Signal Transduction and Gene Regulation in the Endothelium

Abstract

Extracellular signals act on G-protein-coupled receptors (GPCRs) to regulate homeostasis and adapt to stress. This involves rapid intracellular post-translational responses and long-lasting gene-expression changes that ultimately determine cellular phenotype and fate changes. The lipid mediator sphingosine 1-phosphate (S1P) and its receptors (S1PRs) are examples of well-studied GPCR signaling axis essential for vascular development, homeostasis, and diseases. The biochemical cascades involved in rapid S1P signaling are well understood. However, gene-expression regulation by S1PRs are less understood. In this review, we focus our attention to how S1PRs regulate nuclear chromatin changes and gene transcription to modulate vascular and lymphatic endothelial phenotypic changes during embryonic development and adult homeostasis. Because S1PR-targeted drugs approved for use in the treatment of autoimmune diseases cause substantial vascular-related adverse events, these findings are critical not only for general understanding of stimulus-evoked gene regulation in the vascular endothelium, but also for therapeutic development of drugs for autoimmune and perhaps vascular diseases.

SPHINGOSINE 1-PHOSPHATE

Lipids are a source of energy and a major structural element of membranes. However, vertebrates have evolved to use metabolites of phospholipids as extracellular signaling molecules (de Carvalho and Caramujo 2018). Among the plethora of bioactive lipids, lysophospholipids such as lysophosphatidic acid (LPA) and sphingosine 1-phosphate (S1P) are the most extensively characterized in mammals. Extracellular S1P circulates in blood (∼1 µM) and lymph (∼0.1 µM), whereas its concentration is much lower in interstitial fluid of tissues (<1 nM) (Schwab et al. 2005; Hla et al. 2008). This sharp gradient of S1P across the vascular barrier allows S1P to regulate immune cell trafficking and inflammatory response in a spatially precise manner (Yanagida and Hla 2017). S1P is also a key regulator of vascular development, maturation, and barrier function. More specifically, S1P outside of the cell binds and activates cell membrane receptors that transmit signals inside the cell. The downstream signaling cascade then regulates various cellular functions such as adhesion, cytoskeleton organization, inflammation, migration, and proliferation, thus ultimately impacting organ-specific endothelial cell (EC) functions. In this review, we will present an overview of the S1P signaling pathway with a focus on regulation of gene expression in the vascular system.

S1P Metabolism

S1P metabolic pathways have been extensively reviewed in the literature (Blaho and Hla 2011; Cantalupo and Di Lorenzo 2016; Lai et al. 2016; Hannun and Obeid 2018). Briefly, S1P originates either from the breakdown of sphingomyelin or from the de novo pathway that combines serine and palmitoyl-CoA. Both pathways lead to the key intermediate, ceramide, which is further degraded to sphingosine by ceramidase enzymes. Sphingosine is then phosphorylated by sphingosine kinases (SPHK1 and SPHK2) to generate S1P. Finally, S1P can be converted back to sphingosine by the activity of the S1P phosphatase or irreversibly degraded to hexadecenal and phosphoethanolamine by the enzymatic activity of the S1P lyase. The level of S1P is determined by the equilibrium between the synthesis and the degradation of S1P, which can be altered by extracellular or cell-intrinsic signals.

S1P is membrane impermeant, thus requiring the action of a transporter for extracellular export. The two validated S1P transporters are SPNS2 and MFSD2B. Red blood cells use the activity of SPHK1 and MFSD2B to provide more than 50% of the plasma circulating S1P (Vu et al. 2017) and increase S1P at sites of thrombosis (Chandrakanthan et al. 2021). Under homeostatic conditions, EC SPNS2 (Fang et al. 2021) contributes to ∼20%–30% of plasma S1P (Fukuhara et al. 2012; Mendoza et al. 2012; Xiong et al. 2014) and ∼80% of lymph S1P (Mendoza et al. 2012). Thus, erythrocytes are the primary source of S1P circulating in plasma while the lymphatic endothelium is the major contributor of lymph S1P and a minor contributor of plasma S1P.

S1P Chaperones

The low water solubility of S1P requires it to be bound to proteins to circulate in the extracellular milieu. Carrier proteins bind S1P and allow circulation at high concentrations (Yanagida and Hla 2017). These carriers also protect the phosphate group of S1P from extracellular phosphatases and enable efficient and precise activation of S1P receptors to generate ligand-dependent biological responses. Thus, extracellular S1P-binding proteins have been designated as “S1P chaperones.” The two major chaperones are apolipoprotein M (APOM) and albumin (ALB). In plasma, ∼65% of S1P is bound to APOM present on a subpopulation (∼5%) of high-density lipoprotein (HDL) particles, while ∼30% of S1P is bound to ALB (Murata et al. 2000; Aoki et al. 2005; Christoffersen et al. 2006, 2011). The plasma S1P was reduced by ∼45% in Apom knockout (KO) mice and was absent from the HDL fraction (Christoffersen et al. 2011). Interestingly, in mice missing both Apom and Alb genes, ∼25% of S1P remains in circulation compared to wild-type animals (Obinata et al. 2019). In this condition, S1P associates with a third chaperone, apolipoprotein A4 (APOA4), likely due to a compensatory mechanism that maintains essential S1P signaling required for embryonic development.

Fundamental differences exist between the two chaperone-bound pools of circulating S1P. S1P binds to APOM with a higher affinity compared to ALB (Fleming et al. 2016). In fact, APOM buries S1P in a binding pocket that shields S1P from S1P-degrading enzymes (Sevvana et al. 2009; Christoffersen et al. 2011), resulting in a stable S1P pool in vivo. Contrarily, ALB associates with S1P on its surface via low affinity (perhaps high capacity) interactions thus resulting in transient receptor activation and short half-life of ALB-S1P in circulation (Venkataraman et al. 2008). Chaperone-dependent S1P association results in different receptor activation modes. For example, Apom KO mice still have 30%–40% of the normal S1P level in plasma, but showed increased lung vascular leakage due to reduced endothelial sphingosine 1-phosphate receptor 1 (S1PR1) signaling, suggesting that ALB-S1P was unable to fully compensate the lack of APOM-S1P in vivo (Christoffersen et al. 2011; Christensen et al. 2016). In addition, HDL particles containing APOM-S1P exhibit a biased signaling mechanism that suppresses inflammatory reactions (Galvani et al. 2015). In other words, it generates a unique cellular response while stimulating the same receptor as ALB-S1P (biased signaling of G-protein-coupled receptors [GPCRs] by agonists is briefly described in the section GPCR Signaling and Gene Regulation and in Wang et al. 2018b; Fernandez et al. 2020; Wingler and Lefkowitz 2020).

S1P Receptors

The extracellular S1P generates various physiological effects through activation of cell-surface GPCRs. In vertebrates, there are five known high-affinity S1P receptors, namely, S1PR1 to S1PR5 (Sanchez and Hla 2004; Chun et al. 2010). The S1PRs present different individual expression signatures and signaling pathways with multiple overlapping elements that define the diverse and redundant S1P functions (Hisano et al. 2012; Blaho and Hla 2014; Cannavo et al. 2017; Jozefczuk et al. 2020). S1PR1-3 are the predominant S1P receptors in the vascular system. S1PR1 signals through Gαi/o, while S1PR2 and S1PR3 can signal using Gαi/o, Gα12/13, and Gαq. S1PR4, mostly expressed in lymphoid tissues, and S1PR5, predominantly expressed in nervous (oligodendrocytes) and immune systems (natural killer cells), both use Gαi/o and Gα12/13 to transmit signals inside the cells. S1PR5 may regulate the blood–brain barrier (BBB) functions in addition to its role in oligodendrocytes (Di Pardo et al. 2018).

