Abstract
The long-acting β2 sympathomimetics salmeterol and formoterol have been presumed to exert their prolonged action either by binding to an accessory binding site (“exo-site”) near the β2 adrenoceptor or by their high affinity for β2 adrenoceptors and correspondingly slow dissociation. Whereas most studies with salmeterol had been done in intact tissues, which have slow diffusion and compartmentation of drugs in lipophilic phases, that restrict drug access to the receptor biophase, we used purified receptor membranes from rat lung and disaggregated calf tracheal myocytes as model systems. Binding experiments were designed to measure the slow dissociation of agonists by means of delayed association of (−)-[125I]iodopindolol. Rat lung membranes were pretreated with high concentrations of agonists (salmeterol, formoterol, isoprenaline) before dissociation was induced by 50-fold dilution. Half-times of association of (−)-[125I]iodopindolol remained unchanged compared with untreated controls, indicating that dissociation of agonists occurred in less than 2 min. Adenylyl cyclase experiments were designed to determine the on and off kinetics of agonists to β2 adrenoceptors by measuring the rate of receptor-induced cyclic AMP (cAMP) formation. Experiments were performed in tracheal membranes characterized by highVmax values of cAMP formation. Adenylyl cyclase activation occurred simultaneously with the addition of the agonist, continued linearly with time for 60 min, and ceased immediately after the antagonist was added. Similarly, when receptor membranes were preincubated in a small volume with high salmeterol concentrations, there was a linear increase in cAMP formation, which was immediately interrupted by a 100-fold dilution of the reaction mixture. This militates against the exo-site hypothesis. On the other hand, dissociation by dilution was much less when membranes were preincubated with a large volume of salmeterol at the same concentration, indicating that physicochemical effects, and not exo-site binding, underlie its prolonged mode of action.
Formoterol (FOR) and salmeterol (SLM) are highly selective β2 adrenoceptor (AR) agonists characterized by a long duration of action. Whereas the long-acting bronchodilation by FOR was found by chance in clinical studies (Löfdahl and Svedmyr, 1989; Anderson, 1993), SLM was the result of a specific research program to design long-acting drugs through molecular modification of the β2AR agonist salbutamol (for review, seeJack, 1991; Johnson et al., 1993; Johnson, 1995).
FOR, which is characterized by high potency for relaxation of airway smooth muscle (Ida, 1976; Lemoine and Overlack, 1992), high-affinity binding to the high- and low-affinity states of pulmonary β2AR (Lemoine, 1992; Lemoine et al., 1992) and of β2AR expressed in COS-7 cells (Green et al., 1996), and high efficacy for stimulation of β2AR-coupled adenylyl cyclase (AC), acts as a full agonist for β2AR (Lemoine and Overlack, 1992). Based on radioligand-binding studies with (±)-[3H]FOR and on the resistance of (±)-[3H]FOR to displacement by high concentrations of βAR agonists or antagonists, it was postulated that the stability of the FOR-βAR complexes may contribute to its long-lasting therapeutical action (Lemoine, 1992).
The SLM-induced relaxation of airway smooth muscle is atypical in that it is slow in onset of action, prolonged in duration, and almost resistant to washout (Ball et al., 1991; Lindén et al., 1991). The tenacity of the action of SLM was first observed in vitro in isolated airway smooth muscle, in which SLM relaxation reappears after blockade by βAR antagonist and successive washout of antagonist without the readdition of SLM, a phenomenon termed “reassertion” (Ball et al., 1991). This unique property of a βAR agonist led to the hypothesis that SLM binds to two sites of the β2AR: the classic βAR site (active site) interacting with the saligenin moiety of SLM and a new site, called the “exo-site,” to which the hydrophobic tail of the molecule is supposed to bind quasi-irreversibly. As a consequence of this hypothesis, a high-affinity binding of the hydrophobic tail to the exo-site is thought 1) to keep SLM in the vicinity of the receptor, thus improving the likelihood that the saligenin head interacts with the active site; and 2) to restore its action by flipping in and out of the active site after withdrawal of antagonists (Johnson et al., 1993).
