Molecular Dynamics Study of Oligonucleotides Containing Difluorotoluene

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

J. Am. Chem. Soc.

2000, 122, 6891-6899

6891

Molecular Dynamics Study of Oligonucleotides Containing


Difluorotoluene
Elena Cubero, Charles A. Laughton, F. Javier Luque, and Modesto Orozco*,
Contribution from the Departament de Bioqumica, Facultat de Qumica, UniVersitat de Barcelona,
Mart i Franque` s 1, Barcelona 08028, Spain, Cancer Research Laboratories, School of Pharmaceutical
Sciences, UniVersity of Nottingham, NG7 2RD, UK, and Departament de Fisicoqumica, Facultat de
Farma` cia, UniVersitat de Barcelona, AVgda Diagonal sn, Barcelona 08028, Spain
ReceiVed January 10, 2000

Abstract: Extended molecular dynamics (MD) and thermodynamic integration (MD-TI) calculations have been
used to determine the structural and energetic changes in DNA that accompany the replacement of thymine
(T) by the nonnatural isostere difluorotoluene (F). In a duplex DNA oligonucleotide, it is found that the TfF
mutation leads to only small changes in the average structure, but to important alterations in flexibility, hydration,
and recognition properties. The TfF mutation in the Watson-Crick or Hoogsteen position of a pyrimidine
purinepyrimidine type DNA triplex does not lead to dramatic changes in the general structure of the triplex,
but again, detailed analysis shows some alterations in flexibility, hydration, and recognition properties. MDTI calculations on the TfF mutation in duplex DNA reproduce the experimentally determined free energy
differences with good accuracy, and detailed analyses of the trajectories have enabled us to rationalize these.
Finally, MD-TI simulations have been used to predict the changes in stability of a triplex due to a TfF
mutation in either the Watson-Crick or Hoogsteen-binding pyrimidine strands. We predict that in either case
the mutation will reduce stability, being most unfavorable in the Watson-Crick strand.

Introduction
DNA replication by DNA polymerase is a high-fidelity
process with an error rate of the order of one mismatch per 104
to 105 bases.1,2 It might be thought that this fidelity has its origins
in the specificity of Watson-Crick hydrogen bonding, but it
has been argued that this is not the case. The free energy
difference between matched and mismatched terminal base pairs
in solution is estimated3 to range between 0.2 and 0.4 kcal
mol-1. This difference is too low to account for the observed
error rate, suggesting that shape-complementarity must also play
an important role in ensuring the fidelity of replication.4
However, other workers5 have argued that within the environment of the polymerase, the free energy difference between
matched and mismatched base pairs may be much greater due
to reduced solvation and so the hydrogen-bonding argument is
sufficient.
These competing hypotheses have been tested using the
thymine mimic difluorotoluene (F, Figure 1). F was designed
as a nonpolar homologue of thymine,6 which lacks H-bonding
capability, even in apolar solvents such as chloroform.7 Despite
its nonpolar nature, it was found that DNA polymerase I would

Facultat de Qumica, Universitat de Barcelona.


University of Nottingham.
Facultat de Farma
` cia, Universitat de Barcelona.
(1) Kornberg, A.; Baker, T. A. DNA Replication, 2nd ed.; W. H.
Freeman: New York.
(2) Goodman, M. F.; Creighton, S.; Bloom, L. B.; Petruska, J. Crit. ReV.
Biochem. Mol. Biol. 1993, 28, 83.
(3) Petruska, J.; Goodman, M. F.; Boosalis, M. S.; Sowers, L. C.; Cheong,
C. Proc. Natl. Acad. Sci. U.S.A. 1988, 85, 6252.
(4) Echols, H.; Goodman, M. F. Annu. ReV. Biochem. 1991, 60, 477.
(5) Johnson, K. A. Annu. ReV. Biochem. 1993, 62, 685.
(6) Guckian, K. M.; Kool, E. T. Angew. Chem., Int. Ed. 1997, 36, 2825.
(7) Moran, S.; Ren, R. X.-F.; Kool, E. T. Proc. Natl. Acad. Sci. U.S.A.
1997, 94, 10506.

