Structure-Activity Profiling of Indazole-based Derivatives as Spinster Homolog 2 (Spns2)-dependent S1P Release Inhibitors

Abstract

Sphingosine-1-phosphate (S1P) signaling is a bona fide drug target as evidenced by the S1P receptor modulator drug class for treating multiple sclerosis and ulcerative colitis. Despite this clinical success, on-target adverse events have been reported. To circumvent these, we hypothesized that interdicting the S1P transport cascade will have the same physiological benefits without undesired events. This led to the first reported S1P release inhibitor (SRI), SLF1081851, that supported this concept. Subsequent structural improvements led to new generations of SRIs. Herein, we report a structure-activity relationship study of novel indazole compounds inspired by modifications on a previously reported scaffold. Studies revealed compounds 6c (IC₅₀ = 651 ± 3 nM) and 6k (IC₅₀ = 719 ± 50 nM) to be viable SRIs with molecular modeling suggesting that conformational effects are observed and potency can be influenced by these effects. Thus, both 6c and 6k provide valuable insight into the chemical space within Spns2 and help influence the next generation of SRIs.

Graphical abstract

Catabolism of complex sphingolipids leads to ceramide, which is hydrolyzed by acid ceramidase to yield a long chain fatty acid and sphingosine. Following transfer out of the lysosome, sphingosine is used to regenerate ceramide via ceramide synthases (CerS1-6) or phosphorylated via sphingosine kinases (SphK1,2)¹ to form S1P. In turn, S1P is dephosphorylated by lipid phosphatases, degraded by S1P lyase or in some cells, transported out via MFSD2B (red blood cells) or Spns2 (endothelial cells, others).^{2, 3}

Once S1P enters the extracellular space, it acts as the endogenous ligand to a series of G-protein coupled receptors (S1P1-5). These receptors are expressed in various cell types and on S1P binding, signal transduction pathways are activated, promoting angiogenesis, regulating cell-matrix adhesion and immune cell trafficking.^{4, 5} It is now known that these pathways contribute to various disease states. For example, lymphocyte trafficking is a key factor in autoimmune diseases such as multiple sclerosis (MS) and ulcerative colitis (UC).^{6, 7} Mechanistically, S1P1 in lymphocytes respond to the S1P gradient and agonist activity leads to endocytosis and either recruit G-protein signaling or initiates lymphocyte migration.⁸

Towards therapeutic development, a class of immunomodulators were developed to target the S1P receptors leading to a drug class known as the S1P receptor modulators (SRMs). Fingolimod is the first oral FDA-approved drug for the treatment of relapsing-remitting MS.⁹ Fingolimod becomes pharmacologically active after phosphorylation by SphK2, facilitating activation to S1P1,3-5.¹⁰ Once bound, phospho-fingolimod acts as a functional antagonist to cause endocytosis and degradation of the receptors rendering the immune cells insensitive to S1P, leading to lymphopenia, which is the primary mechanism of action against MS.¹⁰ Various other modulators have been developed and approved to treat MS and UC, including Siponimod, Ponesimod, Ozanimod, and Etrasimod (Figure 1).¹¹ Although effective, this drug class does have shortcomings with the prominent one being on-target cardiac side effects including bradycardia and arrhythmia.¹¹

Figure 1.

FDA-approved S1P receptor modulators.

A potential alternative target of regulating the S1P pathway is upstream of S1P receptors through blockade of S1P export via Spns2. Spns2 is a membrane-bound protein that acts similarly to a rocker-switch to transport S1P into extracellular space.^{12, 13} Additionally, Spns2 has been recently reported to take part in glucose uptake as an importer.¹⁴ Spns2 is also indicated to contribute in blood-brain barrier stability,¹⁵ vascular aging,¹⁶ and various cancers.^17–19 Due to this, Spns2 is being researched as a new drug target for different disease states.

Of particular advantage is the mitigation of cardiac effects that is an on-target liability of the S1P modulator drug class.²⁰ As Spns2 is commonly found in endothelial and perivascular cells, controlling the release of S1P into the extracellular space could manipulate local S1P signaling.²¹ Indeed, genetically modified Spns2 knockout mice have decreased immune cell trafficking, inducing lymphopenia.²¹ Further, Spns2-null mice show improved clinical scores when induced with experimental autoimmune encephalomyelitis (EAE), a disease model for the progression and mitigation of MS.^{22, 23} Moreover, Spns2-deficient mice also show improvements related to ischemia-reperfusion injury (IRI) with decreased collagen deposition and plasma creatinine levels compared to wild-type mice.²⁴