The high lipophilicity of the compound (Rhodes et al., 1992) led to another plausible hypothesis, the “diffusion microkinetic model” ofAnderson et al. (1994). The essential feature of the microkinetic model is that after the inhalation of SLM, a bulk concentration of the drug enters the plasmalemma lipid bilayer of airway smooth muscle cells and acts as an agonist depot even after withdrawal of the drug. In this model, drug access to the active site of the β2AR occurs via lateral diffusion between the α helices into the β2AR rather than via a direct approach from the extracellular aqueous biophase, thus accounting for the slow onset and long duration of action.
More recently, evidence in favor of the exo-site binding of SLM was published based on site-directed mutagenesis of β2AR by replacing β2AR amino acids 149 through 173 of the transmembrane spanning domain IV with the corresponding β1AR sequences (Green et al., 1996). In the resultant constructs expressed in COS-7 cells and assayed by radioligand binding, SLM binding under washout conditions was reduced by 67%. In contrast were findings with structural analogs of SLM characterized by the same aliphatic side chain known to be responsible for exo-site binding of SLM, which, however, did not compete with SLM for the exo-site (Bergendal et al., 1996). All of the studies supporting exo-site binding and prolonged action cited above were performed in isolated tissues and superflow organ baths or with cell culture techniques that use large volumes of buffer and high amounts of SLM. We decided to use isolated receptor membranes that were pretreated in small volumes with low amounts of SLM and used the βAR-coupled AC activity to assess SLM binding to the receptor. Some of these results have been published in preliminary form (Teschemacher and Lemoine, 1998).
Materials and Methods
Preparation of Rat Lung Membranes
Rat lung membranes were prepared as described by Lemoine et al., (1992). Briefly, lungs were placed in 1 mM ice-cold KHCO3, minced, homogenized with a Polytron (1 × 2-s setting 7, 2 × 20-s setting 3.5), and filtered through Japanese silk. The crude homogenate was centrifuged at 120g for 10 min, the pellet was discarded, and the supernatant was collected at 40,000g for 30 min. Lung membranes were further purified by a sucrose density gradient (20:40% w/w) followed by a 2-h centrifugation at 100,000g to collect membrane particles. The βAR density in purified membranes was approximately 1000 fmol/mg, as estimated in saturation binding experiments with [125I]iodopindolol (IPIN).
Isolation of Smooth Muscle Cells from Calf Trachea
Cells were isolated by enzymatic disaggregation as detailed inLemoine et al., (1989). Briefly summarized, tracheal strips were freed from cartilage, mucosa, and connective tissue at room temperature in Krebs’ solution containing 89 mM NaCl, 30 mM NaHCO3, 5 mM KCl, 1 mM Na2HPO4, and 0.5 mM MgSO4, oxygenated with carbogen (95% O2/5% CO2). The ring muscle layer was then cut into narrow strips and incubated in an isotonic (130 mM) potassium-methanesulfonate solution, buffered with 10 mM HEPES at pH 7.4, and supplemented with 5 mM pyruvate, 5 mM creatine, 0.7 g/liter collagenase D (Boehringer Mannheim, Mannheim, Germany), and 0.1 g/liter Pronase E (Serva, Heidelberg, Germany). During enzymatic disaggregation, muscle strips were gently moved by magnetic stirring. Cells were harvested every 30 min and washed once in isotonic buffer, yielding fusiformed smooth muscle cells. Membranes were obtained after gentle homogenization of cells with a glass-glass homogenizer and centrifugation with 40,000g for 30 min at 4°C. The pellet was resuspended in 3 mM EGTA (pH 7.4) with 0.3 mM ascorbic acid and stored at −80°C until use. The protein content was determined according to Bradford (1976) using BSA as a standard.
