Design, Synthesis, and Preclinical Evaluation of Fused Pyrimidine-Based Hydroxamates for the Treatment of Hepatocellular Carcinoma

Article abstract of DOI:10.1021/acs.jmedchem.7b01465

Class I histone deacetylases (HDACs) are highly expressed and/or upregulated in hepatocellular carcinoma (HCC) and are associated with aggressiveness, spread, and increased mortality of HCC. Activation of phosphatidylinositol 3-kinase-Akt-mammalian target of rapamycin (PI3K/Akt/mTOR) signaling pathway was involved in the development of HCC and acquired resistance to sorafenib. A series of purine or 5H-pyrrolo[3,2-d]pyrimidine based hydroxamates were designed and developed as multitarget drugs to modulate both HDACs and the PI3K/Akt/mTOR pathway. Among 39 cell lines screened, the molecules (e.g., 20e, 20f, and 20q) were the most selective against leukemia, lymphoma, and HCC cells; they also demonstrated target modulation in cancer cell lines and in mice bearing MV4-11 and HepG2 tumors. Compound 20f in particular showed significant single agent oral efficacy in hypervascular liver cancer models (e.g., HepG2, HuH-7, and Hep3B) and was well-tolerated. These encouraging results, along with its favorable target profile and tissue distribution, warrant further development of 20f.

Full text of DOI:10.1021/acs.jmedchem.7b01465

Subscriber access provided by READING UNIV
Article
Design, Synthesis and Preclinical Evaluation of Fused Pyrimidine-
Based Hydroxamates for the Treatment of Hepatocellular Carcinoma
Dizhong Chen, Chang Kai Soh, Wei Huang Goh, and Haishan Wang
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.7b01465 • Publication Date (Web): 23 Jan 2018
Downloaded from http://pubs.acs.org on January 23, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 79
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Design, Synthesis and Preclinical Evaluation of
Fused PyrimidineꢀBased Hydroxamates for the
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Treatment of Hepatocellular Carcinoma
Dizhong Chen,† Chang Kai Soh,† Wei Huang Goh,† and Haishan Wang*,†,‡
†Drug Development Unit, Institute of Molecular and Cell Biology, Agency for Science,
Technology and Research, 61 Biopolis Drive, Proteos, Singapore 138673, Republic of Singapore
‡Probit Pharmaceuticals Pte. Ltd., 10 Anson Road #26ꢀ04, Singapore 079903, Republic of
Singapore
ABSTRACT
Class I histone deacetylases (HDACs) are highly expressed and/or upregulated in hepatocellular
carcinoma (HCC) and are associated with aggressiveness, spread, and increased mortality of
HCC. Activation of phosphatidylinositol 3ꢀkinaseꢀAktꢀmammalian target of rapamycin
(PI3K/Akt/mTOR) signaling pathway was involved in the development of HCC and acquired
resistance to sorafenib. A series of purine or 5Hꢀpyrrolo[3,2ꢀd]pyrimidine based hydroxamates
were designed and developed as multiꢀtarget drugs to modulate both HDACs and the
PI3K/Akt/mTOR pathway. Among 39 cell lines screened, the molecules (e.g., 20e, 20f, and 20q)
were the most selective against leukemia, lymphoma, and HCC cells; they also demonstrated
target modulation in cancer cell lines and in mice bearing MV4ꢀ11 and HepG2 tumors.
Compound 20f in particular showed significant single agent oral efficacy in hypervascular liver
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 79
1
2
3
4
5
6
7
8
cancer models (e.g., HepG2, HuHꢀ7, and Hep3B) and was wellꢀtolerated. These encouraging
results, along with its favorable target profile and tissue distribution, warrant further
development of 20f.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
KEYWORDS. HDAC inhibitor, PI3K inhibitor, PI3K/Akt/mTOR pathway, multiꢀtarget, purine,
fusedꢀpyrimidine, hydroxamic acid, anticancer, xenograft, efficacy, hepatocellular carcinoma,
liver cancer, leukemia, nonꢀHodgkin lymphomas.
Introduction
Hepatocellular carcinoma (HCC) is the third most fatal cancer worldwide with sorafenib being
the first and only approved systemic therapy for advanced HCC since 2007.1, 2 Only recently in
April 2017 was regorafenib, a fluorinated version of sorafenib, approved to treat HCC patients
who fail sorafenib treatment (sorafenib resistance).3, 4 However there is currently no established
systemic therapy for HCC patients who tolerate neither sorafenib nor regorafenib, or fail
regorafenib therapy.5 Class I histone deacetylases (HDACs) are highly expressed and/or
upregulated in hepatocellular carcinoma (HCC) and are associated with aggressiveness, spread,
and increased mortality of HCC.6ꢀ9 Drug resistance to sorafenib involves pathways such as
PI3K/Akt and JAK/STAT, and PI3K/Akt/mTOR signaling pathway activation has been found in
a subset of HCC tumor samples.10ꢀ13 The above data thus suggest that both HDACs and
PI3K/Akt/mTOR are relevant targets for the treatment of advanced HCC.
HDACs are validated oncology targets and there are already approved HDAC inhibitors for
the treatment of lymphoma and multiple myeloma (Figure 1), with new indications, including
HCC, being explored and further validated in clinics.14 In preclinical studies, both HDAC
ACS Paragon Plus Environment

