Intracellular Protein–Lipid Interactions Studied by Rapid-Scan Electron Paramagnetic Resonance Spectroscopy

Protein–membrane interactions play key roles in essential cellular processes; studying these interactions in the cell is a challenging task of modern biophysical chemistry. A prominent example is the interaction of human α-synuclein (αS) with negatively charged membranes. It has been well-studied in vitro, but in spite of the huge amount of lipid membranes in the crowded environment of biological cells, to date, no interactions have been detected in cells. Here, we use rapid-scan (RS) electron paramagnetic resonance (EPR) spectroscopy to study αS interactions with negatively charged vesicles in vitro and upon transfection of the protein and lipid vesicles into model cells, i.e., oocytes of Xenopus laevis. We show that protein–vesicle interactions are reflected in RS spectra in vitro and in cells, which enables time-resolved monitoring of protein–membrane interaction upon transfection into cells. Our data suggest binding of a small fraction of αS to endogenous membranes.

pET11C_ASYN_Δ2-11 was kindly provided by the Leist Lab 2-3 and a cysteine was introduced in full analogy to pT7-7_ASYN_A27C with just 10 amino acids lacking in the N-terminus.

S4
Prior to labeling, DTT was removed using PD-10 desalting columns (GE Healthcare). A 6x molar excess of 3-maleimido-2,2,5,5-tetramethyl-1-pyrrolidinyloxy (M-proxyl, Sigma-Aldrich Chemie GmbH) was added and the protein solution was incubated overnight at 4° C. Subsequently, free label was removed with Microsep™ Advance 3K MWCO filters at the same time as the buffer was exchanged to 10 mM Tris-HCl pH 7.4 containing 150 mM NaCl. The protein solution was concentrated in Amicon Ultra-0.5 centrifugal filter units (3K) to a concentration of about 3.5 mM (Table S1). Spin concentration was determined at an EMXnano benchtop X-band spectrometer (Bruker Biospin), and the labeling efficiency was calculated as the quotient of spin and protein concentration. The obtained labeling efficiency was in the range of 90-100 %. Samples were kept on ice during the washing process to minimize aggregation and stored at -80 °C afterwards.
Vesicle size was checked by dynamic light scattering (DLS, Figure S2). 1 µL of the LUV solution was added to 1 mL Tris-HCl (10 mM, pH 7.4) containing 150 mM NaCl in a 1 cm polycarbonate cuvette. The sample was measured at 20 °C using a Zetasizer Nano-ZS spectrometer equipped with a 4 mW He-Ne laser (vertically polarized incident radiation of 633 nm wavelength, Malvern Panalytical).

Microinjection
Microinjection was performed as previously described. 4 After an incubation time of 30 min at 18 °C, the αS solution was microinjected into the opposite side within the black hemisphere. Detailed concentrations and volumes are given in Table S2. Subsequently, seven microinjected oocytes were collected in a Q-band tube (quartz glass, 1 mm inner diameter, Bruker), and mounted into the resonator of the spectrometer.

Determination of the signal-to-noise ratio (SNR)
Quantitative SNR comparison between RS and CW EPR was performed using an EMXplus (Bruker) equipped with a Super High QE (SHQE) resonator (Bruker) and Elexsys 500 X-band spectrometer (Bruker) equipped with the Rapid-scan (RS) accessory (Bruker). 4.04 µM TEMPOL was dissolved in water, and the solution was degassed and loaded into glass capillaries with 1 mm inner diameter (HIRSCHMANN ® ringcaps ® ). A power sweep was performed for both techniques. ( Figure S1a) Nitroxide saturation was found at 1.26 mW or 100 mW for CW or RS. Accordingly, for CW experiments, a power of 1.26mW was chosen.
For RS experiments, the power was limited to 20mW to avoid heating effects (see section rapid-scan EPR) respectively.
For CW spectroscopy, the experimental parameters were optimized to a modulation amplitude of 1 G at a

Spectral simulations
Absorption spectra were simulated with the use of Matlab R2019b and the toolbox EasySpin 5.

