Pyrimethamine conjugated histone deacetylase inhibitors: Design, synthesis and evidence for triple negative breast cancer selective cytotoxicity

Article abstract of DOI:10.1016/j.bmc.2020.115345

Signal transducer and activator of transcription 3 (STAT3) is an oncogenic transcription factor which has been recognized as a promising cancer therapeutic target. Small molecule pyrimethamine (PYM) is a known direct inhibitor of activated STAT3 and it is

Full text of DOI:10.1016/j.bmc.2020.115345

Journal Pre-proofs
Pyrimethamine Conjugated Histone Deacetylase inhibitors: Design, Synthesis
and Evidence for Triple Negative Breast Cancer Selective Cytotoxicity
Bocheng Wu, Shaghayegh Fathi, Shanee Mortley, Mahir Mohiuddin, Young
C. Jang, Adegboyega Oyelere
PII:
S0968-0896(20)30135-8
https://doi.org/10.1016/j.bmc.2020.115345
BMC 115345
DOI:
Reference:
To appear in:
Bioorganic & Medicinal Chemistry
Received Date:
Revised Date:
Accepted Date:
10 October 2019
12 January 2020
22 January 2020
Please cite this article as: B. Wu, S. Fathi, S. Mortley, M. Mohiuddin, Y.C. Jang, A. Oyelere, Pyrimethamine
Conjugated Histone Deacetylase inhibitors: Design, Synthesis and Evidence for Triple Negative Breast Cancer
Selective Cytotoxicity, Bioorganic & Medicinal Chemistry (2020), doi: https://doi.org/10.1016/j.bmc.
2
020.115345
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover
page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version
will undergo additional copyediting, typesetting and review before it is published in its final form, but we are
providing this version to give early visibility of the article. Please note that, during the production process, errors
may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
©
2020 Published by Elsevier Ltd.

Pyrimethamine Conjugated Histone Deacetylase inhibitors: Design, Synthesis and
Evidence for Triple Negative Breast Cancer Selective Cytotoxicity.
1
†
1, 2†
1
4, 5
3, 4,
Bocheng Wu , Shaghayegh Fathi , Shanee Mortley , Mahir Mohiuddin , Young C. Jang
5,
1, 4*
Adegboyega Oyelere
School of Chemistry and Biochemistry, School of Biological Sciences, Parker H. Petit Institute
for Bioengineering and Bioscience, Georgia Institute of Technology, Atlanta, GA 30332-0400
USA
1School of Chemistry and Biochemistry, Georgia Institute of Technology
2Current Address: University of Chicago
3School of Biological Sciences, Georgia Institute of Technology
4Parker H. Petit Institute for Bioengineering and Bioscience, Georgia Institute of Technology
5Wallace H. Coulter Department of Biomedical Engineering, Georgia Institute of Technology
†These Authors contributed equally to the manuscript
Correspondence to:
Adegboyega K. Oyelere, E-mail: aoyelere@gatech.edu
Keywords: Pyrimethamine, STAT proteins, STAT3, histone deacetylases, histone deacetylase
inhibitors, Triple Negative Breast Cancer, molecular docking
1

Abstract:
Signal transducer and activator of transcription 3 (STAT3) is an oncogenic transcription factor
which has been recognized as a promising cancer therapeutic target. Small molecule
pyrimethamine (PYM) is a known direct inhibitor of activated STAT3 and it is currently under
clinical trial. Also, histone deacetylase (HDAC) inhibition has been shown to indirectly attenuate
STAT3 signaling through inhibition of STAT3 activation. Herein we described the design and
biological profiling of two classes of PYM-conjugated HDAC inhibitors (HDACi). We
observed that the class I PYM-HDACi compounds 12a-c potently inhibited HDACs 1 and 6 in
cell free assays while a lead class II PYM-HDACi compound 23 showed a strong HDAC 6
selective inhibition. In a cell-based assay, 12a-c are preferentially cytotoxic to MDA-MB-231, a
TNBC cell line that is highly STAT3-dependent, while 23 showed no such selective toxicity.
Subsequent target validation studies revealed that a representative class I PYM-HDACi
compound 12c elicited a signature of HDAC and STAT3 pathway inhibition intracellularly.
Collectively, these data suggest that PYM-HDACi compounds are promising leads to develop
targeted therapy for TNBC.
Key words: Triple Negative Breast Cancer, Pyrimethamine, HDAC inhibitor, STAT3 pathway
2

Introduction:
Pyrimethamine (PYM) (Figure 1) is an FDA approved drug which, due to its perturbation
of the functions of several intracellular targets, has found use for the management of various human
1
,2
diseases including toxoplasmosis and malaria. PYM has also been used in chemotherapy along
3
with other drugs such as proguanil for few decades. PYM’s anti-parasitic activity originates from
its ability to specifically bind and inhibit dihydrofolate reductase (DHFR, 5,6,7,8-tetrahydrofolate:
+
4
NADP oxidoreductase, EC 1.5.1.3) in Plasmodium falciparum and other protozoa. DHFR is
critical for folate metabolism and has been a drug target for fungal, protozoal and bacterial
infections and cancer. DHFR facilitates an NADPH-dependent reduction of dihydrofolate to
tetrahydrofolate, a cofactor necessary for the biosynthesis of thymidylate, purine nucleotides, and
5
many other essential amino acids required for protein, RNA, and DNA synthesis. DHFR
inhibition by antifolate compounds interferes with these pathways, resulting in cell cycle arrest
and cell death.6
N
NH2
N
NH2
Cl
Figure 1. Structure of pyrimethamine.
PYM is also an inhibitor of STAT3 (Signal transducer and activator of transcription 3)
7
transcriptional function. STAT3 is a member of STAT proteins comprising of seven sub-family
8
members (STAT1, 2, 3, 4, 5a, 5b, 6). STAT3, an oncogenic transcription factor with critical role
in the signaling of a number of cytokines and growth factors, confers resistance to apoptosis in
9
various cell types and is activated in many cancers including triple-negative breast cancer
3

