Lysine-specific demethylase 1-selective inactivators: Protein-targeted drug delivery mechanism

Article abstract of DOI:10.1002/anie.201303999

Drug drop off: Given that lysine-specific demethylase 1 (LSD1) could be potently and selectively inactivated by delivering phenylcyclopropylamine (PCPA), a weak and nonselective LSD1 inhibitor, directly to the enzyme's active site, a novel series of LSD1 inactivators (1) were designed. Biological and mechanistic studies indicate that 1 inhibits LSD1 potently and selectively. Copyright

Full text of DOI:10.1002/anie.201303999

.
Angewandte
Communications
DOI: 10.1002/anie.201303999
Drug Design
Lysine-Specific Demethylase 1-Selective Inactivators: Protein-
Targeted Drug Delivery Mechanism**
Daisuke Ogasawara, Yukihiro Itoh, Hiroki Tsumoto, Taeko Kakizawa, Koshiki Mino,
Kiyoshi Fukuhara, Hidehiko Nakagawa, Makoto Hasegawa, Ryuzo Sasaki, Tamio Mizukami,*
Naoki Miyata,* and Takayoshi Suzuki*
Reversible histone methylation, a process controlled by two
counteracting enzyme families, the histone methyltransfer-
ases and the histone demethylases, plays a pivotal role in the
regulation of epigenetic gene expression.[1] Lysine-specific
demethylase 1 (LSD1) removes methyl groups from mono-
and dimethylated Lys4 of histone H3 (H3K4me1/2) through
flavin adenine dinucleotide (FAD) dependent enzymatic
oxidation.[2] LSD1 also demethylates H3K9me1/2 in prostate
cell lines and in cells infected with herpesviruses.[3] Further-
more, histone demethylation by LSD1 is suggested to be
associated with certain disease states, including cancer and
herpes simplex infection.[1a,3b,4]
describe the design and synthesis of a series of LSD1
inactivators based upon this concept.
PCPA inhibits LSD1 by a single-electron-transfer mech-
anism (Figures 1 and 2A).[5a] In the active site of LSD1, FAD
first extracts one electron from the nitrogen atom of PCPA to
form a cation radical. Then, opening of the cyclopropyl ring
occurs and subsequent covalent bond formation with FAD. In
the course of LSD1 inactivation, the nitrogen atom of PCPA
is released as an ammonia molecule through hydrolysis of the
imine intermediate. Taking this mechanism into account,
together with our idea of delivering PCPA directly to the
trans-2-Phenylcyclopropylamine
(PCPA/Tranylcypro-
mine),[5] which was originally found as an inhibitor of
monoamine oxidases (MAOs; also FAD-dependent
enzymes), is the best-studied LSD1 inhibitor, and biological
studies using PCPA have uncovered important roles of LSD1
in several diseases.[3b,4a,c,d] In addition, several groups, includ-
ing ours, have reported PCPA derivatives with LSD1
inhibitory activity.[6] While many of these LSD1 inhibitors
have been suggested to be potential lead compounds for
anticancer agents, most of them have various disadvantages,
including poor intracellular activity, insufficient inhibitory
potency, or inadequate selectivity for LSD1 over MAO A and
MAO B. To overcome these issues, we hypothesized that
LSD1 could be potently and selectively inactivated by
delivering PCPA directly to the LSD1 active site. Herein we
Figure 1. Mechanism of LSD1 inhibition by PCPA.
