β-Arrestin1 and 2 differentially regulate PACAP-induced PAC1 receptor signaling and trafficking

A pituitary adenylate cyclase-activating polypeptide (PACAP)-specific receptor, PAC1R, is coupled with multiple signal transduction pathways including stimulation of adenylate cyclase, phospholipase C and extracellular-signal regulated kinase (ERK)1/2. PAC1R has been shown to exert its long-lasting and potent signals via β-arrestin1 and β-arrestin2. However, the precise roles of the two β-arrestin isoforms in PACAP-PAC1R signaling remain unclear. Here we examined the interaction between the two β-arrestin isoforms and PAC1R, β-arrestin-dependent PAC1R subcellular localization and ERK1/2 activation. Upon PACAP stimulation, although PAC1R similarly interacted with β-arrestin1 and β-arrestin2 in HEK293T cells, the complex of PAC1R and β-arrestin2 was translocated from the cell surface into cytosol, but that of β-arrestin1 remained in the cell surface regions in HeLa cells and mouse primary cultured neurons. Silencing of β-arrestin2 blocked PACAP-induced PAC1R internalization and ERK1/2 phosphorylation, but silencing of β-arrestin1 increased ERK1/2 phosphorylation. These results show that β-arrestin1 and β-arrestin2 exert differential actions on PAC1R internalization and PAC1R-dependent ERK1/2 activation, and suggest that the two β-arrestin isoforms may be involved in fine and precise tuning of the PAC1R signaling pathways.


Cell culture
HEK293T cells were maintained in Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 10% fetal bovine serum. HeLa cell line was provided by the RIKEN BRC (Tsukuba, Ibaraki, Japan), through the National Bio-Resource Project of the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan. HeLa cells were maintained in DMEM (high glucose, GlutaMAX) supplemented with 10% fetal bovine serum (Thermo Fisher Science, Tokyo, Japan).
Primary cultures of cortical neurons were prepared as described previously [31]. Pregnant mother mice (strain, ICR) were purchased from JAPAN SLC (Shizuoka, Japan). The mice were deeply anesthetized by intraperitoneal injection of three-types mixed anesthetic agents (medetomidine hydrochloride, 0.75 mg/kg body weight; midazolam, 4 mg/kg body weight; butorphanol tartrate, 5 mg/kg body weight) and were removed fetuses (embryonic day 16). All animal care and handling procedures were performed in accordance with protocols approved by the Animal Care and Use Committee of the Graduate School of Pharmaceutical Sciences, Osaka University. Dissociated cortical cells were plated in Neurobasal medium (Invitrogen, Carlsbad, CA, USA), supplemented with B27 (Thermo Fisher Scientific) and L-glutamine (0.5 mM) at 4.0×10 5 cells/well in 96-well dishes coated with poly-l-lysine (for NanoBiT assay), 1.2×10 6 cells/well in 6-well dishes (for co-immunoprecipitation assay) or 2.5×10 4 cells/well in 35 mm glass-bottom dishes (for time-lapse imaging).
All animal care and handling procedures were performed in accordance with protocols approved by the Animal Care and Use Committee of the Graduate School of Pharmaceutical Sciences, Osaka University.
For Time-lapse imaging, PAC1R-SmBiT-pQM512B lentivirus together with β-arrestin1-Lg-BiT-pQM512B or LgBiT-β-arrestin2-pQM512B lentivirus were infected into HeLa cells or primary cultured cortical neurons of 3 days in vitro (DIV 3). After Nano-Glo Live Cell Reagent (Promega) was added, the luminescence was measured using Olympus LV200 bioluminescence imager (Olympus, Tokyo, Japan) and the time-lapse images were processed either using Metamorph software (Molecular Devices Japan, Tokyo, Japan) or the ImageJ software (https:// imagej.net/). To assess the luminescence intensity at the plasma membrane and the cytoplasm, we defined the shape of the whole-cell (region of interest (ROI) A) and the cytoplasm region (ROI B) by reducing the size by 7 pixels and determined the luminescence in the both ROIs. The amount of luminescence at the vicinity of the plasma membrane was defined by subtracting the amount of luminescence in ROI A by that in ROI B. All analyses were performed in a blind manner.

