Negative cooperativity across β1-adrenoceptor homodimers provides insights into the nature of the secondary low-affinity CGP 12177 β1-adrenoceptor binding conformation

At the β1-adrenoceptor, CGP 12177 potently antagonizes agonist responses at the primary high-affinity catecholamine conformation while also exerting agonist effects of its own through a secondary low-affinity conformation. A recent mutagenesis study identified transmembrane region (TM)4 of the β1-adrenoceptor as key for this low-affinity conformation. Others suggested that TM4 has a role in β1-adrenoceptor oligomerization. Here, assessment of the dissociation rate of a fluorescent analog of CGP 12177 [bordifluoropyrromethane-tetramethylrhodamine-(±)CGP 12177 (BODIPY-TMR-CGP)] at the human β1-adrenoceptor expressed in Chinese hamster ovary cells revealed negative cooperative interactions between 2 distinct β1-adrenoceptor conformations. The dissociation rate of 3 nM BODIPY-TMR-CGP was 0.09 ± 0.01 min−1 in the absence of competitor ligands, and this was enhanced 2.2- and 2.1-fold in the presence of 1 µM CGP 12177 and 1 µM propranolol, respectively. These effects on the BODIPY-TMR-CGP dissociation rate were markedly enhanced in β1-adrenoceptor homodimers constrained by bimolecular fluorescence complementation (9.8- and 9.9-fold for 1 µM CGP 12177 and 1 µM propranolol, respectively) and abolished in β1-adrenoceptors containing TM4 mutations vital for the second conformation pharmacology. This study suggests that negative cooperativity across a β1-adrenoceptor homodimer may be responsible for generating the low-affinity pharmacology of the secondary β1-adrenoceptor conformation.—Gherbi, K., May, L. T., Baker, J. G., Briddon, S. J., Hill, S. J. Negative cooperativity across β1-adrenoceptor homodimers provides insights into the nature of the secondary low-affinity CGP 12177 β1-adrenoceptor binding conformation.

According to classic receptor theory, the EC 50 of a partial agonist is expected to be similar to its binding affinity. The discrepancy between these 2 values for CGP 12177 and also other b-adrenoceptor ligands, such as pindolol, led to the classification of nonconventional b-adrenoceptor agonists (16). Furthermore, the CGP 12177agonisteffectappeared resistant to b-blocker antagonism at concentrations normally sufficient to block catecholamine-mediated agonist effects (4,6,12,14,17). As such, the affinities of a range of badrenoceptor antagonistshavebeenreportedtobe$1order of magnitude lower when inhibiting responses mediated by CGP 12177 at the b 1 -adrenoceptor compared with those mediated by catecholamines in both animal and human tissue preparations (10,14,18). It is noteworthy that the use of recombinant cell systems and cardiac tissue isolated from b 2and b 1 -/b 2 -adrenoceptor knockout mice clearly showed that the b 1 -adrenoceptor alone was responsible for the observed CGP 12177 pharmacology (1,3,4). This led to the proposal that there are 2 active conformations of the b 1 -adrenoceptor: a primary high-affinity endogenous catecholamine site and asecondarylow-affinity CGP 12177 site (1,4).
A b 1 -adrenoceptor homodimer complex would possess 2 structurally identical orthosteric b 1 -adrenoceptor sites, to which ligands would be expected to bind with similar affinities. However, negative cooperative interactions between the 2 orthosteric b 1 -adrenoceptor binding sites may provide an explanation of the lower affinity observed for the secondary b 1 -adrenoceptor protomer, if indeed this occurs as a dimer (23). Negative cooperativity across a homodimer interface has previously been described for the human A 3 adenosine receptor (23). In this example, negative cooperativity was demonstrated in single living cells by following the impact of orthosteric unlabeled ligands binding to one protomer of an A 3 -homodimer on the dissociation of a fluorescently labeled agonist (which was enhanced) from the orthosteric site of the other A 3 -receptor protomer (23).
We previously showed that the fluorescent CGP 12177 analog bordifluoropyrromethane-tetramethylrhodamine- (6) CGP 12177 (BODIPY-TMR-CGP) can be used to label both conformations of the b 1 -adrenoceptor (24). In this study, we used this fluorescent CGP 12177 analog to investigate the potential for allosteric interactions across a homodimer interface of the b 1 -adrenoceptor using kinetic measurements of BODIPY-TMR-CGP binding in single living cells.

Materials
Cell culture plastics were purchased from Thermo Fisher Scientific (Loughborough, United Kingdom), and cell culture reagents were from Sigma-Aldrich (Gillingham, United Kingdom) except for fetal calf serum, which was obtained from PAA Laboratories (Pasching, Austria). Lipofectamine 2000 transfection reagent and Opti-MEM medium were from Invitrogen (Paisley, United Kingdom), and SNAP-Surface 488 was from New England Biolabs (Ipswich, MA, USA). BODIPY-TMR-CGP was from Molecular Probes (Leiden, The Netherlands), and unlabeled CGP 12177 and propranolol were from Tocris Cookson (Avonmouth, Bristol, United Kingdom). All other reagents were from Sigma Chemicals (Poole, United Kingdom).

Generation of b 1 -adrenoceptor constructs
The b 1 -yellow fluorescent protein (YFP) N and b 1 -YFP C receptor constructs were generated by fusing either the N-terminal fragment of YFP (YFP N; amino acids 1-155) or the C-terminal fragment of YFP (YFP C ;aminoacids156-239) to the C-terminal end of the full-length wild-type human b 1 -adrenoceptor. The SNAP-b 1 construct was generated by fusing the SNAP-tag (New England Biolabs, Ipswich, MA, USA) to the N-terminal end of the wild-type human b 1 -adrenoceptor. The D138A mutation (7) was introduced into the b 1 -YFP C and the SNAP-b 1 sequence using the QuikChange sitedirected mutagenesis kit (Agilent Technologies, Cheshire, United Kingdom). All sequences were confirmed by DNA sequencing. All receptor constructs were subcloned into pcDNA3.1 vectors.

