Agonist‐induced desensitisation of β₃‐adrenoceptors: Where, when, and how?

Abstract

β₃‐Adrenoceptor agonists have proven useful in the treatment of overactive bladder syndrome, but it is not known whether their efficacy during chronic administration may be limited by receptor‐induced desensitisation. Whereas the β₂‐adrenoceptor has phosphorylation sites that are important for desensitisation, the β₃‐adrenoceptor lacks these; therefore, it had been assumed that β₃‐adrenoceptors are largely resistant to agonist‐induced desensitisation. While all direct comparative studies demonstrate that β₃‐adrenoceptors are less susceptible to desensitisation than β₂‐adrenoceptors, desensitisation of β₃‐adrenoceptors has been observed in many models and treatment settings. Chimeric β₂‐ and β₃‐adrenoceptors have demonstrated that the C‐terminal tail of the receptor plays an important role in the relative resistance to desensitisation but is not the only relevant factor. While the evidence from some models, such as transfected CHO cells, is inconsistent, it appears that desensitisation is observed more often after long‐term (hours to days) than short‐term (minutes to hours) agonist exposure. When it occurs, desensitisation of β₃‐adrenoceptors can involve multiple levels including down‐regulation of its mRNA and the receptor protein and alterations in post‐receptor signalling events. The relative contributions of these mechanistic factors apparently depend on the cell type under investigation. Which if any of these factors is applicable to the human urinary bladder remains to be determined.

Linked Articles

This article is part of a themed section on Adrenoceptors—New Roles for Old Players. To view the other articles in this section visit http://onlinelibrary.wiley.com/doi/10.1111/bph.v176.14/issuetoc

1. INTRODUCTION

In 1967, Lands, Arnold, McAuliff, Luduena, and Brown (1967) proposed a subdivision of β‐adrenoceptors into β₁ and β₂. Soon thereafter, it emerged that some β‐adrenergic‐like responses were not mediated by either of these two subtypes (Furchgott, 1972). This included lipolytic responses in rat white adipose tissue (Harms, Zaagsma, & van der Wal, 1974) and smooth muscle relaxation responses in rat colon (Bianchetti & Manara, 1990) and human urinary bladder (Nergardh, Boreus, & Naglo, 1977), which looked to be mediated by β‐adrenoceptors based on the order of potency of catecholamines. However, the existence of a third subtype, the β₃‐adrenoceptor, only became fully accepted after it was cloned in 1989 (Emorine et al., 1989). β₃‐adrenoceptors have a more restricted expression pattern in humans than β₁‐ or β₂‐adrenoceptors (Michel & Gravas, 2016) and have a unique and species‐dependent ligand recognition profile, including a low affinity for propranolol and several other classic antagonists previously considered to block all β‐adrenoceptor subtypes (Cernecka, Sand, & Michel, 2014). Some agonists including mirabegron and solabegron have higher affinity for the human than the rodent β₃‐adrenoceptor, whereas others including BRL 27,344 and ritobegron have higher affinity for the rodent receptor. Similarly, some agonists including CL 316,243 have greater efficacy at the rodent than the human receptor (Baker, 2005). While β₃‐adrenoceptors play a key role in lipolysis in rodents, particularly in brown adipose tissue, brown adipose tissue is largely lacking in adult humans, and lipolysis in white adipose tissue involves β₃‐adrenoceptor only to a minor extent, if at all, in humans (Lönnqvist et al., 1993). Finally, the mouse β₃‐adrenoceptor gene has splice variants, whereas the human orthologue does not (Evans, Papaioannou, Hamilton, & Summers, 1999).

Recently, β₃‐adrenoceptor agonists have been introduced clinically for the treatment of the overactive bladder syndrome. While they effectively improve bladder symptoms and are well tolerated (Chapple, Cardozo, Nitti, Siddiqui, & Michel, 2014; Ohlstein, von Keitz, & Michel, 2012; Yoshida, Takeda, Gotoh, Nagai, & Kurose, 2018), they are not curative, implying a need for long‐term treatment. Other β‐adrenoceptor subtypes exhibit desensitisation upon extended exposure to an agonist, for instance, in the use of β₂‐adrenoceptor agonists for treatment of preterm labour (Engelhardt et al., 1997; Frambach et al., 2005; Michel, Pingsmann, Nohlen, Siekmann, & Brodde, 1989). This can limit the effectiveness of this treatment for women with preterm labour (The Canadian Preterm Labor Investigators Group, 1992) and animal models thereof (Caritis, Chiao, & Kridgen, 1991; Lye, Dayes, Freitag, Brooks, & Casper, 1992). β₂‐adrenoceptors also exhibit desensitisation upon chronic use in patients with asthma (Salpeter, Ormiston, & Salpeter, 2004), although it is debatable whether this is a treatment‐limiting effect. Therefore, the question arises as to whether β₃‐adrenoceptors similarly undergo agonist‐induced desensitisation and how this affects their therapeutic efficacy in long‐term use.

The β₂‐adrenoceptor is the prototype for studying agonist‐induced regulation of GPCRs (Lefkowitz, 1998). Agonist‐induced β₂‐adrenoceptor desensitisation can occur within seconds to minutes, which is often studied by determining the proxy parameter, cAMP production or receptor internalisation (January et al., 1997). It can also occur after more prolonged agonist exposure (hours to days), which often involves down‐regulation of the receptor at the protein level as a result of internalisation followed by lysosomal degradation and/or reduced mRNA abundance, which in turn can be due to decreased transcription and/or mRNA stability (Bouvier et al., 1989). Moreover, it can involve regulation of G protein expression (Adie & Milligan, 1994) and the function of ACs (Feldman, 1989; Medina‐Martinez & Garcia‐Sainz, 1993) and PDEs (Ortiz, Dasi, Cortijo, & Morcillo, 2000). The desensitising stimulus may be endogenous catecholamines, which exhibit elevated levels in disease states such as congestive heart failure (Maisel et al., 1989; Thomas & Marks, 1978), but may also be exogenously administered catecholamines and xenobiotics, for example, for the treatment of preterm labour (Engelhardt et al., 1997; Frambach et al., 2005; Michel et al., 1989) or congestive heart failure (Mauro & Mauro, 1986). While the mechanisms found to underlie agonist‐induced β₂‐adrenoceptor desensitisation have been qualitatively confirmed for many other GPCRs, the relative roles of the contributing mechanisms as well as the time courses of desensitisation differ among GPCRs. Even closely related receptors such as β₂‐ and β₁‐adrenoceptors may exhibit differential regulation (Marullo, Nantel, Strosberg, & Bouvier, 1995; Michel, Feth, Sundermann, Rascher, & Brodde, 1993).

Based on the relative lack of phosphorylation sites in its C‐terminus (Emorine et al., 1989; Lelias et al., 1993; Muzzin et al., 1991; Nahmias et al., 1991), it was originally assumed that β₃‐adrenoceptors are resistant to agonist‐induced desensitisation. While this assumption was supported by some early studies (Granneman, 1992; Nantel et al., 1993), later work shows that desensitisation can occur in at least some settings (Nantel, Bouvier, Strosberg, & Marullo, 1995). Therefore, this review will concentrate on tissues and cell types in which β₃‐adrenoceptor desensitisation occurs upon prolonged agonist exposure, and what possibly differentiates these from those where desensitization does not occur. We will discuss desensitisation of tissue and cell responses, including those of signal transduction, and the underlying molecular and cellular mechanisms. We will also discuss the implications of β₃‐adrenoceptor desensitisation for chronic treatment with corresponding agonists. Of note, some studies reporting on desensitisation of responses apparently occurring via β₃‐adrenoceptors have not unequivocally established that this receptor is really mediating the response under investigation, for example, in neonatal rat cardiomyocytes. For ease of reading, descriptions of β₃‐adrenoceptor ligands and cell lines mentioned in the manuscript are shown in Table 1.

Table 1.

Characterisation of β₃‐adrenoceptor (AR) ligands and cell lines mentioned in this article

β3‐Adrenoceptor ligands
Ligand	Property
BRL 26,830	β3‐AR‐selective agonist
BRL 37,344	β‐AR with poor selectivity for β3‐AR
CGP 12,177	β1/β2‐adrenoceptor antagonist; partial β3‐AR agonist; β1/β2‐AR radioligand, lower affinity for β3‐AR
CL 316,243	β3‐AR‐selective agonist
D7114	β3‐AR‐selective agonist
Fenoterol	β2‐AR‐selective agonist
FR 149,175	β3‐AR‐selective agonist
Iodocyanopindolol	β1/β2‐AR radioligand, lower affinity for β3‐AR
Isoprenaline	β‐AR antagonist, all subtypes
L 748,337	β3‐AR‐selective antagonist/biased agonist; also radioligand
Mirabegron	β3‐AR‐selective agonist
Ritobegron	β3‐AR‐selective agonist
Ro 16‐8714	β3‐AR‐selective agonist
Salbutamol	β2‐AR‐selective agonist
SB 206,606	β3‐AR radioligand
Solabegron	β3‐AR‐selective agonist
SR 59,119	β3‐AR‐selective agonist
SR 59,230	β‐AR antagonist, all subtypes

Cell lines
Cell line	Origin
3T3‐F442	Mouse adipocyte
CHO	Hamster ovary
CHW	Hamster fibroblast
HEK 293	HEK
L	Mouse sarcoma
Ltk−	Mouse fibroblast
M1	Mouse kidney cortical duct
SK‐N‐MC	Human neuroblastoma

Note: For references on ligands, see Baker (2005, 2010) and Hoffmann, Leitz, Oberdorf‐Maass, Lohse, and Klotz (2004).

