Amino acids for diels-alder reactions in living cells

Article abstract of DOI:10.1002/anie.201108231

Under tension: A set of genetically encoded unnatural amino acids can be used for biocompatible site-specific labeling of proteins with fluorogenic dyes. The new compounds have norbornene and trans-cyclooctene units that react with tetrazine derivatives i

Full text of DOI:10.1002/anie.201108231

.
Angewandte
Communications
DOI: 10.1002/anie.201108231
Protein Labeling
Amino Acids for Diels–Alder Reactions in Living Cells**
´
Tilman Plass, Sigrid Milles, Christine Koehler, Je˛drzej Szymanski, Rainer Mueller,
Manfred Wießler, Carsten Schultz,* and Edward A. Lemke*
The endeavour to perform tailored chemical reactions in the
challenging environment of the intact cell delves deeply into
the biological sciences. Requirements include strict bio-
orthogonality of the reactants and reactions that occur
spontaneously and quickly in an aqueous environment or at
the interface of membranes. Commonly used reactions that
meet these criteria are Staudinger ligations and various forms
of click chemistry.[1] The most prominent among the latter is
the Huisgen-type [3+2] cycloaddition between azides and
alkynes.[1,2] Through the seminal work of the Bertozzi group,
this reaction was stripped of its need for CuI catalysis by
straining the alkyne group, thereby making this chemistry
(termed strain-promoted alkyne–azide chemistry, SPAAC)
viable in intact cells as well as in living animals.[3] These
reactions have been widely used to label molecules on cell
surfaces and, in a few cases, inside the cell, for instance to
label lipids,[4] nucleotides,[5] or carbohydrates.[6] Another
exciting click variant is strain-promoted inverse-electron-
demand Diels–Alder cycloaddition (SPIEDAC), which can
exhibit accelerated reaction rates by using strained reactants
and furthermore is irreversible because of the loss of N2
(Scheme 1).[7] This chemistry has been used in cells to label
small molecules and is magnitudes faster than the classical
Huisgen-type cycloadditions.[8]
Scheme 1. a) Structures of strained alkene and alkyne UAAs. b) Reac-
tion scheme showing orthogonality and cross-reactivity of SPIEDAC
and SPAAC with fluorogenic tetrazine-functionalized dyes (gray sphere)
and azide-functionalized dyes (green sphere). Dyes coupled to tetra-
zine are only fluorescent (green) after successful labeling.
To date, most biological applications of SPAAC or
SPIEDAC do not involve modifications of proteins but
instead alter cellular molecules that are not genetically
encoded, such as metabolically incorporated sugars.[6] Current
tools for site-specific labeling of proteins within the cell use
fluorescent protein fusions,[9] self-alkylating protein addi-
tions,[10] or high-affinity binding domains.[11] The smallest size
of artificial protein modifications currently available to
introduce fluorescent labels are tetracysteine motifs consist-
ing of six amino acids.[12] Ideally the modification unit would
be only a single artificial amino acid suitable for specific
chemistry in cells. The introduction of such unnatural amino
acids (UAAs) is possible by codon reassignment[13] or by
suppression of the Amber stop codon.[14] For fluorescent
labeling, genetically encoded azides can be used, but azides
typically suffer from intracellular reduction.[15] Furthermore,
encoding the azide jeopardizes the design of a fluorogenic
labeling scheme.[13a,16] Fluorogenicity is of particular rele-
vance for high-contrast imaging and super-resolution tech-
niques, since dyes are turned on only after successful labeling,
while nonspecifically attached dyes remain quenched. Rather
than encoding azides, a more suitable approach is the use of
an amino acid that carries the strained reactant, that is,
a cyclooctyne group, thereby leaving the nitrogen-bearing
reactants to serve as part of a fluorogenic probe. If suppres-
sion of the Amber stop codon is used, a single residue in
a specified protein can then be replaced with the strained
alkyne. This type of protein labeling by using an artificially
introduced cyclooctyne amino acid and fluorogenic azides
´
[*] T. Plass, S. Milles, C. Koehler, Dr. J. Szymanski, Dr. R. Mueller,
Priv.-Doz. Dr. C. Schultz, Dr. E. A. Lemke
Structural and Computational Biology Unit and
Cell Biology and Biophysics Unit, EMBL
Meyerhofstrasse 1, 69117 Heidelberg (Germany)
E-mail: schultz@embl.de
lemke@embl.de
Prof. Dr. M. Wießler
Biological Chemistry, DKFZ
Im Neuenheimer Feld 280, 69120 Heidelberg (Germany)
[**] We thank P. G. Schultz for a mammalian RS expression plasmid, V.