Rapid-scan EPR
Either 50 µL aqueous solution or 7 oocytes were loaded into Q-band tubes (quartz glass, 1 mm inner diameter, Bruker) and rapid-scan spectra were recorded using an Elexsys 500 X-band spectrometer (Bruker) equipped with the rapid-scan (RS) Accessory (Bruker). The RS Accessory comprises: (i) water-cooled RS coils mounted on a 10'' magnet for modulating the magnetic field, (ii) RS coil driver to generate the waveform for the scan and control the current of the RS coils, (iii) a capacitor circuit for resonant mode operation, (iv) the microwave front end for detection and amplification of the EPR signal, (v) the RS acquisition unit for signal digitalization, and (vi) dedicated RS resonator that is transparent to the rapidly changing magnetic fields thus avoiding generation of eddy currents.
Rapid-scan spectra were acquired applying sinusoidal rapid magnetic-field scans at a frequency of 20 kHz with a scan width of 200 G (scan rate of 12.6 MG/s). The scan was centered at a static magnetic field of 3280 G with a microwave power of 20 mW. Temperature stability was achieved with a nitrogen flow and temperature stability was ensured by monitoring the stability of the AFC during tuning. It was stable at experimental attenuation which indicates that the resonator and sample were not heated. Data was acquired using a 2D field vs. delay experiment with 270 slices and a minimum delay between slices. Each field slice was acquired in 10 s and consisted of 200400 averages.

Continuous wave EPR
In vitro samples were loaded into glass capillaries with 1 mm inner diameter (HIRSCHMANN ® ringcaps ® ) and measured on an EMXnano benchtop X-band spectrometer (Bruker Biospin) (9.637 GHz) with optimized parameters. ( Figure S11) using a microwave power of 3.162 mW, a modulation amplitude of 0.8 G at a modulation frequency of 100 kHz at room temperature.

Circular dichroism
The Circular dichroism (CD) spectra were measured with a JASCO J-715 spectropolarimeter at 20 °C.
Protein samples (20 µM) were mounted in 0.5 mm demountable cuvettes. Baseline correction and subtraction of the background spectrum (buffer without protein) were applied to the raw data. CD data measured at a HT value above 550 V were removed, as the signal got noisy and unreliable.

Continuous wave EPR
Spectra were corrected according to the microwave frequency, baseline-corrected, and normalized to the double integral with the help of Matlab 2019b. The measured spectra were analyzed using Matlab R2019b (The MatWorks, Inc. Natrick,MA). Field-modulated EPR spectra were computed with the toolbox EasySpin 5.2.25. 7 All spectra were baseline-corrected with a second order polynomial function, smoothed using a Savitzky-Golay filter with a second order polynomial and a frame length of 51. Depending on the spin concentration, 10 to 30 scans were accumulated and normalized to the double integral.

Processing of raw spectra
The detected time domain EPR signal was processed on board into the field domain with a baseline After background-subtraction (I(field)bgcorr(field)), the magnetic field-axis was recalculated to the microwave-frequency of 9.493204 GHz, and the spectra were normalized to the maximum amplitude of the center field peak.

Quantification of the binding process
As revealed from Figure 1b, 2a, and 3, spectral broadening can most prominent be seen in the low-field peak. Therefore, the integral of the inter-peak space between lowfield and centerfield peak (3371-3375 G) was chosen to calculate a protein−lipid interaction factor ξ, the area below a baseline interval (3320-3324 G) represents effects caused by noise and was used to calculate error bars. To compensate for nitroxide reduction effects, normalization to the area below the centerfield peak (3378.3-3379.7 G) was performed ( Figure S5).
In order to reveal intracellular lipid binding with small impact on the spectral shape, time-resolved measurements ( Figure 4b) were fitted with the following function: Resulting parameters are given in table S3.      to membranes. Therefore, translation of ξ into b is not possible for this negative control. Empirical fits (Table S4) were applied as guide to the eye. S14 Table S4. . Nitroxide signal depletion as indicated by the intensity of the centerfield peak. Since the spectral shape did not change over time, not only the spectral integral but also the peak intensity was directly proportional to the number of spins. However, we decided for the intensity as a measure because it was less prone to errors caused by decreasing SNR. S15 Figure S10. Ascorbate reduction assay. M-proxyl labeled αS A27C (222 µM) was incubated with POPG LUVs (13.3 mM) for 30 min. Then, ascorbate was added (6.6 mM in water, pH 7.4) and spectra were recorded subsequently after sample mounting (dark blue) and 30 min later (light blue). (a) The peak-to-peak amplitude was reduced from 779 a.u. to 303 a.u. within 30 min. (b) Normalization to the maximum amplitude of the spectra shows that the spectral shape did not change. This observation indicates that all spin labels are to the same extent prone to reduction.