[*] D. Ogasawara, Dr. Y. Itoh, Prof. T. Suzuki
Graduate School of Medical Science
E-mail: mizukami@nagahama-i-bio.ac.jp
Dr. R. Sasaki
Kyoto Prefectural University of Medicine, 13 Nishitakatsukasa-cho
Taishogun Kita-ku, Kyoto 603-8334 (Japan)
E-mail: suzukit@koto.kpu-m.ac.jp
Frontier Pharma Inc., Shiga 526-0829 (Japan)
Prof. T. Suzuki
PRESTO (Japan) Science and Technology Agency (JST)
Saitama 332-0012 (Japan)
D. Ogasawara, Prof. H. Nakagawa, Prof. N. Miyata
Graduate School of Pharmaceutical Sciences
Nagoya City University
3-1 Tanabe-dori, Mizuho-ku, Nagoya, Aichi 467-8603 (Japan)
E-mail: miyata-n@phar.nagoya-cu.ac.jp
[**] We thank Mie Tsuchida, Dr. Yasunao Hattori, and Prof. Kenichi Akaji
for their technical support. This work was supported in part by JST
PRESTO program (T.S.), a Grant-in-Aid for Scientific Research from
the Japan Society for the Promotion of Science (T.S. and Y.I.), Ho-
ansha Foundation (T.S), Takeda Science Foundation (T.S.), Naito
Foundation (T.S.), and NOVARTIS Foundation (Japan) for the
promotion of science (T.S.).
Dr. H. Tsumoto
World-Leading Drug Discovery Research Center, Kyoto University
Kyoto 606-8501 (Japan)
Dr. T. Kakizawa, Dr. K. Fukuhara
Organic Chemistry National Institute of Health Sciences
Tokyo 158-8501 (Japan)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201303999.
Dr. K. Mino, Dr. M. Hasegawa, Prof. T. Mizukami
Graduate School of Bio-Science
Nagahama Institute of Bio-Science Technology
1226 Tamura-cho, Nagahama, Shiga 526-0829 (Japan)
8620
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 8620 –8624

Angewandte
Chemie
Figure 2. A) Mechanism of LSD1 inhibition by PCPA. B) Putative mechanism of LSD1-targeted delivery of PCPA by 1.
LSD1 active site, we designed LSD1 inactivators 1 (Fig-
ure 2B), in which PCPA is coupled to a lysine carrier moiety
at the nitrogen atom. Since methylated lysine is the substrate
of LSD1, we expected that the lysine moiety of 1 would be
efficiently recognized by LSD1, thus affording high selectivity
over MAO A and MAO B. After the PCPA moiety of 1 is
carried to the active site of LSD1, we expected that the PCPA
moiety would inactivate LSD1 in a similar manner to PCPA
itself, through single-electron transfer, radical opening of the
cyclopropyl ring, and covalent bond formation with FAD
(Figures 1 and 2). Then, the lysine moiety 1ꢀ is expected to be
released through hydrolysis of the imine intermediate (Fig-
ure 2B). Thus, the lysine moiety of 1 serves as a carrier that
delivers PCPA into the active site of LSD1 selectively and
efficiently. This approach can be regarded as an LSD1-
targeted drug delivery system.
As a proof of concept study, we designed PCPA-Lys-4 H3-
21 (1a; Figure 3), which bears a PCPA moiety at Lys-4 of a 21-
amino-acid LSD1 substrate peptide (H3-21), and evaluated
its LSD1-inhibitory activity. We chose the H3-21 peptide as
a carrier of PCPA because it is known to be a substrate of
LSD1,[2a,7] and several H3-21-peptide-based LSD1 inhibitors
were reported.[8] The peptide 1a was synthesized by displace-
ment of the mesyl group of MesylK4 H3-21 (2)[8] with PCPA
(see Scheme S1 in the Supporting Information). In a horse-
radish peroxidase coupled assay,[7] 1a strongly inhibited LSD1
(IC50 = 0.16 mm) in a time- and concentration-dependent
manner, but did not inhibit MAO A or MAO B (IC50 >
Figure 3. Compounds designed, synthesized, and evaluated in this
study.
Angew. Chem. Int. Ed. 2013, 52, 8620 –8624
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
8621

.
