Discovery of carbazole derivatives as novel allosteric MEK inhibitors by pharmacophore modeling and virtual screening

Article abstract of DOI:10.1016/j.ejmech.2019.06.027

We report in this work the discovery of novel allosteric MEK inhibitors by pharmacophore modeling and virtual screening. Two out of 13 virtual hit compounds were identified as MEK kinase inhibitors using a MEK1 binding assay. Structural derivations on the

Full text of DOI:10.1016/j.ejmech.2019.06.027

Accepted Manuscript
Discovery of carbazole derivatives as novel allosteric MEK inhibitors by
pharmacophore modeling and virtual screening
Dandan Xi, Yan Niu, Hongyue Li, Stefan M. Noha, Veronika Temml, Daniela
Schuster, Chao Wang, Fengrong Xu, Ping Xu
PII:
S0223-5234(19)30553-7
DOI:
https://doi.org/10.1016/j.ejmech.2019.06.027
EJMECH 11430
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 25 February 2019
Revised Date: 27 May 2019
Accepted Date: 10 June 2019
Please cite this article as: D. Xi, Y. Niu, H. Li, S.M. Noha, V. Temml, D. Schuster, C. Wang, F. Xu, P.
Xu, Discovery of carbazole derivatives as novel allosteric MEK inhibitors by pharmacophore modeling
and virtual screening, European Journal of Medicinal Chemistry (2019), doi: https://doi.org/10.1016/
j.ejmech.2019.06.027.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT
Discovery of Carbazole Derivatives as Novel Allosteric MEK
Inhibitors by Pharmacophore Modeling and Virtual
Screening
Dandan Xi1, Yan Niu1*, Hongyue Li1, Stefan M. Noha2, Veronika Temml2, Daniela
Schuster2,3*, Chao Wang1, Fengrong Xu1, Ping Xu1*
1Department of Medicinal Chemistry, School of Pharmaceutical Sciences, Peking University
Health Science Center, 38 Xueyuan Road, Beijing 100191, China.
2Institute of Pharmacy/Pharmaceutical Chemistry and Center for Molecular Biosciences Innsbruck,
University of Innsbruck, Innrain 80-82, 6020 Innsbruck, Austria
3Institute of Pharmacy, Department of Pharmaceutical and Medicinal Chemistry, Paracelsus
Medical University Salzburg, Strubergasse 21, 5020 Salzburg, Austria
*Corresponding authors.
Tel: +86 10 82805281. Email: yanniu@bjmu.edu.cn (Y. Niu),
Tel: +43 512 50758253. Email: daniela.schuster@uibk.ac.at (D. Schuster),
Tel: +86 10 82802632. Email: pingxu@bjmu.edu.cn (P. Xu)
Abstract
We report in this work the discovery of novel allosteric MEK inhibitors by
pharmacophore modeling and virtual screening. Two out of 13 virtual hit compounds
were identified as MEK kinase inhibitors using a MEK1 binding assay. Structural
derivations on the hit compound M100 (IC50 = 27.2 ± 4.5 µM in RAF-MEK
cascading assay) by substituent transformation and bioisosterism replacement have
led to the synthesis of a small library of carbazoles. The enzymatic studies revealed
the preliminary structure-activity relationships and the derivative 22k (IC50 = 12.8 ±
0.5 µM) showed the most potent inhibitory effect against Raf-MEK cascading.
Compound 7 was discovered as toxic as M100 to tumor cells whereas safer to
HEK293 cells (IC50 > 100 µM) than M100 (IC50 = 8.9 ± 2.0 µM). It suggests that
carbazole is a good scaffold for the design of novel MEK inhibitors for therapeutic
uses. More importantly, the developed pharmacophore model can serve as a reliable
criterion in novel MEK inhibitor discovery.
1 / 45

ACCEPTED MANUSCRIPT
Keywords: pharmacophore modeling; virtual screening; MEK inhibitor; carbazoles;
bioisosterism
1. Introduction
Mitogen-activated protein kinases (MAPKs) are a highly conserved family of
serine/threonine protein kinases involved in a variety of fundamental cellular events
such as survival, proliferation, differentiation, stress response, and apoptosis[1]. In
mammals, more than a dozen MAPK enzymes have been identified, among which the
best known ones are the conventional MAPKs, including the extracellular
signal-regulated kinases 1 and 2 (ERK1/2), c-Jun amino-terminal kinases 1 to 3
(JNK1 to 3), p38 (α, β, γ, and δ), and ERK5 families[2]. Each group of conventional
MAPKs is composed of a set of three kinases, a MAPK, a MAPK kinase (MAPKK),
and a MAPKK kinase (MAPKKK), which in turn phosphorylate and activate MAPKs
relaying extracellular signals to intracellular responses.
The ERK1/2 signaling pathway is, by far, the best-characterized MAPK pathway
consisting of the MAPKKKs A-Raf, B-Raf, and Raf-1, the MAPKKs MEK1 and
MEK2, and the MAPKs ERK1 and ERK2. This pathway plays a central role in the
control of cell survival and proliferation. Growth factors and mitogens use the
Ras/Raf/MEK/ERK signaling cascade to transmit signals from their receptors to
regulate gene expression and prevent apoptosis[3]. Components of the ERK1/2
signaling cascade are frequently mutated and highly up-regulated in cancer cells, with
approximately 1/3 of human tumors expressing a constitutively activated mutant of
RAS[4] and approximately 8% of tumors expressing an activated form of BRAF[5, 6]
.
These findings prompted the development of small molecule inhibitors targeting the
ERK1/2 signaling pathway as anti-cancer agents[7]. So far, two RAF inhibitors
(vemurafenib and dabrafenib) and three MEK inhibitors (trametinib, cobimetinib,
binimetinib) have been approvedby the FDA of US for the treatment of melanoma
expressing the mutated RAF paralogue BRAF (V600E or V600K )[8-10]. In addition,
selumetinib, a highly selective MEK inhibitor has been granted Orphan Drug
2 / 45

ACCEPTED MANUSCRIPT
Designation for adjuvant treatment of patients with stage III or IV differentiated
thyroid cancer[11]
.
MEK1 and MEK2 are dual-specificity tyrosine/threonine proteinkinases, which
play an integral role in the Ras/Raf/MEK/ERK signaling module[12].They are
attractive targets for ERK1/2 pathway modulation because ERK1/2 are known to be
the only substrates for MEK1/2, and MEK1/2 specific modulator could thus confine
its function within the ERK1/2 pathway leaving other pathways unaffected[13]
Additionally, co-crystal structures revealed that MEK1/2 has an allosteric site[14, 15]
.
,
which endows inhibitors with high selectivity against MEK1/2 by a non-ATP
competitive mechanism. The binding site, which is adjacent to, but different from, the
ATP-binding site undergoes several conformational changes to lock unphosphorylated
MEK1/2 in a catalytically inactive state. Undesired adverse effects can be largely
avoided owing to this specific binding to MEK kinase[16]
.
Although numerous small molecule inhibitors have been discovered within the
past two decades, the diversity of the available MEK1/2 inhibitors remains limited.
Most of the reported MEK1/2 inhibitors can be structurally categorized as
biarylamines[17,18]
for example, the FDA-approved trametinib, cobimetinib, and
,
binimetinib, the in-clinical-trailed compounds pimasertib, refametinib, PD184352,
and PD0325901 (Fig 1). A good degree of similarity can be seen among the majority
of these inhibitors. Unlike the closely related biarylamines, the non-biarylamine
inhibitors are quite dissimilar to one and another. For example, the scaffold of U0126
is a symmetric vinylogous cyanamide which undergoes cyclization and isomerization
in aqueous solution[19], and mostly used as an in vitro laboratory tool. Derivation from
PD98059 has generated a series of chromone-based inhibitors against MEK1 with
IC50 values in nanomolar range[20]. Compound RO5126766, which is also known as
CH5126766, represents a series of chromen-2-one-based dual Raf/MEK inhibitors
that can suppresses feedback reactivation of RAF activity[21]. Similarly, RO5068760
and its hydantoin-based analogues can also prevent feedback reactivation in addition
to MEK/ERK dual inhibition[22,23] (Fig. 1).
3 / 45