2. DESENSITISATION AFFECTING CELL/TISSUE FUNCTION

Desensitisation of cell and tissue function mediated by β₃‐adrenoceptors has been explored in a small number of studies based on pretreatment of animals in vivo or of isolated tissue ex vitro (Table 2). In an early study, hamsters were treated for 6 days with osmotic minipumps delivering noradrenaline or with daily injections of noradrenaline or adrenaline, followed by assessment of lipolytic responses to isoprenaline, noradrenaline, adrenaline, the β_2/3‐adrenoceptor agonist BRL 37,344, or the β_1/2‐antagonist and partial β₃‐adrenoceptor agonist CGP 12,177 in isolated adipocytes (Carpene et al., 1993). While responses to these agonists were unchanged upon treatment, those to the β₁‐adrenoceptor agonist dobutamine and the β₂‐adrenoceptor agonist procaterol were reduced, indicating that in contrast to β₁‐ and β₂‐adrenoceptors, the β₃‐adrenoceptor was refractory to desensitisation. Similarly, oxygen consumption in response to the β₃‐adrenoceptor agonist FR 149,175 remained stable after 2 weeks of treatment with the same compound in Zucker fatty rats (Hatakeyama et al., 2004).

Table 2.

In vivo and in vitro studies on desensitisation of β₃‐adrenoceptor function

Tissue	Species	Response	Intervention	Outcome	Reference
In vivo treatments
White adipocytes	Hamster	Lipolysis	Noradrenaline 5 μg·min−1·kg−1 infusion, 6 days	No change	Carpene et al., 1993
White adipocytes	Rat	Oxygen consumption	FR 149,175, 0.1–3.2 mg·kg−1 twice daily orally, 2 weeks	No change	Hatakeyama, Sakata, Takakura, Manda, & Mutoh, 2004
Ileum	Mouse	Relaxation	CL 316,243, 1 mg·kg−1 injection, 1 hr	No change	Hutchinson, Evans, & Summers, 2000
			CL 316,243, 1 mg·kg−1 injection, 4 or 24 hr	Reduced
Pulmonary artery	Rat	Relaxation	Isoprenaline 0.3 mg·kg−1·day−1 injection, 7 days	No change	Davel, Victorio, Delbin, Fukuda, & Rossoni, 2015
In vitro treatments
Myometrium	Human	Relaxation	SR 59,119, 10 μM, 5 or 15 hr	No change	Rouget et al., 2004
Urinary bladder	Rat	Relaxation	Isoprenaline, CL 316,243, mirabegron 10 μM each, 6 hr	No change	Michel, 2014

Note: For details, see Section 2.

In mouse ileum, a 4‐ or 24‐hr treatment with 1 mg·kg⁻¹ of the β₃‐adrenoceptor‐selective agonist CL 316,243 reduced maximum relaxation in response to freshly added CL 316,243 by 52% and 56%, respectively, whereas the potency of the agonist was not affected; a 1 hr pretreatment did not affect the subsequent relaxation response (Hutchinson et al., 2000). These data indicate the occurrence of desensitisation upon long‐term stimulation but not shorter time stimulation. In vivo treatment with the antagonist SR 59,230 (1 mg·kg⁻¹) for 4 hr increased maximum relaxation to freshly added CL 316,243 by 15% with a minor increase in agonist potency (pEC₅₀ from 7.27 to 7.49). In contrast, a 4‐hr in vivo treatment with 3 mg·kg⁻¹ forskolin did not affect relaxation responses to CL 316,243. Of note, the changes that occurred following treatment with CL 316,243 occurred without changes in receptor binding or mRNA, whereas the changes with SR 59,230A were associated with an increase in β₃‐adrenoceptor mRNA. These data were interpreted as showing agonist‐induced desensitisation of β₃‐adrenoceptor tissue function and sensitisation by an antagonist. The lack of effect of forskolin may indicate that the agonist‐induced desensitisation may not be cAMP‐mediated.

Relaxation of rat isolated pulmonary artery was assessed after pretreatment with vehicle or 0.3‐mg·kg⁻¹·day⁻¹ isoprenaline for 7 days (Davel et al., 2015). As an internal control of the action of isoprenaline, the treatment induced right and left ventricular hypertrophy. The authors used phenylephrine‐contracted artery rings to test responses to various vasodilators including the β‐adrenoceptor agonists, isoprenaline, metaproterenol, and the β₃‐adrenoceptor‐selective agonist mirabegron, the endothelium‐dependent ACh, and the receptor‐independent NO donor nitroprusside. Whether this represents a β₃‐adrenoceptor‐mediated response has not been fully established. Pretreatment with isoprenaline did not alter relaxation responses to any of the three β‐adrenoceptor agonists or to nitroprusside but caused some degree of sensitisation of the ACh response (E_max from 78.4% to 92.7%, pEC₅₀ 7.03 to 7.39). These findings were interpreted to represent a lack of β‐adrenoceptor desensitisation, including that of β₃‐adrenoceptors based on the finding that the response to mirabegron was unaffected, but enhanced NO modulation of vascular tone after isoprenaline treatment. In line with this interpretation, NO levels and cGMP content as well as immunoreactivity for endothelial NOS in the pulmonary artery were increased after isoprenaline treatment. Thus, in vivo studies in three different tissues of two species did not yield a consistent pattern with desensitisation being observed in one but not two other studies; moreover, duration of agonist exposure appears to be a critical factor.

Desensitisation of β₃‐adrenoceptor‐mediated smooth muscle relaxation following ex vivo pretreatment of organ strips with agonists has been explored in two studies. Pretreatment of human near‐term myometrial strips for 5 or 15 hr with β₂‐adrenoceptor agonist salbutamol right‐shifted the concentration–response curve to freshly added salbutamol, indicating desensitisation of the β₂ component of the response, but did not affect relaxation by the β₃‐adrenoceptor agonist SR 59,119; pretreatment with SR 59,119 for the same times did not affect either response, indicating resistance to desensitisation of the β₃‐adrenoceptor component (Rouget et al., 2004). Pretreatment of rat urinary bladder strips with isoprenaline or the β₂‐adrenoceptor agonist fenoterol for 6 hr led to a desensitisation of the β₂‐adrenoceptor component of the relaxation (response to fenoterol), whereas pretreatments with isoprenaline, fenoterol, CL 316,243, or mirabegron caused less if any desensitisation of the β₃‐adrenoceptor component of relaxation (response to CL 316,243 or to mirabegron; Michel, 2014).

Thus, most studies on tissue and cell responses other than signal transduction failed to detect desensitisation of β₃‐adrenoceptor responses. However, the total number of studies on cell and tissue responses other than signal transduction is too limited to draw generalisable conclusions in the face of heterogeneity of species, tissues, and treatment regimens that were investigated.

3. β₃‐ADRENOCEPTOR‐INDUCED DESENSITISATION OF cAMP SIGNALLING

Stimulation of AC leading to the formation of cAMP is the canonical signalling pathway of β‐adrenoceptors, including the β₃‐adrenoceptor subtype (Bylund et al., 1994). While stimulation of AC can be measured directly in studies with cell membrane preparations, for convenience, most investigators have explored desensitisation of β₃‐adrenoceptor‐mediated cAMP formation as cAMP accumulation in intact cells. However, use of cAMP accumulation assays has limitations that may affect the interpretation of the resulting data. Most importantly, cAMP is readily degraded by PDEs. Most investigators try to limit the impact by including PDE inhibitors such as IBMX in their assays. However, how well such inhibitors prevent cAMP degradation and whether they may have other effects, for example, by acting on adenosine receptors, is not well understood for most model systems (Michel & Seifert, 2015). Moreover, accumulating cAMP can leave cells, making intracellular concentrations lower than they should be and concomitantly act on adenosine receptors to inhibit cAMP formation (Pacini, Sanders‐Silveira, & Godinho, 2018). Due to a combination of these two factors, cellular cAMP accumulation is not necessarily linear over time. An alternative only rarely used in desensitisation studies are FRET‐ or BRET‐based cAMP probes (Barak et al., 2008), but this has been used in only one study in the field (Milano et al., 2018). Another complication of AC and cAMP accumulation assays is that β₃‐adrenoceptors not only couple to G_s proteins to stimulate AC but can also couple to G_i proteins to inhibit it (Germack & Dickenson, 2006; Hadi et al., 2013; Sato et al., 2012; Soeder et al., 1999). Thus, unless G_i proteins are inhibited, for instance, by pretreatment with pertussis toxin, AC and cAMP accumulation assays reflect a net effect of stimulation via G_s and inhibition via G_i; in some models, this net effect may be an inhibition rather than a stimulation of cAMP accumulation (Germack & Dickenson, 2006). Of note, changes in cAMP formation may at least partly result from changes in the expression or function of G proteins, AC, or PDEs (see Section 1), but only a few studies have explored this. These factors must be taken into account in the interpretation of desensitisation data at the AC/cAMP level.

