A Concise and Highly Enantioselective Total Synthesis of (+)-anti- and (-)-syn-Mefloquine Hydrochloride: Definitive Absolute Stereochemical Assignment of the Mefloquines

Article abstract of DOI:10.1002/anie.201507304

A concise asymmetric (>99:1 e.r.) total synthesis of (+)-anti- and (-)-syn-mefloquine hydrochloride from a common intermediate is described. The key asymmetric transformation is a Sharpless dihydroxylation of an olefin that is accessed in three steps from commercially available materials. The Sharpless-derived diol is converted into either a trans or cis epoxide, and these are subsequently converted into (+)-anti- and (-)-syn-mefloquine, respectively. The synthetic (+)-anti- and (-)-syn-mefloquine samples were derivatized with (S)-(+)-mandelic acid tert-butyldimethylsilyl ether, and a crystal structure of each derivative was obtained. These are the first X-ray structures for mefloquine derivatives that were obtained by coupling to a known chiral, nonracemic compound, and provide definitive confirmation of the absolute stereochemistry of (+)-anti- as well as (-)-syn-mefloquine.

Full text of DOI:10.1002/anie.201507304

.
Angewandte
Communications
International Edition: DOI: 10.1002/anie.201507304
German Edition: DOI: 10.1002/ange.201507304
Total Synthesis
Hot Paper
AConcise and Highly Enantioselective Total Synthesis of (+)-anti- and
(À)-syn-Mefloquine Hydrochloride: Definitive Absolute Stereochem-
ical Assignment of the Mefloquines
Ettore J. Rastelli and Don M. Coltart*
Abstract: A concise asymmetric (> 99:1 e.r.) total synthesis of
(+)-anti- and (À)-syn-mefloquine hydrochloride from
a common intermediate is described. The key asymmetric
transformation is a Sharpless dihydroxylation of an olefin that
is accessed in three steps from commercially available materi-
als. The Sharpless-derived diol is converted into either a trans
or cis epoxide, and these are subsequently converted into
(+)-anti- and (À)-syn-mefloquine, respectively. The synthetic
(+)-anti- and (À)-syn-mefloquine samples were derivatized
with (S)-(+)-mandelic acid tert-butyldimethylsilyl ether, and
a crystal structure of each derivative was obtained. These are
the first X-ray structures for mefloquine derivatives that were
obtained by coupling to a known chiral, nonracemic com-
pound, and provide definitive confirmation of the absolute
stereochemistry of (+)-anti- as well as (À)-syn-mefloquine.
Figure 1. The (+)- and (À)-anti-mefloquine hydrochloride and (+)- and
(À)-syn-mefloquine hydrochloride salts.
M
alaria is the deadliest parasitic disease to affect humans.
including one instance in which it was obtained by resolution
of the racemate.[9] Of the remaining approaches to enantio-
merically enriched (+)-anti-mefloquine hydrochloride, three
employed asymmetric hydrogenation to set the C11 stereo-
genic center (88–96% ee).[10] This was followed by substrate-
controlled catalytic hydrogenation to set the C12 stereogenic
center, which gave 85:15 diastereoselectivity in favor of the
correct diastereomer.[11] In another approach to enantiomer-
ically enriched (+)-anti-mefloquine hydrochloride, an orga-
nocatalytic aldol addition was used (71% ee).[12] Further
synthetic operations conducted on the aldol product led to an
increase in the enantiomeric purity, producing (+)-anti-
mefloquine hydrochloride in 95% ee. (+)-anti-Mefloquine
hydrochloride was also prepared in unspecified enantioselec-
tivity starting from (S)-(À)-1-N-Boc-2-piperidinecarboxylic
acid.[13] The groups of Hall[14] and subsequently Leonov[15]
independently reported routes to both syn- and anti-meflo-
quine hydrochloride in enantiomerically enriched form. In
each case, these routes were used to access all four stereo-
isomers. In Hallꢀs synthesis, a novel enantioselective (99% ee)
allyl boration reaction was used as a key step to provide
access to a syn-vicinal amino alcohol precursor that was used
to obtain syn-mefloquine hydrochloride.[16] Synthesis of the
anti isomer required the syn-vicinal amino alcohol to be
oxidized to a vicinal keto amine, which then underwent
diastereoselective (10:1 d.r.) reduction to give an anti-vicinal
amino alcohol, which was finally converted into anti-meflo-
quine hydrochloride. A key step in Leonovꢀs synthesis was the
diastereoselective (12:1 d.r.) addition of trimethylsilyl acety-
lide to an enantiomerically enriched aldehyde derived from
pipecolinic acid. The major diastereomer from this reaction
was advanced through a Sonogashira/6p electrocyclization
domino reaction to give anti-mefloquine hydrochloride as the
In 2013, an estimated 198 million new cases were reported
worldwide, leading to 584000 deaths.[1] anti-Mefloquine
hydrochloride[2,3] (1; Figure 1) is a highly effective drug that
has been used both for malaria treatment and prophylaxis.
