Capturing RAS oligomerization on a membrane

Significance

RAS proteins regulate various intracellular signaling pathways, and mutations in these proteins are associated with several human cancers. A longstanding question is the existence of RAS oligomers, which are thought to play critical roles not only in the activation of downstream proteins but also suppression of mutant RAS-driven cancers. Despite conflicting reports on RAS assemblies, there is much excitement about disrupting RAS multimers, including membrane localization, as an alternative approach to target cancer. The discrepancies regarding RAS stoichiometries have been compounded by the use of indirect methods with varying spatial resolution. Using native mass spectrometry, we capture RAS dimerization on membranes that is directly influenced by lipid composition, posttranslational modifications, nucleotides, and small molecules.

Abstract

RAS GTPases associate with the biological membrane where they function as molecular switches to regulate cell growth. Recent studies indicate that RAS proteins oligomerize on membranes, and disrupting these assemblies represents an alternative therapeutic strategy. However, conflicting reports on RAS assemblies, ranging in size from dimers to nanoclusters, have brought to the fore key questions regarding the stoichiometry and parameters that influence oligomerization. Here, we probe three isoforms of RAS [Kirsten Rat Sarcoma viral oncogene (KRAS), Harvey Rat Sarcoma viral oncogene (HRAS), and Neuroblastoma oncogene (NRAS)] directly from membranes using mass spectrometry. We show that KRAS on membranes in the inactive state (GDP-bound) is monomeric but forms dimers in the active state (GTP-bound). We demonstrate that the small molecule BI2852 can induce dimerization of KRAS, whereas the binding of effector proteins disrupts dimerization. We also show that RAS dimerization is dependent on lipid composition and reveal that oligomerization of NRAS is regulated by palmitoylation. By monitoring the intrinsic GTPase activity of RAS, we capture the emergence of a dimer containing either mixed nucleotides or GDP on membranes. We find that the interaction of RAS with the catalytic domain of Son of Sevenless (SOS^cat) is influenced by membrane composition. We also capture the activation and monomer to dimer conversion of KRAS by SOS^cat. These results not only reveal the stoichiometry of RAS assemblies on membranes but also uncover the impact of critical factors on oligomerization, encompassing regulation by nucleotides, lipids, and palmitoylation.

The RAS isoforms, HRAS (Harvey Rat Sarcoma viral oncogene), KRAS (Kirsten Rat Sarcoma viral oncogene) splice variants KRAS4A and KRAS4B (hereafter referred to as KRAS), and NRAS (Neuroblastoma oncogene), are small GTPases (1–5). They function as molecular switches, toggling between active guanosine triphosphate(GTP)-bound and inactive guanosine diphosphate(GDP)-bound states to regulate cell proliferation and differentiation (6). RAS propagates signals through a variety of cellular pathways, such as the mitogen-activated protein kinase (MAPK) pathway where RAS activates the downstream effector Raf (6–8). RAS is regulated by guanine nucleotide exchange factors (GEFs) that activate RAS by facilitating the exchange of GDP with GTP, and GTPase-activating proteins (GAPs) that promote hydrolysis of GTP, returning RAS to the inactive GDP-bound state (9, 10). RAS is the most frequently mutated protein in cancer, representing about 20% of human cancers (11–13). Oncogenic RAS mutants display impaired biophysical properties, such as insensitivity to GAPs, that renders them constitutively active, contributing to tumor progression (9, 14).

Studies from many laboratories have suggested that RAS proteins oligomerize on biological membranes (Fig. 1A), but the oligomeric state remains a key question in RAS biology [for review (15–18)]. RAS association with the membrane is essential for biological function (19–21) and is mediated by posttranslational modification (PTM) of the unstructured C-terminal hypervariable region (HVR) (Fig. 1B) (22, 23). Despite the high sequence similarity of the RAS G-domain, the HVR of RAS isoforms share a common C-terminal S-farnesylcysteine carboxymethyl ester modification but differ in sequence and additional PTMs (Fig. 1B). These differences have been ascribed to the distinct lipid interactions and preferences of RAS (17), such as KRAS being selective for phosphatidylserine (PS) lipids (24). A battery of approaches have been used to show that RAS isoforms form dimers and other higher-order assemblies, commonly referred to as nanoclusters of approximately six RAS proteins, that are often dependent on lipids and the nucleotide-bound state of RAS (25–33). In support of RAS multimers is the fact that wild-type RAS isoforms can act as suppressors of mutant RAS-driven tumors (33, 34). Adding to the conundrum are several studies reporting the absence of RAS multimers, even on membranes (35–37). The underlying issues regarding the oligomeric state of RAS are compounded by the use of indirect methods with varying spatial resolution, studies in live cells compared to artificial membranes, and fusion to fluorescent reporters for visualization and immunolabeling, leading to an array of reported oligomeric states (15, 16). Nevertheless, disruption of RAS membrane association and multimerization, such as farnesyl transferase inhibitors and small antibody mimetics, represents an alternative and attractive therapeutic approach to target RAS-driven cancers (15, 16, 18, 38).

