RNA molecules can fold into noncanonical structures such as the four-stranded structures known as G-quadruplexes. G-quadruplexes in the transcriptome have recently emerged as relevant regulatory elements of gene expression. Conformational transitions in RNA molecules offer an important way to regulate their biological functions. Here we report on the competition between a canonical hairpin structure and a G-quadruplex structure within an RNA molecule. We show that the conformational preference strongly depends on the relative amounts of mono- and divalent metal ions present in solution. In our system, the G-quadruplex, whose formation is not predicted by available predictive RNA folding programs, is the major conformer at physiologically relevant K(+) and Mg(2+) concentrations. Furthermore, we show that a synthetic small molecule can displace the structural dynamic equilibrium in favor of the hairpin conformer. This work highlights a new and important level of complexity in RNA folding that could be relevant to the biological functions and targeting of RNAs comprising G-quadruplex motifs.
RNA molecules can fold into noncanonical structures such as the four-stranded structures known as G-quadruplexes. G-quadruplexes in the transcriptome have recently emerged as relevant regulatory elements of gene expression. Conformational transitions in RNA molecules offer an important way to regulate their biological functions. Here we report on the competition between a canonical hairpin structure and a G-quadruplex structure within an RNA molecule. We show that the conformational preference strongly depends on the relative amounts of mono- and divalent metal ions present in solution. In our system, the G-quadruplex, whose formation is not predicted by available predictive RNA folding programs, is the major conformer at physiologically relevant K(+) and Mg(2+) concentrations. Furthermore, we show that a synthetic small molecule can displace the structural dynamic equilibrium in favor of the hairpin conformer. This work highlights a new and important level of complexity in RNA folding that could be relevant to the biological functions and targeting of RNAs comprising G-quadruplex motifs.
The biological functions of
RNA molecules generally depend on their folding properties.[1] Some RNAs can adopt several stable folds that
coexist in an intramolecular conformational equilibrium.[2,3] Changes in RNA conformation in response to fluctuations in intracellular
conditions provide a mechanism that can regulate gene expression.[3,4] This is exemplified by naturally occurring RNA regulatory elements
that respond to extrinsic stimuli such as temperature, pH, and ionic
conditions or to interacting trans-acting factors such as proteins,
nucleic acids, and small-molecule metabolites.[3,4]Besides canonical purine (Pu)–pyrimidine (Py) helix-based
structural elements, some RNA sequences may also adopt non-canonical
base-pairing schemes leading to alternative folds that are fundamentally
different from helix-based RNA structures and would be unexpected
if only Pu/Py base pairs are considered. This includes guanine (G)-rich
four-stranded structures, called G-quadruplexes, which arise from
Hoogsteen-type hydrogen bonds between Gs.[5] RNA G-quadruplexes have recently emerged as motifs that can regulate
gene expression and have been implicated in several aspects of RNA
metabolism such as transcription termination, splicing, and translation.[6] Furthermore, they have also been used as switchable
modules that allow artificial control of the function of attached
RNA sequences.[7] Therefore, it is becoming
increasingly important to obtain a greater understanding of the properties
of such RNA folds.Biophysical studies on RNA G-quadruplexes
published to date have
focused on small pieces of G-rich RNAs that unambiguously fold into
G-quadruplex structures[8] but have not addressed
the question of RNA G-quadruplex formation in competition with helix-based
structures, which could occur in longer stretches of RNA. Herein we
report the first description of the competitive formation of a helix-based
hairpin structure and a G-quadruplex structure within an RNA molecule.We designed an RNA molecule, HpQd, that has the potential to fold
into two mutually exclusive intramolecular structures: a hairpin structure
(Hp) and a G-quadruplex structure (Qd) (Figure 1). The hairpin-forming sequence was derived from a previously documented
hairpin structure belonging to the VPK pseudoknot structure in the
mouse mammary tumor virus, whose the thermodynamic stability strongly
depends on the Mg2+ concentration.[9] The G-quadruplex-forming sequence was derived from a G-quadruplex
motif previously identified within the 5′-UTR of humanN-RAS messenger RNA.[10] To design
HpQd, we introduced (i) three G/C base permutations within the stem-forming
sequence of the wild-type VPK hairpin sequence and (ii) two mutations
(A to U and G to C) within the third loop of the wild-type NRAS G-quadruplex
(Figure S1 in the Supporting Information). These modifications adapted the wild-type sequences in such a
way that the mutated sequences (denoted as Hp′ and Qd′,
respectively) share an identical 11 nucleotide (nt) sequence element
(rGCGGGAGUGGG), leading to structural ambivalence between the two
conformers within HpQd (Figure 1 and Figure S1). Circular dichroism (CD)[12] and thermal difference[13] spectra of the individual Hp′ and Qd′ reference sequences
(Figure 1) confirmed that they retained the
hairpin and the G-quadruplex structures, respectively, adopted by
the parent RNA sequences (Figure S2). It
is notable that even though HpQd can possibly fold into a G-quadruplex
structure, available RNA folding prediction programs such as the popular
MFold[11] do not yet include a consideration
of Hoogsteen base pairs and thus cannot predict G-quadruplex folds
(Figure S3).
