Synthesis and biological evaluation of guanidino compounds endowed with subnanomolar affinity as competitive inhibitors of maize polyamine oxidase

Article abstract of DOI:10.1021/jm900371z

Previous studies on agmatine and its derivatives suggested that the presence of hydrophobic groups on the guanidine moiety was a crucial key for inhibitory activity of maize polyamine oxidase. Accordingly, new lipophilic agmatine and iminoctadine derivatives were synthesized and tested for their ability to inhibit this enzyme. Several compounds showed an affinity in the nanomolar range, while a cyclopropylmethyl derivative of iminoctadine was found to be the most potent inhibitor of maize polyamine oxidase reported so far (K_i = 0.08 nM).

Full text of DOI:10.1021/jm900371z

4774 J. Med. Chem. 2009, 52, 4774–4785
DOI: 10.1021/jm900371z
Synthesis and Biological Evaluation of Guanidino Compounds Endowed with Subnanomolar Affinity as
Competitive Inhibitors of Maize Polyamine Oxidase
Fabrizio Manetti,†,§ Alessandra Cona,‡,§ Lucilla Angeli,†,§ Claudia Mugnaini,† Francesco Raffi,† Caterina Capone,‡
Elena Dreassi,† Alessandra Tania Zizzari,† Alessandra Tisi,‡ Rodolfo Federico,‡ and Maurizio Botta*,†
†Dipartimento Farmaco Chimico Tecnologico, Universitaꢀdegli Studi di Siena, Via Alcide de Gasperi 2, I-53100 Siena, Italy, and ‡Dipartimento di
Biologia, Universitaꢀ Roma Tre, Viale G. Marconi 446, I-00146 Roma, Italy. § These authors contributed equally to this work.
Received March 24, 2009
Previous studies on agmatine and its derivatives suggested that the presence of hydrophobic groups on
the guanidine moiety was a crucial key for inhibitory activity of maize polyamine oxidase. Accordingly,
new lipophilic agmatine and iminoctadine derivatives were synthesized and tested for their ability to
inhibit this enzyme. Several compounds showed an affinity in the nanomolar range, while a cyclopro-
pylmethyl derivative of iminoctadine was found to be the most potent inhibitor of maize polyamine
oxidase reported so far (Ki = 0.08 nM).
Introduction
interest was turned on the polyamine catabolic pathway
because of the attribution of programmed cell death
(apoptosis) induction to the PAO-mediated H2O2 production
in different tumor cell types.13-15 Likewise, by means of both
pharmacological and small interfering RNA strategies, it was
demonstrated that spermine oxidase (SMO) mediated oxida-
tive stress causes macrophage apoptosis in epithelial stomach
cells infected by Helicobacter pylori16 and can potentially
contribute to gastric cancer development.17 SMO expression
and activity are induced by H. pylori, and inhibition of SMO
activity with the inhibitor MDL 72527 (Chart 1) reduces both
H2O2 production and apoptosis.16 Since oxidative stress is
potentially carcinogenic because of its ability to damage DNA,
it was proposed that SMO-dependent H2O2 production re-
presents a potential mechanism for inflammation-induced
carcinogenesis.17,18 Consistently, it was demonstrated that
tumor necrosis factor-R (TNF-R), a proinflammatory cyto-
kine, induces H2O2 production and oxidative DNA damage in
epithelial cells through SMO-catalyzed spermine oxidation.18
Analogously, the PAO-dependent H2O2 production in plants
was shown to contribute to hypersensitive cell death occurring
after pathogen attack.19
In addition, the neurotoxicity of reactive aldehydes, which
are mediators of neurodegeneration in a number of neurolo-
gical disorders such as Alzheimer’s disease, is now well
established. Particularly, 3-aminopropanal is an endogenous
mediator of neuronal and glial cell death during cerebral
ischemia.20
Zea maysPAO (ZmPAO, formerlyMPAO) is the first PAO
whose tertiary structure has been determined,21,22 and over
the past years it has become the reference enzyme as far as
structure-activity relationships are concerned. Indeed, mole-
cular modeling of mouse SMO,23 Hordeum vulgare PAO1
(HvPAO1, formerly BPAO1), Hordeum vulgare PAO2
(HvPAO2, formerly BPAO2), and Arabidopsis thaliana
PAO1 (AtPAO1)5 revealed that the global fold of all the four
proteins strictly resembles that of ZmPAO. Consistently,
PAOs from different sources show similar affinity toward
Polyamine oxidases (PAOsa) are flavin adenine dinucleo-
tide (FAD) dependent enzymes involved in the catabolism of
ubiquitous polyamines via oxidative deamination of spermi-
dine (Spd), spermine (Spm), and/or their acetylated deriva-
tives at the secondary amino group.1-3 PAOs represent a
heterogeneous family of enzymes whose substrate specificity,
kinetic properties, and mode of substrate oxidation vary
depending on the biological source.1-3 The chemical identity
of the reaction products reflects the mode of oxidation of the
substrate, which is alternatively cleaved at the carbon located
on the endo-side or on the exo-side of the N4-nitrogen, giving
rise to polyamine terminal catabolism (PAO from Zea mays,4
Hordeum vulgare,5 Avena sativa)6 or back-conversion path-
way (animal N1-acetyl-PAO,7 spermine oxidase,8 yeast
Fms1,9 Arabidopsis thaliana PAO1,10 Arabidopsis thaliana
PAO3),11 respectively. Aminoaldehydes and H2O2 are common
products to both pathways of polyamine oxidation.
The great interest devoted to PAOs is derived from the
physiological relevance of their substrates, which are essential
growth factors, aswellasfrom the cytotoxic properties of their
reaction products, such as aminoaldehydes and H2O2. In this
regard, both the modulation of polyamine homeostasis and
the production of toxic compounds confer to PAO a crucial
regulative role in cell proliferation and death in normal and
cancer cells.
The search for selective inhibitors of polyamine biosynthesis
enzymes started with the purpose to develop pharmacological
strategies able to inhibit tumor growth.1,12 Only recently,
*To whom correspondence should be addressed. Phone: þ39 0577
234306. Fax: þ39 0577 234333. E-mail: botta@unisi.it.
a Abbreviations: PAO, polyamine oxidase; Spd, spermidine; Spm,
spermine; SMO, spermine oxidase; TNF, tumor necrosis factor;
ZmPAO, Zea mays polyamine oxidase; HvPAO1 (formerly BPAO1),
Hordeum vulgare polyamine oxidase-1; HvPAO2 (formerly BPAO2),
Hordeum vulgare polyamine oxidase-2; AtPAO1, Arabidopsis thaliana
polyamine oxidase-1; G3, N-prenylagmatine; APAO, animal N1-acetyl-
polyamine oxidase; MD, molecular dynamics; GA, genetic algorithm.
r
pubs.acs.org/jmc
Published on Web 07/10/2009
2009 American Chemical Society

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15 4775
Chart 1. Structures of Polyamine Inhibitors of PAO
Table 1. Structures of Agmatine and Iminoctadine Derivatives and
Their Affinity toward ZmPAO
different ligands,24 and therefore, ZmPAO protein represents
an important model protein in designing new inhibitors
suitable for every animal or plant PAOs.
The availability of selective and powerful PAO inhibitors
would appreciably contribute to the study of polyamine
homeostasis and to design new anticancer drugs. Inhibitors
of PAO activity belong to four main classes: (i) linear primary
diamines lacking secondary amino groups (such as 1,8-diami-
nooctane, 1, Table 1)25 and diamine derivatives;26 (ii) agma-
tine (2) and its analogues, including N-prenylagmatine
(G3, 3),27 compounds 4-9,28 and compounds 10-16; (iii)
diguanidino inhibitors, such as iminoctadine (17, a compo-
nent of the guazatine mixture)27 and its analogues (18-24),
which represent the most powerful PAO inhibitors; (iv) poly-
amine analogues lacking terminal amino groups including the
Spm analogue MDL 72527 (Chart 1)26 and three new power-
ful PAO inhibitors (namely, SL-11144, SL-11156, and
SL-11061) described by Maiale and co-workers.6
A previous analysis of the ZmPAO inhibition properties of
agmatine derivatives suggested a hydrophobic substituent as
the main determinant for inhibitory activity.25,28 In fact, the
addition of a lipophilic moiety (in particular, a prenyl sub-
stituent) to one of the terminal nitrogen atoms of the guani-
dine group of 2 (Table 1) greatly increased the inhibitory
activity of about 200-fold (compare 2 and its N-prenyl analo-
gue 3), owing to the presence of a hydrophobic pocket located
at the wide catalytic tunnel opening of ZmPAO, able to bind
the prenyl group with high affinity.25 Consistently, the sig-
nificant reduction of the hydrophobic pocket could explicate
the lower affinity displayed by 3, 17, and MDL 72527 for N1-
acetyl-PAO (APAO) and SMO compared to ZmPAO.24
Moreover, the lengthening of the aliphatic chain from four
(3) to five and six (4 and 5, respectively) carbon atoms did not
significantly affect activity, although 4 was found to be the
most potent agmatino-like derivative, according to the evi-
dence that the ZmPAO catalytic tunnel is a favorable hydro-
phobic environment for the binding of aliphatic molecules of
variable length.22
a In the experimental conditions utilized, since the lowest exploitable
concentration of ZmPAO, as imposed by the enzyme assay sensitivity, is
0.2 nM, Ki values for ZmPAO inhibition by some of the iminoctadine
derivatives (especially 19) are overestimated and thus represent higher
limit values.
groups different from the prenyl substituent of 4. The remain-
ing inhibitors 14-16 were characterized by modifications of
the aliphatic chain, while the prenyl group on guanidine was
kept fixed.
Moreover, during the fulfilment of this project, a LC-MS
methodology was developed for the analysis of guazatine,29 a
nonsystemic contact fungicide that disturbs the membrane
function of fungi and that is commercially available as a
complex mixture of polyamines and guanidines. Several of
the mixture components, already studied for their antifungal
properties,29 were tested for their ability to inhibit ZmPAO.
On the basis of these results, 4 was kept as a molecular
template to further investigate the structure-activity relation-
ships of agmatine analogues by design and synthesis of second
generation derivatives. In particular, some of them, with the
aminopentyl chain unchanged, were obtained by replacement
of the prenyl group with various saturated and unsaturated
rings (10-13) and designed to explore the ability of the
hydrophobic pocket of ZmPAO to accommodate lipophilic