PAC1R internalization
Internalization of PAC1R was quantitatively assessed using HaloTag technology according to the manufacturer's protocol (Promega). HEK293T cells were transfected with PAC1R-Haloexpressing vector together with β-arrestin1 siRNA, β-arrestin2 siRNA or control siRNA; the cells were then labeled with the cell-impermeable Alexa Fluor 488 ligand (Promega) in Opti-MEM for 15 min at 37˚C. Clathrin-mediated endocytosis inhibitor, 250 μg/ml ConA or 15 μM Pitstop2, were pretreated for 30 min. After 30 min, the cells were stimulated with 1 nM PACAP or saline, and were then washed with phosphate-buffered saline and fixed in 4% paraformaldehyde. The cell images were obtained using FV1000D confocal microscope (Olympus) in a sequential mode and membrane protein internalization was quantified using the ImageJ software. To assess the internalization ratio of PAC1R, we defined the shape of the whole-cell (ROI A) and the cytoplasm region (ROI B) by reducing the size by 5-10 pixels and determined the fluorescence in the both ROIs. The internalization ratio (%) was defined by dividing the amount of luminescence in ROI B by that in ROI A. All analyses were performed in a blind manner.

Chemical crosslinking and co-immunoprecipitation
For the detection of PACAP-induced association of PAC1 and β-arrestin1 or β-arrestin2, covalent protein cross-linking with a chemical crosslinker, Dithiobis succinimidylpropionate (DPS; Thermo Fisher Scientific) was used as described previously [32,33]. HEK293T cells were washed with PBS containing 10 mM Hepes (pH 7.4) and were incubated with 2.5 mM DSP for 30 min at room temperature. The reactions were quenched by the addition of 0.1 ml of 1 M Tris (pH 7.5). For prevention of co-elution of IgG fragments, the immobilized antibodies were covalently cross-linked to protein G-Sepharose beads. Antibody-coupled protein G-Sepharose beads were resuspended in 5 mM BS3 (Thermo Fisher Scientific) and incubated for 30 min at room temperature. Reactions were quenched by the addition of 0.1 ml of 1 M Tris (pH 7.5). Cells were lysed in RIPA buffer and the resultant lysates were incubated with the indicated antibody-coupled protein G-Sepharose beads for 2 h at 4˚C. The beads were then washed three times with RIPA buffer and suspended in SDS sample buffer.

Statistical analysis
Experimental data were analyzed using one-way or two-way analysis of variance (ANOVA). Fisher-PLSD post hoc tests were also performed after significant main effects for drug, time or luminescence intensity were observed. The criterion for statistical significance was p < 0.05. Statistical analyses were performed using Stat View software (version 5.0; SAS Institute, Cary, NC, USA).