VanDelinder, and A. Ori for helpful discussions. This study was
technically supported by the EMBL-ALMF, -PEPCF, and -PCF core
facilities. T.P. is a VCI, S.M. a BIF, and J.S. an EMBO fellow. E.A.L.
acknowledges funding by the Emmy Noether program and C.S.
from Transregio83 of the DFG.
Supporting information (including detailed experimental methods
and compound synthesis) for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201108231.
4166
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 4166 –4170

Angewandte
Chemie
was recently achieved in our labs.[17] Herein we expand strain-
promoted click reactions to the direct labeling of proteins by
using Diels–Alder reactions between strained dienophiles
and tetrazines serving as dienes (Scheme 1). Fluorophores
can be made intrinsically fluorogenic by coupling them to the
strongly quenching tetrazine.[8b] This quenching is not easily
addressed, when the tetrazines are genetically encoded.[18] We
synthesized a set of three new dienophilic amino acids and
incorporated them into proteins in E. coli and in mammalian
cells through suppression of the Amber stop codon.
Recently it was found that the trans-cyclooctene deriva-
tives have much faster reaction kinetics for SPIEDAC than
norbornene derivatives.[8a,b] Fast but highly specific reaction
kinetics are desirable for labeling studies in vivo. Owing to the
close similarity to 4, we thus aimed to genetically encode the
trans-cyclooctene UAA 3. The exploration of the promiscuity
of double-mutant synthetase pair RSAF gave typical expres-
sion yields of 4 mg GFPTAG !3 from 1 L E. coli culture.
Compound 3 can be synthesized in five steps starting from
commercially available 9-oxabicyclo[6.1.0]non-4-ene with an
overall yield of 38% (average yield per step above 83%). One
key step in the preparation of gram quantities of 3 is the
coupling of the trans-cyclooctenyl-4-nitrophenyl carbonate to
fluorenylmethyloxycarbonyl (Fmoc) lysine in dimethyl sulf-
oxide and N,N-diisopropylethylamine, because other stan-
dard conditions that were used to generate a carbamate bond
were inefficient in yielding pure Fmoc-protected 3. Subse-
quent deprotection was carried out using 20% piperidine in
N,N-dimethylformamide. Attempts to synthesize compound 3
by using the well-established Boc synthetic strategy resulted
in poorer yields because of less-efficient formation of the
carbamate linkage and the lack of suitable Boc-removal
conditions that neither catalyze double-bond isomerization
nor cleave the sidechain carbamate.
Summarizing our expression results, we find that UAAs 1,
2, and 3 are efficiently incorporated by the double-mutant
synthetase pair RSAF, while 1 is also accepted with reduced
efficiency by the wildtype synthethase pair RSWT (Figure 1;
for masspectrometric data see Table S1 in the Supporting
Information). We next aimed to further explore the potential
of UAAs 1, 2, and 3 for in vivo labeling, also in comparison to
the reactivity of 4. E. coli cultures expressing GFPTAG !UAA
were supplemented with 50 mm dye, washed, and then
analyzed using SDS PAGE (masspectrometric conformation
given in Table S1 in the Supporting Information). Figure S1 in
the Supporting Information shows that in line with Scheme 1,
UAAs 1–4 all reacted with tetrazines but only 4 also with
azides. To further test if SPIEDAC and SPAAC reactions
were orthogonal, we grew two cultures expressing MBPTAG !3
(maltose-binding protein) and MBPTAG!4 separately and then
mixed them prior to labeling with tetramethylrhodamine
(TAMRA) azide, which is predicted not to react with
dienophiles (see Scheme 1). Finally the mixed culture was
reacted with coumarin tetrazine. Figure 2a shows selectively
labeled green- and red-fluorescent E. coli (see Figures S1 and
S2 in the Supporting Information for additional controls).