Desensitisation studies at the level of AC/cAMP accumulation have been reported in four models representing endogenously expressed β₃‐adrenoceptors, rat adipose tissue, neonatal rat cardiomyocytes, human myometrium, and human SK‐N‐MC neuroblastoma cells (Table 3). In rat white adipose tissue, stimulation for 1 hr with isoprenaline or BRL 37,344 did not desensitise the β₃‐adrenoceptor component of AC stimulation but did markedly desensitise the β₁‐adrenoceptor component of the response (Granneman, 1992). In a follow‐up study, pretreatment with isoprenaline or the β₃‐adrenoceptor agonist BRL 26,830 in vivo, reduced BRL 37,344‐induced AC stimulation ex vivo (Granneman & Lahners, 1992). However, AC responses evoked by the G protein stimulator NaF were not altered, suggesting that desensitisation may have occurred at the receptor level. Cold exposure reduced AC responses to BRL 37,344 in brown adipose tissue of young and senescent rats, although to a smaller extent in the latter (Scarpace et al., 1999), indicating that the ability to undergo desensitisation within a given system may be modulated by additional factors such as ageing. In that model, desensitisation also extended to AC stimulation induced by GTP but not to that mediated by forskolin, suggesting that it may at least partly have occurred at the G protein level. Thus, data obtained in rat adipose tissues are not consistent.

Table 3.

Studies on desensitisation of β₃‐adrenoceptor‐mediated cAMP formation.

Intervention	Outcome	Reference
In vivo treatments of animals
Rat white adipose tissue
Isoprenaline 2 × 200 μg or BRL 26,830, 2‐mg injection, 8 hr	Decrease	Granneman & Lahners, 1992
Rat brown adipose tissue
Cold exposure, 48 hr	Decrease	Scarpace, Matheny, & Thümer, 1999
In vitro treatments of tissues and cells natively expressing β3‐adrenoceptor
Rat brown adipocytes
Isoprenaline 100 μM, 1 hr	No change	Granneman, 1992
Neonatal rat cardiomyocytes
Noradrenaline 100 μM, 24 hr	No change	Germack & Dickenson, 2006
Human myometrium
SR 59,119, 10 μM, 5 or 15 hr	No change	Rouget et al., 2004
SK‐N‐MC human neuroblastoma cells
Isoprenaline 100 μM, 1 hr	Decrease	Chaudhry & Granneman, 1994
Isoprenaline 10 μM, 30 min	No change	Curran & Fishman, 1996
In vitro treatments of cells transfected with β3‐adrenoceptors
Ltk− cells
Isoprenaline 10 μM, 30 min	No change	Nantel et al., 1993
Isoprenaline 10 μM, 0–24 hr	No change	Nantel et al., 1995
CHW cells
Isoprenaline 100 μM, 30 min	No change	Liggett, Freedman, Schwinn, & Lefkowitz, 1993
Isoprenaline 10 μM, 30 min	No change	Nantel et al., 1993
Isoprenaline 10 μM, 0–24 hr	No change	Nantel et al., 1995
CHO cells
Isoprenaline 100 μM, 24 hr	Decrease	Chambers et al., 1994
Isoprenaline 100 μM, 1 hr	No change	Chaudhry & Granneman, 1994
Isoprenaline 10 μM, 15 min to 6 hr	Decrease	Candelore et al., 1996
Isoprenaline 10 μM, 24 hr	No change	Michel‐Reher & Michel, 2013
Isoprenaline 10 μM, 24 hr	Decrease	Okeke, Michel‐Reher, & Michel, 2018
HEK 293 cells
Isoprenaline 100 μM, 1 hr	Decrease	Chaudhry & Granneman, 1994
Isoprenaline 10 μM, 24 hr	Decrease	Vrydag, Alewijnse, & Michel, 2009
Isoprenaline 10 nM to 10 μM, 2–24 hr	Decrease	Michel‐Reher & Michel, 2013
Isoprenaline, L 755,507, CL 316,243, solabegron 10 μM, 24 hr	Decrease	Okeke, Michel‐Reher, & Michel, 2017
Mouse renal epithelial cells
Mirabegron 10 nM, 1–12 hr	No change	Milano et al., 2018
Mirabegron 10 nM, 24 hr	Decrease

Note: cAMP accumulation has been measured as proxy for cAMP formation in some studies. Studies with transfected cells are based on human receptor except for Chaudhry and Granneman (1994), who used both rat and human receptors with similar results. For details, see Section 3.

In neonatal rat cardiomyocytes, treatment with noradrenaline markedly reduced cAMP accumulation in response to noradrenaline, isoprenaline, dobutamine, or procaterol; responses deemed to occur via β₁‐ and/or β₂‐adrenoceptors (Germack & Dickenson, 2006). When β₁‐ and β₂‐adrenoceptors had been blocked during pretreatment, stimulation by isoprenaline turned into inhibition, apparently a β₃‐adrenoceptor‐mediated and pertussis toxin‐sensitive response. This inhibitory β₃‐adrenoceptor response did not exhibit desensitisation. In isolated human myometrial strips, a 15 hr pretreatment with 10 μM salbutamol reduced cAMP accumulation in response to freshly added salbutamol, reflecting desensitisation of the β₂‐adrenoceptor in this tissue; however, pretreatment with either salbutamol or SR 59,119 did not affect cAMP responses to freshly added SR 59,119, suggesting that the β₃‐adrenoceptor in this tissue did not undergo desensitisation (Rouget et al., 2004). In SK‐N‐MC cells, treatment with 100 μM isoprenaline for 1 hr was reported to desensitise the β₃ component of AC stimulation (Chaudhry & Granneman, 1994), whereas treatment with 10 μM isoprenaline for 30 min did not (Curran & Fishman, 1996). However, interpretation of data from SK‐N‐MC cells is complicated by their co‐expression of β₁‐adrenoceptors, which also undergo agonist‐induced desensitisation (Michel et al., 1993). Thus, desensitisation and a lack thereof were both reported in tissues and cells with native expression of β₃‐adrenoceptors However, the number of models being studied is too small to allow robust conclusions, particularly since each model has been studied only once.

Given the heterogeneous results in models endogenously expressing β₃‐adrenoceptors, several investigators have turned to transfected cells as simpler models, potentially allowing more detailed dissection of underlying molecular pathways (Table 3). Upon expression in mouse fibroblast Ltk⁻ cells, two studies from the same group of investigators reported a lack of change in maximum AC stimulation upon pretreatment with isoprenaline for 30 min to 24 hr, but under at least some conditions, agonist potency decreased in pretreated cells (Nantel et al., 1993, 1995). In contrast, a marked reduction of maximum responses was observed in Ltk⁻ cells transfected with β₂‐adrenoceptors. A chimeric receptor consisting of a β₃‐adrenoceptor with substitution of its third cytoplasmic loop and carboxyl‐terminal tail for the corresponding region of the β₂‐adrenoceptors resulted in an intermediate phenotype with a minor reduction in maximum AC stimulation along with a major reduction in agonist potency. Thus, these two regions of the receptor may play a role in desensitisation but apparently do not entirely explain the difference between the two subtypes. Upon expression of β₃‐adrenoceptors in Chinese hamster fibroblast CHW cells, a largely similar observation was reported by the same investigators (Nantel et al., 1993, 1995); no change in maximum AC responses upon pretreatment with isoprenaline for 30 min to 24 hr, whereas these responses were markedly reduced in cells transfected with β₂‐adrenoceptors. However, an additional experiment provided an interesting insight: if cAMP accumulation in intact cells was studied, a time‐dependent desensitisation of the β₃‐adrenoceptor response became detectable; while this was smaller than that of the β₂ response, it pointed to the possibility that desensitisation in intact cells may involve changes in cAMP breakdown or extrusion. Overall, these results indicated that the β₃‐adrenoceptor was less prone to desensitisation than the β₂‐adrenoceptor for stimulation times varying between 30 min and 24 hr and that, in addition to a relative lack of phosphorylation sites present in the third intracellular loop and the carboxyl tail of the β₂‐adrenoceptor, other determinants contribute to the relative resistance of the β₃‐adrenoceptor to desensitisation. Little agonist‐induced desensitisation of cAMP formation has also been observed in murine renal cortical collecting duct cells transfected with the human β₃‐adrenoceptor (Milano et al., 2018).

