β3-Adrenoceptor Control the Cystic Fibrosis Transmembrane Conductance Regulator through a cAMP/Protein Kinase A-independent Pathway*

In human cardiac myocytes, we have previously identified a functional β3-adrenoceptor in which stimulation reduces action potential duration. Surprisingly, in cardiac biopsies obtained from cystic fibrosis patients, β3-adrenoceptor agonists produced no effects on action potential duration. This result suggests the involvement of cystic fibrosis transmembrane conductance regulator (CFTR) chloride current in the electrophysiological effects of β3-adrenoceptor stimulation in non-cystic fibrosis tissues. We therefore investigated the control of CFTR activity by human β3-adrenoceptors in a recombinant system: A549 human cells were intranuclearly injected with plasmids encoding CFTR and β3-adrenoceptors. CFTR activity was functionally assayed using the 6-methoxy-N-(3-sulfopropyl)quinolinium fluorescent probe and the patch-clamp technique. Injection of CFTR-cDNA alone led to the expression of a functional CFTR protein activated by cAMP or cGMP. Co-expression of CFTR (but not of mutated ΔF508-CFTR) with high levels of β3-adrenoceptor produced an increased halide permeability under base-line conditions that was not further sensitive to cAMP or β3-adrenoceptor stimulation. Patch-clamp experiments confirmed that CFTR channels were permanently activated in cells co-expressing CFTR and a high level of β3-adrenoceptor. Permanent CFTR activation was not associated with elevated intracellular cAMP or cGMP levels. When the expression level of β3-adrenoceptor was lowered, CFTR was not activated under base-line conditions but became sensitive to β3-adrenoceptor stimulation (isoproterenol plus nadolol, SR 58611, or CGP 12177). This later effect was not prevented by protein kinase A inhibitors. Our results provide molecular evidence that CFTR but not mutated ΔF508-CFTR is regulated by β3-adrenoceptors expression through a protein kinase A-independent pathway.

␤ 3 -Adrenoceptors differ from ␤ 1 -and ␤ 2 -adrenoceptor subtypes by their molecular structure and pharmacological profile (for review see Ref. 1). The ␤ 3 -adrenoceptor gene contains two introns (2,3) leading to alternative splice isoforms, whereas ␤ 1and ␤ 2 -adrenoceptor genes are intronless. ␤ 3 -Adrenoceptors are G protein-coupled receptors that interact with either G s or G i proteins (4,5). Depending on the tissue, ␤ 3 -adrenoceptor stimulation produces functional effects that are either comparable with or opposite to those produced by ␤ 1 -and ␤ 2 -adrenoceptor stimulation. For instance, in adipose tissue, ␤ 3 -adrenoceptor stimulation increases lipolysis through an elevation in intracellular cAMP concentration (6,7) as does ␤ 1 -or ␤ 2 -adrenoceptor stimulation. In the human heart, we have previously demonstrated that ␤ 3 -adrenoceptors mediate negative inotropic effects (5) in stark contrast to the classical positive inotropic effects caused by ␤ 1 -and ␤ 2 -adrenoceptor stimulation. Negative inotropy as produced by ␤ 3 -adrenoceptors stimulation is unlikely to be related to stimulation of the cAMP pathway but rather to stimulation of the cGMP pathway (8) and is associated with an acceleration in the relaxation phase of the twitch and with a shortening of the action potential duration (5).
The present study issued from the observation that in myocardial samples from cystic fibrosis patients, ␤ 3 -adrenoceptor stimulation produced negative inotropic effects but remarkably did not shorten the action potential. Cystic fibrosis is a genetic disease caused by mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) 1 gene encoding an ATPbinding cassette protein (9) with anionic channel properties (10) and expressed in the human heart (11). Therefore, the simplest explanation accounting for our observation was that ␤ 3 -adrenoceptors in non-cystic fibrosis tissues control a repolarizing CFTR chloride conductance lacking in cystic fibrosis tissues. We confirmed this hypothesis using various techniques in a heterologous mammalian expression system.

