Development of the First Generation of Disulfide-Based Subtype-Selective and Potent Covalent Pyruvate Dehydrogenase Kinase 1 (PDK1) Inhibitors

Article abstract of DOI:10.1021/acs.jmedchem.6b01245

Pyruvate dehydrogenase kinases (PDKs) are overexpressed in most cancer cells and are responsible for aberrant glucose metabolism. We previously described bis(4-morpholinyl thiocarbonyl)-disulfide (JX06, 16) as the first covalent inhibitor of PDK1. Here, on the basis of the scaffold of 16, we identify two novel types of disulfide-based PDK1 inhibitors. The most potent analogue, 3a, effectively inhibits PDK1 both at the molecular (k_inact/K_i = 4.17 × 10³ M^-1 s^-1) and the cellular level (down to 0.1 μM). In contrast to 16, 3a is a potent and subtype-selective inhibitor of PDK1 with >40-fold selectivity for PDK2-4. 3a also significantly alters glucose metabolic pathways in A549 cells by decreasing ECAR and increasing ROS. Moreover, in the xenograft models, 3a shows significant antitumor activity with no negative effect to the mice weight. Collectively, these data demonstrate that 3a may be an excellent lead compound for the treatment of cancer as a first-generation subtype-selective and covalent PDK1 inhibitor.

Full text of DOI:10.1021/acs.jmedchem.6b01245

Subscriber access provided by University of Newcastle, Australia
Article
Development of the First Generation of Disulfide-Based Subtype-Selective
and Potent Covalent Pyruvate Dehydrogenase Kinase 1 (PDK1) Inhibitors
Yifu Liu, Zuoquan Xie, Dan Zhao, Jin Zhu, Fei Mao, Shuai Tang,
Hui Xu, Cheng Luo, Mei-Yu Geng, Min Huang, and Jian Li
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.6b01245 • Publication Date (Web): 23 Feb 2017
Downloaded from http://pubs.acs.org on February 24, 2017
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 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Development of the First Generation of Disulfide-Based
Subtype-Selective and Potent Covalent Pyruvate Dehydrogenase
Kinase 1 (PDK1) Inhibitors
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
a, †
b, †
c
a
a
b
b
Yifu Liu , Zuoquan Xie , Dan Zhao , Jin Zhu , Fei Mao , Shuai Tang , Hui Xu , Cheng
c
b
b,
a,
Luo , Meiyu Geng , Min Huang *, Jian Li *
a
Shanghai Key Laboratory of New Drug Design, School of Pharmacy, East China University
of Science and Technology, Shanghai 200237, China
b
Division of Antitumor Pharmacology, State Key Laboratory of Drug Research, Shanghai
Institute of Materia Medica, Chinese Academy of Sciences, Shanghai 201203, China
c
Drug Discovery and Design Center, State Key Laboratory of Drug Research, Shanghai
Institute of Materia Medica, Chinese Academy of Sciences, Shanghai 201203, China
†
These authors contributed equally to this work.
*
To whom correspondence should be addressed: jianli@ecust.edu.cn or
mhuang@simm.ac.cn
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 70
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
ABSTRACT
Pyruvate dehydrogenase kinases (PDKs) are overexpressed in most cancer cells and are
responsible for aberrant glucose metabolism. We previously described Bis(4ꢀmorpholinyl
thiocarbonyl)ꢀdisulfide (JX06, 16) as the first covalent inhibitor of PDK1. Here, based on the
scaffold of 16, we identify two novel types of disulfideꢀbased PDK1 inhibitors. The most
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
ꢀ1 ꢀ1
i
potent analog, 3a, effectively inhibits PDK1 both at the molecular (kinact/K = 4.17×10 M s )
and the cellular level (down to 0.1 ꢁM). In contrast to 16, 3a is a potent and subtypeꢀselective
inhibitor of PDK1 with >40ꢀfold selectivity for PDK2–4. 3a also significantly alters glucose
metabolic pathways in A549 cells by decreasing ECAR and increasing ROS. Moreover, in
the xenograft models, 3a shows significant antitumor activity with no negative effect to the
mice weight. Collectively, these data demonstrate that 3a may be an excellent lead compound
for the treatment of cancer as a firstꢀgeneration subtypeꢀselective and covalent PDK1
inhibitor.
ACS Paragon Plus Environment

Page 3 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
INTRODUCTION
As a major public health problem worldwide, cancer is a very complex set of diseases.
Unlike their normal counterparts, most cancer cells have a preference for cytoplasmꢀbased
glycolysis rather than mitochondrial oxidative phosphorylation, even under normoxia. This
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
unique reprograming of cellular metabolic pathways, first recognized by Otto Warburg in
1
1924, meets the survival and proliferation needs during the process of tumor progression. In
recent years, a more intricate understanding of the metabolic alterations and adaptations of
cancer cells has refocused efforts to target tumor metabolism as a selective anticancer
2
strategy.
The pyruvate dehydrogenase complex (PDC) is a gatekeeper of glucose oxidation, and
3
by converting pyruvate to acetylꢀCoA, it links glycolysis to the Krebs cycle. An emerging
concept suggests that the Warburg effect may be partly due to the attenuation of
mitochondrial function, which results from the inhibition of PDC via an increased expression
4
of pyruvate dehydrogenase kinase. Four pyruvate dehydrogenase kinase isoenzymes
(PDK1–4) have been identified in mammals with tissueꢀspecific expression. They negatively
regulate the activity of the PDC by reversible phosphorylation, which occurs independently
5–7
on three different serines (Ser232, Ser293, and Ser300). PDK1 is the only isoform reported
8
–9
to phosphorylate all three serines and is closely associated with cancer malignancy. In
cancer cells, the expression of the PDK1 gene is upregulated by the oncogenes cꢀMyc,
hypoxiaꢀinducible factorꢀ1α (HIFꢀ1α) and BRAFVꢀ600E to control metabolic and malignant
10–12
phenotypes.
A recent study revealed that PDK1 is regulated by Wnt/βꢀcatenin signaling
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 70
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
1
3
in colon cancer cells. In addition, PDK1 activity is increased at the postꢀtranscriptional level
by diverse oncogenic tyrosine kinases. This enhancement of PDK1 activity results in a
1
4
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
promotion of the Warburg effect and tumor growth. Furthermore, overexpression of PDK1
or PDK4 protects cells from anoikis, while depletion of the enzymes restores the
15
susceptibility to anoikis and lowers the metastatic potential. These evidences suggest that
the inhibition of PDK (especially PDK1) may offer a new therapeutic option for the treatment
of cancer.
16–17
Before our recent publications on two new classes of PDK inhibitors,
classes of PDK inhibitors had been reported and distinguished by different binding sites
Figure 1). The wellꢀknown PDK inhibitor dichloroacetate (DCA, Figure 1), an analog of
pyruvate, rapidly entered phase II clinical trials after its anticancer activity was first
three major
(
18
18–19
reported. However, the high effective dosage limits its further clinical application.
Although there are several ingenious modifications of DCA, the results are far from
20–22
23
24
satisfactory.
Nov3r and AZD7545 (Figure 1) represent a class of PDK inhibitors that
25
target the lipoylꢀbinding pocket of the enzymes; and recently CPI613 (Figure 1), a lipoate
derivative, demonstrates certain anticancer effects in an in vivo human xenograft tumor, and it
has been investigated in phase I clinical trials for patients with relapsed or refractory
2
3–25
26
16
27
hematological malignancies.
Radicicol, M77976 and its analogs, and PS10 (Figure 1)
belong to a class of ATPꢀcompetitive inhibitors that target the GHKL (gyrase, Hsp90,
16, 26–27
histidine kinase, MutL) domain in PDKs.
However, the high similarity of the
ATPꢀbinding pocket of kinases may lead to offꢀtarget effects, resulting in side effects and
ACS Paragon Plus Environment

