Agonist-induced phosphorylation bar code and differential post-activation signaling of the delta opioid receptor revealed by phosphosite-specific antibodies

The δ-opioid receptor (DOP) is an attractive pharmacological target due to its potent analgesic, anxiolytic and anti-depressant activity in chronic pain models. However, some but not all selective DOP agonists also produce severe adverse effects such as seizures. Thus, the development of novel agonists requires a profound understanding of their effects on DOP phosphorylation, post-activation signaling and dephosphorylation. Here we show that agonist-induced DOP phosphorylation at threonine 361 (T361) and serine 363 (S363) proceeds with a temporal hierarchy, with S363 as primary site of phosphorylation. This phosphorylation is mediated by G protein-coupled receptor kinases 2 and 3 (GRK2/3) followed by DOP endocytosis and desensitization. DOP dephosphorylation occurs within minutes and is predominantly mediated by protein phosphatases (PP) 1α and 1β. A comparison of structurally diverse DOP agonists and clinically used opioids demonstrated high correlation between G protein-dependent signaling efficacies and receptor internalization. In vivo, DOP agonists induce receptor phosphorylation in a dose-dependent and agonist-selective manner that could be blocked by naltrexone in DOP-eGFP mice. Together, our studies provide novel tools and insights for ligand-activated DOP signaling in vitro and in vivo and suggest that DOP agonist efficacies may determine receptor post-activation signaling.

poly-L-lysine-coated 60-mm dishes and grown for 48 h to 80% confluency. Cells were lysed with detergent buffer (50 mM Tris-HCl, pH 7.4; 150 mM NaCl; 5 mM EDTA; 10 mM NaF; 10 mM disodium pyrophosphate; 1% Nonidet P-40; 0.5% sodium deoxycholate; 0.1% SDS) in the presence of protease and phosphatase inhibitors after treatment with agonists, directly. Where indicated, cells were preincubated with antagonists or GRK2/3 inhibitor compound 101 for 30 min before agonist exposure. Cells were centrifuged for 30 min at 4 °C followed by receptor enrichment using wheatgerm-lectin-agarose beads. Samples were inverted for 2 h at 4 °C. Where indicated, DOP receptor was dephosphorylated using lambda protein phosphatase (Santa Cruz; Heidelberg, Germany) for 1 h at 30 °C. After washing three times, proteins were eluted using SDS sample buffer for 30 min at 50 °C. Proteins were separated on 7.5% or 12% SDS-polyacrylamide gels. After electroblotting, membranes were incubated with either 0.1 µg/ml anti-pT361 (5038) or anti-pS363 antibodies over night at 4 °C. Subsequently, blots were washed followed by detection using enhanced chemiluminescence detection (ECL) (Thermo Fisher Scientific; Schwerte, Germany) of bound antibodies. Blots were thereafter stripped and reprobed with the anti-HA antibody (2238) to ensure equal loading of the gels.

Analysis of Dop receptor internalization.
HEK293-cells stably expressing HA-tagged DOP or receptor mutants were plated onto poly-L-lysine-coated coverslips and grown for 24 h. Next, cells were incubated with rabbit anti-HA antibody (2238) in serum-free medium for 2 h at 4 °C. Cells were fixed with 4% paraformaldehyde and 0.2% picric acid in phosphate buffer (pH 6.9) for 30 min at room temperature after agonist or antagonist exposure for 30 min at 37 °C. Subsequently, cells were washed three times with phosphate buffer (22.6 ml/L 1 M NaH 2 PO 4 •H 2 O; 77.4 ml/L 1 M Na 2 HPO 4 •H 2 O; 0.1% Triton X-100, pH 7.4) and permeabilized. After incubation with an Alexa488-coupled goat anti-rabbit antibody (Invitrogen; Darmstadt, Germany), cells were mounted and receptor internalization was examined using a Zeiss LSM510 META laser scanning confocal microscope (Jena, Germany). For quantitative internalization assays, stably HA-hDOP receptor transfected HEK293-cells were plated onto 24-well plates and grown overnight. After preincubation with anti-HA antibody (2238) for 2 h at 4 °C, cells were exposed to agonists or antagonists for 30 min at 37 °C and subsequently fixed for 30 min at room temperature. Cells were washed several times with PBS and incubated with a peroxidase-conjugated secondary antibody (Santa Cruz; Heidelberg, Germany). After additional washing steps, the HRP-substrate ABTS was added and optical density was measured at 405 nm using an iMark ™ Microplate Absorbance Reader (BioRad, Munich, Germany).