EXPERIMENTAL PROCEDURES
Action Potential Recordings-Human ventricular biopsies were obtained from three cystic fibrosis (CF) patients homozygous for the ⌬F508 mutation who underwent cardiopulmonary transplantation and from five non-CF patients used as control. Heart and tissues were placed in a transport solution and conveyed rapidly to the laboratory. Preparations were dissected and placed in an experimental chamber. They were superfused at a flow rate of 5 ml/min with oxygenated (95% O 2 , 5% CO 2 ) Tyrode solution (37 Ϯ 0.5°C) composed as follows: 120 mM NaCl, 5 mM KCl, 2.7 mM CaCl 2 , 1.1 mM MgCl 2 , 0.33 mM NaH 2 PO 4 , 5 mM glucose, and 20 mM NaHCO 3 . Tissues were equilibrated for 60 min and then subjected to field stimulation at a pacing cycle length of 1,700 ms. Action potentials were recorded as described previously (12) using conventional 3 M KCl-filled microelectrodes coupled to an Ag-AgCl electrode connected to an amplifier (Biologic VF 102; Claix, France). The tissue chamber was grounded through an Ag-AgCl electrode. The action potentials were recorded on a digital tape recorder (Biologic DTR-1200) for off-line analysis.
Cell Cultures-The human lung epithelial-derived cell line A549, the African green monkey kidney-derived cell line COS-7, and the human colonic carcinoma cell line T84 were obtained from the American Type Culture Collection (Manassas, VA). A549 and COS-7 cells were cultured as previously reported (13). T84 cells were grown in a 1:1 mixture of Dulbecco's modified Eagle's medium and Ham's F-12 medium supplemented with 10% fetal calf serum, 2 mM L-glutamine, and antibiotics (100 IU/ml penicillin and 100 g/ml streptomycin; all from Life Technologies, Inc., Paisley, Scotland), maintained in a humidified incubator (95% air, 5% CO 2 ) at 37°C and passaged weekly.
Plasmids-Transgene cDNAs were subcloned into pECE or pcDNA 3 mammalian expression vectors under the control of a SV40 enhancer/ promoter or a cytomegalovirus promoter, respectively. pECE-CFTR (a kind gift from P. Fanen, INSERM, Créteil, France) and pcDNA 3 -CFTR (a gift from J. Ricardo, Lisbon, Portugal) plasmids encoded the wildtype CFTR protein. pcDNA 3 -⌬F508-CFTR plasmid (also from J. Ricardo) encoded the mutated ⌬F508-CFTR protein. Alternative splicing of the ␤ 3 -adrenoceptor mRNA could generate three isoforms with different C-terminal regions: (i) the A isoform is the shortest protein; (ii) the B isoform, which corresponds to the A isoform extended by 12 amino acids in the C-terminal region, is the only isoform expressed in rat and is supposedly not expressed in human; and (iii) the C isoform corresponds to the A isoform extended by 6 amino acids in the C-terminal region and is the prevalent isoform in humans (2,14). As the C-terminal region of the ␤ 3 -adrenoceptor is involved in the coupling with G proteins (1), plasmids encoding either for the A (pcDNA 3 -␤ 3 A) and C (pcDNA 3 -␤ 3 C) isoforms present in human tissues were generated and expressed in mammalian cells. For control experiments, we used a pECE plasmid lacking the insert and a pcDNA 3 -ROMK 2 plasmid encoding a renal outer medulla potassium channel (a gift from S. C. Hebert, Vanderbilt University, Nashville, TN). For microcytofluorometry and patch-clamp experiments, cells settled on glass coverslips (Nunclon; InterMed Nunc, Roskilde, Denmark) were microinjected with plasmids at day 1 or 2 after plating. Our protocol to intranuclearly microinject individual cells using the Eppendorf ECET microinjector 5246 system and the ECET micromanipulator 5171 system has been reported in detail elsewhere (13). Plasmids were diluted at various final concentrations (0.01-130 g/ml) in an injection buffer made of 50 mM NaOH, 40 mM NaCl, 50 mM HEPES, pH 7.4, with