New anthracenedione derivatives with improved biological activity by virtue of stable drug-DNA adduct formation

J. Med. Chem. 2010, 53, 6851–6866 6851

DOI: 10.1021/jm901894c

New Anthracenedione Derivatives with Improved Biological Activity by Virtue of Stable Drug-DNA

Adduct Formation

Oula C. Mansour,^†Benny J. Evison,^†Brad E. Sleebs,^‡Keith G. Watson,^‡Abraham Nudelman,^§Ada Rephaeli,

Damian P. Buck,^{^}J. Grant Collins,^{^}Rebecca A. Bilardi,^†Don R. Phillips,^†and Suzanne M. Cutts*^,†

^†Department of Biochemistry, La Trobe University, Victoria 3086, Australia, ^‡The Walter and Eliza Hall Institute of Medical Research, Parkville,

Victoria, 3052, Australia, ^§Chemistry Department, Bar-Ilan University, Ramat-Gan, 52900, Israel, Sackler Faculty of Medicine, Felsenstein

Medical Research Center, Tel-Aviv University, Petach-Tikva 49100, Israel, and ^{^}School of Physical, Environmental and Mathematical Sciences,

University of New South Wales, Australian Defence Force Academy, ACT 2600, Australia

Received December 22, 2009

MitoxantroneisananticanceragentthatactsasatopoisomeraseIIpoison, however, itcanalsobeactivated

by formaldehyde to form DNA adducts. Pixantrone, a 2-aza-anthracenedione with terminal primary

amino groups in its side chains, forms formaldehyde-mediated adducts with DNA more efficiently than

mitoxantrone. Molecular modeling studies indicated that extension of the “linker” region of anthracene-

dione side arms would allow the terminal primary amino greater flexibility and thus access to the guanine

residues on the opposite DNA strand. New derivatives based on the pixantrone and mitoxantrone

backbones were synthesized, and these incorporated primary amino groups as well as extended side chains.

The stability of DNA adducts increased with increasing side chain length of the derivatives. A mitoxantrone

derivative bearing extended side chains (7) formed the most stable adducts with ∼100-fold enhanced

stability comparedto mitoxantrone. This finding is of great interest because long-lived drug-DNA adducts

are expected to perturb DNA-dependent functions at all stages of the cell cycle.

Introduction

Numerous classes of anticancer drugs are represented

by compounds that form stable chemical bonds with DNA.

A large number of such compounds have been synthesized

and tested for anticancer activity, and many of them are in

clinical use.¹The cytotoxicity of these drugs is mainly caused

by an apoptotic response, which is triggered by the formation

of DNA adducts.²

Mitoxantrone (1, Figure 1) is a synthetic anticancer agent

that belongs to the anthracenedione class of compounds.^3,4

The initial motive behind its synthesis was to find an anthra-

cycline analogue with the same potent cytotoxicity but with

less cardiotoxicity.^3,51 lacked the severe cardiotoxic side

Figure 1. Chemical structures of AN-9 3 and the anthracenediones

studied. Mitoxantrone 1; Pixantrone 2; 1-derived analogues 4-7;

2-derived analogues 8-10.

effects of the anthracyclines and has subsequently been used

against a wide range of solid tumors and myeloid cancers.⁶As

an anthracycline analogue, 1 maintains the planar ring struc-

ture that allows intercalation between base pairs of DNA.^7,81

is classified as a typical intercalating agent that accumulates

in the nucleus where it acts as a topoisomerase II poison,⁶

thereby inhibiting the topological modification of DNA. The

drug blocks the strand scission step of the topoisomerase II

catalytic cycle, stabilizes the “cleavable complex”, and leads

to the generation of DNA double-strand breaks (DSB^a).⁹

Although 1 exhibits lower toxicity compared to doxorubicin,

it still causes a range of side effects, including myelosuppres-

sion and mild cardiotoxicity.^6,10

1, like doxorubicin, can be activated by formaldehyde to

generate covalent 1-DNA adducts.^11-13These adducts were

originally detected by interstrand cross-linking assays as they

stabilize duplex DNA.¹²Both formaldehyde and 1 were

required to form the adducts, confirming a key role for

formaldehyde in activating 1.¹²Production of formaldehyde

is relatively more active in cells of myeloid origin¹⁴as well as in

solid tumors where it is generated from biologically available

polyamines.^15,16These findings may explain why 1 is more

effective against myeloid and solid tumors.

1-DNA adducts have a half-life of approximately 50 min,

indicating that the adducts are intrinsically unstable. An

energy-minimized model of 1-DNA adducts demonstrated

that these adducts were likely to be mediated by a linkage

between the secondary amino portion of one side arm of 1 and

the exocyclic N-2 of guanine,¹⁷with the drug chromophore

*To whom correspondence should be addressed. Phone: þ61-03-

9479-1517. Fax: þ61-03-9479-2467. E-mail: s.cutts@latrobe.edu.au.

^aAbbreviations: DSB, double-strand breaks; MTT, 3-(4,5-dimethy-

lthiazol-2-yl)-2,5-diphenyltetrazolium bromide; TE, Tris-EDTA buffer;

TC, transcription buffer; CI, combination index.

r

2010 American Chemical Society

Published on Web 09/22/2010

pubs.acs.org/jmc

6852 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19

Mansour et al.

intercalated between C and G residues of a CpG dinucleo-

tide. The substitution of the secondary amino by a primary

amino function has been suggested^18,19based on studies of

the DNA binding mode of other formaldehyde activated

compounds.^13,20,21

Time Dependence of Adduct Formation Displayed by Formal-

dehyde-Activated Anthracenediones. The time dependence of

adduct formation by the derivatives was subsequently investi-

gated. Anthracenediones (7.5 μM and 25 μM for all 1 and 2

derivatives, respectively) were reacted with end-labeled pCC1

DNA and 2 mM formaldehyde. All derivatives formed DNA

adducts in a time-dependent manner (Figure 3A,B). 7 achieved

a maximal level of dsDNA stabilization, reaching 100% within

10 h while 1 exhibited significantly lower reactivity, failing to

stabilize any detectable dsDNA even at 24 h. Formaldehyde-

activated 5 and 6 stabilized an intermediate level of dsDNA at

20 and 40% dsDNA, respectively (Figure 3C). In contrast to the

1 group of derivatives where 7 was the more reactive derivative,

2 was the most active of its group, with 100% dsDNA stabiliza-

tion formed in the first 6-7 h (Figure 3D). All 2 derivatives did

not reach more than 50% dsDNA stabilization in the 24 h

period of incubation (Figure 3D).

Formaldehyde Concentration-Dependence of Adduct For-

mation Displayed by Formaldehyde-Activated Anthracene-

diones. To confirm that DNA adducts formed by 1, 2, and

their derivatives were dependent on formaldehyde, a for-

maldehyde concentration-dependence study was performed.

All derivatives (15 μM) were reacted overnight with DNA

and a range of increasing concentrations of formaldehyde

(0-3 mM). 1, 2, and their derivatives did not form adducts

in the absence of formaldehyde, confirming the absolute

requirement for formaldehyde in mediating a covalent aminal

linkage to DNA. The formation of DNA adducts induced by

1, 2, and the derivatives were clearly dependent on formal-

dehyde concentration with increasing levels of adducts in the

presence of increasing levels of formaldehyde (Figure 4A,B).

7 was again the most effective derivative in forming stable

DNA adducts, followed by 6 and then 5 (Figure 4C). In con-

trast, 2 derivatives were all less effective than 2, with 10 being

more reactive than the other derivatives (Figure 4D).

Pixantrone 2 (Figure 1), a novel 2-aza-anthracenedione

with primary amino groups in its side chains, has recently

demonstrated significant anticancer efficacy. The drug fails to

induce any detectable cardiotoxicity and is now in phase III

clinical trials for the treatment of aggressive non-Hodgkin’s

lymphoma.^22,23It was recently demonstrated that formalde-

hyde-activated 2, due to the primary amino groups incorpo-

rated into its side arms, formed stable DNA adducts much

more effectively than 1, forming a covalent linkage at CpG

sequences in DNA with at least one accessible guanine at

the N-2 position.^24,25On the basis of this finding and on the

energy-minimized model described above, anthracenedione

derivatives were designed in an attempt to develop com-

pounds with better DNA reactivity, cross-linking potential,

and stability. The side arm moieties of 1 were replaced by

methylene tethers, of varying lengths, with terminal primary

amine functionalities to yield compounds 4-7 (Figure 1).

Similarly, the side chains of 2 were extended to yield the three

derivatives 8-10 (Figure 1).

Each derivative was initially tested for its capacity to bind

covalently with DNA in the presence of formaldehyde. The

stability of each drug-DNA lesion was subsequently explored

and their DNA sequence specificity determined. The cell

growth inhibition effects of 1 derivatives in combination with

the formaldehyde-releasing prodrug AN-9 3²⁶(Figure 1) were

also evaluated as a preliminary study.

Results

Drug Concentration Dependence of Adduct Formation

Displayed by Formaldehyde-Activated Anthracenediones.

Each anthracenedione was initially tested in a cross-linking

assay for the capacity to form DNA adducts in the presence of

formaldehyde. When anthracyclines and anthracenediones

are covalently bound to DNA, duplex DNA is stabilized

and exhibits an altered thermal melting profile where the mid-

point melting temperature is increased.²⁷The optimal dena-

turation temperature for the detection of anthracenedione

derivative-DNA adducts was determined as 52 ꢀC (data not

shown). DNA was initially reacted with a range of increasing

concentrations of each drug in the presence of formaldehyde.

Samples were subjected to a phenol/chloroform cleanup pro-

cedure to remove unbound and intercalated drug and dena-

tured in formamide before separation by electrophoresis.

The phosphorimage in Figure 2A shows that all 1-based

compounds formed high levels of stabilized double-stranded

DNA (dsDNA) at low concentrations with the exception of

1 itself, which required around 50 μM to achieve the same

level of stabilization (Figure 2C) as previously documented.²⁴

7 was slightly more potent than the other derivatives, with

90% dsDNA stabilization achieved at 7.5 μM, while at the

same concentration, 6 and 5 formed 70% and 45% dsDNA

stabilization, respectively. In contrast, 2-based derivatives

were all less effective than 2 and did not form high levels

of DNA adducts (Figure 2B). At 7.5 μM, 2 yielded 100%

dsDNA stabilization, while at the same concentration, 8-10

caused less than 20% dsDNA stabilization. 10 was slightly

more effective in stabilizing dsDNA than the other deriva-

tives (Figure 2D).

