Peptide and Protein Dynamics and Low-Temperature/DNP Magic Angle Spinning NMR

Abstract

In DNP MAS NMR experiments at ~80–110 K, the structurally important −¹³CH₃ and −¹⁵NH₃⁺ signals in MAS spectra of biological samples disappear due to the interference of the molecular motions with the ¹H decoupling. Here we investigate the effect of these dynamic processes on the NMR line shapes and signal intensities in several typical systems: (1) microcrystalline APG, (2) membrane protein bR, (3) amyloid fibrils PI3-SH3, (4) monomeric alanine-CD₃, and (5) the protonated and deuterated dipeptide N-Ac-VL over 78–300 K. In APG, the three-site hopping of the Ala-C_β peak disappears completely at 112 K, concomitant with the attenuation of CP signals from other ¹³C’s and ¹⁵N’s. Similarly, the ¹⁵N signal from Ala-NH₃⁺ disappears at ~ 173 K, concurrent with the attenuation in CP experiments of other ¹⁵N’s as well as ¹³C’s. In bR and PI3-SH3, the methyl groups are attenuated at ~95 K, while all other ¹³C’s remain unaffected. However, both systems exhibit substantial losses of intensity at ~243 K. Finally, with spectra of Ala and N-Ac-VL, we show that it is possible to extract site specific dynamic data from the temperature dependence of the intensity losses. Furthermore, ²H labeling can assist with recovering the spectral intensity. Thus, our study provides insight into the dynamic behavior of biological systems over a wide range of temperatures, and serves as a guide to optimizing the sensitivity and resolution of structural data in low temperature DNP MAS NMR spectra.

Graphical abstract

INTRODUCTION

Magic angle spinning (MAS) NMR spectroscopy is now established as a versatile and essential tool in structural biology.^1–9 In particular, advances in sample preparation, methodology,^10–19 and labeling strategies have dramatically improved the resolution of MAS spectra, thus making possible structural studies of large biomolecules not accessible with other techniques.^20–24 Nevertheless, the primary limiting factor of MAS NMR is its inherently low sensitivity. An approach to circumvent this limitation is operation at cryogenic temperatures, since the Boltzmann population scales as 1/T.^25,26 An even greater gain in sensitivity can be achieved by integrating dynamic nuclear polarization (DNP) into the MAS NMR experiments, where orders of magnitude enhancements of NMR signal intensities have been reported for peptide and protein samples.^27–30 This permits experiments that are otherwise difficult or impossible to perform.^31–35

To take advantage of this increased sensitivity, DNP/NMR experiments at 80–120 K are becoming widely accessible and heavily utilized. However, to date, the tremendous sensitivity gain overshadows the effects of molecular motions present in the ambient and low temperature spectra. For example, in many cases, side chain and backbone resonances of structural importance are absent for reasons that are glibly referred to as “dynamics” but are not clearly delineated or understood. Early ²H NMR studies on model compounds revealed 3-fold hopping (at 10³–10⁶ s⁻¹) by −CD₃ and −ND₃⁺ groups at 130–200 and 200–310 K temperatures, respectively,^36–39 and there have been a few investigations of these 3-fold processes and 2-fold flips of aromatic rings in peptides and proteins at low temperatures.^40–47 However, to the best of our knowledge, none of them provide a detailed description of the effects of these processes on the accompanying loss in signal intensity in MAS spectra, especially in the 80–120 K regime. Furthermore, recent observation of heteronuclear polarization transfer under DNP has been attributed to methyl dynamics in proteins.⁴⁸ Thus, the purpose of the experiments reported here is to provide an overview of the global and site-specific spectral intensity losses as a function of temperature. Thus, we combine data from ¹H−¹³C/¹⁵N cross-polarization (CP), ¹³C Bloch decay MAS experiments recorded from a series of temperature-dependent MAS NMR spectra on five systems including (1) the microcrystalline tripeptide alanyl-prolyl-glycine (APG), (2) the membrane protein bacteriorhodopsin (bR), (3) amyloid fibrils of phosphatydinal-inositol-3-kinase SH3 domain (PI3-SH3), (4) monomeric −CH₃ and −CD₃ alanine, and (5) the protonated and deuterated dipeptide N-acetyl-valyl-leucine (N-Ac-VL) over the temperature range 78–300 K.

In APG, we find that the motion of the −NH₃⁺ and −CH₃ interferes with ¹H decoupling and also compromises CP efficiencies, leading to specific and complete attenuation of spectral lines from these two groups at ~173 and ~112 K, respectively. In addition, these intensity losses propagate throughout the sample, causing a global loss of spectral intensity. At temperatures around 80 K, the interfering motions approach the rigid lattice limit and the signal intensity fully recovers. In this regime, the gain in signal intensity is purely due to the Boltzmann factor.

