The β-Hydroxybutyrate-GPR109A Receptor Regulates Fasting-induced Plasticity in the Mouse Adrenal Medulla

Abstract

During fasting, increased sympathoadrenal activity leads to epinephrine release and multiple forms of plasticity within the adrenal medulla including an increase in the strength of the preganglionic → chromaffin cell synapse and elevated levels of agouti-related peptide (AgRP), a peptidergic cotransmitter in chromaffin cells. Although these changes contribute to the sympathetic response, how fasting evokes this plasticity is not known. Here we report these effects involve activation of GPR109A (HCAR2). The endogenous agonist of this G protein–coupled receptor is β-hydroxybutyrate, a ketone body whose levels rise during fasting. In wild-type animals, 24-hour fasting increased AgRP-ir in adrenal chromaffin cells but this effect was absent in GPR109A knockout mice. GPR109A agonists increased AgRP-ir in isolated chromaffin cells through a GPR109A- and pertussis toxin–sensitive pathway. Incubation of adrenal slices in nicotinic acid, a GPR109A agonist, mimicked the fasting-induced increase in the strength of the preganglionic → chromaffin cell synapse. Finally, reverse transcription polymerase chain reaction experiments confirmed the mouse adrenal medulla contains GPR109A messenger RNA. These results are consistent with the activation of a GPR109A signaling pathway located within the adrenal gland. Because fasting evokes epinephrine release, which stimulates lipolysis and the production of β-hydroxybutyrate, our results indicate that chromaffin cells are components of an autonomic-adipose-hepatic feedback circuit. Coupling a change in adrenal physiology to a metabolite whose levels rise during fasting is presumably an efficient way to coordinate the homeostatic response to food deprivation.

Acute periods of fasting trigger the activation of a neuronal and endocrine reflex that prevents hypoglycemia by maintaining circulating levels of high-energy metabolites (1-3). A prominent component of this counterregulatory reflex is an increase in sympathoadrenal activity that includes the chromaffin cells in the adrenal medulla (4). These modified postganglionic neurons secrete the catecholamine hormones epinephrine and norepinephrine into the systemic circulation. During fasting both in humans and mice, the levels of epinephrine in the blood rise several-fold (5, 6). By stimulating hepatic glycogenolysis and gluconeogenesis, epinephrine elevates glucose release and contributes to the maintenance of euglycemia (7, 8). However, if fasting is prolonged, the supply of glucose can become limiting. To maintain central nervous system activity, which typically uses glucose as a primary energy source, many peripheral tissues switch to metabolizing ketone bodies as a means to generate adenosine 5′-triphosphate. Stored triglycerides in white adipose tissue are catabolized to free fatty acids and glycerol. Following their uptake by the liver, fatty acids are converted into ketone bodies (β-hydroxybutyrate, acetoacetate, and acetone) and subsequently released into the blood (9, 10). In addition to negative regulation by insulin and leptin, ketogenesis is also stimulated by sympathetic activity—by norepinephrine released from postganglionic neurons that innervate white adipose tissue and by epinephrine released from adrenal chromaffin cells (11, 12).

In addition to acutely stimulating epinephrine release, fasting leads to long-term changes in adrenal function (6, 13, 14), as do other forms of stress (15-17). Chromaffin cells synthesize multiple peptide cotransmitters including neuropeptide Y (NPY) and agouti-related peptide (AgRP) (14, 17-20), and the levels of both are elevated after food deprivation (6, 14, 21, 22). In mice all chromaffin cells express NPY, and approximately one-fifth of these cells express AgRP (14, 17). We have previously shown that fasting increases the strength of the preganglionic → chromaffin cell synapse and this effect is mediated by NPY release from chromaffin cells and the activation of adrenal Y5 receptors. When synaptic strengthening is prevented, epinephrine release is reduced, and the mice are hypoglycemic (6). In contrast, AgRP weakens the strength of the preganglionic → chromaffin cell synapse (by inhibiting presynaptic MC3/4 receptors) and thus the role of AgRP may be to prevent excessive catecholamine release during fasting (14). AgRP also inhibits adrenal glucocorticoid release consistent with a role in responding to an energy deficit (23-25). Curiously, while NPY and AgRP appear to have antagonistic actions at the preganglionic → chromaffin cell synapse, the adrenal levels of both peptides are elevated after food deprivation (6, 14, 21, 22). These findings indicate that there is a bidirectional connection between adrenal activity and metabolism. The output of the adrenal medulla regulates glucogenesis and ketogenesis but adrenal function itself is also sensitive to metabolic state.