Stability of Adducts Induced by Formaldehyde-Activated

Anthracenediones. One of the most desirable attributes of the

new derivatives was the ability to form stable drug-DNA

adducts at a physiologically relevant temperature. To ascer-

tain the stability of the derivatives, drug-DNA adducts were

generated by reacting those derivatives at a concentration

that would initially yield 60-90% dsDNA stabilization.

Interestingly, the stability of adducts increased dramatically

with increasing side arm length, with 1 derivatives displaying

slightly better stability than 2 derivatives (Figure 5A,B). The

half-life of 7 was more than 2.5 days, while that of 1 was less

than 60 min (Figure 5C and summarized in Table 1). 6 and 5

exhibited half-lives of 30 and 14 h, respectively (Figure 5C;

Table 1). The same trend was seen with 2 derivatives, which

exhibited the following half-lives: approximately 35 h for 10,

16 h for 9, 13 h for 8, and 2 h for 2 (Figure 5D; Table 1). As

7 displayed exceptionally high DNA adduct stability, an

extended stability study was performed over a period of 12 days.

Samples were incubated for defined time periods up to 12 days

at 37 ꢀC. There was a slow loss of adducts with an extended

half-life (Figure 5E).

pH Dependence of Adduct Formation by Formaldehyde-

Activated Anthracenediones. The pH dependence of forma-

tion of adducts by 1 and 2 derivatives was also established.

Different concentrations of derivatives that were expected to

yield between 40 and 60% dsDNA stabilization at neutral

pH were used (Figure 6A,B). 1-DNA adduct formation was

pH dependent, with neutral pH required for optimal forma-

tion as previously demonstrated.²⁴In contrast, 1 derivatives

Article

Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19 6853

Figure 2. Drug concentration dependence of adduct formation displayed by 1 and 2 derivatives. (A,B) Anthracenediones (0-50 μM) were

reacted with pCC1 DNA (25 μM_bp) in the presence of 2 mM formaldehyde in 1ꢀ PBS for 12 h at 37 ꢀC. Control ssDNA represents unreacted

DNA that was thermally denatured while the dsDNA control was not subjected to thermal denaturation. Images shown are representative of

those obtained from each of two independent experiments. (C) Quantitation of gel (A). The percentage of dsDNA stabilized by formaldehyde-

activated 1 (O), 5 ( ꢀ ), 6 (0), and 7 (9) was quantitated and is expressed as a function of drug concentration. (D) Quantitation of gel (B). The

percentage of dsDNA stabilized by formaldehyde-activated 2 (O), 8 ( ꢀ ), 9 (0), and 10 (9) was quantitated and is expressed as a function of

drug concentration. Results shown are the average of two independent experiments, with error bars representing the range.

were not pH dependent, although slightly lower levels of

adducts typically formed under alkaline conditions and there

appeared to be a slight increase of the optimal pH with

decreasing side chain length (Figure 6C). Adduct formation

with 2 and related derivatives were also essentially indepen-

dent of pH (Figure 6D).

DNA Adduct Formation Studies with Derivative 4. 7 was

clearly more effective than 1 in rapidly forming stable,

non-pH dependent DNA adducts at low concentrations. In

contrast, 2, with a shorter side chain length, proved to be

more effective than 2 derivatives in forming DNA adducts. 4,

a new derivative based on the 1 backbone but with two

6854 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19

Mansour et al.

Figure 3. Time dependence of adduct formation displayed by 1 and 2 derivatives. A 7.5 μM concentration (A) and 25 μM concentration

(B) were used with all compounds which were reacted with pCC1 DNA (25 μM_bp) and 2 mM formaldehyde in PBS at 37 ꢀC for increasing time

periods up to 24 h as indicated. (C) Quantitation of gel (A). The percentage of dsDNA stabilized for 1 (O), 5 ( ꢀ ), 6 (0), and 7 (9) reacted DNA

samples was quantitated and expressed as a function of time. (D) Quantitation of gel (B). The percentage of dsDNA stabilized for 2 (O), 8 ( ꢀ ), 9

(0), and 10 (9) reacted DNA samples was quantitated and expressed as a function of time.

methylene groups in each side chain, was therefore synthe-

sized to enable direct comparison to 2 and determine if

shorter chain length corresponds to a greater efficiency of

adduct formation for each backbone studied.

4 was compared to 7 and 2 in drug concentration depen-

dence experiments. All compounds behaved similarly, forming

DNA adducts at relatively low concentrations (Figure 7A).

At a concentration of 2.5 μM, the three compounds attained

approximately 40% dsDNA stabilization. The formaldehyde

concentration dependence was then examined for 4, and 7 was

used for comparison (both compounds at 15 μM). Both

compounds showed a similar dependence on formaldehyde

concentration, with no adducts formed in its absence. Less than

1 mM formaldehyde was needed for both drugs to achieve

50% dsDNA stabilization, and 100% was achieved with 2 mM

formaldehyde (Figure 7B). Stability studies indicated that 4

Article

Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19 6855

Figure 4. Formaldehyde concentration dependence of adduct formation displayed by 1 and 2 derivatives. (A,B) Drugs (at 15 μM

concentration) were reacted with pCC1 DNA (25 μM_bp) and increasing concentrations of formaldehyde ranging from 0 to 3 mM in PBS

for 16 h at 37 ꢀC. Control ssDNA is the unreacted DNA thermally denatured, while the dsDNA control is the unreacted native dsDNA. A third

control (ct) is the no-drug reacted DNA incubated with 2 mM formaldehyde. (C) Quantitation of gel (A). The percentage of dsDNA stabilized

for 1 (O), 5 ( ꢀ ), 6 (0), and 7 (9) reacted DNA samples was quantitated and expressed as a function of formaldehyde concentration. (D) The

percentage of dsDNA stabilized for 2 (O), 8 ( ꢀ ), 9 (0), and 10 (9) reacted DNA samples was quantitated and expressed as a function of

formaldehyde concentration.

adducts were rapidly lost (half-life approximately 2.3 h) relative

to 7adducts (half-life more than 2.5 days) (Figure 7C). 4adduct

formation was time dependent, with increasing adducts formed

with increasing time of incubation, with 50% dsDNA stabiliza-

tion being attained within 5 h (Figure 7D) similar to the result

shown for 7 (Figure 3C). Therefore, for the 1 backbone, chain

length does not correlate with adduct formation efficiency as

observed with the 2 backbone.

In Vitro Transcription Studies Demonstrate the Sequence

Specificity of Formaldehyde-Activated Anthracenediones. An

in vitro transcription assay was used to obtain the preferred

sequence of binding sites for each derivative. The assay is

based on blocking the progression of RNA polymerase along

dsDNA. Escherichia coli RNA polymerase moves along

dsDNA and is blocked each time it encounters a drug

occupied site, thereby building up a number of truncated

6856 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19

Mansour et al.

Figure 5. Stability of adducts induced by 1 and 2 derivatives. (A,B) 1 (36 μM), 5 (12 μM), 6 (8 μM), 7 (4 μM), 2 (5 μM), and 8-10 (40 μM each)

were reacted with DNA in the presence of 2 mM formaldehyde at 37 ꢀC for 16 h then subjected to a phenol/chloroform cleanup and left, as

indicated, for time periods up to 24 h at 37 ꢀC, then exposed to a second phenol/chloroform extraction and characterized after overnight

electrophoresis. Control ssDNA is the unreacted DNA thermally denatured, while the dsDNA control was native dsDNA. (C) Quantitation of

gel (A). The percentage of dsDNA stabilized for 1 (O), 5 ( ꢀ ), 6 (0), and 7 (9) reacted DNA samples was quantitated and expressed as a function

of time. (D) Quantitation of gel (B). The percentage of dsDNA stabilized for 2 (O), 8 ( ꢀ ), 9 (0), and 10 (9) reacted DNA samples was

quantitated and expressed as a function of time. (E) 7 (4 μM) was reacted with DNA in the presence of 2 mM formaldehyde at 37 ꢀC for 16 h and

then subjected to a phenol/chloroform cleanup and left, as indicated, for time periods up to 12 days at 37 ꢀC. The percentage of dsDNA

stabilized for 7-reacted DNA samples was quantitated and expressed as a function of time.

Article

Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19 6857

Table 1. DNA Adduct Stability (In Vitro Crosslinking Assay)

was evaluated in MCF-7 breast adenocarcinoma, PC-3

prostate, and HL60 acute promyelocytic leukemia cells using

growth inhibition assays. As single agents, drug-induced

growth inhibition generally decreased with increasing side-

chain length (Table 2). The growth inhibition assays were

also performed in the presence of 3 (as a formaldehyde-

releasing prodrug) and CI values calculated to determine the

types of drug interactions. Generally, there was a tendency

for CI values to decrease as side-chain length increased for

the compound series 4-7. This trend was most evident in the

HL60 cells for the combination with 3, where 4 demonstrated

antagonism, 5 demonstrated slight antagonism, 6 showed

slight synergism, and 7 displayed moderate synergism. In

PC-3 cells, the same trend was observed in combination

experiments where 6 displayed synergism, however, only mod-

erate synergism was observed for 7. The trend was least

obvious in MCF-7 cells, where increasing side chain length

only led to a slight decrease in CI values. It should be noted that

because IC₅₀values generally increased with increasing side-

chain length, the molar excesses of 3 resulted in comparatively

higher 3 concentrations used. However, CI calculations take

into account the intrinsic contribution of each drug to growth

inhibitory effects.

compd

chain length

adduct half-life (h)^a

1

NA

2

0.5^b

4

2.3 ( 0.1

14 ( 0.4

30 ( 0.9

53 ( 7

5

3

6

4

7

5

2

2.3 ( 0.1

13 ( 0.1

16 ( 0.2

35 ( 0.1

8

3

9

10

4

5

^aThe adduct half-life is shown for each compound. Error values

indicate the error of the exponential decay fit to the data. ^bThis figure

was taken from ref 24.

transcripts. The location of the drug binding sites can be

obtained from the length of these truncated RNAs. When

there are no DNA adducts, the RNA polymerase can move

along the entire DNA template, generating 379 nucleotide

full length transcripts.

Recently, it was established that formaldehyde-activated 2

generates DNA lesions which block the progression of RNA

polymerase during transcription and cause drug-induced

blockages at CpG and CpA dinucleotide sequences.²⁵DNA

adducts induced by formaldehyde-activated 1 and 2 derivatives

were assessed using the in vitro transcription assay. A 512 bp

fragment containing the lac UV5 promoter was reacted with

each derivative in the presence of formaldehyde (Figure 8A).

