GPCR Intracellular Loop Regulation of Beta-Arrestin-Mediated Endosomal Signaling Dynamics

Abstract

G protein-coupled receptors (GPCRs) are currently appreciated to be routed to diverse cellular platforms to generate both G protein-dependent and -independent signals. The latter has been best studied with respect to β-arrestin-associated receptor internalization and trafficking to signaling endosomes for extracellular signal-regulated kinase (ERK) activation. However, how GPCR structural and conformational variants regulate endosomal ERK signaling dynamics, which can be central in neural development, plasticity, and disease processes, is not well understood. Among class B GPCRs, the PACAP-selective PAC1 receptor is unique in the expression of variants that can contain intracellular loop 3 (ICL3) cassette inserts. The nervous system expresses preferentially the PAC1Null (no insert) and PAC1Hop (28-amino acid Hop insert) receptor variants. Our molecular modeling and signaling studies revealed that the PAC1Null and PAC1Hop receptor variants can associate with β-arrestin differentially, resulting in enhanced receptor internalization and ERK activation for the PAC1Hop variant. The study amplifies our understandings of GPCR intracellular loop structure/function relationships with the first example of how the duration of endosomal ERK activation can be guided by ICL3. The results provide a framework for how changes in GPCR variant expression can impact developmental and homeostatic processes and may be contributory to maladaptive neuroplasticity underlying chronic pain and stress-related disorders.

G protein-coupled receptors (GPCRs), the largest family of transmembrane proteins in the human genome, function to interpret diverse extracellular regulators and cues, including hormones, neurotransmitters, odorants, tastants, and light, into intracellular signals for homeostatic regulatory responses. GPCRs can transduce the signals through G protein-dependent or -independent signaling pathways; in contrast to plasma membrane delimited G protein-dependent signaling, the G protein-independent signaling mechanisms – best studied with respect to β-arrestin-associated receptor internalization and trafficking to signaling endosomes for extracellular signal-regulated kinase (ERK) activation (Luttrell and Luttrell 2003; Shenoy and Lefkowitz 2011; Smith and Rajagopal 2016) (Fig. 1) – often drive long-lasting cellular effects. In particular, GPCR vesicular endocytosis allows a reprogramming of cell surface receptor signaling events to endosomal platforms that can deliver second messengers to intracellular sites with high spatiotemporal resolution and dynamics (Sposini and Hanyaloglu 2017; Walther and Ferguson 2015; West and Hanyaloglu 2015). Despite implications of sustained endosomal GPCR ERK signaling in neural development, plasticity, and participation in maladaptive physiological responses, how the many GPCRs can deliver variable endosomal ERK signaling dynamics remains largely elusive.

Fig. 1.

Biochemical and structural overview of PAC1R system. A Activation of the ERK signaling pathway though PACAP binding to PAC1R proceeds via G-protein coupled-dependent and independent pathways. Abbreviations are adenylate cyclase (AC), protein kinase A (PKA), phosphoinositide phospholipase C (PLC), protein kinase C (PKC), mitogen-activated protein kinase (MEK), and extracellular signal-regulated kinases (ERK). Homology models of PACAP (magenta) bound PAC1 (green) null B and HOP C receptor isoforms bound to β-arrestin (yellow). In C the HOP, ICL3 is highlighted in blue and the two clathrin binding sites on β-arrestin are circled with the unstructured N-domain clathrin-binding box site (i) and the second C-edge site (ii). D The two PAC1R isoform ICL3 sequences are shown; the 28-amino acid HOP insert in blue letters

Among GPCRs, the differences in receptor-β-arrestin-mediated endocytosis and signaling dynamics can be dependent on the distinct conformational states of the GPCR-β-arrestin complexes. The more than 800 GPCRs share a heptahelical transmembrane structure and variations particularly in the intracellular loops (ICLs) that can participate in transducer docking, may be critical determinants driving diverse signaling responses. As the length of the ICL3 segment can range from 20 to more than 100 amino acids, the conformation of ICL3 and its interactions with β-arrestin in the GPCR/β-arrestin plasma membrane complex may have critical roles in generating diverse yet tunable endosomal signals.

To evaluate these possibilities, we have investigated the roles of the endogenous ICL3 variants of a model class B GPCR – the human PAC1 receptor (ADCYAP1R1; Fig. S1) for the pituitary adenylate cyclase activating polypeptides (PACAP, ADCYAP1). PACAP belongs to the VIP/secretin/glucagon family of peptides and has well established roles in regulating endocrine hormone production and secretion, cardiovascular responses, nociception, gastrointestinal motility, glucose metabolism, micturition, and germ cell development (Blechman and Levkowitz 2013; Harmar et al. 2012; Sherwood et al. 2000; Vaudry et al. 2009). In the central nervous system (CNS), high PACAP expression in hypothalamic and limbic neurocircuits has important roles in regulating feeding/satiety in body weight maintenance, autonomic responses, nociceptive sensitivity, cognition and learning/memory, and stress-related behavioral responses (Harmar et al. 2012; May and Parsons 2017; Vaudry et al. 2009). Maladaptive PACAP/PAC1 receptor signaling has been associated with chronic stress-related psychopathologies including posttraumatic stress disorder (PTSD) (Hammack et al. 2009; Hammack and May 2015; Johnson et al. 2020a, 2020b; May and Parsons 2017; Ressler et al. 2011; Roman et al. 2014), and notably in recent work, PACAP function in the spino-parabrachioamygdaloid tract has been associated with nociceptive hypersensitivity to convey the emotional component of chronic pain (Missig et al. 2017, 2014). Although PACAP and VIP can share function at the VPAC1 and VPAC2 receptors (VIPR1 and VIPR2 GPCRs respectively), these maladaptive disorders are recapitulated by the PAC1-selective agonist maxadilan and not VIP, demonstrating that the responses are mediated specifically by the PAC1 receptor. Hence, the PAC1 receptor has been considered an emerging therapeutic target for chronic pain and stress-related disorders, and represents a model to study how ICL3 variants within GPCR systems can alter signaling for homeostatic regulation or drive disease states.

