Targeting breast cancer stem cells by novel HDAC3-selective inhibitors

Article abstract of DOI:10.1016/j.ejmech.2017.08.069

Although histone deacetylase (HDAC) inhibitors have been known to suppress the cancer stem cell (CSC) population in multiple types of cancer cells, it remains unclear which HDAC isoforms and corresponding mechanisms contribute to this anti-CSC activity. Pursuant to our previous finding that HDAC8 regulates CSCs in triple-negative breast cancer (TNBC) cells by targeting Notch1 stability, we investigated related pathways and found HDAC3 to be mechanistically linked to CSC homeostasis by increasing β-catenin expression through the Akt/GSK3β pathway. Accordingly, we used a pan-HDAC inhibitor, AR-42 (1), as a scaffold to develop HDAC3-selective inhibitors, obtaining the proof-of-concept with 18 and 28. These two derivatives exhibited high potency and isoform selectivity in HDAC3 inhibition. Equally important, they showed in vitro and/or in vivo efficacy in suppressing the CSC subpopulation of TNBC cells via the downregulation of β-catenin.

Full text of DOI:10.1016/j.ejmech.2017.08.069

Accepted Manuscript
Targeting breast cancer stem cells by novel HDAC3-selective inhibitors
Hao-Yu Hsieh, Hsiao-Ching Chuang, Fang-Hsiu Shen, Kinjal Detroja, Ling-Wei Hsin,
Ching-Shih Chen
PII:
S0223-5234(17)30680-3
10.1016/j.ejmech.2017.08.069
EJMECH 9713
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 3 July 2017
Revised Date: 30 August 2017
Accepted Date: 30 August 2017
Please cite this article as: H.-Y. Hsieh, H.-C. Chuang, F.-H. Shen, K. Detroja, L.-W. Hsin, C.-S. Chen,
Targeting breast cancer stem cells by novel HDAC3-selective inhibitors, European Journal of Medicinal
Chemistry (2017), doi: 10.1016/j.ejmech.2017.08.069.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT
1
2
Targeting Breast Cancer Stem Cells by Novel HDAC3-Selective Inhibitors
3
Hao-Yu Hsieh,1,2,3† Hsiao-Ching Chuang,2† Fang-Hsiu Shen,3 Kinjal Detroja,3 Ling-
Wei Hsin,1* and Ching-Shih Chen*2,3
4
5
1School of Pharmacy, College of Medicine, National Taiwan University, Taipei,
Taiwan; 2Division of Medicinal Chemistry and Pharmacognosy, College of Pharmacy,
6
3
7
The Ohio State University, Columbus, OH, USA; Institute of Biological Chemistry,
8
Academia Sinica, Taipei, Taiwan;
9
(†Equal contributions; *Co-corresponding authors)
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
1

ACCEPTED MANUSCRIPT
1
2
3
4
5
6
7
8
9
Abstract
Although histone deacetylase (HDAC) inhibitors have been known to suppress the
cancer stem cell (CSC) population in multiple types of cancer cells, it remains unclear
which HDAC isoforms and corresponding mechanisms contribute to this anti-CSC
activity. Pursuant to our previous finding that HDAC8 regulates CSCs in triple-
negative breast cancer (TNBC) cells by targeting Notch1 stability, we investigated
related pathways and found HDAC3 to be mechanistically linked to CSC homeostasis
by increasing β-catenin expression through the Akt/GSK3β pathway. Accordingly,
we used a pan-HDAC inhibitor, AR-42 (1), as a scaffold to develop HDAC3-selective
10 inhibitors, obtaining the proof-of-concept with 18 and 28. These two derivatives
11 exhibited high potency and isoform selectivity in HDAC3 inhibition. Equally
12 important, they showed in vitro and/or in vivo efficacy in suppressing the CSC
13 subpopulation of TNBC cells via the downregulation of β-catenin.
14
15 Keywords
16 Histone deacetylase 3 (HDAC3), cancer stem cell (CSC), triple-negative breast cancer
17 (TNBC), β-catenin
18
19 INTRODUCTION
20
The concept of cancer stem cells (CSCs) or tumor-initiating cells has provided a
21 new paradigm for the functional heterogeneity of cancer cells [1, 2]. Because of their
22 tumorigenic properties and capacity for self-renewal and differentiation, this small
23 subpopulation of cancer cells drives initiation, progression, and metastasis in many
24 types of tumors [3]. Moreover, recent evidence suggests that epithelial-to-
25 mesenchymal transition acts as a critical regulator of the drug-resistant phenotype of
2

ACCEPTED MANUSCRIPT
1
2
3
4
5
6
7
8
9
CSCs [4, 5]. Consequently, CSCs are more resistant to chemotherapeutic agents than
the bulk population of non-CSCs within a tumor, allowing the surviving CSCs to
repopulate the tumor and leading to tumor relapse. From a therapeutic perspective,
there is an urgent unmet need for CSC-targeting therapeutic agents to achieve optimal
patient outcomes. To date, a series of therapeutic agents targeting key signaling
pathways, especially those mediated by Notch1, Hedgehog, and Wnt, have proved to
be efficacious in eradicating CSCs in preclinical settings [6-8]. In addition, there is
accumulating evidence of the in vitro and/or in vivo efficacy of pan- and class I
HDAC inhibitors, such as suppressing the CSC subpopulation in different cancer cell
10 lines [9]. However, two issues remain unclear with respect to the anti-CSC activity of
11 HDAC inhibitors. First, as there are 11 Zn2+-dependent isoforms in the HDAC
12 family, it is unclear which of these isoforms contribute to the regulation of CSCs.
13 Second, the mechanism by which pan- or class I HDAC inhibitors suppress the CSC
14 subpopulation has not been clearly defined.
15
Previously, we reported a non-epigenetic function of HDAC8 in activating CSC-
16 like properties in triple-negative breast cancer (TNBC) cells by upregulating the
17 expression of Notch1 [10]. We demonstrated that HDAC8 protected Notch1 from
18 Fbwx7-facilitated protein degradation, thereby leading to the accumulation of Notch1
19 and the Notch intracellular domain NICD. Nevertheless, this finding could not
20 account for the ability of the class I HDAC inhibitor MS-275 to effectively block
21 mammosphere formation in breast cancer cell lines [10]. Because MS-275 is deficient
22 in HDAC8 inhibitory activity, this suggested the involvement of another HDAC
23 isoform, namely, HDAC3. In the present study, we obtained evidence that HDAC3
24 plays a pivotal role in regulating CSCs through a distinct mechanism, i.e., to increase
25 β-catenin protein stability by activating the Akt/glycogen synthase kinase (GSK)3β
3

