New ligand combinations for the efficient stabilization of short nucleic acid hairpins

Article abstract of DOI:10.1016/S0040-4020(02)00876-1

Short nucleic acid hairpins (one or two base-pair stems) were strongly stabilized by simple chemical modifications. Non-nucleosidic pyrene or naphtalene diimide derivatives were appended at both 3′ and 5′ nearby ends of 2′-OMe RNA hairpins, yielding a very large increase in melting temperatures of the modified structures (from +21 to +55°C). The excimer formation between the two consecutive pyrene units is in favor of end-stacked pyrenyl rings.

Full text of DOI:10.1016/S0040-4020(02)00876-1

Tetrahedron 58 (2002) 7975–7982
New ligand combinations for the efficient stabilization of short
nucleic acid hairpins
Justine Michel,a Katell Bathany,b Jean-Marie Schmitter,b Jean-Pierre Montic and Serge Moreaua
,
*
a
´
´
IFR Pathologies Infectieuses, INSERM U-386, Bat 3a, Universite Victor Segalen Bordeaux 2,146 rue Leo Saignat, 33076 Bordeaux, France
´
b
Institut Europeen de Chimie et Biologie, FRE CNRS 2247, 16, avenue Pey Berland, 33607 Pessac, France
´ ´
UFR Sciences Pharmaceutiques EA 491, Universite Victor Segalen, 146 rue Leo Saignat, 33076 Bordeaux, France
c
Received 15 April 2002; revised 14 June 2002; accepted 15 July 2002
Abstract—Short nucleic acid hairpins (one or two base-pair stems) were strongly stabilized by si0 mple chemical modifications. Non-
nucleosidic pyrene or naphtalene diimide derivatives were appended at both 30 and 50 nearby ends of 2 -OMe RNA hairpins, yielding a very
large increase in melting temperatures of the modified structures (from þ21 to þ558C). The excimer formation between the two consecutive
pyrene units is in favor of end-stacked pyrenyl rings. q 2002 Published by Elsevier Science Ltd.
1. Introduction
The factors contributing to the thermodynamic stability of
DNA or RNA double helices include hydrogen bonding and
base stacking. Both are important in stabilizing non-
covalent interactions of nucleic acid complexes. An initial
experimental estimation which relies on terminal unpaired
bases was made on RNA strands and concluded in a
relatively equal contribution of base stacking and hydrogen
bonding to the stability of a base pair.1 Intercalation of
heteroaromatic rings between two base pairs in a DNA
double helix is a well-known example of the stacking
properties of flat aromatic structures.2 Our aim was to
Herein we examined the stacking properties of various end
combinations of these two aromatic cycles on short-stem
loop oligomers, which are potential ligands of TAR (an
RNA structural element of the HIV genome).7
develop the combination of flat aromatic rings enabling the
stabilization of very short double strands (one or two base
pairs) through end-stacking. By looking for synthetic
derivatives which might exhibit such properties, we
investigated a pyrene system which0 is th0ought to improve
duplex stability when appended to 3 or 5 oligonucleotides
ends.3 We also examined the properties of a naphtalene-
tetracarboxylic diimide derived compound which has been
shown to allow the construction of stacked aromatic cores.4
Furthermore, examination of the fluorescence properties of
pyrene should allow their stacking arrangement to be
checked.5 The conjugation of these aromatic cycles to short
nucleic acid oligomers was performed by using an
appropriately functionalized serinol backbone as a minimal
nucleotide replacement unit.6 The phosphoramidite of
derivatives P and N featuring either a pyrene or a naphtalene
were synthesized and used in the chemical synthesis of
nucleic acid sequences.
We expected these ligands to be stabilized through the
formation of a closed hairpin, as shown on the hypothetical
diagram shown above.
2. Chemical synthesis
The synthetic routes for the diols P and N and their
corresponding phosphoramidites are depicted in Scheme 1.
The non-symmetrical diimide compound 1 was obtained by
successive addition of 0.5 equiv. of glycine aminoacid and a
slight excess of methylamine to a solution of the
commercially available naphtalenedianhydride in DMF.
The crude acid 1 was isolated by acidic precipitation and
Keywords: nucleic acid hairpins; stacking interactions; naphtalene diimide;
pyrene derivatives; fluorescence; excimer.
*
Corresponding author. Tel.: þ33-5-57-5710-16; fax: þ33-5-57-5710-15;
e-mail: serge.moreau@bordeaux.inserm.fr
0040–4020/02/$ - see front matter q 2002 Published by Elsevier Science Ltd.
PII: S0040-4020(02)00876-1

7976
J. Michel et al. / Tetrahedron 58 (2002) 7975–7982
Scheme 1. Chemical synthetic pathway for N and P compounds and their corresponding phosphoramidites. (a) HOBt, DIPC, DMF, rt, 10 min, then Serinol,
DIEA, 408C, 16 h. (b) Glycine, DIEA, DMF, 908C, 16 h then CH3NH2, rt, 3 h. (c) PyBOP, HOBt, TEA, Serinol, DMF 408C, 3 h. (d) DMTCl, pyridine, rt, 2 h.