4776 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15
Manetti et al.
Scheme 1a
a Reagents and conditions. (i) For 26a and 26g: RBr, TBAB, KOK, CH2Cl2/CH3CN, 19/1, room temp. For 26b-f and 26h,i: ROH, Ph3P, DIAD,
THF, 0 °C f reflux. (ii) For 27b-e: R-NH2, THF/H2O, 50 °C. For 27a and 27f: R-NH2, CH3CN. (iii) CF3COOH, CH2Cl2.
Scheme 2a
As a result, 17 (Table 1) showed Ki = 7.5 nM and was then
used as the template to design a number of lipophilic deriva-
tives (compounds 18-24, Table 1) characterized by the pre-
sence of a hydrophobic terminal portion previously suggested
to be important for the activity of ZmPAO inhibitors.
Chemistry
The N-Boc-protected S-methylisothiourea 2 (Scheme 1)
was alkylated with the appropriate bromide by transfer phase
catalysis to give the corresponding alkylated and protected S-
methylisothioureas 26a and 26g, while compounds 26b-f and
26h,i were obtained by treating 25 with the appropriate
alcohol under the Mitsunobu conditions. Compounds 26a-
e were then submitted to a nucleophilic displacement of the
a Reagents and conditions: (i) 1,4-diaminobutane, THF/H2O, 50 °C;
methylthio group by four different amines. While 1,5-penta-
(ii) 25, THF/H2O, 50 °C; (iii) CF3COOH, CH2Cl2.
nediamine and 4-aminobenzylamine were monoprotected
before their use, the protection of bis(3-aminopropyl)amine
was particularly troublesome, and for this reason it was used
unprotected. Compounds 27b-e were obtained in a good
80% yield starting from 26a-d and using a slight excess of the
amine in THF/H2O at 50 °C. Guanylation of 4-aminobenzy-
lamine and 3-methoxybenzylamine was accomplished in
CH3CN at room temperature, giving 27a and 27f with a
90% and 40% yield, respectively. Compound 28 (Scheme 2)
was obtained in two steps starting from 26a, which was
initially reacted in a THF/H2O mixture with a large excess
of 1,4-diaminobutane to give the high polar intermediate 27g,
which was guanylated by reaction with 25. Finally, the target
compounds 10-16 were obtained by simple deprotection of
the amino groups of 27a-g usingTFAindryCH2Cl2. Insome
cases, becuase of the high number of protected amino
groups in the molecule, multiple additions of TFA and long
reaction times were necessary to drive the reaction to comple-
tion.
Compounds 18-24 (Scheme 3) were obtained starting from
2.6 equiv of 1,17-diamino-9-azaheptadecane (29), which was
monoguanylated by the use of 1 equiv of the appropriate 26 to
give the intermediates 30a-f. These compounds were then
submitted to a second guanylation step in the presence of 1,3-
bis(tert-butoxycarbonyl)-2-triflylguanidine, affording the
Boc-protected derivatives 31a-f. On the other hand, treat-
ment of 1 equiv of 29 with 2 equiv of 26a afforded 31g in a
single synthetic step. Deprotection of 31a-g in the presence of
10% TFA in CH2Cl2 gave the final compounds 18-24.

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15 4777
Scheme 3a
a Reagents and conditions. (i) For 30a-f: 0.4 equiv of 26a or 26c or 26f-i, THF/MeOH, 50 °C, 16 h. For 31g: 2 equiv of 26a, THF/MeOH, 50 °C, 16 h.
(ii) 1,3-Bis(tert-butoxycarbonyl)-2-triflylguanidine, Et3N, CH2Cl2, room temp, 6 h. (iii)10% TFA, CH2Cl2, room temp, 24 h.
Biological Assays
Excellent results were obtained with the series of iminoc-
tadine (17) analogues. Iminoctadine, a component of the
guazatine mixture, strongly inhibited ZmPAO activity with a
Ki value of 7.5 nM, identical to that previously reported for
the guazatine mixture.27 In particular, the introduction of a
prenyl chain on one of the guanidine moieties of 17 led to 18
showing a very high affinity toward the enzyme, with a Ki
value in the nanomolar range (3.0 nM). The introduction of a
second prenyl group as in 24 (Ki = 1.7 nM), as well as
changing the prenyl chain of 17 with different unsaturated
moieties such as a γ-methylallyl (20, Ki = 1.1 nM), a
propargyl (23, Ki=0.7 nM), and a β-methylallyl group (21,
Ki=0.5 nM), slightly ameliorated the affinity of the mole-
cules with respect to the parent compound. Moreover,
replacement of the prenyl chain of 17 with a cyclopropyl-
methyl substituent resulted in 19 with Ki=0.08 nM. On the
basis of its affinity, compound 19 is the most potent inhibitor
of ZmPAO reported so far. Finally, a benzyl chain instead of
the prenyl group of iminoctadine led to 22, which retained
nanomolar affinity (Ki=1.0 nM).
ZmPAO was purified as previously described.4 Enzyme
activity was measured spectrophotometrically by following
the formation of a pink adduct resulting from the H2O2-
dependent oxidation of 4-aminoantipyrine catalyzed by
horseradish peroxidase and the subsequent condensation of
oxidized 4-aminoantipyrine with 3,5-dichloro-2-hydroxyben-
zenesulfonic acid.30 ZmPAO activity was assayed at pH 6.5
with Spd as the substrate in the presence or absence of test
compounds. Ki values were determined according to the
Dixon graphical method31 (see also Supporting Information).
Results and Discussion
Inhibition of ZmPAO by Agmatine and Iminoctadine Deri-
vatives. Agmatine (2) and its analogues (3-9) competitively
inhibited Spd oxidation catalyzed by ZmPAO.27,28 The
presence of a prenyl substituent on the guanidine moiety
(3-5) was the major determinant of the inhibitory activity
because of the presence of a hydrophobic pocket in the
ZmPAO active site, which is able to bind the prenyl group
with high affinity.25
Considering that 4 (Ki =10 nM)28 had proved to be the
most active compound among the analogues of 3, we de-
signed and synthesized compounds 10-13 with the five-
membered alkylamino chain kept fixed while the prenyl
moiety was replaced with cycloaliphatic or unsaturated/
aromatic groups. They showed comparable inhibitory activ-
ity (Ki values spanning from 1150 to 2580 nM) but were more
than 2 orders of magnitude less active than the parent
compound 4, thus indicating the prenyl group as the optimal
substituent and suggesting a high selectivity of the hydro-
phobic binding pocket in accommodating lipophilic substit-
uents of different nature. On the other hand, with the prenyl
substituent fixed on the guanidine and with modification of
the aminoalkyl portion of the molecule, the best result was
obtained with a 4-membered alkyl chain bearing a terminal
guanidine moiety (16, Ki=700 nM) and with the introduction
of a secondary nitrogen atom in a seven-membered alkyl chain
(15, Ki=1210 nM), while a benzylamino group (14) produceda
significant reduction of affinity (Ki=18 000 nM).
In summary, compounds 3-5 (Table 1) appeared to be the
most potent inhibitors among agmatine analogues (2-16)
while compounds belonging to the class of iminoctadine-like
inhibitors (18-24) showed similar potency, comparable to or
higher than that of iminoctadine itself.
Prediction of Pharmacokinetic Properties. Several physi-
cochemical properties (octanol/water partition coefficient,
aqueous solubility, apparent Caco-2 cell permebility, and
percent of human oral absorption) of the new PAO inhibi-
tors were predicted (QikProp software)32 to assess for their
druglikeness (Table 2). All the inhibitors showed a log P
in the range of the recommended values (-2.0 and 6.5 define
the range of log P values reported for 95% of known drugs).
The predicted aqueous solubility (log S) was also optimal for
all the molecules with the exceptions of 2 (out of the range)
and 9 and 24 (borderline values). The predicted apparent
Caco-2 cell permeability value (PCaco, an additional para-
meter of permeability that is used as a model for the gut-
blood barrier)33 was <25 nm/s for 2, 15, 17, 18, 20-23
(suggesting poor permeability) and 25-500 nm/s for the