incorporate F across from A, and A across from F, in a precise


fashion.7,8 Despite this specificity, thermal denaturation studies9
show that replacing T by F destabilized DNA duplexes by 3.03.6 kcal mol-1. The significance of these results was challenged
by Evans and Seddon,10 who argued on the basis of quantum
mechanical calculations that F was a significantly polar molecule
and could form nonclassical hydrogen bonds. Recently, this
conclusion has been disputed by Wang and Houk,11 whose
quantum mechanical and molecular mechanical calculations
support the view that F does not hydrogen bond.
The structure of an AF-containing DNA dodecamer has been
recently determined by NMR.12 The structure refinement
involved numerous short (approximate 25 ps) molecular dynamics (MD) simulations with NMR-derived distance restraints. No
unusual behavior of the dodecamer was observed during the
MD simulations, and the refined structure showed standard
B-type characteristics. On the theoretical side, very recent MD
simulations13 have confirmed that the TfF mutation does not
induce major changes in the average structure of a duplex, but
revealed important changes in its flexibility. It was possible to
detect, within a 10 ns trajectory, several breathing movements
of the AF base pair, whereas normal AT and GC base pairs
only breath on the microsecond time scale.
In contrast to the wealth of information on the effect of TfF
mutations on the structure and stability of duplex DNA, there
(8) Liu, D.Y.; Moran, S.; Kool, E. T. Chem. Biol. 1997, 4, 919.
(9) Moran, S.; Ren, R. X.-F.; Rumney, S.; Kool, E. T. J. Am. Chem.
Soc. 1997, 119, 2056.
(10) Evans, T. A.; Seddon, K. R. J. Chem. Soc., Chem. Commun. 1997,
2023.
(11) Wang, X.; Houk, K. N. J. Chem. Soc., Chem. Commun. 1998, 2631.
(12) Guckian, K. M.; Krugh, T. R.; Kool, E. T. Nature Struct. Biol. 1998,
5, 954.
(13) Cubero, E.; Sherer, E. C.; Luque, F. J.; Orozco, M.; Laughton, C.
A. J. Am. Chem. Soc. 1999, 121, 8653.

10.1021/ja000117n CCC: $19.00 2000 American Chemical Society


Published on Web 07/08/2000

6892 J. Am. Chem. Soc., Vol. 122, No. 29, 2000

Cubero et al.

Figure 1. Chemical structure of 1,3-difluorotoluene (F) and thymine bound to adenine in the Watson-Crick and Hoogsteen orientations.

Scheme 1. DNA Duplex and Triplex Sequences Used in This Study

have been, to our knowledge, no similar studies looking at


triplex DNA. In this paper we present an extension of our
previous MD simulations of F-containing oligonucleotides. In
addition to a 10 ns trajectory of a duplex containing a TfF
mutation (and the corresponding 1.5 ns of trajectory for a control
duplex containing no mutation), we have produced 3 ns
trajectories for DNA triplexes containing the TfF mutation in
both the Watson-Crick (d(T-AF) motif) and the Hoogsteen
(d(F-AT) motif) positions, and a 1.5 ns control trajectory for
the parent unmutated triplex. In addition, a series of MD-TI
simulations have been performed to study in more detail the
influence of the TfF mutation on the stability of both duplex
and triplex DNA. MD-TI simulations are expected to be very
useful to determine the stability of nucleic acid structures
containing unnatural bases pairs,14,15 but due to technical
problems MD-TI calculations on these systems are scarce.
Calculations provide a detailed view on the changes in
structure, flexibility, reactivity, and stability induced in duplex
and triplex DNA by the TfF mutation. The implications of
(14) Soliva, R.; Luque, F. J.; Orozco, M. Nucleic Acid. Res. 1999, 27,
2248.
(15) Hernandez, B.; Soliva, R.; Luque, F. J.; Orozco, M. J. Am. Chem.
Soc. 2000. Submitted for publication.

the results in the design of new antigene and antisense strategies


are discussed.
Methods
MD Simulations. All molecular dynamics simulations were performed using the AMBER 5.0 suite of programs16 in association with
the AMBER 95 force field. Missing AMBER parameters for difluorotoluene were those previously determined in our previous studies.13
All simulations were performed at constant temperature (300 K) and
pressure (1 atm). Long-range electrostatic interactions were handled
using the particle mesh Ewald (PME)17 method. SHAKE18 was used
to constrain all bonds, allowing the use of a 2 fs time step.
The DNA sequences studied are shown in Scheme 1. The duplex
structures 1 and 2 were constructed in standard B-type conformation.19
Triplexes 3, 4, and 5 were generated using our equilibrated structure
(16) Case, D. A.; Pearlman, D. A.; Caldwell, J. W.; Cheatham, T. E.;
Ross, W.S.; Simmerling, C. L.; Darden, T. A.; Merz, K. M.; Stanton, R.
V.; Cheng, A. L.; Vincent, J. J.; Crowley, M.; Ferguson, D. M.; Radmer,
R. J.; Seibel, G. L.; Singh, U. C.; Weiner, P. K.; Kollman, P. A. AMBER
5, University of California: San Francisco, 1997.
(17) Cheetham. T. E.; Miller, J. L.; Fox, T.; Darden, T. A.; Kollman, P.
A. J. Am. Chem. Soc. 1995, 117, 4193.
(18) Ryckaert, J. P.; Ciccotti, G.; Berendsen, H. J. C. J. Comput. Phys.
1977, 23, 327.
(19) Arnott, S.; Hukins, W. L. Biochem. Biophys. Res. Commun. 1972,
47, 1504.

