D.H. O’ Donovan et al.
Bioorganic & Medicinal Chemistry Letters 39 (2021) 127904
1
◦
◦
◦
Scheme 1. Synthesis of compound 2 with R = F. (a) iPrMgCl, B(OCH
3
)
3
, H
2
O
2
, THF, ꢀ 20 C to rt, 2 h, 100% (b) Cs
2
◦
CO
3
, 3-bromoprop-1-ene, DMF, 60 C, 2 h, 86%
◦
◦
(
c) o-DCB, 180 C, 16 h, 86% (d) K
2
OsO
4
⋅2H
2
O, NaClO
4
, MeCN, Acetone, H
2
O, 0 C, 2 h, 42% (e) NaBH
4
, MeOH, 0 C, 1 h, 99% (f) PPh
3
, DIAD, THF, 0 C, 1 h, 92%
◦
(
g) Pd
2
(dba)
3
, Zn(CN)
2
, Xantphos, TMEDA, DMF, 130 C, 16 h, 70% (h) H
2
atm, 10 mol% Pd/C, Boc
2
O, MeOH, rt, 16 h, 93% (i) HCl, dioxane, MeOH, rt, 2 h, 92% (j)
◦
◦
neat, 60 C, 48 h, 37% (k) 2-methyl-3-(Bpin)-pyridine, Xphos Pd G2, K
2
CO
3
, dioxane, H
2
O, 100 C, 3 h, 64%.
synthetic investment, with a step count of 11 transformations as the
longest linear sequence and an overall yield of just 4% (Scheme 1).
The low yield and lengthy route incurred in the preparation of 2
prompted a revised synthetic approach. Subsequent analogues were
prepared following the principles of late-stage functionalisation (LSF) by
introducing the required functionality on the common synthetic inter-
mediate 15 (Scheme 2) which is either commercially available or can be
synthesised on multi-gram scale from inexpensive 3-bromo-4-fluorophe-
prepared compound 22 with R1 = OMe which was predicted to bind
poorly.
To expedite their chemical synthesis, the second round of com-
pounds was prepared using the iridium-catalyzed CH borylation chem-
istry developed by Hartwig et al. (Scheme 2).21 Boc protection of 15
followed by borylation provided a versatile synthetic intermediate 25 as
the major regioisomer which could be transformed into the target
compounds using established organoboron chemistry. Thus, compound
1
20 with R1 = NH
nol. Thus, compound 16 with R = Cl was synthesised via the N-chlor-
2
was accessible via copper-mediated amination.
1
osuccinimide mediated chlorination of trifluoroacetate-protected amine
Compound 21 with R = OH was prepared via oxidative hydroboration
using an O-allyl group to protect the nascent alcohol, which underwent
spontaneous palladium-mediated deprotection during the final Suzuki
1
7; the chlorinated product could then be transformed into the target
molecule in just two additional steps.
Gratifyingly, in agreement with our FEP calculations, compounds 2
and 16 (R1 = F, Cl) afforded potencies comparable with the parent
coupling. Compound 19 with R1 = CN was prepared using Zn(CN)
2
,
while the methoxy compound 22 was easily accessible via methylation
of phenol intermediate 28.
1
compound 1 with R = H (Table 1), thereby confirming that substitution
at the C7 position could be tolerated. Given the success of these initial
predictions, we explored a wider set of analogues in a second FEP
campaign, once again restricting our search to groups which would
explore the steric constraints in this environment. In an effort to address
the limited solubility of compound 1, we also incorporated polar sub-
stituents in this second round of calculations. Thus, compounds 20 and
In agreement with our predictions, compound 20 (R1 = NH
2
) ex-
hibits excellent affinity by SPR comparable with compound 1 (Fig. 2),
albeit with slightly reduced potency in the cellular trimethylation assay.
Furthermore, this compound improves solubility 14-fold versus the 1
while retaining favorable in vitro metabolic properties (Table 1). For
compound 21 (R1 = OH), the cell potency is within 3-fold of the
unsubstituted analogue 1 and the lower logD results in a net improve-
1
1
2
1 (R = NH
2
and OH) were predicted to be roughly equipotent to R =
1
22
H, while the analogue with R = CN was predicted to be slightly weaker
ment in lipophilic-ligand efficiency (LLE). Although the solubility is
1
and within ~ 10-fold of the starting compound with R = H. In contrast,
improved versus 1, the OH compound 21 also suffers from increased
clearance in rat hepatocytes, likely due to oxidative metabolism of the
phenol moiety.23 Compound 19 with R = CN was also well predicted by
FEP and incurred a 20-fold reduction in cellular potency, a result which
may be partially explained by the decrease in logD which may limit
lipophilic interaction with the protein. Unfortunately despite this
1
examples with R = CF
3
, OMe and cyclopropyl were predicted to have
1
much lower affinity (compounds 1b, 1c and 22, Table 1), with a > 10-
1
fold reduction compared to compound 1 with R = H. On this basis we
selected compounds with R1 = CN, NH
, and OH for synthesis (com-
2
pounds 19, 20 and 21 respectively). As a negative control, we also
3