Size-expanded analogues of dG and dC: Synthesis and pairing properties in DNA

Article abstract of DOI:10.1021/jo048357z

(Chemical Equation Presented) We describe the completion of the set of four benzo-fused expanded DNA (xDNA) nucleoside analogues. We previously reported the development of benzo-fused analogues of dA and dT and their inclusion in an exceptionally stable n

Full text of DOI:10.1021/jo048357z

Size-Expanded Analogues of dG and dC: Synthesis and Pairing
Properties in DNA
Haibo Liu, Jianmin Gao, and Eric T. Kool*
Department of Chemistry, Stanford University, Stanford, California 94305-5080
kool@stanford.edu
Received September 16, 2004
We describe the completion of the set of four benzo-fused expanded DNA (xDNA) nucleoside
analogues. We previously reported the development of benzo-fused analogues of dA and dT and
their inclusion in an exceptionally stable new four-base genetic system, termed xDNA, in which
the base pairs were expanded in size. Here we describe the preparation and properties of the second
half of this nucleotide set: namely, the previously unknown dxC and dxG nucleosides. The dxC
analogue was prepared from the previously reported dxT nucleoside in three steps and 57% yield.
The large-sized deoxyguanosine analogue was prepared from an intermediate in the synthesis of
dxA, yielding dxG in 14 steps overall (2.4%). Suitably protected versions of the deoxynucleosides
were prepared for oligonucleotide synthesis following standard procedures, and they were readily
incorporated into DNA by automated synthesizer. “Dangling-end” measurements revealed that the
benzo-fused homologues stack considerably more strongly on neighboring DNA sequences than do
their natural counterparts. Base pairing experiments with xC or xG bases showed that they pair
selectively with their Watson-Crick partners, but with mild destabilization, due apparently to
their larger size. Overall, the data suggest that the fluorescent xG and xC bases may be useful
probes of steric effects in the study of biological nucleotide recognition mechanisms. In addition,
the completion of the xDNA nucleoside set makes it possible in the future to construct full four-
base xDNA strands that can target any sequence of natural DNA and RNA.
Introduction
selectivity during replication of the cellular genome.2 Of
course, recognition of chemical differences in the bases
themselves is paramount in such processes, and thus,
steric effects, solvation effects, and hydrogen-bonding
effects could all possibly play roles in this selectivity.3
We have previously explored the importance of hydrogen
bonding and solvation effects by the molecular strategy
The biological recognition and processing of nucleotides
requires efficient and selective binding of nucleobases.
Differential recognition of GTP and ATP, for example,
regulates metabolic energy usage in cells and switching
and signaling in biochemical pathways.1 Similarly, the
four natural deoxynucleotidessdATP, dCTP, dTTP, and
dGTPsmust be distinguished from one another with high
(2) (a) Goodman, M. F.; Creighton, S.; Bloom, L. B.; Petruska, J.
Crit Rev. Biochem. Mol. Biol. 1993, 28, 83-126. (b) Kunkel, T. A.;
Bebenek, K. Annu. Rev. Biochem. 2000, 69, 497-529. (c) Kool, E. T.
Annu. Rev. Biochem. 2002, 71, 191-219.
(1) Alaimo, P. J.; Shogren-Knaak, M. A.; Shokat, K. M. Curr. Opin.
Chem. Biol. 2001, 5, 360-367.
(3) Kool, E. T. Annu. Rev. Biophys. Biomol. Struct. 2001, 30, 1-22.
10.1021/jo048357z CCC: $30.25 © 2005 American Chemical Society
Published on Web 12/23/2004
J. Org. Chem. 2005, 70, 639-647
639

Liu et al.
of using nonpolar isosteres of natural nucleosides,4 and
with them have gained useful insights into mechanisms
of DNA replication and DNA repair enzymes.5,6 However,
a complementary approach to study of such biomolecular
recognition would be to directly test steric effects by
maintaining hydrogen bonding groups, but changing the
sizes of the nucleobases. Here, we describe the completion
of a full set of nucleoside analogues for such an approach.
We recently reported the synthesis of two nucleosides,
dxT and dxA, in which the nucleobases of dT and dA are
expanded in size by benzo-homologation.7 This is a
strategy adapted from early work by Leonard, who
described ATP analogues having lin-benzoadenine as a
base.8 Analogous to the addition of a benzene ring to the
framework of the purine base adenine, we designed a
size-expanded analogue of the pyrimidine T by use of the
bicyclic quinazolinedione skeleton. Studies demonstrated
that the two analogues could be successfully incorporated
into DNA oligonucleotides, and their pairing properties
in natural DNA were explored.9 Finally, we showed that
entire duplexes consisting of size-expanded xA-T and
xT-A pairs could be formed and that these large-
diameter “xDNA” duplexes were more stable than natu-
ral DNAs of the analogous sequence.10
However, that early work left a number of problems
unsolved. First, for universal application of expanded
nucleotide analogues as biochemical tools, one would
need variants of all four bases, while the initial studies
described only two of them, leaving C and G analogues
unexplored. Second, the high thermodynamic stability of
xDNA duplexes suggests the possibility of four-base
xDNA oligomers as efficient targeting agents for natural
sequences of DNA and RNA. However, in previous
studies only two of the four xDNA nucleosides were
reported, making general sequence recognition impos-
sible. Finally, the development of xDNA analogues of C
and G would make it conceptually possible to form a
novel eight-base genetic system using combinations of
four natural and four expanded bases. Thus, the current
study addresses these problems by the preparation and
study of dxC and dxG nucleosides.
Experimental Section
Preparation of dxG Nucleoside.
6-Amino-3,7-dihydroimidazo[4,5-g]quinazolin-8-one
(12). A mixture of 3-acetyl-3,5-dihydroimidazo[4,5-g]quinazo-
line-6,8-dione and 1-acetyl-1,5-dihydroimidazo[4,5-g]quinazo-
line-6,8-dione11 (6.974 g) and Na2CO3 (6.59 g) was suspended
in 100 mL of CH3CN and 25 mL of water and heated at reflux
3 h. After being cooled to room temperature, the reaction
mixture was evaporated and the residue was suspended in 100
mL of water. After the pH was adjusted to 7.0, the mixture
was stored at 4 °C overnight. After filtration, washing with
water, and drying, 3.093 g of product (54%) was obtained: 1H
NMR (DMSO-d6, 500 MHz) δ 8.20 (s, 1H), 8.16 (s 1H), 7.31 (s,
1H); HRMS calcd for C9H8N5O (M + H) 202.0729, found
202.0737.
N6-Isobutyryl-6-amino-3,7-dihydroimidazo[4,5-g]-
quinazolin-8-one (13). To a suspension of 2.732 g of 6-amino-
3,7-dihydroimidazo[4,5-g]quinazolin-8-one 12 in 45 mL of
anhydrous pyridine was added 5.8 mL of isobutyryl chloride.
After being heated at reflux for 4.5 h, the reaction mixture
was cooled to room temperature and evaporated. Fifty milli-
liters of ethanol was added and the solution heated at reflux
for 1.5 h. After ethanol was removed by rotary evaporation,
100 mL of water was added and the pH was adjusted to 7.0.