Oligonucleotides Containing Difluorotoluene

J. Am. Chem. Soc., Vol. 122, No. 29, 2000 6893


Table 1. Summary of MD/TI Simulationsa
mutation
FfT
FfT
FfT
FfT
FfT
TfF
TfF
FfT
TfF(WC)
TfF(H)
FfT(WC)
FfT(WC)
FfT (H)
FfT (H)

system
SS (3 steeps)
SS (3 steeps)
SS (3 steeps)
SS (3 steeps)
SS (5 steeps)
duplex d(AT)
duplex d(AT)
duplex d(AF)
triplex d(AT-T)
triplex d(AT-T)
triplex d(AF-T)
triplex d(AF-T)
triplex d(AT-F)
triplex d(AT-F)

structure

length of MD
psb

equil. for 340


equil. for 340 psc
equil. for 340 ps
annealing (520 ps)d
equil. for 340 psb
equil. for 1500 ps
equil. for 1500 ps
equil. for 1500 ps
equil. for 1000 ps
equil. for 1000 ps
equil. for 1000 ps
equil. for 1000 ps
equil. for 1000 ps
equil. for 1000 ps

420 ps
420 ps
840 ps
420 ps
420 ps
420 ps
840 ps
420 ps
420 ps
420 ps
420 ps
840 ps
420 ps
840 ps

a For more details of the simulations see text. SS means single


stranded, WC refers to the Watson-Crick position, and H refers to
the Hoogsteen position. b B-type single stranded structures were generated, surrounded by counterions and around 850 water molecules.
Structures were then minimized, heated, and equilibrated for 340 ps.
c As in footnote b, but a large box of 2851 water molecules (similar to
those used for duplex structures) was used. d As in footnote b, but a
180 ps simulated annealing (heating at 500 K for 80 ps and cooling at
300 K for 100 ps) was used to obtain a different starting conformation.

Figure 2. Thermodynamic cycles used in the MD/TI calculations.


Average results obtained from the simulations are displayed with their
standard errors (all in kcal/mol).

of the d(AT-T) triplex.20 Periodic boxes were constructed around the


systems containing sodium counterions (to achieve neutral systems)
and 2482 (duplex structures) or 3278 (triplex structures) TIP3P water
molecules.21 The systems were energy minimized, heated, and equilibrated using our standard equilibration process.20 The equilibrated
systems were subjected to 1.5 (1, 3), 10 (2), and 3 (4, 5) ns unrestrained
MD simulations.
Energy analysis was done using AMBER-5.0 as well as in-house
programs. Helical parameters were calculated using CURVES.22 Where
CURVES yields both global and local versions of parameters, the local
parameter values are quoted.
Free Energy Calculations. MD-TI calculations and standard
thermodynamic cycles (see Figure 2) were used to determine the effect
of the TfF mutation on the stability of duplexes and triplexes. Starting
structures for the MD-TI calculations were determined after extensive
MD equilibration (see Table 1). Mutations were typically performed
in both the TfF and FfT directions to verify the reversibility of the
mutation pathways. Simulations were performed using 21 windows of
10/20 ps of equilibration and 10/20 ps of collection for a total of 420
or 840 ps. In all cases, free energy estimates obtained using the first
half (equilibration) and second half (data collection) of each window
were collected. The existence of several estimates of the free energy
(20) Shields, G. C.; Laughton, C. A.; Orozco, M. J. Am. Chem. Soc.
1997, 119, 7463.
(21) Jorgensen, W. L.; Chandresekhar, J.; Madura, J.; Impey, R. W.;
Klein, M. L. J. Chem. Phys. 1983, 79, 926.
(22) Lavery, R.; Sklenar, J. J. Biomol. Struct. Dyn. 1988, 6, 63.

