- Email: [email protected]
The Landscape of Atypical and Eukaryotic Protein Kinases
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
Feature Review
The Landscape of Atypical and Eukaryotic Protein Kinases Georgi K. Kanev,1,2 Chris de Graaf,1,3 Iwan J.P. de Esch,1 Rob Leurs,1 Thomas Wu¨rdinger,2 Bart A. Westerman,2,5,* and Albert J. Kooistra1,4,5,* Kinases are attractive anticancer targets due to their central role in the growth, survival, and therapy resistance of tumor cells. This review explores the two primary kinase classes, the eukaryotic protein kinases (ePKs) and the atypical protein kinases (aPKs), and provides a structure-centered comparison of their sequences, structures, hydrophobic spines, mutation and SNP hotspots, and inhibitor interaction patterns. Despite the limited sequence similarity between these two classes, atypical kinases commonly share the archetypical kinase fold but lack conserved eukaryotic kinase motifs and possess altered hydrophobic spines. Furthermore, atypical kinase inhibitors explore only a limited number of binding modes both inside and outside the orthosteric binding site. The distribution of genetic variations in both classes shows multiple ways they can interfere with kinase inhibitor binding. This multilayered review provides a research framework bridging the eukaryotic and atypical kinase classes.
Two Classes of Protein Kinases: Atypical and Eukaryotic Kinases Protein kinases regulate cell signaling by catalyzing the transfer of the terminal phosphate group of ATP to substrate proteins [1]. Primarily due to their pivotal role in cancer, protein kinases have been subject to drug development efforts that resulted in the approval of more than 50 small-molecule kinase inhibitors (Box 1) by the US FDA. The 555 members of the human kinase superfamily have been groupedi,ii into a main class of 497 ePKs (see Glossary) and, due to lack of sequence similarity, 58 aPKs, which include the lipid kinases [2]. The ePK class is further subdivided into nine groups: AGC, CAMK, CK1, CMGC, Other, RGC, STE, TK, and TKL [3,4]. In the past three decades, extensive structural knowledge about this superfamily has been acquired. Somewhat surprisingly, the first aPK structures [5] revealed that several of these atypical kinases also share the prototypical ePK fold despite their lack of sequence similarity (Figure 1, Key Figure) [6]. Based on these insights, 26 of the 58 aPKs have been further classified as protein kinase-like (PKL) as they share the same structural fold as ePKs [7]. More recent aPK crystallographic highlights include the structures of PIK3CD [8,9], mTOR [10], and ADCK3 [11]. Until now, half of all protein kinases in the human kinome (268 ePKs and 15 aPKs) have been solved. In total, there are more than 4000 unique human protein kinase structures. On a structural level, the prototypical kinase fold of both ePKs and aPKs (see middle panel in Figure 1B) can be easily recognized by the two main lobes; namely, the N-lobe, which largely comprises b-sheets (I–V), and the C-lobe, which largely comprises a-helices (aD–aF; Figure 1B). The linker, a segment of three to five residues (shown in cyan in Figure 1), connects the two domains. ATP binds in the cleft that is formed between the C- and N-lobes [12]. The overwhelming majority of kinase inhibitors compete with ATP by binding to the same adenine pocket near the so-called hinge region. The most important regions directly involved in the binding of these inhibitors include the hinge region, HRD motif, DFG motif, G-rich loop, aC-helix, and catalytic and activation loops (Figure 1) [13,14]. Several of these segments are flexible and can adopt different conformations, thereby providing a large variety of options for inhibitor binding [15]. Moreover, the conserved Asp-Phe-Gly residues of the DFG motif can rotate (‘flip’) and are responsible for the activation and inactivation of the protein kinases. This motif is also involved in the (dis)assembly of the regulatory hydrophobic spine (R-spine) and indirectly the catalytic hydrophobic spine (C-spine) [16,17]. Both spines contribute to phosphorylation by stabilizing the protein kinase fold in its active form [16–18]. The conserved His-Arg-Asp residues of the HRD motif sustain kinase activity with their involvement in the transfer of a phosphate group of ATP to a bound substrate [19].
Highlights Despite their low sequence similarity, most atypical kinases share the same characteristic eukaryotic protein kinase fold. The atypical kinases lack the highly conserved kinase motifs (e.g., the GxGxxG motif) and contain a mirrored catalytic HRD motif. Atypical kinases have altered regulatory and catalytic hydrophobic spines compared with the eukaryotic kinases. SNPs and cancer mutations occurring in the catalytic cleft of the kinases are mutually exclusive in location in both atypical and eukaryotic kinases and can interfere with inhibitor binding. Atypical kinase inhibitors currently explore a limited number of binding modes compared with eukaryotic kinase inhibitors. Selective kinase inhibitors can be designed by rationally exploiting selectivity hotspots or by targeting alternative allosteric binding sites. 1Division of Medicinal Chemistry, Amsterdam Institute for Molecules, Medicines and Systems (AIMMS), Vrije Universiteit Amsterdam, De Boelelaan 1108, 1081 HZ Amsterdam, The Netherlands 2Department of Neurosurgery, Amsterdam University Medical Centers, Cancer Center Amsterdam, Brain Tumor Center Amsterdam, De Boelelaan 1117, 1081 HV Amsterdam, The Netherlands 3Current address: Sosei Heptares, Steinmetz Building, Granta Park, Great Abington, Cambridge, CB21 6DG, UK 4Current address: Department of Drug Design and Pharmacology, University of Copenhagen, Universitetsparken 2, 2100 Copenhagen, Denmark 5Shared
senior authors
*Correspondence: [email protected], [email protected]
Trends in Pharmacological Sciences, --, Vol. --, No. -- https://doi.org/10.1016/j.tips.2019.09.002 ª 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
1
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
Box 1. Drug Approvals: Growing Numbers for Both Kinase Classes
Glossary
The FDA approval of imatinib in 2001 formed the start of the kinase inhibitor era, which has been steadily rising from nine FDA drug approvals in the first decade and another 47 FDA approvals in the past decade leading to a total of 56 FDA-approved kinase inhibitors as reported in PKIDBvii [84] (Figure I). The stem of a nonproprietary drug name is linked to the mechanism of action or class of targets (see the list stems for International Nonproprietary Namesviii). Drugs targeting the tyrosine kinase family have -tinib as the suffix and form the largest category of kinase inhibitors with 39 drugs. With the approval of sorafenib in 2005, BRAF kinase inhibitors (drugs with -rafenib as suffix) were added to the repertoire, followed by angiogenesis inhibitors (drugs with -anib as suffix) such as pazopanib in 2009. Despite all drug development efforts, it was not until 2014 that the first drug was approved that targeted an aPK – namely, idelalisib, which targets the PI3K family (all drugs with -lisib as suffix); since then three more PI3K inhibitors have been approved (copanlisib in 2017, duvelisib in 2018, and alpelisib in 2019). In that period, also approved have been cyclin-dependent kinase inhibitors (drugs with -ciclib as suffix), netarsudil, a vasodilator (drugs with -sudil as suffix), and the broad-spectrum kinase inhibitor midostaurin.
aC-helix: helix in the N-terminal lobe of the kinase fold that frequently forms a salt bridge with the lysine at position 17 (KbIII.17) in the structural kinase database KLIFSi. Activation loop: the first of two loops that combined form the activation segment ranging from the DFG motif to the APE motif in ePKs. The activation loop stops after the phosphorylation sites (serine, threonine, or tyrosine residues) that once phosphorylated typically activate the protein kinase and is followed by the P+1 loop. Adenine pocket: part of the orthosteric pocket of the catalytic kinase domain where the adenine of ATP binds. Allosteric binding site: in protein kinases, binding site located outside the binding site of ATP. Atypical kinases: kinases with very low sequence similarity compared with the ‘normal’ ePKs, forming a separate cluster in the phylogenetic tree for kinases (kinome). Back pockets: all (sub)pockets located in the back cleft (Figure 4A); that is, between the gatekeeper, the aC-helix, and the DFG motif. Catalytic domain: the domain of a protein kinase where the catalysis occurs (i.e., the transfer of the gamma phosphate from ATP to bound substrate). Catalytic loop: the loop between b-sheets VI and VII starting with the conserved HRD motif (HisArg-Asp) and ending with the conserved asparagine (Asn, Nc.l.75). Both the HRD motif and Nc.l.75 are involved in the catalysis of ATP to a bound substrate. DFG motif: a highly conserved motif of three amino acids (AspPhe-Gly; positions 81–83 in the structural kinase database KLIFSi) at the beginning of the activation loop involved in the regulation of the activation state of the protein kinase. Eukaryotic protein kinases (ePKs): large protein superfamily with a highly conserved kinase domain that is involved in the regulation of many proteins by phosphorylation. Fingerprints: string of bits (zeros and ones), each of which indicates either the presence or the
Figure I. Number of FDA-Approved Small-Molecule Kinase Inhibitors Per Year Over the Past Two Decades. The gray line shows the overall growth whereas the icon color indicates the category of the approved kinase inhibitor as defined by the suffix of the nonproprietary drug name. Circle icons indicate inhibitors for atypical kinases and triangle icons indicate inhibitors for eukaryotic kinases.
In cancer, lack of (lasting) therapeutic efficacy is frequently observed in the clinic due to already present or de novo resistance mechanisms as a result of therapy [20,21]. An example of a widely studied mutation responsible for drug resistance is the gatekeeper mutation T790M in the EGFR receptor [22]. This mutation significantly decreases the efficacy of first-generation (e.g., gefitinib, erlotinib) and second-generation (e.g., afatinib, neratinib) tyrosine kinase inhibitors (TKIs) and resulted in the development of third-generation, mutant-specific, TKIs (e.g., osimertinib, olmutinib). A similar path led to the development of vemurafenib, an inhibitor that specifically targets BRAF with the V600E mutation [23,24]. Upcoming approaches in an attempt to break this iterative cycle by preventing these escape mechanisms are the identification of synergistic drug combinations and the rational design of inhibitors with a specific polypharmacological inhibition profile that matches the genetic fingerprint of the patient [25]. By tuning the inhibition of multiple pathways (i.e., multiple kinases), a treatment might restrict or even gridlock adaption mechanisms of oncogenic cells yielding synergistic inhibitory effects while avoiding drug resistance [26]. Of specific interest are the PI3K–mTOR [27] and Ras–Raf–MEK–ERK pathways [28–30], as parallel inhibition of these pathways often leads to a synergistic effect [31,32] by hampering the pathway crosstalk [30,33].
2
Trends in Pharmacological Sciences, --, Vol. --, No. --
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
Key Figure
The Two Primary Classes of Kinases: Eukaryotic and Atypical Protein Kinases
Figure 1. (A) Sankey-type flow diagram showing the nine eukaryotic and one atypical protein kinase groups with their numbers of kinases, structures, and inhibitors (shown in black) as well as mutations (shown in red) and SNPs (shown in green). Here, the atypical protein kinases are defined as all noneukaryotic protein kinases including also the nonprotein kinase-like kinases. (B) Most atypical and eukaryotic protein kinases share the same fold of the catalytic binding site (schematically shown in the middle). (C) The various topics investigated in this review of eukaryotic and atypical protein kinases.
absence of a feature; for example, a specific interaction. Front pockets: all (sub)pockets located in the front cleft (Figure 4A), the ATP-binding region. Gatekeeper: amino acid (position 45 in the structural kinase database KLIFSi) that acts as a gatekeeper between the front and back pockets. This residue is frequently mutated in oncogenic cell lines, which can induce drug resistance. G-rich loop: short loop of six amino acids (positions 4–9 in the structural kinase database KLIFSi) between b-sheets I and II containing three highly conserved glycines (GxGxxG motif). Hinge region: a segment of three amino acids (positions 46–48 in the structural kinase database KLIFSi) used as an anchor point by ATP and most protein kinase inhibitors. HRD motif: a highly conserved segment of three amino acids (His-Arg-Asp at positions 68–70 in the structural kinase database KLIFSi) in the catalytic loop involved in the catalysis of ATP. Orthosteric binding site: in protein kinases this is the binding site for ATP. Pathway crosstalk: communication between different pathways that allows cells to adapt to inhibition of certain pathways. Polypharmacological profile: simultaneously targeting multiple protein targets (kinases). Resistance mechanisms: adaptations of the cell (e.g., mutations in a protein kinase) to become resistant to drug treatment. Synergistic drug combination: combination of two or more drugs that achieves a greater effect than the sum of the effects of the individual drugs.
In this review, we provide a (structural) presentation of six layers (Figure 1) of kinase data providing a better understanding of the similarities and differences between atypical and protein kinases. Moreover, we provide a unifying, side-by-side presentation of the two kinase families for all topics in this review. Note that throughout this review we utilize the KLIFSi residue nomenclature for the 85 residues defining the binding cleft (see Figure I in Box 2) [34,35] whenever applicable (e.g., Flinker.50 refers to a phenylalanine at KLIFS position 50 of the binding site, which is located in the linker segment).
Trends in Pharmacological Sciences, --, Vol. --, No. --
3
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
From Sequence to Structure and from Mutation to Inhibitor Binding Understanding the mode of action of small-molecule kinase inhibitors on multiple targets is a highly complex and demanding task that requires the integration of structural, chemical, genetic, and phenotypic data [36]. To improve this understanding, we review all topics (Figure 1) for both ePKs and protein kinase-like aPKs, starting with (i) sequence and structural alignments of all ePKs and aPKs, which is followed by (ii) the spatially conserved kinase folds. Subsequently, the kinase classes are discussed in the light of (iii) genetic data in the form of single-nucleotide polymorphisms (SNPs) and oncogenic mutations along with kinase inhibitor binding to (iv) orthosteric and (v) allosteric sites. Finally, we present examples of (vi) the structural design of selective atypical kinase inhibitors.
Primary Sequence Alignment Shows That Atypical Kinases Lack the Conserved Glycines in the G-Rich Loop and Contain an Inverted HRD Motif The finding that protein structures can be conserved even at low sequence identity [37] certainly holds true for ePKs and most aPKs. Although they are highly divergent on the primary sequence level, at the tertiary level the structures of their catalytic domains are highly conserved, in both the N-lobe (G-rich loop, b-sheets I–IV, aC, gatekeeper, and hinge region) and the C-lobe (aD, aE, b-sheets VI–VIII, catalytic loop, and DFG motif). The first catalytic kinase domain structures were of ePKs [12,38] and revealed the now well-known fold of the catalytic kinase domain (Figure 1). Given that ePKs and aPKs are often coactivated in cancer [39], understanding sequence and structural differences and their consequences for inhibitor binding is important for drug discovery. The conserved glycines in the G-rich loop (commonly referred to as the p-loop for aPKs [9]) are important for the catalytic activity of eukaryotic kinases [13], which is very different for the aPKs as the ePK conserved glycine pattern (GxGxxG) is missing (Figure 2). These variations in the amino acids of the G-rich loop give rise to a selectivity pocket that is present only in aPKs, advancing the selectivity of aPK inhibitors. Despite these variations, the loop fulfills the same function as in ePKs; namely, the positioning and coordination of ATP [5,40]. Another major difference is observed at the HRD motif where the histidine and arginine are mirrored around the aspartate creating the atypical DRH motif (Figure 2). In this case, the histidine rather than the aspartate is suggested to act as a catalytic base during phosphorylation [41,42]. An example of the conformation of the HRD motif in ePKs as well as the DRH motif in aPKs is shown in Figure 2D.
Box 2. Consistent Residue Numbering for the Binding Site Throughout this review, the binding site numbering scheme of the structural kinase database KLIFSi is utilized to enable direct comparison between ePKs and aPKs. This numbering system was introduced as part of the KLIFS databasei [34,35] to facilitate the analysis of interactions between inhibitors and ePKs. To this end, 85 residues shown to interact with inhibitors binding to the cleft between the two lobes were selected (Figure I). For the purposes of this review, the numbering scheme and database were extended to also cover aPKs. The scheme starts with three residues in b-sheet I (positions 1–3) followed by the GxGxxG motif of the G-rich loop (positions 4–9) and is continued with four more residues of b-sheet II (positions 10–13). Then, the scheme cuts to six residues of b-sheet III (positions 14–19), including the VAVK/VAIK motif at positions 14–17. This is followed by 11 residues from the aC-helix (positions 20–30) with the highly conserved glutamate at position 24. The succeeding seven residues (positions 31–37) comprise the back loop (b.l.) connecting to four residues of b-sheet IV (positions 38–41). The next segment goes from b-sheet V (positions 42–44) via the gatekeeper (position 45), hinge (positions 46–48), and linker region (positions 49–52) all the way to the aD-helix (positions 53–59). After a short interval, the scheme is continued with the last turn of the aE-helix (positions 60–64) and moves via b-sheet VI (positions 65–67) and the catalytic loop (c.l.) (positions 68–75) to b-sheet VII (positions 76–78). The HRD motif in the catalytic loop can be found at positions 68–70 and the highly conserved asparagine at position 75. The last segment starts with one residue from b-sheet VIII (position 79) followed by the xDFG motif (positions 80–83) and two additional residues of the activation loop (position 84–85). When referring to a specific residue, the nomenclature is as follows:. (e.g., gk.45 refers to the gatekeeper at position 45). This can be combined with the annotation of a specific amino acid (e.g., Tgk.45) and UniProtix sequence number (e.g., T790gk.45). Additionally, the numbering scheme has an accompanying color profile that is used in all figures. The color coding is as follows: b-sheets, yellow; helices, red; G-rich and back loop, green; gatekeeper, gold/light orange; hinge region, magenta; linker region, cyan; catalytic loop, dark orange; and activation loop (including the xDFG motif), blue.
4
Trends in Pharmacological Sciences, --, Vol. --, No. --
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
Trends in Pharmacological Sciences
Figure I. The Residue Numbering Scheme Used to Navigate Eukaryotic and Atypical Protein Kinases. The numbering scheme of KLIFSi covers 85 residues known to interact with kinase inhibitors binding to the cleft between the N- and C-lobes. A table with both KLIFS numbering and UniProtix numbering is shown for two eukaryotic kinases (PRKACA and EGFR) and three atypical protein kinases (ATM, mTor, and PIK3CA).
By contrast, the residues KIII.17 and EaC.24, Nc.l.75, and the DFG motif are mostly conserved. KIII.17 and EaC.24 form a salt bridge that is important for maintaining kinase activity and interact with the a- and b-phosphates of ATP [43]. In several aPKs, there is an aspartate at position 24 (DaC.24), which due to its shorter side chain is incapable of forming the ePK-trademark salt bridge with KIII.17 [41]. Moreover, KIII.17 is part of the eukaryotic Val-Ala-Val/Ile-Lys (VAVK/VAIK) motif of which the highly conserved
Trends in Pharmacological Sciences, --, Vol. --, No. --
5
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
(A)
(B)
(C)
Trends in Pharmacological Sciences
Figure 2. Eukaryotic and Atypical Kinases Share Structurally Conserved Binding Sites with Only Some Sequence Motifs Also Being Conserved. (A) Sequence alignment of the binding sites shows the conservation of lysine at position 17, asparagine at position 76, and the DFG motif. The conserved glycines in the G-rich loop (green) are missing in the atypicals and the HRD motif is flipped (orange). The coloring of the backbones and residues throughout the paper is consistent with b-sheets in yellow, a-helices in red, G-rich loop in darker green, back loop (b.l.) in lighter green, gatekeeper in lighter orange, hinge in purple, linker in cyan, catalytic loop in darker orange, and DFG motif in blue. Residues not discussed in detail are colored light gray for clarity purposes. (B) Structural alignment from KLIFSi of the highly conserved binding sites of eukaryotic and atypical protein kinases. (C) An overlay of all orthosterically binding ligands with, in the middle, examples of binding modes for a promiscuous inhibitor (staurosporine) to a eukaryotic and an atypical protein kinase. Vectors showing the direction from the Ca to the Cg atoms of amino acids 12 and 35 show the distinct side-chain orientations in the structures of eukaryotic and atypical protein kinases. In addition, the residues of the flipped HRD/DRH motif in the catalytic loop are shown in an orange stick representation.
AIII.15 together with VII.11 forms the capping residues for the adenine ring of ATP [44]. For the aPKs, this motif is observed only in members of the ABC1/ADCK and the RIO kinase family. For other aPK members (the PIK and PIKK families) either the valine at position II.11 is replaced by a leucine or isoleucine having a
6
Trends in Pharmacological Sciences, --, Vol. --, No. --
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
similar function, or a larger hydrophobic residue instead of an alanine is seen at position III.15 with its side chain protruding into the space normally occupied by II.11 in ePKs [10,44]. A trademark feature of ePKs is a conserved Ala-Pro-Glu (APE) motif at the end of the activation segment that interacts with a conserved arginine in the loop between the aH and aI helices, which secures the activation loop near the aF-helix [45]. This motif is not observed for any of the aPKs [46]. Moreover, the activation segment of aPKs is, similar to more primitive eukaryotic-like protein kinases [45,46], much shorter than for ePKs. Instead, the activation loop of most aPKs (from the PIK and PIKK families) contain a conserved Pro-Phe-any-Leu-Thr (PFxLT) motif (not shown in the sequence conservation in Figure 2 as it lies outside the residue nomenclature scheme; Box 2). The threonine of this motif is known to be phosphorylated in PI3K kinases [47] and thereby marks the end of the activation loop. Moreover, the lysines in this loop have been shown to determine substrate specificity in PI3K kinases [48]. Immediately after the PFxLT motif, the aPKs have a short helix of two turns. This helix together with the activation loop present in most aPKs replaces the whole activation segment and aF-helix present in ePKs [45] (Figure 1).
Atypical Kinases Share a Spatially Conserved Kinase Fold with Eukaryotic Kinases The overlay of the binding sites of the ePKs and aPKs indicates remarkable structural similarity (Figure 2). Despite their overall structural similarity, the shape of the binding sites can differ substantially between ePKs and aPKs. In general, we see that although the average ePK binding site is wider (seen from the aD-helix to the aC-helix), the binding sites of aPKs are on average deeper (from the HRD motif to the hinge) and more open. When looking at the orientation of the side chains of the binding site residues, we find that uniquely for the aPKs the side chains of amino acid 12 are oriented towards the pocket, providing unique interaction opportunities for aPK inhibitors by partly forming the so-called specificity pocket [9] (Figure 2). In these cases, surrounding residues such as II.11 can be flipped thereby enabling the aforementioned protrusion of the capping residue III.15 into the space normally occupied by II.11. Amino acid 38 shows a similar trend, with most side chains in the aPKs being rotated roughly 90 around the backbone compared with the side chains of the ePKs at the same position (Figure 2). When not rotated, orthosterically binding aPK ligands can interact with the amino acid at position 38 in the socalled affinity pocket [9]. This in contrast to the ePKs, where the gatekeeper blocks access to amino acid 38, but due to a shift of the aPK gatekeeper it provides enough space to allow this interaction.
Atypical Kinases Have Altered Hydrophobic Spines Hydrophobic spines form two blocks of hydrophobic and aromatic amino acids inside the catalytic domains that regulate activation (R-spine) as well as priming for catalysis (C-spine) [17,18]. Residue–residue interactions in the Protein Contacts Atlas [49] show, as expected, that most of the DFG-in ePK and aPK structures are involved in highly conserved interaction motifs, with the exception of the amino acids at positions 11 and 15 in the ePKs. However, when ATP is bound it completes the C-spine by interacting via its adenine moiety with these two amino acids. The histidine residue of the HRD motif is one of the most conserved residues in the ePKs, making in 93% of all DFG-in structures contact with the phenylalanine of the DFG motif (position 82). Despite the flipped HRD motif in aPKs, position 68 still interacts with the DFG motif in virtually all aPK structures [49]. While the spines of aPKs in the binding sites seem similar to the ePKs, outside the binding sites substantial differences are observed. PI3K, mTOR, PI4K, ATR, DNAPK, and ATM all miss the entire aF-helix that causes rearrangement of the lower part of the C-spine. However, in all of the mentioned kinases there is an alternative cluster of hydrophobic residues in that region that interacts with the other residues of the C-spine. Only RIOK1 and ADCK3 are similar to the ePKs, as they have an aF-helix with hydrophobic residues at the conserved positions [50]. It should be noted that hydrophobic mutations in or surrounding the hydrophobic spines have been shown to affect kinase regulation with potential oncogenic effects [51].
Trends in Pharmacological Sciences, --, Vol. --, No. --
7
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
Frequently Occurring Mutations and SNPs Are Mutually Exclusive in Location To structurally compare the current landscape of mutations and SNPs in ePKs and aPKs, we investigated publicly available mutation and SNP data from COSMICiii [52], cBioPortaliv [53], and dbSNPv [54]. In total, 104 000 ePK and 16 000 aPK mutations (unique) and 18 000 ePK and 2700 aPK SNPs (unique) are listed in these databases. When mutation frequencies per residue are compared with SNPs, a mutually exclusive pattern was seen both for ePKs and aPKs: conserved regions (i.e., containing low SNP frequencies) show a relatively high occurrence of tumor-driving lesions (Figure 3). This points to a possible essential role of amino acids that are frequently mutated and not (often) compromised by SNPs [55]. Interestingly, SNPs can also be present in and around the inhibitor binding site (Figures 3 and 4), which potentially could directly interfere with the binding of inhibitors. The spatial organization of the mutations on the kinase structure shows that the two most frequent mutations inside the binding site of the ePKs are found at positions 4 (G-rich loop) and 45 (gatekeeper) (Figure 3). Mutations of the first glycine of the G-rich loop (position 4) impact kinase activity [13]. The most frequently observed mutations at that position occur in the EGFR: G719A, G719S, and G719C, with frequencies of 11%, 8%, and 5%, respectively (relative to the most frequent mutation – at position 45). Gatekeeper mutations are responsible for acquired drug resistance in many protein kinases [56]. The gatekeeper mutation T790M in the EGFR is the third most frequently occurring mutation (n = 1300) and is responsible for roughly 50% of acquired EGFR inhibitor resistance [56]. The frequencies of hyperactivating V600K and V600E BRAF mutations located in the activation loop follow directly after T790M [57]. The most frequent SNP variants for ePKs are found at position 40, which lies in b-sheet IV with its side chain pointing away from the binding site and thus does not directly affect the binding site itself. With an exceptionally high allele frequency of almost 50%, the S511N (serine to asparagine) mutation at this position for PAK5 is the most common (n = 2431) missense SNP in the ePK binding site, making PAK5 more similar to PAK4. Similar to ePKs, mutations in the G-rich loop of aPKs at position 4 are also frequently observed (65% relative to the most frequent mutated; i.e., the residue at position 22). Gatekeeper mutations are seldom observed in the aPKs (Figure 3). Frequently mutated positions in the aPKs are: position 22 (n = 40), position 32 (80% compared with position 22), and position 69 (70%). In PI3K, residue 22 (RaC.22) acts as hydrogen bond donor to the surrounding of E702 and Q705 thereby further stabilizing D836 and KIII.17 [58]. The arginine of the HRD motif (or rather, the DRH motif for aPKs) is involved in aligning the phosphorylated activation segment for substrate binding [59,60]. Point mutation of this arginine was previously found to significantly reduce kinase activity in yeast PKA [61]. The most frequently occurring mutations at that position are R743Q in EPHB1, R836C in EGFR, and G914R in PIK3CA. The most frequent missense SNP in the binding site of aPKs is observed at position 35 for RIOK2. This V175I (valine to isoleucine) mutation is located deep in the back loop and thereby only affects the back pocket of this kinase. As shown in the ranking of most frequently mutated kinases, the closely related ERBB4 and EGFR are the most mutated [62,63]. Among the aPKs, TRRAP and ATM have the highest numbers of unique mutations (1132 and 1130, respectively). ATM has the highest number of mutations inside the binding domain (145). The notable aPKs mTOR and PIK3CA are in the top ten with 744 and 714 mutations, respectively. PIK3CA has the highest number of mutations in the kinase fold (239), some of which are frequently found in cancers [28,64]. The most frequently occurring mutations in PIK3CA are found at positions 1047 (i.e., H1047R and H1047L) and 545 (i.e., E545K and E545A). Although these positions are outside the binding site [64], these mutants result in increased catalytic activity and the ability to induce tumors [65,66].
Atypical Kinase Inhibitors Explore a Limited Number of Binding Modes Analyzing the inhibitor binding modes of the ePKs and aPKs revealed that the inhibitors target a much wider portion of the binding site of ePKs (including the front cleft, the gatekeeper region, the back cleft, and even multiple sites outside the catalytic domain) than aPKs. All inhibitors currently solved
8
Trends in Pharmacological Sciences, --, Vol. --, No. --
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
(A)
(B)
(C) Mutations
Trends in Pharmacological Sciences
Figure 3. Differences in Mutation and SNP Patterns in Eukaryotic and Atypical Protein Kinases. (A) Schematic representations of the mutations and SNPs inside the binding site of the eukaryotic and atypical protein kinases and their frequencies. The gatekeeper (position 45) is the most frequently mutated in eukaryotic protein kinases while position 22 is the most frequently mutated in atypical protein kinases. The highest occurrences of SNPs are found at positions 40 and 35 in eukaryotic and atypical kinases, respectively. Scales are adjusted to the maximal numbers of unique mutations/SNPs. (B). Scatter plots of SNP versus mutation frequencies show mutually exclusive patterns for frequently occurring positions (i.e., high SNP rate and low mutation rate and vice versa). (C) Top ten most-mutated genes with their numbers of mutations and SNPs in the full protein and in the binding site. Unique mutations (e.g., H1047R and H1047L in PIK3CA) at the same position in a kinase were counted individually.
in aPK structures belong to either type I, which targets exclusively the front cleft (the same binding site as ATP) inside the binding site (Figure 4), or type I1/2, which targets both the front cleft and part of the back cleft in DFG-in conformations [4]. To the best of our knowledge, there are currently no known type II atypical inhibitors and also no aPK structures in the DFG-out conformation. Also, there are currently no known atypical type III inhibitors (i.e., inhibitors that are non-ATP competitive and exclusively target the back cleft or adjacent pockets).
Trends in Pharmacological Sciences, --, Vol. --, No. --
9
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
(A)
(B) Gatekeeper
(C)
Mutations
Mutations
(D)
Trends in Pharmacological Sciences
Figure 4. Limited Binding Mode Diversity for Orthosteric Atypical Kinase Inhibitors Compared with Eukaryotic Kinase Inhibitors. (A) A schematic overview and examples of type I, I1/2, II, and III inhibitors and their binding (sub)pockets. (B) The surface of all type I eukaryotic kinase inhibitors combined compared with atypical kinase inhibitors combined. (C). The combined inhibitor surface of type I1/2 eukaryotic and atypical kinase inhibitors. Currently, inhibitors of types II and III occur only in eukaryotic protein kinases. (D) The number of interactions per residue for each of the four different types of eukaryotic and atypical kinase inhibitors and the number of mutations/SNPs per position (expressed as % in relation to the highest mutated position). The color coding (yellow, green, red, orange, purple, cyan, and blue) under the residue numbers corresponds to segments of the binding site as defined in KLIFSi (Figure 2).
10
Trends in Pharmacological Sciences, --, Vol. --, No. --
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
Type I protein kinase inhibitors represent the largest group of inhibitors targeting this superfamily. These inhibitors are ATP competitive and exclusively target the front pocket using hydrogen bonds with the hinge region as an anchor point (Figure 4) [4,35]. In addition, the amino acids II.12, KIII.17, IV.38, g.k.45, linker.52, c.l.74, and DxDFG.81 commonly interact with this type, of which II.12 and IV.38 interact predominantly with inhibitors of the aPKs. The amino acids II.12 and IV.38 are involved in the formation of both the specificity pocket in the front cleft (positions 4 and 12) [9] and the affinity pocket (formed by the residues at positions 27, 24, and 38) in the aPKs [9]. A common challenge for type I inhibitors is obtaining selectivity for a specific (set of) kinase(s), primarily due to the highly conserved adenine pocket [67]. The selectivity could be improved by targeting distinct subpockets or nonconserved amino acids. One noticeable example among the eukaryotic inhibitors is CAM4066 (HET: 551) that targets a cryptic pocket of CK2a1 between the aD and the catalytic loop [68]. This compound and its derivatives are the only crystallized inhibitors that interact with the amino acid at position 73 [69]. Idelalisib (HET: 40L), the first aPK inhibitor approved by the FDA (2014; Box 1), is an ATP-competitive inhibitor of which the purine moiety makes a hydrogen bond with the hinge region of PI3Kd while the fluoroquinazolinone moiety penetrates the specificity pocket [70].
Gaining Selectivity in the Back Pocket Efforts to gain selectivity were also made by exploring the back pockets of the protein kinases. A variant of the type II kinase inhibitor STI-571 [71] was extended with a pyrazine moiety to reach deep into the back pocket [72] and became imatinib, the first FDA-approved small-molecule kinase inhibitor (Box 1) [73]. This inhibitor makes the common type II bonds with residues 67 and 68. In addition, type II inhibitors also make bonds with amino acids 31, 35, and 66. Between type I and type II, there is a type of inhibitor, type I1/2, that binds to both the front and the back like type II, but unlike this type, binds to the kinases in DFG-in conformation. An example of aPK type I1/2 is a pyrazolopyrimidine inhibitor [74], which is over 600 times more selective for mTOR than PI3Ka [74]. In the aPKs, it is only this inhibitor type that makes interactions with amino acid 24 (Figure 4) as the affinity pocket reaches the back of the protein kinases. In the ePKs, these inhibitors make selective bonds with residues 21, 28, and 38. The non-ATP-competitive type III inhibitors bind only to the back cleft and are not currently known for aPKs. Type III examples for ePK include inhibitor A1 [75] and a dibenzodiazepine inhibitor [76]. The latter makes an aromatic interaction with amino acid 60 that is observed only among this type of inhibitor. While the above-mentioned type I, I1/2, II, and III all directly target the catalytic cleft or an adjacent pocket, type IV inhibitors bind to allosteric sites away from this cleft [4]. In the current kinase structures, eight different type IV allosteric sites in ePKs and four in aPKs have been targeted (Figure 5). Of the eight ePK sites, five target the C-lobe and the remaining three the N-lobe [15]. A similar trend is seen for the aPKs, with three sites in the C-lobe and only one site in the N-lobe [77,78]. Currently, there are 250 Protein Data Bankvi (PDB) structures solved with type IV inhibitors, comprising 165 unique eukaryotic and 17 atypical type IV kinase inhibitors.
Selectivity Hotspots in the PI3K Family Although selectivity is hard to achieve, especially within very closely related kinases, it can be achieved by exploiting the correct hotspots within each kinase. The same goes for the PI3K family, where most inhibitors inhibit all four subtypes. Despite this challenge, selectivity hotspots for each of the PI3K kinases have been found and four examples are given in Figure 5, one for each subtype. The inhibitors belong to type I and type I1/2 and thus all make hydrogen bond(s) with the hinge region at amino acid 48 (Vhinge.48). Alpelisib (Figure 5), a highly selective PI3Ka inhibitor, makes hydrogen bonds with Slinker.50 and QaD.55 and a face-to-edge aromatic interaction with YIV.38 [79]. This compound is up to 50-fold more selective for PIK3Ca and its E545K and H1047R mutants over the rest of the PI3K (b/d/g) isoforms [80]. The selectivity driver of this inhibitor is the unique bidentate hydrogen bond with Q859aD.55, as this amino acid is not observed in the other subtypes [81]. SAR260301 makes hydrophobic contact with KI.2, a face-to-edge interaction with WII.12, and a hydrogen bond with the aspartate of the DFG motif. Despite having a binding mode similar to that of INK666,
Trends in Pharmacological Sciences, --, Vol. --, No. --
11
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
Figure 5. Gaining Selectivity by Targeting Distant Allosteric Pocket or Utilizing Selectivity Hotspots. (A) Type IV inhibitors: targeting allosteric binding pockets away from the ATP pocket for eukaryotic and atypical kinase inhibitors. One example of an allosteric ligand is shown in stick representation and a colored molecular surface per allosteric pocket. A few selected ligands are highlighted in detail by their 2D chemical structure. (B) Orthosteric atypical inhibitors gaining selectivity by interacting with selective pockets and/or not-conserved amino acids. At the top, four examples of binding modes of isoform-selective PI3K inhibitors and their interactions.
this compound is over 20 and 70 times more active for PIK3CB than for PIK3CD and PIK3CA and has no observed affinity for PI3KCG [82]. Its selectivity was reasoned to come from the KI.2 that restricts the selectivity pocket while its carbon atoms also provide a hydrophobic outline for the indoline moiety. INK666 also targets the selectivity and affinity pockets in PIK3CD, but makes use of the smaller threonine instead of a lysine at position 2 allowing a different angle and bulkier substituents. Additionally, INK666 makes a hydrogen bond with the aC-helix and a unique ionic interaction with the Dlinker.51 in the linker region [9]. This inhibitor is over 270 times more active for PIK3CD than PIK3CA and between
12
Trends in Pharmacological Sciences, --, Vol. --, No. --
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
ten and 20 times over PIK3CG and PIK3C. Another notable inhibitor is 2WH [83]. This inhibitor is over 200 times more selective for PI3Kg than the other isoforms [83] due to the small hydrophobic Alinker.50. In all other isoforms, the slightly bigger and polar serine at the same position prevents the inhibitor from assuming this binding mode.
Concluding Remarks By comprehensively reviewing all available human eukaryotic and atypical kinase structures, we provide a complete overview to allow the design of a next generation of kinase inhibitors with, for instance, optimal polypharmacological profiles that might be able to overcome drug resistance. Structurally, the eukaryotic and atypical kinases are highly similar with notable differences at the sequence level. This information enables matching of the unique features of small molecules to obtain an optimal selectivity profile. The mutation and SNP data revealed important hotspots and their location in the protein kinase fold. Moreover, when mutation frequencies per position are compared with SNPs, a mutually exclusive pattern was seen for both ePKs and aPKs, which is likely to reflect evolutionary cost–benefit relations. This integrated kinase review merging the two primary kinase classes provides clues for next-generation personalized medicine, for which several hurdles have yet to be overcome (see Outstanding Questions).
Acknowledgments This work is supported by a grant of the Amsterdam Academic Alliance (Amsterdam Data Science) as provided by the Boards of the VU and UvA universities, and by the Cancer Center Amsterdam - project CCA2018-2-19.
Outstanding Questions Can we use structural data to generate new lead compounds with desired bioactivity profiles? Is it possible to target deep into the back pockets of the atypical kinases and how would that affect the selectivity profile? Can desired combinations of eukaryotic and atypical kinases be simultaneously targeted with a single inhibitor? What would be the best approach to translate the wealth of kinase crystal structures into de novo predictions of bioactivity profiles for existing compounds? Does simultaneous targeting of atypical and eukaryotic protein kinase mutations prevent therapy resistance? Can therapy resistance through pathway redundancy/compensation be simultaneously prevented by polypharmacological drugs?
Resources i
http://klifs.vu-compmedchem.nl www.kinase.com/web/current/kinbase/ iii https://cancer.sanger.ac.uk/cosmic iv www.cbioportal.org v www.ncbi.nlm.nih.gov/snp/ vi www.rcsb.org vii www.icoa.fr/pkidb/ viii www.who.int/medicines/services/inn/stembook/en/ ix https://uniprot.org ii
Which of the existing drugs inhibit tumor driving mutations as well as resistance mechanisms? How can we optimally match the bioactivity profiles of single drugs for maximal therapy effects based on the patient’s target profile?
References 1. Cohen, P. (2002) The origins of protein phosphorylation. Nat. Cell Biol. 4, E127 2. Heath, C.M. et al. (2003) Lipid kinases play crucial and multiple roles in membrane trafficking and signaling. Histol. Histopathol. 18, 989–998 3. Manning, G. et al. (2002) The protein kinase complement of the human genome. Science 298, 1912–1934 4. Roskoski, R., Jr. (2015) A historical overview of protein kinases and their targeted small molecule inhibitors. Pharmacol. Res. 100, 1–23 5. Walker, E.H. et al. (1999) Structural insights into phosphoinositide 3-kinase catalysis and signalling. Nature 402, 313 6. Yamaguchi, H. et al. (2001) Crystal structure of the atypical protein kinase domain of a TRP channel with phosphotransferase activity. Mol. Cell 7, 1047–1057 7. Scheeff, E.D. and Bourne, P.E. (2005) Structural evolution of the protein kinase-like superfamily. PLoS Comput. Biol. 1, e49 8. Heffron, T.P. et al. (2016) The rational design of selective benzoxazepin inhibitors of the alphaisoform of phosphoinositide 3-kinase culminating in the identification of (S)-2-((2-(1-isopropyl-1H-1,2,4-
9. 10. 11. 12.
13. 14. 15. 16.
triazol-5-yl)-5,6-dihydrobenzo[f]imidazo[1,2-d][1,4] oxazepin-9-yl)oxy)propanamide (GDC-0326). J. Med. Chem. 59, 985–1002 Berndt, A. et al. (2010) The p110 delta structure: mechanisms for selectivity and potency of new PI3K inhibitors. Nat. Chem. Biol. 6, 117–124 Yang, H. et al. (2013) mTOR kinase structure, mechanism and regulation. Nature 497, 217–223 Stefely, J.A. et al. (2015) Mitochondrial ADCK3 employs an atypical protein kinase-like fold to enable coenzyme Q biosynthesis. Mol. Cell 57, 83–94 Knighton, D. et al. (1991) Crystal structure of the catalytic subunit of cyclic adenosine monophosphate-dependent protein kinase. Science 253, 407–414 Hemmer, W. et al. (1997) Role of the glycine triad in the ATP-binding site of cAMP-dependent protein kinase. J. Biol. Chem. 272, 16946–16954 Greenman, C. et al. (2007) Patterns of somatic mutation in human cancer genomes. Nature 446, 153–158 Wu, P. et al. (2015) Allosteric small-molecule kinase inhibitors. Pharmacol. Ther. 156, 59–68 Kornev, A.P. et al. (2006) Surface comparison of active and inactive protein kinases identifies a conserved
Trends in Pharmacological Sciences, --, Vol. --, No. --
13
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
17. 18. 19.
20. 21. 22.
23. 24. 25. 26.
27. 28. 29. 30. 31.
32. 33. 34. 35. 36.
37.
38. 39.
14
activation mechanism. Proc. Natl Acad. Sci. U. S. A. 103, 17783–17788 Kornev, A.P. et al. (2008) A helix scaffold for the assembly of active protein kinases. Proc. Natl Acad. Sci. U. S. A. 105, 14377–14382 Taylor, S.S. and Kornev, A.P. (2011) Protein kinases: evolution of dynamic regulatory proteins. Trends Biochem. Sci. 36, 65–77 Adams, J.A. (2003) Activation loop phosphorylation and catalysis in protein kinases: is there functional evidence for the autoinhibitor model? Biochemistry 42, 601–607 Ramos, P. and Bentires-Alj, M. (2015) Mechanismbased cancer therapy: resistance to therapy, therapy for resistance. Oncogene 34, 3617–3626 Holohan, C. et al. (2013) Cancer drug resistance: an evolving paradigm. Nat. Rev. Cancer 13, 714–726 Yun, C.H. et al. (2008) The T790M mutation in EGFR kinase causes drug resistance by increasing the affinity for ATP. Proc. Natl Acad. Sci. U. S. A. 105, 2070–2075 Davies, H. et al. (2002) Mutations of the BRAF gene in human cancer. Nature 417, 949 Chapman, P.B. et al. (2011) Improved survival with vemurafenib in melanoma with BRAF V600E mutation. N. Engl. J. Med. 364, 2507–2516 Knight, Z.A. et al. (2010) Targeting the cancer kinome through polypharmacology. Nat. Rev. Cancer 10, 130–137 Mir, S.E. et al. (2010) In silico analysis of kinase expression identifies WEE1 as a gatekeeper against mitotic catastrophe in glioblastoma. Cancer Cell 18, 244–257 Fruman, D.A. et al. (2017) The PI3K pathway in human disease. Cell 170, 605–635 Samuels, Y. et al. (2004) High frequency of mutations of the PIK3CA gene in human cancers. Science 304, 554 Thorpe, L.M. et al. (2015) PI3K in cancer: divergent roles of isoforms, modes of activation and therapeutic targeting. Nat. Rev. Cancer 15, 7–24 Mendoza, M.C. et al. (2011) The Ras–ERK and PI3K– mTOR pathways: cross-talk and compensation. Trends Biochem. Sci. 36, 320–328 Engelman, J.A. et al. (2008) Effective use of PI3K and MEK inhibitors to treat mutant Kras G12D and PIK3CA H1047R murine lung cancers. Nat. Med. 14, 1351–1356 LoPiccolo, J. et al. (2008) Targeting the PI3K/Akt/ mTOR pathway: effective combinations and clinical considerations. Drug Resist. Updat. 11, 32–50 Sanchez-Vega, F. et al. (2018) Oncogenic signaling pathways in The Cancer Genome Atlas. Cell 173, 321–337 Kooistra, A.J. et al. (2016) KLIFS: a structural kinase– ligand interaction database. Nucleic Acids Res. 44, D365–D371 van Linden, O.P. et al. (2014) KLIFS: a knowledgebased structural database to navigate kinase–ligand interaction space. J. Med. Chem. 57, 249–277 Kanev, G.K. et al. (2017) Structural chemogenomics databases to navigate protein– ligand interaction space. In Comprehensive Medicinal Chemistry III, S. Chackalamannil et al., eds. (Elsevier), pp. 444–471 Sander, C. and Schneider, R. (1991) Database of homology-derived protein structures and the structural meaning of sequence alignment. Proteins 9, 56–68 Hubbard, S.R. et al. (1994) Crystal structure of the tyrosine kinase domain of the human insulin receptor. Nature 372, 746 Jaeger, S. et al. (2017) Quantification of pathway cross-talk reveals novel synergistic drug
40. 41. 42. 43. 44. 45.
46.
47.
48.
49. 50.
51. 52. 53. 54. 55. 56. 57. 58.
59.
60. 61.
combinations for breast cancer. Cancer Res. 77, 459–469 Baretic, D. and Williams, R.L. (2014) The structural basis for mTOR function. Semin. Cell Dev. Biol. 36, 91–101 Vadas, O. et al. (2011) Structural basis for activation and inhibition of class I phosphoinositide 3-kinases. Sci. Signal. 4, re2 Miller, S. et al. (2010) Shaping development of autophagy inhibitors with the structure of the lipid kinase Vps34. Science 327, 1638–1642 Endicott, J.A. et al. (2012) The structural basis for control of eukaryotic protein kinases. Annu. Rev. Biochem. 81, 587–613 Taylor, S.S. et al. (2019) Evolution of a dynamic molecular switch. IUBMB Life 71, 672–684 Taylor, S.S. et al. (2012) Evolution of the eukaryotic protein kinases as dynamic molecular switches. Philos. Trans. R. Soc. Lond. B Biol. Sci. 367, 2517–2528 Kannan, N. and Neuwald, A.F. (2005) Did protein kinase regulatory mechanisms evolve through elaboration of a simple structural component? J. Mol. Biol. 351, 956–972 Wang, B.D. et al. (2017) Alternative splicing promotes tumour aggressiveness and drug resistance in African American prostate cancer. Nat. Commun. 8, 15921 Pirola, L. et al. (2001) Activation loop sequences confer substrate specificity to phosphoinositide 3kinase alpha (PI3Ka). Functions of lipid kinasedeficient PI3Ka in signaling. J. Biol. Chem. 276, 21544–21554 Kayikci, M. et al. (2018) Visualization and analysis of non-covalent contacts using the Protein Contacts Atlas. Nat. Struct. Mol. Biol. 25, 185–194 Coun˜ago, R.M. et al. (2017) Development of narrow spectrum ATP-competitive kinase inhibitors as probes for BIKE and AAK1. bioRxiv. Published online March 2, 2017. https://doi.org/10.1101/094631 Hu, J. et al. (2015) Kinase regulation by hydrophobic spine assembly in cancer. Mol. Cell. Biol. 35, 264–276 Forbes, S.A. et al. (2017) COSMIC: somatic cancer genetics at high-resolution. Nucleic Acids Res. 45, D777–D783 Gao, J. et al. (2013) Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci. Signal. 6, pl1 Sherry, S.T. et al. (2001) dbSNP: the NCBI database of genetic variation. Nucleic Acids Res. 29, 308–311 Ikediobi, O.N. et al. (2006) Mutation analysis of 24 known cancer genes in the NCI-60 cell line set. Mol. Cancer Ther. 5, 2606–2612 Sequist, L.V. et al. (2011) Genotypic and histological evolution of lung cancers acquiring resistance to EGFR inhibitors. Sci. Transl. Med. 3, 75ra26 Karoulia, Z. et al. (2017) New perspectives for targeting RAF kinase in human cancer. Nat. Rev. Cancer 17, 676–691 Carter, H. et al. (2014) Prioritization of driver mutations in pancreatic cancer using cancer-specific high-throughput annotation of somatic mutations (CHASM). Cancer Biol. Ther. 10, 582–587 Krupa, A. et al. (2004) Structural modes of stabilization of permissive phosphorylation sites in protein kinases: distinct strategies in Ser/Thr and Tyr kinases. J. Mol. Biol. 339, 1025–1039 Johnson, L.N. et al. (1996) Active and inactive protein kinases: structural basis for regulation. Cell 85, 149–158 Gibbs, C.S. and Zoller, M.J. (1991) Rational scanning mutagenesis of a protein kinase identifies functional regions involved in catalysis and substrate interactions. J. Biol. Chem. 266, 8923–8931
Trends in Pharmacological Sciences, --, Vol. --, No. --
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
62. Hynes, N.E. and Lane, H.A. (2005) ERBB receptors and cancer: the complexity of targeted inhibitors. Nat. Rev. Cancer 5, 341–354 63. Gazdar, A.F. (2009) Activating and resistance mutations of EGFR in non-small-cell lung cancer: role in clinical response to EGFR tyrosine kinase inhibitors. Oncogene 28, S24 64. Liu, S. et al. (2014) The structural basis of PI3K cancer mutations: from mechanism to therapy. Cancer Res. 74, 641–646 65. Kang, S. et al. (2005) Phosphatidylinositol 3-kinase mutations identified in human cancer are oncogenic. Proc. Natl. Acad. Sci. U. S. A. 102, 802–807 66. Meyer, D.S. et al. (2013) Expression of PIK3CA mutant E545K in the mammary gland induces heterogeneous tumors but is less potent than mutant H1047R. Oncogenesis 2, e74 67. Morphy, R. (2010) Selectively nonselective kinase inhibition: striking the right balance. J. Med. Chem. 53, 1413–1437 68. De Fusco, C. et al. (2017) A fragment-based approach leading to the discovery of a novel binding site and the selective CK2 inhibitor CAM4066. Bioorg. Med. Chem. 25, 3471–3482 69. Brear, P. et al. (2016) Specific inhibition of CK2a from an anchor outside the active site. Chem. Sci. 7, 6839– 6845 70. Somoza, J.R. et al. (2015) Structural, biochemical, and biophysical characterization of idelalisib binding to phosphoinositide 3-kinase delta. J. Biol. Chem. 290, 8439–8446 71. Schindler, T. et al. (2000) Structural mechanism for STI-571 inhibition of Abelson tyrosine kinase. Science 289, 1938–1942 72. Nagar, B. et al. (2002) Crystal structures of the kinase domain of c-Abl in complex with the small molecule inhibitors PD173955 and imatinib (STI-571). Cancer Res 62, 4236–4243 73. Dagher, R. et al. (2002) Approval summary: imatinib mesylate in the treatment of metastatic and/or unresectable malignant gastrointestinal stromal tumors. Clin. Cancer Res. 8, 3034–3038
74. Zask, A. et al. (2009) ATP-competitive inhibitors of the mammalian target of rapamycin: design and synthesis of highly potent and selective pyrazolopyrimidines. J. Med. Chem. 52, 5013–5016 75. Furuya, N. et al. (2017) The juxtamembrane region of TrkA kinase is critical for inhibitor selectivity. Bioorg. Med. Chem. Lett. 27, 1233–1236 76. Karpov, A.S. et al. (2015) Optimization of a dibenzodiazepine hit to a potent and selective allosteric PAK1 inhibitor. ACS Med. Chem. Lett. 6, 776–781 77. Miller, M.S. et al. (2017) Identification of allosteric binding sites for PI3Ka oncogenic mutant specific inhibitor design. Bioorg. Med. Chem. 25, 1481– 1486 78. Hon, W.C. et al. (2011) Regulation of lipid binding underlies the activation mechanism of class IA PI3kinases. Oncogene 31, 3655 79. Furet, P. et al. (2013) Discovery of NVP-BYL719 a potent and selective phosphatidylinositol-3 kinase alpha inhibitor selected for clinical evaluation. Bioorg. Med. Chem. Lett. 23, 3741–3748 80. Fritsch, C. et al. (2014) Characterization of the novel and specific PI3Ka inhibitor NVP-BYL719 and development of the patient stratification strategy for clinical trials. Mol. Cancer Ther. 13, 1117–1129 81. Frazzetto, M. et al. (2008) Dissecting isoform selectivity of PI3K inhibitors: the role of nonconserved residues in the catalytic pocket. Biochem. J. 414, 383–390 82. Certal, V. et al. (2014) Discovery and optimization of pyrimidone indoline amide PI3Kb inhibitors for the treatment of phosphatase and tensin homologue (PTEN)-deficient cancers. J. Med. Chem. 57, 903–920 83. Collier, P.N. et al. (2015) Structural basis for isoform selectivity in a class of benzothiazole inhibitors of phosphoinositide 3-kinase gamma. J. Med. Chem. 58, 517–521 84. Carles, F. et al. (2018) PKIDB: a curated, annotated and updated database of protein kinase inhibitors in clinical trials. Molecules 23, E980
Trends in Pharmacological Sciences, --, Vol. --, No. --
15
Trends in Pharmacological Sciences
Feature Review
The Landscape of Atypical and Eukaryotic Protein Kinases Georgi K. Kanev,1,2 Chris de Graaf,1,3 Iwan J.P. de Esch,1 Rob Leurs,1 Thomas Wu¨rdinger,2 Bart A. Westerman,2,5,* and Albert J. Kooistra1,4,5,* Kinases are attractive anticancer targets due to their central role in the growth, survival, and therapy resistance of tumor cells. This review explores the two primary kinase classes, the eukaryotic protein kinases (ePKs) and the atypical protein kinases (aPKs), and provides a structure-centered comparison of their sequences, structures, hydrophobic spines, mutation and SNP hotspots, and inhibitor interaction patterns. Despite the limited sequence similarity between these two classes, atypical kinases commonly share the archetypical kinase fold but lack conserved eukaryotic kinase motifs and possess altered hydrophobic spines. Furthermore, atypical kinase inhibitors explore only a limited number of binding modes both inside and outside the orthosteric binding site. The distribution of genetic variations in both classes shows multiple ways they can interfere with kinase inhibitor binding. This multilayered review provides a research framework bridging the eukaryotic and atypical kinase classes.
Two Classes of Protein Kinases: Atypical and Eukaryotic Kinases Protein kinases regulate cell signaling by catalyzing the transfer of the terminal phosphate group of ATP to substrate proteins [1]. Primarily due to their pivotal role in cancer, protein kinases have been subject to drug development efforts that resulted in the approval of more than 50 small-molecule kinase inhibitors (Box 1) by the US FDA. The 555 members of the human kinase superfamily have been groupedi,ii into a main class of 497 ePKs (see Glossary) and, due to lack of sequence similarity, 58 aPKs, which include the lipid kinases [2]. The ePK class is further subdivided into nine groups: AGC, CAMK, CK1, CMGC, Other, RGC, STE, TK, and TKL [3,4]. In the past three decades, extensive structural knowledge about this superfamily has been acquired. Somewhat surprisingly, the first aPK structures [5] revealed that several of these atypical kinases also share the prototypical ePK fold despite their lack of sequence similarity (Figure 1, Key Figure) [6]. Based on these insights, 26 of the 58 aPKs have been further classified as protein kinase-like (PKL) as they share the same structural fold as ePKs [7]. More recent aPK crystallographic highlights include the structures of PIK3CD [8,9], mTOR [10], and ADCK3 [11]. Until now, half of all protein kinases in the human kinome (268 ePKs and 15 aPKs) have been solved. In total, there are more than 4000 unique human protein kinase structures. On a structural level, the prototypical kinase fold of both ePKs and aPKs (see middle panel in Figure 1B) can be easily recognized by the two main lobes; namely, the N-lobe, which largely comprises b-sheets (I–V), and the C-lobe, which largely comprises a-helices (aD–aF; Figure 1B). The linker, a segment of three to five residues (shown in cyan in Figure 1), connects the two domains. ATP binds in the cleft that is formed between the C- and N-lobes [12]. The overwhelming majority of kinase inhibitors compete with ATP by binding to the same adenine pocket near the so-called hinge region. The most important regions directly involved in the binding of these inhibitors include the hinge region, HRD motif, DFG motif, G-rich loop, aC-helix, and catalytic and activation loops (Figure 1) [13,14]. Several of these segments are flexible and can adopt different conformations, thereby providing a large variety of options for inhibitor binding [15]. Moreover, the conserved Asp-Phe-Gly residues of the DFG motif can rotate (‘flip’) and are responsible for the activation and inactivation of the protein kinases. This motif is also involved in the (dis)assembly of the regulatory hydrophobic spine (R-spine) and indirectly the catalytic hydrophobic spine (C-spine) [16,17]. Both spines contribute to phosphorylation by stabilizing the protein kinase fold in its active form [16–18]. The conserved His-Arg-Asp residues of the HRD motif sustain kinase activity with their involvement in the transfer of a phosphate group of ATP to a bound substrate [19].
Highlights Despite their low sequence similarity, most atypical kinases share the same characteristic eukaryotic protein kinase fold. The atypical kinases lack the highly conserved kinase motifs (e.g., the GxGxxG motif) and contain a mirrored catalytic HRD motif. Atypical kinases have altered regulatory and catalytic hydrophobic spines compared with the eukaryotic kinases. SNPs and cancer mutations occurring in the catalytic cleft of the kinases are mutually exclusive in location in both atypical and eukaryotic kinases and can interfere with inhibitor binding. Atypical kinase inhibitors currently explore a limited number of binding modes compared with eukaryotic kinase inhibitors. Selective kinase inhibitors can be designed by rationally exploiting selectivity hotspots or by targeting alternative allosteric binding sites. 1Division of Medicinal Chemistry, Amsterdam Institute for Molecules, Medicines and Systems (AIMMS), Vrije Universiteit Amsterdam, De Boelelaan 1108, 1081 HZ Amsterdam, The Netherlands 2Department of Neurosurgery, Amsterdam University Medical Centers, Cancer Center Amsterdam, Brain Tumor Center Amsterdam, De Boelelaan 1117, 1081 HV Amsterdam, The Netherlands 3Current address: Sosei Heptares, Steinmetz Building, Granta Park, Great Abington, Cambridge, CB21 6DG, UK 4Current address: Department of Drug Design and Pharmacology, University of Copenhagen, Universitetsparken 2, 2100 Copenhagen, Denmark 5Shared
senior authors
*Correspondence: [email protected], [email protected]
Trends in Pharmacological Sciences, --, Vol. --, No. -- https://doi.org/10.1016/j.tips.2019.09.002 ª 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
1
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
Box 1. Drug Approvals: Growing Numbers for Both Kinase Classes
Glossary
The FDA approval of imatinib in 2001 formed the start of the kinase inhibitor era, which has been steadily rising from nine FDA drug approvals in the first decade and another 47 FDA approvals in the past decade leading to a total of 56 FDA-approved kinase inhibitors as reported in PKIDBvii [84] (Figure I). The stem of a nonproprietary drug name is linked to the mechanism of action or class of targets (see the list stems for International Nonproprietary Namesviii). Drugs targeting the tyrosine kinase family have -tinib as the suffix and form the largest category of kinase inhibitors with 39 drugs. With the approval of sorafenib in 2005, BRAF kinase inhibitors (drugs with -rafenib as suffix) were added to the repertoire, followed by angiogenesis inhibitors (drugs with -anib as suffix) such as pazopanib in 2009. Despite all drug development efforts, it was not until 2014 that the first drug was approved that targeted an aPK – namely, idelalisib, which targets the PI3K family (all drugs with -lisib as suffix); since then three more PI3K inhibitors have been approved (copanlisib in 2017, duvelisib in 2018, and alpelisib in 2019). In that period, also approved have been cyclin-dependent kinase inhibitors (drugs with -ciclib as suffix), netarsudil, a vasodilator (drugs with -sudil as suffix), and the broad-spectrum kinase inhibitor midostaurin.
aC-helix: helix in the N-terminal lobe of the kinase fold that frequently forms a salt bridge with the lysine at position 17 (KbIII.17) in the structural kinase database KLIFSi. Activation loop: the first of two loops that combined form the activation segment ranging from the DFG motif to the APE motif in ePKs. The activation loop stops after the phosphorylation sites (serine, threonine, or tyrosine residues) that once phosphorylated typically activate the protein kinase and is followed by the P+1 loop. Adenine pocket: part of the orthosteric pocket of the catalytic kinase domain where the adenine of ATP binds. Allosteric binding site: in protein kinases, binding site located outside the binding site of ATP. Atypical kinases: kinases with very low sequence similarity compared with the ‘normal’ ePKs, forming a separate cluster in the phylogenetic tree for kinases (kinome). Back pockets: all (sub)pockets located in the back cleft (Figure 4A); that is, between the gatekeeper, the aC-helix, and the DFG motif. Catalytic domain: the domain of a protein kinase where the catalysis occurs (i.e., the transfer of the gamma phosphate from ATP to bound substrate). Catalytic loop: the loop between b-sheets VI and VII starting with the conserved HRD motif (HisArg-Asp) and ending with the conserved asparagine (Asn, Nc.l.75). Both the HRD motif and Nc.l.75 are involved in the catalysis of ATP to a bound substrate. DFG motif: a highly conserved motif of three amino acids (AspPhe-Gly; positions 81–83 in the structural kinase database KLIFSi) at the beginning of the activation loop involved in the regulation of the activation state of the protein kinase. Eukaryotic protein kinases (ePKs): large protein superfamily with a highly conserved kinase domain that is involved in the regulation of many proteins by phosphorylation. Fingerprints: string of bits (zeros and ones), each of which indicates either the presence or the
Figure I. Number of FDA-Approved Small-Molecule Kinase Inhibitors Per Year Over the Past Two Decades. The gray line shows the overall growth whereas the icon color indicates the category of the approved kinase inhibitor as defined by the suffix of the nonproprietary drug name. Circle icons indicate inhibitors for atypical kinases and triangle icons indicate inhibitors for eukaryotic kinases.
In cancer, lack of (lasting) therapeutic efficacy is frequently observed in the clinic due to already present or de novo resistance mechanisms as a result of therapy [20,21]. An example of a widely studied mutation responsible for drug resistance is the gatekeeper mutation T790M in the EGFR receptor [22]. This mutation significantly decreases the efficacy of first-generation (e.g., gefitinib, erlotinib) and second-generation (e.g., afatinib, neratinib) tyrosine kinase inhibitors (TKIs) and resulted in the development of third-generation, mutant-specific, TKIs (e.g., osimertinib, olmutinib). A similar path led to the development of vemurafenib, an inhibitor that specifically targets BRAF with the V600E mutation [23,24]. Upcoming approaches in an attempt to break this iterative cycle by preventing these escape mechanisms are the identification of synergistic drug combinations and the rational design of inhibitors with a specific polypharmacological inhibition profile that matches the genetic fingerprint of the patient [25]. By tuning the inhibition of multiple pathways (i.e., multiple kinases), a treatment might restrict or even gridlock adaption mechanisms of oncogenic cells yielding synergistic inhibitory effects while avoiding drug resistance [26]. Of specific interest are the PI3K–mTOR [27] and Ras–Raf–MEK–ERK pathways [28–30], as parallel inhibition of these pathways often leads to a synergistic effect [31,32] by hampering the pathway crosstalk [30,33].
2
Trends in Pharmacological Sciences, --, Vol. --, No. --
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
Key Figure
The Two Primary Classes of Kinases: Eukaryotic and Atypical Protein Kinases
Figure 1. (A) Sankey-type flow diagram showing the nine eukaryotic and one atypical protein kinase groups with their numbers of kinases, structures, and inhibitors (shown in black) as well as mutations (shown in red) and SNPs (shown in green). Here, the atypical protein kinases are defined as all noneukaryotic protein kinases including also the nonprotein kinase-like kinases. (B) Most atypical and eukaryotic protein kinases share the same fold of the catalytic binding site (schematically shown in the middle). (C) The various topics investigated in this review of eukaryotic and atypical protein kinases.
absence of a feature; for example, a specific interaction. Front pockets: all (sub)pockets located in the front cleft (Figure 4A), the ATP-binding region. Gatekeeper: amino acid (position 45 in the structural kinase database KLIFSi) that acts as a gatekeeper between the front and back pockets. This residue is frequently mutated in oncogenic cell lines, which can induce drug resistance. G-rich loop: short loop of six amino acids (positions 4–9 in the structural kinase database KLIFSi) between b-sheets I and II containing three highly conserved glycines (GxGxxG motif). Hinge region: a segment of three amino acids (positions 46–48 in the structural kinase database KLIFSi) used as an anchor point by ATP and most protein kinase inhibitors. HRD motif: a highly conserved segment of three amino acids (His-Arg-Asp at positions 68–70 in the structural kinase database KLIFSi) in the catalytic loop involved in the catalysis of ATP. Orthosteric binding site: in protein kinases this is the binding site for ATP. Pathway crosstalk: communication between different pathways that allows cells to adapt to inhibition of certain pathways. Polypharmacological profile: simultaneously targeting multiple protein targets (kinases). Resistance mechanisms: adaptations of the cell (e.g., mutations in a protein kinase) to become resistant to drug treatment. Synergistic drug combination: combination of two or more drugs that achieves a greater effect than the sum of the effects of the individual drugs.
In this review, we provide a (structural) presentation of six layers (Figure 1) of kinase data providing a better understanding of the similarities and differences between atypical and protein kinases. Moreover, we provide a unifying, side-by-side presentation of the two kinase families for all topics in this review. Note that throughout this review we utilize the KLIFSi residue nomenclature for the 85 residues defining the binding cleft (see Figure I in Box 2) [34,35] whenever applicable (e.g., Flinker.50 refers to a phenylalanine at KLIFS position 50 of the binding site, which is located in the linker segment).
Trends in Pharmacological Sciences, --, Vol. --, No. --
3
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
From Sequence to Structure and from Mutation to Inhibitor Binding Understanding the mode of action of small-molecule kinase inhibitors on multiple targets is a highly complex and demanding task that requires the integration of structural, chemical, genetic, and phenotypic data [36]. To improve this understanding, we review all topics (Figure 1) for both ePKs and protein kinase-like aPKs, starting with (i) sequence and structural alignments of all ePKs and aPKs, which is followed by (ii) the spatially conserved kinase folds. Subsequently, the kinase classes are discussed in the light of (iii) genetic data in the form of single-nucleotide polymorphisms (SNPs) and oncogenic mutations along with kinase inhibitor binding to (iv) orthosteric and (v) allosteric sites. Finally, we present examples of (vi) the structural design of selective atypical kinase inhibitors.
Primary Sequence Alignment Shows That Atypical Kinases Lack the Conserved Glycines in the G-Rich Loop and Contain an Inverted HRD Motif The finding that protein structures can be conserved even at low sequence identity [37] certainly holds true for ePKs and most aPKs. Although they are highly divergent on the primary sequence level, at the tertiary level the structures of their catalytic domains are highly conserved, in both the N-lobe (G-rich loop, b-sheets I–IV, aC, gatekeeper, and hinge region) and the C-lobe (aD, aE, b-sheets VI–VIII, catalytic loop, and DFG motif). The first catalytic kinase domain structures were of ePKs [12,38] and revealed the now well-known fold of the catalytic kinase domain (Figure 1). Given that ePKs and aPKs are often coactivated in cancer [39], understanding sequence and structural differences and their consequences for inhibitor binding is important for drug discovery. The conserved glycines in the G-rich loop (commonly referred to as the p-loop for aPKs [9]) are important for the catalytic activity of eukaryotic kinases [13], which is very different for the aPKs as the ePK conserved glycine pattern (GxGxxG) is missing (Figure 2). These variations in the amino acids of the G-rich loop give rise to a selectivity pocket that is present only in aPKs, advancing the selectivity of aPK inhibitors. Despite these variations, the loop fulfills the same function as in ePKs; namely, the positioning and coordination of ATP [5,40]. Another major difference is observed at the HRD motif where the histidine and arginine are mirrored around the aspartate creating the atypical DRH motif (Figure 2). In this case, the histidine rather than the aspartate is suggested to act as a catalytic base during phosphorylation [41,42]. An example of the conformation of the HRD motif in ePKs as well as the DRH motif in aPKs is shown in Figure 2D.
Box 2. Consistent Residue Numbering for the Binding Site Throughout this review, the binding site numbering scheme of the structural kinase database KLIFSi is utilized to enable direct comparison between ePKs and aPKs. This numbering system was introduced as part of the KLIFS databasei [34,35] to facilitate the analysis of interactions between inhibitors and ePKs. To this end, 85 residues shown to interact with inhibitors binding to the cleft between the two lobes were selected (Figure I). For the purposes of this review, the numbering scheme and database were extended to also cover aPKs. The scheme starts with three residues in b-sheet I (positions 1–3) followed by the GxGxxG motif of the G-rich loop (positions 4–9) and is continued with four more residues of b-sheet II (positions 10–13). Then, the scheme cuts to six residues of b-sheet III (positions 14–19), including the VAVK/VAIK motif at positions 14–17. This is followed by 11 residues from the aC-helix (positions 20–30) with the highly conserved glutamate at position 24. The succeeding seven residues (positions 31–37) comprise the back loop (b.l.) connecting to four residues of b-sheet IV (positions 38–41). The next segment goes from b-sheet V (positions 42–44) via the gatekeeper (position 45), hinge (positions 46–48), and linker region (positions 49–52) all the way to the aD-helix (positions 53–59). After a short interval, the scheme is continued with the last turn of the aE-helix (positions 60–64) and moves via b-sheet VI (positions 65–67) and the catalytic loop (c.l.) (positions 68–75) to b-sheet VII (positions 76–78). The HRD motif in the catalytic loop can be found at positions 68–70 and the highly conserved asparagine at position 75. The last segment starts with one residue from b-sheet VIII (position 79) followed by the xDFG motif (positions 80–83) and two additional residues of the activation loop (position 84–85). When referring to a specific residue, the nomenclature is as follows:
4
Trends in Pharmacological Sciences, --, Vol. --, No. --
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
Trends in Pharmacological Sciences
Figure I. The Residue Numbering Scheme Used to Navigate Eukaryotic and Atypical Protein Kinases. The numbering scheme of KLIFSi covers 85 residues known to interact with kinase inhibitors binding to the cleft between the N- and C-lobes. A table with both KLIFS numbering and UniProtix numbering is shown for two eukaryotic kinases (PRKACA and EGFR) and three atypical protein kinases (ATM, mTor, and PIK3CA).
By contrast, the residues KIII.17 and EaC.24, Nc.l.75, and the DFG motif are mostly conserved. KIII.17 and EaC.24 form a salt bridge that is important for maintaining kinase activity and interact with the a- and b-phosphates of ATP [43]. In several aPKs, there is an aspartate at position 24 (DaC.24), which due to its shorter side chain is incapable of forming the ePK-trademark salt bridge with KIII.17 [41]. Moreover, KIII.17 is part of the eukaryotic Val-Ala-Val/Ile-Lys (VAVK/VAIK) motif of which the highly conserved
Trends in Pharmacological Sciences, --, Vol. --, No. --
5
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
(A)
(B)
(C)
Trends in Pharmacological Sciences
Figure 2. Eukaryotic and Atypical Kinases Share Structurally Conserved Binding Sites with Only Some Sequence Motifs Also Being Conserved. (A) Sequence alignment of the binding sites shows the conservation of lysine at position 17, asparagine at position 76, and the DFG motif. The conserved glycines in the G-rich loop (green) are missing in the atypicals and the HRD motif is flipped (orange). The coloring of the backbones and residues throughout the paper is consistent with b-sheets in yellow, a-helices in red, G-rich loop in darker green, back loop (b.l.) in lighter green, gatekeeper in lighter orange, hinge in purple, linker in cyan, catalytic loop in darker orange, and DFG motif in blue. Residues not discussed in detail are colored light gray for clarity purposes. (B) Structural alignment from KLIFSi of the highly conserved binding sites of eukaryotic and atypical protein kinases. (C) An overlay of all orthosterically binding ligands with, in the middle, examples of binding modes for a promiscuous inhibitor (staurosporine) to a eukaryotic and an atypical protein kinase. Vectors showing the direction from the Ca to the Cg atoms of amino acids 12 and 35 show the distinct side-chain orientations in the structures of eukaryotic and atypical protein kinases. In addition, the residues of the flipped HRD/DRH motif in the catalytic loop are shown in an orange stick representation.
AIII.15 together with VII.11 forms the capping residues for the adenine ring of ATP [44]. For the aPKs, this motif is observed only in members of the ABC1/ADCK and the RIO kinase family. For other aPK members (the PIK and PIKK families) either the valine at position II.11 is replaced by a leucine or isoleucine having a
6
Trends in Pharmacological Sciences, --, Vol. --, No. --
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
similar function, or a larger hydrophobic residue instead of an alanine is seen at position III.15 with its side chain protruding into the space normally occupied by II.11 in ePKs [10,44]. A trademark feature of ePKs is a conserved Ala-Pro-Glu (APE) motif at the end of the activation segment that interacts with a conserved arginine in the loop between the aH and aI helices, which secures the activation loop near the aF-helix [45]. This motif is not observed for any of the aPKs [46]. Moreover, the activation segment of aPKs is, similar to more primitive eukaryotic-like protein kinases [45,46], much shorter than for ePKs. Instead, the activation loop of most aPKs (from the PIK and PIKK families) contain a conserved Pro-Phe-any-Leu-Thr (PFxLT) motif (not shown in the sequence conservation in Figure 2 as it lies outside the residue nomenclature scheme; Box 2). The threonine of this motif is known to be phosphorylated in PI3K kinases [47] and thereby marks the end of the activation loop. Moreover, the lysines in this loop have been shown to determine substrate specificity in PI3K kinases [48]. Immediately after the PFxLT motif, the aPKs have a short helix of two turns. This helix together with the activation loop present in most aPKs replaces the whole activation segment and aF-helix present in ePKs [45] (Figure 1).
Atypical Kinases Share a Spatially Conserved Kinase Fold with Eukaryotic Kinases The overlay of the binding sites of the ePKs and aPKs indicates remarkable structural similarity (Figure 2). Despite their overall structural similarity, the shape of the binding sites can differ substantially between ePKs and aPKs. In general, we see that although the average ePK binding site is wider (seen from the aD-helix to the aC-helix), the binding sites of aPKs are on average deeper (from the HRD motif to the hinge) and more open. When looking at the orientation of the side chains of the binding site residues, we find that uniquely for the aPKs the side chains of amino acid 12 are oriented towards the pocket, providing unique interaction opportunities for aPK inhibitors by partly forming the so-called specificity pocket [9] (Figure 2). In these cases, surrounding residues such as II.11 can be flipped thereby enabling the aforementioned protrusion of the capping residue III.15 into the space normally occupied by II.11. Amino acid 38 shows a similar trend, with most side chains in the aPKs being rotated roughly 90 around the backbone compared with the side chains of the ePKs at the same position (Figure 2). When not rotated, orthosterically binding aPK ligands can interact with the amino acid at position 38 in the socalled affinity pocket [9]. This in contrast to the ePKs, where the gatekeeper blocks access to amino acid 38, but due to a shift of the aPK gatekeeper it provides enough space to allow this interaction.
Atypical Kinases Have Altered Hydrophobic Spines Hydrophobic spines form two blocks of hydrophobic and aromatic amino acids inside the catalytic domains that regulate activation (R-spine) as well as priming for catalysis (C-spine) [17,18]. Residue–residue interactions in the Protein Contacts Atlas [49] show, as expected, that most of the DFG-in ePK and aPK structures are involved in highly conserved interaction motifs, with the exception of the amino acids at positions 11 and 15 in the ePKs. However, when ATP is bound it completes the C-spine by interacting via its adenine moiety with these two amino acids. The histidine residue of the HRD motif is one of the most conserved residues in the ePKs, making in 93% of all DFG-in structures contact with the phenylalanine of the DFG motif (position 82). Despite the flipped HRD motif in aPKs, position 68 still interacts with the DFG motif in virtually all aPK structures [49]. While the spines of aPKs in the binding sites seem similar to the ePKs, outside the binding sites substantial differences are observed. PI3K, mTOR, PI4K, ATR, DNAPK, and ATM all miss the entire aF-helix that causes rearrangement of the lower part of the C-spine. However, in all of the mentioned kinases there is an alternative cluster of hydrophobic residues in that region that interacts with the other residues of the C-spine. Only RIOK1 and ADCK3 are similar to the ePKs, as they have an aF-helix with hydrophobic residues at the conserved positions [50]. It should be noted that hydrophobic mutations in or surrounding the hydrophobic spines have been shown to affect kinase regulation with potential oncogenic effects [51].
Trends in Pharmacological Sciences, --, Vol. --, No. --
7
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
Frequently Occurring Mutations and SNPs Are Mutually Exclusive in Location To structurally compare the current landscape of mutations and SNPs in ePKs and aPKs, we investigated publicly available mutation and SNP data from COSMICiii [52], cBioPortaliv [53], and dbSNPv [54]. In total, 104 000 ePK and 16 000 aPK mutations (unique) and 18 000 ePK and 2700 aPK SNPs (unique) are listed in these databases. When mutation frequencies per residue are compared with SNPs, a mutually exclusive pattern was seen both for ePKs and aPKs: conserved regions (i.e., containing low SNP frequencies) show a relatively high occurrence of tumor-driving lesions (Figure 3). This points to a possible essential role of amino acids that are frequently mutated and not (often) compromised by SNPs [55]. Interestingly, SNPs can also be present in and around the inhibitor binding site (Figures 3 and 4), which potentially could directly interfere with the binding of inhibitors. The spatial organization of the mutations on the kinase structure shows that the two most frequent mutations inside the binding site of the ePKs are found at positions 4 (G-rich loop) and 45 (gatekeeper) (Figure 3). Mutations of the first glycine of the G-rich loop (position 4) impact kinase activity [13]. The most frequently observed mutations at that position occur in the EGFR: G719A, G719S, and G719C, with frequencies of 11%, 8%, and 5%, respectively (relative to the most frequent mutation – at position 45). Gatekeeper mutations are responsible for acquired drug resistance in many protein kinases [56]. The gatekeeper mutation T790M in the EGFR is the third most frequently occurring mutation (n = 1300) and is responsible for roughly 50% of acquired EGFR inhibitor resistance [56]. The frequencies of hyperactivating V600K and V600E BRAF mutations located in the activation loop follow directly after T790M [57]. The most frequent SNP variants for ePKs are found at position 40, which lies in b-sheet IV with its side chain pointing away from the binding site and thus does not directly affect the binding site itself. With an exceptionally high allele frequency of almost 50%, the S511N (serine to asparagine) mutation at this position for PAK5 is the most common (n = 2431) missense SNP in the ePK binding site, making PAK5 more similar to PAK4. Similar to ePKs, mutations in the G-rich loop of aPKs at position 4 are also frequently observed (65% relative to the most frequent mutated; i.e., the residue at position 22). Gatekeeper mutations are seldom observed in the aPKs (Figure 3). Frequently mutated positions in the aPKs are: position 22 (n = 40), position 32 (80% compared with position 22), and position 69 (70%). In PI3K, residue 22 (RaC.22) acts as hydrogen bond donor to the surrounding of E702 and Q705 thereby further stabilizing D836 and KIII.17 [58]. The arginine of the HRD motif (or rather, the DRH motif for aPKs) is involved in aligning the phosphorylated activation segment for substrate binding [59,60]. Point mutation of this arginine was previously found to significantly reduce kinase activity in yeast PKA [61]. The most frequently occurring mutations at that position are R743Q in EPHB1, R836C in EGFR, and G914R in PIK3CA. The most frequent missense SNP in the binding site of aPKs is observed at position 35 for RIOK2. This V175I (valine to isoleucine) mutation is located deep in the back loop and thereby only affects the back pocket of this kinase. As shown in the ranking of most frequently mutated kinases, the closely related ERBB4 and EGFR are the most mutated [62,63]. Among the aPKs, TRRAP and ATM have the highest numbers of unique mutations (1132 and 1130, respectively). ATM has the highest number of mutations inside the binding domain (145). The notable aPKs mTOR and PIK3CA are in the top ten with 744 and 714 mutations, respectively. PIK3CA has the highest number of mutations in the kinase fold (239), some of which are frequently found in cancers [28,64]. The most frequently occurring mutations in PIK3CA are found at positions 1047 (i.e., H1047R and H1047L) and 545 (i.e., E545K and E545A). Although these positions are outside the binding site [64], these mutants result in increased catalytic activity and the ability to induce tumors [65,66].
Atypical Kinase Inhibitors Explore a Limited Number of Binding Modes Analyzing the inhibitor binding modes of the ePKs and aPKs revealed that the inhibitors target a much wider portion of the binding site of ePKs (including the front cleft, the gatekeeper region, the back cleft, and even multiple sites outside the catalytic domain) than aPKs. All inhibitors currently solved
8
Trends in Pharmacological Sciences, --, Vol. --, No. --
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
(A)
(B)
(C) Mutations
Trends in Pharmacological Sciences
Figure 3. Differences in Mutation and SNP Patterns in Eukaryotic and Atypical Protein Kinases. (A) Schematic representations of the mutations and SNPs inside the binding site of the eukaryotic and atypical protein kinases and their frequencies. The gatekeeper (position 45) is the most frequently mutated in eukaryotic protein kinases while position 22 is the most frequently mutated in atypical protein kinases. The highest occurrences of SNPs are found at positions 40 and 35 in eukaryotic and atypical kinases, respectively. Scales are adjusted to the maximal numbers of unique mutations/SNPs. (B). Scatter plots of SNP versus mutation frequencies show mutually exclusive patterns for frequently occurring positions (i.e., high SNP rate and low mutation rate and vice versa). (C) Top ten most-mutated genes with their numbers of mutations and SNPs in the full protein and in the binding site. Unique mutations (e.g., H1047R and H1047L in PIK3CA) at the same position in a kinase were counted individually.
in aPK structures belong to either type I, which targets exclusively the front cleft (the same binding site as ATP) inside the binding site (Figure 4), or type I1/2, which targets both the front cleft and part of the back cleft in DFG-in conformations [4]. To the best of our knowledge, there are currently no known type II atypical inhibitors and also no aPK structures in the DFG-out conformation. Also, there are currently no known atypical type III inhibitors (i.e., inhibitors that are non-ATP competitive and exclusively target the back cleft or adjacent pockets).
Trends in Pharmacological Sciences, --, Vol. --, No. --
9
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
(A)
(B) Gatekeeper
(C)
Mutations
Mutations
(D)
Trends in Pharmacological Sciences
Figure 4. Limited Binding Mode Diversity for Orthosteric Atypical Kinase Inhibitors Compared with Eukaryotic Kinase Inhibitors. (A) A schematic overview and examples of type I, I1/2, II, and III inhibitors and their binding (sub)pockets. (B) The surface of all type I eukaryotic kinase inhibitors combined compared with atypical kinase inhibitors combined. (C). The combined inhibitor surface of type I1/2 eukaryotic and atypical kinase inhibitors. Currently, inhibitors of types II and III occur only in eukaryotic protein kinases. (D) The number of interactions per residue for each of the four different types of eukaryotic and atypical kinase inhibitors and the number of mutations/SNPs per position (expressed as % in relation to the highest mutated position). The color coding (yellow, green, red, orange, purple, cyan, and blue) under the residue numbers corresponds to segments of the binding site as defined in KLIFSi (Figure 2).
10
Trends in Pharmacological Sciences, --, Vol. --, No. --
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
Type I protein kinase inhibitors represent the largest group of inhibitors targeting this superfamily. These inhibitors are ATP competitive and exclusively target the front pocket using hydrogen bonds with the hinge region as an anchor point (Figure 4) [4,35]. In addition, the amino acids II.12, KIII.17, IV.38, g.k.45, linker.52, c.l.74, and DxDFG.81 commonly interact with this type, of which II.12 and IV.38 interact predominantly with inhibitors of the aPKs. The amino acids II.12 and IV.38 are involved in the formation of both the specificity pocket in the front cleft (positions 4 and 12) [9] and the affinity pocket (formed by the residues at positions 27, 24, and 38) in the aPKs [9]. A common challenge for type I inhibitors is obtaining selectivity for a specific (set of) kinase(s), primarily due to the highly conserved adenine pocket [67]. The selectivity could be improved by targeting distinct subpockets or nonconserved amino acids. One noticeable example among the eukaryotic inhibitors is CAM4066 (HET: 551) that targets a cryptic pocket of CK2a1 between the aD and the catalytic loop [68]. This compound and its derivatives are the only crystallized inhibitors that interact with the amino acid at position 73 [69]. Idelalisib (HET: 40L), the first aPK inhibitor approved by the FDA (2014; Box 1), is an ATP-competitive inhibitor of which the purine moiety makes a hydrogen bond with the hinge region of PI3Kd while the fluoroquinazolinone moiety penetrates the specificity pocket [70].
Gaining Selectivity in the Back Pocket Efforts to gain selectivity were also made by exploring the back pockets of the protein kinases. A variant of the type II kinase inhibitor STI-571 [71] was extended with a pyrazine moiety to reach deep into the back pocket [72] and became imatinib, the first FDA-approved small-molecule kinase inhibitor (Box 1) [73]. This inhibitor makes the common type II bonds with residues 67 and 68. In addition, type II inhibitors also make bonds with amino acids 31, 35, and 66. Between type I and type II, there is a type of inhibitor, type I1/2, that binds to both the front and the back like type II, but unlike this type, binds to the kinases in DFG-in conformation. An example of aPK type I1/2 is a pyrazolopyrimidine inhibitor [74], which is over 600 times more selective for mTOR than PI3Ka [74]. In the aPKs, it is only this inhibitor type that makes interactions with amino acid 24 (Figure 4) as the affinity pocket reaches the back of the protein kinases. In the ePKs, these inhibitors make selective bonds with residues 21, 28, and 38. The non-ATP-competitive type III inhibitors bind only to the back cleft and are not currently known for aPKs. Type III examples for ePK include inhibitor A1 [75] and a dibenzodiazepine inhibitor [76]. The latter makes an aromatic interaction with amino acid 60 that is observed only among this type of inhibitor. While the above-mentioned type I, I1/2, II, and III all directly target the catalytic cleft or an adjacent pocket, type IV inhibitors bind to allosteric sites away from this cleft [4]. In the current kinase structures, eight different type IV allosteric sites in ePKs and four in aPKs have been targeted (Figure 5). Of the eight ePK sites, five target the C-lobe and the remaining three the N-lobe [15]. A similar trend is seen for the aPKs, with three sites in the C-lobe and only one site in the N-lobe [77,78]. Currently, there are 250 Protein Data Bankvi (PDB) structures solved with type IV inhibitors, comprising 165 unique eukaryotic and 17 atypical type IV kinase inhibitors.
Selectivity Hotspots in the PI3K Family Although selectivity is hard to achieve, especially within very closely related kinases, it can be achieved by exploiting the correct hotspots within each kinase. The same goes for the PI3K family, where most inhibitors inhibit all four subtypes. Despite this challenge, selectivity hotspots for each of the PI3K kinases have been found and four examples are given in Figure 5, one for each subtype. The inhibitors belong to type I and type I1/2 and thus all make hydrogen bond(s) with the hinge region at amino acid 48 (Vhinge.48). Alpelisib (Figure 5), a highly selective PI3Ka inhibitor, makes hydrogen bonds with Slinker.50 and QaD.55 and a face-to-edge aromatic interaction with YIV.38 [79]. This compound is up to 50-fold more selective for PIK3Ca and its E545K and H1047R mutants over the rest of the PI3K (b/d/g) isoforms [80]. The selectivity driver of this inhibitor is the unique bidentate hydrogen bond with Q859aD.55, as this amino acid is not observed in the other subtypes [81]. SAR260301 makes hydrophobic contact with KI.2, a face-to-edge interaction with WII.12, and a hydrogen bond with the aspartate of the DFG motif. Despite having a binding mode similar to that of INK666,
Trends in Pharmacological Sciences, --, Vol. --, No. --
11
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
Figure 5. Gaining Selectivity by Targeting Distant Allosteric Pocket or Utilizing Selectivity Hotspots. (A) Type IV inhibitors: targeting allosteric binding pockets away from the ATP pocket for eukaryotic and atypical kinase inhibitors. One example of an allosteric ligand is shown in stick representation and a colored molecular surface per allosteric pocket. A few selected ligands are highlighted in detail by their 2D chemical structure. (B) Orthosteric atypical inhibitors gaining selectivity by interacting with selective pockets and/or not-conserved amino acids. At the top, four examples of binding modes of isoform-selective PI3K inhibitors and their interactions.
this compound is over 20 and 70 times more active for PIK3CB than for PIK3CD and PIK3CA and has no observed affinity for PI3KCG [82]. Its selectivity was reasoned to come from the KI.2 that restricts the selectivity pocket while its carbon atoms also provide a hydrophobic outline for the indoline moiety. INK666 also targets the selectivity and affinity pockets in PIK3CD, but makes use of the smaller threonine instead of a lysine at position 2 allowing a different angle and bulkier substituents. Additionally, INK666 makes a hydrogen bond with the aC-helix and a unique ionic interaction with the Dlinker.51 in the linker region [9]. This inhibitor is over 270 times more active for PIK3CD than PIK3CA and between
12
Trends in Pharmacological Sciences, --, Vol. --, No. --
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
ten and 20 times over PIK3CG and PIK3C. Another notable inhibitor is 2WH [83]. This inhibitor is over 200 times more selective for PI3Kg than the other isoforms [83] due to the small hydrophobic Alinker.50. In all other isoforms, the slightly bigger and polar serine at the same position prevents the inhibitor from assuming this binding mode.
Concluding Remarks By comprehensively reviewing all available human eukaryotic and atypical kinase structures, we provide a complete overview to allow the design of a next generation of kinase inhibitors with, for instance, optimal polypharmacological profiles that might be able to overcome drug resistance. Structurally, the eukaryotic and atypical kinases are highly similar with notable differences at the sequence level. This information enables matching of the unique features of small molecules to obtain an optimal selectivity profile. The mutation and SNP data revealed important hotspots and their location in the protein kinase fold. Moreover, when mutation frequencies per position are compared with SNPs, a mutually exclusive pattern was seen for both ePKs and aPKs, which is likely to reflect evolutionary cost–benefit relations. This integrated kinase review merging the two primary kinase classes provides clues for next-generation personalized medicine, for which several hurdles have yet to be overcome (see Outstanding Questions).
Acknowledgments This work is supported by a grant of the Amsterdam Academic Alliance (Amsterdam Data Science) as provided by the Boards of the VU and UvA universities, and by the Cancer Center Amsterdam - project CCA2018-2-19.
Outstanding Questions Can we use structural data to generate new lead compounds with desired bioactivity profiles? Is it possible to target deep into the back pockets of the atypical kinases and how would that affect the selectivity profile? Can desired combinations of eukaryotic and atypical kinases be simultaneously targeted with a single inhibitor? What would be the best approach to translate the wealth of kinase crystal structures into de novo predictions of bioactivity profiles for existing compounds? Does simultaneous targeting of atypical and eukaryotic protein kinase mutations prevent therapy resistance? Can therapy resistance through pathway redundancy/compensation be simultaneously prevented by polypharmacological drugs?
Resources i
http://klifs.vu-compmedchem.nl www.kinase.com/web/current/kinbase/ iii https://cancer.sanger.ac.uk/cosmic iv www.cbioportal.org v www.ncbi.nlm.nih.gov/snp/ vi www.rcsb.org vii www.icoa.fr/pkidb/ viii www.who.int/medicines/services/inn/stembook/en/ ix https://uniprot.org ii
Which of the existing drugs inhibit tumor driving mutations as well as resistance mechanisms? How can we optimally match the bioactivity profiles of single drugs for maximal therapy effects based on the patient’s target profile?
References 1. Cohen, P. (2002) The origins of protein phosphorylation. Nat. Cell Biol. 4, E127 2. Heath, C.M. et al. (2003) Lipid kinases play crucial and multiple roles in membrane trafficking and signaling. Histol. Histopathol. 18, 989–998 3. Manning, G. et al. (2002) The protein kinase complement of the human genome. Science 298, 1912–1934 4. Roskoski, R., Jr. (2015) A historical overview of protein kinases and their targeted small molecule inhibitors. Pharmacol. Res. 100, 1–23 5. Walker, E.H. et al. (1999) Structural insights into phosphoinositide 3-kinase catalysis and signalling. Nature 402, 313 6. Yamaguchi, H. et al. (2001) Crystal structure of the atypical protein kinase domain of a TRP channel with phosphotransferase activity. Mol. Cell 7, 1047–1057 7. Scheeff, E.D. and Bourne, P.E. (2005) Structural evolution of the protein kinase-like superfamily. PLoS Comput. Biol. 1, e49 8. Heffron, T.P. et al. (2016) The rational design of selective benzoxazepin inhibitors of the alphaisoform of phosphoinositide 3-kinase culminating in the identification of (S)-2-((2-(1-isopropyl-1H-1,2,4-
9. 10. 11. 12.