GPCR SIGNALING AND GENE REGULATION

G-protein-coupled receptors (GPCRs) are a well-studied class of proteins with approximately 800 members encoded in the human genome. Drugs targeting GPCRs represent approximately one-third of all FDA-approved drugs (Schiöth and Fredriksson 2005; Hauser et al. 2017). GPCR ligands are astonishingly diverse: amines, ions, lipids, nucleotides, odorants, peptides, photons, or proteins (Fredriksson et al. 2003). The intracellular GPCR signaling cascades involve the activation of heterotrimeric G proteins Gα, Gβ, and Gγ (Fig. 1; Downes and Gautam 1999). The interaction of the GPCR with its ligand changes the receptor conformation, which triggers the dissociation of Gα from the Gβγ dimer (Hilger et al. 2018). The Gα proteins are subdivided into four main categories (i.e., Gαs, Gαi/o, Gαq/11, and Gα12/13) and target different downstream effectors such as adenylyl cyclases, cGMP phosphodiesterase, phospholipase C, and RhoGEFs, respectively (Campbell and Smrcka 2018; Hilger et al. 2018). On the other hand, the Gβγ dimer can also transmit downstream signals (Khan et al. 2013; Tennakoon et al. 2021). In addition to these classical GPCR signaling pathways, several other signaling mechanisms have been described—for example, β-arrestin-dependent biased signaling, receptor dimerization, and transactivation of other signaling mechanisms, etc. (for reviews, see Wang et al. 2018b; Fernandez et al. 2020; Wingler and Lefkowitz 2020).

Figure 1.

G-protein-coupled receptors (GPCRs) signaling cascade. The GPCRs associate with heterotrimeric G proteins (Gα, Gβ, and Gγ). The interaction of the GPCR with its cognate ligand changes the receptor conformation, which triggers the dissociation of Gα from the Gβγ dimer. The activated Gα subunit initiates the classical intracellular signaling cascade by activating small GTPases. The dimer Gβγ also activates downstream effectors. The activated small GTPases then transmit the signal to kinases and other effectors that have the ability to modulate both transcription modulators and post-translational signals that will modify the target cell gene expression and cell phenotype, respectively. Both phenotypic and gene-expression changes are able to influence each other, creating a closed regulation loop between GPCR signaling, phenotype, and gene expression. (AC) Adenylyl cyclase, (PLC) phospholipase C, (CGB-PDE) cGMP phosphodiesterase, (PKA) protein kinase A, (MAPK) mitogen-activated protein kinase, (ECM) extracellular matrix. (Created with BioRender.com.)

The aforementioned GPCR signaling cascades occur rapidly after ligand binding and evoke largely post-translational changes in the cell interior. These cytosolic signals are relayed into the nucleus to impact gene transcription. For example, GPCR signaling activates Gαs- or Gαi/o-dependent cAMP/protein kinase A (PKA)/cAMP-response element-binding protein (CREB) alterations. Activated CREB then translocates into the nucleus, binds to cAMP-response elements (CREs) in DNA, and modulates gene transcription (Zhang et al. 2020). A wide variety of transcriptional modules, including CRE, NF-κB, Hippo/YAP/TAZ, STAT3, and AP-1 have been described (Ye 2001; Ho et al. 2009). In several cases, the relationships between GPCR signaling and its biological function or phenotype have been well established, but how downstream regulation of gene expression contributes to the phenotype has remained elusive. In this review, we will focus on this poorly understood aspect of GPCR signaling with a focus on S1P and the vascular system.

Established Examples of GPCR Regulation of Gene Transcription and Cellular Phenotype Change

The study of Dictyostelium discoideum, a unicellular eukaryote, has provided a classical example to understand how GPCRs regulate gene transcription and modify developmental phenotypes (Gomer et al. 1986; Artemenko et al. 2014; Pergolizzi et al. 2017). During its life cycle, D. discoideum grows as a single-celled organism while in a nutrient-rich environment and relies on mitosis for reproduction. When nutrients are scarce, D. discoideum initiates a sporulation phase that requires the cooperation of multiple cells to ensure their survival via the propagation of spores, hence its name of “social amoeba.” GPCR signaling is essential for chemotaxis, cell differentiation, and cell–cell interaction events required to complete the sporulation process. Briefly, waves of extracellular cAMP act as potent chemoattractants via the activation of GPCR (cAR1) at the surface of neighboring cells. This signaling pathway directs the convergent migration of individual cells to initiate the formation of multicellular aggregates, which then release even more cAMP. The cAMP signaling also contributes to increased cell–cell adhesion observed at the transition from initial aggregates to larger and tighter mound-like aggregates. The multicellular masses intensify the cAMP secretion that promotes cell differentiation and transition to more organized structures (slug state). This process culminates in the formation of the fruiting body responsible to release the spores, essential for the dispersion and survival of the amoeba. Both cell migration and differentiation involve GPCR–ligand interactions coupled to intracellular signaling cascades that lead to differential gene expression (Pergolizzi et al. 2017). For example, the periodic activation of the cAMP GPCR regulates the nuclear-cytoplasmic shuttling of the GATA transcription factor GtaC via the activity of GskA, an ortholog of GSK3 (Cai et al. 2014). The repeated successions of active/nuclear and inactive/cytoplasmic GtaC contribute to create a biological circuit where the transition from low to high cAMP activates the circuit. This allows the steady-state expression level of developmental marker genes, like csaA, to be determined by the number of cAMP cycles rather than the concentration of cAMP or the overall time since initiation of development (Cai et al. 2014).

In vertebrates, opioid receptors regulate gene expression in neurons. The repetitive activation of these GPCRs during opioid addiction leads to differential gene expression likely to contribute to the pathology (Saad et al. 2019), which involves time-dependent drug dependence and withdrawal effects. Analysis of gene expression in the brain in relation to drug abuse showed that Fos/Jun and CREB are among the transcription factors that activate the promoters of targeted genes (Zhou et al. 2014). Acute opiate action inhibits neurons via the opioid GPCR/Gαi/o, which reduces intracellular cAMP/PKA activity that results in a weak activation of the CREB transcription factor. However, chronic exposure to morphine, for example, induces an up-regulation of key elements of the pathway, namely, adenylate cyclase, PKA, and CREB, which contribute to the phenotype of drug addiction (Chao and Nestler 2004; McPherson and Lawrence 2007). It is thought that epigenetic modifications and transcriptional changes contribute to the long-term effects of addiction (Madsen et al. 2012).

These classical examples demonstrate that GPCR-dependent gene-expression changes are critical for phenotypic alterations in both invertebrates and vertebrates. Our review will focus on S1PR-dependent gene-expression changes in the vascular system during development, homeostasis, and disease.

S1P SIGNALING IN THE VASCULAR SYSTEM

Embryonic Development

Blood Vessels and S1P

The global deletion of the S1pr1 gene is embryonic lethal in mice at ∼E14.5 resulting from defective maturation of blood vessels, hemorrhage, and edema (Liu et al. 2000). This phenotype was recapitulated in the endothelium-specific S1pr1 KO (Allende et al. 2003) and may be due to S1PR1 regulation of EC and pericyte interactions in the dorsal aorta and the central nervous system (CNS) arteries (Paik et al. 2004). Combined deletion of S1pr1 and S1pr2 or S1pr1, S1pr2, and S1pr3 worsened the aberrant vascular phenotype with embryos dying earlier at E12.5 and E11.5, respectively (Kono et al. 2004). A similar phenotype was observed in animals lacking the S1P ligand (Sphk1 and Sphk2 double-KO mice) (Mizugishi et al. 2005). These data suggest that cooperative S1P signaling via multiple S1PRs is an essential step in vascular system development.