Membrane potential assay. Membrane potential change was measured as previously described 94  5.5 mM glucose. Afterwards, cells were incubated with membrane potential dye (FLIPR Membrane Potential kit BLUE, Molecular Devices, Biberach, Germany) for 45 min at 37 °C. The final injection volume of compounds or vehicle was 20 µl and the initial volume in the wells was 180 µl (90 µl buffer and 90 µl dye). Finally, 20 µl of compound was added to the cells (final volume in the well was 200 µl resulting in a 1:10 dilution of the compound). The compounds were prepared at 10x concentrations. Compounds or buffer were injected after 60 seconds baseline measurement. Right after injection of compounds or vehicle membrane potential change was measured at 37 °C using a FlexStation 3 microplate reader (Molecular Devices; Biberach, Germany). The buffer-only trace for each corresponding data point was subtracted from the data after normalization to the baseline. Data analysis. Protein bands detected on Western blots were quantified using ImageJ 1.47 v software (National Institute of Health, Bethesda, MD, USA). Data (Western blots and ELISA) were analyzed using GraphPad Prism 5 software (La Jolla, CA, USA). Statistical analysis was carried out with one-way ANOVA followed by Bonferroni correction. P values < 0.05 were considered statistically significant. Dose response curves were analyzed and compiled with OriginPro.

Results
Characterization of phosphosite-specific antibodies to analyze agonist-induced phosphorylation of T361 and S363 in the carboxyl-terminal tail of DOP receptor. There are several potential phosphate acceptor sites in the intracellular loops and within the carboxyl-terminal tail of human DOP receptor (Fig. 1A). To examine the temporal and spatial dynamics of DOP receptor phosphorylation in its carboxyl-terminal tail, we generated a polyclonal phosphosite-specific antibody for the carboxyl-terminal residue T361, and we used the commercially available anti-S363 antibody (Fig. 1A). In addition, we generated an anti-HA antibody which binds to an amino-terminally fused hemagglutinin epitope tag (HA-tag) in order to detect HA-DOP receptor independent of phosphorylation status (Fig. 1A). First, antisera were affinity purified against their immunizing peptides and specificity was then verified with corresponding synthetic peptides using dot-blot assays. All antibodies, which clearly detected their respective peptide were further characterized using Western blot analysis. The anti-pT361 [5038] and anti-pS363 antibodies specifically detected the respective T361-or S363phosphorylated form of DOP receptor after DADLE stimulation of HEK293 cells stably transfected with human HA-DOP receptor (Fig. 1B). Both phosphosite-specific antibodies were no longer able to detect their cognate forms of phosphorylated DOP receptor after treatment with lambda phosphatase (Fig. 1B), but the receptor protein was still detectable using the anti-HA antibody [2248] (Fig. 1B). Different DOP receptor mutants were generated for further characterization of the phosphosite-specific antibodies (Fig. 1C). As expected, no phosphorylation signal for pT361 and pS363 was detectable after DPDPE stimulation in the T361A/S363A mutant as well as after global mutation of all serine and threonine residues present in the carboxyl-terminal tail (7S/T-A) (Fig. 1D). Also, internalization after DPDPE incubation was reduced in the T361A/S363A mutant (Fig. 1E,F). A stronger inhibition of DOP receptor internalization was detectable in the 7S/TA mutant (Fig. 1E,F). These results show that the phosphosite-specific antibodies directed against T361 and S363 in the carboxyl-terminal tail clearly recognize only the respective phosphorylated form of DOP receptor and that the mutation of only T361 and S363 is sufficient to diminish DOP receptor internalization.

