A Transient Product of DNA Alkylation can be Stabilized by Binding Localization

Article abstract of DOI:10.1021/ja036943o

A 9-aminoacridine conjugate of a silyl-protected bis(acetoxymethyl)phenol (bisQMP) was synthesized and evaluated as an inducible cross-linking agent of DNA to test our ability to harness the chemistry of reactive quinone methide intermediates (QM). The ac

Full text of DOI:10.1021/ja036943o

Published on Web 10/28/2003
A Transient Product of DNA Alkylation Can Be Stabilized by
Binding Localization
Willem F. Veldhuyzen, Praveen Pande, and Steven E. Rokita*,†
Contribution from the Department of Chemistry and Biochemistry, UniVersity of Maryland,
College Park, Maryland 20742
Received June 27, 2003; E-mail: rokita@umd.edu
Abstract: A 9-aminoacridine conjugate of a silyl-protected bis(acetoxymethyl)phenol (bisQMP) was
synthesized and evaluated as an inducible cross-linking agent of DNA to test our ability to harness the
chemistry of reactive quinone methide intermediates (QM). The acridine component was chosen for its
ability to delivery an appendage to the major groove of DNA, and the silyl-protected component was chosen
for its ability to generate two quinone methide equivalents in tandem upon addition of fluoride. This design
created competition between reaction of (1) the 2-amino group of guanine that reacts irreversibly to form
a stable QM adduct and (2) the more nucleophilic N7 group of guanine that reacts more efficiently but
reversibly to form a labile QM adduct. This lability was apparently compensated by co-localization of the
N7 group and QM in the major groove since the N7 adduct appeared to dominate the profile of products
formed by duplex DNA. The controlling influence of acridine was also expressed in the sensitivity of the
conjugate to ionic strength. High salt concentration inhibited covalent reaction just as it inhibits intercalation
of the cationic acridine. As expected for QM formation, the presence of fluoride was indeed necessary for
initiating reaction, and no direct benzylic substitution was observed. The conjugate also cross-linked DNA
with high efficiency, forming one cross-link for every four alkylation events. Both alkylation and cross-
linking products formed by duplex DNA were labile to hot piperidine treatment which led to ∼40% strand
scission and ∼50% reversion to a material with an electrophoretic mobility equivalent to the parent DNA.
All guanines exhibited at least some reactivity including those which were recalcitrant to cross-linking by
an oligonucleotide-bisQMP conjugate designed for triplex formation [Zhou, G.; Pande, P.; Johnson, A. E.;
Rokita, S. E. Bioorg. Med. Chem. 2001, 9, 2347-2354].
Alkylating agents that react irreversibly with DNA generate
product profiles derived from kinetic competition between
accessible nucleophiles. Accordingly, reagents such as dimethyl
sulfate and ethyl nitrosourea alkylate the most reactive N and
O nucleophiles independent of the relative stability and ther-
modynamics of the nascent products.1 In contrast, alkylating
agents that react reversibly with DNA may form kinetic
intermediates that ultimately give way to an equilibrium
distribution of products. Highly functionalized synthetic and
biological compounds often bind to regions of DNA that
facilitate formation of the thermodynamically favored products
and avoid generating covalent intermediates associated with
alternative kinetic products.2
Notable exceptions illustrate that covalent attachment to DNA
need not prevent migration of particular adducts from sites of
kinetic reactivity to those of thermodynamic stability established
by optimum covalent and noncovalent association. For example,
ecteinascidin 743 (Et 743) alkylates the 2-amino group of
guanine (GN2) within 5′-AGT and 5′-AGC sequences with equal
efficiency.3 However, release of Et 743 is faster from the AGT
vs AGC sequence and consequently adducts accumulate at AGC
sequences for which association is preferred. This reversibility
may also provide a mechanism for recycling this promising drug
candidate once its DNA adduct is removed by cellular repair
processes.3 The toxicity of malondialdehyde may likewise be
enhanced to our detriment by its ability to form adducts with
guanine reversibly and return to DNA after excision of its
pyrimidopurinone adduct.4,5 Reversible adduct formation is
particularly critical for efficient cross-linking of duplex DNA
since initial alkylation anchors an intermediate to a sequence
that may not necessarily contain the appropriate nucleotides for
subsequent cross-linking. This may in part explain the low levels
of cross-linking vs monoalkylation achieved by reagents such
as N-mustards6 and diepoxybutane7 that operate under kinetic
control. In contrast, bizelesin containing two cyclopropylpyr-
roloindole units efficiently cross-links duplex DNA through
reaction at the N3 position of two adenosines (AN3) separated
by four base pairs.8 This efficiency is certainly aided by the
(3) Zewail-Foote, M.; Hurley, L. H. J. Am. Chem. Soc. 2001, 123, 6485-
6495.
(4) Dedon, P. C.; Plastaras, J. P.; Rouzer, C. A.; Marnett, L. J. Proc. Natl.
Acad. Sci. U.S.A. 1998, 95, 11113-11116.
† Current address: Enzo Biochem, Inc., 60 Executive Blvd., Farmingdale,
NY 11735.
(1) Singer, B.; Grunberger, D. In Molecular Biology of Mutagens and
Carcinogens; Plenum: New York, 1983, ch. 4, pp 45-141.
(2) Tomasz, M. Chem. Biol. 1995, 2, 575-579.
(5) Plastaras, J. P.; Riggins, J. N.; Otteneder, M.; Marnett, L. J. Chem. Res.
Toxicol. 2000, 13, 1235-1242.
(6) Rink, S. M.; Solomon, M. S.; Taylor, M. J.; Rajur, S. B.; McLaughlin, L.
W.; Hopkins, P. B. J. Am. Chem. Soc. 1993, 115, 2551-2557.
(7) Millard, J. T.; White, M. M. Biochemistry 1993, 32, 2120-2124.
9
10.1021/ja036943o CCC: $25.00 © 2003 American Chemical Society
J. AM. CHEM. SOC. 2003, 125, 14005-14013
14005

A R T I C L E S
Veldhuyzen et al.
Scheme 1
Scheme 3
Scheme 2
ability of bizelesin to migrate from sites supporting only
monoalkylation to alternative sites supporting cross-linking.9
The highly reactive and transient electrophilic intermediate
quinone methide (QM) and related species have been the focus
of our laboratory’s interest in DNA alkylation and cross-linking
for the past decade. Early reports from others describing the
reversible10,11 and transient12 formation of QM adducts became
more instructive when our model studies indicated that the
observed specificity of QM for weak nucleophiles was based
on the thermodynamics rather than the kinetics of reaction
(Scheme 1).13 Selective preassociation between one of the
simplest QM (o-benzoquinone methide) and DNA is not easily
imagined, and therefore the preferential modification of GN2
by this QM in duplex DNA can be ascribed to the fundamental
thermodynamics of this adduct. Certainly its formation is not
kinetically favorable.13,14 A QM conjugate directed toward this
site should then act in synergy with the reaction thermodynamics
by increasing the effective concentration of the transient
intermediate and providing noncovalent binding stabilization to
the final product. The outcome of a complementary strategy of
directing QM to a kinetic rather than thermodynamic adduct is
not as easily predicted since either target binding or covalent
bond formation might dominate the ultimate product profile.
The potential conflict between adduct stability and site-
directed targeting became a concern while investigating sequence-
selective cross-linking of DNA by a conjugate constructed of a
triplex-forming oligonucleotide and a quinone methide precursor
designed for tandem alkylation (Scheme 2).15 A parent bis-
functionalized precursor (bisQMP, L ) CH3) derived from an
earlier monofunctionalized precursor (QMP) had been designed
to share a five-atom bridge between alkylation sites common
to N-mustards which cross-link GN7s in the major groove.6,16
and only one atom longer than the four-atom bridge of
mitomycin C that cross-links GN2’s in the minor groove.17 This
parent was competent at cross-linking DNA, but the extent of
reaction at GN7 was ambiguous and could have suffered from
competition between thermodynamic and kinetic products.18 An
oligonucleotide-bisQMP conjugate was then expected to
enhance reaction at GN7 due to its selective binding within the
major groove adjacent to a sequence 5′-CG that is favored for
bisQMP reaction in the major groove.18 Surprisingly, only
monoalkylation and no cross-linking was detected for the
conjugate.15 A bisQMP-acridine conjugate has now been
synthesized and characterized to address whether results with
the triplex-forming conjugate were unique to this initial trial or
more representative of a potential limitation in directing a
reversibly reactive group to form adducts lacking intrinsic
thermodynamic stability.
