Discovery of Lanraplenib (GS-9876): A Once-Daily Spleen Tyrosine Kinase Inhibitor for Autoimmune Diseases

Abstract

Spleen tyrosine kinase (SYK) is a critical regulator of signaling in a variety of immune cell types such as B-cells, monocytes, and macrophages. Accordingly, there have been numerous efforts to identify compounds that selectively inhibit SYK as a means to treat autoimmune and inflammatory diseases. We previously disclosed GS-9973 (entospletinib) as a selective SYK inhibitor that is under clinical evaluation in hematological malignancies. However, a BID dosing regimen and drug interaction with proton pump inhibitors (PPI) prevented development of entospletinib in inflammatory diseases. Herein, we report the discovery of a second-generation SYK inhibitor, GS-9876 (lanraplenib), which has human pharmacokinetic properties suitable for once-daily administration and is devoid of any interactions with PPI. Lanraplenib is currently under clinical evaluation in multiple autoimmune indications.

Spleen tyrosine kinase (SYK) is a nonreceptor cytoplasmic tyrosine kinase that mediates immunoreceptor signaling in multiple cell types involved in inflammatory disease. SYK plays an essential role in coupling the B cell receptor (BCR) to B cell survival, migration, and activation.^1,2 SYK also plays a critical role in signaling initiated from Fc-epsilon receptors (FcεR) on mast cells and basophils and Fc-gamma receptors (FcγR) on monocytes, macrophages, and neutrophils.³⁻⁵ Importantly, SYK is only found at very low levels in mature T cells, and this is hypothesized to limit the potential for excessive immunosuppression with selective SYK inhibition.^6,7 Upon immunoreceptor ligation, SRC family kinases phosphorylate immunoreceptor tyrosine-based activation motifs (ITAMs) in the cytoplasmic domains, leading to recruitment of SYK via the tandem SH2 domains. Subsequent activation of SYK leads to autophosphorylation and phosphorylation of downstream targets (e.g., B-cell linker [BLNK] protein and phospholipase-C gamma 2 [PLCγ2]) and activation of multiple signaling pathways, including phosphatidylinositol 3-kinase (PI3K), Bruton’s tyrosine kinase (BTK), and mitogen-activated protein kinase (MAPK) pathways.

Systemic lupus erythematosus (SLE) is a chronic heterogeneous autoimmune disease characterized by the production of autoantibodies and autoimmune attack on multiple organs, including the skin, heart, lungs, and kidneys. SLE accounts for approximately 70% of all cases of lupus.⁸ Lupus nephritis (LN) is a major clinical complication of SLE, characterized by inflammation and deposition of immune complexes in the kidneys that induce tissue damage and lead to kidney failure.⁹ Current treatment options for LN are very limited. In vitro and in vivo data suggest that SYK inhibition may decrease several pathologically active mechanisms implicated in SLE, including B-cell activation, T-cell costimulation, cytokine release, and FcγR-mediated myeloid cell activation. Phosphorylated SYK has been observed in tissues from SLE patients,⁹ and SYK inhibition by fostamatinib prevented proteinuria and improved glomerular pathology in the NZBxW and MRL-lpr models of lupus.¹⁰ More recently, we have shown that treatment of NZB/W F1 mice with a selective SYK inhibitor prevented progression of proteinuria and reduced renal inflammation.¹¹ The study also found that the B- and T-cell subsets identified in peripheral blood mononuclear cells from SLE human subjects have a similar phenotype to NZB/W F1 diseased mice.¹²

In recent years, several SYK inhibitor chemotypes have been reported,¹³ including heteroaryl carboxamides,^14,15 pyrazolopyrazines,¹⁶ and naphthyridines¹⁷ (Figure 1). From these efforts, compounds 1,¹⁸2,¹⁹ and entospletinib 3(20) have progressed into human clinical trials. Fostamatinib 4 was approved by the FDA in 2018 for the treatment of immune thrombocytopenia and is the only SYK inhibitor approved for any indication.²¹ We have previously disclosed 3 as a selective SYK inhibitor with clinical efficacy in chronic lymphocytic leukemia (CLL) and acute myeloid leukemia (AML).^22,23 Development of this compound in inflammatory conditions was precluded by pH-dependent solubility that limits absorption and leads to a drug–drug interaction with proton pump inhibitors as well as its short plasma half-life, resulting in a BID dosing regimen.^24,25 Herein, we describe the discovery of lanraplenib (GS-9876, 39), a selective SYK inhibitor suitable for once-daily oral dosing and devoid of any drug–drug interaction with proton pump inhibitors.

Figure 1.

Clinical stage SYK inhibitors.

