Structural insights into the complex of oncogenic KRas4BG12V and Rgl2, a RalA/B activator

Abstract

About a quarter of total human cancers carry mutations in Ras isoforms. Accumulating evidence suggests that small GTPases, RalA, and RalB, and their activators, Ral guanine nucleotide exchange factors (RalGEFs), play an essential role in oncogenic Ras-induced signalling. We studied the interaction between human KRas4B and the Ras association (RA) domain of Rgl2 (Rgl2^RA), one of the RA-containing RalGEFs. We show that the G12V oncogenic KRas4B mutation changes the interaction kinetics with Rgl2^RA The crystal structure of the KRas4B^G12V: Rgl2^RA complex shows a 2:2 heterotetramer where the switch I and switch II regions of each KRas^G12V interact with both Rgl2^RA molecules. This structural arrangement is highly similar to the HRas^E31K:RALGDS^RA crystal structure and is distinct from the well-characterised Ras:Raf complex. Interestingly, the G12V mutation was found at the dimer interface of KRas4B^G12V with its partner. Our study reveals a potentially distinct mode of Ras:effector complex formation by RalGEFs and offers a possible mechanistic explanation for how the oncogenic KRas4B^G12V hyperactivates the RalA/B pathway.

Figure S1.. Rgl2 Ras-binding domain interacts with active KRas4B^G12V.

(A) A schematic diagram of the domain organisation of human Rgl2 as defined in the UniProt database (UniProt number: O15211). A domain spanning amino acid residues 648–735 is annotated as “Ras-associating” by PROSITE annotation rule PRU00166. In this work, we call this domain Rgl2^RA. (B) Rgl2^RA interacts with active KRas4B^G12V. Bacteria recombinant Rgl2^RA fragment spanning the amino acid residues 643–740 of Rgl2 was fused with GST, and the fusion protein was fixed on glutathione beads. Recombinant KRas4B^G12V 1–169, loaded with either GDP or non-hydrolysable GTP analogue, GMPPNP, was applied on the beads to examine the protein–protein interaction. GST–Rgl2^RA interacted only with the GMPPNP-loaded KRas4B^G12V. (C) KRas4B^G12V:Rgl2^RA complex was purified using size exclusion chromatography. The fractions were analysed by SDS–PAGE gels (lower panel) according to the elution profile (upper panel). Fractions indicated by the green double-arrow line were used for crystallisation trials. (D, E) The size exclusion chromatography elution profiles of the KRas4B–Rgl2^RA complex. Rgl2^RA and KRas4B^G12V (D) or KRas4B^WT (E) were mixed at the 3:1 M ratio and applied to Superdex 75 10/300 Gl. The elution profiles (upper panels) show little difference between the two samples. Peak fractions were analysed by SDS–PAGE gel (lower panel).

Figure 1.. KRas4B^G12V exhibits a different binding kinetics towards Rgl2^RA than KRas4B^WT and is structurally more stable.

