Synthesis and preliminary evaluation of amiloride analogs as inhibitors of the urokinase-type plasminogen activator (uPA)

Article abstract of DOI:10.1016/j.bmcl.2011.09.044

A known side-activity of the oral potassium-sparing diuretic drug amiloride is inhibition of the enzyme urokinase-type plasminogen activator (uPA, K _i = 7 μM), a promising anticancer target. Several studies have demonstrated significant antitum

Full text of DOI:10.1016/j.bmcl.2011.09.044

Bioorganic & Medicinal Chemistry Letters 21 (2011) 6760–6766
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Synthesis and preliminary evaluation of amiloride analogs as inhibitors
of the urokinase-type plasminogen activator (uPA)
Hayden Matthews a, Marie Ranson b, Joel D.A. Tyndall c, Michael J. Kelso a,
⇑
a School of Chemistry, University of Wollongong, NSW 2522, Australia
b School of Biological Sciences, University of Wollongong, NSW 2522, Australia
c National School of Pharmacy, University of Otago, Dunedin 9054, New Zealand
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 25 July 2011
Revised 10 September 2011
Accepted 13 September 2011
Available online 18 September 2011
A known side-activity of the oral potassium-sparing diuretic drug amiloride is inhibition of the enzyme
urokinase-type plasminogen activator (uPA, Ki = 7 M), a promising anticancer target. Several studies
l
have demonstrated significant antitumor/metastasis properties for amiloride in animal cancer models
and it would appear that these arise, at least in part, through inhibition of uPA. Selective optimization
of amiloride’s structure for more potent inhibition of uPA and loss of diuretic effects would thus
appear as an attractive strategy towards novel anticancer agents. The following report is a preliminary
structure–activity exploration of amiloride analogs as inhibitors of uPA. A key finding was that the
well-studied 5-substituted analogs ethylisopropyl amiloride (EIPA) and hexamethylene amiloride
(HMA) are approximately twofold more potent than amiloride as uPA inhibitors.
Keywords:
Urokinase-type plasminogen activator
uPA
Inhibitor
Ó 2011 Elsevier Ltd. All rights reserved.
Amiloride
Structure–activity relationships
Orally active non-cytotoxic agents capable of inhibiting the
progression of primary tumors to metastasis have long been a goal
in anticancer drug discovery.1 Upregulation of the plasminogen
activation system (PAS) is known to play a critical role in tumor
invasion and metastasis2,3 and several components of the PAS offer
attractive drug targets in this context.4–6 In its simplest form, the
PAS comprises the serine protease urokinase-type plasminogen
activator (uPA), its cognate glycosylphosphatidylinositol
(GPI)-anchored cell surface-bound receptor (uPAR) and two endog-
enous serpins; plasminogen activator inhibitor 1 (PAI-1) and plas-
gether these lead to pericellular proteolysis and remodelling of
the local extracellular environment—key events required for
metastasis (reviewed in Refs. [7,8]). To highlight the importance
of the PAS in metastasis, upregulated uPA and PAI-1 have been
shown to be the strongest known prognostic biomarkers of short-
ened disease-free survival and overall survival in breast cancer and
the most accurate predictors of metastasis in lymph-node-negative
tumours.9
One approach towards dampening PAS activity and potentially
inhibiting metastasis is to target the serine protease activity of
uPA. Several classes of uPA inhibitors have been reported and se-
lected analogs have undergone preclinical evaluation as non-
cytotoxic antitumor/metastasis agents.10 An orally active uPA
inhibitor prodrug MESUPRONÒ (Wilex AG, Germany) is currently
undergoing two phase II trials for the treatment of breast and
pancreatic tumors.11 A general problem with uPA inhibitors, how-
ever, is that many contain highly basic amidines or guanidines
which often confer poor drug-like properties.12 These positively
charged groups are necessary for making a key salt-bridge contact
with Asp189 located at the base of uPA’s S1-pocket. Scaffolds
which could employ less basic groups to make this crucial interac-
tion are thus sought after for elaboration into uPA inhibitors with
improved properties.
minogen activator inhibitor
2 (PAI-2). Receptor-bound uPA
efficiently cleaves and thus activates co-localised plasminogen at
the cell surface, which reveals the broad spectrum serine protease
plasmin. Plasmin activates downstream extracellular proteases
(e.g., matrix metalloproteases) and latent growth factors and to-
Abbreviations: EIPA, ethylisopropyl amiloride; GPI, glycosylphosphatidylinosi
tol; HMA, hexamethylene amiloride; HMW-uPA, high molecular weight urokinase-
type plasminogen activator; NBS, N-bromosuccinimide; NCS, N-chlorosuccinimide;
NHE1, sodium hydrogen exchanger 1; NIS, N-iodosuccinimide; PAI-1, plasminogen
activator inhibitor 1; PAI-2, plasminogen activator inhibitor 2; PAS, plasminogen
activation system; rp-HPLC, reverse phase high performance liquid chromatogra
phy; SOSA, selective optimisation of side-activity; tPA, tissue-type plasminogen
activator; uPA, urokinase-type plas minogen activator; uPAR, urokinase-type
plasminogen activator receptor.
