Structural basis for the design and synthesis of selective HDAC inhibitors

Article abstract of DOI:10.1016/j.bmc.2013.04.036

Histone Deacetylases are considered promising targets for cancer epigenetic therapy, and small molecules able to modulate their biological function have recently gained an increasing interest as potential anticancer agents. In spite of their potential application in cancer therapy, most HDAC inhibitors unselectively bind the several HDAC isoforms, giving rise to different side-effects. In this context, we have traced out the structural elements responsible of selective binding for the therapeutically relevant different HDAC isoforms. The structural analysis has been carried out by molecular modeling, docking in the binding pockets of HDAC1-4 and HDAC6-8, 36 inhibitors presenting a well defined selectivity for the different isoforms. As quick proof of evidence, we have designed, synthesized and experimentally tested three selective ligands. The experimental data suggest that the obtained structural guidelines can be useful tools for the rational design of new potent inhibitors against selected HDAC isoforms.

Full text of DOI:10.1016/j.bmc.2013.04.036

Accepted Manuscript
Structural basis for the design and synthesis of selective HDAC inhibitors
Simone Di Micco, Maria Giovanna Chini, Stefania Terracciano, Ines Bruno,
Raffaele Riccio, Giuseppe Bifulco
PII:
S0968-0896(13)00362-3
DOI:
http://dx.doi.org/10.1016/j.bmc.2013.04.036
BMC 10776
Reference:
To appear in:
Bioorganic & Medicinal Chemistry
Received Date:
Revised Date:
Accepted Date:
11 January 2013
9 April 2013
10 April 2013
Please cite this article as: Di Micco, S., Chini, M.G., Terracciano, S., Bruno, I., Riccio, R., Bifulco, G., Structural
basis for the design and synthesis of selective HDAC inhibitors, Bioorganic & Medicinal Chemistry (2013), doi:
http://dx.doi.org/10.1016/j.bmc.2013.04.036
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Graphical Abstract
Structural Basis for the design and synthesis
of selective HDAC inhibitors
Leave this area blank for abstract info.
Simone Di Micco, Maria Giovanna Chini, Stefania Terracciano, Ines Bruno, Raffaele Riccio, and Giuseppe
Bifulco*
Dipartimento di Scienze Farmaceutiche e Biomediche, Università degli Studi di Salerno, Via Ponte Don
Melillo, 84084 Fisciano (SA), Italy

Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com
Structural basis for the design and synthesis of selective HDAC inhibitors
Simone Di Micco, Maria Giovanna Chini, Stefania Terracciano, Ines Bruno, Raffaele Riccio, and Giuseppe
Bifulco*
Dipartimento di Farmacia, Università degli Studi di Salerno, Via Ponte Don Melillo, 84084 Fisciano (SA),Italy
ARTICLE INFO
ABSTRACT
Article history:
Received
Received in revised form
Accepted
Histone Deacetylases are considered promising targets for cancer epigenetic therapy, and small
molecules able to modulate their biological function have recently gained an increasing interest
as potential anticancer agents. In spite of their potential application in cancer therapy, most
HDAC inhibitors unselectively bind the several HDAC isoforms, giving rise to different side-
effects. In this context, we have traced out the structural elements responsible of selective
binding for the therapeutically relevant different HDAC isoforms. The structural analysis has
been carried out by molecular modeling, docking in the binding pockets of HDAC1-4 and
HDAC6-8, thirty six inhibitors presenting a well defined selectivity for the different isoforms.
As quick proof of evidence, we have designed, synthesized and experimentally tested three
selective ligands. The experimental data suggest that the obtained structural guidelines can be
useful tools for the rational design of new potent inhibitors against selected HDAC isoforms.
Available online
Keywords:
Antitumor agents
HDAC selective inhibitors
histone deacetylase
molecular modeling
DFT
2009 Elsevier Ltd. All rights reserved.
1. Introduc1tion
Please, insert Table 1
Gene expression is highly regulated by post-translational
modifications of histone proteins, including acetylation,
methylation, phosphorylation, ubiquitylation, sumoylation, ADP-
ribosylation, glycosylation, biotinylation and carbonylation.1 In
particular, histone acetyl groups are removed by metalloenzymes
called Histone Deacetylases (HDACs).1 These enzymes, encoded
by a family of 18 genes, are grouped into four separate classes.1
Class I includes HDAC1-3 and HDAC8, which are localized in
the nucleus, while class II HDACs, subdivided in Class IIA and
Class IIB, is formed by member proteins shuttling between the
nucleus and in the cytoplasm. Class IIA includes HDAC4,
HDAC5, HDAC7, and HDAC9, and Class IIB comprises
HDAC6 and HDAC10. The third class of HDACs are the human
sirtuins (SIRT), that are NAD+ dependent enzymes.2 There are
seven human SIRT proteins (SIRT1–7) whose cellular
localization include various organelles depending on the function
of each protein.3 Class IV only comprises HDAC11, located in
the nucleus.
Since the increased activity of HDACs is associated with
carcinogenesis and tumor progression, HDAC inhibitors have
been attracting the attention of the researchers as promising
anticancer agents.5,6 Several bioactive small molecules of natural
or synthetic origin, have been investigated as HDAC inhibitors,7,8
and the suberoylanilide hydroxamic acid (SAHA)9 and FK228
(Istodax)10 has been approved by the Food and Drug
Administration for the treatment of cutaneous
T
cell
lymphoma.9,10 Besides SAHA, other inhibitors have also
demonstrated a therapeutic potential use. Most of the currently
HDAC inhibitors, tested in clinical trials, are rather unselective,
inhibiting either all or at least several members of the HDAC
family. Clinical studies show that pan-HDAC inhibitors may give
rise to numerous side effects:5,10 bone marrow depression,
diarrhoea, weight loss, taste disturbances, electrolyte changes,
disordered clotting, fatigue, and cardiac arrhythmias. However,
the toxicity profile is different for the various pan-HDAC
inhibitors, giving rise to different and controversial opinions of
their use in cancer therapy. Moreover, the very efficient anti-
tumour effect of HDAC inhibitor should be attributed to the
simultaneous inhibition of the different isoforms by blocking
different functions of cancer cells. These undesirable effects
could be ascribed to the important role of HDACs to regulate
different proteins involved in diverse biological processes.5 For
this reason, selective HDAC ligands may be preferred to pan
inhibitors in therapeutic applications.12 Thus, the next step in the
development of HDAC inhibitors is to target selectively
It has been observed that overexpression of HDACs is
correlated to cancerous pathologies,4 and the different isoforms
of HDAC are expressed in several tumor tissues with specific
biological function (Table 1).5
———
*Corresponding author. Tel.: +39 089969741; fax: +39 089969602.
E-mail address: bifulco@unisa.it (G. Bifulco).

individual HDAC isoforms, with the aim of interfering with
critical oncogenic function in cancer cells and without affecting
normal cells. It should be also verified if selective inhibitors
maintain the efficacy of pan inhibitors and present reduced side
effects. In this context, as the enzymes work in cells with other
protein partners, it is difficult to device small molecules able to
selectively bind HDACs.13 Indeed, HDACs are the catalytic
centre of multiprotein clusters.13 In literature, studies devoted to
develop selective ligands for specific isoforms are reported, even
though such studies only consider a restricted number of
isoforms.14 Moreover, most of the published works focused on
selective ligands are only based on biological profiles, lacking
the structural investigation aimed to disclose the ligand and
protein elements potentially responsible of class and isoform
selectivity.15
and HDAC6 homology models building. HDAC6 differs from
other HDACs, for the presence of two catalytic domains
(HDAC6 CD I and HDCA6 CD II) sharing 46% sequence
identity and 60% similarity. In general, CD1 and CD2 show the
same relevant amino acid residues in the active site, whereas
more differences can be observed in the loop regions. A recent
study, using natural and synthetic substrates, showed that the
second catalytic site is the major functional domain of HDAC6.23
In particular, the authors demonstrated that the inhibition of
HDAC6 can be solely ascribed to the interaction of ligands with
the second binding domain. Moreover, recently, docking studies
on HDAC6 have been performed using the second catalytic
domain of HDAC6.24 The resulting alignments were used as
input for the automated homology modeling program
MODELER.25 The number of generated loops was set to five
along with high optimization level for models and loops. The
generated models of HDAC1 and HDAC6, showing the lowest
energy and minor number of restraint violations, were selected.
On the obtained homology models, hydrogen atoms were added
by using the graphical interface Maestro version 6.0,
Schrödinger, LLC, New York, NY, 2003. The charges of side
chains were assigned considering their pKa at physiological pH
of 7.4. In particular, Arg and Lys were positively charged,
whereas Asp and Glu were negatively charged. The His were
neutral. The geometry of the added hydrogen atoms by OPLS
force field [26] and steepest descent method (500 steps and
convergence threshold 0.5 kJ mol-1 Å-1) by using the
MacroModel 8.5.27 The quality of the obtained models for
HDAC1 and HDAC6 (see supplementary information) were
validated using the software PROCHECK.28
On this basis, in the present contribution, we have traced out
the structural elements responsible of selective binding in the
whole landscape of the therapeutically relevant HDAC isoforms
(Scheme 1).
Please, insert Scheme 1
In particular, we have tried to rationalize a number of
experimental observations and tried to systematically add new
structural insights for a targeted design of selective inhibitors of
the different HDAC isoforms, focusing our attention on HDAC1-
4, and HDAC6-8. HDAC9-11, for which few information so far
are available in literature on expression and function in tumor
cells, and HDAC5, missing a concrete ligand inhibitory profile,5
have not been considered in our investigation.
Our structural analysis was performed by molecular docking
calculations, using as ligands pan and class selective HDAC
inhibitors reported in the literature (1-36, Scheme 2), presenting a
well defined profile of HDACs inhibition. Based on the obtained
structural guidelines, we designed (37-39, Scheme 2),
synthesized and experimentally tested selective molecular probes.
2.2. Quantum mechanical calculations
In order to have an accurate weight of the electrostatics, we
derived the partial charge of Zn2+ and of the amino acids
constituting the catalytic center (Table S2-S9) by DFT
calculations m0529 level by the 6-31+G(d) basis set and ChelpG
method30 for population analysis (Gaussian 03 Software
Package).31 By using the same theoretical level, the partial
charges of 37-39 were achieved (TableS10-S12). All the partial
charges derived from DFT theoretical level were used in the
subsequent docking calculations.
Please, insert Scheme 2
2. Methods
2.1. Homology modeling
The amino acid sequences of Human HDAC1 (Genbank
Accession Number Q13547,482aa), HDAC2 (Genbank
Accession Number AAH-31055,488aa), HDAC4 (Genbank
Accession Number BAA22957,1097aa), and HDAC6 (Genbank
Accession Number Q6NT75,1215aa), were extracted from the
NCBI protein sequence database. The BLAST (Basic Local
Alignment Search Tool)16 search was performed to find
homologous proteins in the PDB database, applying the
BLOSUM6217 (BLOcks of amino acid SUbstitution Matrix)
matrix. The search of homologous proteins was run by the
Chimera 1.5.3 package.18 The resulting alignments were
examined and modified manually. HDAC1 has additional
segment in his C-terminal domain that is about 50-110 amino
acids long. When this portion of the sequence was subjected to a
BLAST search, no alignment was possible, and no similar
sequences (other than itself) were found. Moreover, the function
of these residues was proposed to recruit other enzymes to large
protein complexes that may regulate their activities.19,20 Thus, it
may have less influence on the substrate/inhibitor binding. Due to
the lack of structural information on this portion, it was omitted
in the model building. The three-dimensional structure of
HDAC2 chain A (PDB code: 3MAX)21 and HDAC4 (PDB code:
2VQM)22 were respectively used as templates for human HDAC1
2.3. Ligands preparation
All ligands structures were built using the graphical interface
Maestro version 6.0, Schrödinger, LLC, New York, NY, 2003,
and their geometries optimized through MacroModel 8.5 and
using the MMFFs force field.32 For the 36, the tertiary amine on
the tricyclic ring system was protonated and the two enantiomers
were considered in our theoretical studies. Monte Carlo Multiple
Minimum (MCMM) method (10000 steps) of the MacroModel
module was used in order to allow a full exploration of the
conformational space. The so obtained geometries were
optimized using the Polak-Ribier conjugate gradient algorithm
(maximum derivative less than 0.001 kcal/mol). A GB/SA
(generalized Born/surface area) solvent treatment33 was used,
mimicking the presence of H2O in the geometry optimization and
in the conformational search steps.
2.4. Molecular docking calculations

