Total synthesis of the bacterial RNA polymerase inhibitor ripostatin B

Article abstract of DOI:10.1002/anie.201200871

A modular and highly stereoselective synthesis of the title compound was developed. Key steps in the assembly of the carbon framework of ripostatin B (1; see scheme) were a stereoselective Paterson aldol reaction and a high-yielding ring-closing metathesis mediated by Grubbs first generation catalyst. The C15 hydroxy group was established through Tishchenko-Evans reduction in excellent yield and selectivity. Copyright

Full text of DOI:10.1002/anie.201200871

Angewandte
Chemie
DOI: 10.1002/anie.201200871
Ripostatin B (3)
Total Synthesis of the Bacterial RNA Polymerase Inhibitor
Ripostatin B**
Florian Glaus and Karl-Heinz Altmann*
The resistance of bacterial pathogens to antibiotic treatment
has emerged as one of the central threats to public health in
the 21st century.[1] At the same time, the surge of bacterial
drug resistance has been paralleled by a decline in antibacte-
rial drug discovery research in most of the pharmaceutical
industry,[1,2] a decline which has led to a general current
paucity of agents (or the complete absence thereof) with
satisfactory activity against major drug-resistant bacteria.[1,3]
Thus, there is an urgent need for the development of new
types of antibiotics, including the continuous chemical and
biological assessment of new structural leads with promising
antibacterial activity.
Bacterial RNA polymerase (RNAP) is the target of the
rifamycin class of antibiotics, with rifamycin derivatives such
as rifampicin, rifabutin, and rifapentin being essential drugs
for the treatment of tuberculosis (TB).[3] RNAP is highly
conserved across gram-positive and gram-negative bacteria,
but differs significantly from mammalian RNA polymerases I,
II, and III, thus providing the basis for the selective inhibition
of bacterial growth and survival.[4] Thus, RNAP is a highly
relevant, if underexploited,[5,6] target for antibacterial drug
discovery, especially in light of the recent discovery of a new
mode of RNAP inhibition by a number of natural pro-
ducts,[4a,7] including the myxobacterial macrolide riposta-
tin A.[8,9] The binding of these compounds to the “switch
region” of RNAP stabilizes a closed conformation of the
enzyme, thus preventing access of the dsDNA substrate to the
active-site cleft. This inhibition mode is fundamentally differ-
ent from that of rifamycin-type inhibitors;[4a,7] as a result,
switch-region binders exhibit very limited cross-resistance
with rifampicin.[7,10]
Ripostatins A and B are 14-membered macrolides that
were first isolated in 1995 from the myxobacterium Soran-
gium cellulosum (strain So ce 377) at the Helmholtz Centre
for Infection Research (formerly the GBF).[8] Both com-
pounds show a narrow spectrum of antibiotic activity with
MICꢀs of less than 1 mgmLꢀ1 only against S. aureus and the
hyperpermeable E. coli strain DH21tolC; ripostatin A inhib-
its RNAP with submicromolar activity with no effect on
eukaryotic RNA polymerase II.[8] No explicit data for the
inhibition of RNAP by ripostatin B have been reported, but
based on its structural relationship with ripostatin A and its
similar antibiotic spectrum, an identical mode of action can be
inferred for both compounds.
[*] MSc F. Glaus, Prof. Dr. K.-H. Altmann
Natural ripostatins are not suitable drug candidates
themselves, owing to both their limited antibacterial activity
and their unfavorable biopharmaceutical properties, but these
issues may be addressed by appropriate structural modifica-
tions.[5] To provide the chemical basis for structure–activity
relationship (SAR) studies with ripostatins and concurrent
lead optimization efforts, we embarked on the stereoselective
total synthesis of ripostatin B. We used a strategy that was
built around the construction of the macrolide ring by ring-
closing olefin metathesis (RCM). Herein we report on the
total synthesis of ripostatin B (1) through the successful
implementation of this approach. No total synthesis of either
ripostatin A or B had been reported in the literature prior to
our own work and the concurrent independent work by
Christmann and co-workers, and Prusov and Tang, which is
also described in this issue of Angewandte Chemie.[11]
Swiss Federal Institute of Technology (ETH) Zꢀrich
Department of Chemistry and Applied Biosciences
Institute of Pharmaceutical Sciences
HCI H405, Wolfgang-Pauli-Strasse 10, 8093 Zꢀrich (Switzerland)
E-mail: karl-heinz.altmann@pharma.ethz.ch
Homepage: http://www.pharma.ethz.ch/institute_groups/institu-
te_groups/pharmaceutical_biology/
[**] This work was supported by the European Community Framework
Programme 7, More Medicines For Tuberculosis (MM4TB), grant
agreement no. 260872. We are indebted to Dr. Bernhard Pfeiffer
(ETH Zꢀrich) for NMR spectroscopy support and to Louis Bertschi
from the ETHZ LOC MS Service for HRMS spectra. We are
particularly grateful to Dr. Evgeny Prusov, Helmholtz Centre for
Infection Research, Braunschweig (Germany), for sharing with us
spectroscopic data for ripostatin B prior to publication and to Dr.