After storage at 4 °C overnight, 3.014 g of product (82%) was
obtained after filtration and drying: 1H NMR (DMSO-d6, 500
MHz) δ 8.35 (s, 1H), 8.29 (s, 1H), 7.61 (s, 1H), 2.82-2.75 (m,
1H) (d, 1H, Hz); HRMS calcd for C13H14N5O2 (M + H)
272.1148, found 272.1155.
[N6-Isobutyryl-6-amino-3,7-dihydroimidazo[4,5-g]-
quinazolin-8-one]-1′-â-D-ribofuranoside 2′,3′,5′-Triacetate
(14a and N7 Regioisomer 14b). The isobutyryl-protected
base 13 (1.051 g, 3.88 mmol) was dried in vacuo overnight and
was then suspended in 16.0 mL of anhydrous CH3CN. An 11.6
mL portion of N,O-bistrimethylsilyl acetamide was then added
to the reaction mixture. The starting material dissolved under
heating at reflux for 15 min. Then 23.3 mL of a solution of
0.20 mM â-D-ribofuranose 1,2,3,4-tetraacetate in anhydrous
CH3CN was added, followed by addition of trimethylsilyl
triflate (5.8 mL). After the reaction was heated at reflux for 5
h, it was cooled to room temperature, concentrated, and
redissolved in 140 mL ethyl acetate. The organic phase was
washed with water, 5% sodium bicarbonate, and brine (70 mL
each). The organic layer was then dried over sodium sulfate,
concentrated, and purified by silica gel column chromatogra-
phy (0-3-5% MeOH in CH2Cl2). An inseparable ∼1:1 mixture
of N1 and N3 coupling products was obtained in 70% yield
(1.436 g): 1H NMR (CDCl3, 500 MHz) δ 8.69 (s, 0.5H), 8.45 (s,
0.5H), 8.39 (s, 0.5H), 8.32 (s, 0.5H), 7.91 (s, 0.5H), 7.65 (s,
0.5H), 6.20 (d, 0.5H, J ) 6.5 Hz), 6.13 (d, 0.5H, J ) 5.0 Hz),
5.62-5.60 (m, 1H), 5.47-5.41 (m, 1H), 4.51-4.50 (m, 1H),
4.42-4.389 (m, 2H), 2.77 (m, 1H), 2.18-2.07 (m, 9H), 1.25 (d,
6H, J ) 7.0 Hz); 13C NMR (CDCl3, 125 MHz) δ 179.0, 178.7,
170.7, 170.7, 169.9, 169.9, 169.7, 169.6, 162.1, 161.8, 149.6,
145.5, 145.1, 144.9, 144.1, 143.9, 142.7, 138.0, 131.7, 119.6,
117.5, 117.1, 116.5, 108.8, 106.5, 87.6, 87.0, 81.0, 80.5, 73.6,
73.1, 70.8, 70.3, 63.2, 63.0, 36.9, 36.9, 21.2, 21.1, 20.9, 20.8,
20.7, 20.7, 19.3, 19.3; HRMS calcd for C24H28N5O9 (M + H)
530.1887, found 530.1878.
(4) Schweitzer, B. A.; Kool, E. T. J. Org. Chem. 1994, 59, 7238-
7242.
(5) (a) Moran, S.; Ren, R. X.-F.; Rumney, S.; Kool, E. T. J. Am. Chem.
Soc. 1997, 119, 2056-2057. (b) Moran, S.; Ren, R. X.-F.; Kool, E. T.
Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 10506-10511. (c) Morales, J.
C.; Kool, E. T. Nature Struct. Biol. 1998, 5, 950-954. (d) Morales, J.
C.; Kool, E. T. J. Am. Chem. Soc. 1999, 121, 2723-2724. (e) Matray,
T. J.; Kool, E. T. Nature 1999, 399, 704-708. (f) Dzantiev, L.;
Alekseyev, Y.; Morales, J. C.; Kool, E. T.; Romano, L. J. Biochemistry
2001, 40, 3215-3221. (g) Delaney, J. C.; Henderson, P. T.; Helquist,
S. A.; Morales, J. C.; Essigmann, J. M.; Kool, E. T. Proc. Natl. Acad.
Sci. U.S.A. 2003, 100, 4469-4473.
(6) (a) Schofield, M. J.; Brownewell, F. E.; Nayak, S.; Du, C.; Kool,
E. T.; Hsieh, P. J. Biol. Chem. 2001, 276, 45505-8. (b) Drotschmann,
K.; Yang, W.; Brownewell, F. E.; Kool, E. T.; Kunkel, T. A. J. Biol.
Chem. 2001, 276, 46625-9. (c) Begley, T. J.; Haas, B. J.; Morales, J.
C.; Kool, E. T.; Cunningham, R. P. DNA Repair 2002, 2, 107-120. (d)
Washington, M. T.; Helquist, S. A.; Kool, E. T.; Prakash, L.; Prakash,
S. Mol. Cell. Biol. 2003, 23, 5107-5112. (e) Francis, A. W.; Helquist,
S. A.; Kool, E. T.; David, S. S. J. Am. Chem. Soc. 2003, 125, 16235-
16242.
[N6-Isobutyryl-6-amino-3,7-dihydro-imidazo[4,5-g]-
quinazolin-8-one]-1′-â-D-ribofuranoside (regioisomers 15a
and 15b). The mixture of isomers 14a and 14b (1.420 g, 2.68
mmol) was dissolved in 30.0 mL anhydrous MeOH, to which
was added 10.0 mL of 2.0 M ammonia in MeOH. After stirring
at room temperature for 7.5 h, the reaction was evaporated.
The residue was suspended in ethyl acetate, filtered, washed
with EtOAc, and dried in vacuo, which furhished 0.846 g (95%)
(7) Liu, H.; Gao, J.; Maynard, L.; Saito, Y. D.; Kool, E. T. J. Am.
Chem. Soc. 2004, 126, 1102-1109.
(8) Bommarito, S.; Peyret, N.; SantaLucia, J., Jr. Nucleic Acids Res.
2000, 28, 1929-1934.
1
free dxG (2). This crude product was essentially pure by H
NMR. Without further purification, this intermediate was
(9) Gao, J.; Liu, H.; Kool, E. T. J. Am. Chem. Soc. 2004, 126, 11826-
11831.
(10) (a) Liu, H.; Gao, J.; Lynch, S. R.; Saito, Y. D.; Maynard, L.;
Kool, E. T. Science 2003, 302, 868-871. (b) Liu, H.; Gao, J.; Kool, E.
T. J. Am. Chem. Soc. In press.
(11) Leonard, N. J.; Sprecker, M. A.; Morrice, A. G. J. Am. Chem.
Soc. 1976, 98, 3987-3994.
640 J. Org. Chem., Vol. 70, No. 2, 2005

Expanded Analogues of dG and dC
subjected to N6 protection. An 840 mg (2.54 mmol) portion of
this intermediate was coevaporated with anhydrous pyridine
(3 × 5 mL) and then was suspended in 30.0 mL of anhydrous
pyridine, followed by addition of TMSCl (4.2 mL, 32.6 mmol).
