Molecular dynamics simulation to investigate differences in minor groove hydration of HNA/RNA hybrids as compared to HNA/DNA complexes

Article abstract of DOI:10.1021/ja973721f

Hexitol nucleic acids (HNA) are oligonucleotides built up from natural nucleobases and a phosphorylated 1,5-anhydrohexitol backbone. Molecular associations between HNA and RNA are more stable than between HNA and DNA and between natural nucleic acids (dsD

Full text of DOI:10.1021/ja973721f

J. Am. Chem. Soc. 1998, 120, 5381-5394
5381
Molecular Dynamics Simulation To Investigate Differences in Minor
Groove Hydration of HNA/RNA Hybrids As Compared to HNA/
DNA Complexes
H. De Winter,† E. Lescrinier, A. Van Aerschot, and P. Herdewijn*
Contribution from the Laboratory of Medicinal Chemistry, Rega Institute, Catholic UniVersity of LeuVen,
Minderbroedersstraat 10, B-3000 LeuVen, Belgium
ReceiVed October 27, 1997
Abstract: Hexitol nucleic acids (HNA) are oligonucleotides built up from natural nucleobases and a
phosphorylated 1,5-anhydrohexitol backbone. Molecular associations between HNA and RNA are more stable
than between HNA and DNA and between natural nucleic acids (dsDNA, dsRNA, DNA/RNA). 1H NMR
analysis of a HNA dimer confirms the axial orientation of the base moiety with respect to the hexitol ring, and
this was used as starting conformation for a molecular dynamics study of HNA/RNA and HNA/DNA duplexes.
Both complexes show an A-type geometry and very similar hydrogen bonding patterns between base pairs.
The relative stability of HNA/RNA versus HNA/DNA might be explained by a difference in minor groove
solvatation.
Introduction
The formation of the helical structure is driven by stacking
interactions between the bases, allowed by conformational
changes in the phosphate backbone. This helical stabilization
of the structure is not possible with equatorially oriented bases,
as found in pyranose nucleic acids.7,8 The six-membered hexitol
ring can be considered as a mimic of a furanose ring, frozen in
its 2′-exo,3′-endo conformation.9-11 This is largely the con-
formation of the ribofuranose ring found in the A-form of
dsRNA and dsDNA. It is not surprising, therefore, that initial
CD studies demonstrate that HNA/DNA and HNA/RNA
duplexes adopt a A-type conformation,3 and this observation is
now further investigated by NMR and X-ray studies. An
intriguing observation is that the HNA/RNA duplex is more
stable than an HNA/DNA duplex.1-3 Dependent on the
sequence studied, the latter duplex is sometimes of lower
stability than a dsDNA hybrid.3 The HNA/RNA duplex,
however, is invariably more stable than all other associations
of natural nucleic acids (dsDNA, dsRNA, DNA/RNA).3 The
order of duplex stability is given by HNA/HNA > HNA/RNA
> HNA/DNA. The higher stability of duplexes with RNA as
compared with DNA duplexes is generally explained by the
propensity of the duplex to take an A-form-like structure. The
present investigation describes a molecular dynamics study of
HNA/RNA and HNA/DNA duplexes in an effort to try to
explain the difference in stability between both complexes. This
kind of study may lead to a better understanding of oligomeric
complexes and, thus, to the design of more efficient RNA
receptors.11 The study itself was carried out on a HNA sequence
(GCGTAGCG) containing all four natural bases.
Hexitol nucleic acids (HNA) represent a new oligomeric
structure able to hybridize sequence selectively as well with
DNA, RNA, and with itself.1-3 It was previously demonstrated
that HNA forms hightly selective and exceptionally stable
duplexes with RNA2,3 and that they are stable toward nuclease
degradation.2 From this point of view, HNA is superior to its
DNA and RNA analogues as an antisense construct.3 However,
the HNA/RNA duplex is a poor substrate for RNase H,3 and
its antisense activity will most probably be due to a physical
blockage of the targeted mRNA.3 Hexitol nucleic acids are built
up of phosphorylated 1,5-anhydro-D-arabino-2,3-dideoxyhexitol
building blocks with a base moiety positioned in the 2-position.
The conformational preference of the hexitol monomers are
driven by the fact that the ring oxygen atom only contains
unshared pairs (and thus a smaller substituent than a hydrogen
atom) and that steric strains should be avoided in selecting the
energetically most favorable conformation of a molecule (West-
heimer model). As a result, the base moiety of the hexitol
nucleosides are axially oriented.4-6 When these monomers are
polymerized, oligomers are obtained that form helical duplex
structures as well with DNA and RNA and with itself (HNA),
with a geometry resembling those of the Watson-Crick pairing
natural nucleic acids.2,3
† Present address: Janssen Research Foundation, Turnhoutseweg 30,
B-2340 Beerse, Belgium.
(1) Van Aerschot, A.; Verheggen, I.; Hendrix, C.; Herdewijn, P. Angew.
Chem., Int. Ed. Engl. 1995, 34, 1338-1339.
(2) Hendrix, C.; Verheggen, I.; Rosemeyer, H.; Seela, F.; Van Aerschot,
A.; Herdewijn, P. Chem. Eur. J. 1997, 3, 110-120.
(3) Hendrix, C.; Rosemeyer, H.; De Bouvere, B.; Van Aerschot, A.; Seela,
F.; Herdewijn, P. Chem. Eur. J. 1997, 3, 1513-1520.
(4) Verheggen, I.; Van Aerschot, A.; Toppet, D.; Snoeck, R.; Janssen,
G.; Balzarini, J.; De Clercq, E.; Herdewijn, P. J. Med. Chem. 1993, 36,
2033-2040.
(5) Verheggen, I.; Van Aerschot, A.; Van Meervelt, L.; Rozenski, J.;
Wiebe, L.; Snoeck, R.; Andrei, G.; Balzarini, J.; Claes, P.; De Clercq, E.;
Herdewijn, P. J. Med. Chem. 1995, 38, 826-835.
(6) Declercq, R.; Herdewijn, P.; Van Meervelt, L. Acta Crystallogr. 1996,
C52, 1213-1215.
(7) Augustyns, K.; Van Aerschot, A.; Urbanke, C.; Herdewijn, P. Bull.
Soc. Chim. Belg. 1992, 101, 119-130.
(8) Eschenmoser, A.; Dobler, M. HelV. Chim. Acta 1992, 75, 218-259.
(9) Herdewijn, P. Liebigs Ann. 1996, 1337-1348.
(10) Maurinsh, Y.; Schraml, J.; De Winter, H.; Blaton, N.; Peeters, O.;
Lescrinier, E.; Rozenski, J.; Van Aerschot, A.; De Clercq, E.; Busson, R.;
Herdewijn, P. J. Org. Chem. 1997, 62, 2861-2871.
(11) Herdewijn, P. Carbohydrate mimics: concepts and methods;
Chapleur, Y., Ed.; Verlag Chemie: Weinheim, Germany, 1998; pp 553-
579.
S0002-7863(97)03721-9 CCC: $15.00 © 1998 American Chemical Society
Published on Web 05/19/1998