Figure 3. RMS deviations in the trajectories of the d(AAGAAAGAAAAG) duplex containing T or F bound to the adenine in position
6. The continuous line shows the deviation with respect to the average
structure obtained from the same trajectory. The dashed line shows
the deviation with respect to the average structure from the other
trajectory.
change associated with a particular mutation allows us to obtain a good
estimate of the statistical errors in the free energy estimates.
Trajectories for MD-TI calculations were performed considering
long-range contributions as introduced by the PME method, and
considering all inter- and nonbonded intramolecular contributions. All
the technical details of simulations used for TI calculations are identical
with those used in standard MD simulations (see above).

Results and Discussion


MD Simulations. (a) Duplex DNA. MD simulations of the
d(AAGAAAGAAAAG) DNA duplex with T or F paired to
adenine at position six lead to stable trajectories in which the
conformation of the duplex remains closer to B than to A DNA.
Comparison of the two trajectories shows (Figure 3) that from
a general structural point of view they are very similar. Thus,
the average RMS deviation between the MD-averaged structure
of 2, and the trajectory of 1 is 1.9((0.3) . Conversely, the

6894 J. Am. Chem. Soc., Vol. 122, No. 29, 2000


Table 2. Selected MD-Averaged Helical Parameters for the
Central 10 Base Pairs of the Dodecamers Containing a D(AT) and
a D(AF) Stepa
parameters
X-displacement ()
inclination (deg)
rise ()
roll (deg)
twist (deg)
propeller twist (deg)
phase angle (deg)
puckering amplitude (deg)
major groove width ()
minor groove width ()

duplex with d(AT) duplex with d(AF)


-1.7 (0.7)
-8.2(4.7)
3.5(0.2)
6.3(2.5)
30.3(2.0)
-11.2(5.3)
119(31)
40.9(6.0)
21.5(1.7)
12.3(0.9)

-1.4(0.6)
-8.5((4.9)
3.5(0.2)
6.9(2.3)
30.0(1.5)
-11.1(4.5)
116 (35)
40.7(6.2)
22.1(1.7)
12.8(0.9)

a
Averages are done using the last 5 ns of the trajectory of the duplex
with the d(AF) step and the last 1 ns of the trajectory for the control
duplex.

average RMS deviation between the MD-averaged structure of


1 and the trajectory of 2 is 2.1((0.4) . These values are almost
identical with those found when the MD-averaged structures
of 1 and 2 are compared with their own trajectories (1.5((0.4)
for 1 and 1.8((0.4) for 2), demonstrating the identity
between the different trajectories.
Results in Table 2 confirm the B-like nature of the two
duplexes. It is also clear that the TfF mutation leads to very
small changes in the helical parameters of the duplex. The
grooves are also very similar in normal DNA and that containing
F. All sugars show a B-type puckering with major population
of the East and South regions. As typically found in simulations
of DNA with AMBER-95 force field,16 twist values are slightly
underestimated with respect to a canonical B-type duplex, but
in any case almost the same twist is found for both duplexes,
confirming the small impact the TfF mutation has on the
average structure of the duplex.
Despite the close similarity in the general structure of the
two duplexes, there are detectable differences in the region of
the mutation. Thus, the TfF mutation leads to small changes
in the recognition characteristics of duplex DNA, as evident in
the MIP 20,23 map (Figure 4). This shows that, compared to 1,
2 has a reduced affinity for cationic probes in the minor groove.
However, this altered affinity is very localized to the mutation
site. MD-derived solvation maps20 (Figure 4) also show a
reduced solvation in the minor groove in the region close to
the AF pair, as compared to the AT one. These differences may
be rationalized by considering first the alteration in the molecular
electrostatic potential (see Figure 5) due to the lack of the lone
pair at position 2 in difluorotoluene and second the existence
of breathing movements (see below) which lead to a partial
disruption of the solvent atmosphere surrounding the mutated
base pair.
Though we observe no significant difference between 1 and
2 in terms of their global structures, there is a clear difference
between them in terms of their dynamics. The most important
aspect of this difference relates to the existence of a breathing
motion, which leads to a partial or total reversible opening of
the AF pair, in contrast to the rigid behavior of the AT pair.
The breathing events occur on a nanosecond time scale (see
Figure 6 and ref 13), and have been shown to originate from
the related to the reduced stability of the AF Watson-Crick
dimer compared to that of the AT pair.
(b) Triplex DNA. Three triplexes have been studied, with
the sequences shown in Figure 1. Triplex 3 contains the
unmodified d(AT-T) trio at position 6, while the other two
(23) Shields, G.; Laughton, C. A.; Orozco, M. J. Am. Chem. Soc. 1998,
120, 5895.