Molecular docking studies were performed using AutoDock
3.0.5,34 which has been successfully used in the interpretation of
the inhibitory activity of several HDAC ligands.19,35,36
0.1% TFA in water (A) and 0.1% TFA in acetonitrile (B). The
absorbance was detected at 220-240 nm. All the solvents used for
the synthesis were HPLC grade; they were purchased from
Aldrich, Fluka, Carlo Erba. All reagents were purchased from
commercial suppliers and used as received. All the reactives were
purchased from Aldrich.
The homology models for HDAC1 and HDAC6 were used in
molecular docking calculations, along with the X-ray structures
of HDAC2 (3MAX),19 HDAC3 (4A69),37 HDAC4 (2VQM),22
HDAC7 (3C0Z)38 and HDAC8 (3F0R).39 In particular, the crystal
structure of HDAC8 complexed with CRAA-A (1VKG)40 was
also considered in the calculations, in order to explore the large
sub-pocket created by the shift of Phe152 upon ligand binding,
and located in the hydrophobic active site channel.
2.5.2. Synthesis of amides: general procedure
The carboxylic acid 40 or 41 (1 equiv.) and the appropriate
ammine 42 or 43 (2 equiv.) were dissolved in DMF. N,N-
triethylamine (TEA), N-hydroxybenzotriazole (HOBt) and N,N′-
diisopropil-carbodiimide (DIC) (2 equiv. of each) were added.
The mixture was leaved at room temperature for 48 hours under
stirring. When TLC showed the consumption of the carboxylic
acids 40 or 41, the reaction was stopped adding HCl 1N (10 mL).
The aqueous phase was extracted with ethyl acetate (3 x 10 mL)
and the organic phase was washed firstly with a saturate solution
of NaHCO3 and then with brine. Afterward the organics were
dried over Na2SO4, filtered and concentrated in vacuo. The crude
was purified by flash chromatography (10 % ethyl-acetate/n-
hexane to 50 % ethyl-acetate/n-hexane) to obtain the desired
amidic derivatives as yellow powder (Yields: 78-82 %).
HDACs are metalloproteins, so a non-bonded model for
metallic center according to the non-bonded Zn2+ parameters of
Stote41 (Zinc Radius = 1.10 Å, well depth = 0.25 kcal/mol) were
used in the docking calculations.
To achieve a representative conformational space during the
docking studies and taking into account the variable number of
active torsions, 10 calculations consisting of 256 runs were
performed, obtaining 2560 structures for each ligand (1-39). The
Lamarkian genetic algorithm was employed for dockings. An
initial population of 600 randomly placed individuals was used.
The maximum number of energy evaluations and of generations
was set up at 5 x 106 and 6 x 106, respectively. A mutation rate of
0.02 and a crossover rate of 0.8 were used, and the local search
frequency was set up at 0.26. Results differing by less than 2 Å in
positional root-mean-square deviation (rmsd) were clustered
together and ranked by free energy of binding. For all the docked
structures, all bonds were treated as active torsional bonds except
the amide bonds. The grid box was sized, by using 0.375 Å
between the grid points, and the coordinates of grid center were
varied for each protein (see supplementary data for the three-
dimensional coordinates of the homology models). In details, in
docking calculations with HDAC1 as receptor, a grid box size of
66 x 64 x 72 with a grid centre having the following x, y, and z
coordinates respectively: 49.75, 5.0 and 105.0. For HDAC2 it
was used a grid box size of 66 x 64 x 68 with the grid center
positioned on the following x, y, and z coordinates respectively:
66.265, 31.0 and -3.7. For HDAC3 a grid box size of 68 x 86 x
76 with grid center x, y, and z coordinates at 34.423, 60.188 and
26.892, respectively. For HDAC4 a grid box size of 48 x 64 x 90
presenting a grid center with the following x, y, and z coordinates
respectively: 16.376, -10.087 and 2.96. For HDAC6 a grid box
size of 54 x 60 x 62 with a grid center positioned on the
following x, y, and z coordinates respectively: 18.908, -9.0 and -
0.126. For HDAC8 (3F0R) a grid box size of 66 x 66 x 72 was
used, and it presented a grid center with the following x, y, and z
coordinates respectively: 107.7, 59.3 and 26.383. For HDAC8
(1VKG, pdb ID) the grid box size was 62 x 62 x 62 and the grid
center at the following x, y, and z coordinates respectively:
21.859, 69.0 and 72.5.
4-hydroxy-N-(naphthalen-2-ylmethyl)-3,5-dinitrobenzamide
1
(45): H-NMR δ (600 MHz; CH3OH) 8.51 (2H, s), 8.11 (d, 1H),
7.86 (d, 1H), 7.78 (d, 1H), 7.51-7.43 (m, 4H), 5.01 (2H, s); ES-
MS calcd. for C18H13N3O6 367.1; found m/z 368.0 [M+H]+.
(4-hydroxy-3,5-dinitrophenyl)(pyrrolidin-1-yl)methanone
(46): 1H-NMR δ (600 MHz; CH3OH) 8.51 (2H, s), 3.62 (4H, m),
2.01 (4H, m); ES-MS calcd. for C11H11N3O6 281.1; found m/z
282.2 [M+H]+.
1
N-benzyl-2-(4-hydroxy-3,5-dinitrophenyl)acetamide (47): H-
NMR δ (600 MHz; CH3OH) 8.22 (2H, s), 7.28 (5H, m), 3.63
(2H, s); ES-MS calcd. for C15H13N3O6 331.1; found m/z 332.0
[M+H]+.
2.5.3. Reduction of nitro-derivatives to amines: general
procedure
A solution of 45-47 and stannous chloride dihydrate (5 equiv.)
in ethanol was stirred at 80 °C for about 30 min. The solvent was
removed in vacuo. Ethyl-acetate (10 mL) was added to the
residue and washed with saturated sodium bicarbonate (2 x 10
mL). The organic layer was dried over Na2SO4 and concentrated
to obtain yellow solids. The obtained compounds 48-50 were
directly employed in the next step without further purification.
2.5.4. Reductive amination: general procedure
2.5. Chemistry
Under inert atmosphere (N2), 2 equiv. of aldehyde 51-53 were
dissolved in anhydrous CH3OH (1 mL/0.15 mmol of amine). The
mixture was kept under stirring at room temperature; anhydrous
amine 42-44 (1 equiv.), ZnCl2 (1 equiv.) and NaCNBH3 (2
equiv.) were added. After 1.5 h, when the amine disappeared, the
reaction was stopped and 10 mL of an aqueous solution of NaOH
0.1 M was added. After concentration of CH3OH in vacuo, the
aqueous phase was extracted with ethyl acetate (3 x 10 mL) and
the organics were dried over Na2SO4, filtered and concentrated in
vacuo. The obtained oil was purified on semipreparative RP-
HPLC (5% acetonitrile to 100% acetonitrile) to obtain the desired
products as yellow powder with a purity > 98 %.
2.5.1. General Experimental Procedures
All the NMR spectra (1H, HMBC, HSQC) were recorded on a
Bruker Avance DRX600 at T = 298 K. The compounds (37-39)
were dissolved in 0.5 mL of 99.8% d4-MeOH (Aldrich, 99.8+
Atom % D). Chemical shifts are expressed in parts per million
(ppm) on the delta (δ) scale. Electrospray mass spectrometry (ES-
MS) was performed on a LCQ DECA TermoQuest (San Josè,
California, USA) mass spectrometer. Analytical and
semipreparative reverse phase HPLC was performed on a Jupiter
C-18 column (250 x 4.60 mm, 5μ, 300 Å, flow rate = 1 mL/min;
250 x 10.00 mm, 10 μ, 300 Å, flow rate = 4 mL/min
respectively). The binary solvent system (A/B) was as follows:

