Overcoming Secondary Mutations of Type II Kinase Inhibitors

Abstract

Type II kinase inhibitors bind in the “DFG-out” kinase conformation and are generally considered to be more potent and selective than type I inhibitors, which target a DFG-in conformation. Nine type II inhibitors are currently clinically approved, with more undergoing clinical development. Resistance-conferring secondary mutations emerged with the first series of type II inhibitors, most commonly at residues within the kinase activation loop and at the “gatekeeper” position. Recently, new inhibitors have been developed to overcome such mutations; however, mutations activating other pathways (and/or other targets) have subsequently emerged on occasion. Here, we systematically summarize the secondary mutations that confer resistance to type II inhibitors, the structural basis for resistance, newer inhibitors designed to overcome resistance, as well as the challenges and opportunities for the development of new inhibitors to overcome secondary kinase domain mutations.

Graphical Abstract

1. INTRODUCTION

1.1. Protein Kinase.

A kinase is a type of enzyme that transfers a phosphate group from ATP to a substrate.¹ The human kinome encodes approximately 538 kinases. Based on the chemical properties of their substrates, kinases can be categorized into protein kinases, lipid kinases, carbohydrate kinases, and others.² The primary sites for phosphorylation in the human proteome are tyrosine, serine, and threonine. Consequently, protein kinases are further classified as tyrosine kinases, serine-threonine kinases, dual-specificity kinases, histidine kinases, among others.³ Most kinases play a crucial role in regulating cellular activities, including survival, proliferation, differentiation, and migration.⁴ Therefore, over-expression or gain-of-function mutations in kinases can lead to cancer development. Moreover, dysregulation of kinases can also cause multiple physiological dysfunctions, affecting immune, neurological, and infectious activities.⁵

1.2. Protein Kinase Inhibitors Classification.

As of July 31, 2023, the FDA has approved a total of 76 small molecule kinase inhibitors. These inhibitors are categorized into five classes based on their binding modes to kinases: type I, type II, type III, type IV, and covalent inhibitors. Type I inhibitors are ATP-competitive. They bind to the ATP-binding pocket and compete with ATP in a kinase-catalyzed phosphorylation reaction (Figure 1A). On the other hand, type II inhibitors are non-ATP-competitive inhibitors. They bind to kinases and stabilize an inactive conformation, characterized by the Asp-Phe-Gly (DFG) motif pointing toward the outside of the kinase (Figure 1A and B). This state is also known as the DFG-out state.⁶ Typically, type II kinases bind to kinases at both the ATP-binding pocket and extend to an adjacent hydrophobic pocket, located behind the gatekeeper residue. Both type III and type IV inhibitors are allosteric inhibitors. Type III inhibitors bind to allosteric sites adjacent to ATP-binding sites (Figure 1A), exemplified by MEK1 inhibitors such as trametinib.⁷ Type IV inhibitors bind to allosteric pockets that are distant from ATP-binding pockets (Figure 1A), exemplified by the BCR::ABL1 inhibitor asciminib.^8,9 Finally, inhibitors that form covalent bonds with their targets are categorized as covalent inhibitors, exemplified by the BTK inhibitor ibrutinib.^10-12

Figure 1.

Illustrations of kinase and kinase inhibitors. (A) The binding modes of type I, II, III, and IV inhibitors. The inhibitors are depicted as the orange shapes. (B) The conformations of the activation loop in the active state and inactive state. The activation loops are depicted as the solid lines. The DFG motifs are depicted as three sequential green circles marked with “D”, “F”, and “G″.

Till now, the totality of reports on crystallographic studies have produced an extensive collection of cocrystal structures, showing inhibitors bound to kinases.¹³ These structures provide a convenient basis for classifying kinase inhibitors according to the structures of their drug-enzyme complexes.¹⁴ Generally, type I inhibitors bind to kinases in an active state, characterized by the DFG-motif pointing toward the interior of the kinase and aligned with the ATP binding pocket (DFG-in state). However, it is not entirely suitable to classify kinase inhibitors solely based on cocrystal structures, as some type I inhibitors have been found to bind to kinases in their DFG-out state in crystallographic studies. For example, gilteritinib is an ATP-competitive FLT3 kinase inhibitor, so it is a type I inhibitor;¹⁵ however, in a crystallographic study, it binds to FLT3 kinase in a DFG-out inactive state.¹⁶ Therefore, a more appropriate approach is to classify kinase inhibitors based on their enzymological inhibitory analysis.

1.3. Advantages of Type II Kinase Inhibitors.

The ATP binding pocket and the hydrophobic pocket within the DFG-out conformation are common features found in nearly all kinases. As a result, the majority of FDA-approved kinase inhibitors fall under type I or II categories. Conversely, type III and IV inhibitors are more empirical in nature, and covalent inhibitors heavily rely on the specific presence of target amino acid residues for covalent binding. Here, we make a comparison between type I and type II inhibitors, in terms of selectivity.

