A fast-acting reagent for accurate analysis of RNA secondary and tertiary structure by SHAPE chemistry

Article abstract of DOI:10.1021/ja0704028

Selective 2′-hydroxyl acylation analyzed by primer extension (SHAPE) chemistry allows local nucleotide flexibility to be quantitatively assessed at single nucleotide resolution in any RNA. SHAPE chemistry exploits structure-based gating of the nucleophili

Full text of DOI:10.1021/ja0704028

Published on Web 03/17/2007
A Fast-Acting Reagent for Accurate Analysis of RNA Secondary and Tertiary
Structure by SHAPE Chemistry
Stefanie A. Mortimer and Kevin M. Weeks*
Department of Chemistry, UniVersity of North Carolina, Chapel Hill, North Carolina 27599-3290
Received January 18, 2007; E-mail: weeks@unc.edu
RNA sequences fold back on themselves to form structures that
are difficult to predict, especially if only a single sequence is
known.1 Current algorithms correctly predict 50-70% of known
base pairs on average.2 Predicted secondary structure models
achieving 50-70% accuracy tend to have regions in which the
overall topology differs significantly from the correct one, making
it difficult or impossible to develop robust biological hypotheses.
Knowledge of which nucleotides are likely to be paired or single-
stranded can significantly improve prediction accuracies.3
The ideal technology for chemically assisted RNA structure
analysis would (1) be experimentally straightforward, (2) use a
single reagent that reacts generically with all four nucleotides, (3)
employ a self-quenching reagent, (4) involve short reaction times,
and (5) be proven to yield accurate results with complex RNAs of
reactivity of the 2'-hydroxyl group is selectively enhanced at flexible
known structure.
Figure 1. Mechanism of RNA SHAPE chemistry. (A) The nucleophilic
positions.4 (B) Parallel reaction of N-methylisatoic anhydride derivatives
with RNA 2′-hydroxyl groups and with water. [reagent]0 was 5 mM.
Selective 2′-hydroxyl acylation analyzed by primer extension
(SHAPE) chemistry4 takes advantage of the discovery that the
nucleophilic reactivity of a ribose 2′-hydroxyl group is gated by
local nucleotide flexibility. At nucleotides constrained by base
pairing or tertiary interactions, the 3′-phosphodiester anion and other
interactions reduce reactivity of the 2′-hydroxyl.4a In contrast,
flexible positions preferentially adopt conformations (Figure 1A)
that react with N-methlyisatoic anhydride (1, NMIA) to form a 2′-
O-adduct (Figure 1B). NMIA reacts generically with all four
nucleotides, and the reagent undergoes a parallel, self-inactivating,
hydrolysis reaction (Figure 1B). Thus SHAPE chemistry meets the
first three of the five criteria outlined above.
Figure 2. Comparative reactivity of 1M7 and NMIA via hydrolysis (left)
and 2′-O-adduct formation with pAp-ethyl (right).
However, NMIA is relatively unreactive and requires tens of
minutes to react to completion. To address the final two criteria
for a near-ideal RNA structure interrogation technology, we design
a significantly more useful, fast acting, reagent for SHAPE
chemistry. We then show that structural constraints obtained using
this reagent allow the secondary and tertiary structure of a large
RNA to be assessed with high accuracy.
The reactive carbonyl of NMIA is in the benzylic position relative
to the aromatic ring system and should be sensitive to substituents
in the meta or the para positions, owing to a direct resonance effect.
We therefore evaluated the 2′-O-adduct-forming and hydrolysis
activities of 1-methyl-7-nitroisatoic anhydride (2, 1M7) (Figure 1B).
The para nitro substituent is strongly electron-withdrawing (σp )
0.81)5 and should increase adduct formation and hydrolysis rates
in two ways: via a ground-state effect by increasing the electro-
philicity of the reactive carbonyl and via a transition-state effect
by stabilizing the negative charge in the developing tetrahedral
reaction intermediate. We first monitored reagent hydrolysis as the
increase in UV absorbance of the aminobenzoate products. 1M7 is
significantly more labile toward hydrolysis than NMIA. 1M7
undergoes hydrolysis with a half-life of 14 s and therefore the
reaction is complete in ∼70 s; in contrast, NMIA requires over
20 min to react to completion (left panel, Figure 2).
ethyl contains a 2′-hydroxyl and 3′-phosphodiester monoanion and
is a good analogue for an unstructured RNA nucleotide.4a 1M7
reacts significantly more rapidly with pAp-ethyl than does NMIA;
however, the final extent of 2′-O-adduct formation for the two
compounds is identical, within error (right panel, Figure 2).
Identical extents of reaction for NMIA and 1M7, despite the
much faster reactivity of 1M7, indicate that the rates of hydrolysis
and of 2′-hydroxyl acylation have increased by precisely the same
20-fold increment. These experiments indicate that 1M7 has the
ideal chemical characteristics for a fast-acting and self-quenching
reagent for RNA SHAPE chemistry.
We next evaluated the extent to which 1M7 provides accurate
and quantitative information regarding RNA structure using the
specificity domain of the Bacillus subtilis RNase P enzyme. This
domain was chosen because it is a large (154 nt) RNA with a known
structure6 that does not contain pseudoknots, which are not well
predicted by current algorithms. This RNA spans numerous typical
base-pairing and stacking interactions, a tetraloop-receptor tertiary
interaction (involving L12 and P10.1) common to many large
RNAs, and two large internal loops (J11/12 and J12/11) stabilized
by an extensive series of noncanonical interactions.6
A SHAPE experiment was performed on the RNase P domain
under conditions that stabilize the native tertiary fold (6 mM MgCl2,
100 mM NaCl, pH 8.0) by treating the RNA with 6.5 mM 1M7.
We next evaluated the ability of each compound to react with
3′-phosphoethyl-5′-adenosine monophosphate (pAp-ethyl). pAp-
9
4144
J. AM. CHEM. SOC. 2007, 129, 4144-4145
10.1021/ja0704028 CCC: $37.00 © 2007 American Chemical Society

