Abstract
GPR88 is an orphan G-protein–coupled receptor (GPCR) highly expressed in striatal medium spiny neurons (MSN), also found in cortical neurons at low level. In MSN, GPR88 has a canonical GPCR plasma membrane/cytoplasmic expression, whereas in cortical neurons, we previously reported an atypical intranuclear localization. Molecular size analysis suggests that GPR88, expressed in plasma membrane of MSN or in nuclear compartment of cortical neurons, corresponds to the full-length protein. By transfection of cortical neurons, we showed that GPR88 fluorescent chimeras exhibit a nuclear localization. This localization is contingent on the third intracytoplasmic loop and C-terminus domains, even though these domains do not contain any known nuclear localization signals (NLS). Using yeast two-hybrid screening with these domains, we identified the nuclear proteins ATRX, TOP2B, and BAZ2B, all involved in chromatin remodeling, as potential protein partners of GPR88. We also validated the interaction of GPR88 with these nuclear proteins by proximity ligation assay on cortical neurons in culture and coimmunoprecipitation experiments on cortical extracts from GPR88 wild-type (WT) and knockout (KO) mice. The identification of GPR88 subcellular partners may provide novel functional insights for nonclassical modes of GPCR action that could be relevant in the maturating process of neocortical neurons.
Introduction
According to the classical view, G-protein–coupled receptors (GPCR) transduce extracellular signals by binding hormones and neurotransmitters at the plasma membrane outer face, and G proteins at the inner face, forming complexes that dissociate to activate intracellular pathways and generate rapid nongenomic signals. Recent evidence has opened up exciting views upon new modes of action for several GPCRs, related so far to genomic modulations. Some receptors were described as inserted into the nuclear membrane and efficiently coupled with several nucleoplasmic signaling factors (O’Malley et al. 2003; Sergin et al. 2017). Receptor intranuclear translocations were also observed upon ligand binding from an extracellular source or by endogenously produced nonsecreted ligands (Jong et al. 2005; Gobeil et al. 2006; Kumar et al. 2008).
GPR88 is a neuronal orphan G-protein–coupled receptor preferentially expressed in synaptic sites of striatal GABAergic MSNs (Massart et al. 2009). This indicates a potential role for GPR88 in MSNs’ neurotransmission, which plays a central role in psychomotor functions subserved by the basal ganglia. Indeed, previous studies reported variations of GPR88 expression in cortical and limbic areas following treatments with various antidepressants, mood regulators, or addictive drugs (Ogden et al. 2004; Brandish et al. 2005; Böhm et al. 2006; Conti et al. 2007). Moreover, chronic stress paradigms and anxiety-/depression-like behaviors are associated with a down-regulation of GPR88 expression (Kim and Han 2006). Recent data from connectivity studies in GPR88-deficient mice also revealed extensive remodeling of intracortical and corticosubcortical networks (Arefin et al. 2017). These and other preclinical studies indirectly link GPR88 to psychiatric disorders (Ingallinesi et al. 2015; Komatsu 2015). Consistent with the view that GPR88 activity has a serious impact on sensorimotor and cognitive functions, homozygous deleterious mutation of GPR88 in humans has been linked to manifestation of speech delay, learning disabilities, and chorea in children (Alkufri et al. 2016). Human genetic studies have also shown an association between GPR88 and bipolar disorder and major psychoses (Del Zompo et al. 2014). Overall, these studies suggest that GPR88 could constitute an interesting and potential target in mental illness.
We previously reported a particular intranuclear localization of GPR88 in most neocortical neurons in the rat, mice, monkey, and human brains (Massart et al. 2016). Data by Ehrlich et al. confirmed this observation, as they published a confocal photomicrograph showing that their Venus-GPR88 chimera exhibited a nuclear localization (see supplementary figure 6 in Ehrlich et al. 2018), even though they did not comment this fact. This morphological data confirmed the nuclear distribution of GPR88 by a completely different molecular approach. GPR88 intranuclear localization appears in rats around birth, more precisely during the cortical lamination, and remains until adulthood with the same expression pattern. Just before the intranuclear presence of GPR88 in cortical neurons settling in its laminar position, GPR88 displayed a plasma membrane/cytoplasmic localization in the undifferentiated cortical plate neurons. This observation suggests that the nuclear presence of GPR88 results from dynamic molecular adaptations in maturating cortical neurons. Considering the absence of a known nuclear localization signal (NLS) in GPR88 sequence, it is highly probable that several protein partners expressed in differentiating neocortical neuron are responsible for its intranuclear localization and/or function. Therefore, an in-depth analysis of GPR88 expression in the cerebral cortex should provide a better understanding of its involvement in cerebral function.
In the present report, we used the mouse GPR88 model to investigate potential nuclear protein partners of GPR88. We first confirmed the intranuclear distribution of GPR88 in the mouse cortex using electron microscopy. Next, we demonstrated that GPR88 nuclear localization depends on its third intracellular loop (IL3) and C-terminus domains. We finally identified three potential nuclear protein partners, by yeast two-hybrid screening, whose interaction with GPR88 in the nuclei of cortical neurons was further characterized by proximity ligation assay and coimmunoprecipitation. Considering that these protein partners, ATRX, TOP2B, and BAZ2B are known to be implicated in chromatin remodeling, their interaction with GPR88 might be essential to elucidate its putative nuclear function in mouse cortical neurons.
