Abstract
Activation of beta-2 adrenoceptors (BAR) in smooth muscle preparations is associated with a rapid, reversible and incomplete receptor desensitization, resulting in a steady-state relaxation response to BAR agonists. Based on results from cell culture studies, we hypothesize that, in the isolated guinea pig trachea, this steady state is a result of a concurrent resensitization of desensitizing BAR. In tracheal segments maintained at mechanical tone (4–6 g), isoproterenol (ISO) and the partial BAR agonist salbutamol (SALB) elicited a monotonic, rapid (1–3 min) and reproducible relaxation response that could be maintained for up to 45 min and was completely reversed by propranolol. Similarly, tissues preconstricted with 0.1 μM carbachol (CARB) responded with a sustained relaxation response to ISO. In contrast, in tissues preconstricted with 0.3 to 10 μM CARB or with 75 mM KCl, the relaxation elicited by ISO was followed by a slow (20–30 min) and partial restoration of muscle tone (“fade”). The relaxation and fade were observed when CARB-constricted tissues were relaxed with SALB (0.2 or 10 μM) or 10 μM salmeterol. No response to SALB was observed when tissues were preconstricted with KCl. The fade met criteria for its classification as a homologous desensitization of the relaxation response at the BAR level. In desensitized washed tissues, a complete recovery of the original relaxation response could be detected within 60 min of drug removal. A propranolol- and ICI 118–551-sensitive steady state was achieved 30 to 35 min after the addition of BAR agonists to the isolated tissues. A three-compartment phenomenological kinetic model accurately described the observed data, defining one steady-state and three rate constants, describing relaxation (k 1), desensitization (k 2) and resensitization (k 3). The values ofk 2 and k 3 for the response to SALB and to salmeterol were significantly larger than those observed for ISO. In the presence of KCl, the values ofk 2 and k 3 for the response to ISO were indistinguishable from those measured in the presence of CARB. Given the parameters defined by our model, we propose that desensitization and resensitization of BAR in the isolated guinea pig trachea are distinct concurrent processes whose net result actively maintains a sustained partial relaxation response to ISO, SALB or salmeterol. The component of resensitization in the presence of agonist may account for the clinical efficacy of inhaled BAR agonists.
Modulation of the BAR in airway smooth muscle, through desensitization and resensitization, has been well studied both in vitro and in vivo (Barnes, 1995, and references cited therein). Although clinical desensitization (tolerance and tachyphylaxis) of the long-term therapeutic effects of BAR agonists has been reported both in patients with obstructive airway disease (Repsher et al., 1984; Georgopoulos et al., 1990) and in normal human subjects (Galant et al., 1978), desensitization of BAR during a short exposure to agonists does not appear to represent a significant phenomenon. Recent data suggest that the absence of detectable desensitization may be due to the large BAR reserve in airway smooth muscle (reviewed by Barnes, 1995). A rapid fade of the relaxation response to BAR agonists in the isolated guinea pig trachea has been reported (Raper and Malta, 1973; Jeppsson et al., 1992; Linden et al., 1993); however, it was characterized as neither a desensitization nor a time-dependent process.
Much effort has been directed toward the understanding of the molecular structure and regulation of BAR in cells in culture. Hall and collaborators performed second messenger measurements of the desensitization of the BAR in cultured human airway smooth muscle myocytes (Hall et al., 1992, 1993; Hall and Kotlikoff, 1995). Furthermore, hypotheses based on these cellular models have been tested in tissues from multiple organs and species, including rabbit aorta (Keitz et al., 1990), guinea pig stomach fundus (Ballet al., 1991), human bronchi (Nials et al., 1993) and guinea pig lung parenchyma (Zaagsma et al., 1983). The high density of BAR in the guinea pig trachea makes this tissue a sensitive model for the study of BAR agonist activity and for the modulation of airway responses to these drugs (van der Heijden et al., 1984; Fernandes et al., 1988).
A resensitization of desensitizing BAR at the cellular level has been described (Yu et al., 1993; Pippig et al., 1995), but the contribution of the resensitization process to the clinical efficacy of inhaled BAR agonists has yet to be demonstrated. In the current study, we hypothesized that, after BAR activation, a steady state was established between desensitization and resensitization of BAR in the isolated guinea pig trachea. We used a three-compartment phenomenological model to assess the kinetics of these processes and proposed that desensitization and resensitization of BAR were distinct concurrent events whose net result actively maintained a sustained partial relaxation response to BAR agonists.
Methods
Isolated guinea pig trachea.
Male Hartley guinea pigs (300–500 g) were sacrificed, by CO2 asphyxiation, up to 16 hr before an experiment, and the entire trachea from larynx to carina was rapidly excised. The trachea was immediately cleaned of surrounding muscle and connective tissue and was placed into Krebs-bicarbonate buffer saturated at low temperature (0–4°C) with 95% O2/5% CO2. The trachea remained in this buffer at 4°C until it was used. At that time, each trachea was divided into four approximately equal segments. In some experiments, the tracheal epithelium was mechanically removed with a cotton-tipped applicator; epithelial removal was confirmed histologically. The four tracheal segments were suspended between stainless steel hooks in 20-ml organ baths. The baths contained Krebs-bicarbonate buffer, through which was bubbled 95% O2/5% CO2 to maintain a pH of 7.4 ± 0.1 at 37 ± 1°C. The tissues were initially set to 2 g of mechanical tension (“basal tone”) and the buffer was drained and replaced (“washed”) a minimum of three times over a period of not less than 1 hr, with periodic readjustment to 2-g tension. After the tissues had been stabilized at 2-g tension, they were primed by eliciting a contraction with 10 μM CARB. Drugs were removed from the bath by replacing the Krebs buffer (washing).
Solutions.
