Conformational control and regulation of the pseudokinase KSR via small molecule binding interactions

Abstract

Pseudokinases often operate through functionally related enzymes and receptors. A prime example is the pseudokinase KSR (Kinase Suppressor of RAS), which can act as both an amplifier and inhibitor of members in the RAS-MAPK (Mitogen Activated Protein Kinase) signaling pathway. KSR is structurally related to the active RAF kinases over multiple domains; moreover, the pseudokinase domain of KSR forms physical and regulatory complexes with both RAF and MEK through distinct interfaces. Characterization of small molecule interactions on KSR has been used to uncover novel chemical tools and understand the mechanism of action of clinical drugs. Here, we elaborate on assays and structural methods for measuring binding at orthosteric and interfacial binding sites on KSR. These distinct small molecule pockets provide therapeutic paths for targeting KSR1 and KSR2 pseudokinases in disease, including in RAS and RAF mutant cancers.

1. KSR identification

Kinase Suppressor of RAS (KSR) was originally identified in forward genetic screens conducted in the fly and worm as a modulator of RAS signaling (Kornfeld, Hom, & Horvitz, 1995; Sundaram & Han, 1995; Therrien et al., 1995). Specifically, KSR was identified by virtue of mutations that suppress phenotypes associated with oncogenic RAS signaling. Epistatic analysis further suggested that KSR functioned downstream of RAS but upstream or parallel to RAF (Karim et al., 1996; Sundaram & Han, 1995). We now know that KSR and RAF are highly related, with KSR characterized as the pseudokinase homolog of the RAF kinases (Lavoie & Therrien, 2015). Indeed, all metazoan species that we have analyzed possess at least a single KSR and RAF paralog (Khan et al., 2020; Kondo et al., 2019; Lavoie et al., 2018; Liau et al., 2020; Mysore et al., 2021; Park et al., 2019), demonstrating broad coexistence and potentially conserved pseudokinase-kinase functions within the RAS-mitogen-activated protein kinase/extracellular signal-regulated kinase (MAPK/ERK) signaling system.

2. KSR domain structure

Humans possess two KSRs (KSR1 and KSR2) and three RAFs (ARAF, BRAF, and CRAF). Below we focus on KSR1 and BRAF as exemplary, but also highlight some general features within each subfamily. KSR and RAF proteins share several elements including C-terminal kinase domains (Fig. 1). The N-terminal regions differ whereby the RAF kinases possess the prototypical RAS-binding domain (RBD), which participates in direct interactions with active GTP-bound RAS (Herrmann, Martin, & Wittinghofer, 1995; Nassar et al., 1995; Van Aelst, Barr, Marcus, Polverino, & Wigler, 1993; Weber, Slupsky, Kalmes, & Rapp, 2001; Zhang et al., 1993). Among the RAF homologs, BRAF also contains a specific N-terminal region (BRS) that is involved in homotypic and heterotypic dimerization interactions, including with KSR1 or KSR2 (Lavoie et al., 2018). In contrast, the KSR proteins possess an N-terminal coiled-coiled sterile alpha motif (CC-SAM) that has been shown to participate in targeting KSR1 to the plasma membrane (Koveal et al., 2012). Both KSR and RAF kinases possess a cysteine-rich domain (CRD), which has been shown to be required for interactions and also activation of RAF1 by KSR1 (Michaud et al., 1997) as well as to participate directly in binding to RAS (Brtva et al., 1995; Cutler, Stephens, Saracino, & Morrison, 1998; Tran et al., 2021). Recent structures have further demonstrated that the CRD folds back onto the kinase domain in the auto-inhibited structures of BRAF via interactions that are reinforced through a 14-3-3 dimer (Heppner & Eck, 2021; Park et al., 2019). 14-3-3 has also been demonstrated to directly associate with KSR1 through several phosphorylated residues, including Ser297 and Ser392 within the areas flanking the CRD (Cacace et al., 1999). Although not experimentally established, but by analogy to BRAF, these and other phospho-regulatory sites may be important in maintaining an auto-inhibited and putative monomeric conformation of full-length KSR.

Fig. 1.

Domain structure, regulatory motifs, and interaction sites between RAS/RAF/KSR/MEK/14-3-3. (A) The domain structures of human KRAS, BRAF, KSR1, and MEK1 are shown and numbered according to the listed Uniprot accession numbers. Domain boundaries are marked based on structural data available in the PDB or are based on structure-based sequence alignments. 14-3-3 binding sites are based on published studies (Müller, Ory, Copeland, Piwnica-Worms, & Morrison, 2001; Xing, Kornfeld, & Muslin, 1997); in BRAF, these binding sites have been recently confirmed via structural studies (Kondo et al., 2019; Liau, Wendorff, et al., 2020; Park et al., 2019). In KSR1, the 14-3-3 binding sites are based on experimental data or predicted based on structural alignments (Cacace et al., 1999). (B) A cartoon depiction of the putative signalsome including KSR1 highlighting expected interactions sites based on known structures of fragments, as well as functional data (Brennan et al., 2011; Dhawan, Scopton, & Dar, 2016; Khan et al., 2020; Kondo et al., 2019; Lavoie et al., 2018; Liau, Wendorff, et al., 2020; Park et al., 2019).

3. Higher-order KSR–RAF–MEK complexes: Heterodimers and heterotetramers

At the moment, experimental structures for the “KSR signalosome” are lacking, however, insights from fragments reveal the complexity of binding between distinct components of the RAS-MAPK cascade and imply the potential for overlapping and possible competing interactions between KSR, RAF and MEK to regulate RAS–MAPK pathway output (Khan et al., 2020; Kondo et al., 2019; Lavoie et al., 2018; Liau, Wendorff, et al., 2020; Mysore et al., 2021; Park et al., 2019). Current models suggest that the signalosome occurs when KSR bridges to RAS through heterodimerization with RAF kinases and thereby indirectly through Ras-binding domain (RBD) interactions. Within the larger assembly, KSR directly interacts with both RAF and MEK through the pseudokinase domain (Brennan et al., 2011; Khan et al., 2020; Lavoie et al., 2018; Rajakulendran, Sahmi, Lefrançois, Sicheri, & Therrien, 2009).

Crystal structures have revealed that the pseudokinase domain of KSR1/2 binds to MEK via the lower C-lobe centered on interactions mediated by helix αG and the activation loop (Fig. 2; Khan et al., 2020; Dhawan et al., 2016), whereas binding to RAF occurs through the dimer interface centered on interactions in the N-lobe and helix αC (Rajakulendran et al., 2009). In the inactive state conformation of KSR1 and KSR2, helix αC adopts an OUT configuration (Khan et al., 2020; Brennan et al., 2011). Differences between inhibitor and ATP/AMP-PNP structures have revealed additional variations in the P-loop and N-terminal portion of the activation segment that have highlighted potential intermediate-state conformations of KSR (Dhawan et al., 2016). For example, in previous work we proposed that the ATP-bound state permits the transition from the inactive to active state of KSR through alterations in the conformation of the P-loop and an “induced” locked within the KSR activation segment that maintains the Ser218–Ser222 sites on bound MEK within a protected arrangement (Dhawan et al., 2016). The first characterized inhibitor pocket on KSR is referred to as the orthosteric site as this pocket overlaps directly with the ATP-binding site.

Fig. 2.

Structural features of KSR:MEK and RAF–MEK complexes. Tetrameric assembly of KSR1:MEK (A) and BRAF:MEK (B). The available crystal structures of KSRI-bound MEK (PDB ID: 7JV1; Khan et al., 2020) provide a snap-shot of KSR:KSR homodimerization centered around R665. MEK binds KSR on the flanking sides of the KSR homodimer. The left inset represents the heterodimerization interface between KSR1 and MEK1 centered around the respective helix ⊠G (W831 of KSR1 is shown as spheres) and the activation loops, which form a 2-stranded antiparallel β-sheet. The right inset highlights the location of the KSR1 orthosteric pocket with AMP-PNP bound and interfacial glue pocket. (B) A crystal structure of BRAF:MEK (PDB ID: 4MNE; Haling et al., 2014) highlighting BRAF homodimerization centered around R509. The lower insets highlight the homodimerization interface of KSR1 and BRAF. Whereas KSR1 homodimerization involves an extensive interaction of residues at the N-lobe of the two KSR1 molecules that results in a back-back dimer, two molecules of BRAF are assembled in an inverted fashion from N- to C-lobe. (C) Orthosteric pockets (left;bound to ADP, PDB ID 7JUQ; Khan et al., 2020) and interfacial binding pockets (middle left;bound to APS-2-79, PDB ID 5KKR; Dhawan et al., 2016). In KSR2, APS-2-79 directly projects toward a back pocket that is lined by hydrophobic residues Y714 (helix ⊠C), F725 (strand β4) and F825 (DFG motif). The binding of APS-2-79 results in the restructuring of the activation loop, downstream of the DFG motif (not shown) and disordering of P-loop (Dhawan et al., 2016). Right two panels show interfacial binding pockets of KSR1-MEK (PDB ID 7JV1) and KSR2-MEK (PDB ID 7JUV) occupied with trametiglue. The terminal sulfamide motif of trametiglue occupies the interfacial space between KSR and MEK, with complementarity between trametiglue and KSR spanning the pre-helix ⊠G loop and the tip of the helix ⊠G (Khan et al., 2020).

