
==== Front
Nucleic Acids ResNucleic Acids ResnarNucleic Acids ResearchNucleic Acids Research0305-10481362-4962Oxford University Press 1716998810.1093/nar/gkl1073Structural BiologyInfluence of cationic molecules on the hairpin to duplex equilibria of self-complementary DNA and RNA oligonucleotides Nakano Shu-ichi 1Kirihata Toshimasa 2Fujii Satoshi 12Sakai Hiroshi 2Kuwahara Masayasu 34Sawai Hiroaki 3Sugimoto Naoki 12*1Frontier Institute for Biomolecular Engineering Research (FIBER)8–9–1 Okamoto, Higashinada–ku, Kobe 658–8501, Japan2Department of Chemistry, Faculty of Science and Engineering, Konan University8–9–1 Okamoto, Higashinada–ku, Kobe 658–8501, Japan3Faculty of Engineering, Gunma University1-5-1 Tenjin-chou, Kiryu, Gunma 376-8515, Japan4PRESTO, Japan Science and Technology Agency (JST)Saitama 332-0012, Japan*To whom correspondence should be addressed. Tel: +81 78 435 2497; Fax: +81 78 435 2539; Email: sugimoto@konan-u.ac.jp1 2007 14 12 2006 14 12 2006 35 2 486 494 27 7 2006 09 11 2006 10 11 2006 © 2006 The Author(s).2006This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/2.0/uk/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.A self-complementary nucleotide sequence can form both a unimolecular hairpin and a bimolecular duplex. In this study, the secondary structures of the self-complementary DNA and RNA oligonucleotides with different sequences and lengths were investigated under various solution conditions by gel electrophoresis, circular dichroism (CD) and electron paramagnetic resonance (EPR) spectroscopy and a ultraviolet (UV) melting analysis. The DNA sequences tended to adopt a hairpin conformation at low cation concentrations, but a bimolecular duplex was preferentially formed at an elevated cationic strength. On the other hand, fully matched RNA sequences adopted a bimolecular duplex regardless of the cation concentration. The thermal melting experiments indicated a greater change in the melting temperature of the bimolecular duplexes (by ∼20°C) than that of the hairpin (by ∼10°C) by increasing the NaCl concentration from 10 mM to 1 M. Hairpin formations were also observed for the palindrome DNA sequences derived from Escherichia coli, but association of the complementary palindrome sequences was observed when spermine, one of the major cationic molecules in a cell, existed at the physiological concentration. The results indicate the role of cations for shifting the structural equilibrium toward a nucleotide assembly and implicate nucleotide structures in cells.
==== Body
INTRODUCTION
A self-complementary nucleotide sequence potentially adopts a unimolecular hairpin as well as a bimolecular duplex. This is problematic for oligonucleotide designs, such as nucleotide hybridization with a complementary sequence, molecular beacons, dsRNAs for RNAi technology and ribozymes that work either in a bimolecular or a unimolecular conformation (1,2). In biological systems, a palindrome sequence or inverted repeats may adopt a hairpin structure by dissociating a number of base pairs in a genome DNA and form a cruciform structure that is relevant to the frameshift mutations. The hairpin structure may also affect the supercoiling degree of the plasmid DNA and be the recognition sites for the transcription and replication proteins (3–7). In addition, most RNAs in biological systems fold into a unimolecular structure accompanied by a formation of hairpin loops, and the loop nucleotides may further interact with other RNAs to form an intermolecular assembly and controlled periodic structural patterns (8).

The relative stability between the unimolecular hairpin and the bimolecular duplex determines the secondary structure of the self-complementary nucleotides. The hairpin (H)-duplex (D) equilibrium mediated by a single strand (S) can be represented by Equations 1 and 2.
 1 S ⇄ H 
 2 2S ⇄ D. 
Although the equilibria between a unimolecular hairpin and a single-strand state and between a bimolecular duplex and two single strands have been extensively investigated (9–20), the equilibrium between the hairpin and the bimolecular duplex has not been well studied except for several reports: the precedence studies used DNA oligonucleotides that showed a structural transition from a hairpin to a bimolecular duplex by adding NaCl, and most of them had a continuous AT sequence around the dyad of the palindrome sequence (21–26).

In this study, the secondary structures of the DNA and RNA oligonucleotides of fully and partly self-complementary sequences were systematically investigated in the absence and presence of the cationic molecules of NaCl, MgCl2, spermidine, spermine and a basic oligopeptide containing lysine residues. We prepared the oligonucleotides containing a continuous AT or GC sequence around the dyad of the palindrome sequence. As a result, the fully self-complementary as well as the partly self-complementary DNA sequences adopted a unimolecular hairpin, while only the partly self-complementary RNA sequences formed the hairpin structure. Although the nucleotides tended to adopt a unimolecular hairpin at low cation concentrations, a bimolecular duplex was formed at an elevated cationic strength. In addition, it was indicated that the DNAs with longer lengths formed a greater amount of the hairpin structure, which was unexpected based on the nearest-neighbor parameters (14–17). Our observations implicate the role of cationic molecules for shifting the structural equilibrium toward the intermolecular assembly.

MATERIALS AND METHODS
Preparation of oligonucleotides and buffer solutions
DNA oligonucleotides with a high-performance liquid chromatography (HPLC) purification grade were purchased from Hokkaido System Science (Japan). The RNA oligonucleotides were synthesized on a solid support using phosphoramidite chemistry by automated DNA synthesis (Model 3400, Applied Biosystems, USA) and purified by reversed-phase HPLC (Model PX-8020, TOSOH, Japan) after removal of the protection groups. The spin-labeled deoxyuridine analog phosphoramidite, 5-(2,2,6,6-tetramethyl-4-ethynylpiperidyl-3-ene-1-oxyl)-5′-(4,4′-dimethoxyltriphenyl)-uridine phosphoramidite, was synthesized, according to previously reported methods (27,28). The spin-labeled deoxyuridine analogue was incorporated into a DNA strand by the automated chemical synthesis, and the oligonucleotide purified with the HPLC was confirmed by the MALDI-TOF MS (matrix assisted laser desorption/ionization time of flight mass spectra) measured by a Voyager DE MALDI-TOF mass spectrometer (Applied Biosystems).

