Metabolic Benefit of Chronic Caloric Restriction and Activation of Hypothalamic AGRP/NPY Neurons in Male Mice Is Independent of Ghrelin

Abstract

Aging is associated with attenuated ghrelin signaling. During aging, chronic caloric restriction (CR) produces health benefits accompanied by enhanced ghrelin production. Ghrelin receptor (GH secretagogue receptor 1a) agonists administered to aging rodents and humans restore the young adult phenotype; therefore, we tested the hypothesis that the metabolic benefits of CR are mediated by endogenous ghrelin. Three month-old male mice lacking ghrelin (Ghrelin−/−) or ghrelin receptor (Ghsr−/−), and their wild-type (WT) littermates were randomly assigned to 2 groups: ad libitum (AL) fed and CR, where 40% food restriction was introduced gradually to allow Ghrelin−/− and Ghsr−/− mice to metabolically adapt and avoid severe hypoglycemia. Twelve months later, plasma ghrelin, metabolic parameters, ambulatory activity, hypothalamic and liver gene expression, as well as body composition were measured. CR increased plasma ghrelin and des-acyl ghrelin concentrations in WT and Ghsr−/− mice. CR of WT, Ghsr−/−, and Ghrelin−/− mice markedly improved metabolic flexibility, enhanced ambulatory activity, and reduced adiposity. Inactivation of Ghrelin or Ghsr had no effect on AL food intake or food anticipatory behavior. In contrast to the widely held belief that endogenous ghrelin regulates food intake, CR increased expression of hypothalamic Agrp and Npy, with reduced expression of Pomc across genotypes. In the AL context, ablation of ghrelin signaling markedly inhibited liver steatosis, which correlated with reduced Pparγ expression and enhanced Irs2 expression. Although CR and administration of GH secretagogue receptor 1a agonists both benefit the aging phenotype, we conclude the benefits of chronic CR are a consequence of enhanced metabolic flexibility independent of endogenous ghrelin or des-acyl ghrelin signaling.

The GH secretagogue receptor (GHSR1a) was identified as an orphan G-protein coupled receptor that mediates the action of a small molecule (MK-0677) designed to rejuvenate the amplitude of episodic GH release in the elderly (1, 2). Subsequently, GHSR1a was “deorphanized” by the discovery of an octanoylated peptide in the stomach named ghrelin (3). Ghrelin O-acyl transferase (GOAT) on the endoplasmic reticulum of ghrelin producing stomach cells selectively uses luminal medium chain fatty acids from the diet as substrate to convert des-acyl ghrelin (DAG) to ghrelin (4, 5). Both ghrelin and DAG are found in the circulation. GHSR1a is the only known receptor for ghrelin, whereas DAG neither binds to, nor activates GHSR1a (3). A DAG receptor, as defined by biological activity linked to specific high-affinity, limited capacity binding, has yet to be identified.

Ghrelin-GHSR1a signaling is associated with regulation of metabolism and glucose homeostasis at multiple levels (2, 6–9). Viewed broadly, the actions of ghrelin suggested a key role for the ghrelin-GHSR1a axis in mediating gut-brain communication according to energy balance. Studies using short-term restricted feeding paradigms reported an increase in the total amount of circulating ghrelin in rodents (10, 11), and humans (12–18), whereas other investigators found that fasting and postprandial total ghrelin levels do not change after short-term energy restriction (19, 20). However, measuring total ghrelin peptide (ghrelin + DAG) rather than active ghrelin does not reflect the metabolic importance of ghrelin-GHSR1a signaling. GHSR1a is expressed abundantly in regions of the brain that control appetite, and ghrelin administration enhances food intake in nonfasted animals; nevertheless, in contrast to the pharmacological effects of ghrelin, recent evidence suggests preprandial concentrations of endogenous ghrelin are insufficient to stimulate food intake (21).

