|Home | About | Journals | Submit | Contact Us | Français|
Estrogens regulate body weight and reproduction primarily through actions on estrogen receptor-α (ERα). However, ERα-expressing cells mediating these effects are not identified. We demonstrate that brain-specific deletion of ERα in female mice causes abdominal obesity stemming from both hyperphagia and hypometabolism. Hypometabolism and abdominal obesity, but not hyperphagia, are recapitulated in female mice lacking ERα in hypothalamic steroidogenic factor-1 (SF1) neurons. In contrast, deletion of ERα in hypothalamic pro-opiomelanocortin (POMC) neurons leads to hyperphagia, without directly influencing energy expenditure or fat distribution. Further, simultaneous deletion of ERα from both SF1 and POMC neurons causes hypometabolism, hyperphagia and increased visceral adiposity. Additionally, female mice lacking ERα in SF1 neurons develop anovulation and infertility, while POMC-specific deletion of ERα inhibits negative feedback regulation of estrogens and impairs fertility in females. These results indicate that estrogens act on distinct hypothalamic ERα neurons to regulate different aspects of energy homeostasis and reproduction.
Ovarian estrogens exert important anti-obesity effects in women and female mammals. Lower levels of estrogens in postmenopausal women or in ovariectomized (OVX) animals are associated with obesity (Carr, 2003; Rogers et al., 2009). Estradiol-17β replacement in rodents prevents OVX-induced obesity by decreasing food intake and increasing energy expenditure (Gao et al., 2007). Hormone replacement therapy reverses the progression of obesity and metabolic dysfunctions in postmenopausal women (Wren, 2009). However, current hormone replacement therapy is often associated with increased prevalence of heart disease and breast cancer (Billeci et al., 2008). Because estrogens have both positive and negative effects on disease progression which are likely mediated by estrogen receptors (ERs) expressed in a variety of tissues, identification of the critical ERs and their site of actions is imperative in order to develop selective estrogen-based therapies which can selectively treat diseases associated with obesity.
Effects of estrogens on energy balance are primarily mediated by estrogen receptor-α (ERα), as women or female mice with mutations in the ERα gene display hyperadiposity (Heine et al., 2000; Okura et al., 2003), characteristically seen in postmenopausal women and OVX animals. However, the critical ERa sites that mediate estrogenic effects on energy homeostasis have not been identified. In the present study, we generated genetic mouse models with ERα selectively deleted in the central nervous system (CNS), in hypothalamic steroidogenic factor-1 (SF1) neurons, in pro-opiomelanocortin (POMC) neurons, or in both SF1 and POMC neurons, respectively. These models allowed us to identify ERα neuronal populations that regulate food intake, energy expenditure, fat distribution, and reproduction.
To determine if CNS ERα is required for body weight control, we crossed mice carrying loxP-flanked ERα alleles (ERαlox/lox) to the Nestin-Cre transgenic mice. These crosses produced mice lacking ERα in most of brain regions (ERαlox/lox/Nestin-Cre) and their control littermates (ERαlox/lox). Using immunohistochemistry, we demonstrated almost complete absence of ERα in the hypothalamus (and other brain regions) in the ERαlox/lox/Nestin-Cre mice (Supple Fig. 1).
Compared to controls, both male and female ERαlox/lox/Nestin-Cre mice displayed significant increases in body weight (Fig. 1A and 1B). Further characterizations in female mice revealed that increases in body weight were mainly reflected by increased body fat mass (Fig. 1C). We further demonstrated that ERαlox/lox/Nestin-Cre female mice had significantly higher visceral fat distribution (% of the whole body fat), but lower subcutaneous fat distribution (Fig. 1D). These data indicate that CNS ERα is required to regulate body weight, adiposity and fat distribution.
ERαlox/lox/Nestin-Cre mice displayed hyperphagia (Fig. 1E) and decreased heat production (Fig. 1F). The lower energy expenditure may be partly caused by decreased physical activity, as the ambulatory movements and rearing activities in ERαlox/lox/Nestin- Cre mice were significantly reduced (Fig. 1G and 1H). These results demonstrate that CNS ERα is required to mediate estrogenic effects on feeding, energy expenditure, and physical activity.
Notably, we found that circulating estradiol- 17β was significantly elevated in ERαlox/lox/Nestin-Cre mice (Fig. 1I). Our observations that mice lacking ERα only in the CNS develop obesity despite the elevated levels of estradiol- 17β in the circulation suggest that compared to ERα expressed in the peripheral tissues, CNS ERα appears to play more predominant roles in the regulation of energy balance.
