Residue-resolved monitoring of protein hyperpolarization at sub-second time resolution

The study addresses the challenge of observing protein processes with high structural (residue-level) and temporal resolution in solution by NMR. Conventional NMR requires signal averaging due to intrinsically weak signals, limiting time-resolved measurements and typically emphasizing equilibrium conditions. Hyperpolarization, especially dissolution dynamic nuclear polarization (d-DNP), can enhance NMR signals by orders of magnitude and enable millisecond-to-second time resolution. However, applying d-DNP to biomolecular NMR often sacrifices residue resolution because rapid, multidimensional acquisition is incompatible with the transient lifetime of hyperpolarized states. The authors propose using hyperpolarized water to selectively enhance signals of solvent-exposed residues and combining this with selective detection to achieve residue-resolved, real-time monitoring under near-physiological conditions, thereby addressing both the limitation of non-physiological experimental conditions and the need for sub-second time resolution.

Prior work has shown that d-DNP can significantly enhance NMR signals for biomolecules, including proteins and nucleic acids, enabling time-resolved and sensitivity-enhanced experiments. However, most biomolecular d-DNP applications faced trade-offs between time resolution and residue resolution: rapid 1D detection is possible, but 2D/3D residue-resolved spectra typically take too long for time-series acquisition within the hyperpolarization lifetime. Earlier studies established polarization transfer from hyperpolarized water to proteins via chemical exchange and solvent-relayed NOEs, with residue-specific enhancements depending on exchange rates and solvent exposure. Water-selective NOESY (WS-NOESY) has been used to quantify solvent-to-protein polarization transfer at residue resolution. Additional developments include ultrafast 2D NMR for real-time monitoring of reactions, and improvements in water hyperpolarization and dissolution protocols. Collectively, these works motivate a strategy that leverages hyperpolarized solvents and selective detection to regain residue resolution while preserving time resolution.

The approach combines hyperpolarized buffers produced by dissolution d-DNP with selective detection to sparsify spectra and enable rapid, residue-resolved readout. Hyperpolarization and mixing: Water protons were hyperpolarized ex situ at cryogenic temperature and high magnetic field, then rapidly dissolved and transferred for in situ mixing with the protein in a high-field NMR spectrometer. A representative workflow included: (1) hyperpolarization at T_DNP = 1.2 K and B0,DNP = 6.7 T using off-resonance microwave irradiation (~187.7–188.4 GHz) on a 15 mM TEMPOL solution (180 µL, 0.85:0.15 v/v H2O:glycerol) vitrified in liquid helium; microwave modulation at 1 kHz over 100 MHz; hyperpolarization build-up time ~3 h (τ_build-up^−1 ≈ (2.1 ± 0.1)×10^−4 s^−1). (2) Dissolution with 5 mL D2O at 180 °C (453 K) under 1.05 MPa and rapid transfer (≈1.5 s) through a 0.9 T magnetic tunnel over ~4 m using He gas at 0.7 MPa into a detection spectrometer at B0,NMR = 18.8 T (800 MHz 1H) and T_NMR = 310 K (37 °C). The hyperpolarized water was mixed in situ with 150 µL of the target protein solution; delay for mixing and turbulence settling was ~3 s. After dissolution, H2O:D2O ≈ 0.03:1. Protein and buffer: 15N-labeled ubiquitin (Ubq) expressed in E. coli BL21-(DE3)-pLysS, purified by ion exchange and gel filtration, concentrated to 1 mM in physiological saline at pH 7.4. Detection schemes: Time series of 15N-edited 1H 1D spectra were recorded without water suppression to preserve the hyperpolarized water pool. Selective proton pulses were used (PC90 1000 µs for 90°, REBURP 2000 µs for inversion), carrier set to 10 ppm to avoid water excitation, so only amide proton shifts >8 ppm were detected. Single-scan acquisitions with no phase cycling; FID detection of 0.1 s under GARP decoupling; interscan delay d1 = 0.5 s; d2 = 0.00345 s; d0 = 0.00002780 s (not incremented). Water-selective NOESY (WS-NOESY): A sequence comprising protein presaturation (deleting protein polarization), mixing time T_m to allow solvent-to-protein polarization transfer, and 1H-15N Fast-HSQC readout. Water-selective 180° pulses and gradients were used (e.g., 7.5 ms Gaussian at 4.7 ppm for water; 15N carrier at 118.5 ppm). For WS-NOESY, experiments were on a 600 MHz Bruker NEO with a Prodigy TCI probe at 298 K; 128 complex t1 increments; data processed with shifted sine-bell apodization and NMRPipe baseline correction. Selective detection strategy: By tuning the mixing/interscan delay (T_m ≈ d1 = 0.5 s), only residues with fast solvent-mediated polarization transfer (via chemical exchange and NOE) achieve sufficient polarization for single-scan detection, rendering the spectra sparse enough that 1D lines become resolvable and assignable. Data analysis: Lorentzian line-shape fits (ten Lorentzians) were used to deconvolute 1D spectra; time traces fitted with monoexponential decays to extract apparent relaxation/decay rates (R1,app). All experiments were replicated three times with similar results.