ACCEPTED MANUSCRIPT
1
2
3
4
5
6
7
8
9
signaling pathway.
Based on this biological finding, we embarked on developing HDAC3-selective
inhibitors as anti-CSC agents via structural modifications of AR-42 [formerly, (S)-
HDAC42; 1]. We had a three-fold rationale for employing this hydroxamate-based
pan-HDAC inhibitor developed in our laboratory [11, 12]. First, AR-42 shows high
potency in ablating cancer stem cells in acute myelogenous leukemia [13] and triple-
negative breast cancer (TNBC) [10].
pharmacokinetic properties in animal models [14].
bioavailability of AR-42 was estimated to be 26% and 100% in mice and rats,
Second, AR-42 exhibits favorable
For example, the oral
10 respectively. Third, AR-42 has recently completed a phase I trial in patients with
11 multiple myeloma and T-cell lymphoma, which showed that oral AR-42 was well
12 tolerated with no apparent dose-limiting toxicities [15]. From a translational
13 perspective, AR-42-derived HDAC3 inhibitors might retain drug-like properties with
14 lesser toxicity. Here, we report the successful conversion of AR-42 into a series of
15 HDAC3-selective inhibitors, represented by 18 and 28, by replacing the hydroxamate
16 moiety with o-aminoanilides as the Zn2+-chelating motif, followed by altering the
17 structure of the cap group. Compounds 18 and 28 exhibited high potency and isoform
18 selectivity toward HDAC3, and equally important, were able to mimic the suppressive
19 effect of HDAC3 depletion on CSCs in vitro and/or in vivo through the inhibition of
20 the Akt/β-catenin signaling pathway. β-Catenin, a downstream effector of Wnt
21 signaling, has been shown to play a critical role in the maintenance of the CSC
22 population in many types of cancer [16, 17], including TNBC [18]. This regulation is
23 mediated, in part, through the transcriptional regulation of target genes associated
24 with tumor initiation and cell renewal, including c-Myc [19, 20] and BMI-1 [19], by
25 interacting with a complex network of binding partners [16, 17].
4

ACCEPTED MANUSCRIPT
1
2
3
4
5
6
7
8
9
In this paper, we first describe our biological finding on the role of HDAC3 in
regulating the CSC subpopulation in TNBC cells, followed by the structural
modification of AR-42 to develop HDAC3-selective inhibitors and the evaluation of
their in vitro and/or in vivo efficacies in suppressing CSCs.
RESULTS
Evidence that HDAC3 is involved in regulating breast CSCs via the Akt-GSK3β-
β-catenin signaling pathway. To shed light onto the functional role of HDAC3, we
used two different shRNAs (#4993 and #6267) to deplete HDAC3 in two TNBC cell
10 lines, MDA-MB-231 and SUM-159 (Figure 1A). We isolated two stable clones from
11 each treated cell type to assess the consequent effects, vis-à-vis parental cells, on
12 different cellular functions, including cell proliferation, colony formation, and
13 mammosphere formation, a surrogate measure of CSC expansion [21]. Stable
14 knockdown of HDAC3 via these two shRNAs gave rise to consistent results in both
15 cell lines (MDA-MB-231, Figure 1B; SUM-159, Figure 1C). As shown, HDAC3
16 silencing led to decreases in cell proliferation rate (left), clonogenic survival (center),
17 and mammosphere formation (right) as compared to control.
18
19
20
21
<Insert Figure 1>
To understand the mechanism by which HDAC3 regulates CSCs, we analyzed
22 various signaling pathways in both cell lines by Western blot analysis after stable
23 HDAC3 knockdown. Among the different pathways examined, the ability of HDAC3
24 depletion to reduce β-catenin expression via the inhibition of the Akt-GSK3β
25 signaling axis was especially noteworthy. As shown, ablation of HDAC3 led to
5

ACCEPTED MANUSCRIPT
1
2
3
reduced phosphorylation of Akt and GSK3β, accompanied by parallel changes in the
expression of β-catenin in both cell lines (Figure 1A). Moreover, the suppressive
effect of HDAC3 depletion on β-catenin expression was corroborated by parallel
4
5
decreases in β-catenin target gene products associated with the maintenance of breast
CSCs, including c-Myc [19, 20] and BMI-1 [19]. This finding is consistent with
recent reports that blocking β-catenin signaling is effective in inhibiting breast CSCs
[18, 19, 22]. Conversely, enforced expression of HDAC3 in MDA-MB-231 cells
resulted in the activation of Akt-GSK3β signaling, leading to the accumulation of β-
catenin and the aforementioned CSC regulators (Figure 1D).
6
7
8
9
10
As CSCs are characterized by their capacity to form tumors from low cell
11 numbers, we assessed the effect of HDAC3 depletion on tumor-initiation using MDA-
12 MB-231 cells. Female NOD/SCID mice were injected bilaterally in the mammary fat
13 pads with 50,000 MDA-MB-231 cells stably expressing HDAC3 shRNA (HDAC3KD;
14 Figure 1E) on one side and an equal number of control MDA-MB-231 cells on the
15 opposite side (N = 9). As shown, HDAC3 depletion suppressed breast tumorigenicity
16 in vivo. For example, at six days post-injection, 90% of the mice had tumors
17 generated from the control cells, while only 40% of the mice had HDAC3KD tumors
18 (P < 0.025; log-rank test). Moreover, the HDAC3KD tumors grew at a slower rate
19 than the control tumors (P = 0.0015).
20
Together, these data supported the function of HDAC3 in regulating breast CSCs,
21 in part via the Akt-GSK3β-β-catenin signaling pathway. Involvement in this pathway
22 provided a mechanistic basis to develop HDAC3-selective inhibitors as anti-CSC
23 agents. Accordingly, we embarked on the development of HDAC3-selective
24 inhibitors by using compound 1 as a starting point according to the strategy delineated
25 in the next section.
6

ACCEPTED MANUSCRIPT
1
2
Use of the pan-HDAC inhibitor 1 (AR-42) as a scaffold to develop class I- and
HDAC-3-selective inhibitors. There is substantial evidence that the Zn2+-chelating
motif is critical for determining the selectivity of HDAC inhibitors [23, 24]. For
example, compound 1 and most other hydroxamate-based HDAC inhibitors, such as
SAHA, TSA, and LBH589, are broad-spectrum inhibitors, while many class I HDAC
inhibitors (MS-275, CI994, and mocetinostat) and HDAC3-selective inhibitors (BG-
45 and RGFP966) contain N-(2-aminophenyl)-benzamide and o-aminoanilide as their
Zn2+-chelating motifs, respectively (Figure 2A).
3
4
5
6
7
8
9
10
11
12
13
<Insert Figure 2>
We hypothesized that the isoform selectivity of the aforementioned class I
14 and HDAC3-selective inhibitors might be attributable to the bulkiness of the
15 respective Zn2+-chelating motifs relative to the hydroxamate moiety, which hinders
16 their access to the catalytic Zn2+-binding domain of non-class I HDAC isoforms.
17 Based on this reasoning, we replaced the –NHOH moiety of 1 with o-aminoaniline,
18 yielding 2. In addition, we prepared 3, which is the (R)-enantiomeric isomer of 2, to
19 examine potential stereochemical effects on the potency/selectivity in inhibiting
20 different HDAC isoforms. General procedures for the syntheses of 2, 3, and
21 subsequent new derivatives are depicted in Scheme 1.
22
23
24
<Insert Scheme 1>
7