The intrinsic selectivity of type II inhibitors compared to type I inhibitors has been a subject of prolonged discussion. In essence, the rationale behind the notion that type II inhibitors possess better selectivity is supported by four key factors. 1. Evolutionary conservation of catalytic regions: The amino acid sequences and structures of catalytic regions, including the ATP-binding pocket, are highly conserved through evolution. However, the amino acid sequence surrounding the hydrophobic pocket generated in the DFG-out state is not as highly conserved. This discrepancy opens opportunities for type II inhibitors to exhibit greater selectivity by targeting the less conserved regions.^17,18 2. Influence of gatekeeper residues: Different kinases have varying gatekeeper residues, which play a crucial role in the binding of type II inhibitors.¹⁹ Smaller gatekeeper residues generally lead to fewer steric effects, making it easier for type II inhibitors to accommodate and interact with the kinase. For instance, many type II inhibitor-bound complexes feature threonine as the gatekeeper residue, while phenylalanine is the only large amino acid observed as a gatekeeper in such complexes.²⁰ 3. Kinase-specific tendencies: Various kinases have distinct propensities to adopt the DFG-out inactive conformation. Some kinases easily form this inactive state, while others can only do so with difficulty (e.g., thermodynamic penalty). As a result, the energy states for the DFG-out conformation and type II inhibitor binding are specific to each kinase.¹⁹ This uniqueness can explain the selectivity of certain inhibitors, such as imatinib, which exhibits specificity toward ABL1 kinase over SRC.^21,22 4. Nonclassical DFG-out inactive conformation: Researchers, including R. S. K. Vijayan et al, have observed that some kinase structures in structural databases adopt a “nonclassical” DFG-out inactive conformation. In these cases, the hydrophobic pocket may be either small or blocked, limiting its ability to accommodate type II inhibitors. This structural variability can impact the selectivity of type II inhibitors toward specific kinases.²³ However, there is an opposing viewpoint regarding the intrinsic selectivity of type II inhibitors compared to type I inhibitors, which arises from studies conducted byN.S. Gray’s group. In their research, they screened 36 type II inhibitors against a comprehensive panel of more than 400 kinases and found that some of these inhibitors exhibited nonselective behavior.²⁴ On the other hand, a separate study carried out by J.R. Peterson, R.L. Dunbrack, and R.M. Levy’s groups argued that, with only a few exceptions, most type II inhibitors demonstrated significantly better selectivity than type I inhibitors.²³ In practice, several clinically approved type II inhibitors, such as sorafenib and regorafenib, are considered nonselective as they target multiple kinases. Conversely, type II inhibitors like nilotinib, dabrafenib, and quizartinib are renowned for their high selectivity. From our perspective, the structural variability of the kinase DFG-out conformation, which tends to be more diverse than the active conformation, allows for greater potential variation in interactions between type II inhibitors and different kinases. As a result, although in some worst cases type II kinase inhibitors might not be selective, type II inhibitors have promising potential for high selectivity over that of type I inhibitors.

1.4. Clinically Approved Type II Kinase Inhibitors and Their Targets.

To date, the FDA has approved nine type II kinase inhibitors (Figure 2, and Table 1). These nine type II kinase inhibitors typically employ two primary mechanisms of action for cancer therapy: 1) inhibiting kinases that regulate angiogenesis, and 2) inhibiting kinases that have gain-of-function mutations.

Figure 2.

Structures of clinically approved type II kinase inhibitors.

Table 1.

List of Clinically Approved Type II Kinase Inhibitors

Name	Chemical name	Year approved	Target kinase	Target disease
Imatinib	STI571	2001	BCR::ABL1	Hypereosinophilic syndrome,31 Philadelphia chromosome-positive CML or ALL,32 aggressive systemic mastocytosis,33 chronic eosinophilic leukemia,34 dermatofibrosarcoma protuberans,35 gastrointestinal stromal tumors,36 myelodysplastic/myeloproliferative disease37
Sorafenib	BAY 43-9006	2005	VEGFR1/2/3	Hepatocellular carcinoma,38 renal cell carcinoma,39 thyroid cancer (differentiated)40
Nilotinib	AMN-107	2007	BCR::ABL1	Philadelphia chromosome-positive CML41
Regorafenib	BAY 73-4506	2012	VEGFR1/2/3	Colorectal cancer,42 hepatocellular carcinoma43
Ponatinib	AP24534	2012	BCR::ABL1	Philadelphia chromosome-positive CML or ALL44
Pexidartinib	PLX3397	2019	CSF1R	Tenosynovial giant cell tumors45
Ripretinib	DCC-2618	2020	KIT/PDGFR	GIST46
Tivozanib	AV-951; KRN951	2021	VEGFR2	RCC 47
Quizartinib	AC220	2023	FLT3	FLT3-ITD mutation-positive AML48