(A) KRas4B^WT and KRas4B^G12V samples used for biolayer interferometry (BLI) were confirmed to be loaded with GTP. The nucleotide-binding status of KRas4B^WT and KRas4B^G12V were examined by denaturating the proteins and detecting the released nucleotides by anion exchange chromatography. Samples of pure GDP or GTP were used as references. (B) KRas4B^WT and KRas4B^G12V show different binding kinetics towards Rgl2^RA. BLI was used to measure the binding kinetics of KRas4B (analyte) to immobilized GST-Rgl2^RA (ligand). GST-Rgl2^RA was immobilized on the bio-sensors, and varying concentrations of free KRas4B^WT (left panel) and KRas4B^G12V (right panel) were provided, and the interactions were measured at 20°C. The 1:1 binding model was fitted to the data using Octet Analysis Studio 13.0 (Sartorius). The fitted curves are shown in red. The deduced K_D values of these two cases are similar: ∼1.48 μM (WT) and 1.49 μM (G12V). Meanwhile, the kon values are ∼137 (M⁻¹ms⁻¹) (WT) and 238 (M⁻¹ms⁻¹) (G12V), and koff values are ∼0.20 (s⁻¹) (WT) and 0.36 (s⁻¹) (G12V), indicating that the G12V mutation may cause the interaction more dynamic. The high residual sum of squares value (90.5) for the wildtype case indicates that the 1:1 binding model fitting for the wildtype case is not as good as for the G12V case. (C) KRas4B^WT and KRas4B^G12V show highly similar binding kinetics towards BRAF^RBD. (B) BLI was conducted in the same manner as (B), except using GST–BRAF^RBD as the ligand. The 1:1 binding model fitting (shown in red) predicts the K_D values of these two cases to be about 299 nM (WT) and 259 nM (G12V), respectively. The kon and koff values are comparable between the two cases, and the residual sum of squares values ensure the 1:1 binding model fitting is suitable for both the wildtype and the G12V cases. (D) KRas4B^WT and KRas4B^G12V samples used for circular dichroism (CD) were confirmed to be loaded with GTP. (A) The nucleotide-binding status of KRas4B^WT and KRas4B^G12V were examined in the same way as stated in (A). (E) CD spectra of KRas4B^WT and KRas4B^G12V (20 μM) at 20°C. (F) CD signal intensity at 220 nm as a function of temperature from 20°C to 90°C.

Figure 2.. Crystal structure of the KRas4B^G12V:Rgl2^RA 2:2 heterotetramer.

Cartoon representation of the structure of the heterotetramer complex of KRas4B^G12V and Rgl2^RA with top view and side view; the two KRas4B^G12V molecules are shown in dark and pale cyan and the two molecules of Rgl2^RA in pink and violet. Switch I and switch II regions of KRas4B^G12V are shown in green and yellow, respectively, and the α-helix and β-sheets are numbered for each chain. The Mg²⁺ is shown as a grey sphere. The structure shows that each KRas4B^G12V molecule interacts with two Rgl2^RA molecules (referred to as I and II) at switch I and switch II individually.

Figure S2.. Analysis of KRas4B–effector binding kinetics.

(A, B) Analyses of binding kinetics of GST–Rgl2^RA and KRas4B^WT or KRas4B^G12V. (A, B) The BLI sensorgram data presented in Fig 1B for the binding kinetics between Rgl2^RA and KRas4B^WT (A) or KRas4B^G12V (B) were fitted for the 2:1 heterogeneous binding and the 1:2 bivalent binding models using Octet Analysis Studio 13.0 (Sartorius). The fitted curves are shown in red, and the deduced K_D, kon, koff, and the residual sum of squares are shown below the sensorgrams. For the 2:1 heterogeneous binding model, the values for the two binding types (binding type 1 and 2) are shown. The proportions of each binding type (1 or 2) varied among the samples of different concentrations; hence, the varied percentage ranges are indicated. For the 1:2 bivalent binding model, the KRas4B^G12V data could not be fitted. (C) Analyses of binding kinetics of GST–BRAF^RBD and KRas4B^WT or KRas4B^G12V. The BLI sensorgram plots presented in Fig 1C were fitted for the 2:1 heterogeneous binding model using Octet Analysis Studio 13.0 (Sartorius). The fitted curves are shown in red, and the deduced K_D, kon, koff, and residual sum of squares are shown below the sensorgrams. The proportion of each binding type (binding type 1 or 2) varied among the samples of different concentrations; hence, the varied percentage ranges are indicated.

Figure S3.. KRas4B^G12V:Rgl2^RA complex is highly similar to HRas^E31K:RALGDS^RA complex.

