|Home | About | Journals | Submit | Contact Us | Français|
MAS solid-state NMR is capable of determining structures of protonated solid proteins using proton-detected experiments. These experiments are performed at MAS rotation frequency of around 110kHz, employing 0.5mg of material. Here, we compare 1H, 13C correlation spectra obtained from protonated and deuterated microcrystalline proteins at MAS rotation frequency of 111kHz, and show that the spectral quality obtained from deuterated samples is superior to those acquired using protonated samples in terms of resolution and sensitivity. In comparison to protonated samples, spectra obtained from deuterated samples yield a gain in resolution on the order of 3 and 2 in the proton and carbon dimensions, respectively. Additionally, the spectrum from the deuterated sample yields approximately 2–3 times more sensitivity compared to the spectrum of a protonated sample. This gain could be further increased by a factor of 2 by making use of stereospecific precursors for biosynthesis. Although the overall resolution and sensitivity of 1H, 13C correlation spectra obtained using protonated solid samples with rotation frequencies on the order of 110kHz is high, the spectral quality is still poor when compared to the deuterated samples. We believe that experiments involving large protein complexes in which sensitivity is limiting will benefit from the application of deuteration schemes.
Magic Angle Spinning (MAS) solid-state NMR for biological samples has taken a large leap forward during the past years1–8. Advancements in NMR technology such as the development of ultrahigh magnetic fields (ω0/2π>1GHz), emergence of ultrafast spinning probes (ωr/2π>100kHz), isotopic labelling strategies and development of new NMR methods are some of the key points behind this success. Solid-state NMR samples are spun in rotors at the magic angle (Arctan [√2]=54.74°) with respect to the main magnetic field. The maximum achievable frequency of sample spinning at the magic angle is inversely proportional to the diameter of the solid-state NMR rotor9. Therefore, faster spinning rotors imply smaller volumes with less amount of sample and thus, a reduced sensitivity of the NMR spectra to start out with. Other parameters such as the diameter of the rf coil, detection of protons instead of heteronuclei may counteract the decrease of sample volume10.
Protons yield the highest sensitivity for detection due to their high gyromagnetic ratio. However, the strong proton-proton dipole couplings yield significant line broadening. Using deuterated samples helps to recover the spectral resolution while retaining high sensitivity. Several deuteration strategies have been proposed for solids during the last decade, that are now widely used. In general, deuteration strategies have to be adopted according to the available NMR hardware, in particular the achievable MAS frequency. Specific labelling schemes allow to study exchangeable as well as non-exchangeable protons. At a maximum MAS frequency of 24kHz (3.2 mm MAS rotor, 10–20mg sample), partial deuteration (H:D ≈ 20:80) at exchangeable amide sites in an extensively deuterated environment is required to obtain optimal spectral quality11–13. The emergence of 1.3 mm probes (2mg of sample, maximum ~60kHz MAS) allowed to employ 100% protons at exchangeable sites in an otherwise deuterated sample without having to sacrifice sensitivity or resolution14–17. On the other hand, the proton density in the aliphatic sidechain is significantly higher compared to the amide backbone. For such non-exchangeable protons, several fractional deuteration protocols have been implemented for solid-state NMR such as random deuteration at the aliphatic sites or RAP18, selective ILV-methyl labelling19, SAIL20, iFD21 and proton-cloud22 schemes. It was shown that fractional labelling of side chain protons yields excellent spectral resolution even at moderate MAS frequencies23–27. Finding the optimal deuteration scheme for non-exchangeable sites at a given MAS frequency poses a greater challenge in comparison to backbone amides, since deuteration itself limits the observation of side chain resonances. In addition, deuteration of a protein often significantly reduces protein expression yields28.
It has been suggested recently that sample rotation with a frequency of ~110kHz (in 0.7 mm MAS rotor, 0.5mg of material), which corresponds to the maximum MAS frequency achieved to-date, eliminates the need for deuteration to achieve high resolution29, 30. On the other hand, simulations indicate that MAS frequencies on the order of 250kHz are necessary to yield comparable proton resolution31.
In this work, we compare 1H, 13C correlation spectra obtained for protonated and deuterated samples of a microcrystalline protein at a MAS rotation frequency of 111kHz, and show that the spectral quality obtained from deuterated samples is superior to those acquired using protonated samples in terms of resolution and sensitivity.
Figure 1 shows the 1H,13C correlation spectra obtained for a protonated and two types of deuterated samples of a microcrystalline sample of the α-spectrin SH3 domain. The 25% and 5% deuterated RAP samples are prepared using u-[2H,13C]-glucose, and a D2O based M9 growth medium that has been diluted with 25% and 5% H2O to yield selective incorporation of protons at the aliphatic sites, respectively24. The SH3 sample in which pro-R or pro-S methyl groups of valine and leucine residues are specifically protonated was produced using α-ketoisovalerate as a precursor for amino acid biosynthesis, yielding CH3 isotopomers)32, 33.