NaOH, supplemented with 0.5% fluorescein isothiocyanate-dextran to visualize successfully injected cells. The method we used for cell transfection with the 22-kDa polyethylenimine synthetic vector (a gift from J. P. Behr, CNRS, Illkirch, France) has been reported in detail elsewhere (15). Cells were seeded either on glass coverslips for microcytofluorometry experiments or in 6-well plates (Poly Labo, Strasbourg, France) for cAMP and cGMP assays. The area of coverslips and wells was identical. COS-7 cells were transfected at a polyethylenimine/DNA ratio of 2 equivalents with 4 g of plasmids/well. pcDNA 3 -CFTR was used at 2 g/well, pcDNA 3 -␤ 3 A at 1 g/well, and pcDNA 3 -␤ 3 C at 1 g/well. To identify successfully transfected cells, we also added a plasmid encoding a green fluorescent protein (pTR-UF 2 -green fluorescent protein; a gift from P. Lory, CNRS, Montpellier, France) to obtain a final amount of 4 g of plasmid/well. SPQ Fluorescence Assay-The Cl Ϫ channel activity of CFTR was assessed using the halide-sensitive fluorescent probe SPQ (Molecular Probes, Eugene, OR) as described previously (13). 24 h post-injection or transfection, cells placed on glass coverslips were loaded with SPQ using a hypotonic shock procedure. The coverslip was mounted on the stage of an inverted microscope (Diaphot; Nikon, Japan) equipped for fluorescence and illuminated at 360 nm. The emitted light was collected at 456 Ϯ 33 nm by a high resolution image intensifier coupled to a video camera (Extended ISIS camera system; Photonic Science, Robertsbridge, United Kingdom) connected to a digital image processing board controlled by FLUO software (Imstar, France). Cells were maintained at 37°C and continuously superfused with the extracellular solution containing: 145 mM NaCl, 4 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 5 mM HEPES, 5 mM glucose, pH 7.4, with NaOH. A microperfusion system allowed local application and rapid change of the different experimental mediums. I Ϫ and NO 3 Ϫ mediums were identical to the extracellular solution except that I Ϫ and NO 3 Ϫ replaced Cl Ϫ as the dominant extracellular anion. I Ϫ and NO 3 Ϫ mediums also contained 10 M bumetamide (Sigma) to inhibit the Cl Ϫ /Na ϩ /K ϩ co-transporter. To standardize fluorescence intensity, the initial fluorescence level in the presence of I Ϫ was taken as zero. The membrane permeability to halides (p) was determined as the rate of SPQ dequenching upon perfusion with NO 3 Ϫ medium (see Fig. 2A).
Current Recordings-Whole cell currents were recorded with the ruptured patch configuration of the patch-clamp technique as described previously (16). Cells were placed on the stage of an inverted microscope and bathed at 37°C in the same extracellular solution as used in SPQ experiments. Patch pipettes with a tip resistance of 2.5-5 M⍀ were electrically connected to a patch-clamp amplifier (Axopatch 200A; Axon Instruments, Foster City, CA). Stimulation, data recording and analysis were performed by Acquis I software made by Gérard Sadoc (distributed by Biologic) through an analog-to-digital converter (Tecmar TM100 Labmaster; Scientific Solution, Solon, OH). The pipette solution contained 74.5 mM CsCl, 70.5 mM aspartic acid, 5 mM HEPES, 1 mM BAPTA, 5 mM MgATP, pH 7.2, with CsOH. During Cl Ϫ current recording, the cell was locally perfused with a Cl Ϫ free solution containing 149 mM CsCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 5 mM glucose, 5 mM HEPES, pH 7.4, with CsOH. Two stimulation protocols were used: voltage ramps applied at a frequency of 0.2 Hz from Ϫ80 to ϩ60 mV (depolarization rate, 46.7 mV/s; holding potential, Ϫ60 mV) and voltage steps imposed for 500 ms every 2 s from Ϫ60 mV to various potentials between Ϫ100 and ϩ60 mV.