Krebs-bicarbonate buffer contains the following components (in mM), in glass-distilled water: NaCl, 110; KCl, 5; MgSO4, 1.2; CaCl2, 2.35; KH2PO4, 1.2; NaHCO3, 25; glucose, 11. A modified Krebs solution, with 40 mM NaCl and 75 mM KCl, was used to elicit a contractile response to KCl. All drug solutions were prepared in glass-distilled water on the day of the experiment. The cumulative volume of added drugs never exceeded 300 μl or 1.5% of the organ bath volume.
Sequential responses to BAR agonists.
The relaxation response to (−)-ISO or to (±)-SALB was elicited and assessed with tissues maintained at basal tone and with tissues preconstricted with CARB or KCl. A typical protocol is illustrated in figure1. After the 1 μM CARB-elicited constriction reached a plateau, 130 nM ISO was added to the organ bath. A rapid (over 350 sec) decrease of the tissue tone (relaxation) was followed by a slower (over 1500 sec) partial reestablishment of the tone (fade; desensitization). After the tissue tone had reached a steady-state level, the tissues were washed with fresh Krebs buffer. During the following 1 hr, no drugs were added, and the tissues were washed three to six times to completely remove the drugs from the tissue and to allow for a full recovery of tissue responsiveness. This protocol allowed for several reproducible assays to be conducted on the same tracheal segment. The magnitude of the first response of a given tracheal segment to BAR agonists was consistently smaller than that of the subsequent responses. Therefore, all data in this study were obtained from tissues that were treated twice, with 1) exposure to 1 μM CARB and, 2) after 20 to 30 min of wash, exposure to 1 μM CARB followed by the saturating concentration of the tested BAR agonist. With this protocol of pretreatment of the tissues, the subsequent responses of tracheal segments to BAR stimulation were consistent and reliable and could be used for multiple assays for >12 hr.
Collection of data points.
Isometric contractions were measured with Grass FT03C linear force-displacement transducers connected to a Grass polygraph (model 7D; Grass Co., Quincy, MA). Signals from the amplifier driver were fed through a universal interface module (UIM100; BIOPAC Systems, Inc., Goleta, CA) to a 16-channel digital-to-analog converter and data acquisition unit (MP100; BIOPAC). The signals were then processed on an Apple Macintosh SE with the software package AcqKnowledge 3.0 (BIOPAC) to control data acquisition and handle file management commands. Data points corresponding to tension were collected at a frequency of 1 Hz. Digital values were transferred to an Apple Macintosh Power PC 7100/66 or 7100/80 for nonlinear regression analysis with KaleidaGraph (Synergy Software, Reading, PA).
Data analysis.
Figure 1 illustrates the time course of the response to 130 nM ISO of a trachea segment preconstricted with 1 μM CARB. We considered several kinetic models and chose the simplest one that could describe the observed kinetics. The following is the three-compartment phenomenological model:
Determination of time 0.
The kinetic model of the response (eq. 2) describes most of the observed time course of the response. However, it is not applicable to a short time period at the start of the relaxation response during which there is a lag of the response, which begins at the time of BAR agonist addition and ends at the inflection point of the initial relaxation response (see fig. 9). The time lag, which is not considered by the three-compartment model, may be caused by drug diffusion and by an initial nonhomogeneous distribution of the tested BAR agonist in the tissue. This discrepancy between the model and the observed data required that the theoretical curve be fitted to the observed data without the time lag; also, instead of using the actual time of addition of a drug as time 0, a corrected time, t0, was used to produce the best fit. Thus, an additional parameter was introduced with A(t),B(t) and C(t), i.e., telap. For example, eq. 3 shows the B(t) function with the inclusion of the telap parameter.A(t) and C(t) functions were similarly altered. In each assay, the collection of data points started a few seconds before the addition of the constrictor (CARB or KCl) to the organ bath. The value of telap was defined as the time period that started with this addition and ended at t0, such that, at t0, t − telap = 0 (see figs. 1 and 9 and “”).
At infinite time, Rss (fig. 1) is:
Statistics.
The following statistical methods were used: 1) cumulative frequency plot of the residuals (e.g., fig. 10) to assess the goodness of the fit, 2) χ2 test of the residuals, 3) Kruskal-Wallis H test (Mendenhall et al., 1986), 4) linear regression analysis and 5) nonlinear regression analysis (KaleidaGraph; Synergy Software, Reading, PA). A short description of these methods is found in the “.”
Drugs.
(−)-ISO, CARB, (±)-SALB, epinephrine, (±)-PROP, forskolin, NECA, (±)-metaproterenol and indomethacin were obtained from Sigma Chemical Co. (St. Louis, MO); (±)-atenolol and ICI 118–551 were obtained from RBI (Natick, MA). SALM was generously donated by the Glaxo Pharmaceutical Company. All other chemicals were of the highest grade available.
Results
BAR agonists elicited a sustained, PROP-sensitive, relaxation response in tracheal segments maintained at basal tone.
When tested on a mechanically applied basal tissue tone of 2 g, 130 nM ISO elicited a robust relaxation response that completely eliminated this tone (n = 36 tissues). Therefore, we tested the relaxation response to this concentration of ISO on tissues set to 4 to 6 g of mechanical tension. Figure 2A illustrates digitized data collected from one tissue. Addition of 130 nM ISO elicited an initial rapid relaxation that was completed within 500 sec, followed by a further, slower loss of tissue tone (500–2000 sec). The relaxation response was assayed for up to 45 to 60 min. The original basal tone was restored after removal of ISO from the organ bath by washing. Addition of 3 μM (±)-PROP to an ISO-relaxed tissue reversed the relaxation response (fig. 2A). The relaxation response could be reelicited three to five times over 10 to 12 hr in a given tissue without a detectable loss of the relaxation magnitude. Consistently, within a range of mechanically applied tone intended to mimic the range of tone inducible by 1 μM CARB (up to 7 g), there was no fade of the response to ISO within the 45- to 60-min assay. Similar sustained, PROP-sensitive relaxation responses were elicited from tissues maintained at basal tone by other BAR agonists, such as 10 μM SALB and 20 μM (±)-metaproterenol (data not shown).
