immunoprecipitation (qChIP) assay was performed on MSC-Prdm16 cells and the control cell line MSC-Ctl with an assay kit (Millipore) according to the manufacturer's instructions. E2F1 (Santa Cruz; sc-193) and Flag M2 (Sigma) antibodies were used for immunoprecipitation. ChIP primer sets (Supporting Information Table S1) amplified the promoters of p107 (−214 to −32) or positive control angiotensinogen (Agt) (−127 to +23). The PCR conditions were 40 cycles for 30 seconds at 95°C denaturation, 30 seconds at 56°C annealing, and 30 seconds at 72°C extension.
MSC-Prdm16 or MSC-p107 cells were untransduced (control) or transduced with p107 or Prdm16 retrovirus, respectively, all containing 8 µg/ml polybrene. After 24 hours, cells were washed and refed. Once confluent, 72 hours post-transduction, cells were differentiated and lysed after 7 days for qPCR analysis.
Measurement of Mitochondrial O2 Respiration
Cells at 0, 3, and 7 days differentiation were first counted, resuspended in mitochondrial respiration media (MiRO5); 130 mM sucrose, 60 mM K-MES, 1 mM ethylene glycol tetraacetic acid, 3 mM MgCl2, 10 mM K2HPO4, 20 mM HEPES, 1 mg/ml bovine serum albumin (pH 7.4), and permeabilized with 40 µg/ml saponin for 5 minutes at 37°C rotation. Saponin concentration was optimized by comparing the effect of a range of concentrations on State 3 respiration with exogenously added mitochondrial membrane intactness factor cytochrome c in all cell lines. After permeabilization, high resolution respirometry was conducted at 37°C with the Oroboros Oxygraph-2K (Innsbruck, Austria, www.oroboros.at) on 106 cells/chamber in MiRO5 buffer. For measurement of ADP-dependent oxidative phosphorylation (State 3 respiration) the following were added sequentially: (a) 5 mM glutamate + 2 mM malate (NADH generation for complex I-supported respiration), (b) 4 mM ADP, (c) 5 mM pyruvate (complex I substrate), and (d) 5 mM succinate (FADH2 generation for complex II supported respiration). Two methods were used to assess mitochondrial O2 consumption due to proton leak. First, measurements were made in the absence of ADP-stimulation of ATP synthase (State 2 respiration) with the addition of 5 mM glutamate and 2 mM malate to the permeabilized cells. Second, 10 µg/ml oligomycin was added after the addition of the complex 1 and II components of State 3 above.
One-way ANOVA used a Tukey post hoc test. Student's paired t test was used unless otherwise stated. Results were considered to be statistically significant when p < .05.
p107 Is Not Expressed in Terminally Differentiated Adipocytes and in BAT
To begin exploring the role for p107 in adipocyte lineage determination, we first examined p107 expression levels in adipose depots of adult mice relative to their epididymal pads. We found that p107 is expressed at high levels in visceral WAT (epididymal and retroperitoneal) (Fig. 1A). It is not expressed in interscapular BAT depots and is expressed at much lower levels in subcutaneous WAT, a depot known to contain more beige adipocytes (Fig. 1A) . To determine whether p107 is differentially expressed in precursors or mature adipocytes, we fractionated adipose tissues into the terminally differentiated adipocytes and the SVF, containing MSCs and preadipocytes . Interestingly, Western blot analysis revealed that p107 was uniquely expressed in the SVF and was undetectable in the mature adipocyte fraction of all depots (Fig. 1B). Moreover, p107 was significantly increased in the SVF of visceral compared to subcutaneous WAT. Thus, confirming that lower expression levels of p107 in resident stem cells within the subcutaneous adipose tissue of adult BALB/c mice are correlated with higher levels of beige adipocytes. Together, these results suggest that p107 functions in stem or progenitor cells to control the fate of differentiated adipocytes.