The ventromedial hypothalamic nucleus (VMH) is important for regulation of body weight (King, 2006). We hypothesized that ERα in the VMH is required to mediate the estrogenic effects on energy balance. Because steroidogenic factor-1 (SF1), a transcription factor, is expressed exclusively in the VMH within the brain (Ikeda et al., 1995), we used a SF1-Cre transgenic mice (line 7) (Dhillon et al., 2006) to generate mice lacking ERα in the VMH (ERαlox/lox/SF1-Cre). In control (SF1-Cre/rosa26GFP) mice, 48.3±5.7% of ERαneurons in the VMH co-express SF1 and 12.0±1.9% of SF1 neurons express ERα. In ERαlox/lox/SF1-Cre/rosa26GFP mice, the majority of ERα was selectively deleted from the VMH SF1 neurons (Supple Fig. 2). Importantly, we found that the numbers of SF1neurons in ERαlox/lox/SF1-Cre/rosa26GFP mice and in control mice were comparable (Supple Fig. 2), suggesting that deletion of ERα did not cause loss of SF1 population in the VMH.
When fed on regular chow, male ERαlox/lox/SF1-Cre and control (ERαlox/lox) littermates had comparable body weight and body composition (Fig. 2A and 2B). The chow-fed ERαlox/lox/SF1-Cre females had modest but significant increases in body weight (Fig. 2C) and trended increases in both fat mass and lean mass (Fig. 2D). When fed with HFD, male ERαlox/lox/SF1-Cre and control mice showed comparable body weight (Fig. 2E) and fat/lean mass (Fig. 2F). On the other hand, HFD-fed ERαlox/lox/SF1-Cre females had significant increases in body weight (Fig. 2G), which were reflected by increases in both fat mass and lean mass (Fig. 2H). Collectively, these results indicate that ERα expressed by SF1 cells is required to maintain body weight homeostasis in females, but not in males.
ERαlox/lox/SF1-Cre females displayed significantly higher visceral fat distribution but lower subcutaneous fat distribution than controls (Fig. 2I). Interestingly, 30% of adult ERαlox/lox/SF1-Cre females had a massive accumulation of gonadal adipose tissue as demonstrated by appearance (Fig. 2J) and weight (Fig. 2K). Although it is unclear why the gonadal fat pads in subsets of mutant mice form this peculiar shape, this phenomenon is consistent with the abdominal obese phenotype. Given that increased visceral fat is commonly associated with impaired glucose homeostasis (Lafontan and Girard, 2008), we performed glucose tolerance tests, and found that ERαlox/lox/SF1-Cre females were glucose intolerant (Fig. 2L). Collectively, these results indicate that ERα in SF1 neurons is required for the regulation of fat distribution. Loss of ERα in SF1 neurons leads to abdominal obesity, which subsequently causes glucose dysregulation.
We further characterized the white adipose tissue (WAT) in these mutant females. We found that the average adipocyte size of gonadal WAT was larger in ERαlox/lox/SF1-Cre mice (Fig. 2M and 2N). Importantly, compared to controls, ERαlox/lox/SF1-Cre females stored more energy in the gonadal WAT, demonstrated by significantly increased triglyceride content (normalized by fat weight) (Fig. 2P). Consistently, mRNA levels of genes promoting lipogenesis and triglyceride storage, including stearoyl-CoA desaturase-1 (SCD1), lipoprotein lipase (LPL) and acetyl-CoA carboxylase α (ACCα), were significantly elevated in gonadal WAT of ERαlox/lox/SF1-Cre females (Fig. 2Q). These findings indicate that ERα in SF1 neurons is required for the regulation of energy partitioning in the gonadal WAT. On the other hand, the inguinal WAT of ERαlox/lox/SF1-Cre mice and controls had comparable adipocyte size (Fig. 2M and 2O), triglyceride content (Fig. 2P), and gene expression profile (Fig. 2Q). No significant difference was observed in the expression of fatty acid synthase (FAS), ERα, and ERβ in all the fat pads (Fig. 2Q).
The increased body weight and adiposity in ERαlox/lox/SF1-Cre females was not due to differences in energy intake, because young or adult female ERαlox/lox/SF1-Cre and control mice consumed comparable levels of calories (Fig. 3A). In contrast, ERαlox/lox/SF1-Cre females were hypometabolic, as demonstrated by significant decreases in heat production (Fig. 3B). Components of total energy expenditure include energy required for physical activities, basal metabolism and diet-induced thermogenesis (Castaneda et al., 2005). In particular, ERαlox/lox/SF1-Cre females showed normal ambulatory movements and rearing activities (Fig. 3C and 3D), indicating that ERα in SF1 neurons is not required to regulate physical activity. We further assessed the basal metabolic rate by measuring the minimal heat production during the light cycle (Kaiyala et al., 2010), and found that ERαlox/lox/SF1-Cre females had significant reductions in basal metabolic rate (Fig. 3E). To assessed diet-induced thermogenesis, we monitored energy expenditure in response to a fasting-re-feeding paradigm. The increased heat production in ERαlox/lox/SF1-Cre mice was significantly reduced (Fig. 3F and 3G). Thus, our results suggest that ERα expressed by VMH SF1 neurons is required to regulate basal metabolic rate and to mediate appropriate thermogenic responses to feeding.