ACCEPTED MANUSCRIPT
Fig. 1. Representative MEK1/2 inhibitors.
Co-crystal structure has revealed that although bind to the same allosteric site,
novel non-biarylamine inhibitors like RO5126766[24], adopt binding modes different
from those biarylamines. For example, the inhibitors miss some of the key
interactions with Val127 or Lys97 while extending to a cavity surrounded by Arg189,
Asn221, Ser222, and others. We thus wondered, if there were any novel MEK
inhibitors that can not only retain the key interaction normally formed by
biarylamines, but also form some additional interactions or occupy some specific
sub-pockets of MEK in the allosteric site. To this end, we decided to build a
pharmacophore model starting from the active conformations of biarylamines isolated
from co-crystal structures in complex with MEK1, followed by analysis of the hit
compounds out of database screening, and search for the MEK inhibitors with novel
binding modes.
2. Results and discussion
2.1 Pharmacophore modeling and virtual screening
Up to this work was initiated, there were 45 crystal structures of MEK1 released
in the Protein Data Bank[25], among which 26 were in complex with ligands closely
related to biarylamines. We collected co-crystal structures of 11 classic biarylamine-
MEK1 complexes to build pharmacophore models. The PDB entries of the 11
4 / 45

ACCEPTED MANUSCRIPT
co-crystal structures[26-34] and the ligand structures are listed in Fig 2.
Fig. 2. The PDB entries of the 11 co-crystal and ligand structures used to build pharmacophore model.
For each of the 11 PDB complexs, a structure-based pharmacophore model was
generated automatically by using the LigandScout 3.02 software[35]. All the automated
models possessed 7-10 chemical features, in addition to many exclusion volumes. For
example, pharmacophore model 2P55 was composed of six hydrophobic features, two
hydrogen bond (HB) acceptors，and one HB donor，in addition to 19 exclusion
volumes (Fig 3a and 3b). Two of the six hydrophobic features were generated from
the biaryl rings within the 2P55 ligand (ring A and ring B in Fig. 3a and 3b). For
easier description in the following context, ring A refers to the one connected with the
linear side chain and ring B is the ring substituted with one or two halogens. With the
aim of finding novel ligands with more structural diversity, the exclusion volumes
were consequently not considered in the following model validation and database
screening. The topologies of the remaining pharmacophore features of the 11 models
were very similar and they overlapped very well when aligned by the “reference
points” with LigandScout (Fig 3c).
5 / 45

ACCEPTED MANUSCRIPT
Fig. 3. Structure-based pharmacophore models generated by LigandScout. a) Alignment of 2P55 ligand
with its pharmacophore model in the allosteric pocket of MEK1; b) Model from the ligand-protein
complex 2P55; c) The overlapping of all eleven models. Among the features within models, yellow
spheres represent hydrophobic features, red arrows represent hydrogen bond acceptors, and green
arrows represent hydrogen bond donors. Exclusion volumes are shown in grey and hidden for clarity in
Fig. 3c.
For model validation, we organized a test dataset composed of 142 active, 5
inactive, and 6041 decoy compounds collected from the ChEMBL Database[36]. Here,
decoys refer to those compounds with reported activities not related to MEK, whereas
their physical properties, including molecular weight, ALogP, mumber of rotatable
bonds, and number of HB acceptor and donors, were similar to those of known MEK
inhibitors. Table 1 shows the values for each physical property considered in the
decoy compound selection, and these values came from the analysis of the above
mentioned 142 active compounds. The quality of the pharmacophore model, namely
the ability to enrich the active compounds from the data set, was evaluated by
calculating enrichment factors (EF) using the equation below, where TP is the number
of active compounds fitting into the model, n is the total number of molecules
(including both active compounds and decoys) that were returned as hits by the
pharmacophore-based screening, A is the number of active compounds in the entire
validation database, and N is the number of all compounds in the validation database.
Table 1. Physicochemical properties considered in decoy compound selection
Physical Properties
Value
6 / 45

ACCEPTED MANUSCRIPT
Molecular Weight
ALogP
≥ 400 and ≤ 500
≥ 1.4 and ≤ 4.6
Number of Rotatable Bonds
Number of Hydrogen Bond Acceptors
Number of Hydrogen Bond Donnors
≥ 4
≥ 4
≥ 4
Possessing many features, the individual models were too specific so that almost
none of the compounds in the active dataset could be correctly classified (Table S1 in
Supporting Information). The exceptions were models 3MBL and 3PP1, having
only 7 features with no hydrophobic features on ring A (Fig. 3b). These two models
hit more than 5 compounds in the active dataset, which indicated that the hydrophobic
features on ring A might not be required for MEK inhibition. Therefore, the
hydrophobic features on ring A were all deleted from each of the automated models, if
present, for refinement. Based on the observation that the hydrophobic pocket formed
by Met143, Ile141, Leu118 and Phe209 had relatively limited space to accommodate
more of the hydrophobic substitutions on position 2 of ring B, one of the
corresponding hydrophobic features was also deleted, although in most of the cases it
was occupied by halogens in the biarylamine ligands. As for the HB acceptors, the
number of this feature in each automated model varied from 2 to 4, among which two
were shared by all models. One of them referred to the halogen or carbonyl oxygen in
position 4 of ring A interacting with the peptidic backbone of Ser 212 (Fig. 3a), the
other was the oxygen or carbonyl group on the substitution at position 2 of ring A
bonding with Lys 97. These two HB acceptors seemed to be more important
compared with other features, therefore, they were kept in all models during
refinement.
Among the automated models, 1S9J, 3MBL and 3PP1 were observed forming
one additional HB with Val 211, in adjacent to the HB formed with Ser 212. Since this
HB acceptor was only observed in 3 of the 11 automated models, it was likely to be
less important, thus be set as “optional” if available. In addition,Lys 97 was found
forming HB with individual ligand in 1S9J, 3DV3, and 3ORN at some upper
positions, and therefore, these HB acceptors were included in the refined models and
were set as “optional”.
7 / 45

ACCEPTED MANUSCRIPT
It can be seen from Fig. 3b that directions of the HB donor features (green
arrows) varied a lot. These HBs were observed forming with ATP instead of with
MEK1. These direct hydrogen bondings may restrict ATP-γS hydrolysis due to
limited access to the water required for such reaction. It might not be required for
MEK inhibition, because this direct hydrogen bonding with ATP was not observed in
1S9J, 3DV3, 3DY7, and 3ORN at all, and many compounds in the active dataset
didn’t have corresponding substructure to form this HB, either. Therefore, we set this
HB donor feature as “optional” during refinement.
The refined models were used to screen the testset compounds again with results
shown in Table 2. The enrichment factors ranged from 12.82 to 26.28, which were
much more improved than those of the automated ones (Table S1).
Table 2 Number of hit compounds in the test set by the refined models.
Model
code
No. of Hits
No. of Hits
No. of Hits
Decoy
68
EF
Inactive
Active
52
8
1S9J-r-1
1S9J-r-2
2P55-r
1
0
0
0
0
1
0
0
0
0
0
0
0
0
18.87
14.51
26.28
24.19
26.45
12.82
18.10
21.02
18.49
22.70
21.21
20.39
20.97
15.40
16
67
60
65
58
74
70
62
49
57
59
13
58
44
3DV3-r-1
3DV3-r-2
3DY7-r
3E8N-r
3EQB-r
3EQC-r
3EQG-r
3MBL-r
3ORN-r-1
3ORN-r-2
3PP1-r
48
42
139
104
75
84
45
60
67
14
106
To indicate how the refined models could hit the active compounds, a heat map
was drawn (Fig. S1 in Supporting Information). Accordingly, we have tried many
different combinations of the models to increase the possibility to hit more active
compounds, and finally, a collection of 1S9J-r-1, 2P55-r and 3DV3-r-1 (Fig. 4) were
selected representing all the possible binding modes of biarylamine ligands. They
have hit 80 compounds in the active dataset (142 in total, 56.3% hit rate) and
8 / 45

