Synthesis and characterization of 8-methoxy-2
'
-deoxyadenosine-containing oligonucleotides to probe the
syn
glycosidic conformation of 2
'-deoxyadenosine within DNA
Synthesis and characterization of 8-methoxy-2 ' -deoxyadenosine-containing oligonucleotides to probe the syn glycosidic conformation of 2 '-deoxyadenosine within DNA
Robert G.
Eason
,
Dawn M.
Burkhardt
,
Shirley J.
Phillips
,
David P.
Smith
and
Sheila S.
David*
Department of Chemistry and Biochemistry, University of California,
Santa Cruz
, CA 95064,
USA
Received November 2, 1995;
Revised and Accepted January 11, 1996
ABSTRACT
The synthesis of 8-methoxy-2
'
-deoxyadenosine (moA) protected at N
6
as an
N
,
N
-dimethylformamidine derivative and incorporation of the modified
nucleoside into oligodeoxynucleotides via the phosphoramidite method are
described. UV thermal denaturation studies were conducted on duplexes
containing moA:G, moA:C and moA:T base pairs to determine the thermodynamic
stability of duplexes containing moA relative to their adenosine (A)-containing counterparts. In the case of moA:G base pairs the effect of moA
substitution is sequence dependent. In A:G mismatch-containing sequences, which have been shown by structural characterization
to have a
syn
conformational preference at the glycosidic bond of A, moA substitution results
in stabilization of the duplex. In contrast, in sequences where the A in the
A:G mismatch has been shown to prefer the
anti
conformation moA substitution is destabilizing to the duplex. Thus moA may be a
useful probe for investigating the conformational preferences of the
N
-glycosidic bond of adenosine within DNA. In addition, moA nucleoside is
more resistant to acid-catalyzed depurination than previously described 8-bromo-2
'
-deoxyadenosine, allowing for facile incorporation into oligonucleotides
via automated solid phase DNA synthesis.
INTRODUCTION
In B-form DNA the
anti
conformation of the
N
-glycosidic bond (C1'-N) predominates in normal Watson-Crick base pairs. However, the
syn
glycosidic conformation of guanine residues in DNA has been observed. For
example, in left-handed Z-DNA the guanine nucleotides adopt the
syn
conformation (
1
,
2
). In addition, high resolution structures of G quartet structures derived from
several telomeric DNA sequences (
3
,
4
) and a thrombin binding aptamer
(
5
,
6
) show that the glycosidic conformations of sequential guanine residues
alternate between
syn
and
anti
along the polynucleotide chain. Mismatches in DNA, and in particular A:G
mismatches, have been observed in a number of structural motifs differing in
conformation at the
N
-glycosidic bond. Observed NMR and X-ray crystallographic structures for A:G mismatches include A
anti
:G
anti
, AH
+
anti
:G
syn
and A
syn
:G
anti
base pairs (Fig.
1
) (
7
-
12
).
In vivo
experiments have indicated that A:G mispairs are among the least efficiently
repaired, possibly due to a failure of the repair enzyme to recognize all the
structural conformations of the A:G mismatch (
13
). DNA damage may also result in modifications of the nucleotides which alter
the glycosidic conformation. For example, oxidative stress can result in
formation of 7,8-dihydro-8-oxoguanine, which will preferentially adopt the
syn
conformation in an A:G base pair (
14
). Thus DNA repair enzymes might key upon a particular conformation-dependent recognition element at a mismatched base pair or at a damaged
base.
MATERIALS AND METHODS
General methods
2'-Deoxyadenosine was purchased from Fisher. 2-Cyanoethyl-
N
,
N
-diisopropyl chlorophosphoramidite was obtained from Aldrich. Other
chemicals were purchased from Fluka. TLC was conducted on silica GF
254
plates (Analtech). Flash column chromatography was performed on 200-425 mesh silica (Davisil type 60A; Fisher). HPLC separations and analyses
were conducted on a Waters 625LC system with a Waters 996 photodiode array
detector. Oligonucleotides were purified by anion exchange HPLC on a Waters Gen-Pak Fax column (DEAE, 100 * 4.6 mm) and analyzed for purity by reversed phase HPLC on an
Alltech Nucleosil C
18
column (250 * 4.6 mm). Nucleoside analyses from enzymatic digestions were conducted
under similar reversed phase conditions.
1
H-,
13
C- and
31
P-NMR spectra were recorded on a Brüker 250A spectrometer; chemical shifts are in p.p.m. relative to
internal tetramethylsilane (
1
H or
13
C) or external 85% H
3
PO
4
(
31
P). DNA thermal denaturation experiments were conducted on a Hewlett Packard
8452A diode array UV-VIS spectrophotometer equipped with an HP89090A Peltier heat control unit.
