Development of Small-Molecule Cryptochrome Stabilizer Derivatives as Modulators of the Circadian Clock

Article abstract of DOI:10.1002/cmdc.201500260

Small-molecule probes have been playing prominent roles in furthering our understanding of the molecular underpinnings of the circadian clock. We previously discovered a carbazole derivative, KL001 (N-(3-(9H-carbazol-9-yl)-2-hydroxypropyl)-N-(furan-2-ylme

Full text of DOI:10.1002/cmdc.201500260

DOI: 10.1002/cmdc.201500260
Full Papers
Development of Small-Molecule Cryptochrome Stabilizer
Derivatives as Modulators of the Circadian Clock
Jae Wook Lee,[a, c] Tsuyoshi Hirota,*[b, d, e] Anupriya Kumar,*[b, f] Nam-Jung Kim,[a, g]
Stephan Irle,[b, f] and Steve A. Kay*[b, d]
Small-molecule probes have been playing prominent roles in
furthering our understanding of the molecular underpinnings
of the circadian clock. We previously discovered a carbazole
derivative, KL001 (N-(3-(9H-carbazol-9-yl)-2-hydroxypropyl)-N-
(furan-2-ylmethyl)methanesulfonamide), as a stabilizer of the
clock protein cryptochrome (CRY). Herein we describe an ex-
tensive structure–activity relationship analysis of KL001 deriva-
tives leading to the development of a highly active derivative:
2-(9H-carbazol-9-yl)-N-(2-chloro-6-cyanophenyl)acetamide
(KL044). Subsequent 3D-QSAR analysis identified critical fea-
tures of KL001 derivatives and provided a molecular-level un-
derstanding of their interaction with CRY. The electron-rich car-
bazole, amide/hydroxy linker, sulfonyl group, and electron-
withdrawing nitrile moieties contribute to greater biological
activity. The hydrogen bonding interactions with Ser394 and
His357 as well as stronger CH–p interactions with Trp290 make
KL044 a better binder than KL001. KL044 lengthened the circa-
dian period, repressed Per2 activity, and stabilized CRY in re-
porter assays with roughly tenfold higher potency than KL001.
Altogether, KL044 is a powerful chemical tool to control the
function of the circadian clock through its action on CRY.
Introduction
In mammals, many physiological processes such as body tem-
perature, hormone secretion, metabolism, and sleep/wake be-
havior are controlled by the circadian clock in a daily rhythmic
manner. Cell-autonomous circadian rhythms are generated
through clock genes, which are orchestrated by interconnect-
ed transcriptional and translational regulatory loops.[1] The core
clock loop is constituted by two transcriptional activators,
CLOCK and BMAL1, and two repressors: Cryptochrome (CRY1
and CRY2) and Period (PER1 and PER2). The CLOCK–BMAL1
heterodimer activates the transcription of Cry and Per though
binding to E-box enhancer elements. Translated PER and CRY
proteins dimerize and translocate into the nucleus to inhibit
CLOCK–BMAL1 function, resulting in rhythmic gene expression
pattern. The loop regulates a large number of clock-controlled
genes in circadian physiology. The stabilities of clock proteins
are further controlled by posttranslational modification and
ubiquitination.[2] The phosphorylation of PER protein by casein
kinase 1 (CK1d/e) promotes its degradation through the ubiq-
uitin–proteasome pathway, dependent on F-box proteins b-
TrCP1 and b-TrCP2.[3] The stability of CRY protein is separately
modulated by degradation through the ubiquitin–proteasome
pathway. The SCFFBXL3 ubiquitin ligase complex plays an impor-
tant role in CRY ubiquitination and degradation to regulate the
negative feedback loop.[4] Recent X-ray crystallographic analysis
revealed details of CRY stability control at the atomic level. The
F-box protein FBXL3 recognizes CRY2 by occupying the FAD
binding pocket of CRY2 with its C-terminal tail and also covers
the PER binding domain of CRY2.[5]
To develop novel small-molecule modifiers of the circadian
clock, we previously screened a collection of 60000 structurally
[a] Dr. J. W. Lee,+ Dr. N.-J. Kim
[e] Dr. T. Hirota+
Department of Chemistry
PRESTO, Japan Science and Technology Agency, Nagoya, 464-8601 (Japan)
[f] Dr. A. Kumar,+ Prof. S. Irle
The Scripps Research Institute, La Jolla, CA, 92037 (USA)
[b] Dr. T. Hirota,+ Dr. A. Kumar,+ Prof. S. Irle, Prof. S. A. Kay
Institute of Transformative Bio-Molecules
Nagoya University, Nagoya, 464-8601 (Japan)
E-mail: thirota@itbm.nagoya-u.ac.jp
anupriya.veerman@gmail.com
[c] Dr. J. W. Lee,+
Department of Chemistry, Graduate School of Science
Nagoya University, Nagoya, 464-8602 (Japan)
[g] Dr. N.-J. Kim
Collage of Pharmacy
Kyung Hee University, Seoul, 130-701 (Republic of Korea)
[+] These authors contributed equally to this work.
Natural Product Research Center
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201500260: synthesis of KL001 derivatives,
cell-based assays, and 3D-QSAR studies.
Korea Institute of Science and Technology
Department of Biological Chemistry
University of Science and Technology (Republic of Korea)
ꢀ 2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA.
This is an open access article under the terms of the Creative Commons At-
tribution Non-Commercial NoDerivs License, which permits use and distri-
bution in any medium, provided the original work is properly cited, the use
is non-commercial and no modifications or adaptations are made.
[d] Dr. T. Hirota+ Prof. S. A. Kay
Molecular and Computational Biology Section
University of Southern California, Los Angeles, CA, 90089 (USA)
E-mail: stevekay@usc.edu
ChemMedChem 2015, 10, 1489 – 1497
1489 ꢀ 2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
analyzed using a cell-based Bmal1-dLuc circadian assay by test-
ing 12 points of threefold dilution series to obtain EC50 values
(Supporting Information Figure S1).
The synthetic routes followed for the preparation of carba-
zole derivatives 3, 7, 10, and 13 are depicted in Schemes 1 and
2. At first, the carbazole moiety was replaced with purine, imi-
dazole, phenylindole, and pyridoindole. For the synthesis of in-
termediate 2, furfuryl amines were sulfonylated with methane-
sulfonyl chloride under basic conditions followed by alkylation
with epichlorohydrin and sodium hydride in DMF (Scheme 1A).