The buffer solution contained 25 mM HEPES [4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid] (pH 7.0), 0.1 mM Na2EDTA, and an appropriate amount of NaCl. For the experiments using MgCl2, spermidine, spermine, and Lys3Tyr, 10 mM NaCl and 0.1 mM Na2EDTA were also included in the buffer. The oligopeptide of Lys3Tyr was synthesized on a solid support by automated peptide synthesis using Fmoc chemistry (Model 433A, Applied Biosystems) and purified with reversed-phase HPLC after removal of the protection groups. The product was confirmed by MALDI-TOF MS spectra, and the concentration of the oligopeptide was adjusted by the absorbance of the tyrosine residue.

Polyacrylamide gel electrophoresis (PAGE) of DNA and RNA
After annealing of 20 μM nucleotides in the buffer solution at 90°C for 5 min, the sample was cooled to 4°C at the rate of −1°C/min, followed by loading on the native 20% polyacrylamide gel (acrylamide:bisacrylamide = 19:1) in a cold room at 4°C, unless otherwise mentioned. The running buffer used for the electrophoresis was the 1× TBE buffer containing 0.09 M Tris, 0.09 M boric acid and 2 mM Na2EDTA at pH 8.3. It was confirmed that the annealing temperature and the cooling rate did not affect the results. After the electrophoresis, nucleotides in the gel were stained by GelStar (TaKaRa) or CyberGreen II (TaKaRa), and the fluorescence emission of the stains was visualized by FUJIFILM FLA-5100. The fraction of a bimolecular duplex was evaluated based on the assumption that the efficiency of the staining of the base pairs in a hairpin was similar to that in a bimolecular duplex. This assumption evidently provided similar results to those determined from the circular dichroism (CD) profile and those obtained by using the fluorescein-labeled nucleotides (see below). The 6-FAM (6-carboxyfluorescein) labeled oligonucleotides were used for the gel electrophoresis under the same conditions except that the gel electrophoresis image was detected by the fluorescein emission.

Measurements of CD and electron paramagnetic resonance (ERR) spectra
The CD spectra were obtained with a JASCO J-820 spectropolarimeter equipped using a temperature controller. The measurements were carried out with 50 μM oligonucleotides in a buffer containing the appropriate concentration of NaCl, 25 mM HEPES (pH 7.0) and 0.1 mM Na2EDTA. All spectra were measured at 4°C and the nucleotides were heated to 90°C before use.

The electron paramagnetic resonance (EPR) measurements were carried out using a JEOL JES-TE100 spectrometer. The spectra of 100 μM DNA containing the deoxyuridine analog tethering 4-ethynyl-2,2,6,6-tetramethyl-3,4-dehydro-piperidine-1-oxyl as a spin labeled moiety in an appropriate buffer solution were measured at room temperature. Instrument settings were adjusted for a microwave power of 8.0 mW, modulation width of 0.2 mT at 100 kHz, time constant of 0.1 s and sweep time of 2 min. The effective rotational correlation time (τR) of the spin probe was calculated as reported previously (29,30).

Measurement of the ultraviolet (UV) melting curves
The UV absorbance of the nucleotides was measured with a Shimadzu 1650 or 1700 spectrophotometer equipped with a temperature controller. The melting curves were monitored at 260 nm and at a rate of 0.5 or 1°C/min, and the melting temperature (Tm) at which half of the folded structure was denatured and determined (31). The extinction coefficient of the oligonucleotides was calculated based on the nearest-neighbor approximation (32), and those containing the deoxyuridine analog were assumed to be the same as those containing deoxythymidine.

RESULTS AND DISCUSSION
Secondary structures of the self-complementary DNAs and RNAs at low and high NaCl concentrations
A self-complementary nucleotide can adopt a unimolecular hairpin and a bimolecular duplex depending on the sequence and the experimental conditions. Although, the computer programs for the nucleic acid folding and hybridization can predict the nucleotide secondary structures from its sequence using the nearest-neighbor parameters determined in solutions containing 1 M NaCl (14–16,33–36) and the amounts of individual species of the hairpin and the duplex in equilibrium (37), it is difficult to predict which secondary structure of a self-complementary sequence is adopted under the different salt conditions. To experimentally investigate the secondary structure and the equilibrium between the hairpin and the bimolecular duplex under various solution conditions, fully and partly self-complementary DNA and RNA sequences indicated in Figure 1A were designed. The DNA nucleotides of d16cg, d14cg, d12cg and d10cg contain a continuous GC sequence and d14ta, d12ta and d10ta contain a continuous AT sequence in the middle of the sequence, all of which are fully self-complementary sequences forming a fully matched duplex when adopting a bimolecular duplex. One or two base substitutions in the d14cg sequence lead to the partly self-complementary sequences of d14cg-caag, d14cg-ccag and d14cg-ccga, and these nucleotides contain two mismatched sites when adopting a bimolecular duplex. A similar sequence set was also prepared for RNAs as shown in Figure 1A.

Figure 1 (A) Oligonucleotide sequences and their abbreviations. Underline indicates the base substitutions from d14cg or r14cg. (B–E) Gel electrophoresis at 4°C for 20 μM DNA (B and C) and RNA nucleotides (D and E), prepared in 10 mM (B and D) or 1 M NaCl (C and E). The nucleotides on the marker lanes are the non-self complementary duplex of 5′-ACTGAGGACCTA-3′/5′-TAGGTCCTCAGT-3′ (MD), the stable tetraloop structure of 5′-GGAGCTTGCTCC-3′ (MH) (12), and a single strand of 5′-ACTGAGGACCTA-3′ (MS). D and H on the right side of the gel images indicate the bimolecular duplex and the hairpin, respectively.