Within the class B receptor family, there are several elements unique to the PAC1 receptor system. In the large PAC1 receptor extracellular domain (ECD), there is a unique 21-amino acid insert between the β3/β4 antiparallel sheets that appears to form a zipper with extracellular loop 3 (ECL3) to help maintain the receptor in an open state (Blechman and Levkowitz 2013; Liao et al. 2019b, 2017; Pantaloni et al. 1996). Critically, however, there are also PAC1 receptor variants resulting from the absence or presence of two 84-base pair Hip and/or Hop cassette inserts (28 amino acids) into regions corresponding to the GPCR ICL3 (Blechman and Levkowitz 2013; Braas and May 1999; Harmar et al. 2012; Spengler et al. 1993). Despite the number of potential PAC1 receptor variants from alternative splicing, i.e., receptor with Null (neither Hip nor Hop), Hip, Hop1/2, or HipHop inserts, the nervous system preferentially expresses the PAC1Null and PAC1Hop receptors (Figs. 1 and S1). In short, the PAC-1Null ICL3 is 27 amino acids in length and the PAC1Hop ICL3 segment is 55 amino acids long. The PAC1 receptor ICL3 variants have been suggested to potentially fine-tune downstream Gs and Gq signaling for adenylyl cyclase/cAMP (AC/cAMP) and phospholipase C (PLC)/DAG/IP3 activation, respectively (Spengler et al. 1993).

As with many GPCRs, PAC1 receptors are also potently and efficaciously coupled to ERK activation. ERK signaling within the nervous system has been best studied with respect to its trophic functions in development and injury, including the promotion of neuronal survival, proliferation, cell fate specification, differentiation and maturation, and neural and synaptic plasticity. Whereas these ERK-mediated effects help maintain physiological homeostasis, ERK-mediated plasticity can also be maladaptive contributing to several disorders. Enhanced ERK signaling, for example, has been associated with stress/fear memory consolidation, stress-related behavioral abnormalities (Atkins et al. 1998; Duvarci et al. 2005; Maldonado et al. 2014; Meller et al. 2003; Schafe et al. 2000, 2008; Shen et al. 2004), and chronic pain (Carrasquillo and Gereau 2007; Fu et al. 2008; Ji et al. 1999, 2009; Li et al. 2011; Melemedjian and Khoutorsky 2015; O’Brien et al. 2015; Yarwood et al. 2017). In particular, endosomal ERK signaling through a variety of GPCRs has been shown to facilitate prolonged visceral pain and other maladaptive physiological responses (Gottesman-Katz et al. 2021; Jensen et al. 2017; Jimenez-Vargas et al. 2020, 2018; Manglik et al. 2016; Ramírez-García et al. 2019; Stoeber et al. 2018). Specifically for PACAP/PAC1 receptor endosomal ERK signaling, our in vivo studies have shown that the PACAP-induced anxiety-related behaviors, pain hypersensitivity, and stress-induced relapse in addiction paradigms could be attenuated by PAC1 receptor antagonism, and by blockade of PACAP/PAC1 receptor-mediated MEK/ERK signaling or clathrin-mediated endocytic processes (Johnson et al. 2020a, 2020b; May et al. 2021; Miles et al. 2018; Missig et al. 2017, 2014; Roman et al. 2014). Yet, despite much work on PAC1 receptor signaling and function, there are still significant voids in understanding of how the receptor structural variants (and resulting conformational dynamics) can alter signaling patterns to potentially transform homeostatic responses to maladaptive neuroplasticity.

In the current work, molecular modeling and simulations revealed that the 28-amino acid ICL3 Hop insert can enhance PAC1Hop/β-arrestin interactions to impact receptor internalization and endosomal ERK signaling. In accord, pulse signaling studies showed that the duration of PAC1Hop receptor-mediated ERK activation was significantly extended compared to that for the PAC1Null variant which correlated with the experimentally observed temporal onset, extent, and dwell time of internalized PAC1Hop receptors in endosomal structures. Acute versus sustained ERK signaling may be contributory to cellular programs in development, homeostatic regulation, and maladaptive neuroplasticity. Accordingly, these findings provide insights on how PAC1Null/Hop variants can modulate intracellular signaling patterns and suggest that targeting GPCR ICL3 segments for therapeutics may be a means to fine-tune GPCR signaling for homeostatic regulation.

Materials and Methods

Computational Model Preparation, Simulation Setup, and Data Analyses

Modeling of the PAC1Null receptor was described previously (Liao et al. 2019a, 2017). The PACAP-bound active conformation of human PAC1Null receptor was achieved in our study using simulations with enhanced sampling via adaptive tempering (Zhang and Ma 2010). All-atom MD simulations were performed using GROMACS (Abraham et al. 2015) with the CHARMM36m-cmap force field (Best et al. 2012), with the POPC lipid bilayer membrane and the TIP3P water model. Membrane builder in CHARMM-GUI (Jo et al. 2008) was used to embed the PAC1 transmembrane domain (7TM) of each PACAP/PAC1/β-arrestin1 complex in a lipid bilayer consisting of 200 to 240 1-palmitolyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) molecules, 24,000 to 35,000 TIP3P water molecules, counter ions, and 0.12 M NaCl (see Table S1 for number of atoms and system sizes). Alignment in the membrane was conducted using the OPM PPM server (Lomize et al. 2012). Each system was preprocessed with energy minimization before 285 ns equilibration and subsequent production runs in the NPT ensemble (310 K, 1 bar, Nose–Hoover thermostat and semi-isotropic Parrinello-Rahman barostat) with a time-step of 2 fs. All of the bond lengths to hydrogen atoms in each protein/membrane system were constrained. The particle mesh Ewald (PME) technique was used for electrostatic calculations. The van der Waals and short-range electrostatics were cut off at 12.0 Å with switch at 10.0 Å. The PAC1Null/β-arrestin1 and PAC1Hop/β-arrestin1 models were then simulated under constant temperature (310 K) and pressure (1 atm) for three replicates (see Table S1 for simulation lengths).

