Angewandte
Chemie
The hydroxy substructure turned out to be a significant, but
non-critical site as ketone (GO152) and alkoxy analogues
sustained the period-lengthening activity, albeit with less
potency. The furan part was found not to be a critical site as
thiophene analogue GO061 and a phenyl-substituted ana-
logue also exhibited period-changing effects. Fortunately, as
expected, it was found that additional substituents on the
C5 position led to an increase in period-changing activities.
For example, the C5-bromo-substituted derivative GO214
induced a higher period-lengthening effect at the same
concentration. C5-aryl-substituted derivative GO216 had an
activity similar to that of KL001. Interestingly, we discovered
that some aryl substituents at the C4 position could induce the
opposite effect (period shortening), albeit with low efficiency.
For example, when a 4-n-butylphenyl group was attached at
the C4 position (GO044), the period was shortened by
0.4 hours. Given the opposite activity of C5 isomer GO216,
the rotation around the thiophene ring may be rather
restricted at the binding site. We also found that the
substituent effect at the C4 position was rather elusive.
Whereas the introduction of some para-substituted aryl
groups (GO200 and GO211) also induced a period-shortening
effect, a small para substituent, such as the methoxy group
(GO058), led to a period-lengthening effect.
With these exciting preliminary results in hand, we
examined the effects of the following representative com-
pounds in further detail: one period-lengthening compound,
GO214, and three period-shortening compounds, GO044,
GO200, and GO211. We first studied the effects of these four
compounds at three different concentrations (Figure 3a,b).
At 10 mm, the period-lengthening effect of GO214 was
significantly greater than that of KL001 (Supporting Infor-
mation, Figure S1). On the other hand, GO044, GO200, and
GO211 shortened the period by 0.3 to 1 hour at 10 mm
(Figure 3a,b). These effects were also confirmed in the
mPer2 promoter-driven luciferase assay (Figure S2).
We next added period-lengthening and -shortening com-
pounds simultaneously to a cell culture to examine whether
these compounds compete with each other. When we added
a 1 mm solution of KL001 together with a 1 mm solution of
either GO044, GO200, or GO211, the period-lengthening
effect of KL001 was attenuated, and the resulting period was
comparable to that observed in vehicle-treated cells (Fig-
ure 3c and Figure S3a). The period-lengthening effect of
GO214 was also cancelled out by the addition of GO200 or
GO211 (Figure S3b). These results suggest that the period-
shortening compounds may compete with KL001 to bind to
the same FAD-binding pocket.
Therefore, to elucidate the binding mode of the deriva-
tives to CRY in more detail, we performed docking simu-
lations for the representative compounds (period-lengthening
GO214 and period-shortening GO044, GO200, and G0211)
with the CRY protein structure (PDB X-ray structure
4MLP).[10] Technical details of these simulations are described
in the Supporting Information. The optimal docking pose of
the ligands was obtained by performing a conformational
search allowing for torsional flexibility. Using this method,
the binding pose of the crystal structure for KL001 was
successfully reproduced. The best docking pose and the most
Figure 3. Discovery of period-lengthening and -shortening molecules.
a) Representative luminescence traces and b) the changes in the
circadian period in Bmal1-dLuc U2OS cell lines in the presence of each
compound. Measurements were done in triplicate or quadruplicate.
Data are presented as meanÆSD (* P<0.05, ** P<0.01; one-way
ANOVA vs. vehicle). c) Competition assay between KL001 and period-
shortening compounds. Bmal1-dLuc U2OS cells were simultaneously
treated with KL001 (1 mm) and either GO044 or GO200 (1 mm).
Measurements were done in triplicate or quadruplicate. Data are
presented as meanÆSD. Different characters (a–d) indicate significant
differences (P<0.05; one-way ANOVA). For each compound, we
performed three independent experiments, and representative results
are shown.
important amino acid residue interactions of GO214, GO044,
GO200, and G0211 are shown as snapshots of the binding
pocket in Figures 4a, 4b, S4a, and S4b, respectively.
The binding free energies of KL001, GO214, GO044,
GO200, and GO211 were determined to be À10.41, À11.39,
À13.58, À8.44, and À12.27 kcalmolÀ1, respectively, indicating
that all four compounds bind to the FAD-binding pocket of
CRY. In a similar manner to KL001, the hydroxy moiety and
the sulfonyl moiety in GO214 and GO044 form a hydrogen
bond with S414 and H377, respectively. Furthermore, the
mesyl moiety forms a CH–p interaction[19] with W417 and
W310 (within < 4 ꢁ), which is similar to the interaction of
P426 in the FBXL3–CRY complex (PDB: 4I6J).[20] In the case
of GO200, two hydrogen bonds were observed, one between
the mesyl group and H377 and the second one between the
hydroxy moiety and H373 instead of S414, owing to a torsional
change to fit into the active site. The superimposition of the
Angew. Chem. Int. Ed. 2015, 54, 1 – 6
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3
These are not the final page numbers!