Synthesis, biological evaluation, and three-dimensional quantitative structure-activity relationship study of small-molecule positive modulators of adrenomedullin

Article abstract of DOI:10.1021/jm050021+

Adrenomedullin (AM) is a peptide hormone implicated in blood pressure regulation and in the pathophysiology of several diseases such as hypertension, cancer, diabetes, and renal disorders, becoming an interesting new target for the development of drugs. I

Full text of DOI:10.1021/jm050021+

4068
J. Med. Chem. 2005, 48, 4068-4075
Synthesis, Biological Evaluation, and Three-Dimensional Quantitative
Structure-Activity Relationship Study of Small-Molecule Positive Modulators
of Adrenomedullin
Mario A. Garc´ıa,† Sonsoles Mart´ın-Santamar´ıa,† Mo´nica Cacho,†,‡ Fernando Moreno de la Llave,† Miguel Julia´n,†
Alfredo Mart´ınez,§ Beatriz de Pascual-Teresa,*,† and Ana Ramos*,†
Departamento de Quı´mica, Facultad de Farmacia, Universidad San Pablo CEU, Urbanizacio´n Monteprı´ncipe, 28668 Boadilla
del Monte, Madrid, Spain, and Departamento de Neuroanatomı´a y Biologı´a Celular, Instituto Cajal, Av. Doctor Arce, 37,
28002 Madrid, Spain
Received January 10, 2005
Adrenomedullin (AM) is a peptide hormone implicated in blood pressure regulation and in the
pathophysiology of several diseases such as hypertension, cancer, diabetes, and renal disorders,
becoming an interesting new target for the development of drugs. In a recent high-throughput
screening study, a positive modulator with a bistriazole structure has been identified.1 In this
work, a new series of structurally related compounds has been synthesized by reaction of
phenoxyacetic acid with the corresponding dihydrazide, followed by treatment of the formed
bisoxadiazoles with benzylamine. The affinity toward AM of the lead compound, and a
structurally related family obtained from the small-molecule NCI library together with the
synthesized series, has been determined. A three-dimensional quantitative structure-activity
relationship (3D-QSAR) study and conformational and molecular dynamics simulations have
shown that the presence of a free NH and a phenyl group is essential for the interaction of
these compounds with AM.
Introduction
NCI has collected since 1955. The first phase involves
a screening to search compounds that inhibit the
binding between AM and its monoclonal antibody. All
the compounds that gave a positive response to this
assay were subjected to a secondary screening that
consisted of an analysis of their ability to modify the
production of cAMP, a second messenger elicited by the
specific receptor system. This assay allowed classifying
the molecules in positive or negative modulators, de-
pending on their ability to elevate or reduce cAMP
levels, respectively.
Further assays characterized the physiological impact
of selected small molecules in AM-mediated responses.
Validation of the biological activity of those compounds
was carried out, both as potential antiproliferative
agents for the negative modulators and as hypotensive
agents for the positive modulators.1
Adrenomedullin (AM) is a 52 amino acid peptide that
belongs to the calcitonin/calcitonin gene-related peptide
(CGRP)/amylin/AM superfamily and was isolated in
1993 from a human pheochromocytoma (adrenal tu-
mor).2 In humans, it is expressed by many cell types
and exerts a variety of physiological roles, including
vasodilatation, bronchodilatation, hormone regulation,
antimicrobial activity, and growth regulation.3,4 Exten-
sive reviews have recently been published that sum-
marize the state of the art in our collective knowledge
on AM science.5
AM levels are dysregulated in many human patholo-
gies such as hypertension, heart failure, sepsis, cancer,
and diabetes.4 Several studies have demonstrated that
changes in AM levels have opposite effects depending
on the particular disease studied. Thus, while AM is a
protective agent against cardiovascular disorders,6 it
behaves as a stimulating factor in other pathologies
such as cancer7 and diabetes.8 Therefore, AM is a new
and promising target in the development of molecules
that, through their ability to positively or negatively
regulate AM activity, could be used in the treatment of
these pathologies.
Results and Discussion
One of the positive modulators that showed promising
behavior by both screening steps mentioned above was
bistriazole 1a. Structurally related chemical family
members such as 1b-i were also evaluated, and in most
cases they maintained the activity or even showed
stronger activity (1h) than the lead compound. In this
work the synthesis of analogues 1j-n and the study of
the structure-activity relationship using computational
techniques of compounds 1a-n have been carried out,
with the aim of determining the structural requirements
for efficient interaction with AM.
The only attempt in this area has been carried out
recently by Mart´ınez et al.1 This article describes a new
and efficient method to detect nonpeptidic modulators
of AM from a large library of small molecules that the
* To whom the correspondence should be addressed. For B.d.P.-T.:
phone, 34913724796; fax, 34913510475; e-mail, bpaster@ceu.es. For
A.R.: phone, 34913724782; fax, 34913510475; e-mail, aramgon@ceu.es.
† Universidad San Pablo CEU.
Chemistry. The synthesis of bistriazoles 1j-n was
carried out following the route outlined in Scheme 1.
Thus, phenoxyacetic acid was treated with the corre-
‡ Current address: Instituto de Qu´ımica Me´dica, C/ Juan de la
Cierva, 3, 28006 Madrid, Spain.
§ Instituto Cajal.