In addition, conformational analysis, number of contacts, and distance analysis were performed with TCL scripts implemented in VMD (Humphrey et al. 1996) and plotted with matplotlib (Hunter 2007). The length of alpha-helical segments in ICL3 (α_C) was calculated using the Plumed package (Tribello et al. 2014). Polar contacts within 3.6 Å were shown by Pymol (Schrödinger, Inc.) and defined as contacts between a donor (N, O, S with hydrogen atom attached) and an acceptor (O) within 4 Å. Short-range hydrophobic contacts were defined as contacts between atoms with the absolute charge less than 0.3 (not including hydrogens) within 6.5 Å.

Cell Cultures, Treatments, and Imaging

Stable HEK293 human PAC1Null-EGFP and PAC1Hop-EGFP receptor cell lines were prepared using methods previously described (May et al. 2014, 2010). Briefly, the cells were transfected using Lipofectamine 2000 reagent and cultured in Dulbecco’s modified Eagle’s medium (DMEM)/Ham’s F-12 supplemented with 8% fetal bovine serum containing 500 ug/ml G418 (Geneticin) for stable cell selection. Individual colonies were selected and expanded, and functional expression of the receptor was determined by green fluorescent protein (EGFP) expression and second messenger activation. From transcript analyses, PAC1Null-EGFP and PAC1Hop-EGFP expression levels were comparable. The human PAC1Hop ICL3 KRSKE mutant receptor construct (PAC1Hop mutKRSKE-EGFP) was prepared and cloned into pEGFP-N1 expression vector (BioBasic, Markham, Canada); the HEK293 cells were similarly transfected and selected as described above and single colonies were expanded to create stable cell lines. The cultures were seeded into 24-well culture plates and maintained in serum-supplemented DMEM/F-12 medium. In drug treatment experiments, the cultures were pretreated with inhibitors for 15 min before the addition of 25 nM PACAP27 for 5 min at 37 °C. In acute pulse PACAP treatments, the cultures (70% confluence) were exposed to 25 nM PACAP27 for 5 min at 37 °C and rinsed in 2 complete changes of warmed medium before continued incubation for the indicated times. KT5720, BimI (bisindoyl-maleimide I) (both from Calbiochem EMD Biosciences, La Jolla CA), and Pitstop 2 (N-[5-(4-bromobenzylidene)-4-oxo-4,5-dihydro-1,3-thiazol-2-yl]naphthalene-1-sulfonamide) (Abcam Biochemicals, Cambridge, UK) were prepared as stocks in DMSO; final DMSO concentrations in the cultures were < 0.01%.

For β-arrestin immunocytochemistry, the PAC1 receptor HEK cells were fixed with 4% paraformaldehyde, permeabilized and incubated overnight with rabbit monoclonal β-arrestin1/2 antibody (D24H9, 1:1000; Cell Signaling Technologies, Danvers, MA) for target visualization with Cy3-conjugated donkey anti-rabbit IgG (1:250; Jackson Immunoresearch, West Grove, PA). For endosomal trafficking, the coverslip cultures were transfected for the expression of Rab5a-RFP (early endosome marker) or Rab7-RFP (late endosome marker; CellLights reagent, 2 ul reagent per well; Life Technologies, Carlsbad, CA). Twenty-four hours after transfection, the cultures were rinsed, treated as described above, and fixed with 4% paraformaldehyde. Fluorescent images were captured on an Olympus BX51 microscope/QImaging camera using Image Pro 7.0. For ImageJ quantitative analyses, the fluorescence of the total cell area (region of interest A; ROI A) and from the cytoplasm (area within 2 µm of the cell membrane; ROI B) was determined; the percent receptor internalization was determined by the ROI B/ROI A ratio (Shintani et al. 2018).

Primary neonatal rat superior cervical ganglion (SCG) cultures were prepared exactly as described previously and cultured in complete serum-free medium (Braas and May 1999; May and Braas 1995) for PAC1Null-EGFP or PAC1Hop-EGFP vector transfection by biolistic gene transfer(May et al. 2010). The use of neonatal rats was approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Vermont.

Gel Electrophoresis and Western Blotting

Control and treated cultures were extracted as previously described (May et al. 2014, 2010) with 75 µl RIPA buffer (50 mM Tris–HCl, pH 8.0, 120 mM NaCl, 5 mM EDTA, 1% NP-40, and 0.1% SDS) containing 0.3 mg/ml phenylmethyl-sulfonylfluoride, protease inhibitors (16 µg/ml benzamidine, 2 µg/ml leupeptin, 50 µg/ml lima bean trypsin inhibitor, and 2 µg/ml pepstatin A) and phosphatase inhibitor mix (5 mM EDTA, 5 mM EGTA, 1 mM sodium orthovanadate, 10 mM sodium pyrophosphate, and 50 mM sodium fluoride). Total sample proteins were determined using the Bradford reagent (Thermos Fisher). For Western analyses, protein samples (30 µg) were fractionated on 4–12% SDS-PAGE gels, transferred onto Immobilon-P PVDF membranes (Millipore, Billerica, MA), blocked and incubated with pan or phospho-specific ERK1/2 antisera (Cell Signaling Technology, Beverly, MA) for quantitative infrared imaging (LiCor Biosciences, Lincoln, NE). All pERK data were normalized to total ERK levels in the same samples.

Statistics

Statistics were performed using SigmaPlot 12 statistical software (version 5.4; La Jolla, CA). Differences among means were compared by one-way ANOVA followed by Student–Newman–Keuls or Holms-Sidak posthoc analysis. Data are presented as mean SEM; p < 0.05 was considered significant.

