Improved thermodynamic parameters and helix initiation factor to predict
stability of DNA duplexes
Improved thermodynamic parameters and helix initiation factor to predict stability of DNA duplexes
Naoki
Sugimoto*
,
Shu-ich
Nakano
,
Mari
Yoneyama
and
Kei-ich
Honda
Department of Chemistry, Faculty of Science, Konan University, 8-9-1 Okamoto, Higashinada-ku,
Kobe
658,
Japan
Received July 26, 1996;
Revised and Accepted September 13, 1996
ABSTRACT
To improve the previous DNA/DNA nearest-neighbor parameters, thermodynamic parameters (
[Delta]
H
o
,
[Delta]
S
o
and
[Delta]
G
o
) of 50 DNA/DNA duplexes were measured. Enthalpy change of a helix initiation factor is also considered though the parameters
reported recently did not contain the factor. A helix initiation factor for
DNA/DNA duplex determined here was the same as that of RNA/RNA duplex (
[Delta]
G
o
37 = 3.4 kcal/mol). The improved nearest-neighbor parameters reproduced not only these 50 experimental values used
here but also 15 other experimental values obtained in different studies. Comparing
[Delta]
G
o
37
values of DNA/DNA nearest-neighbor parameters obtained here with those of RNA/RNA and RNA/DNA, RNA/RNA duplex was generally the most stable of the three kinds of duplexes with the same nearest-neighbor sequences. Which is more stable between DNA/DNA and RNA/DNA duplexes is sequence dependent.
INTRODUCTION
Thermodynamics of enthalpy change ([Delta]
H
o), entropy change ([Delta]
S
o) and free energy change ([Delta]
G
o) for helix formation of nucleic acids and melting temperature (
T
m
) for helix melting can be explained by considering nearest-neighbor interactions. Nearest-neighbor parameters ([Delta]
H
o, [Delta]
S
o and [Delta]
G
o) of DNA/DNA and RNA/RNA double helices (
1
,
2
) were obtained on the basis of the nearest-neighbor model (
3
). Recently, we also reported the nearest- neighbor parameters for RNA/DNA duplexes (
4
). Thus, thermodynamics for double helix formation of DNA/DNA, RNA/RNA and
RNA/DNA can be estimated with the nearest-neighbor parameters (
1
,
2
,
4
,
5
) and their parameters are used to predict stable secondary structures and
active centers of nucleic acids (
6
-
9
).
However, it was reported that there were sometimes larger differences between
measured and predicted thermodynamics of DNA/DNA double helices (
10
,
11
) when the published DNA/DNA nearest-neighbor parameters (
1
) were used. In order to make a much better prediction, as all DNA duplexes consist of 10 nearest- neighbor sequences of dAA/dTT, dAT/dAT, dCG/dCG, dCT/dAG, dGA/dTC, dGC/dGC, dGG/dCC, dGT/dAC, dTA/dTA and dTG/dCA, these 10 DNA
nearest-neighbor parameters should be improved. Also, an initiation factor should
be improved because enthalpy change of a helix initiation factor has been
assumed to be 0 kcal/mol for DNA/DNA and RNA/RNA parameters (
1
,
2
). Although new nearest-neighbor parameters for predicting DNA/DNA duplex stability were reported on the basis of a small number of
thermodynamic results of DNA oligomers (
12
), free energy changes for duplex initiation (1.82 kcal/mol for G/C pair and 2.8 kcal/mol for A/T pair) were quite different from that of RNA (3.4 kcal/mol). Such a large difference between DNA/DNA and RNA/RNA duplex
initiations is amazing. The enthalpy change for DNA/DNA duplex initiation has
to be considered for the improvement, because it was shown that the enthalpy contribution to the duplex initiation was more important than the
entropy contribution in the case of RNA/DNA duplexes (
4
).
Here, we report improved DNA nearest-neighbor parameters determined by measuring thermodynamic parameters for 50 DNA duplexes in 1 M NaCl buffer. The new nearest-neighbor parameters are compared with the previous ones (
1
,
12
).
