Discovery of potent HDAC inhibitors based on chlamydocin with inhibitory effects on cell migration

Article abstract of DOI:10.1002/cmdc.201300372

The histone deacetylase (HDAC) family is a promising drug target class owing to the importance of these enzymes in a variety of cellular processes. Docking studies were conducted to identify novel HDAC inhibitors. Subtle modifications in the recognition domain were introduced into a series of chlamydocin analogues, and the resulting scaffolds were combined with various zinc binding domains. Remarkably, cyclo(L-Asu(NHOH)-L-A3mc6c-L-Phe-D-Pro, compound 1 b), with a methyl group at positionsa 3 or 5 on the aliphatic ring, exhibited better antiproliferative effects than trichostatina A (TSA) against MCF-7 and K562 cell lines. In addition to cell-cycle arrest and apoptosis, cell migration inhibition was observed in cells treated with compound 1 b. Subsequent western blot analysis revealed that the balance between matrix metalloproteinasea 2 (MMP2) and tissue inhibitors of metalloproteinasea 1 (TIMP1) determines the degree of metalloproteinase activity in MCF-7 cells, thereby regulating cell migration. The improved inhibitory activity imparted by altering the hydrophobic substitution pattern at the bulky cap group is a valuable approach in the development of novel HDAC inhibitors.

Full text of DOI:10.1002/cmdc.201300372

CHEMMEDCHEM
FULL PAPERS
DOI: 10.1002/cmdc.201300372
Discovery of Potent HDAC Inhibitors Based on
Chlamydocin with Inhibitory Effects on Cell Migration
Shimiao Wang,[a] Xiaohui Li,*[a] Yingdong Wei,[a] Zhilong Xiu,[a] and Norikazu Nishino[b]
The histone deacetylase (HDAC) family is a promising drug
target class owing to the importance of these enzymes in a va-
riety of cellular processes. Docking studies were conducted to
identify novel HDAC inhibitors. Subtle modifications in the rec-
ognition domain were introduced into a series of chlamydocin
analogues, and the resulting scaffolds were combined with
various zinc binding domains. Remarkably, cyclo(l-Asu(NHOH)-
l-A3mc6c-l-Phe-d-Pro, compound 1b), with a methyl group at
positions 3 or 5 on the aliphatic ring, exhibited better antiproli-
ferative effects than trichostatin A (TSA) against MCF-7 and
K562 cell lines. In addition to cell-cycle arrest and apoptosis,
cell migration inhibition was observed in cells treated with
compound 1b. Subsequent western blot analysis revealed that
the balance between matrix metalloproteinase 2 (MMP2) and
tissue inhibitors of metalloproteinase 1 (TIMP1) determines the
degree of metalloproteinase activity in MCF-7 cells, thereby
regulating cell migration. The improved inhibitory activity im-
parted by altering the hydrophobic substitution pattern at the
bulky cap group is a valuable approach in the development of
novel HDAC inhibitors.
Introduction
Posttranslational modifications such as the acetylation and
methylation of histones are implicated in various cellular pro-
cesses, including chromatin remodeling, expression silencing,
and DNA repair.[1,2] Aberrant histone deacetylation has been
shown to be associated with tumorigenesis and tumor devel-
opment. The dynamics of acetylation at histone N-terminal
tails (such as H3K4 and H4K16) is mediated by histone acetyl-
transferases (HATs) and histone deacetylases (HDACs).[3] Eight-
een human HDAC members have been identified so far, and
these are divided into four classes based on size, cellular locali-
zation, number of catalytic sites, and homology to yeast pro-
teins. Class I (HDACs 1, 2, 3, and 8), class IIa (HDACs 4, 5, 7, and
9), class IIb (HDACs 6 and 10), and class IV (HDAC11) HDACs
are all zinc-dependent enzymes, whereas class III HDACs are
strictly dependent on NAD+ for their activities. Further knock-
out analyses of various class I and II HDAC proteins have indi-
cated that class I HDACs play a key role in cell survival and pro-
liferation, and class II HDACs may have tissue-specific func-
tions.[4]
in cells treated with HDAC inhibitors.[7] Furthermore, HDAC6,
a unique HDAC containing two intact catalytic domains, has al-
ready been identified as a key regulator of cell migration.[8] In
addition, there is increasing evidence to suggest that HDAC4
plays a significant role in cell migration by regulating the ex-
pression of matrix metalloproteinase (MMP) family mem-
bers.[9,10]
There are typically three functional domains in the structure
of HDAC inhibitors: a zinc binding domain (ZBD, usually a hy-
droxamate or thiol moiety), a surface recognition domain (SRD,
generally an aromatic moiety), and a linker domain (typically
an aliphatic chain). Based on the structure of natural and syn-
thetic compounds such as trichostatin A (TSA), vorinostat (sub-
eroylanilide hydroxamic acid, SAHA), chlamydocin, and FK228
(Figure 1), numerous HDAC inhibitors have been developed as
potential anticancer agents. There are currently several HDAC
inhibitors undergoing clinical trials.[11,12] Two of them, SAHA
and FK228, have already been approved by the US Food and
Drug Administration (FDA) for the treatment of cutaneous T-
cell lymphoma (CTCL).
To date, thousands of non-histone proteins have been iden-
tified as HDAC substrates.[5,6] As HDACs are well-validated anti-
cancer targets, it is well known that HDAC inhibitors mediate
cell death via multiple pathways. Cell-cycle arrest, apoptosis,
differentiation, and anti-angiogenic effects have been observed
In our previous work, a novel chlamydocin–hydroxamic acid
analogue, cyclo(l-Asu(OBn)-Aib-l-Phe-d-Pro) (Ky-2), was identi-
fied as a potential anticancer agent and was shown to induce
apoptosis in myeloma cells.[13] Therefore, docking studies were
initially carried out to analyze the binding mode of HDAC in-
hibitors to HDAC. Consistent with the previous study, the inter-
actions between Ky-2 and HDAC2 further confirmed that the
changes brought about in the aminoisobutyric acid (Aib) site
or to the aromatic ring of l-Phe would affect HDAC inhibitory
activity.[14,15] In this study, unnatural amino acids 1-amino-n-
methyl-1-cyclohexanecarboxylic acid (Anmc6c), l-2-amino-
chloro-n-aliphatic acid (l-Acn), or n-chlorophenylalanine (l-Phe-
(nCl)) were used to substitute the corresponding residues Aib
[a] S. Wang, Dr. X. Li, Y. Wei, Z. Xiu
School of Life Science and Biotechnology
Dalian University of Technology, 2 Linggong Road, 116024 Dalian (China)
E-mail: lxhxh@dlut.edu.cn
[b] Dr. N. Nishino
Graduate School of Life Science and Systems Engineering
Kyushu Institute of Technology, 808-0196 Kitakyushu (Japan)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201300372.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 627 – 637 627

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Figure 1. Classic HDAC inhibitors and target compounds.
(marked as AA2 according to its position in Ky-2) or l-Phe
(marked as AA3), and the resulting scaffolds were combined
with either hydroxamate or a thioacetyl group (Figure 1). The
inhibitory activities against HDAC isoforms and antiproliferative
activities of these compounds were then evaluated. Further-
more, key regulators that mediate cell cycle and cell migration
were explored to determine the underlying mechanisms of
HDAC-inhibitor-induced cell death.
Results and Discussion
Molecular design
Figure 2. Binding mode of Ky-2 (stick model) in HDAC2 as predicted by
docking; HDAC2 residues that interact with Ky-2 are indicated.
It is well known that HDACs, especially the class I subfamily
(HDACs 1, 2, and 3), are generally expressed in nearly all
cancer tissue types. HDAC1 and HDAC2 are highly similar, shar-
ing an overall sequence identity of ~82%, suggesting either of
them is qualified to represent the class I enzymes.[16,17] There-
fore, to explore the binding mode of the chlamydocin–hy-
droxamic acid analogue to HDAC, a docking study was per-
formed with the HDAC2–Ky-2 complex. In the docking of Ky-2,
the compound binds very well to the active site of HDAC2: the
aliphatic chain of Ky-2 occupies the long and narrow channel
of HDAC2, the Aib and l-Phe residues from the bulky SRD in-
teract with the residues at the entrance of the active site,
while the hydroxamic acid group chelates to the zinc ion at
the bottom of the catalytic pocket (Figure 2). In HDAC2–Ky-2
complex, the aromatic ring of the phenyl group is oriented
into the groove formed by Tyr209 and Phe210, thus contribu-
ting to the stabilization of the enzyme–inhibitor complex.
