Differential S-palmitoylation of the human and rodent β3-adrenergic receptors

With few reported exceptions, G protein–coupled receptors (GPCRs) are modified by Cys palmitoylation (S-palmitoylation). In multiple GPCRs, S-palmitoylation targets a canonical site within the C-terminal cytoplasmic tail adjacent to the C terminus of the seventh transmembrane domain, but modification of additional sites is exemplified by the β-adrenergic receptors (βARs). The β1AR is S-palmitoylated at a second, more distal site within the C-terminal tail, and the β2AR is modified at a second site within the third intracellular loop, neither of which is conserved in other βAR isoforms. The functional roles of S-palmitoylation of disparate sites are incompletely characterized for any GPCR family. Here, we describe S-palmitoylation of the β3AR. We compared mouse and human β3ARs and found that both were S-palmitoylated at the canonical site within the C-terminal tail, Cys-358 and Cys-361/363 in mouse and human β3ARs, respectively. Surprisingly, the human β3AR was S-palmitoylated at two additional sites, Cys-153 and Cys-292 within the second and third intracellular loops, respectively. Cys-153 is apparently unique to the human β3AR, and Cys-292 is conserved primarily in primates. Mutational substitution of C-tail Cys in human but not mouse β3ARs resulted in diminished ligand-induced cAMP production. Substitution of Cys-153, Cys-292, or Cys-361/363 within the human β3AR diminished membrane-receptor abundance, but only Cys-361/363 substitution diminished membrane-receptor half-life. Thus, S-palmitoylation of different sites differentially regulates the human β3AR, and differential S-palmitoylation distinguishes human and rodent β3ARs, potentially contributing to species-specific differences in the clinical efficacy of β3AR-directed pharmacological approaches to disease.

It is estimated that Ͼ10% of mammalian proteins are subject to modification by long-chain fatty acid acylation, the predominant form of which is S-palmitoylation, the dynamic (reversible) ligation of a 16-carbon saturated fatty acid to a Cys residue via thioesterification (1). S-Palmitoylation operates throughout metazoan phylogeny. In mammalian cells, S-palmitoylation is mediated by the 23-24 members of the DHHC family of palmitoyltransferases (1), and the dynamic nature of S-palmitoylation reflects the operation of depalmitoylating enzymes, including the acyl-protein thioesterases APT1 2 and APT2 as well as other, less well-characterized members of the ABHD17 family of serine hydrolase/lysophospholipases (2).
S-Palmitoylation was originally characterized as a mechanism for membrane localization of otherwise cytosolic proteins but was subsequently shown to modify multiple integral membrane proteins where insertion of the palmitate moiety into the membrane would alter protein configuration and thereby function (3). G protein-coupled receptors (GPCRs), a eukaryotic superfamily that comprises about 800 members and that represents the largest family of membrane receptors, are seventransmembrane-spanning integral membrane proteins. The first example of S-palmitoylation of a ligand-activated GPCR was provided by the prototypical GPCR, the ␤ 2 -adrenergic receptor (␤ 2 AR). S-Palmitoylation was shown to target Cys-341 of the human ␤ 2 AR, located within the C-terminal tail in a juxtamembrane position adjacent to the cytosolic terminus of the seventh transmembrane segment (4) (unless indicated, all residue numbering here and below is in accordance with human sequences). Where reported, almost all GPCRs examined subsequently were shown to be subject to S-palmitoylation, including S-palmitoylation of one to three Cys located within the C-terminal tail adjacent to the cytosolic terminus of the seventh transmembrane segment (5). However, not all GPCRs possess Cys within the C-terminal tail; S-palmitoylation of sites in addition to or other than this canonical site was shown to occur in a number of GPCRs more distally within the C-terminal tail (6 -14), and S-palmitoylation of a site within either the first, second, or third intracellular loop has been demonstrated directly in a number of cases (15)(16)(17). Understanding of the different functional roles of S-palmitoylation of disparate sites within a given GPCR or across members of a GPCR family remains incomplete.
These generalities are well instantiated in the case of the ␤ARs. The ␤ 1 AR is S-palmitoylated at an additional site within the C-terminal tail distal to the canonical site, comprising two Cys that are not conserved in the ␤ 2 AR or ␤ 3 AR (9), whereas the ␤ 2 AR is S-palmitoylated at an additional site within the third intracellular loop that is not conserved within the ␤ 1 AR or ␤ 3 AR (15).
We describe here S-palmitoylation of the ␤ 3 AR. We examined mouse and human ␤ 3 ARs and found that both are modified at the canonical C-tail site (Cys-361/363 and Cys-358 in human and mouse, respectively). Surprisingly, however, the human ␤ 3 AR was modified at two additional sites, Cys-153 and Cys-292 within the second and third intracellular loops, respectively, which are not conserved in the ␤ 1 AR or ␤ 2 AR. Furthermore, Cys-153 is apparently unique in phylogeny to the human receptor, and Cys-292 is conserved primarily among primates. S-Palmitoylation of the canonical site within the ␤ 3 AR facilitated receptor-effector coupling in the human but not in the mouse ␤ 3 AR. S-Palmitoylation of the additional sites within the human ␤ 3 AR increased receptor membrane abundance, but only S-palmitoylation of Cys-361/363 enhanced the half-life of the receptor at the membrane. Thus, S-palmitoylation of different sites within the human ␤ 3 AR differentially regulates receptor disposition. The unique pattern of S-palmitoylation of the human ␤ 3 AR provides a molecular distinction between human and rodent ␤ 3 ARs, which may be associated with the differential response of humans and rodents to ␤ 3 AR agonists as therapeutic agents.

