Long-range distance determination in fully deuterated RNA with pulsed EPR spectroscopy

Abstract

Pulsed electron-electron double resonance (PELDOR or DEER) spectroscopy is powerful in structure and dynamics study of biological macromolecules by providing distance distribution information ranging from 1.8 to 6 nm, providing that the biomolecules are site-specifically labeled with paramagnetic tags. However, long distances up to 16 nm have been measured on perdeuterated and spin-labeled proteins in deuterated solvent by PELDOR. Here we demonstrate long-range distance measurement on a large RNA, the 97-nucleotide 3′SL RNA element of the Dengue virus 2 genome, by combining a posttranscriptional site-directed spin labeling method using an unnatural basepair system with RNA perdeuteration by enzymatic synthesis using deuterated nucleotides. The perdeuteration removes the coupling of the electron spins of the nitroxide spin labels from the proton nuclear spin system of the RNA and does extend the observation time windows of PELDOR up to 50 μs. This enables one to determine long distances up to 14 nm for large RNAs and their conformational flexibility.

Significance

Pulsed electron-electron double resonance (PELDOR) spectroscopy is powerful in structure and dynamics study of RNAs by providing distance distribution information ranging from 1.8 to 6 nm. We here present a long-range distance measurement method for PELDOR in RNA by combination of unnatural basepair system-based site-specific spin labeling and enzyme-mediated RNA perdeuteration. The present method significantly extends the observation time window of PELDOR and allows distance measurements up to 14 nm in RNA, which will broaden the application of PELDOR in structural study of large RNAs such as long noncoding RNA.

Introduction

Noncoding RNAs play an important role in many biological processes, for example, with ribozymes involved in catalytic processes and riboswitches in gene regulation. Like on riboswitches, many of these RNAs have to adapt several distinct secondary and tertiary structures, triggered by specific interactions with ligands, to accomplish their regulatory function. The knowledge of structures and interactions as well as conformational dynamics is essential to understand their functions and in developing RNA-based pharmaceuticals (1,2). Structural characterization of RNAs using techniques such as x-ray crystallography and Cryo-EM remains a challenge due to RNA’s inherent high flexibility (3,4). Solution techniques, such as NMR, small angle x ray, and neutron scattering (SAXS/SANS), allow describing the flexibility and obtaining unique structural and dynamical information of the RNA or RNA-ligand complex. However, NMR is limited to short RNA (on average 50–60 nucleotides (nts)) and SAXS/SANS has relatively low resolution (5,6). There is a growing trend in the field to characterize the structure, interaction, and dynamics of RNAs using integrative methods, which combines data from multiple sources or techniques with computational modeling. Among these techniques, pulsed electron paramagnetic resonance (EPR) methods, such as PELDOR (pulsed electron-electron double resonance) also called DEER (double electron-electron resonance), x-ray scattering interferometry, and single molecule fluorescence resonance energy transfer (smFRET), are molecular rulers that can measure intra- or intermolecular distance distributions at nanometer range, provided that the biomolecules are site-specifically labeled with spin labels, gold nanoparticles, or fluorophores, respectively (7, 8, 9, 10, 11). A number of studies have demonstrated that PELDOR-measured distance restraints, coupled with NMR and computational modeling, are highly informative in investigating the structure and dynamics of RNA and RNA-ligand complexes (12, 13, 14).

Two paramagnetic spin labels are required to be site-specifically attached to the biomolecule to perform PELDOR measurements. In most cases, nitroxide-based spin labels are used, well-known from EPR work developed on proteins, where site-directed spin labeling to cysteine residues has been first explored. On RNAs, several different strategies covalently attaching the spin labels are described in the literature. Different kind of nitroxides were covalently linked to the nucleotide base, the phosphate backbone, or the sugar moiety. With short RNAs, mainly two strategies were explored to obtain the double spin-labeled oligonucleotides. The spin-labeled nucleotides could be directly used in the solid-phase chemical synthesis of the oligonucleotides, or postsynthetic spin labeling of prefunctionalized specific nucleotides was performed by chemical reactions via Sonogashira cross-coupling, click chemistry, or a photolabel protection group (15, 16, 17, 18). The postsynthetic method avoids reduction of the spin labels by the reagents in the solid-phase synthesis. However, with solid-phase synthesis, only short RNAs with fewer than 100 nucleotides are easily accessible. Longer spin-labeled RNAs have been achieved by ligation techniques (19,20) or, more recently, by posttranscriptional (10) or co-transcriptional (21) spin labeling using the NaM-TPT3 unnatural base pairs (UBPs) system (Fig. 1 A). This opens up the possibility to investigate long RNAs by EPR spectroscopy.