Two classes of larger protein systems commonly studied at low temperature and with DNP include membrane and amyloid proteins.^49–52 Examples of these systems include bR in its native purple membrane and PI3-SH3. In both of these cases, the signal minimum due to the dynamics of −CH₃ groups occurs at a lower temperature (95 K) and appears to be localized as opposed to the case of APG. In addition, other spectral regions such as the carbonyl and aromatic resonances are not affected and exhibit improved signals. Finally, we show that intensity losses due to the dynamic process can be partially and in some cases completely recovered by the introduction of ²H labeling. In particular, the introduction of −CD₃ groups in Ala, N-Ac-VL, and the use of perdeuterated bR permit observation of these groups in cases where spectral lines from the −CH₃ moiety are completely absent.

EXPERIMENTAL SECTION

Sample Preparation

Both uniformly ¹³C, ¹⁵N-labeled and ¹⁵N-labeled APG samples were diluted to 10% with the corresponding natural abundance APG to suppress any intermolecular couplings. This was accomplished by dissolving the mixture of labeled APG/natural abundance APG in a minimal amount of water (~45 mg/mL) followed by slow crystallization in a desiccator, and crystals forming in about a week. 40 mg of each APG sample was packed into a 4 mm Revolution NMR zirconia rotor.

Bacteriorhodopsin (bR), in its native purple membrane, was purified from Halobacterium salinarum grown in uniformly ¹³C, ¹⁵N-labeled peptone medium.⁵³ Peptone was obtained from the anaerobic acid hydrolysis of Methylophilus methylotrophus cells grown on ¹³C-labeled methanol and ¹⁵N-labeled ammonium sulfate.⁵⁴ The purple membranes were isolated using the method of Oesterhelt and Stoechenius.⁵⁵ The sample was washed three times with 300 mM guanidine hydrochloride at pH 10.0. The sample was pelleted after every wash by centrifugation for 2 h at ~43,000g. The washed pellet was mixed with 5 mM AMUPol⁵⁶ or 15 mM TOTAPOL⁵⁷ in “DNP juice” consisting of d₈-glycerol/D₂O/H₂O (60/30/10 volume ratio) and centrifuged once more.

The phosphatidyl-inositol-3-kinase SH3 domain PI3-SH3 fibril sample was uniformly ¹³C, ¹⁵N-labeled at the F, V, Y, and L residues. The fibrils were grown from a solution of monomeric protein by incubation at pH 2.0 and 25 °C for 14 days.⁵⁸ “DNP juice” adjusted to pH 2.0 and supplemented with 15 mM TOTAPOL was added to the gel-like fibrils for cryoprotection and DNP experiments. Both bR and PI3-SH3 samples were packed into a 4 mm sapphire rotor for DNP/ NMR experiments.

[U−¹³C, ¹⁵N] and [−CD₃−U−¹³C, ¹⁵N] alanine were purchased from Cambridge Isotope Laboratories (CIL), dissolved in water, and then crystallized in a desiccator. [U−¹³C, ¹⁵N] N-acetyl-l-Val-l-Leu (N-Ac-VL) and [1,2-¹³C] acetic anhydride were purchased from CIL. N-Ac-VL was synthesized by New England Peptide (Gardner, MA), using standard solid-phase methods, and purified by HPLC. N-Ac-VL was crystallized from a 1:1 (v/v) H₂O:acetone solution.

NMR Spectroscopy

MAS spectra were recorded using a custom-designed triple resonance (¹H, ¹³C, ¹⁵N) cryogenic MAS probe equipped with a sample exchange system⁵⁹ on a home-built NMR spectrometer operating at 380 MHz ¹H frequency (courtesy of Dr. D. J. Ruben). Two types of 1D NMR experiments were conducted for APG: ¹H−¹³C/ ¹⁵N cross-polarization (CP) and ¹³C Bloch decay. For CP experiments, a spin-lock field of 50 kHz was employed on the proton channel. Experimental parameters including recycle delay, CP Hartmann-Hahn matching conditions, CP duration, and TPPM decoupling have all been optimized for all temperature dependent experiments. ¹H TPPM decoupling fields of ω_1H/2π = 83 kHz or 100 kHz and ω_r/2π = 4.83 kHz were used unless stated otherwise.

Liquid nitrogen boil-off gas was used for both the bearing and drive gas streams, and the spinning frequency was controlled by a Bruker MAS controller.⁵⁹ Both bearing and drive streams were cooled using a custom-designed heat exchanger, and the temperature was subsequently controlled using heating elements inside vacuum jacketed transfer lines with two PID controllers (Lakeshore, Westerville, OH). The sample temperature was monitored using a fiber optic temperature sensor (Neoptix, Quebec, Canada) that extends to the inside of the MAS stator. The fiber optic thermometers were calibrated by immersion in liquid nitrogen at 77 K. Note, during MAS experiments, we have consistently recorded temperatures in the range 72–77 K presumably due to Joule-Thompson cooling on the expansion of the N₂ gas from the jets of the drive cup and bearings of the stator.