Treatment of animals with β-adrenergic activators is well known to increase the thermogenic program by increasing the number of beige adipocytes in subcutaneous WAT depots [28, 47]. We found that treatment of adult mice for 7 days with the synthetic β3-adrenergic agonist, CL-316,243, significantly (p < .05) reduced p107 protein levels in the subcutaneous (inguinal) adipose depot (Fig. 1C). This reduction coincided with the increased protein expression of the brown adipocyte marker Ucp1 in the subcutaneous (inguinal) adipose depot (Compare Fig. 1C with Fig. 1D). Thus, reduction of p107 expression in MSCs or their committed progenitors is correlated with increased beige adipocyte formation in vivo.
p107 KO Stem Cells Readily Differentiate into Brown-type Adipocytes In Vitro
We next tested whether p107 expression levels control adipocyte lineage commitment in a stem cell-autonomous manner. To accomplish this, we first assessed the SVF from the subcutaneous fat pads, which contain adult stem cells, for their capacity to control the fate of adipocyte differentiation. As had been published, oil red O staining 7 days postdifferentiation revealed that adipocyte differentiation was higher for the SVF of p107 KO over wild-type (Wt) mice  (Fig. 1E). Gene expression analysis of general adipogenic markers Adipoq and Fabp4 showed no differences in levels, but Pparγ levels were significantly increased in the differentiated SVF of p107 KO mice (Fig 1F). This increase might be reflective of the known enhanced adipocyte differentiation capacity of the p107 KO SVF . The oxidative state of the differentiated cells was assessed by measuring the expression levels of various brown-type fat-selective genes (Fig. 1G). Convincingly, we found that adipocytes derived from p107 KO SVF had at least a sixfold (p < .0001) significantly higher gene expression level of the brown-type adipocyte marker Ucp1 (Fig. 1G). The significantly higher expression levels of other brown-type markers, Cidea, Cox8b, Elovl3, and Pgc-1α, in differentiated p107 KO SVF corroborated the thermogenic character (Fig. 1G). These results suggest that p107 functions in stem and/or progenitor cells to block brown-type adipocyte differentiation.
p107 KO MEFs Readily Differentiate into Brown-type Adipocytes In Vitro
We confirmed that p107 expression levels control adipocyte lineage commitment within stem cells by assessing MEFs, an enriched source of primary MSCs. Both Wt and p107 KO MEFs underwent efficient conversion into adipocytes as observed by oil red O staining 7 days after induction of differentiation (Fig. 2A). Equivalent differentiation of Wt and p107 KO MEFs was confirmed by quantitative gene expression analysis for various general adipocyte markers, Adipoq, Fabp4, and Pparγ (Fig. 2B). These results indicate that the deletion of p107 does not influence the ability of the cells to differentiate into adipocytes. However, after 7 days, we found that adipocytes derived from p107-deficient MEFs had at least a threefold (p < .01) higher mRNA level of the brown-type adipocyte marker Ucp1 (Fig. 2C). We validated the brown-type adipocyte phenotype by assessing the gene expression levels of other brown-type versus white fat-selective markers, Cidea, Cox8b, Elovl3, and Pgc-1α, which were all significantly higher in p107-deficient cultures (Fig. 2D). However, gene expression levels for Ucp2, a homolog of Ucp1, were not affected (Fig. 2D). These results suggest that brown-type adipocyte differentiation is suppressed by p107 in primary stem cells.
p107 KO MEFs Readily Differentiate into Brown-type Adipocytes In Vivo
We examined whether p107 levels affect adipocyte fate during the in vivo differentiation of MSCs by transplanting p107 KO MEFs or control Wt MEFs alone or transduced with p107 into the sternum region of BALB/c mouse hosts. The formation of new adipose tissue pads can easily be discerned within 6–8 weeks postinjection of cells at this location, which has negligible endogenous adipose stores [33, 48, 49]. At 6 weeks post-transplantation, the newly formed adipose tissue, derived from transplanted cells, was excised for gene expression and histological analysis. Gross visualization of the tissue revealed that p107 KO MEFs readily produced multilocular appearing cells, which is a morphological hallmark of brown-type adipocytes (Fig. 2E) . By contrast, Wt MEFs gave rise to unilocular appearing white adipocytes. The brown-type adipocyte formation of p107 KO MEFs was confirmed by gene expression analysis for Ucp1, which was significantly (p < .05) elevated in the ectopic tissue derived from p107 KO MEFs relative to Wt transplants (Fig. 2F). The sustained expression of p107 by viral transduction of the transplanted MEFs did not significantly alter the Ucp1 gene expression pattern in the newly formed tissue. These results establish that p107 controls the adipocyte lineage fate in culture and in vivo.