SNAP-tag labeling and confocal imaging
Confocal microscopy was performed using a Zeiss LSM710 laser scanning microscope with a 340 1.3 NA oil immersion lens. CHO-K1 cells were grown to 70% confluence in 8-well Labtek borosilicate chambered-cover glasses (Nalgene Nunc International, Fisher Scientific) and transiently transfected with SNAP-b 1 or SNAPb 1D138A recombinant DNA using Lipofectamine 2000 and Opti-MEM medium according to the manufacturer'sinstructions.The next day, a 1 mM concentration of the benzyl-guanine labeled SNAP-tag substrate SNAP-Surface 488 (BG-488) was prepared in fresh cell culture medium, added to these cells, and incubated in the dark for 30 min (room temperature). The cells then were washed twice in imaging buffer (147 mM NaCl, 24 mM KCl, 1.3 mM CaCl 2 , 1 mM MgSO 4 , 10 mM 4-(2-hydroxyethyl)-1piperazineethanesulfonic acid, 2 mM sodium pyruvate, 1.43 mM NaHCO 3 4.5 mM D-glucose, pH 7.4). After this, 3 nM BODIPY-TMR-CGP was added to the cells in the dark for 10 min (room temperature), after which the cells were imaged immediately (1024 3 1024 pixels, averaging at 4 frames, pinhole diameter 1 airy unit); 561 nm diode and 488 nm argon lasers were used to excite BODIPY-TMR-CGP and BG-488, respectively. A variable spectral detection system was used to capture emission at 565-605 and 495-535 nm, respectively. Confocal settings for laser power, offset, and gain were kept constant throughout each experiment set.
BODIPY-TMR-CGP kinetic studies using the confocal perfusion system

BODIPY-TMR-CGP association and dissociation kinetics
Live cell fluorescence imaging was performed on the Zeiss LSM510 laser scanning confocal microscope using a Zeiss Plan-Neofluar 340 1.3 NA oil-immersion objective in conjunction with a temperature-controlled (37°C) perfusion system to allow the visualization and quantification of BODIPY-TMR-CGP dissociation kinetics under infinite dilution (ID) conditions in single living cells (25). Kinetic experiments were performed as described by May et al. (25). In brief, cells were grown to near confluence on 32 mm glass coverslips in 6-well plates 1 d prior to experimentation. On the day of experimentation, the coverslip was placed into a tightly closed imaging chamber on a heated (37°C) microscope stage, where it was connected to tubes to facilitate the flow of imaging buffer through the imaging chamber in the absence and presence of BODIPY-TMR-CGP and/or unlabeled ligands at a constant flow rate of $4ml/min. For association and dissociation kinetic experiments of 10-100 nM BODIPY-TMR-CGP in CHO-b 1 and CHO-CS cells, the cells were exposed to imaging buffer only (30 s baseline fluorescence recording), then BODIPY-TMR-CGP (4.5 min association), and followed again by imaging buffer only (4.5 min dissociation). BODIPY-TMR-CGP was excited using a 543 nm helium-neon laser with emission collected through a 565 nm long-pass filter every 3 s throughout each experiment (512 3 512 pixels, averaging at 2 frames). The pinhole diameter (1 airy unit), laser power (2%), offset, and gain remained constant between the 3 BODIPY-TMR-CGP concentrations and the 2 cell lines used. Membraneassociated BODIPY-TMR-CGP fluorescence was measured by drawing regions of interest (ROIs) around the membranes of 10 single cells of each imaged coverslip, and changes in the average pixel intensity values for each ROI over time were analyzed to obtain association and dissociation rates for each experiment.

Measuring the influence of unlabeled ligands on the BODIPY-TMR-CGP dissociation rate
Live cell fluorescenceimagingusing3nMBODIPY-TMR-CGPwas performed on the Zeiss LSM710 laser scanning confocal microscope using a Zeiss Plan-Neofluar 340 1.3 NA oil-immersion objective in conjunction with a perfusion system as described above for the Zeiss LSM510 laser scanning confocal microscope. For association and dissociation kinetic experiments using 3 nM BODIPY-TMR-CGP in CHO-b 1 and CHO-CS cells, the cells were exposed to imaging buffer only (30 s baseline fluorescence recording), then BODIPY-TMR-CGP (4 min association), and followed again by imaging buffer only (4 min dissociation). Influences of unlabeled ligands on the BODIPY-TMR-CGP dissociation rate in CHO-b 1 cells were determined by perfusion of imaging buffer (30 s baseline read), 3 nM BODIPY-TMR-CGP (4 min association), and imaging buffer (4 min dissociation) in the absence or presence of CGP 12177 (0.01-10 mM) or propranolol (0.1-10 mM).
In experiments using CHO-b 1 TM4 cells, the cells were first exposed to 3 nM BODIPY-TMR-CGP for 3.5 min in a 6-well plate prior to placing the coverslip into the imaging chamber to achieve asignificantbutlowleveloflabelingofthereceptor.Onceplaced onto the microscope stage, the cells were perfused with BODIPY-TMR-CGP (30 s baseline), before perfusing imaging buffer in the absence or presence of 1 mM CGP 12177 or 1 mM propranolol (dissociation).
