Measurement of the Change in Twist at a Helical Junction in RNA Using the Orientation Dependence of FRET

Abstract

Indocarbocyanine fluorophores attached via the 5′ terminus of double-stranded nucleic acids have a strong propensity to stack onto the terminal basepair. We previously demonstrated that the efficiency of fluorescence resonance energy transfer between cyanine 3 and 5 terminally attached to duplex species exhibits a pronounced modulation with helix length. This results from a systematic variation in the orientation parameter κ² as the relative rotation of the fluorophore transition moments changes due to the helical geometry. Analysis of such profiles provides a rich source of orientational information. In this work, we applied this methodology to the structure of a three-way helical junction that plays an important role in the hepatitis C virus internal ribosome entry site. By comparing matched pairs of duplex and junction species, we were able to measure the change in rotation at the junction. The data reveal a 29.5° overwinding and a small axial extension. This shows the power of this approach for measuring orientational information in biologically important RNA junctions.

Introduction

Helical junctions are key architectural elements in RNA (1). Most functional RNA species can be considered as a collection of helical segments that are connected by junctions, and thus the conformational properties of the junctions determine the global folding of the RNA. In this work, we show how the orientation dependence of fluorescence resonance energy transfer (FRET) can be used to obtain valuable information about the conformation of a helical junction.

The hepatitis C virus (HCV) RNA contains a 341 nt segment that allows initiation of translation to occur independently of a capped 5′ terminus (2–4). The internal ribosome entry site (IRES) adopts a folded structure in the presence of divalent metal ions (5,6). The secondary structure (7–10) includes a number of helical junctions that organize the overall structure and mediate some tertiary contacts (Fig. 1 A). The structures of some parts of the IRES RNA have been determined (11–14).

Figure 1.

Three-way junction of the HCV IRES element. (A) Schematic of the secondary structure of the IRES. The three-way helical junction studied here is circled. (B) Sequence of the three-way junction. We have designated the helical arms as C, D, and E. (C) Scheme of the constructs used in this study. The black strand is common to both species, to form either a duplex (left) or junction (right). Both are labeled with Cy3 (C3) and Cy5 (C5) fluorophores as indicated.

The formally 3HS₄ junction that includes helix IIId is very important in the function of the IRES (highlighted in Fig. 1 A). The terminal loop of helix D binds to the 40S ribosomal subunit (15–17), and certain mutations in the loop and internal bulge can abolish HCV IRES binding (18). We previously studied the structure and folding of this junction (Fig. 1 B) (19). Analysis of the global structure indicated that in the presence of Mg²⁺ ions, this junction adopts a conformation based on the coaxial alignment of the C and E helices, with helix D oriented to generate an acute angle with helix C (19). Single-molecule FRET detected no transitions to other conformations down to a timescale of 8 ms. Rather, the junction appears to adopt a very stable fold with a high population of the species based upon the C and E alignment, probably coaxially stacked. However, our previous experiments provided no information on the relative rotational setting of helices C and E about their joint axis, and it is likely that the junction with the D helix introduces a significant change in rotation angle at that point. We therefore set out to examine this using the orientational dependence of FRET.

FRET arises from the dipolar coupling between the TDM vectors of fluorophores, and is therefore a function of both distance and orientation. The efficiency of FRET (E_FRET) is given by (20):

$$E_{FRET}=1/\left(1+R^6/\left(\left(0.529\times {\kappa }^2\times {\phi }_D\times J\left(\lambda \right)\right)/N\times n^4\right)\right)$$ (1)

where R is the separation of donor and acceptor fluorophores (in centimeters), Φ_D is the quantum yield of the donor, N is the Avogadro number, n is the index of refraction of the medium, and J(λ) is the spectral overlap between donor emission and acceptor absorption. κ² describes the relative orientation of the fluorophores, as defined in the Supporting Material (Fig. S1). If the transition moments are coaxial and constrained to parallel planes, κ² can vary between 0 and 1 (see Fig. S1). The photophysics of the cyanine fluorophores attached to nucleic acids have been well characterized (21–23). Significantly, when cyanine fluorophores are attached to the 5′ termini of double-stranded DNA (dsRNA) or RNA (dsRNA), they have a strong propensity to stack on the end of the duplex (24–26), whereupon their transition moments approximate to the above condition.