Dop receptor phosphorylation and internalization occur in a time-and
concentration-dependent manner with S363 as primary phosphorylation site. We then examined the DPDPE-induced DOP receptor internalization using fluorescence microscopy and for quantification we used a cell-surface enzyme-linked immunosorbent assay (ELISA). DOP receptor internalization was initiated after treatment with 10 nM DPDPE and reached a maximum after incubation with 10 µM DPDPE ( Fig. 2A,B). We then examined the time-course of DPDPE-induced T361 and S363 phosphorylation and receptor internalization. After DPDPE stimulation, a robust phosphorylation at S363 was detectable within 2 min, which remained at high levels throughout the 30 min treatment period. T361 phosphorylation was first detectable after 5 min following DPDPE incubation and increased throughout the 30 min treatment period (Fig. 2C). To determine the DOP receptor phosphorylation time-course in more detail, agonist was added to the cells at room temperature (RT) for shorter time periods. Under these conditions, S363 phosphorylation occurred within 40 s, whereas T361 phosphorylation became first detectable after 10 min, suggesting that S363 is the primary site of phosphorylation, followed by T361 (Fig. 2D).
Internalization of DOP receptor was first detectable after 5 min DPDPE treatment and reached a maximum after 20 min (Fig. 2E,F). DPDPE-induced DOP receptor phosphorylation and internalization occur in a time-dependent manner with S363 as primary phosphorylation site followed by T361. Also, receptor internalization takes place in a concentration-dependent manner.

DOP receptor phosphorylation is mediated by GRK2 and GRK3. Phosphorylation of GPCRs
can be mediated by different types of kinases, G protein-coupled receptor kinases (GRKs) and second messenger-activated kinases (e.g. PKA, PKC). To examine if DOP receptor phosphorylation could also be mediated heterologously by PKA or PKC, we therefore incubated cells with phorbol-12-myristat-13-acetat (PMA) or forskolin. Neither forskolin nor PMA induced any detectable phosphorylation at T361 or S363 (Fig. 3A). To evaluate which GRK isoforms are mediating the DPDPE-induced DOP receptor phosphorylation, we used the chemical GRK2/3 selective inhibitor compound 101 (cmpd101) as well as siRNA knockdown experiments. DPDPE-induced phosphorylation at T361 and S363 is reduced in a concentration-dependent manner after inhibition of GRK2/3 activation using compound 101 (Fig. 3B). Treatment with specific GRK2 or GRK3 siRNA sequences also led to a significant reduction of DPDPE-induced phosphorylation at T361 and S363 (Fig. 3C). It is possible that the loss of either GRK2 or GRK3 could be compensated for by the remaining isoform, because of the close relationship between the two GRK isoforms. Therefore, we evaluated the inhibitory effect of siRNA knockdown of both GRK2 and GRK3 on DOP receptor phosphorylation. A combination of siRNA knockdown of both GRK isoforms produced a stronger inhibition of T361 and S363 phosphorylation, indicating that GRK2 and GRK3 function as a redundant system for DPDPE-induced DOP receptor phosphorylation (Fig. 3C). To rule out that also GRK5 and GRK6 were involved in DPDPE-induced DOP receptor phosphorylation, we performed the same siRNA knockdown experiments for the two GRK isoforms. As expected, knockdown of GRK5, GRK6 or a combination of both, could not reduce the DPDPE-induced phosphorylation signal neither at T361 nor at S363 (Fig. 3D). These results suggest that GRK2 and GRK3 are responsible for DPDPE-induced DOP receptor phosphorylation at T361 and S363.

Dop receptor agonists induce varying levels of receptor phosphorylation and internalization.
We next surveyed a large selection of chemically diverse selective DOP receptor ligands and clinically used opioids for their ability to induce DOP receptor phosphorylation and internalization. We consistently observed that endocytotic activity of these selective DOP receptor agonists and common opioids correlated with their ability to induce receptor phosphorylation at T361 and S363 (Fig. 4). Robust phosphorylation and were preincubated with antibody to HA-tag and subsequently exposed to 10 µM DPDPE or vehicle for 30 min at 37 °C. Cells were fixed, permeabilized, immunofluorescently stained and examined using confocal microscopy. Images are representative from one of three independent experiments. Scale bar, 20 µm. (F) Receptor internalization was quantified by ELISA. Cells described in (D) were preincubated with antibody to HA-tag and stimulated with 10 µM DPDPE or vehicle at 37 °C for 30 min. Cells were fixed and labeled with a peroxidaseconjugated secondary antibody. Receptor internalization was measured by enzyme-linked immunosorbent assay and quantified as the percentage of internalized receptors in DPDPE-treated cells. Data are mean ± SEM of five independent experiments performed in quadruplicate. Results were analyzed by one-way ANOVA followed by Bonferroni's post-hoc test (*p < 0.05).
