( D )- (cid:1) -Hydroxybutyrate Inhibits Adipocyte Lipolysis via the Nicotinic Acid Receptor PUMA-G*

As a treatment for dyslipidemia, oral doses of 1-3 grams of nicotinic acid per day lower serum triglycerides, raise high density lipoprotein cholesterol, and reduce mortality from coronary heart disease (Tavintharan, S., and Kashyap, M. L. (2001) Curr. Atheroscler. Rep. 3, 74-82). These benefits likely result from the ability of nicotinic acid to inhibit lipolysis in adipocytes and thereby reduce serum non-esterified fatty acid levels (Carlson, L. A. (1963) Acta Med. Scand. 173, 719-722). In mice, nicotinic acid inhibits lipolysis via PUMA-G, a Gi/o-coupled seven-transmembrane receptor expressed in adipocytes and activated macrophages (Tunaru, S., Kero, J., Schaub, A., Wufka, C., Blaukat, A., Pfeffer, K., and Offermanns, S. (2003) Nat. Med. 9, 352-355). The human ortholog HM74a is also a nicotinic acid receptor and likely has a similar role in anti-lipolysis. Endogenous levels of nicotinic acid are too low to significantly impact receptor activity, hence the natural ligands(s) of HM74a/PUMA-G remain to be elucidated. Here we show that the fatty acid-derived ketone body (D)-beta-hydroxybutyrate ((D)-beta-OHB) specifically activates PUMA-G/HM74a at concentrations observed in serum during fasting. Like nicotinic acid, (D)-beta-OHB inhibits mouse adipocyte lipolysis in a PUMA-G-dependent manner and is thus the first endogenous ligand described for this orphan receptor. These findings suggests a homeostatic mechanism for surviving starvation in which (D)-beta-OHB negatively regulates its own production, thereby preventing ketoacidosis and promoting efficient use of fat stores.

For the generation of stable cell lines, 5 ϫ 10 6 CHO-K1 cells were transfected with 12 g of plasmid DNA (pCDNA3.1, Invitrogen) containing either HM74a, HM74, or PUMA-G expressed from the cytomegalovirus promoter. Two days after transfection, the growth medium was supplemented with 400 g/ml G418 to select for antibiotic-resistant cells. Clonal CHO-K1 cell lines that stably express HM74, HM74a, or PUMA-G were selected based on the ability of nicotinic acid (HM74a and PUMA-G) or S711589 (an Arena HM74-specific agonist; data not shown) to inhibit forskolin-induced cAMP production.
Calcium Mobilization-CHO-K1 cells expressing an NFAT-␤-lactamase reporter and the promiscuous G␣ subunit G qi5 (kind gift of K. Sullivan, Merck Research Laboratories) were stably transfected with either empty vector (pCDNA3.1, Invitrogen) or vector expressing PU-MA-G, HM74a, or HM74. Cells were seeded at 10,000 cells/well in 384-well culture plates and grown overnight at 37°C, 5% CO 2 in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 2 mM L-glutamine, 10 mM HEPES, pH 7.4, 0.1 mM MEM non-essential amino acids solution, 1 mM sodium pyruvate, 0.6 mg/ml hygromycin B, 0.5 mg/ml zeocin, and 1 mg/ml geneticin (BD Biosciences). Cells were washed four times with Hanks' balanced salt solution containing 10 mM HEPES, pH 7.4, and loaded with calcium-sensitive dye by incubating with an equal volume of Molecular Devices calcium assay kit loading buffer at 37°C for 1 h. Calcium response in the fluorometric imaging plate reader assay was measured according to the directions from Molecular Devices.
[ 35 S]GTP␥S Binding Assay-Membranes from untransfected CHO-K1 cells or cells stably expressing PUMA-G, HM74a, or HM74 (20 g/assay) were diluted in assay buffer (20 mM HEPES, pH 7.4, 100 mM NaCl, 10 mM MgCl 2 ) in Wallac Scintistrip plates and preincubated with test compounds diluted in assay buffer containing 40 M GDP (final * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
[GDP] was 10 M) for ϳ10 min before addition of [ 35 S]GTP␥S to 0.3 nM. To measure the agonist activity of free acids, rather than sodium salts (Table I), the HEPES concentration was increased to 60 mM; this increase had no effect on the EC 50 of nicotinic acid (data not shown). Binding was allowed to proceed for 1 h before centrifuging the plates at 4000 rpm for 15 min at room temperature and subsequent counting in a Packard TopCount scintillation counter. Non-linear regression analysis of the binding curves was performed in GraphPad Prism version 4.
[ 3 H]Nicotinic Acid Binding Competition Assay-Assays were performed with the same preparations of membrane used for the [ 35 S]GTP␥S assay. Equilibrium binding of [ 3 H]nicotinic acid was done with membranes (30 g/assay) and test compounds diluted in assay buffer (20 mM HEPES, pH 7.4, 1 mM MgCl 2 , and 0.01% CHAPS) in a total volume of 200 l. After 4 h at room temperature, reactions were filtered through Packard Unifilter GF/C plates using a Packard Harvester and washed eight times with 200 l of ice-cold binding buffer. Nonspecific binding was determined in the presence of 250 M unlabeled nicotinic acid. Competitive binding assays were performed in the presence of 50 nM [ 3 H]nicotinic acid.
In Vitro Lipolysis-Isolation of mouse primary epididymal adipocytes and in vitro determination of NEFA release was performed according to the method of Rodbell as adapted by Tunaru et al. (3,23).