10.1021/jm050021+ CCC: $30.25 © 2005 American Chemical Society
Published on Web 05/24/2005

Modulators of Adrenomedullin
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 12 4069
Table 1. Positive Modulators of AMa
Scheme 1. Synthesis of Series 2, 3, and 1j-na
a Reagents: (a) (i) phenoxyacetic acid and POCl3 in CH3CN, ∆;
(ii) concentrated NH3; (b) (i) POCl3 in CH3CN, ∆; (ii) concentrated
NH3; (c) PhCH2NH2, ∆.
sponding commercially available dihydrazides in the
presence of POCl3 to yield compounds 2j-n. Further
refluxing of these with POCl3 in CH3CN for 7 h brought
about the formation of bisoxadiazoles 3j-n. Attempts
to undergo the reaction in one step, following the
reaction conditions described by Kudari et al.9 for the
synthesis of bisisoxazoles (series 3) and other analogue
compounds, led to the isolation of the open-chain bishy-
drazides. Finally, reaction of bisoxadiazoles 3j-n with
benzylamine yielded the desired bistriazoles 1j-n.
Biological Activity. The affinity of the different
compounds for AM was quantified by measuring the
interference of the compounds in the binding between
the peptide and its blocking monoclonal antibody. A
neutralizing monoclonal antibody will bind to an epitope
on the peptide that is critical for receptor recognition.
Thus, molecules that disrupt peptide-antibody binding
may be good candidates as modulators of peptide
physiology. Internal controls were placed in every plate
as described in Experimental Section (wells without any
coating, wells where no potential modulators were
added, and wells with a positive inhibition control). The
affinity was quantified by comparison with the internal
controls (Table 1).
a See Experimental Section for compounds 1j-n, 2, and 3.
Compounds 1a-i belong to the NCI’s small-molecule repository.
QSAR Analysis. In cases such as this, in which the
target receptor structure is not available, structure-
activity relationship (SAR) analysis can often help to
elucidate the essential requirements of a potential
ligand to interact favorably with its target and indirectly
to provide insight into the features that describe the
binding site. In this study, a three-dimensional quan-
titative structure-activity relationship (3D-QSAR) model
has been derived for a series of AM ligands, utilizing
molecular interaction fields (MIF) calculated in the well-
known methodology GRID10 and the subsequent analy-
sis of these fields by use of alignment-independent
descriptors in ALMOND.11 MIF identify regions where
Seven compounds (1a,c-h), all of them bearing an
NH group, significantly inhibited binding of the anti-
body, suggesting a high degree of affinity for AM.
However, compounds 1j-n, where the NH group has
been replaced by a methylene group, significantly
decreased their AM affinity when compared to com-
pounds with an NH group. In addition, all synthetic
intermediates (2 and 3) were tested and showed poor
affinity values.

4070 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 12
Garcı´a et al.
Figure 2. Loading plot (which shows the pattern of descrip-
tors) of the first (PC1) and the second (PC2) principal compo-
nents for the 3D-QSAR model. Energy descriptors are col-
ored: (11) DRY-DRY; (22) O-O; (33) N1-N1; (44) TIP-TIP;
(12) DRY-O; (13) DRY-N1; (14) DRY-TIP; (23) O-N1; (24)
O-TIP; (34) N1-TIP.
rate compounds belonging to the different series in tight
clusters. Furthermore, series 1a-i is split into two
groups: compounds with R′ ) NHPh (1a, 1f, 1g, 1h,
and 1i) with high affinity with the exception of 1i;
compounds with R′ ) NH2 (1b, 1d, and 1e) with
moderate affinity. Compound 1c was not considered at
this point to be used later as external set.
Figure 1. Score plot (which shows the pattern of compounds)
of the first (PC1) and the second (PC2) principal components
for the 3D-QSAR model.
the ligand can interact favorably with certain chemical
groups of a potential receptor, defining a “virtual
receptor site” (VRS).12 On one hand, this approach can
assist in the knowledge of the chemical aspects required
to bind AM and in the further design of more potent
analogues. On the other hand, it can contribute to the
scrutiny of the anchoring points that compose the
binding site.
The chosen probes were DRY, N1, and O to represent
important groups of the peptide AM. The DRY probe
represents hydrophobic interactions, the O probe (car-
bonyl oxygen) represents hydrogen bond acceptor groups,
and the N1 probe (amide nitrogen) represents hydrogen
bond donor groups. Shape complementarity was also
integrated by means of the molecular shape field (TIP
“probe”). The compounds show three types of chemical
modifications: the length of the aliphatic chain linking
the rings (series 1j-n, 2, and 3), the nature of the ring
(1,3,4-oxadiazole or 1,2,4-triazole), and the substituents
R and R′ (Table 1). Compounds 2 and 3 were obtained
as intermediates in the synthesis of compounds 1j-n
and were also included in the model. Ten correlograms
of 92 variables each were obtained, thus producing a
matrix of 920 variables and 23 objects.
