Lavendamycin antitumor agents: Structure-based design, synthesis, and NAD(P)H:quinone oxidoreductase 1 (NQO1) model validation with molecular docking and biological studies

Article abstract of DOI:10.1021/jm701066a

A 1H69 crystal structure-based in silico model of the NAD(P)H:quinone oxidoreductase 1 (NQO1) active site has been developed to facilitate NQO1-directed lavendamycin antitumor agent development. Lavendamycin analogues were designed as NQO1 substrates utilizing structure-based design criteria. Computational docking studies were performed using the model to predict NQO1 substrate specificity. Designed N-acyllavendamycin esters and amides were synthesized by Pictet-Spengler condensation. Metabolism and cytotoxicity studies were performed on the analogues with recombinant human NQO1 and human colon adenocarcinoma cells (NQO1-deficient BE and NQO1-rich BE-NQ). Docking and biological data were found to be correlated where analogues 12, 13, 14, 15, and 16 were categorized as good, poor, poor, poor, and good NQO1 substrates, respectively. Our results demonstrated that the ligand design criteria were valid, resulting in the discovery of two good NQO1 substrates. The observed consistency between the docking and biological data suggests that the model possesses practical predictive power.

Full text of DOI:10.1021/jm701066a

3104
J. Med. Chem. 2008, 51, 3104–3115
Lavendamycin Antitumor Agents: Structure-Based Design, Synthesis, and NAD(P)H:Quinone
Oxidoreductase 1 (NQO1) Model Validation with Molecular Docking and Biological Studies
Mary Hassani,† Wen Cai,‡ Katherine H. Koelsch,† David C. Holley,§ Anthony S. Rose,‡ Fatemeh Olang,‡ Jayana P. Lineswala,‡
William G. Holloway,‡ John M. Gerdes,§ Mohammad Behforouz,*,‡ and Howard D. Beall*,†
Center for EnVironmental Health Sciences, Department of Biomedical and Pharmaceutical Sciences, The UniVersity of Montana, Missoula,
Montana 59812, Molecular Computational Core Facility, Center for Structural and Functional Neuroscience, Department of Biomedical and
Pharmaceutical Sciences, The UniVersity of Montana, Missoula, Montana 59812, Department of Chemistry, Ball State UniVersity,
Muncie, Indiana 47306
ReceiVed August 28, 2007
A 1H69 crystal structure-based in silico model of the NAD(P)H:quinone oxidoreductase 1 (NQO1) active
site has been developed to facilitate NQO1-directed lavendamycin antitumor agent development. Laven-
damycin analogues were designed as NQO1 substrates utilizing structure-based design criteria. Computational
docking studies were performed using the model to predict NQO1 substrate specificity. Designed
N-acyllavendamycin esters and amides were synthesized by Pictet-Spengler condensation. Metabolism and
cytotoxicity studies were performed on the analogues with recombinant human NQO1 and human colon
adenocarcinoma cells (NQO1-deficient BE and NQO1-rich BE-NQ). Docking and biological data were found
to be correlated where analogues 12, 13, 14, 15, and 16 were categorized as good, poor, poor, poor, and
good NQO1 substrates, respectively. Our results demonstrated that the ligand design criteria were valid,
resulting in the discovery of two good NQO1 substrates. The observed consistency between the docking
and biological data suggests that the model possesses practical predictive power.
Chart 1
Introduction
NQO1a is a widely distributed homodimeric flavoenzyme
composed of two closely associated monomers of 273 residues,
each containing one molecule of FAD cofactor that is required
for NQO1 catalytic activity.1–5 NQO1 catalyzes an NAD(P)H-
dependent, two-electron reduction of quinones.6,7 A number of
quinone-based chemotherapeutic agents including quinoline-
quinones, such as our previously studied lavendamycin ana-
logues8 and streptonigrin,9–11 mitomycin C,12,13 ꢀ-lapachone
(a 1,2-naphthoquinone analogue),14,15 and various indolequi-
nones16–18 and aziridinylbenzoquinones,19–25 can be bioactivated
by NQO1. NQO1 overexpression in many human solid tumors
has been established,26–30 and NQO1 is considered a key target
for the design of quinone-based chemotherapeutic agents that
can be bioactivated by the enzyme to achieve selective toxicity
toward NQO1-rich tumors.
Lavendamycin (1) (Chart 1), a naturally occurring 7-amino-
quinoline-5,8-dione antitumor antibiotic, was first isolated from
the fermentation broth of Streptomyces laVendulae by Balitz et
al.,31 and its structure was determined by Doyle et al.32 The
latter study determined that 1 is a pentacyclic structure with
two moieties including quinoline-5,8-dione and indolopyridine
(ꢀ-carboline) (Chart 1).32 Compound 1 has been the focus of
several synthetic studies to elucidate the structural features that
are required for its cytotoxic activity and to develop improved
analogues with potent antitumor properties and lower animal
toxicity. Initial SAR studies have demonstrated that the essential
moiety for the cytotoxic activity of 1 is the 7-aminoquinoline-
5,8-dione moiety.33 It has also been reported that lavendamycin
analogues possess low animal toxicity.34,35
Bioreductive enzyme-directed antitumor agent development
depends on the identification of chemotherapeutic agents with
high substrate specificity for target reductases.36 Structure-based
ligand design is an efficient approach in modern drug develop-
ment for targets with resolved three-dimensional structures.37–39
In structure-based design, the information obtained from the
interactions and composite structure of the target protein–ligand
is utilized to design improved ligands with high binding affinity
for target proteins implicated in diseases.37–40 Deeper insight
of ligand-protein interactions within the active site of the target
greatly contributes to more refined SAR studies, which in turn
are the crucial components in the context of structure-based
ligand design.41 Computer-aided docking techniques serve as
* Corresponding authors for chemistry (M.B.) and biology (H.D.B.). M.
Behforouz: Phone, 765-285-8070; Fax, 765-285-6505; E-mail, mbehforo@
bsu.edu. H. D. Beall: Phone, 406-243-5112; Fax, 406-243-5228; E-mail,
howard.beall@umontana.edu.
†
Center for Environmental Health Sciences, The University of Montana.
Molecular Computational Core Facility, The University of Montana.
Department of Chemistry, Ball State University.
§
‡
a Abbreviations: FAD, flavin adenine dinucleotide; NQO1, NAD(P)H:
quinone oxidoreductase 1; RMSD, root-mean-square deviation; SAR,
structure–activity relationship.
10.1021/jm701066a CCC: $40.75
2008 American Chemical Society
Published on Web 05/06/2008

LaVendamycin Antitumor Agents
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 11 3105
Table 1. Structures of Lavendamycin Analogues, Yields, and Reaction Conditions
no.
R1
R2
R3
% yield
solvent
h (°C)a
11
12
13
14
15
16
28
29
30
31
CH3CONH
CH3CONH
CH3CONH
n-C3H7CONH
2-furylCONH
NH2
NH2
NH2
CH3CONH
NH2
CO2(CH2)2OH
CONH(CH2)3CH3
CONHCH(CH3)C2H5
CO2(CH2)2CH(CH3)2
CO2CH3
CO2(CH2)2OH
CO2C8H17-n
CH2OH
H
H
H
H
H
H
H
H
H
H
87–94
43
DMF/anisole
anisole
anisole
xylene
xylene
hydrolysis of 11
see ref 35
see ref 8
see ref 8
see ref 35
2 (reflux)35
20 (reflux)
15 (reflux)
17 (reflux)34
17 (reflux)
35
70
53.5
30
CO2C5H11-i
CONH2
a For the synthesis of 11–15, the reaction mixtures were slowly heated to reflux over a period of approximately 3 h. Compound 16 was obtained by the
acid hydrolysis of 11 and 28–31 were synthesized according to our previously reported methods.
time- and cost-efficient tools for structure-based design making
for more efficient syntheses and biological testing of new
analogues.38,42 These techniques greatly contribute to this field
because they facilitate a greater understanding of the binding
events and the molecular basis of ligand-protein interactions
as well as prediction of binding orientations and affinities of
candidate compounds.37–39,43 The information is then utilized
to design more efficacious new analogues.38
docking data acquired using our recently developed in silico
model of the NQO1 active site were consistent with the
biological data. We also sought to investigate the predictive
power of the model to correctly distinguish between good and
poor NQO1 substrates. This is the first study to perform
structure-based lavendamycin analogue design using molecular
docking and SAR data that serve to validate the developed
model with computational and biological studies.
Crystal structure complexes with bound ligands and cofactors
can be used as composite reference structures for docking studies
and the creation of active site models.37,38,40 The crystal
structures of the apo human NQO11,4 and human NQO1 in
complex with several agents have been reported.4,44–46 The
active site of the enzyme is a hydrophobic and plastic pocket
capable of forming van der Waals and hydrogen bond interac-
tions with quinone compounds.4,8,45 The substrate binding
pocket sequentially binds NAD(P)H and the quinone substrate
during an obligatory two-electron reduction process (ping-pong
mechanism).4,5,45 During this process, a hydride ion from
NAD(P)H is transferred to the N5 of the FAD followed by the
release of NAD(P)+.5,47,48 The hydride donation from the FAD
N5 to the hydride-acceptor quinone can then be accomplished
followed by hydroquinone release.5,45,47,48 NQO1 crystal structure-
based molecular modeling studies have been recommended for
structure-based design efforts for the optimization of quinone
antitumor agents.4,45,49
Complete synthesis of the methyl ester of lavendamycin (2)
(Chart 1) was first reported by Kende et al. in 1984,50 and the
second synthesis of this compound was described by Boger et
al. in 1985.51 However, our 199352 and 199653 reported
syntheses of 2 are much shorter and more efficient compared
to the methods reported by the first two groups.50,51 Because
of the practicality of our synthetic methods, specifically that
reported in 1996,53 we have been able to synthesize a large
number of substituted lavendamycins and study their various
biological activities.8,34,35,54 In this report, we describe the
synthesis of rationally designed analogues 12, 13, 15, and 16
(Table 1).
Results and Discussion
Structure-Based Design. Lavendamycin analogues 12, 13,
15, and 16 (Table 1) were designed utilizing the criteria obtained
from our previous SAR and modeling study8 and other recent
NQO1-related literature. The structure-based design criteria for
substituents at key ligand positions (Chart 1) are discussed in
the following sections.
Quinolinedione-7 Position (R1). (1) Substituents should be
small to medium in size, preferably NH2 or NHCOCH3 groups,8
that do not produce steric interactions with the key residues of
the active site including the internal wall (Trp-105/Phe-106)
(applied to 12, 13, and 16).40,55,56 The substituents should also
be capable of hydrogen bond formation with the FAD cofactor
and/or the key amino acid residues of the active site including
Tyr-126, -128, and His-161.8 Faig et al. determined that
positions of 24 (RH1; Chart 2), 2,5-diaziridinyl-3-(hydroxy-
methyl)-6-methyl-1,4-benzoquinone,25 that point to the inner part
of the NQO1 active site could accommodate only small
substituents.45 (2) Substituents that can intercalate between and/
or form van der Waals interactions with the Trp-105/Phe-106
minipocket (applied to 15).40,55 A previous study demonstrated
that an aziridinyl group at the C5 position of 25 (EO9; Chart
2), 3-hydroxymethyl-5-aziridinyl-1-methyl-2-(1H-indole-4,7-
dione)-propenol,57 can form favorable van der Waals interactions
with Trp-105.55 Additionally, van der Waals interactions
between the C5 aziridine ring of 26 (CB1954; Chart 2),
5-aziridinyl-2,4-dinitrobenzamide,1 and Trp-105 are important
to the binding of this prodrug.1 Unsubstituted pyrrolo and pyrido
rings in dipyrroloimidazobenzimidazole and dipyridoimida-
zobenzimidazole compounds are able to sandwich between Trp-
105/Phe-106 residues and form van der Waals interactions to
increase NQO1 substrate specificity.40 (3) No large substituents
are allowed due to increased steric hindrance with the internal
The present study was conducted to design more optimal
lavendamycin substrates for NQO1 with increased selective
cytotoxicity toward NQO1-rich cells using computational active
site docking studies and correlated SAR data. The objective was
to determine whether our design criteria were valid and whether

