Evidence for a Non-β 2-Adrenoceptor Binding Site in Human Lung Tissue for a Subset of β 2-Adrenoceptor Agonists

The aim of this study was to compare the binding profile of a range of β2-adrenoceptor (β2-AR) agonists and antagonists in human lung tissue. Radioligand saturation and competition binding experiments were performed by filtration with a β2-AR antagonist ([H]propranolol) or agonist ([H]vilanterol) radioligand and membrane fragments generated from lung parenchyma in the presence of 100 μM guanosine 5’-[β,γ-imido]triphosphate (Gpp(NH)p). In membranes prepared from human lung parenchyma, carmoterol, formoterol, ICI118551, propranolol and salbutamol resulted in inhibition of [H]vilanterol binding to levels that were significantly different from indacaterol, salmeterol and vilanterol (ANOVA, Bonferroni post-test, P < 0.001 except formoterol vs indacaterol where P < 0.01). Indacaterol and salmeterol resulted in inhibition of [H]vilanterol binding to levels that were not significantly different from vilanterol (ANOVA, Bonferroni post-test, P > 0.05). Indacaterol, salmeterol and vilanterol resulted in full inhibition of [H]propranolol binding to levels not significantly different from ICI118551 (ANOVA, Bonferroni post-test, P > 0.05). Indacaterol, salmeterol and vilanterol bind to an additional site in human lung parenchyma membranes that is distinct from the β2-AR.


Introduction
Inhaled β 2 -adrenoceptor (β 2 -AR) agonists are used in the treatment of both asthma [1] and chronic obstructive pulmonary disease (COPD) [2] by causing relaxation of the airways and increased airflow into the lungs.Early drug discovery efforts in this area yielded short acting β 2 -AR agonists like salbutamol in the late 1960s [3] and during subsequent decades a number of pharmaceutical companies focussed on developing molecules with longer duration of action for better control of symptoms and lung function.The fruits of this labour included formoterol [4] and salmeterol [5] that provided relief of symptoms for at least 12 hours.These drugs were branded long acting β 2 -AR agonists (LABAs) and are still used routinely as twice daily bronchodilators in both asthma and COPD, in combination with a corticosteroid.The next generation of LABAs, that include vilanterol and indacaterol, were developed to be fast acting and have 24-hour duration of action as this is predicted to improve patient convenience, and therefore compliance, within these patient populations [6].
The mechanism by which both the established (formoterol and salmeterol) and recently developed (indacaterol, olodaterol and vilanterol) LABAs achieve their long duration of action has never been fully elucidated despite almost 20 years of literature dedicated to it.A number of hypotheses have been put forward that include the "microkinetic" theory [7], "exosite" theory [8] and slow dissocation kinetics from the high affinity agonist receptor state [9].The "microkinetic" theory or model describes a highly lipophilic molecule partitioning into cell membrane and forming depots of drug maintaining active concentrations of drug in the tissue for longer.The "exosite" theory puts forward the hypothesis of a second, distinct binding site on the β 2 -AR itself that interacts with the long hydrophobic chains of salmeterol and vilanterol, trapping the agonist in the vicinity of the orthosterically active binding site.Recently it was also suggested for the LABA olodaterol that the duration of action observed in tissue studies was due to a slow dissociation rate from the β 2 -AR high affinity agonist receptor state.However, the validity of this study remains debatable due to the non-physiological temperature (room temperature) the experiments were completed at, as much faster kinetics would be predicted at 37˚C [10].As proposed in a recent review of these hypothesises, framed in the context of explaining salmeterol's duration of action [11], further in-depth studies in sub-cellular systems are required to aid in the determination of the exact mechanism/s of action that account for the duration of action of LABAs.One potential method to further investigate the interaction at the sub-cellular level is radioligand binding studies in membranes prepared from human lung tissue.Aside from the saturation binding studies completed on a radiolabelled form of formoterol in human lungs [12] these types of study have not been routinely completed with the LABAs.
As part of the vilanterol drug development programme a radiolabelled form of this LABA was generated to determine its β 2 -AR binding characteristics in recombinant systems [13].In this study [ 3 H]vilanterol has been used as a tool radioligand, in parallel with [ 3 H]propranolol, to investigate the binding characteristics of a range of β 2 -AR agonists and antagonists in membranes generated from human lung tissue in an effort to provide further evidence to explain the duration of action of the LABAs.