5382 J. Am. Chem. Soc., Vol. 120, No. 22, 1998
Winter et al.
Scheme 1. Chemical Structure and Atom Numbering of the
Scheme 2. Schematic Drawing of the Two Hybrid
Anhydrohexitol Nucleosides Used in This Studya
Complexes Used in This Studya
a The lowercase letter at the beginning of each residue name stands
for the particular residue type: ‘h’ for anhydrohexitol, ‘d’ for
deoxyribose, and ‘r’ for ribose. The residue numbering starts at the
7′-terminal side of the HNA chain and ends at the 3′-end of the DNA
(HNA/DNA complex) or RNA strand (HNA/RNA complex). Residue
12 can be either thymidine (HNA/DNA duplex) or uridine (HNA/RNA
duplex). The black boxes linking adjacent nucleotides indicate phos-
phodiester bridges, and hydrogen bonds between opposite residues are
drawn by thin lines.
a The base moiety can be adenine, guanine, cytosine, or thymine.
Methods
Nomenclature. The chemical structure and atom naming of
anhydrohexitol nucleosides as used throughout the modeling and NMR
experiments is given in Scheme 1. They are composed of a six-
membered anhydrohexitol ring that is substituted at C1′ by one of the
four different heterocycles attached by a C1′-N linkage. The
heterocycles are the purine bases adenine (A) and guanine (G) and the
pyrimidine bases cytosine (C) and thymine (T). In anhydrohexitol
oligonucleotides (HNA), the individual anhydrohexitol residues are
linked via 3′,7′-phosphodiester bonds. Such HNA oligonucleotides can
form antiparallel duplexes with DNA or RNA oligonucleotides, and
the general structure of the two hybrid complexes used in the present
study are shown in Scheme 2.
(C-6′), 64.2 (C-4′, d, J ) 3.8 Hz), 68.4 (C-1′), 80.4 (C-5′, d, J ) 8
Hz), 86.2 (Ph3C), 110.5 (C-5), 138.5 (C-6), 151.2 (C-2), 163.9 (C-4),
113.0, 126.8, 127.4, 127.7, 130.4, 135.5, 144.3, 158.4 (aromatic C)
ppm.
1,5-Anhydro-4-O-benzoyl-2-(N6-benzoyladenin-9-yl)-2,3-dideoxy-
D-arabino-hexitol (2b). An amount of 257 mg (0.4 mmol) of 1b12
was co-evaporated with pyridine and subsequently dissolved in 20 mL
of anhydrous pyridine after which 110 µL (0.8 mmol) of benzoyl
chloride was added. The mixture was stirred overnight at room
temperature and quenched with 2 mL of methanol. After addition of
some NaHCO3, the mixture was concentrated and partitioned between
CH2Cl2 and 5% of aqueous NaHCO3. The organic phase was washed
once more with 75 mL of aqueous NaHCO3 and dried on Na2SO4.
Evaporation left an oil that was co-evaporated with toluene. The foam
was treated with 40 mL of 80% aqueous HOAc and left at RT for 1 h.
Evaporation gave 400 mg of a light brown foam that was purified on
silica gel with a methanol gradient in CH2Cl2 (100 to 96:4). The
product-containing fractions were pooled affording 86 mg (0.18 mmol,
45%) of the title product 2b as a white powder. LSIMS (ThGLy) m/z
474 (MH+, 25), 240 (BH2+, 7). 1H NMR (CDCl3): δ 2.21-2.39 (ddd,
1H, H-3′ax), 2.84-2.98 (dm, J ) 13.5 Hz, 1H, H-3”eq), 3.64-3.91 (m,
3H, H-5′, H-6′, H-6”), 4.13 (dd, 1H, J ) 2.6 and 13.1 Hz, H-1′ax),
4.52 (dm, 1H, J ) 12.5 Hz, H-1′eq), 5.07-5.14 (m, 1H, H-2′), 5.18-
5.33 (m, 1H, H-4′), 7.35-7.65 (m, 6H, aromatic H), 7.91-8.08 (m,
4H, aromatic H), 8.70 (s, 1H) and 8.80 (s, 1H) (H-2, H-8), 9.63 (br s,
1H, NH) ppm. 13C NMR (CDCl3): δ 33.1 (C-3′), 50.7 (C-2′), 61.4
(C-6′), 64.4 (C-4′), 69.1 (C-1′), 80.5 (C-5′), 123.2 (C-5), 142.0 (C-8),
149.6 (C-4), 151.7 (C-6), 152.5 (C-2), 164.8, 165.4 (2 × CO) ppm +
aromatic signals.
Synthesis
1,5-Anhydro-6-O-monomethoxytrityl-2-(thymin-1-yl)-2,3-dideoxy-
D-arabino-hexitol-4-yl-hydrogenphosphonate, Triethylammonium
Salt (3a). To 270 µL (3 mmol) of phosphorus trichloride in 30 mL of
anhydrous CH2Cl2 cooled on an icebath was added 3.3 mL (30 mmol)
of N-methylmorpholine and 690 mg (10 mmol) of 1,2,4-triazole. The
mixture was stirred for 30 min at room temperature and subsequently
cooled again, and 320 mg (0.6 mmol) of 1,5-anhydro-6-O-monomethoxy-
trityl-2-(thymin-1-yl)-2,3-dideoxy-D-arabino-hexitol (1a)12 dissolved in
10 mL of CH2Cl2 was added dropwise over 15 min. The mixture was
stirred for 10 min more at room temperature and poured into 25 mL of
1 M triethylammonium bicarbonate (TEAB) buffer. After separation
of both layers, the aqueous phase was extracted once more with CH2Cl2,
and the organics were dried and purified by flash chromatography on
20 g of silica gel (gradient of CH2Cl2-TEA 99:1 to CH2Cl2-TEA-
MeOH 89:1:10). Product-containing fractions were washed once with
25 mL of 1 M TEAB buffer and dried, affording 344 mg (0.495 mmol,
82%) of the title product 3a as a white foam. LSIMS (ThGly doped
with NaOAc) m/z 659 (M + 3Na+, 1), 637 (M + 2Na+, 2), 273
(MMTr+, 100). 1H NMR (CDCl3): δ 1.27 (t, 9H, J ) 7 Hz, CH2CH3),
1.74 (s, 3H, CH3), 1.94-2.11 (m, 1H, H-3′ax), 2.61 (dm, 1H, H-3′′eq),
3.00 (q, 6H, CH2CH3), 3.34 (dd, 1H, J ) 4.5 and 10.4 Hz, H-5′), 3.45-
3.60 (m, 2H, H-6′, H-6”), 3.79 (s, 3H, OCH3), 3.89 (dd, 1H, J ) 3.4
and 13.4 Hz, H-1′ax), 4.16 (d, 1H, J ) 13.6 Hz, H-1′eq), 4.57 (m, 1H,
H-4′), 4.73 (m, 1H, H-2′), 6.66 (d, 1H, J ) 625 Hz, P-H), 6.83 (d, J
) 8.7, 2H, aromatic H), 7.15-7.55 (m, 12H, aromatic H), 8.05 (s, 1H,
H-6), 9.50 (br s, 1H, NH) ppm. 13C NMR (CDCl3): δ 8.5 (CH2CH3),
12.7 (CH3), 35.2 (C-3′), 45.3 (CH2CH3), 50.4 (C-2′), 55.1 (OCH3), 62.7
[1,5-Anhydro-2-(thymin-1-yl)-2,3-dideoxy-D-arabino-hexitol]-(4-
6)-[1,5-anhydro-2-(adenin-9-yl)-2,3-dideoxy-D-arabino-hexitol]phos-
phate, Sodium Salt (5). A portion of 86 mg (0.18 mmol) of 2b and
145 mg (0.21 mmol) of the thymine hydrogenphosphonate 3a were
co-evaporated twice with anhydrous pyridine and dissolved in a 6 mL
mixture of anhydrous pyridine and acetonitrile. One milliliter (0.5
mmol) of a dilution of pivaloyl chloride in pyridine was added, and
the reaction mixture was stirred for 10 min at room temperature under
a nitrogen atmosphere. Subsequently 2 mL of a 4% I2 (w/v) solution
in pyridine-water (96:4) was added, and stirring continued for another
10 min at room temperature. The mixture was diluted with 100 mL
(12) De Bouvere, B.; Kerremans, L.; Rozenski, J.; Janssen, G.; Van
Aerschot, A.; Claes, P.; Busson, R.; Herdewijn, P. Liebigs Ann. 1997, 1453-
1461.

