Introduction

The human kinome is encoded by more than 500 genes which lead to the synthesis of functionally diverse kinases. Kinases are different in their domain composition but all contain a structurally conserved phosphotransferase domain (Manning (2009), Kornev and Taylor (2010)). Conventionally, protein kinases function as enzymes that enable the transfer of a phosphate group from ATP to defined amino acids of a target protein. Protein kinases regulate various aspects of cellular functions including cell survival, apoptosis, cell division and metabolism via reversible and tightly regulated phosphorylations (Figure 1A) (Bhullar et al. (2018), Blume-Jensen and Hunter (2001), Manning et al. (2002)). In addition to kinase phosphotransferase activities, kinase domains have scaffolding functions (Shrestha et al. (2020), Reiterer et al. (2014)).

Figure 1.

On the cellular level, kinases act as molecular switches, adopting conformational states that align with an active (ON) or inactive (OFF) kinase state (Huse and Kuriyan (2002), Yamaguchi and Hendrickson (1996), Lopez et al. (2019), Feichtner et al. (2022)). These ON and OFF states of protein kinases signify a switch-like behavior which is governed by a collection of molecular mechanisms (Figure 1A). Funneling the diverse signals underlines the center stage of kinases in signaling pathways, which frequently involve kinase regulation. Thus, one kinase directly modulates the activity level and role of the subsequent kinase in a cascade like structure (Guo et al. (2020), Avruch et al. (2001)). This regulation is facilitated by the formation of multi-protein complexes to spatiotemporally control and amplify signal transmission (Morrison (2001), Pouysségur et al. (2002)). Figure 1A highlights several factors affecting kinase functions. Selected modes of kinase regulation which are of relevance for the presented study are listed below:

First, the activity of a kinase can be altered by post-translational modifications (PTMs) of selected amino acids. PTMs alter kinase characteristics like its activity status, its location within the cell, rate of degradation, and its associations with other proteins (Chou (2020), Cohen (2000), Deribe et al. (2010)).

Second, the scaffolding functions of kinase domains accompany the process of phosphotransferase reactions. These play central roles in the activation and deactivation process of interacting kinase protomers. This feature is key for pseudokinase activities, for relaying signaling inputs without catalytic functions (Weinlich and Green (2014), Morrison and Davis (2003), Boudeau et al. (2006)).

Third, kinase functions depend on decisive regulatory protein interdependencies, for which several modes of regulation have been described. Intramolecular auto-inhibitory modules (AIM) alter kinase activity states by reducing the accessibility of the substrate protein for the subsequent kiss-and-run phosphotransferase reaction (Mayrhofer et al. (2020), Pufall and Graves (2002), Xu et al. (2002)). In some cases, kinase domain activities are controlled by interacting with substrate-like sequences. Pseudo-substrate stretches bind to the catalytic cleft of the kinase and hinder the phosphorylation of the substrate. This binding to the catalytic site is altered in response to input signals and manifested in kinase conformation changes, which in turn coordinate protein kinase activation cycles (Schmitt et al. (2022), Kemp et al. (1994)). Besides intramolecular inhibition, activating and inactivating regulatory protein interactions have been described for prototypical kinases such as PKA and CDKs (Boudeau et al. (2003), Taylor et al. (2005)).

Fourth, kinase regulation is strongly dependent on the expression pattern and how the protein is stabilized in regard to the cell fate within the respective cell system or compartment (Capra et al. (2006)).

Fifth, regulatory protein interactions of kinases depend on small molecule interactions. Besides different types of second messengers (e.a. Ca²⁺, IP3, cAMP) (Newton et al. (2016), Kasai and Petersen (1994)) a collection of metabolites and ions contribute in orchestrating cellular kinase functions (Ramms et al. (2021)).

Alterations of the ON and OFF modes of regulating kinases are also pertinent to kinase-related disorders. Deregulation of kinase functions is associated with the development of numerous diseases, such as cancer, inflammatory, infectious and degenerative conditions (Shchemelinin et al. (2006), Köstler and Zielinski (2015), Ferguson and Gray (2018)). Kinase malfunctions result from gene mutations, deletions, fusions or increased or aberrant expressions. These mechanisms result in gain or loss of function of the involved kinase pathways, thus driving disease etiology and progression (Ochoa et al. (2018), Van et al. (2021), Cicenas et al. (2018)).

For this technical report, we have incorporated disease-relevant full-length phosphotransferases which exhibit different modes of regulation into the kinase conformation (KinCon) reporter system. The KinCon reporter platform is a Renilla luciferase (RLuc) based protein fragment complementation assay, which fuses two fragments of RLuc to the N and the C terminus of a full-length kinase. This reporter can be used to track conformational changes of kinases due to mutations, PTMs, PPIs, or inhibitor binding. We discuss the KinCon reporter principle at the beginning of the results section. The kinases listed below are characterized by distinctive modes of regulation, are viable targets for inhibition, or are so far difficult to assess directly through traditional biochemical measurements: The liver kinase B1 (LKB1, STK11), the receptor-interacting serine/threonine-protein kinase 1 (RIPK1) and the cyclin-dependent kinases 4 and 6 (CDK4/6).

LKB1 is active as trimeric cytoplasmic protein complex. It is the upstream regulator of the AMP-activated protein kinase (AMPK), which is the macromolecular key kinase complex in regulating the cellular energy metabolism (Rodríguez et al. (2021)). LKB1 promotes AMPK signaling by forming an heterotrimeric complex with the pseudokinase STE20-related adaptor alpha (STRADα) and the scaffolding protein Mouse protein-25 (MO25) through phosphorylation and thus activation of cytoplasmic AMPK (Boudeau et al. (2004), Narbonne et al. (2010)). Mutations in LKB1 can lead to the autosomal dominant disease Peutz-Jeghers syndrome (PJS) (Mehenni et al. (1998), Beggs et al. (2010)). Inactivating mutations of LKB1 are frequently observed in non-small-cell lung cancer (NSCLC), cervical carcinoma and malignant melanoma (Wingo et al. (2009), Ndembe et al. (2022)).

RIPK1 acts as a central stress sensor to control cell survival, inflammation and cell death signaling (Clucas and Meier (2023)). Deregulation of RIPK1 and RIPK3-involved signaling cascades have been linked to inflammatory bowel diseases, rheumatoid arthritis, autoimmune conditions and neuroinflammatory diseases such as Alzheimer’s and Parkinson’s disease (Martens et al. (2020), Li et al. (2019), Speir et al. (2021), Clucas and Meier (2023)). This is believed to be related to deregulations of both catalytic and scaffolding functions. Thus, a collection of small molecule blockers has been identified to interfere with RIPK1 signaling and function. It’s intriguing to note that RIPK1 and RIPK3 may share a similar mechanism of action with BRAF, as their structural arrangements and the dimerization of their kinase domains bear resemblance (Raju et al. (2018)). In addition to this, auto-phosphorylation of RIPK1 and RIPK3 represents a central regulatory element, with similar ties to BRAF (Laurien et al. (2020)).

