European Respiratory Society

Beyond the dogma: novel β2-adrenoceptor signalling in the airways

M. A. Giembycz, R. Newton


β2-Adrenoceptor agonists evoke rapid bronchodilatation and are the mainstay of the treatment of asthma symptoms worldwide. The mechanism of action of this class of compounds is believed to involve the stimulation of adenylyl cyclase and subsequent activation of the cyclic adenosine monosphosphate (cAMP)/cAMP-dependent protein kinase cascade.

This classical model of β2-adrenoceptor-mediated signal transduction is deeply entrenched, but there is compelling evidence that agonism of β2-adrenoceptors can lead to the activation of multiple effector pathways, which now compels researchers in academia and the pharmaceutical industry alike to think beyond the traditional dogma. Therefore, the regulation by β2-adrenoceptor agonists of responses, including airways smooth muscle tone and the secretory capacity of the epithelium and pro-inflammatory/immune cells, may be highly complex, involving both cAMP-dependent and -independent mechanisms that, in many cases, may act in concert.

In this article, the current status of β2-adrenoceptor-mediated signalling in the airways is reviewed in the context of understanding mechanisms that may underlie both the beneficial and detrimental effects of these drugs in asthma symptom management.


Edited by K.F. Chung and I.M. Adcock

Number 5 in this Series

Short-acting β2-adrenoceptor agonists are the most effective and safest bronchodilators currently available, and are the drugs of choice for the rapid alleviation of bronchoconstriction that occurs in asthma. This class of compounds relaxes airways smooth muscle (ASM) irrespective of the constrictor stimulus and are well tolerated by most patient groups 13. The introduction, in the early 1990s, of long-acting β2-adrenoceptor agonists and their widespread inclusion into primary healthcare regimens (e.g. 2003 British Thoracic Society guidelines 4), as well as the clinical development of other cyclic adenosine monosphosphate (cAMP)-elevating compounds, such as phosphodiesterase (PDE) 4 inhibitors, has led to the realisation that cAMP elevation in relevant cells may also provide additional benefits, including possible anti-inflammatory activity 5. Furthermore, the widely reported beneficial effects in asthma of long-acting β2-adrenoceptor agonists in combination with inhaled glucocorticoids has led to renewed interest in the molecular basis underlying this enhanced therapeutic effect 6. Despite improved asthma treatment options, adverse effects are not reported infrequently and increased numbers of asthma-related deaths associated with high-level use of certain agonists are well documented 7. Although these adverse clinical events may be due to receptor or even whole pathway desensitisation 8, studies since the late 1990s suggest that promiscuous coupling of the β2-adrenoceptor to effectors other than the traditional cAMP/cAMP-dependent protein kinase (PK)A cascade may also be involved 9. More fundamentally, the molecular mechanisms by which β2-adrenoceptor agonists cause ASM to relax are still not completely defined. It is likely that areas of deficiency in the understanding of β2-adrenoceptor function and action will provide explanations for many of these observations and continued research may lead to improved therapeutic approaches. In the present article, β2-adrenoceptor signalling in the airways is reviewed and related, where possible, to the clinical effects of selective agonists.


It is well established that the β2-adrenoceptor can couple via the heterotrimeric stimulatory G-protein (Gs) to adenylyl cyclase (AC), thereby enhancing the rate of cAMP biosynthesis. One primary consequence of this signalling is the activation of PKA, which, according to conventional dogma, mediates the ability of cAMP to cause smooth muscles to relax by a variety of complementary mechanisms (fig. 1a) 2, 3, 1014. However, incremental advances in the understanding of β2-adrenoceptor-mediated signal transduction since the mid-1990s have shown that this deceptively simple signalling cascade is, in fact, highly complex, involving multiple variants of AC, PKA and PDE, as well as scaffolds, such as PKA anchoring proteins (AKAPs), which bind PKA and play a role in targeting the kinase to specific intracellular locations 1519. In addition, data obtained from studies using inhibitors of AC and PKA suggest the existence of cAMP- and PKA-independent mechanisms of ASM relaxation evoked by β2-adrenoceptor agonists (fig. 1b–e) 2022. Likewise, there are reports of PKA-independent effects of cAMP-elevating agents in respect of repression of cytokine release and apoptotic responses 23, 24. Therefore, it is highly likely that further nonclassical, yet biologically significant, mechanisms of β2-adrenoceptor signalling remain to be elucidated 19, 21, 2528.

Fig. 1—

Multiple pathways of cyclic adenosine monophosphate (cAMP)-dependent signalling. a) Agonism of β2-adrenoceptors (β2-AR) in the membrane of human airways smooth muscle (ASM) and other cells results in the liberation of stimulatory G-protein (Gs) subunit α from the αβγ heterotrimeric guanine nucleotide-binding protein. The free α subunit then augments the basal activity of one or more isoforms of adenylyl cyclase (AC), resulting in an increase in the formation of cAMP from adenosine triphosphate (ATP). cAMP binds to the regulatory subunits of protein kinase (PK) A and, thereby, promotes the release of the catalytic subunits, which phosphorylate target proteins to bring about changes in cell function. b) Targets of PKA include potassium channels (e.g. large-conductance calcium-activated potassium channel and ATP-sensitive potassium channel) that open upon phosphorylation, resulting in the efflux of K+ from the cell down its concentration gradient leading to membrane rectification (repolarisation). In ASM, K+ efflux reduces the excitability of the cell and facilitates ASM relaxation. Gating of K+ channels may also be effected by the direct interaction of Gsα with the channel independently of cAMP and PKA. c) β2-AR agonists can also promote the cAMP-mediated activation of PKG (so-called cross-activation), leading to functional responses in the airways such as smooth muscle relaxation. d, e) More novel β2-AR signalling cascades include the activation of the tyrosine kinase, Src, via either inhibitory G-protein (Gi) subunit α or through Gsα leading to Ras, Raf-1, mitogen-activated protein kinase kinase (MKK) 1 and extracellular signal-regulated kinase (ERK) activation (d) and the binding and activation by cAMP of exchange proteins directly activated by cAMP (Epac) independently of PKA, leading to Rap-1-dependent responses (e).


Adrenoceptors are members of a large family of related G-protein-coupled receptors (GPCRs) and are activated by the endogenous hormones adrenalin and noradrenalin. There are two main groups of adrenoceptor and these have been classified as α- and β-subtypes, which are encoded by at least nine distinct genes (α1A, α1B, α1D, α2A/D, α2B,2C, β1, β2 and β3) 29. α1-Adrenoceptors can couple via the G-protein, Gq, to phospholipase C (PLC), ultimately leading to an increase in the cytosolic free Ca2+ concentration ([Ca2+]c) 30. Accordingly, α1-adrenoceptor agonists promote Ca2+-dependent responses, typically smooth muscle contraction. Conversely, agonism of α2-adrenoceptors can lead to an inhibitory G-protein (Gi)-mediated inhibition of AC that may act to prevent smooth muscle relaxation 30. Despite evidence for α-adrenoceptors in the lung, neither receptor subtype has a clear role in regulating human ASM tone 31. This is in contrast to all β-adrenoceptor subtypes, which can activate AC through Gs and could, in theory, promote smooth muscle relaxation 30.

It is noteworthy that the β3-adrenoceptor can also couple to Gi and, therefore, inhibit AC. This finding may have significance for epithelial cell function, where functional β3-adrenoceptor has been described, but not for ASM, which seemingly lacks expression of this subtype 3234. In addition to being widely expressed on ASM 31, β2-adrenoceptors are also expressed on many pro-inflammatory and immune cells, including mast cells 35, macrophages 36, neutrophils 37, lymphocytes 38, eosinophils 39, epithelial and endothelial cells 40, 41, and both type I and type II alveolar cells 10, 42, 43. Thus, many cell types within the lung that are implicated in the pathogenesis of asthma are potential targets of inhaled β2-adrenoceptor agonists.