Fig. 1.

RAS posttranslational modifications and dimerization on membranes. (A) Schematic of KRAS in the inactive (GDP-bound, PDB 7C40), active (GTP-bound, PDB 7KFZ), and oligomerization of KRAS on the membrane surface. The C-terminal farnesylation of KRAS is denoted by an orange line. (B) Alignment of RAS C-terminal HVRs. Cysteine residues are palmitoylated (green) or farnesylated along with methylation of the C terminus (orange). (C) Mass spectrum of KRAS bound to GTP in DDM. (D) Illustration of the introduction and activation of proteoliposomes using native MS. (E) KRAS•GTP ejected from POPC liposomes. The asterisk denotes signal for a lipid cluster corresponding to four POPC molecules. (F) KRAS•GDP ejected from POPC liposomes. (G) BI2852 induces dimerization of KRAS•GDP on POPC liposomes. (H and I) NS1 and RBDCRD disrupt KRAS•GTP dimerization and eject as a 1:1 complex with KRAS. (J) KRAS interaction assay using the NanoBiT reporter. BI2825, NS1, RBDCRD, and SOS^CAT were added to KRAS•GTP proteoliposomes. Reported are the mean and SD (n = 3).

Herein, we use mass spectrometry to interrogate RAS assemblies on membranes of defined composition. We show that RAS can form monomers and dimers on membranes, and these assemblies are dependent on nucleotide, lipid composition, and posttranslational modifications. In combination with a protein interaction assay, we demonstrate that the small molecule BI2852 promotes dimerization of KRAS bound to GDP and GTP. Moreover, the binding of an engineered monobody and RAS-binding domain and cysteine-rich domain of RAF1 disrupts KRAS dimers. Furthermore, we monitor in real time the intrinsic GTPase activity and activation of RAS by the catalytic domain of Son of Sevenless (SOS^cat), revealing the presence of dimers containing GDP or mixed nucleotides as well as monomer to dimer conversion on membranes.

Results and Discussion

Native Mass Spectrometry Captures KRAS Oligomerization on Membranes.

To characterize KRAS on membranes using native mass spectrometry, we expressed and purified farnesylated and methylated KRAS from insect cells (39). After optimizing the expression construct, KRAS was extracted from purified membranes using n-Dodecyl-β-D-maltopyranoside (DDM), a nonionic detergent. The mass spectrum of the purified protein in DDM indicated that KRAS was fully modified, monomeric, and bound to GTP (Fig. 1C and SI Appendix, Tables S1–S3). We next reconstituted KRAS with liposomes composed of 1-palmitoyl-2-oleoyl-glycero-3-phosphocholine (POPC), a lipid abundant in cellular membranes (40). For native mass spectrometry analysis, proteoliposomes are introduced into the mass spectrometer using nanoelectrospray ionization, and membrane proteins are ejected from the liposome by the application of collisional energy (Fig. 1D and SI Appendix, Table S1) (41). Proteoliposomes containing KRAS loaded with GTP (KRAS•GTP) ejected from proteoliposomes as a distribution of monomer and dimers (Fig. 1E). The hump centered around ~6,200 m/z corresponds to signal for activated liposomes, consistent with our previous study of proteoliposomes containing different membrane proteins (41), and in this congested region no signal for higher-order KRAS oligomers was found (SI Appendix, Fig. S1). Under these conditions, KRAS•GTP ejects from membranes as monomers and dimer, not larger oligomers.

As KRAS•GTP is monomeric and dimeric on a membrane, we examined the impact of the bound nucleotide and other effectors on the distribution of stoichiometries. KRAS in the inactive state, loaded with GDP (KRAS•GDP), ejected from liposomes as monomers (Fig. 1F and SI Appendix, Fig. S2), indicating a nucleotide dependence on dimerization, a result consistent with earlier reports (25–33). The addition of BI2852 (42), a small molecule that promotes dimerization of KRAS (43, 44), to proteoliposomes containing KRAS•GDP or KRAS•GTP resulted in the ejection of only dimers, no signal for monomeric KRAS was present (Fig. 1G and SI Appendix, Fig. S3A). Structural studies suggest BI2852 promotes dimerization through binding at the interface of a KRAS dimer (43). However, the mass for the ejected KRAS dimers reveals that BI2852 is not bound. The promotion of a KRAS dimer is not due to the solvent, DMSO in this case, but the presence of B12852. The lower charge state of the KRAS dimer compared to the control suggests the drug may dissociate from the dimer during desolvation and ejection of KRAS from proteoliposomes (SI Appendix, Fig. S3). In the presence of NS1, an engineered antibody mimetic reported to disrupt RAS oligomers (45), KRAS•GTP ejected from proteoliposomes with NS1, forming a 1:1 complex (Fig. 1H). No signal for monomeric and dimeric KRAS was observed. The RAS-binding domain and cysteine-rich domain of RAF1 (residues 52–186, RBDCRD) (46) also disrupted KRAS dimers, ejecting as a 1:1 complex with KRAS (Fig. 1I). In summary, native MS captures distinct stoichiometries of KRAS that are influenced by the nucleotide-bound state and binding of effectors.

