Sphingosine-1 phosphate induces cAMP/PKA-independent phosphorylation of the cAMP response element-binding protein (CREB) in granulosa cells

Background and aims: Sphingosine-1 phosphate (S1P) is a lysosphingolipid present in the ovarian follicular fluid. The role of the lysosphingolipid in gonads of the female is widely unclear. At nanomolar concentrations, S1P binds and activates five specific G protein-coupled receptors (GPCRs), known as S1P1-5, modulating different signaling pathways. S1P1 and S1P3 are highly expressed in human primary granulosa lutein cells (hGLC), as well as in the immortalized human primary granulosa cell line hGL5. In this study, we evaluated the signaling cascade activated by S1P and its synthetic analogues in hGLC and hGL5 cells, exploring the biological relevance of S1PRstimulation in this context. METHODS AND RESULTS. hGLC and hGL5 cells were treated with a fixed dose (0.1 μM) of S1P, or by S1P1and S1P3-specific agonists SEW2871 and CYM5541. In granulosa cells, S1P and, at a lesser extent, SEW2871 and CYM5541, potently induced CREB phosphorylation. No cAMP production was detected and pCREB activation occurred even in the presence of the PKA inhibitor H-89. Moreover, S1Pdependent CREB phosphorylation was dampened by the mitogen-activate protein kinase (MEK) inhibitor U0126 and by the L-type Ca2+ channel blocker verapamil. The complete inhibition of CREB phosphorylation occurred by blocking either S1P2 or S1P3 with the specific receptor antagonists JTE-013 and TY52156, or under PLC/PI3K depletion. S1P-dependent CREB phosphorylation induced FOXO1 and the EGF-like epiregulinencoding gene (EREG), confirming the exclusive role of gonadotropins and interleukins in this process, but did not affect steroidogenesis. However, S1P or agonists did not modulate granulosa cell viability and proliferation in


Introduction
Sphingosine-1-phosphate (S1P) is a lysosphingolipid present at high concentrations in plasma and lymph. While implicated in several intracellular processes such as cell growth, survival, motility and migration (Strub et al., 2010), S1P can be transported out of the cell, where it acts in autocrine and paracrine mode via specific membrane receptors. Extracellular S1P is associated with apolipoprotein M (ApoM), which is a constituent of high-density lipoproteins (HDL) (Potì et al., 2014). HDL-associated S1P is involved in a number physiological processes, acting both on vascular and immune cell targets (Obinata and Hla, 2012). In endothelial cells, S1P induces cell migration (Liu et al., 2000) and survival, promotes DNA synthesis and enhances endothelial barrier integrity (Liu et al., 2000;MacLennan et al., 2001). The positive impact on cell proliferation and motility may be exerted by S1P and its precursor sphingosine via protein kinase C activation (Merrill and Stevens, 1989;Igarashi, 1997).
In the present study we aimed to further evaluate the biological relevance of S1P signaling for the granulosa cell function by exploring the lysosphingolipid-induced signaling cascades. To this purpose, primary granulosa lutein cells (hGLC) and the hGL5 cell line were used in parallel. We demonstrate for the first time that S1P and its synthetic analogues induce phosphorylation of the cAMP-responsive element binding protein (pCREB) and thereby affects steroidogenesis, gene expression and cell proliferation.

Human samples and patient selection
Human primary granulosa lutein cells (hGLC) were isolated from the follicular fluids of women undergoing oocyte retrieval for assisted reproduction at the Santa Maria Nuova hospital (Reggio Emilia, Italy). Patients had to match the following criteria: absence of endocrine abnormalities, viral or bacterial infections; age between 25 and 45 years. The study was approved by the local Ethics Committee (protocol n. 2014(protocol n. /0015,349, June 12, 2014 and each patient provided the informed consent.

