Discovery of VU6015929: A Selective Discoidin Domain Receptor 1/2 (DDR1/2) Inhibitor to Explore the Role of DDR1 in Antifibrotic Therapy

Abstract

Herein, we report the discovery of a potent and selective dual DDR1/2 inhibitor, 7e (VU6015929), displaying low cytotoxicity, good kinome selectivity, and possessing an acceptable in vitro DMPK profile with good rodent in vivo pharmacokinetics. VU6015929 potently blocks collagen-induced DDR1 activation and collagen-IV production, suggesting DDR1 inhibition as an exciting target for antifibrotic therapy.

Receptor tyrosine kinases (RTKs) are a large family of cell-surface receptors responsible for regulating cellular proliferation, differentiation, survival, metabolism, and life cycle.¹ RTKs are divided into 20 subfamilies sorted by structural similarity and associated endogenous ligand. Discoidin domain receptors 1 and 2 (DDR1 and DDR2) are RTKs, which, unlike other members of the receptor family, are activated by collagens as opposed to peptide-like growth factors.² DDRs are widely expressed in developing organs, while their expression in adult organs is low or undetectable. DDR1 expression is predominant in epithelial cells, while DDR2 is most commonly found in cells of mesenchymal origin.³⁻⁶ Upon ligand binding, these receptors can control various cell functions including cell adhesion, migration, survival, and collagen production. Both receptors are well documented to play critical roles during embryonic development of mice and humans, however, their role during adulthood is limited.³ Expression of DDR1 is low in regions of the adult kidney such as the glomerulus; however, its expression can increase following glomerular injury.⁴ Upregulation of DDR1 has been observed in patients with common kidney diseases such as lupus nephritis and Goodpasture’s syndrome. Genetic mouse models of chronic kidney disease show that deletion of DDR1 improves survival and reduces fibrosis and inflammation.⁵ In addition to DDR1’s pro-inflammatory effect, this receptor has also been implicated as having a deleterious role in kidney injury via regulation of extracellular matrix production by resident cells.⁶

Blocking DDR1 activity with a small molecule inhibitor has emerged as an attractive option to treat DDR1 related progressions of fibrotic disease.^2,6 Several small molecule DDR1 antagonists have been reported to date (Figure 1). Inhibitor 1 was originally reported as a selective DDR2 antagonist (IC₅₀ = 18.6 nM), although it displayed similar inhibition of DDR1 (IC₅₀ = 12.4 nM).⁷ Efforts to identify DDR1 selective antagonists have yielded compounds such as 2 (DDR1-IN-1),⁸ with approximately 4-fold selectivity for DDR1 versus DDR2 and 3 with approximately 20-fold selectivity.⁹ Allosteric (type II/III) inhibitors, i.e., 4, of DDR1/DDR2 have also been reported,¹⁰ but the recent Roche compound 5 is the most selective to date for DDR1 (∼63-fold selective, but with a MW of 565).^11,12 While these ligands have garnered interest in DDR1 as an antifibrotic therapeutic agents, either cytotoxicity, kinome selectivity, broader selectivity, or poor DMPK profiles (or a combination of all four) have precluded more definitive proof of concept studies from being performed, and additional tools are needed.⁶⁻¹²

Figure 1.

Structures of reported compounds 1–5 that display inhibitory activity against DDR1.

While awaiting new high-throughput screening (HTS) hits, we utilized 1, due to its clean ancillary kinome panel,⁷ and showed that inhibition of DDR1 by 1 significantly reduced DDR1-mediated collagen IV production;⁶ however, the DMPK profile of 1 proved unacceptable along with noted cytotoxicity at modest concentrations. Thus, we elected to perform a multidimensional optimization campaign focused on 1, surveying a broad range of chemical diversity (Figure 2).

Figure 2.

Multidimensional optimization strategy for 1, exploring a wide range of structural diversity and assessing the minimum pharmacophore.