4778 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15
Table 2. Predicted Physicochemical Parameters of Compounds 1-24
Manetti et al.
remaining compounds (suggesting moderate permeability).
Only 1 was predicted to have a great permeability (PCaco >
log S b PCaco
-0.42 688.03
0.85 23.51
human oral absorption (%)d
a
c
compd log Po/w
500 nm/s). The most active inhibitor (19) showed an optimal
value for either log P or log S, and its Caco-2 cell perme-
ability was predicted to be better than that of the other
iminoctadine derivatives (with the exception of 24), although
quite poor.
1
1.04
-1.42
0.99
1.36
1.65
0.36
1.61
0.08
3.51
2.11
0.94
2.31
1.90
1.46
0.58
0.81
1.47
3.97
3.33
3.57
3.61
4.23
3.05
6.18
70.84
30.20
70.10
72.47
73.92
67.13
75.10
64.97
80.66
79.13
74.63
79.44
82.31
72.59
41.90
57.40
30.65
57.35
58.68
57.10
58.61
61.80
51.85
76.03
2
3
-1.22 128.48
-1.89 125.05
-2.03 121.43
-0.36 134.30
-1.72 146.03
-0.11 125.66
-6.22 378.84
-2.34 168.20
-0.61 227.12
-2.62 150.10
-1.70 295.90
-2.15 118.32
4
5
6
Cell Growth and Viability Assay on 19. Compound 19, the
most active ZmPAO inhibitor reported in the literature, was
also evaluated for its effect on cell growth and viability in
tumor cells. The effectiveness in toxicity and in reducing
cancer cell growth was demonstrated by MTT survival and
Trypan blue assays on HeLa and U373-MG cell lines
characterized by a high and low proliferation rate, respec-
tively. In fact, cell proliferation, viability, and activity of
mitochondrial electron transport chain were determined
through the capacity of cells to reduce 3-(4,5-dimethylthia-
zol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) to forma-
zan, while the Trypan blue method was applied to distinguish
between living and dead cells. Cell viability measured by
MTT assay was significantly diminished in both cell lines
upon treatment with 1 μM 19 for 48, 72, and 96 h (Figure 1,
panels c and d). In addition, the Trypan blue dye exclusion
assay suggested a cytostatic effect rather than toxic (Figure 1,
panels a and b). However, a noticeable toxic effect was
instead revealed at 10 μM (Figure 1, panels c and d), the
lowest tested concentration that significantly induced cell
death (in a similar way, it was previously demonstrated that
other polyamine analogues could be cytotoxic at a 10 μM
concentration).34 Accordingly, Trypan blue count revealed a
number of viable cells appreciably lower than relative con-
trol to zero time cell-plating (data not shown). Therefore,
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
-0.62
23.39
-2.26 145.09
-3.21
-6.0
13.29
2.82
-3.92
-4.94
-4.90
-5.58
-4.76
25.60
17.37
20.48
19.41
13.14
-6.51 146.86
a Predicted octanol/water partition coefficient (-2.0 and 6.5 define
the range of log P values reported for 95% of known drugs). b Predicted
aqueous solubility in mol dm-3 is the concentration of the solute in a
3
saturated solution that is in equilibrium with the crystalline solid (-6.5
and 0.5 define the range of log P values reported for 95% of known
drugs). c Predicted apparent Caco-2 cell permeability in nm/s (<25, poor
permeability; >500, great permeability). d Percentage of predicted
human oral absorption (>80%, high absorption; <25%, poor
absorption).
Figure 1. Measurement of living cell number: HeLa (a) and U373-MG (b) cells were incubated in the absence (control) or presence of 1 μM 19
for 0, 24, 48, 72, and 96 h. The addition of Trypan blue allows us to distinguish viable, unstained cells from nonviable, blue-stained cells. Data
were expressed as (number of living cells) ꢀ107. Statistical analysis was performed with a two-way ANOVA test comparing compound
treatment versus control (//, P<0.001 was considered as statistically significant). MTT evaluation of cells proliferation: HeLa (c) and
U373-MG (d) cells were incubated in the absence (control) or presence of 0.1, 1, or 10 μM 19 and processed for MTT assay after 0, 24, 48, 72,
and 96 h. The relative rate of cell proliferation was expressed as percent of absorbance versus untreated control. Statistical analysis was
performed with a two-way ANOVA test. From a comparison of 1 μM compound treatment versus control, the MTT reduction rate gradually
decreases, being statistically significant (P<0.001) at 48-96 h.

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15 4779
Figure 2. (A) Graphical representation of the binding mode of 15 in P orientation in both charged (left) and uncharged (right) forms. (B) L
orientation of 15 in both charged (left) and uncharged (right) forms.
such a preliminary investigation demonstrated both toxic
effects and inhibition of proliferation on U373-MG and HeLa
cell lines treated with 19. In particular, a 10 μM concentration
was clearly toxic for tumor cells, already after a 24 h treatment
(Figure 1, panels c and d), whereas a 1 μM concentration
seemed to activate a significant blockage of cell proliferation.
Analysis of Binding Mode. As previously reported,25 atom-
ic coordinates of ZmPAO used in modeling simulations were
derived from the structure of the complex between the
enzyme and 17, refined at 1.9 A resolution,22 that is, along
with the 3D structure of yeast spermine amine oxidase,35 the
only member of the family whose tridimensional structure is
known. One of the most striking features of ZmPAO is a 30 A
long, U-shaped catalytic tunnel that hosts the FAD isoallox-
azine ring and the crucial catalytic residue Lys300, which is
highly conserved in all the proteins of APAO and SMO
families.24 The tunnel shows two openings of different sizes.
One of them, lined mainly by carboxylates of glutamic and
aspartic residues (namely, Asp88, Asp90, Asp117, Glu120,
Glu121, Glu124, Asp194, and Asp195) and forming the so-
called “carboxylate ring”, constitutes the large entrance of
the tunnel. Differently, the second opening is too narrow to
admit the ligands. As a consequence, both substrates and
inhibitors can get into the tunnel only by passing the largest
entrance. Moreover, since the tunnel has a diameter ranging
between 3.8 and 4.3 A,21 the ligands cannot invert their
orientation upon binding and are forced to keep the same
orientation once they had entered the tunnel. The tunnel is
mainly bordered by hydrophobic and aromatic residues,
with the exception of Glu62 and Glu170, that protrude in
front of the flavin and are within hydrogen bond distance from
each other and from active site ligands. The Glu62-Glu170
pair may represent the structural element that properly aligns
the substrate within the tunnel.22 Aromatic residues such as
Phe403, Tyr439, and Tyr298 are also crucial elements of the
catalytic tunnel. The side chains of Phe403 and Tyr439 are
positioned parallel to each other and flank the tunnel on
opposite sites, forming an aromatic sandwich that makes
interactions with enzyme-bound inhibitors. The Tyr298 side
chain also appears within hydrogen bond distance from active
site ligands.
Because of the uncertainty on the protonation state of
ligands within the enzyme binding site, computer simula-
tions on both the charged and uncharged forms of inhibitors
have been performed. As an example, the protonated form of
15 showed its terminal charged edge directed toward the
narrow entrance of the catalytic tunnel, suggesting that the
ligand penetrated the tunnel through its terminal ammonium
group (this binding mode is thereafter referred to as the P
orientation, Figure 2A, left side. Further details on the
binding mode of ZmPAO inhibitors are in Supporting
Information).
In addition, among the conformers found by docking
calculations, an opposite orientation (thereafter referred to
as the L orientation, Figure 2 B, left side) into the catalytic
tunnel, characterized by a comparable energy with respect to
the global minimum, was also found. In detail, the prenyl
group of 15 was located in proximity to the narrow entrance
of the tunnel, while the guanidino moiety formed a cation-π
interaction with the aromatic rings of Phe403 and Tyr439.

