Background
Sphingosine 1-phosphate (S1P) is a bioactive lipid and G protein-coupled receptor (GPCR) ligand formed within the brain from sphingomyelin and is also present at high levels in blood where it is bound to lipoproteins and stored in erythrocytes [
1‐
5]. There are five S1P receptor subtypes [
6,
7] with S1P
1, S1P
2, S1P
3, and S1P
5 (and in some reports S1P
4) expressed in the CNS [
8‐
13]. Astrocytes are activated in response to CNS injury and diseases like multiple sclerosis (MS) and undergo astrogliosis characterized by increases in proliferation, hypertrophy, and glial fibrillary acidic protein (GFAP) expression [
14‐
20]. S1P induces astrogliosis when injected into the brain as evidenced by increases in GFAP expression and astrocyte proliferation [
21‐
23]. The importance of S1P receptors in disease is highlighted by the widespread acceptance of Fingolimod (FTY720; Gilenya) as a first line oral drug to treat MS [
24‐
27]. Phosphorylated fingolimod functions as an S1P analogue that blocks lymphocyte egress through functional inhibition of S1P
1 signaling [
28,
29]. Its efficacy in the EAE mouse model of MS has also been linked to signaling through S1P
1 on astrocytes [
30].
The predominant S1P receptor subtype detected by quantitative-PCR (q-PCR) in cortical astrocytes is S1P
3, although S1P
1 is also expressed on astrocytes from rat and mouse brain [
8,
12,
31]. The potential importance of S1P
3 signaling in astrocytes is suggested by the finding that this receptor is upregulated in MS lesions and in response to inflammatory stimuli [
32‐
34]. In a mouse model of Sandhoff disease characterized by neuronal death and astrocyte proliferation, deletion of S1P
3, along with the enzyme sphingosine kinase (Sphk) which catalyzes the synthesis of S1P, decreased astrogliosis and disease severity [
35]. Importantly, whereas S1P
1 exclusively couples to the G protein Gα
i, S1P
3 couples promiscuously and its coupling to Gα
12/13 activates the small G-protein RhoA [
36‐
38]. Previous work from our laboratory documented the importance of RhoA activation in inducing astrocyte proliferation, gene expression, and inflammation in response to stimulation of GPCRs for thrombin and S1P [
39‐
44].
Here, we ask whether stimulation of the S1P3 receptor on astrocytes activates RhoA, is responsible for inflammatory gene expression, or can be locally engaged by endogenously formed S1P in an in vitro model of neuroinflammation. We demonstrate that S1P3, and not S1P1, mediates induction of interleukin-6 (IL-6) and vascular endothelial growth factor A (VEGFa) mRNA, and cyclooxygenase-2 (COX-2) mRNA and protein in mouse astrocytes and that this occurs through S1P receptor coupling to Gα12/13 and RhoA. We also demonstrate that simulated inflammation in vitro leads to increases in expression of Sphk1 and S1P3 which could contribute to autocrine inflammatory astrocyte signaling.
Discussion
Neuroinflammation, which underlies many neurodegenerative processes including those involved in Alzheimer’s disease, Parkinson’s disease, and MS, is increasingly recognized as a hallmark of CNS pathology [
19,
20,
55‐
58]. Astrocytes were once considered as structural elements in the brain but subsequently emerged as functionally important for neuronal guidance, maintenance of the BBB, and structural and metabolic support of neurons [
15,
17,
18]. In addition, astrocytes, like microglia, are now known to contribute to neuroinflammation [
14,
17,
19,
56,
59]. The lysophospholipid S1P regulates astrogliosis and inflammatory responses in the CNS; however, the role of the individual S1P receptor subtypes in these processes has not been clearly delineated [
23,
30,
31,
33,
34,
60,
61].
Astrocytes contribute to neuroinflammation by upregulating proinflammatory mediators such us IL-6, MCP-1, TNF-α, iNOS, and COX-2 [
15,
17,
40,
62]. Induction of COX-2 in astrocytes increases generation of reactive oxygen species (ROS), as well as formation of prostanoids that play a prominent role in inflammation, and thus further contribute to neuronal cell death and demyelination in diseases such as MS [
20,
56]. Moreover, astrocytes produce VEGF which plays a role in the breakdown of the blood-brain barrier, a step critical to the entry of pathogenic lymphocytes into the brain [
63‐
67]. Our data demonstrate that an important mechanism for induction of inflammatory cytokines and cytotoxic mediators such as IL-6, COX-2, and VEGFa in astrocytes is through their exposure to S1P and activation of S1P
3.
