When trafficking and signaling mix: How subcellular location shapes G protein-coupled receptor activation of heterotrimeric G proteins

Abstract

G protein-coupled receptors (GPCRs) physically connect extracellular information with intracellular signal propagation. Membrane trafficking plays a supportive role by “bookending” signaling events: movement through the secretory pathway delivers GPCRs to the cell surface where receptors can sample the extracellular environment, while endocytosis and endolysosomal membrane trafficking provide a versatile system to titrate cellular signaling potential and maintain homeostatic control. Recent evidence suggests that, in addition to these important effects, GPCR trafficking actively shapes the cellular signaling response by altering the location and timing of specific receptor-mediated signaling reactions. Here, we review key experimental evidence underlying this expanding view, focused on GPCR signaling mediated through activation of heterotrimeric G proteins located in the cytoplasm. We then discuss lingering and emerging questions regarding the interface between GPCR signaling and trafficking.

1 |. INTRODUCTION

G protein-coupled receptors (GPCRs) comprise the largest family of signaling receptors and integral membrane proteins encoded by mammalian genomes, and it is estimated that about one-third of all drugs presently used in the clinic target GPCRs.^1–4 GPCRs are seven-transmembrane proteins that are cotranslationally inserted into the endoplasmic reticulum membrane and glycosylated in the endoplasmic reticulum and Golgi apparatus. GPCRs are so-named because many of their signaling effects are mediated through ligand-dependent allosteric coupling to heterotrimeric GTP-binding proteins (G proteins) that associate with cytoplasmic membrane leaflets.^5,6

The beta-adrenergic receptor as a model of GPCR signaling-trafficking relationships.

Early evidence that membrane trafficking is relevant to GPCR signaling emerged through the course of studying desensitization of the beta-adrenergic receptor-coupled adenylyl cyclase activity transduced by the trimeric G protein, Gs.⁷ Short-term agonist exposure (on the order of minutes) resulted in a reversible reduction in cyclic AMP (cAMP) production without changing the total number of receptors detected in the cell, while more prolonged exposure (on the order of several hours or more) produced a downregulation of receptor number⁸ that required new protein synthesis to reverse.⁹ A nondestructive process of receptor “sequestration,” detected by subcellular fractionation and ligand accessibility assays, occurred over an intermediate time scale^10–13 and was later shown to represent the iterative cycling of receptors between plasma membrane and endosomes in the presence of agonist.¹⁴ Agonists were found to stimulate this cycle primarily by promoting ligand-dependent accumulation of receptors in clathrin-coated pits (CCPs) that subsequently internalize irrespective of continued agonist binding,^14,15 and the slower process of proteolytic downregulation was shown to be mediated by receptor sorting to lysosomes from a shared endosome intermediate.^16–18 Ligand-dependent accumulation of beta-adrenergic receptors in CCPs—the key event initiating GPCR entry to the endocytic network—was found to be promoted through receptor phosphorylation mediated by a family of GPCR kinases (GRKs) followed by interaction with arrestin (or beta-arrestin) proteins that act as endocytic adaptors by binding also to the clathrin lattice structure and PIP2.^19–21 GPCR phosphorylation and binding to arrestins were shown previously to attenuate G protein activation,^22–25 and acidification of the endosome lumen was generally recognized to destabilize ligand binding to receptors.²⁶

We focused above on beta-adrenergic receptors as an early and well-studied model of the GPCR signaling-trafficking interface, but considerable diversity can be observed even among closely related GPCR paralogues and splice variants.^27,28 Nevertheless, there are now many examples of GPCRs that undergo regulated endocytosis as described above. General takeaways from studies of such GPCRs are that (a) ligand-induced activation of GPCRs and G proteins is initiated at the plasma membrane, and (b) GPCR endocytosis is associated with events that reduce or terminate G protein activation. These takeaways support a traditional model of the signaling-trafficking interface in which G protein activation is restricted to the plasma membrane and receptors are inactive after endocytosis (Figure 1, left side). It now appears that some GPCRs can retain or recover the ability to activate G proteins after they internalize, and some GPCRs can activate heterotrimeric G proteins from the biosynthetic pathway (Figure 1, right side). Receptor signaling reactions organized in discrete membrane compartments, and dynamically connected through membrane trafficking, is a widespread theme in cell biology.²⁹ Moreover, G protein activation on endomembranes has been recognized since early studies of phototransduction initiated by the light-activated GPCR rhodopsin.³⁰ Ligand-activated GPCRs, however, were long thought to operate exclusively at the plasma membrane or to use endomembranes only to initiate G protein-independent signaling cascades.^31,32 The present review focuses on evidence regarding the hypothesis that ligand-activated GPCRs can activate cytoplasmic G proteins from endosomes and membrane compartments in the biosynthetic pathway, as well as the plasma membrane. We point the reader to other reviews that discuss additional aspects of this concept and mechanisms of G protein-independent signal activation from endosomes.^33–36

FIGURE 1.

