Differential regulation of corticotropin releasing factor 1α receptor endocytosis and trafficking by β-arrestins and Rab GTPases

Authors


Address correspondence and reprint requests to Stephen S. G. Ferguson, Cell Biology Research Group, Robarts Research Institute, 100 Perth Drive, PO Box 5015, London, Ontario, Canada, N6A 5K8.
E-mail: ferguson@robarts.ca

Abstract

The corticotropin releasing factor (CRF) type 1α receptor, a member of the G protein-coupled receptor (GPCR) subfamily B, is involved in the aetiology of anxiety and depressive disorders. In the present study, we examined the internalization and trafficking of the CRF1α receptor in both human embryonic kidney (HEK)293 cells and primary cortical neurons. We found that CRF1α receptor activation leads to the selective recruitment of β-arrestin2 in both HEK293 cells and neurons. We observed distinct distribution patterns of CRF1α receptor and β-arrestin2 in HEK293 cells and cortical neurons. In HEK293 cells, β-arrestin2-green fluorescent protein (GFP) co-localized with CRF1α receptor in vesicles at the plasma membrane but was dissociated from the receptor in endosomes. In contrast, in primary cortical neurons, β-arrestin2 and CRF1α receptor were internalized in distinct endocytic vesicles. By bioluminescence resonance energy transfer, we demonstrated that β-arrestin2 association with CRF1α receptor was increased in cells transfected with G protein-coupled receptor kinase (GRK)3 and GRK6 and decreased in cells transfected with GRK2 and GRK5. In both HEK293 cells and cortical neurons, internalized CRF1α receptor transited from Rab5-positive early endosomes to Rab4-positive recycling endosomes and was not targeted to lysosomes. However, CRF1α receptor resensitization was blocked by the overexpression of wild-type, but not dominant-negative, Rab5 and Rab4 GTPases. Taken together, our results suggest that β-arrestin trafficking differs between HEK293 cells and neurons, and that CRF1α receptor resensitization is regulated in an atypical manner by Rab GTPases.

Abbreviations
AT1AR

angiotensin II type 1A receptor

β2AR

β2-adrenergic receptor

BRET

bioluminescence resonance energy transfer

cAMP

cyclic AMP

CRF

corticotropin releasing factor

GFP

green fluorescence protein

GPCR

G protein-coupled receptor

GRK

G protein-coupled receptor kinase

HA

hemagglutinin

HEK

human embryonic kidney

PBS

phosphate-buffered saline

Rluc

Renilla luciferase

YFP

yellow fluorescence protein

Integration of the endocrine, autonomic and behavioural responses to stress is mediated in part by a 41-amino acid neuropeptide, corticotropin releasing factor (CRF) (Vale et al. 1981; Rivier and Vale 1983). CRF was first identified on the basis of its ability to stimulate the release of adrenocorticotropic hormone and exerts this effect via the activation of two G protein-coupled receptors (GPCRs), CRF1 and CRF2 (Dautzenberg and Hauger 2002; Grammatopoulos and Chrousos 2002; Reul and Holsboer 2002). CRF1 receptor is expressed primarily in the brain and pituitary, and several pharmacological and genetic studies in mice have linked CRF1 receptor signaling to both anxiety and depressive disorders (Stenzel-Poore et al. 1994; Smith et al. 1998; Timpl et al. 1998; Okuyama et al. 1999; Smagin and Dunn 2000). This link between CRF function and depressive states has been supported by recent findings that CRF1 receptor mRNA expression was reduced in the brains of suicide victims diagnosed with major depressive disorder (Merali et al. 2004).

Although the pharmacological regulation of CRF1 receptor activity has been the subject of extensive investigation, relatively few studies have examined the mechanisms governing the desensitization, endocytosis, trafficking and resensitization of CRF1 receptors. The canonical model for homologous GPCR desensitization and endocytosis involves receptor phosphorylation by G protein-coupled receptor kinases (GRKs) followed by the binding of β-arrestin proteins. β-Arrestins not only uncouple the receptor from heterotrimeric G proteins, but also function as endocytic adaptor molecules (reviewed by Krupnick and Benovic 1998; Ferguson 2001). β-Arrestins specifically target receptors for internalization by clathrin-coated vesicles. Internalized GPCRs are either retained in the endosomal compartment of the cell, targeted for degradation in lysosomes, or are dephosphorylated and recycled back to the cell surface.

Recent studies have determined that many GPCRs (e.g. the β2-adrenergic receptor, β2AR), preferentially associate with β-arrestin2, whereas other GPCRs (e.g. the angiotensin II type 1A receptor, AT1AR), bind equally well to both β-arrestin1 and 2 and internalize as a complex with β-arrestins (Zhang et al. 1999; Oakley et al. 1999, 2000; Anborgh et al. 2000). GPCRs that do not internalize with β-arrestin are dephosphorylated in endosomes, and are rapidly recycled and resensitized, whereas GPCRs that internalize with β-arrestin bound are either recycled slowly, are retained in the endosomal compartment or are targeted to lysosomes for degradation (Oakley et al. 1999; Trejo and Coughlin 1999; Anborgh et al. 2000; Seachrist et al. 2002). However, the relative contribution of β-arrestin1 and 2 to the regulation of CRF1α receptor endocytosis in both heterologous expression systems and primary neuronal cultures has not been investigated.