Amiloride.HCl 1 is an orally administered potassium-sparing
diuretic that has been used clinically for several decades, most com-
monly in combination with thiazide (e.g., hydrochlorothiazide) or
⇑
Corresponding author. Tel.: +61 2 4221 5085; fax: +61 2 4221 4287.
E-mail address: mkelso@uow.edu.au (M.J. Kelso).
0960-894X/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2011.09.044

H. Matthews et al. / Bioorg. Med. Chem. Lett. 21 (2011) 6760–6766
6761
(b)
(a)
S1β subsite: Lys143, Ser146, Gln192, Cys191-Cys220
disulfide, Gly216, Arg217,Gly219
H
His57 (Catalytic Histidine)
HN
N
Cl
H2
N
O
Gly219
Amiloride
2.6 Å
3.3 Å
N
2.8 Å
2.7 Å
O
N
H2
H
O
S
O
N
N
N
HN
O
O
NH2
2.9 Å
O
O
Sulfate of
Crystallization
H2N
HN
2.8 Å
2.6 Å
O
O
H
O
3.2 Å
OH
Asp189
3.0 Å
H
N
3.0 Å
2.7 Å
H
3.0 Å
O
O
O
Gly193
N
H
O
Ser195 (Catalytic Serine)
NH
Ser190
Figure 1. (a) X-ray crystal structure of the amiloride 1:uPA complex (binding site only); PDB 1f5l.16 (b) Summary of the binding interactions between amiloride (blue), uPA
(black), an adventitious sulfate of crystallization (red) bound in the oxyanion hole and a water molecule (green) bound deep within the S1 pocket.
loop diuretics (e.g., frusemide) as an antikaliuretic.13 Studies have
shown that amiloride is also a moderately potent competitive inhib-
bond to an adventitious sulfate ion bound in the oxyanion hole.
The sulfate ion is further held in place by hydrogen bonds to the
Ser195 hydroxyl, His57 imidazole nitrogen and Gly193 amide
NH. The chlorine atom at C6 partially occupies the S1b subsite
and participates in interactions with residues Gln192, Gly216,
Gly219 and the Cys191-Cys220 disulfide bond.
The structural features of amiloride that were examined in this
study are summarised in Figure 2. All compounds were purified by
preparative rp-HPLC to greater than 95% purity (analytical rp-
HPLC) and evaluated for uPA inhibitory potency using a 96-well
plate in vitro enzyme assay. The commercially available colorimet-
ric uPA substrate Spectrozyme UK (American Diagnostica Inc.) was
used with active high molecular weight uPA (HMW-uPA) in all as-
says. Briefly, compounds were dissolved in DMSO to create 20 mM
stock solutions which were diluted with buffer in series in the
plates to give final DMSO concentrations of less than 2%. Two com-
pounds were assayed per plate with triplicate measurements taken
for each inhibitor concentration. Assay blanks (no enzyme) were
included to account for the colour of some inhibitors. Amiloride
itor of uPA (Ki = 7 l
M)14 and that uPA inhibition correlates with sig-
nificant antitumor/metastasis effects of the compound in vivo.15 The
published X-ray co-crystal structure of amiloride bound to uPA16
(Fig. 1) reveals that the drug positions its acylguanidine unit deep
within the S1 pocket to make the key salt-bridge interaction with
Asp189. The low lM uPA inhibitory potency of amiloride combined
with its demonstrated antitumor/metastasis effects in animal mod-
els, the reduced basicity of its acylguanidine (pKa = 8.8),17 its long-
standing use as a safe oral K+-sparing diuretic and its known
selectivity for uPA over closely related off-target serine proteases
(i.e., tissue-type plasminogen activator (tPA), plasmin, thrombin
and kallikrein)14 suggest it is an excellent starting point for a selec-
tive optimisation of side-activity (SOSA)18 study aimed at producing
a more potent, orally active uPA inhibitor for use in the cancer set-
ting. The following report is a preliminary structure–activity explo-
ration of amiloride analogs as inhibitors of uPA.
The reported amiloride:uPA X-ray structure showed that amilo-
ride occupies the S1 and S1b sites of uPA by making a network of
hydrogen bonds and van der Waals contacts in addition to the
salt-bridge interaction with Asp189 (Fig. 1).16 The Ser190 hydroxyl
group forms a hydrogen bond to a terminal nitrogen atom of the
acylguanidine, which in turn forms a hydrogen bond with a water
molecule bound deep within the S1 pocket. The Ser190 hydroxyl
also forms a hydrogen bond to this water molecule, as does the
acylguanidine carbonyl oxygen. The Gly219 carbonyl oxygen forms
a weak hydrogen bond (OÁ Á ÁN distance 3.3 Å) to the acylguanidine
amide NH. The Ser195 hydroxyl group is hydrogen bonded to the
amino group at C3, while the amino group at C5 forms a hydrogen
was included in all assays as a positive control (IC50 = 11 lM under
the assay conditions). Plates were incubated at 37 °C in a plate
reader and absorbances read at 405 nm over time. Absorbance val-
ues were recorded for a time point taken from the linear region of
plots of absorbance vs time and IC50 values were calculated from
sigmoidal dose response curves of absorbance versus log[inhibitor]
using GraphPad Prism V. 5.01 software.