DNP Experiments

In order to investigate methyl group dynamics at DNP temperatures, ¹H, ¹³C, ¹⁵N CP and homonuclear experiments were performed using a home-built DNP gyrotron instrument operating at 250 GHz/380 MHz with ~14W of microwave power. ^60,61 Enhancement factors (ε) were calculated by comparing the signals obtained with and without µw irradiation.

Simulations

The simulations of the Ala-CH₃ group dynamics in APG were performed using GAMMA⁶² with 100 powder orientations chosen using the ZCW scheme.⁶³ Details of the parameters used can be found in the Supporting Information. In order to simulate the three-site hopping mechanism, the simulations were carried out in a composite Liouvillian space that facilitates a mutual-exchange mechanism. The dimensions of the exchange matrices for the four-spin CH₃ and CD₃ spin systems are 256 × 256 and 2916 × 2916, respectively. The amount of time required to simulate the FID of one crystallite orientation using one CPU core is ~1 min and ~1.5 days for CH₃ and CD₃, respectively. All simulations took ~2 weeks to compute using the ETH Brutus cluster with 384 CPU cores. The parameters chosen for the simulations are ω_r/2π = 4.651 kHz, TPPM decoupling ω_1H/2π = 83 kHz with 6.9 µs pulses, and phases ±15 degree. The size of the quadrupole coupling used for ²H nuclei is 167 kHz.

RESULTS

The 1D ¹³C spectra in Figure 1 illustrate the spectral resolution of two microcrystalline APG samples with differing isotopic labeling schemes at 80 K. In both cases, the spectra exhibit a resolution comparable to that obtained at room temperature. Specifically, Figure 1a shows a ¹³C CP MAS NMR spectrum of [U−¹²C, ¹⁵N]-APG and in the absence of ¹³C−¹³C J-couplings; the line width of the Gly-C_o is as narrow as 27 Hz (0.28 ppm). In comparison, the line widths in the spectrum of [U−¹³C, ¹⁵N]-APG (Figure 1b) increase by a factor of 2 or more primarily due to one-bond J-couplings and higher order cross-terms that arise from the denser ¹³C network. Nevertheless, resolved J-splittings (~50 Hz) can still be distinguished in two out of three carbonyls, each of which has only one J-coupled neighboring ¹³C. On the other hand, each ¹³C in the aliphatic region is J-coupled to multiple neighbors, which obscures the splitting. The aliphatic carbons also have stronger CH dipolar coupling than that of carbonyl carbons, and would require stronger decoupling and higher spinning frequencies to be averaged out effectively. The data confirm that the optimal resolution in MAS spectra can be obtained with sparsely labeled ¹³C samples.⁶⁴ In addition, they demonstrate that high resolution in well-ordered materials can be obtained at cryogenic temperatures (~80 K).

Figure 1.

¹³C CP-MAS spectrum of [U−¹⁵N] APG (a) and [U−¹³C, ¹⁵N] APG (b) at 80 K. Both samples were diluted to 10% in unlabeled APG to minimize the intermolecular couplings. The line width of the Gly-C_o is 27 Hz (0.28 ppm). In part b, the line width is broadened by ¹³C−¹³C J-coupling evident in the doublet splitting in the carbonyl region with the Gly-C_o and Ala-C_o showing one-bond C_α−C_o J-couplings of ~50 Hz. The fact that these resonances are not resolved in the aliphatic region is likely due to the presence of multiple J-couplings.

Figure 2 illustrates the temperature dependence of the ¹H−13C/¹⁵N CP-MAS spectra acquired in the range from 73 to 295 K, where we observe several interesting spectral changes. First, in the transition between regions I and II at 225 K, there is a doubling in all of the side chain signals that is especially obvious on the Ala-C_β line (δ ≈ 18 ppm at 295 K), as well as Pro-C_β and Pro-C_γ (30 and 35 ppm at 295 K, respectively). A recent calorimetric study of APG crystals revealed that the spectral changes at 225 K, although in the vicinity of the famous protein glass transition,⁶⁵ are likely the result of a polymorphic phase transition.⁶⁶ Similar spectral changes at ~200 K were observed for the peptide N-f-MLF-OH.³² Second, in region III, the intensity of Ala-C_β exhibits a local minimum that coincides with the disappearance of the ¹⁵N signal from the −NH₃⁺ group at ~173 K. Third, at lower temperatures, in region IV between 148 and 96 K, a more significant loss of Ala-¹³C_β signal intensity occurs. The intensity decreases dramatically and the Ala-¹³C_β line disappears into the baseline at ~112 K, and there is a concurrent loss of spectral intensity in the ¹⁵N spectra at this temperature. We attribute the two signal minima in regions III and IV at ~173 and ~112 K to the 3-fold hopping rates of the −NH₃⁺ and −CH₃ groups, respectively, matching the ¹H decoupling frequency. Finally, at lower temperatures, the hopping rate enters the slow exchange limit and the −CH₃ and −NH₃⁺ lines reappear and narrow and the intensity recovers fully at 73 K.

Figure 2.