After 30 min of stirring, the reaction solution was treated with
isobutyric anhydride (0.83 mL, 5.27 mmol). After being stirred
at room temperature for 24 h, the reaction was cooled to 0 °C
and treated with 5.0 mL of water. After 5 min, 5.0 mL of 25%
ammonium hydroxide was added. After an additional 15 min
stirring at 0 °C, volatiles were removed by evaporation. The
product was purified in nearly quantitative yield by silica gel
column chromatography (10-20% MeOH in CH2Cl2): 1H NMR
(CD3OD, 500 MHz) δ 8.78 (s, 0.4H), 8.70 (s, 0.6H), 8.54 (s,
0.4H), 8.49 (s, 0.6H), 7.83 (s, 0.6H), 7.81 (s, 0.4H), 6.06 (d, 0.4H,
J ) 7.0 Hz), 6.02 (d, 0.6H, J ) 7.5 Hz), 4.51-4.48 (m, 1H),
4.32-4.29 (m, 1H), 4.18-4.15 (m, 1H), 3.93-3.87 (m, 1H),
3.84-3.79 (m, 1H), 2.75-2.72 (m, 1H), 1.24 (d, 1H, J ) 8.5
Hz); 13C NMR (DMSO-d6, 125 MHz) δ 181.0, 180.9, 149.8,
147.7, 146.7, 142.6, 138.9, 132.1, 117.5, 116.6, 116.3, 109.3,
89.6, 86.48, 86.2, 84.1, 74.7, 74.4, 70.8, 70.7, 61.9, 61.7, 35.5,
19.7; HRMS calcd for C18H22N5O6 (M + H) 404.1570, found
404.1575.
3′,5′-Tetraisopropyldisiloxane-Protected [N6-Isobutyryl-
6-amino-3,7-dihydroimidazo[4,5-g]quinazolin-8-one]-1′-
â-D-ribofuranoside (Regioisomers 16a and 16b). The
amine-protected riboside mixture 15a,b (496 mg, 1.23 mmol)
and imidazole (300 mg) were dissolved in 20.0 mL of anhydrous
DMF, to which was added 481 µL (1.48 mmol) of 1,3-dichloro-
1,1,3,3-tetraisopropylsiloxane. After being stirred at room
temperature for 24 h, the solvent was removed by evaporation.
The residue was dissolved in 250 mL of ether. After being
washed with 125 mL of water and 125 mL of aqueous
saturated NaHCO3, the organic phase was dried over NaSO4,
filtered, concentrated, and purified by silica gel column chro-
matography (50-100% EtOAc in hexanes). A white solid was
obtained as product in 71% yield (560 mg): 1H NMR (CDCl3,
500 MHz) δ 12.00 (s, 0.5H), 11.93 (s, 0.5H), 8.75 (s, 0.5H), 8.47
(s, 0.5H), 8.42 (s, 0.5H), 8.39 (s, 0.5H), 7.95 (s, 0.5H), 7.54 (s,
0.5H), 6.03 (s, 0.5H), 5.99 (s, 0.5H), 4.60-4.55 (m, 1H), 4.34-
4.20 (m, 3H), 4.16-4.12 (m, 1H), 3.32 (s, 1H), 3.129 (s, 1H),
2.72-2.67 (m, 1H), 1.34-1.27 (m, 6H), 1.14-0.99 (m, 28H);
13C NMR (CDCl3, 125 MHz) δ 170.7, 170.6, 170.0, 169.9, 169.8,
169.7, 149.3, 145.3, 144.0, 142.5, 137.9, 131.5, 119.4, 117.2,
116.7, 108.9, 87.5, 87.1, 80.9, 80.4, 73.5, 73.2, 70.7, 70.2, 63.2,
62.9, 36.6, 21.1, 21.0, 20.9, 20.8, 20.7, 20.7, 19.3, 19.3; HRMS
calcd for C30H47N5O7Si2 (M + H) 646.3087, found 646.3082.
3′,5′-Tetraisopropyldisiloxane-Protected [N6-Isobutyryl-
6-amino-3,7-dihydroimidazo[4,5-g]quinazolin-8-one]-1′-
â-D-Ribofuranoside-2′-O-thiocarbonate Phenyl Ester (17).
Starting material 16a,b (859 mg, 1.33 mmol) and DMAP (974
mg, 7.98 mmol) were dissolved in 40 mL of anhydrous CH3CN,
to which was added phenyl chlorothionoformate (368 µL, 2.66
mmol). After being stirred at room temperature 24 h, the
reaction mixture was evaporated to dryness. The residue was
dissolved in 120 mL of EtOAc, and the organic phase was
washed with saturated NaHCO3 (60 mL) and brine (60 mL).
After being dried over NaSO4, the organic layer was concen-
trated and purified by silica gel column chromatography (25-
50-67% EtOAc in hexanes). Separation of imidazole regio-
isomers was successful in this step. The desired product (17)
was obtained as a pale yellow foamy solid in 50% yield (516
mg): 1H NMR (CDCl3, 500 MHz) δ 11.99 (s, 1H), 8.67 (s, 1H),
8.72 (b, 1H), 8.51 (s, 1H), 7.59 (s, 1H), 7.463 (t, 2H, J ) 7.5
Hz), 7.35 (t, 1H, J ) 7.0 Hz), 7.179 (d, 2H, J ) 8.0 Hz), 6.21
(s, 1H), 6.03 (d, 1H, 7.0 Hz), 4.84-4.81 (m, 1H), 4.36-4.33 (m,
1H), 4.28-4.26 (m, 1H), 4.26-4.12 (m, 1H), 2.67-2.65 (m, 1H),
1.28 (d, 6H, J ) 7.0 Hz), 1.16-1.00 (m, 28H); 13C NMR (CDCl3
125 MHz) δ 194.1, 178.6, 162.2, 153.6, 145.38, 144.9, 143.6,
142.8, 137.8, 129.9, 127.2, 121.9, 121.2, 119.6, 117.0, 105.9,
87.8, 84.2, 82.7, 68.6, 60.7, 60.2, 36.9, 19.3, 17.7, 17.6, 17.6,
17.6, 17.3, 17.20, 17.2, 17.0, 14.5, 13.64, 13.2, 13.0; HRMS calcd
for C37H52N5O8Si2S (M + H) 782.3070, found 782.3058.
The other regioisomer was eluted from the silica column
with 100% EtOAc in 21% yield (214 mg).
3′,5′-Tetraisopropyldisiloxane-Protected [N6-Isobutyryl-
6-amino-3,7-dihydroimidazo[4,5-g]quinazolin-8-one]-1′-
â-D-2′-deoxyribofuranoside (18). Starting material 17 (480
mg, 0.61 mmol) and AIBN (45 mg) were dissolved in 18.0 mL
of toluene, to which was added 2.0 mL of n-Bu3SnH. After
being heated at reflux for 2 h, the reaction mixture was cooled.
