NMR sample preparation
For the preparation of K-Ras·GTP samples, purified protein was buffer exchanged first in 20 mM HEPES and 15 mM EDTA buffer (pH 7.0), followed by another buffer exchange in 20 mM HEPES buffer (pH 7.0) before being concentrated. Importantly, phosphate buffers were avoided in order to prepare stable GTP-complex samples. After the protein concentration was measured, the protein solution was diluted to 100 µM with 20 mM HEPES buffer (pH 7.0), and GTP ligand (Fisher Scientific) was added to a final concentration of 10 mM for further buffer exchange, a step that was then repeated twice. The final concentrated protein solution (650–750 µM) was supplemented with 5 mM MgCl2, 5 mM BME, and 5% D2O for NMR work.
NMR experiments
NMR experiments were carried out on a Bruker Avance III 850 MHz spectrometer (Bruker Inc., Billerica, MA), equipped with a 5mm triple resonance HCN cryogenic cooled TCI probe and Z-axis gradient. A set of six triple-resonance experiments (Sattler et al. 1999) were recorded using sensitivity-enhanced gradient coherence selection (Palmer et al. 1991; Kay et al. 1992), semi-constant time acquisition in the 15N dimension (Grzesiek and Bax 1993), and non-uniform sampling (NUS) following a Poisson-gap sampling schedule (Hyberts et al. 2010). Briefly, the direct free-induction decay (FID) dimension typically contained 2048 complex points with 86ms acquisition time, while the maximal number of complex points along the 15N × 13C indirect dimensions and, hence, the total acquisition times, as well as the sparsity level were as follows: HNCO, 302* (50ms) × 138* (25ms), 2.6%; HN(CA)CO, 302* (50ms) × 138* (25ms), 3.8%; HNCA, 240* (40ms) × 180* (14ms), 3.9%; HN(CO)CA, 302* (50ms) × 180* (14ms), 1.6%; HNCACB, 240* (40ms) × 460* (13.4ms), 2.9%; and CBCA(CO)NH, 300* (50ms) × 220* (6.4ms), 2.4%. Additionally, 3D 15N HSQC-NOESY and 3D CNH-NOESY (Diercks et al. 1999) were recorded with a mixing time of 180ms, and the uniformly sampled complex data matrix (acquisition times) were as follows with the direct FID dimension listed first: 1H 2048 (86ms) × 15N 60 (10ms) × 1H 180 (6.6ms), 1H 2048 (86ms) × 15N 64 (10.6ms) × 13C 116 (3.4ms). Each sample took about 12–14 days to complete. The experimental temperature was set to 288K for G12C-GTP and 283K for both WT-GTP and G12D-GTP. To aid the transfer of the backbone NH assignments to the respective 15N HSQC at room temperature, the 3D HNCO experiment was repeated at 298K on all the samples, and an additional 3D HNCA was recorded on G12D-GTP only. All the data were processed using NMRPipe/SMILE (Delaglio et al. 1995; Ying et al. 2017) and visualized with NMRViewJ (Johnson and Blevins 1993) both via NMRBox (Maciejewski et al. 2017).
Assignments
The sequential assignments followed the established 13Cα/13Cβ/13C'-based approach using the high-resolution triple-resonance spectra conferred by the NUS method. It was further aided by 3D 15N-edited NOESY and 3D CNH-NOESY experiments, where the NOE patterns and the associated 1H or 13C chemical shifts are informative regarding the nature of the amino acids in addition to offering clues for the sequential assignments since the NOEs are generally dominated by intra-residual as well sequential proton pairs. This approach led to virtually complete sequence-specific assignments in all the samples.
We also re-assigned K-Ras in its inactive forms, namely the WT-GDP, G12C-GDP, and G12D-GDP complexes that have been reported previously and, hence, are not reported here (e.g. Vo et al. 2013; Sharma et al. 2019; and Palfy et al. 2020). First, this was to ensure unequivocal sequence-specific assignments to be used in the dynamics work (Hansen et al. 2023), considering that the amide cross-peaks are highly sensitive to small changes of the buffer pH. Secondly, the work served as a prelude to the more challenging task of the resonance assignments of the respective GTP-complexes, as the analyses in these GDP forms could lend themselves to the GTP work. In summary, we have achieved all the assignments of the backbone 1HN, 15N, 13C', 13Cα, and 13Cβ of the three GDP-bound forms, with the only exception of S17 13Cβ in the two mutants. The results are consistent with previous reports, e.g. G12C-GDP (Sharma et al. 2018). In addition, the previously missing 13Cβ assignments of S17, T20, and T158 in WT-GDP (Palfy et al. 2020), which overlap with the respective 13Ca resonances, were resolved by an additional HN(CA)CB experiment.
By contrast, the NMR work on the GTP-complexes is more challenging due to the in situ hydrolysis of GTP, even in the absence of guanine nucleotide exchange factors and GTPase activating proteins. Therefore, there is significant line-broadening arising from conformational exchange. Recent work with the NMR experiments conducted at room temperature only reports around 80% backbone NH assignments (Menyhard et al. 2020), and the missing assignments were largely clustered in the conformationally flexible Switch I and Switch II regions, which are the regions of highest functional interest. To alleviate this problem, in the present work the measurements were conducted at a lower temperature and in HEPES buffer instead of the common phosphate buffer. Specifically, the experimental temperature was set to 288K for G12C-GTP, and further lowered to 283K for WT-GTP and G12D-GTP. By recording 15N HSQCs in-between 3D NMR experiments, the samples appeared to be only weakly perturbed or unchanged over the course of 3D data collection. Nevertheless, some resonances, particularly those in Switch I and Switch II, were very weak or even beyond detection in the triple-resonance experiments. Therefore, their assignments were aided by the NOE patterns and the chemical shifts of the cross-peaks in the NOESY spectra, for example, I36-E37-D38 of WT-GTP as illustrated in Fig. 1. Among the three samples, G12D-GTP showed the best behavior that led to virtually complete assignments of backbone 15N, 1H, 13Cα, and 13C' (only missing D33). Additionally, about 85.5% of its 13Cβ were assigned (Table 1).