Angewandte
Communications
targeted PCPA delivery mechanism (Figure 2B). Next, we
evaluated its activity at the cellular level against cervical
HeLa and neuroblastoma SH-SY5Y cancer cell lines. We
chose these cell lines because LSD1 is overexpressed in both
of them, and LSD1 inhibitors have been reported to suppress
their cell growth.[4a,6c,d] However, the cellular activity of 1a
was weak (Table S1). We surmise that 1a has poor cell-
membrane permeability, likely a result of the high polarity of
its peptide structure. Thus, based on this proof of concept of
our LSD1-targeted PCPA delivery strategy, we next sought to
apply this strategy to design nonpeptide, small-molecule
LSD1 inactivators which would show activity in cell-based
assays.
We designed small-molecule, drug-delivery-type LSD1
inactivator candidates, guided by the X-ray crystal structure
of LSD1.[10] We designed 1b, bearing a benzoyl group (R) and
benzylamino group (R’; Figure 3), because these groups were
expected to fit in the hydrophobic pockets near the entrance
of the LSD1 active-site cavity, and should bear the binding
affinity to LSD1. We also expected that these small hydro-
phobic groups would enhance cell membrane permeability. A
binding simulation of 1b with LSD1 using Molegro Virtual
Docker 5 confirmed that the benzoyl group of 1b is located in
the hydrophobic pocket delineated by Phe 538, Ala 539, and
Phe 692, and the benzylamino group is located in the
hydrophobic pocket delineated by Ile 356, Leu 677, and Cys
360 (see Figure S2A in the Supporting Information). More-
over, the PCPA moiety of compound 1b appears to be located
in almost the same place as the FAD–PCPA adduct in the
reported X-ray crystal structure (see Figure S3 in the
Supporting Information).[5b] This is consistent with the idea
that the PCPA moiety of 1b would be selectively recognized
by LSD1 and oxidized by FAD, which would lead to selective
inactivation of LSD1 (Figure 2B).
We synthesized 1b from commercially available Boc-
Lys(Cbz)-OH (3) as shown in Scheme S2 in the Supporting
Information, and the inhibitory activity of 1b towards LSD1
was evaluated. The compound 1b exerted potent LSD1
inhibitory activity (IC50 = 0.30 mm), being at least 100-fold
more potent than PCPA itself (IC50 = 31 mm, which is com-
parable to that of 1a; Table S2 in the Supporting Information)
without inhibiting MAO A (IC50 > 25 mm). We also evaluated
its cellular activity against HeLa and SH-SY5Y cancer cell
lines. As expected, 1b showed relatively high cell growth
inhibitory activity against these cancer cell lines (GI50 for
HeLa = 35 mm, GI50 for SH-SY5Y= 17 mm; Table S2).
Encouraged by this finding, we next sought to optimize the
lysine moiety of 1b to improve the activity.
Figure 4. Time- and concentration-dependent inhibition of LSD1 by
1a. A) Steady-state progress curves obtained for the inactivation of
~
&
^
*
LSD1 by 0 ( ), 0.025 ( ), 0.05 ( ), 0.1 (ꢀ), 0.2 ( ), 0.4 ( ) mm 1a.
*
B) Rate constants (kobs) for the time-dependent inactivation of LSD1 by
1a were extracted from steady-state progress curve in (A) by single
exponential fitting.
100 mm; see Table S1 in the Supporting Information and
Figure 4). These results indicate that 1a is an irreversible
LSD1 inactivator, as we had hoped.[9] As expected, the
Ki value of 1a was much smaller than that of PCPA, which
indicates that 1a has a greater binding affinity for LSD1 than
does PCPA (Table S1). Next, we performed MALDI MS
analysis of the LSD1/1a mixture. According to the putative
mechanism of LSD1 inhibition by peptide 1a, the FAD–
PCPA adduct and 1’a should be generated (Figure 2B).[5a] As
shown in Figure 5, peaks with m/z 918 and 900, corresponding
Figure 5. Mass spectrometric detection of FAD PCPA adducts. Mass
spectra were obtained from LSD1/1a mixture (negative mode).
