In recent years, protein kinases have been one of the most pursued drug targets. These determined efforts have resulted in ever increasing numbers of small-molecule kinase inhibitors reaching to the market, offering novel treatment options for patients with distinct diseases. One essential component related to the activation and normal functionality of a protein kinase is the regulatory spine (R-spine). The R-spine is formed of four conserved residues named as RS1–RS4. One of these residues, RS3, located in the C-terminal part of αC-helix, is usually accessible for the inhibitors from the ATP-binding cavity as its side chain is lining the hydrophobic back pocket in many protein kinases. Although the role of RS3 has been well acknowledged in protein kinase function, this residue has not been actively considered in inhibitor design, even though many small-molecule kinase inhibitors display interactions to this residue. In this minireview, we will cover the current knowledge of RS3, its relationship with the gatekeeper, and the role of RS3 in kinase inhibitor interactions. Finally, we comment on the future perspectives how this residue could be utilized in the kinase inhibitor design.

Protein kinases are dynamic proteins that regulate a multitude of cellular signalling processes. They control the activity of their downstream targets mainly by phosphorylation, and their own activity is also usually controlled in the same manner. The human kinome comprises more than 500 protein kinases [1], and nearly 500 proteins contain a typical kinase domain [2]. Still, the biological role of many protein kinases is largely unknown, and there are ongoing efforts aiming to characterize these poorly understood protein kinases [3]. Although protein kinases display high similarity in their kinase domain, there is a higher level diversity in their structures; while some kinases consist (almost) solely of the kinase domain (e.g. MAPK14, GSK3B), other are larger multidomain proteins (e.g. LRRK2 [4]). The structure and function of the protein kinase domain is well-established (Figure 1A). For a comprehensive view of structural history of protein kinases, the reader is recommend a recent review by Taylor et al. [5].

In the protein kinase domain, one of the key dynamic elements in regulating protein kinase function is the hydrophobic regulatory-spine (R-spine), which was discovered already 15 years ago in 2006 [6]. The R-spine consists of four residues, named RS1–RS4, which connects the two lobes of the kinase domain (Figure 1A,B). Two of these residues, RS1 and RS2, belong the C-lobe. RS1 is His (sometimes Tyr) residue from the HRD (or YRD) motif [7]. RS2 is Phe (or Leu) from the DFG-motif, which is part of the activation loop of a protein kinase. The other two R-spine residues belong to the N-lobe of the protein kinase. RS4 is a residue from the β4-strand, which is less conserved but frequently Leu can be found in this position. Finally, RS3 is located four residues C-terminal from the αC-helix Glu that forms a salt-bridge to the Lys of β3-sheet. RS3 is usually (not always) accessible from the ATP binding site as its side chain lines the active site cleft. Overall, αC-helix, where RS3 is located, has a central role in the kinase activation process [8]. In the catalytically active form of a protein kinase, R-spine is assembled and in the inactive state it is disassembled (Figure 1C). In the active state the location of RS2 as part of the assembled R-spine results in an open and extended conformation of the activation loop (A-loop), while a closed A-loop configuration is preferred in the inactive state. Notably, additional stabilization of the R-spine, such as via in-frame insertions or RS3 mutations [9,10], may result in increased catalytic activity of the protein kinase.

Next to the R-spine in the N-lobe are located three conserved residues, named as Shell (SH) residues (Figure 1C) [9]. These residues, which are usually hydrophobic, have a role in supporting R-spine and are therefore important for kinase activity. One of these residues, SH2, is found close to RS3. This SH2 residue is more commonly known as the gatekeeper residue, which is named due to its role in controlling access to the hydrophobic pocket [11]. This shell residue participates in regulating R-spine dynamics, and gatekeeper mutations may stabilize the R-spine promoting the kinase activation [12].

In recent years, ever increasing efforts have been conducted by the pharmaceutical industry to target protein kinases [13]. These efforts have resulted in numerous small-molecule kinase inhibitors, totalling now over 70 FDA approved small-molecule inhibitors (Figure 1D). According to the Protein Kinase Inhibitors in Clinical Trials database (PKIDB) [14,15], approximately 300 small-molecule kinase inhibitors are either in clinical trials or already approved. Comprehensive reviews of the kinase inhibitor drug discovery and kinase inhibitor development are available [16–18]. Currently, oncology is dominating indication for the kinase inhibitors, but there is potential also in other therapeutic areas such as autoimmune and inflammatory diseases, and degenerative disorders [19].

Here, we review the characteristics of RS3 and its relationship with the neighbouring gatekeeper (SH2) based on the publicly available structural data. We also have a look at RS3 interactions to small-molecule kinase inhibitors, including the approved drugs. Finally, we end the review with available mutational data of RS3.

A majority of the human protein kinases with publicly disclosed structures display a nonpolar aliphatic RS3 residue (Figure 2). Nearly half of these kinases exhibit Leu in RS3, and almost a third have Met (Figure 2A). In the overall human proteome Leu is also the most abundant residue (9.97%), while Met has the second lowest frequency (2.13%) of all amino acids [23]. Following the abundant Leu and Met in RS3, next preferred are aromatic residues. Tyr, His and Phe appear with the frequencies of 3–6%. Cys and Gln exist in 2%, Ile and Ser in 1%. Even more rare residues that are observed in this location are Val, Thr, Asn and Ala. The charged residues, Asp, Glu, Lys and Arg, as well as structurally more unique Trp, Gly or Pro are not present in the analysed set of human protein kinases in RS3. Based on the sequence alignment of a larger set of eucaryotic protein kinases these residues have been suggested to exist as RS3 residues, although rarely [7]. Regardless of the high existence of hydrophobic residues in RS3, there exists no clear trend related to the hydrophobicity ranking of the residues and their observed frequencies [24,25].

Protein kinases of different groups and families display distinct preferences for RS3 residues (Figure 2B). The majority of the kinases (72%) belonging to the TK group display Met RS3. More than a fifth (22%) of this group present Leu in this position, including protein kinases belonging to JakA family (JAK1–3; TYK2) and Trk family (NTRK1–3 also known as TRKA–C). Four protein kinases have either Ile (ALK; ERBB3 (ErbB3)) or a Phe (PTK7 (CCK4); LMTK3 (LMR3)). In EGFR family Met is preferred in RS3, except ErbB3 has Ile. Interestingly, ErbB3 has been identified to display considerably lower kinase activity [26,27]. However, ErbB3 displays also other unique characteristics that differ from other EGFRs (for instance, instead of αC-Glu ErbB3 has a His in this location). In the PDGFR family, KIT displays Leu instead of Met that is observed in other family members (FLT, PDGFRA, CSF1R (FMS)).