Unlike the auto-inhibitory mechanisms of kinase regulation, some kinases dynamically engage with activating and deactivating polypeptides (Zhang et al. (2021)). One of the best studied examples is the interaction of regulatory proteins with the catalytic subunits of the cAMP-controlled serine/threonine protein kinase A (PKA) (Taylor et al. (2012), Zhang et al. (2020)). In a similar manner CDKs form active and inactive protein complexes to directly promote the cell cycle (Goel et al. (2018)). The central role of CDK4/6 lies in regulating cell cycle progression by phosphorylating and activating the key substrate retinoblastoma protein (Rb) that promotes G1 to S phase transition. On the molecular level this is controlled by complex formation of CDK4 and CDK6 with regulatory polypeptides (James et al. (2008)). Binding to p16^INK4a, one of the most frequently mutated tumor suppressor proteins, blocks CDK4/6 functions (Quelle et al. (1997)). Cancer mutations in p16^INK4a counteract this. As a result, kinase activating cyclinD proteins bind to CDK4/6 to promote carcino-genesis (VanArsdale et al. (2015)). Thus, CDK4/6 inhibitors (CDK4/6i) found the way into the clinic, in particular for treating breast cancer patients (Nebenfuehr et al. (2020),Yu et al. (2006). The development of drug resistance upon CDK4/6 inhibitors therapies underscores the need for personalized treatments that consider the patient’s genetic profile and the underlying alterations of CDK4/6 complexes present in their cancer cells (Álvarez-Fernández and Malumbres (2020), Knudsen and Witkiewicz (2017)).

Various cellular mechanisms are employed to control the molecular switch-like-behavior of protein kinases to temporally lock them into either an active or inactive conformation. These mechanisms include the PTMs of specific residues, the binding of regulatory proteins or co-factors, and allosteric changes induced by ligands (Bhullar et al. (2018), Taylor et al. (2021), Newton (2001)). This precisely controlled mode of action can be hindered, amongst others by patient-specific mutations or modified by bio-active small molecules (Figure 1A). Conventional methods often fall short in capturing the dynamic functioning of kinases within their native cellular environments (Klaeger et al. (2017), Croce et al. (2019)). This makes innovative biotech approaches essential for a comprehensive understanding of their roles in the cell. Here we show that genetically encoded KinCon reporters are extendable to many more kinases, to enable systematic monitoring of cellular kinase activity states in live cells. In addition to its predictive capabilities, the KinCon technology serves as a valuable tool for uncovering cellular factors influencing drug efficacy (Mayrhofer et al. (2020), Fleischmann et al. (2023), Röck et al. (2019)). Such insights into the molecular dynamics of kinases upon interactions with small molecule inhibitors or regulatory proteins is necessary for the design of more effective therapeutic strategies.

Results

Kinases act as molecular switches to integrate, amplify, restrict and/or relay signal propagation in spatiotemporal fashion and in cell-type specific manner. The precise coordination of usually oscillating kinase activities is a prerequisite for proper signal transmission (Taylor et al. (2012), Pan and Heitman (2002)). When investigating pathological kinase functions for therapeutic purposes, it is crucial to take the cellular elements that influence kinase activities into account. In figure 1A we list a collection of factors which affect kinase activity states. It is thus essential to closely monitor the physiological and pathophysiological kinases functions of activation and deactivation in intact cell settings in the presence and absence of patient mutation and drug exposure.

One cell-based technology for studying cellular kinase activity states is the KinCon reporter technology (Röck et al. (2019), Mayrhofer et al. (2020), Fleischmann et al. (2023)). It is a highly sensitive assay for tracking kinase activities which is mirrored by the alterations of full-length kinase structures in cells. With in-silico predictions it was projected that more than 200 protein kinases of the human kinome contain cis-regulatory elements. In many cases these sequence stretches may act as AIMs (Mayrhofer et al. (2020), Yeon et al. (2016)). In figure 1B we depict exemplary how underlying kinase ON and OFF states can be tracked using KinCon reporter technology (Enzler et al. (2020). The read out is based on the molecular motion of the full-length kinase containing diverse cis-regulatory elements such as AIMs. Cell-type specific KinCon measurement allow the testing of the influence of mutations, signaling pathway activation, binary protein interactions and drug binding on the respective reporter dynamics. We have previously applied the technology to gain insights into the functioning of two kinases that belong to the MAPK pathway. In these proof of concept studies, we showed that BRAF and MEK1 KinCon reporters are direct real-time readouts for kinase activities in intact cell settings caused by mutation and drug treatments (Röck et al. (2019), Mayrhofer et al. (2020), Fleischmann et al. (2021), Fleischmann et al. (2023)).

The overall construction principle of the KinCon reporter is modular. For the generation of the genetically encoded KinCon reporter the sequence of the kinase of choice is inserted into the multiple cloning site (MCS) of a mammalian expression construct. The MCS is flanked by the coding regions of a split luminescent protein for cellular over-expression experiments of the encoded hybrid reporter protein (see the KinCon reporter protein domain structure at the top of figure 1C). In many cases it is sufficient to fuse to the N and C terminus of the full-length kinase sequence the two fragments of the respective reporter protein (with intervening flexible linker stretches shown in red, Figure 1B, C).

In this study we used the protein-fragment complementation fragments (PCA, -F[1] and -F[2]) of the Renilla luciferase (RLuc-PCA) (Stefan et al. (2007)). The KinCon reporters are constructed to facilitate the intramolecular complementation of appended Rluc PCA fragments. Transient expressions offer the flexibility to analyze different time frames for KinCon reporter expressions and drug candidate exposures, in either low or high throughput format in intact cells (Figure 1C). Besides applying wild-type reporters it allows for the ’personalization’ of the sensor set-up by integrating patient-specific mutations, co-expressing regulatory proteins, or making systematic changes to post translational modification sites.

Following KinCon reporter expression along with co-expression of interacting molecules in the appropriate cell plate format, systematic perturbations can be applied. Following addition of the luciferase substrate to cells grown in a mono-layer or in suspension cellular bioluminescence signals are emanating from complemented RLuc PCA fragments (Figure 1B, C) (Röck et al. (2019), Mayrhofer et al. (2020)). Light recordings and subsequent calculations of time-dependent dosage variations of bioluminescence signatures of parallel implemented KinCon configurations aid in establishing dose-response curves. These curves are used for discerning pharmacological characteristics such as drug potency, effectiveness of drug candidates, and potential drug synergies (Figure 1C). In order to enhance our understanding of kinase dynamics we selected a group of kinases which activities are altered in different pathological settings. These examples emphasize how mutations, PTMs, PPIs, or kinase drugs induce context-dependent effects on the conformation states of kinases. In figure 1D we present a schematic and simplified depiction of the kinase’s ON and OFF conditions for complexes emanating from the kinases BRAF, LKB1, RIPK1 and CDK6. Exemplary, we show how protein complex formation and patient mutations contribute or perturb kinase activation cycles. As starting point, we illustrated the high sensitivity of the reporter system for tracking basal activity conformations of the kinases BRAF-V600E, LKB1, RIPK1, and CDK6 respectively. We showed that transient over-expression of these KinCon reporters for 10h, 16h, 24h or 48h in HEK293T cells delivers consistently of increasing signals for all KinCon reporters (Figure 1E, Figure Supplement 1A). Immunoblotting cell lysates after luminescence measurements showed expression levels of the reporters in the range and below the endogenous expressed kinases (Figure 1E).

