Structure-based drug discovery of novel fused-pyrazolone carboxamide derivatives as potent and selective AXL inhibitors

Abstract

As a novel and promising antitumor target, AXL plays an important role in tumor growth, metastasis, immunosuppression and drug resistance of various malignancies, which has attracted extensive research interest in recent years. In this study, by employing the structure-based drug design and bioisosterism strategies, we designed and synthesized in total 54 novel AXL inhibitors featuring a fused-pyrazolone carboxamide scaffold, of which up to 20 compounds exhibited excellent AXL kinase and BaF3/TEL-AXL cell viability inhibitions. Notably, compound 59 showed a desirable AXL kinase inhibitory activity (IC₅₀: 3.5 nmol/L) as well as good kinase selectivity, and it effectively blocked the cellular AXL signaling. In turn, compound 59 could potently inhibit BaF3/TEL-AXL cell viability (IC₅₀: 1.5 nmol/L) and significantly suppress GAS6/AXL-mediated cancer cell invasion, migration and wound healing at the nanomolar level. More importantly, compound 59 oral administration showed good pharmacokinetic profile and in vivo antitumor efficiency, in which we observed significant AXL phosphorylation suppression, and its antitumor efficacy at 20 mg/kg (qd) was comparable to that of BGB324 at 50 mg/kg (bid), the most advanced AXL inhibitor. Taken together, this work provided a valuable lead compound as a potential AXL inhibitor for the further antitumor drug development.

Graphical abstract

A class of novel AXL inhibitors has been discovered, in which compound 59 exhibited potent AXL inhibition, high kinase selectivity and significant in vivo anti-tumor efficacy with desirable pharmacokinetic profile.

1. Introduction

AXL, a member of TAM (TYRO3, AXL and MER) subfamily of receptor tyrosine kinases (RTKs), is generally activated by binding to the endogenous ligand growth arrest specific protein 6 (GAS6), followed by self-dimerization and auto-phosphorylation¹. The GAS6/AXL signaling could promote tumor progression through affecting multiple downstream signaling transduction including PI3K-AKT, STAT, RAS-RAF-MEK-ERK, and so on (Supporting Information Fig. S1)^2,3. Moreover, AXL-RTKs could cross-talk to rewire the kinase signaling nodes in various malignancies⁴. GAS6/AXL axis is involved in multiple biological process of tumorigenesis and progression^2,5, 6, 7, 8, 9, such as tumor cell survival, motility, invasion and epithelial–mesenchymal transition (EMT). In particular, AXL is required for multiple steps of the metastasis cascade, thus blockage of AXL could decrease tumor metastasis10, 11, 12, 13, 14. Aberrant AXL signaling has also been found in several types of cancers and associated with high tumor recurrence and poor prognosis15, 16, 17, 18, 19. Accumulating evidences have highlighted the importance of GAS6/AXL axis in the tumor microenvironment (Fig. S1)20, 21, 22, where it regulates the expression of major histocompatibility complex I (MHC-I) and programmed death ligand-1 (PD-L1), as well as promotes secretion of immunosuppressive chemokines23, 24, 25. Besides, GAS6/AXL signaling promotes infiltration of macrophages, monocytes and myeloid-derived suppressor cells (MDSCs), but reduces infiltration of CD4⁺, CD8⁺ T-cells and conventional dendritic cells (DC) in the tumor^21,26. Furthermore, AXL high expression could cause innate as well as acquired resistance to anticancer therapies27, 28, 29, 30. In short, AXL is a promising anticancer therapeutic target since it plays an important role in promoting tumorigenesis and immunosuppressive microenvironment as well as mediating drug resistance.

Currently available small-molecule AXL inhibitors are mainly classified into U-shaped type I and linear type II ATP-competitive inhibitors based on their structure characteristics and binding modes with AXL kinase (Fig. 1)31, 32, 33. BGB324 (1) also called Bemcentinib, was considered as the first type I selective AXL inhibitor with an IC₅₀ of 14 nmol/L³⁴, and its monotherapy and combination therapies have currently entered phase II clinical trials for several cancers including NSCLC and AML with positive clinical benefits35, 36, 37, 38. Cabozantinib (2) represents a classic type II kinase inhibitor, which can occupy the additional allosteric pocket of kinase formed by the aspartate-phenylalanine-glycine motif far away from the active site (DFG-out), and was approved for the therapy of MTC, renal cell carcinoma and hepatocellular carcinoma³⁹. However, Cabozantinib is not a selective AXL inhibitor, and it has very strong inhibitory activities against multiple kinases including VEGFR2, c-Met, KIT, FLT3, etc., which may lead to some potential toxicities and side effects. Some similar derivatives, such as BMS-777607 (3)⁴⁰, compounds 4–641, 42, 43 and other reported AXL inhibitors^31,32,44, 45, 46, 47, 48, 49, 50, have been in different stages of preclinical and clinical trials for various cancer indications. In view of the importance of AXL in tumorigenesis and progression, the development of selective AXL inhibitors is of great significance for the clinical treatment of tumors.

Figure 1.

Chemical structures of reported representative AXL inhibitors.

Although AXL inhibitors could be a promising and powerful approach for various malignancies including solid tumors and hematological malignancies, the development of selective AXL inhibitors as anticancer agents still remains a great challenge. In this work, we have developed a novel class of selective AXL inhibitors featuring a fused-pyrazolone carboxamide scaffold, and demonstrated that most of them had more excellent AXL inhibitory activities and antitumor efficiencies than the most advanced selective AXL inhibitor BGB324.