At the outset we were interested in establishing the importance
of the acylguanidine:Asp189 interaction and, specifically, whether
this interaction might be providing the bulk of amiloride’s uPA
affinity with other contacts perhaps playing only a minor role. A
Cl
NH2
O
Ar
N
H
NH2
O
R
N
N
N
O
Exploration of aroylguanidine core
NH2
NH2
N
H
N
H2N
NH2
Variations
at C6
Acylguanidine
isosteric replacements
Cl
O
NH2
NH2
1
N
6
5
2
3
Cl
Cl
N
N
H
N
N
Br
N
N
Cl
R
H2N
N
4
NH2
N
N
R1
NH2 H2N
Amiloride
Substituted
amines at C5
Removal of
3- or 5-amino groups
1
Figure 2. Summary of the amiloride structural features examined.

6762
H. Matthews et al. / Bioorg. Med. Chem. Lett. 21 (2011) 6760–6766
Table 1
ole and acylhydrazide isosteric replacements for the acylguanidine,
Aroylguanidines with varied aryl cores
respectively. Compound 9 was accessed in 39% yield by refluxing
hydroxyguanidine hemisulfate19 and sodium in ethanol with the
commercially available methyl ester 11.20 Compound 10 was
prepared in 81% yield by refluxing 11 with hydrazine hydrate in
ethanol. Analogs 9 and 10 were both found to be inactive
Cl
(i) Guanidine.HCl,
Na, iPrOH/Δ
O
O
NH2
R
R
N
H
NH2
O
(ii)HCl
2-8
(IC50 >1000 lM) (Scheme 1).
The importance of the amino groups at the 5- and 3-positions of
amiloride were investigated next with the deaminated analogs 12
and 13 (Scheme 2). Chlorination of the commercially available pyr-
azine methyl ester 14 by heating with N-chlorosuccinimide (NCS)
in DMF at 80 °C gave 15 in 72% yield. Guanidinylation of 15 using
the aforementioned conditions provided 12 in 87% yield. Guanid-
inylation of the commercially available methyl ester 16 afforded
compound 13 (28%). The chloro analogue of 16 was not commer-
cially available so the direct 3-deamino analogue of amiloride
(i.e., 6-chloro-13) was not pursued; however it was later noted that
substitution of the 6-chloro group of amiloride 1 with bromide (i.e.,
6-Br-amiloride 58, (Table 3) resulted in a twofold increase in
activity. This suggests that 6-chloro 13 would most likely show
very similar or only slightly increased activity relative to 13.
It was found that removal of either 3- or 5-amino groups re-
Compound
R
Yield %
IC50, lm
2
89
>1000
>1000
>1000
3
4
5
42
27
56
N
N
N
N
>1000
>1000
Cl
6
27
Cl
N
Cl
sulted in a total loss of activity (IC50 >1000 lM). Loss of activity
7
8
62
93
222
upon removing the amino group at C3 (i.e., 13) may (partially) re-
sult from losing the hydrogen bond between this amine and the OH
group of the catalytic Ser195 residue (Fig. 1b). Explaining the total
loss of potency after removing the 5-amino group (i.e., 12) is diffi-
cult since this group appears not to form any direct contacts with
the enzyme, although it does form a hydrogen bond with the
adventitious sulfate of crystallization bound in the oxyanion hole
(Fig. 1b).
A major part of the study focused on amiloride analogs carrying
substituents at C5. This was due to their ease of synthesis and be-
cause many 5-substituted analogs are known which show reduced
diuretic effects relative to amiloride.21 Application of the SOSA
approach to identify amiloride analogs with high uPA inhibitory
potency for use as antitumor/metastasis agents would (ideally)
require that the evolved compounds show reduced diuretic activity
and minimal effects on K+-levels. While many amiloride analogs
bearing hydroxyl, alkoxy, mercapto, alkylmercapto, alkyl, aryl
and other substituents at the 5-position are reported,21,22 we chose
to examine those carrying substituted amines at this position after
the 5-deaminated derivative 12 was found to be inactive.
