Sedimentation of large, soluble proteins up to 140 kDa for 1H-detected MAS NMR and 13C DNP NMR − practical aspects

Solution NMR is typically applied to biological systems with molecular weights < 40 kDa whereas magic-angle-spinning (MAS) solid-state NMR traditionally targets very large, oligomeric proteins and complexes exceeding 500 kDa in mass, including fibrils and crystalline protein preparations. Here, we propose that the gap between these size regimes can be filled by the approach presented that enables investigation of large, soluble and fully protonated proteins in the range of 40−140 kDa. As a key step, ultracentrifugation produces a highly concentrated, gel-like state, resembling a dense phase in spontaneous liquid-liquid phase separation (LLPS). By means of three examples, a Sulfolobus acidocaldarius bifurcating electron transfer flavoprotein (SaETF), tryptophan synthases from Salmonella typhimurium (StTS) and their dimeric β-subunits from Pyrococcus furiosus (PfTrpB), we show that such samples yield well-resolved proton-detected 2D and 3D NMR spectra at 100 kHz MAS without heterogeneous broadening, similar to diluted liquids. Herein, we provide practical guidance on centrifugation conditions and tools, sample behavior, and line widths expected. We demonstrate that the observed chemical shifts correspond to those obtained from μM/low mM solutions or crystalline samples, indicating structural integrity. Nitrogen line widths as low as 20−30 Hz are observed. The presented approach is advantageous for proteins or nucleic acids that cannot be deuterated due to the expression system used, or where relevant protons cannot be re-incorporated after expression in deuterated medium, and it circumvents crystallization. Importantly, it allows the use of low-glycerol buffers in dynamic nuclear polarization (DNP) NMR of proteins as demonstrated with the cyanobacterial phytochrome Cph1.

1.00; |2|: 13.77; |3|: 2.86.The water content was roughly estimated as follows: The integral of the region between 4.2 and 6 ppm was considered to reflect largely water, and the areas to the right and left protons of the protein.Protein signals under the water lead to an error, yet the water line had a hump extending towards 3 ppm, leading to an error in the opposite direction.We assumed a compensation of both.For the fresh sample, the integral of the water area is 3.8 times larger than the sum of the other two that are reflecting the protein.Since we are interested in the water/protein ratio, it is fair to assume that the protein is represented by integrals 1 and 3 that reflect 4862 protons (for one molecule), and the water signal contains 2 protons times x molecules.With an integral ratio of 3.8 and molecular weights of 66843 g/mol (protein) and 18 g/mol (water) a 2.5-fold excess of water (w/w) is obtained for the fresh sample.The relative intensities are given as percentages.(a) Measurements using iFD-labelled protein in a 0.7 mm rotor at 100 kHz.The 1 H bulk T1 time was determined to be 750 ms, a recycle delay of 1 s was used.(b) Measurements using 2 H, 13 C, 15 N-labelled and back-exchanged sedimented into a 1.3 mm rotor at 55 kHz MAS.The 1 H bulk T1 time was determined to be 800 ms, a recycle delay of 1 s was used.CP-based hNH @ 289 K (Fig. 4) CP-based hNH @ 313 K (Fig. 4)

Fig. S1
Fig. S1 1D 1 H spectra used to determine the water content and its change over time.A 0.7 mm diameter rotor containing sedimented SaETF was measured at 900 MHz 1 H Larmor frequency and at 100 kHz MAS.Spectra and integration regions of freshly filled sample (a and b) and after 1 year (c and d) are shown.The determined integrals are: (a and b) |1|: 1.00; |2|: 12.35; |3|: 2.24, (c and d) |1|:1.00;|2|: 13.77; |3|: 2.86.The water content was roughly estimated as follows: The integral of the region between 4.2 and 6 ppm was considered to reflect largely water, and the areas to the right and left protons of the protein.Protein signals under the water lead to an error, yet the water line had a hump extending towards 3 ppm, leading to an error in the opposite direction.We assumed a compensation of both.For the fresh sample, the integral of the water area is 3.8 times larger than the sum of the other two that are reflecting the protein.Since we are interested in the water/protein ratio, it is fair to assume that the protein is represented by integrals 1 and 3 that reflect 4862 protons (for one molecule), and the water signal contains 2 protons times x molecules.With an integral ratio of 3.8 and molecular weights of 66843 g/mol (protein) and 18 g/mol (water) a 2.5-fold excess of water (w/w) is obtained for the fresh sample.

