ABSTRACT
Modified oligodeoxyribonucleotides (ODNs) that have unique hybridization
properties were designed and synthesized for the first time. These ODNs, called
selective binding complementary ODNs (SBC ODNs), are unable to form stable
hybrids with each other, yet are able to form stable, sequence specific hybrids
with complementary unmodified strands of nucleic acid. To make SBC ODNs,
deoxyguanosine (dG) and deoxycytidine (dC) were substituted with deoxyinosine
(dI) and 3-(2
'
-deoxy-
[beta]
-D-ribofuranosyl)pyrrolo-[2,3-
d
]-pyrimidine-2-(3
H
)-one (dP), respectively. The hybridization properties of several otherwise identical complementary ODNs containing one or both of these nucleoside analogs were studied by both UV monitored thermal denaturation and non-denaturing PAGE. The data showed that while dI and dP did form base
pairs with dC and dG, respectively, dI did not form a stable base pair with dP. A self-complementary ODN uniformly substituted with dI and dP acquired single-stranded character and was able to strand invade the end of a duplex
DNA better than an unsubstituted ODN. This observation implies that SBC ODNs
should effectively hybridize to hairpins present in single-stranded DNA or RNA.
Oligodeoxyribonucleotides (ODNs) do not effectively hybridize to complementary
sequences which are already base paired. Without the assistance of recombinase
enzymes such as recA (
1
), accessibility of ODNs to double-stranded DNA (dsDNA) is usually restricted to homopurine runs (
2
) or to extruded single-stranded sequences in supercoiled DNA (
3
). Although less of an issue with single-stranded DNA (ssDNA) or RNA, hybridization of ODNs to many sequences in
these molecules can be compromised by intramolecular base pairing (
4
,
5
). While numerous hybridization strategies have been described to overcome or
exploit secondary structure, none provides a general solution to the problem.
Examples include modified ODNs which form unusually stable hybrids (
6
-
12
), ODNs which form triple-stranded complexes (
13
), ODNs which hybridize to hairpins or contiguous flanking sequences (
14
-
18
), and the use of `effector' ODNs (
19
) and `tethered' ODNs (
20
) to improve binding affinity through cooperative interactions.
A pair of uniquely modified complementary ODNs (or a single self-complementary ODN) that do not hybridize to each other, yet do hybridize
to unmodified complementary sequences might offer a general solution to the
challenge of targeting any site in DNA or RNA. If such a pair of ODNs could be
synapsed to a homologous region in dsDNA by recombination, a complement-stabilized or double D-loop (
21
-
22
) would be formed (Fig.
1
a). Unlike a simple D-loop, the double D-loop is relatively stable and might inhibit gene expression.
Alternatively, the same type of paired ODNs could be hybridized to a unique
sequence in long, single-stranded nucleic acid. To the extent that sequence is involved in secondary structure (such as a localized hairpin; Fig.
1
b), the paired ODNs should have an advantage over a standard ODN. Whether such ODNs
are used as probes or antisense agents, their hybridization to a target should
generate more new base pairs than an unmodified ODN. This is depicted in Figure
1
.
Materials and their sources were as follows: DNA synthesis reagents, Glen Research; phosphodiesterase I (
Crotalus adamanteus
venom), alkaline phosphatase (calf intestinal) and DNase I, Amersham Life
Science; T4 polynucleotide kinase (10 U/[mu]l), Promega; [[gamma]-
32
P]ATP, NEN Research. Commercial reagents were used as received.
1
H-NMR spectra were determined on a Varian Gemini-300. Elemental analysis was performed by Quantitative Technologies Inc. (Whitehouse, NJ). UV spectra were measured on a Beckman DU-40 spectrophotometer or a Perkin Elmer Lamda 2S UV/VIS
spectrophotometer.
5-Ethynyl-2'-deoxyuridine (3 g, 11.9 mmol) (
23
) and copper (I) iodide (500 mg, 2.6 mmol) in a 250 ml two-necked round-bottomed flask were dried in vacuo for 3 h, placed under argon, and suspended in anhydrous DMF (35 ml) and triethylamine (15 ml). The solution was vigorously stirred at 120oC under argon and every 30 min fresh copper (I) iodide (250 mg, 1.3 mmol) was added until most of the starting material had reacted. After 2 h, the
resulting mixture was filtered and the filtrate was concentrated
in vacuo
to dryness. The residue was suspended in acetone (100 ml) and stirred
overnight. The desired product was filtered, washed with acetone (20 ml), and
dried
in vacuo
to afford 2.2 g of
dF
as a slightly yellowish solid. The remaining product in mother liquor was
further purified by silica gel column chromatography (elution solvent: 25% MeOH
in EtOAc) to afford an additional 0.3 g of
dF
(total yield: 2.5 g, 83%): mp 167-168oC; UV (0.05 M KHPO
4
/NaOH, pH 7) [lambda]
max
322 nm ([epsilon] 12 500). Anal. calcd for C
11
H
12
N
2
O
5
: C, 52.38; H, 4.80; N, 11.11. Found: C, 52.11; H, 4.81; N, 10.91.