3106 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 11
Hassani et al.
Chart 2
to examine the predictive power of the model, to relate the
model and docking studies with the metabolism and cytotoxicity
results, and to gain deeper insight into the binding events and
themolecularbasisoflavendamycinanalogue-NQO1interactions.
Synthetic Chemistry. Table 1 displays the structures of
lavendamycin analogues 11-16 and 28-31. Analogues 12, 13,
15, and 16 were designed and synthesized. These lavendamycin
analogues as well as previously synthesized compound 14 were
the subject of our metabolism, cytotoxicity, and modeling studies
in this report. As shown in Scheme 1, Pictet-Spengler
condensation of 7-N-acylamino-2-formylquinoline-5,8-diones
3–5 with trytophan esters 6, 9, and 10 and amides 7 and 8
yielded the corresponding lavendamycin derivatives 11–15. The
synthesis of compounds 11, 14, and 28-31 has already been
reported by us.8,34,35
Demethllavendamycin ꢀ-hydroxyethyl ester (16) was prepared
by the acid hydrolysis of acetyl derivative 11 in 30% yield.
Aldehydes 3 and 4 necessary for the synthesis of lavenda-
mycins 11–14 were prepared through our reported procedures.35,53
Aldehyde 5 required for the synthesis of lavendamycin 15 was
obtained via reactions shown in Scheme 2.
Tryptophan esters 6, 9, and 10 were prepared according to
our previously reported method.8,35 Tryptophan amides 7 and
8 were prepared according to Scheme 3, via similar methods to
those previously reported by us.8 Benzyloxycarbonyl succin-
imide ester 21 was obtained8 and then condensed with n-
butylamine or s-butylamine to produce 22 and 23 respectively
in yields of 68 and 90%. Deprotection of 22 and 23 with dry
ammonium formate in the presence of 10% Pd/C yielded
tryptophan amides 7 and 8 in 77% and 75% yields, respectively.
Docking Studies. Our laboratory recently developed an 1H69
crystal structure-based in silico model of the NQO1 active site.8
To further determine the predictive power of the model and
correlation of the docking data with biological measures, we
performed computational and comparative docking studies on
the structure-based designed lavendamycin analogues 12, 13,
15, and 16 and the previously synthesized compound 14. The
molecular modeling studies were performed using SYBYL 7.0
software suite60 (Tripos, Inc., St. Louis, MO). Flexible docking
was performed using the FlexX software module within the
SYBYL environment, where FlexX is capable of determining
30 possible conformations (poses) for each docked ligand.61,62
The docked conformations of ligands 12-16 were evaluated
and ranked using the CSCORE module, with ranking scores
between 0 and 5, where 5 was the best fit to the model. Table
2 displays the number of conformations of the ligands in each
score group of CSCORE function.
Ligands 12 and 16 possessed higher number of poses with
optimal CSCORE values compared to 13, 14, and 15 (Table
2). Visual screening of the binding orientations of the poses
and geometric postdocking analyses were performed. The
analyses included distance measurements and pose geometries
that determined: (1) hydrogen-bonding and van der Waals
interactions of ligand poses with FAD and the key residues of
the NQO1 active site including Trp-105, Phe-106, Tyr-126, -128,
Gly-149, -150, His-161, and Phe-232, and (2) hydride ion
transfer from the N5 of the FAD isoalloxazine ring to the ligands
at carbonyl oxygens (O5 or O8), ring carbon, or substituent
atoms. Residue numbers in this paper are those used in the
Protein Data Bank coordinates, PDB ID code: 1H69.45 Only
poses of the ligands with CSCORE g 4 were considered for
further detailed postdocking analyses.
wall that can result in unfavorable positioning of the quinoline-
5,8-dione moiety of lavendamycin analogues for hydride ion
reception from FAD and quinone reduction (applied to 12, 13,
15, and 16).8,40 A study by Suleman et al. demonstrated that
dipyrroloimidazobenzimidazole compounds with both pyrrolo
rings bearing bulky substituents were excluded from the active
site due to steric interactions.40
Quinolinedione-6 Position. Absence of substituents in this
position is highly preferred. The simultaneous placement of
substituents at both R1 and the quinolinedione-6 position
increases steric interactions of the lavendamycin ligands with
the internal wall of the active site.8 It has been reported that
1,4-naphthoquinones with small substituents such as an aziridine
ring or CH3 at C2 and no substituents at C3 (C2 and C3 positions
point to the inside of the active side) are good substrates for
NQO1.58 Increased bulkiness of substituents at the C5 position
of indolequinones dramatically reduced rates of reduction by
NQO1.45,55 Previous SAR studies have demonstrated that
7-aminoquinoline-5,8-dione is an essential moiety in determining
thecytotoxicandantitumoractivityofquinolinedioneantibiotics.33,59
Therefore, substituent placement at the R1 position over
quinolinedione-6-position is highly desirable (applied to 12, 13,
15, and 16).
Indolopyridine-2′ Position (R2). (1) Substituents should be
capable of hydrogen bond formation with the FAD cofactor and/
or the key residues of the active site including Gly-149 and
Gly-150 (applied to 15 and 16).8 The 3-hydroxymethyl group
of 27 (ARH019; Chart 2), 3-(hydroxymethyl)-5-(2-methylaziri-
din-1-yl)-1-methyl-2-phenylindole-4,7-dione,18 that points to-
ward the outside of the active site forms a hydrogen bond with
the Tyr-128 OH.45 (2) Substituents (including aliphatic chains)
should be capable of formation of van der Waals interactions
with residues of the NQO1 active site (applied to 12 and 13).
Compound 28, demethyllavendamycin n-octyl ester,35 possesses
a large n-octyl ester substituent at the R2 position and is a good
NQO1 substrate with high selective toxicity toward NQO1-rich
cancer cells.8
Utilizing the above criteria, we designed compounds 12, 13,
15, and 16 with small or medium size substituents at R1, no
substituent at quinolinedione-6-position, and small to large
substituents at R2. The substituents at R1 and R2 were expected
to form hydrogen bond and/or van der Waals interactions with
FAD and/or the amino acid residues of the NQO1 active site
such as Trp-105, Phe-106, Tyr-126, -128, -Gly-149, -150, and
His-161. Docking studies with compounds 12, 13, 15, and 16
using an X-ray derived, in silico model were performed to
predict the substrate specificity of the compounds for NQO1,
Ligands 12 and 13 were designed to possess highly similar
chemical structures with only a minor difference in the sub-

LaVendamycin Antitumor Agents
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 11 3107
Scheme 1
Scheme 2a
One crucial determining factor of NQO1 substrate binding
strength in the NQO1 active site is the ability to form hydrogen-
bonding and/or van der Waals interactions with FAD and/or
amino acid residues of the active site.1,40,45,47 Good substrates
for NQO1 such as 24, 25, and 26 are capable of hydrogen-
bonding interactions with the key amino acid residues of the
NQO1 active site.1,45 Duroquinone, a tetramethyl analogue of
benzoquinone, binds to the NQO1 active site through interac-
tions with FAD and several hydrophilic and hydrophobic
residues.4 For ligand 12, poses 1, 2, 4, 5, and 6 formed the
highest number of hydrogen bonds and van der Waals interac-
tions in the active site of the enzyme. For pose 1 of 12, the
Tyr-128 OH formed hydrogen bonds with the carbonyl oxygen
O8, the NH of the indole ring of the indolopyridine moiety,
and the carbonyl oxygen of the quinolinedione-7 position
substituent (Figure 1b). The carbonyl oxygen O5 formed a
hydrogen bond with the NH of His-161 (Figure 1b). The NH
of the quinolinedione-7 position substituent also formed hy-
drogen bonds with the N1, N5, and N10 of FAD (Figure 1b).
The CONH(CH2)3CH3 group of the indolopyridine moiety
further stabilized the binding by forming van der Waals
interactions with Phe-232 (Figure 1b). Poses 2, 4, 5, and 6
displayed the same interactions as pose 1 (only pose 1 is shown)
(Figure 1b and Table 3). The docked model determined a high
number of poses of 12 with optimum CSCORE (g4) that are
capable of hydrogen bond and van der Waals formation in the
NQO1 active site, suggesting that this compound is a good
substrate for NQO1.
a Reagents and conditions: (a) Pd-C 5%, H2 (41 psi), HCl-H2O, 24 h,
rt, 82%; (b) 2-furylCOCl, NaOAc, Na2SO3, 2 h, 0 °C, 75%; (c) K2Cr2O7,
HOAc, THF, H2O 20 h, rt, 64%; (d) SeO2, dioxane, 26 h, reflux, 93%.
Scheme 3a
a Reagents and conditions: (a) s-butylamine, Et3N, EtOH, CHCl3, 2 h,
rt; (b) dry MeOH, dried ammonium formate, 10% Pd/C, Ar, 30 min, rt.
Table 2. Number of Poses of Ligands 12, 13, 14, 15, and 16 in Each
Score Group of CSCORE Function
However, of the thirty poses of 13, only poses 1, 2, and 4
merited further considerations (CSCORE ) 4), and the remain-
der of the poses possessed CSCORE e 3 (Table 2). None of
the poses of 13 had a CSCORE ) 5 (Table 2). Poses 1, 2, and
4 of 13 entered the active site by the 5,8-dione moiety similar
to the poses of 12 (Figure 2a and Table 4). Although these poses
formed hydrogen bonds with FAD and the residues of the NQO1
active site, the number of total hydrogen bonds was lower than
that for the poses of 12 (hydrogen bonds not shown). Further-
more, poses 1, 2, and 4 of 13 were found not to form van der
Waals interactions with Phe-232, compared to the poses of 12,
providing a rationale for the lower binding affinity of 13
compared to 12 in the NQO1 active site (Figure 2a,b). The in
silico docking studies ranked 13 as a poor substrate for NQO1.
Overall, a higher number of more favorable poses of ligand 12
were found compared to 13 that formed key hydrogen bonding
interactions (van der Waals interactions only in poses of 12)
and had favorable binding orientations for hydride ion reception
and quinone reduction. The in silico model was found to be
capable of distinguishing the differences between the respective
good and poor NQO1 substrate qualities of 12 and 13.
number of poses
CSCORE
12
13
14
15
16
0
1
2
3
4
5
7
4
5
5
4
5
6
9
1
11
3
0
5
10
6
4
5
9
9
7
3
1
1
5
3
2
5
10
5
0
stituents at R2 (R2 substituents of 12 and 13 are structural
isomers) in order to examine the role that R2 substituents play
in the affinity of lavendamycin substrates for NQO1. Ligand
12 possessed nine poses of CSCORE g 4 compared to ligand
13, with only three poses of CSCORE g 4 (Table 2). Poses 1,
2, 4, 5, and 6 (CSCORE ) 5) of 12 fell into one cluster in
which the RMSD of the poses equaled zero for atoms of the
quinolinedione and indolopyridine moieties, and the difference
was in the binding orientation of the CONH(CH2)3CH3 group
in the NQO1 active site (Figure 1a). All of the poses of 12
entered the active site by the 5,8-dione moiety, where the
carbonyl oxygen O8 compared to O5 was positioned closer to
Tyr-126, -128, and FAD (Figure 1a and Table 3).45,55 The
binding orientations of the poses of 12 in the cluster are
considered as preferred binding orientations of 12 because these
are the binding orientations with maximum CSCORE ) 5
(Figure 1a).
Compound 15 was designed to investigate whether an
aromatic amide group at R1 is capable of intercalating between
Trp-105 and Phe-106 residues and forming van der Waals
interactions to increase NQO1 substrate specificity. Docking
studies of 15 were performed to observe how the model would