The intermediate 2 was treated with carbazole substituents
and sodium hydride to provide the derivatives 3. Replacement
of the carbazole ring by purine (KL005), imidazole (KL006), and
phenylindole (KL007) moieties led to the loss of effect of these
compounds on the circadian period (Supporting Information
Table S1). The pyridoindole derivatives KL008 and KL010, which
are structurally similar to carbazole compounds KL001 and
KL009, showed severely decreased period effect. Therefore, it is
Figure 1. A) Structures of previously identified KL001 derivatives. B) Modifica-
tion sites of KL001 for structure–activity relationship (SAR) studies.
diverse compounds assembled
from an in-house compound li-
brary using a cell-based Bmal1-
dLuc reporter assay. We identi-
fied carbazole derivatives KL001,
KL002, and KL003 as period-
lengthening compounds (Fig-
ure 1A). Subsequent studies
using pull-down and proteomic
target identification revealed
that KL001 interacts with the
clock protein CRY. Interestingly,
KL001 competes with FAD for
CRY binding and stabilizes CRY
by inhibiting FBXL3-dependent
ubiquitination.[6] The co-crystal
structure of the KL001–CRY2
complex revealed that KL001 oc-
cupies the FAD binding site of Scheme 1. Modifications of KL001. A) Synthesis of modified carbazole. Reagents and conditions: a) MeSO2Cl, py, RT,
24 h, 85.3%; b) NaH, DMF, 0–508C, ~16 h, 52.1%; c) NaH/DMF, 608C, 2 h. B) The synthesis of 3-acylation derivatives
with acyl halide. Reagents and conditions: a) 1. NaH/DMF, 2. THF, RT, ~16h; b) DIEA, RT, ~16h; c) carbazole, NaH/
DMF, 508C, 2 h.
CRY2 and interferes with the
binding of FBXL3 C-terminal to
CRY2.[7] KL001 is the first com-
pound to target CRY, and its fur-
ther characterization is important to understand the regulation
of CRY function. Herein we describe the structural require-
ments for KL001 activity and the molecular basis for a highly
potent derivative that we discovered.
clear that the carbazole ring is essential for control of the circa-
dian period.
We next modified the methanesulfonyl group by removing
its sulfonyl group (compounds KL011 and KL012). Similar to
KL004, the compounds showed significantly lowered period-
lengthening activity (Table 1 and Supporting Information
Table S1). The derivatives with an aliphatic and aromatic acyl
group appeared to have decreased period effect (Tables 1 and
S1). We therefore evaluated compounds with aliphatic and aro-
matic acyl groups at the nitrogen atom, or possessing other
sulfonyl groups (Scheme 1B). Intermediate 5 was obtained
from carbazole using sodium hydride for deprotonation of
a carbazole followed by reaction with furfuryl amine. The acy-
lated derivatives KL013–KL019 were synthesized by acylation
Results and Discussion
KL001 consists of a cabazole structure linked to 2-hydroxy-
propyl and furan groups (Figure 1B). Our preliminary SAR
study identified a less active analogue, KL004, which lacks the
terminal methanesulfonyl group (Figure 1A).[6] To systemically
optimize each subsection of KL001, we designed and synthe-
sized the KL001 derivatives considering four modification sites
(Schemes 1 and 2). The period effect of these derivatives was
ChemMedChem 2015, 10, 1489 – 1497
www.chemmedchem.org
1490 ꢀ 2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Table 1. Representative structures of KL001 derivatives and their period-lengthening effects.
ID
Structure
Period effect
ID
Structure
Period effect
log(EC50 [m])
log(EC50 [m])
KL001
À6.16
À5.57
KL011
À4.82
À4.73
KL013
KL019
KL016
KL021
À5.61
À5.49
KL022
KL032
À5.70
À4.25
KL028
KL033
À4.94
À4.23
KL034
KL046
À6.01
À6.46
KL044
KL048
À7.32
À6.12
of intermediate 5 under basic conditions. KL013 and KL014,
which were acylated with relatively short aliphatic and aromat-
ic groups, showed tolerance to modification. However, deriva-
tives with nicotinic groups (KL016 and KL017) showed a further
decrease in activity. KL020 and KL021, whose sulfonyl moieties
were modified with phenyl or cyclopropyl groups, showed re-
duced period effects.
tives with benzyl and phenyl groups showed lower activity
(Tables 1 and S1). Interestingly, KL034, which has a methoxyeth-
yl group, exhibited potency similar to that of KL001. However,
other derivatives with ethylaminoacetyl and dimethylami-
noethyl groups (KL031–KL033) showed weaker activity. There-
fore, a methoxyethyl group can be a replacement for furan.
Further modification of the sulfonyl group did not improve po-
tency (compounds KL036–KL038). Derivatives with phenyl and
cyclopropylacyl or sulfonyl moieties (KL039–KL042) showed
a decreased period effect.
We then replaced furan with phenyl, benzyl, or aliphatic
moieties (Scheme 2A). Intermediate 9 was obtained from reac-
tion of aromatic or aliphatic amine with methanesulfonyl chlo-
ride followed by reaction with sodium hydride and epichloro-
hydrin. In the next step, intermediate 9 was linked with a carba-
zole moiety to generate derivatives KL022–KL035. The deriva-
To explore the role of the linker, we introduced an acet-
amide linker with
a variety of anilines (KL043–KL064;
Scheme 2B). Intermediate 12 was obtained by reaction with
ChemMedChem 2015, 10, 1489 – 1497
www.chemmedchem.org
1491 ꢀ 2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
cy in terms of Àlog(EC50) in molar units (pEC50), rang-
ing from 3.94 to 7.32. For 3D-QSAR studies, a random
method was used for dividing the data set into two
sets: 54 compounds in the training set, and six com-
pounds in the test set (Table S1). Technical details of
structure preparation, docking, pharmacophore tem-
plate generation, and 3D-QSAR are given in Support-
ing Information Methods and Figure S2. Figure 2B
gives a representation of the electrostatic and steric
coefficients of the 3D-QSAR model. The linear rela-
tionship between experimental and predicted activity
of both the training and test sets (Figure 2C) indi-
cates the successful construction of a reliable model
with high predictive power. We also performed
leave-many-out cross-validation by leaving out 20%
of the molecules at each step for the same set of
training and test set compounds as used for leave-
one-out cross-validation. The values for leave-one-out
(LOO) cross-validated correlation coefficient (q2), non-
Scheme 2. Modifications of KL001. A) Synthesis of furan-modified derivatives with aro-
matic and aliphatic amines. Reagents and conditions: a) 1. pyridine, RT, overnight,
2. NaH/DMF, 508C, overnight; b) carbazole, NaH/DMF, 608C, 2 h. B) Synthesis of acet-
amide linker derivatives. Reagents and conditions: a) DMF, RT, or 408C, or Et3N/CH2Cl2, RT,
overnight; b) carbazole, NaH/DMF, 608C, overnight.
