Genetic deletion of ghrelin does not decrease food intake but influences metabolic fuel preference

Abstract

Ghrelin is a recently identified growth hormone (GH) secretogogue whose administration not only induces GH release but also stimulates food intake, increases adiposity, and reduces fat utilization in mice. The effect on food intake appears to be independent of GH release and instead due to direct activation of orexigenic neurons in the arcuate nucleus of the hypothalamus. The effects of ghrelin administration on food intake have led to the suggestion that inhibitors of endogenous ghrelin could be useful in curbing appetite and combating obesity. To further study the role of endogenous ghrelin in appetite and body weight regulation, we generated ghrelin-deficient (ghrl^–/–) mice, in which the ghrelin gene was precisely replaced with a lacZ reporter gene. ghrl^–/– mice were viable and exhibited normal growth rates as well as normal spontaneous food intake patterns, normal basal levels of hypothalamic orexigenic and anorexigenic neuropeptides, and no impairment of reflexive hyperphagia after fasting. These results indicate that endogenous ghrelin is not an essential regulator of food intake and has, at most, a redundant role in the regulation of appetite. However, analyses of ghrl^–/– mice demonstrate that endogenous ghrelin plays a prominent role in determining the type of metabolic substrate (i.e., fat vs. carbohydrate) that is used for maintenance of energy balance, particularly under conditions of high fat intake.

Ghrelin is a 28-aa peptide produced predominantly in the stomach (1, 2) that has recently been identified as a ligand of the growth hormone (GH) secretogogue (GHS) receptor (GHS-R). Like other GHSs, activation of the receptor stimulates GH secretion from the pituitary gland (1). In addition to inducing GH release, administration of exogenous ghrelin also stimulates food intake and body weight gain (3–7), increases gastric motility and acid secretion (8, 9), and decreases lipid metabolism in mice and rats (3, 4). The effects of centrally administered ghrelin on food intake are independent of its ability to induce GH release and thought to result from its direct actions on the arcuate nucleus of the hypothalamus. Furthermore, recent studies have demonstrated that plasma ghrelin levels increase preceding meals and during fasting (10, 11). Thus, it has been suggested that ghrelin stimulates appetite and that inhibitors of endogenous ghrelin, therefore, could prove useful in reducing food intake and combating obesity (11).

Supporting the possibility that ghrelin acts as a key regulator of appetite and food intake by actions on the hypothalamus, GHS-R is colocalized with neuropeptide Y (NPY)/agouti-related protein (AgRP) neurons (12) in the arcuate nucleus, a region that is responsive to circulating peripheral nutrients and hormones and critically involved in the regulation of food intake (13). Indeed, ghrelin stimulates the spontaneous activity of these neurons (14), and central ghrelin administration increases NPY and AgRP gene expression (15). Moreover, ghrelin-immunoreactivity has been reported in the hypothalamus (14, 16). Thus, circulating ghrelin released from the stomach, or centrally released ghrelin acting in a paracrine manner, could function to modulate appetite and body weight via an interaction with these orexigenic neurons. There is considerable evidence that exogenous ghrelin can markedly increase food intake and body weight when administered directly into the ventricles, and increases in plasma ghrelin levels observed before meals and during fasting have circumstantially linked ghrelin to the hunger response. However, peripheral administration of ghrelin exerts, at best, a very modest increase in food intake and body weight in rodents (3), and plasma ghrelin levels are reduced, rather than elevated, in genetically obese and hyperphagic rodents (17) and obese humans (18, 19). Certain mutations in the ghrelin/preproghrelin gene have been tenuously linked with early onset obesity, but the functional significance of these mutations remains unclear (20, 21). Moreover, a preliminary study of ghrelin-deficient mice indicates that ghrelin may not function as a critical regulator of food intake (22). Thus, the role of endogenous ghrelin in the regulation of food intake and body weight has not been definitively established.

To further define the role of endogenous ghrelin, we generated and characterized mice in which the ghrelin gene was deleted and replaced with a reporter gene (lacZ). The results of these studies indicate that endogenous ghrelin is not essential for the maintenance of normal levels or patterns of food intake, or increases in food intake after a fast. Thus, ghrelin, at most, appears to play a redundant role in the regulation of appetite. In contrast, our studies demonstrate that endogenous ghrelin plays a prominent role in regulating energy substrates (i.e., fat vs. carbohydrate) used for maintenance of energy balance, particularly under conditions of high fat intake.