4780 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15
Manetti et al.
In contrast, simulations performed on the uncharged form
of ligands led to results difficult to be interpreted. In fact, the
P and the L orientations are both represented in the lower
energy clusters, avoiding the identification of a significantly
preferred orientation. The binding mode of unprotonated 15
in the L orientation (Figure 2 B, right side) is very similar to
that found for its charged form, while the P orientation
(Figure 2 A, right side) showed an interaction pattern within
the catalytic tunnel that is slightly different (see Supporting
Information).
Modeled complexes between docked inhibitors and the
ZmPAO catalytic binding site allowed us to rule out a
preliminary structure-affinity relationship of the new com-
pounds. In particular, the major determinant for activity of
agmatine derivatives 3-8 seemed to be their prenyl moiety
that in general was accommodated within the hydrophobic
pocket close to the large entrance of the tunnel, where the
most profitable region of interaction with a C1d probe of
GRID36 was found (Figure 3). In fact, insertion of the prenyl
group into the structure of agmatine (2) led to 3 with a
20-fold improved affinity, while replacement of the same
moiety by different unsaturated cycloalkyl or aromatic
groups (6-8 and 10-13) was detrimental for affinity, with
Ki values at least 1 order of magnitude higher. This trend was
accounted by docking calculations showing that hydropho-
bic substituents of 6-8 and 10-13 had shape and size that
lack several contacts with the major hydrophobic pocket of
the enzyme binding site. In contrast, variation of the length
of the alkyl spacer from four to five and six carbon atoms
(4 and 5, respectively) did not influence affinity, further
supporting the hypothesis that the hydrophobic moiety of
agmatine derivatives played a very important role in defining
affinity of compounds. In agreement with that, compounds
4 and 5, although characterized by longer alkyl chains with
respect to 3, showed similar interactions between their
terminal ammonium group and glutamates 62 and 170
(hydrogen bonds) or aromatic residues (cation-π inter-
actions) of the binding site.
On the other hand, keeping fixed the prenylguanidine
moiety of 3 and replacing its aminobutyl terminal with
various basic side chains, affinity underwent a 50- to 1200-
fold decrease (14-16). The low affinity of 14 and 16 seemed
to depend on the lack of a quaternary nitrogen atom (able to
bind glutamates and to interact with aromatic residues as in
3) acting as an anchor point to orientate the ligand within the
binding pocket. In fact, docked complexes did not show a
preferential binding mode for such compounds. On this
basis, however, it is impossible to rationalize the affinity of
15 (1210 nM), which possessed such a structural feature and
showed interactions similar to that of better compounds.
The importance of a protonatable nitrogen atom able to
contact Glu62, Glu170, and near neighboring aromatic
residues was also supported by the X-ray crystallographic
structure of ZmPAO in complex with 17, which showed such
interactions. Moreover, the nanomolar affinity of 17 also
depended on the long alkyl chains contacting aromatic
residues of the catalytic tunnel and on cation-π interactions
between guanidino groups and aromatic amino acids (all
reproduced by docking simulations, thus giving support for
the reliability of the computational protocol). Derivatives of
17, obtained by insertion of a lipophilic moiety at one or both
of its guanidino edges, were all characterized by affinity values
better than that of the parent compound. As expected, the
preferred orientation of docked complexes of iminoctadine
Figure 3. Graphical representation of the lowest energy ZmPAO-
24 complex (ball-and-stick notation, carbon atoms in cyan) as
obtained by docking simulations. Regions of minimum energy as
derived from molecular interaction field calculations are also dis-
played for the C1d probe (green spheres).
derivatives 18-24 suggested that they could penetrate the
tunnel by their charged edge (the guanidine moiety), similar
to agmatine derivatives, while the terminal lipophilic group
was located within the hydrophobic pocket close to the large
entrance of the tunnel. However, affinity values did not
significantly depend on the lipophilic substituent, affinity
ranging from 0.5 to 3 nM, with the sole exception of 19
(0.08 nM). This was probably due to the huge number of
profitable interactions that iminoctadine derivatives made
with the catalytic tunnel, much higher than those involving
smaller molecules, such as prenylagmatine derivatives.
Compound 19, which exhibited the highest activity among
the new guanidine derivatives and, to the best of our knowl-
edge, among all known ZmPAO inhibitors, penetrates the
tunnel by means of the free guanidine moiety and, in addition
to the polar interactions (see Supporting Information), it
showed peculiar lipophilic contacts. In fact, after minimiza-
tion, the χ angle of the side chain of Leu186 shifted from
-82.73° to -145.17°, assuming a different rotameric state
that created a hydrophobic pocket (defined by Ile140,
Val184, Leu186, Phe167, and Phe171) not present in the
crystallographic complex between ZmPAO and iminocta-
dine. This pocket, mapped by a GRID minimum of the DRY
probe, allowed the cyclopropyl moiety to have very favor-
able lipophilic interactions (Figure 4). In particular,
although the DRY probe identified the first part of catalytic
tunnel as favorable for lipophilic contacts, three major
interaction zones could be described: (1) a pocket lined by
Trp60, Phe89, and Val196 interacting with one of the poly-
methylene spacer, (2) the region in front of the aromatic ring
of Phe189, in contact with the second polymethylene spacer,
and (3) the space between side chains of Leu186 and Phe171,
accommodating the cyclopropyl moiety.
To check for the stability of the 19-ZmPAO complex and
of the lipophilic interactions between the cyclopropyl moiety
and the pocket defined by Ile140, Val184, Leu186, Phe167,
and Phe171, a molecular dynamics (MD) simulation was
performed. During the initial steps of calculations, the
hydrophobic interactions observed in the minimized com-
plex were maintained, with the χ angle of Leu186 always
found in a trans conformation. Next, such a dihedral
angle shifted from a trans to a gaucheþ conformation, thus

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15 4781
interactions (-4.3 kcal/mol) for such a probe was found in a
region close to Phe89, Trp60, and Tyr169, inaccessible to
prenyl groups.
In summary, docking simulations and GRID calculations
further support the hypothesis that the protonated form of
inhibitors could be preferential with respect to the corres-
ponding uncharged form, in agreement with that previously
found by us on agmatine derivatives,24 as well as suggested
that charged inhibitors preferentially approached the tunnel
through their charged terminal moiety (P orientation).
Moreover, while the prenyl moiety and the terminal amino
group of agmatine derivatives appeared as the crucial keys
for affinity, lipophilic terminal moieties of iminoctadine
analogues should be considered as structural elements for
fine-tuning of affinity toward the enzyme.
In conclusion, previous studies on agmatine derivatives
suggested structural modifications to improve affinity of
compounds toward PAO. New agmatine and iminoctadine
analogues have been synthesized and tested for their inhibi-
tory properties against maize polyamine oxidase, allowing us
to identify a hit compound found to be the most potent
inhibitor known so far (Ki = 0.08 nM). Because of the great
interest devoted to PAOs and their inhibitors in cell prolif-
eration and death in normal and cancer cells, the new
compounds described in this paper could be considered as
very interesting hits in the search of new agents able to inhibit
cell proliferation in particular tumor cell lines. However,
although the global fold of computational models previously
reported for several plant and animal PAOs strictly resem-
bles that of ZmPAO,23 we are aware of the fact that
structural differences might occur at their binding pockets,
thus leading compounds to have different affinity toward
these enzymes. On the basis of these considerations and on
the lack of PAO structures from experimental sources
(except for ZmPAO and yeast Fms1), additional efforts
should be made to generate reliable models of mammalian
and human PAO to be used for the design of new enzyme
inhibitors and for rationalizing pre-existing biological data.
Figure 4. Binding mode of 19 (ball-and-stick notation, carbon
atoms in green) into the ZmPAO catalytic binding site. Pink spheres
represent regions of minimum energy as derived from molecular
interaction field calculations with the GRID DRY probe. Inter-
molecular hydrogen bonds are represented as black dashed lines.
Residues lining the catalytic tunnel are shown.
Figure 5. Distance between the centroids of aromatic ring of
Phe171 and cyclopropyl moiety of compound 19 during the course
of molecular dynamics simulations.
Experimental Section
Chemistry. Starting materials were purchased from Aldrich-
Italia (Milan, Italy). Melting points were determined with a
Buchi 530 apparatus and are uncorrected. 1H NMR spectra
were recorded on a Varian Gemini 200 (200 MHz) instrument.
Chemical shifts are reported as δ (ppm) relative to TMS as
internal standard, and J is in Hz. 1H patterns are described using
the following abbreviations: s=singlet, d=doublet, t=triplet,
m = multiplet, b = broad. Analyses for C, H, N were with-
in (0.4% of the theoretical value. No residue remained after
combustion, thus indicating a purity of >95%.
opening a gap that allowed the cyclopropylmethyl moiety to
penetrate more deeply into the lipophilic pocket and to make
a profitable stacked interaction with the aromatic ring of
Phe171, as shown in Figure 5, depicting the distance in A
between the centroid of the Phe171 ring and the cyclopropyl
group versus the frames of the MD run. Such a result
confirmed the importance of the lipophilic interactions with
Leu186 and Phe171 (found by docking simulations and by
GRID calculations). In fact, despite the high flexibility of 19
and the wide size of the catalytic tunnel entrance, the
cyclopropyl moiety maintained lipophilic interactions and,
upon large fluctuations found from the beginning to the
middle of the simulation (Figure 5), it was also able to
reinforce contacts with Phe171.
Compound 24, bearing a prenyl group at each of its
molecular edges, interacted by the guanidino moieties with
Asp117 and Glu170, while the secondary amino group
formed a hydrogen bond with the side chain of Glu62. Prenyl
moieties contacted two of the three zones preferred by the
C1d probe (Figure 3), corresponding to Phe171, Leu184,
and Thr183 (-4.0 kcal/mol) and to Phe171, Glu120,
and Phe167 (-3.7 kcal/mol), although the most profitable
General Procedure for the Synthesis 26a and 26g. To a
suspension of KOH (157 mg, 2.8 mmol) in a CH2Cl2/CH3CN
(19/1) mixture (3.5 mL) were added tetrabutylammonium bro-
mide (65 mg, 0.2 mmol) and N,N0-bis(tert-butoxycarbonyl)-S-
methylisothiourea (290 mg, 1 mmol). After a few minutes, a
solution of the appropriate bromide (2.4 mmol) in CH2Cl2/
CH3CN (19/1, 3.5 mL) was added dropwise and the reaction
mixture was stirred at room temperature for 16-18 h. The
reaction mixture was poured into ice, and the aqueous layer
was extracted once with CH2Cl2. The organic phase was washed
with brine and dried over Na2SO4. The solvent was evaporated
under reduced pressure, and the crude mixture was purified by
flash chromatography.
N1,N2-Bis(tert-butoxycarbonyl)-N1-(γ,γ-dimethylallyl)-S-methy-
lisothiourea (26a). Reaction time: 16 h. Oil, 70% yield. 1H NMR
(CDCl3, 200 MHz) δ 5.14-5.10 (m, 1H), 4.0-3.97 (m, 2H), 2.22 (s,

