Diastereoselective nickel-catalyzed reductive Aldol cyclizations using diethylzinc as the stoichiometric reductant: Scope and mechanistic insight

Article abstract of DOI:10.1021/ja0775624

In the presence of diethylzinc as a stoichiometric reductant, Ni(acac) ₂ functions as an efficient precatalyst for the reductive aldol cyclization of α,β-unsaturated carbonyl compounds tethered to a ketone electrophile through an amide or an ester linkage. The reactions are tolerant of a wide range of substitution at both α,β-unsaturated carbonyl and ketone components and proceed smoothly to furnish β-hydroxylactams and β-hydroxylactones with generally high diastereoselectivities. A series of experiments, including deuterium-labeling studies, was carried out in an attempt to gain some insight into the possible reaction mechanisms that might be operative.

Full text of DOI:10.1021/ja0775624

Published on Web 05/16/2008
Diastereoselective Nickel-Catalyzed Reductive Aldol
Cyclizations Using Diethylzinc as the Stoichiometric
Reductant: Scope and Mechanistic Insight
Pekka M. Joensuu,† Gordon J. Murray,† Euan. A. F. Fordyce,† Thomas Luebbers,‡
and Hon Wai Lam*,†
School of Chemistry, UniVersity of Edinburgh, Joseph Black Building, The King’s Buildings,
West Mains Road, Edinburgh, EH9 3JJ, United Kingdom, and F. Hoffmann-La Roche Ltd.,
Grenzacherstrasse 124, CH-4070 Basel, Switzerland
Received October 1, 2007; E-mail: h.lam@ed.ac.uk
Abstract: In the presence of diethylzinc as a stoichiometric reductant, Ni(acac)2 functions as an efficient
precatalyst for the reductive aldol cyclization of R,ꢀ-unsaturated carbonyl compounds tethered to a ketone
electrophile through an amide or an ester linkage. The reactions are tolerant of a wide range of substitution
at both R,ꢀ-unsaturated carbonyl and ketone components and proceed smoothly to furnish ꢀ-hydroxylactams
and ꢀ-hydroxylactones with generally high diastereoselectivities. A series of experiments, including
deuterium-labeling studies, was carried out in an attempt to gain some insight into the possible reaction
mechanisms that might be operative.
Introduction
species such as alkynes, 1,3-dienes, 1,3-enynes, and allenes with
an array of electrophiles that includes aldehydes, ketones,
Nickel-catalyzed reductive coupling and cyclization reactions
have recently emerged as powerful methods for accessing
numerous polyfunctionalized products.1,2 Typically operating
under mild conditions, these reactions enable rapid increases
in molecular complexity from relatively simple starting materi-
als, with often high levels of diastereo- and enantiocontrol. Using
reductants such as triethylborane, diethylzinc, and silanes, a
variety of reactions have been developed that couple unsaturated
imines, and epoxides.2
Another important class of reaction partners for nickel-catalyzed
reductive coupling/cyclization reactions are R,ꢀ-unsaturated car-
bonyl compounds. With these substrates, the majority of examples
described to date result in carbon-carbon bond formation at the
ꢀ-position. For example, Montgomery and co-workers have
described reductive cyclizations of electron-deficient alkenes with
a range of tethered unsaturation where coupling at the ꢀ-position
was the predominant pathway,1t and Tanaka and co-workers have
reported similar carbocyclizations of enones onto vinyl sulfoxides.1i
In addition, the Montgomery group has recently described inter-
molecular [3 + 2] reductive cycloadditions of enals with alkynes1d
and intermolecular reductive couplings of enones with alkynes,1b
where coupling also occurs at the ꢀ-carbon of the R,ꢀ-unsaturated
carbonyl component.
In contrast, corresponding reactions where coupling occurs
at the R-position of an R,ꢀ-unsaturated carbonyl compound have
been rare;3 although this deficiency has now been partially
addressed by the recent report of nickel-catalyzed reductive aldol
reactions between acrylate esters and aldehydes, where aryl
halides were found to be essential reaction ingredients.4 Given
the wealth of literature that exists in the area of catalytic
reductive aldol couplings and cyclizations of R,ꢀ-unsaturated
carbonyl substrates mediated by other metals,5 the relative dearth
of corresponding nickel-catalyzed transformations is surprising.
The development of such processes would considerably expand
the repertoire of synthetic methods available to organonickel
† University of Edinburgh.
‡ F. Hoffmann-La Roche Ltd.
(1) For recent, representative examples, see: (a) Chaulagain, M. R.;
Sormunen, G. J.; Montgomery, J. J. Am. Chem. Soc. 2007, 129, 9568–
9569. (b) Herath, A.; Thompson, B. B.; Montgomery, J. J. Am. Chem.
Soc. 2007, 129, 8712–8713. (c) Yang, Y.; Zhu, S.-F.; Duan, H.-F.;
Zhou, C.-Y.; Wang, L.-X.; Zhou, Q.-L. J. Am. Chem. Soc. 2007, 129,
2248–2249. (d) Herath, A.; Montgomery, J. J. Am. Chem. Soc. 2006,
128, 14030–14031. (e) Samih ElDouhaibi, A.; Lozanov, M.; Mont-
gomery, J. Tetrahedron 2006, 62, 11460–11469. (f) Kimura, M.; Ezoe,
A.; Mori, M.; Iwata, K.; Tamaru, Y. J. Am. Chem. Soc. 2006, 128,
8559–8568. (g) Moslin, R. M.; Miller, K. M.; Jamison, T. F.
Tetrahedron 2006, 62, 7598–7610. (h) Knapp-Reed, B.; Mahandru,
G. M.; Montgomery, J. J. Am. Chem. Soc. 2005, 127, 13156–13157.
(i) Maezaki, N.; Sawamoto, H.; Ishihara, H.; Tanaka, T. Chem.
Commun. 2005, 3992–3994. (j) Miller, K. M.; Jamison, T. F. Org.
Lett. 2005, 7, 3077–3080. (k) Luanphaisarnnont, T.; Ndubaku, C. O.;
Jamison, T. F. Org. Lett. 2005, 7, 2937–2940. (l) Kimura, M.; Miyachi,
A.; Kojima, K.; Tanaka, S.; Tamaru, Y. J. Am. Chem. Soc. 2004, 126,
14360–14361. (m) Patel, S. J.; Jamison, T. F. Angew. Chem., Int. Ed.
2004, 43, 3941–3944. (n) Mahandru, G. M.; Liu, G.; Montgomery, J.