3108 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 11
Hassani et al.
Figure 1. Molecular models of the poses of ligand 12 docked into the NQO1 active site. (a) View of the superposition of the docked poses 1, 2,
4, 5, and 6 of 12 (magenta, cyan, yellow, salmon, and blue, respectively) (CSCORE ) 5) in the NQO1 active site. (b) Molecular model of the pose
1 of 12 (magenta) (CSCORE ) 5) docked into the NQO1 active site. In (b) the Tyr-128 OH formed hydrogen bonds with the carbonyl oxygen O8,
the NH of the indole ring of the indolopyridine moiety, and the carbonyl oxygen of the quinolinedione-7 position substituent. The carbonyl oxygen
O5 formed a hydrogen bond with the NH of His 161. The NH of the quinolinedione-7 position substituent also formed hydrogen bonds with the
N1, N5, and N10 of FAD. The CONH(CH2)3CH3 group of the indolopyridine moiety further stabilized the binding by forming van der Waals
interactions with Phe-232. In (a) and (b) residues of the active site (green), FAD (blue), and 12 are represented as stick models. In (b) the rest of
the structure is represented as a secondary structure cartoon. The atoms are colored: red, oxygen atoms; blue, nitrogen atoms; orange, phosphorus
atoms; and white, hydrogen atoms. Hydrogen bonds are represented as black dashed lines.
Table 3. Geometric Post-Docking Analysis and Measurements of Five Poses (CSCORE ) 5) and Four Poses (CSCORE ) 4) of Ligand 12 in the
NQO1 Active Site
Pa Cb O5-Tyr-126 (Å) O5-Tyr-128 (Å) O8-Tyr-126 (Å) O8-Tyr-128 (Å) N5-O5 (Å) N5-O8 (Å) N5-C6 (Å) N5-C7 (Å)
1
2
4
5
6
3
7
14
25
5
5
5
5
5
4
4
4
4
9.174
9.174
9.174
9.174
9.174
9.417
9.260
9.180
9.167
6.143
6.143
6.143
6.143
6.143
6.196
6.085
6.242
6.198
4.862
4.862
4.862
4.862
4.862
4.804
4.774
4.878
4.882
2.077
2.077
2.077
2.077
2.077
1.802
1.899
2.123
2.081
7.786
7.786
7.786
7.786
7.786
7.953
7.772
7.618
7.692
5.006
5.006
5.006
5.006
5.006
5.281
5.187
4.960
5.016
5.515
5.515
5.515
5.515
5.515
5.714
5.532
5.375
5.427
4.622
4.622
4.622
4.622
4.622
4.863
4.697
4.515
4.560
a P ) Pose. b C ) CSCORE.
Figure 2. Molecular models of the poses of ligand 13 docked into the NQO1 active site. (a) View of the superposition of the docked poses 1, 2,
and 4 of 13 (magenta, yellow, and salmon, respectively) (CSCORE ) 4) in the NQO1 active site. (b) Molecular model of the pose 1 of 13
(magenta) (CSCORE ) 4) docked into the NQO1 active site. In (a) and (b) residues of the active site (green), FAD (blue), and 13 are represented
as stick models. In (b) the rest of the structure is represented as a secondary structure cartoon. The atoms are colored: red, oxygen atoms; blue,
nitrogen atoms; orange, phosphorus atoms; and white, hydrogen atoms.
rank this compound as an NQO1 substrate. Only poses 7 and
30 of ligand 15 possessed CSCORE g 4 (Table 2). These poses
entered the active site by the 5,8-dione moiety similar to the
poses of 12 (Supporting Information Figure 1a and Table 1).
The other 28 poses with CSCORE e 3 were found to be
excluded from the NQO1 active site including pose 2 (CSCORE
) 3) (only pose 2 is shown; Supporting Information Figure 1b).
The NHCO-2-furyl group of poses 7 and 30 did not intercalate
between the Trp-105 and Phe-106 residues, suggesting the lack
of Van der Waals interactions with the Trp-105/Phe-106

LaVendamycin Antitumor Agents
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 11 3109
Table 4. Geometric Post-Docking Analysis and Measurements of Three Poses of Ligand 13 (CSCORE ) 4) in the NQO1 Active Site
Pa Cb O5-Tyr-126 (Å) O5-Tyr-128 (Å) O8-Tyr-126 (Å) O8-Ty-r128 (Å) N5-O5 (Å) N5-O8 (Å) N5-C6 (Å) N5-C7 (Å)
1
2
4
4
4
4
9.402
9.402
9.268
6.193
6.193
6.084
4.802
4.802
4.692
1.828
1.828
1.770
7.933
7.933
7.739
5.246
5.246
5.240
5.689
5.689
5.504
4.831
4.831
4.691
a P ) Pose. b C ) CSCORE.
Table 5. Metabolism of Lavendamycin Analogues by Recombinant
Human NQO1 Monitored by Spectrophotometric Cytochrome c Assay
minipocket and presence of possible steric interactions with the
residues Trp-105 and Phe-106 (Supporting Information Figure 1a).
Previous studies have indicated that the Trp-105/Phe-106
minipocket can play a crucial role in impacting the substrate
specificity of NQO1 ligands.40,55,63 Our model ranked ligand
15 as a poor substrate for NQO1.
Compound 16 was designed after the good NQO1 substrate,
compound 29,8 decarboxy-2′-(hydroxymethyl)-demethyllaven-
damycin,8 to create another good substrate. Fifteen poses of
ligand 16 were found to possess CSCORE values g4 (Table
2). Poses 1, 2, 4, and 7 (CSCORE ) 5) of 16 entered the active
site by the 5,8-dione moiety similar to the poses of 12
(Supporting Information Figure 2a and Table 2). Among five
poses of 16 with CSCORE ) 5, pose 8 entered the active site
with an orientation opposite to poses 1, 2, 4, and 7 (Supporting
Information Figure 2b and Table 2). Different binding orienta-
tions of quinone substrates that can facilitate hydride ion
reception from the FAD N5 to the substrates can be tolerated
in the active site.45,47 Among the poses of ligand 16, poses 1,
2, 4, and 7 formed the highest number of hydrogen bonds in
the active site of the enzyme (for a detailed description of
hydrogen bonds that were formed by ligand 16, see Supporting
Information Figure 2c legend). Pose 8, among poses of 16 with
CSCORE ) 5, formed the lowest number of hydrogen bonds,
suggesting that the binding orientations of poses 1, 2, 4, and 7
are the preferred binding orientations for ligand 16 relative to
pose 8 (Supporting Information Figure 2a,b). Taken together,
the in silico model recognized 16 as a good substrate for NQO1.
Compound 14 was selected from previously synthesized
lavendamycin analogues, and according to our previous SAR
studies,8 this compound was thought to be a poor substrate for
NQO1 due to the presence of the large NHCOC3H7-n group at
the R1 position that could create steric interactions inside the
NQO1 active site. Therefore, compound 14 was selected for
docking studies to observe how the model would rank it as an
NQO1 substrate. None of the conformations of 14 had a
CSCORE ) 5 (Table 2), and all of the poses of 14, including
poses 2, 5, 6, 21, and 25 (CSCORE ) 4), fell into one cluster
that yielded an RMSD equal to 0 Å for atoms of the
quinolinedione and indolopyridine moieties (Supporting Infor-
mation Figure 3a). However, the binding orientations of the NH-
COC3H7-n and CO2C2H4CH(CH3)2 groups of 14 were found
to be less optimal (Supporting Information Figure 3a), and all
30 poses of 14 were excluded from the NQO1 active site (only
pose 2 is shown; Supporting Information Figure 3b). The site
exclusion is most likely related to the steric hindrance produced
by the large NHCOC3H7-n substituent against the Trp-105/Phe-
106 wall that could be further enhanced by the contributing steric
effect produced by the large CO2C2H4CH(CH3)2 group at R2.
Ligand 14 was ranked as a poor substrate for NQO1 by the in
silico model.
metabolism by NQO1
(µmol/min/mg)
(cytochrome c reduction)
no.
R1
R2
12 CH3CONH
13 CH3CONH
14 n-C3H7CONH CO2C2H4CH(CH3)2
15 2-furyl-CONH CO2CH3
CONH(CH2)3CH3
CONHCH(CH3)C2H5
143 ( 11
4.9 ( 2.9
3.4 ( 1.7
7.0 ( 1.5
60 ( 8
16 NH2
CO2(CH2)2OH
Biological Studies. Metabolism Studies. Metabolism of the
lavendamycin analogues by recombinant human NQO1 was
examined. Reduction rates by NQO1 were measured using a
spectrophotometric assay that employs cytochrome c as the
terminal electron acceptor55 and gives initial rates of lavenda-
mycin analogue reduction (Table 5). The initial reduction rates
(µmol cytochrome c reduced/min/mg NQO1) were calculated
from the linear portion (0–30 s) of the reaction graphs.
Compound 12 with the NHCOCH3 and CONH(CH2)3CH3
groups at R1 and R2 positions, respectively, displayed the highest
metabolism rate by NQO1 among the compounds (Table 5).
The NHCOCH3 group, a medium size substituent, did not
produce steric hindrance with the internal wall of the NQO1
active site, resulting in favorable positioning of 12 for hydride
ion reception from FAD and quinone reduction. Our docking
studies determined that this group was also capable of hydrogen
bonding with FAD and the key residues of the active site. These
studies also demonstrated that the CONH(CH2)3CH3 group at
the R2 position (pointing toward the outside of the active site)
was capable of forming van der Waals interactions with the
Phe-232 residue of the NQO1 active site. This could be a
contributing factor to the substrate specificity of this compound.
Compound 28 with the straight chain n-octyl ester substituent
at the R2 position (CO2C8H17-n) was determined to be a good
substrate for the enzyme similar to 12.8 The docking studies
also indicated ligand 12 as a good substrate for NQO1 consistent
with the metabolism data.
Although 13 possessed a similar chemical structure to 12 (the
best substrate), it was a poor substrate for NQO1 (Table 5) in
accordance with the docking data. Compounds 12 and 13 are
structural isomers. The CONH(CH2)3CH3 group at the R2
position of 12 is a large, nonbulky, and straight chain aliphatic
group, whereas CONHCH(CH3)C2H5 at the R2 position of 13
is a large, bulky, and branched structural isomer of the former.
According to our docking studies, the branched configuration
at R2 of 13 was not capable of forming van der Waals
interactions with the Phe-232 residue of the NQO1 active site
because the sec-butyl group is no longer in proximity with the
residue due to the inverted orientation of this substrate.
The molecular docking studies demonstrated that ligands 12
and 16 possessed an increased number of possible poses with
optimal CSCORE values and favorable binding orientations to
promote hydrogen bonding and van der Waals interactions,
hydride ion reception, and quinone reduction compared to
ligands 13, 14, and 15.

