Glycyrrhetic Acid Synergistically Enhances β2-Adrenergic Receptor-Gs Signaling by Changing the Location of Gαs in Lipid Rafts

Glycyrrhetic acid (GA) exerts synergistic anti-asthmatic effects via a β2-adrenergic receptor (β2AR)-mediated pathway. Cholesterol is an important component of the structure and function of lipid rafts, which play critical roles in the β2AR-Gs-adenylate cyclase (AC)-mediated signaling pathway. Owing to the structural similarities between GA and cholesterol, we investigated the possibility that GA enhances β2AR signaling by altering cholesterol distribution. Azide-terminal GA (ATGA) was synthesized and applied to human embryonic kidney 293 (HEK293) cells expressing fusion β2AR, and the electron spin resonance (ESR) technique was utilized. GA was determined to be localized predominantly on membrane and decreased their cholesterol contents. Thus, the fluidity of the hydrophobic region increased but not the polar surface of the cell membrane. The conformations of membrane proteins were also changed. GA further changed the localization of Gαs from lipid rafts to non-raft regions, resulting the binding of β2AR and Gαs, as well as in reduced β2AR internalization. Co-localization of β2AR, Gαs, and AC increased isoproterenol-induced cAMP production and cholesterol reloading attenuated this effect. A speculation wherein GA enhances beta-adrenergic activity by increasing the functional linkage between the subcomponents of the membrane β2AR-protein kinase A (PKA) signaling pathway was proposed. The enhanced efficacy of β2AR agonists by this novel mechanism could prevent tachyphylaxis.


Introduction
Lipid rafts are small (10 nm to 200 nm), heterogeneous, highly dynamic, cholesterol and sphingolipid-enriched microdomains of plasma membranes [1]. Numerous membrane proteins are known to partition into lipid rafts. Such proteins include glycosylphosphatidyl inositol (GPI)-anchored proteins [2,3], doubly acylated proteins, such as Src-family kinases and the a-subunits of heterotrimeric G proteins [4], cholesterol-linked proteins, and palmitoylated proteins [5]. Lipid rafts have been implicated in the regulation of various cell signaling pathways, and recent studies have indicated that many multi-component signaling pathways are coordinated by co-localization in lipid rafts, including the immunoglobulin E [6,7], T-cell antigen receptor [8,9], Ras [10], and G protein-coupled receptor (GPCR) [11] signaling pathways. While the functional significance of lipid rafts has not been completely understood, the rafts appear to enrich receptors and effectors, thus facilitating the interaction of signaling components, as in serotonin-1A receptor and neurokinin-1 receptor signaling regulation [12,13]. However, despite the positive role of lipid rafts, raft disruption can also enhance signal transduction through epidermal growth factor receptors [14], revealing a potential negative regulatory role in signal transduction.
Cholesterol serves as a spacer between the saturated chains of sphingolipids and is essential for maintaining the liquid-ordered phase of rafts and segregating embedded proteins from the rest of the membrane [15,16]. Small changes in the plasma membrane cholesterol content near the physiological set point may alter a variety of large biological responses [17,18]. Accumulating evidence also suggests that membrane cholesterol could impinge on signal transduction by directly affecting receptor affinity [11] and protein activity via sterol-sensing domains [19]. In the canonical signaling pathway mediated by seven transmembrane b 2 -adrenergic receptors (b 2 AR), cholesterol depletion was found to increase b 2 AR-stimulated cAMP production [20,21]. Cholesterol supplementation has an inverse effect on such production [22]. Some natural cholesterol derivatives, including steroid hormones, phytosterols, or saponins, can also alter signal transduction by changing the integrity of lipid rafts [23,24]. Considering the properties of plasma membrane, cholesterol or cholesterol-like substances may play important roles in signal transduction.
Liquorice, one of the best-known herbaceous plants, has been widely used in traditional Chinese medicine for thousands of years. Glycyrrhizin, a triterpene saponin and the major pharmacologically active compound in liquorice, possesses various pharmaco-logical effects, such as anti-inflammatory [25], anti-viral [26,27], and anti-carcinogenic activities [28]. Glycyrrhizin also exhibits anti-allergic effects and has been demonstrated to be useful for treating airway and lung disorders [29,30]. In a previous study, we demonstrated that co-treatment with glycyrrhizin and b 2 AR agonists yields synergistic anti-asthmatic effects, and that glycyrrhizin enhances b 2 AR signaling by increasing b 2 AR agoniststimulated cAMP accumulation [31]. Pharmacological studies have revealed that after oral administration, glycyrrhizin is hydrolyzed to glycyrrhetic acid (GA), the aglycone of glycyrrhizin, by human intestinal bacteria prior to absorption through a specialized b-glucuronidase [32]. However, GA has higher bioavailability than glycyrrhizin [33], and whether or not GA has functions similar to glycyrrhizin remains to be determined.
In this study, we investigated the underlying mechanism of GA regulatory effect on b 2 AR-Gas-adenylate cyclase (AC) signal transduction. GA was demonstrated to change the localization of Gas from lipid rafts to the non-raft region. The results show GA reduced level of b 2 AR internalization, thereby increasing downstream cAMP production. The effect of GA on b 2 AR-Gas-AC signaling is linked to its ability to decrease the cholesterol content of lipid. This effect is evidenced by the finding that the targeting of a GA probe to membrane and enhanced b 2 AR-Gas-AC signaling are strongly attenuated by cholesterol reloading.