In short order, ∼200 analogues were synthesized and screened against both purified DDR1 and DDR2 kinase domains using a LanthaScreen Eu kinase binding assay binding assay,^6,13 and ∼80% of these proved inactive, highlighting core elements of the essential DDR1 pharmacophore (Figure 3). While frustrating that SAR was steep, data derived from this initial effort identified a smaller range of productive targets (7) and focused subsequent efforts.

Figure 3.

Regions and substituents explored in analogues 6 that were inactive as DDR1/DDR2 antagonists (<30% tracer binding in the presence of 100 nM inhibitor, single point assay),⁶ leading to a basic DDR1 pharmacophore 7.

The synthesis of analogues 7 (Scheme 1) required only four steps from commercial starting materials. Here, either 4-methyl-3-nitrobenzoic acid or 4-fluoro-3-nitrobenzoic acid 8 underwent a HATU-mediated amide coupling with 3- or 3,4-disubstituted anilines 9 to afford, after a microwave-assisted nitro reduction with SnCl₂, anilino amides 10 in yields ranging from 32 to 96% for the two steps. Treatment of anilino amides 10 with 5-bromonicotinaldehyde under reductive amination conditions yields 11 in 43–77% isolated yields. Finally, a standard Suzuki coupling reaction with 11 and various heteroaryl boronic acids delivers analogues 7 in 38–63% yields.¹³

Scheme 1. Synthesis of Analogues 7.

Reagents and conditions: (a) HATU, N,N-diisopropylethylamine, DCM, rt, 17 h, 49–91%; (b) SnCl₂, EtOAc/EtOH, 110 °C (μW), 25 min, 67–95%; (c) 5-bromonicotinaldehyde, NaBH₃CN, AcOH, MeOH, 50 °C, 2 h, 43–77%; (d) Aryl-B(OH)₂, Pd(dppf)Cl₂, Cs₂CO₃, 1,4-dioxanes/H₂O, 90 °C, 3 h, 38–63%.

The most robust series of active compounds are highlighted in Table 1, and all active analogues were dual DDR1/DDR2 inhibitors. We were pleased that a simple methylene homologation afforded active compounds, and our medicinal chemistry strategy explored several avenues on further improving compounds; here, one motivation was to find a suitable isosteric replacement for the carboxamide moiety. The matched-pair homologated analogue of 1, the aminomethyl pyridine 7a, proved more potent than the parent 1 (DDR1 IC₅₀ = 3.97 nM, DDR2 IC₅₀ = 4.79 nM). Replacement of the carboxamide with a pyrrole as in 7b, was 4–5-fold less active (DDR1 IC₅₀ = 15.2 nM, DDR2 IC₅₀ = 11.0 nM), and adding an eastern 4-position substituent to 7b, to afford 7c, lost ∼3-fold in DDR inhibitory activity (DDR1 IC₅₀ = 36.4 nM, DDR2 IC₅₀ = 44.1 nM). Replacement of the pyrrole with a 3-N-methylpyrazole, such as 7d, showed improved activity (DDR1 IC₅₀ = 8.53 nM, DDR2 IC₅₀ = 4.56 nM) relative to 1. Within the 7d core replacement of the methyl group in the central phenyl ring with a fluorine, alongside substitution of the eastern aryl methoxy to an aryl trifluoromethyl ether, provided another very potent analogue 7e (DDR1 IC₅₀ = 4.67 nM, DDR2 IC₅₀ = 7.39 nM). Removal of the western heterocycle altogether, as with the unsubstituted pyridine 7f, only lost ∼2-fold (DDR1 IC₅₀ = 10.5 nM, DDR2 IC₅₀ = 10.6 nM). Finally, a simple OCH₃ moiety, 7g, was also a viable heterocyclic/carboxamide replacement (DDR1 IC₅₀ = 5.89 nM, DDR2 IC₅₀ = 6.36 nM). As all analogues 6 were potent inhibitors of DDR1 and DDR2, we employed a panel of in vitro DMPK assays to further triage these compounds prior to deeper profiling. As a reference, the lead 1 displayed low free fraction in plasma (rat f_u = 0.018) and high predicted hepatic clearance from liver microsomes (rat CL_hep = 60 mL/min/kg). Pyrrole-based 7b possessed an inferior profile (rat f_u = 0.001 and rat CL_hep = 62 mL/min/kg) and thus was precluded from further advancement, as were the majority of analogues. Pyrazole 7d showed a comparable profile relative to 1 profile (rat f_u = 0.016 and rat CL_hep = 46 mL/min/kg), while 7e was superior (rat f_u = 0.089 and rat CL_hep = 43.2 mL/min/kg) as a potential rodent tool compound. On the basis of these data, 7e was further evaluated in a rat IV (0.5 mg/kg)/PO (3 mg/kg) PK study in a 10% EtOH/40% PEG400/50% saline vehicle (solution). Here, 7e displayed a good in vitro:in vivo correlation (IVIC), with moderate in vivo clearance (CL_p = 34.2 mL/min/kg), an ∼3 h half-life, moderate volume of distribution at steady state (V_ss = 4.3 L/kg), and 12.5% oral bioavailability with a rapid T_max (0.75 h). As 1 showed cytotoxicity in mesangial cells expressing Hu-DDR1b at low concentrations (∼1 μM) after a 24 h incubation period, 7e was assessed under similar conditions and found to be without cytotoxicity at 10 μM. Finally, we evaluated the kinome selectivity of 7e in a panel of 371 wild-type and mutant kinases at a concentration of 1 μM. Similar to the high selectivity displayed by 1, 7e showed (Figure 4) potent inhibition of only 27/371 kinases (7.2%).¹¹ Thus, on the basis of the DDR1/DDR2 binding potency, the rat in vitro and in vivo DMPK profiles, a lack of cytotoxicity, and a favorably clean kinome profile, 7e was advanced into key cell-based assays.