Volatiles were removed by evaporation, and compound 18 was
purified by silica gel column chromatography (50-75% EtOAc
in hexanes). The product was obtained as a white foamy solid
in 71% yield (276 mg): 1H NMR (CDCl3 500 MHz) δ 12.04 (s,
1H), 8.75 (s, 1H), 8.35 (s, 1H), 7.47 (s, 1H), 6.24 (dd, 1H, J )
3.5, 7.0 Hz), 4.73-4.72 (m, 1H), 4.12-3.97 (m, 3H), 2.70-2.65
(m, 2H), 2.61-2.59 (m, 1H), 1.32 (d, 6H, J ) 6.5 Hz), 1.12-
0.99 (m, 28H); 13C NMR (CDCl3 125 MHz) δ 178.5, 162.2, 145.3,
144.5, 143.8, 142.9, 138.1, 119.5, 116.6, 85.6, 84.1, 69.9, 61.8,
40.2, 36.9, 19.3, 17.8, 17.7, 17.6, 17.4, 17.3, 17.1, 13.7, 13.4,
13.2, 12.7; HRMS cald. for C30H48N5O6Si2 (M + H) 630.3138,
found 630.3182.
[N6-Isobutyryl-6-amino-3,7-dihydroimidazo[4,5-g]-
quinazolin-8-one]-1′-â-D-2′-deoxyribofuranoside (19). Start-
ing material 18 (276 mg, 0.44 mmol) was dissolved in 12.0 mL
of anhydrous THF and 4.0 mL of anhydrous pyridine, to which
was added 70% HF in pyridine (350 µL). After the reaction
mixture was stirred at room temperature for 80 h, it was
concentrated and purified by silica gel column chromatography
(10-20% MeOH in CH2Cl2). The product was obtained as a
white solid in essentially quantitative yield: 1H NMR (CD3OD,
500 MHz) δ 11.98 (s, 1H), 11.51 (s, 1H), 8.68 (s, 1H), 8.33 (s,
1H), 7.71 (s, 1H), 6.45 (t, 1H, J ) 7.0 Hz), 5.41 (d, 1H, J ) 4.0
Hz), 5.01 (t, 1H, J ) 5.0 Hz), 4.44-4.42 (m, 1H), 3.89 (q, 1H,
J ) 4.0 Hz), 3.61-3.54 (m, 2H), 2.81-2.78 (m, 1H), 2.67-2.62
(m, 1H), 2.37-2.33 (m, 1H), 1.13 (d, 6H, J ) 6.5 Hz); 13C NMR
(DMSO-d6, 125 MHz) δ 181.0, 161.6, 146.5, 146.2, 144.9, 142.6,
139.0, 117.5, 116.2, 107.1, 88.3, 85.4, 71.1, 62.1, 35.5, 19.7;
HRMS calcd for C18H22N5O5 (M + H) 388.1621, found 388.1631.
Base-Protected Phosphoramidite Derivative of 19.
Synthetic details and characterization are given in the Sup-
porting Information.
Synthesis of dxC Nucleoside.
3′,5′-Diacetyl-1′-â-[8-(6-methylquinazoline-2, 4-dione)]-
D-2′-deoxyribofuranoside (4). 1′-â-[8-(6-Methylquinazoline-
2, 4-dione)]-2′-D-deoxyribofuranose 37 (200 mg, 0.69 mmol) was
dissolved in 5 mL of dry pyridine. To the solution was added
200 µL of acetic anhydride. The mixture was stirred at room
temperature for 8 h. Volatiles were removed under vacuum,
and crude product was purified by silica column chromatog-
raphy (hexanes/ethyl acetate 1:1 followed by hexanes/ethyl
acetate 1:2) to give the product as a white foam (195 mg,
76%): 1H NMR (CDCl3, 500 MHz), δ) 9.74 (S, 1H), 9.39 (s,
1H), 7.93 (s, 1H), 7.28 (s, 1H), 5.26 (m, 2H), 4.52 (m, 1H), 4.33
(m, 2H), 2.38 (s, 3H), 2.40-2.29 (m, 2H), 2.21 (s, 3H), 2.15 (s,
3H); 13C NMR (CDCl3, 125.7 MHz) δ 171.3, 170.9, 163.3, 150.2,
136.5.06, 133.0, 128.3, 123.9, 116.0, 83.9, 80.7, 75.6, 64.0, 39.7,
21.3, 21.2, 20.8; HRMS (EI+) calcd for C18H21N2O7 [M + H]+
m/z ) 377.1349, found 377.1349.
3′,5′-Diacetyl-1′-â-[8-(6-methyl-4-thio-2-quinazolone)]-
D-2′-deoxyribofuranoside (5). Compound 4 (70 mg, 0.19
mmol) and phosphorus pentasulfide (211 mg, 0.95 mmol) were
charged into a round-bottom flask equipped with condenser.
Four milliliters of dry pyridine was added to form a suspension,
which turned into a clear solution upon heating to reflux. TLC
showed the reaction was complete after 4 h. Solvent was
removed under vacuum, and the residue was suspended in 15
mL of warm water. The aqueous suspension was extracted
with 3 × 20 mL dichloromethane. The organic layer was
collected, dried, and concentrated. Silica column chromatog-
raphy (hexanes/ethyl acetate 1:1) gave product 5 as yellow
foam (56 mg, 77%): 1H NMR (CDCl3, 500 MHz) δ 10.74 (s,
1H), 9.62 (s, 1H), 8.33 (s, 1H), 7.29 (s, 1H), 5.25 (m, 2H), 4.52
(m, 1H), 4.34 (m, 2H), 2.40 (s, 3H), 2.36 (m, 1H), 2.30 (m, 1H),
J. Org. Chem, Vol. 70, No. 2, 2005 641

Liu et al.
2.21 (s, 3H), 2.16 (s, 3H); 13C NMR (CDCl3, 125.7 MHz) δ 191.9,
171.2, 170.8, 147.0, 135.5, 133.9, 133.6, 131.8, 123.9, 122.0,
84.0, 80.7, 75.5, 63.9, 39.7, 21.3, 21.2, 20.96; HRMS (EI+) calcd
for C18H21N2O6S [M + H]+ m/z 393.1120, found 393.1112.
and free energy values for the complexes. The free energies
were also calculated from van’t Hoff plots by plotting 1/Tm vs
ln(C/4); in all cases, close agreement was observed, indicating
that the two-state approximation is reasonable for these
specific sequences.
1′-â-[8-(4-Amino-6-methyl-2-quinazolone)]-D-2′-deoxy-
ribofuranoside (1). In a high-pressure reaction tube, com-
pound 5 (106 mg, 0.27 mmol) was dissolved in 10 mL of
anhydrous methanol and chilled to 0 °C with an ice-water
bath. Ammonia gas was bubbled into the solution for 1 min.
The tube was sealed and heated to 100 °C in an oil bath. The
reaction mixture was stirred at 100 °C for 8 h and then was
allowed to cool to room temperature. Methanol was evaporated
under vacuum, and residue was suspended in 10 mL of
dichloromethane. The solid was collected by filtration to give
product 1 as pale solid (77 mg, 97%): 1H NMR (DMSO-d6, 500
MHz) δ 9.80 (s, broad, 1H), 7.79 (s, 1H), 7.9-7.5 (broad peak,
2H), 7.45 (s, 1H), 5.27 (dd, 1H, J ) 11 Hz, 5 Hz), 5.13 (s, broad,
1H), 4.25 (m, 1H), 3.86 (m, 1H), 3.58 (m, 2H), 2.31 (s, 3H),
2.09 (m, 1H), 1.96 (m, 1H); 13C NMR (DMSO-d6, 125.7 MHz)
δ 164.6, 156.4, 138.7, 133.8, 130.2, 126.6, 124.2, 109.3, 88.73,
78.6, 72.9, 62.2, 42.1, 21.1; HRMS (EI+) calcd for C14H18N3O4
[M + H]+ m/z 292.1297, found 292.1295.
