Discovery of macrocyclic peptides armed with a mechanism-based warhead: Isoform-selective inhibition of human deacetylase SIRT2

Article abstract of DOI:10.1002/anie.201108118

Designed to inhibit: By using the random nonstandard peptide integrated discovery (RaPID) system, highly potent isoform-selective inhibitors can be identified from a library of nonstandard macrocyclic peptides (see picture). These inhibitors, which contain a mechanism-based warhead residue, are active against the human deacetylase SIRT2, with IC₅₀ values in the low nanomolar region. Copyright

Full text of DOI:10.1002/anie.201108118

Angewandte
Chemie
DOI: 10.1002/anie.201108118
Inhibitors
Discovery of Macrocyclic Peptides Armed with a Mechanism-Based
Warhead: Isoform-Selective Inhibition of Human Deacetylase
SIRT2**
Jumpei Morimoto, Yuuki Hayashi, and Hiroaki Suga*
Mass screening of available compound libraries by means of
in vitro or cell-based assays is the most prevalent approach for
the discovery of enzyme inhibitors. The major drawback of
this approach is generally a poor cost–performance ratio,
often giving many false-positive hits and therefore requiring
extensive reevaluations of the hits to identify true inhibitors.
Enzyme-mechanism-based drug design is an alternative
approach,[1–3] in which an inhibitory functional group is
ingeniously designed on the basis of knowledge of mechanism
and embedded into an appropriate scaffold, such as a native
enzyme substrate. Even though this approach represents
a strategy with a high cost–performance ratio to obtain an
initial hit(s), the primary design of such an inhibitor is rarely
potent enough, particularly against an enzyme family that
catalyzes various substrates or consists of multiple isoforms.
In any case, in order to improve the potency of the initial
hit(s) further elaboration of its structure is generally con-
ducted by classical medicinal chemistry.
One example of mechanism-based designed inhibitors
against yeast sirtuins is derived from a substrate peptide, H2N-
KSTGG-KAc-APRKQ-OH, in which the e-N-acetyl group on
the lysine residue (KAc) was replaced with e-N-trifluoroacetyl
group (KTfa).[4] Sirtuin belongs to a family of deacetylases,
homologs of which are widespread in prokaryotes and
eukaryotes, and deacetylates the e-N-acetyl group on KAc in
various cellular substrates in the presence of nicotinamide
adenine dinucleotide (NAD+) as a cosubstrate. Mechanisti-
cally, the carbonyl oxygen of the e-N-acetyl group first attacks
the C1’ of the N-ribose in NAD+ to form an O-alkylamidate
intermediate upon the release of a nicotinamide group, and
following several reaction steps give the free e-amino-K
residue and 2’-O-acetyl-ADP-ribose (Scheme 1a).[5,6] The
replacement of KAc with KTfa in the peptide marginally
increases its affinity to yeast sirtuins, but reduces the rate of
the formation of the O-alkylamidate intermediate by nearly
six orders of magnitude, thus resulting in its inhibitory
behavior (Scheme 1b).[4,7,8] Various other mechanism-based
inhibitors against sirtuins, such as e-N-thioacetyl-K (KTac
)
containing peptides,[4,9] have been reported, however, none of
them showed spectacular inhibitory activities; for example,
IC50 values of KTfa and KTac peptides were 61 mm and 1 mm,
respectively.
Because of a biological importance of mammalian sirtuins
(SIRTs) that are involved in the regulation of diverse cellular
functions, it is of great interest to devise more potent
inhibitors against SIRTs than the present examples. In this
study, we chose human SIRT2, which is known to deacetylate
KAc residues in a-tubulin[10] and histone H4K16,[11] thereby
regulate cell-cycle progression. Since there are three isoforms
(SIRT1–3) of class I SIRTs[12] and each SIRT plays distinct
biological roles, we were interested in devising not only
a SIRT-selective but also an isoform-selective inhibitor as an
epigenetic probe and a potential therapeutic agent. We herein
report a new methodology to construct a library of non-
standard macrocyclic peptides that contain a KTfa residue in
the middle of the sequence by using a custom-made trans-
lation apparatus that enables genetic code reprogramming,
and rapidly select potent inhibitors by using an in vitro display
format. The nonstandard macrocyclic peptides that were
identified exhibit a high inhibitory potency against SIRT2
with Kd (and IC50) values in the low (single digit) nanomolar
region as well as a remarkable isoform selectivity.