“Layers” in the GPCR trafficking/signaling relationship. Layer 1 (left side): Membrane trafficking processes “bookend” receptor signaling by trimeric G proteins from the plasma membrane (step 2), using biosynthetic (step 1) and endocytic (step 3) transport to modulate the number of functional surface receptors. Layer 2 (right side): Additional signaling can be initiated by receptor coupling to cytoplasmic G proteins from intracellular membranes. We propose that these layers function together to determine the integrated cellular response

2 |. G PROTEIN SIGNALING FROM ENDOSOMES: A CONTINUATION OR NEW BEGINNING?

As noted above, the traditional view of cellular GPCR signaling mediated by activation of heterotrimeric G proteins did not require any biochemical activity of receptors in endosomes. Indeed, early efforts to assess the potential of adrenergic receptors to initiate signaling from endosomes detected receptors and adenylyl cyclase activity in the same fraction but failed to detect functional coupling between them.^13,37 Over the last decade, however, the hypothesis that GPCRs can initiate G protein-coupled signaling from endosomes as well as the plasma membrane has gained considerable experimental support.^38–55

Evidence for GPCR-G protein signaling from endosomes.

Broadly considered, four experimental approaches have produced evidence supporting endosomal GPCR-G protein signaling. In the first, agonist application followed by washout showed a persistent component of the cellular response after agonist removal from the extracellular medium.^{39,40,42,45,48,54} In a second approach, the ability of membrane-permeant relative to membrane-impermeant antagonists to reverse the GPCR signaling was assessed; incomplete reversal by the membrane-impermeant antagonist was found.^48,54 In a third approach, GPCR endocytosis was inhibited using genetic or chemical manipulations; endocytic blockade was found to reduce the strength and/or duration of downstream cellular responses.^{39,40,42,45,46,49,50} In a fourth approach, biosensors derived from single-domain antibodies (nanobodies) were used to localize active-conformation GPCRs as well as detect conformational activation of G protein; these studies reported a second phase of GPCR and G protein activation in endosomes, with a brief (seconds to about a minute) refractory period separating the arrival of receptors in endosomes from the second activation phase.^{46,53,54,56,57}

Limitations of the present evidence.

While there is now reasonable evidence that some GPCRs initiate G protein signaling after endocytosis, endomembrane signaling by G proteins is not proven and the present evidence supporting it has limitations and caveats. A potential caveat of agonist washout experiments is that the ligand of interest may not be fully removed. Depending on the ligand and system, complete agonist washout is not trivial to achieve or verify.^58–60 A caveat of genetic manipulations to inhibit endocytosis is that effects develop over a period of days, exceeding the time required for extensive remodeling of the plasma membrane and of the cellular proteome more broadly^61–65; accordingly, genetic manipulations of endocytosis may have more widespread effects on cellular signaling than those resulting directly from blocking endocytosis of a particular GPCR. Chemical inhibitors of endocytosis act more rapidly but have demonstrated potential to produce additional off-target effects that complicate experimental interpretation.^66–68 Studies using conformational biosensors to assess the activation state of GPCRs or G proteins are useful in that they can produce a “direct” location-specific readout, but a limitation of existing conformational biosensors is that they do not report functional signaling. Another caveat is that, depending on experimental conditions, such tools may significantly perturb the conformational landscape of the target that they are intended to sense or block critical signaling interactions.

The field is still grappling with how to deal with these problems. One approach is to combine ligand washout with endocytic inhibitor approaches, so that off-target effects of chemical/genetic inhibitors of endocytosis—as well as possible inefficiencies in ligand washout—can be internally controlled.^39,45,49,54 Another is to deliberately exploit the potential of conformationally selective nanobodies to block GPCR signaling reactions, specifically localizing them to a defined target membrane using chemical recruitment tools and assessing effect on the cellular response.⁶⁹ A third approach, which may also have therapeutic potential with further development, is to generate antagonist ligands that accumulate in endosomes to inhibit activation in this compartment selectively.^49,50

Consequences of endosomal GPCR-G protein signaling.