Temperature-dependent ¹H−¹³C CP (left column) and ¹H−¹⁵N CP (right column) spectra of [U−¹³C, ¹⁵N] APG. Several important spectral changes are observed. In region III, between 213 and 148 K, the intensity of the Ala-NH₃⁺ peak is buried under the noise level at ~173 K, which couples with the partial attenuation of the ¹³C spectrum. Region IV, below 148 K, exhibits the disappearance of the Ala-CH₃ signal at ~113 K, coincident with the signal dip in the ¹⁵N spectrum. The spectral changes are further analyzed in Figure 3. The spectra were acquired with ω_r/2π = 4.83 kHz, ω_1H/2π = 83 kHz for TPPM decoupling.

The temperature dependence of the signal intensity is a product of the Boltzmann factor, which has a T⁻¹ dependence, and the intensity loss due to molecular motions, which varies with temperature. In order to isolate the effect of molecular dynamics, the contribution from the Boltzmann factor is removed by multiplying the integrated signal intensity by the corresponding temperature. Figure 3a shows the normalized signal intensity of Ala-C_β (−CH₃ group) with (red solid circles) and without (blue open circles) the correction for the Boltzmann factor. To correct for the Boltzmann factor, the normalized intensity was calculated as (I/T)/(I₀/T₀), where I₀ is the intensity at the highest temperature T₀, where experiments were performed and I is the intensity at temperature T. To facilitate data presentation, without the correction for the Boltzmann factor, the normalized intensity was calculated differently as (I/I₀), where I₀ is the intensity at the lowest temperature at which we perform experiments. Note that the integrated signal intensities were adjusted when necessary by their relative ¹H or ¹³C T₁’s measured at each temperature and incorporated into the intensity calculations. T_1H and T_13C values are plotted as a function of temperature in Figure S1 in the Supporting Information. The minima at ~112 and ~173 K in Figure 3 are visible even without this correction and are greatly amplified with the correction. Below 90 K and above 225 K, the adjusted signals in red are essentially the same, indicating that the signal intensities are unaffected by the effect of molecular dynamics in these regions. In particular, the signal intensity in these regions follows closely the T⁻¹ dependence, resulting in a factor of 4 times higher intensity at 73 K compared to 295 K. Figure 3b illustrates that other sites such as Pro-C_α and Gly-C_o also exhibit two signal minima in ¹H−¹³C CP experiments and the entire spectrum is uniformly attenuated at 112 and 173 K. Figure 3c presents the data from the single-pulse ¹³C Bloch decay signals detected with ¹H decoupling that show a minimum at 112 K, and a set of ¹³C Bloch decay spectra is included in Figure S2. Furthermore, at this temperature, the Ala-CH₃ is also completely attenuated (Figure 3a), whereas other sites are nearly unaffected. In combination, the data in Figure 3b and c suggest that the signal minimum at ~112 K is associated with the 3-fold hopping rate of the Ala-CH₃, which approximates to the Rabi frequency of the ¹H RF fields during decoupling and/or CP. The minimum at ~173 K is attributed to the hopping of the Ala-NH₃⁺ which manifests itself in the disappearance of its ¹H−¹⁵N CP signal, as shown in Figure 3d. In comparison to the ¹H−¹³C CP data (Figure 3b), the ¹H−¹⁵N CP data also exhibit two minima but in reverse intensity order: nearly uniform signal attenuation at ~112 K and complete signal attenuation of Ala-NH₃⁺ at ~173 K.

Figure 3.

Temperature dependence of the integrated peak intensities of [U−¹³C, ¹⁵N]-APG. The vertical dashed lines indicate the minima in the signal intensities at ~112 and ~173 K due to the hopping of −CH₃ and −NH₃⁺, respectively. (a) Ala-C_β from ¹H−¹³C CP with (red solid circles) and without (blue open circles) Boltzmann correction at each temperature. (b) Adjusted spectrum intensities of several peaks of APG plotted as a function of temperature from ¹H−¹³C CP. Again, notice the two minima, ~112 K where the Ala-C_β peak broadens beyond detection and ~173 K. (c) ¹³C Bloch decay experiments show the first minimum, indicating interference between methyl hopping and ¹H decoupling. (d) Spectral intensities of ¹H−¹⁵N CP alanine NH₃⁺ as a function of temperature. The second minimum in part b is due to the hopping of the Ala-NH₃⁺ group.