Genetic code reprogramming is a new means to riboso-
mally express various nonstandard peptides with desired
structural diversities, that is executed by the reassignment of
codons from proteinogenic amino acids to nonproteinogenic
or artificial amino acids.[13–15] To facilitate genetic code
reprogramming, we have devised a custom-made cell-free
translation system integrated with the flexizyme (flexible
tRNA acylation ribozyme) technology, referred to as FIT
(flexible in vitro translation) system.[16] In the FIT system,
arbitrary proteinogenic amino acids are omitted from the
translation components to create vacant codons, and tRNAs
that read the vacant codons are charged with nonproteino-
genic amino acids by flexizyme; thus nonproteinogenic amino
[*] J. Morimoto, Dr. Y. Hayashi, Prof. H. Suga
Department of Chemistry, School of Science
The University of Tokyo
7-3-1 Hongo, Bunkyo, Tokyo, 113-0033 (Japan)
E-mail: hsuga@chem.s.u-tokyo.ac.jp
J. Morimoto
Department of Chemistry and Biotechnology
School of Engineering, The University of Tokyo
7-3-1 Hongo, Bunkyo, Tokyo, 113-8656 (Japan)
[**] We thank Drs. H. Murakami, T. Katoh, and Y. Goto for discussions
on the methodology of the RaPID system, and Drs. M. Yoshida and
A. Ito for discussions on in vitro sirtuin assays. We also thank Dr.
Goshima for a gift of a human genome entry clone used for the
SIRT2 expression. This work was supported by a Grants-in-Aid of the
Japan Society for Promotion of Science (JSPS), the Specially
Promoted Research (21000005), the Industrial Science and Tech-
nology Program in the New Energy and Industrial Technology
Development Organization (NEDO) to H.S., and a Grants-in-Aid for
JSPS Fellows (21-9079) to J.M.
Supporting information for this article (including experimental
details) is available on the WWW under http://dx.doi.org/10.1002/
anie.201108118.
Angew. Chem. Int. Ed. 2012, 51, 3423 –3427
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3423

.
Angewandte
Communications
factor 1 (RF1), but is supple-
mented with two aminoacyl-
tRNAs, in which ClAclY or
ClAcdY and KTfa were charged
onto initiator tRNAfMet
and
CAU
orthogonal elongator tRNAAsn-
E2
CAU
[25]
,
respectively, by means of
enhanced flexizyme (eFx).[15]
The mRNA sequence library
was designed to have a mixture of
lengths consisting of AUG-
(NNC)m-AUG-(NNC)n-UGC, in
which the combination of [m,n]
is [3,4], [4,4], [4,5], [5,5], and
[5,6], and UGC assigned cysteine
(C) residue (see Figure 1a, and
Table S1 in the Supporting Infor-
mation). In this library design, five
amino acids (M, Q, E, K, and W)
would not appear in the random
NNC region, and, more impor-
tantly, the AUG codon would also
not appear in it; thus, ClAclY/
ClAcdY and KTfa are ensured to
Scheme 1. Schematic illustration of a) deacetylation of acetylated lysine (KAc) on peptide catalyzed by
sirtuin, and b) mechanism-based inhibition of deacetylation by a KTfa-containing peptide. a) The
carbonyl oxygen atom of KAc attacks NAD+ to form an O-alkylamidate intermediate, followed by the
breakdown of the intermediate to give the corresponding deacetylated peptide. b) Substitution of KAc
with KTfa leads to an unproductive intermediate, which significantly slows the initial attack of the
oxygen atom of Tfa, thus resulting in inhibition of the activity of sirtuin.
acids are introduced to the translation system to express
peptides that contain these nonproteinogenic amino acids.
