A regulatory role of membrane by direct modulation of the catalytic kinase domain

ABSTRACT

Cell membrane modulates the function and activity of specific proteins and acts more than just a non-specific scaffolding machinery. In this review, I focus on studies that highlight a direct membrane-mediated modulation of the catalytic kinase domain of a variety of kinases thereby regulating the kinase activity. It emerges that membrane provides a second level of regulation once kinase domain is relieved of its inactive auto-inhibitory state. For the first time a generalized regulatory role of membrane is proposed that governs the kinase activity by modulating the catalytic kinase domain. Striking similarities among a variety of multi-domain kinases as well as single-domain lipidated enzymes such as RAS proteins are presented.

GRAPHICAL ABSTRACT

Introduction

Modulation of protein function by various components of plasma membrane is well known [1,2]. Among others, this includes binding of specific signalling lipids to membrane-interacting protein domains (e.g. PH, C1) via lipid-binding pockets [2]; cholesterol-mediated modulation of GPCR function [3]; membrane micro-domain dependent isoform-specific localization and function [4,5]. From recent studies emerges a unique membrane-binding mode adopted for instance by lipidated enzyme KRAS where the solvent-exposed protein surface, lacking any defined lipid-binding pockets/grooves, interacts with the specific anionic phospholipids residing in the cell membrane via basic/polar surface patches [6–11]. As a result, functional modulation is likely achieved either by occluding or exposing the binding site required for interacting with its downstream binding partners (Figure 1). Quite interestingly, a combined overview presented in this review article from recent studies indicates that catalytic kinase domains (kinaseD) of a variety of multi-domain kinases such as cRaf, PI3K, EGFR and others interact with the phospholipids residing in the cell membrane via solvent-exposed polybasic clusters [12–18]. I present in this review, the likely mechanism by which the kinaseD–membrane interaction plays a regulatory role after the release of auto-inhibition and that this mechanism is generalizable across a variety of kinases; and is not only limited to kinases but is also known to occur in other lipidated membrane-bound peripheral proteins such as KRAS. In this review, first I will briefly discuss the role of membrane in modulating the function of lipidated small GTPases followed by studies that focus on the direct interaction between kinase domain and membrane phospholipids. Towards the end, I will present interesting similarities between the modes of membrane interaction of small GTPases and kinases that suggests how nature preserves a general regulatory mechanism across different proteins.

Figure 1.

Cartoon illustration of the two preferred binding modes of KRAS bound to membrane (grey). KRAS is shown as an oval with lobe1 (blue) and lobe2 (pink) and hypervariable region (HVR) in yellow. The switch region (active surface) that binds the downstream binding partners lies in the lobe1. The active surface is occluded in (A) and solvent-exposed in (B) and hence capable of interacting with its downstream binding partners in (B)

Direct interaction of RAS G-domain with anionic membrane – functional role of membrane

RAS family of small GTPases are molecular switches that toggle between an active ‘on’ (GTP-) and inactive ‘off’ (GDP-bound) state [19,20]. In its active form RAS interacts with its downstream binding partners, protein kinases, which upon phosphorylation activates a cascade of downstream signalling thereby regulating key cellular activities such as cell proliferation, progression, survival and growth [19,20]. RAS (three mammalian isoforms, H-, N- and K-Ras) is C-terminally lipidated and attaches itself to the inner leaflet of the plasma membrane for its function. Single point mutations at residues 12, 13 and 61 render RAS constitutive active and mutant RAS is associated with a large number of cancers [20,21]. Different isoforms result in differential functional outputs [20]. Single-domain RAS enzyme is composed of a catalytic G-domain followed by a flexible hypervariable region (HVR) containing the lipidation [19,22,23]. The G-domain is subdivided into an effector-binding (lobe 1) and allosteric lobe (lobe 2) [24]. The downstream binding partners of RAS bind the functionally-critical switch region lying within the lobe1 and this protein-protein interaction is communicated to lobe2 and to the C-terminally located membrane attachment site giving rise to different isoform-specific functional outputs [4,24,25]. Mutant KRAS is one of the deadliest RAS isoforms and anionic phospholipids are required for its proper function [26,27].