MATERIALS AND METHODS
Materials
All deoxyribonucleotides (87 nucleotides) were synthesized chemically on a solid support using phosphoramidite procedures and purified with high-performance liquid chromatography (HPLC) after deblocking operations (
13
). The synthesized oligonucleotides were further purified and desalted with a C-18 Sep-Pak cartridge. The final purity was confirmed to be >98% by HPLC.
These oligoucleotides consist of the 10 kinds of nearest-neighbor sequences of Watson-Crick base pairs described above, and their lengths range from 5 to
14 nucleotides. The sequences were selected to have many different combinations
of the nearest-neighbor base pairs.
All experiments were conducted in a buffer including 1 M NaCl, 10 mM Na
2
HPO
4
and 1 mM Na
2
EDTA (pH 7.0). Each oligonucleotide concentration was determined by measuring
the absorbance at 260 or 280 nm at a high temperature as described previously (
14
). Single strand extinction coefficients were calculated from mononucleotide and dinucleotide data by using a nearest-neighbor approximation (
15
), though it should be noted that the data included not a small error. The DNA
strand and its complementary DNA strand were mixed with a 1:1 concentration
ratio to obtain each DNA double helix.
UV measurement
Absorbance measurements in the UV region were made on Hitachi U-3200 and U-3210 spectrophotometers. Melting curves (absorbance versus
temperature curves) were measured at 260 or 280 nm. Cell holders were
thermostated by Hitachi SPR-7 and SPR-10 thermoprogrammers. The water condensation on the cuvette exterior at the low-temperature range was avoided by flushing with a constant
stream of dry N
2
gas. The heating rate was 0.5 or 1.0oC/min.
The melting curves of all the DNA/DNA duplexes examined here showed normal
melting behaviors as well as RNA/RNA and RNA/DNA duplexes (
2
,
4
). Melting data were collected and fitted with NEC PC9801 computers.
T
m
values of Watson-Crick base pairs were obtained with the curve fitting procedure (
16
). Thermodynamic parameters ([Delta]
H
o, [Delta]
S
o and [Delta]
G
o) for DNA/DNA double-helix formation were determined as the average values obtained by
T
m
-1
versus ln(
C
t
) plot and the curve fitting procedure (
2
,
4
).
Calculation of nearest-neighbor parameters
According to the nearest-neighbor model, a free energy change at 37oC ([Delta]
G
o
37
) of the helix formation consist of three terms: (i) a free energy change for
helix propagation as a sum of each subsequent base pair, (ii) a free energy
change for helix initiation and (iii) a free energy change of an entropy effect
when the duplex is composed of self-complementary strands (
1
,
2
,
4
). Nearest-neighbor parameters of DNA/DNA duplexes were determined by using
thermodynamic data for 50 DNA/DNA duplexes measured here and the data for 15
DNA/DNA duplexes examined previously (
1
,
10
,
17
) on the computer program (
4
). End effect by terminal dA/dT pair were not considered, because destability of
the fraying was not so large and it affected the whole duplex stability by only
0.4 kcal/mol (
12
). Each nearest-neighbor parameter of [Delta]
G
o
37
and [Delta]
H
o was described on the program by 0.1 kcal/mol steps so as to predict the
stabilities of all the DNA/DNA duplexes applying to a two-state model with minimum sum of square-error values (
4
). The initiation parameter of [Delta]
G
o
37
was once fixed at 3.4 kcal/mol, because the initiation factor is included to
all 65 DNA/DNA duplexes so that the change of the initiation parameter
significantly affected all base-set parameters. Once the base-set parameters were determined, all nearest-neighbor parameters including the initiation factor were calculated
again. Nearest-neighbor parameters of [Delta]
H
o were obtained in the same way as for [Delta]
G
o
37
determination. The values of [Delta]
S
o were estimated by the determined values of [Delta]
G
o
37
and [Delta]
H
o. Predicted melting temperatures at 100 [mu]M strand concentration were also calculated.
RESULTS
Thermodynamics of DNA/DNA double helices
Measured free energy changes at 37oC ([Delta]
G
o
37
) of d(CTAGTGGA)/d(TCCACTAG), d(GCCAGTTA)/d(TAACTGGC) and d(GGTGCCAA)/d(TTGGCACC), for
example, were -7.5, -8.2 and -9.0 kcal/mol, respectively. One can predict the
thermodynamics of the double helix formation of these DNAs with the previous
nearest-neighbor parameters (
1
). We have compared these measured values with predicted ones by the parameters.