More recently, the FIT system has
been coupled with an mRNA dis-
play method,[17,18] devising a new
platform technology referred to as
RaPID (random nonstandard pep-
tide integrated discovery) system.[19]
In comparison with other in vitro or
in vivo display methodologies,[20–24]
the RaPID system allows us to
select active nonstandard peptides
against desired targets from their
highly diverse libraries that consist
of multiple unique nonproteino-
genic amino acids in the peptide
chain. Herein we have utilized this
RaPID system to construct such
a library of peptides that contain
KTfa residues as a mechanism-based
warhead, and aimed at discovering
anti-SIRT2 inhibitors.
appear only in the designated positions. RNA sequences for
the ribosome binding site and (GS)3 linker peptide were
In this study, we reprogrammed
the AUG codon to two distinct
artificial amino acids, one of which
was a-N-(2-chloroacetyl)-l-tyrosine
Figure 1. Schematic illustration of the construction of a library of macrocyclic peptides that contain
a KTfa residue by the FIT system and selection of the nonstandard peptides. a) An mRNA sequence
library containing two AUG codons. The initiation AUG codon is suppressed by ClAcl(d)Y-tRNAfMet
or a-N-(2-chloroacetyl)-d-tyrosine
(ClAclY or ClAcdY) assigned to
the initiator AUG codon, and the
other of which was KTfa assigned to
the elongator AUG codon. These
double assignments of amino acids
to a single AUG codon could be
achieved by using an FIT system
CAU
and the elongation AUG codon is suppressed by KTfa-tRNAAsn-E2CAU. b) The translated peptide library
contains ClAcl(d)Y at the N terminus and KTfa at the middle of the random sequences. The peptide
and its encoding mRNA are connected by puromycin, thus resulting in the peptide-displayed
mRNAs. c) The ClAc group and cysteine residue spontaneously react with each other after
translation to give a macrocyclic peptide that contains a single KTfa residue. d) SIRT2-binding
peptides are selected from the library on the basis of SIRT2-binding activity. Selected mRNA
sequences are amplified by reverse transcription and PCR (RT-PCR), and the amplified DNAs are
which lacks methionine and release transcribed to give an enriched mRNA library with the desired SIRT2-binding property.
3424
www.angewandte.org ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2012, 51, 3423 –3427

Angewandte
Chemie
embedded upstream and downstream of the mRNA library
sequences, respectively, and this mRNA library was termi-
nated with a UAG codon followed by a linker RNA region for
the installation of a puromycin-CC-(poly(ethylene glycol)
linker)-DNA fragment. Because RF1 was absent in the FIT
system, the desired peptide–mRNA fusion with puromycin
was effectively formed at the UAG ribosome-stalling position
in generally greater than 10% yield, thus giving a complexity
of the peptide library of 1012 or higher after all processes of
the nonstandard-peptide library synthesis (Figure 1b and
Table S1).
conditions depending on sequence compositions, they all
exhibited inhibitory activities to some extent (Figure S4), thus
validating the effectiveness of our designed-library strategy.
In order to quantitatively determine how strongly selected
macrocyclic peptides could bind to SIRT2, we chemically
synthesized two representative peptides with a C-terminal
carboxyamide modification, S2Li8 and S2iD7, both of which
shared the R(I/V)-KTfa-RY motif, albeit originating from the
different library sets, and analyzed their binding kinetics
against SIRT2 by using a surface plasmon resonance (SPR)
method. Interestingly, both peptides behaved nearly the same
kinetically, and gave nearly identical Kd values of 3.8 and
3.7 nm, respectively (Table 1, S2iL8 and S2iD7; Figure 2a and
b; Figure S5a and b).
Translation of the mRNA library was initiated with
ClAclY and ClAcdY to give two independent libraries,
l
d
referred to as Y and Y libraries. The C residue(s) installed
at the UGC codon after the
Table 1: Binding and inhibitory properties of peptides studied.[a]
random region or UGC appear in
the random region would sponta-
neously react with the N-terminal
chloroacetyl group to form various
macrocyclic structures closed by
a thioether bond (Figure 1c).[26]
The displayed peptide libraries
were then subjected to selection
against His-tagged SIRT2 immobi-
lized on magnetic beads on the
basis of strong affinity to SIRT2
(Figure 1d; see Figure S1 for more
details of the RaPID selection
procedures), resulting in saturated
enrichment of active populations at
the fifth and sixth rounds in lY and
dY libraries, respectively (Fig-
ure S2).
