Inhibition of store-operated calcium channels by N-arachidonoyl glycine (NAGly): no evidence for the involvement of lipid-sensing G protein coupled receptors

N-arachidonoyl glycine (NAGly) is an endogenous lipid deriving from the endocannabinoid anandamide (AEA). Identified as a ligand of several G-protein coupled receptors (GPCRs), it can however exert biological responses independently of GPCRs. NAGly was recently shown to depress store-operated Ca2+ entry (SOCE) but its mechanism of action remains elusive. The major aim of this study was to gain a better knowledge on the NAGly-dependent impairment of SOCE in neurons of the central nervous system (CNS) from mice. First, we examined the expression of genes encoding for putative lipid sensing GPCRs using transcriptomic data publicly available. This analysis showed that the most abundant GPCRs transcripts present in the cerebral cortices of embryonic brains were coding for lysophosphatidic acid (LPA) and sphingosine-1 phosphate (S1P) receptors. Next, the presence of functional receptors was assessed with live-cell calcium imaging experiments. In primary cortical cells S1P and LPA mobilize Ca2+ from internal stores via a mechanism sensitive to the S1P and LPA receptor antagonists Ex26, H2L5186303, or Ki16425. However, none of these compounds prevented or attenuated the NAGly-dependent impairment of SOCE. We found no evidence for the requirement of lipid sensing GPCRs in this inhibitory process, indicating that NAGly is an endogenous modulator interfering with the core machinery of SOCE. Moreover, these data also raise the intriguing possibility that the depression of SOCE could play a role in the central effects of NAGly.

primary cultures of cortical neurons. Cells were dissociated from cerebral cortices collected from embryonic (E13) mice (with the vaginal plug as E0) according to 9,23,24 . Briefly, tissues were placed in a 1.5 mL Eppendorf tube containing 1 mL of an ice-cold Ca 2+ -and Mg 2+ -free Hank's solution supplemented with 33 mM glucose, 4.2 mM NaHCO 3 , 10 mM HEPES, and 1% penicillin/streptomycin. Cells were isolated by a mechanical trituration of the medium containing the cerebral cortices. The cell suspension was filtered through a 40-µm cell strainer before plating the cells on 16 mm ∅ glass coverslips. They were kept in a Neurobasal medium supplemented with B27 (2%) and glutamine (500 µM) and maintained in a 5% CO 2 atmosphere at 37 °C. All the experiments were conducted on cells kept 2 or 3 days in vitro.
Calcium imaging experiments with Fluo-4. The culture medium was removed and replaced by a saline containing (in mM) 150 NaCl, 5 KCl, 1 MgCl 2 , 2 CaCl 2 , 5.5 glucose, 10 HEPES (pH 7.4, NaOH). LPAand S1P-induced Ca 2+ responses were analyzed with Fluo-4. Cells were incubated with 5 μM Fluo-4/AM for 20 min following procedures described previously 23,24 . Images were obtained by a CCD CoolSnap HQ2 camera (Princeton Instruments, Roper Scientific, France) mounted on an inverted Zeiss A1 microscope (Carl Zeiss, France). Cells were excited at 470 nm and emission was collected at 525 nm using a DG-4 wavelength switcher (Princeton Instruments, Roper Scientific, France). MetaFluor (Universal Imaging, Roper Scientific, France) was used for image acquisition and analysis. All experimental procedures were conducted at room temperature. Time-lapse changes in Fluo-4 fluorescence intensity were collected at a frequency of 0.2 Hz from 30-45 cell bodies per dish and analyzed off-line by defining regions of measurements. Results were expressed as F/F0, with F being the fluorescence at each time point and F0 being the mean baseline fluorescence that was monitored at the beginning of each experiment for 1 min before the addition of any substance. Culture dishes were discarded at the end of the recording and never re-used. A positive LPA (or S1P)-induced calcium response was determined as one F/F0 greater than 0.02 that develops within 50 s upon the application of the agonist. Fluo-4 responses were measured as area under curve (AUC).