The mouse ␤ 3 AR (m␤ 3 AR) is S-palmitoylated exclusively at the canonical site
To examine S-palmitoylation of the ␤ 3 AR, we used the acyl-RAC method (resin-assisted capture of fatty-acylated proteins) in which the thioester bond linking palmitate to Cys is cleaved with neutral hydroxylamine and the resultant free thiol is coupled to thiopropyl-Sepharose for pulldown and subsequent Western blot analysis (18). This method directly assesses the presence of palmitate-modified Cys, and omission of hydroxylamine provides a rigorous control for false positives. Human embryonic kidney (HEK) 293 cells were used in all experiments. We have shown previously that 19 of the ϳ23 mammalian DHHC palmitoyltransferases as well as the protein depalmitoylases APT1/2 are expressed in HEK293 cells (15). We first used acyl-RAC with HEK293 cells stably expressing wildtype (WT) or Cys 3 Ala mutant ␤ 3 ARs to assess the occurrence of S-palmitoylation and to locate the sites of modification. Analysis was restricted to clones in which levels of receptor expression varied by Ͻ10%.
Analysis of HEK293 cells expressing FLAG-tagged WT m␤ 3 AR revealed that the m␤ 3 AR is basally S-palmitoylated (Fig. 1A). The m␤ 3 AR contains 10 Cys, two of which are predicted to be cytoplasmic: Cys-272 and Cys-358. One of these, Cys-358, is predicted to have a juxtamembrane location adjacent to the seventh transmembrane domain and thus represents the canonical site of GPCR S-palmitoylation ( Fig. 1B and see also Fig. 2, D and E). Mutation of Cys-358 to Ala eliminated S-palmitoylation of the m␤ 3 AR (Fig. 1, C and D). Thus, the m␤ 3 AR is S-palmitoylated solely at the canonical C-tail site. Cys-358 within the m␤ 3 AR was identified previously as a site of S-palmitoylation in a global survey of the mouse brain palmitoyl proteome (19).
Alternative splicing generates a "b" isoform of the m␤ 3 AR (expressed in parallel but at substantially lower levels in multiple tissues) that contains an alternative stretch of residues near the C terminus, one of which is a Cys (Cys-400) (20). Cys-400 is conserved in the mouse ␤ 1 AR and ␤ 2 AR and in the human ␤ 1 AR but apparently is not S-palmitoylated (9).

