Actions of a picomolar short-acting S1P₁ agonist in S1P₁-eGFP knock-in mice.

Sphingosine 1-phosphate receptor 1 (S1P(1)) is critical for lymphocyte recirculation and is a clinical target for treatment of multiple sclerosis. By generating a short-duration S1P(1) agonist and mice in which fluorescently tagged S1P(1) replaces wild-type receptor, we elucidate physiological and agonist-perturbed changes in expression of S1P(1) at a subcellular level in vivo. We demonstrate differential downregulation of S1P(1) on lymphocytes and endothelia after agonist treatment.

S1P1 (Edg1) is a high-affinity G protein-coupled receptor (GPCR) for the zwitterionic lysophospholipid Sphingosine 1-Phosphate (S1P)[1] and is essential for vascular development[2,3] and integrity[4], and modulates both lymphocyte development and recirculation[5]. Selective agonists of S1P1 induce effective lymphocyte sequestration of both T and B cells within the lymph node[6,7], and have been investigated as potential therapeutics in transplantation[8] and multiple sclerosis. FTY720 (fingolimod)[9], a prodrug whose phosphorylated form is an agonist of S1P1,3–5, but is thought to strongly downmodulate S1P1 expression[10], has recently received regulatory approval as the first oral treatment of relapsing, remitting multiple sclerosis[11-13].

Here we have generated both a short-acting, low picomolar agonist of S1P1 and a mouse model where, rather than modulating S1P1 expression[10,14-16], enhanced green fluorescent protein fused to S1P1 protein (S1P1-eGFP) replaces wild-type S1P1 in order to better understand both normal and agonist-perturbed expression of S1P1 (Detailed methods are shown in Supplementary Methods). Utilization of mice expressing S1P1-eGFP enabled detection of S1P1 protein ex vivo by both quantitative biochemical and flow cytometric analysis at high resolution, circumventing limitations of antibody detection and demonstrating differential regulation of S1P1 in lymphocytes and endothelial cells following short-duration agonist treatment[17]. Lymphocytes exhibited rapid, sustained downregulation, whereas endothelia demonstrated slower loss of S1P1. Additionally, mice expressing S1P1-eGFP allowed in vivo visualization of changes in subcellular localization of S1P1-eGFP using two-photon microscopy.

While exploring the chemical space around the allosteric S1P1 agonist, CYM-5442 (1)[7], we noticed that adding S1P-like headgroup interactions enhanced potency by 100-fold and shifted ligand binding into an orthosteric binding mode, leading to the short-acting agonist RP-001 (2) (Fig. 1a, Supplementary Results, Supplemental Scheme 1). RP-001 activated S1P1 in vitro with an EC50 of 9pM, while having little activity on S1P2-S1P4 and only moderate affinity for S1P5 (Supplemental Table 1). No differences were observed between the racemate and the resolved S- enantimoer across all five S1P receptors, demonstrating no chiral preference. RP-001, in contrast to CYM-5442, was competitive with the selective S1P1 antagonist W146 (3) (Fig. 1b). RP-001 induced sustained signaling while in internalized vesicles, as reported for FTY720 (4)[18] and in contrast to the natural ligand S1P (5) (Fig. 1c). It also induced dose-dependent internalization and polyubiquitination of S1P1 in vitro similar to other S1P1 agonists[7] (Supplemental Fig. 2). In vivo, RP-001 caused dose-dependent rapid lymphopenia with an EC50 of 0.03 mg/kg, where maximal lymphocyte sequestration from blood occured within 2 hours and returned to untreated levels by 8 hours following treatment with 0.3mg/kg (Figs. 1d,e). The short duration of lymphopenia and short half life rendered RP-001 unsuitable for clinical use, but allowed examination of changes in S1P1 expression during both induction and recovery of lymphopenia, an ability unique to such a short-acting agonist.

Figure 1

RP-001 is an orthosteric, short-duration S1P1 selective agonist

(a) Chemical structures of CYM-5442 and its derivative, RP-001. (b) Left graph represents dose- response curves of RP-001 activation of S1P1, expressed as a percent of S1P-induced activation. Right graph is a Schild Plot demonstrating competition between RP-001 and the S1P1 antagonist W146. (c) RP-001 induces sustained S1P1 signaling in internalized vesicles. S1P1 CRE-β-lactamase expressing CHO-K1 cells were pretreated with 1μM of the indicated compounds for 1 hour, washed, and rested for 5 hours. Cells were then treated with 2μM forskolin with or without 10nM FTY720 as indicated, and cyclic AMP induced beta lactamase expression was detected by cleavage of CC4-AM. (d) RP-001 elicits dose-dependent lymphopenia of CD4+ T cells. (e) RP-001 induces acute lymphopenia with rapid recovery to untreated levels. Graph for d and e display number of CD4+ T cells per mL of blood, while e additionally displays concentration of RP-001 found in the blood as quantified by mass spectrometry with a dashed line. The bottom horizontal line represents the lower limit of detection of RP-001 (1nM). All intravital experiments are representative of at least 3 experiments, 3–4 mice per group per experiment and are presented as mean ±SEM.

A parallel genetic approach was taken to the system-wide analysis of physiological S1P1 function. To avoid the pitfalls of gene deletion, or the use of receptors with altered signaling properties, we generated Edg1 mice by inserting the coding sequence for eGFP directly 3′ to the coding sequence of Edg1, leading to expression of S1P1-eGFP as a fusion protein under normal regulatory mechanisms (Fig. 2a, Supplemental Fig. 3a–c). We observed Mendelian ratios of wild-type, heterozygous, homozygous mice, and generated hemizygous mice possessing only a single copy of the Edg1 allele (Supplemental Fig. 3d), demonstrating that S1P1-eGFP was fully functional. Direct in-gel fluorescence and immunoblotting for eGFP from homogenized tissues showed that S1P1-eGFP was abundantly expressed across tissues, with highest expression in brain and lung (Fig. 2b, Supplemental Fig. 4a). We observed roughly equal amounts of both wild-type S1P1 and S1P1-eGFP protein in heterozygous mice by using an antibody specific to a carboxyl-terminal portion of the S1P1 protein on immunoprecipitated brain tissues (Fig. 2c), and found that S1P1-eGFP was N-glycosylated in vivo (Fig. 2d), suggesting normal surface export.

Figure 2

Expression and function of S1P1-eGFP in Edg1 mice

(a) Schematic for the final locus for mice expressing Edg1-eGFP from the Edg1 locus. E1 and E2 represent exons 1 and 2 of Edg1, triangle represent loxP recombination site (b) S1P1 is broadly expressed across many tissues by Western blot for eGFP from tissues listed under non- denaturing conditions. (c) S1P1-eGFP is expressed similarly to wild-type S1P1. Brain lysates were immunoprecipitated with either an antibody specific to the carboxyl-terminus of S1P1 or an antibody specific to GFP. These lysates were separated by SDS-PAGE, transferred onto membrane, and incubated with antibodies specific to either S1P1 or GFP. (d) S1P1-eGFP is N- Glycosylated in vivo. Homogenized brain or lung tissues from Edg1 mice were immunoprecipitated with an antibody specific to GFP then incubated with or without PNGase F. The smaller size of S1P1-eGFP following incubation with PNGase F indicates loss of N-linked sugars from S1P1-eGFP. (e) T and B lymphocytes from Edg1 and Edg1 mice are sequestered following S1P1 agonist RP-001 treatment. Lymphocyte counts were taken two hours after intraperitoneal injection of either 0.3mg/kg RP-001 or vehicle. (f) Enhancement of vascular leakage following S1P1 antagonism. Graphs represent leakage of Evans’ Blue Dye into the lungs 1 hour after treatment with 3mg/kg W146 or vehicle intraperitoneally, standardized to wild-type mice treated with W146. Graphs in e and f represent mean +/− SEM for 2 pooled experiments, 3–4 mice per group per experiment. *p<0.05, **p<.01, ***p<0.001 for e and f using unpaired t-test.

Wild-type and homozygous mice exhibited indistinguishable sequestration of T and B cells from the blood two hours after treatment with 0.3mg/kg RP-001 (Fig. 2e). Antagonism of S1P1 by the selective antagonist W146 causes significant leakage of fluid into the lungs[4]. Homozygous knock-in mice showed similar lung leakage two hours following treatment with 3 mg/kg W146, with no difference in the basal lung leakage between wild-type and homozygous animals (Fig. 2f). The equivalent effects of S1P1 activation and inactivation in Edg1 mice and wild- type mice indicate that S1P1-eGFP substitutes fully for wild-type S1P1 in vivo.

The role of S1P1 in lymphocyte egress is proposed to be due to S1P1 expression on lymphocytes[19] and/or lymphatic sinus endothelial cells[20]. S1P1-eGFP was expressed on CD4 and CD8+ T cells as well as CD19+ B cells within the lymph node (Fig. 3a), and was expressed on mature thymocytes (Supplemental Fig. 5). The eGFPhi population found in CD8+ T cells results from accumulation of cleaved eGFP in peripheral CD8+ T cells, which may indicate unique regulation of S1P1 in CD8+ T cells (Supplemental Fig. 4e,f). S1P1-eGFP was expressed on lymphocytes similarly in both blood and secondary lymphoid tissues, indicating a lack of substantial receptor degradation even in S1P rich blood (Supplemental Fig. 6). Contrary to lymphocytes, S1P1-eGFP was not expressed on myeloid-lineage cells within the spleen, as cells from Edg1 and Edg1 mice exhibited indistinguishable fluorescence intensities. (Supplemental Fig. 7b).

Figure 3

RP-001 causes changes in lymphcytic and endothelial S1P1-eGFP expression and localization

(a) Flow cytometry histograms for eGFP fluorescence on specified lymphocyte populations isolated from lymph nodes of Edg1 (colored) and wild-type (gray) mice at indicated timepoints following IP injection of RP-001. (b) Lymphocytes were homogenized and mixed 1:1 with 90% sucrose. These lysates were placed in centrifuge tubes and overlaid sequentially with equal volumes of 35% sucrose followed by 5% sucrose. Following centrifugation at 100,000×g for 16 hours, twelve equal volume fractions were collected and Western blotted for GFP, Caveolin-1 (Cav1), Early Endosomal Antigen 1 (EEA-1). c) Equal numbers of lymphocytes were isolated and centrifuged as above. S1P1-eGFP in Cav1-rich fractions is highly preserved following treatment with RP-001, but endosomal-associated S1P1-eGFP is lost, particularly at 1mg/kg. (d) Left panel represents scatter plot of GP38 and CD31 expression on CD45.2− cells isolated from Collagenase/DNaseI digested lymph nodes. Right panels are histograms for eGFP fluorescence on blood endothelial cells (BECs), lymphatic endothelial cells (LECs), and fibroblast reticular cells (FRCs). (e) Flow cytometry histograms for eGFP from lung BECs and LECs from the same mice as in a. (f) Snapshots from intravital two-photon microscopy of lymph nodes treated as indicated. Internalization of S1P1-eGFP is not detected at 0.1mg/kg at timepoints up to 80 minutes. Flow cytometry plots in a and e are representative of 3 experiments, 3 mice per group per experiment.

The ability to detect total S1P1-eGFP expression allowed us to examine the effect that RP-001 had on S1P1-eGFP expression. S1P1-eGFP expression on lymphocytes in the lymph node was significantly decreased two hours following treatment with 0.1 mg/kg RP-001, suggesting degradation of S1P1-eGFP (Fig. 3a, Supplemental Table 2). Lymphocytic S1P1-eGFP did not fully recover to untreated levels by eight hours after treatment, despite restoration of normal lymphocyte recirculation, indicating that decreased lymphocytic S1P1 is not the driving force behind maintenance of lymphocyte sequestration by RP-001. 0.2mg/kg FTY720 treatment also induced S1P1-eGFP degradation on lymphocytes corresponding with sequestration of T and B lymphocytes from the blood, which partially recovered by 48 hours following treatment (Supplemental Fig. 8a). Due to the carboxyl-terminal location of eGFP, we were unable to distinguish by flow cytometry whether S1P1-eGFP was on the cell surface or internalized, reinforcing the need for widely available extracellular S1P1 antibodies, the development of which continue to be troublesome.

We examined whether RP-001 was able to induce changes in the subcellular localization of S1P1-eGFP on lymphocytes by sucrose gradient density centrifugation, similar to previous work with COS7 cells expressing tagged S1P1 in vitro[21]. S1P1-eGFP on lymphocytes from untreated mice was found predominantly in Caveolin-1-rich (Cav1) fractions, with significant S1P1-eGFP being found in fractions that colocalized with the early endosomal marker EEA-1 (Fig. 3b), demonstrating significant receptor reserve in Cav1-rich fractions as well as constitutive activation and internalization into endosomal compartments. Following treatment of mice for 2 hours with 0.1 mg/kg or 1 mg/kg RP-001, lymphocytic Cav1-associated fractions demonstrated no significant loss of S1P1-eGFP, but endosomal fractions exhibited a loss of S1P1-eGFP, suggesting lysosomal sorting and degradation of S1P1-eGFP (Fig. 3c, Supplemental Fig. 9). Cav1-associated S1P1-eGFP could potentially be heterogeneous, consisting of both cell-surface and internalized S1P1-eGFP, which is likely shifted by RP-001 treatment.

We next investigated S1P1-eGFP expression on endothelial cells within the lymph node by flow cytometry following enzymatic digestion[22], finding that both CD31+GP38− blood and CD31+GP38+ lymphatic vessels were very abundant in S1P1-eGFP (Fig. 3d). S1P1-eGFP was readily observed on both lymphatic and blood endothelial cells within the lymph node by confocal microscopy, however lymphocytes were largely below the level of detection (Supplemental Fig. 10). Expression of S1P1-eGFP on high endothelial venules (HEVs) was highly expressed on the luminal side of the cell, ideally situated for sensing changes in S1P levels in the blood. Treatment of mice with RP-001 caused changes in S1P1-eGFP expression on both blood and lymphatic endothelial cells within the lung, demonstrating a small loss of S1P1- eGFP two hours following 0.1mg/kg RP1138 treatment, but substantial degradation eight hours following treatment, when no agonist was present in circulation (Fig. 3e, Supplemental Table 3).

The generation of mice expressing S1P1-eGFP also allowed us to directly observe S1P1 expression in vivo using two-photon microscopy. Initial experiments confirmed our ex vivo findings, demonstrating expression on different types of endothelial cells within the lymph node and relatively little expression on lymphocytes other than motile eGFPhi CD8+ T cells (Supplemental Fig. 11a, Supplemental Video 1). Treatment of mice with 0.1mg/kg RP-001, a dose that causes significant lymphocyte sequestration, did not lead to internalization of S1P1- eGFP on endothelial cells within the lymph node within 80 minutes (Fig. 3f, Supplemental Fig. 11b). By explanting lymph nodes and incubating them in the presence of vehicle alone (Supplemental Video 2) or 250nM RP-001, a concentration 10 fold higher than the Cmax of RP- 001 in vivo at 0.1mg/kg, we observed rapid ligand-induced internalization of S1P1-eGFP on endothelial cells (Supplemental Fig. 3c, Supplemental Video 3), indicating that while higher concentrations of RP-001 could achieve full receptor occupancy on endothelial cells and cause significant internalization of S1P1, this was not required for induction of lymphopenia.

Clarifying S1P1 expression also allows insights into how S1P1 agonist treatment can affect the pathogenesis of both multiple sclerosis[23] and influenza virus infection[24,25] by allowing direct analysis of S1P1 expression on the variety of cell types that are involved in these pathologies, and how this expression changes following treatment with S1P1 modulatory drugs. The study of the complex regulation of S1P1 requires a full set of methodologies to provide a clearer view of the multiple roles that S1P1 can play in many physiological systems. Edg1 mice, together with continued optimization of S1P1 modulatory compounds, can help validate proposed roles of S1P1, find novel roles for S1P1 in physiological and pathophysiological systems, and aid in the clinical treatment of diseases such as multiple sclerosis.