It is the profound bronchodilator response effected by β2-adrenoceptor agonists that is most critical to asthma therapeutics. Several, possibly complementary, mechanisms have been advanced to explain how agonism of β2-adrenoceptors on ASM cells promotes relaxation (see below), although none are entirely satisfactory. This lack of a unifying concept strongly suggests that understanding of this important process is fundamentally incomplete and prompts speculation that additional components of β2-adrenoceptor signalling, intimately involved in regulating ASM tone, remain to be defined.

Reduced sensitivity of the contractile proteins to calcium

Generically, agonist-dependent contraction of ASM requires an increase in [Ca2+]c, which can originate extracellularly (entering through Ca2+ channels in the plasma membrane) or from intracellular stores (fig. 2). This effect results in the Ca2+/calmodulin(CaM)-dependent activation of myosin light chain (MLC) kinase (MLCK), phosphorylation of the 20-kDa MLC (MLC20) at serine (Ser)19 and subsequent muscle contraction (fig. 2). Conversely, many texts cite the phosphorylation of MLCK by PKA as being an event intimately related to β2-adrenoceptor-mediated relaxation (fig. 2) 44. Indeed, this phosphorylation event reduces the ability of MLCK to phosphorylate MLC20 at Ser19 by increasing its requirement for Ca2+/CaM (i.e. it reduces its sensitivity to Ca2+; fig. 3) 4547. This mechanism also adequately explains the finding that both forskolin and isoprenaline reduce force generation in intact depolarised tracheal smooth muscle at a constant [Ca2+]c; similar data have also been obtained with the addition of PKA to Triton X-100-permeabilised tracheal smooth muscle 4850. The additional finding that contraction of tracheal smooth muscle by the M3 muscarinic receptor agonist, carbachol, correlates with MLC20 phosphorylation and that this effect is reduced by isoprenaline also supports the above hypothesis 51. However, at longer time points, [Ca2+]c and MLC20 phosphorylation decline towards resting levels, whereas smooth muscle force is maintained. These data indicate that β2-adrenoceptor-mediated relaxation of ASM is a complex process involving mechanisms other than, or in addition to, phosphorylation of MLCK 51, 52. Several studies have provided data that are completely inconsistent with the concept that PKA-mediated phosphorylation of MLCK reduces smooth muscle force. For example, it has been reported that, in bovine tracheal smooth muscle, β2-adrenoceptor agonists have no effect on the proportion of MLCK in the activated state at concentrations that caused relaxation 52. Furthermore, in vitro, PKA phosphorylated MLCK at two distinct sites, which are defined by separate tryptic fragments corresponding to amino acids 990–1002 and 1003–1017 in rabbit smooth muscle MLCK (fig. 3) 11, 46, 53, 54. These fragments have been designated A and B, respectively, and it appears that only phosphorylation within the A fragment, which maps next to the CaM-binding domain, is responsible for the inactivation of MLCK, by reducing Ca2+/CaM binding (fig. 3) 11, 46, 53, 54. Importantly, only minimal change in the phosphorylation within the A site, or indeed MLCK, occurs following exposure of the tissue to cAMP-elevating agents under conditions in which force was shown to be greatly diminished. Thus, there is an absence of a causal relationship between MLCK and muscle relaxation 52, 53. Therefore, the role of MLCK phosphorylation is equivocal and may not represent the primary mechanism of PKA-mediated Ca2+ desensitisation 53. It is worth noting that the peptides, A and B, contain a number of other potential phospho-acceptor sites. Indeed, the MLCK A site has been shown to be a substrate for CaM-dependent protein kinase II (CaMKII; fig. 4) 55, 56. As CaMKII is highly Ca2+ sensitive, the regulation of MLCK activity via this kinase may normally represent a mechanism of classical negative feedback control rather than a primary site for PKA-mediated relaxation (fig. 4) 54. Finally, MLCK can also become phosphorylated on other residues in response to β-adrenoceptor agonists, raising the possibility of alternative unexplored regulatory mechanisms 53.

Fig. 2—

Contraction of airways smooth muscle. Contractile agonists, such as acetylcholine (ACh), bind to G-protein (Gq)-coupled muscarinic M3 receptors on the plasma membrane, resulting in the activation of phospholipase C (PLC) and the subsequent production of inositol 1,4,5-trisphosphate (IP3) as a product of inositol phospholipid hydrolysis. IP3 binds to ligand-gated IP3 receptors (IP3Rs) on the endoplasmic reticulum (ER) to release Ca2+ into the cytoplasm. In addition, many agonists may also promote, either directly or indirectly, the entry of Ca2+ from the exterior of the cell via the action of Ca2+ channels in the plasma membrane. The rise in intracellular Ca2+ causes Ca2+ to bind to calmodulin (CaM; and subsequently myosin light chain kinase (MLCK)) with the formation of a catalytically active (Ca2+)4-CaM-MLCK complex, which then phosphorylates (P) serine 19 of the 20-kDa regulatory light chain of myosin (MLC20) to promote actin–myosin cross-linking and smooth muscle contraction. PKA: protein kinase A.

Fig. 3—

Activation of myosin light chain kinase (MLCK) by calcium-calmodulin (CaM). a) Schematic representation of the 1,147-amino-acid rabbit smooth muscle MLCK. The catalytic core (amino acids 703–951; ▒) and CaM-binding domain (▪) are shown, along with an expanded sequence showing part of the CaM-binding domain and its proximity to the tryptic peptides A and B, which are phosphorylated by protein kinase (PK)A. The putative PKA phosphorylation sites (*: serine (Ser) residues 992 and 1005). b) Phosphorylation (P) of a Ser residue (probably Ser992) within the peptide A region reduces the ability of (Ca2+)4-CaM to bind and activate MLCK. Thus, MLCK phosphorylated in this region requires a higher [Ca2+], at a fixed CaM concentration, to achieve enzyme activation. Unphosphorylated enzyme is more readily activated by Ca2+ at a fixed CaM concentration. R: arginine; K: lysine; W: tryptophan; Q: glutamine; T: threonine; G: glycine; N: asparagine; A: alanine; V: valine; I: isoleucine; L: leucine; S: serine; M: methionine; P: proline; E: glutamic acid. (Modified from 11.)

Fig. 4—

Opposing activities of myosin light chain kinase (MLCK) and myosin light chain phosphatase (MLCP) in the control of airways smooth muscle contraction. The rise in intracellular Ca2+ concentration in response to contractile agonist (fig. 3) results in the formation of (Ca2+)4-calmodulin (CaM) complexes, which, in turn, activate MLCK. Active MLCK phosphorylates (P) serine (Ser)19 of the 20-kDa regulatory light chain of myosin (MLC20) to promote smooth muscle contraction. However, (Ca2+)4-CaM can also activate CaM-dependent protein kinase II (CaMKII), which phosphorylates MLCK in the A site (Ser992) in order to prevent activation by CaM. This is the previously described site for phosphorylation by protein kinase (PK)A. Ser19 phosphorylation of MLC20 is also regulated by the activity of MLCP. MLCP is a multi-protein complex consisting of protein phosphatase (PP)1c, a 20-kDa subunit (M20), the function of which is unknown, and a targeting subunit known as myosin phosphatase target subunit (MYPT1). The primary activity of MLCP is regulated by phosphorylation of MYPT1 by a number of pro-contractile kinases, resulting in a reduced affinity of the phosphatase for myosin and leading to the inactivation of MLCP. In addition, the activity of MLCP towards myosin is also repressed by PKC-activated PP inhibitor 17 (CPI-17), which is itself activated by phosphorylation at threonine (Thr) 38 by pro-contractile kinases. Finally, both PKG and PKA may maintain MLCP activity by targeting MYPT1, at Ser695, thereby preventing the adjacent Thr696 phosphorylation, which is responsible for switching off MLCP activity. In addition, PKG appears to prevent CPI-17 phosphorylation and activation via as yet uncharacterised mechanisms. ILK: integrin-linked kinase; ROCK: Rho-associated protein kinase; PAK: p21-activated kinase. –––|: inhibition.