We have demonstrated this experimentally by measuring the efficiency of energy transfer from sulfoindocarbocyanine 3 (Cy3) to Cy5 attached to the two 5′ termini in a series of duplex species with lengths varying between 10 and 22 bp (27). Due to the helical twist, the transition dipoles of the stacked fluorophores will pass through parallel and crossed orientations twice in one complete turn, thereby modulating E_FRET as a function of length. In dsDNA, this generates peaks of E_FRET at 14 and 19 bp. These data could be well simulated using a model based on the geometry of B-form DNA, with some lateral flexibility of the fluorophores. The pattern of modulation is strongly sensitive to the helical parameters. When a DNA-RNA hybrid was used, the E_FRET-versus-length profile underwent a marked shift fully consistent with a more unwound A-form helical geometry (27). Changing the tether linking the fluorophore to dsDNA resulted in a marked phase shift in the E_FRET-versus-length profile consistent with a 30° reorientation of the cyanine fluorophores relative to the terminal basepairs (22). This was subsequently confirmed in NMR studies (26), providing proof of principle that one can extract reliable orientation information from FRET data on nucleic acids using these fluorophores. We therefore chose to apply the method to examine a branched nucleic acid structure, i.e., the structure of the IRES three-way helical junction.

Materials and Methods

Synthesis of oligonucleotides and attachment of fluorophores

RNA oligonucleotides were synthesized using t-butyldimethylsilyl (t-BDMS) phosphoramidite chemistry (28) as described in Wilson et al. (29). Oligonucleotides were terminally labeled with the 5′ trifluoroacetyl (TFA)-amino modifier C₆ phosphoramidite (Link Technologies, Bellshill UK), and sulfoindocarbocyanine fluorophores were conjugated as N-hydroxysuccinimide esters (GE Healthcare, Little Chalfont, UK), generating a total tether length of 13 atoms. All oligonucleotides were purified by gel electrophoresis in polyacrylamide, and recovered from gel fragments by electroelution or diffusion in buffer followed by ethanol precipitation. Fluorescently labeled RNA was subjected to further purification by reversed-phase high-performance liquid chromatography on a C18 column using an acetonitrile gradient with an aqueous phase of 100 mM triethylammonium acetate pH 7.0. Equimolar quantities of complementary strands were mixed in 90 mM Tris-borate (pH 7.0) and 25 mM NaCl, annealed by heating at 85°C, and cooled slowly to 4°C. The hybridized duplexes were purified by electrophoresis in 20% polyacrylamide in 90 mM Tris-borate (pH 8.3) and 25 mM NaCl, and recovered by electroelution and ethanol precipitation.

Phospholipid encapsulation

A 100:1 molar ratio of unmodified and biotinylated phospholipids was prepared by evaporating a number of aliquots of a mixture of 2.5 mg L-α-phosphatidylcholine and 35 μg 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-(cap biotinyl) (Avanti Polar Lipids, Alabaster, AL) from chloroform in a stream of argon. The aliquots were hydrated in 250 μL of 50 mM NaCl, 10 mM Tris-HCl (pH 8.1) (TN50 buffer) without (one aliquot, used for surface coating of slides) or with addition of 200 nM of a given fluorescent RNA duplex and 10 mM MgCl₂.

RNA was encapsulated in phospholipid vesicles by repeated extrusion through a polycarbonate membrane containing 200 nm pores using a mini-extruder (both from Avanti Polar Lipids), creating 200-nm-diameter unilamellar vesicles (30,31). The ratio of RNA to phospholipid resulted in most vesicles being empty, and thus most encapsulations involved a single RNA molecule. Slides were prepared by injecting RNA-free phospholipid vesicles into a narrow channel made between a quartz microscope slide and a coverslip (No. 1.5; VWR International) using double-sided adhesive tape, and left at 4°C for 1 h to allow supported bilayer formation. The sample chamber was then washed with TN50 buffer, treated with 0.2 mg/ml neutravidin (Pierce, Rockford, IL) for 10 min, and washed. A 1/20th dilution of a chosen encapsulated RNA was injected and allowed to bind to the neutravidin for 15 min. Imaging was performed under the same buffer conditions used for vesicle preparation with an oxygen-scavenging system consisting of 1.6 mg/ml glucose oxidase, 0.2 mg/ml catalase, 6% (w/v) glucose, and 1 mM trolox.

Total internal reflection single-molecule microscopy

Fluorescence intensities were acquired from single RNA duplexes or junctions encapsulated in phospholipid vesicles using prism-based total internal reflection fluorescence microscopy. A microscope quartz slide containing the sample was mounted on the stage of an inverted microscope (IX70; Olympus). The sample chamber was chilled to 8°C ± 2°C.