(TNBC), acute myeloid leukemia (AML), chronic lymphocytic leukemia (CLL) and small
lymphocytic lymphoma (SLL).10,11
STAT3 activation through its tyrosine phosphorylation by JAK (Janus kinase) or IL-6
1
2
signaling cascade enhances its dimerization and translocation from cytoplasm to nucleus where
it can bind to certain DNA sequences and regulate genes expression involved in various cellular
processes. Tyrosine phosphorylation is not the only way to activate STAT3. It can be activated
through other processes such as serine phosphorylation, acetylation, methylation, and
glutathionylation.1
3,14
Once STAT3 is activated, it enhances various cell processes such as cell
proliferation, differentiation, survival, and angiogenesis that contribute to malignant
1
5
transformation and progression in many cancers such as breast, ovary, and prostate. Although
STAT3 also has non-transcriptional responsibilities, such as regulation of mitochondrial function,
most of its oncogenic activities are related to its gene regulation in the nucleus.16
Activation of STAT3 is tightly regulated in normal conditions; however, in cancer, it is
1
7
highly activated and leads to malignant cancer cells phenotype. Therefore, inhibiting STAT3
activation is a promising strategy for cancer therapy, as several cancer types depend on activated
STAT3 for their survival. In fact, PYM is currently in phase I/II clinical trials as a standalone agent
for the treatment of relapsed CLL and SLL (ClinicalTrials.gov Identifier: NCT01066663).
Interestingly, however, it has been observed that STAT3 inhibitors exhibit a synergistic effect with
other therapeutic agents in inhibiting tumor stem cells, leading to improved therapeutic indices for
these agents.18
Histone deacetylase (HDAC) enzymes are a class of proteins that play an important role in
1
9
regulating STAT3 activation. HDACs, along with histone acetyltransferase (HAT), control gene
expression, chromatin condensation and play an essential role in transcriptional activation by
4

2
0
regulating acetylation state of histone proteins. In addition to histones, the expression levels and
acetylation status of several non-histone proteins, including transcription factors (E2F, STAT3,
P53, NF-ҝB), estrogen receptor (ERα), androgen receptor (AR), α-tubulin, and chaperons
2
1
(
HSP90), are regulated by HDAC and HAT activity. Due to their critical roles in regulating a
wide range of cellular pathways, HDACs are considered as promising drug discovery targets, and
their inhibition has emerged as a potential strategy in treating various diseases including
2
2
neurological diseases, malaria, leishmania and cancer. To date, there are four US FDA approved
HDAC inhibitors (HDACi), namely, SAHA (suberoylanilide hydroxamic acid, vorinostat)
2
3
approved in 2006 for relapsed and refractory cutaneous T-cell lymphoma (CTCL), romidepsin
(
(
FK228) approved in 2009 for relapsed/refractory peripheral T-cell lymphoma,24 belinostat
PXD101) approved in 2014 for relapsed/refractory peripheral T-cell lymphoma,25 and
2
6
panobinostat (LBH589) approved in 2015 for treating acute myeloma (Figure 2). Chidamide
CS055) is another HDACi which is approved in China for treating relapsed/ refractory peripheral
(
2
7
T-cell lymphoma. Chidamide is in phase II clinical trials in the US. Inhibition and knockdown
of class I HDACs have been shown to result in inhibition of STAT3 activation, through
upregulation STAT3 Lys685 acetylation and attenuation of STAT3 Tyr705 phosphorylation,
resulting in the inhibition of the survival of pSTAT3-positive lymphoma (DLBCL) cells.28
Therefore, targeting STAT3-positive cancer cells with HDACi is another potentially viable
therapeutic option for managing these tumors. Inhibition of STAT3 pathway has been mediated
by HDACi SAHA through the acetylation on bromodomain protein 4 (BRD4) to down-regulate
Leukemia Inhibitory Factor Receptor (LIFR).29
5

O
O
H
O
O
O
N
OH
S
OH
N
NH2
N
N
N
H
H
H
N
O
H
H
N
SAHA
O
O
Belinostat
Chidamide
F
O
O
N
HN
H
HN
HN
S
O
S
N
H
NHOH
O
N
O
O
H
Panobinostat
Romidepsin
Figure 2. Structures of approved HDACi.
We hypothesized that designed multiple ligands comprising PYM and HDAC inhibition
chemotype would integrate direct STAT3- and HDAC-inhibition within a single molecular
template. These PYM-HDACi compounds are anticipated to be efficient inhibitors of proliferation
of tumors which are exquisitely dependent on STAT3 signaling pathway. Herein we demonstrate
that PYM-HDACi compounds inhibit representative HDACs, downregulate the expression of
selected STAT3 target proteins and are selectively cytotoxic to MDA-MB-231, a triple-negative
breast cancer (TNBC) cell that is highly dependent on STAT3 Pathway for its proliferation and
metastasis.
Results and discussion
Design of PYM-HDACi compounds
6