Early postnatal deletion of S1pr1 in ECs induces hypersprouting of the retinal vasculature (Gaengel et al. 2012; Jung et al. 2012). In contrast, EC-specific overexpression of S1PR1 suppresses retinal sprouting and branching compared to control animals (Jung et al. 2012). Combined deletion of S1pr1, S1pr2, and S1pr3 also resulted in a more severe phenotype in the retinal vasculature development compared to single receptor deletion, recapitulating the vascular defects observed in the embryos. The inactivation of the S1pr1 orthologs in zebrafish Danio rerio using morpholinos also shows impaired vascular development (Mendelson et al. 2013). The hypersprouting and increased filopodia numbers observed at the level of the cardinal vein plexus highly resemble the phenotype of the genetic deletion of S1PR1 in the murine retinal endothelium at the primary vascular plexus. The injection of S1pr1 mRNA in zebrafish, presented an opposite phenotype with decreased filopodia, thus validating the morpholino experiments (Mendelson et al. 2013). However, CRISPR/Cas9-mediated gene deletion efforts that led to maternal zygotic deletion of all S1P receptor genes in the zebrafish model did not lead to vascular defects, attesting to the compensatory mechanisms that can substitute for S1P effects in this model of embryonic vascular system development (Hisano et al. 2015). Together, these studies attest to the evolutionary conservation of S1P signaling systems in vascular development.

Lymphatic Vessel Development

During lymphangiogenesis, DLL4, a tip cell marker, promotes sprouting by increasing VEGFC-VEGFR3 signaling, probably via the activation of Notch1 (Geng et al. 2020). The fluid shear stress from the circulation of lymph in the newly formed vessels block the expression of tip cell markers such as DLL4. Interestingly, shear stress also promotes VEGFC-VEGFR3 signaling, which is counteracted by S1P-S1PR1 signaling, preventing ectopic spouting (Geng et al. 2020). Thus, S1P signaling may regulate the maturation of lymphatic vessel development.

Regulation of Postnatal Vascular Functions by S1P

Barrier Function and Inflammation

The S1P-S1PR1 signaling pathway is a major positive regulator of vascular barrier function thus suppressing inflammation. However, in the context of autoimmune diseases, S1P-S1PR1 signaling in lymphocytes promotes inflammation in tissue parenchyma (Brinkmann 2009; Tsai et al. 2019). In fact, S1PR1 functional antagonists are currently used in the clinic to treat numerous autoimmune diseases (Pérez-Jeldres et al. 2021; Roy et al. 2021).

Lung Endothelium

S1PR1 signaling in ECs stabilizes adherens junctions by stimulating homotypic VE-cadherin interactions, which closes intercellular gaps (Lee et al. 1999; Garcia et al. 2001; Camerer et al. 2009). In vivo, the pharmacological inhibition of S1PR1 and the EC-specific deletion of S1pr1 were both shown to increase Evans blue dye (EBD) extravasation in mouse lungs (Oo et al. 2011; Blaho et al. 2015; Burg et al. 2018), suggesting that S1PR1 signals prevent lung vascular leakage. The absence of the plasma S1P chaperone APOM also led to a detectable increase in vascular leak in the lung. The administration of an S1PR1-specific agonist to Apom^–/– mice reduced EBD leakage back to the level of control animals (Christensen et al. 2016). These data suggest that circulating S1P stimulation of endothelial S1PR1 is an important homeostatic mechanism. Using a lipopolysaccharide (LPS)-induced acute lung injury model, the administration of S1P or S1PR1-specific agonist was able to reduce lung edema, EBD extravasation, bronchoalveolar lavage (BAL) protein concentration, and leukocyte infiltration, which suggests a potential therapeutic use for S1PR1 agonists to protect lung vascular barrier during pathological inflammation (Peng et al. 2004; Sammani et al. 2010). The delivery of nebulized LPS to mice activated SPHK1 and increased S1P concentrations in the lungs (Tauseef et al. 2008), which was presumed to be a compensatory mechanism. The genetic deletion of Sphk1 increased lung edema and infiltrating neutrophils, supporting the role of S1P-S1PR1 in the vascular barrier protection of the lungs (Tauseef et al. 2008).

During acute lung injury induced by influenza A virus (IAV) infection, S1PR1 contributes to lung protection (Teijaro et al. 2011; Zhao et al. 2019). First, the administration of S1PR1 agonist suppressed the cytokine storm, blocked innate immune cell activation, and increased mice survival (Teijaro et al. 2011). Second, it was demonstrated that EC-specific S1PR1 deletion is associated with an increased lung edema, accompanied by a higher level of total protein in the BAL 5 days after infection, as well as an elevated BAL fluid (BALF) cytokines/chemokines concentration (Zhao et al. 2019). The deletion of S1pr1 in the endothelium was also associated with an increased number of infiltrating immune cells. Interestingly, administration of an S1PR1 agonist in control mice, but not in S1pr1-ECKO mice, had the opposite effect on edema, BALF total protein, BALF cytokines/chemokines concentrations, and the amount of immune cell infiltration.

The stimulation of S1PR1 signaling for its protective function on acute lung injury is an intriguing therapeutic option. Because neurogenic pulmonary edema following brain death is a major problem in donor lung availability for transplants, it was suggested that administration of S1PR1 agonists may provide a viable solution (Sammani et al. 2011). This opens the question of whether the endothelial protective effect of S1PR1 could be used in organ transplantation (Kummer et al. 2020; Perez Daga et al. 2020) or vein grafts (Pashova et al. 2020). Stone et al. used a lung ischemia reperfusion injury model to show that an S1PR1 agonist (with S1PR3 antagonist properties) was protective of the lung vascular barrier (Stone et al. 2015). This effect was reverted by cotreatment with an S1PR1 antagonist. Because the molecule used in this study was also shown to induce lymphopenia (Zhu et al. 2007), it would require further investigations to confirm that the stimulation of S1PR1 at the surface of ECs is responsible for this protective effect. In a reverse Arthus reaction (RAR) model, the lungs of S1pr1-ECKO mice had increased vascular permeability, hemorrhage, and increased neutrophil recruitment (Burg et al. 2018). Using the same RAR model, the treatment with an S1PR1-specific agonist reduced the RAR in wild-type mice (Burg et al. 2018). Together, these data suggest that S1PR1 protects the integrity of the lung vasculature and reduces the recruitment of proinflammatory neutrophils.

Proinflammatory cytokines can induce expression of S1P receptors. For example, S1PR2 is induced by LPS or TNF-α in microvascular ECs in vitro (Du et al. 2012). Because S1PR2 signaling contributes to the weakening of the vascular barrier (Sanchez et al. 2007), the use of the S1PR2-specific antagonist may be useful for protecting the endothelium from inflammation-induced damage.

Sepsis

HDL-S1P protects endothelial functions and reduces lung injury during sepsis in a rat model (Fan et al. 2020). Overexpression of Apom in mice improved the survival rate during sepsis, and this effect was partially reversed by an S1PR1/3 antagonist (Kurano et al. 2018). Therefore, the protection offered by HDL-S1P is dependent on the stimulation of EC-S1PR1.

An S1PR1-specific agonist, SEW2871, was shown to improve renal microcirculation when administered after sepsis triggered by cecal ligation and puncture in mice (Wang et al. 2015). Using the rat cremaster muscle and histamine-induced microvascular leakage model, it was shown that S1P was incapable of reducing the leakage (Lee et al. 2009). On the other hand, S1PR1-specific agonist SEW2871, or the combination of S1P and an S1PR2 antagonist (JTE013) were able to reduce the vascular leakage. The data show the complexity of the S1P signaling system in which different receptors (i.e., S1PR1 vs. S1PR2 and S1PR3) can have opposite effects on the endothelium (Lee et al. 2009).

