Far-reaching effects of tyrosine64 phosphorylation on Ras revealed with BeF3– complexes

Tyrosine phosphorylation on Ras by Src kinase is known to uncouple Ras from upstream regulation and downstream communication. However, the mechanisms by which phosphorylation modulates these interactions have not been detailed. Here, the major mono-phosphorylation level on tyrosine64 is quantified by 31P NMR and mutagenesis. Crystal structures of unphosphorylated and tyrosine64-phosphorylated Ras in complex with a BeF3− ground state analogue reveal “closed” Ras conformations very different from those of the “open” conformations previously observed for non-hydrolysable GTP analogue structures of Ras. They deliver new mechanistic and conformational insights into intrinsic GTP hydrolysis. Phosphorylation of tyrosine64 delivers conformational changes distant from the active site, showing why phosphorylated Ras has reduced affinity to its downstream effector Raf. 19F NMR provides evidence for changes in the intrinsic GTPase and nucleotide exchange rate and identifies the concurrent presence of a major “closed” conformation alongside a minor yet functionally important “open” conformation at the ground state of Ras. This study expands the application of metal fluoride complexes in revealing major and minor conformational changes of dynamic and modified Ras proteins.

T he three human oncogenes, HRas, NRas and KRas, encode highly related membrane-bound Ras protein isoforms that act as "molecular switches" cycling between GDP-bound and GTP-bound forms 1,2 .Guanine nucleotide-exchange factors (GEFs) are recruited to promote the intrinsically slow GDP to GTP nucleotide exchange, which is coupled to distinct conformations at Switch I (residues 30-38) and Switch II (residues 59-72) near GDP/GTP binding site 3 .The GTP-bound active form has a high affinity to downstream effector proteins, such as Raf, and triggers the mitogen-activated protein kinase (MAPK) pathway 3 .This GTPase signalling is terminated by accelerated GTP hydrolysis catalysed by GTPase-activating proteins (GAPs) 4 .Deregulation of the GTPase cycle in Ras is commonly associated with cancer initiation and progression 5 .
Protein tyrosine kinase Src was recently shown to phosphorylate Ras proteins on tyrosine residues 32 and 64 6,7 .It was first reported that Src predominantly phosphorylates Y32 in the GDPbound GST-tagged Ras and was thought to allow the more favoured displacement of downstream effector Raf by GAP, which then increases GAP-catalysed GTP hydrolysis 6,8 .That led to a focused investigation 8 on the effect of SHP2-mediated dephosphorylation, only on Y32.However, it was later found that Ras in both GDP and GTP-bound forms can be phosphorylated on Y64 in Switch II faster than Y32 in Switch I by Src which reduces the activity of GAP 7 .Therefore, phosphorylation of these sites has been proposed as the mechanism for uncoupling phosphorylated Ras from upstream regulation in vivo, especially by attenuation both of the nucleotide exchange catalysed by GEF and of GTP hydrolysis activity catalysed by GAP 7 .Particularly, Y64 has been identified as one of the major hotspots for effector interaction 9 .Nonetheless, contradictory findings regarding Srcmediated phosphorylation levels on Y32 and Y64 have given rise to divergent hypotheses.These hypotheses revolve around whether phosphorylation enhances or diminishes GAP binding, and subsequently, how this modulation influences GTP hydrolysis rate and downstream effector interactions 6,7 .To date, only a handful of computational investigations have centred on the conformational ramifications of mono-phosphorylation on Y32 or dual phosphorylation on both Y32 and Y64 concerning effector and inhibitor interactions [10][11][12] .Nonetheless, the accurate quantification of each tyrosine phosphorylation level and the specific ways in which phosphorylation influences intrinsic Ras nucleotide exchange, GTP hydrolysis, and interactions with downstream effectors all remain elusive 7,13 .