Results
Synthesis of bisQMP, Its Acridine Conjugate and Related
Derivatives. Attempts to elaborate our original silyl-protected
bis(bromomethyl)phenol18 for coupling to site-directing ligands
was stymied by the instability of the benzylic bromides.
Consequently, these substituents were replaced with acetates
which allowed for facile synthesis of the desired activated ester
and its coupling to primary amines. Hydroxyphenylpropionic
acid 1 was first hydroxymethylated with aqueous formaldehyde
under basic conditions to provide the bis(hydroxymethyl)
derivative 2 (Scheme 3).19 Selective silylation was not required
in this scheme since the benzylic acetoxy groups could be
introduced efficiently into the trisilylated species 3a through a
method developed by Ganem and Small.20 Accordingly, 3a was
treated with FeCl3 and Ac2O to produce the bis(acetoxymethyl)
compound 4a in 94% yield. The carboxylic acid was next
activated for coupling by preparation of its N-hydroxysuccin-
imidyl ester intermediate. This was condensed with 9-[N-(2-
aminoethyl)]aminoacridine21-23 5 to generate the desired acri-
dine conjugate 6a and with ammonia to generate a nonconjugated
(8) Sun, D.; Hurley, L. H. J. Am. Chem. Soc. 1993, 115, 5925-5933.
(9) Lee, S.-J.; Seaman, F. C.; Sun, D.; Xiong, H.; Kelly, R. C.; Hurley, L. H.
J. Am. Chem. Soc. 1997, 119, 3434-3442.
(10) Angle, S. R.; Yang, W. Tetrahedron Lett. 1992, 33, 6089-6092.
(11) Modica, E.; Zanaletti, R.; Freccero, M.; Mella, M. J. Org. Chem. 2001,
66, 41-52.
(12) Lewis, M. A.; Yoerg, D. G.; Bolton, J. L.; Thompson, J. A. Chem. Res.
Toxicol. 1996, 9, 1368-1374.
(13) Veldhuyzen, W. F.; Shallop, A. J.; Jones, R. A.; Rokita, S. E. J. Am. Chem.
Soc. 2001, 123, 11126-11132.
(14) Veldhuyzen, W.; Lam, Y.-f.; Rokita, S. E. Chem. Res. Toxicol. 2001, 14,
1345-1351.
(17) Tomasz, M.; Lipman, R.; Chowdary, D.; Pawlak, J.; Verdine, G. L.;
Nakanishi, K. Science 1987, 235, 1204-1208.
(15) Zhou, Q.; Pande, P.; Johnson, A. E.; Rokita, S. E. Bioorg. Med. Chem.
2001, 9, 2347-2354.
(18) Zeng, Q.; Rokita, S. E. J. Org. Chem. 1996, 61, 9080-9081.
(19) James, J. W.; Stevens, R. GB Patent # 1, 225, 290.
(16) Mattes, W. B.; Hartley, J. A.; Kohn, K. W. Nucleic Acids Res. 1986, 14,
2971-2987.
(20) Ganem, B.; Small, V. R. J. Org. Chem. 1974, 39, 3728-3730.
9
14006 J. AM. CHEM. SOC. VOL. 125, NO. 46, 2003

Stabilizing Labile Adducts of DNA
A R T I C L E S
Figure 1. Reaction of single- and double-stranded DNA with the bisQMP
conjugate 6a. 5′-[32P]-OD1 (30 nCi, 3.0 µM, lane 1) and 5′-[32P]-OD2 (30
nCi, 3.0 µM, lane 7) were alternatively incubated with 6a (60 µM) for 4 h
(20 °C) in 25% aqueous acetonitrile (6 mM MES, pH 7) and KF (10 mM)
(lane 2 and 8, respectively). Duplex DNA was formed by alternately
annealing 5′-[32P]-OD1 (30 nCi, 3.0 µM) with OD2 (3.3 µM, lane 3) and
5′-[32P]-OD2 (30 nCi, 3.0 µM) with OD1 (3.3 µM, lane 5) and incubated
with 6a under the conditions described above (lanes 4 and 6, respectively).
Figure 2. Cross-linking efficiency of bisQMP derivatives in the presence
and absence of fluoride. Duplex DNA formed by 5′-[32P]-OD1 (30 nCi,
3.0 µM) and OD2 (3.3 µM) (lane 1) was incubated for 12 h (20° C, 6 mM
MES, pH 7, 25% aqueous acetonitrile) in the alternate presence of 6a (60
µM, lanes 2 and 6), 7 (60 µM, lanes 3 and 7), 7 and 8 (60 µM each, lanes
4 and 8), or 6b (60 µM, lanes 5 and 9). Reactions were initiated by addition
of either KF (10 mM KF, lanes 2-5) or KCl (10 mM, lanes 6-9).
control derivative 7. The labile silyl group of 6a was also
replaced with a stable methyl group to distinguish between DNA
alkylation resulting from direct benzylic substitution and QM
formation. This additional derivative 6b was constructed in
parallel to the silyl species 6a starting from permethylation of
2,24 saponification and benzylic substitution.20
The bisQMP-Acridine Conjugate 6a Cross-Links Duplex
DNA Efficiently. Activities of both conjugated and nonconju-
gated derivatives of the bisQMP were examined with the same
oligonucleotide duplex used previously for investigating delivery
of the bisQMP by triplex formation (OD1/OD2, Figure 1).15
This permitted direct comparisons between the two bisQMP
conjugates and focused attention on competition between kinetic
and thermodynamic product formation while simultaneously
avoiding potential complications caused by use of different
nucleotide sequences. The original design incorporated a poly-
purine tract for triple strand recognition and oligonucleotides
of different lengths for distinguishing between strand alkylation
and cross-linking by gel electrophoresis.
Samples of single- and double-stranded DNA (3 µM) were
almost completely consumed within 4 h (20° C) in the presence
of excess bisQMP conjugate 6a and fluoride (Figure 1). The
mobilities of the parent oligonucleotides and their alkylation
products were unique and readily separated by denaturing gel
electrophoresis. In contrast, cross-linked DNA containing both
strands should express identical gel mobilities regardless of
whether OD1 or OD2 is labeled. This common mobility was
indeed observed after treating duplex DNA with conjugate 6a
(Figure 1, lanes 4 and 6). Interestingly, this conjugate cross-
linked the duplex with little accumulation of monoalkylation.
Acridine Conjugation and Quinone Methide Formation
Are Essential for Efficient Cross-Linking. The high efficiency
of cross-linking was not innate to the simple bisQMP component
(7) but rather depended on the attached acridine derivative to
direct and localize the reactive intermediate.
The bisQMP conjugate 6a generated cross-linked products
with a conservatively estimated yield of 64% as suggested by
quantifying the material migrating only within the indicated
region for cross-linked products relative to the total material
distributed throughout the analytical separation (Figure 2, lane
2). In contrast, only 1% of the duplex at most was cross-linked
under comparable conditions with the nonconjugated derivative
7 (Figure 2, lane 3). This low yield is consistent with the activity
of another nonconjugated derivative (L ) CH3, Scheme 2) that
required significantly higher concentrations (450 µM vs 60 µM)
to generate a maximum of 7.5% cross-linking.18 In addition,
the activity of 7 was not detectably affected by addition of an
untethered acridine derivative (8) (Figure 2, lane 4), and thus
intercalation by itself neither inhibited nor promoted reaction.
Similarly, intercalation and concomitant stabilization of duplex
DNA is not likely responsible for the low gel mobility of the
material ascribed to be cross-linked DNA since the products
formed by the bisQMP containing (6a) and lacking (7) acridine
migrated analogously.
