A worldwide epidemic of obesity and its associated metabolic disorders raise the significance of adipocytes, their origins and characteristics. Our previous study has demonstrated that interscapular brown adipose tissue (BAT), but not intramuscular adipose, is derived from the Pax3-expressing cell lineage. Here, we show that various depots of subcutaneous (SAT) and visceral adipose tissue (VAT) are highly heterogeneous in the Pax3 lineage origin. Interestingly, the relative abundance of Pax3 lineage cells in SAT depots is inversely correlated to expression of BAT signature genes including Prdm16, Pgc1a (Ppargc1a) and Ucp1. FACS analysis further demonstrates that adipocytes differentiated from non-Pax3 lineage preadipocytes express higher levels of BAT and beige adipocyte signature genes compared with the Pax3 lineage adipocytes within the same depots. Although both Pax3 and non-Pax3 lineage preadipocytes can give rise to beige adipocytes, the latter contributes more significantly. Consistently, genetic ablation of Pax3 lineage cells in SAT leads to increased expression of beige cell markers. Finally, non-Pax3 lineage beige adipocytes are more responsive to cAMP-agonist-induced Ucp1 expression. Taken together, these results demonstrate widespread heterogeneity in Pax3 lineage origin, and its inverse association with BAT gene expression within and among subcutaneous adipose depots.
Adipose tissues are traditionally classified as white and brown adipose tissues (WAT and BAT). Brown adipocytes contain multilocular droplets and uniquely express UCP1 that uncouples electron transmission and ATP production in the inner membrane of mitochondria, leading to heat production and energy dissipation (Cannon and Nedergaard, 2004). Unlike brown adipocytes, white adipocytes contain monolocular oil droplet and are responsible for storing energy as triglycerides. WAT are located in multiple subcutaneous and visceral locations of body, in the form of distinct fat depots, and contribute to overweight and obesity. Obesity is associated with many metabolic disorders, including type 2 diabetes, high blood pressure, high blood cholesterol, cardiovascular diseases and certain types of cancer (Oh and Olefsky, 2010). It is the visceral adipose tissue (VAT) but not the subcutaneous adipose tissue (SAT) that is associated with insulin resistance and type 2 diabetes (Yamamoto et al., 2010).
Adult humans possess adaptive adipocytes that can be activated to express BAT-specific genes and function as BAT in response to cold exposure (Cypess et al., 2009; van Marken Lichtenbelt et al., 2009; Virtanen et al., 2009). The cold exposure induced brown adipocyte-like cells in the SAT are called ‘beige’ or ‘brite’ adipocytes (Waldén et al., 2012). In addition to cold-induced beige adipocyte formation and thermogenesis, the adipocyte derived hormone Leptin can stimulate sympathetic nerves to release catecholamines, which bind to β-adrenergic receptors of adipocytes and increases intracellular cAMP levels (Commins et al., 1999). The increased cAMP levels are sufficient to activate PKA-CREB pathway which upregulates Ucp1 transcription in adipocytes (Cao et al., 2001). In parallel, activated PKA phorsphorylates a series of substrates, promotes lipid mobilization and fatty acid oxidation, and stimulates heat production and may lead to the formation of beige cells (Altarejos and Montminy, 2011; Wu et al., 2012). Beige adipocytes express not only brown fat markers such as Prdm16, Cidea and Ucp1, but also a set of unique markers distinct from established white or brown fat markers (Waldén et al., 2012; Wu et al., 2012). Whether beige cells are derived from preexisting brown preadipocytes or transdifferentiated from white preadipocytes is unknown.
