method to search for novel hits. However, in many cases the identified inhibitors
were not optimal due to weak inhibitory activity or unspecific mode of action
[49]. Below, several VS campaigns with biologically validated hits suitable for
further optimization are described.
Price et al. published a study in which they docked a virtual library of
644 hydroxamic acids, generated from available in-house carboxylic acids to a crystal
structure of bacterial HDAC-like protein [50]. Based on the docking poses and scores
obtained with FlexX docking program, 75 compounds were selected for biological
testing. A promising hit, namely, ADS100380 (3, Fig. 1), was identified with an IC 50
of 0.75 μM in an in vitro HDAC assay and a weak functional activity (IC 50 of
11.4 μM) in a cell proliferation assay. Structure-based optimization of the hit guided
by molecular docking studies resulted in a series of compounds with improved
potency, showing up to single-digit nanomolar activity in a HDAC assay and
double-digit nanomolar activity in an antiproliferative assay (4, Fig. 1) [50].
Tang et al. reported their quantitative structure-activity relationship (QSAR)based VS implemented to search for human HDAC1 inhibitors. Chemical descriptors were calculated for 59 known HDAC inhibitors to generate QSAR models,
which were used to screen a virtual library of 9.5 million compound structures. Out
of the 45 retrieved screening hits, 4 structurally diverse compounds were selected for
experimental validation. Three of them, which were hydroxamic acids, showed
micromolar activity against HDAC1 and HDAC6 [51].
Park et al. identified novel HDAC inhibitors by applying structure-based VS
[52]. A virtual library of around 460,000 compounds was pre-filtered by Lipinski’s
rule of five [53] and structural similarity. The remaining 180,000 compounds were
docked to a homology model of HDAC1 using AutoDock program and scored with a
modified scoring function which included a novel solvation model. Around
100 top-scored virtual hits were selected for pretesting in a HDAC inhibition
assay. The IC 50 values of the six final hits ranged between 4 and 100 μM. Interestingly, the most active hit (5, Fig. 1) was a non-hydroxamate HDAC inhibitor with a
putative thiadiazole sulfonamide zinc-binding group [52].
Another successful structure-based VS was conducted by Kannan et al.
[54]. The ZINC database of around 15 million compounds was pre-filtered to
select only compounds containing one of several known zinc-binding groups
(hydroxamic acid, anilinobenzamide, thiazole sulfonamide). About 5,000 retrieved
zinc binders were docked and scored with Glide (Glide SP scoring function) and
GOLD (GoldScore, ChemScore, ChemPLP, and ASP scoring) docking programs
using a homology model of the anti-parasitic target Schistosoma mansoni HDAC8
(smHDAC8). This docking protocol was previously validated by re-docking of the
cocrystallized human HDAC8 inhibitors. Based on the docking poses and calculated scores, 75 compounds were purchased and tested on smHDAC8. For the six
most active hydroxamic acid hits, the IC 50 values were determined on smHDAC8
(1–6 μM), as well as human HDAC8 (0.6–3 μM), HDAC1 (3–30 μM), and
HDAC6 (0.02–4 μM) [54]. Two of the most promising fragment-like compounds
T5979345 (6) and T6072858 (7, Fig. 1) could be cocrystallized with the
target protein [55]. Based on the binding mode of 6, a library of open-ring analogs
Structure-Based Design of Epigenetic Inhibitors
461
were not optimal due to weak inhibitory activity or unspecific mode of action
[49]. Below, several VS campaigns with biologically validated hits suitable for
further optimization are described.
Price et al. published a study in which they docked a virtual library of
644 hydroxamic acids, generated from available in-house carboxylic acids to a crystal
structure of bacterial HDAC-like protein [50]. Based on the docking poses and scores
obtained with FlexX docking program, 75 compounds were selected for biological
testing. A promising hit, namely, ADS100380 (3, Fig. 1), was identified with an IC 50
of 0.75 μM in an in vitro HDAC assay and a weak functional activity (IC 50 of
11.4 μM) in a cell proliferation assay. Structure-based optimization of the hit guided
by molecular docking studies resulted in a series of compounds with improved
potency, showing up to single-digit nanomolar activity in a HDAC assay and
double-digit nanomolar activity in an antiproliferative assay (4, Fig. 1) [50].
Tang et al. reported their quantitative structure-activity relationship (QSAR)based VS implemented to search for human HDAC1 inhibitors. Chemical descriptors were calculated for 59 known HDAC inhibitors to generate QSAR models,
which were used to screen a virtual library of 9.5 million compound structures. Out
of the 45 retrieved screening hits, 4 structurally diverse compounds were selected for
experimental validation. Three of them, which were hydroxamic acids, showed
micromolar activity against HDAC1 and HDAC6 [51].
Park et al. identified novel HDAC inhibitors by applying structure-based VS
[52]. A virtual library of around 460,000 compounds was pre-filtered by Lipinski’s
rule of five [53] and structural similarity. The remaining 180,000 compounds were
docked to a homology model of HDAC1 using AutoDock program and scored with a
modified scoring function which included a novel solvation model. Around
100 top-scored virtual hits were selected for pretesting in a HDAC inhibition
assay. The IC 50 values of the six final hits ranged between 4 and 100 μM. Interestingly, the most active hit (5, Fig. 1) was a non-hydroxamate HDAC inhibitor with a
putative thiadiazole sulfonamide zinc-binding group [52].
Another successful structure-based VS was conducted by Kannan et al.
[54]. The ZINC database of around 15 million compounds was pre-filtered to
select only compounds containing one of several known zinc-binding groups
(hydroxamic acid, anilinobenzamide, thiazole sulfonamide). About 5,000 retrieved
zinc binders were docked and scored with Glide (Glide SP scoring function) and
GOLD (GoldScore, ChemScore, ChemPLP, and ASP scoring) docking programs
using a homology model of the anti-parasitic target Schistosoma mansoni HDAC8
(smHDAC8). This docking protocol was previously validated by re-docking of the
cocrystallized human HDAC8 inhibitors. Based on the docking poses and calculated scores, 75 compounds were purchased and tested on smHDAC8. For the six
most active hydroxamic acid hits, the IC 50 values were determined on smHDAC8
(1–6 μM), as well as human HDAC8 (0.6–3 μM), HDAC1 (3–30 μM), and
HDAC6 (0.02–4 μM) [54]. Two of the most promising fragment-like compounds
T5979345 (6) and T6072858 (7, Fig. 1) could be cocrystallized with the
target protein [55]. Based on the binding mode of 6, a library of open-ring analogs
Structure-Based Design of Epigenetic Inhibitors
461