4782 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15
Manetti et al.
3H), 1.58-1.54 (m, 6H), 1.36 (s, 9H), 1.33 (s, 9H). MS (ESI) m/z:
739.1 (2M þ Na)þ, 381.0 (M þ Na)þ, 359.1 (M þ H)þ.
N1,N2-Bis(tert-butoxycarbonyl)-N1-(β-methylallyl)-S-methy-
lisothiourea (26g). Reaction time: 18 h. Oil, 75% yield. 1H NMR
(CDCl3, 200 MHz) δ 4.63 (s, 2H), 4.05 (s, 2H), 2.30 (s, 2H), 1.68
(s, 3H), 1.44 (s, 9H), 1.40 (s, 9H). MS (ESI) m/z: 710.9 (2Mþ
Na)þ, 367.0 (M þ Na)þ, 345.1 (M þ H)þ.
NMR (CDCl3, 200 MHz) δ 3.69-3.54 (m, 2H), 3.21-3.11 (m,
4H), 1.73-1.12 (mþ3s, 44H), 0.98-0.80 (m, 2H).
Synthesis of 4-[N2,N3-Bis(tert-butoxycarbonyl)-N3-(γ,γ-dime-
thylallyl)guanidine]-1-[N5,N6-bis(tert-butoxycarbonyl)guanidino]-
butane (28). To a solution of 1,4-diaminobutane (739 mg, 8.38
mmol) in a THF/H2O, 50/1, mixture (20 mL) was added drop-
wise a solution of 26a (1.00 g, 2.79 mmol) in THF (15 mL), and
the reaction mixture was heated at 50 °C for 3 h. After removal of
the solvent, the residue was partitioned between CH2Cl2 and
10% NaHCO3 and the aqueous layer was extracted twice with
CH2Cl2. The combined organic phases were washed with brine
and dried over Na2SO4, and the residue obtained by removal of
the solvent was dissolved in a THF/H2O, 50/1, mixture. To this
solution, a solution of 25 (0.81 g, 2.79 mmol) in THF (10 mL) was
added dropwise, and the reaction mixture was heated at 50 °C for
16 h. The residue obtained by standard workup using CH2Cl2
and H2O waspurified by flash chromatography to give 6 (0.625 g,
35% yield) as a brown oil. 1H NMR (CDCl3, 200 MHz) δ 5.28-
5.12 (m, 1H), 4.26-4.14 (m, 2H), 3.28-3.02 (m, 4H), 1.78-1.26
(mþ4s, 46H).
General Procedure for the Synthesis 26b-f and 26h,i. To a
stirred solution of N,N0-bis(tert-butoxycarbonyl)-S-methyli-
sothiourea (2.90 g, 10 mmol) in dry THF (36 mL) were added
Ph3P (3.93 g, 15 mmol) and the appropriate alcohol (13 mmol).
The reaction mixture was cooled to 0 °C, and DIAD (2.95 mL,
15 mmol) was added dropwise. After being stirred at reflux
temperature for 17 h, the reaction mixture was concentrated and
diluted with H2O and CH2Cl2, the aqueous layer was extracted
twice with CH2Cl2, and the combined organic phases were
washed with brine and dried over Na2SO4. The solvent was
evaporated under reduced pressure, and the crude mixture was
purified by flash chromatography.
Example. N1,N2-Bis(tert-butoxycarbonyl)-N1-(cyclohexylethyl)-
S-methylisothiourea (26b). Yellow oil, 95% yield. 1H NMR
(CDCl3, 200 MHz) δ 3.53-3.45 (m, 2H), 2.34 (s, 3H), 1.79-1.40
(m, 6H), 1.47 (s, 9H), 1.44 (s, 9H), 1.32-1.08 (m, 5H), 1.05-0.86
(m, 2 H).
General Procedure for the Synthesis of 10-16. Freshly dis-
tilled CF3COOH (0.917 mL, 12.28 mmol) was added to a
solution of 27 (for 10-15) or 28 (for 16) (1 mmol) in dry CH2Cl2,
and the reaction mixture was stirred at room temperature for
3-16 h depending on the number of protecting groups to be
removed. Removal of the solvent under vacuum gave a residue
which was washed with several times petroleum ether to give
10-16 as pure compounds.
General Procedure for the Synthesis 27a and 27f. To a solution
of 26 (4 mmol) in CH3CN (10 mL) was added the appropriate
amine (1 mmol), and the reaction mixture was stirred for 16 h at
room temperature. After removal of the solvent, the residue was
partitioned between CH2Cl2 and H2O and the aqueous layer was
extracted twice with CH2Cl2. The combined organic phases were
washed with brine and dried over Na2SO4. Removal of the
solvent under vacuum gave a residue which was purified by flash
chromatography.
Example. N1-Benzylamine-N3-(γ,γ-dimethylallyl)guanidine Bis-
1
(trifluoroacetate) (14). Oil, 99% yield. H NMR (CD3OD, 200
MHz) δ 7.54 (d, J=8.56 Hz, 2H), 7.36 (d, J=8.59 Hz, 2H), 5.30-
5.27 (m, 1H), 4.12 (s, 2H), 3.87 (d, J=6.74 Hz, 2H), 1.71 (s, 3H),
1.62 (s, 3H). MS (EI) m/z 233 (M þ H).
N1-Benzylamine-N3-(γ,γ-dimethyallyl)-N2,N3,N4-tris(tert-bu-
toxycarbonyl)guanidine (27a). Oil, 92% yield. 1H NMR (CDCl3,
200 MHz) δ 7.29-6.94 (m, 4H), 5.26 (m, 1H), 4.39-4.12 (m,
4H), 1.70 (s, 3H), 1.66 (s, 3H), 1.51 (s, 9H), 1.45 (s, 9H), 1.43 (s,
9H).
General Procedure for the Synthesis of 30a-f. To a stirred
solution of 1,17-diamino-9-azaheptadecane (4.08 g, 15.06 mmol)
in 5/3 THF/CH3OH (80 mL) at 50 °C, a solution of N,N0-bis(tert-
butoxycarbonyl)-N-(alkyl)-S-methylisothiourea (5.02 mmol) in
THF (25 mL) was added dropwise over 1 h. After 18 h, the
reaction mixture was concentrated under reduced pressure
and the residue was purified by flash chromatography (6%
methanol, 4% triethylamine, 90% ethyl acetate), affording pure
30a-f.
N1-(3-Methoxybenzyl)-N3-(5-aminopentyl)-N2,N3,N4-tris(tert-
butoxycarbonyl)guanidine (27f). Oil, 56% yield. 1H NMR
(CDCl3, 200 MHz) δ 7.29-6.72 (m, 4H), 4.36 (s, 2H), 3.78 (s,
3H), 3.64-3.52 (m, 2H), 3.12-3.29 (m, 2H), 1.65-1.42 (mþ3s,
33H).
Example. 1-Amino-17-(N1,N2-bis(tert-butoxycarbonyl)-N1-(γ,γ-
dimethylallyl)guanidine)-9-azaheptadecane (30a). Oil, 80% yield.
1H NMR (CDCl3, 200 MHz) δ 7.22 (bs, NH), 5.13 (t, J=6,6 Hz,
1H), 4.15-4.12 (m, 2H), 3.05-2.98 (m, 2H), 2.61-2.46 (m, 6H),
1.60-1.58 (m, 6H), 1.45-1.42 (m, 22H), 1.32 (bs, 20H). MS (ESI)
m/z 582 (M þ H)þ.
Synthesis of N1-[(30-Aminopropyl)-3-aminopropyl]-N3-(γ,γ-
dimethylallyl)-N2,N3-bis(tert-butoxycarbonyl)guanidine (27b). To
a solution of N-(3-aminopropyl)-1,3-propanediamine (1.2 mL,
8.37 mmol) in a 50/1 THF/H2O mixture, a solution of 26a (1.00 g,
2.79 mmol) in THF (15 mL) was added dropwise over 1 h. The
reaction mixture was heated at 50 °C for 3 h, and then it was
concentrated under vacuum and diluted with H2O and CH2Cl2.
The aqueous phase was extracted twice with CH2Cl2. Then the
combined organic solutions were washed with brine and dried
over Na2SO4. The residue obtained after the removal of the
solvent was purified by flash chromatography to give 27b (689
mg, 56% yield) as a yellow oil. 1H NMR (CDCl3, 200 MHz) δ 5.10
(t, J=6.9 Hz, 1H), 4.11 (d, J=7.0 Hz, 2H), 3.62-3.48 (m, 1H),
3.34-3.05 (m, 3H), 2.69-2.47 (m, 4H), 1.68-1.47 (m, 10 H), 1.38
(s, 9H), 1.36 (s, 9H).
General Procedure for the Synthesis of 31a-g. Et3N (67 μL,
0.48 mmol) and 1,3-bis(tert-butoxycarbonyl)-2-triflylguanidine
(188 mg, 0.48 mmol) were added to a stirred solution of 30 (0.44
mmol) in CH2Cl2 (8 mL), and the mixture was stirred at room
temperature for 6 h. The mixture was concentrated under
reduced pressure and purified by flash chromatography
(MeOH/Et3N/EtOAc, 3/2/95), affording 31a-g as pure com-
pounds.
Example. 1-(N1,N2-Bis(tert-butoxycarbonyl)guanidine)-17-(N1,
N2-bis(tert-butoxycarbonyl)-N1-(γ,γ-dimethylallyl)guanidine)-9-aza-
heptadecane (31a). Oil, 82% yield. H NMR (CDCl3, 200 MHz)
1
General Procedure for the Synthesis of 27c-e. To a solution of
N-(tert-butoxycarbonyl)-1,5-pentandiamine (1.17 g, 5.81 mmol)
in THF (10 mL) was added dropwise a solution of 26
(2.90 mmol) in THF (10 mL), and the reaction mixture was
heated at 50 °C for 18 h. After removal of the solvent, the residue
was partitioned between CH2Cl2 and H2O and the aqueous
phase was extracted twice with CH2Cl2. The combined organic
phases were washed with brine and dried over Na2SO4. The
residue obtained after removal of the solvent was purified by
flash chromatography to give the desired pure products.
Example. (N1-5-Aminopentyl)-N3-(cyclohexylethyl)-N1,N2,N3-
tris(tert-butoxycarbonyl)guanidine (27c). Oil, 95% yield. 1H
δ8.20(bs,NH),5.13(t,J= 6.21 Hz, 1H), 4.21-4.18 (m, 2H), 3.25-
3.18 (m, 4H), 3.12-3.05 (m, 2H), 2.79-2.66 (m, 4H), 1.65-1.63
(m, 6H), 1.46-1.43 (m, 44H), 1.27 (bs, 16H). MS (ESI) m/z 824
(MþH)þ.
General Procedure for the Synthesis of 18-24. A 10% solution
of freshly distilledTFA inanhydrous CH2Cl2 (3 mL) was added to
31 (0.102 mmol) under argon atmosphere. The reaction mixture
was stirred at room temperature for 24 h and concentrated under
reduced pressure to give 18-24 in quantitative yield.
Example. 1-(Guanidino)-17-(N1-(γ,γ-dimethylallyl)guanidino)-9-
azaheptadecane tris(trifluoroacetate) (18). Oil. 1H NMR

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15 4783
((CD3)2CO, 200 MHz) δ 8.62 (bs, NH), 7.85 (bs, NH), 7.32 (bs,
NH), 5.25-5.23 (m, 1H), 3.41-3.32 (m, 2H), 3.73-3.65 (m, 2H),
3.41-3.35(m, 2H), 3.05-2.96 (m, 4H), 1.69-1.65 (m, 6H), 1.63 (bs,
8H), 1.45 (bs, 16H). MS (ESI) m/z 424 (M þ H)þ.
from the complex, with the exception of the following molecules
which were considered in all the calculations (reported in
parentheses is the numeration of the corresponding water
molecules of complex ZmMPAO-1,8-diaminooctane used to
validate the docking protocol; see below): (i) HOH158
(HOH148), present in the oxidized enzyme in the proximity of
Lys300 and the N5 atom of FAD and constituting the so-called
Lys300/water/flavin N5 motif; (ii) HOH253 (HOH247) and
HOH94 (HOH84), found at the entrances of the catalytic site
in the proximity of Asn437 and Tyr165, respectively, and (iii)
HOH10 (HOH9), HOH23 (HOH263), HOH232 (HOH224),
HOH264 (HOH261), HOH265 (HOH2), and HOH266
(HOH262), located in the FAD binding site. Hydrogen atoms
were added to their idealized positions using Maestro40 in such a
way to protonate all the guanidino and amino groups of Arg and
Lys residues, as well as both the N-terminal primary amino
groups and the guanidino moieties of the inhibitors, and to
deprotonate all the carboxy groups of Glu and Asp side chains,
as well as the C-terminal carboxy group.41
The atomic partial charges for the inhibitors and for the
cofactor FAD present in the catalytic tunnel of ZmMPAO were
calculated with MOPAC using the AM1 approximation.
The program Autodock was used to evaluate the binding
mode of the inhibitors and to explore their binding conforma-
tions within the ZmPAO structure. A grid point spacing of
0.375 A and 90ꢀ90ꢀ90 points were used. The grid was centered
on the macromolecule. Parameters for docking runs are re-
ported in Supporting Information.
Enzyme Activity and Inhibition Assays. ZmPAO was purified
as previously described.4 The ZmPAO activity was measured
spectrophotometrically by following the formation of a pink
adduct (ε515nm=2.6 ꢀ 104 M-1 cm-1) as the result of the H2O2-
dependent oxidation of 4-aminoantipyrine (Sigma-Aldrich)
catalyzed by horseradish peroxidase (Sigma-Aldrich) and the
subsequent condensation of oxidized 4-aminoantipyrine with
3,5-dichloro-2-hydroxybenzenesulfonic acid (Sigma-Aldrich).30
The ZmPAO activity was assayed at pH 6.5 (0.2 M sodium
phosphate buffer) and 25 °C with Spd as the substrate in the
presence or absence of test compounds. Enzyme activities were
expressed in International Units (U; 1 unit is the amount of
enzyme that catalyzes the oxidation of 1 μmol of substrate per
minute). In the enzyme assays, the ZmPAO concentration
ranged from 0.2 to 4.0 nM (i.e., [E]<[S] and [E] < [I], where
[E], [S], and [I] represent the enzyme, substrate, and inhibitor
concentrations, respectively), while the substrate concentration
ranged from 0.1Km to 10Km (where Km is the Michaelis con-
stant). Specifically, Spd concentration ranged from 0.5 to 4 10-5
M, while ZmPAO concentration ranged from 0.2 to 0.7 nM for
17 and its derivatives 18-24 and from 1 to 4 nM for agmatine
analogues 10-16. Moreover, the inhibitor concentration ranged
from 0.1Ki to 10Ki (where Ki is the apparent dissociation
equilibrium constant for the formation of the reversible en-
zyme-inhibitor complex).
A different protocol was applied for the inhibitors with more
than eight degrees of freedom. In particular, a series of separate
calculations were performed and all the results of the single
docking runs were merged and clustered with a 1.5 A rms
tolerance.
Next, on the basis of the fact that Autodock does not perform
any structural optimization and energy minimization of the
complexes found, a molecular mechanics/energy minimization
(MM/EM) approach was applied to refine the Autodock out-
put. The computational protocol applied consisted of the ap-
The initial reaction rate for the ZmPAO catalyzed conversion
of the substrate was unaffected by the ZmPAO/inhibitor in-
cubation time, prior to substrate addition. Ki values were
determined according to the Dixon graphical method.31 The
data reported are the average of values obtained in three
different experiments, each with two replicates. Standard error
for each point was less than 8%.
Colorimetric MTT Assay and Trypan Blue Exclusion Test of
Cell Viability. Human solid tumor U373-MG (glioblastoma/
astrocytoma) and cervical carcinoma HeLa cell lines were
seeded in Dulbecco’s modified Eagle’s medium (DMEM) and
10% FBS (Lonza) in 96-well plates (100 μL/well at a density of
2.5 ꢀ 103 cells/well). Cells were allowed to adhere and then
exposed to various concentrations of the test compound 19
(0.1-10 μM) for 24, 48, 72, and 96 h. MTT assay (Sigma-
Aldrich) was performed as indicated by the manufacturer’s
instructions at the end of each incubation period. The optical
density (OD) of each sample was measured by an ELISA reader
at 550 nm.37 Cell viability was calculated according to [(OD of
drug treatment)/(OD of control)] ꢀ 100. The dye exclusion test
with Trypan blue, in association with MTT assay, was used to
determine the number of U373-MG and HeLa living cells
present after treatments with the compound. Trypan blue
(Sigma-Aldrich) staining was performed according to routine
procedures. Cells of each sample were counted two times in a
Neubauer chamber, and the assay was performed in duplicate.
The dye-excluding cells were indicated as the number of viable
cells. All the experiments were performed independently at least
three times.
Computational Details. All calculations and graphical mani-
pulations were performed on Linux computers using the soft-
ware packages Autodock 3.0.538 and MacroModel 8.5.39 Atom-
ic coordinates of ZmPAO used during molecular modeling
simulations were derived from the structure of the complex
between ZmMPAO and 1722 (1.9 A resolution, RCSB Protein
Data Bank entry 1H82). To set the initial coordinates for the
docking studies, the residues belonging to chains A and B, two
N-acetyl-D-glucosamine, two R-D-fucose, and two R-D-mannose
residues present in the C chain of the crystal structure of
ZmPAO were removed and excluded from all calculations.
Similarly, the crystallographic water molecules were removed
ꢀ
plication of 100 000 steps of the Polak-Ribiere conjugate
gradients (PRCG) or until the derivative convergence was
0.05 kJ/mol. Moreover, because of the large number of atoms
in the model, to correctly optimize the ZmPAO-inhibitor
complexes obtained by docking, the following additional
constraints had to be imposed: (i) the cofactor FAD and the
oxygen atoms of the water molecules were frozen; (ii) the back-
bone of the entire protein was constrained with a force constant
of 100 kJ/mol; (iii) the inhibitor and all the amino acid side
chains were unconstrained during energy minimization to allow
for reorientation and proper hydrogen-bonding geometries and
van der Waals contacts.
In addition to docking calculations, GRID maps36 were
calculated for some atom or group probes to evaluate the
regions of best interaction between the inhibitors and the
macromolecule. Box dimensions were defined to accommodate
all the residues constituting the binding site. The used probes
were DRY (hydrophobic probe), C1d (sp2 CH aromatic or
vinyl), N1: (sp3 NH with lone pair), N2: (sp3 NH2 with a lone
pair), N3þ (sp3 amine NH3 cation), N1d (sp2 amine NH
cation), and N2d (sp2 amine NH2 cation). A value of 3 was
used for the GRID keyword NPLA, which is the number of
planes of grid points per A and determines the resolution of the
computation; the grid points were 0.333 A apart. A value of
0 was used for the GRID keyword MOVE, thus allowing lone
pairs and tautomeric hydrogen atoms to move in response to the
probe. In order to achieve the points of the minimum of
molecular interaction fields (MIFs), GRID calculations were
performed for each probe separately and using the export
GRIDKONT option. The kont files corresponding to each
MIF were then processed by means of the minim and filmap
programs (both implemented in the GRID package) that extract

