|Home | About | Journals | Submit | Contact Us | Français|
While erythropoietin is the cytokine known that regulates erythropoiesis, erythropoietin receptor (EpoR) expression and associated activity beyond hematopoietic tissue remain uncertain. Here we show that mice with EpoR expression restricted to hematopoietic tissues (Tg) develop obesity and insulin resistance. Tg-mice exhibit a decrease in energy expenditure and an increase in white fat mass and adipocyte number. Conversely, erythropoietin treatment of wild-type mice increases energy expenditure and reduces food intake and fat mass accumulation but showed no effect in body weight of Tg-mice. EpoR is expressed at a high level in white adipose tissue and in the proopiomelanocortin neurons of the hypothalamus. While Epo treatment in wild-type mice induces the expression of the polypeptide hormone precursor gene, proopiomelanocortin, mice lacking EpoR show reduced levels of proopiomelanocortin in the hypothalamus. This study provides the first evidence that mice lacking EpoR in nonhematopoietic tissue become obese and insulin resistant with loss of erythropoietin regulation of energy homeostasis.
Obesity occurs when a surplus of body energy is sustained due to an increase in food intake, or decrease in energy expenditure1,2. Despite growing awareness of obesity worldwide, knowledge of normal physiology of energy homeostasis and fat mass accumulation is still limited. Identifying regulatory factors in energy homeostasis and adipogenesis will provide valuable information for development of effective therapeutic and preventive strategies for human obesity.
Erythropoietin (Epo), a cytokine produced in the fetal liver and adult kidney, is required for erythropoiesis3. Epo binding to its receptor (EpoR) on erythroid progenitor cells activates Janus kinase 2 (JAK2)4 and downstream signaling pathways including signal transducer and activator of transcription 5 (STAT5), extracellular-signal-regulated kinase (ERK) and p38 mitogen-activated protein kinase (p38MAPK)5-7, which promote survival, proliferation and differentiation. Mice lacking Epo or EpoR die in utero due to severe anemia8, but also show impaired angiogenesis and defective brain and heart development9-11, suggesting Epo activity beyond hematopoietic tissue.
A GATA1-promotor/EpoR cDNA transgene (TgEpoRE) provided erythroid specific EpoR expression and rescued the phenotype of EpoR−/−-mice12. These EpoR−/−TgEpoRE (Tg) mice had no gross abnormal morphology but showed susceptibility to hypoxia-induced retinopathy of prematurity and to pulmonary hypertension13,14, suggesting disruption of normal tissue integrity is linked to loss of specific EpoR expression in non-hematopoietic tissue.
In this study we found EpoR expression in wild type (WT) C57B6 mice was high in white adipose tissue (WAT) and hypothalamus compared to other non-hematopoietic tissues. Tg-mice exhibited significant increase in body weight gain due to increased fat mass and a decrease in total activity and energy expenditure compared to WT-mice. Tg- mice were glucose intolerant and became insulin resistant with development of obesity. Further analysis demonstrated a regulatory role of Epo in differentiation of preadipocyte cultures and in vivo in adipocyte number, hypothalamus POMC expression and energy homeostasis. These observations provide the first evidence for an essential role of endogenous Epo in maintaining normal fat mass in mice.
Hematopoietic rescue of the EpoR−/− phenotype in Tg-mice overcomes severe anemia in utero. However, after the first postnatal week we observed a disproportionate increase in body weight. Tg-mice weekly body weight gain was greater by 60% in females and 30% in males compared with WT-mice during the first year, and was due to an increase in fat mass with no change in lean mass (Fig. 1a-e). Whole-body X-ray computed tomography (CT) showed a dramatic increase in Tg-mice visceral and subcutaneous white fat (Fig. 1f-h). Increased body weight was also observed in a second transgenic line for erythroid rescue of EpoR−/−-mice12, indicating that the increase in body weight is not related to transgene insertion site. Although EpoR−/−-mice die in utero, EpoR+/−-mice with or without the transgene (Fig. 1i-j) and transgenic mice on a WT background (EpoR+/+TgEpoRE) survive through adulthood with normal hematocrit and no morphological defects8,15 and exhibit normal body weight.
Development of Obesity in Tg-mice suggested Epo involvement in the regulation of fat mass accumulation. Therefore, we treated adult WT-mice (6 month) fed a high fat diet (HFD) with high dose Epo for 3 weeks. Body weight increased less than 4% in Epo-treated mice with hematocrit increased up to 80%, but body mass increased more than 18% in saline-treated mice by 3 weeks (Fig. 2a). While lean mass remained the same in both groups, fat mass in saline-treated mice increased by more than 112%, but only by 42% in Epo-treated mice (Fig. 2b,c), indicating that Epo treatment greatly suppressed fat mass accumulation.
