J. A. Cadieux et al. / Bioorg. Med. Chem. Lett. 22 (2012) 90–95
91
moiety present in 1 led to the corresponding pyridyl analog 2 with
slightly improved potency against DMT1. The benzylidene function
present in 2 imparted several undesirable features, most notably
an extended conjugation motif affording an intense orange color
with 2-benzothiazole or 2-benzimidazole functions (9a–d) affor-
ded modest increases in potency, along with high microsomal
metabolic stability. On the other hand, comparable substitution
with a 2-thiazole ring led to a reduction in potency (9e). The
presence of a 2-pyridmidyl (9g) or 2-isoquinolyl (9i) substituent
was not tolerated, nor was the insertion of a methylene spacer
between the pyrazole and pyridine rings (9f).
as well as Michael acceptor properties at the benzylidene
a-car-
bon. The initial aim of the ensuing hit-to-lead efforts was the
replacement of the pyrazolone 4-benzylidene substituent with
benzamide or aryl functions in order to improve the potency and
ADME properties of 2.
The introduction of a cyclic constraint by means of tethering C-3
of the pyrazole ring and C-2 of the 4-aryl substituent with an ethyl-
ene linker (12a–o, Table 4) was then carried out. In addition to pro-
viding a series of novel analogs for further optimization, it was
hoped that the ensuing partial rigidification would in turn afford
an increase in target potency. In general, these analogs were equi-
potent to their corresponding ‘acyclic’ 4-aryl-5-hydroxypyrazoles
(Tables 2 and 3), less soluble and more metabolically labile (poten-
tially on account of benzylic oxidation at the ethylene linker). The
strong preference for electron-withdrawing substituents observed
for the 4-aryl-5-hydroxypyrazoles 8a–j did not extend to this rigid-
ified series (cf. 12j–o), as the presence of the ethylene linker pre-
vents optimal electronic interaction between the pyrazole and
aryl rings.15 Conversion of the active 2-pyridyl analog 12f into its
corresponding N-oxide 12h was not tolerated, nor was the intro-
duction of a methyl group (12i) to the 4-position of the benzothia-
zole ring in 12b. None of the analogs in Tables 1–4 were found to be
Treatment of the pyrazolones 5 obtained via condensation9 of
arylhydrazines 3 and ethyl acetoacetate 4 (Scheme 1) with aryl iso-
cyanates under basic conditions afforded the 4-amido-5-hydrox-
ypyrazoles 6a–l.10 Pd-catalyzed
a
-arylation of ethyl acetoacetate
4,11 followed by condensation of the resultant
a-aryl-b-ketoesters
7 with a variety of hydrazines led to the formation of 4-aryl-5-
hydroxypyrazoles 8a–i and 9a–j. The constrained analogs 12a–o
were in turn prepared by treating the sodium enolates of the 6-
substituted-2-tetralones 10 with diethyl carbonate12 followed by
suitable aryl or heteroaryl hydrazines.
The 4-amido derivatives 6a–l (for which some conformational
rigidity imparted by intramolecular hydrogen bonding was envis-
aged10b) generally exhibited acceptable solubility in phosphate-
buffered saline and were relatively resistant to metabolism in rat
liver microsome incubations (Table 1). This series was devoid of
appreciable CYP3A4 inhibition (in a recombinant human enzyme
assay), with the exception of the methylenedioxyphenyl analog
6k.13 However, permeability (Caco-2) data suggest that some of
these analogs (e.g., 6d, 6h and 6k) with Papp ratios in excess of 2
may be subject to efflux transport. In terms of potency against
DMT1, a variety of substituents on the 4-aryl ring were tolerated.
Electron-withdrawing substituents (6a, 6b, 6d) were slightly pre-
ferred over electron-donating ones (6g, 6k). Moreover, substitution
at the aryl ring’s 2-position was disfavored over substitution at the
corresponding 3- or 4-positions (cf. 6i vs 6b as well as 6j vs 6a and
6d). Replacement of the 2-pyridyl group at the pyrazole N-1 posi-
tion with a phenyl group (6l) was, however, not tolerated.