The amiloride:uPA X-ray co-crystal structure suggests that
substituted amines at C5 should extend away from the pyrazine
ring towards the oxyanion hole being occupied in the structure
by the adventitious sulfate ion (Fig. 1).16 Analogs incorporating
H2N
N
O
>1000
HN
series of aroylguanidines 2–8 with varied aryl cores were thus syn-
thesized by refluxing their commercially available methyl ester
precursors with guanidine.HCl (pre-neutralised with Na) in isopro-
panol. Methyl esters which were not commercially available were
synthesised by methylation of their available carboxylic acids by
using CH3I/Cs2CO3 in N,N-dimethylformamide (DMF). As shown
in Table 1, the aroylguanidines were all essentially inactive with
IC50 values above 1 mM (except 7, IC50 = 222 lM). It was significant
that 2-pyrazinoyl guanidine 5 was inactive as this structure forms
the core of amiloride. These findings are evidence that the 2-pyraz-
inoyl guanidine core of amiloride is not on its own responsible for
uPA affinity and that the 3- and 5-amino and 6-chloro groups, in
combination with the acylguanidine, contribute significantly to
binding. The improved potency of the substituted benzoylguani-
dine 7 (which bears an aryl ring substitution pattern similar to
amiloride) relative to benzoylguanidine 2 supports this conclusion.
The importance of the acylguanidine unit was further explored
with analogs 9 and 10, which incorporate 3-amino-1,2,4-oxadiaz-
O
O
NCS
O
DMF, 80 oC
N
N
Cl
N
N
O
R
Cl
N
O
72%
NH2
NH2
H2N
N
11
NH2
14
15
12
R = OMe
(i) Guanidine.HCl
Na, iPrOH, Δ
Cl
NH2
=
1/2 SO4
(ii) HCl
87%
NH2
Na,
NH2NH2.H2O
EtOH, Δ
HN
NH2
EtOH, Δ
HO
N
H
NH2
IC50 > 1000 μM
39%
81%
O
Cl
N
O
(i) Guanidine.HCl
NH2
O
O
NH2
N
N
NH2
Cl
N
Cl
H2N
i
Na, PrOH, Δ
N
N
H
Br
N
N
Br
N
N
H
NH2
O
(ii) HCl
28%
H2N
NH2
N
NH2
H2N
H2N
N
9
10
13
16
IC50 > 1000 μM
IC50 > 1000 μM
IC50 > 1000 μM
Scheme 1. Analogs incorporating acylguanidine isosteres.
Scheme 2. Analogs with 3- and 5-amino groups removed.

H. Matthews et al. / Bioorg. Med. Chem. Lett. 21 (2011) 6760–6766
6763
negatively charged groups attached to the amine via 1- or 2-carbon
linkers (i.e., carboxylates 18 and 20, sulfonate 21, phosphonates 23,
25, 26 and tetrazoles 29 and 31) were thus explored under the
hypothesis that these groups could potentially mimic the sulfate
ion and pick up favourable contacts within the oxyanion hole. Un-
charged synthetic precursors of these inhibitors (carboxylate esters
Table 2
Analogs with substituted amines at C5
Cl
Cl
O
O
NH2
O
NH2
(i) Amine
(i) Guanidine.HCl
Cl
Cl
N
Cl
Cl
N
N
Cl
R
N
N
O
Na, iPrOH, Δ
N
H
NH2 DMF, 100
°C
N
H
NH2
N
NH2
(ii) HCl
(ii) HCl
84%
NH2
NH2
48
17-47
49
Compound
R
Yield %
IC50
(
l
M)
Compound
R
Yield %
97
IC50
8
(lM)
NH
NH
1
NH2
a
11
32
33
MeO
NH
17
18
19
20
82
26
110
25
63
77
81
78
38
17
6
N
O
HO
NH
NH
77b
69
34
35
36
O
N
O
NH
N
F
EtO
HO
NH
O
NH
83c
102
6
NH
NH
O
S
21
48
>1000
37
84
16
NH
NH
HO
O
O
EtO
N
22
23
24
25
26
27
28
P
NH
45
75d
53
37
>1000
44
38
39
40
41
42
43
44
88
83f
78
14
3
OEt
O
H
N
HO
N
N
NH
P
OH
OEt
P
15
28
30
21
15
NH
EtO
HO
NH
NH
O
OH
P
H
N
22e
32e
24
>1000
>1000
10
52
NH
NH
O
OH
P
N
69
NH
EtO
EtO
O
O
H
N
S
92
P
NH
NH
NH
OEt
O
S
NH
N
58
50
90g
O
N
O
NH
NH
H
N
N
N
29
63
177
45
89
35
NH
NH
S
NH
N
O
N
30
31
61
74
16
61
46
47
78
26
6
6
N
N
N
N
NH
a
b
c
d
e
f
Amiloride.HCl 1 was purchased from Sigma–Aldrich.
Yield of 18 following hydrolysis (K2CO3, MeOH, H2O) of 17.
Yield of 20 following hydrolysis (K2CO3, MeOH, H2O) of 19.
Yield of 23 following hydrolysis (TMSCl) of 22.
Yield of 25/26 following hydrolysis (TMSCl) of 24.
Yield and IC50 correspond to the unexpected bicycle 39 formed form the 20 amine cyclizing onto C6 and displacing the Cl.