J. Am. Chem. Soc. 2004, 126, 3698–3699. (o) Sato, Y.; Saito, N.;
Mori, M. J. Org. Chem. 2002, 67, 9310–9317. (p) Sato, Y.; Sawaki,
R.; Saito, N.; Mori, M. J. Org. Chem. 2002, 67, 656–662. (q) Kimura,
M.; Ezoe, A.; Tanaka, S.; Tamaru, Y. Angew. Chem., Int. Ed. 2001,
40, 3600–3602. (r) Shibata, K.; Kimura, M.; Shimizu, M.; Tamaru,
Y. Org. Lett. 2001, 3, 2181–2183. (s) Kimura, M.; Fujimatsu, H.; Ezoe,
A.; Shibata, K.; Shimizu, M.; Matsumoto, S.; Tamaru, Y. Angew.
Chem., Int. Ed. 1999, 38, 397–400. (t) Montgomery, J.; Oblinger, E.;
Savchenko, A. V. J. Am. Chem. Soc. 1997, 119, 4911–4920.
(2) For a review, see: Montgomery, J. Angew. Chem., Int. Ed. 2004, 43,
3890–3908.
(3) The Zhou and Tamaru groups have reported isolated examples of both
inter- and intramolecular reductive couplings of dienyl esters with
aldehydes, where coupling occurred R to the ester carbonyl. However, it
is the diene functionalities of these substrates that are essential for reaction
to proceed, rather than the R,ꢀ-unsaturated ester. See refs 1c, 1f, and 1q.
(4) Chrovian, C. C.; Montgomery, J. Org. Lett. 2007, 9, 537–540.
9
7328 J. AM. CHEM. SOC. 2008, 130, 7328–7338
10.1021/ja0775624 CCC: $40.75
2008 American Chemical Society

Nickel-Catalyzed Reductive Aldol Cyclizations
A R T I C L E S
chemistry and would provide complementary approaches for
executing reductive aldol reactions.
13), and phenyl (entries 8-11) ketones all participating suc-
cessfully. Variation of the nitrogen protecting group from benzyl
(entries 1-5 and 9-11) to para-methoxyphenyl (entries 6, 7,
12, and 13) or ortho-methoxyphenyl (entry 8) was also
permitted, and a pre-existing stereocenter in substrates 3l and
3m led to high levels of internal asymmetric induction to provide
bicyclic δ-lactams 4l and 4m, respectively (entries 12 and 13).
Overall, the results presented in Table 1 are comparable to those
reported previously using Co(acac)2 ·2H2O as the precatalyst.6
Simple ꢀ-unsubstituted acrylamides are less competent sub-
strates in these reactions, due to their propensity to undergo
alkylatiVe aldol cyclization9 in preference to reductive cycliza-
tion. For example, acrylamide 5 provided the desired product 6
in only 17% yield, with the major product obtained being 7,
derived from ethyl conjugate addition and aldol cyclization (eq
2). This result contrasts with that obtained previously using
Co(acac)2 ·2H2O as the precatalyst in place of Ni(acac)2, where
the reductive cyclization product 6 was obtained in 88% yield
and none of the alkylative product 7 was detected.6 However,
substitution at the R-position of the acrylamide re-established
reductive cyclization as the dominant pathway, as illustrated
by the cyclization of methacrylamide 8 to provide lactam 9
containing two contiguous quaternary centers in 82% yield (eq
3). Efforts to extend the scope of these reactions to more highly
substituted R,ꢀ-unsaturated amides were partially successful;
although tiglic amide 10 did not undergo cyclization (furnishing
elimination product 11 instead) (eq 4), R,ꢀ-unsaturated amide
12 provided lactam 13 in 54% yield (eq 5). It should be noted
that attempted cyclizations of substrates 8, 10, and 12 using
Co(acac)2 ·2H2O6 were completely unsuccessful.
With these considerations in mind, we recently initiated a
program targeted at the development of new reductive aldol
cyclizations and couplings catalyzed by nickel (among other
metals), where emphasis was placed on less reactive substrate
classes such as ꢀ-substituted R,ꢀ-unsaturated esters, ꢀ-substi-
tuted R,ꢀ-unsaturated amides, and ketones that might otherwise
prove difficult to couple using existing catalytic methodologies.5
As part of these studies, we recently disclosed that cobalt salts
function as effective precatalysts in the diethylzinc-mediated
reductive aldol cyclization of substrates of general structure 1,
containing an R,ꢀ-unsaturated carbonyl moiety tethered to a
ketone through an amide (eq 1).6,7 These reactions delivered
ꢀ-hydroxylactams 2 in 47 to >99% yield, and with generally
high levels of diastereoselection (8:1 to >19:1).
In this Article, we demonstrate that the combination of
Ni(acac)2 and diethylzinc not only functions in much the same
capacity but also promotes the reactions of substrates that are
either completely inert to the cobalt conditions6 or react with
much lower efficiencies and/or reproducibilities. This signifi-
cantly expanded substrate scope includes cyclization precursors
containing ester tethers, R-substituted R,ꢀ-unsaturated amides,
ꢀ,ꢀ-disubstituted R,ꢀ-unsaturated amides, and ꢀ-ester-substituted
R,ꢀ-unsaturated amides, allowing access to a correspondingly
broader array of mono- and bicyclic products in diastereose-
lective fashion.
Results and Discussion
Reaction Scope. Our investigations into nickel-catalyzed
reductive aldol cyclizations commenced with reactions that
provided six-membered ꢀ-hydroxylactams as the products (Table
1). Using 5 mol % of Ni(acac)2 and 2 equiv of Et2Zn, a range
of substrates 3a-3m underwent smooth cyclization to furnish
4-hydroxypiperidin-2-ones 4a-4m. A variety of substituents
at the R,ꢀ-unsaturated amide component that encompassed linear
and branched alkyl (entries 1-3, 8-10, and 12), aromatic
(entries 4, 7, and 13), and heteroaromatic (entries 5, 6, and 13)
groups was accommodated to provide cyclized products in good
to excellent yields and with generally high diastereoselectivities
(g9:1 by 1H NMR analysis of the unpurified reaction mixtures).8
The reactions were also tolerant of variation of the ketone
component, with alkyl (entries 1-7), cycloalkyl (entries 12 and
Substrates 14a and 14b afforded bicyclic products 16a and
16b, respectively, as a result of the presumed tertiary zinc
(5) For a seminal reference, see: (a) Revis, A.; Hilty, T. K. Tetrahedron
Lett. 1987, 28, 4809–4812. For an extensive collection of reports of
catalytic reductive aldol reactions, see references cited within: (b) Ngai,
M.-Y.; Kong, J.-R.; Krische, M. J. J. Org. Chem. 2006, 72, 1063–
1072. For relevant reviews, see: (c) Nishiyama, H.; Shiomi, T. Top.
Curr. Chem. 2007, 279, 105–137. (d) Jang, H.-Y.; Krische, M. J. Eur.
