Dissociations in the Effects of β2-Adrenergic Receptor Agonists on cAMP Formation and Superoxide Production in Human Neutrophils: Support for the Concept of Functional Selectivity

In neutrophils, activation of the β2-adrenergic receptor (β2AR), a Gs-coupled receptor, inhibits inflammatory responses, which could be therapeutically exploited. The aim of this study was to evaluate the effects of various β2AR ligands on adenosine-3′,5′-cyclic monophosphate (cAMP) accumulation and N-formyl-L-methionyl-L-leucyl-L-phenylalanine (fMLP)-induced superoxide anion (O2 •−) production in human neutrophils and to probe the concept of ligand-specific receptor conformations (also referred to as functional selectivity or biased signaling) in a native cell system. This is an important question because so far, evidence for functional selectivity has been predominantly obtained with recombinant systems, due to the inherent difficulties to genetically manipulate human native cells. cAMP concentration was determined by HPLC/tandem mass spectrometry, and O2 •− formation was assessed by superoxide dismutase-inhibitable reduction of ferricytochrome c. β2AR agonists were generally more potent in inhibiting fMLP-induced O2 •− production than in stimulating cAMP accumulation. (−)-Ephedrine and dichloroisoproterenol were devoid of any agonistic activity in the cAMP assay, but partially inhibited fMLP-induced O2 •− production. Moreover, (−)-adrenaline was equi-efficacious in both assays whereas the efficacy of salbutamol was more than two-fold higher in the O2 •− assay. Functional selectivity was visualized by deviations of ligand potencies and efficacies from linear correlations for various parameters. We obtained no evidence for involvement of protein kinase A in the inhibition of fMLP-induced O2 •− production after β2AR-stimulation although cAMP-increasing substances inhibited O2 •− production. Taken together, our data corroborate the concept of ligand-specific receptor conformations with unique signaling capabilities in native human cells and suggest that the β2AR inhibits O2 •− production in a cAMP-independent manner.

Classic models of G-protein-coupled receptor (GPCR) activation postulate the existence of a single active (R*) and an inactive (R) state [20][21][22]. In the active R* state, the receptor is assumed to activate its cognate G protein and regulate down-stream effectors. However, over the past 15-20 years, compelling evidence from various groups has accumulated that the R/R* dichotomy is too simplicistic. These studies comprise biochemical, pharmacological and biophysical approaches . Accordingly, it is now generally assumed that any given ligand stabilizes a ligand-specific receptor conformation with unique signaling capabilities, resulting in ligand-specific activation of G-proteins and/or b-arrestin [27,28]. Stabilization of ligand-specific receptor conformations with unique signaling capabilities is also referred to as functional selectivity or biased signaling [20][21][22][29][30][31][32][33]. To this end, most of the evidence for ligand-specific receptor conformations has been obtained in studies with recombinant systems [34][35][36] or purified receptor proteins [31,[37][38][39], but studies on the relevance of ligand-specific receptor conformations in native human cells are still largely missing. Reasons for this lack of knowledge are inherent difficulties to manipulate human cells genetically. In addition, after isolation, human blood cells such as neutrophils survive only for limited period of time [40]. However, since the concept of functional selectivity implies that certain ligands can be clinically more efficacious in a given setting while displaying less unwanted effects, it is of paramount importance to probe functional selectivity in native cells.
Recently, we have reported on the functional selectivity of another G s -coupled receptor, the histamine H 2 receptor, in two native cell systems, human eosinophils and neutrophils [40]. The pharmacological profiles of H 2 R agonists as well as H 2 R antagonists do not match by comparing their effects on eosinophils and neutrophils as well as by comparing these parameters with data obtained in a recombinant test system. Moreover, even in the same cell type, differences were observed when ligands were characterized determining two different parameters. Each ligand triggers unique effects depending on the test system and parameters measured which is of importance for further drug development.
The aim of the present study was the characterization of the b 2 AR on human neutrophils with a series of structurally diverse b 2 AR ligands and thereby, to probe the concept of ligand-specific receptor conformations on one of the most important and bestcharacterized GPCRs in a physiologically relevant native human cell system. Two distinct parameters were chosen for the characterization of the b 2 AR on neutrophils. The first parameter was measurement of the cAMP content in neutrophils. The second parameter was monitoring of the b 2 AR-mediated effects on fMLPstimulated O 2 N2 production. We also examined the effects of various pharmacological tools including protein kinase inhibitors, AC inhibitors and activators and various cAMP analogues on fMLP-induced O 2 N2 production in order to obtain further insights into the mechanisms underlying inhibition of NADPH oxidase.