Results

PAC1 Receptor ICL3 Hop Insert Enhances β-Arrestin Interactions

Stemming from prior work (Dror et al. 2009, 2011; Goldfeld et al. 2011; Li et al. 2013; Liao et al. 2019a, b), the PAC1 receptor and β-arrestin interactions were modeled with microsecond-long molecular dynamics (MD) simulations (Liao et al. 2021; Mafi et al. 2020; Rodríguez-Espigares et al. 2020; Wu et al. 2020). The PAC1Null and PAC1Hop homology models were generated using templates from the PAC1 receptor ECD crystal structure (PDB ID: 3N94) (Kumar et al. 2011) and the transmembrane domain of the glucagon receptor (PDB ID: 4L6R) (Siu et al. 2013). While activated PACAP and G protein bound structures have recently been obtained using cryogenic electron microscopy (Kobayashi et al. 2020; Liang et al. 2020; Wang et al. 2020), these structures lack the 21-amino acid ECD insert and the Hop insert (Fig. 1). The 21-amino acid ECD insert, also missing in the ECD crystal structure, was inserted in both PAC1 receptor models before all-atom simulations of the receptors in explicit solvent/membrane environments. Furthermore, the complex models were created by aligning our PACAP-bound PAC1Null or PAC1Hop receptor model and the β-arrestin1(L) homology model to the rhodopsin-arrestin crystal structure (PDB ID: 4ZWJ) (Kang et al. 2009). Each complex model was simulated for 2.0 µs in each of 3 replicas which was sufficient to attain stable structures.

Our simulations demonstrated how the distinct ICL3 sequences between the PAC1Null and PAC1Hop receptors can result in marked differences in β-arrestin interactions. Whereas the ICL3 of the PAC1Null receptor presented as a random coil, MD simulations and loop predictions (Prime, Schrödinger, Inc.) suggested that the ICL3 of the PAC1Hop variant, in the absence or presence of β-arrestin, was always in dynamic equilibrium between coiled and partial helical conformational states (Figs. 1 and 2A). Distinct from the random coiled ICL3 in PAC1Null, PAC1Hop was found to form the Hop helix from A362^ICL3 to M369^ICL3 in the ICL3 region, which allowed extensive local interactions with the N-domain loops of β-arrestin1 (i.e., K358^ICL3/E92^{β−arrestin1} and E371^ICL3/R52^{β−arrestin1}; Fig. 2B and C). As a result, the number of interactions was increased by 72% compared to that for the PAC1Null receptor (Fig. 2B). Furthermore, our analysis of conformational populations indicated that β-arrestin binding stabilized the helical conformations of the PAC1Hop ICL3, which in turn enhanced the stability of the complex. In addition to the increased local interactions between PAC1Hop ICL3 and β-arrestin N-domain, we also observed an allosteric effect which augmented the interactions between the receptor C-terminal helix 8 (H8) and the 160-loop of β-arrestin (E155^{β−arrestin1}/R441^H8 and E156^{β−arrestin1}/K440^H8) by 80% compared to the PAC1Null receptor (Fig. 2B and C).

Fig. 2.

ICL3 structural characterization from molecular dynamics simulations. A Histograms of the alpha helical content of the HOP isoform of ICL3 (αc) for free PAC1Hop (blue bars) and PAC1Hop receptor bound to β-arrestin (red bars) obtained from multiple micro-second simulations (Hop segment in blue). B Polar (dark) and non-polar (light) contacts between different parts of PAC1 Null (blue) and HOP (red) receptor with β-arrestin. Error bars are from independent simulations. C Key polar contacts observed between PAC1Hop receptor (green/blue) and β-arrestin (yellow) (see text)

PAC1Hop Receptor Activation Elicits Sustained ERK Signaling

In our MD simulations, the enhancements of PAC1Hop receptor interactions with β-arrestin may anticipate mechanistic and functional differences in ERK activation compared to the PAC1Null/β-arrestin including (i) faster ERK activation onset; (ii) enhanced PAC1Hop receptor internalization and dwell time in endosomes; and (iii) increased ERK activation duration. These parameters were evaluated in our HEK PAC1Null-EGFP and PAC1Hop-EGFP stable cell lines. When the cultures were treated with PACAP for 10 min, there were no apparent differences between the two receptor variants in ERK activation with respect to PACAP potency (EC₅₀ < 10 nM) and efficacy (6- to tenfold at 25 nM PACAP). Consistent with previous observations, inhibition of cAMP/PKA with KT5720 (1 µM) did not attenuate PACAP-stimulated ERK activation for either receptor isoforms (Fig. S2); by contrast, for both cell lines, treatments with endocytosis inhibitor Pitstop2 (PS; 20 µM) or PKC inhibitor BimI (15 µM) blocked ERK activation approximately 50–70% (Fig. S2). Notably, the endosomal ERK signaling pathway appeared preferential across different PACAP concentrations or treatment times (Fig. S3) (May et al. 2021).

To assess whether there were temporal differences in the onset of ERK activation between the receptor variants, the PAC1Null and PAC1Hop cell lines were treated with 25 nM PACAP for 0.5 to 5 min; at the end of each treatment, the replicate cultures were immediately placed on ice and extracted. Following 2.5 min exposure to PACAP, both PAC1Null and PAC1Hop receptors stimulated ERK activation (Fig. 3A), consistent with the temporal parameters of internalized receptor mechanisms. There were no apparent temporal differences in activated ERK efficacy between the receptor variants; the responses were already maximal at the first few minutes of stimulation and sustained for at least 10 min.

Fig. 3.