The developmental origins of various adipose depots have recently begun to be elucidated. Lineage mapping indicates that the interscapular BAT is derived from progenitors that have expressed classic myogenic markers Pax3, Pax7 and Myf5 (Lepper and Fan, 2010; Liu et al., 2012a; Seale et al., 2008), suggesting that BAT share a common developmental origin with skeletal muscles. WAT has been shown to be derived from non-Myf5 lineages, though a recent study argues that subsets of WAT adipocytes may have been derived from the Myf5 lineage (Sanchez-Gurmaches et al., 2012; Seale et al., 2008). It remains unclear, however, if anatomically distinct WAT depots share similar developmental origin, and if heterogeneous origins confer different gene transcription programs and cellular functions.
Pax3 is a paired box homeodomain transcriptional factors critical for the embryonic development of skeletal muscles and neurogenesis (Buckingham and Relaix, 2007). Pax3 mRNA is detected in a group of cells as early as day 8.5 in the dorsal region of neural tube and the adjacent dermomyotome (Goulding et al., 1991). At day 9.5 of mouse embryogenesis, Pax3 is expressed in the somites and dorsal mesoderm (Daston et al., 1996). From day 10 to day 12, Pax3 is expressed in the undifferentiated mesenchyme of both forelimb and hindlimb, and in non-somitic neural crest cells (Goulding et al., 1991). Embryonic somitic Pax3 expressing cells later give rise to skeletal muscle, skin, brown fat and blood vessel smooth muscle cells (Atit et al., 2006; Buckingham and Relaix, 2007; Djian-Zaouche et al., 2012; Esner et al., 2006; Goupille et al., 2011). Our previous study demonstrates that interscapular BAT, but not the intramuscular adipose, is derived from the Pax3 lineage (Liu et al., 2012a). Whether Pax3 expressing cells give rise to white adipocytes is unknown. Given the close association of adult adipocyte progenitor cells with blood vessels (Gupta et al., 2012; Shan et al., 2013b; Tang et al., 2008), we hypothesize that at least some preadipocytes are derived from the Pax3 lineage. By lineage tracing analysis, here we show that Pax3 lineage cells contribute to both SAT and VAT depots but the extent of contribution is highly heterogeneous among depots. Within the same SAT depots, non-Pax3 lineage adipocytes express higher levels of BAT and beige markers. These studies demonstrate that differential lineage origins underlie the phenotypic heterogeneity in various WAT depots.
Results and Discussion
WAT depots are heterogeneous in Pax3 lineage origin
Pax3 is one of the earliest transcription factors expressed in embryonic somites that give rise to skeletal muscle, dermis, cartilage and adipose. To dissect the Pax3 lineage contribution to various WAT depots distributed throughout the body, we performed Cre/LoxP mediated genetic cell lineage tracing. Heterozygous Pax3Cre/+ knockin mice were used as the lineage driver, and the Cre-inducible mTmG dual color or tdTomato reporter mice as the lineage reporter (Engleka et al., 2005; Madisen et al., 2010; Muzumdar et al., 2007). In the Pax3Cre/+/mTmG mice, all cells derived from the Pax3 lineage should be labeled with membrane targeted Green Fluorescent Protein (mG) and non-Pax3 lineage cells should be labeled with membrane targeted Tdtomato (mT, a Red Fluorescent Protein) (Fig. 1A). In the Pax3Cre/+/tdTomato mice, all Pax3 lineage cells should be labeled with tdTomato (a red fluorescent protein; supplementary material Fig. S1).
We examined various subcutaneous and visceral WAT depots following a recently reported nomenclature (Waldén et al., 2012). Four SAT depots examined are sWAT (skin WAT attached on the skin around the hindbrain and neck), bsWAT (back subcutaneous WAT covering the interscapular BAT), asWAT (anterior subcutaneous WAT, bilateral superficial depots above forelimbs) and ingWAT (inguinal WAT, posterior bilateral superficial depots associated with hindlimb). In addition, we examined four VAT depots: thWAT (thymus WAT), inWAT (intestinal WAT), rWAT (retroperitoneal WAT, beside the kidney and attached on the dorsal wall of back) and eWAT (epididymal WAT) (supplementary material Fig. S2).