2. Results and discussion

2.1. Rational design of novel AXL inhibitors

Considering the significant therapeutic values of AXL inhibitors in overcoming drug resistance, tumor metastasis and immunosuppression, we first conducted a thoroughly biological screening of our in-house compound library to discover a new class of AXL inhibitors with a novel scaffold for hit-to-lead development. Fortunately, three fused-pyrazolone carboxamide derivatives 7–9 displayed acceptable inhibitory activities against AXL kinase, especially the inhibition rate of compound 9 reached 75.4 % at 100 nmol/L. More importantly, compound 9 displayed a relative high selectivity against AXL in the kinase screening at 500 nmol/L, suggesting that compound 9 could be a novel hit compound for further modification (Fig. 2).

Figure 2.

Discovery of selective AXL inhibitors featuring a novel fused-pyrazolone carboxamide scaffold.

To find more potent AXL inhibitors, we adopted structure-based drug design and bioisosterism strategies to guide the structure design of compounds. We firstly performed the molecular docking analysis of compound 9 in the AXL DFG-out homology model (constructed with PDB ID: 7AAY and chain B of PDB ID: 7AVX as the templates) (Fig. 3). The results showed that the dual carbonyl motif of the fused-pyrazolone carboxamide in the DHBA (dual hydrogen bond acceptor) group formed two hydrogen bonds with K567 and D690 located in the allosteric pocket of AXL kinase, respectively, and its fused-piperidine ring occupied the allosteric cavity relatively well, which were speculated to be the key to the activity and selectivity of AXL inhibitors. In addition, the pyrrolo [2,3-d]pyrimidine in the head region of compound 9 could form a hydrogen bond with M623 located in the ATP binding region, enhancing the interaction between compound 9 and AXL kinase. The linker occupied the hydrophobic channel, which perfectly connected the DHBA moiety and head region. As the fused-pyrazolone carboxamide structure in the DHBA group is particularly important for the biological activity and selectivity, we retained this scaffold in the design of target compounds. Based on the above molecular docking analysis, we rationally carried out structural modifications on the head region using different aromatic heterocyclic rings, the linker and the phenyl in the DHBA moiety.

Figure 3.

Rational design of novel selective AXL inhibitors by employing the structure-based drug design and bioisosterism strategies.

Lead optimization was initiated with the head region, which was firstly explored by introducing diversified fused or single nitrogen-containing heterocycles (10–15) to form more appropriated interactions with the residues of AXL kinase hinge region. Then, single or double substituents were introduced into the phenyl ring of the DHBA moiety to investigate the hydrophobic interaction between different substituted benzene rings and the allosteric pocket (16–28). After that, different linkers connecting the head region and DHBA moiety were investigated to design compounds 29–55. In the end, given the fact that quinoline rings often serve as adenine analogs to competitively bind to the hinge region of kinases, we further selected 6,7-dimethoxyquinoline as the head region and designed compounds 56–63.

2.2. Synthesis of novel AXL inhibitors

Generally, all target compounds were prepared by a condensation of acid fragments 69a–m and the corresponding amine fragments 74a–b, 80a–o, 82 and 85.

The synthesis of key acid fragments 69a–m were described in Scheme 1⁵¹. The free substituted phenylhydrazines, released from their hydrochlorides 64a–m under Na₂CO₃ aqueous solution, were condensed with 5-chloropentanoyl chloride to form intermediates 65a–m, followed by another condensation with ethyl 3-chloro-3-oxopropanoate to afford 66a–m. Next, they were sequentially treated by sodium hydride (NaH) and 1,8-diazabicyclo [5.4.0]undec-7-ene (DBU) in the subsequent intramolecular nucleophilic substitution and aldol condensation to provide the corresponding ester derivatives 68a–m, which were hydrolyzed to conveniently generate the fused-pyrazolone acid fragments 69a–m.

Scheme 1.

Synthesis of key acid fragments 69a–m. Reagents and conditions: (a) 5-Chloropentanoyl chloride, 10% Na₂CO₃ (aq), DCM, 0 °C–rt, overnight, 29%–58%; (b) Ethyl 3-chloro-3-oxopropanoate, Na₂CO₃, DCM, rt, overnight, 56%–80%; (c) NaH, DMF, 0 °C–rt, 5 h, 83%–90%; (d) DBU, 50 °C, 5 h, 85%–90%; (e) 2 N KOH (aq), EtOH, reflux, 1 h; (f) 2 N HCl (aq), rt, 0.5 h, 61%–79%.

Compounds 7–9 were synthesized according to the procedures outlined in Scheme 2. Commercially available 4-chloro-5-iodo-7H-pyrrolo [2,3-d]pyrimidine 70 was firstly protected with (2-(chloromethoxy)ethyl)trimethylsilane (SEMCl) to afford the intermediate 71, which subsequently underwent a Suzuki coupling with phenylboronic acid, an intermolecular nucleophilic substitution with 4-nitrophenol or 2-fluoro-4-nitrophenol, and a nitro reduction to give the amine fragments 74a–b, followed by a condensation with 69a or 69i, and a two-step deprotection to generate the target compounds 7–9, respectively.

Scheme 2.

Synthesis of compounds 7–9. Reagents and conditions: (a) SEMCl, NaH, DMF, 0 °C–rt, overnight, 80%; (b) Phenylboric acid, Pd(PPh₃)₄, Na₂CO₃, H₂O, toluene, Ar, 90 °C, overnight, 71%; (c) 4-Nitrophenol or 2-fluoro-4-nitrophenol, PhCl, 140 °C, 8 h, 56%–59%; (d) NiCl₂·6H₂O, NaBH₄, THF/MeOH (3:1, v/v), 0 °C–rt, 2 h, 64%–66%; (e) 69a or 69i, HATU, DIPEA, DCM, rt, overnight, 69%–75%; (f) TFA, DCM, rt, overnight, 82%–86%; (g) NaOH, THF/H₂O (5/4, v/v), rt, overnight, 75%–79%.