The native PAGE was examined to detect the secondary structure of the nucleotides. Since the interconversion of a hairpin and a bimolecular duplex and their dissociations into a single-strand are slow at low temperature, identification of the secondary structures was carried out by gel electrophoresis performed at 4°C. Figure 1B show the secondary structures of the DNAs prepared in the buffer containing 10 mM NaCl. The control 12mer nucleotides on the marker lanes indicated that the migration of the hairpin structure was faster than that of the bimolecular duplex and that the staining of the single-stranded nucleotide was faint due to the stain dye preferentially associating with the base paired sites. It was demonstrated that d16cg, d14cg, d12cg and d10cg adopted the bimolecular duplex, although these nucleotides except for d10cg also formed the unimolecular hairpin. On the other hand, d14ta and d12ta predominantly formed the hairpin rather than a fully matched duplex. Since the hairpin structure stably exists when the loop comprises 3–5 nt (9–13,18), substitutions of the central GC nucleotides by AT nucleotides decrease the stability of the bimolecular duplex greater than that of the hairpin form. The nucleotide of d10ta showed less stain, suggesting that the base pairings were dissociated into a single-strand during the electrophoresis due to its short nucleotide length. Existence of the single-stranded d10ta and the absence of single-strands of other sequences were confirmed by the experiments using the fluorescein-labeled oligonucleotides as described later. Hairpin formations were also observed for the partly self-complementary nucleotides of d14cg-caag, d14cg-ccag and d14cg-ccga, consistent with the idea that the nucleotide substitutions in the d14cg sequence destabilize their bimolecular duplexes.

On the other hand, all the fully self-complementary DNA sequences predominantly adopted the bimolecular duplex at 1 M NaCl (Figure 1C), and d14cg-caag and d14cg-ccga as well as d16cg, d14cg and d12cg increased the fraction of the bimolecular duplex compared with those at 10 mM NaCl. Intriguingly, the secondary structure of d14ta and d12ta was changed from the hairpin to the bimolecular duplex by increasing the NaCl concentration from 10 mM to 1 M.

In contrast to the DNAs, the fully self-complementary RNA sequences except for r14ua and r10ua predominantly adopted the bimolecular duplex both at 10 mM (Figure 1D) and 1 M NaCl (Figure 1E), and the bimolecular duplex fraction of r14ua increased with the increasing NaCl concentration from 10 mM to 1 M, although r10ua showed less stain. Furthermore, r14cg-caag, one of the partly self-complementary RNAs, clearly changed the structure from the hairpin at 10 mM NaCl to the bimolecular duplex at 1 M NaCl, as observed for the DNA nucleotides of d14ta and d12ta. Therefore, it appears to be a general feature that NaCl can shift the hairpin-duplex equilibrium of the self-complementary nucleotides toward the bimolecular duplex, and that the RNAs tend to adopt the bimolecular duplex even at the low NaCl concentration, probably due to the greater base pair stabilities in RNA than in DNA of the same sequence (14,19).

Cation concentrations for forming the bimolecular duplex
The secondary structures of d14ta, d12ta and r14cg-caag, which adopted the hairpin at 10 mM NaCl and the bimolecular duplex at 1 M NaCl, were further investigated at different NaCl concentrations (Figure 2). The fraction of the bimolecular duplex increased as the NaCl concentration increased, and the NaCl concentration at which half of the nucleotides formed the bimolecular duplex was estimated to be 66 ± 6, 123 ± 7 and 79 ± 8 mM for d14ta, d12ta and r14cg-caag, respectively (Figure 2D).

Figure 2 Dependence of the fractions of the bimolecular duplex and the hairpin on the NaCl concentration. The gel electrophoresis of (A) d14ta and (B) d12ta and (C) r14cg-caag at 20 μM with an increase in the NaCl concentration; 10, 20, 40, 50, 70, 100, 150, 200, 400 and 1000 mM from left to right. (D) The profiles of the fraction of the bimolecular duplex versus NaCl concentration obtained by gel electrophoresis for d14ta (gray squares), d12ta (white triangles), r14cg-caag (black circles), based on the CD spectra change at 265 nm for r14cg-caag (black triangles), and by gel electrophoresis with the fluorescein-labeled d14ta (gray diamonds).

The equilibrium shift toward the bimolecular duplex was also observed when MgCl2, spermidine, and the basic oligopeptide of Lys3Tyr were used (Supplementary Figure S1). Notably, only 10 mM or less magnesium ion, and 100 μM or less spermidine, spermine and Lys3Tyr were adequate for affecting the secondary structure of the DNAs, suggesting that the equilibrium shift was primarily caused by electrostatic interactions between the cationic molecules and the nucleotides. Although the cationic molecules stabilize both the bimolecular duplex and the hairpin, our results suggest that the stabilization to the bimolecular duplex structure is more effective. Our observations are consistent with the polyelectrolyte behavior of nucleotides (38), that is, the number of base pairs of a hairpin is less than that of a bimolecular duplex, which is relevant to the number of cations associated with the nucleotides. Because there is relatively less interphosphate repulsion at the hairpin loop nucleotides than at the base paired nucleotides, the greater stabilization for the bimolecular duplex than for the hairpin with the increasing NaCl concentration is expected, and this is confirmed by the Tm measurements as described later.

A high concentration of cosolutes, such as [poly(ethyleneglycol)] (PEG), which are often used to mimic intracellular molecular crowding, can also modulate the stability of the nucleotide base pairs even though they do not directly interact with nucleotides (39–41). The structural transitions of d14ta, d12ta and r14cg-caag with the changing NaCl concentration were observed even in the presence of 40 wt% PEG300 or 20 wt% PEG8000 (an average molecular weight of 300 or 8000, respectively), and the NaCl concentration at which half of the nucleotides formed the bimolecular duplex changed by no >2-fold difference with the addition of the PEGs (Supplementary Table S1). This observation indicates a less significant effect of the PEGs on the hairpin-duplex equilibrium, and implies no substantial change in the sodium ion activity upon the addition of PEG. Although PEG decreases the water activity (40) that may change the nucleotide hydration, a reduced significance of the hydration status on the equilibrium shift is suggested by the data, consistent with the observation of an equilibrium shift by a low concentration of the multivalent cations (Supplementary Figure S1).