Table 1
Summary of chemical shift assignments on kRas-GTP complexes
Sample | 15N Assigned/Total Number (Missing residues) | 1HN Assigned/Total Number (Missing residues) | 13C' Assigned/Total Number (Missing residues) | 13Ca Assigned/Total Number (Missing residues) | 13Cb Assigned/Total Number (Missing residues) |
G12C-GTP (288K) | 162/169 (P34, E63, Y64, M72, P110, P121, P140) | 163/165 (E63, Y64) | 166/169 (D33, E62, E63) | 168/169 (E63) | 151/159 (S17, T20, E62, E63, S65, M67, R68, T158) |
G12C-GTP (298K) | 162/169 (P34, Q61, Y64, M72, P110, P121, P140) | 162/165 (Q61, Y64, M72) | 161/169 (D33, G60, E63, Y71, V109, L120, I139, K169) | N/A | N/A |
G12D-GTP (283K) | 165/169 (P34, P110, P121, P140) | 165/165 | 168/169 (D33) | 169/169 | 136/159 (V8, V9, K16, S17, T20, Q22, V29, D30, D33, T35, I36, E63, Y64, S65, M67, R68, M72, F78, R97, K101, L113, K117, T158) |
G12D-GTP (298K) | 165/169 (P34, P110, P121, P140) | 165/165 | 163/169 (D33, V109, L120, I139, K169) | 168/169 (P34) | N/A |
WT-GTP (283K) | 164/169 (P34, Y64, P110, P121, P140) | 164/165 (Y64) | 165/169 (D33, Q61, E63, V109) | 167/169 (E62, E63) | 130/158 (S17, T20, V29, Y32, D33, P34, T35, I36, L56, D57, A59, Q61, E62, E63, Y64, S65, M67, R68, D69, Y71, M72, F78, D92, I93, I100, V103, T158, R161) |
WT-GTP (298K) | 162/169 (P34, Y64, S65, M72, P110, P121, P140) | 162/165 (Y64, S65, M72) | 162/169 (D33, E63, Y64, V109, L120, I139, K169) | N/A | N/A |
To transfer the backbone NH assignments to the physiologically more relevant experimental condition at 298K, at which our dynamics study was performed, a series of temperature-dependent 15N HSQC spectra were initially acquired to track the trajectories of the cross-peaks. As uncertainties arose in the overlapping regions, we repeated 3D HNCO at 298K - instead of 3D 15N-edited NOESY - to resolve ambiguities, since the former took significantly less time and the 13C' chemical shifts hardly changed over this temperature range. For example, T50 and V81 of G12C-GTP, strongly overlapped in the 15N HSQC at 288K, are well separated and easily assigned at 298K (Fig. 2) as the preceding residues E49 and C81 have 13C' chemical shifts of 175.56 and 172.81ppm, respectively. We also collected an additional HNCA for G12D-GTP at 298K, which confirmed the correctness of the 13C' information for the assignment transfer associated with the variable temperature experiments. Figure 3 shows the 15N HSQC of this sample at 298K with the assigned residues labeled.
The 13Cα/13Cb chemical shifts were used to calculate the secondary structure propensities (SSP) (Marsh et at. 2006). While the results for the GDP complexes overall agree with the crystal structure (Hunter et al. 2014) and the previous NMR work (Sharma et al. 2018), significant differences were observed between GDP and GTP complexes in the regions of Switch I and Switch II (Fig. 4). For example, in the GDP complexes, Switch I residues P34 and T35 have SSP indices close to zero, which is consistent with a high degree of intrinsic disorder. By contrast, in the three GTP-bound forms, the same residues have an average value of 0.36, ranging from 0.31 to 0.42. Consistently, the combined backbone NH chemical shift differences between the GDP and GTP forms are also more pronounced in these two regions, second only to a couple of residues in the P-loop (Fig. 5).
Extent of assignment and data deposition
The completeness of assignments is summarized in Table 1. Briefly, the backbone NH is 99.4%, 100%, and 98.8% assigned for WT-GTP at 283K (Y64 missing), G12D-GTP at 283K, and G12C-GTP at 288K (E63 and Y64 missing), respectively. For all three samples, both the 13C' and 13Cα were more than 97.6% assigned, while the 13Cβ were more than 82.2% assigned. The missing 13Cβ assignments are largely co-localized in Switch I and Switch II. At 298K, the backbone amides are 98.2% assigned for WT-GTP (Y64, S65 and M72 missing), 100% assigned for G12D-GTP, and 98.2% assigned for G12C-GTP (E63, Y64, and M72 missing). The results have been deposited to the BMRB database (https://bmrb.io) with access codes 52021 (WT-GTP, 283K), 52023 (G12D-GTP, 283K), and 52024 (G21C-GTP, 288K).