ACCEPTED MANUSCRIPT
generated a final EF of 17.16.
1S9J-r-l
2P55-r
3DV3-r-l
Fig. 4 The pharmacophore model collection selected for the large-scale virtual screening.
The generated collection of pharmacophore models was used for virtual screening of the
Specs database (version Apr 2010, www.specs.net) containing 200,158 compounds in total.
The numbers of virtual hits for each model were listed in Table 3.
Table 3 Number of hit compounds identified by the individual pharmacophore model in the collection.
Model name
1S9J-r-l
Hits number
507
Hit rate %
0.25
Number of top hits selected
42
17
2P55-r
6142
3.08
3DV3-r-1
3063
1.53
304
The total number of virtual hits was 507 + 6,142 + 3,063 = 9,712, among which some
could be hit by more than one model. These virtual hits datasets were merged and clustered
by structural similarity with Pipeline Pilot 7.5 software (Accelrys Inc., San Diego, CA, USA).
Representative compounds with better lead-likeness were selected manually from the
structural clusters with more members than others. Finally, 13 hit compounds (Table 4) were
selected, and purchased from Specs for further MEK1 inhibitory evaluations.
Table 4 The 13 hit compounds selected for further validation by MEK1 inhibitory assays.
Hit pharmacophore
model
Pharmacophore
fit score
Compound name
Specs entry
M100
M101
M103
M104
M105
M106
M108
M109
AP-906/41639971
AF-399/41703146
AN-988/15131219
AG-690/33045045
AP-853/42877451
AN-329/42286028
AG-205/03751018
AO-365/43113290
3DV3-r-1
56.07
1S9J-r-l, 3DV3-r-1
3DV3-r-1
66.42/57.49
56.81
3DV3-r-1
57.21
3DV3-r-1
56.32
1S9J-r-l, 3DV3-r-1
1S9J-r-l, 3DV3-r-1
3DV3-r-1
66.67/55.63
65.54/56.58
55.73
9 / 45

ACCEPTED MANUSCRIPT
M110
M111
M113
M115
M116
AO-365/43300996
AK-198/36417052
AG-205/36953117
AG-690/10031030
AK-968/13027684
3DV3-r-1
2P55-r, 3DV3-r-1
1S9J-r-l
55.60
55.36/55.59
65.77
1S9J-r-l/3DV3-r-1
1S9J-r-l, 3DV3-r-1
66.05/46.82
66.45/56.57
2.2 Validation of the computational work by in vitro enzymatic assays
After all compounds were collected, their ability to bind with phosphorylated
MEK1 was firstly measured by LanthaScreen® Eu Kinase Binding Assay. Then, to
further validate their potency to inhibit the Raf-MEK-ERK signaling, both Raf-MEK
and MEK-ERK cascading assay were performed at the enzymatic level.
Table 5. Results of in vitro enzymatic assaysa
Binding assay normalized to
Raf-MEK
MEK-ERK
AZD6244 at 100 µM
(%inhibition)
cascading inhibition
cascading inhibition
Compound 20 µM (%)
2 µM (%)
20 µM (%)
2 µM (%) 20 µM (%) 2 µM (%)
U0126
M100
M101
M103
M104
M105
M106
M108
M109
M110
M111
M113
M115
M116
-
-
87.4
56.1
22.6
49.4
40.7
±
±
±
±
±
4.4
6.6
7.6
4.6
6.9
41.7
4.4
±
4.9
2.6
8.6
2.9
2.3
3.6
5.5
1.9
0.5
3.7
80.9
32.0
29.8
10.4
13.7
26.7
12.7
6.0
±
±
±
±
±
±
±
1.2 26.9
8.5 3.9
±
1.0
2.3
90.3
35.6
14.4
29.6
16.2
39.3
±
3.9
2.9
0.1
2.4
0.5
1.8
32.1
32.6
8.7
±
±
0.6
0.7
0.5
0.3
0.1
0.0
±
±
±
±
±
±
±
5.1
±
9.1 -10.4
1.8 -0.9
4.2 -5.4
±
2.0
2.2
3.1
1.9
2.2
2.1
±
10.5
5.2
±
±
±
9.6
±
±
±
8.0
32.1
±
5.9
-7.1
11.7
7.8
±
5.8
3.5 -1.6
2.9 -3.7
3.3 -12.1
5.3 4.1
3.3 ±
23.4
±
45.7
31.9
32.8
35.0
32.3
35.1
57.0
58.2
±
5.3
6.0
1.3
2.7
3.5
8.9
6.5
4.8
±
±
±
-128.1
13.52
±
0.5 -50.2
±
0.4
0.3
0.7
1.3
0.3
1.3
0.2
±
±
±
±
±
±
±
±
±
±
0.3
14.8
8.9
±
4.1
±
±
15.2
8.6
±
±
1.8
34.5
22.9
28.0
60.1
24.4
±
1.5
1.7
2.3
4.1
0.6
±
7.2
±
±
8.2
±
4.7
±
-0.04
±
4.3 19.9
±
8.2 -13.1
±
1.9
7.8
±
±
±
12.9
11.9
26.1
±
0.2
6.3
±
9.6
4.4
4.2
3.2
±
9.7 -17.5 ±
±
±
±
35.0
11.7
±
±
7.5
5.1
±
1.4
2.3
14.7
±
6.1 -2.9 ±
a All values are an average of three parallel experiments. bND stands for not determined.
Among all the 13 tested compounds, M100 and M115 showed the desired
10 / 45

ACCEPTED MANUSCRIPT
activities (for the alignment of the two compounds with the refined pharmacophore
model refer to the Supporting Information Fig. S2). M100 showed 90% affinity to
pMEK compared to the biarylaminal inhibitor AZD6244 (selumetinib), and both
M110 and M115 inhibited Raf-MEK cascading by more than 50%. Therefore, we
measured the IC50s for the two compounds in each of the assays. The chemical
structures of M100 and M115 and the respective IC50s are depicted in Figure 5.
Fig. 5 The chemical structures of active hit compounds and their IC50s for RAF-MEK and
MEK-ERK cascade inhibition.
In the following part of this paper, we report the preliminary deviations on M100
in order to further validate the reliability of M100 as a MEK inhibitor.
2.3 Design and synthesis of carbazoles derived from active hit compound M100
As a commercially available compound, there are already some existing reports
about the bioactivities of M100 which is a moderate proteasome inhibitor against β5
subunit[37] and its derivatives bearing a naphthol scaffold are selective Mcl-1
inhibitors[38]. Before we started our chemical deviations on M100 toward improved
potency against MEK1, its cytotoxicity was measured in CellTiter-Glo® luminescent
cell viability assay[39] by using the HEK293 cell line. Results showed that M100 was
cytotoxic (IC50 = 8.89 ± 2.0 µM), which is highly possibly related to the naphthol core
within its structure.
Therefore, we started our work from replacing the naphthol scaffold with
carbazole with considerations including: 1) carbazole could be regarded as a
“ring-closure” form of biarylamine which is a classic structure in allosteric MEK1
inhibitors; 2) the carbazole scaffold was found in many synthetic therapeutic agents,
11 / 45

ACCEPTED MANUSCRIPT
e.g. a blocker of β-adrenergic receptor, and regarded as a privileged structure for drug
discovery[40]. Focused design and modifications on the side chain at the 3-position of
the carbazole core and the sulfonamide moiety were made with the aim to explore the
structure-activity relationship against MEK1.Through substituent transformation and
bioisosterism replacement, the sulfur atom on the side chain was designed and
replaced by an oxygen or a sulfonyl group (SO2), while substitutes for carboxyl group
included ester (COOEt), amide (CONH2), or a variety of heterocycles[41]. The
structures of the designed carbazole derivatives were summarized in Figure 6.
X = O, S, SO2; R1 = COOH, tetrazole, 3-methyl-1,2,4-oxadiazole, 5-methyl-1,3,4-oxadiazole,
5-amino-1,3,4-oxadiazole, COOEt, CONH2; R2 = p-Me, p-NHAc, m-NHAc; R3 = H, OH.
Figure 6. Structure and substituted groups of designed carbazole derivatives.
M100 was synthesized following a reported procedure[42] in 2 steps (Scheme 1).
Commercially available p-toluenesulfonamide was regio-selectively coupled to
naphthoquinone intermediate 1 in the presence of titanium (IV) chloride and
triethylamine with microwave assisted heating. Substitution of the sulfonamide
intermediate 2 with thioglycolic acid as a nucleophile was carried out in presence of
pyridine. After reduction in situ with Na2S2O4, M100 was obtained in a yield of
74.1%.
Reagents and conditions: (a) TsNH2, 1 M TiCl4 in DCM, Et3N, THF, microwave, 65 °C, 15 min,
58.6%; (b) i) thioglycolic acid,pyridine, THF, r.t., 30 min; ii) Na2S2O4, EtOAc, and H2O (mixing
in a separating funnel), r.t., 74.1%.
Scheme 1. Synthetic route of M100
12 / 45