Hydrogen bonds and hydrophobic contacts between HDAC2
and Ky-2 were then analyzed (Figure 3). In the HDAC2–Ky-2
complex, the oxygen atom from the hydroxamate group of Ky-
2 establishes a hydrogen bond with the Ne2 atom of His145,
while the nitrogen atom of Ky-2 establishes hydrogen bonds
with Ne2 of His145, Ne2 of His146, and Od2 of Asp181. Be-
cause the hydrogen bonds formed between the hydroxylamine
group in the metal binding domain and the corresponding res-
idues around the active pocket of HDAC2 were expected to be
crucial for HDAC inhibitory activity, the hydroxamate moiety
was maintained in the newly designed compounds 1a–
e (Figure 1).
Previous studies have demonstrated that thiols are potent
zinc binding groups (ZBGs).[18] Docking studies further revealed
that the sulfur atom of the thiol group can effectively interact
with His183 and Tyr308 of HDAC2 through hydrogen bonds.[19]
In the meantime, evidence emerged to suggest that thioacetyl
derivatives might be as effective as thiol derivatives.[20] In con-
trast to disulfides, mass spectrometry revealed that thioacetyl
groups remain intact after inhibition.[20] Therefore, to develop
potent non-hydroxamate HDAC inhibitors, a thioacetyl group
was introduced into desired compounds 2a–c (Figure 1).
The hydrophobic interactions between Aib and l-Phe resi-
dues of Ky-2 and His183, Glu208, Tyr209, Phe210, and Gly306
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 627 – 637 628

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
into the active pocket of HDAC2 (Supporting Information fig-
ure S1). All target compound–HDAC2 complexes were predict-
ed to be stable, as their binding energies were similar to that
of the reference compound Ky-2 (Supporting Information
table S1). In particular, compounds 1a–c and 2b,c showed sig-
nificantly higher HDAC2 binding potential than Ky-2, and
therefore were expected to have improved inhibitory activity
over Ky-2.
Chemistry
All target compounds were prepared by solution-phase pep-
tide synthesis as illustrated in Scheme 1. Briefly, C-terminal-pro-
tected proline (NH2-d-Pro(OtBu)) was coupled with Z-l-AA3-OH
to give the dipeptide Z-l-AA3-d-Pro(OtBu). After removal of the
Z protecting group via catalytic hydrogenation, the same cou-
pling procedure was used to obtain the tripeptide Z-AA2-l-
AA3-d-Pro(OtBu) and the tetrapeptide Boc-l-AA1-AA2-l-AA3-d-
Pro(OtBu). The corresponding N-terminal (Boc) and C-terminal
(OtBu) protecting groups of the linear tetrapeptide were sub-
sequently removed in the presence of trifluoroacetic acid
(TFA), and then cyclized in N,N-dimethylformamide (DMF)
using
O-(7-azabenzotriazol-1-yl)-1,1,3,3-tetramethyluronium
hexafluorophosphate (HATU) and N,N-diisopropylethylamine
(DIEA) as coupling reagents to give cyclo(l-AA1-AA2-l-AA3-d-
Pro(OtBu).
To obtain the hydroxamic acid derivatives, the OBn protect-
ing group of 3a–e was removed by catalytic hydrogenation,
and the hydrogenated residues were subsequently treated
with hydroxylamine hydrochloride (HCl·NH2OBn) in DMF to
give 5a–e. The OBn-protected hydrochlorides 5a–e were final-
ly transformed by catalytic hydrogenation into the target com-
pounds 1a–e. For the preparation of non-hydroxamate ZBD,
potassium thioacetate was added to 4a–c with stirring to give
the desired thioacetyl derivatives 2a–c. All target compounds
1
were characterized by H NMR and HR-FAB-MS. The purity of
the final products was confirmed by HPLC analysis, and all
Figure 3. Hydrogen bonds and hydrophobic contacts between HDAC2 and
Ky-2. Distances between hydrogen bond donors and acceptors less than
3.5 ꢁ are represented as dashed lines.
compounds showed purity >97%.
Inhibitory activity against HDAC isoforms
residues of HDAC2 are favorable to the stabilization of the
enzyme–inhibitor complex (Figure 3). These findings suggest
that further modification of the SRD to affect hydrophobic
properties may ultimately alter HDAC inhibitory activity.[21]
Therefore, an aliphatic ring, aliphatic chain, or further modifica-
tions to the aromatic ring of l-Phe were introduced into our
desired compounds to alter surface hydrophobicity. Specifical-
ly, Anmc6c was used to replace Aib to obtain compounds 1a–
c, l-Acn was used to substitute l-Phe for compounds 1d,e,
and l-Phe(nCl) groups were used to replace l-Phe to obtain
compounds 2a–c (Figure 1).
The inhibitory activities of all target compounds against class I
HDAC1, class IIa HDAC4, and class IIb HDAC6 were measured
by using whole-cell extracts prepared from transfected 293T
cells. TSA and Ky-2 were used as positive controls. In general,
all target compounds exhibited potent HDAC1 inhibitory activ-
ities, with IC50 values in the nanomolar range (Table 1). Hy-
droxamic acid derivatives 1a–e showed better HDAC inhibitory
activities than those of the thioacetyl derivatives 2a–c. Com-
pared with TSA and Ky-2, the hydroxamic acid derivatives 1a–
e showed increased inhibitory activities against HDACs 1 and
4. Although their HDAC inhibitory activities varied among
HDACs 1, 4, and 6, compounds 1a–e displayed pan-inhibitory
activity similar to SAHA and TSA against class I and II
HDACs.[7,22] In particular, compound 1a, with respective IC50
values of 10 and 6.4 nm against HDACs 1 and 4, showed
To determine whether all the designed compounds can bind
to the active pocket of HDACs and thus have HDAC inhibitory
activity, docking studies were conducted before their synthesis.
Similar to Ky-2, all target compounds were successfully docked
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 627 – 637 629

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Scheme 1. Synthesis of target compounds. Reagents and conditions: a) HOBt/DCC in DMF, RT, overnight; b) Pd/C, H2 in AcOH, RT, overnight; c) TFA, 08C, 2 h;
d) HATU/DIEA in DMF, RT, 1.5 h; e) Pd/C, H2 in MeOH, RT, overnight; f) HCl·NH2OBn, DCC/HOBt in DMF, 08C, 2 h; g) Pd/BaSO4, H2, RT, overnight; h) KSCOCH3 in
DMF, RT, 4 h.
Table 1. HDAC inhibitory, p21 promoter, and antiproliferative activities of target compounds.
activity against HDAC6 than TSA,
their IC50 values toward HDAC6
were slightly higher than that of
Ky-2. The unique tandem-ar-
ranged catalytic pockets of
HDAC6 provide an explanation
for the observed selectivity, sug-
gesting that small molecules
such as TSA might be more fa-
vorable to the HDAC6–inhibitor
interaction.[25,26] Furthermore, the
aliphatic ring from hydroxamic
acid derivatives 1a–c and the ali-
phatic ring from 1d,e are more
hydrophobic than the corre-
sponding moieties in the precur-
sor compound Ky-2 (Supporting
Compd
IC50 [nm]
p21 EC1000 [nm][a]
Antiprolif. IC50 [nm][b]
HDAC1
HDAC4
HDAC6
MCF-7
HeLa
K562
TSA
Ky-2
1a
1b
1c
1d
1e
2a
2b
2c
23
18
10
15
11
14
10
27
28
38
34
17
6
14
12
19
15
ND
ND
ND
65
230
130
160
170
20.0ꢀ2.7
18.0ꢀ0.54
8.0ꢀ0.29
2.9ꢀ0.93
4.3ꢀ1.5
210.0ꢀ48
36.0ꢀ2.6
ND
141ꢀ4
ND[c]
39ꢀ7
ND
309ꢀ6
48ꢀ1
232ꢀ7
ND
560ꢀ55
ND
876ꢀ182
104ꢀ14
863ꢀ87
ND
31ꢀ3
26ꢀ1
42ꢀ4
110
96
>1000
>1000
>1000
>1000
ND
ND
>1000
380
>1000
>1000
480ꢀ10
630ꢀ12
>5000
>5000
>5000
ND
ND
820ꢀ27
[a] Compound concentration at which the induced luciferase activity is 10-fold higher than the basal level.
[b] Cells were treated for 48 h. [c] ND: no data. All data represent the mean ꢀSD of at least three independent
experiments.
a three- to fivefold increase in enzyme inhibitory activity rela-
tive to reference compounds, respectively.
Information table S1), thus leading to better HDAC6 inhibitory
activity.