In contrast to experiments on tissues whose tone was increased by mechanical stretching, the relaxation response to BAR agonists underwent an incomplete fade when elicited on tissues preconstricted with CARB or with KCl.
Cumulative concentration-response curves for CARB yielded two response parameters, i.e., EC50 (pEC50 = 6.56 ± 0.25,n = 4; EC50 = 0.31 μM) and slope index (0.96 ± 0.2, n = 4) (for definition of the slope index, see Clancy and Maayani, 1985). Similarly, 75 mM KCl was found to elicit a maximal contractile response.
At 1 μM (3 × EC50) CARB, a steady-state level of constriction was elicited within 7 to 9 min (fig. 1) and was stable for >60 min. Addition of 130 nM ISO at the plateau of the CARB-induced constriction resulted in a rapid loss of the tissue tone (relaxation). In contrast to the sustained relaxation observed in tissues relaxed from applied mechanical tone (fig. 2A), the relaxation elicited on preconstricted tissue was followed by a partial restoration of tissue tone (fade). The completion of the fade required a time period 5 to 10 times longer than that needed for the completion of the relaxation response (fig. 1). At the end of the fade, addition of 3 μM PROP to the organ bath reversed the remaining relaxation and reestablished the original level of the constriction elicited by CARB (fig. 2B). A response similar to that elicited by ISO, i.e., a relaxation followed by an incomplete fade, was induced by other BAR agonists, such as 10 μM SALB, 10 μM SALM (see below) or 20 μM metaproterenol (n = 10), on trachea segments preconstricted with 1 μM CARB. The relaxation response to 1 μM ISO assayed on tissues preconstricted with 75 mM KCl displayed a similar fade (see below).
In control experiments, several sequential relaxation responses to 130 nM ISO were elicited on the same tissues in alternating order, first preconstricted tissue and then mechanical tone and vice versa. Consistently, within the time frame tested (up to 45–60 min), no fade followed the relaxation elicited on the mechanical tone, whereas a robust fade followed the relaxation of the preconstricted tissue (n = 32 tissues from eight animals).
The relaxation response and its incomplete fade are mediated through the BAR subtype.
The receptor that mediates a relaxation response in mammalian airways (Barnes, 1995), including guinea pig trachea (Jeppsson et al., 1992; Kallstrom et al., 1994), has been classified as the BAR subtype. Because of antagonist interference with the kinetics of agonist action (Clancy et al., 1987), we confirmed this classification with two selective BAR partial agonists, SALB and SALM (beta-2/beta-1 = 1,375 and 85,000, respectively; reviewed by Johnson and Coleman, 1995). We also characterized interactions between the responses to these drugs; the relaxation response to both drugs was followed by a partial fade (fig. 3). After a complete response to 10 μM SALM (relaxation and partial fade), subsequent addition of 10 μM SALB did not elicit a relaxation response (n = 6; data not shown), whereas a robust response was elicited by this concentration of SALB in the absence of SALM (fig. 3). This response indicates that SALB and SALM address the same site of action. Finally, the response to an EC50 concentration of the nonselective BAR agonist ISO (200 nM) was abolished in tissues preexposed to 10 μM SALB (fig.4). Taken together, these results are consistent with the hypothesis that the contribution of the beta-1 adrenoceptor to the observed kinetics of the response to the tested BAR agonists is negligible.
The fade met criteria for its classification as a homologous desensitization at the BAR level.
Seven experiments were performed to characterize the nature of the fade. 1) We first tested the possibility that the fade was a result of a gradual, time-dependent formation of a stable BAR “antagonist.” The Krebs solution was rapidly transferred at the completion of a relaxation and a fade into another organ bath containing a preconstricted tissue whose solution had been drained (bath replacement experiment). The second tissue developed a response to ISO (relaxation followed by a fade) with a magnitude consistent with control experiments in which the tissue was challenged with CARB and ISO in the standard fashion (n= 3). 2) We tested whether the added ISO was eliminated during the fade and, if so, whether this elimination was responsible for the partial restoration of tissue tone. A second addition of the agonist to an organ bath, after the full relaxation and fade response of a tissue to an initial exposure, did not elicit a second relaxation response (fig.5). Of note, the protocol illustrated in figure 5 called for saturating concentrations of CARB (30 μM) as well as ISO (30 μM). This study was repeated multiple times (n = 36), varying the BAR agonist [i.e., (−)-epinephrine, SALB or metaproterenol]. 3) We tested the possibility that the fade resulted from the release of a stable constrictor that functionally antagonized the stable relaxation response. Addition of 3 μM PROP reversed the remaining relaxation only to the original constriction level, implying the absence of a stable constrictor and confirming the presence of intact and functioning BAR (fig. 2B). Similar results were observed with ICI 118–551 (fig. 6) (Bilski et al., 1983). It should be noted that 1 μM CARB (3 × EC50) did not elicit the maximal contractile response of the tissue, because other constrictors, such as 100 μM histamine, or higher concentrations of CARB could elicit additional constriction on top of the existing CARB response. 4) Addition of either 10 μM forskolin (n = 3; data not shown) or 10 μM NECA (an adenosine A2 agonist) (fig. 5) to ISO-exposed tissues, after the relaxation and fade, resulted in additional relaxation responses consistently similar to those of tissues not exposed to ISO. This response implies that the relaxation elicited by activation of adenosine A2 receptors with NECA or by direct activation of adenylyl cyclase with forskolin was not altered after the fade of the response to BAR agonists. 5) Mechanical removal of the epithelial cells from the trachea segments (n = 9; see “Methods”) did not affect the qualitative or time-dependent characteristics of either the relaxation or the fade, indicating that these responses do not depend on the presence of intact epithelial cells. 6) Similarly, 30-min incubation with 1 μM indomethacin did not affect the qualitative characteristics of either the relaxation or the fade (n = 8), indicating that these responses are not completely dependent on prostaglandin synthesis. Of interest, both removal of epithelium and inhibition of prostaglandin synthesis did modulate quantitative aspects of the relaxation and fade responses to ISO. These findings are currently under investigation. 7) Cross-reactivity between BAR agonists was investigated through the addition of a second BAR agonist to the organ bath at the end of a fade caused by a first BAR agonist. Neither ISO (30 μM) nor SALB (30 μM) elicited a second relaxation response after an initial relaxation and fade response to 30 μM ISO (fig. 5). We tested several combinations of two BAR agonists (ISO, SALB and epinephrine), all with similar qualitative results (fig. 5). This apparent cross-reactivity between all tested BAR agonists is consistent with the fact that these drugs share general pharmacological characteristics, are thought to act through the same receptor and should therefore be similar with respect to their ability to elicit a relaxation and fade response.