To quantitatively analyze the different reactions, E. coli
lysate was incubated with TAMRA tetrazine or TAMRA
azide. The reaction was followed by using a fluorescence
spectrometer by exciting GFP (Figure 2). Initially, only green
fluorescence was observed. Upon reaction with the TAMRA
compounds, orange fluorescence appeared owing to Foerster
resonance energy transfer (FRET) from GFP to TAMRA in
agreement with the close proximity after covalent coupling of
TAMRA to GFP.[17] Figure 2c summarizes the data from all
experiments, and Figure S3 and Table S2 in the Supporting
Information list the observed reaction kinetics. For the
SPIEDAC reaction of UAA 3 with TAMRA tetrazine we
found a rate constant of k = (35000 ꢀ 3000)mꢁ1 sꢁ1, which is
In our previous work, we genetically encoded UAA 4 by
means of an engineered tRNA/synthetase pair from Meth-
anosarcina mazei that naturally encodes pyrrolysine.[17,19]
While the wildtype synthetase pair (tRNAPyl/PylRSWT, here
termed RSWT) was not efficient in recognizing the unnatural
amino acid 4 as a substrate, a rationally designed double
mutant (Y306A, Y384F; here termed tRNAPyl/PylRSAF
,
RSAF) efficiently incorporated 4 in response to the Amber
codon.[17] We now aimed to explore the possible promiscuity
of double mutant RSAF to other unnatural amino acids with
similarly bulky side chains. To objectively evaluate our new
compounds, we also included unnatural amino acid 4 as
a standard for which we reported an expression yield of about
10 mg GFPTAG!4 per 1 L of E. coli culture (GFP = green
fluorescent protein).[17] We first synthesized the norbornene
derivatives, compounds 1 and 2, because they are known
dienophiles for SPIEDAC.[7d,20] Unnatural amino acids 1 and
2 were prepared in two steps starting from the commercially
available alcohol via a chloroformate intermediate followed
by coupling to tert-butoxycarbonyl (Boc)-protected lysine and
subsequent deprotection using 60% formic acid, which
prevents cleavage of the sidechain carbamate. Both com-
pounds were obtained in an overall yield of more than 91%.
We then tested the Amber suppression efficiency of the RSWT
and RSAF systems by co-expressing a GFP reporter construct
with an Amber TAG mutation (GFPTAG). GFP fluorescence
occurred only in response to efficient Amber suppression.
Full-length GFP protein was purified using a C-terminal
purification handle. Figure 1 shows that norbonene UAAs
1 and 2 are efficiently recognized by the RSAF mutant (typical
GFP yields from a 1 L E. coli culture are 10 mg for 1 and 2;
Table S1 in the Supporting Information lists the confirming
results from masspectrometric determination).
Figure 1. Fluorescence images of microcentrifuge tubes with E. coli
suspensions expressing GFPTAG!UAA in the absence (ꢁ) and presence
(+) of UAAs 1–4 with the corresponding Coomassie-stained SDS
PAGE gel after purification of GFPTAG!UAA for RSWT (a) and RSAF (b).
Angew. Chem. Int. Ed. 2012, 51, 4166 –4170
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
4167

.
Angewandte
Communications
mCherry-GFPTAG fusion protein. In this construct, mCherry
expression was visible by orange fluorescence in the nucleus,
thereby signifying successful transfection of the plasmid,
while green fluorescence reported on successful expression of
GFP owing to Amber suppression. By using an automated
microscope procedure (see Figures S4–S9 in the Supporting
Information), we analyzed the GFP expression depending on
the concentration of unnatural amino acids 1–4 for RSWT and
RSAF. Figure 3 summarizes the data, showing for all com-
Figure 2. a) Separate E. coli cultures expressing MBPTAG!3 and
MBPTAG!4 were mixed in a 1:1 ratio and reacted first with TAMRA
azide (red) and then with coumarin tetrazine (green). The overlay of
the two color channels plus the differential interference contrast (DIC)
channel visualizes differential labeling, and the inlay shows that 50%
of cells were labeled red or green. Scale bar 10 mm. b) GFPTAG!3 was
incubated with TAMRA tetrazine, and the increase of TAMRA fluores-
cence that is due to FRET (i.e. due to labeling) was followed over time
(dark to light blue). c) Summary for reaction of GFPTAG !UAA: red =1,
*
*
*
*
orange =2, cyan =3, blue =4 with TAMRA tetrazine and blue
*
=4 with TAMRA azide.