The hydroxamic acid derivatives with variations at the ali-
phatic ring (compounds 1a–c) showed no significant differen-
ces in HDAC1, 4, and 6 inhibitory activities. Likewise, the thioa-
cetyl derivatives bearing variations at the aromatic ring (2a–c)
showed similar activity in HDAC1 inhibitory activities. Dramatic
differences were observed, however, among the thioacetyl de-
rivatives 2a–c in HDAC6 inhibitory activities. Compound 2b,
bearing a meta-chlorine on the aromatic ring, exhibited higher
HDAC6 inhibitory activity than the other two, suggesting
modifications at the meta position as suitable for improving in-
hibitory activity. Substitution of the chlorine atom on the aro-
matic ring for phenyl or furyl groups is expected to produce
even more potent HDAC6-selective inhibitors.[23,24]
p21 promoter activation
All the hydroxamic acid derivatives 1a–e including reference
compounds TSA and Ky-2 were tested for their ability to acti-
vate the p21 promoter. All tested compounds effectively acti-
vated p21 gene expression, with EC1000 values in the nanomo-
lar range (Table 1). The p21 promoter activation efficiency is
ranked in the order: 1b > 1c > 1a >Ky-2 > TSA > 1e > 1d.
Notably, compound 1b exhibited the highest activation effi-
ciency, giving an EC1000 value of 2.9 nm, which is at least sixfold
lower than that of TSA and Ky-2. Relative to their HDAC inhibi-
tory activities, compounds 1d,e showed unexpectedly low acti-
vation efficiency, which is presumably due to an excess in tor-
All tested compounds showed selectivity for HDAC1 over
HDAC6. Although compounds 1a–e displayed lower inhibitory
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 627 – 637 630

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
sional energy that must be provided to stabilize HDAC–1d/1e
complex (Supporting Information table S1).
Antiproliferative activity against cancer cell lines
Based on HDAC inhibitory activity and p21 promoter activation
efficiency, MTT assays were performed for the hydroxamic acid
derivatives 1a–c and the thioacetyl derivatives 2a–c to deter-
mine their antiproliferative activities against human breast
cancer (MCF-7), human cervical cancer (HeLa), and human leu-
kemia cancer (K562) cell lines. TSA was used as a reference
compound.
In general, the hydroxamic acid derivatives 1a–c displayed
better antiproliferative activities than the thioacetyl com-
pounds 2a–c. Similar to their HDAC isoform inhibitory activi-
ties, compounds 1a–c exhibited increased antiproliferative ac-
tivity over TSA, with their IC50 values in the nanomolar range
(Table 1). In contrast, the thioacetyl derivatives 2a–c showed
moderate antiproliferative activities, with their IC50 values
barely in the micromolar range. Among these three cell lines,
the antiproliferative activities of 2a–c were clearly lower
against K562. In consideration of the critical function of HDAC6
in HDAC inhibitor-induced leukemia cell death,[27] the unex-
pectedly low activities are probably due to their poor HDAC6
inhibitory activity. In addition, aberrant epigenetic modifica-
tions on the p21 promoter in leukemia cells may also be re-
sponsible for their poor inhibitory activities against the K562
line.[28] Consistent with their enzyme inhibitory activities, no
drastic differences were observed among compounds 2a–c,
suggesting that the position of chlorine on the phenyl ring
does not have a significant effect on cellular potency.
Figure 4. Effects of compound 1b on cell-cycle arrest in MCF-7 cells. a) MCF-
7 cells were treated with indicated the concentrations of 1b for 24 h, then
stained with PI to determine cell-phase distribution. b) Cells were treated
with 26 nm 1b for the indicated times, then analyzed by western blot; b-
actin (actin) was used as internal control.
centage of cells in G0/G1 phase gradually increased from 52.16
to 62.18%, while that of the sub-G1 phase increased from 2.27
to 18.73%. During treatment, the expression of cyclin D1 was
decreased rapidly in the first 12 h, whereas the expression of
p21 was gradually increased under the same conditions, both
appearing in a time-dependent manner (Figure 4b). These re-
sults indicate that compound 1b induces cell-cycle arrest in
MCF-7 cells by simultaneously up-regulating the expression of
p21 while down-regulating the expression of cyclin D1.
Furthermore, p21 and cyclin D1 act as cell-cycle checkpoints
that are fatal for cell survival: on the one hand, cell-cycle
checkpoints enhance cell survival after DNA damage; on the
other hand, programmed cell death pathways like apoptosis
might be triggered during DNA repair.[31] In this study, evi-
dence for a 16% increase in apoptotic peak (sub-G1 phase)
was observed in cells treated with compound 1b. As apoptosis
could be triggered and affected through multiple pathways,
further studies are still in progress in our laboratories.
The antiproliferative activities of compounds 1a–c is in the
order: MCF-7 > HeLa > K562. In particular, the most active
compound, 1b, gave an IC50 value of 26 nm, which is at least fi-
vefold lower than that of TSA (140 nm) against MCF-7 cells.
Relative to 1a and 1c, attachment of the methyl group at po-
sitions 3 or 5 of the aliphatic ring led to increased antiprolifera-
tive activity. These results further support this position of the
aliphatic ring as a preferred modification site for the design of
more potent HDAC inhibitors. Given the clear advantage in cel-
lular potency observed in this assay, compound 1b was select-
ed for subsequent biological evaluation.
Inhibition of cell migration
Tumor growth and metastasis require a coordinated series of
events including degradation of the extracellular matrix, cellu-
lar adhesion, migration, and invasion.[32] Here we used a soft
agar assay, which is regarded as an effective substitute for in
vivo trials,[33] to investigate the effect of compound 1b on cell
migration. As shown in Figure 5a, the number of cell colonies
significantly diminished after treatment with compound 1b.
After incubation with compound 1b at 13 nm for 24 h, no col-
onies were observed. As shown in the enlarged images (Fig-
ure 5b), cell masses formed from untreated MCF-7 cells were
observed, whereas only a single cell or cell debris were ob-
served after exposure to compound 1b.
Colony-forming capacity, which is implicated in cancer de-
velopment and often initiates following cell migration, is
speculated to be influenced by the migratory activity of cancer
cells.[34,35] Accordingly, the effect of compound 1b on colony
formation was subsequently examined. Similar to the results
observed in the soft agar assay, the number of MCF-7 colonies
Cell-cycle arrest induced by compound 1b
It has been widely accepted that the inhibition of HDACs leads
to cell-cycle arrest in various human cancer cell lines. Cyclin D1
and p21 are both key regulators of cell proliferation. By bind-
ing with cyclin-dependent kinase (CDK), the cyclin D1–CDK
complex promotes cell-cycle progression, whereas p21 blocks
cell-cycle progression by inhibiting the activity of the cy-
clin D1–CDK complex.[29,30] Accordingly, cell-cycle distribution
and the expression level of relevant regulators p21 and cy-
clin D1 in MCF-7 cells were analyzed.
After incubation with compound 1b, cell-cycle phase distri-
bution dramatically changed, appearing in a concentration-de-
pendent manner (Figure 4a). After incubation for 24 h, the per-
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 627 – 637 631

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
MMP2 was subsequently decreased if cells were successively
treated for more than 12 h. After 24 h treatment with com-
pound 1b, the activity of MMP2 was barely detectable, sug-
gesting cell migration was inhibited.
TIMP1 is known as a key regulator in the cleavage of pro-
MMP2 to its active form. It is believed that the balance be-
tween MMP and TIMP determines the actual metalloproteinase
activities and controls ECM degradation.[32] To understand the
mechanism underlying the effect of compound 1b on cell mi-
gration inhibition, the expression levels of MMP2 and TIMP1
were then measured. Consistent with the activity variation of
MMP2 during treatment, the expression level of MMP2 was
slightly increased in cells treated with compound 1b during
the first 12 h (Figure 6b). However, if cells were successively
treated for another 12 h, the expression level of MMP2 was
dramatically reduced. In contrast, the expression level of TIMP1
was generally increased during the corresponding treatment
period. These data indicate that compound 1b simultaneously
regulates the expression levels of MMP2 and TIMP1, thereby
disrupting the balance between MMP2 and TIMP1 and leading
to cell migration inhibition.
Figure 5. Effects of compound 1b on cell migration. MCF-7 cells were suc-
cessively cultured in soft agar for 14 days until the resulting colonies were
large enough to be visualized. a) Cells were stained with MTT for colony vis-
ualization; each spot represents a single colony. b) Enlarged view of a single
cell colony (square fields indicated in panel a).
gradually decreased following prolonged treatment time (Sup-
porting Information figure S2). Nearly 40% of total cell colonies
vanished after treatment for 24 h.