Materials and Methods

Generation of ghrl-Deficient Mice and Experimental Procedures. Mice were generated by using the high-throughput homologous recombination Velocigene technology described in ref. 23. Briefly, bacterial artificial chromosome (BAC)-based targeting vectors, in which the coding region of the ghrl locus (from ATG initiation codon to the termination codon) was precisely deleted and replaced with an in-frame lacZ reporter gene and neomycin selectable marker, were electroporated into embryonic stem (ES) cells. Correctly targeted ES cells, as well as eventual heterozygote and homozygous mice derived from these ES cells, were identified by a real-time PCR-based “loss-of-native-allele” assay as described in ref. 23. Two sets of primers were used, the first of which specifically amplified the wild-type/native ghrl^+/+ allele (F1 5′-TAAAGGGGTTGGGGTATGGAGG-3′ and R1 5′-ACCAGAGAGGAAGGTAGAAGGAGTG-3′) and the second of which specifically amplified the null ghrl^–/– allele (F2 5′-GGTCAATCCGCCGTTTGTTC-3′ and R2 5′-ACCATTTTCAATCCGCACCTC-3′). After germ-line transmission was established, mice were backcrossed to C57BL6/J to generate N2 breeding heterozygote pairs that were used to generate homozygous null mice. All experiments reported were conducted on such N2F2 littermates that were housed under 12 h of light per day in a temperature-controlled environment. All procedures were conducted in compliance with protocols approved by the Regeneron Institutional Animal Care and Use Committee. Animals had free access to either standard chow (#5020, Purina) or high-fat diet (45% fat, #93075, Harlan Teklad, Madison, WI) unless otherwise specified.

Indirect Calorimetry, Food Intake, and Body Composition. Metabolic measurements and body composition were assessed both on standard chow (at 8–10 weeks of age) and after 6 weeks exposure to a high-fat diet (at 14–16 weeks of age). Metabolic parameters were obtained by using an Oxymax (Columbus Instruments, Columbus, OH) open circuit indirect calorimetry system as described in ref. 24. Briefly, O₂ consumed (ml/kg/h) and CO₂ generated (ml/kg/h) by each animal were measured for a 48-h period, and metabolic rate (VO₂) and respiratory quotient (RQ) (ratio of VCO₂/VO₂) were then calculated. Activity (counts) was also measured during the 48-h period. Energy expenditure (or heat) was calculated as the product of the calorific value of oxygen (= 3.815 + 1.232 × RQ) and the volume of O₂ consumed. Food intake was also assessed by automated measurements in metabolic cages, body composition was determined pre- and post-high-fat diet for each individual animal by using dual-emission x-ray absorption (pDEXA, Norland Medical System, Fort Atkinson, WI), and the percentage of lean muscle and fat mass was calculated. Experiments were performed in one group of male mice (eight ghrl^+/+ and eight ghrl^–/– mice) and one group of female mice (five ghrl^+/+ and nine ghrl^–/– mice).

Tissue and Serum Analysis. Basal serum samples were taken between 1000 and 1200 h and analyzed for glucose, triglycerides, and cholesterol by using the Bayer (Tarrytown, NY) 1650 blood chemistry analyzer. Nonesterified free fatty acids (NEFA) were analyzed by a diagnostic kit (Wako, Richmond, VA) and insulin levels by LincoPlex (Linco Research Immunoassay, St. Charles, MO). Data were generated by using one group of five ghrl^+/+ and five ghrl^–/– male mice. Tissues for Northern blot analysis were rapidly dissected and immediately frozen at –80°C until RNA was isolated by using TRIzol reagent (Invitrogen) as described in ref. 24.

Histology. Adult mice were deeply anesthetized (240 mg/kg ketamine, 48 mg/kg xylazine, i.m.) and exsanguinated with ice-cold heparinized saline. Tissue was fixed by transcardial perfusion of 2% paraformaldehyde (for β-galactosidase staining) or 4% paraformaldehyde (for immunohistochemistry) in 0.1 M phosphate buffer, postfixed for 2 h, and cryoprotected for at least 24 h in two changes of buffered 30% sucrose at 4°C with agitation before sectioning.