To determine how hematocrit affects the Epo response in fat mass accumulation, Tg- and WT-mice were treated with Epo for 3 weeks and hematocrit increased from 45% up to approximately 80% in both WT- and Tg-mice (Fig. 2d,e). In WT-mice Epo treatment decreased body weight by 15% compared with a 4% increase with saline treatment (Fig. 2f). In contrast, Tg-mice showed no significant difference in body weight between Epo and saline treatment (Fig. 2g), suggesting the Epo effect on fat mass accumulation is not related to changes in hematocrit and is dependent on EpoR expression in non-hematopoietic tissues. This was further illustrated when ob/ob-mice were subjected to phlebotomy beginning after two weeks of Epo treatment to maintain the hematocrit in the normal range (Fig. 2h). Body weight gain and fat mass accumulation in ob/ob-mice treated with Epo were similar with and without phlebotomy and were significantly decreased compared to saline treatment (Fig. 2i,j).
With increased fat mass, Tg-mice also showed increased plasma insulin levels (Fig. 3a,b). Glucose intolerance was identified by 2 months (Fig. 3c-e). At 8 months fasting plasma insulin levels, plasma triglyceride and fasting blood glucose were significantly greater than those in WT-mice (Fig. 3f-h). Tg-mice also showed exacerbated glucose intolerance and a markedly blunted response to insulin stimulation despite a significantly elevated insulin level (Fig. 3i,j). The increase in body fat and resultant obesity preceded the onset of insulin resistance (Supplementary Fig. S1), suggesting a type 2 diabetes-like phenotype. In comparison, EpoR+/−- and EpoR+/−TgEpoRE-mice did not become obese (Fig. 1i,j) and exhibited normal glucose and insulin tolerance at 6 and 12 months. Interestingly, in WT-mice (3 months) fed HFD, Epo treatment for 6 weeks reduced fat mass accumulation, fasting plasma glucose and insulin levels, and glucose stimulated insulin secretion, and improved glucose tolerance (Supplementary Fig. S2).
Although no differences in body temperature or food intake were observed when Tg- and WT-mice were on a regular diet, body weight gain normalized to food intake was significantly higher in Tg-mice (Supplementary Fig. S3a-c), suggesting decreased energy expenditure in Tg-mice. We assessed locomotor activity and metabolic rate in young Tg-mice (2-3 months) prior to development of noticeable obesity. Tg-mice were less active, showed significantly decreased total moving distance and total movement time, and increased total resting time in the open field test (Fig. 4a-c). An indirect calorimetry study showed decreased total activity in Tg-mice and reduced total oxygen consumption and respiratory quotient (Fig. 4d,e; Supplementary Fig. S3d,e). Energy efficiency in Tg-mice was significantly higher than that of WT-mice (Fig. 4f). Similarly, when mice were fed a HFD, while cumulative food intake was comparable between WT- and Tg-mice, locomotor activity, oxygen consumption and respiratory quotient in Tg-mice were significantly lower and energy efficiency was markedly higher (Fig. 4g-j; Supplementary Fig. S3f,g).
In comparison, Epo treatment of WT-mice on HFD significantly reduced food intake of each WT-mouse (Fig. 4k), suggesting that the Epo effect on regulating food intake is dose dependent. These mice exhibited significantly increased locomotor activity (Fig, 4l) and resting oxygen consumption (by 9.2%, P=0.01, Student’s t test) while their rectal temperature was unchanged. Total oxygen consumption showed an increasing trend compared to saline-treated counterparts (Fig. 4m; Supplementary Fig. S3h,i). Epo treatment also decreased energy efficiency by 4-fold (Fig. 4n), while energy efficiency increased in Tg-mice with restricted Epo response. In both Tg-mice and Epo-treated WT-mice, disrupted Epo signaling equilibrium led to a breakdown in energy homeostasis. Although an Epo-induced increase in muscular fat oxidation by supra-physiologic levels of Epo has been suggested16, its effect in regulating energy homeostasis may be critical in the development of obesity in Tg-mice and in suppressed fat mass accumulation in Epo-treated WT-mice.
