Derivation of nearest-neighbor properties from data on nucleic acid oligomers. I. Simple sets of independent sequences and the influence of absent nearest neighbors

The constraints on combinations of nearest neighbors in nucleic acid sequences and the numbers of independent sequences needed to describe nearest-neighbor properties of oligomers and polymers are derived and summarized. It has been pointed out in previous work [D. M. Gray and I. Tinoco, Jr. (1970) Biopolymers, Vol. 9, pp. 223-244; R. F. Goldstein and A. S. Benight (1992) Biopolymers, Vol. 32, pp. 1679-1693] that these constraints restrict the information available from measurements of properties of sequence combinations. The emphasis in this paper is on the properties of oligomer sequences that vary in length, where each nucleotide or base pair at the end of the sequence makes a significant contribution to the measured property by interacting with its boundary of fixed sequence or solvent. In such cases it is not be possible to determine values of properties of individual nearest neighbors, except for the like neighbors [e.g., d(A-A), d(G-G), d(T-T), and d(C-C) nucleotide neighbors in single-stranded DNA or d(A-A)/d(T-T) and d(G-G)/d(C-C) base pair neighbors in double-stranded DNA], solely from measurements of properties of different sequences. Even values for properties of the like neighbors cannot be determined from such oligomeric sequences if the sequences are all of the same length. Nearest-neighbor properties of oligomer sequences that vary in length can be summarized in terms of the values for independent sets of sequences that are nearest neighbors and monomers all with boundaries of the fixed sequence or solvent. Straightforward combinations of the values for the independent sequences will give the values of the property for any dependent sequence, without explicit knowledge of the individual nearest-neighbor values. These considerations have important consequences for the derivation of widely used thermodynamic parameters, as discussed in the following paper.     

Thermodynamic and structural properties of pentamer DNA.DNA, RNA.RNA, and DNA.RNA duplexes of identical sequence

Four pentamers with the general sequence 5'CU(T)GU(T)G/5'CACAG have been prepared by chemical synthesis in order to generate duplex structures with common sequences. The four duplexes studied include the DNA.DNA duplex (5'dCACAG/5'dCTGTG) and the RNA.RNA duplex (5'rCUGUG/5'rCACAG) as well as the two corresponding DNA.RNA heteroduplexes (5'rCUGUG/5'dCACAG and 5'CACAG/5'dCTGTG). The measured entropy, enthalpy, and free energy changes upon melting are reported for each pentamer and compared to the predicted values where possible. Results show that the two DNA.RNA heteroduplexes are destabilized (delta G degrees 25 = -4.2 +/- 0.4 kcal/mol) relative to either the DNA.DNA duplex (delta G degrees 25 = -4.8 +/- 0.5 kcal/mol) or the RNA.RNA duplex (delta G degrees 25 = -5.8 +/- 0.6 kcal/mol). Circular dichroism spectra indicate that the RNA and the two heteroduplexes adopt an A-form conformation, while the DNA conformation is B-form. Imino proton NMR spectra also show that the heteroduplex structures resemble the RNA.RNA duplex.     

Thermodynamic parameters for an expanded nearest-neighbor model for the formation of RNA duplexes with single nucleotide bulges

Thirty-four RNA duplexes containing single nucleotide bulges were optically melted, and the thermodynamic parameters deltaH degrees, deltaS degrees, deltaG degrees (37), and T(M) for each sequence were determined. Data from this study were combined with data from previous thermodynamic data [Longfellow, C. E., Kierzek, R., and Turner, D. H. (1990) Biochemistry 29, 278-85] to develop a model that will more accurately predict the free energy of an RNA duplex containing a single nucleotide bulge. Differences between purine and pyrimidine bulges as well as differences between Group I duplexes, those in which the bulge is not identical to either neighboring nucleotide, and Group II duplexes, those in which the bulge is identical to at least one neighboring nucleotide, were considered. The length of the duplex, non-nearest-neighbor effects, and bulge location were also examined. A model was developed which divides sequences into two groups: those with pyrimidine bulges and those with purine bulges. The proposed model for pyrimidine bulges predicts deltaG degrees (37,bulge) = 3.9 kcal/mol + 0.10deltaG degrees (37,nn) + beta, while the model for purine bulges predicts deltaG degrees (37,bulge) = 3.3 kcal/mol - 0.30deltaG degrees (37,nn) + beta, where beta has a value of 0.0 and -0.8 kcal/mol for Group I and Group II sequences, respectively, and deltaG degrees (37,nn) is the nearest-neighbor free energy of the base pairs surrounding the bulge. The conformation of bulge loops present in rRNA was examined. Three distinct families of structures were identified. The bulge loop was either extrahelical, intercalated, or in a "side-step" conformation.     

