Molecular and neuroimmune pharmacology of S1P receptor modulators and other disease-modifying therapies for multiple sclerosis

Abstract

Multiple sclerosis (MS) is a neurological, immune-mediated demyelinating disease that affects people in the prime of life. Environmental, infectious, and genetic factors have been implicated in its etiology, although a definitive cause has yet to be determined. Nevertheless, multiple disease-modifying therapies (DMTs: including interferons, glatiramer acetate, fumarates, cladribine, teriflunomide, fingolimod, siponimod, ozanimod, ponesimod, and monoclonal antibodies targeting ITGA4, CD20, and CD52) have been developed and approved for the treatment of MS. All the DMTs approved to date target immunomodulation as their mechanism of action (MOA); however, the direct effects of some DMTs on the central nervous system (CNS), particularly sphingosine 1-phosphate (S1P) receptor (S1PR) modulators, implicate a parallel MOA that may also reduce neurodegenerative sequelae. This review summarizes the currently approved DMTs for the treatment of MS and provides details and recent advances in the molecular pharmacology, immunopharmacology, and neuropharmacology of S1PR modulators, with a special focus on the CNS-oriented, astrocyte-centric MOA of fingolimod.

1. Introduction

Neurological disorders are medically defined as diseases of the nervous system, including stroke, Alzheimer’s disease and other dementias, Parkinson’s disease, multiple sclerosis (MS), epilepsy, brain tumors, and others. They are the leading cause of disability-adjusted life-years (DALYs) and one of the major causes of death (Collaborators, 2019; Kang et al., 2022), and thus the market for neurological disorders is continuously growing. Among neurological diseases, MS is unique in that it involves peripheral immune cells, both B and T cells, that attack the central nervous system (CNS) (Frohman, Racke, & Raine, 2006; Nakahara, Maeda, Aiso, & Suzuki, 2012; Noseworthy, Lucchinetti, Rodriguez, & Weinshenker, 2000). Clinical presentation is variable over time and includes optic neuritis, dizziness, ataxia, sensory loss, urinary incontinence, fatigue, depression, cognitive decline, and so on. Currently, the diagnosis of MS is standardized by the McDonald criteria (Kihara, 2019; McDonald et al., 2001; Thompson et al., 2018). Based on the pattern of progression and symptom changes over time, MS can be grouped into relapsing remitting MS (RRMS), secondary progressive MS (SPMS), and primary progressive MS (PPMS) (Frohman et al., 2006; Nakahara et al., 2012; Noseworthy et al., 2000). RRMS is the most common form of MS, with unpredictable, repeated cycles of exacerbations and remissions. Most cases of RRMS enter a progressive phase with relapses but no remission, i.e., SPMS. PPMS is defined prospectively or retrospectively by the accumulation of disability for >1 year without remission. Clinically isolated syndrome (CIS) is a single episode of MS symptoms with MRI features that may or may not be indicative of future MS. Epidemiology shows differences in gender (female:male = 2–3:1 (Compston & Coles, 2002)), geography (high latitude > equator (Tao, et al., 2016)), and ethnicity (White, Black > Hispanic > Asian (LangerGould, Gonzales, Smith, Li, & Nelson, 2022)). The prevalence of MS in the United States (U.S.) is estimated to be approximately 1 million people and 2.3 million people worldwide (Lane, Ng, Poyser, Lucas, & Tremlett, 2022). MS is characterized by plaques (lesions) in the CNS, which can be detected by magnetic resonance imaging (MRI) (Aslam et al., 2022). MS lesions in the white matter are histopathologically characterized by myelin loss (demyelination) and/or inflammation and are classified as preactive (normal myelin density and morphology with clusters of activated microglia), active (demyelination with myelin-laden macrophages), chronic active (demyelinated and hypocellular center with macrophages at the edge of the lesions), or chronic inactive (complete demyelination without macrophages/microglia) (Jonkman et al., 2015). The exact cause of MS is still unknown, but it is thought to be a combination of genetic and environmental factors. Genetic predisposition, including human leukocyte antigen (HLA) genes and interleukin 2 or 7 receptor (IL2R, IL7R) genes, have been identified as risk genes (Goris, Vandebergh, McCauley, Saarela, & Cotsapas, 2022). Environmental factors, including exposure to the Epstein-Barr (EB) virus and vitamin D deficiency have been associated with increased MS risk (Bjornevik, Munz, Cohen, & Ascherio, 2023; Janousek et al., 2022). Although no drug can completely cure the disease, medications have been developed to prevent relapses and slow down the progression of MS, known as disease-modifying therapies (DMT).

2. Disease-modifying therapies (DMTs) for MS

As of late 2022, multiple DMTs have been approved by the U.S. Food and Drug Administration (FDA) and the European Medicines Agency (EMA) (Table 1), resulting in a steady increase in the global availability of MS drugs. Because MS is an immune-mediated demyelinating disease, all FDA-approved MS drugs target immunomodulation to reduce relapse rates, delay disability progression, and prevent new disease activity in MS patients. DMTs are categorized by route of administration, such as injection, infusion, and oral, and by molecular weight, such as small, middle, and macromolecular (biologic) drugs (Table 1). This section briefly introduces the FDA-approved DMTs and their mechanism of action (MOA) via the immune system and/or the CNS. A detailed discussion of sphingosine 1-phosphate (S1P) receptor (S1PR) modulators is provided in a separate section.

Table 1.

DMTs for MS.^*1

Year (FDA)	Year (EMA)	Brand name	Generic name	Route	Type
1993	1995	Betaseron®	IFN-β1b	Injection	Biopharmaceutical
1996	1997	Avonex®	IFN-β1a	Injection	Biopharmaceutical
1996	2004	Copaxone®	Glatiramer acetate	Injection	Middle
2000	1998	Novantrone®	Mitoxantrone (DNA cross-links)	Infusion	Small
2002	1998	Rebif®	IFN-β1a	Injection	Biopharmaceutical
2006	2006	Tysabri®	Natalizumab (anti-β4 integrin Ab)	Infusion	Biopharmaceutical
2009	2008	Extavia®	IFN-β1b	Injection	Biopharmaceutical
2010	2011	Gilenya ®	Fingolimod (S1PR modulator)	Oral	Small
2012	2013	Aubagio®	Teriflunomide (DHODH inhibitor)	Oral	Small
2013	2014	Tecfidera®	Dimethyl fumarate (Nrf2 activator)	Oral	Small
2014	2014	Plegridy®	IFN-β1b	Injection	Biopharmaceutical
2014	2013	Lemtrada®	Alemtuzumab (anti-CD52 Ab)	Infusion	Biopharmaceutical
2015	n.a.	Glatopa®	Glatiramer acetate (generic)	Injection	Middle
2017	2018	Ocrevus®	Ocrelizumab (anti-CD20 Ab)	Infusion	Biopharmaceutical
2019	2017	Mavenclad®	Cladribine (Purine analogue)	Oral	Small
2019	2020	Mayzent ®	Siponimod (S1PR modulator)	Oral	Small
2019	n.a.	Vumerity®	Diroximel fumarate (Nrf2 activator)	Oral	Small
2020	n.a.	Bafiertam™	Monomethyl fumarate (Nrf2 activator)	Oral	Small
2020	2020	Zeposia ®	Ozanimod (S1PR modulator)	Oral	Small
2020	2021	Kesimpta®	Ofatumumab (anti-CD20 Ab)	Injection	Biopharmaceutical
2021	2021	Ponvory™	Ponesimod (S1PR modulator)	Oral	Small
2022	n.a.	Briumvi™	Ublituximab (anti-CD20 Ab)	Infusion	Biopharmaceutical

^*1Daclizumab was withdrawn in 2018 due to safety concerns.

2.1. Interferons (IFNs)

IFNs were discovered in 1957 as secreted proteins in response to viral infections (Isaacs & Lindenmann, 1957; Isaacs, Lindenmann, & Valentine, 1957), followed by identification of anti-viral activity by type I IFNs (IFN-α and β) (Streuli, Nagata, & Weissmann, 1980; Taniguchi et al., 1980) and anti-tumor effects by type II IFN (IFN-γ) (Borden et al., 2007). In 1986, recombinant human IFN-α2 was FDA-approved as the second biopharmaceutical of a receptor modulator for treatment of chronic B cell leukemia, (human insulin (Humulin^®, Eli Lilly) was the first FDA-approved biopharmaceutical in 1982) (Kinch, 2015). Based on the findings of reduced IFN production in cells obtained from MS patients (Neighbour & Bloom, 1979), and a proposed potential linkage between viral infection and MS etiology (Borden et al., 2007), several clinical trials were designed to test the efficacy of IFNs in MS patients, resulting in positive outcomes for IFN-β (Jacobs, O’Malley, Freeman, & Ekes, 1981), detrimental effects of IFN-γ (Panitch, Hirsch, Haley, & Johnson, 1987; Panitch, Hirsch, Schindler, & Johnson, 1987), and inconclusive results in IFN-α (Knobler et al., 1984). In 1993, the FDA approved the subcutaneous administration of recombinant IFN-β (Betasteron^®, Bayer) for RRMS patients, which reduces relapse and MRI-assessed subclinical lesions. Later, additional IFN-β drugs were approved including Avonex® (Biogen), Rebif^® (EMD Serono), Extabia^® (Novartis), and Plegridy^® (Biogen).

Multifactorial MOAs of IFN-β have been proposed that are basically immune modulators, including increased anti-inflammatory IL-10, decreased proinflammatory cytokines, inhibition of leukocyte trafficking to the brain, and promotion of neurotrophic factor production (Kieseier, 2011). However, no definitive MOA has been determined. Although the peripheral administration of IFN-β (~22.5 kDa) is unlikely to reach the CNS efficiently, endogenous IFN-β signaling in the CNS plays key roles in neuroprotection (Blank & Prinz, 2017) based upon multiple observations: Type I IFN signaling in astrocytes protects mice from experimental autoimmune encephalomyelitis (EAE; a well-established animal model of MS) by activating the aryl hydrocarbon receptor (AhR) and suppressing cytokine signaling 2 (SOCS2) (Rothhammer et al., 2016) and through regulation of cognitive function in healthy animals by modulating glutamate homeostasis (Hosseini et al., 2020). Moreover, IFN-β is endogenously produced by microglia/macrophages in response to EAE insults (Khorooshi et al., 2015; Kocur et al., 2015). Thus, the direct action of IFN-β in the CNS may protect MS patients from progression, which might be more effectively achievable by intranasal administration (Gonzalez et al., 2021; Ross et al., 2004) that was recently tested in clinical trials for treatment of coronavirus disease 19 (COVID-19: ClinicalTrials.gov Identifier; NCT04988217, NCT05054114).