J. Org. Chem. 2004, 3953–3958. (e) Jang, H.-Y.; Krische, M. J. Acc.
Chem. Res. 2004, 37, 653–661. (f) Huddleston, R. R.; Krische, M. J.
Synlett 2003, 12–21. (g) Chiu, P. Synthesis 2004, 2210–2215. (h)
Motherwell, W. B. Pure Appl. Chem. 2002, 74, 135–142.
(8) The products 4a, 4b, 4d, 4e, 4g, 4i, 4j, 4l, and 4m have been described
previously (see ref 6). Product 4h has been described previously: Lam,
H. W.; Murray, G. J.; Firth, J. D. Org. Lett. 2005, 7, 5743–5746. The
relative stereochemistries of the remaining products in Table 1 were
assigned by analogy
(9) For examples of alkylative aldol cyclizations (sequential 1,4-conjugate
addition-intramolecular aldol reactions), see: (a) Caube, D. F.; Gipson,
J. D.; Krische, M. J. J. Am. Chem. Soc. 2003, 125, 1110–1111. (b)
Bocknack, B. M.; Wang, L.-C.; Krische, M. J. Proc. Natl. Acad. Sci.
U.S.A. 2004, 101, 5421–5424. (c) Agapiou, K.; Cauble, D. F.; Krische,
M. J. J. Am. Chem. Soc. 2004, 126, 4528–4529.
(6) Lam, H. W.; Joensuu, P. M.; Murray, G. J.; Fordyce, E. A. F.; Prieto,
O.; Luebbers, T. Org. Lett. 2006, 8, 3729–3732.
(7) For intermolecular variants of these reactions, see: Lumby, R. J. R.;
Joensuu, P. M.; Lam, H. W. Org. Lett. 2007, 9, 4367–4370.
9
J. AM. CHEM. SOC. VOL. 130, NO. 23, 2008 7329

A R T I C L E S
Joensuu et al.
Table 1. Nickel-Catalyzed Reductive Aldol Cyclization Furnishing 4-Hydroxypiperidinonesa
a Reactions were conducted using 0.20 mmol of substrate in THF (1.5 mL) and hexane (0.4 mL). b Determined by 1H NMR analysis of the
unpurified reaction mixtures. c Isolated yield of major diastereomer. d Here dr ) (major isomer):(Σ other isomers). PMP ) para-methoxyphenyl. OMP
) ortho-methoxyphenyl.
Scheme 1. Sequential Reductive Cyclization-Lactonization
alkoxides 15a and 15b produced on initial reductive cyclization
undergoing lactonization onto the pendant ester (Scheme 1).
Notably, none of the possible alternative regiosomeric products
17a and 17b, resulting from cyclization R to the ester rather
than the amide, were detected in the product mixtures. These
examples highlight the high chemoselectivity of these transfor-
mations. Again, it should be noted that attempted cyclizations
of 14a and 14b using Co(acac)2 ·2H2O6 in place of Ni(acac)2
were unsuccessful.
7), aromatic (eq 8), and ester groups (eq 9) successfully
undergoing the reaction. In the case of substrate 18d, <10% of
lactonization was observed (compare with Scheme 1).
These reactions are also applicable to the synthesis of five-
membered lactams, as illustrated by eqs 6–9. These results once
again demonstrate the tolerance of these reactions to variation
of the substituent at the ꢀ-position of the R,ꢀ-unsaturated
carbonyl component, with substrates containing alkyl (eqs 6 and
The cyclizations of oxygen-linked precursors were studied
next (Table 2). In contrast to the use of Co(acac)2 ·2H2O as the
9
7330 J. AM. CHEM. SOC. VOL. 130, NO. 23, 2008

Nickel-Catalyzed Reductive Aldol Cyclizations
A R T I C L E S
Table 2. Nickel-Catalyzed Reductive Aldol Cyclization Furnishing
ꢀ-Hydroxylactonesa
Mechanistic Considerations. Numerous mechanistic possibili-
ties (often speculative) have been advanced to explain the
outcome of nickel-catalyzed reductive couplings and cycliza-
tions.2 The generally accepted active oxidation state of nickel
in these reactions is 0, and we assume that Ni(0), generated by
the well-known reduction of Ni(acac)2 with Et2Zn,2 is also
responsible for catalysis of the reactions described herein.
Nevertheless, with Ni(acac)2 as a precatalyst, there exists the
possibility that the active catalyst is actually a Ni(II) species.
We therefore repeated two representative reactions using
Ni(COD)2 in place of Ni(acac)2, with the expectation that, if
Ni(0) is the true active species, cyclization would occur. In the
event, cyclization was observed in both cases (eqs 12 and 14),
thus supporting the notion that Ni(0) is the catalytically
competent species. However, the efficiencies of these reactions
were markedly different. In the reaction of cinnamic amide 3d,
Ni(COD)2 was almost identical to Ni(acac)2, providing lactam
4d in high diastereoselectivity (>19:1) and high conversion (eq
12, compare with Table 1, entry 4). On the other hand,
Ni(COD)2 functioned poorly compared to Ni(acac)2 with ester-
tethered substrate 20d, providing the product 21d in <20%
conversion (eq 14, compare with Table 2, entry 4). Although
the catalytic efficiencies of Ni(COD)2 and Ni(acac)2 have
sometimes proven to be essentially indistinguishable in other
classes of reactions involving main group organometallics, such
as the cross-coupling of cyclic anhydrides with organozinc
reagents developed by the Rovis group,12 the outcome of eq
14, along with the reductive aldol reactions developed by the
Montgomery group,4 demonstrates that this is not always the
case. Given the increasing awareness that olefins, present either
as ligands in precatalyst sources or as exogenous additives, do
not always act as innocent bystanders in transition-metal-
catalyzed reactions,13 these results are not entirely surprising,
and we suspect that the cyclooctadiene ligands present in
Ni(COD)2 are responsible for the different behavior of Ni(acac)2
and Ni(COD)2.
a Reactions were conducted using 0.20 mmol of substrate in THF (1.5
mL) and hexane (0.4 mL). b Determined by 1H NMR analysis of the
unpurified reaction mixtures. c Isolated yield of major diastereomer.
d Isolated as an inseparable 5.5:1 mixture of diastereomers. e Accompa-
nied by <5% of alkylative cyclization product as an inseparable impurity.