(e) 2-Cyano-N,N-diisopropylphosphoramidochloridite, DIEA, CH2Cl2, rt, 1 h.
then used without further purification. The coupling of
either 2-pyrene acetic acid or 1 to serinol easily yielded the
P and N diols which were purified through flash chroma-
tography. The corresponding phosphoramidites 3a and 3b
were obtained by using classical procedures in two steps.
First a monotritylation using 1 equiv. of 4,40-dimethoxy-
trityl chloride followed by flash silicagel chromatography
yielded the monotritylated compounds 2a and 2b. Second a
phosphitylation reaction with 2-cyanoethyl-N,N-diiso-
propylchloro-phosphoramidite in the presence of N,N-
diisopropylethylamine afforded the desired compounds 3a
and 3b. These compounds are obtained as mixtures of
stereoisomers and are used as such in solid phase nucleic
acid synthesis. The structures of these various compounds
were checked by mass spectrometry (FAB) and NMR
spectrometry (1H, 13C and 31P). All the spectral character-
istics of these compounds are given in Section 8.
couple the various phosphoramidite monomers. The
coupling times of the monomer P and N were increased to
15 min. Under these conditions the coupling yields were
similar to those of standard 20-O-Me monomers, as
estimated from the detritylation profiles. For the 30
conjugation of N and P monomers, a universal support
(commercially available from Glen Research) was used. In
order to eliminate any amine reactions on the diimide
groups of N,8 a sodium hydroxide deprotection procedure
was chosen, avoiding prolonged contact with the amine
groups. In all cases the chemical integrity of the oligomers
was checked by MALDI-TOF mass spectra analysis
(Table 1).
4. Hairpin stability
The contributions of N and P moieties to the stability of the
hairpins were analyzed by using UV thermal denaturation
experiments. The oligomers H10 and H12 may fold as
hairpins with short stems of 1–2 base pairs. As expected
these oligonucleotides exhibited a single transition in UV
thermal monitored melting experiments. Experiments con-
ducted on H12 and 12P5P3 oligomers at concentrations
ranging from 0.5 to 7.5 mM revealed no variations in Tm
data (50^18C for H12 and 72^18C for 12P5P3. This is
indicative of a monomolecular equilibrium, thus showing
that the observed Tm are related to the fusion of a hairpin
3. Oligonucleotide synthesis
The phosphoramidites of N and P were used to incorporate
the two aromatic nuclei into oligonucleotide sequences. The
oligomer sequences were conceived from previous data on
the characterization of aptameric ligands for the TAR RNA7
and were constructed using 20-O-Me RNA nucleoside units.
The synthesized oligomeric sequences are shown in Table 1.
The standard solid phase synthesis procedure was used to

J. Michel et al. / Tetrahedron 58 (2002) 7975–7982
7977
Table 1. Hairpin sequences
Ida
Oligonucleotide sequencesb
Calcd mass
Found
Tm (DTm) 260 nmc
H% (Tm) 343 nmd
H12
H10
CGGUCCCAGACG
GGUCCCAGAC
nd
nd
nd
nd
50.0
20.0
12P3
12N3
12P5
12N5
12P5P3
12N5N3
12N5P3
12P5N3
10P5
CGGUCCCAGACGP
CGGUCCCAGACGN
PCGGUCCCAGACG
NCGGUCCCAGACG
PCGGUCCCAGACGP
NCGGUCCCAGACGN
NCGGUCCCAGACGP
PCGGUCCCAGACGN
PGGUCCCAGAC
4370.8
4450.0
4342.0
4418.8
4767.4
4928.3
4845.4
4845.4
3663.6
4088.9
4767.4
4484.3
4484.3
4622.9
4372.2
4442.3
4344.7
4419.9
4771.4
4929.4
4848.5
4850.7
3665.1
4089.8
4771.4
4490.7
4484.3
4631.8
66.5 (16.5)
62.0 (12.0)
60.0 (10.0)
50.0 (0.0)
72.0 (21.0)
.70.0 (.21.0)
70.0 (21.0)
62.0 (12.0)
41.5 (21.5)
62.0 (42.0)
66.0 (46.0)
75.0 (55.0)
75.0 (55.0)
56.0
13
10 (60.0)
19 (72.0)
nd
10P3
GGUCCCAGACP
16 (62.0)
16 (67)
13
10P5P3
10PP5P3
10P5PP3
H14
PGGUCCCAGACP
PPGGUCCCAGACP
PGGUCCCAGACPP
50ACGGUCCCAGACGU30
16
a
Id: identification of oligonucleotides. The number stands for the length of the oligomer; the letters P and N stand for the ligands pyrene and NDI, respectively.
5 and 3 indicate the substituted end(s).
Complementary base pairs are underlined.
The melting temperature Tm (^0.58C) was obtained under the conditions described in Section 8. DTm¼Tm(modified oligomers)2Tm(unmodified parent
hairpin). The unmodified hairpins used as references are termed H10 for 10-mer sequences and H12 for 12-mer sequences.