One reported effect of GPCR-G protein signal initiation from endosomes is a prolonged or sustained cellular response, and another is an increased overall (or integrated) response magnitude.^39,40 Ligand trapping in the endosome lumen is thought to prolong or enhance GPCR responses by favoring ligand rebinding after dissociation from the receptor,⁷⁰ and G protein activation at endosomes may differ inherently from that at the plasma membrane because of location-specific differences in activation or regulatory machineries engaged (see below). GPCR-G protein activation in endosomes (or the trans-Golgi network after endocytic delivery) can also preferentially promote a subset of downstream effects such as GPCR-dependent transcriptional responses.^{39,40,49,56,57,71,72} Studies of GPCR-dependent transcriptional induction identified an additional function of endosomal G protein activation by GPCRs in enhancing cellular discrimination between similar ligands and increasing reliability of the cellular response.^71,73

3 |. UNANSWERED QUESTIONS REGARDING ENDOSOMAL G PROTEIN SIGNALING

Is the mechanism of G protein activation at endosomes different from that at the plasma membrane?

A number of GPCRs that produce a persistent endosome-initiated response also mediate ligand-dependent recruitment of arrestins to this compartment.³³ Whereas arrestin recruitment to the plasma membrane is associated with rapid termination of G protein activation, it has been proposed that arrestin recruitment to endosomes sustains it^42,48 through formation of a complex in which the G protein and arrestin are simultaneously bound to the GPCR.⁴⁸ Not all GPCRs that produce sustained responses from endosomes (such as the delta opioid and luteinizing hormone receptors^45,54) recruit arrestin to endosomes, however, and other GPCRs that activate G proteins at endosomes (such as beta-adrenergic and D1 dopamine receptors ^46,47) produce a transient rather than sustained cellular response. These findings raise the additional question of whether there might exist more than one mechanism of endosomal G protein activation by GPCRs, depending on the particular GPCR activated and/or the time scale of the endosome-initiated response.

What are other players in the signaling cascade and where are they working?

If GPCRs activate G proteins at the endosome, a next important question is whether relevant effectors and cofactors are present in sufficiently close physical proximity to mediate downstream signaling. There is evidence indicating that some G protein-dependent effector proteins are indeed present in the endosome limiting membrane but much remains unknown, and there is also evidence that heterotrimeric G proteins have the potential to traffic or translocate between membranes separately from GPCRs.^74,75 Moreover, considering that plasma membrane and endosome membranes differ broadly in protein and lipid compositions, it is conceivable that there exist numerous mediators or regulators of compartment-specific GPCR signaling and trafficking that are presently unknown. Traditionally, the discovery of GPCR interaction partners has relied on genetic screens or biochemical fractionation. Newer technologies, such as proximity labeling combined with quantitative proteomics, have the potential to enable unbiased interrogation of GPCR-linked interaction networks in intact cells, while achieving both temporal and spatial resolution.^76,77

How can location matter for signaling via diffusible second messengers?

As noted above, endosomal activation of G proteins has been reported to confer selectivity on GPCR-mediated transcriptional control.^49,57,71 It is not clear how this is possible, particularly for cAMP-dependent transcriptional control, because this second messenger chemical is thought to diffuse very rapidly in the cytoplasm. Local signaling by cAMP is known to be enhanced by enforced proximity based on scaffolding, as well as by buffering and local hydrolysis of second messenger.^78–80 It remains unknown if these strategies are sufficient to explain the high degree of signaling selectivity observed in compact cells where diffusion, based on theoretical considerations, would be expected to overwhelm them.^79,81 Further work will be required to more fully delineate how spatial precision of second messenger signaling from the cytoplasm to the nucleus is achieved, particularly in compact cells lacking morphological elaborations capable of restricting diffusional access.

How is endosomal signaling regulated and terminated?

Multiple GPCRs have now been reported to mediate sustained signaling after agonist is removed from the extracellular environment.^{39,40,42,45,48–50,54} These findings raise a critical question: once an endosome is loaded with agonist and activated GPCR, how is activation terminated? Multiple mechanisms for terminating GPCR-G protein signaling from endosomes have been proposed and remain under investigation. A well-studied mechanism for regulating GPCR coupling to G proteins, as discussed above, is through phosphorylation and dephosphorylation of the receptor. This raises the question of the phosphorylation state(s) of GPCRs in endosomes. Beta-adrenergic receptor phosphorylation was found using metabolic labeling techniques to be rapidly induced in response to receptor activation but transient, with receptors isolated from endosomes in a net dephosphorylated state relative to receptors at the plasma membrane.⁸² However, studies carried out using phosphorylation site-specific antibodies reported persistent phosphorylation of GRK sites in the receptor tail and accumulation of GRK-phosphorylated receptor species in endosomes.⁸³ Additional mechanisms that are proposed to terminate GPCR-G protein signaling from endosomes include ligand dissociation or destruction by endosomal acidification or proteolysis of the ligand,²⁶ sequestration of GPCRs from the limiting membrane through transfer to intra-luminal vesicles (ILVs) mediated controlled by ubiquitin-dependent and -independent mechanisms,^84–89 and binding of GPCRs to arrestin-like proteins. GPCR binding to arrestin-like proteins may directly inhibit receptor interaction with G proteins or control signaling indirectly by changing the residence time of receptors in the endosome limiting membrane before tubular exit.^{41,44,90–93} GPCR residence time in the endosome limiting membrane (or in subdomains thereof ) is known to vary considerably between different receptors and to produce effects both on the strength and duration of endosome-initiated signaling.^57,94–97