Electrospray ionization mass spectroscopic (ESI-MS) data were collected on a Perkin-Elmer Sciex API-1 instrument, running either in positive ion (nucleosides) or
negative ion (oligonucleotides) mode. High resolution FAB-MS data were provided by the UIUC Mass Spectrometry Facility (Urbana, IL).
8-Bromo-2
'
-deoxyadenosine (1)
Bromine (2.40 ml, 46.4 mmol) was added to freshly prepared acetate buffer, pH
4.2 (60 ml) and stirred for 1-2 h until fully dissolved. The bromine solution was added in portions
over 20 min to a slurry of 2'-deoxyadenosine (10.0 g, 37.2 mmol) in acetate buffer (150 ml). The
reaction was complete by TLC after 90 min. Saturated sodium bisulfite was added
to destroy excess bromine, then the reaction mixture was neutralized with 2 N
NaOH. The product precipitated quickly and was collected, washed with cold
water and dried. Yield of a beige solid was 6.45 g (52 %). TLC (CHCl
3
/MeOH, 80:20):
R
f
0.56. [lambda]
max
(MeOH) 214, 266 nm.
1
H-NMR (DMSO-d
6
): 8.10 (s, 1H, 2-H), 7.53 (s, 2H, 6-NH
2
), 6.28 (t, 1H, 1'-H), 5.32 (br, 2H, 3'-OH and 5'-OH), 4.47 (m, 1H, 4'-H), 3.87 (m, 1H, 3'-H), 3.67-3.43 (m,
2H, 5'-H), 3.23 and 2.18 (m, 2H, 2'-H).
8-Methoxy-2
'
-deoxyadenosine (2)
Sodium (1.38 g, 60.0 mmol) was dissolved in anhydrous methanol (24 ml). The
sodium methoxide was then added to a solution of
1
(5.00 g, 15.1 mmol) in anhydrous methanol (100 ml) at 45oC. The reaction was complete by HPLC after 2 h. Water (100 ml) was added
and the solution was neutralized carefully with 1 N HCl (44 ml). Solvent was
removed and the residue recrystallized from a minimal amount of water. Yield of
an off-white solid in two crops was 2.43 g (57%). TLC (CHCl
3
/MeOH, 80:20):
R
f
0.52. [lambda]
max
(MeOH) 210, 260 nm.
1
H-NMR (DMSO-d
6
): 8.00 (s, 1H, 2-H), 6.95 (s, 2H, 6-NH
2
), 6.17 (t, 1H, 1'-H), 5.26 (br, 2H, 3'-OH and 5'-OH), 4.39 (m, 1H, 4'-H), 4.11 (s, 3H, 8-MeO-), 3.81 (m, 1H,
3'-H), 3.60-3.43 (m, 2H, 5'-H), 2.99 and 2.08 (m, 2H, 2'-H).
13
C-NMR (DMSO-d
6
): 163.63, 163.48, 160.07, 158.22, 124.35, 97.35, 92.07, 80.73, 71.79, 66.74,
46.41. ESI-MS:
m
/
z
(relative intensity) 282 (MH
+
, 100), 166 ([purine + H]
+
, 89).
N
6-(
N
,
N
-Dimethylaminomethylene)-8-methoxy-2
'
-
deoxyadenosine (3)
Compound
2
(1.00 g, 3.56 mmol) was co-evaporated with dry DMF (5 ml) and dried overnight under vacuum. The dry
residue was then redissolved in DMF (25 ml). To this solution was added
N
,
N
-dimethylformamide di-
n
-butyl acetal (2.17 g, 10.7 mmol). After stirring for 18 h at 40oC the reaction mixture was partitioned between CH
2
Cl
2
and 1 N NaHCO
3
and the aqueous phase extracted with CH
2
Cl
2
(4 * 20 ml). The combined organic portions were dried over Na
2
SO
4
and then solvent was removed under vacuum. Yield of a hygroscopic, waxy solid was 1.09 g, (91%). TLC (CHCl
3
/MeOH/NEt
3
, 90:9:1):
R
f
0.34. [lambda]
max
(MeOH) 312, 240 nm.