Nevertheless, how a fasting condition is communicated to the adrenal is not known. Food deprivation and hypoglycemia lead to changes in the systemic levels of multiple counterregulatory hormones and signaling molecules, many of which could potentially affect adrenal physiology (3, 26). For example, fasting increases the level of circulating ketone bodies to millimolar (mM) levels (3, 9). Of particular interest is β-hydroxybutyrate, the principal ketone body. Epinephrine stimulates lipolysis and thus indirectly regulates β-hydroxybutyrate production (9, 27). Although it was originally thought to be solely a circulating metabolite, it is now known that β-OHB also acts as a hormone (28, 29). It is the physiological agonist for GPR109A/HCAR2, a G_i/o G protein–coupled receptor (GPCR) (30-33). This receptor is also activated by the antidyslipidemic and antiatherogenic drug, nicotinic acid, and is expressed by multiple cell types including adipose, immune, and epithelial cells. In adipose cells, receptor activation inhibits lipolysis while its actions on immune cells are generally anti-inflammatory (28, 29, 34). Within the nervous system, GPR109A is expressed by microglial cells and its activation has recently been shown to slow disease progression in a mouse model of Alzheimer disease (35).

Here we tested whether GPR109A and β-hydroxybutyrate are involved in the adrenal response to fasting. We find that receptor activation controls both the increase in AgRP expression and the strength of the preganglionic → chromaffin cell synapse. Coupling a change in adrenal physiology to a metabolite whose levels rise during fasting is presumably an efficient way to coordinate the neuroendocrine response to food deprivation.

Materials and Methods

Animals and Food Deprivation Paradigm

C57BL/6J wild-type (WT) mice were bred from a laboratory colony. GPR109A knockout (KO) mice were previously described (32). Individually housed mice were food deprived for 24 hours from the start of the dark cycle; water was available ad lib. Animals of both sexes were used. All experiments were approved by the institutional animal care and use committee at Louisiana State University Health Sciences Center.

Biochemical Measurement of Circulating Factors

β-Hydroxybutyrate and leptin were measured from blood samples collected from the body cavity of decapitated animals. Samples were centrifuged at 1000g at 4 °C for 15 minutes in tubes containing 1 µL of 1 mM EDTA. Plasma was then collected and stored at –80 °C. β-Hydroxybutyrate and leptin were measured using a colorimetric assay (Wako Diagnostics) and enzyme-linked immunosorbent assay (R&D Systems), respectively. Blood glucose was quantified using a handheld monitor (OptiumEZ, Abbott Diabetes Care Inc). Epinephrine and corticosterone were measured in urine samples collected over 24 hours from mice singly housed in metabolic cages (Tecniplast). Urine samples were acidified by addition of an equal volume of 0.01 M HCl and stored at –80 °C. Samples were quantified using commercial enzyme-linked immunosorbent assay kits (Abnova and Assay Designs, respectively). The levels of epinephrine and corticosterone were subsequently normalized to urine creatinine that was quantified using the Jaffe reaction (Cayman Chemicals).

Cell Culture

Chromaffin cells were dissociated from the adrenal medulla of P24-60 mice of both sexes, as previously described (17). In brief, the medulla was separated from the cortex and incubated at 37 °C for 15 minutes in HEPES-buffered salt solution (in mM: 138 NaCl, 5.3 KCl, 0.44 KH₂PO₄, 4 NaHCO₃, 0.3 Na₂HPO₄, 20 HEPES, and 5.5 glucose, pH 7.25) containing 1 mg/mL collagenase type IA plus 6 mg/mL bovine serum albumin (BSA) followed by 30 minutes in 1 mg/mL trypsin plus 6 mg/mL BSA. After digestion the medulla was triturated and cells were plated on poly-d-lysine–coated coverslips in culture medium (Dulbecco’s modified Eagle’s medium/10% fetal calf serum). In experiments in which cells were treated with agonists, the incubations were for 24 hours (to mimic the duration of the in vivo fasting paradigm). In these experiments the culture medium was exchanged to a modified version containing 6.5 mM glucose.

In Vitro Immunocytochemistry

Cells were fixed in 4% paraformaldehyde for 20 minutes at room temperature and then stained as described previously (36). In brief, cells were permeabilized for 15 minutes in 0.3% Triton in phosphate-buffered saline (PBS), blocked in 0.25% immunoglobulin G–free BSA in PBS for 30 minutes and incubated overnight at 4 °C in primary antibody. After washing in PBS, the cells were incubated in secondary antibody for 90 minutes at room temperature, then washed with PBS and mounted in Vectashield. Primary antibodies were goat anti-AgRP (1:100; Santa Cruz Biotechnology; sc-18634; RRID: AB_2258141), rabbit anti-NPY (1:40 000; Peninsula Laboratories; T4070; RRID: AB_518504), and rabbit anti-TH (1:10 000; Millipore; ab152; AB_390204; RRID: AB_390204). Secondary antibodies were donkey anti-goat DyLight 549 (1:100; Jackson ImmunoResearch), donkey anti-rabbit DyLight 488 (1:100; Jackson ImmunoResearch), and donkey anti-rabbit Alexa 488 (1:200; Invitrogen).