RESULTS
To ask whether ␤-OHB is a ligand for HM74a, we used Chinese hamster ovary (CHO) cells that stably express the chimeric G-protein ␣ subunit G qi5 (18) and harbor either a control vector or vectors that expresses either HM74a or its paralog HM74, which is 95% identical at the amino acid level to HM74a but has ϳ1000-fold less affinity for nicotinic acid (4,5). Use of G qi5 allows the normally G i/o -coupled HM74 and HM74a to signal via the G q pathway leading to Ca ϩ2 mobilization. Nicotinic acid elicited Ca ϩ2 mobilization only in cells expressing HM74a (Fig. 1A). In contrast, Acifran, an agonist on both HM74 and HM74a (4), elicited a response from both receptors demonstrating that HM74 was functional and could be used as a specificity control in this assay. Given that ketone bodies can reach millimolar concentrations in serum, we tested these compounds at 15 mM. Both (D)-and (L)-␤-OHB, but not acetoacetate or free acetone, elicited a Ca ϩ2 response in cells expressing HM74a but not HM74. Whereas the D-isomer of ␤-OHB is the sole form encountered in high concentrations physiologically (7), both D-and L-isomers have been shown to inhibit lipolysis in vitro (17), consistent with the results shown here. Similar results were observed for cells expressing murine PUMA-G (data not shown).
Next, we determined the half-maximal concentration (EC 50 ) of ␤-OHB required to stimulate receptor-mediated guanine nucleotide exchange on G␣ using a [ 35 S]GTP␥S binding assay with membranes prepared from untransfected CHO cells or CHO cells stably expressing either mouse PUMA-G, human HM74a, or HM74 ( Fig. 1B and Table I). Nicotinic acid stimulated [ 35 S]GTP␥S binding only in membranes from cells expressing HM74a (EC 50 104 Ϯ 5 nM) or the mouse ortholog PUMA-G (EC 50 ϭ 43 Ϯ 3 nM), whereas Acifran was active on all three receptors (EC 50 ϭ 1127 Ϯ 59 nM for HM74a, 358 Ϯ 30 nM for PUMA-G, and 7039 Ϯ 598 nM for HM74). Racemic (DL)-␤-OHB also showed some degree of activity on all receptors; however, it was more potent on HM74a (EC 50 ϭ 0.8 Ϯ 0.06 mM) and its ortholog PUMA-G (EC 50 ϭ 0.7 Ϯ 0.04 mM) than on HM74. The EC 50 for the L-enantiomer was ϳ2-fold higher than that of the physiologically relevant D-enantiomer (Table I). Both the sodium and lithium salts of ␤-OHB were active, indicating that the anion is the active component. Moreover, sodium salts of other small monocarboxylic acids with similar pK a values to ␤-OHB (␣-hydroxybutyrate and lactate) were not significantly active in this assay (Table I). At high concentrations, lithium acetoacetate, but not acetone or lithium chloride, elicited [ 35 S]GTP␥S binding to all three receptors but not to membranes from untransfected cells, suggesting that acetoacetate is a weak agonist of these receptors.
Short-chain fatty acids were recently identified as ligands for the G-protein-coupled receptor GPR41, which is, like HM74a/ PUMA-G, expressed in adipocytes (19 -21). Given their similarity to ␤-OHB, we determined whether small fatty acids are ligands for HM74a/PUMA-G. Indeed, a structure-activity relationship was observed for fatty acids of varying chain length  (Table I) Table I). However, it is unlikely that these small fatty acids reach sufficient concentrations in serum to activate these receptors (19), and so their low affinity for PUMA-G/HM74a and HM74 is probably not physiologically relevant. Consistent with this, while both HM74a/PUMA-G and GPR41 have similar chain length preferences, the latter receptor has at least 10-fold higher affinity for these acids (Table I and Ref. 20). We note as well that in another study racemic (DL)-␤-OHB was not a GPR41 agonist at concentrations up to 10 mM (20). Similarly, another study found "L-OH-butyrate" (the exact molecular structure of which was not described) to be only a very weak (EC 50 ϳ5 mM) ligand of human GPR41 and GPR43, another receptor for short-chain fatty acids expressed predominantly in leukocytes (19). Finally, ␤-OHB was not an agonist of GPR81 (22), the next most phylogenetically related receptor to HM74a/PUMA-G (data not shown).
We employed an equilibrium [ 3 H]nicotinic acid binding competition assay to ask whether nicotinic acid and ␤-OHB compete for the same binding site on the receptor. [ 3 H]Nicotinic acid bound specifically and saturably to membranes from cells expressing HM74a, but not HM74 (data not shown), with a calculated K d of 105 Ϯ 9 nM (data not shown), a value in good agreement with published results (3,4). Homologous competition with unlabeled nicotinic acid yielded a K i of 130 Ϯ 14 nM, similar to the observed K d (Fig. 1C). As in the [ 35 S]GTP␥S binding assay, AcAc displayed low apparent affinity for HM74a, but (DL)-␤-OHB bound the receptor with physiologically relevant affinity (K i ϭ 0.7 Ϯ 0.06 mM; Fig. 1C). Similar results were observed for PUMA-G (K i ϭ 0.7 mM Ϯ 0.1 mM; data not shown). Taken together, these data show that (D)-␤-OHB is a HM74a/PUMA-G agonist and that the serum concentrations of this ketone body observed as early as 2-3 days into a fast in humans (7,8) will result in significant receptor occupancy and activity.
Previous studies with isolated adipocytes from knock-out mice demonstrated that PUMA-G mediates the anti-lipolytic effect of nicotinic acid (3); we performed analogous experiments with (D)-␤-OHB. Both nicotinic acid and sodium (D)-␤-OHB suppressed free fatty acid efflux from isoproterenol-stimulated primary adipocytes from wild-type mice, the latter at concentrations consistent with the affinity determined previously (approximate EC 50 for (D)-␤-OHB-mediated lipolysis inhibition ϳ2 mM, Fig. 2A). This suppression was PUMA-G-mediated, as nicotinic acid and (D)-␤-OHB were without significant effect in adipocytes from PUMA-G knock-out mice (Fig. 2B). Adipocytes from PUMA-G knock-out mice were not merely refractory to lipolysis inhibition per se, as ADP inhibited fatty acid efflux equally well in wild-type and knock-out cells (Fig. 2B). The adenosine analogue phenylisopropyladenosine also inhibited lipolysis equally well in wild-type and knock-out cells (data not shown). We note that the rate of both basal and isoproterenolstimulated lipolysis was routinely lower in adipocytes from knock-out mice compared with those from wild-type animals ( Fig. 2B; compare "No stimulation" and "Iso"). It is unlikely that this difference is due to a deficiency of lipid in the knockout adipocytes, as wild-type and knock-out cells had approximately the same overall size as determined by microscopy (data not shown), and wild-type and knock-out mice had the same overall fat mass as determined by NMR spectroscopy. 2