Fluoride- and Ionic Strength-Dependence of DNA Cross-
Linking. Cross-linking exhibited a specific requirement for
fluoride that could not be satisfied with chloride (Figure 2, lanes
2-4 vs 6-8). This is consistent with an initial loss of the silyl-
protecting group from 6a and subsequent formation of an
intermediate QM prior to DNA reaction.18 If benzylic displace-
ment had alternatively been responsible for cross-linking, then
substitution of the silyl group with a methyl group (6b) should
have allowed for efficient cross-linking that was independent
(21) Prakash, A. S.; Denny, W. A.; Gourdie, T. A.; Valu, K. K.; Woodgate, P.
D.; Wakelin, L. P. G. Biochemistry 1990, 29, 9799-9807.
(22) Kohn, K. W.; Orr, A.; O’Conner, P. M. J. Med. Chem. 1994, 37, 67-72.
(23) Atwell, G. J.; Cain, B. F.; Baguley, B. C.; Finlay, G. J.; Denny, W. A. J.
Med. Chem. 1984, 27, 1481-1485.
(24) Johnstone, R. A. W.; Rose, M. E. Tetrahedron 1979, 35, 2169-2173.
9
J. AM. CHEM. SOC. VOL. 125, NO. 46, 2003 14007

A R T I C L E S
Veldhuyzen et al.
of fluoride. On the contrary, no alkylation or cross-linking of
DNA was detected after incubation of OD1/OD2 with 6b
(Figure 2, lanes 5 and 9). The low level of cross-linking
produced by the silyl derivative 6a in the absence of fluoride
(Figure 2, lane 6) is then best explained by the potential of 6a
to undergo a limited degree of spontaneous desilylation during
long periods of incubation. Reaction time of these experiments
(12 h) was extended well beyond that necessary for complete
reaction of 6a (4 h, Figures 1 and 3) to offer ample time for
possible alternative modes of reaction to generate observable
products.
The response of DNA cross-linking to ionic strength provided
complementary evidence of acridine’s dominant role in directing
reaction of its attached bisQMP in 6a. Intercalation of cationic
dyes is very sensitive to ionic strength, and high concentrations
of Na+ and K+ effectively compete for DNA binding.25-27
Increasing the concentration of KF used to initiate reaction of
6a under standard conditions from 10 to 100 mM decreased
the yield of cross-linking from 65 to 56%. A further increase
of KF concentration to 500 and 1000 mM suppressed the yield
even more to 26 and 13%, respectively. Such a decrease is
consistent with inhibition of intercalation since neither QM
formation or its alkylation of DNA is inhibited by high ionic
strength.18,28 Interestingly, DNA cross-linking by 6a (20 µM)
was not affected by addition of an 8-fold higher concentration
of the free acridine derivative 8 (160 µM). Reversible and
noncovalent interactions of 8 were consequently not competitive
with the reversible and covalent alkylation of 6a.
Figure 3. Time dependence of reaction between the bisQMP-acridine
conjugate 6a and double- and single-stranded DNA. Duplex DNA formed
by 5′-[32P]-OD1 (30 nCi, 3.0 µM) and OD2 (3.3 µM) was incubated (20
°C) with 6a (60 µM) in 25% aqueous acetonitrile 6 mM MES pH 7 for the
indicated time (lanes 1-7). Addition of KF (10 mM) was used to initiate
reaction. Identical conditions were used to examine the reaction of single-
stranded DNA (5′-[32P]-OD1, 30 nCi, 3.0 µM) in the absence of OD2 (lanes
8-14). The extent of cross-linking and covalent attachment of acridine was
measured as described in the Experimental section.
of both OD1 and its cross-linked products as a function of
reaction time were presumably the result of a limited degree of
further alkylation. Within 3 h, OD1 was consumed, and a
majority of cross-link had formed in the presence of a 20-fold
excess of 6a (Figure 3, lane 6). Maximum cross-linking was
achieved by 4 h (Figure 3, lane 7).
Kinetics and Stoichiometry of DNA Alkylation and Cross-
Linking. The rate and stoichiometry of DNA reaction were
examined to identify the origin of the low-mobility products
and detect if a threshold of alkylation was required prior to cross-
linking. These experiments also served as a necessary prereq-
uisite for determining the specificity of DNA modification
described below. Although cross-linking of duplex DNA was
expected to generate products of low mobility in denaturing
gel electrophoresis, changes in mobility due to covalent attach-
ment of acridine derivatives were less predictable. Alkylation
by the simple, nonconjugated bisQMP had not previously
affected DNA mobility,18 but products of reaction between
single-stranded DNA and conjugate 6a certainly did exhibit
retarded mobility (Figure 1).
The mobility change of single-stranded DNA caused by
reaction with 6a also neared completion at the end of the same
4 h incubation although consumption of single-stranded parent
OD1 was accomplished in less than 1.5 h (Figure 3, lanes 11
and 14). In contrast to the electrophoretic behavior of the
products formed by double-stranded DNA, those from single-
stranded DNA presented a continuum of progressively lower
mobility (Figure 3, lanes 9-14). This result is indicative of a
buildup of alkylation at multiple sites within OD1. Alternative
formation of intrastrand cross-links is possible, but again these
would likely have exhibited a distinct, rather than continuous,
change in gel mobility.
The stoichiometry of DNA modification was followed by the
gain in UV absorbance at 415 nm due to the attached acridine
versus UV absorbance at 260 nm due to both DNA and acridine.
The level of incorporating the bisQMP-acridine conjugate 6a
in duplex DNA consistently suggested over the time of
incubation that cross-linking occurred at approximately 25%
of the rate of monoalkylation (Figure 3). Each cross-linked
duplex on average contained no more than 4 acridine equivalents
(acridine bound/cross-link). Incorporation of the conjugate 6a
in single-stranded DNA was more rapid and accumulated to a
greater extent. Attachment of an average of two acridine
derivatives per OD1 inhibited gel migration noticeably, and a
maximum effect was observed after attachment of 8-10 acridine
derivatives per strand (Figure 3, lanes 10, 13, and 14). The lack
of conformational constraints of single-stranded DNA and its
greater accessibility to reagents are presumably responsible for
its enhanced reactivity with 6a.14,29
A small shift in mobility under denaturing conditions was
also evident by gel electrophoresis for the labeled strand (OD1)
in duplex DNA as the duration of its exposure to the bisQMP-
acridine conjugate 6a increased (Figure 3, lanes 1-7). Concur-
rently, a distinct low-mobility product was observed from the
earliest time of analysis, even before the mobility of OD1 had
changed significantly (0.5 h, Figure 3, lane 2). This low-mobility
product is hence not likely caused by the gradual accumulation
of multiple adducts but rather caused by an interstrand cross-
link as previously suggested by reaction of 6a and duplex DNA
containing alternatively labeled strands (Figure 1). Appearance
of a cross-link from the outset suggests that this process is quite
competitive with monoalkylation. The small changes in mobility
(25) Armstrong, R. W.; Kurucsev, T.; Strauss, U. P. J. Am. Chem. Soc. 1970,
92, 3174-3181.
(26) Zimmermann, H. W. Angew. Chem., Int. Ed. Engl. 1986, 25, 115-130.
(27) Medhi, C.; Mitchell, J. B. O.; Price, S. L.; Tabor, A. B. Biopolymers 2000,
52, 84-93.
(29) Pande, P.; Shearer, J.; Yang, J.; Greenberg, W. A.; Rokita, S. E. J. Am.
Chem. Soc. 1999, 121, 6773-6779.
(28) Li, T.; Rokita, S. E. J. Am. Chem. Soc. 1991, 113, 7771-7773.
9
14008 J. AM. CHEM. SOC. VOL. 125, NO. 46, 2003

Stabilizing Labile Adducts of DNA
A R T I C L E S
lability. For example, addition of a model QM (Scheme 1) to
AN1 is rapidly reversible.13 An equivalent CN3 adduct is more
stable but ultimately decomposes as well,12 and finally the GN7
adduct discussed above actually partitions between QM elimina-
tion and deglycosylation.14,32 Only the deglycosylation is easily
detected through its sensitivity to piperidine-induced strand
scission as illustrated in Figure 4A.
