Diastereodivergent de-epimerization in catalytic asymmetric allylic alkylation

Article abstract of DOI:10.1002/anie.201202853

Diastereomers made to order: In an unprecedented ligand-controlled process a racemic mixture of four stereoisomers can be converted with high selectivity into each one of the diastereomers of the product, at will (see scheme). The mechanism of this deracemization of epimers, that is, a de-epimerization, was also studied. Copyright

Full text of DOI:10.1002/anie.201202853

Angewandte
Chemie
DOI: 10.1002/anie.201202853
Asymmetric Catalysis
Diastereodivergent De-epimerization in Catalytic Asymmetric Allylic
Alkylation**
Davide Audisio, Marco Luparia, Maria Teresa Oliveira, Dina Klꢀtt, and Nuno Maulide*
The deracemization of a racemic mixture has become
a powerful alternative to classical resolution techniques, and
dynamic kinetic resolution (DKR) and dynamic kinetic
asymmetric transformation (DYKAT) represent the cutting-
edge technologies in this field.[1] The ability to overcome the
main drawback of simple kinetic resolutions (namely the
maximum yield of enantiopure product of 50%) by directly
transforming a racemic mixture into a single enantiopure
product in 100% theoretical yield justifies the popularity of
DKR and DYKAT. Despite impressive advances in this area,
the deracemization of epimers or “de-epimerization”—in
other words, the enantioselective and quantitative conversion
of racemic mixtures of diastereomers—still presents a formi-
Scheme 1. a) Previous results on the diastereodivergent deracemiza-
tion of (rac)-1. b) Design of a new system suitable for further
investigation.
dable challenge, and only rare examples are known. All prior
reports in this field allow access to only one of the possible
diastereoisomers of the product.[1d,2]
We have recently developed an unprecedented, palla-
dium-catalyzed ligand-controlled diastereodivergent dera-
cemization, with which the racemic lactone 1[3] can be
converted into any one out of four stereoisomeric products,
in high selectivity (Scheme 1a).[4] At the outset, we wondered
whether the diastereodivergent deracemization was a sub-
strate-specific phenomenon or extendable to other systems.
To address this issue, it was critical to evaluate two aspects:
1) the role played by internal coordination from the pendant
carboxylate in the putative allyl–palladium intermediate;
2) the contribution of strain-release intrinsic to the bicyclic
framework of lactone 1 (Scheme 1b).
into each and every one of four possible stereoisomers of the
product.
The cis-4-chlorocyclobut-2-ene carboxylic acid 2[5]
emerged as an ideal candidate for our studies (Scheme 2),
since it possesses the same stereochemical pattern as 1 (cis
Herein we present our results on these investigations and
mechanistic insight into this phenomenon, as well as the
discovery of a unique and hitherto unknown case of de-
epimerization, termed diastereodivergent de-epimerization,
wherein a racemic mixture of diastereoisomers is converted
Scheme 2. Preparation of cyclobutenes cis-2 and trans-3.
configuration) and exhibits no marked release of ring strain
associated with the departure of the leaving group, and the
latter is not expressed in the final product. Compound cis-2
would also provide the opportunity to manipulate the
carboxylate moiety and study the reactivity of derivatives
(Scheme 1b).
[*] Dr. D. Audisio, Dr. M. Luparia,[+] M. T. Oliveira,[+] D. Klꢀtt,
Dr. N. Maulide
Max-Planck-Institut fꢀr Kohlenforschung
Kaiser-Wilhelm-Platz 1, 45470 Mꢀlheim an der Ruhr (Germany)
E-mail: maulide@mpi-muelheim.mpg.de
Homepage: http://www.kofo.mpg.de/maulide
An efficient stereoselective synthesis of pure racemic cis-2
and trans-3 chloro carboxylic acid precursors was developed
(Scheme 2).[5–7] Compounds 2 and 3 proved reasonably stable,
in contrast to the highly labile nature of 1. Moreover, whereas
cis-2[7] might be seen as a surrogate of lactone 1, the trans
diastereomer 3 would represent a new system that might
advance our understanding of this process.
We first examined the reactivity of racemic cis-2. Pleas-
ingly, it was observed that a palladium-catalyzed diastereo-
divergent deracemization was operative. Under optimized
conditions,[6,8] phosphoramidites[9] L1a and L1b were highly
cis-selective, affording substituted cyclobutenes with excel-
[+] These authors contributed equally to this work.
[**] We thank the Max-Planck-Society and the Max-Planck-Institut fꢀr
Kohlenforschung for support of our research programs. This work
was funded by the Deutsche Forschungsgemeinschaft (grant MA
4861/3-1), the European Research Council (ERC Starting Grant
278872), and the Alexander van Humboldt Foundation (fellowship
to M.L.). We further acknowledge the invaluable support of our
Analytical Departments.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201202853.
Angew. Chem. Int. Ed. 2012, 51, 1 – 5
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1
These are not the final page numbers!