receptor internalization were detectable after treatment with DPDPE, DADLE, SNC80, ADL5859, AR-M1000390, deltorphin I, deltorphin II, [Met]-enkephalin, [Leu]-enkephalin and norbuprenorphine (Fig. 4). Fentanyl and (−)-methadone as well as morphine-6-glucuronide, an active metabolite of morphine, induced phosphorylation only at S363, whereas morphine and buprenorphine failed to induced DOP receptor phosphorylation (Fig. 4). The DOP-selective agonists ADL5859 and AR-M1000390 had been previously described as non-or weakly internalizing agonists 79,95 . A significantly higher concentration (10 µM) was used for all compounds in order to distinguish between DOP agonists which induces a strong, partial or no internalization. These results confirm the consensus model that GPCR phosphorylation is highly correlated with, and a prerequisite for, subsequent internalization. www.nature.com/scientificreports www.nature.com/scientificreports/ Correlation analysis between G protein-dependent (GIRK activation) and arrestin-dependent (internalization) effects of the tested compounds revealed positive linear correlation (r = 0.8551; Fig. 4D), suggesting that DOP receptor agonists do not display any bias in intracellular coupling.
Chemically diverse DOP receptor agonists show varying efficacies in GIRK channel activation (G protein-dependent signaling). A previously described fluorescence-based membrane potential assay, that detects Gβγ-mediated activation of inwardly rectifying potassium (GIRK) channels was used to examine G protein signaling of DOP receptor at high temporal resolution 94 . We studied the ability of different DOP receptor ligands to activate GIRK channels (Fig. 5, Table 1). Notably, deltorphin II (EC 50 of 0.11 ± 0.09 nM) and DADLE (EC 50 of 0.16 ± 0.04 nM) were the most potent agonists tested and compared to DPDPE (EC 50 of 0.39 ± 0.14 nM) (  (Fig. 5B, Table 1). ADL5859 and AR-M1000390 exhibited partial agonist activity with EC 50 values of 57.02 ± 10.94 nM and 36.45 ± 10.14 nM (Fig. 5C,D). Morphine and fentanyl showed only a weak activity with a remarkably reduced maximal effect compared to DPDPE (Fig. 5E,F; Table 1). Together with our results from phosphorylation and internalization studies, these observations suggest that DOP agonists range from partial to full agonism without intracellular signaling bias (Fig. 4D).

DpDpe-induced phosphorylation and internalization is inhibited by the Dop receptor antagonist.
Both, the non-selective opioid receptor antagonist naloxone and naltrexone and the selective DOP receptor antagonists naltrindole [96][97][98] and naltriben 99-101 block DPDPE-induced phosphorylation at T361 and S363 as well as receptor internalization (Fig. 6A-C). Moreover, addition of naltrindole induced a reversal of the DPDPE-induced hyperpolarization towards baseline level in the GIRK channel activation assay (Fig. 6D).
Agonist-selective Dop receptor phosphorylation is also observed in vivo in mouse brain. The fact that corresponding mouse DOP receptor phosphorylation sites are located at equivalent positions www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ compared to the human DOP receptor (Fig. 7A) enabled us to use the phosphosite-specific antibodies to analyze agonist-induced phosphorylation in vivo in more detail. DOP-eGFP mice were treated with different types of systemically active DOP receptor ligands. The selective DOP full agonist SNC80 induced a strong and concentration-dependent phosphorylation at T361 and S363 which could be blocked by naltrexone (Fig. 7B-D), similar to observations made in vitro for this combination of compounds (Suppl. Fig. 1). A weaker increase of in vivo DOP phosphorylation signal was detectable after administration of ADL5859 and AR-M1000390 in comparison to saline treatment (Fig. 7B). However, the ADL5859 and AR-M1000390 doses used were much higher than behaviorally necessary (effective dose in most behavior assays: 10 mg/kg AR-M1000390 and ADL5859 95,102,103 ) (Fig. 7B), albeit inducing weaker receptor phosphorylation than SNC80. Therefore, we selected SNC80 for dose-response in vivo experiments (Fig. 7C). These results indicate that DOP receptor phosphorylation occurs after stimulation by selective agonists in a concentration-dependent manner in vivo. We also observed significant constitutive phosphorylation of T361 (unstimulated controls in Fig. 7B-D) that could be related to the previously reported high constitutive activity of DOP receptors 104 .  