3110 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 11
Hassani et al.
Table 6. Cytotoxicity of Lavendamycin Analogues towards BE (NQO1-Deficient) and BE-NQ (NQO1-Rich) Human Colon Adenocarcinoma Cell Lines
cytotoxicity IC50 (µM)
selectivity ratio
no.
R1
R2
BE-NQ
BE
[IC50 (BE)/IC50 (BE-NQ)]
12
13
14
15
16
CH3CONH
CH3CONH
n-C3H7CONH
2-furyl-CONH
NH2
CONH(CH2)3CH3
CONHCH(CH3)C2H5
CO2C2H4CH(CH3)2
CO2CH3
1.7 ( 0.1
>50
50 ( 2
>50
30
>50
>50
47 ( 3
0.5 ( 0.1
49 ( 7
3.3 ( 0.2
1
7
CO2(CH2)2OH
Compound 15 exhibited a low metabolism rate by NQO1
(Table 5) due to the possible failure of the NHCO-2-furyl group
at the R1 position to intercalate between and form van der Waals
interactions with the internal wall residues. Docking studies
demonstrated that the lack of R1-substituent intercalation
capability led to the unfavorable positioning of 15 and loss of
required hydrogen bonding and/or van der Waals interactions
with the active site residues. This compound was also ranked
as a poor substrate for NQO1 by the docking studies in
accordance with the biological data.
determined by the colorimetric MTT assay. We previously
demonstrated an excellent positive linear correlation between
the IC50 values (the chemosensitivity results) of the clonogenic
and MTT assays for lavendamycin analogues.8 We utilized the
BE human colon adenocarcinoma cells stably transfected with
human NQO1 cDNA.25 The BE cells had no measurable NQO1
activity, whereas activity in the transfected cells (BE-NQ) was
337 nmol/min/mg total cell protein using dichlorophenolin-
dophenol as the standard electron acceptor. In this study, the
cytotoxicity of the lavendamycin analogues (Table 6) has been
compared in these cell lines.
Designed analogues 12 and 16 that were the best substrates
for NQO1 (Table 5) were also more toxic to the NQO1-rich
BE-NQ cell line than the NQO1-deficient BE cell line (Table
6). Compounds 12 and 16 had the greatest differential toxicity
with a selectivity ratio [IC50 (BE)/IC50 (BE-NQ)] of 30 and 7,
respectively (Table 6). Our previous study also determined that
good lavendamycin substrates for NQO1 were selectively toxic
toward BE-NQ versus BE cells.8 Compounds 28, 29, 31,35 and
252 exhibited high selective toxicity toward BE-NQ cells
(selectivity ratios ) 10, 11, 9, and 9, respectively).8 Compound
31 was also reported to highly reduce the colony outgrowth of
A549 human lung carcinoma cells,54 and it displayed promising
cytotoxic and antitumor activities in the National Cancer
Institute’s 60-cell line panel and in vivo hollow fiber tumori-
genesis assay.54
Lavendamycin analogues, 13, 14, and 15, that were poor
substrates for NQO1 (Table 5), demonstrated no selective
toxicity toward BE-NQ cells or had no measurable cytotoxicity
(IC50 > 50 µm) (Table 6). Overall, our results suggested that
the best lavendamycin substrates for NQO1 were also the most
selectively toxic to the high NQO1 BE-NQ cell line compared
to NQO1-deficient BE cells, consistent with our previous study.8
It is important to note, however, that factors other than reduction
by NQO1 are most likely involved in the toxicity of the
lavendamycins. We are currently studying the cellular effects
of the lavendamycin analogues in NQO1-rich and NQO1-
deficient cancer cells in an effort to elucidate mechanisms of
toxicity.
Compound 16 with a small NH2 group at R1 and
CO2(CH2)2OH at R2 displayed a good reduction rate by the
enzyme (Table 5). According to the docking studies, favorable
positioning of 16 in the active site was facilitated by lack of
steric interactions of the substituents with the residues of the
active site and by hydrogen bond formation with FAD and the
residues of the NQO1 active site. Our docking studies also
indicated high binding affinities for 16. A previous study
determined that good indolequinone substrates for NQO1
including 25 possessed a hydroxymethyl group at the analogous
C3 position.55 Furthermore, compound 24, which is a good
benzoquinone substrate for NQO1, has a CH2OH group at the
C3 position.2,25 We also determined that compound 29 that
possessed an NH2 group at R1 and CH2OH at R2 was a good
substrate for NQO1.8
Previously synthesized compound 14 with NHCOC3H7-n and
isoamyl ester groups at the R1 and R2 positions, respectively,
exhibited the lowest metabolism rate by NQO1 and ranked as
the poorest substrate (Table 5). The recently studied lavenda-
mycin analogue 30, 7-N-acetyldemethyllavendamycin isoamyl
ester,8 with an acetamide group at the R1 position and isoamyl
ester group at the R2 position displayed a reduction rate 12-
fold higher than 14. The decreased reduction rate of 14
compared to 30 can be explained by apparent steric hindrance
between the quinolinedione moiety of 14 and the NQO1 active
site caused by the large NHCOC3H7-n group at R1 compared
to NHCOCH3 in 30. This steric interaction could result in
exclusion of 14 from the active site with subsequent poor
hydride ion reception and quinone reduction capability. Com-
pound 14 was also ranked as a poor substrate by our docking
studies.
Conclusions
The best substrates were the 2′-CONH(CH2)3CH3-7-NH-
COCH3 (12) and 2′-CO2(CH2)2OH-7-NH2 (16) derivatives with
reduction rates of 143 ( 11 and 60 ( 8 µmol/min/mg NQO1,
respectively (Table 5).
In Vitro Cytotoxicity. Cytotoxicity studies were also per-
formed on the lavendamycin analogues with cell survival being
Specific lavendamycin analogues of novel design were
synthesized and biologically evaluated in order to further assess
NQO1 SAR substrate qualities. The novel ligands were evalu-
ated a priori with docking studies employing an X-ray derived
NQO1 active site computational model to discern explicit