Primary Antibody Specificities

The specificities of the antibodies were described in a recent publication (14). In brief, the AgRP antibody stains INS-1 cells transfected with a plasmid encoding the mouse AgRP precursor and does not cross-react with the NPY antibody (37). The NPY antibody stains NPY-expressing chromaffin cells in an NPY(GFP) mouse and staining is absent in an NPY KO mouse (17). The tyrosine hydroxylase (TH) antibody stains the cytoplasm of the catecholaminergic chromaffin cells in the mouse adrenal medulla in a manner consistent with the known expression of this enzyme (17).

Image Analysis

Fluorescent images were obtained with a Nikon TE2000U microscope with a 60× oil-immersion (1.4 numerical aperture) objective, a Lambda DG-4 wavelength switcher (Sutter Instruments), and a Retiga 1300 monochrome camera. Images were captured using NIS-Elements (Nikon) software. Origin Pro7 and Excel were used for data analysis. For quantification of staining in vitro, single chromaffin cells were identified using phase contrast and the focus was adjusted to the level of the nucleus. Fluorescent images were then taken. This approach maintained a consistent plane of focus and avoided an inadvertent selection of only the brightest AgRP immunoreactive (AgRP-ir) cells. Although AgRP-ir in some figures appears diffuse, we have previously shown that AgRP-ir in chromaffin cells is punctate as expected for a molecule in the regulated secretory pathway (14). In most experiments 50 to 80 cells were photographed per treatment. The mean pixel intensity was then calculated from a region of interest (ROI; the entire cell) without background subtraction. In all staining experiments in which the level of immunoreactivity was quantified, slides were blinded until the analysis was complete. In figures where individual data points are plotted, black lines connect the average values from paired control and experimental animals (ie, 2 littermates of the same sex that were individually housed and treated in parallel). For each figure, representative cells were chosen from the 50th percentile in the cumulative intensity distribution. When intensity adjustments were made to figures, these were uniformly applied to all cells in the same panel. All quantitative analysis was performed on raw data.

In Situ Hybridization

Matched (same-sex) littermates were individually housed for 3 days, then fasted for 24 hours or left undisturbed. Mice were then euthanized by decapitation, the adrenal glands were removed, and the glands were fixed for 24 hours at 4 °C in 4% paraformaldehyde, then transferred to PBS/30% sucrose overnight. Glands were subsequently embedded in Optimal Cutting Temperature Compound and 30-µm cryosections were prepared and mounted on charged (Superfrost Plus) slides and stored at –80 °C. Adrenal AgRP messenger RNA (mRNA) was then localized using an RNAScope 2.5 Chromogenic Assay (Advanced Cell Diagnostics) following the manufacturer’s instructions with minor modifications. Probes (Mm-AgRP No. 400711) were localized using 3,3′-diaminobenzidine, and staining was intensified by incubation in 0.5% copper (II) sulphate/PBS solution for 5 minutes (38). Staining was quantified from 3 to 9 blinded sections per treatment. Individual sections were photographed in brightfield with a 10× objective. For each section, ROIs were drawn using Fiji/ImageJ (39) to select the medulla and to avoid cell-free areas. Staining was then quantified as (∑ integrated signal intensity/ ∑ ROI area) after background subtraction (measured from ROIs placed on unstained areas of medulla).

Reverse Transcription Polymerase Chain Reaction and Molecular Biology

mRNA was extracted from adrenal medulla and brown adipose tissue using TRIzol (Invitrogen) following the manufacturer’s instruction. Samples were treated with DNase (37) and purified using an RNeasy purification kit (Qiagen). Complementary DNA was synthesized by reverse transcription (RT) of 200 ng of mRNA and used in polymerase chain reaction (PCR) reactions. Primers and PCR protocols were GPR109A: forward: 5′ CTAATGGCTGACCCCAGTG 3′, reverse: 5′ GAG CAGAACTGAGGTGTGTC 3′ (386 bp, annealing Tm 60.2 °C, 35 cycles). DNA bands were excised from the agarose gel, purified using a Geneclean kit (Bio 101), and sequenced.