Fluorescent emissions were collected with the use of a 1.2 NA 60X water immersion objective (Olympus) and the incident excitation light was removed by a 550 nm long-pass filter. Donor and acceptor fluorescence emissions were separated by a 645 nm dichroic mirror (Chroma Technology). These two channels were focused side by side into a back-illuminated EMCCD camera cooled to −90°C (iXON; Andor Technology, Belfast, UK) (32). A 645 nm long-pass filter was present in the acceptor channel. Up to several hundred molecules could be recorded simultaneously using an image area of 8.2 × 8.2 mm (512 × 512 active pixels). Data were acquired at 10 frames/s⁻¹ using software provided by Andor Technology. Molecules were identified and the uncorrected intensity data for donor and acceptor channels were extracted with the use of a MATLAB (The MathWorks, Natick, MA) routine. The data for single molecules were then visualized in MATLAB, where appropriate correction factors were applied (27).

Description of the model and fitting procedures

The model incorporates parameters that describe the interaction between the fluorophores and terminal basepair. This includes the fraction of molecules stacked to RNA, the distance of each fluorophore from the terminal basepair, the angle between the transient dipole moment of the fluorophores and terminal basepair, and the standard deviation (SD) of the statistical distribution of this angle (describing the amplitude of the lateral motion). The fraction of stacked molecules was derived from fluorescence lifetime and time-resolved fluorescence anisotropy of Cy3 attached to a 5′ uridine on a dsRNA helix. The remaining three parameters were fitted to experimental data acquired on RNA duplex species using the model described below. The distance separating the fluorophore and terminal part of RNA was constrained by an accessible volume defined by the length of the linker and the hydrated size of the probe. The fitting routine was written in MATLAB using a bound-constrained simulated annealing algorithm to search for the global minimum.

The global minimum is calculated in a manner similar to that described by Ouellet et al. (22). We calculate the interfluorophore distance for each duplex R as

$$R=\left(\left(L-1\right).H\right)+D$$ (2)

where L is the length of the helix (in basepairs), H is the helical rise per basepair step, and D is the additional axial separation for the two fluorophores. The mean angle between the transition dipole moments (TDM) A is calculated as

$$\mathrm{A}=\left(\left(L-1\right).T\right)+C3\mathrm{A}+C5\mathrm{A}$$ (3)

where T is the twist angle for each basepair step, and C3A and C5A are the rotations of Cy3 and Cy5, respectively, relative to the terminal basepairs. For A-form RNA, H = 2.8 Å and T = 32°.

For each species, we generate a distribution of angles (AA) with a defined mean and SD. κ² and subsequently R₀ are calculated for each length. The value of E_FRET is calculated for each angular position (EAA), and the resulting distribution is integrated over the full interval of 360°, i.e.,

$$E_{FRET}=\underset{0°}{\overset{360°}{\int }}EAA\cdot P$$ (4)

where P is the probability distribution function of Gaussian distribution, defined as

$$P=f\left(x|\mu ,\sigma \right)=\frac{1}{\sigma \sqrt{2\pi }}{e}^{\frac{{\left(x-\mu \right)}^2}{2{\sigma }^2}}$$ (5)

where μ is the mean and σ is the SD of the angle between the TDM and the terminal basepair. The data were fitted using this model and the fixed parameters, resulting in an additional axial separation of 8 Å and rotation of Cy3 and Cy5 relative to the terminal basepairs of 31°.

These parameters acquired from the analysis of the duplex species were then applied as fixed values to the analysis of the three-way junction. For this, we used only two parameters, namely, axial extension (AE) and helical winding (HW) at the junction. The values of R and A for the junction species are defined by

$$R=\left(\left(L-1\right).H\right)+D+AE$$ (6)

and

$$\mathrm{A=}\left(\left(\mathit{L-}\mathrm{1}\right).T\right)+C3\mathrm{A}+C5\mathrm{A}+HW$$ (7)

Data sets measured on the three-way junction species were then fitted using a bound-constrained simulated annealing algorithm. To avoid falling into possible local minima, and to show the reliability of our structural interpretation, we also compared the experimental data with full-grid simulations within physically reasonable ranges of the two aforementioned parameters (see Fig. 4).

Figure 4.

Exploration of the physically reasonable range of axial extension and winding of the junction, statistically evaluated using Pearson’s χ² test. There are two minima within the range of core extension from approximately −3 Å to 3 Å and winding from −180° to 180°. Expanded cross sections of the critical region of the global minimum are shown to demonstrate the level of confidence. To see this figure in color, go online.

Results and Discussion

Constructs employed in this study

To study the conformation of the HCV IRES junction, we prepared two sets of 5′-Cy3- and 5′-Cy5-labeled species with one common strand, which formed the continuous strand that passed through the coaxially aligned C and E helices in the junction (black strand in Fig. 1 C). This strand varied between 12 and 22 nt in length (all RNA sequences are presented in the Supporting Material). Cy3 was attached to the 5′ terminus of each strand via a 13-atom linker generated by conjugating as an N-hydroxysuccinimide ester to a C₆ amino linker as described previously for DNA (22). In contrast to our earlier studies, the strand terminated with a 5′ uridine. Each strand was hybridized to its complementary RNA strand (gray in Fig. 1 C), which was similarly labeled with Cy5, to generate a set of RNA duplex species terminally labeled with donor and acceptor. The same Cy3-labeled strands were also hybridized with an alternative set of RNA strands to generate the IRES three-way junction, such that the E helix was 5′-labeled with Cy5.