Page 5 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
toxicity issues.
As the starting point of our efforts to discover a new type of PDK1 inhibitors, an
inꢀhouse library of ~600 old drugs was screened using an enzymatic screening assay.
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
1
7
Thiram (Figure 1), an existing pesticide with anticancer activity, was found to be capable of
remarkably inhibiting PDK1 activity, and further chemical efforts led to the discovery of a
17
more potent PDK1 inhibitor, 16 (JX06, Figure 1). Using 16 as a chemical probe, we found
that the covalent inhibitory action mode of 16 and its analogs was through targeting a highly
conserved cysteine residue (Cys240), and we identified the cancer subset responsive to PDK1
inhibition based on the ECAR (extracellular acidification rate)/OCR(oxygen consumption
17
rate) ratio.
Figure 1
Based on the therapeutic potential of 16, to obtain novel analogs with independent
intellectual property rights and to explore the SAR (structureꢀactivity relationship) of the 16
analogs, we designed two types of novel disulfideꢀbased analogs, bis(thiazolyl)disulfides
(
1a–m) and bis(thiocarbonyl)disulfides (2a–s, 3a–j), derived from 16, under the premise
minimizing the structural changes (Figure 2). Finally, a potent homomorpholinylꢀderived
PDK1 inhibitor with enhanced enzymatic, cellular inhibitory activities, and better
performance in vivo, that is, 3a, was identified as a unique lead compound for the
development of antitumor drugs through the selective and covalent inhibition of PDK1.
Figure 2
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 70
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
CHEMISTRY
To further improve the pharmacologic activity of the lead compound 16 and to obtain
novel analogs, chemical modifications were performed in three cycles (Figure 2). First,
thirteen bis(thiazolyl)disulfides 1a–m (Table 1) were designed to determine whether the
thiocarbamyl moiety of 16 was necessary for PDK1 inhibitory activity. Second, we
substituted the morpholine with various steric and polar groups to design eighteen analogs
(2a–f and 2h–s) (Table 2). Third, eleven ring contraction and ring expansion analogs (2g and
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
a–f, Table 2) were prepared to determine whether the ring size affected PDK1 inhibitory
activity.
In this study, all the disulfides were obtained through oxidation of the corresponding
thiols or dithiocarboxylic acids. The synthesis of analogs 1b–m is shown in Scheme 1, and
28
the procedure generally followed the method reported by K. Ramadas. Treatment of
commercially available thiols 4a–c with ammonium persulfate yielded target analogs 1b and
1d–e. Then, nitration of 1b afforded target analog 1c. Ring closure of αꢀbromo ketones 5a–g
with ammonium dithiocarbamate gave rise to intermediates 4d–k (hydrolysis of 4i produced
4j), and the subsequent oxidation led to target analogs 1f–m.
Scheme 1
Synthesis of target analogs 2a–g and 3a started from commercially available morpholine
derivatives 6a–f and 7 and is shown in Scheme 2. The synthesis generally followed the
29
method developed by Neelakantan. As an exception, target analog 2g used ammonium
ACS Paragon Plus Environment

Page 7 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
persulfate as the oxidant. Furthermore, acid 2c underwent esterification to provide target
analog 2d.
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 2
Scheme 3 depicts the synthetic route for the preparation of target analogs 2h, 2j–k and
30
2
s. Oꢀethylation of compound 8a formed 8b, and chlorination of 8a provided compound 9,
which was further converted to 8c by alcoholysis. Debenzylation of 8a–c was performed by
conventional catalytic hydrogenation using palladium. The resulting secondary amines were
2
treated with CS without further purification, and the following oxidation by ammonium
persulfate afforded target analogs 2h and 2j–k. The synthesis of 2s was used the same route
of 2k, except for the different chiral material 8d.
Scheme 3
31
Synthesis of 2i and 2l–r started from compound 10, as shown in Scheme 4.
Nucleophilic substitution of compound 10 by sodium methoxide, phenols and water
generated intermediates 11a, 11b–d and 11e, respectively. Reaction of 11e with iodoethane or
acetyl chloride in the presence of base produced 11f and 11g. Replacement of the benzyl of
1
1e with tꢀbutyloxy carbonyl led to 12, which was converted to 11h by acylation with
dimethylcarbamic chloride. Deprotection of 11a–h yielded the corresponding secondary
amines, which were used without further purification. Finally, target analogs 2i, 2l–m, and
2p–r were prepared in the same manner described in the preparation of 2a; however, target
analogs 2n–o were obtained by oxidation of the dithiocarboxylic acids, as described in the
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 70
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
preparation of 2h.
Scheme 4
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Target analogs 3b–j were prepared in parallel, as shown in Scheme 5. Nucleophilic
substitution of compound 13 (Supporting Information) by sodium methoxide, phenols and
water gave 14a, 14b–c, and 14d, respectively. The subsequent modification of 14d resulted in
14e–i. Deprotection of intermediates 14a–i led to the free secondary amines, which were used
as nucleophiles in the reaction with CS
2
2
. Oxidation by NaNO /HCl or ammonium persulfate
yielded target analogs 3b–j.
Scheme 5
RESULTS AND DISCUSSION
Design and Synthesis of Disulfide-Based Analogs of 16
In total, thirtyꢀtwo novel analogs of 16 (1a–m, 2a–s, and 3a–j) were designed and
synthesized. Their chemical structures are shown in Tables 1–2. These analogs were
synthesized through the routes outlined in Schemes 1–5, and the details of the synthetic
procedures and structural characterization are described in the Experimental Section. All
analogs were confirmed to be ≥95% purity (Table S1, Supporting Information).
Tables 1–2
The IC Values of Analogs 1a–m, 2a–r, and 3a–j against PDK1
5
0
For the primary assay, the disulfideꢀbased analogs were analyzed for their inhibitory
activity against PDK1 using enzymeꢀlinked immunosorbent assay (ELISA), and 16 was used
ACS Paragon Plus Environment