Immunohistochemical staining of ghrelin was performed on slide-mounted sections of stomach and small intestine (16 μm) and on free-floating coronal sections of brain (40 μm), as described in refs. 24 and 25, by using a rabbit serum against rat octanoylated ghrelin (Phoenix, Belmont, CA; 1:1,000 to 1:6,000). To visualize β-galactosidase, sections were incubated in buffered 1 mg/ml X-Gal (Molecular Probes) for 12–24 h at 37°C as described in ref. 26. The sections were counterstained with eosin.

Statistical Analysis. Data are expressed as mean ± SEM. Comparison of means was carried out by using a t test or ANOVA, where appropriate, with the program statview (SAS Institute, Cary, NC). P values <0.05 were considered significant.

Results and Discussion

Targeted Disruption of the ghrl Locus. ghrl^–/–-null mice were generated by constructing and using BAC-based targeting vectors via the Velocigene technology described in ref. 23. The targeting vector contained a precise deletion of the entire ghrelin coding region between the ATG start and stop codon, with insertion of the lacZ reporter gene (Fig. 1A). Correct targeting of the BAC vector with simultaneous loss of the wild-type/native ghrelin allele in ES cells was determined by using a quantitative loss-of-native-allele assay as described in ref. 23. Loss of allele was further confirmed by PCR analysis of DNA (Fig. 1B), as well as Northern blot analysis of RNA from wild-type (ghrl^+/+), heterozygous (ghrl^+/–), and homozygous (ghrl^–/–) mice (Fig. 1C; and see below for further discussion). We observed a normal birth ratio of ghrl^+/+, ghrl^+/–, and ghrl^–/– mice as predicted by Mendelian genetics, and all ghrl^–/– mice appeared grossly normal and reached normal development milestones during the first 8 weeks of age (data not shown). Further, male ghrl^+/+ and ghrl^–/– mice had similar body lengths at 8 weeks of age (10.2 ± 0.2 cm vs. 9.8 ± 0.2 cm, P = 0.12) and showed no difference in serum GH levels (Table 1).

Fig. 1.

Generation and validation of ghrl^–/– mice. (A) Schematic diagram of the murine wild-type ghrl allele, and the targeting vector used to generate a null ghrl allele by precise substitution of the lacZ reporter gene as well as a neo selectable marker. B, BamH1; K, KpnI; P, PstI; also depicted are primers F1, R1, F2, and R2 for PCR assays shown in B.(B) PCR assays, as described in Materials and Methods, distinguish and identify the wild-type and null ghrl alleles found in ghrl^+/+, ghrl^+/–, and ghrl^–/– mice; primers F1 and R1 (as depicted in A) detect the 774-bp wild-type fragment in the ghrl^+/+ and ghrl^+/– mice but not in the ghrl^–/– mice, whereas primers specific for the introduced lacZ gene on the null allele (F2 and R2 as depicted in A) detect a fragment of a 697-bp product in the ghrl^+/– and the ghrl^–/– mouse DNA. (C) RNA was isolated from stomach, gastrocnemius muscle, adipose, and hypothalamic tissue of ghrl^+/+ and ghrl^–/– mice and probed by Northern blotting for ghrl expression with a full-length ghrl cDNA probe.

Table 1. Serum parameters in male ghrl^+/+ and ghrl^-/- mice maintained on standard chow in both nonfasted and fasted states.

	Nonfasted		Fasted
Parameter	Ghrl+/+ Nonfasted	Ghrl-/- Nonfasted	Ghrl+/+ Fasted	Ghrl-/- Fasted
GH, ng/ml	10.0 ± 1.02	8.9 ± 0.45	No data	No data
Glucose, mg/dl	303 ± 25	326 ± 43	188 ± 10	206 ± 10
Insulin, ng/ml	0.77 ± 0.17	0.55 ± 0.09	0.61 ± 0.25	0.53 ± 0.12
Triglycerides, mg/dl	113 ± 17	104 ± 4	64 ± 6	56 ± 3
Cholesterol, mg/dl	132 ± 17	121 ± 10	86 ± 6	82 ± 8
NEFA, meq/liter	1.33 ± 0.25	1.02 ± 0.10	0.83 ± 0.03	0.72 ± 0.09