Motivated by the dramatic changes in signal intensities in the APG spectra, we extended our studies to the membrane protein bR, and to amyloid fibrils formed by PI3-SH3, to determine if a similar behavior is observed in these systems. Figure 4a illustrates that attenuation of methyl ¹³C resonances occurs in [U−¹³C, ¹⁵N]-bR. However, because of spectral overlap in this 248 amino acid, uniformly ¹³C/¹⁵N labeled protein, the extraction of accurate intensity data for these resonances is imprecise. Multidimensional experiments at higher fields are required to obtain better resolution. In contrast, the spectra in Figure 4c of [¹³C, ¹⁵N-FVYL]-PI3-SH3 are greatly simplified, which permits the extraction of the intensity of different functional groups, especially the Val-CH₃ and Leu-CH₃. In general, the spectra show that the overall intensity of carbonyl and aromatic ¹³C’s in both samples increases as temperature decreases except around 243 K (Figure 4b and d). In addition, the methyl groups exhibit another minimum at ~95 K in the case of [¹³C, ¹⁵N-FVYL]-PI3-SH3, but the intensities of the two −CH₃ containing residues are partially recovered at 87 K and are expected to fully recover at lower temperatures. Incorporation of DNP into the experiments in this temperature regime will, in addition to the Boltzmann factor, boost the sensitivity. In Figure S5, we show 1D ¹H−¹³C CP spectra of [¹³C, ¹⁵N-FVYL]-PI3-SH3 obtained with a signal enhancement of 35. A 2D RFDR spectrum of [¹³C, ¹⁵N-FVYL]-PI3-SH3 fibril with τ_mix = 1.6 ms was acquired in approximately 4 h. Without DNP, the same spectrum would require one month to achieve the same S/N.

Figure 4.

Temperature-dependent ¹H−¹³C CP spectra of (a) [U−¹³C, ¹⁵N] bacteriorhodopsin and (c) [¹³C, ¹⁵N-FVYL]-PI3-SH3 amyloid fibrils without Boltzmann correction. In parts b and d, we show plots of normalized intensities for carbonyl, aromatic, and aliphatic regions in bR and PI3-SH3 fibrils as a function of temperature. Both the carbonyl and aromatic regions experience a minimum in intensity around 243 K followed by a steady increase as the temperature is decreased for both systems. The intensity of −CH₃’s in the valines and leucines in PI3-SH3 also reveals a minimum at ~243 K, followed by a second minimum at ~95 K, where the intensity losses are again due to −CH₃ hopping interfering with ¹H decoupling and spin-lock fields. Both samples were cryoprotected in d₈-glycerol/D₂O/H₂O (60/30/10 volume ratio). Full sets of spectra at different temperatures are provided in Figures S3 and S4. The spectra were acquired with ω_r/2π = 4.83 kHz for bR and 7 kHz for PI3-SH3. In both cases, ω_1H/2π = 83 kHz for TPPM decoupling.

In Figure 4b and d, the intensities are normalized to those obtained at 275 K for bR and 283 K for PI3-SH3. We note that the normalized intensities of the aromatic ¹³C’s at ~80–90 K can be larger than the contribution from the Boltzmann factor. This is due to the fact that at 300 K there is already a significant loss in signal intensity in aromatic ring spectra due to the 2-fold ring flips,^67–70 which affects the polarization transfer during CP and interferes with decoupling. This phenomenon was observed in N-f-MLF-OH where the δ, δ′ and ε, ε′ signals from the Phe ring are absent in the 300 K 1D spectra. They start to reappear at 250–225 K and are fully developed at lower temperatures.³² In addition, in ZF-TEDOR spectra of PI3-SH3, the aromatic region is essentially empty at 300 K but is intense and well resolved at 90 K.⁷¹ Thus, the broad signal minimum around 243 K probably extends to the higher temperatures, ~280 K. Note also that protein samples used for DNP are cryoprotected by the glass-forming mixture of glycerol/water (60/40 volume ratio), as is the case for the bR and PI3-SH3 samples used in our study. Therefore, the signal minimum at 243 K coincides with the freezing of the glycerol/water mixture and a slowing of the 2-fold flips of the 24 aromatic rings in bR. We also note that this temperature is close to that of the protein glass transition, although it is generally centered at the somewhat lower temperature (~200 K).

It would clearly be desirable to recover the signal loss due to the interference between molecular dynamics and decoupling, and accordingly, we have examined the possibility of labeling methyl groups as −CD₃’s.^36,37 The rationale behind this approach is that the first-order ²H quadrupole coupling is inhomogeneous, the second-order ²H coupling is small at high fields, and ²H−²H dipole couplings, proportional to γ_I², are a factor of 42 smaller than ¹H−¹H dipole couplings. Thus, MAS itself should average the ²H−²H and ²H−¹³C dipolar couplings and attenuate the intensity losses. Figure 5 shows temperature-dependent spectra obtained from the monomeric amino acid alanine containing either a −CH₃ or −CD₃ group. As expected, the ¹³C MAS spectra obtained from Ala-¹³CH₃ exhibit dramatic intensity losses around 165 K. Note that the intensity also decreases uniformly across the spectrum and both the C_α and C_o resonances are effectively suppressed, as was the case in APG (Figure 2). However, with the −¹³CD₃ present, the Ala methyl line is not suppressed at 165 K and the intensity recovers at 81 K. Nevertheless, there is still considerable intensity loss at 165 K, and as discussed below, this is probably due to the −CD₃ methyl hopping rate (~10⁴–10⁵ s⁻¹) being similar to the magic angle spinning frequency. By interpolating the data from the Arrhenius plot reported by Beshah et al. for Ala-CD₃, we obtain hopping rates of 1.4 × 10⁵ at 165 K and 1.6 × 10³ at 127 K,³⁸ which confirms our hypothesis.