Stability of HNA/RNA Hybrids
J. Am. Chem. Soc., Vol. 120, No. 22, 1998 5383
Scheme 3. Synthesis of Hexitol Nucleic Acids TA Dimer Using Solution Phase H-Phosphonate Chemistrya
a a, B* ) thymin-1-yl. b, B* ) N6-benzoyladen-9-yl. iii, benzoyl chloride, pyridine; 80% HOAc. iv, PCl3, N-methylmorpholine, triazole, TEAB
buffer. v, 3a (1.15 equiv), 2b pivaloyl chloride, pyridine-acetonitrile; iodine, TEAB buffer. vi, ammonia-ethanol; 80% HOAc; RP-HPLC; Dowex
50X8-200 (Na+ form).
of CH2Cl2 and washed with 50 mL of 10% aqueous Na2S2O3 and with
50 mL of 1 M TEAB. The organic phase was dried, evaporated, and
purified by column chromatography on 20 g of silica gel (gradient of
CH2Cl2-TEA 99:1 to CH2Cl2-TEA-MeOH 93:1:6) affording 200 mg
(0.17 mmol, 95%) of the fully protected dimer. LSIMS (ThGly doped
with NaOAc) m/z 1108 (M + 2Na+, 2) and 273 (MMTr+, 100). The
white foam obtained was dissolved in 20 mL of a 3:1 mixture of
concentrated ammonia and ethanol and heated overnight at 40 °C. The
mixture was evaporated and co-evaporated with 1,4-dioxane. The
residue was dissolved in 20 mL of 80% aqueous acetic acid and heated
at 60 °C for 1 h. After evaporation, the residue was partitioned between
10 mL of 0.1 M triethylammonium acetate (TEAA) solution and 10
mL of ether. The aqueous phase was washed 3 times more with 10
mL of diethyl ether and concentrated and purified by reverse phase
HPLC on a polystyrene(divinyl benzene) support (PLRP-S, 250 × 9
mm) with an acetonitrile gradient in 0.1 M TEAA. Product-containing
fractions were pooled, and the TEA salt was exchanged for the sodium
salt by chromatography on a Dowex 50 × 8-200 cation-exchange resin
under its sodium form. Lyophylization afforded 52 mg (85 µmol, 47%
overall) as a white voluminous powder. LSIMS (ThGly) m/z 628 (M
sodium salt + Na+, 30), 606 (M sodium salt + H+, 4).
NMR Analysis of a HNA TA-Dimer. A 7.5 mM solution of h(TA)
was prepared in a 10 mM phosphate buffer (pH 7.0). The sample was
lyophylized and redissolved twice in 99.996% D2O. A Varian 500
Unity spectrometer with gradient indirect detection probe, operating
at 499.6 MHz (1 H), and 33 °C was used. Data were processed on a
SunSparc workstation with the standard VNMR software (version 5.1)
and on a Silicon Graphics Indigo2 R10000 workstation (IRIX version
6.2) using the Felix95.0 software package (Biosym Technologies, San
Diego). All proton spectra are internal referenced to the water signal
at 4.64 ppm. The 31P signal is external referenced to H3PO4 85%.
Gradient double quantum filtered phase sensitive cosy 1H-1H correla-
tion spectra (GDQFCOPS13) consisted of 2048 pairs of real and
imaginary points in t1 and 4096 data points in t2. To obtain maximum
resolution, a reduced sweep width of 2000 Hz was used, covering only
the anhydrohexitol signals. The data were apodized with a skewed
sinebel squared function in both dimensions. The 31P was decoupled
on resonance and in the continuous mode. The NOESY14 spectrum
(mix ) 300 ms) was performed with 512 pairs of real and imaginary
points in t1 and 2048 data points in t2. To cover the full spectrum, a
spectral width of 4800 Hz was used. Data were zero-filled in t1 to
2048 and apodized with a skewed sinebel squared function in both
dimensions.
were calculated using the RESP methodology,16 involving a two-stage
fit of the charges to the 6-31G*-derived electrostatic potential calculated
using GAMESS.17 The partial atomic charges of the HNA residues
are shown in Scheme 4. The appropriate atom types for the HNA atoms
were taken from the AMBER 4.1 parameter database and shown in
Scheme 4. Force field parameters were also taken from AMBER 4.1
and used without modification. In the absence of experimental data
regarding the relative stabilities and energy barriers between the
different puckering forms of the anhydrohexitol rings, we could not
thoroughly evaluate the assumption that the default AMBER param-
etrization performs properly for anhydrohexitol nucleosides as well.
However, several test MD runs of more than 10 ns each on isolated
anhydrohexitol nucleosides immersed in water boxes compared well
with the average conformations derived from NMR experiments (data
not shown). During these test calculations, the bases remained in axial
orientation with respect to their anhydrohexitol moieties, and good
agreement with the NMR conformations was found for the O7′-C7′-
C4′-C3′ (γ) and C6′-C1′-N1-C2 (ø) torsion angles (data not shown).
Model Building. (i) HNA/RNA complex: A model of d(GCG-
TAGCG)/r(CGCUACGC) in Arnott’s canonical A-RNA geometry was
used as the starting point. Using MidasPlus,18 the p(GCGTAGCG)
strand was subsequently docked onto the DNA strand of the DNA/
RNA model.19 The entire HNA strand was then energy minimized
while its base atoms were template forced onto the corresponding
positions of the DNA strand using a force constant of 10 kcal/mol‚Å2
on each base atom. Finally, the DNA strand was deleted, and the
resulting HNA/RNA complex was energy minimized down to an rms
energy gradient of 0.1 kcal/mol‚Å by imposing a parabolic restraint
on the base pair hydrogen bonds between 2.5 and 4.0 Å using a 10
kcal/mol‚Å2 force constant and a linear restraint at larger distances.
This was followed by unrestrained minimization until the rms gradient
dropped below 0.1 kcal/mol‚Å. Care was taken to model each
(15) Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz, K.
M.; Ferguson, D. M.; Spellmeyer, D. C.; Fox, T.; Caldwell, J. W.; Kollman,
P. A. J. Am. Chem. Soc. 1995, 117, 5179-5197.
(16) Bayly, C. I.; Cieplak, P.; Cornell, W. D.; Kollman, P. A. J. Phys.
Chem. 1993, 97, 10269-10280.
(17) Schmidt, M. W.; Baldridge, K. K.; Boatz, J. A.; Elbert, S. T.;
Gordon, G. S.; Jensen, J. H.; Koseki, S.; Matsunaga, N.; Nguyen, K. A.;
Su, S.; Windus, T. L.; Dupuis, M.; Montgomery, J. A. J. Comput. Chem.
1993, 14, 1347-1363.
(18) Ferrin, T. E.; Huang, C. C.; Jarvis, L. E.; Langridge, R. J. Mol.
Graphics 1988, 6, 13-27.
(19) We also experimented to model the HNA/RNA duplex in a B-like
geometry but were unsuccessful to obtain starting structures with reasonable
backbone torsion angles along the HNA strand. Comparison of the
conformation of the HNA six-membered sugar ring with each of the two
possible low-energy conformations of a five-membered deoxyribose sugar
ring (C2′- and C3′-endo) indicates that the HNA sugar resembles the C3′-
endo conformation much better than the C2′-endo puckered form (data not
shown). In fact, after superimposition of the corresponding nucleoside base
atoms, a rms deviation between the sugar-phosphate backbone atoms (O5′-
C5′-C4′-C3′-O3′ for the deoxyribose sugar versus O7′-C7′-C4′-C3′-
O3′ for the HNA sugar) of <1 Å was calculated when the deoxyribose
was modeled as C3′-endo and >3 Å for the deoxyribose in a C2′-endo
conformation.
Computational Chemistry. Force Field Parameters and Partial
Charges. (i) RNA and DNA: Parameters and also the partial charges
for the nucleoside bases, sugars, and phosphate moieties of the RNA
and DNA strands were taken from the AMBER 4.1 force field parameter
database.15 (ii) HNA: Partial atomic charges of the HNA strand atoms
(13) Davis, A. L.; Laue, E. D.; Keeler, J.; Moskou, D.; Loohman, J. J.
Magn. Reson. 1991, 94, 637-644.
(14) States, D. J.; Haberkorn, R. A.; Ruben, D. J. J. Magn. Reson. 1982,
48, 286-292.