(A) The structures of the KRas4B^G12V:Rgl2^RA complex and the HRas^E31K:RALGDS^RA complex (PDB: 1LFD) were superimposed by aligning the RA chains. (B, C) Superimposition of each corresponding chain of the KRas4B^G12V:Rgl2^RA complex and HRas^E31K:RALGDS^RA complex individually shows that each chain is similar to its respective counterpart structure. (B) KRas4B^G12V and HRas^E31K chains are in cyan and grey. (C) Rgl2^RA and RALGDS^RA are in purple and light pink. Amino acid sequence similarities of human Rgl2^RA and rat RALGDS^RA are shown below the RA cartoon representations, and the conserved amino acids in Rgl2^RA and RALGDS^RA are shaded in light green and dark green, respectively. The structural similarity between Rgl2^RA and RALGDS^RA was seen beyond the regions composed of the conserved amino acids. The RALGDS^RA sequence is derived from PDB 1LFD, which is rat RALGDS^RA, and the numbering was carried out according to UniProt Q03386. Root mean square deviation of atomic position values were generated using PyMol.

Figure 3.. The interacting interface of KRas4B^G12V and Rgl2^RA of the KRas4B^G12V:Rgl2^RA 2:2 heterotetramer.

(A) The overview of the interacting interface of KRas4B^G12V and Rgl2^RA, highlighting the residues involved in hydrogen and hydrophobic interactions between β1, β2, and α1 of Rgl2^RA1 (violet sticks) and KRas4B^G12V switch I (green sticks, enlarged in the green box) or switch II (yellow sticks, enlarged in the yellow box). An orange dashed line shows a salt bridge formed between E37 of KRas4B^G12V and R653 of Rgl2^RA. (B) Upper panel: blow-up images of the KRas4B^G12V:Rgl2^RA interface involving switch I. Lower panel: a schematic representation of the intermolecular contacts where hydrogen bonds (green dashed lines), a salt bridge (orange dashed line), and hydrophobic contacts (spiked arches) were predicted by LIGPLOT (Wallace et al, 1995) (lower panel). Numbers indicate atomic distances in Å. (C) A blow-up image of the KRas4B^G12V:Rgl2^RA interface involving switch II (left panel) and its LIGPLOT representation (right panel).

Figure 4.. The interacting interface of KRas4B^G12V:KRas4B^G12V and Rgl2^RA:Rgl2^RA of the KRas4B^G12V:Rgl2^RA 2:2 heterotetramer.

(A) The interface of KRas4B^G12V:KRas4B^G12V, highlighting the residues involved in hydrogen and hydrophobic interactions between two KRas4B^G12V molecules. The interface comprises switch I, switch II, and α3 of the two KRas4B^G12V molecules. The oncogenic mutation, V12, is annotated in red letters. A schematic representation of the intermolecular contacts predicted by LIGPLOT is presented in the right panel. (B) The interacting interface of Rgl2^RA: Rgl2^RA, highlighting the residues involved in hydrogen and hydrophobic interactions between two Rgl2^RA molecules, shaded in pink and violet. The interaction occurs at the N-terminal between the two anti-parallel β1. A LIGPLOT diagram is shown below.

Figure S4.. Schematic representation of the intermolecular contacts observed in the X-ray structure.

(A) The summary of the diagrams is presented in Fig 3. The residues in KRas4B^G12V switch I region are shown in green, in the switch II region are shown in the yellow background, and those for Rgl2^RA molecules 1 and 2 are shown in pink. V12 oncogenic mutation is in red. Hydrogen bonds, salt bridges, and hydrophobic interactions were predicted by LIGPLOT and represented by green dashed lines, solid orange lines, and blue dotted lines, respectively. (B) The electrostatic surface charge shows that the negatively charged switch I and switch II regions of KRas4B^GV interact with the positively charged surface of Rgl2^RA-1 and Rgl2^RA-2. Surface charge potential was computed using the PyMol vacuum electrostatics function, and the negatively charged and positively charged areas are shown in red and blue, respectively.

Figure S5.. KRas4B^G12V:Rgl2^RA complex formation in solution.

(A) Rgl2^RA solution structure determined by NMR. The 2D ¹H-¹⁵N-HSQC spectrum of the assigned residues of the Rgl2^RA is shown (left). The Rgl2^RA retains the ββαββαβ ubiquitin-fold structure (right). 20 Rgl2 NMR structures are superimposed. (B) Comparison between the human Rgl2^RA and mouse Rlf^RA (PDB 1RLF) solution structures. Rgl2^RA is presented in magenta, and 20 NMR structures are superimposed. Rlf^RA is in cyan, and 10 NMR structures are superimposed. The primary structures of the two constructs are listed below.