Evgeny Prusov and Prof. Mathias Christmann, Technical University
Dortmund, for agreeing to delay publication of their accepted
manuscripts on the total synthesis of ripostatins until our own work,
as described in this paper, had been reviewed.
Our complete retrosynthesis of ripostatin B (1) is sum-
marized in Scheme 1. Among the various possible target sites
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201200871.
for RCM-based ring closure,[12] the C5 C6 double bond was
ꢀ
Angew. Chem. Int. Ed. 2012, 51, 3405 –3409
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3405

.
Angewandte
Communications
Scheme 2. a) Mg, THF, reflux, 15 min, then RT, 1.5 h, then PhCH2CHO,
08C, 1 h; b) CH3C(OEt)3, 1458C, 3 h, 49% (two steps); c) N,O-
dimethylhydroxylamine·HCl, CH3MgBr, THF, ꢀ58C, 1.5 h, then RT,
14 h, 78%.
corresponding Grignard reagent; reaction of the latter with
phenyl acetaldehyde gave a secondary alcohol that was
directly transformed into ester 2 by an ortho-ester Claisen
rearrangement in 49% overall yield (based on 2-bromopro-
pene).[16] Treatment of 2 with CH3MgBr and N,O-dimethyl-
hydroxylamine hydrochloride, in accordance with the proce-
dure of Williams et al.,[17] directly furnished the desired
ketone 3 in 78% yield (via the corresponding Weinreb
amide).
The elaboration of epoxide 4 (obtained from d-aspartic
acid in 3 steps and 60% overall yield)[14] into aldehyde 9 is
outlined in Scheme 3 and was initiated by epoxide opening
with ethyl propiolate; subsequent protection of the resulting
Scheme 1. Retrosynthesis of ripostatin B (1). PMB=para-methoxy-
benzyl, TBS=tert-butyldimethylsilyl.
selected for disconnection, as this provides the only diene
precursor with two unencumbered terminal double bonds
(which, therefore, was assumed to be the most reactive
variant). This diene was envisioned to be accessed from vinyl
=
iodide 14 by Stille coupling with Bu3SnCH2CH CH2; after
RCM, the primary hydroxy group would be deprotected
selectively and then converted into the C28 carboxylate prior
to final deprotection of the secondary hydroxy groups at C11
and C15. Vinyl iodide 14 can be further disconnected into acid
13[13] and the partially protected triol 12, which could be
obtained through a stereoselective aldol reaction between
methyl ketone 3 and protected b-hydroxy aldehyde 9,
followed by stereoselective anti 1,3-reduction of the resulting
aldol product. Aldehyde 9 was projected to be derived from
chiral epoxide 4[14] by epoxide ring opening with an appro-
priate carbon nucleophile and subsequent elaboration of the
diene moiety.
ꢁ
Scheme 3. a) HC CC(O)OEt, nBuLi, BF3·OEt2, THF, ꢀ788C, 45 min,
86%; b) TBSOTf, 2,6-lutidine, CH2Cl2, 08C, 30 min, 96%; c) PhSH,
NaOMe (5 mol%), MeOH, RT, 14 h, 85%; d) CuI, CH3MgBr, THF,
ꢀ788C!ꢀ308C, 30–40 min, 95%; e) DIBAL-H, CH2Cl2, ꢀ788C,
30 min, 87%; f) CBr4, PPh3, 2,6-lutidine, CH2Cl2, ꢀ788C, 2 h, 85%;
=
g) Bu3SnCH CH2, AsPh3, [Pd2(dba)3] (12 mol%), DMF, RT, 75 h, 79%;
h) TMSI, NEt3, CH2Cl2, 08C, 1.5 h, then K2CO3, MeOH, RT, 5 min,
92%; i) (COCl)2, DMSO, NEt3, CH2Cl2, ꢀ788C, 1 h, 98%. dba=diben-
zylideneacetone, DIBAL-H=diisobutylaluminium hydride, DMF=N,N-
dimethylformamide, DMSO=dimethylsulfoxide, nBuLi=n-butyl-
lithium, Tf=trifluoromethanesulfonyl, TMS=trimethylsilyl.
The particular order of steps projected for the assembly of
the precursor diene for the ring-closure reaction from 12, 13,
=
and Bu3SnCH2CH CH2 was chosen based on the results of
the only prior study on the synthesis of ripostatins by
Kirschning and co-workers (which, however, did not lead to
a total synthesis).[15] Thus, although a Stille coupling of 13 (or
secondary hydroxy group with TBS then furnished TBS ether
5 in 83% overall yield. Treatment of 5 with thiophenol and
a catalytic amount of sodium methoxide in MeOH[18] gave
(Z)-thioether 6 in 85% yield. Treatment of 6 with CH3MgI/Cu
resulted in replacement of the thioether moiety by a methyl
group in excellent yield (95%) with complete retention of
configuration.[18] The reaction proved extremely sensitive to
the exact experimental conditions, such that the desired
product was only obtained if the reaction mixture was allowed
to warm to ꢀ308C in the cooling bath over a period of 30–
=
a suitable ester thereof) with Bu3SnCH2CH CH2 prior to
esterification with 12 would represent a more convergent
approach to the RCM substrate, Kirschning and co-workers
had been unable to implement this strategy, partly owing to
the pronounced tendency of the dienoic acid for double-bond
migration under basic conditions.[15]
As illustrated in Scheme 2, the synthesis of ketone 3
started from 2-bromopropene, which was converted into the
3406 www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 3405 –3409

Angewandte
Chemie
40 min, after addition of 6 to the suspension of dimethylcup-
rate at ꢀ788C. No conversion was observed upon slow
warming to room temperature or if 6 was added at ꢀ408C or
08C. The coupling product was then reduced to alcohol 7 with
DIBAL-H in CH2Cl2 at ꢀ788C; subsequent conversion of 7
into the corresponding bromide with CBr4/Ph3P in the
presence of 2,6-lutidine[19] followed by Stille coupling with
=
Bu3SnCH CH2 gave diene 8 in 55% overall yield (from 6).
The elaboration of 8 into the desired aldehyde 9 turned
out to be more challenging than expected, because the
attempted cleavage of the primary PMB ether under standard
oxidative conditions with DDQ resulted in complete decom-
position. Among the various other methods investigated, the
most efficient approach to the clean removal of the PMB
group from 8 involved treatment with TMSI[20] followed by
cleavage of the ensuing silyl ether with K2CO3 in MeOH, to
give the free primary alcohol in an excellent yield of 92%.
Oxidation of the primary hydroxy group under Swern
conditions then furnished the desired aldehyde 9 in 98%
yield.
With building blocks 3 and 9 in hand, the crucial aldol
reaction was then investigated under Paterson conditions with
the chiral boron enolate derived from methyl ketone 3 by
Scheme 4. a) (+)-Ipc2BCl, NEt3, CH2Cl2, ꢀ788C, then 9, ꢀ788C!
ꢀ208C, 12 h, then H2O2, MeOH, pH 7 buffer, 08C, 90 min, 62% (94%
de); b) CH3CH2CHO, SmI2 (35 mol%), THF, ꢀ208C, 1 h, 93%
(d.r.>20:1); c) TBSCl, imidazole, DMF, RT, 20 h, 94%; d) DIBAL-H,
CH2Cl2, ꢀ788C, 1 h, 88%; e) 13, 2,4,6-trichlorobenzoylchloride, DMAP,
NEt3, toluene, ꢀ788C!ꢀ408C, 3 h, 80%. DMAP=4-dimethylamino-
pyridine, (+)-Ipc2BCl=(+)-diisopinocamphenyl chloroborane.
[21]
treatment with Et3N and (+)-Ipc2BCl in CH2Cl2
(Scheme 4). When a 1.5-fold excess of the enolate was
employed at ꢀ788C to ꢀ208C, the reaction proceeded with
good diastereoselectivity (d.r. > 10:1) and the desired stereo-
isomer could be isolated in 62% yield and 94% de.
The subsequent Evans–Tishchenko reduction[22] of the
aldol product 10 proceeded smoothly and in the presence of
35 mol% of SmI2 furnished the desired b-hydroxy ester 11 in
excellent yield (93%). The latter was then elaborated into the
bis(TBS)-protected triol 12 by reaction with TBSCl and
subsequent reduction of the ester moiety with DIBAL-H; 12
was obtained in 83% yield from 11.[23] Initial attempts at the
esterification of alcohol 12 with acid 13[13] under standard
Yamaguchi conditions at room temperature[24] produced
ꢀ788C followed by slow warming to ꢀ408C, but not above.
With this protocol, 14 could finally be obtained in 80% yield.
Stille coupling between vinyl iodide 14 and
=
Bu3SnCH2CH CH2 proceeded smoothly to furnish diene 15
in 86% yield (Scheme 5), thus setting the stage for ring
closure by RCM. The efficiency of the ring-closure reaction
was initially assessed in a series of small-scale experiments
disappointingly
(< 30%), with
low
yields
significant
amounts of starting alcohol
remaining unreacted. A range
of alternative esterification
methods was then investigated,
none of which provided ester 14
in yields significantly in excess of
30% (if any product was formed
at all). However, a careful re-
assessment of the reaction con-
ditions for the Yamaguchi
esterification showed that the
reaction was significantly more
efficient at lower temperature.
Based on this finding, an opti-
mized protocol for the esterifi-
cation of 12 with 13 was devel-
oped; this protocol involved
mixing all the components (12,
=
Scheme 5. a) Bu3SnCH2CH CH2, AsPh3, [Pd2(dba)3] (15 mol%), DMF, RT, 16 h, 86%; b) Grubbs I
(12 mol%), CH2Cl2, RT, 3.5 h, then DMSO, RT, 17 h, 77%; c) HF·py, pyridine, THF, 08C, 4 h, 81%;
d) DMP, THF, RT, 30 min, then NaClO2, NaH2PO4·H2O, tBuOH, H2O, 2-methyl-2-butene, RT, 20 min,
81%; e) 5% aq. HF in CH3CN, 08C, 3 h, 88%. DMP=Dess–Martin periodinane, Grubbs I=bis(tricyclo-
13,
trichlorobenzoylchloride,
Et3N, DMAP) in toluene at hexylphosphine)benzylidene ruthenium(IV)dichloride.
Angew. Chem. Int. Ed. 2012, 51, 3405 –3409
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 3407

.
Angewandte
Communications
(2 mg of 15) with either Grubbs II or Hoveyda–Grubbs II
catalyst.[12] Rapid consumption of diene 15 was observed in all
cases (in CH2Cl2, ClCH2CH2Cl, and toluene), with MS
analysis of the reaction mixtures indicating the presence of
a product with the mass of the desired macrolactone 16
(Scheme 5); however, TLC analysis after concentration con-
sistently showed the formation of three different products
with similar Rf values that were not characterized further. In
contrast, the use of Grubbs I catalyst in CH2Cl2 at room
temperature gave clean conversion of 15 into 16 and
eventually provided 16 in 77% yield.[25] The best results
were obtained when the reaction was terminated by addition
of DMSO before full consumption of the starting material;
the latter could be readily separated from 16 by flash
chromatography.