Leptin and adiponectin, adipokines secreted by WAT, regulate endogenous glucose production, fatty acid oxidation and food intake through binding to their specific receptors in target organs such as brain, liver and muscle17,18. Serum leptin levels were higher in Tg-mice and reduced in Epo-treated WT-mice compared to control WT-mice, and appeared proportional to WAT mass while leptin mRNA levels were unchanged in WAT of these animals (Fig. 5a-e). Similarly, Tg-mice at a younger age (before 4 month) showed significantly higher serum adiponectin levels and no changes in their mRNA levels compared to WT-mice (Fig. 5e,f). In contrast, Epo-treated WT-mice showed markedly lower serum adiponectin level compared to control WT-mice (Fig. 5g). Furthermore, Epo treatment (24 hours) did not change leptin and adiponectin gene expression in both undifferentiated (d0) and differentiated (d7) WT-MEF cells (Fig. 5h,i). These results suggest that Epo has no effect on leptin and adiponectin gene expression. Since Epo treatment of ob/ob-mice significantly decreased fat mass accumulation compared to saline-treated counterparts (Fig. 5j), the Epo effect on fat mass accumulation was not dependent on leptin activity.
We examined expression of other adipokines in WAT of WT-, Tg- and ob/ob-mice and observed increased expression of proinflammatory adipokines Ccl1 and TNFα decreased expression of NOS2 (iNOS), and no change in Csf1 (M-CSF), IL-1β and IL-6 expression in both Tg- and ob/ob-mice compared to WT control mice (Supplementary Fig. S4). Interestingly, we found reduced IL-10 expression in Tg-WAT. The increase in proinflammatory adipokines and decrease in IL-10 raises the possibility that loss of EpoR in non-hematopoietic tissue contributes to increased inflammation giving rise to insulin resistance.
EpoR gene expression was analyzed in multiple tissues including white fat, brown fat, liver, kidney, muscle, heart and spleen (Figure 6a). Spleen, a hematopoietic tissue in mice, exhibited the highest level of EpoR expression. Unexpectedly, a high level of EpoR expression (60% of spleen) was observed in WAT. In comparison, EpoR expression in other non-hematopoietic tissue was no greater than 10% the level in spleen. Analysis of WAT fractions from inguinal fat pads of WT-mice showed comparable EpoR expression between stromal vascular cells and adipocytes (Fig, 6b).
High EpoR expression in WAT of WT-mice (WT-WAT) and increased white fat mass in Tg-mice (Tg-WAT) that lack EpoR expression in WAT led us to hypothesize that Epo signaling may affect adipogenesis. Primary mouse embryonic fibroblast (MEF) cells were induced to differentiate into adipocytes by treatment with IBMX, dexamethasone and insulin (MDI) for 3 days followed by insulin supplementation. Epo addition to MEF cultures from WT-mice (WT-MEF) decreased preadipocyte differentiation by 23% at 5U/ml and by 37% at 10U/ml, determined by Oil red O staining (Fig. 6c,d). In comparison, differentiation of MEF cells from Tg-mice (Tg-MEF) was unaffected by Epo treatment and comparable to WT-MEF cells without Epo (Fig. 6d). In 3T3-L1 cells, Epo treatment reduced preadipocyte differentiation and PPARγ and aP2 expression (Fig. 6e). Further analysis showed although Epo did not affect 3T3-L1 proliferation in standard culture media, Epo reduced the terminal expansion of cell number (by 25%) obtained with induction of adipocyte differentiation by MDI (Fig 6f,g). Epo also decreased the terminal expansion of WT-MEF cell number (by 14%) during induction of preadipocyte differentiation but not in Tg-MEF cultures (Fig. 6h). Both TUNEL and flow cytometry analysis showed this Epo effect was not due to increased cell death (Fig. 6i,j; Supplementary Fig. S5). These results demonstrate that Epo has a significant effect in reducing preadipocyte expansion during differentiation induction in culture.
To assess the mechanism of Epo activity, we examined potential EpoR downstream signaling molecules, such as STAT5a, ERK and p38MAPK. Activation of MAPKs has been shown to inhibit the activity of PPARγ, a master regulator of adipogenesis19. Both expression and phosphorylation of ERK44/42, P38MAPK and PPARγ were comparable in WT-MEF and Tg-MEF cells under normal culture conditions (Supplementary Fig. S6a). Epo treatment of WT-MEF cells during differentiation, but not in Tg-MEF cells, induced significant increase in phosphorylation of ERK42 (p-ERK42), ERK 44 (p-ERK44) and p38MAPK (P-p38MAPK) and led to an increase in PPARγ phosphorylation (p-PPARγ) at day 7 and 9, while overall expression levels remained comparable (Fig. 7a; Supplementary Fig. S6b). STAT5a expression increased significantly during late adipocyte differentiation, as previously reported20, but showed no difference between WT- and Tg-MEF cells (Supplementary Fig. S6c,d). PPARγ phosphorylation induced by Epo stimulation suggests that Epo may regulate PPARγ phosphorylation through ERK and p38MAPK signaling pathways during preadipocyte differentiation.