LaVendamycin Antitumor Agents
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 11 3111
850 mL of ice–water (1/2, ice/water). The mixture was allowed to
stand at room temperature for 36 h, and the solid was filtered off.
Recrystallization of the crude mixture from hot MeOH-H2O (V/V
20/80), gave pure crystals of 19, which were dried under vacuum
(9.04 g, 75%): mp 246.5–247 °C; Rf ) 0.66 (1/20, MeOH/CH2Cl2).
1H NMR (DMSO-d6) δ: 2.74 (s, 3H), 6.75 (dd, 2H, J ) 3.4, 1.6),
7.33–7.49 (m, 3H), 7.92–8.16 (m, 4H), 9.52 (s, 1H), 10.34 (s, 1H).
HRMS m/e calcd for C20H15N3O5, 377.1012; found, 377.1013.
7-N-Furoylamino-2-methylquinoline-5,8-dione (20). In a 500
mL round-bottomed flask equipped with a magnetic stirring bar,
5,7-difuroyl-amino-8-hydroxy-2-methylquinoline (19, 2.00 g, 0.005
mol) was suspended in 110 mL of glacial acetic acid and 10 mL
of THF. A solution of potassium dichromate (3.10 g, 0.01 mol) in
55 mL of water was added, and the resulting dark solution was
stirred at room temperature for 20 h. To the reaction mixture, 100
mL of water was added then was extracted with dichloromethane
(12 × 50 mL). The organic extracts were combined and washed
with 5% sodium bicarbonate (200 mL) then washed with saturated
sodium chloride solution (3 × 100 mL), dried with magnesium
sulfate, and evaporated in vacuo to give an orange-yellow product.
The solid was washed with 3 mL of acetone to give a light-yellow
solid. After drying, the pure product weighed 1.29 g (64.5%): mp
249–250 °C; Rf ) 0.75 (1/1 EtOAc/CH2Cl2). 1H NMR (CDCl3) δ:
2.79 (s, 3H), 6.63 (dd, 1H, J ) 3.5, 1.6), 7.35 (d, 1H, J ) 3.3),
7.58 (d, 1H, J ) 8.0), 7.65 (d, 1H, J ) 2.1), 8.02 (s, 1H), 8.34 (d,
1H, J ) 8.0), 9.35 (s, 1H). HRMS m/e calcd for C15H10N2O4,
282.0641; found, 282.0641.
analogue-active site interactions that are critical for ligand
substrate qualities. Taken together, the investigation has served
to validate the predictive qualities of the NQO1 computational
model. The study determined that our structure-based analogue
design criteria were valid, resulting in the design of two
analogues with high substrate specificity and selective toxicity
toward NQO1-rich cells. It also demonstrated that the in silico
model of the NQO1 active site correctly distinguished good and
poor NQO1 substrates, confirming that the model possessed
practical predictive power. Specifically, the addition of NH2 or
NHCOCH3 groups at R1 and CO2(CH2)2OH or CONH-
(CH2)3CH3 at R2 had the greatest positive impact on substrate
specificity compared to other substituents at these positions. The
best substrates were the 2′-CONH(CH2)3CH3-7-NHCOCH3 (12)
and 2′-CO2(CH2)2OH-7-NH2 (16) derivatives, which were also
the most selectively toxic to the NQO1-rich BE-NQ cell line
compared to the NQO1-deficient BE cell line. The docking data
were in agreement with the biological results, and therefore the
in silico model of the NQO1 active site can be considered
predictive and potentially of great value for the discovery and
development of new NQO1-directed lavendamycin antitumor
agents. Current efforts in our laboratories are utilizing the robust,
predictive NQO1 substrate model to iteratively design a third
generation of novel analogues with enhanced, selective, and
potent antitumor activity.
7-Furoylamino-2-formylquinoline-5,8-dione (5). In a 50 mL
round-bottomed flask equipped with a magnetic stirring bar, water
cooled reflux condenser, and an argon filled balloon was placed
7-furoylamino-2-methylquinoline-5,8-dione (20, 1.25 g, 4.42 mmol),
selenium dioxide (0.62 g, 5.54 mmol), 19 mL of dry distilled 1,4-
dioxane, and 0.55 mL of water. This mixture was stirred and slowly
heated to reflux over a 2 h period. The reaction was monitored by
TLC and allowed to stir for an addition 26 h until completion. Then
20 mL of 1,4-dioxane was added to the reaction mixture, and this
was stirred and refluxed for 15 min. The entire mixture was filtered
and the selenium metal was washed with 10 mL of CH2Cl2. To the
filtrate was added 100 mL of water, and then the aqueous layer
was extracted with CH2Cl2 (5 × 100 mL). The organic extracts
were combined and washed with 100 mL of 5% sodium bicarbonate
solution, dried with magnesium sulfate, and then evaporated in
vacuo. The product was dried under vacuum overnight and gave a
yellow solid weighing 1.23 g (93%): mp 234.5–235 °C; Rf ) 0.2
Experimental Section
Chemistry. General Methods. Chemical reagents and solvents
were purchased from Aldrich, Sigma, and Fisher Chemical com-
panies and used as received unless otherwise noted. 1,4-Dioxane,
xylene, and anisole were dried and distilled over sodium/benzophe-
none. Ethanol and chloroform were dried and distilled over CaH2.
DMF was dried over molecular sieves. Ammonium formate was
dried in a vacuum desiccator (CaCl2). Melting ranges were recorded
with a Thomas-Hoover capillary melting point apparatus. They are
in Celsius and are uncorrected. NMR spectra were recorded on a
JEOL Eclipse 400 or Varian Gemini 200 spectrophotometer using
CDCl3 or DMSO-d6 with TMS as internal standard. Chemical shifts
(δ) are in ppm and coupling constants (J) are in Hz. Low and high
resolution mass spectra were obtained at Indiana University Mass
Spectrometry Laboratory, Bloomington, IN. Analytical TLC was
performed on Baker silica gel and alumina strip with fluorescent
indicator. HPLC tracings were obtained with an Alltech Alltima
C18 (3 µm, 33 mm × 7 mm) column and Waters HPLC system
(2487 Dual λ Absorbance detector, two 515 HPLC pumps,
Empower2 software). The solvent program was isocratic acetoni-
trile:H2O (80:20 or 70:30) at 1 mL/min. Elemental analyses were
performed by Midwest Microlabs, Ltd.
1
(1/1 EtOAc/CH2Cl2). H NMR (CDCl3) δ: 6.65 (dd, 1H, J ) 3.5,
1.6), 7.39 (d, 1H, J ) 8.0), 7.67 (d, 1H, J ) 2.1), 8.17 (s, 1H),
8.35 (d, 1H, J ) 8.0), 8.66 (d, 1H, J ) 8.0), 9.35 (s, 1H), 10.32 (s,
1H). HRMS m/e calcd for C15H8N2O5, 296.0433; found, 296.0436.
Benzyloxycarbonyltryptophan n-Butyl Amide (22). In a 100
mL round-bottomed flask equipped with an argon flow and magnetic
stirring bar, benzyloxycarbonyltryptophan succinimide ester35 (21,
435 mg, 1 mmol), n-butylamine (0.11 mL, 1 mmol), 0.14 mL of
purified triethylamine, 13 mL of absolute ethanol, and 12 mL of
chloroform were combined. The reaction mixture was stirred for
1 h at room temperature while monitoring its progress with TLC.
Upon completion, the mixture was evaporated in vacuo to give a
solid. The solid was dissolved in 90 mL of EtOAc and washed
with 30 mL of water, 10% citric acid (2 × 30 mL), and then
neutralized with 15 mL of 1N NaHCO3. The organic layer was
then washed with 5 mL of water, dried over sodium sulfate, and
evaporated under vacuo until a colorless solid was obtained. The
product was dried under vacuum at 65 °C for 2 days to give 0.27 g
5,7-Diamino-8-hydroxy-2-methylquinolie Dihydrochloride
(18). In a 500 mL heavy-walled hydrogenation bottle, finely
powdered 8-hydroxy-2-methyl-5,7-dinitroquinoline52 (17, 8.99 g,
3.6 mmol), and 5% Pd/C (2.0 g) were suspended in 135 mL of
water, and 15 mL of concentrated hydrochloric acid. In a Parr
hydrogenator, this mixture was shaken under 41 psi of hydrogen
for 24 h. The reaction mixture was filtered, and the filtrate was
evaporated in vacuo to give a bright-orange solid. After drying under
1
vacuum, the pure solid product weighed 7.78 g, (82%): H NMR
(DMSO-d6) δ: 2.75 (s, 3H), 3.98–6.10 (br s, 7H), 6.54 (s, 1H),
7.10 (d, 1H, J ) 8.06), 8.67 (d, 1H, J ) 8.06).
1
(68%) of 22: mp 142–143 °C; Rf ) 0.80 (1/5 EtOH/EtOAc). H
5,7-Difuroylamino-8-hydroxy-2-methylquinoline (19). In a 250
mL two-necked round-bottomed flask equipped with a magnetic
stirring bar and an argon balloon, 8-hydroxy-2-methylquinoline-
5,7-diammonium dichloride (18, 8.00 g, 0.030 mmol), sodium
acetate (16 g, 0.20 mol), and sodium sulfite (16.00 g, 0.12 mol)
were placed in 60 mL of dry DMF. The solution was stirred and
cooled in an ice bath. To this solution, 2-furoyl chloride (6.0 mL,
0.06 mol) was added and stirred in the ice bath for 2 h. The mixture
was filtered, and the filtrate was placed in a 2 L beaker containing
NMR (CDCl3) δ: 0.80 (t, 3H, J ) 7.0), 1.04–1.26 (m, 4H),
3.04–3.19 (m, 2H), 3.31–3.40 (m, 2H), 4.44–4.47 (m, 1H), 5.12
(s, 2H), 5.5 (br s, 2H), 7.03 (s, 1H), 7.10–7.35 (m, 3H), 7.35 (s,
5H), 8.08 (br s, 1H). HRMS (EI) calcd for C23H27N3O3, 393.2047;
found, 393.2040.
Tryptophan n-Butyl Amide (7). In a 50 mL round-bottomed
flask equipped with a magnetic stirring bar and argon flow,
benzyloxycarbonyltryptophan butyl amide (22, 0.5 g, 1.3 mmol)