1
H NMR (DMSO-d
6
): the same as reported by Kumar
et al
. (
24
).
dF
(2.17 g, 8.6 mmol) was dried
in vacuo
at 60oC overnight and then added to 4,4'-dimethoxytrityl-chloride (3.51 g, 10.4 mmol) and anhydrous triethylamine
(2.4 ml) in pyridine (30 ml). After 2 h at room temperature under argon, the
resulting mixture was diluted with an equal volume of water and extracted with two 150 ml portions of ether. The ether layer was dried over anhydrous sodium sulfate and
evaporated to dryness. The residue was dissolved in dichloromethane (20 ml) and the desired product (4.6 g) was precipitated by adding the solution to 400 ml of rapidly stirred hexanes.
Filtration yielded 4.6 g (96%) of a white solid.
Chloro-[([beta]-cyanoethoxy)-
N
,
N
-diisopropylamino]-phosphine (2.9 g, 12.5 mmol) was added dropwise over 30 s to an anhydrous mixture of 5'-
O
-(4,4'-dimethoxytrityl)-
dF
(4.6 g, 8.3 mmol), diisopropylethyl amine (5.8 ml), and dichloromethane (27 ml)
under argon (
25
). After 30 min at room temperature the reaction was stopped by adding anhydrous
methanol (0.3 ml). The reaction mixture was extracted with 5% aqueous NaHCO
3
(2* 15 ml) and saturated aqueous NaCl (2* 15 ml). The organic layer was dried over anhydrous sodium
sulfate, filtered and then evaporated under reduced pressure to afford a brown
oil. This crude product was further purified by silica gel column
chromatography using hexanes:CH
2
Cl
2
:EtOAc:Et
3
N (4:3:2:1 by vol) as the solvent system. Fractions containing the desired
product were combined, evaporated to dryness, and redissolved in EtOAc (10 ml).
Precipitation from rapidly stirred hexanes (400 ml) yielded 5.9 g (94%) of
purified material.
1
H NMR (CDCl
3
) [delta] 8.88 (d, J = 18.6 Hz, 1H), 7.5-7.2 (m, 10H), 6.81 (m, 4H), 6.32 (m, 1H), 5.62 (d of d, J = 9.5,
3.9 Hz, 1H), 4.70 (m, 1H), 4.19 (m, 1H), 3.8-3.4 (m, 12H), 2.77 (m, 1H), 2.59 (t, J = 9.4 Hz, 2H), 2.44 (m, 2H), 1.25-1.0 (m, 12H).
dF
(1 g, 3.96 mmol) was dissolved in 30% aqueous ammonium hydroxide (30 ml). After
overnight at room temperature, the resulting solution was concentrated
in vacuo
to dryness. The residue was suspended in acetone (50 ml), stirred overnight,
and the undissolved product filtered to afford 850 mg. The mother liquor was
concentrated to dryness and the residue was suspended in acetone (10 ml)
overnight with stirring to yield an additional 100 mg of insoluble product
(total 950 mg, 95.4%). This compound was analyzed by HPLC, UV and NMR and shown
to be identical to authentic 3-(2'-deoxy-[beta]-D-ribofuranosyl)pyrrolo-[2,3-
d
]-pyrimidine-2(3
H
)-one (
dP
).
ODNs containing modified bases were synthesized on 1 [mu]mol scale using standard procedures for an ABI-394 DNA synthesizer. ODNs with the dimethoxytrityl group were purified by HPLC using a
Hamilton PRP-1 (7.0 * 305 mm) reverse phase column employing a gradient of 5 to 45% CH
3
CN in 0.1 M Et
3
NH
+
OAc
-
, pH 7.5, over 20 min with a 2 ml/min flow rate. After detritylation with 80%
acetic acid, the ODNs were precipitated by addition of 3 M sodium acetate and 1-butanol. The resulting ODNs were dried and further purified by using 20%
denaturing PAGE as described by Hopkins
et al
. (
26
).