We first set out to optimize the benzoyl group of 1b. We
designed and synthesized 1c–j which bear substituents at
either the meta or para position of the benzoyl group of 1b
(Figure 3; Figures S2B and S2C, and Schemes S3 and S4 in the
Supporting Information). As shown in Table S2 in the
Supporting Information, compounds with hydrophobic sub-
stituents at the benzoyl group (1c–h) showed strong LSD1
inhibitory activity as well as improved growth inhibitory
activity towards HeLa and SH-SY5Y cell lines.
to the FAD–PCPA adduct and dehydrated adduct, respec-
tively, were observed in the mixture of LSD1 and 1a. In
contrast, no peak corresponding to the FAD–PCPA adduct
was observed in control experiments without LSD1 (see
Figure S1A in the Supporting Information). Furthermore, the
peak corresponding to the released peptide 1’a (Figure 2B)
was also detected in the same mixture (Figure S1B). These
mechanistic data strongly support the proposed LSD1-
Next, we set out to further optimize the benzylamino
group of 1g, which was the most promising compound among
8622 www.angewandte.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 8620 –8624

Angewandte
Chemie
1c–h in terms of potency in the cancer cell growth inhibition
assays. We designed and synthesized 1k–t with hydrophobic
substituents at either the meta or para position of the
benzylamino group of 1g (Figure 3; Figure S2D and Schem-
es S5 and S6 in the Supporting Information). As shown in
Table S2 in the Supporting Information, 1k–t showed LSD1
inhibitory activity and further improved cancer cell growth
inhibitory activity. Among them, 1s was the most potent (GI50
for HeLa = 3.7 mm, GI50 for SH-SY5Y= 1.7 mm).
To confirm the selectivity of these inactivators, we tested
the inhibitory activity of selected compounds against MAO A
and MAO B. As we expected, while PCPA itself potently
inhibited MAO A and MAO B (IC50 for MAO A = 2.5 mm,
IC50 for MAO B = 2.4 mm), 1g, 1r, 1s, and 1t did not (see
Table S3 in the Supporting Information). These results
suggest that the lysine moiety of these LSD1 inactivators
does not deliver PCPA to the active site of MAO A and
MAO B, but rather delivers PCPA specifically to the LSD1
active site with high efficiency.
Having optimized the structure of the lysine moiety of 1b
as a PCPA carrier targeting LSD1, we next investigated the
LSD1 inactivation mechanism to confirm that the resulting
inactivators indeed inhibit LSD1 by delivering PCPA to the
LSD1 active site, as peptide 1a does. We used 1g and 1s in the
following mechanistic studies because they showed potent
activity in both enzyme assays and cell-based assays. First, we
examined whether 1g and 1s inhibit LSD1 in a time-
dependent manner. As shown in Figure S4 in the Supporting
Information, 1g and 1s were found to be time-dependent
LSD1 inactivators, in accordance with the irreversible mech-
anism we proposed (Figure 2B).[9] The kinetic parameters of
these compounds are shown in Table S4 in the Supporting
Information. The kinact/Ki values of 1g and 1s were much
larger than that of PCPA and comparable to that of 1a, thus
confirming that 1g and 1s, as well as 1a, are much more
potent LSD1 inactivators than PCPA. As in the case of
peptide 1a, the Ki values of 1g and 1s were greatly improved
(more than 120-fold and 48-fold smaller, respectively) com-
pared to that of PCPA. Significantly, the kinact values of 1g and
1s remained almost the same as that of PCPA, and the
kinact value of 1a was five-fold less than that of PCPA
(Tables S1 and S4). These kinetic parameters indicate that
the lysine moiety of 1g and 1s contributes to the enhanced
LSD1 inhibitory activity of these inactivators by increasing
the binding affinity for LSD1 without decreasing the reaction
rate of the PCPA moiety with FAD. We surmise that the
conformational flexibility of these small-molecule LSD1
inactivators allowed their PCPA moiety to react with FAD
in a similar manner to PCPA itself. Next, we performed
MALDI MS analysis of the inactivated mixture of LSD1 with
1g and 1s. It is expected that inhibition of LSD1 by these
inactivators will yield the FAD–PCPA adduct (Figure 2B),[5a]
regardless of the structure of the lysine moiety. As shown in
Figure S5 in the Supporting Information, peaks with m/z 918
and 900, corresponding to the FAD-PCPA adduct and the
dehydrated adduct, respectively, were observed in the cases of
both LSD1/1g and LSD1/1s. We also successfully detected 1’g
and 1’s, the lysine moieties released from LSD1/1g and LSD1/
1s, respectively (Figure S6 in the Supporting Information).