Electrophysiology

Evoked synaptic currents were quantified in chromaffin cells in adrenal slices as previously described (6, 14). Mice (P30-40) were euthanized using pentobarbital (150 mg/kg) and the adrenal glands were removed and embedded in low melting point–temperature agarose. Adrenal sections (350 µm) were prepared using a microtome (Leica VT1000). Sections were prepared in slicing solution (in mM: 81.2 NaCl, 2.4 KCl, 0.5 CaCl₂, 6.7 MgCl₂, 1.4 NaH₂PO₄, 23.4 NaHCO₃, 23.3 glucose, and 69.9 sucrose, pH 7.4) then transferred to ACSF (in mM: 125 NaCl, 2.5 KCl, 0.5 CaCl₂, 1 MgCl₂, 1.25 NaH₂PO₄, 26 NaHCO₃, and 25 glucose, pH 7.4) for 100 minutes at 37 °C, then maintained at room temperature. All solutions were bubbled with 95% O₂/5% CO₂. Recordings (31-33 °C) were made using a Multiclamp 700B amplifier (Molecular Devices) from chromaffin cells voltage clamped at –60 mV. The patch electrode contained intracellular solution (in mM: 120 Cs acetate, 10 EGTA-Cs, 0.4 MgCl₂, 1 CaCl₂, 1.5 Na₂ATP, 0.4 Na₂GTP, 2.5 MgATP, 1 lidocaine N-ethyl bromide (QX-314), 5 tetraethylammonium chloride, and 10 HEPES, pH 7.25, with CsOH). The preganglionic input was stimulated using a focal electrode filled with ACSF that was placed close to the recorded cell in the adrenal medulla. Series resistance was compensated by 80%. If the value changed more than 20%, the recorded cell was excluded from analysis. Synaptic currents were evoked with a paired pulse protocol (paired 1 ms depolarizations separated by 50 ms repeated every 10 seconds) in ACSF containing 0.5 mM Ca²⁺. The paired-pulse ratio (PPR) of excitatory postsynaptic currents (EPSCs; analyzed using Clampfit) was calculated by dividing the amplitude of the second EPSC by that of the first (PPR = EPSC2/EPSC1). To determine the effect of nicotinic acid on transmission, adrenal slices were incubated in ACSF (containing 0.5 mM Ca²⁺) in 100 µM nicotinic acid for 3 to 6 hours (at room temperature) and recordings were subsequently made in slices superfused with the same solution.

Statistical Tests

The Kolmogorov-Smirnov test was used to analyze cumulative frequency distributions and exact P values are noted in each figure legend. T test and one-way analysis of variance (Tukey post hoc) were used to compare normally distributed data. The nonparametric Mann-Whitney U test was used for data that was not normally distributed (assessed using the Shapiro-Wilk test). OriginPro and Clampfit were used for analysis. All values are mean ± SEM and P less than .05 was considered statistically significant.

Results

GPR109A, a Ketone Body Receptor, Regulates Adrenal Expression of Agouti-related Peptide

We have previously shown that food deprivation for 1 day leads to an increase in AgRP-ir in chromaffin cells in the mouse adrenal medulla (14). Using in situ hybridization, we confirmed that fasting also led to a large increase in AgRP mRNA in the adrenal medulla (Fig. 1A and 1B), as previously shown by RT-PCR of whole adrenal (21, 22). To test the idea that GPR109A receptors are involved in a fasting-induced increase in AgRP synthesis, chromaffin cells in vitro were incubated in 5 mM β-hydroxybutyrate (a concentration similar to that observed during fasting) for 24 hours. This led to a significant increase in the level of AgRP-ir as shown by the rightward shift in the cumulative intensity distribution (Fig. 2A). The same effect was observed following treatment with 100 µM nicotinic acid, a selective agonist of the GPR109A receptor (32) (Fig. 2B). Acipimox (100 µM) and monomethylfumarate (100 µM), which also activate the GPR109A receptor (32, 33, 40), similarly led to an increase in AgRP-ir (Fig. 2C and 2D). Studies have shown that β-hydroxybutyrate can also modulate GPR41, a short-chain fatty acid receptor (41, 42). However, incubation in 1 mM propionate, a GPR41 agonist, led to a decrease rather than an increase in AgRP-ir (Fig. 2E), indicating that GPR41 is unlikely to mediate the β-hydroxybutyrate-induced increase in AgRP expression.

Figure 1.

Fasting increases the expression of agouti-related peptide (AgRP) messenger RNA (mRNA) in the adrenal medulla. A, Examples of in situ hybridization showing expression of AgRP mRNA in the adrenal gland of fed and fasted mice. AgRP expression was present in the medulla (Med) and absent from the cortex (Ctx). B, Group data showing that fasting led to a large increase in AgRP mRNA in the adrenal medulla. Symbols show data from individual mice and lines connect data from paired animals (see “Materials and Methods”). IU; intensity units. (n = 5, mean ± SEM). P = .008, Mann-Whitney U test. Scale bar 100 µm.

Figure 2.

GPR109A agonists increase the expression of agouti-related peptide (AgRP) in chromaffin cells in vitro. A, Examples of AgRP-ir in chromaffin cells treated with 5 mM β-hydroxybutyrate for 24 hours in vitro. Right panel is a cumulative frequency distribution showing that β-hydroxybutyrate led to a significant increase in AgRP-ir (n = 3 experiments, ≥ 377 cells per distribution, P = 1.03E-10, Kolmogorov-Smirnov [K-S] test). B, 100 µM nicotinic acid led to a significant increase in the levels of AgRP-ir (n = 3 experiments, ≥ 253 cells per distribution, P = 3.4E-14, K-S test). C and D, Cumulative frequency distributions showing that 100 µM acipimox (n = 3 experiments, ≥ 246 cells per distribution, P = 6.29E-07, K-S test) and 100 µM monomethylfumarate (n = 3 experiments, ≥ 248 cells per distribution, P = 3.92E-11, K-S test) increased the levels of AgRP-ir. E, Unlike GPR109A agonists (panels A to D), which increased the levels of AgRP-ir, 1 mM propionate, a GPR41 agonist, led to a significant decrease in the levels of AgRP-ir in chromaffin cells in vitro (n = 3 experiments, ≥ 245 cells per distribution, P = 8.24E-38, K-S test). **P less than .01. Scale bar 10 µm.