Sorafenib, regorafenib, and tivozanib are generally considered to impede tumor progression mainly by inhibiting the tyrosine kinase targets that are associated with angiogenesis functions such as VEGFR 1, 2, and 3, but each has distinct minor targets. Sorafenib inhibits VEGFR1/2/3, and also inhibits RAF-1, B-RAF, PDGFR-β, FLT3, and KIT.²⁵ Regorafenib is a derivative of sorafenib, with an additional fluorine attachment on the central phenyl ring of sorafenib (Figure 2). Interestingly, the attachment of this fluorine induced surprisingly increased the inhibitory effects of the chemical structure on multiple kinases.²⁶ Tivozanib is more potent against VEGFR2 than VEGFR1 and 3. It also potently inhibits EPHB2, PDGFR-α, PDGFR-β, KIT, and TIE2.^27,28 Angiogenesis inhibitors produce clinical benefits in a variety of cancer types but followed by resistance mediated by multiple adaptive mechanisms.^29,30 Unlike inhibitors that target kinases harboring gain-of-function mutations, inhibitors that target angiogenesis rarely encounter on-target resistant mutations.

In contrast, other inhibitors impede the tumor progression by inhibiting the kinases harboring gain-of-function mutations. Notably, imatinib, nilotinib, and ponatinib are BCR::ABL1 inhibitors, specifically designed to target the BCR::ABL1 fusion protein. Pexidartinib, ripretinib, and quizartinib, on the other hand, are inhibitors of CSF1R, KIT/PDGFR, and/or FLT3. While these inhibitors have shown significant clinical benefits for cancer therapy initially, they often encounter resistance over time. The development of resistance mechanisms for these inhibitors includes two main factors: on-target secondary mutations and off-target activation of downstream or alternative pathways. Here, we give a review focusing on the development of BCR::ABL1 and FLT3 inhibitors, and provide a perspective on ways to overcome secondary mutations of type II inhibitors.

2. BCR::ABL1 INHIBITORS

2.1. Imatinib and Its Secondary Mutations.

BCR::ABL1 arises as a consequence of a reciprocal chromosome translocation between chromosome 9 and 22 - t(9, 22), resulting in the fusion of the BCR and ABL1 genes.⁴⁹ The BCR::ABL1 gene fusion causes unregulated ABL1 tyrosine kinase activity, and is found in >95% of chronic myeloid leukemia (CML),^50,51 20—30% of acute lymphoblastic leukemia (ALL),⁵² and a minority of primary AML cases. The first FDA-approved BCR::ABL1 inhibitor was imatinib.³² It is also the first type II inhibitor to be identified. As shown in a crystal structure, imatinib binds to the hydrophobic pocket, gatekeeper area and ATP binding pocket of the ABL1 kinase domain in the DFG-out inactive conformation⁵³ (Figure 3A). Besides BCR::ABL1, imatinib also inhibits KIT and PDGFR, with IC₅₀s against ABL1, KIT and PDGFR-β of 188 nM, 413 nM and 386 nM, respectively.^54,55 During the clinical application of imatinib for treating CML, resistance occurred due to multiple secondary point mutations, including at least 25 residues in the BCR::ABL1 kinase domain.^56,57 These point mutations render BCR::ABL1 insensitive to imatinib. Other genetic mechanisms for drug resistance include amplification, which has been largely overcome by the increased potency of the second generation BCR::ABL1 inhibitors.

Figure 3.

Binding of ABL1 kinase domain with imatinib. (A) Crystal structure of ABL1 kinase domain in complex with imatinib (PDB: 1IEP). (B) The interaction of imatinib with gatekeeper T315 (PDB: 1IEP). (C) Imatinib with gatekeeper mutation T315I. (D) The amino acid residues M351T, E355G, and F359 V at kinase domain (PDB: 1IEP). (E) The amino acid residues on the G-loop (PDB: 1IEP). (F) The amino acid residue H396 at the A-loop (PDB: 1IEP). (G) The amino acid residue V289 at the c-helix (PDB: 1IEP). The ribbons representing G-loop, A-loop, and c-helix are colored in blue, red, and yellow, respectively. The hydrogen bonds, salt bridge, π–π stacking interaction, and steric clashes are generated by Maestro 11.8 according to the structures and depicted as dashed lines colored in yellow, purple, blue, and red, respectively. The structure exhibited in (C) was generated by Maestro 11.8 with the “single point mutation” module (T315 to I315) from PDB 1IEP.

Imatinib-resistant point mutation hotspots include residues that directly bind to imatinib, as well as ones that do not. These critical residues are situated within four essential structural regions: the contact site, C-lobe, G-loop (also referred to as the P-loop), and A-loop.⁵⁸

Gatekeeper.