Radioligand Binding
Experiments were carried out in rat lung membranes at 37°C in 50 mM Tris·HCl (pH 7.4), 2 mM MgCl2, 1 mM EGTA, and 0.1 mM ascorbic acid and in the presence of a nonhydrolyzable GTP analog {10 μM guanosine-5′-(β,γ-imido)triphosphate [Gpp(NH)p]}) to convert the receptors in the low-affinity state for agonist binding. In addition, 1 μM CGP 20,712 A was added, a concentration known to selectively block β1AR (Lemoine and Kaumann, 1991; Lemoine and Overlack, 1992). To measure slow dissociation of agonists, lung membranes were pretreated with a nonradioactive agonist and used, after dilution, for association binding with a radioligand ([125I]IPIN). Untreated membranes were used as a control. A delayed association of the radioligand (L*) with the receptors (R) was expected for agonists (A) with slow dissociation (Fig. 1). Membranes were preincubated for 30 min with agonist in a small volume. After 30 min, 3-μl samples of pretreated membranes were transferred into a volume of 150 μl (dilution of 1:50) containing radioligand (0.2–0.4 nM [125I]IPIN), 10 μM Gpp(NH)p, and 1 μM CGP 20,712 A. Thus, dissociation of the agonist was induced by a 50-fold dilution and the simultaneous addition of the radioligand. Binding observed in the presence of 0.2 mM (−)-isoproterenol was regarded to be nonspecific (Bns ≪ 10%). Bound radioligand was separated from free radioligand by rapid vacuum filtration through Whatman GF/A glass-fiber filters. Bound radioligand was counted in a gamma counter with an efficiency of 70%. Association kinetics of [125I]IPIN (L*) were analyzed (see also Lemoine, 1992) by nonlinear regression as a pseudo-first order kinetic according to
To determine equilibrium dissociation constantsKD of agonists for binding (B) to β2AR, competition binding experiments were performed with [125I]IPIN (L*) in the presence of increasing concentrations of agonists (A) and analyzed by nonlinear regression (Ehle et al., 1985; Lemoine et al., 1985) according to
AC Assay
Assays were carried out in 60 μl of a reagent (MIX) containing 0.04 mM [α-32P]ATP (250–350 cpm·pmol−1), 100 mM Tris·HCl (pH 7.4), 2 mM MgCl2, 1 mM EGTA, 0.1 mM ascorbic acid, 0.01 mM GTP, 1 mM [3H]cyclic AMP (cAMP; 10.000 dpm/tube), 20 mM creatine phosphate as substrate of the ATP-regenerating enzyme system, and the enzymes (IU·ml−1) 15 mM creatine phosphokinase and 9.8 mM myokinase. The reaction was performed at 32.5°C and terminated by a quick freeze stop in liquid nitrogen and the addition of 50 μl of a stop solution containing 10 mM cAMP, 40 mM ATP, and 1% SDS. Cyclic [32P]AMP was purified by double-column chromatography as described by Salomon et al. (1974) and measured by liquid scintillation counting.
AC Kinetics
For the analysis of receptor interactions with long-acting SLM in comparison to other β2 SYM, kinetic experiments were designed based on the close coupling of receptors to AC; the association of an agonist with a receptor is directly translated to an increase in cyclase activity and after the dissociation of the agonist AC activity immediately ceases. Thus, the dissociation of SLM was analyzed under several conditions and compared with that of other agonists.
Dissociation by Addition of a Blocker.
The assay was started by addition of 20 μl of native receptor membranes to 40 μl of AC buffer supplemented with the indicated agonists (see Fig. 3). The βAR antagonist (−)-bupranolol (BU; 1 μM) was added to the test tubes 5 min after the start of the reaction to induce the dissociation of agonists.
Dissociation by Dilution.
Dissociation of drug/receptor complexes was induced by a 100-fold dilution of membranes preincubated with various agonists (ISO, FOR, SLM) in a small volume (20 μl, Fig.4A step 1) (see Figs. 4A, 5, and 6). Drug concentration was 10 K (condition HIGH and Dissociation) and 0.1 K (condition LOW). After a 15-min preincubation at 32.5°C, a 100-fold volume (2 ml) of MIX reagent was added containing a drug concentration of 10, 0.1, or 0 K for the conditions HIGH, LOW, and Dissociation (Fig. 4A step 2), respectively. Aliquots of 50 μl were taken at the times indicated in the Figs. 5 and 6.
Dissociation by Dilution after SLM Loading.
SLM-loaded membranes were obtained by pretreatment with 20 nM SLM in a large volume of 40 ml for 1 h at 10°C (Fig. 4B step 1) (see Figs. 7and 8). Membranes were collected by centrifugation at 40,000g and 4°C for 30 min. After resuspension of pellets, membranes were preincubated with 20 nM SLM in the absence (condition HIGH) or presence of 100 KB blocker [80 nM BU; condition LOW] or in the absence of all drugs (condition Dissociation) in a volume of 20 μl (step 2). After 15 min at 32.5°C, AC reaction was started by the addition of 2 ml of MIX reagent (Fig. 4B step 3) containing the same concentrations of SLM and BU as used for preincubation. Aliquots were taken and stopped at the times indicated.
Comparison of Differently Pretreated Membranes.