- Hyperpolarized water selectively boosts signals of solvent-exposed residues, enabling residue-resolved 1D NMR time series at 2 Hz sampling (interscan delay 0.5 s) under near-physiological conditions (physiological saline, pH 7.4, 37 °C). - Spectral sparsification via selective detection and rapid polarization replenishment yields only 13 cross-peaks in WS-NOESY at T_m = 0.5 s, projecting onto 10 discernible lines in the 15N-edited 1H 1D spectra; seven residues were unambiguously assigned. - Time-dependent enhancements and decay kinetics were measured for individual residues over ~20 s post-mixing, revealing residue-specific apparent decay rates influenced by local environment and exchange rates. - Enhancement factors (ε) and apparent decay rates (R1,app, s^−1) for the ten fitted lines (Table 1): • L8: ε = 2.7; R1,app = 0.14 s^−1 • A46: ε = 2.2; R1,app = 0.29 s^−1 • Q2: ε > 11.6; R1,app = 0.25 s^−1 • T14: ε = 6.2; R1,app = 0.70 s^−1 • L71/T12/Q49/R42 (overlapped): ε = 12.4; R1,app = 0.15 s^−1 • D39/I23 (overlapped): ε = 9.1; R1,app = 0.16 s^−1 • R74: ε = 12.8; R1,app = 0.16 s^−1 • L69: ε = 5.0; R1,app = 0.13 s^−1 • n/a (unassigned line): ε > 11.6; R1,app = 0.19 s^−1 • G47: ε = 3.4; R1,app = 0.11 s^−1 - Qualitative trend: residues in hydrophobic/buried regions show faster apparent decays (reduced replenishment), whereas flexible/solvent-exposed residues decay more slowly. - Demonstrated that WS-NOESY residue intensities at T_m = 0.5 s predict which residues are observed in the d-DNP time series with d1 = 0.5 s, confirming the role of chemical exchange and solvent-relayed NOE in site-specific enhancements. - Achieved strong signal boosts immediately after mixing, with decay toward thermal equilibrium; many resonances fall below detection threshold without hyperpolarization under these physiological conditions. - Overall signal enhancements accessible with d-DNP can reach up to ~10,000-fold under suitable conditions.

The method overcomes a central limitation in biomolecular NMR by reconciling sub-second temporal resolution with residue-level readout under near-physiological conditions. By exploiting the residue-dependent kinetics of proton exchange and solvent-relayed NOE, selective enhancement highlights a subset of residues sufficiently for single-scan 1D detection, enabling deconvolution and assignment without resorting to slow multidimensional acquisitions. The strong correspondence between WS-NOESY intensities at short T_m and residues observed in d-DNP time series confirms that solvent interaction efficiency governs hyperpolarization transfer and observable enhancements. Apparent decay rates vary with residue environment: buried residues show faster decay due to limited replenishment from the solvent pool, while flexible, solvent-exposed residues decay more slowly. Buffer properties (pH, temperature, ionic strength) modulate exchange rates and therefore both enhancement and decay; higher pH and temperature generally increase exchange and enhancements up to an optimum before relaxation losses dominate. Solvent-relayed NOE can further amplify residues with protic side chains (e.g., D39, Q49, K63). Collectively, the findings validate that hyperpolarized water combined with selective detection provides a practical route to simultaneously track multiple diagnostic residues in real time, offering new access to kinetic information otherwise inaccessible with thermally polarized NMR.

The study presents a practical strategy for residue-resolved, real-time protein NMR enabled by hyperpolarized water and selective detection. By tuning the interscan/mixing time to favor fast-exchanging residues, spectra become sufficiently sparse to deconvolute and assign 1D lines, achieving 2 Hz sampling over tens of seconds under near-physiological conditions. The approach complements established multidimensional protein NMR by providing sub-second temporal resolution without sacrificing residue specificity for many folded proteins. Potential applications include monitoring protein dynamics, enzymatic processes, and ligand binding events in real time. Future work can extend the methodology to different proteins, optimize buffer conditions and pulse sequences for broader residue coverage, integrate with alternative detection schemes (e.g., SOFAST-HMQC/BEST-HNCO), and apply it directly to time-resolved interaction studies and drug discovery assays.

- Applicability is best for well-folded proteins with sufficient 1H chemical-shift dispersion; intrinsically disordered proteins (IDPs) with narrow 1D proton dispersion may still require 2D/3D detection, limiting time resolution. - Hyperpolarization lifetimes (seconds to minutes) constrain the duration and dimensionality of acquisitions; multidimensional residue-resolved experiments remain challenging within the hyperpolarized window. - Enhancements depend sensitively on buffer pH, temperature, and ionic strength; too-rapid exchange can diminish net signal due to relaxation losses during pulse sequence evolution and detection. - Selective detection emphasizes solvent-exposed/fast-exchanging residues; buried/slow-exchanging sites may be underrepresented, potentially biasing structural/kinetic interpretation. - Overlap among residue signals may persist even in sparse spectra, leaving some lines ambiguous or unassigned.