In bimolecular fluorescence complementation (BiFC) experiments, CHO-K1 cells were seeded onto coverslips on day 1 and transiently transfected with YFP N -andYFP C -tagged b 1 -adrenoceptor recombinant DNA (750 ng total) using Lipofectamine 2000 and Opti-MEM medium according to the manufacturer's instructions on day 2. The following day, the transfection medium was removed and replaced with fresh CHO growth medium, before the cells were placed back into the cell culture incubator (37°C, 5% CO 2 /95% air atmosphere). After ;6 h, the cells were moved into a 30°C incubator overnight (5% CO 2 /95% air atmosphere) to allow the maturation of the fluorophore following correct protein folding (26). On day 4, the cells were used for experimentation, and the dissociation of 3 nM BODIPY-TMR-CGP in the absence and presence of unlabeled ligands was determined as described above.
BODIPY-TMR-CGP and YFP (when used) were excited using a 561 nm diode and 488 argon laser, respectively, every 2 s throughout each experiment (512 3 512 pixels, averaging at 2 frames). A variable spectral detection system was used to capture BODIPY-TMR-CGP and YFP emission at 565-605 and 495-535 nm, respectively. The pinhole diameter (1 airy unit) and laser power (5%) remained constant between all experiments, but the gain and offset were adjusted for each experiment for optimal detection of 3 nM BODIPY-TMR-CGP, and kinetic data were expressed in percentage fluorescent intensity to allow data to be grouped and compared.
ROIs were drawn around the membranes of 3-10 single cells of each imaged coverslip, and changes in average pixel intensity values for each ROI over time were analyzed to obtain BODIPY-TMR-CGP dissociation rates for each experiment. In BiFC experiments, membrane-associated BODIPY-TMR-CGP fluorescence was measured by drawing ROIs around membranes of cells that were identified to express BiFC-constrained homodimers by examination of the YFP fluorescence.

Data analysis
GraphPad Prism 5.0 (GraphPad Software, San Diego, CA, USA) was used to fit all data presented in this study. Association kinetic data were fitted using the following monoexponential association equation: where Y 0 is the level of BODIPY-TMR-CGP binding at time t =0 (i.e., baseline fluorescence), Plateau is the level of BODIPY-TMR-CGP binding at infinite time, and k onobs is the rate of observed association. BODIPY-TMR-CGP fluorescence intensities measured at time t 4.5min in CHO-b 1 cells and CHO-CS cells were plotted against the BODIPY-TMR-CGP concentrations used and fitted to a 1-site total binding saturation equation (Eq. 2) or a nonspecific binding linear regression equation (Eq. 3), respectively where B MAX is the maximum specific BODIPY-TMR-CGP binding, [B] is the BODIPY-TMR-CGP concentration, K D is the BODIPY-TMR-CGP concentration that achieves 50% specific binding, M denotes the slope of the linear regression component, and C is the background fluorescence intensity. The dissociation kinetic data were analyzed using the following monoexponential decay equation, if nonspecific binding of ,10% of total binding was observed for a given BODIPY-TMR-CGP concentration where Y 0 , Plateau, and t are the same as defined above with t 0 representing the start of dissociation in Y 0 (i.e., the binding of BODIPY-TMR-CGP at time zero). The k off is the dissociation rate of BODIPY-TMR-CGP. Where a level of nonspecific binding .10% of total binding was observed, the dissociation kinetic data were fitted to a 2-phase exponential decay function where Plateau and t are as defined above, and Span fast and Span slow represent the proportion of Y 0 -Plateau accounted for by the fast (k off (fast) )andslow(k off(slow) ) dissociation rate, respectively. Within this analysis, k off(fast) and Plateau was constrained to the average rate of dissociation and the average Plateau (in %) reached by BODIPY-TMR-CGP in control CHO-CS cells (i.e., cell not expressing the receptor of interest to determine the nonspecific BODIPY-TMR-CGP binding component). Equation 6 was then used to calculate the association rate constants (k on )u s i n gt h ek onobs and k off(slow) determined above for each BODIPY-TMR-CGP concentration [B] The negative logarithms of the equilibrium dissociation constant (pK D ) were obtained using the above determined kinetic parameters in the following equation: Experiments investigating the kinetic parameters of 3 nM BODIPY-TMR-CGP binding to CHO-b 1 cells were analyzed using a global fit of its association (Eq. 8a) and dissociation traces (Eq. 8b) where B MAX is the maximum specific BODIPY-TMR-CGP binding, K D is the BODIPY-TMR-CGP concentration that achieves 50% specificbinding,NSisnonspecific binding, T(ID) is the time at which infinite dilution was commenced, and k on , k off, and t are as described above. Concentration-dependent cooperative effects of unlabeled ligands CGP 12177 and propranolol on the dissociation rate of 3 nM BODIPY-TMR-CGP were fit to the following equation: where E MAX is the maximal increase in BODIPY-TMR-CGP dissociation rate, [B] is the concentration of unlabeled ligand, and K D(site 2) is the concentration of the unlabeled ligand that achieves 50% binding to a secondary binding site that exerts cooperative effects on the BODIPY-TMR-CGP dissociation from the primary binding site. A measure of cooperativity between 2 binding sites is provided by the cooperativity factor a, which was calculated using the following equation: where K D(site 1) is the equilibrium dissociation constant of a given ligand for the primary (site 1) binding site of an unbound receptor, and K D(site 2) is the equilibrium dissociation constant of the same ligand for the secondary (site 2) binding site of a ligandbound (at site 1) receptor. All data are represented as means 6 SEM from n separate experiments. Statistical analysis was performed where appropriate and as detailed in the text, with P , 0.05 reflecting statistical significance.