Membranes were loaded with SLM as described in Fig. 4B and compared with untreated controls (sham membranes) (see Figs. 8 and 9). As a third condition to test a putative wash-out effect, SLM-loaded membranes were washed by incubation in 40 ml of fresh buffer (3 mM Tris, 3 mM EGTA, pH 7.4) before being collected by centrifugation. Membrane pellets in all three conditions had a volume of 100 μl and were resuspended in 10 ml. AC assays were started without preincubation by the addition of 20 μl of membrane suspensions to 40 μl of AC buffer in the presence or absence of drugs. The reaction was stopped after 20 min by freeze stop.
Estimation of Free SLM Concentration
Bioassay for SLM Concentration.
The free concentration of SLM in SLM-loaded membranes can be estimated 1) directly by comparison of preloaded membranes to native membranes stimulated with SLM in a concentration-dependent manner and 2) indirectly from the inhibition of AC activity in preloaded membranes by an antagonist. For this purpose, the concentration-effect curves for agonists and antagonists must be fitted by nonlinear regression according to the following set of equations. Concentration-effect curves for AC stimulation by SLM (see Fig. 9, top) were analyzed according to:
Estimation of the SLM Concentration with Partition Coefficients.
Rhodes et al. (1992) determined membrane partition coefficients Kp[mem] according to
Drugs and Materials
[α-32P]ATP (22.2 TBq/mmol) and [125I]IPIN (81.4 TBq/mmol) were purchased from Du Pont de Nemours (Dreieich, Germany). Cyclic [3H]AMP (1.33 TBq/mmol) was purchased from the Radiochemical Center (Amersham, UK). ATP, cAMP, GTP, Gpp(NH)p, creatine phosphate, myokinase, (−)-ISO HCl, and forskolin were obtained from Sigma Chemical Co. (St. Louis, MO). Creatine phosphokinase was obtained from Calbiochem (La Jolla, CA). The antagonists BU and CGP 20,712 A (1-[2(3-carbamoyl-4-hydroxy phenoxy)-ethylamino]3-[4-(1-methyl-4-trifluoromethyl-2 imidazolyl)phenoxy]-2-propanol methanesulfonate) were purchased from Schwarz-Sanol (Monheim, Germany) and Ciba-Geigy (Basel, Switzerland). (−)-FOR and (±)-SLM were generous gifts from Ciba-Geigy (Basel, Switzerland) and Glaxo (Hamburg, Germany), respectively.
Results
Radioligand Binding.
The dissociation of SLM, which is not a promising candidate to be developed as a radioligand because of its high lipophilicity, can be studied only indirectly (i.e., by a delayed association of a radioligand that itself is characterized by fast association kinetics) as shown in Fig. 1. We chose [125I]IPIN for this purpose and studied its receptor binding kinetics in rat lung membranes that possess a high density of βAR. The receptor density was measured with saturation binding, resulting in a Bmax value of 1093 ± 24 fmol/mg protein and a dissociation constant (−log, M) of 9.67 ± 0.09 for [125I]IPIN (experiment not shown). To study receptor -binding kinetics, the subdominant β1 subtype of lung βAR (Lemoine, 1992) was antagonized with 1 μM CGP 20,712 A, a concentration known to prevent binding to β1AR without influencing binding to β2AR (Lemoine and Kaumann, 1991). A nonhydrolyzable GTP analog [10 μM Gpp(NH)p] was used to convert the receptors from the high-affinity to the low-affinity binding state for agonists. Thus, it was possible to study the association and dissociation kinetics of [125I]IPIN to a homogeneous class of low-affinity β2AR. The kinetics followed a simple exponential relation and were fitted by nonlinear regression to eqs. 1 and 2, respectively. The parameter estimates listed in Table1 illustrate fast association and slow dissociation of [125I]IPIN with half-times (T1/2) of approximately 1.0 and 5.2 min, respectively. The ratio of kinetic constants (−Logkoff/kon= 10.25 ± 0.24) matched the dissociation constant pKD of 9.67 ± 0.09 determined by saturation binding, confirming the high affinity of [125I]IPIN for β2AR. For the agonists of interest (ISO, FOR, SLM), competition binding experiments (not shown) were performed in the presence of 10 μM Gpp(NH)p. The dissociation constants (pKD) estimated by nonlinear regression (eq. 4) were 6.62 ± 0.04 for ISO, 8.06 ± 0.04 for FOR, and 8.44 ± 0.04 for SLM.