No selective antagonists for GPR109A have been identified but because the receptor is G_i/o coupled, its downstream actions are inhibited in the presence of pertussis toxin (32). Consistent with a G_i/o-dependent response, the β-hydroxybutyrate-mediated increase in AgRP-ir was prevented by co-incubation with pertussis toxin (Fig. 3A and 3B). RT-PCR experiments revealed the presence of GPR109A mRNA in adrenal medulla from WT but not littermate GPR109A KO animals (Fig. 3C). Brown adipose tissue, which expresses high levels of GPR109A (32), was used as a positive control (see Fig. 3C).

Figure 3.

Effect of β-hydroxybutyrate on adrenal chromaffin cells is mediated by a pertussis-toxin sensitive G protein–coupled receptor (GPCR). A, Examples of AgRP-ir in chromaffin cells treated with 5 mM β-hydroxybutyrate, an agonist of GPR109A receptors and pertussis toxin, an inhibitor of G_i/o-coupled receptors. B, Cumulative frequency distributions confirmed that β-hydroxybutyrate incubation led to an increase in AgRP-ir and that this effect was abolished by coincubation in 100 µg/mL pertussis toxin (≥ 143 cells per distribution, n = 3 experiments). C, Reverse transcription polymerase chain reaction showing that wild-type (wt) (GPR109A^+/+) adrenal medulla contains GPR109A messenger RNA whereas amplicons were not generated using tissue from GPR109A knockout (ko) mice. Brown adipose tissue (BAT), which expresses high levels of GPR109A, was used as a positive control. **P = 5.58E-07; ns, not significant; Kolmogorov-Smirnov test. Scale bar 10 µm.

The Food Deprivation–Induced Increase in Adrenal Agouti-related Peptide Expression Is Lost in GPR109A Knockout Mice

To determine whether GPR109A receptors regulate AgRP expression in vivo, we quantified the levels of AgRP-ir in chromaffin cells acutely isolated from GPR109A KO mice that had been fed ad lib or food-deprived for 24 hours. Fasting did not increase AgRP-ir in GPR109A KO mice and in fact a small decrease was observed (Fig. 4A and 4B). In contrast, the levels of AgRP-ir were significantly higher in chromaffin cells from fasted GPR109A WT mice compared to fed animals (Fig. 4D and 4E). These experiments are consistent with the idea that the food deprivation–induced increase in adrenal AgRP requires the activity of the GPR109A receptor. Food deprivation induced a significant increase in the plasma levels of β-hydroxybutyrate both in GPR109A KO and WT littermates (Fig. 4C and 4F). There was also no difference in the levels of β-hydroxybutyrate when comparing fed and fasted KO and WT littermates (Fig. 4C and 4F; fed: P = .62, fasted: P = .32, n = 6; unpaired t test).

Figure 4.

Knockout (ko) of GPR109A receptors prevents the fasting-induced increase in agouti-related peptide (AgRP) expression in adrenal chromaffin cells. A, Examples of AgRP immunoreactivity (AgRP-ir) in chromaffin cells from fed and fasted GPR109A^–/– mice. B, Cumulative frequency distributions showing that fasting led to a small but significant decrease in the levels of AgRP-ir (n = 5 experiments, 263 cells per distribution; P = 1.43E-06; Kolmogorov-Smirnov [K-S] test) in ko mice. Right panel shows the change in mean AgRP-ir for each independent experiment (n = 5). C, Plasma levels of β-hydroxybutyrate show a significant increase in fasted GPR109A^–/– mice (n = 6, mean ± SEM). D, Examples of AgRP-ir in chromaffin cells from fed and fasted GPR109A^+/+ mice. E, Cumulative frequency distributions showing that fasting led to a significant increase in the levels of AgRP-ir (n = 6 experiments, 329 cells per distribution, P = 2.04E-10, K-S test). Right panel shows the mean change in AgRP-ir for each independent experiment (n = 6). F, Plasma levels of β-hydroxybutyrate show a significant increase in fasted GPR109A^+/+ mice (n = 6, mean ± SEM). G, Food deprivation leads to a significant decrease in the plasma levels of leptin both in GPR109A wild-type (wt) and ko littermates (n = 6, mean ± SEM). H, Urine levels of corticosterone show a significant increase after fasting both in GPR109A wt (GPR109A^+/+) and ko (GPR109A^–/–) littermates (n = 6 independent experiments, mean ± SEM). I, Examples of AgRP-ir in chromaffin cells from GPR109A^–/– mice treated with 100 µM nicotinic acid. J, Cumulative frequency distributions showing that nicotinic acid led to a significant decrease in the levels of AgRP-ir (n = 3 experiments, 163 cells per distribution, P = 2.84E-4, K-S test). Right panel shows the change in mean AgRP-ir for each independent experiment (black lines connect data points from individual experiments). K, Examples of AgRP-ir in chromaffin cells from GPR109A^+/+ mice treated with 100 µM nicotinic acid. L, Cumulative frequency distributions showing that nicotinic acid led to a significant increase in the levels of AgRP-ir (n = 3 experiments, 150 cells per distribution, P = 4.96E-3, K-S test). Right panel shows the change in mean AgRP-ir for each independent experiment (black lines connect data points from individual experiments). *P less than .05; **P less than .01 (C and F, t test; G and H, Mann-Whitney U test). Scale bar 10 µm.