PAC1Hop receptor variant displays increased ERK activation duration. Stable HEK PAC1Null-EGFP and PAC1Hop-EGFP cell lines were treated with 25 nM PACAP27 as described in Methods for phosphorylated ERK signaling by Western assays. A The cultures were treated acutely for 0.5–5 min before harvest. For both receptor variants, the onset of ERK activation was apparent at 2.5 min and appeared maximal by 5 min. B By contrast, in cultures acutely pulse treated with PACAP for 5 min, followed by washout and harvest after extended times, the duration of ERK activation in the PAC1Hop cultures was extended markedly compared to that in PAC1Null receptor cultures. All pERK data normalized to total ERK levels in the samples and expressed as fold change from untreated control cultures. *, significantly different from untreated control samples; + , significantly different from PACAP-treated samples (15 min); n = 3, data represent mean fold change ± SEM. Representative PAC1Null and PAC1Hop receptor culture pERK/total ERK Western blot data shown at bottom of each panel

Unlike the ERK initiation data, there were marked differences between PAC1Null and PAC1Hop receptors in the duration of ERK activation. To establish how transient PACAP exposure may sustain ERK phosphorylation, the cells were exposed to a 5 min pulse of 25 nM PACAP, washed and incubated at 37 °C for harvest at subsequent time points. Under this paradigm, acute PACAP/PAC1Null receptor stimulation sustained ERK signaling for 15 min before returning to near basal levels (Fig. 3B). Importantly, in comparison, PACAP/PAC1Hop receptor activation increased the duration of maximal ERK signaling at least twofold (to more than 30 min) before signal attenuation to basal levels at 60 min. The current results demonstrated that PAC1Hop does not require continuous PACAP exposures for sustained ERK responses, and that once initiated, the endosomal ERK signaling cascade was able to proceed autonomously to completion. Consistent with previous work, the long temporal pERK responses above implicated PAC1 receptor internalization and endosomal signaling as the principal pathway to ERK activation (May et al. 2014; May and Parsons 2017; Merriam et al. 2013; Tompkins et al. 2016). In conjunction with MD simulations, these observations suggested that the differences in ICL3, and the resulting alterations in docked β-arrestin orientations to the PAC1Null and PAC1Hop variants, had no apparent effects in the onset of ERK signaling; however, the enhancements and resulting alterations in ICL3 Hop interactions with β-arrestin significantly extended the duration of PACAP/PAC1Hop receptor-mediated endosomal ERK activation.

Internalized PAC1Null and PAC1Hop Receptor Associations with β-Arrestin in Endosomes Correlate with ERK Activation

The PAC1Null and PAC1Hop receptor constructs used in the cell culture ERK signaling studies above were tethered to EGFP so that the temporal changes in variant receptor internalization and trafficking could be correlated directly with biochemical data. GPCR internalization and endosomal ERK signaling are well choreographed processes typically initiated by β-arrestin binding for signaling complex scaffolding and subsequent complex trafficking through endosomal compartments. Under unstimulated control conditions, intense PAC1Null-EGFP and PAC1Hop-EGFP receptor expression in the cultures was found on the cell surface, whereas β-arrestin immunoreactivity was weak and distributed diffusely in the cytosol (Fig. 4, left panels). Using the same 25 nM PACAP pulse exposure and washout paradigm as described above, a significant fraction of the surface PAC1Null-EGFP receptor expression (55%) was internalized and colocalized (< 5 min) with bright β-arrestin-immunoreactivity in small intracellular endosomal vesicles (Fig. 4, panels in top half; PAC1Null (green), β-arrestin (red), merged (yellow); Fig. S4). The cellular PAC1Null-EGFP/β-arrestin colocalization pattern was maintained at 15 min, which correlated temporally with PAC1Null receptor-mediated ERK activation, after which the β-arrestin immunoreactivity dissociated and dissipated from the vesicular structures (30–60 min). Notably, during the entire treatment paradigm, a significant fraction of the PAC1Null-EGFP receptor appeared to remain on the cell surface.

Fig. 4.

Activated PAC1Null and PAC1Hop receptor cells show differential receptor colocalization patterns with β-arrestin. PAC1Null-EGFP and PAC1Hop-EGFP cells were acutely pulse treated with PACAP (5 min) and washed as described in Methods and Fig. 4 (above). At various times following washout, the cultures were fixed and processed for β-arrestin immunocytochemistry. For untreated control PACNull (top) and PAC1Hop (bottom) receptor cells, the receptors (green) were localized to the cell membrane and β-arrestin-immunoreactivity (red) appeared diffuse in the cytoplasm. For acutely treated PAC1Null receptor cells, PACAP-induced receptor internalization which colocalized well with bright β-arrestin-immunoreactive puncta at 5 and 15 min (EGFP-tagged receptor and red β-arrestin signals appear yellow in meged image); the colocalization signals dissipated at later time periods. Note that for the PAC1Null receptor cells, a fraction of the receptors remained on the cell surface during the experimental period. By contrast, the activated and internalized PAC1Hop receptors were well colocalized with β-arrestin-immunoreactivity during throughout the 60 min treatment period. The activated PAC1Hop receptors demonstrated near complete internalization compared to the PAC1Null receptors; the PAC1Hop receptors demonstrated surface distribution again after 60 min