The three-motif HDACi pharmacophoric model consists of a recognition cap group, linker
3
0
group, and zinc-binding group (ZBG) (Figure 3). PYM structure closely resembles the HDACi
aryl-derived cap group. Our molecular docking analysis (discussed below) suggested that the PYM
halogen group could be replaced by the HDACi linker and ZBG groups without significantly
impacting STAT3 binding. Interestingly, the substitution of the PYM halogen group with alkyl,
3
1
aryl, and ring systems have been shown to be compatible with its biological activities. Based on
these observations, we designed two classes of PYM-HDACi compounds using PYM as a
surrogate for HDACi cap group. The restriction on the length of the linker group of class I
compounds is based on our previous study which revealed that five and six methylenes are an ideal
length for linker group for aryl triazolyl HDACi which inspired the design of this class of
2
2
compounds. Relatedly, the linker group of class II compounds is based on a similar moiety in the
approved HDACi belinostat and panobinostat.
7

O
(a)
H
N
SAHA
NHOH
ZBG
O
Cap group
Linker group
(b)
NH2
N
N
N
N
O
H N
Class I
A-C
2
N
NHOH
n
n= 3, 4, 5
NH2
N
N
Class II
D
H N
O
2
NHOH
NH2
N
H N
O
2
E
N
NHOH
Figure 3. (a) Pharmacophoric model of HDACi using SAHA as a prototypical HDACi. (b)
Structures of the designed two classes of PYM-HDACi compounds.
Molecular docking study
3
2
We first performed an unbiased molecular docking, using AutoDock Vina, to determine
the potential docking poses of PYM on the structure of STAT3 (PDB ID: 1BG1). The docking
outputs revealed that PYM binds to two pockets (P1 and P2) within the DNA binding domain
(DBD) and one solvent-exposed, shallow pocket (P3) between the connector (a part of DBD) and
the SH2 domains of STAT3 (Figure 4). Although PYM is accommodated through stabilizing H-
8

bonding and hydrophobic interactions with key residues within P1, P2 and P3, the binding energies
of PYM at the three sites revealed a strong preference for P2. In P1 and P2, the halogen moiety of
PYM is oriented in regions adjoining solvent-accessible grooves/sub-pockets which could
potentially accommodate the substitution of the halogen by the HDACi linker and ZBG groups of
the designed PYM-HDACi compounds (Figure 3). Conversely, in P3, the halogen moiety of PYM
is tucked into a shallow pocket, an orientation which may necessitate an extensive change in the
binding orientation for the PYM-HDACi to be accommodated at this location. Subsequent docking
of a representative class I PYM-HDACi compound B revealed that it adopts similar docking poses
as PYM at P1 and P2. At P3, however, the phenylpyrimidinediamine end of B is forced out into
the protein surface to accommodate its HDAC inhibition moiety. Interestingly, B binds to P1 and
P2 with enhanced binding affinities relative to PYM (Figure 4). This observation strongly suggests
that the replacement of the PYM halogen group by the designed HDAC inhibiting moieties is
compatible with the STAT3 binding attributes of PYM at P1 and P2.
9

a
bi
bii
P1
ci
cii
P2
1
0

di
dii
P3
Figure 4. Molecular docking of PYM and class I PYM-HDACi compound B to STAT3
(
PDB:1BG1). (a) The 4 domains of the STAT3 – 4-Helix bundle (blue), barrel (red), connector
green) and SH2 (yellow) domains. Docked poses of PYM and class I compound B at P1 (bi-ii),
(
P2 (ci-ii) and P3 (di-ii). The binding energies of PYM at P1, P2 and P3 are -5.6kcal/mol, -
.4kcal/mol and -6.0kcal/mol respectively, while B bound to P1, P2 and P3 with binding energies
6
of -9.2kcal/mol, -9.3kcal/mol and -7.3kcal/mol, respectively. Note that the binding of compound
B to P1 could be potentially stabilized by H-bonding interactions with Ser-381, Asn-420, His-437,
Thr-440, and Glu-455. In P2, compound B could form stabilizing H-bonding interactions with
Asp-334, Ile-467, Cys-468, Asn-472 and Asp-566. In P3, compound B could form H-bonding
interactions with Gln-524, Glu-582 and Tyr-584.
To confirm that the PYM could act as a surrogate for HDACi cap group, we performed
molecular docking analyses to interrogate the interaction between the designed PYM-HDACi and
selected HDAC isoforms. We observed that, in addition to engaging in stabilizing H-bonding
interaction with enzymes’ outer rim residues, class I compounds B and C adopt poses that may
allow effective chelation of zinc ion in the active sites of HDAC 1 and HDAC 6 (Figures 5 and 6).
In both HDAC 1 and HDAC 6, class I compound B shows strong evidence of zinc ion chelation
in the pocket. Presumably due to its flexible linker, class I compound B is able to facilely access
1
1

