Sphingosine 1‐phosphate elicits a ROS‐mediated proinflammatory response in human endometrial stromal cells via ERK5 activation

Endometriosis is a chronic gynecological disease affecting ~10% women in the reproductive age characterized by the growth of endometrial glands and stroma outside the uterine cavity. The inflammatory process has a key role in the initiation and progression of the disorder. Currently, there are no available early diagnostic tests and therapy relies exclusively on symptomatic drugs, so that elucidation of the complex molecular mechanisms involved in the pathogenesis of endometriosis is an unmet need. The signaling of the bioactive sphingolipid sphingosine 1‐phosphate (S1P) is deeply dysregulated in endometriosis. S1P modulates a variety of fundamental cellular processes, including inflammation, neo‐angiogenesis, and immune responses acting mainly as ligand of a family of G‐protein‐coupled receptors named S1P receptors (S1PR), S1P1–5. Here, we demonstrated that the mitogen‐activated protein kinase ERK5, that is expressed in endometriotic lesions as determined by quantitative PCR, is activated by S1P in human endometrial stromal cells. S1P‐induced ERK5 activation was shown to be triggered by S1P1/3 receptors via a SFK/MEK5‐dependent axis. S1P‐induced ERK5 activation was, in turn, responsible for the increase of reactive oxygen species and proinflammatory cytokine expression in human endometrial stromal cells. The present findings indicate that the S1P signaling, via ERK5 activation, supports a proinflammatory response in the endometrium and establish the rationale for the exploitation of innovative therapeutic targets for endometriosis.


| INTRODUCTION
Endometriosis is a gynecological disorder defined as the presence of endometrial glands and stroma outside the uterus. 1,2 It is a complex, estrogen-dependent, chronic, inflammatory syndrome that is estimated to affect about 10% of reproductive-age women. 3 Endometriosis lacks definitive treatments because surgery and hormonal drugs are associated with high rates of symptom recurrence. 1,[4][5][6] Extensive investigations on the pathogenesis of this disease 7 are showing the occurrence of complex molecular mechanisms. On the other hand, a key role of inflammation in initiation and progression of the disorder is supported by increased levels of proinflammatory cytokines in endometriotic tissue and in peritoneal fluid of patients with endometriosis. [8][9][10][11][12][13] Sphingosine 1-phosphate (S1P) is a pleiotropic sphingolipid implicated in the regulation of many physiological as well as pathological processes including inflammation, immune response, and tumorigenesis. 14 Its cellular levels are tightly regulated, depending on the relative rate between its biosynthesis and degradation. 15 Most of the S1P biological actions are evoked after its extracellular release through transporters, such as spinster homolog 2 (Spns2), and by its binding to specific G-protein-coupled receptors named S1PR (S1P [1][2][3][4][5] in an autocrine/paracrine manner. 16 Interestingly, S1P signaling axis has been recently reported to be altered in endometriosis, [17][18][19][20][21] and S1P levels have been shown to be significantly augmented in the peritoneal fluid of endometriotic patients. 19 Moreover, the inhibition of sphingosine kinase (SK) 1, one of the two isoenzymes responsible for S1P generation, significantly reduces the development of endometriotic lesions in an animal model of the disease. 21 The therapeutic potential of the S1P axis has been already efficiently exploited. For instance, several S1PR selective antagonists have been developed and used in clinical settings, such as the S1P 1 antagonist KRP-203 in subacute lupus erythematosus, and the immunomodulator fingolimod, a functional antagonist of S1P 1,3,4,5 , in multiple sclerosis and other autoimmune and inflammatory disorders. 22,23 The extracellular signal-related kinase 5 (ERK5) belongs to the mitogen-activated protein kinase (MAPK) family. In response to growth factors and cellular stresses, ERK5, activated by MEK5, can orchestrate gene transcription. 24 Like other MAPK, ERK5 can also be activated by G-protein-coupled receptors (GPCR). 25,26 ERK5 is involved in several biological responses, including cell survival, anti-apoptotic signaling, angiogenesis, differentiation and proliferation of several cell types. 24 In recent years, the ERK5 pathway has emerged as a key player in the onset and development of chronic inflammatory diseases as well as cancers, including endometrial cancer, [27][28][29] and, with the aim to dampen the negative effects of this signaling axis, many small-molecule inhibitors of ERK5 or of its upstream activator MEK5 have been developed.
The present study evaluated the functional interaction between S1P signaling and ERK5 in human endometrial stromal cells and its effect on the inflammatory process.

