ABSTRACT
A dialkyl-substituted anthraquinone derivative was synthesized and ligated to a
sequence-directing oligodeoxynucleotide to examine its efficiency and specificity for cross-linking to complementary sequences of DNA. The anthraquinone appendage stabilized spontaneous hybridization of the target and probe sequences through non-covalent interactions, as indicated by thermal denaturation
studies. Covalent modification of the target was induced by exposure to near UV
light ([lambda]
> 335 nm) to generate cross-linked duplexes in yields as great as 45%. Reaction was dependent on the first unpaired nucleotide extended beyond the duplex formed by association of the target and probe. A specificity of C > T > A
[approx]
G was determined for modification at this position. The overall site and
nucleotide selectivity seems to originate from the chemical requirements of
cross-linking and does not likely reflect the dominant solution structure of the
complex prior to irradiation.
Therapeutic intervention at the level of nucleic acids holds considerable promise, but technical challenges remain. Application of antisense and antigene strategies have been limited
in vivo
, for example by problems with delivery and stability of the necessary nucleic
acid derivatives (
1
-
5
). The challenges associated with stability alone are numerous and range from
concerns of metabolic lifetime to target affinity and reversibility of target association. Many significant advances have relied on modification of a reagent's oligonucleotide backbone, even though the resulting derivative often exhibits weaker binding to its target sequence. However, affinity can usually be regained and metabolic stability further enhanced by appending an intercalator or alkylating agent to
a terminus of the oligomer (
6
). Psoralen has been used widely in this manner, since it can covalently cross-link single- and double-stranded nucleic acids upon exposure to near UV light (
7
-
10
). Our research has focused on a series of related reagents that similarly express an inducible ability to alkylate nucleic acids. In each
case, highly reactive quinone methide intermediates have been generated under alternative control of light (
11
,
12
), enzymatic reduction (
12
,
13
), anion concentration (
14
) and the biological target itself (
15
). Much of our initial effort relied on a series of naphthoquinone derivatives;
but due to their general instability, target modification was inefficient (
11
-
13
). This report now describes the synthesis and characterization of an anthraquinone derivative that exhibits superior properties as an affinity reagent and retains
a potential for application
in vivo
.
Anthraquinone derivatives have long been associated with nucleic acid binding
and cancer chemotherapy. The activity of many natural and synthetic anthracycline antibiotics is derived from their anthraquinone component (
16
). Numerous bis(aminoalkyl) anthraquinones have also been prepared as anticancer agents based on their
strength as intercalators (
17
-
19
) and this binding property has since been used to promote platinum-DNA reaction through construction of anthraquinone-cisplatin conjugates (
20
). Anthraquinone derivatives have similarly been coupled to antisense and antigene reagents in order to strengthen their association with target DNA,
as well as to enhance their cellular uptake and decrease their sensitivity to
nucleases (
21
-
27
).
The oxidation-reduction and photochemical activity of anthraquinones offer additional opportunities for target recognition and modification. Bis(aminoalkyl)anthraquinones act as photosensitizers to generate singlet oxygen and may find use in photodynamic therapy (
28
). Other anthraquinones are capable of photochemically oxidizing DNA by electron transfer and hydrogen atom abstraction (
29
-
31
). Finally, 1-methylanthraquinone readily photoenolizes to form a transient quinone methide (
32
) in an equivalent manner to the naphthoquinone derivatives previously applied to site-directed modification of DNA (
11
-
13
). The appropriate anthraquinone-oligodeoxynucleotide conjugate thus has the potential to stabilize target
hybridization and inducibly modify adjacent nucleotide residues via a number of
alternative pathways. Our studies began with the photochemical activity of a dialkyl-substituted anthraquinone.
All oligodeoxynucleotides, including the aminolinker-containing derivative, were synthesized using standard solid phase automated procedures. Coupling between hydrophobic activated esters and hydrophilic
DNA remains one of the more difficult tasks in preparing sequence-directed reagents. Our laboratory has consistently had greatest success by mixing equal volumes of DNA dissolved in 0.25 M 3-(
N
-morpholino)propane sulfonate (MOPS), pH 7.5, and the activated ester dissolved in dimethylformamide (DMF).
After incubation at room temperature, the coupled product was purified by
reverse-phase chromatrography in a standard yield of 34%.