Cell Lines and Chemosynthesis
Hemagglutinin (HA)-b 2 AR HEK293 cells were derived from HEK293 cells. b 2 ARs were expressed with an N-terminal HA epitope tag. The enhanced green fluorescent protein-b 2 AR (b 2 AR-EGFP) plasmid was constructed in our laboratory [31]. The synthesis experimental procedures of the compounds azideterminal GA (ATGA), 4-ethynyl-N-ethyl-1, 8-naphthalimide and linker-biotin are described in the Supporting Information files.

Analysis of Fluorescent Labeling in Cells by Confocal Microscopy
HA-b 2 AR HEK293 cells were cultured and supplied with 0.1 mM GA (J&K Chemical, Beijing, China) or 0.1 mM ATGA for 1 h. After fixing with 4% paraformaldehyde, the cells were subjected to a probe labeling reaction (0.1 mM linker-biotin, 0.1 mM Tris-triazoleamine, 0.1 mM CuSO 4 , and 1 mM sodium ascorbate in phosphate-buffered saline, PBS) at room temperature for 30 min. Subsequently, the cells were labeled with mHA.11 antibody (Covance, CA, USA) and visualized using fluorescein isothiocyanate-conjugated secondary antibody and Cy3-conjugated streptavidin. Fluorescent images were captured by a LEICA TCS SP5 laser scanning confocal microscopy system (Leica Microsystems, Heidelberg, Germany).

Detergent-free Purification of Lipid Raft Membrane Fractions
The detergent-free purification of cholesterol-enriched microdomains was carried out according to the protocol described by Song et al. [34]. After two washes in cold PBS, HA-b 2 AR-HEK293 cells treated with or without 0.1 mM GA for 2 h were scraped into 500 mM sodium carbonate (pH 11.0) and homog-enized by sonication. The homogenate was adjusted to 40% sucrose, placed at the bottom of an ultracentrifuge tube (Beckman Instruments), and overlaid with a 5% to 35% discontinuous sucrose gradient.

Cholesterol and ATGA Assays
The cholesterol content of the sucrose gradient fractions was measured using a Cholesterol/Cholesteryl Ester Quantitation kit (Biovision, CA, USA). Each sucrose gradient fraction was extracted with chloroform/isopropanol/nonyl phenoxypolyethoxylethanol (NP-40) (7:11:0.1). The lipids were extracted, and all traces of organic solvents were evaporated prior to resuspending the lipids in the cholesterol assay buffer. The assays were performed according to the manufacturer's instructions. For the analysis of ATGA content, the samples were prepared as for cholesterol analysis, except that the assay buffer contained 0.1 mM 4-ethynyl-N-ethyl-1,8-naphthalimide, 0.1 mM Tris-triazoleamine catalyst, 0.1 mM CuSO 4 , and 1 mM sodium ascorbate. Then, the reaction mixture was transferred to 96-well plates and scanned by a Thermo Scientific Varioskan Flash spectral scanning multimode reader (Varioskan, Thermo Electron Co, MA, USA) with SkanIt software.