To investigate the therapeutic potential of Spns2, various S1P release inhibitors (SRIs) have been developed. The first generation SLF1081851 has a IC₅₀ of 1.93 ± 0.04 μM and reduces absolute lymphocyte counts (ALC) when administered to mice, mimicking a phenotype of Spns2 knockout mice (Figure 2).²⁵ SLF1081851 showed improvements in a mouse model of acute kidney injury (ischemia-reperfusion injury (IRI)) with decreased plasma creatinine levels, collagen deposition, and pro-inflammatory markers.²⁴ While useful as a tool compound, SLF1081851 is overtly toxic to mice when dosed at 30 mg/kg.²⁵ Therefore, a new generation of inhibitors were developed to improve potency and mitigate toxicity. Prior work from our laboratories indicated two successful structural changes; a ring fusion of the phenyl linked to the oxadiazole led to benzoxazole SLB1122168 (IC₅₀ = 94 ± 6 nM)²⁶ and a prior SAR study established the piperazine group as the optimal head group with a new urea-based linker led to phenylurea SLF80821178 (IC₅₀ = 51 ± 3 nM)²⁷ (Figure 2).²⁰ Both compounds improved potency compared to SLF1081851 and demonstrated in vivo target engagement using decreased lymphocyte count as a pharmacodynamic biomarker.^{26, 27}

Figure 2.

Structural improvements on Spns2 inhibitors.

Taking inspiration from these prior studies, along with imidazole based inhibitor SLF80721166,²⁸ we designed a hybrid ring fusion that includes an indazole core (Figure 3). This indazole core introduces both H-bond donor and acceptor moieties within the fused ring system, which is not observed in SLF80821178 or SLB1122168.^{26, 27} Of note, SLB1122168 suffered from low oral bioavailability, which could have been caused by a metabolic liability or poor aqueous solubility.²⁹ We hypothesize that these shortcoming could be addressed with the indazole core^{30, 31} Furthermore, this heterocycle samples a different region in space (carbonyl is placed on the 3-position) and can provide some insight into binding affinity within Spns2. In this study, we performed the structure-activity relationship study investigating the heterocyclic core, the position of substituents, as well as the optimal aliphatic tail length while maintaining the piperazine head group and amide linker from SLF80821178.

Figure 3.

Design rationale of indazole derivatives.

The synthetic route began with an amide coupling of indazoles 1a-b and N-Boc-piperazine using 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride (EDC) to generate amides 2a-b. Protection of indazole with 2-(trimethylsilyl)ethoxymethyl chloride (SEM) using sodium hydride led to 3a-b. A two-step, one-pot hydroboration of various alkenes followed by a Suzuki-Miyaura cross coupling furnished 4a-l. Deprotection of the SEM group using tetrabutylammonium fluoride (TBAF) provided 5a-l. Finally, hydrochloric acid was used to deprotect the Boc group to provide the products as hydrochloride salts 6a-l.

With the first set of indazole-based compounds synthesized, their inhibitory activity to block S1P transport was investigated using HeLa cells.³² In this assay, S1P lyase activity is inhibited by 4-deoxypyridoxine while lipid phosphate phosphatases 1-3 (LPP_1-3) were blocked by NaF and sodium orthovanadate. The blockade of S1P degradation maximizes intracellular S1P for transport. Thus, in presence of putative Spns2 inhibitors, S1P extruded in the extracellular media decrease, which can be measured by LC-MS.

The initial screening campaign utilized 1 μM putative inhibitors. SLF80821178 was used as the positive control with 79% inhibition when dosed at 0.3 μM. As shown in Figure 4, indazole derivative containing a short pentyl tail (6a) on the 5-position had no significant S1P transport inhibition. However, Spns2 inhibitory activity was observed beginning with a hexyl tail (6b) and maximized with a heptyl tail (6c). As the chain length increased to C10, a concomitant decrease in activity was observed. We next investigated placing the alkyl tail on the 6-position (Figure 5). Similarly, as the alkyl chain length increased from five to 10 carbons, improvement in activity was observed and peaked at a nonyl chain (51% inhibition). However, the extent of S1P export inhibition was less pronounced at this 1 μM concentration. Taken together, our studies suggest that alkyl chain positioning on the indazole core plays a more nuanced role for potency than initially believed.

Figure 4. Tail homologation on 5-indazoles.

Investigation of compounds 6a-f for 5-position on indazole. All compounds were tested in duplicate at a concentration of 1.0 μM in HeLa cells. Spns2 inhibition is reported as a percentage of extracellular S1P levels compared to S1P level in control experiment with no inhibitor present. S1P levels are quantified via LC/MS.

Figure 5. Tail homologation on 6-indazoles.

Tail homologation investigation of compounds 6g-l for 6-position on indazole. All compounds were tested in duplicate at a concentration of 1.0 μM in HeLa cells. Spns2 inhibition is reported as a percentage of extracellular S1P levels compared to S1P level in control experiment with no inhibitor present. S1P levels are quantified via LC/MS.

To explore this nuance, we performed a dose-response investigation of the top compound for each series (Figure 6). Thus, HeLa cells were treated with increasing concentration (1 nM → 10 μM) of 6c and 6k and quantified exported S1P levels via LC-MS.²⁰ Our results indicate that 6c had a IC₅₀ = 651 ± 3 nM while 6k had IC₅₀ = 719 ± 50 nM, suggesting a minor potency difference based on tail position. While 6c (SLH1102405) is about 10-fold less potent than SLB1122168 (94 ± 6 nM) and SLF80821178 (51 ± 3 nM), 6c significantly improved the tool compound SLF1081851 (1.93 ± 0.04 μM).