The phosphorylation status of MLC20 at Ser19 54 is also controlled by myosin-bound protein phosphatase 1 (PP-1M) 57, which opposes the activity of MLCK. PP-1M, often referred to as MLC phosphatase (MLCP) 58, or smooth muscle myosin phosphatase 59, consists of a catalytic subunit of 38 kDa, which is identical to protein phosphatase 1c (PP1c), and two other proteins of ∼20 kDa (M20) and ∼130 kDa (fig. 4) 59, 60. The function of the smaller subunit is currently unclear, whereas the larger subunit, myosin phosphatase target subunit (MYPT1), or myosin-binding subunit, is responsible for providing substrate specificity by virtue of binding both PP1c and myosin 59. As stated above, it is the relative activities of MLCK and MLCP that regulate the sensitivity of the contractile apparatus to Ca2+. Although it is well established that MLCK activity is controlled by Ca2+/calmodulin, it is now apparent that MLCP activity may also be regulated to a high degree, and this represents a primary mechanism involved in the regulation of smooth muscle contractility 58, 59. Thus contractile agonists lead to the phosphorylation of MYPT1 and another protein, PKC-activated protein phosphatase inhibitor 17 (CPI-17), which appears to be primarily responsible for inactivating MLCP in smooth muscle, with a consequent enhancement of force development 6163. Interestingly, the phosphorylation of CPI-17 at threonine (Thr38) 64, 65 and MYPT1 at Thr696 can occur via a range of kinases, including PKC 64, integrin-linked kinase 66, p21-activated kinase 67 and Rho-associated protein kinase (ROCK). Therefore, MLCP and hence MLC20 phosphorylation are potentially under the control of multiple kinase signalling cascades (fig. 3) 58, 6870. In addition, it has been reported that phosphorylation of MYPT1 by ROCK at other sites (Thr853 in humans) promotes its dissociation from myosin and thereby reduces the activity of MLCP (fig. 4) 70. With respect to ASM relaxation, there are data to suggest that β-adrenoceptor agonists may activate MLCP and that this may play an important role in reducing tone 71. Moreover, MLCP can also be activated by the cyclic guanosine monophosphate (cGMP)/cGMP-dependent PKG pathway. Thus, relaxation of arterial smooth muscle by the nitric oxide donor, sodium nitroprusside, is associated with reduced MLC20 phosphorylation 72. These events correlate with elevated levels of cGMP, increased MLCP activity and a concomitant loss of CPI-17 phosphorylation 72; the cGMP analogue, 8-bromo-cGMP, is also reported to reduce CPI-17 phosphorylation (fig. 4) 73. Taken together, these data are potentially significant as cAMP can cross-activate PKG in smooth muscle (fig. 1c) 74, and studies using cyclic nucleotide analogues support a role for PKG 75 rather than PKA in reducing tracheal smooth muscle tone (see below). A further mechanism that would prevent agonist-induced downregulation of MLCP was also recently proposed for PKG and PKA 76. Thus phosphorylation of the MLCP targeting subunit, MYPT1, at Ser695, was not found to affect MLCP phosphatase activity. Instead, Ser695 phosphorylation blocked the subsequent phosphorylation at a nearby site (Thr696) that is involved in phosphatase inactivation by other kinases (fig. 4). Thus cyclic nucleotide-dependent phosphorylation of MYPT1, at Ser695, prevents the ability of pro-contractile kinases to inactivate the phosphatase via phosphorylation at Thr696. In this manner, cyclic nucleotides may maintain MLCP activity to promote smooth muscle relaxation.

More recent studies have suggested that mitogen-activated protein (MAP) kinases may also impact on contractile responses in airways 64, 7779 and other types of smooth muscle 80, 81. In particular, MLC20 are the proposed substrate of extracellular signal-regulated kinase (ERK)1 and 2 8284. In the context of β2-adrenoceptor-mediated relaxation, MAP kinase phosphatase-1 (MKP-1), which dephosphorylates and inactivates both ERK and p38 MAP kinase, is transcriptionally activated by cAMP-elevating agents, including the β-adrenoceptor agonist, isoprenaline, by a PKA-dependent mechanism 85, 86. Therefore, the induction of MKP-1 may, as has been recently described for glucocorticoids 87, represent a novel mechanism by which β2-adrenoceptor agonists, upon repeated use, can suppress smooth muscle contraction.

Calcium-activated potassium channels and membrane hyperpolarisation

Another biologically significant effect of β2-adrenoceptor agonists is membrane hyperpolarisation 11. In ASM, this response is mediated through the activation of K+ channels in the plasma membrane, which counteracts the electrical excitation and subsequent Ca2+ influx that contribute Ca2+ to contraction 88. Of the four main classes of K+ channel, it is the Ca2+-activated K+ channels that appear to be the most critical 88, 89, in particular the large- or big-conductance channel (BKCa) 90. These channels, which are composed of four subunits, are abundant in ASM and are a substrate for PKA 89, 9193. Furthermore, their role in regulating ASM contractility is suggested from the finding that pharmacological blockade of these channels with charybdotoxin, and the more selective iberiotoxin, prevents hyperpolarisation and β2-adrenoceptor-mediated relaxation 9396.

More recent studies on the role of BKCa in β2-adrenoceptor-mediated relaxation point to the existence of both cAMP-dependent and -independent mechanisms, as well as a role for the cGMP/PKG signalling cascade 97, 98. Indeed, there is evidence for direct coupling of BKCa to Gs, which provides an explanation for the cholera toxin- and β2-adrenoceptor-mediated relaxation of guinea pig trachea that prevails in the presence of AC inhibitors (fig. 1b) 21, 22, 97. Similarly, in vascular smooth muscle, isoprenaline is reported to modulate BKCa activity in a membrane-delimited manner by a mechanism involving Gi 99, suggesting promiscuous coupling of the β2-adrenoceptor. Interestingly, in that tissue, the degree to which BKCa-blocking drugs were effective at preventing β2-adrenoceptor-mediated relaxation was highly dependent upon the concentration of agonist 99.

More direct evidence for the coupling of β2-adrenoceptor signalling to BKCa is the finding that co-expression of these proteins in Xenopus oocytes resulted, in the presence of agonist, in channel activation that was abolished either by inhibitors of PKA or following mutation of the consensus PKA phosphorylation sites within the BKCa α subunit (fig. 1b) 100. More recent experiments have identified a multi-protein signalling complex that is formed from a direct association of the BKCa α subunit, β2-adrenoceptor and AKAP79/150 (now classified as AKAP5) through a mechanism that may involve leucine-zipper protein–protein interactions 101, 102. This unique signalling complex may also include an L-type voltage-gated Ca2+ channel (Cav1.2), resulting in a highly localised signalosome that mediates Ca2+- and phosphorylation-dependent modulation of BKCa currents 101.

Notwithstanding the fact that some investigators contend the importance of BKCa in β2-adrenoceptor-mediated relaxation of ASM 88, a potentially significant level of complexity is raised by the knowledge that the pore-forming α subunit of BKCa, encoded by the slowpoke gene (slo1), exists as multiple splice-variants. One of these α splices (the ZERO isoform) is activated following phosphorylation by PKA, whereas another variant expressing a 59-amino-acid cysteine-rich exon at splice site 2, called the STREX-1 isoform, is inhibited 103. Moreover, in ASM, these variants of the BKCa α subunit show differential sensitivity to PKA and PKG 104. Interestingly, activation of BKCa α subunit activity by PKA requires phosphorylation of all four members of the α tetramer, whereas inactivation appears to require only a single PKA-mediated phosphorylation of the respective isoform 105. Even more fascinating is the finding that the glucocorticoid, dexamethasone, is able to block the PKA-dependent inhibition of the STREX-1 isoform, but is without effect on the PKA-mediated activation of standard (ZERO) α subunits 106. These extraordinary data provide a mechanism by which glucocorticoids may selectively enhance BKCa channel activity and provides yet another insight into the rationale behind the possible enhanced therapeutic benefit of combining a glucocorticoid and a β2-adrenoceptor agonist in the treatment of asthma 6.