ACCEPTED MANUSCRIPT
1
2
3
4
5
6
7
8
9
Compounds 2 and 3, each at 1 µM, were subjected to HDAC isoform-profiling
analysis (HDAC1-8) by a commercial vendor (Reaction Biology Cooperation,
Malvern, PA). The isoform profiling data revealed that, compared to compound 1,
this substitution of the Zn2+-chelating motif resulted in decreased, yet more selective,
inhibitory activities against the various HDAC isoforms examined (Table 1).
<Insert Table 1>
For example, while 1 inhibited HDAC1-3, HDAC6, and HDAC8 with high potencies
10 (data provided by Arno Therapeutics), 2 and 3 showed substantially lower activities
11 relative to 1 in inhibiting HDAC6 and HDAC8. Of note, 3 displayed higher potency
12 than 2 toward HDAC1-3, indicating the stereochemical preference of the (R)-α-
13 isopropyl substituent in binding to the catalytic Zn2+-binding domain of these HDAC
14 isoforms. In addition to the above in vitro deacetylase assays, we measured three
15 cell-based biomarkers by Western blot, including histone H3 acetylation (both pan-
16 and H3K9-acetylation), tubulin acetylation [25], and Notch1 expression [10], to
17 monitor the inhibitory activities of individual derivatives against histone
18 deacetylation, HDAC6, and HDAC8, in MDA-MB-231 cells. As shown, treatment of
19 cells with 1, 2, or 3 at the indicated concentrations increased histone H3 pan- and
20 H3K9 acetylation in a consistent manner, demonstrating the inhibition of histone
21 deacetylase activity. Compound 1 was highly potent in inhibiting HDAC6 and
22 HDAC8; at 0.25 µM it facilitated tubulin acetylation and decreased the expression of
23 Notch1 (Figure 3).
24
25
<Insert Figure 3>
8

ACCEPTED MANUSCRIPT
1
2
3
4
5
6
7
8
9
Consistent with the low in vitro HDAC6 and HDAC8 inhibitory activities, 2 and
3 effected no apparent changes in the levels of tubulin acetylation and Notch1
expression, with the exception of 3 at 2.5 µM against Notch1. Throughout the course
of this structural modification, we used these two biomarkers to measure HDAC6-
and HDAC8-inhibitory activities of individual derivatives.
Previously, it has been reported that fluorine-containing benzamides as the Zn2+-
chelating motif could enhance HDAC3 selectivity [26, 27]. Consistent with this
finding, the isoform selectivity of 2 and 3 could be further improved by adding a
10 fluorine atom at the para-position of the aminoaniline moiety, resulting in 4 and 5,
11 respectively. As shown, 5 was deficient in HDAC6- and HDAC8-inhibitory activities
12 (<10% inhibition at 1 µM; Table 1), further confirmed by the lack of appreciable
13 changes in tubulin acetylation and Notch1 expression, respectively (Figure 3). In
14 particular, 5 exhibited a slight preference for inhibiting HDAC3 over HDAC1 and
15 HDAC2 [Table 1; IC50, 0.75, 1.2 and 2 µM, respectively; determined by Reaction
16 Biology Cooperation (Figure 2B)]. However, substituting the fluorine atom of 5 with
17 CF3 completely abolished the HDAC inhibitory activity of the resulting compound 6
18 (Table 1), consistent with the lack of increase in histone H3 acetylation in drug-
19 treated cells (Figure 3). This loss of HDAC inhibitory activity might be associated
20 with the bulkiness of the CF3 group, which restricted ligand access to the active site
21 pocket. From a structural perspective, 5 provided a proof-of-concept that HDAC3-
22 selective inhibitors could be developed from a pan inhibitor by replacing hydroxamate
23 with N-(2-amino-4-fluorophenyl)carboxamide as the Zn2+-chelating motif. We
24 therefore conducted further structural modifications of 5 to augment its HDAC3
9

ACCEPTED MANUSCRIPT
1
2
3
4
5
6
7
8
9
inhibitory potency and isoform selectivity by altering the structure of the cap group
and/or linker.
It is noteworthy that substitution of the α-substituent of the cap group of 5, i.e.,
(R)-isopropyl, with either (S)- or (R)-ethyl (7 and 9, respectively) or methyl (8 and 10,
respectively) resulted in multifold increases in the potency in HDAC3 inhibition
(Figure 2B; IC50, 5, 1 µM; 7, 0.17 µM; 8, 0.23 µM; 9, 0.11 µM; 10, 0.21 µM).
Equally important, this increase in HDAC3 inhibitory activity was also accompanied
by improved isoform selectivity. There were multifold increases in the selectivity
ratio, defined as the ratio of IC50 of HDAC1 to that of HDAC3 (IC50HDAC1/IC50HDAC3),
10 of individual compounds (5, 1.2; 7, 14.7; 8, 13; 9, 10.9; 10, 8.6).
11 To assess the effect of altering the linker on the potency and/or selectivity, we
12 replaced the 4-aminobenzoyl linker of the above three enantiomeric pairs of
13 compounds (4 and 5, 7 and 9, and 8 and 10) with a heterocyclic linker, 5-
14 aminopicolinoyl, yielding the corresponding optically active derivatives 11-16. This
15 modification led to increases in HDAC3 inhibitory potencies in 11, 13, 14, and 15.
16 Compound 14, in particular, had a nearly tenfold increase in activity relative to 5
17 (IC50, 0.08 µM versus 1 µM). 14 also showed improved isoform selectivity with a
18 selectivity ratio of 6.5. In contrast, 12 and 16 had a modest loss of HDAC3-inhibitory
19 activity, but an increased HDAC1-inhibitory activity, as compared to 7 and 10.
20 Interestingly, replacing the 5-aminopicolinoyl linker of 14 with an isomeric 6-
21 aminonicotinoyl moiety to produce compound 17 resulted in a substantial loss of
22 activities against not only HDAC3 (IC50, 1.7 µM), but also HDAC1 (> 10 µM) and
23 HDAC2 (8.4 µM) (Figure 2B). Together, these data suggest subtle interplay between
24 the linker and α-substituent in interacting with the binding pockets of different HDAC
25 isoforms. Because there were no apparent changes in tubulin acetylation and Notch 1
10

ACCEPTED MANUSCRIPT
1
2
3
4
5
6
7
8
9
expression, it was evident that compounds 11 and 13-16 lacked inhibitory activities
toward HDAC6 and HDAC8 (Figure 3).
Pursuant to the above findings, we further used achiral p-substituted 2-
phenylacetyl groups (R = H, F, and CF3) as the cap group in conjunction with either
4-aminobenzoyl (to generate derivatives 18-20) or 5-aminopicolinoyl (to generate
derivatives 21-23). Of these six derivatives, 18 exhibited the highest HDAC3-
inhibitory potency (IC50, 0.18 µM) as well as isoform selectivity of HDAC3 over
HDAC1 (2.3 µM) and HDAC2 (3.2 µM). The modifications to generate 19-23 did
not improve potency or isoform selectivity relative to 18, and in the case of 23, even
10 abolished the HDAC-inhibitory activity (IC50 > 10 µM for all three isoforms). This
11 intricate structure-activity relationship (SAR) was also illustrated by a switch in
12 isoform selectivity from HDAC3 to HDAC1 when the F atom was removed from the
13 Zn2+-chelating motif of 20 to generate 24 (Table 1). The structural basis of this
14 selectivity switch is as yet unclear. Moreover, replacement of the phenylacetyl group
15 of 18 and 21 with a 3-phenylpropionyl moiety decreased and abrogated the HDAC3-
16 inhibitory potency in the resulting compounds 25 (IC50, 0.8 µM) and 26 (no
17 appreciable activity at 10 µM), respectively. We proceeded to synthesize a series of
18 derivatives with conformationally restricted cap groups, 27 - 31, among which 28 was
19 notable for its high HDAC3-inhibitory potency (IC50, 0.35 µM) and isoform
20 selectivity (Table 1; Figure 2B). Relative to 28, 27 exhibited lower HDAC3-
21 inhibitory potency and isoform selectivity, which reiterates the integral role of the F
22 atom on the Zn2+-chelating motif in interacting with the catalytic pockets of HDAC3
23 versus other HDAC isoforms.
24
11

ACCEPTED MANUSCRIPT
1
2
In vitro and/or in vivo efficacy of 18 and 28 in suppressing breast CSCs. Based on
the above SAR, we evaluated 18 and 28 for their in vitro and/or in vivo efficacies in
inhibiting breast CSCs. Compounds 18 and 28 exhibited only a modest suppressive
effect on cell viability (IC50 ≥ 10 µM) by MTT assays (Figure 4A, left), which was
also confirmed by the growth inhibition assay using crystal violet staining (right; IC50
≥ 10 µM). In contrast, both compounds were highly effective in blocking colony
formation and mammosphere formation in suspension medium (IC50, 0.5 – 1 µM)
(Figure 4B).
3
4
5
6
7
8
9
10
11
12
<Insert Figure 4>
This discriminative effect suggests the ability of these two HDAC3 inhibitors to
13 selectively target the CSC subpopulation in MDA-MB-231 cells. Consistent with
14 effects of HDAC3 depletion, treating MDA-MB-231 cells in a monolayer culture with
15 18 or 28 in the presence of 5% FBS led to concurrent downregulation of Akt
16 phosphorylation and β-catenin expression, accompanied by increases in histone H3
17 acetylation (Figure 4C). Moreover, RT-PCR analysis revealed no changes in mRNA
18 expression, indicating that this downregulation of β-catenin was mediated at the
19 protein level and suggesting that this decrease in β-catenin expression was attributable
20 to protein degradation associated with drug-induced Akt deactivation.
21
We next obtained two lines of evidence verifying the inhibitory effect of 28 on
22 breast CSCs. First, the CD44+/CD24low breast cancer population is known to be
23 enriched for breast cancer initiating stem cells [28]. Flow cytometric analysis
24 demonstrated the concentration-dependent, suppressive effect of 28 on this
25 subpopulation of MDA-MB-231 cells (Figure 4D). Second, we tested the in vivo
12

ACCEPTED MANUSCRIPT
1
2
3
4
5
6
7
8
9
efficacy of 28 in suppressing tumor initiation by MDA-MB-231 cells in NOD/SCID
mice. After pretreating with 5 µM 28 or DMSO vehicle for 48 h, 50,000 MDA-MB-
231 cells were injected into the mammary fat pads of female NOD/SCID mice (N =
10 for the control and drug-treated groups, respectively). Mice were treated with 28
at 100 mg/kg or vehicle by oral gavage, starting day 2 post-injection and continuing
through day 12. Consistent with HDAC3 depletion effects, treatment with 28
exhibited a suppressive effect on the tumorigenicity of MDA-MB-231 cells relative to
vehicle control, though the P value of 0.053 bordered on being statistically
significant. In addition, the drug-treated tumors grew at a slower rate than the control
10 tumors (P = 0.0038) after 26 days of treatment.
11
12
13 Discussion
14
Although the potential of using HDAC inhibitors to eradicate the CSC
15 subpopulation has been demonstrated in multiple types of cancer [9], the underlying
16 mechanism has not been fully elucidated. We previously demonstrated that HDAC8 is
17 involved in regulating TNBC CSCs by maintaining the expression of Notch1 via
18 protein stabilization [10]. Following this connection, we investigated the mechanistic
19 link between HDAC3 and CSC homeostasis in TNBC cells. We obtained evidence
20 that the suppressive effect of genetic depletion or pharmacological inhibition of
21 HDAC3 on CSCs might be attributable to the inhibition of the Akt-GSK3β β-catenin
22 signaling axis. It is well recognized that the activation status of Akt is controlled by a
23 myriad of signaling effectors, including upstream kinases, phosphatases, and negative
24 regulators [29]. Considering the importance of Akt in mediating cellular functions, the
25 mechanism by which HDAC3 regulates Akt activation represents an intriguing
13

ACCEPTED MANUSCRIPT
1
2
3
4
5
6
7
8
9
therapeutic target. We speculate that the link between HDAC3 and Akt activation
might not mediated through a general chromatin modeling mechanism based on the
following rationale. From a mechanistic perspective, HDAC3 may mediate diverse
biological functions in a cell context-specific manner via two distinct mechanisms
[30]. First, HDAC3 forms complexes with the corepressors SMRT and N-CoR to
facilitate the transcriptional repression of genes, especially those encoding nuclear
receptors. Second, HDAC3 can also regulate transcriptional activity of various
transcription factors through deacetylation, including RelA, SRY, p53, and
CBP/p300. In principle, HDAC3 may regulate the expression of an upstream
10 regulator of Akt phosphorylation via one of these two mechanisms, which is currently
11 under investigation.
12
13 types of cancer, including that of breast, liver, lung, skin, and pancreas
14 (www.proteinatlas.org/ENSG00000171720-HDAC3/cancer), suggesting its
According to the Human Protein Atlas, HDAC3 is overexpressed in multiple
15 involvement in tumorigenesis. The present study further demonstrated that HDAC3
16 represents a potential anti-CSC target. Accordingly, we used the pan-HDAC inhibitor
17 1 previously developed in our laboratory as a starting point for synthesizing HDAC3-
18 selective inhibitors. From a drug development perspective, this strategy exploits the
19 advantages of retaining the drug-like properties of the parent molecule in the course
20 of structural modifications. The proof-of-concept of this strategy was in the efficacy
21 of the two HDAC3-selective inhibitors 18 and 28, both of which showed high
22 HDAC3-inhibitory potency and isoform selectivity. Based on the previous reports
23 that use of F-containing benzamides as the Zn2+-chelating motif could enhance
24 HDAC3 selectivity [26, 27], a key element of this “pan-to-HDAC3-selective
25 inhibitor” conversion was the replacement of the hydroxamate moiety with N-(2-
14

ACCEPTED MANUSCRIPT
1
2
3
4
5
6
7
8
9
amino-4-fluorophenyl)carboxamide, which resulted in the dissociation of HDAC6-
and HDAC8-inhibitory activities in resulting derivatives. Comparing the isoform
selectivity between the compound pair of 27 versus 28 clearly indicates that the
importance of the fluorine atom at the Zn2+-binding motif in improving ligand
recognition at the HDAC3 binding pocket. Moreover, optimal isoform selectivity
toward HDAC3 was achieved by varying the α-substituent on the 2-phenylacetyl
moiety, revealing the importance of this cap group in determining ligand interactions
with HDAC isoforms. In contrast, replacing the 4-aminobenzoyl linker of 5 and 9
with a 5-aminopicolinoyl substructure was not beneficial. Although the resulting
10 compounds 14 and 15 had improved HDAC3-inhibitory activity (IC50, 80 nM and 60
11 nM, respectively), this change also gave rise to increased HDAC1- and HDAC2-
12 inhibitory activities, therefore compromising the isoform selectivity.
13
The suppressive effect of 18 and 28 on CSCs was confirmed by their inhibition of
14 clonogenic survival, mammosphere formation, and/or the CD44+/CD24low
15 subpopulation of MDA-MB-231 cells (Figure 4B). Consistent with the HDAC3
16 knockdown data, this anti-CSC effect correlated with the ability of 18 and 28 to
17 inhibit Akt signaling, thereby downregulating the protein expression of β-catenin
18 (Figure 4C). Equally important, oral 28 at 100 mg/kg was able to mimic the
19 suppressive effect of HDAC3 depletion on breast tumorigenicity in vivo in the MDA-
20 MB-231 animal model. Together, these data provide a proof-of-concept for the
21 translational potential of HDAC3 inhibitors in combination with chemotherapeutic
22 agents to target the CSC subpopulation in cancer treatment.
23
24 CONCLUSION
25
Based on our finding that HDAC3 regulates TNBC CSCs by targeting the Akt-
15

ACCEPTED MANUSCRIPT
1
2
3
4
5
6
7
8
9
GSK3β-β-catenin pathway, we used the pan HDAC inhibitor 1 as a scaffold to
develop HDAC3-selective inhibitors by replacing hydroxamate with the N-(2-amino-
4-fluorophenyl)carboxamide as the Zn2+-chelating motif, obtaining the proof-of-
concept with 18 and 28. These two derivatives exhibited high potency and isoform
selectivity in HDAC3 inhibition, and equally important, showed in vitro efficacy in
suppressing the CSC subpopulation of TNBC cells via the downregulation of β-
catenin. Moreover, oral 28 mimicked the suppressive effect of HDAC3 knockdown
on in vivo tumorigenesis in nude mice. These findings have simulated further
evaluation of 28 in enhancing the in vivo efficacy of different chemotherapeutic
10 agents and the use of 28 as a lead compound to develop more potent and selective
11 HDAC3 inhibitors. Based on the aforementioned SAR analysis, we will focus on
12 modifying the cap group of 28 by replacing the α,α,-dimethyl-phenylacetyl moiety
13 with rigid, non-aromatic functional groups to improve the HDAC3-inhibitory
14 potency, which is currently underway.
15
16 EXPERIMENTAL SECTION
17
All commercially available reagents were used without further purification unless
18 otherwise stated. Anhydrous THF was obtained by distilling commercial reagent over
19 sodium, and anhydrous DMF was commercially available. Silica gel for column
20 chromatography was purchased from Merck Silica gel 60 (0.040 – 0.063 mm).
1
13
21 Routine H and C nuclear magnetic resonance spectra were recorded on the Bruker
22 NMR AV 400 or Bruker AVII 500 NMR. Samples were dissolved in deuterated
23 chloroform (CDCl3) or dimethyl sulfoxide (DMSO-d6), and tetramethylsilane (TMS)
24 was used as a reference. Mass spectrometry analyses were performed with a Waters,
25 LCT orthogonal acceleration time-of-flight (oa-TOF) LC/MS system. All compounds
16

ACCEPTED MANUSCRIPT
1
1
2
for bioassays were identified with H NMR, 13C NMR and HRMS, and purities
confirmed to be higher than 95%. Purities of all tested compounds were determined
by a Shimadzu LCMS-2020 system (comprised of a LC-20AD pump, a SPD-M30A
detector, an SIL-20AC autosampler and a Shim-pack GIST C18 2µm, 100 X 2.1 mm
I.D. C18 column). The general procedures for the synthesis of compounds 2-31 are
depicted in Scheme 1, and detailed synthesis is described in the Supplementary
Information. HDAC isoform profiling and IC50 determination of individual
compounds were conducted by Reaction Biology Cooperation (Malvern, PA).
3
4
5
6
7
8
9
10
Cell Lines, Cell Culture, Reagents, and Antibodies. The breast cancer cell
11 lines, MDA-MB-231 and SUM-159, were purchased from American Type Culture
12 Collection (Manassas, VA) and Asterand Biosciences (Detroit, MI), respectively.
13 MDA-MB-231 cells were maintained in DMEM medium containing 10% fetal bovine
14 serum, and SUM159 cells were cultured in Ham’s F-12 (Life Technologies)
15 supplemented with 5% fetal bovine serum, 5 µg/ml insulin, and 1 µg/ml
16 hydrocortisone. All cells were cultured at 37°C in a humidified incubator containing
17 5% CO2. AR-42 (1) was a kind gift from Arno Therapeutics (Flemington, NJ).
18 Antibodies for various proteins were purchased from following sources: HDAC3, β-
19 catenin, Histone H3, tubulin, acetyl-tubulin and β-actin (Santa Cruz, CA); p-Ser473-
20 Akt, AKT, p-Ser9-GSK3β, GSK3β, c-Myc, BMI-1, and Ac-H3K9 (Cell Signaling
21 Technology; Beverly, MA); pan-Ac-H3 (Millipore; Billerica, MA); APC-conjugated
22 CD44 and PE-conjugated CD24, (Thermo Fisher Scientific, Waltham, MA);
23 horseradish peroxidase-conjugated anti-mouse IgG or anti-rabbit IgG (Jackson
24 ImmunoResearch Laboratories; West Grove, PA).
25
17

ACCEPTED MANUSCRIPT
1
2
3
4
5
6
7
8
9
Generation of HDAC3-depleted stable clones or HDAC3-overexpressing
MDA-MB-231 cells. For lentivirus generation, psPAX2 (#12260, Addgene),
pMD2.G (#12259, Addgene) and shRNA of HDAC3 (TRCN0000194993,
TRCN0000196267; RNAi Core of Academia Sinica, Taipei, Taiwan) were
cotransfected in 293T cells with Lipofectamine 2000 (Life Technologies) according
to the manufacturers’ instructions. Viral particles were used for infection of target
cells and stable clones were maintained with puromycin (Life Technologies; Grand
Island, NY). pLAS.Void plasmid served as negative control. The HDAC3 expression
plasmid (#13819, Addgene; Cambridge, MA) was transfected with Lipofectamine
10 2000 in MDA-MB-231 cells according to the manufacturers’ instructions.
11
12
Cell Lysis and Immunoblotting. Drug-treated cells were collected and then
13 lysed in a buffer containing 1% sodium dodecyl sulfate (SDS), 10 mM EDTA, 50
14 mM Tris–HCl (pH 8.1), and 1% protease and phosphatase inhibitor mixture. Lysates
15 were sonicated and then centrifuged at 13,000 rpm for 10 min. Protein concentrations
16 in the supernatants were determined (Micro BCA Protein Assay Kit, Pierce
17 Biotechnology, Rockford, IL) and equal amounts of proteins were resolved via SDS-
18 PAGE and transferred to a nitrocellulose membrane (GE Healthcare Life Sciences,
19 Pittsburgh, PA). The membrane was washed twice with Tris-buffered saline
20 containing 0.1% Tween-20 (TBST), blocked with TBST containing 5% non-fat milk
21 for 30 min, and then incubated with primary antibody (1:1000 dilution) in TBST at
22 4°C overnight. After washing with TBST, the membrane was incubated with goat
23 anti-rabbit or anti-mouse IgG–HRP conjugates (1:5000 dilution) for 1 h at room
24 temperature. The immunoblots were visualized by enhanced chemiluminescence.
25
18

ACCEPTED MANUSCRIPT
1
2
3
4
5
6
7
8
9
RNA Extraction and RT-PCR. Total RNA was isolated with TRIzol (Thermo
Fisher Scientific) according to the manufacturer's protocol. From each sample, 2 µg
total RNA was reverse-transcribed into cDNA using the iScriptTM cDNA Synthesis
Kit (Bio-Rad; Hercules, CA) and the cDNA were separated by electrophoresis. PCR
products were resolved by electrophoresis in 2% agarose gels and visualized by
ethidium bromide staining. The sequences of primers used for RT-PCR were as
follows. β-catenin, forward primer: GCTGATTTGATGGAGTTGGA; reverse primer:
GCTACTTGTTCTTGAGTGAA;
β-actin,
forward
primer:
GCTCGTCGTGGACAACGGCTC, reverse primer: CAAACATGATCTGGGTCAT-
10 CTTCTC.
11
12
Colony Formation Assays. MDA-MB-231 or SUM-159 cells were seeded in
13 six-well plates at a density of 1,000 cells per well and were left to attach overnight.
14 Vehicle control (DMSO) or increasing concentrations of test agents were added to the
15 cells in the presence of 5% FBS for 48 h. The drug mixture was washed out, and the
16 plates were incubated with media for 7-14 days until colonies were visible. Each drug
17 concentration was assessed in triplicate. The colonies were fixed with 4%
18 formaldehyde (Sigma-Aldrich) and stained with crystal violet (5 mg/ml in 2%
19 ethanol, Sigma-Aldrich). Colonies containing more than 50 cells were counted. Cell
20 survival is expressed as a percentage and was determined from the numbers of
21 colonies present in the drug-treated groups relative to the vehicle-treated control
22 group.
23
24
Mammosphere formation assays. MDA-MB-231 (1,000 cells/well) and
25 SUM159 (500 cells/well) cells were seeded onto ultra-low attachment 24-well flat
19

ACCEPTED MANUSCRIPT
1
2
3
4
5
6
7
8
9
bottom plates (Corning) in serum-free culture medium (MammoCult™, STEMCELL
Technologies; Vancouver, Canada) supplemented with 1 µM hydrocortisone
(MammoCult™) and 2 µg/ml heparin (MammoCult™). Cells were then treated with
28 at the indicated concentrations in triplicate for 7 days. The number of
mammospheres was counted at 100X magnification.
3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium
bromide
(MTT)
Assays. To determine the drug effects on cell viability, MDA-MB-231 cells were
seeded onto 96-well plates at a density of 5,000 cells per well in the presence of 10%
10 FBS. After overnight incubation, cells were exposed to test agents in the presence of
11 5% FBS for 24, 48, and 72 h. After treatment, cells were incubated with MTT
12 (Biomatik, Wilmington, DE) for an additional 1 h. The medium was then removed
13 from each well and replaced with DMSO to dissolve the reduced MTT dye for
14 subsequent colorimetric measurement of absorbance at 560 nm. Cell viabilities are
15 expressed as percentages of viable cells relative to the corresponding vehicle-treated
16 control group. For the cell proliferation analysis of HDAC3KD stable clones versus
17 parental MDA-MB-231 cells, a total of 500 cells per well were seeded in 96-well
18 plates and incubated for the indicated time at 37°C. The cell proliferation was
19 analyzed by detecting the absorbance of reduced MTT dye.
20
21
Growth Inhibition Assay. Inhibition of cell growth was determined by a
22 modified crystal violet assay [31]. MDA-MB-231 cells were plated in 96-well culture
23 plates (5000 cells/well). After overnight incubation, cells were treated with individual
24 test agents at the indicated concentrations in the presence of 5% FBS for 24, 48 and
25 72hrs. After treatment, cells were washed with PBS and stained with crystal violet
20

ACCEPTED MANUSCRIPT
1
2
3
4
5
solution (0.5% crystal violet in 20% methanol). After 20 min, the wells were washed,
dried and the stain was solubilized with 2% SDS. Absorbance at 550nM was
measured by a plate reader. The percentage of growth inhibition was determined
using the equation (1-Nt/Nc) x 100, where Nt and Nc are the absorbencies of stain in
treated and vehicle control wells, respectively.
6
7
Flow cytometry. DMSO- or 28-treated MDA-MB-231 cells were harvested,
washed with stain buffer (BD, Franklin Lakes, NJ), and incubated with specific FITC-
or phycoerythrin-conjugated monoclonal antibodies for CD44 and CD24 on ice in the
8
9
10 dark for 30 min. Cells were washed with stain buffer, and the pellets were collected
11 and suspended in PBS. The CD44+/CD24low subpopulation was analyzed using a
12 FACSCalibur (BD Biosciences) flow cytometer.
13
14
In Vivo Tumorigenesis Study. Female NOD/SCID mice (NOD.CB17-
15 Prkdcscid /NCrHsd; 5–6 weeks of age; Harlan, Indianapolis, IN) were group-housed
16 under constant photoperiod (12-h light/12-h dark) with ad libitum access to sterilized
17 food and water. All experimental procedures were done according to protocols
18 approved by The Ohio State University Institutional Animal Care and Use
19 Committee. To assess the effect of HDAC3 knockdown on tumor initiation in vivo,
20 stable HDAC3KD clone #6267 cells [50,000 cells/0.1 ml in 50% Matrigel (BD
21 Biosciences)] were implanted into the 4th inguinal mammary fat pads of NOD/SCID
22 mice. An equal number of MDA-MB-231 cells transfected with control lentiviral
23 constructs were injected into the contralateral mammary fat pads to serve as negative
24 control. Tumors were measured with calipers and the volumes were calculated using
25 V = (width2 x length) x 0.52. To assess the effect of 28 on tumor initiation in vivo,
26 MDA-MB-231 cells were pretreated with 5 µM 28 for 48 h, and control cells were
21

ACCEPTED MANUSCRIPT
1
2
3
4
5
6
7
8
9
treated with the same amount of DMSO. Treated cells were then collected by trypsin
(Gibco), the cell density was adjusted to 50,000 cells/0.1ml in 50% Matrigel, and
implanted into the left 4th inguinal mammary fat pads of NOD/SCID mice. The day
after cell implantation, mice were treated with 28 at 100 mg/kg body weight or
vehicle (0.5% methylcellulose/0.1% Tween 80 [v/v] in sterile water) once daily by
oral gavage.
Statistical Analysis. In vitro experiments were performed at least three times
and data are presented as means ± SD. Group means were compared using one-way
10 ANOVA followed by Student’s t tests. For the in vivo experiments, differences in
11 tumor incidence and tumor volume were analyzed by log-rank test and Student’s t-
12 test, respectively. Differences were considered significant at P < 0.05.
13
14 ASSOCIATED CONTENT
15 Supporting Information
16
17
18
19
20
I. Synthetic procedures of 2-31 and purity of individual derivatives as
determined by HPLC
II. NMR and HRMS spectra, and HPLC chromatograms
III. Images of all Western blots in full from Figure 3
21 AUTHOR INFORMATION
22 Co-corresponding Authors
23 Professor Ching-Shih Chen, E-mail: chencs@gate.sinica.edu.tw or
24 chen.844@osu.edu; Professor Ling-Wei Hsin, E-mail: lwhsin@ntu.edu.tw
25 Author Contributions
22

ACCEPTED MANUSCRIPT
1
2
3
4
5
6
7
8
9
The manuscript was written through contributions from all authors. All authors have
given approval of the final version of the manuscript. H.Y.H. and H.C.C. contributed
equally to this manuscript.
Notes
CSC is the inventor of AR-42, which was licensed to Arno Therapeutics by the Ohio
State University Foundation for clinical development. He has received payments
associated with the licensing of this agent according to the University guidelines. The
other authors declare no competing financial interest.
10 ACKNOWLEDGMENTS
11 This work was supported by grant # MOST 105-2321-B-001-064 from the Team of
12 Excellent Research Program of the Ministry of Science and Technology (Taiwan), the
13 National Health Research Institutes (Taiwan) grant NHRI-EX105-10521BI, and
14 Lucius A. Wing Endowed Chair Fund from The Ohio State University Medical
15 Center. We thank Dr. Cindy Lee at the Institute of Biological Chemistry at Academia
16 Sinica for the scientific editing of this manuscript. We thank Dr. Xiaokui Mo for the
17 statistical analysis of the difference in tumor incidence.
18
19 ABBREVIATIONS USED
20 CSC, cancer stem cells; TNBC, triple-negative breast cancer; HDAC, histone
21 deacetylase; GSK3β, glycogen synthase kinase 3β; MTT, 3-(4,5-dimethylthiazol-2-
22 yl)-2,5-diphenyltetrazolium bromide; SAR, structure-activity relationship
23
24 REFRENCES
25 [1] A. Pietras, Cancer stem cells in tumor heterogeneity, Adv Cancer Res, 112
26
(2011) 255-281.
23