A secondary pocket on KSR exists directly at the interface for MEK binding and is mediated through the region directly preceding helix αG. We refer to this pocket as the interfacial “glue” site. This site enables binding and direct interactions with allosteric MEK inhibitors, including the clinical anti-cancer drug trametinib and is exemplified by our structures of the KSR1–MEK1 and KSR2–MEK1 complexes bound to a tool compound that we created called trametiglue (Khan et al., 2020). Trametiglue is distinct from trametinib in that the former can form high affinity and glue-like interactions with both KSR–MEK and, as we have speculated, RAF–MEK complexes. Moreover trametiglue is markedly enhanced in terms of potency of growth inhibition in KRAS mutant cell lines relative to trametinib and other MEKi including CH5126766. The enhanced activity of trametiglue has been correlated to reduced “rebound” signaling. This form of adaptive drug resistance has emerged as a major limitation of current MEKi including trametinib (Yaeger & Corcoran, 2019). More recently, structures of BRAF-MEK1 bound to a series of MEKi have provided further evidence for binding of MEKi within RAF-MEK complexes (Pino et al., 2021), and have demonstrated functional agreement of differential binding of MEKi to isolated MEK and BRAF-MEK (Ishii et al., 2013; Lito et al., 2014).

Both KSR and RAF form homotypic dimers through similar interfaces on the N-lobe and helix αC, but via distinct quaternary arrangements. For example, in the inactive state conformations of KSR1 and KSR2, we found that KSR-KSR dimerization is centered on Arg665 and Arg718 in KSR1 and KSR2, respectively (Khan et al., 2020; Brennan et al., 2011). The quaternary arrangement of the KSR-KSR dimer is related to the RAF-RAF side-to-side dimer in that the conserved Arg at the base of helix αC forms part of the dimer interface (Haling et al., 2014; Rajakulendran et al., 2009; Thevakumaran et al., 2015; Wan et al., 2004). However, the RAF-RAF arrangement is rotated relative to the KSR-KSR dimer (Haling et al., 2014; Khan et al., 2020). We previously suggested that KSR and RAF most likely would dimerize in a RAF-RAF-like side-to-side arrangement (Brennan et al., 2011; Dhawan et al., 2016); in this configuration helix αC of KSR would be expected to rotate toward the IN state, which we further predicted would disengage inhibitory interactions of the activation segments between KSR and MEK and thereby release the Ser218–Ser222 sites. We predicted that this release mechanism induced through KSR-RAF heterodimerization would enable phosphorylation on the Ser218–Ser222 sites in MEK by separate RAF protomers; data to support this model was generated through biochemical reconstitution and chemical genetic experiments (Brennan et al., 2011). More recent data using a cellular reconstitution system and point mutations have suggested that catalytic RAF kinases prefer to phosphorylate unbound MEK over KSR-bound MEK, implying that KSR may not physically bridge the RAF→MEK phosphorylation step directly (Lavoie et al., 2018). However, given that RAF also forms similar inhibitory interactions on bound MEK, it is likely that both KSR-bound MEK and RAF-bound MEK can serve as adaptors and scaffolds for separate catalytic RAF protomers within large macromolecular assemblies. Further structural data on RAF–KSR complexes will be needed to better understand how exactly heterodimerization enables phosphorylation on MEK.

4. Is KSR an active kinase?

While KSR possesses a conserved kinase domain, like many pseudokinases, it lacks a physiological substrate and several lines of evidence suggest that its role in RAS-MAPK signaling is predominantly mediated via scaffolding, allostery, and higher order interactions. For example, the mammalian homologs of KSR, including human KSR1 and KSR2, include an arginine residue (R639 in human KSR1 and R692 in human KSR2) at the invariant lysine of subdomain II of the kinase domain; this residue is critical for orienting ATP in the active site of canonical kinases and mutations to arginine have been shown to severely compromise phospho-transferase activity of canonical kinases (e.g., Carrera, Alexandrov, & Roberts, 1993; Iyer, Garrod, Woods, & Taylor, 2005; Robinson et al., 1996). That the mammalian homologs of KSR naturally possess a lysine-to-arginine substitution at this critical position suggested that KSR may function independent of catalytic activity, and this notion was in fact supported in early studies. Stewart et al. were able to demonstrate that several catalytic site mutants of KSR, including at the conserved aspartic acid residue in the DFG motif, were able to rescue KSR loss-of-function in animals as evidenced by RAS-dependent vulval induction in C. elegans (Stewart et al., 1999). Notably, a KSR allele within the KSR-RAF heterodimerization interface mutated could not rescue, supporting the specificity of the tested catalytic site mutants and further suggesting that KSR phospho-transferase activity is either weak or unnecessary to support RAS signaling. However, while some studies suggested that KSR does not utilize catalytic function, other studies in fact argued toward the importance of KSR catalytic activity at least in some contexts (e.g., Therrien, Michaud, Rubin, & Morrison, 1996; Zhang et al., 1997).

In support of catalytic function, we previously demonstrated that KSR can phosphorylate MEK1 in vitro using purified KSR2:MEK1 complexes (Brennan et al., 2011). Intriguingly, while the efficiency and stoichiometry of KSR2→MEK1 phosphorylation was low, we identified that heterodimerization with BRAF could accelerate KSR2 activity, again in vitro using reconstitution assays. However, further studies testing a quadruple site mutant of MEK1 (S18A + T23A + S24A + S72A) that removed all of the identified KSR phosphorylation sites found that the mutant could be activated and signal downstream to ERK at levels that were essentially identical to wild-type MEK1, leading us to surmise that phosphorylation on KSR-mediated sites within MEK1 were dispensable for MAPK signaling (Dhawan et al., 2016). Moreover, we conducted experiments using two distinct types of inhibitors (ASC24 and APS-2-79) and demonstrated that inhibition of KSR2 catalytic activity alone was insufficient to antagonize KSR2-accelerated BRAF→MEK phosphorylation in vitro (Dhawan et al., 2016). Shaw, Taylor and colleagues identified an ATP-binding site mutant (A587F in mKSR1) that has been modeled to fill the adenosine pocket in KSR so as to stabilize an active and closed conformation of the pseudokinase domain (Filbert, Taylor, & Shaw, 2011). This mutant has been studied in both KSR and RAF family kinases (Filbert et al., 2011; Hu et al., 2013), and suggested to mimic the ATP-bound state through stabilization of the hydrophobic spine; this mutant would also be expected to impede conformational changes within the pseudokinase domain of KSR, including by disfavoring switching between the open and closed states of the pseudokinase domain. Intriguingly, the A587F mutant in KSR1 could form complexes with both RAF and MEK but not support MAPK/ERK signaling as assessed by cell transformation and colony formation (Filbert et al., 2011).