Since Epo stimulation activates ERK and p38MAPK signaling pathways in differentiating preadipocytes, we speculate that ablation of EpoR would reduce the activity of these signaling pathways in WAT. Indeed, differences in WAT from WT- and Tg-mice were analogous to those observed in MEF cells. Phosphorylation of p38MAPK, ERK44, ERK42 and PPARγ in Tg-WAT was significantly lower than in WT-WAT, although protein expression levels were unchanged (Fig. 7b,c). These results suggest an role of Epo in regulating MAPKs activity and PPARγ phosphorylation in WAT.
Assessment of adipocyte size distribution in gonadal fat pads showed a shift to smaller cells despite a marked increase in white fat mass in Tg-mice. Consistent with the histological characteristics revealed in tissue sections, direct analysis of adipocyte size in WAT by flow cytometry provided further evidence that fat cell size in Tg-mice is smaller than in WT-mice (Fig. 7d,e). The increase in adipocyte number in Tg-WAT compared with WT-WAT is consistent with preadipocyte culture studies where preadipocyte differentiation and cell number increased in the absence of Epo stimulation.
In contrast to high EpoR expression in WAT, we detected only a low level of EpoR in brown fat and no differences in PPARγ expression and phosphorylation or in Epo signaling in brown adipose tissues (BAT) from WT- and Tg-mice. Similarly, no alterations in mRNA levels of BAT specific genes, Ucp1, Pgc1α and Dio2, were found between WT- and Tg-mice (Fig. 7f), suggesting that BAT is not Epo responsive.
Decreased locomotor activity and energy expenditure in Tg-mice, and decreased food intake and increased locomotor activity and energy expenditure in WT-mice with Epo treatment suggested that Epo contributes to the central regulation of energy homeostasis. We found high level of EpoR expression in hypothalamus comparable to WT-WAT (Fig. 8a). EpoR expression was increased in WAT from WT-mice on HFD and ob/ob-mice (Supplementary Fig. S7), but absent in Tg-WAT, indicating that EpoR expression in WAT does not generally correlate directly with fat mass accumulation which is primarily related to changes in energy homeostasis. However, EpoR expression in the hypothalamus was absent in Tg-mice and significantly reduced in ob/ob-mice compared to littermate control and WT-mice (Fig. 8b), suggesting an inverse relationship with fat mass accumulation. Gene expression analysis of hypothalamic neuropeptides, that regulates activity, food intake and energy homeostasis21, revealed that expression of proopiomelanocortin (POMC) decreased more than 50% in Tg-mice compared to WT-mice while comparable levels of expression were observed for agouti-related protein (AGRP), neuropeptide Y (NPY), melanin-concentrating hormone (pro-MCH) and prepro-orexin (Fig. 8c). Furthermore, Epo treatment in WT-mice increased POMC mRNA 3-fold, but had no effect on hypothalamic expression of NPY, AGRP, pro-MCH and prepro-orexin (Fig.8d). Immunostaining for EpoR on WT-mouse brain sections revealed specific staining on neurons in the hypothalamus that were identified to be exclusively POMC neurons by double immunostaining (Fig. 8e). These findings led us to speculate that in brain, Epo directly stimulates POMC neurons to induce α-melanocyte-stimulating hormone(α-MSH) production that acts through binding to melanocortin receptor MC3R and MC4R, and exerts a direct effect on food intake and energy homeostasis22. To confirm that Epo can directly regulate POMC gene expression, adult hypothalamic neurons were prepared and cultured with or without Epo for 24 hours. Epo treatment increased POMC gene expression by 4 fold, but did not change the gene expression level of NPY, AGRP, pro-MCH and prepro-orexin (Fig. 8f). These data suggest that EpoR expression in POMC neurons permits a direct response to exogenous Epo stimulation. Thus, affecting energy homeostasis by increasing POMC production in the hypothalamus of WT-mice and obese, albeit at a lower level, but not in Tg-mice that lack EpoR in the hypothalamus.
EpoR−/−-mice that die in utero due to disruption of definitive erythropoiesis8 are rescued by the erythroid expressing TgEpoRE transgene (Tg-mice) that restores erythropoiesis. Although Tg-mice show no gross developmental or morphological defect, we show here that they exhibit abnormally increased weight gain from the first week after birth. By10 months, the body weight of Tg-mice exceeded that of WT-mice by 60% in females and 30% in males. The greater accelerated weight gain in female Tg-mice may be related to estrogen-dependent Epo production that has been implicated in estrus cycle-dependent uterine angiogenesis23.