Knockdown of p107 in Stem Cells Stimulates Brown Fat Cell Differentiation
To avoid potential developmental effects caused by chronic loss of p107 in MEFs and to maintain a homogenous population of cells, we also used siRNA technology to knockdown p107 acutely in C3H10T1/2 MSCs. The C3H10T1/2 cell line has the potential to differentiate into different cell types including osteocytes and chondrocyte lineages, and they have been widely used to study brown and white adipocyte lineage potential [48, 50, 51]. We used retrovirus to deliver a short hairpin (sh) sequence for targeting p107 mRNA (shp107) or a nontargeting control (shctl) into C3H10T1/2 cells. Gene and protein expression analyses showed that p107 levels were significantly (p < .01) reduced by expression of shp107 (Fig. 3A). Consistent with our MEF studies, we found loss of p107 did not affect the adipogenic differentiation activity of C3H10T1/2 stem cells. Specifically, 7 days postdifferentiation, substantial lipid accumulation was visualized by staining with oil red O in both the MSC-shctl and MSC-shp107 cells (Fig. 3B). Adipocyte formation coincided with expression of general adipocyte markers Adipoq, Fabp4, and Pparγ that were not significantly different between MSC-shctl and MSC-shp107 cells (Fig. 3C). However, there was a significantly higher protein expression of Ucp1 over the course of 7 days of adipogenesis in p107-depleted MSCs (Fig. 3D). Strikingly, the p107-depleted cells expressed a near 400-fold more Ucp1 mRNA compared to control cells at day 7 of differentiation (Fig. 3E). Importantly, the brown-type adipocyte phenotype was not limited to the expression of Ucp1 since other brown lineage markers, including Cidea, Elovl3, and Pgc-1α, were also significantly increased in p107-depleted cells (Fig. 3E). As in the p107 KO MEFs, the gene expression levels for Ucp2 were not affected in the differentiated p107-depleted MSCs (Fig. 3E). Taken together, these results demonstrate that p107 functions in a stem cell autonomous manner to suppress the brown-type adipocyte fate.
p107 Knockdown Increases the Thermogenic and Oxidative Capacity of Differentiating Cells
To confirm that the thermogenic gene expression levels of the differentiated MSC-shp107 cells are reflected by a greater oxidative capacity, we assessed the O2 consumption of the cells during adipocyte differentiation. High resolution respirometry revealed that during State 3 respiration, supported by both complex I and complex II substrates, the MSC-shp107 cells had significantly greater O2 consumption than control cells in day 0, 3, or 7 days of differentiation (Fig. 4A). In addition, our results revealed a greater respiration due to proton leak when supported by substrates for complex I alone (GM) or both complex I and II (GMPS) with oligomycin for each of the days in differentiation (Fig 4B). Together, these data indicate that the oxidative and thermogenic capacities of the p107 depleted stem cells during differentiation are significantly enhanced.
p107 Knockdown in MSCs Influences Commitment Rather than Differentiation
As p107 is not expressed in terminally differentiated adipocytes (Fig. 1), it suggests that its reduction influences the lineage commitment of MSCs to brown-type adipocytes rather than their terminal differentiation. We tested for this possibility by assessing brown-type adipocyte formation when p107 levels were downregulated in control cells to the same levels as those in MSC-shp107 cells during differentiation. With normal serum conditions, p107 levels in MSC-shp107 cells are negligible by day 2 of differentiation compared to MSC-shctl control cells, and disappear for both cell types by day 7 (Fig. 5A). However, differentiation in serum-free media results in MSC-shctl cells with negligible p107 protein expression levels as in MSC-shp107 cells after 2 days (compare Fig. 5A with Fig. 5B). Although, the expression pattern for p107 was the same during differentiation, the MSC-shctl cells were unable to form brown-type adipocytes up to 7 days postdifferentiation as determined by the absence of Ucp1 (Fig. 5B). These results indicate that p107 expression must be downregulated at a time point prior to or at the very least immediately following brown-type adipocyte differentiation.