Page 3 of 79
Journal of Medicinal Chemistry
1
2
3
4
inhibitors
vorinostat
(Figure
1)
and
Nꢀhydroxyꢀ4ꢀ[[(2S)ꢀ3ꢀmethylꢀ2ꢀ
5
6
15
phenylbutanoyl]amino]benzamide (OSUꢀHDAC42)
were efficacious in HCC models, while
7
8
9
HDAC inhibitor panobinostat (Figure 1) used in combination with sorafenib demonstrated the
highest preclinical efficacy in HCC models.16 In completed clinical trials, HDAC inhibitor
belinostat (Figure 1) demonstrated disease stabilization with a tolerable toxicity profile amongst
patients with HCC,17 while resminostat (Figure 1) 14 in combination with sorafenib was found to
be safe and showed early signs of efficacy in advanced HCC patients.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Both PI3K and mTOR are also validated cancer targets: currently there are two approved
PI3K inhibitors: idelalisib (PI3Kδ selective)18 and copanlisib (panꢀclass I PI3K inhibitor with
preferential activity against PI3Kα and PI3Kδ) 19 (Figure 1). Combination of mTOR inhibitor (5ꢀ
(2,4ꢀbis((S)ꢀ3ꢀmethylmorpholino)pyrido[2,3ꢀd]pyrimidinꢀ7ꢀyl)ꢀ2ꢀmethoxyphenyl)methanol
20
(AZD8055)
with HDAC inhibitor vorinostat in vivo almost completely inhibited tumorꢀ
growth without obvious adverse effects. Furthermore, the combination of dual PI3K/mTOR
inhibitor
1ꢀ[4ꢀ[4ꢀ(dimethylamino)piperidineꢀ1ꢀcarbonyl]phenyl]ꢀ3ꢀ[4ꢀ(4,6ꢀdimorpholinꢀ4ꢀylꢀ
21
1,3,5ꢀtriazinꢀ2ꢀyl)phenyl]urea (PKIꢀ587) with sorafenib targeting both PI3K/AKT/mTOR and
Ras/Raf/MAPK pathways showed synergistic inhibition of HCC cell proliferation. These results
suggest that it is feasible and effective to target both HDACs and PI3K/AKT/mTOR pathways
together for the treatment to advanced HCC.
From a medicinal chemist’s perspective, a multiꢀtarget drug (single molecule) is an alternative
approach to achieve multiꢀtargeted therapy as compared to combination therapy (multiple
22ꢀ
molecules). Based on our knowledge and experience in the development of HDAC inhibitors
24
and PI3K/mTOR inhibitors (e.g., SB2343 (1), Figure 1),25, 26 we designed and synthesized a
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 79
1
2
3
4
5
6
series of fusedꢀpyrimidine based hydroxamates to target both HDACs and the PI3K/Akt/mTOR
pathway for the treatment of cancer as well as nonꢀoncology applications.27
7
8
9
As illustrated in Figure 1, a typical HDAC inhibitor (vorinostat) comprises of three parts: zincꢀ
binding group (ZBG), linker, and a lipophilic and surface recognition “CAP” group. As the CAP
sits outside of the binding pocket of HDAC enzymes,28 it can be replaced with a kinase inhibitor
core structure to afford a novel HDACꢀkinase inhibitor.29 Purine and fusedꢀpyrimidine are
scaffolds used to build up competitive kinase inhibitors (e.g., 1 (SB2343), PI103 (2),30
pictilisib,30ꢀ32 Figure 1) which compete with adenosine triphosphate (ATP). Hence, the HDAC
inhibitor moieties (linker + ZBG) can be used to replace the sugar part of ATP to construct a
kinaseꢀHDAC inhibitor. The general chemical types of the designed HDACꢀPI3K/mTOR
inhibitors are depicted in Figure 2. Previous SAR studies on purineꢀbased PI3K/mTOR inhibitors
(4, as analogues of 1, Figure 2 ) had shown that the 9ꢀposition can accommodate rings and
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
chains,26 thus HDAC inhibitor moieties were introduced to the 9ꢀposition to generate HDACꢀ
26
PI3K/mTOR inhibitors (5). Because both Cpdꢀ38 (6) and Cpdꢀ51 (8) (Figure 2),
two
representative examples of 4, are potent PI3K and mTOR inhibitors, two expanded chemical
series 7 and 9 were designed to establish the SARs for HDAC, PI3K and mTOR inhibitory
activities as well as tune target profile (HDAC vs PI3K vs mTOR) by exploring a variety of
combination of functional groups (i.e., Linker, and R2) (Figure 2). Series 10 was designed to
explore potential morpholine replacement. As the best combination profile of
HDAC/PI3K/mTOR is unknown or yet to be validated, the inhibitors were designed to have a
broad range of potency or strength against each target (i.e., HDAC, PI3K, and mTOR) in order to
mimic the combination of singleꢀtarget inhibitors. The optimal combination was then determined
in preclinical studies. Even though every drug candidate has to address its toxicity preꢀclinically,
ACS Paragon Plus Environment

Page 5 of 79
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
there remain concerns of potential combined toxicity of designed multiꢀtarget drugs in clinics.
Fortunately, the latest Phase I results of CUDCꢀ907 (3) (Figure 1) demonstrated tolerability and
antitumor activity across all dosing schedules studied in patients with relapsed or refractory
9
35
lymphoma or multiple myeloma.33, 34 3 (CUDCꢀ907) is the first HDAC/PI3K dual inhibitor
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
entered clinical trials,34, 36, 37 being considered as a hybrid derivative of PI3K inhibitor pictilisib
and HDAC inhibitor quisinostat (Figure 1). 38, 39
O
CAP
linker
ZBG
OH
H
O
N
H
O
O
N
H
N
H
S
OH
N
OH
N
H
O
N
H
HN
O
boxed warning
Vorinostat
Belinostat
Panobinostat
HDACi
HDACi
HDACi
O
O
N
N
O
OH
N
N
O
N
OH
OH
HN
NH
H
N
H
S
N
N
N
O
N
N
Pracinostat
HDACi
Resminostat
Quisinostat
HDACi
HDACi
H2N
N
N
N
O
HN
O
O
N
N
S
O
N
N
N
HN
N
N
HN
N
S
N
N
HN
N
O
O
N
N
F
O
N
O
Copanlisib
Pictilisib
Idelalisib
PI3K /P3K
PI3K
PI3K/ mTOR (weak)
boxed warning
NH2
O
N
O
from
pictilisib
N
N
N
N
N
N
N
O
N
N
O
N
N
S
N
N
HN OH
HO
N
N
N
N
from quisinostat
O
VS-5584/SB2343 (1)
PI-103 (2)
CUDC-907 (3)
PI3K/mTOR
PI3K/mTOR
HDAC/PI3K
Figure 1. Examples and chemical structure features of HDAC inhibitors (HDACi), PI3K/mTOR
inhibitors, and HDAC/PI3K dual inhibitor. 3 (CUDCꢀ907) is considered to be a hybrid derivative
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 79
1
2
3
4
5
6
of pictilisib and quisinostat. Both idelalisib and panobinostat were issued with a boxed warning
by FDA.
7
8
9
Here we report the synthesis and preclinical evaluation of hydroxamates 7, 9 and 10 series
(Figure 2) which have been designed and developed to modulate both HDACs and the
PI3K/Akt/mTOR pathway. Among 39 cancer cell lines screened, the molecules (e.g., 20e, 20f,
and 20q) were more potent against leukemia, lymphoma, and HCC cells; they also demonstrated
target modulation in cancer cell lines and in mice bearing MV4ꢀ11 and HepG2 tumors.
Compound 20f in particular showed significant single agent oral efficacy in hypervascular HCC
models (e.g., HepG2, HuHꢀ7, and Hep3B) and was wellꢀtolerated. These encouraging results,
along with its favorable target profile and tissue distribution, warrant further development of 20f.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment

Page 7 of 79
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
kinase inhibitor core
keep small groups
when tuning R4
R1
R1
N
tune
PI3K/mTOR
inhibition
N
N
N
N
N
R3
[Linker]
N
R3
N
R2
O
R2
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
HN OH
ZBG
ring and chain
to introduce -Linker-ZBG
R4
CAP
5
Linker
4
NH2
NH2
N
N
N
N
N
9-position
7
N
6
5
4
1N
2
N
N
8
N9
O
N
N
N
N
3
[Linker]
O
O
HN OH
7
6 (Cpd-38)
PI3K /mTOR
IC50 = 17/150 nM
A
O
O
N
N
replace
N
7
N
6
morpholine
2-,7-,9-position
N
N
X
5
1N
N
N
8
2
N
9
N
N
R2
N
R2
4
N
N
O
O
3
[Linker]
[Linker]
H2N
N
HN OH
HN OH
10
9
8 (Cpd-51)
PI3K /mTOR
IC50 = 23/340 nM
X = N, CH
A = -CHOH, -CHCH2OH
Figure 2. Design and chemical types of inhibitors targeting HDACs, PI3Ks and mTOR. IC50
data of Cpdꢀ38 (6) and Cpdꢀ51 (8) were extracted from ref 26.
RESULTS AND DISCUSSION
Chemistry. A wide range of substituted purines and pyrrolo[2,3ꢀd]pyrimidines were prepared
in a straightforward fourꢀ or fiveꢀstep procedure starting from 2,6ꢀdichloropurine (11a) or 2,4ꢀ
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 79
1
2
3
4
5
6
7
8
9
dichloroꢀ7Hꢀpyrrolo[2,3ꢀd]pyrimidine (11b) which were commercially available. As depicted in
Scheme 1, a typical procedure used an alkyl halide in the presence of a suitable base such as
potassium carbonate to alkylate the NH group of compound 11a or 11b. The major products
were Nꢀ9 alkylated 13a–d when purine 11a was used. The two chlorine atoms of 13a–d were
displaced selectively or sequentially under optimal reaction conditions. Under Suzuki reaction
conditions, a boronic ester (14a) reacted with the more reactive chlorine atom to form monoꢀ
chloro compound 15a–d. The 2nd chlorine atom was displaced by an amine (e.g., morpholine) at
elevated temperature to give the esters 16a–d, which were further converted to the hydroxamic
acids 17a–d by hydroxylaminolysis in the presence of excessive hydroxylamine.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 1. Synthesis of Nꢀ9 Substituted Purines 17a–da
aReagents and conditions: (a) NaI (0.2 equiv), anhydrous K2CO3 (2 equiv), Br(CH2)mCO2R5
(12a–d, m = 3,4,5,6; R5 = Et or Me; 1.1 equiv), DMF, 40 °C, 12 h, 61–78%; (b) 14a (1.1
equiv), K2CO3 (2.5 equiv), dioxane/water, Pd(dppf)Cl2·CH2Cl2 (0.05 equiv), 82 °C for 6 h, 24–
72%; (c) morpholine, DMF, 80 °C, 12 h, 56–72%; (d) NH2OH·HCl (10 equiv)/NaOMe (25
equiv)/MeOH, −20 ºC to rt, 21–38%.
ACS Paragon Plus Environment

Page 9 of 79
Journal of Medicinal Chemistry
1
2
3
4
Table 1. SAR of Nꢀ9 Substituted Purines 17a–d
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
HDAC1a
IC50 (nM)
>9257
1754 ± 299
60 ± 58
17 ± 9
50 ± 30
NT
PI3Kαa
MV4ꢀ11b
IC50 (µM)
20.3 ± 2.5
9.81 ± 0.25
2.81 ± 0.39
0.25 ± 0.01
0.43 ± 0.10
NT
PCꢀ3b
MCF7b
Compd
m
IC50 (µM)
>25
IC50 (µM)
12.9 ± 1.9
11.1 ± 2.5
7.37 ± 1.40
2.08 ± 0.48
1.60 ± 0.31
NT
IC50 (nM)
12.8 ± 12.0
8.7 ± 4.6
9.2 ± 6.3
19.0 ± 0.1
NT
3
4
5
6
17a
18.1 ± 6.1
9.28 ± 3.03
2.33 ± 0.46
3.07 ± 1.08
NT
17b
17c
17d
vorinostat
wortmannin
1
10.1 ± 7.1
7.8 ± 3.5
18.7 ± 8.6
NT
0.20 ± 0.03
1.46 ± 1.09
0.18 ± 0.06
0.75 ± 0.36
0.13 ± 0.05
0.22 ± 0.18
pictilisib
NT
aValues are expressed as mean ± standard deviation (SD) of at least two independent
experiments. Values are expressed as mean ± SD of at least two independent duplicate
experiments. NT: not tested.
b
Scheme 2 depicted an alternative displacement sequence of the two chlorine atoms of
compounds 13c–e. The more reactive chlorine was displaced first by morpholine to give 18c–e,
and the second chlorine was used to introduce aryl groups via Suzuki coupling by reaction with
boronic acid or ester 14a–o to afford esters 19a–t which were subsequently converted to the
hydroxamic acids 20a–s. For those aryl groups could not directly derive from the readily
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 79
1
2
3
4
5
6
7
8
9
available 14a–o, one additional step was applied to modify the analogues of 19a–t. The reactive
phenol group in 19t or 19p was alkylated by 2ꢀbromoethanol and subsequently the resulting
product 19t2 or 19u was converted to 20t or 20u, which was extended with a neutral hydroxyl
group. The aldehyde 19v was reduced to the benzyl alcohol 19w, which was subsequently
converted to the final hydroxamate 20v.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment

Page 11 of 79
Journal of Medicinal Chemistry
1
2
3
4
Scheme 2. Synthesis of 2ꢀArylꢀ9ꢀSubstituted Purines and Pyrrolo[2,3ꢀd]pyrimidines 20a–
5
6
va
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
aReagents and conditions: (a) morpholine (3 equiv), DCM, 40 °C, 3 h or rt overnight, 87–
99%; (b) boronic acid or ester 14aꢀo (1.1 equiv), K2CO3 (2.5 equiv), dioxane/water,
Pd(dppf)Cl2·CH2Cl2 (0.05 equiv), 82 °C for 6 h, 28–90%; (c) neat morpholine (excess), 78 °C,
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 79
1
2
3
4
5
overnight; (d) NH2OH·HCl (10 equiv)/NaOMe (25 equiv)/MeOH, ‒20 ºC to rt; (e) 2ꢀ
bromoethanol (3 equiv), K2CO3 (5 equiv), DMF, 90 °C; (f) NaBH4 (10 equiv), MeOH, rt, 95%.
6
7
8
9
Scheme 3 depicted another modification: replacing the commonly used morpholine at the 6ꢀ
positon of the purine ring with other cyclic amines. Both piperidinꢀ4ꢀylmethanol and piperidinꢀ4ꢀ
ol were used to react with 13d to afford 21a and 21b, which were further converted to
hydroxamates 23a and 23b via Suzuki coupling and hydroxylaminolysis.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 3. Synthesis of 6ꢀSubstituted Purines 23a and 23ba
aReagents and conditions: (a) 13d (1 equiv), piperidinꢀ4ꢀylmethanol or piperidinꢀ4ꢀol (6
equiv), Et3N (6 equiv), DCM, 40 °C, overnight; (b) 14a (3 equiv), K2CO3 (4 equiv),
dioxane/water, Pd(dppf)Cl2·CH2Cl2 (0.05 equiv), microwave 150 °C, 10‒30 min; (c)
NH2OH·HCl (10 equiv)/NaOMe (25 equiv)/MeOH, ‒20 ºC to rt.
ACS Paragon Plus Environment

Page 13 of 79
Journal of Medicinal Chemistry
1
2
3
4
Table 2. SAR of 2ꢀArylꢀ9ꢀSubstituted Purines and Pyrrolo[2,3ꢀd]pyrimidines 20a–v
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
HDAC1a
IC50 (nM)
PI3Kαa
MV4ꢀ11b
IC50 (µM)
PCꢀ3b
Compd
Ar
X
m
IC50 (µM)
IC50 (nM)
N
6
5.55 ± 0.38
>10,000
0.78 ± 0.05
3.60 ± 0.90
20a
N
N
N
N
5
6
5
6
12 ± 2
2.9 ± 1.2
35 ± 8
724 ± 156
>1000
6.84 ± 2.66
1.99 ± 0.12
2.39 ± 1.81
17.4 ± 1.5
8.48 ± 0.32
6.97 ± 1.23
20b
20c
20d
20e
>10,000
196±26
0.85 ± 0.34
0.045 ± 0.036 4.01 ± 1.43
0.047 ± 0.030 1.08 ± 0.82
N
N
6
6
2.7 ± 1.2
4.4 ± 5.4
50 ± 5
>1000
20f
0.08 ± 0.09
3.64 ± 1.03
20g
N
6
2.3 ± 1.8
38 ± 27
0.034 ± 0.015 2.26 ± 0.63
20h
N
N
N
6
6
6
1.2 ± 0.8
0.78 ± 0.02
0.84 ± 0.29
>1000
>863
>924
0.061 ± 0.006 6.04 ± 1.19
0.058 ± 0.003 3.53 ± 0.31
0.059 ± 0.004 3.47 ± 0.16
20i
20j
20k
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 79
1
2
3
4
5
HDAC1a
IC50 (nM)
PI3Kαa
MV4ꢀ11b
IC50 (µM)
PCꢀ3b
Compd
Ar
X
m
IC50 (µM)
IC50 (nM)
6
7
8
9
N
N
N
6
0.61 ± 0.06
1.7 ± 1.0
1.4 ± 0.1
>1000
>1000
>6663
0.017 ± 0.002 3.02 ± 0.33
20l
20m
20n
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
6
6
0.31 ± 0.04
1.36 ± 0.51
2.81 ± 0.16
7.54 ± 0.15
N
N
6
2.7 ± 0.4
376
0.062 ± 0.026 2.77 ± 1.00
20o
6
6
6
5
0.49 ± 0.21
1.5 ± 0.5
1.6 ± 0.3
8.5 ± 2.5
619 ± 16
25 ± 10
41 ± 8
1.81 ± 0.16
5.67 ± 1.03
20p
20q
20r
20s
CH
CH
N
0.044 ± 0.027 1.57 ± 0.69
0.13 ± 0.01
7.39±0.41
1.70 ± 0.57
17.2 ± 2.8
317 ± 23
N
6
0.76 ± 0.65
994 ± 22
0.030 ± 0.008 1.68 ± 1.12
20t
N
N
6
6
3.7 ± 1.4
6439 ± 180
218 ± 14
1.88 ± 0.14
0.38 ± 0.06
4.69 ± 0.70
2.00 ± 0.17
20u
20v
0.89 ± 0.41
NT
NT
7.8 ± 3.5
18.7 ± 8.6
NT
0.20 ± 0.03
1.46 ± 1.09
0.43 ± 0.10
0.18 ± 0.06
0.75 ± 0.36
3.07 ± 1.08
1
pictilisib
vorinostat
50 ± 30
ACS Paragon Plus Environment

Page 15 of 79
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
HDAC1a
IC50 (nM)
NT
PI3Kαa
MV4ꢀ11b
IC50 (µM)
NT
PCꢀ3b
IC50 (µM)
NT
Compd
Ar
X
m
IC50 (nM)
10.1 ± 7.1
wortmannin
a,b See footnote of Table 1.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 3. SAR of 6ꢀSubstitutedamino Purines 23a and 23b
R1
N
N
N
N
N
O
(CH2)6
H2N
N
HN
OH
HDAC1a
IC50 (nM)
PI3Kαa
MV4ꢀ11b
IC50 (µM)
PCꢀ3b
MCF7b
Compd R1
IC50 (µM)
IC50 (µM)
IC50 (nM)
N
0.99 ± 0.35 1041 ± 203
0.69 ± 0.10 3955 ± 1688
0.17 ± 0.05
0.13 ± 0.03
1.94 ± 0.19
1.68 ± 0.74
2.56 ± 1.58
23a
23b
HO
N
1.21 ± 0.48
0.95 ± 0.69
HO
2.3 ± 1.8
38 ± 27
0.034 ± 0.015
2.26 ± 0.63
20h
a,b See footnote of Table 1.
Scheme 4 depicted the approaches to introduce aryl linked hydroxamates at the 9ꢀposition of the
purine ring. The more reactive 6ꢀCl of 11a was first replaced with morpholine, and then the
resulting 6ꢀmorpholinopurine 24 was subjected to further alkylation. This approach was also
applied to make individual compounds described in both Schemes 2 and 3. Nitrile 25 was
prepared by Nꢀ9 alkylation of 24 with 2ꢀiodoacetonitrile and was further reduced to amine 26.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 79
1
2
3
4
5
6
7
8
9
By reductive amination with aldehyde and alkylation with alkyl bromide, amine 26 was
converted to intermediates 27 and 32 with aryl linkers, respectively. After Suzuki coupling with
either boronic ester 14a or boronic acid 14d, and subsequent hydroxylaminolysis, 27 and 32
were eventually converted to hydroxamates 29, 31 and 34.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 4. SAR of Nꢀ9 Aryl linked Purine Hydroxamates 29, 31 and 34
HDAC1a
IC50 (nM)
84 ± 40
PI3Kαa
IC50 (nM)
>8956
MV4ꢀ11b
IC50 (µM)
0.24 ± 0.07
5.44 ± 1.73
PCꢀ3b
IC50 (µM)
1.72 ± 0.81
>25
Compd
29
31
34
136 ± 10
307 ± 57
1082 ± 192 4543 ± 1136 5.78 ± 0.64
a,b See footnote of Table 1.
18.5 ± 6.3
The acid (RCO2H) was identified as the major impurity in the final target hydroxamate
(RCONHOH) product. In general, the hydroxamic acid was slightly more polar than the acid and
was separated by preparative HPLC with purity >95%. The acid (20w) corresponding to 20f
was prepared by hydrolysis of the ester 19f (Supporting Information), was a very poor HDAC
inhibitor as compared to the respective hydroxamic acid 20f, it still maintained majority of the
PI3K inhibitory potency, but its contribution to biological activity was insignificant or negligible
(20f vs 20w, Table S1).
Scheme 4. Synthesis of Nꢀ9 Aryl linked Purine Hydroxamates 29, 31, and 34a
ACS Paragon Plus Environment