Materials
(S*,S*)-3-(Isopropylamino)-1-(7-methyl-2,3-dihydro-1H-inden-4-yloxy)butan-2-ol) ((6)-(S*, S*)-ICI 118551) (ICI) and (6) Fig. S1. Stock solutions of ISO, ADR, SAL, DOB, EPH and DCI (10 mM each) were prepared in 1 mM HCl and stock solutions of ICI, MET, ALP, BIS and ATE in Millipore water. Dilution series of all ligands were prepared in Millipore water. Dulbecco's PBS (DPBS, 10 x) without Ca 2+ and Mg 2+ (pH 6.5-7.0) was purchased from PAN Biotech (Aidenbach, Germany) and Biocoll separating solution from Biochrom (Berlin, Germany). Trypan blue solution, ferricytochrome c, cytochalasin B, fMLP and IBMX were from Sigma-Aldrich. Solvents for extraction and HPLC analysis were purchased as follows: HPLC-gradient grade water and methanol from J. T. Baker (Deventer, The Netherlands), ammonium acetate from Sigma-Aldrich and acetic acid from Riedel-de Haen (Hannover-Seelze, Germany). Tenofovir was obtained from the National On human neutrophil granulocytes, the O 2 N2 assay (1610 5 cells per well) and the cAMP assay (5610 5 cells per cup) were performed as described under Materials and Methods. Data were analyzed by non-linear regression and were best fitted to sigmoidal concentration/response curves. Data shown are the means of four to nine independent experiments performed in triplicate. The efficacy (E max ) of ISO in each assay was set to 1.00 and the efficacies of other ligands were referred to this value. n.d. not determined. As DOB at concentrations higher than 500 nM caused reduction of ferricytochrome c per se, the reliable determination of pIC 50 value was not possible. a Efficacy at concentration of 500 nM was taken as E max . b The data were taken from [45]. The reported non-logarithmic EC 50 values were converted into logarithmic pEC 50 values. c The data were taken from [43]. The non-logarithmic EC 50 values shown in mM in parentheses were converted into logarithmic pEC 50

Isolation of Human Neutrophils
This study and the consent procedure were approved by the Ethics Committee of the Medical School of Hannover. Written consent was obtained by all volunteers. The completed and signed consent forms are kept on file in the secretary of the Institute of Pharmacology of the Medical School of Hannover. Human neutrophils were isolated from venous blood of healthy volunteers of either sex (1.6 mg EDTA/ml blood as anticoagulant) or from buffy coat obtained from the Institute for Transfusion Medicine (Medical School of Hannover, Germany). Buffy coat preparations were also obtained from individual donors. All isolation steps were carried out at room temperature. Firstly, 7 ml of venous blood or 5 ml of buffy coat were diluted to 35 ml with 16DPBS and carefully layered onto 15 ml of Biocoll separating solution (density 1.077 g/ml) in a 50 ml-Falcon tube. Following centrifugation (30 min, 4006g), the upper three layers were removed. The residual pellet (,2 ml), which contained erythrocytes and granulocytes, was resuspended in 18 ml of Millipore water and incubated for 1 min under gentle agitation in order to achieve selective lysis of erythrocytes. Afterwards, isotonicity was restored by adding 2.2 ml of 106DPBS, and centrifugation at 3006g for 5 min followed. The lysis step was repeated once to remove residual erythrocytes. The cell pellet was re-suspended in 5 ml of 16DPBS and sedimented by centrifugation at 3006g for 5 min. The resulting cell preparation consisted of viable neutrophils (.98%), as assessed by the trypan blue exclusion test. Finally, neutrophils were suspended in 16PBS (1610 6 cells/ml for the O 2 N2 assay or 1610 7 cells/ml for the determination of cAMP) and stored on ice until use. Experiments were performed within 4 h after completion of isolation because at later time points, viability of cells declined substantially as assessed by trypan blue dye uptake and declined responsiveness to receptor ligands (data not shown).

Superoxide Anion Generation (O 2 N2 Assay)
Reactions were carried out in 96-well plates in triplicate. Standard reaction mixtures (total volume 200 ml) contained 1 mM CaCl 2 , 100 mM ferricytochrome c, 0.3 mg/ml cytochalasin B (priming role by enhancing O 2 N2 formation upon exposure to fMLP) [2], ligands at different concentrations (where indicated, additionally PKA inhibitors, AC inhibitors or cAMP analogs) and 1610 5 neutrophils in 16DPBS. After pre-incubation of the reaction mixtures for 3 min at 37uC, reactions were initiated by addition of fMLP (1 mM). Reference samples contained all components listed above except for fMLP. O 2 N2 formation was continuously measured by monitoring the reduction of ferricytochrome c at 550 nm for 30 min at 37uC, using a Synergy 4 microplate reader (BioTek Instruments, Winooski, VT, USA). The difference in absorbance at 550 nm between 0 min (addition of fMLP) and 30 min was used for subsequent data analysis, in order to assess agonistic activity of examined ligands. With the exception of DOB, all examined test compounds did neither reduce ferricytochrome c nor stimulate O 2 N2 production per se nor acted as radical scavenger as assessed by the lack of effect on phorbol ester-stimulated O 2 N2 production (data not shown). As at DOB concentrations higher than 500 nM, ferricytochrome c reduction took place, the maximum concentration of DOB used in the O 2

N2
assays was 500 nM. Reactions were conducted in triplicate in 1.5 ml Eppendorf reaction vessels in a total volume of 100 ml. Fifty ml of the reaction mixture containing CaCl 2 (1 mM final concentration after addition of neutrophils), IBMX (non-selective phosphodiesterase inhibitor; 100 mM) and the respective ligand at different concentrations in 1 x DPBS were pre-incubated for 5 min at 37uC. Isolated neutrophils suspended in 16DPBS were pre-incubated separately for 10 min at 37uC. Following the addition of 50 ml of neutrophils (5610 5 cells/reaction vessel) to reaction mixture, samples were incubated for 10 min at 37uC. Afterwards, samples were incubated for 10 min at 95uC in order to denature cell proteins and then cooled to 4uC. One hundred ml of ice-cold internal standard (tenofovir; 100 ng/ml) in eluent A (3/ 97 MeOH/H 2 O, 50 mM NH 4 OAc, 0.1% HOAc) were added. The suspension was centrifuged at 20.8006g at 4uC for 5 min in order to remove denatured proteins. The cAMP concentration of the supernatant was determined by reversed phase HPLC coupled to mass spectrometry (HPLC-MS/MS).