Yield of 44 following oxidation (Oxone) of 43.
g

6764
H. Matthews et al. / Bioorg. Med. Chem. Lett. 21 (2011) 6760–6766
H2N NH.H2O
Et3BzN Cl
CCl4: DCM (4:7)
K2CO3
increases in potency (p-fluorobenzyl 36, IC50 = 6
analogs substituted with benzyl pyridines (compounds 33–35) only
the para-substituted pyridine 35 (IC50 = 6 M) showed increased
potency relative to amiloride. Extension of the benzylamine substi-
tuent to phenylethylamine caused a twofold drop in potency (37,
lM). Of the three
O
P
O
P
OEt
l
NH2
EtO
N
H
H
EtO
EtO
90%
Ref 23
50
IC50 = 16 lM vs 32, IC50 = 8 lM).
5-Substituted analogs containing 2°, 3° or 4° nitrogens were
investigated to examine the effects of positively charged groups
Cl
N
N
R
N
H
100%
Ref 24
LiAlH4
94%
in this region. Analogs 26 (IC50 = 14
bearing 3° amino groups showed only slightly reduced potency,
while 42 (pyridinium, IC50 = 30 M) and 41 (2° amine,
IC50 = 28
lM) and 40 (IC50 = 15 lM)
R = CN
l
51
= CH2NH2
l
M) were ꢀ3-fold less potent than amiloride. It was con-
Br
Br
Pyridine
NH2
cluded that positive charges in C5 substituents do not confer in-
creased potency but they are largely tolerated or produce only
minor losses in potency.
NH3
N
Br
86%
Ref 25
52
Analogs 43, 44 and 45, which included sulfur in the C5-substi-
tuent, were examined because some potent uPA inhibitors have
been shown to position sulfonamides in the oxyanion hole.26,27
N
N
Cl
NH2
n
n
NH4OH
N
N
NH
NH
N
N
or
(i) NaN3
(ii) PPh3
53
54
n = 1
= 2
n = 1
= 2
52%
74%
The sulfone 44 (IC50 = 15
than its corresponding sulfide 43 (IC50 = 21
pounds were slightly less potent than amiloride. Sulfonamide 45
(IC50 = 35 M) was more than threefold less potent than amiloride.
lM) was found to be slightly more potent
l
M), while both com-
Scheme 3. Synthesis of non-commercially available amines 50–54.23–25
l
The overall trend with this series indicated that analogs bearing
shorter amines at C5 tended to slightly improve inhibitor potency
while the more extended amines produced no improvement or a
reduction in potency. Inspection of the amiloride:uPA X-ray co-
crystal structure (Fig. 1) suggests that C5-substituted analogs
should place the substituent near the oxyanion hole, with longer
substituents extending out further, probably towards solvent.
The finding that only slight increases (up to twofold) and relatively
minor reductions (except for the negatively charged analogs) in
potency were observed with this diverse set of analogs suggests
that only non-specific contacts are being made between the
C5-substituents and the enzyme. The relatively flat SAR trend
would be compatible with these substituents significantly interact-
ing with the surrounding solvent.
17 and 19, diethylphosphonates 22 and 24 and nitriles 28 and 30,
respectively) were also examined. Other 5-substituted analogs (27,
32–47) were chosen in order to explore a cross-section of inhibi-
tors containing aromatic rings, basic (or quaternary) amines capa-
ble of carrying positive charges, and compounds containing sulfur.
Nucleophilic aromatic substitution reactions of the 5,6-dichlo-
ropyrazinoyl guanidine 49 with requisite amines were used to pre-
pare all 5-substituted analogs (Table 2). Compound 49 was
obtained in 84% yield from the commercially available methyl-
5,6-dichloropyrazine carboxylate 48 by guanidinylation using the
conditions described above.
Amines used in the synthesis of 17, 19, 21, 22, 24, 28, 30, 32–39,
41, 43, 45–47 were all commercially available while others re-
quired synthesis. Diethyl phosphorohydrazidate 50, 2-(benzyl-
methylamino) ethylamine 51 and 1-(2-aminoethyl)pyridinium
bromide 52 (used in the synthesis of 27, 40 and 42, respectively)
were prepared by reported methods (Scheme 3).23–25 Tetra-
zolylalkylamines 53 and 54 (used in the synthesis of 29 and 31,
respectively) were prepared from commercially available alkyl
chlorides by substitution with ammonia (i.e., 53) or by conversion
to the azide followed by Staudinger reduction with PPh3 (i.e., 54)
(Scheme 3).
Esters 17 and 19 were hydrolysed under basic conditions
(K2CO3/MeOH/H2O) to yield the carboxylic acids 18 and 20, respec-
tively. Phosphonate esters 22 and 24 were fully or partially hydro-
lysed using trimethylsilyl chloride to afford 23, 25 and 26.
Oxidation of thioether 43 with Oxone yielded sulfone 44. An at-
tempt to produce 5-(N,N’-dimethylethane-1,2-diamino)-amiloride
by reaction of 49 with N,N’-dimethylethylenediamine instead re-
sulted in formation of a novel bicycle 39, where the free secondary
amine of the initial 5-sustituted product had cyclised onto the
6-position by displacing the chloride.
The hypothesis that negatively charged groups appended at C5
might increase potency by acting as sulfate mimetics able to make
favourable contacts in the oxyanion hole was found to be incorrect.