Figure 5.. NMR analysis of KRas4B^G12V:Rgl2^RA complex formation in solution.

(A) Comparison between the Rgl2^RA structures identified in solution NMR (shown in magenta) and in crystal complex with KRas4B^G12V (shown in cyan). (B) ¹H-¹⁵N HSQC titration analysis of ¹⁵N-labelled Rgl2^RA upon addition of non-labelled KRas4B^G12V. Overlays of 2D ¹H-¹⁵N HSQC NMR spectra from multipoint titrations of ¹⁵N-labelled Rgl2^RA with non-labelled KRas4B^G12V. Rgl2^RA:KRas4B^G12V molar ratios of the titration samples are colour coded as follows; 1:0 – black, 1:0.25 – blue, 1:0.25 – cyan, 1:0.5 – green, 1:1 – yellow, 1:2 – orange, and 1:3 – red. (B, C) The NMR signal intensity changes (upper panel) and the chemical shift perturbation (lower panel) of backbone ¹H^N and ¹⁵N nuclei of Rgl2^RA with non-labelled KRas4B^G12V, presented in (B), are summarised as column diagrams as a function of Rgl2^RA amino acid sequence. Proline and unassigned Rgl2^RA residues are shaded in grey. Rgl2^RA residues involved in the Rgl2^RA:Rgl2^RA interface of the KRas4B^G12V:Rgl2^RA crystal structure are highlighted in pink on the amino acid sequence, and residues involved in the KRas4B^G12V:Rgl2^RA interface are highlighted in blue. Rgl2^RA residue position numbers, according to the UniProt, are indicated at the bottom of the diagram. (Upper panel) The signal intensities of Rgl2^RA residues in the presence of three times molar excess of KRas4B^G12V were divided by the signal intensities in the absence of KRas4B^G12V and plotted as a bar chart graph. Red-dotted lines are drawn at the fold-change values of 0.5 and 1.5 to highlight the residues that show a substantial increase or decrease of the signals upon the addition of KRas4B^G12V. (Lower panel) The chemical shift perturbation of backbone ¹H^N and ¹⁵N nuclei of Rgl2^RA with non-labelled KRas4B^G12V. The mean shift difference Δδ_ave was calculated as ([Δδ¹H^N]² + [Δδ¹⁵N/10]²)^1/2 where Δδ¹H^N and Δδ¹⁵N are the chemical shift differences between Rgl2^RA on its own and in the presence of non-labelled KRas4B^G12V. The bar graphs are colour coded according to the Rgl2^RA–KRas4B^G12V concentration ratio.

Figure 6.. KRas4B^G12V:Rgl2^RA complex formation in solution.

¹H-¹⁵N-HSQC titration analysis of ¹⁵N-labelled KRas4B^G12V upon addition of non-labelled Rgl2^RA supports the KRas4B^G12V:Rgl2^RA tetramer formation. (A, B) The 2D ¹H-¹⁵N-HSQC NMR spectra of KRas4B^G12V. ¹⁵N-labelled KRas4B^G12V was titrated with non-labelled Rgl2^RA. (A) The 2D ¹H-¹⁵N-HSQC spectra of KRas4B^G12V: Rgl2^RA complex mixed with the molar ratio of 1:0 (black, left panel) and 1:2 (red, right panel) are shown. Many signals from ¹⁵KRas4B^G12V residues disappeared upon the addition of Rgl2^RA. (B) Superimposed 2D ¹H-¹⁵N-HSQC NMR spectra of ¹⁵N-labelled KRas4B^G12V:Rgl2^RA titration experiments. The titration samples are colour coded as follows; 1:0 – black, 1:0.25 – blue, 1:0. 5 – cyan, 1:0.75 –green, 1:1 – yellow, 1:1.5 – orange, and 1:2 – red. (C) Fold changes of the signal intensities of ¹⁵N-labelled KRas4B^G12V upon the addition of non-labelled Rgl2^RA. The signal intensities of KRas4B^G12V residues in the presence of two times molar excess of Rgl2^RA were divided by the signal intensities in the absence of Rgl2^RA, and the obtained values were plotted as a column graph. Undetectable residues are shaded in grey. The residue T2 was also shaded grey, as the chemical shift after the addition of Rgl2^RA overlapped with other signals. A red-dotted line is drawn at the fold-change values of 0.5 to indicate that most residues show a substantial decrease in the signals upon the addition of Rgl2^RA. KRas4B^G12V residues in the KRas4B^G12V:KRas4B^G12V interface of the KRas4B^G12V:Rgl2^RA crystal structure is highlighted in orange, and residues in the KRas4B^G12V:Rgl2^RA interface of the crystal structure is highlighted in blue. KRas4B^G12V residue positions according to the UniProt are indicated at the bottom of the diagram.