Improved nearest-neighbor parameters for the stability of RNA/DNA hybrids under a physiological condition

The stability of Watson–Crick paired RNA/DNA hybrids is important for designing optimal oligonucleotides for ASO (Antisense Oligonucleotide) and CRISPR (Clustered Regularly Interspaced Short Palindromic Repeats)–Cas9 techniques. Previous nearest-neighbour (NN) parameters for predicting hybrid stability in a 1 M NaCl solution, however, may not be applicable for predicting stability at salt concentrations closer to physiological condition (e.g. ∼100 mM Na⁺ or K⁺ in the presence or absence of Mg²⁺). Herein, we report measured thermodynamic parameters of 38 RNA/DNA hybrids at 100 mM NaCl and derive new NN parameters to predict duplex stability. Predicted ΔG°₃₇ and T_m values based on the established NN parameters agreed well with the measured values with 2.9% and 1.1°C deviations, respectively. The new results can also be used to make precise predictions for duplexes formed in 100 mM KCl or 100 mM NaCl in the presence of 1 mM Mg²⁺, which can mimic an intracellular and extracellular salt condition, respectively. Comparisons of the predicted thermodynamic parameters with published data using ASO and CRISPR–Cas9 may allow designing shorter oligonucleotides for these techniques that will diminish the probability of non-specific binding and also improve the efficiency of target gene regulation.     

Comparison of thermodynamic stabilities between PNA (peptide nucleic acid)/DNA hybrid duplexes and DNA/DNA duplexes.

We have examined quantitatively stabilities of PNA/DNA hybrid duplexes with identical nearest-neighbor base pairs and compared stabilities between PNA/DNA and DNA/DNA. The average difference of stabilization energy of the short PNA/DNA was 0.9 kcal mol(-1), which suggests that the stability of the hybrids with identical nearest-neighbor base pairs can be predicted with the nearest-neighbor model as well as those of nucleic acid duplexes.     

The ups and downs of nucleic acid duplex stability: structure-stability studies on chemically-modified DNA:RNA duplexes.

In an effort to discover novel oligonucleotide modifications for antisense therapeutics, we have prepared oligodeoxyribonucleotides containing more than 200 different modifications and measured their affinities for complementary RNA. These include modifications to the heterocyclic bases, the deoxy-ribose sugar and the phosphodiester linkage. From these results, we have been able to determine structure-activity relationships that correlate hybridization affinity with changes in oligonucleotide structure. Data for oligonucleotides containing modified pyrimidine nucleotides are presented. In general, modifications that resulted in the most stable duplexes contained a heteroatom at the 2'-position of the sugar. Other sugar modifications usually led to diminished hybrid stability. Most backbone modifications that led to improved hybridization restricted backbone mobility and resulted in an A-type sugar pucker for the residue 5'to the modified internucleotide linkage. Among the heterocycles, C-5-substituted pyrimidines stood out as substantially increasing duplex stability.     

