Functional and Structural Insights into RAS Effector Proteins

Summary

RAS proteins are conserved guanosine triphosphate (GTP) hydrolases (GTPases) that act as molecular binary switches and play vital roles in numerous cellular processes. Upon GTP binding, RAS GTPases adopt an active conformation and interact with specific proteins termed RAS effectors that contain a conserved ubiquitin-like domain, thereby facilitating downstream signaling. Over 50 effector proteins have been identified in the human proteome, and many have been studied as potential mediators of RAS-dependent signaling pathways. Biochemical and structural analyses have provided mechanistic insights into these effectors, while studies using model organisms have complemented our understanding of their role in physiology and disease. Yet, many critical aspects regarding the dynamics and biological function of RAS-effector complexes remain to be elucidated. In this review, we discuss the mechanisms and functions of known RAS effector proteins, provide structural perspectives on RAS-effector interactions, evaluate their significance in RAS-mediated signaling, and explore their potential as therapeutic targets.

eTOC blurb

RAS proteins act as molecular switches by binding GTP and interacting with effectors to facilitate downstream signaling. This review by Mozzarelli et al. discusses the mechanisms, structures, and functions of RAS effectors, their roles in signaling, and their potential as therapeutic targets, emphasizing the need for deeper insights into RAS-effector interactions.

Introduction

The RAS family of small GTPases, comprising 36 members, exhibits homology in their protein sequence. Advancements in sequencing and cloning technologies have revealed their phylogenetic relationship and highlighted critical differences in protein sequences present within their distinct regulatory motifs^1,2. The latter has significant importance in the biological function of RAS GTPases, defining their biochemical properties, interactions with regulatory proteins, subcellular localization, and ability to engage with downstream protein effectors. Most RAS GTPases exhibit high affinity towards guanosine diphosphate and triphosphate (GDP/GTP, respectively) and their intrinsic enzymatic activity promotes GTP hydrolysis to GDP, with RAS proteins being loaded with GTP through a nucleotide exchange reaction. These cyclic reactions are inherently catalytically inefficient, and therefore, guanine exchange factors (GEF) facilitate GTP loading, and GTPase activating proteins (GAP) accelerate the rate of hydrolysis³. The coordinated function of these factors ensures efficient, rapid, and tightly regulated RAS signaling in cells (Figure 1A).

Figure 1. The fundamentals of RAS-effector signaling.

(A) Representative overview of the RAS binary switch. The effector binding domain (highlighted in blue) undergoes extensive conformational changes upon nucleotide exchange facilitating effector complex formation (red). PDB 6MBT (KRAS-GDP), 6MBQ (KRAS-GppNHp), and 6XHB (KRAS and CRAF RBD-CRD). (B) Potential factors affecting the competitive nature of RAS effectors (different effectors depicted as colored shapes). Dissociation constant (K_D) is the main driver of effector affinity, and some studies have systematically assessed the interaction (K_D) between RAS proteins and effectors. For instance, HRAS^GTP K_D (μM) for CRAF (0.09) < RASSF5 (0.24) < RALGDS (2.5) < PLCε (3.7) < PI3Kα (84.3)⁴⁰. The dynamics of the interaction are also affected by association (k_on) and dissociation (k_off) rate constants, affecting the interaction between RAS and different effectors even if their affinity at equilibrium is similar⁷. Local concentrations¹⁸⁸ (or subcellular localization) and post-translational modifications of effectors can also modify their apparent affinity. For instance, RAP1 S11 phosphorylation and H/K/NRAS K147 mono-ubiquitination have been suggested to modify RAF interaction^189,190 (C) Dendrogram of the RAS family of GTPases highlights their phylogenetic relationships and main clades, representing important subfamilies like classical RAS proteins (in blue). (D) Protein sequence alignment in the effector binding domain of different RAS GTPases demonstrates similarities and differences in critical residues that contribute to effector specificity. Amino acids are highlighted as follows: small polar (orange), hydrophobic (green), polar (magenta), negatively charged (red), and positively charged (blue). Red font indicates similar properties to the KRAS amino acid, but distinct amino acid (synonymous change). Amino acid sequences include KRAS (32–40), RRAS (58–66), RALA (43–51), RAP2A (32–40), RIT1 (50–58), ERAS (70–78), RHEB (35–43), RHES (48–56), DIRAS1 (36–44) and GEM (106–114). (E) Structural overlay of RAS protein in RAS-effector complexes, highlighting the conformational similarities of key interacting residues in the effector binding domain.

RAS binding to GTP induces conformational changes, creating a unique binding interface for interaction with specific effector proteins, which possess a conserved ubiquitin fold domain known as the RAS-binding domain (RBD). Other effector domains binding to RAS were initially classified as RAS-association (RA) domains due to their limited sequence similarity to RBDs, with both types typically spanning 80–100 amino acids. However, subsequent structural studies have shown that, despite low primary sequence homology, these domains possess the characteristic ubiquitin fold and a similar RAS interaction mechanism, suggesting their reclassification under the RBD category^4,5. While the presence of RBD/RA indicates the potential for a specific protein to act as a RAS effector, this is not always the case, making the definition of a RAS effector somewhat ambiguous. At a minimum, a hypothetical RAS effector must bind its cognate RAS protein only when GTP-bound. In addition, this interaction should lead to changes in the activity or function of the effector; although there are also effectors that just act as a scaffold. Therefore, confirming a protein’s role as a RAS effector necessitates experimental validation.

Our current view of RAS-effector signaling predominantly follows a linear model and assumes that each RAS protein has multiple downstream effectors to diversify the signaling output. However, a major question in the field is how we can reconcile a model in which different effectors compete for limited GTP-bound RAS molecules⁶. In this conundrum, the importance of the RAS effector kinetics is crucial; not only the affinity towards an effector is important, but how such affinity is affected by different factors (Figure 1B). For instance, are there high-affinity effectors that signal first and block available RAS molecules for other low-affinity effectors? More importantly, do these also modify the association between RAS and GAPs through steric hindrance, given the overlapping nature of their binding interface? Conversely, can low-affinity effectors play a significant role downstream of RAS-GTP if their local concentration is high enough? In this context, systems biology approaches are beginning to tackle these important questions^7–9.

There are critical concepts regarding the biology of RAS effectors that warrant at least reevaluation and, at most, substantial reconsideration. For example, many structural and biophysical studies carried out using RAS effectors use isolated RBD/RA domains, despite reports showing additional binding contacts outside of these domains^10,11. Moreover, isolated RBD/RA do not necessarily capture important functional details of the full-length protein, such as post-translational modifications that could modify affinity, other adjacent domains that might occlude accessibility, or its subcellular localization, among others^10,12,13. Another important aspect that remains understudied is the role of RAS effectors in the in vivo context. While many potential effectors have been characterized in vitro or in cell-based assays, their role during development, disease, or at the organismal level remains poorly understood. In fact, there are limited studies using mouse models in which the suppression of an effector can counteract phenotypes associated with a gain-of-function RAS allele.

The diversity of RAS proteins and their binary switch

The RAS superfamily of small GTPases encompasses over 160 members distributed in 5 distinct families: RAS, RHO, RAB, ARF, and RAN^2,14. Within the RAS family, there are 36 proteins that are further classified into distinct monophyletic groups based on their amino acid sequence conservation¹⁵ (Figure 1C). Notably, the subfamily of RAS GTPases can play dual roles in regulating cellular proliferation, broadly categorized into those promoting cell proliferation or survival and those exhibiting growth-suppressive properties¹⁵. Key members such as classical RAS proteins, H/K/NRAS, are frequently mutated in cancers and, hence, have been extensively studied for their roles in cell proliferation, apoptosis, and differentiation. Similar functions appear to be regulated by the closely related members RRAS/RRAS2/MRAS. Other members, like RALA/B and RAP1/2, are implicated in cell migration, adhesion, and cytoskeletal organization, while RHEB activates the mTORC1 signaling pathway, impacting cell growth and metabolism¹⁶. However, many RAS proteins remain poorly characterized at the biochemical and cellular levels.

RAS proteins are characterized by a globular domain, known as the G-domain, and N- and C-terminal extensions that provide unique properties to these proteins. The C-terminal extensions are particularly important because they generally contain amino acids responsible for subcellular localization and show low sequence conservation across the family; for this reason, this extension is also referred to as the hypervariable region (HVR). Many RAS proteins contain a C-terminal CAAX box (C: cysteine; A: aliphatic residue; X: terminal residue), which is prenylated at the cysteine residue and promotes plasma membrane anchorage. Additionally, some RAS proteins are palmitoylated at one or two cysteines upstream of the CAAX motif, while others, lacking these motifs, associate with the plasma membrane, or endomembranes through charge complementation¹⁷. The role of the HVR is highly relevant in the context of effector signaling because subcellular localization can dictate the ability to bind to specific effectors. Molecular simulations have proposed that different HVRs can position the G-domain in configurations with differing accessibility to certain effectors^18,19.

Within the G-domain, there are several regulatory regions that are important for the coordination of the nucleotide, interaction with effectors, and other regulatory mechanisms. The G-domain of most RAS proteins acts as a molecular binary switch that toggles between conformational states bound to GDP/GTP. The GTP-bound state, also known as the active conformer, allows for effector binding and activation of signaling cascades that regulate essential cellular functions. Structural analyses of RAS have identified two critical switch regions—switch-I (aa 30–38 in H/K/NRAS) and switch-II (aa 59–76) that are central to protein-protein interactions and exhibit dynamic conformational shifts between the GDP/GTP-bound states, often compared to a loaded-spring mechanism^20,21. In the loaded spring state, RAS presents the interface residues essential for effector engagement (mainly the effector binding domain; aa 32–40). After GTP is hydrolyzed and the γ phosphate is released, RAS reverts to a GDP-bound state, causing the switch regions to adopt ‘open’ conformations, releasing effector proteins. Despite the high conservation of the G-domain, variations in regulatory motifs diversify the properties of RAS proteins, particularly those affecting the effector binding motif. These variations lead to changes in affinity and, thus, effector selectivity (Figure 1D, E).

RAS-GTP binds specific domains in RAS effector

The discovery of RAF and PI3 kinases as RAS effectors accelerated our understanding of the molecular mechanisms of effector association. RAF was the first RAS effector discovered, and its exclusive association with RAS-GTP, validated by several studies, confirmed its role in selectively transmitting signals from active RAS^22–26. The N-terminal region of CRAF (RAF1) had been shown to antagonize the transformation effects of oncogenic RAS in cells, and consequently, this region was found to bind RAS-GTP directly through its RBD²⁷. This discovery was integral in understanding the initiation of the RAF-MEK-ERK (MAPK) signaling cascade downstream of RAS. In parallel, PI3K emerged as a second class of validated RAS effector after initial findings linked PI3K activity with RAS²⁸, but the direct interaction between PI3K catalytic subunit with RAS-GTP was only substantiated later²⁹. This interaction was delineated through RAS-GTP binding to p110α, the catalytic subunit of the PI3Kα lipid kinase complex, and subsequently extended to the p110γ isoform, mediated through their respective RBDs^30,31.

Following the identification of RAF and PI3K as RAS effectors, several exploratory efforts were launched to identify additional protein effectors. Utilizing cDNA libraries coupled with yeast two-hybrid assays using RAS GTPases as baits revealed a broad range of RAS effectors. Additionally, using the consensus amino acid sequences characteristic of RBD/RA domains, data mining of the human proteome identified over 50 proteins containing similar domains that could act as potential effectors of RAS-like GTPases (Figure 2). Beyond RAF and PI3K kinases, this diverse family of effectors includes several RAL GTPase GEFs (e.g., RALGDS, RGL1–4), AFDN/Afadin, the RASSF family, RIN1, and PLCε1, among others^32–39. This diversity underscores the complexity of RAS signaling, extending past the known RAF and PI3K-mediated growth and survival pathways. However, detailed studies focusing on the specificity of effector binding within the extensive RAS family of GTPases have been limited, despite their importance in understanding the potential binding preference. Furthermore, the mechanisms by which binding to RAS family GTPases alters the functions of these effector proteins remain largely unexplored for most effectors.

Figure 2. Proposed human RAS effectors.

There are over 50 human proteins in the human proteome that contain a prototypical RBD/RA domain in their amino acid sequence and have been proposed as RAS effectors. The figure shows a dendrogram resulting from the amino acid alignment of the RBD/RA domains of the indicated potential RAS effectors. Heat maps represents the protein expression levels of these effectors in the selected tissues (PRIDE Dataset)¹⁹¹. The values represent protein levels normalized to the highest expressing protein across each tissue (a) and to the highest espressing tissue (b) (and given a value of 100).

RAS-effector interactions

Several factors influence the formation of RAS-effector complexes in cells, including affinity, selectivity, competition, and the potential effect of neomorphic mutations (those changing the molecular properties) in RAS GTPases, among others. Classical RAS proteins exhibit a high binding affinity towards RAF kinases, with dissociation constants (K_D) ranging from 50–200 nM⁴⁰. This affinity is in stark contrast to interactions with other effector molecules, which generally fall into the low micromolar range. Structural models of 54 putative RAS effectors have been used to estimate the binding affinity to KRAS, revealing striking differences⁸. Yet, RAS proteins are tethered to membranes and an effector containing a membrane-targeting domain can enhance affinity by over 100-fold, effectively increasing local concentration and reducing the three-dimensional reaction space to two dimensions⁴¹. In addition, previous studies have shown that RAS proteins organize into transient nanoclusters on the plasma membrane, each containing approximately 5–6 molecules, which affect effector recruitment and activation⁴². Contrary to earlier theories of RAS dimerization influencing RAF kinase activity, current consensus suggests that interactions of HVR with membrane phospholipids drive RAS clustering⁴³. This conclusion is supported by several evidences, which show similar clustering patterns irrespective of the G-domain⁴⁴. Typically, the affinities between RAS-GTP proteins and effectors are measured using their isolated RBD domains; while this approach provides biophysical insights into RAS-RBD interaction, it fails to consider important aspects of the effector’s regulation. For instance, some RBDs are occluded by adjacent domains through autoinhibitory conformations; in other cases, adjacent domains may directly interact with the RAS protein, modifying the overall binding affinity^10,45,46.

Another important aspect is selectivity, the effector’s ability to preferentially bind one RAS protein over another. This selectivity is influenced by specific residues in the effector binding domain and their respective affinities. This is well-exemplified by RAF kinases, which display selective interactions with H/K/NRAS, and exhibit minimal interactions with other RAS proteins. In contrast to the high-affinity binding of RAF kinases, other effectors, such as PI3Kα and RASSF5, demonstrate a more moderate affinity that can lead to a less restricted specificity⁴⁰. For instance, PI3Kα is known to interact with additional RAS GTPases like the members of the RRAS subfamily⁴⁷. Broader interaction profiles could indicate a more versatile role in cellular signaling, contrasting highly selective interactions of RAF. One of the few studies that analyzed interaction and downstream effects of both well-characterized and novel RAS effectors in a large panel of RAS proteins, revealed remarkable differences in their selectivity; certain effectors exhibit more selectivity than others⁴⁸. Importantly, affinity and selectivity are not necessarily interconnected; this is exemplified by the effector Sin1, which shows similar in vitro affinities towards H/K/NRAS, but selectivity for KRAS4a in cells¹⁰. This originates from the unique HVR, likely due to subcellular localization or G-domain accessibility. Therefore, affinity and selectivity need to be assessed experimentally both in vitro and in cells.

Effector proteins play a crucial role in directing downstream signals of RAS GTPases, involving multiple direct effectors that are often co-expressed and co-localized, highlighting the intricate interplay and competition initially recognized as the problem of specificity⁶. Since all effector RBDs target the same switch regions of RAS-GTP, this leads to a competitive dynamic among effectors, especially when the GTPase availability is limited. From a biochemical perspective, the effector with the highest affinity is typically a winner in this competition. However, experimental data in cellular and organismal contexts indicate that lower-affinity effectors also play significant biological roles. Thus, it’s conceivable that additional factors influence this competition. Unfortunately, the complexity of this problem has prevented the development of robust experimental models to test this hypothesis, leaving limited experimental evidence available in this regard other than models from systems biology^8,49,50. To date, it is still unknown how effectors’ activity/function is regulated upon RAS-GTP interaction; some proposed hypotheses include localization, changes in effector conformation, or a combination of these. The dynamic nature of the RAS nucleotide cycling, effector binding and inactivation by GAPs has been reviewed by others⁵⁰ and experimental approaches using in vitro network reconstitution have revealed the intricate relationship between GEFs/GAPs and effectors⁵¹.

Pathogenic mutations in RAS genes result in higher levels of RAS-GTP, increasing the total number of RAS-effector complexes and potentially altering their binding profile. This may result in interaction with lower-affinity effectors or force atypical interactions. These mutations not only favor RAS-GTP by disrupting the GTPase cycle but can also act as neomorphs. This is exemplified with RASopathy alleles, where certain RAS proteins, which typically do not bind RAF kinases, show enhanced signaling through this pathway^52–54. Because RAS oncogenic mutations have been widely used as a model to increase RAS-GTP in cells, one must consider whether the effects caused by effectors using this model are truly due to their physiological association with RAS-GTP or as a result of the neomorphic mutation. Hence, although many effector proteins seem to play crucial roles in cells with mutant H/K/NRAS, it is likely that most of the >50 effectors discovered so far act as physiological effectors for less explored RAS-like GTPases. This has been well-exemplified in several studies that systematically analyzed the interaction of different effectors with distinct RAS GTPases^48,55,56. Thus, a comprehensive understanding of RAS-effector signaling necessitates analyzing all effectors in a biological system, considering the concentrations and affinities of all entities involved for the cellular outcome.

Structural insights into RAS-effector complexes

Structural studies of RAS-GTP in complex with RBD/RA (and adjacent domains) of various effectors have advanced our understanding of RAS-effector interactions (Figure 3).

Figure 3. Structural overview of RAS-effector complexes.

(A) Domain organization of human effector proteins, whose structures have been solved in complex with RAS proteins, is depicted. The effector domain(s), which are part of the RAS-effector complex, are shown in color, while the remaining domains are in gray. (B) The structures of RAS-effector complexes solved so far are shown in cartoon representation. These structures were aligned using RAS proteins, which are depicted in a grey cartoon with switch-I and switch-II regions colored purple and blue, respectively. Nucleotides and Mg ions are represented as sticks and spheres and are colored gray and green, respectively. RBD and adjacent domains are color-coded following the same scheme as in panel A. The structures of CRAF (PDB: 6XI7), PI3Kγ (PDB: 1HE8), GRB14 (PDB: 4K81), and SIN1 (PDB: 7LC1) in complex with H/KRAS were solved with RBD and adjacent domains, whereas the structures of RALGDS (PDB: 1LFD), RGL1 (PDB: 7SCW), RASSF5 (PDB: 3DDC), AFDN (PDB: 6AMB), and PLCε (PDB: 2C5L) in complex with HRAS were primarily solved using RBD/RA domains.

RAF kinases (A/B/CRAF) feature an N-terminal regulatory domain containing RBD and CRD (cysteine-rich domain) and a C-terminal kinase domain (Figure 3A). The RAS-RAF interaction was first elucidated through the crystal structure of RAP1A, a RAS-like GTPase, in complex with the CRAF-RBD, revealing that the RBD adopts a β-sandwich ubiquitin-like fold and forms a continuous β-sheet at the interface⁵⁷. Subsequent research with H/KRAS validated this interaction pattern^11,58. The CRD not only aids RAF in attaching to membranes but also strengthens its binding to RAS¹¹. Crystallographic data reveals that the RBD and CRD form a single elongated entity, with both domains extensively interacting with RAS (Figure 3B). Distinct from the switch-I region-mediated RAS-RBD interface, the RAS-CRD interaction mainly involves the interswitch region and the α5-helix in RAS proteins^11,59. Alterations in the RAS-CRD interface markedly diminish CRAF activation with minimal effect on the binding strength, underscoring the criticality of RAS-CRD interactions for complete RAF activation. Cryo-electron microscopy studies examining BRAF complexes have enhanced our knowledge of RAF’s autoinhibited and active states⁶⁰. These investigations underscore that the autoinhibited state is stabilized through interactions between the kinase domain, the CRD, and 14-3-3 proteins, with pivotal phosphoserines located at both ends of the kinase domain^61–64. Activation of RAF by RAS involves RAS binding to the RBD, leading to the disruption of 14-3-3 interactions and the subsequent release of the CRD. This series of events results in RAF dimerization at the plasma membrane, culminating in its full activation. The full-length, active BRAF–14-3-3 complex in the active state yielded no structural details beyond the BRAF kinase domain, suggesting the necessity of incorporating membrane mimetics in future studies to gain a more comprehensive understanding of the RAF activation mechanism.

The four class-I PI3K isoforms, namely PI3Kα, PI3Kβ, PI3Kγ, and PI3Kδ, each contain an RBD. Active RAS proteins bind to the p110α, p110γ, and p110δ catalytic subunits of the PI3K lipid kinase complex via their respective RBDs. RAS exhibits a GTP-dependent yet comparatively weaker binding affinity to PI3K than RAF⁴⁰. A crystallographic study demonstrated that HRAS binds to the PI3Kγ RBD through its switch-I and II regions, while only the switch-II region interacts with PI3Kγ’s catalytic domain, emphasizing RAS’s dual role in membrane attachment and allosteric activation⁶⁵ (Figure 3B). Unlike the other class-I PI3K isoforms, PI3Kβ does not bind to RAS-like GTPases; instead, it is activated by the RHO GTPases CDC42 and RAC1⁴⁷.

Utilizing yeast two-hybrid, RALGDS, a GEF for RAL GTPases, was identified as RAS effector³⁸. RALGDS links RAS and RAL signaling, thereby regulating cellular functions like vesicular trafficking and migration^38,66. The crystal structure of the HRAS and RALGDS-RA complex revealed a heterotetrameric arrangement, where each HRAS molecule interacts with two RALGDS-RA molecules, one via the switch-I region and another through the switch-II region⁶⁷. However, later studies suggested that the heterotetrameric arrangement was an artifact of crystal packing, with the physiological interaction primarily involving the switch-I region of RAS⁶⁸. In mammals, three paralogues of RALGDS have been identified: RGL1 (RALGDS Like 1), RGL2, and RGL3. Recent study suggested that KRAS complexed with RGL1-RA predominantly forms 1:1 dimers in solution, with tetramerization occurring through a very slow association⁶⁹ (Figure 3B). The interaction of PI3Kγ and RALGDS with RAS is facilitated by forming an antiparallel β-sheet between the RBD and RAS, similar to that observed with RAF. This supports a conserved mode of RBD-RAS recognition, although there are distinctive differences in the relative orientation between the RAS and RBD⁶⁵ (Figure 3B).

The RAS Association Domain Family (RASSF), comprising ten members (RASSF1–10), represents a large group of RBD-containing effector proteins. Recent research revealed that only RASSF5 directly interacts with classical RAS proteins, distinguishing it from other RASSFs that bind non-classical GTPases^55,56. The crystal structure of RAS in complex with the RA domain of RASSF5, containing an additional N-terminal helix, displays the typical RA domain interaction with RAS’ switch-I region seen in other RAS-effector complexes⁷⁰. However, the N-terminal helix of RASSF5-RA uniquely forms a distinct hydrophobic interaction with the switch-II region of activated RAS (Figure 3B).

AFDN (also known as AF6/MLLT4/Afadin) is characterized by two RBDs and plays a crucial role in the formation and maintenance of cell-cell adherens junctions⁷¹ (Figure 3A). The crystal structure of RAS with AFDN’s first RBD shows a classical interaction⁷². However, similar to RASSF5, AFDN’s first RBD is predicted to feature an N-terminal helix that interacts with the switch-II region of RAS. The NMR structure of AFDN’s second RBD showed a typical ubiquitin-like fold, and alone, it has a very weak affinity for RAS or RAP1⁷³. Recent work has shown the interaction of AFDN’s second RBD with RAP2C and, to a lesser extent, HRAS in cells. Interaction of RAS GTPases with AFDN promotes SCRIB interaction, which likely participates in cell contacts and polarity regulation⁷⁴.

Phospholipase Cε (PLCε), containing two RA domains (RA1/2), is activated by RAP1 and RAS and regulates endocytosis, exocytosis, and cytoskeletal reorganization^75,76. NMR studies revealed that both RA domains adopt a ubiquitin fold similar to the RAF-RBD⁷⁷. However, only RA2 binds to RAS, engaging with both switch regions akin to PI3Kγ. Recent studies revealed that the RA1 domain forms extensive interactions with other PLCε domains to maintain structural and functional integrity⁷⁸.

GRB14, belonging to the GRB7–10-14 family of multi-domain cytoplasmic adaptor proteins, acts as a tissue-specific regulator of insulin signaling⁷⁹. GRB14 has been shown to regulate negative feedback on the insulin receptor⁸⁰. The complex structure of GRB14’s RA and pleckstrin homology (PH) domains with RAS reveals that only the RA domain directly interacts with RAS, forming the typical RAS-RBD interaction surface, while the PH domain does not participate in this interaction⁸¹ (Figure 3B). It has been suggested that the PH domain may enhance the RA domain’s positioning for effective interaction with plasma membrane-tethered GTPases, such as RAS and RAP1.

Sin1, encoded by MAPKAP1, is an essential component of the mTORC2 and contains a RBD; in cells, it preferentially binds KRAS4a^10,82. Unlike the RAS-GRB14 (RBD-PH) complex, structural analyses of KRAS-Sin1 (RBD-PH) revealed an expanded interaction interface that encompasses the typical RAS-RBD interaction and an additional interaction interface between the RAS switch-II region and the Sin1 linker-PH domain¹⁰ (Figure 3B). The KRAS-Sin1 interface is notably more extensive than other RAS-effector complexes, highlighting the importance of both the PH domain in Sin1 and the switch-II region in RAS in facilitating RAS-effector interactions.

In general terms, most RAS-effector complexes are formed by antiparallel β-sheet formations between the RBD β1/β2 and the RAS GTPase β2/β3. Most RAS residues engaging with effectors are conserved, particularly notable hotspots in and around the switch-I region, including Q25, D33, I36, E37, D38, S39, Y40, R41, and in the switch-II region, Y64 (H/K/NRAS numbering)⁴⁰. Conversely, the majority of residues in RBD/RA domains interacting with RAS display variability. A comparison of CRAF-RBD and RASSF5-RA revealed only two distantly conserved residues between them⁵⁶. A study analyzing specificity in binding interfaces of RAS with RBDs of CRAF, PI3Kγ, and RALGDS suggested significantly shared binding contacts⁸³. Some RAS residues are crucial for binding all three effectors, while others alter specificity. Mutations in RAS residues D33, D38, I36, and Y40 typically disrupt binding across all three effectors. Specific hydrophobic mutations at I36 impede binding to PI3Kγ and RALGDS but not CRAF, whereas certain changes at L56 enhance CRAF binding and variably affect the others⁸³. Alterations at E37 tend to decrease binding to CRAF and RALGDS but increase it for PI3Kγ. Y64 mutations selectively diminish PI3Kγ and RALGDS binding while maintaining CRAF interaction, and certain mutations at S39 specifically hinder binding to PI3Kγ and CRAF, sparing RALGDS⁸³.

Structural studies on the H/KRAS-CRAF(RBD-CRD) and KRAS-Sin1(RBD-PH) complexes have highlighted crucial interactions with RAS proteins beyond the RBD^10,11,65. Given that most RAS-effector complex structures have been obtained primarily using the RBD alone, it is possible that adjacent domains of the RBD/RA in these multi-domain effector proteins may also engage with RAS proteins.

Atypical RAS effectors and isoform-specific effectors

Given the low protein sequence homology among RBD and RA domains, it is also tempting to speculate that many RAS effectors have not yet been discovered, as these will only be able to be identified by experimental means. This notion is supported by the recent identification of “atypical” effectors lacking classical RBD/RA domains. For instance, like Sin1, Hexokinase 1 (HK1) is a KRAS4a effector¹³. The interaction between HK1 and KRAS4a was attributed to the palmitoylation-depalmitoylation cycle of KRAS4a, facilitating its colocalization with HK1 on the outer mitochondrial membrane. Intriguingly, HK1 lacks the typical RBD/RA sequence or structure. Yet, the full-length HK1 structure reveals a surface-exposed helix–loop–sheet motif that partially resembles the RBD, enabling it to interact with KRAS4a. These findings underscore the possibility that certain RAS GTPases may interact with effectors that do not contain typical RBD. Although there are isoform-specific interactors that have been shown to affect the biology of RAS isoforms, many of these do not satisfy the definition of effector. The identification of novel isoform-specific effectors requires proteomic approaches using different isoforms as baits in both GDP/GTP bound forms, as shown by many recent studies^84–87. Future work will be necessary to assess whether some of these interactors bind in a GTP-dependent manner and if they contain atypical RBD domains.

Signaling through RAS effectors at the organismal level

The function of RAS effectors has been studied across model organisms, from yeast to mammals, given the evolutionary conservation of RAS GTPases and their protein effectors (Figure 4A). Notably, high conservation is observed in the switch-I motif, crucial for interactions with downstream effectors⁸⁸. The budding yeast Saccharomyces cerevisiae has two RAS genes, Ras1 and Ras2, with comparable functions but distinct regulatory mechanisms governing their expression^89–91. Ras1/2 bind and activate their downstream effector Cyr1, an ortholog of adenylate cyclase that contains an RBD-like domain, catalyzing cAMP synthesis. This, in turn, activates Bcy1, a Protein Kinase A subunit, driving cell cycle progression, stress response, and spore morphogenesis^91–93. Unlike S. cerevisiae, the fission yeast Schizosaccharomyces pombe features a single Ras protein (Ras1) and two main effectors, Byr2 and Scd1. Byr2, part of a kinase cascade including Byr1 and Spk1 activated by Ras1 through its RBD-like domain, regulates pheromone signaling, akin to the mammalian MAPK pathway^91,94. While Scd1, a guanine exchange factor (GEF) that regulates Cdc42 and contains a distinct type of RBD, is involved in spindle formation and cytokinesis and is similar to mammalian RALGDS^91,95–97.

Figure 4. Signaling pathways of RAS-effector complexes in model organisms and humans.

(A) Biochemical pathways regulated by RAS effectors in model organisms, including fission and baker’s yeast, slime mold, roundworms and fruit flies. Note the conserved modes of signal transduction, despite the diverse phenotypic outputs. (B) Overview of some of the different RAS effector signaling pathways that have been characterized in humans. Although many of these have been studied in the context of oncogenic KRAS signaling, they might represent pathways physiologically regulated by non-classical RAS GTPases.

In the slime mold Dictyostelium discoideum, ten RAS proteins have been identified, with RasD and RasG akin to human classical RAS proteins^98–100,101. Moreover, the orthologs of several mammalian RAS effectors (PI3K and MAPK pathways) have been identified in this organism^100–102. Dictyostelium’s PI3K pathway regulates chemotaxis upon RasG binding, promoting downstream activation of AKT. Similarly, RasC and Rap1 can bind directly to the Sin1 ortholog RIP3 and modulate the activity of the TorC2 complex. In addition to AKT, activation of PI3K in this organism leads to the recruitment of other PH domain-containing proteins such as cytosolic regulator of adenylyl cyclase (CRAC) and PhdA, which in turn promote pseudopod propulsion through F-actin polymerization¹⁰³.

The genetic characterization of the roundworm Caenorhabditis elegans was pivotal in discovering the role of RAS proteins in larval development, specifically, vulval differentiation^104,105. LET-60 is considered as the main RAS ortholog^106,107. In addition, screenings for vulval development defects have revealed downstream effectors in this organism, underscoring the RAS-MAPK pathway’s role in regulating this phenotype in worms^105,108,109. Indeed, lin-45, mek-2 and mpk-1 were identified as the RAF, MEK and ERK orthologs^110–114. However, other genes have also been identified downstream of LET-60, including Sur-8/soc-2 and ksr-1, which positively regulate the RAS-mediated signaling pathway^{105,115–117}. These genes are conserved in humans as SHOC2 and KSR1 and are now known to play critical roles in this signaling pathway.

The discovery of a RAS ortholog (Ras1) in Drosophila melanogaster has advanced our understanding of RAS signaling in fruit flies. Ras1 is responsible for differentiating progenitor cells into neuronal cells, namely eye photoreceptor cells^118,119. Initiation of the signal transduction cascade through Ras1 is modulated upstream by the receptor tyrosine kinase Sevenless (Sev) and the GEF Son-of-sevenless (Sos), while Gap1 terminates the signaling^{118,119–121}. Drk (GRB2) serves as a scaffold of Sev and Sos, leading to the activation of Ras1 and its subsequent binding to Raf, in a process that is highly conserved to the mammalian pathway^{118,122–124}. Activated RAS primarily facilitates RAF’s plasma membrane localization through its conserved RBD, where RAF activation occurs¹²⁵. Once active, RAF initiates the MAPK module through Dsor and Rolled^{118,126–128}.

The insights gained from studies in model organisms played a pivotal role in uncovering the intricacies of RAS signaling in mammals (Figure 4B). While the precise outcomes of the RAS pathway may differ among organisms, the fundamental mechanisms of effector signaling remain conserved across species, using domains that bind only to RAS-GTP. While the evolutionary conservation of the RAS-MAPK cascade is remarkable, the complexity of the RAS signaling pathway increases in mammals, highlighting the significance of studying RAS-effector signaling pathways in less-complex model organisms. In humans, there are >50 putative RAS effectors, compared to the more limited number of effectors in model organisms^5,72,129. Moreover, the larger number of RAS proteins in humans contributes to the complexity of the signaling network. In mammals, the best-studied downstream pathways of classical RAS proteins are the MAPK and PI3K pathways. In the MAPK pathway, RAS-GTP recruits the RAF kinases, initiating a complex activation mechanism¹³⁰. A critical step in this process is the dephosphorylation of RAF proteins at a site N-terminal to the kinase domain by the MRAS-GTP/SHOC2/PP1c complex, underscoring the regulatory role of non-classical RAS proteins in various signaling pathways. Upon dimerization and activation, RAF kinases phosphorylate downstream kinases MEK1/2, which subsequently phosphorylate ERK1/2. This MAPK cascade ultimately stimulates the phosphorylation of many substrates, including transcription factors that promote cellular proliferation. Moreover, ERK1/2 can also negatively regulate the MAPK pathway by phosphorylating and inhibiting SOS1 or CRAF^131,132.

In the context of PI3K signaling, alongside classical RAS proteins, RAS-like GTPases such as RRAS, RRAS2, MRAS, and ERAS have been shown to interact with PI3Kα, indicating their potential function as physiological activators of the enzyme⁴⁷. Upon binding to RAS-GTP, PI3K increases its catalytic activity synergistically with RTK stimulation, phosphorylating phosphatidylinositol 4,5-biphosphate (PIP₂) to promote the formation of phosphatidylinositol (3,4,5)-triphosphate (PIP₃)¹³¹, a second messenger that leads to the recruitment of AKT to the plasma membrane via its PH domain. Membrane-localized AKT is phosphorylated and activated by the kinases PDK1 and mTORC2, enabling it to phosphorylate numerous substrates involved in cell growth and metabolism, including TSC2, PRAS40, FOXO transcription factors, and others¹³³.

Activation of RALGDS by RAS-GTP facilitates its translocation to the plasma membrane and potentially exposes its REM domain, which possesses GEF activity for RALA/B GTPases. RALA/B proteins also have numerous downstream RAS effectors, including RALBP1 and Sec5, to regulate various cellular phenotypes. RALBP1 acts as a GAP for RAC GTPases¹³⁴, and Sec5 is a critical component of the exocyst complex¹³⁵. The activation of these pathways promotes actin organization, exocytosis and cytokinesis¹³⁶. Like RALGDS, other RAL GEF proteins, such as RGL1–3, contain RA domains that interact with different GTP-bound RAS-like proteins with different degrees of specificity⁴⁸.

RASSF5, also known as NORE1A, was initially identified as the first member of the RASSF family through a yeast two-hybrid screen using mutant RAS as bait^35,137. Recent investigations have revealed that RAS proteins can modulate the Hippo pathway by utilizing RASSF family proteins as effectors¹³⁸. MST kinases were identified as direct interacting partners of RASSF proteins, facilitated by a SARAH domain¹³⁹. Given their pro-apoptotic characteristics¹⁴⁰, these kinases have emerged as promising candidates for mediating RASSF-driven tumor suppression¹⁴¹. MST kinases activate the LATS kinases through phosphorylation, leading to subsequent phosphorylation of the transcriptional regulators YAP and TAZ¹⁴². In addition, although NORE1A plays a role in modulating apoptosis, its primary function may lie in acting as a potent effector of RAS-induced senescence^{137,143–148}, but the precise mechanisms by which RAS regulates these pathways, and the contribution of RASSF proteins, remain poorly understood.

Potential therapeutic strategies targeting RAS effectors

Understanding the importance of RAS effectors is also crucial for its therapeutic implications. Identifying critical effectors downstream of activated RAS proteins can provide novel therapeutic targets for cancer and congenital disorders such as RASopathies. Mutations in genes encoding for classical RAS proteins are found in several cancer types; KRAS is mutated in lung, pancreatic and colorectal adenocarcinomas^{131,149–151}, while N/HRAS mutations are frequent in melanoma, head and neck squamous carcinomas and bladder cancer^{131,152–154}. In RASopathies, activating variants in several RAS genes are also frequent¹⁵⁵. In all these conditions, the diseased phenotype solely depends on downstream effector signaling.

Given the longstanding challenges of direct RAS inhibition, many therapeutic efforts for RAS-driven cancers have focused on inhibiting downstream effectors. Using animal models of KRAS-driven cancer, it was shown that genetic ablation of the MAPK pathway suppressed tumor onset¹⁵⁶. Although ATP-competitive RAF inhibitors have been developed for several tumor types, these inhibitors do not seem to work in the context of RAS-mutant tumors. Indeed, these compounds induce paradoxical activation of the MAPK pathway that results from feedback relief^157–159. To overcome this compensatory mechanism, combinations with MEK inhibitors have been developed, although these have been mostly beneficial in BRAF-mutant tumors, including melanoma, lung and thyroid cancers, but not in RAS-mutant tumors. Studies using next-generation pan-RAF inhibitors show less paradoxical activation, opening new therapeutic strategies for the outcomes of patients with RAS-mutated tumors^160–163. Based on preclinical evidence, a more targeted approach could involve the specific inhibition of CRAF, particularly through targeted degradation, since this effector appears essential for oncogenic KRAS-driven tumors¹⁵⁶. In RASopathies driven by activated RAS, the MAPK pathway seems to be the most important downstream target, and the use of MEK inhibitors in this setting is currently being tested with early reports of efficacy¹⁶⁴.