Cited yield of 21f has been adjusted to reflect this impurity. f Accompa-
nied by ca. 10% of alkylative cyclization product as an inseparable
impurity, making determination of the diastereomeric ratio of 21g dif-
ficult. Cited yield of 21g has been adjusted to reflect this impurity. g Here
dr ) (major isomer):(Σ other isomers).
precatalyst, which generally provides negligible (<5%) conver-
sions with ester-tethered substrates, the Ni(acac)2/Et2Zn com-
bination was found to give productive cyclizations to afford a
range of ꢀ-hydroxylactones 21a-21j in good to excellent yields
with high diastereoselectivities.10,11 Furthermore, compared to
their amide-tethered counterparts (Table 1), these reactions
display similar tolerance of substitution at both the R,ꢀ-
unsaturated carbonyl component and the ketone. With substrates
20f and 20g, small quantities of the corresponding alkylative
aldol cyclization products were observed (entries 6 and 7). A
pre-existing stereocenter in substrates 20h-20j again led to high
levels of internal asymmetric induction (entries 8-10).
An interesting electronic effect was observed with cinnamic
substrates. While p-methoxy-substituted precursor 20c gave only
the reductive aldol product 21c in 76% yield (Table 2, entry
3), less electron-rich substrates 22a and 22b provided significant
quantities of alkylative aldol products 23a and 23b, respectively,
as 1:1 inseparable mixtures of diastereomers (eqs 10 and 11).9
It appears that, as the aromatic group becomes more electron-
deficient, the degree of alkylative aldol cyclization becomes
more significant.
The cyclizations of substrates 3d and 20d using (DME)NiBr2
as the precatalyst provided very similar outcomes to those
obtained using Ni(acac)2 (eqs 13 and 15, compare with Table
1, entry 4 and Table 2, entry 4, respectively). These results lend
further support to the hypothesis that the inferiority of Ni(COD)2
compared with Ni(acac)2 for ester-tethered substrates is due to
a negative effect of the cyclooctadiene ligands in Ni(COD)2
(10) Compare with results obtained using copper bisphosphine catalysts
in conjunction with siloxane reductants: Lam, H. W.; Joensuu, P. M.
Org. Lett. 2005, 7, 4225–4228.
(11) The products 21a-21d have been described previously (see ref 10),
and the relative stereochemistries of the remaining products in Table
2 were assigned by analogy.
(12) Bercot, E. A.; Rovis, T. J. Am. Chem. Soc. 2005, 127, 247–254.
(13) Johnson, J. B.; Rovis, T. Angew. Chem., Int. Ed. 2008, 47, 840–871.
9
J. AM. CHEM. SOC. VOL. 130, NO. 23, 2008 7331

A R T I C L E S
Joensuu et al.
Scheme 2. A Possible Reaction Mechanism Invoking Metallacycle
Participation
be protonated upon workup to give the product), ethylene, and
Ni(0), which would re-enter the catalytic cycle. Within this
mechanistic framework, the relative stereochemistries of the
major diastereomers obtained in these reactions may be ex-
plained by the preference for formation of the bicyclic metal-
lacycle 26 containing a cis-ring junction, as opposed to a likely
higher energy trans-ring junction.
A second, alternative catalytic cycle that does invoke the
intervention of discrete enolate intermediates is illustrated in
Scheme 3 (mechanism B). Interaction of Ni(0) with Et2Zn may
lead to intermediate 30, containing a three-center two-electron
bridging interaction.15a Coordination of 30 to the substrate 25
would then provide 31, which can undergo ꢀ-hydride elimination
to provide nickel hydride 32. Reorganization of 32 can then
occur to provide zinc enolate 33,16 which would undergo aldol
cyclization to 29, and Ni(0), which can re-enter the catalytic
cycle. Here, the observed stereochemical outcome of the
reactions may be explained by preferential formation of the (Z)-
zinc enolate 33, along with a chelated Zimmerman-Traxler-
type transition state 34.17
rather than some beneficial role played by the acetylacetonate
ligands or Zn-acac byproducts resulting from the use of
Ni(acac)2. Having established that Ni(0) is the likely active
oxidation state in the reactions described herein, our attention
turned to consideration of the possible reaction mechanism. Of
the reactive pathways available to organonickel chemistry that
could explain these reactions,2 two in particular seemed to be
plausible.
If mechanism B is operative, the formation of alkylative aldol
cyclization products in varying quantities from precursors 5 (eq
2), 20f and 20 g (Table 2, entries 6 and 7), and 22a and 22b
(eqs 10 and 11) might be accounted for by the particular steric
and/or electronic properties of these particular substrates
enabling conjugate addition from 31 to compete with ꢀ-hydride
elimination to 32. Alternatively, the formation of 7 from
substrate 5 (eq 2) can be explained using mechanism A if
reductive elimination from 27 competes effectively with ꢀ-hy-
dride elimination to 28. However, the isolation of alkylative
cyclization products 23a and 23b (eqs 10 and 11) as 1:1 mixtures
of diastereomers might be more difficult to rationalize by a
metallacycle pathway, assuming all steps in mechanism A are
stereospecific (vide infra).