The synthetic routes of target compounds 10–50, 52–57 and 59–63 were summarized in Scheme 3. The chlorinated heterocyclic raw materials 77a–g proceeded with substituted 4-nitrophenols 78a–g (H, 2-F, 2-Cl, 2-Me, 2-OMe, 2-CF₃ and 3-F) or 4-nitronaphthalen-1-ol 78h to give nitro intermediates 79a–o, which were further transformed into the corresponding amine fragments 80a–o under the Fe/NH₄Cl condition. Finally, these amine fragments were separately condensed with 69a–m to yield the ultimate compounds 10–50, 52–57 and 59–63.

Scheme 3.

Synthesis of compounds 10–50, 52–57 and 59–63. Reagents and conditions: (a) K₂CO₃, DMF, 85 °C, overnight, 58%–73%; (b) PhCl, 140 °C, 8 h, 52%–63%; (c) Fe/NH₄Cl, EtOH/H₂O (1:1, v/v), 75 °C, 1 h, 65%–79%; (d) 69a–m, HATU, DIPEA, DCM, rt, overnight, 80%–90%.

As shown in Scheme 4, the amine fragment 82 was smoothly obtained by treating 4-chloro-6,7-dimethoxyquinazoline 77a with tert-butyl (5-hydroxypyridin-2-yl)carbamate 81, and underwent a subsequent condensation reaction with 69a to yield the ultimate compound 51. Moreover, 6,7-dimethoxyquinolin-4-ol 83 proceeded with 2-fluoro-5-nitropyridine to offer a nitro intermediate 84, which was reduced and subsequently condensed with 69a to offer the desired compound 58.

Scheme 4.

Synthesis of compounds 51 and 58. Reagents and conditions: (a) PhCl, 140 °C, 8 h, 39%; (b) 69a, HATU, DIPEA, DCM, rt, overnight, 85%–88%; (c) 2-Fluoro-5-nitropyridine, Cs₂CO₃, DMF, 0 °C–rt, 2–3 h, 46%; (d) Fe/NH₄Cl, EtOH/H₂O (1:1, v/v), 75 °C, 1 h, 75%.

2.3. Structure–activity relationship (SAR) exploration and their anticancer activities in vitro

The inhibitory activities of all compounds against AXL kinase were determined using an enzyme-linked immunosorbent assay (ELISA), and BGB324 (1), the most advanced (in clinical Phase II) selective AXL inhibitor, served as the positive control. All results were summarized in Table 1 and Table 2. Firstly, in the hope of forming more appropriate interactions with the hinge region of AXL kinase, we introduced diversified heterocyclic fragments at the head region, and obtained compounds 10–15. The results shown in Table 1 revealed that compound 10 containing a quinazoline fragment showed the most potent AXL kinase inhibitory activity with an IC₅₀ value of 43.9 nmol/L, however, other fused heterocycles (11–13) or single heterocycles (14 and 15) as the head region all resulted in great loss of activities against AXL kinase, suggesting that quinazoline fragment was the most appropriate head region to enhance the AXL kinase inhibitory activity. Therefore, we kept quinazoline fragment in the head region, and modified the linker and DHBA group to synthesize compounds 16–28.

Table 1.

Structures of compounds 10–28 and their inhibitory activities against AXL kinase and BaF3/TEL-AXL cell viability^a.

Compd.	R1	R2	R4	IC50 (nmol/L)b or inhibition ratec
				AXL kinase	Cell viabilityd
10	H	H		43.9 ± 4.7	–
11	H	H		−9.5 %	–
12	H	H		−17.7 %	–
13	H	H		−18.9 %	–
14	H	H		−6.0 %	–
15	H	H		6.3 %	–
16	H	F		8.1 ± 0.2	<0.1
17	2-F	F		9.4 ± 3.2	<0.1
18	3-F	F		5.9 ± 0.6	<0.1
19	4-F	F		7.8 ± 1.1	<0.1
20	2-Cl	F		10.5 ± 0.5	7.0 ± 4.2
21	4-Cl	F		42.6 ± 6.2	–
22	2-Me	F		15.7 ± 1.7	7.6 ± 4.1
23	4-Me	F		27.3 ± 2.6	–
24	2,5-di-F	F		5.5 ± 1.6	<0.1
25	2,5-di-Me	F		78.4 ± 2.1	–
26	2-Cl-5-F	F		13.7 ± 3.1	5.2 ± 4.1
27	2,4-di-F	F		25.1 ± 1.6	–
28	2-Cl-4-F	F		13.5 ± 1.8	4.4 ± 2.2
BGB324 (1)				7.7 ± 2.9	95.1 ± 3.7

^aAll data were obtained as mean values from two independent experiments.

^bThe IC₅₀ values were represented as mean ± SD.

^cThe inhibition rate was determined at a concentration of 100 nmol/L.

^dCell viability inhibition was tested in BaF3/TEL-AXL cells, “–” means not evaluated.

Table 2.

Structures of compounds 29–63 and their inhibitory activities against AXL kinase and BaF3/TEL-AXL cell viability^a.