The drug is manufactured commercially and administered in
racemic form, and is marketed under the name Lariam.[4]
While effective and widely used, mefloquine hydrochloride
suffers from serious side effects.[5,6] It has been reported that
the potencies of the two enantiomers of mefloquine hydro-
chloride appear to be unequal, with the (+)-enantiomer being
at least 1.5 times more active than the (À)-enantiomer.[7]
Furthermore, although distributed across many different
types of tissue, evidence suggests that the (À)-enantiomer
has a shorter in vivo half-life owing to higher blood plasma
concentrations.[8] This has prompted interest in the enantio-
selective synthesis of mefloquine to further understand and
potentially benefit from its use as a single enantiomer. Herein,
we describe a concise asymmetric (> 99:1 e.r.) total synthesis
of (+)-anti- and (À)-syn-mefloquine hydrochloride from
a common intermediate. Moreover, we provide definitive
confirmation of the absolute stereochemistry of (+)-anti-
mefloquine as well as the first X-ray crystal structure of (À)-
syn-mefloquine.
Several syntheses of enantiomerically enriched samples of
(+)-anti-mefloquine hydrochloride have been described,
[*] E. J. Rastelli, Prof. D. M. Coltart
Department of Chemistry, University of Houston
Houston, TX (USA)
E-mail: dcoltart@uh.edu
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201507304.
14070
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2015, 54, 14070 –14074

Angewandte
Chemie
major product and syn-mefloquine hydrochloride as the
minor product. (The latter was formed as a side product by
epimerization during the domino reaction.) The anti and syn
diastereoisomers were separated chromatographically, and
the enantiomeric ratio was determined to be either 93:7 or
96:4 depending on the stereoisomer.
We recently reported an asymmetric total synthesis of
(À)-anti-mefloquine hydrochloride (> 99:1 e.r.).[17] A key step
in that synthesis was a cascade azide reduction/epoxide ring-
opening reaction beginning with 3 that, following trapping of
the 3,4-dehydropiperidine intermediate with Boc2O, deliv-
ered 3,4-dehydro-N-Boc-mefloquine (4) in a one-pot reaction
(Scheme 1). Compound 4 subsequently underwent olefin
reduction, acid-mediated removal of the Boc group, and HCl
salt formation to give (À)-mefloquine hydrochloride [(À)-1]
in excellent yield.
Scheme 2. Proposed synthesis of (+)-1 and (À)-2. PG=protecting
group.
Scheme 1. Prior synthesis of (À)-anti-mefloquine hydrochloride [(À)-1].
Boc=tert-butyloxycarbonyl, TFA=trifluoroacetic acid.
We began our studies by converting commercially avail-
able quinolinol 11 into aryl bromide 12 according to
a literature procedure (Scheme 3).[20] Compound 14 was
prepared from phthalimide and 5-hexen-1-ol. We chose to
mask the amine function as a phthalimide rather than an azide
as we had done previously (Scheme 1) to avoid the expected
complications related to the use of phosphine ligands in the
Heck reaction. In the case of the phthalimide-masked amine,
we would initiate the cascade deprotection/cyclization
sequence (see Scheme 1) by treatment with hydrazine (see
below).