2.2. Glatiramer acetate (GA)

GA (MW = 623.7 g/mol) is a synthetic polypeptide copolymer comprising a heterogeneous and random mixture of L-alanine, L-lysine, L-glutamic acid, and L-tyrosine (Weinstock-Guttman, Nair, Glajch, Ganguly, & Kantor, 2017). EAE was developed as an animal model of MS by immunization with whole brain homogenates in complete Freund’s adjuvant (Rivers, Sprunt, & Berry, 1933), followed by identification of myelin (Laatsch, Kies, Gordon, & Alvord Jr., 1962) and myelin basic protein (MBP)-derived peptides as factors for EAE induction (Eylar, Caccam, Jackson, Westall, & Robinson, 1970). In 1971, a synthetic copolymer (Copolymer-1; GA) was developed for the purpose of EAE induction, but it unexpectedly and effectively blocked EAE induction (Teitelbaum, Meshorer, Hirshfeld, Arnon, & Sela, 1971). Intensive clinical research established the efficacy and safety of GA (Kasindi et al., 2022), leading to FDA approval of Copaxone^® (Teva Pharmaceutical) in 1996, followed by Glatopa^® (Sandoz, a Novartis division) and Glatiramer Acetate Injection (Viatris) as generic medicines in 2015.

The MOA of GA has not been fully elucidated but is thought to be immunomodulation in an antigen-specific manner (Schrempf & Ziemssen, 2007; Ziemssen & Schrempf, 2007). GA competitively binds to major histocompatibility complex (MHC) molecules with myelin autoantigens, resulting in inhibition of encephalitogenic T cell responses, induction of GA-reactive T_H2-biased cells and regulatory T cells, and an increase of anti-inflammatory cytokine production. GA induces brain-derived neurotrophic factor (BDNF) not only in T cells, but also in resident CNS cells including neurons and astrocytes (Aharoni et al., 2005), indicating existence of a potential CNS MOA.

2.3. Mitoxantrone (MTX)

MTX (1,4-dihydroxy-5,8-bis[2-(2-hydroxyethylamino)ethylamino]anthracene-9,10-dione; MW = 444.481 g/mol) is an antineoplastic agent that was the first fully synthesized anthraquinone derivative (Ehninger, Schuler, Proksch, Zeller, & Blanz, 1990; Faulds, Balfour, Chrisp, & Langtry, 1991). The FDA approved MTX (Novantrone^®, EMD Serono) in 1987 for acute nonlymphocytic leukemia with subsequent approval for other cancers (Schleyer, Kaufmann, Unterhalt, & Hiddemann, 1994). MTX acts as a powerful immunosuppressive agent, which showed efficacy in a pre-clinical study using EAE (Lublin, Lavasa, Viti, & Knobler, 1987), as well as in clinical trials in MS patients that showed a reduction in disability progression and annualized relapse rate (ARR) of ~60% (Edan et al., 1997; Hartung, et al., 2002; Millefiorini et al., 1997; van de Wyngaert et al., 2001). In 2000, MTX was approved by the FDA for the treatment of RRMS and SPMS patients. Because of its potentially severe adverse effects, MTX is a second-line drug for RRMS, while it serves as a first-line treatment for malignant MS (Morrissey, Le Page, & Edan, 2005).

Originally used for cancer therapies, MTX acts as topoisomerase II inhibitor and delays cell cycle progression by inhibiting DNA synthesis in late S phase. Thus, the MOA of MTX in MS is thought to inhibit proliferating immune cells including macrophages, and both T and B lymphocytes (Neuhaus, Kieseier, & Hartung, 2006). Interestingly, MTX is identified as a nanomolar inhibitor of PIM1 kinase, a proto-oncogene member (Wan et al., 2013), whose inhibition rescues cognitive deficits and improves amyloid-β/Tau pathology in an animal model of Alzheimer’s disease (Velazquez, Shaw, Caccamo, & Oddo, 2016). Given the accumulation of MTX in glioma and metastatic brain tumors rather than in the blood (Pitz, Desai, Grossman, & Blakeley, 2011) and its beneficial effects in SPMS patients, a non-immunological CNS MOA for MTX may exist and would be worth investigating.

2.4. Fumarates

Fumaric acid ((2E)-But-2-enedioic acid) is an intermediate in the tricarboxylic acid (TCA) cycle (or Krebs cycle), and its esters and salts are called fumarates. Fumaric acid was first orally used as a supplement by Walter Schweckendiek in 1959 to treat his own psoriasis, but caused significant gastrointestinal irritation (Balak, 2015). Then, fumarates (a mixture of dimethyl fumarate (DMF) and monoethyl fumarate (MEF)) were tested instead by himself and other patients with psoriasis, resulting in the development of standardized fumarate treatments in combination with diet in the 1970s. However, concerns about efficacy and safety in patients with fumarates were raised, delaying its pharmaceutical development for more than a decade (Raab, 1984). Clinical trials in the early 1990s showed the beneficial effects of fumarates in patients with psoriasis without major safety concerns (Altmeyer et al., 1994; Nugteren-Huying, van der Schroeff, Hermans, & Suurmond, 1990), and a mixture of dimethylester fumarate and salts of ethylhydrogene fumarate (Fumaderm^®, Fumapharm AG/Biogen) was approved in Germany in 1994. Based on the established safety and tolerability of oral fumarates, as well as its immunomodulatory and neuroprotective effects, fumarates became a promising treatment for MS. A second-generation fumarate drug, BG-12 (Tecfidera^®, Biogen), which is an oral formulation of enteric-coated DMF capsules that reduce the burden on the gastrointestinal tract, reduced relapse rates and improved neuroradiologic outcomes in the CONFIRM (Comparator and an oral Fumarate in Relapsing–Remitting Multiple Sclerosis; ClinicalTrials.gov Identifiers: NCT00451451) and DEFINE (Determination of the Efficacy and Safety of Oral Fumarate in RRMS; ClinicalTrials.gov Identifiers: NCT00420212) clinical trials. The FDA approved dimethyl fumarate (Tecfidera^®, Biogen) in 2013, followed by diroximel fumarate (Vumerity^®, Biogen) in 2019, and monomethyl fumarate (Bafiertam™, Banner Life Sciences) in 2020.

DMF (MW = 144.127 g/mol) and diroximel fumarate are prodrugs that are hydrolyzed in vivo to MMF. Fumarates are believed to have neuroprotective and anti-inflammatory effects via activating the Nrf2 transcription factor that induces genes associated with reducing oxidative stress such as NADPH:quinone oxidoreductase (NQO1) and glutathione peroxidase (Giannetti, Niccolini, & Nicholas, 2012). The Nrf2 activation mechanism is mediated by modification of BTB (Broad-Complex, Tramtrack and Bric a brac) domain of a Nrf2 negative regulator, Kelch-like ECH-associated protein 1 (Keap1) (Unni, Deshmukh, Krishnappa, Kommu, & Padmanabhan, 2021). DMF is reported to induce T_H2-poralization of CD4⁺ T cells (Bruck et al., 2018; Wu et al., 2017). Protective roles of Nrf2 were demonstrated in Nrf2 knockout (KO) mice showing severe EAE course as compared to controls (Larabee et al., 2016), though the efficacy of DMF in Nrf2-KO mice was equivalent to those in WT controls (Schulze-Topphoff et al., 2016). On the other hand, DMF efficacy was diminished in EAE-induced mice lacking hydroxycarboxylic acid receptor 2 (HCA2/GPR109A) which belong to the G protein-coupled receptor (GPCR) family (Chen et al., 2014), indicating an Nrf2-independent MOA of DMF. This is supported by the finding that DMF inhibits cell surface expression of integrin α4 in T cells in a Nrf2-independent manner (Kihara, Groves, Rivera, & Chun, 2015). Since HCA2/GPR109A is not expressed in the CNS, fumarates may act through immune GPR109A and both immune and neuronal Nrf2. Further studies of the CNS-mediated fumarate actions are needed.

2.5. Cladribine

Cladribine (2-chlorodeoxyadenosine; (2R,3S,5R)-5-(6-amino-2-chloropurin-9-yl)-2-(hydroxymethyl)oxolan-3-ol) is a purine nucleoside analogue (Kretzschmar, Pellkofer, & Weber, 2016; Rammohan et al., 2020) and was first synthesized in 1972 to selectively inhibit proliferation of lymphocytes (Christensen, Broom, Robins, & Bloch, 1972). Concomitantly, adenosine deaminase (ADA) deficiency was discovered in patients with immunodeficiency (Giblett, Anderson, Cohen, Pollara, & Meuwissen, 1972), followed by the identification of accumulated cytotoxic deoxyadenosine nucleotides in ADA-deficient lymphocytes (Carson, Kaye, & Seegmiller, 1977). These findings led to the FDA-approval of cladribine for the treatment of hairy cell leukemia in 1993 (Saven & Piro, 1994). Cladribine is a pro-drug and in vivo it is phosphorylated by the sequential actions of kinases, including deoxycytidine kinase, to produce active triphosphate deoxynucleotide. Because of the high deoxycytidine kinase and low 5′-nucleotidase (which dephosphorylates and inactivates the compound) activities in lymphocytes, the active metabolite of cladribine is incorporated into DNA strands in lymphocytes, leading to cell death by disrupting DNA synthesis and inhibiting the cell cycle. Small placebo-controlled trials established efficacy of cladribine in both progressive MS (Beutler et al., 1996; Sipe et al., 1994) and RRMS (Romine, Sipe, Koziol, Zyroff, & Beutler, 1999; Sipe et al., 1996). The phase 3 CLARITY trial (Cladribine Tablets Treating Multiple Sclerosis Orally; ClinicalTrials.gov Identifiers: NCT00213135) showed positive results in reducing relapse rates and MRI activity (Giovannoni, et al., 2010), resulting in FDA/EMA-approval of cladribine (Mavenclad^®, Merck KGaA) in 2019 and 2017, respectively.

Cladribine is small enough (MW = 285.69 g/mol) to cross the blood–brain barrier (BBB) and may enter the CNS, as its concentration in the cerebrospinal fluid (CSF) of acute myeloid leukemia patients reached ~25% of plasma concentrations (Liliemark, 1997). In addition, oligoclonal bands in the CSF disappeared in ~50% of RRMS patients receiving parenteral cladribine (Rejdak, Stelmasiak, & Grieb, 2019). In rat microglia, cladribine inhibited proliferation and induced apoptosis via mitochondrial apoptotic pathways (Singh, Voss, Benardais, & Stangel, 2012). These studies suggest potential CNS actions for cladribine.

2.6. Teriflunomide

Teriflunomide (AA77–1726; (Z)-2-cyano-3-hydroxy-N-[4-(trifluoromethyl)phenyl]but-2-enamide) is an active metabolite of leflunomide (Arava^®, Sanofi), an immunosuppressive DMT for arthritis approved by the FDA in 1998 (Aly, Hemmer, & Korn, 2017). Teriflunomide inhibits dihydroorotate dehydrogenase (DHODH; K_D = 12 nM, IC₅₀ = ~650 nM) (Bruneau et al., 1998; Williamson et al., 1995), which is involved in de novo pyrimidine biosynthesis. DHODH inhibition causes cell-cycle arrest in cells that require DHODH for pyrimidine synthesis and thus inhibits lymphocyte proliferation (Cherwinski et al., 1995; Ruckemann et al., 1998; Siemasko, Chong, Williams, Bremer, & Finnegan, 1996). Importantly, cells that rely on salvage pathways for pyrimidine synthesis are spared from the cytotoxicity of teriflunomide (Nwankwo, Allington, & Rivey, 2012). Phase 3 placebo-controlled clinical trial (Teriflunomide Multiple Sclerosis Oral, TEMSO; ClinicalTrials.gov Identifiers: NCT00134563) revealed significant reductions in ARR, disability progression, and disease activity determined by MRI (O’Connor, et al., 2011), resulting in FDA and EMA approval of (Aubagio^®, Sanofi) in 2012 and 2013, respectively.