Enzymatic hydrolysis of ODNs was carried out as described by Woo
et al
. (
27
). The resulting hydrolysate was analyzed by HPLC with dual detection at 260 nm
and 320 nm (Waters 994 Programmable Photodiode Array Detector) using a C-18 reverse phase column (Rainin, Microsorb
TM
Short-One
®
). The solvent gradient was run at 1 ml/min as follows: solvent A, 0.1 M Et
3
NH
+
OAc
-
, pH 7.5; solvent B, CH
3
CN; a linear gradient 0 to 13% B over 10 min, a linear gradient to 100% B over 2
min, then isocratic 100% B for 3 min. Peaks were identified by comparison of
retention times to those of authentic, commercial samples (dA, dG, dT and dC)
and synthetic samples (
dF
and
dP
) prepared by known procedures (
28
).
T
m
values were recorded on a Perkin Elmer Lamda 2S UV/VIS spectrophotometer
equipped with a temperature programmer (PTP-6) and interfaced to an IBM personal computer (PECSS software, Perkin
Elmer). Scan rates were 0.5oC/min. Data were collected at 260 nm in the temperature range from 5 to 90oC. The
T
m
is defined as the temperature at half the maximal hyperchromicity using
baseline correction at high and low temperature extremes (
29
). Samples were prepared by dissolving ODNs in TNM buffer [10 mM Tris-HCl (pH 8.0), 0.1 mM EDTA, 50 mM NaCl, 10 mM MgCl
2
]. To ensure complete hybridization of complementary strands (1:1 molar ratio)
before collecting data, the samples were incubated at 90oC for 2 min and cooled to 3oC over 1 h. The concentration of hybridized ODNs was approximately 2 [mu]M.
ODNs with an asterisk (*) in Figures
5
and
6
were 5'
32
P-labeled using T4 kinase and [[gamma]-
32
P]ATP (
30
) and present at 0.5 [mu]M unless otherwise indicated. Hybrids were formed by incubating the labeled
ODN with a 2-fold molar excess of cold complementary ODN for 60 min at room temperature
in 20 [mu]l TNM buffer. These samples were then mixed with 20 [mu]l loading buffer (0.25% bromophenol blue, 0.25% xylene cyanol, 2.5%
Ficoll type 400) and then kept on ice prior to gel electrophoresis. Aliquots (5
[mu]l) were analyzed in a 12% non-denaturing polyacrylamide gel [19:1 acrylamide: bisacrylamide, 0.35 mm
thick, 20 * 16 cm, polymerized and run in TBE buffer (89 mM Tris-borate/2 mM EDTA) containing 3 mM MgCl
2
]. Pre-electrophoresis in a BioRad Protean
®
II xi apparatus was performed for 1 h at 200 V and 10oC. Samples were loaded, and the gel was run as before until the bromophenol
blue dye had traveled ~15 cm (~5 h). The gel was dried and visualized with a phosphorimager (BioRad
GS-250 Molecular Imager).
The design paradigm for SBC ODNs is modification of complementary dA-dT or dG-dC bases such that the modified bases form only one hydrogen bond
when paired to each other, yet can form two or even three hydrogen bonds when
paired to the natural partner. We report the synthesis of a complementary pair
of G/C-rich SBC 28mers substituted with deoxyinosine (
dI
) in place of dG and 3-(2'-deoxy-[beta]-D-ribofuranosyl) furano-[2,3-
d
]- pyrimidine-6(5
H
)-one (
dF
) in place of dC. As shown in Figure
2
, these modified bases should form two hydrogen bonds, respectively, with dC (2b) or dG (2c), yet only one hydrogen bond with each other (2d).
Although the stabilities of the SBC-DNA hybrids might not be as good as DNA-DNA hybrids, the SBC-SBC hybrids would be much less stable, thus enabling the
design goals.
The dG analog was simply prepared by removal of the N2 exocyclic amino group of
dG to give deoxyinosine (
dI
). This nucleoside analog is known to preferentially pair with dC (
31
-
32
). The modified dC was designed to have no hydrogen bonding ability at the
position equivalent to the N4 exocyclic amino group of dC. We chose the
bicyclic nucleoside
dF
to fulfill this role. It was expected to be better than monocyclic dC analogs (
33
) because of its base stacking ability. The nucleoside
dF
was synthesized by copper (I)-catalyzed cyclization of the known antiviral nucleoside, 5-ethynyl-2'-deoxyuridine (
23
). Unlike
dF
analogs with a substituent at C2 (
23
,
34
-
35
), preparation of the desired compound was very sensitive to solvent and
reaction conditions; for example, if pyridine was used as a solvent the major product was a dimer of the starting material.
dF
was dimethoxytritylated and converted to its cyanoethoxy phosphoramidite by
conventional methods (
25
).