C O M M U N I C A T I O N S
Positive and negative peaks thus indicate an increase or decrease
in local nucleotide flexibility in the absence of Mg2+, respectively
(lower panel, Figure 3). Many sites in the (-) Mg2+ experiment
show increased SHAPE reactivity. Strikingly, increased reactivity
occurs precisely at nucleotides that participate in tertiary interactions
in the RNase P domain (in red, Figure 4B). SHAPE reactivity also
shows that the irregularly stacked P7-P10-P11 helical domain
unfolds when Mg2+ is removed (Figure 4B).
We then evaluated how well SHAPE information can be used
to constrain the output of an RNA secondary structure prediction
algorithm. We calculated prediction accuracies both using the native
secondary structure (Figure 4A) as the target and using a modified
structure that excluded the Mg2+-dependent base pairs in the P7-
P10-P11 domain. When the specificity domain of Bacillus subtilis
RNase P is folded in RNAstructure,3d the lowest free energy
structure contains 52% of the correct pairs and features an overall
topology that is radically different from the correct structure (Figure
S3A). When SHAPE reactivity information is added to constrain
single-stranded and non-internal base pairs, the lowest free energy
structure is 76% correct using the native secondary structure as
the target and 91% correct when base pairs in the P7-10-P11 domain
(which do not form in the absence of native tertiary interactions)
are excluded (Figure S3B). Using either target structure, the
SHAPE-constrained prediction features an overall topology that
closely resembles the correct structure.
Figure 3. Histograms and difference plot of absolute nucleotide reactivities
for SHAPE experiments performed with 1M7 in the presence and absence
of 6 mM Mg2+
.
SHAPE chemistry performed with 1M7 accurately reports the
known structure of the RNase P specificity domain under native
conditions. 1M7 reactivity detects nucleotides constrained both by
base pairing and by idiosyncratic, noncanonical tertiary interactions
(Figure 4). SHAPE chemistry enables very precise analysis of the
differences between two structures, such as Mg2+-dependent tertiary
interactions. 1M7 is easily handled in the laboratory and enables
analysis of large RNA structures at single nucleotide resolution in
less than 70 s.
Figure 4. Base pairing and tertiary interactions for the specificity domain
of Bacillus subtilis RNase P.
Sites of 2′-O-adduct formation were identified as stops to primer
extension, using fluorescently labeled DNA primers, resolved by
capillary electrophoresis7 (Figure S1). Absolute SHAPE reactivities
were calculated by subtracting the background observed in
no-reagent control experiments that omitted 1M7. Reactivity at each
nucleotide was classified as high, medium, low, or near-zero (red,
orange, blue, and black columns, respectively, in Figure 3).
Superposition of the quantitative reactivity information on a
secondary structure diagram6 for the RNase P specificity domain
shows that a 70 s reaction with 1M7 accurately reports the known
secondary and tertiary structure for this RNA (Figure 4A).
Essentially all nucleotides involved in Watson-Crick base-pairs
are unreactive; moreover, many noncanonical, but stable, U‚G, A‚