Both S1P
1 and S1P
3 are expressed on astrocytes [
8,
12] and are upregulated on reactive astrocytes that contribute to inflammation associated with CNS disease [
32,
33,
35,
61,
68]. In response to inflammatory stimuli or in CNS pathologies, Sphk1, an enzyme that generates S1P, is also increased in astroglial cells [
23,
34,
35,
69,
70]. Our findings using siRNA and S1P
3 KO astrocytes demonstrate mechanistically that agonist binding to S1P
3 signals to inflammatory responses through S1P
3 coupling to Gα
12/13 and activation of RhoA. We also show here, using an astrocyte scratch injury assay, that S1P
3 and Sphk1 expression are increased by simulated inflammation and demonstrate by their knockout and downregulation, respectively, that they are involved in an autocrine signaling loop to increase COX-2 expression. While S1P
2 could also signal through Gα
12/13 and RhoA to contribute to COX-2 expression ([
37,
71‐
73] and Fig.
4) and appears to serve this role when S1P
3 is downregulated (Fig.
1b), the relatively low expression of this receptor subtype and its lack of upregulation in response to wounding suggests limited involvement in astrocyte inflammatory responses (Fig.
1c). Thus, it appears that S1P
3, and its autocrine activation by S1P generated through Sphk1, are poised to mediate astrocytic inflammatory responses that could contribute to the progression of CNS neuropathology.
S1P signaling in the CNS has important pathophysiological consequences [
21,
28‐
30,
33‐
35,
40,
61,
74]. Much research has focused on S1P
1 as the primary target for the MS drug FTY720 (fingolimod). While a well-recognized effect of fingolimod is to functionally antagonize S1P
1 receptors on lymphocytes and thereby prevent their egress into the blood and access to the brain, S1P
1 localized to astrocytes contributes significantly to the effects of this drug in an experimental model of MS [
75]. The basis for also considering S1P
3 signaling in MS is that this receptor subtype is upregulated in astrocytes during MS and in EAE and that it is a target for fingolimod [
33,
61]. Notably, fingolimod causes transient bradycardia that appears, at least in the mouse, to be due to its agonist actions on S1P
3 [
24,
25,
76‐
78]. While it is clear that fingolimod downregulates S1P
1, and thus acts as a functional antagonist, its ability to similarly downregulate and thus act as a functional antagonist of S1P
3 is controversial [
61,
77,
79,
80]
. A recent study demonstrated that continuous treatment with FTY20, initiated at the onset of disease in an EAE model, reduced S1P
3 expression at day 22 [
61]. While this indicates that S1P
3 is downregulated by FTY720 treatment, this could reflect reversal of the disease process/inflammation (and its accompanying increases in S1P
3 gene expression) rather than downregulation at the receptor level. Our data with FTY720 (like that examining bradycardia) demonstrate that FTY720 acts as an agonist, eliciting COX-2 induction, over a period of at least 6 h. Our data further establish that it is S1P
3-mediated RhoA signaling, not effects of S1P
1 and Gα
i, that lead to maladaptive astrocyte inflammation. Thus, agonism at astrocyte S1P
3 by fingolimod or other drugs could contribute to neuroinflammation and worsen disease progression, particularly when S1P
3 are upregulated and S1P availability increased through activation of sphingosine kinase. Further studies using S1P
1/3 double knockout mice are ongoing and should indicate whether blocking S1P
3, in addition to S1P
1, would have additional therapeutic benefit.
The importance of S1P
3 and RhoA signaling in CNS disease could be logically extended to consideration of any of the myriad GPCRs found on astrocytes [
32] that couple to RhoA signaling. We and others have shown that PAR1, the receptor for thrombin, couples through RhoA to mediate proliferation and inflammatory responses in astrocytes [
39,
40]. Thrombin is also increased in the injured brain [
81,
82], and an antagonist of protease activated receptor 1 (PAR1) reduces clinical symptoms in EAE mice [
83]. Thus, the evidence that S1P
3 and other GPCRs that stimulate RhoA can contribute to sustained inflammatory responses suggests this pathway as a critical target for blocking neuroinflammation in MS and other CNS diseases.
Acknowledgements
We thank Melissa S. Barlow for her capable assistance with the animal breeding and genotyping and Jeffery M. Smith for the technical assistance.