Filipin Fluorescence Staining of Cell Membrane Cholesterol
HA-b 2 AR HEK293 cells were cultured with or without 0.1 mM GA for 2 h at 37uC. After fixing with 4% paraformaldehyde, the cells were stained with 1 ml of a filipin working solution (0.05 mg/ ml in PBS with 10% fetal bovine serum) for 2 h. Images were obtained under a LEICA TCS SP5 laser scanning confocal microscopy system (Leica Microsystems, Heidelberg, Germany) using a UV filter set.
Spin Labeling of Cells with 5-doxyl Stearic Acid (5-DSA), 16-doxyl Stearic Acid (16-DSA), and 3-maleimide-proxyl (3-MP) The spin labels 5-DSA and 16-DSA consist of a radical group (doxyl) and a hydrocarbonated chain (stearic acid) that acts as a radical support. Since spin labels are oriented and linked like the lipids in the lipid bilayers of cell membranes. These nitroxyl radicals in the 5th or 16th positions of the alkyl chain can be used to determine local fluidity in the two main regions of the cell membrane, the region near the polar head group (5-DSA), and the hydrophobic region (16-DSA) within the lipid bilayers.
After treatment with 0.1 mM GA, the HA-b 2 AR HEK293 cell suspension was mixed with spin label 5-doxyl or 16-docyl (30 mg/ ml) and incubated at 37uC for 30 min. The free spin labels were washed out by PBS from the cell suspension until no signal was detected in the supernatant. The electron spin resonance (ESR) measurement conditions were as follows: microwave power, 20 mW; modulation frequency, 100 kHz; modulation amplitude, 2 G; sweep width, 100 G; and temperature, 24uC. The parameter S and rotational correlation time t c are defined as follows: where h 0 and h (21) are the peak heights of the center and high field lines, respectively, DH 0 is the width of the central line; and A I and A H are the parallel and perpendicular hyperfine splitting parameters of the spectrum, respectively. For 3-MP labeling, cells treated with GA were labeled by 3-MP (20 mg/ml) and incubated at 37uC for 3 h. Then, the suspension was washed in PBS until no ESR signal was detected in the supernatant. The measurement conditions were as follows: microwave power, 20 mW; modulation frequency, 100 kHz; modulation amplitude, 1 G; sweep width, 300 G; and temperature, 24uC.

Co-immunoprecipitation of b 2 AR with b-arrestins and Clathrin
Confluent cells were scraped into ice-cold lysis buffer (pH 7.4 50 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA, 0.1% SDS, 1% sodium deoxycholate, 1% Triton X-100, and 1 mM phenylmethylsulfonylfluoride) and incubated for 3 h at 4uC. The total cellular protein mixture was centrifuged at 21 0006g for 15 min. The supernatant (detergent-soluble extract) was then mixed with 20 mL of an anti-HA affinity matrix (Roche Diagnostics, Indianapolis, IN, USA) at 4uC overnight. After washing with ice-cold lysis buffer, the immune complexes were boiled in SDS sample loading buffer and subjected to electrophoresis on a 4-20% Tris-glycine gel. Immunoblotting was executed with anti-b-arrestins or clathrin (Santa Cruz, San Diego, CA, USA) heavy chain polyclonal antibody.

Live Cell Imaging Microscopy for Receptor Internalization Observation
To monitor the internalization of b 2 ARs upon receptor activation, HEK293 cells were transiently transfected with the b 2 AR-EGFP plasmid. The cells with or without 0.1 mM GA pretreatment were stimulated with 0.2 mM isoproterenol (Sigma, St. Louis, MO, USA) for up to 30 min. b 2 AR-EGFP fluorescent signals were recorded under a LEICA TCS SP5 confocal laser scanning microscope (Leach Instruments, Heidelberg, Germany). The excitation and emission wavelengths employed were 488 and 520 nm, respectively. Sequential confocal sections (z plane) were acquired at intervals of 0.5 mm from the middle to the bottom of the cells.