A, and A‚G pairs are unreactive. Nucleotides in P10.1 and in L12
that form the tetraloop-receptor tertiary structure motif are also
unreactive. In contrast, nucleotides in loops or adjacent to bulges
or other irregularities are reactive. Nucleotides in the structurally
idiosyncratic module involving J11/12 and J12/11 show a wide
range of reactivities. Strikingly, the most highly conserved nucleo-
tides in this module (A187, A191, G219-G220, A222), which
participate in stabilizing tertiary interactions,6 also show the lowest
SHAPE reactivities using 1M7.
We performed a similar SHAPE experiment in the absence of
magnesium ion (middle histogram, Figure 3). Control experiments
show that both reaction with the model nucleotide, pAp-ethyl, and
1M7 hydrolysis are independent of Mg2+ concentration (Figure S2).
This Mg2+-independence represents an additional significant im-
provement over the parent compound, NMIA, whose reactivity is
strongly dependent on ionic strength (Figure S2). Thus, observed
changes in SHAPE reactivity with 1M7 reflect changes in RNA
secondary and tertiary structure and not Mg2+-induced differences
in reagent properties.
Acknowledgment. We are indebted to T. Pan and A. Mon-
dragon for a B. subtilis RNase P specificity domain plasmid, to J.
Chattopadhyaya for adenosine 3′-ethyl phosphate, and to J. Morken
for helpful discussions. This work was supported by a grant from
the NSF (Grant MCB-0416941 to K.M.W.).
Supporting Information Available: Experimental procedures and
three figures. This material is available free of charge via the Internet
at http://pubs.acs.org.
References
(1) (a) Tinoco, I.; Bustamante, C. J. Mol. Biol. 1999, 293, 271-281. (b) Eddy,
S. R. Nat. Biotechnol. 2004, 22, 1457-1458. (c) Doshi, K. J.; Cannone,
J. J.; Cobaugh, C. W.; Gutell, R. R. BMC Bioinformatics 2004, 5, 105.
(2) (a) Dowell, R. D.; Eddy, S. R. BMC Bioinformatics 2004, 5, 71. (b)
Mathews, D. H.; Turner, D. H. Curr. Opin. Struct. Biol. 2006, 16, 270-
278.
(3) (a) Peattie, D. A.; Gilbert, W. Proc. Natl. Acad. Sci. U.S.A. 1980, 77,
4679-4682. (b) Ehresmann, C.; Baudin, F.; Mougel, M.; Romby, P.; Ebel,
J.-P.; Ehresmann, B. Nucleic Acids Res. 1987, 15, 9109-9128. (c) Knapp,
G. Methods Enzymol. 1989, 180, 192-212. (d) Mathews, D.; Disney, M.;
Childs, J.; Schroeder, S.; Zuker, M.; Turner, D. Proc. Natl. Acad. Sci.
U.S.A. 2004, 101, 7287-7292. (e) Badorrek, C. S.; Weeks, K. M. Nature
Chem. Biol. 2005, 1, 104-111.
(4) (a) Merino, E. J.; Wilkinson, K. A.; Coughlan, J. L.; Weeks, K. M. J.
Am. Chem. Soc. 2005, 127, 4223-4231. (b) Wilkinson, K. A.; Merino,
E. J.; Weeks, K. M. J. Am. Chem. Soc. 2005, 127, 4659-4667. (c)
Wilkinson, K. A.; Merino, E. J.; Weeks, K. M. Nature Protocols 2006,
1, 1610-1616.
(5) Exner, O. Correlation Analysis in Chemistry; Plenum Press: New York,
1978.
(6) Krasilnikov, A. S.; Yang, X.; Pan, T.; Mondragon, A. Nature 2003, 421,
760-764.
(7) Wilkinson, K. A.; Gorelick, R. J.; Vasa, S. M.; Guex, N.; Rein, A.;
Mathews, D. H.; Giddings, M. C.; Weeks, K. M., submitted for publication,
2007.
We quantified the effect of Mg2+ on the structure of the RNA
using a difference plot in which nucleotide reactivities in the (+)
Mg2+ experiment were subtracted from the (-) Mg2+ experiment.
JA0704028
9
J. AM. CHEM. SOC. VOL. 129, NO. 14, 2007 4145