Numerous past reports have characterized the strong tendency for anthraquinone
derivatives to intercalate into duplex DNA (see for example
16
-
27
) and therefore a similar interaction was expected for the duplexes formed by
the dimethylanthraquinone-oligodeoxynucleotide conjugate and its complementary target. This non-covalent binding was characterized by an increase in the thermal
melting temperature (
T
m
) of duplex DNA in the presence of the quinone (Table
1
). Even the unligated dimethylanthraquinone stabilized duplex formation, as demonstrated by its ability to raise the duplex
T
m
by 7oC. Further stabilization was gained by covalently attaching the
anthraquinone to one DNA strand. Related naphthoquinone conjugates did not
exhibit this activity and covalent attachment of a naphthoquinone had no effect
on duplex hybridization (
12
). Interestingly, psoralen intercalates into duplex DNA, but it does not seem to
stabilize hybridization when linked as a oligodeoxynucleotide-psoralen conjugate (
7
,
34
-
36
).
Table 1
A series of oligonucleotide targets,
OT1
-
OT4
, were designed to test the specificity of the anthraquinone conjugate
OAQ
. Both double- and single-stranded regions were provided as possible reaction sites by
extending the target strand beyond the conjugate sequence (Scheme
2
2). The nucleotide directly adjacent to the duplex was also varied in
anticipation of its key role in modification. This site had been the sole
target of alkylation in a related methylnaphthoquinone system (
12
) and dominated photocyclization in certain psoralen-based systems (
35
). Similar to psoralen and naphthoquinone, anthraquinone absorbs near UV light
(340-360 nm), which allows for its selective activation in the presence of
DNA. Irradiation at [lambda] > 335 nm was consequently used to initiate reaction following
hybridization of
OAQ
and
OT1
-
OT4
and a combination of electrophoresis and autoradiography was used to
characterize the products. Target alkylation would covalently link the duplex
strands and generate derivatives of high molecular weight and low mobility,
whereas oxidative scission would generate strand fragments of low molecular
weight and high mobility.
Only target alkylation was observed over a 2 h irradiation and its yield
approached 40% for a stoichiometric mixture of
OAQ
and
OT1
(Fig.
1
A). Product formation was limited in part by a competing photochemical
decomposition of
OAQ
, since a 1 h pre-irradiation of
OAQ
inhibited subsequent modification of the target by ~50%. Still, the anthraquinone chromophore was much more stable than an
equivalent naphthoquinone analog, which completely decomposed within the first
10 min of irradiation (
11
). Modification of each target sequence was examined under the same standard
conditions as in Figure
1
A and summarized in Figure
1
B. Greatest yields were obtained for N = C (
OT1
) and this was not unique to sequences terminating in a single-stranded sequence -CAG. An alternative target,
OT5
(5'-[
32
P]AGTGCCACCTGACGTCTAAG), with a single-stranded extension of -CTAAG was cross-linked with the same efficiency as
OT1
. In contrast, only slight reaction was detected for N = G or A (
OT3
and
OT4
), even though these nucleobases are known to react with quinone methide intermediates via their
exo
-amino groups (
37
-
39
). The moderate yields of cross-linking for N = T were most surprising, since this base does not even contain an
exo
-amino group.
Certain alkylation products of DNA can be identified by their characteristic lability to heat, alkali or other treatments. Accordingly, the stability of the high molecular weight derivative formed by
OT1
/
OAQ
was examined under a variety of conditions after its isolation by preparative
gel electrophoresis. In contrast to the cross-link formed by a naphthoquinone conjugate (
11
), the equivalent product of the anthraquinone reaction was relatively inert to
standard manipulations and did not readily decompose to a material of similar
gel mobility to the starting material (Fig.
2
, lane 3). Under additional exposure to UV light ([lambda] > 335 nm), the anthraquinone product began to decompose to form
oligonucleotides with mobilities equivalent to that of the starting material
OT1
and its fragment lacking the 3'-terminal -CAG (lane 4). Strand scission at the five bases on the 5'-side of the -CAG was also observed and may have
resulted from anthraquinone's ability to act as a photochemical oxidant (
30
,
31
).
The covalent linkage generated between
OT1
and
OAQ
remained stable under acidic conditions (pH 2; Fig.
2
, lane 5). However, the expected depurination reaction promoted by acid was detected after subsequent treatment with piperidine (90oC; lane 6). Other fragments were generated as well by these conditions and once again
found comparable in gel mobility with the original target and its derivative
lacking the 3'-terminal -CAG. These two species were not necessarily the result of sensitivity to acid or base (lane 7), since heat alone was
sufficient for their production (lane 8). This fragmentation pattern suggested that at least some alkylation occurred at the single-stranded C extended from the duplex formed by
OT1
/
OAQ
. A general specificity for this site was further implicated by the wide range
of yields observed for cross-linking targets
OT1
-
OT4
that only differed at the first unpaired base.