Each anthracenedione blocked RNA polymerase at similar

sequences, indicating that all of these compounds have simi-

lar sequence specificity for adduct formation. The similarity of

these discrete transcriptional blockages to those observed for 1

and 2 indicate that the sequence preference of all derivatives is

CpG and CpA specific. The most intense blockages occurred

at either CpG or CpA dinucleotides (Figure 8B,C,D; for

sequence specificity histograms for all compounds see Support-

ing Information). Transcriptional blockages were also shown

to increase with increasing concentrations of the drugs and with

increasing reaction time (data not shown).

Molecular Modeling of DNA Adducts Induced by Formal-

dehyde-Activated Anthracenediones. Figure 9 shows 7 inter-

calated between the C₃and G₄bases of the hexanucleotide

d(ACCGGT)₂from the minor groove and the drug covalently

bound through an aminal linkage to the N-2 atom of the G₅

residue on each strand. Although it was possible to form bis

adducts with the N-2 atoms from both G₄residues, the G₅bis

adducts gave lower energies. The terminal amine of 7 in the

noncovalent intercalation complex is ideally positioned to

form a covalent adduct with the G₅residues on each strand.

The modeling suggests that the formation of a G₅cross-linked

bis adduct causes very little disruption to the hexanucleotide

structure. Alternatively, a cross-linked bis adduct with the G₄

residues introduces perturbations to the helix.

Discussion

It has previously been shown that formaldehyde is involved

in the activation of 1, which can subsequently bind covalently

to DNA, forming adducts at CpG sequences of duplex

DNA.^11,12These adducts are formed by a linkage between

the secondary amino portion of the side arm of 1 and the N-2

of guanine.¹⁷The potential involvement of a primary amino

group instead of a secondary amino group to create a covalent

N-C-N aminal linkage to the N-2 of guanine was investi-

gated using 2.^18,192, which bears a primary amino group in

each of its side chains, was able to form DNA adducts much

more effectively than 1, forming a covalent linkage at CpG

sequences with DNA caused by reaction with at least one

accessible guanine at the N-2 position.²⁴2-DNA adducts are

more stable than 1-DNA adducts, yet both half-lives are less

than 2 h, indicating that the adducts are unstable.²⁴On the

basis of an energy-minimized structure of both the intercalated

and covalently bound 1,¹⁷it was suggested that the extension

of the side arms of 2 may permit covalent linkage to both DNA

strands, thereby creating a more reactive and possibly more

stable adduct. The same underlying principle was also applied

to derivatives based on a 1 backbone. The capacity of these

derivatives to bind covalently to DNA and to form stable

DNA adducts was then investigated, together with their DNA

sequence specificity and growth inhibitory effects.

Chemistry. The synthesis of analogues 4-7 and 8-10 were

from the respective difluoro anthracenedione precursors, 12 or

16. The difluoro anthracenediones 12 and 16 were synthesized

via a Friedel-Crafts bis-cycloacylation from the correspond-

ing anhydrides 11 and 13. The difluoro anthracenediones 12

and 16 were then reacted under S_NAr conditions, with the

appropriate methylenediamine, to yield the desired substituted

anthracenediones 4-7 and 8-10 (Scheme 1). This methodo-

logy was established by Krapcho et al.^28,29

Formaldehyde-Activated Anthracenediones Form Stable

Covalent Drug-DNA Adducts. All formaldehyde-activated

anthracenedione derivatives were able to stabilize dsDNA by

forming DNA adducts that were detectable using the in vitro

cross-linking assay. The formation of these adducts was

The modeling demonstrated that G₅cross-linked bis ad-

ducts are also preferred for 1 derivatives 4-6. However, the

G₅bis adduct formed with 4 induced noticeable deformation

to the hexanucleotide structure. While it was possible to form

G₅bis adducts with5 and 6 withoutsignificantly distortingthe

helix, it was noted that these adducts were less flexible than

those formed with 7. The reduced flexibility could reduce the

ability of the helix to reform after random structural fluctua-

tions that occur at 37 ꢀC, thereby making the aminal linkages

more susceptible to hydrolysis.

Growth Inhibition Induced by Formaldehyde-Activated

Anthracenediones. The biological activity of each 1 derivative

6858 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19

Mansour et al.

Figure 6. pH dependence of adduct formation displayed by 1 and 2 derivatives. (A,B) 1 (40 μM), 5 (12 μM), 6 (8 μM), 7 (4 μM), 2 (5 μM), 8

(50 μM), 9 (50 μM), and 10 (40 μM) were incubated with 2 mM formaldehyde at 37 ꢀC for 16 h in PBS adjusted to various pHs ranging from

5.4 to 9.6 (as indicated). (C) Quantitation of gel (A). The percentage of dsDNA stabilized for 1 (O), 5 ( ꢀ ), 6 (0), and 7 (9) reacted DNA samples

was quantitated and expressed as a function of pH. (D) Quantitation of gel (B). The percentage of dsDNA stabilized for 2 (O), 8 ( ꢀ ), 9 (0), and

10 (9) reacted DNA samples was quantitated and expressed as a function of pH.

dependent on drug and formaldehyde concentration, with 1

analogues bearing extended side-chains forming adducts at low

concentrations compared to 1. In contrast, 2 formed more

adducts at low concentrations relative to its analogues with

extended side-chains (Figure 2). 4-7 and 2 all formed adducts

to a similar degree, indicating that the side-chain length is not a

major determinant in adduct formation but rather the presence

of a primary amino group is the primary determinant. The pri-

mary amino group is known to be considerably more reactive

than a secondary amino group, thus providing a greater con-

centration of reactive drug-Schiff base precursor. Moreover,

this drug-Schiff base complex is also more stable than second-

ary amino groups, therefore providing greater potential for

reaction with DNA. The presence of the primary amine was

previously shown to be crucial in the formation of 2-DNA

adducts in the presence of formaldehyde as BBR2378 (a

dimethyl N-substituted analogue of 2) was unable to stabilize

dsDNA in vitro nor form monoadducts in the presence of

formaldehyde.²⁵The introduction of the pyridine ring for the

dihydroxyphenylene moiety of the 1 chromophore has been

associated with a compromised DNA binding affinity.³⁰This is

consistent with the results presented in Figure 2: each com-

pound with a 2 backbone is less effective than the correspond-

ing 1-based compound.

For all the derivatives, the formation of adducts was also

dependent upon the level of formaldehyde (Figure 4). Also,

Article

Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19 6859

Figure 7. Formation of 4-DNA adducts. (A) Concentration dependence of adduct formation displayed by 4, compared to 7 and 2.

(B) Formaldehyde concentration dependence of adduct formation displayed by 4 and 7. (C) Stability of adducts induced by 4 and 7. (D) Time

dependence of adduct formation displayed by 4.

there was no adduct formation in the absence of formalde-

hyde, confirming a vital role for formaldehyde in the forma-

tion of drug-DNA adducts. Following nucleophilic attack

by an amine within a side-chain of the drug, formaldehyde

forms a reactive Schiff base intermediate, providing a likely

binding site to DNA. The nature of the 2-DNA covalent

linkage was shown to be a methylene bridge formed by

formaldehyde.²⁵Given the structural similarities shared by all

derivatives, and by analogy with 2, this was also expected to be

the nature of the covalent linkage created between the deriva-

tives and DNA. This was indeed confirmed by mass spectro-

metric analysis of the 8-DNA adduct (data not shown).

1 can exist as different ionic types and conformations in

aqueous solutions and a change in pH (and temperature) can

cause an equilibrium shift between the forms of the 1 mole-

cule. However, it has been established that 1 exists predomi-

nantly in one molecular form in neutral aqueous solutions.³¹

A change in the pH induces a conformational change in 1

where the nitrogen atom in the lateral chain forms a hydro-

gen bond with the deprotonated oxygen atom of the hydro-

xyl group at the tenth position. This may explain the strong

pH-dependence of formaldehyde-activated 1-DNA adduct

formation (Figure 6). Moreover, while the secondary amino

group of 1 could conceivably form an unstable aminal with

formaldehyde, it cannot form an imine. This inability to

form an imine may explain the different behavior of 1 with

variation of pH (Figure 6). This also may explain the

comparative pH-independent response of the derivatives.

A highlight of the present study is the outstanding stability

exhibited by both 7 and 10 drug-DNA adducts relative to

the parent compounds. As anticipated, all 1-DNA adducts

were lost in the first hour (Figure 5C), while 2-DNA adducts

were more stable but were still lost relatively quickly (in 120 min,

Figure 5D). Strikingly, adduct stability increased with increas-

ing side chain length. 7 formed the most stable adducts, with

∼100-fold enhanced stability compared to the 1-DNA lesion

(Table 1).

For a conventional DNA interstrand cross-linking agent

such as cisplatin, the two DNA strands are covalently

joined together by the drug, thereby generating a stable

cross-link. This is also postulated to be the case for stable 7

adducts where aminal linkages to the N-2 of guanine on

both side chains could be accommodated to form a cross-

link within steric constraints, as anticipated by initial

molecular modeling studies¹⁷and shown to be feasible

in the molecular model of a 7-DNA adduct (Figure 9).

Adducts formed by other formaldehyde activated deriva-

tives previously studied act as monoadducts where the

adduct is covalently bound via a single methylene bridge

(provided by formaldehyde) to a single DNA strand and

presumably complemented by hydrogen bonds to the

opposite DNA strand with the chromophore intercalated

6860 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19

Mansour et al.

Figure 8. Sequence specificity of 1 and 2 derivatives. (A) In vitro transcription assay demonstrating the sequence specificity of 1 and 2 derivatives

at 4 h. Drug-reacted samples were subjected to a transcription stop assay. Following initiation from the lac UV5 promoter, the transcription

complex was elongated for 5 min and then terminated. Lanes I and E are controls representing an initiated complex that has not been elongated

and extension of this complex to generate the full length transcript, respectively. Lanes A and G are sequencing lanes which were obtained using

3⁰-O-methoxy-ATP or 3⁰-O-methoxy-GTP during elongation, respectively. The length of selected transcripts is indicated at the left of the

phosphorimage. (B) The sequence selectivity of formaldehyde-activated 4, (C) 7, and (D) 2. The % transcriptional blockage of each transcript in

the drug lanes from nucleotide 30 to 100 in (A) were quantitated and are expressed as a function of the sequence of a portion of the 512 bp DNA

fragment. The sequence of the nontemplate DNA strand is shown. Transcriptional blockages less than an arbitrary value of 0.2% have been

omitted. The numbers at the bottom of histogram (D) indicate the length of transcripts represented at the left of the phosphorimage in (A).

between neighboring base pairs, accounting for the relative

instability.¹³

blockages were produced by most 1 derivatives within the first

4 h (Figure 8A) and increased after 24 h (data not shown). 2

derivatives did not produce high levels of transcriptional

blockages (Figure 8A) because a much higher concentration

was required for adduct formation. Both 4- and 2-generated

lesions caused the same level of transcriptional blockages,

consistent with the same level of adducts caused by the two

compounds in the in vitro cross-linking assay.