Figure 1
A bistable RNA conformational
switch (HpQd) under cation-mediated
control.
A bistable RNA conformational
switch (HpQd) under cation-mediated
control.We then subjected the Hp′ and Qd′
references to UV
melting experiments. The melting temperatures obtained in 10 mM sodium
cacodylate (pH 7.0) containing 1 mM KCl were 71 ± 1 and 62 ±
1 °C, respectively. Van’t Hoff analysis of the melting
profiles[14] indicated that the thermodymamic
stabilities of the two structures differed by ∼5 kJ mol–1, suggesting an 85:15 equilibrium in favor of Hp formation
over Qd formation within the HpQd sequence. Under the same conditions,
the melting profiles of HpQd recorded at 260, 280, and 295 nm all
displayed a reversible hyperchromic sigmoidal curve with a transition
point at 70 ± 1 °C (Figure S4), in agreement with the hairpin structure being the major conformer.[15]However, the preferential formation of
an RNA structure can strongly
depend on the environmental conditions, in particular the relative
abundance of metal ions in solution.[16] For
example, the thermodynamic stability of RNA hairpin structures has
been reported to be dependent on the concentration of divalent cations,
especially Mg2+ (the most abundant divalent cation in the
cytoplasm of a cell), whereas G-quadruplex formation generally depends
on the concentration of monovalent cations, especially K+ (the most abundant monovalent cation in the cytoplasm of a cell).
For DNA, there was an early report by Hardin et al.[17] on a cation-dependent equilibrium between a DNA hairpin
and a tetramolecular G-quadruplex from a d(CGCG3GCG) oligonucleotide
showing that K+ ions strongly favor the G-quadruplex fold.
In addition, short oligonucleotides mimicking the G-rich DNA strands
of the fragile X chromosome repeats have been shown to form either
hairpins or tetraplexes depending on the concentrations of Na+ and K+ ions and the temperature.[18]UV melting studies of the Qd′ reference revealed
that the
melting temperature of the G-quadruplex was strongly influenced by
the concentration of K+ in solution, whereas the Mg2+ concentration had no or little effect (Figure 2). On the other hand, as expected, the concentration of Mg2+ had a strong effect on the thermal stability of the Hp′
hairpin structure, but it had only a moderate effect on the Qd′
G-quadruplex (Figure 2). These initial data
suggested a possible cation-dependent competitive hairpin–quadruplex
equilibrium within the HpQd RNA molecule.
Figure 2
Influence of the K+ and Mg2+ concentrations
on the thermal stabilities of the hairpin (Hp′) and G-quadruplex
(Qd′) RNA structures. Denaturation studies were performed in
10 mM sodium cacodylate (pH 7.0) supplemented with (a) increasing
amounts of KCl (K+ dependence) and (b) 1 mM KCl and increasing
amounts of MgCl2 (Mg2+ dependence).
Influence of the K+ and Mg2+ concentrations
on the thermal stabilities of the hairpin (Hp′) and G-quadruplex
(Qd′) RNA structures. Denaturation studies were performed in
10 mM sodium cacodylate (pH 7.0) supplemented with (a) increasing
amounts of KCl (K+ dependence) and (b) 1 mM KCl and increasing
amounts of MgCl2 (Mg2+ dependence).We then used 1D 1H NMR spectroscopy
to investigate the
propensity of HpQd to adopt either of its two possible conformers
(Hp or Qd) preferentially under different ionic conditions at thermodynamic
equilibrium. The imino proton NMR resonances of nucleic acid structures
generally reflect their base-pairing arrangement, and indeed, 1D NMR
of imino protons has previously been used to study hairpin to hairpin
transitions within small RNA molecules.[19] To start, we recorded the 1H NMR spectrum of HpQd in
10 mM phosphate-buffered saline (PBS) (pH 7.0) in the absence of any
added salt. The spectrum revealed two patterns of imino proton signals
encompassing two distinct spectral regions (10.5–11.8 and 12–13
ppm), suggesting the coexistence of two structural RNA conformers
in solution (Figure 3). On the basis of the 1H NMR spectra recorded for the Hp′ and Qd′ reference
sequences (Figure S6) these imino fingerprints
were attributed to the Qd (below 12 ppm) and Hp (above 12 ppm) structures.