The movement of internalized GPCRs between intracellular vesicular compartments is regulated by Rab GTPases (reviewed by Novick and Zerial 1997; Ferguson 2001; Rosenfeld et al. 2002; Seachrist and Ferguson 2003). However, the role of Rab GTPases in regulating the trafficking of CRF1α receptors between intracellular compartments and on CRF1α receptor desensitization and resensitization has not been determined. We have investigated the hypothesis that CRF1α receptor internalization is differentially regulated by β-arrestin1 and 2, and that Rab GTPases regulate CRF1α receptor resensitization.

Materials and methods

Materials

Human embryonic kidney (HEK)293 cells were obtained from the American Type Culture Collection (Manassas, VA, USA). Tissue culture reagents were obtained from Invitrogen (Burlington, ON, Canada). Effectene, mouse monoclonal anti-hemagglutin (HA) (12CA5) and FITC conjugated anti-mouse IgG were obtained from GE Healthcare (Oakville, ON, Canada) and Sigma (St Louis, MO, USA). AlexaFluor 546 goat anti-mouse antibody and Lipofectamine 2000 were purchased from Invitrogen/Molecular Probes (Burlington, ON, USA). Radioisotopes [32P]ATP and [3H]cyclic AMP, and coelenterazine were ordered through VWR/ICN (Mississauga, ON, Canada). CRF peptide was purchased from Phoenix Pharmaceuticals Inc. (Belmont, CA, USA).

Plasmid constructs

The cDNA for the human CRF1α receptor was first amplified by PCR from the human universal Quick CloneTM library (BD Biosciences/Clontech, Mississauga, ON, Canada). The PCR product generated was digested with BglII/HindIII and subcloned into enhanced green/yellow fluorescent protein (EGFP/EFYP-N1) (BD Biosciences/Clontech) and pRluc-N3 (PerkinElmer, Montreal, QC, Canada), constructs kindly provided by Dr Michel Bouvier. The HA epitope was subsequently inserted immediately after the predicted signal peptide cleavage site by PCR. The HA-tagged CRF1α receptor was cloned into pcDNA3.1 (Invitrogen) as a BamHI/EcoRI fragment. All other plasmids have been described previously (Oakley et al. 2000; Dale et al. 2004).

Cell culture and transfection

HEK293 cells were maintained in Eagle's minimal essential medium containing fetal bovine serum (10% v/v) and gentamicin (100 µg/mL). Cells were transfected using a modified calcium phosphate method as described previously (Ferguson et al. 1995). Primary cortical cultures were prepared from E18 embryos as described previously (Fortin et al. 2001) and were maintained in Neurobasal medium containing B27 and N2 supplements as well as penicillin/streptromycin and l-glutamine. Cells were plated on poly-l-lysine-coated 50-mm glass coverslips in Neurobasal medium for 5 h at 37°C and 5% CO2 in a humidified incubator to permit cell attachment. Neurobasal medium was subsequently replaced with culture medium that was replenished every 3 days. Culture medium consisted of Neurobasal medium supplemented with B27, 0.5 U/mL penicillin, 0.05 mg/mL streptomycin, 10 µm MK-801, 25 mm KCl and 5 pg/mL glial-derived neurotrophic factor. Primary neuronal cultures were astrocyte free and transfected using Lipofectamine 2000 following the manufacturer's instructions. The University of Western Ontario Animal Care Committee approved all animal procedures.

Bioluminescence resonance energy transfer (BRET)

Transfected HEK293 cells were washed twice with phosphate-buffered saline (PBS), resuspended in 5 mL PBS and pelleted by centrifugation for 5 min at 180 g. Cell pellets were resuspended in 1 mL BRET buffer (PBS containing 0.1% glucose and 0.1 µm ascorbic acid). Protein concentrations were measured by spectrophotometry using the Dc protein assay (Bio-Rad, Hercules, CA, USA) and cell suspensions were diluted with BRET buffer to a final protein concentration of 1 µg/µL. BRET was then measured using a PerkinElmer (Boston, MA, USA) Victor multilabel counter. Briefly, 50 µL of cell suspension was aliquoted per well (96-well format) and heated to 37°C. Coelenterazine was added to each well (5 µm final concentration), followed by the addition of 500 nm CRF. Light emission was measured from the luciferase donor and yellow fluorescence protein (YFP) acceptor at 460 and 535 nm wavelengths respectively. The BRET ratio was defined as [(emission at 535 nm) − (emission at 460 nm) × Cf]/(emission at 460 nm) where Cf corresponded to (emission at 535 nm/emission at 460 nm) for the control, Renilla luciferase (Rluc), expressed alone.

Membrane adenylate cyclase assay

Transfected HEK293 cells were lysed and detached from culture dishes using cold lysis buffer (10 mm Tris-HCl, 5 mm EDTA, pH 7.4) and cell scrapers. Membranes were prepared by disrupting cell lysates with a Polytron homogenizer twice each for 20 s at 20 000 rpm followed by centrifugation for 30 min at 40 000 g. Membrane pellets were resuspended in lysis buffer, disrupted by polytron, recentrifuged and then resuspended in cold assay buffer (75 mm Tris-HCl, 2 mm EDTA, 15 mm MgCl2, pH 7.4). Following protein assay, membranes were adjusted to 1–2 µg/µL with assay buffer. Membrane preparations, 20 µL in 50 µL final volume, were then assayed for agonist-stimulated adenylate cyclase activity as described previously (Walseth and Johnson 1979; Hausdorff et al. 1990). For resensitization experiments, before membrane preparation, cells were incubated for 15 min at 37°C in the absence (naive) or presence (desensitized or resensitized) of 500 nm CRF. The cells were then washed to remove agonist and maintained on ice (naive and desensitized) or incubated for 15 or 30 min at 37°C (resensitized).