Thermodynamics of RNA duplexes modified with unlocked nucleic acid nucleotides

Thermodynamics provides insights into the influence of modified nucleotide residues on stability of nucleic acids and is crucial for designing duplexes with given properties. In this article, we introduce detailed thermodynamic analysis of RNA duplexes modified with unlocked nucleic acid (UNA) nucleotide residues. We investigate UNA single substitutions as well as model mismatch and dangling end effects. UNA residues placed in a central position makes RNA duplex structure less favourable by 4.0–6.6 kcal/mol. Slight destabilization, by ∼0.5–1.5 kcal/mol, is observed for 5′- or 3′-terminal UNA residues. Furthermore, thermodynamic effects caused by UNA residues are extremely additive with ΔG°₃₇ conformity up to 98%. Direct mismatches involving UNA residues decrease the thermodynamic stability less than unmodified mismatches in RNA duplexes. Additionally, the presence of UNA residues adjacent to unpaired RNA residues reduces mismatch discrimination. Thermodynamic analysis of UNA 5′- and 3′-dangling ends revealed that stacking interactions of UNA residues are always less favourable than that of RNA residues. Finally, circular dichroism spectra imply no changes in overall A-form structure of UNA–RNA/RNA duplexes relative to the unmodified RNA duplexes.     

Thermodynamic, counterion, and hydration effects for the incorporation of locked nucleic acid nucleotides into DNA duplexes

A locked nucleic acid (LNA) monomer is a conformationally restricted nucleotide analogue with an extra 2'-O, 4'-C-methylene bridge added to the ribose ring. LNA-modified oligonucleotides are known to exhibit enhanced hybridization affinity toward complementary DNA and RNA. In this work, we have evaluated the hybridization thermodynamics of a series of LNA-substituted DNA octamers, modified to various extents by one to three LNA substitutions, introduced at either adenine (5'-AGCACCAG) or thymine (5'-TGCTCCTG) nucleotides. To understand the energetics, counterion effects, and the hydration contribution of the incorporation of LNA modification, a combination of spectroscopic and calorimetric techniques was used. The CD spectra of the corresponding duplexes showed that the modified duplexes adopt an A-type conformation. UV and DSC melting studies revealed that each type of duplex unfolds in a two-state transition. A complete thermodynamic profile at 5 degrees C indicated that the net effect of modification on thermodynamic parameters might be positional and that the neighboring bases flanking the modification might influence the favorable formation of the modified duplexes. Furthermore, relative to the formation of the unmodified reference duplexes, the formation of modified duplexes is accompanied by a higher uptake of counterions and a lower uptake of water molecules.     

RNA:DNA hybrids are more stable than DNA:DNA duplexes in concentrated perchlorate and trichloroacetate solutions.

Rates of formation of RNA:DNA hybrids have been measured as a function of temperature and compared to DNA:RNA duplex denaturation temperatures in 4 M sodium perchlorate, 4 M NaClO4-6 M urea, and 3 M rubidium trichloracetate solvents. The usual bell shaped curves of reaction rate versus temperature were observed. The optimal temperatures for the RNA:DNA association reaction are 5 degrees to 12 degrees greater than the Tm's for DNA:DNA denaturation in these solvents, just as in formamide. R-loops of phi80d3ilv DNA with E. coli rRNA can be formed at high efficiency in these solvents.     

Thermodynamic parameters based on a nearest-neighbor model for DNA sequences with a single-bulge loop

All 64 possible thermodynamic parameters for a single-bulge loop in the middle of a sequence were derived from optical melting studies. The relative stability of a single bulge depended on both the type of bulged base and its flanking base pairs. The contribution of the single bulge to helix stability ranged from 3.69 kcal/mol for a TAT bulge to -1.05 kcal/mol for an ACC bulge. Thermodynamics for 10 sequences with a GTG bulge were determined to test the applicability of the nearest-neighbor model to a single-bulge loop. Thermodynamic parameters for the GTG bulge and Watson-Crick base pairs predict, DeltaH degrees, DeltaS degrees, and T(M)(50 microM) values with average deviations of 3.0%, 4.3%, 4.7%, and 0.9 degrees C, respectively. The prediction accuracy was within the limits of what can be expected for a nearest-neighbor model. This certified that the thermodynamics for single-bulge loops can be estimated adequately using a nearest-neighbor model.     

Comparison of the electrophoretic and hydrodynamic properties of DNA and RNA oligonucleotide duplexes.