The predicted [Delta]
G
o
37
values of d(GCCAGTTA)/d(TAACTGGC) and d(GGTGCCAA)/d(TTGGCACC) were -7.1 and -9.6 kcal/mol, respectively. The differences between predicted and measured
values were not so small [15.5 and 6.3% for d(GCCAGTTA)/ d(TAACTGGC) and d(GGTGCCAA)/d(TTGGCACC), respectively]. Moreover, the predicted [Delta]
G
o
37
value of d(CTAGTGGA)/ d(TCCACTAG) was -5.3 kcal/mol, which was 41.5% different from the measured value of -7.5 kcal/mol. It was also reported that the differences between measured and predicted thermodynamics by the parameters were sometimes larger than expected (
10
,
11
). Accordingly, we improved the nearest-neighbor parameters for DNA double helix formation.
Improved nearest-neighbor parameters for DNA/DNA duplexes
To determine new nearest-neighbor parameters, we measured the thermodynamics of duplex formation for 50 DNAs. Thermodynamic
parameters of all the duplexes determined with both
T
m
-1
versus ln(
C
t
) plots and the curve fitting procedure (
4
) are provided as supplementary material (Tables S1 and S2). Melting behaviors of all 50 DNAs
used here show a two-state transition, so that they are considered to show similar transition
enthalpy changes as calculated from the shape of the calorimetric curves (
18
,
19
). Thermodynamic parameters for another 15 DNA duplexes (
1
,
10
,
17
) were also considered, so data for 65 duplexes in order to determine the nearest-neighbor parameters are listed in Table S3 as supplementary material.
DNA/DNA double helices have 10 nearest-neighbor sequences of dAA/dTT, dAT/dAT, dCG/dCG, dCT/dAG, dGA/dTC, dGC/dGC, dGG/dCC, dGT/dAC, dTA/dTA and dTG/dCA. We considered these 10 values and the enthalpy effect of a helix initiation for the
prediction which had been ignored (
1
,
2
). The energies of an initiation at A/T and G/C base pairs were regarded as the
same values according to the previous assumption (
2
,
4
). As the helix symmetry factor was due to an entropy effect of associating self-complementary strands, the factor was regarded as the same as that of RNA
double helix (
2
). Thus, the negative entropy change of -1.4 cal/mol/K is used for double helix formation of self-complementary strands.
Improved thermodynamic parameters for DNA/DNA double helix initiation and
propagation in 1 M NaCl buffer
a
Sequence
[Delta]
H
o
[Delta]
S
o
[Delta]
G
o
37
kcal/mol
cal/mol/K
kcal/mol
dAA
-8.0
-21.9
-1.2
dTT
dAT
-5.6
-15.2
-0.9
dTA
dTA
-6.6
-18.4
-0.9
dAT
dCA
-8.2
-21.0
-1.7
dGT
dCT
-6.6
-16.4
-1.5
dGA
dGA
-8.8
-23.5
-1.5
dCT
dGT
-9.4
-25.5
-1.5
dCA
dCG
-11.8
-29.0
-2.8
dGC
dGC
-10.5
-26.4
-2.3
dCG
dGG
-10.9
-28.4
-2.1
dCC
initiation
0.6
-9.0
3.4
self-complementary
0.0
-1.4
0.4
non-self-complementary
0.0
0.0
0.0
a
Estimated errors in [Delta]
H
o, [Delta]
S
o and [Delta]
G
o
37
are +-0.3 kcal/mol, +-1.3 cal/mol/K and +-0.1 kcal/mol, respectively.
The optimized propagation factor for 10 nearest-neighbor sequences and an initiation factor are shown in Table
1
. The [Delta]
H
o, [Delta]
S
o, [Delta]
G
o
37
and
T
m
values of 65 DNA duplexes predicted with the improved parameters are listed as supplementary data in Table S3 together with the measured values.