ACCEPTED MANUSCRIPT
1
2
3
[2] A.K. Singh, R.K. Arya, S. Maheshwari, A. Singh, S. Meena, P. Pandey, O.
Dormond, D. Datta, Tumor heterogeneity and cancer stem cell paradigm:
updates in concept, controversies and clinical relevance, Int J Cancer, 136
(2015) 1991-2000.
4
5
6
[3] H. Clevers, The cancer stem cell: premises, promises and challenges, Nat Med,
17 (2011) 313-319.
7
8
[4] T. Shibue, R.A. Weinberg, EMT, CSCs, and drug resistance: the mechanistic
link and clinical implications, Nat Rev Clin Oncol, (2017).
9
10
[5] F. Marcucci, G. Stassi, R. De Maria, Epithelial-mesenchymal transition: a new
target in anticancer drug discovery, Nat Rev Drug Discov, 15 (2016) 311-325.
11 [6] K. Chen, Y.H. Huang, J.L. Chen, Understanding and targeting cancer stem cells:
12
13
therapeutic implications and challenges, Acta Pharmacol Sin, 34 (2013) 732-
740.
14 [7] D.L. Dragu, L.G. Necula, C. Bleotu, C.C. Diaconu, M. Chivu-Economescu,
15
16
Therapies targeting cancer stem cells: Current trends and future challenges,
World J Stem Cells, 7 (2015) 1185-1201.
17 [8] N. Takebe, P.J. Harris, R.Q. Warren, S.P. Ivy, Targeting cancer stem cells by
18
19
inhibiting Wnt, Notch, and Hedgehog pathways, Nat Rev Clin Oncol, 8 (2011)
97-106.
20 [9] M. Dvorakova, T. Vanek, Histone deacetylase inhibitors for the treatment of
21 cancer stem cells, Med Chem Commun, 7 (2016) 2217-2231.
22 [10] M.W. Chao, P.C. Chu, H.C. Chuang, F.H. Shen, C.C. Chou, E.C. Hsu, L.E.
23
24
25
Himmel, H.L. Huang, H.J. Tu, S.K. Kulp, C.M. Teng, C.S. Chen, Non-
epigenetic function of HDAC8 in regulating breast cancer stem cells by
maintaining Notch1 protein stability, Oncotarget, 7 (2016) 1796-1807.
26 [11] Q. Lu, Y.T. Yang, C.S. Chen, M. Davis, J.C. Byrd, M.R. Etherton, A. Umar,
27
28
C.S. Chen, Zn2+-chelating motif-tethered short-chain fatty acids as a novel class
of histone deacetylase inhibitors, J Med Chem, 47 (2004) 467-474.
29 [12] Q. Lu, D.S. Wang, C.S. Chen, Y.D. Hu, C.S. Chen, Structure-based
30
31
optimization of phenylbutyrate-derived histone deacetylase inhibitors, J Med
Chem, 48 (2005) 5530-5535.
32 [13] M.L. Guzman, N. Yang, K.K. Sharma, M. Balys, C.A. Corbett, C.T. Jordan,
33
34
35
36
M.W. Becker, U. Steidl, O. Abdel-Wahab, R.L. Levine, G. Marcucci, G.J.
Roboz, D.C. Hassane, Selective activity of the histone deacetylase inhibitor AR-
42 against leukemia stem cells: a novel potential strategy in acute myelogenous
leukemia, Mol Cancer Ther, 13 (2014) 1979-1990.
37 [14] H. Cheng, Z. Xie, W.P. Jones, X.T. Wei, Z. Liu, D. Wang, S.K. Kulp, J. Wang,
38
39
40
41
C.C. Coss, C.S. Chen, G. Marcucci, R. Garzon, J.M. Covey, M.A. Phelps, K.K.
Chan, Preclinical Pharmacokinetics Study of R- and S-Enantiomers of the
Histone Deacetylase Inhibitor, AR-42 (NSC 731438), in Rodents, AAPS J, 18
(2016) 737-745.
42 [15] D.W. Sborov, A. Canella, E.M. Hade, X. Mo, S. Khountham, J. Wang, W. Ni,
43
44
45
46
47
M. Poi, C. Coss, Z. Liu, M.A. Phelps, A. Mortazavi, L. Andritsos, R.A.
Baiocchi, B.A. Christian, D.M. Benson, J. Flynn, P. Porcu, J.C. Byrd, F.
Pichiorri, C.C. Hofmeister, A phase 1 trial of the HDAC inhibitor AR-42 in
patients with multiple myeloma and T- and B-cell lymphomas, Leuk
Lymphoma, (2017) 1-9.
48 [16] W.H. Lien, E. Fuchs, Wnt some lose some: transcriptional governance of stem
49
cells by Wnt/beta-catenin signaling, Genes Dev, 28 (2014) 1517-1532.
24