www.nature.com/scientificreports www.nature.com/scientificreports/ DOP receptor dephosphorylation occurs in a time-dependent manner with S363 as primary dephosphorylation site. Phosphorylation at S363, the primary phosphorylation site, occurred during the first minute whereas a phosphorylation signal at T361 was first detectable after prolonged stimulation with DPDPE (10 min). Therefore, we examined DOP receptor dephosphorylation, using three different buffer, in order to investigate if distinct temporal dynamics exist between both phosphorylation sites. Washout with citric acid buffer removes high-affnity agonsits and disrupt receptor-ligand-binding more efficiently than phosphate buffer 105 . Naltrindole was added to citric acid buffer to further facilitate displacement of DPDPE from the receptor and thus terminate agonist stimulation. Cells were washed with citric acid after DPDPE stimulation and then incubated in agonist-free medium with or without naltrindole to differentiate the dephosphorylation time in more detail. Dephosphorylation of T361 and S363 occurred more quickly after washout with citric acid buffer and naltrindole. Interestingly, no differences were observed in dephosphorylation time between T361 and S363 (Fig. 8). These results indicate that there is no primary site of dephosphorylation.
Dephosphorylation of DOP receptor is mediated by PP1. Classified by their catalytic subunits, seven families of serine/threonine-specific protein phosphatases, PP1-PP7, have been identified [106][107][108] . Calyculin A is an inhibitor of PP1 and PP2 activity to a similar extent [108][109][110] . In contrast, okadaic acid can block the activity of PP2, www.nature.com/scientificreports www.nature.com/scientificreports/ PP4, and PP5, but has little effect on PP1 activity [108][109][110][111][112][113][114] . Both inhibitors are not able to reduce the activity of PP3. When cells stably expressing DOP receptor were exposed to escalating concentrations of calyculin A or okadaic acid, DOP receptor dephosphorylation was inhibited in a concentration-dependent manner only by calyculin A (Fig. 9A,B). Thus, the present results strongly suggest that PP1 activity is required for DOP receptor dephosphorylation at T361 and S363.
We next performed siRNA knockdown experiments to confirm these results and to evaluate the contribution of the catalytic subunits PP1α, PP1β and PP1γ to DOP receptor dephosphorylation. Simultaneous knockdown of all three PP1 catalytical subunits nearly completely blocked DOP receptor dephosphorylation in DPDPE-treated cells (Fig. 10A). Only inhibition of PP1α and PP1β expression resulted in a robust reduction of dephosphorylation at T361 and S363 (Fig. 10B). In contrast, PP1γ siRNA knockdown did not attenuate T361 and S363 dephosphorylation (Fig. 10B). Moreover, inhibition of PP2α and PP2β had no effect on DOP receptor dephosphorylation (data not shown). These results confirmed that PP1 activity was required for efficient DOP receptor dephosphorylation, most likely mediated by PP1α and/or PP1β.  www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
In the present study, we used phosphosite-specific antibodies for DOP to analyze distinct phosphorylation patterns induced by a large variety of selective DOP agonist and opioids, both in vitro and in vivo. Moreover, we detected hierarchical and temporally controlled receptor multisite phosphorylation and dephosphorylation. Appropriate signaling by GPCRs is dependent on the specific activation of canonical regulatory kinases and phosphatases that together create the overall signaling output. Agonist binding to the receptor triggers activation and signaling through its associated heterotrimeric G protein which involves GRKs or second messenger-dependent protein kinases, such as PKC or PKA. GRK-imprinted phosphorylation barcodes increase the affinity for β-arrestin binding, which uncouple the receptor from its G protein, regulate receptor internalization and subsequent desensitization while simultaneously initiating β-arrestin-dependent signaling. Internalized GPCRs are either sorted to lysosomes for degradation or recycle back to the plasma membrane for reinsertion. Return of GPCRs to their resting state requires dissociation of agonist and β-arrestins but also dephosphorylation of the receptor.