Ghrelin and the lacZ Reporter Gene Are Expressed Robustly in the Stomach but at Negligible Levels in the Hypothalamus. Northern blot analysis of total tissue RNAs confirmed the previously reported high level of ghrelin expression in the stomachs of ghrl^+/+ mice and the expected ablation of such transcripts in ghrl^–/– mice (Fig. 1C). Analysis of hypothalamic RNA, using muscle and adipose RNAs as controls, failed to detect endogenous ghrelin mRNA expression in this tissue (Fig. 1C). To explore the expression pattern and localize the ghrelin-expressing cells with higher resolution, we performed immunostaining for endogenous ghrelin on tissues from ghrl^+/+ mice by using ghrl^–/– mice as specificity controls and compared the pattern of ghrelin-immunoreactivity in ghrl^+/+ littermates to β-galactosidase staining for the lacZ reporter gene in ghrl^+/– and ghrl^–/– mice. Consistent with the Northern blot analysis, dense ghrelin immunostaining was noted in a distinctive cell subpopulation in the stomach in the ghrl^+/+ mice. As expected, ghrelin staining was absent in the stomachs of ghrl^–/– mice and was replaced by strong reporter gene expression (Fig. 2 Top). Occasional ghrelin- or lacZ-positive cells could also be detected in the small intestine (Fig. 2 Middle), commensurate with the low level of ghrelin mRNA expression previously observed by Northern blot analysis in this tissue. In contrast to previous reports (14, 16), we could not detect any specific ghrelin immunostaining in the hypothalamus, in the vicinity of the arcuate nucleus, adjacent to the third ventricle, or between the dorsomedial and ventromedial nuclei of the hypothalamus (Fig. 2 Bottom Left). However, this lack of specific staining could be attributable to reduced specificity of this antisera in hypothalamic tissue (Fig. 2 Bottom Left). Similarly, β-galactosidase staining for the introduced lacZ reporter gene could not be detected in any part of the hypothalamus (Fig. 2 Bottom Right). However, using a TaqMan PCR, we were able to detect low levels of endogenous ghrelin mRNA in the hypothalamus of ghrl^+/+ mice, which were specifically deleted in ghrl^–/– mice.

Fig. 2.

Cellular expression of ghrelin in adult mice revealed by immunostaining and reporter gene expression. To determine the cells expressing ghrelin in adult mice, tissues were subjected to immunostaining with ghrelin-specific polyclonal antibodies (Left and Middle) as well as β-galactosidase staining for the introduced lacZ reporter gene (Right; with dark blue staining indicating reporter gene expression and a pink eosin counterstain). To control for the specificity of the immunostaining, staining of tissues from wild-type ghrl^+/+ mice was compared to staining of tissues from ghrl^–/– mice; only staining specificto ghrl^+/+ mice was considered specific. Comparisons of immunostaining and reporter gene expression revealed matching expression patterns for a large subset of cells in the stomach (Top), and for infrequent cells in the small intestine (Middle); Insets show higher magnification views of expressing cells. In contrast, while some lightly stained cells were identified in the hypothalamus of ghrl^+/+ mice, an identical staining pattern was found in ghrl^–/– mice (Bottom), and no β-galactosidase positive cells were present in the hypothalamus. (Scale bar, 100 μm.) DM, dorsomedial hypothalamus; VMH, ventromedial hypothalamus; Arc, arcuate nucleus.