DISCUSSION
Accuracy of the improved nearest-neighbor parameters
Improved nearest-neighbor parameters have the same tendency as the previous ones though the absolute values are quite different. For
example, the dCG/dCG and dGC/dGC pair have a relatively large stabilization
energy, while dTA/dTA and dAT/dAT have the smallest stability for a helix
formation. Thermodynamics of d(CTAGTGGA)/d(TCCACTAG), d(GCCAGTTA)/d(TAACTGGC) and d(GGTGCCAA)/d(TTGGCACC) which could not be predicted with the previous
parameters as described above are predicted exactly with the improved
parameters. For example, the predicted [Delta]
G
o
37
values were -7.3, -7.8 and -9.2 kcal/mol, respectively, and the differences between the
measured and predicted [Delta]
G
o
37
values were only 2.7, 5.1 and 2.2%, respectively. Recently, SantaLucia Jr
et al
. reported new nearest-neighbor parameters for DNA stability (
12
). Using their parameters for prediction of these duplexes, the predicted [Delta]
G
o
37
values were -6.8, -7.5 and -8.9 kcal/mol, respectively, and the average difference was 6.9%. Thus, our improved nearest-neighbor parameters determined here are more useful to predict DNA
stability than the parameters of SantaLucia Jr
et al
. including the incorrect helix initiation factor described above.
All thermodynamic values of the 65 DNAs were predicted with both improved and
previous parameters, and the values of [Delta]
G
o
37
are shown in Table
2
. The average error of the predicted values by the previous parameters is 23.5%,
while that by the improved parameters is only 4.8%. The improved parameters can
also predict much higher accuracy of the other 15 thermodynamics (average error
of only 5.6%) compared with the previous prediction (that of 27.9%). The
average differences of [Delta]
H
o, [Delta]
S
o and
T
m
of 65 DNA duplexes between measured and predicted values by the improved
parameters are 6.1, 6.8 and 5.7%, respectively. These errors are much smaller
than those by the previous parameters (9.3, 9.4 and 19.5%, respectively). Thus,
the improved parameters can lead to much improved prediction for the stability
of DNA/DNA double helices.
Measured and predicted free energy changes for DNA duplex formations
DNA sequence
a
[Delta]
G
o
37
values (kcal/mol)
Measured
Predicted (I)
b
Predicted (II)
c
AGCCG
-5.6
-5.3
-5.0
ACCGCA
-6.7
-7.0
-6.6
ATGCGC
-7.3
-6.6
-6.7
CGGTGC
-7.2
-7.0
-6.6
CGTGCC
-6.9
-7.0
-6.6
TGCGCA
-6.9
-7.0
-6.9
AATACCG
-5.9
-6.0
-5.7
AGCCGTG
-8.5
-8.5
-7.9
AGCTTCA
-6.1
-6.3
-5.1
GGACTTA
-5.6
-5.3
-3.9
AAAAAAAA
-5.1
-5.0
-5.7
AAGCGTAG
-8.0
-8.3
-7.1
AATCCAGT
-6.8
-7.0
-6.1
ACATATGT
-5.7
-5.3
-3.4
ACCTAGTC
-6.5
-7.1
-4.6
ACGACCTC
-8.7
-9.0
-7.2
AGAGAGAG
-7.4
-7.1
-4.2
AGCGTAAG
-7.8
-8.3
-7.1
AGTCCTGA
-7.5
-7.9
-5.9
ATGCGCAT
-9.0
-8.8
-9.3
CACGGCTC
-10.0
-10.0
-9.3
CCATATGG
-6.8
-6.5
-6.8
CGATATCG
-7.3
-7.5
-7.0
CGCGTATA
-8.4
-8.7
-8.1
CGCTGTAA
-7.9
-8.5
-7.6
CTAGTGGA
-7.5
-7.3
-5.3
CTCACGGC
-9.8
-10.0
-9.3
CTGAGTCC
-8.0
-7.9
-5.9
GAATATTC
-4.3
-4.3
-3.8
GACTAGTC
-6.6
-6.1
-2.8
GAGTACTC
-5.8
-6.1
-2.8
GATTAATC
-4.3
-4.3
-3.8
GCATATGC
-7.8
-6.9
-6.8
GCCAGTTA
-8.2
-7.8
-7.1
GGTGCCAA
-9.0
-9.2
-9.6
GTCGAACA
-8.3
-8.3
-6.6
GTCTAGAC