With 12 and 14 in hand, we attempted the Heck coupling
reaction. As shown in Table 1, of the conditions tried, our best
results were obtained using NHC ligand 16. Purification of 15
from the reaction mixture by column chromatography on
silica gel proved extremely difficult. This was also the case for
We reasoned that it should be possible to utilize the above
cyclization strategy as the basis for a concise synthesis of not
only (+)- and (À)-anti-mefloquine hydrochloride [(+)-1 and
(À)-1, respectively], but also the syn diastereoisomers [(+)-2
and (À)-2]. The general idea is retrosynthetically outlined in
Scheme 2. Cyclization of trans epoxide 5 would lead to
(+)-anti-mefloquine hydrochloride [(+)-1], whereas cycliza-
tion of cis epoxide 6 would give rise to (À)-syn-mefloquine
hydrochloride [(À)-2]. Access to epoxides 5 and 6 would
mark a key divergence point in our strategy, as they would
each be obtained from a common diol intermediate (7), which
would be prepared in enantiomerically enriched form by the
asymmetric Sharpless dihydroxylation[18] of olefin 8. We
expected that formation of the trans epoxide from 7 would
be reasonably straightforward using the Sharpless method.[19]
However, the stereospecific formation of the cis epoxide
would require one of the two hydroxy groups to be
regioselectively converted into a leaving group, with the
other acting as a nucleophile in an SN2 reaction. Given the
strongly electron-withdrawing nature of the aryl system, we
reasoned that the C11 hydroxy group should be considerably
more acidic than the one at the C12 position, which should
allow selective base-mediated functionalization of the C11
hydroxy group (see below). In turn, the olefin (8) required for
the dihydroxylation would be generated by a Heck reaction
between aryl halide 9 and N-protected 1-aminohexene 10.
Scheme 3. Synthesis of the Heck coupling substrates 12 and 14.
DIAD=diisopropyl azodicarboxylate.
Angew. Chem. Int. Ed. 2015, 54, 14070 –14074
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 14071

.
Angewandte
Communications
Table 1: Ligand screen for the Heck coupling of 12 and 14.[a]
Entry
Ligand
Yield [%]
1
2
3
4
P(tBu)3
XPhos
16
–
48
68
65
17
[a] Reaction conditions: Pd(OAc)2 (5 mol%), ligand (10 mol%), Et3N,
DMF, 1108C. XPhos=2-dicyclohexylphosphino-2’,4’,6’-triisopropylbi-
phenyl.
entries 2 and 4 of Table 1. Therefore, the yields given in
Table 1 correspond to material of approximately 95% purity.
At this point, we tried the Sharpless asymmetric dihy-
droxylation on the semi-purified material, suspecting that the
diol product might be easier to purify than the olefin.
Therefore, 15 (approximately 95% pure, as judged by
1H NMR analysis, obtained by the reaction shown in entry 3
of Table 1) was subjected to the dihydroxylation conditions[18]
(15!18, Scheme 4). Gratifyingly, this produced the pure diol
in 47% over two steps based on 12, following column
chromatography on silica gel. The enantiomeric purity of 18
was established by converting it into the corresponding
acetonide[21,22] and comparing it with a racemic acetonide[23]
sample by HPLC analysis on a chiral stationary phase.[24] The
e.r. of the acetonide was found to be 96:4, which we assume to
be a reasonable indication of the enantiomeric purity of diol
18. The enantiomeric purity of 18 was readily increased to
99:1 by a single and efficient (92% yield) recrystallization.
With access to the diol secured, we converted it into trans
epoxide 19 in very good yield by the Sharpless one-pot
procedure (18!19).[19] To ensure that there was no loss of
stereochemical integrity during epoxide formation, it was
compared to a corresponding racemic sample[25] by HPLC
analysis on a chiral stationary phase.[24] The e.r. of epoxide 19
was thus found to be > 99:1. Compound 19 was next treated
with hydrazine followed by acidification with HCl, which, as
anticipated, provided (+)-anti-mefloquine hydrochloride
Scheme 4. Synthesis of (+)-1 and (À)-2. PPTS=pyridinium para-
toluenesulfonate, Ts =para-toluenesulfonyl.
taking advantage of the predicted enhanced acidity of the C11
hydroxy group—which would allow it to be regioselectively
activated—and having it undergo SN2 displacement by the
C12 hydroxy group. After considerable experimentation, we
found that this was most effectively achieved by treatment of
18 with 0.8 equiv of tBuOK and 1.0 equiv of p-TsCl to give the
C11 tosylate. This was then converted into epoxide 20 in 76%
yield over two steps (based on recovered 18) by treatment
with K2CO3. The enantiomeric purity of 20 was determined by
comparison to the corresponding racemate[26] by HPLC
analysis on a chiral stationary phase, and found to be > 99:1
e.r.[24]
With cis epoxide 20 in hand, we were pleased to find that
in an analogous manner to the synthesis of (+)-1 described
above, it could be smoothly converted into the syn diaste-
reomer (À)-2 by treatment with hydrazine followed by HCl.
