Characterization of Histone Deacetylase 8 (HDAC8) Selective Inhibition Reveals Specific Active Site Structural and Functional Determinants

Article abstract of DOI:10.1021/acs.jmedchem.8b01087

Metal-dependent histone deacetylases (HDACs) are key epigenetic regulators that represent promising therapeutic targets for the treatment of numerous human diseases. Yet the currently FDA-approved HDAC inhibitors nonspecifically target at least several of

Full text of DOI:10.1021/acs.jmedchem.8b01087

Subscriber access provided by Kaohsiung Medical University
Article
Characterization of histone deacetylase 8 (HDAC8) selective inhibition
reveals specific active site structural and functional determinants
Martin Marek, Tajith Baba Shaik, Tino Heimburg, Alokta Chakrabarti, Julien LANCELOT, Elizabeth Ramos
Morales, Cyrielle Da Veiga, Dmitrii V. Kalinin, Jelena Melesina, Dina Robaa, Karin Schmidtkunz, Takayoshi
Suzuki, Ralph Holl, Eric Ennifar, Raymond Pierce, Manfred Jung, Wolfgang Sippl, and Christophe Romier
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.8b01087 • Publication Date (Web): 22 Oct 2018
Downloaded from http://pubs.acs.org on October 22, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 57
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Characterization of histone deacetylase 8 (HDAC8) selective
inhibition reveals specific active site structural and functional
determinants
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
†‡
1‡
2
3
4
Martin Marek *, Tajith B. Shaik , Tino Heimburg , Alokta Chakrabarti , Julien Lancelot , Elizabeth
1
5
6
2
2
Ramos-Morales , Cyrielle Da Veiga , Dmitrii Kalinin , Jelena Melesina , Dina Robaa , Karin
3
7
8
5
4
3
Schmidtkunz , Takayoshi Suzuki , Ralph Holl , Eric Ennifar , Raymond J. Pierce , Manfred Jung ,
2
1
Wolfgang Sippl & Christophe Romier *
1
Institut de Génétique et Biologie Moléculaire et Cellulaire (IGBMC), Département de Biologie
Structurale Intégrative, Université de Strasbourg, Centre National de la Recherche Scientifique (CNRS
UMR7104), Institut National de la Santé et de la Recherche Médicale (INSERM U1258), 1 rue Laurent
Fries, 67404 Illkirch Cedex, France.
2
Institute of Pharmacy, Martin-Luther-Universität Halle-Wittenberg, Wolfgang-Langenbeck-Straße 4,
0
6120 Halle/Saale, Germany.
3
4
5
Institute of Pharmaceutical Sciences, Albert-Ludwigs-Universität Freiburg, Albertstraße 25, 79104
Freiburg, Germany.
Université de Lille, CNRS, Inserm, CHU Lille, Institut Pasteur de Lille, U1019 - UMR 8204- CIIL -
Centre d'Infection et d'Immunité de Lille, F-59000 Lille, France.
Architecture et Réactivité de l’ARN, Institut de Biologie Moléculaire et Cellulaire (IBMC), UPR 9004
du CNRS, Université de Strasbourg, 15 Rue René Descartes, 67084 Strasbourg Cedex, France.
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 57
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
6
7
Institute of Pharmaceutical and Medicinal Chemistry, University of Münster, Corrensstr. 48, 48149
Münster, Germany.
Graduate School of Medical Science, Kyoto Prefectural University of Medicine, 1-5 Shimogamohangi-
Cho, Sakyo-Ku, 606-0823 Kyoto, Japan - CREST, Japan Science and Technology Agency (JST), 4-1-
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
8
Honcho Kawaguchi, 332-0012 Saitama, Japan.
8
Institute of Organic Chemistry, Department of Chemistry, University of Hamburg, Martin-Luther-King-
Platz 6, 20146 Hamburg, Germany.
ABSTRACT: Metal-dependent histone deacetylases (HDACs) are key epigenetic regulators that
represent promising therapeutic targets for the treatment of numerous human diseases. Yet, the currently
FDA-approved HDAC inhibitors non-specifically target at least several of the eleven structurally similar
but functionally different HDAC isozymes, which hampers their broad usage in clinical settings.
Selective inhibitors targeting single HDAC isozymes are being developed, but precise understanding in
molecular terms of their selectivity remains sparse. Here, we show that HDAC8-selective inhibitors adopt
a L-shaped conformation required for their binding to a HDAC8-specific pocket formed by HDAC8
catalytic tyrosine and HDAC8 L1 and L6 loops. In other HDAC isozymes, a L1-L6 lock sterically
prevents L-shaped inhibitor binding. Shielding of the HDAC8-specific pocket by protein engineering
decreases potency of HDAC8-selective inhibitors and affects catalytic activity. Collectively, our results
unravel key HDAC8 active site structural and functional determinants important for the design of next-
generation chemical probes and epigenetic drugs.
2
ACS Paragon Plus Environment

Page 3 of 57
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
INTRODUCTION
Acetylation of lysine residues in proteins is a major signaling mark that impacts most cellular processes1-
3. In the cell nucleus, acetylation of histones has been shown to be essential for modulating chromatin
structure and for acting in epigenetic signaling that drives and regulates nuclear mechanisms, cellular
processes and development1, 3, 4. Protein lysine acetylation is a reversible process relying on the opposing
effects of acetyltransferases and deacetylases3, 5, 6. In addition, the acetylation marks on lysines are
recognized by epigenetic readers harboring structural modules (e.g. bromodomains) that enable the
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
, 7
recruitment of cellular effectors to specific subcellular and genomic loci .
Due to the functional importance of acetylation mechanisms, their deregulation has been linked with
multiple human diseases, including cancer4, 8-11. The reversibility of acetylation and the possibility of
modulating recognition of acetylated lysines by bromodomains provide a way to pharmacologically
modulate acetylation pathways. Thus, epigenetic regulators involved in these pathways represent
important therapeutic targets4, 8-10, 12, 13
.
Accordingly, among the currently approved epigenetic drugs, a majority (Vorinostat (SAHA),
Romidepsin, Belinostat, Panobinostat and Chidamide) target lysine deacetylases13-15. The family of
lysine deacetylases has been divided into four classes depending on their folds and their sequence
similarities. Classes I, II (IIa and IIb), and IV adopt an arginase-deacetylase α/β fold and rely on a zinc
6
ion for activity (thereafter referred to as histone deacetylases or HDACs) . Class III deacetylases are
+
6
referred to as sirtuins and adopt a Rossmann fold, relying on NAD for activity . Eleven HDACs and
seven sirtuins are found in humans.
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 57
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
The currently approved drugs against lysine deacetylases target only proteins from the HDAC family.
However, these drugs show poor selectivity against single members of the structurally similar but
functionally different human HDAC isozymes, targeting at least two and generally more than two HDAC
isozymes, thus hampering their broad therapeutic usage13, 15. Several small-molecule inhibitors exhibiting
selectivity for specific HDACs have been developed. PCI-34051 and NCC-149 were among the first
HDAC isozyme-selective inhibitors discovered16-18. These two aromatic hydroxamate derivatives show
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
high selectivity for human HDAC8 (hHDAC8), an HDAC isozyme that has been shown to be
overexpressed in several cancers19-21 and whose mutations can lead to the Cornelia de Lange syndrome22-
24.
Specifically, PCI-34051, which is an indole-based derivative, is currently among the most selective
HDAC8 inhibitors with a selectivity index of 290 and 400 for HDAC6 and HDAC1, respectively, making
it a strong chemical tool for studying the biological role of HDAC8 in vivo16, 25-28. In addition, our work
on HDAC8 from the human-pathogenic flatworm Schistosoma mansoni (smHDAC8) has led to the
design of new selective HDAC8 inhibitors targeting both the human and schistosome enzymes or, for
some of them, showing selectivity towards schistosome HDAC829, 30
.
To date, the experimental structural and mechanistic bases underlying HDAC8 selective inhibition by
PCI-34051, NCC-149 and other related selective compounds remain poorly understood. Structural work
on HDAC8 and computational studies have suggested that selective inhibitors target specific structural
features of HDAC8 active site, potentially making use of the malleability of this active site17, 18, 31-34. Yet,
no clear experimental evidence has been provided so far. To address this issue, we have dissected the
molecular basis of HDAC8-selective inhibition by combining biochemical, biophysical and
crystallographic studies on hHDAC8 and smHDAC8.
4
ACS Paragon Plus Environment