Accumulating evidences from computational, spectroscopic, biophysical and biochemical experiments show that solvent-exposed lobes of Ras G-domain associate with the anionic phospholipids such as phosphatidylserine or PIPs and modulate their function [6–9,28-30]. The first report of G-domain/membrane direct interaction in RAS came from a computational study of HRAS attached to a model DMPC bilayer [31]. This observation was later on confirmed using a number of experimental approaches such as cell-based, synthetic membrane models, synthetic or native cell-membrane based nanodiscs for all the three isoforms of RAS and was shown to modulate the function of the RAS enzyme (e.g. [31,32]). Lobes 1 and 2 of the G-domain interact with the membrane via a mix of basic and polar residues [33,34]. Interaction of lobe 1 with the membrane occludes the functionally critical switch regions thereby blocking the entry of the downstream binding partners rendering RAS signalling-incompetent (Figure 1). In contrast, the orientation of RAS in which lobe2 interacts with the membrane exposes switches towards solvent (away from the membrane) thereby making it available to interact with its downstream binding partners and is therefore signalling-competent. Nature of the anionic phospholipid, bound nucleotide, mutations may favour/disfavour certain membrane orientations than others and there are excellent studies/perspectives that discuss these, for example [7,9,10,35,36]. Such functional modulation is observed for Ras-related GTPases as well [36]. The readers are referred to the excellent articles mentioned here and elsewhere for further in-depth information on RAS-membrane interactions. Since the goal of the current review is primarily centred on the catalytic kinase domain and membrane interaction; I will discuss this in the following sections.

Direct interaction of catalytic kinase domain with anionic membrane

Kinase domain (kinaseD) is bi-lobal and contains an N- and a C-lobe (Figure 2). N-lobe has key functional elements such as the catalytic nucleotide-binding site, activation segment and a regulatory αC-helix the orientation of which is one of the key factors that governs the active/inactive states of the kinase. Literature reports a number of structural studies of kinases in which the catalytic domain (kinaseD) interacts directly with the membrane. These include: EphA2 [12], PI3K [14], EGFR [15], cRaf [13], mTOR [13], FAK [37], KSR [18] and PIP kinase (Figure 3). I discuss each of these kinases briefly with emphasis on the kinaseD-membrane interaction below.

Figure 2.

A canonical structure of the kinase domain (kinaseD) showing two lobes: N- (yellow) and C-lobe (blue). The regulatory αC helix is shown in different orientations that governs the active/inactive state of the kinaseD, among other factors. The residues in sticks show the DFG motif whose alignment is modulated by αC helix

Figure 3.

The domain architecture of kinases discussed in this review in which kinaseD-membrane interactions are reported

EphA2

The largest subfamily of receptor tyrosine kinases (RTK) is erythropoietin-producing hepatoma (EphA) receptors with 14 members each classified as either an A or B-type of receptors [38,39]. EphA’s are found overexpressed in tumours, ovarian cancers, prostrate adenocarcinomas and others [40]. The ligands of EphA receptors are ephrins that are tethered to the membrane [38]. EphA is a membrane-localized kinase containing a single-pass transmembrane domain (TM). Binding of EphA with ephrins occurs extracellularly and is communicated via extracellular domains, TM domain to the intracellular juxtamembrane and kinaseD where the final functional response takes place [39,41] (Figure 3). While the full activity of the kinase is achieved by phosphorylation of phosphorylatable residues in the activation segment of kinaseD, the phosphorylation of two conserved tyrosines within the JM domain is responsible for the release of auto-inhibition [39,42]. Otherwise, in its inactive auto-inhibited state the kinase domain is bound directly with the JM [43] showing also the modulatory role of JM that forms the first regulatory step in kinase activity.