20
D
The optical rotation of (À)-2 was measured as ½aŠ ¼À41.4
(c = 0.91, MeOH), which is in reasonable agreement with the
20
reported value of ½aŠ ¼À49.9 (c = 0.91, MeOH).[14] The
D
enantiomeric purity of our synthetic material was determined
in a similar manner to that described above,[17] and found to
be > 99:1 e.r.
[(+)-1] in excellent yield. The optical rotation of the HCl
We next set out to confirm the absolute stereochemistry of
our synthetic material. Over the years, there has been
controversy regarding the absolute configuration of anti-
mefloquine. Originally, on the basis of circular dichroism
(CD) experiments, (+)-anti- and (À)-anti-mefloquine were
assigned the structures opposite to those shown in Figure 1 by
Carroll and Blackwell.[9a] Subsequently, using X-ray diffrac-
tion studies on anti-mefloquine hydrochloride, Karle and
Karle argued that the correct structures of (+)-anti- and (À)-
20
salt was found to be ½aŠ ¼ + 26.0 (c = 0.45, MeOH), which
D
20
D
matches quite well with the reported value of ½aŠ ¼ + 29.8
(c = 0.45, MeOH).[14] The enantiomeric purity of our synthetic
material was established in the same manner as we described
previously,[17] and was found to be > 99:1 e.r.
We next turned our attention to preparing syn-mefloquine
[(À)-2]. As indicated above, we needed to convert diol 18 into
cis epoxide 20 or its enantiomer. We planned to do this by
14072 www.angewandte.org
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2015, 54, 14070 –14074

Angewandte
Chemie
anti-mefloquine were those shown in Figure 1 [(+)-1 and
(À)-1, respectively].[7] Arguments against this revised assign-
ment were then made by Xie and co-workers[12] based upon
a Mosher analysis[27] carried out as part of their total synthesis
of (À)-anti-mefloquine hydrochloride, which supported the
original Carroll–Blackwell assignment (i.e., the opposite of
what is shown in Figure 1). Recently, another argument
regarding the absolute configuration of (+)- and (À)-anti-
mefloquine was put forth in a study by Reinscheid, Grie-
singer, and co-workers.[2] This analysis was based upon
residual dipolar coupling (RDC) enhanced NMR spectros-
copy in combination with optical rotatory dispersion (ORD)
and CD spectroscopy. The result of this study recognized the
absolute configuration given by Karle and Karle[7] to be
correct [i.e., (+)-1 and (À)-1, Figure 1]. Support for this
assignment based on a synthetic study was subsequently
provided by Hall and Ding.[14]
We next carried out the same type of absolute config-
uration analysis on compound 24, which was again prepared
according to Scheme 4, but without the final HCl salt
formation step. Coupling of 24 with (S)-22 gave 25
(Scheme 6). Again, using the known S configuration of the
mandelic acid derived stereogenic center as a reference, X-ray
crystallographic analysis established the absolute stereochem-
istry of the C11 and C12 positions of 25 to be R and R,
respectively. Again, this is the first X-ray structure reported
for a syn-mefloquine derivative obtained by coupling to
a known chiral, nonracemic compound. On this basis, we were
able to unambiguously establish the absolute stereochemistry
of (À)-syn-mefloquine hydrochloride to be that shown in
Figure 1 [(À)-2].
Notwithstanding the importance of the above studies to
determine the absolute configuration of (+)- and (À)-anti-
mefloquine hydrochloride, we were surprised to find that no
studies had been reported on the derivatization of the
mefloquine enantiomers with a known chiral, nonracemic
compound, followed by X-ray crystallographic analysis of the
derivative. Consequently, we set out to do this.