Page 9 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
as the reference compound. The halfꢀmaximal inhibitory concentration (IC50) values of
the bis(thiazolyl)disulfides (1a–m) are shown in Table 1. Based on the bioisosterism strategy,
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
thiazoles with different substituents were induced to replace the thiocarbamyl moiety in 16.
This set of heteroaryl disulfides displayed promising PDK1 inhibitory activity (IC < 0.5
5
0
17
ꢁ
M), inheriting the activity of 16 (IC50 = 0.049 ꢁM). Fusing aryl or cycloalkyl onto the
thiazole can be tolerated (1a vs 1b–m). For the aryl analogs, introduction of a phenyl ring (1b,
IC50 = 0.379 ꢁM) and pyridine ring (1e, IC50 = 0.476 ꢁM) produced activities similar to 1a
(IC50 = 0.326 ꢁM). An electronꢀwithdrawing group or electronꢀdonating group at the aryl ring
improved the activity (1c–d vs 1b). For the cycloalkyl analogs, the substituent type on the
cycloalkyl ring had little impact on the activity (1h vs 1i–l); however, the introduction of a
large cycloalkyl ring was favorable. In the studied set of the cycloalkyl rings, the potency
increased in the order cycloheptyl > cyclohexyl > cyclopentyl (1m > 1h > 1g). Notably, the
inhibitory activity of analog 1m (IC50 = 0.039 ꢁM) increased a lot compare to that of 1a, even
more potent than that of 16. We also tested the inhibitory activities of two free thiols (the
intermediates 4a and 4i) against PDK1, but no inhibition was observed at up to 10 ꢁM,
demonstrating that the disulfide bond is a critical pharmacophore for PDK1 inhibitory
activity.
Due to the difficulty of the structural modification of morpholinyl rings, we explored a
limited range of substituent types. Analysis of the data in Table 2 revealed that the
introduction of substituents at the 2 or 3 position of the morpholinyl ring of 16 substantially
affected the activity (2a–s). Generally, fused ring substituents (2e–f vs 16) or bulky groups
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 70
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
(
2p–r vs 16) were not beneficial, and the terminal polar substituents (2g–i and 2o vs 16) led
to a loss of activity. To our delight, small steric hydrophobic substituents on the morpholinyl
ring maintained the potency (2a–b, 2d and 2i–m vs 16); especially, the inhibitory activity of
analog 2k (IC = 0.024 ꢁM) was greater than that of 16. Ring contraction (2g vs 16) of the
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
50
morpholinyl ring was not beneficial, but ring expansion of the morpholinyl ring led to a
homomorpholinyl analog (3a, IC50 = 0.026 ꢁM) was favorable. Moreover, the effects of
substituents on the homomorpholinyl ring were similar to the morpholinyl ring analogs.
SAR of Disulfide-Based Analogs
The SAR analysis of a set of thirtyꢀone disulfideꢀbased analogs provided important
insight into the essential structural requirements for effective PDK1 inhibition. A preliminary
analysis base on the IC50 data shown in Tables 1–2 reveals some noteworthy observations of
the SAR for analogs 1a–m, 2a–r, and 3a–j: (1) the disulfide bond rather than the
thiocarbamyl moiety of 16 is indispensable for PDK1 inhibitory activity; replacement of the
thiocarbamyl moiety with a thiazolyl ring can be tolerated (16 vs 1a–m); (2) fusing a
cycloalkyl (especially a larger ring) onto thiazole is favorable for high potency (1g, 1h, and
1m, Table 1), leading to identification of the most potent PDK1 inhibitor (1m) in the thiazole
series; (3) among the different N,Oꢀcontaining aliphatic rings, the potency increases in the
order homomorpholinyl > morpholinyl > azetidinyl (3a > 16 > 2g); (4) the effects of
substituents on the morpholinyl or homomorpholinyl ring are similar. Generally, fused ring
substituents (2e–f), bulky groups (2p–r) and terminal polar substituents (2g–i and 2o) are not
ACS Paragon Plus Environment

Page 11 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
beneficial, but small steric hydrophobic substituents can maintain, even improve the potency
(
2k vs 16).
Analysis of IC50 data has provided some interesting SARs. Considering the specialty of
covalent inhibitors, IC value alone may be not sufficient for SAR discussion. In order to
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
5
0
validate the SAR concluded above, we introduced the secondꢀorder rate constant of target
inactivation (kinact/K ) as the quantitative assessment of potency. Several representative
compounds closely related to the above SAR were tested for their kinact/K and the resultant
values were incorporated into a rectangular coordinate (Table 3, Figure 3). kinact/K of 16, 2k
and 3a (1.94×10 M s , 4.17×10 M s and 5.14×10 M s , respectively), which filled in
quadrant III, suggested the high affinity and moderate specific reactivity. In contrast, kinact/K
i
i
i
3
ꢀ1 ꢀ1
3
ꢀ1 ꢀ1
3
ꢀ1 ꢀ1
i
3
ꢀ1 ꢀ1
3
ꢀ1 ꢀ1
3
ꢀ1 ꢀ1
3
ꢀ1 ꢀ1
of 2e, 2g, 2h and 2r (0.06×10 M s , 0.07×10 M s , 0.54×10 M s , and 0.03×10 M s ,
respectively), as located in quadrant IV, indicated weak affinity and resultant low inhibitory
3
ꢀ1 ꢀ1
3
ꢀ1 ꢀ1
i
potencies. 1a and 1m (kinact/K = 1.53×10 M s and 3.93×10 M s , respectively) were
located in quadrant I, and their high activities likely stemmed from the strong specific
reactivity whereas the affinity was responsible for the SAR. These results were overall in
agreement with IC50 data indicated activities (Table 2) and the corresponding SAR. Moreover,
i
the two isomers (2k and 2s) did not show difference in both IC50 and kinact/K values (Tables
2–3), indicating that the chirality of compounds had no influence on the enzymatic activities.
Taken together, the SARs are obviously regular. A subtle interplay between
N,Oꢀcontaining aliphatic rings and small steric hydrophobic substituents is critical for potent
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 70
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
PDK1 inhibitory activity. In our limited analogs, 2k and 3a represent the best combination of
these factors, leading to ~2ꢀfold potency improvement compared to the lead compound 16.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 3
Figure 3
Cellular Activity of Representative Analogs
Our chemical efforts based on 16 led to the discovery of two types of novel
disulfideꢀbased PDK1 inhibitors. Encouraged by the favorable PDK1 inhibitory activity in
the biochemical kinase activity assay, we assessed the cellular activity using immunoblotting
analysis, which examined the impact on PDHE1 phosphorylation downstream from PDK1.
All analogs were subgrouped at a cutoff IC50 value of 0.3 ꢁM, and the twentyꢀtwo best
analogs (1c–d, 1f–m, 2a–d, 2f, 2i–k, 2m–n, 2p, and 3a) were selected and tested against
A549 lung cancer cell lines. Firstly, we performed the immunoblotting analysis of the 22
analogs at 10 ꢁM. DCA and 16 were used as positive controls. As shown in Figure 4A,
among these samples, analogs 2a–b, 2d, 2i–k, 2m–n, and 3a displayed the strongest cellular
potency. These nine analogs almost abolished PDHE phosphorylation at both S293 and S232,
while the others showed marginal cellular activity or were only effective at one site (analogs
2f and 2p).
For further screening, these nine potent analogs were investigated at much lower
concentrations. As shown in Figure 4B, when the concentration was 1 ꢁM, all the analogs,
except for 2i, maintained similar strong inhibition of phosphorylation at both sites. We then
ACS Paragon Plus Environment

