Non-stabilized nucleophiles in Cu-catalysed dynamic kinetic asymmetric allylic alkylation

Article abstract of DOI:10.1038/nature14089

The development of new reactions forming asymmetric carbon-carbon bonds has enabled chemists to synthesize a broad range of important carbon-containing molecules, including pharmaceutical agents, fragrances and polymers¹. Most strategies to obtain enantiomerically enriched molecules rely on either generating new stereogenic centres fromprochiral substratesor resolving racemicmixtures of enantiomers.Analternative strategy-dynamic kinetic asymmetric transformation-involves the transformation of a racemic starting material into a single enantiomer product, with greater than 50 per centmaximumyield^2,3. The use of stabilized nucleophiles (pK_a<25, where K_a is the acid dissociation constant) in palladium-catalysed asymmetric allylic alkylation reactions has proved to be extremely versatile in these processes^4,5. Conversely, the use of non-stabilized nucleophiles insuch reactions is difficultand remains a key challenge^6-9. Herewe report a copper-catalyseddynamic kinetic asymmetric transformation using racemic substrates and alkyl nucleophiles. These nucleophiles have a pK_a of ≥ 50, more than 25 orders of magnitude more basic than the nucleophiles that are typically used in such transformations. Organometallic reagents are generatedin situ fromalkenes byhydrometallation and givehighly enantioenriched productsunder mild reaction conditions. The method is used to synthesize natural products that possess activity against tuberculosis and leprosy, and an inhibitor ofpara-aminobenzoate biosynthesis. Mechanistic studies indicate that the reaction proceeds through a rapidly isomerizing intermediate.We anticipate that this approach will be a valuable complement to existing asymmetric catalytic methods.

Full text of DOI:10.1038/nature14089

LETTER
doi:10.1038/nature14089
Non-stabilized nucleophiles in Cu-catalysed
dynamic kinetic asymmetric allylic alkylation
Hengzhi You1, Emeline Rideau1, Mireia Sidera1 & Stephen P. Fletcher1
The development of new reactions forming asymmetric carbon– pathway where both enantiomers of the starting materialare converted
carbon bonds has enabled chemists to synthesize a broad range of
important carbon-containing molecules, including pharmaceutical
agents, fragrances and polymers1. Most strategies to obtain enanti-
omerically enriched molecules rely on either generating new stereo-
geniccentresfromprochiralsubstratesorresolvingracemicmixtures
of enantiomers. An alternative strategy—dynamic kinetic asymmetric
transformation—involves the transformation of a racemic starting
material into a single enantiomer product, with greater than 50 per
centmaximumyield2,3. Theuseofstabilizednucleophiles(pKa , 25,
where Ka is the acid dissociation constant) in palladium-catalysed
asymmetric allylic alkylation reactions has proved to be extremely
versatile in these processes4,5. Conversely, the use of non-stabilized
to a common pseudo-prochiral intermediate, and a subsequent enan-
tioselective step creates the new stereogenic centre6. The use of non-
stabilized nucleophiles in such reactionsis along-standing research goal6–9
and efforts to broaden the scope of DYKAT to non-stabilized carbon
nucleophiles now allows certain specific ‘less-stabilized’ partners to be
used in Pd-catalysed asymmetric (for example, Fig. 1c)7,13 and non-
enantioselective procedures8.
One difficulty in expanding the scope of DYKATs is that metal-
catalysed AAAs do not rely on a single mechanistic pathway4. In par-
ticular, there is a significant mechanistic distinction between stabilized
and non-stabilized nucleophiles (Fig. 1d)4,6,7,10. In the case of stabilized
nucleophiles, the key bond-making event occurs outside of the coor-
dination sphere of the metal, so that the stereochemistry is determined
when the nucleophile attacks a carbon atom of a p-allyl-Pd intermedi-
ate14. In contrast, when using non-stabilized nucleophiles bond forma-
tion probably occurs through reductive elimination of an intermediate
in which the nucleophile is bound to the metal centre15.
nucleophilesin such reactionsisdifficultand remainsakey challenge6–9
.
Herewereportacopper-catalyseddynamickineticasymmetrictrans-
formation using racemic substrates and alkyl nucleophiles. These
nucleophiles have a pKa of $50, more than 25 orders of magnitude
morebasicthanthenucleophilesthataretypicallyusedinsuchtrans-
formations. Organometallic reagents are generated in situ fromalkenes
byhydrometallationandgivehighlyenantioenrichedproductsunder
mild reaction conditions. The method is used to synthesize natural
products that possess activity against tuberculosis and leprosy, and
an inhibitor of para-aminobenzoate biosynthesis. Mechanistic studies
indicate that the reaction proceeds through a rapidly isomerizing
intermediate. We anticipate that this approach will bea valuable com-
plement to existing asymmetric catalytic methods.
