Consensus on the RAS Dimerization Hypothesis
, by Dhirendra K Simanshu, Mark R Philips, John F Hancock
Dhirendra Simanshu leads the structural biology efforts at the NCI RAS Initiative, Frederick National Laboratory for Cancer Research. His research focuses on the structural characterization of KRAS in complex with various regulators and targeting druggable pockets using structure-based drug design.
John Hancock is executive dean of McGovern Medical School at the University of Texas Health Science Center at Houston. His research is focused on the plasma membrane interactions of RAS proteins and how the spatiotemporal organization of membrane lipids contributes to the assembly and regulation of RAS signaling complexes.
Mark Philips is a Professor of Medicine, Cell Biology and Biochemistry in the Perlmutter Cancer Center at NYU Grossman School of Medicine. He has directed a molecular cell biology laboratory for 30 years focused on the post-translational modification and membrane targeting of RAS and related small GTPases. His many contributions to the field include characterization of isoprenylcysteine carboxylmethyltransferase, endomembrane trafficking and signaling of RAS, and, most recently, differential localization and function of the splice variants of KRAS.
RAS GTPases regulate various intracellular signaling pathways. Mutations in one of the three RAS genes (HRAS, KRAS, and NRAS) are present in almost 20% of human cancers. These oncogenic mutations cause RAS to switch to the GTP-loaded active state, resulting in the constant activation of downstream effectors, including the RAF kinase. The question of whether RAS dimers exist and their significance in activating RAF kinases is a topic of intense debate in RAS biology. The idea of RAS dimers was initially proposed based on the observation that RAF kinases function as dimers, leading to the notion that RAS dimerization through the G-domain might initiate RAF dimer formation. RAS dimerization has generated considerable interest as a mechanism required for RAS function and a potential target for therapeutics in RAS-driven cancers. However, studies focusing on RAS dimerization have provided conflicting evidence about the proposed dimer interface. RAS dimers are not observed in solution, and cell-based experimental techniques have limited resolution and are subject to false positive results. In June 2022, we organized a debate on RAS dimerization and the role of the G-domain at the 30th Anniversary of the FASEB Summer Research Conference on the Structure, Function, and Regulation of Small GTPases in Vermont. The discussion continued in October 2022 at the Fourth RAS Initiative Symposium in Frederick, MD. RAS researchers generally agreed that clustering two or more RAS proteins do not result from a stable G-domain association but rather from interactions between RAS C-terminal membrane anchors and membrane phospholipids. This consensus supports the primary mechanism for RAS dimerization being lipid-mediated clustering of RAS proteins instead of G-domain-mediated interactions. Recently, we reviewed the evidence for RAS dimerization and summarized the discussions from these two meetings for the broader RAS community in a Perspective published in Molecular Cell.
RAS proteins have a highly conserved G-domain but differ in their C-terminal hypervariable regions (HVR), which undergo post-translational modifications to anchor RAS to the cell membrane for signaling. Studies on the spatial organization of RAS on the inner leaflet of the cell membrane have shown that RAS molecules exist as freely diffusing monomers, pairs, and short-lived transient nanoclusters containing 5-6 RAS molecules (1-4). The nanoclusters are essential for effector recruitment and activation, and different RAS proteins segregate into distinct, transient, lipid-dependent nanoclusters (4-6). Interestingly, it is the C-terminal anchor, not the G-domain, that drives the clustering and distribution of RAS molecules. This conclusion is based on observations that the same spatial distribution is seen even when the G-domain is absent or replaced with other proteins (6-9). For instance, mCherry-tagged KRAS4B and mCherry extended with the HVR of KRAS4B exhibit the same clustering distribution in single molecule tracking studies2,10.
The spatial organization of RAS is determined by the C-terminal anchor's interaction with phospholipids in the plasma membrane (PM), leading to lipid-mediated lateral segregation, with the G-domain playing no direct role in the process. The anchor recruits specific lipids, forming a lipid-protein complex that acts as a building block for higher-order complexes. Different RAS anchors recruit different lipids, resulting in lateral segregation of RAS isoforms unless the lipid composition of the PM is altered. Lipid de-mixing drives the formation of higher-order complexes, similar to lipid raft formation (11,12). KRAS remains monomeric on two-component supported lipid bilayers, but it colocalizes with anionic lipids in large, unilamellar vesicles and is observed in clusters on intact PM sheets (4-6,13,14).
Despite extensive evidence supporting the lipid-mediated clustering model, many studies have focused on characterizing RAS homodimers mediated by G-domain interactions (15-19). Different interaction interfaces on the G-domain have been proposed for RAS dimerization, including interfaces formed by α3-α4 helices, α4-α5 helices, and β1/β2 strands (20). However, molecular dynamic simulations found that RAS-RAS interactions are interfacially nonspecific and that all previously reported interfaces are part of a large ensemble of possible interactions (21). A recent study has shown no significant impact of α4-α5 amino acid substitutions on self-association, signaling, or transformation by oncogenic RAS (22). Additionally, studies on the activation cycle of RAF dimers have argued against the role of RAS dimers (23-25).
The terminology surrounding the proximity of the two RAS molecules at the PM is also a semantic problem. While cell biologists may refer to two proteins as forming a dimer if they are close together, biochemists define protein dimers as two protomers that interact via residues present on a specific interface with a dissociation constant in the nanomolar to micromolar range. It is important to use the term "dimer" only for protomers that conform to the biochemical definition. The current methods for detecting protein proximity in intact cells cannot distinguish biochemical dimers from other forms of proximity and often produce false-positive results. Therefore, rigorous controls are necessary, particularly when the proteins examined are overexpressed. For example, including other members of the RAS superfamily of small GTPases that localize on the PM can confirm RAS specificity.
Based on discussions at two meetings, there is widespread consensus that the proximity of two or more RAS proteins in cells or on isolated or artificial membranes is not due to the self-association of RAS protomers via their conserved G-domains but rather due to RAS-RAS protomer clustering as a result of the HVR membrane anchor and associated phospholipids bringing membrane-anchored RAS proteins into close proximity. This proximity likely occurs in nanoclusters where two or more RAS protomers are close enough to form lipid-based signaling-competent domains that do not require specific G-domain interactions. The overlapping low-affinity G-domain interaction surfaces allow for the non-random assembly of the G-domain "cargo" into HVR-driven clusters. Charge repulsion mutants and monobody binding can disrupt the lipid domain-driven assembly and impair RAS clustering and signal transmission. Mutations in putative dimeric interfaces can also affect anchor conformations, changes in RAS-lipid interactions, or interactions with effectors, GEFs, and/or GAPs, resulting in a reduction or loss of downstream signaling. Therefore, decreased signal output cannot be considered proof of impaired biochemical dimerization.
The rush to accept RAS G-domain-mediated dimerization as a required signaling component has led some RAS biologists to use this model to explain phenomena likely due to other processes. Any study invoking RAS isoform heterodimerization must consider the relative stoichiometry of RAS proteins, as it varies dramatically from tissue to tissue and from tumor to tumor (26). Therapeutic approaches that aim to interfere with GEF or effector interactions or disrupt nanoclustering, e.g., with monobodies and DARPins, hold more promise than targeting specific interactions at one of the proposed dimeric interfaces on the G-domain with small molecules (22,27-29).
Please click on the link below to read the Perspective published in Molecular Cell, which includes a comprehensive review of the evidence for RAS dimerization and a summary of the discussions from two recent meetings attended by RAS researchers.