Measurement of the structural changes by CD and EPR spectra
The structural transitions of d14ta, d12ta and r14cg-caag were further investigated from the viewpoint of their structural aspects. The CD spectra of the nucleotides were measured under the same conditions as the gel electrophoresis. The CD spectrum of r14cg-caag changed with the NaCl concentration (Figure 3A), while those of r14cg-ccag and r14cg-ccga that were found not to undergo a change in their secondary structure by gel electrophoresis, were only slightly affected by changing the NaCl concentration (Supplementary Figure S2). The CD intensity of the RNAs, either adopting a hairpin or a bimolecular duplex, gradually increased as the NaCl concentration was raised, suggesting that the sodium ion may slightly affect the base pair stacking interaction. Remarkably, the NaCl concentration profile of r14cg-caag evaluated by the CD spectra was almost identical to that obtained by the gel electrophoresis (Figure 2D also includes the NaCl concentration profile of the CD intensity at 265 nm for r14cg-caag), indicating that the gel electrophoresis reflects the nucleotide structures in solution.

Figure 3 (A) CD spectra of r14cg-caag at various NaCl concentrations measured at 4°C. The arrows indicate the change in the NaCl concentration from 10 mM to 1 M. (B) Structure of the nitroxide spin-labeled deoxyuridine and the sequences of d14ta and d12ta containing the spin label deoxynucleoside (denoted by S). (C) The EPR spectra of the spin labeled d14ta (upper), d12ta (middle) and a single-strand of 5′-TGAGGSGCTGTTTGTG-3′ as the control (lower) measured in 10 mM NaCl (black) and 1 M NaCl (gray).

Since the CD spectra of the hairpin and the bimolecular duplex of d14ta and d12ta differed less by changing the NaCl concentration (data not shown), the structural transition of the DNAs was detected by the EPR measurements using the nitroxide spin-labeled deoxyuridine unit. The EPR spectroscopy monitors the dynamics of the spin-labeled nucleotide and has been used to examine, e.g. the metal ion-specific structural transition of nucleotides (42) and ribozyme dynamics during the metal ion induced folding (30). It is reported that an oligonucleotide containing the nitroxide moiety does not perturb a DNA duplex structure (43), and that the EPR measurements provide distinguishable spectra among the spin-labeled nucleotides in an unpaired loop, in a duplex stem and in a single strand, that is, the labeled nucleotide in a hairpin loop revealed a more dynamic constraint thus providing a higher τR value than that in a single strand, but smaller than that in a duplex stem (44). The EPR spectra of d14ta and d12ta incorporating the spin-labeled nucleoside in the middle of the sequence (Figure 3B) are shown in Figure 3C. While the NaCl concentration did not change the spectrum of the control single-stranded sequence (τR of 0.14 ns both at 10 mM and 1 M NaCl), an increased spectral width was observed for d14ta and d12ta at 1 M NaCl, reflecting the reduced dynamics of the spin-labeled nucleotides relative to that at 10 mM NaCl (τRs of 0.96 and 1.8 ns for d14ta at 10 mM and 1 M NaCl, respectively, and τRs of 0.95 and 1.3 ns for d12ta at 10 and 1 M NaCl, respectively). The increased spectral width was also obtained when 1 mM spermine was used, giving the τRs of 1.8 ns for d14ta and 1.4 ns for d12ta. The observations are consistent with the nucleotides of d14ta and d12ta adopting the bimolecular duplex at the elevated ionic strength.

Thermal stability of the nucleotide structures
The changes in the secondary structures of d14ta, d12ta and r14cg-caag were further demonstrated from the UV melting experiments in which their Tm values at 10 mM NaCl were nucleotide strand concentration-independent and those at 1 M NaCl were nucleotide strand concentration-dependent (Table 1 and Supplementary Figure S3). It is noted that the difference between the Tm values at 1 M and 10 mM NaCl (ΔTm) of the nucleotides forming the bimolecular duplex (ca. 20°C) are obviously greater than those forming the hairpin (ca. 10°C) (Table 1), e.g. the ΔTm of r14cg forming the bimolecular duplex is 22.1°C and that of r14cg-ccga forming the hairpin is 11.0°C (see the details in the Supplemental Data). The greater stabilization for the bimolecular duplex could result in the equilibrium shift toward the bimolecular duplex at 1 M NaCl. In contrast to the Tm changes with the NaCl concentration, the Tm changes due to the addition of PEGs for the hairpin (at 10 mM NaCl) and the bimolecular duplex (at 1 M NaCl) differed only by <3.2°C (Table 2), indicating that the PEGs affected the stability of the hairpin and the bimolecular duplex to a similar extent, consistent with the NaCl-concentration profile data in the presence of the PEGs shown in Supplementary Table S1.

Table 1 Tm (°C) of the DNA and RNA oligonucleotidesa

Sequence	10 mM NaCl	1 M NaCl	ΔTm	
DNA	
    d16cg	67.7 (63.3)	83.3 (72.1)	15.6 (8.8)	
    d14cg	63.2 (55.0)	80.0 (69.1)	16.8 (14.1)	
    d12cg	52.1 (43.1)	72.5 (61.0)	20.4 (17.9)	
    d10cg	48.2 (39.0)	70.4 (61.1)	22.2 (22.1)	
    d14ta	50.7 (50.0)	64.9 (57.6)	NA b	
    d12ta	37.0 (38.0)	55.9 (47.0)	NA	
    d10ta	20.2 (<10)	46.0 (33.0)	25.8 (>23)	
    d14cg-caag	56.5 (56.8)	67.3 (66.1)	10.8 (9.3)	
    d14cg-ccag	55.2 (55.0)	64.7 (64.5)	9.5 (9.5)	
    d14cg-ccga	54.0 (54.5)c	66.2 (66.0)	12.2 (11.5)	
RNA	
    r14cg	66.5 (59.3)	88.6 (81.7)	22.1 (22.4)	
    r12cg	60.5 (50.4)	85.0 (74.0)	24.5 (23.6)	
    r10cg	63.0 (50.0)	85.6 (67.8)	22.6 (17.8)	
    r14ua	55.0 (44.2)	72.3 (61.5)	17.3 (17.3)	
    r12ua	40.6 (28.2)	61.2 (53.0)	20.6 (24.8)	
    r10ua	32.2 (20.7)	55.3 (46.2)	23.1 (25.5)	
    r14cg-caag	55.0 (55.2)	71.8 (59.0)	NA	
    r14cg-ccag	54.7 (47.2)	74.0 (58.0)	19.3 (10.8)	
    r14cg-ccga	56.2 (55.0)	67.2 (66.3)	11.0 (11.3)	
    r14cg-ccaa	NDd	51.2 (51.9)	ND	
aThe Tm values were determined in the buffer containing 10 mM or 1 M NaCl, 25 mM HEPES (pH 7.0) and 0.1 mM Na2EDTA for a total nucleotide strand concentration close to 100 μM. The values in parentheses are the data obtained for the total nucleotide strand concentration of ∼2 μM. ΔTm indicates the subtraction of the Tm obtained at 10 mM NaCl from that at 1 M NaCl.