Brain Endothelium and Neuroinflammation

In the CNS, the vasculature forms the BBB and blood retinal barrier (BRB). Such selective barriers are crucial to protect the CNS and its defects are thought to contribute to different CNS diseases. EC-specific deletion of S1pr1 increases the vascular permeability of the BBB and BRB. Unlike the lungs, the leakiness was restricted to small molecular mass tracers (<3 kDa) (Yanagida et al. 2017). It was proposed that S1PR1 promotes the function of tight junction complexes containing Claudin-5 and Occludin, thus enabling proper BBB functions.

In a stroke model, EC-S1PR1 signaling is protective (Li et al. 2020). The administration of S1PR1-specific agonist SEW2871 improved both the cerebral blood flow and the neurological outcome, as well as decreased the infarct volume after pMCAO (Iwasawa et al. 2018). More recently, it was shown that EC-S1PR1 maintains perfusion and microvascular patency in ischemic brain (Nitzsche et al. 2021). The authors proposed that BBB-penetrating S1PR1 agonists could provide therapeutic neuroprotection during stroke. Similarly, in a rat model of intracranial aneurysm, the administration of an S1PR1/5-specific agonist was shown to reduce vascular permeability and macrophage infiltration, suggesting a potential protective mechanism (Yamamoto et al. 2017).

However, S1P-S1PR1 signaling has an impact on brain inflammation independent of the BBB (Blaho et al. 2015). In an experimental autoimmune encephalomyelitis (EAE) model, mice that express an internalization-impaired S1PR1 (S1PR1_S5A) present an exacerbated disease (Garris et al. 2013). S1PR1 expression in myeloid cells was shown to be responsible for that phenotype (Tsai et al. 2019).

Tumor Vessels

The contribution of S1PR1 to the establishment of mature vessels was also demonstrated for tumor blood vessels (Sarkisyan et al. 2014; Cartier et al. 2020). EC-specific deletion of S1pr1 in blood vessels in subcutaneous tumor models presented with an elevated number of branch points, dilated vessels, defective pericyte contacts, and increased vascular leak (Cartier et al. 2020). Similarly, pharmacological inhibition of S1PR1 was previously shown to increase tumor blood vessel leakage (Sarkisyan et al. 2014). On the other hand, the specific overexpression of S1pr1 in ECs had an opposing effect, promoting vessel normalization and increasing antitumor drug efficacy (Cartier et al. 2020). It was recently demonstrated that physical exercise reduced hyperpermeability, decreased hypoxia, and improved chemotherapy efficacy in a Ewing sarcoma tumor mouse model (Morrell et al. 2019). Interestingly, vessels in tumors of mice subjected to treadmill exercise had increased expression of S1PR1 compared to tumor vessels from control mice (Morrell et al. 2019). Moreover, lack of S1PR1 in tumor ECs had reduced tumor-infiltrating macrophages and dendritic cells (DCs), suggesting that S1PR1 in the endothelium has an impact on the tumor microenvironment. All together, these data suggest that promoting S1PR1 signaling in tumor ECs represent a potential approach to facilitate the cytotoxic action of chemotherapeutic agents. Because several tumors use the S1P/S1PR1 pathway to stimulate their own cell survival capacity and promote tumor growth (Jin et al. 2016), it will be important to identify the key molecular determinants downstream of S1PR1 signaling that are responsible for this effect on tumor growth and tumor microenvironment.

S1PR1 could also inhibit tumor metastasis by promoting a tighter vascular barrier and preventing circulating tumor cells from invading and colonizing new metastatic sites. The EC-specific deletion of S1pr1 was shown to increase lung tumor foci when B16F10 cancer cells were injected systemically into the tail vein (experimental metastasis model) (Cartier et al. 2020). In a similar model, a nanoliposome-encapsulated, S1PR1-specific agonist was shown to prevent lung colonization using three different cancer cell lines (Chen et al. 2021).

Lymphatics and Lymphoid Organs

In the lymphatic endothelium, S1PR1 promotes the formation of Claudin-5-positive cell–cell junctions (Geng et al. 2020). We recently demonstrated that the LPA-LPAR1 pathway promoted β-arrestin recruitment to S1PR1 and inhibited S1P-induced Gαi signaling (Hisano et al. 2019). In lymph node LECs, the LPA-LPAR1 signaling favors punctate junctions over continuous junctions. Inhibition of LPAR1 was also shown to increase lymphocyte retention in the lymph nodes. Together, these data suggested that LPAR1 and S1PR1 signaling cross talk generates porous junctions that permit efficient lymphocyte trafficking across the LEC of the lymph nodes.

The S1P-S1PR1 signaling pathway is crucial for lymphocyte egress at sinus-lining endothelium. This pathway is also active in lymphatic ECs (LECs) and vascular ECs of lymphoid organs. The genetic deletion of S1P transporter Spns2 using Lyve1^CRE;Spns2^f/f mice presented hypoplastic peripheral lymph nodes with apoptotic and dysfunctional high-endothelial venules (HEVs), which express Lyve1 during their development. Interestingly, the DC and HEV interaction was lost in these mice, a phenotype that was also recapitulated by the EC-specific deletion of S1pr1 or by the treatment with an S1PR1 antagonist (Simmons et al. 2019). The loss of S1P-S1PR1 signaling in HEV was associated with reduced release of CCL21, a DC chemoattractant (Simmons et al. 2019).

Atherosclerosis

The induction of EC-specific deletion of S1pr1 in mice was shown to increase vascular inflammation (Galvani et al. 2015). More precisely, in the absence of S1PR1 in the endothelium, vascular adhesion molecule 1 (VCAM1) and intercellular adhesion molecule 1 (ICAM1) were up-regulated (Galvani et al. 2015). Interestingly, this anti-inflammatory function of S1PR1 was shown to be dependent on HDL-bound S1P. In vitro, the stimulation of human microvascular ECs (HUVECs) with HDL containing APOM-S1P was able to inhibit the TNF-α-induced phosphorylation of p65 compared to HDL isolated from Apom KO mice or to ALB-S1P. Therefore, HDL-bound S1P acts on S1PR1 with a Gi-biased agonist activity that minimizes β-arrestin activation. Interestingly, there are evidences in the literature of a diminished HDL-S1P level in plasma associated with chronic kidney diseases (Prüfer et al. 2015), acute myocardial infarction (Sattler et al. 2010), and type II diabetes (Tong et al. 2013, 2014). In contrast, β-arrestin-biased activation of S1PR1 could potentially exacerbate vascular inflammation observed in these pathologies. A recent study evaluated the role of cord blood–derived HDL (nHDL) in fetoplacental endothelial dysfunction. The treatment of primary fetal placental arterial ECs with healthy nHDL-S1P could reduce TNF-α-induced NF-κB activation and expression of proinflammatory markers (Del Gaudio et al. 2020). This protective effect was reverted by pretreatment with the S1PR1 antagonist W146, supporting the anti-inflammatory potential of the HDL-S1P/S1PR1 signaling axis.

Mice lacking the genes encoding either the low-density lipoprotein (LDL) receptor (Ldlr) or apolipoprotein E (Apoe) and fed a high-fat, Western-style diet for several weeks are two of the most commonly used animal models to study the development of atherosclerosis (Emini Veseli et al. 2017). In these mice models, the lesser curvature of the aortic arch, major branch points of the aorta, and pulmonary and carotid arteries (Emini Veseli et al. 2017) are particularly sensitive for atherosclerotic plaque development. Such regions are also exposed to disturbed shear stress, which promotes endothelial dysfunction, a precursor and risk factor of the disease (Hahn and Schwartz 2009).