Page 13 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
adjusted the concentration to 0.3 ꢁM, and the different cellular potency of the analogs could
be easily distinguished. Among the nine analogs, only 2k and 3a maintained obvious activity
comparable to that of 16. Considering the good performance in cellꢀbased assays, together
with 16, analogs 2k and 3a were selected for the next phase of investigation. For a more
accurate comparison of the three analogs in the cell lines, we performed a concentration
gradient assay (0.1 ꢁM, 0.3 ꢁM and 1 ꢁM), and the semiꢀquantitative analysis was performed
by GrayꢀScale. As illustrated in Figure 4C, the cellular potency increased against a
concentration gradient. Analog 3a exhibited the best inhibition at each concentration at both
sites. Although analog 2k was slightly weaker than 3a, it exhibited stronger inhibition at S293
than that of 16.
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
2k and 3a Inhibit PDK1 by Covalent Interaction
Taking the similar chemical structures and comparable cellular activities into
consideration, we believed that analogs 2k and 3a inhibited PDK1 through the same
17
mechanism as 16, i.e., covalently binding to a conserved cysteine residue (C240). To
confirm this hypothesis, we introduced the reducing thiol group to interrupt the inferential
formation of disulfide bonds between PDK1 and the analogs by adding glutathione (GSH) to
1
7
the system. As with 16, an increasing loss of activity was observed as the GSH
concentration increased in the experiments with analogs 2k and 3a (Figure 5), while DCA
17
17
was not affected. This result was consistent with our previous findings and suggested that
2k and 3a exerted strong inhibitions on the PDK1 activity through covalent interactions.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 70
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Figure 5
2
k and 3a Are Subtype-Selective and Potent PDK1 Inhibitors
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Due to the most significant role of PDK1 among the PDK subtypes in cancer therapy,
we measured the kinact/K value of 16, 2k and 3a against all PDK isoforms using a
i
biochemical enzymatic assay. As shown in Table 4, all three compounds exhibited more or
less selectivity towards PDK1 over other PDK subtypes. Compounds 2k and 3a displayed
better subtypeꢀselectivity than 16. Among them, 3a exhibited ~40ꢀfold selectivity towards
PDK1 over PDK2 and 3, better than 2k and even 10ꢀfold improvement than that of 16. All
three compounds did not inhibit PDK4 at a concentration up to 10 ꢁM. The control DCA had
no subtypeꢀselectivity for PDK1
a) are the first generation of subtypeꢀselective and potent PDK1 inhibitors.
Table 4
–4. To the best of our knowledge, these two analogs (2k and
3
Molecular Docking of 16, 2k and 3a with PDK1–4
To better understand the binding mode of 16 analogs with PDKs, we carried out
molecular docking of 16, 2k and 3a with PDK1–4, respectively. Firstly, 16, 2k and 3a all
suited well in the binding pocket near C240 of PDK1 (Figure 6A), which provided a good
model to explain the improved activities of 2k and 3a towards PDK1. Apparently, the
substituted morpholinyl of 2k and homomorpholinyl ring of 3a allowed stronger hydrophobic
interactions with C240 surrounding residues, with more residues (A236, L239, N246, K282,
M285, P351, L352, and A293) participating in 2k and 3a interactions compared with
1
6ꢀPDK1 binding model. In the meanwhile, distance between the thiol of C240 and the
ACS Paragon Plus Environment

Page 15 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
thiocarbonyl of 2k (3.2 Å) and 3a (3.0 Å) was smaller than that of 16 (3.8 Å), suggesting a
higher possibility to form a disulfide bond (Figure 6A). Secondly, the docking models may
also explain the subtypeꢀselectivity of 16 analogs for PDK1. The binding pockets of 16/2k/3a
in PDK2 and PDK3 (Figures 6B–C) appeared less suitable (i.e. narrow) for compounds entry.
For example, the binding pockets of 16/2k/3a in PDK3 showed rather small hole shapes,
indicating that the compound need to overcome the solvent repulsion before binding and
fitting into the enzyme pockets (Figure 6C). These findings were consistent with 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
i
measured K values of 16/2k/3a towards PDK2 and PDK3 versus PDK1 (Table 4). In the case
of the irreversible binding, the kinact value reflects the reaction rate. In addition to the
distances, many different factors contributed to kinact, including intrinsic chemical reactivity,
32
reactant alignment, enforced local concentration, etc. As show in Figures 6A
–
B, the
distances between the thiocarbonyl of these compounds and the thiol of C148 in PDK2 were
much bigger than that of C240 in PDK1, consistent with the lower kinact of 16/2k/3a towards
PDK2. Similarly, the distances of 16 and 2k to the cysteine in PDK1 and PDK3 were 3.8/3.5
Ǻ and 3.2/3.1 Ǻ, respectively, which were consistent with the higher kinact of 16/2k towards
PDK3 over PDK1 (Table 4). One exception was 3a, whose distance to the cysteine in PDK1
was closer than that of PDK3 (3.0/3.2 Ǻ, respectively), but the kinact value towards PDK3 was
obvious higher. We speculated that the binding pose of 3a in PDK3 may offer an
advantageous trajectory of the thiocarbonyl to the cysteine. Altogether, these results
i
suggested the lower kinact/K value of 16 analogs towards PDK2/3. Likewise, there were no
obvious binding pockets for 16/2k/3a in PDK4 and appeared quite low possibility to form the
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 70
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
disulfide bonds (Figure 6D), explaining the lack of activity against PDK4.
Figure 6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2k and 3a Alter Glucose Metabolic Pathways
Inhibition of PDK1 in cancer cells leads to the recovery of the impaired mitochondrial
function, which switches the metabolism of pyruvate from cytoplasmꢀbased lactate
production to mitochondriaꢀbased oxidative phosphorylation. This metabolic pathway change
is characterized by a decrease in ECAR and an increase in ROS. We tested the influence of
2k and 3a on the two indicators of metabolic pathway change and used 16 as the reference
substance. As shown in Figure 7, ECAR was significantly decreased by analogs 2k and 3a in
a doseꢀdependent manner, which was consistent with the expected reduction of lactate
production in A549 cells. In line with the accelerated oxidative phosphorylation, ROS was
increased by analogs 2k and 3a with increasing concentration. The regulative results of
analogs 2k and 3a for ECAR and ROS were largely the same as lead compound 16. These
data suggested that analogs 2k and 3a affected cell metabolism in the same way as 16.
Figure 7
2k and 3a Exhibit Antiproliferative Effects in Cancer Cells but not in Normal Cells
Encouraged by the above observations, we tested the antiproliferative effects of analogs
2k and 3a by the CCK8 assay. Two cancer cells, A549 (human lung cancer cells) and Kelly
(human neural cancer cells), and two normal cells, GM00637 (human skin fibroblast cells)
and LO2 (human normal liver cells), were employed. DCA was used as the positive control.
All three compounds doseꢀdependently inhibited the proliferation of cancer cells and had no
ACS Paragon Plus Environment