13. 14. 15. 16.
triazol-5-yl)-5,6-dihydrobenzo[f]imidazo[1,2-d][1,4] oxazepin-9-yl)oxy)propanamide (GDC-0326). J. Med. Chem. 59, 985–1002 Berndt, A. et al. (2010) The p110 delta structure: mechanisms for selectivity and potency of new PI3K inhibitors. Nat. Chem. Biol. 6, 117–124 Yang, H. et al. (2013) mTOR kinase structure, mechanism and regulation. Nature 497, 217–223 Stefely, J.A. et al. (2015) Mitochondrial ADCK3 employs an atypical protein kinase-like fold to enable coenzyme Q biosynthesis. Mol. Cell 57, 83–94 Knighton, D. et al. (1991) Crystal structure of the catalytic subunit of cyclic adenosine monophosphate-dependent protein kinase. Science 253, 407–414 Hemmer, W. et al. (1997) Role of the glycine triad in the ATP-binding site of cAMP-dependent protein kinase. J. Biol. Chem. 272, 16946–16954 Greenman, C. et al. (2007) Patterns of somatic mutation in human cancer genomes. Nature 446, 153–158 Wu, P. et al. (2015) Allosteric small-molecule kinase inhibitors. Pharmacol. Ther. 156, 59–68 Kornev, A.P. et al. (2006) Surface comparison of active and inactive protein kinases identifies a conserved
Trends in Pharmacological Sciences, --, Vol. --, No. --
13
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
17. 18. 19.
20. 21. 22.
23. 24. 25. 26.
27. 28. 29. 30. 31.
32. 33. 34. 35. 36.
37.
38. 39.
14
activation mechanism. Proc. Natl Acad. Sci. U. S. A. 103, 17783–17788 Kornev, A.P. et al. (2008) A helix scaffold for the assembly of active protein kinases. Proc. Natl Acad. Sci. U. S. A. 105, 14377–14382 Taylor, S.S. and Kornev, A.P. (2011) Protein kinases: evolution of dynamic regulatory proteins. Trends Biochem. Sci. 36, 65–77 Adams, J.A. (2003) Activation loop phosphorylation and catalysis in protein kinases: is there functional evidence for the autoinhibitor model? Biochemistry 42, 601–607 Ramos, P. and Bentires-Alj, M. (2015) Mechanismbased cancer therapy: resistance to therapy, therapy for resistance. Oncogene 34, 3617–3626 Holohan, C. et al. (2013) Cancer drug resistance: an evolving paradigm. Nat. Rev. Cancer 13, 714–726 Yun, C.H. et al. (2008) The T790M mutation in EGFR kinase causes drug resistance by increasing the affinity for ATP. Proc. Natl Acad. Sci. U. S. A. 105, 2070–2075 Davies, H. et al. (2002) Mutations of the BRAF gene in human cancer. Nature 417, 949 Chapman, P.B. et al. (2011) Improved survival with vemurafenib in melanoma with BRAF V600E mutation. N. Engl. J. Med. 364, 2507–2516 Knight, Z.A. et al. (2010) Targeting the cancer kinome through polypharmacology. Nat. Rev. Cancer 10, 130–137 Mir, S.E. et al. (2010) In silico analysis of kinase expression identifies WEE1 as a gatekeeper against mitotic catastrophe in glioblastoma. Cancer Cell 18, 244–257 Fruman, D.A. et al. (2017) The PI3K pathway in human disease. Cell 170, 605–635 Samuels, Y. et al. (2004) High frequency of mutations of the PIK3CA gene in human cancers. Science 304, 554 Thorpe, L.M. et al. (2015) PI3K in cancer: divergent roles of isoforms, modes of activation and therapeutic targeting. Nat. Rev. Cancer 15, 7–24 Mendoza, M.C. et al. (2011) The Ras–ERK and PI3K– mTOR pathways: cross-talk and compensation. Trends Biochem. Sci. 36, 320–328 Engelman, J.A. et al. (2008) Effective use of PI3K and MEK inhibitors to treat mutant Kras G12D and PIK3CA H1047R murine lung cancers. Nat. Med. 14, 1351–1356 LoPiccolo, J. et al. (2008) Targeting the PI3K/Akt/ mTOR pathway: effective combinations and clinical considerations. Drug Resist. Updat. 11, 32–50 Sanchez-Vega, F. et al. (2018) Oncogenic signaling pathways in The Cancer Genome Atlas. Cell 173, 321–337 Kooistra, A.J. et al. (2016) KLIFS: a structural kinase– ligand interaction database. Nucleic Acids Res. 44, D365–D371 van Linden, O.P. et al. (2014) KLIFS: a knowledgebased structural database to navigate kinase–ligand interaction space. J. Med. Chem. 57, 249–277 Kanev, G.K. et al. (2017) Structural chemogenomics databases to navigate protein– ligand interaction space. In Comprehensive Medicinal Chemistry III, S. Chackalamannil et al., eds. (Elsevier), pp. 444–471 Sander, C. and Schneider, R. (1991) Database of homology-derived protein structures and the structural meaning of sequence alignment. Proteins 9, 56–68 Hubbard, S.R. et al. (1994) Crystal structure of the tyrosine kinase domain of the human insulin receptor. Nature 372, 746 Jaeger, S. et al. (2017) Quantification of pathway cross-talk reveals novel synergistic drug
40. 41. 42. 43. 44. 45.
46.
47.
48.
49. 50.
51. 52. 53. 54. 55. 56. 57. 58.
59.
60. 61.
combinations for breast cancer. Cancer Res. 77, 459–469 Baretic, D. and Williams, R.L. (2014) The structural basis for mTOR function. Semin. Cell Dev. Biol. 36, 91–101 Vadas, O. et al. (2011) Structural basis for activation and inhibition of class I phosphoinositide 3-kinases. Sci. Signal. 4, re2 Miller, S. et al. (2010) Shaping development of autophagy inhibitors with the structure of the lipid kinase Vps34. Science 327, 1638–1642 Endicott, J.A. et al. (2012) The structural basis for control of eukaryotic protein kinases. Annu. Rev. Biochem. 81, 587–613 Taylor, S.S. et al. (2019) Evolution of a dynamic molecular switch. IUBMB Life 71, 672–684 Taylor, S.S. et al. (2012) Evolution of the eukaryotic protein kinases as dynamic molecular switches. Philos. Trans. R. Soc. Lond. B Biol. Sci. 367, 2517–2528 Kannan, N. and Neuwald, A.F. (2005) Did protein kinase regulatory mechanisms evolve through elaboration of a simple structural component? J. Mol. Biol. 351, 956–972 Wang, B.D. et al. (2017) Alternative splicing promotes tumour aggressiveness and drug resistance in African American prostate cancer. Nat. Commun. 8, 15921 Pirola, L. et al. (2001) Activation loop sequences confer substrate specificity to phosphoinositide 3kinase alpha (PI3Ka). Functions of lipid kinasedeficient PI3Ka in signaling. J. Biol. Chem. 276, 21544–21554 Kayikci, M. et al. (2018) Visualization and analysis of non-covalent contacts using the Protein Contacts Atlas. Nat. Struct. Mol. Biol. 25, 185–194 Coun˜ago, R.M. et al. (2017) Development of narrow spectrum ATP-competitive kinase inhibitors as probes for BIKE and AAK1. bioRxiv. Published online March 2, 2017. https://doi.org/10.1101/094631 Hu, J. et al. (2015) Kinase regulation by hydrophobic spine assembly in cancer. Mol. Cell. Biol. 35, 264–276 Forbes, S.A. et al. (2017) COSMIC: somatic cancer genetics at high-resolution. Nucleic Acids Res. 45, D777–D783 Gao, J. et al. (2013) Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci. Signal. 6, pl1 Sherry, S.T. et al. (2001) dbSNP: the NCBI database of genetic variation. Nucleic Acids Res. 29, 308–311 Ikediobi, O.N. et al. (2006) Mutation analysis of 24 known cancer genes in the NCI-60 cell line set. Mol. Cancer Ther. 5, 2606–2612 Sequist, L.V. et al. (2011) Genotypic and histological evolution of lung cancers acquiring resistance to EGFR inhibitors. Sci. Transl. Med. 3, 75ra26 Karoulia, Z. et al. (2017) New perspectives for targeting RAF kinase in human cancer. Nat. Rev. Cancer 17, 676–691 Carter, H. et al. (2014) Prioritization of driver mutations in pancreatic cancer using cancer-specific high-throughput annotation of somatic mutations (CHASM). Cancer Biol. Ther. 10, 582–587 Krupa, A. et al. (2004) Structural modes of stabilization of permissive phosphorylation sites in protein kinases: distinct strategies in Ser/Thr and Tyr kinases. J. Mol. Biol. 339, 1025–1039 Johnson, L.N. et al. (1996) Active and inactive protein kinases: structural basis for regulation. Cell 85, 149–158 Gibbs, C.S. and Zoller, M.J. (1991) Rational scanning mutagenesis of a protein kinase identifies functional regions involved in catalysis and substrate interactions. J. Biol. Chem. 266, 8923–8931
Trends in Pharmacological Sciences, --, Vol. --, No. --
Please cite this article in press as: Kanev et al., The Landscape of Atypical and Eukaryotic Protein Kinases, Trends in Pharmacological Sciences (2019), https://doi.org/10.1016/j.tips.2019.09.002
Trends in Pharmacological Sciences
62. Hynes, N.E. and Lane, H.A. (2005) ERBB receptors and cancer: the complexity of targeted inhibitors. Nat. Rev. Cancer 5, 341–354 63. Gazdar, A.F. (2009) Activating and resistance mutations of EGFR in non-small-cell lung cancer: role in clinical response to EGFR tyrosine kinase inhibitors. Oncogene 28, S24 64. Liu, S. et al. (2014) The structural basis of PI3K cancer mutations: from mechanism to therapy. Cancer Res. 74, 641–646 65. Kang, S. et al. (2005) Phosphatidylinositol 3-kinase mutations identified in human cancer are oncogenic. Proc. Natl. Acad. Sci. U. S. A. 102, 802–807 66. Meyer, D.S. et al. (2013) Expression of PIK3CA mutant E545K in the mammary gland induces heterogeneous tumors but is less potent than mutant H1047R. Oncogenesis 2, e74 67. Morphy, R. (2010) Selectively nonselective kinase inhibition: striking the right balance. J. Med. Chem. 53, 1413–1437 68. De Fusco, C. et al. (2017) A fragment-based approach leading to the discovery of a novel binding site and the selective CK2 inhibitor CAM4066. Bioorg. Med. Chem. 25, 3471–3482 69. Brear, P. et al. (2016) Specific inhibition of CK2a from an anchor outside the active site. Chem. Sci. 7, 6839– 6845 70. Somoza, J.R. et al. (2015) Structural, biochemical, and biophysical characterization of idelalisib binding to phosphoinositide 3-kinase delta. J. Biol. Chem. 290, 8439–8446 71. Schindler, T. et al. (2000) Structural mechanism for STI-571 inhibition of Abelson tyrosine kinase. Science 289, 1938–1942 72. Nagar, B. et al. (2002) Crystal structures of the kinase domain of c-Abl in complex with the small molecule inhibitors PD173955 and imatinib (STI-571). Cancer Res 62, 4236–4243 73. Dagher, R. et al. (2002) Approval summary: imatinib mesylate in the treatment of metastatic and/or unresectable malignant gastrointestinal stromal tumors. Clin. Cancer Res. 8, 3034–3038
74. Zask, A. et al. (2009) ATP-competitive inhibitors of the mammalian target of rapamycin: design and synthesis of highly potent and selective pyrazolopyrimidines. J. Med. Chem. 52, 5013–5016 75. Furuya, N. et al. (2017) The juxtamembrane region of TrkA kinase is critical for inhibitor selectivity. Bioorg. Med. Chem. Lett. 27, 1233–1236 76. Karpov, A.S. et al. (2015) Optimization of a dibenzodiazepine hit to a potent and selective allosteric PAK1 inhibitor. ACS Med. Chem. Lett. 6, 776–781 77. Miller, M.S. et al. (2017) Identification of allosteric binding sites for PI3Ka oncogenic mutant specific inhibitor design. Bioorg. Med. Chem. 25, 1481– 1486 78. Hon, W.C. et al. (2011) Regulation of lipid binding underlies the activation mechanism of class IA PI3kinases. Oncogene 31, 3655 79. Furet, P. et al. (2013) Discovery of NVP-BYL719 a potent and selective phosphatidylinositol-3 kinase alpha inhibitor selected for clinical evaluation. Bioorg. Med. Chem. Lett. 23, 3741–3748 80. Fritsch, C. et al. (2014) Characterization of the novel and specific PI3Ka inhibitor NVP-BYL719 and development of the patient stratification strategy for clinical trials. Mol. Cancer Ther. 13, 1117–1129 81. Frazzetto, M. et al. (2008) Dissecting isoform selectivity of PI3K inhibitors: the role of nonconserved residues in the catalytic pocket. Biochem. J. 414, 383–390 82. Certal, V. et al. (2014) Discovery and optimization of pyrimidone indoline amide PI3Kb inhibitors for the treatment of phosphatase and tensin homologue (PTEN)-deficient cancers. J. Med. Chem. 57, 903–920 83. Collier, P.N. et al. (2015) Structural basis for isoform selectivity in a class of benzothiazole inhibitors of phosphoinositide 3-kinase gamma. J. Med. Chem. 58, 517–521 84. Carles, F. et al. (2018) PKIDB: a curated, annotated and updated database of protein kinase inhibitors in clinical trials. Molecules 23, E980
Trends in Pharmacological Sciences, --, Vol. --, No. --
15