The excessive body weight in Tg-mice was due to increased white fat mass since neither brown fat nor lean mass were altered, suggesting an obese phenotype. Accompanying the disproportionate white fat accumulation was deterioration of insulin sensitivity, evident by 4 months in female mice and 6 months in male mice. These results link Epo activity in non-hematopoietic tissues to white fat mass accumulation and propose an effect of endogenous Epo in maintaining normal fat mass. Hematocrit and weight gain in EpoR+/−-mice with or without the TgEpoRE transgene were indistinguishable from WT-mice providing evidence that the increase of weight gain in Tg-mice is a consequence of loss of EpoR expression in non-hematopoietic tissue rather than transgene-related. Red cell production appeared normal in Tg-mice and Epo treatment in both WT- and Tg-mice increased hematocrit. However, only WT-mice exhibited the decreased fat mass compared with saline-treated control mice. In ob/ob-mice with Epo treatment, phlebotomy prevented the increase of hematocrit but did not affect the Epo-induced decrease of body weight gain in these mice. These data demonstrate that the regulatory effect of Epo on fat mass accumulation is independent of erythropoiesis.
Decreased total activity and rate of oxygen consumption in Tg-mice was detected at 2 months, before the massive increase in fat mass. In contrast, Epo treatment in HFD-fed WT-mice reduced fat mass accumulation, increased total activity and reduced food intake. These changes as consequences of alteration in Epo signaling indicate a breakdown of energy homeostasis leading to disruption of stability in body fat storage. Rodents and humans exposed to high altitude exhibit decreased food intake, loss of body fat with accompanying decreased plasma glucose and insulin24,25. It is tempting to speculate that hypoxia inducible Epo may contribute to hypophagia and loss of body fat associated with high altitude.
The hypothalamus is an organ responsible for regulating energy metabolism. Two populations of neurons in the arcuate nucleus of hypothalamus act in an opposing manner to control of food intake, energy expenditure and glucose metabolism. POMC-expressing neurons sense the change in peripheral regulatory hormones, leptin and insulin, and respond by secreting α-MSH that stimulates melanocortin 3 and 4 receptors in target neurons to decrease food intake and increase energy expenditure26,27. Activation of NPY/ARGP neurons expressing orexigenic neuron peptides NPY and AGRP stimulates orexigens MCH and orexins expression in lateral hypothalamus target neurons to increase food intake and decrease energy expenditure26-28. Surprisingly, we observed high level of EpoR expression in murine hypothalamus localized to POMC neurons, and Epo regulated POMC gene expression but not anabolic hypothalamic neuropeptide gene expression. Furthermore, decreased POMC expression due to lack Epo signaling in Tg-mice hypothalamus is consistent with decreased energy expenditure, increased metabolic efficiency and development of obesity. Increased POMC expression induced by Epo treatment provides an explanation for the decreased food intake, increased energy expenditure and decreased metabolic efficiency. These results show that Epo regulates energy homeostasis at least through its target, POMC neurons, in the hypothalamus. However, Tg-mice with 50% POMC expression are not hyperphagic and the development of obesity is more gradual than that in Pomc−/− mice reflecting a POMC-dose-dependent effect29. Recent observations of weight gain regulation and high fat diet induced leptin resistance by PPARγ in the CNS30,31 suggest additional possible mechanisms for Epo action, given the effect of Epo on PPARγ in WAT and in differentiating preadipocytes. However, PPARγ activity in the rat hypothalamus localized to the paraventricular nucleus31 and not in the arcuate nucleus where we localize EpoR expression in POMC neurons.
The high level of EpoR expression in WT-WAT and the obese phenotype in Tg-mice that lack EpoR expression in WAT led us to explore Epo activities during preadipocyte proliferation and differentiation. Epo binding to the EpoR homodimer in erythroid progenitor cells activates JAK2 and downstream ERK and p38MAPK signaling pathways5-7,32. Similarly, Epo activates ERK and p38MAPK in differentiating preadipocytes, consistent with reports of ERK activity increasing PPARγ phosphorylation and decreasing PPARγ activity19,33. Adipocyte number is largely established during childhood and adolescence34. A recent study also suggests that early adipocyte progenitor cell number plays a critical role in fat mass determination35. Epo treatment during preadipocyte differentiation decreased the terminal expansion of cell number following MDI induction. Expansion of growth-arrested 3T3-L1 cells during differentiation induction is a critical step for subsequent adipogenic gene expression and adipocyte phenotype decision36. The effect of Epo in reducing cell number of differentiating preadipocyte cultures suggests that loss of Epo activity in WAT may result in an increase of total adipocyte number. Indeed, Tg-WAT exhibits marked increase in fat cell number and decrease in fat cell size despite the disproportionate increased fat mass accumulation. Mice homozygous for a PPARγ-2 phosphorylation site mutation (S112A) show a WAT phenotype of greater number and smaller size of adipocytes compared with WT-WAT37, indicating that reduced or absent phosphorylation of PPARγ-2 can lead to increased fat cell number in vivo36. Similarly, endogenous Epo signaling in WAT may contribute to regulation of adipocyte number.