Calcium imaging experiments with Fura-2. The fluorescent Ca 2+ probe Fura-2 was used to study store-operated Ca 2+ entry (SOCE). The experimental conditions and setup were as above except that cells were incubated with 2.5 µM Fura-2 for 20 min at room temperature. They were then washed twice and kept in a Fura-2-free saline solution for >12 min at room temperature. A dual excitation at 340 and 380 nm was used and emission was collected at 515 nm. Images were acquired at a frequency of 0.2 Hz and analyzed off-line. The classical "Ca 2+ add-back" protocol was used to study SOCE. Cells were bathed in a nominally Ca 2+ -free saline containing (in mM) 150 NaCl, 5 KCl, 3 MgCl 2 , 5.5 glucose, 10 HEPES (pH 7.4, NaOH). SOCE activation was triggered by depletion of the ER Ca 2+ pool with 200 nM thapsigargin, which induced a transient elevation in intracellular Ca 2+ concentration before re-admission of 2 mM external Ca 2+ . SOCE responses were analyzed in cells generating a rapid Ca 2+ rise upon the application of a depolarizing saline containing 90 mM KCl. In cultures of embryonic cortical cells, KCl responding cells are identified as neurons whereas KCl-unresponding cells are considered as non-neuronal cells 25 . The depolarizing (K + rich) medium had the following composition (in mM): 65 NaCl, 90 KCl, 1 MgCl 2 , 2 CaCl 2 , 5.5 glucose, 10 HEPES (pH 7.4, NaOH). Ca 2+ changes as a function of time were expressed as delta ratio F340/F380 whereas total Ca 2+ responses were measured as area under curve (AUC).
Stock solutions of Ex26, Ki16425, and BTP2 were prepared in dimethyl sulfoxide (DMSO). Methanol and ethanol were used for preparing stock solutions of S1P and NAGly, respectively. These stock solutions were diluted at least 1000-fold into the recording saline immediately before use so that the final concentration of vehicle never Analysis of gene expression by RnAseq. The RNASeq gene expression data derive from 22 . Raw fastq files are publicly available and can be found on the GEO repository under accession number: GSEXXX.
Data and Statistical analysis. Each experimental condition as well as its appropriate control were tested on the same batch of primary neuronal cell cultures. For the Ca 2+ imaging experiments, all experiments were done ≥3 times (e.g. with ≥3 distinct biological samples) using distinct dishes from different batches of cells (e.g. from distinct pregnant mice). Data are presented as means ± standard error of the mean (SEM) with n being the number of biological replicates. SigmaPlot (version 10.0, Systat Software) and SigmaStat (version 3.5, Systat Software) were used for plotting graphs and statistical analysis, respectively. Differences between several groups of cells were tested using one-way analysis of variance (ANOVA) followed by a Bonferroni's post hoc test. A P value < 0.05 was considered statistically significant.

Results
mRnA expression of lipid sensing GpcRs in the cerebral cortex of embryonic mice. In order to determine whether NAGly is acting via a GPCR, we analyzed the expression of genes encoding for putative lipid sensing GPCRs in the embryonic cerebral cortex. Table 1 provides the list of the 60 mouse genes selected [26][27][28][29][30] . The transcriptomic data were extracted from a recent RNAseq study 22 . The expression pattern of putative lipid sensing GPCRs was analyzed at 3 embryonic ages: E11, E13 and E17. Only genes for which the number of transcripts per million (TPM) was >2 were considered as significantly expressed 31 , therefore when the number of transcripts was <2 TPM, the gene was eliminated from the analysis. This resulted in the selection of 14 genes encoding for putative lipid sensing GPCRs (Fig. 1). In this RNAseq analysis the genes encoding for GPR18, GPR55 and GPR92, 3 putative targets of NAGly, were not expressed. Overall, the most abundant transcripts were coding for cannabinoid receptors type 1 (CB 1 ) (Cnr1 gene), the orphan receptor GPR12, lysophosphatidic acid (LPA) and sphingosine-1 phosphate (S1P) receptors (Fig. 1). Of note, the abundance of CB 1 and GPR12 transcripts increased markedly during the embryonic development of the cerebral cortex whereas the expression of genes encoding for LPA and S1P receptors was repressed. Since all the live-cell Ca 2+ imaging reported previously were conducted on cortical cells isolated from E13 brain cerebral cortices 9 we focused our attention on the most expressed lipid sensing GPCR genes at that embryonic age: S1pr1, Lpar2 and Lpar6 (vertical arrows, Fig. 1). They encode for S1P1, LPA2 and LPA6 receptors, respectively. CB 1 was excluded from our analysis because NAGly has no affinity for CB 1 receptors 32 and the CB 1 antagonist AM251 did not prevent the NAGly-induced responses in cortical neurons 9 , arguing against a role for these receptors. On the other hand, GPR12 was also not considered as a likely target of NAGly because the GPR12 gene was weakly expressed at E13 (Fig. 1). Its expression was strongly upregulated but only at the end of corticogenesis (E17).