Pre-hydrolysis state structures of Ras provide conformational information for substrate binding and explain the molecular origin of intrinsic GTP hydrolysis and nucleotide exchange.They have been depicted by numerous structures of Ras co-crystallised with non-hydrolysable GTP analogues, including guanosine 5'-[β,γ-imido] triphosphate (GMPPNP), guanosine 5'-[β,γ-methylene] triphosphate (GMPPCP), and guanosine 5'-O-[γ-thio] triphosphate (GTPγS).However, because of their non-isopolar chemical changes 14 , these compounds have modified electron densities on their γ-phosphate oxygens resulting in changed protonation states in bound complexes for Ras and significantly modified H-bonding compared to the true ground state of GTP in Ras 15,16 .Metal fluoride complexes (MF x ) have been used extensively in structural biology to monitor conformational changes and address the activation origin of proteins 17 .The ground state analogue (GSA) BeF 3 − (Protein Data Bank (PDB) Chemical ID BEF) mimics the tetrahedral geometry of phosphate before or after phosphoryl transfer. 19F NMR as a highly sensitive technique offers a more direct spectroscopic approach to provide a detailed picture of the charge distribution in the ground and transition states for P-O bond through MF x complexes mimicking the γ-phosphate of GTP in GTPases 18,19 , γ-phosphate of ATP in kinases 20,21 , and bacterial phosphatases and phosphomutases [22][23][24][25][26][27] . 19F NMR has also successfully reported the conformational changes that regulate the phosphatase activity of bacterial histidine kinase with BeF 3 − complexes 26,28 .For Ras, GDP-BeF 3 -could provide a valuable alternative ground state conformation because all three fluorine atoms are capable of accepting H-bonds as a fully deprotonated γ-phosphate in GTP 15 .However, no x-ray structures of the BeF 3 -GSA complex for Ras have been deposited in the PDB to date.
We here show by mutagenesis and 31 P NMR the monophosphorylation level on Y64 in Ras is significantly higher than for other mono-and double-phosphorylated species, ruling out Src being a dual-kinase for Ras.Our work delivers pioneering BeF 3 -GSA complex x-ray structures for unphosphorylated (Ras WT ) and Y64-monophosphorylated Ras (Ras pY64 ) that unveil novel "closed" ground state conformations for the intrinsic hydrolysis of Ras, which are distinct from other nonhydrolysable GTP analogue-bound structures.The phosphorylation on Y64 in Switch II triggers a cascade of conformational changes beyond the active site, potentially impairing the binding of Ras to downstream effector Raf and offering an alternative mechanism to previous proposals 12 .Combined with highresolution x-ray crystal structures, the subtle yet significant 19 F NMR chemical shifts and linewidth changes between Ras WT -GDP-BeF 3 -and Ras pY64 -GDP-BeF 3 -GSA complexes offer viable insights into decreased intrinsic GTPase rate and increased nucleotide exchange rate on phosphorylation 6 .Our study expands the evidence-based understanding of post-translational modification (PTM)-induced conformational changes of Ras.Furthermore, it underscores the capability of 19 F NMR to detect unnoticed conformations and conformational alterations in high-resolution structures, demonstrated here via MF x complexes.

Results and discussion
Phosphorylation of Y64 is the main Ras site for Src kinase.Expression tags have been shown to affect the site preference of tyrosine phosphorylation of Ras 6 .Thus, we recombinantly generated un-tagged Ras.A phosphorylation assay of the GDPbound HRas (hereafter Ras) by Src catalytic domain was set up by using physiologically relevant concentrations of Mg 2+ and ATP 29 .Mass spectroscopy (MS) analysis detected unphosphorylated, mono-, and double-phosphorylated species (Supplementary Figure 1).To quantify the site-specific phosphorylation, ion exchange chromatography was used to separate major mono-phosphorylated Ras species from non-and double-phosphorylated ones; double-phosphorylated species only show minor peaks (Supplementary Figure 2).The sole monophosphorylated Ras fraction was trypsin digested and MS detected phosphorylated peptide fragments only on Y64 (Supplementary Figure 3).To specifically quantify phosphorylation levels, Y32F and Y64F variants of Ras, Ras Y32F and Ras Y64F , were also prepared and phosphorylated by Src under the same condition as for Ras WT . 31P NMR shows the integration of the resonance from the phosphorylated tyrosine (-0.4 ppm) is reduced by 85% for the Ras Y64F variant compared to Ras WT and Ras Y32F variant (Fig. 1).This reveals that Y64 is the main phosphorylation target of Src.It conflicts with the conclusion in previous work that Y32 is the major phosphorylation site, possibly due to the in vitro phosphorylation assay was there carried out with GST-tagged HRas 6 .In addition, the resonance of P B 30 of the bound GDP in the Y64-phosphorylated Ras Y32F variant moved downfield by 0.5 ppm relative to the less-phosphorylated Ras Y64F variant, indicating the mutation of Y32 to F has induced local conformational changes around the P-loop (Fig. 1) 31 .