Figure 6. Dose-response of compounds 6c and 6k.

Dose-response investigation for 6c (IC₅₀ = 651 ± 3 nM) and 6k (IC₅₀ = 719 ± 50 nM) against Spns2 in HeLa cells. Assay was completed in duplicate.

To understand this difference in potency, we performed a computational docking of compounds 6c and 6k with Spns2, hypothesizing the difference is due to overall intrinsic affinity of each compound. Spns2 is a dynamic major facilitator superfamily (MFS) transporter that is theorized to cycle between inward‑ and outward-facing states during transport, allowing inhibitors to bind multiple conformations.³³ To explore the aforementioned hypothesis, we docked 6c and 6k into (i) the molecular dynamics (MD)-derived inward‑facing occluded model²⁷ and (ii) the outward-facing cryo-EM structure (PDB ID: 8EX8).¹² When docked to the inward-facing state, both ligands occupy binding site 1 (BS1) and retain the crucial Arg119 salt bridge (Figure 7). MM/GBSA scores are favorable for both (6k = −76 kcal mol⁻¹, 6c = −56 kcal mol⁻¹). The longer alkyl chain of 6k buries more hydrophobic surface, while 6c adopts a more compact pose that may reduce entropic cost and keep the indazole/piperazine correctly oriented for inhibition as compared to previous compounds.²⁷ Interestingly, when docked to the outward-facing state, the ligands shift toward the extracellular leaflet, revealing a distinct pocket (BS2) (Figure 7). 6c makes additional contacts (H-bond to Tyr246, Van der Waals to Gln144) that 6k lacks, giving 6c a modest energetic advantage (6c = −42 kcal mol⁻¹ vs 6k = −34 kcal mol⁻¹). The 5-alkyl chain of 6c likely balances hydrophobic engagement with optimal polar interactions in both pockets, whereas the 6-substituted chain of 6k may perturb this geometry. Both compounds show improved predicted free energy of interaction values comparable to previously in vitro and in vivo validated Spns2 inhibitors. The slight experimental advantage of 6c is therefore likely due to subtle geometric and conformational effects rather than a large difference in intrinsic affinity.

Figure 7. Molecular docking of compounds 6c and 6k to Spns2.

Aromatic, hydrogen-bonding, and charged or Van der Waals (GLN144) interactions are shown in blue, purple, and red, respectively; gray denotes residues interacting with only one ligand. Compound 6c is shown in red and compound 6k is shown in purple. (Right) Top docked poses in the molecular dynamics (MD) derived Spns2 model with key interaction distances (Å). (Left) Top docked poses in the outward-facing partially occluded structure (PDB ID: 8EX8) with key interaction distances (Å). BS1: binding site 1, BS2: binding site 2.

In conclusion, we conducted a tail homologation study to explore the indazole core. Our studies suggest that tail position can affect potency, albeit compounds 6c and 6k show comparable inhibitory activity. For the 5-position, a heptyl chain is preferred whereas a nonyl chain is optimal for the 6-position. Molecular docking experiments suggest key interaction with Arg119 and piperazine nitrogen and a subtle one in the inward-facing MD-based structure that leads to movement towards the outer leaflet in the outward-facing structure. Nonetheless, the observed conformational differences in correlation to potency provide insight into compound binding within Spns2 and can be further probed in previous and forthcoming scaffolds. Future studies regarding the indazole scaffold will include defining the optimal head group and potential efficacy studies in mouse model of multiple sclerosis.

Scheme 1.

(a) N-Boc-piperazine (1.5 equiv), EDC·HCl (1.5 equiv), DMF, rt, 16 h, 72-85%; (b) 60% NaH (1.2 equiv), SEM-Cl (1.1 equiv), DMF, rt, 16 h, 88-90%; (c) (i) alkene (2.0 equiv), 9-BBN (2.2 equiv), THF, 70 °C, 1 h; (ii) 3a-b (1.0 equiv), Pd(dppf)Cl₂·CH₂Cl₂ (0.15 equiv), Na₂CO₃ (3.0 equiv), 5:1 THF:H₂O, 80 °C, 16 h; (d) 1.0 M TBAF/THF (5.0 equiv), 70 °C, 16 h, 17-63%; (e) 4.0 M HCl/Dioxane, DCM, rt, 16 h, 7-79%.

Highlights.

• Twelve indazole derivatives were synthesized for S1P release inhibition.

• Alkyl chain length trends change based on tail position.

• Compounds 6c and 6k are more potent than SLF1081851.

• Docking study indicated conformational differences between compounds 6c and 6k.

Appendix A: Supplementary Data

Supplementary data can be found online via the following link.

Data Availability

Data will be made available upon request.