Cubero et al.
contain a TfF mutation: d(AF-T) in 4 and d(AT-F) in 5.
In all cases MD simulations produce stable trajectories over
the nanosecond time scale at room temperature and nearphysiological conditions, suggesting that triplexes containing a
single TfF mutation can be stable in physiological conditions.
The triplex conformations remain essentially B-type. Thus,
the RMS deviations from a canonical B-type model average
1.3 while the RMS deviations from a canonical A-type model
lie in the range 2.3-2.4 . In a manner similar to the duplexes,
convergence of the three trajectories is evident from comparison
of RMS deviations. This is shown in Table 3, where the cross
RMS deviations between triplexes containing F and T are 1.01.1 . These values are similar in magnitude to the RMS
deviations found between individual structures from a trajectory
and the MD-averaged structure derived from it.
The TfF mutation does not have any major impact on the
helical characteristics of the triplexes, which are those expected
for a B-type triplex20,24,2524,25 (see Table 4). In all the cases the
twist is around 30 with a rise around 3.3-3.4 . No change
in groove dimensions or sugar puckering (average phase angle
around 110-112) is apparent as a consequence of the TfF
mutation either in the Watson-Crick or Hoogsteen positions.
The general recognition and hydration characteristics of the
triplex remain unaltered upon the TfF mutation in the WatsonCrick position. However, such a mutation weakens the region
of negative potential in the minor groove, and a partial disruption
of the spine of hydration (see Figure 7). There is also a slight
decrease in the apparent density of water in the minor part of
the major groove (for nomenclature see ref 20). As for the
duplexes (see above), these observations can be accounted for
by considering two factors: first the loss of the lone pair at
position 2 of the pyrimidine, resulting from the TfF mutation
(see Figure 5), and second the breathing motions in the AF
pair (see below).
The TfF mutation in the Hoogsteen position does not lead
to any important change in the electronegativity in the grooves
or in their apparent water density (see Figure 7). The lack of
difference in the minor part of the major groove is particularly
interesting, since it should experience a partial loss in H-bonding
capabilities as a result of the TfF mutation. Comparing with
the results for triplex 4, this suggests that the integrity of the
spines of hydration is more sensitive to the existence of
breathing and opening movements (see ref 13 for nomenclature) than to the partial loss of H-bonding capability.
Breathing movements are detected in triplexes containing F
(4 and 5), while they are not found in the reference triplex 3.
The triplex with the TfF mutation in the Watson-Crick
position (4) shows reversible partial opening and breathing
movements (see ref 13 for nomenclature), which lead to partial
or total loss of WC hydrogen bonds due to the displacement of
F to the major groove. Analysis of the trajectory (see Supporting
Information for details) suggests free energy changes of 0.91.3 kcal/mol for partial opening, and 2.0-2.4 kcal/mol for
breathing. These values are only around 0.5 kcal/mol higher
than those found for a duplex DNA with a single TfF
mutation,13 which suggests that the breathing of WatsonCrick base pairs is not dramatically modified by the presence
of a third strand.
The triplex with the TfF mutation in the Hoogsteen position
(5) shows reversible breathing movements, where the F is
displaced to the major groove, leading to the partial or total
(24) Ragunnathan, G.; Miles, H. T.; Sasisekharan, V. Biochemistry 1993,
32, 455.
(25) Howard, F. N.; Miles, H. T.; Liu, K.; Frazier, J.; Raghunathan, G.;
Sasisekharan, V. Biochemistry 1993, 31, 10671.

Oligonucleotides Containing Difluorotoluene

J. Am. Chem. Soc., Vol. 122, No. 29, 2000 6895

Figure 4. Molecular interaction potential (MIP) and solvation maps obtained for the duplexes containing T or F. The MIP map is contoured at
-5.0 kcal/mol, and the solvation map is contoured at a water density of around 2.5 g/cm3.

loss of Hoogsteen hydrogen bonds. The free energies for partial


and total opening are estimated to be around 1.9-2.6 and 3.14.1 kcal/mol, respectively. These values are 1-2 kcal/mol larger
than those found for Watson-Crick breathing of AF pairs. This
strongly suggests that Hoogsteen breathing (at least for AF
pairs) is a less common event than Watson-Crick breathing,
which might suggest a greater stability for the d(AT-F) trio
compared to the d(AF-T) trio.
Free Energy Calculations. Figure 8 shows the free energy
profiles associated with the TTF mutation in single-stranded

oligonucleotides, duplex DNA, and triplex DNA, obtained


considering different starting conformations, both the TfF and
FfT directions, and using trajectories of different lengths (see
Table 1).26 All the free energy profiles (Figure 8) are smooth,
without apparent discontinuities which could signal the existence
of hysteresis. Moreover, the different free energy changes
associated with a given mutation, which are determined from
(26) In all cases the gas phase profile is subtracted to produce a
common reference state for all the simulations.