Figure 3. a) UAA-concentration-dependent expression of an NLS-
mCherry-GFPTAG construct in HeLa cells expressing either RSWT or RSAF.
b) Corresponding representative images (for 2) in the green, red, and
DIC channel. Scale bar 50 mm.
orders of magnitude faster than the next fastest reaction of 4
with TAMRA tetrazine (k = (400 ꢀ 200)mꢁ1 sꢁ1). It should
thus be mentioned that while the reactivity of strained alkynes
towards the tetrazine is tuneable by tetrazine substituents,[21]
this reactivity is generally much higher than that of the
norbornene-bearing amino acids, of which UAA 1 has
recently (note added in proof) also been encoded by the
Chin lab.[22]
The ultimate goal is that bioorthogonal chemical methods
can be used to site-specifically label proteins with small
photostable fluorophores in mammalian cells, thereby bypass-
ing the need for bulky fluorescent proteins that typically
exhibit less-favorable photophysical properties with respect
to photostability and quantum yield. To achieve this goal, one
important milestone is to genetically encode unnatural amino
acids for the bioorthogonal reactions SPAAC or SPIEDAC in
mammalian cells, while another goal is to make small,
photostable dyes coupled to tetrazine or azide that are
membrane-permeant and to wash out unreacted dye from
intact cells. To test if we can achieve the first milestone, we
transferred tRNAPyl/PylRS pairs (RSWT and RSAF) to a mam-
malian expression plasmid, following a published proce-
dure.[23] The tRNAPyl/PylRS pair was co-transfected with an
expression plasmid of a nuclear localization signal (NLS)-
pounds a clear dependence of the GFP fluorescence on the
UAA concentration; as expected, the fluorescence intensity is
low and variable at concentrations below 100 mm. GFP
expression typically showed an optimum around 250 mm
UAA and then frequently decreased again, which is probably
due to mild toxicity of either too efficient Amber suppression
or the amino acids themselves. Our systematic concentration
screen revealed a way to optimize Amber suppression in
mammalian cells, and we are now investigating how Amber
suppression in mammalian culture can be further improved.
Notably, the selectivity that we observed in E. coli is
qualitatively conserved in mammalian cells.
Having now a system at hand that allows for site-specific
encoding of UAAs for SPIEDAC and SPAAC into mamma-
lian cells, we aimed to test if these reactions can be used for
labeling. We expressed an NLS-MBP-GFPTAG construct with
UAAs 1–4 and then labeled the culture for only five minutes
with 100 nm red fluorescent Cy5 tetrazine. Figure 4 shows that
only cells expressing NLS-MBP-GFPTAG!3 show GFP fluo-
4168
www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 4166 –4170

Angewandte
Chemie
[3] a) S. T. Laughlin, J. M. Baskin, S. L. Amacher, C. R. Bertozzi,
Science 2008, 320, 664; b) P. Kele, X. Li, M. Link, K. Nagy, A.
Herner, K. Lorincz, S. Beni, O. S. Wolfbeis, Org. Biomol. Chem.
2009, 7, 3486.
[4] A. B. Neef, C. Schultz, Angew. Chem. 2009, 121, 1526; Angew.
Chem. Int. Ed. 2009, 48, 1498.
[5] a) A. Salic, T. J. Mitchison, Proc. Natl. Acad. Sci. USA 2008, 105,
2415; b) P. M. Gramlich, C. T. Wirges, A. Manetto, T. Carell,
Angew. Chem. 2008, 120, 8478; Angew. Chem. Int. Ed. 2008, 47,
8350.
[6] P. V. Chang, J. A. Prescher, E. M. Sletten, J. M. Baskin, I. A.
Miller, N. J. Agard, A. Lo, C. R. Bertozzi, Proc. Natl. Acad. Sci.
USA 2010, 107, 1821.
[7] a) M. L. Blackman, M. Royzen, J. M. Fox, J. Am. Chem. Soc.
2008, 130, 13518; b) M. Wiessler, W. Waldeck, C. Kliem, R.
Pipkorn, K. Braun, Int. J. Med. Sci. 2009, 7, 19; c) J. Schoch, M.
Wiessler, A. Jaschke, J. Am. Chem. Soc. 2010, 132, 8846; d) J.
Sauer, A. Mielert, D. Lang, D. Peter, Chem. Ber. 1965, 98, 1435.
[8] a) N. K. Devaraj, R. Upadhyay, J. B. Haun, S. A. Hilderbrand, R.
Weissleder, Angew. Chem. 2009, 121, 7147; Angew. Chem. Int.
Ed. 2009, 48, 7013; b) N. K. Devaraj, S. Hilderbrand, R.
Upadhyay, R. Mazitschek, R. Weissleder, Angew. Chem. 2010,
122, 2931; Angew. Chem. Int. Ed. 2010, 49, 2869; c) R. Rossin,
P. R. Verkerk, S. M. van den Bosch, R. C. Vulders, I. Verel, J.
Lub, M. S. Robillard, Angew. Chem. 2010, 122, 3447; Angew.
Chem. Int. Ed. 2010, 49, 3375.
[9] B. N. Giepmans, S. R. Adams, M. H. Ellisman, R. Y. Tsien,
Science 2006, 312, 217.
[10] a) A. Keppler, H. Pick, C. Arrivoli, H. Vogel, K. Johnsson, Proc.
Natl. Acad. Sci. USA 2004, 101, 9955; b) S. Mizukami, S.
Watanabe, Y. Hori, K. Kikuchi, J. Am. Chem. Soc. 2009, 131,
5016.
[11] R. Wombacher, M. Heidbreder, S. van de Linde, M. P. Sheetz,
M. Heilemann, V. W. Cornish, M. Sauer, Nat. Methods 2010, 7,
717.
[12] a) B. A. Griffin, S. R. Adams, J. Jones, R. Y. Tsien, Methods
Enzymol. 2000, 327, 565; b) A. Rutkowska, C. H. Haering, C.
Schultz, Angew. Chem. 2011, 123, 12867; Angew. Chem. Int. Ed.
2011, 50, 12655.
[13] a) K. E. Beatty, F. Xie, Q. Wang, D. A. Tirrell, J. Am. Chem. Soc.
2005, 127, 14150; b) J. K. Montclare, D. A. Tirrell, Angew. Chem.
2006, 118, 4630; Angew. Chem. Int. Ed. 2006, 45, 4518.
[14] a) A. Deiters, P. G. Schultz, Bioorg. Med. Chem. Lett. 2005, 15,
1521; b) C. C. Liu, P. G. Schultz, Annu. Rev. Biochem. 2010, 79,
413.
[15] a) J. V. Staros, H. Bayley, D. N. Standring, J. R. Knowles,
Biochem. Biophys. Res. Commun. 1978, 80, 568; b) S. Milles, S.
Tyagi, N. Banterle, C. Koehler, V. Vandelinder, T. Plass, A. P.
Neal, E. A. Lemke, J. Am. Chem. Soc. 2012, 134, 5187.
[16] a) Z. Hao, Y. Song, S. Lin, M. Yang, Y. Liang, J. Wang, P. R.
Chen, Chem. Commun. 2011, 47, 4502; b) D. P. Nguyen, H.
Lusic, H. Neumann, P. B. Kapadnis, A. Deiters, J. W. Chin, J.
Am. Chem. Soc. 2009, 131, 8720; c) J. W. Chin, S. W. Santoro,
A. B. Martin, D. S. King, L. Wang, P. G. Schultz, J. Am. Chem.
Soc. 2002, 124, 9026; d) T. Fekner, X. Li, M. M. Lee, M. K. Chan,
Angew. Chem. 2009, 121, 1661; Angew. Chem. Int. Ed. 2009, 48,
1633; e) T. Yanagisawa, R. Ishii, R. Fukunaga, T. Kobayashi, K.
Sakamoto, S. Yokoyama, Chem. Biol. 2008, 15, 1187; f) J. T. Ngo,
J. A. Champion, A. Mahdavi, I. C. Tanrikulu, K. E. Beatty, R. E.
Connor, T. H. Yoo, D. C. Dieterich, E. M. Schuman, D. A.
Tirrell, Nat. Chem. Biol. 2009, 5, 715.
Figure 4. HeLa cells expressing NLS-MBP-GFPTAG!3 (upper row) and
HeLa cells expressing NLS-MBP-GFPTAG!4 (lower row) were incubated
for five minutes with Cy5 tetrazine (100 nm). The corresponding
images of nuclear fluorescence from Hoechst staining in blue and GFP
in green are shown. Only when UAA 3 is used, also red colocalizing
Cy5 fluorescence is observed. Scale bar 20 mm.
rescence colocalized with the red fluorescence of Cy5 (see
Figure S10 in the Supporting Information for all other UAAs
and details on the labeling procedure). This experiment
demonstrates the advantage of using much faster reactions for
cell labeling despite the lower expression yield of UAAs 3
versus UAAs 1, 2, and 4. This remarkable result lays a strong
foundation for future use of our now faster and more diverse
set of bioorthogonal chemical reactions to label and manip-
ulate proteins with single-residue precision in mammalian cell
cultures.
In summary, we have genetically encoded UAAs for
a biocompatible chemistry in living cells that is orthogonal to
the previously described cyclooctyne–azide click chemistry.
The basic utility of this new method for labeling of proteins
in vivo was demonstrated in E. coli. A particular focus of
future work will be the development of more hydrophilic
UAAs that facilitate easy washout of unincorporated UAA
and thus high-contrast imaging. However, we were already
able to show the high potential of genetically encoding 3 for
fluorescence imaging of specifically labeled proteins in
mammalian cell culture. This approach will pave the way
for labeling single protein sites with small fluorophores and
other analytical or functional labels in physiologically rele-
vant cells in the future.
Received: November 22, 2011
Revised: February 17, 2012
Published online: March 30, 2012
Keywords: amber suppression · click chemistry · Diels–
.
Alder reactions · protein engineering · unnatural amino acids
[1] C. P. Hackenberger, D. Schwarzer, Angew. Chem. 2008, 120,
10182; Angew. Chem. Int. Ed. 2008, 47, 10030.
[17] T. Plass, S. Milles, C. Koehler, C. Schultz, E. A. Lemke, Angew.
Chem. 2011, 123, 3964; Angew. Chem. Int. Ed. 2011, 50, 3878.
[18] J. L. Seitchik, J. C. Peeler, M. T. Taylor, M. L. Blackman, T. W.
Rhoads, R. B. Cooley, C. Refakis, J. M. Fox, R. A. Mehl, J. Am.
Chem. Soc. 2012, 134, 2898.
[2] a) H. C. Kolb, M. G. Finn, K. B. Sharpless, Angew. Chem. 2001,
113, 2056; Angew. Chem. Int. Ed. 2001, 40, 2004; b) J. C.
van Hest, F. L. van Delft, ChemBioChem 2011, 12, 1309;
c) C. W. Tornøe, C. Christensen, M. Meldal, J. Org. Chem.
2002, 67, 3057.
Angew. Chem. Int. Ed. 2012, 51, 4166 –4170
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 4169

.
Angewandte
Communications
[19] H. Neumann, S. Y. Peak-Chew, J. W. Chin, Nat. Chem. Biol.
2008, 4, 232.
[20] N. K. Devaraj, R. Weissleder, S. A. Hilderbrand, Bioconjugate
Chem. 2008, 19, 2297.
[21] M. R. Karver, R. Weissleder, S. A. Hilderbrand, Angew. Chem.
2012, 124, 944; Angew. Chem. Int. Ed. 2012, 51, 920.
[22] K. Lang, L. Davis, J. Torres-Kolbus, C. Chou, A. Deiters, J. W.
Chin, Nat. Chem. 2012, DOI:10.1038/NCHEM.1250.
[23] P. R. Chen, D. Groff, J. Guo, W. Ou, S. Cellitti, B. H. Geier-
stanger, P. G. Schultz, Angew. Chem. 2009, 121, 4112; Angew.
Chem. Int. Ed. 2009, 48, 4052.
4170
www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 4166 –4170