Materials and Methods
Animals
Generation of GPR88−/− mice was previously described (Meirsman et al. 2016), where GPR88 KO mice results from the deletion of GPR88 exon 2. Wild-type and knockout mice displayed a hybrid 50% C57BL/6J–50% 129Sv genetic background.
Animals were maintained on a 12 h light/dark cycle at controlled temperature (22 ± 1 °C). Food and water were available ad libitum throughout all experiments. All experimental procedures were performed in strict accordance with the European Communities Council Directive of 24 November 1986 (89/609/EEC) and were approved by a local Ethical Committee for animal care and use of laboratory animals. All efforts were made to minimize animal suffering and to reduce the number of animals used in these experiments.
Brain Sections
GPR88-KO mice (9–12 weeks) were anesthetized deeply with Dolethal (182,2 mg of pentobarbital/kg). Mice were perfused transcardially with 50 mL of saline solution (0.9% NaCl warmed at 37 °C), followed by 100 mL of a fixative solution containing 2–3% of paraformaldehyde for immunohistofluorescence or 3% of paraformaldehyde and 0.25% of glutaraldehyde for electron microscopy, in phosphate buffer 0.1 M, pH 7.4 (PB). Brains were then removed and post fixed at room temperature for 1.5 h in the same fixative solution. After two rinses with PB, brains were cut in the coronal plan with a vibratome (VT1000, Leica) to obtain serial free-floating sections of 40 μm, for immunohistofluorescence, or 70 μm thick, for electron microscopy.
Free-floating brain sections for immunoelectron microscopy were soaked in 30% sucrose dissolved in PB. Sucrose solution was renewed every 2 h at least four times. Finally, brain sections were frozen in the same solution and stored at −80 °C until use.
Immunoelectron Microscopy
On the day of the immunostaining protocol, frozen brain sections containing prefrontal or somatosensory cortices from 12 weeks old GPR88 WT (n = 4) and KO mice (n = 3) were simultaneously treated with several cycles of freeze/thaw, in order to make the neuronal membrane permeable to antibody. Subsequently, the free-floating brain sections were washed with decreasing sucrose concentrations to bring them in Tris-buffered saline 50 mM pH 7.4 (TBS). Brain sections were incubated for 1 h at room temperature in TBS containing 1% bovine serum albumin (BSA) and 3% normal donkey serum. Sections were incubated either with the homemade GPR88 primary antibody (1:15 000) for 60 h or with a biotinylated form of this GPR88 primary antibody (1:200) for 72 h. GPR88-purified IgGs were biotinylated using freshly prepared EZ-Link Sulfo-NHS-Biotin (Pierce) with a 20-fold molar excess. The reaction was stopped by addition of 0.1 M glycine followed by multiple dialysis. Then IgGs were concentrated using centrifugation on Microcon YM-50. Both primary antibodies were dissolved in TBS containing 1% BSA and 0.05% Tween20. Sections treated with the nonbiotinylated GPR88 primary antibody were rinsed 3 × 10 min in TBS with 0.05% Tween20 (TBST) before an overnight incubation with the biotinylated donkey antirabbit secondary antibody (1:300; Jackson ImmunoResearch). Following three washes with TBST (10 min each), all sections were incubated at room temperature for 1.5 h with the avidin–biotin–peroxidase complex (Vectastain Elite-ABC kit, Vector Laboratories). The immunoperoxidase staining was revealed by incubating the sections in 50 mM Tris buffer (pH 7.6) containing 0.05% 3,3-diamino-benzidine tetrahydrochloride (Sigma) and 0.01% hydrogen peroxide. The immunoperoxidase-stain intensity was monitored under a microscope and stopped when sections from KO mice still displayed a complete absence of background staining while the WT sections displayed a weak but distinct staining in the neocortex. Immunoperoxidase-stained sections were post fixed for 1 h with 2.5% glutaraldehyde in 0.1% cacodylate buffer and then for 15 min with 2% osmium tetroxide in 0.1 M phosphate buffer, before being dehydrated and flat-embedded in Epon 812.
Small squares (about 1–2 mm) were taken in the prefrontal and the somatosensory cortices from the flat-embedded sections. They were glued on top of the prepolymerized Epon blocks, in order to first make semithin resin sections (0.5 μm) for light microscopic observations, and then ultrathin sections for transmission electron microscopy examinations.
Mouse Primary Culture Neurons
Primary neocortical neurons were prepared from embryonic day 16 GPR88 WT and KO mice. Neurons were plated on 14 mm-coverslips placed in 60 mm-dishes (8 per dish) at a density of 30 000 cells/cm2, in neurobasal medium supplemented with B-27 serum-free supplement, Glutamax supplement, and penicillin/streptomycin (Thermo Fisher Scientific). Cells were maintained at 37 °C in a humidified atmosphere containing 5% CO2 for 14 days (DIV 14).