Our efforts to shed light on these two mechanistic possibilities
began with an experiment to establish whether mechanism B
(Scheme 3) is possible, by subjecting R,ꢀ-unsaturated amide
35 lacking the pendant ketone electrophile to our standard
reaction conditions (Scheme 4). In principle, mechanism B does
not require the participation of the ketone until the zinc enolate
33 is formed, whereas in mechanism A (Scheme 2), the ketone
is an essential component for oxidative cyclization to occur to
provide oxanickellacycle 26. Therefore, if mechanism B is
operative, we might expect to observe the simple conjugate
reduction product 38.18,19 In the event, exposure of 35 to our
standard conditions provided only a complex mixture that
appeared to be composed of oligomeric products, with none of
the amide 38 being observed (Scheme 4). The failure to detect
38 does not rule out mechanism B since the zinc enolate 36
that would be formed from conjugate reduction of 35 could react
A significant number of nickel-catalyzed reductive coupling
and cyclizations are consistent with the intervention of metal-
lacyclic intermediates, formed by the oxidative cyclization of
Ni(0) with two π-components.2 Further support for the involve-
ment of metallacycles in these reactions comes in the form of
oxidative cyclization products that have been isolated and
characterized by X-ray crystallography.14,15 Although the present
reactions are formally reductive aldol processes, it is conceivable
that they actually occur through metallacycle participation rather
than via discrete enolates, and a possible catalytic cycle for such
a pathway is depicted in Scheme 2 (mechanism A). Oxidative
cyclization of Ni(0) with the alkene and the ketone of the
substrate 25 would result in oxanickellacycle 26. By analogy
with a detailed study conducted by Schlegel, Montgomery, and
co-workers,15a we suggest that diethylzinc would facilitate
oxidative cyclization by (i) Lewis acid activation of the ketone
through binding with zinc, and (ii) Lewis basic activation of
Ni(0) through a three-center two-electron bridging interaction
of a zinc-ethyl bond. The rate-accelerating effect of Lewis acids
on nickel-promoted oxidative cyclizations has also been dem-
onstrated by Ogoshi, Kurosawa, and coworkers.14c,d Cleavage
of the oxanickellacycle 26 by transmetalation would provide
nickel-ethyl species 27, which can then undergo ꢀ-hydride
elimination to generate nickel hydride 28. Finally, reductive
elimination of 28 would provide zinc alkoxide 29 (which would
(16) A more detailed discussion of the putative conversion of intermediates
of general structure 31 into zinc enolates is provided in the discussion
of reaction mechanism (see Scheme 12).
(17) Zimmerman, H. E.; Traxler, M.; D, J. Am. Chem. Soc. 1957, 79, 1920–
1923.
(14) (a) Ogoshi, S.; Ikeda, H.; Kurosawa, H. Angew. Chem., Int. Ed. 2007,
46, 4930–4932. (b) Ogoshi, S.; Tonomori, K.; Oka, M.; Kurosawa,
H. J. Am. Chem. Soc. 2006, 128, 7077–7086. (c) Ogoshi, S.; Ueta,
M.; Arai, T.; Kurosawa, H. J. Am. Chem. Soc. 2005, 127, 12810–
12811. (d) Ogoshi, S.; Oka, M.; Kurosawa, H. J. Am. Chem. Soc.
2004, 126, 11802–11803.
(18) Conjugate reductions of sterically hindered di- and trisubstituted enones
in low yields have been observed previously in attempted conjugate
addition reactions using Ni(acac)2/Et2Zn. See: (a) Chaloner, P. A.;
Hitchcock, P. B.; Langadianou, E.; Readney, M. J. Tetrahedron Lett.
1991, 32, 6037–6038. (b) Bolm, C.; Ewald, M.; Felder, M. Chem.
Ber. 1992, 125, 1205–1215.
(15) (a) Hratchian, H. P.; Chowdhury, S. K.; Gutie´rrez-Garc´ıa, V. M.;
Amarasinghe, K. K. D.; Heeg, M. J.; Schlegel, H. B.; Montgomery,
J. Organometallics 2004, 23, 4636–4646. (b) Amarsinghe, K. K. D.;
Chowdhury, S. K.; Heeg, M. J.; Montgomery, J. Organometallics 2001,
20, 370–372.
(19) For Co(acac)2/Et2Zn-mediated conjugate reduction of chalcone, see:
de Vries, A. H. M.; Feringa, B. L. Tetrahedron: Asymmetry 1997, 8,
1377–1378, See also ref 6 for Co(acac)2/Et2Zn-mediated conjugate
reductions of R,ꢀ-unsaturated amides.
9
7332 J. AM. CHEM. SOC. VOL. 130, NO. 23, 2008

Nickel-Catalyzed Reductive Aldol Cyclizations
A R T I C L E S
Scheme 3. A Possible Reaction Mechanism Invoking Zinc Enolate Formation
Scheme 4. Attempted Conjugate Reduction of 35
Scheme 5. Nickellacycle Formation for ꢀ-Tethered Substrates
Such as Alkynyl Enones/Enals
with additional starting material 35 in a Michael addition to
provide a second zinc enolate 37 that could then lead to further
oligomeric products. This type of behavior is known for poorly
chemoselective 1,4-addition reactions of organometallic re-
agents.20 This result differs from that obtained previously using
Co(acac)2 ·2H2O in place of Ni(acac)2, where amide 38 was
formed in 73% yield.6
Since this experiment did not provide concrete evidence
for or against mechanism B, a different approach was adopted
in the hope of gaining more useful insight, which involved
analysis of products obtained through variation of the
connectivity of the substrates. As mentioned previously, most
of the nickel-catalyzed reductive cyclizations reported to date
involving R,ꢀ-unsaturated carbonyl substrates result in
products where carbon-carbon bond formation has occurred
at the ꢀ-position. The principal reason for this preference
lies in the fact that the second reactive functional group in
these substrates is tethered to the R,ꢀ-unsaturated carbonyl
through the ꢀ-carbon. Consequently, as a result of geometrical
constraints imposed by this mode of tethering, the metalla-
cyclic mechanisms often invoked for these reactions must
lead to an overwhelming preference for cyclization at the
ꢀ-position, placing nickel R to the carbonyl carbon (e.g.,
Scheme 5, showing cyclization of an alkynyl enone/enal1t,15).
Because all of the substrates we have described thus far
have the ketone tethered to the R,ꢀ-unsaturated carbonyl via
an amide or an ester linkage, it stood to reason that the
cyclization of substrates where the ketone is tethered to the
R,ꢀ-unsaturated carbonyl through the ꢀ-carbon might provide
valuable information. Accordingly, substrate 39 was prepared
for study (Scheme 6).
If reductive aldol product 41 was obtained in the cycliza-
tion of 39, a mechanism analogous to that shown in Scheme
3, involving a discrete zinc enolate intermediate 40, would
most likely be operative, as a metallacyclic pathway (analo-
gous to mechanism A, Scheme 2) would be precluded on
the basis of geometric constraints. Although this outcome
would not provide rigorous proof, it would then be tempting
to suggest mechanism B (Scheme 3) as the most likely
explanation for the cyclization of the amide- and ester-
tethered substrates. However, if products resulting from
cyclization at the ꢀ-position were formed, a mechanism (akin
to that shown in Scheme 2) involving oxanickellacycle 42
could be responsible. A third option that is relevant in this
case, but which could not explain the cyclization of the
amide- and ester-tethered substrates, involves Et2Zn-assisted
oxidative addition of Ni(0) to the R,ꢀ-unsaturated ester to
form π-allylnickel complex 43, followed by migratory
(20) (a) Fraser, P.; Woodward, S. Chem.—Eur. J. 2003, 9, 776–783. (b)
Blake, A. J.; Shannon, J.; Stephens, J. C.; Woodward, S. Chem.—Eur.