Figure 5.

Temperature dependent ¹³C MAS spectra of (a) [U−¹³C, ¹⁵N]-Ala with a ¹³CH₃; (b) Ala-¹³CD₃; (c) [U−¹H, ¹³C, ¹⁵N] N-acetyl-l-Val-l-Leu; and (d) [U−²H, ¹³C, ¹⁵N] N-acetyl-l-Val-l-Leu. The spectra were acquired with ω_r/2π = 6.2 kHz, ω_1H/2π = 100 kHz for TPPM decoupling.

To explore the intensity losses in two other amino acids, namely, Leu and Val, we recorded spectra of protonated and −CD₃ labeled N-Ac-VL. As shown in Figure 5c, the Val γ₁ and γ₂ lines in protonated N-Ac-VL exhibit significant intensity losses and are absent in the spectrum for T < 116 K. Note also that the line from V^γ2 disappears in the interval 116–173 K, whereas the V^γ1 line persists to ~116 K. This behavior is consistent with the fact that the 3-fold hopping rates, measured with ²H spectra, for the two −CH₃ groups in Val differ by about an order of magnitude.³⁸ Specifically, in N-Ac-dl-Val, they are 4.8 × 10⁵ and 2.8 × 10⁶ at 118 K, and therefore, the signal intensities of the methyl groups at 116 K (Figure 5d) are not severely attenuated and still detectable, as the hopping rates are ~1–2 orders of magnitude away from the interference regime (~10⁴–10⁵ s⁻¹). In contrast, the Leu-CH₃ line does not lose intensity even at 90 K and it can potentially be used for distance measurements at low temperatures. In deuterated N-Ac-VL, both groups are present at 90 K but with reduced intensity.

In DNP enhanced spectra of uniformly labeled bR, the intensity of methyl containing residues is also reduced. Figure 6a shows a 2D RFDR spectrum of ¹H uniformly labeled bR at 190 K with cross-peaks corresponding to the 29 Ala and 18 Thr-CH₃’s in bR that are not resolved at 380 MHz. In contrast, at 90 K (Figure 6b), the Ala C_βα cross-peaks are no longer observed in the H bR sample, while they are detected in the ²H uniformly labeled bR. Similarly, Thr_γ2β cross-peaks are partially attenuated in the ¹H spectrum compared to that at 190 K, but it is fully recovered in the spectrum of ²H bR This further verifies that an effective solution of methyl group attenuation could involve deuteration.

Figure 6.

2D DNP enhanced ¹³C−¹³C RFDR spectra of (a) U−¹H, ¹³C, ¹⁵N bR doped with 5 mM AMUPol at 190 K, (b) 92 K and (c) 2D RFDR of U−²H, ¹³C, ¹⁵N bR containing 15 mM TOTAPOL at 92 K. Parts b and c were acquired with DNP microwave irradiation. Enhancements of 75 and 71 were obtained, respectively. All spectra were acquired with 2 ms mixing, 30 kHz ¹³C pulses, and 100 kHz ¹H decoupling and spinning frequency ω_r/2π = 7 kHz at ω_0H/2π = 380 MHz. Both experiments took 7 h with 6 s recycle delay, 128 t₁ increments, and 32 scans per increment.

Although the loss of signal intensity of the methyl groups at low temperature impedes many NMR experiments, including distance measurements, it encodes useful site-specific information about the dynamics of a group, for instance, the activation energy, E_a. The activation energy contains rich details about the local chemical and structural environment, and they can be extracted by first comparing experimental data at different temperatures with numerical simulations. Figure 7 shows the Arrhenius plot of the −CH₃ group hopping rates of alanine extracted from simulations of the signal intensities in APG measured by observing ¹³C signals directly using a Bloch decay in the presence of ¹H decoupling (data from Figure 3c). The least-squares fit yields an activation energy E_a of 7.2 ± 1 kJ/mol. This value is lower than the literature value of 20.0 kJ/mol³⁹ obtained for the monomeric amino acid alanine. The activation energy of the −CH₃ group in alanine is higher than that in the APG probably due to tighter crystal packing and thus more restricted rotation, hence the higher barrier for 3-fold hopping. However, this analysis is complicated by the fact that the methyl group is not strictly an isolated system, i.e., it is coupled to the nearby proton bath, especially the −NH₃⁺ group via spin diffusion. A more detailed study of this effect on the activation energy can be performed by deuteration of the NH₃⁺ group.

Figure 7.

Arrhenius plot of the three-site hopping rate of the Ala −CH₃ group in APG. The least-squares fit (red line) yields an activation energy E_a of 7.2 ± 1 kJ/mol and a pre-exponential constant A of ~2 × 10⁹ s⁻¹.