Primary Culture Neuron Transfection
Peptides corresponding to the carboxy-terminal tail (Cter; amino acids 331–384) and to the third intracellular loop (IL3; amino acids 216–285) of the human GPR88 were subcloned into pEYFPN1 (BD Biosciences), upstream of the eYFP protein and downstream of a Kozak sequence and an ATG initiation codon. Plasmids encoding Cter-YFP and IL3-YFP were transfected into mouse primary cortical neurons at DIV7 (2 μg/well) with Lipofectamine 2000 (Thermo Fisher Scientific) for 48 h, according to the manufacturer’s instructions.
Yeast Two-Hybrid Screening
Yeast two-hybrid screening was performed on a mouse brain cDNA library using the intracellular third loop and carboxy-terminus of GPR88, fused together, as bait (Hybrigenics, France). Polymerase chain reaction (PCR) was performed to generate a GPR88-GAL4 or GPR88-LexA DNA-binding domain fusion construct containing intracellular third loop (aa217–285) fused to the carboxy-terminal tail (aa332–384) of mouse GPR88, in pB66 or pB27 vector (Hybrigenics). Sixty million interactions were screened with pB66 and 63 million with pB27 vector.
For complementation experiments in yeast, bait and prey constructs were transformed into the haploid yeast strains CG1945 and YHGX13 (respectively) or cotransformed into the diploid yeast strain TOTO (Hybrigenics, France), using the lithium acetate method. Haploid yeast strains were cultured at 30 °C for 3–5 days on YPD-plates lacking tryptophan (bait) or leucine (prey). For diploid yeast strain, we used YPD-plates lacking both tryptophan and leucine.
Yeast mating was performed via overnight incubation of both haploid yeast strains in 0.25 mL of YPD medium with shaking. The resulting mixture was then spotted on plates containing leucine- and tryptophan-deficient medium and was cultured for 3–5 days at 30 °C. For diploid yeast strains, resulting colonies after cotransformation were streaked onto a new plate (replica) and grown for two more days at 30 °C. Specific interactions between bait and prey constructs were verified either by the growth of yeast colonies in leucine-, tryptophan-, and histidine-free medium after 3–5 days at 30 °C or by a β-galactosidase colony-lift filter assay.
The interacting partners were classified by Hybrigenics based on the confidence of interaction into six different classes (A–F) with A indicating high confidence, D low confidence, and F for experimentally proven artifacts.
Double Immunofluorescent Labeling and Confocal Microscopy
Cortical and striatal primary cultures at DIV14 were fixed in 2–3% paraformaldehyde containing 4% sucrose for 10 min at room temperature and washed with PBS. Neurons were blocked for 1 h at room temperature with TBS containing 0.1% Tween20 (T20), 5% normal donkey serum, 0.5% BSA, and 0.1% gelatin before an overnight incubation at 4 °C with primary antibodies diluted in TBS-T20 and 0.5% BSA. Double immunostainings were performed with the following primary antibodies: a homemade GPR88 rabbit polyclonal antibody (1:20 000; Massart et al. 2009) and either TOP2B mouse monoclonal antibody (1:250; 681417v, R&D Systems) or ATRX mouse monoclonal antibody (1:500; sc-55 584, Santacruz Biotechnology) or BAZ2B mouse monoclonal antibody (1:200; H029994-MO7, Abnova).
After primary antibody incubations, cultures were washed 3 × 10 min in TBS-T20 containing 0.1% gelatin and then incubated 1 h at room temperature with a CY3-donkey antirabbit antibody (Jackson ImmunoResearch, Thermo Fisher Scientific) and an AlexaFluor 488-donkey antimouse (Thermo Fisher Scientific), both diluted at 1:250 in TBS-T20. Following secondary antibody incubations, neurons were washed in TBS-T20 containing 0.1% gelatin. Nuclei were stained with Hoechst solution or TO-PRO-3. Finally, coverslips were mounted in Fluoromont-G (SouthernBiotech) on SuperFrost-Plus slides (Menzel-Glaser). The fluorescent immunostaining was examined with a Zeiss photomicroscope equipped with a Sony CCD camera. Some immunofluorescence images were acquired using a LEICA TCS SP5 confocal laser scanning microscope, PL APO 63× oil immersion objective. Software used for image processing was Image J.
Proximity Ligation Assay (PLA)
Protein interactions in primary cultures (DIV14) were detected using the Duolink in situ PLA detection kit (Sigma Aldrich) according to the supplier instructions. Neurons were incubated in a blocking solution for 1 h at room temperature and then with the primary anti-GPR88 (homemade) and ATRX, TOP2B, or BAZ2B antibodies 1.5 h at room temperature. On the following day, samples were washed and incubated with a PLA probe solution for 1 h at room temperature. Neurons were washed and incubated with the ligation solution for 30 min at 37 °C and then with the amplification solution at 37 °C for 100 min. Finally, cell nuclei were counterstained with Hoechst 33258 solution (Sigma-Aldrich), and sections were mounted using the Fluoromont-G (SouthernBiotech).