The model was able to highlight the structural dif-
ferences between the compounds as is shown in the
principal component analysis (PCA) (Figure 1) where
compounds are grouped by their chemical structure. The
first and second principal components (PCs) explained
96% of the variance of the original descriptors. The first
PC extracted, with major contributions from GRIND
variables involving the carbonyl oxygen O probe (O-O,
DRY-O, O-N1, and O-TIP (Figure 2)), clearly distin-
guishes the compounds bearing the NH group (series
1) from the others (series 2 and 3). This can be
rationalized in terms of the strong interactions that NH
groups can establish with the O probe. The second PC,
which is mostly made up by the GRIND variables
involving the hydrophobic DRY probe, is able to sepa-
The partial least squares (PLS) analysis of the
original matrix resulted in a three-component model
with an r2 of 0.93. The cross-validation of the model
produced a small positive q2 value (q2 ) 0.41). A
preliminary inspection of the plot representing predicted
vs experimental affinities allowed us to identify com-
pound 1i as an outlier. With the exclusion of compound
1i, the PLS that was performed on the 22 objects matrix
produced a good predictive model, with an r2 of 0.97 and
a cross-validated q2 of 0.81, for four PCs. Inclusion and
exclusion of series 2 and 3 in the model were carefully
studied. The predictive ability of the model was slightly
improved when considering the whole data, showing
that the enrichment in chemical information was rel-
evant to the predictive quality of the model. To simplify
the interpretation of the model, the auto- and cross-
correlograms that showed low contribution were re-
moved. After the removal of these correlograms, the
model contained O-O, DRY-O, O-N1, and O-TIP
correlograms, producing a matrix of 368 variables. The
PLS of this matrix produced a model of four PCs (r2 )
0.95, q2 ) 0.84, q2LOO ) 0.87). The model was externally
evaluated with compound 1c (Table 1). The predicted
affinity values were very similar to the experimentally
measured ones. Figure 3 shows a scatter plot of the
calculated vs experimental affinities, including also the
affinity prediction.
The inspection of the PLS coefficients showed that all
the interactions were positively correlated with affinity
(Figure 4). To interpret these results, the correlograms
were studied. The O-O autocorrelogram clearly shows
that the intensity of the interaction is correlated with
the affinity (Figure 5). The variables represent pairs of
nodes at different distances where the O probe has
favorably interacted. The greatest interactions (red and
pink colors, Figure 5) were exhibited by compounds
belonging to series 1a-i, which hold a free NH group

Modulators of Adrenomedullin
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 12 4071
Figure 3. Plot of recalculated versus experimental affinities
in the 3D-QSAR model, showing the prediction for compound
1c (9).
Figure 5. Set of 23 superimposed correlograms, correspond-
ing to the DRY-O correlogram, representing interactions for
ligand hydrogen bond donor and hydrophobic regions. Every
point in the correlogram represents the product of two
particular nodes for a certain compound. Points are color-coded
according to the biological activity of the corresponding
compounds; active compounds are in red, intermediate in
white, and inactive in blue. A simple visual inspection shows
that the strength of the interaction is positively correlated with
the biological affinity.
Figure 4. PLS pseudocoefficients histogram for the 3D-QSAR
model represented.
that might act as hydrogen bond donor in the interaction
with the receptor. Effectively, compounds 1a-i showed
an interaction defined by a pair of nodes between both
NH groups. Thus, compounds such 1n, which could be
considered as a potential bioisoster of the lead compound
1a, did not show significant affinity. This finding might
suggest that for this family of compounds the NH group
is a relevant group for the binding to AM through
formation of a hydrogen bond with a residue of the
binding pocket.
Other interactions that were shown to be important
were those involving DRY probe (DRY-O correlogram)
and the molecular shape field (O-TIP correlogram). The
MIF corresponding to those nodes coincided around the
substituents of type R (Figure 6). This result could
indicate that the presence of aromatic rings, which
establish van der Waals interactions with the receptor,
is favorable for the binding and, furthermore, that those
rings must have suitable convexity and shape. Com-
pounds 1a, 1g, and 1h, which presented the highest
affinity, also exhibited intense interactions with DRY
and TIP probes surrounding the phenyl at the NH
group. This could highlight the presence of a secondary
hydrophobic pocket, not essential for AM binding but
that could be playing a relevant role in order to enhance
the affinity toward AM.
Figure 6. Interactions present in compound 1a. The fields
represent interactions of the probe DRY (white) and O (red).
Conformational and Molecular Dynamics Stud-
ies of Compounds 1h and 1i. Despite the structural
similarity with 1h, compound 1i showed a low affinity
value for AM and behaved as an outlier in the QSAR
analysis. On the other hand, the other naphthol isomers
1d and 1e did not exhibit different behavior as AM
modulators, and the QSAR model was able to fit them
adequately. It is worth emphasizing that while 1h and
1i bear a NH substituted by a phenyl group, 1d and 1e
have a free NH2, so the rotational freedom is higher and
there is no steric hindrance. To rationalize the different
affinity shown by isomers 1h and 1i, a thorough
conformational study was carried out.
First, a Monte Carlo based conformational search was
carried out on the four naphthol derivatives (1d, 1e, 1h,
and 1i), using the program MacroModel 7.2.13 Of 1000

4072 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 12
Garcı´a et al.
Figure 7. MD trajectories of conformers 1h.1, 1i.1, and 1i.2: NH‚‚‚O(naphthol) and NH‚‚‚O(ether) distances. Allowed maximum
H‚‚‚acceptor distance (3.20 Å) is represented as a dotted line.
Monte Carlo pushes of compounds 1h and 1i, 454 and
71 unique conformations were found, respectively,
within an energetic range of 67-79 kcal mol-1 for 1h
and 87-99 kcal mol-1 for 1i. The five energetically
lowest conformations for both isomers were inspected,
and all possible intramolecular bonds were examined
by means of monitoring the distances between oxygen
and nitrogen atoms and between polar hydrogens. For
every conformer of compound 1i, eight intramolecular
hydrogen bonds were observed, while each conformer
of compound 1h presented only two hydrogen bonds. In
many of these bonds, the NH group, whose relevance
to the rationalization of the affinity was shown by the
3D-QSAR, was involved as hydrogen bond donor. The
conformational search on the other naphthol deriva-
tives, 1d and 1e, did not show any important differences
(data not shown) between both isomers, as expected.