| Biological sample collection
Tissue samples were collected at Careggi University Hospital, Florence, Italy. Endometriotic lesions were from patients with endometriosis undergoing laparoscopic surgery. Control endometrial specimens were collected during diagnostic hysteroscopy (in the proliferative phase) from nonpregnant women not affected by endometriosis or other uterine disorders. The clinical and imaging investigations excluded endometriosis and other uterine disorders. All samples were histologically characterized. The endometrial cycle phase was confirmed by histologic analysis of endometrial biopsies. There was no difference in age, gravidity, and parity between the study and control groups. The patients stopped hormonal treatment at least 3 months before surgery. The institutional review board (protocol no. 13742) approved the study protocol, and all patients gave informed written consent.

| Cell transfection
Lipofectamine RNAiMAX (Invitrogen, Carlsbad, CA) was used for cell transfection according to the manufacturer's instructions as previously described. 34,35 Lipofectamine RNAiMAX was incubated with siRNAs (siRNA universal negative control SIC001; siS1P 1 SASI_HS_0024; siS1P 2 SASI_HS01_0008; siS1P 3 SASI_HS01_0001, Sigma Aldrich RRID:SCR_008988; siERK5, M0035130200 Dharmacon, Thermo Fisher Scientific Inc.) in DMEM:F12 (1:1) without serum and antibiotics at room temperature for 20 min and then added to cells to a final concentration of 50 nmol/L, in DMEM:F12 (1:1) containing serum. After 56 h, cells were serum-starved overnight and were used for experiments 72 h after the beginning of transfection. The efficacy of target down-regulation was evaluated with the use of quantitative real-time polymerase chain reaction or Western blot analysis. The pcMV5-MEK5DD-HA (a constitutively active form of MEK5) was generously provided by Jiing-Dwan Lee (Scripps Institute, La Jolla, CA, USA). HESC were plated (3 × 10 4 cells/well) and transfected after 24 h with a total amount of 2 μg of plasmid DNA using Lipofectamin2000 (Invitrogen, Thermo Fisher Scientific), following manufacturer's instructions. Cells were used for experiments 24 h after the beginning of transfection.

| Confocal microscopy
Cells were seeded on a glass coverslip and, at the end of treatment, incubated with the probe chloromethyl derivative of 2′,7′-dichlorodihydrofluorescein diacetate (CM-H 2 DCFDA; Thermo Fisher Scientific) 5 μM for 20 min, as an indicator for reactive oxygen species (ROS) in cells. Fluorescence was assessed with confocal microscopy (Leica SP8) and quantified using the ImageJ software (RRID:SCR_003070, National Institutes of Health, USA).

| Flow cytometry
After treatment, cells were incubated with CM-H 2 DCFDA 1 μM for 20 min. Cells were then detached with Trypsin-EDTA solution (Sigma-Aldrich) and cell fluorescence was measured with a FACSCanto flow-cytometer (Becton-Dickinson, Franklin Lakes, NJ, USA).

| Statistical analysis
Statistical analysis and graphical representations of the data were performed using the GraphPad Prism 5 software (RRID:SCR_002798). P values were calculated using Student's t test (two groups), one-way analysis of variance ANOVA (more than two groups) followed by Bonferroni post hoc test or two-way ANOVA followed by Bonferroni post hoc test. p < .05 was considered statistically significant.