the pockets of both enzymes, allowing efficient chelation of the active site zinc ion which is a key
driver of HDAC 1 binding by both classes of compounds. Relative to Class II compounds D and
E, the hydroxamate group of B seems to be positioned to allow a more efficient zinc ion chelation
at HDAC 1 active site. The somewhat less optimal zinc ion chelation by D and E could be partially
compensated by the possibility of formation of  interaction with rings of Phe-155 and Phe-210
(Figure 5). Nevertheless, the rigidity of D and E constrained their phenylpyrimidinediamine
moiety to be presented on HDAC 1 surface where there is no obvious prospect for stabilizing
interactions. In contrast, the flexibility of the alkyl linker of B allows its phenylpyrimidinediamine
moiety to be tucked into a hydrophobic patch on HDAC 1 surface where it is further stabilized by
a hydrogen bonding interaction with Glu-203. This in silico observation suggests that compound
B could be better accommodated at HDAC 1 active site than D and E. Interestingly, the extra
benzene ring of class II compound E is able to overcome the deficiency of class II compound D
as it enables better zinc ion chelation and optimal binding within hydrophobic regions at outer rims
guarding the active sites of HDAC 6. Specifically, compound E could potentially form two

 interactions with Phe-202 and Trp-261 while compound D docked poses could only support
one  interaction with Phe-202. Also, compound B could form H-bonding with the amide
backbone of Ser-259 at the enzyme rim. The extra interactions displayed by B and E could confer
on them better binding affinities for HDAC 6 relative to compound D.
1
2

ai
aii
aiii
bii
bi
*
Note that the red portion of the protein surface indicates hydrophobic and white area indicates the
hydrophilic areas.
Figure 5. Docked poses of PYM-HDACi at the active sites of HDAC 1 (PDB:5ICN). Grey sphere
represents zinc ion in the active site of HDAC isoform. (ai-iii) Docked pose of class I compound
B on HDAC 1. (bi-ii) Overlay of the docked poses of class II compounds D (in purple) and E (in
1
3

grey) on HDAC 1. The compounds are accommodated at enzyme’s active site through a
combination of zinc chelation, H-bonding and hydrophobic interactions.
ai
aii
aiii
bi
bii
1
4

biii
Figure 6. Docked poses of PYM-HDACi at the active sites of HDAC 6 (PDB:5G0G). Grey sphere
represents zinc ion in the active site of HDAC isoform. (ai-iii) Docked pose of class I compound
B on HDAC 6. (bi-iii) Overlay of the docked poses of class II compounds A (color in light pink)
and B (color in cyan) on HDAC 6. The compounds are accommodated at enzyme’s active site
through a combination of zinc chelation, H-bonding and hydrophobic interactions. *Dotted lines
indicate interatomic distance for H-bonding and stacking interaction.
Based on evidence from the previous molecular docking analysis on human DHFR
3
3
(
hDHFR), it seemed that the replacement of the halogen group of PYM may not be compatible
with DHFR binding. Nevertheless, we docked these compounds against the structure of hDHFR
PDB code 1U72). We observed that the PYM moiety of the PYM-HDACi compounds bind
(
hDHFR with similar orientation as PYM (Figure S1). However, the PYM-HDACi compounds’
pyrimidinediamine head does not gain access into the hDHFR binding pocket as efficiently as
PYM, largely due to the interruption caused by their HDACi moiety. This result suggests that the
docked poses of PYM-HDACi may not be favorable thus making DHFR inhibition a less likely
attribute of the PYM-HDACi compounds.
1
5

Chemistry
To synthesize the class I PYM-HDACi compounds, 4-bromophenyl acetonitrile 1 was
3
4
reacted with ethyl propanoate 2 under basic condition to yield β-ketonitrile 3 which was then
3
5
converted to the methoxyphenol 5 using trimethyl orthoformate 4. Pyrimidine ring was formed
3
1
through cyclization reaction of 5 with guanidine hydrochloride to afford compound 6. Boc
protection of amine36,37 groups of 6, to give 7, followed by sonogashira reaction with
3
8
trimethylsilylacetylene 8 resulted in compound 9. Trimethylsilyl group was removed using
potassium carbonate to afford alkyne 10. Subsequently, copper (I) catalyzed azide-alkyne-
3
9
40
cycloaddition (AAC) reaction between alkyne 10 and compounds 11a-c, followed by removal
of trityl- protecting group resulted in the final class I compounds 12a-c. Control compound 12d
was similarly synthesized from Boc deprotected compound 10a and azido ester 11d (Scheme 1).
O
O
3
CH C(OMe)3
O
O
a
CN
2
4
CN
Br
b 53%
CN
c 72%
90%
1
Br
3
5
Br
N
NH2
N
N(Boc)2
N
N(Boc)2
TMS
8
N
N
N
d
86%
f, 36%
e
NH2
N(Boc)2
Br
N(Boc)2
Br
7
9
6
TMS
O
N3
O
n
N
H
N
NH2
n = 3 11a
n = 4 11b
n = 5 11c
N
N
N(Boc)2
N
NH2
g, h 84-88%
N
N
N(Boc)2
N
n
n = 3 12a
n = 4 12b
n = 5 12c
10
HN
HO
O
1
6