Immunofluorescence labeling

HEK293 cells grown on 35-mm confocal dishes or primary cortical cultures grown on round 15-mm glass coverslips were fixed for 30 min in 4% formaldehyde in PBS at 48 h after transfection. Cells were washed twice with PBS and then incubated for 20 min at 22°C in PBS-TB (PBS containing 0.05% Triton X-100 and 3% bovine serum albumin) to permeabilize cells and block non-specific sites. Cells were then incubated in PBS-TB containing anti-HA antibody (1 : 1000) for 45 min at room temperature, washed twice with PBS, and incubated for an additional 45 min in PBS-TB containing AlexaFluor 546 anti-mouse IgG (1 : 500). Cells were then washed twice with PBS and stored at 4°C until imaged.

Confocal microscopy

Confocal microscopy was performed on a Zeiss (North York, ON, Canada) LSM-510 META laser scanning confocal microscope using a Zeiss 63X, 1.3 numerical aperture, oil immersion lens. Live cell imaging was performed on transfected HEK293 cells plated on 35-mm glass-bottom dishes and maintained at 37°C in Hank's balanced salt solution on a heated stage. Neuronal cultures maintained on coverslips were fixed, processed for immunofluorescence, and mounted on glass slides before imaging. Co-localization studies were performed using dual excitation (488, 543 nm) and emission (band pass 505–530 nm and long pass 585 nm for GFP and AlexaFluor 546 respectively) filter sets. The specificity of labeling and absence of signal crossover were established by examination of single-labeled samples.

Quantification of β-arrestin translocation

HEK293 cells transiently transfected with HA-CRF1α and β-arrestin1 or 2 were stimulated with 500 nm CRF for 0 or 15 min and then fixed for 30 min using 4% formaldehyde. Cells were washed twice with PBS and then, using the LSM 510 microscope, all β-arrestin-GFP positive cells were counted in five random fields of view and the percentage of cells displaying β-arrestin translocation to the cell membrane was calculated as the total number of GFP-positive cells.

Data analysis

Data were analyzed using GraphPad (San Diego, CA, USA) Prism software and statistical significance was determined by analysis of variance. Comparisons between groups were made using Tukey's multiple comparison test.

Results

Internalization of HA-tagged and CRF1α-GFP receptors

To ensure that both HA-CRF1α and CRF1α-GFP receptor constructs were fully functionally competent, we compared their ability to stimulate cAMP formation in HEK293 cells with non-epitope-tagged wild-type CRF1α receptor. CRF stimulation of HEK293 cells induced identical dose-dependent increases in cAMP formation in cells expressing wild-type, HA-CRF1α or CRF1α-GFP receptors (Fig. 1a). A response to CRF treatment was also observed in non-transfected cells suggesting the expression of a low-affinity CRF receptor in HEK293 cells (Fig. 1a). Our initial studies examined the internalization of transfected HA-CRF1α and CRF1α-GFP receptors in HEK293 cells and primary cortical neurons (Figs 1b–e). In the absence of agonist stimulation both HA-CRF1α and CRF1α-GFP receptors were predominantly expressed at the plasma membrane in HEK293 cells (Figs 1b and c, 0 min). In response to the treatment of cells with 500 nm CRF, both the HA-CRF1α and CRF1α-GFP receptors were internalized from the cell surface and accumulated in a perinuclear vesicular compartment (Figs 1b and c, 10 and 30 min).

Figure 1.

Characterization of epitope-tagged CRF1α receptor expression and signaling in HEK293 cells and mouse primary cortical neurons. (a) Adenylyl cyclase activity of HEK293 cells expressing tagged or untagged CRF1α receptors in response to increasing concentrations of CRF. Values are mean ± SEM of four independent experiments performed in duplicate and are expressed as a percentage of 10 µm forskolin control. NT, not transfected. HEK293 cells (b, c) or cortical neurons (d, e) transiently transfected with CRF1α-GFP (b, d) or HA-CRF1α (c, e) were imaged by confocal microscopy at 0, 10 and 30 min following stimulation with 500 nm CRF. Scale bar 10 µm.

When both HA-CRF1α and CRF1α-GFP receptors were transfected into primary cortical neurons, the receptor constructs were expressed at the cell surface of the soma, as well as in the dendritic arbour of the neuron (Figs 1d and e, 0 min). Some punctate staining of both HA-CRF1α and CRF1α-GFP receptors was also observed in dendrites in the absence of agonist stimulation. However, in contrast to findings in HEK293 cells, considerable CRF1α-GFP receptor was found to reside intraneuronally in the absence of agonist treatment (Fig. 1d, 0 min). Agonist treatment with 500 nm CRF promoted the internalization and perinuclear accumulation of both HA-CRF1α and CRF1α-GFP receptors (Figs 1d and e, 10 and 30 min) and CRF1α receptor expressed in dendrites developed a notably more punctate distribution. These data demonstrated that CRF1α receptor was internalized to the perinuclear compartment of both HEK293 cells and neurons. However, subtle differences were observed in the subcellular localization of CRF1α-GFP receptor in the absence of agonist in HEK293 cells and neurons.