In summary, no one experiment has clearly answered the question of whether KSR is a physiologically relevant active kinase or phospho-transferase in the RAS-MAPK pathway. However, collective data from many studies suggest that the role of the KSR pseudokinase is primarily linked to its conformational changes that mirror changes pervasive across many active kinases. One such change includes an orientation of helix αC relative to the ATP-binding pocket. Data to support that KSR does not simply serve as a static scaffold for the recruitment of MEK to RAF have been provided through elegant structural and mutagenesis studies suggesting that heterodimerization of KSR and RAF first requires the formation of KSR–MEK complexes to enable RAF-mediated phosphorylation of KSR-free MEK on the essential activating residues Ser218 and Ser222 (Lavoie et al., 2018). ATP binding and/or hydrolysis may play a permissive role in the mechanism and formation of KSR–RAF–MEK complexes, which could explain some of the confounding results testing the importance of KSR catalytic function. Indeed, one possibility is that conformational switching of the KSR pseudokinase domain, and thereby binding to MEK and RAF, is regulated through ATP binding and release cycles that are coupled to ATP hydrolysis but not through phospho-transferase activity per se. Consistent with the importance of ATP in the RAS-MAPK pathway, nucleotide binding has been demonstrated to modulate side-to-side dimerization of BRAF in the absence of bound 14-3-3 (Liau et al., 2020; Liau, Wendorff, et al., 2020); and further a C-terminal regulatory segment of BRAF has been suggested to also regulate ATP binding (Kondo et al., 2019; Kondo, Paul, Subramaniam, & Kuriyan, 2021). Likewise ATP, through its binding and hydrolysis, could play an important role in KSR-dependent MAPK signaling by acting as an endogenous cofactor to regulate conformational changes within the active site that are linked to protein– protein interactions or possibly heterodimerization toward RAF kinases. Such a model would suggest that the pseudokinase domain in KSR shares regulatory principles with enzymes such as GTPases, like K-RAS, or other ATPases. Furthermore, such a model would suggest that small molecules that lock and thereby disrupt conformational cycles in KSR may block specific steps in the RAS-MAPK pathway. We have in fact found two classes of molecules—one targeting the orthosteric site on KSR and a second at the interface of KSR–MEK complexes—that can modulate RAS signaling (Dhawan et al., 2016; Khan et al., 2020). We next describe purification methods to isolate KSR1:MEK1 and KSR2:MEK1 complexes for structural studies. We further describe assays and methods to characterize small molecule binding interactions at the orthosteric and interfacial bindings sites on KSR both in vitro and in live cells.

5. Purification of KSR1–MEK1 and KSR2–MEK1 for biochemical assays and cocrystallography

5.1. Production of KSR:MEK in insect cells

We routinely use insect cell expression systems to produce KSR1:MEK1 and KSR2:MEK1 complexes, and have been successful in isolating full-length MEK1 bound to the pseudokinase domains of KSR1 [residues 591–899] or KSR2 [634–950]. However, for cocrystallography, we most often use a construct that lacks the first 34 residues at the N-terminus of MEK1. We have also utilized mutations that improve the stability of MEK1, including Ser298Asn/Ser299Lys/Tyr300Phe (Fischmann et al., 2009). We have reported the isolation of both sets of complexes using our purification strategies in detail previously (Brennan et al., 2011; Dhawan, Scopton, & Dar, 2016; Khan et al., 2020), which we outline in Fig. 3. Here, we provide additional practical guidelines.

Fig. 3.

Purification of the KSR1:MEK1 complex for biochemical and structural analysis. (A) Copurification of the KSR1 pseudokinase domain and MEK1 kinase domain is shown as an example (Khan et al., 2020). KSR1 [residues 591–899] and MEK1 [residues 33–393] were coexpressed as His-tagged proteins using the Sf21 insect cell system. (B) Cobaltaffinity purification of KSR1-MEK1 complex using the gravity-flow column (left);the corresponding 8–16% SDS-gel is shown on the right. (C) Partially-purified KSR1:MEK1 complex is further purified using anion exchange chromatography to remove excessive MEK. Several impurities are detected on the 8–16% SDS gel (below). (D) KSR1-MEK1 (1:1) complex is obtained after size-exclusion chromatography. Either the KSR1:MEK1 complex can be isolated as His-tagged protein (left) or KSR1-MEK1 can be treated with trypsin (1:1000, incubated overnight at 4 °C) to remove partially unfolded proteins and flexible loops, and hence highly homogenous proteins are obtained for crystallization. SDS gels are shown below, and the protein fractions corresponding to the tetrameric peak (150 kDa—see Fig. 2 for a tetrameric assembly) are pooled, and concentrated for subsequent analysis. (E) The flow-chart is provided to illustrate the procedure for determining structures of the KSR1:MEK1 complexes bound to interfacial binders, including trametinib (Khan et al., 2020). Similar procedures can be used for solving structures of other MEKi. Cocrystallization of the KSR:MEK1 complex in the presence of orthosteric binders can also be used to solve structures (Brennan et al., 2011; Dhawan, Scopton, & Dar, 2016). (left to right) Crystals of purified KSR1:MEK1 or KSR2:MEK1 complexes are first grown in the presence of 5mM AMP-PNP, and typically first appear within 24 h at room temperature. Crystals of KSR2-MEK1 are shown as an example. Next, individual crystals (72–96 h old) are transferred into a fresh drop containing crystallization solution (20% PEG2000, 200mM Mg acetate, 100mM MES pH 6.5) supplemented with 5 mM AMP-PNP and 1 mM trametinib. The drop is re-sealed and further incubated for 48–96 h at room temperature. Next, crystals are harvested and flashfrozen in liquid nitrogen. A representative X-ray diffraction image is shown, followed by a model of the inhibitor binding pocket. The electron density map of KSR1:MEK1 in complex with AMP-PNP and trametinib is displayed (PDB ID: 7JUX).

To generate baculovirus and determine viral titres, we follow standard protocols, detailed in Jarvis (2014). Virus titres can be determined using plaque assays or alternatively using other approaches (Mulvania, Hayes, & Hedin, 2004). Several contract research organizations, including Expression Systems, offer titering as a service. We also test for a functional multiplicity of infection (MOI) of 1, which should be an amount of virus solution that when added to insect cells allows for a single doubling before growth reaches a plateau.

In terms of insect cell lines, we most often use the Sf21 line; however, it is useful to test new constructs (e.g., mutant variants) in additional lines, including Sf9 and Hi5, as the yield of soluble complex from each can vary significantly. Our typical growth format for large scale infection includes 4L of culture, with each liter split into a 2.4L baffled and aerated flask that is grown at 27 °C and at 125rpm shaking. We also find that the highest yields of protein occur after 3–4 passages of cells. Each passage is conducted when cells reach 3–4 × 10⁶cells/mL.

Equipment:

• Biosafety cabinet (Labconco, Cat# 302420001)

• Countess III FL automated cell counter (ThermoFisher, Cat# AMQAF2000)

• P2, P20, P200, P1000 pipettes (Gilson, Cat# F144054M, F144056M, F144058M, F144059M, respectively)

• Serological Pipet-Aid (ThermoFisher, Cat# 03–840-313)

• Sorvall LYNX 6000 centrifuge (ThermoFisher, Cat# 75006590)

• Nalgene 1L Super-Speed Centrifuge Bottles with Sealing Closure (ThermoFisher, Cat# 05-564-25)

• Sonicator (Branson 450 Digital Sonifier)

• Benchtop microcentrifuge (ThermoFisher, Cat# 75002447)

• Benchtop centrifuge (ThermoFisher, Cat# 75008801)

• NanoDrop UV–vis Spectrophotometer (ThermoFisher, Cat# ND-ONEC-W)

• Superloop (Cytiva, Cat# 18102385)

• HiTrap Q anion-exchange column (Cytiva, Cat#17505301)

• HiTrap SP cation exchange column (Cytiva, Cat# 17515701)

• Size exclusion chromatography column: SD200 10/30 GL (Cytiva, Cat# 28990944)

• ÄKTA pure protein purification system (Cytiva, Cat# 29018224)

Consumables:

• Celltreat Scientific Erlenmeyer Flasks with baffled bottom, sterile, various sizes (e.g., ThermoFisher, Cat# 50-202-096)

• Pierce Protease Inhibitor Tablets, EDTA-free (Thermo Fisher, Cat# A32965)

• Cobalt agarose beads (Gold Bio, Cat# H-310-100)

• Gravity flow columns (Bio-rad, Cat# 7372512)

• 2-way Stopcock valve (Bio-rad, Cat# 7328102)

• Slide-A-Lyzer™ Dialysis Cassettes (ThermoFisher, Cat# 66030)

• Amicon^® Ultra Centrifugal Filter Units: various sizes (Millipor, Cat# UFC903008)

• 8–16% Criterion™ TGX™ Precast Midi Protein Gel (Bio-rad, Cat# 5671105)

• Pellet-wash buffer: 20mM Tris-HCl, pH 7.5, 200mM NaCl, 5% Glycerol

• Lysis buffer: 50mM Tris-HCl, pH 7.5, 200mM NaCl, 10% Glycerol, 10mM Imidazole; optional 0.25% Triton X-100, 0.25% Tween 20. To add immediately prior to lysis: 5mM beta mercaptoethanol, and 1 tablet of EDTA-free protease inhibitor

• Immobilized metal affinity chromatography (IMAC) wash buffer: 20mM Tris-HCl pH 7.5, 200mM NaCl, 10% glycerol, 10mM Imidazole

• IMAC elution buffer: 20mM Tris-HCl pH 7.5, 200mM NaCl, 10% glycerol, 500mM Imidazole