DISCUSSION

It is well-known that the hopping rates of −CH₃ and −NH₃⁺ groups can be measured precisely by analyzing the ²H line shape of the deuterated analogues of these groups.^36–39 Alternatively, Long et al. showed a strong correlation between the intensity of Bloch decay signals of ¹⁵N and the hopping rate in −NH₃⁺.³⁷ Using the same approach, we acquired ¹³C Bloch decay signals (Figure 3c) and assigned the signal minimum at ~112 K to the interference between the methyl group hopping and the ¹H decoupling. Furthermore, the small signal attenuation (10%) of Pro-C_α and Gly-C_o from Bloch decay experiments (Figure 3c) cannot account for the large signal loss (60%) of the same resonances from CP experiments (Figure 3b), suggesting that the methyl group hopping also causes inefficient CP at this temperature. This suggests that the hopping rate of the methyl group in APG at ~112 K is ~10⁴–10⁵ s⁻¹, i.e., the same order of magnitude as the MAS frequency and/or ¹H decoupling regime. The hopping rate can be extracted from the fitted Arrhenius plot (Figure 7) and we obtained a hopping rate of ~7 × 10⁵ s⁻¹, thus confirming our hypothesis.

In practice, the decoupling field and the spin-locking field are very close in strength. Thus, it is expected that interferences between these Rabi fields and the three-site hopping occur at the same temperature. Interference between the spin-locking field and the molecular dynamics is clearly the dominant mechanism responsible for the signal minimum at ~173 K in the ¹H−¹³C CP data (Figure 3b), which does not appear in the ¹³C Bloch decay data (Figure 3c). At this matching condition, the spin-lock is inefficient for methyl ¹H’s, causing a short ¹H T_1ρ. Furthermore, due to rapid ¹H−¹H spin diffusion and the combination of intra- and intermolecular contacts, the effect of a short T_1ρ of the methyl ¹H’s readily distributes itself throughout the molecule, resulting in a uniform signal loss at ~173 K in the ¹H−¹³C CP data (Figure 3b). In contrast, the destructive interference effect on the decoupling appears to be more localized to the −CH₃ and −NH₃⁺. This is apparent in the signal minimum at ~112 K in Figure 3b and c as well as in the signal loss at ~173 K in the ¹H−¹⁵N CP data (Figure 3d), which is due to interference of −NH₃⁺ hopping with both the decoupling and spin-lock fields.

The diffusive hopping of −CH₃ causes the first minimum at ~112 K, and the second at ~173 K is also caused by a similar phenomenon involving the −NH₃⁺ group. The temperature or hopping rate at which the signal intensity is minimum is governed by both the E_a and the pre-exponential factor of the Arrhenius equation, A. However, if the values of A are comparable, then by comparing the previously published E_a values one can predict the temperature range at which the intensity is minimum. The higher activation energy of the −NH₃⁺ group is due to its ability to form hydrogen bonds.³⁷ For example, the E_a in Ala for −ND₃⁺ is 40.5 kJ/mol,³⁷ whereas that for the −CD₃ group is 20.0 kJ/mol.³⁹ Long has shown the E_a values of CD₃ and CH₃ groups are comparable. Similar to the case of −CH₃, Long et al. shows that the hopping rate of −NH₃⁺ in Ala reaches 5 × 10⁴ s⁻¹ at 243 K, which is significantly higher than 173 K implied by our data, pointing to a lower activation energy in APG, given that the values of A are comparable in both cases.

Our hypothesis on the correlation between molecular packing/flexibility and E_a is further supported by the data on larger systems including the membrane protein bR (Figure 4a and c) and amyloid fibrils of PI3-SH3 (Figure 4c and d). In contrast to the tripeptide APG, in bR and PI3-SH3, the hopping effect appears to be localized to the methyl groups. In [FVYL-¹³C, ¹⁵N] PI3-SH3, a sample in which only methyl groups of Val and Leu are labeled, we are able to observe the minimum in the intensity of the −CH₃ groups at ~95 K, compared to ~112 K in APG, suggesting a lower E_a.

Our observations have a direct implication for the application of DNP to problems in structural biology. Carbonyl ¹³C’s, together with the aromatic side chains, are the least affected by the molecular dynamics and exhibit excellent sensitivity at low temperatures. This validates the approach using low temperature DNP to obtain long-range intermolecular distances involving aromatic side chains.⁷¹ Similarly, methyl groups are of proven importance in measuring long-range contacts in proteins due to their position at the termini of many amino acid side chains. Interestingly, as demonstrated by the data here, one cannot use DNP experiments in the 80–120 K temperature range to measure distances associated with certain protonated (−CH₃) methyl groups. As hinted by the partial recovery of the methyl carbons at 87 K in Figures 3 and 4, the brute force solution to this problem is to perform experiments at even lower temperatures, which would then require cooling using liquid helium. A promising alternative approach is to fully or partially deuterate the methyl groups, and it was demonstrated some time ago that full deuteration largely prevents the intensity losses observed here.³⁶ Thus, the benefit of this approach is 2-fold. First, it circumvents the detrimental intensity losses while maintaining sufficient CP from adjacent protons. Second, deuteration of proteins has been demonstrated to increase DNP enhancements by a factor of 3–4.³¹

