Mechanisms of drug resistance in kinases

Abstract

Introduction: because of their important roles in disease and excellent 'druggability', kinases have become the second largest drug target family. The great success of the BCR-ABL inhibitor imatinib in treating chronic myelogenous leukemia illustrates the high potential of kinase inhibitor (KI) therapeutics, but also unveils a major limitation: the development of drug resistance. This is a significant concern as KIs reach large patient populations for an expanding array of indications.

Areas covered: we provide an up-to-date understanding of the mechanisms through which KIs function and through which cells can become KI-resistant. We review current and future approaches to overcome KI resistance, focusing on currently approved KIs and KIs in clinical trials. We then discuss approaches to improve KI efficacy and overcome drug resistance and novel approaches to develop less drug resistance-prone KI therapeutics.

Expert opinion: although drug resistance is a concern for current KI therapeutics, recent progress in our understanding of the underlying mechanisms and promising technological advances may overcome this limitation and provide powerful new therapeutics.

Fig. 1. Conformational changes mediating Src-family protein tyrosine kinase (SFK) activation

(A) SFK primary structure depicting conserved domains. Green, myristoyl- or farnesyl-conjugated N-terminus. SH3, SH2, KD, Src-homology 3, 2 or KD, respectively. (B-G) Schematic depictions (B,C) and crystal structures (D-G) of the inactive (B,D,F) or active (C,E,G) SFK tertiary structures. The crystal structures shown are hHck-SH2-SH3-PP1 (Pdb accession number 1QCF, D,F), hc-Src-des-methyl-Imatinib (1Y57, E) and hLck-Furanopyrimidine (2OF2, G). (F,G) Kinase domains only. Highlighted in (B-G) and annotated in (B,C) are typical characteristics of the inactive and active SFK conformations, respectively. All structures were rendered and colored in Swiss-PdbViewer (www.expasy.org/spdbv). Domains and interdomain linker regions are indicated and color-coded. Bordeaux, SH3 domain; black, SH3-SH2 interdomain linker; orange, SH2 domain; gray, SH2-KD linker, light blue, KD N-lobe with αC helix (yellow) and G-loop (pink); dark blue, C-lobe with activation (A)-loop (brown); salmon, C-terminal tail (C-Tail). Also indicated are key amino acid (AA) side-chains involved in catalysis, or whose orientation differs markedly among the different conformations in Src or ABL family kinases. Red, D and F of the A-loop DFG motif, D/E_αC within the αC helix which forms a salt-bridge with conserved K_β (green) in N-lobe β-sheet 3 in active SFKs, Y_A in the A-loop which is auto-phosphorylated into YP (red sphere in C) in active kinases, Y_C in the C-terminus which is phosphorylated into YP (red sphere in B) by Csk and binds to the SH2-domain in inactive SFKs. Also shown is A-loop K_A (green) which may form a salt-bridge with D/E_αC in the “αC-out” conformation of inactive SFKs (B,D,F) and of ABL in the SFK-like inactive structure (Fig. 2H). Cyan, bound ATP-competitive inhibitor.

Fig. 2. Conformational changes mediating ABL/Arg-family protein tyrosine kinase activation

(A) ABL and oncogenic BCR-ABL primary structures depicting conserved domains. Green, myristoyl-conjugated N-terminus only present in ABL1b (ABL-1b) due to alternative splicing. SH3, SH2, KD, Src-homology 3, 2 or KD, respectively. (B,C) Schemes (B) and crystal structures (C) of myristate-bound, autoinhibited full length hAbl-PD166326 (Pdb accession number 2FO0, left), or hAbl-P16 (1OPL chain B, right), which illustrates potential N-lobe/SH2 domain interactions in active ABL. Domains and features are indicated and color-coded as in fig. 1. Cayenne, N-terminal cap region (Cap) folding back over and interacting with the SH2 domain through a phosphorylated serine. Green, myristate (Myr) binding site including myristate moiety and involving a unique αI helix. The region between Cap and Myristate is disordered and harbors a deletion. (D-I) Schematic depictions (D-F) and crystal structures (G-I) of the following complexes: (D,G) hABL-Imatinib (1IEP) in the type 2 inhibitor-binding ABL-type inactive conformation. (E,H) hABL-ATPγS-substrate peptide (2G1T) in the type 1 inhibitor-binding competent SFK-like inactive conformation. (F,I) hABLVX-680 (2F4J) in the type 1 inhibitor-bound active conformation. The active conformation most likely results from synergy between a H396P mutation, which destabilizes the ABL-type inactive conformation, and binding of VX-680, which favors the active conformation through hydrogen-bonding and steric effects. VX-680 binds Abl in a mode that accommodates the T315I “gatekeeper” mutation. Cyan, bound ATP-analog. Bordeaux stick model, substrate peptide in (H). Key characteristics of each conformation and other kinases for which the respective conformation has been reported are summarized underneath the respective structures.