Multiple S-palmitoylation of the human ␤ 3 AR (h␤ 3 AR)
Analysis by acyl-RAC of FLAG-tagged h␤ 3 AR stably expressed in HEK293 cells demonstrated S-palmitoylation ( Fig.  2A). However, unlike the m␤ 3 AR, mutation of either or both of the Cys residues comprising the canonical site, Cys-361/363, did not significantly diminish S-palmitoylation as detected with acyl-RAC (Fig. 2B), indicating the existence of additional sites of S-palmitoylation.
The h␤ 3 AR contains 16 Cys, eight of which are predicted to be cytoplasmic. Of those, four are predicted to be juxtamembrane, including, in addition to the canonical Cys-361/363, Cys-153 located adjacent to the N terminus of the fourth transmembrane domain and Cys-292 located adjacent to the N terminus of the sixth transmembrane domain (see Fig. 2, D and E). Mutation of all four of these Cys (designated C4A) effectively eliminated S-palmitoylation (Fig. 2C). Because the acyl-RAC assay cannot distinguish between one or more sites of S-palmitoylation within a given substrate, we stably expressed h␤ 3 AR in which three of the four Cys that were potential sites of S-palmitoylation were mutated to Ala. The h␤ 3 AR containing either Cys-153, Cys-292, Cys-361, or Cys-363 was S-palmitoylated, demonstrating that all four Cys were sites of S-palmitoylation (Fig. 2C). Note that the populations of S-palmitoylated receptors as revealed by acyl-RAC were diminished relative to WT receptor when only a single site of S-palmitoylation was present (Fig. 2C), suggesting that individual receptors may be differentially S-palmitoylated at one or more sites under these conditions. Notably, on the basis of comparison of available sequences of the ␤ 3 AR, Cys-153 is apparently unique to the ␤ 3 AR in humans, and Cys-292 is restricted largely to primates ; n ϭ 4 -11. *, p Ͻ 0.0001 with respect to WT by ANOVA. D, Cys-153 is unique to human, and Cys-292 is conserved largely in primates, whereas Cys comprising the canonical site of S-palmitoylation, Cys-361/363, are conserved across vertebrate phylogeny. E, a schematic summary comparing localization of S-palmitoylation within the m␤ 3 AR and h␤ 3 AR. Within the h␤ 3 AR, Cys-153 and Cys-292 are located within the second and third intracellular loops, respectively, and are predicted to have a juxtamembrane location. Cys shown to be S-palmitoylated in the present study are indicated in blue and red; cytoplasmic Cys not subject to S-palmitoylation are indicated in green and white.

S-Palmitoylation of the h␤ 3 AR but not the m␤ 3 AR facilitates receptor-effector coupling
It was reported that mutation of the canonical C-tail target of S-palmitoylation within the human ␤ 2 AR (Cys-341) diminished ligand-induced production of cAMP, interpreted as disrupted coupling of the receptor via G s to G protein-activated adenyl cyclase (4). To examine the role of S-palmitoylation on receptor-effector coupling of the ␤ 3 AR, we transiently expressed, in HEK293 cells, WT or Cys-mutant h␤ 3 AR or m␤ 3 AR and, 24 h after transfection, assessed by ELISA cAMP production elicited by exposure for 5 min to the ␤ 3 AR-specific synthetic ligand mirabegron (21).
To avoid interpretive difficulty resulting from different levels of expression of different constructs, we calculated the EC 50 following exposure to mirabegron over the effective concentra-tion range of 10 Ϫ2 -10 4 nM. Mirabegron did not elicit cAMP production in the absence of transfected receptor (Fig. 3A). The EC 50 of cAMP production by WT m␤ 3 AR (3.66 nM) was unaffected by mutation of the sole site of S-palmitoylation, Cys-358 (Fig. 3A). In contrast, mutation of all four sites of S-palmitoylation within the h␤ 3 AR substantially increased the EC 50 from 2.17 nM in WT to 4.82 nM in the absence of S-palmitoylation (Fig. 3, B and C). Transfection with the h␤ 3 AR lacking only Cys-153, Cys-292, or Cys-361/363 revealed that the canonical C-tail sites of S-palmitoylation (Cys-361/363) accounted for the effects of S-palmitoylation on receptor-coupled cAMP production (Fig. 3, B and C). The ϳ2.2-fold increase in EC 50 is larger than the ϳ1.5-fold increase reported following mutation of the canonical S-palmitoylated Cys within the human ␤ 2 AR (4). We confirmed that the effect of S-palmitoylation on EC 50 observed with transient transfection was comparable in HEK293 cells stably expressing C4A versus WT h␤ 3 AR (ϳ2.2-fold increase) (Fig. 3D). Thus, S-palmitoylation of different sites within the h␤ 3 AR mediates different functions, and S-palmitoylation of Within each individual experimental run, the significance of the variance between the EC 50 value of WT and palmitoylation-deficient mutants was tested. EC 50 values are presented as means with 95% confidence intervals (CI). Mirabegron-induced production of cAMP was not significantly diminished by C358A mutation of the m␤ 3 AR (two-tailed unpaired Student's t test), whereas cAMP production was significantly diminished (Ͼ2.2-fold) by C4A mutation and by C153A mutation of the h␤ 3 AR (one-way ANOVA, post hoc Dunnett). D, diminished cAMP production by C4A mutantion of the h␤ 3 AR was confirmed in HEK293 cells stably expressing either WT or C4A mutant h␤ 3 AR (two-tailed unpaired Student's t test; n ϭ 4 -5). In A, B, and D, error bars represent S.D.