ACCEPTED MANUSCRIPT
1
2
3
4
5
6
7
[17] T. Zhan, N. Rindtorff, M. Boutros, Wnt signaling in cancer, Oncogene, 36
(2017) 1461-1473.
[18] S.G. Pohl, N. Brook, M. Agostino, F. Arfuso, A.P. Kumar, A. Dharmarajan,
Wnt signaling in triple-negative breast cancer, Oncogenesis, 6 (2017) e310.
[19] J. Xu, J.R. Prosperi, N. Choudhury, O.I. Olopade, K.H. Goss, beta-Catenin is
required for the tumorigenic behavior of triple-negative breast cancer cells,
PLoS One, 10 (2015) e0117097.
8
9
10
[20] J. Xu, Y. Chen, D. Huo, A. Khramtsov, G. Khramtsova, C. Zhang, K.H. Goss,
O.I. Olopade, beta-catenin regulates c-Myc and CDKN1A expression in breast
cancer cells, Mol Carcinog, 55 (2016) 431-439.
11 [21] G. Dontu, W.M. Abdallah, J.M. Foley, K.W. Jackson, M.F. Clarke, M.J.
12
13
Kawamura, M.S. Wicha, In vitro propagation and transcriptional profiling of
human mammary stem/progenitor cells, Genes Dev, 17 (2003) 1253-1270.
14 [22] G.B. Jang, J.Y. Kim, S.D. Cho, K.S. Park, J.Y. Jung, H.Y. Lee, I.S. Hong, J.S.
15
16
Nam, Blockade of Wnt/beta-catenin signaling suppresses breast cancer
metastasis by inhibiting CSC-like phenotype, Sci Rep, 5 (2015) 12465.
17 [23] F.F. Wagner, U.M. Wesmall yi, M.C. Lewis, E.B. Holson, Small molecule
18
19
inhibitors of zinc-dependent histone deacetylases, Neurotherapeutics, 10 (2013)
589-604.
20 [24] F. Thaler, C. Mercurio, Towards selective inhibition of histone deacetylase
21
22
isoforms: what has been achieved, where we are and what will be next,
ChemMedChem, 9 (2014) 523-526.
23 [25] C. Hubbert, A. Guardiola, R. Shao, Y. Kawaguchi, A. Ito, A. Nixon, M.
24
25
Yoshida, X.F. Wang, T.P. Yao, HDAC6 is a microtubule-associated
deacetylase, Nature, 417 (2002) 455-458.
26 [26] H. Jia, J. Pallos, V. Jacques, A. Lau, B. Tang, A. Cooper, A. Syed, J. Purcell, Y.
27
28
29
30
31
Chen, S. Sharma, G.R. Sangrey, S.B. Darnell, H. Plasterer, G. Sadri-Vakili, J.M.
Gottesfeld, L.M. Thompson, J.R. Rusche, J.L. Marsh, E.A. Thomas, Histone
deacetylase (HDAC) inhibitors targeting HDAC3 and HDAC1 ameliorate
polyglutamine-elicited phenotypes in model systems of Huntington's disease,
Neurobiol Dis, 46 (2012) 351-361.
32 [27] X. Li, Y. Zhang, Y. Jiang, J. Wu, E.S. Inks, C.J. Chou, S. Gao, J. Hou, Q. Ding,
33
34
35
36
J. Li, X. Wang, Y. Huang, W. Xu, Selective HDAC inhibitors with potent oral
activity against leukemia and colorectal cancer: Design, structure-activity
relationship and anti-tumor activity study, Eur J Med Chem, 134 (2017) 185-
206.
37 [28] M. Al-Hajj, M.S. Wicha, A. Benito-Hernandez, S.J. Morrison, M.F. Clarke,
38
39
Prospective identification of tumorigenic breast cancer cells, Proc Natl Acad Sci
U S A, 100 (2003) 3983-3988.
40 [29] Y. Liao, M.C. Hung, Physiological regulation of Akt activity and stability, Am J
41 Transl Res, 2 (2010) 19-42.
42 [30] P. Karagianni, J. Wong, HDAC3: taking the SMRT-N-CoRrect road to
43 repression, Oncogene, 26 (2007) 5439-5449.
44 [31] M. Feoktistova, P. Geserick, M. Leverkus, Crystal Violet Assay for Determining
45
46
47
Viability of Cultured Cells, Cold Spring Harb Protoc, 2016 (2016) pdb
prot087379.
48
25

ACCEPTED MANUSCRIPT
1
2
3
Scheme 1. General Synthetic Procedures for 2- 31.
4
5
6
7
8
9
Figure Legends
Figure 1. Evidence that HDAC3 plays a role in regulating the CSC subpopulation in
TNBC cells. (A) Western blot analysis of the effect of the stable depletion of
HDAC3 by two different shRNAs (#4994 and #6267) versus control shRNA
treatment (Ctl) on the expression and/or phosphorylation levels of HDAC3, Akt,
GSK3β, and β-catenin and its target gene products c-Myc and BMI-1 in MDA-MB-
10 231 and SUM-159 cells. (B & C) Functional analyses of the effect of the stable
11 depletion of HDAC3 by two different shRNAs versus control on cell proliferation
12 (left; N = 6), colony formation (center; N = 6), and mammosphere formation (right; N
13 = 6) in (B) MDA-MB-231 and (C) SUM-159 cells. (D) Western blot analysis of the
14 effect of the ectopic expression of HDAC3 versus control on the expression and/or
15 phosphorylation levels of HDAC3, Akt, GSK3β, and β-catenin and its target gene
16 products c-Myc and BMI-1 in MDA-MB-231. (E) Effect of stable HDAC3
17 knockdown (HDAC3KD) versus control shRNA on (left) tumor-initiating ability by
18 monitoring tumor incidence, and (right) xenograft tumor growth, measured at Day 27
19 after implantation, of MDA-MB-231 cells. Data are expressed as means ± S.D. (N =
20 9).
21
22 Figure 2. (A) Structures of 1 (AR-42) versus various class I HDAC inhibitors and
23 HDAC3-selective inhibitors. (B) Structure of 2-31 and the respective activities in
24 inhibiting the deacetylase activities of HDAC1, HDAC2, and HDAC3 (HDAC1/2/3),
25 which were expressed in two ways: a, % inhibition at 1 µM; b, IC50 values.
26
26

ACCEPTED MANUSCRIPT
1
2
3
4
5
6
7
8
9
Figure 3. Western blot analyses of the effects of representative derivatives at
indicated concentrations on histone H3 pan and H3K9 acetylation, tubulin acetylation,
and Notch 1 expression in MDA-MB-231 cells after 48 h of treatment, in which 1
(0.25 µM) was used as a positive control in each blot. Images of some of these blots,
including those depicting 1 versus 3, 1 versus 10 and 16, 1 versus 15 and 13, and 1
versus 18, 24, 20, 27, and 28, were not contiguous because lanes of repeated samples
were cropped out from the blots. Images of these full blots are shown in
Supplemental Information.
10 Figure 4. The in vitro and/or in vivo efficacy of 18 and 28 in suppressing the CSC
11 subpopulation of MDA-MB-231 cells, in part through the downregulation of β-
12 catenin expression. (A) Concentration- and time-dependent effects of 18 and 28 on
13 cell viability by MTT assays (left) and growth inhibition by crystal violet staining
14 (right). Value, means + S.D. (N = 6). (B) Concentration-dependent suppressive
15 effects of 18 and 28 on (upper) colony formation and (lower) mammosphere
16 formation. Value, means + S.D. (N = 6). (C) Upper, Western blot analysis of
17 concentration-dependent suppressive effects of 18 and 28 on Akt phosphorylation, β-
18 catenin expression, and histone H3 pan and H3K9 acetylation. Lower, RT-PCR
19 analysis of concentration-dependent effects of 28 on β-catenin mRNA expression. (D)
20 Flow cytometric analysis of concentration-dependent suppressive effects of 28 on the
21 CD44+/CD24low subpopulation. (E) Effect of daily 28 at 100 mg/kg via oral gavage
22 (N = 10) versus vehicle control (N = 10) on (left) tumor-initiating ability by
23 monitoring tumor incidence, and (right) xenograft tumor growth, measured at Day 27
24 after implantation of MDA-MB-231 cells. Data are expressed as means ± S.D.
25
27

ACCEPTED MANUSCRIPT
Table 1. HDAC isoform inhibition profiles of 1 and representative derivatives, each
at 1 µM.
HDAC isoforms; % Inhibition at 1 µM*
HDAC1 HDAC2 HDAC3 HDAC4 HDAC5 HDAC6 HDAC7 HDAC8
1
2
95%
55%
80%
42%
6%
72%
31%
57%
26%
4%
93%
30%
76%
78%
0%
21%
0%
9%
0%
3%
0%
0%
6%
0%
0%
0%
0%
9%
0%
0%
0%
0%
0%
0%
3%
100%
20%
16%
0%
22%
1%
9%
0%
0%
0%
0%
0%
0%
0%
87%
7%
3
38%
6%
5
6
0%
0%
20
24
27
28
29
41%
78%
49%
8%
32%
38%
34%
5%
36%
33%
36%
87%
82%
0%
5%
0%
0%
4%
0%
4%
0%
0%
9%
0%
14%
* Each value represents the average of two determinations