1
H-NMR (DMSO-d
6
): 8.81 (s, 1H, -N=CH-), 8.29 (s, 1H, 2-H), 6.25 (t, 1H, 1'-H), 5.75 (br, 2H, 3'-OH and 5'-OH), 4.44 (m, 1H, 4'-H), 4.15
(s, 3H, 8-MeO-), 3.84 (m, 1H, 3'-H), 3.68-3.43 (m, 2H, 5'-H), 3.22 and 3.16 (2s, 6H, Me
2
N-), 3.08 and 2.16 (m, 2H, 2'-H). ESI-MS:
m
/
z
(relative intensity) 337 (MH
+
, 36), 221 ([purine + H]
+
, 100). HRMS (FAB): calculated for C
14
H
21
N
6
O
4
(MH
+
) 337.1624, found 337.1626.
5
'
-
O
-(4,4
'
-Dimethoxytrityl)-
N
6-(
N
,
N
-dimethylamino-
methylene)-8-methoxy-2
'
-deoxyadenosine (4)
Compound
3
(0.685 g, 2.04 mmol) was co-evaporated with anhydrous pyridine (3 * 10 ml). To this was added (as freshly dried powder) 4, 4'-dimethoxytrityl chloride (0.83 g, 2.45 mmol), followed
by pyridine (15 ml). The mixture was stirred under argon at 25oC. After 30 min CH
2
Cl
2
(20 ml) and 1 N NaHCO
3
(20 ml) were added and the aqueous phase was extracted with CH
2
Cl
2
(2 * 20 ml). The combined organic fractions were washed with water, dried
over Na
2
SO
4
and then solvent was removed under vacuum. The crude oil was purified by flash
column chromatography, eluting with a 0-5% gradient of MeOH in CH
2
Cl
2
/NEt
3
, 99:1. Yield of a white solid was 1.37 g, (75 %). TLC (CHCl
3
/MeOH/NEt
3
, 90:9:1):
R
f
0.42. [lambda]
max
(MeOH) 210, 236, 312 nm.
1
H-NMR (CDCl
3
): 8.76 (s, 1H, -N=CH-), 8.27 (s, 1H, 2-H), 7.39-6.76 (m, 13H, DMT Ar-H), 6.40 (t, 1H, 1'-H), 4.83 (m, 1H, 4'-H), 4.12 (s, 3H, 8-MeO-), 4.04 (m,
1H, 3'-H), 3.77 (s, 6H, DMT MeO-), 3.34 (m, 2H, 5'-H), 3.22 and 3.16 (2s, 6H, Me
2
N-), 3.25 and 2.28 (m, 2H, 2'-H). ESI-MS:
m
/
z
(relative intensity) 639 (MH
+
, 100), 337 ([M-DMT]
+
, 90). HRMS (FAB): calculated for C
35
H
39
N
6
O
6
(MH
+
) 639.2931, found 639.2932.
5
'
-
O
-(4,4
'
-Dimethoxytrityl)-
N
6-(
N
,
N
-dimethylamino-
methylene)-8-methoxy-2
'
-deoxyadenosine-3
'
-
O
-[(2-cyano-
ethyl)-
N
,
N
-diiisopropylphosphoramidite] (5)
Compound
4
(0.285 g, 0.455 mmol) was co-evaporated with anhydrous pyridine (3 * 10 ml). To the solid residue was added anhydrous CH
2
Cl
2
(4 ml), triethylamine (0.136 ml, 0.910 mmol) and 2-cyanoethyl-
N
,
N
-diisopropyl chlorophosphoramidite (0.130 ml, 0.55 mmol). The mixture was
stirred under argon for 50 min at 25oC, then quenched with anhydrous MeOH (2 ml). After stirring for 15 min
solvent was removed and the crude oil purified by flash chromatography using a
0-2% gradient of MeOH in CH
2
Cl
2
/NEt
3
, 99:1. Yield of a white solid was 0.240 g, (63 %). TLC (CHCl
3
/MeOH/NEt
3
, 90:9:1):
R
f
0.71.
1
H-NMR (CDCl
3
): 8.76 (s, 1H, -N=CH-), 8.17 (s, 1H, 2-H), 7.28-6.72 (m, 13H, DMT Ar-H), 6.27 (m, 1H, 1'-H, two diastereomers), 4.86 (m, 1H,
4'-H), 4.04 (s, 3H, 8-MeO-), 3.69 (s, 6H, DMT MeO-), 3.15 and 3.08 (2s, 6H, Me
2
N-).
31
P-NMR (CDCl
3
): 148.84 and 148.68 (two diastereomers). ESI-MS:
m
/
z
(relative intensity) 839 (MH
+
, 100), 303 ([DMT]
+
, 27), 786 ([MH-acrylonitrile]
+
, 11). HRMS (FAB): calculated for C
44
H
56
N
8
O
7
P (MH
+
) 839.4010, found 839.4009.