Teriflunomide (MW = 270.21 g/mol) crosses the BBB at ~0.17% of blood concentration (Lycke, Farman, & Nordin, 2023; Miller, 2017). Leflunomide inhibits inducible nitric oxide synthase in astrocytes (Miljkovic et al., 2001). Teriflunomide promotes oligodendrocyte differentiation (Gottle et al., 2018). These studies support a potential CNS MOA of teriflunomide/leflunomide possibly independent of mechanisms from the DHODH inhibition and through inhibition of protein tyrosine kinase inhibition (Elder et al., 1997; Gonzalez-Alvaro et al., 2009; Siemasko et al., 1998), cyclooxygenase-2 (COX-2) (Hamilton, Vojnovic, & Warner, 1999), and other pathways.

2.7. Monoclonal antibodies

Monoclonal antibodies are biopharmaceuticals and are representative of molecularly targeted drugs. So far, five medications including natalizumab (anti-α4 integrin Ab; Tysabri^®, Biogen), alemtuzumab (anti-CD52 Ab; Lemtrada^®, Sanofi), ocrelizumab (anti-CD20 Ab; Ocrevus^®, Roche), ofatumumab (anti-CD20 Ab; Kesimpta^®, Novartis) and ublituximab (anti-CD20 Ab; Briumvi^™, TG Therapeutics) were approved by the FDA (Linker, Kieseier, & Gold, 2008; Voge & Alvarez, 2019). Each drug targets a cell surface molecule that is highly expressed on lymphocytes and rarely found in the CNS (Mariottini, Muraro, & Lunemann, 2022; Morrow et al., 2022; Saidha et al., 2023). Since these biopharmaceuticals (~150 kDa) are normally unable to cross the BBB, the CNS MOA of these monoclonal antibodies appears to be negligible. However, disruption of the BBB in MS patients (Balasa, Barcutean, Mosora, & Manu, 2021) allows monoclonal antibodies to reach disease lesions and thus may play some role in disease modification.

2.8. Sphingosine 1-phosphate (S1P) receptor (S1PR) modulators

S1P is a bioactive lipid that acts as a regulator of various physiological and pathophysiological processes (Cartier & Hla, 2019; Kihara, Maceyka, Spiegel, & Chun, 2014) by binding to five cognate S1PRs (S1P_1,2,3,4,5) belonging to the rhodopsin family of G protein-coupled receptors (GPCRs) (Fig. 1A) (Mizuno & Kihara, 2020). S1P₁, S1P₂, and S1P₃ are ubiquitously expressed in the human body, while S1P₄ and S1P₅ are mainly expressed in lymphoid tissues and in the CNS, respectively (Fig. 1B). All the S1PRs couple to G_αi, S1P_2–5 couple to G_α12/13, and S1P_2,3 couple to G_αq. S1P is produced from sphingosine by the action of sphingosine kinase 1/2 (SPHK1/2), which are ubiquitously expressed in the human body with distinct subcellular localization (SPHK1 in the cytoplasm and SPHK2 in the nucleus, endoplasmic reticulum, and mitochondria) (Bryan, Kordula, Spiegel, & Milstien, 2008; Maceyka, Harikumar, Milstien, & Spiegel, 2012). Enzymatic activity of SPHK1 towards sphingosine (Km = 7 ± 1 μM, Vmax = 3850 ± 170 pmol/min/mg) is equivalent to that of SPHK2 (Km = 7 ± 1 μM, Vmax = 940 ± 50 pmol/min/mg) (Paugh, Payne, Barbour, Milstien, & Spiegel, 2003). Intracellularly generated S1P is released through S1P transporters (SPNS2 (Kawahara et al., 2009)) and MFSD2a (Wang et al., 2020)), followed by activation of S1PRs, in an autocrine/paracrine manner. S1P forms a spatial gradient, i.e., low S1P levels in lymphoid organs by S1P lyase-mediated degradation and high S1P levels in blood (~1 μM) and lymph (~0.1 μM) circulation (Yanagida & Hla, 2017) by binding to albumin or apolipoproteins (ApoM (Christoffersen et al., 2011) and ApoA4 (Obinata et al., 2019)). Biased signaling of S1P₁ occurs naturally in a S1P chaperonedependent manner; albumin-bound S1P inhibits adenylyl cyclase via the G_αi pathway, but high density lipoprotein (HDL)-bound S1P does not; and HDL (ApoM)-bound S1P increases the association of S1P₁ with β-arrestin (Galvani et al., 2015). S1P₁ is sensitive to receptor internalization and desensitization, which is precisely regulated by GPCR kinase 2 (GRK2), dynamin, and moesin (Arnon et al., 2011; Nomachi et al., 2013; Thangada et al., 2010; Willinger, Ferguson, Pereira, De Camilli, & Flavell, 2014). Reviews for S1P biology are available elsewhere (Blaho & Chun, 2018; Blaho & Hla, 2014; Burg, Salmon, & Hla, 2022; Cartier & Hla, 2019; Kunkel, Maceyka, Milstien, & Spiegel, 2013; Maceyka et al., 2012; Nishi, Kobayashi, Hisano, Kawahara, & Yamaguchi, 2014; Proia & Hla, 2015; Saba, 2019; Saba & Hla, 2004; Spiegel & Milstien, 2011; Yanagida & Hla, 2017).

Fig. 1.

(A) S1PR signaling and (B) S1PR expression in human tissues (Sjostedt et al., 2020).

Among the FDA-approved MS drugs, fingolimod, siponimod, ozanimod, and ponesimod are known as S1PR modulators (Fig. 2). Of note, fingolimod has been among the top 50 pharmaceutical products by global sales since 2011. The proposed and widely accepted MOA of S1PR modulators, like other DMTs, is immunomodulation (sequestration of pathogenic lymphocytes from the circulation to the secondary lymphoid organ), while direct CNS activities (discussed below) are also likely. Details on the molecular, immuno-, and neuropharmacology of S1PR modulators are discussed in Section 3.

Fig. 2.

Chemical structures of S1PR modulators.

2.8.1. Fingolimod

Fingolimod (FTY720: 2-amino-2-[2-(4-octylphenyl)ethyl]propane-1,3-diol: Fig. 2) was developed by chemical modification of ISP-1 (it also known as myriocin or thermozymocidin), a fungal (Isaria sinclairii (Fujita et al., 1994)) metabolite with immunosuppressive properties. The historical aspects of fingolimod are well summarized in our reviews (Chun, Kihara, Jonnalagadda, & Blaho, 2019; Kihara, Mizuno, & Chun, 2015) as well as the expert reviews (Chiba, 2005, 2020; Chiba, Matsuyuki, Maeda, & Sugahara, 2006). Fingolimod is considered to be a structural analogue of sphingosine and thus serves as a substrate for both SPHK1 (Km = 11 ± 1 μM, Vmax = 29 ± 1 pmol/min/mg) and SPHK2 (Km = 13 ± 1 μM, Vmax = 170 ± 5 pmol/min/mg), generating an active metabolite, fingolimod phosphate (fingolimod-P) (Paugh et al., 2003). Requirement of SPHK1/2 activity in fingolimod efficacy was validated by EAE studies (Imeri et al., 2016; Kihara et al., 2022). Fingolimod-P is released extracellularly via SPNS2 (Hisano, Kobayashi, Kawahara, Yamaguchi, & Nishi, 2011) and binds to S1P₁, S1P₃, S1P₄, and S1P₅ (Table 2). The mean accumulation ratio (R_ac; which is defined as a ratio of drug concentration between steady state vs. a single dose) of fingolimod is approximately 8. Fingolimod accumulation in the CNS was clearly demonstrated in rats receiving radioactive fingolimod (Foster et al., 2007). Fingolimod-P is dephosphorylated by lipid phosphate phosphatases (LPPs), particularly LPP1a (Km = 4.1 μM, Vmax = 2000 pmol/mg protein/min) and LPP3 (Km = 6.6 μM, Vmax = 833 pmol/mg protein/min) (Mechtcheriakova et al., 2007; Yamanaka, Anada, Igarashi, & Kihara, 2008), but not by S1P lyase (SGPL1) (Bandhuvula, Tam, Oskouian, & Saba, 2005).

Table 2.

Affinity of S1PR modulators for each receptor (nM).