Fig. S2
Fig. S2 Technical drawing of the 0.7mm filling tool, overview.

Fig. S3
Fig. S3Technical drawings of the 0.7mm filling tool, detailed view.

Fig. S4
Fig. S4 Effect of magic angle spinning on the sample inside the NMR rotor.(a) Top view of a 0.7 mm diameter rotor directly after filling StTS by ultracentrifugation.Due to the filling procedure, the bottom cap is attached.(b) Top view of the identical 0.7 mm rotor after multiple sessions of MAS NMR measurements at 100 kHz spinning after removing both caps.Enlarged images are shown on the right.

Fig
Fig. S5 (a) B1 profile of a 0.7mm rotor.(b) profile perpendicular to the rotor axis, taken in the center (see white arrows in (a)).

Fig. S6
Fig. S6 1D slices along a 1 H chemical shift of 8.95 ppm from 1 H-15 N correlations of deuterated SaETF by solid-state NMR (CP-based, yellow) and solution NMR (INEPT-based, blue).Full spectra are shown in main text Fig. 3.

Fig. S7
Fig. S7 Comparison of CP-based 1 H-15 N SaETF spectra to examine the change of the sample over time.Measurements were conducted with a 0.7 mm rotor spinning at 100 kHz MAS on a spectrometer with 900 MHz 1 H Larmor frequency directly after filling (orange, from the main text) and after one year of storage (blue).

Fig. S10
Fig. S10 Comparisons of transfer efficiencies on PfTrpB at 700 MHz 1 H Larmor frequency.Individual bulk intensities were determined on the first 1D free induction decay with 256 scans and compared to the bulk intensity of the first 1D experiment of a 2D hNH experiment (also measured with 256 scans).The relative intensities are given as percentages.(a) Measurements using iFD-labelled protein in a 0.7 mm rotor at 100 kHz.The 1 H bulk T1 time was determined to be 750 ms, a recycle delay of 1 s was used.(b) Measurements using 2 H, 13 C, 15 N-labelled and back-exchanged sedimented into a 1.3 mm rotor at 55 kHz MAS.The 1 H bulk T1 time was determined to be 800 ms, a recycle delay of 1 s was used.

Fig. S11
Fig. S11 DNP enhancement obtained by the radical BcTol-M on Cph1Δ2 from Synechocystis at 800 MHz 1 H Larmor frequency spinning at 20 kHz MAS.The spectra shown were acquired with (blue) and without (red) microwave irradiation.

Table S1
Acquisition parameters for the 2D 1 H-15 N spectra of SaETF.

Table S2
RF fields used to obtain the 2D 1 H-15 N spectra of SaETF.

Table S3
CP conditions used to obtain the hNH spectra of SaETF.

Table S4
Acquisition parameters for the spectra of StTS.

Table S5
RF fields and CP conditions used for the spectra of StTS.

Table S6
Acquisition parameters for the 2D hNH spectrum of the u-[ 13 C, 15 N] labeled sample of the 2B9 mutant of PfTrpB spun at 100 kHz MAS in a 0.7 mm rotor as shown in main text Fig.6.

Table S7
RF fields and CP conditions used to obtain the 1 HN bulk intensities in main text Fig.6and Fig.S9for the u-[ 13 C, 15 N] labeled sample of the 2B9 mutant of PfTrpB spun at 100 kHz MAS in a 0.7 mm rotor.

Table S8
Acquisition parameters for the 2D hNH spectrum of the u-[ 2 H, 13 C, 15 N] labeled sample of the 2B9 mutant of PfTrpB spun at 55 kHz MAS in a 1.3 mm rotor as shown in main text Fig.6.

Table S9
RF fields and CP conditions used to obtain the 1 HN bulk intensities in main text Fig.6and Fig.S9for the u-[ 2 H, 13 C, 15 N] labeled sample of the 2B9 mutant of PfTrpB spun at 55 kHz MAS in a 1.3 mm rotor.