Time-lapse cell imaging of PAC1R and β-arrestin coupling and translocation in HeLa cells and primary cultured cortical neurons
We examined whether β-arrestin1 and β-arrestin2 are implicated in the PAC1R internalization by time-lapse cell imaging to visualize PAC1R and the two β-arrestin isoforms coupling and translocation in HeLa cells. At as early as 3 min after 1 μM PACAP stimulation, the NanoBiT signals from β-arrestin1 and β-arrestin2 (β-arrestin1-LgBiT and LgBiT-β-arrestin2) coexpressed with PAC1R-SmBiT were clearly detectable and localized around the plasma membrane and cytoplasm. At 15 min or later, however, the signal from PAC1R-β-arrestin1 complex remained localized around the plasma membrane, whereas that from PAC1R-β-arrestin2 complex localized predominantly at the cytoplasm (Fig 2; S1 and S2 Movies). Time-dependent changes of line-scan images and quantitative analysis of time course changes in luminescence at the vicinity of the plasma membrane and the cytoplasm showed that PAC1R-β-arrestin2 complex was translocated from the plasma membrane to the cytoplasm and unevenly clustered, while PAC1R-β-arrestin1 complex remained around the plasma membrane (Fig 2C-2F To confirm the PAC1R and β-arrestin coupling in more biologically relevant cells, we also used mouse primary cultured cortical neurons. PACAP increased the luciferase luminescence with similar dose-dependency and time-course in cortical neurons infected with PAC1R-Sm-BiT-pQM512B lentivirus in combination with either β-arrestin1-LgBiT-pQM512B or LgBiTβ-arrestin2-pQM512B lentiviruses (Fig 3A-3D). These dose-and time-dependent changes were virtually similar as those observed in HEK293T cells (Fig 1C-1F). In addition, time-lapse cell imaging and time-dependent changes of line-scan images of these cortical neurons were very similar to those observed in HEK293T cells (Fig 3E-3H; S3 and S4 Movies). These results suggest that PAC1R is internalized via distinct pathways in complex with β-arrestin1 and β-arrestin2.

Silencing of β-arrestin2, but not β-arrestin1, inhibits prolonged ERK1/2 activation
We addressed whether β-arrestin2 is also implicated in the PACAP-induced ERK1/2 activation in HEK293T cells (Fig 5). In the absence of PACAP, phosphorylated ERK1/2 levels were not significantly changed by the silencing of β-arrestin1 or β-arrestin2. PACAP (1 μM)-increased ERK1/2 phosphorylation at an early time point (3 min) was not changed by the β-arrestin2 silencing but rather increased by the β-arrestin1 silencing compared with the control siRNA. At a later time point (25 min), phosphorylated ERK1/2 levels were decreased by the β-arrestin2 silencing while they were increased by the β-arrestin1 silencing compared with the control siRNA. To examine whether PACAP-induced ERK1/2 phosphorylation is required for the PAC1R internalization, the cells were pretreated with 250 μg/ml ConA or 15 μM Pitstop2 before treatment with 1 μM PACAP. Pretreatment with ConA significantly decreased the PACAPincreased ERK1/2 phosphorylation (Panels F and G in S3 Fig). Likewise, Pitstop2 almost completely inhibited the PACAP-increased ERK1/2 phosphorylation (Panels F and G in S3 Fig).