To corroborate our native MS findings, we developed a KRAS interaction assay using the protein-fragment complementation reporter NanoBiT (Fig. 1J) (47). A mixture of KRAS genetically fused to either the large BiT (17.6 kDa subunit) or small BiT (1.3 kDa peptide), loaded with GTP and solubilized in DDM, showed no luminescence, indicating no association of the KRAS fusion proteins (SI Appendix, Fig. S4A). The same mixture reconstituted with POPC liposomes resulted in robust luminescence, signifying the presence of KRAS oligomers (Fig. 1J). In contrast, a significant decrease in luminescence was observed when the KRAS fusions are bound to GDP. For proteoliposomes containing KRAS•GTP fusions, the addition of BI2852 enhanced signal, whereas NS1 and RBDCRD decreased signal. No significant change in luminescence was observed in the presence of SOS^cat, a GEF (SI Appendix, Fig. S4B) (48, 49). While results from the KRAS interaction assay agree with those obtained using native MS, the assay requires bulky fusion proteins, which may perturb RAS oligomerization, and most importantly, fails to report on stoichiometry.

Impact of Lipid Composition on KRAS Dimerization.

As KRAS oligomerization has been reported to be influenced by certain lipids (17), we performed similar experiments for KRAS•GTP reconstituted with different liposome formulations (Fig. 2). For these studies, we used liposomes containing POPC doped with a percentage of another lipid. The choice of the percentage was based on abundance of the particular lipid in cellular membranes (40). KRAS•GTP associated with POPC liposomes containing 5% 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphate (POPA) ejected as a near equal mixture of monomers and dimers (Fig. 2A). Proteoliposomes containing 10% 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine (POPE) or 2.5% 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-(1′-rac-glycerol) (POPG) shifted the equilibrium toward monomeric KRAS-GTP (Fig. 2 B and C). Interestingly, liposomes containing 20% 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-L-serine (POPS), a lipid reported to be associated with KRAS nanoclusters (24), promoted dimerization of KRAS•GTP. These findings demonstrate that dimerization of KRAS•GTP is dependent on lipid composition and, of the lipids investigated, PS displayed the most significant impact on dimer formation.

Fig. 2.

Lipids influence KRAS dimerization. (A–D) Mass spectrum of KRAS•GTP ejected from POPC liposomes containing (A) 5% POPA, (B) 10% POPE, (C) 2.5% POPG, or (D) 20% POPS. (E) Mole fraction of monomeric and dimeric KRAS•GTP determined for different liposome compositions. Reported are the mean and SD (n = 3).

Modifications of NRAS Influence Dimerization.

We then shifted our focus to other RAS isoforms (HRAS and NRAS) that, unlike KRAS, contain palmitoylation modification(s). HRAS purified from insect cells was heterogenous, containing mixed PTMs or truncated at the C-terminus, making it difficult to assign mass spectra of the ejected complexes (SI Appendix, Fig. S5 and Supporting Text). On the other hand, purified NRAS contained the C-terminal farnesylation carboxymethyl ester modification but varied in the presence or absence of palmitoylation at C181 (Fig. 3A and SI Appendix, Table S2). Most interestingly, NRAS loaded with GTP ejected from POPC liposomes revealed that the fully modified NRAS was monomeric, whereas the form missing the palmitoylation (NRAS^-palm) was dimeric (Fig. 3B). NRAS•GTP reconstituted with liposomes containing mixed lipids, except for POPA, showed the same distribution as that observed for POPC liposomes (SI Appendix, Fig. S6). In the case of liposomes containing 5% POPA, the fully modified NRAS was monomeric but also formed a dimer (Fig. 3 C and D). Again, only a dimer was observed for NRAS^-palm. NRAS loaded with GDP was observed as both monomers and dimers, but the NRAS^-palm was only found as dimers (Fig. 3C). In the case of the other liposome compositions, the dimer of NRAS^-palm bound to GDP persisted (SI Appendix, Fig. S6). Furthermore, addition of RAF1 RBDCRD disrupted NRAS dimers, ejecting as a 1:1 complex with NRAS (Fig. 3E). These results underscore the importance of PTMs on NRAS, particularly the unexpected role of palmitoylation in regulating NRAS dimerization.

Fig. 3.

Dimerization of NRAS is influenced by palmitoylation and lipids. (A) Deconvoluted mass spectrum shows that a mixture of NRAS contains the C-terminal farnesylcysteine but varies in palmitoylation at C181. Posttranslationally modified lipid anchors are denoted as orange (farnesylation) and green (palmitoylation) lines. (B) Mass spectrum of NRAS•GTP ejected from POPC liposomes. NRAS ejected from liposomes containing 5% POPA bound to (C) GDP and (D) GTP. Deconvolution of mass spectra is shown to the right. (E) RAF1 RBDCRD disrupts NRAS•GTP dimerization and ejects as a 1:1 complex with NRAS.