We find that the methyl cross peak intensities in the 1H,13C correlation spectra obtained using the protonated sample vary significantly in comparison to the spectra obtained using the α-ketoisovalerate (CH3) labelled sample. A spectrum comprising the full spectral width is provided in the Supporting Information (Fig. S1). In particular, several methyl peaks are broadened or missing in the spectrum from the protonated sample. Using the RAP labelled samples, high resolution Hα,Cα correlation spectra are obtained. The resolution of the spectrum from the 25% RAP labelled sample is only compromised due to the presence of three methyl isotopomers CH3, CH2D, CHD2 which yield isotope induced chemical shift changes34. This problem can be alleviated by using 5% RAP labelled sample. In the protonated sample, the α-carbon spectral region seems to be rather well dispersed at first sight. However, cross peaks significantly overlap in particular in the region of around 4.7/55 ppm proton/carbon chemical shifts.
To quantify the differences between the spectra, we analysed the experimental 1H and 13C line widths and the sensitivity in detail. The results are summarized in Fig. 2. Representative traces for two residues along the proton and carbon dimension of the correlation experiment are shown in Fig. 2A. Comparing the α-ketoisovalerate (CH3) labelled sample and the fully protonated sample, we observe an average gain in resolution on the order of 3 and 2 in the proton and carbon dimensions, respectively (Fig. 2C,D). To account for the difference in the amounts of material in the two rotors (Fig. S2), we quantified the amount of sample by performing 1D-13C experiments using direct 13C excitation. This ratio (~1.76 times) is used to normalize the cross peak intensities represented in Fig. 2B. In total, we find approximately a 2–3x higher sensitivity for the α-ketoisovalerate (CH3) labelled sample. This enhancement could potentially be further increased by an additional factor of 2 by making use of stereospecific precursors for amino acid biosynthesis35.
We attribute differences in peak intensities to the effective intra- and inter-methyl proton-proton dipolar couplings in the sample. To estimate the influence of the proton network around a given proton site, we quantified the spin density, following the convention of Zorin et al.36. In solids, a given spin is affected by many internuclear couplings of different magnitudes. Therefore, d RSS, which is defined as the square root of the sum of squared dipolar couplings, provides a good approximation for an effective coupling experienced by a given spin
In the deuterated protein sample (back-substituted with 100% protons at exchangeable sites), the average proton, proton dipolar couplings for amides are on the order of 7kHz, and thus approximately 3x smaller in comparison to the effective dipolar couplings in the protonated sample. In the deuterated protein sample, high d RSS values are obtained primarily in the turn regions (RT-loop around residue 20, the N-src-loop at residue 40, the distal loop with residues 47–49), for which short amide-amide distances are found. Apparently, a MAS rotation frequency of around 50kHz is sufficient to yield efficient averaging of d RSS for the amide protons in perdeuterated samples. Under these conditions, average 1HN line widths are on the order of 50Hz14.
For the selectively methyl protonated sample, the effective proton, proton dipolar coupling is on the order of 27kHz for methyl protons (Fig. 3). This value is similar in magnitude to the Hα d RSS values for a protonated sample (Figure S3). The 1Hα,13Cα correlation spectrum for a protonated sample at a MAS rotation frequency of 111kHz is reasonably well dispersed with an average proton line width on the order of 100Hz (Figure S3), suggesting that this spinning frequency is sufficient to yield efficient averaging of d RSS in selectively methyl protonated samples. In contrast, the d RSS values for methyl protons in a fully protonated sample are approximately 1.5x larger, indicating that MAS frequencies on the order of 160kHz and above are necessary to yield methyl 1H,13C correlation spectra with maximum resolution. Effective dipolar interactions are apparently not efficiently averaged in a protonated sample at a MAS frequency of 110kHz. This in agreement with a previous study, where it has been suggested that homogeneous interactions require MAS frequencies that are significantly larger (10 times) than the strength of the interaction31.
In order to test whether d RSS is a good descriptor for the proton spin network in MAS solid-state NMR, the calculated effective proton, proton dipolar couplings is represented as a function of the normalized cross peak intensities in Fig. 4. In general, higher Hα peak intensities are found for smaller d RSS values. Proton isotropic chemical shifts are not included in the analysis. n=0 rotational resonance effects may potentially yield line broadening, which can be alleviated at higher magnetic fields. In a typical protein sample, the effective proton, proton dipolar interactions are on the order of 15–45kHz. In that sense, rotation frequencies greater than 300kHz are required in order to efficiently average all homonuclear proton dipolar couplings.
Even though fully protonated samples would be ideal for spectroscopy, a brute-force approach bears risks for structure determination protocols: Resonances of protons, which are involved in strong dipolar interactions, are potentially weak or not observable. These residues do not yield long-range distance restraints (>5Å). If especially long-range distance restraints are missing, structure calculation protocols might potentially converge into a wrong fold.