Modulators \ S1PR	S1P1	S1P2	S1P3	S1P4	S1P5	Assay	Reference
S1P	0.47 ± 0.34	0.31 ± 0.02	0.17 ± 0.05	95.0 ± 25.0	0.61 ± 0.39	Binding IC50	(Mandala et al., 2002)
	50.1– 100	–	1.26– 3.98	794– 1259	10.0– 39.8	GTPγ[35S]/EC50	(Brinkmann et al., 2002)
	0.9	2.9	1.1	–	43.9	GTPγ[35S]/EC50	(Im, Clemens, Macdonald, & Lynch, 2001)
	33.1 ± 0.13	–	3.55 ± 0.09	–	–	GTPγ[35S]/EC50	(Deng et al., 2007)
	0.63 ± 0.22	–	0.45 ± 0.15	–	–	Binding Ki
	0.39 ± 0.04	–	0.23 ± 0.08	–	–	Binding KD
	0.90 ± 0.70	8.9 ± 2.3	0.16 ± 0.11	8.6 ± 3.8	11.0 ± 9.0	GTPγ[35S]/EC50	(Pan et al., 2006)
Median	0.765	2.9	0.34	51.8	22.3
Fingolimod-P (Fingolimod t1/2 = 6– 9 days tmax = 12 h)	0.21 ± 0.17	>10,000	5.00 ± 2.70	5.90 ± 2.30	0.59 ± 0.27	Binding IC50	(Mandala et al., 2002)
	6.31– 7.94	–	0.40– 15.8	12.6– 251	0.79– 63.1	GTPγ[35S]/EC50	(Brinkmann et al., 2002)
	0.27 ± 0.04	>10,000	0.90 ± 0.50	–	0.50 ± 0.08	GTPγ[35S]/EC50	(Scott et al., 2016)
	–	–	–	345 ± 39	–	β-arrestin/EC50
	0.27 ± 0.02	–	–	–	–	cAMP/EC50
	0.20 ± 0.01	>1000	1.33 ± 0.19	2.06 ± 0.30	0.49 ± 0.07	GTPγ[35S]/EC50	(Selkirk, Bortolato, Yan, Ching, & Hargreaves, 2022)
	1.12 ± 0.22	–	–	–	–	β-arrestin/EC50
	0.10 ± 0.01	–	–	–	1.93 ± 0.07	Binding Ki
	0.29 ± 0.15	–	1.35 ± 0.08	–	–	GTPγ[35S]/EC50	(Deng et al., 2007)
	1.10 ± 0.08	–	2.69 ± 0.19	–	–	Binding Ki
	0.30 ± 0.10	>10,000	3.10 ± 0.45	0.60 ± 0.20	0.30 ± 0.20	GTPγ[35S]/EC50	(Pan et al., 2006)
Median	0.27	N/A	2.02	3.98	0.5
Siponimod t1/2 = 30 h tmax = 3– 4 h	0.39 ± 0.07	>10,000	>1000	750 ± 487	0.98 ± 0.43	GTPγ[35S]	(Gergely et al., 2012)
	0.39 ± 0.29	>10,000	>10,000	–	0.38 ± 0.09	GTPγ[35S]/EC50	(Scott et al., 2016)
	–	–	–	920 ± 215	–	β-arrestin/EC50
	0.21 ± 0.02	–	–	–	–	cAMP/EC50
	0.46 ± 0.05	>10,000	>1000	384 ± 67.8	81.5 ± 2.40	GTPγ[35S]/EC50	(Selkirk et al., 2022)
	9.30 ± 2.79	–	–	–	–	β-arrestin/EC50
	0.74 ± 0.09	–	–	–	9.30 ± 2.79	Binding Ki	(Kihara, Jonnalagadda, et al., 2022)
	0.80 ± 0.97	–	–	–	–	Binding KD
Median	0.46	N/A	N/A	750	5.14
Ozanimod t1/2 = 17– 21 h tmax = 8 h	0.41 ± 0.16	>10,000	>10,000	–	11.0 ± 4.3	GTPγ[35S]/EC50	(Scott et al., 2016)
	–	–	–	>1000	–	β-arrestin/EC50
	0.16 ± 0.06	–	–	–	–	cAMP/EC50
	0.40 ± 0.03	>10,000	>1000	1486 ± 306	5.84 ± 0.51	GTPγ[35S]/EC50	(Selkirk et al., 2022)
	1.12 ± 0.22	–	–	–	–	β-arrestin/EC50
	0.50 ± 0.03	–	–	–	1.93 ± 0.07	Binding Ki
Median	0.41	N/A	N/A	N/A	5.84
Ponesimod t1/2 = 30 h tmax = 2– 4 h	5.70 ± 1.20	>10,000	105 ± 1.30	>1000	59.1 ± 1.90	GTPγ[35S]	(Bolli et al., 2010)
	6.00 ± 1.40	>10,000	>1000	>1000	142 ± 1.20	Binding IC50
	3.42 ± 1.17	>10,000	89.5 ± 14.3	>10,000	26.5 ± 2.50	GTPγ[35S]/EC50	(Selkirk et al., 2022)
	2.66 ± 0.47	–	–	–	–	β-arrestin/EC50
	9.37 ± 1.13	–	–	–	17.4 ± 4.32	Binding Ki
	2.09 ± 0.27	–	–	–	–	Binding KD	(Kihara, Jonnalagadda, et al., 2022)
Median	4.56	N/A	97.25	N/A	42.8

Details of fingolimod’s clinical trials have been reviewed (Khatri, 2016). Phase 1 clinical studies of oral fingolimod was conducted in stable renal transplant patients, revealing the reduction of peripheral blood lymphocytes that reversed within several weeks after medication discontinuation, revealing a linear correlation between dosing and concentration, no interactions with cyclosporin A, and acceptable safety (Budde et al., 2002; Kahan et al., 2003). However, a phase 3 clinical trial to evaluate the efficacy of fingolimod in combination with cyclosporin A in renal transplant patients did not provide benefits over standard care using mycophenolate mofetil plus cyclosporin A (Budde et al., 2006; Tedesco-Silva, et al., 2006). The first Phase 2 clinical trial of fingolimod for RRMS (Phase 2) started in 2003, where patients receiving 5 mg and 1.25 mg of fingolimod had a significant reduction in the number of gadolinium-enhanced lesions on MRI and ARR as compared to placebo controls (Kappos, et al., 2006). Three independent phase 3 clinical trials (FREEDOMS, FTY720 Research Evaluating Effects of Daily Oral therapy in Multiple Sclerosis (Kappos, et al., 2010); FREEDOMS II (Calabresi et al., 2014); and TRANSFORMS, Trial Assessing Injectable Interferon versus FTY720 Oral in Relapsing–Remitting Multiple Sclerosis (Cohen, et al., 2010); ClinicalTrials.gov Identifiers: NCT00289978, NCT00355134, NCT00340834, respectively) showed that 0.5 mg of fingolimod, which is the approved dose per day, significantly reduced ARR in patients with RRMS as compared to placebo (Kappos, et al., 2010) and interferon groups (Cohen, et al., 2010). Fingolimod was approved as the first orally available MS drug by the FDA in 2010 and by the EMA in 2011.

The LONGTERMS study (ClinicalTrials.gov Identifiers: NCT01201356), which included an extension study of phase 2/3/3b trials and additional post marketing open-label studies (FIRST (Gold et al., 2014), VERIFY (Kappos et al., 2015), TOFINGO (Kappos, Radue, et al., 2015)), confirmed the long-term safety and efficacy, as well as no tolerability concerns, of fingolimod for RRMS patients (Cohen, Tenenbaum, Bhatt, Zhang, & Kappos, 2019). Pooled analyses of the FREEDOMS and TRANSFORMS studies showed transient reductions in heart rate to about 8–11 bpm at nadir in a dose dependent manner, reaching 4–5 h after the first dose (known as bradycardia) (DiMarco et al., 2014). This event is mediated through activation of G protein-coupled inwardly-rectifying potassium channels (GIRKs) via S1P₃ in rodents (Forrest et al., 2004; Sanna et al., 2004) and S1P₁ in humans (Gergely et al., 2012). Although no cases of serious bradycardia were reported beyond 24 h, patients receiving fingolimod for the first time need to undergo first-dose observation for the first 6 h. Although the LONGTERMS study revealed the low levels of disease activity and progression in RRMS patients receiving fingolimod (Cohen, Tenenbaum, et al., 2019), fingolimod failed to show efficacy in patients with PPMS in the INFORMS trial (Investigating FTY720 oral in PPMS ClinicalTrials. gov Identifiers: NCT00731692) (Lublin et al., 2016). The negative results of fingolimod in PPMS (Lublin et al., 2016) suggested distinct disease mechanisms occurring in RRMS vs. PPMS. A recent study of single nucleus RNA-seq (snRNA-seq) in normal-appearing MS brains that compared gene expression profiles between relapsing vs. progressive MS documented significant downregulation of SPHK1/2 in astrocytes of SPMS brains (Kihara, Zhu, et al., 2022), which may partly explain the failure of the INFORMS trial (Lublin et al., 2016).

2.8.2. Siponimod

Siponimod (BAF-312; 1-({4-[(1E)-1-({[4-Cyclohexyl-3-(trifluoromethyl)phenyl]methoxy}imino)ethyl]-2-ethylphenyl}methyl)azetidine-3-carboxylic acid: Fig. 2) was designed from fingolimod using a medicinal chemistry approach (Gergely et al., 2012; Pan et al., 2013). Siponimod contains a benzyloxy oxime moiety instead of the n-octyl moiety of fingolimod, which increases scaffold rigidity and reduces S1P₃ selectivity. Moreover, the replacement of the amino phosphate moiety in fingolimod-P to amino carboxylic acids resulted in shorter in vivo half-lives compared to fingolimod (~30 h). Siponimod shows selective binding to S1P₁ and S1P₅ with subnanomolar affinity (Table 2) and a long-lasting internalization of S1P₁, validating it as a functional antagonist of S1P₁ (Gergely et al., 2012). Siponimod reduces circulating lymphocytes, including both T (naïve T cells > central memory T cells > peripheral effector memory T cells, in that order) and B cells, but has little effect on monocytes. The mean accumulation ratio (R_ac) of siponimod is in the range of 1.9–2.7. After discontinuation of siponimod, it takes 6–7 days to completely wash-out and for normal lymphocyte counts to recover. Siponimod, as well as fingolimod (Koyrakh, Roman, Brinkmann, & Wickman, 2005), decreases heart rate and causes bradycardia via S1P₁-mediated activation of GIRK at an EC₅₀ of 15.8 nM (S1P, EC₅₀ = 1.9 nM) in myocytes (Gergely et al., 2012), while the heart rate reducing effect can be avoided by a dose titration treatment (Legangneux, Gardin, & Johns, 2013). A phase 2 clinical trial (BOLD, BAF312 on MRI lesion given once-daily; ClinicalTrials.gov Identifier: NCT01185821) that tested the safety and efficacy of siponimod in RRMS patients showed reductions in MRI-determined gadoliniumenhancing T1 lesions and new T2 lesions, as well as the ARR, over 24 months (Kappos et al., 2016; Selmaj et al., 2013). In the Phase 3 clinical trial (EXPAND, exploring the efficacy and safety of siponimod in patients with secondary progressive multiple sclerosis; ClinicalTrials. gov Identifier: NCT01665144), siponimod reduced 3- and 6-month confirmed disability progression (hazard ratios are 0.79 and 0.78, respectively) in patients with SPMS as compared to a placebo control group (Cree et al., 2022; Kappos et al., 2018). Siponimod (Mayzent^®) was approved by the FDA in 2019 and the EMA in 2020 for the treatment of RRMS including clinically isolated syndrome, and SPMS with active disease.

2.8.3. Ozanimod

Ozanimod (RPC1063; (S)-5-(3-(1-((2-hydroxyethyl) amino)-2,3-dihydro-1H-inden-4-yl)-1,2,4-oxadiazol-5-yl)-2-isopropoxybenzonitrile hydrochloride: Fig. 2) was discovered by researchers at The Scripps Research Institute, and was licensed to a startup company, Receptos Inc. (that was acquired by Celgene Corp., then by Bristol Myers Squibb). Ozanimod contains a 1,2,4-oxadiazole heterocyclic ring, a unique oxadiazole isomer found in natural products and used in ataluren (Translarna^RT, a drug for Duchenne muscular dystrophy), that become an attractive target for drug development (Biernacki et al., 2020). Ozanimod was validated as an S1P₁/S1P₅-selective agonist in primates (S1P₁-selective in rodents due to lower S1P₅ potency (Selkirk et al., 2021)) and showed a similar potency as fingolimod-P and siponimod (Scott et al., 2016). As with other S1P receptor modulators, ozanimod acts with functional antagonist activity against S1P₁ (Table 2) (Scott et al., 2016). Multiple-ascending doses of ozanimod for over 7 days reduces peripheral lymphocytes, particularly chemokine receptor 7 (CCR7)-positive naïve and central memory T cells, which can be fully recovered by 6 days after ozanimod discontinuation (Tran et al., 2017). The R_ac of ozanimod is estimated at approximately 2. Ozanimod exhibits the same heart rate reduction as observed in other S1PR modulators (Tran et al., 2017), with cardiac safety including its active metabolites on QT/QTc intervals in healthy subjects (Tran et al., 2018). Importantly, ozanimod is sequentially metabolized by cytochrome P450 (CYPs) and monoamine oxidase to generate an active metabolite, CC112273 (5-(3-(1-oxo-2,3-dihydro-1H-inden4-yl)-1,2,4-oxadiazol-5-yl)-2-isopropoxybenzonitrile), which shows a longer in vivo half-life (t_1/2 = 195 h) than ozanimod (t_1/2 = 17–21 h) and an equivalent potency against S1P₁ (~3 nM) compared to ozanimod (~1 nM) (Surapaneni et al., 2021). In addition to CC112273, other in vivo metabolites show S1P_1/5 agonistic activity, suggesting a complex MOA of ozanimod that is not completely understood. Two phase 3 clinical trials, RADIANCE (ClinicalTrials.gov Identifier: NCT02047734) (Cohen et al., 2019) and SUNBEAM (ClinicalTrials.gov Identifier: NCT02294058) (Comi et al., 2019), showed that ozanimod is well tolerated and reduces relapse rate more effectively than IFN-β1a. Ozanimod (Zeposia^®) was approved by the FDA and EMA in 2020 for treatment of RRMS patients, and was also approved for the treatment of ulcerative colitis in 2021.