A liposome pull-down assay using varying lipid compositions showed an inherent ability of the kinase domain of EphA2 to preferentially bind with PIP2 anionic phospholipid [12]. Interestingly, the long microsecond-level simulation of EphA2 kinase domain bound to PIP2-containing membrane via JM and TM domain showed that an extended JM domain folds within hundreds of nanoseconds and pulls the kinaseD towards the membrane. KinaseD of EphA2 favours two predominant modes of orientations with respect to the PIP2-rich membrane model [12]. In mode1, the activation segment (containing phosphorylatable residues) is solvent-exposed and is therefore accessible to undergo phosphorylation. In contrast, both N- and C-lobe interact directly with the membrane thereby making the activation loop inaccessible in mode2. From the multiple simulations performed majority of kinaseD-membrane orientations preferred mode1 and the protein-membrane interface was formed by interactions between basic residues of N-lobe: H609, R615, K617, K629, K633, K638 and K639 with anionic lipid [12]. Mutation of these residues resulted in shifting the predominant mode of interaction from mode1 (activation segment exposed) to mode2 (activation segment occluded). In contrast, C-lobe residues R790, Y791, K792, R858 interact with membrane in mode2. The accessibility of the activation loop is directly related to the extent of phosphorylation and hence kinase activity appears to be regulated by interaction with specific phospholipids within the cell membrane.

The preferential modes of EphA2 kinase domain orientations bound to a PIP2-containing model membrane shares an interesting similarity with the preferred orientations of KRAS attached at an anionic membrane [6]. Interestingly, it appears that the membrane plays a direct regulatory role by either occluding or exposing the active/interacting surface of both, the kinase domain and RAS. JM domain is unstructured, flexible and contains basic residues (apt for membrane interaction) that may facilitate its interaction with the membrane and is shown to drive the association between the kinaseD and membrane [43]. The HVR of KRAS is predominantly flexible, contains polybasic cluster and is shown to govern the orientations adopted by KRAS with respect to PS-containing membrane [6,28,29].

FAK

Focal adhesion kinase (FAK) is a protein tyrosine kinase involved in a variety of cellular signalling pathways [44]. The multi-domain FAK kinase constitutes following domains: FERM, kinase domain, Pro-rich region and a FAT domain (Figure 3). FERM and kinaseD are linked via a linker of ~50 amino acids. In its cytosolic auto-inhibited state the catalytic kinaseD of FAK is masked by the FERM domain [45]. The direct interaction between FERM and PIP2-rich membrane micro-domain releases the auto-inhibition thereby relieving the sequestered kinase domain [46,47] The phosphorylation of Y397 that lies in the linker region results in partial release of the kinaseD autoinhibition. Src docking further phosphorylates the Y576 and Y577 lying in the activation segment of the kinaseD thereby leading to the full activation [48]. FERM interacts with PIP2 directly via a basic patch KAKTLRK [46,49].

Using long-scale microsecond-level simulations two independent computational studies showed that kinaseD of FAK kinase interacts directly with the anionic membrane and likely regulates the activity of the kinase by modulating its phosphorylation [49,50]. Localization of the FAK kinase on the cell membrane via FERM-PIP2 interaction that occurs through a PIP2 binding motif of FERM involving four basic residues K216, K218, R221 and K222 (from the patch KAKTLRK) releases the sequestered kinaseD. The kinase domain made direct interactions with the membrane via residues in the vicinity of the ATP-binding region such as R508, R514, K515, K621 and K627 [50]. Simulation of the isolated (autoinhibition released) kinaseD of the FAK kinase preferred two orientations, state I and III, with respect to an anionic membrane model [49]. The sets of residues stabilizing the two states include R426, K587, K621 and R413, R541, R665, respectively. While the active site is occluded by the membrane in state I; it is solvent-exposed and thus available for phosphorylation in state III.