| S1P stimulation increases ROS production in human endometrial stromal cells
Although S1P metabolism and signaling are dysregulated in endometriosis, 17,18 the molecular mechanisms implicated in the action of this sphingolipid in the pathogenesis of the disease have not been profoundly investigated so far. To deepen this issue, we first investigated whether human endometrial stromal cells (HESC) express the enzymes involved in S1P metabolism and the S1PR (Figures 1 and S1). qPCR analysis showed that both isoforms of sphingosine kinase (SK), SK1 and SK2, and the enzymes involved in S1P catabolism, S1P lyase (SPL) and S1P phosphatase (SPP)1, but not SPP2, were expressed in HESC ( Figure 1A). In addition, HESC expressed the SK1activating protein CIB1 ( Figure 1A). At the mRNA level, the S1P transporter Spns2 and all S1PR were expressed, being S1P 3 the most expressed receptor ( Figure 1B). The F I G U R E 1 S1P metabolism enzyme, transporter, and S1PR expression levels in HESC. HESC were seeded in 60 mm dishes at 70% confluence. qPCR analysis was performed using TaqMan gene expression assay probes specific for human SK1, SK2, CIB1, SPL, SPP1, and SPP2 (A) and S1P 1-5 and Spns2 (B). Results, analyzed with the 2 −ΔΔCt method, were normalized to βactin housekeeping gene and expressed as fold changes over a reference gene, SK1 (A), and S1P 1 (B). nd, not detected. expression of SK1, SK2, SPL, Spns2 as well as S1P 1 /S1P 2 / S1P 3 was confirmed at the protein level by Western blot analysis ( Figure S1). Given the key role of ROS in the progression of inflammatory disorders, we investigated whether S1P was able to induce ROS formation in HESC. Intracellular ROS levels were detected by confocal immunofluorescence analysis performed using CM-H 2 DCFDA. We found that the treatment with 100 nM S1P for 10 min potently increased ROS formation in HESC (Figure 2A). In agreement, cytofluorimetric ROS quantification highlighted a significant increase of about 40% following 10 min treatment with 100 nM S1P ( Figure 2B).

| S1P activates ERK5 through S1P 1 and S1P 3 receptor engagement and downstream activation of SFK and MEK5
We then examined whether, in HESC, S1P was capable of activating ERK5, a pivotal MAPK involved in ROS modulation, 40 expressed in endometrial tissues and endometriotic lesions as determined by qPCR ( Figure 3A). To this end, HESC were treated with 100 nM S1P for different time intervals (from 5 to 60 min) before being lysed and subjected to Western blot analysis. ERK5 activation was then evaluated on the basis of the presence of a bandshift due to decreased ERK5 electrophoretic mobility following extensive phosphorylation as the consequence of MEK5-dependent phosphorylation in the conserved threonine-glutamic acid-tyrosine (TEY) motif of the catalytic domain and MEK5-independent phosphorylation at the C-terminus. 27,41 As depicted in Figure 3B, S1P transiently activated ERK5 with a maximal overall phosphorylation at 10 min of treatment. In particular, treatment with the sphingolipid determined MEK5-dependent phosphorylation, as witnessed by increased phosphorylation of the TEY motif. ERK5 activation was confirmed by the increased p90RSK phosphorylation and MEF2D expression that are ERK5-regulated downstream targets. 42,43 Dose-dependence studies showed that ERK5 activation was induced by S1P at all the tested concentrations (10-1000 nM) being maximal starting from 100 nM ( Figure 3C). Accordingly, S1P induced a dose-dependent increase of ERK5 phosphorylation on TEY motif, of p90RSK phosphorylation and of MEF2D expression.
We next examined the possible S1PR involved in S1Pinduced ERK5 activation. For this purpose, S1P 1 , S1P 2 , and S1P 3 were efficiently knocked-down by specific siRNA ( Figure 4A). Moreover, to further verify that S1PR silencing affected their functional activity, the activation of known S1PR downstream targets 37,44 has been analyzed. In particular, while S1P 1 and S1P 3 silencing significantly reduced S1P-induced ERK1/2 phosphorylation, that of S1P 2 inhibited the S1P-induced ERM activation in HESC ( Figure S2). Notably, as shown in Figure 4B, siRNA directed against S1P 1 and S1P 3 significantly reduced the enhancement of ERK5 activation elicited by S1P. Conversely, the down-regulation of S1P 2 did not alter the S1P-induced effect ( Figure 4B). To confirm the role of S1P 1 and S1P 3 in S1P-induced ERK5 activation, pharmacological inhibition F I G U R E 2 S1P stimulates ROS production. (A) HESC seeded on microscope slides were serum-starved for 18 h before being stimulated with 100 nM S1P for 10 min. Upper panel: images were obtained by confocal microscopy analysis using CM-H 2 DCFDA to stain ROS (images are representative of three independent experiments). Lower panel: quantification of the intensity of CM-H 2 DCFDA fluorescence normalized to cell number. Data are mean ± SEM of four fields quantified in three independent experiments. S1P increases ROS formation in a statistically significant manner by t-test **p < .01. (B) HESC were serum-starved for 18 h before being stimulated with 100 nM S1P for 10 min. ROS were quantified by cytofluorimetric analysis using CM-H 2 DCFDA. Results are reported as mean ± SEM of eight independent experiments, fold change over control set as 1. S1P increases ROS formation in a statistically significant manner by t-test (***p < .001). MFI, mean fluorescence intensity. of both S1PR was performed using the S1P 1/3 antagonist VPC23019. As shown in Figure 4C, the activation of ERK5, induced by S1P, was prevented by preincubation with VPC23019. All together, these data support the view that S1P activates ERK5 through S1P 1 and S1P 3 engagement.
To gain insight into the implicated molecular mechanism, we considered pathways that are known to be activated by G-protein-coupled receptors and, in turn, are involved in ERK5 activation such as SRC-family kinases (SFK), MEK5, and ERK1/2. 28 As reported in Figure 5A, 10 min stimulation with 100 nM S1P was responsible for MEK5, SRC, and ERK1/2 activation in HESC. Subsequently, HESC were pretreated with the MEK5 inhibitor BIX02189 (5 μM), the SFK inhibitor PP2 (10 μM), or the ERK1/2 inhibitor SCH772984 (1 μM). As shown in Figure 5B the band-shift of ERK5, induced by S1P, was prevented by preincubation with BIX02189 or PP2, but not by SCH772984, demonstrating that the bioactive sphingolipid activates ERK5 through SFK and MEK5.