All inhibitors bearing C5-substituents capable of carrying negative
charges (i.e., 18, 20, 21, 23, 25, 26, 29, 31) showed dramatically re-
duced or total loss of activity relative to amiloride, while uncharged
synthetic precursors of these inhibitors (i.e., 17, 19, 22, 24, 28 and
30) universally showed less dramatic losses in potency (IC50 = 25–
Compounds 46 (commonly known as ethylisopropyl amiloride,
EIPA) and 47 (hexamethylene amiloride, HMA) were of consider-
able interest in this study because they are two of the most potent
known inhibitors of the sodium hydrogen exchanger 1 (NHE1), an-
other emerging antitumor/metastasis drug target.28 Amiloride is
also a moderately potent NHE1 inhibitor (K0.5 = 7 lM) but EIPA
46 and HMA 47 are 140Â and 524Â more potent, respectively.17,29
The likelihood that the antitumor/metastasis effects of amiloride
(and analogs) in vivo arises through dual inhibition of both NHE1
and uPA was recently reviewed,15 and it was noted there that the
inhibitory potencies of EIPA and HMA against uPA had not yet been
reported. Compounds 46 (IC50 = 6 lM) and 47 (IC50 = 6 lM) are
now shown to be equipotent uPA inhibitors and ꢀ2-fold more po-
tent than amiloride.
The amiloride: uPA X-ray structure (Fig. 1) indicates that the C6
chloro group of amiloride projects toward and partially occupies
the S1b site.16 A selection of 6-substitued amiloride analogs which
either remove or replace the chloro group were thus prepared to
explore structure–activity relationships about the S1b site.
Hydrodehalogenation (H2, Pd/C, MgO)21 of amiloride 1 and methyl
ester 11 provided dechlorinated compounds 55 (59%) and 56 (98%),
respectively. Reaction of 55 with N-iodosuccinimide (NIS) and N-
bromosuccinimide (NBS) yielded aryl halides 57 (72%) and 58
(81%), and iodination of 56 provided the aryl iodide 59 in 98% yield
(Scheme 4). Compounds 60 and 61 were produced by hydrode-
halogenation (H2, Pd/C,MgO) of 32 and 37 in 78% and 81% yields,
respectively. Stille and Sonogashira couplings were carried out on
iodide 59 with phenyltributylstannane and phenylacetylene. Gua-
nidinylation of the resultant esters afforded the 6-substituted ana-
logs 62 and 63 (71% and 46% yields over two steps, respectively).
50 lM). Only diethylphosphorohydrazidate 27 (IC50 = 10 lM) re-
tained the potency of amiloride. Benzylic extensions (compounds
32–36) were well tolerated and produced slight (up to twofold)

H. Matthews et al. / Bioorg. Med. Chem. Lett. 21 (2011) 6760–6766
6765
O
O
O
N
N
R3
R1NH
N
N
Pd/C,
MgO, H2
Cl
N
N
R2
R2
R2
NIS or NBS
R1NH
NH2
NH2
R1NH
NH2
R1 = H
R
R1 = H
2 = NHC(NH2)2Cl
R1 = H,
2 = NHC(NH2)2Cl
R
R
2 = NHC(NH2)2Cl,
55
1
R3 = I
3 = Br
57
58
59
R
R1 = H, R2 = OMe
R1 = H, R2 = OMe
R1 = H, R2 = OMe, R3 = I
11
32
56
60
(i) Pd(PPh3)4, PhSnBu3, CuI
(ii) Guanidine.HCl, Na, iPrOH
R1 = PhCH2,
R1 = PhCH2,
R2 = NHC(NH2)2Cl
R
1 = H, R2 = NHC(NH2)2Cl, R3 = Ph
R2 = NHC(NH2)2Cl
62
63
(i) Pd(PPh3)4, PhCCH, CuI, DIPEA
(ii) Guanidine.HCl, Na,iPrOH
R1 = Ph(CH2)2,
R
R1 = Ph(CH2)2,
R1 = H, R2 = NHC(NH2)2Cl, R3 = PhCC
2 = NHC(NH2)2Cl
R
2 = NHC(NH2)2Cl
37
61
Scheme 4. Synthesis of analogs with variations at C6.
(62 IC50 = 10
amiloride.
lM) or higher potency (63 IC50 = 2
lM) relative to
Table 3
IC50 values of analogs with variations at C6
Amiloride shows significant antitumor/metastasis effects in
in vivo animal models and it would appear that these probably
arise from dual inhibition of uPA and NHE1.15 The in vitro potency
of amiloride against these two targets is relatively modest, how-
ever, so it is conceivable that an analog optimised for higher po-
tency against both targets might yield a superior anticancer
compound. Deliberately optimising compounds to potently inter-
act with two or more independent targets (i.e., polypharmacology)
is increasingly being recognised as a viable drug development
strategy.30 The NHE1 inhibitory potency of many amiloride analogs
has been reported and some show greatly increased potency rela-
tive to amiloride, including the well studied compounds EIPA 46
and HMA 47. In order to advance our long-term goal of identifying
potent dual NHE1/uPA inhibitors it is necessary to characterise the
potency of amiloride analogs as uPA inhibitors.