Prdm16 Downregulates p107 Expression
Prdm16 can drive the expression of a brown fat gene program in BAT and WAT [11, 33, 35]. Hence, we asked whether p107 plays a pivotal role in brown-type adipocyte lineage determination within the Prdm16 pathway. Indeed, the expression pattern for p107 is inversely proportional to Prdm16 in the subcutaneous WAT depots, with the difference increasing in response to treatment with probrowning, β3-adrenergic agonists (Fig. 1) .
To examine whether Prdm16 represses p107 expression or activity in MSCs, we created Flag-tagged Prdm16-expressing cell lines using retrovirus in CH310T1/2 cells. Prdm16-expressing cells readily express Prdm16 RNA and protein as visualized by qPCR and Western blot, respectively (Fig. 6A). As anticipated, Prdm16-expressing MSCs differentiate into brown-type adipocytes expressing significant higher levels of Ucp1, as compared to control cultures (Fig. 6B) . Interestingly, Western blot analysis showed that p107 protein levels were markedly reduced by Prdm16 expression in MSCs, prior to induction of differentiation (Fig. 6C). The effect of Prdm16 on p107 levels appeared to be transcriptional since p107 mRNA levels were at least threefold lower in Prdm16-expressing relative to control MSCs. (Fig. 6D). By contrast, reducing p107 levels did not impact Prdm16 mRNA (Fig. 6E) or protein (Fig. 6F) expression levels, in either the fibroblast stage or during a 7-day time course of adipogenic differentiation. Thus, Prdm16 reduces p107 expression levels, but reductions in p107 levels do not affect Prdm16 expression, suggesting that p107 is a downstream target of Prdm16 in the brown-type adipocyte lineage commitment pathway.
We next evaluated whether Prdm16 regulates p107 transcription. For this purpose, we used a construct containing 900 base pairs of the 5′ regulatory region and promoter proximal to the transcriptional start site of p107 fused to a luciferase reporter gene . As previously shown, E2f1 expression significantly increased luciferase activity from this p107 construct (Fig. 6G). By contrast, increasing amounts of Prdm16 robustly suppressed p107 promoter activity (Fig. 6G), suggesting that Prdm16 might directly control p107 transcription.
Using qChIP assays, we assessed whether Prdm16 binds the promoter region of p107 (Fig. 6H). As C3H10T1/2 cells have negligible levels of Prdm16 (Fig. 6A), we used Flag-Prdm16-expressing or control vector-expressing MSC lines for binding assays. Prdm16-bound DNA regions were immunoprecipitated and purified using anti-Flag antibodies. Specific binding was evaluated by qPCR with primer sets within the 900 base pair (bp) promoter of p107 and to the positive control promoter, angiotensinogen (Agt) (Fig. 4H) . As expected, E2f1 bound to p107 and not the Agt promoter sequences (Fig. 6H). Importantly, Prdm16 was enriched at both the positive control angiotensinogen (Agt) and p107 promoters, but only in the cells engineered to express Flag-Prdm16 (Fig. 6H). Together, these data suggest that Prdm16 interacts with the p107 promoter region to downregulate p107 gene expression in precursor cells undergoing commitment to brown adipocytes.