A phenomenological kinetic model accurately describes the observed data.
A phenomenological model (see “Methods”) enabled us to fit three kinetic parameters (k 1,k 2 and k 3) and a steady-state parameter (Tmin) to the observed data. The kinetic parameters allowed for the assessment of Rss (eq.4). Shown in figure 1 are data of the response of one tissue segment to 1 μM CARB, followed by 130 nM ISO. Superimposed upon the digitized data is a computer-generated nonlinear regression fit to eq. 11 (see “”). Using the three-compartment model, the time-dependent change from state A, a state of tissue constriction, to state B, a state of tissue relaxation, and state C, a loss of the relaxation, was simulated from the fitted values of the observed response (see legend to fig. 1). The initial rapid elimination of state A occurs concurrently with a slower formation of states B and C, whose steady state is described by Rss. The ratio Rss/Rmax represents the fraction of the response that remains active at steady state. Table 1summarizes the mean values (and their standard deviation) of these parameters collected from 89 tissues from 32 animals.
Despite the same stimulus for constriction (1 μM CARB), a large variation was observed across the 89 trachea segments tested (Tmax = 1.0–6.5 g; mean, 2.74 ± 1.04 g). Similarly, the calculated value of the relaxation response to 130 nM ISO (Rmax) spanned a wide range (1.03–8.98 g; mean, 3.36 ± 1.03 g). Several intriguing correlations were observed between the values of Tmax and both the values of the three kinetic parameters and the values of some steady-state parameters (see table 1, footnote). The rate constant of the relaxation, k 1, decreased with the increase in Tmax, whereas the values of both k 2and k 3 increased with Tmax. Of note, the slope of the linear regression analysis fork 2 was about 9 times larger than that of thek 3 values. The functional antagonism between the stimulus for constriction and that for relaxation, as assessed by the ratio Rmax/Tmax or Rss/Tmax decreased with an increase in Tmax values. Furthermore, Rmax frequently exceeded Tmax for those tissues exposed to ISO. This finding reflects the ability of ISO to induce a relaxation response both in tissues preset to a mechanically applied level of basal tone and in those actively preconstricted with CARB or KCl (fig. 2). Of note, Rmax never completely consumed both the preset mechanical constriction of 4-g basal tone applied to all tissues and the Tmax (Rmax < Tmax + 4).
The PROP- and ICI 118–551-sensitive Rss was 21 ± 9.4% of Rmax. It is proposed to represent an equilibrium between the simultaneous processes of desensitization and resensitization in the continued presence of BAR agonist.
The interaction between the stimulus for constriction and the kinetics of the response to a BAR agonist was investigated by altering CARB concentrations.
Figure 7 illustrates sequential responses of a single tissue to a saturating concentration (10 μM) of SALB after constriction with several CARB concentrations (0.1–10 μM). Constriction with CARB concentrations below the EC50 (e.g., 0.1 μM) did not support partial desensitization; in contrast, CARB concentrations at or above the EC50 altered the rate and magnitude of desensitization and resensitization. Consequently, the steady-state portions of the curves were reached more rapidly at higher CARB concentrations. Similar results were observed with 3 μM ISO (n = 3; not shown). The same sequential addition of KCl could not be performed due to the lack of a graded constrictor response to this compound.
The kinetics of desensitization and resensitization of the response to the partial BAR agonists SALB and SALM were faster than those observed with the response to ISO.
Figure 3 illustrates sequential responses of a single, preconstricted (3 μM CARB) tracheal segment to saturating concentrations of ISO (10 μM), SALB (10 μM) and SALM (10 μM). In another set of experiments (data not shown), the kinetics of the response to ISO and to SALB were assessed at concentrations of 1 × EC50 (130 nM and 200 nM, respectively) on tissues preconstricted with 1 μM CARB. A summary of mean values of the kinetic and steady-state parameters for ISO and SALB is presented in table 2. The partial agonist nature of SALB vs. that of ISO was confirmed at both concentrations as tested by the calculated ratio between values of the steady state parameters, Rmax (0.39 and 0.61 for the EC50 and saturating concentrations, respectively) (table 2), Rss (0.67 and 0.32) and Rmax/Tmax (0.56 and 0.59). By using the Kruskal-Wallis H test (Mendenhall et al., 1986), we found that the rate constant of relaxation (k 1) was statistically indistinguishable between the two drugs. In contrast, compared with ISO, and at both concentrations, the values of the rate constants of desensitization (k 2) and resensitization (k 3) were significantly larger with SALB. These results were confirmed by studies with SALM; at saturating concentrations, both the partial agonist nature of SALM (calculated ratios, Rmax = 0.44, Rss = 0.48 and Rmax/Tmax = 0.36) and its larger values ofk 2 and k 3, compared with ISO, were demonstrated (fig. 3).