cAMP and cGMP Assays-Determination of intracellular cAMP and cGMP contents was performed 24 h after transfection. COS-7 cells were incubated in the extracellular solution supplemented with 100 M 3-isobutyl-1-methylxanthine (Sigma) for 1 h at 37°C. For the cAMP assay, cells were incubated at 37°C with the same medium containing either 0 or 10 M forskolin (Sigma) for 5 min. For the cGMP assay, cells were incubated in the presence or absence of 500 M cGMP analog for 20 min. At the end of the incubation period, cells were lysed by three cycles of freezing and thawing. All samples were subsequently boiled at 90°C for 5 min and then centrifuged at 12,000 ϫ g for 15 min at 4°C. The supernatants were assayed for cAMP or cGMP by using a cAMP or cGMP enzyme immunoassay kit (Cayman Chemical Company, Ann Arbor, MI). Data are the means Ϯ S.E. of duplicate assays and normalized to total cell protein content determined by the method of Lowry et al. (17).
Statistics-Data are expressed as the means Ϯ S.E. of n number of experiments. Statistical significance of the observed effects was assessed by a Student's t test.

␤ 3 -Adrenergic Stimulation Does Not Shorten Action Potential in CF Cardiac
Myocytes-In control human endomyocardial tissues from non-CF patients ( Fig. 1), the specific ␤ 3 -adrenoceptor agonist BRL 37344 (0.3 M) reduced the action potential duration by Ϫ13.8 Ϯ 3.8% (n ϭ 5; p Ͻ 0.01) and also slightly reduced the action potential amplitude (Ϫ3.7 Ϯ 0.4%; p Ͻ 0.05). These effects were not observed in eight myocardial preparations obtained from ⌬F508/⌬F508 CF patients ( Fig. 1), suggesting that the electrophysiological effects of ␤ 3 -adrenoceptor stimulation in non-CF tissues were mediated by activation of a chloride repolarizing current flowing through CFTR channels that are not functional in CF cardiac muscle.
Activation of CFTR by Recombinant ␤ 3 -Adrenoceptors-CFTR channels are known to activate under intracellular cAMP (18,19) or cGMP elevation (20,21). The sensitivity of recombinant CFTR protein to intracellular cAMP and cGMP was investigated in A549 cells intranuclearly injected with a CFTR encoding plasmid (100 g/ml) and monitored with the SPQ fluorescence assay. A549 cells were chosen because this cell line was previously demonstrated to lack endogenous CFTR protein (22). Under base-line conditions, cells injected with plasmids alone or containing the CFTR expression cassette exhibited a low permeability to halide ( Fig. 2A and Fig. 3, left panel). Cells expressing CFTR but not cells injected with the plasmid lacking the insert displayed a Ϸ6-fold increase in the rate of SPQ dequenching upon application of the cAMPstimulating mixture ( Fig. 2A and Fig. 3, left panel). Similarly, in cells expressing CFTR, the rate of SPQ dequenching was increased approximately 3-fold by preincubation with 500 M CPT-cGMP (Fig. 2, B and C). As shown in Fig. 2C, the effect of cGMP stimulation was partially reversible upon CPT-cGMP washout. From this first set of experiments, we concluded that recombinant CFTR proteins produced by intranuclear injection of CFTR plasmid were sensitive to stimulation through both the cAMP-and cGMP-dependent pathways.