The PAC1Hop-EGFP receptor also internalized rapidly into intensely labeled β-arrestin-immunoreactive vesicles (Fig. 4, panels in bottom half), but in contrast to the PAC-1Null receptor, nearly all of the activated PAC1Hop-EGFP receptors were internalized from the cell membrane (> 85%; Figs. 4 and S4) within 15 min – i.e., for many cells, the surface was essentially devoid of PAC1Hop-EGFP expression – and remained completely colocalized with β-arrestin immunore-activity for at least 30 min. Some dissociations of PAC1Hop-EGFP receptors from the β-arrestin vesicles were apparent by 60 min, at which time PAC1Hop-EGFP receptor expression also returned to the plasma membrane (Fig. 4, panels at bottom half). As the internalization of PAC1Hop-EGFP receptor cleared the receptors from the plasma membrane and the resulting PAC1Hop-EGFP receptor vesicles colocalized temporally with the early endosomal marker Rab5 during the entire treatment period (Fig. 5A, middle panels), these observations implicated that the sustained ERK signals were generated from PAC1Hop receptor/β-arrestin endosomal structures rather than plasma membrane mechanisms. The clathrin inhibitor Pitstop 2 has been shown to block PAC1Hop receptor endocytosis and ERK activation ^13,57,58 and in congruence, Pitstop 2 appeared to lock PAC1Hop receptor/β-arrestin at the internal leaf of the plasma membrane (Fig. 5A, bottom panels). Note, the punctate β-arrestin-immunoreactivity structures (red) that align and rim the inner surface of the plasma membrane, in apposition to the PAC1Hop-EGFP receptor signals. Hence, clathrin-mediated PAC1 receptor endocytosis appeared requisite for long term PACAP/ERK signaling. As protein biosynthesis and intracellular vesicular trafficking for secretion/routing to the plasma membrane can require hours, the return of PAC1Hop-EGFP to the cell surface within 60 min likely represented receptor recycling mechanisms rather than de novo synthetic mechanisms.

Fig. 5.

Internalized PAC1Hop receptor is associated with endosomes, blocked by clathrin inhibitors and found in primary neurons. A The internalization of activated PAC1Hop receptors at 30 min is colocalized with β-arrestin-immunoreactivity (red; top row) and with early endosomal marker Rab5a (red, middle row). However, if the cells were pretreated with the clathrin inhibitor Pitstop 2 (PS; 20 µM), the activated PAC1Hop receptors, even at 30 min, remained on the cell surface and appeared to be docked with β-arrestin (red, bottom row) at the membrane. B Primary rat SCG neurons were transfected with PAC1Null-EGFP or PAC1Hop-EGFP vectors. Comparable to HEK cells, the receptors were localized predominantly to the cells surface; following 5 min PACAP treatment, the receptors demonstrated apparent surface clustering. After 30 min, most of the SCG PAC1Hop receptors appeared internalized while some PAC1Null receptors appeared retained on the cell surface

PAC1Hop Receptor ICL3 – β-Arrestin Interactions Are Determinants of Sustained Endosomal Signaling

There are several potential mechanisms underlying the observed long term PAC1Hop receptor activation of ERK. One mechanism may be dependent on PAC1Hop receptor preferential association with particular β-arrestin subtypes and variants. There are β-arrestin1 and β-arrestin2 subtypes, and further, the β-arrestin1 long variant (β-arrestin1L) contains an additional LLGDLASS clathrin binding sequence in the 344-loop of the C-edge (independent of the primary consensus N-domain LφLφD/E clathrin binding motif) which may serve to deeply anchor the β-arrestin molecule in to the membrane to alter signaling dynamics. By contrast, both the p-short β-arrestin1S and β-arrestin2 are similar in that the LLGDLASS sequence is missing. All of the β-arrestin variants are present in the cell lines and the structural differences among the variants suggested that the PAC1Hop receptor may preferentially associate with β-arrestin1(L) to extend PAC1Hop receptor association with endosomal membranes. However, knockdown of β-arrestin1 failed to change the temporal dynamics of PAC1Hop receptor ERK signaling which appeared consistent with previous work describing PAC1 receptor associations with β-arrestin2 for ERK signaling (Shintani et al. 2018). Hence to assess the roles of PAC1Hop ICL3 activities, some of the key ICL3 residues were mutated to disrupt ICL3 helical or random-coil interactions with β-arrestin. Informed from molecular modeling (Fig. 1), the targeted PAC1Hop ICL3/β-arrestin salt-bridge and hydrogen-bond disruptions included K358A^ICL3/E118^Arr, R361A^ICL3/D93^Arr/E92^Arr, S366G^ICL3/N83^Arr, K368A^ICL3/E50^Arr/E92^Arr, and E371A^ICL3/Y54^Arr/R52^Arr. Stable HEK-PAC1Hop mut-KRSKE-EGFP cell lines were subsequently created and as for the wild type receptors, the mutant receptor was highly expressed and distributed on the plasma membrane (Fig. 6A). As the interactions between the GPCR C-terminal tail (H8) and β-arrestin appear sufficient to induce receptor internalization (Nguyen and Lefkowitz 2021), activation of the PAC1Hop mutKRSKE-EGFP receptor was anticipated to result in receptor endocytosis and endosomal signaling. But importantly, in the same PACAP-pulse treatment paradigm described above, disrupting the PAC1Hop ICL3 interactions with β-arrestin significantly diminished the duration of ERK activation (Fig. 6B) to that observed in the Null receptor variant. These observation are unequivocal in demonstrating how ICL3 can influence signaling outcomes.

Fig. 6.

Stimulated PAC1Hop ICL3 mutant receptor exhibits more transient activated ERK duration than wildtype receptor. Stable HEK PAC1Hop mutKRSKE-EGFP cells expressing the receptor with ICL3 amino acid substitutions that disrupt salt-bridge and hydrogen bond interactions with β-arrestin were prepared as described in Methods and text. A PAC1Hop mutKRSKE-EGFP receptors are expressed on the cell surface as in wildtype PAC1Null and PAC1Hop receptor cells. The PAC1Hop mutKRSKE-EGFP receptors circumscribe the cells resulting in cultures with a cobble-stone appearance. B PACAP-pulse treatment of the PAC1Hop mutKRSKE-EGFP cells, as in described above, demonstrated abbreviated ERK activation duration compared to wildtype PAC1Hop receptors (compare with Fig. 4B). All pERK data normalized to total ERK levels in the samples and expressed as fold change from untreated control cultures. *, significantly different from untreated control samples; +, significantly different from PACAP-treated samples; n = 3, data represent mean fold change ±SEM. Representative PAC1Hop mutKRSKE receptor culture pERK/total ERK Western blot data shown at bottom