Role of other potassium channels

Of the other three main groups of K+ channel, it is only the adenosine triphosphate (ATP)-dependent variants (KATP) 107109 that are currently believed to be functionally important in ASM 90, 110, 111. However, although KATP channel openers oppose bronchoconstriction, attenuate airways hyperreactivity and may play a role in β2-adrenocepotor-mediated relaxation of vascular smooth muscle, the effect of selective agonists, per se, on these channels in human ASM is not well studied 112114. Nevertheless, KATP channels are phosphorylated and activated by PKA in response to cAMP analogues, forskolin and isoprenaline 115, 116, and could, therefore, play a role in β2-adrenoceptor-mediated relaxation of ASM (fig. 1b). This possibility needs to be revisited.

Modulation of cytosolic free calcium concentration

In addition to the well-described mechanisms for reducing the Ca2+ sensitivity of the contractile apparatus, β2-adrenoceptor agonists are generally believed to reduce Ca2+ influx into ASM cells 14, 117120. Consistent with this effect, a number of studies have found that the inositol 1,4,5-trisphosphate (IP3) receptor (IP3R) is phosphorylated by PKA, apparently resulting in reduced Ca2+ release from the endoplasmic reticulum (ER) in response to IP3 121, 122. It is now known that there are at least three types of IP3R and that, in ASM, cAMP and cGMP both promote the PKG-dependent phosphorylation at Ser1755 of the type 1 IP3R that predominates in this tissue 123. However, the significance of these findings is questionable as it was subsequently found that phosphorylation of Ser696 of another protein called IP3R-associated cGMP-dependent protein kinase substrate, which is intimately associated with the IP3R, also inhibited IP3-induced Ca2+ release from the ER and that this effect was lost if Ser696 was mutated to alanine.

Of particular interest is the recent finding that isoprenaline, in single isolated ASM cells, increases the Ca2+ content in the peripheral cytosol of the cell by a mechanism that is dependent upon external Ca2+ concentration and blocked by ryanodine 124. In contrast, in the same cell, isoprenaline reduces the concentration of Ca2+ in the inner cytosol 124. Thus, β2-adrenoceptor agonists can produce discrete spatially distinct changes in the Ca2+ concentration within the cytosol of individual cells that presumably have profound functional implications, including effects on tone.

Another mechanism leading to the release of Ca2+ from the ER is via one or more ryanodine receptor (RyR), of which there are at least three variants 125, 126. RyR are intracellular Ca2+ release channels and are not usually associated with nonexcitable cells 125, 126. However, a recent study by Du et al. 127 identified both RyR1 and RyR3 in murine ASM. Moreover, Ca2+ release through these channels mediated agonist (carbachol)-induced tension development, a finding that was consistent with a previous report by Hay et al. 128. Taken together, these data suggest that RyR may represent novel targets for suppressing bronchoconstriction, although it should be noted that the effect of activation of the cAMP/PKA cascade on RyR function appears to be variable 126. Thus, studies specifically designed to examine the role of RyR in the regulation of ASM tone are clearly warranted.


Asthma deaths, adverse events and effect of glucocorticoids

Long-term use of β2-adrenoceptor agonists is common in the management of asthma and airways hyperresponsiveness (AHR), which is defined as an increase in the sensitivity of the ASM to constrictor stimuli 129132. Similarly, the ability of short- and long-acting β2-adrenoceptor agonists to protect the airways against bronchoconstrictor stimuli and to promote bronchodilatation may be partially lost with time following long-term use 2, 129, 131, 133140. Indeed, there is a greater tendency for bronchodilator tolerance to develop to long-acting β2-adrenoceptor agonists compared to their short-acting counterparts, which may relate to some aspect of their prolonged duration of agonism 141, 142. The alteration in ASM sensitivity to spasmogens is believed to predispose asthmatic subjects who use β2-adrenoceptor agonist on a regular basis to episodes of acute bronchoconstriction or “a loss of asthma control” 143. Thus, long-term β2-adrenoceptor agonist use is associated with an increased incidence of asthma exacerbations and other markers of morbidity and mortality 143145. Indeed, this finding is consistent with the many studies that have found that regular use of β2-adrenoceptor agonists, in particular full agonists e.g. fenoterol, is associated with elevated bronchial responsiveness and asthma-associated deaths 7, 146. Accordingly, the current recommendation of most national asthma guidelines is to use short-acting β2-adrenoceptor agonists as relievers (i.e. on an as-needed basis) and not as a regular prophylactic therapy. More recently, the issue of β2-adrenoceptor agonist safety has arisen again in respect of long-acting compounds 141. Thus, studies published in 1993 indicated that, although patients taking salmeterol generally showed better control of their asthma, a small but nonsignificant increase in the number of deaths in the salmeterol treatment group was noted 147. That result led to the Salmeterol Multi-centre Asthma Research Trial (SMART), which also revealed a slightly increased risk of death in subjects of certain genetic backgrounds 148. The issue of genetic susceptibility was again highlighted by findings indicating that certain β2-adrenoceptor polymorphisms may affect patient phenotype and outcome in response to β2-adrenoceptor agonist treatment 149, implying that genetic analysis may, in the future, be diagnostically useful in tailoring individual treatment regimens. The results of the SMART further prompted a warning by the US Food and Drug Administration, and the current recommendation from most key sources is that long-acting β2-adrenoceptor agonists should only be used in conjunction with inhaled glucocorticoids 150153. Although the balance of expert opinion is that the benefits of long-acting agonists, such as salmeterol, when taken alongside glucocorticoids, far outweigh the potential risk, there is still a pressing need to provide a rational molecular basis for these adverse effects. This will be especially important given the probable introduction before 2010 of ultra-long-acting β2-adrenoceptor agonists such as carmoterol, arfomoterol (R,R-formoterol), indacaterol and GSK 159797 (GlaxoSmithKline, Stevenage, UK) 142.

Mechanistic basis for adverse effects

Receptor desensitisation

One widely touted explanation for the reduction in both the bronchodilator and bronchoprotective actions of these drugs is that regular treatment leads to β2-adrenoceptor desensitisation (i.e. a state of refractoriness that ensues following prolonged exposure or repeated application of an agonist 129, 141). Receptor desensitisation (as discussed below) can occur via a variety of processes, including uncoupling of the receptor from Gs, receptor internalisation 3, 10, 13, upregulation of cAMP PDE 154157 and downregulation of Gs 158. However, receptor desensitisation may not fully account for all of the effects of prolonged high-level dosing with β2-adrenoceptor agonists 9. One unexpected observation, which may be of clinical relevance, is the finding that agonism of the β2-adrenoceptor can lead to overexpression of PLCβ1 in ASM 159. Using β2-adrenoceptor knockout mice, McGraw et al. 159 made the novel observation that methacholine (MCh)- and 5-hydroxytryptamine (5-HT)-induced bronchoconstriction were markedly reduced compared with the same response evoked in wild-type animals. Conversely, MCh- and 5-HT-induced bronchoconstriction in transgenic animals overexpressing two-fold cell surface β2-adrenoceptors was significantly enhanced. Thus, contrary to expectation, agonism of β2-adrenoceptors in mice augments signalling through GPCRs utilised by contractile agonists. Subsequent studies confirmed that inositol phosphate accumulation evoked by MCh, 5-HT and U-46619 (a thromboxane mimetic) was suppressed and augmented in knockout and transgenic overexpressing mice, respectively 159. These data led to the idea that β2-adrenoceptors antithetically regulate the actions of Ca2+-mobilising contractile agonists and can activate mechanisms that simultaneously promote bronchoconstriction and bronchodilatation. Further investigations have implicated PLCβ1 as the point of regulation since expression of this enzyme was enhanced and inhibited, respectively, in β2-adrenoceptor transgenic and knockout mice 159. Taken together, these results provide an unanticipated and intriguing explanation for the clinical finding that long-term treatment of asthmatic subjects with β2-adrenoceptor agonists can promote AHR and increase the incidence of exacerbations. The mechanism(s) by which β2-adrenoceptor agonists upregulate PLCβ1 are unknown. However, classical pharmacology would predict that this effect is related to the intrinsic efficacy of the agonist and that partial agonists might be less prone to antithetically regulating signalling through PLCβ1. Despite the elegance of these studies, it should be borne in mind that much of these data were obtained in mice following manipulation of cell surface β2-adrenoceptor number, and it is currently unclear to what extent such processes occur in human ASM in an in vivo setting.