ACCEPTED MANUSCRIPT
As shown in Scheme 2, 4-hydroxycarbazole derivative 7 was prepared in 4 steps.
Starting from 3 and N-chlorosuccimide (NCS)[43], 4-hydroxycarbazole chloride 4 was
yielded, which was further oxidized to afford carbazole-1,4-dione 5[44]. Following the
same two-step procedure for M100 synthesis, intermediate 5 was converted to desired
derivative 7.
aReagents and conditions: (a)N-chlorosuccimide (NCS), MeCN, r.t., 1.5 h, 47.6%; (b) Fremy’s
Salt, KH2PO4, water-acetone, r.t., 1.5h, 94.7%; (c) TsNH2, 1 M TiCl4 in DCM, Et3N, THF,
microwave, 65 °C, 15 min, 69%; (d) i) thioglycolic acid,pyridine, THF, r.t., 30 min; ii) Na2S2O4,
EtOAc, and H2O (mixing in a separating funnel), r.t., 21.3%.
Scheme 2. Synthetic route of derivative with a 4-hydroxycarbazole corea
The key steps for preparation of key intermediates 12, 14 and 15 were shown in
Scheme 3. Reaction of 4-amino-3-nitro-phenol 8 with ethyl 2-bromoacetate in the
presence of base (K2CO3) provided amine 9[45] which was then diazotized and reduced
to yield a substituted hydrazine analogue 10[46]. Fischer indolization between
hydrazine 10 and cyclohexanone occurred subsequently to afford tetrahydrocarbazole
11[47]. Then, oxidation of 11 by 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ)
was conducted to give key intermediate 12[48]. After hydrolysis of the ester 12 with
LiOH[49], the resulting carboxylic acid 13 was then treated with acetamide oxime,
EDCI and HOBt to afford key intermediate 14[50]. Ring-closure reaction between 13,
acetohydrazide and POCl3 gave key intermediate 15[51]
.
13 / 45

ACCEPTED MANUSCRIPT
Reagents and conditions: (a) BrCH2COOEt, K2CO3, acetone, r.t., overnight, 75.2%; (b) i) NaNO2,
conc. HCl, -10 °C, 0.5 h; ii) SnCl2, r.t., overnight; (c) cyclohexanone, EtOH/conc. HCl (1:1), reflux, 3h,
41.7%; (d) DDQ 2.4 equiv., PhMe, reflux, 2 h, 96.2%; (e) LiOH, THF/water, r.t., 4.5 h, 99.8%; (f)
HOBt, EDCI, acetamide oxime, MeCN, r.t., 0.5h, then reflux overnight, Ar, 53.8%; (g) i)
acetohydrazide, HOBt, EDCI, DMF, r.t., 12h, 74.7%; ii) POCl3, sealed tube, 100 °C, 1h, 65.8%.
Scheme 3. Synthetic Route to key intermediates 12, 14, 15
Key intermediate 21 was synthesized through another route in 5 steps (Scheme
4). Compound 16 underwent diazotization, reduction and Fischer indolization to give
tetrahydrocarbazole 18, which was converted into carbazole 20 after esterification[52]
and aromatization with DDQ. Finally, key intermediate 21 was generated through
coupling reaction using palladium catalyst[53]
.
Reagents and conditions: (a) i) NaNO2, conc. HCl, -10 °C, 0.5 h; ii) SnCl2, r.t., 3h; (b) cyclohexanone,
EtOH, reflux, overnight, 38.1%; (c) Tf2O, 2,6-Lutidine, DCM, Ar, r.t., 24 h, 99.3%; (d) DDQ, PhMe,
reflux, overnight, 77.5%; (e) Pd2(dba)3 0.25 equiv. , Xantphos 0.5 equiv. , DIPEA, ethyl
mercaptoacetate, 1,4-dioxane, Ar, reflux, overnight, 75.6%.
Scheme 4. Synthetic Route to Key Intermediates 21
With key intermediates (12, 14, 15, 21) in hand, synthesis of a series of final
compounds (22a, 22b, 22c, 22d, 22e, 22f, 22g and 22h) were accomplished (Scheme
14 / 45

ACCEPTED MANUSCRIPT
5) by reduction of -NO2 to -NH2 and substitution with diverse sulfonyl chlorides[54]
.
Corresponding carboxylic acid 22i, 22j and 22k were generated from the hydrolysis
of ester 22g, 22h and 22a[49]. Aminolysis of 22a with NH4OH provided carboxamide
22l. And then, terminalamide 221 was easily converted to tetrazole 22m following a
reported procedure[38]. 5-amino-1,3,4-oxidiazole 22o was generated from 22a by
using a 2-step method[55]. By oxidation with oxone[42], thioether 22g was converted to
sulfone 22p which underwent hydrolysis to provide carboxylic acid 22q. All
1
structures were confirmed by H and 13C-NMR spectroscopy, EI-MS or EI-HRMS
spectra (for details, see Experimental Section).
R1
R1
X
X
X=O
R2= p-Me
i) a or b
ii) c
22m: R1= tetrazole
22l: R1= CONH2
22a: R1= COOEt
22n: R1= CONHNH2
22o: R1= 5-amino-1,3,4-
oxadiazole
f
O
N
N
H
NO2
H
HN
R2
e
g
h
S
O
12: X=O, R1= COOEt
14: X=O, R1= 3-methyl-1,2,4-oxadiazole
15: X=O, R1= 5-methyl-1,3,4-oxadiazole
21: X=S, R1= COOEt
22a:X=O, R1 = COOEt, R2 = p-Me
22b:X=O, R1 = COOEt, R2 = p-NHAc
22c:X=O, R1 = 3-methyl-1,2,4-oxadiazole, R2 = p-Me
22d:X=O, R1 = 5-methyl-1,3,4-oxadiazole, R2 = p-Me
22e:X=O, R1 = 5-methyl-1,3,4-oxadiazole, R2 = p-NHAc
22f:X=O, R1 = 5-methyl-1,3,4-oxadiazole, R2 = m-NHAc
22g:X=S, R1 = COOEt, R2= p-Me
22g: X=S, R1= COOEt, R2= p-Me
22p: X=SO2, R1= COOEt, R2= p-Me
22q: X=SO2, R1= COOH, R2= p-Me
d
i
j
d
22h:X=S, R1 = COOEt, R2= p-NHAc
d
22i:X=S, R1 = COOH, R2= p-Me
22j:X=S, R1 = COOH, R2= p-NHAc
22k:X=O, R1 = COOH, R2= p-Me
Reagents and conditions: (a) Pd-C, H2, MeOH, r.t., overnight; (b) Na2S2O4, K2CO3, DCM/MeOH,
water, r.t., 2 h, Ar; (c) R2PhSO2Cl, pyridine, r.t., overnight, 21.6-63.9%; (d) NaOH (1N), THF/water, r.t.,
1h, 88.4%; (e) NH4OH (25%), r.t., 1 h, 94.9%; (f) NaN3, SiCl4, CH3CN(dry), sealed tube, 80 °C, 14 h,
Ar, 62.1%; (g) hydrazine hydrate, EtOH, reflux, 1 d, 83.0%; (h) BrCN, EtOH, reflux, 3 h, 22.8%; (i)
oxone, acetone/water, r.t., overnight, 49.3%; (j) NaOH (1N), THF/water, r.t., 1h, 87.7%.
Scheme 5. Synthetic Route to Final Compounds
2.4 Biological evaluations and preliminary SAR of carbazole derivatives as
MEK1 inhibitors
After all designed compounds were prepared, they were analyzed with the
Raf-MEK and MEK-ERK cascading assays and the results were listed in Table 6. For
Raf-MEK cascading, they were first tested at 20 µM and the IC50 values were
determined for the 5 compounds with inhibition >50% and the best IC50 value (12.8
µM) was observed from 22k and 22i. None of the acquired compounds exhibited >50%
15 / 45