Coimmunoprecipitation (Co-IP) Experiments and Western Blot Analyses
Fresh nonfixed cortex (9–12 weeks) was dissected from WT and KO mice. We obtained nuclear fractions following the instruction of the nuclear complex Co-IP Kit (Active Motif). This kit allowed us to modulate and optimize the Co-IP conditions for each protein partner. For each Co-IP, 2 mg of nuclear extract, quantified using a BCA protein assay kit (Pierce), was incubated overnight at 4 °C in 500 μL IP solution with 45 μL detergent from the kit, protease inhibitors, DTT (1 mM), and the appropriate antibody (1 μg). In our optimized conditions, we used the low IP buffer for BAZ2B and TOP2B, and the high IP buffer with 7.5 μL NaCl 5 M for ATRX. The next day, immunoprecipitated extracts were incubated at 4 °C for 3 h with 15 μL of protein-G magnetic beads (Pierce), prewashed in the defined IP buffer for each Co-IP. Co-IPs were then washed 3 times with their respective IP buffer supplemented with 0.1% BSA and finally with the IP buffer alone. Finally, each Co-IP was eluted with 40 uL of Laemmli 4X + DTT (0.1 M) and heated 15 min at 56 °C. Proteins were size fractioned on a 5–20% polyacrylamide gels and electrophoretically transferred to a PVDF membrane. GPR88, ATRX, TOP2B, and BAZ2B were detected using their respective primary antibody (our rabbit biotinylated anti-GPR88 at 1:500; mouse monoclonal anti-TOP2B at 1:250 [681417v, R&D systems]); rabbit polyclonal anti-ATRX at 1:500 (NBP1–32851, Novus biological); and rabbit polyclonal anti-BAZ2B at1:200 (RO189-2a, Abiocode). We then used HRP-conjugated secondary antibodies, followed by enhanced chemiluminescence (Clarity Western ECL Substrates, Biorad). Membranes were then scanned on a ChemidocMP (Biorad).
Statistics
Data were analyzed with Student’s t-test using GraphPad Prism 6.0 Software. Differences were considered significant at P < 0.05. Values are presented as mean ± standard error of the mean (SEM).
Results
Intranuclear GPR88 Localization in Mice Cortex
We previously established the subcellular intranuclear localization of GPR88 in rat cortical neurons using a specific rabbit polyclonal antibody raised against the GPR88 carboxy-terminal part (Cter; Massart et al. 2009, 2016). In the current study, we validated the precise intranuclear distribution of GPR88 in mice cortical neurons using immunoelectron microscopy analysis in GPR88 WT and control GPR88 KO mice (12 weeks). For this purpose, we used the biotinylated form of the primary polyclonal anti-GPR88 antibody with an amplification step to increase signal intensity and achieve specific staining with minimal background. As expected, examination of immunostained brain sections revealed the absence of immunoperoxidase signal in the GPR88-KO, contrasting with the light but distinct labeling detected in the neocortex of WT mice (Supplementary Fig. 1). This low intensity for GPR88 labeling in cortex is consistent with the low level of GPR88 expression previously reported in the cortex of rodents (Massart et al. 2009; Ehrlich et al. 2018). Analysis of ultrathin section with transmission electron microscopy also demonstrated electron-dense peroxidase labeling that appears as numerous small dark dots barely visible scattered throughout the cell nucleus and in some membrane portions of the nuclear envelope of numerous cortical neurons (Fig. 1). In contrast, no immunoperoxidase staining was observed in KO cortical neurons (Fig. 1). Some unlabeled nuclei, presumably from glial cell, were also seen in the proximity of immuno-positive neuronal nuclei in WT mice (not shown), which is consistent with our previously reported immunoelectron study in the rat brain (Massart et al. 2009). Immunoreactive dendritic spines with discrete labeling in the postsynaptic density and the spine apparatus were also detected in the neocortical neuropil (Supplementary Fig. 2A,B). In addition, some perinuclear organelles including the rough endoplasmic reticulum, the Golgi apparatus, and vesicles were observed with immunolabeling (Supplementary Fig. 2C,D), indicating GPR88 expression in both perinuclear cytoplasmic and intranuclear compartments. These ultrastructural findings are consistent with previously reported subcellular GPR88 expression (Massart et al. 2016; Ehrlich et al. 2018).
Electron microscopic immunodetection of GPR88 labeling in cortical neurons of WT mice. (A) The immunoperoxidase labeling of GPR88 appears as little electron-dense dots throughout the nucleus and near the inner nuclear membrane in WT cortical neurons. Circles delimit part of the nuclear immunolabeling. Heterochromatin appears as larger and irregular dense clusters (white asterisks). (B) No GPR88 immunoperoxidase labeling is observed in KO mouse cortical neurons. (GC) Glial cell.