Progenitors from the stromal vascular fraction (SVF) of Pax3Cre/+/mTmG mice were cultured and induced to differentiate. Overall, all SVF cells except for those from sWAT gave rise to both Pax3 lineage (mG+) and non-Pax3 lineage (mT+) adipocytes (Fig. 1B). Specifically, 55–80% adipocytes in bsWAT, asWAT, thWAT and rWAT cultures were mG+ (Fig. 1B), suggesting that most adipocytes in these depots are derived from the Pax3 lineage. By contrast, no more than 20% of adipocytes in ingWAT, inWAT, eWAT and sWAT cultures were mG+ (Fig. 1B), suggesting that adipocytes in these depots are predominantly non-Pax3 derivatives. Additionally, we performed the in vivo lineage tracing experiments to examine the abundance of Pax3 lineage cells in asWAT, ingWAT, rWAT, and eWAT (supplementary material Fig. S1). There were about 56% and 62% Pax3 lineage adipocytes in asWAT and rWAT, respectively (supplementary material Fig. S1A,C). Consistently, less than 15% of adipocytes in ingWAT and eWAT were raised from Pax3 lineage (supplementary material Fig. S1B,D). Therefore, developmental origins of WAT depots are highly heterogeneous in Pax3 lineage contribution.
The large variations of Pax3 lineage contribution to various subcutaneous and visceral WAT suggest that different WAT depots are derived from distinct cell populations. This conclusion is consistent with recent studies that white adipocytes can be derived from E-cadherin-expressing endothelial cells, perivascular cells and Myf5 lineage (Sanchez-Gurmaches et al., 2012; Tran et al., 2012).
Pax3 lineage contribution correlates negatively to BAT gene expression in various SAT depots
BAT specific Prdm16 and Ucp1 genes exhibit elevated expression in SAT compared to VAT (Seale et al., 2011; Waldén et al., 2012). Given that BAT adipocytes are exclusively derived from the Pax3 lineage cells (Liu et al., 2012a), we hypothesized that Pax3 lineage adipocytes are responsible for the thermogenic gene program of SAT. The Pax3 lineage contributed to 60–75% of adipocytes in bsWAT and asWAT, much higher than its contribution to ingWAT (<10%) (Fig. 2A). Unexpectedly, the BAT specific genes Pgc1a, Prdm16 and Ucp1 are unanimously expressed at significantly lower levels in the bsWAT and asWAT compared to ingWAT (Fig. 2B,C). Thus, the Pax3 lineage adipocytes appear to be inversely correlated to the thermogenic gene program of SAT depots.
Under normal culture condition (37°C), the expression levels of thermogenic gene Pgc1α, Prdm16 and Ucp1 in ingWAT cultures are ∼1/7, 1/2 and 1/130 of those in BAT cell cultures (supplementary material Fig. S3). However, under cold environment, the thermogenic gene program is robustly activated in SAT depots, especially ingWAT (Nedergaard and Cannon, 2013). Our results indicate that the ability of SAT depots to function as adaptive thermogenic organs is related to their lineage origin.
Pax3 lineage adipocytes express lower levels of BAT genes than do non-Pax3 lineage cells within the same SAT depot
To further dissect the phenotypic differences between Pax3 and non-Pax3 lineage cells, we next aimed to analyze these two populations of adipose progenitor cells from the same SAT depot. We chose the asWAT as it contains roughly equal numbers of Pax3 and non-Pax3 lineage cells (Fig. 1B). We used Pax3Cre/+/Rosa-tdTomato mice, in which all cells derived from the Pax3 lineage should exhibit tdT (red) fluorescence. Freshly isolated SVF cells were purified by Fluorescence Activated Cell Sorting (FACS) based on Lin− (CD31, CD45, Ter119, and CD11b), Sca1+, and tdT+/− (Fig. 3A). Thus, the Pax3 and non-Pax3 lineage progenitors are Lin−Sca1+tdT+ and Lin−Sca1+tdT−, respectively.