Compd.	X	Linker	R1	IC50 (nmol/L)b or inhibition ratec
				AXL kinase	Cell viabilityd
29	N		H	10.0 ± 1.4	2.3 ± 0.4
30	N		2-F	13.4 ± 0.8	<0.1
31	N		4-F	30.5 ± 4.9	–
32	N		2,5-di-F	37.0 ± 1.4	–
33	N		H	16.9 ± 2.5	1651.3 ± 443.7
34	N		2-F	30.9 ± 1.4	–
35	N		4-F	19.6 ± 1.9	–
36	N		2,5-di-F	6.6 ± 2.9	3632.1 ± 493.6
37	N		H	23.2 %	–
38	N		2-F	32.3 %	–
39	N		4-F	15.6 %	–
40	N		2,5-di-F	35.7 %	–
41	N		H	56.4 %	–
42	N		2-F	28.0 %	–
43	N		4-F	19.9 %	–
44	N		2,5-di-F	59.6 %	–
45	N		H	59.4 %	–
46	N		2-F	50.4 %	–
47	N		4-F	51.9 %	–
48	N		2,5-di-F	62.4 %	–
49	N		2-F	47.9 %	–
50	N		2,5-di-F	9.1 ± 0.2	7.8 ± 0.7
51	N		H	16.8 ± 3.5	3.3 ± 1.4
52	N		H	41.0 %	–
53	N		2-F	41.4 %	–
54	N		4-F	39.0 %	–
55	N		2,5-di-F	50.4 %	–
56	CH		H	4.0 ± 0.4	1.0 ± 0.2
57	CH		H	2.5 ± 0.3	<0.1
58	CH		H	50.9 %	–
59	CH		4-F	3.5 ± 0.8	1.5 ± 0.4
60	CH		2,5-di-F	1.2 ± 0.5	<0.2
61	CH		2-Cl-5-F	0.7 ± 0.1	<0.2
62	CH		2,4-di-F	6.1 ± 0.8	0.7 ± 0.3
63	CH		2-Cl-4-F	2.3 ± 0.2	<0.2
BGB324 (1)				7.7 ± 2.9	95.1 ± 3.7

^aAll data were obtained as mean values from two independent experiments.

^bThe IC₅₀ values were represented as mean ± SD.

^cThe inhibition rate was determined at a concentration of 100 nmol/L.

^dCell viability inhibition was tested in BaF3/TEL-AXL cells, “–” means not evaluated.

Encouragingly, compound 16 with 2-fluoro-4-aminophenol as the linker had a more excellent inhibitory potency against AXL kinase with an IC₅₀ value of 8.1 nmol/L. Meanwhile, compound 16 strongly inhibited the BaF3/TEL-AXL cell viability (IC₅₀ < 0.1 nmol/L). Therefore, we further investigated the hydrophobic effects of different substituted benzene rings at R₁ position. Introduction of a single F atom at the ortho, meta, and para positions of the benzene ring alone (17–19) resulted in the similar AXL kinase inhibitions (IC₅₀ < 10 nmol/L), and compound 18 possessed the strongest AXL kinase inhibition with an IC₅₀ value of 5.9 nmol/L. It is worth mentioning that compounds 16–19 possessed very strong anti-BaF3/TEL-AXL cell viability potencies with their IC₅₀ < 0.1 nmol/L. We further explored Cl and Me substituents at R₁ position, and found that the activities of ortho-substituted compounds 20 and 22 were superior to that of 21 and 23 substituted at para-position of the benzene ring (20 IC₅₀: 10.5 nmol/L vs 21 IC₅₀: 42.6 nmol/L), respectively. Besides, compounds 24–28 with double substituents on the benzene ring also showed desirable AXL kinase inhibitions with IC₅₀ ranging from 5.5 to 78.4 nmol/L.

Given the positive effects of 2-fluoro-4-aminophenol linker, we continued to explore the effects of the position of F atom in the linker and its bioisosteric groups such as Cl, Me, OMe, CF₃ groups on AXL kinase inhibitory activity. The results in Table 2 revealed that changing the position of F atom maintained the potent AXL inhibitory activities of compounds 29–32 with IC₅₀ ranging from 10.0 to 37.0 nmol/L. Then, among compounds 33–36 featuring a 2-chloro-4-aminophenol linker, the IC₅₀ value of 36 (2,5-di-F for R₁) reached 6.6 nmol/L. However, when introducing Me, OMe or CF₃ substituted 4-aminophenol as the linker, the corresponding compounds 37–48 showed poor inhibitory activities against AXL kinase at a concentration of 100 nmol/L, as did for 52–55 with a 4-nitronaphthalen-1-ol linker, which may be attributed to that the flat and slender hydrophobic channel was not enough to accommodate these larger substituents. Moreover, when 4-aminophenol was used as a linker, only compound 50 bearing 2,5-di-F substituents at the benzene ring of the DHBA moiety showed a significant AXL kinase inhibition (IC₅₀: 9.1 nmol/L). Notably, the 6-aminopyridin-3-ol, as in 51 with an IC₅₀ value of 16.8 nmol/L, could also be regarded as a potential linker.

It is well known that quinoline and its analogs could be bioisosteric groups of quinazoline, and they are a class of privileged scaffolds in many anticancer active compounds and drugs52, 53, 54, 55, 56, 57, 58, 59. Therefore, we designed and synthesized compounds 56–63 by replacing 6,7-dimethoxyquinazoline with 6,7-dimethoxyquinoline as the head region to explore their effects on AXL inhibitory activity. It revealed that compounds 56 and 57 with 4-aminophenol and 2-fluoro-4-aminophenol linkers, respectively, displayed excellent and comparable biological activities (56 IC₅₀: 4.0 nmol/L and 57 IC₅₀: 2.5 nmol/L), which were far superior to that of compound 58 with a 5-aminopyridin-2-ol linker. Based on these results, we further fixed R₁ as the 4-F substituent and synthesized compound 59, which also significantly blocked AXL kinase with an IC₅₀ value of 3.5 nmol/L. Meanwhile, compound 59 showed a potent BaF3/TEL-AXL cell viability inhibition with an IC₅₀ value of 1.5 nmol/L. Besides, we further introduced double halogen substituents, such as 2,5-di-F (60), 2-Cl-5-F (61), 2,4-di-F (62) and 2-Cl-4-F (63), to R₁ position and demonstrated their excellent in vitro inhibitory effects against AXL with IC₅₀ values less than 10 nmol/L. Overall, 6,7-dimethoxyquinoline was demonstrated to be beneficial for the activities of these AXL inhibitors. In summary, in the SAR discussion, we successfully designed and synthesized in total 54 novel AXL inhibitors featuring a fused-pyrazolone carboxamide scaffold, of which up to 20 compounds including compounds 18, 24 and 59 exhibited more potent AXL inhibitory activities and anti-BaF3/TEL-AXL cell viability potencies than the most advanced AXL inhibitor BGB324 (1).