various anilines and chloroacetyl chloride in DMF. Compound
13 was synthesized by nucleophilic reaction with chloroacetyl
anilines 12 and carbazole using NaH/DMF. Derivatives with
ortho-substituted phenyl groups (KL043–KL053) exhibited rela-
tively high period effects, except KL045, with ortho- and para-
substituted phenyl groups (Tables 1 and S1). We found that
KL044 with an acetamide linker showed ~10-fold higher po-
tency than KL001. Interestingly, nitrile group substitution at
the meta position of the phenyl group (KL055) results in com-
plete loss in activity (Table S1). Therefore, ortho substitution at
the phenyl group is important
cross-validated correlation coefficient (r2), number of compo-
nents (N), predicted non-cross-validated correlation coefficient
(rpred2), and predicted RMSE were 0.639, 0.966, 6, 0.827, and
0.442, respectively. The respective analogous values for leave-
many-out (LMO) parameters q2, r2, N, rpred2, RMSEpred were 0.606,
0.966, 6, 0.827, and 0.462. Hence, the stability of the 3D-QSAR
model is robust, as indicated by the similar predictive values
obtained by both LOO and LMO cross-validation methods.
Based on the model, the higher activity corresponds to the fol-
lowing features: 1) negative electrostatics (cyan tetrahedrons)
for
the
period-lengthening
effect.
To obtain
understanding of our SAR study,
we further conducted 3D-
a molecular-level
a
QSAR study which uses a ligand-
field-based approach.[8] The mo-
lecular fields include electrostat-
ic, steric, hydrophobic, and
van der Waals fields calculated
by the interaction of the ligand
with positive, negative, and neu-
tral atoms as probes. These
ligand fields can be used as mo-
lecular features, yielding a so-
called “pharmacophore”, to iden-
tify structural regions in the li-
gands crucial for bioactivity. In
our study, field points of KL001
and KL044 were used for gener-
ating
a pharmacophore (Fig-
ure 2A) in FieldTemplater avail-
able in the Forge software pack-
age,[9] which was used to align
a diverse range of KL001 deriva-
tives. The data included a set of
Figure 2. A) Model template generated by the conformational search of KL001 and KL044. B) 3D-QSAR model gen-
erated by using pEC50 values of 54 training compounds and six test set compounds. C) Plot of experimental and
60 compounds with their poten- predicted pEC50 values.
ChemMedChem 2015, 10, 1489 – 1497
www.chemmedchem.org
1492 ꢀ 2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
contributed by electron-rich carbazole, amide/hydroxy linker,
and sulfonyl moieties; 2) positive electrostatics (red tetrahe-
drons) mediated by the electron-withdrawing nitrile and hy-
droxy groups; and 3) steric bulk associated with the furyl
moiety (green tetrahedrons). The purple tetrahedrons show
areas where it is best to avoid bulkier functional groups.
pocket depicted by the green surface of the contour map
drawn for the van der Waals repulsive term. In contrast, the un-
favorable steric field points (9–14) are located at the edge of
the contour map away from this empty pocket. In Figures 3B
and 3B’, the amide moiety in KL044 overlaps with the hydro-
philic map. The positive electrostatic field points (6–8) and
a negative electrostatic field point (4) are within 3 Š of the
ligand. In Figures 3C and 3C’, the hydrophobic surface map
overlaps with the carbazole as well as the disubstituted ben-
zene moiety of KL044. The negative electrostatic field points
(1–3, 5) are within 3 Š of the ligand. This indicates that KL044
is a good fit in the pocket of the CRY binding site, and field
points indicate the opportunities for obtaining additional inter-
actions in the hydrophilic and hydrophobic pockets. Unfavora-
ble steric interactions with residues Trp290, Trp397, Val390 and
Phe294 should be avoided. Additionally, Figure S3 (Supporting
Information) shows the superimposition with the CRY protein
structure containing FBXL3 protein bound at the FAD binding
site, (PDB ID: 4I6J[5]). The surface contour (Figure 3A) extends
in the space occupied by FBXL3 residues such as Trp428 and
Thr427. Moreover, the position of FBXL3 P426 bound to CRY
protein overlaps with a negative electrostatic field point
number 5. Overall, we found that the 3D-QSAR results and the
nature of the amino acid residues in the CRY ligand binding
domain are very well correlated.
For studying the complementarity of the binding pocket of
the CRY protein, we further superimposed the selected field
points with the higher coefficient values from the 3D-QSAR
model and KL044 on the CRY protein structure (PDB ID:
4MLP[7]) as shown in Figure 3. The binding site analysis for CRY
protein was carried out using Sitemap,[10] and the superimposi-
tion was performed with Maestro.[10] The surfaces were gener-
ated depicting the contour maps obtained by van der Waals
repulsion, and hydrophilic and hydrophobic regions. The
values for SiteScore, Dscore (druggability) and the ratio of hy-
drophobic and hydrophilic character were 1.109, 1.051, and
1.050, respectively, indicating that CRY protein has a druggable
tight-binding pocket. The good ligand could be a better ac-
ceptor than a donor of hydrogen bonds, indicated by the ratio
of donor verses acceptor as 0.692. In Figures 3A and 3A’, fa-
vorable steric field points (15–19) overlap with the empty
It is important to understand how a field pattern of a particu-
lar molecule contributes to its predicted activity. Figure 4A
shows the medicinal chemistry evolution of KL044 based on
the contribution of the field points to the predicted activity.