To measure a putative slow dissociation of long-acting SLM, receptor membranes were pretreated with a high concentration of SLM (100 nM), resulting in a receptor occupation by SLM of greater than 90% (Fig.2, bottom). Receptor membranes pretreated with a hydrophilic catecholamine (5 μM ISO, receptor occupancy > 90%) were coanalyzed (Fig. 2, top). Small samples of pretreated membranes (3 μl) were diluted by adding 150 μl of reaction buffer containing a low concentration (<0.4 nM) of [125I]IPIN, thereby combining the dissociation of cold ligands by dilution (50-fold) with the association of the radioligand (Fig. 1). The association of [125I]IPIN was fast in both cases, independent of whether the membranes had been pretreated with SLM or ISO. Similar experiments (not shown) were performed with higher concentrations of agonists (500 nM SLM, 15 μM ISO) and with 3 and 10 nM concentrations of a high-affinity antagonist [BU, pKD = 9.3; Lemoine et al., 1985]. Association kinetics were analyzed by nonlinear regression according to eq. 1 resulting in estimates of kappvalues for association (Table 1). To correct for the variation of the concentration of radioligand, konvalues were calculated assuming that the time constant of dissociationkoff of [125I]IPIN was unaffected by the presence of nonradioactive ligands (Table 1). The comparison ofkon values shows that pretreatment with either agonist, SLM or ISO, slows the association of [125I]IPIN slightly by a factor of about 2 to 3, whereas pretreatment with a high-affinity antagonist (BU) retards the association by a factor of about 5.
AC Assay
In regard to the exo-site hypothesis (Johnson et al., 1993), the fast dissociation of SLM induced by a radioligand is not sufficient to disprove the hypothesis; that is why AC experiments were designed that also allow the dissociation of SLM to be determined in the absence of antagonists. Thus, dissociation of SLM was induced by a 100-fold dilution of membranes preincubated in a small (20 μl) or a large (40 ml) volume of buffer containing 20 nM SLM and compared with dissociation kinetics induced by a βAR antagonist. For AC experiments, membranes of disaggregated tracheal cells were used that can be stimulated maximally up to 10-fold over basal by βAR agonists (Lemoine et al., 1989).
Blocker-Induced Dissociation of SLM.
Stimulation of AC activity by maximum effective concentrations of ISO and SLM induced an immediate increase of cAMP production that was linear with time up to 20 min (Fig. 3). Maximal AC stimulation by ISO was 4.7 times higher than basal activity, and that by SLM was 2.5 times higher. Thus, SLM was characterized as a partial agonist. Dissociation of SLM was induced by the addition of a surplus (10 μM) of the βAR blocker BU added 5 min after the start of the reaction with SLM. With the kinetic constants measured with the [3H] derivative of BU (kon = 1.21 min−1·nM−1 of the radioligand, koff = 0.26 min−1; Lemoine et al., 1985), it could be calculated that the association of BU at a concentration of 10 μM would occur with a half-time of less than 30 ms (i.e., the association of the competing ligand with the receptors was not rate limiting). The experiment depicted in Fig. 3 (triangles) shows that the dissociation of SLM was fast and complete, resulting in a line parallel to basal activity. Thus, it had to be concluded that the dissociation of SLM from the β2AR was complete within less than 1 min. Similar experiments (not shown) performed with 100 nM (−)-FOR and 5000 nM (−)-ISO showed an immediate association by the addition of the respective agonist and an immediate dissociation by the addition of BU.
Dissociation of SLM Induced by Dilution.
Membranes of tracheal myocytes were preincubated with 20 nM SLM in a very small volume (20 μl; Fig. 4A); dissociation was induced by a 100-fold dilution with MIX reagent containing [α-32P]ATP to start the AC reaction simultaneously (Fig. 5). A concentration of 20 nM (10 K) was chosen to cause a submaximal (about 90%) stimulation of AC activity. The basal rate of cAMP formation of 98.4 ± 3.7 pmol·mg−1·min−1 was stimulated to 368.8 ± 5.2 pmol·mg−1·min−1 with 200 μM ISO. With 10 K SLM, the cAMP formation rate was 314.4 ± 10.6, and that with 0.1 K SLM was 131.0 ± 4.0 pmol·mg−1·min−1. As with dissociation by adding a blocker (Fig. 3), the dissociation by dilution (Fig. 5) caused an immediate cessation of AC stimulation by SLM, in contrast to the predictions of the exo-site hypothesis. The kinetic after dilution matched the kinetic stimulated with 0.2 nM SLM, corresponding to the 0.1 K condition. Both kinetics could be fitted with straight lines. It can be concluded from the small difference between the two lines that the dissociation of SLM is complete within 1 min and that there is no hint at a prolonged action of SLM on the level of receptor-coupled AC.