Data obtained from transfected CHO cells yielded more variable data. Desensitisation of cAMP accumulation was observed in some studies with short‐term (Candelore et al., 1996) and long‐term pretreatment (Chambers et al., 1994; Okeke et al., 2018), whereas no such desensitisation was observed in other studies with short‐term (Chaudhry & Granneman, 1994) or long‐term pretreatment (Michel‐Reher & Michel, 2013; Figure 1). In HEK 293 cells, five studies consistently demonstrated agonist‐induced desensitisation of AC activity and cAMP accumulation with rat (Chaudhry & Granneman, 1994) and human receptors (Chaudhry & Granneman, 1994; Michel‐Reher & Michel, 2013; Okeke et al., 2017; Vrydag et al., 2009; Figure 1). Desensitisation was also observed with a truncated version of the human receptor that lacked the six C‐terminal amino acids encoded by the second exon, which include two serines potentially serving as acceptor sites for phosphorylation. Thus, these and studies with chimeric receptors expressed in Ltk⁻ cells question the role of the C‐terminus of the β₃‐adrenoceptor in desensitisation responses.

Figure 1.

Desensitisation of cAMP accumulation mediated by human β₃‐adrenoceptors transfected into CHO and HEK 293 cells. Cells were treated for 24 hr with 10 μM isoprenaline or vehicle, followed by washout and exposure to fresh isoprenaline. Desensitisation was observed in HEK 293 but not in CHO cells. Reproduced with permission from Michel‐Reher and Michel (2013)

In conclusion, agonist‐induced desensitisation of β₃‐adrenoceptor‐stimulated cAMP formation is consistently absent in some models, for example, Ltk⁻ cells, and consistently present in others, for example, HEK 293 cells, whereas inconsistent data have been reported in several other models including CHO cells. However, in studies where it was compared with the β₂‐adrenoceptor, the β₃‐adrenoceptor was generally found to be less prone to desensitisation than its closely related subtype. These findings, as well as limited findings on cell/tissue function other than signal transduction (see Section 2), highlight the probability that the presence of β₃‐adrenoceptor desensitisation is largely determined by the cell type in which it is expressed.

3.1. Changes in G proteins, AC, and PDEs

The formation of cAMP upon β‐adrenoceptor stimulation is modulated by the abundance and activity of α‐subunits of G_s and G_i and of AC isoforms, whereas the fate of the cAMP formed is modulated by the abundance and activity of PDEs. Studies with β₁‐ and β₂‐adrenoceptors have shown that prolonged agonist exposure may cause a down‐regulation of the α‐subunit of G_s and/or an up‐regulation of the α‐subunit of G_i proteins (Adie & Milligan, 1994; Eschenhagen et al., 1992; Kimura, Miyamoto, & Oshika, 1993; Marzo et al., 1991). Therefore, several groups have explored whether a similar regulation also occurs upon β₃‐adrenoceptor stimulation. To test this involvement functionally, one group treated rats with isoprenaline or BRL 26,830 and found that AC responses to freshly added BRL 37,344 were reduced in adipose tissue; whereas, responses to the direct G protein activator NaF were not altered, suggesting that the attenuated response to the agonist did not occur at the G protein or AC level (Granneman & Lahners, 1992). Others exposed young rats to cold temperature and found that responses to BRL 37,344 and the direct G protein stimulator GTP were similarly desensitised in white adipose tissue, whereas those to forskolin were not, suggesting that desensitisation may have largely occurred at the G protein level (Scarpace et al., 1999). Similar changes did not occur in senescent rats.

More direct assessments have been based on immunoblotting of G protein α‐subunits. In CHO cells transfected with human β₂‐ or β₃‐adrenoceptors, a 7‐hr treatment with isoprenaline concentration‐dependently reduced immunoreactivity of G_s by approximately 50% (Chambers et al., 1994). The time course of the down‐regulation of G_s was comparable with both receptor subtypes, but the concentration–response curve exhibited a lower potency for down‐regulation via the β₃‐adrenoceptor. This could explain the observed desensitisation of isoprenaline‐stimulated AC activity, as no changes in β₃‐adrenoceptor protein expression were observed. A study in HEK 293 cells transfected with human β₃‐adrenoceptors reported a reduction of G_s immunoreactivity upon a 24‐hr treatment with isoprenaline, but the extent of this down‐regulation was only about 20% (Michel‐Reher & Michel, 2013). Within the same study, isoprenaline treatment reduced immunoreactivity of G_i1 by about 50% but did not change that of G_i2, G_i3, or G_q/11. As cAMP formation induced by forskolin desensitised to a similar extent as that by isoprenaline upon pretreatment with isoprenaline, the down‐regulation of G_sα was unlikely to be a major contributor to functional desensitisation of the receptor. Of note, the authors reported immunoreactivity of G protein α‐subunits based on loaded protein content, that is, without correcting for a reference protein. In a follow‐up communication on re‐analysis of the same data (Michel‐Reher & Michel, 2015), isoprenaline treatment in these cells was found to reduce GAPDH immunoreactivity by approximately 30%. Had they normalised to GAPDH expression, as many authors do, the down‐regulation of G_s and G_i1 would have been masked, and the lack of significant regulation of G_i3 would have turned into an apparent up‐regulation. A study with in vivo pretreatment of rats with isoprenaline for 7 days reported a lack of change of expression of G_i1/2 and G_i3 protein in pulmonary artery, when normalised for α‐actin expression (Davel et al., 2015). Taken together, two studies in cell lines consistently reported a down‐regulation of G_s upon treatment with isoprenaline, but it is unclear to what extent this may have contributed to the observed desensitisation of cAMP formation. However, in vitro and in vivo studies did not find an up‐regulation of G_i, suggesting that this did not contribute to desensitisation.

In one of the above studies, performed in HEK 293 cells, pretreatment with isoprenaline reduced cAMP responses to forskolin to a similar extent as those to freshly added isoprenaline, suggesting that desensitisation of the cAMP formation could fully be explained by an alteration in the level of AC (Michel‐Reher & Michel, 2013). Reduced AC activity was confirmed to have a major role in desensitisation in a follow‐up study from these investigators (Okeke et al., 2017). However, the same investigators observed no change in forskolin‐stimulated relaxation in rat urinary bladder after 6 hr of agonist treatment (Michel, 2014); similarly, rat ileum relaxation induced by forskolin was also found to be unaltered after 4 hr of agonist treatment (Hutchinson et al., 2000). A study in pulmonary artery demonstrated that a 7 day pretreatment of rats with isoprenaline reduced PDE5 expression (Davel et al., 2015). While this may be relevant to enhanced endothelial NOS activity in the same study, it is unlikely to have contributed to β‐adrenoceptor function because PDE5 is selective for cGMP, not cAMP; accordingly, relaxation responses to agonists including the β₃‐adrenoceptor selective mirabegron were unchanged. An in vitro study with human myometrium did not detect changes in PDE4 upon pretreatment with β₂‐selective salbutamol or β₃‐adrenoceptor selective SR 59,119 (Rouget et al., 2004). Taken together, these studies suggest that prolonged stimulation of β₃‐adrenoceptors can change the activity and/or expression of AC or PDEs, but the number of reports available is insufficient for a robust conclusion in the face of the divergent outcomes across models.

The local intracellular concentration of cAMP resulting from the net effect of the above factors leads to activation of more distal signalling responses such as activation of PKA or Epac or modulation of expression of various transcription factors. However, these have been investigated only rarely. One study used CHW and mouse sarcoma L cells, each transfected with either β₂‐ or β₃‐adrenoceptors and treated for 10 min or 24 hr with isoprenaline (Nantel et al., 1995). In CHW cells, both short and extended treatment increased PKA activity, irrespective of which subtype had been stimulated. However, in L cells, both subtypes increased PKA activity only after short but not after prolonged stimulation, again similarly with both subtypes.

In conclusion, some studies have presented evidence that components of the signal transduction chain downstream of the β₃‐adrenoceptor may exhibit desensitisation upon prolonged agonist exposure. However, such changes do not appear consistent across model systems and have been studied in too few instances to allow robust conclusions.

β₃‐adrenoceptors couple not only to stimulation of cAMP formation but also to that of other signalling pathways, including phosphorylation of the MAPKs ERK and p38, and to extracellular acidification rates (Gerhardt, Gros, Strosberg, & Issad, 1999; Hadi et al., 2013; Robay, Toumaniantz, Leblais, & Gauthier, 2005; Sato, Horinouchi, Hutchinson, Evans, & Summers, 2007, 2008; Soeder et al., 1999; Steinle, Booz, Meininger, Day, & Granger, 2003). Interestingly, only one report published in abstract form has explored desensitisation of signalling pathways other than cAMP formation (Okeke et al., 2018). While the authors detected desensitisation of ERK phosphorylation in transfected CHO cells, this exhibited a different pattern from that of cAMP accumulation within the same series of experiments. These data suggest that different signalling responses to β₃‐adrenoceptor stimulation may undergo differential regulation upon agonist exposure.