Page 17 of 79
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
aReagents and conditions: (a) morpholine (3 equiv), rt, THF; (b) ICH2CN (2 equiv), anhydrous
K2CO3 (2 equiv), MeCNꢀDMSO (19:1, v/v), 60 ºC; (c) NiCl2·6H2O (0.1 equiv), NaBH4 (7
equiv), THF/MeOH (1:2); (d) methyl 3ꢀ(4ꢀformylphenyl)acrylate (1.2 equiv), NaBH(OAc)3 (1.5
equiv), HOAc (1 equiv), DCE, rt; (e) 14a or 14d (2.0 equiv, Scheme 2), K2CO3 (2.0 equiv),
dioxane/water, Pd(dppf)Cl2·CH2Cl2 (0.05 equiv), microwave 15–30 min; (f) NH2OH·HCl (10
equiv)/NaOMe (25 equiv)/MeOH, −20 ºC to rt; (g) pꢀBrCH2C6H4CO2CH3 (1.1 equiv), Et3N (1.2
equiv), THF, rt, 2 h.
Inhibition of HDAC1 and PI3Kα, SAR and Antiꢀproliferative Activity. HeLa nuclear
extract and HDAC isozyme HDAC1 as well as PI3Kα (p110α/p85α) were used as routine
enzymatic screening tests for HDAC and PI3K inhibitory activities. Cancer cell lines MV4ꢀ11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 79
1
2
3
4
5
6
7
8
9
(acute monocytic leukemia), PCꢀ3 (prostate cancer) and MCF7 (breast cancer) were used to
evaluate antiꢀproliferative activities as well as modulation of targets (HDACs, PI3K, mTOR) in
cancer cells. HDAC inhibitor vorinostat and PI3K inhibitor pictilisib were used as positive
control in all experiments.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
To establish SAR, we tried to optimize HDAC inhibitory activity first and started from a PI3K
inhibitor core structure with fixed functional groups which were good for PI3K inhibition.
According to our previous study, purineꢀbased PI3K/mTOR inhibitor 1 (Figure 1) or 4 (Figure
2), the 2ꢀaminopyrimidinꢀ5ꢀyl group was good for PI3Kα and mTOR inhibition (e.g., 6 and 8),
and 9ꢀposition could accommodate substituents such as rings or chains of reasonable size, 26 and
thus a batch of compounds 17a–d (Table 1) were first synthesized. Their PI3Kα inhibitory IC50
values were comparable to that of 1, however their HDAC IC50 values were correlated with the
length of carbon linker. It was found that C6 (six methylenes) or six atoms linker was optimal;
23
these results were in line with the previous reported SAR for acylurea linked hydroxamates.
The antiꢀproliferative potencies were also improved along with the linker length, but even the
best compound 17d which showed better HDAC inhibitory potency than vorinostat was less
potent than vorinostat in cells. Besides target potency, the physicochemical properties may also
affect cellular potency. Polar group 2ꢀaminopyrimidinꢀ5ꢀyl at 1ꢀposition and linkerꢀhydroxamic
acid at 9ꢀposition of the purine molecule may not be the best arrangement. Next we tried to swap
the morpholine and 2ꢀaminopyrimidine groups to create molecules like 8 or 9 (Figure 2). A
series of compounds 20a–v were thus made to further confirm the optimal linker for HDAC
inhibition and explore PI3K/mTOR inhibitory potency SAR (Table 2).
The C6 linker for hydroxamate was still preferred (20b vs 20c, 20d vs 20e, and 20s vs 20p),
and the HDAC1 IC50 value went down to 1 nM, this was still valid for most of the compounds
ACS Paragon Plus Environment