Dissociation of (−)-FOR and (−)-ISO Induced by Dilution.
As with SLM, the dissociation of the long-acting β2SYM (−)-FOR and of the short-acting catecholamine ISO was investigated by dilution of pretreated membranes of tracheal myocytes. Membranes were pretreated with concentrations of 500 nM ISO (Fig. 6, bottom) and 50 nM (−)-FOR (Fig. 6, top), which were submaximally effective with Vmax values approximately 3.5 times basal rates. As in the previous experiments, a 100-fold dilution caused an immediate cessation of AC activity by βAR agonists; reaction kinetics after dilution were linear and matched linear kinetics stimulated with 0.1 K concentrations.
Absence of Dissociation after Loading of Membranes with SLM.
To examine whether the total amount of SLM offered to the membranes during pretreatment has an influence on the dissociation of SLM, membranes were incubated in 40 ml of Tris buffer with 20 nM SLM for 60 min at 10°C, a process termed “loading”. After collection of the membranes by centrifugation, the AC assay was performed as depicted in Fig. 4B. In addition to the submaximal stimulation (474.9 ± 3.8 pmol cAMP·min−1·mg protein−1) at a concentration equivalent to 10 K (20 nM SLM) and to the 100-fold dilution, a third condition was chosen, which showed that 80 nM BU nearly completely reversed prestimulation with SLM (Fig. 7), reducing activity to basal rates (137.2 ± 2.4 pmol cAMP·min−1·mg protein−1). In contrast to the blocker-induced reversal of AC stimulation in preloaded membranes and to the reversal by a 100-fold dilution after pretreatment of membrane in a small volume (Fig. 5), a 100-fold dilution of preloaded membranes was quite ineffective (Fig. 7). The cAMP production after dilution was almost as high as that in the membranes stimulated with 20 nM SLM. From this experiment, it becomes apparent that the exo-site binding thought to be accompanied by slow dissociation kinetics can be mimicked by incubating the membranes in a large volume of buffer containing SLM, presumably by trapping of SLM in the phospholipid bilayer of the receptor membranes.
Ineffectiveness of Washing of Preloaded Membranes.
Sham membranes (Fig. 8) were compared with membranes submitted to the loading procedure with SLM (“preloaded”) and with preloaded membranes that were washed by centrifugation (“loaded & washed”). AC activity was stimulated via βAR using ISO or SLM, or directly with forskolin and was compared with basal conditions in the absence and presence of the βAR antagonist BU. The experiment clearly indicates 1) that the basal rate in sham membranes matched the rate in the presence of the βAR antagonist in preloaded membranes; 2) that the elevated rate of cAMP formation in SLM-preloaded membranes persisted without the addition of SLM during incubation, thereby confirming the results of Fig. 7; and 3) that the washing procedure did not reduce the rate of cAMP formation in preloaded membranes, hinting at the effectiveness of trapping of SLM by the membranes.
Estimation of Unknown SLM Concentration in Preloaded Membranes by Bioassay.
A bioassay was designed to determine the unknown SLM concentration of preloaded membranes with the help of the SLM-induced AC stimulation. As detailed in Materials and Methods, the unknown SLM concentration was determined (Fig.9) as a shift of inhibition curves by an antagonist (BU), comparing the inhibition in preloaded membranes (pIC50 = 8.46 ± 0.08) with sham membranes in the presence of 20 nM SLM (pIC50 = 8.08 ± 0.07). Taking into account the estimated dissociation constant of BU (pKB = 9.16, eq. 7), the pEC50 value for SLM of 8.74 (Fig. 9, top), and the volume of buffer used for resuspension of pelleted membranes (100 μl of membranes were resuspended in 30 ml of buffer), the unknown SLM concentration in preloaded membranes was approximately 7.3 nM (eq. 8). This is the concentration that is effective under AC assay conditions, and it corresponds to a SLM concentration of 2.4 μM (i.e., 240 pmol in 100 μl) in the membrane pellet and to a 120-fold accumulation during the loading process (Fig. 4B).