-6.5
-6.1
-2.8
TAGGCCTA
-8.2
-7.5
-7.0
TATGCATA
-5.9
-5.5
-4.8
AAAAAAAAA
-6.2
-6.2
-7.4
ATAACTGGC
-9.0
-8.7
-8.3
ATCTATCCG
-8.7
-8.7
-8.2
CGCTGTTAC
-9.9
-10.0
-8.7
GCCAGTTAA
-8.8
-9.0
-8.8
AAAAAAAAAA
-6.7
-7.4
-9.1
CGGCAAGCGC
-13.3
-15.6
-17.4
TAGGTTATAA
-7.0
-7.7
-7.0
ACGTATTATGC
-10.4
-11.2
-10.6
ATTGGATACAAA
-10.3
-11.4
-12.0
ACATTATTATTACA
-11.3
-12.0
-11.7
GCGCGC
d
-9.2
-8.7
-9.4
CAAAAAG
d
-4.4
-4.6
-4.7
CGTCGACG
d
-9.8
-10.6
-9.3
GAAGCTTC
d
-6.7
-6.9
-6.0
GGAATTCC
d
-7.4
-6.7
-7.4
GGTATACC
d
-5.5
-6.1
-5.0
CAAAAAAAG
d
-7.3
-7.0
-8.1
CAAACAAAG
d
-7.9
-7.8
-7.6
CAAAGAAAG
d
-7.3
-7.6
-7.4
CAAATAAAG
d
-6.5
-6.4
-6.7
GCGAATTCGC
d
-12.9
-12.7
-14.0
CGCGTACGCGTACGCG
d
-29.1
-27.7
-28.6
CACAG
e
-3.6
-3.0
-0.8
CAACCAACCAAC
f
-14.1
-14.0
-14.2
CTTCCTTCCTTC
f
-14.1
-13.4
-13.6
a
The DNA duplex consists of the denoted DNA strand and its complementary DNA
strand.
b
The values predicted with our parameters in Table 1.
c
The values predicted with the parameters of ref. 1.
d
Data from Breslauer
et al
. (1).
e
Data from Hall
et al
. (10).
f
Data from Ratmeyer
et al
. (17).
Comparison of DNA/DNA nearest-neighbor parameters with RNA/RNA and RNA/DNA parameters
Free energy change of the initiation factor for DNA/DNA was determined to be 3.4
kcal/mol in this study, which was the same as that of RNA/RNA (
2
). As [Delta]
G
o
37
of RNA/DNA was 3.1 kcal/mol for the duplex initiation (
4
), the duplex initiation processes for DNA/DNA, RNA/RNA and RNA/DNA seem to be
not so different energetically. It is reasonable that these helix initiations are similar because each duplex associates two oligonucleotides in the same manner. But, though the enthalpy change of the helix formation had been
ignored for the nearest-neighbor parameters of DNA/DNA (
1
) and RNA/RNA duplexes (
2
), we determined the enthalpic energy for the helix initiation process for RNA/DNA hybrids (
4
). The enthalpy change was 1.9 kcal/mol and the free energy change was 3.1
kcal/mol for RNA/DNA duplexes. That is, the enthalpy change seems to be
dominant for RNA/DNA helix initiation (
4
). On the other hand, [Delta]
H
o and [Delta]
G
o
37
of the initiation factor for DNA/DNA duplexes are 0.6 and 3.4 kcal/mol,
respectively. Thus, at a helix initiation process of DNA/DNA duplexes, an entropy
contribution is dominant for the free energy change.
Figure
1
shows the [Delta]
G
o
37
values of the nearest-neighbor pairs of DNA/DNA duplexes with those of RNA/RNA and RNA/DNA ones
(
2
,
4
). Generally, as shown in Figure
1
, RNA/RNA double helix is the most stable with the same nearest-neighbors. Which is more stable between DNA/DNA and RNA/DNA duplexes
depends on its sequence. Nearest-neighbor values of dCPu/dPyG of DNA/DNA duplexes are about 1.0 kcal/mol
more stable than those of RNA/DNA hybrids. The values of dPyT/dAPu and
dTPu/dPyA of DNA/DNA are also ~0.8 and 0.2 kcal/mol, respectively, more stable than those of RNA/DNA
hybrid. All of these pairs of DNA/DNA duplexes are more stable than those of
RNA/DNA because of their enthalpy advantages. In contrast, the values of
dPuC/dGPy and dPuG/dCPy of DNA/DNA duplexes are ~0.6 kcal/mol less stable than those of RNA/DNA hybrids.