Solid phase synthesis of oligonucleotides
Automated syntheses were conducted on a Millipore Cyclone Plus DNA synthesizer
on a scale of either 0.2 or 1.0 [mu]mol with a final detritylation step. 5'-
O
-Dimethoxytritylated nucleoside [beta]-cyanoethyl phosphoramidites were purchased from Perseptive
Biosystems, with DNA bases protected either as isobutyryl (G) or benzoyl (A and
C) derivatives. Oligonucleotides were cleaved from the support and deprotected
with 28% NH
4
OH for 24 h at 25oC. The crude oligonucleotides were purified by anion exchange HPLC in 20 mM
NH
4
H
2
PO
4
, 10% AcCN, pH 8, employing a salt gradient of 0-30% 1.0 M NaCl. Representative oligonucleotides were analyzed by one of
two methods: (i) enzymatic digestion with snake venom phosphodiesterase and
bacterial alkaline phosphatase (
25
), with constituent nucleosides then identified by reversed phase HPLC in 100 mM
TEAA, pH 6.5, with an AcCN gradient of 0-40%; (ii) ESI-MS with an operating potential of -3500 V and orifice potential of 75-90 V. Note that 1% piperidine was added to the
desalted aqueous samples to enhance the signal.
Thermal denaturation studies
Oligonucleotides containing the moA residue were annealed at a 1:1 ratio to
their appropriate complementary strands in buffer containing 0.10 M phosphate,
1.0 M NaCl, 0.001 M EDTA, pH 7. All final solutions contained 8-11 [mu]M duplex DNA. In the case of non-self-complementary oligonucleotides individual strand concentrations were determined spectrophotometrically
using
c
=
A
/[epsilon]
l
. The extinction coefficient at 25oC, [epsilon], was calculated by the nearest-neighbor method, summing the known [epsilon]
260
values for the individual residues in di- and trinucleotides (
26
). The [epsilon]
260
value for the free moA nucleoside was experimentally found to be 18.1 * 10
3
M
-1
cm
-1
in water at 25oC. Concentration errors introduced by the presence of moA residues in the
duplex were estimated to be insignificant, based upon the relative similarity in extinction coefficient to the normal A nucleoside (14.4 * 10
3
M
-1
cm
-1
under similar conditions) and the fact that no more than one modified residue
per 12 was present. In cases where oligonucleotides were self-complementary and individual strand concentrations could not be
spectroscopically measured, duplex concentrations were determined by estimating
[epsilon]
260
= 6600 per base. DNA was annealed by heating to 95oC for 10 min followed by slow cooling to 4oC. Solutions were thoroughly degassed prior to each melting
experiment. Measurements of UV absorbance at 260 nm were made in a 3 ml quartz
cell (1 cm path length) at temperature intervals of 1.0oC over the range 15-75oC, with precise internal cell temperature monitored by an
internal probe. Data reflect the average of three melting curves per
oligonucleotide duplex. Raw melting curves were converted to plots of single-strand fraction (
f
) versus temperature (
T
) using linear regression analysis to find lower and upper baseline slopes and
y
-intercepts of double- and single-stranded forms. Thermodynamic parameters were calculated
assuming a two-state transition. Values for [Delta]
H
(van't Hoff) were determined from analysis of plots of
f
versus
T
using the equation [Delta]
H
= 6
R
(
T
m
)
2
(slope at
T
m
).
T
m
is defined as the temperature where
f
= 0.5 and is not necessarily the point of greatest slope on the curve.
Entropies were estimated by [Delta]
S
= ([Delta]
H
/
T
m
)-
R
ln(
C
T
/
n
), where
R
is the gas constant at 25oC,
C
T
is the total oligonucleotide strand concentration and
n
= 4 for non-self-complementary strands or
n
= 1 for self-complementary strands (
26
). Free energy was calculated using [Delta]
G
= [Delta]
H
-
T
[Delta]
S
, where
T
is temperature (Kelvin). Uncertainty in
T
m
is estimated at +-0.5oC and in [Delta]
G
at +-10%.
Nucleoside stability studies
The relative rate of depurination in acidic medium of moA was compared with
those of A and BrA. Solutions of the nucleosides (~0.10 M) in 50% aqueous acetic acid were monitored for appearance of the
respective purine resulting from hydrolysis of the glycoside bond at 25 and 50oC. Relative amount of nucleoside remaining after various time intervals was
determined by removing aliquots of each hydrolysis mixture at various times
ranging from 0.5 to 24 h, quenching by addition of excess phosphate buffer and
subsequent analysis by reversed phase HPLC with UV detection at 220-300 nm.
