Dynamic Nuclear Polarization of Biomembrane Assemblies

Abstract

While atomic scale structural and dynamic information are hallmarks of nuclear magnetic resonance (NMR) methodologies, sensitivity is a fundamental limitation in NMR studies. Fully exploiting NMR capabilities to study membrane proteins is further hampered by their dilution within biological membranes. Recent developments in dynamic nuclear polarization (DNP), which can transfer the relatively high polarization of unpaired electrons to nuclear spins, show promise for overcoming the sensitivity bottleneck and enabling NMR characterization of membrane proteins under native-like conditions. Here we discuss fundamental aspects of DNP-enhanced solid-state NMR spectroscopy, experimental details relevant to the study of lipid assemblies and incorporated proteins, and sensitivity gains which can be realized in biomembrane-based samples. We also present unique insights which can be gained from DNP measurements and prospects for further development of the technique for elucidating structures and orientations of membrane proteins in native lipid environments.

Keywords: Dynamic Nuclear Polarization (DNP); Nuclear Magnetic Resonance (NMR); membrane active peptides; membrane proteins; solid-state nuclear magnetic resonance (ssNMR).

Figure 1

(A): Electron and nuclear polarization as a function of temperature and magnetic field. The polarization of an electron (red solid line) and ¹H (black line) at 100 K and 14.1 T are highlighted (red dashed line). The ¹H polarization gain with moving to a higher magnetic field (blue line) is shown for comparison. The general scheme of dynamic nuclear polarization (DNP) is to irradiate at the resonance of an unpaired electron spin (395 GHz at 14.1 T), resulting in the transfer of polarization to nuclear spins, thereby giving DNP enhancement. (B): 600 MHz DNP installation at the National High Magnetic Field Laboratory. The magic angle spinning (MAS)-DNP solid-state nuclear magnetic resonance (ssNMR) system is pictured in the upper right corner along with the gyrotron in the lower right corner. The gyrotron supplies high power microwave irradiation to two magnets via the quasioptic tables in the center of the photograph. A complete description of this instrumentation has been published [20].

Figure 2

(A) Commercially available water-soluble biradicals for DNP. (B) Likely polarization transfer pathways via cross-effect DNP (CE) and spin diffusion (SD) in biomembrane samples incorporating AMUPol. Green diamonds indicate the presence of a glassing agent to maintain AMUPol dispersion during sample freezing.

Figure 3

Cartoon schematic of proteoliposome sample preparation for MAS-DNP. The lipid/peptide multilamellar lipid vesicle (MLVs) are produced following traditional membrane peptide or protein sample preparations for ssNMR. Here, the lipid pellet is first hydrated in excess buffer and subjected to 15X freeze/thaw cycles to form homogenous MLVs and pelleted by ultracentrifugation. Following this, steps 1–3 illustrate addition of polarizing agents (PA), and additional freeze/thaw cycles to distribute the PA and glassing agent prior to sample transfer into the NMR rotor.

Figure 4

Pulse sequence for DNP buildup time (T_B) measurements and example data for a sample containing lipid bilayers and 10 mM AMUPol as the polarizing agent.

Figure 5

(A): Continuous wave (CW) EPR X-band (9.5 GHz) spectra of 10 mM AMUPol in a sample containing 5.5/2.7/2/1 DPPC-d₆₂/DPPC/POPG/cholesterol liposomes with 2 mol% KL₄ peptide. EPR spectra shown in red and black were collected at 298 K and 100 K, respectively. (B): EPR spectra at 298 K for three liposome preparations containing 10 mM AMUPol prepared in parallel. These EPR spectra provide a qualitative assessment of PA distribution between identically prepared biomembrane samples. Sample one contains DPPC-d₆₂/DPPC/POPG/cholesterol liposomes without peptide; samples 2 and 3 contain lipids and 2 mol% KL₄ peptide.

Figure 6

¹³C chemical shift spectra collected with MW off (black) or microwave on (red) for a sample containing 5.5/2.7/2/1 DPPC-d₆₂/DPPC/POPG/cholesterol MLVs with 2 mol% K₄ equilibrated in a buffer containing 10% DMSO-d₆ and 10 mM AMUPol. Spectra were scaled based on the number of transients averaged. The “MW on” spectrum is an average of 32 transients collected in ~2 min; the “MW off” spectrum is an average of 512 transients collected in ~35 min. MAS sidebands are indicated by an *. Since the DMSO is deuterated, solvent resonances contribute minimally to the spectrum.

Figure 7

Comparison of ¹³C chemical shift spectra for a DNP-enhanced sample containing 5.5/2.7/2/1 DPPC-d₆₂/DPPC/POPG/cholesterol and 2 mol% K₄ made with buffer containing 10% DMSO and 10 mM AMUPol (red) vs. a DNP spectrum for a control sample made without peptide (black) and an NMR spectrum for a control sample made without polarizing agent (grey). The DNP spectra are an average of 32 transients collected in ~2 min; the control NMR spectrum is an average of 512 transients collected in ~35 min. Spectra are scaled to the lipid -CH₂ resonance in the lipid only sample and no appreciable differences in resolution were detected between the samples.