Discussion
In the present study, we have investigated PAC1R internalization that depends on β-arrestin1 and β-arrestin2 using the NanoBiT system, HaloTag technology and siRNA-mediated silencing of endogenous β-arrestins in HEK293T cells, HeLa cells and primary cultured cortical neurons. For this purpose, we used time-lapse luminescence microscopy and processed the images to detect intracellular translocation. In addition, we examined the roles of β-arrestin1 and β-arrestin2 involved in the PACAP-stimulated PAC1R-mediated prolonged ERK1/2 activation relevant to the PAC1R internalization.
Here we showed that PACAP-stimulated PAC1R interacted with β-arrestin1 and β-arrestin2 with similar PACAP dose-and time-dependency and that subcellular distribution of the PAC1R and β-arrestin2 complex changed time-dependently, whereas the complex with β-arrestin1 seemed to remain at the cell surface regions. We also showed that silencing of β-arrestin2 significantly reduced PACAP-induced PAC1R internalization and prolonged ERK1/ 2 activation, but silencing of β-arrestin1 did not affect PAC1R internalization and rather increased ERK1/2 phosphorylation. Moreover, we found evidence suggesting that PACAPinduced ERK1/2 phosphorylation was dependent on PAC1R internalization using inhibitors of clathrin-mediated endocytosis, ConA and Pitstop2. Taken together, these results suggest mean (C and D) of three independent experiments each conducted in duplicate. Ã P < 0.05, ÃÃ p < 0.001 and ÃÃÃ p < 0.001 vs. 0 nM PACAP, one-way ANOVA followed by Fisher-PLSD test. See also S3 and S4 Movies.
https://doi.org/10.1371/journal.pone.0196946.g003 PAC1 receptor internalization differentially regulated by β-arrestin1 and 2 that β-arrestin2 but not β-arrestin1 is critically involved in the PAC1R endocytosis and subsequent activation of the ERK1/2 signalosome. These findings contribute to the mechanistic understanding of previous studies showing that PAC1R endocytosis is implicated in the PACAP-increased ERK1/2 activation [20,27]. However, the residual β-arrestin1 after β-arrestin1 knockdown may be enough for the PAC1R endocytosis and subsequent ERK1/2 activation. Further studies with genome editing technology are needed for elucidating the precise functional diversity of β-arrestin1 and β-arrestin2 in the PACAP signaling.
In the current study, we observed that silencing of β-arrestin2 inhibits prolonged ERK1/2 activation (25 min after PACAP stimulation), whereas, in contrast, β-arrestin1 siRNA increases ERK1/2 activation (3 min after PACAP stimulation) in HEK293T cells. The same reciprocal activity of the two β-arrestin isoforms on ERK1/2 activation has been demonstrated for the angiotensin II type 1A receptor-mediated activation of ERK1/2, which is increased under β-arrestin1 silencing but is eliminated under β-arrestin2 silencing in HEK-293 cells [40]. In contrast to our current results, Gupte et al. have demonstrated that PACAP-induced ERK1/2 activation mediated by PAC1R (the hop1 and hop2 splicing variants) is abolished by β-arrestin1 silencing, but not by β-arrestin2 silencing, in HEK293T cells [41]. The reason for the opposing results in the β-arrestin isoform-dependent regulation of ERK1/2 activation is currently unknown. The discrepancies might be explained by the differential experimental conditions; Gupte et al. used the overexpressed PAC1R and β-arrestins, while we use endogenous PAC1R and β-arrestins in this study. The residual β-arrestin after β-arrestin knockdown should also be taken into account. May et al. demonstrated that PACAP-stimulated ERK phosphorylation is diminished by Pitstop 2 and dynasore, inhibitors of clathrin-mediated endocytosis [20]. They have also demonstrated that PACAP-stimulated protein kinase C signaling contributes to ERK phosphorylation [20]. In the present study we observed that a protein kinase C inhibitor, D-sphingosine, shows no effect on the interaction of PAC1R with either β-arrestin1 or β-arrestin2 (Panels C and D in S2 Fig). These results are consistent with those reported by May et al. [20] and suggest that protein kinase C may be required for PACAPstimulated ERK1/2 phosphorylation downstream of PAC1R internalization. Moreover, we found that treatment with inhibitors of clathrin-mediated endocytosis, ConA and Pitstop2, reduces PACAP-stimulated ERK1/2 activation. These results suggested that PACAP-induced ERK1/2 phosphorylation was dependent on PAC1R internalization. We also observed that β-arrestin2 plays a role in PAC1R internalization using β-arrestin2 siRNA. These results indicate that PAC1R-activation generally leads to β-arrestin-scaffolding of the ERK1/2 cascade and enhances ERK1/2 activity, while the two β-arrestin isoforms involved in this pathway depend on tissues and physiological conditions. The present results suggest that β-arrestin1 and β-arrestin2 might be involved in fine and precise tuning of the PAC1R signaling pathways.
The finding of the β-arrestin-mediated signaling implicated signaling-biased ligands that favor G-protein signaling over arrestin recruitment or vice versa, which proposes therapeutic potential [21,24,[42][43][44]. In the case of antipsychotic drugs, G protein-and β-arrestin2-biased functional selectivity at dopamine D 2 receptor is expected to have the potential to correct both positive and cognitive symptoms of schizophrenia [45]. Because PACAP and VIP signals might be related to psychiatric and neurological disorders, the present observations as well as the assay systems generated are expected to contribute to therapeutic drug development.