H%: percent of hyperchromism observed at 343 nm (nd: not determined).
b
c
d
structure and not a bulged duplex. In the following
0.88. The 10P5P3 hairpin exhibited similar fluorescence
features with a ratio IE/IM rising to 5.7. The tris pyrenyl
conjugates 10PP5P3, 10P5PP3 spectra were almost entirely
constituted by excimer emission (Fig. 1(c)).
experiments, Tm data were recorded at a 1 mM concen-
tration of oligomers. The H10 hairpin, although showing a
weakly cooperative curve, was c0haracterized by a melting
transition at 208C. All the 30 or 5 modified hairpins except
oligomer 12N5 exhibited a very significant increase in Tm
relative to the corresponding unmodified ones. Thus, a
single pyrene residue in the 30 position of H10 or H12
resulted in a DTm of þ42 and þ16.58C, respectively (10P3,
12P3, Table 1). Multiple N or P substitutions at 30 and 50
revealed even higher increases since DTm values of þ55.08C
could be reached with the combination of three P at 30 and 50
ends of H10 (see 10PP5P3, 10P5PP3, Table 1). When
monitored at 343 nm, the long wavelength absorption band
of the pyrene cycle, the UV thermal denaturation experi-
ments on 12P5 and 12P5P3 revealed a significant hyper-
chromic contribution (10 and 19%, respectively). In the
latter cases, Tms were observed at the same values as those
at 260 nm (Table 1). The hyperchromic effect was always
significant at 343 nm, but a clear cooperative transition was
not always detected for some oligomers, thereby preventing
a clear-cut melting transition to be determined at this
wavelength (Table 1).
6. Molecular modeling studies
To further understand the structures of our hairpin species,
we conducted molecular mechanics studies. Our objective
was to evaluate the stacking or intercalating abilities of the
P and N pending aromatic cycles. To construct a three base-
pair model, we chose an RNA double strand mimicking the
stem of the H12 hairpin with a terminal0CG base pair. P and
N substituents were introduced at the 3 or 50 ends. Stacked
or intercalated species were constructed. During the
minimizations, the base pairs were fully maintained as
well as the puckering of the ribose units. The mono-
substituted species 30P, 50P, 30N, 50N all proved capable of
stacking (Fig. 2(b)) over the last CG pair or of intercalating
(Fig. 2(a)) between the first two pairs. The N–CH3
substituent of the NDI nucleus did not prevent intercalation
(Fig. 2(a)). According to the fluorescence data, we focused
our attention on the bis substituted P species, looking at both
possible conformations (30 over 50, or vice-versa). No
particular steric constraints were detected by the minimiz-
ation calculation, thus indicating very similar energies for
both stacked structures. An illustration of possible stacking
geometries is shown in Fig. 2(c). The two consecutive P
units overlap each other, with the stacking area correspond-
ing roughly to 75% of the aromatic pyrene surface.
5. Fluorescence studies
The fluorescence properties of pyrene were used to check
for the stacked arrangements of the chromophores. Upon
excitation at 349 nm, the 12P3 oligomer showed a
fluorescence emission characteristic of a monomer (two
main peaks at 379 and 395 nm and a weak shoulder at
421 nm5 (Fig. 1(a)). Similar results were observed for 12P5.
The 12P5P3 hairpin, which incorporates two nearby
pyrenes, exhibited fluorescence emission properties indica-
tive of excimer formation. When excited at 349 nm, the
hairpin emission spectra showed a new broad band centered
at 485 nm, together with the monomer emission bands
(Fig. 1(b)). The ratio of intensities (IE/IM) of excimer
emission (485 nm) to that of the monomer (379 nm) was
7. Discussion
The present molecular modeling data indicate that the short
flexible linker allows end stacking or intercalation whatever
the position or the residue. The experimental data (Tm,
hyperchromism) regarding the monosubstituted species are
consistent with both intercalation and dangling positions of

Figure 1. (a) Excitation fluorescence spectrum (- - -) and emission spectrum (__) of hairpin 12P3. The fluorescence excitation was recorded for an emission wavelength set at 379 nm for 0.1 mM of 12P3 in the following
buffer: 20 nM sodium cacodylate, 140 mM potassium chloride, 20 mM sodium chloride (pH 7.3), 3 mM magnesium chloride. The emission spectrum (lexc 349 nm) was obtained under the same conditions. (b) Emission
spectrum of 12P5P3 (0.1 mM) (lexc 349 nm). The IE/IM (0.88) ratio was measured at 379 and 485 nm. (c) Emission spectrum of 10PP5P3 (0.1 mM) (lexc 349 nm). In both cases the buffer was: 20 nM sodium cacodylate
(pH 7.3), 140 mM potassium chloride, 20 mM sodium chloride, 3 mM magnesium chloride.

J. Michel et al. / Tetrahedron 58 (2002) 7975–7982
7979
Figure 2. Illustrations of possible intercalating or stacking geometries of N or P units on the 3 base-pair model used in this study (wireframe representation).