6896 J. Am. Chem. Soc., Vol. 122, No. 29, 2000

Cubero et al.

Figure 5. Ab initio HF/6-31G(d) molecular electrostatic potential map computed at 3.4 above the aromatic plane. Colored areas correspond to
MEP values from -10 (deep red) to +10 (deep blue) kcal/mol.

Figure 6. Structure of three snapshots showing normal AF pair (black)


and pairs with breathing of A (dark gray) and breathing of F (light
gray).
Table 3. RMS Deviations (in ) between Trajectories (rows) and
MD-Averaged Structures (columns)a
d(AT-T)
d(AF-T)
d(AT-F)
a

d(AT-T)

d(AF-T)

d(AT-F)

0.97(0.16)
1.11(0.21)
1.03(0.18)

1.05(0.18)
1.10(0.21)
1.07(0.17)

1.08(0.22)
1.11(0.19)
1.01(0.18)

Standard deviations in the averages are displayed in parentheses.

Table 4. Selected MD-Averaged Helical Parameters for the


Central 10 Base Pairs of the Triplexes Containing a d(AT-T), a
d(AF-T), and a d(AT-F) Stepa
parameters

triplex
d(AT-T)

triplex
d(AF-T)

triplex
d(AT-F)

X-displacement ()
inclination (deg)
rise ()
roll (deg)
twist (deg)
propeller twist (deg)
phase angle (deg)
puckering amplitude (deg)
MM groove width ()
mM groove width ()
m groove width ()

-3.0(0.5)
-3.2(4.7)
3.4(0.1)
3.2(1.8)
29.6(0.7)
-12.2(4.7)
111(29)
41.5(5,8)
15.6(1.2)
9.0(0.3)
11.5(0.5)

-3.0(0.6)
-1.6(5.0)
3.4(0.1)
3.6(1.8)
29.5(0.6)
-12.8(4.7)
110(38)
41.5(5.8)
15.1(1.0)
9.0(0.3)
12.1(0.5)

-3.1(0.5)
-1.4(4.3)
3.4(0.1)
3.1(1.8)
29.7(0.7)
-13.5(4.8)
113(26)
41.7(5.7)
15.3(0.8)
9.0(0.3)
11.5(0.5)

a Averages are done using the last 1.5 (reference triplex) and 2.0
(triplexes with mutation) ns of the trajectories. MM means major-major
groove, mM means minor-major groove, and m refers to the minor
groove. See ref 20 for nomenclature.

different simulations, are in very close agreement (below 0.3


kcal/mol), giving us confidence in the statistical quality of these
estimates.
Free energy differences around -8.1 kcal/mol are found for
mutations in single-stranded DNA (irrespective of the size of
the oligonucleotide) favoring hydration of T in front of F (see

Figures 2 and 8). This value agrees well with SCRF estimates
(-8.6 kcal/mol) obtained at the AM1/MST level,27 which gives
us further confidence in the quality of the MD/TI calculations.
Simple algebra with free energy differences in Figure 2 gives
us all the thermodynamic data associated with the effect of TfF
mutations on duplex and triplex stability. Results in Table 5
show that the TfF mutation destabilizes the duplex DNA by
about 5 kcal/mol, in good agreement with experimental values
found by Kool and co-workers (values around 4 kcal/mol from
refs 6, 7, and 9). Analysis of the central d(GpApA) sequence
of the MD-averaged structures (see Table 6) shows that the
largest difference between T and F in terms of duplex stability
lies in the H-bonding term, which is near 10 kcal/mol more
stable for duplexes containing the d(AT) pair.28
In summary, MD and MD/TI simulations strongly suggest
that the presence of a single d(AF) step destabilizes the duplex,
but the resulting DNA is still stable at room temperature, in
good agreement with all known experimental evidence.6,7,9 In
practice, the largest differences between a DNA duplex containing a d(AF) pair and the natural one should be related to (i)
the change in the intrinsic reactivity in the minor groove, which
might modify the ability of DNA to interact with small
molecules or some DNA minor-groove binding proteins, and
(ii) the existence of breathing events which might also affect
the reactivity of DNA, and contribute to the activation of DNArepairing systems.
The formation of most parallel-motif triplexes is a two-step
process. First a DNA duplex is formed, and second the triplexforming oligonucleotide is bound29 to give the triplex. The
transition from a duplex with a d(AT) pair to a triplex with a
d(AT-T) trio is around 3.2 kcal/mol more favorable than the
transition from the same duplex to a triplex with a d(AT-F)
trio (see Table 5). Interestingly, if the duplex contains a d(A
F) pair, the formation of a d(AF-T)-containing triplex is not
disfavored with respect to the formation of a triplex with a d(A
T-T) trio from a duplex with the d(AT) pair (differential free
(27) (a) Luque, F. J.; Negre, M.; Orozco, M. J. Phys. Chem. 1993, 97,
4386. (b) Luque, F. J.; Bachs, M.; Orozco, M. J. Comput. Chem. 1994, 15,
847. (c) Orozco, M.; Bachs, M.; Luque, F. J. J. Comput. Chem. 1995, 16,
564-575. (d) Miertus, S.; Scrocco, E.; Tomasi, J. Chem. Phys. 1981, 55,
117. (e) Miertus, J.; Tomasi, J. Chem. Phys. 1982, 65, 239.
(28) AM1/MST calculations show that hydration disfavors AT and A
F pairings by around 8.1 and 3.6 kcal/mol. This implies that solvation favors
relative binding of F to A, with respect to the binding of T to A, by around
4.5 kcal/mol, thus reducing the apparent large difference in stability
suggested from H-bonding considerations alone.
(29) Marky, L. A.; Breslauer, K. J. Biopolymers 1987, 26, 1601.