When the isopropyl moiety is introduced to KL046, the activity
is significantly decreased (KL058, DpEC50 2.2) due to unfavora-
ble electrostatics (orange cubes) and steric clash (purple
cubes). This emphasizes the importance of having a hydrogen
bond donor such as the ÀNH moiety. By introducing Cl at the
ortho position of the benzyl group, KL044 with favorable elec-
trostatics and sterics can be obtained.
For validating the QSAR model, we predicted the activities
for five compounds, KL065–KL069 (Supporting Information,
Figure 3. Overlap of QSAR field points on KL044 and surface maps generat-
ed for the binding pocket of CRY (PDB ID: 4MLP) by Sitemap depicted as:
A) surface contour obtained by van der Waals repulsion, B) hydrophilic, and
C) hydrophobic. Color code for field point coefficients represented as num-
bered spheres: negative electrostatic (field points 1–5, cyan); positive elec-
trostatic (field points 6–8, red); unfavorable steric (field points 9–14, purple);
favorable steric (field points 15–19, green).
Figure 4. Evolution of medicinal chemistry based on the contribution of the
field points to the predicted activity in 3D-QSAR model for A) KL044 and
B) KL065. Experimental pEC50 values are shown in brackets. Color code for
field point contributions: favorable electrostatic (green), unfavorable electro-
static (orange), favorable steric (cyan), unfavorable steric (purple).
ChemMedChem 2015, 10, 1489 – 1497
www.chemmedchem.org
1493 ꢀ 2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Methods). We synthesized these compounds and analyzed
their activities. An excellent agreement between predicted and
experimental bioactivity further supported the reliability of the
model (Table S1). Figure 4B shows a comparison of the predict-
ed compound KL065 with KL001 and KL034. The van der Waals
radii for chloro (1.8 Š) and methyl (2.0 Š) groups are similar,
and these are well-known isosteres. The chloro moiety in
KL044 and methyl group in KL065 overlap very well. Hence, we
found that replacement of the furyl moiety (KL001) and me-
thoxyethane group (KL034) with methyl retains the activity, as
shown by the favorable electrostatic (cyan cubes) and de-
creased sterics (purple cubes). The logP values for KL001 and
KL065 are 3.0 and 2.4, respectively, indicating that KL065 has
higher solubility due to the less bulky substituent at the trigo-
nal nitrogen atom. Figure S4 provides other examples for con-
tribution of a field pattern on the predicted activity.
To further understand why KL044 provides the best activity
among the tested compounds, differences in the electrostatic
potential maps, which can be used to explain the differences
in higher or lower activity of the compounds, were analyzed
by using Activity Miner.[11] The ligand binding site residues in
CRY protein and electrostatic potential (ESP) map difference
between KL044 and KL001 are shown in Figure 5. In KL044, the
positive potential cloud (red) due to the ÀNH moiety, as well
as the negative cloud (cyan) due to the disubstituted benzyl
with the electron-withdrawing groups at the ortho position
and ÀC=O correlated with in-
Figure 5. Ligand binding site residues in CRY protein interacting with
A) KL044 and B) KL001 (distances given in Š). Electrostatic map difference
between C) KL044 and D) KL001 based on field similarity.
dLuc and Per2-dLuc cell-based reporter assays. KL044 length-
ened the circadian period and also suppressed Per2-dLuc activi-
ty much stronger than KL001 (Figures 6A and 6B), suggesting
an activation of CRY, a repressor of the Per gene. Consistent
with this observation, we revealed that KL044 potently stabi-
lized the CRY1–LUC fusion protein in cell-based degradation
assay without affecting LUC stability (Figure 6C). The parallel
relationship between KL044 and KL001 against CRY1 stability,
creased activity over KL001. This
highlights the importance of in-
teractions such as the ÀNH and
ÀC=O groups forming hydrogen
bonds with Ser394 (3.06 Š) and
His357 (2.35 Š), respectively.
Moreover, the disubstituted
benzyl group forms a CH–p in-
teraction with Trp290 (3.71 Š). In
KL001, the positive cloud due to
trisubstituted nitrogen and the
negative cloud due to the sulfo-
nyl group result in decreased ac-
tivity relative to KL044. The im-
portant interactions for KL001
are the hydrogen bond formed
by the ÀOH group with Ser394
(2.75 Š) and CH–p interactions
formed by the methanesulfonyl
moiety with Trp290 (4.50 Š) and
Trp397 (4.10 Š). Hence, these
characteristics of the compound
make KL044 a better fit than
KL001 in the CRY binding
pocket.
In this experimental–theoreti-
cal study, we discovered a highly
potent derivative KL044. We
therefore characterized its bio-
Figure 6. Effects of KL001 and KL044 in A) Bmal1-dLuc and B) Per2-dLuc cell-based circadian assays. C) Effects of
logical activities by using Bmal1- KL001 and KL044 in CRY1-LUC and LUC degradation assays in the HEK293 stable cell line.
ChemMedChem 2015, 10, 1489 – 1497
www.chemmedchem.org
1494 ꢀ 2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
2.23 ppm (s, 3H); 13C NMR (CDCl3, 100 MHz): d=173.4, 148.8, 142.7,
140.6, 125.9, 123.0, 120.3, 119.2, 110.2, 109.0, 108.6, 71.0, 52.3, 47.4,
47.0, 21.7 ppm; ESI-MS: 363.1 [M+H], 385.1 [M+Na].
Per2 repression, and period lengthening supports the mecha-
nism of action of KL044 as a CRY stabilizer for circadian regula-
tion.