We have initiated the investigation of such an approach. In particular, in Figure 5, we compared spectra of alanine with −CD₃ and −CH₃ methyl groups as well as between protonated and fully deuterated N-acetyl-l-Val-l-Leu. The methyl group −CH₃ of alanine disappears completely at 165 K, while the alanine −CD₃ signal experiences some level of attenuation but nevertheless survives. In N-Ac-VL, the Val C_γ1 and C_γ2 possess different activation energies, resulting in their disappearance at different temperatures, 90 and 116 K, respectively. This behavior is delayed in the perdeuterated N-Ac-VL spectra, and more importantly, valine C_γ1 and C_γ2 signals can still be observed at 90 K. These two carbons in N-Ac-Val were reported to have distinct E_a’s of 15.3 and 22.2 kJ/mol.³⁸ Our results suggest that it is possible to maintain the signals from methyl groups at low temperature by deuteration. Further investigations including the incorporation of deuterium decoupling are underway and will be the topic of the subsequent studies.

It is worth noting that the signal loss at ~243 K can be as large as ~70% in the case of bR and occurs in the neighborhood of, but above, the temperature normally associated with the protein glass transition. It appears that two mechanisms could be responsible for this effect. First, this intensity loss could be due to the freezing of glycerol/water mixture. Thus, the signal attenuation is related to the hydration of the sample and protein–water interactions and is, therefore, ubiquitous in biological samples. Second, the loss could also be due to the two-site flipping of phenyl rings. bR contains 13 Phe’s and 11 Tyr’s in the amino acid sequence and preliminary ²H NMR data show intermediate exchange flipping rates at this temperature. We believe this is important, since protein samples are frequently studied at temperatures slightly lower than room temperature to slow some dynamic processes. Our results suggest that this approach may be suboptimal in terms of the overall sensitivity. It is therefore desirable to perform such experiments at even lower temperatures, but to date, these temperatures are not always achievable due to instrumental limitations.

Finally, the main focus of this Article is to address the effect of temperature on the line intensity of the −CH₃ group, along with solutions to alleviate the line-broadening effect at certain temperatures due to interference between the hopping mechanism with MAS frequency and/or ¹H decoupling and/ or CP. While such interference is problematic for some experiments, this temperature-dependent phenomena can be exploited and used to extract useful dynamic information about the system. Thus, we present here a new method to extract the activation energy of the −CH₃ group (Figure 7) by monitoring the change of ¹³C signal intensities as a function of temperature. We are currently applying this novel approach to investigate the site-specific motions of other biological macromolecules like proteins, which could reveal important relations between the dynamics and functions of proteins.

CONCLUSION

In summary, we report effects of molecular motions on NMR signals of peptide and protein samples at cryogenic temperatures. In the microcrystalline tripeptide APG, a first-order polymorphic phase transition occurs at 225 K, which manifests itself in the line doubling in ¹³C NMR spectra. At lower temperatures, we observe a destructive interference effect from the 3-fold jump diffusion of the Ala-CH₃ (at ~112 K) and of the Ala-NH₃⁺ (at ~173 K) on the proton decoupling and/or the CP spin-lock fields. The effect on the decoupling appears to be localized, whereas the effect on the CP is readily transmitted throughout the molecule due to fast ¹H−¹H spin diffusion in a strongly coupled ¹H bath.

We extend these experiments to larger biological systems consisting of the membrane protein bacteriorhodopsin and amyloid fibrils formed from PI3-SH3. At ~243 K, both samples exhibit significant signal loss, in the neighborhood of the protein glass transition. Similar to the case of APG, the methyl groups are attenuated at low temperatures. However, the effect does not propagate in an obvious way to other parts of the spectrum, which supports the approach of using low temperature DNP to obtain long-range distances involving aromatic side chains. Furthermore, the intensity minima occur at lower temperatures in these biological systems than in APG (95 K vs 112 K). Simulations of the intensity losses for the −CH₃ group in APG suggest that it should be possible to use similar data to extract site-specific information on molecular dynamics. Finally, our study suggests deuteration of the methyl groups as −CD₃ as another probe of long-range distance constraints using DNP at liquid nitrogen temperatures.

ABBREVIATIONS

MAS

magic angle spinning

NMR

nuclear magnetic resonance

ssNMR

solid state NMR

CP

cross-polarization

DNP

dynamic nuclear polarization

RF

radio frequency

RFDR

radio-frequency-driven recoupling

S/N

signal-to-noise ratio

APG

alanyl-prolyl-glycine

bR

bacteriorhodopsin

PI3-SH3

phosphatydinal-inositol-3-kinase SH3 domain