ACCEPTED MANUSCRIPT
inhibition against MEK-ERK cascading, which was in agreement with their
postulated mechanism that the allosteric MEK kinase inhibitors were supposed to
inhibit MEK phosphorylation in Raf-MEK assay, instead of inhibiting the process of
phosphorylated MEK to activate ERK in MEK-ERK assay.
Table 6. Inhibition of Raf-MEK and MEK-ERK cascading.
Raf-MEK
MEK-ERK
% Inhibition
(20 µM)
Compd.
X
R1
R2
R3
% Inhibition
IC50 ± SD
(µM)
(20 µM)
86.2 ± 2
U0126
M100
7
0.21± 0.02
27.2 ± 4.5
31.7 ± 1.3
23.7 ± 0.7
NDa
85.8 ± 5.7
12.7 ± 3.2
35.6 ± 14.6
28.9 ± 4.9
18.2 ± 2.1
21.5 ± 2.7
27.9 ± 1.9
-
-
-
-
OH
H
53.1 ± 2.5
43.7 ± 0.2
56.6 ± 0.7
27.4 ± 2.7
65.8 ± 3.9
41.9 ± 3.3
S
COOH
COOEt
COOEt
COOH
CONH2
p-Me
p-Me
p-NHAc
p-Me
p-Me
22a
O
O
O
O
22b
22k
22l
H
H
12.8 ± 0.5
ND
H
22m
O
p-Me
H
46.8 ± 0.3
ND
45.1 ± 13.4
22o
O
p-Me
H
33.0 ± 3.8
ND
21.4 ± 8.6
22c
22d
22e
22f
O
O
O
O
p-Me
p-Me
H
H
H
H
38.2 ± 2.2
48.6 ± 3.1
50.7 ± 1.2
29.5 ± 0.1
ND
ND
37.8 ± 15.9
25.8 ± 9.6
20.6 ± 8.3
-10.0 ± 9.6
p-NHAc
m-NHAc
15.3 ± 0.6
ND
22g
22h
22i
S
S
COOEt
COOEt
COOH
COOH
COOH
p-Me
p-NHAc
p-Me
H
H
H
H
H
45.2 ± 2.5
35.8 ± 4.0
67.2 ± 0.0
44.2 ± 0.0
45.0 ± 4.4
ND
ND
39.2 ± 0.2
21.8 ± 0.7
21.4 ± 2.1
27.7 ± 7.5
35.5 ± 11.6
S
12.8 ± 1.9
ND
22j
22q
S
p-NHAc
p-Me
SO2
ND
16 / 45

ACCEPTED MANUSCRIPT
a ND = not determined.
Compared with the starting compound M100 (IC50 = 27.2 µM), hopping of the
naphthol core with carbazole maintained the inhibitory activity (Compd. 7, IC50
=
31.7 µM). As for the various R1, compounds with a carboxyl group (22k)
demonstrated the best activity (IC50 = 12.8 µM), and the sequence of the substituents
for inhibitory activity was COOH (22k) > COOEt (22a)> CONH2 (22l) and
heterocycles. As for the various R2, phenyl ring with an acetylamino group substituted
at para-position (22e) showed better potency than meta- position (22f). Compounds
with a methyl group at para-position of the phenyl ring (22a, 22g) were more
favorable than the acetylamino substituted ones (22b, 22h), suggesting that steric
hindrance on R2 might reduce binding affinity. Substitution of -OH at R3 seemed not
contribute much to the activity because compared with comd. 7 (IC50 = 31.7µM), 22i
demonstrated even slightly improved potency (IC50 = 12.8 µM). Compounds with
either a thioether side chain or an ether side chain at position 3 displayed equal
inhibitory activity (22k and 22i, IC50 = 12.8 µM). However, replacement of the S or O
atom by sulfonyl group (22q) has led to a total lost of activity. This is supposed due to
the lost of flexibility of this side chain.
The above five active carbazoles were further measured for cytotoxicity against
HEK293 cells, and their anti-proliferation effects were also evaluated against
representative cancer cells lines, A549, A375 and HL60, with data listed in Table 7.
As expected, all five carbazoles exhibited much reduced cytotoxicity compared with
M100. Interestingly, although the phenolic -OH was retained as -R3 in compd. 7, its
cytotoxicity remained very low with IC50 higher than 100 µM. In addition, the
anti-proliferation effects of compd. 7 in tested cell lines were also comparable to those
of M100. The above results have confirmed our speculation that carbazole is a better
scaffold than naphthol to develop druggable molecules.
Table 7. The anti-proliferative and cytotoxic profilesa and docking scores of selected
compounds.
IC50b (µM)
17 / 45

ACCEPTED MANUSCRIPT
Compd.
HEK293
>100
A549
A375
HL60
Glide gscore
DMSO
>100
-
>100
-
>100
-
-
Staurosporine 0.0037 ± 0.0010
-
U0126
M100
7
-
32.2 ± 7.5
21.6 ± 6.1
1.1 ± 0.2
7.7 ± 1.1
0.3 ± 5.3
17.2 ± 6.6
26.3 ± 3.4
>100
-
8.9 ± 2.0
>100
>100
>100
>100
>100
-5.302
-7.677
-7.369
-7.95
-6.28
-6.28
12.8 ± 1.9 14.7 ± 0.6
70.7 ± 0.9 70.7 ± 0.9
22a
22k
22e
>100
>100
>100
>100
43.7 ± 0.6
>100
>100
>100
22i
>100
aThe cytotoxicity was measured on HEK293 cells; bIC50 values are the means of three independent
experiments.
2.5 Effect of ATP concentrations on the Raf-MEK cascading inhibitory potency
The ATP competitive profile of these inhibitors was evaluated by using
representative compound 22k in BRaf/MEK1 cascade inhibition assay by varying
ATP concentrations. The IC50 values were 13.2 µM and 12.1 µM respectively when
the ATP concentration varied from 100 µM to 300 µM. No shift in IC50 (Figure 7)
was observed when ATP concentration was even increased up to 300 µM, and it
suggested that compound 22k inhibited the BRaf/MEK1 cascading in an ATP
non-competitive manner.
Figure 7. The dose-response curve of 22k in BRaf-MEK cascading assay at various ATP
concentrations
2.6 Docking Studies
Both the virtual hit M100 and several selected derived carbazoles were docked
into the allosteric site MEK1 (PDB entry: 3DV3) by using Schödinger suite 2016-2
(Schrödinger, LLC, New York, NY, USA) to evaluate the binding modes. As shown in
18 / 45

ACCEPTED MANUSCRIPT
Table 7, a significantly improved docking score was achieved by 22k (-7.950)
compared with M100 (-5.302). Three HBs were observed forming between M100 and
MEK1, among which two were formed by the sulfonyl oxygen with Lys97, and the
other was between the carboxylic oxygen and Asp 208 (Fig. 8a). However, the
orientation of the 4-methylbenzenesulfonamide moity within 22k was reversed, and
therefore, the HB with Lys97 was formed by the carboxylic oxygen of 22k. The
increased binding affinity of 22k was supposed largely due to the additional HBs
formed by the carbazole and sulfomide nitrogen with Phe209, together with two
additional π-π stacking between the carbazole ring and Phe209 (Fig. 8b). The
sulfonyl oxygen of 22k was also observed forming critical HB with Ser212. In
summary, it seemed that the binding of 22k was more favorable than the lead
compound M100. To further increase affinity based on 22k, halogen can be
introduced to the carbazole ring to mimic the halogen bond formed by the native
ligand of 3DV3 and Val127. Alternatively, negative charged substitutions can be
introduced at the para-position of the benzenesulfonamide moity of 22k to fit into the
positive charge pocket unoccupied.
Fig. 8 The predicted binding mode of M100 (a) and 22k (b) with MEK1 (PDB entry: 3DV3) in the
allosteric site. The native ligand of 3DV3 is presented in green sticks, while M100 in purple and 22k in
yellow. The hydrogen bonds are shown in yellow dashed lines, and the magenta dashed lines represent
halogen bond. The surface of the binding site is shown either in solid or mesh in which red represent
negative charge and blue represent positive charge.
19 / 45

ACCEPTED MANUSCRIPT
3. Conclusions
A structure-based pharmacophore model of MEK1 inhibitors was constructed
and used as the criterion in a large scaled virtual screening. Two active compounds
with moderate potency were identified as novel MEK inhibitors out of 13 virtual hits.
Deviations on one hit, M100, by using substituent transformation and bioisosterism
replacement, have generated several active carbazole derivatives with IC50s at µM
level against Raf-MEK cascading, and with greatly reduced cytotoxicity to HEK293
cells. These results suggested the liability of our pharmacophore models, and that
carbazole can serve as a good starting point of novel MEK1 inhibitor design.
4. Experimental section
4.1 Computational works
The pharmacophore models were generated with LigandScout 3.02 software
which is a useful tool for generating 3D pharmacophore models from structural data
of macromolecular complexes, and the models generated can either be manually
refined or directly used to screen databases.
The physicalchemical properties of compounds were calculated by using
“Calculate Molecular properties” function within Discovery Studio software
(Accelrys, Discovery Studio Vers. 2.5, San Diego, CA, USA.). Clustering was
performed with the “Find Diverse Molecules” function by using the predefined set
FCFP4 for fingerprint comparison. For filtering and analysis of libraries, Accelrys’
Pipeline Pilot (PP) version 7.5 was used.
Docking studies were performed in Schödinger Suite (Schrödinger, LLC, New
York, NY, USA) using the standard precision mode and the co-crystallized ligand of
PDB entry 3DV3 [28] as center of the grid box.
4.2 Chemistry
1
Melting points were measured with an X4 apparatus and were uncorrected. H
NMR spectra were recorded on a Bruker (BrukerBioSpin AG, Fällanden, Switzerland)
Avance III 400 MHz system. Chemical shifts were reported in parts permillion (ppm)
relative to tetramethylsilane (TMS) as internal standard. The spin multiplicities were
20 / 45