Thus, compound 21 was prepared according to Scheme 4,
but without the final HCl salt formation step that was used to
prepare (+)-1. It was then coupled with (S)-(+)-mandelic acid
tert-butyldimethylsilyl ether (22)[28] under non-epimerizing
conditions to give 23 (Scheme 5). By reference to the known
S configuration of the mandelic acid derived stereogenic
center, X-ray crystallography determined the absolute ste-
reochemistry of the C11 and C12 positions of 23 to be S and R,
respectively. This analysis confirmed the absolute stereo-
chemistry of (+)-anti-mefloquine to be that shown in Figure 1
[(+)-1], thus verifying the assignment by Karle and Karle.[7]
Scheme 6. Synthesis and X-ray crystal structure of 25.
In conclusion, we have developed a concise and highly
enantioselective (> 99:1 e.r.) approach to both (+)-anti- and
(À)-syn-mefloquine hydrochloride [(+)-1 and (À)-2, respec-
tively]. A key step in the synthesis of each compound is the
enantioselective conversion of olefin 15 into diol 18, which is
then converted into either a trans or syn epoxide and
ultimately into (+)-anti-mefloquine hydrochloride [(+)-1] or
(À)-syn-mefloquine hydrochloride [(À)-2], respectively. As
a result of X-ray crystallographic and optical rotation studies
on our synthetic material, the absolute stereochemistry of
(+)-anti-mefloquine hydrochloride has been definitively
confirmed as that reported by Karle and Karle[7] [(+)-1,
Figure 1]. Furthermore, a similar analysis has unambiguously
established the absolute stereochemistry of (À)-syn-meflo-
quine hydrochloride [(À)-2, Figure 1].
Scheme 5. Synthesis and X-ray crystal structure of 23. HATU =O-(7-
azabenzotriazol-1-yl)-N,N,N’,N’-tetramethyluronium hexafluorophos-
phate, TBS=tert-butyldimethylsilyl.
Angew. Chem. Int. Ed. 2015, 54, 14070 –14074
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 14073

.
Angewandte
Communications
Zutter, Pure Appl. Chem. 1996, 68, 131 – 138; b) C.-y. Chen,
R. A. Reamer, J. R. Chilenski, C. J. McWilliams, Org. Lett. 2003,
5, 5039 – 5042; c) W. P. Hems, W. P. Jackson, P. Nightingale, R.
Byrant, Org. Process Res. Dev. 2012, 16, 461 – 463.
[11] E. Brober, W. Hofneinz, A. Meili (Hoffmann – La Roche), Eur.
Pat. 553778, 1993.
[12] Z.-X. Xie, L.-Z. Zhang, X.-J. Ren, S.-Y. Tang, Y. Li, Chin. J.
Chem. 2008, 26, 1272 – 1276.
[13] G. Zhou, X. Liu, X. Liu, H. Nie, S. Zhang, W. Chen, Adv. Synth.
Catal. 2013, 355, 3575 – 3580.
Acknowledgements
We are grateful to James Korp of the University of Houston
for determining the X-ray crystal structures of 23 and 25. We
thank Professor K. C. Nicolaou (Rice University) for gen-
erously allowing us use of his laboratory for polarimetry
measurements. We also thank the University of Houston and
the NSF for financial support (NSF 1300652).
[14] J. Ding, D. G. Hall, Angew. Chem. Int. Ed. 2013, 52, 8069 – 8073;
Angew. Chem. 2013, 125, 8227 – 8231.
Keywords: asymmetric synthesis · malaria · mefloquine ·
Sharpless dihydroxylation · total synthesis
[15] N. Schützenmeister, M. Müller, U. M. Reinscheid, C. Griesinger,
A. Leonov, Chem. Eur. J. 2013, 19, 17584 – 17588.
[16] As part of this work, Hall and Ding determined that the syn-
mefloquines are also active against Plasmodium falciparum; see
Ref. [14].
How to cite: Angew. Chem. Int. Ed. 2015, 54, 14070–14074
Angew. Chem. 2015, 127, 14276–14280
[1] a) World Health Organization, T. R. Soc. Trop. Med. H. 2000, 94
Suppl. 1, S1/1 – S1/90; b) World Health Organization Fact Sheet
No. 94; http://www.who.int/mediacentre/factsheets/fs094/en/;
accessed August 2015.