Brown adipose tissue (BAT) plays crucial roles in mediating thermogenesis (Dulloo, 2002). Although the weight of BAT was not different between genotypes (ERαlox/lox: 86.67±7.68 mg vs ERαlox/lox/SF1-Cre: 97.86±7.16 mg, P>0.05, N=12 or 15/genotype), histological analyses revealed a large amount of lipid deposition in the ERαlox/lox/SF1-Cre BAT (Fig. 3H). Consistently, UCP1 mRNA levels in the ERαlox/lox/SF1-Cre BAT were significantly reduced (Fig. 3I). In addition, mRNA levels of peroxisome proliferator-activated receptor γ (PPARγ), PPARγ co-activator-1α (PGC-1α) and β3 adrenergic receptor, factors known to stimulate UCP1 expression (Seale et al., 2007), were significantly lower in ERαlox/lox/SF1-Cre BAT (Fig. 3I). The levels of PRDM16 and ERα in BAT were not altered (Fig. 3I). Taken together, these findings suggest that reduced ERα signals in VMH SF1 neurons led to a reduction in thermogenic functions of BAT by suppressing UCP1 expression.
ERαlox/lox/SF1-Cre females had significantly lower plasma norepinephrine levels than controls; plasma epinephrine levels were not significantly different (Fig. 3J). These findings suggest that ERα in VMH SF1 neurons is required to maintain normal central sympathetic outflow to the peripheral tissues. The decreased sympathetic tone in ERαlox/lox/SF1-Cre mice could contribute to decreased BAT thermogenesis and the increased lipid accumulation in gonadal WAT.
Messenger RNA of leptin receptors was significantly reduced in the VMH of ERαlox/lox/SF1-Cre females (Fig. 3K). These findings support the possibility that ERα signals in SF1 neurons are required to maintain normal leptin sensitivity by regulating transcription of leptin receptors. No significant changes were found in the mRNA levels of SF1, ERβ, thyroid hormone receptor-α (THRα) and growth hormone secretagogue receptor (GHSR, ghrelin receptors) in the VMH of ERαlox/lox/SF1-Cre females (Fig. 3K).
In addition to the VMH in the brain, SF1 cells are also found in the pituitary, adrenal gland and gonads (Zhao et al., 2001). Therefore, ERα may be deleted in these endocrine organs which may potentially confound the metabolic phenotypes. We found that ERα mRNA levels were comparable in these organs from the two genotypes (Supple Fig.3A). These results suggest that ERα may not be expressed in SF1 cells in these peripheral organs in wildtype mice, and therefore SF1-Cre-induced recombination did not affect ERα expression. Alternatively, the loss of ERα from these peripheral SF1 cells, if any, may be compensated by elevated ERα expressed by non-SF1 cells in the same organs. We further examined the impact of the possible ERα deletion on the functions of these organs. We found that there were no significant changes in the plasma estradiol-17β, corticosterone, T3/T4, progesterone, FSH or LH (Supple Table 1). No major abnormalities were observed in the morphology of the pituitary and adrenal gland (Supple Fig. 3B). However, ERαlox/lox/SF1-Cre ovaries mice showed increased number of antral follicles and lack of corpus luteum (Supple Fig. 3B), consistent with our findings that ERαlox/lox/SF1-Cre females were infertile (data not shown). Nevertheless, because the hormones secreted from the pituitary, adrenal gland and ovary were not significantly altered, it is unlikely that the metabolic phenotypes outlined above are due to the possible deletion of ERα in these peripheral organs, although this possibility cannot be fully excluded.