(a) Intercalated 30 NDI unit (view from the major groove). (b) 50 end stacked pyrene cycle above the last GC pair. (c) Wireframe view of the 3 base-pair model
duplex with the 2 stacked pyrenyl units. Left: view from the major groove. Right: view looking down the helical axis, dark bold is the 50 P unit, light bold is the
30 P unit above the last GC pair (wireframe).
P and N residues. Therefore, it is not possible to decide
which of the two hypotheses is more likely and cannot rule
out the occurrence of an equilibrium between the two
structures. NMR structural studies would have to be
conducted to obtain further experimental evidence.
for the H10 derivatized hairpin.5 The fluorescence data of
the tris pyrenyl derivatives (10PP5P3, 10P5PP3) reinforce
these conclusions since almost only excimer emissions were
observed in these hairpins. Taken together, these data (Tm,
hyperchromism at 343 nm, fluorescence) clearly suggest a
successive stacking arrangement of the aromatic cycles
above the last base pair of the hairpin stem. Among the four
stacked combinations studied here P5P3, N5N3, and N5P3
behave similarly. Additional stabilizations was clearly
provided in the case of the bis N-substituted hairpin
12N5N3 (DTm¼þ218C) in comparison with 12N3 and
12N5 (DTm¼þ12 and 08C respectively). A similar situation
occurred with 12P5P3 (DTm¼þ218C) and 12N5P3
(DTm¼þ208C), whereas 12P5N3 did not manifest any
additional contribution. Molecular modeling studies on
stacked bis pyrenyl entities (Fig. 2) revealed a possible
overlap of the pyrenyl cycles. The observed geometry does
not fully correspond to the optimum overlap (almost
complete) expected for the ground state excimer structure.5
However, the variations in intensities of excimer emission
might be due to variations in the overlap of the aromatic
pyrenic rings in the H10 or H12 hairpins and to the relative
orientations of the rings. The tris pyrenyl species (10PP5P3,
10PP3P5) led to high excimer emission as the two last
pyrenyl units are free to adopt a fully overlapped structure
(not shown).
30 or 50 substitutions re0vealed different contributions.
Whatever the residue, the 3 position provided more efficient
stabilization with the non-polar pyrene residue being the
best one in this case. Such observations have0 been reported
for various non-natural nucleosides at the 5 end of DNA
duplexes. For example it has been shown that the best
stacking is obtained with non-polar molecules through
the hydrophobic effect.3 Newcomb and Gellman demon-
strated that optimal stacking properties can be reached
between polar and apolar aromatic rings.9 Geometrical
analysis of our structure failed to indicate significant
differences in the overlap surfaces of 30 or 50 positions
(not shown).
Simultaneous substitutions at 50 and 30 generally led to an
additional stabilization of the hairpins, revealing inter-
actions between the substituents. Thus the two opposing
pyrene units might constitute an attractive interstrand
contribution, as can be inferred from the fluorescence
data. The formation of an excimer revealed the occurrence
of stacked interactions between the pyrenyl nuclei. The ratio
of intensities of excimer emission to monomer emission was
significantly different for H12 or H10 derivatives (a relative
increase of 5.7-fold was observed in the latter case). This
suggests first a ground state dimer association of the pyrenes
and therefore a better arrangement of the stacked pyrenes
In conclusion, the various substituents introduced at the 30 or
50 ends of the hairpin H10 and H12 stabilize the structures.
The contribution made by the terminal residues in most
cases exceeds that of the additional AU base pair included in
the H14 hairpin (a three base-pair stem, last line Table 1).

7980
J. Michel et al. / Tetrahedron 58 (2002) 7975–7982
Thus a Tm of 568C was observed for this hairpin resulting in
an increase of þ68C compared to the H12 hairpin.
Appended pyrene residues at the 30 and 50 ends of the H10
hairpin (one GC base-pair stem) resulted in a Tm of 668C
(DTm¼þ468C in comparison with H10 and þ108C in
comparison with H12). A combination of three pyrene units
led to an even more stable hairpin (Tm¼758C), 198C more
than the 3 base-pair H14 hairpin. On the whole the polar
naphtalene diimide moiety led to a similar stabilization to
that provided by the apolar pyrene when conjugated at the
two nearby hairpin ends. The major determining factor in
such stacked compounds is very probably related to the
overlap surface on stacking.3 Although more polar and more
hydrated the naphtalene diimide behaves like a hydrophobic
compound. This might be due to its structural analogy with
the polar nucleic acid bases, where the polar functional
groups lie on the edges rather than on the aromatic faces of
the rings.3
8.2. Chemical procedure
8.2.1. N-(2-Hydroxy-1-hydroxymethyl)-ethyl-1-yl-acet-
amide (P). 2-Pyreneacetic acid (500 mg) was dissolved in
dimethylformamide (15 ml) and preactivated for 15 minutes
at room temperature (rt) with hydroxybenzotriazole
(282 mg) and 1.1 equiv. of diisopropylcarbodiimide
(327 ml). Then, serinol (173 mg) and 1 equiv. of diiso-
propylethylamine (327 ml) were added. The reaction was
stirred overnight at 408C. The mixture was concentrated,
dissolved in CH3OH/CH2Cl2 (1/2), and washed twice with
water. The organic phase was then dried over MgSO4and
evaporated. The diol was purified by flash chromatography
with a gradient of methanol in dichloromethane to afford P
(320 mg, yield¼50%).