Two strategies for generating single enantiomer compounds using
enantioselective catalysishave become widelyused: thefirststrategygen-
erates a chiral product by a selective reaction which introduces asymmetry
to a prochiral substrate (Fig. 1a). This prochiral approach has proven
importantinthedevelopmentofcatalyticasymmetricreactions, despite
the rather limited availability of prochiral substrates when compared
to chiral substrates. The second widely used strategy is to start from a
racemic mixture of chiral starting materials. Here the catalyst selec-
tively reacts with one of the two enantiomers allowing differentiation
(or resolution) of the enantiomers, but the yield is necessarily limited
to 50%, as the undesired enantiomer remains as starting material. An
efficientvariationofthissecondstrategyistocoupleenantiomerdiffer-
entiation with interconversion of the enantiomers, commonly called
dynamickineticresolution. Thisstrategyallowsyieldsofmorethan50%
from chiral starting materials2,3.
Catalytic asymmetric reactions with non-stabilized nucleophilesoften
use copper catalysts16,17. These reactions have proven much more dif-
ficult to study mechanistically than their Pd-catalysed counterparts, as
non-stabilized organometallic reagents are typically very reactive, hin-
dering the isolation of intermediates and characterization of pathways.
Copper-catalysedallylicalkylationsaregenerallyacceptedtooperatevia
a
O
O
Nu
Nu
Nu
R
X
R
Catalyst
Catalyst
Nu
Asymmetric allylic alkylation
Asymmetric conjugate addition
b
c
OCO2Me
H3COOC
COOCH3
O
O
Pd-catalyst
+
H3CO
OCH
3
M
+
PdLn
(
)
O
N
Base,
Pd-catalyst
N
N
O
N
Me
(
)
Nu
M
Nu
M
d
M
or
Nu
Transition-metal-catalysed asymmetric allylic alkylation (AAA) reac-
tions have proven to be powerful tools for creating new carbon–carbon
bonds4,9–11. Twentyyears ago, palladium-catalysedAAAreactionswere
reported in which racemic mixtures were converted to single enantio-
mer products with high yield and enantioselectivity (Fig. 1b)12. The use
of stabilized nucleophiles(pKa , 25)inmetal-catalysed AAAreactions
is now well established in catalysis and synthesis5,9. These reactions are
dynamic kinetic asymmetric transformations (DYKATs) as they form
one enantiomer of a new product from both enantiomers of a racemic
startingmaterial, and no‘resolved’ startingmaterialis recovered3,5. Mech-
Stabilized nucleophiles
Non-stabilized nucleophiles
Figure 1
|
Asymmetric allylic alkylation (AAA) procedures. a, Typical
examples of the prochiral approach to asymmetric catalysis, in which a species
selectively adds to one face of a prochiral substrate, exemplified (on the left)
by asymmetric conjugate addition and (on the right) by AAA. Nu, nucleophile;
R, generic substituent (or generic group); X, leaving group. b, A prototypical
Pd-catalysed DYKAT, where AAA with a stabilized carbon nucleophile
transforms a racemic mixture of starting materials into a single highly
enantioenriched product. c, An example of DKYAT in AAA using heterocyclic
nucleophiles with a pKa . 25. d, Transition-metal-catalysed allylic substitution
anistic studies reveal that Pd-catalysed DYKATs operate through a mechanisms with stabilized and non-stabilized nucleophiles.
1Department of Chemistry, Chemistry Research Laboratory, University of Oxford, 12 Mansfield Road, Oxford OX1 3TA, UK.
1 5 J A N U A R Y 2 0 1 5
| V O L 5 1 7 | N A T U R E | 3 5 1
Macmillan Publishers Limited. All rights reserved
©2015

RESEARCH LETTER
regiospecific anti-addition mechanisms, where selective rate-determining variety of reaction conditions. The use of Pd- and Ir-based catalysts did
SN29oxidativeadditions give copper(III) intermediates, andfast reduc- not lead to product 3a, but when Cu-catalysts were used, product 3a
tive eliminations form the final product16,17. A key distinction between
was obtained with varying degrees of enantioselectivity. The combination
of 3-chloro-cyclohexene 1d,phosphoramiditeC11, and (tBuCN)2Cu?OTf
Cu- and Pd-AAA reactions is a lack of an efficient s- to p- isomeriza-
tion18 in the case of copper, so that if reductive elimination is fast, a
in dichloromethane provided 3a in quantitative yield with 71% enan-
tiomericexcess(e.e.; Table1, entry 6). Variationof the coppercounterion
under these conditions showed that CuI gave the highest e.e. (89% e.e.,
entry 10) and chloroform gave the best results of the solvents exam-
ined (93% e.e., entry 11). Varying the temperature (entries 12 and 13)
showed that slightly higher enantioselectivity was obtained by cooling
the reaction to 0 uC (93% versus 95% e.e.) but we chose to perform the
rest of our studies at room temperature because of operational conve-
nience and under these conditions full conversion was always observed
overnight. Examination of bromide 1a under these conditions (entry14)
showed that the less reactive chloride 1c gave superior results.
The scope ofthe alkylcoupling partners wasinvestigatedusing avariety
of simple gaseous alkenes (giving products 3b–d) and unfunctiona-
lized primary alkenes (3e–g) (Fig. 2). High yields and consistently high
enantioselectivities were obtained. Functionalizedalkenes (3h–3r)were
alsoexamined:halogens, alkynes, aromaticrings,ethers, protectedalco-
hols, trifluoromethyl groups and preformed stereogeniccentres wereall
found to be compatible with the procedure. Electron rich alkenes such
asallylsilane(togive3i)andstyrenes (togive3m–o) alsoworkwell. We
also examined the use of 5-membered ringswiththisprocedure;wefound
that it is suitable for the direct preparation of highly enantioenriched
cyclopentenes (4a–c) from an easily prepared racemic 5-membered
allylchloride.
To probe practical aspects of this chemistry, scale-up reactions were
performed (Fig. 3a). Despite excellentconversion(.90%),theyieldof4b
ona15 mmolscalewaslowbecausetheproduct andtheresidualalkene
have similar polarities and boiling points, complicating isolation. Even
so, we were easily able to obtain more than 1.8 g of pure 4b (58% yield,
94% e.e.). When 1-hexene was used (to give 4a), we arbitrarily reduced
theratioofalkeneandSchwartzreagentused(to2and1.8equiv. respec-
tively) and observed improved yield (86% yield, 91% e.e., 1.2 g) relative
to using our standard test conditions (Fig. 2). Functionalized 3n con-
taining a bromide (Fig. 3b) was also synthesized on a preparative scale
(88% yield, 87% e.e., 0.75 g) and a subsequent Heck reaction to provide
cis-fusedtricycle5(88%yield)demonstratesthatthesereactions canbe
used to rapidly access complex molecular frameworks.
To demonstrate that this method allows rapid access to important
structural motifs, we performed asymmetric syntheses of biologically
active cyclopentene-containing natural products (Fig. 3c). This approach
provides short, straightforward, flexible syntheses which compare fav-
ourably to those previously reported25, and will facilitate the study of
structural analogues. Hydnocarpic acid (6) and chaulmoogric acid (7)
are leprosy treatments used in ancient and traditional medicine: in both
cases, the cyclopentenyl ring is a requirement for biological activity.
Hydnocarpic acid is believed to act by blocking the activity of biotin or
inhibiting microbial biotin synthesis26, while the activity of chaul-
moogric acid is probably due to incorporation into the cell wall lipids
of Mycobacterium leprae27. Recent reports show that chaulmoogric acid
and anthelminthicin C (8) significantly inhibit Mycobacterium tuber-
culosis growth with minimum inhibitory concentration values of 9.82
and 4.38 mM, respectively28. Anthelminthicin C has also recently been
found to inhibit para-aminobenzoic acid biosynthesis (pABAB)28; inhib-
itors of pABAB are important leads for the development of new anti-
biotics, as this pathway is not found in humans28.
p-allyl-Cu species will not form and reactions involving racemic start-
ing materials will lead to racemic products.
Copper-catalysedAAA on prochiral materials, where selective oxid-
ative addition occurs on one enantiotopic face of the substrate, can be
readily achieved10,16–18. However, the generally accepted Cu-catalysed
mechanismswouldseem to preclude efficient asymmetricreactions with
chiral racemic substrates, where addition would be controlled by the
stereogeniccentreofthe substrate, notthecatalyst18. Wenotethat race-
mization of enantiopure starting materials during Cu-catalysed allylic
alkylation has been reported, suggesting the possibility of generating
useful p-allyl-CuspeciesandDYKATs19. Buildingonthiswork, promis-
ing initial results have been obtained with Grignard reagents20, but
despiteextensiveexamination, andthe elucidationofinterestingmech-
anistic pathways, generally useful procedures have not emerged21,22
Based on the observation that alkylzirconium reagents undergo
highly enantioselective conjugate addition at ambient temperature23,24
.
,
whereasprocedureswithtraditionalnon-stabilizedcarbonnucleophiles
(based on Mg, Al, and Zn species) are normally performed at very low
temperatures16,17, we hypothesized that alkylzirconium reagents may
allow efficient AAA from racemic substrates. Here we describe AAA of
racemic cyclicallylchlorideswith alkylzirconocenes initiatedbycopper
catalysts. Highly enantioenriched products are obtained at room tem-
perature in convenient procedures, in which all of the reaction com-
ponents are commercially available.
We began by exploring the coupling of racemic 3-substituted cyclo-
hexenes1and the alkylzirconium species generated insitufrom 4-phenyl-
1-butene (2 in Table 1) and the Schwartz reagent (Cp2ZrHCl) under a
Table 1
|
Selected optimization experiments
Lg
Cp2ZrHCl, CH2Cl2;
Solvent
10 mol% copper
+
10 mol% ligand
(
)-1
2
3a
OCH3
tBu
tBu
tBu
OCH3
O
O
O
P
P
tBu
P
P
O
O
P
N
tBu
OCH3
tBu
O
tBu
tBu
B
A
C
OCH3
Entry
Leaving group
Ligand
Copper source
Solvent
Temp.
e.e. (%)
(uC)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Br (1a)
OP(O)(OEt)2 (1b)
O2CCF3 (1c)
Cl (1d)
A
A
A
A
B
C
C
C
C
C
C
C
C
C
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
CHCl3
25
25
25
25
25
25
25
25
25
25
25
0
4
(tBuCN)2CuNOTf
(tBuCN)2CuNOTf
(tBuCN)2CuNOTf
(tBuCN)2CuNOTf
(tBuCN)2CuNOTf
(tBuCN)2CuNOTf
CuCl/AgSbF6
CuCl
10
10
35
1
71
51
49
73
89
93
95
77
77
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
CuBr
CuI
CuI
CuI
CuI
CuI
On the basis of previously reported reactions it seemed likely that
DYKAT was occurring by one of the four following pathways: (1) the
reaction proceeds through a pseudo-meso- or prochiral p-allyl-Cu
intermediate19,20 similar to the Pd- or Ir-intermediates observed with
stabilized nucleophiles6; (2) a related scenario, in which the reaction
proceeds through rapidly interconverting intermediates derived from
1d (such as s-allyl-Cu species), one enantiomer of which undergoes a
selectivereaction5; (3)the twoenantiomersof 1dundergo two different
CHCl3
CHCl3
CHCl3
60
25
Br (1a)
Optimization was performed for the AAA of racemic materials bearing different leaving groups and
4-phenyl-1-butene, which was hydrometallated using the Schwartz reagent. Different copper sources,
ligands, solvents and temperatures were examined. The enantiomeric excess (e.e.) was determined by
HPLC. Full conversion was observed in all cases shown.
3 5 2
|
N A T U R E
|
V O L 5 1 7
|
1 5 J A N U A R Y 2 0 1 5
Macmillan Publishers Limited. All rights reserved
©2015

LETTER RESEARCH
Cl
CuI, ligand C, CHCl3
R
Cp2ZrHCl
CH2Cl2
R
R
ZrClCp2
n
n
3g
3b*
3c*
3d*
93%, 91% e.e.
3f
3e
56%, 93% e.e.
78%, 95% e.e.
67%, 94% e.e.
89%, 92% e.e.
84%, 95% e.e.
Ph
Cl
Ph
Si
Ph
3j
3h
3a
3i
3k
71%, 91% e.e.
76%, 91% e.e.
86%, 93% e.e.
52%, 93% e.e.
90%, 81% e.e.
O
Si
O
3m
Br
F3C
3l
3n
3o
3p*
66%, 87% e.e.
93%, 87% e.e.
78%, 87% e.e.
63%, 95% e.e.
45%, 94% e.e.
Si
O
Ph
O
Si
Cl
Ph
3q
3r*
97%, 91% e.e.
4a*
4b
4c*
Cl
55%, 89% e.e.
77%, 90% e.e.
72%, 90% e.e.
57%, 92% e.e.
Figure 2
|
Dynamic kinetic AAA with alkylzirconium nucleophiles
and enantiomeric excess of illustrated products. Yield refers to pure isolated
compounds; the enantiomeric excess was either determined directly by
high-performance liquid chromatography, or on the corresponding epoxides
by gas chromatography, using a chiral non-racemic stationary phase.
*Prepared using (S,S,S)-C (the enantiomer of the ligand shown in Table 1).
generated from alkenes. Top row, alkenes react with the Schwartz reagent
(Cp2ZrHCl) in CH2Cl2 to generate alkylzirconocene species in situ which
undergo AAA with racemic cyclic allylic chlorides. Cp, cyclopentadienyl.
Conditions: 2.5equiv. alkene, 2 equiv. Cp2ZrHCl, 10 mol% CuI, 10 mol%
ligand C, room temperature overnight, argon atmosphere. Lower rows, yield
reactions, both giving the same product (an enantio-convergent trans- conditions, and we are unable to rule out the possibility of such species
formation)21;or(4)1dracemizesduringthe reaction2 and AAA isselec- playing a role in the enantioselectivity of these reactions.
tive for one of the two enantiomers.
Examination of allyliodide 10 generated from 1d (by CuI and C in
To gain insight into the mechanism, we followed reactions in situ CDCl3) using 2-dimensional nuclear Overhauser effect NMR spectro-
using 1H NMR spectroscopy. During these experiments, we observed scopy (NOESY) shows that the Ha and Hc protons of 10 rapidly inter-
clean conversion of 1d to either 3a or 3f, and the formation of low, but convert (as indicated by cross-peaks between these proton signals, Fig. 4b,
fairly constant, concentrations of allyliodide 10 (Fig. 4a, also see Sup- top right). At ambient temperature 1d did not undergo observable
plementary Information). The initial drop in the concentration of 1d dynamic processes, but at 55 uC (see Supplementary Information) we
is an artefact of inconsistent NMR field inhomogeneities (‘shimming’) were able to detect interconversion between 1d and 10, demonstrating
at early stages of the reaction. We were unable to observe the formation that the allyliodide and allylchloride interconvert, but not fast enough
of alkyl-copper species (that would be produced by transmetallation to be detected on the NMR timescale at room temperature. Under these
from zirconium to copper) or s- or p-allyl-copper species under these conditions 1d was not observed to isomerize directly. Monitoring an
a Cl
b
Cl
Cp2ZrHCl (2 equiv.),
CH2Cl2
Cp2ZrHCl (2 equiv.),
CH2Cl2
+
+
CuI, ligand C, CHCl3
4b (1.8 g)
CuI, ligand C, CHCl3
Br
3n* (0.75 g)
88%, 87% e.e.
(
)
2.5 equiv.
2.0 equiv.
58%, 94% e.e.
Br
(
)
1.5 ml
15 mmol
Cl
2.5 equiv.
0.36 ml
3.2 mmol
Cp2ZrHCl (1.8 equiv.),
CH2Cl2
+
H
Pd(OAc)2, dppp, K2CO3
Toluene, reflux 28 h
4a (1.2 g)
86%, 91% e.e.
CuI, ligand C, CHCl3
(
)
Br
0.9 ml
9 mmol
H
5
88%, 87% e.e.
c
NaHCO3, DMSO, H2O;
DMP, CH2Cl2;
Cl
)
O
O
HO
O
HO
+
Br
8
NaClO2, NaH2PO4,
H2O, tBuOH, 79%
Hydnocarpic acid (6)
Anthelminthicin C (8)
OH
HO
(
HO
O
DCC, DMAP, CH2Cl2, 0 °C;
O
Amberlyst-15, MeOH, 90%
O
DMP, CH2Cl2;
O
Br
NaClO2, NaH2PO4, H2O, tBuOH
OH
Mg, I2, CuI (10 mol%)
THF, 0 °C, 76%
8
8
4d
42%, 92% e.e.
90%
Chaulmoogric acid (7)
9
Figure 3
|
Scale-up and applications of the method to access tricyclic
6 (by hydrolysis, then oxidation), or to 9 (by epoxide opening chain extension)
structures and natural products. a, Experiments on a preparative scale to give followed by oxidation to 7. Coupling 7 with racemic solketal, then acid
.1 g of cyclopentene products. b, Preparative-scale reaction on a 6-membered catalysed deprotection, gave anthelminthicin C (8). *Made using (S,S,S)-C (the
ring followed by Heck reaction to give cis-fused tricycle 5. c, Concise
asymmetric synthesis of hydnocarpic acid (6), chaulmoogric acid (7) and
anthelminthicin C (8). The syntheses start from commercially available
11-bromo-1-undecene to give bromo-substituted 4d which is converted to
enantiomer of the ligand shown in Table 1). DMP, Dess-Martin periodinane;
DCC, dicyclohexylcarbodiimide; dppp, 1,3-bis(diphenylphosphino)propane;
DMAP, 4-dimethylaminopyridine. Amberlyst 15 is a polymeric acid resin.
See Supplementary Information for full details.
1 5 J A N U A R Y 2 0 1 5
| V O L 5 1 7 | N A T U R E | 3 5 3
Macmillan Publishers Limited. All rights reserved
©2015

RESEARCH LETTER
a
0.4
0.38
0.36
Cl
1d
1d
3f
0.34
0.04
I
C5H11
CuI, ligand C, CDCl3
ZrCp2Cl
0.02
0
10
10
3,500
C5H11
0
500
1,000
1,500
2,000
2,500
3,000
Hc(I)
Time (s)
3f
Hb(I)
Ha(I)
b
H
bH
CuI
ligand C
c I
Cl
Ha
CDCl3
Hc(I)
10
1d
Ha(I)
Hb(I)
Hb
Hc
I
Ha
Hb(Br) Ha(Br)
Hc(Br)
Ha(I)
Hc(I)
Hb(I)
H
bH
bH
H
Ha
cBr
CuI
ligand C
CDCl3
Ha
c I
10
Hc(Br)
1a
Hc(I)
Hb
Hb
Ha(I)
Hb,a(Br)
Hb(I)
Hc
Hc
Br
Ha
I
Ha
CuLn
d
e
c
X
X
X
(R)
syn SN2′
Cu-catalysed
AAA
R
I
Cl
Cl
(S)
Slow
Slow
Slow
[rel]
Time
Observed
enantiomer
(>95:5 e.r.)
Ph
Me
H
0.20
0.15
0.10
0.05
Fast
I
N
Very slow
Minor
enantiomer
0.00
[ppm]
4.40
4.60
4.55
4.50
4.45
Figure 4
|
Mechanistic analysis. a, Reaction kinetics as monitored by in situ Cross peaks between Ha(Br)/Hc(I) and Ha(I)/Hc(Br), but not between
1H NMR spectroscopy. Formation of an allyliodide intermediate (10) is
observed. Plot at right shows concentration versus time. b, Examination of
intermediates 10 (generated from CuI and C in CDCl3) by NOESY: when
generated from 1d (top, proton chemical shifts on both axes, shift ranges as
follows: vertical, 4.5–6.3 p.p.m.; horizontal, 6.2–5.05 p.p.m.), allyliodide
protons labelled Ha and Hc exchange. When generated from 1a (bottom,
proton chemical shifts on both axes, shift ranges as follows: vertical, 4.1–
6.3p.p.m.; horizontal, 6.2–4.6 p.p.m.), rapid Ha/Hc isomerization is observed
within both 1a and 10. Rapid exchange between 10 and 1a is also observed.
Ha(Br)/Ha(I) or Hc(Br)/Hc(I), suggest interconversion occurs by an SN29
pathway. c, Proposed mechanism for the DYKAT: as 1d racemizes via 10, one
enantiomer of 1d undergoes selective AAA. d, Possible racemization
mechanism: metal-assisted syn-SN29 attack. e, 1H chemical shifts (vertical,
relative intensity, 0.00–0.25; horizontal, chemical shift, 4.65–4.00p.p.m.) of the
benzylic protons of C are consistent with a change in metal-ligand aggregation,
as a function of time (colours are for different spectra, taken at different times),
from the initial [LnCuI]n species.
AAAreactiontoform3afrom1dbyinsituNOESY(seeSupplementary
Taken together, these studies point to a mechanism (Fig. 4c) where
Information) demonstrates that allyliodide 10 undergoes dynamic iso- 1d racemizes, via 10, and selective AAA of one of the enantiomers of
merization during the AAA itself.
1d gives a product with high enantiomeric excess. When using copper
halides, higher enantioselectivities are obtained when the halide is a better
When allylbromide 1a is dissolved in CDCl3 at room temperature,
nodynamic processes areobserved byNOESY(seeSupplementaryInfor- nucleophile and leavinggroup (that is, I . Br . Cl, Table 1, entries 8–10),
facilitating interconversion, but the overall reaction rates with CuCl
observed (Fig. 4b, bottom right). Allyliodide 10 is formed, and both 1a and CuI are similar, suggesting that it is 1d that undergoes AAA.
and 10 rapidly isomerize (as indicated by cross peaks between Ha(Br)/ Racemization probably occurs through syn-SN29 displacement reac-
mation), butinthepresenceof CuIand C a variety offast exchangesare
Hc(Br) of1aand Ha(I)/Hc(I)of10). Rapidinterconversionof1aand10 tions: SN29 pathways are observed by NOESY, and anti-SN29 reactions
occurs as indicated by cross peaks between Ha(Br)/Hc(I) and Ha(I)/ would not change the absolute sense of stereochemistry of the allylic
Hc(Br). As cross peaks between Ha(Br)/Ha(I) and Hc(Br)/Hc(I) are not halides. Many allylic substitutions are syn-SN29 selective, and ‘ion-pair’
observed, interconversion appears to exclusively occur by SN29 mech- nucleophiles undergoselectivesyn-SN29reactionsthroughcyclictransi-
anisms, although SN2 exchange may occur through minor pathways tion structures29. Our studies are consistent with the copper-mediated
not observable on the NMR timescale.
syn- SN29 pathway shown in Fig. 4d.
3 5 4
|
N A T U R E
|
V O L 5 1 7
|
1 5 J A N U A R Y 2 0 1 5
Macmillan Publishers Limited. All rights reserved
©2015

LETTER RESEARCH
6. Pfaltz, A. & Lautens, M. in Comprehensive Asymmetric Catalysis ii Vol. 2 (eds
Jacobsen, E. N., Pfaltz, A. & Yamamoto, H.) Ch. 24, 833–884 (Springer, 1999).
7. Trost, B. M. & Thaisrivongs, D. A. Strategy for employing unstabilized nucleophiles
in palladium-catalyzed asymmetric allylic alkylations. J. Am. Chem. Soc. 130,
14092–14093 (2008).
8. Sha, S. C., Zhang, J. D., Carroll, P. J. & Walsh, P. J. Raising the pKa limit of ‘‘soft’’
nucleophiles in palladium-catalyzed allylic substitutions: application of
diarylmethane pronucleophiles. J. Am. Chem. Soc. 135, 17602–17609 (2013).
9. Lu, Z. & Ma, S. Metal-catalyzed enantioselective allylation in asymmetric synthesis.
Angew. Chem. Int. Edn 47, 258–297 (2008).
10. Geurts, K., Fletcher, S. P., van Zijl, A. W., Minnaard, A. J. & Feringa, B. L. Copper-
catalyzed asymmetric allylic substitution reactions with organozinc and Grignard
reagents. Pure Appl. Chem. 80, 1025–1037 (2008).
11. Teichert, J. F. & Feringa, B. L. Phosphoramidites: privileged ligands in asymmetric
catalysis. Angew. Chem. Int. Edn 49, 2486–2528 (2010).
12. Trost, B. M. & Bunt, R. C. Asymmetric induction in allylic alkylations of
3-(acyloxy)cycloalkenes. J. Am. Chem. Soc. 116, 4089–4090 (1994).
13. Misale, A., Niyomchon, S., Luparia, M. & Maulide, N. Asymmetric palladium-
catalyzed allylic alkylation using dialkylzinc reagents: a remarkable ligand effect.
Angew. Chem. Int. Edn 53, 7068–7073 (2014).
14. Trost, B. M. & Verhoeven, T. R. Allylic substitutions with retention of
stereochemistry. J. Org. Chem. 41, 3215–3216 (1976).
15. Matsushita, H. & Negishi, E. Anti-stereospecificity in the palladium-catalyzed
reactions of alkenyl-metal oraryl-metal derivatives with allylic electrophiles. Chem.
Commun. 160–161 (1982).
We followed the enantiomeric excess of 3f formed as a function of
time using CuCl, CuI and CuOTf. When using CuOTf, the reaction goes
to completion much faster (,30 min) than with copper halides (over-
night) and we did not detect any change in e.e. with time, so we tenta-
tivelyspeculatethat CuOTfandCu-halide catalysedreactions (Table1,
entry 6) aremechanistically distinct. Reactions usingCuClshow a decrease
of e.e. with increasing time (,82% e.e., 10 min; 68% e.e., 2 h; 54% e.e. at
completion) and wespeculate thatwhen CuCl is used, the rapidly react-
ing enantiomer of 1d is consumed, and not replenished by fast race-
mization. The reaction with the minor enantiomer therefore becomes
significant and reduces the e.e. of 3f as time progresses. We observed
non-racemic 1d (at least 11% e.e. at 2 h) during the reaction as judged
by gas chromatography analysis (see Supplementary Information). In
the case of CuI the initial reaction rate is slow, and only very small
amounts of 3f are formed after 10 and 30 min. Here the e.e. increases
withtime(,85%e.e., 10 min;90%e.e., 2 h;95%e.e. atcompletion)sug-
gesting that a more enantioselective system is generated as the reaction
progresses.
At least part of the role of CuI is to racemize 1d via 10, but the iden-
tity of the proposed highly enantioselective catalyst is unclear. Both
CuCl and CuI give lower initial values of e.e. than are observed at later
stages of the CuI-catalysed reaction. It may be that copper aggregates
containing both ‘Cl’ and ‘I’ are more selective than those containing
only one halogen. An experiment performed using 10% C 1 5% CuCl
1 5% CuI ambiguously gave 3f with 90% e.e. (rather than 95% e.e.) at
completion. Alternatively, salt effects may increase the e.e. of 3f as time
progresses by favouring the formation of more highly enantioselective
copper-ligand complexes; the reaction produces Cp2ZrCl2 which may
affect the relative solubility of other reaction components.
16. Harutyunyan, S. R., den Hartog, T., Geurts, K., Minnaard, A. J. & Feringa, B. L.
Catalytic asymmetric conjugate addition and allylic alkylation with Grignard
reagents. Chem. Rev. 108, 2824–2852 (2008).
¨
17. Alexakis, A., Backvall, J. E., Krause, N., Pamies, O. & Dieguez, M. Enantioselective
copper-catalyzed conjugate addition and allylic substitution reactions. Chem. Rev.
108, 2796–2823 (2008).
18. Langlois, J. B. & Alexakis, A. in Topics in Organometallic Chemistry Vol. 38, Transition
Metal Catalyzed Enantioselective Allylic Substitution in Organic Synthesis (ed.
Kazmaier, U.) 235–268 (Springer, 2012).
19. Norinder, J. & Ba¨ckvall, J. E. Dynamic processes in the copper-catalyzed
substitution of chiral allylic acetates leading to loss of chiral information. Chem.
Eur. J. 13, 4094–4102 (2007).
20. Langlois, J. B. & Alexakis, A. Dynamic kinetic asymmetric transformation in copper
catalyzed allylic alkylation. Chem. Commun. 3868–3870 (2009).
21. Langlois, J. B., Emery, D., Mareda, J. & Alexakis, A. Mechanistic identification and
improvement of a direct enantioconvergent transformation in copper-catalyzed
asymmetric allylic alkylation. Chem. Sci. 3, 1062–1069 (2012).
22. Giacomina, F. & Alexakis, A. Construction of enantioenriched cyclic compounds by
asymmetric allylic alkylation and ring-closing metathesis. Eur. J. Org. Chem. 2013,
6710–6721 (2013).
1H NMRspectroscopystudies suggestthatcopper-ligandaggregates
change during the reaction: we observe a shift in the benzylic proton
signal of ligand C, from being characteristic of [LnCuI]n complexes,
towards newspeciesastime progresses(Fig. 4e, seealsoSupplementary
Information). Cu-halide phosphoramidite aggregates exhibit complex
dynamic equilibria between several solution- and solid-phase species,
which are sensitive to solvent as well as halide and salt effects30, and so
determining the actual composition of the copper species generated in
the reaction will require further work. How the Cu-ligand complex selects
for one enantiomer of 1d, and how the Cu-ligand complex interacts
with the alkylzirconium species, are also both currently unclear.