O
N3
n
O
N
NH2
n = 5 11d
N
NH2
N
N
NH2
n = 5 12d
g
90%
N
NH2
0a
N
N
n
1
O
O
Scheme 1. (a) Potassium tert-pentylate 25% in toluene,THF, rt, 20 min, 90%; (b) 6 h, neat
reaction 120 °C, 53%; (c) Guanidine hydrochloride, NaHCO , DMSO, 100 °C, 6 h, 72%; (d)
3
Boc O, THF, DMAP, 45 °C, THF, 86%; (e) Hunig’s base, Pd(PPh ) , acetonitrile, CuI, 75 °C,
2
3 4
overnight; (f) K CO , MeOH, 0 °C, 2 h, 36% (e and f); (g) CuI, Hunig's base, rt, overnight; (h)
2
3
TFA, DCM, rt, 2 h, 84-88% (g and h).
To synthesize the monophenyl class II PYM-HDACi compound, a Heck reaction was
performed on intermediate 6 with methyl acrylate 13.41 After the Boc protection to give
intermediate ester 15 which was subsequently converted to carboxylic acid 16 using sodium
4
2
hydroxide. The desired hydroxamic acid compound 18 was synthesized through the coupling of
carboxylic acid 16 and O-trityl hydroxylamine 17 followed by trityl deprotection (Scheme 2).40
O
N
NH2
N
NH2
N
N(Boc)2
O
N
N
13
N
a
55%
b, 56%
NH2
NH2
O
Br
N(Boc)2
15
O
6
14
O
O
N
N(Boc)2
N
NH2
NH OCPh
2
3
N
17
N
c
60%
d, e
N(Boc)2
NH2
HO
HOHN
24%
O
O
18
16
1
7

Scheme 2 . (a) Tri-O-tolyl phosphine, Pd(OAc) , TEA, DMF, 120 °C, overnight, 55%; (b) Boc O,
2
2
THF, DMAP, 45 °C, THF, 56%; (c) NaOH, H O, Dioxane, 20 °C, 12 h, 60%; (d) EDCI, HOBT,
2
DCM, rt, 6 h; (e) TFA, DCM, rt, 2 h, 24% (d and e).
To synthesize the biphenyl class II PYM-HDACi compound, a Suzuki coupling reaction
of Boc protected intermediate 7 with 4-hydroxyphenyl boronic acid furnished compound 19
which was converted to triflate compound 20 using standard protocol. Suzuki coupling between
compound 20 and potassium vinyl trifluoroborate furnished aryl vinyl compound 21. Cross
metathesis reaction of 21 with N-(trityloxy)acrylamide using Hoveyda-Grubbs 2nd generation
4
3
catalyst afforded compound 22 which upon treatment with TFA and TIPS furnished the desired
class II compound 23 (Scheme 3).
N
N(Boc)2
N
N(Boc)2
N
N(Boc)2
a
b
N
N
c
N
N(Boc)2
N(Boc)2
Br
N(Boc)2
7
HO
19
TfO
20
N
N(Boc)2
N
N(Boc)2
N
NH2
N
d
N
e
N
N(Boc)2
N(Boc)2
NH2
H
H
21
N
N
TrtO
22
HO
23
O
O
o
Scheme 3. (a) 4-hydroxyphenyl boronic acid, Pd(PPh ) , Cs CO , 80 C, 4 h, 81%; (b)
3
4
2
3
o
Trifluoromethanesulfonic anhydride, pyridine, -20 C, 1 h, 77%; (c) Pd(PPh ) , Potassium vinyl
3
4
o
trifluoroborate, Cs CO , DMF, 80 C, 4 h, 90%; (d) Hoveyda-Grubbs 2nd generation catalyst, N-
2
3
o
[
Tris(hydroxymethyl)-methyl]acrylamide, DCM, 33 C, overnight, 33%; (e) TFA, TIPS, rt, 2 h,
8
0%.
1
8

HDAC inhibition study
PYM-HDACi compounds were tested against HDAC isoforms 1, 6 and 8. These
compounds inhibited the HDAC isoforms tested with IC s ranging from low nanomolar to
5
0
micromolar. Specifically, class I compounds 12a-c broadly inhibited HDACs 1 and 6 but are less
potent against HDAC 8 (Table 1). Within this class, there is a linker length dependency in HDAC
1
1
6
and 6 inhibition potency which optimal for compound 12b. The monophenyl class II compound
8 is a relatively weaker HDACi which displayed the strongest inhibitory effect towards HDAC
. The inclusion of an additional phenyl ring, however, broadened and enhanced potency as the
biphenyl class II compound 23 inhibited both HDAC 6 (Table 1). This HDAC inhibition pattern
is in agreement with the predictions from the in silico docking study shown in Figures 5 and 6.
a
Table 1: HDAC inhibition activities of PYM-HDACi compounds (IC in µM).
5
0
Compound
2a
HDAC1
HDAC6
0.046
HDAC8
1
1
1
1
2
0.26
0.045
0.21
2.2
2.8
2b
2c
8
0.017
0.021
0.40
0.78
NTb
1.8
NT
3
3.7
0.073
0.034
SAHA
0.042
2.8
a
b
Performed through contractual agreement with BPS Bioscience. NT: Not Tested.
1
9