Both cell populations robustly differentiated into mature adipocytes upon induction, and tdT expression did not change during subsequent culture (Fig. 3B). Real-time PCR analysis showed that compared to the Pax3 lineage adipocytes, non-Pax3 lineage adipocytes express undetectable levels of tdT, but ∼3 folds of Prdm16, Pgc1α, and Cox8b, ∼8 folds of Cox7a, ∼30 folds of Cidea and ∼45 folds of Ucp1 (Fig. 3C). In addition, beige adipocyte specific Tmem26 and Tbx1 genes were expressed 10- and 25-fold higher in the non-Pax3 than the Pax3 lineage adipocytes (Fig. 3D). Furthermore, UCP1 and PGC1α proteins were also expressed more abundantly in the non-Pax3 compared to the Pax3 lineage adipocytes (Fig. 3E). These results together demonstrate that non-Pax3 lineage adipocytes express much higher levels of established mitochondrial thermogenesis related genes unique to the classic BAT and newly defined beige adipocytes.
Non-Pax3 lineage contributes more abundantly to beige adipocytes than does Pax3 lineage
The higher expression levels of BAT/beige specific genes in non-Pax3 lineage adipocytes suggest that the non-Pax3 lineage cells contribute more robustly to beige adipocytes in SAT. To examine this possibility, we immunostained SVF cell cultures from Pax3Cre/+/tdTomato SAT with UCP1 primary antibody. UCP1 positive adipocytes in the SAT SVF cultures were predominantly (>85%) tdT− (RFP−), suggesting that UCP1 expressing beige cells are primarily derived from non-Pax3 lineage (Fig. 4A,B).
We also used FACS to purify the beige cells from cultured asWAT SVF of Pax3Cre/+/tdTomato mice. SVF cells were cultured, passaged 2–3 times and separated based on expression of TMEM26 or CD137 (Fig. 4C–E), two cell surface markers of beige preadipocytes (Wu et al., 2012). Interestingly, freshly isolated SVF cells did not express detectable levels of these proteins by microscopic and FACS analysis (data not shown). Only after the SVF cells were cultured and passaged 2–3 times, can we detect the expression of these markers by immunofluorescence and FACS. This observation indicates that early adipose progenitors are not committed to beige cells, and in vitro proliferation of preadipocytes activates the expression of beige adipocyte specific genes. Importantly, only ∼15% of FACS purified CD137+ or TMEM26+ beige preadipocytes were also tdT+ (Fig. 4F,G), suggesting that ∼85% of beige preadipocytes originate from non-Pax3 lineages. By contrast, there were twice as many (∼30%) tdT+ cells in the CD137− or TMEM26− fraction non-beige cells (Fig. 4F,G). These results suggest that the beige cells in SAT SVF cultures are predominantly derived from non-Pax3 lineage.
To confirm that non-Pax3 lineages contribute more significantly to beige cell formation within the same depot, we performed genetic ablation analysis to selectively remove the Pax3 lineage in cultured preadipocytes of asWAT. A Pax3Cre/+/Rosa26-iDTR mouse model was used for selective ablation of Pax3 lineage cells. In this model, Pax3-Cre induces the expression of DT receptor which is not endogenously expressed in mouse cells. DT treatment would bind to the DT receptor and kill the Pax3 lineage cells in the mixed preadipocyte culture (supplementary material Fig. S4A). Compared to the control cells, the DT treated cells express higher levels of beige cell marker Tbx1 and TMEM26 (P<0.05) (supplementary material Fig. S4B). DT treatment also moderately increased BAT-specific Ear2 (P = 0.1) and Ucp1 (P = 0.20) genes (supplementary material Fig. S4B). The observed modest increases in Ucp1 and beige markers after Pax3 lineage ablation is probably due to its small contribution (∼15%, Fig. 4F,G) to beige adipocytes. The results demonstrate that removal of Pax3 lineage adipocytes leads to enrichment of beige cell markers, which is consistent with our conclusion that non-Pax3 lineage cells contribute more significantly to beige adipocytes.