Fig. 3. Types and structural features of small-molecule inhibitor binding sites in ABL/Arg-family protein tyrosine kinases

Shown are (A-C) the crystal structures and (D-F) schemes of the compound-bound ATP-binding sites of (A,D) the type 2 inhibitor imatinib-bound human (h) ABL kinase domain (KD, 1IEP), (B,E) the ATPγS/substrate peptide bound hABL KD in the SFK-like inactive conformation (2F1T) and (C,F) the type 1 inhibitor PD166326-bound hABL KD in the active conformation (1OPL). Domains and structural features are color-coded as in fig. 1,2. Colored spheres highlight the positions of the following key sites involved in inhibitor-interactions: Hydrophobic pocket 1 (pink) or 2 (green), ATP-adenine binding region (yellow), ATP-ribose binding region (violet), ATP-triphosphate binding region (blue), type 2/3 allosteric site (gray) and the myristate-binding region as an example for a type 4 allosteric site remote from the ATP-binding region (orange). Also indicated are the hinge region, which forms conserved hydrogen-bonds (dashed lines) with the ATP-adenine or adenine-analogous moieties of ATP-competitive inhibitors, the T315 “gatekeeper” residue which can control access to hydrophobic pocket 2 and type 2/3 allosteric site, and characteristic hydrogen-bonds between imatinib and residues in αC-helix and in the DFG-motif at the beginning of the A-loop^{, , , ,} .

Fig. 4. Mechanisms of kinase inhibitor (KI) drug-resistance and approaches to overcome it

(A) Generalized mechanisms that can reduce KI drug efficacy or cause drug-resistance. Pharmacokinetic (PK) factors such as drug sequestration in extracellular space or plasma, suboptimal delivery, absorption, tissue penetration, enhanced metabolic turnover, clearance and excretion primarily affect drug cellular availability and efficacy (Fig. 4A,B). In pre-clinical models, drug-binding to plasma α1-acid glycoprotein (AGP) reduced drug delivery into cancer cells. However, the clinical relevance is unclear. Imatinib-release by the AGP competitive binder erythromycin could provide an avenue to overcome this potential problem^{, ,} . Pharmacogenomic factors can profoundly affect almost every PK/pharmacodynamic (PD) property of a drug. For example, polymorphisms in the drug-importer OCT-1 or the drug exporters MDR-1/P-glycoprotein (Pgp) and ABCG2 may affect imatinib, gefitinib or erlotinib pharmacokinetics and toxicity, although conflicting data render the clinical relevance unclear^{, ,} . Drug-resistance can be caused by target-cell extrinsic “systemic” mechanisms, or by a broad array of target cell autonomous/intrinsic mechanisms. In addition to those discussed in the text, the upregulation of cellular target protein activators, or the downregulation of target-protein inhibitors might contribute to drug-resistance or reduced drug efficacy^{, , -, ,} . (B) Schematic depiction of factors influencing KI efficacy or drug-resistance, exemplified by ABL-inhibitors^, . To maximize ease of use and patient compliance, drugs (blue stars) are preferably applied orally as tablets/capsules (1). In the gut (2), they are released and enter the blood via the hepatic portal vein (3). Unabsorbed drugs are excreted in the faeces. KIs enter their target cells (cells of the immune system in the case of ABL-inhibitors) via import-proteins, including the organic cation importer hOCT1 in case of imatinib (4). Cellular sequestration or metabolism may affect cellular drug efficacy. KIs are exported by ATP binding-cassette transporter families B and G2, including AGP in case of imatinib. Illustrated by BCR-ABL, several cell-intrinsic, auto- and paracrine effects can cause drug-resistance, discussed in detail in the text. Many of them are facilitated by the genomic instability of tumor cells. A major role of target gene missense mutations complements contributions by target-gene amplification, overexpression or epigenetic activation, or by the deregulation of redundant or downstream pathways. A recent review of KI PK properties suggests that overall, KIs reach their maximum plasma levels relatively fast, have an unknown absolute bioavailability, are extensively distributed and highly bound to plasma-proteins such as AGP in the blood (5,6). KIs are primarily metabolized (red stars) in the liver by cytochrome P450 (CYP) 3A4 (5) and excreted primarily via the biliary-fecal route (3). Only a minor fraction is eliminated with the urine (6,7). In a small study, elevated CYP3A activity and production of the therapeutically active metabolite CGP74588 associated with higher imatinib molecular responses. CYP1A1 induction by cigarette smoke may decrease erlotinib exposure. No significant contribution of CYP3A4/5 polymorphisms to KI efficacy/toxicity has been reported . Finally, the interesting KI ability to inhibit some of their own metabolizing enzymes and transporters renders steady-state metabolism and drug-drug interactions complex and unpredictable. (C) General approaches to improve drug efficacy and overcome drug-resistance. For details, see text and^{, , -, , , -}.

Fig. 5. Topological locations of known drug-resistance causing point mutations in ABL

Shown is the crystal structure of myristate-bound, autoinhibited full length hAbl-PD166326 (Pdb accession number 2FO0). (A) shows the entire structure including regulatory subunit interactions. (B) shows the KD only, oriented as those in fig. 2,3. Domains, structural features and kinase-inhibitor interaction sites are labeled and color-coded as in fig. 1-3. The colored spheres indicate the positions of drug-resistant point mutations clinically observed in ABL only (red), clinically observed in ABL and other kinases (light green), observed preclinically in ABL and clinically in other kinases (light blue), or only observed preclinically in ABL to date (brown, tab. 4,5).