The profile of products generated from single-stranded DNA
and 6a were additionally examined with piperidine treatment,
and the results highlight important similarities and differences
between reaction of double- and single-stranded DNA. Gel
mobility of the single-stranded products again increased after
incubation with piperidine, but only a small fraction reverted
to a mobility equivalent to that of the parent oligonucleotide
(Figure 4B).31 Approximately 40% of the products present after
a 4 h incubation with 6a and OD2 still migrated more slowly
than unmodified OD2. Single-stranded, but not duplex, DNA
can then likely form a significant number of stable QM adducts
by reacting at GN1, GN2 and AN6.13,14 Another 45% of the
alkylated products were subject to scission by piperidine and
yielded fragments that also indicated a high selectivity for GN7.
However, the relative distribution of alkylation among the G
residues did not mimic the mild sequence selectivity detected
in the double-stranded reaction (Figures 4C vs 4D). Only G4
exhibited an apparent high reactivity in single-stranded DNA.
However, this observation is also consistent with an accumula-
tion of adducts typically associated with conditions that exceed
single hit kinetics. The relative selectivity of G4 and, to a lesser
extent, G6 increased over the incubation period as expected for
multiple alkylations within each target strand. Indeed, an
estimated 8-10 equiv of 6a alkylated single-stranded DNA after
a 4 h reaction. Accordingly, few conclusions pertaining to
specificity may be drawn from the smallest fragments of this
analysis.33 Note that comparable reaction of OD2 in duplex
DNA did not result in an excess of small DNA fragments
(Figure 4C) despite evidence for ∼2 adducts per strand (4 per
duplex). These seemingly contradictory results can be reconciled
if QM release is more facile than deglycosylation for adducts
of GN7.
Figure 4. Site specificity of the bisQMP-acridine conjugate 6a for double-
and single-stranded DNA. Identical incubations containing 6a and either
(A) double- or (B) single-stranded DNA as described in Figure 3 were
treated with hot piperidine, analyzed by denaturing gel electrophoresis,
compared to standard G ladders (lanes 1 and 9) and quantified by
phosphoimagery. The resulting strand scission at G24 ([), G22 (9), G20
(2), G6 (O), and G4 (×) in the (C) double- and (D) single-stranded samples
was then calculated relative to the total material of each sample. See Figure
1 for the nucleotide sequences of OD1 and OD2.
Site Specificity of Alkylation and Cross-Linking. Once the
efficiency of DNA alkylation and cross-linking by 6a was found
to be substantially controlled by its attached acridine, the site
specificity of reaction was also expected to be controlled by
this same component. The anticipated selectivity was confirmed
by treating product mixtures with hot piperidine, a common
method used to induce strand scission at GN7, AN3, and AN7
alkylation.30 All of the sites of alkylation and cross-linking
within duplex DNA were sensitive to this secondary treatment.
Of those products present after 4 h under conditions identical
to those of Figure 3, approximately 50% reverted to a mobility
equivalent to that of the parent strand in denaturing gel
electrophoresis (Figure 4A).31 Another 40% underwent strand
scission consistent with alkylation or cross-linking at GN7, and
only a trace level of scission at other residues was also observed.
The relative levels of each DNA strand fragment indicated that
a mild selectivity for the 5′‚‚‚GCGCG‚‚‚ sequence of OD2
remained constant during incubation with 6a (Figure 4C). Less
information can be gleaned from the fraction of oligonucleotide
products that reverted to parent-like mobilities, although these
products most probably derive from QM adducts of known
Discussion
A triplex binding oligonucleotide conjugate of bisQMP had
previously been constructed in hopes of enhancing the expected
efficiency of the major groove reaction of its nonconjugated
parent derivative. Although fewer equivalents of this conjugate
were necessary for reaction relative to that of the parent, the
conjugate was only capable of alkylating individual strands of
its target, and no interstrand cross-linking was detected.15 The
constraints of sequence specificity enforced by the attached
oligonucleotide were then eliminated in the acridine conjugate
6a to investigate the potential conflict between thermodynamic
and kinetic processes that may limit targeted alkylation of DNA.
An acridine conjugate was chosen as an alternative to the
oligonucleotide conjugate for a variety of reasons including the
ability of acridine to intercalate with little sequence selectiv-
ity,21,27,34,35 ease of preparation,22,23 ability to direct reagents to
the major groove,21,22 and its general applicability to coupling
(32) Frankenfield, K.; Veldhuyzen, W. F.; Weinert, E.; Rokita, S. E., in
preparation.
(33) Rokita, S. E. In Current Protocols in Nucleic Acid Chemistry; Glick, G.,
Ed.; Wiley: New York, 2001, p 6.6.1-6.6.16.
(30) Burrows, C. J.; Muller, J. G. Chem. ReV. 1998, 98, 1109-1151.
(31) Data on the complementary strand are included in Supporting Information.
9
J. AM. CHEM. SOC. VOL. 125, NO. 46, 2003 14009

A R T I C L E S
Veldhuyzen et al.
with a wide variety of DNA reagents.35-39 Most important to
our interests were the acridine conjugates of aliphatic and
aromatic N-mustards.21,22 The efficacy of DNA alkylation and
cytotoxicity of these conjugates increased orders of magnitude
over those of their nonconjugated derivatives. Such results were
somewhat anticipated since the attached acridine enhances
binding of the N-mustards to DNA and localizes them in the
major groove near their preferred site of reaction, GN7. The
extent to which acridine might promote reaction of the bisQMP
(Scheme 2) was not as obvious since the energy gained from
intercalation would need to offset the inherent instability of a
major groove reaction at GN7 and overcome competition from
minor groove reaction at GN2 that yields a thermodynamically
favored product.14 To complicate predictions even further,
conjugates appended to the amino group of 9-aminoacridine
have also been shown to reside in the minor groove, depending
on the linker and conjugate structure.21,40
The bisQMP-Acridine Conjugate 6a Cross-Links DNA
in the Presence of Fluoride To Initiate Quinone Methide
Formation. Interstrand cross-linking of duplex DNA was the
most readily observable product of reaction with 6a (Figure 1).
Conjugation of bisQMP to acridine enhanced its cross-linking
efficiency by at least 64-fold (Figure 2, lanes 2 vs 3) in analogy
to acridine’s effect on the N-mustards noted above.21,22 The
response of 6a to ionic strength also illustrated acridine’s
controlling influence on the bisQMP. Intercalation by acridine
is inhibited by high ionic strength25-27 and so too is the ability
of 6a to cross-link DNA. In contrast, the nonconjugated parent
bisQMP and a monofunctional QMP conjugate of single-
stranded DNA were not inhibited by high ionic strength or
excessive concentrations of KF (1 M).18,28
The bisQMP component again performed as designed for
generating the desired quinone methide intermediate. Reaction
depended on the presence of fluoride (10 mM) to remove the
silyl-protecting group (Figure 2, lanes 2 vs 6). No evidence was
detected to support an alternative mechanism based on direct
benzylic substitution even though this might have been promoted
if an optimal orientation of the nucleophile/electrophile pair had
occurred. The lack of such activity is best illustrated by the
inability of 6b, in which the silyl group is substituted by methyl,
to react with DNA (Figure 2, lane 5). Although the ratio of
cross-linking to simple alkylation was not measured or easily
determined for the parent bisQMP,18 the conjugate 6a yielded
approximately one cross-link out of every four initial alkylations
as indicated by the ratio of equivalents of bound acridine vs
equivalents of cross-linking (Figure 3). This ratio is quite
desirably small when compared to an estimate of 1:20 for a
N-mustard acridine conjugate.21 Certainly, the attached acridine
promotes initial alkylation for both types of conjugates. Its effect
on the subsequent interstrand reaction is less clear, and
consequently, differences in cross-linking vs alkylation ef-
ficiency may reflect intrinsic differences in the nature of the
Scheme 4
QM and N-mustard. Of course, the reversibility of QM
addition10-14 is one important characteristic that does not seem
to be shared by N-mustards, or at least no data have yet to
suggest this possibility.