The use of S1PR1 agonists to protect the vasculature from the development of atherosclerosis could be challenging due to the chronic nature of the pathology. On one hand, S1PR1 at the surface of the endothelium seems to enhance endothelial function. On the other hand, multiple cell types are capable of expressing S1PR1 in the atherosclerotic plaques, including vascular smooth muscle cells and immune cells like macrophages (Daum et al. 2009). It was recently demonstrated that S1PR1 expression is elevated in human femoral atherosclerotic plaques and in aortic plaques in murine models of atherosclerosis, mostly in regions presenting foam cells accumulation (Liu et al. 2018).

S1P Receptor Regulation of Gene Expression

In this section, we will review how gene expression is regulated downstream of S1PRs with a focus on the vasculature.

S1PR1

Arterial Endothelium

Because S1pr1-ECKO mice developed worse atherosclerosis and had specific patterns of activation in the mouse aorta (Jung et al. 2012; Galvani et al. 2015), we investigated the impact of S1PR1 signaling on gene expression in the adult mouse aortic ECs (MAECs) (Engelbrecht et al. 2020). We used a combination of bulk RNA-seq, ATAC-seq, and scRNA-seq on freshly isolated and FACS-sorted MAECs to describe the first detailed in vivo transcriptomic data downstream of S1PR1 signaling. We took advantage of two mouse models. The first was S1PR1-GS (S1pr1^knockin; H2B-GFP reporter allele) (Kono et al. 2004), which reports the recruitment of β-arrestin to S1PR1 by the expression of nuclear H2B-GFP. Interestingly, although S1PR1 protein is expressed across the whole aortic endothelium, the expression of H2B-GFP reporter protein is limited to a subset of MAECs (Galvani et al. 2015; Engelbrecht et al. 2020). Briefly, the GFP⁺ nuclei are sparse in the greater curvature of the aortic arch and the thoracic aorta. On the other hand, high GFP⁺ MAECs can be observed in the aortic arch lesser curvature and around the intercostal branch points of the thoracic aorta. The frequency of GFP⁺ MAECs is positively correlated with regions of the aorta exposed to disturbed flow where S1PR1 presents increased internalization into endosomal structures in vivo (Jung et al. 2012; Galvani et al. 2015). The second model used was S1pr1 ECKO (S1pr1^f/f; Cdh5-Cre^ERT2) (Jung et al. 2012), which offers a loss-of-function of this GPCR specifically in the endothelium.

The transcriptome of S1pr1 ECKO MAECs is significantly more similar to the GFP^high MAECs than the GFP^low cells (Engelbrecht et al. 2020). This suggests that a strong expression of the H2B-GFP reporter gene mainly reports a cumulative effect of desensitization of S1PR1 high signaling followed by down-regulation, and that the sum of gene-expression changes resemble the S1PR1 loss-of-function EC. Both the S1pr1-ECKO and the GFP^high MAECs were enriched in mRNAs associated with inflammation (Engelbrecht et al. 2020). Interestingly, a reanalysis with gProfiler (biit.cs.ut.ee/gprofiler) of the genes that were up-regulated in both the S1pr1 ECKO and in the GFP^high MAECS (∼150 genes) showed enrichment of the KEGG pathways “viral protein interaction with cytokine and cytokine receptor,” “NF-κB signaling pathway,” “cytokine–cytokine receptor interaction,” and “chemokine signaling pathway.” Genes like Ccl2, Ccl5, Ccl7, Ccl21, Cxcl2, Eda2r, Il2rg, Il7, Il33, Lbp, Ptgs2, Prkcq, Gng2, and Vav3 were all up-regulated in these two cell populations (Fig. 2). The transcriptome of S1pr1-ECKO MAECs had 2.5 times more genes significantly up-regulated compared to down-regulated. Thus, S1PR1 signaling appears to repress proinflammatory gene expression in aortic ECs. It is not known whether these are direct target genes of S1PR1 signaling or secondary to some physiology changes regulated by S1PR1 such as junction tightening.

Figure 2.

Sphingosine 1-phosphate (S1P)/sphingosine 1-phosphate receptor 1 (S1PR1) S1PRs signaling in endothelial cells (ECs) and gene regulation. In mouse aortic ECs, S1P/S1PR1 signaling modulates different transcription factors. The S1PR1 signaling is associated with reduced NF-κB/p65 activation and lower expression of genes such as Ccl2, Ccl5, Il2rg, and Il7. Similarly, in the developing retinal vasculature, the deletion of S1pr1-3 promoted the expression of putative NF-γB/p65 target genes. The S1P/S1PRs signaling in the retinal vasculature also promotes assembly of VE-cadherin, which represses JUNB protein expression. In the nucleus, JUNB contributes to the expression of tip cell–associated genes in the developing retinal vasculature. The gradient of AP-1/JUNB transcriptional activity, high in the sprouting front and low in the nascent vascular network, is essential for the organotypic EC specialization. JUNB targets genes such as Pdgfc, Sema3a, and Loxl2, which are up-regulated in the S1pr-gTKO and down-regulated in the Junb-ECKO retinal ECs. (Created with BioRender.com.)

The chromatin accessibility data provided insights into the mechanisms of gene regulation by S1PR1 signaling. The chromatin regions that presented higher accessibility in S1pr1 ECKO and in GFP^high MAECs were both enriched for transcription factor-binding sites related to the NF-κB pathway (HOMER motifs NFKB-p65-Rel, NFKB-p65, NFKB-p50p52), ETS (HOMER motifs: ETS-1, ETV2, EWS:ERG-fusion, Fli1), and bZIP (HOMER motif: Fosl2) (Engelbrecht et al. 2020). These transcription factor binding sites have a high probability of being used to regulate several of the enriched transcripts correlated to S1PR1 decreased signaling capacity. For example, ETS-1 was reported to regulate expression of MCP-1 (encoded by Ccl2) in the endothelium during vascular injury (Zhan et al. 2005; Russell and Garrett-Sinha 2010; Feng et al. 2014; Shiu and Jaimes 2018). Stimulation of brain ECs (bEnd.3) with CCL2 was shown to induce S1pr1 expression via activation of Ets-1 transcription factor (Stamatovic et al. 2006). This could suggest a potential feedback loop in some vascular ECs. Also, NF-κB was shown to regulate the expression of genes such as Ccl2 (MCP-1) and Ccl5 (RANTES) (Yeligar et al. 2009; Qiu et al. 2016). Therefore, in MAECs, the regulation of gene expression by S1PR1 is accompanied by changes in accessibility in regions of the chromatin enriched for binding site of transcription factors with the potential to regulate these genes.

The findings from that genome-wide analysis agree with several other studies. For example, ECs, freshly isolated from mice infected with IAV and treated with the S1PR1 agonist CYM-5442 were shown to have reduced mRNA levels of Ccl2, Ccl5, and Cxcl10 compared to ECs isolated from vehicle-treated animals (Teijaro et al. 2011). These mRNA data were also correlated with the level of these proteins present in the BAL. Direct in vitro infection of human pulmonary microvascular ECs with IAV-H1N1 presented an increased mRNA level for several cytokines/chemokines genes as well as adhesion molecules (Jiang et al. 2017). The authors demonstrated that the S1PR1 agonist CYM-5442 was able to inhibit the IAV-induced expression of ICAM1. Moreover, mechanistic experiments suggested that β-arrestin was involved downstream of S1PR1 to inhibit the phosphorylation of p65, a key step in the activation and translocation of the NF-κB.