UV (MeOH) lmax (nm) (1) 339 (76,500), 322 (50,200), 272
(79,700); IR (KBr) 1630 cm21 1H NMR (200 MHz, DMSO-
d6) d (ppm): 3.45 (s, 4H, CH2OH), 3.77 (m, 1H, CH), 4.2 (s,
2H, CH2CO), 8.1 (m, CH ar). 13C NMR (50.32 MHz,
DMSO-d6) d (ppm): 40.0 (CH2CO), 52.9 (CH), 60.0
(CH2OH), 123.9, 124.1, 124.8, 125.0, 126.1, 126.7, 127.0,
127.3, 127.7 (CHar), 128.6, 128.9, 129.6, 130.3, 120.8,
131.2 (Cquat ar), 170.0 (CO). MS (FAB) m/z 334.13
(MHþ); HRMS (FAB) calcd for C21H20NO3 (MHþ)
334.1443, found 334.1437.
Although the structural parameters of the various hairpins
are not completely defined, the ligands N and P we used in
this study (while leading to stereoisomeric mixtures) make a
very high stacking contribution to the stability of double-
stranded nucleic acid structures. The interstrand interaction
provided by two opposite stacked aromatic ligands can
exceed the contribution of two additional base pairs in the
duplex stabilization. As mentioned in the introduction, the
hairpins used in this study are potential ligands for the TAR
RNA structure. Work is in progress to evaluate the
biological properties of these modified hairpins. The
combination of N and P derivatives might be a useful tool
for the efficient closure of synthetic short-stem oligonucleo-
tides and might potentially impair exonuclease activities.
They should be evaluated as substitutes for cyclization in
decoy strategies.
8.2.2. N-(2-Hydroxy-1-(-O-4,40-dimethoxytritylmethyl-
ethyl))-2-pyren-1-yl-acetamide (2a). P (297 mg) was
carefully dried by three co-evaporations with pyridine and
then dissolved in anhydrous pyridine (15 ml). Dimethoxy-
tritylchloride (1.1 equiv., 334 mg) was added under argon
atmosphere and mixed at room temperature for 2 h. The
mixture was poured on water and the organic phase was
extracted twice with dichloromethane, dried over MgSO4
and concentrated. 2a was purified by flash chromatography
with methanol/dichloromethane/triethylamine, 5/94/1
(v/v/v) to afford pure oil (236 mg, yield¼35%).
8. Experimental
8.1. General
1H NMR (200 MHz, CDCl3) d (ppm): 3.0 (d, 2H, CH2O, d,
1H, OH), 3.14 (dd, 2H, CH2OH), 3.77 (m, 1H, CH), 3.8 (s,
6H, 2OCH3), 4.48 (s, 2H, CH2CO), 8.23 (m, 9H, CHar); 13
C
Thin layer chromatography (TLC) was performed on Merck
silica gel 60 F254 aluminium-backed plates, and visualiza-
tion was by UV illumination. Flash chromatography refers
to column chromatography performed on a BIOTAGE
system. UV spectra were obtained on a KONTRON Uvikon
933 spectrophotometer and IR spectra were measured on a
BRUKER FT/IR spectrophotometer. NMR spectra were
recorded on a BRUKER AC200 spectrometer working at
NMR (50.32 MHz, CDCl3) d (ppm): 41.7 (CH2CO), 51.7
(CH), 54.8 (2OCH3), 63.3 (CH2O), 85.7 (Cquat DMT),
112.7, 126.4, 127.5, 126.4, 127 (CHar DMT), 122.7, 123.6,
125.0, 125.2, 126.0, 127.3 (Cquat ar DMT), 122.7, 123.6,
125.0, 125.2, 126.0, 127.3 (CHar pyrene), 128.0, 129.2,
130.5, 130.7, 131.1 (Cquat ar pyrene), 171.5 (CONH). MS
(FAB) m/z 658.26 (MþNaþ); HRMS (FAB) calcd for
C42H37NNaO5 (MþNaþ) 658.2574, found 658.2569.
8.2.3. N-(2-Cyanoethoxy(diisopropylamino)-1-(-O-4,40-
dimethoxytritylmethyl-ethyl))-2-pyren-1-yl-acetamide
(3a). 2a (140 mg) was carefully dried by three co-evapora-
tions with pyridine and then was diluted in anhydrous
dichloromethane (2 ml, stabilised with amylene). Diiso-
propylethylamine (2.5 equiv., 57 ml) and 2-cyano-N,N-
diisopropylphosphoramidochloridite (1.5 equiv., 78 mg)
dissolved in dichloromethane (0.5 ml) was added to the
mixture and stirred at room temperature for 1 h. The
reaction was monitored by TLC (ethyl acetate/hexane/
triethylamine, 45/45/1, v/v/v). The crude material was
poured onto ethyl acetate and the organic phase was washed
1
200 MHz for H, 50.32 MHz for 13C and 81.02 MHz for
31P. The chemical shifts are expressed in ppm using TMS as
1
internal standard (for H and 13C data) and 85% H3PO4 as
external standard (31P data). FABMS spectra and high-
resolution FABMS analysis were performed in m-nitro-
benzyl alcohol on a VG Autospec spectrometer. The mass
spectra of modified oligonucleotides were obtained on a
MALDI-ToF mass spectrometer (Reflex III, Bruker)
operating in the linear mode for negative ion detection
and using a mixture of oligonucleotides d(T12)–d(T18)
(Sigma) for external calibration mass spectrometry (Brucker
Reflex III). The matrix used was THAP (2,4,6-trihydroxy-
acetophenone).