Binding mode of 1b in HDAC4
It has been speculated that class II HDACs, especially HDAC4
and HDAC6, are associated with cell migration.[9] The expres-
sion profile revealed that HDAC4 expression is up-regulated in
breast cancer, whereas HDAC6 expression might be associated
with improved survival.[37,38] Therefore, a docking study was
conducted for the HDAC4–1b complex, and the hydrogen
bonds and hydrophobic contacts between HDAC4 and 1b
were then analyzed (Figure 7).
Regulating the balance between MMP2 and TIMP1 to inhibit
cell migration
MMPs comprise a family of proteases that are capable of de-
grading the extracellular matrix (ECM). MMP2, a major MMP
family member, is secreted as latent enzyme (pro-type,
72 kDa). After cleavage, the obtained active form of MMP2
(66 kDa) is able to modulate cell-matrix interaction, thus modu-
lating cell migration.[36] To determine the role of MMP2 in com-
pound-1b-induced cell migration inhibition, the activity of
MMP2 was measured using gelatin zymography. As shown in
Figure 6a, the catalytic activity of MMP2 dramatically increased
during the first 6 h of treatment. However, the activity of
Compound 1b binds very well to the active site of HDAC4,
exhibiting a binding mode similar to that of the HDAC2–Ky-2
complex. Specifically, in the HDAC4–1b complex, the oxygen
atom from the hydroxamate group of 1b establishes a hydro-
gen bond with the Ne2 atom of His159 and the OH atom of
Tyr332, while the nitrogen atom of 1b establishes hydrogen
bonds with Ne2 of His159. Hydrophobic interactions provided
by residues such as Asp115, Phe227, Leu299, and Pro298
around the active site of HDAC4 contribute to stabilizing the
HDAC4–1b complex. In summary, these results are in line with
the observed HDAC inhibitory activity of 1b and provide struc-
tural explanations for the clear lack of selectivity of 1b for
HDAC1 over HDAC4.
Conclusions
The binding energy and hydrophobic interactions in the
HDAC2–Ky-2 complex predicted by docking studies provides
a structural basis for the design of novel HDAC inhibitors. By
the introduction of subtle modifications to the SRD, novel
cyclic tetrapeptide-based HDAC inhibitors bearing either hy-
droxamate (1a–e) or thioacetyl moieties (2a–c) as their ZBG
were developed. Chlamydocin–hydroxamic acid analogues 1a–
e, containing either Anmc6c or l-Acn in their framework, ex-
hibited impressive HDAC inhibitory potential over TSA and Ky-
Figure 6. Effects of compound 1b on regulating the balance between
MMP2 and TIMP1. MCF-7 cells were treated with 26 nm 1b for the times in-
dicated. a) The extracellular activity of MMP2 is gauged as the intensity of
transparent bands against a dark background. b) The intracellular expression
levels of MMP2 and TIMP1 are presented as dark bands against a grey back-
ground. b-actin (actin) was used as internal control.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 627 – 637 632

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
Experimental Section
Materials and measurements
Unless otherwise noted, all solvents and reagents were reagent
grade and used without purification. All compounds obtained in
each reaction were routinely checked by thin-layer chromatogra-
phy (TLC) and HPLC. Analytical HPLC was performed on a Hitachi
instrument (Japan) equipped with a Chromolith Performance RP-
18e column (4.6ꢂ100 mm, Merck). The mobile phases used were:
A: H2O and 0.1% TFA, B: CH3CN with 0.1% TFA using a solvent gra-
dient conditions changing linearly from 0% B to 100% B over
15 min at a flow rate of 2 mLminꢁ1; elution was detected by UV
absorbance at l 220 nm. TLC detection was performed on alumi-
num-backed silica gel plates (Merck DC Alufolien Kieselgel 60 F254
,
Germany) with spots visualized under UV light and after heating.
Flash chromatography was performed with silica gel 60 (230–
400 mesh) with eluting solvents as indicated. High-resolution fast-
atom bombardment mass spectrometry (HR-FAB-MS) data were
collected on a JEOL JMS-SX 102A instrument. NMR spectra were re-
corded on a Varian INOVA 400 MHz spectrometer. All NMR spectra
1
were measured in CDCl3 with reference to TMS. All H NMR shifts
are given in ppm (s=singlet, d=doublet, t=triplet, m=multiplet).
General procedure for the preparation of target compounds
1a–e
All compounds were synthesized according to conventional solu-
tion-phase methods;[14] dicyclohexylcarbodiimide (DCC) and N-hy-
droxybenzotriazole (HOBt) were used as coupling reagents
throughout the synthesis. O-(7-Azabenzotriazol-1-yl)-1,1,3,3-tetra-
methyluronium hexafluorophosphate (HATU) was used as cycliza-
tion reagent.
cyclo(l-Asu(NHOH)-(ꢀ)A2mc6c-l-Phe-d-Pro) (1a): This compound
was synthesized according to conventional solution-phase meth-
ods. DCC (5 mmol) and HOBt (5.0 mmol) were added to a solution
of H-d-Pro(OtBu) (5 mmol) and Z-l-Phe-OH (5 mmol) in DMF
(10 mL), and the mixture was stirred overnight to give the dipep-
tide Z-l-Phe-d-Pro(OtBu). After evaporation of DMF, the Z-protect-
ed dipeptide was dissolved in EtOAc and washed with critic acid
(10%), NaHCO3 (4%) and brine. After evaporation of EtOAc, the
residue was purified by silica gel chromatography (CHCl3/MeOH
99:1). The Z-protected dipeptide was subsequently dissolved in
AcOH (5 mL), Pd/C (150 mg) was added, and the mixture was
stirred under H2 pressure (150 kPa) overnight to remove the Z
group. After filtration and evaporation of AcOH, the N-terminal
freed dipeptide was then dissolved in EtOAc and washed with
NaHCO3 (4%). After evaporation of EtOAc, the obtained H-l-Phe-d-
Pro(OtBu) was then coupled with Z-(ꢀ)A2mc6c-OH using HOBt/
DCC in the same manner as described earlier. Tripeptide Z-
(ꢀ)A2mc6c-l-Phe-d-Pro(OtBu) was deprotected by catalytic hydro-
genation to remove the Z group and purified as described earlier
to give H-(ꢀ)A2mc6c-d-Phe-d-Pro(OtBu). To the N-terminal freed
tripeptide, Boc-protected aminosuberic acid benzyl ester (Boc-l-
Asu(OBn)) and HOBt/DCC were added to give linear tetrapeptide
Boc-l-Asu(OBn)-(ꢀ)A2mc6c-l-Phe-d-Pro(OtBu). Boc-protected tetra-
peptide was dissolved in TFA (3 mL) at 08C and kept for 2 h. The
obtained TFA salt was then dissolved in DMF (250 mL), DIEA
(1.7 mmol), and highly diluted HATU (0.65 mmol in 100 mL DMF)
was added in five separate portions every 30 min as cyclization re-
agents to give cyclo(l-Asu(OBn)-(ꢀ)A2mc6c-l-Phe-d-Pro) in 70%
yield. After evaporation of DMF, cyclo(l-Asu(OBn)-(ꢀ)A2mc6c-l-
Phe-d-Pro) was washed with citric acid, NaHCO3, and brine as men-
Figure 7. Hydrogen bonds and hydrophobic contacts between HDAC4 and
1b. Distances between hydrogen bond donors and acceptors less than 3.5 ꢁ
are represented as dashed lines.
2. Among them, compound 1b exhibited the highest antiproli-
ferative activity, with an IC50 value of 26 nm against MCF-7
cells. Similar to TSA and Ky-2, cell-cycle arrest and apoptosis
were both observed in MCF-7 cells treated with compound
1b. Furthermore, compound 1b could effectively inhibit the
activity of MMP2, thereby inhibiting cell migration. Western
blot analysis further confirmed that 1b-induced migration in-
hibition is regulated by interrupting the balance between
MMP2 and TIMP1, which is in agreement with previous stud-
ies.[39,40] Recent studies further revealed that HDAC4 silencing
would increase the synergistic anticancer activity of combina-
tion treatments involving an HDAC inhibitor and cisplatin.[41]
As an effective inhibitor of HDAC1 and HDAC4, compound 1b
could find use as a promising tool for investigating the corre-
sponding molecular mechanisms underlying migration inhibi-
tion and combination treatments.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 627 – 637 633

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
tioned above. The OBn-protected cyclic tetrapeptide was dissolved
in MeOH (5 mL), and Pd/C (150 mg) was added to remove the OBn
group. The mixture was stirred overnight under H2 pressure
(150 kPa). After filtration and evaporation of MeOH, cyclo(l-Asu-
(ꢀ)A2mc6c-l-Phe-d-Pro) was obtained. The cyclic tetrapeptide was
then dissolved in DMF, and HCl·NH2OBn (1.5 mmol) and HOBt/DCC
(0.4 mmol) were added to give cyclo(l-Asu(NH-OBn)-(ꢀ)A2mc6c-l-
Phe-d-Pro). The OBn group was removed under H2 pressure with
Pd/BaSO4 as catalyst in the presence of AcOH to yield the desired
cyclic tetrapeptide cyclo(l-Asu(NHOH)-(ꢀ)A2mc6c-l-Phe-d-Pro).