Preconstriction of guinea pig trachea with KCl slowed the relaxation, compared with that with CARB, but did not change the kinetics of desensitization and resensitization of the response to ISO.
Potassium chloride at 75 mM elicited a smaller contractile response (2.02 ± 0.44 g; n = 11), compared with that observed with 1 μM CARB (2.93 ± 0.86 g;n = 10) (table 3). The kinetics of the response to 1 μM ISO were assessed on tissues preconstricted sequentially with CARB and KCl (fig. 8). The rate constant of the relaxation response to ISO (k 1) with 75 mM KCl was 3.7-fold smaller than that observed with 1 μM CARB (table 3). In contrast, the rate constants of desensitization (k 2) and resensitization (k 3) were statistically indistinguishable, according to the Kruskal-Wallis H test. The calculated Rmax for ISO elicited in the presence of KCl (2.69 g) was significantly smaller than that observed in the presence of CARB (4.69 g), despite the somewhat smaller constriction elicited by KCl. Of interest, even maximal concentrations of the partial agonist SALB were not sufficient to functionally antagonize the KCl-induced response; no response to 10 μM SALB was observed on tissues preconstricted with 75 mM KCl. Attempts to use lower concentrations of KCl to optimize this functional antagonism with SALB resulted in erratic and unreliable responses.
Discussion
The results of this study allow us to tentatively identify and, using a phenomenological model, to quantify the kinetics of three concurrent processes elicited by three BAR agonists in the isolated guinea pig trachea, i.e., a relaxation response, its short-term desensitization and its resensitization. Desensitization is defined as a decreasing response to a constant stimulus, assessed at either the cellular or tissue level; the clinical terms are tachyphylaxis or tolerance (Gershengorn, 1994). Resensitization is defined as the restoration of the desensitized responsiveness, either in the continued presence or in the absence of the stimulus. Molecular details of BAR desensitization and resensitization have been elucidated in cells in culture (Perkins et al., 1991, and references cited therein), including the kinetics of the response to BAR activation (Shear et al., 1976), desensitization (Suet al., 1980) and resensitization (Kurz and Perkins, 1992;Yu et al., 1993; Pippig et al., 1995). Cellular studies of desensitization of BAR in cultured human airway smooth muscle myocytes were reported by Hall and collaborators (Hall et al., 1992, 1993; Hall and Kotlikoff, 1995). These studies of the underlying biochemical mechanism of BAR desensitization appear to provide information similar to that reported in established cultured cell lines. Some of these cellular hypotheses involving desensitization and resensitization of BAR have been addressed in isolated tissues (Fernandes et al., 1988), but the counteracting dynamic nature of these events has not been previously reported in isolated tissues. We have created a kinetic model for the analysis of the short-term response to BAR agonists in the isolated guinea pig trachea. In addition to the kinetics of the relaxation response, we assessed those of its desensitization and resensitization. The results of this analysis appear to support the notion that, in the ongoing presence of BAR stimulus, there is a concurrent resensitization of desensitizing BAR. We propose that the protocol used in this study, coupled with the phenomenological model used for data analysis, are powerful tools for assessing the kinetics of these and other overlapping responses in tissue preparations.
The guinea pig trachea preparation has been widely used as a model to study the physiology and the pharmacology of BAR agonists in vitro (Castillo and De Beer, 1947; Watanabe et al., 1976; Douglas et al., 1977; Fernandes et al., 1988). The high density of BAR in the guinea pig trachea (Zaagsmaet al., 1984) makes it a sensitive model for the study of BAR and its modulation by BAR agonists. Furthermore, the ability of BAR in the isolated trachea to undergo short-term desensitization offers an integrated system whereby cellular events can be probed at the level of a relatively intact organ system (van der Heijden et al., 1984; Fernandes et al., 1988).
We described the relaxation, desensitization and resensitization responses by a phenomenological kinetic model, consisting of states A, B and C (see “Methods”). We have tested several more complex models, such as the one in which state A converts to state B in a reversible manner and in which state C converts irreversibly to state A. However, the model we have chosen (eq. 1) not only is the simplest one but also is an identifiable model, in which the three parameters proposed (k 1, k 2 andk 3) may be identified with known cellular processes. In contrast, more complex models define large sets of adjustable parameter values that may not identify specific cellular processes. Currently, we are performing additional experiments using more complex protocols to exclude these alternative models.
We propose that the fade of the relaxation response to BAR agonists is a receptor-mediated event, consistent with desensitization of BAR. A series of control experiments supports this proposition. There was neither formation of a tentative stable antagonist or a constrictor nor elimination of the BAR agonist during the fade. Although both removal of epithelium and inhibition of prostaglandin synthesis failed to eliminate desensitization, these manipulations did alter some characteristics of the response. Thus, the hypotheses that the epithelium (Goldie et al., 1986) or de novosynthesis of prostaglandins (Berti et al., 1982; Ominiet al., 1984) contributes to the desensitization response could not be rejected. The desensitization was homologous; relaxation responses to both NECA (fig. 5) and forskolin were not altered by BAR desensitization. Desensitization induced by one BAR agonist resulted in cross-desensitization to other BAR agonists (van der Heijden et al., 1984) (figs. 4 and 5). Finally, BAR selectivity was confirmed. We previously reported on the interference of antagonist kinetics with those of agonists (Clancy et al., 1987). Consequently, we have currently confirmed this classification by testing two selective BAR agonists, SALB and SALM (see “Results”). Under the selected experimental conditions, SALB appeared to activate the BAR and not the beta-1 adrenoceptor, because the proportion of beta-1/beta-2 binding sites in airways is less than 1:4 (Barnes, 1995). Furthermore, there is no receptor reserve for the partial agonist SALB on the BAR. Thus, if the ratio between the binding sites reflects the true ratio between the densities of the two active receptor subtypes, then 200 nM SALB and 10 μM SALM activated 1,375 and 85,000 × 4-fold more BAR thanbeta-1 adrenoceptors, respectively. However, we cannot rule out the contribution of beta-1 adrenoceptors in the observed response to ISO. The lack of a response to the EC50concentration of ISO (200 nM) in the presence of 10 μM SALB (fig. 4) supported the notion that the response to ISO was also mediated through activation of the BAR.
