Gd(III)-Gd(III) RIDME for In-Cell EPR Distance Determination

In-cell distance determination by EPR reveals essential structural information about biomacromolecules under native conditions. We demonstrate that the pulsed EPR technique RIDME (relaxation induced dipolar modulation enhancement) can be utilized for such distance determination. The performance of in-cell RIDME has been assessed at Q band using stiff molecular rulers labelled with Gd(III)-PyMTA tags and microinjected into X. laevis oocytes. The overtone coefficients are determined to be the same for protonated aqueous solutions and inside cells. As compared to in-cell DEER (double electron-electron resonance, also abbreviated as PELDOR), in-cell RIDME features approximately 5 times larger modulation depth and does not show artificial broadening in the distance distributions due to the effect of pseudo-secular terms.

In-cell distance determination by EPR reveals essential structural information about biomacromolecules under native conditions. We demonstrate that the pulsed EPR technique RIDME (relaxation induced dipolar modulation enhancement) can be utilized for such distance determination. The performance of in-cell RIDME has been assessed at Q band using stiff molecular rulers labelled with Gd(III)-PyMTA tags and microinjected into X. laevis oocytes. The overtone coefficients are determined to be the same for protonated aqueous solutions and inside cells. As compared to in-cell DEER (double electron-electron resonance, also abbreviated as PELDOR), in-cell RIDME features approximately 5 times larger modulation depth and does not show artificial broadening in the distance distributions due to the effect of pseudo-secular terms.

TOC GRAPHICS
KEYWORDS RIDME, Xenopus laevis oocytes, in-cell distance determination, dipolar spectroscopy, Gd(III)-PyMTA, spin labels, in-cell EPR Electron paramagnetic resonance (EPR) spectroscopy provides means to determine distances between magnetically coupled electron spins. By combination of pulsed EPR techniques with site-directed spin labelling, distances in the range up to 8 nm, and under the very special condition of a perdeuterated biomacromolecule up to 16 nm, can be determined. 1 reduce the artefacts at short distances by using chirp pulses 28 and a dual-mode cavity, 29 these improvements can be achieved only with an additional comprehensive and expensive equipment of the EPR spectrometer with a high-power microwave amplifier, an arbitrary-waveform generator, a broad-band or a dual-mode cavity.
Alternatively, the downsides of conventional Gd(III)-Gd(III) DEER can be overcome in a related experiment abbreviated RIDME (relaxation induced dipolar modulation enhancement). [30][31][32] RIDME is a single-frequency technique and makes use of relaxation-induced spin flips, whereas DEER is a two-frequency technique which detects at the observer frequency while flipping coupled spins in a controlled way by a -pulse at the pump frequency. Thus RIDME is technically less demanding than DEER and has no limitations with respect to the excitation bandwidth. As an example, consider a rectangular pump pulse of 24 ns (excitation bandwidth of 40 MHz), which flips only a small part of the spins out of the 2-GHz-broad spectrum of Gd(III).
While the contribution to the DEER signal is determined by this part of excited Gd(III) spins, the contribution to the RIDME signal is determined by the spins from the entire spectrum, which flip due to relaxation. Consequently, the modulation depth of in vitro Gd(III)-Gd(III) RIDME is increased by the factor of 10 compared to in vitro Gd(III)-Gd(III) DEER. 26,[33][34][35] These features of RIDME can also be beneficial for distance measurements inside cells. While the performance of Gd(III)-Gd(III) RIDME has recently been demonstrated for model systems in deuterated solvents at Q-and W-band frequencies, [33][34][35] the in-cell application remained elusive. To the best of our knowledge, the only in-cell RIDME experiment was reported for trityl spin labels, which are S=1/2 systems and have very different relaxation properties as compared to Gd(III) complexes. 36 Here, we utilize stiff molecular rulers spin labelled with a Gd(III)-PyMTA complex ( Fig. 1A and B) and demonstrate that the 5-pulse dead-time free RIDME sequence 32 (Fig. 1C) is suitable for in-cell distance measurements between Gd(III)-based spin labels. Dead-time free RIDME pulse sequence.
In-cell RIDME implies that the measurement is performed in protonated media, which results in shorter relaxation times, as compared to deuterated solvents. Additionally, RIDME features a steeper signal decay than DEER because of the hyperfine spin diffusion. Therefore, accurate choice of experimental conditions with respect to relaxation and spin diffusion is crucial to realize in-cell RIDME. The measurement of relaxation times for Gd-ruler-3.0 in H2O/glycerol (8/2 by volume) were performed in the temperature range between 10 and 30 K (Fig. S1). The phase memory time decreases from 1.4 to 0.8 μs and the ratio T2/T1 increases from 0.02 to 0.09 upon increasing the temperature. Though the highest ratio T2/T1 is favourable for RIDME, we performed all measurements at 10 K to have the longest phase-memory time. The pulse lengths were set at 8 and 16 ns for and , respectively. The mixing time of 8 μs provided a reasonable tradeoff between the steepness of the RIDME background and the dipolar modulation depth. Figure 2 shows the results of in vitro RIDME measurements in frozen protonated aqueous solutions (8/2 mixture of H2O and glycerol) for two molecular rulers that bear Gd-PyMTA as the spin labels with distances of 2.1 and 3.0 nm between the two Gd(III) ions. [37][38][39] The RIDME form factors display modulation depths of 15% and 21% and visible oscillations, which indicate narrow distance distributions.
For determination of the interspin distances from the RIDME data, it was taken into account that relaxation-induced spin flips with Δ ≥ 1 are possible for Gd(III) ions which have S = 7/2.
Thus, apart from the fundamental frequency of the dipolar coupling , which corresponds to spontaneous spin flips with Δ = 1, overtones of this fundamental frequency occur, which originate from spin flips with Δ > 1. It has been shown for Gd(III) ions that only the first two overtones, i.e., 2 and 3 , contribute significantly to the RIDME form factor and that the overtone fractions can be assumed constant for distances above 3 nm, but may vary if the distance is shorter. 35 The RIDME data were processed with the OvertoneAnalysis software package, which explicitly includes the overtone fractions into the kernel function to extract the distance distribution by the Tikhonov regularization procedure. 35 The fractions, which are represented by the coefficients , , and (for the fundamental frequency and its overtones at 2 and 3 , respectively), were determined by their systematic variations towards a narrow single-peak distance distribution, as is expected for the studied Gd-rulers. 26   In general, in vitro RIDME measurements applied to Gd(III) spin labelled compounds provide larger modulation depths and more accurate distance distributions than in vitro DEER measurements. 34,35 In-cell measurements are performed under the challenging conditions of a protonated environment. In the presence of protons, the relaxation processes become faster and the hyperfine spin diffusion stronger. This results in a faster decay of the RIDME signal and shortens the length of the dipolar evolution time trace. If the experimental settings are chosen to maximize the length of the dipolar evolution time trace, i.e., low temperature and short mixing time, a RIDME time trace of at least 2 μs can be recorded, be it at the cost of the modulation depth. In this case, the modulation depths of in-cell and in vitro RIDME are in the range of 13-21 %, which is still less than the modulation depth of 50 % for RIDME measurements in deuterated solutions 33 but is a significant improvement over an in-cell DEER measurement using the same Gd-ruler-3.0 (a modulation depth of about 4 %). 18 The overtone coefficients in a protonated aqueous solution and inside cells are the same for Gd-PyMTA spin labels at a given interspin distance. The distance distributions extracted from RIDME do not contain artefacts caused by pseudo-secular terms, as opposed to DEER. 20, 26 We find that the overtone coefficients for aqueous solutions are significantly different for Gd(III)-Gd(III) distances of 2.1 and 3.0 nm, in agreement with what was reported for deuterated solvents. 35 This implies that for broad distance distributions, which span distances below and above 3 nm, the dependence of the overtone coefficients on the interspin distance has to be included in the processing of the RIDME data.
In conclusion, we demonstrated for the first time the performance of Gd(III)-Gd(III) RIDME for distance measurements inside cells. It is suggested that the overtone fractions remain the same for samples in aqueous solution and inside cells, being primarily determined by the presence of protons in the medium. At the current state, Gd(III)-Gd(III) RIDME can only be applied to distance distributions where the overtone fractions can be assumed constant. Further methodological developments are required to account for variable overtone fractions during data processing and also to slow down the decay of the RIDME signal during data acquisition. Along these lines, current developments of other Gd(III)-based spin labels [40][41][42][43][44][45] and the application of shaped pulses 46 are very promising. Giving its inherent advantages, i.e., precise distance determination and large modulation depth, RIDME has the potential to become a method of choice for in-cell distance measurements.