Oligonucleotides Containing Difluorotoluene

J. Am. Chem. Soc., Vol. 122, No. 29, 2000 6897

Figure 7. Molecular interaction potential (MIP) and solvation maps obtained for the triplexes containing T or F. The MIP maps are contoured at
-6.0 kcal/mol, and the solvation maps are contoured at a water density of around 3.5 g/cm3

energy -0.6 kcal/mol). This finding can be explained considering that the characteristics of the major groove of the duplex
containing the d(AF) pair are very similar to those of the
reference duplex (see above and Table 2). The similarity of
triplex-forming characteristics of the parent and TfF mutated
duplex strongly suggests that a duplex DNA containing a d(A
F) pair would have, a priori, a similar ability to interact with
most major groove-binding DNA proteins as the parent
duplex.
Perhaps more interesting is the study of the effect of the TfF
mutation on the thermodynamics of the whole assembly process,
from single-stranded species to triplex. As seen in Figure 8 and
Table 5, the presence of one F destabilizes the resultant triplex
with respect to the reference triplex (d(ATT)) by around 3.2
kcal/mol when F is in the Hoogsteen position, and 4.7 kcal/

mol when it is in the Watson-Crick position. Therefore, a


triplex containing a d(AT-F) trio will be around 1.5 kcal/mol
more stable than that containing a d(AF-T) trio. Energy
analysis of the MD-averaged structures suggests (Table 7) that
the largest difference in stability due to the presence of T instead
of F lies in the H-bonding term. The TfF mutation in the
Watson-Crick position adversely affects H-bonding to a greater
extent than the same mutation in the Hoogsteen position, which
may help to explain the higher stability of the d(AT-F) triplex
compared to the d(AF-T) triplex found in our MD/TI
calculations. The better hydrogen bonding of the d(AT-F)
triplex (unexpected from optimizations in the gas phase30) can
be mostly attributed to backbone effects which make possible
a better interaction geometry for the Hoogsteen A-F than for
the Watson-Crick AF pairing.

6898 J. Am. Chem. Soc., Vol. 122, No. 29, 2000

Cubero et al.

Figure 8. Free energy profiles obtained from different trajectories. Average values (in kcal/mol) are shown with their standard errors. See also
Figure 2.
Table 5. Changes in the Free Energy of the DuplexfTriplex,
SinglefDuplex, and SinglefTriplex Process Associated with the
TfF Mutationa
G
SE
(kcal/mol) (kcal/mol)

folding
process

sequence

duplexf triplex
duplexf triplex
duplexf triplex
singlef duplex
singlef duplex
singlef triplex
singlef triplex
singlef triplex

d(T)+d(AT)f d(AT-T)
d(T)+d(AF)f d(AF-T)
d(F)+d(AT)f d(AT-F)
d(T)+d(A)fd(AT)
d(F)+d(A)fd(AF)
d(T)+d(A)+d(T)f d(AT-T)
d(F)+d(A)+d(T)fd(AF-T)
d(T)+d(A)+d(F)fd(AT-F)

0.0
-0.6
3.2
0.0
5.3
0.0
4.7
3.2

0.0
0.3
0.2
0.0
0.1
0.0
0.3
0.2

Standard errors for the different estimates are displayed. See Figure
8 and text for details.
Table 6. Interaction Terms (in kcal/mol) for the Central Three
Steeps (d(GpApA)) of the Duplexes Containing the d(AT) and
d(AF) Paira

interaction

d(AT)

d(AF)

stacking (intrastrand)
stacking (interstrand)
stacking (total)
hydrogen bonding
total interaction

-27.6
-4.6
-32.3
-54.5
-86.8

-24.8
-6.2
-31.0
-44.2
-75.2

Calculations are done using the MD-averaged structure.