GPR88 Nuclear Localization Depends on Cter and IL3 Domains
Following the ultrastructural localization in mice cortical neurons, we investigated which portion of GPR88 was involved in its nuclear localization. The GPR88-predicted structure comprises seven transmembrane domains, the fifth and sixth being separated by the long intracellular third loop (IL3). Nuclear localization sequences (NLS) have been identified in the carboxyl-terminal region or the third loop of various GPCR (Lee et al. 2004), but for the GPR88, no mono- or bipartite basic NLS motif has been discovered yet. However, we previously demonstrated the nuclear localization of a chimeric YFP-GR88 protein, when transfected in cortical organotypic cultures (Massart et al. 2016).
Therefore, we performed transfections of mice cortical primary neurons (DIV7) with either the C-terminus domain (44 last residues, 331–384, Cter-YFP) or the putative intracellular third loop (70 residues, 216–285, IL3-YFP) fused to YFP, and compared with YFP alone. YFP alone exhibited essentially a cytoplasmic localization extending from the soma to the distal part of neurites and a faint nuclear labeling (Fig. 2A). When we fused the IL3 domain to YFP, the signal was more restricted to the soma and the nucleus (Fig. 2B), whereas the Cter-YFP was exclusively expressed in the nucleus, similar to the labeling of GPR88 with our homemade antibody (Fig. 2C).
Transfection of mouse cortical neurons with YFP, IL3-YFP, or Cter-YFP chimeras. Transfection with plasmids containing (A) YFP alone, (B) IL3-YFP, or (C) Cter-YFP. YFP alone exhibited essentially an overall cytoplasmic localization, while chimeras displayed a perinuclear and nuclear distribution for IL3-YFP and an exclusive nuclear localization for Cter-YFP. Scale bars = 10 μm.
Thus, these experiments suggested that nuclear localization of GPR88 depends on both the C-terminus and IL3 domains. To explore this hypothesis, we performed a two-hybrid screening to identify potential partner proteins in interaction with these domains of GPR88.
Identification of GPR88-Interacting Partners
We used the yeast two-hybrid system to screen a mouse brain library using the entire intracellular third loop of GRP88 (70AA) fused to the carboxy-terminal tail (44AA), as a bait.
From 123 million interactions that were screened, 27 positive clones were found. Unfortunately, we did not identify any potential trafficking and/or targeting protein. Therefore, we focused on the three nuclear proteins with good confidence in the interaction, namely: ATRX, TOP2B, and BAZ2B (Gene ID: 22589, 407 823, and 21 974, respectively). In particular, ATRX was found in both Gal4 and LexA screening, the latest being more stringent. Considering the involvement of these proteins in mechanisms of chromatin remodeling, we made the hypothesis that they could play a role in the nuclear function of GPR88.
To rule out the possibility of a false-positive interaction, we tested several control constructs in the yeast two-hybrid system. Only transformants bearing the bait plasmid pB66-GPR88 and the prey plasmids pP6-Atrx, pP6-Top2b, or pP6-Baz2b were positive for the β-galactosidase and the histidine phenotype selection. These results thus suggest that ATRX, TOP2B, and BAZ2B are direct partner of the GPR88.
Colocalization of GPR88 with ATRX, TOP2B, and BAZ2B
To interact with each other, proteins must colocalize within the same cell. We investigated whether GPR88 colocalizes with the partner proteins identified in the two-hybrid screening using both immunocytochemistry (IHC) on cortical and striatal primary neuron cultures and IHC on brain sections of GPR88 WT and KO mice (9–12 weeks).
GPR88 labeling is consistent with previously reported IHC results (Massart et al. 2009, 2016), with a membrane/cytoplasmic localization in the striatal neurons (Supplementary Fig. 3) and a main nuclear expression pattern in cortical neurons (Fig. 3A–L). The immunostaining of TOP2B or ATRX is visible throughout the nucleoplasm and highly concentrated in spots of heterochromatin (Fig. 3G,H,J,K), without any labeling in the cytoplasm, as reported in the literature (Cowell et al. 2011; Clynes et al. 2013). Regarding BAZ2B, the labeling is condensed in the nucleus with spots of heterochromatin and a faint labeling of the cytoplasm (Fig. 3I,L). Since no reported data for BAZ2B immunostaining are available, we cannot conclude reasonably whether the cytoplasmic signal generated by the antibody is specific.
Colocalization between GPR88 and ATRX, TOP2B, or BAZ2B in cortical neurons. (A–F) Double immunostaining of GPR88 (left panels, in red) and its nuclear partners (middle panels, in green) in primary cultures of cortical neurons derived from GPR88-WT (A–C) and -KO (D–F) mice. Merge images (right panels) show a clear colocalization, in the nucleus of wild-type cortical neurons, between GPR88 and ATRX (A), TOP2B (B) or BAZ2B (C). (G–L) Double immunostaining of cortical neurons in brain slices from GPR88-WT (G–I) and -KO (J–L) mice. Colocalization between GPR88 (left panels, in red) and ATRX, TOP2B, or BAZ2B (middle panels, in green) is specifically observed in brain cortical neurons of WT mice (G–I; right panels) compared with KO mice (J–L; right panels). Scale bars = 10 μm
We showed a clear colocalization in the nucleus for GPR88 with ATRX, TOP2B, or BAZ2B on WT cortical neurons in primary cultures, which is specific compared with KO mice cultures, where there is no GPR88 labeling (Fig. 3A–F). Colocalization is also specifically observed on brain cortical sections of WT mice compared with KO (Fig. 3G–L). Colocalization seems to be predominant outside heterochromatin spots, which is in agreement with previous results showing a colocalization of GPR88 with euchromatin markers but not with heterochromatin markers (Massart et al. 2016).