GHSR1a is also expressed in pancreatic islets, and under conditions of negative energy balance ghrelin regulates glucose homeostasis. Ghrelin suppresses pancreatic insulin secretion by activating a heteromeric complex between GHSR1a and somatostatin receptor-5 in β-cells and enhancing glucagon secretion by α-cells (22). Caloric restriction (CR) (50%) of Ghrelin−/− and Ghsr−/− mice produced hypoglycemia, but gradual introduction of CR allowed compensatory pathways to restore euglycemia (23). Mice with acyl ghrelin deficiency due to GOAT knockout exhibited life-threatening hypoglycemia in response to acute CR (60%) that can be rescued by acyl ghrelin or GH treatment (24). Although another group failed to confirm the severity of CR-induced hypoglycemia in Goat−/− mice (25), the disparity is likely explained by differences in environment and/or mouse strains. Collectively, these results suggest ghrelin and GHSR1a are sensors of negative energy balance which provide pivotal counter-regulatory mechanisms to maintain glucose homeostasis under conditions of famine to prevent glucopenia (26).

Aging is associated with metabolic changes that become evident in C57BL/6J mice at age 12–15 months. Chronic CR is a demonstrated means of dissociating age from age-related metabolic impairment (27). However, despite significant efforts, elucidation of the molecular mechanisms and identification of mediators that precisely mimic the beneficial effects of CR have remained elusive. Ghrelin is a GH secretagogue, and like GH, ghrelin is implicated in physiological changes that accompany aging (28). GHSR1a signaling decreases with age (29), but pharmacological treatment with GHSR1a agonists reverse aging of the pituitary/hypothalamic GH axis, as well as the effects of aging on thymus and liver functions (2, 30, 31). In old mice, chronic CR increased the size and weight of the stomach and enhanced ghrelin production, which implicated ghrelin as a mediator of the salutary effects of CR (32). Until now, the effect of long-term CR after ghrelin and GHSR1a ablation had not been investigated. By employing Ghrelin−/− and Ghsr−/− mice, here we tested the hypothesis that ghrelin-GHSR1a signaling is critical for the beneficial metabolic effects of chronic CR.

Materials and Methods

Animals

All experimental protocols were approved by the Institutional Animal Care and Use Committee of the Scripps Research Institute-Florida. Male C57Bl/6J mice were individually housed in AAALAC-approved facilities with 12-hour light, 12-hour dark cycles, fed daily, and given free access to water. Congenic Ghrelin−/− mice and Ghsr−/− were generated as described previously and by backcrossing for more than 15 generations with C57BL/6J mice (23, 33, 34). In each case, data were collected and compared with the relevant littermate controls: Ghrelin−/− with Ghrelin+/+ mice, and Ghsr−/− with Ghsr+/+ mice.

Caloric restriction

All mice were individually housed for the duration of the study. Food intake (standard chow) of 12-week-old male C57Bl/6J mice was calculated for 5 days, and mice were randomly assigned to ad libitum (AL) or 40% CR groups (n = 4–11). Intake of CR mice was reduced 10% week 1, 20% week 2, and 40% starting at week 3. Mice were metabolically phenotyped and killed 1 year after restricted feeding was initiated.

Plasma analyses

Blood glucose was measured before killing using an automated glucometer (One touch Ultra; LifeScan, Inc). Immediately after killing with CO₂, blood was collected via cardiac puncture and placed in chilled EDTA-lined tubes with 4 mM 4-(2-aminoethyl) benzene sulfonyl fluoride (Sigma-Aldrich) added immediately and centrifuged. Separated plasma was acidified with 200-μL 1N HCl/mL and stored at −20°C overnight. For the Ghsr−/− cohort: n = 7 for AL-Ghsr+/+, AL-Ghsr−/−, CR-Ghsr+/+; n = 5 CR-Ghsr−/−. Ghrelin−/− cohort: n = 6 for AL-Ghrelin+/+, AL-Ghrelin−/−, CR-Ghrelin+/+; n = 7 AL-Ghrelin−/−. The following day, samples were sent to University of Virginia (Thorner lab) on dry ice. Ghrelin and DAG were assayed separately using sensitive and specific 2-site sandwich assays developed by the Thorner lab as previously described (35).

Indirect calorimetry

Subsets of mice (n = 4 per group) were placed in metabolic chambers (oxymax CLAMS system) designed to measure O₂ consumption, CO₂ production, and ambulatory movement (photobeam break system) in real time. Mice were placed in individual metabolic cages with access to water and food (AL groups only) for 5 days. The first 48 hours were considered an acclimatization period followed by a 72-hour data collection period. Oxygen consumption (VO₂) was calculated per kilogram of lean body mass, which was determined using nuclear magnetic resonance (NMR) (Minispec LF50; Bruker Corp).