Figure S6.. Backbone resonance assignment of GDP-bound and GMPPNP-bound KRas4B^G12V.

The 2D ¹H-¹⁵N-HSQC spectra of GDP-bound and GMPPNP-bound KRas4B^G12V. Cross peaks are labelled with their corresponding backbone assignments. Residues which could not be assigned are shaded in grey in the primary sequence shown at the bottom.

Figure S7.. T₁, T₂, and {¹H}-¹⁵N NOE for backbone ¹⁵N resonances of ¹⁵N-KRas4B^G12V:Rgl2^RA complex.

(A) The longitudinal relaxation times (T₁), the transverse relaxation times (T₂), and the steady-state heteronuclear {¹H}-¹⁵N NOEs were measured using uniformly ²H/¹³C/¹⁵N-labeled KRas4B^G12V with non-labelled Rgl2^RA. Residues without data are shown in grey. (A, B) From the data presented in (A), the overall rotational correlation time τ_c, effective correlation times τ_e, and generalized order parameters S² for each ¹⁵N–¹H vector were estimated by a Lipari-Szabo model-free analysis. The deduced τ_c was ∼14.7 nsec. Meanwhile, theoretical τ_c values for a hypothetical dimer and a tetramer were calculated as follows. The two complex conformations were assumed to be spherical, and the radii of gyration were set to ∼25 and 35 Å, referring to the crystal structures of the heterodimer and the heterotetramer. By applying these values of radii to the Stokes–Einstein–Debye equation, the rotational correlation times for the hypothetical dimer and tetramer complexes were estimated to be ∼14.2 and 38.8 nsec, respectively.

Figure 7.. KRas4B^G12V:Rgl2^RA complex can form a heterotetramer in solution.

Mass photometry analysis of KRas4B^G12V:Rgl2^RA complex in solution. (A) KRas4B^G12V:Rgl2^RA complex was purified using size exclusion chromatography. The fractions were analysed with 15% SDS–PAGE gel (upper panel) according to the elution profile (lower panel). Fraction 5 was chosen and measured using mass photometry (OneMP, Refeyn). (B) The cartoon representing the different complex configurations possibilities indicates that only heterodimers (∼31 kD) and heterotetramers (∼62 kD) can be detected by the system. Histogram of the frequency counts against the purified KRas4B^GV:Rgl2^RA complex with fitting for Gaussian distribution (red). The duration of the video analysed was 60 s, and it shows that the population identified has an average mass of ∼68 kD, which is in good agreement with the expected MW of the heterotetramer (61.7 kD). (C) KRas4B^G12V and Halo-tagged Rgl2^RA (Rgl2^RA–Halo) complex formation. The cartoon represents MWs of possible complexes. KRas4B^G12V and Rgl2^RA–Halo were separately prepared and were mixed to generate a premix 2 μM sample, which was further diluted to 20, 50, 100, and 250 nM before the measurement. At 250 nM, Rgl2^RA–Halo without KRas4B^G12V showed about 55 kD, within the range of the predicted MW (∼46 kD). In contrast, the mixed samples showed an increase in the MW as the concentration was increased. At 50 nM, the observed MW was about 90 kD, which may represent the predicted dimer, and at 100–250 nM, the observed MW peaked at about 130 kD, which coincided with the predicted MW for the tetramer.