The stability of the NH-acceptor interactions was
checked by performing unrestrained molecular dynam-
ics (MD) simulations in water for compounds 1h and
1i, using the AMBER program.14 The initial geometries
were the energetically lowest conformations, and the
following ones were under 1.5% of the energetic mini-
mum value. This criterion led to the selection of the
lowest energy conformation for compound 1h (1h.1) and
two conformers for 1i, the lowest energy conformer (1i.1)
and the conformer 1i.2, which was 0.13 kcal mol-1
higher than 1i.1. Bond distances and angles, where
oxygen, nitrogen, and polar hydrogen atoms are in-
volved, were monitored. Figure 7 shows selected results
for bonds involving NH hydrogens. The study of these
data revealed that 1h.1 holds a stable intramolecular
hydrogen bond along the trajectory of the MD simula-
tion between the hydroxyl group and the neighboring
ether oxygen. However, NH hydrogen did not take part
in any stable bond involving hydroxyl oxygen or ether
oxygen (magenta color, Figure 7). The other atoms
studied did not exhibit bond distance and angle values
corresponding to intramolecular hydrogen bonds. On the
contrary, both 1i conformers presented several intramo-
lecular bonds, with those involving NH hydrogens being
of special interest. One of the NH presented bonds to
hydroxyl oxygen or ether oxygen or both during most
of the MD simulation (Figure 7, 1i.1 conformer, brown
and cyan colors; Figure 7, 1i.2 conformer, green and red
colors). Bond angle values were in agreement with
stable hydrogen bond geometries.
These findings reveal that compound 1i presents
intramolecular bonds involving atoms that are neces-
sary to establish interactions with AM, as NH hydrogen.
Accordingly, these interactions might prevent NH from
being available to interact with the receptor and would
explain the different affinity between 1.h and 1.i. Also,
these results would be in agreement with the 3D-QSAR
analysis, which pointed out the relevance of a free NH
to interact with the receptor.
Conclusions
The synthesis, biological evaluation, and 3D-QSAR
analysis of a series of positive modulators of AM have
been carried out. This research has allowed us to figure
out some key aspects of the structural requirements to
bind AM, a new and promising target for the develop-
ment of new molecules with several pharmacological
activities. In this particular case, the derived 3D-QSAR
model has been able to predict the affinity with a high
predictive quality (q2LOO ) 0.87), as well as to highlight
two relevant features for the binding to AM: the
presence of a free NH, which is able to establish a
hydrogen bond with the receptor, and a phenyl group,
which is able to establish hydrophobic interactions.
Additional conformational and MD studies have shown
the presence of intramolecular bonds in compound 1i
that could compromise this crucial NH group, thus

Modulators of Adrenomedullin
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 12 4073
N,N′-Bis(phenoxyacetyl)sebacic Dihydrazide (2n). From
sebacic dihydrazide (9.80 g, 42.61 mmol), phenoxyacetic acid
(12.95 g, 85.22 mmol), and POCl3 (52.32 g, 340.85 mmol) in
CH3CN (270 mL), 2n (4.99 g, 25%) was produced as a white
solid, mp 188-190 °C (EtOH).
General Procedure for the Preparation of Bisoxadia-
zoles 3j-n. To a stirred suspension of the corresponding
dihydrazide (1 equiv) in CH3CN was added a solution of POCl3
(8 equiv) in CH3CN dropwise over 30 min at room temperature.
The resulting mixture was heated at reflux temperature for 7
h. Then, the crude was cooled to room temperature and the
solvent was removed under reduced pressure. The obtained
residue was poured into ice-water and neutralized with
concentrated ammonium. After the mixture was stirred for one
night at room temperature, the resulting precipitate was
filtered off and purified by flash column chromatography on
sicila gel to give the corresponding oxadiazole.
justifying its behavior as outlier in the 3D-QSAR model
and the unexpected low affinity of this compound.
The SAR methodology employed has proved to be a
useful tool to elucidate some of the structural require-
ments to bind AM and to predict affinity. Similar work
on negative modulators of AM is in progress in our
laboratory.
Experimental Section
Small-Molecule Library. Compounds 1a-i were obtained
from the NCI’s small-molecule repository. The construction of
the library has been described15 and can be viewed at http://
cactus.nci.nih.gov/ncidb2. All compounds were provided diluted
in dimethyl sulfoxide (DMSO). NCI codes are 697165 for 1a,
697161 for 1b, 697164 for 1c, 697162 for 1d, 697163 for 1e,
697166 for 1f, 697169 for 1g, 697167 for 1h, and 697168 for
1i.
1,4-Bis(5-phenoxymethyl[1,3,4]oxadiazol-2-yl)butane
(3j). From 2j (2.50 g, 5.66 mmol) and POCl3 (6.95 g, 45.28
mmol) in CH3CN (35 mL), a white solid was produced. Flash
chromatography of the crude product using hexane-AcOEt
(1:1) as eluent gave 2j (1.75 g, 76%), mp 75-78 °C (EtOH). IR
General Methods. Melting points (uncorrected) were
determined on a Stuart Scientific SMP3 apparatus. Infrared
(IR) spectra were recorded with a Perkin-Elmer 1330 infrared
spectrophotometer. 1H and 13C NMR data were recorded on a
Bruker 300-AC instrument. Chemical shifts (δ) are expressed
in parts per million relative to internal tetramethylsilane;
coupling constants (J) are in hertz. Mass spectra were run on
a Bruker Esquire 3000 spectrometer. Elemental analyses (C,
H, N) were performed on a Perkin-Elmer 2400 CHN apparatus
at the Microanalyses Service of the University Complutense
of Madrid. Unless otherwise stated, all reported values are
within (0.4% of the theoretical compositions. Thin-layer
chromatography (TLC) was run on Merck silica gel 60 F-254
plates. Unless stated otherwise, starting materials used were
1
(KBr): 1600, 1570 cm-1. H NMR (CDCl3): δ 1.93 (4H, br s,
2CH2), 2.93 (4H, br s, 2CH2CNO), 5.24 (4H, s, 2CH2O), 7.02
(6H, m, ArH), 7.31 (4H, m, ArH). 13C NMR (CDCl3): δ 24.9,
25.4, 59.8, 114.8, 122.2, 129.7, 157.5, 162.6, 167.4. MS (ESI):
m/z 407 [M + H]+. Anal. (C22H22N4O4) C, H, N.