Monitoring RAS GTPase Activity on Membranes.

To determine how the intrinsic GTPase activity of RAS influences dimerization, we measured the mass of the monomeric and dimeric complexes in real time using mass spectrometry (Fig. 4). Hydrolysis of GTP is monitored by a mass shift due to the loss of inorganic phosphate (~90 Da) but GDP remains bound to the enzyme (50). At the initial time point, KRAS•GTP on POPC liposomes populates a distribution of monomers and dimers. Deconvolution of the mass spectrum, removing the charge component to obtain molecular masses (51), reveals that GTP-bound to KRAS is hydrolyzed over time to GDP (Fig. 4A). Interestingly, careful examination of the data reveals that the KRAS•GTP dimer sequentially hydrolyzes GTP over time (Fig. 4B). More specifically, KRAS dimers were observed containing KRAS•GDP and KRAS•GTP, which eventually hydrolyzed GTP to form a dimer consisting of two KRAS•GDPs. The KRAS•GDP dimer remained throughout the duration of the experiment. This result is most interesting as only monomers were observed for KRAS loaded with GDP on membranes (Fig. 1F). Given the results for KRAS, we next performed similar experiments for NRAS. Like KRAS, a dimer of NRAS•GTP, in which NRAS is not palmitoylated, is present at the start of the reaction (Fig. 4C). Over time, two populations of dimeric NRAS appear with near equal abundance. One contains mixed nucleotides, GTP and GDP, and the other containing two GDPs (Fig. 4C). A dimer of NRAS•GDP remained at longer incubation times. By monitoring mass in real time, mass spectrometry reveals that a dimer of KRAS and NRAS, initially formed while bound to GTP, on membranes can persist while bound to mixed nucleotides and even bound to GDP.

Fig. 4.

Monitoring RAS GTPase activity on a membrane. (A) Mass spectra of KRAS•GTP reconstituted with POPC liposomes recorded at different time points. Shown from top to bottom are time points at 0, 24, and 48 h, respectively. (B) Deconvolution of mass spectra shown in panel A. Monomeric KRAS•GDP appears over time as GTP is hydrolyzed. Dimeric KRAS•GTP undergoes sequential hydrolysis of GTP, forming dimeric complexes containing mixed nucleotides and only GDP. (C) Deconvoluted mass spectra of NRAS•GTP acquired at different times. Like KRAS, sequential hydrolysis of GTP in the dimeric NRAS•GTP complex is observed.

Assemblies of SOS^cat with RAS on Membranes.

RAS interacts with SOS, a guanosine nucleotide exchange factor that facilitates reloading of RAS-GDP with GTP, in a mutant-dependent manner (52, 53) and disrupting this protein–protein interaction is emerging as an attractive therapeutic strategy (54–56). Therefore, we investigated how RAS associated with membranes affects its interaction with the catalytic domain of SOS1 (SOS^cat). SOS^cat representing the minimal functional unit (49, 57) can bind two RAS molecules: one at the active site where RAS is reloaded with GTP; and the other at the allosteric site, in which binding of RAS•GTP enhances the nucleotide exchange rate. The addition of SOS^cat to KRAS•GTP associated with POPC liposomes resulted in a binary complex, KRAS•GTP bound to SOS^cat, and a ternary complex, SOS^cat bound to KRAS•GTP and KRAS (Fig. 5A). The abundances of these complexes was similar to that observed for the truncated form of KRAS in the absence of a membrane (52). KRAS•GTP and SOS^cat ejected from liposomes containing 20% POPS were strikingly similar to results for POPC liposomes (SI Appendix, Fig. S7A). KRAS•GTP and SOS^cat ejected from liposomes containing POPG showed that the abundance of the binary complex was significantly decreased and the ternary complex was absent (Fig. 5B). The same mixture but with liposomes containing 5% POPA or 10% POPE showed no complexes of KRAS•GTP with SOS^cat (Fig. 5C and SI Appendix, Fig. S7B). KRAS•GDP associated with POPC liposomes was monomeric and upon incubation with SOS^cat and GTP promoted dimerization of KRAS (Fig. 5D).

Fig. 5.

KRAS and SOS^cat molecular assemblies and activation on membranes. (A–C) Mass spectra of GFP fusion KRAS bound to GTP complexed with SOS^cat in 3:1 molar ratio of RAS:SOS^cat ejected from (A) 100% POPC, (B) 2.5% POPG, and (C) 5% POPA liposomes. (D) GFP fusion KRAS bound to GDP ejected from POPC liposome before (Left) and after (Right) incubation with SOS^cat and GTP in 10:1 molar ratio of RAS:SOS^cat. (E) Representative fluorescence readout of SOS^cat-mediated nucleotide exchange on KRAS bound to mant-GDP and associated with liposomes containing, 2.5% POPG (pink), 10% POPE (teal), 5% POPA (turquoise), and 20% POPS (magenta) acquired at room temperature (Left). The plot of the nucleotide exchange rate is shown (Right). Reported are the mean and SD (n = 3).