After filtration, the cyclic tetrapeptide was then purified by gel
chromatography and lyophilization. Compound 1a was obtained
as a white powder (328 mg, 11%); tR: 7.36 min (HPLC); mp: 93.4–
1H), 4.88 (ddd, J=7.5, 10, 2.9 Hz, 1H), 6.48 (s, 1H), 7.16 (d, J=
10 Hz, 1H), 7.45 ppm (d, J=10 Hz, 1H); HRMS: m/z [M+H]+ calcd
for C22H36ClN5O6: 502.2433, found: 502.2417. CD spectra can be
found in Supporting Information figure S3.
cyclo(l-Asu(NHOH)-Aib-l-Ac6-d-Pro) (1e): This compound was
synthesized according to 1d using Z-l-Ac6-OH instead of Z-l-Phe-
OH. Compound 1e was obtained as a white powder (206 mg, 8%);
tR: 5.24 min (HPLC); mp: 76.2–77.18C; 1H NMR (400 MHz, CDCl3):
d=1.24 (t, 2H), 1.36 (s, 3H), 1.43 (d, J=6.7 Hz, 1H), 1.47 (d, J=
6.7 Hz, 1H), 1.49 (t, 2H), 1.16–1.74 (m, 6H), 1.76 (s, 3H), 1.78–1.88
(m, 4H), 2.16–2.40 (m, 4H), 2.88 (s, 1H), 2.96 (s, 1H), 3.53 (m, 3H),
3.72 (m, 1H), 4.24 (ddd, J=8, 9.9, 2.4 Hz, 1H), 4.77 (t, J=7.9 Hz,
1H), 4.85 (ddd, J=7.1, 10, 2.9 Hz, 1H), 6.40 (s, 1H), 7.18 (d, J=
10 Hz, 1H), 7.40 ppm (d, J=10 Hz, 1H); HRMS: m/z [M+H]+ calcd
for C23H38ClN5O6: 516.2580, found: 516.2620. CD spectra can be
found in Supporting Information figure S3.
1
93.88C; H NMR (400 MHz, CDCl3): d=0.78 (d, J=7.2 Hz, 3H), 1.05
(d, J=6 Hz, 1H), 1.24–1.40 (m, 8H), 1.60–1.86 (m, 9H), 2.04 (s, 1H),
2.14 (s, 2H), 2.29 (d, J=7.2 Hz, 1H), 3.19 (m, 1H), 3.24 (m, 1H), 3.47
(s, 1H), 3.50 (m, 1H), 3.84 (m, 1H), 3.98 (m, 1H), 4.32 (m, 1H), 4.73
(m, 1H), 5.20 (t, J=9.2, 8.0 Hz, 1H), 6.41 (s, 1H), 6.46 (s, 1H), 7.18
(d, J=7.2 Hz, 1H), 7.20–7.28 (m, 5H), 7.53 ppm (d, J=10, 1H);
13C NMR (100 MHz, CDCl3): d=14.21, 18.68, 21.05, 24.57, 24.86,
25.07, 25.13, 27.28, 27.41, 28.77, 29.32, 32.76, 36.07, 46.98, 52.88,
53.22, 54.91, 57.97, 65.04, 65.83, 125.85, 128.74, 128.74, 129.23,
129.26, 137.09, 171.36, 173.26, 175.22, 175.86 ppm; HRMS: m/z
[M+H]+ calcd for C30H43N5O6: 570.3292, found: 570.3306.
cyclo(l-Asu(NHOH)-Aib-l-Phe-d-Pro) (Ky-2): This compound was
synthesized according to previously reported procedures.[13] Ky-2
was obtained as a white powder (95 mg, 9%); tR: 14.3 min (HPLC);
1H NMR (400 MHz, CDCl3): d=1.77–1.25 (m, 16H), 2.13 (m, 2H),
2.31 (m, 1H), 2.94 (m, 3H), 3.16 (m, 2H), 3.86 (m, 1H), 4.22 (d, 1H),
4.68 (d, 1H), 5.12 (m, 1H), 6.31 (d, 1H), 7.11–7.29 (m, 7H), 7.56 ppm
(d, 1H); HRMS: m/z [M+H]+ calcd for C26H38N5O6 516.2822, found:
516.2819.
cyclo(l-Asu(NHOH)-l-A3mc6c-l-Phe-d-Pro) (1b): This compound
was synthesized according to cyclo(l-Asu(NHOH)-(ꢀ)A2mc6c-l-Phe-
d-Pro) (1a) using Z-l-A3mc6c-OH instead of Z-(ꢀ)A2mc6c-OH.
Compound 1b was obtained as a white powder (430 mg, 15%); tR:
1
7.80 min (HPLC); mp: 106.8–107.58C; H NMR (400 MHz, CDCl3): d=
General procedure for the preparation of target compounds
2a–c
0.88 (dd, J=4.0, 6.0 Hz, 4H), 1.16 (m, 3H), 1.19–1.30 (m, 6H), 1.48–
1.72 (m, 10H), 2.04 (s, 1H), 2.13 (m, 2H), 2.30 (d, J=8.4 Hz, 1H),
2.95 (m, 1H), 3.21 (m, 1H), 3.47 (m, 1H), 3.98 (m, 1H), 4.18 (m, 1H),
4.68 (t, J=6.0 Hz, 1H), 5.12 (m, 1H), 6.38 (s, 1H), 6.46 (s, 1H), 7.18
(d, J=9.2 Hz, 1H), 7.21–7.32 (m, 5H), 7.49 ppm (d, J=10 Hz, 1H);
13C NMR (100 MHz, CDCl3): d=22.50, 24.78, 24.99, 25.15, 28.16,
29.04, 29.41, 29.60, 32.69, 33.17, 34.11, 36.07, 41.56, 42.83, 47.11,
53.22, 54.57, 57.85, 62.48, 62.63, 126.94, 128.72, 128.78, 129.25,
129.32, 137.09, 171.22, 172.48, 173.33, 174.36 ppm; HRMS: m/z
[M+H]+ calcd for C30H43N5O6: 570.3292, found: 570.3257.
cyclo(l-Am7(SAc)-Aib-l-Phe(oCl)-d-Pro) (2a): This compound was
synthesized according to 1d using Boc-l-Ab7 (Boc-l-a-amino-7-
bromoalkanoic acid) instead of Boc-l-Asu(OBn), and Z-l-Phe(oCl)-
OH instead of Z-l-Ac5-OH. After cyclization, the obtained cyclo(l-
Ab7-Aib-l-Phe(oCl)-d-Pro) (0.8 mmol) was then dissolved in DMF
(3 mL). Potassium thioacetate (1.2 mmol) was added to DMF and
stirred for 4 h. After purification by silica gel chromatography and
lyophilization, cyclo(l-Am7(SAc)-Aib-l-Phe(oCl)-d-Pro) (0.5 mmol)
was obtained as a white power (289 mg, 10%); tR: 8.51 min
(HPLC); mp: 70.0–70.68C; 1H NMR (400 MHz, CDCl3): d=1.27 (m,
2H), 1.32 (s, 3H), 1.38 (m, 2H), 1.58 (m, 2H), 1.76 (s, 3H), 1.80 (m,
2H), 1.96 (d, J=8 Hz, 2H), 2.20 (m, 1H), 2.33 (s, 3H), 2.86 (t, 2H),
3.22 (m, 2H), 3.37 (m, 2H), 3.87 (m, 1H), 4.21 (m, 1H), 4.69 (d, J=
7 Hz, 1H), 5.34 (m, 1H), 6.11 (s, 1H), 7.11 (d, J=10 Hz, 1H), 7.17 (t,
2H), 7.25 (m, 1H), 7.34 (t, 1H), 7.53 ppm (d, J=10 Hz, 1H); 13C NMR
(100 MHz, CDCl3): d=23.65, 24.90, 25.02–25.41 (2C), 26.48, 28.50,
28.91–29.28(2C), 29.40, 30.79, 33.96, 47.19, 51.70, 54.55, 57.92,
58.94, 127.04, 128.50, 129.72, 131.90, 134.47, 134.87, 172.00,
172.83, 174.56, 175.77, 196.16 ppm; ESIMS: m/z [M+H]+ calcd for
C27H37ClN4O5S: 564.22, found: 565.30.