Food deprivation produces a decrease in serum leptin and an increase in circulating corticosterone levels (43, 44). Both hormones regulate hypothalamic AgRP expression (45-49), and a similar mechanism may be operating in the adrenal because leptin receptors have been identified in chromaffin cells (50). AgRP also indirectly inhibits corticosterone release from adrenal cortical cells in vitro (23-25). However, a fasting-induced change in leptin and corticosterone levels was present both in GPR109A KO mice and WT littermates (Fig. 4G and 4H). Thus, the loss of the fasting-induced increase in AgRP-ir in the GPR109A KO mice was due to the absence of the GPR109A receptor and was not caused by a change in the fasting levels of β-hydroxybutyrate, leptin, or corticosterone.

Given the presence of GPR109A mRNA in the adrenal medulla (see Fig. 3C), one possibility is that during fasting, circulating β-hydroxybutyrate acts directly on the adrenal to modulate AgRP expression. To test this idea we quantified the levels of AgRP-ir in vitro from GPR109A^–/– and GPR109A^+/+ chromaffin cells after incubation in 100 µM nicotinic acid. This treatment did not increase AgRP-ir in GPR109A^–/– cells (Fig. 4I and 4J). In fact, a consistent decrease in AgRP-ir was seen (see Fig. 4J). In contrast, nicotinic acid treatment led to a significant increase in AgRP-ir in chromaffin cells from GPR109A^+/+ mice (see Fig. 4K) (50). Because GPR109A agonists increased AgRP-ir in chromaffin cells in vitro, the most parsimonious explanation is that the actions of β-hydroxybutyrate are mediated directly on the adrenal.

GPR109A Agonists Can Influence the Synthesis of Multiple Transmitters In Chromaffin Cells

During food deprivation, chromaffin cells release multiple transmitters including epinephrine and the cotransmitter NPY. Fasting elevates the adrenal level of NPY but does not increase the level of TH, the rate-limiting enzyme for catecholamine synthesis (6). We next asked whether the levels of these modulators were also altered by activation of GPR109A receptors. Incubation of chromaffin cells for 24 hours in 100 µM nicotinic acid or 5 mM β-hydroxybutyrate led to a large increase in the levels of NPY-ir (Fig. 5A and 5B). In contrast, the same agonists led to a decrease in the level of TH-ir (Fig. 5C and 5D). Thus, the effects of GPR109A agonists in vitro broadly mimic the effects of fasting in vivo (ie, an increase in the adrenal levels of AgRP and NPY and no change, or a decrease in the level of TH).

Figure 5.

GPR109A agonists alter the expression of multiple transmitter pathways in chromaffin cells in vitro. A, Examples of neuropeptide Y immunoreactivity (NPY-ir) in chromaffin cells treated with 5 mM β-hydroxybutyrate for 24 hours in vitro. Right panel is a cumulative frequency distribution showing that β-hydroxybutyrate led to a significant increase in NPY-ir (n = 3 experiments, ≥ 52 cells per experiment; P = 2.06E-20; Kolmogorov-Smirnov [K-S] test). B, Cumulative frequency distributions showing that 100 µM nicotinic (Nic.) acid increased the levels of NPY-ir (n = 3 experiments, ≥ 52 cells per experiment, P = 2.11E-09, K-S test). C, TH-ir in chromaffin cells treated with 5 mM β-hydroxybutyrate for 24 hours in vitro. Right panel is a cumulative frequency distribution showing this led to a decrease in TH-ir (n = 3 experiments, ≥ 50 cells per experiment, P = 3.96E-42, K-S test). D, Cumulative frequency distributions showing that 100 µM nicotinic acid treatment increased the levels of NPY-ir (n = 3 experiments, ≥ 50 cells per experiment, P = 5.31E-08, K-S test). **P less than .01. Scale bar 10 µm.