Interestingly, the binary complex of FERM-kinaseD adopt similar orientations on the anionic membrane as adopted by isolated kinaseD that is states I and III [49]. In contrast, auto-inhibition relieved (simulated as isolated FERM domain) FERM interacts with the membrane via the same surface that otherwise sequesters kinase domain that is its auto-inhibition interface [49]. Surprisingly, the isolated FERM domain and membrane interaction does not occur via its PIP2-binding motif, KAKTLRK but rather via the interface that interacts with the kinase domain. This suggests that the reactive surface (the one forming interface with the kinase domain) of FERM acts as a hot-spot and competes between the membrane binding and protein-protein interaction. In addition, it is likely that the orientation of the FERM-kinaseD complex on the anionic membrane is largely governed by the orientational preference of the kinaseD alone with respect to the membrane and not driven much by the FERM domain.

PIP Kinase

PIP kinases are involved in the synthesis of the doubly phosphorylated phosphoinositol (PI) from mono-phosphorylated PI derivatives [51,52]. While the three types of PIPKs share sequence identity within the kinase domain yet they lead to differential outputs [52]. For example, type I and II PIPK generates PI(4,5)P from PI4P and PI5P, respectively. PIP5K belongs to the type I kinase and plays an important role in a wide variety of signalling and trafficking and is localized on the plasma membrane. A crystallographic study reported a structural motif, DLKGS, unique to the PIPK family that is inserted between the N- and C-lobes of the kinase domain and proposed this motif as a PIP binding motif (PIPBM) [53]. It was also shown that the activation segment and an interplay between the PIPBM and the activation segment plays an important role in the substrate specificity and phosphorylation of the PIPKs [53,54].

A microsecond-level computational study investigated the conformational and structural dynamics of PIP5KA1, a type I PIPK, and showed that PIP5KA1 distinctly binds a substrate-containing anionic membrane model (PC/PS/PI4P) and neither PC alone membrane nor PC in combination with the PS [16]. In the PI4P membrane, the DLK₂₃₈GSXXXR₂₄₄ motif and the activation loop made direct and stable contacts with the membrane and preferred two distinct modes of interactions with respect to the membrane [16]. In mode 1, the activation loop alone is in contact with the PI4P lipids which undergo reorientations on the membrane to adopt a more ‘productive’ mode 2 where the PIPBM, the region flanking the PIPBM (DLKGSXXXR) and the activation loop interact with the PI4P molecules [16]. It has been shown that the lipid specificity is encoded in the defined structures/conformations adopted by the otherwise flexible peptides and is not merely driven by non-specific electrostatic interactions [27]. This combined with the unresolved density of the activation segment in the crystal structure of PIPK suggests a likelihood of a preferred conformation of activation segment when bound to PI4P lipids. This is further supported by the adoption of an alpha-helix by the activation segment in a DMPC micellar environment ([16] and references therein). The dependence of orientation modes (i.e. either mode1 or mode2) on the membrane interaction of activation segment shows a likely involvement of the unstructured region in differential membrane orientation modes. The coarse-grained simulations of the dimeric PIP5K1A on the membrane surface showed that only one protomer interacts with the membrane at a time suggesting that membrane distortion or major protein conformational changes may be required for the dual site binding of dimeric PIP5K1A on the membrane surface.