Page 19 of 79
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
with C6 linker in Table 2. The IC50 value didn’t change significantly with a variety of
substituents at 2ꢀposition of the purine nor pyrrolo[2,3ꢀd]pyrimidine ring, as these lipophilic
purines or pyrrolo[2,3ꢀd]pyrimidines were excellent CAPs for HDAC inhibitors. Therefore there
was leeway to tune PI3K/mTOR inhibitory potency in these SARs. 2ꢀAminopyrimidinꢀ5ꢀyl
group was still one of the best groups for PI3K inhibition (e.g., 20h and 20q), metabolically
labile meta phenol group was also good (20r) for PI3Kα inhibition (similar to 2), and meta
benzyl alcohol (20f) was found to be a good replacement of the phenol and demonstrated
comparable PI3Kα potency. The binding site cannot tolerate big or bulky groups as we reported
earlier,26 such compounds (20i–l, 20n, 20t, and 20u) didn’t show any appreciable potency
against PI3Kα. Adding para substituent to meta phenol (20p and 20s) or benzyl alcohol (20v)
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
actually reduced the PI3Kα potency. Compound 20g with para benzyl alcohol group was no
longer active against PI3Kα. Compounds 20f, 20h, 20q, 20e, and 20o were very potent HDAC
inhibitors with variable PI3Kα potency and showed good cellular activates as well, were worth
of further profiling and evaluation.
40
Morpholine ring was commonly used in PI3K inhibitors,
it also showed good activity at
purine 6ꢀposition in above mentioned series. A few substituents were tried to replace the
morpholine, such as 23a and 23b, unfortunately they significantly destroyed PI3Kα affinity, but
still kept the HDAC inhibitory effect intact.
Besides aliphatic linkers, aryl linker was also commonly used for HDAC inhibitors. We made
several compounds with aryl linker at Nꢀ9 position to investigate the HDAC potency. Both 29
and 31 with cinnamic linkers showed moderate HDAC potency, but the PI3Kα potency was
affected even with the best 2ꢀaminopyrimidine group, the Nꢀ9 substituent was too big and basic
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 79
1
2
3
4
5
6
(secondary amine). 34 was linked with a shorter benzylamine linker unit and showed even poorer
HDAC inhibitory potency.
7
8
9
Molecular Docking Study. To better understand the above observed SAR for PI3Kα
inhibition, molecular docking study was performed using AutoDock Vina software41 and crystal
structure of PI3Kα–3 complex (PDB ID code: 4L23)42 (Supporting Information, Figure S2). For
these morpholinoꢀpyrimidine based PI3K inhibitors, there are two major binding modes. One is
represented by 2 (Figure S2A and Figure S2B), and the other is represented by 1 (Figure S2D).
The backbone NH group of Val851 in the hinge region is a hydrogen bond donor, it forms a
hydrogen bond with the 1ꢀposition nitrogen atom of the ATP purine ring, and this interaction is
critical to kinase activity. The oxygen of the morpholine group of 2 is a hydrogen bond acceptor
and forms a hydrogen bond with the NH group of Val851 (Figure S2B). Forming this bond
enables 2 to compete with ATP in the ATP binding pocket. The hydroxyl group of the phenol
ring of 2 also forms another two key hydrogen bonds with the residues of Tyr836 and Asp810.
Compound 10r, which has a meta phenol group as well, shows a similar binding mode (Figure
S2E). 20p has similar interactions with the key residues, but the inhibitory potency is lower,
probably due to the unfavorable interaction between the fluorine atom and the hydroxyl group
(Figure S2F). The meta benzyl alcohol group of 20f interacts with the residues of Asp810,
Tyr836 and Asp933 (Figure S2G), which makes 20f a potent PI3Kα inhibitor. In contrast, the
para benzyl alcohol group of 20g is far away from Tyr836, and only forms a hydrogen bond with
Asp810. Furthermore, the oxygen of benzyl alcohol group is unfavorably close to the carboxylic
group of Asp933 (Figure S2H). This is why 20g loses PI3Kα inhibitory activity dramatically as
compared to 20f (Figure S2I). Pictilisib does not have a hydroxyl group, but its 1Hꢀindazolꢀ4ꢀyl
group forms two hydrogen bonds with Tyr836 and Asp810 (Figure S2C), and therefore it is also
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment

Page 21 of 79
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
a potent PI3Kα inhibitor. Compound 20n also has an indazole group, but it is 1Hꢀindazolꢀ6ꢀyl
and thus only forms a hydrogen bond with Asp810 (Figure S2J). The hydrogen bond with
Tyr836 is very important in achieving a ligand’s inhibitory activity, for example, an Y836A
mutation resulted in 44ꢀfold reduction of 2’s potency against PI3Kα.42
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Compound 1 represents the second binding mode, its para amino group of the 2ꢀ
aminopyrimidine group forms two hydrogen bonds with Asp810 and Asp933, and the two
nitrogen atoms in the pyrimidine ring form additional hydrogen bonds, one with Lys802, and the
other with the NH group of Asp933 or hydroxyl group of Tyr836 (Figure S2D). Compounds
17d (Figure S2K), 20h (Figure S2L) and 20q (Figure S2M) all have 2ꢀaminopyrimidine groups:
they show similar binding mode as that of 1 and are also potent PI3Kα inhibitors. 20e has an
aminoꢀpyridine group and it forms fewer hydrogen bonds as compared to 2ꢀaminopyrimidine
group (Figure S2N), therefore its potency is lower than 20h or 20q. 23a (Figure S2O) and 23b
(Figure S2P) are two analogues of 20h, with the morpholine group being replaced by piperidines.
In order to make similar contact with Val851, the hydroxyl group repels the entire molecule
away from the binding pocket (i.e., Trp780, Met772, Met922, Ile800, Ile932, Tyr836 and
Asp933), but pushes it slightly closer to Lys802 as compared to 20h. Furthermore, the piperidine
has to adopt a conformation notably different from that of the morpholine in 20h. In contrast, all
rings including the morpholine of 20h superpose well on 2 (Figure S2L). Hence, 23a and 23b are
poor PI3Kα inhibitors. Interestingly, 23a is about 4ꢀfold as potent as 23b. Its flexible primary
alcohol may allow it to adopt a more favorable conformation in order to form a hydrogen bond
with the NH group of Val851 in contrast to that of the rigid secondary alcohol in 23b. It is thus
clear that piperidine is not a suitable replacement for the morpholine group in the current
chemical series.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 79
1
2
3
4
5
6
7
8
9
In Vitro Profiling against HDACs and Kinases. Besides in house assays, the HDAC isoform
profiling was done at BPS Bioscience. In additional to routine controls such as vorinostat and
trichostatin A (TSA), 3 and pracinostat were also included for comparison. The data for 3,
except for class IV HDAC11, were generally in line with reported data generated with different
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
substrate.36
IC50 values of pracinostat were also in line with the trend of our previous reported
Ki values which were derived from IC50 values, except for HDACs 4, 5, 7, 8 and 11 which were
tested using class 2a substrate. 20e, 20f and 20q were potent nanomolar inhibitors for HDACs 1,
2, 3, 6, and 10, but poor inhibitors for classes IIa and IV HDACs.
Selected compounds 20f and 20q were also profiled against a panel of 97 kinases: both were
found selective against lipid kinases (Supporting Information Table S2), and their Kd values were
determined (Table 6). The PI3K and mTOR inhibitory IC50 values were determined at both
Reaction Biology Corp and Life Technologies, the data agreed well from two different assay
platforms. Additional kinase profiling was done at Life Technologies with 3 as positive control.
Pictilisib was the most potent PI3K inhibitor, but a poor mTOR inhibitor. 3 showed reasonable
potency against mTOR, and was still a panꢀinhibitor for PI3K. Compounds 20h, 20e and 20q
were active against mTOR, 20f was a weaker mTOR inhibitor thus more selective for PI3K, was
also potent against PI3K2β, which was confirmed by two different assay platforms.
Compounds 20f, 20h, 20e, 20o, and 20q achieved dual or multi inhibitions of HDACs and
PI3K/mTOR. They were very potent panꢀHDAC inhibitors with a different degree of inhibitory
potency against PI3K/mTOR, thus were good examples for further evaluation in order to identify
the optimal combination of HDAC and PI3K/mTOR.
ACS Paragon Plus Environment