In contrast with the TKs, in the CMGC group Leu is clearly dominating RS3 (90%). Only three protein kinases have Met in RS3 (MAPK8 (JNK1); MAPK10(JNK3); GSK3B), and MAPK6 (Erk3) displays Ile in this position.

The TKLs prefer quite diversely Leu, Met, Tyr and Phe in their RS3. Members of the STKR family prefer an aromatic residue in this position: Tyr is observed in ACVR1 (ALK2), ACVR2A (ACTR2), ACVRL1 (ALK1), BMBR1B, BMBR2 and TGFBR1 (TGFbR1); Phe in ACVR2B (ACTR2B) and TGFBR2 (TGFbR2). Aromatic Phe is also present in two MLK family members: MAP3K9 (MLK1) and MAP3K21 (MLK4), while other kinases in this family display Leu (ILK; MAP3K12 (DLK); MAP3K20 (ZAK); MAP3K7 (TAK1); TNNI3K (HH498)).

Kinases of the STE group display mainly Leu and Met in RS3. However, also other RS3 residues are observed in this group, including His (STE11) and Ser (STRADA (STLK5); MAP2K6), Gln (MAP3K8 (COT)), Cys (MAP3K14 (NIK)) and Val (MAP2K4; MAP2K7). From these, MAP3K14 (NIK) and MAP3K8 (COT) belong to the STE-Unique family. Three members of the STE7 family display rare RS3 residues. These kinases are MAP2K6 (Ser), MAP2K4 also known as MKK4 (Val) and MAP2K7 also known as MKK7 (Val). Interestingly, unique autoinhibited conformational states have been reported for these kinases state [30–32]; with MKK4 this state may be related to its dimer form [33].

In the CK1 group, Tyr is dominating in RS3 (73%). VRK3 displays an aromatic Phe in this position. TTBK family kinases (TTBK1, TTBK2) are more diverse in this group with their aliphatic Leu in RS3.

The majority of the AGC and CAMK group kinases exhibit Leu in RS3, which is followed by Met with lower frequencies. Two kinases with an aromatic RS3 (Phe) are observed in AGC group, in PKN2 and PRKCI (PKCi). In CAMK group, kinases of the CAMK2 family (CAMK2A, CAMK2B, CAMK2D, CAMK2G) display Cys in RS3, as well as CASK. His is observed in MAPKAP family (MAPKAP2, MAPKAP3). TRIB1 (Trb1) represents Ile in RS3; however, in this pseudokinase the neighbouring Tyr may actually occupy the canonical RS3 position [34,35].

The protein kinases that are not belonging to any specific group display also family specific preferences. For instance, protein kinases of WEE family and PLK family exhibit His in RS3 (located above and below CK1 group in the kinome tree). Of the Atypical kinases, COQ8A that is also known as ADCK3 [36], is the only structure in the dataset that displays Ala in RS3.

Not only hydrophobic RS3 residues exist, but also polar residues are observed in this position. AURKA is an example of a widely studied kinase that has a polar RS3 (Gln) is. It was disclosed by Levinson et al. that this polar residue has a specific role in AURKA activation via a water-network [37]. Similarly, AURKB and AURKB have also Gln as their RS3. In addition, Gln is observed in MAP3K8 (COT, TPL2) and ATM. MAP3K8 controls inflammation [38] and ATR DNA damage responses [39].

In the available data, Ser is observed in three kinases. While in STRADA (STLK5, STE20) this residue appears unreachable from the binding cleft (PDB ID: 3gni [40], 2wtk [41]), in Haspin Ser is accessible (participates in water coordination next to αC-Glu (PDB ID: 4ouc [42]). RS3 Ser may also be accessible in MAP2K6 when it is not in its autoinhibited state (PDB ID: 3fme). In the autoinhibited state its neighbouring Met appears to take the regular RS3 position (PDB ID: 3vn9 [32]).

Two unique polar RS3 residues are present in the data. Asn is observed in the RS3 of CHK1, while all the other kinases belonging to the same CAMKL family have either a lipophilic Leu or Met in this position. CHK1 inhibition could be useful in the treatment of KRAS driven pancreatic ductal adenocarcinoma [43]. Thr is observed in ULK4, while other members of the ULK family (ULK1–ULK3) have Leu in the respective position. ULK4 is a pseudokinase and has an unusual structural characteristic in its αC-helix: it exhibits Trp residue in the location of αC-Glu. This Trp appears to participate in its R-spine formation [44].

The access towards the hydrophobic pocket (and towards RS3) is controlled by gatekeeper, also known as SH2 residue. This residue may also influence R-spine dynamics and it can be found in close contact to RS3 (Figure 3A). Generally, protein kinases prefer Met, Leu, Phe and Thr gatekeepers (Figure 3B,C). In the available structures, Met is the most abundant gatekeeper (40%), followed by Leu (18%), Phe (16%) and Thr (15%). Less frequent gatekeepers — but presented in more than eight kinase domains — are Ile 4%, Tyr 3%, Val 3%.

Kinases with different RS3 residues display also distinct gatekeeper preferences (Figure 3B,C). With Met in RS3, less Phe gatekeepers are observed in comparison with when RS3 is Leu. Bulky aromatic RS3 residues (Tyr, His, Phe) do not exist in combination with aromatic gatekeepers. The kinase domains with Cys in RS3 appear to prefer Phe gatekeepers. Polar Gln and Asn display either Leu or Met as their gatekeepers. While Thr is rare RS3 residue, as a gatekeeper it is common with Leu, Met and Tyr in RS3. In addition to Thr, other polar gatekeepers do exist. Asn gatekeeper is observed only in combination with His RS3 in WEE1 (Wee1) (PDB ID: 3biz [45]) and WEE2 (Wee1B) (PDB ID: 5vdk [46]). Gln gatekeeper is observed in four kinases, including MAPK1 (Erk2) (PDB ID: 5hnv [47]), MAPK3 (Erk1) (PDB ID: 4qtb [48]), MAPK6 (Erk3) (PDB ID: 7aqb [49]), and pseudokinase domain JAK2-b (PDB ID: 4fvq [50]). Pseudokinase domain Jak1-b has a unique polar Glu as its gatekeeper (PDB ID: 4l00 [51]). Overall, thirteen protein kinases in the analysed dataset display unique RS3-gatekeeper combinations (Figure 3C).