5384 J. Am. Chem. Soc., Vol. 120, No. 22, 1998
Winter et al.
Scheme 4. Force Field Atom Types and Partial Charges of the Four Different Anhydrohexitol Nucleotides Used in This Study
anhydrohexitol ring in a chair-like conformation with the base moieties
oriented axially, but during minimization it was found that the hG1
anhydrohexitol ring had flipped from a chair- to a twistboat-like
conformation, while keeping the guanine base in an axial orientation.
The resulting duplex was soaked in a rectangular box of TIP3P
waters that extended approximately 12 Å in each direction from the
nucleic acid atoms. This seems to be adequate to reproduce the global
structure and dynamics in solution.20 Fourteen water molecules whose
oxygens were located at positions of highest negative electrostatic
potential were replaced by sodium ions. This yielded a total of 3752
waters. (ii) The HNA/DNA complex was prepared from the minimized
coordinates of the HNA/RNA system after deletion fo the 2′-hydroxyl
groups of the ribose sugars and proper adjustment of partial charges
and atom types. After minimization, the whole system was submerged
in a rectangular box of waters. Sodium counterions were added in a
similar fashion as for the HNA/RNA complex. In total, 3781 TIP3P
waters were included in the HNA/DNA simulation. The initial box
size was ∼58 by ∼50 by ∼47 Å3 for both simulations.
runs were initiated. The simulations were continued for 1.1 ns each,
and conformations were saved every 50 fs. Final box dimensions were
∼56 by ∼48 by ∼45 Å3 for both simulations. Only the final 1 ns
were used for analysis, except where noted otherwise.
Spatially Resolved Solvent Densities. Spatially resolved solvent
densities were calculated on the final 0.5 ns of each trajectory.
Rotations and translations of the complexes within their boxes were
removed from the trajectories by using the average structure of each
double helix to establish a solute-fixed coordinate frame. Solvent
densities were calculated by determining water oxygen positions from
rms coordinate fit frames in 0.5 × 0.5 × 0.5 Å3 grid volumes. The
value of each grid volume represents the number of times a water
oxygen atom was found within the 0.125 Å3 of that particular grid
volume. For 10 000 frames (the final 500 ps of each simulation), the
expected number of water oxygens per grid volume, assuming bulk
water density, is 41.6. The value at each grid volume was then
converted to a density relative to the expected bulk water density. As
an example, 83.2 and 124.8 hits per 0.125 Å3 volume element are equal
to 2 and 3 times the expected bulk water density. These grid values
were then contoured in a manner analogous to the contouring of electron
density in crystallographic diffraction studies. Results were visualized
using Quanta (Biosym/MSI, San Diego).
Molecular Dynamics. Periodic boundary conditions were applied
using constant temperature (300 K) and constant pressure (1 bar)
conditions. Berendsen temperature coupling was used with separate
but equal heat bath coupling time constants of 0.4 ps-1 for solute and
solvent atoms.21 The compressibility of the systems was set to 44.6 ×
Solvent Accessible Surface Areas. The solvent accessible surface
10-6 bar-1, and the pressure relaxation time was defined at 0.4 ps-1 21
.
area (SASA) of the double helices were calculated at each dynamics
Simulations were run with SHAKE on all bond lengths,22 a 2 fs time
step, a 9 Å cutoff for Lennard-Jones interactions, and the nonbonded
pairlist updated every 10 steps. Calculations were performed with the
SANDER module of AMBER 4.1.15 The results were analyzed with
the CARNAL module. No restraints were applied during the production
runs. The particle mesh Ewald (PME) method was used for the
treatment of long-range electrostatic interactions.23 The PME charge
grid spacing for both simulations was approximately 1.0 Å in each
direction. The order of the â-spline interpolation was 4, and the direct
sum tolerance was set to 10-6. The systems were minimized until the
rms energy gradient dropped below 0.1 kcal/mol‚Å. Subsequently, the
systems were heated to 300 K over 10 ps, after which time production
frame using the ‘dms’ module of the MidasPlus suite of programs.18
A
solvent probe radius of 1.4 Å and van der Waals radii of 1.9, 1.4, 1.5,
and 1.9 Å for C, O, N, and P, respectively, were used.
Results
Chemical Synthesis of the HNA TA Dimer. The chemical
synthesis of the protected hexitol nucleoside monomers has been
described before.12 Assembly of the monomers into the h(TA)
dimer was done according to standard H-phosphonate chemistry
and was carried out in solution as described in Scheme 3. The
tritylated anhydrohexitol nucleoside analogue 1a was converted
to the triethylammonium hydrogen phosphonate salt 3a in 82%
yield using phosphorus trichloride.24 The adenine counterpart
1b was reacted with benzoyl chloride, and the obtained crude
was treated with 80% aqueous acetic acid for 1 h at RT. Silica
(20) Norberto de Souza, O.; Ornstein, R. L. Biophys. J. 1997, 72, 2395-
2397.
(21) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; DiNola,
A.; Haak, J. R. J. Chem. Phys. 1984, 81, 3684-3690.
(22) van Gunsteren, W. F.; Berendsen, H. J. C. Mol. Phys. 1977, 34,
1311-1327.
(23) Essmann, U.; Perera, L.; Berkowitz, M. L.; Darden, T.; Lee, H.;
Pedersen, L. G. J. Chem. Phys. 1995, 103, 8577-8593.
(24) Froehler, B. C.; Ng, P. G.; Matteucci, M. D. Nucleic Aicds Res.
1986, 14, 5399-5407.

Stability of HNA/RNA Hybrids
J. Am. Chem. Soc., Vol. 120, No. 22, 1998 5385
Figure 2. Expansion of the NOESY spectrum showing the interactions
of the hT1:H6 and hA2:H8 protons with the anhydrohexitol moieties.
Observed intraresidue NOE cross-peaks correspond to an anhydrohexitol
conformation depicted in Scheme 5. The interresidue cross-peak of hA2:
H8 f hT1:H21′ is marked in gray.
Scheme 5. Idealized Conformation of the HA Unitya
Figure 1. Overview of the non-31P decoupled [1H,1H]GDQFCOPS
spectrum. The hT1 signals are connected with dotted lines; the hA2
signals are connected with dashed lines. The long-range H61′-H21′
cross-peaks are marked in boxes.
Table 1. Measured in D2O at 33 °C, Internal Referenced to the
Water Signal (4.64 ppm)a
Section A:
Section B:
List of Chemical Shiftsa
Coupling Constants in Hzb
a Observed intraresidue NOE interactions which confirm the chair
conformation of the anhydrohexitol unit are indicated by arrows.
chemical shifts
JH-H
hT1
hA2
1b
hT1
hA2
constants listed in Table 1, SectionB. The large coupling
constants JH22′-H3′ and JH3′-H4′ (8-11 Hz) indicate an axial
position of the involved protons where the dihedral angles are
close to 180°, while the small couplings in the pattern of the
H1′ (2-3 Hz) suggest an equatorial position of the involved
proton. These data correspond with an anhydrohexitol in a chair
conformation with the base substituent in an axial position.4-6
The “W” conformation of the atoms H61′-C6′-C1′-C2′-H21′
in this chair is confirmed by the long-range coupling JH61′-H21′
in both hexitol rings (Figure 1). This result is according to what
is described on the individual hA monomer.4 A NOESY
spectrum of h(TA) showed cross-peaks of hT1:H6 and hA2:
H8 base protons with H3′, H21′ (only in hT1), and H61′ of
their anhydrohexitols respectively (Figure 2). Between H4′,
H22′ and H61′ were also neocontacts observed (data not shown).
The occurrence of these NOE interactions is in agreement with
the assignment of the anhydrohexitol conformation, which was
based on values of coupling constants (Scheme 5). This NOE
cross-peak pattern also indicates that the bases are in an anti
position (ø ) -120 ( 60°) where the H8/H6 of adenine and
thymine, respectively, are located above the anhydrohexitol ring.
These data support the selection of the starting conformation
(i.e., HNA with axially positioned base moiety) for the modeling
experiment.
Geometric Features and Flexibility of the HNA/RNA and
HNA/DNA Helices. Rms Deviation. A stereo figure of the
average structures of HNA/DNA and HNA/RNA is shown in
Figure 3. Over the course of the entire simulation, the HNA/
RNA complex remains close to its starting A-DNA-like
geometry with an average rms deviation 1.4 Å from the start
conformation (Figure 4a). The average minor groove width
calculated for this complex, 13 Å, is closer to the 11 Å of A-type
DNA than it is to the 6 Å of B-DNA (Figure 4b). In the case
of the HNA/DNA duplex, however, the structure remains close
to an A-DNA form for the first 300 ps but then deviates on the
average more than 2.0 Å from the A-DNA-like starting
H1′
4.52
2.48
2.09
4.20
3.51
4.13
3.92
3.86
3.78
7.74
4.88
2.36
2.02
3.82
3.54
4.26
4.01
4.13
4.04
8.32
8.18
1′-21′
1′-22′
1′-61′
1′-62′
21′-61′
21′-22′
21′-3′
22′-3′
3′-4′
4′-71′
4′-72′
61′-62′
71′-72′
2.5
2.5
2.0
2.9
2.0
13.7
5.0
10.7
8.8
2.9
5.9
2.6
2.9
2.0
2.0
2.0
13.7
4.9
11.7
9.8
2.9
1.9
H21′
H22′
H3′
H4′
H61′
H62′
H71′
H72′
H6/H8
H2
CH3
P
1.77
0.88
13.7
12.7
12.7
11.7
a The phosphorus signal is referenced to H3PO4 (85%). b Estimated
in a 31P-decoupled GDQFCOPS spectrum, using first-order analysis.
gel purification afforded 45% of the 4′-protected building block
2b. Under activation by pivaloyl chloride 3a (1.15 equiv) was
reacted with 2b for 10 min at RT, followed by the addition of
an aqueous pyridine solution of iodine and further stirring for
10 min at RT. The described phosphate diester 4 was isolated
as the triethylammonium salt by silica gel chromatography in
95% yield. Following deprotection (28% aqueous ammonia-
EtOH 3:1, 15 h, 40 °C; 80% aqueous HOAc, 1 h, 60 °C), the
dimer 5 was purified by RP HPLC and converted to its sodium
salt in 47% overall yield.
NMR Analysis Confirms the Axial Orientation of Base
Moieties in an h(TA) Dimer. The conformation of the
individual hexitol nucleosides in a dimer was studied on an TA
sequence. The resonance assignment in h(TA) was done with
a [1H,1H]GDQFCOPS spectrum (Figure 1). The JH-P in the
pattern of the signals hT1:3′, hT1:21′ (long range), hA2:4′ (long
range), hA2:h71′, and hA2:72′ arise from the 3′ to 7′ fosfodiester
linkage between hT1 and hA2.25
GDQFCOPS spectrum was used to determine the coupling
A
31P decoupled [1H,1H]-
(25) Blommers, M. J. J.; Nanz, D.; Zerbe, O. J. Biomol. NMR 1994, 4,
595-601.