In a second set of experiments, various cell lines co-expressing either recombinant or endogenous CFTR and recombinant ␤ 3 -adrenoceptors were investigated. In A549 cells injected with ␤ 3 -adrenoceptor isoform A alone, the base-line membrane permeability to halide was not different from cells injected with the plasmid lacking the insert (Fig. 3, left panel) either in the absence or presence of cAMP. Surprisingly, cells co-injected with CFTR plus the A or the C isoform of ␤ 3 -adrenoceptor exhibited an 8 -10-fold increase in p base line that was not further increased upon cAMP stimulation (Fig. 3, left panel). To ensure that this effect was related to ␤ 3 -adrenoceptor expression, cells were co-injected with CFTR cDNA and with a cDNA encoding a K ϩ channel (ROMK 2 ). In these cells, p base line was similar to that in control cells and markedly increased in response to cAMP elevation. In cells expressing the mutated ⌬F508-CFTR protein, there was no increase in p base line related to co-expression with ␤ 3 -adrenoceptor A or C isoforms. As expected, ⌬F508-CFTR proteins were not sensitive to cAMP even in the presence of the ␤ 3 -adrenoceptor (Fig. 3, left panel). Similar results were also obtained in COS-7 cells injected (not illustrated) or transfected (Fig. 3, middle panel) with CFTR and ␤ 3 -adrenoceptor plasmids. In transfected COS-7 cells, p base line increased about 3-fold in cells co-expressing CFTR and either the A or C isoform of ␤ 3 -adrenoceptor. This effect was not observed in cells expressing CFTR or ␤ 3 -adrenoceptor alone. In COS-7 cells cotransfected with CFTR and the A or C isoform of ␤ 3 -adrenoceptor, cAMP stimulation produced a small albeit significant increase in halide permeability (Fig. 3, middle panel). T84 cells endogenously express the CFTR protein (23). Again, T84 cells injected with the A isoform of ␤ 3 -adrenoceptor cDNA exhibited a 2-fold increase in p base line as compared with noninjected cells (Fig. 3, right panel). This set of experiments shows that recombinant ␤ 3 -adrenoceptor activates both endogenous and recombinant CFTR irrespectively to the transfection method used.
Patch-clamp experiments were performed in cells co-expressing ␤ 3 -adrenoceptor and CFTR to further characterize the halide conductance responsible for the increased membrane permeability. A549 cells co-expressing CFTR and the A isoform of ␤ 3 -adrenoceptor displayed a high amplitude time-independent Cl Ϫ current in the absence of cAMP stimulation (Fig. 4). This Cl Ϫ current possessed characteristics reminiscent to the CFTR Cl Ϫ current recorded under cAMP stimulation in cells expressing CFTR alone. On average, in the absence of cAMP stimulation, the current amplitude at ϩ60 mV was 25.2 Ϯ 5.4 pA/pF (n ϭ 12) in cells injected with CFTR plus ␤ 3 -adrenoceptor but Variable Expression Levels of CFTR and ␤ 3 -Adrenoceptors-To modulate the effects of ␤ 3 -adrenoceptor on CFTR activity, the levels of expression of CFTR and ␤ 3 -adrenoceptor were varied independently. In A549 cells, varying the level of CFTR expression in the absence of ␤ 3 -adrenoceptor expression did not modify the base-line permeability to halide (Fig. 5, left  panel). In contrast, in the presence of a constant expression level of ␤ 3 -adrenoceptor isoform A (100 g/ml), p base line increased as the CFTR plasmid concentration in the injected medium increased from 3 to 100 g/ml (Fig. 5, right panel).
In another set of experiments, we varied the injected concen-tration of ␤ 3 -adrenoceptor isoform A plasmid from 0.03 to 3 g/ml in the presence of a constant CFTR concentration (30 g/ml). In these cells, ␤ 3 -adrenoceptors were selectively activated using 10 M isoproterenol, a nonselective ␤-adrenoceptor agonist, in the presence of 10 M nadolol, a ␤ 1 -and ␤ 2 -adrenoceptor antagonist (24). It is noteworthy that cells expressing CFTR alone were not responsive to isoproterenol in the presence of nadolol (data not shown), suggesting that A549 cells lack endogenous ␤ 3 -adrenoceptors. As the expression level of ␤ 3 -adrenoceptor isoform A was increased, p base line increased, the effects of ␤ 3 -adrenoceptor stimulation with isoproterenol plus nadolol were progressively reduced, and the effects of the cAMP-activating mixture were also reduced (Fig. 6). So, at a low injected ␤ 3 -adrenoceptor cDNA concentration (i.e. 0.1 g/ ml), the base-line halide permeability was not different from control cells but increased 3-4-fold upon ␤ 3 -adrenoceptor stim- , mutated ⌬F508-CFTR cDNA (⌬F508-CFTR), plasmid lacking the insert (mock), the A or C isoform of ␤ 3 -adrenoceptor cDNAs (␤ 3 A and ␤ 3 C), and K ϩ channel cDNA (ROMK 2 ). For cDNA injection experiments, each plasmid was used at 100 g/ml, except the plasmid lacking the insert (mock; 130 g/ml). For transfection, CFTR cDNA was used at 2 g/coverslip and ␤ 3 -adrenoceptor cDNAs at 1 g/coverslip. A green fluorescent protein plasmid was also added so as to obtain a final amount of 4 g of plasmid/coverslip. ulation. Inversely, at the highest injected ␤ 3 -adrenoceptor cDNA concentration (i.e. 3 g/ml), the increase in p base line was such that the effects of ␤ 3 -adrenoceptor stimulation or direct cAMP elevation were abolished.