Racemic formulations

All β2-adrenoceptor agonists used clinically are racemates, in which the R-enantiomer is generally the more active component relative to the S-enantiomer 160. The ability now to produce highly pure stereoisomers has led to the view that the less active or inactive S-enantiomer may be responsible for evoking some of the unwanted paradoxical responses by an unknown mechanism(s) that is unrelated to β2-adrenoceptor desensitisation (see above). Moreover, pharmacokinetic studies have found that the metabolic clearance of S-salbutamol occurs at a significantly lower rate than that of R-salbutamol, such that the unwanted effects may persist after the beneficial actions have worn off 161. Evidence is available that S,S-formoterol (inactive) may also accumulate in vivo relative to the active R,R-stereoisomer due to differential rates of metabolism 162. In view of these data, it is perhaps not surprising that many new long-acting β2-adrenoceptor agonists in clinical development are pure pharmacologically active enantiomers 142.

Promiscuous G-protein coupling

Another mechanism worthy of consideration that could contribute to the undesirable actions of long-term β2-adrenoceptor agonist treatment is the finding that β2-adrenoceptors can couple to multiple effectors via distinct G-proteins 9, 163. Studies conducted in the 1980s and early 1990s established that purified β2-adrenoceptor or a peptide corresponding to the third intracellular loop of the protein could activate pure Gi in a reconstituted cell-free system 164, 165, indicating that this promiscuous coupling could also occur in intact cells and, arguably, in vivo. It is now firmly appreciated that β2-adrenoceptors may couple to multiple guanosine triphosphate (GTP)-binding proteins, including Gi 166, 167, leading to Gβγ-dependent activation of the MAP kinase/ERK cascade (fig. 1d). Since this pathway is undoubtedly activated in asthma and chronic obstructive pulmonary disease and may contribute to the inflammation and even remodelling of the airway wall 168, it is tempting to speculate that further activation of ERK-dependent signalling may account for some of the adverse clinical observations made with β2-adrenoceptor agonists described above 169, 170. Such effects, as are often the case, require appropriate validation in primary cells relevant to the airways or suitable animal models.

Other explanations

Undesirable effects of β2-adrenoceptor agonists may also be related to several other actions. For example, chronic treatment of human bronchi with fenoterol enhances contractile responses to endothelin via an effect that may be due to the sensitisation of the transient receptor potential vallinoid 1 channel 171, 172. In addition, repression of eosinophil apoptosis 173, 174, induction of neurokinin (NK)-2 receptor expression 175, 176, and upregulation of histamine H1 receptor expression 177 are all examples of responses that would be considered undesirable in the context of asthma pathogenesis. Interestingly, glucocorticoids tend to reverse many of these effects, demonstrating the highly beneficial potential of combined β2-agonist/corticosteroid therapy 6. Indeed, glucocorticoids enhance β2-adrenoceptor expression and function 178180, resensitise β2-adrenoceptors 181, increase Gsα expression 182, enhance eosinophil apoptosis 173 and reverse the upregulation of NK-2 receptor expression 175.


There are numerous reports in many cells, including ASM and pro-inflammatory cells, documenting the existence of cAMP-dependent, yet PKA-independent, responses. Data accrued since the mid-1990s also provide intriguing evidence that GPCRs may not signal individually but as homo- or even heterodimers or higher-order oligomers. Another major advance has been the unequivocal demonstration of compartmentalisation of cAMP-dependent hormone action. In the following sections, these recent advances are discussed in the context of β2-adrenoceptor signalling and asthma therapy together with data obtained from studies employing transgenic mice overexpressing pulmonary β2-adrenoceptors.

β2-Adrenoceptor oligomerisation

Traditionally, it was thought that GPCRs existed and functioned as discrete monomers, and that the stoichiometry of receptor, G-protein and effector interaction was equivalent (i.e. 1:1:1) 183. However, there are many studies demonstrating cooperativity in agonist binding to GPCRs, which led to the suggestion that each receptor may, indeed, assemble and signal as part of a larger multi-receptor array. Since the early 1990s, the concept of GPCR dimers or higher-order oligomers has gained general acceptance, and persuasive evidence for this phenomenon in transformed cell systems is now available 184. It should be noted that oligomerisation has not yet been demonstrated in nontransformed cells relevant to airways biology and so the significance of the findings discussed below is currently unclear. Nevertheless, the possibility that GPCRs can function as oligomers seems likely and is worthy of discussion in this article. As most current methods cannot distinguish between dimers and oligomers, the former term is used throughout the rest of this discussion as it represents the minimum oligomeric form of the receptor 185. Nevertheless, this term is not meant to exclude the possibility of the existence of higher-order GPCR complexes.

Using the technique of saturated bioluminescence resonance energy transfer (BRET), it has been estimated that 82% of human β2-adrenoceptors expressed on human embryonic kidney (HEK) 293T cells exist as dimers 186, 187. Moreover, studies by Hebert et al. 188 demonstrated that a peptide corresponding to transmembrane domain VI of the β2-adrenoceptor, which features a putative motif (Gly276-X3-Gly280-X3-Leu284) for intramolecular receptor: receptor interactions at the C-terminus, disrupted dimerisation in a concentration-dependent manner. Significantly, this effect was associated with impaired agonist, but not forskolin- or sodium fluoride-, induced AC activation, indicating that dimeric β2-adrenoceptors may be the minimum functional signalling unit.

An important focus of research following the discovery of GPCR dimerisation was the mechanism(s) regulating the formation of receptor complexes. Angers et al. 189 advanced three possible scenarios: 1) GPCR dimers are stable preformed complexes that are constitutively expressed at the cell surface and are unaffected by ligand binding; 2) GPCR dimers are constitutively expressed at the cell surface but ligands can alter the degree of dimerisation; and 3) ligand binding is a prerequisite for GPCR dimerisation. With respect to the β2-adrenoceptor, early studies found a high degree of constitutive dimerisation in Sf9 cells expressing the recombinant protein, which was increased (albeit modestly) by isoprenaline, an observation that favours scenario two 187, 188. However, an emerging consensus for family A GPCRs, which include the β2-adrenoceptor, is that ligands bind to constitutively expressed dimers and that the apparent increase in ligand-induced dimerisation, implied from BRET studies, probably represents conformational changes of pre-existing dimers rather than the formation of new receptor complexes 187, 190.

Thus the question remains, where and how is β2-adrenoceptor dimerisation regulated? It is well established that the assembly within the ER of proteins in general is a common form of quality control used by a cell to permit the export of correctly folded complexes 191. This process also appears to apply to GPCRs. Indeed, immature forms of the β2-adrenoceptor have been recovered as dimers from ER-enriched fractions of HEK 293T cells, suggesting that complex formation takes place early during receptor biosynthesis 192, 193. Further support for this hypothesis comes from the fact that mutant β2-adrenoceptors that lack the normally expressed heterologous export motif do not leave the ER en route to the plasma membrane. Similarly, β2-adrenoceptors harbouring the retention signal, which ordinarily is masked if the protein is folded correctly (i.e. is functionally competent), still dimerise with wild-type receptors, but, again, do not move to the cell surface. Moreover, disruption of the putative dimerisation motif in transmembrane domain VI (see above) prevents normal trafficking of the receptor to the plasma membrane. Thus, dimerisation is seemingly part of the maturation process of β2-adrenoceptors, providing an important mechanism that permits the production of functionally competent ligand recognition and signalling units.