PAC1Null and PAC1Hop Receptor Internalization Patterns Are Different in Primary Sympathetic Neurons

Nearly all sympathetic postganglionic neurons in the superior cervical ganglion (SCG) endogenously and uniquely express PAC1Hop receptors for physiological sympathetic responses, including long term ERK signaling (Braas and May 1999). Congruent with previous studies, the transfection of primary neonatal SCG neurons with PAC1Hop-EGFP vectors resulted in cell surface receptor expression on the soma and fiber processes, which upon PACAP exposure, clustered (5 min) and internalized completely into long lasting vesicular structures (> 30–60 min) that correlated temporally with long term ERK and Akt activation (Fig. 5B) (May et al. 2010). Neural tissues (and cell lines) do not uniquely express PAC1Null receptors for homologous PAC1Null-EGFP receptor expression, but as SCG neurons can route PAC1 receptors, primary sympathetic neurons were again transfected for PAC1Null-EGFP receptor expression and internalization studies. PAC1Null-EGFP receptor expression was again observed predominantly on the neuronal cell surface (Fig. 5B). PACAP treatment of the sympathetic neurons resulted in rapid PAC1Null-EGFP receptor clustering, but as in the HEK cell lines, the internalization was unlike that for the PAC1Hop-EGFP receptor and a significant fraction of the PAC1Null receptor remained on the cell surface. Hence, the interactions of PAC1 receptor ICL3 loop with β-arrestin are determinants of the extent of PAC1 receptor internalization from the cell surface and the dwell time in endosomes for sustained second messenger generation.

Discussion

More than half of all GPCRs identified have splice variants (Markovic and Challiss 2009) and many of the resulting structural differences in the intracellular loops can potentially modulate the docking of transducers to fine-tune intracellular signaling events. In addition to G proteins, the β-arrestin transducer molecules have roles in receptor internalization/desensitization and can be critical protein scaf-folds for endosomal signaling, especially the generation of activated ERK. ERK signaling can participate in a variety of trophic and regulatory responses in development, regeneration and homeostasis, but notably, can also mediate maladaptive neuroplasticity. For the PAC1 receptor, sustained endosomal ERK signaling has been implicated in chronic pain and psychopathologies; hence understanding how the PAC1 ICL3 receptor variants impact ERK generation can be centrally important with respect to future therapeutics.

Among the PAC1 receptor splice variants, all brain regions including limbic structures preferentially express the PAC1Null and PAC1Hop receptor isoforms in an approximate ratio of 2:1; some studies have shown that the PAC1Hop receptor variant can be increased by stress to modulate behavioral responses (Amir-Zilberstein et al. 2012; Blechman and Levkowitz 2013). Notably, our molecular simulations revealed that the PAC1Null receptor ICL3 (no cassette insert) adopts a random coil while the PAC1Hop ICL3 can transition between coiled and helical structures. Accordingly, the resulting differences in PAC1Null/β-arrestin and PAC1Hop/β-arrestin complex formation anticipated differences in β-arrestin-mediated receptor internalization, trafficking, and ERK activation. Although in vitro stimulation of the PAC1Null receptor resulted in receptor endocytosis, significant levels of PAC1Null receptors (45%) remained on the cell surface. By contrast, much of the activated PAC1Hop receptors were rapidly internalized; further-more, the internalized PAC1Hop receptors were colocalized and retained in β-arrestin-immunoreactive endosomes longer than that for PAC1Null receptors.

In correlation with these studies, we examined how the PAC1 ICL3 receptor variants may differentially impact ERK signaling. For both PAC1Null and PAC1Hop receptors, PACAP treatments increased phosphorylated ERK levels in less than 2 min, in accord with previous temporal studies of GPCR endosomal ERK activation (Kotowski et al. 2011; Sposini and Hanyaloglu 2017). However, even though the onset of ERK activation did not appear different between the variants, the duration of endosomal ERK signaling to a pulse exposure of PACAP was increased substantially in the Hop variant compared to that of the Null receptor. For PAC1Null and PAC1Hop receptors, the duration of ERK activation was directly related to PAC1 receptor/β-arrestin association and resident time in Rab5 endosomes. The PAC1Null receptor-mediated ERK activation was brief and correlated temporally with the vesicular colocalization of internalized receptors with β-arrestin-immunoreactivity; the colocalization of PAC1Null with β-arrestin dissipated after 15 min in parallel with termination of the ERK signal. Similarly, the prolonged PAC1Hop receptor association time with β-arrestin (> 30 min) was reflected in an increase in ERK signaling duration. These results appear coherent with other recent studies suggesting that ICL3 sequences can significantly impact GPCR signaling responses (Verweij et al. 2020). Interestingly, the application of Pitstop2 to inhibit clathrin terminal domain binding to β-arrestin did not block β-arrestin apparent translocation and association with PAC1 receptors at the inner leaf of the plasma membrane, yet ERK activation was markedly attenuated. Although β-arrestin can be a scaffold to more than 50 proteins related to cellular trafficking and signaling, including kinases, phosphatases, phosphodiesterases, and ubiquitin ligases, these observations suggested that the docking of β-arrestins to activated PAC1 receptors alone at the plasma membrane was insufficient for Raf1/MEK assembly and ERK signaling. Clathrin binding to β-arrestins does not appear to occlude Raf1/MEK binding (Bourquard et al. 2015); hence clathrin-mediated receptor endocytosis and trafficking appeared to be prerequisites to alter β-arrestin conformation to expose scaffolding sites or facilitate receptor/β-arrestin vesicular routing to signaling endosomes.