Agonist-induced phosphorylation usually involves a specific pattern of serine and threonine residues and is considered the first step in receptor regulation. Receptor phosphorylation was studied in great detail for several GPCRs, most notably the β2-adrenoceptor (β2-AR), MOP and NOP receptor as well as somatostatin sst2, sst3 and sst5 receptor subtypes [71][72][73]92,93,[115][116][117][118][119][120][121][122][123][124] . A primary phosphate-acceptor site has been identified for each GPCR, and phosphorylation often correlates with receptor internalization levels. The prevailing hypothesis that complete GPCR phosphorylation is required for maximal internalization is supported by these observations and a hallmark of full agonists. In case of DOP receptor, mutagenesis studies have localized multiple serine and threonine residues located in the carboxyl-terminal domain as major phosphorylation sites [74][75][76] . Using phosphosite-specific antibodies, we found that DOP receptor phosphorylation proceeds in a temporal hierarchy, with S363 as primary phosphorylation site followed by T361 phosphorylation. Stimulation with DPDPE induced maximal receptor internalization with a similar kinetic as seen for MOP receptor 72,121 . Mutation of T358, T361 and S363 has no influence on the capacity to recruit β-arrestins 84 but reduced receptor internalization which was abolished when all serine and threonine residues in the C-terminal tail were mutated to alanine. Earlier studies have shown, that T358 and S363 are substrates for GRK2 74,78 . The data presented here are in line with these previous findings and also demonstrate that T361 is phosphorylated by GRK2/3 after DPDPE stimulation in vitro. In addition, DOP receptor can also undergo heterologous site-specific PKC-dependent phosphorylation at S344 85 . Accordingly, we did not observe any impact on T361 or S363 phosphorylation level either after PKC-or PKA-activation. Subsequently, cells were washed three times with ice cold citrate buffer and then incubated in serum-free medium in the absence of DPDPE for 0, 20 or 40 min in the presence of the above indicated concentrations of (A) calyculin A or (B) okadaic acid at 37 °C. Lysates were immunoblotted with antibody to pT361 or pS363. Blots were stripped and reprobed with the anti-HA antibody. Densitometry readings, above the blots, were normalized to those in DPDPE-stimulated cells (0 nM and 0 min), which were set to 100%. Data are mean ± SEM from three independent experiments. *p < 0.05 vs. DPDPE controls by one-way ANOVA with Bonferroni's post-hoc test. (2020) 10:8585 | https://doi.org/10.1038/s41598-020-65589-7 www.nature.com/scientificreports www.nature.com/scientificreports/ Like MOP receptor agonists [71][72][73] , most DOP receptor agonists stimulated receptor phosphorylation to a degree that correlated with the level of internalization. The enkephalin analogs DPDPE and DADLE are highly selective DOP receptor agonists 125,126 . Both compounds induced robust activation of GIRK channels, phosphorylation and receptor internalization equally to the natural DOP receptor agonists, the enkephalins and deltorphins. SNC80 is a full agonist in G protein-dependent cAMP assays with high DOP receptor selectivity and antinociceptive, antidepressant and anxiolytic properties, but also producing convulsions 36,37,40,55,[127][128][129][130] . We found that SNC80 also induced robust GIRK channel activation and showed strong DOP receptor phosphorylation. Our data confirmed previous studies which also showed DOP internalization after SNC80 exposure 79,128,130 . The SNC80 derivative AR-M1000390 and the spirocyclic agonist ADL5859 were previously described as potent, highly selective and orally available DOP receptor agonists 41,131 . Both compounds reduced inflammatory and neuropathic pain and were devoid of proconvulsive activity 79,95,131,132 . We found that ADL5859 and AR-M1000390 induced GIRK channel activation with similar potency and produced significant DOP receptor phosphorylation and internalization at saturating concentrations in vitro. However, in DOP-eGFP knock-in mice both compounds were not able to induce DOP receptor internalization in vivo 79,95,102 . It should be mentioned that we used unphysiologically high concentrations of both compounds to see whether they can elicit any DOP receptor phosphorylation or internalization. Therefore, our in vitro observations are not comparable to the in vivo situation. However, it was at least demonstrated for ADL5859 and AR-M1000390 that substantial DOP receptor internalization can occur at very high concentrations in vitro 79 . In contrast, fentanyl and (−)-methadone failed to induce any robust activation of GIRK channels but produced phosphorylation at DOP receptors only at S363, but not at T361, followed by weak receptor internalization. Morphine failed to