An additional level of complexity is that a GPCR may dimerise with a different, but closely related, receptor. BRET studies have found that β1- and β2-adrenoceptors can form heterodimers with a binding affinity that is similar to homodimeric β2-adrenoceptor complexes 186. Dimerisation of the β2-adrenoceptor with the angiotensin II type 1 receptor 194 and both δ- and κ-opioid receptors has also been described 195. The latter finding has important consequences for signalling, as each member of the dimer may couple to a different G-protein (Gs and Gi, respectively). For example, δ-opioid receptors co-expressed in HEK 293T with β2-adrenoceptors undergo rapid isoprenaline-induced endocytosis. Similarly, etorphine (a δ-agonist) promotes the internalisation of β2-adrenoceptors in the same system 195. However, β2-adrenoceptors, when co-expressed with κ-receptors, undergo neither opioid- nor isoprenaline-induced endocytosis 195. Moreover, isoprenaline promotes the phosphorylation of ERK1 and ERK2 in cells expressing heterodimers composed of β2-adrenoceptors and δ-receptors, which is essentially lost in cells in which the opioid binding partner is replaced with the κ-receptor 195. Thus, in expression studies, the composition of GPCR dimers has a marked influence on G-protein coupling, receptor signalling, agonist-induced desensitisation and ligand pharmacology, and could help provide explanations for many of the effects described in the present review 185, 196. Accordingly, this phenomenon may have significant implications for drug design in the future. Indeed, the crystal structure of the GPCR rhodopsin illustrates that its cytoplasmic surface is too small to accommodate, simultaneously, more than a single point of contact with either the α and βγ subunits of a G-protein 196, 197. This has consequently led to the suggestion that two receptors might be necessary to satisfy the binding requirements of a single G-protein 197, 198. It is clear from the preceding discussion that a number of key questions need to be addressed. These include the relevance and prevalence in native tissues of GPCR homo- or heterodimers 199, and whether the biosynthesis of these novel signalling arrays and the functional responses they mediate are affected by a multitude of factors, including airways disease, sexual dimorphism, drug therapy, age or genetic polymorphisms.

Studies using transgenic mice overexpressing functional β2-adrenoceptors

Elegant studies by McGraw et al. 200 have revealed new insights into the regulation of ASM tone by β2-adrenoceptors. In airway myocytes harvested from the trachea of transgenic mice overexpressing, ∼75-fold, β2-adrenoceptors, basal AC activity and cAMP content are significantly greater than in smooth muscle cells taken from the trachea of nontransgenic litter mates. In addition, isoprenaline-stimulated cAMP accumulation and AC activity are enhanced further in transgenic animals, indicating that, despite the markedly increased receptor density, maximal constitutive activation of AC was not attained. Controversially, these data imply that the β2-adrenoceptor is the rate-limiting component of the receptor/Gs/AC signalling cascade, at least in murine airway myocytes and challenges the contention, for which unequivocal empirical data are lacking, that there exists on ASM a large receptor reserve, at maximal response, for β2-adrenoceptor agonists used in clinical practice 201203. Indeed, if the spare-receptor hypothesis is correct, an increase in receptor density would not be expected to have any major impact on the maximum agonist-induced response.

Functionally, targeted overexpression of β2-adrenoceptors on ASM has marked consequences. Thus MCh-induced bronchoconstriction is significantly reduced in transgenic mice in the absence of a β2-adrenoceptor agonist compared with animals not expressing the transgene 200. Furthermore, the magnitude of bronchoconstriction evoked by MCh in nontransgenic animals in the presence of salbutamol is greater than that achieved in transgenic mice not treated with agonist 200. As identified by the authors, these data are consistent with a multi-state model of GPCRs, in which the effector, AC in this case, is activated by the receptor in the absence of agonist. Obviously, in order to account for this model, it is necessary to propose that the equilibrium in the absence of agonist favours the inactive conformation. In the experiments described by McGraw et al. 200, 75-fold overexpression of ASM β2-adrenoceptors allowed sufficient spontaneous coupling to severely limit MCh-induced tone in the absence of agonist. Assuming the pharmacological behaviour of the murine β2-adrenoceptor can be extrapolated to humans, these transgenic animals exhibit what may be described as an antiasthma phenotype, which tempts speculation that targeted overexpression of β2-adrenoceptors to ASM cells could provide a genetic therapy for asthma 200.

The density of β2-adrenoceptors on airway epithelial cells also has an impact on the tone of the underlying smooth muscle. Using the rat Clara cell secretory protein promoter to enhance, two-fold, cell surface β2-adrenoceptor number on airway epithelial cells in mice, McGraw et al. 204 found that the dose of MCh required to increase, over baseline, airways resistance by 200% was significantly higher compared with that in nontransgenic litter mates. As in the smooth muscle study described above, the protection afforded against MCh-induced bronchoconstriction in the transgenic animals was the same as that produced by inhaled delivery of salbutamol to mice not expressing the transgene 204. These data demonstrate that the density of β2-adrenoceptors on epithelial cells can regulate ASM tone in the absence of agonist, implying that the degree of spontaneous coupling to AC is increased in transgenic mice. The mechanism underlying this protective effect is unknown, but it is not apparently due to the enhanced release from the epithelium of nitric oxide or prostaglandin (PG)E2 204. Thus, the targeted expression of β2-adrenoceptors to the airways epithelium could also provide a novel gene therapy for asthma.

Src tyrosine kinases as effectors of β2-adrenoceptor agonism

In addition to activating the classical cAMP/PKA cascade shown in figure 1a, β2-adrenoceptor agonists have also been shown to stimulate MAP kinase signalling, although the mechanism underlying this effect is not completely understood. In the original description of this phenomenon, Daaka et al. 169 reported that exposure of HEK 293 cells to isoprenaline led to cAMP/PKA-dependent phosphorylation and desensitisation of the β2-adrenoceptor and a consequent switch in receptor coupling from Gs to Gi. Activation of MAP kinase then occurred by the sequential activation of a Src→Son of Sevenless (SOS)→Ras→ERK-dependent signalling cascade initiated by Giβγ subunits 205207. However, in a subsequent study using the same cells, no evidence for Gs:Gi switching was found; indeed, Friedman et al. 170 reported that ERK is phosphorylated by a Src-dependent mechanism mediated by the classical Gsα→AC→PKA cascade (fig. 1d). Regardless of the precise mechanism of isoprenaline-induced ERK activation, these data clearly implicate Src tyrosine kinases in β2-adrenoceptor-mediated signalling. Significantly, this finding is not restricted to HEK 293 cells and seemingly occurs in primary cells of the lung and airways. For example, PP2, a Src tyrosine kinase inhibitor, was recently reported to partially block β2-adrenoceptor-mediated actin depolymerisation of ASM cells 208, 209, by a mechanism that was insensitive to pertussis toxin and the MAP kinase kinase (MKK) 1/2 inhibitor, PD098059. The additional observation that cholera toxin mimicked the effect of isoprenaline 209 is consistent with the data of Friedman et al. 170 that a Gs/Src-mediated pathway in ASM is responsible for this effect (fig. 1d) 209, 210.