PI3K

The PI3K are multi-domain signalling proteins that phosphorylate the 3-hydroxyl of phosphoinositides [55]. Mammalian PI3Ks are divided into three classes of which class I PI3Ks are a downstream effector of RAS enzyme. There are two types of class I PI3Ks: IA and IB. Class IA consists of a catalytic (p110α,δ,β) and a regulatory subunit (p85) that are large multi-domain proteins. p110 contains the catalytic kinase domain in addition to an ABD (adaptor-binding domain), RBD (Ras-binding domain) and C2 domain [56,57]. ABD mediates a tight interaction with the p85 subunit, RBD binds to Ras, C2 domain forms inhibitory contacts with p85 and is responsible for membrane binding and the kinase domain carries out the catalytic function of phosphorylating PIP2 to PIP3 (Figure 3). Early structural studies of PI3K predicted a putative membrane binding interface at the crevice between the N- and C-lobe of the kinaseD [56]. Based upon simple structural models of PI3K’s interaction with membrane embedded Ras, studies suggested the involvement of specific kinaseD residues for example F902, K903, K756, Y757, F975, K973 [57] and K723, K729, K863 and K867 [58] in the protein-membrane binding. More direct evidence of the role of kinaseD-membrane interaction came from a hydrogen-deuterium exchange mass spectroscopy (HDX-MS) of PI3K in the presence of a covalently-linked HRAS reconstituted in a vesicle containing PIP2/PC/PS/PE [14]. The HDX-MS showed that while both N- and C-lobes of the kinaseD has potential to bind with the membrane (N-lobe: residues 720 to 744 and C-lobe: residues 880 to 890); the N-lobe of the kinase domain showed large changes in the HDX-MX in the presence of the reconstituted HRAS vesicles. The probable functionally relevant membrane-binding motif involved the following N-lobe residues: K720, K723, K724, K729, K733, R740 and R741. Some of these residues are in good agreement with previous predictions [58]. The activation of PI3K by Ras is enhanced by membrane interaction [14]. Quite interestingly, a direct Ras-independent kinaseD-membrane interaction is reported as one of the major factors controlling the oncogenicity of a PI3K mutant, H1047 R [59]. According to this study, the Arg mutant recruits PI3K on the plasma membrane via Arg-membrane interaction and results in the constitutive activation of the PI3K pathway in a Ras-independent manner.

EGFR

Epidermal growth factor receptor (EGFR) is a receptor tyrosine kinase (RTK) that activates downstream signalling pathways such as RAS/MAPK, PI3K/AKT upon ligand stimulation [60]. EGF binding at the extracellular region results in conformational changes that are communicated via transmembrane segment to the catalytic kinase domain. While there are several excellent articles explaining the mechanism of the EGFR receptor activation (e.g. [61,62]) here I will only focus on selected structural studies that discuss its catalytic domain.

Ligand-induced dimerization of the intracellular kinase domains of two EGFR protomers is required for its function. Jura et al [63] showed that the active dimer of kinase domain is asymmetrical and is formed by a direct mediation from the C-terminal of the juxtamembrane domain (JM). This observation is similar to the related HER2 receptors and also FAK kinases discussed above [64]. In contrast, ligand-independent dimerization of kinase domains results in an inactive symmetrical dimer [63]. A millisecond level MD simulation showed that monomer and inactive symmetrical dimer of kinase domains interact directly with the anionic membrane via N-lobe or both N- and C-lobes thereby occluding the interface containing the ATP-binding site and the activation segment [15]. The burial of activation segment prevents the activation and renders monomer and the symmetrical dimer inactive. In contrast, in case of the asymmetrical active dimer activation segment is solvent-exposed and dynamic and hence available for its substrates [15]. Quite interestingly, the kinase domain of EGFR in its non-productive or inactive orientation buries its regulatory αC helix as well [15]. A similar burial of regulatory helix has been reported in case of monomeric cRaf due to interactions with its PA-containing membrane [13].