Granulosa-lutein cell isolation and culture
hGLC were purified using a 50% Percoll gradient (GE Healthcare, Little Chalfont, UK), thus separating these cells from other cellular components by centrifugation, as previously described Casarini et al., 2017). A hemolysis buffer was added to remove red blood cells contamination and finally blocked with complete medium containing DMEM/F12 (Gibco, Thermo Fisher Scientific, Waltham, MA, USA), 5% fetal bovine serum (FBS), 100 IU/ml penicillin, 50 μl/ml streptomycin, 2 mM glutamine and 2.5 μg/ml Amphotericin B (all from Sigma-Aldrich Corporation, Saint Louis, MO, USA). Cells were maintained in an incubator at 37 • C and 5% CO 2 for about 6 days before each experiment to allow recovery of G protein-coupled receptor gene expression (Nordhoff et al., 2011) and serum-starved over-night before stimulation.

Time-course experiments
hGLC and hGL5 cells were seeded in 24-well plates (10 × 10 4 cells/ well) in 500 μl of culture medium. Cells were treated with the lowest effective concentration detected by dose-response experiments, i.e. 0.1 μM, over 0-120 min, which is consistent with the 50% effective concentrations of these compounds previously described (Jo et al., 2012;Gonzalez-Cabrera et al., 2008) and falling within the nM range.

Evaluation of cAMP production
hGL5 cells were treated with the phosphodiesterase inhibitor 3-isobutyl-1-methylxanthine (IBMX) (Sigma-Aldrich) 20 min before 3-h exposure to increasing S1P, SEW2871 and CYM5541 concentrations (nM-μM range). Reactions were blocked by rapid freezing at − 80 • C. Total cAMP production was evaluated by ELISA following the manufacturer's instruction (#H019-H5; Arbor Assays, Ann Harbor, MI, USA) and values represented in a graph as means ± standard deviation (SD).
For protein extraction, cells were lysed in ice-cold RIPA buffer added with protease and phosphatase inhibitors.
Lysates from over-night, 0.1 μM S1P-and 10 μM LY294002-treated cells were also loaded into acrylamide gel wells and used for total protein pattern analysis by 1-h Coomassie Brilliant Blue (Sigma-Aldrich) staining, followed by 20-min washing, after electrophoresis. The molecular weight standard was a Plus Protein™ Dual Color Standard (Bio-Rad Laboratories Inc., Hercules, CA, USA). S1P-, CYM5541-and SEW2871-induced pCREB activation was evaluated by Western blotting after 12% SDS-PAGE, using a specific antibody (#9198; Cell Signaling Technology, Boston, MA, USA). Total ERK1/2 was used as a protein loading control (#4695; Cell Signaling Technology Inc.). Signals were revealed by ECL chemiluminescent compound (GE HealthCare, Chicago, IL, USA), after incubation of the membranes by a secondary anti-rabbit horseradish peroxidaseconjugated antibody (#NA9340V; GE HealthCare). Western blotting images were acquired by the Molecular Imager VersaDoc™ MP 4000 System and QuantityOne analysis software (Bio-Rad Laboratories Inc.) and semi-quantitatively evaluated by the ImageJ software (U.S. National Institutes of Health, Bethesda, MD, USA).

Intracellular Ca 2+ measurements by BRET
hGL5 cells were transfected by aequorin Ca 2+ biosensor-encoding cDNA plasmid carriers two days before BRET measurements (Adamczyk et al., 2001). Transient transfections were performed by reverse transfection in 96-well plate (25 × 10 3 cells/well) using 2.5 μl/well Lipofectamine (Thermo Fisher Scientific), following the manufacturer's protocol, and 1000 ng/well of plasmid. Then, culture medium was removed and cells were treated 2 h by 100 μl/well PBS, 1 mM HEPES and 5 μM coelenterazine. The solution was replaced by 50 μl HBSS, 1 mM HEPES and 5 μM coelenterazine H, and maintained under incubation 45 min. Kinetics of intracellular Ca 2+ increase was evaluated over at least 150 s upon injection of 50 μl/well of 10 μM S1P, SEW2871 and CYM5541. The injection was set at the 20 s time-point. Measurements were performed in the presence or in the absence of 100 μM verapamil, 100 μg/ml fucoidan and 10 μM phospholipase C (PLC) inhibitor U73122 (Morley et al., 1992). Cells treated with PBS-1 mM HEPES or 5 μM thapsigargin (Sigma-Aldrich), acted as negative or positive controls, respectively.