Pro-opiomelanocortin (POMC) neurons secrete α-melanocyte-stimulating hormone (α-MSH) to reduce food intake and increase energy expenditure (Morton et al., 2006). Estrogens activate POMC neurons (Gao et al., 2007; Malyala et al., 2008). However, the physiological significance of ERα expressed by POMC neurons has not been directly tested. We crossed ERαlox/lox mice to the POMC-Cre transgenic mice to generate mice lacking ERα in POMC neurons (ERαlox/lox/POMC-Cre). In control (POMC-Cre/rosa26GFP) mice, 21.5±2.5% of POMC neurons in the ARH and 20.9±4.8% of POMC neurons in the NTS co-expressed ERα (Supple Fig. 4). In ERαlox/lox/POMC-Cre/rosa26GFP mice, only 1.9±0.4% of POMC neurons in the ARH and none in the NTS co-expressed ERα, confirming that the majority of ERα was selectively deleted from POMC neurons (Supple Fig. 4). Importantly, deletion of ERα from POMC neurons did not cause loss of POMC neurons (Supple Fig. 4).
Male ERαlox/lox/POMC-Cre and control (ERαlox/lox) littermates showed comparable body weight (Fig. 4A). On the other hand, ERαlox/lox/POMC-Cre females had significant increases in body weight (Fig. 4B), mainly reflected by increases in lean mass (Fig. 4C). No significant difference were observed in the visceral/subcutaneous fat distribution (Fig. 4D), the weight of gonadal fat pads (Fig. 4E) and the adipocyte size in gonadal and inguinal WAT (Fig. 4F-4H) between the ERαlox/lox/POMC-Cre females and controls.
ERαlox/lox/POMC-Cre females displayed a chronic hyperphagia. This is demonstrated by significant increases in food intake in chow-fed ERαlox/lox/POMC-Cre females at 4 and 6 months of age (Fig. 5A). When normalized by lean mass, food intake was significantly increased at 6 months (Fig. 5B). We further examined the effects of anorexigenic hormones in these females. While leptin (5 mg/kg, i.p.) significantly reduced food intake in controls, these anorexigenic effects were blunted in ERαlox/lox/POMC-Cre females (Fig. 5C). In contrast, anorexia induced by MTII, a melanocortin agonist, was indistinguishable in ERαlox/lox/POMC-Cre and controls (Fig. 5D). Collectively, these findings demonstrated that ERα in POMC neurons is required to maintain normal feeding behavior, at least partly through interacting with the anorexigenic leptin signals.
Unexpectedly, at libitum ERαlox/lox/POMC-Cre females showed increased heat production (Fig. 5E). In addition, the ambulatory movements during the dark cycle were significantly elevated in the ERαlox/lox/POMC-Cre females (Fig. 5F), which may at least partly contribute to the increased energy expenditure. The rearing activities were comparable between the two genotypes (Fig. 5G). To further confirm the unexpected hypermetabolic phenotypes, we examined the energy expenditure during fasting. Consistent with results from at libitum mice, ERαlox/lox/POMC-Cre mice showed increased heat production during 24 hr fasting (Fig. 5H), which led to a significantly greater weight loss (Fig. 5I). Interestingly, ERαlox/lox/POMC-Cre females had significantly higher plasma norepinephrine than control mice, while plasma epinephrine levels were comparable (Fig. 5J). The elevated sympathetic tone may be the underlying mechanisms for elevated metabolism seen in these ERαlox/lox/POMC-Cre mice.
Although we cannot fully exclude the possibility that the increases in sympathetic outflow and energy expenditure in ERαlox/lox/POMC-Cre females are directly due to ERα deletion from POMC neurons, it is more likely that these phenotypes may result from compensatory actions of ERα in other CNS sites. Supporting this notion, we found that ERαlox/lox/POMC-Cre females have significantly higher plasma estradiol-17β levels (Fig. 6A). Presumably, the elevated estradiol- 17β would act on other CNS ERα sites (including SF1 neurons) to increase sympathetic outflow, energy expenditure and physical activity.
Plasma progesterone in ERαlox/lox/POMC-Cre females trended to increase but failed to reach statistical significance (Fig. 6B). Although the basal FSH and LH levels in gonad intact females were not significantly altered (Supple Table 2), the estrogen-induced suppressions of FSH and LH were blunted in ERαlox/lox/POMC-Cre females. Specifically, estradiol-17β replacement in OVX control females significantly suppressed expression of FSH and LH subunits in the pituitary, whereas these negative feedback effects were blunted in ERαlox/lox/POMC-Cre females (Fig. 6C and 6D). Similar patterns were also observed in plasma FSH and LH (Fig. 6E and 6F). These results suggest that ERα in POMC neurons is required to mediate the negative feedback regulation. Of note, the plasma FSH/LH levels after OVX appeared to be lower in ERαlox/lox/POMC-Cre compared to control mice (Fig. 6E and 6F), while the pituitary mRNA levels of FSH/LH subunits in the same mice were comparable (Fig. 6C and 6D). This discrepancy suggests that the secretion of FSH/LH or the stability/degradation of plasma FSH/LH may be differently regulated in the OVX ERαlox/lox/POMC-Cre mice.