These mechanistic data strongly support the idea that these
small-molecule LSD1 inactivators inhibit LSD1 through
efficient and selective delivery of PCPA to the active site of
LSD1 with the assistance of their lysine moiety.
In conclusion, we have designed and synthesized a series
of novel LSD1 inactivators based on our new concept of
LSD1-targeted PCPA delivery. We tested this concept with
peptide 1a and confirmed that 1a does indeed inhibit LSD1
selectively and efficiently by delivering PCPA directly to the
LSD1 active site. The enzyme is then inactivated by FAD-
PCPA adduct formation, as in the case of PCPA itself, with
release of the carrier peptide 1’a. We also demonstrated that
this strategy could be applied to the design of nonpeptide,
small-molecule LSD1 inactivators. Mechanistic studies con-
firmed that the small-molecule inactivators also inactivate
LSD1 by LSD1-targeted PCPA delivery, similar to peptide
1a. Biological evaluations revealed that these small-molecule
compounds are potent LSD1-selective inactivators (compa-
rable to peptide 1a) in enzyme assays, and they also exhibit
potent cell growth inhibitory activities against cancer cell
lines. We believe this is the first time that a drug-carrier-type
system has been employed to target a specific molecule. These
novel LSD1-selective inactivators are considered to be
candidates for anticancer agents, as well as bioprobes.
Received: May 10, 2013
Published online: July 3, 2013
Keywords: biological activity · drug delivery · drug design ·
.
kinetics · proteins
[1] a) Y. Shi, Nat. Rev. Genet. 2007, 8, 829 – 833; b) A. J. Bannister,
T. Kouzarides, Nature 2005, 436, 1103 – 1106; c) S. Kubicek, T.
Jenuwein, Cell 2004, 119, 903 – 906.
[2] a) Y. Shi, F. Lan, C. Matson, P. Mulligan, J. R. Whetstine, P. A.
Cole, R. A. Casero, Y. Shi, Cell 2004, 119, 941 – 953; b) J. C.
Culhane, P. A. Cole, Curr. Opin. Chem. Biol. 2007, 11, 561 – 568.
[3] a) E. Metzger, M. Wissmann, N. Yin, J. M. Muller, R. Schneider,
A. H. Peters, T. Gunther, R. Buettner, R. Schule, Nature 2005,
437, 436 – 439; b) Y. Liang, J. L. Vogel, A. Narayanan, H. Peng,
T. M. Kristie, Nat. Med. 2009, 15, 1312 – 1317.
[4] a) J. H. Schulte, S. Lim, A. Schramm, N. Friedrichs, J. Koster, R.
Versteeg, I. Ora, K. Pajtler, L. Klein-Hitpass, S. Kuhfittig-Kulle,
E. Metzger, R. Schule, A. Eggert, R. Buettner, J. Kirfel, Cancer
Res. 2009, 69, 2065 – 2071; b) Y. Huang, T. M. Stewart, Y. Wu,
S. B. Baylin, L. J. Marton, B. Perkins, R. J. Jones, P. M. Woster,
R. A. Casero, Jr., Clin. Cancer Res. 2009, 15, 7217 – 7228; c) S.
Lim, A. Janzer, A. Becker, A. Zimmer, R. Schꢁle, R. Buettner, J.
Kirfel, Carcinogenesis 2010, 31, 512 – 520; d) T. Schenk, W. C.