The S1P 1 receptor antagonist Ex26 (1 µM) 43 reduced the peak amplitude of the S1P-induced Ca 2+ signals and diminished the number of responsive cells with only 12 cells out 220 tested (⁓5%) generating a Ca 2+ signal in response to 10 µM S1P (Fig. 2D). In each instance, depleting the ER with thapsigargin prevented the development of a Ca 2+ rise upon LPA or S1P application (Fig. 2B,D).
Previous reports showed that LPA and S1P receptors are mainly found in proliferative regions of the immature cerebral cortex, with few post-mitotic neurons responding to LPA and S1P 35 . This latter point was checked by using a depolarizing saline solution containing 90 mM KCl to evoke KCl-dependent Ca 2+ rises. Acutely cultured cells were undifferentiated cells. When cultured for several days, some of these differentiate into neurons (post-mitotic) responding to high-K + whereas non-differentiated cells are not high-K + sensitive. In cultures of embryonic cortical cells, KCl responding cells are identified as neurons whereas KCl-unresponding cells are considered as non-neuronal cells 25 . Overall, only 10 of 67 LPA sensitive cells (⁓15%) generated an intracellular Ca 2+ rise in response to KCl. These data are consistent with a previous report showing that in the embryonic cerebral cortex LPA receptors are predominantly expressed by neural precursor cells with only a small minority of neurons responding to LPA 35 . On the other hand, 5 of 25 S1P sensitive cells (20%) were KCl-responsive cells. This indicates that the S1P-sensitive cells are also mainly found in KCl-insensitive cells 34 . Taken together, LPA or S1P mobilizes Ca 2+ from the ER in a subset of cells (<20%). These functional LPA-and S1P-sensitive receptors are essentially expressed by non-neuronal cells 35 www.nature.com/scientificreports www.nature.com/scientificreports/ Before testing the contribution of LPA and S1P receptors in the NAGly-dependent alteration of SOCE, it was important to check whether the receptor antagonists Ki16425 and Ex26 could alter SOCE on their own. In the following experiments, the ratiometric Ca 2+ probe Fura-2 was used to analyze SOCE in cells that responded to the KCl challenge (i.e. post-mitotic neurons). Cells, bathed in a nominally Ca 2+ -free medium, were challenged with thapsigargin to deplete ER Ca 2+ stores. A subsequent re-admission of external Ca 2+ was followed by an intracellular elevation of Ca 2+ (open circles, Fig. 3A) 9,24 . This entry of Ca 2+ was sensitive to the CRAC channel blocker BTP2 45,46 (1 µM, gray up triangles, Fig. 3A). The thapsigargin-evoked Ca 2+ release was unaffected by Ex26 (1 µM, filled down triangles) or Ki16425 (10 µM, gray squares) (Fig. 3A). The SOCE response was however upregulated by Ki16425 but not by Ex26. This is further illustrated in Fig. 3B showing the Ca 2+ release and entry analyzed as area under the curve for each condition tested. Ki16426 enhanced the SOCE signal by nearly 30% (n = 5, p < 0.05) (Fig. 3B, gray bar). Altogether, these data show that the LPA and S1P receptor antagonists used did not alter the ER Ca 2+ release. The SOCE response was also unaffected by Ex26 but augmented by Ki16426. This potentiating effect was not investigated further.

NAGly depresses SOCE independently of LPA and S1P receptors.