The gatekeeper of a kinase is the first residue of the hinge region, connecting the N- and C-terminal lobes. It regulates inhibitor access to the hydrophobic pocket in the back cleft by altering the size and shape of the ATP-binding pocket. Gatekeeper T315I in BCR::ABL1 is the most common secondary mutation that arises after imatinib treatment, and is highly cross-resistant to dasatinib, nilotinib and bosutinib. T315 is a critical contact site for these inhibitors. T315I confers imatinib resistance by two mechanisms: 1) elimination of the hydrogen bond from between imatinib amino-pyridine and the hydroxyl group of T315; 2) introduction of steric clash on site due to the replacement of a bulky isoleucine residue (Figure 3B and C). In addition, another mutation hotspot is at F317, also located at the contact site.

C-Lobe.

The mutation hotspots that occur in the C-lobe involve M351T, E355G, and F359V. These secondary mutations have no direct interactions with imatinib but can affect ABL1 kinase conformations. For example, M351T is the second most common secondary mutation, which allosterically changes the conformation of the hydrophobic pocket that imatinib binds, therefore conferring imatinib resistance⁵⁹ (Figure 3D).

G-Loop.

G-loop (glycine-rich loop) is also named P-loop (phosphorylation loop). It is characterized to have a conserved amino acid sequence GXGXXG, where G represents glycine, and X for any amino acid. Mutations at variable positions within the G-loop confer resistance by stabilizing the active conformation and destabilizing the inactive conformation.⁶⁰ Secondary mutation hotspots L248, G250, Q252, Y253 and E255 are located in the G-loop. Of particular significance, E255 forms a salt bridge with K247, contributing to the stabilization of the G-loop conformation. Mutations at this position can increase the instability of the G-loop, leading to reduced binding of imatinib and ultimately resulting in resistance.⁶¹ Additionally, Y253 is involved in a π–π stacking interaction with imatinib. This interaction is crucial for imatinib’s effective binding to the ATP-binding pocket of BCR::ABL1. Mutations at Y253 can disrupt this π—π stacking interaction, further contributing to resistance (Figure 3E).

A-Loop.

Mutation hotspot H396 is located at the A-loop (activation loop). Mutations at this location suppress the kinase tendency to adopt the DFG-out inactive conformation, making ABL1 kinase to adopt the active conformation more easily. Since type-II inhibitors only bind to kinases with DFG-out conformation, BCR::ABL1 with H396P/R secondary mutation confers imatinib resistance (Figure 3F).

C-Helix.

In addition of above-mentioned four locations, V289S/I mutation at C-helix also confers imatinib resistance by destabilizing the inactive DFG-out conformation⁶²·⁶³ (Figure 3G). However, secondary mutations at this location rarely occur clinically.

2.2. Next Generation BCR::ABL1 Inhibitors.

Second Generation BCR::ABL1 Inhibitors.

Dasatinib, nilotinib and bosutinib are second generation BCR::ABL1 inhibitors. They were approved by the FDA in 2006, 2007, and 2012, respectively (Figure 4A). Interestingly, dasatinib and nilotinib are good alternatives for each other (with the exception of T315I) in the treatment of imatinib-resistant CMLs with different secondary mutations, because of their largely distinct interactions with BCR::ABL1.⁶⁴ Dasatinib is a type I inhibitor, and it exhibits a different inhibitory profile on BCR::ABL1 secondary mutations than imatinib⁶⁵ (Figure 4B). Since dasatinib can bind to ABL1 in both active and inactive conformations, it overcomes imatinib resistance caused by secondary mutations at G-loop, A-loop, and C-helix.⁶⁶ In addition, because it only binds to the ATP-binding pocket and gatekeeper, it can also overcome secondary mutations located at the kinase domain, particularly those influencing the conformation of the hydrophobic pocket. However, dasatinib is still vulnerable to contact site mutations V299L, F317L/V and T315I/A. On the contrary, nilotinib is a type II inhibitor and structural derivative of imatinib (Figure 4C). Nilotinib shares a similar binding mode as imatinib but exhibits greater potency. Nilotinib demonstrates efficacy in treating multiple secondary mutants, including H396R at the A-loop, F317L at the ATP-binding pocket, and M351T at the kinase domain, although it exhibits a slightly reduced inhibitory effect when compared to BCR::ABL1 primary mutations.⁵⁹ However, select mutations at the G-loop, F359V at the kinase domain, and T315I at the gatekeeper, which confer the greatest degree of resistance to imatinib, also render BCR::ABL1 insensitive to nilotinib. In summary, dasatinib and nilotinib are good alternatives and can overcome most imatinib-resistant secondary mutations; but neither of them retains inhibitory effects on T315I. Besides ABL1, bosutinib also potently inhibits SRC⁶⁷ and shares mutational vulnerabilities with dasatinib. It also overcomes several imatinib-resistant secondary mutations but not T315I.^68,69

Figure 4.

Binding of BCR::ABL1 inhibitors with ABL1 kinase domain. (A) Chemical structures of BCR::ABL1 inhibitors. (B) Crystal structure of dasatinib bound to activated ABL1 kinase domain (PDB: 2GQG). (C) Human ABL1 kinase in complex with nilotinib (PDB: 3CS9). (D) Crystal structure of ponatinib in complex with the ABL1-T315I mutant kinase (PDB: 3IK3). (E) ABL1 kinase domain in complex with asciminib (PDB: 5MO4). The ribbons representing G-loop, A-loop, and c-helix are colored in blue, red, and yellow, respectively. The hydrogen bonds are depicted as dashed lines in yellow.