RESULTS AND DISCUSSION
Nucleoside acid-stability studies
Depurination of adenosine residues in synthetic DNA is not normally a problem,
but in handling BrA-containing strands we have on occasion observed some degradation of the
oligonucleotide, particularly when attempting to remove a final trityl
protecting group following reversed phase HPLC purification. The final
detritylation is normally accomplished by treating the oligonucleotide with 80%
acetic acid for 20-30 min at ambient temperature. Under these conditions we found that the
free nucleoside BrA underwent >95% hydrolysis of the glycoside bond to give 8-bromoadenine. In order to compare relative resistance to depurination of
BrA, moA and A we prepared 0.1 M solutions of each of the free nucleosides in
50% acetic acid and followed the hydrolysis reactions at 25 and 50oC by HPLC. After 1 h at 25oC we found 22% conversion of BrA to 8-bromoadenine, whereas A and moA were each converted to their
corresponding purines by <4%. After 24 h BrA was completely hydrolyzed, while A and moA were only 38 and
25% depurinated, respectively. At 50oC BrA was completely depurinated, in 2 h, while A and moA were 75 and 67%
hydrolyzed, respectively.
Recently depurination rate constants for a variety of 8-substituted-2'-deoxypurine nucleosides in dilute acid have been
reported (
24
). These studies indicate that at 80oC in dilute aqueous acid (pH 5.2) BrA was depurinated at a rate >500 times
that of A. While moA was not examined in this study, it was found that 8-methylthio-2'-deoxyadenosine was depurinated at a rate 14 times that
of A under these same conditions. This study, as well as our own data, lead us
to conclude that while moA and A are similar in resistance to depurination, BrA
is significantly more susceptible. Since A and moA are similar in this regard,
we further expect they may have similar electronic properties, including base
pairing ability. The enhanced acid stability of moA relative to BrA suggests
that moA may be the better candidate for a
syn
conformational probe for A in duplex DNA.
Synthesis of phosphoramidite (5)
8-Methoxy-2'-deoxyadenosine was synthesized by a modification of a
literature procedure (
27
). Since acid-catalyzed depurination is likely to be a complication in automated
oligonucleotide synthesis, we elected to protect the N
6
of moA as the
N
,
N
-dimethylformamidine derivative, as previously described for the protection of the N
2
of 8-methoxy-2'-deoxyguanosine (moG) (
28
).
8-Bromo-2'-deoxyadenosine, (see Scheme
1
) (
1
), prepared by reaction of deoxyadenosine with bromine water (
29
), was then converted to moA (
2
) in 52% yield on treatment with sodium methoxide in methanol for 1 h at 45oC. Employing longer reaction times as given in the literature (18 h) we
were unable to obtain the desired product and over-substitution of the adenine by methoxide was apparent. The N
6
position was next protected as the formamidine derivative
3
by reaction with
N
,
N
-dimethylformamide di-
n
-butyl acetal in 91% yield and this intermediate was then utilized without
purification. Dimethoxytritylation (
30
) of the 5'-OH to give
4
was accomplished in acceptable yield (75% after chromatography). Finally,
reaction with 2-cyanoethyl-
N
,
N
-diisopropyl chlorophosphoramidite for 50 min at ambient temperature
followed by a methanol quench gave the desired moA phosphoramidite
5
in 50% yield after chromatography.
Phosphoramidite
5
was incorporated into oligonucleotides using standard solid phase automated
chemistry. Coupling efficiency at the moA step was estimated at >95% from
spectrophotometric determination of trityl cation (495 nm) formed upon
deprotection of the 5'-OH group. To establish that moA was successfully incorporated into
DNA without degradation or further modification, representative
oligodeoxynucleotides were analyzed by electrospray ionization mass
spectrometry (EIS-MS) (
31
). The expected negative ion series was observed in each case. Further evidence
for successful incorporation of the moA nucleoside was obtained by enzymatic
digestion of a representative test oligonucleotide with snake venom
phosphodiesterase and alkaline phosphatase, with the resulting nucleoside
mixture then analyzed by reversed phase HPLC. The experimental data yielded the
expected five peaks for (d)A, T, G, C and moA, with integrated areas in
agreement with the calculated nucleoside composition of the DNA sequence.
UV thermal denaturation studies of moA-containing duplexes with moA:T, moA:C and moA:G base pairs
Phosphoramidite
5
was incorporated into DNA duplexes across from three of the four naturally
occurring bases to form moA:T, moA:G and moA:C base pairs in the same sequence
environment. UV melting curves were used to compare the resulting stability of
these DNA duplexes with their A:T, A:G and A:C counterparts, as shown in Table
1
. In addition, we compared the effects of moA with BrA by investigating the
stability of BrA:T and BrA:G base pairs in sequences
3
and
6
respectively. For all three base pairs (in sequences
1
-
8
) containing moA or BrA instead of 2'-deoxyadenosine, considerably lower
T
m
values are observed, indicating significantly lower thermodynamic stability.
A comparison of sequences
1
and
2
, which contain A:T and moA:T base pairs, respectively, indicates a 6oC destabilization ([Delta][Delta]
G
= 2.4 kcal/mol) of the duplex containing the moA nucleoside. This amount of
destabilization is comparable with that observed for substitution of the normal
Watson-Crick A:T base pair in sequence
1
with a mismatched A:G base pair in sequence
4
, which resulted in a 7oC destabilization ([Delta][Delta]
G
= 3.1 kcal/mol) of the duplex. In the case of an A:T base pair the A
preferentially adopts the
anti
conformation to preserve Watson-Crick hydrogen bonding. An 8-methoxy or 8-bromo substituent on the A would be expected to destabilize
the structure, since the nucleotide must be accommodated in the duplex in the
anti
conformation, with considerable energy cost associated with the steric demand
of the bulky methoxy or bromo substituent. The amount of destabilization ([Delta]
T
m
[approx] 6oC) that is observed in this case for moA or BrA substitution in an A:T
base pair is strikingly similar to that of BrG substitution into sites which
contained the preferred G
anti
conformation in the telomeric G quartet structure ([Delta]
T
m
[approx] 6oC) (
22
).
.
T
m
(oC) and [Delta]
G
25
(kcal/mol) values for mismatch-containing undecamers
No
Sequence
pH
T
m
-[Delta]
G
25
5'-GAGCT
X
GTGGC-3'
3'-CTCGA
Y
CACCG-5'
1
X = A
7
53
15.9
Y = T
2
X = moA
7
47
13.5
Y= T
3
X = BrA
7
48
14.8
Y = T
4
X = A
7
46
12.8
Y = G
5
43
11.5
5
X = moA
7
44
11.6
Y = G
5
44
12.0
6
X = BrA
7
43
11.8
Y = G
7
X = A
7
44
11.5
Y = C
5
46
12.5
8
X = moA
7
41
12.2
Y = C
5
41
11.4
UV absorbance measurements (260 nm) were recorded at temperature increments of
1.0oC. Solutions contained 8-11 [mu]M duplex oligonucleotide, buffer pH 5.0 or 7.0 as indicated, 1.0
M NaCl, 0.010 M phosphate, 0.001 M EDTA. Data represent the average from three
melting curves per duplex.
At pH 7, substitution of moA for A in an A:C mismatch (
7
and
8
) results in only a 3oC destabilization of the duplex. At neutral pH the A:C mismatch has a
single hydrogen bond and can adopt a variety of structures and, therefore, may
be able to accommodate the moA substitution with little additional
destabilization. At pH 5 the A:C pair is usually more stable, since protonation
of the A can result in formation of two hydrogen bonds to form an AH
+
anti
:C
anti
base pair. In this case the destabilizing effect of moA substitution is larger
than at pH 7, due to the preference for the A
anti
conformation in this base pair (
32
). The A:G mismatch has been shown to exist in a variety of conformations,
depending upon the sequence environment and pH. Comparing the moA- and BrA-containing sequences
5
and
6
to the normal A-containing sequence
4
we find a small amount of destabilization (2-3oC) for those containing the 8-substituted adenine. This may indicate a slight preference for
the
anti
conformation by the A in this sequence environment, however, the A:G mismatch-containing sequence
4
has not been structurally characterized by NMR or X-ray crystallography.
Substitution of moA for A in structurally characterized A:G-containing duplexes
In order to investigate the effects of moA substitution on the glycosidic
conformation in the A:G mismatch we examined the effects on duplex stability of
moA substitution in a variety of sequences where the structure of the mismatch
had previously been defined by NMR or X-ray crystallography. The relevant sequences and
T
m
values are shown in Table
2
. Sequence
9
is related to sequence
4
by inversion at the A:G mismatch. This sequence contains an A:G pairing at a
position which is a known `hot-spot' for G -> T transversion mutations in the K-
ras
oncogene sequence (
33
). Furthermore, sequence
9
has been shown by NMR to exist in the A
anti
:G
anti
conformation at pH 7, with a p
K
a
of 6 for conversion to the AH
+
anti
:G
syn
conformation (
10
). Substitution of moA for A in this sequence (
10
) induces considerable destabilization ([Delta]
T
m
= 6oC, [Delta][Delta]
G
= 1.3 kcal/mol) at pH 7, consistent with the preference for an
anti
conformation by A in this sequence. A similar trend is observed in a comparison
of duplex
11
with
12
. X-ray characterization of crystals of
11
grown at pH 6.6 have shown that the A:G mismatch in this sequences adopts the
AH
+
anti
:G
syn
structure (
11
). Consistent with these results, substitution of moA for A in this sequence (
12
) causes significant destabilization, with a [Delta]
T
m
value of at least 5oC per mismatch. The
T
m
value for
12
was too low for us to precisely determine; the values in Table
2
represent upper limits for the
T
m
and [Delta]
G
values for this oligonucleotide. In both of these examples substitution of moA
into A positions where the
anti
conformation is preferred results in destabilization of the duplex. The
magnitude of this destabilization ([Delta][Delta]
G
[approx] 2-3 kcal/mol) is similar to that observed for moA substitution in an
A:T base pair.
In contrast, substitution of moA for A in the self-complementary dodecamers
13
or
16
results in increases in stability. X-ray analyses of duplexes
13
and
16
have shown the adoption of the unusual A
syn
:G
anti
conformation for the mismatch (
7
,
8
). Substitution of moA for A (
14
)
results in a minor increase in stability (1oC) relative to duplex
13
, rather than the decrease
in stability which had been observed previously. This indicates that in this
sequence environment the
syn
preference of the moA in the moA:G base pair is easily accommodated. This
result is completely consistent with the structural studies on duplex
11
, indicating the
syn
conformation for the A in the A:G base pair (
11
). Substitution of moA for A in duplex
16
gives
17
, which shows an even larger increase in stability of the duplex ([Delta]
T
m
= 6oC/mismatch, [Delta][Delta]
G
= 2.3 kcal/mol/mismatch). These results also correlate with the X-ray analysis, which indicates a
syn
preference for the adenine in the A:G pair. Interestingly, the degree of
stabilization observed in duplex
17
compared with
16
is of the same magnitude as the degree of destabilization for duplex
10
compared with
9
.
.
T
m
(oC) and [Delta]
G
25
(kcal/mol) values for mismatch-containing duplexes and conformation of the mismatched base pair where the
structure is known
No
Sequence
T
m
-[Delta]
G
25
Conformation
Reference
5'-GAGCT
X
GTGGC-3'
3'-CTCGA
Y
CACCG-5'
9
X = G
47
12.3
G
anti
:A
anti
10
Y = A
(NMR)
10
X = G
41
11.0
Y = moA
(5'-CGC
X
AATT
G
GCG-3')
2
11
X = A
38
9.2
G
syn
:A
anti
11
(X-ray)
12
X = moA
<28
<7.6
(5'-CGC
Y
AATT
X
GCG-3')
2
13
X = A
55
11.0
G
anti
:A
syn
8
Y = G
(X-ray)
14
X = moA
56
11.2
Y = G
15
X = moA
55
11.7
Y = I
(5'-CGC
X
AGCT
Y
GCG-3')
2
16
X = A
42
9.7
G
anti
:A
syn
7
Y = G
(X-ray)
17
X = moA
54
14.3
Y = G
18
X = moA
51
11.3
Y = I
In duplex
13
incorporation of moA for A does not destabilize the structure, but neither is
any significant stabilization of the structure observed. This is likely due to
the ability of A:G mismatches to adopt a variety of structures. In sequence
13
it is likely that in solution both the A
anti
:G
anti
and A
syn
:G
anti
conformations show similar stability, while in duplex
16
there is a preference for the A
syn
:G
anti
conformation of the mismatch over the A
anti
:G
anti
conformation and therefore the presence of moA results in a significant
increase in stability. Theoretical studies have suggested that the energy
difference between the different conformers of the A:G mismatch is ~1 kcal/mol (
34
-
36
), suggesting that moA substitution could influence the conformational
preference in a given sequence. In the case of duplex
16
the moA nucleotide is already poised to adopt the
syn
conformation and, since the sequence prefers the A
syn
:G
anti
conformer, an increase in stability is observed. In duplex
13
the duplex can adopt both structures with very similar stabilities and,
therefore, if the thermodynamic cost is `prepaid' (as in the case of the moA
duplex), the moA
syn
:G
anti
base pair is not destabilizing. In contrast, in sequences which prefer the A
anti
conformation an appreciable energy investment may be needed to induce a base
with a
syn
preference into an
anti
conformation so as to be best accommodated in the duplex.
Inspection of the structural elements of A:G mismatches gave us concern that
substitution at the 8-position in moA might result in a steric clash between the methoxy group
and the 2-amino group of the guanine. If so, this could be negating some of the
expected stabilization we would observe for moA substitution in duplex
13
(to give
14
). To test this hypothesis we investigated the thermal stability of duplexes
15
and
18
, in which inosine (i.e. guanosine without the 2-amino group) is substituted for guanosine. In moA:I-containing duplex
15
we observed a small degree of destabilization relative to duplex
14
, indicating that the 2-amino group of the guanine exerts little influence in this conformation,
either in steric demand or in hydrogen bonding. In the case of moA:I-containing duplex
18
we observed a [Delta]
T
m
of -3oC relative to moA:G analog
17
, suggesting that the 2-amino group of guanine may contribute to duplex
stability
in some manner, perhaps through bridging intrastrand hydrogen bonds. Thus the
moA nucleoside can fairly easily be accommodated in the DNA helix in the
syn
conformation and steric clashes with the 2-amino group are not problematical, probably because base pair propeller twist allows accommodation of the methoxy substituent.
Conclusions
We have developed a convenient synthesis of moA-containing oligonucleotides and found that this modified nucleoside can be
accommodated within duplex DNA. In the case of A:G mismatches our results have
shown that moA substitution for A can be stabilizing or destabilizing,
depending upon sequence. The effects on stability of moA:G versus A:G duplexes
correlates with the structurally characterized conformational preferences of
the A nucleoside in the mismatch. moA substitution for A results in duplex
stabilization in sequences with a preferred A
syn
glycosidic conformation and conversely moA substitution for A results in duplex
destabilization in sequences with a preferred A
anti
glycosidic conformation. Based on these results, moA substitution could be used
to diagnose the conformational preferences of A nucleosides in DNA. For
example, moA substitution into sequence
4
to give sequence
5
results in a small amount of destabilization. No structural information on this
sequence exists, however, based on the destabilizing effect of moA
substitution, it is likely that the A:G mismatch in this sequence exists in the
G
anti
:A
anti
conformation. Interestingly, the stability of sequence
9
, which is related to sequence
4
by inversion at the A:G mismatch, is much more sensitive to moA substitution,
indicative of slight differences between these sequences. Thus moA substitution
may also be useful in locating subtle differences in structural properties of
unusual DNA structures, like A:G mismatches. DNA repair glycosylases which
recognize damaged or mismatched DNA may be sensitive to such subtle differences
in conformational preferences.
Our data have also indicated that the ability of the moA conformational probe to
adopt the
syn
conformation is dictated by the base pair and sequence context and is most
likely to occur in unusual circumstances, such as in non-standard base pairs. NMR studies of moG-containing oligonucleotide duplexes with moG opposite C have shown
that the moG can be accommodated in the
anti
conformation to preserve Watson-Crick base pairing (
22
). Our data suggest that moA could also exist in the
syn
conformation in moA:G base pairs in specific sequence contexts. At the present
time there are no solution structures of A
syn
:G
anti
base pairs in DNA. NMR experiments are currently underway in our laboratory to
establish the structures of moA:G-containing duplexes where our melting data in conjunction with the
previously established crystal structures of the corresponding A:G duplexes
implicates the presence of the A
syn
:G
anti
conformation.
ACKNOWLEDGEMENTS
This work was supported by NIH grant CA67985 and a Young Investigator Award to
SSD from the Arnold and Mabel Beckman Foundation. RGE was supported by a UC
Biotechnology Training Grant.
REFERENCES
1 Wang,A.H.J., Quigley,G.J., Kolpak,F.J., Crawford,J.L., van Boom,J.H., van der Marel,G. and Rich,A. (1979) Nature, 282, 680-686.MEDLINE Abstract
2 Wang,A.H.J., Quigley,G.J., Kolpak,F.J., van Boom,J.H., van der Marel,G. and Rich,A. (1981) Science, 211, 171-176.MEDLINE Abstract
3 Smith,F.W. and Feigon,J.W. (1992) Nature, 356, 164-168.MEDLINE Abstract