A GPR109A Agonist Increases the Strength of the Preganglionic → Chromaffin Cell Synapse

In addition to postsynaptic plasticity, fasting also leads to a long-lasting increase in the strength of the preganglionic → chromaffin cell synapse (6). This effect is evident as a decrease in the PPR of evoked synaptic currents recorded in chromaffin cells (Fig. 6A, fed vs fasted). We have previously shown that strengthening involves the activation of NPY5 receptors within the adrenal gland, likely located on the presynaptic terminals of the cholinergic input (6). We next determined whether activation of GPR109A receptors could also lead to synaptic strengthening. Adrenal slices were incubated in 100 µM nicotinic acid and PPR at the preganglionic → chromaffin cell synapse was quantified. Both in male and female mice, this treatment led to a significant decrease in the PPR, indicating an increase in synaptic strength, similar to that observed after fasting in WT mice (Fig. 6A).

Figure 6.

GPR109A agonist modulates preganglionic → chromaffin cell synaptic strength. A, Synaptic strength was monitored at the preganglionic → chromaffin cell synapse in adrenal slices. Shown are representative examples of synaptic currents evoked in response to paired presynaptic depolarizations. Food deprivation for 24 hours led to synaptic strengthening evident as a decrease in the paired pulse ratio (PPR; fed vs fasted). Incubation of adrenal slices in 100 µM nicotinic acid also led to a decrease in PPR in adrenal slices (fed vs nicotinic acid) from both male and female mice (n = 6-8 cells from 3-6 mice). B, Model pathway leading to the fasting-dependent regulation of adrenal medulla function. Fasting-induced sympathetic activity evokes epinephrine release from chromaffin cells in the adrenal medulla. Epinephrine contributes to increased lipolysis and fatty acid release from white adipose tissue during periods of negative energy balance. Hepatic ketogenesis metabolizes fatty acids to ketone bodies, including β-hydroxybutyrate, the endogenous agonist for GPR109A. Activation of this GPCR, likely in the adrenal, elevates agouti-related peptide (AgRP) and neuropeptide Y (NPY) and inhibits tyrosine hydroxylase (TH) expression in chromaffin cells. By acting at GPR109A receptors, β-hydroxybutyrate also leads to an increase in preganglionic → chromaffin cell synaptic strength. Not shown is the GPR109A-mediated negative feedback pathway that also regulates lipolysis at the level of white adipose tissue (51). Adrenal GPR109A receptors thus regulate communication between 3 tissues (adrenal-adipose-liver) that are involved in the peripheral response to food deprivation. *P less than .05; **P less than .01; ***P less than .001 (A, left and right, respectively: analysis of variance, Tukey post hoc; t test).

Discussion

Our results indicate that fasting increases adrenal AgRP expression via a novel pathway that involves activation of GPR109A receptors by circulating β-hydroxybutyrate, a ketone body whose levels rise during food deprivation. This effect may underlie previous reports showing β-hydroxybutyrate can alter AgRP mRNA levels in hypothalamic cell lines (52, 53). Regulation of AgRP expression has been extensively studied in the hypothalamus, where the fasting-induced increase involves elevated plasma levels of corticosterone and fatty acids (54, 55) and decreased levels of circulating leptin. Arcuate NPY/AgRP neurons express the leptin receptor, activation of which prevents the fasting-induced increase in AgRP expression (47-49). In contrast, this hormone does not appear to be the primary regulator of adrenal AgRP expression because fasting did not alter AgRP expression in GPR109A KO mice in spite of a decline in the plasma levels of leptin. Thus, even if leptin receptors are involved they must be upstream of the actions of β-hydroxybutyrate and GPR109A.

The adrenal GPR109A signaling pathway appears to be selectively engaged during periods of metabolic stress. The half maximal effective concentration for GPR109A activation is approximately 0.8 mM (30) and the levels of β-hydroxybutyrate increase from approximately 0.1 mM in fed animals to approximately 2 mM during food deprivation for 24 hours (Fig. 4) (56, 57). Thus, it is likely that the receptor primarily regulates AgRP expression in response to a metabolic challenge rather than by adjusting basal levels of expression. The intracellular signaling pathways that modulate AgRP expression have been extensively studied and multiple pathways have been identified. In the arcuate nucleus of the hypothalamus the leptin-regulated pathway is thought to involve STAT3 and FoxO1 signaling. STAT3 can bind to the AgRP promoter and inhibits its expression, and this is prevented by nuclear FoxO1 (58). Knockout of FoxO1 in AgRP-expressing cells impairs the increase in hypothalamic AgRP mRNA and peptide levels that is observed after fasting (59). Although the GPR109A receptor is not known to signal via this mechanism, FoxO1 is expressed in the adrenal (unpublished observations). The increase in arcuate AgRP and NPY mRNA that occurs during fasting requires the induction of peroxisome proliferator–activated receptor γ, likely at the level of the hypothalamus (60). Finally, an increase in the circulating levels of fatty acids during food deprivation stimulates autophagy in hypothalamic neurons and this has also been shown to elevate AgRP synthesis (54). Whether, or how, these diverse pathways interact with GPR109A signaling during fasting is not known.