Specificity of CRF1α receptor-induced β-arrestin-GFP translocation

Internalized CRF1α receptor was reported to co-localize with β-arrestin2-GFP in endosomes in heterologous cell expression systems (Perry et al. 2005), and β-arrestin1 was reported to translocate to the membrane in response to CRF1α receptor activation (Rasmussen et al. 2004). However, a comparison of β-arrestin1 and β-arrestin2 translocation in response to CRF1α receptor activation in heterologous cells was not made and the regulation of CRF1α receptor internalization has not been studied in primary cortical neurons. Therefore, we compared β-arrestin-GFP translocation in both HEK293 cells and cortical neurons.

In HEK293 cells transfected with β-arrestin1-GFP and HA-CRF1α receptor, the β-arrestin1-GFP signal was expressed diffusely throughout the cytoplasm and nucleus, whereas HA-CRF1α receptor staining was limited to the plasma membrane in the absence of CRF treatment (data not shown). Treatment of cells with 500 nm CRF for 30 min induced HA-CRF1α receptor internalization without β-arrestin1-GFP translocation in the majority of cells (71%) (Fig. 2a).

Figure 2.

Translocation of β-arrestin1-GFP in HEK293 cells and mouse primary cortical neurons. (a) Lack of β-arrestin1-GFP translocation in HEK293 cells transiently transfected with HA-CRF1α following a 30-min exposure to 500 nm CRF. (b, c) β-Arrestin1-GFP translocation response in primary cortical neurons transfected with HA-CRF1α receptor before (b) and after (c) a 30-min treatment with 500 nm CRF. At the time points shown cells were fixed and immunolabeled using anti-HA antibody. Images were captured by confocal microscopy and are representative of three independent experiments. Scale bar 10 µm.

To further characterize the translocation of β-arrestin1 following CRF1α receptor activation, we examined β-arrestin1-GFP translocation responses in mouse primary cortical neurons co-transfected with HA-CRF1α. In the absence of agonist treatment, β-arrestin1-GFP was diffusely distributed throughout the soma and dendrites of primary cortical neurons (Fig. 2b) and treatment of cells with 500 nm CRF induced HA-CRF1α receptor internalization without any change in β-arrestin1-GFP distribution (Fig. 2c).

In contrast to findings for β-arrestin1-GFP translocation in response to CRF1α activation in HEK293 cells, we found that CRF treatment elicited a membrane translocation response in 87% of cells expressing β-arrestin2-GFP and HA-CRF1α, (Fig. 3a). Specifically, β-arrestin2-GFP appeared to co-localize with HA-CRF1α receptor in vesicles at or close to the cell surface but was not co-localized with HA-CRF1α in the perinuclear compartment of the cell. These results differed from those in a study by Perry et al. (2005), in which β-arrestin2-GFP was observed to co-localize with CRF1α in internalized vesicles.

Figure 3.

Translocation of β-arrestin2-GFP in HEK293 cells and mouse primary cortical neurons. (a) β-Arrestin2-GFP translocation in HEK293 cells transiently transfected with HA-CRF1α following a 30-min exposure to 500 nm CRF. (b–e) β-Arrestin2-GFP translocation response in primary cortical neurons transfected with HA-CRF1α receptor following stimulation for 0 min (b), 10 min (c) and 30 min (d, e). Cells were fixed and immunolabeled using anti-HA antibody. Translocation of βarrestin2-GFP is also shown in the dendrites of transfected neurons (e). Images were captured by confocal microscopy and are representative of three independent experiments. Scale bar 10 µm.

In mouse primary cortical neurons expressing both HA-CRF1α receptor and β-arrestin2-GFP, β-arrestin2-GFP was diffusely localized throughout the soma and dendrites (Fig. 3b). Treatment of the cells with 500 nm CRF for 10 min stimulated the movement of β-arrestin2-GFP to the plasma membrane where it was first diffusely localized with the HA-CRF1α receptor (Fig. 3c). Prolonging the agonist treatment to 30 min produced a redistribution of both HA-CRF1α receptor and β-arrestin2-GFP into intracellular vesicles (Fig. 3d). However, co-localization of HA-CRF1α receptor and β-arrestin2-GFP was small in proportion to the total amount of β-arrestin2-GFP, which was extensively localized to vesicles that were not positive for receptor (Fig. 3d). In dendrites, both β-arrestin2-GFP and HA-CRF1α receptor fluorescence was punctate, but exhibited little co-localization (Fig. 3e). These observations illustrated two points. First, in both neurons and HEK293 cells the CRF1α receptor more readily induced the translocation of β-arrestin2 over β-arrestin1. Second, in neurons the intracellular trafficking of CRF1α and β-arrestin2 diverged at later time points of endocytosis.

To determine whether β-arrestin2-GFP translocation in response to HA-CRF1α receptor activation was not solely the consequence of receptor overexpression, we treated cortical neurons expressing only β-arrestin2-GFP with 500 nm CRF to activate endogenously expressed receptor. In response to CRF stimulation, we observed rapid translocation of β-arrestin2-GFP to the plasma membrane where the signal became increasingly punctate and vesicular with time in both the soma and dendrites (Fig. 4). This indicated that β-arrestin2-GFP translocation in response to CRF treatment was not solely related to receptor overexpression.

Figure 4.