Table 1. Structures and DDR1/DDR2 Activities of Selected Analogues 7^a.

^aTR-FRET LanthaScreen Eu kinase binding assay with DDR1 and DDR2.⁶

Figure 4.

Kinome selectivity and structure of 7e (VU6015929).

The high affinity binding of 7e to DDR1 kinase detected by the LanthaScreen Eu kinase binding assay (Table 1) suggested that 7e would potently inhibit collagen-induced DDR1 phosphorylation in cells. To analyze the ability of 7e to inhibit DDR1 phosphorylation in cells, we incubated serum-starved HEK293 expressing full length human DDR1 cDNA (HEK293-DDR1) with various amounts of lead compound 1 or 7e together with collagen I. After 24 h, the levels of total and phosphorylated DDR1 were analyzed in cell lysates by Western blot analysis (Figure 5A). Both 1 and 7e inhibited collagen I-induced DDR1 phosphorylation in a dose dependent manner. However, lower doses of 7e were required to achieve better inhibition of DDR1 phosphorylation, visible at the lowest dose of 4 nM (Figure 5A,B). To better determine the IC₅₀ values, we used a capture ELISA assay using HEK293 cells expressing flagged human full length DDR1 cDNA (HEK293-DDR1-FLAG) incubated with inhibitors and collagen I for 18 h. Analysis of the phosphorylated DDR1/total DDR1 ratio revealed an IC₅₀ for 7e of 0.7078 ± 0.3533 nM, while the IC₅₀ for 1 was ∼7-fold weaker at 4.799 ± 2.066 nM. The low IC₅₀s for 1 and 7e confirm that both compounds are potent inhibitors of DDR1 (Figure 5C,D).

Figure 5.