S-Palmitoylation of the ␤ 3 -adrenergic receptor
the canonical site within the m␤ 3 AR and h␤ 3 AR differs in functionality. We note that the WT m␤ 3 AR and h␤ 3 AR differ substantially in the EC 50 of cAMP production by mirabegron (3.66 versus 2.17 nM) (Fig. 3C), but we have not explored this difference with alternative agonists.

S-Palmitoylation of the ␤ 3 AR influences receptor disposition
S-Palmitoylation of GPCRs (which may be co-and/or posttranslational) has been implicated in the governance of receptor plasma-membrane abundance that may reflect receptor processing/membrane targeting as well as plasma-membrane stability (turnover) (5). We first assessed steady-state plasmamembrane abundance of WT versus Cys-mutant m␤ 3 AR or h␤ 3 AR in HEK293 cells transiently expressing FLAG-tagged receptor. The FLAG tag in all constructs was localized to the N terminus of the ␤ 3 AR and was therefore extracellular in plasma-membrane-localized receptors. Twenty-four hours following transfection, cells were labeled with anti-FLAG Ab without membrane permeabilization followed by a fluores-cently labeled secondary Ab and analysis by fluorescenceactivated cell sorting (FACS). Fluorescence immunocytochemistry confirmed that very little intracellular labeling occurred with this protocol (see Fig. 4D).
C358A mutation of the m␤ 3 AR resulted in a small (ϳ8%) but consistent decrease in steady-state levels of plasmamembrane-localized receptor (Fig. 4A). C4A mutation of the h␤ 3 AR or mutation of either Cys-153, Cys-292, or Cys-361/363 resulted in a larger decrease that was of similar magnitude for each mutation (ϳ20 -32% decrease) (Fig. 4A). We asked whether diminished levels of plasma-membrane-localized h␤ 3 AR might reflect a decrease in overall receptor abundance. We transiently expressed FLAG-tagged WT or mutant receptor, and 24 h following transfection, cells were fixed, permeabilized to allow intracellular access of antibody, and labeled with anti-FLAG Ab conjugated with fluorescent dye followed by FACS. C4A mutation significantly diminished h␤ 3 AR abundance, and mutation of individual Cys revealed that this decrease was due to abrogated S-palmitoylation of Cys-292 Cys-361/363 resulted in a larger (ϳ20 -32%) decrease in membrane-receptor abundance; n ϭ 4. *, p Ͼ 0.006 by one-way ANOVA, post hoc Dunnett. B, to assess total receptor abundance, cells were fixed, permeabilized and exposed to anti-FLAG Ab conjugated with Alexa Fluor 488 followed by FACS. C358A mutation of the m␤ 3 AR had no effect on receptor abundance (mouse, at left). C4A or C292A mutation of the h␤ 3 AR significantly decreased total receptor abundance (human, at right); n ϭ 4; *, p Ͻ 0.004 by one-way ANOVA, post hoc Dunnett. C, none of the mutations examined in either the m␤ 3 AR or h␤ 3 AR resulted in a change in ␤ 3 AR mRNA as assessed by quantitative PCR; n ϭ 4. ns, not significant (p Ͼ 0.05). In A-C, error bars represent S.D. D, HEK293 cells transiently expressing WT or mutant h␤ 3 AR were labeled with anti-FLAG ␤ 3 AR Ab conjugated to fluorescent dye and visualized by confocal fluorescence microscopy. In the left column, cells were exposed to anti-FLAG Ab conjugated to Alexa Fluor 488 (green) prior to fixation (selective staining of plasma-membrane receptor). Note that exposure of cells to Ab without permeabilization did not result in staining other than that at the plasma membrane. In the middle column, cells were then fixed, permeabilized, and exposed to anti-FLAG Ab conjugated to Alexa Fluor 594 (red) to visualize total h␤ 3 AR. In the right column, merged images are shown with filter settings that also allow visualization of nuclear staining (Hoechst 33342; blue). Scale bars, 10 m. (Fig. 4B). The abundances of m␤ 3 AR and h␤ 3 AR mRNAs were unaltered by any mutation as assessed by quantitative PCR (Fig.  4C). Diminished levels of C292A h␤ 3 AR would be consistent with deficient processing resulting from abrogation of cotranslational S-palmitoylation of Cys-292 that triggers (endoplasmic reticulum-based) proteostatic degradation and/or with posttranslational destabilization.