The selective cleavage of the primary TBS ether in 16 was
first attempted with sodium periodate in aqueous THF,[26] but
the reaction was found to be concentration and scale
dependent; the yield of alcohol 17 varied between 50% and
70%, with significant amounts of the doubly deprotected
species being formed with the larger-scale reactions. Gratify-
ingly, however, treatment of 16 with pyridine-buffered HF·py
at 08C led to clean formation of primary alcohol 17 in 81%
yield, with only trace amounts of doubly deprotected material
being detected. Subsequent attempts at the direct oxidation of
17 to the corresponding carboxylic acid 18 under a variety of
reaction conditions met with complete failure, thus leading us
to resort to a two-step sequence that involved the initial
conversion of 17 into the aldehyde followed by further
oxidation to the carboxylic acid with a different oxidant. Thus,
17 was submitted to Dess–Martin oxidation conditions, which
resulted in spot-to-spot conversion to a less-polar product.
Unfortunately, however, smearing was observed on TLC after
an aq. Na2S2O3/NaHCO3 work-up, and subsequent Pinnick–
Kraus oxidation of this material gave the desired carboxylic
acid 18 in varying yields of only less than 5 to 34%. At the
same time, however, by NMR spectroscopy the aldehyde was
found to be stable under the conditions of the Dess–Martin
oxidation for at least one hour, thus suggesting that its further
oxidation without isolation might lead to cleaner and higher
yielding conversion into the carboxylic acid. This approach
could be successfully implemented, such that the addition of
NaClO2, NaH2PO4·H2O, tBuOH, H2O, and 2-methyl-2-butene
to the Dess–Martin oxidation mixture furnished the desired
carboxylic acid 18 in an excellent yield of 81%.[27] Cleavage of
the two secondary TBS ethers in 18 with 5% aqueous HF in
CH3CN then gave ripostatin B (1) in 88% yield.
different parts of the ripostatin structure and thus provides
a sound basis for future SAR studies and lead optimization
efforts. Work along these lines has been initiated in our
laboratory.
Received: February 1, 2012
Published online: February 29, 2012
Keywords: antibiotics · ring-closing metathesis · ripostatin ·
.
RNA polymerase · total synthesis
[1] a) A. R. M. Coates, G. Halls, Y. Hu, Br. J. Pharmacol. 2011, 163,
184 – 194; b) C. Walsh, Nat. Rev. Microbiol. 2003, 1, 65 – 70.
[2] P. G. Ambrose, Nat. Med. 2011, 17, 772.
[3] a) A. Koul, E. Arnoult, N. Lounis, J. Guillemont, K. Andries,
Nature 2011, 469, 483 – 490; b) C. E. Barry III, Curr. Top. Med.
Chem. 2011, 11, 1216 – 1225.
[4] For recent reviews see: a) A. Srivastava, M. Talaue, S. Liu, D.
Degen, R. Y. Ebright, E. Sineva, A. Chakraborty, S. Y. Druzhi-
nin, S. Chatterjee, J. Mukhopadhyay, Y. W. Ebright, A. Zozula, J.
Shen, S. Sengupta, R. R. Niedfeldt, C. Xin, T. Kaneko, H.
Irschik, R. Jansen, S. Donadio, N. Connell, R. H. Ebright, Curr.
Opin. Microbiol. 2011, 14, 532 – 543; b) R. Mariani, S. I. Maffioli,
Curr. Med. Chem. 2009, 16, 430 – 454.
[5] D. Haebich, F. von Nussbaum, Angew. Chem. 2009, 121, 3447 –
3451; Angew. Chem. Int. Ed. 2009, 48, 3397 – 3400.
[6] An exception is the natural product lipiarmycin (fidaxomicin),
which was recently approved by the FDA and the EMEA for the
treatment of C. difficile infections.
[7] J. Mukhopadhyay, K. Das, S. Ismail, D. Koppstein, M. Jang, B.
Hudson, S. Sarafianos, S. Tuske, J. Patel, R. Jansen, H. Irschik, E.
Arnold, R. H. Ebright, Cell 2008, 135, 295 – 307.
[8] H. Irschik, H. Augustiniak, K. Gerth, G. Hçfle, H. Reichenbach,
J. Antibiot. 1995, 48, 787 – 792.
[9] H. Augustiniak, H. Irschik, G. Hçfle, H. Reichenbach, Liebigs
Ann. 1996, 1657 – 1663.
[10] A. OꢀNeill, B. Oliva, C. Storey, A. Hoyle, C. Fishwick, I. Chopra,
Antimicrob. Agents Chemother. 2000, 44, 3163 – 3166.
