4574
K. J. Filipski et al. / Bioorg. Med. Chem. Lett. 23 (2013) 4571–4578
Glucokinase at 20 mg/mL was crystallized in the presence of
N
40 mM glucose and 1 mM 1 or 2. Crystallization drops in a 1:1 ratio
were set up over wells containing 0.1 M TrisHCl pH 7.0,
80–200 mM glucose, and 19–26% PEG-4000. Structures were
solved using molecular replacement method. Human liver micro-
somal clearance (HLM),14 rat liver microsomal clearance (RLM),
kinetic aqueous solubility, passive permeability flux performed in
a low transporter-expressive cell line, MDCKII-LE,15 and logD
performed by the shake flask method, were also used to further
characterize final compounds. The kinetic solubility assay was
performed as follows: 50 mM sodium phosphate buffer, at pH
6.5, was freshly prepared from NaH2PO4 and Na2HPO4 and filtered.
a
O
N
OPMB
15
OH
28
Br
N
N
N
N
b
N
O
N
H
O
N
O
29
Buffer (294 lL) was combined with 6 lL of 30 mM DMSO stock
O
N
solution in a Millipore polycarbonate filter solubility plate (part#
MSSLBPC10) for a final DMSO concentration of 2%. The plates were
heat sealed with a polypropylene seal. After 24 h of 200 rpm
shaking at room temperature (23–25 °C), the plates were vacuum
filtered into deep well receiver plates. Filtrates were injected into
a chemiluminescent nitrogen detector (Antek 8060) for quantifica-
tion. Apparent concentration was divided by the number of nitro-
gens in the sample (from the molecular formula) to determine the
concentration.
O
30
Scheme 5. Synthesis of representative analog 30. Reagents and conditions: (a) (i)
11, Pd(dppf)Cl2, Na2CO3, 3:1 DME:H2O, 70 °C, 19 h; (ii) TBAF (1 M in THF), THF,
23 °C, 90 min, 37%; (b) (i) 29, CuCl, 2,2,6,6-tetramethylheptane-3,5-dione, Cs2CO3,
DMF, 120 °C, 15 h, sealed tube; (ii) TFA, CH2Cl2, 23 °C, 10 min, 65%.
Our design hypothesis of replacing the heterocyclic amide with
a pyridone was confirmed with 4 (Table 1) having sub-micromolar
potency, while reducing molecular weight and increasing polarity
as compared to amide 3, leading to a modestly higher lipophilic
efficiency (LipE).16 Several other properties also were improved,
namely passive permeability, kinetic solubility, and human micro-
N
TBDMSO
Br
b
HO
a
N
OPMB
18
OTBDMS
OH
OH
somal clearance. Gratifyingly, this compound also had a higher
a
32
31
(i.e., less substantial reduction in the enzyme’s Km for glucose),
indicating the potential for increased glucose dependency relative
benchmark candidate 3, as discussed above.
(R)
(R)
c
OH
d
A nitrogen-walk around the pyridine of 4 was first performed.
The pyrimidin-2(1H)-one, 17, and pyrazinone, 21, were much less
active, while the pyrimidin-4(3H)-one, 25, was equipotent with
still desirable activation profile. Since the logD was reduced from
the pyridone lead, this compound had improved LipE, as well as
improved permeability and solubility, and so was deemed to be a
superior motif and was used in subsequent optimization. Based
on the crystal structure and the overlap with the heterocyclic
amide motif it was envisioned that building off the 5-position
would be sterically tolerated while the 6-position appeared less
promising. This hypothesis played out with the 5-methyl analog,
26, having improved potency and ligand efficiency (LE = 0.33 for
26 vs 0.31 for 25),17 and maintained LipE; while the 6-methyl
homolog, 27, lost significant activity. Therefore additional substitu-
tions were examined in the 5-position as shown in Table 2.
An SAR trend with small alkyl substituents, 26, 37, and 38,
shows that ethyl, 37, is optimal for potency. Replacing the ethyl
with the isosteric methoxy group, 39, shows a loss in potency.
Noteworthy, is that electron-donating substituents, as in 26 and
37–39, are significantly more potent than electron-withdrawing
substituents 40 and 41, even for the similarly sized and lipophilic
chloro compared to methyl or ethyl. This suggested the importance
of the strength of the H-bonding between the pyrimidone carbonyl
and NH with the binding site residue. Specifically, the anticipated
pKa of the pyrimidone NH with an electron-withdrawing group
present is low enough where at physiological pH a significant per-
centage of the deprotonated compound will be present; eliminat-
ing the H-bond interaction and creating an electronic repulsion.18
Exploration of SAR around the O-aryl ring is shown in Table 3.
Several heterocycles and substituents were incorporated as was
done in related series.10,19 Sulfone (25), sulfonamide (42), and
amide substituents (30, 43, and 44) all showed good to moderate
activity, while large improvements in LipE and kinetic solubility
were demonstrated with pyridyl (43), pyrimidyl (30), and
TiPSO
HO
33
34
N
(S)
O
N
(S)
HO
N
H
O
O
e
HO
N
OPMB
O
OH
S
O
35
O
36
Scheme 6. Synthesis of representative analog 36. Reagents and conditions: (a)
TBDMS-Cl, imidazole, DMF, 23 °C, 3d, 89%; (b) (i) tBuLi (1.7 M in pentane), B(OMe)3
THF, ꢁ78 to 23 °C, 16 h; (ii) 11, Pd(dppf)Cl2, Na2CO3, DME, H2O, 65 °C, 16 h; (iii)
TBAF, THF, 23 °C, 1 h, 59%; (c) TiPS-Cl, DIEA, imidazole, CH2Cl2, DMF 23 °C, 2d, 84%;
(d) PPh3, DIAD, THF, 23 °C, 24 h, 33%; (e) (i) 1-(ethylsulfonyl)-4-fluorobenzene,
K2CO3, DMF, 80 °C, 3 h; (ii) TFA, 23 °C, 1 h.
concentrations (0–100 lM) and 16 different glucose concentra-
tions (0–100 mM). A nonessential activator model was utilized to
determine an activator’s maximum fold effect on reducing the glu-
cokinase Km for glucose, defined as
on altering the Vmax, defined as b.13 The
glucokinase Km at maximum activator concentration divided by
the enzyme’s Km in the absence of activator. Values of range from
0 to 1 with lower values representing more substantial reductions
in the enzyme’s glucose Km. The b value is defined as the ratio of
the enzyme’s velocity at maximum activator concentration divided
by the enzyme’s velocity in the absence of activator. Values >1
indicate increases in the Vmax with activator present, b = 1 indicates
no effect on Vmax, and values <1 indicate suppression of the Vmax
with activator present. The EC50 was formally defined as the con-
centration of activator affording a half-maximal reduction in Km.
a
, and the maximum fold effect
a
value is the ratio of the
a