S-Palmitoylation of the ␤ 3 -adrenergic receptor
We then further examined the influence of S-palmitoylation on receptor disposition with fluorescence immunocytochemistry, comparing transiently expressed WT FLAG-h␤ 3 AR with C4A receptor as well as with receptors with individual Cys mutation (Fig. 4D). In cells that were permeabilized before staining (as above), we observed consistently that the C4A receptor accumulated in the cytoplasm where it exhibited predominantly a punctate distribution (Fig. 4D). Examination of the effects of selective mutation of Cys-153, Cys-292, and Cys-361/363 revealed that only mutation of Cys-153 replicated this intracellular accumulation (Fig. 4D). Mutation of Cys-292 resulted in a diffuse intracellular distribution consistent with receptor degradation. Thus, the decrease in plasma-membranelocalized C153A receptor, demonstrated by FACS (Fig. 4A), is due at least in part to deficient receptor targeting to the plasma membrane, consistent with a requirement for Cys-153 S-palmitoylation in efficiently engaging the transport mechanism that conveys receptors from the Golgi to the plasma membrane (22). Our results therefore indicate that the diminished plasmamembrane abundance of the h␤ 3 AR resulting from mutation of Cys-153 and Cys-292 (Fig. 4A) reflects at least in part different mechanisms.
On the basis of mutational analysis, S-palmitoylation of Cys-361/363 within the h␤ 3 AR also diminishes plasma-membrane abundance (Fig. 4A) but does not result in enhanced receptor degradation (diminished total levels of receptor) (Fig. 4B) or diminished transport of receptor to the plasma membrane (signified by cytoplasmic accumulation of receptor) (Fig. 4D). Membrane-receptor abundance will be determined at least in part by membrane-receptor turnover (membrane stability). To examine directly a possible role for S-palmitoylation in membrane-receptor stability, we labeled cells transiently expressing WT or Cys-mutant m␤ 3 AR or h␤ 3 AR with anti-FLAG Ab and, at subsequent intervals over 25 h, labeled unpermeabilized cells with fluorescent secondary Ab followed by FACS.
We observed that the half-life of the plasma-membranelocalized m␤ 3 AR was about 4 times greater than that of the

. Diminished S-palmitoylation of the canonical site diminishes plasma-membrane receptor stability of the h␤ 3 AR but not the m␤ 3 AR. A, WT
or S-palmitoylation-deficient FLAG-m␤ 3 AR was transiently expressed in HEK293 cells, and plasma-membrane-localized receptors were labeled with anti-FLAG Ab without permeabilization followed by labeling with a florescent secondary Ab at the intervals indicated and FACS. Data are fitted to a one-phase decay curve; error bars represent S.D. B, WT or C4A FLAG-h␤ 3 AR was transiently or stably expressed, and plasma-membrane-localized receptors were assessed as in A (analysis by one-way ANOVA, post hoc Tukey; error bars represent S.D.; n ϭ 3-4). C, WT or S-palmitoylation-deficient FLAG-h␤ 3 AR was transiently expressed, and data are presented as in A and B. D, analysis of the results shown in A and C demonstrates that the stability of the m␤ 3 AR at the plasma membrane as revealed by calculated half-life was not affected by mutation of Cys-358 (analysis by two-tailed paired Student's t test). The half-life of the h␤ 3 AR was significantly diminished (ϳ29%) by mutation of Cys-361/363, whereas mutation of Cys-153 or Cys-292 had no effect (analysis by one-way ANOVA, post hoc Dunnett; error bars represent S.D.). CI, confidence interval.

S-Palmitoylation of the ␤ 3 -adrenergic receptor
h␤ 3 AR (29.3 versus 7.82 h) (Fig. 5). Mutation of Cys-358 did not affect the membrane half-life of the m␤ 3 AR (Fig. 5, A and  C). Thus, the small diminishment of plasma-membrane receptor abundance associated with mutation of Cys-358 within the m␤ 3 AR (Fig. 4A) is likely to reflect deficient receptor processing/targeting.
In contrast, transient expression of C4A mutant h␤ 3 AR substantially diminished membrane-receptor half-life (55% decrease), which we confirmed in cells stably expressing C4A mutant h␤ 3 AR (Fig. 5B). Analysis of the effects of individual mutations demonstrated that diminished membrane-receptor half-life could be ascribed to Cys-361/363 (Fig. 5, C and D). Thus, S-palmitoylation of different sites within the h␤ 3 AR regulates membrane-receptor abundance differentially: S-palmitoylation of Cys-153 and Cys-292 within the second and third intracellular loops, respectively, is required for efficient receptor processing/targeting, whereas S-palmitoylation of Cys-361/ 363 stabilizes the receptor at the plasma membrane.