J. Michel et al. / Tetrahedron 58 (2002) 7975–7982
7981
twice with NaHCO3 solution and water. Pure phosphor-
amidite (96 mg, yield¼50%) was obtained after purification
by flash chromatography (ethyl acetate/hexane/triethyl-
amine, 45/45/1, v/v/v).
CH2CONH), 4.64 (t, J 3¼5.4 Hz, 2H, OH), 8.0 (d,
J 3¼7.9 Hz, 1H, NH), 8.52 (s, 4H, Har); 13C NMR
(50.32 Hz, DMSO-d6) d (ppm):26.9 (CH3N), 42.6 (CH2-
CONH), 53.2 (CH), 60.4 (CH2OH), 125.7–126.5 (Cquat
ar), 130.2–130.6 (CHar), 162.3–162.7 (CON), 166.0
(CONH). MS (FAB) m/z 412.14 (MHþ); HRMS (FAB)
calcd for C20H17N3O7 (MHþ) 412.1144, found 412.1146.
1H NMR (200 MHz, CDCl3) d (ppm): 1.0 (m, 12H,
CHi3Prop), 2.44 (m, 2H, CH2CN), 3.56 (m, 8H, CHiProp,
CH2bCN, 2CH2O), 3.8 (s, 6H, CH3O DMT), 3.8 (m, 1H,
CH), 4.6 (s, 2H, CH2CO), 6.7–7.2 (m, 13H, Har DMT),
87.2 (m, 9H, Har NDI); 13C NMR (50.32 MHz, CDCl3) d
(ppm): 19.8–19.9 (CH2CN), 24.2 (CHi3Prop), 42.0
(CH2CO), 42.6–42.8 (CHiProp), 49.7 (CHCH2O), 54.9
(OCH3), 57.7–58.1 (OCH2CH2), 60.8, 61.0, 61.4, 61.7
(CHCH2O), 85.5 (Cquat DMT), 117.5 (CN), 122.8, 124.9,
125.3, 126.1, 127.3, 127.5 (CHar pyrene), 112.7, 126.4,
127.4, 128.2, 129.6 (CHar DMT), 124.4, 124.9, 129.3,
130.6, 130.7, 131.1 (Cquat ar pyrene), 135.4, 144.5, 158.1
(Cquat ar DMT), 170.3 (CO). 31P NMR (200 MHz, CDCl3)
d (ppm): 146.0–146.2. MS (FAB) m/z 858.37 (MþNaþ);
HRMS (FAB) calcd for C51H54N3NaO6P (MþNaþ)
858.3644, found 858.3647.
8.2.6. N-Methyl-N0-(N-(2-hydroxy-1-(-O-4,40-dimethoxy-
tritylmethyl-ethyl)-acetamide))-1,4,5,8-naphtalenetetra-
carboxylic diimide (2b). N (250 mg) was carefully dried by
three co-evaporations with pyridine and then dissolved in
anhydrous pyridine (30 ml). Dimethoxytrityl chloride
(0.8 equiv., 163 mg) was added under argon atmosphere
and mixed at room temperature for 2 h. The reaction was
monitored by TLC (ethyl acetate/hexane, 70/30, v/v). The
mixture was poured onto water (30 ml), and the organic
phase was extracted twice by MeOH/CH2Cl2 (1/2, v/v) and
was then washed, dried and concentrated. Pure 2b was
obtained after purification by flash chromatography with a
gradient of ethyl acetate in hexane (70–100%) in the
presence of 1% triethylamine (200 mg, yield¼46%).
1H NMR (200 MHz, CDCl3) d (ppm): 3.2 (d, 2H, CH2OH),
3.4 (s, 2H, CH2O), 3.54 (s, 3H, CH3N), 3.76 (s, 6H, CH3O
DMT), 4.0 (m, 1H, CH ), 4.8 (s, 2H, CH2CO), 6.67–7.25
(m, 13H, CHar DMT), 8.6 (s, 4H, CHar NDI); 13C NMR
(50.32 MHz, CHCl3) d (ppm): 27.3 (CH3N), 43.0 (C H2CO),
51.4 (CH), 55.1 (CH3O DMT), 62.8–63.2 (2CH2O), 86.2
(Cquat DMT), 113.1, 126.6, 127.9, 129.8 (CHar DMT),
130.7–131.1 (CHar NDI), 126.0–126.8 (Cquat ar NDI),
135.5, 144.3, 158.4 (Cquat ar DMT), 162.4–162.8 (CON),
166.5 (CONH). MS (FAB) m/z 726.23 (MþNaþ); HRMS
(FAB) calcd for C41H35N3NaO9 (MþNaþ) 736.2270, found
736.2263.