Fixed-cell ELISA Assays for Cell Surface b 2 AR Detection
HA-b 2 AR-HEK 293 cells growing on 96-well microplates were exposed to 0.1 mM salbutamol for up to 2 h at 37uC, fixed with 4% paraformaldehyde, and blocked with 4% skimmed milk. Antibody mHA.11 and HRP-labeled second antibody were used to detect HA-b 2 AR. Finally, the content of HA-b 2 AR was determined by a 3, 39, 5, 59-tetramethylbenzidine solution.

Luciferase Reporter Assay for cAMP Accumulation
Cells were co-transfected with pCRE-Luc reporter plasmid and pRL-TK plasmid (Promega, Madison, USA). After 24 h, cells were treated with different concentrations of the indicated drugs for 5 h. In the glycyrrhizin pretreatment group, cells were pretreated with 0.1 mM GA for 3 h, and the reaction was stopped by washing with PBS buffer. The cells were lysed and the luciferase activity was measured using a Dual-Luciferase reporter assay system (Promega). Luminescence was detected using a Modulus Luminometer from Turner Biosystems (Turner design, Sunnyvale, CA, USA). The ratio of firefly luciferase activity to renilla luciferase activity was used to normalize for differences in the transfection efficiency.

Statistical Evaluation
Data were expressed as mean 6 S.E.M. of triplicate determinations from at least three independent experiments. Statistical analysis was performed using Student's t-test or two-way analysis of variance (ANOVA). p,0.05 was considered statistically significant.

Cellular Localization of ATGA
To identify the cellular targets of GA, we prepared an ATGA probe. The synthesis of ATGA is outlined in Materials S1. Owing to the azido modification, ATGA could undergo a Cu (I)-catalyzed [3+2] cycloaddition reaction with an alkyne group (Fig. 1) [35]. Fig. 2A shows the location of ATGA. The blue fluorescence is clearly concentrated on the cell membrane (b and e). In contrast, only a low level of background fluorescence was observed in the GA-treated control group (a and d), supporting the specificity of the click-chemistry reactions between the azide (ATGA) and alkyne groups (4-ethynyl-N-ethyl-1,8-naphthalimide), which is synthesized through the process described in Materials S2. The pretreatment of cells with GA significantly inhibited the uptake of ATGA, indicating that ATGA and GA compete for the same cellular targets (c and f). These observations suggest that ATGA mainly localizes in the cell membranes and provides a suitable mimic for GA. Furthermore, a fluorescence microscopy of ATGA and b 2 AR was performed. ATGA was reacted with biotin linked alkyne (Materials S3) and visualized by Cy3 labeled strepavidin (red) and the b 2 ARs were visualized with a proper antibody (green). The week co-location of ATGA and b 2 AR shows in Fig. 2B indicated that ATGA and b 2 AR may exist in different microdomains.

GA Decreased the Cholesterol Content of Lipid Rafts
The staining of cholesterol (Fig. 2C) shows the effect of GA on cholesterol localization. As cholesterol is the primary content of lipid rafts, it was mainly detected on cell membrane. Under GA treatment, cholesterol was dispersed as a much weaker signal compared with the control experiment. To confirm the decrease effect of GA on cholesterol on cell membrane, we evaluated the cholesterol content in different cell fractions. Triton X-100 sucrose-gradient fractions were analyzed for total protein, cholesterol, and ATGA contents. As shown in Fig. 2D, the bulk of the cholesterol was found in the low-density fractions at the upper end of the gradient (fractions 2 to 4) which should correspond to the lipid rafts distribution, and the content significantly decreased when treated with GA. The distribution of ATGA was also determined by reaction with 4-ethynyl-N-ethyl-1, 8-naphthalimide. The fluorescence intensity indicated that ATGA distributes predominantly in light-density fractions (fractions 2 to 4), which also correspond to cholesterol (Fig. 2E). This result indicates that the ATGA might decrease the cholesterol on membrane by a displacement effect.