Page 5 of 57
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Our results reveal that HDAC8-selective inhibitors bind into a specific HDAC8-selective pocket formed
by the active site catalytic tyrosine and by residues from the L1 and L6 loops. This specific enzyme-
ligand recognition is favored by the constrained L-shaped conformation of HDAC8-selective inhibitors.
This selective binding relies on a specific conformation of the HDAC8 L6 loop and a shorter L1 loop
that are not observed in any other HDAC isozyme. Our mutational studies further reveal HDAC8
structural determinants that support HDAC8 selective inhibition and function. Collectively, our results
highlight the structural/functional similarities and dissimilarities between the various HDAC isozymes
and pave the way for the development of new HDAC isozyme-selective chemical probes for cell
biological research and inhibitors to treat human diseases.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
RESULTS
In vitro and in vivo effects of PCI-34051 and NCC-149
PCI-34051 and NCC-149 have been developed to target human HDAC8 (hHDAC8) selectively16-18. To
investigate whether these inhibitors also target Schistosoma mansoni HDAC8 (smHDAC8), we have
looked at their inhibition and binding to smHDAC8. As a comparison, we have used the highly potent
but non-selective HDAC inhibitor Quisinostat (QSN) which is in Phase II of clinical trials35, 36
.
Measurements of the maximal-half inhibitory concentration (IC ) showed that all three inhibitors
5
0
possess activity in the submicromolar range against hHDAC8 and smHDAC8. NCC-149 showed the
most potent inhibition, followed by QSN and PCI-34051 (Figure 1). Measurement of the thermodynamic
parameters using Isothermal Titration Calorimetry (ITC) confirmed the inhibition results obtained, the
equilibrium dissociation constant (Kd) values determined being in the same range as the corresponding
IC values, with the exception of QSN that showed a lower Kd value for smHDAC8 (Figures 1 and 2).
5
0
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 57
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
The biological effects, especially the anti-cancer properties of PCI-34051 and NCC-149 in various cell
types have been characterized16-18, 25-28, and we have previously shown that pan-HDAC inhibitors affect
schistosome pathogens30, 37. We therefore asked whether PCI-34051 and NCC-149 could also have anti-
parasitic effect on schistosomes. Our various biological assays confirmed that PCI-34051 and NCC-149
affect the pathogens, triggering their apoptosis (Figure S1). These results demonstrate that both hHDAC8
and smHDAC8 can be used for studying HDAC8 inhibition by PCI-34051 and NCC-149.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
So far, few structures of HDACs in complex with selective inhibitors have been solved. Moreover, in
many HDAC/inhibitor structures, the active site of the HDAC and the bound inhibitor are involved in
extensive crystal packing contacts. This complicates the delineation between biologically-relevant and
crystal packing-driven conformations and interactions of the HDAC active site loops and the inhibitors.
Therefore, in addition to the co-crystallization attempts of hHDAC8 with inhibitors, we have used the
possibility offered by our apo smHDAC8 crystals to look at HDAC8/inhibitor interactions in a crystal
lattice-open environment30.
Despite intensive efforts, we were not able to obtain well-diffracting crystals of hHDAC8 in complex
with PCI-34051, NCC-149 and QSN. In contrast, soaking experiments of apo smHDAC8 crystals with
all three inhibitors were successful and yielded high resolution complex structures (Figure 3; Table S1).
Binding mode of QSN to smHDAC8
Analysis of the smHDAC8/QSN structure revealed that QSN adopts a straight conformation as its
piperidine-pyrimidine linker allows limited conformational flexibility (Figure 3A). The QSN
hydroxamate warhead coordinates the catalytic zinc and simultaneously interacts via hydrogen bonding
with the histidine dyad, H141 and H142 (hHDAC8 H142 and H143) and with the catalytic tyrosine Y341
(hHDAC8 Y306) hydroxyl, as commonly observed for most other hydroxamate-containing HDAC
6
ACS Paragon Plus Environment

Page 7 of 57
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
inhibitors. Furthermore, the QSN piperidine-pyrimidine linker is sandwiched between the side chains of
smHDAC8 F151 (hHDAC8 F152) and F216 (hHDAC8 F208), where it forms planar π-π stacking and
non-polar contacts.
Specifically, the QSN piperidine ring adopts a chair conformation, which allows the QSN methyl-amino-
methyl linker to form a hydrogen bond (2.4 Å) with the carboxyl group of smHDAC8 D100 (hHDAC8
D101), a conserved class I HDAC residue that has been shown to interact with the backbone of incoming
acetylated peptides33, 38. Finally, the QSN capping methyl-indole group is solvent exposed, making
minimal non-polar contacts with Y99 (hHDAC8 Y100). Interestingly, we previously observed a very
similar binding mode to smHDAC8 for another pan-HDAC inhibitor, M344, including an interaction
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
30
between D100 and the M344 internal amide group . The M344 conformation is less constrained by its
linker, which suggests that this binding mode is common to and favored by many pan-HDAC inhibitors.
In agreement, a similar binding mode was also observed upon SAHA binding to human HDAC2 in a
crystal lattice-free environment39.
Binding mode of PCI-34051 to smHDAC8
The PCI-34051 hydroxamate warhead interacts with the catalytic zinc and active site residues as
observed for QSN (Figure 3B). However, in contrast to QSN, the hinge connecting the central indole-
based spacer and the methoxyphenyl group of PCI-34051 favours binding of its capping group onto the
side chain of smHDAC8 Y341 (hHDAC8 Y306). This tyrosine, together with the catalytic zinc, has been
shown in hHDAC8 to be involved in catalysis by polarizing the leaving acetyl group of the incoming
33
acetylated lysine . Here, the methoxyphenyl capping group is perpendicularly (86°) oriented over the
aromatic ring of this tyrosine, which favours T-shaped π-π stacking (4.9 Å). Thus, the binding of PCI-
34051 onto Y341 is favored by the L-shape of this inhibitor.
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 57
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
The methoxyphenyl capping group of PCI-34051 is positioned in close vicinity to the smHDAC8 L6
loop, being inserted in a small pocket shaped by the side chains of P291 and H292 (hHDAC8 P273 and
M274). While the methoxy group forms non-polar contacts with the pyrrolidine ring of P291, the phenyl
ring of the inhibitor interacts (4.3 Å) via either π-π or cation-π interaction with H292, depending upon
the protonation state of the histidine (Figure 3B).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The smHDAC8 and hHDAC8 differ by one residue in their active sites, where hHDAC8 M274 is
replaced by smHDAC8 H292. Since this latter residue is involved in inhibitor binding, we asked whether
the smHDAC8-H292M mutant binds PCI-34051 in the same way as the wild-type (WT) enzyme. The
crystal structure of PCI-34051 bound to the “humanized” smHDAC8-H292M mutant reveals that PCI-
3
4051 still adopts a L-shaped conformation when bound to the smHDAC8-H292M mutant, but this
conformation is slightly different from the one adopted with the WT enzyme (Table S1; Figure S2A-C).
Specifically, PCI-34051 still lies over Y341 catalytic tyrosine but appears more centred in the pocket
created by smHDAC8 Y341, F151 and the L6 loop. In contrast to the WT enzyme, the central indole
group of PCI-34051 is axially rotated by an angle of ~20°, which favors the positioning of the capping
methoxyphenyl group over the aromatic ring of Y341 (4.8 Å), effecting nearly parallel (8.9°) π-π
stacking. As a consequence, the PCI-34051 capping group interacts differently with the L6 loop than is
observed with the WT enzyme, but still making close non-polar contacts with the aliphatic ring of P291
and the side chain of M292 in this loop (Figure S2A).
Binding mode of NCC-149 to smHDAC8
NCC-149 hydroxamate also binds in a canonical way to the catalytic zinc and active site residues, and
the rest of the L-shaped inhibitor is turned towards and interacts with Y341 and the smHDAC8 L6 loop
(Figure 3C). Specifically, the 1,2,3-triazole ring of the linker is oriented in a position (4.9 Å) that is
8
ACS Paragon Plus Environment