RESULTS
Kinetic parameters of BODIPY-TMR-CGP at the human b 1 -adrenoceptor The association and dissociation of 10, 30, and 100 nM BODIPY-TMR-CGP at the human b 1 -adrenoceptor were examined in CHO-b 1 and CHO-CS cells to determine total and nonspecific binding levels, respectively. Using the same microscope settings for all BODIPY-TMR-CGP concentrations in both cell lines allowed direct comparison of fluorescence intensities and thus BODIPY-TMR-CGP binding levels ( Fig. 1). The fluorescence intensity levels increased with increasing BODIPY-TMR-CGP concentrations in both CHO-b 1 and CHO-CS cells; however, the nonspecific cell membrane-associated fluorescence intensity in CHO-CS cells was markedly lower than the total binding cell membrane-associated fluorescence intensity measured in CHO-b 1 cells ( Fig. 2A, B). A plot of the binding levels against BODIPY-TMR-CGP concentration was best described by a saturable and linear component following 4.5 min association in CHO-b 1 cells (Fig. 2C). In contrast, the increase in BODIPY-TMR-CGP binding levels with increasing concentration of fluorescent label in CHO-CS cells was best described by a linear relationship (Fig. 2D) ,w h i c hi sc h a r a c t e r i s t i co f nonspecific binding components. Interestingly the nonspecific binding component appeared smaller in the cells expressing the human b 1 -adrenoceptor (Fig. 2C).
The association and dissociation traces obtained in CHO-CS could only be accurately analyzed for 30 and 100 nM BODIPY-TMR-CGP, and revealed rapid observed association (k onobs ) and dissociation (k off )r a t e s ( Table 1), w h i c hp r o v i d e sf u r t h e re v i d ence that BODIPY-TMR-CGP binding interactions in CHO-CS were nonspecific (27).
The association of 10, 30, and 100 nM BODIPY-TMR-CGP in CHO-b 1 cells was monoexponential, and a plateau of BODIPY-TMR-CGP binding was reached following 4.5-min exposure of CHO-b 1 cells to 30 and 100 nM, but not 10 nM, BODIPY-TMR-CGP (Fig. 3C). The concentration dependence of the observed association rates of 10, 30, and 100 nM BODIPY-TMR-CGP at the b 1 -adrenoceptor was clearly seen for the normalized group data (Fig. 3A), as well as the individual cell data (Fig. 3C). The derived observed association rates (k onobs )increasedfrom0.636 0.08 (n = 5) to 1.64 6 0.08 (n =6)and3.086 0.11 min 21 (n =4) for 10, 30, and 100 nM BODIPY-TMR-CGP, respectively. The dissociation of 10 nM BODIPY-TMR-CGP was also monoexponential yielding a dissociation rate of 0.09 6 0.01 min 21 (n = 5). By contrast, 2 dissociation components, a fast (k off(fast) )andaslow(k off(slow) )component, were detected in the dissociation of 30 and 100 nM BODIPY-TMR-CGP. The rate of the fast dissociating component was comparable to the dissociation rate determined in CHO-CS cells and was thus defined as a nonspecificcomponent, which was constrained to the average dissociation rate obtained in CHO-CS cells (2.47 min 21 )inthefitting of the 2-component dissociation rates. The slow dissociation component determined for 30 and 100 nM BODIPY-TMR-CGP in CHO-b 1 cells was 0.09 6 0.01 (n =6) and 0.14 6 0.02 min 21 (n = 4), respectively. Interestingly, the dissociation rate determined for 100 nM BODIPY-TMR-CGP was significantly faster than the rate determined for 10 and 30 nM BODIPY-TMR-CGP (P , 0.05, 1-way ANOVA followed by Tukey's multiple comparisons test; Table 1). We have previously shown that 100 nM BODIPY-  Table 1. Data are means 6 SEM with n representing the number of separate experiments carried out. In each experiment, ROIs were drawn around the membrane of 10 cells. NA, not applicable. *Statistical significance (P , 0.05) from the value determined for 10 nM and # P , 0.05 from 30 nM BODIPY-TMR-CGP in CHO-b 1 cells (1-way ANOVA followed by Tukey's multiple comparisons test).
TMR-CGP binds to the secondary low affinity conformation of the b 1 -adrenoceptor, and as such may cause allosteric e f f e c t st h a tr e s u l ti na ne n h a n c e dd i s s o c i a t i o nr a t eo f 100 nM BODIPY-TMR-CGP (i.e., negative cooperativity) from the primary high-affinity b 1 -adrenoceptor conformation. This was not evident for 10 and 30 nM BODIPY-TMR-CGP, which is likely caused by lower occupancy levels of these BODIPY-TMR-CGP concentrations at the secondary b 1 -adrenoceptor conformation compared with 100 nM BODIPY-TMR-CGP (24). The concentration independence of the dissociation rates of 10 and 30 nM BODIPY-TMR-CGP and the concentration dependence of the observed association rates of 10, 30, and 100 nM BODIPY-TMR-CGP at the b 1 -adrenoceptor were clearly seen for normalized grouped data (Fig. 3A, B), as well as single cell data (Fig. 3C, D). The observed association (k onobs ) and b 1 -adrenoceptor-specific dissociation rate (k off(slow) ) of each BODIPY-TMR-CGP concentration were used to determine the association rate constants (k on ). The k on and k off(slow) were then used to calculate the equilibrium dissociation constant (K D ) for each BODIPY-TMR-CGP concentration and are summarized in Table 1.
Cooperative interactions between the high-and lowaffinity conformation of the human b 1 -adrenoceptor The dissociation rate of a ligand should not be altered in the presence of a second ligand if the two compete for the same binding site. However, if the second ligand binds to a separate second binding site, a resulting conformational change could lead to cooperative (allosteric) effects and thus affect the dissociation rate of the first ligand (23,(28)(29)(30). To assess whether the faster dissociation rate observed for 100 nM BODIPY-TMR-CGP was in fact caused by cooperative interactions between the high-and low-affinity conformation of the b 1 -adrenoceptor, we investigated the dissociation rates of 3 nM BODIPY-TMR-CGP in the absence and presence of increasing concentrations of CGP 12177 and propranolol. We previously showed that 3 nM BODIPY-TMR-CGP predominantly labels the high-affinity conformation over the secondary low-affinity conformation of the b 1 -adrenoceptor [;86% and 3% occupancy, respectively, based on the affinity of BODIPY-TMR-CGP for the high-and low-affinity b 1adrenoceptor conformations determined in functional assays (24)]. This limits competition of labeled and unlabeled ligands at the secondary b 1 -adrenoceptor conformation and therefore ensures that any observed effects on the BODIPY-TMR-CGP dissociation rate are caused by the unlabeled ligand used.