A distinguishing aspect of the transcriptional blockages

is that they all occurred in a sequence-dependent manner

(Figure 8A). The strong similarity of these distinct transcrip-

tional blockages to those observed for 1 and 2 indicate

that the sequence preference of all the derivatives is CpG

and CpA specific.^11,25This was confirmed by quantitation

of truncated transcripts (Figure 8B,C,D) where the most

intense blockage sites for derivatives was at CpG and

CpA sequences (the full data set is available as Supporting

Formaldehyde-Activated Anthracenedione Derivatives Form

CpG- and CpA-Specific Adducts. DNA-ligand interactions

are of crucial importance to biological processes such as DNA

replication and transcription. It has become clear that DNA

sequence specificity is an essential factor contributing to the

cytotoxic influence of numerous antitumor agents.^32-34

Recently, the DNA sequence involved in the formaldehyde-

mediated 2-DNA adducts was characterized, and formalde-

hyde activated 2 was shown to form CpG and CpA specific

adducts,²⁵revealing a prominent resemblance to 1 in sequence

selectivity.^11,25In vitro transcription studies showed that each

derivative produced DNA lesions capable of blocking the

progression of RNA polymerase (Figure 8), resulting in an

overall inhibition of transcription and the generation of trun-

cated transcripts. Relatively high levels of transcriptional

Article

Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19 6861

Figure 9. Modeling of 7 intercalated into DNA compared to the covalently bound form. 7 intercalated between the C₃and G₄bases of the

hexanucleotide d(ACCGGT)₂from the minor groove (left-hand side) and the compound covalently bound through aminal linkages to the N-2

atom of the G₅residue on each strand. 7 is shown in green, and the additional carbon resulting from formaldehyde activation is red. The adenine

residues are shown in blue, the thymine residues in cyan, the guanine residues in white, and the cytosine residues in yellow. The DNA backbones

are shown in purple.

Table 2. Growth Inhibition Induced by 1 Derivatives As Single Agents

and in Combination with a Formaldehyde-Releasing Prodrug^a

Scheme 1. Synthesis of the 1-Derived 4-7 and the 2-Derived

8-10 Analogues^a

MCF-7

compd

IC₅₀(nM)

IC₅₀þ 3 (nM)^d

CI

1

4

5

6

7

3

71 ( 30

190 ( 88

2900 ( 2300

12000 ( 3000

6700 ( 600

107 ( 7^c

240 ( 120

190 ( 100

990 ( 520

840 ( 110

900 ( 320

N/A

3.56

1.16

1.27

0.86

0.98

PC-3

compd

IC₅₀(nM)

IC₅₀þ 3 (nM)^d

CI

^aReagents and conditions: (a) hydroquinone, AlCl₃, NaCl, 200 ꢀC,

5 h, 60%; (b) 1,4-difluorobenzene, AlCl₃, vV, 20 h, 84%; (c) oleum (30%

SO₃) 135 ꢀC, 3 h, 86%; (d) 4 equiv NH₂(CH₂)_nNH₂, THF, 50 ꢀC, 20 h.

1

4

5

6

7

3

680 ( 160

590 ( 480

6300^b

10800^b

400 ( 160

480 ( 320

1200 ( 280

730 ( 120

1290 ( 30

0.76

1.03

0.73

0.39

0.85

may be similar to 1. In fact, the guanine base of DNA is well-

known to contain reactive and accessible sites.^36-38As other

formaldehyde-activated drugs like 1, doxorubicin and 2 bind

to guanine via the exocyclic N-2 amino group,^13,17,25it is likely

that all of the formaldehyde-activated derivatives bind cova-

lently via a methylene group to the DNA through the exocyclic

N-2 amino group of guanine, leading to the formation of

N-C-N aminal linkages.

In addition to the importance of recognizing appropriate

DNA sequences in overcoming the problem of generalized

toxicity of a drug, the CpG recognition by the formaldehyde-

activated derivatives is significant because this dinucleotide

holds a distinctive function in the mammalian genome. The

possibility of finding any dinucleotide in a given DNA

sequence is 1/16 (about 6%). In humans, however, the mea-

sured frequency of the CpG dinucleotide is very low, an event

referred to as CG suppression. Despite this, some genomic

regions contain a high frequency of CpG dinucleotide known

as CpG islands, which typically occur in clusters within 300-

3000 bp regions. These are frequently associated with the

promoter region of many genes (in about 70% of human

promoters),^39-42making them an excellent selective target

for anthracenedione derivative-DNA adducts outlined in

this work.

4800 ( 1000

230 ( 40^c

HL60

compd

IC₅₀(nM)

IC₅₀þ 3 (nM)^e

CI

1

4

5

6

7

3

7 ( 2

9 ( 3

260 ( 110

1700 ( 190

3800 ( 1100

22 ( 13^c

11 ( 6

17 ( 10

205 ( 10

360^b

1.74

1.82

1.17

0.87

0.73

350 ( 60

^aIC₅₀values are shown as the concentration as which 50% of cells

survived. The interaction of the drugs in combination was evaluated from

IC₅₀values by calculating the combination index. The combination index

(CI) was calculated as described in the Experimental Section. Two experi-

ments were performed independently in quadruplicate. Data is shown as

mean ( SD. ^bDenotes data from a single experiment performed in

quadruplicate. ^cDenotes that the concentration is μM. ^dA 100-fold molar

excess of 3 was used. ^eA 40-fold molar excess of 3 was used.

Information). Because 1 has been found to intercalate prefer-

entially at CG and CA sites in DNA,³⁵it therefore appears that

this enables subsequent activation by formaldehyde at these

sites, leading to the covalent binding to adjacent nucleotides on

the DNA where a reactive site is available. This also suggests

that the sequence specificity of intercalation by the derivatives

6862 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19

Mansour et al.

Molecular Modeling Studies. In an effort to explain the

exceptional stability of 7-DNA adducts compared to

adducts formed with other derivatives, modeling studies were

undertaken. In the energy minimized model (Figure 9), it is

apparent that the intercalation optimum for 7 is symmetri-

cal, which contrasts with the earlier study with the parent

compound 1.¹⁷Similar models of 1 intercalated between the

C₃and G₄bases of the hexanucleotide d(ACCGGT)₂could

not produce a symmetrical intercalation optimum. An ob-

vious explanation for the increased stability of 7-DNA

adducts is the potential for intercalated 7 to form aminal

linkages at two accessible guanine N-2 positions with minimal

distortion to the helix. Furthermore, with the less stable

drug-DNA adducts, the aminal bonds would be more read-

ily hydrolyzed in a molten DNA state because of the con-

tribution of base-pairing to the electronic state of the gua-

nine ring. Cytosine residues lend electron density to the

guanine ring through multiple hydrogen bonds, contributing

to the reactivity of the unmodified guanine N-2. Formation

of an aminal adduct at N-2 would make N-2 withdraw

electrons less strongly from the adjacent aminal carbon,

making the aminal carbon less susceptible to nucleophilic

attack. In this way, the longer cross-link induced by 7 may

contribute to chemical stability by contributing to the DNA

duplex thermostability

Interactions with the Formaldehyde-Releasing Prodrug 3.

Significant synergism was demonstrated in cultured cells

treated with a combination of formaldehyde releasing pro-

drugs and doxorubicin.^19,43However, this synergy is not

observed for 1 in combination with formaldehyde-releasing

prodrugs (unpublished results) and this is most likely due to

the instability of DNA adducts. Derivatives 4-7 form increas-

ingly stable DNA adducts following formaldehyde activation

in vitro (Figure 5 and Table 1), and this generally appeared to

be associated with decreasing CI values (Table 2). 4, which

achieved a similar level of reactivity as 7 (Figure 7) but formed

much less stable DNA adducts, did not exhibit synergy in

combination with 3 in any of the cell lines studied. This may

reflect the importance of inducing stable DNA adducts which

can maximize DNA damage and potentially impair crucial

cellular processes such as DNA replication and transcription.

However, in PC-3 cells, 7 (that formed the most stable DNA

adducts) did not conform to the general trend. This may be due

to our observation that increased side chain length appears to

favor lysosome versus nuclear localization in cell lines with

extensive lysosomal networks (unpublished results), and this is

a subject of future investigations.

A problem with the development of new anticancer com-

pounds based on traditional chemotherapeutic compounds,

such as the anthracyclines and anthracenediones, is their limited

selectivity for tumor vs nontumor cells. A potential solution can

now be presented with respect to the new types of compounds

outlined in the present study;an inherent dependence on

formaldehyde for biological activity means that tumor-targeted

formaldehyde is a potential treatment strategy in conjunction

with anthracenedione therapy. Another key problem in the

development of anticancer drugs is the large gap between pro-

mising findings in preclinical in vitro models and the results of

clinical trials. Drug combinations that displayed evidence of

synergy in this study will next be subjected to rigorous analysis,

particularly with respect to their anticancer activity in clono-

genic assays in vitro and effects on tumors in vivo.

Experimental Section

Materials. 3 was synthesized and provided by Professor

Abraham Nudelman, Ramat-Gan University, Israel. Formalde-

hyde solution (40% w/v) and chloroform were purchased from

Merck (BDH). Calf thymus DNA was obtained from Worthing-

ton Biochemical Corporation. Tris-saturated ultrapure phenol

and 1 Kb plus DNA marker were purchased from Invitrogen.