Tissue collection

Mice were killed with CO₂ and body weight and length recorded. Bilateral adipose tissue depots (sc inguinal, epididymal, interscapular brown) were excised and weighed before processing for gene expression and histological studies as described below. Brains were removed and hypothalami were carefully dissected.

TaqMan real-time PCR

For the Ghsr−/− cohort: n = 11 for AL-Ghsr+/+, AL-Ghsr−/−, CR-Ghsr+/+; n = 9 CR-Ghsr−/−. Ghrelin−/− cohort: n = 6 for AL-Ghrelin+/+, AL-Ghrelin−/−, CR-Ghrelin+/+; n = 7 AL-Ghrelin−/−. Liver and hypothalamus were placed in TRIzol for RNA extraction using commercial spin column kits (RNeasy; QIAGEN) with the DNAse step. RNA quantity and purity was determined using a Nanodrop spectrophotometer (Nanodrop 2000), and first strand cDNA was generated from 1 μg of RNA (SSIII; Invitrogen). cDNA was diluted 1:20 and used for TaqMan real-time QPCR in 384-well plate format (7900HT; Applied Biosystems). Each reaction (in duplicate) consisted of 2 μL of cDNA (total reaction volume, 10 μL). Assay IDs (Life Technologies) are Mm00435874_m1 (Pomc), Mm00475829_g1 (Agrp), Mm00445771_m1 (Npy), Mm00432554_m1 (Cidea), Mm00617672_m1 (Cidec) Mm00440940_m1 (Pparg2), Mm00440939_m1 (Ppara), Mm00431814_m1 (Apoa4), and Mm03038438_m1 (Irs2). Fold changes were calculated as 2^ΔΔCT with thiodoredoxin-binding protein (liver) or 36B4 (hypothalamus) used as the endogenous controls.

Histology

Excised liver was fixed overnight with Z-fix (Anatech LTD). The Scripps Research Institute histology core embedded samples in paraffin, sectioned (3 μm) tissue, stained sections with hematoxylin and eosin, and generated images.

Statistical analyses

All data are presented as mean ± SEM. Statistical comparisons were done using two-way ANOVA with Bonferroni post hoc testing, with P ≤ .05 considered significant (GraphPad Prism v5).

Results

Ghrelin+/+, Ghrelin−/−, Ghsr+/+, and Ghsr−/− mice display similar changes in body composition in response to chronic CR

After 1 year of 40% CR, Ghrelin+/+, Ghrelin−/−, Ghsr+/+, and Ghsr−/− mice were all approximately 1 cm smaller in naso-anal length and weighed approximately 45% less than corresponding AL-fed groups. Likewise, circulating concentrations of glucose were similarly reduced with CR in all 4 genotypes (Table 1). Total body fat mass, determined by NMR, was markedly reduced in all dietary restricted mice, but two-factor (diet, genotype) ANOVA analyses determined in each strain showed a significant effect of diet (P < .0001), but not genotype (P = .60, P = .08) or diet and genotype (D × G) interaction (P = .72, 0.10) in Ghrelin (Figure 1A) and Ghsr (Figure 1B) cohorts, respectively. Post hoc analyses revealed reduced fat mass in AL-Ghsr−/− compared with AL-Ghsr+/+ mice (P < .05) (Figure 1B). Long-term CR also decreased lean body mass, but as with fat mass, this was observed irrespective of genotype (Figure 1, A and B). Consistent with NMR results, accumulation of sc white adipose tissue (sWAT) and epididymal WAT (eWAT), as well as brown adipose tissue (BAT), was similar in all 4 groups of CR mice (Table 1). For the Ghsr mice only, D × G interaction P values were significant (sWAT and BAT, P < .05) or approaching significance (eWAT, P = .07), and all 3 depots were significantly smaller in AL-Ghsr−/− vs AL-Ghsr+/+ mice (P < .05) (Table 1).

Table 1.

