3824
A. Kaptein et al. / Bioorg. Med. Chem. Lett. 21 (2011) 3823–3827
O
the (hetero)-aryl and the pyrimidyl is essential for proper binding
TMB
TMB
to the hinge and front pocket regions. Despite the low nanomolar
biochemical potency of many of these compounds, substantial cel-
lular inhibition was only obtained if the lipophilicity was above a
certain level (c Log P >1). Examples 16–20 are probably not suffi-
ciently lipophilic to produce a decent cellular effect. Introduction
of aliphatic groups, such as a cyclopentyl in 21 (as compared to
unsubstituted 20), restored the cellular activity somewhat. In-
creased lipophilicity did not necessarily correlate with membrane
permeability as determined in our PAMPA and Caco-2 assays,
and permeability properties were disappointingly low for all 4-
piperidyl analogues (Table 2).
O
NH2
O
HN
N
MeO
MeO
O
a
b, c
e
and d
N
N
N
A
B
C
Boc
Boc
Boc
O
N
O
Cl
N
N
TMB
N
NH
N
N
f and g
N
N
N
N
H
H
Cl
Ar
O
Br
N
H
E
D
3 - 21
We reasoned that the lack of permeability could be the prime
cause of both low cellular activity and absence of oral bioavailabil-
ity in this series, as the microsomal and hepatocyte stabilities of
our MK2 inhibitors were generally very good (Table 2). The basicity
of the unsubstituted 4-piperidyl moiety was determined experi-
mentally; the pKa’s of 1 and 3a were 9.3 and 9.1, respectively (Ta-
ble 2). We therefore sought to lower the basicity with the aim to
improve the above mentioned parameters. In general, pKa values
can potentially be reduced via introduction of (electron-withdraw-
ing) aliphatic substituents. Another, less conventional possibility,
was to move the basic nitrogen closer to the electron-deficient pyr-
role ring. By doing so, a net electron-withdrawing effect could low-
er the pKa. Consistent with this hypothesis, the calculated log P
values of 4- and 3-piperidyl analogues with identical molecular
formulas differed on average by 0.3–0.4 log unit (compare e.g., 1
and 23, and 3a and 22 in Table 3).
Preparation of the 3-piperidyl analogues started with iodome-
thylation of commercially available F (Scheme 2). Subsequent
nucleophilic substitution with sodium azide, and reduction to the
primary amine gave intermediate H. Reductive N-alkylation with
trimethoxybenzaldehyde, and construction of the oxolactam ring
according the methods described for the 4-piperidyl derivatives,
furnished intermediate K. Final test compounds were again pre-
pared in analogy of Scheme 1.
Boc
Scheme 1. Reagents and conditions: (a) 2,4,6-trimethoxybenzaldehyde, NaB-
H(OAc)3, EtOH, rt (100%); (b) ClCOCH2COOEt, DMAP, pyridine, DCM, rt (100%); (c)
NaOMe, MeOH, 60 °C (100%); (d) ACN/H2O (1:1) 80 °C (98%); (e) Compd E, NH4OAc,
EtOH (35%); (f) arylboronic acid, PdCl2 (dppf), K3PO4Á7H2O, dioxane, 140 °C, MW; (g)
TFA, rt (4–48% in two steps).
group on the bridgehead resulting in drastically improved isolated
yields. Also, it provided UV-positive intermediates that were easy
to track during the workup and purification stages. Acylation with
ethyl (chloroformyl)acetate, a Dieckmann condensation and decar-
boxylation subsequently rendered intermediate C. Paal–Knorr con-
densation was carried out with chloropyrimidine derivative E (see
insert Scheme 1),6 followed by a Suzuki-coupling to introduce the
various aryl moieties in the front pocket. Final compounds were
generated by a one-step acidic removal of the Boc- and TMB-pro-
tective groups. The N-alkylated analogues 3b and 3c were synthe-
sized via standard reductive amination procedures on the
unsubstituted 4-piperidyl parent molecule 3a.