| ERK5 is involved in S1P-dependent intracellular ROS formation
We next examined the potential role of S1P-dependent activation of ERK5 in the induction of intracellular ROS formation. For this purpose, HESC were treated with 5 μM XMD8-92 or 5 μM JWG-071, pharmacological inhibitors of ERK5, before being challenged with 100 nM S1P for 10 min. Confocal fluorescence analysis ( Figure 6A) as well as cytofluorimetric quantification ( Figure 6B) clearly demonstrated that the enhanced formation of ROS induced by S1P was completely reversed by ERK5 inhibition. To further confirm the role of ERK5 in the modulation of ROS levels evoked by S1P, ERK5 was knocked-down by RNA interference. The specific siRNA treatment efficaciously reduced basal expression levels of ERK5 ( Figure S3), and significantly reduced S1P-induced intracellular ROS increase measured by cytofluorimetric analysis ( Figure 6C). Of note, activation of ERK5 by the expression of a constitutively active form of MEK5 (MEK5DD) significantly increased intracellular ROS, confirming the role of MEK5-dependent phosphorylation of ERK5 in the modulation of ROS levels in HESC ( Figure S4).
In order to demonstrate whether S1P-induced ROS production is a S1PR-mediated event, we down-regulated by RNA interference S1P 1 and S1P 3, previously shown to be F I G U R E 3 S1P activates ERK5. (A) Quantitative mRNA analysis was performed by qPCR in total RNA extracted from healthy endometrium (n = 2) and endometriotic lesions (n = 2). Results, analyzed with the 2 −ΔCt method, were normalized to 18S housekeeping gene. HESC were serum-starved for 18 h and then treated for different time intervals (5-60 min) with 100 nM S1P (B) or with S1P at the indicated concentrations (10-1000 nM) for 10 min (C). Western blot analysis was performed using anti-ERK5, −pERK5 T218/Y220 , −pp90RSK S380 , -MEF2D. Migration of molecular weight markers is indicated on the left (kDa). Upper panels: blot representative of at least three independent experiments with analogous results is shown. Lower panels: densitometric analysis of at least three independent experiments. Data are fold change over control (set as 1) of the mean ± SD of the ratio between the slower migrating upper ERK5 band (indicated by the arrow) and the lower ERK5 band normalized to vinculin. One-way ANOVA followed by Bonferroni's post hoc test (*p < .05 **p < .01, treated vs. control). involved in the activation of ERK5 elicited by the sphingolipid ( Figure 4B,C). Cytofluorimetric analysis demonstrated that S1P-induced intracellular ROS increase was significantly reduced by siRNA directed against S1P 1 or S1P 3 ( Figure 6D) or pharmacological blockade of S1P 1/3 using VPC23019 ( Figure 6E).