N-(3-(9H-Carbazol-9-yl)-2-hydroxypropyl)-N-(furan-2-ylmethyl)pi-
colinamide (KL016). A solution of 1-(9H-carbazol-9-yl)-3-((furan-2-
ylmethyl)amino)propan-2-ol (50 mg, 0.16 mmol) in DMF (4 mL) was
treated with DIEA (35 mL, 0.22 mmol), HATU (76 mg, 0.2 mmol), and
picolinic acid (25 mg, 0.2 mmol). After stirring at room temperature
overnight, the reaction mixture was extracted with CH2Cl2 (50 mL)
and H2O (30 mL). The organic layer was dried with Na2SO4 and
evaporated under reduced pressure. The crude material was puri-
fied by column chromatography to give 41.5 mg (62.6%) of KL016;
1H NMR (CDCl3, 400 MHz): d=8.03 (d, J=7.6 Hz, 2H), 7.65–7.62 (m,
2H), 7.50–7.38 (m, 5H), 7.24–7.19 (m, 3H), 7.05–7.01 (m, 1H), 6.22
(dd, J=3.2, 2.0 Hz, 1H), 6.11 (d, J=2.8 Hz, 1H), 5.13 (d, J=15.2 Hz,
1H), 4.58–4.49 (m, 2H), 4.38–4.27 (m, 2H), 3.52 (dd, J=14.8,
12.0 Hz, 1H), 3.19 ppm (dd, J=15.2, 3.2 Hz, 1H); 13C NMR (CDCl3,
100 MHz): d=167.8, 151.9, 149.9, 145.7, 142.4, 141.1, 138.0, 125.7,
125.5, 124.9, 122.9, 120.1, 119.1, 110.5, 109.5, 109.4, 68.0, 50.6, 47.5,
41.5 ppm; ESI-MS: 426.2 [M+H], 448.2 [M+Na].
Conclusions
CRY is a key regulator not only for the circadian clock but also
for the hepatic gluconeogenesis,[12] suggesting a potential use
of KL001 derivatives as circadian rhythm modulators as well as
antidiabetics. KL044 provides a potent chemical probe for the
regulation of CRY protein function. Our SAR and 3D-QSAR
studies in combination with the published CRY–KL001 co-crys-
tal structure aid in our understanding of the mechanism of
CRY regulation in terms of molecule–protein interactions.
Future diversification of the compounds by combining the
pharmacophoric features of KL044 and KL001 will provide
even better compounds for modulating circadian function and
gluconeogenesis. Together with a recently published study in
which period-shortening derivatives of KL001 through CÀH ac-
tivation were discovered,[13] the development of KL001 ana-
logues clearly demonstrates the power of chemical biology ap-
proaches in the research field of the circadian clock.
N-(3-(9H-Carbazol-9-yl)-2-hydroxypropyl)-2-cyano-N-(furan-2-yl-
methyl)benzamide (KL019). Same procedure as described for the
synthesis of KL016. 1-(9H-Carbazol-9-yl)-3-((furan-2-ylmethyl)ami-
no)propan-2-ol (50 mg, 0.16 mmol), DIEA (35 mL, 0.22 mmol), HATU
(76 mg, 0.2 mmol), and 2-cyanobenzoic acid (29 mg, 0.2 mmol)
were combined in DMF (4 mL) to give 38.1 mg (56.6%) of KL019;
1H NMR (CDCl3, 400 MHz): d=8.10 (d, J=8.0 Hz, 2H), 7.65 (d, J=
8.0 Hz, 2H), 7.62 (d, J=8.0 Hz, 2H), 7.47–7.37 (m, 4H), 7.28–7.23
(m, 2H), 7.13 (s, 1H), 5.95 (s, 1H), 5.46 (s, 1H), 4.41–4.23 (m, 3H),
4.17–4.08 (m, 2H), 3.89 (dd, J=14.0, 8.4 Hz, 1H), 3.36 ppm (d, J=
14.4 Hz, 1H); 13C NMR (CDCl3, 100 MHz): d=171.9, 149.0, 147.7,
144.5, 143.1, 140.5, 139.5, 132.5, 128.0, 126.2, 126.0, 123.0, 120.4,
119.3, 114.0, 110.3, 109.7, 108.9, 70.5, 51.2, 47.7, 47.1 ppm; ESI-MS;
450.2 [M+H], 472.1 [M+Na].
Experimental Section
Characterization of representative KL001 derivatives. 1H and
13C NMR spectra were recorded on a Bruker 400 MHz spectrometer.
Chemical shifts are reported relative to internal CDCl3 (CH4Si, d=
0.0), CD3OD (CH4Si, d=0.0), and CD3CN (CH4Si, d=0.0). All com-
pounds were identified by LC–MS (Agilent Technology) using a C18
column (20”4.0 mm) with 20 min elution using a gradient solution
of CH3CN/H2O (5:95 v/v containing 0.05% trifluoroacetic acid
(TFA)), with UV detector and an electrospray ionization source.
General procedure for the synthesis of compound 10. We fol-
lowed the previously described procedure for the synthesis of
compound 9.[6] A solution of carbazole (250 mg, 1.5 mmol) in DMF
(5 mL) was treated with NaH (41 mg, 1.7 mmol, 1.2 equiv) and
stirred for 30 min at room temperature, followed by compound 9
(242 mg, 0.85 mmol) in DMF (2 mL). The reaction mixture was
stirred for 2 h at 608C. The reaction mixture was diluted with
EtOAc, washed with 5% HCl and brine. The crude material was pu-
rified by flash column chromatography to give the final products.
1-(9H-Carbazol-9-yl)-3-((thiophen-2-ylmethyl)amino)propan-2-ol
(KL011). The reaction mixtures of 9-(oxiran-2-ylmethyl)-9H-carba-
zole (29.2 mg, 0.13 mmol) and thiophen-2-ylmethanamine (200 mL,
2.0 mmol) were heated at 558C. The crude material was purified by
preparative HPLC (0.05% TFA/H2O / 0.05% TFA/CH3CN) to give
1
26 mg (81.3%) of KL011; H NMR (CD3OD, 400 MHz): d=8.16 (d, J=
8 Hz, 2H), 7.61 (d, J=8.5 Hz, 2H), 7.54–7.51 (m, 3H), 7.30 (t, J=
8 Hz, 2H), 7.17 (d, J=3 Hz, 1H), 7.07 (dd, J=5, 3.5 Hz, 1H), 4.54–
4.46 (m, 5H), 3.17 (dd, J=12.5, 10 Hz, 1H), 3.10 ppm (dd, J=12.5,
2.5 Hz, 1H); 13C NMR (CD3OD, 100 MHz): d=141.0, 131.1, 128.6,
127.7, 127.6, 127.5, 126.0, 123.4, 120.2, 119.5, 109.1, 101.8, 66.1,
49.5, 48.0, 46.0 ppm; ESI-MS: 337.1 [M+H].