5386 J. Am. Chem. Soc., Vol. 120, No. 22, 1998
Winter et al.
in the HNA/RNA helix and 26.0(43.2) and 133.9(99.1)° for the
HNA/RNA double helix. These average values are indicative
of a chair conformation, although significantly flattened and
tending toward a boat conformation. However, if the terminal
anhydrohexitol residues (hG1 and hG8) are excluded from the
calculations, then the average value of the θ and æ phase angles
are 10.3(5.4) (θ) and 142.0(102.5)° (æ) and 10.6(5.4) (θ) and
134.8(98.8)° (æ) for the HNA/RNA and HNA/DNA double
helices, respectively. These values are typical for a slightly
flattened chair conformation, with the large standard deviations
of the æ angles being indicative of small deviations from the
chair conformation toward boat- and twist boat-like conforma-
tions with almost equal probability of occurrence.
Figure 5 shows the time course and polarplots of the θ and
æ pucker phase angles of residues hG1 and hT4 in the HNA/
RNA and HNA/DNA double helices. The phase angles of hG1
in the HNA/RNA and HNA/DNA duplexes are shown in Figure
5, panels a and b, respectively, while hT4 is represented in
Figure 5, panels c (HNA/RNA) and d (HNA/DNA). The
puckering phase angles of residues hC2-hG8 are very similar
to the values calculated for residue hT4; hence, hT4 was taken
as a representative in this figure for the rest of the residues.
From the data in Figure 5, it is clear that the anhydrohexitol
ring of hT4 (and thus also residues hC2-hG8) in both the HNA/
RNA and HNA/DNA duplexes remains close to a chair-like
conformation (θ ∼ 0°; æ ∼ indefinite) throughout the simula-
tion, with the base moiety oriented axially onto the anhydro-
hexitol ring. This is not entirely surprising as this conformation
was taken as the starting conformation for residues hC2-hG8.
On the other hand, in both the HNA/RNA and HNA/DNA
duplexes, the anhydrohexitol ring of residue hG1 was initially
placed in a boat-like conformation, and this was found to have
significant consequences on the puckering behavior of this
residue during the two MD simulations. In the HNA/RNA
simulation (Figure 5a), hG1 undergoes already after a few
picoseconds a quick transition from boat (θ ∼ 90°; æ ∼ 360°)
to inverted chair, in which the guanine base gets reoriented from
axial to equatorial (θ ∼ 180°; æ ∼ indefinite). Also, a similar
transition takes place in the HNA/DNA simulation (Figure 5b),
but here the initial boat conformation (θ ∼ 90°; æ ∼ 0°) remains
for ∼0.6 ns before converting to inverted chair.
Figure 3. Stereo figure of the average structure of HNA/DNA (a) and
HNA/RNA (b).
Ribose and Deoxyribose Puckering. The ribose sugars of
the RNA strand in the HNA/RNA duplex remain close to a
C3′-endo conformation (P ∼ 18°), which is typical for the
canonical A-family, while the deoxyribose sugar puckering of
the HNA/DNA duplex interchanges between C2′-endo (P ∼
162°) and C3′-endo, giving an average puckering phase angle
of ∼90° (Table 2). The time course and polarplots of the
individual sugar pucker phase angles of the DNA and RNA
strands in both duplexes are shown in Figure 6. The RNA strand
in the HNA/RNA heteroduplex is represented in Figure 6a for
each nucleotide going from the 5′- (rC9) to the 3′-end (rC16).
The corresponding DNA strand in the HNA/DNA complex is
shown alongside in Figure 6b. From the data in Figure 6a, it
is obvious that the RNA strand is clearly not repuckering on a
nanosecond time frame, except for the few repuckering events
of the terminal rC16. However, it is likely that this latter event
represents an artifact of the simulation, as rC16 is base-paired
to hG1, which is repuckering from a boat to an inverted chair
conformation after a few picoseconds in the simulation (see
above). As this repuckering reorients the guanine base from
an axial to an equatorial orientation, it is therefore likely that
the observed repuckering of the rC16 ribose sugar is a direct
consequence of its interaction with hG1.
Figure 4. (a) Variation of the rms deviation from the start conformation
as a function of time for the HNA/RNA complex (left) and HNA/DNA
complex (right). (b) Variation of the minor groove width as a function
of time for the HNA/RNA complex (left) and HNA/DNA complex
(right). Minor groove widths were calculated by taking the perpendicular
separation of helix strands drawn through phosphate groups, diminished
by 5.8 Å to account for van der Waals radii of phosphate groups.
conformation during the second half of the simulation. This
increase in rms deviation is accompanied by a decrease in minor
groove width from 13 to less than 12 Å (Figure 4b).
Anhydrohexitol Puckering. Puckering parameters and
standard deviations calculated for the six-membered anhydro-
hexitol rings (HNA strands) in the HNA/RNA and HNA/DNA
double helices are given at the end of Table 2. The average
value and standard deviation of the phase angles θ and æ,
calculated over all of the eight anhydrohexitol residues, is
29.5(51.3) and 155.0(105.6)° for the anhydrohexitol residues