• Dialysis buffer: 20mM Tris-HCl pH 7.5, 200mM NaCl, 10% glycerol, 5mM DTT

• Ion-exchange (IEX) buffer A: 20mM Tris-HCl, pH 7.5, 10% Glycerol

• IEX buffer B: 20mM Tris-HCl, pH 7.5, 10% Glycerol, 500mM NaCl

• Size exclusion chromatography (SEC buffer: 20mM Tris-HCl pH 7.5, 200mM NaCl, 5mM DTT, 1mM TCEP

• Adenylyl-imidodiphosphate (AMP-PNP) (Sigma, Cat# 10102547001)

• Trametinib 10mM solution (Selleckchem, Cat# S2673)

• Crystallization condition C9 from JBS Kinase Screen (Jena Biosciences, Cat# CS-204 L)

• TPCK treated trypsin (ThermoScientific, Cat# 20233)

Here, we are detailing the steps for production of KSR1:MEK1 and KSR2:

MEK1 complexes:

1. Infection: On the day before an infection, for 1L of cell culture per growth flask, inoculate at 0.5–1 × 10⁶ cells/mL (counted using trypan blue dye). Next day, when cells are at 1–2 × 10⁶ cells/mL, add the virus at a MOI of ~1. (Note: At the beginning of an infection, cells are usually 90–95% viable. After 72h, when cells are harvested, we can measure viability, as low as ~50%. This decrease in viability is often a marker of a strong infection and can be associated with larger yields of recombinant protein).

2. Harvest the insect cells using low-speed centrifugation (1000 × g), and wash the pellets using TBS buffer. (Note: typically, a 10–20g pellet is achieved from 1L insect cell culture.) For both the KSR1:MEK1 and KSR2:MEK1 complexes, we have developed expression vectors based on the pFASTBac-Dual cassette. In this system 6 ×-His-MEK1 is produced under the influence of the early p10 promoter, whereas 6 ×-His-KSR1 or 6 ×-His-KSR2 are expressed from the later PH promoters. Tags on both proteins to isolate the complex are not essential. However, we find the dual-tagging strategy advantageous as this allows for capture of KSR-bound MEK and also excess free MEK. Moreover, due to the timing of the promoters, and the relative expression levels, the amount of free His-MEK1 can exceed the levels of His-KSR1/2 by several fold, as can be observed on SDS-PAGE

3. Protein extraction: (Note: Optimum yield from infected pellets is the highest without extensive or long-term cold storage). All the purification steps are performed at 4 °C. Lyse the cell pellets in 30mL buffer per 10g pellet in Lysis buffer on ice by sonication using a Branson 450 Digital Sonifier, installed with 3mm diameter horn, using 2 rounds of 60 pulses with 0.3 s burst of 45% output and 0.7 s resting time

4. Centrifugation and incubation of the cleared lysate with cobalt-agarose beads for IMAC: an initial high-speed centrifugation step (45,000 × g, 20min) separates soluble protein from cellular debris. However, downstream processing at this stage can be difficult so one can conduct batch purification in two different ways: either (A) protamine sulfate (final 1mg/mL) is added to the supernatant after the first step and the protein is precipitated via a second high-speed centrifugation step before the supernatant is decanted onto cobalt agarose beads (5mL of beads per 20g of starting pellets) in a gravity flow-column; or (B) the highly-viscous supernatant is directly incubated for 2h with cobalt agarose beads to sequester the histidine-tagged proteins in a 50mL falcon tube, which are then sedimented by a low-speed centrifugation step (100 × g) and the viscous supernatant can be poured off and discarded. The cobalt beads with the bound proteins are then suspended in wash buffer, and poured onto the gravity columns attached to a valve to control the flow of buffer

5. Elution of His-KSR1/2:His-MEK1 complexes from cobalt beads. After washing the beads with 3 rounds of 50mL wash buffer, the bound proteins are eluted using 25 mL elution buffer. Immediately after the elution, add EDTA to a final concentration of 1mM to chelate any leached cobalt in the eluate. After an incubation of 5min on ice with EDTA, add DTT to a final concentration of 5mM

6. Dialysis: Perform using dialysis cassette for minimally 4h against 2L of dialysis buffer

7. SDS-PAGE analysis: Use 8–16% gradient SDS-PAGE to assess supernatant, pellets, washes, elutions, and beads following elution for expression and solubility of KSR–MEK complexes over the initial purification steps (see Fig. 3B).

8. Ion-exchange chromatography: this step is required for additional purification and to separate excess free MEK from KSR-bound MEK. For KSR1:MEK1, use Q anion exchange chromatography, and for KSR2:MEK1, use an SP cation exchange column

9. Sample loading onto ion-exchange column: Remove the dialysis retentate from the dialysis cassette and dilute the sample in buffer A to reduce the concentration of sodium chloride to 50mM. Load the sample onto the IEX column at a flow-rate of 1mL/min using a 150mL superloop attached to an FPLC (e.g., AKTA) purification system

10. Washing and elution step of IEX: wash the column with 10 CV of buffer containing 50mM NaCl. Elute the bound protein using a gradient of buffer A and buffer B over a volume of 40mL. Collect 2mL fractions in tubes or deep well plates. Note: In the case of KSR1:MEK1, the complex will elute at ~250mM NaCl concentration. In the case of KSR2: MEK1, the complex will elute at ~250mM NaCl, with free MEK1 in the flow through or low salt washes (i.e., <100mM NaCl).

11. SDS-PAGE analysis: Use 8–16% gradient SDS-PAGE to confirm which fractions contain stoichiometric KSR:MEK complex

12. Concentration of KSR:MEK complex: Pool the fractions including 1:1 KSR:MEK complex and add fresh DTT at a final concentration of 5mM. Pour the sample into a 15 mL Amicon Ultra centrifugal device (with 20kDa MWCO), and centrifuge the sample at 4000 rpm for 10min. (Note: pipette the sample inside the centrifugal device up and down between each spin to mix the sample evenly). Repeat the centrifugation to achieve protein at a concentration of 4–5 mg/mL

13. Concentration estimation: use nanodrop to measure the concentration of KSR:MEK complex with buffer as a blank. For both KSR1:MEK1 and KSR2:MEK1, 1.0 Absorbance units at 280nm equates to 1 mg/mL of complex

14. Size exclusion chromatography (SD200 10/30 GL) to isolate the KSR1:MEK1 or KSR2:MEK1 heterotetramer. The isolation of the larger molecular weight heterotetrametric complex ensures reproducible biochemical results and crystal formation. Collect the sample into 0.5mL fractions. Estimate purity by SDS-PAGE analysis. Concentrate the fractions containing 1:1 KSR-MEK complex to 5–7 mg/mL. (Note: Following gel filtration, the purified complexes can either be flash frozen in liquid nitrogen or used immediately for crystallization trials). Typical yields are 0.2 mg of tetrameric complex from 50g of pellet for KSR1: MEK1 and 0.8 mg from 50 g of pellet for KSR2:MEK1

5.2. Crystallization of KSR:MEK complexes to map orthosteric and interfacial binders

Crystals of purified KSR1:MEK1 and KSR2:MEK1 complexes can be generated using the hanging drop vapor diffusion method (Bergfors, 2011), with a 1:1 μL mix ratio of protein and crystallization mother liquor above crystallization well-solutions.

1. For crystallization, KSR:MEK complexes typically in the range of 5–7 mg/mL are supplemented with 10mM MgCl₂ and 5 mM ATP or AMP-PNP

2. Conditions for optimal crystals can be found from focused screens; for example, slight variations on the concentration of PEG and pH around the following: 20% PEG 2000, 200mM magnesium acetate, 100mM MES, pH 6.5. In our hands, the best crystals have come from the condition C9 in the Jena Biosciences Kinase screen

3. We also found serendipitously that proteases can cleave the KSR:MEK complex in crystallization drops, and the crystals that subsequently appeared from these drops were of better quality in terms of size and X-ray diffraction properties (Khan et al., 2020). Testing a series of proteases, we determined that trypsin is ideal to control this process; we assume the protease removes flexible loops and/or partially folded proteins. Eventually, the trypsinization step can be included in the purification prior to size exclusion chromatography. Trypsinization is done at 1:1000 ratio of TPCK treated trypsin to protein (IEX-purified) overnight at 4 °C. Subsequent SEC allows a separation of hetero-tetrameric KSR:MEK (150 kDa) from trypsin (27 kDa). Intriguingly, from our determined crystal structures, we have not identified an obvious cleavage site suggesting that limiting amounts of trypsin does not cleave at crystallographically-visible lysine or arginine residues in the determined structures of KSR–MEK complexes (Khan et al., 2020).