In contrast, GPR88 does not exhibit double labeling with any of these three proteins in striatal neuron cultures or in striatal neurons of WT mice (Supplementary Fig. 3A–L). This lack of double labeling in the striatum, where GPR88 has a typical cytoplasmic membrane distribution, confirmed the specificity of the colocalization of GPR88 with these three nuclear partners in cortical neurons.
GPR88 Interacts with Nuclear Proteins ATRX, TOP2B, and BAZ2B
To demonstrate that GPR88 and each of these nuclear partners belong to the same protein complex (two-by-two), we used proximity ligation assay (PLA) to show the physical proximity of GPR88 with ATRX, TOP2B, or BAZ2B.
We performed PLA for each interaction on WT and KO cortical cultures (Fig. 4A1–A3, left panels). Quantification of the number of dots per cortical cell nucleus on 100 of neurons for each putative partner confirmed the specific interaction of GPR88 with ATRX, with TOP2B, or with BAZ2B on primary cultures of cortical neurons from WT mice compared with KO (Fig. 4A, right panels). The PLA signal detected for each partner is not very abundant; nevertheless, it is significantly higher in WT compared with KO, and reproducible in three different cultures. Moreover, we were able to establish its specificity through controls such as the use of primary cultures of striatal WT and KO neurons (Supplementary Fig. 4), where the expression of GPR88 remains confined to the cell surface. In this control experiment, quantification of dots (not shown) demonstrated that the signals of the three controls (cortex KO, striatum WT, and KO) were comparable and thus defined the level of the background of the PLA.
Interaction between GPR88 and ATRX, TOP2B, or BAZ2B in cortical neurons. (A) PLA signal (red dots) indicates interaction between GPR88 and ATRX (A1), TOP2B (A2) or BAZ2B (A3) in GPR88-WT and -KO mice primary cultures of cortical neurons. Histograms represent the quantification of the number of red dots per nuclei from three different cultures and about 100 neurons for each interaction. Nuclei were stained with TO-PRO3. (B) Coimmunoprecipitation of endogenous GPR88 and ATRX, TOP2B or BAZ2B from fresh unfixed nuclear extract of brain cortical tissue from GPR88-WT and -KO mice, analyzed by western blot and revealed with chemiluminescence. Input: 50 μg; IP: 1 mg.
Finally, to consolidate our results, we performed coimmunoprecipitations (Co-IP) of native GPR88 with ATRX, TOP2B, or BAZ2B. First, we were able to detect protein partners and GPR88 from nuclear extracts (input 50 μg; Fig. 4B). Then, we validated the immunoprecipitation of GPR88 specifically from WT nuclear cortical extracts without any signal in KO extracts (Fig. 4B). Finally, we showed that ATRX, TOP2B, and BAZ2B coimmunoprecipitated each specifically with GPR88 in WT nuclear cortical extracts (1 mg; Fig. 4B). These signals showed that the interactions were preserved in nuclear extracts using our optimized conditions. As a control, partners were not immunoprecipitated in KO nuclear cortical extracts from fresh brain tissue (Fig. 4B). As in the PLA, the signal intensity of the Co-IP of the partners is weak but distinct from the KO and reproducible in three independent experiments.
In conclusion, our results, altogether, demonstrated that GPR88 colocalizes and directly interacts with ATRX, TOP2B, and BAZ2B, three nuclear proteins involved in chromatin remodeling.
Discussion
In the present study, we first verified by electron microscopy the ultrastructural GPR88 distribution, previously reported in the rat cortex (Massart et al. 2009, 2016), using GPR88-WT and control-KO mice. Our study demonstrated an identical GPR88 distribution to that previously found in the rat cortex. GPR88 immunosignal is principally distributed throughout the nucleoplasm and the nuclear membrane segments in cortical neurons of wild-type mice. This subcellular distribution pattern is in agreement with a recent study showing a very similar distribution of a chimeric GPR88-Venus protein in cortical slices from the GPR88-Venus knock-in mice (confocal microscopy illustration from the supplementary figure 6 of Ehrlich et al. 2018). Although not commented, the microphotograph illustrates GPR88-Venus fluorescence perfectly colocalized with the DAPI-stained nuclei, indicating that GPR88 is targeted in the nuclei of cortical neurons, while the same GPR88-Venus construct displays exclusive somatodendritic distribution in striatum. Thus, this result provides an additional confirmation of nuclear GPR88 addressing in cortical neurons by a completely different molecular approach. Our study also highlighted GPR88 immunolabeling on Golgi apparatus and RER mainly distributed in the perinuclear region, as well as in dendritic spines in the neuropile, in agreement with previous reported data (Massart et al. 2016; Ehrlich et al. 2018). Taken together, these observations support the inference that the endogenous GPR88 actually accumulates in the nuclei of differentiated neocortical neurons.