The electrophoretic behavior of defined DNA and RNA oligonucleotide duplexes from 10 to 20 bp in length has been investigated as a function of salt conditions, gel concentration, and temperature. The RNA oligomers migrated much more slowly than the DNA oligomers of the same sequence under all conditions. From sedimentation equilibrium and velocity measurements, the apparent partial specific volume in 0.1 M KCI, 20 mM NaPi, pH 7, was determined as 0.56 +/- 0.015 ml g(-1) for DNA and 0.508 ml g(-1) for RNA. The translational friction coefficients were determined and compared with the values calculated for cylinders. Taking into account the shape factors, the solution density, and partial specific volumes, the effective degree of hydration was estimated as 0.8-1 g g(-1) DNA. There was no significant difference in the frictional coefficients of the DNA and RNA oligomers, indicating that the effective sizes of DNA and RNA are very similar in solution. The differential electrophoretic mobility of DNA and RNA must arise from the differences in interaction with counterions, which is probably a global property of the oligonucleotides.     

Thermodynamic and structural characterization of 2'-nitrogen-modified RNA duplexes

2′-aminonucleosides are commonly used as sites of post-synthetic chemical modification within nucleic acids. As part of a larger cross-linking strategy, we appended alkyl groups onto the N2′ position of 2′-amino-modified RNAs via 2′-ureido and 2′-amido linkages. We have characterized the thermodynamics of 2′-amino, 2′-alkylamido and 2′-alkylureido-modified RNA duplexes and show that 2′-ureido-modified RNAs are significantly more stable than analogous 2′-amido-modified RNAs. Using NMR spectroscopy and NMR-based molecular modeling of 2′-modified RNA duplexes, we examined the effects that 2′-nitrogen modifications have on RNA helices. Our data suggest that the 2′-ureido group forms a specific intra-nucleoside interaction that cannot occur within 2′-amido-modified helices. These results indicate that 2′-ureido modifications are superior to analogous 2′-amido ones for applications that require stable base pairing.     

Analysis of conformational parameters in nucleic acid fragments. II. Co-crystal complexes of nucleic acid bases.

Studies have been made of conformational parameters in co-crystal complexes and compounds of nucleic acid bases in which there is the possibility of formation of hetero-base-pairs. Using published data extracted from the Cambridge structural database, a total of 37 base-pairs were found, of which 25 were hetero-pairs and 12 homo-pairs. These base-pairs were subject to analysis to reveal hydrogen bond parameters, propeller twist, buckle and C1'-C1' separation (or a similar parameter if C1' atoms were not present). Hetero-pairs were found to show larger twists than homo-pairs, the magnitude of twist being unrelated to hydrogen bond parameters or buckle value. The propeller twisting is less pronounced in these nucleic acid bases than in nucleosides, but still has a significant magnitude. Propeller twisting in hetero-pairs is found to be larger than in homo-pairs. Hetero-pairs appear to be formed preferentially in competitive situations.     

Cleavage of DNA.RNA hybrids by type II restriction enzymes.

The action of a number of restriction enzymes on DNA.RNA hybrids has been examined using hybrids synthesised with RNAs of cucumber mosaic virus as templates. The enzymes EcoRI, HindII, SalI, MspI, HhaI, AluI, TaqI and HaeIII cleaved the DNA strand of the hybrids (and possible also the RNA strand) into specific fragments. For four of these enzymes, HhaI, AluI, TaqI and HaeIII, comparison of the restriction fragments produced with the known sequences of the viral RNAs confirmed that they were recognising and cleaving the DNA strand of the hybrids at their correct recognition sequences. It is likely that the ability to utilise DNA.RNA hybrids as substrates is a general property of Type II restriction enzymes.     

Thermodynamic parameters for loop formation in RNA and DNA hairpin tetraloops.

We determined the melting temperatures (Tm) and thermodynamic parameters of 15 RNA and 19 DNA hairpins at 1 M NaCl, 0.01 M sodium phosphate, 0.1 mM EDTA, at pH 7. All these hairpins have loops of four bases, the most common loop size in 16S and 23S ribosomal RNAs. The RNA hairpins varied in loop sequence, loop-closing base pair (A.U, C.G, or G.C), base sequence of the stem, and stem size (four or five base pairs). The DNA hairpins varied in loop sequence, loop-closing base pair (C.G, or G.C), and base sequence of the four base-pair stem. Thermodynamic properties of a hairpin may be represented by nearest-neighbor interactions of the stem plus contributions from the loop. Thus, we obtained thermodynamic parameters for the formation of RNA and DNA tetraloops. For the tetraloops we studied, a free energy of loop formation (at 37 degrees C) of about +3 kcal/mol is most common for either RNA or DNA. There are extra stable loops with delta G degrees 37 near +1 kcal/mol, but the sequences are not necessarily the same for RNA and DNA. The closing base pair is also important; changing from C.G to G.C lowered the stability of several tetraloops in both RNA and DNA. These values will be useful in predicting RNA and DNA secondary structures.     