4 |. G PROTEIN SIGNALING FROM THE BIOSYNTHETIC PATHWAY

While much attention has been focused on the potential of GPCRs to initiate G protein signaling from endosomes, the endosome membrane is not the only internal site at which GPCR-G protein activation has been detected or inferred. There is also evidence that some GPCRs have the potential to signal from the Golgi apparatus as well as the endoplasmic reticulum in some cases. Activation is thought to occur in two basic ways: In the first, the Golgi (or trans-Golgi network (TGN))-associated receptor pool arrives by cotrafficking with ligand from the plasma membrane.⁵⁶ In the second, the relevant pool of GPCRs is derived by retention or retrieval during biosynthetic trafficking; in this case, ligand access is thought to be achieved through transmembrane transporters^69,98 or, for sufficiently hydrophobic ligands, direct permeation through the membrane bilayer.^54,69

A glutamate-activated GPCR, mGluR5, localizes at steady state primarily to the endoplasmic reticulum and outer nuclear membrane of neurons.^99,100 Evidence that the intracellular receptor pool produces a functional response via cytoplasmic G proteins emerged by comparing effects of membrane-permeant and -impermeant ligands on calcium signaling; membrane-impermeant antagonists only partially blocked the cellular calcium response induced by glutamate, and a full response to glutamate required organic anion transporter and/or exchanger activities that enable this ligand to cross membranes.⁹⁸ The beta-1 adrenergic receptor was found to be activated by its physiological agonist, epinephrine, at the Golgi apparatus of cardiac muscle cells, and ligand access was found to require an organic cation transporter activity.⁶⁹ Sufficiently hydrophobic ligands, such as the clinically relevant drug dobutamine, were found to selectively activate the Golgi-associated pool of adrenergic receptors when transporter was inhibited. A Golgi-localized pool of opioid receptors was found to be activated only by small-molecule drugs, apparently by passive partitioning through membranes, while a panel of peptide ligands (including several natively produced opioid agonists) produced activation at the plasma membrane and in endosomes but not at the Golgi apparatus.⁵⁴ Such observations further expand the concept of endomembrane GPCR signaling and suggest new opportunities for achieving selectivity in drug action based on the subcellular location of GPCR activation.

5 |. WHERE ARE WE NOW, AND WHERE DO WE GO FROM HERE?

There is presently considerable experimental support for the hypothesis that GPCRs can activate signaling via cytoplasmic G proteins from internal membrane compartments as well as from the plasma membrane. However, a broad caveat is that much of this knowledge is based on data derived from simplified cultured cell models. Such models are unlikely to fully mimic signaling reactions or regulation that occur in native cell types and tissues, and such models are also notoriously susceptible to distortion by protein overexpression. Improvements in technologies for achieving endogenous gene editing have potential to address the latter point, but this technology carries additional caveats such as clonal bias or adaptation effects that require careful control.¹⁰¹ Such (valid) concerns notwithstanding, simplified cell model systems offer significant advantages for mechanistic study, and we note that many aspects of GPCR signaling and trafficking that were first elucidated using such models have been subsequently validated in native systems. Nevertheless, a critical challenge moving forward is to develop improved approaches for investigating GPCR function under native conditions, while still enabling rigorous control and measurement. Recent efforts to combine genetic mouse models with advanced fluorescence imaging to examine GPCR ligand availability, activation and signaling in intact tissues^45,102 suggest one promising direction for future investigation. Another interesting future direction is toward developing ligands that are able to selectively stimulate or inhibit GPCR activation at one subcellular location relative to another,⁴⁹ both to experimentally manipulate location-based signaling and possibly as a basis to develop new drugs for the clinic with improved therapeutic specificity or efficacy.