Indeed, imino proton peaks from 12 to 14 ppm are characteristic of
Watson–Crick base pairs, whereas those from 10 to 12 ppm are
characteristic of Hoogsteen base pairs and indicate G-tetrad formation.[20] Next, we titrated the HpQd sequence with increasing
concentrations of KCl (0–100 mM). As shown in Figure 3a, this led to the disappearance of the imino peaks
characteristic of the hairpin motif along with remodelling and sharpening
of the peaks below 12 ppm, indicating a displacement of the equilibrium
in favor of a unique stable G-quadruplex motif. For KCl concentrations
above 25 mM, the characteristic signals of the hairpin structure were
no longer detectable.[21] On the other hand,
titration of HpQd with increasing amounts of MgCl2 (0–100
μM) led to the disappearance of the Hoogsteen imino proton resonances
and sharpening of the 1H signals between 12 and 13 ppm
(Figure 3b), indicating a displacement of the
structural equilibrium in favor of the hairpin. At 100 μM MgCl2 concentration, the signals from the quadruplex conformer
almost disappeared, and only the hairpin signals remained. These results
illustrated that the structural preference of HpQd is controlled by
the cations present in solution, with K+ promoting the
formation the G-quadruplex motif and Mg2+ promoting the
hairpin form.
Figure 3
1H NMR titrations of the HpQd sequence with
increasing
amounts of (a) KCl and (b) MgCl2. The initial spectra were
acquired in 10 mM PBS (pH 7.0) in the absence of added KCl or MgCl2. Signals labeled with red stars and blue circles correspond
to the Qd and Hp structures, respectively.
1H NMR titrations of the HpQd sequence with
increasing
amounts of (a) KCl and (b) MgCl2. The initial spectra were
acquired in 10 mM PBS (pH 7.0) in the absence of added KCl or MgCl2. Signals labeled with red stars and blue circles correspond
to the Qd and Hp structures, respectively.We next assessed the competitive formation of the
two RNA structures
in an environment that better mimics a cellular context where both
K+ and Mg2+ ions were present at high concentration.
The 1H NMR spectrum of HpQd recorded in 10 mM PBS (pH 7.0)
and 100 mM KCl indicated the preferential formation of the Qd G-quadruplex
structure (Figures 3a and 4a). Upon addition of increasing amounts of MgCl2 (0–3 mM), no significant modification of the imino proton
NMR spectrum of HpQd was observed (Figure 4a), indicating that the structural equilibrium was not displaced
and that, once formed in the presence of 100 mM KCl, the G-quadruplex
motif is the more stable conformer even at near-physiological millimolar
Mg2+ concentrations. When HpQd was pre-equilibrated in
10 mM PBS (pH 7.0) and 3 mM MgCl2, titration with increasing
concentrations of KCl (0–100 mM) revealed the disappearance
of the imino peak signals attributed to the hairpin motif and the
emergence of a imino proton envelope below 12 ppm corresponding the
Qd G-tetrad resonance signals (Figure 4b).
It is noteworthy that the NMR spectra obtained at the ends of the
two titrations were identical. In summary, these data demonstrated
that HpQd preferentially folds into a G-quadruplex conformation under
near-physiological conditions in presence of 100 mM KCl and 3 mM MgCl2, regardless of its initial state (Figure
S7).
Figure 4
1H NMR titrations of the HpQd sequence with increasing
amounts of (a) MgCl2 and (b) KCl. The initial spectra were
acquired in 10 mM PBS (pH 7.0) in the presence of (a) 100 mM KCl and
no MgCl2 and (b) 3 mM MgCl2 and no KCl. Signals
labeled with red stars and blue circles correspond to the Qd and Hp
structures, respectively.
1H NMR titrations of the HpQd sequence with increasing
amounts of (a) MgCl2 and (b) KCl. The initial spectra were
acquired in 10 mM PBS (pH 7.0) in the presence of (a) 100 mM KCl and
no MgCl2 and (b) 3 mM MgCl2 and no KCl. Signals
labeled with red stars and blue circles correspond to the Qd and Hp
structures, respectively.RNA folds can also be manipulated by external triggers
such as
synthetic small molecules that can act as regulators for chemical
biology studies.[22] We next investigated
the use of a synthetic molecule to control the equilibrium between
the two secondary structures. While the G-quadruplex conformation
is the favored conformation of HpQd under near-physiological ionic
conditions, we sought to determine whether a small-molecule nucleic
acid binder could influence this structural preference. We selected
a triarylpyridine (TAP) derivative, 1 (Figure 5), that we previously showed can disrupt DNA G-quadruplex
formation.[23] We first performed a titration
of the Qd′ reference sequence with 1. Increasing
concentrations of 1 led to a progressive decrease in
the imino proton resonance envelope between 10 and 12 ppm, which almost
completely disappeared after the addition of 3 equiv of 1 (Figure S8a), indicating small-molecule-mediated
disruption of the quadruplex motif. In a control experiment, titration
of the Hp′ reference revealed only minor alterations of the
imino signals, and importantly, the imino resonances corresponding
to the hairpin structure still persisted even after the addition of
3 equiv of 1, indicating that the hairpin fold was not
disrupted (Figure S8b). Next, we titrated
HpQd with 1 under ionic conditions favoring the formation
of the G-quadruplex motif. We observed that increasing the concentration
of 1 led to a decrease in the imino signal envelope below
12 ppm with a concomitant increase of the signal above 12 ppm (Figure 5). Structural interconversion was confirmed by integrating
the changes in the imino resonance envelope areas triggered by the
addition of 1 (Figure S9).
Collectively, these results demonstrated the ability of the TAP derivative
to displace the structural equilibrium in favor of the hairpin motif
by unfolding the G-quadruplex motif.
Figure 5
1H NMR titration of the HpQd
sequence with increasing
amounts of 1. The initial spectrum was recorded in 10
mM PBS (pH 7.0) containing 10 mM KCl and 200 μM MgCl2.
1H NMR titration of the HpQd
sequence with increasing
amounts of 1. The initial spectrum was recorded in 10
mM PBS (pH 7.0) containing 10 mM KCl and 200 μM MgCl2.In conclusion, we have shown that a reversible
hairpin G-quadruplex
structural equilibrium within an RNA molecule is controlled by cations
and can be manipulated by a synthetic small molecule. In view of the
fact that the regulatory functions of RNA molecules are often linked
to conformational transitions,[3] these observations
have important implications for the inevitable competition between
such structures and their functions in nature as well as their targeting.
Thus, a bioinformatic search for naturally occurring hairpin/G-quadruplex
switchable elements in the transcriptome now represents a challenging
but exciting perspective (Figure S10).
Authors: Zi-Fu Wang; Andrei Ursu; Jessica L Childs-Disney; Rea Guertler; Wang-Yong Yang; Viachaslau Bernat; Suzanne G Rzuczek; Rita Fuerst; Yong-Jie Zhang; Tania F Gendron; Ilyas Yildirim; Brendan G Dwyer; Joseph E Rice; Leonard Petrucelli; Matthew D Disney Journal: Cell Chem Biol Date: 2018-11-29 Impact factor: 8.116
Authors: Zhaoming Su; Yongjie Zhang; Tania F Gendron; Peter O Bauer; Jeannie Chew; Wang-Yong Yang; Erik Fostvedt; Karen Jansen-West; Veronique V Belzil; Pamela Desaro; Amelia Johnston; Karen Overstreet; Seok-Yoon Oh; Peter K Todd; James D Berry; Merit E Cudkowicz; Bradley F Boeve; Dennis Dickson; Mary Kay Floeter; Bryan J Traynor; Claudia Morelli; Antonia Ratti; Vincenzo Silani; Rosa Rademakers; Robert H Brown; Jeffrey D Rothstein; Kevin B Boylan; Leonard Petrucelli; Matthew D Disney Journal: Neuron Date: 2014-08-14 Impact factor: 17.173
Authors: Tania F Gendron; Kevin F Bieniek; Yong-Jie Zhang; Karen Jansen-West; Peter E A Ash; Thomas Caulfield; Lillian Daughrity; Judith H Dunmore; Monica Castanedes-Casey; Jeannie Chew; Danielle M Cosio; Marka van Blitterswijk; Wing C Lee; Rosa Rademakers; Kevin B Boylan; Dennis W Dickson; Leonard Petrucelli Journal: Acta Neuropathol Date: 2013-10-16 Impact factor: 17.088
Authors: Henry Albert Day; Elisé Patricia Wright; Colin John MacDonald; Andrew James Gates; Zoë Ann Ella Waller Journal: Chem Commun (Camb) Date: 2015-09-25 Impact factor: 6.222