Using mouse models of KRAS and EGFR mutant-driven lung tumors, it was demonstrated that PI3K is required for tumor onset and maintenance¹⁶⁵. PI3K inhibitors that specifically target this pathway have been tested in KRAS cancers; however, as monotherapy, these inhibitors do not work in RAS-driven preclinical models nor in clinical trials. Although the combination of MEK and PI3K inhibitors seemed more promising in mouse models¹⁶⁶, clinical trials testing this combination showed significant toxicity and questionable efficacy^131,167,168.

Targeting other effectors has been tested preclinically as potential therapies for RAS-driven cancers, although no drugs have been developed yet. For example, depleting RALGDS in tumor models reduces cancer incidence, size, and malignant progression in a multistage skin carcinogenesis model¹⁶⁹. Similarly, TIAM1 knockout mice were more resistant to HRAS-driven transformation in such skin models and embryonic fibroblasts formed less foci upon KRAS G12V expression¹⁷⁰. Mice in which PLCε had been knocked out exhibited opposing results regarding susceptibility to tumor formation upon the skin carcinogenesis protocol, so its role remains questionable^171,172. Screenings in different cell lines driven by RAS have uncovered synthetic lethal interactors¹⁷³. Of these, many are not direct effectors, but CRAF and SHOC2 appear to be essential in these models and are common hits in these screenings¹⁷³. Although SHOC2 has not been considered a RAS effector, its binding specificity to RAS-GTP forms and recruitment of catalytic activity via protein phosphatase 1 (PP1) proteins makes it act like a RAS effector. Another independent study validated the crucial role of SHOC2 in a KRAS-driven lung cancer model, as the ablation of SHOC2 inhibited tumor development and prevented resistance to MEK inhibitors¹⁷⁴. A potential strategy for targeting all RAS effectors involves disrupting the RAS-RBD interaction. Preclinical studies indicated that rigosertib, mimicking activated RAS, disrupts the binding of RAF, PI3K, and RALGDS to RAS-GTP, thereby impeding downstream signaling^175,176. However, clinical trials failed to show its efficacy, and subsequent studies raised concerns regarding its specificity^177,178. An interesting development in targeting effector RBDs is exemplified by a small molecule (BBO-10203) that binds the PI3Kα-RBD near the RBD-RAS interface, effectively blocking its interaction with classical RAS proteins¹⁷⁹. These compounds lay the groundwork for developing similar inhibitors that target the RBDs of other effector proteins, thereby inhibiting their activation by RAS-like GTPases.

Directly targeting allosteric pockets in RAS or effector RBDs near the interface is currently the most promising strategy for treating RAS-driven cancers. Recent advancements include the development of KRAS G12C (GDP) covalent inhibitors^180,181, currently approved for treating KRAS G12C mutant lung cancer. Targeting only the inactive KRAS G12C may induce resistance by increasing active GTP-bound KRAS levels; thus, inhibitors targeting both active and inactive states may overcome this resistance¹⁸². Recently, inhibitors targeting both nucleotide states of KRAS G12C have been reported^183,184. Similarly, multiple efforts have also focused on targeting other RAS mutant alleles and pan-RAS or pan-KRAS inhibitors^182,185. These inhibitors either lock RAS in its inactive state or prevent RAS-effector interactions. In this context, a recent chemical strategy that leverages the formation of a tri-complex between the inhibitor, RAS, and cyclophilin A has been reported¹⁸⁶; these molecular glues prevent the association with RAS effectors due to steric hindrance. Given that RAS-GTP utilizes overlapping surfaces for interacting with effectors and GAP proteins, approaches aimed at stabilizing mutant RAS-GAP interactions via molecular glues¹⁸⁷, preferably also increasing GTPase activity, could effectively block RAS-effector interactions.

Future directions and challenges

Despite the extensive work done to date, a comprehensive and multifaceted experimental approach will be essential to enhance our understanding of RAS-effector interactions. There are several aspects of their biology that will need to be reevaluated: (1) Structural analysis of RAS complexed with full-length effector proteins could reveal the interaction between RAS and RBD-adjacent domains and their collective influence on downstream signaling pathways. (2) The structural aspects of RAS-effector complexes in membrane-mimicking environments are critical, as this aligns closely with physiological conditions, offering a more accurate depiction of these interactions in cellular membranes. (3) Understanding the specific biological roles of classical RAS isoforms and RAS-like GTPases in effector activation, since many effectors are likely activated by non-classical RAS proteins. (4) Study whether oncogenic RAS proteins might activate some of these less-characterized effectors in a neomorphic manner as the active RAS population becomes significantly higher. (5) Discovery of potential isoform-specific interactions with effectors by mechanisms that extend beyond the G-domain binding. (6) Assessing the dynamics of RAS-effector complex formation at the organismal level. Exploring innovative experimental strategies could significantly advance our understanding of RAS-effector interactions. Techniques such as proximity ligation and in vivo tagging of RAS GTPases (and their constitutively GTP-bound mutants) offer promising avenues to address current limitations. Integration of these techniques with robust biochemical and structural analyses will help address key open questions in the field.