The bisQMP-acridine conjugate, like other acridine conju-
gates, had the potential to sample all possible cross-linking sites
of its duplex target through noncovalent preassociation. In
contrast, the previous triplex-forming bisQMP derivative was
held to a single region of its target duplex, and this may have
restricted interstrand cross-links regardless of the reversibility
of QM alkylation.15 A N-mustard held through triplex formation
was previously shown to exhibit a similar ability to act only as
a mild alkylating agent and not as a cross-linking agent.41 The
possibility that the initial monoalkylated and cross-linked
derivative of DNA formed by 6a could continue to sample
alternative covalent structures is currently under investigation.
Rather than translocating via monoalkylation and release as
described previously for bizelesin,9 the QM may “walk” from
unfavorable to favorable cross-linking sites via transient forma-
tion of a monoalkylated intermediate (Scheme 4). This alterna-
tive is more reasonable for the bisQMP since its alkylation and
cross-linking are coupled in a manner not shared by bizelesin.
Whether or not QM migration occurs in this system, the
individual adducts of GN7 are still likely to remain labile as
expected from model studies.12-14 The attached acridine may
stabilize the overall association of the QM conjugate by
intercalation but cannot be expected to alter the dynamics of
the QM-GN7 bond.
Acridine Provides Rapid Delivery of bisQMP to GN7 in
the Major Groove of Duplex DNA. The acridine conjugate
6a cross-linked DNA from the earliest time of analysis and did
not first require a significant accumulation of monoalkylation
(Figure 3, lane 2). The ratio of total acridine attached per duplex
vs cross-linking varied little from 4.3 after 1 h to 3.2 upon
completion of reaction (4 h). If substantial monoalkylation
instead of cross-linking had occurred, then the reaction products
would have exhibited a gel mobility intermediate between that
of the parent DNA and its cross-linked derivative and similar
to the alkylation products of single-stranded DNA (Figures 1
and 3).
(34) Bailly, C.; Denny, W. A.; Mellor, L. E.; Wakelin, L. P. G.; Waring, M. J.
Biochemistry 1992, 31, 3514-3524.
(35) Bentin, T.; Nielsen, P. E. J. Am. Chem. Soc. 2003, 125, 6378-6379.
(36) Berthet, N.; Constant, J.-F.; Demeunynck, M.; Michon, P.; Lhomme, J. J.
Med. Chem. 1997, 40, 3346-3352.
(37) Ross, S. A.; Pitie´, M.; Meunier, B. Eur. J. Inorg. Chem. 1999, 557, 7-563.
(38) Delcros, J.-G.; Tomasi, S.; Carrington, S.; Martin, B.; Renault, J.;
Blagbrough, I. S.; Uriac, P. J. Med. Chem. 2002, 45, 5098-5111.
(39) Kuzuya, A.; Machida, K.; Mizoguchi, R.; Komiyama, M. Bioconj. Chem.
2002, 13, 365-369.
(40) Coppel, Y.; Constant, J.-F.; Coulombeau, C.; Demeunynck, M.; Garcia,
J.; Lhomme, J. Biochemistry 1997, 36, 4831-4843.
Treating the products of duplex reaction with piperidine
caused reversal of the cross-link and fragmentation of the
phosphodiester backbone in approximately equal proportions
(Figure 4A). Both results are consistent with alkylation of GN7
(41) Reed, M. W.; Lukhtanov, E. A.; Gorn, V.; Dutyavin, I.; Gall, A.; Wald,
A.; Meyer, R. B. Bioconj. Chem. 1998, 9, 64-71.
9
14010 J. AM. CHEM. SOC. VOL. 125, NO. 46, 2003

Stabilizing Labile Adducts of DNA
A R T I C L E S
Scheme 5
acridine intercalation, 6a can act more akin to its parent
nonconjugated derivative.
The bisQMP-acridine conjugate 6a demonstrates that ef-
ficient cross-linking based on quinone methide formation need
not rely on the presence of a single-stranded region extending
from duplex DNA as previously observed with a bisQMP
lacking the acridine.18 Likewise, cross-linking is not prohibited
from sites within the major groove as implied by data on a
bisQMP-oligonucleotide conjugate.15 The acridine component
of 6a appears to deliver and hold the bisQMP in the major
groove in a highly productive manner for cross-linking GN7, a
strong nucleophile that has the potential to add to QMs
reversibly. Intercalation of the acridine may (1) provide suf-
ficient binding stabilization to overcome the intrinsic instability
of QM alkylation at GN7 vs QM alkylation at GN2 and AN6
that forms irreversible adducts and/or (2) create a kinetic barrier
that prevents migration of the bisQMP to GN2 and AN6. In either
event, noncovalent interactions can overcome potential conflicts
between thermodynamics and kinetics of bond formation. For
the bisQMP-acridine conjugate, reversible QM formation seems
to be anchored in the major groove (Scheme 4). In a comple-
mentary manner, minor groove binding has previously been
shown to anchor reversible reaction of bizelesin.9 Related
mechanisms may also account for the apparent specificity of
other synthetic and natural products that react with DNA in a
reversible manner.
since its product can undergo elimination of the QM as well as
depurination and ultimately strand scission (Scheme 5).14,32
However, products formed by QM addition to CN3 and AN1
are also expected to eliminate the QM during piperidine
treatment. While alkylation at these sites has been observed with
N-mustards in special cases,42,43 equivalent reaction with at least
the simple o-benzoquinone methide is dramatically suppressed
in double- vs single-stranded DNA.14,29 Alkylation of AN7, a
weaker nucleophile than GN7 but accessible in the major
groove, is not evident after incubation with 6a since this too
would have induced strand scission upon piperidine treatment.
In contrast, this site has previously demonstrated competence
in nucleophilic reaction since it is a target of alkylation for
conjugates containing a long linker between acridine and an
aniline-based N-mustard.21
Since a single piperidine-labile bond in DNA cross-linking
(e.g., QM-GN7) is sufficient for the reversion and fragmentation
illustrated in Figure 4, little information is available on the
second site of attachment. However, cross-linking through two
GN7 positions is most satisfying. Participation of nucleophiles
in the major groove that can form stable QM adducts (AN,6
CN4)12-14 would only be possible if they acted in concert with
GN7. Cross-linking two AN6 or CN4 positions would be
expected to remain inert to piperidine treatment. Alternative
reaction between sites in the major and minor groove would
also be unlikely due to the steric constraints of duplex DNA.
Finally, cross-linking within the minor groove alone could
involve the nucleophiles AN3 and GN2 but no evidence suggests
reaction at either of these sites within duplex DNA. Alkylation
at AN3 would have been observed by strand fragmentation after
the hot piperidine treatment,44 and alkylation at GN2 would have
resisted reversal or strand fragmentation after this treatment.30,45
Unlike double-stranded DNA, single-stranded DNA was
observed to form products of both reversible and irreversible
alkylation. Reaction still occurred at GN7 as detected by strand
fragmentation after piperidine treatment (Figure 4B). Other
products of reaction retained a somewhat diminished gel
mobility (vs the parent strand) after equivalent treatment,
suggesting the presence of adducts at sites such as GN2 and
AN6. Thus, in the absence of the strong directing influence of
Experimental Section
General Materials. Organic reagents and starting materials were
purchased from Aldrich, Sigma, and Lancaster Chemical Co. and used
without further purification. Solvents, buffers, and salts were purchased
from Fisher and Sigma. HPLC grade acetonitrile was distilled over
calcium hydride under a nitrogen atmosphere prior to use. Silica gel
(230-400 mesh) for flash column chromatography was purchased from
EM sciences. Oligonucleotides were synthesized by Invitrogen Life
Technologies (Rockville, MD), purified by gel electrophoresis under
standard conditions,46 and labeled at their 5′ terminus with 5′-[γ-32P]-
ATP as directed by the enzyme supplier (New England Biolabs,
Beverly, MA). 4-Morpholineethanesulfonic acid (MES) was purchased
from Sigma and Calbiochem. All aqueous solutions were made with
water that had been purified by a standard purification system to yield
a resistivity of between 17.8 and 18 MΩ.