Ghrelin Deletion Does Not Impair Spontaneous Food Intake or Alter Basal Expression of Hypothalamic Orexigenic Neuropeptides. We hypothesized that if ghrelin does indeed play a major role in promoting appetite or initiating feeding, then genetic ablation of ghrelin should decrease body weight and food intake. When fed standard chow, no significant differences in the body weights of male or female ghrl^–/– mice were observed as compared with ghrl^+/+ littermates (Fig. 3A). Consistent with this, there were no significant differences in total 24-h food intake (Fig. 3B), or the normal circadian pattern of spontaneous food intake, when mice were assessed at 8–10 weeks of age (Fig. 3C). To further evaluate the possibility that ghrelin may regulate appetite signals, we next studied the expression of hypothalamic neuropeptides in the ghrl^–/– mice. Northern blot analysis did not reveal any differences in basal expression levels of AgRP, melanin-concentrating hormone (MCH), proopiomelanocortin, VGF, or NPY between ghrl^–/– and ghrl^+/+ mice (Fig. 4A). Moreover, after a 24-h fast, food intake was not impaired in either male (Fig. 4B) or female ghrl^–/– mice (24-h food intake after fast, 5.0 ± 0.5 g in female ghrl^+/+ mice vs. 4.93 ± 0.2 g in female ghrl^–/– mice), indicating that neural pathways that positively regulate appetite after a food deprivation challenge respond in a physiologically normal manner. In fact, there was a consistent trend for the male ghrl^–/– mice to eat more than their ghrl^+/+ littermates under conditions of ad libitum access to food as well as refeeding after a fast. No differences in locomotor activity were observed that would account for the trend toward increased food intake (data not shown).

Fig. 3.

Absence of ghrelin does not decrease body weight or food intake, or alter metabolic parameters or body composition on a standard chow diet. (A) Body weights for male ghrl^–/– mice and ghrl^+/+ littermate mice were determined every week from birth. Data represent the mean of at least 5–24 mice of each genotype at each time point; weights for male littermates are depicted, and no differences were noted for female ghrl^–/– and ghrl^+/+ littermates (data not shown). Twenty-four-hour food intake (B), food intake per interval (C), BMR (D), and RQ (E) were determined in metabolic cages over a 48-h period on 8- to 10-week-old male mice maintained on a standard chow diet. BMR was calculated from the mean oxygen consumed (ml/kg/h) during the light period for each genotype. Dual-emission x-ray absorption calculated percentage lean (F) and fat (G) mass for each genotype. For B–G, data represent mean ± SEM of n = 8 mice. Bars in C and E represent the dark period.

Fig. 4.

Ghrelin-deficient mice show normal regulation of hypothalamic orexigenic signals. (A) Basal hypothalamic neuropeptide expression in nonfasted male ghrl^+/+ and ghrl^–/– mice was not altered by ghrelin deletion. (B) Food intake after a 24-h fast, as measured in metabolic cages, was not decreased but slightly elevated in ghrl^–/– mice compared with ghrl^+/+ littermates. Data represent mean ± SEM of n = 8 mice.

Metabolic Rate and Fuel Preference Are Not Significantly Altered in ghrl^–/– Mice on Normal Diet. We next examined the metabolic characteristics of ghrl^–/– mice by using indirect calorimetry. On a standard chow diet, ghrl^–/– mice showed no significant differences in their basal metabolic rate (BMR) (Fig. 3D) or differential utilization of carbohydrates or fats as indicated by the RQ (Fig. 3E) compared to ghrl^+/+ littermates. Moreover, under these conditions, we could not detect any differences in serum glucose, insulin, triglycerides, cholesterol, or nonesterified fatty acids (Table 1), or differences in body composition, as determined by dual-emission x-ray absorption analysis in ghrl^–/– mice compared with ghrl^+/+ littermates (Fig. 3 F and G).

ghrl^–/– Mice Exhibit Altered Utilization of Substrates When Subjected to High-Fat Diet. When placed on a high-fat diet (45% fat) for 6 weeks, body weight increased to a similar extent in female (not shown) and male ghrl^–/– and ghrl^+/+ littermates (Fig. 5A). Total food intake (Fig. 5B) and food intake per interval (Fig. 5C) were also similar in ghrl^+/+ and ghrl^–/– mice. However, indirect calorimetry revealed that although there was no change in BMR (Fig. 5D), RQ was significantly decreased in the ghrl^–/– mice compared with ghrl^+/+ littermates [Fig. 5E; repeated measures ANOVA over entire time course, effect of genotype, F_(1–14) = 6.07, P = 0.027]. A similar decrease in RQ was observed in the female mice (data not shown). Decreases in RQ indicate a greater utilization of fat as an energy substrate, revealing that ghrl^–/– mice overuse fat as a fuel source when placed on a high-fat diet. Consistent with this possibility, the ghrl^–/– mice also tended to have a greater percentage of lean body mass and a smaller percentage of body fat after the high-fat diet exposure (Fig. 5 F and G; percentage lean mass, P = 0.09; percentage fat mass, P = 0.06).

