The impact of sphingosine-1-phosphate receptor modulators on COVID-19 and SARS-CoV-2 vaccination

Background Sphingosine-one phosphate receptor (S1PR) modulation inhibits S1PR1-mediated lymphocyte migration, lesion formation and positively-impacts on active multiple sclerosis (MS). These S1PR modulatory drugs have different: European Union use restrictions, pharmacokinetics, metabolic profiles and S1PR receptor affinities that may impact MS-management. Importantly, these confer useful properties in dealing with COVID-19, anti-viral drug responses and generating SARS-CoV-2 vaccine responses. Objective To examine the biology and emerging data that potentially underpins immunity to the SARS-CoV-2 virus following natural infection and vaccination and determine how this impinges on the use of current sphingosine-one-phosphate modulators used in the treatment of MS. Methods A literature review was performed, and data on infection, vaccination responses; S1PR distribution and functional activity was extracted from regulatory and academic information within the public domain. Observations Most COVID-19 related information relates to the use of fingolimod. This indicates that continuous S1PR1, S1PR3, S1PR4 and S1PR5 modulation is not associated with a worse prognosis following SARS-CoV-2 infection. Whilst fingolimod use is associated with blunted seroconversion and reduced peripheral T-cell vaccine responses, it appears that people on siponimod, ozanimod and ponesimod exhibit stronger vaccine-responses, which could be related notably to a limited impact on S1PR4 activity. Whilst it is thought that S1PR3 controls B cell function in addition to actions by S1PR1 and S1PR2, this may be species-related effect in rodents that is not yet substantiated in humans, as seen with bradycardia issues. Blunted antibody responses can be related to actions on B and T-cell subsets, germinal centre function and innate-immune biology. Although S1P1R-related functions are seeming central to control of MS and the generation of a fully functional vaccination response; the relative lack of influence on S1PR4-mediated actions on dendritic cells may increase the rate of vaccine-induced seroconversion with the newer generation of S1PR modulators and improve the risk-benefit balance Implications Although fingolimod is a useful asset in controlling MS, recently-approved S1PR modulators may have beneficial biology related to pharmacokinetics, metabolism and more-restricted targeting that make it easier to generate infection-control and effective anti-viral responses to SARS-COV-2 and other pathogens. Further studies are warranted.

In contrast fingolimod (Figure 1), which is a lymphocyte migration inhibitor, was not associated with a worse outcome following natural SARS-CoV-2 infection [22][23][24][25]35]. Perhaps consistent with this, the majority of people on fingolimod appeared to seroconvert, albeit sometimes with lower antibody titres following a natural SARS-CoV-2 infection [36][37][38][39][40][41]. Similarly, modest reductions in antibody titres notably with other vaccines, have been seen in fingolimod-treated individuals [42][43][44]. However, fingolimod consistently inhibits both seroconversion and peripheral blood T cell responses following SARS-CoV-2 vaccination, although there was some variability between studies in part due to the: individuals, viral strain, past infection, vaccine type; time of assay relative to infection/vaccination, different assays and different functional and physical targets [12,13,30,31,45,46]. Importantly, this had biological impact because in comparison to other MS treatments, use of either fingolimod or CD20-depleting antibodies was sometimes associated with COVID-19 disease breakthrough following vaccination [47][48][49][50]. Importantly, this was seen even before the time when circulating SARS-CoV-2 variants of concern, notably omicron variants, required high vaccine-induced antibody titres to protect from infection compared to the initial SARS-CoV-2 variants [16,[50][51][52]. As this breakthrough was associated with agents with poor seroconversion, it supports the view that viral neutralizing antibodies are particularly important in preventing infection/re-infection [10,50]. This indicates that the clinical responses observed can be attributed to the chemistry and biology of the different agents. It therefore remained to be seen whether there could be any differences between fingolimod and the other sphingosine-one-In the United States of America: fingolimod, siponimod, ozanimod and ponesimod all have a similar utility and are licensed for clinically-isolated syndrome, relapsing MS and active secondary progressive MS [53][54][55]57]. These are all characterised by bout attacks and/or new T2 or gadolinium enhancing T1 lesion formation [2]. However, in Europe, differences in the licensing exist that may influence use in practice. As such, fingolimod is a second-line treatment for highly-active relapsing MS, siponimod is licenced for active, secondary progressive MS, whereas both ozanimod and ponesimod have recently been approved as first-line treatments for active relapsing MS [53][54][55]57]. Fingolimod has been used and studied most extensively and forms the basis for most COVID-19-related information. Fingolimod exhibits a long half-life and so peripheral lymphocyte recover slowly after treatment cessation (Table 1). However, once cells repopulate, disease may relapse and therefore this requires an appropriately-timed switch to an alternative treatment [60].
Although there were initial recommendations to stop fingolimod treatment following SARS-CoV-2 infection, the virus would be naturally cleared before therapeutic levels have been eliminated [10].
This probably prompted some people to switch to S1PR modulators with a more rapid clearance, in case people exhibited COVID-19 symptoms.
Siponimod, ozanimod and ponesimod have relatively short half-lives compared to fingolimod (Table   1)  Furthermore, as ozanimod is metabolised to compounds with long half-lives, it will exhibit similar issues to fingolimod in terms of slow lymphocyte recovery following treatment cessation (Table 2) [54,63]. In contrast ponesimod has a relatively short life, with a relatively rapid repopulation of lymphocytes (Table 1) [64] and therefore could offer advantages following infection or in using short treatment breaks to promote better vaccination responses. However, information on the time window before disease reactivation occurs after cessation is currently limited [65*], but effective vaccination following discontinuation that avoids disease breakthrough seems feasible with shorthalf live agents [66,67*,68*].
Fingolimod binds to S1PR1, S1PR3, S1PR4, and S1PR5, which have distinct tissue distributions that will impact on its function ( Figure 1; Table 2) [69]. However, it is clear that the major therapeutic impact on lymphocyte migration is mediated by S1PR1 [70]. Ponesimod targets largely S1PR1, with lower affinities and partial activity for other receptors, notably S1PR5, and inhibits relapsing MS ( Figure 1, Table 2) [55,71]. Siponimod and ozanimod both target S1P1R and S1PR5, to notably to limit perceived S1PR3-mediated side effects encountered with fingolimod [5,70,72]. They also target S1PR5 on oligodendrocytes and their precursors to potentially better influence remyelination ( Figure 1, Table 2) and do not require the action of phosphorylating S1P kinases for activity [73][74][75]. Although remyelination effects are largely unproven in MS, oligodendrocyte actions are unlikely to be of major importance to COVID-19, therefore targeting this pathway is unlikely to impact on SARS-CoV-2 infection or vaccination responses. However, S1PR5 modulators may affect natural killer cell function, which may influence COVID-19 biology [76][77][78]. However, given that the impact of these agents on natural killer cell numbers is often minimal [8] and that fingolimod, which also targets S1PR5 and is not associated with a worse prognosis following SARS-CoV-2 infection [22,23], indicates that likewise, siponimod, ozanimod and ponesimod are unlikely to cause a worse prognosis following COVID-19 infection. Indeed, this appears to be the case in the few individuals that are reported to be infected with SARS-CoV-2 who are taking these drugs [35,79,80*,81*].
Natural killer cells are unlikely to exhibit a major effect on the generation of T and B cell responses and this suggests that siponimod, ozanimod and ponesimod may behave similarly regarding vaccination.
In addition, S1PR1 is also expressed by the vascular system and brain endothelial cells, hence S1PR modulation can further inhibit leucocyte trafficking into the CNS to prevent disease [85,86]. This may be further influenced by astrocytic S1PR1/S1PR3 activity, as astrocytes are known to be involved in blood-brain barrier formation and targeting astrocytes probably serves to help inhibit Although it is clear that CD4, CD8 and CD19 expressing T and B lymphocytes are markedly inhibited following S1P1R internalization, it is evident that there is differential inhibition of lymphocyte subsets notably due to S1PR1 and CCR7 chemokine receptors Importantly, fingolimod, like most other MS-disease modifying treatments, targets the adaptive immune response and does not induce marked changes to the peripheral innate immune response, which are sentinels located within the affected tissues and appear to be central to SARS-CoV-2 removal [10,98,100]. This is facilitated by the cytotoxic T cell response and the subsequent generation of cytopathic and neutralizing antibodies that can help protect against re-infection [10,20,50]. Antibody responses can be generated within the lymphoid tissue from immature and naïve B cells, which seem to express many S1PR ( Figure 2) and thus may not require re-circulation to tissues to induce antibody-producing plasma cells that could facilitate removal of the SARS-CoV-2 virus [101,102]. However, S1P1R is involved in the release of immature B cells from bone marrow and B cell migration within lymphoid tissues that involves shuttling of B cells from marginal zones and B cell follicles using S1PR1 and CXCR5, responding to CXCL13 [90, 103,104]. This could contribute to the reduction of SARS-CoV-2 B cell responses as seen with fingolimod [12,13].
Vaccine responses during fingolimod treatment are blunted compared to untreated individuals with a seroconversion rate of 60.2% in a meta-analysis of n=785 people treated with S1PR modulators, largely taking fingolimod n=764 [12]. This was supported in an additional meta-analysis examining only fingolimod treatment, which reported an antibody response in n=160/220 (72.7%) vaccinated and n=152/198 (76.8%) mRNA-vaccinated, fingolimod-treated individuals (Table 3) (Table 3). This suggests an important impact of S1PR1 on vaccine-induced antibody responses.
Interestingly, although the numbers of studies on non-fingolimod, S1PR modulators are relatively small and the differences observed may be part of the variability between studies, including the nature of the vaccines and the immune-response detection assays used, it seems that there are better seroconversion rates seen in the majority of people treated with siponimod, ozanimod and ponesimod (Table 3). This contrasts with studies on fingolimod that often report that the minority of people seroconvert following vaccination [12,13,67*,105-109,110*,111*,112*,113*]. This could suggest that S1PR3 and S1PR4, which are widely expressed by the immune system ( Figure 2), contribute to lower antibody titres following vaccination, as suggested by the underlying biology.
Consistent with other studies [11], the level of seroconversion is influenced by the nature of administered vaccine ( Therefore, the demographics of individuals vaccinated will potentially influence study outcome. Furthermore, it could also be argued that the possible subtle differences reported between fingolimod and the more recently approved variants may relate to biology created by the changing circulating SARS-CoV-2 variants of concern and thresholds of immunity required for immune protection [50,114]. However, the information reported here was largely based on full vaccination (typically two cycles) with the original index-SARS-CoV-2 virus-based vaccines. This was also collected during periods when SARS-COV-2 alpha and delta variants of concern were prevalent [115] and most people appeared to be natural-infection naïve (Table 3) and respond consistently over time and between vaccine cycles [30,31,116]. Therefore, the circulating SARS-CoV-2 variant, may have had limited impact on the vaccine responses seen (Table 3).
However, it is likely that the threshold of assay detection of SARS-CoV-2 antibodies is important in determining the level of seroconversion. Therefore, it is perhaps of interest that the high frequency of seroconversion seen notably in ozanimod-treated individuals was largely detected in studies using the SARS-CoV-2 receptor binding domain ECLIA Elecsys® assay (Table 3)  perhaps other S1PR modulators, may allow a higher antibody titre to develop and thus create a potentially more effective vaccination response. Larger studies, meta-analysis of numerous smaller studies [12,13] or ideally clinical or experimental head to head studies will be required to determine whether there are indeed any real differences between the vaccine responses of the different S1PR modulators. However so far, this idea is suggested by some recent studies that contain responses to multiple different S1PR modulators (Table 3) 116,118], the threshold level needed for immune protection requires further study and may vary with time, notably related to the changing circulating SARS-CoV-2 variant of concern, as seen with antibody protection from infection [11,50,114].

Influence of non-sphingosine-1-phopshate 1 receptors controlling antibody responses.
None of the current S1PR modulators target S1PR2 (Table 2) and this may be beneficial for vaccine responses as SP1R2 and the CXCR5 chemokine regulate the localization of follicular helper T cells into the B cell follicles to promote antibody responses and are particularly important for germinal centre reactions, where they can function to help antibody responses to novel antigens, through production of cytokines and co-stimulatory molecules [119,120]. Furthermore, S1PR2 is expressed by B cells within follicles ( Figure 2) and this regulates entry into a plasma cell or recycling germinal centre cell fate and maintains the homeostasis of germinal centre B cells [121][122][123][124][125]. As such, S1PR2 may inhibit some S1PR1-mediated functions [126].
In contrast, it is also possible to speculate that potential differences between fingolimod and the other S1P modulators could occur due to an activity on S1PR3. Indeed, it has been suggested that S1PR3 controls B cell function [103,127,128]. Notably, this receptor has been associated with the development of progenitor cells; positioning of immature B cells within bone marrow sinusoids and migration of B cells within the bone marrow and lymphoid tissues [128][129][130]. Importantly, it is involved in B cell capture of antigens in the marginal zones and shuttling to B cell follicles for the development of antibodies [103,127]. However, this view may only reflect the case in rodents, as there is a paucity of evidence to suggest an S1PR3-mediated B cell activity in humans. As such, there may be subtle differences in the migration cues between rodent and humans [131].
Importantly, whilst mouse B cells express S1PR3 [128,132] it appears that human B cells express limited S1PR3 mRNA [132,133] ( Figure 2). This may have parallels with the cardiac side-effect activity that was originally attributed to S1PR3, based on rodent studies [5,70]. It is now evident that these issues are mediated by S1PR1 in humans [5]. As such all approved S1P modulators, including those with no/limited S1PR3 activity can induce cardiac arrhythmias [53-55,57]. However, S1PR3 can be pathologically regulated within lymphoid tissues and may have some element to play in B cell development, notably as it has been suggested that S1PR3 contributes to vascular and dendritic cell function within B cell areas [129, [134][135][136]. As such S1PR3 can control dendritic cell migration into secondary lymphoid tissues and monocyte activity [137,138] and may influence the generation of primary immune responses that ultimately lead to a vaccination response ( Figure 2). Indeed, it is evident that S1PR4 is widely expressed by immune cells subsets (Figure 2), including platelets [139] (Figure 2) and that S1PR4 modulation may mediate effects on T and B cell migration and may modulate S1PR1 function [126,[140][141][142][143], S1pr4-gene deficient mice have normal lymphocyte numbers and regular architecture of secondary lymphoid organs [144]. In contrast, there was a marked impact of S1PR4 depletion on dendritic cell migration and cytokine secretion leading to reduced Th17 T-cell differentiation and inhibition of mouse and human dendritic cell activity [144][145][146]. There is S1PR-mediated control of dendritic (Langerhans) cell migration from the skin [147]. This could impact antigen-presenting cell function in vaccine-induced responses and may affect progenitor development in the bone marrow that limit the occurrence of dendritic cells [146,148]. Furthermore, S1PR4 is required for plasmacytoid dendritic cell differentiation [148].
Plasmacytoid dendritic cells secrete high levels of interleukin-6 and type I interferons in response to infection, which are protective against the SARS-CoV-2 virus and can help induce the differentiation of B cells into IgG-secreting plasma cells [149][150][151]. The innate immune system and endothelial express the S1P kinases, notably SPHK1 and can respond to fingolimod [152,153].
However, it is important to note that the affinity of fingolimod for S1PR4 may be low in some functional assays (Table 2) and therefore differences between this and other agents may only be incremental. Likewise, it is possible that differences between fingolimod and other agents could relate to the requirement for phosphorylation that may vary due to the potential differential expression of SPKH1 and SPKH2 in tissues influencing fingolimod activity [69,73] The potential differences observed between seroconversion after infection or vaccination during fingolimod use may simply reflect variability between studies and it is important to note in both cases the level of response is often diminished compared to untreated individuals. However, whilst vaccination is induced via the skin, natural infection with SARS-COV-2 occurs via the pulmonary and mucosal surfaces over time and so the range of antigen-presenting cells and lymphoid tissues involved may be broader and may account for potentially better seroconversion in these immunosuppressed people [151][152][153]. We hypothesise that the apparent, subtle, differences between fingolimod and the newer generation S1PR modulators are most likely due to differences on SIPR3 and notably SP1R4 modulation on the germinal centre formation and function, which are critical for neoantigen antibody responses (Figure 3).  [159][160][161][162]. There, is limited expression of S1PR receptors, except for S1PR4, by polymorphonuclear neutrophils and neutropenia is not typically associated with S1PR modulation and may control egress from inflamed tissues [8, 163,164]. Although fingolimod, siponimod and ozanimod could perhaps influence natural killer (NK) cell function secondary to effects notably on S1P5R, and in the case of fingolimod S1P4R, associated migration [76,77]. Again, peripheral NK depletion is modest following S1PR modulation [8, 165,166]. This probably reflects a more limited activity on the CD16+, CD56 dim NK cell subset that are dominant in the periphery and are important for antibodymediated cytotoxicity [8, 164,165]. There is therefore limited reason to believe that S1PR would directly influence viral activity via antibodies or a direct effect on viral activity. However, poor viral elimination in SARS-CoV-2 neutralizing antibody-treated and immunosuppressed people can support the selection of immune escape variants that can render SARS-CoV-2-specific neutralizing antibodies such as casirivimab/imdevimab, tixagevimab/cilgavimab and sotrovimab to become rather ineffective as the SARS-CoV-2 virus evolves [17,19,114,159,160,162,167]. Therefore, alternative strategies are needed.

Small molecule anti-viral agents
have been identified and/or generated to block essential viral function that are distinct from SARS-CoV-2 receptor-binding domain targeting antibodies.
These chemicals currently include: remdesivir infusions for hospitalised individuals [168] and oral molnupiravir [169,170*,171] and ritonavir-enhanced nirmatrelvir [171,172] that are used for COVID-19 prophylaxis or for a rapid, post-infection application [172]. Remdesivir prodrug is metabolised by the cytochrome P450 CYP3A4 variant leading to increased bioavailability of other CYP3A4 substrates that include some MS-related drugs [173,174]. Molnupiravir, is another prodrug and is rapidly converted to active drug in plasma, which is excreted with limited hepatic metabolism [175]. This should not unduly influence S1PR modulators. In contrast, ritonavir-enhanced nirmatrelvir may augment the pharmacokinetics of many drugs, including nirmatrelvir, because ritonavir is a potent inhibitor of CYP3A4, CYP2D6 and a number of drug transporters [176].
Ponesimod is metabolised by a number of cytochrome enzymes including: CYP2J2, CYP3A4, CYP3A5, CYP4F3A, and CYP4F12 without major contribution by any single enzyme and as such is considered unlikely that any major impact on ponesimod metabolism will occur [177]. Likewise, fingolimod is metabolised mainly by CYP4F2 and CYP4F3B enzymes and specific CYP3A4 inhibition did not impact fingolimod distribution [178]. Ozanimod is extensively metabolised, notably by CYP2C8 and to a small extent by CYP3A4, as such CYP3A4 inhibition had limited impact on ozanimod levels [179]. However, ozanimod and its active metabolites (notably CC112273) are substrates for p-glycoprotein (ABCB1) and breast cancer resistance protein (ABCG2), which can be inhibited by ritonavir and could increase ozanimod exposure. However, this would be within safe limits observed in ozanimod trials [176]. Siponimod carries a warning that it should be avoided with CYP2C9 and CYP3A4 inhibitors [180]. Whilst this may lead some people to avoid use of nirmatrelvir/ritonavir, the contraindication is based on the combination of both CYP2C9 and CYP3A4 inhibition, whereas CYP3A4 inhibition alone exhibits limited impact on siponimod concentrations [61]. Likewise, as with ozanimod, any enhanced levels of siponimod would be within safe limits tested in trials. Whilst ritonavir can induce CYP2C9 [176], problems will be avoided as people sensitive to the influences of CYP2C9 alleles, notably CYP2C9*3 homozygotes as poor metabolizers, are excluded from siponimod treatment as part of drug screening and those with CY2C9*2*3 and CYPC9*1*3 genotypes are differentially dosed [180]. Currently there is no information to determine whether CYP2C9 genotypes influences vaccine responses. However, given the short course of anti-viral and the half-lives of the agents, there does not seem to be a major risk-benefit barrier to offering antiviral agents to people taking S1PR modulators, if considered to be useful.

Conclusions:
The success of fingolimod as a therapeutic in MS, has led to the development of a number of other S1PR modulators for the treatment of MS and other immune conditions, which have different binding and pharmacokinetic activities. Whilst their selection may in part be hampered by their licence/label, from a biological perspective these differences can influence: dosing and activity; the strategy required to switch to other agents in case of treatment failure and, as highlighted here, the ability to facilitate vaccination and infection control. Whilst vaccination issues have become evident due to the COVID-19 pandemic, annual vaccinations such as with the current influenza vaccines, mean that this element of S1P and S1PR biology will remain an issue.
Therefore, ways to improve/optimize vaccination responses will be beneficial  [185,186]. This was seen in animals treated with ponesimod [68*] and people given other vaccines during siponimod treatment and potentially in a small number of people given the SARS-CoV-2 vaccine [66,67*]. This is consistent with the observations that vaccine-responsiveness correlated with lymphocyte numbers, notably CD19+, CD27-B cell numbers [158*,184]. However, the primary aim should be to effectively control disease and with anti-viral agents available, potentially less virulent and vaccine immune-escape SARS-COV-2 variants circulating, thought/evidence is needed to justify any change to standard practice [187]. However, more extensive studies to optimise responses whilst ensuring efficacy and safety are needed to avoid disease-breakthrough and cardiac issues for all S1P modulators [188]. This is important as the power to rapidly generate vaccines to infections is likely to increase in the future.  Information about the pharmacokinetics of sphinogsine-1-phosphate receptor (S1PR) modulators were obtained from the Summary of Medical Product Characteristics from the European Medicines Agency website. CC112273 is a metabolite of ozanimod. Cmax maximum concentration Table 2. Receptors specific cities of approved S1PR modulators Treatment Sphingosine-1-phophate (S1P) receptor binding affinities S1PR1 S1PR2 S1PR3 S1PR4 S1PR5 Reference The S1P1R binding affinities of sphinogsine-1-phosphate (S1P) and the S1PR modulators were extracted from the literature. The results report the a IC50 or b,c EC50 binding levels using either a competitive radio-ligand binding, b gamma GTPS or c beta-arrestin binding assays. d Maximal effect at 10,000nM on S1PR4/S1PR5 was 18/42%, respectively of the effect on S1P response, so was not only less efficacious but also less potent than S1P. The standard daily doses are: 0.5mg fingolimod, 1mg siponimod, 0.92mg ozanimod or 20mg ponesimod. Table 3. Influence of S1PR modulation on SARS-CoV-2 vaccine responses.  The S1PR mRNA expression distributions from cells and tissues were extracted from Affimetrix RNAseq data in the human Primary Cell Atlas or the mouse GeneAtlas MOE430 gcrma datasets at www.biogps.org, using the indicated S1PR-specific probes. The results represent the mean ± standard error of the mean expression of 2-21 individual samples of normalised expression data.
arbitrary units. The human natural killer cells subset examined, expressed CD56, CD62 antigens.