2.8.4. Ponesimod

In 2010, Actelion Pharmaceuticals Ltd., that was acquired by Johnson & Johnson in 2017, reported the discovery and structure-activity relationships of ponesimod (ACT-128800; (2Z,5Z)-5-{3-Chloro-4-[(2R)-2,3-dihydroxypropoxy]benzylidene}-3-(2-methylphenyl)-2-(propylimino)-1,3-thiazolidin-4-one: Fig. 2) as a novel class of S1P₁ agonist (Bolli et al., 2010). High-throughput screening identified a series of iminothiazolidinone derivatives showing agonistic activities to S1P₁ and S1P₃. Potency and selectivity were optimized by altering substituents on the thiazolidine ring including 1) a propyl chain at the 2-imino position, 2) 2- and/or 3-substituted phenyl ring with methyl group or chlorine atom at the N3 position, and 3) 3- and 4-substituted benzylidene moiety (3-chloro or 3-methyl in combination with 4-hydroxy, 4-hydroxyalkyl, or 4-dihydroxyalkoxy group) at the position 5. Ponesimod is less potent (~10-fold) than other S1PR modulators (Table 2) but acts as a potent functional antagonist of S1P₁ by showing prolonged S1P₁ internalization (Kihara et al., 2022). The mean R_ac of ponesimod is in the range of 2.0–2.6. Ponesimod, like other S1PR modulators, decreases heart rate transiently. Single dose ponesimod administration (75 mg) takes about a week to reach normal lymphocyte counts (Brossard et al., 2013). A Phase 2 study with 464 patients showed a beneficial effect with significant reduction of new T1-weighted gadolinium-enhanced lesions on MRI (Olsson et al., 2014). The efficacy of ponesimod was directly compared with teriflunomide in the Phase 3 clinical trial (OPTIMUM; Oral Ponesimod Versus Teriflunomide in Relapsing Multiple Sclerosis, ClinicalTrials.gov Identifier: NCT02425644), which showed the superiority of ponesimod in reducing ARR, fatigue, and MRI activity. Ponesimod (Ponvory^™) was approved by the FDA and EMA in 2021.

3. Pharmacology of S1PR modulators

3.1. Molecular pharmacology of S1PR modulators

After administration of fingolimod to animals, the dominant metabolite of +80 amu was detected in plasma and identified as fingolimod-P by nuclear magnetic resonance analysis (Mandala et al., 2002). The structural similarity between fingolimod/fingolimod-P vs. sphingosine/S1P led to the identification of the direct action of fingolimod-P on S1P receptors (S1P_1,3,4,5) (Brinkmann et al., 2002; Mandala et al., 2002) and the discovery of the involvement of SPHKs in the phosphorylation of fingolimod (Brinkmann et al., 2002; Paugh et al., 2003).

S1PR modulators initially act as agonists for S1P₁, inducing intracellular calcium mobilization, adenylate cyclase inhibition, and mitogen-activated protein kinase (MAPK) activation. However, S1PR modulators preferentially induce β-arrestin-mediated S1P₁ internalization into the endosomal pathway (Graler & Goetzl, 2004; Liu et al., 1999), inhibit S1P₁ recycling to the plasma membrane, and induce S1P₁ degradation (Graler & Goetzl, 2004; Jo et al., 2005) like HDL-bound S1P (Galvani et al., 2015). However, physical binding of fingolimod-P to ApoM remains unclear (Pournajaf, Dargahi, Javan, & Pourgholami, 2022). The sustained downregulation of S1P₁ mimics receptor antagonism, and this effect is referred to as “functional antagonism” of S1PR modulators. The intracellular C-terminus of S1P₁ is essential for receptor internalization (Liu et al., 1999) and contains a conserved serine-rich region and several lysine residues that are targets for phosphorylation and ubiquitination, respectively. The phosphorylation of these residues was discovered in MS brains by proteomics analyses (Garris et al., 2013). Fingolimod-P induces phosphorylation of serine residues of S1P₁ by GRK2, and adjacent lysine residues are subsequently polyubiquitinated by WWP2 (NEDD4-like E3 ubiquitinprotein ligase) (Oo et al., 2011), leading to S1P₁ degradation.

Structural studies provide a molecular machinery of differential responses between S1P vs. S1PR modulators to S1P₁. S1P₁ was the first crystallized lipid GPCR (PDB: 3V2W, 3V2Y), which was solved as a fusion protein with T4-lysozyme in complex with a S1P₁-selective antagonist, ML056 ((R)-3-amino-(3-hexylphenylamino)-4-oxobutylphosphonic acid) (Hanson et al., 2012). S1P₃ as a Fab As55 complex in complex with a natural ligand, S1P (PDB:7C4S) (Maeda et al., 2021), and S1P₅ as a fusion protein with an apocytochrome b562RIL (Lyapina et al., 2022) were also crystalized in complex with an agonist, ONO-5430608 (PDB: 7YXA). Recent advances of cryo-EM technology enabled solutions to more structures of S1PRs in complex with heterotrimeric G proteins (Chen et al., 2022; Liu et al., 2022; Xu et al., 2022; Yu et al., 2022; Yuan et al., 2021; Zhao et al., 2022), including S1P₁-G_αiβ1γ2 complex bound to S1P (PDB: 7WF7, 7TD3, 7VIE), FTY720-P (PDB: 7VIF, 7EO2), siponimod (PDB: 7TD4, 7EO4), and CBP-307 (PDB: 7VIH), S1P₂-G_α13β1γ2 complex bound to S1P (PDB: 7T6B), S1P₃-G_αiβ1γ2 complex bound to S1P (PDB: 7EW3), FTY720-P (PDB: 7EW2), and CYM-5541 (PDB: 7EW4), and S1P₅-G_αiβ1γ2 complex bound to siponimod (PDB: 7EW1). Recently, a cryo-EM structure of CD69-bound S1P₁ coupled to the heterotrimeric G_αi complex was reported (PDB: 8G94) (Chen, Qin, Chou, Cyster, & Li, 2023).

Agonist binding, as compared to antagonist-bound structures of S1P₁, induces 1) an outward shift of transmembrane 6 (TM6) that is more pronounced for S1P (9.0 Å) as compared to fingolimod-P (7.2 Å) and siponimod (5.9 Å), 2) remarkable displacement of F265^6.44 in the PIF motif toward TM5, 3) a displacement of N307^7.49 in the NPxxY motif to the center, and 4) extension of the side chain of R142^3.50 in the DRY motif toward the center of the cavity (Fig. 3A). The polar head of a highly hydrophobic-zwitterionic S1P locates in the upper pocket and interacts with hydrophilic residues located in the cap and TM2/TM3/TM7 of S1P1 (Y29^N-term, K34^N-term, N101^2.60, S105^ECL1, T109^ECL1, R120^3.28, E121^3.29, W269^6.48, R292^7.34, and E294^7.36). The polar head of fingolimod-P has the same binding configuration as that of S1P. Mutagenesis studies identified the key residues (N101^2.60, S105^ECL1, R120^3.28, E121^3.29), of which R120^3.28, N101^2.60, and E121^3.29 form a tight interaction that is required for S1P and fingolimod-P-mediated S1P₁ activation. On the other hand, the polar group of siponimod (carboxylic acid moiety) only forms a minor interaction with E121^3.29 of S1P₁. Residues on the N-terminal helix cap (Y29^N-term and K34^N-term) that interact with an S1P₁ antagonist (ML056) are not essential for agonist binding, suggesting a different binding mode between agonists and antagonists. The alkyl chain of S1P and fingolimod-P and the cyclohexyl group of siponimod extend into a highly hydrophobic cavity of the S1P₁ orthosteric binding pocket, forming a unique trefoil shape (Fig. 3B). The trifluoromethyl group of siponimod occupies one of three trefoil sites formed by the CWxP motif in TM6, known as the toggle switch that is highly conserved in class A GPCRs. Rotation of the side chains on W269^6.48 and F273^6.25 in the CWxP motif opens a side binding pocket, resulting in strong interactions between W269^6.48 and the trifluoromethyl group of siponimod (Fig. 3B). The same side pocket is most likely occupied by the 2-propylimino group of ponesimod and the nitrile group of ozanimod.

Fig. 3.

Structural features of S1PRs. (A) Comparison between active (PDB: 7TD3) vs. inactive (PDB: 3V2W) S1P₁ structures. (B) Trefoil shape binding pocket of the active S1P₁ structure (PDB: 7TD3 and 7TD4) and the toggle switch (W269^6.48) movement between active (PDB: 7TD4) vs. inactive (PDB: 3V2W) S1P₁ structures. (C) Structure of the S1P₅-G_αiβ1γ2 complex bound to siponimod (PDB: 7EW1) and a comparison between the active S1P₁ (PDB: 7TD4) vs. active S1P₅ (PDB: 7EW1) structures. (D) F274^7.39 of S1P₂ (PDB: 7T6B) and F263^6.55 of S1P₃ (PDB: 7EW2) may cause collisions with side chains of siponimod, ozanimod, and ponesimod.

Biased agonism of S1PR modulators toward β-arrestin signaling might be explained by several key residues and their interactions including W269^6.48 of the toggle switch, N307^7.49 in the NPxxY motif, V132^3.40/L213^5.50/F265^6.44 in the PIF motif, and L135^3.43. A proposed model of S1P₁ activation and biased signaling is as follows: S1P₁ agonists flip F210^5.47 from the interfaces of TM3-TM5 to TM5-TM6 and flip down the indole ring of W269^6.48, F265^6.44 in the PIF motif and L135^3.43, allowing siponimod and other S1PR modulators to access the trefoil sites. S1P₁ agonists that disrupt the interactions of N307^7.49 with F265^6.44 preferentially couple to G_αi, whereas those that increase the interactions between N307^7.49 and L135^3.43 with S1PR modulators, as well as those that fail to disrupt the N307^7.49-F265^6.44 interactions, bias the signals toward β-arrestin. The involvement of L^3.43, which is highly conserved among S1PRs, in β-arrestin recruitment is supported by the β2 adrenergic receptor (β2AR) (Picard, Schonegge, & Bouvier, 2019). Although these studies clearly show the mechanisms underlying S1P₁ activation and biased signaling, structural information about the fingolimod-P/S1P₁/β-arrestin complex and beyond is still lacking.