3,5-Bis-[(furan-2-ylmethyl)-amino]-4-hydroxy-N-naphthalen-
2-ylmethyl-benzamide (37): RP-HPLC: from 10% B to 80% B
over 55 min: tR = 31.67 min.; Yield: 49%; 1H-NMR δ (600 MHz;
CH3OH): 8.11 (2H, d), 7.91 (2H, d), 7.85 (2H, d), 7.58-7.45 (8H,
m), 7.16 (1H, s), 5.02-4.97 (6H, m); 13C-NMR δ (150 MHz;
CH3OH): 172.7, 129.4, 128.9, 127.1-125.1, 124.1, 117.6, 43.0-
42.6. ES-MS calcd. for C28H25N3O4: 467.18; found 468.0
[M+H]+.
HDAC7: 962 nM of HDAC7 and 50 μM of substrate are in
the reaction buffer and 1% of DMSO final concentration of tested
compounds. Incubation for 2 hours at 30ºC.
HDAC8: 119 nM of HDAC8 and 50 μM of substrate are in
the reaction buffer and 1% of DMSO final concentration of tested
compounds. Incubation for 2 hours at 30ºC.
3. Results and Discussion
{4-Hydroxy-3,5-bis-[(thiazol-2-ylmethyl)-amino]-phenyl}-
3.1. Structural analysis.
pyrrolidin-1-yl-methanone (38): RP-HPLC: from 10% B to 50%
B over 50 min: tR = 18.15 min.; Yield: 46%; H-NMR δ (600
1
All the considered compounds in the present contribution
were analyzed by molecular docking calculations on HDAC1-4
and HDAC6-8 (Scheme 2). We refined some enzyme calculation
parameters related to the electrostatic and Van der Waals terms
of binding energy, following the successful strategy adopted for
the structural studies on azumamide E and two stereochemical
variants.35 This strategy allowed to reach a good qualitative
accordance between theoretical KD and biological assays results,
and it was also validated by its useful application to the study of
other HDAC ligands.36 In particular, for the electrostatic
contribution, the partial charges of the zinc ion and the amino
acids constituting the catalytic site of each isoforms were
calculated at DFT/M0529 theory level by using the 6-31+g(d) as
basis set and the ChelpG method30 for the population analysis and
were used in the subsequent docking calculations. Similarly, for
the van der Waals term the well depth and zinc radius proposed
by Stote and Karplus were applied.41
MHz; CH3OH): 7.79-7.66 (4H, br s), 7.50 (2H, br s), 4.66 (4H,
s), 3.50 (1H, m), 3.42 (1H, m), 3.21 (1H, m), 3.08 (1H, m), 1.92
(1H, m), 1.86 (1H, m), 1.81 (1H, m), 1.73 (1H, m); 13C-NMR δ
(150 MHz; CH3OH): 174.5, 142.7-141.0, 120.4-119.0, 50.9, 50.8
47.5, 47.3, 27.5, 27.13, 27.0, 26.9. ES-MS calcd. for
C19H21N5O2S2: 415.11; found 416.1 [M+H]+.
N-benzyl-2-(4-hydroxy-3,5-bis((thiophen-2-
ylmethyl)amino)phenyl)acetamide (39): RP-HPLC: from 10% B
1
to 60% B over 45 min: tR = 30.96 min.; Yield: 47%; H-NMR δ
(600 MHz; CH3OH): 7.27 (2H, m), 7.22-7.17 (5H, m), 6.88-6.81
(5H, m), 6.75 (1H, s), 4.39 (4H, s), 4.33 (2H, s), 3.36 (2H, s);
13C-NMR δ (150 MHz; CH3OH): 172.0, 129.2-126.0, 127.4-
127.1, 44.3, 44.1 . ES-MS calcd. for C25H25N3O2S2: 463.14;
found 464.1 [M+H]+.
2.6. Biological assay
3.2. Common structural features of all isoforms
The screening was performed by Reaction Biology Corp.
(www.reactionbiology.com/). Compounds 37-39 were tested in
10-dose mode with 3-fold serial dilution starting at 100 μM in
DMSO. TSA (reference compound) was tested in a 10-dose with
3-fold serial dilution starting at 10 μM, and starting at 20 μM
with Class2A substrate. In the screening, HDAC1, HDAC2,
HDAC4 and HDAC6-8 were used as isolated recombinant
human protein; whereas HDAC3 was used as HDAC3/NcoR2
complex. The substrate for HDAC1-3, and HDAC6 was the
fluorogenic acetylated peptide based on residues 379-382 of p53
(Arg-His-Lys-Lys(Ac)). For HDAC8, the substrate was the
diacetylated peptide based on residues 379-382 of p53 (Arg-His-
Lys(Ac)-Lys(Ac)). For HDAC4 and HDAC7, the substrate was
Acetyl-Lys(trifluoroacetyl)-AMC The Reaction Buffer was: 50
mM Tris-HCl, pH 8.0, 137 mM NaCl, 2.7 mM KCl, 1 mM
MgCl2, 1 mg/ml BSA. The assay conditions were as follow.
As well defined by previous studies,7,8 the general structure of
HDAC inhibitors can be dissected (1, Scheme 2) in: a cap group
involved in the molecular recognition process with surface amino
acids; a linker, usually hydrophobic; a zinc-chelating group. Each
structural moiety contributes to the binding event and biological
activity of the small molecules. In particular, a fundamental
structural element to inhibit these biological targets is the metal
binder. Many chelating agents, such as hydroxamic acids,
carboxylate, α-hydroxy-ketone were introduced, and theoretical
and experimental evidence showed their ability to coordinate the
prosthetic group of the investigated enzymes. Class I and II
proteins present a considerable sequence similarity in the
catalytic site. By comparing homology and experimental models,
we observed the presence of two parallel phenylalanine units
delineating the channel which accommodates the acetylated
lysine of the histone (Table S1, and Figure S1). As revealed by
our docking calculations on all considered HDACs, the switching
from a linear carbon chain (TSA (1), 3, 4, 5, 23-35) to an
aromatic linker (2, 6-22, 36) causes an increased affinity with the
targets, thanks to favourable π-π interactions with the above
mentioned two phenylalanine side chains.
HDAC1: 75 nM of HDAC1 and 50 μM of substrate were in
the reaction buffer and 1% of DMSO final concentration of tested
compounds. Incubation for 2 hours at 30ºC.
HDAC2: 5 nM of HDAC2 and 50 μM of substrate were in the
reaction buffer and 1% of DMSO final concentration of tested
compounds. Incubation for 2 hours at 30ºC.
HDAC3: 2.3 nM of HDAC3 and 50 μM of substrate were in
the reaction buffer and 1% of DMSO final concentration of tested
compounds. Incubation for 2 hours at 30ºC.
3.3. General features and differences of Class I and Class II
In our analysis we observed that HDAC1-3 and HDAC8
present a CO of a glycine and side chain of a tyrosine pointing
inside the 11 Å channel: Gly149 and Tyr303 for HDAC1,
Gly154 and Tyr308 for HDAC2, Gly143 and Tyr298 for
HDAC3, and Gly151 and Tyr306 for HDAC8. The class I
isoform selective compounds 7-16 present an amide bond
between the linker and the metal binder (Figure 1). The NH of
the amide functionality can establish a hydrogen bond with the
carbonyl of the glycine. Moreover, the hydrophobic channel of
class II proteins does not display acceptors and donors of
hydrogen bonds. In this perspective, the linker could be modified
HDAC4: 665 nM of HDAC4 and 50 μM of substrate were in
the reaction buffer and 1% of DMSO final concentration of tested
compounds. Incubation for 2 hours at 30ºC.
HDAC6: 12.6 nM of HDAC6 and 50 μM of substrate are in
the reaction buffer and 1% of DMSO final concentration of tested
compounds. Incubation for 2 hours at 30ºC.