Representative compound 59 with potent AXL inhibitory activities was further progressed into a docking analysis with the AXL DFG-out homology model (generated from PDB ID: 7AAY and chain B of PDB ID:7AVX). As shown in Fig. 4, the quinoline N atom bound to the residue M623 located in the hinge region via a key hydrogen bond interaction. The 4-aminophenol was located at the hydrophobic tunnel of AXL kinase. Moreover, the two carbonyl groups in the novel fused-pyrazolone carboxamide scaffold favorably played a DHBA role to form two hydrogen bonds with residues K567 and D690 in the allosteric pocket induced by the DFG-out conformation, respectively. More importantly, the fused-piperidine ring occupied the allosteric cavity relatively well. These results rationalized the potent AXL inhibitory activities of compound 59.

Figure 4.

The binding mode of 59 (green) with the AXL DFG-out homology model (generated from PDB ID: 7AAY and chain B of PDB ID:7AVX). Hydrogen bonds were indicated by red dashed lines.

2.4. Pharmacokinetic (PK) profiles of compounds 18, 24 and 59 in mice

Based on comprehensive consideration of the structural characteristics and their potent in vitro AXL inhibitory activities, compounds 18, 24 and 59 were selected to test their PK properties in mice through oral administration and intravenous injection to provide a dosage reference for in vivo efficacy studies. As shown in Table 3, compounds 18, 24 and 59 showed the similar high oral in vivo exposures (18: 15,903 ng h/mL vs 24: 13,500 ng h/mL vs 59: 14,800 ng h/mL) and low clearance rates (18: 2.05 mL/min/kg vs 24: 1.70 mL/min/kg vs 59: 3.86 mL/min/kg), but the oral bioavailabilities of 22.9% for 18 and 36.1% for 59 indicated that they had better gastrointestinal permeabilities than that of compound 24. Compounds 18 and 59 also exhibited the relative higher peak concentrations (18 C_max: 2970 ng/mL and 59 C_max: 2300 ng/mL), and compound 59 had a shorter time to peak (T_max: 1.0 h), indicating that it worked quickly. Overall, these results confirmed that compounds 18 and 59 possessed relatively desired pharmacokinetic properties.

Table 3.

Pharmacokinetic parameters of compounds 18, 24 and 59 in mice^a.

Compd.	Admini.	Tmax	Cmax	AUC0–t	AUC0–∞	t1/2	CL	F
		(h)	(ng/mL)	(ng·h/mL)	(ng·h/mL)	(h)	(mL/min/kg)	(%)
18	po	2.0	2970	15,903	16,007	3.24	/	22.9
	iv	/	/	34,754	42,066	10.40	2.05
24	po	2.0	1490	13,500	14,200	6.77	/	13.8
	iv	/	/	48,900	49,100	3.38	1.70
59	po	1.0	2300	14,800	14,900	2.94	/	36.1
	iv	/	/	20,500	21,600	1.67	3.86

^aAll data for each group were obtained as mean values from three animals. po at dose of 10 mg/kg iv at dose of 5 mg/kg.

2.5. In vivo antitumor activities of compounds 18 and 59

Encouraged by the excellent in vitro AXL inhibitory potencies and outstanding PK properties of compounds 18 and 59, we further established the BaF3/TEL-AXL cell-derived mouse xenograft model to investigate the in vivo antitumor activities (Fig. 5). Mice were orally administered with 18 and 59 at different doses of 4 and 20 mg/kg once a day for 7 consecutive days, respectively. As shown in Fig. 5A, compound 18 in both 4 mg/kg dose group and 20 mg/kg dose group exhibited significant inhibitions on tumor growth, with TGI of 89.5 % and 93.4 %, respectively. Compound 59 markedly suppressed tumor growth in a dose-dependent manner, and its antitumor efficacy at 20 mg/kg (TGI: 89.4 %) was comparable to that of BGB324 (the most advanced selective AXL inhibitor) at a dose of 50 mg/kg twice a day (TGI: 92.0 %). In addition, no significant body weight loss of tumor-bearing mice during the treatment demonstrated the well tolerance of compounds 18 and 59 (Fig. 5B).

Figure 5.

Compounds 18 and 59 significantly inhibited tumor growth in BaF3/TEL-AXL mouse xenograft model. (A, B) Tumor volume (A) and body weight (B) changes during the treatment period. The data were expressed as the mean ± SEM. Significant differences from the vehicle group were determined using student's t-test analysis, ∗P < 0.05. (C) Compound 59 inhibited intratumoral AXL phosphorylation of subcutaneous BaF3/TEL-AXL tumor tissues. Subcutaneous BaF3/TEL-AXL tumor tissues were obtained 1 h after the last dosage. Then, intratumoral AXL phosphorylation level was determined by immunoblotting analysis.

Given that compound 59 exhibited good and dose-dependent in vivo anti-tumor effects, at the end of treatment, subcutaneous BaF3/TEL-AXL tumor tissues were obtained. Then, intratumoral AXL phosphorylation level was tested to validate in vivo AXL targeting-activity of compound 59. The robust inhibitions of p-AXL were observed in both treatment groups (Fig. 5C). Taken together, compound 59 exhibited strong in vitro AXL inhibitory activities and dose-dependent in vivo antitumor activities, along with favorable pharmacokinetics profiles, indicating that compound 59 represented a potent AXL inhibitor and could be used for the further preclinical in-depth evaluation.