Quantitation of cAMP by HPLC-MS/MS
In this study, cAMP levels were determined by HPLC-MS/ MS which is characterized by extremely high sensitivity and specificity [41][42]. Since this method is not yet commonly known and used, we describe the experimental protocol in some detail. The chromatographic separation was performed on an Agilent 1100 Series HPLC System (Agilent Technologies, Santa Clara, CA, USA) equipped with a binary pump system and with a 100 ml sample loop. A combination of Supelco Column Saver (2.0 mm filter, Supelco Analytical, Bellafonte, CA, USA), Security Guard Cartridge (C18, 462 mm) in an Analytical Guard Holder KJO-4282 (Phenomenex, Aschaffenburg, Germany) and an analytical Zorbax Eclipse XDB-C16 column (5064.6 mm, 1.8 mm particle size, Agilent Technologies), temperature controlled by a HPLC column oven at 25uC, were used. The binary pump system supplied eluent A (50 mM ammonium acetate and 0.1% (v/v) acetic acid in a methanol/ water mixture (3/97 (v/v)) and eluent B (50 mM ammonium acetate and 0.1% (v/v) acetic acid in a methanol/water mixture (97/3 (v/v)). The injection volume was 50 ml and the flow rate of 0.4 ml/min remained constant throughout the chromatographic run. From 0 to 5 min, the gradient of eluent B was linearly increased from 0 to 50% of eluent B, and reequilibrium of the column to 100% of eluent A was achieved from 5 to 8 min. Retention times of the analyte cAMP and the internal standard tenofovir were 6.2 and 5.4 min, respectively. The internal standard was used to mathematically correct the loss of cAMP during preparation as well as possible variabilities in HPLC-MS/MS measurement. Analyte detection was conducted on an AB Sciex QTRAP 5500 triple quadrupole mass spectrometer (AB Sciex, Foster City, CA, USA) using selected reaction monitoring (SRM) analysis in positive ionization mode. For this purpose nitrogen was used as collision gas. Using a 50 ms dwell time, SRM transitions were monitored as follows: cAMP +330/136 and +330/312, tenofovir +288/176 and +288/159. The transition +330/136 was the most intense transition of cAMP and therefore used for quantification. Additionally the +330/312 transition of cAMP was used as qualifier. The transition +288/176 of tenofovir was used as quantifier and the transition +288/159 as qualifier. The mass spectrometer parameters were as follows: ion source voltage: 4500 V, ion source temperature: 600uC, curtain gas: 30 psi and collision gas: 9 psi. cAMP in samples was quantified by applying the standard curve, obtained by analysis of known amounts of pure cAMP at: 0.0262, 0.066, 0.164, 0.41, 1.024, 2.56, 6.4, 16, 40, 100, 250 pmol/tube.

Miscellaneous (Data Analysis, Statistical Analysis and GTPase Assay)
Chromatograms, obtained by the HPLC-MS/MS analysis, were analyzed with the Analyst Software 1.5.1 (AB Sciex). Steady-state GTPase activity assay, using membrane preparations of Sf9 insect cells, expressing fusion protein b 2 AR-G saS , was performed as described previously [43]. Data from the O 2 N2 , cAMP and GTPase assays were analyzed with the Prism 5.01 software (GraphPad, San Diego, CA, USA). The means 6 S.E.M. were always determined by the analysis of at least three The efficacy (E max ) of ISO in each assay was set to 1.00 and the efficacies of other ligands were referred to this value. The pK B values for b 2 AR antagonists were calculated according to Cheng and Prusoff [44] using the following equation: pK B = -log(IC 50antagonist /(1+(c ISO /x))); IC 50antagonist -IC 50 value of an antagonist, determined in antagonist mode, c ISO -used concentration of ISO, 6-IC 50 (O 2 N2 assay) or EC 50 value (cAMP assay and GTPase assay) for ISO, determined in agonist mode (data from Table 1).