Several lines of evidence, at both the cellular and tissue levels, are consistent with the hypothesis that the short-term desensitization of responses to BAR agonists is incomplete, despite full receptor occupancy. Our kinetic studies on isolated rabbit aorta (Keitz et al., 1990; Osman et al., 1990) and on guinea pig trachea (this study) have demonstrated that the initial relaxation response to a saturating ISO concentration undergoes a slower partial desensitization, culminating in a PROP-sensitive (fig. 2A) and ICI 118–551-sensitive (fig. 6) steady state. In both tissues, we found that removal of ISO restored the original response within 60 min, confirming the work of others with guinea pig trachea (Fernandeset al., 1988; Herepath and Broadley, 1992). Similarly, short-term, partial, reversible desensitization of the response to BAR agonists has been demonstrated in cells in culture (Perkins et al., 1991). There is a reported range of 50 to 55% (Su et al., 1980; Kelsen et al., 1995) to 82% (Mukherjeeet al., 1976; Fernandes et al., 1988) loss in the ability of the cells to respond to ISO stimulation after incubation with ISO, a process that is fully reversible after removal of agonist (Yu et al., 1993; Pippig et al., 1995).
Although our study does not explore the molecular nature of BAR desensitization or resensitization, our data do support the general notion of concurrent resensitization of desensitizing BAR. The strongest evidence for the existence of the resensitization process is based on the properties of Rss. The initial, unstable relaxation response to both ISO and SALB (k 1 and Rmax) evolved into a smaller, more stable relaxation response (Rss). The properties of both responses were characteristic of BAR activation; however, the initial response partially desensitized, whereas Rss did not. Desensitization of the initial relaxation response was incomplete, regardless of the level of BAR occupancy for both tested BAR agonists (Rss/Rmax values) (tables 1–3). We propose that there is a counteracting process that prevents full desensitization of the response, allowing for 18 to 50% activity at steady state. We further propose that the short-term desensitization of BAR in the isolated trachea is incomplete due to a concurrent, balancing, resensitization process that continuously restores the responsiveness of the desensitizing BAR.
The kinetic parameters summarized in table 1 may be expressed as half-life values (t1/2 = ln 2/k) and thus may be compared with t1/2 values reported for cells in culture, where the time course of cyclic AMP formation in cells is a measure of the kinetics of the response to BAR agonists. The fastest response in the trachea is the relaxation (k 1), with t1/2 = 1.7 min. This t1/2 value is similar to that found in B50 neuroblastoma cell monolayers (t1/2 = 0.44 min) (McCrea and Hill, 1993) and in S49 mouse lymphoma cells (Shear et al., 1976). The kinetics of the desensitization (k 2) of the relaxation response in the trachea (t1/2 = 4.3 min) are similar to reported kinetics in cells (t1/2 = 3 min) (Su et al., 1979, 1980; Waldo et al., 1983). However, resensitization (k 3) of desensitized BAR, elicited in cells by removal of the BAR agonist, proceeded with t1/2 of 3 to 7 min (Su et al., 1980; Yu et al., 1993). Kurz and Perkins (1992) described similar kinetics of receptor externalization, in both the presence and absence of agonist (t1/2 = 3–4 min). These kinetics are different from those in the isolated trachea, in which t1/2 is 28 min. Thus, although the kinetics of relaxation and desensitization of BAR in the trachea appear to be similar to those reported in cells, the resensitization of BAR in cells in culture displayed faster kinetics than seen in the isolated trachea. This discrepancy may be explained by an interaction that occurs between receptors and their effectors in the intact tissue system, accelerating or slowing various processes, in contrast to single cells. We propose, however, that, despite these different kinetics, the effective resensitization of desensitized BAR is an intrinsic property of this receptor in both systems.
Each of the three kinetic parameters (eq. 1) may be related to a subset of biochemical processes elicited by BAR activation. The subset of processes defined collectively by k 1 may include rates of events such as cyclic AMP production and opening of K+ channels (Lohse, 1993; Barnes, 1995, and references cited therein). Similarly, k 2 may represent a series of cellular events implicated in BAR desensitization, such as kinase-activated receptor phosphorylation by protein kinase A and/or BAR kinase (Lefkowitz and Williams, 1978; Benovic et al., 1986; Pitcher et al., 1992), β-arrestin binding to the phosphorylated receptor (Pippig et al., 1993) or sequestration of phosphorylated receptors to internal membrane compartments (Lohse et al., 1990; Lohse, 1993), which is associated with a diminished adenylyl cyclase response. This agonist-induced sequestration may be the first step in the set of events characterizing k 3, the third rate constant. The internal vesicles are rich in phosphatases that dephosphorylate BAR and reactivate them. They are then recycled to the cell membrane (Barak et al., 1994; Pippig et al., 1995), a process that has been shown to occur even in the continued presence of agonist (Kurz and Perkins, 1992). Furthermore, there is good temporal correlation between the recycling of internalized BAR back to the plasma membrane and the recovery of ISO-stimulated adenylyl cyclase activity (Garcia-Higuera and Mayor, 1994).