Multiple hormones including insulin, leptin and adiponectin regulate food intake and energy expenditure through the CNS17,21. Plasma insulin was significantly elevated in Tg-mice when their obese phenotype became noticeable, suggesting that insulin resistance may be a consequence of obesity in Tg-mice. Serum leptin and adiponectin levels appeared proportional to fat content - elevated with increased fat mass in Tg-mice, and reduced with decreased fat mass in Epo treated WT-mice. Both leptin and adiponectin mRNA level in WAT are comparable between WT- and Tg-mice. Epo treatment in WT-MEF cell cultures shows no effect in regulation of leptin and adiponectin gene expression both before and after differentiation. In addition, Epo treatment significantly reduced fat mass in leptin deficient ob/ob-mice, indicating that leptin signaling may contribute but is not required for Epo regulation of body weight. Adiponectin is implicated in regulation of gluconeogenesis, fatty acid oxidation, food intake and energy expenditure by stimulating AMP-activated protein kinase and peroxisome proliferator-activated receptor-α18,38, and declines in obese patients and animal models that are insulin resistant, suggesting that decreased adiponectin level contributes to development of obesity and insulin resistance39. However, serum adiponectin levels in young Tg-mice with modest obesity prior to onset of insulin resistance were higher than those in control WT-mice. Hence, adiponectin is an unlikely causal factor of obesity and insulin resistance in Tg-mice. However, EpoR expression is elevated in WAT from obese WT-mice and ob/ob-mice and we observed an increase expression of proinflammatory adipokines in Tg- and ob/ob-WAT and reduced IL-10 in Tg-WAT. The improved insulin resistance with Epo treatment in ob/ob-mice and WT-mice on HFD may relate to the immune-modulatory effects associated with Epo40 in non-hematopoietic tissue that is absent in Tg-mice.
Reports of Epo stimulated activation of an EpoR interaction with the GM-CSF/IL-3/IL-5 receptor common β-chain in hematopoietic and non-hematopoietic cells or tissues raised the possibility of an alternate Epo receptor41-43. The studies presented here do not directly address the nature of the non-hematopoietic Epo receptor. However, the relatively high level of EpoR expression that we observed in the hypothalamus and WAT are consistent with the hypothesis that EpoR expression level determines Epo response and formation of the “classical” EpoR in these non-hematopoietic tissues.
While the essential role of Epo in erythropoiesis has been well established, the biological activity of Epo in non-hematopoietic tissues remains poorly understood. In the current study, we found significant EpoR expression in WAT and hypothalamus. Epo regulates POMC expression in hypothalamus and ablation of Epo signaling in hypothalamus disrupts energy homeostasis and leads to development of obesity and insulin resistance in Tg-mice. The effects of Epo in suppressing preadipocyte differentiation in vitro are associated with increased PPARγ phosphorylation and reduced preadipocyte expansion during differentiation. The lack of Epo signaling in WAT results in an increase in adipocyte number, which may lead to predisposition to this central defect-driven obesity. Together, these data indicate a role of Epo in maintaining normal fat mass in mice and suggest that further investigations are warranted to fully understand the central regulation of Epo in energy homeostasis.
Tg-mice containing the TgEpoRe transgene (GATA-1 locus hematopoietic regulatory domain driving mouse EpoR cDNA12) were established on the EpoR−/− (C57BL/6) background (Jackson Laboratories, Bar Harbor, ME, USA) to obtain EpoR−/−TgEpoRe-mice. Heterozygous EpoR+/−-mice and EpoR+/−TgEpoRe-mice with the transgene were generated. Genotypes were identified by PCR (Neo gene (339bp), wild type EpoR (387bp) and transgene (303bp)) (Supplementary Table S1).
Mice including WT-mice (C57BL/6), ob/ob-mice (Jackson Laboratories) and lean littermate control-mice were maintained on a 12-h light/dark cycles and fed regular chow NIH-07 diet (Zeigler Brothers, Inc., Gardnerds, PA). Animal procedures following National Institutes of Health guidelines were approved by NIDDK Animal Care and Use Committee. The mouse strains and the number, gender and age are listed in Supplementary Table S2.
Body weight was measured using the same balance by the same person. Body composition was measured using the EchoMRI 3-in-1™ (Echo Medical Systems).