Progesterone stimulation protocol and measurements
hGLC and hGL5 cells were seeded in 48-well plates (3 × 10 5 cells/ well) and serum starved 12 h before treatments. Cells were treated with 0.1 μM S1P, SEW2871 and CYM5541, while 50 μM forskolin and 0.05 μM FSH stimuli were used for positive controls. Further control experiments were performed using the hGL5 cell line, which was exposed to U0126, H-89, W146 and TY52156 before treatments. Reactions were blocked by immediate sample freezing at the 24, 48 and 72 h time-point, as indicated in the figures. Total progesterone levels were measured by immunoassay (ARCHITECT 2nd Generation Progesterone system; Abbot Diagnostic, Chicago, IL, USA).

S1P induces CREB phosphorylation in granulosa cells
As determined in dose-response experiments (Fig. S2), the optimal concentration for cell treatment with native S1P as well as SEW2871 and CYM5541, two selective S1P 1 and S1P 3 agonists, respectively, was 0.1 μM. This concentration was subsequently used for the temporal evaluation of ERK1/2, AKT and CREB phosphorylation. hGLC and hGL5 cells were exposed for various times (0-120 min) to each agonist and protein phosphorylation was detected by Western blotting. Signals were quantified by densitometry and are presented in graphs using total ERK as a normalizer. S1P induced pERK1/2 activation in hGL5 cells, although achieving the maximum between 5 and 15 min upon cell treatment and downward tendency thereafter (Fig. 1). By contrast, the ERK1/2 phosphorylation induced by SEW2871 and, especially, CYM5541 was prolonged and still detectable up to 30 min compared to S1P (panel A, B). Reflecting results obtained from hGLC ( Fig. S3), all ligands failed to induce pAKT activation in hGL5 (panel A, C). However, S1P markedly elevated the levels of phosphorylated CREB in these cells with the time course resembling primary granulosa cells, while CYM5541 and SEW2871 exerted no effect on CREB phosphorylation in hGL5 cells (panel A, D). It is worth noting that neither S1P nor selective agonists affect the expression of S1P receptors in our experimental setting (Fig. S1).