Measurement of FRET efficiency

The value of E_FRET was measured for each species as single molecules encapsulated within phospholipid vesicles individually tethered to a quartz slide at 8°C, using prism-based total internal reflection to excite Cy3 (Fig. 2). E_FRET was measured from the intensities of Cy3 and Cy5 fluorescence, and plotted as a histogram for each species after correction for channel cross talk and back reflection as explained in the Supporting Material. These were fitted to Gaussian distributions from which the mean E_FRET was calculated for each species.

Figure 2.

Histograms of corrected FRET efficiency for each member of the duplex (upper panel) and junction (lower panel) series. Each panel plots the number of molecules (n) of given interval of FRET efficiency and is labeled with the length of the e strand (i.e., the length of the duplex or the sum of the lengths of the C plus E helices for the junctions, all in basepairs). All histograms comprise the actual FRET peak and a zero peak generated by missing or inactive Cy5 molecules. Both peaks are well fitted by Gaussian distributions (shown as lines).

Modulation of FRET efficiency for the RNA duplex series

The data for the duplex series are plotted in Fig. 3. The values of E_FRET are clearly modulated in a periodic manner, with a minimum at 15 bp and a maximum at 17 bp. This is very similar to the data obtained for the DNA-RNA hybrid that we presented previously (27), as would be expected since both polymers should adopt an A-form helical geometry.

Figure 3.

FRET efficiency as a function of helix length for a duplex and the HCV IRES three-way RNA junction. The error bars show the SD from the fitting of the E_FRET histograms. The data for the duplex (open squares) and junction (solid circles) have been fitted (lines) to the model described in the text, with the best fits shown.

In this study, we developed a fitting procedure that is very similar to the simulations used previously (see Supporting Material for details). This procedure is based on a model representing the standard A-form conformation of RNA, with both fluorophores stacked onto the ends of the helix with a Gaussian distribution of lateral flexibility about the mean rotation with respect to the terminal basepair. In addition, 20% of the molecules were taken to have one or both fluorophores unstacked at any given time with κ² = 0.67, based on earlier measurements of fluorescent lifetime for these species (T. Fessl and D.M.J. Lilley, unpublished data). The best fit placed both fluorophores with a rotation of their long axis (effectively the transition dipole) at 30° relative to the terminal basepair. This contrasts with our previous studies of DNA duplexes using the same combination of Cy3 and Cy5 with the 13-atom tether, where the long axes of the fluorophores and terminal basepairs were parallel (22,26). The work presented here differs from the earlier study in that the polymer is RNA, and each strand terminates in 5′ uridine; either difference might result in a reorientation of the fluorophores. The SD for the variation in fluorophore lateral position was 19°.

Analysis of the three-way junction

The data for the junction series are also plotted in Fig. 3. It is immediately apparent that there is a change in phase, but not in periodicity, between the duplex and the junction series. The junction data exhibit a minimum at 13 bp and a maximum at 15 bp. It should be emphasized that the sequences of the C-E helices and each corresponding duplex of the same length are identical; the continuous strand is common to both in each case. Therefore, the difference can only be a local change in helical winding at the junction itself. Since the entire curve for the junction is shifted toward a shorter distance, this indicates a change in rotation of the helix at the junction with the third helix in the same direction as the helical winding, i.e., an overwinding.

The data were fitted using the same geometric parameters (except for the SD of lateral motion) taken from the fitting of the duplex data, but with an additional two variable parameters corresponding to a local change in twist about the C-E helical axis and a translation of the C helix relative to the E helix along this axis. A surface showing the χ² value for the agreement between the calculated and experimental E_FRET values as these parameters are varied is shown in Fig. 4. The deepest minimum is observed for an axial extension of 1.6 Å and a winding of 29.5°. These values were used to generate the fit to the junction data in Fig. 3. There was also an increase in lateral flexibility for the junction such that the SD was 33°.

Conclusions

Our data unequivocally demonstrate that there is a substantial local overwinding of the helix at the junction. This might indicate that one of the unpaired nucleobases becomes stacked into the structure between the C and E helices, extending the end-to-end distance by ∼2 Å and leading to a rotation in the same sense as the right-handed helix, i.e., effectively an overwinding. Further structural analysis will be required to identify the origin of the change in rotation. In the meantime, this study demonstrates the power of extracting orientational information from FRET studies in nucleic acids by employing indocyanine fluorophores.