Page 17 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
obvious influence on normal cells (IC50 > 10 ꢁM, Table 5). As the best analog in the
immunoblotting analysis (Figure 4), analog 3a sustained its advantage, and inhibited Kelly
cell proliferation with a promising IC50 of 0.057 ꢁM (Table 5), leading to ~5ꢀfold potency
improvement compared to the lead compound 16, whereas analog 2k possessed less than
satisfactory potency (IC50 = 1.426 ꢁM). In addition to Kelly cells, analog 3a effectively
inhibited A549 cell proliferation (IC50 = 0.225 ꢁM, Table 5), to a greater extent than both 16
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
(
IC50 = 0.480 ꢁM) and analog 2k (IC50 = 0.957 ꢁM). Of note, compounds 2k and 3a barely
inhibited the growth of two normal cell lines.
Table 5
Compounds 2k and 3a Attenuates Tumor Growth in A549 Xenograft Models
To further prove the therapeutic potential of compound 3a, we compared antitumor
efficacy of 16, 2k and 3a in A549 subcutaneous xenograft mice models (Figure 8).
Tumorꢀbearing mice were randomly subgrouped and intraperitoneally received vehicle
control or 16 analogs at 40 and 80 mg/kg per day, respectively. A 21ꢀday continuous
treatment of 16 and 3a considerably reduced tumor volume at both dosages compared with
the vehicle control (16, TGI = 27.7% at 40 mg/kg, TGI = 50.4% at 80 mg/kg; 3a, TGI = 26.2%
at 40 mg/kg, TGI = 47.8% at 80 mg/kg; Figures 8A and 8C), but the effect of 2k was
marginal (TGI = 11.0% at 40 mg/kg, TGI = 20.0% at 80 mg/kg; Figure 8B). Likewise,
endpoint tumor weights in the 16 and 3a treated group were significantly less than those
treated with vehicle control (Figure 8D). In this experiment, compound 3a showed similar
antitumor efficacy as 16. The twoꢀfold improvement in the enzymatic activity of 3a (Table 4)
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 70
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
was not translated to in vivo efficacy in inhibiting tumor growth (Figure 8E). None of the
results resulted in mice bodyweight loss even at higher dosage.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
CONCLUSIONS
In summary, we have discovered two novel types of PDK1 inhibitors, i.e.,
bis(thiazolyl)disulfides (1a–m) and bis(thiocarbonyl)disulfides (2a–s, 3a–j). On the basis of
the structure of the lead compound 16, thirtyꢀone completely novel disulfideꢀbased analogs
have been synthesized and tested by a PDK1 enzymatic inhibitory assay. Twentyꢀtwo analogs
(1c–d, 1f–m, 2a–d, 2f, 2i, 2j–k, 2m–n, 2p, and 3a) showed potent PDK1 inhibitory activity
(IC50 < 0.3 ꢁM). Preliminary SARs were obtained base on IC50 data and were further
validated by kinact/K , showing that the disulfide bond and bulky N,Oꢀcontaining aliphatic
i
rings of 16 are indispensable for PDK1 inhibitory activity, and terminal small steric
hydrophobic substituents can substantially improve the potency.
The PDK1 inhibition assays at the cellular level further confirmed that nine analogs (2a–
b, 2d, 2i–k, 2m–n, and 3a) were potent PDK inhibitors and effectively abolished PDHE
phosphorylation at both S293 and S232. Among the nine analogs, analog 3a exhibited the
best inhibition (down to 0.1 ꢁM) at both S293 and S232 and the PDKꢀselective profile of 3a
was observed. 3a was also identified to inhibit PDK1 by covalently binding to a conserved
cysteine residue. Compound 3a demonstrated approximately 2 times higher potency in
i
inhibiting PDK1 than that of 16 according to the kinact/K value. In the meanwhile, 3a had a
~
40ꢀfold selectivity for PDK1 over PDK2 and 3 and showed no inhibition of PDK4 at a
ACS Paragon Plus Environment

Page 19 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
concentration up to 10 ꢁM. Subtypeꢀselectivity of 3a was 10 times higher than that of 16.
Further detailed analysis of molecular docking of 16, 2k and 3a with PDK1ꢀ4 gained a better
understanding of the subtypeꢀselectivity for PDK1ꢀ4 of these compounds.
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 3a significantly altered glucose metabolic profile in a doseꢀdependent
manner in A549 cells by decreasing ECAR and increasing ROS. Moreover, the in
vitro antiproliferative assays showed that 3a had more potent inhibitory effects than that of 16
on the proliferation of two human cancer cell lines (A549: IC50 = 0.225
ꢁ
M and Kelly: IC50
M), without inhibition activity against normal cell lines GM00637 and LO2 (IC50 > 10
M). Together with 16, compound 3a also showed significant antitumor activity in the in vivo
=
0.057 ꢁ
ꢁ
A549 xenograft models which was better than that of 2k. Altogether, these data demonstrate
that 3a may be a promising starting point for the discovery of potential therapeutic drugs to
specifically inhibit PDK1. Considering the small number of PDK1 inhibitors reported thus far,
3a could be regarded as the first generation of subtypeꢀselective and covalent PDK1
inhibitors. Further characterization of 3a as a potential cancer therapeutic drug will be
performed in the future.
EXPERIMENTAL SECTION
Chemistry. General Chemistry
The synthetic starting materials, reagents and solvents were purchased from Alfa Aesar,
Acros, Adamasꢀbeta, Energy Chemical, J&K, Shanghai Chemical Reagent Co. and TCI at the
highest commercial quality and were used without further purification. Analytical thinꢀlayer
chromatography (TLC) was performed on HSGF 254 (150−200 ꢁm thickness; Yantai Huiyou
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 70
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Co., China), and the components were visualized by observation under UV light (254 nm and
3
65 nm). Melting points were determined on an SGW Xꢀ4 melting point apparatus without
correction. Optical rotation of the chiral compounds was measured using a RUDOLPH
AUTOPOL V) automatic polarimeter. The products were purified by recrystallization or
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(
column chromatography on silica gel (200–300 mesh). The reaction yields were not
optimized. Nuclear magnetic resonance (NMR) spectroscopy was performed on a Bruker
AMXꢀ400 NMR using deuterated chloroform (CDCl
3 3
), deuterated methanol (CD OD), or
deuterated dimethyl sulfoxide (DMSOꢀd ) as the solvent. The chemical shifts were reported
6
in parts per million (ppm, δ) downfield from tetramethylsilane. The proton coupling patterns
were described as singlet (s), doublet (d), triplet (t), quartet (q), multiplet (m), and broad (br).
Lowꢀ and highꢀresolution mass spectra (LRMS and HRMS) were performed with electric and
electrospray ionization (EI and ESI) with a Finnigan MATꢀ95 and an LCQꢀDECA
spectrometer. HPLC analysis of compounds 1a–m, 2a–s, and 3a–j was performed on an
Agilent 1100 with a quaternary pump and a diodeꢀarray detector (DAD). The peak purity was
checked by UV spectra. All analogs were confirmed to be ≥95% purity.
Bis(thiazol-2-yl)disulfide (1a)
This analog is commercially available from J&K Scientific Ltd.
Bis(benzo[d]thiazol-2-yl)disulfide (1b)
An aqueous solution of ammonium persulfate (342 mg, 1.5 mmol in 5 mL water) was
added dropwise at room temperature to a stirred solution of benzo[d]thiazoleꢀ2ꢀthiol (4a, 167
mg, 1 mmol) in THF (10 mL), and the reaction was monitored by TLC. After 0.5 h, the
mixture was separated between EtOAc (50 mL) and water (50 mL). The organic layer was
washed with brine, dried over anhydrous sodium sulfate, and concentrated under reduced
ACS Paragon Plus Environment

Page 21 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
pressure. The residue was then purified by column chromatography on silica gel to afford
1
analog 1b as a white solid (150 mg). Yield: 90%. M.p. 179–181 ºC. HꢀNMR (400 MHz,
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
CDCl
3
): δ 7.94 (d, J = 8.2 Hz, 2H), 7.77 (d, J = 8.0 Hz, 2H), 7.49–7.45 (m, 2H), 7.38–7.34
+
(m, 2H); HRMS (EI) m/z calcd. C H N S (M ) 331.9570, found 331.9576.
14
8
2 4
Bis(6-nitrobenzo[d]thiazol-2-yl)disulfide (1c)
Fuming nitric acid (0.016 mL, 0.35 mmol) was added to a stirred solution of 1b (100 mg,
0.3 mmol) in concentrated sulfuric acid (5 mL) maintained at 0 ºC. The solution was stirred at
room temperature for 1 h and was then added to ice water (50 mL). The precipitated product
was collected and crystallized from EtOAc to give analog 1c as a yellow solid (90 mg). Yield:
1
71%. M.p. 218–220 ºC; HꢀNMR (400 MHz, DMSOꢀd
6
): δ 9.13 (d, J = 2.3 Hz, 2H), 8.35 (dd,
+
J = 9.0, 2.3 Hz, 2H); HRMS (EI) m/z calcd. C14
Bis(6-ethoxybenzo[d]thiazol-2-yl)disulfide (1d)
d was obtained in the same manner as described in the preparation of 1b from
ꢀethoxybenzo[d]thiazoleꢀ2ꢀthiol (4b) as a white solid. Yield: 85%. M.p. 132–135 ºC.
H
6
N
4
O
4
S
4
(M ) 421.9272, found 421.9277.
1
6
1
3
HꢀNMR (400 MHz, CDCl ): δ 7.83 (d, J = 9.0 Hz, 2H), 7.65 (d, J = 2.2 Hz, 2H), 7.10 (dd, J
= 9.0, 2.4 Hz, 2H), 4.06 (q, J = 6.9 Hz, 4H), 1.34 (t, J = 6.9 Hz, 6H); HRMS (EI) m/z calcd.
+
C
18
H
16
N
2
O
2
S
4
(M ) 420.0095, found 420.0096.
Bis(thiazolo[5,4-b]pyridin-2-yl)disulfide (1e)
1
e was obtained in the same manner as described in the preparation of 1b from
1
thiazolo[5,4ꢀb]pyridineꢀ2ꢀthiol (4c) as a white solid. Yield: 60%. M.p. 147–148 ºC. HꢀNMR
(400 MHz, DMSOꢀd
6
): δ 8.61 (d, J = 4.6 Hz, 2H), 8.39 (d, J =8.3 Hz, 2H), 7.65–7.58 (m, 2H);
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 70
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
+
HRMS (EI) m/z calcd. C12
H
6
N
4
S
4
(M ) 333.9475, found 333.9476.
Bis(6,7-dihydro-4H-pyrano[4,3-d]thiazol-2-yl)disulfide (1f)
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Ammonium dithiocarbamate (110 mg, 1 mmol) was added to a solution of
3
ꢀbromooxanꢀ4ꢀone (5a, 179 mg, 1 mmol) in EtOH (20 mL), and the mixture was heated to
reflux for 1 h and concentrated under reduced pressure. The residue was separated between
EtOAc (50 mL) and water (50 mL), the organic layer was washed with brine and dried over
anhydrous sodium sulfate, and the solvent was removed to give the crude product
6
,7ꢀdihydroꢀ4Hꢀpyrano[4,3ꢀd]thiazoleꢀ2ꢀthiol (4d, 180 mg), which was used in the next step
without further purification.
Analog 1f was obtained in the same manner as described in the preparation of 1b from
1
the crude product 4d as a white solid. Yield: 52% over two steps. M.p. 125–127 ºC. HꢀNMR
(400 MHz, DMSOꢀd
6
): δ 4.75 (s, 4H), 3.96–3.88 (m, 4H), 2.83–2.75 (m, 4H); HRMS (EI)
+
m/z calcd. C12
H
12
N
2
O
2
S
4
(M ) 343.9782, found 343.9781.
Bis(5,6-dihydro-4H-cyclopenta[d]thiazol-2-yl)disulfide (1g)
g was obtained in the same manner as described in the preparation of 1f from
1
2ꢀbromocyclopentanone (5b) as a white solid. Yield: 70% over two steps. M.p. 115–118 ºC.
1
HꢀNMR (400 MHz, DMSOꢀd
6
): δ 2.91 (t, J = 7.2 Hz, 4H), 2.77 (t, J = 7.4 Hz, 4H), 2.46–
+
2.37 (m, 4H); HRMS (EI) m/z calcd. C12
H
12
N
2
S
4
(M ) 311.9883, found 311.9884.
Bis(4,5,6,7-tetrahydrobenzo[d]thiazol-2-yl)disulfide (1h)
1h was obtained in the same manner as described in the preparation of 1f from
2
ꢀbromocyclohexanone (5c) as a white solid. Yield: 85% over two steps. M.p. 102–105 ºC.
ACS Paragon Plus Environment

Page 23 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
HꢀNMR (400 MHz, DMSOꢀd
6
): δ 2.78–2.71 (m, 4H), 2.71–2.63 (m, 4H), 1.84–1.72 (m, 4H);
+
HRMS (EI) m/z calcd. C14
H
16
N
S
2 4
(M ) 340.0196, found 340.0197.
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
Bis(6-methyl-4,5,6,7-tetrahydrobenzo[d]thiazol-2-yl)disulfide (1i)
i was obtained in the same manner as described in the preparation of 1f from
1
2ꢀbromoꢀ4ꢀmethylcyclohexanone (5d) as a light yellow solid. Yield: 80% over two steps. M.p.
1
1
10–112 ºC. HꢀNMR (400 MHz, CDCl
3
): δ 2.93–2.79 (m, 4H), 2.79–2.68 (m, 2H), 2.40–
2.29 (m, 2H), 1.94–1.79 (m, 4H), 1.56–1.43 (m, 2H), 1.10 (s, 3H), 1.08 (s, 3H); HRMS (EI)
+
m/z calcd. C16
H
20
N
2
S
4
(M ) 368.0509, found 368.0508.
Bis(6,6-dimethyl-4,5,6,7-tetrahydrobenzo[d]thiazol-2-yl)disulfide (1j)
j was obtained in the same manner as described in the preparation of 1f from
1
2ꢀbromoꢀ4,4ꢀdimethylcyclohexanone (5e) as a light yellow solid. Yield: 77% over two steps.
1
M.p. 110–111 ºC. HꢀNMR (400 MHz, CDCl
3
): δ 2.83–2.75 (m, 4H), 2.53 (s, 4H), 1.67–1.58
+
(
m, 4H), 1.03 (s, 12H); HRMS (EI) m/z calcd. C18
H
24
N
2
S
4
(M ) 396.0822, found 396.0829.
Bis(6-ethoxycarbonyl-4,5,6,7-tetrahydrobenzo[d]thiazol-2-yl)disulfide (1k)
k was obtained in the same manner as described in the preparation of 1f from ethyl
1
3ꢀbromoꢀ4ꢀoxocyclohexanecarboxylate (5f) as a white solid. Yield: 67% over two steps. M.p.
1
1
29–131 ºC. HꢀNMR (400 MHz, CDCl
3
): δ 4.18 (q, J = 7.0 Hz, 4H), 3.08–2.99 (m, 4H),
2.99–2.87 (m, 2H), 2.87–2.74 (m, 4H), 2.32–2.21 (m, 2H), 2.07–1.90 (m, 2H), 1.28 (t, J = 7.1
+
Hz, 6H); HRMS (ESI) m/z calcd. C H N O S [M+H] 485.0697, found 485.0700.
2
0
25
2
4 4
Bis(6-carboxyl-4,5,6,7-tetrahydrobenzo[d]thiazol-2-yl)disulfide (1l)
A mixture of ethyl 2ꢀmercaptoꢀ4,5,6,7ꢀtetrahydrobenzo[d]thiazoleꢀ6ꢀcarboxylate (243
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 70
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
2
mg, 1 mmol) and LiOH (100 mg, 4 mmol) was combined in 4 mL of THF and H O solvent
(V
THF:VH2O = 3:1) at room temperature for 0.5 h, and was acidified to pH 5.5. Then, the
precipitate was collected, washed with O, and dried to afford
ꢀmercaptoꢀ4,5,6,7ꢀtetrahydrobenzo[d]thiazoleꢀ6ꢀcarboxylic acid (4j, 200 mg, 93%) as a
white solid.
l was obtained in the same manner as described in the preparation of 1f from 4j as a
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
H
2
2
1
1
white solid. Yield: 95%. M.p. 222–223 ºC. HꢀNMR (400 MHz, CDCl
3
): δ 12.48 (brs, 2H),
3
2
.07–2.97 (m, 2H), 2.95–2.83 (m, 2H), 2.83–2.68 (m, 6H), 2.19–2.06 (m, 2H), 1.93–1.79 (m,
+
H); HRMS (EI) m/z calcd. C H N O S (M ) 427.9993, found 427.9991.
1
6
16
2
2 4
Bis(5,6,7,8-tetrahydro-4H-cyclohepta[d]thiazol-2-yl)disulfide (1m)
m was obtained in the same manner described in the preparation of 1f from
ꢀbromocycloheptanone (5g) as a white solid. Yield: 60% over two steps. M.p. 87–88 ºC.
1
2
1
HꢀNMR (400 MHz, CDCl
3
): δ 2.98–2.89 (m, 4H), 2.83–2.75 (m, 4H), 1.89–1.81 (m, 4H),
+
1
.77–1.64 (m, 4H); HRMS (EI) m/z calcd. C H N S (M ) 368.0509, found 368.0510.
16 20
2 4
Bis(2-methylmorpholin-4-yl-thiocarbonyl)disulfide (2a)
A mixture of 2ꢀmethylmorpholine (6a, 505 mg, 5 mmol), carbon disulfide (0.6 mL, 10
mmol), potassium hydroxide (616 mg, 11 mmol), and water (50 mL) was stirred at 50 ºC
overnight. After cooling to room temperature, the solution was mixed with sodium nitrite
(345 mg, 5 mmol) and MeOH (0.3 mL), and concentrated HCl (1 mL) was added dropwise at
0 ºC under vigorous stirring. After 0.5 h, the system was separated between EtOAc (200 mL)
and water (100 mL), and the organic layer was washed with brine, dried over anhydrous
ACS Paragon Plus Environment

Page 25 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
sodium sulfate, and concentrated under reduced pressure. The residue was then purified by
column chromatography on silica gel to afford analog 2a as a colorless oil (818 mg). Yield:
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
9
3%. HꢀNMR (400 MHz, CDCl
3
): δ 5.63–4.65 (m, 4H), 4.10–3.92(m, 2H), 3.88–3.65 (m,
4
H), 3.62–3.36 (m, 2H), 3.32–2.94 (m, 2H), 1.27 (s, 3H), 1.26(s, 3H); HRMS (EI) m/z calcd.
+
C
12
H
20
N
2
O
2
S
4
(M ) 352.0408, found 352.0405.
Bis((2R,6S)-2,6-dimethylmorpholin-4-yl-thiocarbonyl)disulfide (2b)
2b was obtained in the same manner as described in the preparation of 2a from (2S,
25
D
6
R)ꢀ2,6ꢀdimethylmorpholine (6b) as a white solid. Yield: 90%. M.p. 124–125 ºC. [α]
=
1
ꢀ0.010 (c 1, methanol). HꢀNMR (400 MHz, CDCl ): δ 5.30 (br s, 2H), 4.82 (br s, 2H), 3.77
3
+
(
br s, 4H), 3.09 (br s, 2H), 294 (br s, 2H); HRMS (EI) m/z calcd. C14
80.0721, found 380.0723.
Bis(2-carboxyl-morpholin-4-yl-thiocarbonyl)disulfide (2c)
H
24
N
2
O
2
S
4
(M )
3
2c was obtained in the same manner as described in the preparation of 2a from
1
morpholineꢀ2ꢀcarboxylic acid (6c) as a white solid. Yield: 90%. M.p. 112–113 ºC. HꢀNMR
(400 MHz, DMSOꢀd6): δ 13.30 (brs, 2H), 5.14–4.90 (m, 1H), 4.79–4.17 (m, 5H), 4.16–3.73
+
(m, 6H), 3.73–3.63 (m, 2H); HRMS (ESI) m/z calcd. C12
found 434.9784.
H
16
N
2
O
6
S
4
Na [M+Na] 434.9789,
Bis(2-methoxycarbonylmorpholin-4-yl-thiocarbonyl)disulfide (2d)
Concentrated sulfuric acid (0.05 mL) was added to a stirred solution of 2c (412 mg, 1
mmol) in MeOH (50 mL), and the solution was stirred at room temperature for 10 h. The
system was separated between EtOAc (200 mL) and water (100 mL), the organic layer was
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 70
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
washed with water and brine, dried over anhydrous sodium sulfate, and concentrated under
reduced pressure. The residue was then purified by column chromatography on silica gel to
1
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
afford analog 2d as a white solid (380 mg). Yield: 86%. M.p. 150–152 ºC. HꢀNMR (400
MHz, CDCl ): δ 5.50–5.00 (m, 2H), 4.96–4.72 (m, 2H), 4.34 (dd, J = 9.4, 2.9 Hz, 2H), 4.23–
3
+
4
.11 (m, 2H), 3.93–3.72 (m, 12H); HRMS (EI) m/z calcd. C12
found 440.0205.
Bis(3,4-dihydro-2H-benzo[b]morpholin-4-yl-thiocarbonyl)disulfide (2e)
H
20
N
2
O S
2 4
(M ) 440.0204,
2e was obtained in the same manner as described in the preparation of 2a from
1
3
,4ꢀdihydroꢀ2Hꢀbenzo[b]morpholine (6d) as a colorless oil. Yield: 65%. HꢀNMR (400 MHz,
DMSOꢀd6): δ 8.05 (d, J=8.2 Hz, 2H), 7.27 (t, J = 8.0 Hz, 2H), 7.16–7.04 (m, 4H), 4.28 (t, J =
+
4
.8 Hz, 2H), 4.03 (t, J = 4.8 Hz, 2H); HRMS (EI) m/z calcd. C18
found 420.0093.
Bis(trans-octahydro-2H-benzo[b][1,4]oxazin-4-yl-thiocarbonyl)disulfide (2f)
f was obtained in the same manner as described in the preparation of 2a from
H
16
N
2
O
S
2 4
(M ) 420.0095,
2
transꢀoctahydroꢀ2Hꢀbenzo[b][1,4]oxazine (6e) as a light yellow solid. Yield: 70%. M.p. 55–
1
56 ºC. HꢀNMR (400 MHz, CDCl
3
): δ 5.27–5.16 (m, 2H), 4.36–4.24 (m, 2H), 4.23–4.15 (m,
2H), 4.10–4.06 (m, 2H), 4.00–3.89 (m, 2H), 3.86–3.76 (m, 2H), 2.88–2.72 (m, 2H), 2.04–
28 2 2 4
1.98 (m, 2H), 1.88–1.78 (m, 4H), 1.27–1.25 (m, 8H); HRMS (EI) m/z calcd. C18H N O S
+
(
M ) 432.1034, found 432.1036.
Bis(3-hydroxyazetidin-1-yl-thiocarbonyl)disulfide (2g)
Carbon disulfide (0.6 mL, 10 mmol) was added to a solution of azetidinꢀ3ꢀol (6f, 730 mg,
ACS Paragon Plus Environment