inserting hydrogen bond acceptors and donors in order to give
interactions with these two residues constituting the hydrophobic
channel, sustaining selectivity for class I HDACs.
By our investigation we observed a tighter access to the catalytic
site of HDAC1 compared to HDAC2 (Figure 2), due to the
different spatial arrangement of residues leading to the zinc ion
and bordering the 11 Å channel. In particular, this diverse
orientation reflects the replacement of Met233, Pro361 and
Met364 in HDAC2 by Leu228, Asn356 and Leu359 in HDAC1.
Please, insert Figure 1
The analysis of crystal structures of bacterial homologues
(HDLP) of class I42 along with human HDAC2,21 HDAC337 and
HDAC839 reveals the presence of a 14 Å internal cavity at the
bottom of the 11 Å hydrophobic channel, close to the zinc active
site.43 The high sequence similarity of HDAC1 with HDLP,
HDAC2, HDAC3 and HDAC8 confirm for all class I HDACs the
presence of the 14 Å internal cavity is expected. Our homology
model of HDAC1 HDAC3 in fact showed the presence of the
internal channel observed in HDLP, HDAC2, HDAC3 and
HDAC8, also in agreement with previously reported homology
modelling studies.19 With one exception (HDAC8), the residues
forming the 14 Å internal cavities are identical across the
different proteins of class I. Recently, 2-amino benzamides were
proposed as metal binders, and compounds presenting this
functionality showed selectivity for class I enzymes.44 In
particular, in agreement with reported experimental data, the
analysis of docking results on the compounds presenting a
benzamide as metal binder (7,45 Scheme 2) revealed a preference
for class I enzymes, in particular for HDAC1 and HDAC2. The
coordination of the enzyme prosthetic group by the NH2 of
benzamide requires a side accommodation of phenyl ring, unlike
the common metal binders, such as hydroxamic acid, carboxylate
group, α-hydroxy-ketone. The consequence is the requirement of
a side room at the bottom of the 11 Å channel, offered by the 14
Å internal cavity.
In literature it is reported that natural cyclopeptides46 are
selective inhibitors of class I proteins, and the most important of
these ligands were considered in our studies. For example, the
azumamide E,35 Fk22810 and apicidine47 (3-5, Scheme 2) show
selectivity for class I, with significant affinity for HDAC8 and
superior affinity for HDAC1-3. From our theoretical analysis of
cyclopeptides and the other ligands, we found that the
macrocycle can be accommodated on a shallow pocket located on
the protein surface at the entrance of 11 Å hydrophobic cavity,
establishing Van der Waals interactions and hydrogen bonds
(Figure S2). These interactions contribute to the complex
stability, favouring the binding for the HDACs of class I over the
isoforms of class II. Such observation is suggested from the
docking analysis of the binding mode of azumamide E and
apicidine with HDACs of class II, showing that the macrocycle
does not interact with macromolecular counterparts on the
surface of HDAC4, HDAC6 and HDAC7, thus not contributing
to the affinity for the protein.
Please, insert Figure 2
As reported above, all proteins of class I present an internal
hydrophobic pocket at the bottom of the channel accommodating
the substrate. We observed that the amino acids constituting this
internal cavity are identical or conservatively substituted, but
differences in the surrounding amino acids of these internal
cavities can be pointed out. In particular, bulkier residues in
HDAC3 and HDAC8 (see below) prevent the accommodation of
the chelating agent with an appendage. Indeed, in the case of
HDAC1 and HDAC2 the docking results on 7 highlighted that
the metal binder is well harboured and the zinc is coordinated by
the CO and NH2 functionalities. Our docking studies show
(Figure S3) that the phenyl ring interacts with the surrounding
hydrophobic residues and along with the bidentate coordination
increases the affinity for HDAC1 and HDAC2. Our findings are
in agreement with published biological assays,45 which show the
selectivity of MS275 (7) for HDAC1 with this preference:
HDAC1 > HDAC3 and HDAC1 >> HDAC8, confirming the
presence of the benzamide metal binder for directing isoform
selectivity.7,8,44 The insertion of a substituent on the benzamide
leads to a further discrimination among HDACs of class I.
Indeed, HDAC1 and HDAC2 were able to harbour a substituent
on the benzamide (8-16, Scheme 2), such as thiophene or
phenyl.49 Our docked pose (Figure S4) suggests the same
considerations for compound 1750 where the phenethyl replaces
the benzamide interacting with the internal cavity, and
strengthening the coordination of the zinc to the carbonyl group.
3.5. HDAC2
As outlined in the previous section, many structural elements
of compounds having high affinity for HDAC1 are similar for the
isoform 2. By comparing the models of HDAC1 and HDAC2, a
structural differences, even though small, are detectable. As
described above, residues bordering the ≈ 11 Å channel, present a
different spatial arrangement of Tyr204 in HDAC1 with respect
to Tyr209 in HDAC2, in turn depending on the presence of
Leu228, Asn356 and Leu359 in HDAC1 and Met233, Pro361
and Met364 in HDAC2. On the basis of this different
arrangement, a deeper cavity is present for HDAC2 compared to
the shallower cavity of isoform 1, formed by amino acids His183,
Tyr209, Phe210 and Leu276. Together with this first
discrimination, and even though the amino acids constituting the
14 Å internal cavity are identical for HDAC1 and 2, differences
could be found in the neighbouring residues. In details, we
observed that the Val19 (HDAC1) is substituted by the Ile24 in
HDAC2, influencing the arrangement of surrounding residues. In
particular, it was observed a different disposition of Met35,
Phe114 and Leu144 giving rise to a larger room compared to
HDAC1 (Figure 3). This can justify the observed slightly
preference (about 10 fold)7,50 of benzamides for HDAC1 vs.
HDAC2, due to closer contacts with the internal cavity of
isoform 1. Moreover, we calculated volume (CASTp server)51 of
catalytic sites of isoforms 1 and 2 observing for HDAC2 414.8
Å3 vs. 281.7 Å3 of HDAC1. The isoforms 2 showed also a wider
pocket area of 409.2 Å2 vs. 327.7 Å2. The opening mouth area is
23.3 Å2 for HDAC2 and 12.9 Å2 for HDAC1. We designed and
tested new potential selective ligand for HDAC2 taking into
The pan inhibitors TSA (1) and NVP-LAQ824 (2),48 for
example, did not present this structural bulky cap group,
highlighting the role of the peptidic macrocycle in dictating the
selective class I binding. Compared to other analyzed class I
selective ligands, the cyclopeptides presented a wider cap group,
which established more extended contacts with proteins surface.
3.4. HDAC1
The differences between HDAC1 and HDAC2 are very small
due to an 85 % sequence identity and 93 % sequence similarity,
which confer very similar shapes for all the protein regions. The
comparison of the three-dimensional models of HDAC1 and
HDAC2 reveals some small differences, which could be
exploited to discriminate the recognition of these two isoforms.

account the combination of a metal binder able to interact with
the 14 Å internal cavity and of an adapted capping moiety (see
below).
compounds 8-17 endowed of larger appendages, the prosthetic
group of the enzyme is not coordinated. On the other hand, well
tailored chelating moieties can afford selectivity for isoform 8, as
showed by compounds 23 and 24.14a Indeed, the N-thiomethyl-
azetidin-2-one contained in 23 and 24 interacts with Trp141,
allowing a correct coordination of the zinc binding region of
HDAC8 (Figure S7).
Please, insert Figure 3
3.6. HDAC3
As shown in the previous two sections, the relevant structural
elements to discriminate isoforms of class I are appendages of
metal binder, able to interact with the internal cavity at the
bottom of the 11 Ǻ channel. By the analysis of the amino acids
surrounding the internal cavity close to the catalytic site, we
observed that in HDAC3 Tyr107 replaces Ser113 in HDAC1 and
Ser114 in HDAC2. In the isoform 3 the presence of the bulkier
side chain of Tyr107, forces the Leu133 to point towards the
center of the internal cavity. This shift of Leu133 causes a steric
clash preventing the accommodation of bulky metal binders. Our
docking results on 8-17 highlighted that the metal binder was not
3.8. HDAC4
Based on docked poses of all considered ligands (Scheme 2),
we tried to trace out the structural elements responsible for the
selectivity of HDAC4. In particular, aryl pyrrolyl hydroxamide
(APHA, 25, 26, Scheme 2) compounds54 are reported as selective
HDAC4 inhibitors. These compounds show a lower general
HDAC affinity compared to TSA, but present higher selectivity
for this enzyme. It was observed that substitution of chlorine with
fluorine at phenyl C3 position improved the selectivity for
HDAC4 from 78-fold to 176-fold over class I.54 Moreover, non-
halogenated or differently substituted APHA derivatives did not
show any selectivity towards this isoform, highlighting the
importance of the position of the halogen in the capping group.
Our theoretical model showed that the halogen is involved in a
hydrogen bond with the side chain of Tyr170 (Figure 5).
Moreover, there is an interaction between NH group of Phe168
with the electron π system of the halogenated phenyl ring, along
with a π-π interaction of C-halogen bond with CO of Phe168 that
can contribute to the specific recognition for HDAC4. Even
thought the Phe is conserved in the other isoforms, the APHA
cap group interacts only with backbone amide of Phe168 of
HDAC4 due to different shape of residue at rim of catalytic
channel. Our theoretical investigation also suggested that the
presence of an aromatic linker in 25 and 26 gives rise to π-π
interactions with side chains of Phe168 and Phe227 (Figure 5).
well accommodated in the internal cavity, as experimentally
7,8,45,50
confirmed.
Thus, the design of an adapted chelating agent
able to match the HDAC3 macromolecular counterparts and to
give effective contacts, is necessary for gaining selectivity toward
this isoform.
Our investigation allowed to appreciate differences on the
protein surface near the catalytic site. In particular, we observed
the presence of Phe199 in HDAC3 and Tyr204 and Tyr209 in
HDAC1 and HDAC2, respectively. In the isoforms 1 and 2 the
OH group of tyrosine establishes a hydrogen bond with the CO
of Leu271 in HDAC1 and Leu276 in HDAC2. This hydrogen
bond is absent in HDAC3, giving rise to a shallower hydrophobic
cavity delimited by the amino acids Ile171, His172, Phe199,
Phe200, Gly267 and Cys268 (Figure S5). The absence of the OH
implies no possible hydrogen bonds formation with the ligands.
Moreover, HDAC3 differs from all other isoforms for the
presence of a solvent-exposed tyrosine (Tyr198) on the surface of
the enzyme (Figure 4).
Please, insert Figure 5
Please, insert Figure 4
Not surprisingly, the hydroxamic acid of 25 and 26
coordinated the zinc ion, but also established hydrogen bonds
with the Nε2 of His158 and His159 by the OH and NH groups,
respectively (Figure 5).
This residue is very close to the active site tunnel, changing
the shape of the protein surface in that point. In particular, its
carbonyl group of the backbone points could be involved with
hydrogen bonds with the ligands.
As reported above,54 a moderate biological activity has been
shown for these HDAC4 selective inhibitors, thus structural
modifications are required to increase the affinity of new
candidate molecules. From our analysis, hydrophobic cavity
delimited by residues His198, Phe226, Phe227, Leu299 may host
a larger group, able to establish a hydrogen bond with the NH of
His198, in replacement of the methyl group in the 25 and 26.
From the docked poses of 25 and 26 we observed the cap group
near a small pocket formed by Pro165, Met166, Gly167, Tyr170,
and Cys169. Thus, along with a halogen, a hydrophobic group
can be inserted to establish Van der Waals contacts with the
described pocket on protein surface. Our comparison of
electrostatic potential maps of all considered isoforms on the
surface around the channel leading to the zinc ion revealed that
HDAC4 displayed a positive charged area, whereas the other
proteins presented negative or neutral regions (Figure S8). Thus,
the cap moiety can be elongated to favour electrostatic
interactions with side chains of positively charged amino acids.
3.7. HDAC8
The resolved X-ray structures of HDAC8 complexed with
TSA (1)39 and 1852 have suggested structural elements to
selectivity bind this isoform. When bound to 18, HDAC8 shows
a shift from the normal position of Phe152, which is located
along the hydrophobic 11 Ǻ channel. This shift creates a sub
pocket that can contain hydrophobic groups protruding from the
linker moiety, extending Van der Waals contacts with the
channel of HDAC8. These further strong interactions contribute
to the complex stability and favour a selective binding to
HDAC8. Compounds 1914b, 2015b and 2253 were designed on the
structural considerations obtained from 18 bound to HDAC8, and
molecular docking calculations showed, as expected, high
selectivity for HDAC8 thank to the interactions with the induced
hydrophobic cavity. By comparing the amino acids constituting
the internal cavity found in all isoforms of class I, we observed
that HDAC8 presents Trp141, in place of a leucine in HDAC1-3.
The presence of the bulky side chain of Trp141 hinders the
appropriate accommodation of chelating portions endowed with
appendages, compared to HDAC1-3. Indeed, our docking results
on 7 with HDAC8 showed that the Trp141 limited the
accommodation of the benzamide, causing the NH2 to coordinate
the zinc ion in a monodentate manner (Figure S6). Moreover, for
3.9. HDAC6
As already reported in previous studies on selective HDAC6
ligands,14d,14e,15a we confirmed through our analysis that the linker
length is a crucial structural element for achieving selectivity.