The major site of target alkylation was identified by a method of hydroxyl
radical footprinting that had previously been used to characterize numerous
interstrand reactions (
13
,
40
,
41
). This radical provides a non-specific method of strand scission via oxidation of the phosphoribose
backbone. Fragmentation of the modified and unmodified duplexes was compared by
gel electrophoresis (and densitometry) to determine the position at which the
two profiles converge. This in turn indicated the site of cross-link on the labeled target strand (
42
). The hydroxyl radical cleavage patterns for the covalent and non-covalent complexes formed by [
32
P]
OT5
and
OAQ
were comparable through C16, the first unpaired nucleotide (Fig.
3
). An equivalent result was observed for [
32
P]
OT1
and
OAQ
(data not shown) and confirmed a preferential modification of this nucleotide by the anthraquinone.
The anthraquinone derivative described in these studies was developed for site-specific and inducible alkylation of nucleic acids and designed to overcome deficiencies with a previous naphthoquinone analog. Both chromophores were able to cross-link target sequences when irradiated at [lambda] > 335 nm and no oxidative degradation of the nucleotides
was evident until prolonged exposure to light. As expected, the anthraquinone
appendage was more stable and provided higher yields of target modification
than the original naphthoquinone system. Although each quinone reacted at the
first nucleotide extended beyond the duplex formed by probe-target hybridization, their nucleotide specificity differed. The naphthoquinone derivative selectively modified nucleotides containing
exo
-amino groups, as predicted for a reaction dependent on a quinone methide intermediate (
12
). In contrast, the anthraquinone derivative appeared selective for the
pyrimidines C and T and independent of the presence of an
exo
-amine. The mechanism of this process is currently under investigation and
may provide new opportunities in the design of site-directed modification of biopolymers.
1
H-NMR spectra were recorded on a QE-300 (General Electric) at 300 MHz and chemical shifts ([delta], p.p.m.) were determined from an internal standard of CHCl
3
. Low resolution mass spectra (LRMS) were measured using electron impact with an
HP5980A (Hewlett-Packard); high resolution mass spectra (HRMS) were measured with a MS890
(Kratos). Synthetic procedures were performed under a nitrogen atmosphere and
solvents were freshly distilled (CH
3
CN, diethyl ether and CH
2
Cl
2
over calcium hydride and THF over sodium). Silica gel (230-400 mesh) for flash column chromatography was obtained from Fisher
Scientific Co. Pyridinium dichromate (
43
), tris(triphenylphosphine)rhodium(I) chloride (
44
), 1,2-bis(1-oxo-2-butynyl)benzene
1
(
33
) and 1,4-dimethylanthraquinone (
45
) were prepared according to published procedures. All other organic chemicals,
buffers and reagents were obtained as the highest commercial grade available
and used without further purification. T4 kinase was obtained from US Biochemical and Gibco BRL. [[gamma]-
32
P]ATP was purchased from Amersham Corporation.
1,2-Bis(1-oxo-2-butynyl)benzene
1
(37 mg, 170 [mu]mol) was added to tris(triphenylphosphine)rhodium(I) chloride (180 mg, 190 [mu]mol) in 2 ml xylene under nitrogen at 140oC. After stirring for 50 h, the reaction mixture was evaporated,
redissolved in ethyl acetate and subjected to silica gel flash chromatography.
The desired product was isolated in a low but predictable (
33
) yield of 15% (24 mg).
1
H NMR (CDCl
3
) [delta] 2.10 (6H, s), 8.04-7.37 (34H, m).
The rhodium complex above (24 mg, 26 [mu]mol) was added to 5-acetoxy-1-pentyne (4 mg, 31 [mu]mol, prepared by esterification of 4-pentyn-1-ol with acetic anhydride) in 2 ml
benzene. The reaction mixture was maintained at 50oC for 15 h and then evaporated to dryness. The residue was redissolved in hexane:ethyl acetate (1:1) and subjected to silica gel flash chromatography to yield 4
mg product (45% yield). m.p. 119-120oC (uncorrected);
1
H NMR (CDCl
3
) [delta] 1.91 (2H, m), 2.11 (3H, s), 2.75 (6H, s), 2.76 (2H, t), 4.19 (2H,t),
7.32 (1H, s), 7.65 (2H, m), 8.11 (2H, m); LRMS
m
/
z
(relative intensity) 336 (M
+
, 87.4), 261 (100), 249 (23.5), 235 (29.46), 189 (24.7); HRMS calculated for C
21
H
20
O
4
(M
+
) 336.1362, found 336.1362.