Data for receptor ligands were analyzed using one-way ANOVA, followed by Bonferroni's multiple comparison test, in order to compare pK B values of the examined b 2 AR antagonists between the O 2 N2 , cAMP and GTPase assay. Data for AC-and PKA inhibitors were analyzed using one-way ANOVA, followed by Dunnett's multiple comparison test. Statistical significance was defined as p , 0.05 (95% confidence interval).

Results
Characterization of the b 2 AR on Human Neutrophils with b 2 AR Agonists b 2 AR agonists with efficacies varying from very weak partial to full agonism [45] were examined. The effects of b 2 AR agonists were measured as inhibition of fMLP-stimulated O 2 N2 production ( Fig. 1) and as cAMP accumulation (Fig. 2). In both cases, the inter-experimental variability was high ( Figs. 1 and 2). Note that in Figs. 1C and 2B, data from different individuals are depicted. High inter-individual variability of human neutrophil function was observed previously [46]. However, when neutrophils from a given donor were analyzed on different days, data in the two test systems generally varied by less than 20% (data not shown). Thus, interindividual data variability is a much greater issue with neutrophils than intraindividual day-to-day variability. Accordingly, in order to allow comparison of results from experiments with different donors in this study, data were normalized with 1.00 representing the maximal effect of the b 2 AR agonist ISO and 0.00 representing the basal activity.
Potencies and efficacies of the examined b 2 AR agonists in the O 2 N2 and the cAMP assay are listed in Table 1 and concentrationresponse curves are depicted in Fig. 3. Additionally, the EC 50 and E max values of ligands determined in steady-state GTPase activity assays using membrane preparations of Sf9 insect cells expressing the b 2 AR-G saS fusion protein [43,45] are listed in Table 1. The b 2 AR-G saS fusion protein is artificial but has become a standard system for the pharmacological analysis of the b 2 AR [43,45]. For a detailed discussion on the advantages and disadvantages of the fusion protein technique as well as additional references relevant for this approach, the reader is referred to ref. 43. Potencies of ISO, ADR and SAL were higher in the O 2 N2 assay than in the cAMP assay. EPH and DCI were lacking agonistic activity in the cAMP assay at concentrations up to 100 mM, whereas inhibitory effects of both ligands on fMLP-stimulated O 2 N2 production were readily observed. The efficacy of ADR was comparable in both test systems, but the efficacy of SAL more than two times higher in the O 2 N2 assay relative to the cAMP assay. When the data from the recombinant test system were included in the comparison, the rank order of potency of ligands was cAMP assay,GTPase assay,O 2 N2 assay, and the rank order of efficacy was cAMP assay,O 2 N2 assay < GTPase assay (Fig. 4). In case of the two-state model postulating a single active state, we would have expected linear correlations for agonists with respect to efficacies and potencies, regardless of which parameters are compared. However, Fig. 4 shows that the correlations are, in general, rather poor, regardless of which comparisons are being made. The worst correlations in terms of deviation from the theoretically expected slope of 1.00 in case of identity of parameters were observed for the comparison of pEC 50 values in the GTPase and O 2 N2 assay (Fig. 4B) and pEC 50 values in the GTPase and cAMP assay (Fig. 4C). A limitation of our study is that we studied only a limited number of agonists, but an advantage is that the ligands cover a broad range of efficacies and potencies so that clustering of the data in one spot is avoided. In fact, this type of two-dimensional comparison of ligand potencies and efficacies has been repeatedly used to support the concept of ligand-specific receptor conformations in various test systems [24,27,32,40].
A trivial explanation for the differing effects of b 2 AR agonists in the O 2 N2 assay and cAMP assay could be that the agonists exhibit O 2 N2 scavenging properties on fMLP-stimulated O 2 N2 production. However, when O 2 N2 production in neutrophils was triggered with phorbol-12-myristate-13-acetate (activator of protein kinase C, 100 nM) instead of fMLP, the examined b 2 AR agonists had no effect on O 2 N2 production at all (data not shown).
Characterization of the b 2 AR on Human Neutrophils with b 2 AR Antagonists According to conventional models of GPCR activation, potency of an antagonist for a given receptor is constant irrespective of the tissue or recombinant system selected for the characterization, the agonist used for the stimulation of GPCR and downstream signaling event monitored [20][21][22]47]. However, by monitoring the cAMP accumulation and cAMP response element-mediated reporter gene transcription in Chinese hamster ovary (CHO) cells, different K B values were determined for b 2 AR antagonists [48]. These data indicate that antagonists, like agonists, may stabilize functionally distinct receptor conformations. Likewise, we obtained evidence for functional selectivity of antagonists at various recombinant histamine receptor subtypes [49]. Hence, the question arose whether parameter-dependent b 2 AR antagonist potency is also apparent in a native test system, namely in neutrophils.