4. CHANGES IN β₃‐ADRENOCEPTOR PROTEIN EXPRESSION, PHOSPHORYLATION, AND LOCALISATION

Two main tools have been used for the detection of receptor expression at the protein level, radioligand binding, and immunodetection; while the former is easier to quantify, the latter often allows better spatial resolution in histological or cytological studies. However, both approaches can be problematic when applied to β₃‐adrenoceptors. Standard β‐adrenoceptor radioligands such as [³H]‐CGP 12,177 or [¹²⁵I]‐iodocyanopindolol have much lower affinity for β₃‐adrenoceptors as compared to β₁‐ and β₂‐adrenoceptors (Niclauß, Michel‐Reher, Alewijnse, & Michel, 2006). The high concentrations required to use them as labels for β₃‐adrenoceptors imply saturation of the other two subtypes, high non‐specific binding and binding to non‐β‐adrenoceptor sites including 5‐HT receptors (Hoyer, Engel, & Kalkman, 1985a,b) and a protein later identified as a member of the TM9SF multi‐spanning membrane protein family (Sugasawa et al., 2001, 1997). Thus, only one group has extensively used [³H]‐CGP 12,177 in intact transfected cells (Baker, 2005, 2010). In transfected cells, not expressing any of the other subtypes, [¹²⁵I]‐iodocyanopindolol appears more suitable and is routinely used (Emorine et al., 1989; Hoffmann et al., 2004; Niclauß et al., 2006; Vrydag et al., 2009). It has also been used in some tissues with native expression of the receptor, but the considerable non‐specific binding has limited interpretation of the data (Hutchinson, Chernogubova, Sato, Summers, & Bengtsson, 2006; Rathi et al., 2003; Roberts, Molenaar, & Summers, 1993; Rouget et al., 2005; Schneider & Michel, 2010). [³H]‐SB 206,606, a tritiated form of the active RR‐enantiomer of the agonist BRL 37,344, has been introduced as a β₃‐adrenoceptor‐selective radioligand (Muzzin et al., 1994) but the reported K _d values of 38–93 nM limit its use. Accordingly, [³H]‐SB 206,606 has only rarely been used (Deng et al., 1996) but has been employed in desensitisation studies (Klaus et al., 1995). More recently, a tritiated version of the antagonist L 748,337 has been validated as a β₃‐adrenoceptor‐selective radioligand (van Wieringen, Michel‐Reher, Hatanaka, Ueshima, & Michel, 2013). Its affinity of 2–3 nM makes [³H]‐L 748,337 the radioligand of choice for the labelling of human β₃‐adrenoceptors but due to a lower affinity in rodents, may be less useful in mice and rats, species that are often used to study β₃‐adrenoceptors. [³H]‐L 748,337 has also been used in desensitisation studies of the human β₃‐adrenoceptor (Michel‐Reher & Michel, 2013).

While many β₃‐adrenoceptor antibodies lack useful target specificity, some have recently been validated based on stringent criteria (Cernecka, Ochodnicky, Lamers, & Michel, 2012; Cernecka, Pradidarcheep, Lamers, Schmidt, & Michel, 2014; Kullmann et al., 2009; Limberg et al., 2010); however, even the usefulness of these antibodies is limited to some species and applications. A reported monoclonal antibody with good validation data (Chamberlain et al., 1999) unfortunately has never become widely distributed and is no longer available even from the original authors. Therefore, antibodies have only played a minor if any role in desensitisation studies. Use of transfected, tagged receptors may be an interesting alternative as they can be detected immunologically in a more robust way (Milano et al., 2018).

GPCRs can be present in a high‐ and low‐affinity state for agonists (van Wieringen, Booij, Shalgunov, Elsinga, & Michel, 2013). The high‐affinity state reflects the receptor‐G protein coupled form and the low affinity, the uncoupled form of the receptor. Accordingly, desensitisation can also be reflected by a shift between these two states. Studies with transfected L cells expressing little if any of other subtypes did not find a shift in high‐ versus low‐affinity state of the β₃‐adrenoceptor upon treatment with 10 μM isoprenaline for 30 min, despite the presence of agonist‐induced down‐regulation of β₃‐adrenoceptors (Nantel et al., 1993). A lack of agonist‐induced shift of agonist affinity state was also observed with a chimeric receptor consisting of a β₃‐adrenoceptor in which the third cytoplasmic loop and the C‐terminus had been replaced with corresponding stretches of the β₂‐adrenoceptor. In contrast, β₂‐adrenoceptors exhibited a reduction in agonist high‐affinity states upon agonist exposure.

Chronic exposure to agonists can lead to a reduction in total cellular receptor number, which has been widely observed with β₁‐ and β₂‐adrenoceptors (Engelhardt et al., 1997; Matthews, Falckh, Molenaar, & Summers, 1996). Only few studies have explored this for the β₃‐adrenoceptor, mostly using [¹²⁵I]‐iodocyanopindolol as the radioligand (Table 4). Treatment with the β₃‐adrenoceptor agonist Ro 16‐8714 for 24–48 hr decreased receptor protein expression in brown adipose tissue in lean and obese Zucker rats (Revelli et al., 1992). In contrast, treatment with CL 316,243 for up to 24 hr did not reduce β₃‐adrenoceptor protein expression in mouse ileum, despite causing desensitisation of relaxation responses (Hutchinson et al., 2000). Several studies in rats (Sun et al., 2011), dogs (Cheng et al., 2001), and patients with congestive heart failure (Moniotte et al., 2001; Napp et al., 2009) reported an increased abundance of β₃‐adrenoceptor protein, although only some of them were based on validated antibodies. Results from a study in a rat model of colitis (Vasina et al., 2008) are difficult to interpret for two reasons. First, it is based on immunohistochemistry; while using a validated antibody (Cernecka, Pradidarcheep, et al., 2014), the use of immunohistochemistry resulted in qualitative rather than quantitative conclusions on expression levels. Second, treatment with the β₃‐adrenoceptor agonist SR 58,611 for 7 days improved the colitis, and the disease associated down‐regulation of the receptor, making it difficult to determine whether the apparent increase in receptor protein expression was a direct consequence of agonist exposure or reflected improvement in the pathology that had caused the down‐regulation.

Table 4.

Studies on regulation of β₃‐adrenoceptor protein and mRNA by agonist exposure

Intervention	Protein	mRNA	Reference
In vivo treatments of animals
Rat brown adipose tissue
Ro 16‐8741, 0.87 mg·kg−1 orally, every 24 hr	Decrease	Decrease	Revelli, Muzzin, & Giacobino, 1992
Noradrenaline 2 × 250 μg injection, 8 hr	n.r.	Decrease	Granneman & Lahners, 1992
Isoprenaline 2 × 200 μg injection, 8 hr	n.r.	Decrease
BRL 26,830, 2 × 2 mg injection, 8 hr	n.r.	Decrease
Mouse brown adipose tissue
CL 316,243, 1 mg·kg−1 injection, 1–24 hr	n.r.	Decreasea	Hutchinson et al., 2000
Rat white adipose tissue
Noradrenaline 2 × 250 μg injection, 8 hr	n.r.	Decrease	Granneman & Lahners, 1992
Isoprenaline 2 × 200 μg injection, 8 hr	n.r.	Decrease
BRL 26,830, 2 × 2 mg injection, 8 hr	n.r.	Decrease
BRL 26,830, 10 mg·kg−1 injection, 4 hr	n.r.	Decrease	Granneman & Lahners, 1995
FR 149,175, 0.1–3.2 mg·kg−1 twice daily p.o., 2 weeks	n.r.	No change	Hatakeyama et al., 2004
Mouse white adipose tissue
CL 316,243, 1 mg·kg−1 injection, 1–24 hr	n.r.	Decrease	Hutchinson et al., 2000
Mouse ileum
CL 316,243, 1 mg·kg−1 injection, 1–24 hr	No change (63–77)	No change	Hutchinson et al., 2000
In vitro treatments of tissues and cells natively expressing β3‐adrenoceptor
Neonatal rat cardiomyocytes
Noradrenaline 100 μM, 24 hr	Increase	Increase	Germack & Dickenson, 2006
Noradrenaline 100 mM, 72 hr	n.r.	Increase	Watts et al., 2013
Hamster brown adipocytes
Isoprenaline 1 μM, 2–8 hr	n.r.	Decrease	Klaus et al., 1995
Noradrenaline 1 μM, 2–8 hr	n.r.	Decrease
D7114, 1 μM, 2–8 hr	n.r.	Decrease
Mouse brown adipocytes
Noradrenaline 100 nM, 24 hr	n.r.	Decrease	Bengtsson, Redegren, Strosberg, Nedergaard, & Cannon, 1996
3T3‐F442 mouse adipocytes
Isoprenaline 100 μM, 4–30 hr	Increase (39)	Increase	Thomas, Holt, Schwinn, & Liggett, 1992
Isoprenaline 100 μM, 2–8 hr	n.r.	Decrease	Granneman & Lahners, 1994
Isoprenaline 100 μM, 4 hr	n.r.	Decrease	Granneman & Lahners, 1995
Human myometrial strips
Salbutamol or SR 59,119, 10 μM, 15 hr	n.r.	No change	Rouget et al., 2004
SK‐N‐MC human neuroblastoma cells
Isoprenaline 100 μM, 2–6 hr	n.r.	No change	Granneman & Lahners, 1994
Isoprenaline 1–100 nM, 1–24 hr	No change (189)	n.r.	Curran & Fishman, 1996
In vitro treatments of cells transfected with β3‐adrenoceptor
CHW cells
Isoprenaline 100 μM, 6–24 hr	No change (5,900)	n.r.	Liggett et al., 1993
Isoprenaline 10 μM, 1–24 hr	No change (281)	No change	Nantel, Marullo, Krief, Strosberg, & Bouvier, 1994
Isoprenaline 10 μM, 2–24 hr	No change (200–350)	n.r.	Nantel et al., 1995
CHO cells
Isoprenaline 100 μM, 0–24 hr	No change (3,000)	n.r.	Chambers et al., 1994
HEK 293 cells
Isoprenaline 100 μM, 1 hr	No change (750)	n.r.	Chaudhry & Granneman, 1994
Isoprenaline 10 μM, 24 hr	No change (306)	n.r.	Michel‐Reher & Michel, 2013
L cells
Isoprenaline 10 μM, 1–24 hr	Decrease (122)	Decrease	Nantel et al., 1994
Isoprenaline 10 μM, 3–24 hr	Decrease (200–350)	n.r.	Nantel et al., 1995
M1 cells
Mirabegron 10 nM, 6–24 hr	Decrease	n.r.	Milano et al., 2018