Upon zinc coordination, the linker length is responsible of
correctly directing the extended interactions of the capping
moiety with the macromolecular counterparts. HDAC6 presents a
wider entrance of the binding pocket conducting to the prosthetic
group, formed by several non-polar residues. The selective
inhibitors (27-35) of HDAC6 follow the protein shape by their
structural moieties and can assume an extended conformation
interacting with the non-polar residues on protein surface (Figure
S9). In the case of other considered HDAC isoforms, due to the
smaller entrance of the hydrophobic channel, the cap groups are
not well accommodated in the surface cavities giving fewer
contacts with amino acids, and suggesting a consequent lower
contribution to the complex stability. Moreover, in our docked
poses the linker chain of 27-35 folds to allow contacts between
cap group and amino acids on the surface of HDAC1-4, HDAC7
and HDAC8, but this entropic loss is not compensated by
extended interactions with the macromolecular counterparts.
These theoretical findings agree with experimental observations
HDAC7. The docked poses of 1-36 did not show interactions
with this macromolecular counterpart by their cap groups. Thus,
a capping moiety, able to establish contacts with this unique
pocket on the surface, can be another structural element
addressing selectivity for this enzyme.
4. Quick proof of concept: Design, synthesis and biological
evaluation of selective binders
4.1. Design of 37-39
On the basis of our analysis aimed to find the structural
elements responsible for a specific recognition of HDAC
isoforms, we designed three molecular probes (37-39, Scheme 2)
for the selective inhibition of HDACs. In particular, our design
relied on the use of the weak carbonyl group as chelating agent in
order to emphasize the contribution to selective binding by the
other structural moieties. We designed a metal binder with three
different appendages in order to probe the influence on the class I
and isoform selectivity. As cap group, we introduced three five-
membered heterocycles, able to form hydrogen bonds with the
protein. Our calculations suggest that all the three heterocycles
established hydrogen bonds with the enzymes with comparable
contribution to the binding affinities. For this reason, , we did not
consider all the combination of cap groups and chelating portion,
and we focused on the exploration of the chemical diversity of
the synthesized compounds.
55
that all selective inhibitors for HDAC6, tubacin (27),
mercaptoacetamides (28-31),15b and their analogues (32-35)56
presented longer spacers, differently from class I selective
inhibitors whose optimal linker length is six carbons.7,8,35,36 As
recently reported (36, Scheme 2),57 the right combination of
linker length with a large and rigid cap group can also dictate the
selectivity for the isoform 6. The 36 presents a shorter linker
length compared to the other selective HDAC6 compounds (27-
35), but the tolyl linker combined with the tricycle confers a bent
conformation to 36, favouring tight interactions with the rim of
hydrophobic channel of this isoform. On the contrary, the docked
poses of 36 in the binding cavity of HDAC1-4, HDAC7 and
HDAC8 shows steric clashes with amino acids on the proteins
surface, leading to unfavourable ligand-enzyme binding.
The three compounds (37-39, Scheme 2) were docked on all
considered HDACs. As revealed by the theoretical investigation,
the metal binder showed a Class I recognition preference. In fact,
the structure of the metal binder presents an appendage able to
establish contacts with the internal cavity near the catalytic site.
Moreover, changing the structure of the decorative element of the
metal binder, we observed isoform selectivity as predicted by the
investigation of 1-36. Indeed, compound 37 selectively reached
the catalytic site of HDAC2. In all other isoforms, 37 was not
able to coordinate the zinc ion, suggesting an exclusive binding
to HDAC2 as confirmed by the biological assays (see next
section). In particular, the carbonyl moiety coordinates the zinc
ion, whereas the NH of the amide group, as suggested by our
analysis, establishes hydrogen bonds with the CO of Gly154
(Figure 6). The naphthalene was accommodated in the 14 Ǻ
internal cavity, giving hydrophobic interactions with the
surrounding protein amino acids: Met35, Phe114, Gly143,
Leu144, Gly305, Tyr308 (Figure 6). The aromatic linker
interacted by π-π contacts with the hydrophobic channel
conducting to the zinc ion. One of the two furans was
accommodated in a shallow pocket delimited by residues His183,
Tyr209, Phe210 and Leu276, and it also established π-π
interactions with Tyr209 (Figure 6) and was hydrogen bonded
with His183. The other furan ring established anion-π interaction
with Asp104. The arrangement of these two aromatic groups of
the capping moiety, induced by the rim shape of the channel,
favoured the coordination of the zinc. It is noteworthy that we
designed this cap group in order to discriminate between the
isoforms of class I. Indeed, besides the HDAC1 and HDAC2
selectivity driven by the metal binder, we observed that this cap
group interacts differently with HDAC1 and HDAC2, leading to
a preference towards isoform 2. Moreover, the resolved X-ray
structure of HDAC2 (3MAX, pdb ID) presents the protein bound
to a N-(2-aminophenyl)benzamide. The comparison of 37 with
this N-(2-aminophenyl)benzamide in the catalytic site of HDAC2
revealed a similarity in the binding mode (Figure S11). Both
binders present a superimposable appendage at chelating portion
that fills the internal cavity. The two small molecules have an
3.10. HDAC7
Up to date, there are not selective inhibitors of HDAC7. Thus,
the docking results of compounds 1-36, the structural features of
the catalytic domain and of the surface pockets of HDAC7 may
suggest interesting elements for designing selective inhibitors of
this isoform. In detail, the unique sequence of HDAC7 gives rise
to a novel zinc binding motif. This protein domain is formed by a
β-hairpin positioned by two antiparallel β-strands (β3 and β4) and
the loop between helices α1 and α2, which outlines a distinct and
only groove contiguous to the opening of the active site
channel.38 This enlarged active site of HDAC7 could be able to
harbour a well tailored metal binder, conferring selectivity and
improving the affinity of new inhibitors for this enzyme.
Selective inhibitors 8-17 presented a metal binder with an
appendage able to interact with the 14 Ǻ internal cavity of class I
proteins. Our docking results on HDAC7 showed that these
selective inhibitors did not coordinate the prosthetic group of the
enzyme and did not fill the enlarged active site due to the steric
hindrance. Our predicted bioactive conformation of 7 is bound to
the zinc ion in a monodentate fashion. On the contrary, the
remainder docked inhibitors, presenting a classical chelating
agent without decorative appendages, coordinate the zinc ion.
Thus, in the hypothesis to design selective binders of HDAC7,
new protuberances decorating the metal binder should be
projected to selectively match the active site of HDAC7. It could
be suggested to insert two flexible appendages flanking the metal
binder of the putative ligand. We observed a deep hydrophobic
pocket near the Phe679, which is a constituting residue of
hydrophobic channel harbouring the acetylated lysine. This
cavity is delimited by the amino acids His531, His541, Pro542,
Glu543, Ile628 and Phe679 (Figure S10), and it is a peculiarity of