DISCUSSION
In this work we have shown that the ketone body (D)-␤-OHB specifically binds to and activates the adipocyte-expressed GPCRs HM74a/PUMA-G with an affinity that is well within the range of serum concentrations observed for this metabolite after ϳ2-3 days of starvation in humans and ϳ1-2 days in mice (7,10). The effect of (D)-␤-OHB is, like nicotinic acid, Inactive Inactive Inactive a Some compounds displayed activity at the highest concentration tested but not over a full dose response from which an EC 50 could be determined. Therefore, the EC 50 for these compounds is denoted as greater than the highest concentration tested.
b Fatty acids with Ͼ10 carbons were tested at 10 M only.

(D)-␤-Hydroxybutyrate Inhibits Lipolysis via PUMA-G
anti-lipolytic (1, 2, 6). (D)-␤-OHB is thus the first endogenous ligand described for the orphan receptor PUMA-G. During starvation, the rate of hepatic ketone body synthesis is at least partially limited by the rate of adipocyte lipolysis (7). That (D)-␤-OHB is itself anti-lipolytic suggests a homeostatic negative feedback mechanism in which this metabolite regu-lates its own production by decreasing the serum level of fatty acid precursors available for hepatic ketogenesis. Indeed, in a study of the serum NEFA-lowering effect of ␤-OHB infused into humans, Senior and Loridan (9) proposed that during starvation ketone bodies exert "a fine regulatory adjustment" of their own synthesis by inhibiting adipocyte lipolysis. Such a mechanism would potentially conserve fat stores during extended starvation and attenuate excessive formation of ketoacids from unrestrained lipolysis and ketogenesis. This model implies that PUMA-G knock-out mice would experience higher rates of lipolysis and ketogenesis during a fast, as (D)-␤-OHB does not inhibit lipolysis in the absence of this receptor. In preliminary experiments, we have not observed significant reproducible differences between wild-type and PUMA-G knock-out animals in rate of body fat depletion or serum levels of NEFA or (D)-␤-OHB during 24 and 48 h of fasting. 3 However, we note that adipocytes from knock-out animals are refractory to stimulation with isoproterenol (Fig. 2B), perhaps suggesting a fundamental difference in lipolysis regulation in the knock-out cells that may compensate for lack of PUMA-G. Studies with inducible knock-outs of PUMA-G in adult animals may shed light on this question.