We searched the KLIFS database [52] and complemented our search using Protein–Ligand Database (PLDB) tool of Maestro (Schrödinger LLC) to map out all the existing protein kinase–ligand complexes that have contacts between RS3 and the ligand (Figure 4). In KLIFS, RS3 is named as residue #28 (αC-Glu is #24) [53]. Over 100 protein kinases have structures where RS3–ligand interactions are observed (Figure 4B). In total, more than 1000 structures with RS3–ligand interactions are available.

RS3–ligand interactions are observed with kinases containing Leu, Met, Tyr, Asn, Val, Gln, Ile or Ser in their RS3. No direct RS3–ligand contacts were observed for kinases with Ala, Cys, Phe, His and Thr. Polar interactions to RS3 are extremely rare. H-bond interaction occurs between inhibitor and RS3 only in 12 structures, including AURKA (Gln) (PDB ID: 4uzd [57]); CHK1 (Asn) (PDB IDs: 4fsq, 4fst, 4ftk, 4ftl, 4ftm, 4ftn); Haspin (Ser) (PDB ID: 6z5a); VRK1 (Tyr) (PDB IDs: 6btw, 6cfm, 6cnx, 6npn). Hydrophobic interactions are abundant, and hydrophobic contacts to Leu and Met appear in 551 and 410 structures, respectively. This is not surprising, based on their high frequency in RS3 (79%) among the available structures. Interactions to Tyr appear in 41 structures, while other RS3 residues with interactions are represented each with 10 or less structures.

Interactions to RS3 appear independent on the kinase conformation (Figure 4C,D) [55]. Based on the KinaMetrix [56], ‘αC-helix in' conformations are dominating in the structures. CIDI is the most populated with 459 structures and CIDO appears in 323 structures. CODI and CODO structures with αC-helix out configuration exist in 148 and 42 structures, respectively. The ambiguous ωCD occurs in 47 structures. Leu and Met RS3–ligand interaction structures display all configurations, albeit less structures of CODI (16%), CODO (4%) and ωCD (4%) conformations exists. With less structural information containing RS3 residues the conformational representation does not cover all configurations. Nevertheless, all conformations are present in the structures with RS3–ligand interactions. As there exists distinct protein kinase conformation classifications, we also analysed the conformational distribution of RS3 contact structures with Kincore [58]. Table 1 shows the conformational distribution of these structures assigned with the Kincore. Overall, majority of the available compounds with the RS3 interactions exist in DFGin and DFGout spatial classes, covering different conformational classes (dihedrals).

A publicly available structure exists for 49 out of the 71 FDA-approved small-molecule kinase inhibitors. From these, interaction to RS3 is displayed by 26 inhibitors (55%). These structures include targets with Met (Figure 5) and Leu RS3 residues (Figure 6). Inhibitors which exhibit RS3 interactions represent all types of kinase inhibitors that bind to the ATP-binding cleft. Of note, inhibitors of different type may engage RS3 site in a different manner. For instance, type II inhibitors, which bind the kinase in its inactive conformation, reach beyond the RS3 on the αC-helix side, and thereby can interact with the side chain to its ‘side' or ‘head' or both. Conversely, type I inhibitors, which bind to the active conformation of the kinase, interact mainly with the head of the RS3 side chain.

In occupying the region next to RS3, mainly six membered aromatic ring containing structural moieties are preferred (Figures 5, 6). Correspondingly, these aromatic rings participate in RS3 interactions in most cases. Tivozanib is the only inhibitor that presents a five-membered heteroaromatic ring in this location. Furthermore, aromatic ring attached halogens, Cl with RS3 Met (bosutinib; afatinib) and Br or I with RS3 Leu (vandetanib; trametinib, cobimetinib), display contacts to RS3. Vemurafenib and lenvatinib are the only exceptions that utilize nonaromatic moieties to interact with RS3. In addition to VEGFRs (Leu), lenvatinib also binds and inhibits FGFR-1 (Met). To both, Leu or Met, lenvatinib is in contact from its cyclopropyl urea moiety (PDB IDs: 3wzd [59], 5zv2 [60]). Vemurafenib displays contacts to RS3 Leu from aliphatic carbons of its propylsulfonamide group (PDB ID: 4rzv [61]).

Some of these drugs display interactions with other kinases RS3 that are distinct from their main target. Bosutinib displays hydrophobic contacts to RS3 with MST3 (Leu) (PDB ID: 4qmn [62]), as well as dasatinib to Leu with MAPK14 (p38a) (PDB ID: 3lfa). Ponatinib displays hydrophobic interactions to Leu with RIPK2 (PDB ID: 4c8b [63]), KIT (PDB ID: 4u0i [64]) and BRAF (PDB ID: 6p3d [65]). Nilotinib to Leu with MAPK11 (p38b) (PDB ID: 3gp0). Imatinib displays contact to Leu when bound to KIT (PDB ID: 1t46 [66]) and MAPK14 (p38a) (PDB ID: 3hec [67]). Also, Pexidartinib exhibits contact to RS3 Leu with KIT (PDB ID: 7khg [68]). For inhibitors which main targets display Leu in RS3, apart from lenvatinib no Met RS3 interaction containing structures are available.

RS3 interaction is kinase dependent. Approved drugs with interactions to RS3 do not necessarily exhibit contacts to RS3 with other kinases that they bind to. For instance, gefitinib has been co-crystallized with GAK (Met), where it displays no contact to RS3 (PDB ID: 5y7z [69]). Ibrutinib has been also co-crystallized with MAP2K7 that has Val in RS3, but it does not display any contacts to this residue in this complex (PDB ID: 6yg2 [70]).

According to the Catalogue of Somatic Mutations In Cancer (COSMIC; v.95) database [71], no clear tendency for mutations in RS3 exists. In total 82 kinases display at least one mutation (missense or silent) (Table 2). Only with ALK, several mutations at this location appear in the data. These mutations include, I1171N, I1171T and I1171S. For BRAF, L505H mutation is found in eight samples. Perhaps the low number of observed RS3 mutations is not surprising, due to the crucial role of this residue in the kinase function. In comparison, RS2 mutations are also rare, with BRAF F595L (13 samples in COSMIC v.95) being the most frequent in the analysed kinases. Meanwhile, RS2 flanking residues are common oncogenic drivers; for instance, BRAF V600E is found in 52 733 samples and EGFR L858R is present in 10 642 samples. Mutations at αC-helix may activate the kinase via destabilizing the kinase inactive conformation [72], but they are mainly found in other locations on the αC-helix than on RS3 [73].