1,5-Bis(5-phenoxymethyl[1,3,4]oxadiazol-2-yl)pen-
tane (3k). From 2k (1.32 g, 2.89 mmol) and POCl3 (3.55 g,
23.13 mmol) in CH3CN (20 mL), a solid was produced. Flash
chromatography of the crude product using CH2Cl2-MeOH
(99:1) as eluent gave 3k (716 mg, 59%), mp 71-72 °C (EtOH).
1,6-Bis(5-phenoxymethyl[1,3,4]oxadiazol-2-yl)hex-
ane (3l). From 2l (3.60 g, 7.66 mmol) and POCl3 (9.40 g, 61.24
mmol) in CH3CN (50 mL), a white solid was produced. Flash
chromatography of the crude product using hexane-AcOEt
(1:1) as eluent gave 3l (920 mg, 28%), mp 57-60 °C (EtOH).
1,7-Bis(5-phenoxymethyl[1,3,4]oxadiazol-2-yl)hep-
tane (3m). From 2m (2.00 g, 4.13 mmol) and POCl3 (5.07 g,
33.03 mmol) in CH3CN (25 mL), an oil was produced. Flash
chromatography of the crude product using hexane-AcOEt
(1:1) as eluent gave 3m (1.17 g, 63%) as an oil, which
crystalized on standing to form a white solid, mp 48-52 °C
(EtOH).
1,8-Bis(5-phenoxymethyl[1,3,4]oxadiazol-2-yl)octane
(3n). From 2n (1.30 g, 2.61 mmol) and POCl3 (3.20 g, 20.85
mmol) in CH3CN (16 mL), a white solid was produced. Flash
chromatography of the crude product using toluene-AcOEt
(8:2) as eluent gave 3n (844 mg, 70%), mp 95-99 °C (EtOH).
General Procedure for the Preparation of Bistriazoles
1j-n. A mixture of the corresponding bisoxadiazole and an
excess of benzylamine was heated at 150 °C for 30 h under
argon. The reaction mixture was cooled to room temperature,
and CH2Cl2 (25 mL) was added. The solution was washed with
10% aqueous HCl, saturated aqueous NaHCO3, and brine. The
extract was dried (MgSO4), filtered, and evaporated to dryness,
and the obtained crude product was purified by flash column
chromatography on silica.
1
high-grade commercial products. Complete H and 13C NMR,
IR, MS, and elemental analysis data are given in the Sup-
porting Information.
General Procedure for the Preparation of Dihy-
drazides 2j-n. To a stirred suspension of either adipic,
pimelic, suberic, azelaic, or sebacic dihydrazide (1 equiv) and
phenoxyacetic acid (2 equiv) in CH3CN was added a solution
of POCl3 (8 equiv) in CH3CN dropwise over 30 min at room
temperature. The reaction mixture was heated at reflux
temperature until the solution became clear and then for 30
additional minutes. Then, the crude was cooled to room
temperature and the solvent was removed under reduced
pressure. The obtained residue was poured into ice-water and
neutralized with concentrated ammonium. After the mixture
was stirred for one night at room temperature, the resulting
precipitate was filtered off and recrystallized from absolute
EtOH.
N,N′-Bis(phenoxyacetyl)adipic Dihydrazide (2j). From
adipic dihydrazide (5.00 g, 28.73 mmol), phenoxyacetic acid
(8.73 g, 57.46 mmol), and POCl3 (35.30 g, 229.97 mmol) in CH3-
CN (70 mL), 2j (2.00 g, 16%) was produced as a white solid,
mp 237-239 °C (EtOH). IR (KBr): 3420, 3250, 1700, 1660
1
cm-1. H NMR (DMSO-d6): δ 1.54 (4H, br s, 2CH2), 2.50 (4H,
br s, 2CH2CO), 4.58 (4H, s, 2OCH2CO), 6.97 (6H, m, ArH),
7.30 (4H, t, J ) 7.9, ArH), 9.83 (2H, s, 2NH), 10.08 (2H, s,
2NH). 13C NMR (DMSO-d6): δ 24.6, 32.9, 65.9, 114.7, 121.2,
129.4, 157.7, 166.6, 170.9. MS (ESI): m/z 465 [M + Na]+.
N,N′-Bis(phenoxyacetyl)pimelic Dihydrazide (2k). From
pimelic dihydrazide (14.53 g, 77.29 mmol), phenoxyacetic acid
(23.50 g, 154.60 mmol), and POCl3 (94.90 g, 618.24 mmol) in
CH3CN (180 mL), 2k (8.10 g, 23%) was produced as a white
solid, mp 198-200 °C (EtOH).