Figure 1.

Spin labeling of the 97-nt DENV2 3′SL. (A) Chemical structures of the parental NaM-TPT3 unnatural basepair. (B) Conjugation of the TPT3^CO-modified RNA with the azide-modified nitroxide (AZ-TMIO) via click chemistry. (C) Secondary structure of the DENV2 3′SL. The spin labeling sites are indicated by black circles. (D) 3D atomic model of the DENV2 3′SL (22). Two pairs of spin labeling sites, 67/86 and 58/97, are indicated, and the distances between the C1′ atoms of the respective nucleotide pairs are measured as 5.5 nm and 10.5 nm, respectively.

The maximum distance accessible by PELDOR spectroscopy is given by the time window τ₂ spanned by the microwave excitation pulses (Fig. 2). By Fourier arguments the maximum distance between the two spin labels that can be safely detected in such an experiment is given by the following (23,24):

$$R_{max}\left[nm\right]=4\sqrt[3]{{\tau }_2\left[\mu s\right]}$$ (Equation 1)

Figure 2.

The four-pulse PELDOR/DEER experiment. Upper left: Microwave pulse settings and delays. The gray inversion pulse is stepped in the experiment. Upper right: Symbolic representation of the two spin labels A and B. Lower left: Dipolar time trace S(t). Lower right: Distance distribution function P(R) obtained from S(t).

A longer observation time window is required to quantify the width and shape of the obtained distance distribution P(R). This is especially the case for broad distance distributions, which might result from very flexible RNA molecules. One of the reasons is that the mathematical problem to disentangle P(R) from S(t) is ill-posed. Typically, this is solved by Tikhonov regularization approaches, where the regularization parameter has to be carefully chosen (25). The second reason is that intermolecular dipolar interactions to spins of other molecules in the sample contribute to the time domain signal S(t). This contribution can only be well separated if the observation time window exceeds the dipolar oscillation period of the intramolecular coupling of the spin pair. The maximum achievable time window τ₂ is limited by the phase-memory time T_m of the used spin labels. This time is mainly limited due to the dipolar coupling network between the unpaired electron spin and proton nuclear spins in the surroundings for nitroxide spin labels at the experimental temperature of 50–70 K. It has been shown that proton spins in the range of 0.6–1 nm are responsible for the electron spin phase-memory time T_m, which is typically in the range of 2–4 μs for nitroxide spin labels in protonated solvent and protein or nucleic acid (26,27). A careful choice of the time windows of τ₁ and τ₂ can help to optimize the observation time window. In a similar way, Carr-Purcell-like extended pulse sequences can be used to decouple the nuclear spin network and prolong the observation time window (28, 29, 30). A chemical approach is deuteration of the solvent and/or the biomolecule, reducing the dipolar coupling strength about a factor of 6. This has been demonstrated for proteins, where long observation time windows have been achieved with fully deuterated protein and deuterated solvents (31,32).

Here, we show that by taking advantage of the UBP-based posttranscriptional spin labeling strategy (10) and RNA perdeuteration (Fig. 1 A and B), we could record faithfully distances larger than 10 nm with PELDOR experiments on a fully deuterated long RNA, the 97-nucleotide 3′SL element of the dengue virus serotype 2 (DENV2) genomic RNA (Fig. 1 C and D). Synthesis of perdeuterated RNA is achieved by in vitro transcription using deuterated nucleotides. RNA perdeuteration and deuterated solvents together allow extending the observation time windows of PELDOR up to 50 μs, thus enabling one to determine long distances up to 14 nm for large RNAs and their conformational flexibility.

Materials and Methods

Materials

The phosphoramidites of dNaM and dTPT3, triphosphorylated nucleotides of dNaM, dTPT3, rTPT3^CO, and azide modified nitroxide (5-azide-1,1,3,3-tetramethylisoindolin-2-yloxyl) (Az-TMIO) were synthesized as previously described (10). Fully deuterated rNTP mix (ATP, GTP, UTP, CTP) (U-D, 98%) was purchased from Cambridge Isotope Laboratories.

Preparation of UBP-modified DNA templates

The DNA sequence encoding the DENV2 3′SL RNA is the same as previously reported (22). All DNA oligonucleotide primers (containing natural and/or unnatural nucleotides) were synthesized by regular solid-phase chemical synthesis and purified by OPC purification by TSINGKE Biological Technology (Beijing, China). The preparation of dsDNA templates containing one or two NaMs at the specific sites of the template strands for in vitro transcription were generated by following a two-step overlap extension PCR protocol as described previously (10). The native and unnatural DNA primers used in this study are shown in Table S1.