4. Once crystals have appeared (first within 24 h, and grown for 3–5 days) we then transfer them into soaking solutions that include compounds (typically ~1mM, with 5–10% DMSO final) for 24–96 h at 20 °C to allow the diffusion of the compounds

5. After soaking, crystals are harvested in crystallization well solutions (20% PEG2000, 200mM magnesium acetate, 100mM MES, pH 6.5) supplemented with 25% ethylene glycol as cryoprotectant. One can also attempt 20–25% glycerol as a cryoprotectant

6. Crystals are then flash frozen in liquid nitrogen and preserved for data collection

7. We find that we need to screen on average 5 crystals to find a crystal that provides diffraction data to a resolution of ~3.5 Å or better. However, we have screened in some cases up to 20–30 crystals to find a single strong diffractor. High solvent content in the crystal form that we most often obtain for KSR1:MEK1 and KSR2:MEK1, as well as marked sensitivity to radiation damage, likely contribute to the variability in crystal quality and data collection that commonly occurs. However, given that we repeatedly obtain crystals with P6 symmetry, we can collect relatively small data wedges. Moreover, techniques such as low-dose of X-rays, reduced degree of oscillation (0.2 deg.) per frame, exposure time of 1s/deg., vector data collection and grid screening while employing the software-driven strategy of data collection (McCoy et al., 2007) at modern beamlines, such as AMX-NSLS2, FMX-NSLS2 and LS-CAT-APS, where we have collected most of our data, have helped overcome some of the challenges in solving KSR1:MEK1 and KSR2: MEK1 structures bound to ligands

Standard software packages are employed for data processing (XDS; Kabsch, 2010), phase determination (Phaser; McCoy et al., 2007) and refinement (Phenix; Afonine et al., 2012).

6. Assays to measure binding of ATP-competitive compounds within the orthosteric site on KSR1 and KSR2

The original forward genetic screens conducted in the fly and worm, which led to the identification of KSR, found point mutations in KSR that could suppress signaling specifically by oncogenic mutants of RAS. Several years ago, we mapped the mutations from these screens onto the predicted structures of the RAF:KSR:MEK complex (Fig. 4A) and found that a number of the alleles spanned the RAF and MEK interfaces extending to the ATP-binding pocket (Dhawan et al., 2016). This allowed us to speculate that we may identify ATP-competitive binders of KSR that could function equivalently to genetic suppressor alleles or, as we hypothesized, “chemical suppressors of RAS.” To identify such compounds, we used a probe for labeling the ATP binding pocket of purified KSR2:MEK1 complexes. This probe, referred to as ATP-biotin (Fig. 4B), places an acyl-phosphate within the active site of a kinase to label proximal lysine residues (Dhawan et al., 2016; Patricelli et al., 2007). The probe is available commercially (Pierce 88,311) or can be generated based on published procedures (Patricelli et al., 2011). When using ATP-biotin we find it important to calibrate labeling based on varying time and concentration of the probe as well as the target; this can also be a good quality control step for new batches of purified proteins. For example, in Fig. 4C and D, we show labeling on isolated MEK1, isolated BRAF, and the KSR1:MEK1, and KSR2:MEK1 complexes.

Fig. 4.

An assay for orthosteric binders at the ATP-binding pocket of purified KSR-MEK complexes. (A) Putative structure of RAF (blue)–KSR (green)–MEK (red) complex based on structural and functional studies (Brennan et al., 2011; Burack & Shaw, 2000; Dhawan et al., 2016; Khan et al., 2020; Lavoie et al., 2018; Michaud et al., 1997; Morrison, 2001; Nguyen et al., 2002; Rajakulendran et al., 2009; Roy, Laberge, Douziech, Ferland-McCollough, & Therrien, 2002; Therrien et al., 1996). KSR mutations that suppress oncogenic RAS signaling (shown in red) localized around the ATP-binding pocket (yellow) as well as the RAF and MEK interfaces (Dhawan et al., 2016). (B) Structure of desthiobiotin-ATP: ATP (left) linker (middle) desthiobiotin (right). (C) Schematic for labeling of the orthosteric sites of KSR and MEK with ATP-biotin in the presence and absence of competing ligands probe. (D) Representative western blot of recombinant MAPK proteins (MEK1, BRAF, KSR1:MEK, and KSR2:MEK) labeled with increasing concentrations of the ATP-biotin probe. (E) Probe-labeled kinases (MEK, BRAF, KSR1:MEK, and KSR2: MEK) treated with increasing concentrations of inhibitors (5, 10 and 20 μM) were measured via western blotting. 2.5 μM of probe was used to label each protein. DMSO was used as a negative control (No inhibitor) and free ATP (100, 1000, and 10,000 μM) was used as a positive control for blocking labeling. (F) The KSR2-MEK1 complex with APS-2-79 identified the inhibitor bound in the orthosteric site of KSR2 (Dhawan et al., 2016).

Small molecule binding at the orthosteric site on KSR can be detected in the competition format with free inhibitor or nucleotide. Moreover, an initial assessment of inhibitor specificity can be conducted using this assay as well; as we demonstrate BRAF-specific inhibition using dabrafenib as an example in Fig. 4E. However, it is important to note that we run ATP-biotin competition assays without physiological concentrations of free ATP, which can lead to potentially false positive measurements of small molecule-target interactions. For example, we have found that several of our probes, including APS-2-79, can strongly inhibit labeling of both MEK1 and KSR2 in vitro (Fig. 4E); however, we have only been able to validate binding of APS-2-79 to KSR2 in vivo, using orthogonal assays (Dhawan et al., 2016). Moreover, when APS-2-79 was soaked or cocrystallized with KSR2:MEK1, we only found density for APS-2-79 within the KSR2 binding pocket and not MEK1 (Fig. 4F). Again, these data suggest that APS-2-79 binds preferentially to KSR2 within the KSR2:MEK1 complex. Another potential source of false positives for this assay can occur from compounds that possess a free amine (−NH) group that can react with the probe and quench the critical acyl-phosphate moiety; although this is rare and we have not observed reaction with buffer components such as Tris with the probe. False-positives can be assessed by testing for inhibitor+probe adducts using LC-MS. Additionally, competition experiments conducted using control kinases can rule out non-specific reactions between the probe and test compounds.

6.1. Equipment

• P2, P20, P200, P1000 pipettes (Gilson, Cat# F144054M, F144056M, F144058M, F144059M, respectively)

• Isotemp Heat Block (Fisher Scientific, Cat# 11-71806)

• Criterion™ Cell and PowerPac™ Basic Power Supply (BioRad, Cat# 1656019)

• Criterion™ Blotter with Plate Electrodes (BioRad, Cat# 1704070)

• PowerPac™ HC High-Current Power Supply (Bio-Rad, Cat# 1645052)

• Magnetic Stirrer Plate: 5 L Load Capacity (Velp Scientific, Cat# 56LK76)

6.2. Materials and reagents

• Gel loading pipet tips (Fisherbrand, Cat# 02-707-181)

• PCR tube strips (ThermoFisher, Cat# AB2000)

• 1.5 mL microcentrifuge tubes

• Pipette tips

• Assay Buffer (200mM NaCl, 50mM Tris, 5% Glycerol, 1mM DTT, and 1 mM EDTA)

• Proteins: MEK1 and BRAF (Dhawan et al., 2016; Khan et al., 2020), MEK1:KSR2 (Refer to Section 5)

• Magnesium chloride hexahydrate (Sigma-Aldrich, Cat# M9272-500G)

• Pierce Kinase Enrichment Kit with ATP Probe (Thermo Scientific, Cat# 88310)

• Dimethyl Sulfoxide (DMSO) (Fisher Scientific, Cat# BP231-100)

• Dabrafenib (GSK2118436) (Selleckchem, Cat# S2807)

• APS-2-79

• Adenosine 5′-triphosphate disodium salt hydrate (Sigma-Aldrich, Cat# A2383)

• Molecular biology grade water (Corning, Cat# 46-000-Cl)

• 6 × Laemmli Buffer (Alfa Aesar, Cat# J61337)

• 4–15% Bis-tris glycine 26-well gel (Bio-Rad, Cat# 340125)

• 20 × MOPS SDS Running Buffer (ThermoFisher Scientific, Cat# NP0001)

• Nitrocellulose blotting membrane (GE Healthcare, Cat# 10600002)

• Western blotting filter paper (Thermo Scientific, Cat# 84784)

• Tris-Glycine Transfer Buffer (25mM Tris, 192mM Glycine, and 20% methanol)