X-ray computed tomography (MicroCat II, Imtek, Inc) was used for imaging rodents at 50-100 microns resolutions under anesthesia. Three-dimensional reconstructions of microCT images utilized the Hamming algorithm for soft tissue and semi-automatic algorithms to identify fat by anatomic sites (visceral, subcutaneous and nuchal-scapular brown fat) with the established Hounsfield Units. Amira 4.1 software for three-dimensional image analysis (Mercury ComputerSystems, Chelmsford, MA) was used to determine the total volume for each fat type.
For Epo treatment (3,000 Units/Kg; Epoetin alpha, Amgen Manufacturing, Thousand Oaks, CA), mice received intraperitoneal injection three times/week. Hematocrit was measured manually before and every week after Epo application.
Triglycerides were determined using Infinity™ Triglycerides reagents (Thermo DMA, Louisville, CO). Insulin, leptin and adiponectin concentrations were determined by radioimmunoassay (Linco Research Inc. St. Charles, MO).
Mice fasted overnight were injected intraperitoneally with glucose (2.0g/kg body weight) or insulin (Humulin R, 0.75 mU/kg). Blood glucose levels were measured before (0 min) and up to 120 min after the injection (3 μl from tail vein; Elite Glucometer (Bayer, Elkhart, IN)). For determination of insulin secretion in response to glucose stimulation, 30 μl of blood was collected before (0 min) and up to120 min after glucose injection; serum was prepared and frozen on dry ice immediately for later assessment.
Spontaneous locomotor and explorative activities were measured using an open field system (AccuScan Instruments, 8”× 8” configuration). Mice are tested between 6 am and 11 am for 3 consecutive days. Data were recorded for 60 min after 10 min acclimation to new environment and are expressed as averages from 3 days of testing.
Indirect calorimetry including oxygen consumption and activity measurements were assessed 44. Oxygen consumption and CO2 production were measured at 24C° every 20 min for 47 hours using an eight-chamber Oxymax System (Columbus Instruments, OH; 2.5 l chambers with wire mesh floors and gauze as bedding, using 0.6 l/min flow rate, 90 sec purge, and 30 sec measure; one mouse/chamber). The system was calibrated using a defined mixture of O2 and CO2. Mice were allowed to adapt to individual housing for three weeks. During testing food and water were provided ad libitum. Motor activities were determined by infrared beam interruption (Opto-Varimex mini; Columbus). Resting metabolic rate was calculated with ambulation equal zero. The respiratory exchange ratio (RER) was calculated as the ratio between VCO2 and VO2. Data were expressed as averages of 46 hours excluding the first hour of the experiment. Oxygen consumption data were normalized to (body weight)0.75.
Total RNA extracted from cells or tissues using TRIzol (Invitrogen) was treated with Turbo DNase (Ambion) and 2 μg was reverse transcribed (MultiScribe Reverse Transcriptase (ABI)) for quantitative PCR assays. Relative mRNA quantification of adipokine genes used Taqman gene expression assays (Supplementary Table S3) (Invitrogen). For absolute quantification of mouse EpoR mRNA in fat and hypothalamus, probe-based Taqman PCR and mouse EpoR and S16 cDNA plasmid were used (16S, internal control). For relative mRNA quantification of all other genes, SYBR green real-time RT-PCR was used with normalization to house keeping genes β-actin and 18S using the Delta-Delta CT method. Primers and probes used in Taqman PCR and SYBR green real-time RT-PCR are listed in Supplementary Table S4.
Hearts from postnatal 1 day mice were finely minced. Tissue was digested with collagenase/pancreatin at 37°C for 10 min, disrupted by gently pipetting (10X), and the supernatant was transferred into cold PBS with 50% FBS. After three cycles of digestion, cells were plated in cardiomyocyte culture medium SmGM®-2 medium (CC3181 and CC4149, Lonza), non-adherent cells (cardiomyocytes) were collected after 40 min. and replated into collagen I coated dishes.
Inguinal fat pads were harvested, digested with 1mg/ml collagenase (type II) (37°C, 50 min), and reacion stopped with equal volume of 50% FBS in DMEM. After centrifugation (600g, 10 min), floating mature adipocytes were separated from the pellet of stromal vascular (SV) cells. Cells were plated and analyzed after 24 hours.
Cell-size distribution was measured in gonadal fat pad from Tg- and WT-mice using Beckman Coulter Multisizer III45. Tissue samples from multiple areas and critically the same locations in each fat pad were immediately fixed in osmium tetroxide (37°C, 48 h), and adipose cell size was determined with a 400 μm aperture (20-240 μm cell size range). 6,000 events were counted for each sample.