4784 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15
Manetti et al.
all the points of the minimum of each MIF and retain only those
falling under a user-defined cutoff. Threshold values of 0 kcal/
mol were used for all three probes.
Supporting Information Available: Spectral and elemental
analysis data of intermediates and final compounds; details of
Dixon analysis; details of molecular modeling. This material is
available free of charge via the Internet at http://pubs.acs.org.
In a preliminary step, the reliability of the docking protocol
was checked by simulations of the binding modes of 17 and 1
and a comparison of the modeled complexes with the avail-
able 3D structures derived by X-ray crystallography (1H82 and
1H83, respectively).22 The double validation was necessary
because the inhibitors in our hands could be classified into
two different structural classes: (1) compounds with two
guanidino moieties and with many (over 10) degrees of free-
dom, structurally similar to 17 of the 1H82 complex and
(2) compounds with less than 10 torsionals, structurally
similar to 1 found in the 1H83 complex. On this basis, it was
crucial to verify that the computational protocol to be applied
to inhibitors is also able to reproduce the binding modes of both
17 and 1.
References
(1) Wallace, H. M.; Fraser, A. V.; Hughes, A. A perspective of
polyamine metabolism. Biochem. J. 2003, 376, 1–14.
(2) Seiler, N. Catabolism of polyamines. Amino Acids 2004, 26, 217–
233.
(3) Cona, A.; Rea, G.; Angelini, R.; Federico, R.; Tavladoraki, P.
Functions of amine oxidases in plant development and defence.
Trends Plant Sci. 2006, 11, 80–88.
(4) Federico, R.; Alisi, C.; Forlani, F. Properties of the polyamine
oxidase from the cell wall of maize seedlings. Phytochemistry 1989,
28, 45–46.
(5) Cervelli, M.; Cona, A.; Angelini, R.; Polticelli, F.; Federico, R.;
Mariottini, P. A barley polyamine oxidase isoform with distinct
structural features and subcellular localization. Eur. J. Biochem.
2001, 268, 3816–3830.
Structures of inhibitors were built using Maestro 3D-sketch-
er, minimized with OPLS_AA force field42 using the PRCG
algorithm (0.001 kJ/(A mol) convergence) and finally docked in
ꢁ
(6) Maiale, S. J.; Marina, M.; Sanchez, D. H.; Pieckenstain, F. L.;
Ruiz, A. O. In vitro and in vivo inhibition of plant polyamine
oxidase activity by polyamine analogues. Phytochemistry 2008, 69,
2552–2558.
3
the native protein. As a result, the experimental binding con-
formations of the reference compounds were successfully repro-
duced by applying the above-mentioned Autodock settings. In
particular, for the crystallographic complex between ZmPAO
and 17, eight docking runs were performed. All the resulting
conformers were merged and clustered (1.5 A rms). The first
ranked docked conformation belonging to the first cluster
(25 conformers) was able to reproduce the X-ray conformation
of 17 with a root-mean-square deviation (rmsd) of 0.08 A on
backbone atoms, 1.61 A on heavy atoms, and 2.56 A on all the
atomic coordinates. Moreover, a comparison between calcu-
lated and experimental affinity showed that 17 had an estimated
affinity of 33 nM versus an experimental value of 7.5 nM.
The validation step on 1,8-diaminooctane was performed in
both the charged and uncharged form of the inhibitor, since, in
the process of recognition of substrates and inhibitors by
ZmPAO active site, the protonation state of the substrate and
inhibitors within the tunnel has not yet been deterrmined
unambiguously. In this case, for both the charged and un-
charged forms of 1, only one docking computation was carried
out. When the inhibitor was docked in its protonated form, the
best conformations belonging to the two most populated clus-
ters (67 and 52 conformers, respectively) reproduced the two
alternative crystallographic poses with 0.84 and 0.84 A rmsd
calculated on all the atomic coordinates. Simulation on the
uncharged form showed that the best conformers were char-
acterized by a marked increase of the rmsd values against the
X-ray conformation of 1 oriented toward the narrow entrance of
the catalytic tunnel, while the alternative orientation was not
found by Autodock.
(7) Wu, T.; Yankovskaya, V.; McIntire, W. S. Cloning, sequencing,
and heterologous expression of the murine peroxisomal flavopro-
tein, N1-acetylated polyamine oxidase. J. Biol. Chem. 2003, 278,
20514–20525.
(8) Vujcic, S.; Diegelman, P.; Bacchi, C. J.; Kramer, D. L.; Porter, C.
W. Identification and characterization of a novel flavin-containing
spermine oxidase of mammalian cell origin. Biochem. J. 2002, 367,
665–675.
(9) Landry, J.; Sternglanz, R. Yeast Fms1 is
a FAD-utilizing
polyamine oxidase. Biochem. Biophys. Res. Commun. 2003, 303,
771–776.
(10) Tavladoraki, P.; Rossi, M. N.; Saccuti, G.; Perez-Amador, M. A.;
Polticelli, F.; Angelini, R.; Federico, R. Heterologous expression
and biochemical characterization of a polyamine oxidase from
Arabidopsis involved in polyamine back conversion. Plant Physiol.
2006, 141, 1519–1532.
(11) Moschou, P. N.; Sanmartin, M.; Andriopoulou, A. H.; Rojo, E.;
Sanchez-Serrano, J. J.; Roubelakis-Angelakis, K. A. Bridging the
gap between plant and mammalian polyamine catabolism: a novel
peroxisomal polyamine oxidase responsible for a full back-conver-
sion pathway in Arabidopsis. Plant Physiol. 2008, 147, 1845–1857.
(12) Seiler, N. Pharmacological aspects of cytotoxic polyamine analogs
and derivatives for cancer therapy. Pharmacol. Ther. 2005, 107,
99–119.
(13) Ha, H. C.; Woster, P. M.; Yage, J. D.; Casero, R. A. The role of
polyamine catabolism in polyamine analogue-induced pro-
grammed cell death. Proc. Natl. Acad. Sci. U.S.A. 1997, 94,
11557–11562.
(14) Hu, R. H.; Pegg, A. E. Rapid induction of apoptosis by deregulated
uptake of polyamine analogues. Biochem. J. 1997, 328, 307–316.
(15) Devereux, W.; Wang, Y.; Stewart, T. M.; Hacker, A.; Smith, R.;
Frydmann, B.; Valasinas, A. L.; Reddy, V. K.; Marton, L. J.;
Ward, T. D.; Woster, P. M.; Casero, R. A. Induction of the
PAOh1/SMO polyamine oxidase by polyamine analogues in hu-
man lung carcinoma cells. Cancer Chemother. Pharmacol. 2003, 52,
383–390.
Although in several cases Autodock may be unable to cor-
rectly locate ligands within binding sites, results from simula-
tions on 17 and a comparison with the corresponding X-ray
structure led us to assume the docking protocol as a reliable
procedure to be applied to the new inhibitors belonging to the
class of iminoctadine derivatives. Moreover, Autodock calcula-
tions on N-prenylagmatine derivatives gave results, in terms of
orientations and interactions of the ligands, in good agreement
with those found by means of a pseudo Monte Carlo statistical
conformational analysis previously performed on the same
compounds.25
Molecular Dynamics (MD) Simulations. The complex be-
tween 19 and ZmPAO, as derived from docking simulations
and energy minimization, was submitted to a MD protocol
using MacroModel and the OPLS 200142 force field. GB/SA,
with water parameters, was used to simulate the solvent. The
starting structure was relaxed by performing 20 ps of MD at
310 K (equilibration phase). Next, the system was submitted to
1 ns of MD simulations (310 K), extracting a snapshot every
10.0 ps. Dynamics time step was set to 1.5 fs.
ꢀ
(16) Chaturvedi, R.; Cheng, Y.; Asim, M.; Bussiere, F. I.; Xu, H.;
Gobert, A. P.; Hacker, A.; Casero, R. A.Jr.; Wilson, K. T.
Induction of polyamine oxidase 1 by Helicobacter pylori causes
macrophage apoptosis by hydrogen peroxide release and mito-
chondrial membrane depolarization. J. Biol. Chem. 2004, 279,
40161–40173.
ꢀ
(17) Xu, H.; Chaturvedi, R.; Cheng, Y.; Bussiere, F. I.; Asim, M.; Yao,
M. D.; Potosky, D.; Meltzer, S. J.; Rhee, J. G.; Kim, S. S.; Moss,
S. F.; Hacker, A.; Wang, Y.; Casero, R. A.Jr.; Wilson, K. T.
Spermine oxidation induced by Helicobacter pylori results in
apoptosis and DNA damage: implications for gastric carcinogen-
esis. Cancer Res. 2004, 64, 8521–8525.
(18) Babbar, N.; Casero, R. A.Jr. Tumor necrosis factor-R increases
reactive oxygen species by inducing spermine oxidase in human
lung epithelial cells: a potential mechanism for inflammation-
induced carcinogenesis. Cancer Res. 2006, 66, 11125–11130.
(19) Yoda, H.; Hiroi, Y.; Sano, H. Polyamine oxidase is one of the key
elements for oxidative burst to induce programmed cell death in
tobacco cultured cells. Plant Physiol. 2006, 142, 193–206.
(20) Wood, P. L.; Khan, M. A.; Kulow, S. R.; Mahmood, S. A.;
Moskal, J. R. Neurotoxocity of reactive aldehydes: the concept