The pK B values for ICI, MET, ALP, BIS and ATE were determined in the cAMP and O 2 N2 assay by applying a submaximally effective concentration of ISO and increasing concentrations of b 2 AR antagonists. Concentration-response curves for antagonists are shown in Fig. 5. In Table 2, the results are summarized and compared with antagonistic activity of the same ligands in the recombinant test system (GTPase assay using membranes of Sf9 insect cells expressing b 2 AR-G saS ). The statistical analysis of the obtained data revealed no difference of pK B values between the two different parameters in neutrophils, BIS being an exception. However, in comparison with data on neutrophils, the potencies of all antagonists were significantly reduced on the recombinantly expressed b 2 AR-G saS . Fig. 6 shows correlations of the pK B values of antagonists for the various parameters analyzed. It is evident that the correlations for antagonists are much better than the corresponding correlations for agonists shown in Fig. 5.
Do AC-and PKA-activation Interfere with fMLPstimulated O 2 N2 Production?
As already mentioned in the introduction, b 2 AR-signaling is very complex, depending on ligand and test system [15,16,18,26,27]. b 2 AR-mediated inhibition of fMLP-stimulated O 2 N2 production is thought to be due to cAMP production and PKA activation [50][51]. In order to address this question we studied the effects of three structurally distinct and well-established PKA inhibitors. Among the inhibitors are an isoquinoline sulfonamide [52], a cAMP analog [53] and a microbial metabolite [54]. However, in our experiments, we failed to obtain evidence for the hypothesis that activation of PKA after b 2 AR stimulation is a crucial event for inhibition of fMLP-stimulated O 2 N2 production (Fig. 7). Specifically, compounds H89 and KT5720, widely-used and effective cell-permeable competitive inhibitors of the ATPbinding to the ATP-binding pocket of the PKA in other test systems [55][56][57][58], did not reverse ISO-induced inhibition of fMLPstimulated O 2 N2 production (also when neutrophils were pretreated 15 min with H89 or KT5720). Even the cAMP antagonist Rp-cAMPS which competes with cAMP for the binding to the regulatory subunits of PKA [57,59], did not interfere with the effect of ISO of O 2 N2 production in human neutrophils (Fig. 7). Moreover, we tried to assess the involvement of the cAMP signaling pathway in the fMLP-stimulated O 2 N2 production by applying the AC inhibitor SQ 22536 [60]. Surprisingly, SQ 22536 enhanced rather than diminished the inhibitory effect of ISO on fMLP-induced O 2 N2 production (Fig. 8A). Additionally, SQ 22536 exhibited unexpected inhibitory effects on fMLP-induced O 2 N2 production on its own. SQ 22536 by itself did not increase cAMP levels in neutrophils, and the compound also did not inhibit the ISO-induced cAMP increase (Fig. 8B). Pleiotropic and AC- N2 assay, the cAMP assay and the GTPase assay. The data for comparison were taken from Table 1   On human neutrophil granulocytes, the O 2 N2 assay (1610 5 cells per well) and the cAMP assay (5610 5 cells per sample) were performed as described under Materials and Methods. Steady-state GTPase activity assay, using membrane preparations of Sf9 insect cells, expressing fusion protein b 2 AR-G saS , was performed as described in [43]. O 2 N2 and cAMP production as well as GTP hydrolysis were determined at submaximally effective concentration of ISO (100 nM in the O 2 N2 and cAMP assay, 10 nM in the GTPase assay) in the presence of increasing concentrations of b 2 AR antagonists. Data were analyzed by non-linear regression and were best fitted to sigmoidal concentration/response curves. Data shown are from four to five independent experiments performed in triplicate. The pK B values were calculated from the IC 50 values according to Cheng and Prusoff [44]. pK B values were compared with each other using one-way ANOVA, followed by Bonferroni's multiple comparison test (pK B significantly different to: *O 2 N2 assay, + cAMP assay; one symbol: p,0.05, two symbols: p,0.01, three symbols: p,0.001). Non-logarithmic K B values in mM are shown in parentheses. doi:10.1371/journal.pone.0064556.t002 independent effects of SQ 22536 have been observed repeatedly [61].