bNA = not applicable due to the different conformation at 1 M and 10 mM NaCl.

cAlthough the transition was not two-state, the Tm values could be obtained.

dND = not determined due to the non-two-state transition.

Table 2 Difference in the Tm values (°C) obtained in the absence and presence of PEGa

Sequence	PEG	10 mM NaCl	1 M NaCl	ΔΔTm	
d14ta	PEG300	−12.4 (−11.5)	−14.7 (−14.4)	−2.3 (−2.9)	
	PEG8000	0.2 (1.7)	−0.2 (−1.0)	−0.4 (−2.7)	
d12ta	PEG300	−9.2 (−9.6)	−12.4 (−10.9)	−3.2 (−1.3)	
	PEG8000	1.5 (2.0)	0.1 (0.7)	−1.4 (−1.3)	
r14cg-caag	PEG300	−4.0 (−4.2)	−5.3 (−1.0)	−1.3 (3.2)	
	PEG8000	−1.0 (−0.1)	−1.4 (−0.4)	−0.4 (−0.3)	
aThe Tm values were obtained in the buffer containing NaCl, 25 mM HEPES (pH 7.0), 0.1 mM Na2EDTA and 40 wt% PEG300 or 20 wt% PEG8000 for a total nucleotide strand concentration close to 100 μM. The values in parentheses are the data obtained for the total nucleotide strand concentration of ∼2 μM. ΔΔTm indicates the subtraction of the value obtained at 10 mM NaCl from that at 1 M NaCl.

Influence of the nucleotide length on the secondary structures
The order of the hairpin fraction at 10 mM NaCl and 4°C shown in Figure 1B was d16cg > d14cg > d12cg > d10cg, suggesting that the longer nucleotides formed a greater amount of the hairpin structure. To further study the influence of the nucleotide length, the 6-FAM labeled nucleotides at the 5′ end were examined for the native PAGE under the same conditions described above except that the gel electrophoresis image was detected by the fluorescein emission. The DNA sequences not only of those shown in Figure 1A but also of the longer nucleotides of d22cg, d20cg and d18cg indicated in Figure 4A were examined. The results of the gel electrophoresis shown in Figure 4B (obtained at 10 mM NaCl) and 4C (obtained at 1 M NaCl) are essentially the same as shown in Figure 1B and C, respectively, except that the single strands of d10cg and d10ta were observable by the fluorescein labeling. The base pairs of these 10mer DNAs were confirmed to dissociate into a single-strand during the electrophoresis due to the fast dissociation rate of the base pairs rather than due to a low thermal stability (see the Tm data in Table 1). It is also noted that the fluorescein-labeled d14ta gave the NaCl concentration profile similar to that determined from the stains with GelStar (the plots are also included in Figure 2D), indicating less influence of the fluorescein attachment with the nucleotides upon the equilibrium.

Figure 4 (A) Sequences of the DNA nucleotides in addition to those indicated in Figure 1A. Gel electrophoresis at 4°C of (B) 20 μM fluorescein-labeled DNAs at 10 mM NaCl, (C) 20 μM DNAs at 1 M NaCl and (D) 1 μM DNAs at 10 mM NaCl. The marker lanes (MD, MH and MS) are the fluorescein-labeled nucleotides identical to those examined in Figure 1.

Although the nearest-neighbor model predicts that the nucleotide length does not change the free energy cost for the hairpin-to-duplex transition in which two hairpins associate to form the bimolecular duplex (see the Supplementary Data), our observations indicate that the longer nucleotides form a greater amount of the hairpin structure at 10 mM NaCl. It is possible that the hairpin conformation with a long stem is kinetically trapped at low temperature. However, the bimolecular duplexes were formed even at 1 M NaCl at which the kinetic trap would be more significant due to a formation of more stable base pairs, and the equilibrium of the hairpin-to-duplex was shifted by changing the nucleotide strand concentration (Figure 4D), thus reaching equilibrium under the stated conditions. An alternative explanation for the longer nucleotides forming the more of the hairpin structure at 10 mM NaCl is the deviation from the nearest-neighbor model at the low cation concentration. A more significant terminal effect is expected for a hairpin conformation than for a fully matched duplex because a lower amount of cations can condense near the ends of a shorter base paired stem due to a lower electronegative potential of the phosphates (45), that may make the hairpin structure with a short sequence relatively unstable at the low cation concentration.

