Identification of highly selective SIK1/2 inhibitors that modulate innate immune activation and suppress intestinal inflammation

Abstract

The salt-inducible kinases (SIK) 1-3 are key regulators of pro- versus anti-inflammatory cytokine responses during innate immune activation. The lack of highly SIK-family or SIK isoform-selective inhibitors suitable for repeat, oral dosing has limited the study of the optimal SIK isoform selectivity profile for suppressing inflammation in vivo. To overcome this challenge, we devised a structure-based design strategy for developing potent SIK inhibitors that are highly selective against other kinases by engaging two differentiating features of the SIK catalytic site. This effort resulted in SIK1/2-selective probes that inhibit key intracellular proximal signaling events including reducing phosphorylation of the SIK substrate cAMP response element binding protein (CREB) regulated transcription coactivator 3 (CRTC3) as detected with an internally generated phospho-Ser329-CRTC3-specific antibody. These inhibitors also suppress production of pro-inflammatory cytokines while inducing anti-inflammatory interleukin-10 in activated human and murine myeloid cells and in mice following a lipopolysaccharide challenge. Oral dosing of these compounds ameliorates disease in a murine colitis model. These findings define an approach to generate highly selective SIK1/2 inhibitors and establish that targeting these isoforms may be a useful strategy to suppress pathological inflammation.

Keywords: immunological disorders; inflammatory bowel disease; kinase inhibitors; medicinal chemistry; structure-based drug design.

Fig. 1.

Optimization of SIK-selective inhibitors using structure-based design. (A) Structure of the screening hit (compound 1), SIK2 K_i (Top), and docking pose of 1 bound to SIK2 (Bottom) with a modeled two-point hinge binding interaction shown as hashed lines. (B) Structure of compound 2, SIK2 and LCK K_i (Top), and model of 2 bound to SIK2 (Bottom). The nitrile substituent of 2 occupies the space adjacent to Thr96 and in addition to the two-point hinge binding interaction shown in A, a modeled pyridine nitrogen hydrogen bonding interaction with Lys49 is shown as a hashed line. (C) Structure of compounds 3 and 4 having basic groups capable of interacting with the unique SIK Glu103 residue and SIK2 and LCK K_i (Top). X-ray crystal structure of compound 3 bound to a MARK2/SIK2 chimera (Bottom), with hydrogen bonds shown as hashed lines and a salt bridge interaction between the pyrrolidine substituent and the SIK2 Glu103 residue shown as a dotted line. Calculated SIK2 K_i values are based on a SIK2 ADP-Glo^TM assay at K_m ATP and reported as mean ± SD with n ≥ 3 independent experiments. Calculated LCK K_i values are based on a Eurofins Cerep KinaseProfiler^TM assay at K_m ATP with n = 1.

Fig. 2.

SIK1/2-selective inhibitors modulate cytokine responses and upregulate CREB target genes during innate immune activation in vitro. (A) SIK1/SIK2-selective inhibitor JRD-SIK1/2i-3 modulates cytokine production in primary human macrophages derived from healthy donors. Macrophages were differentiated from PBMC monocytes for 7d with M-CSF, exposed to SIK inhibitors for 2 h, and stimulated with LPS for another 3 h before determination of IL-10 (Left) and TNF concentrations (Right) in the cell culture media by AlphaLISA® proximity assays. (B) JRD-SIK1/2i-3 suppresses LPS-induced TNF production in BMDCs from Il10 KO mice. BMDCs derived from mice with the indicated genotypes were treated with the indicated concentrations of JRD-SIK1/2i-3 for 2 h followed by stimulation with LPS for 4 h and quantification of IL-10 (Left) and TNF (Right) by cytokine bead array or AlphaLISA® proximity assays, respectively. Each point represents the mean ± SD (n = 3 biological replicates per genotype) at a given compound concentration. Half-maximal effective concentrations were calculated using non-linear regression analysis. Results from one of two experiments are shown (see SI Appendix, Fig. S3F for the second experiment). (C) sc-RNA-seq analysis was performed in PBMCs treated with JRD-SIK1/2i-3 for 1 h followed by stimulation with LPS for an additional 1 h. The cells were visualized using UMAP plots with the Upper panel showing all cells for each condition (unstimulated, 5 ng/mL LPS, 5 ng/mL LPS + 1 µM JRD-SIK1/2i-3, 5 ng/mL LPS + 10 µM JRD-SIK1/2i-3; n = 2 donors per condition) and the Lower panel depicting only cells for the unstimulated and the 5 ng/mL LPS conditions, as indicated (no compound). The cells largely separated by cell type, but the myeloid compartment included additional distinct clusters upon LPS stimulation. (D) Numbers of significant (absolute log2FC > 0.5 and FDR < 0.05) differentially expressed genes detected in each cell type and for the given comparison from the experiment described in (C). Numbers of genes that were upregulated by LPS or LPS plus 1 µM or 10 µM JRD-SIK1/2i-3 are shown in cyan, while the number of genes downregulated are shown in pink. Fractions of genes that are also putative CREB targets are depicted in a darker shade, with the numerical fractions indicating the number of putative CREB genes over the total number of differentially expressed genes in the associated direction.

Fig. 3.