MD-TI calculations strongly suggest that the TfF replacement in Watson-Crick or Hoogsteen positions destabilizes the
triplex DNAs by 3-5 kcal/mol with respect to the parent triplex
3. With use of the empirical approach of Roberts and Crothers,31
(30) B3LYP/6-31G(d, p) geometry optimization suggests stabilization
energies (after BSSE correction) of -3.1, and -2.9 kcal/mol for Hoogsteen
and Watson-Crick AF pairs. These values are remarkably similar to those
obtained using AMBER parameters: -3.3 (A-F, H) and -3.2 (A-F, WC)
kcal/mol.
(31) Roberts, R. W.; Crothers, D. M. Proc. Natl. Acad. Sci. U.S.A. 1996,
93, 4320.

Table 7. Interaction Terms (in kcal/mol) for the Central Three


Steps (d(GpApA)) of the Triplexes Containing the d(AT-T),
d(AF-T), and d(AT-F) Triosa
interaction

d(AT-T)

d(AF-T)

d(AT-F)

stacking (intrastrand)
stacking (interstrand)
stacking (total)
hydrogen bonding
total interaction

-40.1
-20.6
-60.7
-115.0
-175.7

-35.9
-20.1
-56.0
-96.5
-152.5

-38.7
-17.5
-56.2
-104.9
-161.1

Calculations are done using the MD-averaged structure.

the standard free energy for triplex formation at 37 C and pH


5.0 (we model all third strand cytosines as fully protonated) is
-8.85 kcal/mol. On this basis, we would expect both triplexes
4 and 5 to be stable at room temperature, as suggested by 3 ns
MD trajectories (see above). Experimental work should be done
to verify the stability of these triplexes.
Conclusions
The replacement of a single thymine base by a difluorotoluene
mimic clearly destabilizes the structures of duplex and triplex
DNAs. However, at least for the sequences considered here, all
structures are stable in MD simulations at room temperature
for at least several nanoseconds. These findings agree with
previous experimental data for F-containing DNA duplexes, but
are a prediction for F-containing triplexes, for which no
experimental data exist. These theoretical suggestions are
expected to encourage experimental effort to confirm the
stability of F-containing triplexes.
The TfF mutation does not lead to major modifications in
the structure or recognition properties of either duplex or triplex
DNA, but leads to important changes in dynamics, particularly
the emergence of breathing events on the nanosecond time
scale. These breathing movements appear for all F-modified

Oligonucleotides Containing Difluorotoluene


structures, but are more common when the TfF mutation occurs
in the Watson-Crick position than when it occurs in the
Hoogsteen position. Breathing movements might have important
implications to modulate the interactions of DNAs containing
difluorotoluene with proteins and drugs, and might be involved
in the activation of damage signals for DNA-repairing systems.
State of the art MD/TI simulations provide robust estimates
for the change in duplex/triplex stability produced by the TfF
mutation which agree well with available experimental evidence.
The calculations suggest that the TfF mutation is more
destabilizing in the Watson-Crick than in the Hoogsteen
position. This suggests that it is a reasonable strategy to consider
the development of nonstandard and apolar bases to be used in
triplex-forming oligonucleotides for antigene strategies, where

J. Am. Chem. Soc., Vol. 122, No. 29, 2000 6899


the lack of H-bonding ability might represent a clear advantage
to design stable triplexes in nonhomopurine sequences.
Acknowledgment. We are indebted to E. Sherer, Dr. G.
Shields, J. C. Morales, and Dr. E. T. Kool for valuable
suggestions and acknowledge the financial support of the
Spanish DGICYT (PB96-1005), the Catalonian Supercomputer
Center (CESCA), and the Wellcome Trust. This is a contribution
from the Centre Especial de Recerca en Qumica Teorica.
Supporting Information Available: Figure showing the
variation of the H-bond distances for triplexes 3, 4, and 5 (PDF).
This material is available free of charge via the Internet at
http://pubs.acs.org.
JA000117N

You might also like