Discussion
We describe here S-palmitoylation of the mammalian ␤ 3 AR. Both the mouse and human ␤ 3 ARs are S-palmitoylated at the canonical site within the C-terminal cytoplasmic tail, but the h␤ 3 AR is S-palmitoylated at two additional Cys within the second and third intracellular loops. The site within the second intracellular loop, Cys-153, appears to be unique to humans, and the site within the third intracellular loop, Cys-292, is conserved principally among primates. This observation provides, to the best of our knowledge, the first example of human/ primate-specific modification of any GPCR.
It has been demonstrated previously that differential S-palmitoylation distinguishes ␤ARs. Both the human ␤ 1 AR and ␤ 2 AR have been shown to be S-palmitoylated at the canonical C-tail site (in general, other species have not been examined), but the ␤ 1 AR is S-palmitoylated at an additional Cys more distal within the C-terminal tail (9) and the ␤ 2 AR is S-palmitoylated at an additional Cys within the third intracellular loop (15), neither of which are conserved across the ␤ARs.
Our results indicate that S-palmitoylation of the C-tail versus intracellular loop sites within the h␤ 3 AR subserve different functions: S-palmitoylation of the canonical-site Cys affects both receptor-effector coupling (as readout by ligand-induced cAMP production) and plasma-membrane receptor stability, whereas S-palmitoylation of the sites within the second and third intracellular loops affects receptor processing/targeting. The intracellular loop sites within the h␤ 3 AR are S-palmitoylated under basal conditions, whereas S-palmitoylation of the targeted Cys within the third intracellular loop of the ␤ 2 AR is strictly dependent upon ligand-induced activation of the receptor and subsequent phosphorylation, internalization, and trafficking to the Golgi (15). The possible functional roles(s) of changes in conformation of the plasma-membrane-localized receptor that would result from S-palmitoylation of Cys-153 and Cys-292 remains to be explored.
S-Palmitoylation of Cys-361/363 within the h␤ 3 AR also affects plasma-membrane receptor stability and thereby provides a third S-palmitoylation-dependent mechanism for regulating plasma-membrane receptor abundance. The ␤ 3 AR is well recognized as a tertium quid among ␤ARs in that it lacks the sites of phosphorylation that regulate the protein-protein interactions mediating ligand-coupled receptor internalization and is not internalized following activation (23,24). Generally, little if anything is known about the mechanisms that might influence ␤ 3 AR plasma-membrane abundance. Our finding that S-palmitoylation of the canonical C-terminal site within the h␤ 3 AR regulates plasma-membrane stability and thereby plasma-membrane abundance provides the first demonstration of which we are aware of regulation of ␤ 3 AR disposition by post-translational modification.
Our finding that S-palmitoylation of the h␤ 3 AR influences receptor processing, targeting, and plasma-membrane stability is broadly consistent with extensive prior literature describing the roles of S-palmitoylation of GPCRs (5,25). Our discovery that S-palmitoylation of Cys-292 and Cys-153 is required for proper receptor processing and targeting, respectively, whereas S-palmitoylation of Cys-361/363 stabilizes the receptor at the plasma membrane provides a demonstration of different functionality of S-palmitoylation of different sites within the h␤ 3 AR, which extends previous descriptions of different functionality of S-palmitoylation of different sites within the (human) ␤ 1 AR (9) and ␤ 2 AR (15).
Following the initial report that mutation of the Cys at the canonical C-tail site of S-palmitoylation within the (human) ␤ 2 AR diminished ligand-induced and G protein-dependent cAMP production, multiple mutational analyses in different GPCRs have shown that S-palmitoylation of Cys comprising the canonical site either does or does not suppress ligand-effector coupling. In the case of ␤ARs, mutation of S-palmitoylated Cys within the (human) ␤ 1 AR, unlike the case in the (human) ␤ 2 AR (4), had no effect on ligand-induced cAMP production (9). Our finding that mutation of Cys comprising the canonical site diminishes ligand-induced cAMP production via the human but not the mouse ␤ 3 AR provides the first example of which we are aware in which S-palmitoylation of the canonical site in a particular GPCR differentially regulates receptor-effector coupling in different mammalian species.
The ␤ 3 AR is expressed in multiple tissues, including adipocytes and the heart (26,27). ␤ 3 ARs are the principal effectors of the sympathetic influence on adipocytes, and in rodents, ␤ 3 AR agonists increase energy expenditure and fatty acid oxidation, deplete fat stores, preserve lean body mass, and improve insulin sensitivity (28). ␤ 3 AR agonists therefore represent promising therapeutic agents in obesity and in type 2 diabetes. However, human trials to date have failed (28), pointing to functional differences between the human and mouse ␤ 3 ARs. In the myocardium, multiple studies in mouse models point to a potential beneficial role of ␤ 3 AR stimulation in heart failure, and clinical trials designed to assess the efficacy of ␤ 3 AR agonism in heart failure are underway (29). Our results demonstrate multiple differences between the h␤ 3 AR and m␤ 3 AR assessed in vitro. The half-life at the plasma membrane of the m␤ 3 AR is greatly extended vis-à-vis the h␤ 3 AR, and the half-life of plasma-membrane-localized h␤ 3 AR but not the m␤ 3 AR is determined at least in part by S-palmitoylation. More generally, the disposition of the h␤ 3 AR but not the m␤ 3 AR is substantially determined by S-palmitoylation. Furthermore, S-palmitoyla-S-Palmitoylation of the ␤ 3 -adrenergic receptor tion affects receptor-effector coupling of the h␤ 3 AR but not of the m␤ 3 AR. Our demonstration that the h␤ 3 AR has a unique molecular signature among mammalian ␤ 3 ARs supports increased caution in the design of ␤ 3 AR-directed therapeutic approaches based largely on animal models.