2.6. In vitro antitumor studies of compound 59 and its kinase selectivity

2.6.1. Inhibitory effects of compound 59 against GAS6/AXL-mediated cancer cell invasion, migration and wound healing

It has been widely demonstrated that AXL overactivation is closely related to tumor invasion and metastasis^5,8. Thus, we evaluated the inhibitory effects of compound 59 against GAS6/AXL axis mediated tumor cell invasion and migration. SNU449 hepatocellular carcinoma cell line and NCI-H1299 lung cancer cell line were used for the cell invasion and migration assays, respectively. Compound 59 dose-dependently suppressed the invasion of SNU449 cells, and exhibited a significant inhibition against cell invasion at a concentration of 20 nmol/L (Fig. 6A). Similarly, a significant reduction of the migration of NCI-H1299 cells was observed after a treatment with 20 nmol/L compound 59 (Fig. 6B). We also performed wound healing assay in NCI-H1299 cells to validate the effect of compound 59 on cell motility. As shown in Fig. 7, significant acceleration of wound healing was observed upon GAS6 stimulation, while compound 59 treatment obviously inhibited GAS6-accelaerated wound healing, still showing significant scratch gaps during the same period. In addition, under the same conditions of cell motility, we found that compound 59 had no significant inhibitory effects on the cell viability of these two cell lines (Supporting Information Fig. S5), excluding that its significant inhibitions on cell migration and invasion were due to the direct cellular toxic effect. Together, compound 59 potently attenuated GAS6/AXL-mediated tumor cell invasion and migration, indicating its therapeutic potential on tumor invasion and metastasis.

Figure 6.

Compound 59 suppressed AXL-mediated tumor cell invasion and migration. (A) Effect of 59 on the GAS6-stimulated SNU449 cell invasion. Scale bars, 10 μm. (B) Effect of 59 on the GAS6-stimulated NCI-H1299 cell migration. Scale bars, 10 μm.

Figure 7.

Compound 59 inhibited NCI-H1299 cell wound healing. NCI-H1299 cells were stimulated with 500 ng/mL GAS6 alone or co-treated with different concentrations of compound 59 or BGB324 for 24 h. Scratch healing at 0 and 24 h was observed and recorded. Representative images of triplicates were shown. Scale bars, 600 μm.

In addition, we also detected the influence of compound 59 on the E-cadherin expression, which is a canonical epithelial hallmark, and loss of E-cadherin is a hallmark of EMT^60,61. Interestingly, GAS6 stimulation obviously downregulated E-cadherin expression, and such downregulation was significantly reversed upon compound 59 treatment (Supporting Information Fig. S6), suggesting that reversion of EMT may contribute to the inhibitory effect of compound 59 on tumor cell metastasis.

2.6.2. Inhibitory effects of compound 59 on cellular AXL signaling pathway

The inhibitory effect of compound 59 against the cellular AXL signaling pathway was investigated by Western blotting. Firstly, the classic gain of function model cell BaF3/TEL-AXL was used. As shown in Fig. 8A, compound 59 dose-dependently inhibited the phosphorylation of AXL and its key downstream signaling molecule AKT and STAT3, and showed a bit weaker inhibtion on p-ERK. We also tested the inhibitory effect of compound 59 against the ligand GAS6-stimulated AXL activation in NCI-H1299 cells. Consistently, the phosphorylation of AXL induced by GAS6 was significantly inhibited by compound 59 (Fig. 8B). Similarly, accompanied with p-AXL inhibition upon compound 59 or BGB324 treatment, p-AKT and p-STAT3 levels were significantly inhibited, however, p-ERK was marginally inhibited upon compound 59 treatment. These results revealed that compound 59 could inhibit cellular AXL signaling. And, consistent with its inhibition on BaF3/TEL-AXL cell viability, the inhibitory potencies of compound 59 against cellular AXL signaling were much higher than that of BGB324, the most advanced selective AXL inhibitor. Interestingly, upon GAS6 stimulation in NCI-H1299 cells, acompanied with significant enhancement of p-AXL level, only p-AKT was obviously augmented while there was no obvious enhancement on p-STAT3 and p-ERK level. Our results together with other studies62, 63, 64, 65, 66 suggested that the cellular downstream signaling upon AXL activation or AXL-targeting inhibiton may be different among cell lines.

Figure 8.

The effects of compound 59 on AXL signaling pathway in BaF3/TEL-AXL and NCI-H1299 cells. (A) BaF3/TEL-AXL cells were treated with compound 59 or BGB324 at indicated concentrations for 1 h, followed by immunoblotting analysis. (B) NCI-H1299 cells were deprived serum for 24 h, treated with compound 59 or BGB324 for 1 h, and then stimulated with 500 ng/mL GAS6 for 15 min, followed by immunoblotting analysis.

2.6.3. Kinase selectivity profile of compound 59

Kinase selectivity profile of compound 59 was also performed on a panel of 62 human protein kinases highly related to oncology at a concentration of 100 nmol/L. As shown in Fig. 9, compound 59 only significantly inhibited the target kinase AXL, MER (a TAM subfamily kinase) and c-Met (an AXL homology kinase) with inhibition rates over 90 %. No obvious inhibitory effect was observed in other 55 tested kinases with inhibition rates less than 50 %. A further IC₅₀ study indicated that the inhibitory activity of compound 59 against c-Met kinase with an IC₅₀ of 17.3 nmol/L was weaker than that against AXL kinase with an IC₅₀ of 3.5 nmol/L. In addition, we further evaluated the effect of compound 59 on c-Met-mediated cell viability in a human lung cancer cell line EBC-1 that harbors MET amplification, and the results demonstrated that compound 59 inhibited the viability of EBC-1 cells with an IC₅₀ value of 534.9 nmol/L. Overall, compound 59 displayed a good kinase selectivity.

Figure 9.

Kinase profile map of compound 59 drawn with KinMap⁶⁷.