Ultrapure dNTPs, 3⁰-O-methyl NTPs, GpA dinucleotide, [R-³²P]

dATP, [R-³²P] dCTP, [R-³²P] UTP (3000 Ci/mmol), RNAguard

RNase inhibitor, and ProbeQuant G-50 columns were all from

GE Healthcare. TEMED and the two restriction enzymes

HindIII and PvuII were purchased from Promega. E. coli RNA

polymerase and glycogen were both obtained from Roche. The

Klenow fragment of DNA polymerase I and BSA were obtained

from New England Biolabs. Acrylamide:bisacrylamide solution

was purchased from Amresco and molecular biology grade urea

from MP Biomedicals. Formamide, heparin, and the tetrazolium

salt MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium

bromide] were all obtained from Sigma Chemical Co. A Maxi

Plasmid purification kit was purchased from Qiagen and DTT

and ammonium persulfate were both from Bio-Rad.

Methods. Drug Preparation for Biological Assays. All drugs

were dissolved in Milli-Q water (distilled water purified on a

Millipore Q system) to a final concentration of between 0.5 and

2 mM and subsequently stored at -20 ꢀC. On the day of each

experiment, the precise concentrations of the drugs were deter-

mined spectrophotometrically on a Cary 118C spectrophotometer

using ε = 19200 M^-1cm^-1at 608 nm for 1 and its analogues and

ε = 16500 M^-1cm^-1at 641 nm for 2 and its derivatives. 3 was

diluted in DMSO in glass vials using glass syringes. Formalde-

hyde and 3 solutions were freshly prepared at the time of each

experiment.

DNA Source and End-Labeling. The plasmid pCC1²⁷was

initially restriction digested with HindIII at 37 ꢀC for 2 h and

subsequently end-labeled at the 5⁰-overhang with [R³²P] dCTP

using the Klenow fragment of DNA polymerase I. A G-50

ProbeQuant chromatography column was then used to remove

unincorporated label from the labeled DNA fragment. The

sample was subsequently subjected to phenol/chloroform

extraction, ethanol precipitation, and resuspension in 1ꢀ TE

(10 mM Tris, 1 mM EDTA, pH 8.0). The final DNA concentra-

tion was adjusted to 400 μM_bpby the addition of sonicated calf

thymus DNA.

In Vitro Cross-Linking Assay. The end-labeled DNA was used

for in vitro DNA binding reactions where covalently bound

drug-DNA adducts were formed. Each reaction mixture typi-

cally contained 25 μM_bpend-labeled DNA, formaldehyde, and

the anthracenedione of interest in phosphate-buffered saline,

pH 7.0 (PBS), and were reacted at 37 ꢀC for the desired amount

of time. Noncovalently bound drug was removed from the

reaction mixture by two extractions with an equal volume of

Tris-saturated phenol followed by a single extraction with an

Another factor that might be contributing to the synergy

displayed by some of the derivatives with 3 is the esterase

hydrolysis of 3 and the release of the cellular histone deace-

tylase inhibitor butyric acid.⁴⁴Histone deacetylase (HDAC)

inhibitors have now emerged as a new class of potential anti-

cancer agents for the treatment of solid and hematological

malignancies.⁴⁵HDAC inhibitors change the expression of

genes by causing an increase in histone acetylation, thus

regulating chromatin structure and transcription.^46,47Inhi-

bitors such as suberoylanilide hydroxamic acid (SAHA)

have been tested in combination with a variety of conven-

tional cytotoxic chemotherapeutics.⁴⁸It is well documented

that cultured cells pretreated with HDAC inhibitors display

enhanced activity of DNA damaging agents.⁴⁹However,

given our knowledge of anthracenedione activation and

observation of doxorubicin/3 interactions,⁴³the formaldehyde

contribution to the observed synergy is more likely.

Article

Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19 6863

equal volume of chloroform. Drug-reacted DNA samples were

then precipitated with a mixture containing ethanol, sodium

acetate, and glycogen as a carrier to improve recovery. Sample

pellets were subsequently resuspended in 1ꢀ TE buffer and

denatured in two volumes of loading dye (90% formamide,

10 mM EDTA, 0.1% bromophenol blue, 0.1% xylene cyanol) at

52 ꢀC (or a range of temperatures where the denaturation

temperature was determined) for 5 min. Samples were subjected

to electrophoresis through 0.8% agarose gels overnight at 40 V

in 1ꢀ TAE (Tris-acetate EDTA) buffer. Gels were vacuum-

dried in a Bio-Rad model 583 gel drier and subsequently

exposed to a phosphor screen for 4 h. A Typhoon Trio Phos-

phorImager (GE Healthcare) was used to generate a phosphor-

image of the dried gel, and ImageQuant software was used for

analysis and quantitation.

Experiments assessing drug-DNA adduct stability were per-

formed in an identical manner, however, each sample was sub-

jected to an additional extraction procedure following incubation

at 37 ꢀC for defined time periods. The additional step utilized an

extraction once with phenol, once with chloroform, and a final

ethanolprecipitation. ResuspendedDNA samples weresubjected

to electrophoresis as described above.

Molecular Modeling. Molecular modeling was carried out

using HyperChem Release 7.5 Hypercube Inc. Point charges for

the 1 derivatives and the various covalent adducts were based

on AM1 optimizations. The duplex B-type hexanucleotide

d(ACCGGT)₂was generated from the nucleic acid database

and the 1 derivative manually intercalated between the C₃and

G₄bases. The drug-oligonucleotide complex was then mini-

mized in vacuo using the Amber 99 force field and a Polak-

Ribiere conjugant-gradient algorithm with a 5 ꢀ 10^-5kcal/

˚

(A mol) convergence criteria. Covalent adducts, involving the

N-2 of G₄and G₅, were prepared and energy minimized in a

similar fashion.

Cell Culture and Growth Inhibition Assay. MCF-7 breast

adenocarcinoma cells (obtained from Dr Rosanna Supino,

Istituto Nazionale Tumori, Milano, Italy), PC-3 prostate cancer

cells (obtained from Dr Carleen Cullinane, Peter McCallum

Cancer Institute, Melbourne, Australia), and HL60 promyelo-

cytic leukemia cells (purchased from ATCC, VA, USA) were

maintained in RPMI 1640 media (JRH Biosciences) supplemen-

ted with 10% fetal calf serum (Trace Scientific) at 37 ꢀC in a

humidified atmosphere with 5% CO₂.

Cells were seeded at a density of 4 ꢀ 10³cells/well (MCF-7

and PC-3) or 10⁴cells/well (HL60) into individual 96-well plates

in 100 μL of complete media (RPMI 1640 with 10% FCS) and

incubated overnight at 37 ꢀC with 5% CO₂. The following day, a

serial dilution of drugs was prepared and 50 μL/well of each

drug was added in quadruplicate. Drug treatment consisted of

serial dilutions of one anthracenedione, 3, or a combination of

both anthracenedione and prodrug. For single drug treatments,

50 μL of complete media was added to the wells to adjust the

volume to 200 μL/well. For 4 h treatments (MCF-7 and PC-3),

compounds were removed after 4 h and 200 μL fresh media

replaced. After 72 h, 100 μL of the tetrazolium salt MTT

dissolved in PBS (1 mg/mL and filtered through a 0.22 μm

Millipore filter) was added to each well containing MCF-7 or

PC-3 cells for 3-4 h at 37 ꢀC. Overlying media was then removed

by aspiration, and 100 μL/well of DMSO was added to solubi-

lize the formazan (metabolized MTT). A Molecular Devices

Spectra Max 250 microplate reader was used to measure the

absorbance at 570 nm. HL60 cell growth inhibition was deter-

mined by MTS metabolism. MTS (Promega, WI, USA) and

phenazine methosulfate (PMS) (Sigma, MO, USA) were each

dissolved in PBS at a concentration of 1.6 mg/mL and sterile

filtered. MTS and PMS were mixed at a ratio of 9:1 and 50 μL of

this added to each well and the plates incubated for 3-4 h at

37 ꢀC. Absorbance at 490 nm was determined in each well using

a microplate reader.

In Vitro Transcription. Plasmid pCC1 was subjected to restric-

tion digestion with PvuII and HindIII for 2 h at 37 ꢀC to liberate

a 512 bp fragment containing the lac UV5 promoter.^27,50The

512 bp restriction fragment was subsequently isolated using electro-

phoresis followed by electroelution as previously described.⁵¹The

DNA fragment was finally purified by phenol/chloroform extrac-

tion, ethanol precipitation, and resuspension in 1ꢀ TE buffer in

preparation for drug reaction.

The 512 bp DNA fragment (25 μMbp) was reacted with the

drug of interest in the presence of formaldehyde in PBS (pH 7.0)

at 37 ꢀC for defined time periods. Drug-reacted DNA samples

were then ethanol precipitated and resuspended in 1ꢀ TC

(transcription buffer: 40 mM Tris pH 8.0, 100 mM KCl, 3 mM

MgCl₂, and 0.1 mM EDTA). Each sample was subsequently

incubated with a transcription mixture containing 0.025 U/μL

E. coli RNA polymerase, 10 mM DTT, 125 μg/mL BSA, and

2 U/μLRNAguardRNase inhibitor in1ꢀ TCbufferfor 15minat

37 ꢀC. Heparin (400 μg/mL) was then added for 5 min to remove

RNA polymerase from nonspecific binding sites on the DNA. An

initiation mixture containing 200 μM GpA, 5 μM ATP, 5 μM

GTP, and [R³²P] UTP in TC was then added to each sample again

for 5 min at 37 ꢀC. Transcripts were allowed to elongate by the

subsequent addition of an elongation mixture containing 6 mM

of each NTP and 1.2 mM KCl in 1ꢀ TC for 5 min at 37 ꢀC. Trans-

cription was terminated by the addition of an equal volume of

loading/termination buffer (9 M urea, 10% sucrose, 40 mM

EDTA, 0.1% xylene cyanol, and 0.1% bromophenol blue in

2 ꢀ TBE). All transcription samples were denatured at 90 ꢀC for

5 min and immediately quenched on ice.

Sequencing of the RNA transcripts was achieved by incubat-

ing an unreacted batch of the 512 bp DNA fragment with the

transcription mixture as described above. Samples were then

allowed to elongate in the presence of 90 μM 3⁰-O-methoxy-ATP

(or 3⁰-O-methoxy-GTP), 10 μM ATP (or GTP), 2 mM CTP,

GTP (or ATP), UTP, and 400 mM KCl at 37 ꢀC for 5 min. Reac-

tions were terminated by the addition of an equal volume of

loading/termination buffer, denatured at 90 ꢀC for 5 min, and

quenched on ice.