Next, we analyzed the BRAF kinase activity conformations using wild-type and mutated KinCon BRAF reporters. The V600E mutation, found primarily in melanoma patients, effectively immobilized BRAF and the respective KinCon reporter in its and opened and active conformation (Davies et al. (2002), Lavoie et al. (2020), Karoulia et al. (2017), Lito et al. (2013), Röck et al. (2019)). Previously we have shown that FDA-approved melanoma drugs (Vemurafenib, Encorafenib, Dabrafenib) and one drug candidate from clinical studies (PLX8394) converted the opened BRAF-V600E reporter back to the more closed and thus inactive conformation (Röck et al. (2019), Mayrhofer et al. (2020), Yao et al. (2019)). Using this readout, we showed that at expression levels of the BRAF KinCon reporter below the immunoblotting detection limit, one hour of drug exposure exclusively converted BRAF-V600E to the more closed conformation (Figure 1F, G, Figure Supplement 1B). These data underline that at expression levels far below the endogenous kinase, protein activity conformations can be tracked in intact cells. This may represent the more authentic (patho)physiological context and takes the molecular interactions with endogenous factors into consideration.

Next, we adapted the KinCon biosensor technology to investigate the correlation between conformation and activity-regulation of key kinase pathways. We analyzed three different kinase pathways displaying different modes of kinase ON-OFF regulation.

Trimeric LKB1 complexes

Upon heterotrimeric complex formation with the pseudokinase STRADα and the scaffolding protein MO25, the kinase LKB1 contributes to the activation of AMPK by phosphorylation at the position Thr172 (Figure 2A). This activation mechanism acts as a key step in regulating glucose and lipid metabolism in response to cellular energy fluctuations in the cell. Additionally, AMPK has been shown to have a conserved role in cell polarity. (Shackelford and Shaw (2009), Boudeau et al. (2004),Baas et al. (2003)). It is the pseudokinase domain of STRADα that directly interacts with the kinase domain of LKB1, thus triggering the activation of LKB1‘s tumor-suppressing phosphotransferase functions (Figure 2B, C). It is assumed that upon ATP binding STRADα occupies an active conformation. In this scenario LKB1 affinities for binding rise and it binds as a pseudosubstrate. MO25 acts as scaffold for the kinase dimer to promote the activated LKB1 conformation state (Zeqiraj et al. (2009a), Zeqiraj et al. (2009b)). Further, MO25 binding stabilizes this trimeric cytoplasmic complex (Figure 2A, B) (Boudeau et al. (2003), Baas et al. (2003)).

Figure 2.

The capacity of STRADα for allosterically modulating LKB1 functions through direct interaction underlines that pseudokinases are more than inert scaffolds. They are involved in modulating interacting enzyme entities (Rajakulendran and Sicheri (2010), Reiterer et al. (2014)). This is in line with the common belief that pseudokinases can employ switch-like transitions to regulate signaling networks (Kung and Jura (2019), Shrestha et al. (2020)). Given the discovery of LKB1 inactivating mutations in diseases like PJS, NSCLC, and colorectal cancer, there is growing interest in exploring strategies to therapeutically restore the function of mutated LKB1 (Sanchez-Cespedes (2007), Launonen (2005), Kitajima et al. (2019)).

In contrast to blockers of kinase functions it is more challenging to identify activator molecules for promoting reactivation of the LKB1-AMPK axis. Thus, we investigated the impact of trimeric complex formation of LKB1:STRADα:MO25 on downstream activity of LKB1. We adopted the Kin-Con technology for measuring involved kinase conformation states. First, we transiently over-expressed the three flag-tagged polypeptides LKB1, STRADα and MO25 in HeLa cells. Next, we employed the pT172-AMPK/AMPK ratio as a measure of cellular LKB1 activity, observing the highest increase of LKB1 mediated downstream phosphorylation of AMPKα-T172 following co-expression of both, MO25 and STRADα respectively. Using the chosen cellular setting and transfection protocol with a 1:1:1 ratio of transfected expression constructs, a similar effect was observed when solely over-expressing STRADα. No impact on AMPK phosphorylation was observed when MO25 was over-expressed alone (Figure 2D).

First, we started to investigate the impact of LKB1 and MO25 co-expression on STRADα KinCon dynamics. Notable is the fact that only faint STRADα KinCon signals were detected in the absence of MO25 and LKB1 co-expression. To our surprise we have observed a more than 10-fold elevation of the reporter signals following co-expression of both interacting partners in HEK293T cells (48h of expression, Figure 2E). This data supports the notion that STRADα engages an opened conformation, where C- and N-termini are separated and thus almost no bioluminescence signal can be detected under basal conditions. Upon interaction with LKB1 this conformation shifts to a partially closed intermediate conformation state. The trimeric complex further promotes structure closing. We have observed the same tendency - but to a lower extent - using LKB1 KinCon readouts in the presence and absence of both interacting proteins (Figure 2F).

To validate these findings, we then tested the impact of LKB1-binding deficient STRADα-H231A/ F233A (HF) mutant proteins (Boudeau et al. (2004)) in co-expression experiments with the LKB1 KinCon reporter. It has previously been reported that this mutation prevents STRADα-LKB1 dimer and greatly hinders STRADα-MO25-LKB1 trimer formation and thereby AMPK activation (Zeqiraj et al. (2009a)). We observed that the so-called HF double mutation of STRADα is sufficient to abolish the elevating effect of trimer complex formation (Figure 2G). Therefore, we assume that the active, opened LKB1 conformation is not present upon coexpression of STRADα-HF. This might me due to a lack of onteraction between LKB1 and STRADα-HF. Since both mutations are located at the LKB1/STRADα binding interface, we hypothesize that allosteric communication between LKB1 and STRADα is essential for its function (Qing et al. (2022), Zeqiraj et al. (2009a)) (Figure 2H). These findings underline that indeed the trimeric complex formation alters KinCon dynamics. With this reporter technology we monitored the interaction based activities of of LKB1 directly in a living cell setting.

LKB1-loss of function mutations have been identified in a plethora of pathological conditions including autosomal diseases and many different forms of cancers (Sanchez-Cespedes (2007), Launonen (2005), Molaei et al. (2022)). Thus, we set out to analyze the impact of patient mutations on alterations of involved kinase conformations (Figure 2C).

In figure 2C we have listed the assumed functions of the tested mutations. Three ’tool’ mutations (K48R, R74A and K78I) (Zubiete-Franco et al. (2019), Qing et al. (2022), Yang et al. (2019)) and three patient mutations (D176N, D194N and W308C, in red) (Ui et al. (2014), Al Bakir et al. (2023), Islam et al. (2019)) were analyzed.