Peptide
Sequence
ka
kd
Kd
IC50 [nm]
[ꢀ106 msÀ1
]
[ꢀ10À3 sÀ1
]
[nm] SIRT2 SIRT1 SIRT3
S2iL8
S2iD7
1.2
1.3
4.7
4.8
3.8 3.2
3.7 3.7
47
32
480
240
lin-S2iL8
lin-S2iD7
0.98
1.6
13
17
13
10
82
6.1
5.5
31
n.d.
n.d.
280
n.d.
n.d.
RIKTfaRY
SDKTfaTI
0.82
n.d.
67
1000
n.d.
n.d.
n.d. >104 n.d.
[a] The values of ka, kd, and Kd were determined by SPR, and IC50 values were determined by fluorescence-
based in vitro assays. n.d.=not determined.
The enriched cDNA pools from
lY and dY libraries were cloned, and the peptide-encoding
sequences of 21 and 16 clones, respectively, were analyzed
(Table S2). Interestingly, the lY library gave 14-mers consist-
Encouraged by the strong affinities observed in S2iL8 and
S2iD7, we measured their inhibitory activities by titrating the
SIRT2-catalyzing deacetylation rate of KAc in the aforemen-
tioned fluorescence-based assay. The IC50 values in both
peptides are again very similar (3.2 and 3.7 nm, respectively)
and consistent with the Kd values (Table 1, S2iL8 and S2iD7;
Figure 2c and d). To evaluate their isoform selectivity against
class I SIRTs, the same titration experiments were performed
against SIRT1 and SIRT3 with their corresponding peptide
substrates (RHKKAc and QPKKAc, respectively). Both mac-
rocyclic peptides possess similar isoform selectivity with
approximately 10-fold and 100-fold elevated IC50 values
against SIRT1 and SIRT3, respectively. However, S2iL8
seems to be more isoform selective over S2iD7; other residues
outside of the R(I/V)-KTfa-RY motif possibly contribute to the
observed selectivity (Table 1, S2iL8 and S2iD7; Figure 2c and
d).
The macrocyclic form of the peptides might preorganize
the structures and might thus contribute to the increase in
their target-binding affinities.[19,27,28] Therefore, to evaluate
such an effect on the selected peptides, we synthesized linear
counterparts of S2iL8 and S2iD7, lin-S2iL8 and lin-S2iD7,
respectively (Table 1), in which the N-terminal chloroacetyl
group is substituted with an acetyl group, and the C-terminal
cysteine residue is alkylated with iodoacetamide (CR). SPR
d
ing of Ac-lY-(X)5-KTfa-(X)6-C, while Y library gave 13- and
14-mers consisting of Ac-dY-(X)5-KTfa-(X)5,6-C. Despite the
fact that we did not observe convergent groups of sequences
among these clones, the careful comparison of apparent
peptide sequences showed nearly conserved residues (in
a rate of over 70% in all clones) adjacent to the warhead KTfa
,
consisting of (I/V)-KTfa-RY in both libraries. Furthermore,
among the clones that have the (I/V)-KTfa-RY sequence, the
R(I/V)-KTfa-RY sequence appeared most frequently (10
clones out of 15), thus leading to the hypothesis that this
sequence region might interact with the SIRT2 active site.
In order to see if selected peptides were able to inhibit
SIRT2 enzymatic activity, every clone was expressed by the
FIT system and the inhibitory activity of the individual
peptide was assayed by a conventional fluorescence-based
method by using
a
known SIRT2-substrate peptide
(QPKKAc). MALDI–TOF analysis of the individual peptides
clearly showed that all possessed the designated KTfa residue
and macrocyclic structure (see Figure S3 for mass spectra of
four representative clones and Table S3 for observed mass
values of other clones). Although the expression level of all
the peptides could be varied in 10–50 nm level under the assay
Angew. Chem. Int. Ed. 2012, 51, 3423 –3427
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3425

.