To assess the lipid dependence in nucleotide exchange of KRAS mediated by SOS^cat, liposomes varying in lipid compositions were prepared and reconstituted with KRAS bound to GDP modified with a fluorophore (mant-GDP) to monitor nucleotide exchange at room temperature. The slowest SOS^cat-mediated nucleotide exchange rate was observed for KRAS on POPE and POPG containing liposomes (Fig. 5E and SI Appendix, Fig. S8). In contrast, the fastest nucleotide exchange rates were observed for KRAS associated with liposomes containing POPS and POPA (Fig. 5E and SI Appendix, Fig. S8). Interestingly, KRAS on POPS and POPA liposomes showed the most drastic differences in the complexes formed with SOS^cat (Figs. 2E and 5 and SI Appendix, Fig. S7). However, KRAS on POPS and POPA liposomes formed dimers. These results show that the lipid-dependent dimerization of KRAS on membranes enhances their activation by SOS^cat.

Together, our native mass spectrometry and biochemical data reveal an equilibrium of RAS monomers and dimers on membranes that is dependent on the bound nucleotide, lipid composition, and PTMs. We demonstrate that GTP-bound KRAS dimerizes on the lipid bilayer and the binding of the effector protein, RBDCRD dissociates dimerization by forming a 1:1 complex. Surprisingly, dimerization of NRAS is dependent on PTMs, which oligomerization is enhanced in the presence of POPA but independent of the nucleotide-bound state. Native MS captures distinct stoichiometries of RAS dimers bound to a mixture of GTP and GDP. This is a result of the sequential hydrolysis of the bound GTP to GDP. Liposome composition plays not only a significant role in KRAS oligomerization but also the complexes formed with SOS^cat, impacting the rate of nucleotide exchange. While some reports have suggested RAS forms larger assemblies on membranes (25–33), we find KRAS and NRAS populate a distribution of monomer and dimers that is influenced by several factors. However, there remains a possibility that larger RAS oligomers are dissociated upon the application of collision energy required to eject RAS from the membrane. Collectively, we capture RAS dimerization on a lipid bilayer, uncovering the role of nucleotides, lipids, and palmitoylation that govern oligomerization.

Materials and Methods

Full-Length RAS Expression and Purification.

The genes encoding HRAS (residues 1–189, UniProt P01112) and NRAS (residues 1–189, UniProt 90111) were subcloned from the RAS mutant clone collection kit (Addgene Cat. 1000000089) into a pACEBac1 (Geneva Biotech) with a N-terminal His₆-tag followed by TEV protease cleavable GFP or mCherry. KRAS4B (KRAS) gene (residues 1–188, UniProt P01116-2) was obtained from Frederick National Laboratory for Cancer Research, MD, as a part of the expression clone R989-X05-636 containing a construct: His₆-MBP-TEV-G-Hs.KRAS4b(1–188). The KRAS gene was subcloned into a pACEBac1 (Geneva Biotech) with a N-terminal His₆-tag followed by TEV protease cleavable GFP or mCherry. The plasmids were transformed into DE100 cells obtained from Frederick National Laboratory for Cancer Research, MD, USA. The recombinant bacmid was purified and used to transfect Spodoptera frugiperda (Sf9) cells using PEI Max (Polysciences) transfection reagent to generate P1 virus (58). Ras proteins were expressed in Trichoplusia ni (Tni) cells with the addition of baculovirus for 72 h at 27 °C while shaking. Cells were harvested by centrifugation at 4,000×g for 10 min and stored at −80 °C. Cell pellets were resuspended in lysis buffer (300 mM NaCl, 20 mM HEPES, and 5 mM MgCl₂, pH 7.4, at room temperature) and lysed using a Microfluidics M-110P microfluidizer operating at 25,000 psi. The cell debris was removed by centrifugation at 20,000×g for 25 min at 4 °C. Membranes were pelleted through ultracentrifugation at 100,000×g for 2 h at 4 °C. The pelleted membranes were resuspended in resuspension buffer (300 mM NaCl, 20 mM HEPES, and 5 mM MgCl₂, pH 7.4, at room temperature), and at this point, resuspended membrane could be stored at −80 °C for future use. For extraction of RAS protein, 1% (w/v) n-dodecyl-beta-maltoside (DDM) was added to the resuspended membrane for 2 h. The extracted proteins were centrifugated at 20,000×g at 4 °C for 10 min, and the supernatant was filtered with a 0.45-μm syringe filter (Pall Corp.). The cleared extract was loaded onto a 2-mL bed volume of Ni-NTA Superflow resin (Qiagen) gravity-flow column pre-equilibrated in NHA-DDM (300 mM NaCl, 20 mM HEPES, 20 mM imidazole, 5 mM MgCl₂, and 0.025% DDM, pH 7.4, at room temperature), washed with NHA-DDM, and the RAS protein was then eluted with NHB-DDM (NHA-DDM with 500 mM imidazole). The eluent is then buffer exchanged using a HiPrep 26/10 desalting column (GE Healthcare) equilibrated into storage buffer (300 mM NaCl, 20 mM HEPES, 5 mM MgCl₂, 10% w/v glycerol, and 0.025% DDM, pH 7.4, at room temperature). The buffer exchanged protein was then digested with TEV protease at 4 °C and incubated overnight to remove the N-terminal affinity tag. The proteins were passed over a 2-mL bed volume of Ni-NTA Superflow resin (Qiagen) gravity-flow column pre-equilibrated in storage buffer, and flow-through containing tag-less protein was collected. Glycerol was added to reach 20% w/v, and the protein was concentrated using a 10K MWCO concentrator (MilliporeSigma), aliquoted, flash-frozen in liquid nitrogen, and stored at −80 °C for future use.

Small BiT and Large BiT KRAS Expression and Purification.

For small BiT (SmBiT) RAS fusion, the gene encoding KRAS (residues 1–188, UniProt P01116-2) was subcloned into a pACEBac1 (Geneva Biotech) with the following construct: His₆-SSG-HRV3C-GG-VTGYRLFEEIL-GSGGGSGT-TEV-S-KRAS, where VTGYRLFEEIL is the amino acid sequence for the SmBiT. For large BiT (LgBiT) Ras fusion, the gene encoding KRAS (residues 1–188, UniProt P01116-2) was subcloned into a pACEBac1 (Geneva Biotech) with the following construct: His₆-SSG-HRV3C-GG-LgBiT-GSGGGSGT-TEV-S-KRAS. Expression and purification of the small and large BiT KRAS were performed as described above with minor modifications. Upon eluting from Ni-NTA Superflow resin (Qiagen) gravity-flow column, proteins were desalted using a HiPrep 26/10 desalting column (GE Healthcare) into storage buffer. Glycerol was added to reach 20% w/v, and the protein was concentrated using a 10K MWCO concentrator (MilliporeSigma), aliquoted, flash-frozen in liquid nitrogen, and stored at −80 °C for future use.

SOS^cat Expression and Purification.

SOS1^cat (residues 558–1049) plasmid (a kind gift from Prof. John Kuriyan at the University of California Berkeley) was used to express in Escherichia coli Rosetta 2(DE3) cells (Novagen) as previously described (52, 57, 59). Cells containing the expression plasmid were grown in LB medium supplemented with either 100 μg/mL ampicillin or 34 μg/mL chloramphenicol. Protein expression was induced with 500 µM IPTG once the cells reached an OD₆₀₀ of 0.6 to 0.8 and grown at 18 °C overnight. Cells were harvested by centrifugation (5,000×g), and pellets were stored at −80 °C. Cell pellets were thawed on ice and resuspended in buffer A (300 mM NaCl and 20 mM Tris, pH 7.4) supplemented with 1 mM MgCl₂ and 5 mM β-mercaptoethanol (β-ME). The cell suspension was lysed by a microfluidizer (Microfluidics M-110P) operating at 20,000 psi. The lysate was clarified by centrifugation at 40,000×g for 20 min. All purification steps were performed at 4 °C. The clarified lysate was loaded onto a 5 mL HisTrap HP column (GE Healthcare) pre-equilibrated in buffer A containing 20 mM imidazole. The protein was eluted with buffer A containing 500 mM imidazole. The fractions containing protein were pooled and desalted using a HiPrep 26/10 desalting column (GE Healthcare) equilibrated in buffer A. The desalted protein was then digested with TEV protease at 4 °C and incubated overnight to remove the N-terminal affinity tag. The proteins were passed over a 5 mL HisTrap HP column equilibrated with buffer A, and flow-through containing tag-less protein was collected. The protein collected in the flow-through was concentrated and subjected to size exclusion chromatography using a Highload 16/600 Superdex 75 pg column (GE Healthcare) equilibrated in buffer C (150 mM NaCl, 5 mM β-ME, 10% w/v glycerol, 0.5 mM MgCl₂, and 20 mM Tris, pH 7.4). Peak fractions containing the target protein were pooled, and glycerol was added to a final concentration of 20% w/v, concentrated using a centrifugal concentrator [Millipore, 30K molecular weight cutoff (MWCO)], flash-frozen using liquid nitrogen, and stored at −80 °C.

RBDCRD Expression and Purification.

The gene encoding RAF1 (52–186) was subcloned into a pET28 vector with the following construct: His₆-MBP-TEV-RAF1(52-186). E. coli Rosetta 2(DE3) cells were used for protein expression as previously described (60). Cells were grown in LB medium containing 50 μg/mL kanamycin and 300 µM ZnCl₂. Then, 0.5 mM IPTG was added to the cells to induce expression when the cells reached an OD₆₀₀ of 0.6 to 0.8 and grown at 18 °C overnight. Cells were harvested by centrifugation (5,000×g), and pellets were stored at −80 °C. Cell pellets were thawed on ice and resuspended in buffer A (20 mM HEPES, 500 mM NaCl, and 10% (w/w) glycerol, pH 7.3) supplemented with 5 mM TCEP. The cell suspension was passed through a microfluidizer (Microfluidics M-110P) operating at 20,000 psi. Insoluble material was removed by centrifugation at 40,000×g for 20 min. The sample was loaded onto a 5 mL HisTrap HP column (GE Healthcare) equilibrated in buffer A containing 35 mM imidazole. Buffer A containing 500 mM imidazole (buffer B) was used to elute the protein. Peak fractions were pooled, desalted using a HiPrep 26/10 desalting column (GE Healthcare) equilibrated in buffer A, and digested with TEV protease at 4 °C overnight. The sample was loaded to a 5 mL HisTrap HP column equilibrated with buffer A. The column was washed to baseline with buffer A, then with a 5 CV gradient from 5 to 10% buffer B, followed by 3 CV gradient of 10 to 50% buffer B, and 2 CV of 100% buffer B. The tag-less protein was collected in the shallow gradient (5 to 10% buffer B). Concentrated protein was injected onto a Highload 16/600 Superdex 75 pg column (GE Healthcare) equilibrated in buffer A. Glycerol was added to the pooled fraction to a final concentration of 20% w/v prior to concentrating using a centrifugal concentrator (Millipore, 10 K MWCO). The concentrated protein was flash-frozen and stored at −80 °C.

NS1 Monobody Expression and Purification.

The amino acid sequence encoding NS1 monobody (45) was codon optimized for E. coli, synthesized as gBlock gene fragments (IDT), and cloned into a modified pCDF-1b (Novagen) expression construct. The resulting constructs were DNA sequence confirmed. The protein was expressed with an N-terminal TEV protease cleavable 6xHis tag in Lemo21-(DE3) E. coli (New England Biolabs). Cells containing the expression plasmid were grown in LB medium and induced with 500 µM IPTG once the cells reached an OD₆₀₀ of 0.6 to 0.8 and grown overnight at 18 °C. Cells were centrifuged (5,000×g), and pellets were stored at −80 °C. Thawed cell pellets were resuspended in buffer A (300 mM NaCl and 20 mM Tris, pH 7.4) containing 1 mM MgCl₂ and 5 mM β-mercaptoethanol (β-ME). The cells were disrupted using a microfluidizer (Microfluidics M-110P) operating at 20,000 psi. Insoluble material was removed by centrifugation at 40,000×g for 20 min. The lysate was loaded onto a 5 mL HisTrap HP column (GE Healthcare) equilibrated in buffer A containing 20 mM imidazole. Buffer A containing 500 mM imidazole was used to elute the protein. Peak fractions containing protein were pooled and desalted using a HiPrep 26/10 desalting column (GE Healthcare) equilibrated in buffer A. The purified protein was then digested with TEV protease to remove the N-terminal affinity tag and incubated overnight at 4 °C. The digested protein was loaded onto a 5 mL HisTrap HP column equilibrated with buffer A, flow-through collected and concentrated, injected onto a size exclusion chromatography equilibrated in buffer C (150 mM NaCl, 5 mM β-ME, 10% w/v glycerol, 0.5 mM MgCl₂, and 20 mM Tris, pH 7.4). Glycerol was added to a final concentration of 20% w/prior to concentrating the peak fractions using a centrifugal concentrator (Millipore, 10 K MWCO). The concentrated protein was flash-frozen using liquid nitrogen and stored at −80 °C.

Preparation of RAS Containing Proteoliposomes.

Previously reported proteoliposome preparation for native MS was implemented with some modifications (41). In brief, lipid stocks in chloroform were dried under nitrogen overnight in a vacuum chamber to form lipid films. Lipid films were then rehydrated using lysis buffer (300 mM NaCl, 20 mM HEPES, and 5 mM MgCl₂, pH 7.4, at room temperature) to 10 mM final concentration. This lipid stock was then extruded using a 50 nm membrane filter (Cytiva) with multiple passages, or until the solution turned translucent, to form large unilamellar vesicles (LUVs). To solubilize the LUVs, 2× critical micelle concentration (CMC) of detergent was added (0.025% DDM) and the lipid-detergent mixture was equilibrated at 4 °C while rotating for 1 h or until the mixture became transparent/translucent. RAS protein was added to the solubilized lipids at a protein to lipid molar ratio of 1:1,000. The liposome-protein mixture was diluted with buffer to reach a final lipid concentration of 5 mM and was equilibrated for 1 to 2 h at 4 °C while rotating. Detergents were removed by the addition of Bio-beads (Bio-Rad) with 1 h incubation at 4 °C while rotating. The proteoliposome mixture was pipetted out, avoiding Bio-beads, and was extruded using a 50 nm pore size membrane (Cytiva) to generate unilamellar proteoliposomes. To remove residual detergent and salt, proteoliposomes were dialyzed using a 10 kDa MWCO dialysis membrane overnight against 1 L of 200 mM ammonium acetate, pH 7, at 4 °C. RAS proteins can be preloaded using previously established methods (50) with intended nucleotide prior to proteoliposome preparation. The nucleotide of choice (GDP, GTP, or mant-GDP) was added during the incubation time and during dialysis to ensure the full loading of the intended nucleotide.