.
Angewandte
Communications
Scheme 4. Scope of the diastereodivergent process from the trans-3.
Conditions A: 2.5 mol% [{Pd(allyl)Cl}2], 7.5 mol% ligand L1a, CH3CN,
08C. Conditions B: 5 mol% [{Pd(allyl)Cl}2], 30 mol% ligand L2, THF,
308C.
Scheme 3. Scope of the diastereodivergent process from the cis-2.
Conditions A: 2.5 mol% [{Pd(allyl)Cl}2], 7.5 mol% ligand L1a, CH3CN,
08C. Conditions B: 2.5 mol% [{Pd(allyl)Cl}2], 15 mol% ligand L2, THF,
408C, slow addition of (rac)-cis-2. The reaction that led to 4e was
performed with L1b in THF.
each ligand leads to a preferred product regardless of the
stereochemical information of the starting material, and both
ligands can mediate either overall retention or overall
inversion (depending on the electrophile considered).
In order to probe the influence of the internal carboxylate
moiety in this ligand-dependent process, esters 6 and 7 were
prepared.[6] In the presence of L1b, the cis esters 6 were
transformed into cis-disubstituted cyclobutenes 8a–d in good
yields and selectivities (Scheme 5). Surprisingly, L2 was not
a competent ligand for this transformation (the outcome of
this reaction did not differ from the corresponding back-
ground reaction in the absence of catalyst).[6]
lent diastereo- and enantioselectivity for stabilized nucleo-
philes such as malonates and azlactones (Scheme 3). On the
other hand, when phosphine-oxazoline[10] L2 was employed,
the trans isomer could be prepared with a high degree of
efficiency (Scheme 3). From a theoretical point of view, yields
higher than 50% along with high enantioselectivities suggest
that a dynamic deracemization is operative.[11] Therefore, ring
strain is not a prerequisite for this process, though we believe
that its significant attenuation in cis-2 might be responsible
for the increased reaction times when L2 is employed.
We subsequently investigated the reactivity of the diaste-
reoisomeric, racemic trans-3 (Scheme 4). To our surprise,
under similar conditions L1a proved again to be cis-selective
and products 4a–i were isolated in very good to excellent
levels of selectivity. This outcome was unexpected as we
originally assumed that L1a mediated stereoretentive allylic
alkylation by palladium catalysis.[12] As before, the substrate
scope was broad and common functional groups were
tolerated.[13]
When L2 was employed, the reaction proved to be
sluggish. Nevertheless, the trans-cyclobutene 5a was obtained
in modest yield as the major diastereomer (d.r. = 95/5) with
reasonable enantioselectivity (Scheme 4).[14] The absence of
detectable epimerization to cis-2 or formation of lactone
1 during background experiments[6] supports the notion that
trans-3 is indeed the real substrate of the transformation.
From these results, it appears that it is the stereochemical
identity of the final product (i.e. cis or trans) and not the
stereochemical outcome of the reaction (i.e. overall retention
or inversion) which is controlled by the ligand. Strikingly,
Scheme 5. Catalytic deracemization of esters (rac)-cis-6.
In sharp contrast to the results obtained with trans acid 3,
L1a smoothly mediated the conversion of trans esters 7 into
trans-cyclobutenes 9a–d (Scheme 6). Once more, ligand L2
led only to trace amounts of racemic background products 9.[6]
2
www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 1 – 5
These are not the final page numbers!