Preparation of protonated or deuterated 3′SL RNAs with rTPT3^CO modification

The rTPT3^CO-modified RNAs were prepared by following the same protocol as previously reported (10). Briefly, the rTPT3^CO-modified RNAs were prepared by in vitro transcription using the respective dsDNA templates and the rNTP mix supplemented with nondeuterated rTPT3^COTP. Fully protonated or deuterated rNTP mix was used to prepare protonated or deuterated RNAs, respectively. For 3′SL spin-labeled at positions of 67 and 86, both protonated and deuterated RNA samples were prepared. For 3′SL spin-labeled at positions of 58 and 97, only deuterated RNA sample was prepared. The transcription supernatants were directly applied to a Superdex 75 column for size exclusion chromatography (SEC) with running buffer containing 20 mM HEPES (pH 7.2), 100 mM KCl, and 1 mM MgCl₂. Fractions containing target RNAs were collected and concentrated with the Amicon Centrifugal Filter Units and stored at −80°C until use. The concentrations of RNA samples were determined by UV-Vis absorption at 260 nm using a NanoDrop 2000 (Thermo Scientific). The molar extinction coefficients of RNAs were calculated from the primary RNA sequences using an OligoAnalyzer Tool (https://sg.idtdna.com/pages/tools/oligoanalyzer).

Site-specific spin labeling of 3′SL RNAs

The procedure for site-directed spin labeling of 3′SL RNA is similar as previously reported (10). Briefly, purified DENV 3′SL containing two rTPT3^CO at the specific sites were first precipitated with ethanol, and then dissolved with appropriate RNase-free water to a final concentration of 100 μM. The reactive Az-TMIO nitroxide (dissolved in 100% DMSO) was attached to the modified RNAs by click chemistry with the molar ratio of 1:200 for double labeling of RNAs. The reaction mixtures were incubated at 25°C for 6 h, and were quenched by addition of EDTA to a concentration of 5 mM. The mixtures were buffer-exchanged to the SEC buffer using Amicon Ultra Centrifugal Filter Devices (MWCO 10K, Millipore), and the spin-labeled RNAs were purified with SEC by passing the mixtures through a Superdex 75 column. Fractions containing the spin-labeled RNAs were collected and concentrated for further use.

Small angle x-ray scattering

The wild-type and UBP-modified RNA samples were all prepared by in vitro transcription using protonated rNTP mix. SAXS experiments for all RNA samples were carried out in a buffer containing 20 mM HEPES (pH 7.2), 100 mM KCl, 1 mM MgCl₂, 5 mM DTT, and -3′% (v/v) glycerol. The data collection and processing procedures are similar to that described before (22). Briefly, SAXS measurements were carried out at room temperature at the beamline 12 ID-B of the Advanced Photon Source, Argonne National Laboratory. The setups were adjusted to achieve scattering q values of 0.005 < q < 0.89 Å-1 (12ID-B), where q = (4π/λ) sinθ, and 2θ is the scattering angle. Thirty two-dimensional images were recorded and reduced for each buffer or sample, and no radiation damage was observed. Scattering profiles of the RNAs were calculated by subtracting the background buffer contribution from the sample buffer profile using the program PRIMUS3.2 following standard procedures (33). Guinier analysis was performed to calculate the forward scattering intensity I(0) and the radius of gyration (R_g), which were also estimated from the scattering profile with a broader q range of 0.006–0.30 Å⁻¹ using the indirect Fourier transform method implemented in the program GNOM4.6 (34), along with the pair distance distribution function (PDDF), p(r), and the maximum dimension of the protein, D_max. The volume of correlation (V_c) was calculated using the program Scatter, and the molecular weights of solutes were calculated on a relative scale using the R_g/V_c power law developed by Rambo et al. (35), independent of the RNA concentration and with minimal user bias.

Continuous-wave X-band EPR spectroscopy

5 μL samples of the spin-labeled RNAs were mixed with 4 μL glycerol, 1 μL 5× SEC buffer to record continuous-wave (cw)-X-band EPR spectra on a Bruker EMX X-band spectrometer equipped with an ER-041X microwave bridge and a high sensitivity cavity (ER-4119HS, Bruker BioSpin). All cw-EPR spectra were acquired at room temperature with a microwave power of 2 mW, a modulation frequency of 100 kHz, and a modulation amplitude of 1 G. Spin counting was performed based on the second integral values of the baseline-corrected but not normalized cw-EPR spectra.