Stimulation of endogenous CRF receptor induces translocation of β-arrestin2-GFP in transiently transfected primary neuronal culture. Live neurons transfected with β-arrestin2-GFP were stimulated with 500 nm CRF and imaged by confocal microscopy over a 600-s time series. Scale bar10 µm.

BRET analysis of CRF1α receptor–β-arrestin interactions

To further assess the specificity of β-arrestin1 and β-arrestin2 interactions with the CRF1α receptor, we used BRET to detect β-arrestin binding to the receptor. In cells expressing YFP-tagged CRF1α receptor and either RLuc-tagged β-arrestin1 or β-arrestin2, no significant increase in BRET ratio was observed over time in the absence of agonist stimulation (Fig. 5a), indicating that there was little association between CRF1α receptor and β-arrestins in the absence of agonist. However, in response to treatment with 500 nm CRF there was a significant increase in the BRET ratio for β-arrestin1-RLuc and β-arrestin2-RLuc with CRF1α receptor-YFP (Fig. 5a). The maximal BRET ratio at equilibrium for β-arrestin2–RLuc interactions with CRF1α receptor-YFP was 2.6 times greater than that obtained for β-arrestin1-RLuc. These data were consistent with the confocal data and suggested that β-arrestin2 preferentially interacted with CRF1α receptor.

Figure 5.

Effect of GRK expression on β-arrestin interactions with CRF1α receptor in HEK293 cells as measured by BRET. (a) Increase in association between CRF1α receptor and β-arrestin1 and 2 following agonist stimulation. (b) Effect of GRK1–6 expression on CRF1α receptor–β-arrestin1 association. (c) Effect of GRK1–6 expression on CRF1α receptor–β-arrestin2 association. BRET assays were carried out on live, transiently transfected HEK293 cells following transfer to 96-well plates. Data for BRET ratio calculation were collected by plate reader before and after stimulation with 500 nm CRF. Values are mean ± SEM of four independent experiments performed in duplicate.

The association of β-arrestins with most GPCRs is thought to be GRK phosphorylation dependent (Krupnick and Benovic 1998; Ferguson 2001). Therefore, we examined whether the co-expression of GRK1–6 might facilitate β-arrestin interactions with the CRF1α receptor. We found that the co-expression of GRK1, GRK3, GRK4 and GRK6 increased BRET in response to agonist treatment, whereas GRK2 and GRK5 overexpression had no effect upon CRF1α receptor-YFP–β-arrestin1-RLuc interactions (Fig. 5b). In cells expressing CRF1α receptor-YFP and β-arrestin2-RLuc, GRK2 and GRK5 overexpression attenuated BRET obtained in response to agonist stimulation (Fig. 5c). The BRET ratio obtained for CRF1α receptor-YFP–β-arrestin2-RLuc interactions was increased by the expression of GRK3 and GRK6 and was unaltered by the expression of either GRK1 or GRK4 (Fig. 5c). Overall, GRK6 appeared to be the most effective kinase for increasing the association of both β-arrestin1-RLuc and β-arrestin2-RLuc with CRF1α receptor-YFP (Figs 5b and c).

Intracellular trafficking of CRF1α receptor to Rab5-positive endosomes

Rab GTPases play a key role in regulating the trafficking of proteins between distinct intracellular membrane compartments that contribute to the dephosphorylation and recycling of GPCRs (Pfeffer 2003; Seachrist and Ferguson 2003). Therefore, we examined whether internalized CRF1α receptors were internalized to Rab5-positive endosomes. In the absence of agonist stimulation, HA-CRF1α receptor immunofluorescence was found at the cell surface and GFP-Rab5a was localized to intracellular vesicles distributed throughout the cytoplasm of HEK293 cells (Fig. 6a). However, following agonist treatment of cells with 500 nm CRF, we observed extensive co-localization of HA-CRF1α receptor and GFP-Rab5a in the endosomal compartment (Fig. 6b).

Figure 6.

CRF1α receptor co-localizes with Rab5a-positive vesicles following CRF stimulation of transiently transfected HEK293 cells (a, b) or mouse primary cortical neurons (c, d). Following stimulation with 500 nm CRF for 0 min (a, c) and 30 min (b, d), cells were fixed and immunolabeled using anti-HA antibody. Arrows indicate co-localization of HA-CRF1α receptor and Rab5a-GFP before stimulation (c). Images were captured by confocal microscopy and are representative of three independent experiments. Scale bar 10 µm.

In cortical neurons, GFP-Rab5a fluorescence was localized to the perinuclear compartment of the soma as well as in puncta within the dendrites (Fig. 6c). In the absence of agonist there appeared to be a small degree of co-localization between HA-CRF1α receptor and GFP-Rab5a in the dendrites (Fig. 6c, arrows). However, following agonist stimulation of the neuronal cultures, we observed extensive co-localization between HA-CRF1α receptor and GFP-Rab5a in both the soma and dendrites (Fig. 6d). Thus, the targeting of CRF1α receptors to Rab5-positive early endosomes in both HEK293 cells and cortical neurons appeared to be very similar.

Intracellular trafficking of CRF1α receptor to Rab4-positive recycling endosomes

The recycling of some GPCRs from Rab5-positive early endosomes is mediated by either fast Rab4-regulated recycling endosomes or a slower Rab11-regulated endosomal pathway (Pfeffer 2003; Seachrist and Ferguson 2003). Therefore, we examined whether internalized HA-CRF1α receptor was mobilized to either Rab4- or Rab11-positive recycling endosomes. In HEK293 cells, we found little co-localization between HA-CRF1α receptor and GFP-Rab4 in the absence of agonist stimulation (Fig. 7a). However, following a 30-min exposure to 500 nm CRF, we observed extensive co-localization between immunostained HA-CRF1α receptor and GFP-Rab4 (Fig. 7b). In HEK293 cells, localization of HA-CRF1α receptor to GFP-Rab11-positive vesicles was not observed (data not shown).