Absorbance values were plotted against the log of the drug

concentration, and sigmoidal fits were applied to yield IC₅₀

values using Origin software (Microcal, MA, USA). IC₅₀values

were determined as the concentration at which 50% of cells were

metabolically active. Drug synergism was evaluated using the

method of Chou and Talalay,⁵²where a combination index⁴³

<1 indicates synergy, a CI = 1 indicates additivity, and a CI >1

indicates antagonism. The CI was calculated using the formula:

CI ¼ ðIC₅₀ðdrug 1 in combination with drug 2Þ=

IC₅₀ðdrug 1 aloneÞÞ þ

ðIC₅₀ðdrug2 in combination with drug 1Þ=IC₅₀ðdrug 2 aloneÞÞ

RNA drug-induced blocked transcripts were separated on a

12% acrylamide denaturing gel (19:1 acrylamide:bisacrylamide

containing 7 M urea). The high resolution sequencing gel was

prepared in 1ꢀ TBE buffer and subjected to pre-electrophoresis

at 1800 V for 1-2 h. Denatured samples were loaded onto the gel

and electrophoresed at 2000 V for approximately 2 h. The gel

was then fixed with 10% acetic acid/10% MeOH, dried on a Bio-

Rad gel drier, and exposed to a phosphor screen for 4 h.

Phosphorimaging analysis and quantitation were performed

as described for the in vitro cross-linking assay.

Chemistry. General Experimental. All nonaqueous reactions

were performed in oven-dried glassware under an atmosphere of

dry nitrogen, unless otherwise specified. All anhydrous solvents

were purified using a Braun purification system. All other

solvents were reagent grade. Petroleum ether describes a mixture

of hexanes in the bp range 40-60 ꢀC. Analytical thin-layer chro-

matography was performed on Merck silica gel 60F₂₅₄aluminum-

backed plates and were visualized by fluorescence quenching

under UV light. All NMR spectra were recorded on a Bruker

Avance DRX 300 with the solvents indicated (¹H NMR at

6864 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19

Mansour et al.

300 MHz). Chemical shifts are reported in ppm on the scale,

referenced to the appropriate solvent peak. HRESMS were

recorded at the Australian National University Mass Spectrometry

Facility using a Waters LCT Premier XE (ESI TOF mass spectro-

meter). LCMS was recorded on a Waters ZQ 3100 system using a

Waters 2996 diode array detector. LCMS conditions used to assess

purity of compounds were as follows: column, RP XBridge TM

C18 5 μM 4.6 mm ꢀ 100 mm; injection volume, 10 μL; flow rate,

1.5 mL/min; gradient, 0-100% of B over 10 min, (solvent A,

water; solvent B, AcCN, 0.1% formic acid). All compounds were

g95% pure.

Experimental. 1,4-Dihydroxy-5,8-bis(2-(2-hydroxyethylamino)-

ethylamino)anthracene-9,10-dione 1. Mitoxantrone hydrochloride

1 was obtained commercially from Sigma Chemical Co.

6,9-Bis(2-aminoethylamino)benzo[g]isoquinoline-5,10-dione

dimaleate 2. Pixantrone dimaleate 2 was kindly provided by Cell

Therapeutics Europe (Bresso, Italy).

(M þ H) 357.1559; C₁₈H₂₀N₄O₄requires (M þ H), 357.1563. ¹H

NMR (DMSO) δ 10.40 (2H, t), 8.21 (6H, bs), 7.64 (2H, s), 7.17

(2H, s), 3.85-3.76 (4H, m), 3.08-3.00 (4H, m). LCMS: T_r= 4.24

min, 95% purity; m/z (%) = 357 (100), [M þ H]^þ296 (100).

1,4-Bis(3-aminopropylamino)-5,8-dihydroxyanthracene-9,10-

dione Dimaleate 5. A mixture of the difluoro dione 12 (100 mg,

0.36 mmol) and 1,3-diaminopropane (302 μL, 3.6 mmol) in THF

(5 mL) was stirred at 50 ꢀC for 20 h. The solution was cooled, and

the precipitate was filtered off and washed with THF. The filtrate

was concentrated in vacuo and the residue dried under high

vacuum to remove excess amine. The residue was dissolved in

MeOH(2mL), anda solutionof maleicacid(126 mg, 1.1 mmol)in

MeOH (1.5 mL) was added. A sufficient volume of EtOAc was

added and a precipitate developed. The solid was filtered off,

washing with EtOAc to afford the dimaleate of 5 as a dark-blue

solid (101 mg, 45%).⁵³IR ν_max2953, 1606, 1555, 1454, 1353, 1205,

823 cm^-1. HRESMS found: (M þ H) 385.1876; C₂₀H₂₄N₄O₄

requires (M þ H), 385.1876. ¹H NMR (DMSO) δ 10.52 (2H, t),

7.69 (6H, bs), 7.56 (2H, s), 7.17 (2H, s), 6.01 (4H, s), 3.64-3.56

(4H, m), 2.92 (4H, t), 1.97-1.91 (4H, m). LCMS: T_r= 4.46 min,

95% purity; m/z (%) = 385 (100), [M þ H]^þ328 (60).

3,6-Difluorophthalic Anhydride 11. 3,6-Difluorophthalic

anhydride was prepared in an identical manner to that of Krapcho

et al.²⁸

1,4-Difluoro-5,8-dihydroxyanthracene-9,10-dione 12. The pro-

cedure followed was similar to that of Krapcho et al.²⁸Hydro-

quinone (18.9 mmol) and 3,6-difluorophthalic anhydride 11 (18.0

mmol) were added portionwise to a stirred melt of AlCl₃(198.0

mmol) and sodium chloride (72.0 mmol) at 200 ꢀC under a nitro-

gen atmosphere. The mixture was stirred for 5 h at 200 ꢀC. The

solution was allowed to cool, and ice-water was added. Concen-

trated HCl (40 mL) was added, and the solution was stirred at

room temperature for 3 h. The resulting precipitate was filtered

off, washed with water, and dried in a vacuum oven to afford the

dione 12 as a red solid (60%). ¹H NMR (DMSO) δ 7.87 (2H, m),

7.42 (2H, s).²⁸

1,4-Bis(4-aminobutylamino)-5,8-dihydroxyanthracene-9,10-dione

Dimaleate 6. A mixture of the difluoro dione 12 (75 mg, 0.27 mmol)

and 1,4-diaminobutane (272 μL, 2.7 mmol) in THF (5 mL) was

stirred at 50 ꢀC for 20 h. The solution was cooled, and the

precipitate was filtered off and washed with THF. The filtrate was

concentrated in vacuo, and the residue dried under a stream of

nitrogen to aid in removal of excess amine. The residue was

dissolved in MeOH (2 mL), and a solution of maleic acid (94 mg,

0.81 mmol) in MeOH (1.5 mL) was added. A sufficient volume of

EtOAc was added, and a precipitate developed. The solid was

filtered off and washed with EtOAc to give the dimaleate of 6 as a

dark-blue solid (78 mg, 45%).⁵³IR ν_max2942, 1607, 1556, 1449,

1202, 860 cm^-1. HRESMS found: (Mþ H) 413.2189; C₂₂H₂₈N₄O₄

requires (M þ H), 413.2189. ¹H NMR (DMSO) δ 10.53 (2H, t),

7.66 (6H, bs), 7.56 (2H, s), 7.15 (2H, s), 6.01 (4H, s), 3.58-3.50

(4H, m), 2.90-2.82 (4H, m), 1.70-1.65 (8H, m). LCMS: T_r= 4.66

min, 95% purity; m/z (%) = 413 (90), [M þ H]^þ325 (100).

1,4-Bis(5-aminopentylamino)-5,8-dihydroxyanthracene-9,10-

dione Dimaleate 7. A mixture of the difluoro dione 12 (75 mg,

0.27 mmol) and 1,5-diaminopentane (130 μL, 1.1 mmol) in THF

(5 mL) was stirred at 50 ꢀC for 20 h. The solution was cooled, and

the precipitate was filtered off and washed with THF. The filtrate

was concentrated in vacuo, and the residue dried under a stream

of nitrogen to aid in removal of excess amine. The residue was

dissolved in MeOH (2 mL), and a solution of maleic acid (142 mg,

0.93 mmol) in MeOH (1.5 mL) was added. A sufficient volume of

EtOAc was added and a precipitate developed. The solid was

filtered off and washed with EtOAc to give the dimaleate of 7 as

a dark-blue solid (73 mg, 40%). IR ν_max2940, 1609, 1562, 1456,

1353, 1200, 863 cm^-1. HRESMS found: (M þ H) 441.2496;

C₂₄H₃₂N₄O₄requires (M þ H), 441.2502. ¹H NMR (DMSO) δ

10.63 (2H, t), 7.55 (2H, s), 7.15 (2H, s), 6.00 (4H, s), 3.53-3.45

(4H, m), 2.95-2.78 (4H, m), 1.70-1.41 (12H, m). LCMS: T_r=

4.88 min, 95% purity; m/z (%) = 441 (35), [M þ H]^þ213 (100).

6,9-Bis(3-aminopropylamino)benzo[g]isoquinoline-5,10-dione

Dimaleate 8. A mixture of the difluoro dione 16 (100 mg, 0.41

mmol) and 1,3-diaminopropane (340 μL, 4.1 mmol) in THF

(5 mL) were stirred at 50 ꢀC for 20 h. The solution was cooled,

and the precipitate was filtered off and washed with THF. The

filtrate was concentrated in vacuo and the residue dried under

high vacuum to remove excess amine. The residue was dissolved

in MeOH (2 mL), and a solution of maleic acid (142 mg, 1.2

mmol) in MeOH (1.5 mL) was added. A sufficient volume of

EtOAc was added, and a precipitate developed. The solid was

filtered off and washed with EtOAc to give the dimaleate of 8 a

dark-blue solid (174 mg, 73%).²⁹IR ν_max3052, 1566, 1523, 1461,

1361, 1156, 860 cm^-1. HRESMS found: (M þ H) 354.1917;