Pulsed Q-band EPR spectroscopy

We executed all pulsed EPR experiments on a Bruker ELEXSYS E580 EPR spectrometer (Bruker BioSpin, Rheinstetten, Germany) at Q-band (∼34 GHz) frequencies. The spectrometer was equipped with a Flexline probe head, an ER5106QTII resonator (Bruker), and a 150-W TWT-amplifier (model 187 Ka, Applied Systems Engineering, Fort Worth, TX, USA). The temperature was adjusted to 50 K by using a CF935 helium gas-flow cryostat (Oxford Instruments, Abingdon, UK) and an iTC502 temperature controller (Oxford Instruments). All data were acquired using quadrature detection.

Regarding the pulse settings for PELDOR, PELDOR experiments were performed using the standard four-pulse sequence π/2(νA)–τ₁–π(νA)–(τ₁+t)–π(νB)–(τ₂-t)–π(νA)–τ₂–echo. The magnetic field was set to the maximum of the nitroxide spectrum and the pump pulse applied at νB (34 GHz) in the center of the cavity resonance profile. The detection frequency νA was set lower than νB by an offset of 70 MHz. Detection pulse lengths and the microwave power were adjusted to obtain π/2 and π pulses with the same length (typically 32 ns). The length of the pump pulse was adjusted by a transient nutation experiment in such a way that a maximal inversion of the Hahn echo was obtained (typically 20 ns). We used the full height of the ER5106QTII cavity to get the best S/N on each sample, which required about 60 μL sample volume. We used the AWG combined with the unlocked ELDOR source. For the interpulse delay τ₁, the initial time value was set to the first maximum in the two-pulse ESEEM trace (232 ns). To suppress deuterium ESEEM in the PELDOR trace, a modulation averaging procedure was applied incrementing τ₁ eight times by 16 ns and summing the individual traces (nuclear modulation averaging). The dipolar evolution time τ₂ was adjusted to the actual spin distance or the maximal reachable time (of about 2% echo signal in the ESEEM time trace). The echo integration was positioned symmetrically around the refocused Hahn echo. The integration window was set to 50 ns. For every point on the PELDOR time trace, 20 shots were averaged and the shot repetition time was set to 2 μs. Two-step phase cycling of the π/2 pulse was used to eliminate undesired echoes and receiver baseline offsets.

The optimum spin concentration was reached with a 20 μM RNA concentration. The aforementioned sample concentrations have been reduced by maintaining the 3:1 ration of D₂O: glycerol-d8 and 1 mM Mg²⁺ concentration.

PELDOR time trace analysis

Each recorded PELDOR time trace was analyzed by Tikhonov regularization using DeerAnalysis 2019 (25). The Tikhonov analysis was done with standard parameters, a three-dimensional (3D) background (mono exponential), the Tikhonov regularization L-curve was calculated, and the optimum of the L_c criteria selected. In addition, each time trace was fitted using the DEERNet (generic settings) method to get a confidence interval of the distance distribution (36). The DEERNet fit includes the dimensionality of the background, which was usually close to three dimensional.

Results

High-yield spin labeling of perdeuterated large RNA

DENV2 3′SL consists of a small hairpin followed by a large stem loop (SL), which is folded to have an extended rod-like 3D structure in solution estimated by SAXS and computational modeling (Fig. 1 B and C) (22). Based on the 3D atomic model, two pairs of labeling sites are selected in the single-stranded regions, and the nitroxide spins are site-specifically labeled to DENV2 3′SL using the UBP-based strategy as reported recently (Fig. 1 B and C) (10). Briefly, the DNA templates encoding DENV2 3′SLs are site-specifically modified with the dNaM-dTPT3 UBP, dNaMs in the DNA template strands direct site-specific incorporation of rTPT3^CO into the fully protonated or deuterated 3′SL RNAs during in vitro transcription catalyzed by T7 RNA polymerase using protonated or deuterated nucleotides (Table S1, Fig. S1). SAXS profiles demonstrated that the incorporation of UBP has the minimal perturbation to the global folding of 3′SL RNAs (Fig. S2, Table S2). Posttranscriptional spin labeling to rTPT3^CO is achieved by click chemistry with an azide-functionalized nitroxide. cw-EPR analyses show that the nitroxide labels are attached to the 3′SL RNAs with high efficiency (>90%) (Fig. S3).