Third Generation BCR::ABL1 Inhibitors.

Ponatinib is a third generation type II BCR::ABL1 inhibitor approved by the FDA in 2012 (Figure 4A).⁷⁰ In order to overcome T315I resistance, a less sterically demanding alkyne linker was incorporated into ponatinib to avoid the steric clash between the larger gatekeeper isoleucine and the amino group on the pyrimidine ring in imatinib and nilotinib.⁷¹ The alkyne linker is also engaged in hydrophobic interactions with I315 (Figure 4D). Ponatinib also potently inhibits several other kinases such as FLT3, KIT and FGFR.^72,73 However, in the clinical use of ponatinib, concerning side effects were also observed, most notably involving the cardiovascular system.⁷⁴ Asciminib is a first in class allosteric BCR::ABL1 inhibitor approved by the FDA in 2021 (Figure 4A). Distinct from all other approved BCR::ABL1 inhibitors, it is a type-IV inhibitor, binding to an allosteric pocket of ABL1 kinase distant from the ATP-binding pocket, named the myristoyl binding pocket⁸ (Figure 4E). Therefore, asciminib is predicted to retain potency against imatinib-resistant secondary mutations at the G-loop, A-loop, gatekeeper and kinase domain.⁹ Secondary mutations in the myristoyl pocket such as V468F, A337V, and P465S are insensitive to asciminib, but are all sensitive to imatinib, dasatinib, nilotinib, bosutinib, and ponatinib. We we recently published data demonstrating that several N-lobe mutations, including in the P-loop unexpectedly confer resistance to asciminib.⁷⁵

3. FLT3 INHIBITORS

3.1. Quizartinib and Its Secondary Mutations.

FMS-like tyrosine kinase 3 (FLT3) belongs to the tyrosine kinase superfamily and is mainly expressed in hematopoietic cells. Activating mutations in FLT3 occur in approximately 30% of acute myeloid leukemia (AML) cases, primarily internal tandem duplications in the juxtamembrane domain (FLT3-ITD) and less commonly point mutations in the tyrosine kinase domain (FLT3-TKD). The first FDA-approved FLT3 inhibitor midostaurin demonstrated minimal activity as monotherapy, and is considered nonselective. Midostaurin is only approved for use in combination with chemotherapy agents. ^76,77 Quizartinib was described in 2009 as a FLT3 inhibitor with a K_d value of 1.1 nM, and clinically approved in Japan in 2019 and by the FDA in 2023 (Figure 5A).^78-80 However, in clinical applications, resistance rapidly arose due to secondary mutations including D835Y and F691L.²⁰

Figure 5.

Binding of FLT3 inhibitors with FLT3 kinase domain. (A) Chemical structures of FLT3 inhibitors. (B) Crystal structure of the FLT3 kinase domain bound to quizartinib (PDB: 4XUF). (C) Crystal structure of FLT3 in complex with gilteritinib (PDB: 6JQR). (D) Predicted binding of pexidartinib with FLT3, generated with PDB: 4XUF. (E) Crystal structure of the FLT3 kinase domain bound to the inhibitor FF-10101 (PDB: 5X02). The ribbons representing G-loop, A-loop, and c-helix are colored in blue, red, and yellow, respectively.

As a typical type II inhibitor, quizartinib binds to FLT3 in the DFG-out conformation⁸¹ (Figure 5B). D835 is located at the activation loop. Importantly, D835Y is not only a secondary mutation, but also occurs in primary AML as one of FLT3-TKD activating mutations. As demonstrated in the crystal structure of quizartinib bound to FLT3, D835 does not appear to have direct interactions with quizartinib. However, D835 stabilizes the inactive DFG-out conformation by forming a hydrogen bond with S838.^20,82 Therefore, mutations at the D835 residue mediate FLT3 gain-of-function activation and quizartinib resistance by affecting the kinase conformational balance between DFG-in and DFG-out. F691 is the gatekeeper residue of FLT3, which stabilizes quizartinib binding with FLT3 by conferring a π–π stacking interaction.⁸¹ The F691L mutation eliminates the π–π stacking interaction and the bulky leucine residue also induces steric clash on site.

3.2. Other FLT3 Inhibitors.

Gilteritinib.

Gilteritinib is a type I FLT3 inhibitor approved in 2018 by the FDA (Figure 5A).^83,84 It retains inhibitory effects on FLT3 with D835 mutations (Figure 5C).¹⁶ As illustrated in the binding prediction, gilteritinib has hydrophobic interactions with FLT3 at F691 site.¹⁵ The F691L mutated FLT3 is less sensitive to gilteritinib due to the reduced binding interactions and increased steric clash.

Pexidartinib.