Stability of HNA/RNA Hybrids
Table 2. Torsion Angles and Puckering Parameters with Standard Deviations in Parentheses Averaged over All the Residuesa
HNA/RNA duplex HNA/DNA duplex
J. Am. Chem. Soc., Vol. 120, No. 22, 1998 5387
HNA strand
RNA strand
HNA strand
DNA strand
A-RNAb
A-DNAb
B-DNAb
314
R (deg)c
â (deg)c
γ (deg)c
δ (deg)c
ꢀ (deg)c
266.6 (54.4)
281.9 (9.1)
183.0 (54.5)
166.7 (8.6)
83.4 (43.8)
80.8 (38.5)
81.0 (27.9)
71.0 (7.1)
219.9 (23.9)
211.5 (11.0)
276.1 (42.2)
243.3 (22.0)
235.4 (13.6)
0.520 (0.041)
0.517 (0.041)
29.5 (51.3)
10.3 (5.4)
280.1 (10.8)
280.2 (10.2)
176.0 (10.5)
174.7 (10.2)
65.5 (10.4)
65.9 (9.5)
255.1 (71.5)
283.1 (8.9)
173.9 (23.8)
164.5 (8.4)
88.9 (48.6)
84.5 (43.2)
81.8 (26.2)
72.6 (6.6)
215.3 (22.6)
208.6 (10.9)
278.5 31.5)
241.1 (18.5)
237.0 (14.0)
0.525 (0.055)
0.513 (0.040)
26.0 (43.2)
10.6 (5.4)
283.7 (11.0)
284.2 (10.4)
175.3 (10.5)
173.7 (9.7)
67.7 (32.0)
60.4 (10.6)
107.0 (23.8)
102.9 (22.7)
191.6 (10.1)
191.8 (9.8)
277.6 (13.0)
229.2 (20.7)
229.6 (22.1)
0.371 (0.061)
0.373 (0.059)
292
310
178
54
172
41
213
36
83.8 (18.5)
78.4 (6.5)
82
79
157
155
202.6 (11.2)
203.3 (10.6)
289.4 (9.1)
210.7 (13.7)
208.6 (13.1)
0.381 (0.058)
0.387 (0.054)
207
214
ú (deg)c
ø (deg)c
289
202
282
206
264
262
pucker amplitude (Å)d
pucker phase angle θ (deg)d
pucker phase angle g (deg)d
pucker phase angle P (deg)e
155.0 (105.6)
142.0 (102.5)
133.9 (99.1)
134.8 (98.8)
31.1 (43.0)
17.3 (11.5)
97.5 (55.5)
90.7 (53.8)
C3′-endo
C3′-endo
C2′-endo
a First value in each entry or with the Four Terminal Residues (hG1, hG8, d/rC9, d/rC16) excluded from the calculations (second value in each
entry). The various duplex structures are specified. The average values were calculated by determining the values for each dynamic frame during
the final last nanoseconds of the simulation followed by averaging. The values in parentheses represent the standard deviation of this average.
bData for these duplexes were taken from Saenger.30 cTorsion angles defined by O3′-P-O7′-C7′ (R), P-O7′-C7′-C4′ (â), O7′-C7′-C4′-C3′
d
(γ), C7′-C4′-C3′-O3′ (δ), C4′-C3′-O3′-P (ꢀ), C3′-O3′-P-O7′ (ú), C6′-C1′-N1-C2 (ø). Calculated according to Cremer and Pople.32
eCalculated according to Altona and Sundaralingam.34
On the other hand, from the data in Figure 6b, it is clear that
the DNA strand is repuckering between C2′-endo and C3′-endo
throughout the entire simulation. The time course of the
transition from C3′-endo to C2′-endo pucker is also evident from
this figure. All of the DNA sugars have repuckered within
<300 ps.
complexes remain strongly hydrogen bonded throughout the
entire simulation, and the individual hydrogen bond distances
and standard deviations for HNA/DNA and HNA/RNA are very
similar. The only significant differences are found along the
termini of the helices, where larger standard deviations are
observed in the case of the HNA/DNA complex as compared
to HNA/RNA.
Backbone Torsion Angles. The data in Table 2 show that
the HNA/RNA and HNA/DNA helices are characteristic of the
canonical A-family of oligonucleotide structures. The γ torsion
angles of the HNA strands in the two hybrid duplexes show
large standard deviations, and this is caused by the large
deviations of this angle in the residues along the 6′-end of the
HNA strands, i.e., hG1 and hC2. While the average value of γ
of residues hG3-hG8 centers around ∼65°, the average value
of γ at hC2 is ∼170° in both duplexes. In the case of hG1, a
transition in γ (going from ∼60° to ∼180°) is observed in both
complexes after ∼0.6 ns (data not shown). In the HNA/DNA
simulation, this transition is accompanied by a simultaneous
repuckering of the hG1 anhydrohexitol ring from boat to
inverted chair (see above and also Figure 5b), while in the case
of the HNA/RNA simulation, this transition is accompanied by
a relaxation of the æ puckering phase angle (Figure 5a).
Interestingly, the RNA strand of the HNA/RNA duplex is
less flexible than the DNA strand of the HNA/DNA duplex, as
judged by looking at the standard deviations in backbone angles
presented in Table 2. Especially δ, ú, and ø show enhanced
fluctuation in the DNA strand as compared to the corresponding
values in the RNA chain. This is in agreement with the
increased flexibility of the deoxyribose sugar ring of the DNA
strand as compared to the more rigid ribose puckering observed
for the RNA chain (Table 2 and Figure 6).
Solvation of the HNA/RNA and HNA/RNA Double
Helices. Solvent accessible surface area. A decomposition
of the total solvent accessible surface area (SASA) in polar and
hydrophobic components is tabulated in Table 4. To do this,
the area formed by N, O, and P atoms was defined as polar,
whereas the area defined by C atoms was taken as being
hydrophobic. Hydrogen atoms were not included in the surface
area calculations. The total SASA is approximately identical
for the HNA/RNA and HNA/DNA complexes and fluctuates
around 2440 Å2. Nevertheless, the SASA of the HNA/RNA
complex is significantly more polar than that of the HNA/DNA
complex. This difference is mainly due to the ribose moiety
of the RNA strand whose SASA is more polar than the
corresponding area of the deoxyribose moiety of the DNA
strand. When the contributions of the 2′-OH groups to the
SASA are ignored by assuming a van der Waals radius of 0 Å
for these oxygens, the polar and hydrophobic fractions of the
total SASA become almost identical for the two complexes
(Table 4).
In Figure 7, the SASA of each of the two averaged structures
is color-coded according to the type of its underlying atom: N,
blue; O, red; P, purple; and C, gray. A view into the minor
groove of HNA/RNA and HNA/DNA is shown in Figure 7,
panels a and b, respectively, while the corresponding major
grooves are shown in Figure 7, panels c and d. Inspection of
these latter two panels shows that the major grooves of HNA/
RNA and HNA/DNA are virtual identical in terms of hydro-
phobic/hydrophilic character when comparing the relative
amount of blue, red, and gray coloring. Hence, it can be
expected that the solvation pattern of each of these two major
Base Pair Hydrogen Bonds. A summary of the hydrogen
bond lengths and their standard deviations between the base
pairs in each of the two complexes is given in Table 3. It is
clear from this table that there are no fundamental differences
in hydrogen bonding properties of the HNA/DNA and HNA/
RNA complexes. The corresponding base pairs in the two

5388 J. Am. Chem. Soc., Vol. 120, No. 22, 1998
Winter et al.
Figure 5. Sugar pucker θ and æ phase angles of residues hG1 and hT4 in the HNA/RNA and HNA/DNA double helices. The θ and æ phase angles
of hG1 in the HNA/RNA and HNA/DNA duplexes are shown in panels a and b, respectively, while hT4 is represented in panels c (HNA/RNA)
and d (HNA/DNA). The left side of each plot depicts a phase angle versus time graph, and the right side shows a polarplot of the phase angle. Time
increases when going from the inner circle of the polarplot (0 ns) toward its circumference (1.11 ns).
grooves will be highly comparable to each other. On the other
hand, however, the 2′-OH group of the RNA strand, which is
flanking the minor groove of the HNA/RNA duplex, but which
is not present in the HNA/DNA complex, is responsible for the
enhanced red coloring of the HNA/RNA SASA (Figure 7a) as
compared to the surface of HNA/DNA (Figure 7b). It is
therefore likely that substantial differences in hydration patterns
between the two duplexes can be expected along the minor
groove sides of the two double helices and that the 2′-OH group
of the RNA strand should play an important contribution in
this.
hydration between the HNA/DNA and HNA/RNA complexes.
Hydration patterns around the phosphate moieties and in both
the minor and major grooves were investigated.
(1) Phosphate Groups. Very little preferential hydration of
the phosphate groups is visible. There is no occurrence of single
water molecules simultaneously hydrogen bonding to each of
the free oxygen atoms of the phosphates, which is in agreement
with the A-DNA crystal structure of d(ICGCG)26 and with the
calculations on A-RNA by Cheatman and Kollman,27 but in
disagreement with theoretical calculations done by Clementi and
(26) Conner, B. N.; Takano, T.; Tanaka, S.; Itakura, K.; Dickerson, R.
E. Nature 1982, 295, 294-299.
(27) Cheatham, T. E., III; Kollman, P. A. J. Am. Chem. Soc. 1997, 119,
4805-4825.
Spatially Resolved Solvent Densities. Analysis of the spatial
distribution of the water oxygen atoms around each double helix
was carried out to further investigate possible differences in

Stability of HNA/RNA Hybrids
J. Am. Chem. Soc., Vol. 120, No. 22, 1998 5389
Figure 6. Time course and polarplots of the puckering phase angle P of (a) the ribose rings of the RNA strand in the HNA/RNA complex and (b)
the deoxyribose rings of the DNA strand in the HNA/DNA complex. Time increases when going from the inner circle of each polarplot (0 ns)
toward its circumference (1.11 ns).