Angewandte
Communications
based on the concept of a mechanism-
based warhead, and utilized the
RaPID system to enrich “binding-
active” sequences that possess KTfa
as a warhead against SIRT2. The
selected peptides are not just binders
but also potent inhibitors of SIRT2
and gave consistent Kd and IC50 values
of approximately 3–4 nm. Remarka-
bly, these peptides are SIRT2-selec-
tive inhibitors that also discriminate
against other isoforms. Thus, this
proof-of-concept experiment vali-
dates that introduction of a mecha-
nism-based warhead into random
sequences of peptides facilitates
inhibitor discovery. The methodology
developed herein is rationally appli-
cable to the development of inhibitors
against various other posttransla-
Figure 2. Binding and inhibitory properties of macrocyclic peptides. SPR sensorgrams of a) S2iL8
and b) S2iD7 against SIRT2. The peptide was used in five different concentrations (1, 2, 4, 8, and
16 nm) with a SIRT2-immobilized sensor chip. The binding curve was analyzed by the single-cycle
kinetics method. The observed data were superimposed to theoretical fitting curves based on a 1:1
binding model (see Figure S5 to distinguish the observed data in black and the theoretical fitting
tional-modification
enzymes
by
recruiting appropriate mechanism-
based warheads, for example,
hydroxamic acid against deacety-
lases,[29–31] and propargylamine against
demethylases.[32]
*
curve in red). Inhibitory potencies of c) S2iL8 and d) S2iD7 against SIRT2 ( ) and its isoforms
&
^
SIRT1 ( ) and SIRT3 ( ). The inhibitory activities were measured by means of in vitro
fluorescence-based assays (see details in material methods section in the Supporting Information)
and plotted as the fractional activities of SIRT2 or its isoforms at varying inhibitor concentrations.
The standard deviation of each data point was generated by four independent experiments, and
the data were fitted to a sigmoidal curve.
Received: November 18, 2011
Revised: January 31, 2012
Published online: February 28, 2012
analysis showed that both linear peptides have slightly higher
dissociation rates (kd) from SIRT2 than cyclic peptides, thus
reflecting the elevation of Kd values by approximately three
times. Inhibitory activities against SIRT2 were also consistent
with the observed Kd values, giving slightly higher IC50 values
than those of macrocyclic ones (Table 1 and Figure S5c–f).
Keywords: FIT system · inhibitors ·
macrocyclic peptides · peptides ·
RaPID system
.
[1] V. Sandanayaka, A. Prashad, Curr. Med. Chem. 2002, 9, 1145 –
1165.
[2] J. Zhong, W. Groutas, Curr. Top. Med. Chem. 2004, 4, 1203 –
1216.
[3] P. Cole, Nat. Chem. Biol. 2008, 4, 590 – 597.
[4] B. C. Smith, J. M. Denu, Biochemistry 2007, 46, 14478 – 14486.
[5] A. A. Sauve, V. L. Schramm, Curr. Med. Chem. 2004, 11, 807 –
826.
[6] A. Sauve, C. Wolberger, V. Schramm, J. Boeke, Annu. Rev.
Biochem. 2006, 75, 435 – 465.
[7] B. C. Smith, J. M. Denu, J. Am. Chem. Soc. 2007, 129, 5802 –
5803.
[8] B. C. Smith, J. M. Denu, J. Biol. Chem. 2007, 282, 37256 – 37265.
[9] D. G. Fatkins, A. D. Monnot, W. Zheng, Bioorg. Med. Chem.
Lett. 2006, 16, 3651 – 3656.
[10] B. J. North, B. L. Marshall, M. T. Borra, J. M. Denu, E. Verdin,
Mol. Cell 2003, 11, 437 – 444.
[11] A. Vaquero, M. B. Scher, D. H. Lee, A. Sutton, H. L. Cheng,
F. W. Alt, L. Serrano, R. Sternglanz, D. Reinberg, Genes Dev.
2006, 20, 1256 – 1261.
The data suggested that in these particular peptides the
macrocyclic structure indeed increased the affinity to SIRT2,
but the degree of increased affinity turned out to be rather
less significant. As aforementioned, the comparison of clone
sequences has suggested that R(I/V)KTfaRY sequence might
be the critical motif responsible for binding activity. To this
end, a short peptide, RIKTfaRY bearing an N-terminal acetyl
and a C-terminal amide group, was synthesized and its
binding and inhibitory activities against SIRT2 were deter-
mined. Similar to the linear peptides, the dissociation rate of
RIKTfaRY was further elevated, but the Kd value remained
below 100 nm (Figure S6a). Most importantly, the IC50 values
of RIKTfaRY against SIRT2, SIRT1, and SIRT3 were 31 nm,
280 nm, and 1000 nm, respectively (Table 1, RIKTfaRY and
Figure S6b), thus exhibiting the selective inhibitory potency
against SIRT2 over other isoforms, such as the full-length
macrocyclic peptides. In sharp contrast, a 5-mer peptide
derived from an a-tubulin substrate motif (Table 1, SDKTfaTI
and Figure S6c) had an IC50 value that was nearly three orders
of magnitude higher, thus confirming that the selected
sequence motif is critical for the observed potency.