The cryo-EM structure of S1P₅ in complex with siponimod showed the almost equivalent orthosteric binding pocket with S1P₁, except for a rotamer displacement of F^3.33 that creates differences in the cavity (Fig. 3C) (Yuan et al., 2021). Major differences between S1P₁ vs. S1P₅ are found in the receptor-G_αi interface. Residues on the C-terminal α5 helix of the G_αi subunit (D350, C351 and D341) interact with R78^2.37, R142^3.50, and K250^6.29 of S1P1 or R139^ICL2, R133^3.50, and R245^6.29 of S1P₅, respectively. Moreover, the intracellular loop 2 (ICL2) in S1P₁ is tightly packed into the α5 helix of the G_αi subunit as compared to S1P₅. Although M149^ICL2 of S1P₁ forms strong hydrophobic interactions with I343, I344, and T340 of G_αi, S1P₅ shows a potential weaker formation of the salt bridge between R140^ICL2 of S1P₅ and D193 of G_αi. These residues may play key determinants for G_αi coupling. Importantly, the C-terminus as well as the intracellular loops of S1P₅ lack lysine residues, suggesting that S1P₅ may escape from the ubiquitination and play more constitutive roles. The S1P₅ activity of S1PR modulators remains unclear and needs to be elucidated not only to understand the biological significance of S1P₅ but also to develop more effective and selective S1PR modulators.

Structures of S1P₂-G_α13β1γ2 and S1P₃-G_αiβ1γ2 complexes identified unique residues of F274^7.39 in S1P₂ and F263^6.55 in S1P₃ that are not conserved in other S1PRs (Chen et al., 2022; Zhao et al., 2022). The selectivity of S1P_1/5 modulators appear to be supported by the presence of these unique residues in S1P₂ (F274^7.39), S1P₃ (F263^6.55) and S1P₄ (M289^7.35), that cause collisions with the 2-ethylphenylmethyl structure of siponimod, the 2,3-dihydro-1H-inden of ozanimod and the 3-chloro-benzylidene of ponesimod (Fig. 3D) (Liu et al., 2022). On the other hand, these phenylalanine residues in S1P₂ and S1P₃ appear to stabilize the phenyl group of fingolimod, proposing a fingolimod-P-mediated S1P₂ activation that has not been supported by binding or functional assays. However, a TGF-α shedding assay (Inoue et al., 2012), which measures GPCR activation, clearly demonstrated that fingolimod-P preferentially activates the G_αi and G_α12/13, but not the G_αq, signaling pathway through the 274^7.39 residue of S1P₂ (Chen et al., 2022). The clinical significance of fingolimod’s S1P₂-mediated function requires further investigation.

3.2. Immunopharmacology of S1PR modulators

ISP-I (also known as thermozymocidin or myriocin) is the parent compound of fingolimod (FTY720), which was first isolated as an antifungal antibiotic with immunosuppressive properties (Kluepfel, Bagli, Baker, Charest, & Kudelski, 1972), primarily inhibiting serine palmitoyltransferases (SPTLC1/2/3) that serves as the first enzyme in sphingolipid biosynthesis (Miyake, Kozutsumi, Nakamura, Fujita, & Kawasaki, 1995). ISP-I suppressed T cell-dependent antibody production, alloantigen-specific cytotoxic T cells (Fujita et al., 1994), and IL-2-dependent T cell proliferation (Miyake et al., 1995; Nakamura et al., 1996), and prolonged allograft survival in animal models (Fujita et al., 1996), while it was severely toxic in vivo at above 1 mg/kg (Chiba, 2020). Chemical modification of ISP-I, including elimination of chiral centers and removal of side chains, resulted in fingolimod that showed more potent immunosuppressive activity and less toxicity than myriocin (Chiba, 2005). Fingolimod prolonged allograft survival in animal models (Chiba et al., 1996; Hoshino et al., 1996; Kawaguchi et al., 1996) and ameliorated autoimmune disease models including EAE (Fujino et al., 2003; Kataoka et al., 2005; Webb et al., 2004), arthritis (Matsuura, Imayoshi, Chiba, & Okumoto, 2000; Matsuura, Imayoshi, & Okumoto, 2000), and lupus nephritis (Alperovich et al., 2007; Wenderfer, Stepkowski, & Braun, 2008) without inhibiting SPTLC1/2/3 (Fujita et al., 1996) but was found to decrease peripheral blood lymphocyte counts and subsequently accumulate lymphocytes in the secondary lymphoid organs (Chiba et al., 1998; Yanagawa et al., 1998). Fingolimod accumulates mature medullary single-positive thymocytes (Yagi et al., 2000) with reducing CD69 expression (Rosen, Alfonso, Surh, & McHeyzer-Williams, 2003), indicating the blockade of S1P₁-mediated T cell emigration from the thymus. This was supported by increased single-positive thymocytes and peripheral T cell deficiency in S1P₁^−/− fetal liver chimeras (transplantation of E12.5 liver cells from S1P₁^−/− donors into lethally irradiated wild-type mice) as compared to control chimeras (Matloubian et al., 2004) (S1P₁-KO mice are embryonic lethal at embryonic day 14.5 due to incomplete vascular maturation (Liu et al., 2000)). Furthermore, transfer of single-positive S1P₁^−/− fetal liver chimeric thymocytes into wild-type recipients resulted in rapid homing in secondary lymphoid organs while disappearing from the blood and lymph (Matloubian et al., 2004). These phenotypes were well captured by fingolimod-treated mice (Matloubian et al., 2004). S1P₁-mediated T cell egress from lymph nodes requires overriding of retention signals by CCR7 and other G_αi-coupled GPCRs (Pham, Okada, Matloubian, Lo, & Cyster, 2008). Fingolimod also impairs T cell adhesion to the sinus via G_αi2 signaling (Zhi et al., 2011). Since endothelial S1P-S1P₁ signal increases tight-junction formation, fingolimod is proposed to inhibit lymphocyte egress by closing an egress gateway (Sanna et al., 2006; Wei et al., 2005). Since the deletion of the S1P₁ gene caused hemorrhage and embryonic lethality, S1P₁^flox/flox mice were generated (Allende, Yamashita, & Proia, 2003). Endothelial cell-specific S1P₁-KO mice (Tie2-Cre:S1P₁^flox/flox) phenocopied the global S1P₁-KO mice (Allende et al., 2003), suggesting the key roles of endothelial S1P₁ in vascular development. Tamoxifen-inducible endothelial cell-specific S1P₁-KO mice (Cdh5-CreER^T2:S1P₁^flox/flox), which are viable, did not influence EAE severity (Blaho et al., 2015), indicating that S1P₁-mediated vascular barrier modulation is not essential for immune-mediated neurodegeneration.

CD4⁺ T cell-specific S1P₁-KO mice (CD4-Cre:S1P₁^flox/flox) showed accumulation of mature single positive thymocytes as expected, and increases of Foxp3⁺ regulatory T (T_reg) cells in thymus (Liu et al., 2009). Impairment of T_reg development was validated in human CD2 promoter-driven S1P₁ transgenic (S1P₁-Tg) mice that developed age-related autoimmunity that was due to T_reg cell deficiency (Liu et al., 2009). Fingolimod affected T_reg cell trafficking with accumulation in the spleen, T_reg cell functionality with increased suppressive activity, and T_reg cell proliferation (Liu et al., 2012; Sawicka et al., 2005; Wolf et al., 2009). S1P₁ inhibits thymic T_reg development by blocking the differentiation of CD4⁺CD25⁺Foxp3⁻ cells via the Akt-mTOR pathway (Liu et al., 2009; Sun et al., 2011). S1P₁ also inhibits antigen-specific induced T_reg cell differentiation, but drives T_H1 differentiation, whose reciprocal differentiation is regulated by antagonism of TGFβ-Smad3 signaling (Liu, Yang, Burns, Shrestha, & Chi, 2010). Fingolimod controls the development of both naturally occurring and induced T_reg cells through the S1P₁-mTOR axis, which interferes with TGFβ-Smad3 signaling (Liu, Yang, et al., 2010). Loss of S1P₁ in T_reg cells (Foxp3-Cre:S1P₁^flox/flox) showed systemic autoimmunity due to retention of T_reg cells in lymphoid organ and an increase of apoptosis in CD44^highCd62L^low effector memory T_reg cells (Eken et al., 2017). MS patients treated with fingolimod had increased memory T_reg cells (CD4⁺CD45RO⁺CD25^highCD127^low and CD103⁺) as compared to non-treated or DMF treated patients (Eken et al., 2017). Thus, S1P₁ inhibition may be involved in conversion of central memory to effector memory T_reg cells in both mouse and human. Tamoxifen-inducible T_reg-specific S1P₁-KO mice (Foxp3-CreER^T2:S1P₁^flox/flox) exhibited more severe EAE course than controls, and T_H17-specific S1P₁-KO mice (IL-17A-Cre:S1P₁^flox/flox) were resistant to EAE (Eken et al., 2017). On the other hand, S1P₁-S5A mice that harbor alanine mutations in S1P₁ C-terminal serine residues, which retains S1P₁ on the cell surface (Liu et al., 1999; Thangada et al., 2010), showed severe EAE and increased STAT3-mediated T_H17 polarization than controls (Garris et al., 2013; Zehra Okus, Busra Azizoglu, Canatan, & Eken, 2022). An ex vivo study identified S1PR modulators that inhibit T_H1/T_H17 cell differentiation with increased Treg differentiation (Dominguez-Villar, Raddassi, Danielsen, Guarnaccia, & Hafler, 2019; Zehra Okus et al., 2022). Taken together, S1PR modulators inhibit T cell egress from thymus, retain T cells in secondary lymphoid organs, inhibit T_H1 differentiation, suppress T_H17 responses, and increase T_reg cell numbers and functionality by downregulating S1P₁ on lymphocytes, resulting in amelioration of disease.