In order to provide proof of principle that an increase in intracellular cAMP concentration is capable of inhibiting fMLPstimulated O 2 N2 production, we examined the effects of cAMP and various cAMP analogs on NADPH oxidase activation. DB-cAMP is lipophilic and penetrates the plasma membrane. Inside the cell, DB-cAMP is converted to the biologically active 6-MB-cAMP [62]. In accordance with previous data [9], DB-cAMP reduced fMLP-stimulated O 2 N2 production, whereas the control compound sodium butyrate was ineffective (Fig. 9). In addition, the monobutyrylated control compound 6-MB-cAMP did not robustly inhibit NADPH oxidase, most likely due to inefficient membrane penetration. Sp-cAMPS is less lipophilic than DB-cAMP but does not require bioactivation [59]. Sp-cAMPS was similarly effective at inhibiting O 2 N2 production as DB-cAMP. These data show that cAMP does have the potential to inhibit NADPH oxidase. However, we also noted that very high concentrations of DB-cAMP are required to elicit inhibition, probably exceeding the intracellular cAMP concentrations achieved following b 2 AR stimulation.
cAMP itself also slightly inhibited O 2 N2 production. This could be due to extracellular degradation of cAMP to adenosine by phosphodiesterases and ectonucleotidases and subsequent activation of adenosine A2-receptors by adenosine [63]. Import of cAMP via multidrug resistance protein transporters (MRPs) into cells and subsequent PKA activation could be involved as well [64]. Likewise, the small inhibitory effects of the PKA inhibitor Rp-cAMPS could be due to adenosine liberation from the parent compound. We do not have a satisfactory explanation for the small but significant stimulatory effect of the mono-butyrylated control compound 29-O-MB-cAMP on O 2 N2 production. However, we confirmed that the compound per se did not activate O 2

N2
production or reduced ferricytochrome c independently of NADPH oxidase (data not shown). The activator of the cAMP effector protein Epac, 8-pCPT-29-O-Me-cAMP [65], did not inhibit fMLP-induced O 2 N2 production, arguing against an involvement of Epac in NADPH oxidase regulation. We also observed that the PKA inhibitors H89, KT5720 and Rp-cAMPS showed no reversing effect on the inhibition of O 2 N2 production caused by DB-cAMP (Fig. 7). These data raise questions whether a hitherto unidentified cAMP-binding protein is involved in the inhibition of O 2 N2 production by DB-cAMP and Sp-cAMPS.

Discussion
The two-state model of receptor activation implying an active (R*) and an inactive (R) state has now been superseded by a more complex model involving multiple active receptor conformations that lead to ligand-specific receptor activation, also referred to as functional selectivity or biased agonism [19][20][21][22][23][24][25][26][27][28][29][30][31][32][33]. Functional selectivity has been reported for numerous GPCRs such as dopamine D 1 and D 2 receptors, the histamine H 2 and H 4 receptor, adenosine A 1 and A 3 receptors, the a 2A -adrenoceptor and the b 2 AR [15][16][17][18][19][20][21][22][23][24][25][26][27][28][29][30][31][32][33]40,49]. So-called biased ligands can differently activate G protein-dependent and -independent signaling such as the b-arrestin pathway [27,32,34,35], can discriminate between G s , G i , G q and other G protein-mediated pathways [15,66] or even selectively modulate e.g. G i1 , G i2 and G i3 protein subtype activities [67]. Therefore, it is not surprising that any given ligand possesses multiple potencies and efficacies depending on the down-stream pathway analyzed [33]. This concept was also confirmed in our study with b 2 AR agonists in human neutrophils using the cAMP assay and the O 2 N2 assay as parameters and by comparison of the results with literature data obtained in recombinant test system (Fig. 4). If only a single active b 2 AR conformation existed, we would have expected linear correlations following the dotted lines in Fig. 4 between potencies and efficacies (relative to the reference compound ISO) of agonists, regardless of which parameters are considered. However, this was not the case. In accordance with our data, fluorescence studies with purified b 2 AR nuclear magnetic resonance studies provided evidence for ligand-specific conformations [31,37,39].
With respect to b 2 AR antagonists, effects were similar in the cAMP and O 2 N2 assay on neutrophils (Table 2 and Fig. 5), indicating that in neutrophils, functional selectivity is predominantly observed for b 2 AR agonists. In contrast to neutrophil parameters, potencies of antagonists were generally lower at the  (Table 2), and there were also ligand-specific differences. The trend towards lower antagonist (inverse agonist) potencies at the recombinant b 2 AR could be due to higher constitutive activity of the recombinant than of the native system [20].
The vast majority of reports about functional selectivity originate from studies with recombinant test systems or purified receptors (see, e.g., [31,[34][35][36][38][39]). On the contrary, functional selectivity in native test systems has been rarely studied so far, e.g. for the histamine H 4 receptor on isolated human eosinophils [68] and for the histamine H 2 receptor on isolated human eosinophils and neutrophils [40]. Here, we report on functional selectivity of the b 2 AR in human neutrophils. In accordance with our data, there is evidence for functional selectivity of b 2 AR ligands in cardiomyocytes [69]. In this system, stereoisomers of fenoterol differentially activate G i -and G s -proteins.