Page 9 of 57
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
slightly off perpendicular (~83°) to the aromatic ring of Y341, indicating their π-π contacts. At the same
time, the 1,2,3-triazole ring packs against L6 loop H292, which allows their mutual T-shaped (~67°)
aromatic interactions. In addition, and as observed for PCI-34051, the phenylthiomethyl capping group
of NCC-149 is inserted in the small subpocket of the HDAC8 L6 loop, where it effects both upright
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(~76°) π-π stacking with H292 and hydrophobic contacts (3.6 Å) with P291.
We also solved the structure of NCC-149 bound to the smHDAC8-H292M mutant. Here, the
hydroxamate and linker of NCC-149 bind very similarly to the smHDAC8-H292M mutant and to the
WT enzyme, and show fewer conformational changes than observed with PCI-34051. Interestingly, the
1,2,3-triazole ring is closer to the L6 loop, where it interacts with M292 via a sulphur-aromatic interaction
(3.7 Å), suggesting a similar interaction with hHDAC8 (Figure S2D-F). This binding mode still favors
T-shaped (82°) π-π stacking between the 1,2,3-triazole and Y341 (4.7 Å), as well as hydrophobic contacts
between the internal benzene ring and the two phenylalanines, F151 and F216. However, the terminal
phenylthiomethyl capping group of the inhibitor changes its position and is turned away from the L6
loop, lying in another binding subpocket formed by smHDAC8 K20 and F21 (L1 loop) and Y341 and
F343 (L7 Loop) whose hydrophobic character is conserved in hHDAC8 (Figure S2D).
The conformational adaptation of NCC-149 to the smHDAC8-H292M selective pocket is eased by the
intrinsically higher conformational flexibility of this inhibitor that allows the repositioning of its capping
group. In the case of PCI-34051, which is more rigid as it contains only a one-atom hinge, a major part
of the inhibitor had to be repositioned. Yet, these changes do not affect the major interaction of HDAC8-
selective inhibitors with the uncovered aromatic ring of the catalytic tyrosine and with residues of the L6
loop, showing the importance of these elements as key binding surfaces for these selective inhibitors.
Selective inhibition of smHDAC8 over other human HDACs
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 57
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Previous work on the selective inhibition of smHDAC8 has yielded the development of an inhibitor series
of 3-benzamido-benzohydroxamates that show strong selectivity for smHDAC8 and hHDAC8 over other
2
9
human HDACs . The structure of smHDAC8 with the simplest inhibitor of this series (1) revealed that
the capping benzamido moiety lays over Y341 (3.9 Å), its benzene ring capping group further making
non-polar contacts with the smHDAC8 L6 loop, notably with P291 (3.6 Å) (Figure 4A). This inhibitor
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
9
also exploits smHDAC8-specific interactions with residues K20 and H292 (hHDAC8 K33 and M274) .
Many of the 3-benzamido-benzohydroxamate inhibitors that were subsequently developed displayed
2
9
higher potency than 1 in inhibiting smHDAC8 . To understand the molecular basis of these
observations, we have further solved the structures of smHDAC8 bound to several of these inhibitors
(compounds 2-11) (Figure 4; Figure S3; Table S2; Table S3).
All these compounds showed a similar mode of binding to smHDAC8 as 1. Yet, slight differences in
chemical composition impacted specific interactions, potentially relating to the differences in the IC50
values observed. Compounds 2-4 only have different substituents at the para position of the
benzohydroxamate moiety compared to 1. These compounds bind very similarly to smHDAC8 as 1
(
Figure S3). Their lower IC values most likely stem from the additional contacts of their substituent
50
groups with F216, as well as the possible stabilization by these groups of the non-canonical geometry of
the amide group of these inhibitors.
In the case of 5 and 6, which have respectively larger biphenyl and benzothiophene capping groups, these
latter form more extensive hydrophobic contacts (3.5 Å) with P291 (Figure 4B; Figure S3). In the case
of 7, which only has an inverted internal amide compared to 1, the orientation and the length of the
hydrogen bonds between 7 and smHDAC8 K20 and H292 appear more favorable for interaction (Figure
S3).
1
0
ACS Paragon Plus Environment

Page 11 of 57
1
Journal of Medicinal Chemistry
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Compounds 8-10 are particularly interesting since they also show a higher selectivity for smHDAC8
over hHDAC8 (3-, 4.5-, and 6-fold; respectively) (Figure 4C; Figure S3; Table S2). 8 has one additional
methylene group between the internal amide and the phenyl capping group compared to 1. Compound 8
appears to bind less deeply in the pocket to maximize its interactions with Y341 and the L6 loop. This
change is compatible with the presence of smHDAC8 H292, but would be sterically unfavourable with
hHDAC8 M274, which possibly explains the weaker inhibition observed for the human enzyme.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Compounds 9 and 10 both have a dichlorophenyl rather than a phenyl capping group, and bind perfectly
into the pocket formed by smHDAC8 Y341 and the L1 and L6 loops (Figure 4C; Figure S3). The
presence of the halogen atoms in the capping group of 9/10 forces the inhibitor to be slightly tilted
towards the L6 loop, where it forms cation-π interaction (4.3 Å) with smHDAC8 H292. The bulkier
character of the dichlorophenyl capping group may complicate the adaptation of 9/10 to the active site
of hHDAC8.
Indeed, the predicted docking poses of 9 and 10 in hHDAC8 show that their dichlorophenyl capping
groups are turned around 180° in comparison to their position in smHDAC8 and effects van der Waals
interactions with the hydrophobic residues of the L6 loop P273 and M274 (Figure S4). When comparing
the obtained IC values of compounds 9/10 bearing a dichlorophenyl capping group with their parent
5
0
counterparts 3/4, which have a phenyl capping group, it becomes clear that the observed selectivity of
the former compounds for smHDAC8 over hHDAC8 arises from a significantly decreased inhibitory
activity towards hHDAC8. Compared to compounds 3/4, compounds 9 and 10 show a 7- to 10-fold
decrease in their inhibitory activities towards hHDAC8 (Table S2). To further explain the selectivity of
the dichlorophenyl derivatives, we carried out 100 ns molecular dynamics (MD) simulations for
smHDAC8/10 and hHDAC8/10 complexes, as well as for smHDAC8/4 and hHDAC8/4 complexes for
comparison.
1
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 57
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
MD simulations of smHDAC8/4 crystal structure (Figure S5A) reveal a relatively stable binding mode,
where the phenyl capping group is placed parallel to Y341 in the side pocket, forming π-π stacking
interactions, in addition to cation-π interactions with H292. During the MD simulation of smHDAC8/10
crystal structure (Figure S5B), the dichlorophenyl capping group similarly remains in the side pocket.
Despite the loss of the π-π stacking interactions with Y341, compound 10 is still able to form strong
cation-π interactions with H292, which are stable throughout MD. This might explain why the
dichlorophenyl derivative only shows a slight decrease in inhibitory activity as compared to compound
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4
.
Similarly, MD simulations of the hHDAC8/4 docking complex (Figure S5C) show that the phenyl
capping group mostly remains stable in the side pocket, where it is placed perpendicular to Y306 of the
side pocket, showing strong π-π stacking interactions besides van der Waals interactions with M274. In
contrast, our MD simulation of hHDAC8/10 structure indicates that the predicted binding mode of the
dichlorophenyl derivative 10, where the capping phenyl ring is situated parallel to Y306, is not stable.
Within less than 1 ns of the MD simulation time, the capping group flips by 90° and is placed
perpendicular to Y306 with the o-chloro substituent pointing towards the side chain of Y306. (Figure
S5D). The π-π stacking interactions with Y306, observed with compound 4, are thus lost. Thus, a possible
explanation for the decreased inhibitory activity of 10 for hHDAC8 as compared to compound 4 is that
the dichlorophenyl capping group can only form weak van der Waals interactions between its o-chloro
group and the side chain of Y306 and between its phenyl capping group and the side chain of M274.
We have used another benzohydroxamate inhibitor (11) which has an internal amine rather than an
internal amide in its linker and shows a low nM IC for hHDAC8 but only a low M IC for smHDAC8
5
0
50
(Table S2). The smHDAC8/11 structure reveals that this inhibitor does not form any strong interaction
with K20 and H292 and its capping group interacts less extensively with the L6 loop (Figure S3; Figure
1
2
ACS Paragon Plus Environment