To further assess the CFTR regulation by ␤ 3 -adrenoceptor stimulation, we tested a full ␤ 3 -adrenoceptor agonist, SR 58611, and a partial ␤ 3 -adrenoceptor agonist that possesses also ␤ 1 -and ␤ 2 -adrenoceptor antagonistic properties, CGP 12177 (25). In cells co-expressing CFTR (30 g/ml) and a low level of ␤ 3 -adrenoceptor isoform A (0.1 g/ml), a 15-min application of SR 58611 (1 M) or CGP 12177 (1 M) increased p values 2-or 3-fold, respectively (Fig. 7, middle and right panels). This effect was not observed in cells expressing CFTR alone. We concluded that the effects produced by ␤ 3 -adrenoceptor expression on the CFTR protein depend on the expression level of the ␤ 3 -adrenoceptors: (i) at a low level of ␤ 3 -adrenoceptor expression, the CFTR channels remained in the close state and were sensitive to ␤ 3 -adrenoceptor pharmacological stimulation and (ii) at a high level of ␤ 3 -adrenoceptor expression, CFTR channels were permanently activated and lost their sen-sitivity to either ␤ 3 -adrenoceptor agonists or direct cAMP stimulation.
Lack of PKA Involvement in CFTR Activation by ␤ 3 -Adrenoreceptors-In an attempt to determine whether the PKA activation was involved in CFTR regulation by ␤ 3 -adrenoceptor stimulation, A549 cells were incubated for 20 min with a mixture of two PKA inhibitors, Rp-8-Br-cAMPS (100 M) and Rp-8-CPT-cAMPS (100 M). Rp-8-Br-cAMPS and Rp-8-CPT-cAMPS are specific for PKA type I and II, respectively (26). PKA-dependent CFTR stimulation by isoproterenol (for review see Ref. 27) was used as a control experiment to check efficient PKA inhibition under our experimental conditions. In A549 cells injected with CFTR cDNA alone, isoproterenol, through a ␤ 1 -and/or ␤ 2 -adrenoceptor stimulation, increased the p value 3-fold (Fig. 7, left panel). After incubation with PKA inhibitors, this effect was abolished (Fig. 7, left panel). In contrast, in cells co-expressing CFTR and ␤ 3 -adrenoceptor, pre-treatment with PKA inhibitors had no effects on CFTR activation by ␤ 3 -adrenoceptor agonists (SR 58611 or CGP 12177; Fig. 7, middle and  right panels). These results demonstrated that the regulation of CFTR conductance by ␤ 3 -adrenoceptor stimulation was independent from the PKA pathway.