First, we assessed the association and dissociation kinetics of 3 nM BODIPY-TMR-CGP at CHO-b 1 and CHO-CS cells (Fig. 4). The fluorescence intensities measured in CHO-CS cells were too low to accurately determine observed association and dissociation rates. In line with this, the nonspecific binding component in the 3 nM BODIPY-TMR-CGP dissociation trace obtained in CHO-b 1 cells was also too low to be detected and therefore was analyzed as b 1 -adrenoceptor-specific dissociation using a 1-phase dissociation equation. This gave a dissociation rate of 0.09 6 0.01 min 21 (n = 9) in the absence of unlabeled ligands, which was similar to the dissociation rate obtained for 10 and 30 nM BODIPY-TMR-CGP (P . 0.05, 1-way ANOVA followed by Tukey's multiple comparisons test). To selectively label the high-affinity conformation of the b 1 -adrenoceptor, we chose a concentration of 3 nM BODIPY-TMR-CGP (24) for subsequent dissociation experiments and limited the association of fluorescent ligand to 4 min. As a consequence, the association of 3 nM BODIPY-TMR-CGP did not reach a plateau within this time period, and the association rate could not be accurately determined  Table 1. from these data alone. However, when globally analyzed in conjunction with the dissociation trace, an association rate constant (k on )o f5 . 2 76 0.53 3 10 7 M 21 ×min 21 (n =9 ) , a dissociation rate (k off )of0.086 0.01 min 21 (n =9),and apK D of 8.83 6 0.06 (n =9)wereobtained (Fig.4C).
Next, the influence of unlabeled ligands on the BODIPY-TMR-CGP dissociation rate was examined, and the dissociation rate of 3 nM BODIPY-TMR-CGP was significantly enhanced in the presence of 100 nM (0.21 6 0.02 min 21 , n =5),1mM(0.206 0.02 min 21 , n =7),and10mM(0.226 0.03 min 21 , n =5 )C G P1 2 1 7 7( P , 0.05, 1-way ANOVA followed by Dunnett's multiple comparisons test; Fig. 5A and Table 2). A similar increase in BODIPY-TMR-CGP dissociation rate was observed in the presence of 1 mM (0.19 6 0.01 min 21 , n =6 )a n d1 0mM (0.22 6 0.03 min 21 , n =5 ) propranolol (Fig. 5B). The effect of the enhanced dissociation rate was concentration dependent and saturable, which is characteristic of allosteric interactions (28). A concentration-response curve was fitted through the grouped BODIPY-TMR-CGP dissociation rates (listed in Table 2) plotted against the concentrations of unlabeled CGP 12177 (Fig. 5C) and propranolol (Fig. 5D) used, with the midpoint of the curve providing affinity estimates (pK D ) of the unlabeled ligands for a secondary conformation on the b 1 -adrenoceptor with the fluorescent ligand already bound to the primary orthosteric conformation, which were determined to be 7.79 and 6.65 for CGP 12177 and propranolol, respectively. These values are consistent with those determined from inhibition of functional CGP 12177 responses via the secondary conformation of the b 1 -adrenoceptor (11,24).

Influence of TM4 on cooperative interactions involving the b 1 -adrenoceptor
To further investigate the cooperative effects observed above on the BODIPY-TMR-CGP dissociation rate from the catecholamine conformation by the action of propranolol and CGP 12177 acting at the secondary conformation, we examined the effect of CGP 12177 on the BODIPY-TMR-CGP dissociation kinetics in CHO-b 1 TM4 cells. These cells express b 1 -adrenoceptors that have been mutated such that the residues in TM4 are those of the b 2 -adroceptor ( 1 1 ) .I m p o r t a n t l y ,t h i sm u t a n tb 1 -adrenceptor does not exhibit the secondary CGP 12177 conformation (11). In the absence of unlabeled ligands, the dissociation rate of 3 nM BODIPY-TMR-CGP in CHO-b 1 TM4 cells was 0.066 6 0.005 min 21 (n = 8) in these cells. Interestingly, the dissociation rates of 3 nM BODIPY-TMR-CGP in the presence of 1 mMC G P1 2 1 7 7a n d1mM propranolol were 0.089 6 0.010 (n = 6) and 0.080 6 0.007 min 21 (n = 10), respectively, which are comparable to the dissociation rate in the absence of ligands (P . 0.05, 1-way ANOVA followed by Dunnett's multiple comparisons test; Fig. 6).

Effects of enhancing and disrupting b 1 AR homodimer interactions on cooperative interactions
The cooperative effects observed above at the wild-type b 1 -adrenoceptor clearly highlight the presence of 2 distinct binding conformations to which b-adrenoceptor ligands can bind. Baker et al. (7) conducted mutagenesis studies in which selected mutations (e.g., D138A) in the orthosteric ligand binding domain disrupted both the high-and low-affinity binding conformation (7). Two key mutations in the b 1 -adrenoceptor TM4 region completely abolished the secondary conformation (L195Q and W199Y) (11). Interestingly, these are thought to lie within the TM4-TM5 heterodimer interface of the b 1 -adrenoceptor (19), suggesting a potential role of b 1 -adrenoceptor homodimerization in the secondary conformation CGP 12177 pharmacology.