Figure 2H highlights the location of these mutations within LKB1. K78, D176, and D194 are highly conserved residues within the ATP binding pocket and are critical for kinase activity: K78 forms a salt bridge with E98 on the αC-helix which stabilizes a functional active conformation, the catalytic D176 is crucial for phosphoryl transfer, and D194 is part of the DFG motif involved in binding of the Mg²⁺ ion (Fabbro et al. (2015), Meharena et al. (2016)). The residue W308 is part of a hydrophobic cluster and thus surrounded by lipophilic residues (Zeqiraj et al. (2009a)). R74 is located within the STRADα binding interface, and has been reported to interact with STRADα Q251 (Zeqiraj et al. (2009a)), and K48 is a solvent-exposed acetylation site (Zubiete-Franco et al. (2019), Zeqiraj et al. (2009a)) located on the back of the N-lobe. Among these mutations, only the W308C mutation prevented the closing of the LKB1 conformation when co-expressed with STRADα and MO25 (Figure 2I). This indicates, that the catalytic site residues are critical for enzymatic activity, but play a less important role in maintaining the LKB1 active conformation in the presence of both STRADα and MO25. The W308C mutation has previously been identified to cause a reduction in LKB1 stability when compared to the wild-type variant (Islam et al. (2019)), suggesting the polar side chain of cysteine likely disrupts the interaction network within the hydrophobic cluster. Additionally, the LKB1 mutant W308C diminishes its catalytic activity (Islam et al. (2019), Boudeau et al. (2004), Mehenni et al. (1998)), consistent with the observed disruption of the LKB1 active conformation (Figure 2I). These findings suggest that LKB1-W308C lost its ability to form the heterotrimeric complex, with implications on hindering downstream activation and thus affecting its tumor suppressor functions. Overall these findings underline that KinCon technology can be extended to track the impact of binary protein complexes and related cancer mutations on kinase activity dynamics. Moreover, this data demonstrated that in contrast to the previously published MEK1 and BRAF KinCons (Fleischmann et al. (2021), Fleischmann et al. (2023), Röck et al. (2019), Mayrhofer et al. (2020)) the more closed STRADα and LKB1 KinCon conformations represent the more active kinase states.

Mutations and inhibitors induce RIPK1 conformation changes

RIPK1 is a multi-functional protein and central regulator of cell death, inflammatory processes, and immune responses (Figure 3A). Generally, RIPK1 can promote both, cell survival as well as cell death, highly depending on the regulatory inputs received from the cellular environment (Degterev et al. (2019), Clucas and Meier (2023), Silke and Meier (2013)). Genetic alterations of RIPK1 are linked to immune and autoinflammatory diseases. Examples are missense mutations of aspartate 324, which act as gain-of-function mutations of the protein kinase. Patients harboring these mutations suffer from chronic auto-inflammation caused by hypersensitive RIPK1 (Lalaoui et al. (2020), Tao et al. (2020)). Small molecule mediated inhibition of RIPK1 kinase activity were shown to counteract the necroptotic phenotype in disease models (Tao et al. (2020)). This exemplifies the potential of RIPK1 targeting drugs for the treatment of certain inflammatory diseases.

Figure 3.

We established a KinCon reporter platform for monitoring the conformational changes of kinases and pseudo-kinases involved in the necroptosis signaling pathway. Additionally, we examined the effects of binding of allosteric kinase blockers to both wild-type and mutant RIPK1 reporters. We depict the domain organization of RIPK1, RIPK3 and MLKL. We also included analyses of RIPK2 - which is not part of the necroptosis pathway (Figure 3B). We cloned the human versions of these kinase genes into KinCon reporter expression constructs. After transient over-expression of these kinases in HEK293T cells, we quantified the basal RLuc PCA signals and compared these to the basal signal of the BRAF KinCon reporter (Figure 3C).

As starting point, we evaluated whether RIP kinases 1 and 3 activation would manifest in dynamic conformational changes. We promoted necroptosis introducing stimuli by activating the TNFα pathway with TNFα (Christofferson et al. (2012)), while preventing repressive RIPK1 ubiquitylation with the SMAC mimetic BV-6 (Li et al. (2011)). To prevent the onset of apoptosis and instead induce activation of RIPK3, we blocked caspase 8 activation with the caspase inhibitor zVAD-FMK (Festjens et al. (2007)) (as illustrated in figure 3A).

We used HEK293T cells and transiently expressed the RIPK1 and RIPK3 KinCon reporters for 48 hours. Bioluminescence readouts of the KinCon reporters were quantified after indicated timings of TNFα, BV-6, and zVAD-FMK (TBZ) treatments. We observed successive accumulation of the respective KinCon reporter protein levels over the treatment time course (Figure Supplement 2A). We assume that this is caused by blocking of caspase 8, thus preventing the cleavage of both RIPK KinCon sensors. As consequence we normalized the obtained RLuc PCA signals on reporter expression levels. In doing so we found that both, the RIPK1 as well as the RIPK3 KinCon reporter showed markedly reduced bioluminescence in the necroptotic cellular environment (Figure 3D). These results support the notion that both kinases shift upon pathway activation to a more opened (ON state) full-length kinase conformation.

Conformational modulation of the RIPK1 KinCon following TNFα pathway activation suggests that the reporter is incorporated into complex I. We were interested to test the degree to which the RIPK1 KinCon represents a functional RIPK1 entity, despite the substantial modifications introduced to the polypeptide sequence. For this purpose, we conducted co-expression experiments of various RIPK1 KinCon and flag-tagged RIPK1 variants in RIPK1-deficient HEK293T cells. We used auto-phosphorylation at S166 as the primary readout for RIPK1 catalytic activity. Our results showed that the RIPK1 KinCon reporter was not capable of auto-phosphorylation (Figure Supplement 3B). In the presence of active flag-tagged RIPK1 however, trans-phosphorylation of the KinCon reporter hybrid proteins was evident (Figure Supplement 3B).

To further deepen our understanding of modes of RIPK1 activation we integrated a collection of missense mutations into the KinCon reporter. The assumed functions have been listed in Figure 3E. At the kinase N terminus a stretch of phosphorylation sites are attributed to kinase activation. Auto-phosphorylations at the positions S14, S15 and S166 have been shown to trigger downstream activation (Laurien et al. (2020), Wang et al. (2018)). The localizations are highlighted in the domain organization and the structure depiction of RIPK1 dimers (Figure 3E, F). Additional kinases like IKKα, IKKβ, TBK1, and MK2 have been linked to control the phosphorylation and activity state of RIPK1 (Dondelinger et al. (2019), Xu et al. (2018), Jaco et al. (2017)).

Auto-phosphorylation events have been described for the positions S14/15/166 and are critical for RIPK1 activation (Wang et al. (2021)) thus, we integrated mutations that mimic protein phosphorylation (S to E amino acid substitutions) and those that are deficient of phosphorylation (S to A amino acid substitutions), aiming to track their impact on molecular motions of RIPK1 to simulate the more activate kinase status. In addition, the kinase deficient mutant K45A was investigated (Shutinoski et al. (2016)). Interestingly, K45 in RIPK1 corresponds to K78 in LKB1, highlighting the importance of this lysine residue in the functionality of various kinases. Our KinCon findings unexpectedly demonstrated for RIPK1 that in addition to the kinase-deficient mutant (K45A) all the phosphorylation mimetic (S14/15/166E) and phosphorylation preventing (S14/15/166A) mutations tested led to an opening (ON state) of RIPK1 conformation (Figure 3G).