Cell culture and transfection
HEK293 cells were maintained in Eagle's minimum essential medium (Wako) with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin at 37°C in a humidified 5% CO 2 atmosphere. Cells were transfected at ϳ70% confluence using Lipofectamine 3000 (Invitrogen) according to the manufacturer's instructions. To generate stable cell transfectants, G418 (Nacalai Tesque) was added to a final concentration of 150 g/ml active antibiotic.

Detection of S-palmitoylation by acyl-RAC
The acyl-RAC method was applied essentially as described (18) with minor modifications. Cells in a 6-well plate were harvested and lysed in 400 l of blocking buffer (100 mM HEPES, 1 mM EDTA, 2.5% SDS, 0.1% methyl methanethiosulfonate, pH 7.4), disrupted by sonication, and incubated at 50°C for 10 min. Following two acetone precipitations, the pellets were washed with 70% acetone and resuspended in binding buffer (100 mM HEPES, 1 mM EDTA, 1% SDS, pH 7.4). Total protein was quantified with a bicinchoninic acid assay (BCA; Pierce) using BSA as the standard, and equal amounts of protein (50 -200 g) were rediluted in 150 l of binding buffer. Approximately 60 g of protein from each sample was retained to assess input. When used, an equal volume of freshly prepared 1 M NH 2 OH, pH 7.2, was added followed by ϳ30 l of prewashed thiopropyl-Sepharose. Binding was carried out on a rotator at room temperature for 2 h. The resin was washed four times with binding buffer and eluted in 25 l of SDS sample buffer containing 1% 2-mercaptoethanol at 42°C for 10 min prior to SDS-PAGE and Western blotting.

Western blot analysis and data presentation
Western blotting signals were detected using the ChemiDoc XRS system (Bio-Rad), and densitometric analysis was performed with Quantity One software. All comparisons were made between bands on a single blot. Densitometric values for S-palmitoylated ␤ 3 AR were normalized with respect to total ␤ 3 AR. However, we found that population stoichiometry (i.e. the proportion of ␤ 3 ARs that are S-palmitoylated) cannot be determined from our results. This is because Western blots of FLAG from samples derived from acyl-RAC and from starting material cannot be compared directly: we never saw a ratio of S-palmitoylated ␤ 3 AR/total ␤ 3 AR less than 1, but our analysis shows that the values for S-palmitoylated ␤ 3 AR following acyl-RAC are in fact always proportionately greater than those for total ␤ 3 AR. This is because multiple proteins in the range of ϳ40 -70 kDa are present in the starting material samples (but not the acyl-RAC samples) and evidently (epitope-) mask the Western blotting signal from the ␤ 3 AR in starting material samples. All Western blotting data shown in the figures are from single blots. Within-blot cuts, for clarity of presentation, are indicated by dotted lines.

Assay of cAMP
HEK293 cells transiently or stably overexpressing WT or mutant FLAG-␤ 3 AR were stimulated with mirabegron in the presence of the phosphodiesterase inhibitor 4-(3-butoxy-4methoxybenzyl) imidazolidin-2-one (Ro 20-1724; 20 M). Cells were washed and collected, and cAMP was assayed with a cAMP Parameter Assay kit (R&D Systems) according to the manufacturer's instructions.