| S1P induces IL-1β and IL-6 expression via ERK5 activation and ROS formation
Then, we examined whether S1P was able to induce a proinflammatory response in endometrial cells by increasing the synthesis of cytokines. To this end, the mRNA expression of IL-1β and IL-6 was examined in HESC treated with 100 nM S1P for 6 h. qPCR analysis demonstrated that the sphingolipid increased the expression of the above proinflammatory cytokines ( Figure 7A). This effect was impaired by S1P 1/3 antagonism, using VPC23019 ( Figure 7A), pointing to the involvement of these receptors in S1P-elicited increase of IL-1β and IL-6. In order to establish the potential role of S1P-dependent ERK5 activation in the proinflammatory action of the sphingolipid, HESC were pretreated for 1 h with the ERK5 inhibitor XMD8-92 (5 μM) or transfected with specific siRNA for ERK5 before being challenged with 100 nM S1P for 6 h. qPCR analysis demonstrated that both pharmacological ( Figure 7B) and genetic ( Figure 7C) inhibition of ERK5 significantly diminished the S1P-dependent increase of IL-1β and IL-6. The same effects were elicited upon pretreatment with the ROS scavenger NAC (5 mM) ( Figure 7D), thus providing evidence of the involvement of ROS in the S1P-induced production of the above proinflammatory cytokines. Interestingly, pharmacological inhibition of ERK5 strongly reduced the basal expression levels of the analyzed proinflammatory cytokines ( Figure 7B), in line with the data shown in Figure S4 and with previous reports 40, 45 and further supporting a crucial role for this MAPK in the regulation of the inflammatory response in the endometrium. F I G U R E 4 S1P activates ERK5 via S1P 1 and S1P 3 . (A) Relative quantitative mRNA analysis was performed by qPCR in HESC transfected with nonspecific siRNA (SCR) or with siRNA specific for S1P 1 or S1P 2 or S1P 3 ; the content of housekeeping gene β-Actin was analyzed in parallel. Results are expressed as fold changes according to the 2 −ΔΔCt method, utilizing as calibrator each receptor subtype in cells transfected with SCR siRNA. Data are mean±SEM of three independent experiments performed in triplicate. The effect of S1P 1 -S1P 2 and S1P 3 -siRNA transfection is statistically significant by t-test *p < .05, **p < .01. (B) HESC transfected with SCR-, S1P 1 -S1P 2 -and S1P 3 -siRNA were serum-starved prior to be stimulated with 100 nM S1P for 10 min. ERK5 activation was measured in total cell lysate by Western blotting using anti-ERK5 antibody. Migration of molecular weight markers is indicated on the left (kDa). Data are fold change over control (set as 1) of the mean ± SD of the ratio between the slower migrating upper ERK5 band (indicated by the arrow) and the lower ERK5 band normalized to vinculin. A blot representative of three independent experiments is shown. S1P activates ERK5 in a statistically significant manner (*p < .05). The effect of S1P 1 -or S1P 3 -downregulation on S1P-induced ERK5 activation is statistically significant by two-way ANOVA followed by Bonferroni's post hoc test ( # p < .05). (C) Serum-starved HESC were pretreated with the pharmacological inhibitor of S1P 1/3 VPC23019 (VPC, 10 μM) for 1 h before being challenged with 100 nM S1P for 10 min. ERK5 activation was measured in total cell lysate by Western blotting using anti-ERK5 antibody. Migration of molecular weight markers is indicated on the left (kDa). A blot representative of three independent experiments with analogous results is shown. The histogram represents the densitometric analysis of three independent experiments; data are the mean ± SD of the ratio between ERK5 up-shifted band/ERK5 normalized to vinculin, fold change over control (set as 1). S1P activates ERK5 in a statistically significant manner (**p < .01). The effect of S1P 1/3 blockade by VPC23019 on S1P-dependent ERK5 activation is statistically significant by two-way ANOVA followed by Bonferroni's post hoc test ( ### p < .001).