p107 Blocks Prdm16 Function in Brown Fat Cell Commitment
Our experiments suggested that p107 is required for brown-type adipocyte development in MSCs. In order to assess the importance of p107 to the Prdm16 brown and beige adipocyte determination pathway, we performed “competition” assays. We first asked whether maintaining p107 expression could block the brown fat-inducing action of Prdm16. To accomplish this, MSC lines expressing Prdm16 were transduced with p107 retrovirus and subsequently differentiated (Fig. 7A). p107 over expression did not adversely influence adipocyte differentiation, as shown by staining lipids with oil red O (Fig. 7B). Note that the appearance of general adipogenic markers was at same levels in p107 treated and nontreated Prdm16 cell lines (Fig. 7C). Remarkably, Ucp1 levels were significantly (p < .05) reduced by as much as sevenfold, during adipocyte differentiation (Fig. 7D). Other brown-type adipocyte markers Cidea and Elovl3 were also significantly downregulated by p107 (Fig. 7D). Interestingly, the treatment had no effect on Pgc-1α gene expression levels. These results suggest that loss of p107 is necessary for Prdm16 to bring about the full brown-type adipocyte differentiation program.
Prdm16 Cannot Rescue Brown-type Adipocyte Differentiation in the Presence of p107
The importance of p107 in the Prdm16 brown-type adipocyte-forming pathway was further corroborated by evaluating the phenotype of p107-overexpressing cell lines (MSC-p107), transduced with Prdm16 retrovirus (Fig. 7E). Oil red O and general adipogenic factor expression showed that MSC-p107 overexpressing cell lines differentiated equivalently whether or not they also expressed Prdm16 (Fig. 7F, 7G). However, the excessive Prdm16 levels were unable to increase Ucp1 expression in MSC-p107 cultures, even after 7 days of differentiation (Fig. 7H). Expression levels of other brown adipocyte markers Cidea and Elovl3 were similarly unaffected by Prdm16 (Fig. 7H). Intriguingly, the gene expression levels of Pgc-1α were rescued by Prdm16. This unequivocally demonstrates that p107 is a necessary downstream target of Prdm16, required to promote the full brown-type (brown and beige) adipocyte differentiation program in adult cells.
In this study, we found that p107 controls adipocyte lineage fate decisions of white versus brown-type (beige and brown) adipocytes in a stem cell autonomous manner. We demonstrated that both genetic and shRNA-mediated reduction of p107 in MSCs resulted in a prothermogenic differentiation program in vitro and in vivo. Our results add to the premise that p107 is an important nodal point during stem cell fate decisions. To date, there is evidence that it controls the lineage specification of different neuronal and skeletal muscle cell types during differentiation [45, 52].
The complete absence of p107 within interscapular BAT depots further supports the notion that it is a crucial component for adipocyte lineage fate decisions. A hypothesis confirmed by the conspicuous scarcity of white adipocytes in BAT. The deficiency of p107 in BAT precursors might relate to the different stem cell types and progenitors involved with tissue development compared to WAT [10, 11]. However, tonic levels of sympathetic outflow to BAT may also actively suppress p107 expression . It would be of interest to assess whether p107 expression increases in BAT in response to denervation or thermo neutral conditioning of animals.
Increased Prdm16 expression and beige adipocyte formation in subcutaneous depots corresponds to the significantly lower p107 levels in the stem cell fraction of adult subcutaneous fat that was reduced further with β-adrenergic treatment. Indeed, we provide evidence that Prdm16 targets the downregulation of p107 expression directly to cause commitment to the brown-type adipocyte lineage. Although Prdm16 must downregulate p107 in MSCs for brown fat determination, specifically depleting p107 in MSCs to produce brown-type adipocytes does not require an increase in Prdm16 expression. This observation suggests that reduction of p107 is a major target gene for Prdm16 in lineage commitment, which is able to execute much of the downstream program. It hints at the possibility that other browning agents or adaptations such as Irisin, Fgf21, and exercise might function via reduction of p107 [53-55].
p107 might be a crucial factor that receives external metabolic signals to accordingly adjust how stem cells will react and differentiate to maintain adipocyte homeostasis. Currently, the metabolic cues that direct the upstream signaling pathways for lineage specification through p107 are not known. For example, studies have shown that white adipocyte hyperplasia by recruitment and differentiation of stem cells occurs during high fat feeding and in the obesogenic state [56-58]. Hence, the increase in white adipocyte numbers might be associated with the sustained expression levels of p107, nonetheless the signaling pathway(s) that is activated is not clear. Conversely, the ligand BMP7 has been shown to commit stem cells to the brown-type lineage potentially through the Smad signaling pathway which ultimately relays its signal to Smad 4 [48, 59]. As a potential upstream signaling pathway, it is noteworthy that p107 in complex with E2f4 has been shown to associate with Smad 3 and Smad 4 to interact on the c-myc promoter .