cyclo(l-Asu(NHOH)-A4mc6c-l-Phe-d-Pro) (1c): This compound
was synthesized according to 1a using Z-l-A4mc6c-OH instead of
Z-(ꢀ)A2mc6c-OH. Compound 1c was obtained as a white powder
(398 mg, 14%); tR: 7.60 min (HPLC); mp: 73.5–74.18C; 1H NMR
(400 MHz, CDCl3): d=0.77 (d, J=12 Hz, 1H), 0.84 (d, J=5.6 Hz, 3H),
1.25–1.30 (m, 12H), 1.61–1.74 (m, 6H), 2.05 (s, 1H), 2.13–2.29 (m,
4H), 2.94 (d, J=5.2 Hz, 1H), 2.97 (d, J=5.6 Hz, 1H), 3.29 (m, 1H),
3.97 (m, 1H), 4.23 (m, 1H), 4.68 (t, J=6.0 Hz, 1H), 5.15 (dd, J=9.2,
6.4 Hz, 1H), 6.46 (s, 1H), 6.67 (s, 1H), 7.19 (d, J=6.8 Hz, 1H), 7.21–
7.28 (m, 5H), 7.53 ppm (d, J=10 Hz, 1H); 13C NMR (100 MHz,
CDCl3): d=21.63, 24.77, 24.94, 25.05, 25.26, 28.39, 28.70, 29.18,
31.25, 32.56, 32.75, 34.04, 36.26, 35.78, 36.04, 47.03, 53.16, 54.70,
57.82, 61.95, 126.89, 128.71, 128.71, 129.22, 129.22, 137.08, 173.21,
174.31, 174.39, 174.48 ppm; HRMS: m/z [M+H]+ calcd for
C30H43N5O6: 570.3292, found: 570.3262.
cyclo(l-Am7(SAc)-Aib-l-Phe(mCl)-d-Pro) (2b): This compound was
synthesized according to 2a using Z-l-Phe(mCl)-OH instead of Z-l-
Phe(oCl)-OH. Compound 2b was obtained as a white powder
(285 mg, 10%); tR: 8.70 min (HPLC); mp: 101.9–102.38C; 1H NMR
(400 MHz, CDCl3): d=1.24 (m, 2H), 1.32 (s, 3H), 1.36 (m, 2H), 1.55
(m, 4H), 1.75 (s, 3H), 1.78 (m, 2H), 2.16 (m, 1H), 2.31 (s, 3H), 2.83
(t, 2H), 2.92 (m, 2H), 3.24 (m, 2H), 3.84 (m, 1H), 4.18 (m, 1H), 4.66
(d, J=6 Hz, 1H), 5.11 (m, 1H), 6.11 (s, 1H), 7.10 (m, 1H), 7.18 (s,
1H), 7.21 (d, J=5 Hz, 2H), 7.25 (s, 1H), 7.55 ppm (m, 1H); 13C NMR
(100 MHz, CDCl3): d=23.74, 24.92, 25.18, 26.57, 28.25, 28.51, 28.98,
29.08, 29.41, 30.82, 35.67, 47.19, 53.32, 54.52, 58.05, 59.00, 127.19,
127.44, 129.45, 130.04, 134.50, 139.25, 172.00, 172.65, 174.59,
cyclo(l-Asu(NHOH)-Aib-l-Ac5-d-Pro) (1d): This compound was
synthesized according to 1a using Z-Aib-OH instead of Z-
(ꢀ)A2mc6c-OH, and Z-l-Ac5-OH instead of Z-l-Phe-OH. Compound
1d was obtained as a white powder (176 mg, 7%); tR: 4.95 min
(HPLC); mp: 79.0–79.38C; 1H NMR (400 MHz, CDCl3): d=1.20 (t,
2H), 1.24 (t, 2H), 1.36 (s, 3H), 1.62–1.67 (m, 1H), 1.77 (s, 3H), 1.80–
1.96 (m, 2H), 2.15 (m, 1H), 2.88 (s, 1H), 2.95 (s, 1H), 3.55 (m, 3H),
3.72 (m, 1H), 4.26 (ddd, J=7.1, 10.3, 3.3 Hz, 1H), 4.78 (t, J=7.9 Hz,
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 627 – 637 634

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
175.88, 196.21 ppm; ESIMS: m/z [M+H]+ calcd for C27H37ClN4O5S:
564.22, found: 565.30.
cubated with 40 mg FLAG peptide in restore buffer (200 mL) for 1 h
at 48C to release HDACs. The supernatant was collected by centri-
fugation (12000 rpm, 12800 g) for 10 min at 48C. For inhibitory ac-
tivity assays against HDACs, 10 mL enzyme fraction was added into
the mixture composed of 1 mL fluorescent substrate (2 mm Ac-
KGLGK (Ac)-MCA) and 9 mL restore buffer. After incubation at 378C
for 30 min, 30 mL trypsin solution (20 mgmLꢁ1) was added to the
reaction mixture and incubated at 378C for 15 min to stop the re-
action. The released product, aminomethyl coumarin (AMC), was
measured using a fluorescence plate reader (Thermo Varioskan
Flash, USA) at l 252 nm.
cyclo(l-Am7(SAc)-Aib-l-Phe(pCl)-d-Pro) (2c): This compound was
synthesized according to 2a using Z-l-Phe(pCl)-OH instead of Z-l-
Phe(oCl)-OH. Compound 2c was obtained as a white powder
(313 mg, 10%); tR: 8.68 min (HPLC); mp: 71.8–72.18C; 1H NMR
(400 MHz, CDCl3): d=1.27 (m, 2H), 1.34 (s, 3H), 1.42 (m, 2H), 1.58
(m, 4H), 1.76 (s, 3H), 2.06 (m, 2H), 2.17 (m, 1H), 2.32 (s, 3H), 2.85
(t, 2H), 2.92 (m, 2H), 3.22 (t, 2H), 3.85 (m, 1H), 4.20 (m, 1H), 4.66
(d, J=6 Hz, 1H), 5.12 (m, 1H), 6.20 (s, 1H), 7.10 (d, J=9 Hz, 1H),
7.17 (d, J=7 Hz, 2H), 7.23 (t, 2H), 7.55 ppm (d, J=9 Hz, 1H);
13C NMR (100 MHz, CDCl3); d=23.72, 24.87, 25.13, 26.48, 28.46,
28.89–29.16 (3C), 29.37, 30.79, 35.34, 47.09, 53.36, 54.48, 57.97,
58.92, 128.71–129.14 (2C), 130.45–130.74 (2C), 132.72, 135.63,
171.98, 172.70, 174.58, 175.87, 196.18 ppm; ESIMS: m/z [M+Na]+
calcd for C27H37ClN4O5S: 564.22, found: 587.30.
p21 promoter assay: This assay was performed as described previ-
ously using a stably transformed cell line, namely MFLL-9.[31] MFLL-
9 expresses a low level of luciferase, the activity of which can be
enhanced by TSA in a dose-dependent manner. MFLL-9 cells (1ꢂ
105) were cultured in a 96-well plate for 6 h and then incubated for
18 h under identical conditions in the absence or presence of de-
sired concentrations of test compounds. After treatment, cells
were collected and lysed in the manner as mentioned above. The
luciferase activity of each cell lysate was measured with a LucLite
luciferase reporter gene assay kit and recorded using a Lumines-
cencer-JNR luminometer (ATTO, Tokyo, Japan). All data were nor-
malized to the protein concentration of cell lysate before they
were used for analysis. Concentrations at which a compound indu-
ces the luciferase activity 10-fold higher than the basal level are
presented as the 1000% effective concentration (EC1000).
Biology
Cell lines (MCF-7: human breast cancer; HeLa: human epithelial
cervical cancer; K562: myelogenous leukemia, 293T: human embry-
onic kidney cell line) were obtained from Shanghai Institute of Bio-
chemistry and Cell Biology (SIBS), CAS. RPMI 1640, DMEM, fetal
bovine serum (FBS), and newborn calf serum (NBCS) were pur-
chased from Gibco (USA). TSA was purchased from Sigma (USA).