Fig. 5.

Absence of ghrelin does not decrease food intake or BMR but decreases RQ on a high-fat diet. (A) Body weights for male ghrl^–/– mice and ghrl^+/+ littermate mice were determined every week during a 6-week exposure to a high-fat diet. Twenty-four-hour food intake (B), food intake per interval (C), BMR (D), and RQ (E) were determined in metabolic cages over a 48-h period after the 6-week exposure to a high-fat diet. BMR was calculated from the mean oxygen consumed (ml/kg/h) during the light period for each genotype. Dual-emission x-ray absorption calculated percentage lean (F) and fat (G) mass for each genotype. For all figures, data represent mean ± SEM of n = 8 mice. Bars in C and E represent the dark period.

Conclusions

Our evaluation of ghrl^–/– mice indicates that the principal physiological role of endogenous ghrelin lies in modulating the metabolic substrate (i.e., fat vs. carbohydrate) that is preferentially used for maintenance of energy balance, particularly under conditions of high fat intake. Such a role for endogenous ghrelin is consistent with previous findings that exogenous ghrelin administration decreases fat utilization (3). This is the only action of exogenously administered ghrelin that was reciprocally regulated in our ghrl^–/– mice. Previous studies demonstrate that a high-fat diet decreases ghrelin levels in rodents (27) and that plasma ghrelin levels also are lower in obese humans (18, 19). This reduction in ghrelin secretion in situations of positive energy balance may, together with increased leptin secretion, reflect an adaptive counterregulatory response, to push metabolic fuel preference toward lipid utilization under conditions of nutrient excess. The functional significance of ghrelin in this process is borne out by the present finding showing that when ghrelin is removed altogether, RQ is markedly reduced on a high-fat diet.

The results of the above studies also demonstrate that ghrl-deficient mice do not show appreciable abnormalities in the regulation of appetite or body weight. Although very low levels of ghrelin mRNA were detectable in the hypothalamus of wild-type mice by PCR analysis, it is unlikely that either endogenous central or peripheral ghrelin play an important role in the stimulation of food intake, given the lack of a feeding phenotype in ghrl^–/– mice (see also ref. 22). Here it is important to note that, in contrast to the profound effects on food intake and body weight that are seen with genetic ablation of the leptin and the melanocortin pathways (28, 29), the deletion of other regulators of appetite such as AgRP (30) and NPY (31) also do not have notable repercussions on basal food intake, body weight, or metabolic parameters. Subtle alterations in the feeding phenotypes of AgRP-null mice are revealed, however, in the face of an appropriate physiological challenge, particularly in the pattern of magnitude of refeeding after a fast (M.W.S. and K.E.W., unpublished observation).† However, we have not been able to detect an alteration in feeding patterns in ghrelin-null mice, even after such a challenge, nor have we been able to detect evidence of any compensatory change in the basal expression of other orexigenic or anorexigenic neuropeptides.

In contrast, we find that the constitutive absence of ghrelin causes a distinct shift toward lipid metabolism during consumption of a high-fat diet, a shift that may also be reflected in the trend toward decreased weight and leaner body composition observed in male ghrl^–/– mice after 6 weeks on the high-fat diet. In summary, our results indicate that ghrelin is not a critical orexigenic factor and, rather, support the hypothesis that ghrelin's principle physiological role may be in the determination of the type of metabolic substrate (i.e., fat vs. carbohydrate) that is used for maintenance of energy balance, particularly under conditions of high fat intake.

The ability to efficiently build fat reserves in times of nutritional abundance appears to have resulted from evolutionary pressure to protect against subsequent periods of food scarcity. The tendency to efficiently store fat in times of caloric excess appears to have become paradoxically maladaptive in settings of continuous food availability, as indicated by the present epidemic of obesity in Western societies. Our data suggest that ghrelin may be one of the primary mechanisms by which an individual can sense changes in nutrient availability and trigger biological responses that modulate the efficiency of energy storage (and particularly fat deposition) during periods of fuel overflow. Put in this context, ghrelin receptor antagonists could prove useful in controlling adiposity in human obesity associated with a high-fat diet.