S1P fails to affect cAMP production in granulosa cells
Since stimulation of S1P receptors results in the CREB phosphorylation ( Fig. 1; Fig. S3), which is known to be cAMP-dependent in granulosa cells (Pei et al., 1991), we next evaluated the production of this second messenger under our experimental conditions. In both hGLC and hGL5 cells (panels A and B, respectively), the cAMP total cell content was measured after 3-h treatment with increasing doses of S1P, SEW2871 and CYM5541 and in the presence of 500 μM IBMX to prevent cAMP degradation by phosphodiesterases. Cholera toxin (CTX)-treated cells served as a positive control. While the sensitivity of the experimental setting was demonstrated by the effective response in positive controls, both S1P and S1P receptor agonists failed to induce cAMP production in both hGLC and hGL5 cells (Fig. 2). These data suggest that S1PRs-mediated CREB phosphorylation does not involve cAMP production in granulosa cells.
As expected, in both hGLC and hGL5 cells S1P-induced pCREB activation occurred in the presence of H-89 ( Fig. 3; Supplementary Table S1), confirming that the cAMP/PKA-pathway is not a part of the lysosphingolipid signaling cascade converging on CREB. Moreover, the phosphorylation of CREB was moderately attenuated in the presence of U0126 (Fig. 3, Fig. S4). Likewise, the L-type calcium channel blocker verapamil moderately inhibited the S1P-induced CREB phosphorylation. By contrast, this response was completely abolished by LY294002, 3CAI and U73122 inhibitors, suggesting that PI3K and PLC are prerequisites for the S1PR-mediated CREB phosphorylation. Similar effects of inhibitor treatment were seen in cells exposed to SEW2871 and CYM5541. However, these data should be interpreted with caution, considering the lower pCREB activation levels achieved with selective agonists compared with S1P treatment, as well as the lack of specific protein pattern after Coomassie blue staining of S1P-and LY294002treated cells (Fig. S5).
To further address the identity of the S1P receptors involved in the CREB phosphorylation, granulosa cells were treated with S1P or the agonists in the presence of the selective S1P receptor antagonists W146 (S1P 1 antagonist), JTE-013 (S1P 2 antagonist), and TY52156 (S1P 3 antagonist). As shown in Fig. 3, only marginal effect on the S1P-induced CREB phosphorylation were observed in cells pre-treated with the selective S1P 1 antagonist W146. By contrast, S1P 2 and S1P 3 antagonists JTE013 and TY52156 completely blocked CREB phosphorylation. Fig. 1. Evaluation of pERK1/2, pAKT and pCREB activation by time-course experiment in immortalized human primary granulosa (hGL5) cell line. A) Cells were treated with 0.1 μM S1P, SEW2871, CYM5541 or vehicle for different times (0-120 min). pERK1/2, pAKT and pCREB were evaluated by Western blotting (images representative of three independent experiments). The values were normalized to total ERK. B-D) Semi-quantitative evaluation of pERK1/2, pAKT, pCREB and total ERK Western blotting signals. Bars indicate means ± SD (* = significantly different versus control; two-way ANOVA, p < 0.05; n = 3).

S1P fails to stimulate progesterone production in granulosa cells
CREB phosphorylation in granulosa cells is linked to the enhanced expression of genes encoding steroidogenic enzymes, which culminates in the synthesis of the steroid hormone progesterone. Therefore, the effect of S1P on the 24-h steroidogenesis was next evaluated in hGLC and hGL5 cells treated by the native lysosphingolipid or selective agonists. Treatment with forskolin (50 μM) or FSH (50 nM) FSH-treated cells served as positive controls (Casarini et al., 2014) and progesterone levels in cell culture supernatants were measured by immunoassay. As shown in Fig. 4, neither S1P nor the synthetic S1P mimetics SEW2871 and CYM5541 induced progesterone production in granulosa cells which was instead activated by forskolin (Fig. 4). 1.5-fold increase of progesterone production occurred upon FSH treatment, although this effect was not statistically significant. These results demonstrate that lysosphingolipid-dependent CREB phosphorylation does not induce a steroidogenic signal. As expected, no progesterone synthesis was detected in the hGL5 cell line (not shown), which likely reflects the uncoupling of Gα s protein from the FSH receptor and/or the low adenylyl cyclase enzyme expression levels in this cell line (Casarini et al., 2016a).

S1P produces PLC-independent Ca 2+ influx in granulosa cells
Our results reveal that the activation of PLC is required for the S1P receptor-mediated CREB phosphorylation in granulosa cells and the Ltype calcium channel blocker verapamil moderately inhibits CREB phosphorylation in the same experimental setting (Fig. 5). Therefore, we   next evaluated the direct effect of S1P and its mimetics on the intracellular Ca 2+ mobilization. To this purpose, hGL5 cells, expressing the aequorin Ca 2+ biosensor were treated with S1P, SEW2871 and CYM5541 in the presence of verapamil, fucoidan and U73122. Both S1P and agonists were administered at the concentration of 10 μM (Choi et al., 2009) because of the sub-optimal intracellular Ca 2+ increase, which was previously observed in granulosa cells at lower agonist doses (Murakami et al., 2008). The capability of U73122 in inhibiting the 10 μM S1P-induced cell response was also confirmed (Fig. S6). The kinetics of intracellular Ca 2+ increase was evaluated by BRET over 150 s. Thapsigargin was used as a positive control, while PBS-treated cells indicated the background signal. Intracellular Ca 2+ rapidly increased upon thapsigargin and S1P injection, both in the presence and in the absence of the PLC inhibitor U73122 pre-treatment. However, both the L-type calcium channel blocker verapamil and the inhibitor of the Ca 2+ -response fucoidan (Wu et al., 2018) reduced the surge of S1P-mediated intracellular Ca 2+ . Both SEW2871 and CYM5541 compounds failed to increase intracellular Ca 2+ , regardless of the presence of inhibitors.