General Methods. Melting points were measured with a Thomas-
Hoover Unimelt apparatus and have been corrected. 1H and 13C nuclear
magnetic resonance (NMR) spectra were recorded on AM 400 and DRX
400 (1H: 400.13 MHz, 13C: 100.61 MHz) spectrometers, and coupling
constants are reported in Hz. Low-resolution mass spectra by electron
impact (EI) were determined with Hewlett-Packard 5980A and VG
7070E mass spectrometers. High-resolution mass spectra by EI and
fast atom bombardment (FAB) were also determined with the VG
7070E mass spectrometer.
3-[4′-Hydroxy-3′,5′-bis(hydroxymethyl)phenyl]propionic Acid (2).
Cold aqueous 5 M NaOH was added to 3-(4′-hydroxyphenyl)propionic
acid (2.0 g, 12 mmol) to adjust pH to 11, and the resulting solution
was combined with formaldehyde (37%, 6 mL).19 The reaction was
stirred at 55° C for 17 h, cooled (5° C), and combined with acetone
(100 mL). A resulting orange oil was collected, mixed with methanol
(15 mL) and again poured into acetone (150 mL) to form a white
precipitate. The solid was collected and washed with acetone to yield
(42) Boritzki, T. J.; Palmer, B. D.; Coddington, J. M.; Denny, W. A. Chem.
Res. Toxicol. 1994, 7, 41-46.
(43) Romero, R. M.; Rojsitthisak, P.; Haworth, I. S. Arch. Biochem. Biophys.
2001, 386, 143-153.
(46) Ellington, A.; Pollard, J. D. In Current Protocols in Molecular Biology;
Ausuble, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G.,
Smith, J. A., Struhl, K., Eds.; John Wiley and Sons: New York, 1998; p
section 2.12.
(44) Reynolds, V. L.; Molineux, I. J.; Kaplan, D. J.; Swenson, D. H.; Hurley,
L. H. Biochemistry 1985, 24, 6228-6237.
(45) Chatterjee, M.; Rokita, S. E. J. Am. Chem. Soc. 1990, 112, 6397-6399.
9
J. AM. CHEM. SOC. VOL. 125, NO. 46, 2003 14011

A R T I C L E S
Veldhuyzen et al.
2 as its sodium salt (2.2 g, 72%). To obtain a melting point, an aliquot
of the salt (∼15 mg) was dissolved in water, and the pH of the solution
was adjusted to pH 2 with 6 M HCl. The resulting suspension was
extracted with EtOAc (2 mL), and the organic phase was concentrated
under reduced pressure to yield a solid: mp 128-129° C (lit.19 130-
min, 2 (5.15 g, 22.8 mmol) and methyl iodide (25 mL, 18 mmol) were
added sequentially. This mixture was stirred at room temperature for
1 h, poured into water (100 mL), and extracted with ether (2 × 100
mL). The combined organic solutions were washed with water (2 ×
100 mL) and brine (2 × 100 mL), dried over MgSO4, and concentrated
under reduced pressure. The residue was then dissolved in methanol
(14 mL) and combined with an aqueous solution of LiOH (0.13 g, 3
mmol, 7 mL). This mixture was stirred at 4 °C for 15 h, acidified with
HCl to pH 2.0, and extracted with ether (3 × 30 mL). The combined
organic phases were washed with water (2 × 20 mL) and brine (2 ×
20 mL), dried over MgSO4, and concentrated under reduced pressure.
The residue was subjected to silica gel flash chromatography (hexane:
ethyl acetate, 7:3) to yield 3b (0.27 g, 15%) as light yellow-colored
viscous liquid. 1H NMR (CDCl3) δ 2.64 (t, J ) 7.9, 2H), 2.91 (t, J )
8.0, 2H), 3.40 (s, 6H), 3.76 (s, 3H), 4.48 (s, 4H), 7.20 (s, 2H), 11.2
(bs, 1H). 13C NMR (CDCl3) δ 29.8, 35.2, 57.8, 62.1, 69.1, 129.1, 130.8,
135.81, 154.7, 177.4. MS (FAB) m/z (rel intens) 269 (M + H+, 6.0),
237 (93), 191 (42), 147 (65), 91 (49), 55 (100). HRMS (FAB, glycerol)
m/z 269.1385. Calcd for C14H21O5 (M + H+) 269.1388.
3-(4′-Methoxy-3′,5′-bis(acetoxymethyl)phenyl)propionic Acid (4b).
Solid ferric chloride (0.030 g, 0.17 mmol) was added to a solution of
3b (0.20 g, 0.75 mmol) in acetic anhydride (10 mL) at 20 °C under
nitrogen. The mixture was maintained at 80 °C for 24 h, cooled, and
extracted with ether (3 × 10 mL). The combined ether extracts were
washed with water (2 × 30 mL) and saturated NaHCO3 (4 × 30 mL),
dried over MgSO4, and concentrated under reduced pressure. The
residue was subjected to silica gel flash chromatography (hexane:ethyl
acetate, 7:3) to yield 4b (0.22 g, 82%) as a light yellow-colored liquid.
1H NMR (CDCl3) δ 2.12 (s, 6H), 2.77 (m, 2H), 2.93 (m, 2H), 3.82 (s,
3H), 5.15 (s, 4H), 7.23 (s, 2H). 13C NMR (CDCl3) δ 20.7, 29.4, 36.5,
61.1, 62.7, 129.5, 130.5, 135.6, 155.7, 164.8, 170.6. MS (FAB) m/z
(rel intens) 325 (M + H+, 7.3), 265 (100), 193 (63), 147 (44), 91 (16).
HRMS (FAB, glycerol) m/z 325.1284. Calcd for C16H21O7 (M + H+)
325.1287.
1
133° C). H NMR (CD3OD) δ 2.39 (t, J ) 7.8, 2H), 2.80 (t, J ) 7.8,
2H), 4.66 (s, 4H), 7.00 (s, 2H). 13C NMR (CD3OD) δ 33.7, 42.4, 64.8,
119.3, 128.4, 129.7, 163.2, 171.1.
3-[4′-tert-Butyldimethylsilyloxy-3′,5′-bis(tert-butyldimethylsilyl-
oxymethyl)]propionic Acid (3a). Imidazole (3.30 g, 48.5 mmol) was
added to a solution of tert-butyldimethylsilyl chloride (TBDMSCl, 3.30
g, 21.9 mmol) and the sodium salt of 2 (1.00 g, 4.03 mmol) in 15 mL
of DMF. The reaction mixture was stirred at room temperature for 17
h, diluted with brine (100 mL), and extracted with ether (3 × 100 mL).
The organic phases were combined, dried over MgSO4, and concen-
trated under reduced pressure. The residue was redissolved in methanol
(50 mL) and potassium carbonate (2.00 g) was added in one portion.
The solution was stirred for 1 h and neutralized with 2 M HCl. The
mixture was then diluted with water (100 mL) and extracted with ether
(3 × 100 mL). The organic phases were combined, washed with brine
(3 × 100 mL), dried over MgSO4, and concentrated under reduced
pressure. The residue was purified by silica gel flash chromatography
(hexanes:ethyl acetate, 7:3) to yield 3a as a colorless solid: mp 83-
84 °C (1.51 g, 66%). 1H NMR (CDCl3) δ 0.08 (s, 12H), 0.13 (s, 6H),
0.91 (s, 18H), 0.99 (s, 9H), 2.64 (t, J ) 8.0, 2H), 2.91 (t, J ) 8.0, 2H),
4.66 (s, 4H), 7.17 (s, 2H). 13C NMR (CDCl3) δ -5.3, -3.4, 18.4, 18.7,
25.9, 26.0, 30.3, 35.8, 60.6, 125.6, 131.6, 133.2, 146.5, 179.6. MS (EI)
m/z (rel intens): 570 (M+ + 1, 1.7), 596 (M+, 5.0), 511 (43), 495
(6.4), 395 (14), 379 (100), 323 (19), 265 (27), 205 (34), 146 (67).