cRaf

Raf kinase is a downstream effector of RAS [65]. Cytosolic Raf exist in an auto-inhibited inactive state which is released upon Ras-dependent recruitment of Raf kinase to the plasma membrane [66]. Previous studies hinted towards a possibility of the role of phospholipids in regulating the function of Raf kinase once Raf is membrane-localized and the kinase domain is released from its auto-inhibited inactive state [18,67]. Structure and sequence based bioinformatics analysis combined with a microsecond-level all-atom simulation of the cRaf’s kinaseD bound to an anionic membrane containing POPC and POPA showed a contiguous stretch of residues formed by distantly located functionally critical regions to form a membrane binding motif [13]. This involved the residues from the regulatory αC helix, activation segment and the N-terminal acidic region (NtA). The polybasic region from the αC-helix, RKTR and its role in membrane interaction is in agreement with a previous study [18]. The occlusion of the activation segment and the regulatory helix in cRaf may be a predominant state thereby resulting in inactive monomers (Figure 4(a)). A pseudo kinase KSR showed relatively reduced interaction with lipid vesicles as compared to the cRaf likely due to reduced charge in its αC-helix, RQTR [18]. It is known that the kinases, in general, requires tight structural re-arrangements to become active such as and among others orientation of αC-helix and alignment of specific residues to form R and C-spines. A previous study showed that the R-spine of the cRaf kinase domain is likely extended by a conserved Trp residue of the NtA [68] and quite interestingly the same residue was found embedded in the membrane in another study by Prakash et al [13] (Figure 4(b)). Based upon this one can further speculate the functional role of membrane possibly in orienting the R-spine residues with a possibility of extension provided by one of the membrane-embedded conserved Trp residue of NtA (Figure 4(b)).

Figure 4.

(left) A cartoon illustration of cRaf kinase domain inactive orientation with respect to an anionic membrane. (right) The R-spine is extended by membrane-embedded Trp of the N-terminal acidic region (NtA)

mTOR

Farnesylated Rheb proteins are targeted to endomembrane compartments [69,70] and GTP-bound farnesylated Rheb activates the insulin/TOR/S6K1 pathway by interacting with mTORC1 [69,71,72], one of the isoforms of mammalian target of rapamycin (mTOR). mTOR, is a Rheb effector made up of: N-terminal HEAT domain, FAT domain, kinase and the C-terminal FATC domains (Figure 3). The N-lobe of mTOR-kinaseD consists of an additional FRB (FKBP12-Rapamycin binding) domain in addition to a few other insertions. FRB domain is a Rapamycin binding domain. An experimental lipid vesicle binding study showed that the FRB domains binds PA-specific vesicles and that mutation of the key residue involved in the interaction shows a 60% reduction in its activity likely due to the abolished membrane interaction [17]. A combined bioinformatics and simulation study showed that that the FRB domain of the mTOR contains a cluster of polybasic residues that likely form a membrane-binding patch [13]: H2106, R2109, R2110, K2113 and this was shown to be PA-specific.

A generalized role of membrane in regulating the activity of the key proteins

Accumulating evidence from experimental and computational studies reveal a number of kinases in which the direct interaction between the catalytic kinase domain and the membrane likely regulates the kinase activity [12,14,16,18,49]. A few features that are immediately noticeable among these are listed below:

(1) all kinases discussed here contain either a well-defined membrane-interacting protein domain (FERM, C2 domains in FAK, PI3 K, cRaf, mTOR) or a single-pass transmembrane-domain (EphA2, EGFR) (Figure 3). This suggests that the kinaseD-membrane interaction is likely not the primary site of membrane attachment and is transient yet plays a crucial role in regulating the kinase activity. Based upon the studies highlighted here there are two preferred modes of interaction of kinase domains with respect to the membrane. One, where the active surface of the catalytic domain (activation segment, regulatory αC-helix) is occluded by the membrane and another where it is solvent-exposed and available to undergo auto- or protein-mediated phosphorylation (Figure 4(a)). This is similar to auto-inhibition where the active surface of the kinase domain is sequestered by either a preceding protein domain (e.g. CRD and FERM in cRaf and FAK, respectively) or flexible loop (e.g. juxtamembrane domain). In other words, one can speculate that membrane acts as a second regulatory step during the activation of the kinase after release of the auto-inhibition. Membrane may likely achieve this via either specific signalling lipids, by regulating the concentration of the membrane species such as cholesterol or other yet unknown factors. Thus, it can be proposed that kinase function and activity is regulated by a protein-protein interaction (auto-inhibition) followed by a protein-membrane interaction (kinaseD-membrane).