S1P promotes the expression of pCREB target genes in granulosa cells
Beyond its involvement in steroidogenesis, CREB regulates the expression of several target genes involved in the modulation of the cell cycle as well as pro-and anti-apoptotic events. Therefore, a gene expression analysis was performed in hGLC exposed for 24 h to S1P, SEW2871 and CYM5541. Expression levels of genes known to be induced by CREB were evaluated by real-time PCR. Unstimulated cells were used for estimating basal expression levels (Fig. 6) and normalizing data, while the PI3K inhibitor LY294002 served to demonstrate the involvement of the S1P-induced activation of CREB in regulating target gene expression.
As shown in Fig. 6, the EGF-like factor expressing-gene EREG, the gene encoding cyclooxygenase-2 enzyme PTGS2, and the forkhead in rhabdomyosarcoma FOXO1 were positively regulated by S1P treatment. Moreover, PI3K blockade inhibited the S1P-induced expression of EREG and FOXO1, demonstrating their dependence on pCREB. By contrast, cell treatment with SEW2871 and CYM5541 failed to produce any statistically significant effects as compared to unstimulated cells. The expression of other EGF-like factors encoding genes (AREG), the EGF neuregulin (NRG1), aromatase (CYP19A1), X-linked inhibitor of apoptosis factor (XIAP), B-cell lymphoma 2 (BCL2) and the tyrosine kinase (ERB-B1) were modulated neither by S1P treatment nor by the PI3K inhibitor (Fig. 6, Fig. S7).

S1P does not induce proliferative or anti-apoptotic signals in granulosa cells
The S1P-mediated expression of EREG and FOXO1 point to the  potential involvement of S1P in the regulation of cell death/survival pathways in granulosa cells (Park et al., 2005). Therefore, we next assessed the proliferative potential of S1P in the hGL5 cell line. To this purpose, hGL5 cells were maintained in medium containing the charcoal-stripped serum over 72 h, in the presence or in the absence of S1P, SEW2871 and CYM5541. Under these condition, interferences due to serum hormones are minimized (Casarini et al., 2016a) although hGL5 cells proliferate. Since cultured hGLC are not able to proliferate, they were used as a model for evaluating the anti-apoptotic potential of S1P and agonists. To this purpose, serum-starved primary cells were maintained under continuous stimulation by compounds over 72 h. Cell viability of both the cell models was evaluated by the MTT assay.
In hGL5 cells, S1P and its synthetic agonists only weakly promoted the viability of growing cells, suggesting that their effect on cell proliferation is negligible under these conditions (Fig. 7). hGLC treatment by S1P and agonists did not change the cell viability over 72 h, revealing that these compounds did not counteract the pro-apoptotic effect induced by serum deprivation (Casarini et al., 2017), at least under our experimental conditions (Fig. S8).