In the literature, only few cases of studies including RS3 mutations have been reported. With BRAF, a secondary mutation in RS3 (L505H) induces resistance to vemurafenib [74]. Also, in another study resistance to dabrafenib or vemurafenib was demonstrated with BRAF L505H mutation [10]. Moreover, alectinib resistance was reported with I1171S and I1171N mutations in ALK [75]. The I1171N mutation was demonstrated to increase autophosphorylation level of ALK in vitro [76]. I1171T has been identified to induce crizotinib resistance [77]. Interestingly, these two inhibitors, crizotinib and alectinib do not display direct contacts to RS3. EGFR M766T mutation was reported to induce resistance to gefitinib and erlotinib [78].

The function of the R-spine and the role of RS3 is quite conserved with typical protein kinases. Couple of residues are dominating RS3 in the available protein kinase domain structures. Nevertheless, also unique RS3 residues are observed, and in combination with gatekeeper even more kinase specific profiles for these residues are observed. Obviously, even with identical RS3–gatekeeper combinations the 3D-environment within this region can be quite different between two kinases. Kinase specific angles and absolute positions of these residues may provide important opportunities for selective targeting. Obviously, one must carefully consider this case-by-case, as RS3 is not targetable in all kinases. With pseudokinases [79–81], which compared with regular protein kinases can vary more in their structure in this region, the role of RS3 and its targeting would require further research.

The general understanding of RS3–ligand interactions are quite limited, even though numerous structures that contain these mainly hydrophobic interactions are available. Currently, no studies investigating specific effect of RS3 on ligand binding affinity exist that directly compare a set of ligands with selected mutations of this residue. Further research is needed to disclose the influence of RS3 residue for ligand binding and should be also extended to the cases where no direct contact between the residue and inhibitor exists. Of note, even with hydrophobic interactions (in the case of hydrophobic RS3) this should not be overlooked as these interactions may be crucial for the inhibitor binding [82]. For example, non-canonical interactions play a detrimental role in binding affinity of the ultra-potent small-molecule biotin [83]. There may be good possibilities available to optimize RS3-specific interactions, for instance, with enhanced interactions with the sulfur atom of Met [84].

Infrequency of mutations in RS3 may indicate a defiance against plausible point mutation in this position that could cause drug resistance [85]. Perhaps the somewhat buried location of RS3 in the quite rigid αC-helix position that offers a limited flexibility, renders the mutations in this location (at least in most cases) incompetent to drive kinase activation. This motivates further to optimize protein–ligand interactions for RS3. However, the data at hand may not necessarily cover potential drug therapy induced mutations in cancer patients. We believe that in near future, with accumulation of this data, this information will be more accessible, and a better estimate can be provided.

The full data presented in this review are freely available at https://doi.org/10.5281/zenodo.5796550

  • The role of the conserved R-spine and RS3 residue in protein kinase function is well established.

  • Protein kinases display diversity in their RS3 residue and in its surroundings. Many small-molecule protein kinase inhibitors, including approved drugs, display contacts to RS3.

  • Considering the RS3 residue more carefully in the design of small-molecule kinase inhibitors may offer important advantage for the inhibitor binding and selectivity.

The authors declare that there are no competing interests associated with the manuscript.

This research is financially supported by Academy of Finland GeneCellNano Flagship funding 337120 and T.P. acknowledges funding by Orion Research Foundation sr.

Both authors wrote the manuscript and made the figures. The original concept was conceived by T.P. T.P. directed the project.

ATP

adenosine triphosphate

PLDB

Protein–Ligand Database

1
Manning
,
G.
,
Whyte
,
D.B.
,
Martinez
,
R.
,
Hunter
,
T.
and
Sudarsanam
,
S.
(
2002
)
The protein kinase complement of the human genome
.
Science
298
,
1912
1934
2
Modi
,
V.
and
Dunbrack
,
R.L.
(
2019
)
A structurally-validated multiple sequence alignment of 497 human protein kinase domains
.
Sci. Rep.
9
,
19790
3
Berginski
,
M.E.
,
Moret
,
N.
,
Liu
,
C.
,
Goldfarb
,
D.
,
Sorger
,
P.K.
and
Gomez
,
S.M.
(
2021
)
The dark kinase knowledgebase: an online compendium of knowledge and experimental results of understudied kinases
.
Nucleic Acids Res.
49
,
D529
D535
4
Myasnikov
,
A.
,
Zhu
,
H.
,
Hixson
,
P.
,
Xie
,
B.
,
Yu
,
K.
,
Pitre
,
A.
et al (
2021
)
Structural analysis of the full-length human LRRK2
.
Cell
184
,
3519
3527.e10
5
Taylor
,
S.S.
,
Wu
,
J.
,
Bruystens
,
J.G.H.
,
Rio
,
J.C.D.
,
Lu
,
T.-W.
,
Kornev
,
A.P.
et al (
2021
)
From structure to the dynamic regulation of a molecular switch: a journey over 3 decades
.
J. Biol. Chem.
296
,
100746
6
Kornev
,
A.P.
,
Haste
,
N.M.
,
Taylor
,
S.S.
and
Eyck
,
L.F.T.
(
2006
)
Surface comparison of active and inactive protein kinases identifies a conserved activation mechanism
.
Proc. Natl Acad. Sci. U.S.A.
103
,
17783
17788
7
Meharena
,
H.S.
,
Chang
,
P.
,
Keshwani
,
M.M.
,
Oruganty
,
K.
,
Nene
,
A.K.
,
Kannan
,
N.
et al (
2013
)
Deciphering the structural basis of eukaryotic protein kinase regulation
.
PLoS Biol.
11
,
e1001680
8
Taylor
,
S.S.
,
Shaw
,
A.S.
,
Kannan
,
N.
and
Kornev
,
A.P.
(
2015
)
Integration of signaling in the kinome: architecture and regulation of the αC helix
.
Biochim. Biophys. Acta Proteins Proteomics
1854
,
1567
1574
9
Hu
,
J.
,
Ahuja
,
L.G.
,
Meharena
,
H.S.
,
Kannan
,
N.
,
Kornev
,
A.P.
,
Taylor
,
S.S.
et al (
2015
)
Kinase regulation by hydrophobic spine assembly in cancer
.
Mol. Cell. Biol.
35
,
264
276
10
Yap
,
J.
,
Deepak
,
R.N.V.K.
,
Tian
,
Z.
,
Ng
,
W.H.
,
Goh
,
K.C.
,
Foo
,
A.
et al (
2021
)
The stability of R-spine defines RAF inhibitor resistance: a comprehensive analysis of oncogenic BRAF mutants with in-frame insertion of αC-β4 loop
.
Sci. Adv.
7
,
eabg0390
11
Liu
,
Y.
,
Shah
,
K.
,
Yang
,
F.
,
Witucki
,
L.
and
Shokat
,
K.M.
(
1998
)
A molecular gate which controls unnatural ATP analogue recognition by the tyrosine kinase v-Src
.
Bioorg. Med. Chem.
6
,
1219
1226
12
Nussinov
,
R.
,
Zhang
,
M.
,
Maloney
,
R.
,
Tsai
,
C.-J.
,
Yavuz
,
B.R.
,
Tuncbag
,
N.
et al (
2021
)
Mechanism of activation and the rewired network: new drug design concepts
.
Med. Res. Rev.
42
,
770
799
13
Shih
,
H.-P.
,
Zhang
,
X.
and
Aronov
,
A.M.
(
2018
)
Drug discovery effectiveness from the standpoint of therapeutic mechanisms and indications
.
Nat. Rev. Drug Discov.
17
,
19
33
14
Bournez
,
C.
,
Carles
,
F.
,
Peyrat
,
G.
,
Aci-Sèche
,
S.
,
Bourg
,
S.
,
Meyer
,
C.
et al (
2020
)
Comparative assessment of protein kinase inhibitors in public databases and in PKIDB
.
Molecules
25
,
3226
15
Carles
,
F.
,
Bourg
,
S.
,
Meyer
,
C.
and
Bonnet
,
P.
(
2018
)
PKIDB: a curated, annotated and updated database of protein kinase inhibitors in clinical trials
.
Molecules
23
,
908
16
Ayala-Aguilera
,
C.C.
,
Valero
,
T.
,
Lorente-Macías
,
Á.
,
Baillache
,
D.J.
,
Croke
,
S.
and
Unciti-Broceta
,
A.
(
2021
)
Small molecule kinase inhibitor drugs (1995–2021): medical indication, pharmacology, and synthesis
.
J. Med. Chem.
65
,
1047
1131
17
Cohen
,
P.
,
Cross
,
D.
and
Jänne
,
P.A.
(
2021
)
Kinase drug discovery 20 years after imatinib: progress and future directions
.
Nat. Rev. Drug Discov.
20
,
551
569
18
Attwood
,
M.M.
,
Fabbro
,
D.
,
Sokolov
,
A.V.
,
Knapp
,
S.
and
Schiöth
,
H.B.
(
2021
)
Trends in kinase drug discovery: targets, indications and inhibitor design
.
Nat. Rev. Drug Discov.
20
,
839
861
19
Ferguson
,
F.M.
and
Gray
,
N.S.
(
2018
)
Kinase inhibitors: the road ahead
.
Nat. Rev. Drug Discov.
17
,
353
377
20
Cheung
,
J.
,
Ginter
,
C.
,
Cassidy
,
M.
,
Franklin
,
M.C.
,
Rudolph
,
M.J.
,
Robine
,
N.
et al (
2015
)
Structural insights into mis-regulation of protein kinase A in human tumors
.
Proc. Natl Acad. Sci. U.S.A.
112
,
1374
1379
21
Qin
,
J.
,
Xie
,
P.
,
Ventocilla
,
C.
,
Zhou
,
G.
,
Vultur
,
A.
,
Chen
,
Q.
et al (
2012
)
Identification of a novel family of BRAFV600E inhibitors
.
J. Med. Chem.
55
,
5220
5230
22
Wan
,
P.T.C.
,
Garnett
,
M.J.
,
Roe
,
S.M.
,
Lee
,
S.
,
Niculescu-Duvaz
,
D.
,
Good
,
V.M.
et al (
2004
)
Mechanism of activation of the RAF-ERK signaling pathway by oncogenic mutations of B-RAF
.
Cell
116
,
855
867
23
Kozlowski
,
L.P.
(
2017
)
Proteome-pI: proteome isoelectric point database
.
Nucleic Acids Res.
45
,
D1112
D1116
24
Biswas
,
K.M.
,
DeVido
,
D.R.
and
Dorsey
,
J.G.
(
2003
)
Evaluation of methods for measuring amino acid hydrophobicities and interactions
.
J. Chromatogr. A
1000
,
637
655
25
Simm
,
S.
,
Einloft
,
J.
,
Mirus
,
O.
and
Schleiff
,
E.
(
2016
)
50 years of amino acid hydrophobicity scales: revisiting the capacity for peptide classification
.
Biol. Res.
49
,
31
26
Roskoski
,
R.
(
2019
)
Small molecule inhibitors targeting the EGFR/ErbB family of protein-tyrosine kinases in human cancers
.
Pharmacol. Res.
139
,
395
411
27
Shi
,
F.
,
Telesco
,
S.E.
,
Liu
,
Y.
,
Radhakrishnan
,
R.
and
Lemmon
,
M.A.
(
2010
)
Erbb3/HER3 intracellular domain is competent to bind ATP and catalyze autophosphorylation
.
Proc. Natl Acad. Sci. U.S.A.
107
,
7692
7697
28
Buchanan
,
J.L.
,
Newcomb
,
J.R.
,
Carney
,
D.P.
,
Chaffee
,
S.C.
,
Chai
,
L.
,
Cupples
,
R.
et al (
2011
)
Discovery of 2,4-bis-arylamino-1,3-pyrimidines as insulin-like growth factor-1 receptor (IGF-1R) inhibitors
.
Bioorg. Med. Chem. Lett.
21
,
2394
2399
29
Eid
,
S.
,
Turk
,
S.
,
Volkamer
,
A.
,
Rippmann
,
F.
and
Fulle
,
S.
(
2017
)
Kinmap: a web-based tool for interactive navigation through human kinome data
.
BMC Bioinformatics
18
,
16
30
Matsumoto
,
T.
,
Kinoshita
,
T.
,
Kirii
,
Y.
,
Yokota
,
K.
,
Hamada
,
K.
and
Tada
,
T.
(
2010
)
Crystal structures of MKK4 kinase domain reveal that substrate peptide binds to an allosteric site and induces an auto-inhibition state
.
Biochem. Biophys. Res. Commun.
400
,
369
373
31
Sogabe
,
Y.
,
Hashimoto
,
T.
,
Matsumoto
,
T.
,
Kirii
,
Y.
,
Sawa
,
M.
and
Kinoshita
,
T.
(
2016
)
A crucial role of Cys218 in configuring an unprecedented auto-inhibition form of MAP2K7
.
Biochem. Biophys. Res. Commun.
473
,
476
481
32
Matsumoto
,
T.
,
Kinoshita
,
T.
,
Matsuzaka
,
H.
,
Nakai
,
R.
,
Kirii
,
Y.
,
Yokota
,
K.
et al (
2012
)
Crystal structure of non-phosphorylated MAP2K6 in a putative auto-inhibition state
.
J. Biochem.
151
,
541
549
33
Shevchenko
,
E.
,
Poso
,
A.
and
Pantsar
,
T.
(
2020
)
The autoinhibited state of MKK4: phosphorylation, putative dimerization and R134W mutant studied by molecular dynamics simulations
.
Comput. Struct. Biotechnol. J.
18
,
2687
2698
34
Murphy
,
J.M.
,
Nakatani
,
Y.
,
Jamieson
,
S.A.
,
Dai
,
W.
,
Lucet
,
I.S.
and
Mace
,
P.D.
(
2015
)
Molecular mechanism of CCAAT-enhancer binding protein recruitment by the TRIB1 pseudokinase
.
Structure
23
,
2111
2121
35
Jamieson
,
S.A.
,
Ruan
,
Z.
,
Burgess
,
A.E.
,
Curry
,
J.R.
,
McMillan
,
H.D.
,
Brewster
,
J.L.
et al (
2018
)
Substrate binding allosterically relieves autoinhibition of the pseudokinase TRIB1
.
Sci Signal.
11
,
eaau0597
36
Asquith
,
C.R.M.
,
Murray
,
N.H.
and
Pagliarini
,
D.J.
(
2019
)
ADCK3/COQ8A: the choice target of the UbiB protein kinase-like family
.
Nat. Rev. Drug Discov.
18
,
815
815
37
Cyphers
,
S.
,
Ruff
,
E.F.
,
Behr
,
J.M.
,
Chodera
,
J.D.
and
Levinson
,
N.M.
(
2017
)
A water-mediated allosteric network governs activation of aurora kinase A
.
Nat. Chem. Biol.
13
,
402
408
38
Xu
,
D.
,
Matsumoto
,
M.L.
,
McKenzie
,
B.S.
and
Zarrin
,
A.A.
(
2018
)
TPL2 kinase action and control of inflammation
.
Pharmacol. Res.
129
,
188
193
39
Lee
,
J.-H.
and
Paull
,
T.T.
(
2021
)
Cellular functions of the protein kinase ATM and their relevance to human disease
.
Nat. Rev. Mol. Cell Biol.
22
,
796
814
40
Zeqiraj
,
E.
,
Filippi
,
B.M.
,
Goldie
,
S.
,
Navratilova
,
I.
,
Boudeau
,
J.
,
Deak
,
M.
et al (
2009
)
ATP and MO25alpha regulate the conformational state of the STRADalpha pseudokinase and activation of the LKB1 tumour suppressor
.
PLoS Biol.
7
,
e1000126
41
Zeqiraj
,
E.
,
Filippi
,
B.M.
,
Deak
,
M.
,
Alessi
,
D.R.
and
van Aalten
,
D.M.F.
(
2009
)
Structure of the LKB1-STRAD-MO25 complex reveals an allosteric mechanism of kinase activation
.
Science
326
,
1707
1711
42
Maiolica
,
A.
,
de Medina-Redondo
,
M.
,
Schoof
,
E.M.
,
Chaikuad
,
A.
,
Villa
,
F.
,
Gatti
,
M.
et al (
2014
)
Modulation of the chromatin phosphoproteome by the haspin protein kinase*
.
Mol. Cell. Proteomics
13
,
1724
1740
43
Klomp
,
J.E.
,
Lee
,
Y.S.
,
Goodwin
,
C.M.
,
Papke
,
B.
,
Klomp
,
J.A.
,
Waters
,
A.M.
et al (
2021
)
CHK1 protects oncogenic KRAS-expressing cells from DNA damage and is a target for pancreatic cancer treatment
.
Cell Rep.
37
,
110060
44
Preuss
,
F.
,
Chatterjee
,
D.
,
Mathea
,
S.
,
Shrestha
,
S.
,
St-Germain
,
J.
,
Saha
,
M.
et al (
2020
)
Nucleotide binding, evolutionary insights, and interaction partners of the pseudokinase Unc-51-like kinase 4
.
Structure
28
,
1184
1196.e6
45
Smaill
,
J.B.
,
Lee
,
H.H.
,
Palmer
,
B.D.
,
Thompson
,
A.M.
,
Squire
,
C.J.
,
Baker
,
E.N.
et al (
2008
)
Synthesis and structure-activity relationships of soluble 8-substituted 4-(2-chlorophenyl)-9-hydroxypyrrolo[3,4-c]carbazole-1,3(2H,6H)-diones as inhibitors of the Wee1 and Chk1 checkpoint kinases
.
Bioorg. Med. Chem. Lett.
18
,
929
933
46
Zhu
,
J.-Y.
,
Cuellar
,
R.A.
,
Berndt
,
N.
,
Lee
,
H.E.
,
Olesen
,
S.H.
,
Martin
,
M.P.
et al (
2017
)
Structural basis of wee kinases functionality and inactivation by diverse small molecule inhibitors
.
J. Med. Chem.
60
,
7863
7875
47
Ward
,
R.A.
,
Bethel
,
P.
,
Cook
,
C.
,
Davies
,
E.
,
Debreczeni
,
J.E.
,
Fairley
,
G.
et al (
2017
)
Structure-guided discovery of potent and selective inhibitors of ERK1/2 from a modestly active and promiscuous chemical start point
.
J. Med. Chem.
60
,
3438
3450
48
Chaikuad
,
A.
,
Tacconi
,
E.M.C.
,
Zimmer
,
J.
,
Liang
,
Y.
,
Gray
,
N.S.
,
Tarsounas
,
M.
et al (
2014
)
A unique inhibitor binding site in ERK1/2 is associated with slow binding kinetics
.
Nat. Chem. Biol.
10
,
853
860
49
Schröder
,
M.
,
Filippakopoulos
,
P.
,
Schwalm
,
M.P.
,
Ferrer
,
C.A.
,
Drewry
,
D.H.
,
Knapp
,
S.
et al (
2020
)
Crystal structure and inhibitor identifications reveal targeting opportunity for the atypical MAPK kinase ERK3
.
Int. J. Mol. Sci.
21
,
7953
50
Bandaranayake
,
R.M.
,
Ungureanu
,
D.
,
Shan
,
Y.
,
Shaw
,
D.E.
,
Silvennoinen
,
O.
and
Hubbard
,
S.R.
(
2012
)
Crystal structures of the JAK2 pseudokinase domain and the pathogenic mutant V617F
.
Nat. Struct. Mol. Biol.
19
,
754
759
51
Toms
,
A.V.
,
Deshpande
,
A.
,
McNally
,
R.
,
Jeong
,
Y.
,
Rogers
,
J.M.
,
Kim
,
C.U.
et al (
2013
)
Structure of a pseudokinase-domain switch that controls oncogenic activation of Jak kinases
.
Nat. Struct. Mol. Biol.
20
,
1221
1223
52
Kanev
,
G.K.
,
de Graaf
,
C.
,
Westerman
,
B.A.
,
de Esch
,
I.J.P.
and
Kooistra
,
A.J.
(
2021
)
KLIFS: an overhaul after the first 5 years of supporting kinase research
.
Nucleic Acids Res.
49
,
D562
D569
53
van Linden
,
O.P.J.
,
Kooistra
,
A.J.
,
Leurs
,
R.
,
de Esch
,
I.J.P.
and
de Graaf
,
C.
(
2014
)
KLIFS: a knowledge-based structural database To navigate kinase–ligand interaction space
.
J. Med. Chem.
57
,
249
277
54
Yasuda
,
H.
,
Park
,
E.
,
Yun
,
C.-H.
,
Sng
,
N.J.
,
Lucena-Araujo
,
A.R.
,
Yeo
,
W.-L.
et al (
2013
)
Structural, biochemical, and clinical characterization of epidermal growth factor receptor (EGFR) exon 20 insertion mutations in lung cancer
.
Sci. Transl. Med.
5
,
216ra177
55
Ung
,
P.M.-U.
,
Rahman
,
R.
and
Schlessinger
,
A.
(
2018
)
Redefining the protein kinase conformational space with machine learning
.
Cell Chem. Biol.
25
,
916
924.e2
56
Rahman
,
R.
,
Ung
,
P.M.-U.
and
Schlessinger
,
A.
(
2019
)
Kinametrix: a web resource to investigate kinase conformations and inhibitor space
.
Nucleic Acids Res.
47
,
D361
D366
57
Carry
,
J.-C.
,
Clerc
,
F.
,
Minoux
,
H.
,
Schio
,
L.
,
Mauger
,
J.
,
Nair
,
A.
et al (
2015
)
SAR156497, an exquisitely selective inhibitor of aurora kinases
.
J. Med. Chem.
58
,
362
375
58
Modi
,
V.
and
Dunbrack
,
R.L.
(
2022
)
Kincore: a web resource for structural classification of protein kinases and their inhibitors
.
Nucleic Acids Res.
50
,
D654
D664
59
Okamoto
,
K.
,
Ikemori-Kawada
,
M.
,
Jestel
,
A.
,
von König
,
K.
,
Funahashi
,
Y.
,
Matsushima
,
T.
et al (
2015
)
Distinct binding mode of multikinase inhibitor lenvatinib revealed by biochemical characterization
.
ACS Med. Chem. Lett.
6
,
89
94
60
Matsuki
,
M.
,
Hoshi
,
T.
,
Yamamoto
,
Y.
,
Ikemori-Kawada
,
M.
,
Minoshima
,
Y.
,
Funahashi
,
Y.
et al (
2018
)
Lenvatinib inhibits angiogenesis and tumor fibroblast growth factor signaling pathways in human hepatocellular carcinoma models
.
Cancer Med.
7
,
2641
2653
61
Karoulia
,
Z.
,
Wu
,
Y.
,
Ahmed
,
T.A.
,
Xin
,
Q.
,
Bollard
,
J.
,
Krepler
,
C.
et al (
2016
)
An integrated model of RAF inhibitor action predicts inhibitor activity against oncogenic BRAF signaling
.
Cancer Cell
30
,
485
498
62
Olesen
,
S.H.
,
Zhu
,
J.-Y.
,
Martin
,
M.P.
and
Schönbrunn
,
E.
(
2016
)
Discovery of diverse small-molecule inhibitors of mammalian Sterile20-like kinase3 (MST3)
.
ChemMedChem
11
,
1137
1144
63
Canning
,
P.
,
Ruan
,
Q.
,
Schwerd
,
T.
,
Hrdinka
,
M.
,
Maki
,
J.L.
,
Saleh
,
D.
et al (
2015
)
Inflammatory signaling by NOD-RIPK2 Is inhibited by clinically relevant type II kinase inhibitors
.
Chem. Biol.
22
,
1174
1184
64
Garner
,
A.P.
,
Gozgit
,
J.M.
,
Anjum
,
R.
,
Vodala
,
S.
,
Schrock
,
A.
,
Zhou
,
T.
et al (
2014
)
Ponatinib inhibits polyclonal drug-resistant KIT oncoproteins and shows therapeutic potential in heavily pretreated gastrointestinal stromal tumor (GIST) patients
.
Clin. Cancer Res.
20
,
5745
5755
65
Cotto-Rios
,
X.M.
,
Agianian
,
B.
,
Gitego
,
N.
,
Zacharioudakis
,
E.
,
Giricz
,
O.
,
Wu
,
Y.
et al (
2020
)