GA Changed the Fluidity of Cell Membranes
Cholesterol is a major determinant of membrane fluidity. Previous findings suggest that GA displacement alters the membrane fluidity of lipid rafts similar to other cholesterol derivatives [36]. ESR was utilized to investigate the effect of GA on membrane fluidity and membrane protein conformations. The ESR spectra of spin-labeled membranes with 5-DSA and 16-DSA are shown in Fig. 3A and Fig. 3B. The order parameter S and rotational correlation time t c calculated from the spectra of 5-DSA and 16-DSA were 0.655160.003 and 16.0360.11610 210 (s), respectively. Compared with the control group, the order parameter remained almost unchanged after GA treatment (0.665360.005), whereas t c significantly decreased after GA treatment (15.1360.15, p,0.05). This result indicates that GA significantly increases lipid fluidity in the hydrophobic region of plasma membranes.
To identify change in the properties of membrane proteins, 3-MP spin labels that can specifically bond with the sulfhydryl groups of the proteins were employed. GA treatment decreased the S/W ratio from 5.3160.09 in the control HA-b 2 AR-HEK293 cells to 4.6760.11 after treatment with 0.1 mM GA (p,0.05). This change in S/W indicates that the structures of sulfhydryl binding sites on membrane proteins become tighter after GA treatment owing to conformational changes caused by GA. The results reveal that GA treatment influences the fluidity in the hydrophobic regions of cell membranes, and thereby alters the conformations of many membrane proteins.

GA Changed the Localization of Signaling Molecules Ga s and Increased b 2 AR/G Protein Coupling in Plasma Membrane
We investigated whether or not GA could change the localization of signaling molecules in lipid rafts. Membranes from GA-treated and control HA-b 2 AR-HEK293 cells were fractionated by the sodium carbonate procedure. The signaling molecules present in the fractions were then determined by immunoblotting (Fig. 4A) [34]. The lipid raft marker molecule flotillin and the scaffold protein of caveolae, caveolin-1, were mainly detected in fractions 4 and 5, which are considered to containing lipid raft fractions. The G-protein subunit Ga s was abundant in the lipid raft fractions from control cells. Treatment ofcells withGA resulted a significant loss ofGa s inlipid raft fractions, this part of Ga s might translocated to high-density fractions (fractions 8-12), whereas the distributions of flotillin and caveolin-1 were not affected.
Co-immunoprecipitation experiments were performed to identify the G proteins associated with b 2 ARs. Western blot analysis of b 2 AR immunoprecipitations revealed that Ga s was preassociated with the receptor under the basal condition in untreated cells (Fig. 4B, left panel, 0 min). Upon the stimulation of isoproterenol, Ga s rapidly separated from b 2 AR (Fig. 4B, left panel, 15 min). When the cells were pretreated with GA, we found a significant increase in the association of G proteins with unstimulated b 2 AR (Fig. 4B, 15 min). No significant reduction in the amount of these b 2 AR/Ga s complexes was observed after 15 min of isoproterenol stimulation.

GA Pretreatment Significantly Inhibited the Internalization of b 2 AR
To investigate the influence of GA on the internalization of b 2 ARs in live cells, the subcellular distribution of b 2 AR-EGFP was examined in HEK293 cells. As shown in Fig. 5, isoproterenol treatment induced b 2 ARs to move rapidly into punctate intracellular vesicles (Fig. 6A, 5 min to 30 min) with the appearance of early endosomes [37]. However, this phenomenon was repressed by a pretreatment of GA (Fig. 6B). A quantitative ELISA assay was carried out for the quantitative determination of cell surface b 2 ARs. Treatment with isoproterenol for 2 h caused a maximal 21.11%67.32% loss in cell surface b 2 ARs (Fig. 6C). In contrast, the GA pretreated group showed significantly lower loss. Co-immunoprecipitation of clathrin heavy chain with HA-b 2 AR also revealed that GA could decrease b 2 AR binding to clathrin (Fig. 4B, 15 min to 45 min), a necessary step in   [38]. Taken together, GA inhibited the isoproterenol-induced b 2 AR internalization.

GA-regulated b 2 AR Signaling Efficacy
Finally, to investigate whether or not GA influences the efficacy of the b 2 AR-Ga s -AC signaling cascade, cAMP levels were evaluated in HA-b 2 AR-HEK293 cells using the cAMP-protein kinase A (PKA) reporter system. Indeed, both basal and maximal isoproterenolstimulated cAMP production increased after 0.1 mM GA pretreatment (Fig. 6A). To determine the specific site within the b 2 AR-Ga s -AC signaling cascade sensitized by 0.1 mM GA treatment, we further investigated functional responses to the non-hydrolysable GTP analogue 5-guanylimidodiphosphate (Gpp(NH)p) and to the AC agonist forskolin. Maximal Gpp(NH)p-stimulated cAMP production increased twofold after GA pretreatment (Fig. 6B), suggesting that a G protein may be involved in the GA-mediated response. Forskolinstimulated cAMP production was also significantly increased, which is consistent with the notion that forskolin activates AC in a G proteindependent manner (Fig. 6C) [39]. However, GA could not reverse the blocking effect of 10 mM ICI 118 551 on the b 2 ARs (Fig. 6D).
To test whether the signaling efficacy can be regulated by the GA induced cholesterol loss on membrane, we assessed the influence of cholesterol reloading on the b 2 AR signaling efficacy. Artificially increased membrane cholesterol concentration (100 mM) without prior GA treatment blocked the synergistic effect on b 2 AR signaling (Fig. 6E). This finding indicates that the changes in membrane cholesterol are related with b 2 AR-Ga s -AC signaling.

Discussion
Earlier studies indicate that decreased membrane fluidity in the presence of glycyrrhizin could suppress the infectivity of HIV-1, influenza virus, and vesicular stomatitis virus [40,41]. Recently, Schrofelbauer et al. demonstrated that glycyrrhizin attenuates pro-inflammatory responses by interfering with membrane-dependent receptor signaling [42]. These studies have led to the hypothesis that glycyrrhizin or GA may be incorporated into lipid bilayers to influence receptor signaling by altering the integrity and fluidity of the plasma membrane. In the present work, we treated cells with ATGA, azido-labeled GA, to assess cellular distribution. Based on both the biotin-streptavidin detection and 4-ethynyl-N-ethyl-1, 8naphthalimide probe systems, most of the fluorescence was found to be concentrated at certain areas of the cell membrane and little was observed in the cytoplasm. Furthermore, a displacement effect of GA on cholesterol was detected by imaging of cholesterol location and sucrose gradient separation. Farther, the result of Spin-label ESR spectroscopy showed that GA treatment greatly influenced the hydrophobic regions of cell membranes but had little effect on the polar surface. Although GA has a rigid highly hydrophobic steroid ring structure, it lacks the bulky non-polar hydrocarbon tail of cholesterol. This feature could explain why GA increased the fluidity of the hydrophobic regions of the cell membranes but not the more polar surface domains. A previous study found that glycyrrhizin decreases membrane fluidity under 5-DSA labeling [40]. The influence of glycyrrhizin on the fluidity of the polar surface of cell membranes could be attributed to two more molecules of glucuronic acid at the polar head of GA. The decrease in S/W, which indicates a transition of many membrane proteins to a tighter tertiary conformation, may be caused by the GA-mediated formation of an active ternary complex with stronger sulfhydryl binding than in the inactive state.
Cholesterol plays an important function in membrane lipid raft. It is known to participate in various signal transductions. Changes in cholesterol content affect the localization of proteins in rafts [3,43]. Based on lipid raft preparations, we established that b 2 ARs were excluded from lipid raft domains where were rich of Ga s . After GA treatment, we observed significant translocations of Ga s from lipid raft fractions to the high-density fractions. However, minimal changes were observed in the raft levels of flotillins and  The enhanced coupling of b 2 ARs and Ga s after treatment with GA suggests that GA influences the efficiency of receptor activation. Currently, the most widely accepted model for GPCR activation is the extended ternary complex model (the two-state model) [44,45]. This model proposes that the receptor exists in an equilibrium between an inactive conformation (R) and an active conformation (R*) that is coupled to the G protein. In the absence of an agonist, the inactive R state prevails. However, the energy barrier between the R and R* states is sufficiently low to allow a certain fraction of the receptors to spontaneously assume the R* state. Agonists are predicted to bind with higher affinity to the R* conformation and consequently shift the equilibrium as well as increase the proportion of receptors in R*. In this paper, GA treatment increased the opportunity of molecular collisions between b 2 ARs and Gas to increase through interaction with cholesterol, as well as by releasing a sequestered pool of Ga s from lipid rafts that become available for b 2 ARs engagement. The increased b 2 ARs/Ga s coupling promotes an increase in the active conformation (R*) of b 2 AR, thereby leading to increased cAMP responses to b2AR agonists. In this study, GA was also found increases the maximal Gpp(NH)p and forskolin-stimulated cAMP production, as Ga s was found embedded in lipid raft domains and be excluded under GA treatment, this observation indicated that lipid rafts appear to play a negative regulatory role in b 2 AR signal transductions. Indeed, cholesterol depletion has been found to increase b 2 AR-stimulated cAMP production [20,21], whereas cholesterol supplementation has an inverse effect [22]. This finding was also confirmed by our present data. Cholesterol reloading significantly attenuated the effect of GA on agoniststimulated cAMP production, further suggesting that GA interacts with cholesterol in lipid rafts.
In contrast, several studies established that b 2 ARs undergo rapid phosphorylation by both second messenger dependent protein kinases and GPCR kinases in response to agonist stimulation [38]. This event targets receptors for binding to arrestin proteins that uncouple receptors from their cognate heterotrimeric G proteins, and favor receptor endocytosis via clathrin-coated vesicles into endosomal compartments [46,47]. Our results demonstrate that GA could significantly inhibit isoproterenol-induced b 2 AR internalization, consistent with a previous study demonstrating that saponins could inhibit receptor internalization due to their cholesterol-complexing characteristics [48][49][50][51]. We speculate that the constant coupling between b 2 AR and Ga s allows the receptor to maintain a sterical conformation not suitable for the binding of protein kinases and ensuing arrestin proteins. This speculation could also explain the GA-mediated reduction in the association between clathrin and b 2 AR. Di Certo et al. demonstrated that the fusion of b 2 AR to Ga s slows agonistinduced internalization and strongly affects the recycling of receptors to the plasma membrane [52]. Bertin et al. reported that a b 2 AR/Ga s fusion protein is more resistant to desensitization and enhances the anti-proliferative effect of isoproterenol in transfected S49 cells [53].
Taken together, we speculate that unlike b 2 ARs themselves, a significant fraction of their signaling partners is embedded in lipid raft microdomains of the membrane. GA can decrease the Figure 6. Effects of GA treatment on isoproterenol, forskolin, and GppNHp stimulated cAMP production. HA-b 2 AR-HEK293 cells pretreated with 0.1 mM GA or not were stimulated by isoproterenol (A), GppNHp (B) and forskolin (C). Detect of ICI 118 551 block effect were carry out with 0.1 mM GA pretreatment and 10 mM ICI 118 551 (D). Cholesterol (100 mM) were reloaded while 1 mM and 0.01 mM isoproterenol treatment (E). Further, the cAMP levels were determined by the cAMP-protein kinase A (PKA) reporter system in the graphs. Significant differences between the groups with or without cholesterol reloading were found. **means p,0.001; *means p,0.05, Student's t test (n = 5). doi:10.1371/journal.pone.0044921.g006 cholesterol content in lipid rafts and change the fluidity of the cell membrane, resulting in the release of raft-embedded Ga s as well as increased interaction with b 2 ARs, thereby increasing b 2 AR/Ga s coupling and decreasing receptor internalization. Ultimately, b 2 AR-mediated signal transduction is enhanced.

Supporting Information
Materials S1 Synthesis of azide-terminal glycyrrhetic acid.