Angewandte
Chemie
In the case of ligand L1a, the observed preference for the
anti p-allyl isomer (and therefore the cis product) is consistent
with the minimization of charge repulsion between the
negatively charged carboxylate and a neutral [LPd(Cl)(h3-
allyl)] intermediate. As a consequence, when the correspond-
ing esters are employed this Coulombic interaction is absent
and the aforementioned dynamic behavior is suppressed. The
role played by the free carboxylate is even more crucial when
the reactivity in the presence of ligand L2 is considered.[16]
Since L2 is a bidentate nonsymmetrical P,N ligand,[17] the
weaker s-donation of the imino fragment to the metal may
result in a more significant contribution of cyclometalated h1-
allyl intermediates, in which the carboxylate moiety of cis-2
and trans-3 can act as an additional ligand.[4] Without this
internal chelation (as in the case of esters cis-6 and trans-7),
the reactivity of L2 is therefore shut down, also suggesting
that internal return[18] can become faster than outer-sphere
nucleophilic attack.[19]
A remarkable consequence of the reactions reported
herein is expressed in the following key experiment: A 1:1
mixture of diastereomers cis-2 and trans-3 (each one as
a racemic mixture of enantiomers) was subjected to the
optimized conditions. The corresponding cis- and trans-cyclo-
butene products were obtained in excellent yields and
stereoselectivities (Scheme 8). In the event, a mixture con-
taining a total of four different stereoisomers of the starting
material was converted into each one of four stereoisomers of
the product at will, with high selectivities, upon simple choice
of the appropriate chiral ligand (Scheme 8).
Scheme 6. Catalytic deracemization of esters (rac)-trans-7.
This contrast between the behavior of the free carboxylic
acids (2 and 3) and their esters (6 and 7) is a salient feature of
this system and highlights the crucial role of the carboxylate in
enabling the diastereodivergent phenomena. Thus, the reac-
tions of esters cis-6 and trans-7 proceed with overall retention
of configuration, while still representing notable examples of
DYKAT. The peculiar behavior observed with L2, for which
the absence of the carboxylate moiety completely shuts down
reactivity, shows that the -COOH moiety is actually a man-
datory feature for activation of the Pd–L2 complex (and
eventual trans selectivity).
Comparison of data reported in Scheme 3 and Scheme 4
reveals that, for each nucleophile, the same enantiomer of the
cis product is obtained with equal enantioselectivity (compare
4a and 4c in Scheme 3 and Scheme 4). This suggests that the
key intermediate involved in the enantiodetermining step is
the same when L1a is employed. We therefore propose
a rationale through which the two p-allyl complexes originally
formed from each starting material interconvert in a dynamic
equilibrium (Scheme 7). This would correspond to a scenario
Scheme 8. Diastereodivergent de-epimerization of a complex racemic
mixture containing four stereoisomeric substances.
This represents a unique and, to the best of our knowl-
edge, unprecedented example of the diastereodivergent de-
epimerization of a mixture of two diastereomeric racemates.
The remarkable ability of L1a and L2 to control the
stereochemical outcome of the reaction regardless of the
stereochemistry of the starting material is noteworthy and
unprecedented in palladium-catalyzed asymmetric allylic
alkylation.
Scheme 7. Proposed mechanism for diastereodivergent de-epimeriza-
tion.
characteristic of DYKAT,[1d] which is supported by the
monitoring of ee values during the reaction.[6] The loss of
the stereochemical integrity of the intermediate p-allyl
complexes in allylic alkylation is a well-known phenomenon
previously attributed to a number of factors, most commonly
the direct nucleophilic displacement of palladium from
a metal–allyl complex by a second, incoming palladium(0)
center.[15]
Received: April 13, 2012
Published online: && &&, &&&&
Angew. Chem. Int. Ed. 2012, 51, 1 – 5
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3
These are not the final page numbers!