HRMS (FAB, glycerol) m/z 569.3488. Calcd for C29H57O5Si3 (M +
H+) 569.3513.
3-[4′-tert-Butyldimethylsilyloxy-3′,5′-bis(acetoxymethyl)phenyl]-
propionic Acid (4a). Solid ferric chloride (0.10 mg, 0.62 mmol) was
added to a solution of 3a (1.00 g, 1.76 mmol) in acetic anhydride (20
mL) at 0 °C under nitrogen.20 The reaction mixture was stirred for 30
min and then extracted with ether. The combined organic phases were
washed with water and saturated NaHCO3, dried with MgSO4, and
concentrated under reduced pressure. The residue was subjected to silica
gel flash chromatography (hexane:ethyl acetate, 7:3) and yielded 4a
as a colorless liquid (0.39 g, 94%): 1H NMR (CDCl3) δ 0.17 (s, 6H),
1.00 (s, 9H), 2.07 (s, 6H), 2.74 (m, 2H), 2.91 (m, 2H), 5.06 (s, 4H),
7.13 (s, 2H). 13C NMR (CDCl3) δ -3.8, 18.6, 20.9, 25.8, 29.3, 36.7,
61.6, 127.1, 129.8, 132.9, 149.9, 168.3, 170.7. MS (EI) m/z (rel intens)
424 (M+, 0.06), 421 (0.14), 250 (4.5), 215 (5.2), 186 (26), 142 (100).
HRMS (FAB, glycerol) m/z 425.2006. Calcd for C21H33O7Si (M + H+)
425.1996.
N-Succinimidyl-3-(4′-methoxy-3′,5′-bis(acetoxymethyl)phenyl)-
propionate. N-Hydroxysuccinimide (0.11 g, 0.90 mmol) was added
to a solution of 4b (0.20 g, 0.60 mmol) in DMF (2.0 mL). This mixture
was cooled to 4 °C and combined with EDC (0.13 g, 0.70 mmol). The
mixture was stirred for 20 h at 4 °C, diluted with brine, and extracted
with ether (50 mL). The organic phase was washed with brine (4 × 30
mL), dried over MgSO4, and concentrated under reduced pressure. The
residue was subjected to silica gel flash chromatography (hexane:ethyl
acetate, 7:3) to yield the desired activated ester of 4b (0.18 g, 72%) as
1
a viscous liquid of light yellow color. H NMR (CDCl3) δ 2.03 (s,
6H), 2.74 (s, 4H), 2.83 (t, J ) 7.6, 2H), 2.92 (t, J ) 7.6, 2H), 3.73 (s,
3H), 5.06 (s, 4H), 7.17 (s, 2H). 13C NMR (CDCl3) δ 20.7, 25.3, 29.4,
32.2, 60.9, 62.5, 129.4, 130.2, 135.0, 155.6, 167.5, 168.9, 170.4. MS
(FAB) m/z (rel intens) 422 (M + H+, 8.99), 362 (100), 290 (62), 175
(66), 147 (82), 91 (60). HRMS (FAB, glycerol) m/z 422.1447. Calcd
for C20H24O9N (M + H+) 422.1451.
N-Succinimidyl-3-(4′-tert-butyldimethylsiloxy-3′,5′-bis(acetoxy-
methyl)phenyl)propionate. N-Hydroxysuccinimide (0.035 g, 0.30
mmol) was added to a DMF solution (2.0 mL) of 4a (0.087 g, 0.20
mmol). This mixture was cooled to 4 °C and combined with 1-ethyl-
3-(3-dimethylaminopropyl)carbodiimide (EDC, 0.046 g, 0.24 mmol).
The mixture was then stirred for 20 h at 4 °C, diluted with brine, and
extracted with ether (30 mL). The organic phase was washed with brine
(4 × 20 mL), dried over MgSO4, and concentrated under reduced
pressure. The resulting residue was subjected to silica gel flash
chromatography (hexane:ethyl acetate, 7:3) to yield the desired activated
ester of 4a (0.071 g, 66%) as a viscous colorless liquid. 1H NMR
(CDCl3) δ 0.17 (s, 6H), 1.00 (s, 9H), 2.08 (s, 6H), 2.80 (s, 4H), 2.91
(m, 2H), 2.95 (m, 2H), 5.05 (s, 4H), 7.14 (s, 2H). 13C NMR (CDCl3)
δ -3.8, 18.5, 20.8, 25.5, 25.8, 29.5, 32.5, 61.6, 127.1, 129.8, 132.4,
149.9, 167.7, 168.9, 170.7. MS (FAB, glycerol) m/z (rel. Intensity) 522
(M + H+, 0.44), 462 (5.0), 404 (3.8), 362 (41), 288 (13), 249 (18),
219 (18), 207 (379), 149 (25), 131 (12), 117 (100). HRMS (FAB,
glycerol) m/z 522.2185. Calcd for C25H36O9NSi (M + H+) 522.2159.
3-(4′-Methoxy-3′,5′-bis(methoxymethyl)phenyl)propionic Acid (3b).
After mixing KOH (flakes, 20.6 g, 364 mmol) in DMSO (2 mL) for 5
N-(N′-Acridinyl-2′-aminoethyl)-3-(4′′-tert-butyldimethylsilyoxy-
3′′,5′′-bis(acetoxy-methyl)phenyl)propionamide (6a). Triethylamine
(0.16 g, 1.6 mmol) was added to a suspension of 522,23 (0.060 g, 0.20
mmol) in methanol (10 mL). Once this was homogeneous, the activated
ester of 4a (0.080 g, 0.20 mmol) in acetonitrile (10 mL) was added
dropwise over 1 min, and finally, acetic acid (0.050 g, 0.80 mmol)
was added over 5 min at room temperature. Solvent was removed under
reduced pressure, and the residue was dissolved in CH2Cl2 (50 mL).
This solution was washed with water (1 × 30 mL) and brine (2 × 30
mL), dried (MgSO4), and concentrated under reduced pressure. The
solid residue was recrystallized using methylene chloride and diethyl
ether to yield 6a (0.07 g, 65%) as a yellow solid: mp 187-188 °C. 1H
NMR (CDCl3) δ 0.17 (s, 6H), 1.01 (s, 9H), 2.08 (s, 6H), 2.78 (t, J )
7.4, 2H), 3.04 (t, J ) 7.4, 2H), 3.91 (m, 2H), 4.32 (m, 2H), 5.05 (s,
4H), 7.12 (bs, 1H), 7.20 (s, 2H), 7.46 (t, J ) 7.8, 2H), 8.11 (d, J )
8.5, 2H), 8.20 (d, J ) 8.5, 2H), 8.48 (t, J ) 7.8, 2H), 9.28 (bs, 1H).
9
14012 J. AM. CHEM. SOC. VOL. 125, NO. 46, 2003

Stabilizing Labile Adducts of DNA
A R T I C L E S
13C NMR (CDCl3) δ -3.7, 18.7, 21.0, 25.9, 30.9, 38.1, 39.4, 51.8,
61.9, 112.2, 119.4, 123.5, 124.7, 124.8, 127.1, 130.1, 133.8, 134.1,
139.6, 156.5, 170.9, 175.6. MS (FAB) m/z (rel intens) 644 (M + H+,
44), 600 (3.9) 524 (3.8), 470 (3.1), 264 (6.1), 221 (26), 209 (16), 195
(100), 180 (14), 117 (49). HRMS (FAB, glycerol) m/z 644.3187. Calcd
for C36H46O6N3Si (M + H+) 644.3156.