induce any phosphorylation at DOP receptor and induced only a weak activation in G protein-mediated GIRK assays. Previous studies had shown that DOP receptor was not internalized after morphine exposure, which is consistent with the present findings 81,133 . Glucuronate conjugation at the 6-position in morphine is known to enhance DOP receptor binding 134 . Conversely, we found DOP  panels in A,B). Densitometry readings, above the blots, were normalized to those in SCR-transfected cells, which were set to 100%. Data are mean ± SEM from three to four independent experiments. *p < 0.05 vs. SCR by one-way ANOVA with Bonferroni's post-hoc test. (2020) 10:8585 | https://doi.org/10.1038/s41598-020-65589-7 www.nature.com/scientificreports www.nature.com/scientificreports/ receptor phosphorylation after morphine-6-glucuronide exposure followed by weak receptor internalization. Buprenorphine has a high binding affinity to DOP receptor 135 but no receptor phosphorylation and internalization was detectable in our hands, whereas norbuprenorphine, the major active metabolite of buprenorphine, induced DOP receptor phosphorylation and receptor internalization. Together our data show a strong correlation between GIRK channel activation, phosphorylation and DOP receptor internalization (Fig. 4D). These correlation is also supported by a previous study which showed, that DOP receptor stimulation, GIRK channel undergo arrestin-dependent internalization 136 .
Localization of DOP receptor has been studied using electron microscopy, immunohistochemical detection and in situ hybridization [19][20][21][137][138][139][140][141] . Here, fine-tuning of in vivo phosphorylation of DOP receptor in brain was analyzed for the first time using Western blot. The same techniques were used in recent studies to analyze phosphorylation at MOP and NOP receptors 72,93,120,121 . Both residues, T361 and S363, are phosphorylated in vivo in a dose-dependent manner after agonist injection. Interestingly, T361 is also constitutively phosphorylated which may reflect high constitutive activity of DOP receptors that had been reported before 104 . In comparison to animals injected with saline, in vivo phosphorylation in mouse brains was only weakly increased by ADL5859 and AR-M1000390, which may explain the lack of internalization as well as the absence of proconvulsive activity observed previously 129 .
So far, the molecular mechanisms of DOP receptor dephosphorylation have never been investigated. Here we used phosphosite-specific antibodies in combination with siRNA knock-down screening to identify phosphatases involved in DOP receptor dephosphorylation. We identified PP1α and PP1β as phosphatases which catalyzed T361 and S363 dephosphorylation after agonist removal. Inhibition of PP1α or PP1β expression resulted in increased DPDPE-driven receptor phosphorylation at both residues. Dephosphorylation of T361 occurs slower than dephosphorylation at S363. In comparison, dephosphorylation of DOP receptor occurs at a much slower rate than that observed for MOP receptor and sst5 72,124 . Dephosphorylation of MOP receptor and sst5 involves PP1γ, dephosphorylation of sst2 requires PP1β activity and dephosphorylation of sst3 involves PP1α and PP1β 72,92,124,142 . Our findings indicate that PP1γ mediated dephosphorylation is linked to GPCRs that recycle rapidly to the plasma membrane. PP1 consists of one catalytic subunit and one or more regulatory subunits. The substrate specificity and subcellular localization of PP1s is determined by the regulatory subunit, of which more than 40 exist 106 . However, it remains unclear which mechanisms regulate phosphatase specificity at GPCRs. It is conceivable that carboxyl-terminal phosphorylation sites, specific sequences in the intracellular loops or β-arrestin trafficking patterns may all contribute to selection of phosphatases.
In conclusion, we identified for the first time a specific and hierarchical agonist-induced phosphorylation pattern in the carboxyl-terminal domain of DOP receptor in vitro and in vivo. The phosphorylation pattern correlates with receptor internalization and conceivably provides evidence for a general biochemical mechanism by which the different functional effects of DOP receptor agonists are achieved. Further, differential agonist-induced multi-site phosphorylation patterns suggest that chemically diverse agonists induce distinct receptor conformations that could explain differences in DOP receptor agonist efficacy. This study provides important tools to characterize agonist-dependent regulation of DOP receptor signaling at the cellular and intact animal leyel, which will facilitate the development of DOP receptor agonists for selected therapeutic indications.

Data availability
The datasets in the current study are available from the corresponding author on reasonable request.