Figure 7.

CRF1α receptor co-localizes with Rab4-positive vesicles following CRF stimulation of transiently transfected HEK293 cells (a, b) or mouse primary cortical neurons (c, d). Following 500 nm CRF stimulation for 0 min (a, c) and 30 min (b, d), cells were fixed and immunolabeled using anti-HA antibody. Images were captured by confocal microscopy and are representative of three independent experiments. Scale bar 10 µm.

In primary cortical neurons, Rab4-GFP was found in the perinuclear compartment of the soma as well as in vesicular structures in the dendrites, but exhibited little co-localization with HA-CRF1α receptor (Fig. 7c). Following agonist stimulation for 30 min, we observed extensive co-localization of HA-CRF1α receptor with GFP-Rab4 both in the soma and in dendrites (Fig. 7d). Interestingly, in agonist-stimulated cells, the Rab4-GFP-labelled vesicles were observed as large hollow vesicles that were reminiscent of homotypically fused endosomes. In cortical neurons, co-localization of HA-CRF1α receptor and GFP-Rab11 was not observed (data not shown). Taken together, these data suggest that CRF1α receptors may recycle via the Rab4-regulated rapid recycling endosomal pathway.

CRF1α receptor does not target to lysosomes

In addition to being targeted to recycling endosomes for transport back to the cell surface, it is possible that CRF1α receptors may be targeted for degradation in lysosomes. Therefore, we examined whether HA-CRF1α receptor co-localizes with the lysosomal marker dye LysoTracker Red in both HEK293 cells and cortical neurons. In the absence of agonist treatment there was no co-localization between HA-CRF1α receptor and LysoTracker Red-stained lysosomes. Agonist treatment of cells did not result in the redistribution of HA-CRF1α receptor to lysosomes (Figs 8a and b).

Figure 8.

CRF1α receptor does not co-localize with lysosomes following CRF stimulation of transiently transfected HEK293 cells (a, b) or mouse primary cortical neurons (c, d). Cells were stained with the lysosome-specific dye Lysotracker followed by 500 nm CRF stimulation for 0 min (a, c) and 30 min (b, d) min. Images were captured by confocal microscopy and are representative of three independent experiments. Scale bar 10 µm.

Similarly, there was no co-localization between HA-CRF1α receptor and LysoTracker Red in transfected cortical neurons in the absence of agonist stimulation (Fig. 8c). Identical to observations in HEK293 cells, agonist-stimulated HA-CRF1α receptor internalization in cortical neurons did not result in receptor targeting to lysosomes (Fig. 8d).

CRF1α receptor desensitization and resensitization

Previous studies indicated that trafficking and recycling of GPCRs via Rab5 and Rab4 endosomes preceded receptor resensitization (Seachrist et al. 2000; Odley et al. 2004). Therefore, we examined CRF1α receptor desensitization and resensitization by measuring receptor-stimulated cAMP accumulation. When expressed alone, CRF1α receptor-stimulated cAMP formation was desensitized in response to 500 nm CRF treatment for 15 min, as measured as a mean ± SEM 49 ± 5% decrease in maximal stimulation of cAMP formation (Fig. 9a). When cells were allowed to recover for 15 and 30 min in agonist-free medium, resensitization of CRF1α receptor-responsive adenylyl cyclase activity was observed (Fig. 9a). However, we obtained unexpected results following the overexpression of wild-type Rab5 and a dominant-negative Rab5-S34N mutant. CRF1α receptor-stimulated cAMP responses desensitized normally in the presence of either Rab5 or Rab5-S34N (Fig. 9b). However, only partial resensitization of CRF1α receptor signaling was observed following either wild-type Rab5 or Rab5-S34N overexpression (Fig. 9b). In contrast, full recovery of β2AR signaling was observed following wild-type Rab5 overexpression in matched samples (data not shown). In addition, we found that the overexpression of wild-type Rab4 limited CRF1α receptor resensitization, whereas overexpression of the dominant-negative Rab4-N121I mutant, which blocked the resensitization of the β2AR (Seachrist et al. 2000), did not prevent CRF1α resensitization (Fig. 9c). Taken together, these data suggested that the CRF1α receptor was desensitized in response to agonist exposure and resensitized following the removal of agonist, but exhibited an atypical resensitization profile in the presence of wild-type and dominant-negative Rab5 and Rab4 GTPases.

Figure 9.

Rab5 and Rab4 reduce CRF1α receptor resensitization in transiently transfected HEK293 cells. (a) Time course of HA-CRF1α receptor resensitization following a 15-min 500 nm CRF desensitizing stimulus. (b) Effect of wild-type Rab5 or dominant-negative Rab5 (S34N) on HA-CRF1α receptor resensitization. (c) Effect of wild-type Rab4a or dominant-negative Rab4a (N121I) on HA-CRF1α receptor resensitization. Adenylyl cyclase activity was measured under basal conditions and in the presence of 500 nm CRF. Values are mean ± SEM percentage of control from four independent experiments performed in triplicate.