After having shown the presence of functional receptors sensitive to LPA and S1P, their involvement in the NAGly-induced impairment of SOCE was considered. In the following set of experiments, Fura-2-loaded cells were first stimulated with a K + -rich saline (90 mM KCl) before recording SOCE responses in neurons (i.e. in KCl-responsive cells). Figure 4A shows SOCE without NAGly (open circles) and in the presence of NAGly (10 µM, gray down triangles). As already illustrated 9 , NAGly exerts complex actions on neuronal Ca 2+ signalling: (i) it induces a release of cations (Ca 2+ and Zn 2+ ) that develops prior to thapsigargin addition (phase ➀, Fig. 4A); (ii) it upregulates the thapsigargin-dependent Ca 2+ release (phase ➁); and (iii) reduces the amplitude of SOCE (phase ③). Even in the presence of 1 µM Ex26 (gray up triangles) or 10 µM Ki16425 (filled squares, Fig. 4A), NAGly elevated the Fura-2 fluorescence on its own (phase ➀) and potentiated the thapsigargin-evoked Ca 2+ release (phase ➁). The NAGly-induced inhibition of SOCE (phase ③) was also not affected by Ex26 or Ki16425 (Fig. 4A). NAGly had however no inhibitory action on the entry of Ca 2+ when added together with BTP2 (open squares, Fig. 4A).

Discussion
NAGly inhibits SOCE 20 . This impairment has been observed in every cell type and cell line tested so far like fibroblasts, neurons, EA.hy926 (human endothelial cell line), INS-1 832/13 (rat pancreatic β-cell line), and RBL-2H3 cells (rat basophilic leukemia cell line) 9,20,21 . However, the mechanism by which NAGly alters SOCE is unclear. In the present study we addressed the question of the contribution of lipid sensing GPCRs as targets of NAGly  Table 1). Transcripts of only 14 genes (out of 60) could be detected (e.g. having TPM values ≥ 2). The graph shows the temporal pattern of the mRNA abundance of these 14 genes at 3 embryonic ages: E11, E13 and E17. Genes that were induced (Cnr1, Gpr4, Gpr12, Gpr17, Gper1, Gpr34, Adgrb1) are shown on the left whereas genes that were repressed (Lpar1, Lpar2, Lpar4, Lpar6, S1pr1, S1pr2, S1pr3) appear on the right. Vertical arrows indicate the 3 most abundant transcripts at E13 (except CB 1 , see text for further details).
Five subtypes of S1P receptors are known (S1P [1][2][3][4][5]. They belong to the group of GPCRs and mediate most of the biological actions of the bioactive sphingolipid S1P 30 . Embryonic cerebral cortices displayed a high mRNA level of S1P 1 receptors that declined during embryonic brain development. In addition, cultured cortical cells expressed functional receptors coupled to the release of Ca 2+ from the ER and sensitive to the S1P 1 antagonist Ex26. These findings are in line with previous reports showing that S1P 1 is the major S1P receptor of the murine embryonic brain, followed by S1P 2 and S1P 3 receptors. It is detected as early as E14, highly expressed in proliferative regions (neurogenic ventricular zone) but its expression decreases at E16 and E18 47 . The activation of S1P 1 receptors is coupled to the mobilization of Ca 2+ 33 .
LPA receptors constitute another important family of GPCRs sensitive to bioactive lipids 30,39 . LPA signalling is of particular physiological relevance for the embryonic brain cortex 48 . At E12.5 the most abundant transcripts in the telencephalon are LPA 1 , LPA 2 and LPA 4 35 . In the present work, the main genes expressed at E13 were encoding for LPA 2 and LPA 6 . Moreover, the application of LPA caused the release of Ca 2+ from the ER. These responses were highly sensitive to the LPA 1/3 antagonist Ki16425 but moderately affected by the LPA 2/3 antagonist H2L5186303 30,39 . The pharmacological dissection of the LPA-induced Ca 2+ signalling pointing to LPA 1/3 as the likely receptors responding to LPA is difficult to reconcile with the gene analysis showing that LPA 1 and LPA 3 are, respectively, very weakly expressed and undetected. The pharmacological properties of native LPA receptors of cortical neurons may differ from those of LPA receptors heterogeneously expressed.