Anti-proliferative activity
The PYM-HDACi compounds were tested against three transformed and one normal cell
lines with SAHA, an FDA approved HDACi, as a positive control. The chosen transformed cell
lines were lung (A549), ER-positive (MCF-7) and TNBC (MDA-MB-231) breast cancer cell lines,
while monkey kidney epithelial cell (VERO) was selected as the nontransformed cell line. Our
choice of the transformed cell lines is informed by the STAT3 pathway dependency of these cell
lines. TNBCs, which account for 20% of all breast cancer incidence, lack Estrogen Receptor (ER),
4
4
Human Epidermal Growth Factor receptor 2 (HER2), and Progesterone Receptor (PR). TNBCs
are characterized by high metastasis, chemo-resistance, and poor prognosis with a lower five-year
4
5
survival rate relative to all the other non-TNBCs. Currently, there are no efficient targeted
treatment options for TNBC. Therefore, identification of new drug candidates for TNBC is
urgently warranted.
STAT3 plays a critical role in TNBCs, as it regulates several genes vital to cell survival,
4
6
metastasis, and invasiveness. Constitutive activation of STAT3 in MDA-MB-231 cells
promotes cell survival by regulating the expression of Bcl-2, Bcl-xL, Survivin, cyclin D1, c-
4
7
Myc, and Mcl-1. Conversely, A549 and MCF-7 cell lines have very low levels of constitutively
active STAT3.48-53Due to their combined effect on inhibition of STAT3 pathway, PYM-HDACi
compounds are expected to be more cytotoxic to STAT3-dependent cells such as MDA-MB-231
while somewhat less toxic to cell lines, such as A549 and MCF-7, with low levels of
constitutively active STAT3.
We observed that the two classes of the PYM-HDACi compounds have distinct effects on
the viability of the cell lines tested. Class I compounds 12a-c showed preferential cytotoxicity to
2
0

the STAT3-dependent MDA-MB-231 cells with linker-length-dependent potency that also closely
tracks with their HDAC inhibition potency. Specifically, compound 12a is only cytotoxic to MDA-
MB-231 with an IC of 38.38 ± 1.0 M. Inaddition to being more potent than 12a against MDA-
5
0
MB-231, compounds 12b and 12c are also cytotoxic toward A549 and MCF-7 cells. However,
2b and 12c displayed >3-5-fold selectivity for MDA-MB-231 relative to the other tested cell lines
Table 2). By testing the intermediate ester 12d, we subsequently confirmed that the modification
1
(
that we introduced did not abolish the independent antiproliferative effect of PYM. In fact,
compound 12d is 2.5-4.6-fold more potent than PYM against the two cell lines for which PYM
IC s were measurable within the concentration range we used (Table 2).
5
0
Presumably, due to its poor HDAC inhibitory activity and/or high hydrophilicity (CLogP
0.877) which may negatively impact cell penetration, compound 18 did not show anti-
=
proliferative activity towards all the cell lines tested. The additional phenyl ring in compound 23,
which resulted in the enhancement of its HDAC6 inhibition activity relative to 18 and higher
hydrophobicity (CLogP = 2.765), also results in it having broad antiproliferative activities against
the tested cell lines. However, 23 displayed little or no cell line selectivity (Table 2) which suggests
that its antiproliferative activity may be largely due to HDAC inhibition. Although all the PYM-
HDACi compounds are less potent relative to SAHA, the control HDACi, the selectivity of the
class I compounds 12a-c toward STAT3-dependent MDA-MB-231 suggest the contribution of the
inhibition of STAT3 pathway to the anti-proliferative activities of these compounds since most
non-targeted HDACi are incapable of tumor cells selectivity, a cause of their off-target toxic
2
0
effects. Interestingly, a combination therapy experiment whereby we used fixed concentration of
of PYM (100 M, approx ½ IC against MDA-MB-231 cells) and varying the concetrations of
5
0
SAHA showed only slight to moderate improvements in the potency, relative to SAHA as a
standalone agent, against all cell lines tested, including the non-transformed Vero cells. This
combination did not result in MDA-MB-231 cell-selectivity that we noticed in the designed
mutiple ligands PYM-HDACi 12b and 12c.
2
1

Table 2. Anti-proliferative activity of PYM-HDACi compounds (IC in µM).
5
0
Compound
A549
NI
MCF-7
NI
57.33 ± 3.4
83.34 ± 15.5
96.6 ± 6.85
NI
MDA-MB-231
38.38± 1.0
12.20 ± 2.2
21.4 ± 3.7
94.93 ± 9.67
NI
VERO
ND
40.25 ± 3.58
NI
112.65 ± 8.15
ND
1
1
1
1
1
2
2a
2b
2c
2d
8
65.49 ± 4.6
88.46 ±10.5
113.75 ± 3.95
NI
10.00 ± 1.92
NI
3
9.3 ± 1.6
453
5.43 ± 1.2
238
6.9 ± 0.4
NI
PYM
SAHA
PYM + SAHA
15.73 ± 0.99
10.19 ± 0.87
3.49 ± 0.05
2.29 ± 0.46
3.40 ± 0.20
1.865 ± 0.48
4.465 ± 0.68
1.305 ± 0.12
*
Each value is obtained from a duplicate of three simultaneous experiments. NI: No Inhibition. ND: Not Determined
at a maximum concentration of 100 µM. Comb. means SAHA with variable concentration combined with 100M
PYM
Intracellular Target Validation
To determine the contributions of HDAC and STAT3 pathway inhibition to the antiproliferative
activities of PYM-HDACi compounds, we used immunoblotting to investigate the MDA-MB-
2
31 cells response to one of the lead compounds 12c using SAHA and PYM as positive controls
for HDAC and STAT3 inhibition respectively. For HDAC inhibition, we monitored histone 4
H4) and tubulin acetylation states as markers for HDACs 1 and 6 intracellular inhibition
(
activities, respectively.54,55 GAPDH expression was used as a protein loading control. When
exposed to the cells at ½-IC , IC and 2x-IC , 12c induced accumulation of acetylated H4 and
5
0
50
50
acetylated tubulin. Similarly, SAHA at 1.5 induced H4 and tubulin hyperacetylation (Figure
). PYM has no effect on the acetylation status of tubulin and H4. This result strongly suggests
7
that the PYM-HDACi 12c inhibits these HDACs intracellularly, an attribute which contributes to
their antiproliferative activity.
2
2