Human Lung Parenchyma Membrane (HLPM) Preparation
Non-diseased human lungs from organ donors were obtained from the National Disease Research Interchange (NDRI, Philadelphia, PA, USA) in accordance with local human biological sample management procedures.The human biological samples were sourced ethically and their research use was in accord with the terms of the informed consents.5 -10 g samples of human lung parenchyma tissue obtained from 2 donors were dissected and cleaned of adherent connective and fatty tissue.Tissue samples were suspended in ice-cold assay buffer (50 mM Tris, 154 mM NaCl, 10 mM MgCl 2 and 2 mM EDTA, pH 7.4 (5M HCl)) and homogenised with an Ultra-Turrax homogeniser (IKA, Staufen, Germany) for 20 s followed by 4 × 4 strokes in a glass-teflon homogeniser.
Homogenised tissue was washed in assay buffer and centrifuged at 500 g for 10 min at 4˚C.The supernatant was then harvested and centrifuged at 40,000 g for 15 min at 4˚C with the resulting pellet resuspended in assay buffer and centrifuged a second time at 40,000 g for 15 min at 4˚C.Membrane pellets were then passed 10 × through a 0.22 mm needle, resuspended in assay buffer and protein concentration determined using the bicinchoninic acid method [15] using bovine serum albumin as a standard.The membrane suspensions were frozen in aliquots at −80˚C until required.

Radioligand Binding Assays
All All statistical analyses were completed using SAS® (SAS Institute Inc., NC, USA) and differences of P < 0.05 were considered to be statistically significant.Statistical significance between two data sets was tested using a Student's unpaired t-test.One-way analysis of variance (ANOVA) was used for comparison of more than two datasets to highlight specific inter-group Pvalues, with Holm's method [17] used to adjust P-values for multiple comparisons and so lessen the occurrence of false positive results.

Effect of β 2 -AR Agonists and Antagonists on [ 3 H]vilanterol Binding in HLPMs
Competition binding with unlabelled β 2 -AR agonist and antagonists was determined against [ 3 H]vilanterol at single concentrations (10 µM) in HLPMs following a 1 h incubation period at 37˚C at a concentration of radioligand that ensured measurement of β 2 -AR binding only (i.e.approximately >400-fold lower concentrations shown to engage other endogenous receptors (screened against panel of 7TM receptors and transporters using radioligand binding assays by Eurofins Panlabs Inc. (Bothell, WA, USA), data not shown) including β 1/3 -adrenoceptors [13]).Carmoterol, formoterol, ICI118551, propranolol and salbutamol resulted in inhibition of [ 3 H]vilanterol binding to levels that were significantly different from indacaterol, salmeterol and vilanterol (ANOVA, Bonferroni post-test, P < 0.01).Indacaterol and salmeterol resulted in inhibition of [ 3 H]vilanterol binding to levels that were not significantly different from vilanterol (ANOVA, Bonferroni post-test, P > 0.05) although indacaterol binding inhibition of [ 3 H]vilanterol was shown to be significantly different to salmeterol (ANOVA, Bonferroni post-test, P < 0.01) (Figure 3).CGP20712 did not inhibit the control binding of [ 3 H]vilanterol confirming that this radioligand was not binding to β 1 -ARs in the HLPMs (Figure 3).

Effect of β 2 -AR Agonists and Antagonists on [ 3 H]propranolol Binding in HLPMs
Single concentrations (10 µM) of indacaterol, salmeterol and vilanterol resulted in inhibition of [ 3 H]propranolol binding to levels not significantly different from ICI118551 and propranolol (ANOVA, Bonferroni posttest, P > 0.05) (Figure 4).CGP20712 partially inhibited the binding of [ 3 H]propranolol, compared to control binding, to a significant level (ANOVA, Bonferroni post-test, P < 0.001) confirming that this radioligand was binding to a population of β 1 -ARs in the HLPMs (Figure 4).