HDL-bound S1P reduced vascular inflammation in HUVECs treated with TNF-α via biased activation of S1PR1 accompanied by enhanced β-arrestin signaling and NF-κB inhibition (Galvani et al. 2015). Reduced ICAM1 total protein level was shown in TNF-α-treated HUVECs cotreated with HDL-S1P (APOM⁺ HDL) compared to cells cotreated with vehicle, HDL without S1P (APOM^– HDL), or ALB-bound S1P. In a similar cotreatment experiment, recombinant APOM loaded with S1P was also shown to reduce VCAM1 surface expression. Although this study did not measure the mRNA level of ICAM1 and VCAM1, it is well established that TNF-α treatment on HUVECs induces a transcriptional activation of ICAM1, VCAM1, and other inflammatory markers (Collins et al. 1995). Also, the direct involvement of S1PR1 was not demonstrated in vitro, but well supported by increased expression of these markers in the aorta of S1pr1 EC-specific KO. In human aortic ECs (HAECs), coincubation of recombinant APOM (loaded with S1P) with TNF-α was also shown to have reduced mRNA expression of ICAM1, VCAM1, and SELE compared to TNF-α alone (Ruiz et al. 2017a). Similarly, S1PR1-specific agonist, SEW2871, was shown to reduce VCAM1 expression in the aorta of type-1 diabetic NOD mice (Whetzel et al. 2006). This anti-inflammatory effect was dependent on S1PR1, but not S1PR2 or S1PR3 signaling since only S1PR1 antagonist could block the ApoM protective effect, but not antagonist specific for the latter two receptors (Ruiz et al. 2017a).

ECs are highly heterogenous between and within organs, processes that are termed organotypic and regional specialization, respectively (Chavkin and Hirschi 2020; Marziano et al. 2021). Applying scRNA-seq on freshly isolated and sorted MAECs has revealed a similar finding for the mouse aorta (Engelbrecht et al. 2020). In total, eight clusters of MAECs were identified (six arterial, one venous, and one lymphatic). The arterial EC cluster #1 (aEC1) was shown to be a small number of ECs localized at the intercostal branch points in the thoracic aorta. These are the same cells that present the highest expression of the H2B-GFP reporter in the S1PR1-GS mice aorta (Galvani et al. 2015). It was found that the recruitment of β-arrestin to S1PR1 in these cells is independent of the S1P ligand (Engelbrecht et al. 2020). One hypothesis is that the shear forces from the blood flow could promote the internalization of S1PR1, which was previously shown to respond to flow (Jung et al. 2012; Galvani et al. 2015). One of the key specific markers identified for these cells was ITAG6, which was recently shown to be regulated by flow using extracellular labeling and proteomics (Béguin et al. 2020). Surprisingly, these cells did not present a profound enrichment for inflammatory marker genes, although they are potentially exposed to high shear stress. The second arterial EC cluster (aEC2) was also of interest because these cells also had a high frequency of GFP⁺ nuclei. These MAECs presented enrichment for inflammatory markers such as Cxcl12, Vcam1, and Icam1. Potentially the most intriguing cluster among all of them was the aortic LECs. These were shown to be localized in the aortic adventitia. They present a high frequency of GFP⁺ nuclei and similarly to all other MAECs except the aEC1, the GFP expression was dependent on the presence of S1P (Engelbrecht et al. 2020). The genes enriched in that LEC cluster had the highest overlap with genes up-regulated in the S1pr1 ECKO MAECs (78 transcripts). Genes that are involved in cytokine signaling like Ccl21, Il7, Il33, Nfkbia, and Lbp were among these potential S1PR1 LEC-repressed genes. This suggests that S1PR1 signaling plays an important role in the gene regulation of LECs, potentially having an impact on immune cell chemotaxis.

Microvascular Endothelium of the Retina

The loss of S1P signaling induced a hyper-branching/hyper-sprouting, and immature vasculature phenotype in the mouse retina during postnatal angiogenesis (Jung et al. 2012; Yanagida et al. 2020). This phenotype is well characterized for S1pr1-ECKO mice but poorly studied at the gene-expression level. The contribution of other S1PRs have also been overlooked until recently. The global genetic deletion and the endothelial-specific deletion of S1pr1, S1pr2, and S1pr3 (S1pr-gTKO and S1pr-ecTKO, respectively) was analyzed in the mouse retina model. The loss of all three S1PRs in the retina endothelium induced a severe vascular development defect. The phenotype resembled an exacerbated version of the S1pr1-ECKO phenotype (Yanagida et al. 2020). The RNA-seq analysis of sorted retinal ECs from S1pr-gTKO mice suggested that normal S1P signaling is required for the expression of genes encoding transcription factors (Lef1, Tcf7, and Zic3), transporters (Mfsd2a and Tfrc), and junctional proteins (Lsr, Ocln, and Jam2) abundantly expressed in CNS vasculature (Yanagida et al. 2020). The S1pr-gTKO ECs also presented increased expression of genes associated with tip cell markers such as Esm1, Igfbp3, or Angpt2, which agrees with the hyper-sprouting phenotype observed. A key finding of that study was derived from the analysis of chromatin accessibility data (ATAC-seq). Regions of the genome with increased chromatin accessibility in the S1pr-gTKO ECs were enriched for sequences corresponding to binding sites of transcription factors from the AP-1 family. The regulation of AP-1 transcription factor by S1PR1 was also demonstrated in regulatory T cells, suggesting a conserved connection downstream of S1PR1 signaling (Ishimaru et al. 2012). It was demonstrated that S1P-S1PR signaling in the nascent vascular network was promoting the assembly of VE-cadherin, which represses JUNB protein expression, while the action of VEGF at the sprouting front promotes JUNB protein expression (Fig. 2). Together, this creates a gradient of AP-1 transcription factors that is essential for the organotypic EC specialization of the vascular network. The EC-specific deletion of Junb (Junb-ECKO) showed reduced retinal vascular network progression and decreased expression of tip cell–associated genes. Interestingly, 19 genes were found to be both up-regulated in S1pr-gTKO and down-regulated in Junb-ECKO ECs. Neurovascular guidance genes are highly represented in this data set, for example, Pdgfc, Sema3a, and Loxl2. These studies suggest that S1P-S1PR signaling, which controls JunB spatial gradients in the developing retinal vasculature, is a key mechanism that translates a circulatory lipid/GPCR pathway to a spatially precise spatial structure (Yanagida et al. 2020). It is interesting to note that Junb mRNA expression was not changed significantly in S1pr-gTKO transcriptomics analysis, therefore revealing the enhanced power of combining mRNA expression data to chromatin accessibility analysis to characterize complex biological processes.

A few potential S1PR1-regulated genes identified in the aorta study (Engelbrecht et al. 2020) were also regulated in the retina ECs of S1pr-gTKO mice. For example, Ccl2 and Il2rg were significantly induced in the S1pr1-gTKO ECs. These two transcriptomic analyses (Engelbrecht et al. 2020; Yanagida et al. 2020), combined with a few other observations from the literature, strongly support that S1PR1 signaling in the endothelium negatively regulates the Ccl2 and Ccl5 encoding of the secreted protein MCP-1 and RANTES, respectively (Teijaro et al. 2011). Both of these chemokines are important for the recruitment of immune cells at sites of inflammation, raising the possibility of S1PR1 acting to modulate the flow of immune cells in tissues from the endothelium.

Lessons from Other Models

In ECs, the effect of S1P-S1PR1 signaling on gene expression are associated with anti-inflammatory effects. However, the reverse situation may be involved in other cell types as S1P-S1PR1 signaling has been linked to an increase in TNF-α and IL-1β mRNA expression in LPS-activated microglial cells in vitro (Gaire et al. 2018). This was related to the improved outcome of transient focal cerebral ischemia with the treatment of S1PR1-specific functional antagonist (AUY954) (Gaire et al. 2018). More recently, a study showed that the differentiation of astrocytes into A1 proinflammatory astrocytes by exposition to oxyhemoglobin could be inhibited by the stimulation with Ponesimod, an S1PR1 functional antagonist (Zhang et al. 2021). Ponesimod was also shown to reduce the expression of proinflammatory genes (TNF-α, IL-6, IL-1β, and iNOS) under the same in vitro conditions (Zhang et al. 2021). In vivo, early brain injury (EBI) after subarachnoid hemorrhage (SAH) was reduced in mice receiving Ponesimod compared to vehicle-treated animals with a reduced conversion of A1 astrocytes. Both in vitro and in vivo models suggested that Ponesimod regulated the astrocytes differentiation by inhibiting STAT3 transcription factor. Further validation using S1pr1 astrocytes KO and S1PR1 agonist are required to confirm that the action of Ponesimod was solely due to S1PR1 antagonistic activity, as opposed to a potential activation of S1PR1 (agonist).