In another set of experiments, we determined whether the basal activation of CFTR by a high level ␤ 3 -adrenoceptor expression was dependent on an increase in intracellular cyclic nucleotides. To test this hypothesis, intracellular cAMP and cGMP levels were measured in COS-7 cells co-transfected with CFTR and ␤ 3 -adrenoceptor isoform A. COS-7 cells were used for these experiments because transfection of A549 cells using various synthetic vectors led to a low number of cells expressing the transgene as assessed with a green fluorescent protein reporter. COS-7 cells were transfected under the same experimental conditions as used for functional CFTR tests (Fig. 3,  middle panel). 24 h post-transfection, intracellular cAMP levels were determined under base line and after 5 min-incubation with 10 M forskolin. As illustrated in Fig. 8A, base-line cAMP levels were not significantly modified by expression of either the A or C isoforms of ␤ 3 -adrenoceptor in cells expressing CFTR or not. As expected, pre-incubation with forskolin induced an elevation in cAMP level in every experimental situation. The  Fig. 8B). In cells expressing the CFTR protein and either the A or C isoforms of ␤ 3 -adrenoceptor, base-line cGMP concentration was not different from that of control cells. Comparison between results shown on Fig. 3 (middle panel) and Fig. 8, which were both obtained under the same transfection conditions, shows that ␤ 3 -adrenoceptor-mediated CFTR activation was not related to the activation of either the cAMP or cGMP pathways. DISCUSSION The present study suggests that ␤ 3 -adrenoceptors are functionally coupled to the CFTR protein. Interestingly, coupling between ␤ 3 -adrenoceptors and CFTR depends on the expression level of ␤ 3 -adrenoceptors: (i) expression of a low level of ␤ 3 -adrenoceptors induces CFTR activation in response to ␤ 3adrenoceptor agonists and (ii) expression of a high level of ␤ 3 -adrenoceptors induces permanent CFTR activation independently from ␤ 3 -adrenoceptor stimulation. In addition, we also show that functional coupling between CFTR and ␤ 3adrenoceptors occurs irrespectively to the ␤ 3 -adrenoceptor splice variants expressed in human tissues. Finally, our results show that regulation of CFTR by ␤ 3 -adrenoceptors does not imply activation of the cAMP/PKA pathway.
We have previously reported that intranuclear injection of large quantities of CFTR cDNA into mammalian cells produces hyper-expression of CFTR proteins with altered physiological properties inasmuch as hyper-expressed recombinant CFTR channels are permanently opened and not susceptible to cAMP stimulation (13). This phenomenon, which we attributed to CFTR clustering within the cell membrane, appears for various CFTR cDNA concentrations depending on the cell line. Accordingly, it may be argued that the increase in base-line permeability that we observed in cells co-expressing CFTR and ␤ 3adrenoceptors was caused by a comparable phenomenon. We judge this explanation unlikely for several reasons: (i) in A549 cells, permanent activation of hyper-expressed CFTR protein usually occurs for plasmid concentrations greater than 350 g/ml (13), whereas in the present study, the injected concentration never exceeded 100 g/ml, a concentration that did not lead to permanently activated CFTR channels in the absence of ␤ 3 -adrenoceptor co-expression; (ii) base-line activation of CFTR channels by recombinant ␤ 3 -adrenoceptors was observed in COS-7 cells co-transfected with CFTR and ␤ 3 -adrenoceptors, although permanent activation of CFTR channels was never observed in cells transfected with CFTR alone irrespectively to the quantity of cDNA used for cell transfection; thus, base-line activation of CFTR in the presence of ␤ 3 -adrenoceptors was unlikely to be caused by our intranuclear injection technique; (iii) furthermore, activation of CFTR at base line was not observed when ␤ 3 -adrenoceptor cDNA was replaced by a K ϩ channel cDNA inserted into the same plasmid and injected at the same concentration as ␤ 3 -adrenoceptor cDNA; and (iv) finally and most importantly, CFTR endogenously expressed in T84 cells was also susceptible to activation by ␤ 3 -adrenoceptors, suggesting that agonist-independent activation of CFTR can be observed for level of CFTR expression close to physiological ones.