We also incorporated prevalent patient mutations of RIPK1, observed in distinct inflammatory diseases, into the KinCon biosensor. Besides the patient mutations of the regulatory caspase 8 cleavage site in RIPK1 (D324A/E/H) we integrated the C601Y missense mutation, which causes immunodeficiency and inflammatory bowel disease, into the reporter (Li et al. (2019)). Again, we observed an opening of RIPK1 conformations with all mutations tested (Figure 3H). This data supports the idea that in addition to preventing caspase 8 cleavage the mutations at position 324 affect the activity conformation of RIPK1. The C601Y substitution promotes RIPK1 opening as well.

We next tested RIPK1 KinCon settings with the allosteric RIPK1 modulators GSK547 and Necrostatin (Puylaert et al. (2022), Cho et al. (2011)). When bound to RIPK1, both of these compounds occupy a pocket located behind the ATP binding site (Figure 3I, Figure Supplement 2B). The kinase-deficient mutation (K45A) of the KinCon biosensor displayed the strongest opening of the kinase conformation. For this reason we used it to benchmark the effects of Necrostation and GSK547. Application of both compounds for various time points showed that these inhibitors engaged with all the tested RIPK1 KinCon reporters. We observed an increase of bioluminescence for all RIPK1 inhibitors tested, which reflects the closing of the kinase conformation. The inhibitors also affected the phosphorylation mimetic (S14/15/166E) and phosphorylation preventing (S14/15/166A) KinCon reporters to the more closed configuration when compared to the unrelated control, the MEK1 kinase inhibitor Cobimetinib (Figure 3J, Figure Supplement 3D). This data underlines that application of these two kinase blockers alter the conformations and we assume scaffolding functions of active and inactive RIPK1 protomers.

CDK4 and CDK6 interactions and conformations

CDK4 and CDK6 exhibit proto-oncogenic properties that rely on the presence and interaction with regulatory proteins (Scheiblecker et al. (2020), Eferl and Wagner (2003)). Further, CDK6 functions as a chromatin-bound factor and transcriptional regulator, thereby promoting the initiation of tumori-genesis (Kollmann et al. (2013), Semczuk and Jakowicki (2004), Kollmann and Sexl (2013)). These activities are modulated by its molecular interaction partners, particularly cyclins and proteinogenic inhibitors. CDKs are protein kinases that regulate a plethora of processes, most importantly cell cycle progression (Goel et al. (2018)). On the molecular level CDK4/6 control phosphorylation of the retinoblastoma protein (Rb) which acts as a transcriptional corepressor. This phosphorylation event leads to the dissociation of the Rb-E2F complex, enabling the transcription of target genes that facilitate cell cycle progression (Giacinti and Giordano (2006)). To activate CDK4/6 they form a complex with D-type cyclins (Gao et al. (2020)). The phosphotransferase function of CDK4/6 is inhibited by a number of different proteins, including the INK4 family of proteins (p16^INK4a, p15^INK4b, p18^INK4c, p19^INK4d) and members from the Cip/Kip family (p21^Cip1, p27^Kip1) (Figure 4A) (Lin et al. (2001)). Over 80% of all human cancers (e.g. breast, melanoma and pancreatic cancer) exhibit alterations of different kind in the CDK4/6-cyclinD-INK4-Rb cascade (Ortega et al. (2002)). Deregulation is caused by alterations of protein expression levels and/or by mutations within the crucial regulatory proteins. Prime example is the tumor suppressor protein p16^INK4a which is one of the most frequently mutated genes in human cancers (Romagosa et al. (2011)), Liggett Jr and Sidransky (1998)).

Figure 4.

The CDK4/6-cyclinD protein complex has been found to be hyperactivated in many cancers and therefore promote tumor growth (Choi et al. (2012)). This implies that CDK4/6 serves as a significant therapeutic target, and chemical inhibitors have been developed to target CDK4/6 phosphotransferase activities. Efficient CDK4/6 inhibitors such as Palbocilicb, Ribociclib and Abemacilib are currently in clinical use against breast cancer (Otto and Sicinski (2017)). A major drawback of these therapies is the emergence of drug resistance (Fassl et al. (2022)). The development of precision medicine oriented polypharmacology therapies is one strategy to avoid the advent of such underlying resistance mechanisms (Álvarez-Fernández and Malumbres (2020), Yang et al. (2017)). In this context it needs to be noted that CDK6 possesses a kinase independent function as nuclear transcriptional regulator (Kollmann et al. (2013), Kollmann and Sexl (2013), Scheicher et al. (2015)). This involves feedback loops involving STAT3 and cyclinD regulations of p16^INK4a expression (Kollmann et al. (2013)).

On the molecular level CDK4/6 activity states depend on the formation of PPIs with cyclinD or p16^INK4a, wherein cyclinD activates them, while p16^INK4a inhibits their activity of promoting Rb phosphorylation (Figure 4A). Distinct to the other kinases tested here, the CDK protein coding sequences are composed almost exclusively of the kinase domain with short additional stretches at the N and C termini (Figure 4B). In contrast to BRAF and RIPK1, the regulatory protein motifs are encoded by separate polypeptides. In the following, we tested the binary interaction of p16 with CDK4 and CDK6. Structures of these complexes have been described. It is evident that substitution of R at the position 31 of CDK6 with C (R31C) alters the binding affinities. CDK6-R31 is a crucial amino acid in the binding interface with p16^INK4a. It forms two ionic interactions with p16^INK4a residues D74 and D84, which are lost in the R31C mutant (Figure 4C). Experimental data further supports this notion (Rodríguez-Díez et al. (2014)). We included the respective mutations by generating CDK6-R31C and also the corresponding R24C expression construct in CDK4 (CDK4-R24C). We conceived a work program with PPI and kinase conformation reporters to elucidate the mechanism of dimerization in relation to drug binding.

We started analyzing binary protein interaction of p16 with CDK4 and CDK6. We fused the two RLuc fragments that are used in the KinCon reporter to two different proteins, in this case CDK4 or CDK6 and p16^ink4a. When two proteins interact, the two Rluc PCA fragments form a complemented luciferase and an increased light signal can be detected (Figure 4D, left). Originally the underlying technology had been developed for the dynamic measurements of regulatory kinase interactions of PKA (Stefan et al. (2007)). Besides testing the kinase mutants CDK6-R31C and CDK4-R24C we generated the murine homolog of the p16^INK4a displaying the cancer mutation P40L. P40L showed reduced affinities for CDKs and thus no longer inhibits phosphotransferase functions of CDK4/6 (Yarbrough et al. (1999)). Upon complex formation of CDKs with p16^INK4a the C terminal fused RLuc PCA reporter fragments complement and bioluminescence signals can be detected using intact cells transiently expressing the hybrid proteins. After co-expression of indicated constructs for 48 hours the cells were subjected to bioluminescence measurements (Figure 4D, Figure Supplement 1C). Using this cell based PPI reporter we observed a significant reduction of p16^INK4a:CDK4/6 complex formation when the respective R residues are mutated to C. Integration of the P40L mutation in p16^INK4a prevented complex formation as expected and further validates the reporter system applied (Figure 4D).

Next, we set out to test the impact of these CDK mutations on kinase conformations. We have previously shown that basal CDK4/6 KinCon conformation states can be tracked using the KinCon reporter technology (Mayrhofer et al. (2020)). Thus, we evaluated if a reduction of interaction with inhibitory proteins such as p16^INK4a alter CDK full length conformations. We transiently over-expressed the wt and mutant CDK4/6 KinCon reporter in HEK293T cells. After 48 hours of expression the cells were subjected to bioluminescence measurement. The results showed that the reduced affinity for p16^INK4a binding alters the conformations of both CDKs. CDK4- and CDK6-p16^INK4a binding deficient mutants exhibited a more opened conformation, as indicated by a decrease in bioluminescence (Figure 4E). In the following, all KinCon reporters have been subjected to analyzes upon CDK4/6 inhibitor binding (Yam et al. (2018)). Again we used HEK293T cells and transiently expressed the wt and mutated KinCon reporters for 48 hours. We then exposed the cells to three clinically applied CDK4/6i at a concentration of 1µM for 3h. Overall we could not detect any major effects of drug exposure on the dynamics of the four CDK KinCon reporters which showed different affinities for p16^INK4a binding (Figure 4F). This data underlines that these breast cancer drugs which should bind to the active protein kinase conformation did not induce full length CDK4 and CDK6 conformation changes in the tested cell culture settings.

The inhibitors used in this study, fall into different inhibitor categories (Type I, Type I 1/2, Type II and Type III). The classification of inhibitors is determined by the activation state of the protein kinase, particularly the positioning of the DFG-Asp (active in, inactive out) and the αC-helix (active in, inactive out) (Zhang et al. (2009)). However, type III (allosteric) inhibitors deviate from this pattern. They bind adjacent to the ATP-binding pocket, enabling simultaneous binding of ATP and the inhibitor (Wu et al. (2015)). Type I inhibitors selectively bind to the active conformation of the kinase, characterized by DFG in and αC in. Type I 1/2 inhibitors, on the other hand, bind to the inactive state of the kinase, represented by DFG in and αC out. Type II inhibitors bind to the inactive kinase, where DFG is out and αC can be either in or out (Arter et al. (2022)) (Figure 5A and B). In this study we have observed alterations of KinCon activity conformations upon changes of protein interactions and through type I 1/2 and type III inhibitor binding (Figure 5A-C), but not type I inhibitors. We provide a summary of the changes in KinCon activity conformations. We have previously shown that for both kinases of the MAPK pathway, MEK1 and BRAF the respective inhibitors affect primarily the active kinase conformation (Röck et al. (2019), Mayrhofer et al. (2020), Fleischmann et al. (2021), Fleischmann et al. (2023)). Exemplarily we illustrate that both kinases which are activated through cancer patient mutations (BRAF-V600E and MEK1-K57E) can bind the respective kinase inhibitor, resulting in a change in the activity conformations (Figure 5A, B). The same observation was made with the MEK1 KinCon upon activation through BRAF phosphorylation (Fleischmann et al. (2023)). In contrast it was of interest that for all tested RIPK1 activity conformations we observed that the binding of allosteric RIPK1i promoted a closing of the RIPK1 conformation. In Figure 5B we exemplarily illustrate the impact of inhibitor binding on the kinase-dead version RIPK1-K45A. This observation underlines that also inactive RIPK1 complexes are target of drug binding with feasible consequences on kinase scaffolding functions. Kinase activities of CDK4/6 are regulated via defined PPIs (Nebenfuehr et al. (2020)). We showed that reducing binding affinities for p16^INK4a and related inhibitory proteins triggers the opening of the kinase conformation (Figure 5C). However, using the applied standard protocols we have chosen for all KinCon we have not observed that any of the three clinically applied CDK4/6i tested (such as Abemaciclib) altered the kinase conformation significantly.

Figure 5.

In this context we would like to introduce another kinase example displaying regulatory PPI controlling the catalytic kinase protomer. When the regulatory (R) and the catalytic (C) subunit interact a tetrameric PKA holoenzyme complex (R2:C2) is inactive. When second messenger cAMP molecules bind to a R dimer the complex dissociates and catalytic PKA subunits are activated (Taylor et al. (2013)). Mutations such as L206R lock the catalytic subunit in its active state. Thus, interactions of regulatory subunits are significantly reduced (Mayrhofer et al. (2020), Bolger (2022)). Reduction of binding affinities through the use of the general cAMP elevating agent Forskolin, triggers conformation changes to the more opened state; mechanistically similar to CDK6-R31C. The PKAc-L206R KinCon mutant already engages this opened activity conformation state (Figure 5D). Based on these two examples we assume that binary protein interactions are the prime factors for altering CDK4/6 and PKA-C activity conformations.

In this study we focused on a collection of well-studied kinase pathways. In this regard it is important to note that only a limited number of approved drugs target human kinases. We have depicted this visually by employing an evolutionary-based representation of kinase relationships in the human kinome tree (Figure 5E). The blue squares indicate the kinase branches displaying approved kinase blockers. It is evident that numerous protein kinases lack drug candidates and a more suitable tool for basic research analyses of pathological kinase functions (these kinases are referred to as the ’dark kinome’) (Berginski et al. (2021), Southekal et al. (2021)). In the kinome tree illustration we have integrated our growing catalog of KinCon reporters (red dots, Figure 5E). This represents an extendable toolbox for gaining first but also deeper insights into the dynamic alterations of kinase activity structures induced by different means of inhibitors, activators or mutations. Systematic implementation of this technology in kinase research programs would assist in advancing our understanding of kinase and pseudokinase regulations and functions.

Discussion

In the field of kinase-related diseases, mutations of different kind and at different stages of kinase pathways have prompted the development of kinase-specific and small molecule-based inhibitors as primary therapeutic approach (Ferguson and Gray (2018)). Despite significant efforts in this direction, only a small fraction of the entire kinome has been targeted by FDA-approved kinase blockers so far (Lahiry et al. (2010), Politi et al. (2015)). Once a kinase inhibitor is introduced into clinical use, it often encounters the development of drug resistance mechanisms over time (Longley and Johnston (2005)). Especially in cancer therapy, these facts pose a significant challenge for identifying and applying efficient small molecule based therapies. Hence, it is crucial to gain a more extensive insight into the molecular consequences of kinase patient mutations and the evolving patterns of drug resistance mechanisms. For both endeavors new technologies are indispensable for identifying, accurately monitoring and targeting the dysregulated functions of kinases on the cellular level.

Anticipating and unveiling the cellular mechanism of drug actions may help to develop new treatment concepts by reducing the diminishing of drug efficacies (Chan and Ginsburg (2011), Goetz and Schork (2018)). The KinCon technology, introduced here, seeks to address the previously mentioned challenges. It has the potential to become a valuable asset for tracking kinase functions in living cells which are hard to measure solely via phosphotransferase activities. Overall, it offers an innovative solution for understanding kinase activity conformations, which could pave the way for more novel intervention strategies for kinase entities with limited pharmaceutical targeting potential. This relates to the tracking of kinase-scaffold and pseudo-kinase functions.

At the beginning of our studies, we underlined the exceptional sensitivity of the KinCon-reporter system for tracking kinase activity conformations. We demonstrated that at expression levels far below the endogenous kinase of interest we tracked kinase conformations and their alterations upon drug exposure (Figure 1E-G). Tracking enzyme activities at low expression levels is paramount for understanding the details of cellular functions, as many regulatory and spatiotemporal controlled kinase interactions occur under such settings. Such precise measurements are essential for unraveling and targeting pathological kinases functions.

So far, KinCon reporters have been used to assess different conformational states of kinases which are altered by phosphotransferase-activating patient mutations and/or kinase drug binding (Röck et al. (2019), Mayrhofer et al. (2020), Fleischmann et al. (2021), Fleischmann et al. (2023)). Here we set out to evaluate consequences of inactivating patient mutations on the complex formation emanating from the kinase LKB1. We aimed to broaden the application spectrum of this reporter technology for investigating how regulatory protein interactions influence the tumor suppressor functions of the kinase LKB1 (Partanen et al. (2012)). This holds particular implication for inactivated LKB1, as it suggests that activator compounds could potentially counteract its loss-of-function mutations in cancer. In order to screen or test such re-activator molecules, bioluminescence measurements in living cells would be advantageous, since the complex cellular environment needs to be taken into account. In in-vitro assays, which mostly focus on kinase domain activities, regulatory protein interactions are neglected. We have confirmed the notion that the formation of the trimeric complex between STRADα-LKB1-MO25 was necessary for signal propagation to AMPK. In line with this, the LKB1 and STRADα KinCon bioluminescence signals increased upon complex formation and vice-versa were reduced back to the baseline signal when a complex breaking mutation in the pseudokinase STRADα was introduced (Figure 2G). Further, we showed that tracking structural rearrangements induced by patient or regulatory mutations can be quantified using KinCon technology (Figure 2I). In this setting KinCon measurements of either the kinase (LKB1) or pseudokinase (STRADα) report trimeric complex formation which is pivotal for cytoplasmic kinase pathway activation (Figure 2E, F). In contrast to MEK1 and BRAF KinCons, the closed STRADα and LKB1 KinCon conformations seem to correspond to the more active kinase units.

In the next step, we expanded the application of the KinCon reporter to RIPK. Examining the molecular mechanisms that regulate RIPK1 activity poses a challenge because of the intricate nature of the pathway. Functioning as a molecular scaffold, RIPK1 orchestrates signal transduction by coordinating multiple kinases participating in the NFκB pathway (Moynagh (2005), Meylan et al. (2004)). Moreover, as a regulator of cell death, RIPK1 forms a substantial amyloid-like signalosome in conjunction with RIPK3 and various other factors (Degterev et al. (2019)). We have shown that indeed pathway activation at different levels of the cascade can be tracked. Further, all tested mutations converted RIPK1 to the more opened conformation (Figure 3G, H). Both tested and allosterically acting RIPK1i converted the KinCon reporter back to a more closed conformation (Figure 3J, 5B). This consistent impact on the open kinase conformation, indicates a uniform response in altering the enzymes structural state. We assume that this unexpected drug driven transition of inactive and active RIPK1 complexes to more closed kinase conformation may have relevance for scaffold functions of RIPK1. Further, we speculate that it may affect drug efficacies.

The proto-oncogenic characteristics of CDK4 and CDK6 depend on expression levels and their interactions with a collection of regulatory proteins (Lin et al. (2001). We applied the KinCon technology for tracking activity states of both kinases. We demonstrated that regulatory CDK4/6 interactions induce conformational changes that remain unaltered when clinically used type I inhibitors bind to them (Figure 4F). However, we would like to state that the involvement of cyclinD interactions will need to be taken into account for follow up studies. In general these observations will be relevant for targeting other kinases which are disease relevant and are thus current or future targets for drug discovery efforts. Overall the details of drug binding and action have been elaborated in last section of the results (Figure 5).

In summary, we have extended the KinCon reporter application spectrum. Besides monitoring conformational changes induced by drugs and drug candidates, it can be applied to accurately track the formation of activated and kinase centered protein complexes as well. Apart from its predictive value, the KinCon technology helps consider or identify cellular factors that impact drug candidate efficacies. This understanding of the molecular kinase dynamics as they interact with small molecule inhibitors or regulatory proteins is crucial for designing more effective therapeutic strategies, especially considering that many kinase pathways have so far remained untargeted. The KinCon technology offers a potential avenue to change this.

Materials & Methods

Expression Constructs

All constructs were cloned into the pcDNA3.1 expression vector containing F[1] and F[2] of the KinCon reporter. Linear DNA fragments were produced by PCR using Q5 DNA Polymerase. After removing the PCR overhangs with AgeI and HpaI (NEB) respectively and subsequent DNA fragment gel extraction, the PCR insert fragments were isolated using the innuPREP DOUBLEpure Kit (Analytik Jena). The DNA Fragments were the ligated using T4 DNA ligase (NEB) and amplified using XL10-gold ultracompetent cells. Plasmids were purified by mini- or midiprep (Quiagen) and verified by Sanger Sequencing (Microsynth/Eurofins).

Cell culture

HEK293T cells were obtained from ATCC (CRL-11268). HEK293T cells were grown in DMEM supplemented with 10% FBS. Transient transfections were performed with Transfectin reagent (Biorad, 1703352). HeLa cells were obtained from ATCC (CCL-2). HeLa cells were grown in DMEM supplemented with 10% FBS. HEK293T RIPK1 knock-out cell lines were provided by Pascal Meier. Transient transfections were performed with JetPRIME DNA and siRNA transfection reagent (Polyplus supplied by VWR, 101000046).

All cells are tested regularly for mycoplasma by PCR using suitable primers and/or Universal Mycoplasma Detection Kit (ATCC, 30-1012K).