Page 13 of 57
1
Journal of Medicinal Chemistry
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
S6). The inhibitor binds centrally into the pocket, where it forms hydrophobic contacts with F216 (3.4
Å) and Y341 (3.7 Å), in a conformation identical to that observed for PCI-34051 when bound to the
smHDAC8-H292M mutant (Figure S6). This suggests that HDAC8-selective inhibitors bind more
centrally in this pocket in hHDAC8 than in smHDAC8 due to the slight different physico-chemical
properties of the active sites of these two proteins.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Finally, an unrelated compound, 12, which has a triazole linker, binds similarly to smHDAC8, interacting
with Y341 and the L6 loop (Figure S3). This compound also adopts an L-shaped conformation to bind
to the enzyme. 12 does not make direct contacts to K20 and H292, which might explain its higher IC50
value for smHDAC8 (Table S2). This compound displays, however, a 4-fold higher potency for
smHDAC8 over hHDAC8. Collectively, our results highlight how small chemical variations may be used
to influence inhibition potency.
Structural specificity of the HDAC8-selective pocket
Our findings show an HDAC8 selective inhibition relying on the binding of the selective inhibitors to a
pocket that forms a shallow groove and that we have termed HDAC8-selective pocket. This pocket is
defined by the catalytic tyrosine side-chain (L7 loop), which forms the pocket bottom, and residues from
the L6 loop and, to a lesser extent, of the L1 loop of HDAC8 that both form the sides of the pocket.
The HDAC8-selective pocket is highly specific to this enzyme. Indeed, in class IIa HDACs, the catalytic
tyrosine is replaced by a histidine whose side chain is turned away from the active site and cannot provide
the same interaction surface as HDAC8 catalytic tyrosine (Figure 5). The situation is different for HDAC
isozymes 1, 2, 3, 6 and 10 (thereafter called HDAC1-3,6,10) that all have retained a catalytic tyrosine at
the same position. Specifically, these isozymes have a L6 loop that displays a similar conformation to
that observed in HDAC8 (Figure 5). L6 loops in HDAC1-3,6,10 however protrude slightly more over
1
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 57
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
the catalytic tyrosine side chain than in the case of HDAC8 and could sterically perturb the binding of
HDAC8-selective inhibitors (Figure 5).
Yet, our results with the smHDAC8-H292M mutant show that HDAC8-selective inhibitors can adapt to
small changes within the HDAC8-selective pocket and could potentially overcome a more protrusive L6
loop. However, another specific and conserved feature of HDAC1-3,6,10 is a larger L1 loop that extends
towards the L6 loop. Notably, at the tip of the HDAC1-3,6,10 L1 loop, a proline (or an isoleucine residue
in HDAC10) is present that forms hydrophobic interactions with L6 loop residues and the catalytic
tyrosine, thus forming a lock over the catalytic tyrosine and preventing the formation of a pocket similar
to the HDAC8-selective pocket (Figure 5).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
In HDAC8, the L1 and L2 loops have been shown to display flexibility, being able to change their
conformation to adapt to different inhibitors such as the large Largazole inhibitor and its analogues
(
Figure S7)40, 41. This raises the question of whether the same L1 loop flexibility might exist in HDAC1-
3
,6,10. Current structural data on HDAC1-3,6,10, however, show that their L1 loops make much more
extensive contacts with the rest of the enzyme, which most likely explains this lack of flexibility. Notably,
the recent work on HDAC6/inhibitor complexes demonstrate that L1 loop conformational stability is
important for the interactions of L1 loop residues with inhibitors capping groups to achieve HDAC6
selectivity inhibition42-46
.
These major structural differences between HDAC8 and the other HDAC isozymes would explain how
HDAC8 selective inhibition is achieved by inhibitors such as PCI-34051 and NCC-149. This is
supported by docking studies of PCI-34051 and NCC-149 with other HDAC isozymes (Figure S8).
These studies show that in case of HDAC1-3 and HDAC10, these inhibitors cannot reach properly and
chelate the catalytic zinc ion. This is not as clear for the HDAC6 isozyme. Therefore, to assess the
stability of the predicted docking poses and to further study the selectivity of PCI-34051 and NCC-149
1
4
ACS Paragon Plus Environment

Page 15 of 57
1
Journal of Medicinal Chemistry
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
towards HDAC8, we performed comparative 100 ns MD simulations on human HDAC6 and HDAC8
isoforms complexed with both inhibitors (Figure S8 and S9). In hHDAC8, PCI-34051 shows stable
binding with an RMSD at ~2 Å compared to the predicted docking pose. The capping group shows little
deviation and remains in the side pocket throughout the 100 ns MD simulation (Figure S9A). The same
is observed for NCC-149, which displays a stable RMSD at ~2 Å compared to the docking pose, except
for a short increase in RMSD at 65-70 ns, with the capping group mostly remaining in the side pocket
(Figure S10A).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
In smHDAC8, PCI-34051 shows higher RMSD deviation at ~ 3-4 Å compared to the crystal structure.
The high RMSD arises from the movement of the capping group, which however remains in the side
pocket throughout the simulation time (Figure S9B). The MD simulation of the smHDAC8/NCC-149
crystal structure yielded similar results as observed for the smHDAC8/PCI-34051 complex (Figure
S10B).
In contrast, MD simulations of hHDAC6/PCI-34051 and hHDAC6/NCC-149 complexes, as predicted
by docking, show high RMSD deviations for both ligands with RMSD values at ~5 and up to 7Å,
respectively (Figures S9C and S10C). Throughout the 100 ns simulation, the ligands maintain their
chelation to the catalytic zinc ion. Meanwhile, the capping group of both ligands does not show any
preferred conformation and remain surface exposed throughout the MD simulation, showing little
interactions with the surrounding protein residues (Figure S9C and S10C). Collectively, the MD
simulations confirm the hypothesis made upon docking and are in agreement with the biochemical,
biophysical and crystallographic results obtained.
Essential roles of L1 and L6 loops in HDAC8 catalysis and inhibition
1
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 57
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
To further question the importance of HDAC8 L1 and L6 loops conformation in catalysis and inhibitor
binding, we have performed a mutational analysis of these loops in HDAC8 and studied the effect of
these mutations by biochemical, biophysical and structural means. Despite an identical number of
residues, the HDAC8 L6 loop has a conformation that is slightly different from that adopted by the
corresponding loops in HDAC1-3,6,10. This slight conformational change prevents L6 loop residues
from protruding over the catalytic tyrosine (Figure 5).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Sequence and structural comparisons highlighted two residues that could, at first sight, be responsible
for this specific conformation of HDAC8 L6 loop: hHDAC8 P273/smHDAC8 P291 and hHDAC8
C275/smHDAC8 R293. However, we could not exclude the possibility that larger rearrangements are
required, and we therefore created several different mutants for both hHDAC8 and smHDAC8 (Figure
6
A; Table S4).
First, specific point mutants (hHDAC8 P273R and P273R/C275G and smHDAC8 P291R and
P291R/R293G) were made, where the residues were replaced by their HDAC1 counterparts. Second, we
created mutants where we exchanged the HDAC8 L6 loop completely with that of HDAC1. We also
constructed mutants where not only the L6 loop but also the L1 loop had been exchanged. Finally, triple
mutants were also generated where a leucine (hHDAC8 L179/smHDAC8 L187) was replaced by an
isoleucine, as observed in HDAC1, since this mutation could possibly facilitate an L6 loop
conformational change (Figure 6A; Table S4).
Except for the point mutants, all smHDAC8 mutants turned out to be insoluble. In contrast, all hHDAC8
mutants were soluble. Surprisingly, activity assays showed that all mutants had drastically reduced
activity (Table S4). While smHDAC8 single point mutants lost around one third of their activity,
hHDAC8 point mutants or L6 loop replacement mutants showed around 10-fold activity loss. Mutants
1
6
ACS Paragon Plus Environment

Page 17 of 57
1
Journal of Medicinal Chemistry
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
of hHDAC8 combining L6 loop replacement with either L1 loop replacement or the L179I point mutation
showed a 50-fold loss of activity, and the triple mutant displayed almost no activity.
We used Thermal Shift Assay experiments to assess whether the mutations affect the stability of the
various mutants. All mutants only showed a decreased Tm of about 5°C compared to the WT enzymes
but did not indicate partial or complete unfolding of the proteins (Table S4). Due to the residual activity
of the mutants, we next measured the IC values for PCI-34051, NCC-149 and QSN for all mutants.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0
All inhibitors showed significantly higher IC values indicative of a poorer inhibition capacity (Figure
5
0
6
B; Table S5). This was most pronounced for inhibitor PCI-34051, with a more than 64-fold increase of
the IC value for the triple HDAC8 mutant. The IC values for NCC-149 were also significantly
5
0
50
increased, but not as much as for PCI-34051. QSN also displayed decreased inhibition, albeit to a much
lesser extent, possibly mirroring only the slight stability decrease of the mutants but supporting our
conclusions on the crucial role of L1 and L6 loops for HDAC8 selective inhibition.
To investigate the molecular basis of conformational changes of the L1 and L6 loops following
mutations, we attempted to solve the structures of the different HDAC8 mutants in complex with PCI-
3
4051, NCC-149 and QSN. Although different mutants gave crystals in presence of some of these
inhibitors, only the crystals obtained with the human HDAC8-mL6/QSN complex led to exploitable
structural data (Table S6). To our knowledge this is the first structure of compound QSN bound to human
HDAC8. Here again the capping group of the inhibitor is extensively involved in crystal packing and it
is impossible to understand whether the binding conformation of QSN, which is different to that observed
when bound to smHDAC8, represents a favored binding conformation in solution to hHDAC8 (Figure
S11). Strikingly, the structure of this complex revealed that, despite the complete exchange of the L6
loop, the conformation of this loop remains as observed in hHDAC8 WT (Figure 6C). The hHDAC8
P273R replacement, however, leads to the partial occupation of the HDAC8-selective pocket by the side
1
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 57
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
chain of the arginine replacing the proline (Figure 6). In fact, the only mutation common to all our
smHDAC8 and hHDAC8 mutants is this proline to arginine change. In hHDAC1 and hHDAC3 this
arginine is involved in inositol phosphate binding and is important for activity38, 47, 48. Our results
therefore further highlight the importance of the L6 loop for class I HDACs activity and inhibitor binding
and pinpoint differences between HDAC8 and the other members of this class that can be used for
selective inhibition of the former enzyme.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
DISCUSSION AND CONCLUSIONS
To date, HDAC8 is one of the most investigated HDAC for selective inhibition and numerous
crystallographic studies have been performed on human HDAC8 in complex with, mostly, non-selective
inhibitors to understand the molecular basis of the inhibition of this enzyme33, 34, 40, 41, 49-54. Many of these
studies have revealed how HDAC8 active site residues and loops, notably the L1 and L2 loops, display
flexibility, thus making the design of HDAC8 selective inhibitors a complicated task. In addition, in
many structures, the bound inhibitor is involved in extensive crystallographic contacts, which makes
difficult to assess whether the observed conformations of the inhibitor, notably of its capping group, but
also potentially of the active site loops, are biologically favored.
Yet, HDAC8 selective inhibitors have been designed, showing that selective inhibition of this enzyme
can be achieved. One of the most potent and best known HDAC8 selective inhibitor is PCI-3405116.
However, so far, the molecular basis for its selective inhibition remains poorly understood. We have
previously shown that our smHDAC8 crystals can be used to look at inhibitor binding in a crystal lattice-
unbiased manner29, 30, 55-58. Here, we have used this feature to look at HDAC8 selective inhibition using
a large set of selective inhibitors of this enzyme, including PCI-34051.
1
8
ACS Paragon Plus Environment

Page 19 of 57
1
Journal of Medicinal Chemistry
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Our results provide a comprehensive and detailed molecular view of HDAC8 selective inhibition,
highlighting a specific HDAC8-selective pocket where selective inhibitors form preferential interactions
with HDAC8 catalytic tyrosine and L6 loop residues. These results particularly emphasize the balance
between chemical structure and inherent conformational flexibility of the inhibitors, with important
implications for selective inhibition. In addition, the restricted conformation of HDAC8 L6 loop,
compared to the more flexible character of the L1 loop, points out the importance of the L6 loop for
selective inhibition. Moreover, the constrained conformation of the L7 loop that bears the catalytic
tyrosine also appears as an essential feature of HDAC8 selective inhibition. Specifically, the position of
the tyrosine side chain, which is locked by the interaction of its hydroxyl with the inhibitor’s
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
9
hydroxamate, most likely constrains the reported flexibility of the L7 loop .
Interestingly, in class IIa HDACs the replacement of the catalytic tyrosine by a histidine creates a specific
pocket at the position of the tyrosine side chain. This feature prevents HDAC8-selective inhibitors from
binding to class IIa HDACs and has been exploited to design class IIa-selective inhibitors that cannot
6
0
distinguish, however, between the different class IIa isozymes .
In the case of HDAC1-3,6,10, although the catalytic tyrosine is conserved and the L6 loop is similar in
these isozymes, the HDAC8-selective pocket is not present due to a larger L1 loop in HDAC1-3,6,10
that covers the catalytic tyrosine and forms a lock together with loop L6 over this selectivity pocket. The
structures of HDAC6 in complex with HDAC6-selective inhibitors show that the capping group of these
inhibitors interact with the proline at the tip of the L1 loop as well as with the preceding histidine residue.
This mirrors the interaction observed between the HDAC8-selective inhibitors and the smHDAC8 L6
loop P291-H292 motif. Surprisingly, these inhibitors are selective for HDAC6 although the same motif
is conserved in HDAC1-3 L1 loops that show similar conformations as HDAC6 L1 loop. Selective
inhibition appears to rely on small structural differences between HDAC1-3 and HDAC6 and on small
1
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 57
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
thermodynamic differences in selective inhibitor binding42-46. This parallels our study that demonstrate
how selective inhibition can be affected by small changes in protein conformation and inhibitor
composition.
Interestingly, the HDAC8-selective pocket can adapt to bulkier inhibitors that can also interact with the
back of the pocket, as observed for compound NCC-149 when bound to the smHDAC8-H292M mutant,
thus potentially paving the way for the design of more potent HDAC8-selective inhibitors. Of note, the
reduced conformational flexibility of compound PCI-34051, if it prevents this inhibitor from easily
adapting to small changes within the HDAC8 active site, might also explain why this inhibitor is so
selective for this enzyme.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The existence of a unique selective specific pocket in HDAC8 and the effects observed on activity when
this pocket is partially occupied indicate that it might also be essential for the recognition of specific
targets. Collectively, our results unravel the structural determinants underlying HDAC8 selective
inhibition and pave the way toward the design of more potent HDAC selective inhibitors, the
development of novel epigenetic drugs and the delineation of HDACs specific biological role through
chemical biology approaches.
EXPERIMENTAL SECTION
Small-molecule inhibitors.
The inhibitors PCI-34051 and Quisinostat (QSN) were purchased from Selleck Chemicals (USA). The
NCC-149 inhibitor was purchased from the Tokyo Chemical Company (Japan). The smHDAC8-
selective inhibitors were synthesized and characterized as described previously29, 56 unless stated below.
Compound 6 (3-(benzthiophene-7-carboxamido)-4-chlorobenzohydroxamate) was synthesized as
follows (Figure S12A). Procedure. (a) Benzthiophene-7-carboxylic acid (1.4 mmol) was cooled to 0°C
2
0
ACS Paragon Plus Environment

Page 21 of 57
1
Journal of Medicinal Chemistry
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
and then thionyl chloride (3 mmol) was added dropwise. The mixture was heated under reflux for 30
minutes. After evaporating the excess of thionyl chloride under vacuum, the obtained acid chloride was
dissolved in dry THF (50 ml) and added to a solution of methyl 3-amino-4-chlorobenzoate (1.4 mmol)
and DIPEA (3 mmol) in THF. The reaction was monitored by TLC. Subsequently, the solvent was
evaporated under vacuum and the mixture was dissolved in ethyl acetate (50 ml) and washed with aq.
sodium hydroxide solution. The organic layer was evaporated under reduced pressure and the product
was purified by column chromatography (chloroform/ methanol, 99:1). (Yield: 88.8%). (b) The obtained
amide was dissolved in methanol (25 ml) and 1 M aq. sodium hydroxide solution (10 ml) and heated to
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0°C for 2 h. The reaction was monitored by TLC. Subsequently, the solvent was evaporated under
reduced pressure and the product was dissolved in ethyl acetate and washed with 1 M HCl solution, the
organic layer was finally evaporated under reduced pressure. (Yield: 85.0%). (c) The substituted benzoic
acid derivative (1 mmol) was dissolved in dry THF (50 ml) and PyBOP (1.2 mmol) was added. To the
activated acid, a mixture of NH OTHP (1.5 mmol) and DIPEA (2.5 mmol) in dry THF (5 ml) was added,
2
and the reaction mixture was stirred overnight at room temperature. The solvent was evaporated under
reduced pressure and the mixture was dissolved in ethyl acetate (50 ml) and washed with aq. potassium
hydrogen carbonate solution and brine. The organic layer was evaporated under reduced pressure and
the product was purified by column chromatography (chloroform/methanol/TEA, 99.5:0.45:0.05). The
obtained product was dissolved in THF and a catalytic amount of diluted HCl was added and was stirred
at room temperature. The reaction was controlled by TLC. After that, the solvent was evaporated under
reduced pressure and the mixture was dissolved in ethyl acetate (50 ml) and washed with brine. The
organic layer was evaporated under reduced pressure and the product was purified by column
chromatography (chloroform/methanol/formic acid, 95:4.95:0.05). (Yield: 35%). Analytical data. MS
3
5
37
- 1
m/z: 345.23 (Cl )| 347.24 (Cl )[M-H] . H NMR (400 MHz, DMSO-d ) δ 11.37 (s, 1H), 10.42 (s, 1H),
6
2
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 57
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
9.14 (s, 1H), 8.22 (d, J = 7.3 Hz, 1H), 8.15 (d, J = 7.7 Hz, 1H), 8.02 (s, 1H), 7.85 (d, J = 5.2 Hz, 1H),
.74 – 7.63 (m, 2H), 7.59 (t, J = 7.5 Hz, 1H), 7.53 (d, J = 5.2 Hz, 1H). HRMS m/z: 369.0072 [M+Na]+;
7
+
calculated: C H N O ClSNa 369.0071. HPLC: rt 10.68 min (99.65%). Yield: 120 mg; 0.35 mmol;
1
6
11
2
3
25%.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Compound 12 (1-[5-chloro-2-(4-fluorophenoxy)phenyl]-N-hydroxy-1H-1,2,3-triazole-4-carboxamide)
was synthetized as follows (Figure S12B). Procedure: (a) 2-Amino-4-chlorophenol (7.0 mmol) was
dissolved in 1 M HCl (20 mL). At -5 °C, a solution of sodium nitrite (8.4 mmol) in water (2 mL) was
added dropwise over a period of 5 min. After stirring additional 5 min, urea (50 mg) was added. Then
the mixture was added to a cold solution of sodium azide (14 mmol) and sodium acetate (0.06 mmol) in
water (10 mL). The resulting mixture was stirred for 2 h at -5 °C. Then it was extracted with diethyl ether
(
3×). The combined organic layers were dried (Na SO ), filtered, and the solvent was removed under
2 4
reduced pressure. The residue was purified by flash column chromatography (cyclohexane/ethyl acetate
10/1). (Yield: 86%). (b) The obtained azide (3.0 mmol) was dissolved in a 1:1 mixture of water and
=
tert-butyl alcohol (15 mL). Methyl propiolate (4.0 mmol), sodium ascorbate (0.20 mmol), and copper(II)
sulfate pentahydrate (0.04 mmol) were added and the mixture was stirred for 18 h at room temperature.
Then water was added and the mixture was extracted with ethyl acetate (3×). The combined organic
layers were dried (Na SO ), filtered, and the solvent was removed under reduced pressure. The residue
2
4
was purified by flash column chromatography (cyclohexane/ethyl acetate = 2/1). (Yield: 89%). (c) A 100
mL round-bottom flask was charged with the obtained triazole (0.59 mmol), Cu(OAc) (0.59 mmol), 4-
2
fluorophenylboronic acid (0.71 mmol), and powdered 4 Å molecular sieves. Then dichloromethane (4.5
mL) was added. After the addition of triethylamine (2.9 mmol), the reaction mixture was stirred at
ambient temperature overnight. Then the suspension was filtered and the filtrate diluted with water and
extracted with ethyl acetate (3×). The combined organic layers were dried (Na SO ), filtered, and the
2
4
2
2
ACS Paragon Plus Environment

Page 23 of 57
1
Journal of Medicinal Chemistry
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
solvent was removed under reduced pressure. The residue was purified by flash column chromatography
(cyclohexane/ethyl acetate = 9/1 → 3/1). (Yield: 13%). (d) A 5.4 M solution of sodium methoxide in
methanol (1.1 mmol) was added to a solution of the obtained diphenyl ether (0.11 mmol) and
hydroxylamine hydrochloride (0.55 mmol) in dry methanol (3 mL). The mixture was stirred at ambient
temperature overnight. Then the solvent was removed under reduced pressure and the residue was
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
purified by automatic flash column chromatography using a Biotage purification apparatus (5% → 50%
®
ACN in H O, Biotage SNAP KP-C18-HS 12 g). Fractions containing the desired product were
2
combined, dried from acetonitrile under reduced pressure and then subjected to lyophilization. (Yield:
9
5%).
1
Analytical data: H NMR (DMSO-d ): δ [ppm] = 7.02 – 7.06 (m, 1H), 7.15 – 7.20 (m, 2H), 7.22 – 7.28
6
1
3
(
m, 2H), 7.53 – 7.57 (m, 1H), 7.87 – 7.90 (m, 1H), 8.31 (s, 1H); C NMR (DMSO-d ): δ [ppm] = 116.9
6
(d, J = 23.7 Hz, 2C), 120.4 (1C), 121.1 (d, J = 8.7 Hz, 2C), 123.1 (1C), 125.7 (1C), 127.4 (1C), 128.8
(1C), 130.2 (1C), 147.1 (1C), 148.4 (1C), 151.3 (d, J = 2.4 Hz, 1C), 158.3 (1C), 158.8 (d, J = 241 Hz,
+
1
C); HRMS (m/z): [M+H] calculated for C H ClFN O : 349.0498, found: 349.0522; HPLC: t = 16.0
1
5
11
4
3
R
min, purity 95.7%.
Purity of chemicals.
The purity of all compounds used was assessed by HPLC. All synthesized compounds have a purity
higher than 95%. The purity for the PCI-34051 and Quisinostat (QSN) compounds purchased from
Selleck Chemicals is 99.01% and 99.62%, respectively, as stated by the manufacturer (see HPLC data in
the
following
webpages:
http://www.selleckchem.com/products/pci-34051.html
and
http://www.selleckchem.com/products/JNJ-26481585.html). The purity for the NCC-149 compound
purchased from the Tokyo Chemical Company is 96.00% as stated by the manufacturer (see
2
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 57
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
http://www.tcichemicals.com/eshop/en/sg/commodity/H1340/). The HPLC traces for all other
compounds are provided in Figure S13A-L).
Cloning, expression and purification of HDAC8 proteins.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The full-length cDNA constructs (WT and mutants) for hHDAC8 and smHDAC8 were amplified by
polymerase-chain reaction (PCR) and cloned into bacterial expression vectors. The hHDAC8 (WT and
mutant) gene was inserted between the NdeI and BamHI restriction sites of the pnEA-3HT expression
6
1
vector , where it is in frame with a sequence coding for a N-terminal poly-histidine affinity purification
tag followed by thioredoxin and a protease 3C cleavage site. The smHDAC8 (WT and mutant) gene was
6
1
cloned between the NdeI and BamHI restriction sites of the pnEA-tH expression vector and is in frame
with a sequence coding for a C-terminal thrombin cleavage site followed by a poly-histidine affinity
purification tag.
For wild-type smHDAC8 overproduction, a modified protocol was used compared to that published
previously. Expression was carried out in BL21(DE3) cells in 2xLB medium. Cultures were grown and
induced at 37°C with 0.7 mM IPTG in the presence of 100 μM ZnCl . After overnight incubation at
2
3
7°C, cells were harvested and resuspended in lysis buffer A (150 mM NaCl, 50 mM Tris pH 8.0). Lysis
was done by sonication, the lysate was clarified by centrifugation. The supernatant was loaded onto Talon
Superflow Metal Affinity Resin (Clontech) pre-equilibrated with the lysis buffer A. The his-tagged
protein was released from the Talon resin by thrombin protease treatment in buffer B (50 mM KCl, 10
mM Tris pH 8.0) and subsequently loaded onto a 16/60 Superdex 200 gel filtration column (GE
Healthcare) pre-equilibrated with buffer C (50 mM KCl, 10 mM Tris-HCl pH 8.0 and 2 mM DTT). Peak
fractions were concentrated with an Amicon Ultra centrifugal filter unit. This protocol yielded three times
more protein for the wild type enzyme. smHDAC8 mutants could not be purified using this protocol and
their overproduction was carried out as described initially30, 62
.
2
4
ACS Paragon Plus Environment

Page 25 of 57
1
Journal of Medicinal Chemistry
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
hHDAC8 overproduction was carried out in Escherichia coli BL21(DE3) cells in 2xLB medium. Culture
induction was done at 23°C by adding 0.5 mM final isopropyl-1-thio-β-D-galactopyranoside (IPTG,
Euromedex), in the presence of 100 μM ZnCl . Harvested bacteria were re-suspended in lysis buffer (50
2
mM KCl, 10 mM Tris-HCl pH 8.0) and lysed by sonication. The lysate was clarified by centrifugation
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(17,500 rpm, 50 min, 4°C, Sorvall Lynx 6000 Thermo Scientific). The supernatant was loaded onto Talon
Superflow Metal Affinity Resin (Clontech) pre-equilibrated with lysis buffer. The his-thioredoxin-tagged
protein was released from the Talon resin by 3C protease treatment and subsequently loaded onto a 16/60
Superdex 200 gel filtration column (GE Healthcare) pre-equilibrated with the purification buffer (50 mM
KCl, 10 mM Tris-HCl pH 8.0, and 0.5 mM TCEP). The recombinant protein was concentrated with an
Amicon Ultra centrifugal filter unit.
Mutagenesis experiments.
The mutant constructs were generated using standard PCR-based nested protocols and inserted into the
corresponding expression vectors. The L1 and L6 loops exchange mutants were designed based on
structural comparison. The hHDAC1 L6 sequence S LSGDRLGC was introduced instead of the
2
65
T IAGDPMCS sequence in hHDAC8, to create the hHDAC8-mL6 mutant. In the second step, the
2
68
hHDAC1 L1 sequence Y YGQGHPMK was introduced instead of the L AKI sequence in hHDAC8-
2
3
31
mL6, to create the double hHDAC8-mL1/mL6 mutant. Finally, the triple mutant hHDAC8-
mL1/mL6/L179I was generated by the introduction of an isoleucine residue in the L179 position, using
the double hHDAC8-mL1/mL6 mutant as a template. smHDAC8 mutants were produced as described
for hHDAC8 mutants. Sequences replaced in smHDAC8 were the same as the ones replaced in hHDAC8.
Crystallization and X-ray data collection.
2
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 57
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Diffraction-quality crystals of the native smHDAC8 enzyme were obtained at 17°C after 3 days by
mixing equal volumes of smHDAC8 (2.5 mg/ml) with reservoir solution composed of 21% PEG 3350
+
+
(
Fluka) and 0.05 M Na /K L-tartrate, and crystallized using the hanging-drop vapor diffusion technique.
After 3 days, grown crystals were soaked for 20 hours in mother liquor supplemented with the
corresponding inhibitor (10 mM final concentration of the inhibitor, preparation from a 100 mM stock
in N,N-dimethylformamide (DMF) or DMSO). Crystals used for X-ray data collection were briefly
transferred in reservoir solution supplemented with 22% glycerol and flash-frozen in liquid nitrogen.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Co-crystallization of hHDAC8-mL6 together with QSN (Quisinostat) inhibitor was performed using the
hanging-drop vapor diffusion technique. The hHDAC8-mL6/QSN complex was formed by incubating
the hHDAC8-mL6 mutant protein (5 mg/mL) with QSN (5 mM resuspended in DMF) at 4°C for 1 h.
Diffraction-quality crystals were obtained at 20°C after 3–4 days by mixing equal volumes of the
hHDAC8-mL6/QSN complex with reservoir solution composed of 20% polyethylene glycol 3350
(
Fluka), 0.2 M KNO and 0.1 M Bis-tris propane pH 7.5. Crystals used for X-ray data collection were
3
briefly transferred in reservoir solution supplemented with 22% glycerol and flash-frozen in liquid
nitrogen. Crystallographic data obtained in this project were collected at 100 K on SOLEIL beamline
PROXIMA1, ESRF beamlines ID30b, ID29 and ID23 and SLS PX beamlines.
Structure determination, model building and refinement.
The crystallographic data were processed and scaled using HKL200063 or XDS . Phases for
smHDAC8/inhibitor complexes were obtained by molecular replacement followed by rigid body
refinement against smHDAC8 native structure as a model (4BZ5). The initial models were refined
64
6
5
66
through several cycles of manual building using Coot and automated refinement with Phenix . The
structure of hHDAC8-mL6 complexed with QSN was solved by molecular replacement with Phenix66
using the hHDAC8 structure (PDB 1T67) as a search model. The final models were validated using tools
2
6
ACS Paragon Plus Environment

Page 27 of 57
1
Journal of Medicinal Chemistry
6
5
67
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
provided in Coot and Molprobity . Visualization of structural data was done with Pymol (The PyMOL
Molecular Graphics System, Version 2.0 Schrödinger, LLC), and two-dimensional diagrams
summarizing molecular interaction between inhibitors and HDAC8 enzymes were prepared with the help
6
8
of the LigPlot program . Atomic coordinates and structure factors of the smHDAC8 and hHDAC8-mL6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
complexes were deposited in the Protein Data Bank under the PDB codes 6HQY, 6HRQ, 6HSH, 6HSF,
6
HSG, 6HSZ, 6HT8, 6HTG, 6HTH, 6HTI, 6HTT, 6HTZ, 6HU0, 6HU1, 6HU2, 6HU3, 6HSK.
HDAC activity and inhibition assays.
Catalytic activity and inhibition assays of smHDAC8 and hHDAC8 were performed as described
3
0
earlier . Briefly, the hHDAC8 and smHDAC8 activity testing was carried out with the HDAC8
Fluorimetric Drug Discovery Kit (Fluor de Lys(R)- HDAC8, BML-KI178) from Enzo Life Sciences,
according to the manufacturer’s instructions with a substrate concentration of 50 µM. Fluorescence was
measured in a plate reader (BMG Polarstar) with excitation at λ =390 nm and emission at λ = 460 nm.
IC values were determined with OriginPro (version 9.0.0, Northampton, Massachusetts).
5
0
Studies on Schistosoma mansoni in culture.
The effects of PCI-34051 and NCC-149 compounds on the viability of S. mansoni schistosomula
6
9
(
Puerto-Rican strain) were tested using a microscopy-based assay, as described previously . In brief,
7
0
schistosomula (2,000 per well), prepared by standard mechanical transformation from cercaria , were
maintained in 6-well plates in M199 medium kept at pH 7.4 with 10 mM HEPES and supplemented at
3
7°C in a humid atmosphere containing 5% CO2. Three different concentrations of inhibitors (25, 50
and 100 µM) were tested, the inhibitors were dissolved in DMSO, and the culture medium was refreshed
each day. The assessment of parasite mortality was carried out after microscopic examination, based on
three criteria: a granular appearance, tegumental defects and the absence of motility. At least 300
2
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 57
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
schistosomula were observed at each time point for each condition and results were expressed as a
percentage of viable larvae remaining. Three biological replicates (different batches of larvae) were
examined in duplicate for each condition.
7
0
The stability of adult worm pairs and egg production were assayed as previously described . Adult worm
pairs were obtained from infected golden hamsters (Mesocricetus auratus) by perfusion, washed in M199
medium and ten pairs placed in 2 mL of M199 buffered complete medium (as for schistosomula above)
in each well of a 6-well culture plate. Worms were maintained in culture for 5 days at 37°C (humid
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
atmosphere, 5% CO ) before the addition of HDAC8-selective inhibitors, the application of DMSO alone
2
served as a negative control experiment. Both the culture medium and the inhibitors were refreshed daily.
The number of couples remaining as pairs was determined daily by microscopy and the medium of each
well containing eggs laid by the couples was recovered and centrifuged to allow eggs to be counted under
the microscope. Three biological replicate experiments were performed in triplicate.
TUNEL assay.
Detection of DNA strand breaks in inhibitor-treated S. mansoni schistosomula was done using the
terminal deoxynucleotidyl transferase dUTP nick-end labelling (TUNEL) method using the In Situ Cell
Death Detection Kit TMR Red (Roche). The method designed for cell suspensions was followed as
described in the manufacturer's instructions with few modifications. Briefly, 2,000 schistosomula were
treated with 100 µM PCI-34051 or NCC-149 for 96 h, in 6-well plates containing 2 mL of complete
medium. The treatment with DMSO alone served as negative control experiment. After 96 h incubation,
culture media were removed and the schistosomula were centrifuged (1,000 rpm, 2 min), washed three
times in PBS buffer, and then fixed in 2% formaldehyde for 60 min. Schistosomula were afterwards
washed once more in PBS and permeabilization solution (0.1% Triton X-100, 0.1% sodium citrate) was
added for 10 min on ice. Labeling of schistosomula with 4′,6-diamidino-2-phenylindole (DAPI) and
2
8
ACS Paragon Plus Environment

Page 29 of 57
1
Journal of Medicinal Chemistry
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
TMR Red dUTP was performed according to the manufacturer's instructions and TUNEL-positive
parasites were observed by fluorescence using an AxioImager Z1-Apotome microscope (Zeiss).
Isothermal titration calorimetry (ITC).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ITC experiments were done at 25°C using a PEAQ microcalorimeter (Malvern Instruments). All protein
samples were purified in the same ITC buffer (50 mM KCl, 10 mM Tris pH=8.0 and 0.5 mM TCEP). In
a typical experiment, aliquots of 2.0 µl of HDAC8 protein sample at 200 µM were injected into an
inhibitor solution at 20 µM placed in the ITC cell. Blank experiments were used to retrieve signal due to
solvent (DMSO or DMF) dilution into ITC buffer. Data were analysed with PEAQ-ITC Analysis
Software (Malvern Instruments) and with Affinimeter (Software 4 Science Developments).
Differential scanning fluorimetry.
Thermal stability of HDAC8 proteins was analysed by a label-free differential scanning fluorimetry
(DSF) approach using a Prometheus NT.48 instument (NanoTemper Technologies). Briefly, the shift of
intrinsic tryptophan fluorescence of HDAC8 proteins upon gradual temperature-triggered unfolding
(temperature gradient 20 to 95°C) was monitored by detecting the emission fluorescence at 330 and 350
nm. The measurements were carried out nanoDSF-grade high sensitivity glass capillaries (NanoTemper
Technologies) at a heating rate of 1°C/min. Protein melting points (Tm) were inferred from the first
derivative of the ratio of tryptophan emission intensities at 330 and 350 nm. Finally, the ∆Tm value of
an HDAC8 protein for a particular inhibitor was calculated as the difference between the Tm values of
the inhibitor-bound and inhibitor-free proteins. All the assays were done in triplicate.
Docking experiments.
Crystal structures of hHDAC1-3, hHDAC8 and hHDAC6-CD2 for docking as well as zHDAC10 for
homology modelling of hHDAC10 were downloaded from Protein Data Bank (corresponding PDB IDs
2
9
ACS Paragon Plus Environment