ACCEPTED MANUSCRIPT
given as s (singlet), d (doublet), t (triplet), q (quartet), m (multiplet) or br (broad). MS
and high-resolution mass spectra (HRMS) were obtained using Electrospray
Ionization (ESI) technique on a Bruker’s Fourier Transform Ion Cyclotron resonance
Mass Spectrometer. Thin layer chromatography (TLC) analysis was performed on
silica gel GF254 purchased from Qingdao Haiyang Chemical Co. (Qingdao,
Shandong Province, China) or Merck (Darmstadt, Germany).
4.2.1 N-(3-chloro-4-oxonaphthalen-1(4H)-ylidene)-4-methylbenzenesulfonamide (2)
To a suspension of TsNH2 (1.37 g, 8 mmol) and 2-chloro-1,4-naphthoquinone
(1.54 g, 8 mmol) in dry THF (65 mL) in an ice bath, 1 M TiCl4 solution in DCM (8
mL) was added dropwise, followed by addition of TEA (2.64 mL, 17.6 mmol). The
mixture was heated to 65 °C under microwave irradiation for 15 min. Then, inorganic
salts were filtered off. The filtrate was concentrated and purified by column
chromatography (silica gel, DCM/PE = 1/2) to afford a yellow solid (1.62 g, 58.6%).
m.p.: 135-136 °C. 1H NMR (400 MHz, DMSO-d6): δ 8.44 (s, 1H, ArH), 8.07-8.11 (m,
2H, ArH), 7.97 (d, J = 8.28Hz, 2H, ArH), 7.81-7.89 (m, 2H, ArH), 7.53 (d, J = 8.04Hz,
2H, ArH), 2.45 (s, 3H, CH3).
4.2.2 2-((1-hydroxy-4-(4-methylphenylsulfonamido)naphthalen-2-yl)thio)acetic acid
(M100)
Intermediate 2 (345.5 mg, 1 mmol), thioglycolic acid (0.11 mL, 1.5 mmol) and
pyridine (0.12 mL, 1.5 mmol) were mixed in THF (12 mL) and stirred at room
temperature for 0.5 h. After the solvent was removed in vacuum, the residue was
dissolved in EtOAc (50 mL) and was transferred to a separation funnel. The organic
phase was washed with KHSO4 solution (0.5 M, 50 mL) twice to remove excess
pyridine, and the aqueous phase was removed. Na2S2O4 (1.04 g, 5 mmol) and water
(50 mL) was added, and the organic phase was shaken to colorless before was
collected and washed with water (50 mL), brine in sequence and dried over anhydrous
Na2SO4. The organic phase was concentrated and the residue was recrystallized to
1
obtain a white solid (298.9 mg, 74.1%). m.p.: 153-154 °C. H NMR (400 MHz,
21 / 45

ACCEPTED MANUSCRIPT
DMSO-d6): δ 12.82 (s, 1H, COOH), 9.84 (s, 1H, SO2NH), 8.13 (d, J = 8.00 Hz, 1H,
ArH), 7.93 (d, J = 7.60 Hz, 1H, ArH), 7.52 (d, J = 8.40 Hz, 2H, ArH), 7.50-7.42 (m,
2H, ArH), 7.30 (d, J = 8.00 Hz, 2H, ArH), 6.95 (s, 1H, ArH), 3.47 (s, 2H, CH2), 2.34
(s, 3H, CH3). 13C NMR (100 MHz, DMSO-d6): δ 171.53, 152.47, 143.43, 137.31,
131.62, 129.93, 129.32, 127.42, 127.11, 126.26, 125.51, 124.79, 123.96, 122.79,
113.40, 37.19, 21.43. ESI-MS (C19H17NO5S2): m/z 402.12 (M-H+); HRMS (ESI−) m/z
calculated for C19H16NO5S2 (M-H)−: 402.04699, found 402.04754.
4.2.3 3-chloro-9H-carbazol-4-ol (4)
To a solution of 9H-carbazol-4-ol (918.2mg, 5mmol) in acetonitrile (20 mL),
NCS (667.4mg, 5mmol) was added portionwise in more than 10 min. The reaction
mixture was stirred at ambient temperature for 1.5 h before being concentrated under
reduced pressure. The residue was then purified by column chromatography over
silica gel (gradient eluent = DCM/PE: 1/5-1/2). 3-chloro-carbazol-4-ol was collected
1
as a white solid (518.2mg, 47.6%). m.p.: 218-219 °C. H NMR (400 MHz,
DMSO-d6): δ 11.31 (s, 1H, NH), 9.78 (s, 1H, OH), 8.21 (d, J = 8 Hz, 1H, ArH), 7.46
(d, J = 8 Hz, 1H, ArH), 7.36 (td, J = 1.16 Hz J = 8 Hz, 1H, ArH), 7.30 (d, J = 8 Hz,
1H, ArH), 7.16 (td, J = 0.96 Hz J = 8 Hz, 1H, ArH), 7.00 (d, J = 8 Hz, 1H, ArH). 13C
NMR (100 MHz, DMSO-d6): δ 148.69, 140.44, 139.88, 126.81, 125.50, 122.76,
122.07, 119.23, 113.40, 111.05, 109.35, 104.18.
4.2.4 3-chloro-1H-carbazole-1,4(9H)-dione (5)
A solution of Fremy’s Salt (1.4430g) and potassium dihydrogen orthophosphate
(84.5mg) in water (85mL) was added to a solution of intermediate 4 (410mg,
1.88mmol) in acetone (85mL) at room temperature. The reaction mixture was stirred
at at room temperature for 1.5 h and then evaporated under vacuum. The resulting
precipitate was filtered, washed well with water and dried. A dark purple powder was
1
obtained (413.3mg, 94.7%). m.p.: 263-265 °C. H NMR (400 MHz, DMSO-d6): δ
13.11 (s, 1H, NH), 8.05 (d, J = 8 Hz, 1H, ArH), 7.57 (d, J = 8 Hz, 1H, ArH), 7.43 (td,
J = 0.96 Hz J = 8 Hz, 1H, ArH), 7.37 (td, J = 0.96 Hz J = 8 Hz, 1H, ArH), 7.18 (s, 1H,
22 / 45

ACCEPTED MANUSCRIPT
ArH).
4.2.5N-(3-chloro-4-oxo-4,9-dihydro-1H-carbazol-1-ylidene)-4-methylbenzenesulfona
mide (6)
Intermediate 6 was synthesized using the method similar to that of intermediate 2
except intermediate 5 was used as a starting material. The title compound was
1
obtained as a brown solid (132.8 mg) in yield of 69.0%. m.p.: 234-235 °C. H NMR
(400 MHz, DMSO-d6): δ 12.87 (s, 1H, aromatic NH), 8.11 (s, 1H, ArH), 8.00 (m, 3H,
ArH), 7.54 (d, J = 8 Hz, 3H, ArH), 7.41 (t, J = 8 Hz, 1H, ArH), 7.33 (t, J = 8 Hz, 1H,
ArH), 2.46 (s, 3H, CH3).
4.2.62-((4-hydroxy-1-(4-methylphenylsulfonamido)-9H-carbazol-3-yl)thio)acetic acid
(7)
The title compound (21 mg, 21.3% yield) was obtained as a beige solid with
method similar to M100 except intermediate 6 was used as a starting material. m.p.:
193-195 °C. 1H NMR (400 MHz, DMSO-d6): δ 13.33 (br s, 1H, COOH), 11.06 (s, 1H,
aromatic NH), 9.48 (s, 1H, SO2NH), 8.17 (d, J = 8 Hz, 1H, ArH), 7.57-7.51 (m, 3H,
ArH), 7.34-7.30 (m, 3H, ArH), 7.14 (t, J = 8 Hz, 1H, ArH), 6.65 (s, 1H, ArH), 3.13 (s,
2H, CH2), 2.36 (s, 3H, CH3).13C NMR (100 MHz, DMSO-d6) δ 172.23, 153.62,
143.58, 139.61, 139.44, 136.81, 130.76, 129.88, 127.67, 125.57, 122.51, 122.21,
119.79, 113.36, 112.68, 111.69, 106.30, 21.44. ESI-MS (C21H18N2O5S2): m/z
441.0593 (M-H+); HRMS (ESI−) m/z calculated for C21H17N2O5S2 (M-H)−: 441.05789,
found 441.05734.
4.2.7 ethyl 2-(4-amino-3-nitrophenoxy)acetate (9)
A 100 mL 2-neck flask fitted with a stir-bar and an Ar inlet was charged with
phenol 8 (4.9952 g, 32.4 mmol), K2CO3 (4.9346 g, 35.6 mmol) and acetone (60 mL),
to which was added with ethyl bromoacetate (3.8 mL, 34.0 mmol) dropwise. The
mixture was stirred at room temperature overnight. Salts were filtered through a pad
of celite. The organic phase was concentrated to give solid which was triturated and
23 / 45