LipofectAmine 2000 reagent was obtained from Invitrogen (USA),
pGL3-Basic plasmid was purchased from Promega (USA). BCA kit
was purchased from KeyGen Biotech (China). Protein A/G plus
agarose beads were obtained from Santa Cruz Biotech (USA). Anti-
FLAG M2 antibody and FLAG peptide were obtained from Sigma
(USA). LucLite luciferase Reporter Gene Assay Kit was purchased
from Packard Instrument (USA). Antibodies anti-p21, cyclin D1,
MMP2, TIMP1, and b-actin (actin) were purchased from Santa Cruz
Biotechnology (USA), and horseradish peroxidase (HRP)-conjugated
affinity goat anti-mouse secondary antibodies were obtained from
Invitrogen (USA). The ECL detection reagent was purchased from
Thermo Scientific (USA). MTT, trypsin, and propidium iodide (PI)
were purchased from Amresco (USA).
MTT assay: Cells in logarithmic growth phase were equally seeded
in a 96-well plate at a density of 1ꢂ104 cells per well. Cells were
cultured in complete growth medium for 24 h and then exposed
to various concentrations of test compounds. Cells used as nega-
tive control were exposed to DMSO under the same conditions.
After the 48 h incubation, 20 mL MTT solution (5.0ꢂ103 mgLꢁ1 di-
luted in PBS) was added to each well. For solid tumor cells (MCF-7
and HeLa cells), after incubation with MTT for 3 h, formazan
formed from MTT was dissolved in 200 mL DMSO and mixed for
10 min before measurement. For K562 cells, lysis solution (50%
DMF (v/v), 30% SDS (w/v)) was added to dissolve formazan. The
optical density was measured with an ELISA reader (Sunrise, Japan)
at 570/630 nm. The percent inhibition was calculated as per Equa-
tion (1):
Cell culture: All cell lines were cultured under humidified air con-
taining 5% CO2 at 378C in RPMI 1640 medium supplemented with
10% NBCS (MCF-7 and HeLa cell lines) or in RPMI 1640 medium
supplemented with 10% FBS (transformed mink lung epithelial cell
line, MFLL-9) or in DMEM supplemented with 10% FBS (K562 and
293T cell lines).
% inhib: ¼ ½1ꢁðA570 sampleꢁA630 sampleÞ=ðA570 ctrlꢁA630 ctrlÞꢂ ꢃ 100
ð1Þ
Flow cytometry analysis: MCF-7 cells were seeded at 2ꢂ105 mLꢁ1 in
six-well plated. The adherent cells were treated with increasing
concentrations of compound 1b ranging from 6.5 to 26 nm. After
24 h incubation, cells were detached using trypsin and then har-
vested by centrifugation for 5 min at 2000 rpm (360 g). Cells (1ꢂ
105) were then separated for detection. After washing twice with
cold PBS (48C), the cell pellets were fixed in 75% EtOH at ꢁ208C
overnight. The pellets were washed twice with cold PBS, and then
suspended in 200 mL PI solution (50 mgLꢁ1 PI, 1% (v/v) Triton X-
100 in PBS). After incubation at 48C for 20 min in the dark, cells
were measured using a BD FACSCanto flow cytometer (USA), and
the results were analyzed using Mod Fix LT 3.0 software.
HDAC inhibitory activity: 293T cells were seeded at 2ꢂ106 mLꢁ1 in
100 mm dishes. After culture for 24 h, cells were transiently trans-
fected with 10 g vector pcDNA3-HDAC1 for human HDAC1,
pcDNA3-HDAC4 for human HDAC4, or pcDNA3-mHDA2/HDAC6
for mouse HDAC6 using LipofectAmine 2000 reagent. After succes-
sive incubation in DMEM for 24 h, the transfected cells were
washed with PBS (137 mm NaCl, 2.7 mm KCl, 10 mm Na2HPO4,
4.2 mm KH2PO4, pH 7.2) and then lysed by sonication (150 w,
4 cycles of 30 s sonication, 60 s pause) in 6 mL cold lysis buffer
(50 mm Tris·HCl, 120 mm NaCl, 5 mm EDTA, and 0.5% NP-40,
pH 7.5). The soluble fraction was pre-cleared by incubation with
protein A/G plus agarose beads. After collection by microcentrifu-
gation (12000 rpm, 12800 g) for 2 min at 48C, the pre-cleared su-
pernatant was then incubated with 4 mg anti-FLAG M2 antibody
probing for HDAC1, HDAC4, or HDAC6 for 1 h at 48C. The agarose
beads were washed three times with lysis buffer and once with his-
tone deacetylase restore buffer (20 mm Tris·HCl, 150 mm NaCl, and
10% glycerol, pH 8.0). The obtained immune complex was then in-
Western blots: After treatment, cells were harvested, washed with
cold PBS, and then lysed in lysis buffer (50 mm Tris·HCl, 1% Triton
X-100, 10% glycerol, 1 mm PMSF, 1 mm DTT, 20 mgmLꢁ1 aprotinin,
pH 6.8) at 08C for 1 h. The lysates were clarified by centrifugation
(15000 rpm, 20000 g) for 10 min at 48C. Total protein concentra-
tion of the soluble cell extract was determined using a BCA kit
before electrophoresis. Same amount of protein was run on
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 627 – 637 635

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
a 12.5% (for cyclin D1, MMP2, TIMP1, and b-actin) or 15% (for p21
and b-actin) SDS-PAGE gel and then electrophoretically transferred
onto a PVDF membrane. The membrane was subsequently blocked
with 5% (w/v) non-fat milk in TBST saline (20 mm Tris·HCl, 150 mm
NaCl, 0.05% (v/v) Tween-20) overnight at 48C. The blocked mem-
brane was then incubated successively with primary antibody (di-
luted 1:800–1:1000) at 48C overnight, then HRP-labeled goat anti-
mouse secondary antibody (diluted 1:2000) at 378C for 1 h. After
incubation, the membrane was washed with TBST and the immu-
noreactive bands were visualized using an ECL reagent kit accord-
ing to the manufacturer’s instructions.
Acknowledgements
We acknowledge Dr. Alan Chang for his help in the revision of
this manuscript.
Keywords: cell migration · chlamydocin analogues · cyclic
tetrapeptides · docking · HDAC inhibitors
[1] N. Sengupta, E. Seto, J. Cell. Biochem. 2004, 93, 57–67.
[2] J. S. Lee, E. Smith, A. Shilatifard, Cell 2010, 142, 682–685.
[3] W. S. Xu, R. B. Parmigiani, P. A. Marks, Oncogene 2007, 26, 5541–5552.
[4] M. Yoshida, M. Kijima, M. Akita, T. Beppu, J. Biol. Chem. 1990, 265,
17174–17179.
[5] C. Choudhary, C. Kumar, F. Gnad, M. L. Nielsen, M. Rehman, T. C. Walther,
J. V. Olsen, M. Mann, Science 2009, 325, 834–840.
[6] S. Kim, R. Sprung, Y. Chen, Y. Xu, H. Ball, J. Pei, T. Cheng, Y. Kho, H. Xiao,
L. Xiao, N. V. Grishin, M. White, X. Yang, Y. Zhao, Mol. Cell 2006, 23, 607–
618.
Soft agar assay: For soft agar assay, cells were diluted to the de-
sired concentration (300 per well) in the upper agar layer (0.03%
agar in normal medium) in the absence or presence of compound
1b after the lower layer (2.0% agar in drug-free normal medium)
was plated as previously described.[32] After successive incubation
for 10–14 days, when colonies were large enough for visualization
by microscopy (Olympus, 1X71), MTT (200 mL) was added to each
well for visualization.
[7] M. Dokmanovic, C. Clarke, P. A. Marks, Mol. Cancer Res. 2007, 5, 981–
989.
[8] A. Valenzuela-Fernꢃndez, J. R. Cabrero, J. M. Serrador, F. Sanchez-Madrid,
Trends Cell Biol. 2008, 18, 291–297.
[9] M. Y. Ahn, D. O. Kang, Y. J. Na, S. Yoon, W. S. Choi, K. W. Kang, H. Y.
Chung, J. H. Jung, D. S. Min, H. S. Kim, Cancer Lett. 2012, 325, 189–199.
[10] N. A. P. Franken, H. M. Rodermond, J. Stap, J. Haveman, C. V. Bree, Nat.
Protoc. 2006, 1, 2315–2319.
[11] J. Tan, S. Cang, Y. Ma, R. L. Petrillo, D. Liu, J. Hematol, Oncology 2010, 3,
5.
[12] A. A. Lane, B. A. Chabner, J. Clin. Oncol. 2009, 27, 5459–5468.
[13] S. Fujii, T. Okinaga, W. Ariyoshi, O. Takahashi, K. Iwanaga, N. Nishino, K.
Tominaga, T. Nishihara, Biochem. Biophys. Res. Commun. 2013, 434, 413–
420.
[14] N. Nishino, B. Jose, R. Shinta, T. Kato, Y. Komatsu, M. Yoshida, Bioorg.
Med. Chem. 2004, 12, 5777–5784.