After having shown the presence of functional LPA and S1P receptors, their contribution to the NAGly-dependent depression of SOCE was evaluated. The pharmacological blockade of S1P and LPA receptors with Ex26 or Ki16425 did not abolish or attenuate the NAGly-dependent impairment of SOCE. Some cellular responses of NAGly have been shown to be mediated by the orphan receptor GPR55 11 . However, we found no evidence for the presence of significant levels of GPR55 mRNA. Furthermore, the GPR55 agonist AM251 49 , which induces a GPR55-dependent mobilization of Ca 2+ with an EC 50 of ~0.6 µM 50 , fails to evoke any Ca 2+ release when Figure 3. Effects of Ex26, Ki16425, and BTP2 on the thapsigargin-evoked Ca 2+ release and SOCE. SOCE responses were analysed with Fura-2. Cells were kept in a nominally Ca 2+ -free medium. ER Ca 2+ stores were depleted with thapsigargin (Tg, 200 nM) before re-introducing external Ca 2+ . The resulting increase in intracellular Ca 2+ is due to Ca 2+ entering via the plasma membrane. Panel A shows somatic Ca 2+ responses (expressed as Δ ratio F340/F380) as a function of time, and generated by the sequential addition of Tg (200 nM, horizontal gray bar) followed by the readmission of 2 mM external Ca 2+ (horizontal black bar). Four conditions are shown: without antagonists of LPA and S1P receptors (Control, open circles, n = 7), with 1 µM Ex26 (gray triangles, n = 5), with 10 µM Ki16425 (filled squares, n = 5), and with 1 µM BTP2 (symbols, n = 5). When tested, Ex26 (or Ki16425) and BTP2 were added 4-7 and 11-12 min, respectively, before time 0 and were also present during the recordings. One time point out of 3 is shown. Panel B shows the thapsigargin-evoked Ca 2+ release and SOCE measured as area under the curve (AUC). Mean ± SEM. www.nature.com/scientificreports www.nature.com/scientificreports/ applied to cortical cells at 10 µM. This further suggests that GPR55 does not participate in the NAGly-induced alteration of neuronal Ca 2+ signalling.
In conclusion, our data show that NAGly inhibits a BTP2-sensitive Ca 2+ entry, which is most likely a SOCE. This occurs independently of GPR55, LPA and S1P receptors (present report), and via a mechanism insensitive to the pertussis toxin 9 . It is worth recalling that NAGly regulates voltage-gated Ca 2+ channel activity without acting on GPCRs 8,13 . Although we cannot exclude the possibility that NAGly acts on an orphan lipid sensing GPCR that was not considered in the present study, our report suggests that NAGly disturbs the coupling of the core components of the SOCE machinery (STIM-Orai) 20 . This inhibitory process does not seem to develop in response to an intracellular signalling cascade. These past 9 and present data show that the phytocannabinoid cannabidiol, the endocannabinoid AEA and its derivative NAGly are potent inhibitors of neuronal SOCE. This indicates that NAGly and endocannabinoids are endogenous SOCE modulators, and raises the possibility that the depression of SOCE could play a role in the neuro-behavioural effects of cannabinoids and signalling lipids. Open circles: control conditions (without NAGly) (n = 7). When indicated, 10 µM NAGly was added (vertical arrow) prior to thapsigargin. This elevated the Fura-2 fluorescence (phase ➀) (black triangles, n = 6). Similar experiments were conducted in the presence of NAGly + 10 µM Ki16425 (gray squares, n = 5), NAGly + 10 µM Ex26 (open triangles, n = 4), and NAGly + 1 µM BTP2 (symbol, n = 3). As in Fig. 3, Ki16425 (or Ex26) and BTP2 were added 4-7 and 11-12 min before time 0 and remained present throughout the recordings. One time point out of 3 is shown. Mean ± SEM. Panel B: Area under curve (AUC) measurements of Fura-2 signals under the different conditions tested. Three phases were considered: Ca 2+ signals prior to the addition of thapsigargin (phase ➀), the thapsigargin-induced Ca 2+ release (phase ➁) and SOCE (phase ➂). *p < 0.05 vs NAGlyuntreated cells, one-way ANOVA followed by a Bonferroni's post hoc test.