Discussion
The present study for the first time demonstrates that S1P induces cAMP/PKA-independent phosphorylation of pCREB, plausibly via a PLC/PI3K-dependent mechanism, in human granulosa cell. This is a surprising finding, as it is well established that both cAMP and PKA activation precedes the phosphorylation of CREB, inducing granulosa cell steroidogenesis (Chen et al., 2007). Although the phosphorylation of CREB results in transcription of EGF-like and FOXO1 encoding genes, confirming previous data observed in endothelial and smooth muscle cells (Zhang et al., 2012;Hsu et al., 2015), the impact of this effect on proliferation and starvation-induced cell death is weak, or even absent in vitro. Most importantly, no steroidogenic effect is linked to the S1P 1 -mediated CREB phosphorylation, confirming the exclusive role of gonadotropins and interleukins in triggering granulosa cells production of steroid hormones via activation multiple signaling cascades parallel to the cAMP/PKA-pathway (Casarini et al., 2018a;Dang et al., 2017;Ulloa-Aguirre et al., 2011).
It was previously suggested that S1P targets PKA via a cAMPindependent mechanism in COS cells, and that this mechanism accounts for various biological activities of this lysosphingolipid such as cell cycle control, proliferation and survival (Ma et al., 2005). However, in the present study we found that PKA is not involved in S1P receptor-mediated CREB phosphorylation, as it occurred even in the presence of the specific PKA inhibitor H-89 in granulosa cells. Moreover, upon cell treatment with S1P, CREB phosphorylation was only slightly dampened by the MEK inhibitor U0126, similarly to what was demonstrated in gonadotropin-treated cells maintained under MEK depletion (Casarini et al., 2014), while the L-type Ca 2+ channel blocker verapamil moderately reduced CREB phosphorylation, as previously documented in other cell models (Guo et al., 2016). On the other hand, Ca 2+ /calmodulin-dependent protein kinases were linked to S1P-mediated CREB phosphorylation (Coussin et al., 2003), at least in the central nervous system, suggesting the involvement of this ion in the modulation of S1P receptors signaling. Indeed, while the relationship between S1P 1 and Ca 2+ remains controversial, studies in some few mouse models revealed that the activity of this receptor may be linked to the modulation of Ca 2+ homeostasis (Keul et al., 2016). Interestingly, we demonstrated for the first time that S1P-induced intracellular Ca 2+ increase may be reduced by fucoidan in granulosa cells, a not completely understood mechanism in line with that previously reported using various GPCR agonists in other cell models (Wu et al., 2018). The complete inhibition of CREB phosphorylation occurred by antagonizing S1P 2 , S1P 3 or after blocking PLC/PI3K activity. These data point out to the possible role of S1P 2 , which might play a fundamental role in activating S1P-dependent signaling cascades, in granulosa cells. The key role possibly played by the PLC/PI3K-pathway in mediating S1P-induced CREB phosphorylation was described previously in smooth muscle cells (Hsu et al., 2015), where it increased cell migration and the expression of the PTGS2 gene via FOXO1-mediated mechanisms. Interestingly, in that study, S1P 1 and S1P 3 antagonists prevented CREB phosphorylation and PTGS2 gene transcription, corroborating the idea that S1P might be involved in pro-inflammatory processes in the ovary (Hernández-Coronado et al., 2019).
The activation of intracellular signal transduction by S1P occurs in a nanomolar concentration range, which is considered physiological and which is effective for most of the synthetic S1P analogues. However, our results revealed that S1P is much more potent than SEW2871 and CYM5541 in phosphorylating CREB and ERK1/2 in hGLC, consistently with the ligand-specific gene expression pattern and suggesting that more than one S1P receptor is required for a full activation of signaling cascades. Results of the present study obtained in cells treated with S1P 2 and S1P 3 antagonists demonstrated that their contribution is essential and required for effective triggering CREB phosphorylation, since it was fully abolished in the presence of the specific antagonists JTE-013 and TY52156. These effects could be explained by the modulation of S1Pdependent signals through formation of S1P receptor heterodimers (Van Brocklyn et al., 2002;Zaslavsky et al., 2006;Siehler and Manning, 2002;Kluk and Hla, 2002), which has been previously described in other cells and also for other receptors expressed in granulosa cells (Casarini et al., 2018a). However, we can not exclude that synthetic ligands and S1P may activate different conformational states of the receptors, and hence a distinct profile of intracellular signaling cascades (Troupiotis-Tsaïlaki et al., 2017;Swamy et al., 2018;Jiang and Zhang, 2019).