Note: Where available, baseline β₃‐adrenoceptor density (fmol·mg⁻¹·protein) is shown in parentheses. Note that some of the mRNA expression changes were transient, that is, not found at all time points investigated. Studies in transfected cells refer to the human receptor. For details, see Sections 4 and 5. n.r.: not reported.

^a Decrease observed for β_3a‐ but not β_3b‐adrenoceptor.

Others have used tissue samples or cell lines endogenously expressing β₃‐adrenoceptors with ex vivo or in vitro exposure to agonist (Table 4). Surprisingly, treatment of murine adipocyte 3T3‐F442 cells with isoprenaline time‐dependently increased [¹²⁵I]‐iodocyanopindolol binding (Thomas et al., 1992). A similar increase was observed in neonatal rat cardiomyocytes (Germack & Dickenson, 2006; Watts et al., 2013). Such agonist‐induced increases in β₃‐adrenoceptor expression may relate to the cAMP response elements in the receptor gene (Liggett & Schwinn, 1991; Thomas et al., 1992). In contrast, studies in human myometrial strips (Rouget et al., 2004) or in SK‐N‐MC neuroblastoma cells did not show any alterations in β₃‐adrenoceptor protein expression upon agonist treatment (Curran & Fishman, 1996). The negative studies showed a down‐regulation of β₁‐ or β₂‐adrenoceptors. Taken together, studies with endogenously expressed rat, mouse, or human β₃‐adrenoceptors showed highly heterogeneous results with up‐regulation, no change, and down‐regulation; however, whenever investigated, findings of corresponding mRNA were concordant with those of protein expression (Hutchinson et al., 2000; Revelli et al., 1992; Thomas et al., 1992).

Apparently because of the better detectability of β₃‐adrenoceptor protein (see above), most investigators have explored agonist‐induced regulation at the protein level in transfected cells. In line with the findings in tissues and cells natively expressing β₃‐adrenoceptors, the results appear to be cell type‐dependent (Table 4). A lack of agonist‐induced down‐regulation has consistently been reported in transfected CHW cells (Liggett et al., 1993; Nantel et al., 1995, 1994). While most studies have transfected a β₂‐adrenoceptor as a positive control for a receptor that undergoes agonist‐promoted down‐regulation, a chimeric receptor that comprised the β₃‐adrenoceptor up to proline‐365 of the cytoplasmic tail and the C‐terminus of the β₂‐adrenoceptor was used in one study (Liggett et al., 1993). While the β₃‐adrenoceptor did not exhibit down‐regulation and the β₂‐adrenoceptor did undergo a marked down‐regulation, the chimeric receptor had an intermediate phenotype being closer to that of the β₂‐adrenoceptor, providing additional insight into the molecular basis of down‐regulation. A lack of β₃‐adrenoceptor protein down‐regulation has also been reported in transfected CHO cells (Chambers et al., 1994) and HEK 293 cells (Chaudhry & Granneman, 1994; Michel‐Reher & Michel, 2013). In contrast, agonist‐induced down‐regulation of β₃‐adrenoceptor protein has consistently been observed upon transfection in L cells (Nantel et al., 1995, 1994). This agonist‐promoted down‐regulation of the β₃‐adrenoceptor was found to result from a reduction in the mRNA encoding the receptor and not a change in the half‐life of the protein, pointing to a change in RNA stability as the underlying mechanism (Nantel et al., 1994; see below). Using tagged β₃‐adrenoceptors, their down‐regulation was observed in transfected murine renal epithelial cells; in a time course experiment, this was absent after 1–3 hr, moderate after 6–12 hr, and marked with the 24‐hr treatment (Milano et al., 2018).

Other than down‐regulation, the number of agonist accessible receptors can also change through an altered subcellular localisation. Thus, receptors may migrate from the plasma membrane to an intracellular compartment where they are no longer accessible to hydrophilic agonists such as catecholamines. This process is termed internalisation, sequestration, or endocytosis. It is believed to mainly play a role in short‐term re‐sensitisation and has extensively been explored for β₂‐adrenoceptors (Lefkowitz, 1998). Studies in SK‐N‐MC neuroblastoma cells natively expressing β₃‐adrenoceptors (Curran & Fishman, 1996) and in transfected CHW cells (Liggett et al., 1993; Nantel et al., 1993) did not detect agonist‐induced internalisation of the β₃‐adrenoceptor, with concomitantly expressed β₁‐adrenoceptors (Curran & Fishman, 1996) or transfected β₂‐adrenoceptors (Nantel et al., 1993) serving as an internal positive control. A lack of internalisation was also reported upon expression of human β₃‐adrenoceptors in mouse renal epithelial cells (Milano et al., 2018). The observed lack of β₃‐adrenoceptor sequestration is in‐line with the absence of phosphorylation sites in the C‐terminus of the receptor deemed relevant for desensitisation (Emorine et al., 1989; Lelias et al., 1993; Muzzin et al., 1991; Nahmias et al., 1991), believed to be a molecular prerequisite for internalisation of the other β‐adrenoceptor subtypes (Lefkowitz, 1998).

Phosphorylation of C‐terminal serine and threonine residues by GPCR kinases (GRKs), also known as β‐adrenergic receptor kinases, is considered as an important molecular step in the agonist‐induced desensitisation of β₂‐adrenoceptors since removal of these sites by site‐directed mutagenesis resulted in largely diminished desensitisation (Bouvier et al., 1988; Bouvier, Leeb‐Lundberg, Benovic, Caron, & Lefkowitz, 1987). Also, adding extra phosphorylation sites increases receptor–arrestin interaction (Zindel et al., 2015), an interaction believed to initiate the process of desensitisation by sterically interfering with G protein engagement and promoting receptor endocytosis via the association of β‐arrestin with components of the endocytic machineries (Goodman et al., 1996; Laporte, Oakley, Holt, Barak, & Caron, 2000). The human (Emorine et al., 1989; Lelias et al., 1993), rat (Muzzin et al., 1991) and murine β₃‐adrenoceptor (Nahmias et al., 1991) contain only few C‐terminal serine or threonine residues, and those few are located in positions considered unfavourable for phosphorylation, leading to the hypothesis that this paucity in phosphorylation sites may underlie the generally observed reduced propensity of the β₃‐adrenoceptor to undergo desensitisation when compared with the β₂‐adrenoceptor. To test the functional consequence of this lack of phosphorylation sites, CHW cells were transfected with human β₂‐adrenoceptor, β₃‐adrenoceptor, or chimera thereof (Liggett et al., 1993). In whole‐cell phosphorylation studies with such cells, β₂‐adrenoceptor exhibited isoprenaline‐induced phosphorylation whereas β₃‐adrenoceptors did not, even if a GRK was overexpressed. A chimeric receptor, consisting of the β₃‐adrenoceptor with the cytoplasmic tail of the β₂‐adrenoceptor, however, exhibited phosphorylation. To directly test whether the difference in phosphorylation may be responsible for the different desensitisation observed between the two receptor subtypes, a series of β₃/β₂‐adrenoceptor chimeras were expressed in HEK 293 cells and tested for their ability to undergo internalisation and β‐arrestin recruitment (Angers, 2003). For this study, β‐arrestin recruitment was measured by monitoring agonist‐promoted BRET between β‐arrestin and the receptors (Angers et al., 2000). In contrast to the β₂‐adrenoceptor that promoted a robust recruitment of β‐arrestin and was internalised upon agonist stimulation, no such recruitment or internalisation was observed for the wild‐type β₃‐adrenoceptor (Figure 2). This is consistent with the observation that the β₃‐adrenoceptor does not undergo agonist‐promoted phosphorylation (Liggett et al., 1993), a post‐translational modification needed to favour high‐affinity interaction with β‐arrestin. The introduction of the second intracellular loop and carboxyl tail of the β₂‐adrenoceptor into the β₃‐adrenoceptor (β₃‐3232) restored agonist‐promoted internalisation to some extent (Jockers, Da Silva, Strosberg, Bouvier, & Marullo, 1996; Figure 2) but was insufficient to restore any detectable β‐arrestin recruitment suggesting that a β‐arrestin‐independent mechanism may be involved in the internalisation of the β₃‐3232. A similar lack of β‐arrestin recruitment was also observed for the other chimeras tested (β₃‐3233, β₃‐3323, β₃‐3332, β₃‐3322, and β₃‐3222). These data also show that linear determinants from the β₂‐adrenoceptor, including phosphorylation sites, are not sufficient to instill the high‐affinity interaction with β‐arrestin and suggest that a more complex conformational rearrangement must be required. Interestingly, a chimeric receptor in which the carboxyl tail of the β₃‐adrenoceptor was inserted in the β₂‐adrenoceptor (β2‐2223) showed a reduced but not abolished ability to recruit β‐arrestin. This confirmed a role for the C‐tail phosphorylation sites in the recruitment of β‐arrestin to the β₂‐adrenoceptor. However, it also indicated that while being important to stabilise a high‐affinity interaction, it may not be essential for the initial recruitment. These findings are consistent with a recent observation suggesting that β‐arrestin may have a bitopic mode of interaction involving a distinct domain of the receptor (Cahill et al., 2017; Thomsen et al., 2016).