Our results show that the effects of ␤ 3 -adrenoceptor expression on CFTR activity can be modulated by the density of the receptors within the cell membrane. Increasing the expression of ␤ 3 -adrenoceptor in the presence of a constant CFTR expression level dose-dependently increased base-line permeability and also concomitantly reduced the activating effects of ␤ 3agonists. The ␤ 3 -adrenoceptor belongs to the superfamily of G protein-coupled receptors (28). These receptors are known to exist in the cell membrane in two subpopulations: (i) an inactive subpopulation that requires agonist occupancy for coupling to G protein and (ii) a constitutively active subpopulation that can couple to G protein in the absence of agonist (29 -31). An increase in the constitutively active subpopulation has been reported for mutated G protein-coupled receptors in which mutations induce a conformational change in the receptor that normally requires the binding of an agonist to occur (29,32). Similarly, an increase in the constitutively active receptor subpopulation has been reported for receptors overexpressed at a high level. For example, overexpression of the wild-type ␤ 2adrenoceptor either in Chinese hamster ovary cells or in myocardial cells of transgenic mice produced an elevation of baseline adenylyl cyclase activity (29,30,33). Comparable behavior has been reported with the ␤ 3 -adrenoceptor so that basal adenylyl cyclase activity was raised with ␤ 3 -adrenoceptor density in Chinese hamster ovary cells (34). In the present study, when the expression level of ␤ 3 -adrenoceptors was high, CFTR was activated in the absence of ␤ 3 -adrenoceptor stimulation. In such conditions, intracellular cAMP and cGMP levels were close to normal and in any case much lower than the level required for CFTR activation as shown in Fig. 3 (middle panel). It could be hypothesized that overexpression of ␤ 3 -adrenoceptors led to a greater number of receptors in the active state, resulting in saturation in the ␤ 3 -adrenoceptor signaling capacity in the absence of agonist. Inversely, when the expression of ␤ 3 -adrenoceptors was lowered, CFTR was not activated under base-line conditions and became sensitive to agonists. Activation of CFTR by ␤ 3 -adrenoceptor agonists was not sensitive to PKA inhibitors, ruling out the involvement of the cAMP/PKA pathway. Activation of CFTR independently of the cAMP/PKA pathway has previously been reported for other receptors such as P 2x -subtype of purinergic receptors (35) or mu-opioid receptors (36). Our results suggest that either another second messenger is implied in the coupling pathway between ␤ 3 -adreno- FIG. 8. Intracellular cAMP (A) and cGMP (B) concentrations in COS-7 cells expressing CFTR and ␤ 3 -adrenoceptor isoform A or C. COS-7 cells were transfected with various mixtures of plasmids as indicated on the abscissa. CFTR cDNA (CFTR ϩ) was used at 2 g/well, the plasmid lacking insert (mock) at 2 g/well, A or C isoforms of ␤ 3 -adrenoceptor cDNA (␤ 3 A and ␤ 3 C) at 1 g/well. Green fluorescent protein cDNA was also used to obtain a final amount of 4 g of plasmid/ well. The transfection conditions were the same as those used for functional assay of CFTR activity illustrated in Fig. 3 (middle panel). ceptor and CFTR or a direct interaction exists between the membrane receptor (or an associated G protein) and CFTR. Although direct regulation of CFTR channel protein by G proteins has not yet been reported, it has been shown that G ␣i2 protein modulates its vesicle trafficking and its delivery to the plasma membrane (37). As yet, two coupling pathways have been ascribed to ␤ 3 -adrenoceptors: (i) in adipose tissue and gastrointestinal tract, their stimulation produces an increase in cAMP levels (6,7) and (ii) in human ventricle, the negative inotropic effects mediated by ␤ 3 -adrenoceptor agonists are associated with an increase in NO production and cGMP levels (8). Clearly, the involvement of another mechanism leading to activation of CFTR by ␤ 3 -adrenoceptors needs clarification.
The physiological relevance of our findings should be found in the various organs where ␤ 3 -adrenoceptors and CFTR are co-expressed. In the human heart, ␤ 3 -adrenoceptor agonists produce a negative inotropic effect and a shortening in the action potential duration (5). Our results suggest that the action potential shortening produced by ␤ 3 -adrenoceptor agonists is caused by the activation of a CFTR-related repolarizing chloride current that is nonfunctional in CF patients. ␤ 3 -Adrenoceptors and CFTR are also endogenously co-expressed in other tissues such as airways (38,39) and gallbladder (40) epithelia. In these tissues, ␤ 3 -adrenoceptors may modulate water and salt secretion through apical CFTR channels.