Chen, S. Gçllner, L. Howell, L. Jin, K. Hebestreit, H.-U. Klein,
A. C. Popescu, A. Burnett, K. Mills, R. A. Casero, Jr., L.
Marton, P. Woster, M. D. Minden, M. Dugas, J. C. Y. Wang,
J. E. Dick, C. Mꢁller-Tidow, K. Petrie, A. Zelent, Nat. Med. 2012,
18, 605 – 611; e) L. Shi, S. Cui, J. D. Engel, O. Tanabe, Nat. Med.
2013, 19, 291 – 294.
[5] a) D. M. Schmidt, D. G. McCafferty, Biochemistry 2007, 46,
4408 – 4416; b) M. Yang, J. C. Culhane, L. M. Szewczuk, P. Jalili,
H. L. Ball, M. Machius, P. A. Cole, H. Yu, Biochemistry 2007, 46,
8058 – 8065; c) S. Mimasu, T. Sengoku, S. Fukuzawa, T. Ume-
hara, S. Yokoyama, Biochem. Biophys. Res. Commun. 2008, 366,
15 – 22.
Angew. Chem. Int. Ed. 2013, 52, 8620 –8624
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
8623

.
Angewandte
Communications
[6] a) D. M. Gooden, D. M. Schmidt, J. A. Pollock, A. M. Kabadi,
D. G. McCafferty, Bioorg. Med. Chem. Lett. 2008, 18, 3047 –
3051; b) H. Benelkebir, C. Hodgkinson, P. J. Duriez, A. L.
Hayden, R. A. Bulleid, S. J. Crabb, G. Packhamc, A. Ganesan,
Bioorg. Med. Chem. 2011, 19, 3709 – 3716; c) R. Ueda, T. Suzuki,
K. Mino, H. Tsumoto, H. Nakagawa, M. Hasegawa, R. Sasaki, T.
Mizukami, N. Miyata, J. Am. Chem. Soc. 2009, 131, 17536 –
17537; d) D. Ogasawara, T. Suzuki, K. Mino, R. Ueda, M. N.
Khan, T. Matsubara, K. Koseki, M. Hasegawa, R. Sasaki, H.
Nakagawa, T. Mizukami, N. Miyata, Bioorg. Med. Chem. 2011,
19, 3702 – 3708; e) S. Mimasu, N. Umezawa, S. Sato, T. Higuchi,
T. Umehara, S. Yokoyama, Biochemistry 2010, 49, 6494 – 6503;
f) C. Binda, S. Valente, M. Romanenghi, S. Pilotto, R. Cirilli, A.
Karytinos, G. Ciossani, O. A. Botrugno, F. Forneris, M. Tar-
dugno, D. E. Edmondson, S. Minucci, A. Mattevi, A. Mai, J. Am.
Chem. Soc. 2010, 132, 6827 – 6833; g) T. Suzuki, N. Miyata, J.
Med. Chem. 2011, 54, 8236 – 8250.
[7] F. Forneris, C. Binda, M. A. Vanoni, E. Battaqlioli, A. Mattevi, J.
Biol. Chem. 2005, 280, 41360 – 41365.
[8] a) J. C. Culhane, L. M. Szewczuk, X. Liu, G. Da, R. Marmor-
stein, P. A. Cole, J. Am. Chem. Soc. 2006, 128, 4536 – 4537;
b) J. C. Culhane, D. Wang, P. M. Yen, P. A. Cole, J. Am. Chem.
Soc. 2010, 132, 3164 – 3176.
[9] R. A. Copeland, Enzymes: A Practical Introduction to Structure,
Mechanism, and Data Analysis, 2nd ed., Wiley-VCH, New York,
2000.
[10] M. Yang, J. C. Culhane, L. M. Szewczuk, C. B. Gocke, C. A.
Brautigam, D. R. Tomchick, M. Machius, P. A. Cole, H. Yu, Nat.
Struct. Mol. Biol. 2007, 14, 535 – 539.
8624 www.angewandte.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 8620 –8624