• Tris-Buffered Saline (200mM Tris and 1500mM NaCl) and Tween 20 (TBS-T)

• Bovine Serum Albumin Fatty Acid-free Powder (Fisher Scientific, Cat# BP9704-100)

• Magnetic stir bar

• Streptavidin-horseradish peroxidase (HRP) (Cell Signaling, Cat# 3999S)

6.3. Step-by-step protocol

1. Prepare the assay buffer fresh before each experiment. Dilute each purified protein to a concentration of 200nM (e.g., 8.4ng/L for MEK1, 6.4ng/μL for BRAF kinase domain, and 15.4 ng/μL for KSR2:MEK1) and keep on ice

2. Aliquot 27.3 μL of each protein mix per each condition desired

3. Prepare a 1 M stock solution of MgCl₂ and add 0.3 μL to each condition

4. Prepare a 30 × stock of each inhibitor in 100% DMSO (e.g., 30, 150, and 300 μM) and add 1 μL to the appropriate condition for a final concentration of 1, 5, and 10 μM. Incubate the reaction mixture for 15 min

5. Prepare 30 × free ATP stock concentrations in molecular biology grade water (e.g., 3, 30, and 300mM) and add 1 μL to the appropriate conditions for a final concentration of 100, 1000, and 10,000 μM. Incubate the reaction mixture for 15 min

6. During the incubation, prepare a 166 μM working stock of the ATP-desthiobiotin probe in 60 μL of molecular biology grade water. Add 1.6 μL of the 166 μM stock solution to each reaction mixture at a final concentration of 2.5 μM and incubate for 10 min

7. After the incubation, add 6 μL of 6 × Laemmli SDS sample buffer to each reaction mixture and heat each sample mix to 80 °C for 5 min

8. Load each prepared sample into SDS-PAGE wells of a 4–15% bis-tris glycine 26-well gel and run the gel in 1 × MOPS running buffer for 60min at 125 V at room temperature

9. Next, transfer the gel onto a nitrocellulose membrane in 20% methanol in tris-glycine transfer buffer for 1 h at 90 V at 4 °C. Add a magnetic stir bar at the bottom of the transfer tank (~400 RPM) to help distribute heat and keep the buffer homogenous

10. After confirmation of a successful transfer via Ponceau Red, extensively wash the membrane until the Ponceau Red is completely removed

11. Block the membrane in 5% BSA TBS-T for 45 min at room temperature on a rocker

12. While the membrane is blocking, prepare the Streptavidin-HRP antibody cocktail in 5% BSA TBS-T (1:7500 dilution).

13. Add the antibody cocktail to the membrane and incubate overnight at 4 °C on a rocker

14. The following day, extensively wash the membrane 3 × with TBS-T on a rocker at room temperature

15. Prepare a 1:1 mixture of SuperSignal West Pico Chemiluminescent Substrate and directly add it onto the membrane to detect Streptavidin-HRP specifically bound to the protein of interest

16. Image the membrane by film or using a ChemDoc XRS+ Imaging System

17. Export the desired image exposure using the Image Lab program

18. Integrate band intensities for inhibitor and DMSO treated samples to calculate percent inhibition

7. Assays to measure binding of interfacial compounds of KSR-MEK complexes

Protein–protein interactions are critical to facilitating normal and oncogenic signal transduction. Biochemical characterization of inhibitors is generally focused on measuring interactions of small molecules with single targets. This type of analysis does not capture the physiological factors, such as endogenous competitors or the influence of Protein–protein interactions in the cell. Moreover, inhibitor influence on protein–protein interactions is usually assessed using lytic techniques such as coimmunoprecipitation and cellular thermal shift assays. While robust, these approaches could potentially alter pre-existing or dynamic protein–protein complexes that occur within live cells.

NanoBRET is a technique that overcomes the limitations of in vitro or lytic assays of target engagement by allowing for pharmacological measurements in living cells (Robers et al., 2019; Vasta et al., 2018). In this technique, a protein of interest is transfected into mammalian cell lines (usually HEK293T) as an N- or C-terminal NanoLuciferase (NL) fusion and a small molecule called a tracer is applied to the cell. A tracer is a bifunctional molecule that can participate in bioluminescence resonance energy transfer (BRET) with NL-tagged proteins and consists of a drug moiety targeting a protein(s) of interest, and a Bodipy dye (576/589nm) connected by a linker (usually PEG or alkyl). Once the tracer is bound to its target, the NL substrate, furimazine, is applied to generate blue light (460 nm). If the Bodipy moiety of the tracer is within ~100 Å from the NL (donor) it will undergo BRET and emit red light (acceptor-610 nm). The red and blue light signal can be quantified separately on a plate reader (GloMax Discover plate reader using the BRET 618 protocol) and divided to generate a BRET ratio (acceptor/donor). The emission profiles of light emitted from the Nanoluciferase and the Bodipy moiety have minimal overlap, which minimizes bleed-through of signal between the donor and acceptor channels, and thus, false-positive BRET.

Two examples of NanoBRET assays are steady-state and washout experiments (Fig. 5). The readout of a steady-state experiment is a dose-response curve where varying concentrations of a drug and a constant tracer concentration are introduced to cells transfected with the protein of interest. These dose-response curves can be fitted to generate apparent IC₅₀ values. In a washout experiment, transfected cells are incubated with a drug (at a saturating concentration ~15 × higher than the apparent IC₅₀ determined from steady-state experiments) for 2 h, spun down, washed with Opti-MEM to remove the drug, and introduced to a constant tracer concentration. The readout of this experiment is the buildup of BRET signal over time as the tracer occupies more open targets as they become available due to release of pre-bound drugs. It is important to include DMSO as a negative control and to establish the baseline kinetics of tracer binding in the absence of a drug. Step-by-step instructions for both NanoBRET assays have been previously described (Robers et al., 2019).

Fig. 5.

Live cell assay schematic for allosteric and interfacial binders on KSR–MEK complexes. Schematic examples of steady-state (A) and washout (B) experiments performed on a MEK-NL (pink) construct coexpressed with unlabeled KSR (purple) (Note: other assay configurations to measure pharmacology on the MEK–KSR complex can be found in Fig. 6B). In both assays, BRET can only be observed when the donor (e.g., Nanoluciferase or NL) is in relatively close proximity (<50Å) to an acceptor (e.g., the BODIPY moiety of Tram-bo). Under these conditions, the blue light produced from activity of Nanoluciferase on its substrate furimizine produces high-intensity luminescence (measured at 460 nm) and thereby energy transfer on Tram-bo, which subsequently emits a second signal as red light (measured at 610nm). Competition with free ligands (e.g., MEKi) that overlap with the Tram-bo binding site block the energy transfer and therefore can be recorded as a lack of BRET relative to vehicle control. In the residence time experiments, drugs are added to cells first and then washed out. Build-up of BRET upon addition of Tram-bo to the washed cells then serves as a proxy for the dissociation of free MEKi and thereby k_off MEKi, as binding sites become available for tracer binding on the complex (far right cartoon). Step-by-step instructions to estimate k_off and drug residence time from wash-out experiments can be found in Robers et al. (2019).

When developing a NanoBRET assay to measure pharmacology on a single target and/or protein–protein interaction, it is important to consider (1) the potency and specificity of drug that becomes the basis of a tracer, and (2) NL placement in the context of a fusion construct. MEK inhibitors (MEKi) pharmacology on free MEK and MEK in complex with KSR can be measured using a potent Trametinib-based tracer, which we termed Tram-bo (Fig. 6A). Tram-bo is composed of Trametinib, a two-PEG unit linker, and a Bodipy dye (576/589 nm), and can be used to measure pharmacology of MEKi on free MEK1 using a MEK1-NL construct, and MEK1 in complex with mKSR1 using a mKSR1-NL construct, or MEK1-NL coexpressed with excess unlabeled mKSR1 (Fig. 6B). To compare steady state and washout experiments across multiple complexes, it is important to determine a saturating concentration of a tracer (i.e., a concentration where Tram-bo gives the highest signal and is bound to the same extent across all proteins and protein complexes). Note that this may not be achievable for all tracers since the solubility of tracers only extends generally into the low single-digit micromolar range. For Tram-bo, this was accomplished by generating an EC₅₀ curve for each scenario where the concentration of Tram-bo was varied to determine its binding affinity and highest BRET signal (Fig. 6C). From these assays, we determined a working concentration of Tram-bo above the EC₅₀ on MEK1-NL plus/minus mKSR1 coexpression or in mKSR1-NL assays. Although the EC₅₀ of Tram-bo on MEK1-NL is ~20 times more potent than on the other two MEK1–mKSR1 complex scenarios, using a saturating constant concentration (1 μM) is important to compare MEKi pharmacology on free-MEK1 and the MEK1–mKSR1 complex. Further, since the EC₅₀ of Tram-bo is in the nanomolar range, its tight binding can cause reproducibility issues in steady-state experiments that are dependent on the order of addition of Tram-bo and a drug. This is illustrated by the apparent IC₅₀ and curve height differences resulting from the order of addition of a drug and Tram-bo on MEK1-NL (Fig. 6D). Either adding the drug first, or in combination with a tracer provides the most reproducible results with the largest BRET window.