There are several studies reporting that S1P-S1PR1 signaling can potentially regulate the JAK/STAT pathway and have an impact on target genes expression. In cardiac fibroblasts, S1P/dhS1P and S1PR1 signaling seems to regulate collagen synthesis through cross talk with the JAK/STAT signaling pathway and TIMP1 (Magaye et al. 2020). In primary mouse macrophages, S1P-S1PR1 signaling was shown to promote IL-6 expression and release in a JAK2-dependent manner, which acts as a feedforward loop on S1pr1 expression (Zhao et al. 2018).

Analysis of whole-blood transcriptomic coexpression analysis revealed two sets of genes (62 and 16 genes) that represented S1PR1-related molecular signatures associated with survival of patients with sepsis (Feng et al. 2020). To our knowledge, this type of “metanalysis” has not been performed yet on transcriptomic data obtained from ECs, but should be possible in the near future with scRNA-seq technologies becoming widely used (Lukowski et al. 2019; Adams et al. 2020; Koenitzer et al. 2020; Vila Ellis et al. 2020).

In an activated B-cell-like diffuse large B-cell lymphoma model, S1PR1 contributes to STAT3 activation and promotes protein expression of MCL-1, BXL-X_L, and SURVIVIN (gene related to proliferation and survival) and also MMP-9 (extracellular matrix degradation) (Liu et al. 2012). S1PR1 contributes, through the PI3K pathway, to a potentially oncogenic feedforward loop with the transcription factor basic leucine zipper transcription factor, ATF-like 3 (BATF3) in Hodgkin lymphoma cells (Vrzalikova et al. 2018).

S1P-S1PR1 signaling was proposed to stimulate the phosphorylation of the transcription factor CREB and to increase the expression of downstream target PGC-1α and NRF-1 in activated dopaminergic neurons (MN9D cells), an in vitro model of Parkinson's disease (Sivasubramanian et al. 2015). Because S1PR1 couples with Gαi, it is surprising to see activation of a transcription factor downstream of adenylate cyclase. Perhaps CREB activation was secondary to another element of S1P-S1PR1 signaling cascade.

The expression of multiple S1P receptors at the surface of a plethora of different cells in each tissue and the regulation of their expression in several biological conditions increase the complexity of gene-expression studies. The evidence for a biased signaling when S1PR1 is stimulated by ApoM-S1P on HDL particles also reveals that there could be large differences in gene-expression regulation between different synthetic agonists. Therefore, it is important to validate potential target genes using different approaches such as genetic and pharmacological tools.

Other S1PRs

S1PR2

The action of S1PR2 is generally opposed to S1PR1. S1PR2 signaling is more proinflammatory compared to S1PR1. For example, S1PR1 promotes endothelial barrier formation while S1PR2 decreases the endothelial barrier (Sanchez et al. 2007; Du et al. 2012; Li et al. 2015; Wiltshire et al. 2016; Burg et al. 2018; Cao et al. 2019).

ECs activate the expression of S1PR2 during inflammation. In HMVECs, the treatment with LPS or TNF-α induced S1PR2 mRNA (Du et al. 2012). The treatment with the S1PR2 antagonist JTE-013 was shown to reduce the hyperpermeability and the expression of adhesion molecules induced in ECs either by LPS or TNF-α (Du et al. 2012). During LPS-induced endotoxemia, S1pr2 KO mice were shown to express lower mRNA levels of inflammation marker and coagulation factors (Sele, Vcam1, Icam1, Ccl2, and F3) in liver, lung, and kidney (Zhang et al. 2013a). The activation of the Rho-ROCK pathway downstream of S1PR2 signaling was proposed to activate stress-activated kinase (p38 SAPK) and NF-κB, which would regulate gene expression (Zhang et al. 2013a). Similarly, S1PR2 contributes to the LPS-induced BBB disruption and neutrophil infiltration (Xiang et al. 2021). The S1pr2 KO mice have reduced expression of Sele, Cxcl1, and Cxcl2, which are key factors of neutrophil infiltration. S1PR2 signaling in cultured ECs was shown to promote the expression of adhesion molecules such as VCAM1 and ICAM1, most likely by increasing NF-κB activation (Zhang et al. 2013b; Liu et al. 2016; Ganbaatar et al. 2021). It has also been shown that S1PR2 signaling stimulates the expression of other proinflammatory proteins such as CCL2, IL-6, and PTGS2 (Zhao et al. 2015).

In HUVECs, ox-LDL increases S1PR2 expression (Ren et al. 2017). In this case, S1P-S1PR2 signaling could be implicated in an inflammatory feedforward loop by promoting the production of other proinflammatory factors such as TNF-α, IL-1β, and IL-10. It was demonstrated that S1PR2 expression and signaling impair wound healing, EC migration, and morphogenesis response (Estrada et al. 2008). In vitro senescent ECs, generated by the serial subculture of primary ECs, have elevated expression of S1PR1, S1PR2, and S1PR3 (Estrada et al. 2008). Knockdown of S1PR2 improved the wound-healing defect of senescent ECs, suggesting that S1PR2 signaling plays a role in the senescent EC phenotype (Estrada et al. 2008). Because senescent ECs lack proliferative capacity, S1P-S1PR2 signaling likely inhibited cellular migration thus reducing the repair of the wounded area in this in vitro assay. Senescent ECs have been localized in the vascular endothelium of atherosclerosis (Honda et al. 2021; Ting et al. 2021; Xu et al. 2021). Interestingly, the expression of S1PR2 was increased in atherosclerotic lesions compared to non-lesion areas (Estrada et al. 2008). Also, the treatment of ApoE KO mice with an S1PR2 antagonist was able to improve endothelial dysfunction and to reduce atherosclerotic lesions (Ganbaatar et al. 2021). At this point, it is not clear whether S1PR2 plays a role directly in the atherosclerotic endothelium. It was previously reported that S1PR2 in myeloid cells promotes vascular inflammation and atherosclerosis (Skoura et al. 2011). Because in vitro cultured ECs respond to inflammatory stimuli by increasing S1PR2 expression and S1PR2 signaling promotes expression of proinflammatory genes, it would be interesting to assess the impact of EC-specific deletion of S1pr2 in a mouse model of atherosclerosis.

S1pr2 is induced in the brain vasculature during mouse stroke models (Kim et al. 2015). In an experimental stroke model, S1PR2 promotes cerebrovascular permeability, the development of intracerebral hemorrhage and neurovascular injury. Inhibition of S1PR2 in this model reduces MMP-9 activity in the brain. In human brain microvascular ECs (hBMVECs), S1PR2 expression promotes both MMP-9 mRNA expression and enzymatic activity (Kim et al. 2015).

In anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) patients, S1P level and renal expression of S1PR2-5 were shown to be elevated and associated with the development of the disease (Sun et al. 2017). Treatment with MPO-positive IgG from AAV patients was shown to induce barrier disruption and expression of inflammation makers in cultured primary human glomerular ECs (GEnCs) and S1P would amplify this effect (Sun et al. 2018b). Using the same model, S1P was shown to activate both RAC1 and RHOA via signaling of S1PR1 and S1PR2-5, respectively (Sun et al. 2018a). Consequently, a balance is established in GEnCs between the decrease (S1PR1 and RAC1) and the increase (S1PR2-5 and RHOA) of the inflammation markers ICAM1 and VCAM1, presumably via gene-expression regulation. This would suggest that the change in receptor expression during the disease could contribute to exacerbate the vascular injury by favoring the S1P-RHOA proinflammatory signaling.