| DISCUSSION
Inflammation is one of the key mechanisms in the pathophysiology of endometriosis, involved in lesion formation, fibrosis, and infertility. 3,4,[46][47][48] The investigation of the molecular mechanisms involved in the pathogenesis of the inflammatory process could open new perspectives to identify novel therapeutic strategies for endometriosis. The findings presented here show that the bioactive sphingolipid S1P is responsible for the activation of ERK5, and the identified S1P/ERK5 axis emerged to be crucial in supporting a ROS-mediated proinflammatory effect in endometrial cells. Several lines of evidence support the role of oxidative stress in the development and progression of endometriosis by causing a general inflammatory response in the peritoneal cavity. 49,50 Along this line, S1P has been largely involved in both acute and chronic inflammation, by regulating immune cell recruitment. 51 In human intraperitoneal macrophages, S1P provoked a shift toward an M2-dominant condition and an increased expression of IL-6 and COX2, thus determining an extension of the endometriotic lesion size. 19 In the present study, we demonstrated that S1P stimulation increases ROS production in endometrial cells, thus pointing to the existence of a link between S1P and oxidative stress in endometriosis. This is in agreement with a previous study performed in human pulmonary alveolar epithelial cells showing that S1P increased the inflammatory response through a NADPH oxidase/ROS-dependent NF-κB activation mechanism. 52 Moreover, in bone marrow-derived macrophages, S1P/ S1P 2 signaling axis was responsible for ROS production and NLRP3 inflammasome activation. 53 Thus, our data, in keeping with other reports, 52-54 support the evidence that S1P is a lipid engaging a redox-based signaling. Here, the identification of S1P 1/3 as responsible for ROS generation in endometrial cells in response to S1P treatment further highlights that the signaling mediated by S1PR is cell-specific.
Many studies have emphasized the importance of MAPKs in the development and progression of endometriosis. [55][56][57] Additionally, both in vitro and in vivo investigations have shown that activation of the MEK5/ ERK5 pathway is involved in oxidative stress-elicited effects. 40,[58][59][60] We here reported for the first time that S1P induces ERK5 activation in endometrial cells and that both pharmacological and genetic inhibition of ERK5 reverted the S1P-dependent ROS production and proinflammatory cytokine expression. Very little literature evidence is available on the ability of S1P to activate ERK5. The sphingolipid has been reported to activate this MAPK kinase in mouse embryonic fibroblasts in PTX-independent manner 61 and in HEK293 overexpressing ERK5. 62 Interestingly, in the present study, ERK5 was found to be expressed in endometriotic lesions.
Another interesting finding illustrated here is the identification of the molecular mechanism implicated in S1Pinduced ERK5 activation. Indeed, we demonstrated that S1P 1 and S1P 3 mediate S1P-dependent increase in ERK5 activation.