JRD-SIK1/2i-4 exhibits TNF suppression, IL-10 induction, and proximal target engagement in vivo. (A) Schematic of the acute LPS challenge model used to determine PK/PD relationships for SIK1/SIK2-inhibitors. C57BL/6 mice were orally dosed JRD-SIK1/2i-4 at 1, 5, 20, or 80 mg/kg or YKL-05-099 at 80 mg/kg, or vehicle 1 h before administering a systemic LPS challenge (0.2 mg/kg, IP) and 2 h before harvesting blood and colon tissue for exposure and cytokine analyses. (B) Plasma exposure/response relationships of TNF (Top) and IL-10 (Bottom). Compound plasma concentrations were determined by LC–MS/MS and plasma cytokine concentrations by MSD^® sandwich immunoassays. For JRD-SIK1/2i-4, curves were fitted using four-parameter logistic regression (1/y2 weighting). The top of the TNF and bottom of the IL-10 curves were constrained to the mean TNF and IL-10 concentrations of the vehicle group, respectively (dotted lines). (C) JRD-SIK1/2i-4 dose-dependently inhibits Tnf (Left) and elevates Il10 gene expression (Right) in colon tissue. The PD response to YKL-05-099 is shown for comparison. Gene expression is expressed as fold change over naive using B2m as housekeeping gene. Statistically significant changes for individual cohorts are shown if P < 0.05 (*P < 0.05, ***P < 0.001, ****P < 0.0001 versus vehicle; JRD-SIK1/2i-4 groups versus vehicle: one-way ANOVA followed by Dunnett’s multiple comparisons test, YKL-05-099 versus vehicle: unpaired, two-tailed Student’s t-test). One mouse was removed from the vehicle group because of an issue with the LPS injection. (D) Proximal target engagement in colon extracts as determined by loss of pSer329-CRTC3-phosphorylation. Approximately 5-mm-long tissue samples of flushed colons were lysed in the presence of proteinase- and phosphatase-inhibitors and 30 µg of lysate per sample were analyzed for the presence of either pSer329-CRTC3 or total CRTC3 by MSD®-ECL. Luminescence signals from the pSer329-CRTC3 assay were divided by the luminescence signals from the total CRTC3 assay to normalize to total CRTC3 present in each sample and mean luminescence ratios are shown for each dose group (±SEM; n = 6 mice per group). Significance of differences of compound-treated groups from the vehicle-treated group was determined by one-way ANOVA comparing group means followed by Dunnett’s multiple comparisons test (**P < 0.01, ***P < 0.001, ns = not significant).

Fig. 4.

JRD-SIK1/2i-4 suppresses disease in the anti-CD40-induced colitis model. (A) Schematic of the anti-CD40 colitis model used to determine the ability of SIK1/2-inhibitors to ameliorate pathology in a myeloid-driven inflammatory model. Rag2^−/− mice were injected with 200 µg anti-CD40 antibody per mouse IP (day 0) and orally administered 5, 20, or 80 mg/kg b.i.d. or 40 or 80 mg/kg q.d. JRD-SIK1/2i-4 on days −1 through 6. A cohort dosed once on day −1 with an anti-IL-23p19 antibody (CNTO 3723) served as a positive control. (B) Body weight expressed as percent of baseline weight (day 0). JRD-SIK1/2i-4 dosed at 20 mg/kg (◆) or 80 mg/kg (▲) b.i.d. and 40 mg/kg (◇) or 80 mg/kg (△) q.d. showed statistically significant suppression of body weight loss compared to vehicle b.i.d. (⚫) and vehicle q.d. (○), respectively (††††P < 0.0001 versus vehicle b.i.d., ####P < 0.0001 versus vehicle q.d.; two-way ANOVA with repeated measures, followed by Dunnett’s multiple comparisons test). The error bars denote SEM. (C) Liver histopathology scores (sum of subacute inflammation, acute coagulation necrosis of hepatocytes, portal vein thrombosis, increased hepatocyte mitotic figures, lipid vacuolization of hepatocytes, and extramedullary hematopoiesis, each on a 0–5 scale) on day 7. JRD-SIK1/2i-4 dosed at 80 mg/kg (▲) b.i.d. and 40 mg/kg (◇) or 80 mg/kg (△) q.d. showed statistically significant prevention of liver inflammation compared to vehicle b.i.d. (⚫) and vehicle q.d. (○), respectively (††††P < 0.0001 versus vehicle b.i.d., #P < 0.05 versus vehicle q.d.; Kruskal–Wallis test). (D) Representative photomicrographs of proximal colon at 70× magnification of naive, vehicle b.i.d., JRD-SIK1/2i-4 80 mg/kg b.i.d., and anti-IL-23p19 study groups. Infiltration of neutrophils, lymphocytes, and macrophages (*), as well as gland loss (**) are indicated. Mucosa (M), submucosa (SM), lymphoid aggregate (LA), and tunica muscularis externa (TME) are denoted. The scale bars correspond to 300 µm. (E) Colon histopathology scores (sum of edema, mucosal thickness, inflammation, gland loss, and erosion, each on a 0–5 scale) on day 7. JRD-SIK1/2i-4 dosed at 20 mg/kg (◆) or 80 mg/kg (▲) b.i.d. and 40 mg/kg (◇) or 80 mg/kg (△) q.d. showed statistically significant prevention of colitis compared to vehicle b.i.d. (⚫) and vehicle q.d. (○), respectively. Statistically significant suppression of colitis was also observed in the anti-IL-23p19 cohort against either vehicle group (†P < 0.05 and ††††P < 0.0001 versus vehicle b.i.d.; #P < 0.05, ###P < 0.001, and ####P < 0.0001 versus vehicle q.d.; Kruskal–Wallis test).