Page 27 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
0 mmol) in THF (50 mL), and the mixture was stirred at room temperature for 1 h before an
aqueous solution of ammonium persulfate (2280 mg, 10 mmol in 10 mL water) was added.
After 0.5 h, the mixture was separated between EtOAc (200 mL) and water (100 mL). The
organic layer was washed with brine, dried over anhydrous sodium sulfate, and concentrated
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
under reduced pressure. The residue was then purified by column chromatography on silica
1
gel to afford analog 2g as a colorless oil (800 mg). Yield: 54%. HꢀNMR (400 MHz, CDCl
3
):
δ 4.86–4.80 (m, 2H), 4.79–4.71 (m, 2H), 4.64–4.56 (m, 2H), 4.39–4.31 (m, 2H), 4.25–4.18
+
(m, 2H); HRMS (ESI) m/z calcd. C
8
H
12
N
2
O
2
S
4
Na [M+Na] 318.9679, found 318.9678.
Bis((R)-3- hydroxymethylmorpholin-4-yl-thiocarbonyl)disulfide (2h)
R)ꢀ(4ꢀbenzylmorpholinꢀ3ꢀyl)methanol (8a, 207 mg, 1 mmol) was dissolved in MeOH
50 mL) in a hydrogenation reactor. Then, 10% Pd/C (50 mg) was added, the reactor was
(
(
filled with hydrogen gas to a pressure of 1 atm, and the reaction mixture was stirred overnight.
The spent catalyst was filtered through Celite, and the filtrate was evaporated to dryness. The
residue was stirred with CS (114 mg, 1.5 mmol) in THF (10 mL) for 1 h before an aqueous
2
solution of ammonium persulfate (228 mg, 1 mmol in 1 mL water) was added. After 0.5 h,
the mixture was separated between EtOAc (100 mL) and water (100 mL), and the organic
layer was washed with brine, dried over anhydrous sodium sulfate, and concentrated under
reduced pressure. The residue was then purified by column chromatography on silica gel to
25
afford analog 2h as a colorless oil (158 mg). Yield: 80% over two steps. [α]D = ꢀ0.201 (c 1,
1
methanol). HꢀNMR (400 MHz, CDCl
3
): δ 5.69–4.62 (m, 4H), 4.11–3.93 (m, 10H), 3.80–
+
3.56 (m, 6H); HRMS (ESI) m/z calcd. C12
H
20
N
2
O
4
S
4
Na [M+Na] 407.0204, found 407.0197.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 70
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Bis(2-methanolylmorpholin-4-yl-thiocarbonyl)disulfide (2i)
4ꢀBenzylꢀ2ꢀchloromethylmorpholine (10, 2.25 g, 10 mmol), 10 mL of water and 50 mL
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
of formamide were placed in a 250 mL roundꢀbottomed flask with a condenser. The mixture
was heated at 190 ºC overnight. After returning to room temperature, the medium was poured
into 100 mL of iceꢀcold water, and the resulting mixture was basified with 10 M sodium
hydroxide to pH=12. Extraction was conducted with 400 mL EtOAc, and the organic layer
was washed with brine, dried over anhydrous sodium sulfate, and concentrated under reduced
pressure. The residue was then purified by column chromatography on silica gel to afford
1
4ꢀbenzylꢀ2ꢀ(hydroxymethyl)morpholine (11e) as an amber oil (1.8 g). Yield: 87%. HꢀNMR
(
400 MHz, CDCl
m, 4H), 2.76–2.63 (m, 2H), 2.21 (td, J = 11.4, 3.3 Hz, 1H), 2.03 (t, J = 10.8 Hz, 1H).
1e (207 mg, 1 mmol) was dissolved in MeOH (50 mL) in a hydrogenation reactor.
Then, 10% Pd/C (50 mg) was added, the reactor was filled with hydrogen gas to a pressure of
atm, and the reaction mixture was stirred overnight. The spent catalyst was filtered through
3
): δ 7.57–7.22 (m, 5H), 3.94–3.86 (m, 1H), 3.78–3.65 (m, 2H), 3.64–3.46
(
1
1
Celite, and the filtrate was evaporated to dryness. The residue was stirred with carbon
disulfide (0.06 mL, 1 mmol) and potassium hydroxide (61.6 mg, 1.1 mmol) in 10 mL of
water at 50 ºC overnight. After cooling to room temperature, the solution was mixed with
sodium nitrite (34.5 mg, 0.5 mmol) and MeOH (0.1 mL) before concentrated HCl (0.1 mL)
was added dropwise at 0 ºC under vigorous stirring. After 0.5 h, the mixture was separated
between EtOAc (100 mL) and water (100 mL), the organic layer was washed with brine,
dried over anhydrous sodium sulfate, concentrated under reduced pressure. The residue was
ACS Paragon Plus Environment

Page 29 of 70
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
then purified by column chromatography on silica gel to afford analog 2i as a colorless oil
1
(
140 mg). Yield: 73% over two steps. HꢀNMR (400 MHz, CDCl
3
): δ 5.67–4.42 (m, 4H),
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
4
.10–4.00 (m, 2H), 3.84–3.62 (m, 8H), 3.61–3.30 (m, 4H); HRMS (ESI) m/z calcd.
+
C H N O S Na [M+Na] 407.0204, found 407.0205.
12
20
2
4 4
Bis((R)-3-(ethoxymethyl)morpholin-4-yl-thiocarbonyl)disulfide (2j)
2j was obtained in the same manner as described in the preparation of 2h from
2
D
5
(R)ꢀ4ꢀbenzylꢀ3ꢀ(ethoxymethyl)morpholine (8b) as a light yellow oil. Yield: 75%. [α]
=
1
ꢀ0.185 (c 1, methanol). HꢀNMR (400 MHz, CDCl
3
): δ 5.74–4.50 (m, 4H), 4.15–4.09 (m, 2H),
4
.04–3.94 (m, 2H), 3.93–3.31 (m, 14H), 1.21 (t, J = 6.8 Hz, 6H); HRMS (EI) m/z calcd.
+
C
16
H
28
N
2
O
4
S
4
(M ) 440.0932, found 440.0931.
Bis((R)-3-(methoxymethyl)morpholin-4-yl-thiocarbonyl)disulfide (2k)
k was obtained in the same manner as described in the preparation of 2h from
R)ꢀ4ꢀbenzylꢀ3ꢀ(methoxymethyl)morpholine (8c) as a light yellow solid. Yield: 75%. M.p.
2
(
25
1
1
00–102 ºC. [α]D = ꢀ0.094 (c 1, methanol). HꢀNMR (400 MHz, CDCl ): δ 5.74–4.50 (m,
3
4H), 4.17–4.06 (m, 2H), 4.02–4.01 (m, 2H), 3.89–3.52 (m, 10H), 3.43 (s, 6H); HRMS (EI)
+
m/z calcd. C14
H
24
N
2
O
4
S
4
(M ) 412.0619, found 412.0621.
Bis(2-(ethoxymethyl)morpholin-4-yl-thiocarbonyl)disulfide (2l)
Sodium hydride (60%, 120 mg, 3 mmol) was added slowly to a stirred solution of
(4ꢀbenzylmorpholinꢀ2ꢀyl)methanol (11e, 207 mg, 1 mmol) in DMF (10 mL). The resulting
mixture was stirred for 20 minutes at room temperature. Subsequently, iodoethane (300 mg, 2
mmol) was added to the resulting suspension, followed by stirring at 80 ºC overnight. The
ACS Paragon Plus Environment