Page 23 of 79
1
Journal of Medicinal Chemistry
Table 5. Inhibition of HDAC Isozymes by Selected Compounds
2
3
4
5
IC50 (nM) for HDAC Isozymesa
6
7
8
9
Compd
20e
1
2
3
4
5
6
7
8
184
320
270
NT
9
10
1.7
2.5
2.3
27
11
1.0
1.1
1.5
81
4.0
6.0
4.1
220
NT
290
8.0
2.0
1.1
0.66
34
>10,000 >10,000
4,591 4,800
>10,000 >10,000
4.6
4.2
6.6
12
>10,000
2,305
>10,000
NT
>10,000
1,282
>10,000
NT
>10,000
9,700
6,300
NT
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
20f
20q
vorinostat
TSA
NT
NT
NT
NT
NT
50
NT
70
NT
93
NT
490
2,835
54
NT
NT
82
10,900
>10,000
1,267
pracinostat
3
>10,000
479
5,113
581
>10,000
528
>10,000
639
2.0
3.2
34
4.1
aHDAC profiling was done by BPS Bioscience (6042 Cornerstone Court West, Suite B, San Diego, CA 92121), HDAC Substrate 3 (BPS catalog
#50037) was used for class I HDACs (HDACs 1, 2, and 3) and class IIb HDACs (HDACs 6 and 10) and vorinostat as positive control; substrate
class 2a (catalog # 50040) was used for class IIa HDACs (HDACs 4, 5, 7, and 9) as well as HDACs 8 (class I) and 11 (class IV), using TSA as
positive control. TSA: trichostatin A. NT: not tested.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 79
1
2
Table 6. Inhibition of Lipid and Relevant Kinases by Selected Compounds
3
4
5
Kd (nM)a
IC50 (nM)b
6
7
8
9
Pictilisib
4.7
Kinases
PI3K
20f
3.5
NT
86
20q
12
20h
20e
198
1656
793
301
374
20f
28
20o
20q
11
3
7.9
2
29
136
3.8
10
α
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
PI3Kβ
NT
100
240
NT
NT
260
NT
NT
NT
648
212
177
37
1403
692
87
NT
21
PI3Kγ
98
92
NT
112
6.3
71
PI3Kδ
41
133
331
63
NT
11
mTOR
PI3K2α
PI3K2β
DNAꢀPK
PIK3C3
PI4Kβ
NT
NT
4.9
NT
NT
NT
288
1,946
4,945
101
959
NT
185
NT
141
NT
NT
NT
NT
NT
>1,000
>1,000
>1,000
>10,000 >1,000
>1,000 103
>10, 000 >1,000
>1,000 1,046
NT
NT
>10,000 10,000
>10,000 594
NT
3,673
>10,000
NT
>10,000 >10,000 10,000
>10,000 >10,000 >10,000
>10,000 6,295
>10,000 >10,000
NT
NT
aKd values were determined by DiscoverX Corporation (42501 Albrae Street, Fremont, CA 94538, United States). bPI3K and mTOR IC50 values
were determined by Reaction Biology Corp. (One Great Valley Parkway Suite 2, Malvern, PA 19355, United States) and Life Technologies. NT:
not tested.
ACS Paragon Plus Environment

Page 25 of 79
Journal of Medicinal Chemistry
1
2
3
4
Table 7. AntiꢀProliferative IC50 (µM) of Selected Compoundsa
5
6
7
8
9
Cancer
Cell Lines
K562
Pictilisib
Vorinostat
Sorafenib
20e
20f
20q
Leukemia
0.60 ± 0.30
0.16 ± 0.10
0.045 ± 0.036
0.35 ± 0.15
0.60 ± 0.18
0.23 ± 0.12
1.55 ± 0.15
0.96 ± 0.29
0.50 ± 0.19
1.57 ± 0.24
1.16 ± 0.23
3.15 ± 2.16
1.52 ± 0.63
0.35 ± 0.14
0.14 ± 0.11
0.047 ± 0.030
0.24 ± 0.04
0.40 ± 0.13
0.19 ± 0.06
1.25 ± 0.20
0.59 ± 0.11
0.48 ± 0.13
1.24 ± 0.08
0.79 ± 0.20
2.64 ± 1.50
1.38 ± 0.57
0.44 ± 0.32
0.15 ± 0.07
0.044 ± 0.027
0.28 ± 0.08
0.49 ± 0.18
0.25 ± 0.14
2.87 ± 0.26
1.05 ± 0.25
0.65 ± 0.20
3.05 ± 0.41
1.46 ± 0.54
3.90 ± 0.79
1.66 ± 0.54
6.34 ± 3.09 1.02 ± 0.12 5.08 ± 0.37
0.18 ± 0.04 0.52 ± 0.16 5.94 ± 0.04
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
MOLT4
MV4ꢀ11
1.46 ± 1.09 0.43 ± 0.10
6.50 ± 3.23 1.11 ± 0.48 3.94 ± 0.04
4.86 ± 1.79 0.62 ± 0.07 NT
<0.006
Lymphomas Raji
Ramos
SUꢀDHLꢀ6
Liver cancer Hep3B
0.03 ± 0.01 0.58 ± 0.17 3.82 ± 0.02
2.03 ± 1.02 2.60 ± 0.56 4.33 ± 1.16
6.22 ± 1.83 1.27 ± 0.10 3.66 ± 0.90
1.72 ± 0.02 1.95 ± 0.39 2.68 ± 0.28
3.07 ± 0.74 1.87 ± 0.16 7.51 ± 2.59
4.30 ± 0.24 1.85 ± 0.28 8.37 ± 3.84
HepG2
HuHꢀ7
PLC/PRF/5
SKꢀHEP1
SNUꢀ387
SNUꢀ398
NT
NT
2.45 ± 0.69 8.29 ± 0.75
1.39 ± 0.31 3.38 ± 0.43
aValues are expressed as mean ± SD of at least two independent duplicate experiments. NT: not tested.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 79
1
2
3
4
5
6
7
8
9
In Vitro Antiꢀproliferation Profiling. Lead compounds were screened against 39 cancer cell
lines (Supporting Information Table S3) in antiꢀproliferative assays (cellular IC50) and were
found more sensitive/potent against leukemia, lymphoma and HCC cell lines (Table 7). For
instance, 20e, 20f and 20q were very active against leukemia cell line MV4ꢀ11, and were much
more potent than pictilisib and vorinostat. Sorafenib in general was much less potent than these
compounds as its mechanism is mainly antiꢀangiogenesis in vivo. Sorafenib’s exceptional
activity against MV4ꢀ11 might be due to its inhibition of multiꢀkinase particularly FLT3. Among
liver cancer cell lines, 20f was more potent against HepG2, HuHꢀ7 and SKꢀHEP1.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3. Caspase activity induction and cell death in MV4ꢀ11 cells. (A) Maximal caspase
activity at 24 h. MV4ꢀ11 cells were treated with 20f (▼), vorinostat (●), and pictilisib (♦),
respectively. Caspase 3/7 activity was monitored at 6, 24, 48 and 72 h, respectively. Maximal
activity was found and shown at 24 h, expressed as fold of change (treated vs vehicle, mean ±
SD). (B) Maximal cell death occurred at 48 h. MV4ꢀ11 cells were treated with 20f (▼)
vorinostat (●), and pictilisib (♦), respectively. Cell death was monitored at 24, 48 and 72 h
Maximal cell death as fold of treated vs vehicle was occurred at 48 h. Y Axis is expressed as
mean ± SD. EC50 value was calculated by fitting the doseꢀresponse curve with Prism 4.0.
ACS Paragon Plus Environment