Fig. 6.

Developing NanoBRET to measure the pharmacology of MEK inhibitors on KSR–MEK complexes in live cells. (A) Structures of Trametinib and the NanoBRET probe we call Tram-bo. Within Tram-bo, the parent compound is shown in black, the linker portion is green, and Bodipy is red. (B) Nanoluciferase (−NL) placement on MEK or KSR allows pharmacology to be measured on MEK and the MEK-KSR complex in model systems (e.g., HEK293 cells). Shown on the left is a cartoon for measuring Tram-bo bound to MEK1-NL. Also shown are putative complexes of Tram-bo bound to MEK-KSR formed through MEK1-NL coexpressed with either KSR1 or KSR2 (middle) or from the expression of KSR1-NL or KSR2-NL, which is expected to form complexes with endogenous MEK (far right). (C) Build-up curves to measure the apparent EC50 of Tram-bo on MEK and the MEK–KSR complex. (D) Order of addition matters for tracers that have nanomolar apparent EC50 values as seen for Trametinib on MEK-Nanoluciferase. (E) Steady-state dose-response and washout experiments depend on specific interactions at the interface between MEK and KSR1 as determined by the use of a KSR1 mutant, W781D. (F) Specificity of KSR1-NL assays are evidenced by the lack of a Trametinib dose-response using the W781D mutant. Please see Fig. 2 for the location of the W781 residue at helix αG and at the interface of the KSR1-MEK1 complex. Data in panels (E and F) reproduced from Khan, Zaigham M., Alexander M. Real, William M. Marsiglia, Arthur Chow, Mary E. Duffy, Jayasudhan R. Yerabolu, Alex P. Scopton, and Arvin C. Dar. 2020. “Structural basis for the action of the drug trametinib at KSR-bound MEK.” Nature 588 (7838): 509–14.

When considering protein–protein interactions it is important to use mutants at the interface to test specificity. In the scenario where NL is on MEK1 and is coexpressed with mKSR1, one of the most noticeable observables is a decreased maximum BRET ratio in steady-state and washout experiments that is most likely due to reorientation of the tracer in the complex. Mutation of W781D on mKSR1 generates similar steady-state and washout experiment results to MEK-NL alone indicating the specificity of the complex when testing MEK1-NL in combination with wild-type mKSR1 (Fig. 6E). Harmoniously, specificity of the MEK1:mKSR1 complex can also be tested using the KSR-NL assay configuration where W781D results in the loss of a Trametinib dose-response (Fig. 6F). These data support that Tram-bo based BRET in the context of the mKSR1-bound complex occurs through interactions with endogenous MEK1.

Following tracer optimization and confirmation of complex specificity, inhibitors can be characterized to develop cellular apparent IC₅₀ and residence time values on free MEK1 and complexes (e.g., mKSR1-bound MEK1). When comparing free MEK1 and the mKSR1:MEK1complex, MEKi such as trametinib have a similar apparent IC₅₀ value for both free protein and the complex, whereas a related clinical MEKi cobimetinib displays more potent binding to the complex (Fig. 7A). This change in potency may be related to the sterics that hinder the dissociation of cobimetinib in the MEK1–mKSR1 complex as compared to free MEK1. These differences become less pronounced in washout experiments (Fig. 7B) because most of the MEKi in Fig. 7A display nanomolar apparent binding affinities with long k_off. Even though a drug can have a similar apparent IC₅₀ for free protein vs a complex, differences can be observed with respect to the lifetime of the drug–receptor complex as measured by k_off and residence time (Copeland, 2015; Lu & Tonge, 2010). In NanoBRET experiments, residence time can be approximated and compared between compounds through wash-out assays (Fig. 5B). For highly potent compounds, residence time can be approximated by titrating drugs at subsaturating concentrations. In the case of trametinib, longer residence time (i.e., slower Tram-bo buildup) is observed for the mKSR1:MEK1-NL complex as evidenced by the lack of Tram-bo buildup and recovery and the small changes from titrations of free drug from base-line signal (Fig. 7C). For MEK1-NL alone, equivalent free drug concentrations produced a more progressive movement toward the DMSO control, which is consistent with shorter residence time of the analyzed MEKi on free MEK, supporting the notion that, once bound, mKSR1 serves as a barrier to dissociation.

Fig. 7.

Steady-state and kinetic measures of MEKi on MEK and the KSR–MEK complex in live cells. (A) Dose-response curves and fitted apparent IC50 values of MEKi on free MEK and the KSR–MEK complex. (B) Characterization of MEKi residence time on MEK and the KSR–MEK complex. (C) Residence time differences of Trametinib (i.e., drugs with long k_off values) on MEK and the KSR-MEK complex can be observed using subsaturating drug concentrations in washout experiments (Khan et al., 2020). All panels reproduced from Khan, Zaigham M., Alexander M. Real, William M. Marsiglia, Arthur Chow, Mary E. Duffy, Jayasudhan R. Yerabolu, Alex P. Scopton, and Arvin C. Dar. 2020. “Structural basis for the action of the drug trametinib at KSR-bound MEK.” Nature 588 (7838): 509–14.

Taken together, NanoBRET is a useful tool to interrogate pharmacology of pseudokinase Protein–protein interactions, as well as on-target pharmacology. This assay provided a physiological measure of KSR-bound MEK in living cells and demonstrated the unique pharmacology of MEK inhibitors on the complex versus free forms of the enzyme. The judicious use of mutations at the interface allows one to interrogate in vivo binding signals and interpret the most likely forms of the complex and the binding interface requirements that may strongly influence structure-activity relationships (SAR) between compounds.

7.1. Equipment

• Bioluminescent plate reader (GloMax Discover, Promega Cat# GM3000 or PerkinElmer EnVision Cat# 2105-0010)

• Biosafety cabinet (Labconco, Cat# 302420001)

• Bench top microcentrifuge (ThermoFisher, Cat# 75002447)

• Benchtop centrifuge (ThermoFisher, Cat# 75008801)

• Countess III FL automated cell counter (ThermoFisher, Cat# AMQAF2000)

• P2, P20, P200, P1000 pipettes (Gilson, Cat# F144054M, F144056M, F144058M, F144059M, respectively)

• P20 and P200 12-channel pipettes (Rainin, Cat# 17013808 and 170123810, respectively)

• Serological Pipet-Aid (ThermoFisher, Cat# 03-840-313)

7.2. Material and reagents

• HEK293T cells (ATCC CRL-3216)

• Opti-MEM without Phenol Red (Gibco, Cat# 11058021)

• DMEM (Gibco, Cat# 11965092)

• PBS (Gibco, Cat# 20012027)

• 25 mL reservoirs (ThermoFisher, Cat# 95128093)

• Trypsin-EDTA (0.25%) with phenol red (Gibco, Cat# 25200114)

• White 96-well plates (Corning, Cat# 3990)

• Clear plastic 96-well plate (Celltreat, Cat# 229196))

• Serological pipette tips (2mL aspirating, 2, 5, 10mL)

• Fugene HD Transfection Reagent (Promega, Cat# 2311)

• Carrier DNA (Promega, Cat# E4882)

• MEK inhibitors (Selleckchem)

• DMSO (Sigma, Cat# 275855-100mL)

• Countess slides (ThermoFisher, Cat# C10228)

• Promega NL substate and Inhibitor kit (Promega, Cat# N2161)

• Tracer Dilution Buffer (Promega, Cat# N2191)

• Tram-bo (Khan et al., 2020)

• CMV-mKSR1, CMV-mKSR1-NL, and CMV-MEK1-NL DNA (Khan et al., 2020)

• Stackable pipette tips (Rainin, Cat# 17005873 and 17005875, respectively)

• PCR tube strips (ThermoFisher, Cat# AB2000)

• 1.5 mL microcentrifuge tubes (ThermoFisher, Cat# 05-4080130)

• Trypan Blue (ThermoFisher, Cat# C10282)

• Tissue culture dishes (Celltreat, Cat# 229660)

• 15 mL Falcon tubes (Falcon, Cat# 352196)

7.3. Step-by-step protocol

Day 1: Transfection

1. For MEK1-NL or mKSR1-NL single transfections: prepare a mixture of Carrier DNA and with MEK1-NL or mKSR1-NL DNA in Opti-MEM. For MEK1-NL coexpressed with mKSR1: prepare a mixture of Carrier, MEK1-NL and mKSR1 DNA in Opti-MEM. Add Fugene HD and leave at room temperature for 20min. If using Fugene HD, the final concentration of total DNA should be 10 μg/mL. For single transfections, use 0. 1 μg/mL + 9.9 μg/mL for MEK-NL or KSR-NL and Carrier DNA, respectively. For coexpressions use 0.1 μg/mL + 1:8.9 μg/mL MEK1-NL:mKSR1:Carrier DNA. The total amount of Opti-MEM should be 1/20th the volume of the entire transfection. For example, if 1.1 million cells in 5.28mL of DMEM (200,000cells/mL according to Fugene HD manufacturer’s protocol) need to be transfected to have enough cells on Day 2 for half of a 96-well plate, 0.264 mL of Opti-MEM should be used to generate DNA complexes

2. During the incubation, remove HEK293T cells from incubator, check that they are ~70% confluent, remove DMEM, wash with PBS, trypsinize, add DMEM, and spin down at 300 × g for 5 min at room temperature. Remove the supernatant, adjust the cell density to 200,000cells/mL, and plate in a tissue culture treated plate

3. Add the DNA complex mixture, mix well, and incubate at 37 °C for at least 20 h

Day 2: Steady-state assay setup

1. Thaw 400 × Tram-bo (solution in DMSO), tracer dilution buffer, and 1000 × MEK inhibitors in DMSO. Note that it is good practice to make a large 1000 × stock of Tram-bo and 1000 × stock of each inhibitor (10mM is a good place to start), aliquot into single use PCR tubes to store at −80 °C

2. Warm up DMEM, PBS, Trypsin-EDTA, and Opti-MEM in a 37 °C water bath

3. Prepare a 1000 × dose-response curve in a strip of eight PCR tubes. Pipette 18 μL of DMSO into tubes 2–8, and 5 μL of the 1000 × MEK inhibitor into the first tube. Then take 2 μL from the first tube, and pipette up and down in the second tube. Take 2 μL from the second tube and then pipette up and down in the third tube. Repeat until the end of the strip, and the entire procedure for each MEK inhibitor

4. In a clearplastic 96-well plate, pipette 198 μL of warm Opti-MEM into each well of a column. Using a multichannel pipette, transfer 2 μL of each PCR tube in the strip to the column of Opti-MEM. Pipette up and down to mix. The concentration of the dose-response will be 10 × and enough volume for 20 columns of a 96-well plate

5. Prepare a 20 × stock of Tram-bo aliquoted into a strip of 12 PCR tubes. To prepare enough to add to one 96-well plate mix 28.8 μL of 400 × Tram-bo with 86.4 μL of DMSO and 460.8 μL of Tracer dilution buffer. Mix well and aliquot. Keep in a dark place until use

6. Remove transfected HEK293T cells from the incubator, remove DMEM, wash with PBS, trypsinize, add DMEM, and spin down at 300 × g for 5 min at room temperature. Remove supernatant and resuspend in Opti-MEM

7. Adjust cell density to 235,000 cells/mL using Opti-MEM and use a multichannel pipette to dispense 85 μL of cells into each well of a white NBS Corning 96-well plate from a reservoir

8. Using a multichannel pipette, transfer 10 μL of each 10 × drug dose-response from the clear 96-well plate, to the 96-well plate with cells

9. Next, transfer 5 μL of Tram-bo to each well using a multichannel pipette and pipette up and down

10. Incubate for 2 h at 37 °C

11. Prepare a 3 × solution of NL substrate and inhibitor in Opti-MEM. For one 96-well plate add 33 μL of the NL-substrate and 11 μL of the inhibitor to 5.24mL of Opti-MEM

12. Take plate out of incubator and let it cool to room temperature for 10 min. Then using a multichannel pipette, add 50 μL of Opti-MEM solution to each well

13. Read on bioluminescent plate reader (GloMax Discover) using the stand NanoBRET 618 protocol

14. Export the data as an Excel document, and copy and paste into Prism. The data can be fitted using the non-linear regression option followed by the [Inhibitor] vs response (three parameter) equation

Day 2: Washout assay setup

1. Thaw 1000× solutions of MEK inhibitors in DMSO and add 2 μL into 198 μL of Opti-MEM in a well of a clear 96-well plate to make a 10 × solution. Transfer 30 μL of each solution into a clean microcentrifuge tube

2. Remove transfected HEK293T cells from the incubator, remove DMEM, wash with PBS, trypsinize, add DMEM, and spin down at 300 × g for 5 min at room temperature. Remove supernatant and resuspend in Opti-MEM

3. Adjust cell density to 1.67 million cells/mL using Opti-MEM and transfer 270 μL to each microcentrifuge tube containing MEK inhibitors for a total volume of 300 μL. There will be 450,000 cells in each tube

4. Place tube at 37 °C for 2h

5. Spin tubes down at 300 × g and remove supernatant

6. Add 300 μL of Opti-MEM and spin down again

7. Remove the supernatant, resuspend in 300 μL of Opti-MEM, and transfer 95 μL into three wells of an NBS corning 96-well plate

8. Add 5 μL of a 20 × Tram-bo solution to each well, followed immediately by 50 μL of a 33.3 × solution of NL-substrate and inhibitor

9. Read immediately on a bioluminescent plate reader (GloMax Discover). Adjust the number of reads and time between reads so the length of the experiment will be 2.5 h

10. Export the data into Excel and plot with Prism

8. Final conclusions

KSR-MEK-RAF has proven itself as a model to study signal transduction that occurs via changes in the assembly and regulation of higher order complexes (e.g., Brennan et al., 2011; Burack & Shaw, 2000; Dhawan et al., 2016; Khan et al., 2020; Lavoie et al., 2018; Michaud et al., 1997; Morrison, 2001; Nguyen et al., 2002; Rajakulendran et al., 2009; Roy et al., 2002; Therrien et al., 1996). Moreover, subtle but important structural differences between KSR and RAF have enabled unique pharmacological mechanisms to modulate the RAS-MAPK pathway. For example, our structural and mechanistic data suggests that KSR can serve as a coreceptor for potent and long-lived binding of several clinical MEKi (Khan et al., 2020). The model that emerges from these studies is that certain compounds (e.g., trametinib) favor binding of MEK to KSR while disfavoring binding of MEK to RAF. This proposed mechanism suggests that clinically important compounds can be developed to exploit the defining features of highly conserved enzyme-pseudoenzyme signaling systems. These data also suggest that specific pseudokinases and related complexes (KSR:MEK and BRAF:MEK) may serve as distinct and physiologically relevant targets for therapeutics. Pseudokinase:kinase complexes are pervasive, and the methods described in this chapter that aim to find both ATP-competitive orthosteric inhibitors or interfacial binders could similarly be applied to other related systems. Indeed, we recently used the ATP-biotin assay, combined with additional biophysical and computational approaches, to identify the first type II binders of the pseudokinase STRAD (Smith et al., 2021).

We have highlighted the identification of small molecule binders to exploit or harness specific aspects of KSR function. Such chemical tools, in particular molecular glues for KSR, may be effective at limiting common resistance and drug escape mechanisms from MAPK-targeted therapies. Other pharmacological approaches, such as small molecule degraders and PROTACs, could also be developed. Recent efforts described for BRAF and MEK have demonstrated that degraders can be achieved on both wild-type and mutant forms of these proteins (Alabi et al., 2021; Posternak et al., 2020; Wei et al., 2019). Small molecules to eliminate KSR, or molecular glues and orthosteric inhibitors to lock particular functional states, may yield useful chemical probes and therapeutic leads. Several studies have given somewhat mixed results with respect to the impact of KSR depletion across cancer and disease models (e.g., (Costanzo-Garvey et al., 2009; Germino et al., 2018; Kortum et al., 2006; Lozano et al., 2003; Nguyen et al., 2002; Pearce et al., 2013; Therrien et al., 1996). Future efforts using chemical biology approaches should help deconvolute how KSR may be best exploited for therapeutic applications. This will continue to be an important goal as KSR functions in the pathway and via association with targets for two of the most frequently mutated oncogenes (i.e., BRAF and KRAS) across human cancers.