3112 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 11
Hassani et al.
was suspended in 30 mL of dry methanol. To this mixture, dry
ammonium formate (0.26 g, 4.09 mmol) and 10% Pd/C (0.26 g)
were added. The mixture was allowed to stir for 30 min. TLC was
used to monitor the progress of the reaction. Upon completion, the
reaction mixture was filtered and the filter cake was washed with
10 mL of methanol. The filtrate was evaporated in vacuo to nearly
dry, and the water bath temperature was then raised to 100 °C to
obtain a honey-brown colored gel. The gel was dried under vacuum
in a 50-60 °C oil bath for 2 days to give the final product (0.25 g,
7-N-Acetyldemethyllavendamycin sec-Butyl Amide (13). In
a 250 mL round-bottomed flask equipped with a magnetic stirring
bar and Dean–Stark trap were placed 7-acetamido-2-formylquino-
line-5,8-dione (109 mg, 0.45 mmol) and tryptophan sec-butyl amide
(116.4 mg, 0.45 mmol) along with 180 mL of dry anisole. This
mixture was stirred and heated, using an oil bath, to 167 °C over
a period of 2 h. The solution was refluxed for a total of 15 h, and
the completion of the reaction was verified by TLC. The mixture
was filtered while hot to remove the brown solid impurity. The
filtrate was rotary evaporated to dryness, and the resulting solid
was dissolved in a minimal amount of CHCl3 and acetone. This
solution was kept at room temperature overnight and then filtered
to yield an orange solid (73 mg 35%): mp 306–308 °C; Rf ) 0.73
(0.25/5 acetone/CHCl3). 1H NMR (CDCl3) δ: 1.07 (t, 3H, J ) 7.44),
1.39 (d, 3H, J ) 6.6), 1.60–1.80 (m, 2H), 2.39 (s, 3H), 4.18–4.25
(m, 1H), 7.38–7.48 (m, 1H), 7.63-7.76 (m, 2H), 7.89–7.95 (m,
1H), 8.04 (s, 1H), 8.28 (d, 1H, J ) 7.7), 8.49 (br s, 1H), 8.63 (d,
1H, J ) 8.4), 9.00 (d, 1H, J ) 8.4), 9.10 (s, 1H), 11.80 (br s, 1H).
HRMS (EI) calcd for C27H23N5O4, 481.1745; found, 481.1744.
7-N-(2-Furoyl)demethyllavendamycin Methyl Ester (15). In
a 50 mL three-necked round-bottomed flask, equipped with a reflux
condenser, argon flow, and a magnetic stirring bar was added
7-furoylamino-2-formylquinoline-5,8-dione (5, 29.7 mg, 0.1 mmol),
L-tryptophan methyl ester (21.8 mg, 0.1 mmol), and 100 mL of
dried, distilled xylene. This mixture was slowly heated to 145 °C
over a 3 h period and then was maintained at this temperature for
an additional 17 h. The reaction mixture was then evaporated in
vacuo to give a brownish-orange solid weighing 37.83 mg (77%).
This solid was purified by flash chromatography (1/1 CHCl3/THF).
The final product was an orange solid weighing 26.32 mg (53.5%):
1
77%): mp 74.5–75.0 °C; Rf ) 0.37 (1/5 EtOH/EtOAc). H NMR
(CDCl3) δ: 0.90 (t, 3H, J ) 7.2), 1.21-1.49 (m, 4H), 2.85–3.00
(m, 1H), 3.15–3.30 (m, 2H), 3.30-3.50 (m, 1H), 3.67–3.73 (m,
1H), 7.04 (s, 1H), 7.10–7.29 (m, 2H), 7.29–7.36 (m, 1H), 7.38 (d,
1H, J ) 7.7), 7.66 (d, 1H, J ) 7.7), 8.70 (br s, 1H). HRMS calcd
for C15H21N3O, 260.1763; found, 260.1764.
Benzyloxycarbonyltryptophan sec-Butyl Amide (23). In a 100
mL round-bottomed flask equipped with an argon flow and magnetic
stirring bar was placed benzyloxycarbonyltryptophan succinimide
ester35 (21, 0.870 g, 2 mmol), sec-butyl amine (0.22 mL, 2 mmol),
0.28 mL of purified triethylamine, 26 mL of absolute ethanol, and
24 mL of chloroform. The reaction mixture was stirred at room
temperature for 2 h while monitoring the reaction progress with
TLC. Once completed, the mixture was evaporated in vacuo to a
dry gel. The product was dissolved in 60 mL of EtOAc and washed
with 30 mL of water, 10% citric acid (8 mL), 10 mL of 5%
NaHCO3 solution, and then with 8 mL of brine. The organic solution
was dried over sodium sulfate overnight and then evaporated under
vacuum to dryness. The resulting material was placed under vacuum
for 2 days to give a clear-brown gel (0.71 g, 90%); Rf ) 0.46 (1/3
1
EtOAc/CHCl3). H NMR (CDCl3) δ: 0.79 (t, 3H, J ) 6.6), 0.90
1
mp 301.5–302.5 °C; Rf ) 0.65 (0.1/5 MeOH/CH2Cl2). H NMR
(d, 3H, J ) 6.42), 1.24 (t, 2H, J ) 8.0), 3.13–3.40 (m, 2H), 3.76
(m, 1H), 4.43 (t, 1H, J ) 5.5), 5.12 (s, 2H), 5.20–5.40 (m, 1H) 5.5
(br s, 1H), 7.03 (s, 1H), 7.20–7.28 (m, 2H), 7.30 (d, 1H, J ) 6.6),
7.32–7.40 (m, 5H), 7.71 (d, 1H, J ) 5.9), 8.10 (br s, 1H).
HRMS(EI) calcd for C23H27N3O3, 393.2047; found, 393.2049.
Tryptophan sec-Butyl Amide (8). In a 100 mL round-bottomed
flask equipped with a magnetic stirring bar and argon flow was
placed benzyloxycarbonyltryptophan sec-butyl amide (23, 0.5 g,
1.3 mmol) and 30 mL of dry methanol. To this was added dry
ammonium formate (0.26 g, 5.5 mmol) and 10% Pd/C (0.26 g).
The mixture was allowed to stir for 30 min. TLC was used to
observe the completion of the reaction (15 min). The brownish
precipitate was filtered, the filter cake was washed with 10 mL of
methanol, and the filtrate was evaporated in vacuo to near dryness.
The water bath temperature raised to 100 °C to further dry the
product and then finally heated under vacuum in a 60 °C oil bath
for two days. The product was a white solid and weighed 0.26 g
(75%): mp 72–73 °C; Rf ) 0.11 (0.2/3/2 MeOH/EtOAc/CH2Cl2).
1H NMR (CDCl3) δ: 0.90 (t, 3H, J ) 6.8), 1.09 (d, 3H, J ) 6.5),
1.39 (m, 2H), 2.90–3.45 (m, 2H), 3.68–3.72 (m, 1H), 3.90 (t, 1H,
J ) 6.5), 7.01 (s, 1H), 7.12 (d, 1H, J ) 8.4), 7.05–7.25 (m, 2H),
7.38 (d, 1H, J ) 8.0), 7.71 (d, 1H, J ) 8.0), 8.3 (br s, 1H). HRMS
(EI) calcd for C15H21N3O, 258.1601; found, 258.1610.
(CDCl3) δ: 3.20 (s, 3H), 4.12 (s, 3H), 6.67(dd, 1H, J ) 3.2, 1.4),
7.40–7.45 (dd, 1H, J ) 11.6, 8.0), 7.60–7.75 (m, 2H), 7.81 (d, 1H,
J ) 8.3), 8.13 (s, 1H), 8.27 (d, 1H, J ) 8.0), 8.63 (d, 1H, J ) 8.4),
9.26 (d, 1H, J ) 8.2), 9.44 (s, 1H), 11.92 (br s, 1H). HRMS m/e
calcd for C27H16N4O6, 492.1069; found, 492.1063.
Lavendamycin ꢀ-Hydroxyethyl Ester (16). In a 25 mL two-
necked round-bottomed flask, 7-N-acetyldemethyllavendamycin
ꢀ-hydroxyethyl ester35 (70.5 mg, 0.15 mmol) was added to 7.2 mL
of a 70% solution of sulfuric acid and under an argon atmosphere
was stirred and heated to 60 °C for 3.5 h. The reaction mixture
was carefully basified with a saturated solution of sodium car-
bonate to pH ) 8 and then evaporated in vacuo to dryness. The
residue was extracted with MeOH-CH2Cl2 (10 mL/60 mL) and
then (5 mL/30 mL). The organic extracts were evaporated in vacuo
to a small volume. Vacuum filtration of the resulting precipitate
gave the final product as a dark-red solid (18.9 mg (29.5%): mp >
280 °C; Rf ) 0.31 (0.3/5 MeOH/CH2Cl2). 1H NMR (DMSO-d6) δ:
3.83 (m, 2H), 4.44 (m, 2H), 5.02 (br s, 2H), 5.96 (s, 1H), 7.36–7.45
(m, 1H), 7.68–7.75 (m, 1H), 7.78 (d, 1H, J ) 8.0), 8.51 (d, 1H, J
) 8.0), 8.55 (d, 1H, J ) 8.1), 8.92 (d, 1H, J ) 8.1), 9.12 (s, 1H),
12.01 (br s, 1H). HRMS m/e calcd for C23H16N4O5, 428.1121;
found, 428.1136.
7-N-Acetyldemethyllavendamycin n-Butyl Amide (12). In a
100 mL three-necked, round-bottomed flask, equipped with a reflux
condenser, argon flow, and a magnetic stirring bar, 7-acetamido-
2-formylquinoline-5,8-dione (3, 36.6 mg, 0.15 mmol), tryptophan
n-butyl amide (7, 38.8 mg, 0.15 mmol), and 60 mL of dry anisole
were slowly heated to 165 °C over a 3 h period. The temperature
was maintained for an additional 20 h and then the mixture was
allowed to cool down to room temperature. The resulting dark-
brown precipitate was filtered and washed with a minimal amount
of chloroform to give 38 mg (53%) of 12. Flash chromatography
of the crude product (CHCl3) gave an orange solid (31 mg, 43%):
mp 243–244 °C; Rf ) 0.76 (0.1/5, MeOH/CH2Cl2). 1H NMR
(CDCl3) δ: 1.03 (t, 3H, J ) 7.1), 1.51–1.79 (m, 4H), 2.39 (s, 3H),
3.60–3.63 (m, 2H), 7.39 (dd, 1H, J ) 7.4, 7.6), 7.64 (d, 1H, J )
7.7), 7.64 (dd, 1H, J ) 7.4, 7.7), 8.00 (s, 1H), 8.00–8.17 (m, 1H),
8.19 (d, 1H, J ) 7.3), 8.47 (s, 1H), 8.50 (d, 1H, J ) 8.4), 8.89 (d,
1H, J ) 8.4), 9.00 (s, 1H), 11.63 (br s, 1H). HRMS m/e calcd for
C27H23N5O4, 481.1745; found, 481.1744.
Molecular Modeling. In Silico Model of the NQO1 Active
Site. The coordinates of the crystal structure of human NQO1
complex with bound FAD and indolequinone 27, obtained from
the Protein Data Bank (PDB ID code: 1H6945), were previously
utilized as a reference structure to develop the in silico model of
the NQO1 active site.8 Briefly, the physiological dimer in the crystal
unit was used for docking purposes. To develop the model, we
previously superposed the energy-minimized lavendamycin ana-
logue 29 to the coordinates of the original reference ligand 27 such
that the 3D overlap was optimal. Ligand 29 was energy minimized
in the context of the active site and therefore the position of the
ligand within the pocket was considered optimized for the purpose
of this study. The active site was then defined as all the amino
acid residues confined within a 6.5 Å radius sphere centered about
the superposed ligand 29. The active site coordinates were locally
minimized with attenuated iterations (100) by the Powell standard
method, using the Minimize Subset option within SYBYL. This
option automatically selected 24 seed amino acid residues sur-

LaVendamycin Antitumor Agents
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 11 3113
rounding the superposed ligand 29 to perform the local minimiza-
tion. Default parameters and values within the minimization
dialogue were used except where otherwise mentioned. The active
site minimization procedure yielded a weighted root-mean-square
distance of 0.26 Å between the 24 corresponding residues of interest
in the structures. The composite file containing ligand 29 and FAD
was saved. The composite structure without ligand 29 was utilized
as the in silico model of the NQO1 active site for docking studies.
Ligand 29 served as the reference ligand for the docking studies.
Docking calculations were performed using one of the two identical
active sites.
Ligand Preparation. The structures of ligands were built as
MOL2 files employing the Sketch Molecule module of the SYBYL
7.0 software suite60 (Tripos, Inc., St. Louis, MO). Initially sketched
ligands were subjected to energy minimization (10000 iterations)
by the Powell minimization standard method. Initial optimization
and termination parameters were set to None and Energy Change
options, respectively. Default parameters and values within the
minimization dialogue (Minimize Details) were used except where
otherwise noted. The final ligand conformational coordinates
were stored as MOL2 files within the database.
Docking. Flexible docking was performed using the FlexX
software module61,62 within the SYBYL 7.0 environment.60 The
automatic FlexX docking program assessed the conformationally
flexible ligands by employing the 3D structure of the target protein
in PDB format and was capable of determining 30 possible
conformations for each of the docked ligands. The final rank order
of conformations was based on the free binding energy. The
program automatically selected the base fragment of a ligand (the
ligand core). The base fragment was then automatically placed into
the active site of the target protein using the FlexX algorithmic
pose clustering approach that is based upon a pattern recognition
paradigm. Subsequent incremental reconstruction of the complete
ligand molecule was then performed by linking the remaining
components.61,62 For the in silico model, the active site was defined
as all the amino acid residues confined within a 6.5 Å radius sphere
centered on the superposed ligand 29. FAD was introduced to the
active site as a heteroatom file in MOL2 format.
Scoring Functions. The docked conformations of ligands were
evaluated and ranked using FlexX and four scoring functions
implemented in the CSCORE software module within the SYBYL
environment. The CSCORE module allowed consensus scoring that
integrated multiple well-known scoring functions such as FlexX,
ChemScore,64 D-Score,65 G-Score66 and PMF-Score67 to evaluate
docked ligand conformations. With CSCORE, columns were created
in a molecular spreadsheet that contain raw scores for each
individual scoring function. The consensus column contained
integers that ranged from 0 to 5, where 5 was the best fit to the
defined NQO1 model. A CSCORE threshold of g4 was used to
define those poses worthy of post-docking distance and residue
contact analyses discussed in the Results Section.
Cytochrome c Assay. Lavendamycin analogue reduction was
monitored using a spectrophotometric assay in which the rate of
reduction of cytochrome c was quantified at 550 nm. Briefly, the
assay mixture contained cytochrome c (70 µm), NADH (1 mM),
recombinant human NQO1 (0.1–3 µg) (gift from Dr. David Ross,
University of Colorado Health Sciences Center, Denver, CO), and
lavendamycins (25 µm) in a final volume of 1 mL of Tris-HCl (25
mM, pH 7.4) containing 0.7 mg/mL BSA and 0.1% Tween-20.
Reactions were carried out at room temperature and started by the
addition of NADH. Rates of reduction were calculated from the
initial linear part of the reaction curve (0–30 s), and results were
expressed in terms of µmol of cytochrome c reduced/min/mg of
NQO1 using a molar extinction coefficient of 21.1 mM-1 cm-1
for cytochrome c. All reactions were carried out at least in triplicate.
MTT Assay. Growth inhibition was determined using the MTT
colorimetric assay. Cells were plated in 96-well plates at a density
of 10000 cells/mL and allowed to attach overnight (16 h).
Lavendamycin analogue solutions were applied in medium for 2 h.
Lavendamycin analogue solutions were removed and replaced with
fresh medium, and the plates were incubated at 37 °C under a
humidified atmosphere containing 5% CO2 for 4–5 days. MTT (50
µg) was added, and the cells were incubated for another 4 h.
Medium/MTT solutions were removed carefully by aspiration, the
MTT formazan crystals were dissolved in 100 µL of DMSO, and
absorbance was determined on a plate reader at 560 nm. IC50 values
(concentration at which cell survival equals 50% of control) were
determined from semilog plots of percent of control vs concentra-
tion. Selectivity ratios were defined as the IC50 values for the BE
cell line divided by the IC50 values for the BE-NQ cell line.
Acknowledgment. We acknowledge financial support from
the American Cancer Society, the National Institutes of Health
(NIH grants: R15 CA78232 and P20 RR017670, H.D.B.; P20
RR15583, J.M.G.; and R15 CA74245, M.B.) and Ball State
University. We also would like to acknowledge Colgate-
Palmolive/Society of Toxicology for providing us with the
graduate student research award in alternative methods (MH).
We thank Professor David Williams and the staff at the Mass
Spectroscopy Laboratory of Indiana University for their as-
sistance in obtaining mass spectral data. The expert assistance
of Rohn Wood of the Molecular Computational Core Facility
of The University of Montana is greatly appreciated.
Supporting Information Available: NMR spectra, HPLC
tracings, analytical data, geometric postdocking analyses, and
molecular models of poses of lavendamycin analogues docked into
the NQO1 active site are available free of charge via the Internet
at http://pubs.acs.org.
References
(1) Skelly, J. V.; Sanderson, M. R.; Suter, D. A.; Baumann, U.; Read,
M. A.; Gregory, D. S.; Bennett, M.; Hobbs, S. M.; Neidle, S. Crystal
structure of human DT-diaphorase: A model for interaction with the
cytotoxic prodrug 5-(aziridin-1-yl)-2,4-dinitrobenzamide (CB1954).
J. Med. Chem. 1999, 42, 4325–4330.
(2) Danson, S.; Ward, T. H.; Butler, J.; Ranson, M. DT-diaphorase: A
target for new anticancer drugs. Cancer Treat. ReV. 2004, 30, 437–
449.
Molecular Graphics. The molecular graphics images were
prepared with PyMOL software, version PyMOLX11Hybrid 0.9768
(Delano Scientific, San Carlos, CA). The data for the coordinates
of the NQO1 complex with bound FAD and docked conformations
of ligands were prepared in PDB format as PyMOL input files.
PyMOL session files of the NQO1 active site with docked
conformations of ligands, and the superpositioning of the clustered
conformations were created and afforded the images herein and
within the Supporting Information.
(3) Lind, C.; Cadenas, E.; Hochstein, P.; Ernster, L. DT-diaphorase:
Purification, properties, and function. Methods Enzymol. 1990, 186,
287–301.
Biological Studies. Cell Culture. BE human colon adenocar-
cinoma cells and stably NQO1-transfected BE-NQ cells69 were a
gift from Dr. David Ross (University of Colorado Health Sciences
Center, Denver, CO). Cells were grown in minimum essential
medium with Earle’s salts, nonessential amino acids, L-glutamine,
and penicillin/streptomycin and supplemented with 10% fetal bovine
serum, sodium bicarbonate, and HEPES. Cell culture medium and
supplements were obtained from Gibco, Invitrogen Co. (Grand
Island, NY). The cells were incubated at 37 °C under a humidified
atmosphere containing 5% CO2.
(4) Faig, M.; Bianchet, M. A.; Talalay, P.; Chen, S.; Winski, S.; Ross,
D.; Amzel, L. M. Structures of recombinant human and mouse
NAD(P)H:quinone oxidoreductases: Species comparison and structural
changes with substrate binding and release. Proc. Natl. Acad. Sci.
U.S.A. 2000, 97, 3177–3182.
(5) Li, R.; Bianchet, M. A.; Talalay, P.; Amzel, L. M. The three-
dimensional structure of NAD(P)H:quinone reductase, a flavoprotein
involved in cancer chemoprotection and chemotherapy: Mechanism
of the two-electron reduction. Proc. Natl. Acad. Sci. U.S.A. 1995, 92,
8846–8850.
(6) Ernster, L. DT Diaphorase. Methods Enzymol. 1967, 10, 309–317.