The basic amine function seems to be a prerequisite for MK2
inhibitory activity, since the direct pyrane analogue
2
demonstrates a 180-fold drop in activity (Table 1).7 Benzofurane
derivative 3a was equipotent to the methylenedioxophenyl
substitution, but the higher cellular potency observed in the TNF
a
Compared to their basic amine containing equivalents, N-alkyl
analogues 3b (Me) and 3c (Et) exhibited higher lipophilicity, lower
pKa values and improved Caco-2 permeability at the expense of
biochemical activity (Table 2). This appeared to only marginally
improve oral exposure in rats, at least within the benzofuranyl ser-
production in THP1 and PBMC cells could not be confirmed in the
Hsp27 phosphorylation target engagement assay (pHsp27). N-
Alkylation of the 4-piperidyl moiety with a methyl (3b) or ethyl
(3c) in the benzofurane series resulted in a 2- and 10-fold loss in
biochemical potency. As for 3a, the cellular THP1 readouts did
not correlate very well. Plotting the target engagement (pHsp27)
ies, when comparing the 4-piperidyl NH (AUC <0.006
(0.078 M h), N-Et (0.154 M h). A substantial improvement in
oral exposure in this series was obtained through the 3-piperidyl
(0.472 M h) analogue. This trend towards improved oral bioavail-
lM h), N-Me
l
l
and cytokine production (TNFa) EC50 values in THP1 cells, the high
overall correlation of these two parameters becomes evident
(Fig. 2). This graph also clearly shows that the benzofurane-derived
MK2 inhibitors belong to the few outliers. There is a 4- to 10-fold
l
ability was also observed in many other 3-piperidyl analogues, for
example, 23 and 25. For these compounds, Caco-2 permeabilities
exceeded 20 nm/s, and the pKa values ranged from 8.2 to 8.4. There
is a fairly good correlation between lower pKa values, increased
permeability and improved oral exposure following rapid rat
pharmacokinetic experiments. The biochemical EC50 values of the
3-piperidyl analogues were somewhat compromised, but this is
more than compensated for by this series’ PK properties: The
difference in EC50 values obtained in the pHsp27 and TNFa assays,
while all other compounds are within a 2-fold difference from equi
potency on both parameters. This suggests that a promiscuous
component aids to the inhibition of LPS-induced cytokine produc-
tion in THP1 cells. Since the benzofuranyl series did not reveal any
cytotoxicity in THP1 cells, as determined with Alamar Blue in con-
cert with cytokine production (>10
lM for 3a–3c; data not shown),
AUC in rats after oral administration of 23 was 3.2 lM h along with
off-target contribution to the overall efficacy on LPS-induced TNF
a
acceptable half lives in human and rat liver microsomes. The oral
production is anticipated for this particular subseries. The poten-
tial to inhibit p38 was routinely monitored for these analogues,
and no substantial inhibition was observed (data not shown).
While screening for additional substituents on the (hetero)-aryl
portion in the front pocket, it became evident that the nature of
properties that are accepted in this region is very diverse (Table
1). Hydrogen bond accepting (4, 15–17) as well as donating sub-
stituents (18–21) are allowed. Also, small lipophilic groups on
the meta position, or bulkier aliphatic groups on the para position
can be beneficial. In sharp contrast, ortho substitution is not
allowed except for fluorine as in 8, showing that co-planarity of
bioavailability of 23 was confirmed in mice and with F = 48%
(AUC = 3.5
(F = 0.7%). Also, the cellular activities as shown by both pHsp27
and TNF inhibition EC50 values are still in the low micromolar
lM h) its oral exposure is superior to that of 1
a
range (Table 3, compounds 23–26) and comparable to 4-piperidyl
analogue 1.
The generation of 3-piperidyl analogues provided another inter-
esting avenue, namely separating and testing of enantiomers. This
often provides increased potency, and could generate additional
proof as to whether MK2 inhibition is indeed correlated with
cellular inhibition of pHsp27 and TNF
a in THP1 cells. A couple of