Comparison of the Effects of ad Libitum (AL) and 40% Caloric Restricted (CR) Feeding in Ghrelin+/+, Ghrelin−/−, Ghsr+/+, and Ghsr−/− Mice

	Ghrelin+/+		Ghrelin−/−		Int P Value	Ghsr+/+		Ghsr−/−		Int P Value
	AL	CR	AL	CR		AL	CR	AL	CR
Body weight (g)	42.2 ± 2.7	22.2 ± 0.3c	39.4 ± 2.4	21.8 ± 0.4c	.50	42.1 ± 1.4	21.7 ± 0.4c	38.5 ± 1.7	21.6 ± 0.5c	.13
Body length (cm)	10.1 ± 0.1	9.0 ± 0.0c	10.1 ± 0.1	8.9 ± 0.0c	.98	10.4 ± 0.1	9.2 ± 0.1c	10.3 ± 0.1	9.1 ± 0.1c	.64
Blood glucose (mg/dl)	184 ± 11	145 ± 9a	180 ± 11	131 ± 9b	.62	176 ± 8	126 ± 7b	192 ± 9	128 ± 10c	.26
sWAT (mg)	1337 ± 274	221 ± 16b	1048 ± 285	195 ± 9b	.50	1565 ± 117	234 ± 16c	1135 ± 173d	264 ± 19c	.04
eWAT (mg)	1767 ± 213	179 ± 15c	1479 ± 238	124 ± 11c	.46	2059 ± 110	180 ± 20c	1671 ± 183d	200 ± 19c	.07
BAT (mg)	156 ± 21	39 ± 2c	121 ± 26	37 ± 2b	.32	215 ± 16	44 ± 4c	162 ± 18e	49 ± 5c	.03

Body weight, naso-anal body length, blood glucose levels, subcutaneous white adipose tissue (sWAT), epididymal white adipose tissue (eWAT), and interscapular brown adipose tissue (BAT) weight (in milligrams) after 1 year of CR or AL feeding; n = 6–11.

^aP < .05 CR vs AL.

^bP < .01 CR vs AL.

^cP < .001 CR vs AL.

^dP < .05 −/− vs +/+.

^eP < .01 −/− vs +/+.

Figure 1.

Ghrelin+/+, Ghrelin−/−, Ghsr+/+, and Ghsr−/− mice display similar changes in body composition in response to chronic CR. Fat mass and lean mass after 1 year of ad libitum (AL) and caloric restriction (CR) feeding in (A) Ghrelin+/+ (white bars) and Ghrelin−/− (black bars) and (B) Ghsr+/+ (dashed bars) and Ghsr−/− (checkered bars) mice. Two-way ANOVA results indicated. Bonferroni post hoc testing; ***, P < .001 vs respective AL; #, P < .05 AL-Ghsr−/− vs AL-Ghsr+/+.

Chronic CR alters energy expenditure, anticipatory feeding behavior, and substrate utilization similarly in Ghrelin+/+, Ghrelin−/−, Ghsr+/+, and Ghsr−/− mice

To further investigate metabolic adaptation to long-term CR in mice with and without intact ghrelin signaling, we measured VO₂ (energy expenditure), ambulatory activity, and substrate utilization in real time, and analyzed average values obtained during the light and dark phases using 2-factor (diet, genotype) ANOVA. The average rate of VO₂ (per g lean body mass) was similarly increased (∼15%, P < .05) in all 4 groups of CR vs AL mice during the light phase (Figure 2A, left panels), with differences evident primarily at the end of the light phase, which coincided with the time of daily feeding (Figure 2B). Conversely, there was a decrease in average VO₂ per g lean body mass in CR vs AL mice during the dark phase, again primarily at the end of the phase, but this was significant in Ghrelin+/+ and Ghrelin−/− mice only (Figure 2A, right panels). Coinciding with changes in VO₂, ambulatory activity levels were significantly elevated in CR mice during the light phase (Figure 3A, left panels) around the time of daily feeding (Figure 3B). As with VO₂, changes in ambulatory activity in anticipation of food were indistinguishable across Ghrelin−/−, Ghsr−/−, and wild-type (WT) littermate controls (Figure 3, A and B).

Figure 2.