Thus, the steady state that we observe (Rss) may reflect two simultaneous competing cellular processes, i.e., receptor phosphorylation (short-term desensitization) and receptor sequestration and dephosphorylation (resensitization) (Yu et al., 1993). We propose that these events overlap in time (fig. 1) and that, by applying a phenomenological model, they may be separated and quantified. We previously attempted to implement a phenomenological kinetic model for the analysis of responses to ISO in isolated rabbit aorta (Keitz et al., 1990; Osman et al., 1990). However, the Keitz model assumes a temporal separation, not overlap, between the relaxation and its desensitization. In addition, the Keitz model does not allow for characterization of the resensitization process.
The three-compartment phenomenological model that we describe here addresses these outstanding elements in the description of the observed data, namely the temporal overlap of the three counteracting processes, and allows for assessment of three kinetic and one steady-state parameter (fig. 1). The differential kinetics of the three processes (k 1 > k 2 >k 3) (tables 1–3) resulted in the observed unique time course of the response to ISO (fig. 1) and to SALB (fig.3). A comparison of the t1/2 values for each of these processes, as well as the relative values of the kinetic parameters (e.g., for the response to ISO,k 1/k 2 = 2.64 andk 2/k 3 = 5.50;n = 56) (table 1), indicate that there is, indeed, temporal overlap between them. If this is true, then the magnitude of the observed relaxation response is an underestimate of the correct measure of the potential full relaxation response to ISO stimulation [(Tmax − Tmin)/observed relaxation = 1.7] (fig. 1). The efficiency of the resensitization processes is demonstrated by the high value of the ratio Rss/Rmax that is derived from the values ofk 2 and k 3 (eq. 4). Under all experimental conditions tested in this study (three different concentrations of ISO, two of SALB, two of CARB and one of KCl), Rss varied from 18 to 44% of Rmax (tables 1–3).
Experimental conditions required for detecting BAR desensitization in the isolated trachea appear to be different from those described for cells in culture. In the cell studies, BAR desensitization has been elicited by incubation with a BAR agonist alone. In the isolated trachea, however, desensitization of the response to BAR agonists could be measured only when a constriction was elicited through activation of the muscarinic receptor by CARB at concentrations above its EC50 (fig. 7) or through membrane depolarization of the tissue by 75 mM KCl (fig. 8), and it failed to occur in their absence (fig. 2A) (van der Heijden et al., 1984). Furthermore, thek 2/k 3 ratio may be altered by an interaction between the desensitization response pathways and those of active constriction, because we observed that gradual increases in the CARB stimulus for constriction affected both desensitization and resensitization, as well as progressively decreasing the Rmax/Tmax values (fig. 7). A possible explanation for the apparent link between constriction and the phenomena of desensitization and resensitization may be that cellular events linked to constriction may slow the extremely efficient resensitization mechanism (increasek 2/k 3) and thereby enable the assertion of an otherwise “masked” desensitization process.
Activation of various kinase families, such as protein kinase C, may support the process of desensitization (Barnes, 1995, and references cited therein). If graded muscarinic activation leads to graded kinase activation, then this finding may aid in explaining the incremental nature of the kinetics of desensitization and resensitization observed with increasing CARB concentrations (fig. 7). We propose that the set of events represented by k 2 andk 3 are modulated by the cellular events induced by activation of CARB-linked signaling mechanisms. We further propose a common pathway for CARB and KCl, because they both elicit constrictions that promote a desensitization of the subsequent relaxation response to BAR agonists. Although we currently cannot identify this shared signal, it is possible that an increase in free calcium may have a modulating effect on the desensitization-resensitization process.
The three-compartment model, in addition, may be used to characterize the responses of different BAR agonists. Both kinetic and steady-state parameters were modified in the presence of either partial or full BAR agonists. As shown in figure 3, whereas the partial agonist nature of SALB and SALM were confirmed (Rmax and Rmax/Tmax), the kinetics of desensitization and resensitization of the responses to these partial BAR agonists were, on average, 1.5 to 4 times faster than those of the response to ISO at all tested occupancy levels. We propose that these differences may be explained by an agonist-influenced alteration in the spatial arrangement of the receptor, which, in turn, modifies the characteristics of BAR desensitization. Weiland et al.(1979) studied the effect of temperature on the competition for binding of BAR antagonist by BAR ligand. They demonstrated an increase in the affinity of full and partial BAR agonists with decreasing temperature; furthermore, full agonists displayed a larger increase in affinity than partial agonists [e.g.,Kd (37°C)/Kd (1°C) = 23.3 and 5.3 for ISO, a full agonist, and soterenol, a partial agonist, respectively]. Binding competition protocols were also applied by Kent et al. (1980), who determined the ratio between the high-affinity state (K H) and low-affinity state (K L) dissociation constants of BAR agonists. This ratio,K L/K H, was noted to positively correlate with the intrinsic activity of the agonist, as assessed by the BAR-linked adenylyl cyclase activity, although the individual K L and K Hvalues did not. Gether et al. (1995) labeled purified BAR with a fluorescent probe and found that the full BAR agonist ISO elicited a concentration-dependent decrease in fluorescence from the labeled receptor. In addition, the investigators described a linear correlation between efficacy of BAR agonists and the change in fluorescence of labeled BAR. They concluded that ligand-specific conformational changes occur in this G protein-coupled receptor. Despite the various methodologies, these studies support the notion that full and partial agonists differentially affect the spatial arrangement of BAR. We propose that these molecular changes may influence the characteristics and kinetic properties of the agonist-induced activation and desensitization of BAR. Although the nature of the change in conformation or affinity state has yet to be elucidated, the alteration of the receptor structure induced by the particular agonist may explain our observed change in kinetic parameters.
In summary, the response to BAR agonists in the preconstricted isolated guinea pig trachea preparation is proposed to comprise at least three concurrent processes. Characterization of these processes with a three-compartment phenomenological kinetic model suggests that an efficient resensitization accompanies the desensitization of the relaxation response to these drugs. We further propose that isolated preparations, such as the guinea pig trachea, provide a system that bridges investigations of receptor desensitization and resensitization on cellular and in vivo levels.