1,4-Bis(4-benzyl-5-phenoxymethyl[1,2,4]triazol-3-yl)-
butane (1j). From 3j (200 mg, 0.49 mmol) and benzylamine
(3 mL, 27.48 mmol) and after flash chromatography of the
crude product using CHCl3-MeOH (95:5) as eluent, 1j (84 mg,
30%) was produced as a white solid, mp 190-192 °C (EtOH).
1
IR (KBr): 1600, 1590 cm-1. H NMR (CDCl3): δ 1.72 (4H, br
s, 2CH2), 2.53 (4H, br s, 2CH2CN2), 5.08 (4H, s, 2CH2N), 5.13
(4H, s, 2CH2O), 6.86 (10H, m, ArH), 7.20 (10H, m, ArH). 13C
NMR (CDCl3): δ 24.5, 26.1, 47.0, 60.6, 114.6, 121.8, 126.3,
128.3, 129.1, 129.6, 134.6, 150.8, 155.9, 157.5. MS (ESI): m/z
585 [M + H]+. Anal. (C36H36N6O2) C, H, N.
1,5-Bis(4-benzyl-5-phenoxymethyl[1,2,4]triazol-3-yl)-
pentane (1k). From 3k (100 mg, 0.24 mmol) and benzylamine
(1 mL, 9.16 mmol) and after flash chromatography of the crude
product using CHCl3-MeOH (94:6) as eluent, 1k (31 mg, 22%)
was produced as a white solid, mp 94-97 °C (EtOH).
N,N′-Bis(phenoxyacetyl)suberic Dihydrazide (2l). From
suberic dihydrazide (5.00 g, 24.75 mmol), phenoxyacetic acid
(7.52 g, 49.47 mmol), and POCl3 (30.39 g, 197.98 mmol) in CH3-
CN (60 mL), 2l (4.15 g, 36%) was produced as a white solid,
mp 202-204 °C (EtOH).
N,N′-Bis(phenoxyacetyl)azelaic Dihydrazide (2m). From
azelaic dihydrazide (5.00 g, 23.15 mmol), phenoxyacetic acid
(7.03 g, 46.25 mmol), and POCl3 (28.43 g, 185.21 mmol) in CH3-
CN (70 mL), 2m (2.47 g, 23%) was produced as a white solid,
mp 172-175 °C (EtOH).

4074 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 12
Garcı´a et al.
1,6-Bis(4-benzyl-5-phenoxymethyl[1,2,4]triazol-3-yl)-
hexane (1l). From 3l (620 mg, 1.43 mmol) and benzylamine
(2 mL, 18.32 mmol) and after flash chromatography of the
crude product using CHCl3-MeOH (98:2) as eluent, 1l (236
mg, 27%) was produced as a white solid, mp 145-147 °C
(EtOH).
1,7-Bis(4-benzyl-5-phenoxymethyl[1,2,4]triazol-3-yl)-
heptane (1m). From 3m (300 mg, 0.67 mmol) and benzyl-
amine (1.5 mL, 13.74 mmol) and after flash chromatography
of the crude product using CHCl3/MeOH (98:2) as eluent, 1m
(214 mg, 51%) was produced as a white solid, mp 92-94 °C
(EtOH).
1,8-Bis(4-benzyl-5-phenoxymethyl[1,2,4]triazol-3-yl)oc-
tane (1n). From 3n (315 mg, 0.68 mmol) and benzylamine (1
mL, 9.16 mmol) and after flash chromatography of the crude
product using CHCl3/MeOH (98:2) as eluent, 1n (179 mg, 42%)
was produced as a white solid, mp 109-111 °C (EtOH).
Biological Activity. Synthetic human AM was purchased
from Peninsula (S. Carlos, CA). Neutralizing monoclonal
antibody against AM was produced by Cuttitta et al.16,17 and
labeled with peroxidase using EZ-Link plus activated peroxi-
dase (Pierce, Rockford, IL).
Human AM was attached to PVC 96-well plates by passive
absorption that involved incubating 50 µL of AM (at 1 ng/µL)
in 50 µL of phosphate buffer saline (PBS) per well for 1 h. After
the coating solution was discarded, wells were washed with
200 µL of PBS three times. Immediately after, an amount of
200 µL of 1% BSA in PBS was added and the solution was
incubated for 1 h. Then, this solution was aspirated off and
50 µL containing 1 µM of one of the compounds 1j-n, 2j-n,
or 3j-n in 1% bovine serum albumin (BSA)-0.1% Tween-20
in PBS was added. After 1 h, 50 µL of peroxidase-labeled
antibody (at 2.4 µg/mL) were added and the solution was
allowed to react for 1 h. Following three thorough washes with
200 µL of 1% BSA in PBS, peroxidase activity was developed
using o-phenylenediamine dihydrochloride (Sigma) as a sub-
strate. The reaction product was quantified with a plate reader
(ASYS HiTech Expert 96) at 450 nm. Each plate contained
several internal controls including wells without any coating
that are used to calculate nonspecific binding, wells where no
potential modulators were added, which provided maximum
binding, and wells where the unlabeled antibody (at 1.2 µg/
mL) substituted the small molecule, as a positive inhibition
control. Each compound was added to triplicate wells in the
same plate. A positive hit was defined as a compound able to
significantly reduce the amount of product in three indepen-
dent plates. The intraassay variation was 6%, and the inter-
assay variation was 13%. The sensitivity of the assay, as
calculated with the cold antibody, was 12 nM, and the dynamic
range was between 12 and 54 nM.