[11] a) P. Winter, W. Hiller, M. Christmann, Angew. Chem., 2012,
124, 3452 – 3456; Angew. Chem. Int. Ed., 2012, 51, 3396 – 3400;
b) W. Tang, E. V. Prusov, Angew. Chem., 2012, 124, 3457 – 3460;
Angew. Chem. Int. Ed., 2012, 51, 3401 – 3404.
[12] For a recent review on the metathesis reaction, see: A. H.
Hoveyda, A. R. Zhugralin, Nature 2007, 450, 243 – 251.
[13] M. Kanematsu, M. Shindo, M. Yoshida, K. Shishido, Synthesis
2009, 17, 2893 – 2904.
[14] C. W. Wullschleger, J. Gertsch, K.-H. Altmann, Org. Lett. 2010,
12, 1120 – 1123.
[15] a) C. Kujat, M. Bock, A. Kirschning, Synlett 2006, 419 – 422;
b) C. Kujat, Dissertation, Wilhelm Leibniz Universitꢁt Hann-
over, 2007.
[16] A. E. Guthrie, J. E. Semple, M. M. Joulliꢂ, J. Org. Chem. 1982,
47, 2369 – 2376.
[17] J. M. Williams, R. B. Jobson, N. Yasuda, G. Marchesini, U. H.
Dolling, E. J. J. Grabowski, Tetrahedron Lett. 1995, 36, 5461 –
5464.
[18] T. Murata, M. Sano, H. Takamura, I. Kadota, D. Uemura, J. Org.
Chem. 2009, 74, 4797 – 4803.
[19] D. Enders, J. L. Vicario, A. Job, M. Wolberg, M. Mꢃller, Chem.
Eur. J. 2002, 8, 4272 – 4284.
Although the 1H and 13C NMR spectra of synthetic
1 deviated from those reported for natural 1[9] in some of
the signals, the 1H NMR spectrum fully matches with the one
that has recently been obtained for natural ripostatin B by the
group of Prusov.[11b]
In conclusion, we have developed an efficient modular
total synthesis of the bacterial RNAP inhibitor ripostatin B.
The longest linear sequence in the synthesis comprised 21
steps (from d-aspartic acid; as the starting material for
epoxide 4) and furnished the target molecule in 3.6% overall
yield. The modular nature of our approach lends itself to the
synthesis of analogues with independent modifications in
[20] I. Kadota, Y. Yamagami, N. Fujita, H. Takamura, Tetrahedron
Lett. 2009, 50, 4552 – 4553.
[21] I. Paterson, M. Goodman, Tetrahedron Lett. 1989, 30, 997 – 1000.
[22] D. A. Evans, A. H. Hoveyda, J. Am. Chem. Soc. 1990, 112, 6447 –
6449.
3408 www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 3405 –3409

Angewandte
Chemie
[23] The absolute configuration at C13 and C15 was ascertained: 1)
by Mosher ester analysis with alcohol 12, which proved the
configuration at C13 to be S; and 2) by 13C NMR analysis of the
acetonide derived from 11 by ester cleavage and acetal
formation with 2,2-dimethoxypropane according to Rychnovsky
et al., which proved the relative configuration of the 1,3-diol
moiety to be anti: S. D. Rychnovsky, B. N. Rogers, T. I.
Richardson, Acc. Chem. Res. 1998, 31, 9 – 17.
[24] J. Inanaga, K. Hirata, H. Saeki, T. Katsuki, M. Yamaguchi, Bull.
Chem. Soc. Jpn. 1979, 52, 1989 – 1993.
[25] The coupling constant between the olefinic protons at C5 and C6
was 14.8 Hz. This clearly indicates a trans geometry of the newly
ꢀ
formed C5 C6 double bond.
[26] J. Li, D. Menche, Synthesis 2009, 1904 – 1908.
[27] For another example of a one-pot DMP/Pinnick – Kraus double
oxidation, see: A. Yajima, M. Yamamoto, T. Nukada, G. Yabuta,
Biosci. Biotechnol. Biochem. 2007, 71, 2720 – 2724.
Angew. Chem. Int. Ed. 2012, 51, 3405 –3409
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 3409