Adoption of preferred orientations by the kinaseD with respect to membrane shares striking similarities with the already known preferred orientation states of small GTPase such as KRAS. Of the two preferred states of KRAS one is inactive (functional switch regions occluded by membrane) and another active (switches are solvent-exposed) (Figure 1). Another similarity lies in a competition of the hotspot surface between membrane interaction and protein-protein interaction (Figure 8 of ref # [13]). For instance, in KRAS the same interface that predominantly interacts with the membrane is required for the formation of KRAS dimers [13]. cRaf dimers are formed by the αC helix which directly interacts with the membrane in case of cRaf monomer [13]. Similarly, in FAK kinases the FERM domain interacts with the same surface that otherwise sequesters the kinaseD in an auto-inhibited state. Therefore, both, small GTPases and certain kinases exhibits a competition between the protein-protein and protein-membrane binding by their hotspot surfaces (see Section B2).

(2) The active mode of interaction almost always involves the N-lobe which is expected given N-lobe has the regulatory helix, catalytic site and activation segment (Figure 5).

Figure 5.

Cartoon illustration of kinase domains of selected kinases interacting with a heterogeneous membrane (different colour circles represent different lipids in the membrane). The highlighted residues are those involved in direct interaction with the membrane mainly in the more ‘productive’ mode of the kinase domain and lies in its N-lobe. The N-lobe is yellow, C-lobe is blue, and pink shows insertion in the mTOR that is FRB domain

(3) There exist either a short flexible (FAK, EGFR, EphA2) or a long region (PI3K, cRaf) upstream of the N-lobe (Figure 3). In case of latter, while the long region is structured and forms a helical domain in PI3K the structure of the linker region of cRaf is yet unknown. The flexible juxtamembrane domain contains basic residues that facilitates its membrane interaction and modulates the activation of the kinase domain (e.g. in FAK and EGFR; Sections B2 and B5). The hypervariable region has been reported to modulate the orientation dynamics of KRAS embedded in an anionic membrane [6,28,29,35].

(4) The insertions of sub-structures within the kinase domains houses membrane-binding motifs (e.g. DLKGSxxxR in PIPK and HRRK in FRB domain of mTOR).

The role of membrane in governing the activity of kinases by direct modulation of the catalytic kinase domain will likely serve as a promising therapeutic area in future drug discovery efforts. This is especially true in the case of challenging targets such as Raf kinases. Recent studies have shown that membrane modulates small-molecule binding to an extremely challenging drug target KRAS which is likely one of the primary reason in obtaining better binding affinities as compared to their solution counterparts [73,74]. The mechanism of action likely involves stabilization of the signalling-incompetent orientation of membrane-bound KRAS [73]. Displacing the regulatory αC-helix to adopt an orientation from an active to inactive state via small-molecule inhibitors has been an attractive strategy during the design of kinase inhibitors [75]. Stabilization of inactive orientation where αC helix interacts with the membrane via small-molecules may serve as an alternative therapeutic strategy to design inhibitors for kinases such as Raf. In order to gain further insight into the mechanism of the action of for instance Raf kinases; while the incorporation of membrane may enhance the stability of otherwise flexible linker region in cRaf – active model of the kinase domain bound to membrane may be generated utilizing a similar approach of membrane-anchoring as mentioned in a recent study [13]. This is true especially in case of cRaf where the structural information especially with respect to its long linker region becomes a limiting factor in its complete understanding. The regulatory role of membrane likely possesses hidden therapeutic relevance that can be harnessed as a promising alternative strategy to combat RAS/Raf driven carcinomas.

Disclosure statement

No potential conflict of interest was reported by the author.