It is not clear how p107 functions during lineage commitment of adipocytes. p107 is a transcriptional corepressor that does not bind to DNA directly but through transcription factors, most notably E2f4 . The levels of p107/E2f4 complexes increase the length of time in the G1 phase of the cell cycle by repressing the promoters of genes involved in the G1 to S phase transition. The G1 phase of the cell cycle is an important opportunity for cells to become influenced by extracellular signals, to develop new signaling pathways, and undergo substantive chromatin modifications [62, 63]. One scenario for how p107 might influence stem cell fates is based on the length of time the uncommitted stem cell remains in G1. Time spent at this stage might be a crucial factor for cell fate changes to occur. According to this hypothesis, factors involved in white adipocyte commitment would require a greater length of time in G1 to bring about their lineage-specific effects. In contrast, if the uncommitted stem cells received adipocyte differentiation signals in the absence of p107, their length of time within G1 is significantly reduced such that the white adipocyte fate would not be possible and only brown-type cells would develop instead.
Recently, Rb has been shown to be important in establishing fate choice between bone and brown adipocytes in vivo . However, unlike p107, Rb functions during adipogenesis and for the maintenance of the terminally differentiated state rather than commitment of stem cells. In accordance with this idea is the finding that simian virus 40 (SV40) antigen that inactivates Rb results in the inhibition of adipogenesis in committed white preadipocyte cells . In addition, Rb-deficient fibroblasts are unable to differentiate into white adipocytes [37, 66]. Contrarily, p107 KO primary SVF cells and MEFs readily differentiate into adipocytes [37, 44]. Furthermore, in difference to p107, Rb is expressed in both terminally differentiated brown and white adipocytes in vivo and in vitro [36, 38]. In this capacity, Rb might provide a means to control adipocyte homeostasis and energy expenditure by sensing metabolic signals that impact on its gene expression and/or phosphorylation status. For example, in brown adipocytes, Rb undergoes phosphorylation upon the cold-induced upregulation of Ucp1 . In addition, for terminally differentiated white adipocytes, the conditional deletion of Rb produces a brown-type adipocyte phenotype . Consistent with this are transgenic mice overexpressing SV40 antigen specifically in terminally differentiated adipocytes that transformed white fat into brown-type adipocytes and continuously activated Ucp1 in interscapular BAT .
Studying the intrinsic changes in the expression levels of p107 and its subsequent effect on thermoregulation is critical. This is highlighted during physiological insults, such as traumatic burns or lipodegeneration, where stem cell activation and commitment might ultimately be regulated by p107 function. Moreover, understanding the underlying mechanisms of stem cell commitment to the different adipocyte lineages can lead to effective treatment strategies for the ongoing metabolic problems that plague society. The identification of the role of p107 in specifying the adipose lineages at the level of the stem cell raises the possibility that it might be amenable for therapeutic manipulation for the treatment of type II diabetes and obesity.
We thank Dr. Guillaume Grenier for critical insights and careful reading of the manuscript.
M.D. and D.P.P.: performed experiments, designed experiments and analyzed the data, created unique materials, discussed the results, and commented on the manuscript; A.S.: Performed experiments, designed experiments and analyzed the data, created unique materials, discussed the results, commented on the manuscript, conceived and developed the hypothesis, led the project, interpreted the data, and wrote the manuscript; C.G.R.P.: designed experiments and analyzed the data, discussed the results, and commented on the manuscript; P.S.: created unique materials, discussed the results, and commented on the manuscript.
Disclosure of Potential Conflicts of Interest
The authors indicate no potential conflicts of interest.