Figure S8.. Comparison between KRas4B^G12V:Rgl2^RA switch I contact and KRas^wt:CRAF^RBD complex.

KRas4B^G12V:Rgl2^RA complex formed through KRas4B switch I region was compared with one of the representative Ras:RBD complexes, KRas4B^WT:CRAF^RBD complex (PDB: 6VJJ). Both structures show β2 helices of KRas4B and RA/RBD run parallel to create the interface of the complex. Meanwhile, regarding the spatial arrangements of the α1 of Ras and α1 of RA/RBD, the axes of KRas4B^G12V α1 and Rgl2^RA α1 cross at a wider angle than the axes of KRas4B^WT α1 and CRAF^RBD α1 do. This feature is shared among Ras:RalGDS-family complexes (Eves et al, 2022).

Figure S9.. KRas4B^G12V:Rgl2^RA and HRas^E31K:RALGDS^RA complexes exhibit unique features among other Ras:effector complexes.

The KRas4B^G12V:Rgl2^RA complex was compared with representative Ras:effector complexes. The spectrum colour feature illustrates the orientation and ubiquitin super-fold status for one of the RA/RBD(s). When two RA/RBDs interact with one Ras molecule, the second RA/RBD is coloured pink (Rgl2^RA) or blue (RALGDS^RA). The KRas4B^G12V in the Rgl2^RA complex is in cyan, and other Ras molecules from previously published structures are in grey. The interacting residues involved in Ras switch II and RA/RBDs are depicted with magenta sticks and are labelled. (A) The two Ras:RalGEF complexes (KRas4B^G12V:Rgl2^RA complex and HRas^E31K:RALGDS^RA complex PDB:1LFD) show similar structural features, including the orientation of the RAs, the switch I contact and the mode of usage of switch II. Two RAs (RA-I in the spectrum colour and RA-II in either pink or in blue) interact separately with a single Ras molecule at switch I and switch II regions. (B) Examples of Ras:effector complex crystal structures. KRas4B^WT:CRAF^RBD complex (PDB: 6VJJ) shows that although Raf1^RBD has similar orientations of the ubiquitin-fold structure to Rgl2^RA-1, it does not use Ras switch II in complex formation. PI3Kγ^RBD (PDB:1HE8) and PLCε^RBD (PDB: 2C5L) complexed with Ras use switch II region differently to Ras:RalGEF complexes despite a similar orientation of the ubiquitin-fold structure of RA/RBDs. A single RBD interacts at both switch I and switch II regions of one Ras molecule. HRas^D30E/E31K:RASSF5^RA complex (3DDC) shows yet another unique interaction where the additional α-helix at the N-terminal of the RA/RBD interacts with the switch II region of the Ras molecule, instead of the RA/RBD region itself. Rgl1 is one of the RalGEFs, as Rgl2 and RALGDS are. However, unlike Rgl2 or RALGDS, KRas4B^G12V:Rgl1^RA complex (7SCX), and KRas4B^WT:Rgl1^RA complex (7SCW) show one Rgl1^RA to interact with both switch I and switch II of one Ras molecule.

Figure S10.. Rgl2^RA primary structure is highly similar to RalGEF^RA domains.

Multiple sequence alignment of Ras binding domain (RBD) of human CRAF (residues 56–131) and Ras association (RA) domains of human RalGEFs, RALGDS (residues 798–885), Rgl1 (residues 648–735), and Rgl2 (residues 648–735), as defined by UniProt. The alignment and the Clustal Omega guide tree (shown in a green-framed box) (Sievers et al, 2011), reveals that sequences are relatively divergent between CRAF^RBD and RalGEF family RA domains, although all RBD/RA domains share the common ubiquitin-fold ββαββαβ structure. The RalGEF family RA domains share a high degree of sequence homology.