Estimation of Unknown SLM Concentration in Preloaded Membranes Using Partition Coefficients.
Partition coefficients had been reported to be 22,500 for the membrane partition coefficientKp[mem] and 7600 for the octanol-water coefficient Kp[o/w](Rhodes et al., 1992). Assuming as a first approximation that the mass of membrane lipids equals that of membrane proteins, the mass of membrane lipids that contributed to the trapping of SLM during the loading process (Fig. 4B) was about 2.5 mg. Using this assumption and with a value for Kp[mem] = 22,500, eq. 10 shows that about 58% of the free SLM concentration was trapped in membrane lipids. Given a concentration of 20 nM SLM in a volume of 40 ml (corresponding to a total of 800 pmol of SLM) offered to the membranes during loading, the mass of trapped SLM was approximately 470 pmol. On the basis of octanol-water coefficient, eq. 8 yields an estimate of 32% (260 pmol of SLM) accumulated in the membrane lipids. This latter value for the amount of trapped SLM matches the amount of SLM estimated by bioassay (30% and 240 pmol).
Discussion
Our results using βAR-coupled AC activity as a measure of the occupancy of β2AR by long-acting β2SYM clearly indicate that in isolated receptor membranes 1) the onset of receptor occupation and theon kinetics of AC begin as soon as the agonist is added and 2) the off kinetics of AC, which represent dissociation of receptor-agonist complexes, immediately follow dilution or the addition of antagonist. These findings were made with a hydrophilic catecholamine (ISO, log Kow = 0.64), a weak lipophilic agonist (FOR, log Kow= 1.40), and a strong lipophilic partial agonist (SLM, logKow = 4.15;Kow values are octanol/water partition coefficients calculated according to Meylan and Howard, 1995). In the case of SLM, these findings contradict the exo-site hypothesis, which essentially predicts a delay of dissociation from the exo-site combined with sustained activation of β2AR and receptor-mediated signal transduction.
To elucidate what may have led to the postulation of the exo-site hypothesis, we tried to imitate the experimental conditions that underlaid its postulations. These experiments were done in organ baths with superfused tissues (Ball et al., 1991; Nials et al., 1993) or in cell culture systems (McCrea and Hill, 1993; Green et al., 1996) characterized by the use of high amounts of SLM. For example, in experiments with isolated tracheal strips, 100 nM SLM were superfused for 20 min at a flow rate of 2 ml/min, resulting in a total application of 4000 pmol of SLM to the tissue (Ball et al., 1991); or in cell culture experiments, 1 μM SLM was incubated for 10 min in 35-mm dishes, resulting in a total application of about 2000 pmol of SLM to the cell monolayers (Green et al., 1996). In the experiments we designed to mimic exo-site binding of SLM with receptor membranes, 40 ml with only 20 nM SLM (i.e., 800 pmol of SLM) was preloaded for 60 min at 10°C; nevertheless, the dissociation by dilution failed (Fig. 7). This is in stark contrast to the experiments with membranes preloaded in a small volume of 20 μl (i.e., 0.4 pmol of SLM, Fig. 5) and indicates that it was the manipulation of experimental conditions and not a characteristic behavior of the receptor that was responsible for the persistence of SLM effects.
The same line of evidence was found with radioligand-binding techniques applied to rat lung membranes. Whereas the pretreatment with high concentrations of SLM (10 μM, corresponding to about 1000KD of the drug) completely prevents the access of [125I]IPIN to β2AR (Jack, 1991; Johnson, 1991; Nials et al., 1993), smaller concentrations (100 nM, corresponding to about 10KD), which are supposed to be achieved during therapy of bronchial asthma, have no effect on the access of [125I]IPIN to β2AR (this report). Based on these results, it was concluded that [125I]IPIN, by virtue of its 2-iodo substituent, acquires a molecular size sufficient to deny its access to the binding sites of β2AR when SLM is present (Jack, 1991). In contrast, other groups suppose that [125I]iodocyanopindolol, which is an even bigger molecule than [125I]IPIN, freely interacts with the receptor binding site even though SLM is quasi-irreversibly bound to the exo-site (Clark et al., 1996; Green et al., 1996), hinting at some inconsistencies in exo-site theory.
Recently, the exo-site hypothesis was supported by experiments using site-directed mutagenesis to identify amino acids of the β2 AR involved in the exo-site binding of SLM (Green et al., 1996). The replacement of these amino acids by the analogous β1AR sequences, which are not suspected of contributing to exo-site binding, resulted in mutants that were expressed in COS-7 cells and assayed by radioligand binding with [125I]iodocyanopindolol. However, the reduction of radioligand binding by SLM in cells with wild-type β2AR was only 38.2% [in contrast to a 90% inhibition of binding in rat lung membranes reported by Jack (1991) andNials et al. (1993)], and only part of the 38.2% fraction disappeared in experiments with mutant receptors. Taking into account that these data were derived from saturation binding in different cell cultures with strongly varying receptor densities and normalized to the protein content of the cell preparations, they should not be taken as unequivocal evidence.
Another line of evidence casting doubt on the exo-site hypothesis is based on the failure of SLM analogs to block the exo-site binding of SLM (Bergendal et al., 1996). SLM analogs were designed and synthesized to interact with SLM at the putative high-affinity binding site of the aliphatic side chain. A reasonable assumption of this experimental approach was that a high specificity of a recognition site (exo-site), characterized by high affinity, should constitute the structural basis for competition by related structural analogs. A second line of theoretical considerations focuses on the point that a drug that binds with high affinity to a site near a receptor with a slow onset of association (Jack, 1991; Johnson, 1991) will not interact with the receptor binding sites in a simple competitive manner following the law of mass action but rather will be characterized by steep concentration-effect curves (Hill slope ≫ 1) and other nonequilibrium effects as observed in isolated tissues of guinea pig trachea (Dougall et al., 1991) and human bronchus (Anderson et al., 1994; Naline et al., 1994). Despite these conclusions drawn from the exo-site hypothesis, concentration-effect curves for AC stimulation (Fig. 9; Clark et al., 1996) and binding inhibition curves (this report; Clark et al., 1996) have Hill slopes close to 1.0.
In conclusion, we have found that in receptor membranes incubated (but not flooded) with appropriate concentrations of SLM, neither a slow onset nor a long duration of action can be found. Clinical effects of SLM such as slow onset and long duration of action over 12 h, however, distinguish it from traditional inhaled βAR agonists and qualify the drug for the maintenance treatment of reversible airway obstruction (Ullman and Svedmyr, 1988; Pearlman et al., 1992;Lötvall et al., 1994). Thus, we conclude by exclusion that partitioning of the drug in lipophilic compartments after inhalation along with its high affinity for lung β2AR underlies its long duration of action as outlined in the “diffusion microkinetic model” (Anderson et al., 1994; reviewed by Lindén et al., 1996; and Waldeck, 1996).
Acknowledgments
We appreciate the excellent technical assistance of Patricia Ohly and Frank Renner and the excellent computer work of Andreas Damek and Moritz Becker (Institute for Laser Medicine, University of Düsseldorf). We are indebted to Prof. Dr. S. Cleveland (Institute for Neurophysiology, University of Düsseldorf) for carefully reading and correction of the manuscript.
Footnotes
-
Send reprint requests to: Prof. Dr. H. Lemoine, University of Düsseldorf, Institute for Laser Medicine, Molecular Drug Research Group, Universitätsstr. 1, 40 225 Düsseldorf, Germany. E-mail: lemoine{at}uni-duesseldorf.de
-
↵1 This work was supported by the Deutsche Forschungsgemeinschaft (DFG, LE 552-1).
- Abbreviations:
- AC
- adenylyl cyclase
- BU
- (−)-bupranolol
- CGP 20
- 712 A, 1-[2(3-carbamoyl-4-hydroxy phenoxy)-ethylamino]3-[4-(1-methyl-4-trifluoromethyl-2 imidazolyl)phenoxy]2-propanol methanesulfonate
- FOR
- (−)-formoterol
- Gpp(NH)p
- 5′-guanylylimidodiphosphate
- IA
- intrinsic activity
- ISO
- (−)-isoprenaline
- SLM
- (±)-salmeterol
- AR
- adrenoceptors
- SYM
- sympathomimetics
- IPIN
- (−)-iodopindolol
- Received June 30, 1998.
- Accepted September 28, 1998.
- The American Society for Pharmacology and Experimental Therapeutics