Native Mass Spectrometry.

Proteins for native MS study were prepared as previously reported (61). In brief, proteins were buffer exchanged into 200 mM ammonium acetate (2xCMC DDM was added) with a centrifugal desalting column (Micro Bio-Spin 6, Bio-Rad). Native MS data were collected on a Thermo Exactive Plus with an Extended Mass Range Orbitrap mass spectrometer (Thermo Scientific). The tuning parameters are listed in SI Appendix, Table S1http://www.pnas.org/lookup/doi/10.1073/pnas.2405986121#supplementary-materials. Proteoliposome samples dialyzed against 200 mM ammonium acetate were collected on the same mass spectrometer with the tuning parameters listed in SI Appendix, Table S1http://www.pnas.org/lookup/doi/10.1073/pnas.2405986121#supplementary-materials. For the analysis of various effectors and ligands, 25 µM BI2852 (5:1 molar ratio of drug:protein) (TargetMol), 10 µM NS1 monobody (2:1 molar ratio of monobody:protein), and 5 µM RBDCRD (1:1 molar ratio of RBDCRD:protein) were added, respectively, and incubated for 10 min in ice prior to analysis. BI2852 was dissolved in DMSO, and the final concentration of the solvent in the sample contained 1% DMSO. For proteoliposomes containing GFP-fusion, 1% m-NBA (3-Nitrobenzyl alcohol, CAS no.:619-25-0, Sigma-Aldrich), a supercharging agent, was mixed with proteoliposome prior to introduction into the mass spectrometer. The addition of a supercharging agent has been shown to aid the ejection of proteins from nanodiscs and liposomes (41, 62, 63). A total of three replicate samples were analyzed from different preparations of protein. Native MS data were deconvoluted using UniDec (51).

NanoBiT Assay.

For the luminescence-based RAS oligomerization assay, the NanoBiT approach was used (47). Both Sm and LgBiT KRAS were preloaded with intended nucleotide prior to incorporation onto liposomes. Sm and LgBiT KRAS proteins were added to the solubilized lipids at a protein to lipid molar ratio of 1:1:2,000, where the total combined concentration of Sm and LgBiT KRAS is 5 µM. Proteoliposomes were dialyzed using a 10 kDa MWCO dialysis membrane overnight against 1L of 200 mM ammonium acetate, pH 7, at 4 °C. To measure the luminescence, KRAS liposomes were diluted to a final concentration of 0.25 µM in the presence of 5 µM of coelenterazine 400a (Nanolight Technologies) and incubated for 1 min prior to data acquisition. Luminescence signal intensity at 470 to 20 nm was monitored. For the analysis of various effectors and ligands, 2.5 µM BI2852 (TargetMol), 5 µM NS1 monobody, 5 µM RBDCRD, and 1 µM SOS^cat were added, respectively. A total of three replicate samples were analyzed.

Nucleotide Exchange Assay

The rate of nucleotide dissociation was determined using KRAS loaded with mant-GDP incorporated onto liposomes using Clariostar BMG Labtech. KRAS was preloaded with mant-GDP, and >95% loading was confirmed by native MS prior to incorporation onto liposomes. Proteoliposome samples were prepared following the procedure above. To initiate the exchange, proteoliposomes containing 500 nM KRAS (mant-GDP) were mixed with 100 nM SOS^cat, 10 µM GTP, and 100 µM MgCl₂. Fluorescence was monitored at 430 nm with a 20 nm window (excitation at 370 to 15 nm) over time for 360 s. For KRAS in DDM, blank was substituted to 200 mM ammonium acetate with 0.025% w/v DDM. Total three technical replicate samples were analyzed.

Native Top–Down Proteomics.

Top–down fragmentation on purified HRAS was performed on an ultrahigh mass range orbitrap mass spectrometer (ThermoFisher Scientific, San Jose, CA) by isolating the 2,669 m/z charge state at 100,000 resolution and collisionally activating the ions with 160 to 180 eV of collision energy with a spray voltage of 1.5 kV, capillary temperature of 300 °C, DC source offset voltage of 21.0, desolvation voltage of −25.0, and trapping gas pressure of 2.0. The spectra were deconvoluted with MASH Native using the eTHRASH algorithm with a signal to noise value of 5 and a min fit parameter of 75 (64). Fragment ions were assigned with ProSight Lite v1.4 with a mass error of 5 ppm (65).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.