Figure 2.

Time course of agonist‐promoted internalisation of the β₃‐, β₂‐, and a β₃‐3232 chimeric adrenoceptor. HEK 293 cells stably expressing the β₂‐adrenoceptor, β₃‐adrenoceptor, or the β₃‐3232 (chimeric) were stimulated with 100 μM isoprenaline for the indicated times. Each of the constructs was tagged at their N‐terminus with a Myc‐tag allowing their immuno‐detection using an anti‐Myc antibody. Internalisation was assessed by flow cytometry and expressed as % loss of cell surface receptor immunoreactivity. Reproduced with permission from Angers (2003). The data show that in contrast to the β₂‐adrenoceptor that undergoes a time‐dependent internalisation upon isoprenaline stimulation, the β₃‐adrenoceptor does not. However, introducing the second intracellular loop and the carboxyl‐terminus of the β₂‐adrenoceptor within the β₃‐adrenoceptor backbone (β₃‐3232) restored agonist‐promoted internalisation to some extent

Taken together, these data demonstrate that the presence of agonist‐induced down‐regulation of β₃‐adrenoceptor protein appears to be a cell type‐dependent phenomenon, with a lack of down‐regulation being the more frequent phenotype. In contrast to β₂‐adrenoceptors, β₃‐adrenoceptors apparently do not undergo agonist‐induced internalisation. This may be related to their lack of C‐terminal phosphorylation sites and their limited ability to bind β‐arrestin.

5. CHANGES IN β₃‐ADENOCEPTOR mRNA EXPRESSION

Investigation of mRNA expression can serve to explore whether changes in protein expression can be explained by those of corresponding mRNA abundance. A small number of studies have applied them in this way in the β₃‐adrenoceptor field (Germack & Dickenson, 2006; Revelli et al., 1992; Thomas et al., 1992). However, most investigators have explored alterations in β₃‐adrenoceptor mRNA expression as a proxy for those of protein expression, presumably because mRNA levels are technically easier to quantify for β₃‐adrenoceptors than those of functional protein (see Section 4). Four considerations apply when interpreting mRNA data with regard to functional receptor desensitisation. First, any reduction in mRNA abundance is only likely to translate into changes in receptor responsiveness after hours or days because the half‐life of β₃‐adrenoceptor protein has been estimated at 15–18 hr (Nantel et al., 1994) whereas that of β₃‐adrenoceptor mRNA has been estimated at 30–90 min (Bengtsson et al., 1996; Granneman & Lahners, 1995; Klaus et al., 1995). Hence, changes in mRNA expression have mostly been studied in the context of extended (hours to days) exposure to agonists. Second, any change in mRNA abundance may result from those of transcription and/or mRNA stability. While altered mRNA expression upon agonist treatment mainly reflects mRNA stability for β₂‐adrenoceptors (Danner, Frank, & Lohse, 1998; Hadcock, Wang, & Malbon, 1989), it appears to reflect transcription for β₃‐adrenoceptors (Bengtsson et al., 1996; Granneman & Lahners, 1995). Of note, studies on transfected receptors provide limited insight into possible changes in receptor transcription as this parameter is typically not controlled by the native promoter of the receptor but rather by an artificial one such as that derived from cytomegalovirus. Hence, with few exceptions (Nantel et al., 1994), studies on agonist effects on β₃‐adrenoceptor mRNA abundance have been reported in cells and tissues natively expressing the receptor. Third, the promotor region of the β₃‐adrenoceptor contains four cAMP response elements (Liggett & Schwinn, 1991; Thomas et al., 1992) and an intron containing a putative enhancer element specific for adipose tissue (Granneman, Lahners, & Rao, 1992). This may at least partly explain tissue‐selective regulation of mRNA expression. However, there are also less well‐defined mechanisms leading to inhibition of transcription upon agonist exposure (Granneman & Lahners, 1995). The net effect of these mechanisms in each cell type is difficult to predict. Finally, the predictive value of mRNA levels for corresponding β₃‐adrenoceptor protein expression has been questioned (Nantel et al., 1994).

Several investigators have treated lean and obese Zucker rats (Granneman & Lahners, 1992; Revelli et al., 1992), hamster (Klaus et al., 1995), or mice (Hutchinson et al., 2000) in vivo with agonists including isoprenaline, noradrenaline, BRL 26,830, CL 316,243, D7114, or Ro 16‐8741 for 1–48 hr and mostly observed a down‐regulation of β₃‐adrenoceptor mRNA. In some cases, such down‐regulation was transient and in mice only the β_3a was affected but not the β_3b splice variant of the receptor. Cold exposure, considered to stimulate a release of endogenous noradrenaline, also yielded a transient down‐regulation (at 1–12 hr but not 24 hr) in mice (Bengtsson et al., 1996) and did not exhibit down‐regulation at later time points (48–72 hr) in rats (Granneman & Lahners, 1992; Scarpace et al., 1999). Conversely, surgical denervation increased β₃‐adrenoceptor mRNA expression in rats (Granneman & Lahners, 1992). Acute treatment with noradrenaline similarly reduced β₃‐adrenoceptor mRNA expression in white and brown adipose tissues in rats (Granneman & Lahners, 1992). A study in mice confirmed a similar reduction in white and brown adipose tissues for the β_3a‐adrenoceptor, but a reduction in β_3b‐adrenoceptor mRNA was only observed in white adipose tissue (Hutchinson et al., 2000). A similar regulation of both murine splice variants has also been reported by others in vivo and in rat adipocyte 3T3‐F442 cells (Granneman & Lahners, 1995). In contrast, treatment of Zucker fatty rats with the β₃‐adrenoceptor agonist FR‐149,175 did not change β₃‐adrenoceptor mRNA expression (Hatakeyama et al., 2004).

Investigations with in vitro treatment of tissues and cells natively expressing β₃‐adrenoceptors have yielded conflicting results. Increased mRNA expression was reported in 3T3‐F442 mouse adipocytes following treatment with isoprenaline by some (Thomas et al., 1992), but a decreased expression was reported by others (Granneman & Lahners, 1995). The decrease was mimicked by treatment with 8‐Br‐cAMP and found to be due to reduced transcription and not to altered mRNA degradation. In neonatal rat cardiomyocytes, treatment with noradrenaline increased β₃‐adrenoceptor mRNA expression (Watts et al., 2013). Similarly, in congestive heart failure, a condition associated with an increased sympathetic tone, an increased cardiac β₃‐adrenoceptor mRNA expression is consistently exhibited in rats (Ozakca et al., 2013; Zhao et al., 2013), dogs (Cheng et al., 2001), and humans (Moniotte et al., 2001). In SK‐N‐MC human neuroblastoma cells, treatment with 8‐Br‐cAMP increased β₃‐adrenoceptor mRNA expression, whereas treatment with isoprenaline had no effect (Granneman & Lahners, 1994). Similarly, dexamethasone and the PKC activator phorbol myristate acetate did not change β₃‐adrenoceptor mRNA levels in SK‐N‐MC cells but reduced it in 3T3‐F442 cells (Granneman & Lahners, 1995). This reduction by phorbol myristate acetate in 3T3‐F442 cells was confirmed by other investigators (Feve, Pietri‐Rouxel, El Hadri, Drumare, & Strosberg, 1995). Treatment of human myometrial strips with 10 μM salbutamol or SR 59,119 also failed to change β₃‐adrenoceptor mRNA expression (Rouget et al., 2004). Among the limited studies with transfected β₃‐adrenoceptors, reduced mRNA expression was reported upon expression in L cells (Nantel et al., 1995) but no change with expression in CHW cells (Nantel et al., 1994).