Inhibitors of BRAF dimers using an allosteric site
.
Nat. Commun.
11
,
4370
66
Mol
,
C.D.
,
Dougan
,
D.R.
,
Schneider
,
T.R.
,
Skene
,
R.J.
,
Kraus
,
M.L.
,
Scheibe
,
D.N.
et al (
2004
)
Structural basis for the autoinhibition and STI-571 inhibition of c-Kit tyrosine kinase
.
J. Biol. Chem.
279
,
31655
31663
67
Namboodiri
,
H.V.
,
Bukhtiyarova
,
M.
,
Ramcharan
,
J.
,
Karpusas
,
M.
,
Lee
,
Y.
and
Springman
,
E.B.
(
2010
)
Analysis of imatinib and sorafenib binding to p38alpha compared with c-Abl and b-Raf provides structural insights for understanding the selectivity of inhibitors targeting the DFG-out form of protein kinases
.
Biochemistry
49
,
3611
3618
68
Wagner
,
A.J.
,
Severson
,
P.L.
,
Shields
,
A.F.
,
Patnaik
,
A.
,
Chugh
,
R.
,
Tinoco
,
G.
et al (
2021
)
Association of combination of conformation-specific KIT inhibitors with clinical benefit in patients with refractory gastrointestinal stromal tumors: a phase 1b/2a nonrandomized clinical trial
.
JAMA Oncol.
7
,
1343
1350
69
Ohbayashi
,
N.
,
Murayama
,
K.
,
Kato-Murayama
,
M.
,
Kukimoto-Niino
,
M.
,
Uejima
,
T.
,
Matsuda
,
T.
et al (
2018
)
Structural basis for the inhibition of cyclin G-associated kinase by gefitinib
.
ChemistryOpen
7
,
713
719
70
Schröder
,
M.
,
Tan
,
L.
,
Wang
,
J.
,
Liang
,
Y.
,
Gray
,
N.S.
,
Knapp
,
S.
et al (
2020
)
Catalytic domain plasticity of MKK7 reveals structural mechanisms of allosteric activation and diverse targeting opportunities
.
Cell Chem. Biol.
27
,
1285
1295.e4
71
Tate
,
J.G.
,
Bamford
,
S.
,
Jubb
,
H.C.
,
Sondka
,
Z.
,
Beare
,
D.M.
,
Bindal
,
N.
et al (
2019
)
COSMIC: the catalogue of somatic mutations in cancer
.
Nucleic Acids Res.
47
,
D941
D947
72
Barouch-Bentov
,
R.
and
Sauer
,
K.
(
2011
)
Mechanisms of drug resistance in kinases
.
Expert Opin. Investig. Drugs.
20
,
153
208
73
Kim
,
P.
,
Li
,
H.
,
Wang
,
J.
and
Zhao
,
Z.
(
2021
)
Landscape of drug-resistance mutations in kinase regulatory hotspots
.
Brief Bioinform.
22
,
bbaa108
74
Hoogstraat
,
M.
,
Hooijdonk
,
C.G.G.
,
Ubink
,
I.
,
Besselink
,
N.J.M.
,
Pieterse
,
M.
,
Veldhuis
,
W.
et al (
2015
)
Detailed imaging and genetic analysis reveal a secondary BRAFL505H resistance mutation and extensive intrapatient heterogeneity in metastatic BRAF mutant melanoma patients treated with vemurafenib
.
Pigment Cell Melanoma Res.
28
,
318
323
75
Ou
,
S.-H.I.
,
Klempner
,
S.J.
,
Greenbowe
,
J.R.
,
Azada
,
M.
,
Schrock
,
A.B.
,
Ali
,
S.M.
et al (
2014
)
Identification of a novel HIP1-ALK fusion variant in non-small-cell lung cancer (NSCLC) and discovery of ALK I1171 (I1171N/S) mutations in two ALK-rearranged NSCLC patients with resistance to Alectinib
.
J. Thorac. Oncol.
9
,
1821
1825
76
Bresler
,
S.C.
,
Weiser
,
D.A.
,
Huwe
,
P.J.
,
Park
,
J.H.
,
Krytska
,
K.
,
Ryles
,
H.
et al (
2014
)
ALK mutations confer differential oncogenic activation and sensitivity to ALK inhibition therapy in neuroblastoma
.
Cancer Cell
26
,
682
694
77
Friboulet
,
L.
,
Li
,
N.
,
Katayama
,
R.
,
Lee
,
C.C.
,
Gainor
,
J.F.
,
Crystal
,
A.S.
et al (
2014
)
The ALK inhibitor ceritinib overcomes crizotinib resistance in non-small cell lung cancer
.
Cancer Discov.
4
,
662
673
78
Ruan
,
Z.
,
Katiyar
,
S.
and
Kannan
,
N.
(
2017
)
Computational and experimental characterization of patient derived mutations reveal an unusual mode of regulatory spine assembly and drug sensitivity in EGFR kinase
.
Biochemistry
56
,
22
32
79
Mace
,
P.D.
and
Murphy
,
J.M.
(
2021
)
There's more to death than life: noncatalytic functions in kinase and pseudokinase signaling
.
J. Biol. Chem.
296
,
100705
80
Shrestha
,
S.
,
Byrne
,
D.P.
,
Harris
,
J.A.
,
Kannan
,
N.
and
Eyers
,
P.A.
(
2020
)
Cataloguing the dead: breathing new life into pseudokinase research
.
FEBS J.
287
,
4150
4169
81
Kung
,
J.E.
and
Jura
,
N.
(
2019
)
Prospects for pharmacological targeting of pseudokinases
.
Nat. Rev. Drug Discov.
18
,
501
526
82
Bissantz
,
C.
,
Kuhn
,
B.
and
Stahl
,
M.
(
2010
)
A medicinal chemist's guide to molecular interactions
.
J. Med. Chem.
53
,
5061
5084
83
McConnell
,
D.B.
(
2021
)
Biotin's lessons in drug design
.
J. Med. Chem.
64
,
16319
16327
84
Beno
,
B.R.
,
Yeung
,
K.-S.
,
Bartberger
,
M.D.
,
Pennington
,
L.D.
and
Meanwell
,
N.A.
(
2015
)
A survey of the role of noncovalent sulfur interactions in drug design
.
J. Med. Chem.
58
,
4383
4438
85
Ward
,
R.A.
,
Fawell
,
S.
,
Floc'h
,
N.
,
Flemington
,
V.
,
McKerrecher
,
D.
and
Smith
,
P.D.
(
2021
)
Challenges and opportunities in cancer drug resistance
.
Chem. Rev.
121
,
3297
3351
This is an open access article published by Portland Press Limited on behalf of the Biochemical Society and distributed under the Creative Commons Attribution License 4.0 (CC BY).