Our initial objective was to improve metabolic stability and pH-dependent aqueous solubility while maintaining potency and selectivity. Potency was assessed in three assays: a SYK biochemical assay, inhibition of the αIgM-induced phosphorylation of BLNK (a direct substrate of SYK in B cells) in a Ramos B cell line, and inhibition of αFcεR1-induced CD63 expression on basophils in human whole blood. Metabolite ID studies with 3 showed extensive oxidative metabolism of the morpholine ring, and our initial work sought to functionalize or replace the morpholine ring as a means to disrupt this metabolism (Table 1). An initial series of bridged morpholine compounds 5–8 with various substitutions and ring sizes did not improve metabolic stability, consistent with observations that have since been reported elsewhere.²⁶ Only compound 8, with a one-carbon bridge across the 3- and 5-morpholine positions, maintained stability in human liver microsomes but suffered a 3–4 fold loss in all potency assays relative to 3. The spirocylic, fused, and methylated morpholine derivatives 9, 11, and 12 either improved or maintained microsomal stability relative to 3; however, these compounds lost potency in the cellular and whole blood assays and also failed to improve solubility at physiological pH. While the metabolite ID of 3 could not identify specific sites of metabolism, the improved stability of 12 implicated the C–H bonds adjacent to the morpholine oxygen. However, removal of the ring oxygen atom, as in compounds 13–15, failed to improve metabolic stability. The hydroxymethyl analogue 16 maintained microsomal stability and good potency, although solubility at pH 7.4 remained low. Exchanging the hydroxyl group of 16 with a dimethylamine motif of 17 greatly improved solubility and eliminated the pH-dependence while retaining good whole blood potency. However, this compound had low forward permeability in Caco-2 cells (Papp = 0.5 × 10^–6 cm/s) with a high efflux ratio (ER, 29×), corresponding to poor oral bioavailability in rats (F = 2%). Furthermore, in a functional selectivity assay measuring the proliferation of primary human B or T-cells, 17 showed a reduced selectivity window relative to 3 (T/B EC₅₀ ratio = 5 vs 17 respectively).

Table 1. Modifications of the RHS of Compound 3.

Encouraged by the high solubility and slight improvement in metabolic stability with 17, we pursued a series of piperazine analogues 18–20. The ethyl-piperazine 18 improved microsomal stability as well as solubility at low pH and had 48% oral bioavailability in rats, however the functional selectivity was unacceptably low (T/B ratio = 5). A loss in functional selectivity was a recurring observation with highly basic compounds, so we sought to modulate the pK_a of the piperazine nitrogen through substitution. The N-oxetane derivative 20 improved microsomal stability and solubility at pH 2 while maintaining potency within 2-fold of the morpholine analogue 3. Consistent with the reduced pK_a compared to 18 (calcd pK_a = 8.0), compound 20 (calcd pK_a = 6.4) showed improved functional selectivity with a T/B cell proliferation ratio of 10. In addition, 20 showed favorable Caco-2 permeability (Papp = 5.5 × 10^–6 cm/s, ER = 2.9), improved solubility at pH 2, and good oral bioavailability in rats (42%). Consequently, the N-oxetane piperazine motif was chosen as a morpholine replacement and used in further exploration.

The compounds in Table 1 provided relatively small changes in logD (range 1.9 to 2.7) compared to the parent compound 3 (logD = 2.5) (Figure S1), which we surmised was related to the minimal stability improvements observed. We next turned our attention to the indazole ring system which, although not identified as a site of metabolism, we felt would allow for favorable modulation of molecular properties. The crystal structure of 3 bound to SYK revealed a hydrogen bond between the indazole NH and D512 of the hinge and a van der Waals contact between the bicyclic system and V385.²⁰ The calculated electrostatic potentials due to SYK protein residues at different ligand atom locations were generally large, positive values (Figure 2). Judicious placement of heteroatoms was expected to enhance the electrostatic complementarity between the ligand and the protein, leading to improved binding affinity. Additionally, the hydrogen bond donor strength was also predicted to increase by incorporation of nitrogen atoms into the indazole²⁷ (Figure 2). Because the fundamental nature of the binding mode is not altered by these changes, we speculated that variation of heteroatom placement around the ring offered an avenue to modulate the physicochemical properties of the molecule without adversely impacting potency or selectivity. The results of these investigations are summarized in Table 2.

Figure 2.

Calculated electrostatic potentials (atomic unit of charge/Å) at C-ring sites due to SYK protein residues (A). Hydrogen bond donor moments derived from density functional calculations (B). Stronger donors tend to have more negative values.²⁷

Table 2. C-Ring Modifications of Compound 20.