In addition, we found that the gonad intact ERαlox/lox/POMC-Cre females had abnormal estrous cycles. The length of estrous phase was significantly reduced, while the diestrus phase trended to be elongated in ERαlox/lox/POMC-Cre mice (Fig. 6G). No differences were observed in the length of proestrus phase (Fig. 6G). Further, female ERαlox/lox/POMC-Cre mice had impaired reproductive capacity. In particular, only 30% of ERαlox/lox/POMC-Cre females (6 out of 20) were able to conceive and deliver, while 100% of the age-matched control females gave birth (Fig. 6H). In addition, it took significantly longer for the 6 ERαlox/lox/POMC-Cre dams to conceive than the controls (Fig. 6I). The average litter size from these dams was significantly reduced (Fig. 6J). No major abnormalities were observed in the morphology of ovaries from ERαlox/lox/POMC-Cre females (Supple Fig. 5C). Collectively, these findings indicate that ERα in POMC neurons is required to maintain normal estrous cyclicity and female fertility.
In an attempt to further characterize the effects of estrogen/ERα on the POMC neurons, we examined the gene expression profile in the hypothalamus of ERαlox/lox/POMC-Cre females. However, there was no significant difference in the expression of POMC, AgRP, NPY, CART, leptin receptors, SF1, GnRH, Kiss1 and GPR54 between ERαlox/lox/POMC-Cre females and controls (Fig. 6K).
In addition to the POMC neurons in the brain, POMC (ACTH) cells also exist in the pituitary. We found that ERα mRNA was significantly reduced in the ERαlox/lox/POMC-Cre pituitary (Supple Fig. 5A), indicating that ERα is also deleted from these pituitary ACTH cells. However, we did not find any significant changes in POMC (ACTH) expression in the pituitary (Supple Fig. 5B). In addition, plasma levels of corticosterone at either basal or stressed conditions were comparable between ERαlox/lox/POMC-Cre females and controls (Supple Table 2). Morphologic analysis did not reveal any alterations in the pituitary (Supple Fig. 5B). In addition, we did not detect any significant changes in plasma T3/T4 levels (Supple Table 2). Therefore, it is unlikely that the metabolic and reproductive phenotypes observed in ERαlox/lox/POMC-Cre females are due to loss of ERα from the pituitary ACTH cells, although we cannot fully rule out this possibility.
Our data indicated that ERα in SF1 neurons is required to regulate energy expenditure and fat distribution, while ERα in POMC neurons is required for the regulation of feeding. To further confirm this model, we generated mice lacking ERα in both SF1 and POMC neurons (ERαlox/lox/SF1-Cre/POMC-Cre) and the control (ERαlox/lox) littermates.
Chow-fed male ERαlox/lox/SF1-Cre/POMC-Cre mice and controls showed comparable body weight and body composition (Fig. 7A and 7B). Female ERαlox/lox/SF1-Cre/POMC-Cre mice had significantly increased body weight (Fig. 7C), which was mainly reflected by increases in lean mass (Fig. 7D). Although the whole body fat mass was not different (Fig. 7D), the average weight of gonadal fat was significantly increased in ERαlox/lox/SF1-Cre/POMC-Cre females (Fig. 7E). Similar to ERαlox/lox/SF1-Cre females, subsets of ERαlox/lox/SF1-Cre/POMC-Cre females displayed the massive gonadal fat expansion (Fig. 7F). While 30% of ERαlox/lox/SF1-Cre females developed this phenotype during adulthood (2–3 months), 50% of ERαlox/lox/SF1-Cre/POMC-Cre females developed the same phenotype at earlier ages (as early as 2 weeks of age) (Fig. 7G). Further, ERαlox/lox/SF1-Cre/POMC-Cre females were not only hyperphagic (Fig. 7H), but also showed decreased heat production (Fig. 7I). No significant difference was observed in ambulatory movements and rearing activities in female mice (Fig. 7J and 7K). Finally, ERαlox/lox/SF1-Cre/POMC-Cre females had significantly lower plasma norepinephrine levels than controls; epinephrine levels were not significantly different (Fig. 7L). Collectively, these findings indicate that ERα expressed by SF1 and POMC neurons provides the coordinated control of food intake, energy expenditure and fat distribution.
To rule out the possibility that the Cre transgenes used in the present study may have independently contributed to the metabolic phenotypes, we generated parallel cohorts of transgenic mice carrying only the Nestin-Cre, SF1-Cre or POMC-Cre transgene and their respective wildtype littermates. These mice were produced at the same genetic background as those conditional knock-out mice. No significant change in body weight was observed in these transgenic Cre mice (Supple Fig. 6).