Figure S11.. Heterotetramer formation in crystal structures of KRas4B^G12V:Rgl2^RA and HRas^E31K:RALGDS^RA complexes are distinct from the one in the KRas4B^G12V:Rgl1^RA complex.

Crystal structures of KRas4B^G12V:Rgl2^RA (this study), HRas^E31K:RALGDS^RA (PDB: 1LFD), and KRas4B^G12V:Rgl1^RA (PDB: 7SCX) are compared. All the Ras:RA heterodimers, formed through the Ras switch I region, show a highly similar structural arrangement (the top row). However, the way the second RA molecule interacts with the Ras:RA heterodimer in the KRas4B^G12V:Rgl2^RA is shared only with the HRas^E31K:RALGDS^RA complex (the second row). Consequently, the structural arrangements of the heterotetramers of KRas4B^G12V:Rgl2^RA and the HRas^E31K:RALGDS^RA crystal structures are distinct from the KRas4B^G12V:Rgl1^RA heterotetramer (third row). RA domains of these structures also clearly show that Rgl2^RA and RALGDS^RA are structurally different from Rgl1RA, which interacts with the second Rgl1RA through the C-terminal end (two bottom rows). The C-termini of the first RAs are shown in green.

Figure S12.. Various Ras:Ras interfaces proposed by experimental and computational prediction studies.

The top panel shows the KRas4B^G12V:KRas4B^G12V interface of the KRas4B^G12V:Rgl2^RA heterotetramer complex (this study). The KRas4B^G12V:KRas4B^G12V contacts (shaded in purple) are located in loop regions of switch I, switch II, and α3. The KRas4B^G12V:KRas4B^G12V interface is distinct from previously proposed dimerization interfaces, which can be classified into four categories based on the relevant structural elements; α5–α4, α4–α3, α–β, and β–β. Example images for each category are shown. Single Ras molecules at the Ras:Ras interface are presented with the interacting residues highlighted in purple, and the involved α-helices and β-sheets are annotated. PDB IDs of the image templates are indicated next to each image. In the last example, the dimer formation was mediated by the β–β contact, aided by a small molecule inhibitor BI2852 (shown in purple). The Mg²⁺ is depicted as a dark grey sphere, and the bound nucleotide is shown as a ball-and-stick model.

Figure S13.. The non-canonical Ras:Ras contact seen in three crystal structures, KRas4B^G12V:Rgl2^RA complex (this study), HRas^E31K:RALGDS^RA complex (PDB ID 1LFD), and HRas^WT:HRas^WT crystal structure (PDB ID 5P21).

Neighboring Ras molecules are shown to reveal comparable spatial arrangements. The side view (top panel), top view (middle panel), and the blow-up images of the area Y32 of RAS (bottom panel) are shown. Switch I and switch II regions are indicated in green and yellow. (A) KRas4B^G12V:KRas4B^G12V in the KRas4B^G12V:Rgl2^RA complex (this study). The two Rgl2^RA domains are shown as a ribbon in light magenta. (B) HRas^E31K:HRas^E31K in the HRas^E31K:RALGDS^RA complex (PDB ID 1LFD). The two RALGDS^RA domains are shown as a ribbon in light pink. (C) HRas^WT:HRas^WT in the HRas^WT crystal (PDB ID 5P21). Spatial molecular arrangements around Y32 are similar in all the cases, but the oncogenic V12 provides a larger hydrophobic pocket, which may stabilise the KRas4B^G12V:KRas4B^G12V interface of the KRas4B^G12V:Rgl2^RA heterotetramer. (A, C) E31 of KRas4B^G12V/HRas^WT interact with KRas4B^G12V/HRas^WT K88. In (A), E31 of KRas4B^G12V also interacts with Rgl2^RA K685. (B) The hydrophobic pocket created by Y32 is not reinforced by G12, but K31 of HRas^E31K interacts with rat RALGDS^RA D815, N818, and D820 (the numbering was carried out according to UniProt Q03386), facilitating the complex formation.