Following membrane delimited G protein-dependent signaling, the well-choreographed sequence of steps for activated GPCR internalization and intracellular trafficking have been associated typically with GPCR desensitization/recycling, but critically, can also represent a mechanistic reprogramming for G protein-independent second messenger production in signaling endosomes. class B GPCRs, including the PAC1 receptor, typically have higher affinity for β-arrestins than class A receptors. Inspection of the PAC1 receptor sequence shows the PAC1 receptor C-terminus has a Ser⁴⁵²Ser⁴⁵³Ser⁴⁵⁴ cluster (PAC1Null receptor numbering) that is comparable to the serine cluster in the vasopressin receptor (Milligan 2004; O’Brien et al. 2015; Oakley et al. 2000; Shenoy and Lefkowitz 2003). This cluster may be the principal site for G protein receptor kinase (GRK) phosphorylation that displaces the β-arrestin C-terminal tail from the N-domain, resulting in β-arrestin polar core disruption and adoption of an active conformational state. How-ever, not all class B receptors possess the Ser residue triplet in the cytoplasmic tail and hence the role of these residues in the PAC1 receptor remains to be directly tested. The trafficking of the vesicular PAC1 receptor/β-arrestin signaling platform through endosomal compartments allows sustained second messenger delivery to distinct intracellular sites to differentially regulate cellular function. The early and late endosomes represent two nodes of intracellular sorting (Russell et al. 2006; Scott et al. 2014). The early endosomal membrane contains lipid and protein subsets including the GTPase Rab5 and can incorporate the internalized contents with newly synthesized Golgi-derived components. Since the early endosomal compartment is mildly acidic (approximately pH 6.2), the GPCRs may be largely uncoupled from their ligands, suggesting that the sustained second messenger delivery may be independent of receptor occupancy. Enroute to the perinuclear trans-Golgi network, the internalized receptor endosomes also have alternative routes including tubular network recycling back to the plasma membrane and trafficking to late Rab7 endosomes/lysosomes for degradation processes. The PAC1Hop receptor studies demonstrated that while some PAC1 receptors were trafficked to Rab7 pathways (unpublished observations), PAC1 receptors were also rerouted back to the plasma membrane, as evident from the return of PAC1Hop receptors to the cell surface within 30–60 min, which was better correlated temporally with recycling processes than the protracted hours needed for de novo protein biosynthesis and trafficking.

The ability for GPCRs to signal from diverse intracellular sites has gained wide interest because of stimulated protein and gene expression patterns that appeared unique from those generated from the plasma membrane (Tsvetanova et al. 2021). Endosomal GPCR signaling appears long term; recently, class B GPCRs, including the thyroid-stimulating hormone receptor (TSHR), parathyroid hormone receptor (PTHR), and vasopressin 2 receptor (V2R), have been shown to be capable of generating sustained G protein-mediated cAMP signaling after receptor internalization (Calebiro et al. 2009, 2010; Feinstein et al. 2013; Ferrandon et al. 2009; White et al. 2019) potentially from the formation endosomal GPCR-G protein-β-arrestin megaplexes (Nguyen and Lefkowitz 2021; Nguyen et al. 2019). Rather than G protein dissociation from the activated receptor, the internalized GPCR in this model remains engaged with active G proteins while the receptor C-terminal tail associates with the β-arrestin molecule to drive concomitantly both G protein-dependent and -independent signaling. Whether the PAC1Null and PAC1Hop receptor variants demonstrate differences in ICL3/G protein interactions to alter the duration and efficacy of cAMP second messenger generation is under investigation. However, as the PAC1Hop mutKRSKE receptor also demonstrates impaired cAMP signaling (data not shown), the PAC1 ICL3 will also likely have G protein interactions. Recently, several studies have implicated GPCR endosomal signaling in the regulation of neuronal excitability, including PAC1 receptor modulation of hippocampal dentate gyrus neurons¹⁵ and cardiac neurons (Kotowski et al. 2011; Merriam et al. 2013; Tompkins et al. 2016). Furthermore, chronic pain and the resulting behavioral abnormalities have been associated with sustained ERK signaling, and PAC1 receptor-mediated endosomal ERK signaling appears central in these maladaptive processes (Missig et al. 2017). These observations are consistent with studies of endosomal GPCR protease-activated receptor-2 (PAR₂) and neurokinin-1 receptor (NK₁R) signaling in irritable bowel syndrome (IBS) pain and transient receptor potential (TRP) channel sensitization (Amadesi et al. 2004; Chen et al. 2011; Jensen et al. 2017; Jimenez-Vargas et al. 2020, 2018). While signaling of the GPCR mu opioid receptor at the plasma membrane mediates analgesia via Gαi, mu receptor endosomal ERK signaling has been suggested to contribute to maladaptive tolerance, respiratory depression and gastrointestinal dysmotility leading to constipation (Manglik et al. 2016). Hence from current work, GPCR signaling from endosomes, Golgi, and nuclear membrane sites represent important and novel therapeutic targets for a variety of disorders.

In sum, MD simulations of PAC1Null and PAC1Hop receptor interactions with β-arrestins anticipated differences in ERK activation between the receptor variants. In coherence with the modeling, PAC1Hop receptor internalization and endosome-mediated ERK activation was extensive and enhanced compared to PAC1Null receptor responses. The ratio of PAC1Null to PAC1Hop receptors in nervous tissue is dynamic (Amir-Zilberstein et al. 2012) and hence, depending on context, would influence changes in ERK signaling levels, which may be contributory to homeostatic and maladaptive physiological responses. As in the case for the PAR₂, NK₁R, and mu opioid receptors, current strategies to mitigate maladaptive endosomal and ERK signaling have been the development of biased small molecule agonists/antagonists that impact signaling path-ways selectively or ligands that specifically target endosomal receptor signaling mechanisms (Gottesman-Katz et al. 2021). Our observations suggest that judicious targeting of GPCR variant ICL3 sites might provide opportunities for novel therapeutics; the fine-tuning of PAC1Hop endosomal/ERK signaling dynamics may represent a novel approach to ameliorate pain and the behavioral abnormalities associated with chronic stress without perturbing regulatory homeostatic activities