124.0, 126.4, 135.2, 140.4, 158.4, 172.5. HRMS (FAB) m/z 280.1456
(M + H+); Calcd for C17H17ON3 (M + H+): 280.1450
Oligonucleotide Studies. Duplex DNA was annealed by heating
(90° C) 1.0 equiv of the radiolabeled strand and 1.1 equiv of its
complementary unlabeled strand in 10 mM MES, pH 7, and then slowly
cooling to room temperature (3-4 h). The resulting samples and
alternatively the individual oligonucleotide components were then
diluted by 30% with addition of the bisQMP conjugate 6a or its related
derivatives dissolved in acetonitrile. Alkylation and cross-linking (20°
C) was initiated by diluting the mixtures by another 20% with addition
of KF to a final concentration of 10 mM. At the indicated times, reaction
was stopped by addition of â-mercaptoethanol (0.3 µL/µL of reaction),
and the samples were frozen, lyophilized, resuspended in 10 µL of
formamide loading solution (0.05% bromophenol blue and 0.05%
xylene cyanol FF in formamide), and finally analyzed by 20%
polyacrylamide denaturing gel electrophoresis. Alternatively, samples
were resuspended in 10% aqueous piperidine, heated (90° C, 30 min),
frozen, and lyophilized. The resulting residue was then dissolved in
water (100 µL) and again frozen and lyophilized to remove traces of
piperidine. These samples were then also mixed with the loading
solution and analyzed under equivalent conditions. Radiolabeled DNA
was detected using a Molecular Dynamics PhosphorImager and
quantified with ImageQuant software for determining reaction yields
(% product relative to total material).
N-(N′-Acridinyl-2′-aminoethyl)-3-(4′′-methoxy-3′′,5′′-bis(acetoxy-
methyl)phenyl)propionamide (6b). The synthesis of 6b essentially
followed the same procedure as that described above for coupling an
activated ester and used N-succinimidyl of 4b (0.03 g, 0.07 mmol)
with 522,23 (0.02 g, 0.07 mmol). Recrystallization using methylene
chloride and ether yielded 6b (4.0 mg, 10%) as a yellow solid: mp
1
137-138 °C. H NMR (CDCl3) δ 2.10 (s, 6H), 2.77 (t, J ) 7.3, 2H),
3.05 (t, J ) 7.3, 2H), 3.78 (s, 3H), 3.90 (m, 2H), 4.31 (m, 2H), 5.11
(s, 4H), 7.14 (bs, 1H), 7.26 (s, 2H), 7.46 (t, J ) 7.4, 2H), 8.10 (d, J )
8.5, 2H), 8.21 (d, J ) 8.5, 2H), 8.49 (t, J ) 7.5, 2H), 9.28 (bs, 1H).
13C NMR (CDCl3) δ 21.1, 31.0, 37.8, 39.4, 51.8, 61.3, 62.9, 112.1,
119.3, 123.5, 124.8, 129.5, 130.8, 134.1, 136.7, 139.5, 155.7, 156.6,
170.8, 175.5. MS (FAB) m/z (rel intens) 544 (M + H+, 100), 500 (5.2),
442 (3.9), 264 (3.1), 207 (42), 195 (87), 91 (7.0). HRMS (FAB,
glycerol) m/z 544.2428. Calcd for C31H34O6N3 (M + H+) 544.2447.
3-(4′-tert-Butyldimethylsiloxy-3′,5′-bis(acetoxymethyl)phenyl)pro-
pionamide (7). The activated ester of 4a (0.11 g, 0.19 mmol) was
dissolved in acetonitrile (12 mL) and water (12 mL), and pH was
adjusted to 12 by addition of NH4OH (29% aqueous solution). The
mixture was stirred at room temperature for 5 min, acidified with HCl
to pH 2.0, and extracted with ether (2 × 50 mL). The combined ether
extracts were washed with water (2 × 50 mL) and brine (2 × 50 mL),
dried over MgSO4, and concentrated under reduced pressure. The
residue was subjected to silica get flash chromatography (hexane:ethyl
acetate 1:1) to yield 7 (19 mg, 21%) as a colorless and viscous liquid.
1H NMR (400.13 MHz, CDCl3) δ 0.18 (s, 6H), 1.01 (s, 9H), 2.09 (s,
6H), 2.50 (t, J ) 8.2, 2H), 2.91(t, J ) 8.1, 2H), 5.07 (s, 4H), 5.46 (bs,
1H), 5.58 (bs, 1H), 7.14 (s, 2H). 13C NMR (100.61 MHz, CDCl3) δ
-3.7, 18.6, 21.0, 25.9, 30.5, 37.4, 61.8, 127.1, 129.9, 134.1, 149.8,
170.8, 174.3. MS (FAB) m/z (rel intens) 424 (M + H+, 14), 364 (100),
306 (22), 264 (50), 207 (43), 190 (42), 117 (97). HRMS (FAB, glycerol)
m/z 424.2173. Calcd for C21H34O6NSi (M + H+) 424.2155.
9-[N-(2-Acetylaminoethyl)]aminoacridine (8). Diethylamine (0.50
mL, 6.8 mmol) was added to a solution of 522,23 (25 mg, 0.091 mmol)
in acetonitrile (2.5 mL). The solution was stirred at room temperature
for 1 min, and acetic anhydride (20 µL, 0.19 mmol) was then added in
one portion. The resulting solution was stirred for 2 h at room
temperature, and the reaction was neutralized with 0.1 M HCl. The
mixture was extracted with chloroform (3 × 30 mL), and the organic
phase was dried over MgSO4 and concentrated under reduced pressure.
The residue was purified by silica gel chromatography (4:1 CHCl3:
MeOH). The resulting yellow solid was dissolved in water (5 mL),
and the pH was adjusted to 12 with 1 M NaOH. This solution was
then lyophilized, and the residue was once again dissolved in water (5
mL), neutralized with 0.1 M HCl, and extracted with CHCl3 (3 × 5
mL). The combined organic phases were dried over MgSO4 and
concentrated under reduced pressure to afford 8 as a yellow solid (11
mg, 41%): 1H NMR (400 MHz, DMF-d7) δ 1.97 (s, 3H), 3.78 (q, J )
5.6, 2H), 4.37 (q, J ) 5.6, 2H), 7.55 (t, J ) 8.0, 2H), 7.97 (t, J ) 8.0,
2H), 8.30 (d, J ) 8.0, 2H), 8.83 (m, 2H), 9.02 (bs, 1H), 10.53 (bs,
1H); 13C NMR (101 MHz, DMF-d7) δ 22.6, 39.9, 50.8, 113.2, 119.5,
Stoichiometry of Acridine Coupling to Double- and Single-
Stranded DNA. Reaction conditions used for gel analysis were
maintained for these studies and only the total volume was increased
to 800 µL. At the indicated times, â-mercaptoethanol (100 µL) was
added to stop reaction, and samples were dialyzed (10000 MW cutoff)
against water for 48 h, lyophilized, dissolved in water (1000 µL), and
analyzed by UV to determine the acridine:DNA ratio. Absorbance (415
nm) of the t ) 0 h sample was used as a blank for subsequent A415
determinations of acridine. Since the extinction coefficients of acridine
at 260 and 415 nm are affected by single- and double-stranded DNA,
their corrected values (ꢀ260 ) 29.0 × 10-3 µM-1 cm-1 and 25.5 ×
10-3 µM-1 cm-1, respectively and ꢀ415 ) 5.53 × 10-3 µM-1 cm-1 and
4.59 × 10-3 µM-1 cm-1, respectively) were determined by titrating
acridine with OD2 and calf thymus DNA. The concentration of DNA
recovered from reaction with 6a was determined from the extinction
coefficients provided by the manufacturer (OD1: ꢀ260 ) 2.61 µM-1
cm-1; OD2: ꢀ260 ) 3.64 µM-1 cm-1) and the ∆A260 remaining after
subtracting the contribution of bound acridine.
Acknowledgment. This research was supported in part by
the National Institutes of Health (CA-81571). We thank Profes-
sor J. Davis for helpful discussion and Dr. Q. Zeng for
preliminary studies and synthesis of 9-(N-(2-aminoethyl))-
aminoacridine.
Supporting Information Available: Polyacrylamide gel analy-
sis of alkylation and cross-linking of OD1 after piperidine
treatment and details on acridine equivalents bound to DNA
(PDF). This material is available free of charge via the Internet
at http://pubs.acs.org.
JA036943O
9
J. AM. CHEM. SOC. VOL. 125, NO. 46, 2003 14013