Cl
O
NH2
N
N
N
H
NH2
RNH
NH2
Compound
R
IC50
19
(lM)
55
H
60
61
43
120
Cl
O
NH2
This study initially explored simplified aroylguanidines and
analogs containing acylguanidine isosteres. It was found that sim-
ple aroylguanidines are poor uPA inhibitors and should not be pur-
sued further as alternative uPA inhibitor scaffolds. Analogs bearing
acylguanidine isosteres were found to be totally inactive indicating
that the acylguanidine group should be retained. Analogs lacking
either the 3- or 5-amino substituent were used to establish the
importance of these groups for inhibitor binding. Both compounds
were found to be inactive.
R
H2N
N
N
NH2
H
N
NH2
57
58
|
Br
2
5
62
10
It was postulated that analogs incorporating a negatively
charged sulfate mimetic at C5 could pick up favourable contacts
in the oxyanion hole of uPA. Surprisingly, analogs bearing nega-
tively-charged groups all showed greatly reduced potency relative
to amiloride. Ironically, the uncharged synthetic precursors of
these compounds all showed much lower reductions in potency.
Investigation of alkylaryl extensions at C5 were more fruitful, with
several analogs showing slightly improved potency. Overall,
substituted amines at C5 were generally well tolerated with inhib-
itors being as potent or only slightly less potent than amiloride.
While exploration of these analogs did not yield an inhibitor with
nanomolar potency as hoped, it succeeded in showing that substi-
tuted amines at the 5-position are at least well tolerated. Amiloride
analogs substituted at the 5-position are known to show reduced
diuretic effects.21 If a high potency uPA inhibitor can eventually
be evolved from amiloride it will be useful to know that the
5-position can be varied to reduce diuretic effects and to modulate
physiochemical properties without dramatically altering uPA
affinity.
63
2
Compounds 55, 60 and 61 (i.e., dechloro analogs of 1, 32 and 37,
respectively) were used to explore the contribution of the chloro
group to potency. The three dechloro compounds all showed re-
duced potency (ꢀ2 to 7.5-fold) compared to their 6-chloro variants
(Table 3), confirming the favourable interaction of the chloro group
with the S1b site. The effect of larger halogens at the 6-position
was also investigated. Relative to amiloride, iodide 57 and bromide
58 produced 5- and 2-fold improvements in potency, respectively,
suggesting that the larger halogens interact more favourably with
the S1b site. Previous uPA:inhibitor co-crystal structures showing
that the S1b site can accommodate an aryl ring12 led us to explore
amiloride analogs 62 and 63 carrying phenyl and phenylacetylenyl
substituents at C6. The two compounds were found to have equal

6766
H. Matthews et al. / Bioorg. Med. Chem. Lett. 21 (2011) 6760–6766
Removal of the 6-chloro group was investigated with dechloro
References and notes
analogs; 55, 60 and 61. When compared with their 6-chloro coun-
terparts, the three dechloro compounds each showed significantly
reduced potency indicating that placing a group into the S1b
pocket is important. Other analogs substituted at the 6-position
with larger halogens (i.e., 57, 58) and aryl groups (i.e., 62, 63)
showed improved potency, with 57 and 63 being the most potent
compounds identified in the study (ꢀ6-fold increase in potency
relative to amiloride).
1. Duffy, M. J. Biotherapy 1992, 4, 45.
2. Dass, K.; Ahmad, A.; Azmi, A. S.; Sarkar, S. H.; Sarkar, F. H. Cancer Treat. Rev.
2008, 34, 122.
3. Duffy, M. J. Curr. Pharm. Des. 2004, 10, 39.
4. Rømer, J.; Nielsen, B. S.; Ploug, M. Curr. Pharm. Des. 2004, 10, 2359.
5. Nozaki, S.; Endo, Y.; Nakahara, H.; Yoshizawa, K.; Ohara, T.; Yamamoto, E.
Anticancer Drugs 2006, 17, 1109.
6. Tyndall, J. D. A.; Kelso, M. J.; Clingan, P.; Ranson, M. Recent Pat. Anti-Canc. 2008,
3, 1.
7. Ploug, M. Curr. Pharm. Des. 2003, 9, 1499.
Determining the uPA inhibitory potency of 46 and 47 was of
particular interest since these compounds are two of the most
potent NHE1 inhibitors known.17,29 The two compounds were
shown in the current work to provide twofold improvements
in potency relative to amiloride. While not large increases in po-
tency the result confirms that 5-substituted amiloride analogs
with very high potency against NHE1 are able to maintain po-
tency against uPA (relative to the unsubstituted amiloride)—an
important finding in the search for dual uPA/NHE1 inhibitors.
It also supports that the demonstrated antitumor/metastasis ef-
fects of 46 and 47 most likely arise through dual inhibition of
uPA and NHE1.