8.2.4. N-Methyl-N0-carboxymethyl-1,4,5,8-naphtalene-
tetracarboxylic diimide (1). Naphtalenedianhydride (3 g)
was dissolved in dimethylformamide (120 ml) and glycine
(1 equiv., 838 mg) and diisopropylethylamine (0.5 equiv.,
1.92 ml) were then added. The reaction was stirred
overnight at 908C. The mixture was cooled and methyl-
amine in ethanol (1.5 equiv., 2.04 ml) was added and stirred
at room temperature for 3 h. The crude material was
concentrated and then dissolved in water/methanol (2/1).
The pH was adjusted to 3 with concentrated HCl. The
residue was concentrated and then washed with methanol.
The pure acid 1 was obtained by filtration (3.2 g,
yield¼83%). UV (MeOH) lmax (nm) (1) 373 (19,300),
353 (16,900), 232 (19,600), 205 (19,800); IR (KBr) 1704,
1668 cm21; H NMR (200 MHz, DMSOd6) d (ppm): 3.39
8.2.7. N-Methyl-N0-(N-(2-(2-cyanoethoxy(diisopropyl-
amino))-1-(-O-4,40-dimethoxytritylmethyl-ethyl)-acet-
amide))-1,4,5,8-naphtalenetetracarboxylic diimide (3b).
2b (260 mg) was carefully dried by three co-evaporations
with pyridine and then was diluted in anhydrous dichloro-
methane (3 ml, stabilized with amylene). Diisopropylethyl-
amine (2.5 equiv., 140 ml) and 2-cyano-N,N-diiso-
propylphosphoramidochloridite (1.5 equiv., 115 mg) dis-
solved in dichloromethane (0.5 ml) was added to the
mixture and stirred at room temperature for 1 h. The
reaction was monitored by TLC (ethyl acetate/hexane/
triethylamine, 45/45/1, v/v/v). The crude extract was poured
onto ethyl acetate and the organic phase was washed twice
with NaHCO3 solution and water. Pure phosphoramidite
(190 mg, yield¼50%, Rf¼0.66) was obtained after purifi-
cation by flash chromatography (ethyl acetate/hexane/
triethylamine, 45/45/10, v/v/v).
1
(s, 3H, CH3N), 4.7 (s, 2H, CH2), 8.6 (dd, 4H, Har); 13C
NMR (50.32 MHz, DMSO-d6) d (ppm): 26.9 (CH3N), 41.6
(CH2), 125.2–126.5 (Cquat ar), 130.2–130.8 (CHar),
162.0–162.5 (CO). MS (FAB) m/z 339.22 (MHþ); HRMS
(FAB) calcd for C17H11N2O6 (MHþ) 339.0617, found
339.0620.
8.2.5. N-Methyl-N0-(N-(2-hydroxy-1-hydroxymethyl-
ethyl)-acetamide)-1,4,5,8-naphtalenetetracarboxylic di-
imide (N). 1 (1 g) was dissolved in dimethylformamide
(200 ml), PyBOP (3 equiv., 3.8 g) and then hydroxybenzo-
triazole (HOBt) (2 equiv., 0.665 g), triethylamine (2 equiv.,
0.684 ml) and serinol (2.5 equiv., 0.560 g) were added. The
mixture was stirred at 408C for 3 h. The reaction was
monitored by TLC (methanol/CH2Cl2; 2/80 v/v). The crude
material was concentrated. The product was purified by
flash chromatography with a 10–50% gradient of methanol
in dichloromethane. To eliminate the residual HOBt, the
product was dispersed in acetone (30 ml) and poured onto
water (150 ml). After a few hours, the diol precipitated. N
was then filtered off (0.385 mg, yield¼42%).
1H NMR (200 MHz, CDCl3) d (ppm): 1.12 (m, 12H,
CH3iPr ), 2.52 (m, 2H, CH2CN), 3.52 (m, 11H, CH2O DMT,
CH2OP, CH2bCN, 2CHiPr, CH3N), 3.75 (s, 6H, CH3O
DMT), 4.29 (m, 1H, CH), 4.8 (s, 2H, CH2CONH), 6.7–7.2
(m, 13H, Har DMT), 8.7 (s, 4H, Har NDI); 13C NMR
(50.32 MHz, CDCl3) d (ppm):20.3 (CH2CN), 24.5–24.6
(CHi3Prop), 27.3 (CH3N), 42.9 (CH2CONH), 49.9 (CH),
55.1 (CH3O), 58.6 (CH2bCN), 61.4–62.0 (CH2O), 86.0
(Cquat DMT), 113.0, 126.7, 127.8, 128.1, 129.9 (CHar
UV (MeOH) lmax (nm) (1 ) 374 (14,900), 355 (11,900), 231
1
(21,100); IR (KBr) 1676, 1664 cm21; H NMR (200 MHz,
DMSO-d6) d (ppm): 3.35 (s, 3H, CH3N), 3.4–3.42 (d,
J 3¼5.54 Hz, CH2OH), 3.7 (m, 1H, CH), 4.6 (s, 2H,

7982
J. Michel et al. / Tetrahedron 58 (2002) 7975–7982
DMT), 126.2–126.7 (Cquat ar NDI), 130.8–131.1 (CHar
NDI), 135.7, 144.5, 158.4 (C quat ar DMT), 162.5–163.0
(CON), 165.7 (CONH). 31P NMR (CDCl3) d (ppm): 146.4–
146.6. MS (FAB) m/z 936.33 (MþNaþ); HRMS (FAB)
calcd for C50H52N5NaO10P (MþNaþ) 936.3349,
found.936.3377.
8.5. Fluorescence measurements
The fluorescence of 0.1 mM oligonucleotides was recorded
in the above mentioned buffer on a PTI spectrofluorimeter.
Temperature was controlled and set to 238C with a Peltier
effect device. Spectral data were corrected both in excitation
and emission and recorded at 50 nm min21
.
8.3. Synthesis and purification of modified
oligonucleotides
8.6. Molecular modeling studies
All oligonucleotides were synthesized on a 0.2 mmol scale
on an Expedite 8909 DNA synthesizer using conventional
b-cyanoethyl phosphoramidite chemistry. The P and N
modified phosphoramidites were dissolved in anhydrous
acetonitrile (0.1 M final concentration), and were used with
a coupling time of 15 min. The coupling efficiency was the
same as that of unmodified amidite (.98). The P-containing
oligonucleotides were deprotected by the standard pro-
cedure: standard CPG (P in 50) were treated for 30 min with
AMA (NH4OH/methylamine, 50:50, v/v) (1 ml) at 658C.
Universal supports (GLEN RESEARCH) (P in 30) were
treated overnight with AMA (1 ml) at 558C. The N-contain-
ing oligonucleotides were deprotected using the following
conditions. Standard supports were treated for 4 h at rt in
NH4OH (1 ml), and universal supports were treated for 17 h
at rt with a solution 0.4 M NaOH in MeOH/water (4/1)
(1 ml) and then neutralized by 2 M TEAA (1.5 ml). The
crude oligonucleotides deprotected in AMA or NH4OH
were then evaporated to dryness, and those treated by NaOH
were precipitated in butanol. The crude oligonucleotides
were purified by electrophoresis on denaturing polyacryl-
amide gels. The pure samples were desalted through reverse
phase maxi-clean cartridges (C18, Altech).
Molecular mechanic energies were calculated using the
Amber force field. The simulations were performed with a
distance-dependent dielectric permittivity (1/r) to simulate
the screening induced by solvent molecules.10 Several
initial structures were built as extended, stacked or
intercalated species with the atom charge given by Amber
for hairpin and with atom charges calculated by using the
AM1 semi-empirical method for pyrene and naphtalene
diimide residues. To compare the results, the most stable
structures of each species were retained. The structures were
energy-minimized using 1000 steps of the ‘steepest descent’
¨
algorithm followed by steps of the Polak–Ribiere ‘con-
jugate gradient’ until the energy gradient was smaller than
21
˚
A
0.02 kcal mol21
.
Acknowledgments
´
We thank Dr Jean-Jacques Toulme for helpful discussions.
References
8.4. UV-monitored melting experiments
1. Petersheim, M.; Turner, D. H. Biochemistry 1983, 22,
256–263.
Thermal denaturation of modified oligonucleotides were
monitored on a Cary 1 spectrophotometer interfaced with a
Peltier effect device that controled the temperature to within
^0.18C. Denaturation of the samples (1 mM final concen-
tration) in 20 mM cacodylate buffer, pH 7.3, at 208C
containing 140 mM potassium chloride, 20 mM sodium
chloride and 3 mM magnesium chloride was achieved by
increasing the temperature by 0.58C min21 from 5 to 958C
and was monitored at 260 nm. The melting temperature, Tm,
was taken as the peak of the first derivative of the UV
melting curves. Samples at 1 mM final were boiled for
1 min, kept for 10 min at 48C, and then for 10 min at rt.
The experiment started 30 min later at this temperature
of 58C. At 260 nm a molar extinction coefficient of
2.04£104 mol21 l cm21 was used for P and 4.17£
104 mol21 l cm21 for N.
2. Wang, A. H. J. Nucleic Acids Mol. Biol. 1987, 1, 53–69.
3. Guckian, K. M.; Schweitzer, B. A.; Ren, R. X. F.; Sheils, C. J.;
Tahmassebi, D. C.; Kool, E. T. J. Am. Chem. Soc. 2000, 122,
2213–2222.
4. Nguyen, J. Q.; Iverson, B. L. J. Am. Chem. Soc. 1999, 121,
2639–2640.
5. Winnik, F. M. Chem. Rev. 1993, 93, 587–614.
6. Benhida, R.; Devys, M.; Fourrey, J. L.; Lecubin, F.; Sun, J. S.
Tetrahedron Lett. 1998, 39, 6167–6170.
´
´
7. Duconge, F.; Toulme, J. J. RNA 1999, 5, 1605–1614.
8. Peters, A. T.; Bide, M. J. Dyes Pigment 1985, 6, 349–375.
9. Newcomb, L. F.; Gellman, S. H. J. Am. Chem. Soc. 1994, 116,
4993–4994.
10. Guenot, J.; Kollman, P. A. Protein Sci. 1992, 1, 1185–1205.