amide group, which occupies equivalent spaces and is involved
in the zinc ion coordination. The compound 37 shows a larger
cap group than the N-(2-aminophenyl)benzamides. The latter
coordinates the zinc ion in bidentate fashion, whereas 37 binds
the zinc by only the CO of amide group. In the theoretical model
obtained by docking calculations, we observed that the
interactions of the cap group with the surface counterpart of
HDAC1 prevented the right approaching of metal binder to the
zinc, predicting absence or at least scarce biological activity.
Compound 38 and 39 presented a different appendage of the
metal binder, which caused a different affinity and selectivity for
the investigated isoforms. In compound 38, the reduced
dimensions of the metal binder caused fewer interactions with the
internal cavity, justifying a lower affinity for the HDAC2.
Moreover, the modification of metal binder structure gave rise to
a possible binding to the HDAC8 (Figure S12), as suggested by
our calculations. This was due to the smaller internal cavity of
HDAC8 (see also 3.7) than the remaining class I isoforms, thus
the reduced size of the metal binder appendage could favour the
maximal interactions with protein residues. The same
considerations can be made for the 39 (Figures S13, S14). The 39
presented a metal binder appendage with suitable size for the
limited internal cavity of HDAC8. Moreover, we observed for 38
and 39 that their binding to isoforms 8 was also favoured by the
shift of Phe152 (see also 3.7), along the hydrophobic 11 Ǻ
channel. Respect to 37 and 38, compound 39 presents a
methylene between the chelating agent and the linker. This
structural modification does not favour the interactions given by
the linker and the cap group with HDAC2. Indeed 39 showed a
lower binding affinity with HDAC2, with respect to the other
compounds.
4.3. Biological Activities
The selectivity of 37-39 was evaluated on HDAC1-4 and
HDAC6-8 by experimental assays (Table 2). Compounds 37-39
displayed significant selectivity in the in vitro inhibition tests
against the HDAC1-4 and HDAC6-8 (Table 2). As expected, all
compounds were not able to inhibit class II HDACs, and showed
a selective isoform binding among proteins of class I. As
predicted by the docking studies, 37 showed selectivity on
HDAC2, thus confirming that the structure of cap group and
metal binder is an important determinant for the biological
activity towards isoform 2. In particular, the capping moiety
allowed to discriminate between HDAC1 and HDAC2.
Compared to 37, compound 38 presented no activity against
HDAC2 in agreement to the docking results, due to the less
extended interactions given by the pyrrolidine with the internal
hydrophobic cavity. As for 37, 38 did not present activity of
isoforms 1 and 3. There was a good activity on HDAC8,
probably due to the modified and well tailored metal binder.
Concerning compound 39, the experimental data confirmed the
theoretically expected (see Proof of concept) lower/absent
selectivity of binding to HDAC2 as theoretically foreseen.
Moreover, thank to the cap group an inhibitory activity on
HDAC1 was not detected. Compound 39 showed a comparable
inhibition of HDAC3 and HDAC8. In summary, 37 and 38 are
respectively selective ligands of HDAC2 and of HDAC8, even
though at modest potency. The evaluated inhibitory activity on
these enzymes is in line with the theoretical findings, confirming
the predicted structural observations.
Please insert Table 2
Please, insert Figure 6
5. Conclusions
In the field of epigenetic cancer therapy, inhibitors of Histone
Deacetylases (HDACs) have been gaining a great success. Their
potential application in cancers therapy is limited by different
side-effects, due to the unselective binding towards either all or at
least several HDAC isoforms by the inhibitors. Thus, the further
development of HDAC inhibitors is to target selectively
individual HDAC isoforms. In the present contribution, the
structural elements responsible of the selective binding towards a
specific isoform of HDAC have been investigated, and our
findings are the initial efforts of a scientific project focusing on
the efficacy evaluation of selective binders in cancer therapy.
4.2. Synthesis of 37-39
Virtual screening driven synthesis of the three derivatives 37-
39 relies on the retro-synthetic analysis outlined in Scheme 3(a).
We started from the commercially available scaffold A, which
was subjected as first to an amidation reaction in order to
construct the amido functionality. Then, to insert on the aromatic
ring the two amine functions, reduction of NO2 groups was
performed. Finally, to decorate the aromatic scaffold with
symmetric appendages, a reductive amination reaction, with the
appropriate aldehyde 51-53, was used. The experimental details
of the synthetic procedure are outlined in Scheme 3(b). In
particular, taking advantages of the conventional peptide
synthesis conditions, we used the carboxylic acids 40 or 41 and
the amines 42-44, in presence of HOBt and DIC as carboxylic
group activators, TEA as base and DMF as solvent. Afterward,
we first reduced the nitro-group, on the aromatic ring, employing
SnCl2 in ethanol as solvent and heating the mixture of reaction at
80° C for about 30 minutes. Finally, to generate the desired
derivatives 37-39, we subjected the diaminic intermediates 48-50
to a reductive amination with the appropriate aldehydes (51-53)
using ZnCl2 as carbonyl group activator in presence of
NaCNBH4. This last reaction required a fine adjustment of the
experimental conditions to avoid the by-products formations,
mainly constituted by poly-alkylated amines. Systematically
variations of the aldehydes amount, the dilutions and the reaction
times, allowed to find the optimal conditions for obtaining 37-39
in satisfactory yields (see also Supplementary data).
It should be highlighted that class I and II proteins present a
considerable sequence similarity in the catalytic site. The analysis
of homology and experimental models revealed the presence of
two phenylalanines to shape the channel which accommodates
the substrate. Thus, the insertion of an aromatic linker,
establishing π-π interactions with these two phenylalanine side
chains, could be the starting point to design selective inhibitors of
HDACs. As common structural feature of class I enzymes, we
found the presence of a CO of a glycine and side chain of a
tyrosine pointing inside the 11 Å channel, suggesting the
insertion of hydrogen bond donors and acceptors as structural
elements to increase the affinity for the isoforms of class I. Class
I proteins present an internal pocket at the bottom of the tube-like
channel leading to the zinc ion. This internal cavity is formed by
the same residues for all class I isoforms, except for HDAC8
presenting a Leu replaced by Tyr. Differences are also observed
in the amino acids surrounding the internal cavity, causing
diversity in the shape and dimension of this inside pocket across
class I isoforms. This internal cavity can be useful to dictate a
class I and isoform selectivity, designing a metal binder able to
maximize the interactions with this pocket. It is noteworthy that
Please, insert Scheme 3

the shape at the border of the channel leading to the zinc ion, and
the relative orientation of these surface pockets respect to the
metal and the internal cavity differs in each class I isoforms.
Thus, the rational design of capping group able to fit the surface
cavities of each isoform, without altering the contacts given by
the other structural portions of the ligands can dictate the isoform
selectivity. HDAC1 and HDAC2 presents high similarity and
identity percentage in the amino acid sequence, but we observed
different dimensions in the catalytic pockets of the isoforms 1
and 2. These discrepancies could be ascribable to the differences
in the primary sequences of the residues neighbouring the
catalytic cavity.
theoretical structural observations, and represent a starting point
of a research project aimed to obtain more potent a selective
compounds. Even though 37-39 are weak binders, we were able
to propose new molecular scaffolds, obtainable by few simple
synthetic steps, and they could be new useful chemical platforms
for the design of selective inhibitors of HDACs. In particular, in
the first synthetic step different aromatic platform can be used
allowing to differentiate the obtainable compounds and explore
the chemical diversity. On this basis, new synthetic, theoretical
and experimental studies are in progress to develop more potent
inhibitors against
a selected HDAC isoform. Moreover,
considering that HDACs work in cells as multi-protein
complexes, the new studies will also focus on the efficiency to
inhibit these protein clusters.
The rationalization of binding mode of APHA derivatives,
along with the analysis of all other considered compounds,
suggested the structural elements for the specific binding to
HDAC4. APHA compounds present a halogenated phenyl ring as
cap group, establishing a selective hydrogen bond with the
Tyr170 of isoform 4. Moreover, the hydroxamic acid coordinated
the zinc ion, but also established hydrogen bonds with the side
chains of His158 and His159. Behind the different globular
shape, HDAC4, respect to the other proteins of class I and II,
presented a different charge distribution in the surface regions
bordering the channel which conducts to the zinc ion. Thus, on
the capping moiety could be inserted functionality able to
electrostatically interact with positively charged regions of
HDAC4. Moreover, hydrophobic pockets on the protein surface
were found and can be exploited to increase the affinity for
HDAC4.
Acknowledgments
Financial support by the University of Salerno and by
Ministero dell'Istruzione, dell'Università e della Ricerca (MIUR),
PRIN 2009 ―Design, conformational and configurational analysis
of novel molecular platforms‖.
References and notes
1.
2.
Strahl, B. D.; Allis C. D. Nature 2000, 403, 41–45.
de Ruijter, A. J.; van Gennip, A. H.; Caron, H. N.; Kemp, S.;
van Kuilenburg, A. B. Biochem. J. 2003, 370, 737–749.
Michishita, E.; Park, J. Y.; Burneskis, J. M.; Barrett, J. C.;
Horikawa, I. Mol. Biol. Cell. 2005, 16, 4623–4635.
Bertrand, P. Eur. J. Med. Chem. 2010, 45, 2095-2116.
Witt, O.; Deubzer, H. E.; Milde, T.; Oehme, I. Cancer Lett.
2009, 277, 8–21.
3.
In line with reported studies on selective inhibitors of
HDAC6, we observed that the linker length plays a fundamental
role in the binding preference towards this isoform. Selective
inhibitors of HDAC6 presented a longer linker length than class I
selective binders, and our analysis revealed that this structural
feature of the ligands favour the extended interactions of cap
group with the residues on the surface of the protein. Moreover,
combining the right linker length with a large and rigid cap group
can also lead towards isoform 6 selectivity.
4.
5.
6.
7.
Bolden, J.E.; Peart, M. J.; Johnstone, R. W. Nat. Rev. Drug
Discovery 2006, 5, 769-784.
Bieliauskas, A. V.; Pflum, M. K. H. Chem. Soc. Rev. 2008,
37, 1402–1413.
8.
9.
10.
Wang, D. Curr. Top. Med. Chem. 2009, 9, 241-256.
Marks, P. A.; Breslow, R. Nat. Biotechnol. 2007, 25, 84–90.
Furumai, R.; Matsuyama, A.; Kobashi, N.; Lee, K-H.;
Nishiyama, M.; Nakajima, H.; Tanaka, A.; Komatsu, Y.;
Nishino, N.; Yoshida. M.; Horinouchi, S. Cancer Res. 2002,
62, 4916–4921. (b) VanderMolen, K. M.; McCulloch, W.;
Pearce, C. J.; Oberlies, N.H. J. Antibiot. 2011, 64, 525-531.
Bruserud, O.; Stapnes, C.; Ersvaer, E.; Gjertsen, B. T.;
Ryningen, A. Curr. Pharm. Biotechnol. 2007, 8, 388–400.
(a) Bolden, J. E.; Peart, M.J.; Johnstone, R. W. Nat. Rev.
Drug Discov. 2006, 5, 769–784. (b) Thomas, E. A. Mol.
Neurobiol. 2009, 40, 33–45.
About HDAC7, the selectivity could be obtained by designing
a tailored metal binder able to interact with the unique enlarged
binding site of this isoform. Moreover, we observed a peculiar
surface hydrophobic pocket at the rim of the channel harbouring
the substrate. We suggest that an adapted capping moiety, able to
interact with this found cavity, can further lead the binding
preference for this protein.
11.
12.
13.
Bantscheff, M.; Hopf, C.; Savitski, M. M.; Dittmann, A.;
Grandi, P.; Michon, A-M.; Schlegl, J.; Abraham, Y.; Becher,
I.; Bergamini, G.; Boesche, M.; Delling, M.; Dümpelfeld, B.;
Eberhard, D.; Huthmacher, C.; Mathieson, T.; Poeckel, D.;
Reader, V.; Strunk, K.; Sweetman, G.; Kruse, U.; Neubauer,
G.; Ramsden, N. G.; Drewes, G. Nat. Biotechnol. 2011,
29:255–265.
(a) Galletti, P.; Quintavalla, A.; Ventrici, C.; Giannini, G.;
Cabri, W.; Penco, S.; Gallo, G.; Vincenti, S.; Giacomini, D.
ChemMedChem 2009, 4, 1991–2001. (b) KrennHrubec, K.;
Marshall, B. L.; Hedglin, M.; Verdin, E.; Ulrich, S. M.
Bioorg. Med. Chem. Lett. 2007, 17, 2874–2878. (c) Moradei,
O. M.; Mallais, T. C.; Frechette, S.; Paquin, I.; Tessier, P. E.;
Leit, S. M.; Fournel, M.; Bonfils, C.; Trachy-Bourget, M-C.;
Liu, J.; Yan, T. P.; Lu, A-H.; Rahil, J.; Wang, J.; Lefebvre,
S.; Li, Z.; Vaisburg, A. F.; Besterman, J. M. J. Med. Chem.
2007, 50, 5543-5546. (d) Estiu, G.; Greenberg, E.; Harrison,
C. B.; Kwiatkowski, N. P.; Mazitschek, R.; Bradner, J. E.;
Wiest, O. J. Med. Chem. 2008, 51, 2898–2906. (e) Schäfer,
S.; Saunders, L.; Schlimme, S.; Valkov, V.; Wagner, J. M.;
Kratz, F.; Sippl, W.; Verdin, E.; Jung, M. ChemMedChem
2009, 4, 283–290.
In order to rapidly verify our structural findings, we designed
small molecular probes. In particular, we tried to discriminate
between HDAC1 and 2, which are very similar isoforms. These
designed small molecules were synthesized and tested by in vitro
inhibitory activity on all considered HDAC isoforms. In
particular, we used a weak chelating agent, the carbonyl group, in
order to emphasize the contribution to selective binding by the
other structural moieties. As expected the small molecules
showed a modest potency, but the aim of the presented work is to
open a new avenue for a targeted rational design of selective
inhibitors towards the different HDAC isoforms. In particular, we
found that compound 37 exclusively binds HDAC2, whereas 38
inhibits HDAC8, suggesting that the structural modification of
the appendage of the metal binder can lead to isoform selectivity
in agreement with theoretical analysis. Moreover, our analysis
also suggests the role of cap group to lead to isoforms selectivity.
Compound 39 shows similar inhibitory activity on HDAC3 and
8, confirming the theoretical prediction that structural
modification of the linker alters the interaction of the cap group
with HDAC2. The experimental data confirm as predicted by
14.
15.
(a) Itoh, Y.; Suzuki, T.; Kouketsu, A.; Suzuki, N.; Maeda, S.;
Yoshida, M.; Nakagawa, H.; Miyata, N. J. Med. Chem. 2007,

