2722
J. G. Barlind et al. / Bioorg. Med. Chem. Lett. 23 (2013) 2721–2726
Table 1
for the MGAT2 affinity, no correlation between the pKa (Table 3)
Structure activity relationship in R1 & R2/R3
and the potency was observed. This is in agreement with the ear-
lier discussed tolerance observed towards both electron rich and
electron deficient aromatic rings in R1.
O
O
O
R1
S
NH
R2
R3
N
One main focus of our lead generation program was to identify
compounds for the in vivo model, the mouse oral lipid tolerance
test (OLTT), to validate the ability of an MGAT2 inhibitor to sub-
stantially reduce the plasma TAG levels. An early attempt to use
compound 4 failed due to limitations in exposure at high dose.
The permeability for this particular compound, as measured over
the Caco-2 cells (Table 3), predicts the fraction absorbed to be rel-
atively high. Further, 4, as well as the majority of the described
compounds in the series show low to medium metabolic turnover
in both human and rodent liver microsomes and hepatocytes (HLM
in Table 3). Still a poor bioavailability was observed, which was
rationalized to be due to the low crystalline solid solubility of
H
N
H
O
R1
R2
R3
hMGAT2 pIC50
LLEb
LogD7.4
a
Compd
1
2
3
4
5
6
7
8
Ph
Bn
Et
Et
Et
Et
Et
Et
Et
Et
Et
Et
Et
Et
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
H
6.4
5.1
5.3
7.6
5.7
7.2
7.4
8.4
7.6
8.0
8.2
7.2
5.0
5.3
7.3
7.1
7.3
8.5
3.8
2.3
2.3
4.1
2.3
4.3
3.8
5.7
4.5
5.4
5.6
4.6
2.8
3.1
4.2
4.4
4.0
5.3
1.9
1.8
2.0
2.9
—
1.9
3.1
1.8
2.3
1.8
Cyclo-hexyl
p-Cl-Ph
m-Cl-Ph
o-Cl-Ph
p-CF3-Ph
p-OMe-Ph
p-OCHF2-Ph
o,p-DiF-Ph
o,p-DiF-Ph
o,p-DiF-Ph
Ph
9
2
l
M measured for this compound.
The crystallinity of the compounds in the series was examined.
Relative high melting points (e.g., (S)-10: 228–230 °C, crystalline
solid solubility 68 M) indicate a strong H-bonding in the crystal
10
(S)-10
(R)-10
11
12
13
14
15
16
Me
Et
1.6
1.6
2.2
—
3.0
2.2
l
Ph
lattice. The X-ray structure of (R)-10 confirmed the earlier de-
scribed strong intermolecular H-bond network between the HBD/
HBA pairs of the amides in this bicyclic core.7 In addition, our X-
ray structure disclosed that the sulphonamide in itself was in-
volved in a zigzag intermolecular H-bond pattern to the next row
of molecules in the crystal packing, which potentially adds to the
strength of the crystal packing. The crystalline solid solubility be-
came a key parameter which was monitored in the series going for-
ward, since structural modification to reduce the number of HBDs/
HBAs had a profound effect on the MGAT2 activity.
Permeability as measured over the Caco-2 cells was found to be
in the moderate range for a majority of the compounds in the ser-
ies. A high number of HBDs is known to impair permeability and
indeed, Papp increased more than what could be ascribed from just
adding lipophilicity in all cases where the HBD was removed by
methylation (17–19) (Table 3).8 Fortunately, potent compounds
could be identified from the current scaffold where both the solid
solubility and the Papp were high enough to give satisfactory expo-
sure in the in vivo OLTT target validation. Additional attempts to
modulate Papp in the series is to be disclosed elsewhere.