5390 J. Am. Chem. Soc., Vol. 120, No. 22, 1998
Winter et al.
Table 3. Average Hydrogen Bond Lengths (and Standard Deviations) between Interstrand Base Pairs Calculated over the Last 500 ps of
Each Simulationa
HNA/RNA
(hG1)H2B‚‚‚(rC16)O2
HNA/DNA
(hG1)H2B‚‚‚(dC16)O2
1.93(0.14)
1.98(0.10)
2.01(0.19)
1.87(0.11)
1.93(0.09)
1.99(0.17)
1.86(0.11)
1.96(0.10)
2.03(0.21)
2.00(0.13)
2.00(0.18)
1.97(0.12)
2.11(0.26)
1.87(0.11)
1.96(0.09)
1.98(0.16)
1.88(0.11)
1.96(0.09)
1.98(0.17)
1.93(0.15)
1.97(0.11)
2.06(0.31)
1.97(0.22)
2.01(0.16)
2.17(0.56)
1.89(0.12)
1.95(0.09)
1.99(0.17)
1.88(0.12)
1.96(0.10)
2.03(0.19)
1.98(0.12)
2.05(0.21)
1.98(0.13)
2.10(0.27)
1.90(0.13)
1.97(0.09)
1.98(0.17)
1.88(0.12)
1.95(0.09)
1.97(0.17)
1.94(0.17)
2.03(0.29)
2.20(0.61)
(hG1)H1‚‚‚(rC16)N3
(hG1)O6‚‚‚(rC16)H42
(hC2)O2‚‚‚(rG15)H21
(hC2)N3‚‚‚(rG15)H1
(hC2)H4A‚‚‚(rG15)O6
(hG3)H2B‚‚‚(rC14)O2
(hG3)H1‚‚‚(rC14)N3
(hG3)O6‚‚‚(rC14)H42
(hT4)H3‚‚‚(rA13)N1
(hT4)O4‚‚‚(rA13)H62
(hA5)N1‚‚‚(rU12)H3
(hA5)H6A‚‚‚(rU12)O4
(hG6)H2B‚‚‚(rC11)O2
(hG6)H1‚‚‚(rC11)N3
(hG6)O6‚‚‚(rC11)H42
(hC7)O2‚‚‚(rG10)H21
(hC7)N3‚‚‚(rG10)H1
(hC7)H4A‚‚‚(rG10)O6
(hG8)H2B‚‚‚(rC9)O2
(hG8)H1‚‚‚(rC9)N3
(hG8)O6‚‚‚(rC9)H42
(hG1)H1‚‚‚(dC16)N3
(hG1)O6‚‚‚(dC16)H42
(hC2)O2‚‚‚(dG15)H21
(hC2)N3‚‚‚(dG15)H1
(hC2)H4A‚‚‚(dG15)O6
(hG3)H2B‚‚‚(dC14)O2
(hG3)H1‚‚‚(dC14)N3
(hG3)O6‚‚‚(dC14)H42
(hT4)H3‚‚‚(dA13)N1
(hT4)O4‚‚‚(dA13)H62
(hA5)N1‚‚‚(dT12)H3
(hA5)H6A‚‚‚(dT12)O4
(hG6)H2B‚‚‚(dC11)O2
(hG6)H1‚‚‚(dC11)N3
(hG6)O6‚‚‚(dC11)H42
(hC7)O2‚‚‚(dG10)H21
(hC7)N3‚‚‚(dG10)H1
(hC7)H4A‚‚‚(dG10)O6
(hG8)H2B‚‚‚(dC9)O2
(hG8)H1‚‚‚(dC9)N3
(hG8)O6‚‚‚(dC9)H42
a Values are in Å.
Table 4. Decomposition of Total Solvent Accessible Surface Areas into Contributions of the Base, Sugar, and Phosphate moieties and Polar
and Hydrophobic Atomsa
base
sugar
hydrophobicc
phosphate
polara
polarb
hydrophobicc
polarb
all
HNA/DNA Duplex
133(7)
DNA strand
HNA strand
total
209(6)
238(7)
447(10)
189(10)
164(8)
353(15)
403(11)
461(6)
864(12)
260(7)
272(5)
532(8)
1194(41)
1252(31)
2446(54)
117(5)
250(9)
HNA/RNA Duplex
236(6)
RNA strand
HNA strand
total
208(7)
237(6)
445(10)
170(8)
167(10)
337(14)
302(8)
464(6)
766(10)
277(4)
266(5)
543(7)
1193(33)
1247(32)
2440(49)
113(5)
349(8)
HNA/RNA duplexc
134(5)
RNA strand
HNA strand
total
208(7)
237(6)
445(10)
171(8)
167(10)
338(14)
382(7)
464(6)
846(10)
277(4)
266(5)
543(7)
1172(31)
1247(32)
2419(48)
113(5)
247(7)
a Values are in Å2. Standard deviations are in parentheses. b The solvent accessible surface area formed by O, N, and P atoms. c The solvent
accessible surface area formed by C atoms. d Calculated from the HNA/RNA trajectory but with the van der Waals radius of the ribose 2′-OH
group set to 0 Å.
Corongiu.28 However, adjacent phosphate groups in the HNA
and DNA/RNA strands are linked by single water molecules,
where each bridging water is simultaneously hydrogen bonded
to adjacent phosphate groups in the same oligonucleotide strand
(Figure 8). This observation of single water molecules bridging
adjacent phosphate groups is not only found in the two HNA/
DNA and HNA/RNA complexes but is also observed in
calculations performed on A-RNA.27 The bridging water
molecules, especially along the HNA strands, are positioned in
the major grooves of the complexes. It is interesting to note
that the density of the water molecules bridging the phosphates
of the DNA strand in the HNA/DNA complex is approximately
half the density of the waters along the RNA strand of the HNA/
RNA complex, which are still visible at a contouring level of 8
times the expected water density. This may again be an
indication of the higher flexibility of the DNA backbone in the
HNA/DNA complex as compared to the RNA strand in the
HNA/RNA duplex.
(2) Major Groove. Apart from the above-mentioned bridg-
ing waters between adjacent phosphates, additional hydration
patterns found in the major groove include monodentate binding
of water molecules to the N7 of HNA strand purines and to the
keto oxygens and amino groups of purine and pyrimidine bases
of the HNA chain (Figure 9). Again, as for the phosphate
solvation pattern mentioned above, these additional major
groove hydration schemes are observed in both the HNA/RNA
and HNA/DNA duplexes.
(3) Minor Groove. The minor grooves of the HNA/DNA
and HNA/RNA duplexes can be expected to be solvated
distinctly due to the presence of O2′-hydroxyl groups in the
RNA strand of the HNA/RNA complex and which are not
present in HNA/DNA. Hence, minor groove solvation of HNA/
DNA and HNA/RNA will be treated separately in the following.
(i) HNA/DNA complex: The minor groove in this duplex is
well-hydrated and involves interactions with both strands of the
complex. Most water molecules located in the minor groove
of the HNA/DNA duplex are well-defined, and density remains
easily visible up to a contouring level of at least 6 times the
expected water density. A chain of well-ordered water mol-
(28) Clementi, E.; Corongiu, G. Biopolymers 1979, 18, 2431-2450. (b)
Clementi, E.; Corongiu, G. J. Chem. Phys. 1980, 72, 3979-3992.