Cold induces thermogenesis in a class of beige adipocytes located in SAT depots through activation of cAMP signaling (Cao et al., 2001). In response to forskolin treatment that activates cAMP (Schimmel, 1984), cultured beige cells robustly activate Ucp1 gene expression (Wu et al., 2012). Indeed, adipocytes differentiated from CD137+ SVF cells upregulated Ucp1 gene expression much more robustly than did the CD137− cells (Fig. 4H). We next addressed if Pax3 lineage contribution affects the cAMP-responsiveness of cultured SVF cells. As ingWAT contains higher percentage of non-Pax3 lineage adipocytes than does asWAT (Fig. 1B), we hypothesized that ingWAT is more responsive to forskolin induced thermogenesis. Forskolin significantly elevated the expression of Pgc1α and Ucp1 in both asWAT and ingWAT adipocytes (Fig. 4I). However, Ucp1 induction in the ingWAT is about 2-fold of that in the asWAT (Fig. 4I). This result supports the notion that the relative abundance of inducible beige adipocytes in SAT depots is inversely correlated to Pax3 lineage contribution to these depots.
Together, our study provides phenotypic and molecular evidence that non-Pax3 lineage adipocytes are the main driving force of cold-induced browning in the SAT depots. This conclusion is very surprising given that the interscapular BAT adipocytes are exclusively derived from the Pax3 lineage (Liu et al., 2012a). Since Pax3 mRNA was not detected in the Pax3 lineage adipocytes in the adult, a direct role of Pax3 protein in the determination of white versus brown adipocyte gene program can be excluded. The distinct functions of Pax3 lineage cells in BAT and subcutaneous WAT may be explained by their different embryonic origins. Whereas BAT has been shown to be derived from progenitors in the embryonic somites (Atit et al., 2006), the embryonic origin of various WAT depots has been unknown. It is possible that the Pax3 lineage cells in the subcutaneous WAT are derived from a non-somitic cell population, such as migratory neural crest progenitors.
Materials and Methods
All procedures involving the use of animals were performed in accordance with the guidelines presented by Purdue University's Animal Care and Use Committee. Pax3Cre/+, Rosa26-mTmG, Rosa26-tdTomato and Rosa26-iDTR mice were from Jackson laboratory under stock numbers 005549, 007576, 007914, 007900 (Buch et al., 2005; Engleka et al., 2005; Madisen et al., 2010; Muzumdar et al., 2007). Mouse genotyping was done using PCR primers and protocols described by the supplier.
Primary adipocyte cultures
The isolation of SVF cells from WAT is as previously described (Shan et al., 2013b). Briefly, WAT depots were collected, minced and digested with isolation buffer for proper time at 37°C on a shaker. The digestion was stopped and cells were pelleted. The cells were filtered through 70 µm filter and cultured in growth medium containing DMEM, 20% FBS, 1% penicillin/streptomycin at 37°C with 5% CO2 for 3 days, followed by feeding with fresh medium every 2 days. Upon confluence, cells were exposed to induction medium contains DMEM, 10% FBS, 2.85 µM insulin, 0.3 µM dexamethasone (DEXA) and 0.63 mM 3-isobutyl-methylxanthine (IBMX) for 4 days and differentiation medium contains DMEM, 200 nM insulin and 10 nM T3 for several days until adipocytes mature.
Forskolin treatment and DT treatment
Forskolin (10 µM) was added to the fully differentiated adipocytes 1 hour before sample collection. For cell ablation in culture, the SVF cells from asWAT of the Pax3Cre/+/Rosa26-iDTR mice were treated with a final diphtheria toxin (DT) concentration of 200 ng/ml in culture medium for 48 h. To avoid the effect of cell density on adipogenic differentiation, the control and the DT-treated cells were induced for differentiation at the time when they reach confluence.