In conclusion, data on agonist‐induced regulation of β₃‐adrenoceptor mRNA expression are inconclusive when interpreted across tissues (Table 4). However, they appear more consistent within a tissue, further supporting the view that regulation of β₃‐adrenoceptor presence and function depends critically on cellular context, in the majority of cases yielding a down‐regulation. Of note, agonist‐induced up‐regulation has also been reported in some cases. The molecular basis for this appears to involve cAMP‐mediated inhibitory and stimulatory effects on β₃‐adrenoceptor gene transcription, the balance of which may be cell type‐specific.

6. EVIDENCE OF SYNTHESIS AND CONCLUSIONS

Based upon a lack of consensus phosphorylation sites in its C‐terminal and third intracellular domain, it had been assumed that β₃‐adrenoceptor‐mediated functions are resistant to desensitisation. The above data demonstrate that resistance to desensitisation is present in many but not all tissues and cell types (Figure 3). When desensitisation is observed, it often is less pronounced and/or requires longer agonist exposure as compared to that of β₁‐ and β₂‐adrenoceptors. Moreover, the presence of desensitisation and the molecular level at which it occurs depend strongly on the cell type in which the receptor is expressed.

Figure 3.

Schematic diagram of key aspects of β₃‐adrenoceptor function and regulation. In many if not most tissues and cell types, these do not change upon prolonged agonist exposures, but in some, each of them may decrease (except for G_i, which may increase). For details, see main text

6.1. What are the molecular alterations underlying β₃‐adrenoceptor desensitisation?

All available data suggest that β₃‐adrenoceptors do not undergo internalisation upon agonist exposure (Curran & Fishman, 1996; Liggett et al., 1993; Milano et al., 2018; Nantel et al., 1993). Given that there is strong evidence that internalisation requires phosphorylation (Lefkowitz, 1998) and that such phosphorylation sites are missing in the β₃‐adrenoceptor (Emorine et al., 1989), this is not too surprising.

Down‐regulation of β₃‐adrenoceptor protein upon agonist exposure has been reported only in one in vivo and several studies with transfection into Ltk⁻ cells, whereas most studies have found no change or even increases in receptor expression (Table 4). The technical difficulties of detecting β₃‐adrenoceptors may explain this scarcity of data (see Section 4). Many more studies have demonstrated a down‐regulation of β₃‐adrenoceptor mRNA (Table 4), but in most cases, we do not know whether this translates into changes in expression of functional receptor protein. Apparently, reduced mRNA expression is caused by changes in transcription, not those of mRNA stability (Bengtsson et al., 1996).

Interestingly, desensitisation of β₃‐adrenoceptor‐mediated function can occur despite a documented lack of down‐regulation of receptor protein (Chambers et al., 1994; Chaudhry & Granneman, 1994; Hutchinson et al., 2000; Michel‐Reher & Michel, 2013), indicating that changes at the post‐receptor level must contribute to desensitisation. Such changes at the level of G protein expression (Chambers et al., 1994; Michel‐Reher & Michel, 2013), cAMP formation in the presence of direct G protein stimulators (Scarpace et al., 1999), and AC activity (Michel‐Reher & Michel, 2013; Okeke et al., 2017) have been reported, but examples also exist where desensitisation was not accompanied by changes at these levels (Davel et al., 2015; Granneman & Lahners, 1992; Hutchinson et al., 2000; Scarpace et al., 1999).

β₃‐adrenoceptor desensitisation can be induced in some models by agents not acting on the receptor, for example, forskolin (Michel‐Reher & Michel, 2013) or agonists of other G_s‐coupled GPCRs. This points to a role for cAMP in mediating desensitisation. However, the opposite can also occur; the β₃‐adrenoceptor gene contains cAMP response elements (Liggett & Schwinn, 1991; Thomas et al., 1992), and an elevation of intracellular cAMP can increase receptor expression and/or function (Germack & Dickenson, 2006; Thomas et al., 1992). This raises the possibility that, at least in some cell types, the lack of desensitisation observed does not represent a true resistance but rather the net effect of ongoing sensitisation and desensitisation, but this has not been explored sufficiently for firm conclusions to be reached. Based on limited data, it appears that desensitisation may be independent of β₃‐adrenoceptor gene polymorphisms (Vrydag et al., 2009). As β₃‐adrenoceptors couple not only to cAMP formation but also to other signalling pathways such as ERK or p38 phosphorylation, it is surprising that with one exception (Okeke et al., 2018), desensitisation of these alternative signalling pathways has not been tested.

6.2. Why is desensitisation tissue/cell type specific?

Both the presence of desensitisation and the level at which it occurs appear to be highly cell type‐dependent. While dependency on cell type has been observed for desensitisation of several GPCR that bind arrestin (Tobin, Butcher, & Kong, 2008), it is not fully clear how this can be explained for β₃‐adrenoceptors. The presence of an intron containing a putative enhancer element specific for adipose tissue (Granneman et al., 1992) may at least partly explain tissue‐selective regulation of mRNA expression. Moreover, the presence of desensitisation is modulated within a tissue by additional factors such as age of the animal or innervation status. Understanding the molecular mechanisms underlying such differential desensitisation could help to devise strategies to exploit the presence or absence of desensitisation in a therapeutic manner.

6.3. Desensitisation in disease states

Many GPCRs undergo changes in disease states (Insel, Tang, Hahntow, & Michel, 2007), a phenomenon that has perhaps best been studied for β₁‐adrenoceptors, which are desensitised and down‐regulated in patients with congestive heart failure and animal models thereof (Brodde, 1991; Hammond, 1993). However, the regulation of β₃‐adrenoceptor expression in disease may differ as it appears to be up‐regulated in heart failure at the mRNA and/or protein level (Cheng et al., 2001; Moniotte, Kobzik, et al., 2001; Napp et al., 2009; Ozakca et al., 2013; Sun et al., 2011; Zhang et al., 2005; Zhao et al., 2013). These data have led to the hypothesis that β₃‐adrenoceptor agonists could be useful in patients with heart failure (Rasmussen, Figtree, Krum, & Bundgaard, 2009). An up‐regulation of cardiac β₃‐adrenoceptor expression at the mRNA level has also repeatedly (Amour et al., 2007, 2008; Dinçer et al., 2001; Kayki‐Mutlu, Arioglu Inan, Ozakca, Ozcelikay, & Altan, 2014) but not unequivocally (Haley, Thackeray, Kolajova, Thorn, & DaSilva, 2015; Jiang et al., 2015) been observed in animal models of diabetes or of hypothyroidism (Arioglu, Guner, Ozakca, Altan, & Ozcelikay, 2010). Nevertheless, β₃‐adrenoceptors were also reported to be down‐regulated in an animal model of colitis, where treatment with an agonist improved symptoms and reversed the down‐regulation (Vasina et al., 2008). However, our overall knowledge on changes in β₃‐adrenoceptor expression and function in disease is too limited and needs expansion.

6.4. Conclusions and clinical implications

Based on known β₃‐adrenoceptor functions, it is anticipated that potential clinical uses of ligands for this subtype will be limited to agonists (Dehvari, da Silva Junior, Bengtsson, & Hutchinson, 2018; Michel, Ochodnicky, & Summers, 2010). For such uses, a relative resistance to desensitisation is likely to be a desirable property. For instance, desensitisation is a treatment‐limiting factor in the use of β₂‐adrenoceptor agonists in women with preterm labour (Engelhardt et al., 1997; Frambach et al., 2005; Michel et al., 1989). Tocolysis has also been proposed as an indication for β₃‐adrenoceptor agonists (Bardou et al., 2007; Ursino, Vasina, Raschi, Crema, & De Ponti, 2009); however, whether tocolytic responses to β₃‐adrenoceptor agonists are indeed resistant to desensitisation remains to be studied. The only validated clinical use of β₃‐adrenoceptor agonists is the treatment of the overactive bladder syndrome (Chapple et al., 2014; Ohlstein et al., 2012). While the value of the reported lack of desensitisation in in vitro animal studies (Michel, 2014) remains to be determined, a 1‐year study in patients presented no evidence of desensitisation for the same agonist (Chapple et al., 2013) but may not have included sufficiently early time points to assess desensitisation during the first 2 weeks of treatment. The key translational challenge of β₃‐adrenoceptor desensitisation studies discussed here is the remarkable heterogeneity between species, tissues/cell types, and functional assessments; this makes it difficult to predict what will happen in the target species (humans) and tissue (e.g., urinary bladder or myometrium).

6.5. Nomenclature of targets and ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY (Harding et al., 2018), and are permanently archived in the Concise Guide to PHARMACOLOGY 2017/18 (Alexander, Christopoulos et al., 2017; Alexander, Fabbro et al., 2017).

CONFLICT OF INTEREST

M.C.M. is a consultant to Astellas and a consultant and shareholder of Velicept.

Okeke K, Angers S, Bouvier M, Michel MC. Agonist‐induced desensitisation of β₃‐adrenoceptors: Where, when, and how? Br J Pharmacol. 2019;176:2539–2558. 10.1111/bph.14633