Embryos at day 14p.c. (brain and dark red organs removed) were finely minced and digested with trypsin-EDTA (Gibco 25300) (37°C) at 1ml per embryo with gentle shaking for 30 min (gentle disruption pipetting up and down 10X every 10 min), and reaction stopped with equal volume of cold PBS with 50% FCS. The solution was filtered (falcom 40mm cell strainer) and cells were collected by centrifugation, washed twice with fresh MEF culture medium (DMEM-high glucose, 10% FBS, 1% pen/strep, Gibco), suspended in warm MEF culture medium and plated (1 embryo per dish). Medium was changed at 1 hour to remove the unattached cells. Remaining cells were cultured and then frozen for later use.
MEF cells and 3T3-L1 cells (ATCC, Manassas, VA) were grown in 10 % CO2, in (DMEM, 10% FBS, 1% pen/strep, Gibco). For preadipocyte differentiation, cells at 100% confluency were stimulated in medium DMEM medium supplemented with 0.5mM IBMX, 5ug/ml insulin and 1uM dexamethasone (Sigma) in the absence or presence of Epo. After 3 days, maintenance medium (DMEM, 10% FBS, 1% pen/strep, 5ug/ml insulin) was used and changed every day. Fresh Epo was added at medium changes.
After preadipocyte differentiation, cells were washed with PBS gently and’ fixed (10% formalin, 1 hour, room temperature). Formalin was removed and fresh oil red O solution (0.3% in 60% isopropanol) was added to stain the cells (1 hour, gentle swirling). Cells were rinsed with distilled water thoroughly. For quantification, stained dishes were dried, oil red O was resuspended in isopropanol (1 min) and solution OD was determined at 510nm.
WAT and MEF cells were homogenized, incubated in lysis buffer (50 mM Tris-HCI buffer, pH 7.4, 1% NP40, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% SDS, 1% sodium deoxycholate) with phosphotase and proteinase inhibitors (Roche) (30 min on ice), and cell debris was removed by centrifugation. Protein concentration was determined using Pierce BCA protein assay kit (Thermo Scientific). Sample proteins (30μg/each) were electrophoretically separated, blotted using XCell SureLock™ Mini-Cell system (Invitrogen)(supplemental information) and visualized using protein specific antibodies (Supplementary Table S5) and the Amersham ECL™ Advance Western Blotting Detection System (GE Healthcare Bio-Sciences Corp). Quantitative analysis was performed by measuring integrated density with NIH image J system and normalized with β-actin.
After dewaxing and rehydration, antigen retrieval was performed by autoclaving using antigen unmasking solution (Vector Laboratories). Rabbit anti-mouse POMC antibody (kindly provided by Dr. Peng Loh; Cellular Neurobiology Section, NICHD/NIH) and FITC-conjugated goat anti-rabbit IgG (M30201, Invitrogen) as a secondary antibody were used for labeling POMC neurons. Concerns about commercial EpoR antibodies suggest caution regarding their use46,47. Since we found variability among commercial lots of EpoR polyclonal antibody, we used the monoclonal antibody developed at Genetics Institute against the EpoR extracellular domain (mh2er/16.5.1) that previously provided consistent results for immunohistochemistry48,49 and Alexa594-conjugated goat anti-mouse IgG (A11020, Invitrogen) in detecting mEpoR.
Adult neurons were isolated from hypothalamus50 and cultured for 24 hours, then treated with Epo (5U/ml) or saline for 24 hours. Total RNA were extracted and analyzed for POMC, NPY, AGRP, prepro-orexin and pro-MCH gene expression.
The data are expressed as means ± SEM. Comparisons between two groups were made using two-tailed nonpaired Student’s t-test. Statistical differences between three or more groups were evaluated by one-way analysis of variance (ANOVA) with Dunnet’s multiple comparison post hoc tests at α=0.05. A p value of less than 0.05 was regarded to be statistically significant.
We thank William Jou and Kimya Davani for technical assistance. This work was supported by the Intramural Research Program of the National Institute of Diabetes and Digestive and Kidney Diseases and the National Institutes of Health.
R.T. designed and performed experiments, analysed data and wrote the paper; O.G designed and performed all the indirect calorimetry and open field assays, analysed the data and contributed to writing the paper. N.S. and M.Y. generated the EpoR−/−TgEpoRE lines. T.C. performed radioimmunoassay and indirect calorimetry assays. D.S. performed X-ray computed tomography and S.M. and L.H. analyzed the data. L.H and E.M performed 3T3-L1 differentiation and quantitative PCR analysis of normal mouse tissues. E.M also contributed in writing the paper. D.Y. and S. C performed fat cell size analysis. C.T. N developed the concept, supervised the study and wrote the paper.
Competing financial interests
The authors declare no competing financial interests.