SUPPLEMENTARY MATERIAL
Three tables listing thermodynamic parameters obtained with the
T
m
-1
versus ln(
C
t
) plot (Table S1) and the curve fitting procedure (Table S2), and 65 measured
and predicted thermodynamic data of DNA/DNA duplex formation (Table S3).See supplementary material available in NAR Online.
ACKNOWLEDGEMENTS
We acknowledge M. Katoh, Y. Shintani, T. Ohmichi,.Y. Naraki, D. Miyoshi, I.
Yasumatsu, N. Wakizaka and H. Nakamuta (Konan University) for obtaining
thermodynamic data. This work was supported in part by Grants-in-Aid from the Ministry of Education, Science, Sports and Culture, Japan to N.S.
REFERENCES
1 Breslauer, K. J., Frank, R., Blöcker, H. and Marky, L. A. (1986) Proc. Natl. Acad. Sci. USA83, 3746-3750.
2 Freier, S. M., Kierzek, R., Jaeger, J. A., Sugimoto, N., Caruthers, M. H., Neilson, T. and Turner, D. H. (1986) Proc. Natl. Acad. Sci. USA83, 9373-9377.
3 Tinoco, Jr. I., Uhlenbeck, O. C. and Levine, M. D. (1971) Nature230, 363-367.
4 Sugimoto, N., Nakano, S., Katoh, M., Matsumura, A., Nakamuta, H., Ohmichi, T., Yoneyama, M. and Sasaki, M. (1995) Biochemistry34, 11211-11216.MEDLINE Abstract
5 Turner, D. H., Sugimoto, N. and Freier, S. M. (1990) in Saenger, W. R. (ed.) Landolt-BörnsteinNucleic Acids Vol. 1c, chap. 3.6, Berlin, Germany.
6 Sugimoto, N. and Sasaki, M. (1991) Chemistry Lett. 345-348.
7 Sugimoto, N., Sasaki, M. and Tanaka, A. (1993) Supramolec. Chem.2, 99-102.
8 LeCuyer, K. A. and Crothers, D. M. (1993) Biochemistry32, 5301-5311.
9 Connell, G. J. and Yarus, M. (1994) Science 264, 1137-1141.
10 Hall, K. B. and McLaughlin, L. W. (1991) Biochemistry30, 10606-10613.
11 Sugimoto, N., Honda, K. and Sasaki, M. (1994) Nucleosides Nucleotides 13, 1311-1317.
12 SantaLucia, Jr. J., Allawi, H. T. and Seneviratne, P. A. (1996) Biochemistry35, 3555-3562.MEDLINE Abstract
13 Kierzek, R., Caruthers, M. H., Longfellow, C. E., Swinton, D., Turner, D. H. and Freier, S. M. (1986) Biochemistry25, 7840-7846.MEDLINE Abstract
14 Sugimoto, N., Kierzek, R., Freier, S. M. and Turner, D. H. (1986) Biochemistry25, 5755-5759.MEDLINE Abstract
15 Richards, E. G. (1975) in Fasman, G. D., (ed.) Handbook of Biochemistry and Molecular Biology: Nucleic Acids 3rd. ed., Vol. 1, 597, CRC Press, Cleveland, OH.
16 Petersheim, M. and Turner, D. H. (1983) Biochemistry22, 256-263.MEDLINE Abstract
17 Ratmeyer, L., Vinayak, R., Zhong, Y. Y., Zon, G. and Wilson, D. (1994) Biochemistry33, 5298-5304.MEDLINE Abstract
18 Marky, L. A. and Breslauer, K. J. (1987) Biopolymers26, 1601-1620.
19 Zieba, K., Chu, T. M., Kupke, D. W. and Marky, L. A. (1991) Biochemistry 30, 8018-8026.MEDLINE Abstract