ASSOCIATED CONTENT
Supporting Information is available free of charge.

Notes
The authors declare no competing financial interests.

EPR spectroscopy
All EPR measurements were performed at Q band on an Elexsys E580 spectrometer equipped with an arbitrary waveform generator (Bruker Biospin) and a 150 W TWT amplifier (Applied Systems Engineering). Temperature control was realized with the cryogen-free system consisting of a helium compressor F-70H (Sumitomo Cryogenics of America), a cryocooler ColdEdge (CE-FLEX-4K-0100, Bruker), and a MercuryITC (Oxford Instruments). A Q-band probehead with access for 3 mm tubes (ER5106QT-2, Bruker) was used for both 1.6 mm and 3 mm (o.d.) sample tubes. All the measurements were performed in the dip of the overcoupled resonator at 34 GHz. Rectangular pulses were used with the lengths fixed at 8 and 16 ns for /2 and , respectively.
The primary echo decay ( /2 − − ℎ ) was recorded by increasing the distance between the two microwave pulses with an 8 ns step. The length of the primary echo decay was 15 μs. The inversion recovery ( − /2 − − ℎ ) was recorded by increasing the distance between the first and the second pulse with a 100 ns step. The length of the inversion recovery was 200 ms, In both experiments, the magnetic field was set to 12207 G, which corresponded to the maximum of the absorption curve.
In the RIDME measurement, a 5-pulse scheme was used and an 8-step phase cycle was applied during the acquisition of the refocused virtual echo. 4 The magnetic field was set to the maximum of the absorption spectrum of Gd-PyMTA label. A time delay between the first and the second pulse was 300 ns, a mixing time was 8 μs. The total length of the RIDME dipolar evolution was 2 μs, which was acquired with an 8 ns step.
The RIDME time traces were processed with the OvertoneAnalysis software package for Matlab, which includes a modified kernel function to account for overtone harmonics. 5 The background contributions to every RIDME time trace were approximated by a stretched exponential and eliminated by division. The distance distribution was extracted from the resulting form factor by Tikhonov regularization, where the fractions of overtones were fixed at = 0.4, = 0.3, and = 0.3 (for Gd-ruler-3.0) or at = 0.69, = 0.21, and = 0.10 (for Gd-ruler-2.1). Subsequently, the distance distribution at the optimum alpha-value (ranging from 7 to 39) was subjected to the validation. The validation by varying the background start, the modulation depth, and adding white noise at the level 1.2 to give the uncertainty regions for the distance distribution, which were plotted as grey area on top of the distance distribution. 5