Secondary structures of the palindrome sequences
To further confirm the hairpin formation by long self-complementary sequences, the palindrome sequences derived from the ColE1 sequence, the origin of the DNA replication in Escherichia coli (3) were examined. The palindrome sequence region of a 31 nt length can adopt a cruciform with hairpin loop structures, and the upstream and downstream of the palindrome sequence are the non-self complementary. The nucleotide sets shown in Figure 5A are the complementary sequences of the nucleotide length ranging from 43 to 19mer. Each sequence is partly self-complementary that can adopt the hairpin with the loop sequence of AAATG or CATTT, and the 43mer (43F and 43T) and 37mer nucleotides (37F and 37T) contain the non-self complementary sequence regions at both termini. The gel electrophoresis with the fluorescent labeled DNAs either by 6-FAM or TAMRA (carboxyltetramethylrhodamine) at the 5′ end demonstrated that each nucleotide measured near the physiological condition at 100 mM NaCl and 37°C adopted the hairpin, and the structure was unaffected by the addition of 10 mM spermine because the mobility of the DNAs on the gel varied less (Figure 5B–F and Supplementary Figure S4). As observed for the short oligonucleotides described above, the 31, 25 or 19mer nucleotides showed the hairpin formation at a low cation concentration even when their perfectly matched nucleotides existed (31F/31T, 25F/25T or 19F/19T, respectively). The amount of the bimolecular duplex formed by the complementary nucleotides with the 31, 25 or 19mer increased with an increase in the spermine concentration, and the concentration required for adopting the bimolecular duplex appeared higher for the longer nucleotides (Figure 5D–F), consistent with the idea that the hairpin formation was preferred for longer DNAs. In contrast, comparisons of the PAGE images of 6-FAM, TAMRA and the GelStar stains revealed that 43F/43T and 37F/37T, containing the non-self complementary sequence regions at both termini, formed the bimolecular duplex even in the absence of spermine (Figure 5B and C). Accordingly, the complementary palindrome sequences can associate when spermine exists at the physiological concentration, and the sequences containing the non-self complementary regions, corresponding to the upstream and downstream of the palindrome sequences, can associate even at a low cationic strength.

Figure 5 (A) Palindrome DNA sequences derived from the ColE1 sequence in E.coli. The nucleotides were labeled either with 6-FAM or by carboxytetramethylrhodamine (TAMRA) at the 5′ end. The terminal non-self complementary sequences are indicated in bold. (B) The PAGE images of the 43mer DNAs stained by GelStar (left), visualized by the fluorescence of 6-FAM (middle), and visualized by the fluorescence of TAMRA (right). The 6-FAM images of (C) 37mer, (D) 31mer, (E) 25mer and (F) 19mer DNAs. The equilibrium was reached at 100 mM NaCl and different spermine concentrations indicated on the top of the gel images (mM) at 37°C. M indicates the 10 bp DNA ladder marker. D and H on the right side of the gel images indicate the bimolecular duplex and the hairpin, respectively.

Implications to the nucleotide structures in a cell
There are abundant metal ions and polyamines including spermine and spermidine in cultured animal cells and they directly influence the gene expressions and protein functions (46). Alkali and alkali earth metal ions preferentially bind to nucleotide phosphates, and the polyamines are reported to bind in the duplex grooves (46–48). Moreover, basic amino acid residues in proteins often participate in the DNA binding through the electrostatic interaction, e.g. observed for histone proteins (49,50). Here, we demonstrated that sodium ion, magnesium ion, spermidine, spermine and the basic oligopeptide as a model of the protein shifted the equilibrium of the self-complementary sequence to a duplex structure caused by a greater stabilization of the bimolecular duplex than the hairpin structure at higher cationic strength. It is reported that the loop–loop kissing complex responsible for the dimerization of the genomic HIV RNA displays a stronger dependence on Na+ and Mg2+ concentration than the regular RNA helix (51). Based on the report, it can be supposed that formation of the intermolecular kissing complex is also more favorable at higher cation concentration. In contrast with the results using cationic molecules, although a high concentration of PEG, which can mimic intracellular molecular crowding media caused by large-sized proteins and their assemblies (39–41), modulated the base pair stabilities, the hairpin-to-duplex transition was affected less by PEG because PEG changed the stabilities of the hairpin and the bimolecular duplex to a similar extent. Although the cation concentration required for the structural change changes depending on the nucleotide strand concentration, our observations indicate the tendency that the elevated salt concentration prefers the self-complementary sequences, especially DNAs, to adopt the structure with more number of the base pairings. The results imply that the cationic molecules seen in cells can stabilize the bimolecular form at the palindrome DNA sequences, and that the self-complementary RNAs containing mismatch nucleotides, which are often observed for intracellular RNAs and they enable a tertiary folding and intermolecular interactions, such as loop–loop interactions (51), tend to form a unimolecular structure while the fully matched RNA sequences, which are often used for a target gene regulation by RNAi, adopt a duplex form.

CONCLUSIONS
A self-complementary nucleotide sequence establishes the equilibria between a unimolecular hairpin and a bimolecular duplex. The hairpin structure was favored at a lower cation concentration while the bimolecular duplex was formed at an elevated cationic strength. The different preference in the secondary structure formation depending on the cationic strength can lead to the structural transition from the hairpin to the bimolecular duplex of self-complementary nucleotides. The structural transition by NaCl has been reported for DNA oligonucleotides containing AT nucleotides around the dyad axis of a palindrome sequence (21,22,25,26). Although the moderate stability in the middle part of the sequence, such as an AT-rich sequence, seemed favor for forming a hairpin structure, we found the structural transition even for the DNAs containing a central GC sequence and that the longer nucleotides showed the tendency to form the hairpin at a low salt concentration. The thermal melting analyzes indicated a greater stabilization of the bimolecular duplexes (ΔTm = ∼20°C) than of the hairpin structures (ΔTm = ∼10°C) when changing the NaCl concentration from 10 mM to 1 M, consistent with the polyelectrolyte theory (37). Furthermore, the structural transitions were also obtained when MgCl2, spermidine, spermine and a basic oligopeptide were used, while PEG only slightly affected the transition, suggesting that the equilibrium shift was primarily caused by electrostatic interactions between the cationic molecules and the nucleotides. Because the cations that shield the electronegative potential of the nucleotide phosphates are rapidly exchangeable with bulk ions, it is not easy to investigate cation effects on the conformational transition of nucleotides by X-ray crystallography, an NMR study or molecular dynamics simulations (45). Our findings suggest the role of cations in shifting the structural equilibrium toward the bimolecular duplex due to the different degrees of cations required for forming the hairpin and the bimolecular duplex, and the observations can be extended to the oligonucleotide designs that work either in a bimolecular or a unimolecular conformation, and construct controlled nucleotide assemblies.

SUPPLEMENTARY DATA
Supplementary Data are available at NAR online.

This work was supported in part by Grants-in-Aid for Scientific Research and ‘Academic Frontier’ Project (2004–2009) from MEXT, Japan. Funding to pay the Open Access publication charges for this article was provided by MEXT, Japan.