3114 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 11
Hassani et al.
(7) Ernster, L. DT Diaphorase: A historical review. Chem. Scripta 1987,
27A, 1–13.
Ross, D. Elevated DT-diaphorase activity and messenger RNA content
in human non-small cell lung carcinoma: Relationship to the response
of lung tumor xenografts to mitomycin C. Cancer Res. 1992, 52, 4752–
4757.
(8) Hassani, M.; Cai, W.; Holley, D. C.; Lineswala, J. P.; Maharjan, B. R.;
Ebrahimian, G. R.; Seradj, H.; Stocksdale, M. G.; Mohammadi, F.;
Marvin, C. C.; Gerdes, J. M.; Beall, H. D.; Behforouz, M. Novel
lavendamycin analogues as antitumor agents: synthesis, in vitro
cytotoxicity, structure-metabolism, and computational molecular mod-
eling studies with NAD(P)H:quinone oxidoreductase 1. J. Med. Chem.
2005, 48, 7733–7749.
(28) Mikami, K.; Naito, M.; Ishiguro, T.; Yano, H.; Tomida, A.; Yamada,
T.; Tanaka, N.; Shirakusa, T.; Tsuruo, T. Immunological quantitation
of DT-diaphorase in carcinoma cell lines and clinical colon cancers:
Advanced tumors express greater levels of DT-diaphorase. Jpn. J.
Cancer Res. 1998, 89, 910–915.
(9) Beall, H. D.; Liu, Y.; Siegel, D.; Bolton, E. M.; Gibson, N. W.; Ross,
D. Role of NAD(P)H:quinone oxidoreductase (DT-diaphorase) in
cytotoxicity and induction of DNA damage by streptonigrin. Biochem.
Pharmacol. 1996, 51, 645–652.
(29) Rampling, R.; Cruickshank, G.; Lewis, A. D.; Fitzsimmons, S. A.;
Workman, P. Direct measurement of pO2 distribution and bioreductive
enzymes in human malignant brain tumors. Int. J. Radiat. Oncol. Biol.
Phys. 1994, 29, 427–431.
(10) Sun, X.; Ross, D. Quinone-induced apoptosis in human colon
adenocarcinoma cells via DT-diaphorase mediated bioactivation.
Chem.-Biol. Interact. 1996, 100, 267–276.
(11) Lewis, A. M.; Ough, M.; Du, J.; Tsao, M. S.; Oberley, L. W.; Cullen,
J. J. Targeting NAD(P)H:Quinone oxidoreductase (NQO1) in pan-
creatic cancer. Mol. Carcinog. 2006, 10.1002/mc.20199 .
(12) Siegel, D.; Gibson, N. W.; Preusch, P. C.; Ross, D. Metabolism of
mitomycin C by DT-diaphorase: Role in mitomycin C-induced DNA
damage and cytotoxicity in human colon carcinoma cells. Cancer Res.
1990, 50, 7483–7489.
(13) Siegel, D.; Beall, H. D.; Senekowitsch, C.; Kasai, M.; Arai, H.; Gibson,
N. W.; Ross, D. Bioreductive activation of mitomycin C by DT-
diaphorase. Biochemistry 1992, 31, 7879–7885.
(14) Pink, J. J.; Planchon, S. M.; Tagliarino, C.; Varnes, M. E.; Siegel, D.;
Boothman, D. A. NAD(P)H:Quinone oxidoreductase activity is the
principal determinant of beta-lapachone cytotoxicity. J. Biol. Chem.
2000, 275, 5416–5424.
(15) Planchon, S. M.; Pink, J. J.; Tagliarino, C.; Bornmann, W. G.; Varnes,
M. E.; Boothman, D. A. beta-Lapachone-induced apoptosis in human
prostate cancer cells: Involvement of NQO1/xip3. Exp. Cell Res. 2001,
267, 95–106.
(16) Walton, M. I.; Smith, P. J.; Workman, P. The role of NAD(P)H:
quinone reductase (EC 1.6.99.2, DT-diaphorase) in the reductive
bioactivation of the novel indoloquinone antitumor agent EO9. Cancer
Commun. 1991, 3, 199–206.
(17) Swann, E.; Barraja, P.; Oberlander, A. M.; Gardipee, W. T.; Hudnott,
A. R.; Beall, H. D.; Moody, C. J. Indolequinone antitumor agents:
Correlation between quinone structure and rate of metabolism by
recombinant human NAD(P)H:quinone oxidoreductase. Part 2. J. Med.
Chem. 2001, 44, 3311–3319.
(18) Beall, H. D.; Winski, S.; Swann, E.; Hudnott, A. R.; Cotterill, A. S.;
O’Sullivan, N.; Green, S. J.; Bien, R.; Siegel, D.; Ross, D.; Moody,
C. J. Indolequinone antitumor agents: Correlation between quinone
structure, rate of metabolism by recombinant human NAD(P)H:
quinone oxidoreductase, and in vitro cytotoxicity. J. Med. Chem. 1998,
41, 4755–4766.
(19) Fisher, G. R.; Gutierrez, P. L. Free radical formation and DNA strand
breakage during metabolism of diaziquone by NAD(P)H quinone-
acceptor oxidoreductase (DT-diaphorase) and NADPH cytochrome c
reductase. Free Radical Biol. Med. 1991, 11, 597–607.
(20) Ngo, E. O.; Nutter, L. M.; Sura, T.; Gutierrez, P. L. Induction of p53
by the concerted actions of aziridine and quinone moieties of
diaziquone. Chem. Res. Toxicol. 1998, 11, 360–368.
(21) Siegel, D.; Gibson, N. W.; Preusch, P. C.; Ross, D. Metabolism of
diaziquone by NAD(P)H:(quinone acceptor) oxidoreductase (DT-
diaphorase): Role in diaziquone-induced DNA damage and cytotoxicity
in human colon carcinoma cells. Cancer Res. 1990, 50, 7293–7300.
(22) Nemeikaite-Ceniene, A.; Dringeliene, A.; Sarlauskas, J.; Cenas, N.
Role of NAD(P)H:quinone oxidoreductase (NQO1) in apoptosis
induction by aziridinylbenzoquinones RH1 and MeDZQ. Acta Biochim.
Pol. 2005, 52, 937–941.
(30) Schlager, J. J.; Powis, G. Cytosolic NAD(P)H:(quinone-acceptor)
oxidoreductase in human normal and tumor tissue: Effects of cigarette
smoking and alcohol. Int. J. Cancer 1990, 45, 403–409.
(31) Balitz, D. M.; Bush, J. A.; Bradner, W. T.; Doyle, T. W.; O’Herron,
F. A.; Nettleton, D. E. Isolation of lavendamycin, a new antibiotic
from Streptomyces lavendulae. J. Antibiot. (Tokyo) 1982, 35, 259–
265.
(32) Doyle, T. W.; Balitz, D. M.; Grulich, R. E.; Nettleton, D. E.; Gould,
S. J.; Tann, C.; Moews, A. E. Structure determination of lavendamycin,
a new antitumor antibiotic from Streptomyces laVendulae. Tetrahedron
Lett. 1981, 22, 4595–4598.
(33) Boger, D. L.; Yasuda, M.; Mitscher, L. A.; Drake, S. D.; Kitos, P. A.;
Thompson, S. C. Streptonigrin and lavendamycin partial structures.
Probes for the minimum, potent pharmacophore of streptonigrin,
lavendamycin, and synthetic quinoline-5,8-diones. J. Med. Chem. 1987,
30, 1918–1928.
(34) Behforouz, M.; Cai, W.; Mohammadi, F.; Stocksdale, M. G.; Gu, Z.;
Ahmadian, M.; Baty, D. E.; Etling, M. R.; Al-Anzi, C. H.; Swiftney,
T. M.; Tanzer, L. R.; Merriman, R. L.; Behforouz, N. C. Synthesis
and evaluation of antitumor activity of novel N-acyllavendamycin
analogues and quinoline-5,8-diones. Bioorg. Med. Chem. 2007, 15,
495–510.
(35) Behforouz, M.; Cai, W.; Stocksdale, M. G.; Lucas, J. S.; Jung, J. Y.;
Briere, D.; Wang, A.; Katen, K. S.; Behforouz, N. C. Novel
lavendamycin analogues as potent HIV-reverse transcriptase inhibitors:
Synthesis and evaluation of anti-reverse transcriptase activity of amide
and ester analogues of lavendamycin. J. Med. Chem. 2003, 46, 5773–
5780.
(36) Fourie, J.; Oleschuk, C. J.; Guziec, F., Jr.; Guziec, L.; Fiterman, D. J.;
Monterrosa, C.; Begleiter, A. The effect of functional groups on
reduction and activation of quinone bioreductive agents by DT-
diaphorase. Cancer Chemother. Pharmacol. 2002, 49, 101–110.
(37) Gane, P. J.; Dean, P. M. Recent advances in structure-based rational
drug design. Curr. Opin. Struct. Biol. 2000, 10, 401–404.
(38) Joseph-McCarthy, D. Computational approaches to structure-based
ligand design. Pharmacol. Ther. 1999, 84, 179–191.
(39) Kuntz, I. D. Structure-based strategies for drug design and discovery.
Science 1992, 257, 1078–1082.
(40) Suleman, A.; Skibo, E. B. A comprehensive study of the active site
residues of DT-diaphorase: Rational design of benzimidazolediones
as DT-diaphorase substrates. J. Med. Chem. 2002, 45, 1211–1220.
(41) Erickson, J. A.; Jalaie, M.; Robertson, D. H.; Lewis, R. A.; Vieth, M.
Lessons in molecular recognition: the effects of ligand and protein
flexibility on molecular docking accuracy. J. Med. Chem. 2004, 47,
45–55.
(42) Lyne, P. D. Structure-based virtual screening: An overview. Drug
DiscoVery Today. 2002, 7, 1047–1055.
(43) Kontoyianni, M.; McClellan, L. M.; Sokol, G. S. Evaluation of docking
performance: Comparative data on docking algorithms. J. Med. Chem.
2004, 47, 558–565.
(44) Asher, G.; Dym, O.; Tsvetkov, P.; Adler, J.; Shaul, Y. The crystal
structure of NAD(P)H quinone oxidoreductase 1 in complex with its
potent inhibitor dicoumarol. Biochemistry 2006, 45, 6372–6378.
(45) Faig, M.; Bianchet, M. A.; Winski, S.; Hargreaves, R.; Moody, C. J.;
Hudnott, A. R.; Ross, D.; Amzel, L. M. Structure-based development
of anticancer drugs: Complexes of NAD(P)H:quinone oxidoreductase
1 with chemotherapeutic quinones. Structure 2001, 9, 659–667.
(46) Winski, S. L.; Faig, M.; Bianchet, M. A.; Siegel, D.; Swann, E.; Fung,
K.; Duncan, M. W.; Moody, C. J.; Amzel, L. M.; Ross, D.
Characterization of a mechanism-based inhibitor of NAD(P)H:quinone
oxidoreductase 1 by biochemical, x-ray crystallographic, and mass
spectrometric approaches. Biochemistry 2001, 40, 15135–15142.
(47) Zhou, Z.; Fisher, D.; Spidel, J.; Greenfield, J.; Patson, B.; Fazal, A.;
Wigal, C.; Moe, O. A.; Madura, J. D. Kinetic and docking studies of
the interaction of quinones with the quinone reductase active site.
Biochemistry 2003, 42, 1985–1994.
(23) Dehn, D. L.; Inayat-Hussain, S. H.; Ross, D. RH1 induces cellular
damage in an NAD(P)H:quinone oxidoreductase 1-dependent manner:
Relationship between DNA cross-linking, cell cycle perturbations, and
apoptosis. J. Pharmacol. Exp. Ther. 2005, 313, 771–779.
(24) Dehn, D. L.; Winski, S. L.; Ross, D. Development of a new isogenic
cell-xenograft system for evaluation of NAD(P)H:quinone oxidoreduc-
tase-directed antitumor quinones: evaluation of the activity of RH1.
Clin. Cancer Res. 2004, 10, 3147–3155.
(25) Winski, S. L.; Hargreaves, R. H.; Butler, J.; Ross, D. A new screening
system for NAD(P)H:quinone oxidoreductase (NQO1)-directed anti-
tumor quinones: Identification of a new aziridinylbenzoquinone, RH1,
as a NQO1-directed antitumor agent. Clin. Cancer Res. 1998, 4, 3083–
3088.
(26) Cresteil, T.; Jaiswal, A. K. High levels of expression of the NAD(P)H:
quinone oxidoreductase (NQO1) gene in tumor cells compared to
normal cells of the same origin. Biochem. Pharmacol. 1991, 42, 1021–
1027.
(48) Cavelier, G.; Amzel, L. M. Mechanism of NAD(P)H:quinone reduc-
tase: Ab initio studies of reduced flavin. Proteins: Struct., Funct.,
Genet. 2001, 43, 420–432.
(27) Malkinson, A. M.; Siegel, D.; Forrest, G. L.; Gazdar, A. F.; Oie, H. K.;
Chan, D. C.; Bunn, P. A.; Mabry, M.; Dykes, D. J.; Harrison, S. D.;