Unfortunately, in-depth analysis of functional selectivity in neutrophils is hampered by limited possibilities to block coupling of the b 2 AR to coupling partners. GPCR-G i protein coupling can be interrupted with pertussis toxin [70], whereas there is no pharmacological tool available for the selective inhibition of G s and G q coupling or the b-arrestin pathway in native test systems. Furthermore, since the formyl peptide receptor is coupled to G iproteins and an essential stimulatory component in the O 2 N2 assay, we could not use pertussis toxin to differentiate between b 2 AR ligands in this assay and in the cAMP assay. Moreover, in neutrophils, difficulties for effective genetic manipulation, interindividual variability and relatively short life time impede with more detailed analysis of functional selectivity in this native test system.
Our data obtained by measuring GTP hydrolysis in the recombinant test system reflect coupling of the b 2 AR solely to the short splice variant of the G s protein [43,45]. There is no doubt that the b 2 AR couples to G s in neutrophils, but no information is available regarding the involvement of specific G s splice variants. The question remains what the reason for the generally increased potency of the examined b 2 AR agonists in the O 2 N2 assay compared to the cAMP assay is. Lack of correlation between the cAMP accumulation and inhibition of fMLP-induced O 2 N2 production in neutrophils has been described also by other research groups using other stimuli [71][72][73][74]. Among other reasons, better coupling efficiency of the b 2 AR to the O 2 N2 pathway than to the cAMP pathway is of relevance for the divergence in data. Differences in strength of coupling to different signaling pathways have been reported for other receptors [75]. Additionally, we have no evidence for the involvement of Epac in the signal transduction pathway of the b 2 AR leading to inhibition of NADPH oxidase since an effective Epac activator failed to inhibit fMLP-stimulated O 2 N2 production (Fig. 9). We failed to support an involvement of AC, cAMP and PKA in b 2 AR-mediated inhibition of fMLP-stimulated O 2 N2 production. Thus, it appears that the two measured events (cAMP accumulation and O 2 N2 production) in neutrophils are independent of each other. The lack of effect of DCI and EPH on cAMP accumulation despite inhibitory effects of these ligands on O 2 N2 production supports the concept of cAMP-independent inhibition of NADPH oxidase. Moreover, SAL is more effective at inhibiting NADPH oxidase than at increasing cAMP. Our failure to detect stimulatory effects of DCI and EPH on cAMP accumulation and ineffective stimulation of cAMP accumulation by SAL are not due to cAMP degradation since we included a phosphodiesterase inhibitor into the cAMP assay. Moreover, we applied a highly sensitive and specific MS method to detect cAMP, avoiding notorious cross-reactivity problems of antibodies widely applied in cyclic nucleotide detection [41,42,76]. Thus, our study addressing ligand-specific receptor conformations also casted doubt about the dogma of cAMP-dependent inhibition of NADPH oxidase by the b 2 AR although, in principle, cAMP can inhibit O 2 N2 production ( Fig. 9). Even in case of inhibition of O 2 N2 production by cAMP analogs, we failed to obtain positive evidence for an involvement of PKA (Fig. 7).
Analysis of the signaling pathways responsible for inhibition of O 2 N2 production in neutrophils is hampered by unsuitability of  For comparison, we also studied the effect of the direct AC activator FSK (10 mM). In all tubes, a final concentration of 1% (v/v) DMSO (unavoidable for dissolving SQ 22536 and FSK) was present to achieve comparable results. SQ 22536 had no significant reversal effect on the inhibition caused by ISO in the O 2 N2 assay, and FSK had no significant inhibitory effect on fMLP stimulation in this assay. doi:10.1371/journal.pone.0064556.g008 experimental tools available. Most strikingly, the widely used AC inhibitor SQ 22536 failed to reduce the stimulatory effect of ISO on cAMP levels but further augmented the inhibitory effect of ISO on O 2 N2 production (Fig. 8). Non-specific and pleiotropic effects of SQ 22536 have been subject of a recent review [61]. Quite striking too was the lack of inhibitory effect of FSK on O 2 N2 production and lack of stimulatory effect of FSK on cAMP production (Fig. 8). These data could be explained by a model according to which the FSK-insensitive AC isoform 9 [61] is the functionally predominant AC in neutrophils. This hypothesis needs to be tested in future studies. Again, this is not a trivial task since the quality of AC antibodies is generally poor [61], and we are not aware of the availability of specific AC9 antibodies.
Stimulation of cAMP accumulation and reduction of O 2 N2 production mediated by ISO, was inhibited according to monophasic competition isotherms by ICI, a highly potent and selective b 2 AR antagonist with very low potency on the b 1adrenergic receptor [77]. In case of an exclusive involvement of the b 1 -adrenergic receptor we would have expected low potency of ICI, and in case of an involvement of both b-adrenergic receptors, we would have expected biphasic isotherms. This was clearly not the case (Fig. 5), and moreover, the potency of ICI at the native b 2 AR was even higher than at the recombinant b 2 AR (Fig. 5 and Table 2). Thus, the data obtained with ICI provide strong evidence for the notion that only the b 2 AR but not the b 1adrenergic receptor is functionally expressed in human neutrophils. Moreover, we excluded the possibility that ISO as representative b 2 AR agonist cross-reacts with the histamine H 2 receptor, which is also expressed on human neutrophils [40]. Specifically, the effect of ISO on fMLP-induced O 2 N2 production was not reduced by the histamine H 2 receptor antagonists famotidine, tiotidine and zolantidine (data not shown).