Fluorescence-activated cell sorting (FACS)
FACS was carried out as described (Shan et al., 2013b). SVF cells were isolated from asWAT of Pax3Cre/+/Rosa26-tdTomato mice. For isolating the Pax3+/Pax3− lineage SVF cells, SVF cells from wild-type (WT) mice were used as negative control for gating the tdTomato+ cells. From the Lin− (CD45−, CD11b−, CD31− and TER-119−) cells, we sorted two populations (tdTomato+ Sca1+ and tdTomato− Sca1+) of SVF cells based on endogenous tdTomato expression and Sca1 antibody staining. Antibodies for CD31-PE-Cy7, CD45-PE-Cy7, Ter119-PE-Cy7, CD11b-PE-Cy7, and Sca1-Pacific blue were purchased from eBioscience. For sorting the TMEM26+/TMEM26− and CD137+/CD137− cells, SVF cells were cultured and passaged two to three times following the initial fractionation were trypsinized, washed and centrifuged. The cells were incubated on ice with primary and secondary antibodies for TMEM26 or CD137 (Wu et al., 2012). Cells were subsequently separated based on the cell-surface markers indicated. After sorting, cells were cultured in CO2 incubator at 37°C and differentiated for sample collection.
Quantitative real-time polymerase chain reaction (qPCR)
Total RNA extraction, cDNA synthesis and real-time PCR were performed as described (Shan et al., 2013b). Briefly, total RNA was extracted and purified from adipose tissues or cell cultures. The purity and concentration of total RNA were measured by a spectrophotometer (Nanodrop 3000, Thermo Fisher) at 260 nm and 280 nm. Ratios of absorption (260/280 nm) of all samples were between 1.8 and 2.0. Then 5 µg of total RNA were reversed transcribed using random primers and M-MLV reverse transcriptase (Invitrogen). qPCR was performed by using a light cycler 480 (Roche) machine for 40 cycles and the fold change for all the samples was calculated by 2−ΔΔCt methods. 18s was used as housekeeping genes.
Protein extraction and western blot analysis
The protein extraction and western blot were conducted as previously described (Shan et al., 2013a). Briefly, total protein was isolated from cells using RIPA buffer contains 50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate and 0.1% SDS. Protein concentrations were determined using Pierce BCA Protein Assay Reagent (Pierce Biotechnology, Rockford, IL). Proteins were separated by sodium dodecyl sulfate PAGE (SDS-PAGE), transferred to a polyvinylidene fluoride (PVDF) membrane (Millipore Corporation, Billerica, MA), and incubated with the first antibodies overnight. The UCP1 primary antibody was from Abcam (Abcam), the PGC1a and GAPDH antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA). Immunodetection was performed using enhanced chemiluminescence (ECL) western blotting substrate (Pierce Biotechnology, Rockford, IL) and detected with a Gel Logic 2200 imaging system (Carestream).
Immunostaining and image capture
Immunostaining was performed as described previously (Liu et al., 2012b). The UCP1 primary antibody used for immunostaining was from Santa Cruz Biotechnology. Fluorescent images were captured with a Coolsnap HQ CCD camera (Photometrics, USA) driven by IP Lab software (Scanalytics Inc., USA) using Leica DMI 6000B fluorescent microscope (Mannheim, Germany) with a 20×objective (NA = 0.70).
The data are presented with mean±s.e.m. P-values were calculated using two-tailed Student's t-test. P<0.05 was considered as statistically significant.
W.L., T.S. and S.K. conceived and designed the experiments. W.L., T.S., X.Y., X.L., P.Z. and Y.L. performed the experiments. W.L., T.S. and S.K. analyzed the data. X.L. contributed reagents, materials and analysis tools. W.L., T.S. and S.K. wrote the paper.
The project is partially supported by funding from MDA, National Institutes of Health [grant number NIH R01AR060652] and USDA [grant number USDA 2009-35206-05218] to S.K. Deposited in PMC for release after 12 months.