.
Angewandte
Communications
Keywords: de-epimerization · diastereodivergence ·
.
enantioselectivity · palladium · reaction mechanisms
[1] a) H. Stecher, K. Faber, Synthesis 1997, 1 – 16; b) K. Faber,
Chem. Eur. J. 2001, 7, 5004 – 5010; c) F. F. Huerta, A. B. E.
Minidis, J. E. Bꢀckvall, Chem. Soc. Rev. 2001, 30, 321 – 331; d) J.
Steinreiber, K. Faber, H. Griengl, Chem. Eur. J. 2008, 14, 8060 –
8072; e) H. Pellissier, Tetrahedron 2011, 67, 3769 – 3802.
[2] For examples of DYKAT type III, see: a) M. Edin, J. Steinreiber,
J. E. Bꢀckvall, Proc. Natl. Acad. Sci. USA 2004, 101, 5761 – 5766;
b) A. B. L. Fransson, Y. M. Xu, K. Leijondahl, J. E. Bꢀckvall, J.
Org. Chem. 2006, 71, 6309 – 6316; c) B. Martꢁn-Matute, J. E.
Bꢀckvall, J. Org. Chem. 2004, 69, 9191 – 9195. See also page S58
of the Supporting Information.
Similarly, trans products 5a–c appear to arise directly from cis-2,
since we never observed epimerization of cis-2 to trans-3 in
control experiments (see the Supporting Information).
[12] a) B. M. Trost, D. L. VanVranken, Chem. Rev. 1996, 96, 395 –
422; b) Z. Lu, S. M. Ma, Angew. Chem. 2008, 120, 264 – 303;
Angew. Chem. Int. Ed. 2008, 47, 258 – 297; c) B. M. Trost, T.
Zhang, J. D. Sieber, Chem. Sci. 2010, 1, 427 – 440; d) G.
Helmchen, U. Kazmaier, S. Fçrster in Catalytic Asymmetric
Synthesis (Ed.: I. Ojima), 3rd ed., Wiley, Hoboken, 2010,
pp. 497 – 641; e) J. C. Fiaud, J. Y. Legros, J. Org. Chem. 1987,
52, 1907 – 1911.
[13] When azlactones served as the nucleophiles, no reaction
occurred (unreacted starting trans-3 was recovered when L1b
was employed).
[14] After 48 h, conversion was not complete (48%) and the
recovered starting material showed only a very modest enan-
tioenrichment (6% ee); this ruled out a kinetic resolution
process.
[3] E. J. Corey, J. Streith, J. Am. Chem. Soc. 1964, 86, 950 – 951.
[4] a) M. Luparia, M. T. Oliveira, D. Audisio, F. Frꢂbault, R.
Goddard, N. Maulide, Angew. Chem. 2011, 123, 12840 – 12844;
Angew. Chem. Int. Ed. 2011, 50, 12631 – 12635. For our previous
work on Pd-mediated cyclobutene synthesis see: b) F. Frꢂbault,
M. Luparia, M. T. Oliveira, R. Goddard, N. Maulide, Angew.
Chem. 2010, 122, 5807 – 5811; Angew. Chem. Int. Ed. 2010, 49,
5672 – 5676; c) M. Luparia, D. Audisio, N. Maulide, Synlett 2011,
735 – 740.
[5] W. H. Pirkle, L. H. Mckendry, J. Am. Chem. Soc. 1969, 91, 1179 –
1186.
[6] See the Supporting Information for details and additional
experiments.
[7] The structure of (rac)-cis 2 was confirmed by X-ray analysis.
CCDC 854277 contains the supplementary crystallographic data
for this paper. These data can be obtained free of charge from
The Cambridge Crystallographic Data Centre via www.ccdc.
cam.ac.uk/data_request/cif.
[8] An additional equivalent of nucleophile was added since in situ
deprotonation of the carboxylic acid moiety of cis-2 and trans-3
consumes a similar amount.
[15] Relevant literature on this topic (p. S58) as well as an outline of
relevant experiments (p. S48 and Ref. [4]) can be found in the
Supporting Information.
[16] Though trans-3 is a less competent substrate than cis-2, it is
noteworthy that both processes lead to the same major
enantiomer of the product (compare product 5a in Schemes 3
and 4).
[9] a) B. L. Feringa, Acc. Chem. Res. 2000, 33, 346 – 353; b) J. F.
Teichert, B. L. Feringa, Angew. Chem. 2010, 122, 2538 – 2582;
Angew. Chem. Int. Ed. 2010, 49, 2486 – 2528.
[17] A. Pfaltz, W. Drury, Proc. Natl. Acad. Sci. USA 2004, 101, 5723 –
5726.
[18] L. A. Evans, N. Fey, J. N. Harvey, D. Hose, G. C. Lloyd-Jones, P.
Murray, A. G. Orpen, R. Osborne, G. J. J. Owen-Smith, M.
Purdie, J. Am. Chem. Soc. 2008, 130, 14471 – 14473. When
conditions for attenuation of ion-pair return were employed,
only low yield and selectivity were observed for the product (see
the Supporting Information for details).
[19] Additional experiments further support the proposal that
bidentate coordination plays a key role in the trans diastereo-
selectivity. See the Supporting Information for details.
[10] G. Helmchen, A. Pfaltz, Acc. Chem. Res. 2000, 33, 336 – 345.
[11] Two key experimental observations disprove the intermediacy of
lactone 1 as a possible substrate of the transformation: 1) in
background experiments in the absence of either palladium/
ligand or nucleophile neither 1 nor products derived from it were
detected; 2) when the best performing ligand (L1e) for 1 was
employed, markedly inferior levels of diastereo- and enantiose-
lectivity were obtained:
4
www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 1 – 5
These are not the final page numbers!

Angewandte
Chemie
Communications
Asymmetric Catalysis
Diastereomers made to order: In an
unprecedented ligand-controlled process
a racemic mixture of four stereoisomers
can be converted with high selectivity into
each one of the diastereomers of the
product, at will (see scheme). The mech-
anism of this deracemization of epimers,
that is, a de-epimerization, was also
studied.
D. Audisio, M. Luparia, M. T. Oliveira,
D. Klꢁtt, N. Maulide*
&&&& — &&&&
Diastereodivergent De-epimerization in
Catalytic Asymmetric Allylic Alkylation
Angew. Chem. Int. Ed. 2012, 51, 1 – 5
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5
These are not the final page numbers!