The acetoxy derivative
3
(4 mg, 14 [mu]mol) was dissolved in 4 ml 5% aqueous K
2
CO
3
and MeOH (1:1) and stirred at room temperature for 6 h. The mixture was then
evaporated to 2 ml and extracted with CHCl
3
and H
2
O. The organic layer was dried over MgSO
4
and filtered and the solvent was removed under reduced pressure. The crude
product was purified via silica gel flash chromatography to yield ~4 mg product alcohol.
1
H NMR (CDCl
3
) [delta] 1.64 (2H, m), 2.69 (6H, s), 2.95 (2H, t), 3.42 (2H, m), 4.45 (1H, t),
7.31 (1H, s), 7.55 (2H, m), 8.04 (2H, m); LRMS
m
/
z
(relative intensity) 294 (M
+
, 6.13), 279 (100), 249 (17.3), 235 (33.6), 205 (9.49); HRMS calculated for C
19
H
18
O
3
(M
+
) 294.1256, found 294.1243.
The alcohol
4
(4 mg, 14 [mu]mol) and PDC (184 mg, 49 [mu]mol) were dissolved in DMF (2 ml) and stirred for 10 h at room
temperature. The reaction was then mixed with water (15 ml) and extracted with
ether. The organic phase was dried over MgSO
4
, reduced in volume and subjected to silica gel flash chromatography to yield
the desired acid quantitatively.
1
H NMR (CDCl
3
) [delta] 2.77 (6H, s), 4.30 (2H, t), 4.69 (2H, t), 7.36 (1H, s), 7.68 (2H, m),
8.14 (2H, m), 10.58 (1H, s); LRMS
m
/
z
(relative intensity) 308 (M
+
, 84.59), 291 (14.8), 262 (100), 249 (30.4), 235 (88.6), 220 (14.6), 205 (22.0);
HRMS calculated for C
19
H
16
O
4
(M
+
) 308.1049, found 308.1046.
The acid derivative
5
(4 mg, 13 [mu]mol),
N
-hydroxysuccinimide (18 mg, 16 [mu]mol) and CDI (2 mg, 13 [mu]mol) were dissolved in DMF (40 [mu]l) and maintained at 4oC for 14 h. The reaction mixture was then evaporated to
dryness. The residue was redissolved in CH
2
Cl
2
(3 ml), extracted three times with H
2
O and dried over MgSO
4
. The product was purified by silica gel flash chromatography and yielded 4 mg
activated ester product (75%).
1
H NMR (CDCl
3
) [delta] 2.72 (6H, s), 2.80 (4H, s), 2.88 (2H,t), 3.17 (2H, t), 7.37 (1H, s),
7.67 (2H, m), 8.13 (2H, m).
All oligodeoxynucleotides were synthesized via standard solid-phase cyanoethyl phosphoramidite chemistry (Clontech) and the unmodified species were purified by anion exchange chromatography (Mono Q,
Pharmacia) under strongly denaturing conditions (pH 12) (
46
). The hexamethyleneaminolinker was attached in the last step of the automated
synthesis using a monomethoxytrityl-protected precursor (Clontech). The protecting group was removed by
treating the crude product with 80% acetic acid for 30 min under ambient
conditions and then separated from the by-products with ether extraction. The crude aminolinker preparation was neutralized, dried and used directly in the coupling reaction below.
DNA concentrations were calculated per mol oligonucleotide from their [epsilon]
260
values, estimated from the sum of nucleotide absorptivity as affected by
adjacent bases (
47
). The target sequences
OT1
-
OT5
were radiolabeled at the 5'-terminus by incubation with [[gamma]-
32
P]ATP (50 [mu]Ci, sp. act. 3000 Ci/mmol) and T4 polynucleotide kinase (10 U) in kinase
buffer (0.05 M Tris-HCl, pH 7.6, 0.01 M MgCl
2
, 5 mM DTT, 0.1 mM EDTA and 0.1 mM spermidine) for 45 min at 37oC (
48
). This mixture was then diluted to 2 ml with water and centrifuged in an Amicon
concentrator (10 000 mol. wt cut-off) for 40 min to remove the excess salt and unincorporated [[gamma]-
32
P]ATP. This process of dilution-centrifugation was repeated twice more to provide ~5 [mu]Ci phosphorylated oligodeoxynucleotide.