We have described a copper-catalysed enantioselective addition of
alkyl zirconiumreagentstoracemic cyclic allylic chlorides. The reaction
uses readily available starting materials and catalysts, tolerates a variety
of functional groups and operates under convenient conditions. The
reactions can be performed on gram scales, and we have applied the
reaction to the asymmetric synthesis of biologically active cyclopen-
tene natural products. Mechanistic studies suggest that this dynamic
kinetic asymmetric transformation operates via a rapidly interconvert-
ing intermediate, racemizing the substrate, and the formation of a more
highlyselectivecopper-catalystinsitu. Thisreactionisexpectedtocom-
plement the well-established palladium and iridium-catalysed methods
with stabilized nucleophiles. Additionally, we anticipate that the mech-
anistic insight will inspire future studies in the field.
23. Maksymowicz, R. M., Roth, P. M. C. & Fletcher, S. P. Catalytic asymmetric
carbon-carbon bond formation using alkenes as alkylmetal equivalents. Nature
Chem. 4, 649–654 (2012).
24. Sidera, M., Roth, P. M. C., Maksymowicz, R. M. & Fletcher, S. P. Formation of
quaternary centers by copper-catalyzed asymmetric conjugate addition of
alkylzirconium reagents. Angew. Chem. Int. Edn 52, 7995–7999 (2013).
25. Seemann, M., Scho¨ller, M., Kudis, S. & Helmchen, G. Syntheses of enantiomerically
pure cyclopent-2-ene-1-carboxylic acid and (cyclopent-2-enyl)acetic acid by
enantioselective palladium-catalyzed allylic alkylations — synthesis of
enantiomerically pure (-)-chaulmoogric acid. Eur. J. Org. Chem. 2122–2127
(2003).
26. Jacobsen, P. L. & Levy, L. Mechanism by which hydnocarpic acid inhibits
mycobacterial multiplication. Antimicrob. Agents Chemother. 3, 373–379 (1973).
27. Cabot, M. C. & Goucher, C. R. Chaulmoogric acid-assimilation into the complex
lipids of mycobacteria. Lipids 16, 146–148 (1981).
28. Wang, J. F. et al. Antituberculosis agents and an inhibitor of the para-aminobenzoic
acid biosynthetic pathway from Hydnocarpus anthelminthica seeds. Chem.
Biodivers. 7, 2046–2053 (2010).
29. Streitwieser, A., Jayasree, E. G., Hasanayn, F. & Leung, S. S. H. A theoretical study of
SN29 reactions of allylic halides: role of ion pairs. J. Org. Chem. 73, 9426–9434
(2008).
30. Zhang, H. & Gschwind, R. M. Structure identification of precatalytic copper
phosphoramidite complexes in solution. Angew. Chem. Int. Edn 45, 6391–6394
(2006).
Supplementary Information is available in the online version of the paper.
Received 27 June; accepted 13 November 2014.
Acknowledgements We acknowledge financial support from the EPSRC
(EP/H003711/1, a Career Acceleration Fellowship to S.P.F.). B. Odell and T. Claridge
are thanked for assistance with the NMR experiments.
1. Jacobsen, E. N., Pfaltz, A. & Yamamoto, H. (eds) Comprehensive Asymmetric
Catalysis: Suppl. 2 (Springer, 2004).
2. Huerta, F. F., Minidis, A. B. E. & Ba¨ckvall,J. E. Racemisationinasymmetric synthesis.
Dynamic kinetic resolution and related processes in enzyme and metal catalysis.
Chem. Soc. Rev. 30, 321–331 (2001).
Author Contributions H.Y., E.R. and M.S. performed the experiments. All authors
contributed to designing, analysing and discussing the experiments; S.P.F. conceived
the work and guidedthe research. S.P.F. wrote themanuscriptwith assistance fromH.Y.
All authors contributed to discussing and editing the manuscript.
3. Vedejs, E. & Jure, M. Efficiency in nonenzymatic kinetic resolution. Angew. Chem.
Int. Edn 44, 3974–4001 (2005).
4. Trost, B. M. & VanVranken, D. L. Asymmetric transition metal-catalyzed allylic
alkylations. Chem. Rev. 96, 395–422 (1996).
5. Trost, B. M. & Fandrick, D. R. Palladium-catalyzed dynamic kinetic asymmetric
allylic alkylation with the DPPBA ligands. Aldrichim. Acta 40, 59–72 (2007).
Author Information Reprints and permissions information is available at
www.nature.com/reprints. The authors declare competing financial interests: details
are available in the online version of the paper. Readers are welcome to comment on
the online version of the paper. Correspondence and requests for materials should be
addressed to S.P.F. (stephen.fletcher@chem.ox.ac.uk).
1 5 J A N U A R Y 2 0 1 5
| V O L 5 1 7 | N A T U R E | 3 5 5
Macmillan Publishers Limited. All rights reserved
©2015