Estrogen/ERα system is known to regulate food intake, energy expenditure, physical activity (Gao et al., 2007) and fat distribution (Heine et al., 2000). ERα-expressing cells mediating these estrogenic effects are not identified prior to our study. The phenotypic comparison in four mouse models we generated provides evidence to support a segregation model that ERα expressed by distinct hypothalamic neurons mediates different functions of estrogens in the context of energy homeostasis.
Mice lacking ERα in the CNS have hyperphagia and decreased energy expenditure. These results indicate that CNS ERα is required to suppress food intake and increase energy expenditure. The anorexigenic effects of estrogens are further pinpointed to be mediated by ERα-positive POMC neurons, as mice lacking ERα in POMC neurons or in both POMC and SF1 neurons develop hyperphagia, while deletion of ERα only in SF1 neurons does not affect feeding.
We identify ERα-positive SF1 neurons as the key site where estrogens act to stimulate energy expenditure, since deletion of ERα in SF1 neurons produces hypometabolic phenotypes. The increased energy expenditure seen in ERαlox/lox/POMC-Cre mice is unexpected, as estrogens are known to activate POMC neurons (Malyala et al., 2008), and activation of POMC neurons would increase energy expenditure (Morton et al., 2006). The fact that ERαlox/lox/POMC-Cre mice have elevated estradiol-17β levels suggests that the increased energy expenditure may result from compensatory estrogenic actions in other ER sites. Since mice lacking ERα in the CNS show decreased energy expenditure despite the elevated estradiol-17β levels, these “compensatory” estrogen signals must be neuronal in origin. In addition, decreased energy expenditure in ERαlox/lox/SF1-Cre/POMC-Cre mice further pinpoints that ERα-positive SF1 neurons are at least one site where elevated estradiol- 17β acts to stimulate energy expenditure in ERαlox/lox/POMC-Cre mice.
Estrogen/ERα signals also suppress fat accumulation in the visceral adipose depot (Heine et al., 2000; Rogers et al., 2009). We show that CNS deletion of ERα leads to increased visceral fat distribution. Further, mice lacking ERα in SF1 neurons or in both SF1 and POMC neurons show increased visceral fat distribution, whereas deletion of ERα in POMC neurons alone does not influence fat distribution. Therefore, these findings indicate that ERα in SF1 neurons is required to mediate estrogenic effects on fat distribution.
Collectively, our results indicate that estrogenic effects on food intake, energy expenditure and fat distribution are mediated by segregated hypothalamic ERα populations. Thus, ERα in SF1 neurons is required to maintain normal energy expenditure and fat distribution, while ERα in POMC neurons regulates feeding.
Notably, CNS deletion of ERα leads to hypoactivity, indicating that CNS ERα is required to stimulate physical activity. However, this phenotype is not observed in mice lacking ERα in SF1 neurons, in POMC neurons or in both, which implies that other CNS neurons expressing ERα may be responsible for these effects. Of note, animals with ERα knocked down in the VMH display decreased physical activity (Musatov et al., 2007). Together with our observations, these findings suggest that non-SF1 neurons in the VMH may mediate estrogenic actions to regulate physical activity.
ERα mutations in male mice and men cause obesity (Heine et al., 2000; Smith et al., 1994). Consistently, we show that CNS deletion of ERα also promotes body weight gain in males. These findings highlight the physiological relevance of ERα in male brains in the regulation of energy balance. Notably, deletion of ERα in SF1 neurons, POMC neurons or both, does not affect body weight in male mice. Collectively, these findings indicate that other ERα sites in male brains contribute to the regulation of energy balance. Further efforts are warranted to unravel these unknown ERα sites in male brains.
The major metabolic phenotypes in mice lacking ERα in SF1 neurons include decreased energy expenditure (e.g. BAT thermogenesis) and increased lipid accumulation in gonadal WAT. Both these deficits could be attributed to decreased sympathetic tone (lower norepinephrine). Consistently, electric stimulation of the VMH increases sympathetic inputs to BAT and enhances thermogenesis (Saito et al., 1987). Lipid metabolism in gonadal WAT is also regulated by the sympathetic nervous system. For example, increased lipolysis in gonadal WAT is associated with elevated sympathetic inputs (Plum et al., 2007), while expended gonadal WAT is found to have decreased sympathetic inputs (Ramadori et al., 2010). Collectively, these findings support a model that estrogens act on ERα in VMH SF1 neurons to increase central sympathetic outflow, which leads to increased BAT thermogenesis and inhibits lipid storage in gonadal WAT.
The intracellular mechanisms by which estrogen/ERα regulate SF1 neurons are not fully understood. Park et al. reported that estrogens activate the rapid PI3K–Akt pathway in the VMH via an ERα-dependent mechanism (Park et al., 2011). Remarkably, restoration of ERα-associated rapid signaling pathways (including the PI3K–Akt pathway) at the global ERα null background is sufficient to rescue obesity (Park et al., 2011). In addition, we have previously shown that selective inhibition of PI3K in SF1 neurons leads to obesity (Xu et al., 2010). Collectively, these findings suggest that the PI3K pathway (and/or other rapid signaling pathways) in SF1 neurons may mediate anti-obesity effects of estrogens.
Alternatively, ERα may regulate SF1 neurons as a transcription factor. We show that mice lacking ERα in SF1 neurons have decreased expression of leptin receptors in the VMH. Consistent with a possible role of estrogen/ERα on expression of leptin receptors, these two receptors are found to be co-expressed by VMH neurons (Diano et al., 1998), and an estrogen response element (ERE) was found in the promoter region of leptin receptor gene (Lindell et al., 2001). In addition, selective deletion of leptin receptors in SF1 neurons produces similar obese phenotypes (Dhillon et al., 2006). Collectively, these data suggest that ERα in SF1 neurons may regulate energy homeostasis, at least partly, by modulating leptin sensitivity in the VMH.
Mice lacking ERα in POMC neurons develop hyperphagia, indicating that ERα signals in POMC neurons are physiologically relevant in the regulation of feeding. This regulation may be mediated by estrogenic actions on POMC neural activity, as estrogens have been shown to activate POMC neurons (Gao et al., 2007; Malyala et al., 2008). Alternatively, estrogens may regulate feeding partly via sensitizing responses to other anorexigenic hormones, such leptin. We previously demonstrated that estrogens potentiate the anorexigenic effects of leptin (Clegg et al., 2006). We now extend those findings to demonstrate that the effects of leptin on food intake are blunted in mice lacking ERα in POMC neurons. Importantly, the efficacy of MTII’s anorexia is not affected in these mice, suggesting that the impairments of estrogenic actions are constrained to the POMC neurons. Recent evidence indicates that estrogens, similar to leptin, induce STAT3 phosphorylation in the hypothalamus, and that the anti-obesity effects of estrogens are abolished in mice lacking STAT3 in the brain (Gao et al., 2007). Therefore, it is possible that estrogens may initiate signals in POMC neurons which ultimately converge on the leptin-induced pSTAT pathway to regulate food intake.
While the majority of POMC neurons reside in the ARH of the hypothalamus, a small subset of POMC neurons are also found in the NTS of the brainstem, which are implicated in the regulation of satiety (Fan et al., 2004). In addition, estrogens increase c-fos immunoreactivity in the NTS via ERα-mediated mechanisms (Asarian and Geary, 2007). Notably, our POMC-Cre induced efficient ERα deletion in the NTS, as well as the ARH. Therefore, these findings support an alternative model that estrogens may regulate food intake at least partly by acting on ERα expressed by NTS POMC neurons.
ERαlox/lox/SF1-Cre females develop infertility likely due to anovulation (the lack of ovarian corpus luteum). An important question remains as to whether this ovarian phenotype results from loss of ERα in VMH SF1 neurons or from loss of ERα in ovarian SF1 cells. Although we did not find significant reduction of ERα mRNA in the ovary of ERαlox/lox/SF1-Cre mice, we could not fully exclude the possibility that ERα is deleted in ovarian SF1 cells. On the other hand, it is interesting to note that our ERαlox/lox/SF1-Cre female mice recapitulate the ovarian phenotypes (numerous antral follicles and lack of corpus luteum) seen in ERαlox/lox/CAMKII-Cre mice, a brain-specific ERα knock-out model (Wintermantel et al., 2006). Therefore, it is possible that the ovarian phenotypes are caused by loss of ERα in SF1 neurons in the brain. Supporting this notion, genetic ablation of SF1 neurons in the brain leads to decreased or lacked corpora luteum and subfertility/infertility in female mice (Kim et al., 2010).
Deletion of ERα in POMC neurons impaired negative feedback regulation of estrogens on the HPG axis. The negative feedback is primarily mediated by ERα (Couse et al., 2003). ERα in pituitary LH cells partially mediate this effect, as deletion of ERα in LH cells blunts estrogen-induced inhibition on LH secretion (Singh et al., 2009). Alternatively, estrogens may act on gonadotropin-releasing hormone (GnRH) neurons in the hypothalamus to inhibit the HPG axis. As GnRH neurons do not express ERα (Shivers et al., 1983), interneurons must exist to relay the negative feedback signals to GnRH neurons. Our observations that the negative feedback is impaired in ERαlox/lox/POMC-Cre females indicate that POMC neurons may be one of these interneurons. Supporting this notion, ARH POMC neurons project to the rostral preoptic area, where GnRH neurons are concentrated (Simonian et al., 1999). The possible signals that originate from POMC neurons and synapse on GnRH neurons are unclear. Recent studies from lean hypogonadotropic ewes suggested that α-MSH mediates leptin effects to restore pulsatile LH secretion (Backholer et al., 2010). Additionally, estrogens are implicated to increase the secretion of another POMC gene product, β-endorphin (Lagrange et al., 1994), which inhibits GnRH neurons (Lagrange et al., 1995). These findings support a model that estrogens act on ERα in POMC neurons to modulate secretion of POMC peptides (e.g. α-MSH and/or β-endorphin), which in turn act on GnRH neurons to mediate the negative feedback on HPG axis.
Other ERα populations in the brain may also contribute the negative feedback regulation. For example, Kiss1 neurons are required for the tonic stimulation of GnRH/LH secretion (Dungan et al., 2007). Interestingly, Kiss1 neurons in the ARH co-express ERα (Cravo et al., 2011) and estrogens reduce ARH Kiss1 expression (Smith et al., 2005). Of note, Kiss1 neurons and POMC neurons are exclusively segregated in the ARH (Cravo et al., 2011). Therefore, it is possible that ERα in ARH Kiss1 neurons may provide an alternative pathway, in addition to ERα in POMC neurons, to mediate the negative feedback.
Using mice lacking ERα in the CNS, in SF1 neurons, in POMC neurons, and in both SF1 and POMC neurons, we have substantially narrowed down the critical ERα sites that mediate the estrogenic effects on energy homeostasis and reproduction. These results provide genetic evidence to segregate physiological functions of ERα cells in the context of obesity and infertility, and may provide rational targets for the development of highly selective hormone replacement therapies which may be used to combat obesity and infertility in women.
Care of all animals and procedures were approved by the University of Cincinnati, UT Southwestern Medical Center and Baylor College of Medicine. Mice were housed in a temperature-controlled environment in groups of two to five at 22°C-24°C using a 12 hr light/12 hr dark cycle. Some cohorts were singly housed to measure food intake. The mice were fed either standard phytoestrogen-free chow (#2916, Harlan-Teklad, Madison, WI) or 42% high fat diet (#88137, Harlan Teklad) and water was provided ad libitum.
Body weight was measured weekly. Food intake was measured daily, and average daily food intake was calculated using data from at least 4 continuous days (7-day data was used in 5-week old ERαlox/lox/SF1-Cre group). Body composition was determined using quantitative magnetic resonance (QMR).
Twelve-week old chow-fed female ERαlox/lox/POMC-Cre mice and their ERαlox/lox controls were anesthetized with ketamine (100 mg/kg) and xylazine (10 mg/kg) and received bilateral ovriectomy, followed by subcutaneous implantations of pellets containing estradiol-17β (0.5 µg/day/mouse with 60-day releasing capacity, OVX+E) or containing vehicle (OVX+V). Four weeks later, these mice were sacrificed in the afternoon (between 1 pm to 4 pm) after deep anesthesia. The pituitary was quickly isolated; trunk blood was collected and processed to collect plasma. Expression of FSH and LH in the pituitary and plasma FSH/LH were analyzed as described in Supplemental Materials.
The data are presented as mean ± SEM. Statistical analyses were performed using SigmaStat 2.03. After confirming normal distribution of data, comparisons between two genotypes were made by the unpaired two-tailed Student's t-test; repeated-measures ANOVA were used to compare changes over time between two genotypes. Two-way ANOVA analyses were used to assess the interactions between genotypes and treatments (e.g. saline versus drugs). P < 0.05 was considered to be statistically significant.
We thank the Mouse Metabolic Phenotyping Core at UTSW Center (supported by NIH PL1 DK081182 and UL1 RR024923), Ms. Charlotte Lee, and Ms. Danielle Lauzon for the technical support. This work was supported by grants from American Heart Association (BGI) and from the NIH (R00DK085330, R01DK093587 and P30 DK079638-03 to YX, HD061539 to CFE, R01DK088423, RL1 DK081185, R37DK53301 and R01DK071320 to JKE, and DK073689 to DJC), from American Diabetes Association (YX).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.