S1P can regulate the STAT3 pathway (Degagné et al. 2014). There are several examples of S1P-S1PR2 signaling regulating the activity of STAT3 in non-ECs (Frias et al. 2009; Loh et al. 2012; Smith et al. 2013; Wang et al. 2021). This raises the possibility that S1P-S1PR2 could regulate gene expression by modulating the STAT3 pathway in the endothelium. Whether this action is direct or occurs in collaboration with other factors (i.e., IL-6) is not clear at present.

S1PR3

The expression of SPHK1 and S1PR3 was induced in a mouse model of malaria-associated acute lung injury/acute respiratory distress syndrome (ALI/ARDS). Immunohistochemistry revealed that S1PR3 was induced in ECs, alveolar epithelial cells, and alveolar macrophages (Punsawad and Viriyavejakul 2019) of mouse lung tissue, but the role of S1PR3 signaling remains to be described in this model. Interestingly, the same group has found that in lungs of human patients suffering from severe malaria with pulmonary edema, SPHK1 and S1PR3 were also induced (Viriyavejakul and Punsawad 2020). Similarly, LPS-induced acute lung injury in mice increases expression of S1pr3 mRNA and protein, suggesting that S1PR3 is a biomarker in acute lung injury (Sun et al. 2012). In a mouse model of implant arteriogenesis, it was suggested that S1PR3 promotes the release of angiocrine factors like SDF-1α and preferentially recruits anti-inflammatory macrophages to promote tissue repair (Awojoodu et al. 2013) and contribute to VEGF-induced sprouting (Das et al. 2015). The action of S1PR3 in the angiogenic process is partly happening in ECs, but also in other cells like macrophages or epithelial cells with effects on the regulation of VEGF-A (Yasuda et al. 2021). A potential role for S1PR3 induction during lung injury could be to promote regeneration by stimulating cell proliferation. In that regard, S1P-S1PR3 was proposed to stimulate cell proliferation in mouse endothelial progenitor cells (Wang et al. 2018a), in mouse embryonic stem cells (Ryu et al. 2014), and in HUVECs (Ruiz et al. 2017b). In HUVECs, S1PR3 was shown to up-regulate VEGFR2 protein expression, which promoted cell proliferation, cell migration, and tube formation (Jin et al. 2018). S1PR3 is part of a transcriptional program associated with maturation of fetal pulmonary LECs (Norman et al. 2019). One hypothesis could be that during lung regeneration, there might be reactivation of a developmental gene-expression program that includes S1PR3. A careful in-depth analysis of the transcriptome and chromatin structure in these models combined with S1PR3 genetic models would be required to test that hypothesis.

In the human EC line EA.hy 926, it was suggested that S1PR3 signaling could mediate the inhibition of TNF-α-induced adhesion molecules gene expression (ICAM1 and VCAM1) and the adhesion of immune cells to the ECs (Imeri et al. 2015). It would suggest that S1PR3 signaling, like S1PR1, can interfere with NF-κB activation in some ECs, although in that particular model only S1PR3 showed a significant regulation of the target genes. On the other hand, an earlier study had suggested that S1PR3 was promoting adhesion molecules expression in HUVECs (Kimura et al. 2006). In agreement with that early study, EC S1PR3 promotes leukocyte rolling by inducing the mobilization of P-selectin to the cell surface, which is an opposite effect to S1PR1 signaling (Nussbaum et al. 2015). In addition, S1PR3 was shown to mediate the release of Weibel–Palade bodies in cultured ECs (van Hooren et al. 2014), which could potentially promote thrombosis.

A recent study presented a correlation between high vascular permeability brain metastasis and high expression of S1PR3 in astrocytes (Gril et al. 2018). Both a mouse metastasis model and an in vitro coculture model suggest that S1PR3 signaling promote the secretion of several chemokines, growth factors, and interleukins. The mechanism could be at the gene-expression level, since it was shown that the induction of Ptgs2, Il6, and Vegfa mRNAs by S1P was inhibited in S1pr3-KO mouse astrocytes (Dusaban et al. 2017). Among these, MCP1 (Ccl2) and IL-6 were shown to contribute to the disruption of the blood–tumor barrier (Gril et al. 2018). As we have shown, both of these molecules are potentially regulated by S1PR1 in different cell types, suggesting an overlap in gene targets, but opposite regulation by the two S1P receptors. The remaining question is whether S1PR3 at the surface of ECs can regulate gene expression the same way it does in astrocytes. S1PR3 not only has some functional overlap with S1PR1 but it can also have similar effects compared to S1PR2. For example, it was proposed that both S1PR2 and S1PR3 promote S1P-induced Ptgs2 expression in rat vascular smooth muscle cells (Machida et al. 2016). Overall, there are very few details available about how S1PR3 signaling influences gene expression. The impact of S1PR3 potentially overlap with S1PR1 and S1PR2 depending on the cell type and the context. The fact that S1PR3 can couple with different types of Gα proteins probably contributes to complicate its capacity to modulate gene expression.

S1PR5

Interestingly, S1PR5 appears to be expressed in human brain capillaries (van Doorn et al. 2012). In vitro cultured human brain ECs (hCMEC/D3) express four S1PRs mRNAs, only S1PR4 remained undetectable. S1PR5 is the second most expressed after S1PR1. It was demonstrated that treatment of these cells with an S1PR5-specific agonist (Hanessian et al. 2007) could promote barrier formation and reduce monocyte transmigration and VCAM1 mRNA expression, while promoting a slight increase in CDH5 mRNA expression (van Doorn et al. 2012). The knockdown of S1PR5 using an shRNA had opposite effects compared to the agonist treatment accompanied by an increased activation of NF-κB. Although most S1PRs were down-regulated in the shRNA experiment, overall these data suggest a similar function and gene-expression targets for S1PR1 and S1PR5 (van Doorn et al. 2012). It would be interesting to compare transcriptomic data from brain ECs isolated from mice with EC-specific deletion of S1pr1 or S1pr5 to determine in vivo their respective gene targets.

An S1PR5-specific agonist promotes the BBB integrity and reduces age-related cognitive decline in a mouse model (Hobson et al. 2015). This observation was in agreement with an S1PR5 barrier-stimulating function in brain ECs (van Doorn et al. 2012). In a mouse model of Huntington's disease, the administration of S1PR5 agonist slowed down the progression of the disease, prolonged mice life span, and protected the BBB homeostasis (Di Pardo et al. 2018). Interestingly, when compared to vehicle treated mice, the S1PR5 agonist treated animals had elevated expression of Claudin-5 and Occludin, both of which have been shown to be reduced in Huntington's disease (Drouin-Ouellet et al. 2015; Di Pardo et al. 2017). Therefore, S1PR5 has been proposed as a druggable target in brain diseases (Grassi et al. 2019).

CONCLUDING REMARKS

S1P signaling is critical for the proper development, maturation, and maintenance of the vasculature. These phenotypes were well characterized and linked to different elements of the S1P signaling cascades. We have a fair amount of knowledge of what happens in the EC after the initiation of the signaling cascade. Over a longer period of time, signals lead to chromatin reorganization, transcription factors modulation, and the regulation of gene expression. Such changes determine organotypic endothelial function and phenotype. Deeper understanding in this area is anticipated to help in the therapeutics of vascular diseases using the S1P pathway-targeted approaches.