General procedures

4.3.1 Sample preparation for NMR, CD and UV melting studies

The wildtype d3'-EBS1 (5'-GGAGUAUGUAUUGGAAAUGAGCAUACUCC-3'), d3'- EBS1* (5'-GGAGUAUGUAUUGGCACUGAGCAUACUCC-3') and d3'-TL (5'- GGAGUAUGUGAAAGCAUACUCC) were synthesized by in vitro transcription with T7 RNA polymerase from synthetic double stranded DNA oligonucleotide templates.(292) The reaction mixture contained 5 mM of each NTP, 0.9 µM of the double stranded DNA template, 0.1% Triton X-100, 40 mM Tris-HCl (pH 7.5), 40 mM DTT, 2 mM spermidine and 30 mM MgCl2 for d3'-EBS1, 35 mM MgCl2 for d3'-EBS1* and 50 mM MgCl2 for d3'-TL,

respectively. The amount of T7-polymerase was optimized individually for each polymerase batch. N (29 nucleotides) and n+1 (30 nucleotides) lengths RNAs were obtained in a ratio of approximately 60:40, and care was initially taken to separate them. It was found that the combined RNA species did not affect the chemical shifts in 2D [1H,1H]-NOESY spectra and thus also not the overall structure. Natural isotope abundance as well as fully 15N,13C - enriched samples of all three constructs were transcribed and used for NMR measurements. d3'-EBS1* was also transcribed with selectively deuterated NTPs. The RNA was purified by

Materials and Methods 161

denaturing 15% (w/v) polyacrylamide gel electrophoresis (PAGE), UV-shadowed and excised from the gel, followed by electroelution and ethanol-precipitation. In order to receive a hairpin conformation instead of a duplex, the RNA was dissolved in 100 ml 85°C ddH2O, left

for one minute and cooled down quickly on ice. The RNA was concentrated using Vivaspin 20 at 14 °C and 4'500 g and washed with ddH2O. All samples were lyophilized, resolved in

D2O (200 µl) containing KCl (10 mM for d3'-EBS1* and d3'-TL, respectively, and 110 mM

for d3'-EBS1*·IBS1*) and EDTA (10 µM). The RNA concentrations were determined with a Varian Cary 500 Scan UV-VIS-NIR spectrophometer, using an extinction coefficient at 260 nm (ε260) 341.8 mM–1cm–1 for d3'-EBS1, 325.8 mM–1cm–1 for d3'-EBS1*, 257.7 mM–1cm–1

for d3'-TL, 66 mM–1cm–1 for IBS1, 91.5 mM–1cm–1 for EBS1, 73.6 mM–1cm–1 for IBS1*, 75.5 mM–1cm–1 for EBS1* and 71.1 mM–1cm–1 for IBS1GC. The concentrations of the RNA samples for NMR measurements varied between 0.4 and 1.2 mM. The pH was adjusted to 6.4 for samples in H2O/D2O and to pD 6.8 for samples in D2O, using DCl or NaOD solutions. To

measure the pD value, 0.4 log units were added to the pH meter reading.(73,396) All samples were lyophylized and resuspended in either H2O/D2O (90:10) or 100% D2O prior to the

acquisition of the NMR spectra. Samples for CD measurements and UV melting studies were resuspended in ddH2O.

4.3.2 UV melting studies

Temperature dependent absorption measurements were performed at 260 nm in quartz cuvettes with 1 mm path length (volume 200 µl). Before measuring the samples were degassed for 30 sec and carefully covered with paraffin oil. The melting experiments were carried out by constantly raising the temperature from 5 °C to 80 °C (for EBS1·IBS1, EBS1·IBS1-GC and EBS1*·IBS1*), 5 °C to 90 °C (for d3'-EBS1*·IBS1*) and 15 °C to 75 °C (d3'-EBS1*), with a heating rate of 0.5 °C/min. Absorption spectra were recorded every 0.5 °C during three heating and three cooling cycles. Samples for concentration dependent UV measurements contained 1.0, 1.7, 3.5, 5.8, 13.3, 19.0, 36.5 and 70.1 µM d3'-EBS1* at pH 6.8. RNA concentrations for the other samples varied between 7.3 and 10.9 µM. Sodium ion concentrations ranged from 10 – 200 mM. All spectroscopic data were analysed using Origin® version 6.0 (OriginLabTM Corporation). The van't Hoff analysis was performed with the Hyperchromicity calculation – Thermal, which is part of the Cary WinUV package, with a non-self complementary, bimolecular function for EBS1*·IBS1* and with a self complementary monomolecular function for the d3'-stem.