3. Conclusions

It's increasingly clear that inhibiting the overexpression and overactivation of AXL will represent a promising strategy for the improvement of malignant tumors therapeutic responses. By rational structure-based drug design on the hit compound 9, we designed and synthesized in total 54 novel fused-pyrazolone carboxamide derivatives as potent and selective AXL inhibitors. Twenty compounds exhibited excellent AXL inhibitions and anticancer activities at the nanomolar level. Among them, compounds 18 and 59 displayed desirable PK profiles with excellent peak concentrations (18 C_max: 2970 ng/mL and 59 C_max: 2300 ng/mL), in vivo exposures (18 AUC: 15,903 ng h/mL and 59 AUC: 14,800 ng h/mL) and oral bioavailabilities (18 F: 22.9 % and 59 F: 36.1 %). More importantly, both these two compounds displayed potent antitumor efficiencies in the BaF3/TEL-AXL cell-derived mouse in vivo model, and compound 59, tested as the representative, showed a strong inhibitory activity against intratumoral AXL phosphorylation. Besides, compound 59 exhibited a good kinase selectivity property, and effectively blocked cellular AXL signaling in both BaF3/TEL-AXL and GAS6-stimulated NCI-H1299 cells, and significantly suppressed GAS6/AXL-mediated cancer cell invasion, migration and wound healing in a dose-dependent manner. Moreover, compound 59 significantly reversed the GAS6-stimulated downregulation of E-cadherin, the canonical epithelial hallmark, suggesting that reversion of EMT may contribute to the inhibitory effect of compound 59 on tumor cell metastasis. In conclusion, as a novel potent and selective AXL inhibitor, compound 59 possesses significant in vitro and in vivo antitumor activities, which are more potent than BGB324, as well as desirable pharmacokinetic profiles, and it could be a valuable AXL-targeted lead compound for further optimization.

4. Experimental

4.1. Chemistry

The synthetic procedures and characterization data of all target compounds and key intermediates were described in Supporting Information.

4.2. Pharmacodynamics studies

4.2.1. Kinase inhibition assays

The inhibitory activities of all tested compounds against AXL and c-Met kinases except for the kinase selectivity tests of compounds 9 (Supporting Information Table S1) and 59 (Supporting Information Table S2) were determined using the previously reported ELISA. A modified 4-parameter logistic model was employed to calculate IC₅₀ values from the inhibition curves in at least two separate experiments. When a compound's inhibitory rate is higher than 50 % at the minimum tested concentration, its IC₅₀ will be lower than the tested minimum concentration. Moreover, kinase selectivities of compound 9 (500 nmol/L) evaluated against 65 protein kinases panel and compound 59 (100 nmol/L) evaluated against 62 protein kinases panel were performed at kinase screening platform of Eurofins Discovery (France).

4.2.2. Cell viability assays

BaF3/TEL-AXL, NCI-H1299, SNU449 and EBC-1 cells were seeded in 96-well plates with growth media at a low density, and incubated overnight. For BaF3/TEL-AXL cells, designated concentrations of BGB324 or tested compounds were added to each well, and the cells were further incubated for 48 h. For EBC-1 cells, the cells were treated with different concentrations of compound 59 or INCB28060 for 72 h. For NCI-H1299 and SNU449 cells, the cells were stimulated with 500 ng/mL GAS6 and treated with different concentrations of compound 59 or BGB324 for 24 h. Finally, the effects of tested compounds against BaF3/TEL-AXL, NCI-H1299, SNU449 and EBC-1 cells viability were determined by a cell counting kit (CCK-8) (Dojindo, Japan) assay. IC₅₀ values were calculated by concentration–response curve fitting using a SoftMax pro-based four-parameter method (SoftMax® Pro Software, Version 5.4.1).

4.2.3. Western blot analysis

BaF3/TEL-AXL cells and NCI-H1299 cells were cultured under regular growth conditions to the exponential growth phase. For BaF3/TEL-AXL cells, they were treated with the indicated dose of BGB324 or compound 59 for 1 h. For NCI-H1299 cells, they were serum-deprived for 24 h prior to treatment for 1 h with BGB324 or compound 59, and then stimulated with GAS6 for 15 min. Then, the tested cells were lysed in 1 × sodium dodecyl sulfate (SDS) sample buffer. For mice tumor tissues, they were homogenized in RIPA protein lysate by using an electric homogenizer and then determined protein content. Subsequently, the protein samples were lysed in 2 × SDS sample buffer. The lysates were resolved on 10 % SDS-PAGE and transferred to nitrocellulose membranes. Proteins were probed with specific antibodies [against p-AXL (Y702), AXL, p-AKT (S473), AKT, p-STAT3(Y705), STAT3, p-ERK(Y202/204), ERK and GAPDH] (Cell Signaling Technology, USA) and a subsequent secondary horseradish peroxidase-conjugated antibody. Finally, immunoreactive proteins were detected using an enhanced chemiluminescence detection reagent (Thermo Fisher Scientific, USA).

4.2.4. Cell migration and invasion assays

General procedures for the migration assay as follow. NCI-H1299 cells suspended in serum-free medium (1.5 × 10⁵ cells per well) were seeded in 24-well Transwell plates (pore size, 8 μm; Corning). The bottom chambers were filled with serum-free medium supplemented with GAS6 (500 ng/mL), and designated concentrations of BGB324 or compound 59 were added to both sides of the membrane. The cultures were maintained for further 24 h, and then the non-motile cells at the top of the filter were removed using a cotton swab. The migrating cells were fixed in ethanol (90%) for 1 h and stained with crystal violet (0.1%) for 15 min at room temperature. General procedures for the invasion assay as follow. SNU449 cells were cultured in the top chambers containing matrigel-coated membrane inserts. The subsequent procedures were identical to that of the migration assay. Images were obtained using an Olympus BX51 microscope.

4.2.5. Cell wound healing assay

NCI-H1299 cells were stimulated with 500 ng/mL GAS6 alone or co-treated with different concentrations of compound 59 or BGB324 for 24 h. Scratch healing at 0 and 24 h was observed and recorded. Images were obtained using Incucyte S3.