Energy expenditure is similarly altered with chronic caloric restriction (CR) in Ghrelin+/+, Ghrelin−/−, Ghsr+/+, and Ghsr−/− mice. A, Average VO₂ during the light (left) and dark phase (right) after 1 year of ad libitum (AL) and caloric restriction (CR) feeding in Ghrelin+/+ vs Ghrelin−/− (upper panel), and Ghsr+/+ vs Ghsr−/− mice (lower panel). Two-way ANOVA results indicated. Bonferroni post hoc testing; *, P < .05; **, P < .01 vs respective AL; no significant differences observed between genotypes. B, Changes in VO₂ illustrated over time in AL-Ghsr+/+ (blue), CR-Ghsr+/+ (red), AL-Ghsr−/−, and CR-Ghsr−/− mice, with black bars indicating dark phases.

Figure 3.

Ambulatory activity is similarly altered with long-term caloric restriction (CR) in Ghrelin+/+, Ghrelin−/−, Ghsr+/+, and Ghsr−/− mice. A, Average ambulatory activity counts during the light (left) and dark phase (right) after 1 year of ad libitum (AL) or caloric restricted (CR) feeding in Ghrelin+/+ vs Ghrelin−/− (upper panel), and Ghsr+/+ vs Ghsr−/− mice (lower panel). Two-way ANOVA results indicated. Bonferroni post hoc testing; *, P < .05; ***, P < .001 vs respective AL; no significant differences observed between genotypes. B, Changes in ambulatory activity illustrated over time in AL-Ghsr+/+ (blue), CR-Ghsr+/+ (red), AL-Ghsr−/−, and CR-Ghsr−/− mice, with black bars indicating dark phases.

The ratio of VCO_{2 expired} to VO_{2 consumed}, or respiratory exchange ratio (RER), was used to estimate whole-body substrate utilization. Higher values reflect a preference for glucose (100% glucose, 1.00), and lower values reflect a preference for lipid (100% lipid, 0.70). RER was significantly reduced in CR vs AL mice during the light phase (diet, P < .0001) (Figure 4A), indicating more predominant fat utilization, and this was observed regardless of genotype (Figure 4, A and B). As illustrated in 24-hour RER patterns shown in Figure 4B, all 4 groups of CR mice displayed steep and pronounced changes in RER values when transitioning between the light and dark phases, which highlighted the ability to rapidly cycle between glucose and lipid as the primary oxidative substrate when active and at rest, respectively. In contrast, AL mice displayed much less variable RER values over the course of 24 hours, consistent with loss of “metabolic flexibility.” Importantly, the pronounced differences in 24-hour RER patterns in CR vs AL mice were indistinguishable in Ghrelin−/−, Ghsr−/−, and WT littermates (Figure 4, A and B).

Figure 4.

Substrate utilization is similarly altered with chronic caloric restricted (CR) in Ghrelin+/+, Ghrelin−/−, Ghsr+/+, and Ghsr−/− mice. A, Average RERs during the light (left) and dark phase (right) after 1 year of ad libitum (AL) or caloric restricted (CR) feeding in Ghrelin+/+ vs Ghrelin−/− (upper panel), and Ghsr+/+ vs Ghsr−/− mice (lower panel). Two-way ANOVA results indicated. Bonferroni post hoc testing; **, P < .01; ***, P < .001 vs respective AL; no significant differences observed between genotypes. B, Changes in VO₂ illustrated over time in AL-Ghsr+/+ (blue), CR-Ghsr+/+ (red), AL-Ghsr−/−, and CR-Ghsr−/− mice, with black bars indicating dark phases.

Chronic CR alters hypothalamic gene expression similarly in Ghrelin+/+, Ghrelin−/−, Ghsr+/+, and Ghsr−/− mice

GHSR1a is expressed most abundantly in the hypothalamus; therefore, we assessed whether changes in hypothalamic gene expression associated with long-term CR required ghrelin signaling. As expected, we observed reduced hypothalamic Pomc expression in CR vs AL mice (diet P < .001). However, the effect of CR was similar in all 4 groups of mice (Figure 5A). Consistent with reduced Pomc expression, hypothalamic Agrp and Npy expression were increased in CR vs AL mice (diet P = .03); however, differences were again independent of genotype (genotype, D × G P > .40) (Figure 5, B and C).

Figure 5.