4.3.3 Circular dichroism

CD spectra were recorded over the spectral range of 200-320 nm (3 spectra accumulations) and a scan speed of 50 nm min–1. A quartz cuvette with 1 mm pathlength and 200 µl volume was used. For the metal ion titration studies a solution of 16 µM d3'-EBS1* with 10 mM KCl or 100 mM KCl, respectively, pH 6.8 (solution A), and a solution of 16 µM d3'-EBS1* with 10 mM KCl or 100 mM KCl, respectively, 1 mM MgCl2, pH 6.8 (solution

B), were prepared. The titration was done by removing small aliquots of solution A and adding the equivalent aliquot of solution B, so that the total sample volume remained constant. The same procedure was performed for the titration of d3'-EBS1*·IBS1* with Mg2+, but in these experiments equivalent amounts of d3'-EBS1* and IBS1* (16 µM) were used. For the titrations with IBS1* small aliquots were added to solution A in the absence and presence of 320 µM Mg2+, respectively, up to a final concentration of 32 µM IBS1*. All titrations were measured at 20 °C. The titrations with Cd2+ were performed analogously. For the titration of the d3'-EBS1 with IBS1 a 10 µM d3'-EBS1 solution (100 mM KCl, pH 6.8) was used and small aliquots of IBS1 were added up to a final concentration of 20 µM. Temperature dependent CD measurements were carried out between 20 °C and 84 °C with a heating rate of 1 °C/min and three accumulation cycles at each step.

4.3.4 Dynamic Light Scattering measurements

Dynamic light scattering (DLS) is a good tool to measure the hydrodynamic radius of molecules. DLS was measured at 293 K and 298 K in a 12 µl cuvette using the conditions from the NMR experiments (samples were in D2O, pD 6.8, 10 mM KCl for d3'-TL and d3'-

EBS1*, 100 mM KCl for EBS1*IBS1* and 110 mM KCl for d3'- EBS1*IBS1*). Samples were measured at concentrations of 0.25 mM or 0.5 mM, respectively. Before measuring, the samples were centrifuged for 45 min at 4 °C and 16'100 g prior to the experiment in order to avoid any dust that might hamper DLS measurements. At least ten measurements with ten acquisitions of each sample were recorded.

4.3.5 Preparation of the Pf1 aligned NMR samples to measure RDCs

For RDC measurements, d3'-EBS1* and d3'-EBS1*·IBS1* were aligned in 25.6 mg/ml Pf1 phage. First the magnetic alignment of the phage was checked by measuring the splitting of the HOD signal from the solvent. Therefore, a phage sample only of phages in 90%

Materials and Methods 163

H2O/10% D2O was prepared and a 1D 2H NMR spectrum was recorded at 298 K. The splitted

signal comes from the large deuterium quadrupole moment that is not isotropically averaged for water bound to the aligned phage particles and directly correlates with the ability of the phage to align RNA.(319) For a 25.6 mg/ml concentrated sample a splitting of 24.2 Hz was found for the phage alone. One sample with and one without phages were prepared for d3'- EBS1* or d3'-EBS1*·IBS1*, respectively. The d3'-EBS1* sample without phages was lyophylized and redissolved in 200 µl of 90% H2O/10% D2O to a final RNA concentration of

0.69 mM, including KCl for a final salt concentration of 10 mM KCl and EDTA for a final concentration of 10 µM. The pH was adjusted to 6.85. A droptest with 2 µl of the RNA sample and 1 µl of 50 mg/ml Pf1 phage showed no precipitation. Therefore the second lyophylized sample of d3'-EBS1* was prepared with phages by adding 156 µl of 90% H2O/10% D2O, adjusting the pH to 6.53 and 164 µl of 50 mg/ml Pf1 phages were added to a

final volume of 320 µl. The final RNA concentration was 0.69 mM, salt concentrations were 10 mM KCl and 10 µM EDTA. The final RNA concentration of both samples of d3'- EBS1*·IBS1* was 0.48 mM, KCl concentrations were 50 mM and EDTA concentrations 10 µM. The phage concentration was 25.6 mg/ml in a volume of 320 µl. The pH for d3'- EBS1*·IBS1* without phages was set to 6.45 and with phages to 6.67. The droptest for d3'- EBS1*·IBS1* without phages also showed no precipitation.

4.3.6 NMR spectroscopy

Non-exchangeable proton resonances were assigned from 2D [1H,1H]-NOESY spectra in D2O acquired at 60, 120 and 250 ms mixing times at 288 K, 293 K, 298 K and/or 303 K. 2D

[13C,1H]-HSQCs were recorded separately for the aromatic (sw = 70 ppm, O1 = 135 ppm) and the aliphatic (sw = 120 ppm, O1 = 50 ppm) range of the 13C resonances. 2D [1H,1H]-TOCSY spectra at 45 ms mixing time were used to clarify the sugar pucker conformation. In A-form RNA the sugar pucker conformation is usually 3'-endo and can be distinguished from the 2'- endo form, which gives a strong crosspeak between H1' and H2' and even between H1' and the H3' of the ribose sugar in a 2D [1H,1H]-TOCSY spectrum. Natural abundance 2D [13C,1H]-HSQCs were acquired for the EBS1*·IBS1* construct at 700 MHz over 24 hours.