Currently, there remains an unmet clinical need to overcome the secondary F691L mutation of FLT3. However, promising advancements are being made with the development of other FLT3 inhibitors that show notable inhibitory efficacy against FLT3 with the F691L mutation. One such inhibitor is pexidartinib, classified as a triple kinase inhibitor with IC₅₀ values of 13 nM, 27 nM, and 11 nM on CSF1R, KIT, and FLT3, respectively (Figure 5A). In its inferred binding pose with FLT3 kinase, the central pyridine ring faces the amino acid F691L. Notably, the urea NH group, which is proximal to the phenyl ring in quizartinib, is not compatible with F691L. However, pexidartinib overcomes this limitation by utilizing a shorter linker between the middle and tail ring⁸⁵ (Figure 5D). Because pexidartinib is a type II inhibitor that binds to the DFG-out conformation of FLT3, it is vulnerable to mutations and D835 and other residues in the activation loop.

FF-10101.

In 2018, an irreversible FLT3 inhibitor FF-10101 was reported to bind covalently to a remote cysteine residue C695 (Figure 5A).⁸⁶ This unique binding mode enables FF- 10101 to retain inhibitory effects on FLT3 with secondary mutations F691L, with GI₅₀s of 1.9 nM and 10 nM on 32D-transfectant cell lines with FLT3-ITD and FLT3-ITD-F691L (Figure 5E).

XW-24.

Our previous work also identified an imidazo[1,2-α]pyridine-pyridine derivative (XW-24) which equally inhibited FLT3 and FLT3 with secondary mutations D835Y and F691L. As shown from the molecular modeling analysis, XW-24 retains inhibitory effects on FLT3 with F691L mutation.⁸⁷

3.3. Off-Target Resistance through Activation of Downstream Pathways.

Besides on-target FLT3 secondary mutations, off-target secondary clinical resistance has been observed in patients after treatment with FLT3 kinase inhibitors, mainly due to mutations that activate the PI3K/AKT and/or RaS/MAPK pathways.^88,89 Alternative activation of BCR::ABL1 pathway is also detected in 5% of relapsed FLT3 mutated patients.^89-92 Therefore, FLT3/other pathway dual inhibitors may be of future interest. Our recent work identified a FLT3/BCR::ABL1 dual pathway inhibitor.⁹³

4. KIT INHIBITORS

Activating mutations of KIT kinase occur in 70% - 80% of gastrointestinal stromal tumors (GISTs). Before ripretinib was developed, imatinib (Figure 6A) and other two multitarget kinase inhibitors, sunitinib and regorafenib, were approved by the FDA for the treatment of GIST. However, these three inhibitors can only overcome a limited number of KIT activating mutations. Sunitinib and regorafenib are largely complementary to each other, covering multiple imatinib-resistant KIT secondary mutations, including T670I at gatekeeper, V654A at ATp-binding pocket, and D820A/E/ G/Y, N822H/K, etc. at A-loop; however, secondary mutations at D816 residue in the A-loop are problematic for both of these agents.⁹⁴ Secondary mutations at D816 residue confer resistance to imatinib, sunitinib, and regorafenib by destabilizing the DFG-out inactive conformation, thus changing the equilibrium between the active state and inactive state. In 2019, ripretinib (DCC-2618) was reported to be able to potently inhibit KIT and a series of secondary mutations, including mutations at D816. Ripretinib was reported to bind in a type-II manner. It forms hydrogen bonds with K623, E640, and D810, and has hydrophobic interactions with multiple amino acid residues at both ATP-binding pocket and the adjacent hydrophobic pocket. Ripretinib inhibits KIT kinase by both antagonizing its DFG-in active state and stabilizing its DFG-out inactive state.⁹⁵ Most importantly, ripretinib occupies a crucial part of the hydrophobic pocket, so that F811 of the DFG-motif cannot be accommodated, therefore the KIT DFG-in active state is antagonized (Figure 6B).

Figure 6.

Binding of KIT inhibitors with KIT kinase domain. (A) Crystal structure of KIT in complex with imatinib (PDB: 1T46). (B) Chemical structure of ripretinib. (C) Predicted binding of ripretinib with KIT, generated with PDB: 6MOB.

5. CONCLUSION AND PERSPECTIVE

Compared to type I inhibitors, type II kinase inhibitors may have greater potential for inhibitory potency and selectivity but are more prone to resistance due to the occurrence of secondary mutations. Numerous efforts have been made to develop new inhibitors to overcome these mutations as discussed above. Here, we summarize and offer our perspective for ways to develop novel inhibitors that can overcome secondary mutations.

On-target secondary mutations are located in various regions of kinases. Although there are rare cases where mutations increase ATP affinity (discussed below) most can be simply categorized into two types - mutations of the residues that directly interact with inhibitors and mutations of the residues that indirectly interact with inhibitors.