J. 2007, 13, 2462–2472.
9
J. AM. CHEM. SOC. VOL. 130, NO. 23, 2008 7333

A R T I C L E S
Joensuu et al.
Scheme 6. Possible Cyclization Outcomes for Substrate 39
Scheme 7. Possible Mechanisms for the Formation of 49 from 42 or 44
insertion of the ketone to give 44.2 Therefore, it can be seen
that the cyclization of substrate 39 allows a clearer distinction
between alternative mechanisms than the substrates we have
described thus far. Of course, delineation of the probable
mechanism for the cyclization of 39 would not necessarily
mean that the same mechanism is also at work for the amide-
and ester-tethered substrates. Nevertheless, the reaction of
39 still represented an important experiment, as the result
would at least demonstrate whether enolate formation or
metallacycle participation is possible for substrates containing
both R,ꢀ-unsaturated carbonyl and ketone functional groups
using Ni(acac)2/Et2Zn and would add to the overall weight
of evidence. In this context, it is important to note relevant
precedent from the Montgomery group involving nickel-
catalyzed cyclization of enone-aldehydes 45 and 46 under
somewhat different conditions, which provided formal reduc-
tive homoaldol products 47 and 48, respectively, rather than
aldol products (eqs 16 and 17).1t
In the event, the reaction of 39 under standard conditions
provided bicyclic lactone 49, where carbon-carbon formation
has occurred at the ꢀ-carbon, in 56% yield (eq 18), with no
evidence of reductive aldol product 41. Therefore, an enolate
mechanism can be ruled out for the cyclization of 39, leaving
metallacycle or π-allylnickel pathways as viable options (see
Scheme 6).
Scheme 7 illustrates possible pathways for the formation of
bicyclic lactone 49 from oxanickellacycle 42 or migratory
insertion product 44. If a metallacycle pathway is operative,
transmetalation of 42 to provide nickel enolate 50, followed by
further transmetalation with Et2Zn, would provide zinc enolate
51, which upon workup would result in γ-hydroxyester 52,
which in turn would then undergo spontaneous lactonization to
49. In this case, it is important to note that a ꢀ-hydrogen-
containing organometallic reagent is not required for this formal
reductive homoaldol cyclization-lactonization sequence.21 Al-
ternatively, if a π-allylnickel pathway is operative, the same
zinc enolate intermediate 51 could arise via transmetalation of
44 with Et2Zn. As mentioned previously, the result of eq 18 is
certainly informative insofar that it complements the results of
Montgomery and co-workers1t (eqs 16 and 17) and demonstrates
9
7334 J. AM. CHEM. SOC. VOL. 130, NO. 23, 2008

Nickel-Catalyzed Reductive Aldol Cyclizations
A R T I C L E S
Scheme 8. Expected Stereochemical Outcomes of Reductive Cyclization of Deuterium-Labeled Substrate 53 under Mechanisms A and B
metallacycle formation using R,ꢀ-unsaturated ester and ketone
reaction partners as a possibility. However, this result in no way
constitutes rigorous proof that a metallacyclic pathway is
operative for the cyclizations of the amide- and ester-tethered
precursors, nor does it allows us to confidently exclude an
enolate mechanism for these substrates.
chromatography,23 and this result proved to be repeatable. This
outcome, which is intermediate between the two limiting cases
outlined in Scheme 8, suggests that mechanisms A and B are
probably oversimplified descriptions, and the true mechanism
is appreciably more complex.
Although at first glance the mixture of diastereomers obtained
in eq 19 would appear to provide strong evidence against
mechanism A, a possible scenario that was not discussed earlier,
but which could potentially complicate the analysis of this
deuterium-labeling study, is that the alkene of the R,ꢀ-
unsaturated carbonyl moiety undergoes E/Z equilibration during
these reactions (Scheme 10). Although less thermodynamically
stable (Z)-53 is likely to be present as only a minor component
in an E/Z equilibrium mixture, its influence could nonetheless
be important if E/Z interconversion is rapid, and it displays
comparable or higher reactivity compared with that of (E)-53
(Curtin-Hammett-type kinetics24). This scenario means the
diastereomeric ratio obtained in eq 19 could still result from
mechanism A (Scheme 10). In addition, E/Z isomerization
within a metallacyclic manifold could provide an explanation
for the isolation of alkylative cyclization products 23a and 23b
(eqs 10 and 11) as 1:1 mixtures of diastereomers (vide supra).
To determine whether attainment of E/Z equilibrium is
possible, we prepared (Z)-cinnamic amide (Z)-6025 and ran the
cyclization of this compound to partial completion by using only
0.5 equiv of Et2Zn, as opposed to the 2 equiv used under our
standard conditions (eq 21). Examination of the 1H NMR
spectrum of the unpurified reaction mixture revealed the
presence of lactam 61 in addition to uncyclized material which
Our next attempt to discriminate between mechanisms A and
B involved analysis of the stereochemical outcome of cyclization
of deuterium-labeled substrate 53, which was prepared in
straightforward fashion.22 If metallacycle-based mechanism A
(Scheme 2) is operative, the concerted nature of the oxidative
cyclization would be expected to provide metallacycle 54 with
the relative stereochemistry shown (Scheme 8). An eventual
reductive elimination of a nickel hydride 55 that proceeds with
retention of configuration would be expected to provide only
one diastereomer 56a of the cyclized product. However, if the
alternative mechanism B invoking the intervention of a zinc
enolate 57 is operative, we should in principle expect a 1:1
mixture of two product diastereomers 56a and 56b to be formed
since diastereomeric Zimmerman-Traxler-type transition
states17 58a and 58b differ only with respect to the deuterium
label and would therefore be expected to possess virtually
identical energies.
In the event, the nickel-catalyzed reductive cyclization of 53
provided a 1:1.3 inseparable mixture of diastereomeric products
56a and 56b (relative stereochemistries of major and minor
diastereomers not assigned) as determined by 1H NMR analysis
of the unpurified reaction mixture (Scheme 9, spectrum b, and
eq 19). That a nonequimolar ratio of diastereomers was obtained
was confirmed after removal of trace impurities by column
(24) Seeman, J. I. Chem. ReV. 1983, 83, 83–134.