4.2.6. In vivo anticancer evaluations

All procedures on animals were approved by the Institutional Animal Care and Use Committee of Shanghai Institute of Materia Medica (approval No. 2021-04-DJ-59). BaF3/TEL-AXL cells at density of 5–10 × 10⁶ in 200 μL were implanted subcutaneously (sc) into the right flank of each nude mice and then allowed to grow to a well-developed tumor (700–800 mm³), which was further cut into 1.5 mm³ fragments and transplanted sc into the right flank of nude mice. When the tumor volume reached 50–150 mm³, the mice were randomly assigned into vehicle groups (n = 12) and treatment groups (n = 6). For the in vivo study, compounds 18 and 59 were formulated in 20 % (w/v) Labrosal-1% HPMC (pH 6.8) and BGB324 was formulated in 0.5 % HPMC, respectively. Vehicle groups were given vehicle only (20 % (w/v) Labrosal-1% HPMC (pH 6.8)), and treatment groups received BGB324 (50 mg/kg, bid), compounds 18 and 59 (4, 20 mg/kg, qd) as indicated doses via po administration. The sizes of tumors were measured twice per week using the microcaliper, and the tumor volume (TV) was calculated as Eq. (1):

$$\text{TV}=(\text{Length}\times {\text{Width}}^2)/2$$ (1)

The percent (%) tumor growth inhibition (TGI) value was measured on the final day of study for drug-treated compared with vehicle-treated mice and calculated as Eq. (2):

$$\text{TGI}\left(\%\right)=\left\{1–\left[\left({\text{TV}}_{\text{Treated}\text{final}\text{day}}–{\text{TV}}_{\text{Treated}\text{Day}0}\right)/({\text{TV}}_{\text{Vehicle}\text{final}\text{day}}–{\text{TV}}_{\text{Vehicle}\text{Day}0)}\right]\right\}\times 100$$ (2)

Significant differences between treated groups vs vehicle groups (P < 0.05) were determined using Student's t-test. At the end of treatment, subcutaneous BaF3/TEL-AXL tumor tissues were obtained 1 h after the last dosage. Then, intratumoral AXL phosphorylation level was determined by immunoblotting analysis.

4.3. Pharmacokinetic (PK) studies

The PK properties were determined in ICR mice (6–8 weeks, 18–22 g, n = 3 per group). All procedures on animals were performed in accordance with the established regulations and ethical guidelines, and approved by the Institutional Animal Care and Use Committee at Shanghai Institute of Materia Medica, Chinese Academy of Sciences (approval No. 2021-07-CXY-19 and 2022-01-YY-21). For the iv groups, ICR mice were intravenously administered compounds 18 (DMSO/EtOH/PEG300/0.9 % NaCl = 5/5/40/50, v/v/v/v), 24 (DMSO/Tween 80/0.9 % NaCl = 5/5/90, v/v/v) and 59 (DMSO/Tween 80/0.9 % NaCl = 10/10/80, v/v/v) at doses of 5 mg/kg, respectively. The blood samples were collected at 0.083, 0.25, 0.5, 1, 2, 4, 6, 8 and 24 h after dosing. For the po groups, compounds 18, 24 and 59 were treated with a solution of 20 % (w/v) Labrosal-1% HPMC-pH 6.8, and subsequently administered at doses of 10 mg/kg via oral gavage route, respectively. The blood samples were collected at 0.5, 1, 2, 4, 6, 8 and 24 h after dosing. All serum samples were obtained through common procedures of collected blood samples and stored at −80 °C until analysis, and the concentrations of the compounds in the supernatant were determined by LC–MS/MS.

4.4. Molecular modeling

Given 68.94 % sequence identities between AXL and MER kinase domains, we first aligned their sequences on https://blast.ncbi.nlm.nih.gov, and then confirmed the residues of ligand binding pockets in MER crystal structures (chain B of PDB ID: 7AVX and PDB ID: 7AAY) by Pymol software (Supporting Information Fig. S2). Moreover, Fig. S3 and Fig. S4 showed that three-dimensional conformation and ligand binding pockets of the constructed AXL DFG-out homology model were similar to that of MER crystal structures (chain B of PDB ID: 7AVX and PDB ID: 7AAY). Therefore, PDB ID: 7AAY and chain B of PDB ID: 7AVX were identified as templates. Finally, a homology modeling was performed in PyMod 3.0 on the basis of Modeller software⁶⁸.

The molecular dockings were performed on Glide module of Schro¨dinger software. Specifically, the constructed AXL DFG-out homology model was identified as the protein receptor, and its structure was processed by the Protein Preparation Wizard tool, including filling in missing side chains, adding hydrogens and restrained minimization. The protein grid box was generated with the center of the binding pocket by the Receptor Grid Generation tool. The ligands were processed with the LigPrep tool. Standard precision (SP) mode of Glide was applied to dock the optimized ligands into the constructed AXL DFG-out homology model with the default parameters.

Author contributions

Yu Zhou, Jing Ai, Hong Liu and Meiyu Geng conceived and designed this project. Yu Zhou and Jing Ai designed experiments and supervised data analysis. Feifei Fang, Yang Dai, Xia Peng, Jing Ai and Yu Zhou wrote the manuscript. Feifei Fang and Yang Dai analyzed the data, and together with Xia Peng, Dong Zhang and Dan Zhang drew figures and tables. Feifei Fang, Hao Wang, Yazhou Li and Chunpu Li synthesized the target compounds and key intermediates. Yang Dai, Yinchun Ji, Yangrong Zhao, Xia Peng and Danyi Wang performed in vitro and in vivo experiments of AXL inhibition and antitumor efficacy. Jiyuan Li and Chunpu Li conducted molecular docking. Feifei Fang and Xuewu Liang performed the homology modeling of AXL protein. All authors have given approval to the final version of the manuscript.

Conflicts of interest

The authors declare no competing financial interest.

Appendix A. Supplementary data

The following is the Supplementary data to this article: