ABSTRACT
The Hepatitis Delta Virus (HDV) ribozyme self-cleaving activity in 20 M formamide solutions is unique. Does this
catalytic activity result from the conservation of its tertiary structure in 20
M formamide? We followed the ribozyme structure in formamide solutions by
monitoring the amount of bound Ethidium Bromide (EB). We were able to measure
the quantity of dye bound using time-resolved fluorescence spectroscopy, as an estimate of the ribozyme double helical content. This method, calibrated by using
oligonucleotides with defined tertiary structure and denaturing solvents,
parallels NMR and UV measurements as a function of temperature. Measurements
with the HDV ribozyme lead to three conclusions: (a) both the precursor and
product RNAs are structured to 24 M (95% w/w) formamide or 4 M H
2
O solutions which is equivalent to 4 M H
2
O; (b) the HDV ribozyme is the only RNA sequence investigated in this study that
retains so much structure in formamide; and (c) DNA analogs of formamide
resistant HDV ribozyme sequences lose their structure at less than 15 M
formamide. Thus, the structural integrity of the HDV ribozyme is an intrinsic
property of the RNA molecule and its sequence.
The Hepatitis Delta Virus (HDV) ribozyme exhibits the extraordinary ability to
self-cleave in 20 M (80% w/w) formamide, a highly denaturing condition (
1
-
3
). Is this due to a significant fraction of the HDV ribozyme having structure in
20 M formamide or to a small fraction that is in equilibrium with a large
unstructured RNA population? We measure bound Ethidium Bromide (EB) in the
presence of formamide, taking advantage of the dramatic change in fluorescence
lifetime when EB is bound to nucleic acids. A two-state model yields the fraction bound EB from the fluorescence decay. We
used
1
H-NMR and UV-melting curves on identical samples to demonstrate that bound EB
reflects stacked double helical base paired structure.
The early interest in adding denaturants to nucleic acid aqueous solutions was
to bring a high-helix-coil transition or `melting' temperature (T
m
) down to a more experimentally manageable level, thus avoiding degradation of
the material (
14
). Low levels of denaturant are added to nucleic acid solutions in order to
enhance the stringency of a primer to its target substrate, as in PCR
experiments (
4
) and Southern Transfers (
5
). Denaturant is used when studying ribozyme reactivity because of the cleavage
enhancement observed at low levels of denaturants (
6
), with activity decreasing at higher denaturant concentrations. The enhanced
activity can be explained by assuming that ribozymes can take several
conformations, some of them being inactive (
3
). Denaturant addition lowers the energy barriers between these alternative
conformations allowing the ribozyme molecules to adopt their active form. Some
constructs of the HDV ribozyme have the ability to self-cleave in 20 M formamide (
3
), while some constructs that exhibit no activity in aqueous buffers, start self-cleaving
only
when formamide is present (
3
,
6
). With these considerations, a structural study of the HDV ribozyme would be
most relevant in aqueous buffers containing medium to high formamide concentrations. The experiments reported here constitute a step in this direction.
Our observations with the HDV ribozyme, lead us to conclude that both the
product and precursor of the HDV ribozyme preserve substantial structure in 24
M formamide. The HDV ribozyme is the only nucleic acid among the ones
investigated in this study that retains so much structure in formamide. DNA
analogs with virtually identical sequence showed no structure at formamide
concentrations where RNA sequences were active. Therefore, the structure of the
HDV ribozyme is characteristic of this RNA sequence.
Figures
1
and
2
list the sequences of all oligonucleotides investigated in this study. The
oligonucleotides DHP9 (a DNA hairpin with nine bases), RHP12 (a RNA hairpin
with 12 bases), DHP12, RHP28, DHP28, RPK26, DRI89 (DNA sequence analog of the
ribozyme cleavage product) and DRI94 (DNA sequence analog of the ribozyme) were
synthesized on an Expedite Nucleic Acid Synthesizer using standard
phosphoramidite chemistry. After cleavage from the solid support and
deprotection, RNA and DNA oligonucleotides were passed through a Hamilton PRP-1 column using reverse phase chromatography. All RNA oligonucleotides were
also purified by gel electrophoresis. The bands were cut and eluted overnight
against water or TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 7.4). The yields after purification were measured at
260 nm. Ribosomal 5S RNA was purchased from SIGMA and used with no further
purification. Before fluorescence experiments, ribosomal 5S RNA was dialyzed overnight against water in order to remove salt present in the buffer and then brought up
to the appropriate formamide/buffer mixture. All samples were stored at -20oC.
RNA and DNA samples were snap cooled by heating them at 75oC for 3 min and 95oC for 5 min respectively and then plunging them into an ice bottle for
10 min before a fluorescence measurement. Samples were prepared in either L-buffer (10 mM sodium phosphate, 0.01 mM EDTA, pH 6.5), D-buffer (100 mM sodium citrate, 50 mM Tris-HCl, pH 6.5) or TE-buffer. The RNA pseudoknot solutions contained 5 mM
MgCl
2
. Extinction coefficients were estimated from reported values listed by Puglisi
and Tinoco for 1 M NaCl solutions (
7
). For all samples, the exact amounts of nucleic acid, buffer and EB were
collected in a microfuge tube and lyophilized. The dry pellet was then
resuspended in buffer or formamide solution of the desired final concentration.
This procedure ensured that the same amount of salt was present in all samples.
The weight fraction in formamide, w
f
, was converted to molar concentration using equation
1
for formamide w
f
> 0.05.
[Formamide] = -0.07 + 23(w
f
) + 2.14(w
f
)
2
1
Fluorescence measurements were carried out using a 50 [mu]l cuvette with concentrations of EB and DNA or RNA bases equal to 5 [mu]M and 1.4 mM, respectively. These conditions ensured an excess of 280 bases per EB dye. The NMR experiments used 20 mM of bases (0.7 mM of DHP28) in 300 [mu]l. Each solution contained 20% D
2
O which was considered as water in the formamide/water mixture.
Absorption
as a function of temperature was measured at 280 nm on a Beckman 640 UV-Vis spectrometer with a Peltier temperature controller accessory between 18 and 80oC using a heating rate of 0.5oC/3 min. DNA solutions were covered with a layer of silicon oil to limit evaporation. The difference in sample
absorption measured before and after a melting experiment never exceeded 2%.
The absorption versus temperature curves were fit to a two-state model with sloping baselines from which thermodynamic parameters [Delta]H and [Delta]S were derived (
8
). Parameters were optimized using the Marquardt Levenberg algorithm and the
goodness of the fit was judged from the residuals, auto correlation function of
the residuals and the [chi]
2
parameter (
9
). The melting temperature of hairpins was calculated from equation
2
.
T
m
(oK) = [Delta]H/[Delta]S 2
Fluorescence decay curves were measured on a home built time-correlated single-photon-counting apparatus. Pulsed excitation was obtained from a cavity dumped dye laser, synchronously
pumped by a Nd:YAG laser (Coherent Antares 76-s), with a repetition rate of 4 MHz. The samples were contained in a 50 [mu]l quartz cuvette and experiments were carried out at room temperature.
DNA or RNA bound EB was excited with vertically polarized laser pulses at [lambda]
ex
= 570 nm, and the emission was detected at 90o to the excitation through band pass filters at [lambda]
em
= 610 +- 10 nm. The fluorescence decays of EB bound to DNA or RNA were obtained
by setting the emission polarizer at the magic angle (54.7o with the polarization orientation of the excitation beam). Twenty thousand
counts were stored at the fluorescence decay maximum. All fluorescence decays
were collected over 256 channels using a time per channel of 290 ps/channel.
All experiments were carried out at 25oC.
The fluorescence decay of free EB was measured over the whole range of formamide
concentration and fitted with a single exponential. The lifetimes computed as a
function of formamide content were then fitted with a second order binomial
given in equation
3
:
[tau]
F
= 1.66 + 3.85(w
f
) - 1.15(w
f
)
2
3
which yields the lifetime of free EB at any formamide concentration.
Fluorescence decays of oligonucleotide solutions with EB taken at the magic angle (
33
) for different water/formamide mixtures were clearly non-exponential (Fig.
3
). They were fit with a three-exponential function given in equation
4
that was convoluted with the instrument response function.
All experiments were performed at 25oC with 20% D
2
O using either Bruker DMX 500/600/750 spectrometers with Oxford or JMT magnets.
We first show that we can measure bound EB as a function of nucleic acid
structure. Then we apply our method to ribozyme sequences.
The sequences and the most likely secondary structures of the oligonucleotides
used in this study are listed in Figures
1
and
2
. A typical example resulting from our time-resolved fluorescence experiments is shown in Figure
3
. The fraction of EB bound to a nucleic acid is obtained by analyzing the EB
fluorescence decays with three exponentials, with the short lifetime set equal
to that of free EB in the water/formamide mixture and the two longer ones
attributed to EB bound to the nucleic acid. The ratio of the pre-exponential factors yields the fraction of bound EB from equation
4
. The fraction of EB bound (
f
b
) to a given oligonucleotide in L- and TE-buffers decreases with increasing formamide concentrations (Fig.
4
). The trend shows EB binds more tightly at high formamide concentration for
longer hairpin stems following the sequence DHP28>DHP12>DHP9. To reinforce the conclusion that longer double stranded regions results in tighter EB
binding, we synthesized the double hairpin DTHP28 which is two DHP12 hairpins
tethered by four Ts with same base length as DHP28. Its
f
b
versus [Formamide] profile overlays DHP12s in Figure
4
, demonstrating that EB binding to DNA in formamide is a function of the length
of the double helix and not of the length of the molecule. This result also
suggests that the two hairpins in DTHP28 do not interact, a result confirmed by
fluorescence anisotropy as well as by a UV-melting profile (cf. Table
1
) (
10
).
Relatively few studies have been carried out to understand the nature of the
interaction between a polar denaturing agent (such as formamide) and an
oligonucleotide (
11
-
19
). Using the laser-jump technique, the rates of stacking and unstacking were measured for
poly(adenylic acid) (
14
) and poly(cytidylic acid) (
20
) in various aqueous denaturant mixtures. Base stacking was found to be affected
by the viscosity and polarity of the denaturant, with the higher solvent viscosity resulting in a decrease of the base stacking rate. The unstacking rate is increased by polar denaturants which attack the stacked state and solvate the bases. This effect seems to
overrule the expected stabilization of the stacked state through shielding of
the charged phosphate backbone. The thermodynamic parameters of double helix
formation in aqueous solutions containing different amounts of denaturants were
also measured for the oligonucleotides (dGdC)
3
(
16
) and A
7
U
7
p (
17
). These results combined with the observation that single stranded poly(C) has
a higher melting temperature than single stranded poly(A) (
20
) although the surface area per base is less, all contradict the trend expected
for thermodynamics dominated by cavity terms as an approach based only on the solvophobic force would predict (
21
). The thermodynamic results obtained with (dGdC)
3
and A
7
U
7
p also show that the effects of polar denaturants such as urea, formamide and
dimethyl-formamide depends more on dipole moment than hydrogen bonding ability.
Therefore, formamide being more viscous (viscosity = 4.3 c.p.) and polar (dipole moment = 3.37 D) than water (viscosity = 0.89 c.p. and dipole moment = 1.87 D) (
16
) is a better denaturant, as observed experimentally.
Our time-resolved fluorescence measurement is another option for the investigation
of nucleic acid structure in aqueous buffer/formamide mixtures. These
measurements present several advantages when compared to other techniques.
Since the spectroscopic properties of EB (absorption maximum at 500 nm,
emission maximum at 600 nm) are very different from those of the nucleic acids,
its fluorescence can be used to monitor nucleic acids in the presence of
formamide or a protein which also has UV-absorption. Another advantage is that only 70 nmol of bases in 50 [mu]l are necessary for fluorescence measurements versus 6 [mu]mol of bases in 300 [mu]l for NMR, a 100-fold difference. The measurements observe structure at
the bound ethidium molecule, and a concern could be that the dye perturbs the
structure. Since the imino proton NMR spectra, shown here, and the cleavage activity of the ribozyme both parallel the fluorescence measurements as a function of formamide, we conclude that the latter method is a reasonable
probe of uncomplexed nucleic acid structure.
Our experiments have shown that formamide has varied denaturing effects on nucleic acids, and that among the DNA and RNA oligonucleotides
examined here, the HDV ribozyme is the only one to preserve structure in almost
pure formamide (24 M). For this molecule and other related constructs, high
formamide concentrations can not be regarded as a solvent that gives a
completely unfolded nucleic acid chain in a random coil conformation. This
conclusion, which has already been reported for proteins, draws another
parallel between the behavior of nucleic acids and proteins in the presence of
denaturants (
30
,
31
).
These experiments were done in part on equipment of the Regional Laser and
Biotechnology Laboratories supported by National Institute of Health Research
Resource funding to R.M. Hochstrasser. P. L. acknowledges research grants from
National Aeronautics and Space Administration and United States Army Research
Office and G.D.-G. acknowledges a research grant from National Institute of Health. J.D.
thanks Kathleen Valentine, Ruth Steele and Fabio Almeida for help with the NMR
experiments.
Present addresses:
+
Department of Chemistry, University of Waterloo, Waterloo, Ontario, Canada and
[sect]
Department of Biology, Buffalo State College, Buffalo, NY, USA
REFERENCES
Return