Furthermore, this study showed that the cooperative interactions at the wild-type b 1 -adrenoceptor are completely prevented in CHO-b 1 TM4 cells (Fig. 6). Homodimerization of b 1 -adrenoceptors has been reported to be transient (20,21). Thus, to detect b 1 -adrenoceptor homodimers and allow their pharmacological investigation, we  Data are means 6 SEM of n separate experiments. ND, not determined. *Statistical significance (P , 0.05) in CHO-b 1 cells from control conditions (infinite dilution) for each unlabeled ligand used (1-way ANOVA followed by Dunnett's multiple comparisons test). # Statistical significance (P , 0.05) from the equivalent value in CHO-b 1 cells and @ P , 0.05 from control conditions (infinite dilution) within each cell line (2-way ANOVA followed by Tukey's multiple comparisons test).
used BiFC (31,32) to irreversibly trap and stabilize b 1adrenoceptor homodimers that formed at any given time. We hypothesized that constraining dimers using BiFC would increase the percentage of b 1 -adrenoceptors dimers and as such enhance any dimer-mediated allosteric effects. It has been shown previously that a D138A mutation in TM3 disrupted ligand binding to both the catecholamine and the secondary conformation of the b-adrenoceptor (7). Consequently, any enhanced allosteric effects should be prevented by constraining dimers containing 1 nonligandbinding protomer (b 1D138A ) (7).
BiFC uses 2 nonfluorescent fragments of a fluorescent protein, which reconstitute the functional (i.e., fluorescent) full-length fluorescent protein when in close proximity to one another (32). The N-terminal fragment and the C-terminal fragment of the YFP (YFP N and YFP C ,r espectively) were fused to the C-terminal end of the wildtype or D138A b 1 -adrenoceptor to generate the b 1 YFP N , b 1 YFP C ,a n db 1D138A YFP C receptor constructs. The b 1 YFP N /b 1 YFP C and b 1 YFP N /b 1D138A YFP C constructs were transiently cotransfected into CHO-K1 cells, and clear membrane fluorescence of reconstituted YFP and BODIPY-TMR-CGP binding could be seen (Fig. 7), confirming cell surface expression of wild-type/wild-type and wild-type/D138A b 1 -adrenoceptor homodimers that each contain at least 1 BODIPY-TMR-CGP binding conformation. To confirm that the D138A mutation abolished ligand binding to the b 1 -adrenoceptor, we examined the binding of 3 nM BODIPY-TMR-CGP to a SNAP-tagged D138A b 1adrenoceptor. Indeed, no binding of 3 nM BODIPY-TMR-CGP could be seen in CHO-K1 cells transiently transfected with the SNAP-b 1D138A construct, but clear membrane fluorescence was observed following labeling of the SNAPtag with 1 mM BG-488, confirming the expression of the non-ligand-binding receptor at the cell surface (Fig. 8,  lower left panel). A SNAP-tagged wild-type b 1 -adrenoceptor was transiently transfected as a positive control, and clear fluorescence of the BG-488 labeled SNAP-tag and 3 nM BODIPY-TMR-CGP binding to the wild-type receptor can be seen (Fig. 8, upper panel). This indicates that the lack of BODIPY-TMR-CGP fluorescence seen for the SNAPb 1D138A -transfected cells is caused by the D138A mutation introduced into the b 1 -adrenoceptor.
We then examined the dissociation rate of 3 nM BODIPY-TMR-CGP at irreversibly constrained stable wild-type/wild-type b 1 -adrenoceptor homodimers under ID conditions, which was determined to be 0.02 6 0.01 min 21 (n =5; Fig. 9A). This was significantly slower than the dissociation rate measured in CHO-b 1 cells (P , 0.05, 2-way ANOVA analysis followed by Tukey's multiple comparisons test). The dissociation of 3 nM BODIPY-TMR-CGP binding was enhanced in the presence of 1 mM CGP 12177 and 1 mM propranolol with dissociation rates of 0.186 6 0.008 (n = 6) and 0.189 6 0.007 min 21 (n = 6), respectively (Fig. 9A). This was significantly faster than the dissociation rate determined in the absence of unlabeled ligands (P , 0.05, 2-way ANOVA followed by Tukey's multiple comparisons test; Fig. 9A). The change in the 3 nM BODIPY-TMR-CGP dissociation rate in the absence and presence of unlabeled ligands was ;2-fold in CHO-b 1 cells but was ;10-fold in CHO-K1 cells expressing b 1 YFP N /b 1 YFP C homodimers.
The transient expression of BiFC constructs yielded a mixed population of cells with different expression levels  ( Fig. 7 and Supplemental Fig. S1). To investigate whether higher expression levels of b 1 -adrenoceptor homodimers affected the BODIPY-TMR-CGP dissociation rate meas u r em e n t si nth i ss tu d y ,w ec o m p a r e dt h em ea s ur e m e n t s taken in cells with both high and low expression levels (Supplemental Fig. S1). These data confirmed that the dissociation kinetics were very similar at both expression levels ( Supplemental Fig. S1).

DISCUSSION
In this study, we used a confocal microscopy approach in conjunction with a perfusion system to investigate the dissociation kinetics of BODIPY-TMR-CGP from the human b 1 -adrenoceptor under ID conditions in single living cells. Using this technique, we revealed negative cooperative  interactions between the high-and low-affinity b 1adrenoceptor conformations, which are facilitated by b 1 -adrenoceptor homodimerization.