[2] M. Schmidt, H. Sun, P. Rogne, G. K. E. Scriba, C. Griesinger,
L. T. Kuhn, U. M. Reinscheid, J. Am. Chem. Soc. 2012, 134,
3080 – 3083.
[3] A recent study (see Ref. [2]) revealed that the original stereo-
chemical assignment of (+)- and (À)-mefloquine was likely
incorrect and determined the correct assignment to be that
shown in Figure 1. The designations shown in Figure 1 are used
in this manuscript.
[4] For total syntheses of (Æ)-mefloquine, see: a) C. J. Ohnmacht,
A. R. Patel, R. E. Lutz, J. Med. Chem. 1971, 14, 926 – 928;
b) M. S. Kumar, Y. V. D. Nageshwar, H. M. Meshram, Synth.
Commun. 1996, 26, 1913 – 1919; c) A. Solange, Tetrahedron 1989,
45, 1409 – 1414.
[17] J. D. Knight, S. J. Sauer, D. M. Coltart, Org. Lett. 2011, 13, 3118 –
3121.
[18] K. B. Sharpless, W. Amberg, Y. L. Bennani, G. A. Crispino, J.
Hartung, K.-S. Jeong, H.-L. Kwong, K. Morikawa, Z.-M. Wang,
D. Xu, X.-L. Zhang, J. Org. Chem. 1992, 57, 2768 – 2771.
[19] H. C. Kolb, K. B. Sharpless, Tetrahedron 1992, 48, 10515 – 10530.
[20] A. Jonet, A. Dassonville-Klimpt, S. D. Nascimento, J.-M. Leger,
J. Guillon, P. Sonnet, Tetrahedron: Asymmetry 2011, 22, 138 –
148.
[21] We were not able to resolve the diol (18) enantiomers by HPLC
using the chiral stationary phases at our disposal.
[22] Generated using 2,2-dimethoxypropane. See the Supporting
Information for details.
[23] Generated from 15 using OsO4 and NMO and then 2,2-
dimethoxypropane. See the Supporting Information for details.
[24] See the Supporting Information for details.
[25] Generated from 15 using m-CPBA. See the Supporting Infor-
mation for details.
[26] See the Supporting Information for the synthesis of racemic 20.
[27] T. R. Hoye, C. S. Jeffrey, F. Shao, Nat. Protoc. 2007, 2, 2451 –
2458.
[5] L. H. Chen, M. E. Wilson, P. Schlagenhauf, J. Am. Med. Assoc.
2007, 297, 2251 – 2263.
[6] H. O. AlKadi, Chemotherapy 2007, 53, 385 – 391.
[7] J. M. Karle, I. L. Karle, Antimicrob. Agents Chemother. 2002, 46,
1529 – 1534.
[28] The optical rotation of the S-(+)-mandelic acid sample used to
[8] G. S. Dow, M. L. Koenig, L. Wolf, L. Gerena, M. Lopez-Sanchez,
T. H. Hudson, A. K. Bhattacharjee, Antimicrob. Agents Chemo-
ther. 2004, 48, 2624 – 2632.
[9] a) F. I. Carroll, J. T. Blackwell, J. Med. Chem. 1974, 17, 210 – 219;
b) A. D. Baxter, M. J. Harris, S. Brown, International PCT
patent application PCT/GB2003/005286, 2003.
prepare 22 (see the Supporting Information) was measured as
20
½aŠ ¼ + 153.52 (c = 2.5, H2O), which is consistent with the
D
20
literature value [½aŠ ¼ + 149.0 (c = 2.5, H2O)]; see: E. Hernan-
D
dez, J. A. Soderquist, Org. Lett. 2005, 7, 5397 – 5400.
[10] a) R. Schmid, E. A. Broger, M. Cereghetti, Y. Crameri, J.
Foricher, M. Lalonde, R. K. Müller, M. Scalone, G. Schoettel, U.
Received: August 5, 2015
Published online: September 30, 2015
14074 www.angewandte.org
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2015, 54, 14070 –14074