Investigating RNAs Involved in Translational Control by NMR and SAXS - Biophysical Approaches to Translational Control of Gene Expression - page 138

8.1.1 From RNA Secondary Structure Prediction

to Biophysical Studies

The first step to understanding RNA structure using any solution technique is
identification of the RNA sequence of interest and determination of its secondary
structure. Prediction of RNA secondary structure is routinely accomplished with
Mfold (Zuker 2003 ). Mfold predicts secondary structure of RNA using free energy
minimization based on thermodynamic nearest-neighbor parameters. Mfold is approx-
imately 73% accurate for RNAs less than 700 nt in length (Mathews et al. 1999 ) , and
its predictive power greatly improves for smaller RNAs. While the Mfold algorithm
alone cannot predict base-pairs involved in pseudoknot formation, several alternative
programs have been developed that can (Theis et al. 2008 ; Andronescu et al. 2005 ;
Hart et al. 2008 ). The accuracy of secondary structure prediction by Mfold can be
improved by incorporating experimental data from sources such as chemical probing
(Mathews et al. 2004 ), NMR (Hart et al. 2008 ), and/or microarray (Kierzek et al.
2006 ) experiments. Of these, NMR provides the most direct and rigorous experimen-
tal verification of secondary structure, since it is the only method that can directly
detect the hydrogen-bonded protons involved in base-pair formation. For larger RNA
sequences, accurate prediction of secondary structure is more challenging due to
increased base-pairing possibilities leading to alternative secondary structures with
similar overall free energies. Prediction of these large secondary structures can be
assisted by the use of phylogenetic analysis (Juan and Wilson 1999 ; Mathews et al.
2010 ) and chemical probing (Mathews et al. 2004 ; Weeks 2010 ; Weeks and Mauger
2011 ). If the sequence of interest is derived from a larger RNA, it is critical to verify
that the secondary structure of the RNA was not perturbed due to the truncation. This
can be demonstrated with NMR, by confirming that the chemical shifts for each sub-
domain are consistent with those observed in the context of the larger RNA. This
approach was recently used to examine the secondary structure of a 356 nt HIV-1
genome packaging 5¢ leader RNA (Lu et al. 2011 ) .
In vitro transcription with T7 RNA polymerase is standard for production of the
milligram quantities of RNA required for NMR (Milligan and Uhlenbeck 1989 ;
Hennig et al. 2001 ; Scott and Hennig 2008 ). The majority of the DNA template for
transcription can be single-stranded, but the 20 base-pair T7 RNA polymerase pro-
moter region must be double stranded, as described (Milligan and Uhlenbeck 1989 ) .
Chemical synthesis of the DNA template strand with phosphoramidite chemistry is
customary for synthesis of oligonucleotides 90 nts in length or smaller. For DNA
constructs longer than 90 nts, standard cloning techniques (Sambrook and Russell
2001 ) are used to insert the DNA duplex into a DNA plasmid. Once milligram quan-
tities of plasmid DNA are produced, the plasmid is linearized with a restriction
enzyme that cuts immediately following the end of the DNA template sequence
(Hart et al. 2008 ). This linearized product is then used for in vitro transcription.
Additionally, T7 RNA polymerase requires at least one but preferably two guanos-
ine residues at the 5¢ end of the RNA construct for initiation of transcription (Milligan
and Uhlenbeck 1989 ) .