The 7-azaindazole 21 provided a 2-fold improvement in cellular potency that translated to a significant improvement in whole blood potency compared to the parent indazole 20. Unfortunately, microsomal stability and solubility remained unchanged, deficiencies that were not addressed by the 4-aza or 4,7-diaza analogues 22 and 23. The benzimidazole 24 offered an encouraging combination of pH 7.4 solubility, metabolic stability, and whole blood potency but suffered from poor Caco-2 permeability (Papp = 0.4 × 10^–6 cm/s, ER = 38). Compound 26 had the best permeability profile of the benzimidazole series (Papp = 2.7 × 10^–6 cm/s, ER = 5.8) and also improved microsomal stability but suffered from reduced potency in the whole blood assay. Although the diazaindole 31 provided excellent cellular and whole blood potency, no member of the azaindole series sufficiently improved microsomal stability or solubility at physiological pH.

We next hypothesized that reducing the number of aryl rings or decreasing planarity might be a more fruitful path to improved solubility.^28,29 The benzomorpholine system in 32 introduces 3-dimensionality while retaining the hydrogen bond donor to satisfy the D512 interaction. We were encouraged by the improvement in pH7.4 solubility and believed the loss in potency and stability could be recovered through application of our learnings from the azaindazole series. We found the pH7.4 solubility to be highly dependent on the position of the added nitrogen, with the 8-aza analogue 34 uniquely demonstrating high and pH-independent solubility. Additionally, 34 demonstrated good oral exposure (F = 38%) in rats. The corresponding diazabenzomorpholine 35 provided the best combination of whole blood potency and microsomal stability in this series. This compound maintained good permeability (Papp = 11 × 10^–6 cm/s, ER = 3) and solubility, although oral bioavailability was lower than desired (F = 21% in rats).

Continuing the strategy of minimizing aromatic ring count, we removed the 5-membered ring of indazole 20 altogether to give aniline 36. This significantly improved solubility but at an unacceptable cost to potency in all assays. Encouraged by observations in the bicyclic series, we again introduced nitrogen atoms into the aryl ring and found that all compounds in this monocyclic series exhibited high solubility with reduced pH-dependence, while at the same time providing whole blood potency benefits similar to those seen in the bicyclic series. From this effort, the aminopyrazine 39 provided the best whole blood potency (2-fold improved relative to 3) and had good functional selectivity (T/B ratio = 13). Importantly, this compound was completely stable in human microsomes and hepatocytes and exhibited high solubility with no pH-dependence, thus satisfying the optimization goals of the project. Throughout the optimization process, a general trend for improved metabolic stability with lower logD was observed both in human microsomes and hepatocytes (Figures S2 and S3) with logD lower than 1.8 proving optimal. The N-oxetane piperazine and the aminopyrazine combined to lower the logD from 2.5 in compound 3 to 1.3 in compound 39. Further lowering of the logD beyond this point was observed to be deleterious to permeability and oral bioavailability.

Compound 39 showed moderate clearance in preclinical species that was generally well predicted from in vitro hepatocyte stability assays (Table 3). Together with excellent stability in human hepatocytes, this supported a low predicted in vivo clearance in humans. Consistent with the observed solubility and permeability, 39 was well absorbed with oral bioavailability of 60–100% across species. A key objective of the program was to mitigate the pH-dependent absorption observed with 3. To validate the impact of the improved solubility profile on absorption 39 was dosed as a powder-in-capsule to dogs pretreated with either pentagastrin or famotidine, conditions that maintain gastric pH at ∼2 and ∼5–7, respectively.³⁰ The AUC of 39 was not significantly different with famotidine versus pentagastrin pretreatment (AUC = 10.2 vs 8.8 μM·hr, respectively), in contrast to compound 3, which experienced >90% reduction in AUC when codosed with famotidine.

Table 3. DMPK Profile of Compound 39.

	in vitro			in vivoa
	% free plasma	BPR	Hep Clp (L/h/kg)	Cl (L/h/kg)	Vss (L/kg)	T1/2 (hr)	F (%)b
rat	11.8	0.83	1.4	1.77	2.5	3.7	60
dog	38.5	2.19	0.28	0.33	1.6	5.3	100
monkey	32.2	2.38	0.28	0.58	1.3	2.3	100
human	20.4	1.84	0.05

^aIntravenous administration at 1.0 mg/kg. Oral administration at 5.0 mg/kg.