In previous studies, we identified GPR88 in the cell body of maturating neurons in primordial striatum and cortical plate during prenatal rat development and on somatodendritic synapses in striatal adult neurons (Massart et al. 2009; Ehrlich et al. 2018). In addition to this predicted synaptic expression pattern in striatum, we revealed an unexpected GPR88 nuclear expression in cortex, which begins in differentiating neurons, progressively integrating the deep layer V up to the superficial layer II during corticogenesis and remaining until adulthood (Massart et al. 2016). These GPR88 differential cytoplasmic/nuclear distribution patterns seem to be determined in each neuronal phenotype by specific genetic programs that are spatiotemporally regulated. Previous studies have shown targeting of GPR88 to the primary cilia in primary striatal neuron cultures and in early postnatal cortical neurons in mouse (Marley et al. 2013; Ehrlich et al. 2018). Different GPCRs have also been detected in primary cilia (Berbari et al. 2008; Badgandi et al. 2017) where their signaling activity may differ from the neuronal somatodendritic signaling (Marley et al. 2013). A role for primary cilia in neuronal migration during neocortical development, dendritic refinement, and synaptic integration of adult-born neurons has also been suggested (Kumamoto et al. 2012; Guemez-Gamboa et al. 2014; Sarkisian and Guadiana 2015). Therefore, GPR88 signaling via primary cilium in cortical plate neurons could modulate neuronal migration in the developing cortex and GPR88 deletion may lead to malformations of neocortical development resulting from altered migration. Interestingly in an ongoing study, we detected a high frequency of focal heterotopias, which are compact clusters composed of numerous neuronal and glial cells in the molecular layer of frontal and prefrontal cortex in GPR88-KO mice. Focal heterotopias in the molecular layer are indicative of an excessive neuronal migration (Ramos et al. 2008). Mouse strains with congenital autoimmunity exhibit neocortical heterotopia with striking similarity to those found in the dyslexic brain (Sherman et al. 1985, 1987, 1990). Supporting the link between deficits in cortical neuronal migration with cognitive disruption, the presence of heterotopias in these mice is associated with impaired performance on spatial and nonspatial memory tasks (Denenberg et al. 1991; Boehm et al. 1996; Balogh et al. 1998). Moreover, it might also be linked to the deficits in sensory processing (Clark et al. 2000; Frenkel et al. 2000) and probably the sensorimotor gating deficiency detected in GPR88-KO mice (Logue et al. 2009; Meirsman et al. 2016).
Although the precise mechanism responsible for GPR88 nuclear targeting in cortical neurons is unknown, translocation of several GPCRs into the cell nuclei and perinuclear regions after activation and internalization by extracellular treatment with their related agonists has been described (Lu et al. 1998; Joyal et al. 2014; Bhosle et al. 2016). NLS motifs in GPCR Cter and/or IL3 domains and nuclear importing protein mediators have been typically involved in the nuclear receptor translocation (Senbonmatsu et al. 2003; Lee et al. 2004). The absence of both endogenous ligand and known canonical NLS motifs for GPR88 suggest that at least one mediator protein is involved in its nuclear transport. Our results from transfection with fusion protein between IL3- or Cter-GPR88 domain and YFP demonstrated that putative signals in both domains participate in GPR88 nuclear targeting. Thus, a yeast two-hybrid screening was performed (Hybrigenics, France) using the entire IL3 domain fused to the Cter of GPR88 as a bait (Carrel et al. 2008). Even though no trafficking and/or targeting proteins were spotted, we identified putative GPR88 partners including nuclear proteins ATRX, TOP2B, and BAZ2B, which are expressed at high level and ubiquitously in the mouse brain. Using double immunohistochemistry labeling, we pointed out a significant colocalization between GPR88 and each of its partner proteins in the nucleus of cortical neurons but not in cell body and no colocalization in any neuronal compartments of the striatum. We demonstrated, by PLA experiments on cortical and striatal brain regions, that GPR88 interacts with each of the three nuclear proteins exclusively in the nuclei of cortical neurons. Finally, we confirmed that GPR88 can form, in vivo, a complex with ATRX, TOP2B, or BAZ2B in coimmunoprecipitation experiments with nuclear extracts from cortical tissue of GPR88-WT and -KO mice. As we did not detect evidences of GPR88 interaction in cortical neuronal cell bodies, we can hypothesize that interactions between ATRX, TOP2B, or BAZ2B and GPR88 take place exclusively in nuclear subcompartments in cortex. Furthermore, the observation that nuclear GPR88 targeting is not detected in the striatum despite the concomitant expression of GPR88 and these protein partners in this cerebral region supports this hypothesis. Together, these experiments are in favor of the hypothesis that intranuclear GPR88 addressing is dependent on a specific genetic program of cortical phenotype neurons, which is activated from rodent corticogenesis and remain until adulthood.