Type II inhibitors bind to kinases in the ATP-binding pocket and gatekeeper and extend into the hydrophobic back pocket. Secondary mutations located in these regions usually interact with inhibitors directly. Inhibitor binding to the hydrophobic pocket is generally stabilized by hydrophobic interactions, and secondary mutations are rarely observed in this pocket. The interactions between the residues in the ATP-binding pocket and type-II inhibitors may include hydrophobic interactions, π–π stackings, and hydrogen bonds. Although secondary mutations at this position are often observed, they are usually case-by-case among kinase inhibitors and can be overcome by medicinal chemistry modifications. The interaction between the gatekeeper residue and type II inhibitors varies with different amino acid residues, usually with a hydrogen bond for threonine and a π–π stacking for phenylalanine. Secondary mutations at this position typically lead to the disruption of previous interactions with inhibitors and an increase in steric clash on-site. Notably, secondary mutations at gatekeeper residues are frequently observed in clinical use of type II inhibitors, mediating resistance that is usually difficult to overcome. The current strategies to tackle secondary mutations at the gatekeeper include: 1) avoiding steric clash, like ponatinib,^70-74 and 2) developing inhibitors with new binding modes, like asciminib^8,9 and FF-10101.⁸⁶

Secondary mutations of residues that do not directly interact with type II inhibitors are usually located at structurally important regions, including G-loop, C-helix, A-loop, and kinase domain. Mutations in the G-loop and C-helix usually cause decreased structural stability of the kinase DFG-out inactive conformation, therefore mediating type-II inhibitor resistance. Moreover, the A-loop is directly involved in the kinase conformational change. Therefore, secondary mutations in the G-loop, C-helix, and A-loop are considered mechanistically less sensitive to type II inhibitors. They usually make kinases more likely to adopt a DFG-in active conformation, thereby impairing type II inhibitor binding. In this case, developing new inhibitors with distinct binding modes is attractive. However, although reduced sensitivity has been observed in clinical use of some type II inhibitors with several secondary mutations in the G-loop, the same type II inhibitors may still retain effective inhibitory potency and can still be used to treat these secondary mutations. Moreover, the success of ripretinib demonstrated that even secondary mutations at the A-loop can be overcome by rational design of type II inhibitors.⁹⁵ Therefore, secondary mutations in the G-loop, C-helix, and A-loop do not necessarily require the development of inhibitors with different binding modes.

In addition, secondary mutations in the kinase domain may also confer resistance to type II inhibitors by changing the conformation of the hydrophobic pocket, such as M351T and F359V with BCR::ABL1. These secondary mutations can also be addressed through structural modifications of type II inhibitors by medicinal chemistry methods.

Numerous studies have reported on the frequency of secondary mutations following treatment with type II inhibitors. Although the rankings of mutation frequencies may vary slightly, the gatekeeper mutation T315I consistently emerges as the most frequent mutation after imatinib treatment.^96-98 For quizartinib, mutations associated with the D835 residue in the A-loop are prevalent.^82,99 Notably, mutations at these two locations pose significant challenges for type II inhibitors to overcome and require specific efforts from medicinal chemists.^63,100,101

From our perspective, unlocking the high potential of type II inhibitors requires more efforts to overcome secondary resistance associated with them. Rational design of new type II inhibitors will certainly contribute to this aim. Regarding the gatekeeper residue, the successful example of ponatinib has shown that reducing the steric size of the fragments in proximity to the gatekeeper residue helps retain activity against secondary mutations at that position. Another strategy worth exploring may involve increasing the interactions between the fragment and the mutated gatekeeper. For instance, many type II inhibitors feature either an amide group or a urea group as a connective fragment between the ATP pocket-binding fragments and the hydrophobic pocket-binding fragments. However, this connective group may be incompatible with some of the hydrophobic residues. In such cases, inspired by the design of pexidartinib in comparison with quizartinib, a methylene as the connective group may be more realistic. In addition, this kind of design also helps to reduce steric clash.

Regarding the A-loop, overcoming secondary mutations at locations that destabilize the DFG-out conformation is undoubtedly a challenging task. However, the success of ripretinib offers a potential solution to address secondary mutations at the A-loop - modifications of the hydrophobic pocket-binding fragments. The hydrophobic pocket appears to be less prone to mutations. In a kinase, the hydrophobic pocket’s function is to accommodate the phenylalanine residue of the DFG motif, stabilizing the DFG-in active conformation. If mutations occur at the hydrophobic pocket, the overall kinase activity may be affected. This might explain the rarity of mutations at the hydrophobic pocket. The hydrophobic pocket-binding fragments of type II inhibitors compete with the phenylalanine residue of the DFG motif to bind to the hydrophobic pocket. If the binding of type II inhibitors with the hydrophobic pocket is sufficiently strong, the formation of the DFG-in conformation will be hindered. Therefore, to overcome mutations at the A-loop, it seems reasonable to explore modifications to type II inhibitors that increase their binding affinity with the hydrophobic pocket.

