S.O.R. Greenwood, et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxxx
Table 3
not resolved, suggesting that it does not bind to 2. Given that amides 5i
and 5l (containing polar OH and NH groups respectively that could
engage in hydrogen bonding) were the most potent in the series shown
in Table 2 (IC50 values close to 1 µM) in which the isoindoline ring is
absent, this supported examining the effect of the introduction of a 5-
substituent in an isoindoline series that might also engage in hydrogen
bonding with Asp101.
In Vitro Inhibition of HDAC8 by Substituted Isoindolinyl α-Amino Amides.
The 5-substituted isoindoline series (Table 3) provided further evi-
most potent compounds, amine 14a (IC50 = 0.13 µM) and amide 14b
(IC50 = 0.10 µM), molecular docking poses (Figs. 2b and 2c) clearly
identified distances consistent with hydrogen bonding to Asp101 (ap-
proximately 2.75 Å in both cases) as well as an aromatic edge-to-face
interaction with Phe152. The NH groups of the 5-NHR substituent act as
hydrogen bond donors to the Asp101 sidechain (Fig. 2). However,
as in the reversed amide 21 and the aminomethyl derivative 26 lowers
the ligand binding energy. In amide 21, the amide is more than 4 Å
from Asp101 and the orientation of the NeH bond departs from the
ideal linear hydrogen-bonded assembly.36
As shown in the crystal structure (PDB: 3SFH),19 the aromatic ring
of isoindoline in 2 is well positioned for edge-to-face interactions with
Phe152 of HDAC8, but that is not the case for the corresponding tet-
rahydroisoquinolinyl analog 5r, being more remote from Phe152
(Fig. 2d); additionally, the different topology of a larger and a non-
planar isoquinoline ring would lead to altered interactions with back-
bone amide residues. Such factors would contribute to the lower
binding energy compared to the isoindoline ring system in 2, as re-
flected by a difference of 5.8 kJ mol−1 estimated from IC50 values. The
lack of aromatic interactions and a relatively planar but non-aromatic
ring also account for the weak binding of the piperazine derivatives 8a
and 8b.
In conclusion, general routes to 5-substituted isoindolines have been
developed, and their conversion into the corresponding α-amino
amides has furnished potent HDAC8 inhibitors. Amide 14b was found
to be highly selective for HDAC8 (IC50 = 0.10 µM), compared to
HDAC1 (IC50 = 8.8 µM), HDAC2 (IC50
> 30 µM), and HDAC6
(IC50 > 30 µM). The presence of the 2,4-dichloro substitution at the
phenyl ring of the α-amino amides was shown to confer a gain of ap-
prox. 4.3 kJ mol−1 in binding energy, being an essential unit in this
series of HDAC8 inhibitors, for binding within the acetate-release
channel.19 The need for a linear, planar and rigid ring system, as in-
dicated by the efficacy of a isoindolin-2-yl linker, was underscored by
its replacement with tetrahydroisoquinoline ring system (approx.
5.8 kJ mol−1 lower binding energy than the isoindoline ring system)
which invariably conferred poorer in vitro inhibition of HDAC8. A final
series of α-amino amides was synthesised and shown to give potent in
vitro inhibition of HDAC8, compounds 14a and 14b being somewhat
more potent than the lead HDAC8 inhibitor 5q in the assay used, with
improved binding energy of about 2.0 kJ mol−1 arising from additional
binding interactions within the acetyl-lysine binding tunnel, including
π-π interactions, particularly involving Phe152, and hydrogen bonding
of a 5-isoindoline substituent to Asp101. The importance of Asp101 in
studies of hydroxamic acid inhibitors of HDAC8.30
Fig. 2. (a) Conformation of α-amino amide 2 in HDAC8 (PDB: 3SFH).19 (b)-(d)
Poses of α-amino amides 14a, 14b and 5r respectively docked into HDAC8
(PDB: 3SFH, see Supplementary Material). In addition to edge-to-face interac-
tions of the aromatic ring of isoindoline, (b) and (c) also indicate hydrogen
bonding to Asp101 of the amine substituent in 14b (in magenta), and the amide
substituent 14c (in pink) respectively. Figure (d) shows the markedly different
pose of 5r (brown) containing a fused piperidine ring compared to 2 (overlaid
in green).
the 2-chloro substituent with Gly140, and of the 4-chloro substituent
with Arg37 also contribute to binding in the acetate-release channel
(Fig. 2a). Arg37 has been shown to regulate the release of acetate.31,32
The 4-chloro substituent interacts side-on to Arg37, a standard mode for
a Lewis base. In contrast, the 2-chloro substituent is end-on to the
This study has evaluated the scope and limitations of indolin-2-yl α-
amino amides as HDAC8 inhibitors, and has provided an analysis of
binding energies of the various structural features along with rationales
for the binding of each major structural subunit. Appropriately sub-
stituted isoindolinyl inhibitors have been shown to confer additional
binding of the HDAC8 inhibitor to the acetyl-lysine binding tunnel, the
substituent acting as a capping group that offers additional molecular
diversity. Consequently, there is scope for the development of more
potent and isoform-selective inhibitors of HDACs containing an α-
amino amide as the zinc-coordinating unit within the active site.
For all the α-amino amides in Table 3, the similar or improved
for as follows. In many HDAC isoforms other than HDAC8 a capping
group that binds to the protein periphery is required for potent in-
hibition. The potential for improved binding by adding a capping group
to 2 is also supported by previous findings that Asp101 is essential for
substrate binding in HDAC8 via the formation of a hydrogen bond.34,35
Such hydrogen bonding involving Asp101 is shown in Figs. 2b and 2c,
6