Immunoblotting

Cells were lysed in ice-cold RIPA buffer (50mM Tris-HCl pH 7.4, 1% NP-40, 0,25% Na-Deoxycholate, 120mM NaCl, 1mM EDTA, 1mM PMSF, 1µg/mL Leupeptine/ Aprotinin/ Pepstatin , 1mM Na₃VO₄/ Na₄P₂O₇/ NAF) and mixed with 5xLaemmli Buffer. After heating to 95°C for 10 minutes, the samples were loaded on 10% Acylamide SDS gels for subsequent electrophoresis. Gels were transferred to a PDVF membrane (Roth) using either the Trans-Blot SD Semi-Dry Transfer Cell (Bio-rad) or the Mini Trans-Blot Cell (Biorad), blocked in TBS-T with 2,5% BSA for 30 minutes at room temperature and incubated in the primary antibody over night at 4°C. Blots were incubated with secondary antibodies for 1h at room temperature and washed with TBS-T before imaging. Imaging was performed with a FUSION FX (Vilber). Immunoblot images were analyzed using ImageJ (NHI). The signal of the target protein was then normalized on the indicated loading control (e.g GAPDH or Vinculin). To normalize AMPK phosphorylation (pAMPK) levels to those of total AMPK, GAPDH-normalized pAMPK levels were divided by their respective GAPDH-normalized total AMPK levels from the same experiment. Primary antibodies used were the rabbit anti-GAPDH (14C10) (Cell Signaling, 2118), rabbit anti-AMPKα (Thr172) (40H9) (Cell Signaling, 2532), rabbit-anti-Phospho-AMPKα (Thr172) (40H9) (Cell Signaling 2535), rabbit-anti-LKB1 (D60C5) (Cell Signaling 3047), rabbit-anti-Vinculin (Cell Signaling 4650), rabbit-anti-RIP (D94C12) XP (Cell Signaling 3493), rabbit-anti RIP3 (E1Z1D) (Cell Signaling 13526), anti-CDK6 (DCS83) mouse (Cell Signaling 3136S), mouse-anti-FLAG M2 (Sigma Aldrich F3165-1MG), rabbit-anti-Renilla Luciferase (EPR17792) (Abcam ab185926), mouse-anti-Renilla Luciferase clone 1D5.2 (Millipore MAB4410).

Luciferase PPI Assay

PCA was performed by growing HEK293T cells in DMEM supplemented with 10% FBS in a 24-well plate format. PCA analyses was performed similarly as previously described in (Röck et al. (2019)). HEK293T cells were grown in DMEM supplemented with 10% FBS. The indicated RLuc-tagged constructs, one with RLuc[F1] and one with RLuc[F2], were transiently over-expressed with TransFectin reagent (Bio-Rad, 1703352) in a 24-well plate format. Fourty-eight hours after transfection the medium was carefully aspirated, the cells were washed once with PBS (1mM Sodium phosphate ph 7,2; 15mM NaCl) and after addition of 150µL of PBS to each well, transferred to a 96-well plate (Grainer 96 F-Bottom) . Bioluminescence was measured after addition of 20µL h-Coelentracine (Nanolight Technology) using the PHERAstar FSX (BMG Labtech) (Measurement start time [s]: 0,2 ; Measurement interval time [s]: 10,00; Optic module LUM plus; Gain: 3600; Focal height [mm] 12,5). Data was evaluated using the MARS Data evaluation Software (BMG Labtech).

Luciferase PCA Assay (KinCon Assay)

HEK293T cells, cultured in StableCell DMEM (Sigma Aldrich), were split into 24-well plates at 90.000c/ well and after 24h the indicated plasmids were transfected using TransFectin Lipid reagent (Bio-Rad 1703352) at a total of 50-66ng/well. After 48 hours of Protein expression cells were washed once with PBS (1mM Sodium phosphate ph 7,2; 15mM NaCl) and after addition of 150µL of PBS to each well, transferred to a 96-well plate (Grainer 96 F-Bottom) . Bioluminescence was measured after addition of 20µL h-Coelentracine (Nanolight Technology) using the PHERAstar FSX (BMG Labtech) (Measurement start time [s]: 0,2 ; Measurement interval time [s]: 10,00; Optic module LUM plus; Gain: 3600; Focal height [mm] 12,5 ). Data was evaluated using the MARS Data evaluation Software (BMG Labtech).

Inhibitors

Inhibitors used were Palbociclib (PD 0332991) (MCE Med Chem Express, HY-50767), Abemaciclib (LY2835219) (MCE Med Chem Express, HY-16297A), Ribociclib (LEE011) (MCE Med Chem Express, HY-15777), PLX8394 (MCE Med Chem Express, HY-18972), GSK-547 (MCE Med Chem Express, HY-114492), Necrostatin 2 racemate (1S; Nec-1S; 7-Cl-O-Nec1) (MCE Med Chem Express, HY-14622A), Cobimetinib (GDC-0973; XL518) (MCE Med Chem Express HY13064), TNFα (MCE Med Chem Express, HY-P7090A), z-VAD(OMe)-FMK (MCE Med Chem Express, HY-16658), BV6 (MCE Med Chem Express, HY-16701).

Statistical analyses

The data were analyzed using GraphPad Prism 8.0. If not indicated otherwise, one-sample t-tests were used to evaluate statistical significance. Values are expressed as the mean ±SEM as indicated. Significance was set at the 95% confidence level and ranked as *p<0.05, **p<0.01, ***p<0.001.

Preparation of structures

The LKB1-STRADα-MO25α (PDB code 2WTK (Zeqiraj et al. (2009a))) and CDK6-p16^INK4a (PDB code 1BI7 (Russo et al. (1998))) complex structures were prepared using the default setting of the Protein Preparation Wizard (Madhavi Sastry et al. (2013), Schrödinger (2019)) in Maestro Schrödinger release 2019-4 (Schrödinger Release 2019-4: Maestro, Schrödinger, LLC, New York, NY, 2019). The construct used to generate the LKB1 structure contained the inactivating mutation D194A, intended to prevent Mg²⁺ binding (Zeqiraj et al. (2009a)). This mutation was converted back to the wildtype Asp residue using MOE version v2022.02 (Molecular Operating Environment (MOE), 2022.02; Chemical Computing Group ULC, 1010 Sherbrooke St. West, Suite 910, Montreal, QC, Canada, H3A 2R7, 2022). For the calculation of the CDK6 R31C mutant, only the CDK6 residues within 12Å of p16^INK4a were retained. The mutation was then generated using Osprey v2.2beta (Chen et al. (2009), Gainza et al. (2013)) as described before (Kaserer and Blagg (2018)). Figures were generated using PyMOL version 2.5.0 (The PyMOL Molecular Graphics System, Version 2.5.0 Schrödinger, LLC).

Funding

Austrian Science Fund (FWF: P30441, P32960, P35159, I5406, P34376), FFG Bridge ’MitoKin’ (877163)

Other Chemistry Niceties

Patents

Aspects of the present study are subject of patents and pending patent application.

Author Contributions

Conceptualization, J.T., V.S., P.M., T.K., E.S.; methodology, E.S.; validation, formal analysis, investigation, data curation, V.K., S.S., S.S., A.F., F.E, J.F., T.K.; writing—original draft preparation, V.K., S.S., S.S,. E.S.; writing—review and editing, E.S., P.T., T.K.; supervision, project administration, funding acquisition, E.S. All authors have read and agreed to the published version of the manuscript.

Conflicts of Interest

E.S and P.T. are co-founders of KinCon biolabs.

Acknowledgements

We thank Thomas Nuener for technical, and Erika Lentner and Gabriele Reiter for management support. We thank Alexandra Fritz for their contributions to advance the KinCon technology.