Exchangeable proton resonances were assigned from 2D [1H,1H]-NOESY spectra in 90% H2O/10% D2O acquired at 278 K, 283 K and 293 K with mixing times of 150 ms and

watergate H2O suppression. Assignment was supported by [15N,1H]-HSQC experiments. The

the imino nitrogen of uracil (N3) and guanine (N1) across the H-bond to the N1 of adenine or the N3 of cytosine on the other side of the double helix.

All 1H spectra were recorded with RNA constructs of natural isotope abundance and experiments involving 13C and 15N isoptopes were recorded with fully 15N,13C-enriched samples, except for EBS1*·IBS1*.

One-bond 1H-13C RDCs were measured from the splitting of the peak along the carbon dimension of 2D [13C,1H]-HSQCs (separately for the aromatic and the sugar region) at 298 K and one-bond 1H-15N RDCS in the nitrogen and proton dimensions of 2D [15N,1H]-HSQCs at 278 K for d3'-EBS1* and d3'-EBS1*·IBS1* in isotropic (unaligned) and Pf1-containing (aligned) solution (for sample conditions see section 4.3.5). The couplings for the aromatic (H2, H5, H6 and H8), for H1' sugar protons as well as the imino H1 and H3 protons were extracted with Sparky by determining the difference between 1H-13C and 1H-15N for isotropic and partially aligned samples.

For d3'-EBS1*·IBS1* with d3'-EBS1* hairpin 15N,13C labeled and IBS1* at natural abundance, as well as for a mixture of 15N,13C labeled and unlabeled d3'-EBS1* in a 1:1 ratio, double X half-filtered NOESY-HSQC spectra were recorded at 278 K with a 15N filter in F1 (with watergate H2O suppression) at 278 K with 16 scans and 2048 experiments in F3, 32

experiments in F2 and 132 experiments in F1. The used 3D pulse sequence from the Bruker pulse program library was noesyhsqcf3gpwx13d. The acquisition time was 73 ms, the coupling constants 1J(N-H) (cnst4), 1J(C-H)min (cnst6) and 1J(C-H)max (cnst7) were set to 90

Hz, 160 Hz and 200 Hz, respectively. The relaxation delay (D1) was 1 s, the delay for homospoil/gradient recovery (D16) 200 µs and the mixing time 150 ms.

The 2D X-filter experiments of d3'-EBS1*·IBS1* (d3'-EBS1* labeled, EBS1* unlabeled) were recorded using a phase sensitive 2D w1,w2-15N,13C -filtered NOESY experiment (Bruker

pulse program: noesygpphwgxf.2) with watergate H2O suppression as well as a TOCSY

version of it (Bruker pulse program: dipsigpphwgxf.2).(325-328) These experiments were recorded at 298 K in 90% H2O/10% D2O. The NOESY experiment was recorded with 64

scans and 2048 experiments in F2 as well as 256 experiments in F1 and the TOCSY experiment with with 32 scans and 2048 experiments in F2 as well as 256 experiments in F1. For the NOESY as well as for the TOCSY experiment, the same parameters as for the double X half-filtered NOESY-HSQC experiment (see above) were used.

To calculate the hydrodynamic radius of EBS1*·IBS1*, d3'-TL, d3'-EBS1* and d3'- EBS1*·IBS1* DOSY experiments were acquired. The used 2D pulse sequence from the Bruker pulse program library was stebpgp1s, employing stimulated echo and bipolar gradient

Materials and Methods 165

pulses for diffusion. The diffusion time (∆), the gradient length (δ) and the recovery delay after gradient pulses were set to 350 ms, 2 ms and 200 µs, respectively, and the gradient strength was incremented from 11.8 to 32 G/cm in 64-80 steps. For each FID 64 scans were collected with 21006 data points in F2. The diffusion coefficients were calculated using the DOSY routine of Topspin 1.3. For the calculation of the hydrodynamic radii of the molecules the Stokes-Einstein equation was used:

6 kT D r π η = equation 19

and rearrangement of equation 19 yields:

6 kT r D π η = equation 20

where k is the Boltzmann constant with 1.381·10–23 NmK–1, T the temperature in K at which the experiment was acquired (here 293 K or 298 K, respectively), η is the dynamic viscosity of the solvent (1·10–3 Nsm–2 for H2O at 293 K and 0.89·10–3 Nsm–2 for H2O at 298 K), D is

the diffusion coefficient, which was obtained from the DOSY experiment in m2s–1.

All spectra were processed with XWINNMR and Topspin 1.2, 1.3 or 2.0 (Bruker). 1D spectra were analyzed with MestreC (http://www.mestrec.com/) or with Topspin 1.2, 1.3 or 2.0. Sparky (http://www.cgl.ucsf.edu/home/sparky/) was used for multidimensional spectra assignment. NOE peak volumes were integrated with the Gaussian peak fitting function in Sparky.