a.
b.
DMSO PYM
12c
DMSO SAHA
5
0 100 5
5



Ac-Tubulin
GAPDH
Ac-H4
AC-tubulin
AC-H4
1000
800
*
*
**
8
6
4
2
00
00
00
00
0
**
*
** *** ***
6
4
2
00
00
00
0
*
**
*
*
*
*
*
l
l
o
M
M
M
M
M
M
M
M
o
M
M
M
M
M
M
M
M
r
µ
µ
.5µ
µ
2
.5µ
µ
µ
r
t
µ
µ
µ
µ
2
µ
µ
µ
t
5
.
5
0
n
0
0
3
5
0
n
0
0
.
5
3
5
0
2
5
o
5
0
1
2
2
5
o
1
A
2
1
A
1
1
c
c
C
M
A
H
1
c
c
C
M
A
H
2
2
c
2
2
Y
H
A
c
1
1
Y
H
A
1
1
P
Y
A
S
P
Y
A
S
1
P
1
P
S
S
Figure 7. The Western blot analysis. (a) Immunoblotting of the acetylation status of tubulin and
H4 of MDA-MB-231 treated with DMSO, PYM, SAHA, and 12c for 4 h. Cells were serum-
starved 24 h prior to the treatment. Acetylated tubulin and acetylated H4 are upregulated by
SAHA and 12c but not PYM. (b) Quantification of acetylated tubulin and acetylated H4
obtained by averaging data from two independent experiments. (Bars show mean plus standard
deviation; * P < 0.05; ** P < 0.0021;***P<0.0002)
To elucidate the effect of the PYM-HDACi compounds on the STAT3 pathway, we
probed the effects of 12c, SAHA and PYM on the intracellular expression of STAT3, pSTAT3,
p38, p-p38, Bcl-2, and Bcl-xL. Bcl-2 and Bcl-xL are anti-apoptotic genes whose expressions are
upregulated by constitutive activation of STAT3.56-58 Conversely, inhibition of STAT3
upregulates the intracellular level of p-p38.59-61 As stated earlier, HDACi inhibits STAT3
pathway through direct downregulation of pSTAT3. The exact mechanism of the inhibition of
2
3

the STAT3 pathway by PYM has not been fully elucidated. However, it has been recently shown
that PYM is a direct inhibitor of STAT3 transcriptional activity as PYM could upregulate
pSTAT3 cellular levels while downregulating the expression of pSTAT3 target genes, including
6
2
th
Bcl-2 and Bcl-xL, in TNBC cell lines. We observed that 12c (at 1/4 -, ½- and 1x-IC ) and
5
0
th
SAHA (at 1/4 - and ½-IC ) caused concentration-dependent upregulation of p-p38 and
5
0
downregulation of Bcl-xL in MDA-MB-231 cells. Within this concentration range, 12c slightly
downregulates Bcl-2 levels while SAHA caused no statistically relevant changes to the
intracellular levels of Bcl-2. PYM (at 50 µM and 100 µM) has a similar effect as SAHA, causing
concentration-dependent upregulation of p-p38 and Bcl-xL and no effect on Bcl-2 levels (Figures
8
a and 8b). Interestingly, the effects of 12c and PYM on pSTAT3 levels are closely aligned as
they both caused upregulation of pSTAT3 while SAHA cause a slight downregulation of
pSTAT3. The observed PYM-induced upregulation of pSTAT3 is in agreement with a previous
6
2
observation. To further confirm this effect on STAT3 pathway, we investigated the effects of
2c, SAHA and PYM on the intracellular levels of the cyclin D1, a downstream protein of
1
4
7
STAT3 pathway. We observed that cyclin D1is significantly downregulated with 12c, SAHA
and PYM at approx. ½ - and 1x-IC concentration (Figures 9a and 9b). Collectively, this data
5
0
suggests that the intracellular inhibitions of HDACs and the STAT3 pathway contribute to the
anti-proliferative activity of the PYM-HDACi compounds 12c. However, the mechanism of
STAT3 pathway inhibition of 12c is distinct from that of prototypical HDACi and very similar to
that of STAT3 inhibitor template PYM. This distinction could be the basis for the TNBC cell
selectivity of the PYM-HDACi.
a.
DMSO
PYM
12c
SAHA
5
0µM 100µM
5µM 12.5µM 25µM
0.8µM 1.6µM
2
4

p-STAT3 (Tyr705)
STAT3
p-p38
p-38
GAPDH
Bcl-xL
Bcl-2
b.
p-p38
p-STAT3
**
800
2
1
1
00
50
00
***
***
*
**
6
00
*
400
*
*
*
**
2
00
0
5
0
0
l
rol
o
M
M
M
M
M
M
M
M
M
M
M
M
M
M
u
r
u
u
.8u
.6u
u
.5u
u
t
u
u
u
u
u
u
t
0
0
8
6
5
5
5
n
0
0
5
5
n
.
.
.
5
0
2
o
5
0
0
1
c
2
2
o
0
1
c
2
1
1
2
1
c
C
M
2
1
c
C
M
A
A
2
A
A
1
2
M
1
c
Y
M
H
H
c
1
Y
H
H
1
Y
2
P
Y
2
P
A
A
1
A
A
1
P
P
S
S
S
S
Bcl-2
Bcl-xL
1
1
50
00
1
1
50
00
*
**
***
****
5
0
5
0
0
0
rol
l
M
M
M
M
M
M
M
o
M
M
M
M
M
M
M
r
u
u
.8u
.6u
u
.5u
u
t
u
u
u
u
u
u
u
t
0
0
8
6
5
5
5
n
0
0
5
5
n
.
.
.
5
0
2
o
5
0
0
1
c
2
2
o
0
1
c
2
1
1
2
1
c
C
M
2
1
c
C
M
A
A
A
A
1
2
1
2
M
c
Y
M
H
H
c
1
Y
H
H
1
Y
2
P
Y
2
P
A
A
1
A
A
1
P
P
S
S
S
S
2
5

Figure 8. (a) Western blot analysis of the effects of compounds on the STAT3 pathway in
MDA-MB-231 cells. Cells were serum-starved for 24 h prior to treatment with the tested agents
such that the final DMSO content in the media is 0.1%. Cells were treated with the tested agents
for 24 h. Images show the effects of DMSO, PYM, 12c and SAHA on the intracellular levels of
selected STAT3 pathway markers. (b) Quantification of the Western blots data probing for the
effects of PYM, SAHA, and 12c on the STAT3 pathway in MDA-MB-231 cells. (Bars show
mean plus standard deviation; * P < 0.05; ** P < 0.0021; ***P<0.0002).
a.
Control
DMSO
SAHA
2.5M
12c
12.5M
Control
PYM
100M 200M
5M
5M
25M
Cyclin D1
Actin
b.
Cyclin D1
1
1
50
00
*
**
***
5
0
0
**
****
l
o
M
M
1
M
c
M
2
M
M
M
M
r







t
5
5
5
5
.
5
0
0
n
.
2
0
0
o
2
A
c
2
1
2
C
H
2
1
c
A
M
H
A
1
2
Y
Y
A
S
1
P
P
S
Figure 9. (a) Western blot analysis and (b) quantification of the Western blots data revealed the
effects of 12c on the cyclin D1 expression MDA-MB-231 cells. Experimental conditions are the
same for the data shown in Figure 8. (Bars show mean plus standard deviation; * P < 0.05; ** P
<
0.0021;***P<0.0002; ****P<0.00001).
2
6

Flow Cytometry for Cell Cycle Analysis.
We then performed cell cycle analysis to determine the effects of PYM-HDACi 12c, 12b and
SAHA on MDA-MB-231 cell cycle distribution. We observed that the effects of 12c (12.5 and
2
5 M), 12b (15M) and SAHA (5 M) on cell cycle are very similar as they induced
significant G2 phase arrest (Figure 10 and Figure S2). Previous studies have shown that SAHA
6
3
caused G2 arrest when exposed to breast cancer cells at concentrations above 3.0 μM. The G2
cell cycle arrest induced by these compounds suggests that in addition to their cytotoxicity
effects, they also induce MDA-MB-231 cell apoptosis.
A.
B.
2
7

C.
D.
Figure 10. Effect of (a) DMSO control, (b) 12c (12.5M), (c) 12c (25M), and (d) SAHA (5M)
on MDA-MB-231 cell cycle progression.
2
8

Conclusion
We disclosed in this study two classes of dual-acting compounds designed to inhibit HDACs and
the STAT3 pathway. We observed that the class I PYM-HDACi compounds 12a-c potently inhibit
HDACs 1 and 6 in cell-free assays and are preferentially cytotoxic to MDA-MB-231, a TNBC cell
line that is highly STAT3-dependent. Moreover, target validation studies revealed that a
representative compound 12c elicited a signature of HDAC and STAT3 pathway inhibition
intracellularly. In addition, 12b and 12c show significant selective cell cytotoxicity to TNBC.
Overall, these compounds show promise as leads to develop targeted therapy for TNBC.
Experimental section
Materials and methods
4
-Bromophenyl acetonitrile, ethyl propionate, O-tritylhydroxylamine, methyl acrylate, were
purchased from Sigma-Aldrich. Trimethylsilylacetylene was purchased from Alfa Aesar. All
commercially available starting materials were used without purification. Reaction solvents were
high performance liquid chromatography (HPLC) grade or American Chemical Society (ACS)
grade and used without purification. Analtech silica gel plates (60 F254) were used for analytical
TLC, and Analtech preparative TLC plates (UV 254, 2000 µm) were used for purification. UV
light and anisaldehyde/iodine stain were used to visualize the spots. 200-400 Mesh silica gel was
used in column chromatography. Nuclear magnetic resonance (NMR) spectra were recorded on a
1
Varian-Gemini 400 MHz, Bruker 500 MHz or 700 MHz magnetic resonance spectrometer. H
NMR Spectra were recorded in parts per million (ppm) relative to the residual peaks of CHCl3
(
7.24 ppm) in CDCl or CHD OD (4.78 ppm) in CD OD or DMSO-d (2.49 ppm) in DMSO-d .
3 2 3 5 6
1
3
C spectra were recorded relative to the central peak of the CDCl triplet (77.0 ppm) or CD OD
3
3
2
9