Analogs of 46 and 47 that retain the respective ethylisopropyl
and hexamethylene amines at C5 while replacing the 6-chloro
group with substituents that are able to make more favourable
contacts in the S1b site may lead to more potent uPA inhibitors
that retain potency against NHE1. It is worth noting that 6-Iodo-
46 and 6-Bromo-47 are 313- and 566-fold more potent NHE1
inhibitors than amiloride, respectively, and 1.40- and 1.08-fold
higher in potency than their 6-chloro counterparts 46 and 47.17
The current study has shown that converting amiloride to the 6-
Bromo or 6-Iodo derivatives enhances uPA inhibition. The com-
pounds 6-Iodo-46, 6-Bromo-47 and derivatives of 46 and 47 that
carry other C6-substituents are currently being examined in these
laboratories in pursuit of dual uPA/NHE1 inhibitors.
8. Croucher, D. R.; Saunders, D. N.; Lobov, S.; Ranson, M. Nat. Rev. Cancer 2008, 8,
535.
9. Look, M. P.; van Putten, W. L. J.; Duffy, M. J.; Harbeck, N.; Christensen, I. J.;
Thomssen, C.; Kates, R.; Spyratos, F.; Fernö, M.; Eppenberger-Castori, S.; Sweep,
C. G. J. F.; Ulm, K.; Peyrat, J.-P.; Martin, P.-M.; Magdelenat, H.; Brünner, N., et al
J. Natl. Cancer Inst. 2002, 94, 116.
10. Rockway, T. W.; Giranda, V. L. Curr. Pharm. Des. 2003, 9, 1483.
11. http://www.wilex.de/R&D/Mesupron.htm (accessed on July 7, 2011).
12. Rockway, T. W.; Nienaber, V.; Giranda, V. L. Curr. Pharm. Des. 2002, 8, 2541.
13. Sweetman, S. C., Ed. Martindale: the complete drug reference [online].
Pharmaceutical
Press:
London.
http://www.medicinescomplete.com/
(accessed on 07 July 2011).
14. Vassalli, J.-D.; Belin, D. FEBS Lett. 1987, 214, 187.
15. Matthews, H.; Ranson, M.; Kelso, M. J. Int. J. Cancer 2011, 129, 2051.
16. Zeslawska, E.; Schweinitz, A.; Karcher, A.; Sondermann, P.; Sperl, S.;
Stürzebecher, J.; Jacob, U. J. Mol. Biol. 2000, 301, 465.
17. Kleyman, T. R.; Cragoe, E. J., Jr. J. Membr. Biol. 1988, 105, 1.
18. Wermuth, C. G. J. Med. Chem. 2004, 47, 1303.
19. Walker, J. B.; Walker, M. S. J. Biol. Chem. 1959, 234, 1481.
20. Showell, G. A.; Gibbons, T. L.; Kneen, C. O.; MacLeod, A. M.; Merchant, K.;
Saunders, J.; Freedman, S. B.; Patel, S.; Baker, R. J. Med. Chem. 1991, 34, 1086.
21. Cragoe, E., Jr.; Woltersdorf, O. W.; Bicking, J. B.; Kwong, S. F.; Jones, J. H. J. Med.
Chem. 1969, 10, 66.
22. Bicking, J. B.; Robb, C. M.; Kwong, S. F.; Cragoe, E. J., Jr. J. Med. Chem. 1967, 10,
598.
23. Chen, G. S.; Wilbur, J. K.; Barnes, C. L.; Glaser, R. J. Chem. Soc., Perkin Trans. 2
1995, 2311.
24. Lei, M.; Tao, X.-L.; Wang, Y.-G. Helv. Chim. Acta 2006, 89, 532.
25. Contreras, J.-M.; Rival, Y. M.; Chayer, S.; Bourguignon, J.-J.; Wermuth, C. G. J.
Med. Chem. 1999, 42, 730.
26. Fish, P. V.; Barber, C. G.; Brown, D. G.; Butt, R.; Collis, M. G.; Dickinson, R. P.;
Henry, B. T.; Horne, V. A.; Huggins, J. P.; King, E.; O’Gara, M.; McCleverty, D.;
McIntosh, F.; Phillips, C.; Webster, R. J. Med. Chem. 2007, 50, 2341.
27. Renatus, M.; Bode, W.; Huber, R.; Stürzebecher, J.; Stubbs, M. T. J. Med. Chem.
1998, 41, 5445.
Acknowledgments
28. Harguindey, S.; Arranz, J. L.; Wahl, M. L.; Orive, G.; Reshkin, S. J. Anticancer Res.
2009, 29, 2127.
29. Vigne, P.; Frelin, C.; Cragoe, E. J., Jr.; Lazdunski, M. Mol. Pharmacol. 1984, 25,
131.
This work was partially funded by a University of Wollongong
URC Small Grant (M. Kelso) and by the Otago Medical Research
Foundation (J. Tyndall). An Australian Postgraduate Award to
H. Matthews is gratefully acknowledged.
30. Arnaud, C. H. Chem. Eng. News 2011, 32.