Page 27 of 79
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Caspase Activity Induction and Cell Death. HDAC inhibitors can induce cell apoptosis43
Caspaseꢀ3/7 activity was used as a surrogate for apoptotic cell death induction. Cells were
treated with 20f, vorinostat and pictilisib, respectively and caspase activity was monitored at
different time points. The maximal activity was found and shown at 24 h. The caspase induction
effective concentrations (EC50) were 0.40, 0.90, and 10.7 µM for 20f, vorinostat and pictilisib,
respectively. Maximal cell death as fold of treated vs vehicle was occurred at 48 h, the cell death
potency EC50 was estimated as 0.064, 0.53 and 2.71 µM for 20f, vorinostat and pictilisib,
respectively. 20f was much more potent than vorinostat and pictilisib in terms in induction of
caspase active and cell death in MV4ꢀ11 cells.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Clonogenic and Soft Agar Assays. Anchorageꢀdependent clonogenic assay was used to
determine cell reproductive death after treatment with 20f. Liver cancer cells Hep3B and HuHꢀ7
after treatment with a series dilution of 20f, showed IC50 values of 0.61 µM and 0.39 µM,
respectively, and these values were lower than their respective in vitro cellular assay IC50 values
of 1.25 µM and 0.48 µM, respectively (Table 7). One of the hallmark characteristics of cellular
transformation and uncontrolled cell growth is anchorage independent growth of cancer cells in
soft agar. When 20f was tested in a 3ꢀdimensional (3D) soft agar format, it inhibited colony
formation of NCIꢀH460 and 4T1 cells with IC50 values of 0.31 µM and 0.91 µM, respectively.
Again these IC50 values were lower than the normal in vitro cellular assay IC50 values (0.89 µM
and 1.29 µM for NCIꢀH460 and 4T1, respectively). These data suggested that 20f could inhibit a
variety of cancer cells under both anchorageꢀdependent and anchorageꢀindependent conditions.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 79
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 4. Clonogenic and soft agar assays and IC50 values. 20f inhibited colony formation of
Hep3B and HuHꢀ7 cells under anchorageꢀdependent conditions and inhibited colony formation
of NCIꢀH460 and 4T1 (murine triple negative breast cancer) cells in soft agar assays.
Modulation of HDACs and PI3K/mTOR in Cancer Cells and Tumors. Inhibition of
HDACs will result in hyperacetylation of histones and other substrates such as αꢀtubulin, and
the latter is the result of inhibition of HDAC6. Inhibition of PI3K/Akt/mTOR axis may reduce
the phosphorylated Akt level as well as the phosphorylation levels of downstream proteins such
as P70S6K, S6, and 4ꢀEBP1. The effect on target modulation was first studied in PCꢀ3 (PTEN
deficient) and MCF7 (PIK3CA, E545K mutation) cells and most of the compounds were
screened for target modulation.
In PCꢀ3 cells, after 24 h of drug treatment, 20f, 20g, and 20h were confirmed to be potent
HDAC inhibitors as their effects on hyperacetylation of histone 3 (AcH3) and αꢀtubulin were
ACS Paragon Plus Environment

Page 29 of 79
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
similar to those of vorinostat at 10 µM (Supporting Information Figure S1A). 17d was a much
weaker HDAC inhibitor as compared to 20f–h in terms of enzymatic potency and cellular target
modulation. Pictilisib was a very potent PI3K inhibitor and a weak mTOR inhibitor, it almost
completely depleted pAkt at 1 µM. The order of potency in PCꢀ3 cells for pAkt reduction was
pictilisib > 20f > 17d > 20h > 20g. Vorinostat showed no effect on pAkt level in this experiment.
In MCF7 cells, 20q appeared as a potent effector on pAkt and pS6 reduction, as it was more
potent than 20f and also an mTOR inhibitor (Supporting Information Figure S1B). During the
short period of time of treatment, these compounds didn’t change the total protein levels in
MCF7 cells (e.g., αꢀtubulin, Akt, P70S6K, and S6).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In liver cancer HepG2 cells, 20f and 20q were potent HDAC inhibitors in terms of
hyperacetylation of H3 and αꢀtubulin (Figure 5A), while 3 was a weaker inhibitor on HDAC6.
Acid 20w is a potential impurity or degradation product of 20f, still maintained the majority of
PI3K inhibitory activity, but it showed much weak activity against HDAC6 and HDACs 1, 2, 3
in HepG2 cells. In HuHꢀ7 cells (Figure 5B), the overall effect on phosphorylated Akt, P70S6K,
and S6 was in the order of 20q > 20f > 20e > 20o, which was in line with their respective in vitro
enzymatic IC50 potency against PI3Kα and mTOR. Sorafenib reduced pAkt level (gel
normalized intensity 42% vs vehicle) at lower concentration (1 µM), but enhanced pAkt (S473)
level (252% vs vehicle) at higher concentration (10 µM). These data further support that a
PI3K/mTOR inhibitor is necessary to combine with sorafenib to reduce pAkt level.13, 44, 45
ACS Paragon Plus Environment