N-(3-(9H-Carbazol-9-yl)-2-hydroxypropyl)-N-(furan-2-ylmethyl)cy-
clopropanesulfonamide (KL021). N-(Furan-2-ylmethyl)-N-(oxiran-2-
ylmethyl)cyclopropanesulfonamide (40 mg, 0.156 mmol), carbazole
(52.1 mg, 0.321 mmol), NaH (8.2 mg, 0.34 mmol) were combined in
DMF (5 mL) to give 41.5 mg (62.8%) of KL021. 1H NMR (CDCl3,
400 MHz): d=8.09 (d, J=7.6 Hz, 2H), 7.46 (t, J=8 Hz, 2H), 7.41 (d,
J=7.6 Hz, 2H), 7.25 (t, J=8 Hz, 2H), 7.17 (dd, J=1.6, 0.8 Hz, 1H),
6.13 (dd, J=3.2, 1.6 Hz, 1H), 5.93 (d, J=3.2 Hz, 1H), 4.40–4.31 (m,
4H), 4.27–4.22 (m, 1H), 3.47 (dd, J=14.8, 8 Hz, 1H), 3.28 (dd, J=
14.8, 3.2 Hz, 1H), 2.28 (m, 1H), 1.63–1.12 (m, 2H), 0.93–0.89 ppm
(m, 2H); 13C NMR (CDCl3, 100 MHz): d=149.2, 142.9, 140.6, 125.9,
123.1, 120.3, 119.3, 110.5, 110.1, 109.0, 69.2, 51.6, 46.8, 45.2, 29.0,
5.2, 5.0 ppm; ESI-MS; 425.1 [M+H], 447.1 [M+Na].
N-(3-(9H-Carbazol-9-yl)-2-hydroxypropyl)-N-(furan-2-ylmethyl)a-
cetamide (KL013). A solution of 1-(9H-carbazol-9-yl)-3-((furan-2-yl-
methyl)amino)propan-2-ol (50 mg, 0.16 mmol) in THF (2 mL) was
treated with DIEA (25 mL, 0.16 mmol) followed by acetic anhydride
(25 mL, 0.24 mmol). After stirring at room temperature overnight,
the reaction mixture was extracted with EtOAc (50 mL) and H2O
(20 mL). The organic layer was dried with MgSO4 and then evapo-
rated under reduced pressure. The crude compound was purified
by preparative HPLC (0.05% TFA/H2O / 0.05% TFA/CH3CN) to give
1
32 mg (56.6%) of KL013; H NMR (CDCl3, 400 MHz): d=8.10 (d, J=
N-(3-(9H-Carbazol-9-yl)-2-hydroxypropyl)-N-(4-fluorophenyl)me-
thanesulfonamide (KL022). N-(4-Fluorophenyl)-N-(oxiran-2-ylme-
thyl)methanesulfonamide (100 mg, 0.41 mmol), carbazole (102 mg,
0.6 mmol), and NaH (19.5 mg, 0.81 mmol) were combined in DMF
7.6 Hz, 2H), 7.45 (t, J=8, 2H), 7.39 (d, J=8, 2H), 7.24 (t, J=7.6 Hz,
2H), 7.07 (dd, J=2, 0.8 Hz, 1H), 5.97 (dd, J=3.2, 1.6 Hz, 1H), 5.55
(dd, J=3.2, 0.4 Hz, 1H), 4.46 (brs, 1H), 4.36–4.21 (m, 2H), 4.17–4.10
(m, 3H), 3.79 (dd, J=14.4, 8.8 Hz, 1H), 3.17 (dd, J=14.4, 2 Hz, 1H),
1
(8 mL) to give 52.1 mg (44.6%) of KL022. H NMR (CDCl3, 400 MHz):
ChemMedChem 2015, 10, 1489 – 1497
www.chemmedchem.org
1495 ꢀ 2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
d=8.07 (d, J=7.6 Hz, 2H), 7.44 (t, J=8.4 Hz, 2H), 7.37–7.31 (m,
4H), 7.24 (t, J=8 Hz, 2H), 7.07 (t, J=8 Hz, 2H), 4.45–4.34 (m, 2H),
4.32–4.25 (m, 1H), 3.92 (dd, J=14.4, 7.6 Hz, 1H), 3.76 (dd, J=14.4,
4.8 Hz, 1H), 2.91 ppm (s, 3H); 13C NMR (CDCl3, 100 MHz): d=140.7,
130.4, 130.3, 126.0, 123.1, 120.4, 119.5, 116.9, 116.6, 108.8, 69.1,
55.2, 37.4 ppm; ESI-MS: 413.1 [M+H], 435.1 [M+Na].
NaH (6 mg, 0.24 mmol, 2 equiv) and stirred at 08C for 30 min, fol-
lowed by 2-chloro-N-(2-chloro-6-cyanophenyl)acetamide (27.4 mg,
0.12 mmol) in DMF (1 mL). The reaction mixture was stirred at
808C for 6 h. After reaction completion, the mixture was purified
by preparative HPLC to give KL044 (22.1 mg, 51.4%). 1H NMR
(CD3CN, 400 MHz): d=8.5 (brs, 1H), 8.16 (d, J=7.6 Hz, 2H), 7.74
(dd, J=8, 1.2 Hz, 1H), 7.67 (dd, J=8, 1.6 Hz, 1H), 7.59 (d, J=8 Hz,
2H), 7.52 (t, J=8 Hz, 2H), 7.41 (t, J=8 Hz, 1H), 7.29 (t, J=7.6 Hz,
2H), 5.25 ppm (s, 2H); 13C NMR (CD3CN, 400 MHz): d=168.3, 144.6,
137.2, 135.2, 133.6, 132.7, 129.8, 126.9, 121.0, 120.6, 116.4, 115.1,
110.0, 47.0 ppm; ESI-MS: 360.0 [M+H].