Chronic CR alters hypothalamic gene expression similarly in Ghrelin−/−, Ghrelin+/+, Ghsr−/−, and Ghsr+/+ mice. Pomc (A), Agrp (B), and Npy (C) mRNA expression (endogenous control, 36B4); in hypothalamus from ad libitum or calorically restricted Ghrelin−/−, Ghrelin+/+, Ghsr−/−, and Ghsr+/+ mice. Two-way ANOVA results indicated. Bonferroni post hoc testing; *, P < .05; **, P < .01; ***, P < .001 vs respective AL; no significant differences observed between genotypes.

Effects of chronic CR on hepatic steatosis and gene expression in the liver of Ghrelin−/− and Ghsr−/− mice

Histological analyses of liver tissue sections revealed healthy tissue in all 4 groups of CR mice, and conversely, visible lipid droplets in the 4 groups of AL-fed mice (Figure 6A). Consistent with the tissue morphology across genotypes, hepatic expression of Cidea, Cidec, Pparg2, and Apoa4 were markedly reduced in CR mice, with increased expression of the insulin receptor substrate, Irs2 (Figure 7). Expression of Irs1 was unaffected (Figure 7). As a control for reduced expression of Pparg2 that was predictably associated with low fat accumulation, we measured Ppara expression, because it is involved in fatty acid oxidation rather than adipogenesis; Ppara expression was unaffected by CR (Figure 7). Notably, the degree of hepatic steatosis was visibly lower in AL-Ghrelin−/− and AL-Ghsr−/− mice compared with their AL-WT counterparts (Figure 6A); this was accompanied by corresponding reductions in expression of Cidea and Cidec (Figure 6B). In the case of AL-Ghrelin−/− mice, reduced steatosis was accompanied by reduced expression of Pparg2, and with AL-Ghsr−/− mice reduced steatosis was associated with increased expression of Irs2 (Figure 7).

Figure 6.

AL-fed Ghrelin−/− and Ghsr−/− mice are resistant to age-associated hepatic steatosis. A, Representative hematoxylin and eosin staining of liver sections from ad libitum (AL) or calorically restricted (CR). Ghrelin+/+, Ghrelin−/−, Ghsr+/+, and Ghsr−/− mice. B, Corresponding Cidea (left panel) and Cidec (right panel) mRNA expression (endogenous control, thiodoredoxin-binding protein) in Ghrelin+/+ (white bars), Ghrelin−/− (black bars), Ghsr+/+ (dashed bars), and Ghsr−/− (checkered bars) mice. Two-way ANOVA results with Bonferroni post hoc testing; ***, P < .001 vs AL; #, P < .05; ##, P < .01 vs respective AL-WT group.

Figure 7.

Effects of long-term CR on expression of genes associated with fat accumulation in the liver of Ghrelin+/+, Ghrelin−/−, Ghsr+/+, and Ghsr−/− mice. Corresponding expression of Pparg2, Ppara, Apoa4, Irs2, and Irs1 mRNA (endogenous control, thiodoredoxin-binding protein) in ad libitum (AL) or calorically restricted (CR) mice. Left panels, Ghrelin+/+ (WT, white bars) and Ghrelin−/− (GKO, black bars). Right panels, Ghsr+/+ (WT, white bars) and Ghsr−/− (GHSRKO, black bars). Two-way ANOVA with Bonferroni post hoc testing; *, P < .05; **, P < .01; ***, P < .001 vs AL; #, P < .05; ##, P < .01 vs respective AL-WT group.

Circulating concentrations of ghrelin and DAG are increased with chronic CR

Because there were no major differences in the response of WT and Ghrelin−/− and Ghsr−/− mice to CR, we measured concentrations of ghrelin and DAG in plasma from these mice. Compared with AL-fed animals, circulating concentrations of ghrelin (Figure 8A) and DAG (Figure 8B) were significantly increased with chronic CR in both Ghrelin+/+ (upper panels) and Ghsr+/+ mice (lower panels). As a result of larger increases in DAG, the ratio of ghrelin to DAG was reduced approximately 2-fold in CR vs AL WT mice (Figure 7C), likely reflecting lower levels of octanoyl substrate for DAG under CR conditions. As expected, neither ghrelin, nor DAG, was detected in Ghrelin−/− mice (Figure 8, A and B, upper panel). Similar to Ghsr+/+, Ghsr−/− mice displayed an approximately 6-fold increase in DAG levels in response to CR (4457 ± 1481 vs 790 ± 101 pg/mL, P < .01) (Figure 8B, lower panel). In contrast to CR-Ghsr+/+ mice, CR-Ghsr−/− mice did not produce a significant increase in plasma ghrelin (Figure 8A, lower panel); nevertheless, the relative ratio of ghrelin to DAG was lowered by CR in both Ghsr+/+ and Ghsr−/− groups (Figure 8C, lower panel).