A GDP-BeF 3 -complex delivers conformational changes different from non-hydrolysable GTP analogues.After unsuccessful attempts of crystallising a BeF 3 -GSA complex by cocrystallising Ras WT -GDP with Be 2+ and F -, we adopted a new approach by soaking Ras WT -GDP apo crystals with 50 mM BeCl 2 and 0.8 M NH 4 F for 1 to 2 min followed by flash freezing.We successfully obtained a BeF 3 -structure for Ras WT -GDP-BeF 3 - GSA complex (1.35 Å resolution, PDB: 8CNJ, Table 1, Supplementary Data 1).In this complex, the α2-helix (residues 66-74), which overlaps with Switch II in sequence, deviates from the one in the Ras-GDP structure by 42°but it adopts a conformation highly similar to the GTP-or GTP analogue-bound unphosphorylated Ras structures with an angle difference of <7°( Fig. 2a, Supplementary Figure 4, Table 2).This demonstrates this Ras WT -GDP-BeF 3 -complex is mimicking a GTP-bound GSA state.This is also echoed by the high B-factor of Switch II observed, showing more mobility around the Switch II region than the rest of the protein (Fig. 3b), similar to the observation in the structure for the RhoA/RhoGAP product complex with both the GDP and P i bound, where the inorganic phosphate was also introduced by ligand soaking 32 .However, different from the GMPPCP (PDB: 121P) or GMPPNP-bound (PDB: 5P21) structures of Ras WT , our BeF 3 -structure has well-defined electron densities for both Switch I and Switch II, in which Y32 donates an H-bond to F 3 and also accepts an H-bond from Q61 to the phenolic-OH of Y32.Thus, both Y32 and Q61 adopt a "closed" conformation (Fig. 2b).This is significantly different from the Y32 and Q61 conformations in other GSA structures of Ras with non-hydrolysable GTP analogues or by cryo techniques from a caged GTP analogue 33 (Fig. 3a).In the Ras-GMPPCP (PDB: 121P) and Ras-GMPPNP (PDB: 5P21, 4RSG) GSA complexes, the Y32 and Q61 are in a completely "open" conformation.
In the Ras-GTPγS GSA complex (PDB: 5VQ6), Y32 and Q61 are both disordered and incapable of providing any information.
Since the overall conformations of the GTP analogues are close to that of our BeF 3 -complex structure, this closure of Y32 and Q61 must relate to the electron density of the surrogate γphosphoryl group atoms.Numerous computational studies based on these GTP analogue structures have concluded that the slow, intrinsic hydrolysis of GTP by Ras involves a solvent-assisted pathway via a "2 W" mechanism 34,35 , with partial support from a GTP-bound structure from a caged GTP analogue 33 .Calculations of conformational effects by PTMs and mutations of Ras have also been based on structures using non-hydrolysable GTP analogues 36,37 .Therefore, it is highly significant that in our Ras WT -GDP-BeF 3 -GSA complex structure there is no second water molecule within 7.8 Å of P G .Only the isolated nucleophilic water is in a near-attack conformation (NAC) at P G along with the two waters closely coordinated to the octahedral catalytic Mg 2+ .This NAC water has an in-line angle (O 3B -Be-Ow) 161°a nd is 3.5 Å from the beryllium atom.It donates an H-bond directly to T35 and F 3 and accepts a H-bond from Q61 (Fig. 2b).This highlights the different conformational details provided by GDP-BeF 3 -and GTP analogues might lead to different mechanistic conclusions when alternative H-bonding patterns from residues G60 and Q61 are included in the QM zone 15,[38][39][40] .Furthermore, by comparing the 19 F chemical shifts of the same resonance measured in 10% D 2 O and 90-100% D 2 O, solventinduced isotope shift (SIIS) can be calculated which accurately reflects the number and orientation of H-bond donors around each fluorine 42 .To assess whether the structural changes reflected by the BeF 3 -GSA structure are genuinely caused by Y64 phosphorylation and not by crystallographic artefacts, we investigated this BeF 3 -complex in solution 19 F NMR with presaturation on free fluoride resonance at −120 ppm.The pre-saturated 19 F NMR spectrum of the Ras WT -GDP-BeF 3 -complex shows three major well-resolved peaks for the protein-bound BeF 3 -moiety (Fig. 4a), outside the known range (-163 to -170 ppm) for free BeF x species.The signal of F 1 at −176.3 ppm is significantly more upfield than the other resonances and has 0.3 ppm SIIS, indicating that it has a relatively high electron density and low proton density.It is thus assigned to F 1 , coordinated to the catalytic magnesium 25,26,28 .F 2 at −152.3 ppm and F 3 at −160.4 ppm are also assigned by a combination of SIIS and a partial deuteration strategy (Table 3, Supplementary Figure 5) 19,28 .In particular, the 19 F resonance of F 2 , H-bonded to the K16 ammonium group, is differentially shifted by rotationally averaged HHH, HHD, HDD, and DDD congeners, leading to an unresolved peak in 10%, 50% and 95% D 2 O buffers.By contrast, F 3 , coordinated by two welldefined H-bonds from Y32 and the nucleophilic water, shows a resolved resonance at 50% D 2 O 19,28 .