7e blocks collagen I-induced DDR1 autophosphorylation in DDR1 expressing cells more potently than 1. (A) HEK293-DDR1b or (B) HEK293-DDR1b-FLAG cells were treated with serial dilutions of 1 or 7e then incubated with 50 μg/mL of collagen I. After 18–24 h, cells were lysed and the DDR1 phosphorylation was determined by Western blot analysis (A,B) or by capture ELISA (C,D). A representative Western blot of phosphorylated and total DDR1 is shown in A. pDDR1 and DDR1 bands were quantified as described in the SI methods, and the values are expressed as pDDR/DDR1 ratio. Data are the mean ± SEM of 4–6 independent experiments (B). (C) One representative dose–response curve of four experiments for pDDR1 inhibition by 1 and 7e. For each fitting curve, the IC₅₀ values of 1 and 7e were different with a p < 0.0001. t-Test of IC₅₀ values for four experiments were different between 1 and 7e with p = 0.0079 (D).

DDR1 is a positive regulator of collagen production, and we previously showed that activated/phosphorylated DDR1 promotes collagen IV production in kidney mesangial cells; moreover, we demonstrated that this effect can be inhibited by 1 in a DDR1 dependent manner.⁶ To determine whether 7e can also prevent DDR1-mediated collagen IV, we treated serum-starved DDR1-null mesangial cells reconstituted with human full length DDR1 cDNA with 1 and 7e (both at 3 μM). After 24 h, the levels of collagen IV were analyzed in cell lysates by Western blot analysis. As shown in Figure 6A, both 1 and 7e significantly inhibited collagen IV production compared to cells treated with vehicle only; however, with a lower p value for 7e (Figure 6B). Similar to 1, 7e did not decrease collagen IV production in the DDR1-null mesangial cells (data not shown). These results suggest that 7e (VU6015929) is an improved tool compound over 1 in affecting DDR1 phosphorylation due to a combination of its enhanced DDR1 potency and more drug-like physiochemical and DMPK profiles.

Figure 6.

DDR1 inhibitors 1 and 7e block collagen IV production in a DDR1 dependent manner. Serum-starved DDR1-null mesangial cells expressing human full length DDR1b cDNA were treated with DMSO, 1, or 7e (both at 3 μM) for 24 h and analyzed for the levels of collagen IV (CIV) and Akt (loading control) by Western blot analysis (A). CIV and Akt bands were quantified as described in SI methods. The values are the mean ± SEM of three experiments performed at least in triplicate (*p < 0.05, ** p < 0.0001 compared to no-inhibitor cells) (B) Western blot analysis of CIV and FAK (loading control) levels in serum-starved DDR1-null mesangial cells expressing human full length DDR1b cDNA treated with DMSO or DDR1-null mesangial cells (KO) treated with DMSO or 7e (3 μM) for 24 h. CIV and FAK bands were quantified by densitometry analysis, and values are the mean ± SEM of one representative experiment performed in triplicate (*p < 0.05, compared to DDR1b expressing cells).

In summary, we performed a multidimensional optimization effort around known DDR1/2 dual inhibitor 1, and quickly identified key elements of a DDR1 pharmacophore. Targeted optimization then produced 7e (VU6015929), a new dual DDR1/2 inhibitor with improved binding and cell-based potency, enhanced physiochemical and DMPK properties, and clean kinome profile to serve as a new in vitro and in vivo tool compound. Moreover, 7e was shown to potently inhibit collagen IV production, further validating a role for DDR1 as pro-fibrotic receptor. Results from ongoing in vivo work rodent disease models will be reported in due course.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.9b00382.

• General methods for the synthesis and characterization of all compounds, and methods for the in vitro and in vivo DMPK protocols and figures (PDF)

Author Contributions

D.E.J. and C.M.B. contributed equally. C.W.L. and A.P. drafted/corrected the manuscript and directed the science. D.E.J. performed the chemical synthesis. C.W.L. and A.P. oversaw the target selection and interpreted the biological data. C.M.B. performed the in vitro molecular pharmacology studies. A.L.B. performed the in vitro and in vivo DMPK studies. All authors have given approval to the final version of the manuscript.

The authors declare no competing financial interest.