3D-QSAR Analysis. The 3D geometries for the compounds
of the series were obtained automatically from their structures
in the SMILES format using the program CORINA.18 This
method produces reasonable extended conformations for all
the compounds, which is appropriate for the QSAR analysis
of congeneric series, as the present one. The conformations
obtained from CORINA18 were then analyzed using the grid-
independent descriptors (GRIND) methodology with the pro-
gram ALMOND 3.2,11 which contains version 19 of the
program GRID. DRY, N1, and O probes were chosen in order
to represent potentially important groups of the binding site.
Molecular shape field (TIP “probe”) was also included.19 This
descriptor characterizes the convexity of the molecule, which
can indicate the position and spatial extent of chemical groups
and of the molecule as a whole. The structures and affinities
(y variable) of the AM agonists are presented in Table 1. The
ALMOND parameters were set to default, the ALMD directive
was equal to 1, and the size of the correlograms was automati-
cally established by the program. The baseline was removed
for scaling. Cross-validation was done by using the leave-one-
out (LOO) method or by assigning the compounds randomly
to five groups, performing cross-validation on these groups,
and then repeating the whole procedure 20 times. No relevant
differences were found between both validation methods. All
computations were carried out on an Intel Pentium 4 using
the Linux RedHat 9 operating system.
Conformational Analysis. Conformational studies were
performed using Macromodel 7.2. The MMFF94s20 force field
was used in combination with the GB/SA solvation model.21
The model includes both generalized Born based (GB) solvent
polarization terms and surface area based (SA) terms.22,23
Default values were set for the nonbonded interactions cutoff.
Energy minimizations were performed with the Powell-
Reeves conjugated gradient (PRCG),24 and the derivate con-
vergence criterion was set to 1.0 kJ Å-1 mol-1. Conformational
search was performed by the Monte Carlo method25 for the
random variation of all of the rotable bonds combined with
the so-called low mode conformational search (LMCS) algo-
rithm.26 For each calculation 1000 Monte Carlo steps were
carried out followed by 500 minimization iterations.
Molecular Dynamics of Compounds 1h and 1i. Molec-
ular dynamics were performed using AMBER.14 Highly ac-
curate evaluations of the atomic point charges were carried
out by means of quantum chemical calculations at the HF/6-
31G* level of theory using Gaussian 9827 followed by the RESP
fitting procedure.28 Additional force field parameters of com-
pounds 1h and 1i are given in the Supporting Information.
All the simulations were carried out in the presence of solvent
(water box size ) 35 Å3). After minimization and 30 ps of
equilibration dynamics, the compounds were submitted to a
500 ps restricted dynamics simulation (from 20 to 0 kcal mol-1
along 500 ps) on those atoms that were able to establish
hydrogen bonds in the conformational search performed
earlier. Finally, 1 ns of unrestricted dynamics simulation was
performed for all the compounds.
Acknowledgment. This work was supported by the
Universidad San Pablo-CEU (Grant No. 5/02). S.M.-S.
thanks Ministerio de Educacio´n y Ciencia for a Ramo´n
y Cajal contract. A grant to M.A.G. from Fundacio´n
Universitaria San Pablo CEU is also acknowledged.
A.M. is supported by a grant from the Ministerio de
Educacio´n y Ciencia (Grant BFU 2004-02838/BFI).
Supporting Information Available: 1H and 13C NMR,
IR, and MS data of compounds 2k-n, 3k-n, and 1k-n,
additional force field parameters of compounds 1h and 1i, and
microanalysis data for compounds 1j-n and 3j-n. This
material is available free of charge via the Internet at http://
pubs.acs.org.
References
(1) Mart´ınez, A.; Julia´n, M.; Bregonzio, C.; Notari, L.; Moody, T.;
et al. Identification of vasoactive non-peptidic positive and
negative modulators of adrenomedullin using a neutralizing
antibody-based screening strategy. Endocrinology 2004, 145,
3858-3865.
(2) Kitamura, K.; Kangawa, K.; Kawamoto, M.; Ichiki, Y.; Naka-
mura, S.; et al. Adrenomedullin: a novel hypotensive peptide
isolated from human pheochromocytoma. Biochem. Biophys. Res.
Commun. 1993, 192, 553-560.
(3) Hinson, J. P.; Kapas, S.; Smith, D. M. Adrenomedullin, a
multifunctional regulatory peptide. Endocr. Rev. 2000, 21, 138-
167.
(4) Lo´pez, J.; Mart´ınez, A. Cell and molecular biology of the
multifunctional peptide, adrenomedullin. Int. Rev. Cytol. 2002,
221, 1-92.
(5) Julia´n, M.; Cacho, M.; Garc´ıa, M. A.; Mart´ın-Santamar´ıa, S.; de
Pascual-Teresa, B.; Ramos, A.; et al. Adrenomedullin: a new
target for the design of small molecule modulators with promis-
ing pharmacological activities. Eur. J. Med. Chem., in press.
(6) Dobrzynski, E.; Wang, C.; Chao, J.; Chao, L. Adrenomedullin
gene delivery attenuates hypertension, cardiac remodeling, and
renal injury in deoxycorticosterone acetate-salt hypertensive
rats. Hypertension 2000, 36, 995-1001.
(7) Mart´ınez, A.; Vos, M.; Guedez, L.; Kaur, G.; Chen, Z.; et al. The
effects of adrenomedullin overexpresion in breast tumor cells.