F I G U R E 5 S1P activates ERK5 via SFK and MEK5. (A) HESC
were serum-starved for 18 h and then treated for 10 min with 100 nM S1P. Western blot analysis was performed using antibodies for the phosphorylated forms of MEK5, SRC, and ERK1/2 in cell lysates. Migration of molecular weight markers is indicated on the left (kDa). A blot representative of two independent experiments with analogous results is shown. (B) Serum-starved HESC were pretreated with 5 μM BIX02189, selective inhibitor of MEK5, or 10 μM PP2, specific inhibitor of SFK, or 1 μM SCH772984, specific inhibitor of ERK1/2, for 1 h before being challenged with 100 nM S1P for 10 min. ERK5 activation was measured in total cell lysate by Western blotting using anti-ERK5 antibody. Migration of molecular weight markers is indicated on the left (kDa). A blot representative of three independent experiments with analogous results is shown. The histogram represents the densitometric analysis of three independent experiments; data are the mean ± SD of the ratio between ERK5 up-shifted band/ERK5 normalized to vinculin, fold change over control (set as 1). S1P activates ERK5 in a statistically significant manner (**p < .01). The effect of SFK inhibition by PP2 or MEK5 inhibition by BIX02189 on S1P-dependent ERK5 activation is statistically significant by two-way ANOVA followed by Bonferroni's post hoc test ( # p < .05, ## p < .01).
BIX02189 5µM PP2 10 µM SCH772984 1µM S1P 100 nM - Data are mean ± SEM of four fields quantified in three independent experiments. S1P increases ROS formation in a statistically significant manner (***p < .001). The effect of ERK5 inhibition by XMD8-92 or JWG-071 on S1P-induced ROS formation is statistically significant by twoway ANOVA followed by Bonferroni's post hoc test ( ### p < .001). HESC were serum-starved for 18 h and pretreated or not with the ERK5 inhibitors XMD8-92 (XMD) 5 μM or JWG-071 (JWG) 5 μM (B) or the pharmacological inhibitor of S1P 1/3 VPC23019 (VPC, 10 μM) (E) for 1 h before being stimulated with 100 nM S1P for 10 min. HESC, transfected with SCR-or ERK5-siRNA (C) or with SCR-, S1P 1 -and S1P 3 -siRNA (D), were serum-starved prior to be stimulated with 100 nM S1P for 10 min. ROS were quantified by cytofluorimetric analysis using CM-H 2 DCFDA. Results are reported as mean ± SEM of an experiment representative of four, performed in duplicate, fold change over control set as 1. S1P increases ROS formation in a statistically significant manner (*p < .05; **p < .01). The effect of ERK5 pharmacological inhibition by XMD8-92 or JWG-071 (B) or ERK5 silencing (C) or S1P 1 -or S1P 3 -silencing (D) or S1P 1/3 blockade by VPC23019 (E) on S1P-induced ROS formation is statistically significant by two-way ANOVA followed by Bonferroni's post hoc test ( # p < .05; ## p < .01; ### p < .001). MFI, mean fluorescence intensity. Although it is known that intracellular S1P can evoke inflammatory responses mainly by NF-κB activation, recent studies support the involvement of S1PR and thus of an extracellular pool of S1P in this process. The involvement of S1P 1/3 in the induction of IL-6, COX-2 as well as prostacyclin was also demonstrated in endothelial cells where S1P cooperates with LPS in leukocyte adhesion. 63 Moreover, in different cancer cell types, S1P 1 induces persistent STAT3 activation and IL-6 expression. 64 S1P 1 but also S1P 2 was instead found to mediate NF-κB activation and IL-6 expression in melanoma cells, 65 while S1P 2 was also implicated in the induction of IL-6 expression in endometriotic stromal cells derived from endometrioma. 20 In this work, the involvement of S1P 2 in ERK5 activation was excluded since when the receptor was silenced by RNA interference the effect of the sphingolipid was not significantly altered. Crucially, it was previously shown that a deep remodeling of S1PR occurs at the level of endometriotic lesions when compared with healthy endometrium controls. 17,18 S1P has been previously reported to induce SRC activation, in turn responsible for NADPH oxidase activation and H 2 O 2 production in NIH/3T3 fibroblasts as well as COX-2 expression and PGE2/IL-6 secretion in human tracheal smooth muscle cells. 66 Moreover, SRC has been extensively demonstrated to be an upstream activator of the MEK5/ERK5 module in many cell types. 