Figure 8.

Circulating concentrations of active ghrelin are reduced with long-term caloric restriction (CR). Blood plasma levels of (A) acyl-ghrelin and (B) des-acyl-ghrelin in ad libitum (AL)- or caloric restricted (CR)-Ghrelin+/+, Ghrelin−/−, Ghsr+/+, and Ghsr−/− mice after 1 year. C, Corresponding acyl to DAG ratios for Ghrelin+/+, Ghsr+/+, and Ghsr−/− groups. Two-way ANOVA with Bonferroni post hoc testing; *, P < .05; **, P < .01; ***, P < .001 vs respective AL; #, P < .05; ###, P < .001 vs respective WT.

Discussion

Chronic CR resulting in negative energy balance is a demonstrated means of prolonging health-span and preventing age-related metabolic diseases. Despite extensive efforts, the physiological mechanisms that mediate the beneficial effects of long-term CR are unknown. Because CR is associated with increased circulating levels of ghrelin and many effects of CR are mimicked by ghrelin administration, the present study is the first to investigate whether the ghrelin-GHSR1a axis is required for long-term metabolic adaptation to CR. Consistent with short-term CR studies in rodents (10, 11, 36), we observed an increase in ghrelin and DAG with prolonged CR. Active ghrelin was increased to a lesser degree; hence, the relative ratio of ghrelin to DAG decreased. This is consistent with studies in humans showing that under AL conditions, circulating concentrations of active ghrelin increase pre- and decrease postprandially. In addition to reducing the concentration of medium chain fatty acid GOAT substrate, prolonged fasting down-regulates GOAT expression and lowers the proportion of ghrelin to DAG in the circulation and in the stomach (35, 37, 38). Intriguingly, the increase in circulating ghrelin concentrations in response to CR in Ghsr+/+ mice, with a lesser effect in Ghsr−/− mice, suggested that GHSR1a regulates Goat activity. Although in both Ghsr+/+ and Ghsr−/− mice, DAG levels increased during CR, the identical metabolic phenotype across CR genotypes showed that DAG does not mediate metabolic changes associated with CR.

Across the 4 genotypes tested, 12 months of CR inhibited liver steatosis. In each case, reduced steatosis correlated with reduced expression of Cidea, Cidec, and Pparg2, which are characteristic of fat accumulation. In contrast to Pparg2, which is required for adipogenesis, Ppara, which regulates peroxisome proliferation and fatty acid oxidation in the rodent liver, was unaffected by CR. Intriguingly, the relative tissue distribution of Cidea expression changed during aging; as expression increased in the liver, expression of Cidea in sc fat markedly declined and was attenuated 42-fold in 12-month-old AL-Ghsr+/+ and AL-Ghsr−/− mice (39). The increase in Irs2 expression in response to CR is consistent with enhanced insulin signaling in the liver; Irs1 expression was unaffected. Although both Irs1 and Irs2 play complementary roles in regulating hepatic metabolism, Irs1 is linked to glucose homeostasis, whereas Irs2 is more closely linked to lipid metabolism; hence, the increase in Irs2 expression is consistent with reduced steatosis (40).

In the AL context, Ghrelin−/− and Ghsr−/− mice displayed reduced hepatic steatosis compared with their WT controls. During the study, periodic measurement of AL food intake was performed and found to be identical across genotypes; hence, reduced steatosis was unrelated to either AL food intake, or increased energy expenditure. We suggest the reduced steatosis is a consequence of improved insulin sensitivity exhibited by Ghrelin and Ghsr knockout mice (8, 23, 41). Enhanced insulin sensitivity is more pronounced in Ghsr−/− mice and correlated here with increased expression of Irs2 in the liver of AL-Ghsr−/− mice. Fat mass, lean mass, energy expenditure and locomotor activity were identical across genotypes; similar observations were made by Pfluger et al, although they showed that inactivation of both Ghsr and Ghrelin resulted in reduced fat mass accompanied by increased motor activity and energy expenditure (42). The mechanism that explains the contrasting phenotype in the double knockouts is unknown, but we suggest dopamine dopamine-2 receptor (D2R) and perhaps D1R signaling plays a role, because independent of ghrelin, GHSR1a is essential for regulating dopamine signaling in the brain through GHSR1a:D2R and GHSR1a:D1R heteromers (43, 44).

Despite evidence that nutrient status regulates circulating concentrations of ghrelin, after 1 year of CR, the metabolic phenotype of Ghrelin−/− and Ghsr−/− mice was indistinguishable from that of CR-WT littermates; furthermore, the potential metabolic importance of DAG activity dependent or independent of GHSR1a was also not supported by our results. In direct contrast to AL feeding, regardless of genotype, CR provides remarkably robust metabolic flexibility, which is likely the basis for the health benefits of CR. These data provide the first direct evidence that neither ghrelin, nor DAG signaling are involved in the metabolic benefits of CR.

Exogenous administration of pharmacologic doses of ghrelin enhances food intake and increases Agrp and Npy and decreases Pomc expression. As anticipated, CR increased plasma ghrelin levels in WT mice, which was accompanied by increased expression of hypothalamic Agrp and Npy, with reduced expression of Pomc. Surprisingly, identical effects on Agrp, Npy, and Pomc expression were observed in CR Ghrelin−/− and Ghsr−/− mice; hence, these changes in expression were dependent on diet (P < .05) rather than genotype (P > .4). Food anticipatory behavior in the CR groups was also indistinguishable in Ghrelin−/− and Ghsr−/− mice; hence, in agreement with previous studies conducted in AL-fed Ghrelin, Ghsr, and Goat knockout mice, our new results contradict the widely held belief of a direct physiological role for ghrelin as a regulator of food intake (8, 23, 24, 33). Indeed, in studies, where ghrelin producing cells were ablated in adult mice, when ghrelin was administered to restore plasma ghrelin to preprandial concentrations, food intake was unaffected; rather, stimulation of appetite required supraphysiological concentrations of ghrelin (21). In addition to expressing GHSR1a, Agrp/Npy neurons express insulin receptors, and hypothalamic insulin action counterregulates ghrelin (45, 46). When insulin production is lowered, or when hypothalamic insulin receptors are reduced by siRNA knockdown, Agrp and Npy expression increases producing hyperphagia (47). Because the pharmacologic doses of ghrelin required to stimulate food intake inhibit insulin secretion (8), we speculate that the effect of exogenous ghrelin on appetite is indirect and is a result of reducing insulin tone on hypothalamic neurons.

It is possible that the lack of differences in some of the endpoints observed could be due to developmental compensation for the ablated gene; for example, inactivation of Npy in neonates had only modest effects on food intake, whereas ablation of NPY neurons in adult mice led to starvation (48). However, in the case of Ghrelin−/− mice, the feeding phenotype was identical to WT mice regardless of whether Ghrelin was inactivated in embryonic stem cells or when ghrelin producing cells were eliminated in mature mice (21). To our knowledge, Ghsr has never been inactivated in adult mice to test for possible compensation; nevertheless, indirect evidence argues against compensation for Ghsr deletion. In contrast to WT mice, congenital Ghsr−/− mice are resistant to suppression of food intake by a D2R agonist, and resistant to D1R agonist-induced initiation of hippocampal synaptic plasticity (43, 44). However, both phenotypes were recapitulated by pharmacologically inactivating GHSR1a in adult WT mice (43, 44), suggesting that developmental compensation for Ghsr ablation is unlikely.

In conclusion, by utilizing ghrelin and ghrelin receptor null mice and measuring metabolic parameters under conditions of chronic CR, despite the common benefits of CR and administration of GHSR1a agonists during aging, in the case of CR, the benefits are not dependent on GHSR1a signaling. Furthermore, collectively the data generated are inconsistent with the traditional view of endogenous ghrelin as a physiological regulator of energy intake and metabolism. Indeed, the collective data add credence to the notion that ghrelin regulates fat deposition, and is necessary for ensuring survival under conditions of severe famine (24), or acute food shortage exacerbated by stress caused by low ambient temperatures (49).