Thermodynamic contribution and nearest-neighbor parameters of pseudouridine-adenosine base pairs in oligoribonucleotides

Pseudouridine (Ψ) is the most common noncanonical nucleotide present in naturally occurring RNA and serves a variety of roles in the cell, typically appearing where structural stability is crucial to function. Ψ residues are isomerized from native uridine residues by a class of highly conserved enzymes known as pseudouridine synthases. In order to quantify the thermodynamic impact of pseudouridylation on U-A base pairs, 24 oligoribonucleotides, 16 internal and eight terminal Ψ-A oligoribonucleotides, were thermodynamically characterized via optical melting experiments. The thermodynamic parameters derived from two-state fits were used to generate linearly independent parameters for use in secondary structure prediction algorithms using the nearest-neighbor model. On average, internally pseudouridylated duplexes were 1.7 kcal/mol more stable than their U-A counterparts, and terminally pseudouridylated duplexes were 1.0 kcal/mol more stable than their U-A equivalents. Due to the fact that Ψ-A pairs maintain the same Watson-Crick hydrogen bonding capabilities as the parent U-A pair in A-form RNA, the difference in stability due to pseudouridylation was attributed to two possible sources: the novel hydrogen bonding capabilities of the newly relocated imino group as well as the novel stacking interactions afforded by the electronic configuration of the Ψ residue. The newly derived nearest-neighbor parameters for Ψ-A base pairs may be used in conjunction with other nearest-neighbor parameters for accurately predicting the most likely secondary structure of A-form RNA containing Ψ-A base pairs.     

Local mobility of nucleic acids as determined from crystallographic data. II. Z-form DNA

Directions and magnitudes of the local mobility of the Z-DNA hexamer duplex CpGpCpGpCpG have been determined by crystallographic refinement of anisotropic displacement parameters using the observed X-ray diffraction data. The cytidine and guanosine residues demonstrate different modes of mobility, implying that a dinucleotide is the smallest repeating unit in terms of flexibility as well as structure. Directions of librational and translational mobility of the cytidine and guanosine residues of Z-DNA are similar to those observed for the same nucleotides in B-DNA. This suggests that the local mobility of DNA is primarily determined by the individual nucleotide type and by the constraints of Watson-Crick base-pairing, rather than by helical form. Differences in the magnitudes of mobility may be responsible for some of the different physical properties of B-DNA and Z-DNA. The B to Z transition is discussed in terms of the observed flexibilities of these two helical forms.     

Circular dichroism of nucleic acid monomers. II. Derivation and application to polymers of a consistent set of guanine and cytosine transition parameters

An approximate semiempirical procedure has been developed in order to derive nucleic acid monomer π → π* electronic transition moment parameters. Using the approximate procedure, guanine (G) and cytosine (C) transition moment parameters have been derived from agreement found between calculated weight-averaged and measured CD spectra of cyclic-GMP and cyclic-CMP. The derived base transition moment parameters have been assessed in CD spectral calculations on some G- and C-containing nucleic acids for which reasonably good structural information exists. An attempt was also made at evaluating the likely CD spectral contributions of G and C electric n → π* transition moments whose magnitudes were taken to be the maximum expected. Overall, the results indicate that the derived G and C π → π* transition moment parameters are more successful in nucleic acid CD spectral calculations than those used in previous DeVoe theory CD calculations. In addition, the results indicate that electric n → π* transitions may be of importance in understanding nucleic acid monomer CD spectra but appear to be relatively unimportant in understanding nucleic acid polymer CD spectra. It is concluded that the derived G and C π → π* parameters are more useful in DeVoe theory CD calculations than parameters used previously.