Discussion

In the present study we have characterized molecular components that contribute to the regulation of CRF1α internalization and trafficking in both HEK293 cells and mouse primary cortical cultures. There were several major findings. First, CRF1α receptor preferentially interacts with β-arrestin2 in both HEK293 cells and primary cortical neurons. Second, in HEK293 cells, β-arrestin2 internalizes as a complex with CRF1α receptor, but dissociates from the receptor in the perinuclear compartment. Third, in primary cortical neurons, β-arrestin2 is internalized to intracellular vesicles that do not appear to be positive for the CRF1α receptor. Fourth, the overexpression of either GRK3 or GRK6 promotes the interaction of both β-arrestin1 and β-arrestin2 with the CRF1α receptor, whereas the overexpression of either GRK2 or GRK5 appears to attenuate β-arrestin2–CRF1α receptor interactions. Fifth, the CRF1α receptor is internalized to Rab5-positive early endosomes and is subsequently mobilized to Rab4-positive recycling endosomes in both HEK293 cells and cortical neurons. Finally, the overexpression of wild-type Rab5 and Rab4 GTPases attenuates CRF1α receptor resensitization.

Pervious studies have independently examined the association of β-arrestin1 and β-arrestin2 with the CRF1α receptor (Rasmussen et al. 2004; Perry et al. 2005), but did not make a side-by-side comparison of the two arrestins in either heterologous cell cultures or primary neuronal cultures. We showed here that CRF1α receptors preferentially interact with β-arrestin2 in both HEK293 cells and cortical neurons, but that in a small subset of HEK293 cells (∼30%) β-arrestin1 translocates to the plasma membrane in response to CRF1α receptor activation. These results were validated by BRET, which demonstrated that β-arrestin2 preferentially associates with CRF1α receptor. The weak membrane translocation response for β-arrestin1 in HEK293 cells is consistent with a previous report that β-arrestin1 translocates to the plasma membrane in response to CRF1α receptor activation but is not required for receptor internalization (Rasmussen et al. 2004). Interestingly, in contrast to observations by Perry et al. (2005), we did not observe redistribution of β-arrestin2-GFP to perinuclear endosomes as a complex with the CRF1α receptor. Furthermore, the activation of endogenously expressed CRF1α receptors in primary cortical neurons resulted in β-arrestin2-GFP translocation, demonstrating that β-arrestin translocation is not simply a phenomenon that is associated with receptor overexpression. Taken together, we conclude from these data that the CRF1α receptor is best classified as a class A receptor that does not form a stable complex with β-arrestins and does not bind to β-arrestin with high efficacy.

Of particular interest is the observation that in cortical neurons, but not HEK293 cells, agonist activation of the CRF1α receptor results in the plasma membrane translocation of β-arrestin2 followed by internalization of the CRF1α receptor and β-arrestin2 to distinct endocytic vesicles. Similar results were previously reported for the 5-HT2A receptor in HEK293 cells, in which both β-arrestin1 and β-arrestin2 are differentially sorted to distinct endocytic vesicles (Bhatnagar et al. 2001). The physiological relevance and the mechanism underlying the differential trafficking of arrestins with either CRF1α or 5-HT2A receptors remains unknown. However, our data suggest that this differential trafficking of arrestins and receptor may be a generalized phenomenon. For example, we have previously shown by subcellular fractionation that β-arrestins redistribute to azurophilic granules in response to interleukin-8 stimulation, presumably to facilitate tyrosine kinase-dependent granule release (Barlic et al. 2000). Moreover, recent studies suggest that, in response to stressful stimuli, CRF alters the serotonergic regulation of GABA transmission (Tan et al. 2004). Given recent reports of GPCR heterodimerization (Bulenger et al. 2005), and similarities in the intracellular trafficking of CRF1α and 5-HT2A receptors, it is plausible that the heterodimerization of these receptors functions to modulate the intracellular trafficking of CRF1α receptor and β-arrestin2 in neurons.

The interaction of arrestins with GPCRs is considered to be facilitated by GRK-mediated phosphorylation of serine and threonine residues in the intracellular domains of GPCRs (Krupnick and Benovic 1998; Ferguson 2001). We showed here that there is specificity with regards to the GRK isoforms that facilitate both β-arrestin1 and β-arrestin2 association with CRF1α receptors. Specifically, we found that GRK1, GRK3, GRK4 and GRK6 facilitate β-arrestin1 interactions with the CRF1α receptor, whereas only GRK3 and GRK6 facilitate β-arrestin2 binding to the CRF1α receptor. This is consistent with a previous report showing that GRK3 and GRK6 are the endogenous GRKs in HEK293 cells that mediate CRF1α desensitization (Teli et al. 2005). Therefore, it seems likely that certain GRKs may differentially promote recruitment and association of arrestin molecules to regulate the degree of receptor desensitization. The time course of CRF1α desensitization does seem to differ considerably depending on cell type (Dieterich et al. 1996; Hauger et al. 1997; Roseboom et al. 2001; Teli et al. 2005), which may reflect differences in the cellular complement of GRK and/or arrestin molecules (Menard et al. 1997). Unexpectedly, the overexpression of GRK5 and particularly GRK2 was found to antagonize β-arrestin2 interactions with the CRF1α receptor. Thus, it is possible that the context of phosphorylation, rather than phosphorylation itself, might be essential for β-arrestin binding and CRF1α receptor desensitization. Alternatively, the antagonism of β-arrestin binding to CRF1α receptor following the overexpression of GRK2 and GRK5 suggests that these kinases may compete with β-arrestins for receptor binding. Previous reports have described that GRKs mediate the phosphorylation- and arrestin-independent desensitization of the metabotropic glutamate receptor 1a and GABAB receptor (Dhami et al. 2002, 2004; Perroy et al. 2003). The specificity of GRK interactions with many GPCRs has been the subject of intense interest (see Freedman and Lefkowitz 1998). In general, GRK2, GRK3 and GRK5 contribute to phosphorylation and desensitization of many GPCRs. However, GRK4 contributes to the desensitization of metabotropic glutamate receptor 1a and GRK6 is linked to the regulation of chemokine receptor, muscarinic acetylcholine receptor and dopamine receptor desensitization (Gainetdinov et al. 2003; Willets et al. 2003; Vroon et al. 2004). These observations suggest that the specificity of GRK-mediated phosphorylation is probably the consequence of differences in tertiary structure of different GPCRs and this may also contribute to the specificity of β-arrestin interactions.