The 4-aryl-5-hydroxypyrazoles 8a–j were typically less soluble
and metabolically less stable than the 4-amido-5-hydroxypyraz-
oles 6a–l (Table 2). Analogs in this series were also free of signifi-
cant CYP3A4 inhibition liabilities and certain analogs here also
showed a tendency towards efflux transport. In this series, a
marked preference for electron-withdrawing substituents on the
aromatic ring was noted: the coefficient of determination R2 in a
cytotoxic to HepG2 cells after incubation at 10 l
M for 24 h.16
The calcein quench assay employed in this study and in earlier
investigations of DMT16b is a very sensitive measure of ferrous in-
flux, but inhibition of calcein quench can occur by mechanisms
other than direct effects on the DMT1 transporter. Since ferrous in-
flux involves electrogenic co-transport of protons and ferrous
ions,1e agents which chelate Fe2+, disturb the pH gradient, are re-
dox active or depolarize the cell membrane can all appear active.
One way to establish to establish direct interaction with DMT1 is
to measure the effect of test compounds on ionic current by
whole-cell voltage clamp. This technique allows control of the
intracellular solution and membrane voltage and can rule out all
indirect effects except Fe2+ chelation. In preliminary studies, com-
pounds 6a and 8b were evaluated with this methodology and were
found to be inactive at 10 lM, suggestive of an indirect effect. It
should, however, be noted that unlike some previously reported
antioxidant compounds6b hypothesized to act through modifica-
tion of cellular redox status, the pyrazole analogs in question are
not expected to possess redox activity under physiological
conditions.
In order to rule out the possibility that the majority of the ob-
served activity of these pyrazoles in the calcein quench assay
was attributable to chelation of Fe2+ (e.g., bidentate chelation
through the 5-hydroxy function and the pyridyl N in 8a–j) as op-
posed to intrinsic activity against DMT1, a cell-free competition
variant of this assay was carried out. Test compounds (final well
linear regression of potency (as pIC50
) and the substituent
Hammett constant
r
was 0.815.14 Potency in the 4-aryl series, in
which the 4-substituent is directly bonded to the pyrazole ring,
was far more sensitive to substituent effects than it was for the
4-amido analogs.
In order to probe the SAR at the pyrazole N-1 substituent, a
series of 1-substituted 3-methyl-4-phenyl-5-hydroxypyrazoles
9a–i was then prepared (Table 3). Several heterocycles possessing
concentration 10
were mixed, freshly-prepared Fe2+ (as ferrous ammonium sulfate,
2 lM) was added and the calcein fluorescence was measured after
lM in 12.6% aqueous DMSO) and calcein (1 lM)
heteroatoms
a to the site of linkage to the pyrazole ring were
found to be potent. Replacement of the 2-pyridyl moiety in 8g
incubation for 20 min. As Fe2+ quenches the fluorescence of calcein,
the residual fluorescence persisting after the addition of Fe2+ is a
direct measure of the ability of a compound to chelate Fe2+ away
from calcein. Desferoxamine17 was employed as a positive control.
The pyrazole compounds described herein had responses in this
assay <20% of that of desferoxamine. These data suggest that the
contribution of direct Fe2+ chelation to the observed activity, if
any, was minimal. Moreover, the preference for electron-with-
drawing substituents observed in the 4-aryl-5-hydroxypyrazoles
8a–j is inconsistent with a chelation mechanism; under such a par-
adigm, one would anticipate a preference for electron-releasing
substituents as the latter would increase the Lewis basicity of the
putative chelation sites.18
O
N
N
N
NH
O
N
N
O
N
N
N
N
1
2
IC50 = 2.57 µM
IC50 = 1.53 µM
Figure 1. Structures of the HTS hit 1 and its simplified analog 2.