The described compounds show no CYP inhibition nor any time
dependence on CYP inhibition observed. Metabolite ID experi-
ments revealed that the major site of metabolism was the aniline
moiety R1. Despite the low turnover of these compounds, substan-
tial levels of glutathione adducts were formed in incubations with
glutathione supplemented human liver microsomes. With the ma-
jor metabolic pathways proceeding via reactive metabolites, there
is a substantial risk for a high reactive metabolite body burden and
we therefore designed compounds to understand and address this
issue.9 The levels of glutathione adducts formed, expressed as a
fraction of the glutathione adduct of clozapine incubated in the
same experiment (glutathione ratio), was dependent on the substi-
tution pattern on the aniline moiety R1. For 1, hydroxylation in the
para-position followed by oxidation to a quinoneimine reactive
species, resulted in a glutathione ratio of 0.1. Following O-demeth-
ylation of 8, the same reactive quinoneimine species gave a very
high glutathione ratio (2.2). The rational for this large increase
would be that the O-demethylation of 8 was much more rapid than
the para-hydroxylation on 1. The presence of two fluorine atoms
on the O-methyl moiety in 9 slowed down the rate of the O-
demethylation, but it was not sufficient to prevent the formation
of the quinoneimine species. Substantial levels of glutathione ad-
ducts were still detected (ratio 0.5). Noteworthy, introduction of
o,p-fluorines on the aniline moiety to give compound 10 gave the
same levels of glutathione adducts as compound 1 and higher lev-
els (ratio 1.8) for the more lipophilic analogue 22. The introduction
o,p-DiF-Ph
p-Cl-Ph
p-Cl-Ph
o,p-DiF-Ph
Pr
Me
c-Butyl
c-Pentyl
Et Et
a
pIC50 value from the human MGAT2 (hMGAT2) Rapidfire LCMSÒ assay.
LLE = pIC50ÀclogP.
b
ortho- and/or para-positions of the aromatic ring, whereas the
meta-position was less tolerated, exemplified by mono-chloro ana-
logues (4–6). Both electron donating and electron withdrawing
substituents were found to be well tolerated (4, 7–9). Evaluation
of a large variation of substituted anilines concluded that o,p-di-
fluoro phenyl (10) was among the best substitution patterns found
for R1.
For the more promising racemic compounds their correspond-
ing two enantiomers were isolated, either by separation of the final
compound or by separation of one of the intermediates in the syn-
thetic route, using chiral chromatography. In general, a 10-fold dif-
ference in potency was seen between the two enantiomers of the
3-Et/Me compounds in the series. X-ray crystallography was used
to determine the absolute configuration to be (R) for the least po-
tent enantiomer (+)-10.6 MGAT2 hence showed a preference for
the (S)-configuration at the chiral centre in the Et/Me analogues
of this series.
Concomitant with the determination of the stereoisomeric pref-
erence in the 3-position of the 7-membered ring, we undertook a
more extensive investigation of the SAR in this region. It was soon
realized that the affinity for MGAT2 quickly dropped when the
substituents were modified even slightly from the initial Et/Me
in 1, Table 1 (11–14). The achiral analogues with either a cyclopen-
tyl ring as in 15 or more preferentially a diethyl substitution pat-
tern as in 16 were regarded as viable alternatives to Et/Me,
although the gained affinity for 16 versus 4 was rationalized to
be predominantly lipophilicity driven as LLE is not improved.
To further explore the series, the H-bond donors (HBDs) and
acceptors (HBAs) in the scaffold and the sulphonamide linker were
investigated in a systematic way (Table 2). Methylation of the var-
ious NH positions turned out to be detrimental to the potency
against MGAT2 (17–19). Addressing the HBAs by removing, for
example, the 5-carbonyl was not accepted (20–21) whereas the
2-carbonyl was shown not to be required and 22 has a pIC50 of
8.8. The ether analogue 23 confirmed the necessity of having a
HBD in the 1-position. Attempts to replace the sulphonamide with
an amide (24) or reverse sulphonamide (25) gave inactive com-
pounds. Even though the sulphonamide NH appears to be critical