50, 5425-5438. (b) Kozikowski, A. P.; Chen, Y.; Gaysin, A.;
Chen, B.; D’Annibale, M. A.; Suto, C. M.; Langley, B. C.
2007, J. Med. Chem. 50, 3054-3061. (c) Mai, A.; Massa, S.;
Pezzi, R.; Simeoni, S.; Rotili, D.; Nebbioso, A.;
Scognamiglio, A.; Altucci, L.; Loidl, P.; Brosch, G. J. Med.
Chem. 2005, 48, 3344-3353. (d) Siliphaivanh, P.; Harrington,
P.; Witter, D. J.; Otte, K.; Tempest, P.; Kattar, S.; Kral, A.
M.; Fleming, J. C.; Deshmukh, S. V.; Harsch, A.; Secrist, P.
J.; Miller, T. A. Bioorg. Med. Chem. Lett. 2007, 17, 4619–
4624.
Altschul, S. F., Gish, W.; Miller, W.; Myers, E. W.; Lipman,
D. J. J. Mol. Biol. 1990, 215, 403-410.
Henikoff, S.; Henikoff, J. G. Proc. Natl. Acad. Sci. USA
1992, 89, 10915–10919.
Pettersen, E. F.; Goddard, T. D.; Huang, C. C.; Couch, G. S.;
Greenblatt, D. M.; Meng, E. C.; Ferrin, T. E. J Comput.
Chem. 2004, 25, 1605-1612.
Wang, D. F.; Helquist, P.; Wiech, N. L.; Wiest, O. J. Med.
Chem. 2005, 48, 6936-6947.
Canonico, P. L.; Bifulco, G.; Tron, G. C.; Sorba, G.; Pirali,
T. J. Med. Chem. 2009, 52, 2776–2785. (c) Pirali, T.; Faccio,
V.; Mossetti, R.; Grolla, A. A.; Di Micco, S.; Bifulco, G.;
Genazzani, A. A.; Tron, G. C. Mol. Diversity 2010, 14, 109-
121. (d) Terracciano, S.; Di Micco, S.; Bifulco, G.; Gallinari,
P.; Riccio, R.; Bruno, I. Bioorg. Med. Chem. 2010, 18, 3252–
3260. (e) Terracciano, S.; Chini, M. G.; Bifulco, G.;
D'Amico, E.; Marzocco, S.; Riccio, R.; Bruno, I.
Tetrahedron 2010, 66, 2520–2528.
37.
38.
Watson, P. J.; Fairall, L.; Santos, G. M.; Schwabe, J. W. R.
Nature 2012, 481, 335-340.
16.
17.
18.
Schuetz, A.; Min, J.; Allali-Hassani, A.; Schapira, M.;
Shuen, M.; Loppnau, P.; Mazitschek, R.; Kwiatkowski, N.
P.; Lewis, T. A.; Maglathin, R. L., McLean, T. H.;
Bochkarev, A.; Plotnikov, A. N.; Vedadi, M.; Arrowsmith,
C. H. J. Biol. Chem. 2008, 283, 11355-11363.
Dowling, D. P.; Gantt, S. L.; Gattis, S. G.; Fierke, C. A.;
Christianson, D. W. Biochemistry 2008, 47, 13554-13563.
Somoza, J. R.; Skene, R. J.; Katz, B. A.; Mol, C.; Ho, J. D.;
Jennings, A. J.; Luong, C.; Arvai, A.; Buggy, J. J.; Chi, E.;
Tang, J.; Sang, B-C.; Verner, E.; Wynands, R.; Leahy, E. M.;
Dougan, D. R.; Snell, G.; Navre, M.; Knuth, M. W.;
Swanson, R. V.; McRee, D. E.; Tari, L. W. Structure 2004,
12, 1325-1334.
39.
40.
19.
20.
21.
Ayer, D. E. Trends Cell Biol. 1999, 9, 193-198.
Bressi, J. C.; Jennings, A. J.; Skene, R.; Wu, Y.; Melkus, R.;
De Jong, R.; O’Connell, S.; Grimshaw, C. E.; Navre, M.;
Gangloff, A. R. Bioorg. Med. Chem. Lett. 2010, 20, 3142–
3145.
22.
Bottomley, M. J.; Lo Surdo, P.; Di Giovine, P.; Cirillo, A.;
Scarpelli, R.; Ferrigno, F.; Jones, P.; Neddermann, P.; De
Francesco, R.; Steinkuhler, C.; Gallinari, P.; Carfi, A. J. Biol.
Chem. 2008, 283, 26694-26704.
41.
42.
Stote, R. H.; Karplus, M. Proteins 1995, 23, 12–31.
Finnin, M. S.; Donigian, J. R.; Cohen, A.; Richon, V. M.;
Rifkind, R. A.; Marks, P. A.; Pavletich, N. P. Nature 1999,
401, 188–193.
23.
24.
Zou, H.; Wu, Y.; Navre, M.; Sang, B. C. Biochem. Biophys.
Res. Commun. 2006, 341, 45-50.
Butler, K. V.; Kalin, J.; Brochier, C.; Vistoli, G.; Langley,
B.; Kozikowski, A. P. J. Am. Chem. Soc. 2010, 132, 10842–
10846.
Fiser, A.; Sali, A. Meth. Enzymol. 2003, 374, 461–91.
Kaminski, G. A.; Friesner, R. A.; Tirado-Rives, J.;
Jorgensen, W. L. J. Phys. Chem. B 2001, 105, 6474-6487.
MacroModel, version 8.5, Schrödinger LLC, New York, NY,
2003
Laskowski, R. A.; MacArthur, M. W.; Moss, D. S.;
Thornton, J. M. J. App. Cryst. 1993, 26, 283-291.
Zhao, Y.; Schultz, N. E.; Truhlar, D. G. J. Chem. Phys. 2005,
123, 161103-1-4.
43.
44.
Wang, D. F.; Wiest, O.; Helquist, P.; Lan-Hargest, H. Y.;
Wiech, N. L. J. Med. Chem. 2004, 47, 3409-3417.
(a) Methot, J. L.; Hamblett, C. L.; Mampreian, D. M.; Jung,
J.; Harsch, A.; Szewczak, A. A.; Dahlberg, W. K.;
Middleton, R. E.; Hughes, B.; Fleming, J. C.; Wang, H.;
Kral, A. M.; Ozerova, N.; Cruz, J. C.; Haines, B.; Chenard,
M.; Kenific, C. M.; Secrist, J. P.; Miller, T. A. Bioorg. Med.
Chem. Lett. 2008, 18, 6104–6109.(b) Hamblett, C. L.;
Methot, J. L.; Mampreian, D. M.; Sloman, D. L.; Stanton, M.
G.; Kral, A. M.; Fleming, J. C.; Cruz, J. C.; Chenard, M.;
Ozerova, N.; Hitz, A. M.; Wang, H.; Deshmukh, S. V.;
Nazef, N.; Harsch, A.; Hughes, B.; Dahlberg, W. K.;
Szewczak, A. A.; Middleton, R. E.; Mosley, R. T.; Secrist, J.
P.; Miller, T. A. Bioorg. Med. Chem. Lett. 2007, 17, 5300–
5309.
Saito, A.; Yamashita, T.; Mariko, Y.; Nosaka, Y.; Tsuchiya,
K.; Ando, T.; Suzuki, T.; Tsuruo, T.; Nakanishi, O. Proc.
Natl. Acad. Sci. USA 1999, 96, 4592-4597.
Byrd, J. C.; Marcucci, G.; Parthun, M. R.; Xiao, J. J.;
Klisovic, R. B.; Moran, M.; Lin, T. S.; Liu, S.; Sklenar, A.
R.; Davis, M. E.; Lucas, D. M.; Fischer, B.; Shank, R.;
Tejaswi, S. L.; Binkley, P.; Wright, J.; Chan, K. K.; Grever,
M. R. Blood 2005, 105, 959–967.
Singh, S. B.; Zink, D. L.; Liesch, J. M.; Mosley, R. T.;
Dombrowski, A. W.; Bills, G. F.; Darkin-Rattray, S. J.;
Schmatz, D. M.; Goetz, M. A. J. Org. Chem. 2002, 67, 815-
825.
Atadja, P.; Gao, L.; Kwon, P.; Trogani, N.; Walker, H.; Hsu,
M.; Yeleswarapu, L.; Chandramouli, N.; Perez, L.; Versace,
R.; Wu, A.; Sambucetti, L.; Lassota, P.; Cohen, D.; Bair, K.;
Wood, A.; Remiszewski, S. Cancer Res. 2004, 64, 689–695.
Witter, D. J.; Harrington, P.; Wilson, K. J.; Chenard, M.;
Fleming, J. C.; Haines, B.; Kral, A. M.; Secrist, J. P.; Miller,
T. A. Med. Chem. Lett. 2008, 18, 726–731.
Hu, E.; Dul, E.; Sung, C-M.; Chen, Z.; Kirkpatrick, R.;
Zhang, G-F.; Johanson, K.; Liu, R.; Lago, A.; Hofmann, G.;
Macarron, R.; De Los Frailes, M.; Perez, P.; Krawiec, J.;
Winkler, J.; Jaye, M. J. Pharmacol. Exp. Ther. 2003,
307,720–728.
(a) Liang, J.; Edelsbrunner, H.; Woodward. C. Protein Sci.
1998, 7, 1884-1897. (b) Liang, J.; Edelsbrunner, H.; Fu, P.;
Sudhakar, P. V.; Subramaniam, S. Proteins 1998, 33, 18-29.
(c) Liang, J.; Edelsbrunner, H.; Fu, P.; Sudhakar, P. V.;
Subramaniam, S. Proteins 1998, 33, 1-17.
25.
26.
27.
28.
29.
30.
31.
Breneman, C. M.; Wiberg, K. B. J. Comput. Chem. 1990, 11,
361-373.
45.
46.
Gaussian 03, Revision E.01, Frisch, M. J.; Trucks, G. W.;
Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J.
R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. N.; Burant,
J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.;
Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson,
G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.;
Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda,
Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.;
Hratchian, H. P.; Cross, J. B.; Adamo, C.; Jaramillo, J.;
Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.;
Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.;
Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.;
Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M.
C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari,
K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.;
Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.;
Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox,
D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.;
Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.; Johnson,
B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A
(2004) Gaussian, Inc., Wallingford CT.
Halgren, T. A. J. Comput. Chem. 1996, 17, 616-641.
Still, W. C.; Tempczyk, A.; Hawley, R. C.; Hendrickson, T.
J. Am. Chem. Soc. 1990, 112, 6127–6129.
Morris, G. M.; Goodsell, D. S.; Halliday, R. S.; Huey, R.;
Hart, W. E.; Belew, R. K.; Olson, A. J. J. Comp. Chem.
1998, 19, 1639–1662.
Maulucci, N.; Chini, M. G.; Di Micco, S.; Izzo, I.; Cafaro,
E.; Russo, A.; Gallinari, P.; Paolini, C.; Nardi, M. C.;
Casapullo, A.; Riccio, R.; Bifulco, G.; De Riccardis, F. J.
Am. Chem. Soc. 2007, 129, 3007–3012.
(a) Di Micco, S.; Terracciano, S.; Bruno, I.; Rodriquez, M.;
Riccio, R.; Taddei, M.; Bifulco, G. Bioorg. Med. Chem.
2008, 16, 8635-8642. (b) Grolla, A. A.; Podesta, V.; Chini,
M. G.; Di Micco, S.; Vallario, A.; Genazzani, A. A.;
47.
48.
49.
50.
32.
33.
51.
52.
34.
35.
Somoza, J. R.; Skene, R. J.; Katz, B. A.; Mol, C.; Ho, J. D.;
Jennings, A. J.; Luong, C.; Arvai, A.; Buggy, J. J.; Chi, E.;
Tang, J.; Sang, B-C.; Verner, E.; Wynands, R.; Leahy, E. M.;
Dougan, D. R.; Snell, G.; Navre, M.; Knuth, M. W.;
Swanson, R. V.; McRee, D. E.; Tari, L. W. Structure 2004,
12, 1325-1334.
36.