(21) The cyclization of 39 using Me3Al (2 equiv) in place of Et2Zn also
resulted in the formation of 49 in ca. 80% conversion. Montgomery
and co-workers have shown that bisenone substrates tethered through
their ꢀ-carbons undergo nickel-catalyzed reductive cyclizations in the
presence of organozincs lacking ꢀ-hydrogens. See ref 1t.
(25) Substrate (Z)-60 containing an aromatic N-substituent was chosen to
simplify analysis because acyclic N-aryl tertiary amides exist pre-
dominantly in one rotameric form, leading to less complex 1H NMR
spectra. See: (a) Pederson, B. F.; Pederson, B. Tetrahedron Lett. 1965,
6, 2995–3001. (b) Itai, A.; Toriumi, Y.; Tomioka, N.; Kagechika, H.;
Azumaya, I.; Shudo, K. Tetrahedron Lett. 1989, 30, 6177–6180. (c)
Curran, D. P.; Hale, G. R.; Geib, S. J.; Balog, A.; Cass, Q. B.; Degani,
A. L. G.; Hernandes, M. Z.; Freitas, L. C. G. Tetrahedron: Asymmetry
1997, 8, 3955–3975.
(22) See the Supporting Information for details.
(23) Purification of the product of the reaction depicted in eq 19 by column
chromatography gave 56a/56b in a 1:1.5 ratio (relative stereochem-
istries of major and minor isomers not assigned).
9
J. AM. CHEM. SOC. VOL. 130, NO. 23, 2008 7335

A R T I C L E S
Joensuu et al.
Scheme 9. Reductive Cyclization of Deuterium-Labeled Substrate 53 and 1H NMR Spectra of Unlabeled Reference and Diastereomeric
Products Obtained
Scheme 10. E- to Z-Alkene Isomerization Could Allow a Mixture of
56a and 56b to be Obtained via a Metallacycle Mechanism
hydes 62 in the presence of trialkylsilyl chlorides to provide
isolable π-allylnickel complexes 63 (eq 23).26 In addition,
Ogoshi, Kurosawa, and co-workers have isolated the products
of oxidative addition of Pd(0) to enones in the presence of a
range of Lewis acids.27 π-Allylmetal species of this type have
also been invoked as intermediates in several classes of
transition-metal-catalyzed reactions2,27,28 (see also Scheme
6).
had undergone Virtually complete isomerization (E/Z >19:1)
to the more thermodynamically stable E-isomer, (E)-60. There-
fore, mechanism A cannot be excluded on the basis of the results
of eq 19.
We suggest that alkene isomerization occurs via the
formation of π-allylnickel species. The generation of π-al-
lylmetal complexes by the oxidative addition of low-valent
transition metals to R,ꢀ-unsaturated carbonyl compounds in
On the basis of this precedent, E/Z equilibration is therefore
likely initiated by coordination of Ni(0) to R,ꢀ-unsaturated amide
(E)-64 to provide η2-coordinated complex (E)-65, followed by
Et2Zn-assisted oxidative addition to provide π-allylnickel
complex 66 (Scheme 11). A hapticity change from η3 to η1 to
the presence of Lewis acids is well precedented.26,27
A
particularly relevant example comes in the form of pioneering
studies by MacKenzie and co-workers, who demonstrated that
Ni(0) undergoes oxidative addition to R,ꢀ-unsaturated alde-
9
7336 J. AM. CHEM. SOC. VOL. 130, NO. 23, 2008

Nickel-Catalyzed Reductive Aldol Cyclizations
A R T I C L E S
Scheme 11. Establishment of E/Z Alkene Equilibrium via η3-η1-η3 Isomerization
Scheme 12. Zinc Enolate Formation via π-Allylnickel Species
give 67a would then allow bond rotation to occur to provide
67b. Re-establishment of η3 hapticity to give π-allylnickel
complex 68, followed by reductive elimination to (Z)-65 and
decomplexation, would then furnish the isomerized R,ꢀ-unsatur-
ated amide (Z)-64. The entire process is of course reversible,
and this type of η3-η1-η3 (also known as π-σ-π) isomer-
ization has good precedence in the area of transition-metal-
catalyzed allylic alkylations.29 It should be stated that, for the
purposes of simplicity, no three-center two-electron bridging
interaction15a of the type depicted in structure 31 (see Scheme
3) and in structure 69 (Scheme 12, vide infra) between nickel,
zinc, and an ethyl ligand has been shown for structures 65-68
in Scheme 11. However, an associative interaction of this type
seems highly likely,15a which would also have important
implications for mechanism B (see below).
Lewis basic ketone to the nickel and/or zinc center in any of
the intermediates 70-73 (Scheme 12) is certainly possible,
which would place the ketone on one particular diastereotopic
face. If the degree of this association remains significant until
formation of the zinc enolate (ent-57 in the case of Scheme
12), and aldol cyclization is able to occur before the ketone
has the opportunity to fully establish equilibrium in interchang-
ing between the two diastereotopic enolate faces, a nonequimolar
distribution of diastereomers could certainly result. Other factors
which could complicate the situation further include the E/Z
isomerization discussed above (Scheme 12 only shows the
reaction of (E)-53), the nature/hapticity of the zinc enolate itself
(oxa-π-allyl species, C-bound versus O-bound enolates), and
changes in this hapticity during the course of the reaction.
Therefore, mechanism B remains a strong possibility.
To summarize, these mechanistic investigations have (a)
provided compelling evidence that Ni(0), generated in situ by
the reduction of Ni(acac)2 with Et2Zn, is the catalytically
competent species in these reactions; (b) shown Ni(acac)2 to
be a superior nickel source to the other commonly used
precatalyst for nickel-catalyzed reductive couplings and cy-
clizations, namely, Ni(COD)2; (c) demonstrated that a substrate
Having determined that mechanism A (Scheme 8, left)
remains a possibility, the question of whether mechanism B
(Scheme 8, right) could also give rise to the observed ratio of
56a and 56b from the cyclization of 53 must be addressed. In
this regard, and in light of the above discussion, a more detailed
representation of the mechanism of zinc enolate formation from
53, involving η3 and η1 intermediates, is depicted in Scheme
12. A tacit assumption in the simplified representation of
mechanism B shown in Scheme 8 is that coordination of the
pendant ketone in Zimmerman-Traxler-type transition states
58a and 58b only occurs after zinc enolate 57 has formed, which
should then provide a 1:1 mixture of 56a and 56b. However,
this assumption is not necessarily valid, as association of the
(26) (a) Johnson, J. R.; Tully, P. S.; MacKenzie, P. B.; Sabat, M. J. Am.
Chem. Soc. 1991, 113, 6172–6177. (b) Grisso, B. A.; Johnson, J. R.;
MacKenzie, P. B. J. Am. Chem. Soc. 1992, 114, 5160–5165.
(27) (a) Ogoshi, S.; Yoshida, T.; Nishida, T.; Morita, M.; Kurosawa, H.
J. Am. Chem. Soc. 2001, 123, 1944–1950. (b) Ogoshi, S.; Tomiyasu,
S.; Morita, M.; Kurosawa, H. J. Am. Chem. Soc. 2002, 124, 11598–
11599.
(28) For relevant examples involving Ni(0)-catalyzed reactions, see: (a)
Hirano, K. Yorimitsu, H. Oshima, K. Org. Lett. 2007, 9, 5031–5033.
(b) Hirano, K. Yorimitsu, H. Oshima, K. Org. Lett. 2007, 9, 1541–
1544. (c) Seiber, J. D. Liu, S. Morken, J. P. J. Am. Chem. Soc. 2007,
129, 2214–2215. (d) Ikeda, S.-i. Sato, Y. J. Am. Chem. Soc. 1994,
116, 5975–5976. See also ref 2.
(29) See for example: (a) Trost, B. M.; Van Vranken, D. L. Chem. ReV.
1996, 96, 395–422. (b) Ogasawara, M.; Takizawa, K.-i.; Hayashi, T.
Organometallics 2002, 21, 4853–4861.
9
J. AM. CHEM. SOC. VOL. 130, NO. 23, 2008 7337

A R T I C L E S
Joensuu et al.
39 containing a ketone connected to an R,ꢀ-unsaturated carbonyl
through the ꢀ-carbon provides a reductive homoaldol product
49 rather than a reductive aldol product 41, thus suggesting
metallacycle or π-allylnickel mechanisms as possible pathways;
and (d) uncovered that E/Z isomerization of the R,ꢀ-unsaturated
carbonyl component can occur under these conditions. The
implications of the latter phenomenon have meant that, unfor-
tunately, a deuterium-labeling study designed specifically to
distinguish between alternative metallacyclic (Scheme 8, left)
and enolate (Scheme 8, right) pathways does not allow us to
exclude either of these mechanisms. Although certain mecha-
nistic issues remain unresolved, we believe the insights gained
through these studies will be of benefit to other researchers
engaged in the design, development, and understanding of
related nickel-catalyzed reactions.
the unpurified mixture demonstrates that E/Z interconversion
is also possible using the Co(acac)2 ·2H2O/Et2Zn combination.
Conclusions
We have shown that, in the presence of diethylzinc, Ni(acac)2
serves as a highly effective precatalyst for the reductive aldol
cyclization of substrates containing an R,ꢀ-unsaturated carbonyl
moiety tethered to a ketone through either an amide or an ester.
In these reactions, diethylzinc acts a stoichiometric reducing
agent, formally delivering a hydride to the ꢀ-position of the
R,ꢀ-unsaturated carbonyl component of the cyclization precur-
sor, leading to the formation of ꢀ-hydroxylactams and ꢀ-hy-
droxylactones with good to high levels of diastereoselection. A
notable feature of these reactions is the broad tolerance of
variation of the ketone, the ꢀ-substituent of the R,ꢀ-unsaturated
carbonyl component, and the nitrogen protecting group in the
case of amide-linked precursors to give a range of five- and
six-membered products. Possible mechanisms for these trans-
formations have been discussed, and a deuterium-labeling study
designed to probe the stereochemical outcome of cyclization
within the context of these mechanisms has revealed the
complex nature of these reactions. We anticipate the reactions
described herein will stimulate the development of further
nickel-catalyzed reductive coupling and cyclization reactions
that will be of broad utility.
For comparison purposes, we also repeated the experiments
of these mechanistic investigations using Co(acac)2 ·2H2O as
the precatalyst6 in place of Ni(acac)2. For the cyclization of
substrate 39, Co(acac2)·2H2O was found to be inferior to
Ni(acac)2. Although appreciable quantities of bicyclic lactone
49 were formed on occasions, with none of the reductive aldol
product 41 detected (thus suggesting the ability of cobalt to
participate in metallacycle or π-allylmetal pathways), the
reproducibility of this reaction was variable. The cyclization of
deuterium-labeled substrate 53 using Co(acac)2 ·2H2O again
resulted in a nonequimolar ratio of diastereomeric products 56a
and 56b (eq 20), but compared to the result obtained using
Ni(acac)2 (eq 19), the magnitude of diastereoselection was
increased, and most importantly, the sense of diastereoselection
was reVersed (2.6:1 diastereomeric mixture, see Scheme 9,
spectrum c). While the differences between the Ni(acac)2/Et2Zn
and Co(acac)2 ·2H2O/Et2Zn reagent combinations are highlighted
by the generally broader substrate scope of the nickel system,
and their divergent behavior with simple ꢀ-unsubstituted acry-
lamides such as 5 and with substrates such as 35 lacking a
pendant ketone (vide supra), the results of this isotopic labeling
study also point to differences in their mechanisms. Finally,
the cyclization of (Z)-60 was conducted using Co(acac)2 ·2H2O
and 0.5 equiv of Et2Zn, which resulted in a mixture of recovered
(Z)-60, the cyclized product 61, and (E)-60 (eq 22). Although
this reaction proceeded to lower conversion than the analogous
experiment using Ni(acac)2 (eq 21), the presence of (E)-60 in
Acknowledgment. This research was supported by the Uni-
versity of Edinburgh, Hoffmann-La Roche, and the EPSRC. We
thank AstraZeneca and Merck Sharp & Dohme for unrestricted
research support, Stephen Dixon for assistance with the cyclization
of substrate 14a, Claire L. Oswald for assistance in the preparation
of 39, and the groups of Polly L. Arnold and Jason B. Love for the
use of their glovebox. We thank Dr. Richard C. Hartley at the
Department of Chemistry, University of Glasgow, for helpful
discussions, and the reviewers of this Article for valuable sugges-
tions regarding reaction mechanisms. We gratefully acknowledge
the EPSRC National Mass Spectrometry Centre at the University
of Wales, Swansea, for providing high-resolution mass spectra.
Supporting Information Available: Detailed experimental
procedures, spectroscopic data for all new compounds, and
copies of NMR spectra. This material is available free of charge
via the Internet at http://pubs.acs.org.
JA0775624
9
7338 J. AM. CHEM. SOC. VOL. 130, NO. 23, 2008