Rab GTPases regulate the trafficking of endosomal vesicles between distinct membrane compartments and also serve as markers for these compartments (Pfeffer 2003; Seachrist and Ferguson 2003). Using Rab GTPases as markers, we showed that CRF1α receptor is internalized to the Rab5-positive early endosomes in both HEK293 cells and primary cortical neurons. Subsequently, the receptor appears to transit the early endosome to Rab4 recycling endosomes and does not appear to localize to either the Rab11-regulated recycling pathway or lysosomes in either HEK293 cells or cortical neurons. Previous evidence from our laboratory demonstrated that the dephosphorylation and resensitization of the β2AR occurs as the receptor transits between Rab5- and Rab4-regulated endosomal compartments (Seachrist et al. 2000). Thus, the localization of CRF1α receptor to Rab5 and Rab4 endosomal compartments is consistent with the time course of CRF1α receptor resensitization.

Previous studies demonstrated that the expression of a dominant-negative Rab5-S34N mutant could block the internalization of the β2AR and D2 dopamine receptor (Iwata et al. 1999; Seachrist et al. 2000). Rab5-S34N overexpression also prevented resensitization of the β2AR (Seachrist et al. 2000). In contrast, in the present study both the overexpression of wild-type and dominant-negative Rab5 constructs prevented CRF1α receptor resensitization. Rab5 overexpression has been demonstrated to cause the homotypic fusion of endocytic vesicles (Zerial and McBride 2001). Thus, it is possible that Rab5 overexpression results in the perturbation of CRF1α trafficking from the Rab5-positive early endosome to the Rab4 recycling endosome, thereby preventing receptor resensitization. Moreover, we previously demonstrated that endocytosis of the AT1AR, which forms a complex with Rab5, is not affected by the expression of Rab5-S34N (Seachrist et al. 2002). Similarly, wild-type Rab4 overexpression reduced CRF1α resensitization, whereas the expression of a dominant-negative Rab4-N121I mutant had no effect on CRF1α resensitization. Again, these results are difficult to reconcile with the previous finding that expression of the Rab4-N121I mutant prevented β2AR resensitization (Seachrist et al. 2000). However, disparate results following the overexpression of wild-type Rab4 and Rab4-N121I mutant have been reported previously for the transferrin receptor (van der Sluijs et al. 1992). Specifically, the overexpression of both wild-type and Rab4-N121I mutant prevented trafficking of the transferrin receptor to early endosomes, thereby preventing the acidification of the receptor and the dissociation of its ligand. Because localization of GPCRs to acidified endosomes is considered to be essential for receptor dephosphorylation (Krueger et al. 1997), it is possible that wild-type Rab4 overexpression has the same effect on CRF1α receptor trafficking. In contrast, Rab4-N121I expression may shunt the CRF1α receptor into the Rab11 recycling pathway. This might be similar to the overexpression of Rab11-Q70L, which shunts the AT1AR into recycling endosomes allowing the membrane recycling of the receptor (Dale et al. 2004).

In conclusion, we showed that CRF1α receptor internalization is predominantly β-arrestin2 dependent and that β-arrestin association with the CRF1α receptor is specifically facilitated by the overexpression of either GRK3 or GRK6. In addition, we demonstrated that the selectivity for β-arrestin2 binding to the CRF1α receptor is observed in both HEK293 cells and primary cortical neurons. However, there are differences in the intracellular trafficking fate of β-arrestin2 in HEK293 cells compared with primary neurons. Specifically, in primary neurons β-arrestin2 and CRF1α receptors are trafficked to distinct populations of endocytic vesicles. Finally, we have provided evidence that CRF1α receptor is internalized to Rab5-positive early endosomes and Rab4-positive recycling endosomes in both HEK293 cells and neurons. However, CRF1α receptor resensitization appears to be regulated in an atypical manner in the presence of overexpressed wild-type and dominant-negative Rab GTPases.

Acknowledgements

KDH is the recipient of a Ontario Mental Health Foundation fellowship. AVB is the recipient of a Canadian Hypertension Society/Canadian Institute of Health Research fellowship. SSGF is the recipient of a Canada Research Chair in Molecular Neuroscience and is a Heart and Stroke Foundation of Ontario Career Investigator. This work was supported by CIHR grant MA-15506 and CIHR grant MOP 62738 to SSGF, and CIHR grant MOP 100765 to MOP.

Ancillary