J. Natl. Cancer Inst. 2002, 94, 1226-1237.
(8) Mart´ınez, A.; Elsasser, T. H.; Bhathena, S. J.; Pı´o, R.; Buchanan,
T. A.; et al. Is adrenomedullin a causal agent in some cases of
type 2 diabetes? Peptides 1999, 20, 1471-1478.

Modulators of Adrenomedullin
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 12 4075
(9) Kudari, S. M.; Badiger, S. E. Synthesis of new series of 1,8-bis-
(5-aroloxy substituted 1,3,4-oxadiazol-2-yl)octanes, 1,8-bis(5-
aroloxy substituted 4-amino/anilino 1,2,4-triazol-2-yl)octanes.
Orient. J. Chem. 1997, 13, 245-248.
(10) Goodford, P. J. A computational procedure for determining
energetically favorable binding sites on biologically important
macromolecules. J. Med. Chem. 1985, 28, 849-857.
(11) ALMOND, version 3.2; Molecular Discovery Ltd. (215 Marsh
Road, Pinner, Middlesex, U.K.), 2004.
(21) Still, W. C.; Tempczyk, A.; Hawley, R. C.; Hendrickson, T.
Semianalytical treatment of solvation for molecular mechanics
and dynamics. J. Am. Chem. Soc. 1990, 112, 6127-6129.
(22) Weiser, J.; Shenkin, P. S.; Still, W. C. Approximate atomic
surface from linear combinations of pairwise overlaps (LCPO).
J. Comput. Chem. 1999, 20, 217-230.
(23) Weiser, J.; Shenkin, P. S.; Still, W. C. Fast, Approximate
algorithm for detection of solvent-inaccessible atoms. J. Comput.
Chem. 1999, 20, 586-596.
(12) Pastor, M.; Cruciani, G.; McLay, I.; Pickett, S.; Clementi, S.
GRid-INdependent Descriptors (GRIND): a novel class of align-
ment-independent three-dimensional molecular descriptors. J.
Med. Chem. 2000, 43, 3233-3243.
(13) Mohamadi, F.; Richards, N. G. J.; Guida, W. C.; Liskamp, R.;
Lipton, M.; et al. MacroModelsAn integrated software system
for modeling organic and bioorganic molecules using molecular
mechanics. J. Comput. Chem. 1990, 11, 440-467.
(14) Case, D. A.; Pearlman, D. A.; Caldwell, J. W.; Cheatham, T. E.
I.; Ross, W. S.; et al. AMBER, version 6; University of California
at San Francisco: San Francisco, CA 1999.
(24) Polak, E.; Ribiere, G. Note sur la Convergence de Methodes de
Directions Conjuguees, Serie Rouge (Note on convergence of
conjugated direction methods). Rev. Fr. Inf. Rech. Oper. 1969,
16-R1.
(25) Chang, G.; Guida, W. C.; Still, W. C. An internal coordinate
Monte Carlo method for searching conformational space. J. Am.
Chem. Soc. 1989, 111, 4379-4386.
(26) Kolossvary, I.; Guida, W. C. Low mode search. An efficient,
automated computational method for conformational analysis:
Application to cyclic and acyclic alkanes and cyclic peptides. J.
Am. Chem. Soc. 1996, 118, 5011-5019.
(15) Voigt, J. H.; Bienfait, B.; Wang, S.; Nicklaus, M. C. Comparison
of the NCI open database with seven large chemical structural
databases. J. Chem. Inf. Comput. Sci. 2001, 41, 702-712.
(16) Mart´ınez, A.; Weaver, C.; Lo´pez, J.; Bhathena, S. J.; Elsasser,
T. H.; et al. Regulation of insulin secretion and blood glucose
metabolism by adrenomedullin. Endocrinology 1996, 137, 2626-
2632.
(17) Cuttitta, F.; Carney, D. N.; Mulshine, J.; Moody, T. W.; Fedorko,
J.; et al. Bombesin-like peptides can function as autocrine growth
factors in human small-cell lung cancer. Nature 1985, 316, 823-
826.
(18) Gasteiger, J.; Rudolph, C.; Sadowski, J. Automatic generation
of 3D atomic coordinates for organic molecules. Tetrahedron
Comput. Methodol. 1991, 202, 537-547.
(19) Fontaine, F.; Pastor, M.; Sanz, F. Incorporating Molecular Shape
into the Alignment-Free Grid-Independent Descriptors. J. Med.
Chem. 2004, 47, 2805-2815.
(27) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.;
Robb, M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery,
J. A., Jr.; Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam,
J. M.; Daniels, A. D.; Kudin, K. N.; Strain, M. C.; Farkas, O.;
Tomasi, J.; Barone, V.; Cossi, M.; Cammi, R.; Mennucci, B.;
Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.; Petersson, G.
A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.; Rabuck,
A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz,
J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.;
Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.;
Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.;
Challacombe, M.; Gill, P. M. W.; Johnson, B. G.; Chen, W.; Wong,
M. W.; Andres, J. L.; Head-Gordon, M.; Replogle, E. S.; Pople,
J. A. Gaussian 98, revision A.7; Gaussian, Inc.: Pittsburgh, PA,
1998.
(28) Bayly, C. I.; Cieplak, P.; Cornell, W.; Kollman, P. A. A well-
behaved electrostatic potential based method using charge
restraints for deriving atomic charges: the RESP model. J. Phys.
Chem. 1993, 97, 10269-10280.
(20) Halgren, T. A. MMFF VII. Characterization of MMFF94,
MMFF94s, and other widely available force fields for conforma-
tional energies and for intermolecular-interaction energies and
geometries. J. Comput. Chem. 1999, 20, 730-748.
JM050021+