Article
of “aldehyde load” as demonstrated by neuroprotection with
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15 4785
(31) Dixon, M. The determination of enzyme inhibitor constants.
Biochem. J. 1953, 55, 170–171.
hydroxylamines. Brain Res. 2006, 1095, 190–199.
(21) Binda, C.; Coda, A.; Angelini, R.; Federico, R.; Ascenzi, P.;
Mattevi, A. A 30 A long U-shaped catalytic tunnel in the crystal
structure of polyamine oxidase. Structure 1999, 7, 265–276.
(22) Binda, C.; Angelini, R.; Federico, R.; Ascenzi, P.; Mattevi, A.
Structural bases for inhibition binding and catalysis in polyamine
oxidase. Biochemistry 2001, 40, 2766–2776.
(23) Cervelli, M.; Polticelli, F.; Federico, R.; Mariottini, P. Heterolo-
gous expression and characterization of mouse spermine oxidase.
J. Biol. Chem. 2003, 278, 5271–5276.
(32) QikProp, version 3.0; Schrodinger, LLC: New York, 2007.
(33) Artursson, P.; Karlsson, J. Correlation between oral drug absorp-
tion in humans and apparent drug permeability coefficients in
human intestinal epithelial (Caco-2) cells. Biochem. Biophys. Res.
Commun. 1991, 175, 880–885.
(34) Casero, R. A.Jr.; Mank, A. R.; Xiao, L.; Smith, J.; Bergeron, R. J.;
Celano, P. Steady-state messenger RNA and activity correlates
with sensitivity to N1,N12-bis(ethyl)spermine in human cell lines
representing the major forms of lung cancer. Cancer Res. 1992, 52,
5359–5363.
(35) Huang, Q.; Liu, Q.; Hao, Q. Crystal structures of Fms1 and its
complex with spermine reveal substrate specificity. J. Mol. Biol.
2005, 348, 951–959.
(36) GRID, version 22; Molecular Discovery, Ltd.: Marsh Road, Pinner,
Middlesex, U.K.
(37) Mosmann, T. Rapid colorimetric assay for cellular growth and
survival: application to proliferation and cytotoxicity assays.
J. Immunol. Methods 1983, 65, 55–63.
(38) Morris, G. M.; Goodsell, D. S.; Halliday, R. S.; Huey, R.; Hart, W.
E.; Belew, R. K.; Olson, A. J. Automated docking using a
Lamarckian genetic algorithm and an empirical binding free energy
function. J. Comput. Chem. 1998, 19, 1639–1662.
(39) Mohamadi, F.; Richards, N. G. J.; Guida, W. C.; Liskamp, R.;
Lipton, M.; Caufield, C.; Chang, G.; Hendrickson, T.; Still, W. C.
An integrated software system for modeling organic and bioor-
ganic molecules using molecular mechanics. J. Comput. Chem.
1990, 11, 440–467.
(24) Bianchi, M.; Polticelli, F.; Ascenzi, P.; Botta, M.; Federico, R.;
Mariottini, P.; Cona, A. Inhibition of polyamine and spermine
oxidases by polyamine analogues. FEBS J. 2006, 273, 1115–1123.
(25) Cona, A.; Manetti, F.; Leone, R.; Corelli, F.; Tavladoraki, P.;
Polticelli, F.; Botta, M. Molecular basis for the binding of compe-
titive inhibitors of maize polyamine oxidase. Biochemistry 2004, 43,
3426–3435.
ꢂ
ꢂ
ꢁ
ꢁ
(26) Stranska, J.; Sebela, M.; Tarkowski, P.; Rehulka, P.; Chmelık, J.;
´
ꢂ
Popa, I.; Pec, P. Inhibition of plant amine oxidase by a novel series
of diamine derivatives. Biochimie 2007, 89, 135–144.
(27) Federico, R.; Leone, L.; Botta, M.; Binda, C.; Angelini, R.;
Venturini, G.; Ascenzi, P. Inhibitor of pig liver and Zea mays L.
polyamine oxidase: a comparative study. J. Enzyme Inhib. 2001, 13,
465–471.
(28) Corelli, F.; Federico, R.; Cona, A.; Venturini, G.; Schenone, S.;
Botta, M. Solution and solid-phase synthesis of aminoalkylguani-
dines inhibiting polyamine oxidase and nitric oxide synthase. Med.
Chem. Res. 2002, 11, 309–321.
(29) Dreassi, E.; Zizzari, A. T.; D’Arezzo, S.; Visca, P.; Botta, M.
Analysis of guazatine mixture by LC and LC-MS and antimycotic
activity determination of principal components. J. Pharm. Biomed.
Anal. 2007, 43, 1499–1506.
(30) Artiss, J. D.; Entwistle, W. M. The application of a sensitive
uricase-peroxidase couple reaction to a centrifugal fast analy-
ser for the determination of uric acid. Clin. Chim. Acta 1981, 116,
301–309.
(40) Maestro, version 7.5; Schroedinger: New York, 2006.
(41) Dewar, M. J. S.; Zoebisch, E. G.; Healy, E. F.; Stewart, J. J. P.
AM1: a new general purpose quantum mechanical molecular
model. J. Am. Chem. Soc. 1985, 107, 3902–3909.
(42) Jorgensen, W. L.; Maxwell, D. S.; Tirado-Rives, J. Development
and testing of the OPLS all-atom force field on conformational
energetics and properties of organic liquids. J. Am. Chem. Soc.
1996, 118, 11225–11236.