Conflict of interest statement. None declared.
==== Refs
REFERENCES
1 Fang X.  Mi Y.  Li J.J.  Beck T.  Schuster S.  Tan W.   Molecular beacons: fluorogenic probes for living cell study Cell Biochem. Biophys. 2002 37 71 81 12482132 
2 Kuwabara T.  Warashina M.  Orita M.  Koseki S.  Ohkawa J.  Taira K.   Formation of a catalytically active dimer by tRNA(Val)-driven short ribozymes Nature Biotechnol. 1998 16 961 965 9788354 
3 Murchie A.I.  Lilley D.M.   Supercoiled DNA and cruciform structures Meth. Enzymol. 1992 211 158 180 1406306 
4 Bowater R.P.  Aboul-ela F.  Lilley D.M.   Large-scale opening of A + T rich regions within supercoiled DNA molecules is suppressed by salt Nucleic Acids Res. 1994 22 2042 2050 8029010 
5 Pearson C.E.  Zorbas H.  Price G.B.  Zannis-Hadjopoulos M.   Inverted repeats, stem–loops, and cruciforms: significance for initiation of DNA replication J. Cell. Biochem. 1996 63 1 22 8891900 
6 Timsit Y.  Moras D.   Cruciform structures and functions Q. Rev. Biophys. 1996 29 279 307 9080546 
7 Wadkins R.M.   Targeting DNA secondary structures Curr. Med. Chem. 2000 7 1 15 10637354 
8 Chworos A.  Severcan I.  Koyfman A.Y.  Weinkam P.  Oroudjev E.  Hansma H.G.  Jaeger L.   Building programmable jigsaw puzzles with RNA Science 2004 306 2068 2072 15604402 
9 Marky L.A.  Breslauer K.J.   The melting behavior of a DNA junction structure: a calorimetric and spectroscopic study Biopolymers 1987 26 1621 1634 3663876 
10 Woese C.R.  Winker S.  Gutell R.R.   Architecture of ribosomal RNA: constraints on the sequence of ‘tetra-loops’ Proc. Natl Acad. Sci. USA 1990 87 8467 8471 2236056 
11 Cheong C.  Varani G.  Tinoco I. Jr   Solution structure of an unusually stable RNA hairpin, GGAC(UUCG)GUCC Nature 1990 346 680 682 1696688 
12 Antao V.P.  Lai S.Y.  Tinoco I. Jr   A thermodynamic study of unusually stable RNA and DNA hairpins Nucleic Acids Res. 1991 19 5901 5905 1719483 
13 Hirao I.  Nishimura Y.  Tagawa Y.  Watanabe K.  Miura K.   Extraordinarily stable mini-hairpins: electrophoretical and thermal properties of the various sequence variants of d(GCGAAAGC) and their effect on DNA sequencing Nucleic Acids Res. 1992 20 3891 3896 1508675 
14 Sugimoto N.  Nakano S.  Yoneyama M.  Honda K.   Improved thermodynamic parameters and helix initiation factor to predict stability of DNA duplexes Nucleic Acids Res. 1996 24 4501 4505 8948641 
15 Gray D.M.   Derivation of nearest-neighbor properties from data on nucleic acid oligomers II. Thermodynamic parameters of DNA.RNA hybrids and DNA duplexes Biopolymers 1997 42 795 10904551 
16 Allawi H.T.  SantaLucia J. Jr   Nearest neighbor thermodynamic parameters for internal G.A mismatches in DNA Biochemistry 1998 37 2170 2179 9485363 
17 Xia T.  SantaLucia J. Jr  Burkard M.E.  Kierzek R.  Schroeder S.J.  Jiao X.  Cox C.  Turner D.H.   Thermodynamic parameters for an expanded nearest-neighbor model for formation of RNA duplexes with Watson–Crick base pairs Biochemistry 1998 37 14719 14735 9778347 
18 Nakano M.  Moody E.M.  Liang J.  Bevilacqua P.C.   Selection for thermodynamically stable DNA tetraloops using temperature gradient gel electrophoresis reveals four motifs: d(cGNNAg), d(cGNABg),d(cCNNGg), and d(gCNNGc) Biochemistry 2002 41 14281 14292 12450393 
19 Nakano S.  Kanzaki T.  Sugimoto N.   Influences of ribonucleotide on a duplex conformation and its thermal stability: study with the chimeric RNA–DNA strands J. Am. Chem. Soc. 2004 126 1088 1095 14746477 
20 Owczarzy R.  You Y.  Moreira B.G.  Manthey J.A.  Huang L.  Behlke M.A.  Walder J.A.   Effects of sodium ions on DNA duplex oligomers: improved predictions of melting temperatures Biochemistry 2004 43 3537 3554 15035624 
21 Marky L.A.  Blumenfeld K.S.  Kozlowski S.  Breslauer K.J.   Salt-dependent conformational transitions in the self-complementary deoxydodecanucleotide d(CGCAATTCGCG): evidence for hairpin formation Biopolymers 1983 22 1247 1257 6850063 
22 Wemmer D.E.  Chou S.H.  Hare D.R.  Reid B.R.   Duplex-hairpin transitions in DNA: NMR studies on CGCGTATACGCG Nucleic Acids Res. 1985 13 3755 3772 4011441 
23 Xodo L.E.  Manzini X.  Quadrifoglio F.  van der Marel G.A.  van Boom J.H.   Oligodeoxynucleotide folding in solution: loop size and stability of B-hairpins Biochemistry 1988 27 6321 6326 3219337 
24 Xodo L.E.  Manzini G.  Quadrifoglio F.  van der Marel G.  van Boom J.H.   Hairpin structures in synthetic oligodeoxynucleotides: sequence effects on the duplex-to-hairpin transition Biochimie 1989 71 793 803 2505860 
25 Garcia A.E.  Gupta G.  Soumpasis D.M.  Tung C.S.   Energetics of the hairpin to mismatched duplex transition of d(GCCGCAGC) on NaCl solution J. Biomol. Struct. Dyn. 1990 8 173 186 2275792 
26 Hald M.  Pedersen J.B.  Stein P.C.  Kirpekar F.  Jecobsen J.P.   A comparison of the hairpin stability of the palindromic d(CGCG(A/T)4CGCG) oligonucleotides Nucleic Acids Res. 1995 23 4576 4582 8524645 