Activation of adrenal GPR109A receptors led to a decrease in PPR at the preganglionic → chromaffin cell synapse. Because a decrease in PPR is generally interpreted as a form of presynaptic plasticity, this suggests that the activated receptors are on the terminals of the preganglionic neurons. We have recently shown that fasting also induces a decrease in PPR at this synapse, and a major component of this plasticity is due to the activation of adrenal Y5 receptors (6). Whether the Y5- and GPR109A-mediated effects intersect is not known but this seems feasible given both are G_i-coupled GPCRs. One possibility is that GPR109A receptors evoke NPY release and thus generate fasting-induced synaptic strengthening. When NPY signaling is prevented either genetically or pharmacologically, fasting does not alter preganglionic → chromaffin cell synaptic strength, epinephrine release is suppressed, and the mice are hypoglycemic (6). We attempted to determine whether the effect of fasting on synaptic strength was also absent in GPR109A KO mice. However, in fed KO mice the synapse was already strengthened. Fasting, or incubation of adrenal slices in nicotinic acid, had no additional effect on synaptic strength (Supplementary Fig. 1A) (61). As expected, given this phenotype, epinephrine release during fasting was normal and the mice were euglycemic, consistent with a functioning counterregulatory response (Supplementary Fig. 1B and 1C) (61). This suggests that complete loss of GPR109A is compensated for and determining its role in the adrenal will require a way to acutely silence receptor expression.

Ketone bodies, including β-hydroxybutyrate, can modulate neuronal activity including synaptic transmission (62, 63). Previous studies have examined this type of plasticity in the context of the ketogenic diet and as an anticonvulsant. For example, in the substantia nigra, β-hydroxybutyrate can suppress neuronal excitability via modulation of K(ATP) channels (64). Ketone bodies are high-energy substrates and can regulate neuronal functioning through a cellular metabolic pathway that is likely distinct from the GPCR-mediated effects we found in the adrenal (65).

One issue that we have been unable to resolve is the location of the GPR109A receptors within the adrenal. Receptor agonists increase AgRP and NPY expression in chromaffin cells in vitro, suggesting that these cells are the site of expression. Activation of GPR109A receptors leads to intracellular calcium release and suppression of (3′,5′-cyclic adenosine monophosphate)_i in a variety of cell types (30, 32, 66). However, GPR109A agonists do not appear to activate either of these pathways in chromaffin cells (data not shown). Whether adrenal receptors are coupled to other signaling mechanisms or the actions of GPR109A agonists in vitro are mediated via an intervening cell type (such as immune cells) is not yet known (66-68).

We find that activation of adrenal GPR109A receptors has multiple effects. By increasing NPY expression and the strength of the preganglionic → chromaffin cell synapse, GPR109A receptors are predicted to enhance the response to food deprivation. Conversely the GPR109A-regulated increase in AgRP expression would be expected to impair the response to fasting since we have previously shown AgRP weakens synaptic strength by antagonizing melanocortin receptors (14). Thus, it is not possible to categorize GPR109A signaling within the adrenal as purely facilitating or inhibiting adrenal function. Rather, the net result is likely to be the sum of these effects. One possibility is that the actions of AgRP and NPY are temporally separate, as occurs when the hypothalamic AgRP/NPY arcuate neurons, which control food intake, are activated (69, 70). However, we acknowledge that although the hypothalamic and adrenal chromaffin cells do have features in common (AgRP and NPY expression, activation during an energy deficit), these are cells with distinct roles.

In addition to adrenal cells, GPR109A is also expressed by white adipocytes, where it exerts an antilipolytic role (30, 32). During fasting, lipolysis is stimulated by norepinephrine released from sympathetic postganglionic neurons and epinephrine from the adrenal medulla (11, 12, 71, 72). Increased circulating free fatty acids stimulate hepatic ketogenesis and thus the production of β-hydroxybutyrate. Our experiments show the latter activates adrenal GPR109A receptors leading to multiple effects within the adrenal medulla. By virtue of their expression in white adipocytes and adrenal cells, GPR109A receptors therefore appear positioned to mediate crosstalk between all 3 tissues (adrenal-adipose-liver) that are involved in the peripheral response to food deprivation (Fig. 6B). Using β-hydroxybutyrate as a source of metabolic energy and as an endocrine signaling molecule seems a beautifully efficient way to coordinate the response to food deprivation.

Glossary

Abbreviations

AgRP

agouti-related peptide

BSA

bovine serum albumin

EPSC

excitatory postsynaptic current

GPCR

G protein–coupled receptor

-ir

immunoreactive

KO

knockout

mRNA

messenger RNA

NPY

neuropeptide Y

PBS

phosphate-buffered saline

PPR

paired-pulse ratio

ROI

region of interest

RT-PCR

reverse transcription polymerase chain reaction

TH

tyrosine hydroxylase

WT

wild-type

Financial Support

This work was supported by the National Institutes of Health (grant Nos. DK080441 and DK098134 to M.D.W.).

Disclosures

The authors have nothing to disclose.

Data Availability

Some or all data sets generated during and/or analyzed during the present study are not publicly available but are available from the corresponding author on reasonable request.