Stability of HNA/RNA Hybrids
J. Am. Chem. Soc., Vol. 120, No. 22, 1998 5391
each of these fractions. Although molecular dynamics by itself
is not a suitable tool to study differences in free energy, it could
be used to investigate differences in conformational energy,
base-pair hydrogen bonding, hydration, and conformational
flexibility. Hence, it is interesting to investigate whether the
current calculations might help us to speculate about possible
reasons for the increased stability of HNA/RNA as compared
to HNA/DNA.
Conformational Flexibility. (a) There seems not to be a
significant difference in backbone flexibility of the HNA strands
in HNA/RNA and HNA/DNA, as evidenced by the similar
standard deviations of each of the corresponding backbone
torsion angles (Table 2). In addition, if one excludes residue
hG1 from the comparison, the anhydrohexitol rings of the HNA
strands in HNA/RNA and HNA/DNA remain all in a chair-
like conformation, and the similar standard deviations on the
pucker amplitude and phase angles θ and æ are indicative of
comparable flexibility of the anhydrohexitol rings as well (Table
2). Hence, conformational flexibility of the HNA chain is not
likely a key factor in explaining the observed difference in
stability of HNA/RNA versus HNA/DNA. (b) The RNA and
DNA strands, on the other hand, display more variation in their
flexibility, as indicated by the larger standard deviations on the
DNA backbone torsion angles δ, ú, and ø and puckering
parameters as compared to the RNA values (Table 2 and Figure
6). However, it is unclear whether this difference in flexibility
is correlated with the observed difference in stability or whether
it is merely a characteristic inherently associated with the higher
barrier of sugar repuckering of RNA as compared to DNA.29
In this regard, the observation of a rigid RNA strand and a
flexible DNA strand are also observed in calculations of a RNA/
DNA hybrid structure,27 indicating that this is not a characteristic
of the HNA complexes alone.
Base Pair Hydrogen Bonding. The hydrogen bond distances
and their standard deviations given in Table 3 are unable to
discriminate between the HNA/RNA and HNA/DNA complexes
in terms of relative stability. In each of the two duplexes, both
strands interact strongly with each other through the formation
of stable base-pair hydrogen bonds. Hence, the lower stability
of the HNA/DNA complex in comparison to HNA/RNA is
likely not due to sterical or geometrical reasons, as we would
then expect to observe less stable hydrogen bonds in the case
of HNA/DNA. In fact, there are no indications to believe that
DNA strands are geometrically not capable of adopting an A-like
geometry suitable to bind HNA strands, as there is abundant
evidence of the spontaneous formation of A-type DNA double
helices in low humidity conditions.30
Figure 7. Color coding the total solvent accessible surface areas
(SASAs) by their underlying atom type: N, blue; O, red; P, purple;
and C, gray. (a) A view in the minor groove of the HNA/RNA complex.
(b) The minor groove of HNA/DNA. (c) The major groove of HNA/
RNA. (d) The major groove of HNA/DNA.
ecules forms a first hydration layer by bridging the O4′
deoxyribose ring oxygens of the DNA strand with the O2 or
N3 atoms of the adjacent nucleobase. This layer of water
molecules is linked to a second layer, which serves to complete
the tetrahedral environment and which forms a linkage between
the first water layer and the O2 or N3 atoms of the comple-
mentary anhydrohexitol nucleobases (Figure 10a). (ii) HNA/
RNA complex: In the case of HNA/RNA, a minor groove
hydration pattern similar to the HNA/DNA complex is observed
in casu a well-defined chain of water molecules that links both
strands of the duplex through contacts with the ribose O4′ atoms
of the RNA strand and with the O2 or N3 base atoms of both
strands (Figure 10b). As in the case of HNA/DNA, density
remains visible up to a contouring level of at least 6 times the
expected water density. However, due to the addition of the
O2′ hydroxyl groups of the RNA chain, an additional hydration
pattern is found as well. These O2′ hydroxyls contribute to
the stabilization of water molecules in the minor groove by
forming additional hydrogen bonds to the first hydration layer.
Thus, the first layer of water molecules in the minor groove of
the HNA/RNA complex not only bridges the O4′ atom and the
O2 or N3 atom of the adjacent nucleobase but also bridges this
O4′ atom and the O2′ hydroxyl group of the adjacent ribose
(Figure 10b).
Solvation. The total solvent accessible surface of both
complexes is almost equal and fluctuates around 2430 Å2 (Table
4). However, the minor groove of the duplexes, which in the
case of the HNA/RNA duplex is lined with the 2′-OH groups
of the ribose rings and which is relatively hydrophobic in the
case of the HNA/DNA helix, forms the major difference
between the two solvent accessible surfaces. As a consequence,
the HNA/RNA complex has about 130 Å2 more polar surface
than the HNA/DNA system, which undoubtedly results in
increased stabilization by solvent. In particular, upon formation
of the HNA/DNA hybrid duplex, the complementary DNA
strand is forced in a conformation in which its hydrophobic
deoxyribose ring becomes fully exposed to the solvent present
in the wide minor groove of all the atoms in the deoxyribose
Discussion
In general, HNA/RNA duplexes have higher melting tem-
peratures than the corresponding HNA/DNA complexes, which
is an indication of the increased stability of HNA/RNA as
compared to HNA/DNA. Stability of a complex is determined
by the free energy difference between the product and its
reactants and hence requires knowledge of the free energy of
(29) Olson, W. K.; Sussman, J. L. J. Am. Chem. Soc. 1982, 104, 270-
278.
(30) Saenger, W. Principles of nucleic acid structure; Springer-Verlag:
New York, 1984.

5392 J. Am. Chem. Soc., Vol. 120, No. 22, 1998
Winter et al.
Figure 8. Stereo plot of the typical single water bridges between adjacent intrastrand phosphate groups. Shown are the backbone phosphates
between anhydrohexitol residues hG3, hT4, and hA5 of the HNA/DNA complex, but similar arrangements are also observed for the phosphate
groups of DNA strand and also of the HNA/RNA complex. Dotted lines denote possible hydrogen bonds. Density contouring levels are 8 (dark
yellow), 6 (pale yellow), and 4 (purple) times the expected water density. Red dots are placed at positions of highest density.
Figure 9. Stereo plot of residues hT4-hG8 of the HNA/DNA complex and their interaction with solvent waters in the major groove of the
complex. A similar interaction pattern is observed for the HNA/RNA duplex. Hydrogen bonds are indicated by dotted white lines. Contouring is
done at 4 (yellow) and 3 (red) times the expected water density. Red dots are placed at positions of highest density.
ring, only the O4′ atom interacts with solvent waters (Figure
10a). Clearly, this geometry, which is imposed by the structural
rigidity of the HNA strand, is an unfavorable situation in terms
of solvation. In this regard, it is interesting to note that, in stable
double helices where at least one of the strands is a deoxyribose
oligonucleotide, the minor groove is always narrower than the

Stability of HNA/RNA Hybrids
J. Am. Chem. Soc., Vol. 120, No. 22, 1998 5393
Figure 10. Stereo drawing of the minor grooves of the two complexes showing the well-defined hydration patterns in these grooves. Contouring
is done at 4 (yellow) and 3 (red) times the expected water density. (a) HNA/DNA. (b) HNA/RNA.

5394 J. Am. Chem. Soc., Vol. 120, No. 22, 1998
Winter et al.
12 Å calculated here for the HNA/DNA hybrid. In particular,
minor groove widths around ∼6 Å are typical for B-type double
helical DNA, while for A-type DNA, which is only found in
high salt or crystalline conditions, values of ∼11 Å have been
published.30 In addition, minor groove widths for RNA/DNA
hybrids are in the range of 8-11 Å.31,32
Conclusions
Hexitol nucleic acids are intriguing structures as they belong
to the group of oligomers that hybridize strongly and selectively
with natural RNA. They show stronger base pairing as well as
a greater selectivity of pairing with ribooligonucleotides than
the pairing found between natural nucleic acids. The reason
for the higher stability of HNA/RNA duplexes as compared to
the HNA/DNA complexes was investigated here using molec-
ular dynamics. The selection of the starting conformation used
to carry out the modeling experiments was based on NMR
analysis of a HNA dimer. Both complexes (HNA/RNA and
HNA/DNA) show an A-type geometry and very similar
hydrogen bonding patterns between base pairs. Although it is
certainly not the only factor determining complex stability,
minor groove solvatation contributes to a considerable extent
to duplex stabilization and might explain, at least in part, the
relative stability of HNA/RNA versus HNA/DNA. To further
explore the influence of hydration on the stability of oligo-
nucleotide complexes with a six-membered carbohydrate mimic
in the backbone structure, we are currently modeling and
synthesizing the D-mannitol and D-altritol analogues of the
aforementioned hexitol nucleic acids. These studies might lead
to more efficient antisense constructs, functioning by sterically
blocking the RNA target.
Compared to the HNA/DNA situation, minor groove solvation
is clearly more optimal in the case of HNA/RNA. Just as for
HNA/DNA, the RNA strand is also held in a A-like conforma-
tion when complexed to HNA, and the complex is also
characterized by a wide minor groove that is fully accessible to
solvent. Not only the distribution of water molecules in the
minor groove of HNA/RNA is identical to the pattern in HNA/
DNA, but also the density of the minor groove water molecules
is very similar in both cases. However, the presence of 2′-OH
groups in the RNA strand renders the minor groove of the HNA/
RNA duplex significantly more hydrophilic as compared to the
HNA/DNA system. Not only the O4′ atoms but also the 2′-
OH groups of the ribose rings are all involved in well-defined
interactions with water molecules (Figure 10b). Therefore,
minor groove solvation might be an important factor which
could, at least in part, explain the relative stability of HNA/
RNA systems as compared to HNA/DNA. If this would be
the case, then double helical stability could be improved by
designing oligonucleotides with increased minor groove solva-
tion. Of course, other factors, such as entropy effects, are likely
also important in explaining the relative stabilities, but these
are difficult to estimate from molecular dynamics calculations.
Acknowledgment. This work was supported by a grant of
the K.U. Leuven (GOA 97/11). We thank J. Rozenski for mass
spectrometric analysis. A.V.A. is a research associate, and E.L.
is an aspirant of the Belgian National Fund of Scientific
Research. We are grateful to A. De Bruyn and H. van Halbeek
for their help in interpreting NMR data. We thank M. Vande-
kinderen for editorial help.
(31) Horton, N. C.; Finzel, B. C. J. Mol. Biol. 1996, 264, 521-533.
(32) Arnott, S.; Chandrasekaran, R.; Millane, R. P.; Park, H. S. J. Mol.
Biol. 1986, 188, 631-640.
(33) Cremer, D.; Pople, J. A. J. Am. Chem. Soc. 1975, 97, 1354-1358.
(34) Altona, C.; Sundaralingam, M. J. Am. Chem. Soc. 1972, 94, 8205-
8212.
JA973721F