[15] A. P. Kozikowski, K. V. Butler, Curr. Pharm. Des. 2008, 14, 505–528.
[16] A. De Ruijter, A. H. Van Gennip, H. N. Caron, S. Kemp, A. Van Kuilenburg,
Biochem. J. 2003, 370, 737–749.
Gelatin zymography: This assay was performed as described previ-
ously.[42] Generally, MCF-7 cells were seeded into six-well plate and
maintained for 24 h for adherence. The cells were then treated
with compound 1b in serum-free medium for various times. After
treatment, the conditioned medium was collected, mixed with
loading buffer, and then electrophoresed on a 10% SDS-PAGE gel
containing 0.1% (w/v) gelatin. After electrophoresis, the gel was
washed two times in refolding buffer (50 mm Tris·HCl, 200 mm
NaCl, 10 mm CaCl2, 2.5% (v/v) Triton X-100, pH 7.4) to remove SDS.
Then the refolded gel was soaked in reaction buffer (50 mm
Tris·HCl, 200 mm NaCl, 10 mm CaCl2, 0.02% Brij-35, pH 7.4) at 378C
for 36 h. After reaction, the gel was stained for 3 h with staining
solution (0.1% Coomassie Brilliant Blue, 30% MeOH, and 10%
AcOH), and then destained in destaining solution (30% MeOH and
10% AcOH in distilled H2O) until clear bands against the dark-blue
background could be visualized.
[17] C. J. Millard, P. J. Watson, I. Celardo, Y. Gordiyenko, S. M. Cowley, C. V.
Rpbinson, L. Fairall, J. W. R. Schwabe, Mol. Cell 2013, 51, 57–67.
[18] G. M. Shivashimpi, S. Amagai, T. Kato, N. Nishino, S. Maeda, T. G. Nishino,
M. Yoshida, Bioorg. Med. Chem. 2007, 15, 7830–7839.
[19] D. Huang, X. Li, L. Sun, Z. Xiu, N. Nishino, J. Pept. Sci. 2012, 18, 242–
251.
[20] W. Gu, I. Nusinzon, R. D. Smith, C. M. Horvath, R. B. Silverman, Bioorg.
Med. Chem. 2006, 14, 3320–3329.
[21] T. A. Miller, D. J. Witter, S. Belvedere, J. Med. Chem. 2003, 46, 5097–
5116.
[22] K. B. Glaser, Biochem. Pharmacol. 2007, 74, 659–671.
[23] S. Schꢄfer, L. Sauders, E. Eliseeva, A. Velena, M. Jung, A. Schwienhorst, A.
Strasser, A. Dickmanns, R. Ficner, S. Schlimme, W. Sippl, E. Verdin, M.
Jung, Bioorg. Med. Chem. 2008, 16, 2011–2033.
[24] Y. Chen, M. Lopez-Sanchez, D. N. Savory, D. D. Billadeau, G. S. Dow, A. P.
Kozikowski, J. Med. Chem. 2008, 51, 3437–3448.
[25] E. S. Inks, B. J. Josey, S. R. Jesinkey, C. J. Chou, ACS Chem. Biol. 2012, 7,
331–339.
[26] T. Suzuki, A. Kouketsu, Y. Itoh, S. Hisakawa, S. Maeda, M. Yoshida, H. No-
kagawa, N. Miyata, J. Med. Chem. 2006, 49, 4809–4812.
[27] C. Florean, M. Schnekenbyrger, C. Grandjenette, M. Diederich, Epige-
nomics 2011, 3, 581–609.
Docking: Docking studies were conducted with AutoDock 4.2 using
Lamarckian genetic algorithm.[43] Initial structures of HDAC2 and
HDAC4 were modeled from the atom coordinates of the X-ray crys-
tal structure (PDB IDs: 3MAX for HDAC2, 2VQJ for HDAC4).[44,45]
Target compound was located near the active site of HDAC. A grid
size of 70ꢂ70ꢂ70 points with a spacing of 0.375 ꢁ between the
grid points was implemented to cover the entire surface of HDAC.
For the tested compounds, all single bonds except the amide
bonds and cyclic bonds were treated as active torsional bonds;
150 independent dockings (150 runs) were performed using genet-
ic algorithm searches. A maximum number of 2.5ꢂ106 energy eval-
uations and a maximum number of 5000 generations were imple-
mented during each genetic algorithm run. A mutation rate of 0.02
and a crossover rate of 0.8 were used. Other parameters were all
set as default in AutoDock 4.2. The PyMOL (http://www.pymol.org)
and LigPlot programs were also used to analyze the docking re-
sults, focusing on hydrogen bonds and hydrophobic interac-
tions.[46] Hydrogen bonds with a donor–acceptor distance <3.5 ꢁ
were represented.
Statistics: All data represented at least three independent experi-
ments and are expressed as the mean ꢀSD unless otherwise indi-
cated. Statistical comparisons were made by Student t-test and
one-way ANOVA method using Origin 8.0.
[28] C. Bradbury, F. Khanim, R. Hayden, C. Bunce, D. White, M. Drayson, C.
Craddock, B. Turner, Leukemia 2005, 19, 1751–1759.
[29] K. Harada, G. R. Ogden, Oral Oncol. 2000, 36, 3–7.
[30] D. Dubravka, D. W. Scott, Cell Res. 2000, 10, 1–16.
[31] R. Debret, S. Brassart-Pasco, J. Lorin, A. Martoriati, A. Deshorgue, F. Ma-
quart, I. Rahman, F. Antonicelli, Biochim. Biophys. Acta Mol. Cell Res.
2008, 1783, 1718–1727.
[32] K. Kessenbrock, V. Plaks, Z. Werb, Cell 2010, 141, 52–67.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 627 – 637 636

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
[33] L. M. Coussens, B. Fingleton, L. M. Matrisian, Science 2002, 295, 2387–
2392.
[42] P. A. M. Snoek-van Beurden, J. W. Von den Hoff, BioTechniques 2005, 38,
73–83.
[34] E. Y. Lin, A. V. Nguyen, R. G. Russell, J. W. Pollard, J. Exp. Med. 2001, 193,
727–739.
[35] R. Brunmeir, S. Lagger, C. Seiser, Int. J. Dev. Biol. 2009, 53, 275–289.
[36] H. Jeon, L. T. Zheng, S. Lee, W. H. Lee, N. Park, J. Y. Park, W. D. Heo, M. S.
Lee, K. Suk, Exp. Cell Res. 2011, 317, 2007–2018.
[43] G. M. Morris, D. S. Goodsell, R. S. Halliday, R. Heuy, W. E. Hart, R. K.
Belew, A. J. Olson, J. Comput. Chem. 1998, 19, 1639–1662.
[44] J. C. Bressi, A. J. Jennings, R. Skene, Y. Wu, R. Melkus, R. D. Jong, S.
O’Connell, C. E. Grimshaw, M. Navre, A. R. Gangloff, Bioorg. Med. Chem.
Lett. 2010, 20, 3142–3145.
[37] V. Duong, C. Bret, L. Altucci, A. Mai, C. Duraffourd, J. Loubersac, P. Har-
mand, S. Bonnet, S. Valente, T. Maudelonde, V. Cavailles, N. Boulle, Mol.
Cancer Res. 2008, 6, 1908–1919.
[38] B. Bardeda-Zahonero, M. Parra, Mol. Oncol. 2012, 6, 579–589.
[39] I. M. Clark, T. E. Swingler, C. L. Sampieri, D. R. Edwards, Int. J. Biochem.
Cell Biol. 2008, 40, 1362–1378.
[45] M. J. Bottomley, P. L. Surdo, P. D. Giovine, A. Cirillo, R. Scarpelli, F. Fer-
rigno, P. Jones, P. Neddermann, R. D. Francesco, C. Steinkuhler, P. Galli-
nari, A. Carfi, J. Biol. Chem. 2008, 283, 26694–26704.
[46] A. C. Wallace, R. A. Laskowski, J. M. Thornton, Protein Eng. 1995, 8, 127–
134.
[40] A. Suminoe, A. Matsuzaki, H. Hattori, Y. Koga, E. Ishii, T. Hara, Leukemia
Res. 2007, 31, 1437–1440.
Received: September 15, 2013
Revised: November 6, 2013
[41] E. A. Stronach, A. Alfraid, N. Rama, C. Dalter, J. B. Studd, R. Agarwal, T. G.
Guney, C. Gourley, B. T. Hennessy, G. B. Mills, A. Mai, R. Brown, R. Dina,
H. Gabra, Cancer Res. 2011, 71, 4412–4422.
Published online on November 27, 2013
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 627 – 637 637