In CHO-b 1 cells, the observed association rates for 10, 30, and 100 nM BODIPY-TMR-CGP increased in a concentration-dependent manner. The dissociation rate was monophasic for 10 nM fluorescent ligand, but biphasic for higher concentrations (30 and 100 nM) of BODIPY-TMR-CGP. The fast dissociation component was clearly identified as a nonspecific component as it was comparable to the dissociation rate observed in CHO-CS cells lacking the b 1 -adrenoceptor. The b 1 -adrenoceptor-specificd i ssociation rates were comparable for 10 and 30 nM BODIPY-TMR-CGP, but not for 100 nM BODIPY-TMR-CGP, where a significantly faster dissociation rate was observed. According to classic receptor theory, dissociation rates of a ligand should be independent of the ligand concentration used. However, this analysis assumes that the ligand only binds to 1 receptor binding conformation (25). We recently described BODIPY-TMR-CGP binding to both the high-and low-affinity conformations of the human b 1 -adrenoceptor with affinity values of 0.6 and 87 nM, respectively (24). Thus, the observation of a faster dissociation rate for 100 nM BODIPY-TMR-CGP may be caused by BODIPY-TMR-CGP binding to both the high-and lowaffinity b 1 -adrenoceptor conformations and may be a consequence of negative cooperativity occurring between these 2 ligand-bound conformations.
To further explore the potential for negative cooperativity between different conformations within the b 1adrenoceptor, we examined the dissociation rate of 3 nM BODIPY-TMR-CGP in the absence and presence of increasing concentrations of unlabeled ligands. The affinity value derived from the association and dissociation parameters obtained for 3 nM BODIPY-TMR-CGP (;2.6 nM) compared well to the affinity value of BODIPY-TMR-CGP for the orthosteric binding conformation determined in functional studies (0.6 nM) (24). This strongly suggests that 3 nM BODIPY-TMR-CGP predominantly binds the high-affinity b 1 -adrenoceptor conformation. This therefore allows potential cooperative effects exerted by unlabeled ligands binding to a secondary conformation to be measured. The dissociation rate of a labeled ligand should remain unchanged in the presence of an unlabeled ligand that binds to the same receptor site (28). Therefore, any changes of the dissociation rate of 3 nM BODIPY-TMR-CGP from the high-affinity b 1 -adrenoceptor conformation would be caused by the unlabeled b-adrenoceptor ligands CGP 12177 and propranolol binding to a topographically distinct (allosteric) binding conformation that exerts a cooperative effect on the high-affinity (orthosteric) conformation. Indeed, the BODIPY-TMR-CGP dissociation rate wasenhancedby100nMCGP12177andpropranolol(and concentrations above), clearly highlighting the presence of a secondary topographically distinct b 1 -adrenoceptor binding conformation through which CGP 12177 and propranolol exert negative cooperative effects on the primary high-affinity b 1 -adrenoceptor conformation. The apparent dissociation constants (K D ) of CGP 12177 and propranolol for this secondary (allosteric) conformation could be determined from the concentration dependence of the cooperative effect of each ligand on the BODIPY-TMR-CGP dissociation rates. These were similar to those values determined for these ligands for the low-affinity b 1adrenoceptor conformation in previous functional studies (24). This allosteric conformation could therefore be responsible for the secondary low-affinity CGP 12177 b 1adrenoceptor conformation.
A previous mutagenesis study by Baker et al. (7) reported that the 2 b 1 -adrenoceptor conformations must overlap to some degree, as the introduction of certain single point mutations (e.g., D138A) in the b 1adrenoceptor affected both conformations. However, Baker et al. (11) recently identifiedkeyresiduesinTM4 that only affect the pharmacology of the secondary low-affinity b 1 -adrenoceptor conformation, leaving the catecholamine conformation unaffected. The same mutations in TM4 also removed the cooperative effect observed in this study (Fig. 6).
Furthermore, this clearly highlights that the BODIPY-TMR-CGP dissociation kinetics measured in this study were not influenced by rebinding of the fluorescent ligand to unoccupied b 1 -adrenoceptors on neighboring cells. Rebinding can potentially result in a slowed dissociation rate of the labeled ligand, which is then increased in the presence of a high concentration of a competing ligand that prevents rebinding of the labeled ligand (33,34). In this study, the rapid fluid exchange of the perfusion system was used to prevent the reassociation of the fluorescent ligand (27). Most convincingly, however, high concentrations of unlabeled CGP 12177 and propranolol that in CHO-b 1 TM 4 cells are strictly competitive to BODIPY-TMR-CGP binding did not cause an increased dissociation rate of BODIPY-TMR-CGP.
The TM4 has been highlighted to play a role in dimerization of various class A GPCRs (35), including the b 1 -adrenoceptor (19,36), indicating a potential role of b 1 -adrenoceptor homodimerization in the secondary conformation. Homodimers of b 1 -adrenoceptors have been reported to be transient in nature (20). In an attempt to generate more stable b 1 -adrenoceptor dimers, we used a BiFC approach to lock b 1 -adrenoceptor homodimers into constrained stable dimers of defined composition. Although BiFC does not affect the rate of homodimerization (26), the prevention of dimer dissociation as a consequence of the irreversible nature of BiFC will increase the percentage of b 1 -adrenoceptors that exist as homodimers. The successful trapping of BiFC-constrained b 1 -adrenoceptor dimers was demonstrated in CHO cells cotransfected with YFP N and YFP Ctagged b 1 -adrenoceptor constructs by the clear membrane labeling observed with the reconstituted YFP. In cells expressing these wild-type (b 1 YFP N /b 1 YFP C )B i F Cconstrained b 1 -adrenoceptor homodimers, the BODIPY-TMR-CGP dissociation rate was enhanced ;10-fold by 1 mM CGP 12177 and 1 mM propranolol. In contrast, only a ;3-fold difference in dissociation rate was observed in CHO-b 1 cells (i.e., transient unconstrained dimers), suggesting that the cooperative effects of unlabeled ligands on the dissociation rate of BODIPY-TMR-CGP may be mediated across a b 1 -adrenoceptor homodimer interface (Fig. 10A). It was also notable that the dissociation of 3 nM BODIPY-TMR-CGP from native transient b 1adrenoceptor dimers (k off ,0.09min 21 ) was faster than that from wild-type BiFC-constrained b 1 -adrenoceptor homodimers (k off ,0 . 0 2m i n 21 ). This suggests that the formation of stable homodimers itself leads to significant basal allosteric influences on ligand binding kinetics.