LaVendamycin Antitumor Agents
Journal of Medicinal Chemistry, 2008, Vol. 51, No. 11 3115
(49) Chen, S.; Wu, K.; Zhang, D.; Sherman, M.; Knox, R.; Yang, C. S.
Molecular characterization of binding of substrates and inhibitors to
DT-diaphorase: Combined approach involving site-directed mutagen-
esis, inhibitor-binding analysis, and computer modeling. Mol. Phar-
macol. 1999, 56, 272–278.
(50) Kende, A. S.; Ebetino, F. H. The regiospecific total synthesis of
lavendamycin methyl ester. Tetrahedron Lett. 1984, 25, 923–926.
(51) Boger, D. L.; Duff, S. R.; Panek, J. S.; Yasuda, M. Total synthesis of
lavendamycim methyl ester. J. Org. Chem. 1985, 50, 5790–5795.
(52) Behforouz, M.; Gu, Z.; Cai, W.; Horn, M. A.; Ahmadian, M. A highly
concise synthesis of lavendamycin methyl ester. J. Org. Chem. 1993,
58, 7089–7091.
(53) Behforouz, M.; Haddad, J.; Cai, W.; Arnold, M. B.; Mohammadi, F.;
Sousa, A. C.; Horn, M. A. Highly efficient and practical syntheses of
lavendamycin methyl ester and related novel quinolindiones. J. Org.
Chem. 1996, 61, 6552–6555.
(54) Fang, Y.; Linardic, C. M.; Richardson, D. A.; Cai, W.; Behforouz,
M.; Abraham, R. T. Characterization of the cytotoxic activities of novel
analogues of the antitumor agent, lavendamycin. Mol. Cancer Ther.
2003, 2, 517–526.
(55) Phillips, R. M.; Naylor, M. A.; Jaffar, M.; Doughty, S. W.; Everett,
S. A.; Breen, A. G.; Choudry, G. A.; Stratford, I. J. Bioreductive
activation of a series of indolequinones by human DT-diaphorase:
Structure-activity relationships. J. Med. Chem. 1999, 42, 4071–4080.
(56) Jaffar, M.; Phillips, R. M.; Williams, K. J.; Mrema, I.; Cole, C.; Wind,
N. S.; Ward, T. H.; Stratford, I. J.; Patterson, A. V. 3-Substituted-5-
aziridinyl-1-methylindole-4,7-diones as NQO1-directed antitumour
agents: Mechanism of activation and cytotoxicity in vitro. Biochem.
Pharmacol. 2003, 66, 1199–1206.
(57) Hendriks, H. R.; Pizao, P. E.; Berger, D. P.; Kooistra, K. L.; Bibby,
M. C.; Boven, E.; Dreef-van der Meulen, H. C.; Henrar, R. E. C.;
Fiebig, H. H.; Double, J. A.; Hornstra, H. W.; Pinedo, H. M.;
Workman, P.; Schwartsmann, G. EO9: A novel bioreductive alkylating
indoloquinone with preferential solid tumour activity and lack of bone
marrow toxicity in preclinical models. Eur. J. Cancer 1993, 29A, 897–
906.
1,4-naphthoquinone bioreductive drugs. Biochem. Pharmacol. 2004,
68, 2107–2116.
(59) Kremer, W. B.; Laszlo, J. Comparison of biochemical effects of
isopropylidine azastreptonigrin (NSC-62709) with streptonigrin (NSC-
45383). Cancer Chemother. Rep. 1967, 51, 19–24.
(60) SYBYL Molecular Modeling Software, version SYBYL 7.0; Tripos
Inc.: St. Louis, MO, 2004.
(61) Lemmen, C.; Lengauer, T. Time-efficient flexible superposition of
medium-sized molecules. J. Comput.-Aided Mol. Des. 1997, 11, 357–
368.
(62) Rarey, M.; Kramer, B.; Lengauer, T.; Klebe, G. A fast flexible docking
method using an incremental construction algorithm. J. Mol. Biol.
1996, 261, 470–489.
(63) Skibo, E. B.; Gordon, S.; Bess, L.; Boruah, R.; Heileman, M. J. Studies
of pyrrolo[1,2-alpha]benzimidazolequinone DT-diaphorase substrate
activity, topoisomerase II inhibition activity, and DNA reductive
alkylation. J. Med. Chem. 1997, 40, 1327–1339.
(64) Eldridge, M. D.; Murray, C. W.; Auton, T. R.; Paolini, G. V.; Mee,
R. P. Empirical scoring functions: I. The development of a fast
empirical scoring function to estimate the binding affinity of ligands
in receptor complexes. J. Comput.-Aided Mol. Des. 1997, 11, 425–
445.
(65) Kuntz, I. D.; Blaney, J. M.; Oatley, S. J.; Langridge, R.; Ferrin, T. E.
A geometric approach to macromolecule-ligand interactions. J. Mol.
Biol. 1982, 161, 269–288.
(66) Jones, G.; Willett, P.; Glen, R. C.; Leach, A. R.; Taylor, R.
Development and validation of a genetic algorithm for flexible docking.
J. Mol. Biol. 1997, 267, 727–748.
(67) Muegge, I.; Martin, Y. C. A general and fast scoring function for
protein-ligand interactions: A simplified potential approach. J. Med.
Chem. 1999, 42, 791–804.
(68) DeLano, W. L. The PyMOL Molecular Graphics System; versionPy-
MOL X11Hybrid; DeLano Scientific: San Carlos, CA, 2004.
(69) Winski, S. L.; Swann, E.; Hargreaves, R. H.; Dehn, D. L.; Butler, J.;
Moody, C. J.; Ross, D. Relationship between NAD(P)H:quinone
oxidoreductase 1 (NQO1) levels in a series of stably transfected cell
lines and susceptibility to antitumor quinones. Biochem. Pharmacol.
2001, 61, 1509–1516.
(58) Phillips, R. M.; Jaffar, M.; Maitland, D. J.; Loadman, P. M.; Shnyder,
S. D.; Steans, G.; Cooper, P. A.; Race, A.; Patterson, A. V.; Stratford,
I. J. Pharmacological and biological evaluation of a series of substituted
JM701066A