N-(3-(9H-Carbazol-9-yl)-2-hydroxypropyl)-N-(2-methylbenzyl)me-
thanesulfonamide (KL028). N-(2-Methylbenzyl)-N-(oxiran-2-ylme-
thyl)methanesulfonamide
(120 mg,
0.53 mmol),
carbazole
(133.6 mg, 0.8 mmol), NaH (26.4 mg, 1.1 mmol) were combined in
DMF (8 mL) to give 82 mg (36.4%) of KL028. 1H NMR (CDCl3,
400 MHz): d=8.07 (d, J=8.4 Hz, 2H), 7.47 (td, J=8.4, 1.2 Hz, 2H),
7.24–7.18 (m, 4H), 7.06–6.99 (m, 2H), 6.93 (d, J=7.2 Hz, 1H), 6.87–
6.83 (m, 1H), 4.45 (d, J=13.6 Hz, 1H), 4.19–4.04 (m, 3H), 3.86 (m,
1H), 3.45 (dd, J=15.2, 8.8 Hz, 1H), 3.15 (dd, J=18, 3.2 Hz, 1H), 2.94
(s, 3H), 2.21 ppm (s, 3H); 13C NMR (CDCl3, 100 MHz): d=140.5,
137.2, 132.5, 130.9, 129.5, 128.4, 126.1, 125.8, 123.0, 120.3, 119.3,
108.8, 69.0, 52.1, 51.4, 46.9, 37.1, 19.0 ppm; ESI-MS: 423.1 [M+H],
445.1 [M+Na].
2-(9H-Carbazol-9-yl)-N-(2-cyanophenyl)acetamide (KL046). Same
procedure as described for the synthesis of KL044. 2-Chloro-N-(2-
cyanophenyl)acetamide (20 mg, 0.1 mmol), NaH (10 mg, 0.4 mmol),
and carbazole (16.7 mg, 0.1 mmol) were combined in DMF (3 mL)
to give 22.4 mg (68.9%) of KL046. 1H NMR (CDCl3, 400 MHz): d=
8.21 (d, J=8.4 Hz, 1H), 8.16 (d, J=7.6 Hz, 2H), 7.73 (s, 1H), 7.56–
7.51 (m, 3H), 7.46–7.41 (m, 3H), 7.35 (t, J=8 Hz, 2H), 7.14 (t, J=
7.6 Hz, 1H), 5.13 ppm (s, 2H); 13C NMR (CDCl3, 100 MHz): d=166.9,
140.0, 138.9, 133.9, 132.3, 126.7, 125.8, 123.9, 122.0, 121.0, 120.3,
119.4, 110.6, 108.4, 47.6 ppm; ESI-MS: 326.1 [M+H], 348.1 [M+
Na].
N-(3-(9H-Carbazol-9-yl)-2-hydroxypropyl)-N-(2-(dimethylamino)e-
thyl)methanesulfonamide (KL032). N-(2-(Dimethylamino)ethyl)-N-
(oxiran-2-ylmethyl)methanesulfonamide (103 mg, 0.46 mmol), car-
bazole (153 mg, 0.92 mmol), and 60% NaH (47 mg, 1.15 mmol)
were combined in DMF (7 mL) to give 22 mg (12.3%) of KL032.
1H NMR (CDCl3, 400 MHz): d=8.05 (d, J=7.6 Hz, 2H), 7.50 (d, J=
8.0 Hz, 2H), 7.45 (t, J=8.0 Hz, 2H), 7.22 (t, J=8.0 Hz, 2H), 4.45–4.34
(m, 3H), 3.70–3.65 (m, 1H), 3.53–3.46 (m, 1H), 3.42–3.28 (m, 2H),
3.24–3.11 (m, 2H), 2.78 (s, 3H), 2.74 ppm (m, 6H); ESI-MS: 390.2
[M+H].
2-(9H-Carbazol-9-yl)-N-(2-(trifluoromethyl)phenyl)acetamide
(KL048). Same procedure as described for the synthesis of KL044.
2-Chloro-N-(2-(trifluoromethyl)phenyl)acetamide
(40 mg,
0.17 mmol), 95% NaH (10 mg, 0.4 mmol), and carbazole (28.4 mg,
0.17 mmol) were combined in DMF (4 mL) to give 26.8 mg (42.8%)
1
of KL048. H NMR (CDCl3, 400 MHz): d=8.19 (d, J=8 Hz, 1H), 8.15
(d, J=7.6, 2H), 7.58 (s, 1H), 7.52 (t, J=8, 3H), 7.42 (d, J=8 Hz, 3H),
7.34 (t, J=8 Hz, 2H), 7.17 (m, 1H), 5.11 ppm (s, 2H); 13C NMR
(CDCl3, 100 MHz): d=167.0, 140.0, 132.8, 126.6, 125.0, 124.3, 123.7,
120.8, 120.7, 108.4, 47.6 ppm; ESI-MS: 369.1 [M+H], 391.1 [M+
Na].
N-(2-(N-(3-(9H-Carbazol-9-yl)-2-hydroxypropyl)methylsulfonami-
do)ethyl)acetamide (KL033). N-(2-(N-(Oxiran-2-ylmethyl)methylsul-
fonamido)ethyl)acetamide (80 mg, 0.34 mmol), carbazole (114 mg,
0.68 mmol), and 60% NaH (34 mg, 0.85 mmol) were combined in
DMF (8 mL) to give 24 mg (17.5%) of KL033. 1H NMR (CDCl3,
400 MHz): d=8.06 (d, J=7.6 Hz, 2H), 7.5–7.43 (m, 4H), 7.28–7.20
(m, 2H), 6.33 (s, 1H), 4.48–4.38 (m, 1H), 4.3 (t, J=6 Hz, 2H), 3.45–
3.35 (m, 2H), 3.35–3.20 (m, 3H), 3.18–3.10 (m, 1H), 2.79 (s, 3H),
1.87 ppm (s, 3H); 13C NMR (CDCl3, 100 MHz): d=171.3, 140.8, 126.0,
123.0, 120.3, 119.4, 109.1, 69.6, 54.3, 49.4, 47.3, 38.7, 37.1,
23.1 ppm; ESI-MS: 404.2 [M+H], 426.1 [M+Na].
Acknowledgements
This research was supported by the U.S. National Institutes of
Health (grant R01GM074868 to S.A.K.).