Results
Design Principles. We chose a 4-amino-6-methyl-2-
quinazolone C-nucleoside as a size-expanded analogue of
deoxycytidine. This compound was previously unknown.
A 4-amino-2-quinazolone nucleoside with the glycosidic
attachment on N1 has been reported.12 However, it was
not incorporated into oligonucleotides, and in any case,
with attachment at N1 of the heterocycle, the reported
nucleoside would not have an expanded pairing di-
mension. Space-filling models of 4-amino-6-methyl-2-
quinazolone with electrostatic potentials showed (Figure
1) that it is expected to have similar electrostatic charges
along the Watson-Crick analogous hydrogen-bonding
edge as cytosine, suggesting its capability for forming
hydrogen bonds with guanine.
In a completely analogous way to the design of dxT
nucleoside,7 attachment of 2′-deoxyribose to the C8 of
4-amino-6-methyl-2-quinazolone would give the expanded
cytidine analogue dxC (Figure 1). Compared with the
structure of cytidine, dxC has an extra methyl group
attached to the C6 of 2-quinazolone. Thus, rigorously
speaking, the xC base is an expanded analogue of
5-methylcytosine, rather than cytosine. We decided to
retain the methyl group in our molecular design for ease
of synthesis from an intermediate in the xT preparation
(see below).
Analogous to xT base, the xC base has as number of
possible tautomers as well. Preliminary semiempirical
calculations of heats of formation suggested that the
correct tautomer for pairing with guanine, at least in the
gas phase, was the most stable one (Supporting Informa-
tion). Notably, a second tautomer, an imino variant of
the exocyclic amine, was predicted to be less stable by a
small margin of 1.3 kcal/mol, raising a possible issue of
selectivity in pairing. In fact, further study showed that
xC is highly selective in pairing with G (see below),
indicating that this imino tautomer, if it does exist in
aqueous solution, does not interfere with the pairing
properties of xC.
The design of the xG base was a straightforward
adaptation of the structure of the previously known xA
base. The deoxyriboside derivative of xG (dxG, Figure 1)
was not previously known in the literature, although the
ribose derivative was reported by Leonard 25 years ago.13
As with the xA base, the xG base is larger than G by
benzo-homologation and is longer by approximately 2.4
Å. A space-filling model is shown in Figure 1, and the
calculated electrostatic potentials along the Watson-
Crick-analogous edge are essentially the same as those
on guanine.
Base-Protected Dimethoxytrityl Phosphoramidite De-
rivative of 1. Synthetic details and characterization are given
in the Supporting Information.
Oligonucleotide Synthesis. 5′-O-Dimethoxytritylated phos-
phoramidite analogues of nucleosides were synthesized as
described above and in the Supporting Information. Oligo-
deoxynucleotides were synthesized on a 1.0 µmol scale on a
DNA synthesizer using standard â-cyanoethyl phosphor-
amidite chemistry for DNA synthesis, but with extended (120
s) coupling times for modified nucleosides. Stepwise coupling
yields for nonnatural nucleosides were all greater than 95%
as determined by trityl cation monitoring. Oligomers were
deprotected in concentrated ammonium hydroxide (55 °C, 16
h), purified by preparative 20% denaturing polyacrylamide gel
electrophoresis, and isolated by excision and extraction from
the gel, followed by dialysis against water. The recovered
material was subsequently quantified by absorbance at 260
nm with molar extinction coefficients determined by the
nearest neighbor method. Values for oligomers containing
modified nucleotides were estimated by measuring the molar
extinction coefficients of modified nucleosides at 260 nm and
adding these values to the calculated values of natural DNA
fragments. Extinction coefficients of dxG and dxC at 260 nm
were measured at 8060 and 5800 cm-1‚M-1, respectively. Intact
incorporation of modified nucleotides was confirmed by char-
acterization of short unpurified oligomers (sequence TXT) by
1H NMR (see the spectra in the Supporting Information).
These were also characterized by ESI-MS: calcd for TxGT
(C34H42N9O18P2) [M + H] 926.2, found 926.1; calcd for TxCT
(C34H42N7O18P2) [(M - H)-] m/z ) 898.2, found 898.2.
Thermal Denaturation Studies. Solutions for the ther-
mal denaturation studies contained either a 1:1 ratio of two
complementary oligomers or only a self-complementary oligo-
mer. Total oligomer concentrations ranged from 2.0 to 30 µM.
The buffer contained NaCl (100 mM), MgCl2 (10 mM), and
Na‚PIPES(10 mM), at pH ) 7.0, except where noted otherwise.
After solutions were prepared, they were heated to 95 °C for
10 min and allowed to cool to room temperature over at least
2 h and then cooled at 5 °C for at least 4 h. Melting studies
were carried out in Teflon-stoppered 1 cm path length quartz
cells (under nitrogen atmosphere when temperature was below
20 °C) on a UV-vis spectrophotometer equipped with a
thermoprogrammer and temperature control. Absorbance was
monitored while the temperature was changed at a rate of 0.5
°C/min. Melting data were essentially the same for heating
or cooling at this rate. Experiments were monitored at 260
nm. In most cases, the complexes displayed apparent two-state
transitions, with all-or-none melting curves from bound duplex
to free single strands. Computer fitting of the melting data
using Meltwin 3.0b provided both melting temperatures Tm
The xG base has a number of possible tautomeric
forms. Preliminary heat-of-formation calculations sug-
gested, however (see Supporting Information), that the
desired guanine-like tautomer should be the energetically
(12) Stout, R. G.; Robins, R. K. J. Org. Chem. 1968, 33, 1219-
1225.
(13) Leonard, N. J.; Keyser, G. E. Proc. Natl. Acad. Sci. U.S.A. 1979,
76, 4262-4264.
642 J. Org. Chem., Vol. 70, No. 2, 2005

Expanded Analogues of dG and dC
FIGURE 1. Structures of size-expanded (xDNA) homologues of cytosine and guanine. (A) Structures of the benzo-homologated
deoxyribosides, with natural dC and dG nucleosides shown for comparison. (B) Space-filling models of the free bases xC, C, xG,
and G, respectively, with electrostatic potentials mapped on the surfaces. Geometry optimization and electrostatics calculated by
AM1 using Spartan (Wavefunction, Inc). Colors depict an electrostatic scale of -70 to +50 (red ) negative potential, blue )
positive). (C) Proposed structures of expanded base pairs formed between xC or xG with the Watson-Crick complements.
most stable one relative to two other tautomers, by
relatively large margins of 7-10 kcal/mol.
The expanded deoxyguanosine nucleoside, dxG (2), and
its protected phosphoramidite derivative 21 were pre-
pared as shown in Scheme 2. Transformations from 9 to
11 followed reported procedures.13 In converting 11 to the
free xG base 12 we found it necessary to modify the
reported reaction conditions as NH2CN and t-BuOK gave
unsatisfactory results in our hands. The use of S-
methylisothiouronium sulfate resulted in higher yields.
The exocyclic amine of 12 was then protected with
isobutyryl chloride to give 13 in 82% yield.
The coupling reaction between 13 and tetraacetyl-d-
ribose was carried out in the presence of N,O-bistri-
methylsilyl acetamide and trimethylsilyl triflate in re-
fluxing acenonitrile to afford compounds 14 in 70% yield
as an inseparable mixture of regiosiomers. Treatment of
14 with 2.0 M ammonia in methanol followed by selective
protection of the amine group on the base furnished
isobutyryl-protected dxG nucleoside mixture 15 in 95%
overall yield. Selective protection of the 3′- and 5′-
hydroxyl groups was achieved via 1,3-dichlorotetraiso-
propyl siloxane to give compounds 16 in 71% yield. The
free 2′-OH of 16 was then removed following a two-step
dehydroxylation protocol.14 The first step involved trans-
formation of the free hydroxyl group of 16 to the thio-
carbonate 17 in 50% yield. It was at this stage that the
two regioisomers could be separated by column chroma-
tography. The desired regioisomer was then transformed
to the deoxyribose derivative 18 with tributyltin hydride
in 71% yield. The structures of the isolated regioisomers
Synthesis of dxC and dxG Nucleoside Homo-
logues. Synthesis of dxC proved to be quite straightfor-
ward (see Scheme 1) and proceeded from the previously
reported quinazolinedione intermediate 3.7 The 5′- and
3′-hydroxyls were acylated with acetic anhydride in
pyridine to give the protected nucleoside 4. Regioselective
conversion of C4 carbonyl to thiocarbonyl was accom-
plished by treating compound 4 with phosphorus penta-
sulfide in refluxing pyridine.12 The reaction was closely
monitored to prevent overreaction to give the 2,4-dithio-
quinazolone. The regioselectivity of the sulfurization
reaction was confirmed by heteronuclear multibond
coherence (HMBC) experiments with compounds 4 and
5 (see details in the Supporting Information).
Methanolic ammonia treatment of compound 5 con-
verted the 4-thio group to 4-amino and removed the
diacetyl protecting groups on the sugar, yielding the dxC
free nucleoside 1 in high yield. In preparation for oligo-
nucleotide synthesis, the 4-amino group on quinazolone
was protected with a formamidine group to give com-
pound 6, which was subsequently 5′-dimethoxytritylated
and phosphitylated to afford the desired dxC-phosphor-
amidite 8 in very good yield.
Experiments showed that dxC phosphoramidite 8 was
compatible with standard oligodeoxynucleotide synthesis
protocols. A short test oligonucleotide synthesis (sequence
d(T-xC-T)) showed that good coupling yield (93%) can
be obtained. The product was characterized by NMR and
mass spectrometry for its chemical integrity and purity
(Supporting Information).
(14) Barton, D. H. R.; Ferreira, J. A.; Jaszberenyi, J. C. Prepr.
Carbohydr. Chem. 1997, 151-172.
J. Org. Chem, Vol. 70, No. 2, 2005 643

Liu et al.
SCHEME 1a
a Reagents and conditions: (a) Ac2O, pyridine, 76%; (b) P2S5, pyridine, reflux, 77%; (c) methanolic ammonia, 100 °C, 97%; (d) N,N-
dimethylformamide dimethyl acetal, pyridine, 82%; (e) 4,4′-dimethoxytrityl chloride, DIPEA, pyridine, 78%; (f) 2-cyanoethyl N,N-
diisopropylchlorophosphoramidite, DIPEA, CH2Cl2, 86%.
were identified through NMR experiments. First, HMBC
experiments were used to identify the benzo-group
hydrogens proximal and distal to the ribose in the two
isomers. With this information in hand, we were then
able to identify the two regioisomers based upon NOE
measurements (Supporting Information).
dxG free nucleoside (2) has a long wavelength absorption
at 320 nm (ꢀ ) 3400) and gave fluorescence emission at
413 nm.
Pairing and Stacking Properties in the Context
of Natural DNA. The initial discovery that the first size-
expanded analogue, xA, pairs selectively with T in the
context of DNA9 suggests that DNA backbones are
moderately flexible and that recognition between xA and
T through hydrogen bonding may exist. In the current
study we wished to see if this is generally true for all
size-expanded base analogues, as a general test of DNA
backbone flexibility. Moreover, all four expanded bases
are fluorescent and might conceivably report on matched
or mismatched bases at specific sites by fluorescence
changes. For these reasons we tested pairing stability and
selectivity of the new analogues xC and xG in the context
of natural DNA duplexes.
The pairing properties of nucleosides dxC and dxG
were studied in 12mer duplex DNAs by UV monitored
thermal melting experiments in a pH 7.0 buffer contain-
ing 100 mM NaCl, 10 mM MgCl2, and 10 mM Na‚PIPES.
The same 12 base-pair context used previously for dxA
and dxT was employed9 so that direct comparisons could
be made. To test the effects of a change in nearest
neighbor bases, we tested each expanded base in either
a purine strand or a pyrimidine strand of the same
duplex. The pairing data are given in Table 1.
The synthesis was continued with the single regio-
isomer 18. Removal of the disiloxane protecting group
with pyridine‚HF proceeded in nearly quantitative yield
to give base-protected deoxynucleoside 19; this could be
deprotected with ammonia to yield free nucleoside 2 for
characterization of fluorescence (see below). The iso-
butyryl protected compound 19 was then converted to the
desired phosphoramidite 21 by DMT protection (93%
yield) and phosphoramidite formation (77% yield).
As for dxC, the dxG nucleoside phosphoramidite de-
rivative 21 could be readily incorporated into oligo-
deoxynucleotides by standard automated protocols. A
short test oligomer d(T-xG-T) was prepared and char-
acterized by NMR and mass spectrometry (Supporting
Information), which confirmed intact incorporation of the
homologue of deoxyguanosine.
Fluorescence of dxC and dxG. Experiments showed
that the free nucleoside analogues 1 and 2 were fluores-
cent under illumination by ultraviolet light. Excitation
and emission spectra of the two compounds in methanol
are shown in Figure 2. The dxC nucleoside (1) has, in
addition, to short wavelength absorptions, a longer
wavelength absorption band at 330 nm (ꢀ ) 4100).
Excitation at this wavelength resulted in an emission
band having a maximum at 388 nm. The quantum yield
for fluorescence was measured to be 0.52. Similarly, the
Results showed that, when paired opposite their
Watson-Crick partners, both expanded bases destabi-
lized the duplex by a small amount as compared to the
normally sized base pairs. For example, the duplex
containing an xC-G pair was less stable than the control
644 J. Org. Chem., Vol. 70, No. 2, 2005

Expanded Analogues of dG and dC
SCHEME 2a
a Reagents and conditions: (a) KMnO4, 75 °C, 55%; (b) Ac2O, reflux, 93%; (c) TMSN3, 95 °C; (d) S-methylisothiouronium sulfate, H2O/
THF 1:9, 54% two steps; (e) isobutyryl chloride, Py, reflux, 82%; (f) N,O-Bistrimethylsilylacetamide, trimethylsilyl triflate, ribofuranose
tetraacetate, AcCN, reflux, 70%; (g) (i) 2.0 M NH3 in MeOH, 7.5 h, 95%; (ii) TMSCl, pyridine then isobutyric anhydride, 24 h, 100%; (h)
1,3-dichlorotetraisopropylsiloxane, imidazole, 64%; (i) phenyl chlorothioformate, DMAP, 44%; (j) Bu3SnH, AIBN, PhMe, reflux, 84%; (k)
Py‚HF, THF, 93%; (l) DMTrCl, Py, 93%; (m) 2-cyanoethyl diisopropylchlorophosphoramidite, DIPEA, 77%.
(containing C-G) by a small 1.9 °C in thermal melting
temperature (Tm) and 0.3 kcal/mol in free energy (Table
1). Changing the sequence context (to G-xC) gave a very
similar result. By comparison, the duplexes containing
xG pairs were slightly more destabilized; for example,
the C-xG pair destabilized the duplex by 6.3 °C in Tm
and 1.3 kcal/mol. These values are quite close to those
observed previously for the earlier expanded analogues
xT and xA.9
xC base showed the greatest sequence selectivity against
mismatches, with drops in Tm of 11.3-18.0 °C and losses
of favorable free energy of 2.7-3.8 kcal/mol, as compared
with the matched pairs of xC with G (Table 1). By
comparison, the selectivity of xG was lower, with drops
in Tm of 7.2-11.7 °C. By this measure, the selectivities
of xG are similar in magnitude to those previously
measured for xT and xA,9 while those of xC are consider-
ably higher than the other three expanded bases. To test
the relative destabilizations of the mismatched cases, we
also paired the expanded bases with a tetrahydrofuran
The pairing data also revealed selectivities of the
expanded bases in these two 12mer duplex contexts. The
J. Org. Chem, Vol. 70, No. 2, 2005 645

Liu et al.
TABLE 1. Thermodynamic Data for 12mer DNA
Duplexes Containing a Size-expanded Base Pair or
Mismatch (X-Y) in a Central Positionab
base pair X-Y
Tmc (°C)
φG°37d (kcal/mol)
C-G
43.1 ( 0.5
45.6 ( 0.5
41.2 ( 0.5
25.0 ( 0.5
29.9 ( 0.5
28.5 ( 0.5
26.5 ( 0.5
41.1 ( 0.5
23.1 ( 0.5
26.4 ( 0.5
26.0 ( 0.5
21.8 ( 0.5
43.1 ( 0.5
45.6 ( 0.5
36.0 ( 0.5
25.9 ( 0.5
28.7 ( 0.5
28.8 ( 0.5
28.3 ( 0.5
40.3 ( 0.5
28.6 ( 0.5
32.2 ( 0.5
31.1 ( 0.5
32.8 ( 0.5
-9.7 ( 0.8
-10.4 ( 0.8
-9.4 ( 0.2
-5.9 ( 0.4
-6.6 ( 0.2
-6.7 ( 0.2
-6.0 ( 0.4
-9.4 ( 0.2
-5.6 ( 0.2
-6.0 ( 0.4
-5.9 ( 0.2
-5.4 ( 0.2
-9.7 ( 0.9
-10.4 ( 1.0
-8.2 ( 0.8
-6.1 ( 0.6
-7.0 ( 0.7
-6.3 ( 0.6
-6.4 ( 0.6
-9.3 ( 0.9
-6.6 ( 0.6
-7.2 ( 0.7
-7.1 ( 0.7
-7.3 ( 0.7
G-C
xC-G
xC-A
xC-C
xC-T
xC-φe
G-xC
A-xC
C-xC
T-xC
ø-xC
C-G
G-C
xG-C
xG-A
xG-G
xG-T
xG-φ
C-xG
A-xG
G-xG
T-xG
ø-xG
a Conditions: 100 mM NaCl, 10 mM MgCl2, 10 mM PIPES‚Na
(pH 7.0). b Sequence is 5′-AAGAAXGAAAAG‚5′-CTTTTCYTTCTT.
c Tm values are for 5.0 µM oligonucleotide. d Averages of values
from van’t Hoff and curve fitting methods. e “ø” is tetrahydrofuran
abasic analogue.
FIGURE 2. Fluorescence of the expanded-size free deoxy-
nucleosides in methanol. (A) Excitation and emission spectra
of dxC nucleoside. Excitation for the emission spectrum (solid
line) was at 330 nm; the excitation spectrum (dashed line) was
monitored at 388 nm. (B) Excitation and emission spectra of
dxG nucleoside. Excitation for the emission spectrum (solid
line) was at 330 nm; the excitation spectrum (dashed line) was
monitored at 413 nm.
TABLE 2. Thermodynamic Data for Self-
Complementary Duplexes Containing a 5′ Dangling
Basea
c
d
∆G°37
∆∆G°37
dangling base
Tmb (°C)
(kcal/mol)
(kcal/mol)
abasic analogue15 (Table 1). Results showed that xC or
xG paired opposite this abasic site yielded stabilities
about the same as the least stable mismatches, suggest-
ing that most of the mismatched bases contribute a small
degree of stabilization by remaining stacked in the
duplex. Interestingly, for both xC and xG, the least stable
mismatch was with A.
Since the stability of a pair depends heavily on the
stacking properties of the constituent bases,16 we explic-
itly measured stacking properties for xC and xG in the
context of short self-complementary sequences designed
to form duplexes with a single overhanging base, puta-
tively stacked at the end. The data are shown in Table
2, with comparison to the same duplexes lacking an
overhanging base. In this system, the added stabilization
is expected to arise from stacking of the overhanging base
on a neighboring base pair.16 The results show that both
xC and xG in an overhanging end stabilized their core
core sequence 5′-dCGCGCG
(none)
Ce
41.7 ( 0.5
-8.1 ( 0.2
-9.0 ( 0.2
-9.6 ( 0.3
-10.1 ( 0.3
46.2 ( 0.5
51.5 ( 0.5
53.6 ( 0.5
-0.5 ( 0.2
-0.8 ( 0.3
-1.0 ( 0.3
Ge
xC
core sequence 5′-dACAGCTGT
(none)
C
40.0 ( 0.5
42.8 ( 0.5
44.7 ( 0.5
48.9 ( 0.5
-7.9 ( 0.1
-8.7 ( 0.1
-8.9 ( 0.1
-10.2 ( 0.1
-0.4 ( 0.2
-0.5 ( 0.2
-1.2 ( 0.2
G
xG
a Conditions: 1 M NaCl, 10 mM phosphate (pH 7.0) with 0.1
mM EDTA. b Tm values are for 5.0 µM oligonucleotide. c Averages
of values from van’t Hoff and curve fitting methods. d Values per
each dangling base, relative to the core duplex. e Data from ref
15d.
duplexes more than their nonexpanded counterparts did.
The xG base stabilized the duplex by 1.2 kcal/mol per
substitution, while xC stabilized it by a slightly smaller
1.0 kcal/mol. Thus, combining the data with previous
measurements,9 the relative stacking abilities of the four
expanded bases are xA (2.2 kcal/mol), xT (1.6), xG (1.2),
and xC (1.0). This order is similar to that for the four
natural bases: A (1.0), G (0.6), T (0.5), and C (0.4).16d
However, the four size-expanded (xDNA) bases all stack
more than twice as strongly as their natural (DNA)
counterparts.
(15) Millican, T. A.; Mock, G. A.; Chauncey, M. A.; Patel, T. P.;
Eaton, M. A.; Gunning, J.; Cutbush, S. D.; Neidle, S.; Mann, J. Nucleic
Acids Res. 1984, 12, 7435-7453.
(16) (a) Turner, D. H.; Sugimoto, N.; Freier, S. M. Annu. Rev.
Biophys. Biophys. Chem. 1988, 17, 167-192. (b) SantaLucia, J.; Allawi,
H. T.; Seneviratne, P. A. Biochemistry 1996, 35, 3555-3562. (c)
Guckian, K.; Schweitzer, B. A.; Ren, R. X.-F.; Sheils, C. J.; Paris, P.
L.; Tahmassebi, D. C.; Kool, E. T. J. Am. Chem. Soc. 1996, 118, 8182-
8183. (d) 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. (e) Nakano, S.; Uotani, Y.; Nakashima, S.; Anno, Y.; Fujii, M.;
Sugimoto, N. J. Am. Chem. Soc. 2003, 125, 8086-8087. (g) Lai, J. S.;
Qu, J.; Kool, E. T. Angew. Chem., Int. Ed. 2003, 42, 5973-5977.
Discussion
Our studies demonstrate that syntheses of xC and xG
nucleosides can be accomplished without severe difficul-
646 J. Org. Chem., Vol. 70, No. 2, 2005

Expanded Analogues of dG and dC
ties and that the nucleotides can be incorporated intact
into DNA after deprotection. The synthesis of the xC
nucleoside was achieved in straightforward fashion and
in relatively high yields. It conveniently makes use of the
previous xT nucleoside as its starting point.7 By contrast,
the preparation of the dxG nucleoside is long and rather
low-yielding. In considering possible routes to this previ-
ously unknown compound we considered direct coupling
of a base precursor to a deoxyribose derivative rather
than a ribose precursor, to save steps in deoxygenation
of the 2′ position. However, we chose to adopt the longer
procedure to take advantage of the stereoselectivity and
yield in the more efficient glycosylation with tetraacetyl-
ribose. It remains to be seen whether shorter routes can
be developed to increase access to this compound.
The pairing studies of xC and xG in the context of short
DNAs provide new insights into understanding the
flexibility of DNA backbones to accommodate larger-size
base pairs. Parallel to previous results from xT and xA
investigations,9 the current studies showed that base
pairs involving xC or xG generally destabilized the duplex
stability by 0.3-1.3 kcal/mol. For xC in particular, the
small 0.3 kcal/mol of destabilization is less than that
observed for the other three stretched bases. Without
further structural studies it remains unclear what the
origin of this small difference is.
and 1.3-1.5 log unit difference in the smaller pyrim-
idines.
As for the selectivity of base pairing by xC and xG
bases in the present natural DNA context, the data show
that both new bases do show significant selectivity. The
degree of selectivity of xG is close to the magnitudes of
selectivities seen previously for xT and xA, and these
values are smaller than the selectivities measured for
natural DNA bases. We surmise that the lower selectivity
of the expanded bases in natural DNA arises from the
fact that even the fully matched pairs cause some
distortion (and therefore an energetic penalty) in the
duplex. This is unlike natural Watson-Crick pairs,
where matched pairs cause no distortion of the helix,
while mismatched ones result in a distortion penalty.
This idea is also supported by recent studies of the
pairing selectivity of xT and xA bases in the context of
fully expanded DNAs.10 In that context, xT and xA should
cause little distortion when correctly paired, and the data
show that they do, in fact, give selectivity that is in most
cases as high as, or even higher than, that of natural
bases in natural DNA. Thus, we conclude that selectivity
of pairing of expanded or natural bases does not arise
from the sizes of the bases alone, but rather from the
compatibility of pairs with their surrounding context.
Finally, the successful preparation of the second half
of the xDNA nucleoside set should make possible a
number of future experiments in nucleic acid recognition.
One of these is the possibility that xDNA strands
composed only of size-expanded nucleosides might be
used to bind natural DNA or RNA sequences. xDNA
strands are promising in this respect because expanded-
size helices can be considerably more stable than natural
duplexes.10,17 Thus, the xDNA strands might bind even
to sequences that are involved in secondary structures.
The addition of the dxC and dxG building blocks to this
nucleotide set should make such targeting general for all
possible nucleic acid sequences. Second, xDNA bases are
inherently fluorescent, and it is possible that such
fluorescent bases might be useful as self-reporters of
binding in mechanistic, diagnostic, or imaging applica-
tions. Third, a set of four xDNA bases might be combined
with the natural four bases to make an eight-base genetic
set that encodes more information than the natural
genetic system. Studies are underway to test all these
possibilities.
The observed pairing selectivities of xG for C and xC
for G suggest formation of hydrogen bonded pairs within
the DNA (Figure 1c), although some caution on this point
should be retained until structural studies can confirm
this. Presumably there is an adjustment in backbone
conformation between these larger pairs and the neigh-
boring natural ones to accommodate this transition, and
this suboptimal conformation is reflected in the small
degree of destabilization. We conjecture that the inherent
backbone distortion cost may in fact be larger than this,
but may be compensated for, to some degree, by the
considerably stronger stacking of xC and xG as compared
to C and G.
The completion of the xDNA nucleoside set has allowed
the completion of measurement of their stacking tenden-
cies in DNA. The data show that each expanded nucleo-
base stacks more favorably than its analogue by more
than a factor of 2 in free energy. We hypothesize that
this high affinity arises from two factors: first, the added
size, which comes from the addition of the benzene ring
to each molecule. One face of a benzo ring has a surface
area of roughly 7.5 Å2, increasing overall surface area of
a given base by ca. 50-100%. Increased area of overlap
with neighboring bases would aid stacking by increasing
polarizability, thus enhancing van der Waals interac-
tions. In addition, the addition of a hydrocarbon benzo-
ring to a more polar heterocycle is expected to have the
effect of decreasing overall polarity. This makes it more
likely that solvophobic effects would also contribute
favorably to stacking.16 A simple calculation of polarities
as represented by logP values for the four xDNA bases
gave values of -0.02, +0.91, -0.31, and +0.69 (for xA,
xT, xG, and xC, respectively), which is much less polar
than the natural bases, which gave calculated values of
-1.07, -0.36, -1.36, and -0.76 (for A, T, G, and C,
respectively). Thus, the benzo group is expected to make
a ca. 1.1 log unit difference in polarity for the purines
Acknowledgment. This work was supported by the
National Institutes of Health (GM63587). H.L. and J.G.
acknowledge support from Stanford Graduate Fellow-
ships.
Supporting Information Available: Experimental de-
tails of nucleoside synthesis and characterization; oligo-
nucleotide synthesis and characterization; thermal de-
naturation methods; fluorescence spectrometry methods. This
material is available free of charge via the Internet at
http://pubs.acs.org.
JO048357Z
(17) Lu, H.; Kool, E. T. Angew. Chem., Int. Ed. 2004, 43, 5834-
5836.
J. Org. Chem, Vol. 70, No. 2, 2005 647