53.
54.
Balasubramanian, S.; Ramos, J.; Luo, W.; Sirisawad, M.;
Verner, E.; Buggy, J. J. Leukemia 2008, 22, 1026-1034.
Mai, A.; Jelicic, K.; Rotili, D.; Di Noia, A.; Alfani, E.;
Valente, S.; Altucci, L.; Nebbioso, A.; Massa, S.; Galanello,
R.; Brosch, G.; Migliaccio, A. R.; Migliaccio, G. Mol.
Pharmacol. 2007, 72, 1111–1123.
Haggarty, S. J.; Koeller, K. M.; Wong, J. C.; Butcher, R. A.;
Schreiber, S. L. Chem. Biol. 2003, 10: 383-396.
Suzuki, T.; Kouketsu, A.; Itoh, Y.; Hisakawa, S.; Maeda, S.;
Yoshida, M.; Nakagawa, H.; Miyata, N. J. Med. Chem. 2006,
49, 4809-4812.
isoform
Expression in tumor tissues
gastric, pancreatic, colorectal, prostate
cancer; hepatocellular carcinoma
colorectal cancer: upregulated in
polyps; cervical carcinoma; gastric and
prostate cancer: increased expression
associated with advanced stage
gastric, prostate, colorectal cancers
childhood neuroblastoma
breast cancer
colorectal cancer
oral squamous cell cancer; breast
cancer
HDAC1
Class I
HDAC2
55.
56.
HDAC3
HDAC8
HDAC4
HDAC7
Class IIA
Class IIA
57.
Butler, K. V.; Kalin, J.; Brochier, C.; Vistoli, G.; Langley,
B.; Kozikowski, A. P. J. Am. Chem. Soc. 2010, 132, 10842–
10846.
HDAC6
Supplementary Material
Supplementary material associated with this article can be
found, in the online version.
Scheme 1. Work-flow of the different stages involved in the structural
characterization, design and synthesis of new selective HDAC inhibitors.
Table 1. Expression of HDACs in tumor tissues (see also reference 5).

Scheme 2. Molecular structures of compounds 1-39. 1 and 2, pan inhibitors. 3-7, class I selective inhibitors. 8-17, HDAC1,2 selective inhibitors. 18-24, HDAC8
selective inhibitors. 25 and 26, HDAC4 selective inhibitors. 27-36, HDAC6 selective inhibitors. 37-39, designed compounds (see Proof of concept). 4 is the in
vivo reduced active compound of FK228.

Figure 1. Three-dimensional model of the interactions between 8 and HDAC2. The protein (a) and 8 are represented by tube and their atoms are coloured by
atom type: C, gray; polar H, white; N, dark blue; O, red). For 8, carbon and bonds are depicted in green. The zinc ion is represented in orange cpk. The figure
highlights the hydrogen bond (dashed cyan line) between the NH of the amide functionality and the carbonyl of the Gly154. b) The protein is represented by
molecular surface. The figure shows the coordination of zinc ion (represented in red cpk) and the accommodation of metal binder appendage in the internal
cavity.
Figure 2. View of the top of the ≈11 Ǻ channel. The HDAC1 (a) and HDAC2 (b) are represented by ribbons and their catalytic site by molecular surface
coloured in green and orange, respectively. The zinc ion is depicted by light blue cpk. The chelating amino acids are shown by tube, coloured by atom type: polar
H, white; O, red. The C atoms of the amino acids are coloured as for the ribbons: yellow (a) and tan (b). The figure highlights the difference in shape and
dimensions of the tube like channel, leading to the zinc ion.
Figure 3. View of the bottom of the side channel. The HDAC1 (a) and HDAC2 (b) are represented by ribbons and their catalytic site by molecular surface
coloured in green and orange, respectively. The zinc ion is depicted by light blue cpk. The amino acids are shown by tube, coloured by atom type: polar H, white;
O, red. The C atoms of the amino acids are coloured as for the ribbons: yellow (a) and tan (b). The figure highlights the difference in shape and dimensions of the
internal channel, near to the catalytic site of the enzymes.

Figure 5. Three-dimensional model of the interactions between 26 and
HDAC4. The protein is represented by ribbon and the zinc ion is depicted
in dark orange cpk. The side chains of His158, His159 and Tyr170 (light
green), and 26 (white) are shown by tube. The atoms are coloured as: polar
H, white; N, dark blue; O, red. The yellow dashed lines indicate hydrogen
bonds.
Figure 4. Superimposition of HDAC1 (light blue), HDAC2 (cyan),
HDAC3 (yellow) and HDAC8 (blue). The enzymes are represented by
ribbons and tube. The zinc ion is depicted as purple cpk. The figure
highlights the presence of an extra tyrosine (TYR198) in HDAC3
compared to the other class I proteins.
Figure 6. Three-dimensional model of the interactions between 37 and HDAC2. The protein is represented by molecular surface and the zinc ion is depicted in
dark orange cpk. The side chains of amino acids are represented by tube, and coloured as: gly, white; asp, red; phe, brown; his, light blue; leu, green; cys, yellow;
arg, dark blue; met, olive. The 37 is depicted in purple tube and the atoms are coloured as: C, purple; polar H, white; N, dark blue; O, red.

Scheme 3. a) Retro-synthetic approach for the synthesis of derivatives 37-39. b) Synthetic approach for the synthesis of compounds 37-39.
Table 2. Percentage of residual HDAC activity for compounds 37-39 and TSA as reference, obtained at a concentration of 3.3 x 10-5 M.a,b
HDAC subtype
compound
1
2
3
-
4
-
6
-
7
-
8
-
37
-
-
69.38
-
-
-
-
-
-
-
38
-
-
53.58
60.65
5.72
-
39
53.20
0.95
TSA
0.02
0.07
59.23
5.12
35.37
aEmpty cells indicate percentage of activity > 75 %.
bThe values are calculated at 95% confidence intervals.