[12] R. A. Frye, Biochem. Biophys. Res. Commun. 2000, 273, 793 –
798.
[13] A. C. Forster, Z. Tan, M. N. Nalam, H. Lin, H. Qu, V. W.
Cornish, S. C. Blacklow, Proc. Natl. Acad. Sci. USA 2003, 100,
6353 – 6357.
In conclusion, we have ribosomally expressed a highly
diverse library of thioether-containing macrocyclic peptides
[14] K. Josephson, M. Hartman, J. Szostak, J. Am. Chem. Soc. 2005,
127, 11727 – 11735.
3426 www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 3423 –3427

Angewandte
Chemie
[15] H. Murakami, A. Ohta, H. Ashigai, H. Suga, Nat. Methods 2006,
3, 357 – 359.
[16] Y. Goto, T. Katoh, H. Suga, Nat. Protoc. 2011, 6, 779 – 790.
[17] R. W. Roberts, J. W. Szostak, Proc. Natl. Acad. Sci. USA 1997,
94, 12297 – 12302.
[24] T. Young, D. Young, I. Ahmad, J. Louis, S. Benkovic, P. Schultz,
Proc. Natl. Acad. Sci. USA 2011, 108, 11052 – 11056.
[25] A. Ohta, H. Murakami, E. Higashimura, H. Suga, Chem. Biol.
2007, 14, 1315 – 1322.
[26] Y. Goto, A. Ohta, Y. Sako, Y. Yamagishi, H. Murakami, H. Suga,
ACS Chem. Biol. 2008, 3, 120 – 129.
[27] L. Giebel, R. Cass, D. Milligan, D. Young, R. Arze, C. Johnson,
Biochemistry 1995, 34, 15430 – 15435.
[28] E. M. Driggers, S. P. Hale, J. Lee, N. K. Terrett, Nat. Rev. Drug
Discovery 2008, 7, 608 – 624.
[29] P. Marks, R. Rifkind, V. Richon, R. Breslow, T. Miller, W. Kelly,
Nat. Rev. Cancer 2001, 1, 194 – 202.
[30] J. Bolden, M. Peart, R. Johnstone, Nat. Rev. Drug Discovery
2006, 5, 769 – 784.
[31] S. Minucci, P. Pelicci, Nat. Rev. Cancer 2006, 6, 38 – 51.
[32] J. Culhane, L. Szewczuk, X. Liu, G. Da, R. Marmorstein, P. Cole,
J. Am. Chem. Soc. 2006, 128, 4536 – 4537.
[18] N. Nemoto, E. Miyamoto-Sato, Y. Husimi, H. Yanagawa, FEBS
Lett. 1997, 414, 405 – 408.
[19] Y. Yamagishi, I. Shoji, S. Miyagawa, T. Kawakami, T. Katoh, Y.
Goto, H. Suga, Chem. Biol. 2011, 18, 1562 – 1570.
[20] S. Li, R. W. Roberts, Chem. Biol. 2003, 10, 233 – 239.
[21] S. W. Millward, S. Fiacco, R. J. Austin, R. W. Roberts, ACS
Chem. Biol. 2007, 2, 625 – 634.
[22] C. Heinis, T. Rutherford, S. Freund, G. Winter, Nat. Chem. Biol.
2009, 5, 502 – 507.
[23] C. C. Liu, A. V. Mack, E. M. Brustad, J. H. Mills, D. Groff, V. V.
Smider, P. G. Schultz, J. Am. Chem. Soc. 2009, 131, 9616 – 9617.
Angew. Chem. Int. Ed. 2012, 51, 3423 –3427
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3427