The Ras WT -GDP-BeF 3 -GSA complex exhibited a minor set of 19 F signals comprising ~10% of the population, characterised by three resonances at -155.3 ppm, -162.6 ppm, and -183.0 ppm.Notably, all three resonances are shifted 3-7 ppm upfield relative to those of the major species (Fig. 4a), implying increased shielding for all three fluorines of BeF 3 -in this minor species, a sign of the reduced extent of H-bonding to the BeF 3 -moiety.To investigate whether this minor species originates from the "open" conformation that has been crystallised in the structure of Ras Y32F -GMPPNP (PDB: 3K9N), where the benzyl side chain adopts an "open" conformation, we introduced a Y32F mutation and generated the Ras Y32F -GDP-BeF 3 -GSA complex, monitored using 19 F NMR (Fig. 4b).Applying presaturation on the unbound fluoride resonance at -120 ppm was essential to distinguish the fluorine resonances associated with Ras from the free BeF x resonances present in the solution around -167 ppm (Supplementary Figure 6).The 19 F NMR spectrum of the Ras Y32F -GDP-BeF 3 -GSA complex reveals a distinct overall profile: The trio of resonances, aligned with chemical shifts akin to the minor species in the Ras WT -GDP-BeF 3 -GSA complex, emerged as predominant by integrations.In contrast, fluorine F 3 from the second set of peaks, exhibiting chemical shifts more similar to the resonances of the "closed" conformation, has shifted downfield by 7.5 ppm and fused with F 2 (Table 3).This is likely due to a water molecule assuming the role of an H-bond donor to F 3 , analogous to the phenolic -OH of Y32.This strongly underscores the pivotal role of Y32 -OH in neutralising the negative charge on the γ-phosphate during intrinsic GTP hydrolysis.It is noteworthy that, at equal component concentrations, the significantly lower signal-to-noise ratio in the Ras Y32F -GDP-BeF 3 -GSA implies that the occupancy of BeF 3 -is less than 100%, unlike the Ras WT -GDP-BeF 3 -complex, suggesting that Y32 -OH also contributes to the binding of both BeF 3 -and GTP.The 10% population of Ras WT -GDP-BeF 3 -GSA in "open" conformation explains why we could not crystallise this complex by co-crystallisation after many attempts and why its structure has been missing in the PDB. 19F NMR of metal fluoride complexes identify a structurally invisible effect of phosphorylation of Ras Y64.We next examined the impact of Y64 phosphorylation by investigating the Ras pY64 -GDP-BeF 3 -GSA complex via solution 19 F NMR.The 19 F NMR spectrum shows three major well-resolved resonances.Notably, they show small upfield shifts of 0.6 and 0.5 ppm for F 1 and F 2 , respectively, compared to the Ras WT -GDP-BeF 3 -complex.Conversely, F 3 experiences a 0.5 ppm downfield shift (Table 3, Fig. 4c).Although these chemical shift changes are marginal, they  13.53°R as pY64 -GDP-BeF 3 -, PDB: 8CNN; Ras WT -GDP-BeF 3 -, PDB: 8CNJ; Ras WT -GTP, PDB: 1QRA (caged-GTP analogue); Ras WT -GMPPNP, PDB: 121P; Ras WT -GMPPCP, PDB: 5P21; Ras pY64 -GDP, PDB: 8BOS; Ras WT -GDP, PDB: 4Q21.Data are colours coded by the level of difference for clarity.
show only a small impact of Y64 phosphorylation on the catalytic site with no direct interaction existing between the BeF 3 -moiety and the negatively charged phosphate on the phenolic -OH group of Y64.The average upfield shift of 0.2 ppm across the three resonances, compared to Ras WT -GDP-BeF 3 -, likely signifies a marginal increase in electron density on the three oxygen atoms within the γ-phosphate.This effect might induce a slightly increased repulsion during nucleophilic attack, thus rationalising the observed 3-fold reduction in intrinsic GTPase activity following phosphorylation.Moreover, the three resonances from the Ras pY64 -GDP-BeF 3 -complex have an average 0.70 ppm halfheight line width, larger than the mean 0.48 ppm of the Ras WT -GDP-BeF 3 -complex.This indicates a slightly slower exchange rate of the BeF 3 -moiety with free fluoride and free BeF x species in the Ras pY64 -GDP-BeF 3 -complex on the NMR time scale.Furthermore, a noteworthy but subtle observation emerges: the ratio between major and minor forms, calculated from the averaged integrations of F 1 and F 2 , decreases by ~10%.This signals an increased 'open' conformation population in Ras pY64 -GDP-BeF 3 - following Y64 phosphorylation.Considering the pM affinity binding of GDP to Ras, the accelerated exchange rate of BeF 3 - and increased population of open conformation further strengthens the assertion of a 2.6-fold increase in intrinsic nucleotide exchange rate upon phosphorylation (primarily on Y64), as also corroborated by NMR measurements 7 .
To investigate the impact of Y64 phosphorylation on GAP binding, we also attempted to form an MF x TSA complex with Ras pY64 and RasGAP monitored by 19 F NMR (Supplementary Figure 7).However, our attempts were which stands in stark contrast to the easily formed RasGAP/Ras-GDP-"AlF 3 0 " TSA complex (PDB: 1WQ1) observed by both crystallography 43 and backed up by our 19 F NMR test (Supplementary Figure 8).This confirms phosphorylation on Y64 could disrupt productive interactions with RasGAP, likely due to the steric hindrance because Y64 forms a H-bond with L902 of RasGAP in the RasGAP/Ras-GDP-AlF 3 0 complex structure.
Effect of phosphorylation of tyrosine64 revealed by the structure of Ras pY64 -GDP-BeF 3 -GSA complex.To gain atomistic insight into how phosphorylation of Y64 could influence intrinsic GTP hydrolysis and nucleotide exchange rates with phosphorylated Ras, we initially assessed the stability of phosphorylated Ras.Our results indicated that monophosphorylated Ras remained stable in buffer for over a week, rendering it suitable for subsequent crystallisation trials (Supplementary Figure 9).We first crystallised the apo Ras pY64 -GDP apo structure with commercial crystallisation screens.This structure was solved at 1.32 Å resolution (Table 1, PDB: 8BWG, Supplementary Data 2) but residues 60-64 as a key part of the Switch II loop do not have clear electron density.Thus, it cannot provide the accurate location and coordination of pY64.α2-Helix adopts a conformation that is highly similar to other GDP-bound product structures of Ras with an angle difference of <5°degree (Table 2).
We then obtained a BeF 3 -GSA complex by soaking Ras pY64 -GDP apo crystals with 50 mM BeCl 2 and 0.8 M NH 4 F for 1 to 2 min before flash-freezing.The Ras pY64 -GDP-BeF 3 -GSA complex structure formed diffracts to 1.48 Å resolution (Table 1, PDB: 8CNN, Supplementary Data 3).While different from the Ras pY64 -GDP apo structure, the Ras pY64 -GDP-BeF 3 -complex has welldefined electron densities for the flexible Switch II loop including the whole pY64, and the ordered Switch I loop in a "closed-up" conformation, where Y32 directly donates an H-bond to F 3 (Fig. 5a, Supplementary Figure 10).Compared to the unphosphorylated Ras WT -GDP-BeF 3 -GSA complex (backbone atom RMSD 0.766 Å, Supplementary Figure 11), our Ras pY64 -GDP-BeF 3 -GSA structure exhibits noticeable conformational changes (Fig. 5b).Originating from the phosphate in pY64, these extensive conformational shifts (spanning up to 28 Å) are a result of the alteration (80°) in the rotameric angle of the phenolic ring.This change is facilitated by an interaction with the backbone carbonyl  group of Switch I P34, mediated through a water molecule.In turn, this changes the dihedral angle ψ (O-C-C α -N) of S65 from 39.3°to 173.5°, and the deviation of the α2-helix is further propagated throughout from the N-end of the α2-helix (residues 64-74) to the C-end of the α3-helix (residues 87-104) by a shift of by 2.5 Å (Fig. 5b).Indeed, T74 and M67 have shown noticeable chemical shift changes in 1 H- 15 N HSQC in solution after phosphorylation 7 .The distinctive water-mediated interaction between the negatively charged phosphate on pY64 in Switch II and P34 in Switch I suggests its potential to obstruct the entry of a second water molecule into the active site.This observation offers another plausible rationale for the observed three-fold reduction in intrinsic hydrolysis rate 7 .
The diminished interaction between phosphorylated Ras and Raf-RBD has been explored by molecular dynamics, suggesting that the heightened flexibility of phosphorylated Y32 hinders Raf binding 7 .When Ras pY64 -GDP-BeF 3 -is aligned with Ras-GMPPNP-Raf complex structure (PDB: 4G0N), a noticeable alteration becomes evident.The backbone carbonyl group of pY64 reorients itself by forming a hydrogen bond with a water molecule, which simultaneously participates in H-bonding with the carbonyl group of A66 and the carboxylate sidechain of E37.
The key interactions involving E37 of Ras and R59 and R67 of Raf are disrupted due to the substantial changes in the orientations of both E37 and Y71 following Y64 phosphorylation (Fig. 6).Our structure offers an additional explanation for the significant reduction in affinity between the downstream effector Raf and phosphorylated Ras, mediated by Src.This reduction is attributed to the extensive long-range conformational alterations resulting from the Y64 phosphorylation 7 .

Conclusion
In this study, we have substantiated Y64 as the predominant site of phosphorylation by Src.This finding suggests that Ras likely exerts its biological function primarily in its monophosphorylated form.Making new use of pM GDP binding and readily available Be 2+ and F − salts, we successfully obtained a GDP-BeF 3 − GSA complex for both unphosphorylated and Y64-phosphorylated Ras.The large conformational alterations induced by BeF 3 -within the crystalline structure are of substantial significance.This demonstrates new efficacy and validity of a soaking approach in generating GDP-BeF 3 -GSA complexes for small G proteins from their GDPbound structures.It offers a convenient strategy for structural biological inquiries, without the need to carry out nucleotide exchange with non-hydrolysable GTP analogues.Our investigation reveals that the impact of Src-mediated Y64 phosphorylation on Ras extends beyond the Switch I and II regions.It encompasses broader steric and conformational effects, significant for the biological roles of this crucial PTM of Ras, especially for its interactions with effector proteins such as Raf.These two BeF 3 -GSA complexes both capture a "closed" conformation where Y32 in Switch I directly interacts with Q61 in Switch II by H-bonding, different from other ground state conformations as depicted by caged-GTP, GMPPNP, GMPPCP, or GTPγS.Furthermore, the utility of 19 F NMR is extended to elucidate the underlying molecular basis of conformational changes induced by PTMs, which might not be captured by high-resolution structures.It serves as a precedent for its application to the investigation of other PTMs, such as S-nitrosylation on C118 44 .Indeed, utilising 19 F NMR on BeF 3 - GSA complexes reveals that Switch I predominantly adopts a  "closed" conformation ( ~90% occupancy) in solution.Our study underscores the utility of BeF 3 -complexes to reveal distinct conformational nuances opaque with non-hydrolysable GTP analogues.Finally, BeF 3 -GSA structures fill a crucial gap for molecular docking in Ras-targeted drug discovery and present valuable starting points for computational investigations, quantum mechanics/molecular mechanics (QM/MM) and molecular dynamics (MD), which can deliver disparate mechanistic and conformational insights.
Gene expression and protein purification.HRas(1-166): E. coli BL21(DE3) cells were transformed with ptac-HRas (1-166) plasmid, plated onto LB agar plates (100 μg/mL ampicillin) and grown overnight at 37 °C.A single colony was used to inoculate 30 mL of LB media (100 μg/mL ampicillin) and incubated at 37 °C overnight.This culture was used to inoculate LB medium (100 μg/mL ampicillin) at a 1:100 ratio.The culture was grown at 37 °C to an OD 600 between 0.6 and 0.8 and gene expression was induced with 1.0 mM IPTG.The culture was shaken at 25 °C for 18 h and cells were harvested by centrifugation (7000 g, 4 °C, 20 min).The cell pellet was either stored at -80 °C or processed directly after centrifugation.The cell pellet was resuspended in buffer A (Tris-HCl 25 mM, pH = 7.6, MgCl 2 5 mM, DTT 1 mM) and supplemented with 1 mM PMSF.The cells were lysed by sonication (4 min sonication time, 2 s on and 8 s off) and cell debris was removed by centrifugation (32000 g, 4 °C, 40 min).The supernatant was filtered and loaded on a DEAE column, washed with 3 CV lysis buffer and eluted by applying a gradient of 0-100% elution buffer (Tris-HCl 25 mM, pH = 7.6, NaCl 200 mM, MgCl 2 5 mM, DTT 1 mM) over 8 CV.The eluted protein fractions were pooled and concentrated.Finally, the target protein was purified on a SEC75 26/60 column and stored at -80 °C.The same procedure was used for ptac-HRas (1-166) -Y32F and ptac-HRas (1-166) -Y64F.
RasGAP 334 (714-1047): E. coli BL21(DE3) cells were transformed with pGEX-2T-RasGAP (714-1047) plated onto LB agar plates (100 μg/mL ampicillin) and grown overnight at 37 °C.A single colony was used to inoculate 30 mL of LB media (100 μg/ mL ampicillin) and the culture was shaken at 37 °C overnight.This culture was used to inoculate LB media (ampicillin 100 μg/mL) at a 1:100 ratio.The culture was grown at 37 °C to an OD 600 between 0.6 and 0.8 and expression was induced with 1.0 mM IPTG.The culture was incubated at 18°C for 18 h and cells were harvested by centrifugation (7000 g, 4 °C, 20 min).The cell pellet was resuspended in buffer A (Tris-HCl 50 mM, pH = 7.6, NaCl 150 mM, MgCl 2 5 mM, DTT 1 mM) and supplemented with 1 mM PMSF.Cells were lysed by sonication (4 min sonication time, 2 s on and 8 s off) and cell debris was removed by centrifugation (32000 g, 4 °C, 40 min).The supernatant was filtered and loaded on GST resin (GST-HP resin, Cytiva, United States).The resin was incubated on a tube roller at 4 °C for 60 min and washed with 5 CV buffer A. The target protein was eluted over 3 CV with buffer B (Tris-HCl 50 mM, pH = 7.6, NaCl 150 mM, MgCl 2 5 mM, DTT 1 mM, glutathione 10 mM) and buffer exchanged into buffer A. To cleave the GSTtag the protein solution was incubated at 4 °C with 25 NIH units of thrombin (T7326-1KU, Sigma Aldrich, United States).Progress of the cleavage reaction was controlled via SDS-PAGE and further thrombin was added as required.Upon completion of the cleavage reaction, the protein solution was incubated with GST resin at 4 °C.After 60 min the flow-through was collected and concentrated.Finally, the target protein was purified on a SEC75 26/60 column and stored at -80 °C.
Src(251-533): The catalytic domain of chicken Src was prepared following the protocol described in the literature 45 .E. coli BL21(DE3) cells were co-transformed with pET28-cSrc (251-533) and pCDFDuet-YoPH and incubated on LB agar (50 μg/mL kanamycin and 50 μg/mL streptomycin) overnight at 37°C.A single colony was used to inoculate 30 mL of LB media (50 μg/mL kanamycin and 50 μg/mL streptomycin) and the culture was shaken at 180 rpm at 37 °C overnight.This culture was used to inoculate TB medium (50 μg/mL kanamycin and 50 μg/mL streptomycin) at a 1:100 ratio.The culture was grown at 37 °C to an OD 600 between 1.0 and 1.2 and gene expression was induced with 0.2 mM IPTG.The culture was incubated at 18 °C for 16 h and cells were harvested by centrifugation (7000 g, 4 °C, 20 min).The cell pellet was either processed directly after centrifugation or stored at -80 °C.The cell pellet was resuspended in buffer A (Tris-HCl 50 mM, pH = 8.0, NaCl 500 mM, imidazole 25 mM, glycerol 5% (v/v)) and supplemented with 1 mM PMSF.Cells were lysed by sonication (4 min sonication time, 2 s on and 8 s off) and cell debris removed by centrifugation (32000 g, 4 °C, 40 min).The supernatant was filtered and loaded on a Ni 2+ -NTA column (Cytiva, 5 mL FF HisTrap).After washing with 5 CV buffer A, the target protein was eluted by applying a gradient of 0-50% buffer B (Tris-HCl 50 mM, pH = 8.0, NaCl 500 mM, imidazole 500 mM, glycerol 5% v/v) over 30 CV.The eluted protein fractions were pooled, concentrated, and dialysed at 4 °C overnight against 20 volumes of buffer C (Tris-HCl 20 mM, pH = 8.0, 100 mM, DTT 1 mM, glycerol 5% v/v).The crude kinase was then loaded onto a Q column (Cytiva, HiTrap FF 5 mL) and eluted by applying a gradient of 0-40% buffer D (Tris-HCl 20 mM, pH = 8.0, NaCl 1.0 M, DTT 1 mM, glycerol 5% v/v).Fractions containing the kinase were pooled, further purified on a SEC75 16/60 column and stored at -80 °C.Ras phosphorylation assay by Src.500 µM Ras-GDP was incubated for 2 h at 25 °C in assay buffer (Tris-HCl 25 mM pH = 7.6, NaCl 150 mM, MgCl 2 5 mM, DTT 1 mM, ATP 4 mM, cSrc 20 µM).To remove excess ADP/ATP and quantify each Ras species, monophosphorylated Ras as the major species shown by the chromatogram was separated from unphosphorylated and doublephosphorylated species by ion-exchange chromatography on a 16 × 100 mm Q FF 16/10 column.The chromatogram shows a major monophosphorylated peak, and two double-phosphorylated peaks as identified by MS. (Supplementary Figure 2, Supplementary Figure 12).
Mass spectrometry (MS).Ras WT , Ras variants and all the phosphorylated Ras species were subjected to MS for confirmation (Supplementary Figure 12).Liquid chromatography-mass spectrometry (LC-MS) was performed on a WATERS Synapt G2-Si quadrupole time-of-flight mass spectrometer coupled to a WATERS Acquity H-Class ultraperformance liquid chromatography (UPLC) system.The column was a WATERS Acquity UPLC protein BEH C4 (300 A ̊, 1.7 μm × 2.1 mm × 100 mm) operated in reverse phase and held at 60 °C.The gradient employed was 95% A to 35% A over 50 min, where A is water with 0.1% HCO 2 H and B is acetonitrile with 0.1% HCO 2 H. Spectra were collected in positive ionisation mode and analysed using WATERS MassLynx software version 4.1.Deconvolution of protein-charged states was obtained using the maximum entropy 1 processing software.
Nuclear Magnetic Resonance (NMR) spectroscopy.NMR spectra in this work have been recorded on a 500 MHz Bruker fivechannel liquid-state spectrometer with a high sensitivity QXI cryoprobe.Chemical shifts (δ) are given in parts per million (ppm).All spectra were recorded at 20 °C.For the presaturation of the free fluoride signal, elective 19 F irradiation was achieved with a continuous wave (power level of 42 dB) applied over the 1 s recycle delay at the frequency of free fluoride peak (-119.5 ppm).For samples with 90% D 2 O, this frequency was adjusted to -121.5 ppm.Unless stated otherwise, all protein 19 F NMR spectra were calibrated to an internal fluorobenzene standard at -113.79 ppm 46 .
Stability of phosphorylation on Ras tyrosine.The stability of phospho-Ras was determined by incubating 100 μM monophosphorylated Ras-GDP in crystallisation buffer (HEPES-Na 20 mM, pH = 8.0, MgCl 2 10 mM, NaF 20 mM).At regular intervals, 30 μL aliquots were taken and mixed with SDS-PAGE loading buffer, heated to 95 °C for 3 min and stored at -80 °C until all time points could be analysed by SDS-PAGE (Supplementary Figure 9).Protein x-ray crystallography.Protein crystallisation conditions were set up either using a Douglas Instruments ORYX4 or a SPTlabtech Mosquito Crystal system in either a hanging drop or sitting drop configuration.Microseeding was performed based on literature conditions 47 .
Data collection, structure solution and refinement.The datasets described in this report were collected at the Diamond Light Source, Didcot, Oxfordshire, U.K. on beamline I03.Data were integrated using XDS 48 and scaled/merged using AIMLESS 49 included in the CCP4 software suite xia2 50 .Data collection and refinement statistics are provided in Table 1.The structures were solved by molecular replacement using MOLREP 51 with one monomer of Ras in PDB: 1WQ1 as the model and refined with Refmac 52 .
Reporting summary.Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Fig. 2
Fig. 2 Ras WT -GDP-BeF 3 -GSA complex.a Overlay of Ras WT -GDP (orange, PDB: 4Q21) and Ras WT -GDP-BeF 3 -(purple, PDB: 8CNJ).RMSD for the structure alignment is plotted for each residue.The major conformational changes are induced in the Switch I and Switch II regions.b H-bond interactions for the Ras-GDP-BeF 3 -GSA complex were observed in its structure.

Fig. 5
Fig. 5 The structure of the Ras pY64 -GDP-BeF 3 -GSA complex.a Active site of Ras pY64 -GDP-BeF 3 -.F o -F c omit map for GDP, BeF 3 -and pY64 are contoured at 3σ, 0.330 Å -3 (light blue mesh).Key neighbouring residues are shown in sticks.H-bonds coordinating to BeF 3 -are shown as dashed lines.b Overlay of Ras WT -GDP-BeF 3 -(purple) and Ras pY64 -GDP-BeF 3 - (grey).RMSD for the structure alignment is plotted for each residue.

Table 1 X
-ray data collection, processing, and refinement statistics.

Table 2
Angle differences in α2-helix of the Switch II region for Ras structures.