An elevation of cAMP levels also leads to the inhibition of both growth responses and ERK activation in fibroblasts 211213. Again, this process appears to involve a PKA-dependent phosphorylation of Src, or a Src-like kinase, leading to the activation of the small GTPase, Rap1, and subsequent repression of the Raf→MKK1/2→ERK cascade 214. However, it is important to note that, in a different cell line, this signalling cascade can also be activated via PKA-dependent phosphorylation of Src 215. Clearly, these data demonstrate the potential diversity of β2-adrenoceptor signalling and highlight the importance of characterising physiologically relevant responses under physiological conditions in nontransformed cells of the tissue of interest.

cAMP-guanine nucleotide exchange factors: novel cAMP-dependent effectors

Until relatively recently, most cAMP-dependent functional responses were generally believed to be mediated by one or more isoforms of PKA. However, in 1998, the world of cAMP signalling underwent radical reshaping, with the discovery of cAMP-activated guanine nucleotide exchange factors (GEFs; cAMP-GEFs), also known as exchange proteins directly activated by cAMP (Epacs) 216218. These GEFs function in a manner similar to SOS, which is the GEF responsible for unloading guanosine diphosphate from Ras and promoting Ras-GTP formation and subsequent activation of the Raf→MKK→ERK cascade 218, 219. Thus, binding of cAMP to cAMP-GEFs allows the activation of an associated small GTPase. In the case of cAMP-GEFI (Epac), Rap1 was suggested to be the downstream G-protein, and has been shown to activate MAP kinase kinase kinase, B-raf and downstream MAP kinase cascades (fig. 1e) 216, 220. However, the situation in respect of cAMP-GEFs was complicated from the outset by an initial report describing the existence of two cAMP-binding GEFs, cAMP-GEFI and cAMP-GEFII (Epac 2) 217. In terms of pulmonary physiology, the role of cAMP-GEFs is currently unclear since neither isoform is apparently expressed at a level that is detectable in adult lung (cf. foetal lung) 217. Furthermore, the initial description of a Rap1–B-raf pathway leading to ERK activation may not necessarily occur via cAMP-GEFs. Indeed, the use of a cAMP-GEF-selective cAMP analogue that activated Rap1 failed to influence ERK activity, suggesting that effects of cAMP on ERK are independent of Rap1 221. In addition, there are a number of other signalling processes and proteins that may also be targeted by cAMP-GEFs, indicating the need to elucidate the functional significance of these novel pathways 218. Nevertheless, cAMP-GEFs are now being widely examined with a view to explaining the increasing number of observations of cAMP-dependent PKA-independent effects, especially since a number of other putative cAMP-binding proteins do not seem to bind cAMP 222.

Activation of PKG

Another key signalling pathway that needs to be considered more carefully in the context of β2-adrenoceptor-mediated relaxation is the cGMP/PKG cascade. Activation of PKG is a well-established mechanism leading to relaxation of smooth muscle 223, including ASM 224226, which may involve the direct phosphorylation and activation of BKCa 227. In addition, PKG can lead to the repression of certain genes that may be of potential significance in airways diseases such as asthma, of which inflammation is a characteristic feature 223, 225, 226. Of particular interest is the finding that K+-induced contractions of guinea pig tracheal segments were more potently inhibited by cell permeant cGMP analogues than by analogues of cAMP 75. This inhibitory effect correlated more closely with the ability of the same analogues to activate PKG rather than PKA, suggesting that, in this tissue at least, PKG plays a dominant role in regulating relaxation 75. The activation of PKG by cAMP 228, which is referred to as cross-activation, has been demonstrated in intact smooth muscle over a concentration range that is remarkably similar to that required to activate PKA, suggesting that it is of physiological relevance (fig.1c) 74, 228. Further exploration of cross-activation has revealed that the binding of cAMP to PKG may promote autophosphorylation, leading to an increase in the affinity of cAMP for the enzyme 229, 230. Taken together, these data suggest that PKG is a strong candidate for mediating cAMP-dependent, PKA-independent responses in the airways 25. This conclusion is supported by the finding that inhibitors of PDE5, a cGMP PDE, may be effective bronchodilators 231 and show independent anti-inflammatory activity 232.

Compartmentalisation of β2-adrenoceptor signalling

Agonist-induced acute homologous desensitisation of β2-adrenoceptors involves receptor phosphorylation by PKA and/or one or more GPCR kinases (GRKs), which have the unique ability to recognise the agonist-occupied form of the receptor 233. In addition, the ability of GRKs to disrupt β2-adrenoceptor signalling through phosphorylation is enhanced ∼10-fold following the binding of scaffold proteins called β-arrestins 234, 235. Recently, elegant studies have extended the scaffold functions of β-arrestin to include certain members of the PDE4D family of enzymes, including PDE4D3 and, in particular, PDE4D5 236, 237. Thus, following agonism of β2-adrenoceptors expressed in HEK 293 cells, a β-arrestin/PDE4D complex forms and is recruited to the receptor, where it limits activation of the cAMP/PKA cascade by simultaneously suppressing AC activity, through receptor desensitisation, and accelerating the removal of cAMP, through enhanced degradation 236. It appears that the unique amino-terminal region of PDE4D5 confers preferential interaction with β-arrestins and may represent the normal binding partner of this scaffold in intact cells 237. This targeting of PDE4D to the β2-adrenoceptor complex may have highly discrete functional implications for the cell as it will selectively regulate the activity of a pool of PKA that is co-localised to the same subcellular microdomain via an interaction with an AKAP 238. Both AKAP5 239 and AKAP12 (gravin) 240, 241 have been shown to be recruited to the β2-adrenoceptor, although only the latter species is thought to be functionally relevant 238. Thus, compartmentalisation of signalling allows the level of cAMP to be controlled very tightly within highly discrete intracellular loci, presumably with unique biological consequences, including the extent to which the β2-adrenoceptor undergoes PKA-dependent phosphorylation 242.


A controversial issue that has received considerable attention in the past is whether β2-adrenoceptor agonists exhibit anti-inflammatory activity. In vitro, it has been known for some time that exposure of purified immune and pro-inflammatory cells, such as eosinophils, mast cells, T-lymphocytes and neutrophils, to β2-adrenoceptor agonists generally results in the inhibition of various functional indices of activation 243. Similarly, acute administration of β2-adrenoceptor agonists to humans effectively suppresses inflammatory leukocyte infiltration and stabilises mast cells in response to direct and indirect stimuli 243. However, there is little evidence that regular administration of these drugs can prevent AHR, the late-phase asthmatic response or the activation in vivo of those cells that initiate and perpetuate the chronic inflammation that characterises asthma 2, 244.

The inability of β2-adrenoceptor agonists to resolve asthmatic inflammation may be due to the development of tolerance (or desensitisation) and is consistent with the rapid loss of responsiveness of essentially all pro-inflammatory and immunocompetent cells following prolonged exposure to β2-adrenoceptor agonists in vitro 2, 10. Two major molecular mechanisms that can account for β2-adrenoceptor desensitisation have been extensively described. One of these promotes short-term homologous refractoriness and involves the uncoupling of the receptor from Gs by mechanisms that require phosphorylation of Ser and Thr residues at the C-terminus of the agonist-occupied receptor 245. This reaction is catalysed by at least three GRK family members (GRK2, GRK3 and GRK5), which are attracted to, and anchored at, the plasma membrane by Gsβγ heterodimers that are liberated following agonist-induced activation of Gs. Signalling through the receptor is then halted by the subsequent binding of β-arrestin, a soluble protein which prevents further coupling to Gs 235. The β2-adrenoceptor is similarly desensitised by PKA following phosphorylation of Ser and Thr residues present within the third intracellular loop of the protein in response to an increase in intracellular cAMP 233, 246. Evidence is also available that Gs can activate Src tyrosine kinases, which have been shown to bind both β-arrestins and the phosphorylated form of the receptor, as well as activate GRKs (fig. 1d) 205, 207. Furthermore, the recruitment of kinases (GRKs and PKA) to the receptor complex during the desensitisation process may be specifically targeted, and enhanced, via interactions with certain AKAPs 240, 247. This AKAP-dependent targeting may also be critical in any later receptor resensitisation 241. Finally, it now appears that, in addition to desensitisation of Gs-coupled signalling (above), the recruitment of PDE4D5 via interaction with β-arrestin and AKAP79 (now AKAP5) is also critical in terminating the PKA-dependent switching of the β2-adrenoceptor to Gi-dependent signalling down to ERK 238.