ATRX is a transcriptional regulator with a homeodomain-type zinc finger motif and a DNA-dependent ATPase domain of the sucrose-nonfermenting 2 family, suggesting that it is a chromatin-remodeling protein (Aasland et al. 1995; Eisen et al. 1995). In addition, this protein seems to be a critical mediator of corticogenesis and cell survival during early neuronal differentiation (Bérubé et al. 2005) and its mutation is responsible for an X-linked mental retardation, suggesting an essential role for ATRX in brain development (Gibbons et al. 1995). TOP2B is a topoisomerase involved in the topologic states of DNA during transcription in neurons migration and developing connectivity (Tiwari et al. 2012; Edmond et al. 2017). TOP2B-KO mice exhibit cerebral deficits and premature death of neurons (Lyu and Wang 2003; Xia et al. 2019). Functional loss of ATRX and TOP2B seems to be related, most of the time to cerebral deficits. BAZ2B is an integral component of chromatin remodeling complexes with a potential role in transcriptional activation (Ferguson et al. 2014). Chromatin remodeling exerts fundamental roles in developmental processes such as cortical cell fate determination, transcriptional control of cognitive development, or long-term programming of individual differences in behavioral responses, notably at the postnatal critical period (Hsieh and Gage 2004; Hong et al. 2005; Weaver 2007). Interestingly, we previously identified that GPR88 is associated with euchromatin (Massart et al. unpublished data) probably forming protein complexes with ATRX, TOP2B, or BAZ2B. Complexes of GPR88 and protein partners likely influence, to some extent, the development of intracortical networks and functional communication. Thus, deletion of the receptor in GPR88-KO mice leads to microstructural changes such as reduced density of dendritic spines in prefrontal cortex pyramidal neurons (data not shown), remodeling of intracortical networks in somatosensory cortical areas, and delayed responses in sensory processing tasks (Ehrlich et al. 2018). As well, it produces altered somatomotor connectivity and somatosensory–motor–anterior cingulate cortex functional connections (Arefin et al. 2017). In humans, Gpr88 gene was associated with bipolar disorders and schizophrenia (Del Zompo et al. 2014) accordingly to cognitive alterations with learning delay observed in children, from a consanguineous family carrying a biallelic nonsense mutation of GPR88 (Alkufri et al. 2016). Taken together, all of these observations support the hypothesis that the intranuclear GPR88 participates to protein complexes involved in chromatin remodeling in neurons that contribute to the establishment and maintenance of functional neural networks in the cerebral cortex.
Our results lead us to carry out in silico modeling of GPR88 and nuclear proteins, to identify molecular sites for potential interaction between these nuclear proteins with GPR88. Classical homology model (De Gois et al. 2015) of GPR88 allowed us to predict the potential similar structuration of its IL3 and Cter domains (Supplementary Fig. 5A). These two regions are highly enriched in alanine residues (Supplementary Fig. 5B) leading to partial helix structuration, as previously described (Vanhalle et al. 2016). Virtual protein/protein docking experiment (Discovery Studio, Dassault Systèmes BIOVIA) allows us to identify highly probable binding region on TOP2B (Supplementary Fig. 5C,D). As expected, considering the resemblance between IL3 and Cter, interactions of these domains with TOP2B are clearly similar. For ATRX, our results suggest a strong interaction between GPR88 with the DAXX-binding domain of ATRX (Supplementary Fig. 5E,F), known to interact with the corepressor DAXX. Interestingly, IL3 or Cter interaction with ATRX seems to be similar to the interaction of DAXX corepressor with ATRX (Supplementary Fig. 5G), suggesting that GPR88 and DAXX might compete to form a complex with ATRX (Supplementary Fig. 5D). No particular binding region has been significantly identified concerning BAZ2B, probably due to the size of the protein, the complexity of its domains and functions, and the lack of available structural data. Thus, regarding to common structuration, the high degree of similarity between IL3 and Cter primary sequences and protein/protein-binding clusters, it appeared coherent that both regions are implicated in the same regulation, that is, nuclear protein interaction.
In conclusion, our current data provide the first evidences of interaction between an endogenous GPCR with nuclear proteins in adult mouse brain neurons. Understanding the stability of intranuclear receptors and the biological significance of the nuclear GPCRs remains elusive but this peculiar distribution presupposes GPCR’s new function for gene transcription or chromatin remodeling. It is tempting to think that this GPCR is one of those gene activity modulators that could be characterized as new therapeutic targets in cortical development and psychiatric disorders.
Funding
Hybrigenics for the two-hybrid experiments and analysis of the data, Institut National de la Santé et de la Recherche Médicale (INSERM), Centre National de la Recherche Scientifique (CNRS) and Université de Paris.
Notes
We thank the “Cellular and Molecular facility Platform'', US25 Inserm-3612 CNRS, Faculté de Pharmacie de Paris, Université de Paris, Paris, France, for electron microscopy technical assistance. Conflict of Interest: None declared.
References
Author notes
Michèle Darmon and Jorge Diaz equally contributed to this work.