S1P₁ signaling is highly relevant to lymphocyte trafficking by sensing the S1P gradient that is formed by S1P-degrading enzymes, S1P lyase (SGPL1) and by LPPs. SGPL1 inhibition by 2-acetyl-4-tetrahydroxybutylimidazole increases S1P levels in lymphoid tissues and induces lymphopenia (Zehra Okus et al., 2022). SGPL1 deficiency disrupts the S1P gradient by accumulating S1P in lymphoid tissues, resulting in lymphopenia (Bektas et al., 2010; Vogel et al., 2009; Weber et al., 2009) and milder EAE signs than controls (Billich et al., 2013). Moreover, thymic lymphocyte egress is impaired in global LPP3-KO mice with suppression of cell surface S1P₁ expression due to exposure to high levels of S1P (Breart et al., 2011). Both global SPNS2-KO and endothelial cell-specific SPNS2-KO (Tie2-Cre:SPNS2^flox/flox) mice show defects in lymphocyte egress (Fukuhara et al., 2012; Mendoza et al., 2012), suggesting the involvement of endothelial SPNS2 in establishing the S1P gradient. SPNS2-KO mice are resistant to EAE development (Donoviel et al., 2015). Fingolimod, as well as other S1P₁ agonists (SEW2871 and AUY954), induces lymphopenia in ApoM-KO mice, indicating that lymphocyte trafficking is independent of ApoM (Blaho et al., 2015). ApoM-KO mice showed increased proliferation of hematopoietic progenitors (Lin⁻Sac-1⁺cKit⁺) and common lymphoid progenitors (Lin⁻Kit^lowSca^low IL-7Rα⁺Flt3⁺), and exacerbation of EAE, whereas the opposite trend was found in ApoM transgenic mice (Blaho et al., 2015). Importantly fingolimod and S1P₁ agonists inhibited hyperproliferation of progenitor cells in ApoM-KO mice (Blaho et al., 2015), providing an additional MOA of fingolimod.

Effects of S1PR modulators in other immune cells have also been studied. Fingolimod-mediated S1P₁ inhibition induced migration of marginal zone B cells expressing S1P₁ and S1P₃ into splenic follicles, indicating the requirement of S1P-S1P₁ signaling for marginal zone B cell positioning (Cinamon et al., 2004). Another study identified that marginal zone B cells shuttle back and forth between the marginal zone and follicle by using CXCR5 and S1P_1/3, respectively (Arnon, Horton, Grigorova, & Cyster, 2013; Cinamon, Zachariah, Lam, Foss Jr., & Cyster, 2008). The relocation of marginal zone B cells into follicles by fingolimod is associated with the movement of conventional dendritic cells into the marginal zone (Liu, Wu, An, & Cyster, 2020). Fingolimod, through S1P₁, retains immunoglobulin (Ig) G (IgG) secreting plasma cells (B cells) in secondary lymphoid organs and inhibits their homing to the bone marrow (Kabashima et al., 2006). Moreover, fingolimod inhibits egress of immature B cells from bone marrow (Pereira, Xu, & Cyster, 2010). S1P₂ deficiency causes germinal center-derived diffuse large B cell lymphoma (DLBCL) via G_α12/13-Rho pathway (Cattoretti et al., 2009; Green et al., 2011). S1P₄ deficiency does not alter lymphocyte distribution and function but affects T_H17 differentiation through modulating dendritic cells (Schulze et al., 2011) and causes atypical and reduced formation of proplatelets (Golfier et al., 2010). S1P₅ is abundant in natural killer (NK) cells whose egress from lymph nodes to lymph is suppressed, but not blocked, by fingolimod (Jenne et al., 2009).

Overall, the immunological MOA of S1PR modulators is largely explained by the disruption of lymphocyte trafficking, i.e., the correlation between peripheral lymphocyte counts and disease severity. However, neuropharmacological analyses, discussed in the next section, have shown that the amelioration of EAE and MS by S1PR modulators is completely independent of this correlation and highly dependent on direct effects on the CNS.

3.3. Neuropharmacology of S1PR modulators

MS is a CNS-specific, immune-mediated disease, but a major question exists as to whether CNS cells are actively or passively involved in MS pathogenesis. Since c-Fos is one of the most validated immediate early genes (IEGs) (Bullitt, 1990; Lau & Nathans, 1987), whose activation has been documented in neurons (Masuda et al., 2019), astrocytes (Rubio, 1997), microglia (Eun et al., 2004), oligodendrocytes (Muir & Compston, 1996), and non-neural cell types, this question was addressed by an unbiased screen using a doxycycline-regulated TetTag system (that labels nuclei of c-Fos-activated cells with a green fluorescent protein (GFP) (Tayler et al., 2011))(Groves et al., 2018). TetTag mice showed an expansion of GFP signals in the spinal cord of EAE mice, in contrast with sparse and unchanging patterns of GFP⁺ nuclei in controls, which correlated linearly with EAE severity. Most importantly, the GFP signals largely (>95%) overlapped with an astrocyte marker, glialfibrillary acidic protein (GFAP), identifying immediate-early astrocytes (ieAstrocytes) as the primary CNS cell type activated in response to EAE (Groves et al., 2018). Based on gene expression patterns, reactive astrocytes have been proposed to be classified into neurotoxic A1 astrocytes vs. helpful A2 astrocytes, of which A1 astrocytes accumulate not only in MS lesions, but also in other neurological diseases (Liddelow et al., 2017; Liddelow & Barres, 2017). Although the concept has been extensively investigated, the A1 signature has not been validated by snRNA-seq studies (Kihara, Zhu, et al., 2022; Schirmer et al., 2019). The binary division of reactive astrocytes is questioned by the astrocyte community (Escartin et al., 2021). On the other hand, the snRNA-seq analyses of MS brains identified ieAstrocytes more frequently in MS lesions than in control brains (Schirmer et al., 2019), and in normal-appearing brain regions in RRMS than in SPMS (Kihara, Zhu, et al., 2022), indicating the high relevance of ieAstrocytes in MS etiology.

Early studies suggested direct actions of fingolimod in CNS cells, including that fingolimod-P activates intracellular signaling in astrocytes and oligodendrocyte progenitor cells (Coelho, Payne, Bittman, Spiegel, & Sato-Bigbee, 2007; Jung et al., 2007; Miron et al., 2008; Miron, Hall, Kennedy, Soliven, & Antel, 2008; Mullershausen et al., 2007; Osinde, Mullershausen, & Dev, 2007), enhances remyelination in organotypic cerebellar cultures (Miron et al., 2010), and promotes recovery from spinal cord injury (Lee et al., 2009) and ischemic stroke (Hasegawa, Suzuki, Sozen, Rolland, & Zhang, 2010). Fingolimod is also reported to suppress astrocyte activation (Colombo et al., 2014; Rothhammer et al., 2017), diminish microglial and myeloid cell activation (Airas et al., 2015; Di Dario et al., 2015), promote proliferation and differentiation of oligodendrocyte progenitor cells (Zhang et al., 2015), restore presynaptic defects in EAE mice (Bonfiglio et al., 2017), inhibit brain atrophy in the cerebellum and striatum of EAE mice with an induction of BDNF (Smith et al., 2018), reduce apoptosis of retinal ganglion cells and oligodendrocytes in EAE mice (Yang et al., 2021), and attenuate gait deficits in EAE mice (Kasheke, Holman, & Robertson, 2022). S1P₁ and S1P₃ expression is elevated in MS lesions (Van Doorn et al., 2010). Furthermore, astrocytic function is regulated in part by the S1P signaling pathway. S1P₁, S1P₂, and S1P₃ are abundant in astrocytes, transducing Ca²⁺ and MAPK signals via G_αi and G_αq pathway (Rao et al., 2003), inducing proliferation via G_αi pathway (Bassi et al., 2006), inhibiting gap junctions via G_α12/13-Rho-ROCK pathway (Rouach et al., 2006), and regulating blood brain barrier permeability (Yang et al., 2023). The S1P₂-G_α12/13-Rho pathway inhibits astrocytic glutamate uptake and increases mitochondrial oxygen consumption (Jonnalagadda, Kihara, Rivera, & Chun, 2021). The S1P₃-G_α12/13-Rho pathway is also responsible for pro-inflammatory gene production (COX-2, IL-6, and VEGFα) (Dusaban, Chun, Rosen, Purcell, & Brown, 2017).

Although these studies provide indirect support for the CNS-oriented MOA of fingolimod indirectly, they do not provide definitive evidence. Instead, EAE studies in CNS-specific (Nestin-Cre:S1P₁^flox/flox; S1P₁-CNSCKO), neuron-specific (Synapsin-Cre:S1P₁^flox/flox; S1P₁-NeuCKO), and astrocyte-specific S1P₁-KO mice provided strong evidence for the lymphocyte-independent MOA of fingolimod (Choi et al., 2011). Although fingolimod effectively reduced the number of circulating lymphocytes in these conditional KO mice, the signs of EAE were ameliorated by fingolimod only in S1P₁-NeuCKO mice, but not in S1P₁-CNSCKO and S1P₁-AsCKO mice (Choi et al., 2011). This study clearly demonstrated that a fingolimod-induced reduction in peripheral lymphocyte counts is not always necessary for disease improvement, and that astrocyte S1P₁ is essential for the fingolimod’s efficacy. Astrocyte S1P₁ contributes to ieAstrocyte formation, as pharmacological and genetic inhibition of S1P₁ significantly suppressed ieAstrocytes in EAE lesions (Groves et al., 2018). Gene expression analysis of ieAstrocytes revealed downregulation of a vitamin B₁₂-transcobalamin 2 (TCN2) complex receptor, CD320, in EAE and its restoration by treatment with fingolimod and S1P₁ deficiency. CD320 is responsible for the uptake of vitamin B₁₂-TCN2 complex particularly in the CNS (Lai et al., 2013), the loss of which, as well as dietary restriction of vitamin B₁₂, exacerbated EAE (Jonnalagadda, et al., 2022). Although the relationship between vitamin B₁₂ deficiency and demyelination has been proposed for a long time, the detailed mechanism remains unclear. Fingolimod reduced peripheral lymphocyte counts in CD320-KO mice but did not improve the EAE course (Jonnalagadda, et al., 2022), as was the case in S1P₁-AsCKO mice (Choi et al., 2011). This led to the identification of direct binding of fingolimod and sphingosine to TCN2 with sub-nanomolar affinity (0.14 nM) (Jonnalagadda, et al., 2022), suggesting that TCN2 mediates transfer of fingolimod/sphingosine with vitamin B₁₂ and delivers them to the CNS. Furthermore, fingolimod did not ameliorate EAE in astrocyte-specific SPHK1/2-KO mice (Kihara, Zhu, et al., 2022), suggesting that SPHK1/2 activity in the CNS is necessary and sufficient for fingolimod-mediated disease modification. These results strongly support the CNS-oriented, astrocyte-centric MOA of fingolimod and potentially of other S1PR modulators (Fig. 4).

Fig. 4.

CNS-oriented, astrocyte-centric MOA of fingolimod. Fingolimod induces lymphopenia in the periphery, while it travels with vitamin B₁₂ via TCN2 to the CNS.The vitamin B₁₂-TCN2-fingolimod complex is taken up by astrocytes via CD320. Fingolimod-P is produced by astrocytic sphingosine kinases (SK1/2), possibly released via SPNS2, and functionally antagonizes astrocytic S1P₁, resulting in CD320 production and alleviation of neuroinflammation.