Table
1
shows the hybridization properties of 28mer ODNs containing
dI
for dG,
dP
for dC, or both. The sequence, taken from pBR322 plasmid, had a G-C content of 60.7%. Introduction of either
dI
or
dP
into one or both strands of the duplex decreased its
T
m
by 1.8-2.0 or 0.4-0.7oC, respectively, per modified base pair. When only one strand
of the hybrid was substituted with both
dI
and
dP
, the
T
m
dropped by 1.1-1.6oC per modified base pair. These values reflect a slight
destabilization attributable to the dG-
dP
base pair and a larger destabilization due to the
dI
-dC base pair. When both strands of the hybrid were substituted with
dI
and
dP
, however, the
T
m
drop per modified base pair increased significantly to 3.3oC.
Some of the hybrids were analyzed by non-denaturing PAGE (Fig.
5
). As shown in Table
1
and Figure
5
, the SBC ODNs containing both
dI
for dG and
dP
for dC (Watson in
VIII
, Crick in
IX
and both Watson and Crick in
X
) did not form a stable hybrid with each other at room temperature (hybrid
X
; Fig.
5
, lane 8), yet did form stable hybrids with their unmodified complementary ODN strands (hybrid
IX
; Fig.
5
, lane 11). As a result, these ODNs exhibited selective complementary binding.
Despite the reduced stability of hybrids formed between SBC and normal ODNs,
the normal Watson strand showed no preference for the normal Crick over the SBC
Crick strand when equimolar of these three strands were mixed simultaneously at
room temperature; about equal amount of duplexes
I
and
IX
were formed (Fig.
5
, lane 13). Additionally, there was little, if any, strand displacement or
strand exchange when the pre-formed DNA-SBC duplex
IX
was incubated with the normal homolog of the SBC strand or with SBC-DNA duplex
VIII
; not much single-stranded SBC was formed (Fig.
5
, lanes 12 and 14). These data clearly demonstrate that the SBC ODNs described
above behaved like natural ODNs when hybridized to unmodified complements, yet
did not form stable hybrids with themselves.
To determine whether an SBC ODN could strand invade dsDNA, a 17 base pair
segment of hybrid
X
was synthesized as a single self-complementary ODN (
XIII
; Fig.
6
A). Linking the complementary domains into one ODN was expected to improve the
kinetics of strand invasion and the stability of the product. The chimeric SBC
ODN had a
T
m
of 31oC and hybridized to a partial DNA complement (Watson in
XI
) at room temperature (Fig.
6
B, lane 4). The corresponding unmodified ODN (
XII
), derived from hybrid
I
, had a
T
m
of 80oC and hybridized poorly to the same DNA complement (Fig.
6
B, lane 3). A 48 bp duplex (
XI
) with one end homologous to the self-complementary ODNs was used as a substrate for strand invasion. Annealing
was facilitated by the presence of two 5 base long single-stranded overhangs in
XI
which could hybridize to complementary four base long overhangs in the invading
ODNs. The duplex can be likened to the stem of a hairpin that might exist in a
long ssDNA. After 4 h at room temperature, a 10-fold excess of
XIII
converted 73% of
XI
to a three-way junction compared with 17% for
XII
(Fig.
6
B, lanes 6 and 5). Ongoing studies indicate that the self-complementary SBC ODN has a significant kinetic advantage over the unmodified ODN.
Based on thermal denaturation and non-denaturing gel mobility shift assays, we have designed and synthesized for
the first time modified ODNs which exhibit selective complementary hybridization. A self-complementary version of one of these paired ODNs strand invaded a
homologous double-stranded DNA better than the corresponding unmodified ODN. The possible
diagnostic and therapeutic uses of these ODNs are being explored. Efforts to
improve the hybridization properties of SBC ODNs including the modification of
dA and dT are also underway.
We thank Drs I. V. Kutyavin, A. Gall, V. Gorn and E. Lukhtanov for helpful
discussions. We thank Mr D. Adams and Ms A. Yang for technical contributions.
REFERENCES
Return