Stallaert and coworkers [78] demonstrated on HEK293S cells that b 2 AR-dependent impedance response to ISO is the result of activation of multiple signaling pathways, including G s and G i coupling, Gbc-dependent signaling, cAMP production, extracellular signal-regulated kinase (ERK) 1/2 activation as well as Ca 2+ mobilization. Therefore, when stimulation of the b 2 AR does not activate the PKA-dependent pathway in human neutrophils or this pathway does not interfere with the NADPH oxidase signaling, modulation of the e.g. ERK1/2-pathway could be the explanation for the inhibition of the NADPH oxidase. Interestingly, in other studies on human neutrophils, a correlation between activation of cAMP/PKA signaling pathway and inhibition of ERK phosphorylation was observed, resulting in reduced fMLP-induced O 2 N2 production [79,80]. Intriguingly, the b 2 AR was reported to activate ERK signaling pathway via interaction with G i , Src and/or arrestin proteins in addition to G s proteins in other systems [81]. As it is evident that b 2 AR-signalling strongly depends on the cell system used [16], there is a need to address the correlation between b 2 AR activation, ERK activation and fMLP-stimulated O 2 N2 production in future studies as well. This could provide an explanation for the observed biased effects of some examined b 2 AR ligands on neutrophils.
In preliminary studies we examined a number of pharmacological inhibitors to explore alternative signaling pathways of the b 2 AR; e.g. we tested the p38 inhibitor SB203580, the JNK inhibitors SP600125 and SP600123, the ERK inhibitor PD980598, the PI3 kinase inhibitor LY294002 and the protein kinase C inhibitor Goe 6978. Unfortunately, these compounds per se inhibited fMLP-stimulated O 2 N2 production (data not shown) so that separate effects of these compounds on ISO actions could not be properly dissected. An alternative approach will be the examination of the effects of b 2 AR ligands on protein phosphorylation in neutrophils, using screening approaches encompassing numerous protein kinases as starting point. Lastly, it will also be very important to explore the possibility that the new signaling Figure 9. Effects of cAMP, butyrate and cAMP analogs on fMLP-stimulated O 2 N2 production. O 2 N2 production in human neutrophil granulocytes (1610 5 cells per well) was monitored by measuring superoxide dismutase-inhibitable reduction of ferricytochrome c at 550 nm as described in Materials and Methods. Data shown (mean 6 S.E.M.) are from three to seven independent experiments, performed in triplicate. Data were analyzed for statistical significance relative to the control (-, set to 1.0) using one-way ANOVA, followed by Dunnett's multiple comparison test (***p,0.001). doi:10.1371/journal.pone.0064556.g009 molecules cyclic CMP (cCMP) and cyclic UMP (cUMP) [82] are involved in b 2 AR-mediated signal transduction.
Data from the literature suggest a feasible use of b 2 AR agonists as anti-inflammatory agents [10,83]. In order to extend the data on functional selectivity in native test systems, we compared the effect of b 2 AR ligands on cAMP accumulation and O 2 N2 production in human neutrophils. However, there are many native test systems described, which would afford an expansion of our knowledge about functional selectivity of b 2 AR ligands. One example could be the parallel measurement of cAMP concentration and contractility of cardiomyocytes or endothelial cells as functional parameter [16,69]. Moreover, a comparison of cAMP accumulation and ERK phosphorylation in e.g. mouse embryonic fibroblasts [84] could be used for screening of b 2 AR ligands. All in all, there is a need to assess biased signaling through the b 2 AR in a wide spectrum of native test systems in order to improve the desired therapeutic effect of developed compounds on the one and to minimize side effects on the other. Furthermore, b 2 AR agonists that have been used for many years in the therapy of humans, e.g. as bronchodilators in patients with asthma or chronic obstructive pulmonary disease, should be reassessed using various native test systems, as there is a potential to improve already existing therapies, particularly by minimizing unwanted effects.
As a general approach to study functional selectivity in native cells, it is necessary to construct a systematic data matrix in which multiple ligands are examined at multiple concentrations (enabling determination of precise potencies and efficacies) and for multiple parameters. Previous studies with native human cells may have largely overlooked functional selectivity at GPCRs because there was no appreciation of the necessity to generate such a systematic data collection in order to understand cell biology. It is evident that availability of human cells is an issue for comprehensive pharmacological studies. The human neutrophil, despite its inherent problems, i.e. short survival time after isolation, variable responsiveness and poor accessibility to genetic manipulation, is a suitable model system to test functional selectivity for several reasons. Specifically, neutrophils can be obtained in large quantities, express multiple receptors, display numerous cell functions that can be assessed quantitatively and are pathophysiologically relevant for inflammation.