24 In line with the above findings, we show here that the bioactive F I G U R E 7 S1P induces IL-1β and IL-6 expression via ERK5 activation and ROS formation. HESC were serum-starved for 18 h and pretreated or not with the S1P 1/3 antagonist VPC23019 (VPC, 10 μM) (A) or the ERK5 inhibitor XMD8-92 (XMD, 5 μM) (B) or the ROS scavenger NAC (5 mM) (D) for 1 h before being stimulated with 100 nM S1P for 6 h. HESC transfected with SCR-or ERK5-siRNA (C) were serum-starved prior to be stimulated with 100 nM S1P for 6 h. Quantitative analysis of IL-1β and IL-6 mRNA was performed by real-time PCR. Results are expressed as fold changes according to the 2 −ΔΔCt method, using individual cytokines of the unchallenged specimen as calibrator. S1P increases IL-1β and IL-6 expression in a statistically significant manner (*p < .05; **p < .01). The effect of S1P 1/3 blockade by VPC23019 (A) or ERK5 inhibition by XMD8-92 (B) or ERK5-downregulation by specific siRNA (C) or ROS scavenging by NAC (D) on S1P-induced cytokine mRNA levels was statistically significant by twoway ANOVA followed by Bonferroni's post hoc test ( # p < .05, ## p < .01, ### p < .001). sphingolipid activates ERK5 through SFK. In this respect, it is worth noting that despite ERK5 inhibition almost completely impaired S1P-induced ROS production, we may not exclude that activation of the SFK/MEK5/ERK5 signaling is ROS-dependent. 40,67,68 On the other hand, our data indicate that S1P-induced ERK5 activation is mediated by MEK5, in line with the well-established notion that ERK5 activation is classically operated by its upstream MAPKK, MEK5, that has ERK5 as its only known substrate. 69 To our knowledge, this is the first evidence of MEK5 activation by the bioactive sphingolipid. At variance with what is reported in certain cellular models, 32,70 ERK1/2 does not seem to be involved in S1P-mediated ERK5 activation in endometrial cells. Proinflammatory mediators are involved in the onset and worsening of endometriosis. 71 In the present work, we found that S1P increases the expression of IL-6 and IL-1β in endometrial cells and that ERK5 inhibition abrogated this increase, thus supporting the hypothesis of the involvement of ERK5 in S1P-mediated regulation of the inflammatory response in the endometrium. These findings are in keeping with the fact that serum IL-6 has been identified in high concentrations in endometriosis 72 and that the level of its expression has been linked to the disease severity. 73 Moreover, the expression of IL-1β, which has been associated with the risk of endometriosis, 74 appears to depend on the stage of the disease and stimulates the upregulation of biomolecules that induce pain. 75 In conclusion, our findings support the view that the S1P signaling via ERK5 activation supports a proinflammatory response in the endometrium and establish the rationale for the exploitation of S1P/ERK5 targeting as an innovative therapeutic approach in endometriosis.

AUTHOR CONTRIBUTIONS
CB, CD ER, and IT conceived and designed the research. AM, CB, IS, IT, and MP performed the research and acquired the data. FP and SV collected biological samples; FCa performed the histological characterization of the biological samples. CB, CD, ER, FCe, IS, and IT analyzed and interpreted the data. CB, CD, ER, IS, and IT wrote the manuscript. FP and PB revised the paper.

ACKNOWLEDGMENTS
We took advantage of the newly established Molecular Medicine Facility of the Department of Experimental and Clinical Biomedical Sciences "Mario Serio" at the University of Florence. The Facility was supported by a grant of the Italian Ministry of Education University and Research (MIUR) after the Department was awarded as one of the 180 Departments of Excellence in Italy.

FUNDING INFORMATION
The work was supported by Fondi di Ateneo (ex 60%) to CB, CD, ER, FCe, FP, and PB, by MIUR, Progetto Medicina di Genere to CD, by Fondazione Careggi, Project on Woman's Health, by Fondazione Cassa di Risparmio di Firenze to FP, by Associazione Italiana per la Ricerca sul Cancro (AIRC) to ER and by Università degli Studi di Firenze ("Progetto Competitivo di Ateneo per Ricercatori" funded by MUR and Next Generation EU) to ER.