The other established process that promotes prolonged periods of desensitisation, and which may be of greater clinical relevance 2, is the downregulation of β2-adrenoceptor number, during which physical internalisation and subsequent degradation of the receptor occurs 233. This may involve inhibition of β2-adrenoceptor transcription and/or increased post-transcriptional processing of β2-adrenoceptor mRNA 233. In addition to these well-characterised processes, scrutiny of studies published since the mid-1970s indicates that additional, and in some cases neglected, mechanisms could also play a major role in regulating β2-adrenoceptor signalling and two of these are described below.

Upregulation of phosphodiesterase 4

One mechanism that can contribute to desensitisation is the upregulation of one or more cAMP PDE isoenzymes 248. This can occur through either post-translocational modification (e.g. phosphorylation) of existing enzyme or gene induction 249. With respect to pulmonary β2-adrenoceptor expression, it is the PDE4 isoenzyme family, which is encoded by four genes (PDE4A–PDE4D), that is a primary regulator of cAMP metabolism 250, 251. In this paradigm, tolerance to β2-adrenoceptor agonists is directly related to an increase in PDE activity. This effect would theoretically compromise cell signalling through all Gs-coupled receptors, leading to heterologous desensitisation of susceptible cells to cAMP-dependent events. It is hypothesised that this would occur as a direct consequence of regular treatment with β2-adrenoceptor agonists. Significantly, this model does not exclude the participation of the other established mechanisms of desensitisation, described above. Indeed, phosphorylation of the β2-adrenoceptor by GRKs and PKA could, theoretically, act in concert with cAMP PDE to limit the magnitude and duration of β2-adrenoceptor-mediated signalling 248.

Although generally ignored, the concept of increased cAMP PDE activity as a mechanism of reducing the sensitivity of cells to hormones and other agonists that interact with Gs-coupled receptors is not new. Indeed, evidence that this phenomenon accounts for much of the reduced responsiveness that cells exhibit to chronic hormone exposure was provided in 1978 252, and has since been documented in vitro in many cells implicated in the pathogenesis of asthma, such as T-lymphocytes, neutrophils, monocytes, macrophages, platelets and ASM 154157, 252261. Upregulation of PDE has also been demonstrated empirically in transfection experiments in which the engineered expression of cAMP PDE in yeast and mammalian cells reduces their sensitivity to hormones that augment AC activity 262265.

An important issue that arises from the aforementioned discussion is whether or not induction and/or phosphorylation of PDE4 can be demonstrated in immune/pro-inflammatory cells and in vivo in response to β2-adrenoceptor agonists. Although limited data are available, the answer to both parts of this question is yes. Torphy et al. 154 demonstrated that the β2-adrenoceptor agonist, salbutamol, and the selective PDE4 inhibitor, rolipram, when given in combination to the human monocytic cell line, U937, increased PDE4 activity in a time-dependent manner. Significantly, this effect required new protein synthesis, indicating that the increase in enzyme activity was attributable to the induction of one or more PDE4 isogenes. RT-PCR and Western blot analyses performed by the same authors demonstrated, subsequently, that salbutamol and rolipram increased the expression of PDE4A and PDE4B at both the mRNA and protein level 155. A similar investigation by Verghese et al. 261 essentially confirmed these observations. Thus, exposure of human peripheral blood monocytes and Mono Mac 6 cells to cAMP-elevating agents promoted the transcription of the PDE4A, B and D isogenes, with the generation of at least three distinct mRNA transcripts and proteins. Engels et al. 266 have also reported induction of PDE4 isogenes in U937 and Jurkat T-cells in response to prolonged exposure to dibutyryl cAMP, and, more recently, the same phenomenon was documented in guinea pig macrophages 266, human T-lymphocytes 156, human neutrophils 256 and human ASM cells 157. In the latter study, upregulation of the PDE4D5 splice variant was described, and this may be of particular significance given that this isoform interacts preferentially with β-arrestins and may play a role in β2-adrenoceptor desensitisation (see above) 237, 238.

A consistent and highly significant finding is that the responsiveness of many cells in which PDE4 is induced to cAMP-generating agonists is restored, at least in part, by the addition of a PDE inhibitor, providing compelling evidence that upregulation of PDE is a significant contributory factor to the development of tolerance. In 2000, Finney et al. 267 reported the upregulation of PDE4 in the lungs of rats treated with salbutamol for 7 days. Thus, this phenomenon can be produced in vivo and may be of clinical relevance in the development of tolerance following long-term use of β2-adrenoceptor agonists.

Downregulation of stimulatory G-protein subunit α

Another poorly researched process that could promote heterologous desensitisation of Gs-coupled receptors is a reduction in the abundance of plasma membrane-bound Gsα 268. Finney and co-workers 158, 267 have reported that long-term systemic treatment of rats with salbutamol and salmeterol blocks the ability of these agonists subsequently to protect against acetylcholine-induced bronchoconstriction. Moreover, the bronchoprotective effect of PGE2, which also acts through Gs-coupled prostanoid receptors of the PGE2 receptor 4 subtype in rat airways 269, was similarly abolished, indicating that a state of heterologous desensitisation had been effected. Significantly, further studies found that, in the lungs of rats treated with β2-adrenoceptor agonists, there was a ∼50% reduction in the level of Gsα and an associated impaired ability of cholera toxin to promote cAMP accumulation ex vivo.


A number of additional interactions have been described that extend the multiplicity of β2-adrenoceptor-mediated responses, although none have yet been demonstrated in the airways. In particular, the β2-adrenoceptor features a consensus PDZ domain at its carboxyl terminus that has been shown to interact in an agonist-dependent manner with the PDZ domain of the Na+/H+-exchanger regulatory factor (NHERF) 270. In the absence of β2-adrenoceptor agonist, NHERF binds to the type 3 Na+/H+ exchanger, thereby inhibiting pump activity 271. However, this inhibitory activity is relieved in the presence of agonist, resulting in β2-adrenoceptor-mediated activation of the Na+/H+ exchanger 270. Evidence is also available that the NHERF plays a role in endocytic sorting of β2-adrenoceptors 272. Thus, when bound to NHERF, internalised β2-adrenoceptors are recycled to the plasma membrane, whereas loss of this interaction results in lysosome targeting and receptor degradation 272.

β2-Adrenoceptors have been shown to interact with at least two other proteins: the α subunit of eukaryotic initiation factor 2B 273, which is a nucleotide exchange factor that regulates mRNA translation, and N-ethylmaleimide-sensitive factor 274. The former and latter interactions may have a role in regulating AC activity and β2-adrenoceptor internalisation/recycling respectively.


According to PubMed records, cyclic adenosine monophosphate, since its discovery in 1958, is probably the second-most-studied second messenger, rivalled only by calcium, and has been implicated in a bewildering number of physiological and pathophysiological processes. It is, therefore, perhaps not surprising that the traditional dogma that activators of adenylyl cyclase, exemplified by agonists of the β2-adrenoceptor, exert all of their effects by recruiting a single highly conserved pathway that involves the activation of protein kinase A and the subsequent phosphorylation of target proteins, has been discredited as the sole mechanism of action (fig. 1a). The recently appreciated diversity of β2-adrenoceptor signalling, which is likely to evolve further, may offer clues as to the aetiology of some of the unwanted clinical effects elicited by β2-adrenoceptor agonists and provide opportunities for the future development of novel and safer pharmaceuticals for the treatment of asthma and related respiratory diseases.

  • Received September 27, 2005.
  • Accepted January 25, 2006.


View Abstract