Immunofluorescence staining and confocal imaging
To visualize plasma-membrane receptors, cells in glass-bottom dishes were exposed to anti-DDDDK (FLAG) tag mAb-Alexa Fluor 488 (1:100) for 1 h on ice followed by fixation with 4% paraformaldehyde in PBS and incubation in 10% normal goat serum in PBS containing 0.1% Tween 20 for 30 min before exposure to anti-DDDDK (FLAG) tag mAb-Alexa Fluor 594 (1:100; 1 h). The total population of receptors was visualized identically except that cells were permeabilized with 10% normal goat serum in PBS containing 0.1% Tween 20 for 30 min before initial exposure to anti-DDDDK (FLAG) tag mAb-Alexa Fluor 594 (1:100; 1 h). Hoechst 33342 (0.2 g/ml for 5 min) was used for nuclear staining. Immunostaining was assessed using a confocal immunofluorescence microscope (Carl Zeiss, LSM700).

FACS analysis of total or cell-surface receptor abundance
To examine total receptor abundance, WT or mutant FLAG-␤ 3 AR was transiently expressed in HEK293 cells. One day after S-Palmitoylation of the ␤ 3 -adrenergic receptor transfection, cells were washed with PBS and harvested with 0.05% trypsin, EDTA. Cells were then fixed with 1% paraformaldehyde in PBS for 15 min and incubated in 10% normal goat serum in PBS containing 0.1% Tween 20 for 15 min on ice before exposure to anti-DDDDK (FLAG) tag mAb-Alexa Fluor 488 (1:150; 20 min). After washing and passing through a sieve (catalog number 38030, Falcon), FACS was carried out with a BD Accuri C6 flow cytometer (BD Biosciences). To examine cell-surface receptor abundance, HEK293 cells transiently expressing WT or mutant FLAG-␤ 3 AR in 6-well plates were exposed to anti-DDDDK (FLAG) tag Ab (1:500) for 20 min at 37°C in culture medium. The cells were washed with ice-cold PBS and harvested with 0.05% trypsin, EDTA. The cells were resuspended in ice-cold stain buffer (2% fetal bovine serum and 0.09% NaN 3 in PBS) and exposed to anti-mouse IgG conjugated with Alexa Fluor 488 (1:150) for 20 min. After washing and sieving, FACS was carried out with a BD Accuri C6 flow cytometer.
To examine plasma-membrane receptor stability (turnover), WT or mutant FLAG-␤ 3 AR was transiently or stably expressed in HEK293 cells. Cells were labeled extracellularly with DDDDK (FLAG) tag Ab (1:500) for 20 min at 37°C. Excess Ab was removed by two washes with fresh medium, and cells were then incubated with mirabegron (10 M) at 37°C in a humidified 5% CO 2 atmosphere. Cells were harvested with 0.05% trypsin, EDTA and incubated with anti-mouse IgG conjugated with Alexa Fluor 488 (1:150) for 20 min in ice-cold staining buffer. After washing and passing through a strainer, cell-surface fluorescence was detected with a BD Accuri C6 flow cytometer. Data were fitted to a one-phase decay curve and are presented as mean Ϯ S.D.

Reverse transcription and real-time PCR
RNA was extracted using TRIzol RNA isolation reagents (Invitrogen), and 3 g of RNA was treated with DNase I (catalog number 2270A, Takara Bio Inc.) and used to prepare cDNA with random hexamer oligonucleotide primers using the SuperScript III first-strand synthesis system for RT-PCR (Invitrogen). Gene-specific primers were used for real-time PCR in a LightCycler 480 System (Roche Applied Science) using iTaq SYBR Green Supermix with ROX (Bio-Rad). The expression of GAPDH in each sample was used to normalize the expression of the genes of interest. Sequences of the sense and antisense oligonucleotides, respectively, are 5Ј-GGCCCTCTCTAGTT-CCCAG-3Ј and 5Ј-TAGCCATCAAACCTGTTGAGC-3Ј for mouse ␤ 3 AR, 5Ј-GACCAACGTGTTCGTGACTTC-3Ј and 5Ј-GCACAGGGTTTCGATGCTG-3Ј for human ␤ 3 AR, and 5Ј-ACATCGCTCAGACACCATG-3Ј and 5Ј-TGTAGTTGA-GGTCAATGAAGGG-3Ј for GAPDH. The -fold change in expression was calculated using the comparative Ct method.

Data presentation and statistical analyses
All quantified data are presented as mean Ϯ S.D. Comparisons between groups using Student's t test or one-way ANOVA and Dunnett or Tukey post hoc test as appropriate were conducted using GraphPad Prism 7 (La Jolla, CA). p Ͻ 0.05 was considered significant.