Perdeuteration of RNA prolongs the observation time window

Fig. 3 depicts the phase-memory time measured by a two-pulse Hahn echo sequence at Q-band frequency (∼34 GHz). The signal intensity is monitored as a function of the dephasing time 2τ for protonated and deuterated RNA (both in deuterated buffer solution). Deuteration of the RNA molecule prolongs the echo signal intensity considerably, demonstrating that the RNA protons indeed reduce the spin-label phase-memory time T_m. A reduction of the RNA concentration from 70 μM to 20 μM further drastically increases the phase-memory time to a value of T_m = 25 μs. In this case, reducing the RNA concentration does not prolong the phase-memory time further; however, this is particular for this RNA and is not a general statement. This demonstrates that for fully deuterated surroundings, the intermolecular dipolar interaction network between the spin labels itself limits the phase-memory time T_m for spin-label concentrations larger than about 50 μM, as has already been evaluated on protein samples (31). The blue horizontal line illustrates the possible accessible dipolar evolution times under these conditions. The minimal observable signal intensity of 2% of the maximal signal intensity would translate to an experimental measurement time of about 2 days on our PELDOR setup. The table insert converts this observation time window to maximum observable distances by using Equation (1).

Figure 3.

Hahn echo signal decay of the spin-labeled 3′SL RNAs. Experimental data are shown for a 65 μM protonated RNA sample (light gray), a 70 μM deuterated RNA sample (gray), and a 20 μM deuterated RNA sample (black). The blue horizontal line at 2% of the maximum signal intensity illustrates the accessible time, corresponds to about a 2-day PELDOR experiment time on our setup for suitable S/N. The inset table reports the corresponding τ₂ values of the PELDOR experiments and the maximum accessible distances calculated from Equation (1).

Long-range distance determination by PELDOR

Fig. 4A compares the PELDOR time traces and the therefrom obtained distance distributions for protonated and deuterated 3′SL RNAs doubly spin-labeled at positions 67 and 86. PELDOR experiments with a dipolar evolution time window of 15 μs and a distance of about 5.8 nm are easily accessible for both type of samples. The experimental measurement time for this experiment was 1 h for both samples. The very similar time traces and distance distributions demonstrate that the tertiary structure of the RNA is not changed upon deuteration of the RNA (Figs. S4–S11). Nevertheless, the differences in the S/N observed in the time traces reflect the prolonged phase-memory time of the deuterated RNA sample, which also leads to better defined peaks in the distance distribution (visualized by the shaded area, 95% confidence interval). The distance peak observed at distances <4 nm is probably due to some end-to-end stacking of the RNA molecules (Fig. S12). Fig. 4 B shows that with a 20-μM deuterated RNA sample a dipolar time window as long as 35 μs could be experimentally achieved. This allows one to unambiguously determine the slightly asymmetric distance distribution for a sample with maximum distance larger than 10 nm between both spin labels (C58 and U97).

Figure 4.

PELDOR time traces (left) and distance distributions (right) on doubly spin-labeled DENV2 3′SL RNAs. (A) Protonated (light gray) and deuterated (black) 3′SL RNA spin-labeled at residues 67 and 86. The experiment time is 1 h in both cases. (B) On the left are shown the experimental time trace (black) and the fitted intermolecular background function (light gray) of fully deuterated DENV2 3′SL RNA spin-labeled at residues 58 and 97. On the right, the distance distribution shows a main distance of about 10 nm with an asymmetric shoulder at about 8 nm. Only the long time window of 35 μs accessible with fully deuterated 20 μM RNA allows one to quantitatively obtaining this distance distribution function. Measurement time is 18 h in this case. All distance distribution functions are obtained by using DEERNet (36) within the DeerAnalysis (25) program.

Discussion

Our experiments demonstrate that perdeuteration of RNA substantially prolongs the observation time window for long-range distance determination by PELDOR. Although preparation of perdeuterated protein through recombinant expression in Escherichia coli during growth on highly deuterated media and in bulk D₂O for NMR, EPR, and SANS application is often difficult (37), the enzymatic synthesis of perdeuterated RNA described here is much easier and straightforward. The very efficient posttranscriptional spin labeling protocol established with UBP system together with the perdeuteration of RNA resulted in a >80% spin labeling efficiency (16), as determined by cw-EPR spin counting and calculated from the PELDOR modulation depth. It would be interesting to combine it with advanced pulse decoupling schemes to further prolong the observation time window (Fig. S13). Our results demonstrate the possibility to obtain long-range distance restraints for integrative modeling and quantitative information on the conformational flexibility of biologically relevant large RNAs.

Author contributions

X.F. conceived and designed the project. T.P. supervised the PELDOR experiments. Sample preparation was performed by Y.H. cw-EPR experiments were performed by G.B. and G.L. Pulsed EPR experiments were performed by B.E. The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.