The oligodeoxynucleotide aminolinker (1.2 A
260
units, 9 nmol) was dissolved in 0.25 M 3-(
N
-morpholino)propane sulfonic acid, pH 7.5, (20 [mu]l) and added to a DMF solution (20 [mu]l) of the activated anthraquinonyl ester
6
(2 mg, 5 [mu]mol) and left undisturbed at room temperature for 4 h. The oligodeoxynucleotide conjugate was purified by reverse-phase HPLC (C-18 Spherex column) using a gradient of 45 mM triethylammonium
acetate, pH 6, 10% acetonitrile to 35 mM triethylammonium acetate, pH 6, 30% acetonitrile over 30 min at a flow rate of 1 ml/min. The product
OAQ
was isolated in 34% yield, estimated by the recovery of A
260
units.
Equal concentrations of complementary strands (2.2 [mu]M) in the presence and absence of 1,4-dimethylanthraquinone (2.2 [mu]M) were mixed in potassium phosphate buffer (10 mM, pH 7) and NaCl
(100 mM), heated to 80oC and then cooled slowly to 4oC over 4 h. Samples were then equilibrated in a temperature controlled UV spectrophotometer (Perkin-Elmer [lambda]-5) and absorbance at 260 nm was monitored as a function of
temperature from 10 to 90oC. The
T
m
values were determined from a plot of absorbance (A
260
) versus temperature and assigned as the temperature at 1/2([Delta]A
260
).
Equimolar solutions of
OAQ
and a complementary strand
OT1
-
OT5
(typically 2.2 [mu]M) were dissolved in potassium phosphate (10 mM, pH 7) and annealed by first
heating the solution to 65oC (10 min) and then allowing it to cool slowly back to room temperature
over 3-4 h. The resulting solutions were transferred to conical Pyrex tubes and
irradiated under ambient conditions at the focal point of a 150 W xenon arc
lamp using a 335 nm long pass band filter. Samples were dried, denatured by
addition of 80% formamide and analyzed by denaturing polyacrylamide gel (20%, 7 M urea) electrophoresis. Approximately 2 nCi labeled material was applied to each gel lane. Alkylation was detected by
autoradiography and quantified by excising portions of the gel and measuring their radioactivity by scintillation counting.
The complementary targets and probe (2 [mu]M) were annealed, irradiated (1 h) and subjected to gel electrophoresis
under standard conditions. The cross-linked DNA was located by autoradiography and extracted by routine crush
and soak methods (
48
). This DNA was then precipitated in 75% EtOH and analyzed for purity by gel
electrophoresis.
The cross-linked duplexes were subject to acid, base, heat and irradiation as
described in Figure
2
. Hydroxyl radical footprinting followed published procedures (
42
). Cross-linked products (60 nCi) were dissolved in 2 mM Tris buffer, pH 7, and treated with (NH
4
)
2
Fe(SO
4
)
2
(1 mM), EDTA (4 mM, pH 8) and sodium ascorbate (2 mM). Reactions were initiated
by addition of H
2
O
2
(1 mM), incubated for 3 min at room temperature and quenched with thiourea (80
mM). The resulting product fragments were then analyzed by denaturing gel
electrophoresis, autoradiography and densitometry.
This research was generously supported in part by the Center for Biotechnology,
State University of New York at Stony Brook, in conjunction with the New York
State Science and Technology Foundation.
*To whom correspondence should be addressed at present address: Department of
Chemistry and Biochemistry, University of Maryland, College Park,
MD 20742-2021, USA
+
Present address: Department of Life Science, Yongin University, Samga-ri Yongin-up, Yongin-gun, Kyonggi-do 449-714, Korea
Oligodeoxynucleotide
a
T
m
(oC)
5'-AGTGCCACCTGACGTCAG (
OT1
)
47
3'-TCACGGTGGACTGCA
5'-AGTGCCACCTGACGTCAG (
OT1
)
54
3'-TCACGGTGGACTGCA
+ 1,4-dimethylanthraquinone
5'-AGTGCCACCTGACGTCAG (
OT1
)
57
3'-TCACGGTGGACTGCA-Me
2
AQ (
OAQ
)
REFERENCES
Return