The present study recapitulates previous findings demonstrating ERK1/2 phosphorylation by S1P receptor-stimulation in granulosa cells (Nakahara et al., 2012) that occurs upon treatment in micromolar S1P concentration range. As ERK1/2 phosphorylation is associated with cellular proliferation in both granulosa primary cells in vivo (Donaubauer et al., 2016) and in granulosa cell lines in vitro (Casarini et al., 2016a;Kandaraki et al., 2018) we hypothesized that hGL5 cell treatment by S1P might favorably impact cell viability. As a matter of fact, S1P enhanced the expression of NRG1 and PTGS2, two genes typically linked to proliferative signals (Hashemi Goradel et al., 2018;Ogier et al., 2018). However, our data reveal that hGL5 cell viability is only slightly improved by S1P receptor stimulation. We indeed observed the absence of AKT phosphorylation either by S1P or its synthetic analogues. These data are surprising since CREB phosphorylation is prevented by inhibiting the PI3K, as well as AKT, which likely targets CREB without activating downstream pAKT at detectable levels by Western blotting. As AKT was repeatedly reported to counteract apoptotic events in Fig. 7. Analysis of S1P-induced proliferative and anti-apoptotic signals by MTT. hGL5 were treated 0-72 h in the presence of S1P and agonists. Data are indicated by box and whiskers plot (* = significantly different versus unstimulated day 0; two-way ANOVA and Bonferroni post-test; p < 0.05; n = 18). granulosa cells (Casarini et al., 2012;Hunzicker-Dunn et al., 2012), the lack of its activation might explain the missing improvement of hGLC viability under serum deprivation in the presence of S1P (Fig. S8).
We here described for the first time CREB phosphorylation not involving PKA but rather PLC/PI3K activation in granulosa cells. While pCREB activation is a well-established cAMP/PKA-dependent event in granulosa cells treated by other GPCR agonists, such as gonadotropins (Zhang et al., 2012;Mukherjee et al., 1996), the involvement of PLC in triggering the phosphorylation of CREB is controversially discussed (Salvador et al., 2002). In granulosa cells, activation of pCREB is linked to cell survival (Somers et al., 1999) and positively impacts expression of genes encoding steroidogenic enzymes (Carlone and Richards, 1997), thus stimulating the synthesis of steroid hormones (Casarini et al., 2018b). These data should be interpreted in light of the role of PLC, which may potentiate the cAMP/PKA/CREB-induced expression of steroidogenesis-related genes (Manna et al., 2009). However, we found that the activation of PLC and pCREB together are not sufficient to induce CYP19A1 gene expression (Fig. S7) and progesterone production. While PKA may not be required to activate the synthesis of steroid hormones (Chin and Abayasekara, 2004), except under extraordinary conditions (Manna et al., 2014), it was hypothesized that cAMP provides essential steroidogenic signals in granulosa cells. cAMP-independent gene expression and steroidogenesis was described solely in adrenal cells, where angiotensin II may activate pCREB via serine/threonine protein kinase D (PKD) (Olala et al., 2014). Whether such intracellular signaling cascade operates in granulosa cells, remains to be established. Otherwise, the molecular mechanism underlying failure of S1P receptor-mediated steroidogenic signal might be explained in terms of a specific compartmentalization of the signaling module (Moger, 1991;Sposini and Hanyaloglu, 2018;Sayers and Hanyaloglu, 2018), Actually, previous studies demonstrated that signal compartmentalization may be critical for the effective induction of the production of sex steroid hormones (Chen et al., 2007).
In conclusion, the present study demonstrates for the first time that S1P may trigger cAMP/PKA-independent activation of pCREB, without inducing intracellular steroidogenic signals and progesterone synthesis in human primary granulosa lutein cells. It remains to be investigated whether this lysosphingolipid may act synergistically with gonadotropins in modulating follicle development.