27 Gannett P.M.  Darian E.  Powell J.  Johnson E.M. II  Mundoma C.  Greenbaum N.L.  Ramsey C.M.  Dalal N.S.  Budil D.E.   Probing triplex formation by EPR spectroscopy using a newly synthesized spin label for oligonucleotides Nucleic Acids Res. 2002 30 5328 5337 12466559 
28 Gannett P.M.  Darian E.  Powell J.H.  Johnson E.M.   A short procedure for synthesis of 4-ethynyl-2,2,6,6-tetramethyl-3, 4-dehydro-piperidine-1-oxyl nitroxide Synthetic Commun. 2001 31 2137 2141 
29 Freed J.H.   Berliner L.J.   Theory of slow tumbling EPR spectra for nitroxides Spin Labeling: Theory and Application 1976 Academic Press, NY 53 132 
30 Edwards T.E.  Sigurdsson S.T.   EPR spectroscopic analysis of U7 hammerhead ribozyme dynamics during metal ion induced folding Biochemistry 2005 44 12870 12878 16171402 
31 Puglisi J.D.  Tinoco I. Jr   Absorbance melting curves of RNA Meth. Enzymol. 1989 180 304 325 2482421 
32 Richards E.G.   Fasman G.D.   Use of tables in calculation of absorption, optical rotatory dispersion and circular dichroism of polyribonucleotides Handbook of Biochemistry and Molecular Biology 1975 1 3rd edn Cleveland, OH, USA CRC Press 596 603 
33 SantaLucia J. Jr   A unified view of polymer, dumbbell, and oligonucleotide DNA nearest-neighbor thermodynamics Proc. Natl Acad. Sci. USA 1998 95 1460 1465 9465037 
34 Peyret N.  Seneviratne P.A.  Allawi H.T.  SantaLucia J. Jr   Nearest-neighbor thermodynamics and NMR of DNA sequences with internal A.A, C.C, G.G, and T.T mismatches Biochemistry 1999 38 3468 3477 10090733 
35 Mathews D.H.  Sabina J.  Zuker M.  Turner D.H.   Expanded sequence dependence of thermodynamic parameters improves prediction of RNA secondary structure J. Mol. Biol. 1999 288 911 940 10329189 
36 Zuker M.   Mfold web server for nucleic acid folding and hybridization prediction Nucleic Acids Res. 2003 31 3406 3415 12824337 
37 Longfellow C.E.  Kierzek R.  Turner D.H.   Thermodynamic and spectroscopic study of bulge loops in oligoribonucleotides Biochemistry 1990 29 278 285 2322546 
38 Record M.T. Jr  Anderson C.F.  Lohman T.M.   Thermodynamic analysis of ion effects on the binding and conformational equilibria of proteins and nucleic acids: the roles of ion association or release, screening, and ion effects on water activity Q. Rev. Biophys. 1978 2 103 178 353875 
39 Spink C.H.  Chaires J.B.   Effects of hydration, ion release and excluded volume on the melting of triplex and duplex DNA Biochemistry 1999 38 496 508 9890933 
40 Goobes R.  Minsky A.   Thermodynamic aspects of triplex DNA formation in crowded environments J. Am. Chem. Soc. 2001 123 12692 12693 11741440 
41 Nakano S.  Karimata H.  Ohmichi T.  Kawakami J.  Sugimoto N.   The effect of molecular crowding with nucleotide length and cosolute structure on DNA duplex stability J. Am. Chem. Soc. 2004 126 14330 14331 15521733 
42 Edwards T.E.  Okonogi T.M.  Sigurdsson S.T.   Investigation of RNA-protein and RNA-metal ion interactions by electron paramagnetic resonance spectroscopy. The HIV TAR-Tat motif Chem. Biol. 2002 9 699 706 12079781 
43 Spaltenstein A.  Robinson B.H.  Hopkins P.B.   A rigid and nonperturbing probe for duplex DNA motion J. Am. Chem. Soc. 1988 110 1299 1301 
44 Spaltenstein A.  Robinson B.H.  Hopkins P.B.   DNA structural data from a dynamics probe. The dynamic signatures of single stranded, hairpin-looped, and duplex forms of DNA are distinguishable J. Am. Chem. Soc. 1989 111 2303 2305 
45 Olmsted M.C.  Anderson C.F.  Record M.T. Jr   Importance of oligoelectrolyte and effects for the thermodynamics of conformational transitions of nucleic acid oligomers: a grand canonical Monte Carlo analysis Biopolymers 1991 31 1593 1604 1814506 
46 Tabor C.W.  Tabor H.   Polyamines Annu. Rev. Biochem. 1984 53 749 790 6206782 
47 Deng H.  Bloomfield V.A.  Benevides J.M.  Thomas G.J. Jr   Structural basis of polyamine-DNA recognition: spermidine and spermine interactions with genomic B-DNAs of different GC content probed by Raman spectroscopy Nucleic Acids Res. 2000 28 3379 3385 10954608 
48 Bryson K.  Greenall R.J.   Binding sites of the polyamines putrescine, cadaverine, spermidine and spermine on A- and B-DNA located by simulated annealing J. Biomol. Struct. Dyn. 2000 18 393 412 11149516 
49 Davey C.A.  Sargent D.F.  Luger K.  Maeder A.W.  Richmond T.J.   Solvent mediated interactions in the structure of the nucleosome core particle at 1.9 Å resolution J. Mol. Biol. 2002 319 1097 1113 12079350 
50 Ozdemir A.  Spicuglia S.  Lasonder E.  Vermeulen M.  Campsteijn C.  Stunnenberg H.G.  Logie C.   Characterization of lysine 56 of histone H3 as an acetylation site in Saccharomyces cerevisiae  J. Biol. Chem. 2005 280 25949 25952 15888442 
51 Weixlbaumer A.  Werner A.  Flamm C.  Westhof E.  Schroeder R.   Determination of thermodynamic parameters for HIV DIS type loop-loop kissing complexes Nucleic Acids Res. 2004 32 5126 5133 15459283