To further test whether the secondary (allosteric) b 1adrenoceptor conformation is facilitated by a second b 1 -adrenoceptor protomer in a homodimer complex, BODIPY-TMR-CGP dissociation kinetics were determined in cells expressing BiFC-constrained b 1 -adrenoceptor homodimers of 1 wild-type b 1 -adrenoceptor (b 1 YFP N )and 1 protomer containing a point mutation that abolished binding of b-adrenoceptor ligands at both the primary catecholamine and secondary conformations (b 1D138A YFP C ) (7). The removal of one of the orthosteric binding conformations in a homodimeric b-adrenoceptor pair should remove the potential for negatively cooperative affects across the dimer interface (Fig. 10B). Indeed, in cells expressing b 1 YFP N /b 1D138A YFP C BiFC-constrained homodimers, the effects of CGP 12177 and propranolol on the BODIPY-TMR-CGP dissociation kinetics were reduced and reflected more closely the pharmacology observed in CHO-b 1 cells (i.e., transient wild-type unconstrained dimers). This residual cooperativity is most likely caused by the cooperative effects that can still occur across transient wild-type b 1 YFP N /b 1 YFP N dimers that will be present in the cell population. Two dimerization interfaces have been reported for the b 1adrenoceptor involving transmembrane regions 1 and 2 and helix 8 (TM1/TM2/H8) in the first interface and transmembrane region 4 and 5 and intracellular loop 2 (TM4/TM5/ICL2) in the second interface, the latter of which has been described to make structural rearrangements during receptor activation (19). Key residues (L195, W199) identified by Baker et al. (11) that are responsible for the secondary conformation lie within the dimer interface region of TM4. Using a stable cell line containing the reported b 1 -adrenoceptor TM4 mutations that result in no secondary conformation (11), we observed no effects on the dissociation rate of 3 nM BODIPY-TMR-CGP in the presence of 1 mM CGP 12177 and 1 mM propranolol. The loss of the enhanced dissociation rate in cells expressing b 1adrenoceptor TM4 mutations is consistent with a role of a dimerization interface involving TM4 in the negative cooperativity observed at the b 1 -adrenoceptor (Fig. 10C).
In summary, these data suggest that the secondary low-affinity conformation of the b 1 -adrenoceptor may be a consequence of negative cooperative interactions between 2 orthosteric binding conformations within a b 1adrenoceptor homodimer. Thiscanthenleadtoareduced apparent affinity of ligands for the second protomer of an already ligand-occupied (on the first protomer) dimer [see Supplemental Fig. S1 in May et al. (23)]. The contribution of these cooperative interactions is provided by the cooperativity factor a, and the ligand affinity for the already ligand-bound receptor is described as a ratio of the ligand affinity for the unbound receptor and the cooperativity factor a (K B /a). As such, the cooperativity factor at b 1adrenoceptor dimers can be determined by taking the ratio of the apparent K D value determined for binding to the first protomer (orthosteric b 1 -adrenoceptor conformation 1 K D , i.e., unbound receptor) and that determined for modifying the dissociation rate of BODIPY-TMR-CGP from conformation 1 by an unlabeled ligand binding to the second protomer (allosteric b 1 -adrenoceptor conformation 2 K D , i.e., ligand-bound receptor). Negative cooperativity that leads to an increase in the apparent dissociation observed is reflected in a cooperativity factor smaller than unity. Indeed, the cooperativity factors (a) for CGP 12177 andpropranololwereestimatedtobe0.015and0.010,respectively, using the binding affinities for the orthosteric b 1adrenoceptor conformation determined in Gherbi et al. (24). A mechanistic framework for the secondary b 1adrenoceptor conformation based on homodimer formation opens up new insights into the role of dimerization in altering the molecular pharmacology of GPCRs. Figure 10. Schematic diagram describing negative cooperative interactions at b 1 -adrenoceptor homodimers. A)A b 1 -adrenoceptor homodimer possesses 2 endogenous (i.e., structurally identical) ligand-binding sites for which b 1 -adrenoceptor ligands have the same affinity. However, following binding of a b 1 -adrenoceptor ligand to 1 b 1 -adrenoceptor site (primary, orthosteric binding site; red filled circles) with high affinity, negative cooperativity between the 2 b 1 -adrenoceptor binding sites in a b 1 -adrencoeptor homodimer results in a markedly reduced affinity (increased K D ) of a secondary b-adrenoceptor ligand for the secondary b 1 -adrenoceptor site (secondary, allosteric binding site; black filled circles), which is often described as the low-affinity CGP 12177 binding site. This negative cooperativity is reciprocal between the 2 b 1adrenoceptor binding sites, and ligand binding to the secondary site causes an enhanced dissociation rate of the ligand already bound to the primary site. The cooperativity factor a provides a quantitative estimate of the degree and direction of cooperativity between 2 binding sites and is defined as the ratio of a ligand'saf finity for the free receptor over the affinity of the same ligand for the already ligand-occupied receptor (23,28). Negative cooperativity is indicated by a cooperativity factor smaller than unity. The low-affinity b 1 -adrenoceptor site therefore represents the binding affinity of ligand A for a receptor dimer where 1 protomer is already occupied, and the dissociation constant is given by K D /a (see equation). Effective removal of the orthosteric site from one of the protomers through (B) a point mutation that abolishes ligand binding or (C) disruption of the TM4/TM5 b 1 -adrenoceptor dimerization interface removes any cooperative effects between 2 b 1 -adrenoceptor sites across the homodimer interface. In these 2 situations, the binding equation reverts to a simple mass action equilibrium between ligand A and receptor R.