ACCEPTED MANUSCRIPT
washed with PE/EtOAc=3/1 (50 mL) to remove red impurities and dried under
infrared oven to obtain a yellow solid (5.8554 g, 75.2%). m.p.: 200-201 °C. 1H NMR
(400 MHz, DMSO-d6): δ 7.37 (d, J = 2.24 Hz, 1H, ArH), 7.27 (s, 2H, NH2), 7.22 (dd,
J = 2.68 Hz J = 9.24 Hz, 1H, ArH), 7.01 (d, J = 9.24 Hz, 1H, ArH), 4.74 (s, 2H,
OCH2CO), 4.17 (q, J = 7.08 Hz, 2H, COCH2CH3), 1.21 (t, J = 7.12 Hz, 3H, CH2CH3).
13C NMR (100 MHz, DMSO-d6): δ 169.09, 147.86, 142.71, 129.45, 127.71, 121.18,
107.47, 65.97, 61.11, 14.50.
4.2.8 ethyl 2-((8-nitro-2,3,4,9-tetrahydro-1H-carbazol-6-yl)oxy)acetate (11)
A suspension of 9 (5.8554 g, 24.4 mmol) in conc. HCl (40 mL) was cooled to
-10 °C, to which was added a solution of NaNO2 (2.027 g, 29.28 mmol) in water (3
mL) dropwise. After stirred for 0.5 h, a solution of SnCl2 (11.04 g, 48.8 mmol) in conc.
HCl (10 mL) was added slowly, and the mixture was stirred overnight and was used
directly in the next step.
The phenylhydrazine 10 in conc. HCl was diluted with EtOH (50 mL). To this
solution, cyclohexanone (2.5 mL, 24.4 mmol) was added dropwise. The mixture was
heated to reflux for 3 hours. The acquired orange solid was filtered, washed with
water (10 mL) twice and dried under infrared oven to give a red solid (3.2376 g,
1
41.7%). m.p.: 130-131 °C. H NMR (400 MHz, CDCl3): δ 9.34 (s, 1H, indole NH),
7.66 (s, 1H, ArH), 7.38 (s, 1H, ArH), 4.72 (s, 2H, OCH2CO), 4.32 (q, J = 7.12 Hz, 2H,
COCH2CH3), 2.79 (t, 2H, 1-CH2), 2.67 (t, 2H, 4-CH2), 1.93 (m, 4H, 2,3-CH2CH2)
1.34 (t, J = 7.08 Hz, 3H, CH2CH3).13C NMR (100 MHz, CDCl3): δ 168.83, 150.81,
138.51, 132.31, 131.41, 125.74, 112.82, 111.13, 105.03, 66.90, 61.45, 23.17, 22.86,
22.80, 20.56, 14.18.
4.2.9 ethyl 2-((1-nitro-9H-carbazol-3-yl)oxy)acetate (12)
A suspension of 11 (3.2376 g, 10.17 mmol) and DDQ (5.54 g, 24.41 mmol) in
toluene (50 mL) was refluxed at 110 °C for 2 hours. The mixture was diluted with
EtOAc (100 mL), washed with saturated aqueous NaHCO3 (100 mL) for five times
till aqueous phase turned colorless. The organic phase was separated, dried over
24 / 45

ACCEPTED MANUSCRIPT
anhydrous Na2SO4 and concentrated to afford a reddish brown solid (3.4166 g,
quantitative). m.p.: 134-136 °C. 1H NMR (400 MHz, DMSO-d6): δ 11.97 (s, 1H, NH),
8.37 (s, 1H, ArH), 8.23 (d, J = 7.76 Hz, 1H, ArH), 7.86 (s, 1H, ArH), 7.71 (d, J = 8.16
Hz, 1H, ArH), 7.51 (t, 1H, ArH), 7.27 (t, 1H, ArH), 4.98 (s, 2H, OCH2CO), 4.21 (q, J
= 7.08 Hz, 2H, COCH2CH3), 1.24 (t, J = 7.16 Hz, 3H, CH2CH3). 13C NMR (100 MHz,
DMSO-d6): δ 169.15, 150.61, 141.93, 131.35, 129.40, 128.26, 128.06, 121.81, 121.46,
120.69, 116.03, 113.05, 108.62, 66.58, 61.18, 14.51.
4.2.10 2-((1-nitro-9H-carbazol-3-yl)oxy)acetic acid (13)
Ester 12 (1.255g, 4 mmol) was dissolved in THF (12 mL), and then a solution of
LiOH (338.6 mg, 8 mmol) in water (12 mL) was added dropwise. The mixture was
stirred at room temperature for 4.5 h. The mixture was diluted with EtOAc (80 mL),
extracted with saturated aqueous NaHCO3 (80 mL) for 3 times. The combined
aqueous phase was acidified with 10% HCl until pH = 1. The yellow precipitate
(1.1402 g, 99.8%) was collected by filtration and dried under infrared oven. m.p.:
180-181 °C. 1H NMR (400 MHz, DMSO-d6): δ 13.11 (br s, 1H, COOH), 11.97 (s, 1H,
NH), 8.37 (s, 1H, ArH), 8.24 (d, J = 7.68 Hz, 1H, ArH), 7.85 (s, 1H, ArH), 7.71 (d, J
= 8.04 Hz, 1H, ArH), 7.51 (t, 1H, ArH), 7.27 (t, 1H, ArH), 4.89 (s, 2H, CH2COOH).
13C NMR (100 MHz, DMSO-d6): δ 170.65, 150.84, 141.93, 131.37, 129.31, 128.24,
128.05, 121.82, 121.48, 120.68, 115.95, 113.04, 108.46, 66.35.
4.2.11 3-methyl-5-(((1-nitro-9H-carbazol-3-yl)oxy)methyl)-1,2,4-oxadiazole (14)
Acid 13 (167.0 mg, 0.58 mmol), HOBt (95.7 mg, 0.7 mmol) and EDCI (129.6 mg,
0.67 mmol) were slurried in dry acetonitrile (2 mL). The mixture was stirred at room
temperature for 0.5 h. Methylamidoxime (57.3 mg, 0.61 mmol) in dry MeCN (1.5 mL)
was added, then the mixture was refluxed overnight. After completion the mixture
was concentrated and purified with flash column chromatography (PE/EtOAc = 3/1 as
1
mobile phase) to afford an orange solid (101.8 mg, 53.8%). m.p.: 147-149 °C. H
NMR (400 MHz, DMSO-d6): δ 12.04 (s, 1H, NH), 8.49 (d, J = 1.68 Hz, 1H, ArH),
8.24 (d, J = 7.80 Hz, 1H, ArH), 8.01 (d, J = 1.68 Hz, 1H, ArH), 7.72 (d, J = 8.16 Hz,
25 / 45

ACCEPTED MANUSCRIPT
1H, ArH), 7.52 (t, 1H, ArH), 7.28 (t, 1H, ArH), 5.70 (s, 2H, OCH2), 2.38 (s, 3H, CH3).
13C NMR (100 MHz, CDCl3): δ 174.09, 167.61, 150.33, 140.60, 131.25, 130.04,
128.23, 128.20, 121.76, 121.09, 120.79, 115.47, 111.72, 108.31, 62.55, 11.56.
4.2.12 2-methyl-5-(((1-nitro-9H-carbazol-3-yl)oxy)methyl)-1,3,4-oxadiazole (15)
Acid 13 (561.7 mg, 1.96 mmol), acetohydrazide (291.4 mg, 4 mmol), HOBt
(400.6 mg, 3 mmol) and EDCI (565.4 mg, 3 mmol) were combined in DMF (8 mL).
The mixture was stirred at room temperature for 12 hours. Then the solution was
diluted with water (50 mL). Red precipitate was filtered, washed with water (10 mL)
for 3 times and dried. The red solid (502.0 mg, 74.7%) was used directly in the next
step.
The red solid obtained (173.2 mg, 0.5 mmol) was dissolved in POCl3 (2 mL) in a
sealed tube which was then, heated to 100 °C for 1 h. Then the mixture was poured
into cold water, neutralized with saturated NaHCO3 solution (pH = 12) and extracted
with DCM (50 mL) twice. The combined DCM solution was dried with anhydrous
Na2SO4, concentrated and the crude solid was triturated in DCM/PE=1/1. The
acquired orange solid was filtered and washed with DCM/PE=1/1 to afford pure
1
product (107.9 mg, 65.8%). m.p.: 163-164 °C. H NMR (400 MHz, DMSO-d6): δ
12.02 (s, 1H, NH), 8.47 (s, 1H, ArH), 8.23 (d, J = 7.76 Hz, 1H, ArH), 8.02 (s, 1H,
ArH), 7.72 (d, J = 8.08 Hz, 1H, ArH), 7.52 (t, 1H, ArH), 7.28 (t, 1H, ArH), 5.57 (s,
2H, CH2), 2.55 (s, 3H, CH3). 13C NMR (100 MHz, DMSO-d6): δ 165.30, 162.78,
150.18, 141.98, 131.40, 129.69, 128.33, 128.19, 121.78, 121.46, 120.80, 116.53,
113.11, 108.95, 61.49, 10.97.
4.2.13 8-nitro-2,3,4,9-tetrahydro-1H-carbazol-6-ol (18)
A suspension of aniline 16 (4.6192 g, 30 mmol) in conc. HCl (50 mL) was stirred
at room temperature for 0.5 h till black solid turned into white suspension then cooled
to -10 °C, to which a solution of NaNO2 (2.4897 g, 36 mmol) in water (3.5 mL) was
added dropwise. After stirred for 1 h, a solution of SnCl2 (13.6 g, 60 mmol) in conc.
HCl (10 mL) was added slowly. After completion the mixture was stirred at -10 °C for
26 / 45

ACCEPTED MANUSCRIPT
1h, followed by stirr at room temperature for another 3 hours. The mixture was used
directed in the next step.
The phenylhydrazine 17 solution in conc. HCl was diluted with EtOH (50 mL), to
which was added dropwise with cyclohexanone (3.4 mL, 33 mmol). The mixture was
heated and refluxed overnight. Black solid was filtered, washed with water (10 mL)
twice and dried under infrared oven to give a black solid (2.3455 g, 38.1%). m.p.:
1
205-206 °C. H NMR (400 MHz, DMSO-d6): δ 11.19 (s, 1H, NH), 9.44 (s, 1H, OH),
7.43 (s, 1H, ArH), 7.22 (s, 1H, ArH), 2.74 (t, 2H, 1-CH2), 2.58 (t, 2H, 4-CH2), 1.80 (m,
13
4H, 2, 3-CH2). C NMR (100 MHz, DMSO-d6): δ 149.91, 139.70, 132.79, 131.49,
124.07, 112.58, 109.62, 104.86, 23.40, 23.07, 23.00, 20.69.
4.2.14 8-nitro-2,3,4,9-tetrahydro-1H-carbazol-6-yl trifluoromethanesulfonate (19)
To a suspension of 18 (116.6 mg, 0.5 mmol) in DCM (5 mL), 2,6-Lutidine (64 uL,
0.55 mmol) was added dropwise in an ice bath. Subsequently, Tf2O (68.9 uL, 0.55
mmol) was added over 5 min. The mixture was stirred overnight. Then, another
portion of 2,6-Lutidine (128 uL, 1.1 mmol) and Tf2O (68.9 uL, 0.55 mmol) was added,
and the mixture was stirred for another 5.5 hours. After completion, the mixture was
diluted with DCM (20 mL), washed with 2N HCl (20 mL) twice, brine (20 mL), dried
over anhydrous Na2SO4 and concentrated to afford a yellow solid (181.7 mg, 99.3%).
m.p.: 133-135 °C. 1H NMR (400 MHz, DMSO-d6): δ 11.93 (s, 1H, NH), 7.98 (s, 2H,
ArH), 2.80 (t, 2H, 1-CH2), 2.66 (t, 2H, 4-CH2), 1.82 (m, 4H, 2, 3-CH2). 13C NMR
(100 MHz, DMSO-d6): δ 142.41, 140.63, 132.47, 131.23, 127.63, 118.78 (q, J = 320
Hz), 118.10, 111.63, 110.00, 23.42, 22.77, 22.71, 20.55.
4.2.15 1-nitro-9H-carbazol-3-yl trifluoromethanesulfonate (20)
A suspension of 19 (181.7 mg, 0.5 mmol) and DDQ (272.4 mg, 1.2 mmol) in
toluene (5 mL) was refluxed at 110 °C overnight. The mixture was diluted with
EtOAc (50 mL), washed with saturated aqueous NaHCO3 (50 mL) for 5 times till
aqueous phase turned colorless. The organic phase was separated, dried over
anhydrous Na2SO4 and concentrated. The crude product was purified by flash column
27 / 45

ACCEPTED MANUSCRIPT
chromatography (DCM/PE=1/3) to afford a yellow solid (126.5 mg, 77.5%). m.p:
142-144 °C. 1H NMR (400 MHz, DMSO-d6): δ 12.46 (s, 1H, NH), 8.89 (s, 1H, ArH),
8.37 (s, 2H, ArH), 7.78 (d, 1H, ArH), 7.59 (t, 1H, ArH), 7.35 (t, 1H, ArH).13C NMR
(100 MHz, DMSO-d6): δ 142.16, 140.37, 132.47, 131.24, 129.04, 128.47, 121.93,
121.65, 121.57, 121.45, 118.81 (q, J = 320 Hz), 114.87, 113.47. 19F NMR (376 MHz,
DMSO-d6): δ -72.41.
4.2.16 ethyl 2-((1-nitro-9H-carbazol-3-yl)thio)acetate (21)
Pd2(dba)3 (2311.4 mg, 0.25 mmol) and xantphos (292.4 mg, 0.5 mmol) were
premixed in dioxane (4 mL) under Ar2 for 1 hour. Intermediate 20 (360 mg, 1 mmol)
was added subsequently, the reaction mixture was backfilled with Ar2 (2 cycles). Then
DIPEA (0.35 mL) and ethyl mercaptoacetate (0.11 mL) were added. The mixture was
refluxed overnight (20 hours). After completion, the mixture was diluted with EtOAc
(50 mL), filtered through a pad of celite to remove catalyst, washed with 2N HCl (50
mL) for 3 times, dried over anhydrous Na2SO4 and concentrated. The crude product
was purified by flash column chromatography (EtOAc/PE = 1/20 to 1/10) to afford a
yellow solid (250.0 mg, 75.6%). m.p.: 134-135 °C. 1H NMR (400 MHz, DMSO-d6): δ
12.25 (s, 1H, carbazole NH), 8.76 (s, 1H, ArH), 8.35 (s, 1H, ArH), 8.27 (d, J = 7.76
Hz, 1H, ArH), 7.75 (d, J = 8.12 Hz, 1H, ArH), 7.54 (t, 1H, ArH), 7.32 (t, 1H, ArH),
4.07 (q, J = 7.12 Hz, 2H, CH2CH3), 3.93 (s, 2H, SCH2), 1.10 (t, J = 7.08 Hz, 3H,
CH2CH3).13C NMR (100 MHz, DMSO-d6): δ 169.75, 141.58, 132.83, 131.95, 131.59,
128.42, 128.30, 124.83, 123.72, 121.48, 121.42, 113.22, 61.35, 37.72, 14.38.
4.2.17 ethyl 2-((1-(4-methylphenylsulfonamido)-9H-carbazol-3-yl)oxy)acetate (22a)
To a suspension of 12 (315.1 mg, 1 mmol) in MeOH (25 mL), Pd/C (10%, 65.7
mg) was added. The mixture was stirred in a hydrogen generator (40 psi, room
temperature) overnight. The suspension was filtered through a pad of celite, and the
filtrate was concentrated under reduced pressure. The crude product was used in the
next step without further purification.
To a solution of the obtained aniline in pyridine (10 mL), TsCl (263.1 mg, 1.38
28 / 45