Sex differences in the incidence (3 times more common in women than in men) and clinical presentation of MS have recently received increasing attention (Krysko et al., 2020; McCombe & Greer, 2022; Voskuhl, 2020). EAE in the SJL genetic background captures a relapsing-remitting disease course and an increased disease susceptibility of female mice to disease, whose gene expression analyses identified S1P₂ as a sex-specific disease-relevant molecule (Cruz-Orengo et al., 2014). Pharmacological inhibition of S1P₂ ameliorated EAE by enhancing endothelial barrier function through the Rho/ROCK and CDC42 pathways. Since fingolimod-P may have biased agonism for S1P₂ towards G_αi and G_α12/13 (Chen et al., 2022), BBB disruption may be accelerated by fingolimod treatment. On the other hand, in astrocytes, S1P induced glutamate uptake inhibition through the S1P₂-G_α12/13-Rho signaling pathway, thereby increasing extracellular glutamate and triggering neurotoxicity. Fingolimod-P slightly inhibited glutamate uptake at a higher concentration, possibly through its biased agonism for S1P₂ (Chen et al., 2022). However, because siponimod did not induce glutamate uptake inhibition (Jonnalagadda et al., 2021), the FDA-approved 2nd generation S1PR modulators may have less impact on glutamate-mediated neurotoxicity.

Cuprizone, a copper-chelator, reversibly induces demyelination in a peripheral lymphocyte-independent manner, although the involvement of CXCR2⁺ neutrophils has been reported (Liu et al., 2010), which has the strong advantage of being able to discover the remyelination processes after cuprizone removal from the diet and to study oligodendrocyte biology (Leo & Kipp, 2022; Lubrich, Giesler, & Kipp, 2022; Taylor, Gilmore, Ting, & Matsushima, 2010). The first reported testing of fingolimod in the cuprizone model showed attenuation of demyelination with decreased IL-1β and CCL2 expression, but did not promote remyelination (Kim et al., 2011), which was supported by additional studies (Alme et al., 2015; Hu et al., 2011; Kim et al., 2018; Nystad et al., 2020; Slowik et al., 2015). CNPase-Cre:S1P₁^flox/flox mice, which are considered to be oligodendrocyte-specific S1P₁-KO mice, although CNPase-Cre deletes floxed genes in peripheral lymphocytes. These mutants were susceptible to cuprizone-induced demyelination (Kim et al., 2011). Ozanimod blocked cuprizone-induced demyelination (Selkirk et al., 2021). Global S1P₅-KO mice showed less demyelination and glial activation in the cuprizone model than control mice and the protective effects of siponimod in control mice were absent in S1P₅-KO mice, suggesting that siponimod protects oligodendrocytes through S1P₅ (Behrangi et al., 2022). Ponesimod inhibited cuprizone-induced demyelination and promoted remyelination after cuprizone withdrawal, particularly in the cingulum (Kihara, Jonnalagadda, et al., 2022), a brain region of the limbic system whose demyelination is correlated with the fatigue and cognitive function in MS patients (Bisecco et al., 2016; Novo et al., 2018; Pardini et al., 2015; Sepulcre et al., 2009). In summary, in the cuprizone model, S1PR modulators, all of which can bind to S1P₁ and S1P₅, may inhibit inflammatory responses through their functional antagonism of S1P₁ expressed on astrocytes and microglia, and may exert protective effects directly through modulation of S1P₅ expressed on oligodendrocytes, while its mechanism remains unclear.

Some studies have determined the responses of S1PR modulators in CNS cells from human sources. None of the FDA-approved S1PR modulators induced intracellular Ca²⁺ mobilization or inhibition of cAMP production in human astrocytes. Ponesimod blocks inflammatory responses in human astrocytes, particularly in IFN signaling, ubiquitin-related pathways, antigen presentation, and inhibits ieAstrocyte formation via functional antagonism (Kihara, Jonnalagadda, et al., 2022). In human astrocytes, siponimod is reported to inhibit cytokine-induced NFκB translocation, maintain glutamate transporter expression levels under inflammatory conditions, and activate Nrf2 signaling pathways (Colombo et al., 2020). Fingolimod-P induces neurotrophic factors (LIF, HBEGF, IL11) and inhibits neuroinflammatory genes (CXCL10, BAFF, etc) in human astrocytes (Hoffmann et al., 2015) and enhances survival and differentiation of human oligodendrocyte progenitor cells (Cui, Fang, Kennedy, Almazan, & Antel, 2014). Siponimod induces human microglial migration that prevents cytokine-induced BBB permeability (Spampinato, Costantino, Merlo, Canonico, & Sortino, 2022).

S1PR modulators, particularly fingolimod, have been tested in other neurological disease models. In the Alzheimer’s disease (AD) models (Angelopoulou & Piperi, 2019; Lessmann, Kartalou, Endres, Pawlitzki, & Gottmann, 2023), fingolimod, even at low doses that do not affect circulating lymphocyte counts (Carreras et al., 2019), improves spatial learning and memory (Asle-Rousta, Kolahdooz, Oryan, Ahmadiani, & Dargahi, 2013; Crivelli et al., 2022; Fagan, Bechet, & Dev, 2022), reduces Aβ production in neurons (Takasugi et al., 2013), Aβ-induced neurotoxicity via BDNF production (Asle-Rousta, Kolahdooz, Dargahi, Ahmadiani, & Nasoohi, 2014; Doi et al., 2013; Fukumoto et al., 2014), Aβ deposition, astrogliosis, and microglial accumulation (Aytan et al., 2016; Kartalou et al., 2020), and reverses synaptic deficits (Kartalou et al., 2020; Krivinko, Erickson, MacDonald, Garver, & Sweet, 2022). In a Parkinson’s disease (PD) model (Martinez & Peplow, 2018; Motyl & Strosznajder, 2018), fingolimod shows neuroprotective effects in a PD model mouse (Motyl, Przykaza, Boguszewski, Kosson, & Strosznajder, 2018; Pepin, Jalinier, Lemieux, Massicotte, & Cyr, 2020; Zhao et al., 2017), attenuates degeneration of dopaminergic neurons (Ren et al., 2017), reduces synucleinopathy (Vidal-Martinez et al., 2016), alters behaviors by increasing BDNF (Vidal-Martinez et al., 2019), and attenuates L-dopa-induced dyskinesia (Liu et al., 2021). Fingolimod also prevents neuronal cell death and shows positive effects in brain and spinal cord injuries and hemorrhages (Bonitz et al., 2014; Cheng et al., 2021; Cipriani, Chara, Rodriguez-Antiguedad, & Matute, 2017; Lee et al., 2009; Wang et al., 2020; Wang, Kawabori, & Houkin, 2020; Yamazaki et al., 2020; Yang et al., 2019; Yavuz, Sezik, Ozmen, & Asci, 2017; Zhang, Ding, Wang, Wu, & Xu, 2016), with at least one contrary report (Mencl et al., 2014). Moreover, siponimod and ponesimod show neuroprotective effects in traumatic brain injury (Cuzzocrea et al., 2018) and in subarachnoid hemorrhage (Zhang et al., 2021), respectively. Indeed, several clinical trials are ongoing to test the effects of S1PR modulators in various diseases such as siponimod or fingolimod for intracerebral haemorrhage (ClinicalTrials.gov Identifier; NCT03338998, NCT04088630), and fingolimod for chemotherapy-induced neuropathy (ClinicalTrials.gov Identifier; NCT03943498, NCT03941743) (McGinley & Cohen, 2021).

4. Conclusions

Despite the growing number of therapeutic options with diverse MOAs, the development of MS therapies is still lacking, especially those that directly target the CNS to stop progression in MS and restore neurological function. In this context, S1PR modulators were shown to have CNS MOAs beyond peripheral immunomodulation. S1PR modulators have overlapping pharmacological properties, but also appear to have unique functions that reflect structural differences and pharmacological qualities that include selectivity for S1PR subtypes. Modulators targeting S1P signaling pathways including S1PRs, S1P transporters, and S1P-generating/degrading enzymes – particularly through the engagement of CNS mechanisms – remain an area of continuing potential and promise.

Abbreviations:

AD

Alzheimer’s disease

ADA

adenosine deaminase

AhR

aryl hydrocarbon receptor

ApoA4

apolipoprotein A4

ApoM

apolipoprotein M

ARR

annualized relapse rate

BBB

blood-brain barrier

BDNF

brain derived neurotrophic factor

BOLD

BAF312 on MRI lesion given once-daily

BTB

Broad-Complex Tramtrack and Bric a brac

CCR7

chemokine receptor 7

CIS

clinically isolated syndrome

CLARITY

Cladribine Tablets Treating Multiple Sclerosis Orally

CNS

central nervous system

CONFIRM

Comparator and an Oral Fumarate in Relapsing–Remitting Multiple Sclerosis

COVID-19

coronavirus disease 19

COX-2

cyclooxygenase-2

CSF

cerebrospinal fluid

DALYs

disability-adjusted life-years

DEFINE

Determination of the Efficacy and Safety of Oral Fumarate in RRMS

DHODH

dihydroorotate dehydrogenase

DMF

dimethyl fumarate

DMTs

disease-modifying therapies

EAE

experimental autoimmune encephalomyelitis

EB

Epstein-Barr

EMA

European Medicines Agency

EXPAND

exploring the efficacy and safety of siponimod in patients with secondary progressive multiple sclerosis

FDA

Food and Drug Administration

FREEDOMS

FTY720 Research Evaluating Effects of Daily Oral therapy in Multiple Sclerosis

GA

glatiramer acetate

GFAP

glial fibrillary acidic protein

GFP

green fluorescent protein

GIRK

G protein-coupled inwardly-rectifying potassium channel

GPCR

G protein-coupled receptor

GRK2

GPCR kinase 2

HCA2

hydroxycarboxylic acid receptor 2

HDL

High Density Lipoprotein

ieAstrocyte

immediate early astrocyte

IEG

immediate early gene

IFN

interferon

IL

interleukin

Keap1

Kelch-like ECH-associated protein 1

KO

knockout

LPP

lipid phosphate phosphatase

MAPK

mitogen-activated protein kinase

MBP

myelin basic protein

MEF

monoethyl fumarate

MHC

major histocompatibility complex

MOA

mechanism of action

MRI

magnetic resonance imaging

MS

multiple sclerosis

MTX

mitoxantrone

MW

molecular weight

NQO1

NADPH:quinone oxidoreductase

OPTIMUM

Oral Ponesimod Versus Teriflunomide In Relapsing Multiple Sclerosis

PD

Parkinson’s disease

PPMS

primary progressive MS

R_ac

accumulation ratio

RRMS

relapsing remitting MS

S1P

sphingosine 1-phosphate

S1PR

S1P receptor

SGPL1

S1P lyase

snRNA-seq

single nucleus RNA-seq

SOCS2

suppressor of cytokine signaling 2

SPHK

sphingosine kinase

SPMS

secondary progressive MS

SPTLC

serine palmitoyltransferases

TCA

tricarboxylic acid

TCN2

transcobalamin 2

TEMSO

Teriflunomide Multiple Sclerosis Oral

Tg

transgenic

TM

transmembrane

TRANSFORMS

Trial Assessing Injectable Interferon versus FTY720 Oral in Relapsing–Remitting Multiple Sclerosis

T_reg

regulatory T..

Data availability

No data was used for the research described in the article.