We demonstrated that at MAS rotation frequencies of 110kHz, the spectral quality obtained from selectively methyl protonated samples is superior in comparison to the spectrum obtained from a fully protonated sample. The α-ketoisovalerate labelled (CH3) deuterated sample yields a gain in resolution on the order of 3 and 2 in the proton and carbon dimension, respectively. Sensitivity is enhanced by approximately a factor of 2–3, which could be further increased by a factor of 2 by making use of stereospecific precursors for amino acid biosynthesis. We believe that in particular experiments involving large non-symmetric protein complexes in which sensitivity is limiting will benefit from such deuteration schemes38, 39. The results presented here have been obtained only for a single microcrystalline protein preparation. In comparison to other samples, the α-spectrin SH3 domain yield very narrow line proton spectra. This allows to monitor subtle spectral changes induced by changes in the labelling scheme. We expect that similar effects will be observable for non-crystalline proteins as well, as the physical basis of dipolar interactions is the same, and the average proton density comparable for all proteins.
The perdeuterated, selectively methyl group protonated microcrystalline SH3 sample was prepared as described previously33. In brief, expression was carried out in 100% D2O M9 medium supplemented with 15N-Ammonium Chloride and D-Glucose-13C, d7. α-ketoisovalerate (2-Keto-3-(methyl-d3)-butyric acid-4-13C, 3-d sodium salt, Sigma Aldrich) was added to the M9 minimal medium 1h prior to induction with 1mM IPTG (at OD600=0.5–0.6). yielding methyl CH3 isotopomers in a deuterated matrix. Subsequent to overnight expression, the SH3 domain was purified via anion exchange and size exclusion chromatography as described before. For crystallization, pure protein was lyophilized and dissolved in 100% D2O (final concentration: 8–10mg/ml). Ammonium sulfate (dissolved in 100% D2O) was added to a final concentration of 100mM and the pH was adjusted to 8.0 by adding NaOD. In order to produce the RAP labelled samples, the deuterated M9 minimal medium was supplemented with H2O (25% H2O and 5% H2O for 25% and 5% RAP samples, respectively)24. The protonated sample was prepared employing only protonated chemicals. The protein samples were not unfolded prior to crystallization, and no paramagnetic dopant was used in the crystallization buffer. The amount of protonation in the samples have been characterised in earlier work18, 40.
The proton detected CP based HSQC experiments for the protonated and α-ketoisovalerate labelled samples were carried out at a static magnetic field of 11.8T (1H Larmor frequency 500MHz), employing a 0.7 mm H/C/N MAS probe. The best Hartmann-Hahn condition was found at a nominal RF amplitudes of 70kHz and 40kHz on the 1H and 13C channel, respectively. A linear ramp on the 1H channel was applied for the 1H−>13C CP step (contact time: 800 μs, range of amplitude: 70–100%) and for 13C−>1H CP (contact time: 500 μs, range of amplitude: 100–70%), respectively. For hard pulses, B1 amplitudes of 156kHz and 100kHz were used for 1H and 13C, respectively. Low power decoupling was applied in the evolution periods using swept TPPM during t1 (ω1H/2π=30kHz)41, and WALTZ16 (ω13C/2π=5kHz) in the direct dimension. Water suppression was achieved using MISSISSIPPI with ω1H/2π=30kHz for 300 ms42. Phase-cycling was performed as described in Barbet-Massin et al.17. For all the experiments, the actual sample temperature was calibrated to 288K using DSS (set temperature: 263K, VT gas flow rate: 480L/h). In all experiments, the acquisition times in the F1 and F2 dimensions were set to 70 ms and 35 ms (13C, 1H), respectively. The relaxation delays were set to 0.5s and 1.5s in the protonated and α-ketoisovalerate labelled sample, respectively.
The spectrum of the 25% RAP labelled sample was recorded at 20T (850MHz) setting the MAS frequency to 40kHz. The spectrum of the 5% RAP sample was recorded at 14.1T (600MHz) with the MAS frequency adjusted to 20kHz. For experiments involving RAP labelled samples, scalar coupling based HMQC type experiments were employed. 2–3kHz of WALTZ16 decoupling was used during proton detection.
We acknowledge support from the Helmholtz-Gemeinschaft and the Deutsche Forschungsgemeinschaft (Grants Re1435 and SFB-1035, project B07). In addition, we are grateful to the Center for Integrated Protein Science Munich (CIPS-M) for financial support. ZT acknowledges funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No 657682.
K.X., R.S. performed experiments and analysed data; S.A. provided NMR data on RAP samples; C.M. prepared SH3 samples; DCRC helped in sample preparation; V.D., S.W. performed experiments employing the 0.7 mm MAS probe; K.X., R.S., Z.T., B.R. performed data interpretation, designed research and experimental plans; All authors contributed in writing the manuscript.
The authors declare that they have no competing interests.
Electronic supplementary material
Supplementary information accompanies this paper at doi:10.1038/s41598-017-07253-1
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Riddhiman Sarkar, Email: firstname.lastname@example.org.
Bernd Reif, Email: ed.mut@fier.