Acknowledgments
We thank Dr. B. J. Ebersole and Dr. J. B. Eisenkraft for their thorough and insightful comments. We also thank Dr. C. Bodian and Dr. C. Berger from the Department of Biomathematical Sciences for the stimulating discussions on the statistical analysis of the data and Margaret Roberts for excellent secretarial skills in the preparation of this manuscript.
Three-compartment kinetic model.
Figure 1 illustrates a typical biphasic response to 130 nM ISO of a trachea segment preconstricted with 1 μM CARB. We use a three-compartment phenomenological model described in “Methods” to characterize these responses. The model can be described by two coupled, first-order, differential equations and an equation that constrains the total number of units
Estimation of the time lag.
The observed relaxation response during the first few seconds after the addition of ISO displays a short but consistent time lag before a monotonic relaxation response develops (fig. 9; also see “Methods”). The change in tone over the time lag is always small, in comparison with the change in tone over the entire relaxation (<30% of the y-axis change). Also, the number of time points included in the time lag is smaller than the remainder of the analyzed response, e.g., t = −40 to +30 sec vs. t = +31 to +1848 sec (fig.1). We tested whether over- or underestimation (visually) of the lag time, where a range between 0 and 75% of the y-axis change of the relaxation is eliminated from analysis as time lag, affected the fitted values of the four parameters (k 1,k 2, k 3 and Tmin). When the time lag was increased, e.g., up to 125 sec (75% of the relaxation y-axis change) for the data shown in figure 1, there was no more than a 5% change in any of the fitted values of these parameters or in the estimated error of the parameters. In contrast, an underestimation of the time lag introduced a substantial error in each of the fitted parameters, as well as up to a 35% change in the values of the fitted parameters.
Criteria for rejection of data sets.
Responses to BAR agonists were described by our model in approximately 80% of cases (e.g., 59 acceptable fits of 75 total responses). In these cases, there is good agreement between observed and calculated kinetics. One can usually assess the goodness of a curve fit either visually or by means of statistical tests. Statistically, a curve fit is considered good if the residual values are normally distributed,i.e., the deviations between the observed data and calculated values can be attributed to the experimental uncertainty in the data. In this analysis, the goodness of the fit was tested by means of the cumulative frequency plot of the residuals and the χ2 test of the residuals (Straume and Johnson, 1992). When a curve fit passes these tests, the estimated parameters are retained for further analysis; in the case of a bad fit, the estimated parameters are rejected as outliers.
Figure 10 shows examples of acceptable and unacceptable fits; figure 10, insets, shows cumulative frequency plots. In the case of an acceptable fit, except for the extreme residual values, the points form a linear array in the cumulative frequency plot; in the case of an unacceptable fit, the array of the points is nonlinear. To construct a cumulative frequency plot, the residuals are ordered and numbered sequentially such that
For the χ2 test, a frequency distribution of the residuals is constructed. The observed frequencies,Oi
, are compared with the expected frequencies, Ei
, by the relationship
Linear regression analysis (Neter and Wasserman, 1974).
Kinetic and steady state parameters were obtained from the nonlinear regression analysis of many experiments performed at similar conditions. In table 1, averages of these kinetic and steady state parameters are listed. To study the correlation between Tmax and each of the other parameters, we performed linear regression analysis.
After fitting a straight line to the points defined by each Tmax and the respective value of the other parameter, we determined the slope b and the standard deviation of the slope Sb . The ratiob/Sb follows tdistribution with a degree of freedom of n − 2, wheren is the number of points.
The null hypothesis is that the slope of the fitted line is not 0 because of random experimental errors, i.e., Tmax and the other parameter are not correlated. The null hypothesis is rejected when P < .05, where P is the probability belonging to the calculatedb/Sb value in thet table at n − 2 degrees of freedom.
Kruskal-Wallis H test.
In tables 1, 2 and 3, averages of kinetic and steady-state parameters are listed. Each parameter is the average of the respective parameters obtained from the nonlinear regression analysis of many experiments performed under similar conditions.
To test whether the distributions of each parameter are different under different conditions, we use the Kruskal-Wallis H test. The advantage of this nonparametric statistical test is that no assumptions about the actual form of the probability distributions of the model parameters are required. The assumptions of the test are that 1) there are five or more measurements at each condition and 2) the samples are drawn independently. The null hypothesis of the test is that the probability distributions of the considered model parameter are identical at each condition. If the null hypothesis is rejected, then the distribution of at least at one condition is different from that of the others.
The H value is calculated from the data according to the following equation
Footnotes
-
Send reprint requests to: Saul Maayani, Ph.D., Department of Anesthesiology, Box 1010, The Mount Sinai Medical Center, New York, NY 10029-6574.
-
↵1 This work was supported in part by United States Public Health Service Grants GM34852 and T35-DK07420.
-
↵2 This work was submitted in partial fulfillment of the requirements for graduation with “Distinction in Research” from the Mount Sinai School of Medicine.
- Abbreviations:
- BAR
- beta-2 adrenoceptor(s)
- CARB
- carbachol
- ISO
- isoproterenol
- NECA
- N-ethylcarboxamidoadenosine
- PROP
- propranolol
- Rmax
- maximal relaxation response
- Rss
- relaxation response at steady state
- SALB
- salbutamol
- SALM
- salmeterol
- telap
- time elapsed
- Tmax
- maximal level of tissue tension
- Tmin
- minimal level of tissue tension
- Received April 29, 1996.
- Accepted September 4, 1996.
- The American Society for Pharmacology and Experimental Therapeutics
References
JPET articles become freely available 12 months after publication, and remain freely available for 5 years.Non-open access articles that fall outside this five year window are available only to institutional subscribers and current ASPET members, or through the article purchase feature at the bottom of the page.
|