It is interesting to note that in some studies, the drug-target residence time parameter is suggested to complement classical drug affinity parameters (K_I, K_d, IC₅₀, or EC₅₀) and is believed to influences in vivo efficacy.^102-104 However, as of now, there is no evidence indicating a direct relationship between the length of drug-target residence time and the development of secondary mutations. For instance, covalent inhibitors, which typically have an increased drug-target residence time, do not appear to either trigger more secondary mutations or suppress the emergence of new secondary mutations.

In addition to the above-mentioned secondary mutations, various other secondary mutation types have been observed in clinical practice. For instance, EGFR gatekeeper mutation T790 M is the most frequent secondary mutation associated with EGFR.¹⁰⁵ In addition to disrupting interactions with inhibitors and increasing steric clash, which is common with the secondary mutations at the gatekeeper residues, EGFR T790 M also enhances ATP affinity.^106,107 Therefore, this type of secondary mutation has the most pronounced negative impact on the inhibitory potency of type I inhibitors. In such cases, the development of inhibitors with new binding modes may offer a solution to overcome these secondary mutations.¹⁰⁸ For example, EFGR covalent inhibitor osimertinib has proven effective in overcoming T790 M mutation.^109-112

Lastly, after further inhibition of on-target secondary mutations, off-target mutations sometimes can occur as a result, which activate downstream or parallel pathways. In some cases, off-target activating mutations also occur as a direct result after the primary use of kinase inhibitors, which confers resistance. This issue adds considerable complexity to the problem and may be addressed by the development of dual inhibitors or the combinational use of inhibitors.

SIGNIFICANCE.

Type II kinase inhibitors comprise a significant subset of clinically approved small molecule kinase inhibitors; however, the activity of many type II inhibitors is compromised due to secondary mutations. This study reviews secondary mutations that have emerged for type II inhibitors, current methods to overcome them, and provides perspectives on possible future strategies for overcoming secondary mutations.

ABBREVIATIONS USED

ABL

Abelson murine leukemia

ALL

acute lymphoblastic leukemia

A-loop

activation loop

ATP

adenosine triphosphate

BCR

breakpoint cluster region

CML

chronic myelogenous leukemia

DFG

aspartic acid–phenylalanine–glycine

FDA

U.S. Food and Drug Administration

FLT3

FMS-like tyrosine kinase 3

FMS

feline McDonough sarcoma

GIST

gastrointestinal stromal tumor

G-loop

glycine-rich loop

ITD

internal tandem duplication

KIT

receptor tyrosine kinase

MAPK

mitogen-activated protein kinase

NSCLC

nonsmall cell lung cancer

PDGFR

platelet-derived growth factor receptor

P-loop

phosphorylation loop

Raf

rapidly accelerated fibrosarcoma

RCC

renal cell carcinoma

RET

rearranged during transfection

Src

src-family kinase

TKD

tyrosine kinase domain

VEGFR

vascular endothelial growth factor receptor

Biographies

Xiuqi Wang is a senior Ph.D. student at University of Arkansas for Medical Sciences, USA. He received his B.S. in Pharmacy (Basic Pharmacy) at China Pharmaceutical University in Nanjing. His research interests include the discovery and optimization of novel inhibitors for clinically relevant targets.

Rosa Anna DeFilippis is an Associate Professional Researcher in the laboratory of Neil Shah in the Department of Hematology/Oncology at the University of California, San Francisco. She received her PhD degree from Yale University and performed her postdoctoral work in the laboratory of Thea Tlsty at the University of California, San Francisco. Her research interests include identifying and exploiting novel vulnerabilities in myeloid malignancies and determining mechanisms of resistance to targeted therapies.

Wei Yan is a research assistant professor in School of Medicine, the University of Texas Health San Antonio. He received his Ph.D. from East China University of Science and Technology, Shanghai, China. Following, he worked at WuXi AppTec as a process chemistry scientist before he started his postdoctoral research at the University of Arkansas for Medical Sciences. His research interest lies in the exploration and advancement of innovative chemical compounds to effectively address human cancers.

Neil P. Shah is the Edward S. Ageno Distinguished Professor of Hematology/Oncology at the University of California, San Francisco. He received his MD and PhD degrees from the University of California, Los Angeles. He performed his postdoctoral work in the laboratory of Charles Sawyers. His research interests include identifying vulnerabilities in myeloid malignancies and identifying mechanisms of resistance for myeloid malignancies to targeted therapeutics and next generation compounds with the potential to override resistance.

Hong-Yu Li is a distinguished professor in Medicinal chemistry in School of Medicine, the University of Texas Health San Antonio. He is also a professor in Medicinal Chemistry in School of Pharmacy, the University of Texas Austin. He received his Ph.D. degree from the University of Tokyo and did postdoctoral training at Columbia University and Harvard University. He previously worked at Eli Lilly, the University of Arizona, and University of Arkansas for Medical Sciences. His current research interests are in chemical biology and drug discovery, especially for oncology related targets and phenotypes. In his lab, a robust oncology pipeline is under development.