^bF = oral bioavailability

The crystal structure of the SYK kinase domain in complex with 39 was determined to 1.95 Å resolution. The ligand binds in the active site pocket normally occupied by ATP (Figure 3), forming two hydrogen bonds to the backbone nitrogen and carbonyl of A451 of the hinge. The pyrazine ring sits below the P-loop near residues 377–379, and the primary amine interacts with the side chain of D512 via a hydrogen bond. There are also van der Waals contacts between the pyrazine and V385. The phenyl ring is sandwiched between residues L377 and G454, while the extended N-oxetanepiperazine extends beyond the protein and into a solvent accessible area outside of the pocket.

Figure 3.

Co-crystal structure (1.95 Å) of SYK kinase domain with 39 (PDB code 6VOV). Dashed lines show hydrogen bond contacts in the pocket between compound and protein. Part of the protein is removed to make viewing the compound binding mode easier.

Compound 39 demonstrated high selectivity in a competitive binding assay, showing less than 10% control compound binding for 12 out of 395 nonmutant kinases, including SYK (Table S2). K_d determination showed that 39 inhibited 8 of these 11 off-target kinases with a K_d value within 10-fold of SYK. Follow up in biochemical assays showed JAK2 to be the most potently inhibited off-target kinase with an IC₅₀ of 120 nM, 9-fold higher than the IC₅₀ for SYK. In cellular assays measuring inhibition of α-IgM stimulated BLNK phosphorylation (SYK mediated) and EPO-stimulated pSTAT5 phosphorylation (JAK2 mediated), the selectivity for SYK over JAK2 increased to 48-fold.

Compound 39 is a potent inhibitor (EC₅₀ 24–51 nM) of signaling downstream of the B-cell receptor and completely abrogates the expression of the cell-surface activation markers CD86 and CD69 with EC₅₀ values of 112–164 nM (Table 4). Additionally, proliferation of B-cells following BCR stimulation was completely inhibited by 39 with an EC₅₀ of 108 nM. 39 also reduced immune-complex (IC) stimulated release of proinflammatory cytokines from human macrophages with TNFα and IL-1β inhibited more potently than IL-6.

Table 4. Biological Profile of Compound 39.

cell type	assay	stim	readout	EC50 ± SD (nM)
B-cells	signaling	αIgM	pBLNK-Y96	24
			p-Btk-Y223	26
			p-Akt-S473	34
			pERK-T202/Y204	37
			pMEK-s217/s221	51
	activation		CD69	112 ± 10
			CD86	164 ± 16
	proliferation	αIgM/αCD40	EdU incorporation	108 ± 55
T-cells	proliferation	αCD3/αCD28	EdU incorporation	1305 ± 425
macrophages	cytokine release	immune complex	TNFα	180
			IL-1β	90
			IL-6	700
basophilsa	activation	αIgE	CD63	73 ± 20

^aHuman whole blood.

39 was evaluated in a spontaneous lupus efficacy model in which mice develop characteristic disease pathologies, including body weight loss, proteinuria, glomerulonephritis, and premature death.¹¹ NZB/W F1 mice were treated with 39 for 12 weeks at doses expected to mimic the human clinical target coverage range. Treatment with 39 prevented the increase in proteinuria associated with disease progression and improved survival of the animals. The efficacy of 39 was equivalent to that achieved with cyclophosphamide (Figure 4).

Figure 4.

Efficacy of 39 in the NZB/W F1 murine lupus model. 39 reduced the proteinuria score from weeks 28–40 (A) and increased overall survival at week 40 (B).

The safety and pharmacokinetics of 39 in healthy human volunteers were evaluated at single doses of 2–50 mg and multiple doses of 15–50 mg (manuscript in preparation). 39 was well tolerated with no serious adverse events or clinically significant laboratory abnormalities reported. The median steady-state half-life of 39 was 21.3–24.6 h following a single dose of 2–50 mg, which supports a once-daily dosing regimen. There was no reduction in oral exposure following administration of a single dose of 45 mg 39 to subjects pretreated with 20 mg of omeprazole relative to subjects with no pretreatment (AUC = 6.46 vs 6.19 μM·hr, respectively), demonstrating the absence of any pH influence on absorption. 39 is currently undergoing clinical development in multiple autoimmune disease indications.

Glossary

Abbreviations

BLNK

B-cell linker protein

SLE

systemic lupus erythematosus

CLL

chronic lymphocytic leukemia

AML

acute myeloid leukemia

ER

efflux ratio

EPO

erythropoietin

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.9b00621.

• Experimental details, compound characterization data, kinase selectivity data, and supplementary tables and figures (PDF)

Accession Codes

The coordinates of the crystal structure of compound 39 bound to SYK have been deposited to the RCSB Protein Data Bank under the accession code 6VOV.

These studies were funded by Gilead Sciences, Inc.

The authors declare no competing financial interest.