C₁₉H₂₃N₅O₂requires (M þ H), 354.1930. ¹H NMR (DMSO) δ

11.10 (1H, t), 11.00 (1H, t), 9.45 (1H, s), 8.97 (1H, d), 8.05 (1H, d),

4-(2,5-Difluorobenzoyl)-nicotinic acid 14 and 3-(2,5-Difluo-

robenzoyl)-isonicotinic Acid 15 (3:1 Mixture). The procedure

followed was similar to that of Krapcho et al.²⁹Aluminum

chloride (134.0 mmol) was added portionwise to a stirred solu-

tion of pyridine-3,4-dicarboxylic acid anhydride 13 (33.5 mmol)

and 1,4-difluorobenzene (218.0 mmol) at 90 ꢀC. The solution was

refluxed for 24 h under a nitrogen atmosphere. The excess 1,4-

difluorobenzene was distilled off and can be subsequently

recycled. Ice-water (70 mL) was then added to the resulting residue,

and concentrated HCl (6.3 mL) was added. The resulting pre-

cipitate was filtered off, washed with water, and dried in a vacuum

oven to afford a white solid (84%) as a 3:1 mixture of 4-(2,5-

difluorobenzoyl)-nicotinic acid 14 and 3-(2,5-difluorobenzoyl)-

1

isonicotinic acid 15, respectively. H NMR (DMSO) δ 9.15 (s),

8.91 (d) 8.90 (d), 8.76 (s), 7.85 (d), 7.63-7.22 (m).²⁹

6,9-Difluorobenzo[g]isoquinoline-5,10-dione 16. The proce-

dure followed was similar to that of Krapcho et al.,²⁹A 3:1

mixture of 4-(2,5-difluorobenzoyl)-nicotinic acid 14 and 3-(2,5-

difluorobenzoyl)-isonicotinic acid 15 (11.4 mmol) in fuming

H₂SO₄(7.5 mL, 30% SO₃) was heated at 135 ꢀC for 3 h. The

mixture was allowed to cool, and ice-water was added. The

solution was then neutralized by portionwise addition of solid

NaHCO₃. The resulting precipitate was filtered off and washed

with water. The solid was dried in a vacuum oven to obtain the

dione 16, a pale-yellow solid (86%). ¹H NMR (DMSO) δ 9.26

(1H, s), 9.06 (1H, dd), 7.91 (1H, d), 7.84-7.79 (2H, m).²⁹

1,4-Bis(2-aminoethylamino)-5,8-dihydroxyanthracene-9,10-dione

Dihydrochloride 4. A mixture of the difluoro dione 12 (75 mg, 0.27

mmol) and 1,2-diaminoethane (123 μL, 1.2 mmol) in THF (5 mL)

was stirred at 50 ꢀC for 20 h. The solution was cooled, and the

precipitate was filtered off and washed with THF. The filtrate was

concentrated in vacuo and the residue dried under high vacuum to

remove excess amine. The residue was dissolved in MeOH (2 mL),

and a solution of 1.25 N HCl in MeOH (1.62 mmol) was added. A

sufficient volume of EtOAc was added, and a precipitate devel-

oped. The solid was filtered off and washed with EtOAc to give the

dihydrochloride of 4 as a dark-blue solid (69 mg, 60%).⁵³IR ν_max

2967, 1605, 1560, 1455, 1153, 1055, 791 cm^-1. HRESMS found:

Article

Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19 6865

7.70 (6H, bs), 7.58 (2H, s), 6.00 (4H, s), 3.63-3.56 (4H, m), 2.90

(4H, t), 1.97-1.89 (4H, m). LCMS: T_r= solvent front, 95%

purity; m/z (%) = 354 (95), [M þ H]^þ325 (100).

van Oosterom, A. T.; Smyth, J. F. Mitoxantrone for the treatment of

advanced breastcancer: single-agent therapy in previously untreated

patients. Eur. J. Cancer Clin. Oncol. 1984, 20, 1141–1146.

(7) Durr, F. E. Anthracyclines and Anthracenedione-Based Anticancer

Agents; Elsevier: Amsterdam, 1988.

(8) Pratt, W. B.; Ruddon, R. W.; Ensminger, W. D.; Maybaum, J. The

Anticancer Drugs; Oxford University Press: New York, 1994.

(9) Burden, D. A.; Osheroff, N. Mechanism of action of eukaryotic

topoisomerase II and drugs targeted to the enzyme. Biochim.

Biophys. Acta 1998, 1400, 139–154.

(10) Faulds, D.; Balfour, J. A.; Chrisp, P.; Langtry, H. D. Mitoxantrone.

A review of its pharmacodynamic and pharmacokinetic properties,

and therapeutic potential in the chemotherapy of cancer. Drugs

1991, 41, 400–449.

(11) Parker, B. S.; Cutts, S. M.; Cullinane, C.; Phillips, D. R. For-

maldehyde activation of mitoxantrone yields CpG and CpA

specific DNA adducts. Nucleic Acids Res. 2000, 28, 982–990.

(12) Parker, B. S.; Cullinane, C.; Phillips, D. R. Formation of DNA

adducts by formaldehyde-activated mitoxantrone. Nucleic Acids

Res. 1999, 27, 2918–2923.

(13) Zeman, S. M.; Phillips, D. R.; Crothers, D. M. Characterization of

covalent adriamycin-DNA adducts. Proc. Natl Acad. Sci. U.S.A.

1998, 95, 11561–11565.

(14) Thorndike, J.; Beck, W. S. Production of formaldehyde from

N5-methyltetrahydrofolate by normal and leukemic leukocytes.

Cancer Res. 1977, 37, 1125–1132.

(15) Taatjes, D. J.; Gaudiano, G.; Resing, K.; Koch, T. H. Redox

pathway leading to the alkylation of DNA by the anthracycline,

antitumor drugs adriamycin and daunomycin. J. Med. Chem. 1997,

40, 1276–1286.

(16) Taatjes, D. J.; Gaudiano, G.; Koch, T. H. Production of formal-

dehyde and DNA-adriamycin or DNA-daunomycin adducts,

initiated through redox chemistry of dithiothreitol/iron, xanthine

oxidase/NADH/iron, or glutathione/iron. Chem. Res. Toxicol.

1997, 10, 953–961.

(17) Parker, B. S.; Buley, T.; Evison, B. J.; Cutts, S. M.; Neumann,

G. M.; Iskander, M. N.; Phillips, D. R. A molecular understanding

of mitoxantrone-DNA adduct formation: effect of cytosine methy-

lation and flanking sequences. J. Biol. Chem. 2004, 279, 18814–

18823.

(18) Cutts, S. M.; Nudelman, A.; Rephaeli, A.; Phillips, D. R. The

power and potential of doxorubicin-DNA adducts. IUBMB Life

2005, 57, 73–81.

(19) Cutts, S. M.; Nudelman, A.; Pillay, V.; Spencer, D. M.; Levovich,

I.; Rephaeli, A.; Phillips, D. R. Formaldehyde-releasing prodrugs

in combination with adriamycin can overcome cellular drug resis-

tance. Oncol. Res. 2005, 15, 199–213.

(20) Wang, A. H.; Gao, Y. G.; Liaw, Y. C.; Li, Y. K. Formaldehyde

cross-links daunorubicin and DNA efficiently: HPLC and X-ray

diffraction studies. Biochemistry 1991, 30, 3812–3815.

(21) Bailly, C.; Goossens, J. F.; Laine, W.; Anizon, F.; Prudhomme, M.;

Ren, J.; Chaires, J. B. Formaldehyde-induced alkylation of a

2⁰-aminoglucose rebeccamycin derivative to both A.T and G.C

base pairs in DNA. J. Med. Chem. 2000, 43, 4711–4720.

(22) Borchmann, P.; Reiser, M. Pixantrone (Novuspharma). Idrugs

2003, 6, 486–490.

(23) Beggiolin, G.; Crippa, L.; Menta, E.; Manzotti, C.; Cavalletti, E.;

Pezzoni, G.; Torriani, D.; Randisi, E.; Cavagnoli, R.; Sala, F.;

Giuliani, F. C.; Spinelli, S. Bbr 2778, an aza-anthracenedione

endowed with preclinical anticancer activity and lack of delayed

cardiotoxicity. Tumori 2001, 87, 407–416.

(24) Evison, B. J.; Mansour, O. C.; Menta, E.; Phillips, D. R.; Cutts,

S. M. Pixantrone can be activated by formaldehyde to generate a

potent DNA adduct forming agent. Nucleic Acids Res. 2007, 35,

3581–3589.

(25) Evison, B. J.; Chiu, F.; Pezzoni, G.; Phillips, D. R.; Cutts, S. M.

Formaldehyde-activated Pixantrone is a monofunctional DNA

alkylator that binds selectively to CpG and CpA doublets. Mol.

Pharmacol. 2008, 74, 184–194.

(26) Nudelman, A.; Ruse, M.; Aviram, A.; Rabizadeh, E.; Shaklai, M.;

Zimrah, Y.; Rephaeli, A. Novel anticancer prodrugs of butyric

acid. 2. J. Med. Chem. 1992, 35, 687–694.

6,9-Bis(4-aminobutylamino)benzo[g]isoquinoline-5,10-dione

Dihydrochloride 9. A mixture of the difluoro dione 16 (75 mg,

0.31 mmol) and 1,4-diaminobutane (297 μL, 3.1 mmol) in THF

(5 mL) was stirred at 50 ꢀC for 20 h. The solution was cooled and

the precipitate was filtered off and washed with THF. The filtrate

was concentrated in vacuo and the residue dried under a stream of

nitrogen to aid in removal of excess amine. The residue was dis-

solved in MeOH (2 mL), and a solution of 1.25 N HCl in MeOH

(1.86 mmol) was added. A sufficient volume of EtOAc was added,

and a precipitate developed. The solid was filtered off and washed

with EtOAc to afford the dihydrochloride 9 as a dark-blue solid

(90 mg, 67%).²⁹IR ν_max2934, 1568, 1531, 1279, 1174, 838 cm^-1

.

HRESMS found: (M þ H) 382.2246; C₂₁H₂₇N₅O₂requires (M þ

H), 382.2243. ¹H NMR (DMSO) δ 11.20 (1H, t), 11.10 (1H, t),

9.44 (1H, s), 8.95 (1H, d), 8.05 (1H, d), 7.85 (6H, bs), 7.59 (2H, s),

3.60-3.50 (4H, m), 2.90-2.80 (4H, m), 1.75-1.65 (8H, m).

LCMS: T_r= 4.29, 95% purity; m/z (%) = 381 (2.5), [M þ H]^þ

311 (50), 240 (100).

6,9-Bis(5-aminopentylamino)benzo[g]isoquinoline-5,10-dione

Dihydrochloride 10. A mixture of the difluoro dione 16 (75 mg,

0.31 mmol) and 1,5-diaminopentane (360 μL, 3.1 mmol) in THF

(5 mL) was stirred at 50 ꢀC for 20 h. The solution was cooled,

and the precipitate was filtered off and washed with THF. The

filtrate was concentrated in vacuo and the residue dried under a

stream of nitrogen to aid in removal of excess amine. The residue

was dissolved in MeOH (2 mL), and a solution of 1.25 N HCl

in MeOH (1.86 mmol) was added. A sufficient volume of EtOAc

was added, and a precipitate developed. The solid was filtered off

and washed with EtOAc to yield the dihydrochloride 10 as a dark-

bluesolid(70 mg, 51%). IRν_max2934, 1573, 1391, 1171, 830 cm^-1

.

HRESMS found: (M - H) 408.2387; C₂₃H₃₁N₅O₂requires (M -

H), 408.2400. ¹H NMR (DMSO) δ 11.27 (1H, bs), 11.14 (1H, bs),

9.45 (1H, s), 8.95 (1H, d), 8.09 (1H, d), 7.95 (6H, bs), 7.59 (2H, s),

3.55-3.45 (4H, m), 2.85-2.75 (4H, m), 1.71-1.60 (8H, m),

1.50-1.45 (4H, m). LCMS: T_r= solvent front, 95% purity; m/z

(%) = 409 (10), [M þ H]^þ325 (100).

Acknowledgment. This work was supported by grants from

the National Health and Medical Research Council, Australia

(grant number 487333) (S.M.C., D.R.P., and K.G.W.) and

CASS Foundation, Melbourne, Australia (grant number SM/

08/1971) (S.M.C., B.E.).

Supporting Information Available: The full data set for DNA

sequence specificity of all anthracenedione derivatives. This

material is available free of charge via the Internet at http://

pubs.acs.org.

References

(1) Beljanski, V.; Marzilli, L. G.; Doetsch, P. W. DNA damage-

processing pathways involved in the eukaryotic cellular response

to anticancer DNA cross-linking drugs. Mol. Pharmacol. 2004, 65,

1496–1506.

(2) Goldenberg, G.Tannock, I.Drug resistance and experimental

chemotherapy. In The Basic Science of Oncology; McGraw-Hill:

New York 1998; pp 392-419

(3) Feofanov, A.; Sharonov, S.; Kudelina, I.; Fleury, F.; Nabiev, I.

Localization and molecular interactions of mitoxantrone within

living K562 cells as probed by confocal spectral imaging analysis.

Biophys. J. 1997, 73, 3317–3327.

(4) Feofanov, A.; Sharonov, S.; Fleury, F.; Kudelina, I.; Nabiev, I.

Quantitative confocal spectral imaging analysis of mitoxantrone

within living K562 cells: intracellular accumulation and distribu-

tion of monomers, aggregates, naphtoquinoxaline metabolite, and

drug-target complexes. Biophys. J. 1997, 73, 3328–3336.

(5) De Vita, V. T.; Hellman, S.; Rosenberg, S. A. Cancer;Principles

and Practice of Oncology; J.B. Lippincott: Philadelphia, PA, 1993.

(6) Cornbleet, M. A.; Stuart-Harris, R. C.; Smith, I. E.; Coleman,

R. E.; Rubens, R. D.; McDonald, M.; Mouridsen, H. T.; Rainer, H.;

(27) Cullinane, C.; Phillips, D. R. Thermal stability of DNA adducts

induced by cyanomorpholinoadriamycin in vitro. Nucleic Acids

Res. 1993, 21, 1857–1862.

(28) Krapcho, A. P.; Getahun, Z.; Avery, K. J. The Synthesis of 1,4-

Difluoro-5,8-Dihydroxyanthracene-9,10-Dione and Ipso Substitu-

tionsoftheFluoridesbyDiaminesLeadingto1,4-Bis-[(Aminoalkyl)-

Amino]-5,8-Dihydroxyanthracene-9,10-Diones. Synth. Commun.

1990, 20, 2139–2146.

(29) Krapcho, A. P.; Petry, M. E.; Getahun, Z.; Landi, J. J., Jr.;Stallman,

J.;Polsenberg, J. F.; Gallagher, C. E.;Maresch, M. J.;Hacker, M. P.;

6866 Journal of Medicinal Chemistry, 2010, Vol. 53, No. 19

Mansour et al.

Giuliani, F. C.; Beggiolin, G.; Pezzoni, G.; Menta, E; Manzotti, C.;

Oliva, A.; Spinelli, S.; Tognella, S. 6,9-Bis[(aminoalkyl)amino]-

benzo[g]isoquinoline-5,10-diones. A novel class of chromophore-

modified antitumor anthracene-9,10-diones: synthesis and antitu-

mor evaluations. J. Med. Chem. 1994, 37, 828–837.

nucleosome positions at a single molecule level. Nucleic Acids Res.

2005, 33, e176.

(43) Cutts, S. M.; Rephaeli, A.; Nudelman, A.; Hmelnitsky, I.; Phillips,

D. R. Molecular basis for the synergistic interaction of adriamycin

with the formaldehyde-releasing prodrug pivaloyloxymethyl buty-

rate (AN-9). Cancer Res. 2001, 61, 8194–8202.

(44) Batova, A.; Shao, L. E.; Diccianni, M. B.; Yu, A. L.; Tanaka, T.;

Rephaeli, A.; Nudelman, A.; Yu, J. The histone deacetylase

inhibitor AN-9 has selective toxicity to acute leukemia and drug-

resistant primary leukemia and cancer cell lines. Blood 2002, 100,

3319–3324.

(30) De Isabella, P.; Palumbo, M.; Sissi, C.; Capranico, G.; Carenini,

N.; Menta, E.; Oliva, A.; Spinelli, S.; Krapcho, A. P.; Giuliani,

F. C.; Zunino, F. Topoisomerase II DNA cleavage stimulation,

DNA binding activity, cytotoxicity, and physicochemical proper-

ties of 2-aza- and 2-aza-oxide-anthracenedione derivatives. Mol.

Pharmacol. 1995, 48, 30–38.

(31) Askirka, V. F.; Maskevich, A. A.; Stepuro, V. I.; Maskevich, S. A.

Luminescent properties of mitoxantrone in different microsur-

roundings. J. Appl. Spectrosc. 2004, 71, 39–43.

(45) Vigushin, D. M.; Coombes, R. C. Histone deacetylase inhibitors in

cancer treatment. Anti-Cancer Drug 2002, 13, 1–13.

(46) Cress, W. D.; Seto, E. Histone deacetylases, transcriptional con-

trol, and cancer. J. Cell. Physiol. 2000, 184, 1–16.

(32) Jones, G. B.; Palumbo, M. Advances in DNA Sequence-Specific

Agents, Vol. 3; JAI Press Inc: London, 1998.

(33) Hurley, L. H. DNA and its associated processes as targets for

cancer therapy. Nature Rev. Cancer 2002, 2, 188–200.

(47) Marks, P.; Rifkind, R. A.; Richon, V. M.; Breslow, R.; Miller, T.;

Kelly, W. K. Histone deacetylases and cancer: causes and thera-

pies. Nature Rev. Cancer 2001, 1, 194–202.

(34) Phillips, D. R.; Cullinane, C. Adriamycin. Encyclopedia of Mole-

cular Biology: John Wiley & Sons: New York, 1999; pp 68-72.

(35) Panousis, C.; Phillips, D. R. DNA sequence specificity of mitox-

antrone. Nucleic Acids Res. 1994, 22, 1342–1345.

(48) Nolan, L.; Johnson, P. W.; Ganesan, A.; Packham, G.; Crabb, S. J.

Will histone deacetylase inhibitors require combination with other

agents to fulfil their therapeutic potential? Br. J. Cancer 2008, 99,

689–694.

(49) Kim, M. S.; Blake, M.; Baek, J. H.; Kohlhagen, G.; Pommier, Y.;

Carrier, F. Inhibition of histone deacetylase increases cytotoxicity

to anticancer drugs targeting DNA. Cancer Res. 2003, 63, 7291–

7300.

(50) Phillips, D. R.; Cullinane, C. M.; Crothers, D. M. An in vitro

transcription assay for probing drug-DNA interactions at indivi-

dual drug sites. Mol. Biotechnol. 1998, 10, 63–75.

(51) Phillips, D. R.; Cutts, S. M.; Cullinane, C. M.; Crothers, D. M.

High-resolution transcription assay for probing drug-DNA inter-

actions at individual drug sites. Method. Enzymol. 2001, 340,

466–485.

(36) Lawley, P. D. Effects of some chemical mutagens and carcinogens

on nucleic acids. Prog. Nucleic Acid Res. 1966, 5, 89–131.

(37) Hemminki, K.; Ludlum, D. B. Covalent modification of DNA by

antineoplastic agents. J. Natl. Cancer Inst. 1984, 73, 1021–1028.

(38) Hartley, J. A.; Lown, J. W.; Mattes, W. B.; Kohn, K. W. DNA

sequence specificity of antitumor agents. Oncogenes as possible

targets for cancer therapy. Acta Oncol. 1988, 27, 503–510.

(39) Saxonov, S.; Berg, P.; Brutlag, D. L. A genome-wide analysis of

CpG dinucleotides in the human genome distinguishes two distinct

classes of promoters. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 1412–

1417.

(40) Jones, P. A.; Baylin, S. B. The fundamental role of epigenetic events

(52) Chou, T. C.; Talalay, P. Quantitative analysis of dose-effect

relationships: the combined effects of multiple drugs or enzyme

inhibitors. Adv. Enzyme Regul. 1984, 22, 27–55.

(53) Murdock, K. C.; Child, R. G.; Fabio, P. F.; Angier, R. B.; Wallace,

R. E.; Durr, F. E.; Citarella, R. V. Antitumor agents. 1. 1,4-Bis-

[(aminoalkyl)amino]-9,10-anthracenediones. J. Med. Chem. 1979,

22, 1024–1030.

in cancer. Nature Rev. Genet. 2002, 3, 415–428.

(41) Gardiner-Garden, M.; Frommer, M. CpG islands in vertebrate

genomes. J. Mol. Biol. 1987, 196, 261–282.

(42) Fatemi, M.; Pao, M. M.; Jeong, S.; Gal-Yam, E. N.; Egger, G.;

Weisenberger, D. J.; Jones, P. A. Footprinting of mammalian

promoters: use of a CpG DNA methyltransferase revealing

Article Doi

New anthracenedione derivatives with improved biological activity by virtue of stable drug-DNA adduct formation

DOI: 10.1021/jm901894c

Source and publish data:

Authors:

Article abstract of DOI:10.1021/jm901894c

Full text of DOI:10.1021/jm901894c

Products guided by the article

R&D Labs maybe for 131401-54-2

Relevant to this article

Hot Product