Discussion
LABAs have been used for the last 20 years in the treatment of asthma and COPD to provide continuous relief of symptoms via sustained relaxation of the airways and increased airflow into the lungs.The mechanism accounting for the observed extended duration of airway relaxation following LABA inhalation has never fully been elucidated and although a number of hypotheses have been put forward [7][8][9] none of these have been categorically proven.In this study [ 3 H]vilanterol has been used as a tool radioligand in unison with [ 3 H]propranolol to investigate the binding of a range of β 2 -AR agonists (both short and long acting) and antagonists in sub-cellular membrane preparations generated from human lung tissue.The aim being to provide further evidence to either add weight to or rule out the current theories that have been put forward for the duration of action displayed by LABAs.Historical studies have been completed using membrane preparations from CHO cells recombinantly expressing the human β 2 -AR allowing characterisation of the affinity and maximal inhibition of binding of unlabelled ligands using [ 3 H]vilanterol [13].This allowed the characterisation of binding at the β 2 -AR in the absence of any human lung tissue architecture i.e. measure receptor based binding interactions only.All the agonist and antagonists tested were shown to inhibit binding of [ 3 H]vilanterol to NSB levels defined by ICI118551 in this system suggesting they were all binding to a common binding site on the β 2 -AR [13] i.e. the orthosteric binding site.In order to investigate the binding characteristics of [ 3 H]vilanterol in a more physiological relevant system, radioligand binding was measured in membranes prepared from human lung parenchyma tissue.
Membranes generated from human lung parenchyma contain a range of tissue architecture including membranes from cells making up alveoli, blood vessels and small airways.Therefore, it is worth noting that human lung membranes used in this study will contain a range of receptors in addition to the β 2 -AR, including the β 1 -AR subtype.To aid in the dissection of β 2 -AR versus β 1 -AR subtype binding, a tritiated version of the nonselective β 2 -AR antagonist propranolol [16] was investigated in addition to [ 3 H]vilanterol.When competition binding against both these radioligands was completed in the presence of the β 1 -AR selective antagonist CGP20712, a significant inhibition of [ 3 H]propranolol was observed compared with no inhibition of [ 3 H]vilanterol (Figures 3  and 4).This confirmed that there was indeed a population of β 1 -ARs in the human lung membrane preparations and highlighted the requirement to complete all subsequent [ 3 H]propranolol binding studies in the presence of CGP20712 to ensure measurement of β 2 -AR binding only.It was also a further confirmation of the lack of [ 3 H]vilanterol β 1 -AR binding at the concentrations of this radioligand tested in competition studies (~0.3 nM).
The saturation binding data for [ 3 H]vilanterol and [ 3 H]propranolol showed that both radioligands were labelling the same number of β 2 -AR binding sites in human lung membranes, that would be predicted to be the β 2 -AR orthosteric binding site in its low affinity state due to the presence of Gpp(NH)p.Subsequent single concentration competition binding for a range of β 2 -AR agonists and antagonists against [ 3 H]vilanterol showed a subset of LABAs (indacaterol, salmeterol and vilanterol) inhibiting binding to a significantly greater level than other test agents (Figure 3).In contrast, when tested against [ 3 H] propranolol, indacaterol, salmeterol and vilanterol all inhibited binding to the same level as all other test agents.This showed that [ 3 H]vilanterol/vilanterol was binding to a secondary binding site in human lung membranes distinct from the β 2 -AR, that it shared with indacaterol and salmeterol but not the other β 2 -AR agonists and antagonists tested.With the binding to other endogenous receptors ruled out and this observation not repeated in the recombinant β 2 -AR CHO membranes where binding to the β 2 -AR in isolation is measured [13], this would suggest that the secondary binding site is only present in membranes generated from human lung tissue.
Due to the structural similarities between salmeterol and vilanterol i.e. saligenin head with a long hydrophobic tail (Figure 1), it may not be a surprise that they share this tissue binding site and it could be hypothesised that the long hydrophobic tail could be a contributing factor.The data showing that indacaterol also shares this tissue binding site is a much more interesting observation due to its structural differences with vilanterol and salmeterol.As carmoterol in this study displayed no interaction with the tissue binding site and as it shares a carbostyril head with indacaterol (Figure 1), this suggests it is indacaterol's more hydrophobic tail that is interacting with the tissue binding site.In addition, comparing the cLogP data for the LABAs tested (Table 1) with the inhibition level of [ 3 H]vilanterol binding observed in human lung membranes there is a trend observed that the increased lipophilicity of this structural region (hydrophobic tail) contributes to engagement with the tissue binding site.High lipophilicity per se does not result in an interaction with this secondary site as propranolol and ICI118551 did not bind to the tissue site and have either a comparable or greater cLogP than indacaterol, salmeterol and vilanterol (Table 1).
The novel evidence for a non-β 2 -AR human lung tissue site that has been generated in this study does not provide any further evidence for confirmation nor invalidation of either the "microkinetic" [7] or "exosite" [8] theories put forward for the duration of action of LABAs in the early 1990s.Further research into the characterisation of this human lung tissue site could be extremely beneficial, especially if a structure-activity relationship could be identified by the profiling of an increased number of chemical entities.This could result in the development of inhaled drugs acting in the lungs that target this tissue binding site, in addition to their primary target, as a means of increasing their duration of action.

Conclusion
In summary, it has been shown using β 2 -AR agonist and antagonist tool radioligands that a tissue binding site distinct from the β 2 -AR is present in parenchyma membranes prepared from human lung tissue.This non-β 2 -AR binding site appears to be exclusive to a select number of LABAs (indacaterol, salmeterol and vilanterol), with potentially a link between their lipophilicity and the ability to interact with the site.This may provide a further hypothesis for the duration of action exhibited by these drugs, in addition or as an alternative to the "microkinetic" and "exosite" theories, where binding to a tissue site holds these LABAs in the lung for a longer period of time manifesting in a prolonged activation of the β 2 -AR and subsequent relief of the symptoms of asthma and COPD.

Figure 2 .
Figure 2. Saturation binding of (a) [ 3 H]propranolol and (b) [ 3 H]vilanterol with the β 2 -AR in HLPMs.Specific saturation binding data for (a) [ 3 H]propranolol and (b) [ 3 H] vilanterol were best fitted to a one-site affinity model (extra-sum-of-square F test, P < 0.05).Data shown are the mean ± SD of duplicate points and are representative of 6 individual experiments with similar results (2 donors, n = 3/donor).

Table 1 . Calculated Log P (cLogP) values calculated for the β 2 -AR agonists and antagonists tested in this study. β 2 -AR Agonist/Antagonist cLogP a
a cLogP values calculated by Daylight (Daylight Chemical Information Systems Inc., Laguna Niguel, CA, USA).