Keywords: 3D-QSAR · circadian clock · cryptochrome · protein
degradation · small molecule
N-(3-(9H-Carbazol-9-yl)-2-hydroxypropyl)-N-(2-methoxyethyl)me-
thanesulfonamide (KL034). N-(2-Methoxyethyl)-N-(oxiran-2-ylme-
thyl)methanesulfonamide (100 mg, 0.48 mmol), carbazole (160 mg,
0.96 mmol), and NaH (43.2 mg, 1.08 mmol) were combined in DMF
[1] a) J. Bass, J. S. Takahashi, Science 2010, 330, 1349–1354; b) G. Asher, U.
Schibler, Cell Metab. 2011, 13, 125–137.
1
(10 mL) to give 82 mg (43.2%) of KL034. H NMR (CDCl3, 400 MHz):
[2] M. Gallego, D. M. Virshup, Nat. Rev. Mol. Cell Biol. 2007, 8, 139–148.
[3] a) E. J. Eide, M. F. Woolf, H. Kang, P. Woolf, W. Hurst, F. Camacho, E. L.
Vielhaber, A. Giovanni, D. M. Virshup, Mol Cell Biol. 2005, 25, 2795–
2807; b) T. Shirogane, J. Jin, X. L. Ang, J. W. Harper, J. Biol. Chem. 2005,
280, 26863–26872; c) K. Vanselow, J. T. Vanselow, P. O. Westermark, S.
Reischl, B. Maier, T. Korte, A. Herrmann, H. Herzel, A. Schlosser, A.
Kramer, Genes Dev. 2006, 20, 2660–2672.
d=8.09 (d, J=7.6 Hz, 2H), 7.49–7.46 (m, 4H), 7.27–7.22 (m, 2H),
4.43–4.32 (m, 3H), 3.83 (d, J=3.6 Hz, 1H), 3.5–3.47 (m, 2H), 3.45–
3.28 (m, 5H), 3.23 (s, 3H), 2.85 ppm (s, 3H); 13C NMR (CDCl3,
100 MHz): d=140.8, 125.9, 123.1, 120.3, 119.3, 109.1, 71.4, 69.8,
58.8, 53.9, 49.5, 37.7 ppm; ESI-MS: 377.1 [M+H], 399.1 [M+Na].
2-(9H-Carbazol-9-yl)-N-(2-chloro-6-cyanophenyl)acetamide
[4] a) L. Busino, F. Bassermann, A. Maiolica, C. Lee, P. M. Nolan, S. I. Godinho,
G. F. Draetta, M. Pagano, Science 2007, 316, 900–904; b) S. I. Godinho,
E. S. Maywood, L. Shaw, V. Tucci, A. R. Barnard, L. Busino, M. Pagano, R.
Kendall, M. M. Quwailid, M. R. Romero, J. O’Neil, J. E. Chesham, D. Brook-
er, Z. Lalanne, M. H. Hastings, P. M. Nolan, Science 2007, 316, 897–900;
c) S. M. Siepka, S. H. Yoo, J. Park, W. Song, V. Kumar, Y. Hu, C. Lee, J. K.
Takahashi, Cell 2007, 129, 1011–1023.
[5] W. Xing, L. Busino, T. R. Hinds, S. M. Marionni, N. H. Saifee, M. F. Bush, M.
Pagano, N. Zheng, Nature 2013, 496, 64–68.
[6] T. Hirota, J. W. Lee, P. C. St. John, M. Sawa, K. Iwaisko, T. Noguchi, P. Y.
Pongsawakul, T. Sonntag, D. Welsh, D. A. Brenner, F. J. Doyle III, P. G.
Schultz, S. A. Kay, Science 2012, 337, 1094–1097.
(KL044).
A solution of 2-amino-3-chlorobenzonitrile (50 mg,
0.33 mmol) in DMF (0.5 mL) was treated with chloroacetyl chloride
(44.8 mg, 0.4 mmol, 1.2 equiv) and TEA (50.5 mL, 0.4 mmol,
1.2 equiv). The reaction mixture was stirred at 408C for 6 h. The re-
action mixture was diluted with CH2Cl2 and washed with H2O. The
organic layer was dried under Na2SO4, filtered, and concentrated
under reduced pressure. The crude compound was purified by
flash column chromatography (EtOAc/hexane 1:3) to give 2-chloro-
N-(2-chloro-6-cyanophenyl)acetamide (54.0 mg, 71.8%). A solution
of carbazole (20 mg, 0.12 mmol) in DMF (1 mL) was treated with
ChemMedChem 2015, 10, 1489 – 1497
www.chemmedchem.org
1496 ꢀ 2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[7] S. Nangle, W. Xing, N. Zheng, Cell Res. 2013, 1417–1419.
[8] T. Cheeseright, M. Mackey, S. Rose, A. Vinter, J. Chem. Inf. Model. 2006,
46, 665–676.
ny, S. A. Kay, Nat. Med. 2010, 16, 1152–1157; b) K. A. Lamia, S. J. Papp,
R. T. Yu, G. D. Barish, N. H. Uhlenhaut, J. W. Jonker, M. Downes, R. M.
Evans, Nature 2011, 480, 552–556.
[9] Forge version 10.3.1, Cresset BioMolecular Discovery Ltd., 2014;
www.cresset-group.com/products/forge/.
[10] SiteMap version 2.9, Schrçdinger LLC, New York, NY (USA), 2013;
www.schrodinger.com/SiteMap.
[13] T. Oshima, I. Yamanaka, A. Kumar, J. Yamaguchi, T. Nishiwaki-Ohkawa, K.
Muto, R. Kawamura, T. Hirota, K. Yagita, S. Irle, S. A. Kay, T. Yoshimura, K.
Itami, Angew. Chem. Int. Ed. 2015, 54, 7193–7197; Angew. Chem. 2015,
127, 7299–7303.
[11] Activity Miner in Forge version 10.3.1, Cresset BioMolecular Discovery
Ltd., 2014; www.cresset-group.com/products/activityminer/.
[12] a) E. E. Zhang, Y. Liu, R. Dentin, P. Y. Pongsawakul, A. C. Liu, T. Hirota,
D. A. Nusinow, X. Sun, S. Landais, Y. Kodama, D. A. Brenner, M. Montmi-
Received: June 12, 2015
Published online on July 14, 2015
ChemMedChem 2015, 10, 1489 – 1497
www.chemmedchem.org
1497 ꢀ 2015 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim