Intermolecular enolate heterocoupling: Scope, mechanism, and application

Article abstract of DOI:10.1021/ja804159y

This full account presents the background on, discovery of, and extensive insight that has been gained into the oxidative intermolecular coupling of two different carbonyl species. Optimization of this process has culminated in reliable and scalable proto

Full text of DOI:10.1021/ja804159y

Published on Web 08/05/2008
Intermolecular Enolate Heterocoupling: Scope, Mechanism,
and Application
Michael P. DeMartino, Ke Chen, and Phil S. Baran*
Department of Chemistry, The Scripps Research Institute, 10550 North Torrey Pines Road,
La Jolla, California 92037
Received June 2, 2008; E-mail: pbaran@scripps.edu
Abstract: This full account presents the background on, discovery of, and extensive insight that has been
gained into the oxidative intermolecular coupling of two different carbonyl species. Optimization of this
process has culminated in reliable and scalable protocols for the union of amides, imides, ketones, and
oxindoles using soluble copper(II) or iron(III) salts as oxidants. Extensive mechanistic studies point to a
metal-chelated single-electron-transfer process in the case of copper(II), while iron(III)-based couplings
appear to proceed through a non-templated heterodimerization. This work presents the most in-depth findings
on the mechanism of oxidative enolate coupling to date. The scope of oxidative enolate heterocoupling is
extensive (40 examples) and has been shown to be efficient even on a large scale (gram-scale or greater).
Finally, the method has been applied to the total synthesis of the unsymmetrical lignan lactone
(-)-bursehernin and a medicinally important 2,3-disubstituted succinate derivative.
Introduction
at lower (or higher) oxidation states, as seen in taiwankadsurin
B.6,10 Scrutiny of the structures presented in Figure 1 reveals
Organic chemists’ longstanding endeavor to achieve target-
oriented syntheses as concisely and efficiently as possible has
seen incredible progress over the course of the past decade.
During this time, the development of succinct methods for the
direct functionalization of molecules has begun to dominate the
field.1 While there generally exist a myriad of means, either
direct or indirect, of accessing a particular organic moiety, the
efficiency with which methods can provide a target must
continually be called into question in order to continue to
advance the field. For example, the 2,3-disubstituted-1,4-
dicarbonyl moiety is ubiquitous within organic molecules. Figure
1 provides examples of seemingly unrelated complex natural
products2-7 and medicinal compounds.8,9 While exhibiting great
chemical diversity, they are linked through this common
structural element that is found in these and countless other
natural and medicinal agents. Generally, this motif is readily
apparent, but it can also be masked in the form of carbon atoms
just how valuable a method could be in the construction of such
complex molecular architectures.
The direct, convergent synthesis of unsymmetrical 2,3-
disubstituted-1,4-dicarbonyl compounds from two carbonyl
subunits has proven extremely difficult; several methods for the
synthesis of hypothetical succinate 1 are depicted in Figure 2.9,11
Efficient, enantioselective syntheses of such entities have
escaped synthetic grasp, in spite of their presence in countless
natural products and innumerable medicinal remedies. All of
the methods depicted suffer from one or more of the following
limitations: multistep sequences, installation of requisite dispos-
able functionality on one or both of the monomers, and
stereoselectivity problems with prefunctionalization methods
and/or during the union of the two monomers. No stereoselec-
tivity was observed or necessary for the most efficient of these
methods, the Stetter reaction, as the product was subjected to a
pyrrole synthesis. This report is a full account of a research
program initiated originally to eliminate the first two of these
issues and having since evolved to address the third. By taking
advantage of an underutilized and underappreciated reactivity
of carbonyl enolates, the oxidative heterocoupling of two
enolates joins two different sp3-hybridized carbon centers in a
single step without requiring prefunctionalization of the corre-
sponding monomers. In an effort to foster a more complete
understanding and showcase the broadened utility of this
(1) (a) Campos, K. R. Chem. Soc. ReV. 2007, 36, 1069–1084. (b) Fairlamb,
I. J. S. Annu. Rep. Prog. Chem., Sect. B: Org. Chem. 2007, 103, 68–
89.
(2) Seto, H.; Sato, T.; Urano, S.; Uzawa, J.; Yonehara, H. Tetrahedron
Lett. 1976, 4367–4370.
(3) (a) Omura, S.; Hirano, A.; Iwai, Y.; Masuma, R. J. Antibiot. 1979,
32, 786–790. (b) Furusaki, A.; Matsumoto, T.; Ogura, H.; Takayanagi,
H.; Hirano, A.; Omura, S. J. Chem. Soc., Chem. Commun. 1980, 698.
(4) Kosuge, T.; Tsuji, K.; Hirai, K.; Yamaguchi, K.; Okamoto, T.; Iitaka,
Y. Tetrahedron Lett. 1981, 22, 3417–3420.
(5) Li, S.-H.; Wang, J.; Niu, X.-M.; Shen, Y.-H.; Zhang, H.-J.; Sun, H.-
D.; Li, M.-L.; Tian, Q.-E.; Lu, Y.; Cao, P.; Zheng, Q.-T. Org. Lett.
2004, 6, 4327–4330.
(10) Ayres, D. C.; Loike, J. D. Lignans: Chemical, Biological, and Clinical
Properties; Cambridge University Press: Cambridge, UK, 1990.
(11) (a) Ryter, K.; Livinghouse, T. J. Am. Chem. Soc. 1998, 120, 2658–
2659. (b) Lin, J.; Chan, W. H.; Lee, A. W. M.; Wong, W. Y.
Tetrahedron 1999, 55, 13983–13998. (c) Miura, K.; Fujisawa, N.;
Saito, H.; Wang, D.; Hosomi, A. Org. Lett. 2001, 3, 2591–2594. (d)
Bar, G.; Parsons, A. F.; Thomas, C. B. Synlett 2002, 1069–1072. (e)
Sulsky, R.; et al. Bioorg. Med. Chem. Lett. 2007, 17, 3511–3515.
(6) Shen, Y.-C.; Lin, Y.-C.; Cheng, Y.-B.; Kuo, Y.-H.; Liaw, C.-C. Org.
Lett. 2005, 7, 5297–5300.
(7) McDoniel, P. B.; Cole, J. R. J. Pharm. Sci. 1972, 61, 1992–1994.
(8) Whittaker, M.; Floyd, C. D.; Brown, P.; Gearing, A. J. H. Chem. ReV.
1999, 99, 2735–2776.
(9) Fujisawa, T.; Igeta, K.; Odake, S.; Morita, Y.; Yasuda, J.; Morikawa,
T. Bioorg. Med. Chem. 2002, 10, 2569–2581.
9
11546 J. AM. CHEM. SOC. 2008, 130, 11546–11560
10.1021/ja804159y CCC: $40.75
2008 American Chemical Society

Intermolecular Enolate Heterocoupling
A R T I C L E S
Figure 1. Selected natural products containing 1,4-dicarbonyl moieties.
Figure 2. Prefunctionalization requirements (red ) necessary prefunctionalization).
method, a detailed mechanistic study, expansion of scope, and
application in target-oriented synthesis will be discussed.
ment, further mechanistic studies on molecular halogen-based
oxidative enolate homocoupling culminated in proof that this
process actually proceeds through a single-electron-transfer
(SET) process.13 Although not fully appreciated until many years
later, Ivanoff and Spassoff’s discovery laid the foundation for
what would become a fertile area of research beginning in the
late 1960s14 and continuing to the present. The developments
during this period have been recently summarized,15 and as such,
only highlights pertinent to the current studies will be mentioned
herein. In 1971, Rathke and Lindert reported the use of CuBr2
Background
The seminal report of a carbonyl enolate dimerization was
disclosed in 1935 by Ivanoff and Spassoff (Figure 3).12 In this
disclosure, they treated the sodium salt of a carboxylic acid with
a Grignard reagent, and the carboxylate dianion was generated.
Following treatment with diatomic bromine, the two carboxylate
enolates were fused together at their R-carbons in a net two-
electron oxidation, thus providing the corresponding symmetrical
succinic acid. Although this reaction could simply proceed
through a one-pot R-halogenation and subsequent SN2 displace-
(13) Renaud, P.; Fox, M. A. J. Org. Chem. 1988, 53, 3745–3752.
(14) Okubo, T.; Tsutsumi, S. Bull. Chem. Soc. Jpn. 1964, 37, 1794–1797.
(15) Richter, J. M.; Whitefield, B. W.; Maimone, T. J.; Lin, D. W.;
Castroviejo, M. P.; Baran, P. S. J. Am. Chem. Soc. 2007, 129, 12857–
12869.
(12) Ivanoff, D.; Spassoff, A. Bull. Soc. Chim. Fr. 1935, 2, 76–78.
9
J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008 11547

A R T I C L E S
DeMartino et al.
at least a 3-fold excess of one of the monomers, the union of
ketones and esters was achieved through the action of CuCl2,
affording unsymmetrical succinate derivatives in synthetically
useful yields. At the same time, Itoh and co-workers demon-
strated that the oxidative union of two different esters was
possible by electrolytic means, again using a stoichiometric
advantage (3-fold excess) in order to selectively acquire the
heterodimer.21 These reports and their accompanying synthetic
potential remained dormant for over three decades while the
field was dominated by extensive studies dedicated to the parent
reaction: oxidative enolate homocoupling.22 Selective hetero-
couplings received no further attention until the initial com-
munication of the current method in 2006.23 In this full account,
the discovery, optimization, mechanistic exploration, and ap-
plication to medicinal and natural product chemistry are
discussed for intermolecular enolate heterocoupling.
Discovery and Optimization
At the outset of this research program, one of the key
challenges that needed to be addressed in oxidative heterocou-
pling was chemoselectivity. Indeed, enolate homodimerization
reactions had already been demonstrated as superb standalone
methods for symmetrical succinate synthesis. As previously
discussed, Saegusa and Itoh circumvented this problem through
the use of superstoichiometric quantities of one of the coupling
partners. While this tactic certainly seems sufficient for the
generation of relatively simple systems, its implementation
would be greatly hindered in complex molecule synthesis.
Several solutions to the problem of homo/heterodimer selectivity
have been developed in recent years,15,24 but each of these
methods requires the alteration of carbonyl reactivity through
a variety of different prefunctionalization strategies, rather than
utilization of the innate reactivity of the free carbonyls.
Given this precedent, the prospect of selective heterodimer
formation seemed daunting. Darkening the outlook even further,
homodimerization, while the most troublesome byproduct, is
certainly not the only side reaction that could take place in
attempting this transformation (see Figure 4). Depending on the
types of carbonyls employed, homo- and cross-condensation
products may be envisioned as arising from Claisen or aldol
reactions. This type of reactivity could even potentially dominate
the product distribution if the pKa’s of the two monomers, and
therefore rates of enolization, were vastly dissimilar. Products
resulting from oxidation in an undesired fashion could also be
problematic. For example, carbonyl R-hydroxylation25 and
dehydrogenation26,27 are both possible under the oxidative
conditions of the reaction. Moreover, the homodimers, conden-
sation products, and even the desired heterocoupled product
could be susceptible to overoxidation, thereby greatly suppress-
ing conversion and productive reactivity. It should be noted that
any bimolecular reaction of one carbonyl with another molecule
Figure 3. Pertinent intermolecular oxidative enolate coupling timeline.
and Cu(valerate)2 as suitable oxidants for the homocoupling of
enolates.16 This significant study demonstrated two important
and related advances: soluble metals (in contrast to the electrodes
used in electrolysis17) were viable reagents for the coupling
process, and copper(II) salts were competent oxidants for this
transformation. In a related study in 1980, Frazier and Harlow
illustrated that iron salts in the 3+ oxidation state were also
quite competent for enolate couplings,18 although the scope and
specificity of these two metals has remained unclear. The scope
of enolates used in oxidative couplings was investigated and
expanded over the next 15 years, and in 1995, Kise and co-
workers reported the first example of an oxazolidinone partici-
pating in an oxidative enolate homocoupling.19
(20) (a) Ito, Y.; Konoike, T.; Harada, T.; Saegusa, T. J. Am. Chem. Soc.
1977, 99, 1487–1493. (b) Ito, Y.; Konoike, T.; Saegusa, T. J. Am.
Chem. Soc. 1975, 97, 2912–2914.
The comprehensive list of intermolecular oxidative hetero-
couplings of unfunctionalized carbonyl enolates is actually quite
brief. Saegusa and co-workers were the first to demonstrate that
this process was possible in the mid-1970s (Figure 3).20 By using
(21) Tokuda, M.; Shigei, T.; Itoh, M. Chem. Lett. 1975, 621–624.
(22) Csaky, A. G.; Plumet, J. Chem. Soc. ReV. 2001, 30, 313–320.
(23) Baran, P., S.; DeMartino, M., P. Angew. Chem., Int. Ed. 2006, 45,
7083–7086.
(24) Clift, M. D.; Taylor, C. N.; Thomson, R. J. Org. Lett. 2007, 9, 4667–
4669.
(16) Rathke, M. W.; Lindert, A. J. Am. Chem. Soc. 1971, 93, 4605–4606.
(17) Weinberg, N. L.; Weinberg, H. R. Chem. ReV. 1968, 68, 449–523.
(18) Frazier, R. H., Jr.; Harlow, R. L. J. Org. Chem. 1980, 45, 5408–5411.
(19) Kise, N.; Tokioka, K.; Aoyama, Y.; Matsumura, Y. J. Org. Chem.
1995, 60, 1100–1101.
(25) Baran, P. S.; Hafensteiner, B. D.; Ambhaikar, N. B.; Guerrero, C. A.;
Gallagher, J. D. J. Am. Chem. Soc. 2006, 128, 8678–8693.
(26) Paquette, L. A.; Bzowej, E. I.; Branan, B. M.; Stanton, K. J. J. Org.
Chem. 1995, 60, 7277–7283.
(27) Langer, T.; Illich, M.; Helmchen, G. Synlett 1996, 1137–1139.
9
11548 J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008

Intermolecular Enolate Heterocoupling
A R T I C L E S
Table 1. Oxazolidinone-Propiophenone Coupling Optimizationa
entry
condition
yield (%)
Oxidant ) Fe(acac)3 ) Fe(III)
SolVent
1
2
3
4
5
THF
57
5
16
0
Et2O
DME
CPME
PhMe
51
Temperature
-78 °C
-78 to 25 °C
-40 °C
-40 to 25 °C
0 °C
0 to 25 °C
25 °C
6
7
8
0
21
0
18
16
24
57
9
10
11
12
Concentration
0.05 M
13
14
15
16
17
31
31
57
39
40
0.10 M
0.30 M
0.50 M
1.00 M
Figure 4. Possible reactions to compete with heterocoupling.
Oxidant ) Cu(2-ethylhexanoate)2 ) Cu(II)
of itself would be particularly detrimental to the yield of the
desired heterodimerization; for every one occurrence of these
transformations, two molecules of the limiting reagent would
be consumed in a nonproductive pathway. In spite of the pitfalls
that were sure to be encountered, the potential power of
oxidative enolate heterocoupling gave cause to pursue its
development.
SolVent
THF
18
19
20
21
22
55
0
51
19
9
Et2O
DME
CPME
PhMe
Temperature
-78 °C
-78 to 25 °C
-40 °C
-40 to 25 °C
0 °C
0 to 25 °C
25 °C
Concentration
0.05 M
23
24
25
26
27
28
29
16
42
16
50
39
55
26
N-Acyloxazolidinones were chosen as one of the coupling
partners for initial exploration of the oxidative heterocoupling
reaction for two reasons. First, an initial long-term goal of this
project was to be able to control the stereochemical course of
the reaction, the prospect of which would be more realistic if
the chemistry were developed around appropriate substrates.
Imides of this type have proven competent substrates for
oxidative coupling, as they have been employed repeatedly in
the complementary homodimerization process.19,28,29 Further-
more, simple ketones proved satisfactory for analogous cou-
plings with indoles15,30,31 and pyrroles,15,32 and so propiophe-
none 3 was chosen to complement N-acyloxazolidinone 2. The
choice of oxidant, iron(III) acetylacetonate [Fe(III)], was made
as a result of the recent successful intramolecular coupling of
a tertiary amide (potentially similar to an oxazolidinone) and
an ester in the context of total synthesis.25,33 Fe(III) is also
inexpensive, commercially available or easily prepared if
necessary, and highly soluble in tetrahydrofuran (THF). In the
30
31
32
33
34
39
37
55
50
50
0.10 M
0.30 M
0.50 M
1.00 M
a Diastereomeric ratios determined by 1H NMR integration.
Diastereomeric ratios (for the methyl-bearing carbon) did not change
with altered reaction conditions: Fe(III) entries, 1.8:1.0; Cu(II) entries,
1.0:1.6.
inaugural experiment, an equal stoichiometry of 2 and 3 in the
same reaction vessel was treated with lithium diisopropylamide
(LDA) at -78 °C. Fe(acac)3 was added at this temperature, and
the reaction was then placed in a room-temperature (25 °C)
water bath. The result was 21% yield of the coupled product 4
(entry 7, Table 1), but more importantly, the reaction mixture
consisted of only product, starting materials, and the homodimer
of 3. This was very encouraging, given the initial trepidation
involving potential side reactions (Vide supra), and provided a
baseline on which to improve. A screen of common solvents
indicated that THF, as expected, was indeed the optimal choice,
(28) Langer, T.; Illich, M.; Helmchen, G. Tetrahedron Lett. 1995, 36, 4409–
4412.
(29) (a) Kim, J. W.; Lee, J.-J.; Lee, S.-H.; Ahn, K.-H. Synth. Commun.
1998, 28, 1287–1292. (b) Kise, N.; Kumada, K.; Terao, Y.; Ueda, N.
Tetrahedron 1998, 54, 2697–2708. (c) Nguyen, P. Q.; Schafer, H. J.
Org. Lett. 2001, 3, 2993–2995. (d) Kise, N.; Fujimoto, A.; Ueda, N.
Tetrahedron: Asymmetry 2002, 13, 1845–1847.
(30) Baran, P. S.; Richter, J. M. J. Am. Chem. Soc. 2004, 126, 7450–7451.
(31) (a) Baran, P., S.; Richter, J. M. J. Am. Chem. Soc. 2005, 127, 15394–
15396. (b) Baran, P. S.; Maimone, T. J.; Richter, J. M. Nature 2007,
446, 404–408.
(32) Baran, P. S.; Richter, J. M.; Lin, D. W. Angew. Chem., Int. Ed. 2005,
44, 609–612.
(33) Baran, P. S.; Guerrero, C. A.; Ambhaikar, N. B.; Hafensteiner, B. D.
Angew. Chem., Int. Ed. 2005, 44, 606–609.
9
J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008 11549

A R T I C L E S
DeMartino et al.
Chart 1. Heterocoupling Yield of 2 and 3 as a Function of Oxidant Stoichiometry
although toluene performed admirably as well. Several modes
of oxidant addition were analyzed, including addition as a solid
as well as in solution at different rates, and it was determined
that fast addition of a concentrated solution (0.5 M, limit of
solubility) led to the most reproducible results and highest yields.
An interesting temperature effect was also realized. Oxidant
addition at low temperature, followed by 30 min of stirring at
that temperature before quenching (entries 6 and 8, Table 1),
offered no conversion whatsoever. If, instead, the oxidant-
containing cold reaction vessels were immediately placed into
a room-temperature water bath, minor amounts of coupled
product 4 were noted. As the reaction mixtures are homogeneous
even at low temperatures, oxidant solubility is not the problem;
instead, it seems that higher temperatures are necessary for the
oxidative coupling to occur. This trend continued at moderate
temperature (entries 10 and 11), but the best yields were
observed when the oxidant addition was performed with the
reaction stirring at room temperature (entry 12). A screen of
initial concentrations of the coupling partners demonstrated that
low concentrations suppressed the reaction (entries 13 and 14),
while higher concentrations also lowered yields, but to a lesser
extent. It appeared that 0.3 M (entry 15) was the ideal
concentration for the Fe(III)-based cross-couplings.
carried out at low temperature (entries 23 and 25). Interestingly,
when the optimized conditions for Fe(III) were executed using
Cu(II) (entry 29), the homo-Claisen condensation reaction of 2
became competitive, although it is unclear why the addition of
a solution of Cu(II) would promote this reactivity. Ultimately,
it was found that Cu(II) addition at 0 °C, followed by immersion
of the reaction vessel in a room-temperature water bath, provided
optimal results.
The final reaction parameter studied was the effect of
increasing or decreasing the amount of oxidant relative to a
stoichiometric quantity (two equivalents of metal total, one per
reacting enolate). Similar trends were observed for both metals:
diminished yields at lower stoichiometry, slightly increased
yields at superstoichiometric quantities, and eventual decrease
in yield as a large excess of oxidant is added. The optimized
condition chosen was two equivalents, as the yield improvement
using three equivalents was not significant enough to justify
using an extra equivalent of the oxidant in the generalized
method. Two interesting observations can be gleaned from Chart
1. First, a substoichiometric amount of Cu(II) (1.5 equiv)
provided an identical yield as compared to using the full
stoichiometric amount.34 Second, it is clear that Cu(II) was much
more efficient than Fe(III) when substoichiometric quantities
of oxidant were used. The mechanistic implications of these
findings will be discussed below.
The studies performed on the oxidative coupling of oxazo-
lidinone 2 and ketone 3 culminated in the following generalized
procedure that is efficient, reliable, scalable, and operationally
simple: oxazolidinone (1.0 equiv) and carbonyl compound (1.0
equiv) were dissolved in THF (0.3 M), and the solution was
cooled to -78 °C. LDA (2.1 equiv) was added dropwise by
syringe over 30 s, and the reaction was allowed to stir for 30
min at -78 °C. For Fe(III), the reaction was removed from the
cold bath and allowed to warmed to ambient temperature (5
min), at which time a 0.5 M solution of Fe(acac)3 (2.0 equiv)
was added in one portion. For Cu(II), the reaction was placed
in a 0 °C bath and stirred for 5 min, at which time a 0.5 M
solution of Cu(2-ethylhexanoate)2 (2.0 equiv) was added in one
portion, followed by immediate placement of the reaction vessel
Since the initial disclosure of this method,23 it was discovered
that Cu(II) 2-ethylhexanoate [Cu(II)] is also proficient in
affecting the cross-coupling of oxazolidinones and ketones. This
was a particularly intriguing finding, as this oxidant is also quite
inexpensive and highly soluble in THF. More importantly,
mechanistic investigations of a single transformation using two
different metal oxidants were made feasible, the subtle distinc-
tions of which were not appreciated until this study (Vide infra).
As such, it was prudent to systematically optimize the reaction
conditions for Cu(II) as well.
Very few changes from the method already developed for
Fe(III) were actually necessary. Once again, THF (entry 18)
proved the superior solvent, although dimethoxyethane (DME)
also led to a high yield (entry 20), and parallel concentration
effects (entries 30-34) were noted, with 0.3 M being optimal.
The dependence on temperature was again the most intriguing
of these optimization studies. Similar to the observation with
Fe(III), a rapid increase to room temperature seemed favorable
(entries 24, 26, and 28), though Cu(II) proved modestly capable
of coupling the two monomers even when the reaction was
(34) Two equivalents of oxidant were still chosen for the optimized
procedure, as this proved more reproducible while expanding substrate
scope.
9
11550 J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008

Intermolecular Enolate Heterocoupling
A R T I C L E S
Table 2. Substrate Scope for Oxazolidinone/Oxindole-Ketone
Couplings
Scheme 1. Attempted Quaternary Carbon Formation
propiophenone (11, 12, 4) coupling partners lead to much more
efficient Fe(III)-based couplings. Electron deficiency is much
better tolerated on the propiophenones (13-15) than the
oxazolidinone (7-9), where electron-withdrawing groups sup-
press coupling. Interestingly, the Cu(II)-based couplings showed
the opposite trends; coupling of 4′-methoxypropiophenone (11)
proceeded only modestly, whereas the reactions of the coun-
terpart 4′-methoxy- and 4′-methyloxazolidinones (5, 6) were
stifled. Withdrawing electron density seemed to aid the reaction
for both coupling partners (7-9, 13, 14), with the exception of
4′-trifluoromethylpropiopheneone (15), which was only slightly
inferior to the parent ketone (4), and the 4′-nitro coupling
partners (10, 16). The nitro group is apparently incompatible
with this chemistry, regardless of which metal is used. Isopropyl
oxazolidinones (17-19) were also attempted in the Fe(III)-based
couplings; these proved to be quite proficient, operating under
the same electronic trend as before, and with the bulkier
auxiliary modestly improving the diastereoselectivity. The
oxazolidinones were also cross-coupled with simple enones such
as carvone (20) as well as cyclic aryl ketones such as
4-chromanone (21) in good yield. The synthesis of 21 was
selected to demonstrate that the reaction can be performed on
gram scale with no diminution in yield.
An appealing extension of this chemistry would be the
construction of quaternary carbon centers by employing either
one or two R-methines instead of R-methylenes on the coupling
partners. This advance, however, has eluded discovery to this
point. Appending a small steric entity such as a methyl group
on either of the coupling partners 26 or 27 (Scheme 1) was
found to completely stifle reactivity and led to no reaction with
either oxidant. The formation of quaternary centers in this
context remains a limitation of this chemistry.
Oxindoles are found at the nucleus of countless natural and
medicinal agents,35 and as such, their couplings would be a
useful extension of this chemistry. As seen in Table 2, 1,3-
dimethyloxindole smoothly adheres to carvone (22) in high
yield, as does 1-methoxymethyl-3-prenyloxindole (23); these
are particularly noteworthy, as the product of this transformation
bears a newly formed quaternary carbon center at the C-3
position of the oxindole. If a 2-fold excess of carvone is used
in this transformation, the yield of 22 can be bolstered to 91%.
This result corroborates the early work of Saegusa and Itoh,
showing that a stoichiometric excess of one of the coupling
partners can be used to augment yields in the case of commodity
chemicals. 4-Chromanone was also coupled with 1,3-dimeth-
yloxindole (24) and 1-methoxymethyl-3-prenyloxindole (25) in
very good yields.
a Fe(acac)3 used as oxidant unless otherwise stated. b Isolated yield.
c Diastereomeric ratios determined by 1H NMR integration.
in a room-temperature bath. For both metals, the reaction was
stirred at ambient temperature for 30 min and then quenched.
Scope
As shown in Table 2, changes in the electronic nature of the
aromatic rings of both the oxazolidinones and propiophenones
greatly affected the efficiency of the reaction, the mechanistic
implications of which will be discussed below. Generalizing
these effects suggests that electron-neutral and electron-rich
aromatic rings on both the oxazolidinone (5, 6, 4) and
(35) A SciFinder Scholar search of the oxindole substructure with biological
activity returns over 2200 unique patent references.
9
J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008 11551

A R T I C L E S
DeMartino et al.
Scheme 2. Extension to 2,3-Dialkylsuccinates
but the latter were chosen for generalization so that chemical
differentiation of the carbonyls in subsequent chemistry would
be trouble-free. Significantly, 42 was synthesized on a gram
scale, once again demonstrating the scalability of this method.
Moreover, one gram of this material was donated to the Bristol-
Myers Squibb Co. (BMS) for an ongoing medicinal chemistry
project. In addition, succinates 39 and 40 were also provided
to BMS as intermediates toward potential drug candidates in
an entirely different project. The fact that three unsymmetrical
succinate derivatives were useful in two different medicinal
chemistry projects further demonstrates the utility of this method
in drug discovery. Compound 45 is an interesting example in
this series due to the facility of the enolate coupling, even in
the presence of an indole nucleus. Indoles are extremely
electron-rich heterocycles and, as such, are susceptible to a
variety of oxidative reaction pathways.38 In fact, N-H indoles,15,30
the related N-H pyrroles, and even an N-alkyl pyrrole
(intramolecular)15,32 are capable of participating in oxidative
couplings with enolates! The efficiency with which dihydro-
coumarin 35 is formed implies that lactones are also compatible,
which is not surprising since these moieties likely behave
electronically similarly to their acyclic counterparts. Compounds
47 and 48 demonstrate that modifications of the aromatic core
on the oxazolidinone are acceptable, as is the incorporation of
an alkyl chain in lieu of the aryl nucleus (39-41). The
trifluoromethyl group (38 and 39), indoles (45), biphenyls (44),
and arylfluorides (49 and 50) are all extremely important
pharmacaphores, again suggesting that this method will likely
find use in the pharmaceutical industry.
It is important to note that the syntheses of unsymmetrical
2,3-dialkylsuccinic acid derivatives in Table 3 proceed with
diastereoinduction in accord with that expected from an alky-
lation of the same oxazolidinones, affording a single epimer at
the oxazolidinone R-carbon. Significantly, an anecdotal mention
was made in the original communication of this work that the
oxidative coupling reactions using oxazolidinone 2 proceed with
the opposite stereoinduction at the oxazolidinone R-carbon, as
would be expected (see Scheme 3). At that time, this anomaly
was not understood because there was no reason to expect
the bond formation to occur on the same face as the bulky
steric element of the auxiliary. Clarity to this end has since
been achieved. The pKa of the product N-(phenylacetyl)-
oxazolidinones is apparently still low enough that the products
formed from the heterocoupling reactions are susceptible to a
deprotonation-protonation event, resulting in net epimerization
under the reaction conditions. This was confirmed when a set
of the minor diastereomers from the coupling reaction (14a,
R-phenyl as pictured) were isolated, treated with LDA, and
quantitatively epimerized to the ꢀ-phenyl oxazolidinones 14b,
with no change in diastereomeric ratio at the methyl-bearing
center (Scheme 3). With these results, it is assumed that the
bond-forming process initially occurs directed away from the
steric element on the oxazolidinone and is subsequently epimer-
ized. This result also demonstrates that a single oxazolidinone
enantiomer can provide controllable access to two diastereomeric
sets of products.
Unsymmetrical 2,3-dialkylsuccinic acid derivatives are useful
compounds within medicinal chemistry,8 and the rapid and
efficient synthesis of these molecules has traditionally been a
significant chemical challenge. While many of the solutions
devised have been elegant from a synthetic standpoint,9,36 the
application of the current method to this problem could greatly
streamline access to these chemical entities. As shown in
Scheme 2, the heterocoupling of an oxazolidinone with an ester
would allow for direct access to hypothetical succinyl derivative
30 in the acid/ester oxidation state. Judicious selection of
coupling partners would provide a means of carbonyl dif-
ferentiation to further manipulate and derivitize the molecules
in subsequent synthetic operations. The presence of an additional
saturated carbon atom between the carbonyl and the phenyl ring
of oxazolidinone 28, while a seemingly subtle change, caused
a few unanticipated challenges, necessitating several changes
to the procedure as outlined above. The first interesting
observation was that Fe(III) is not capable of facilitating
oxidative coupling. When Cu(II) was employed instead, the
homodimerization of 28 was suddenly very competitive with
the desired heterocoupling. While the addition of lithium
chloride and use of toluene as a cosolvent seemed to slightly
ameliorate this problem, a modest excess of the ester coupling
partner (1.75 equiv) was employed in order to statistically
facilitate heterodimerization. In addition, the pKa of the R-carbon
is raised substantially (likely 6-7 units37), thereby retarding
the rate of enolization of 28. As alluded to earlier (Vide supra),
this modification led to homo/cross-Claisen condensation prod-
ucts with 28 acting as the electrofuge. Enolate formation in
separate reaction vessels completely resolved this setback,
restoring the efficiency of the heterocoupling process.
As illustrated in Table 3, a wide range of both steric
environments and functional groups are tolerated in this reaction.
For example, increasing the steric demand from methyl (31) to
isopropyl (32), to tert-butyl (33), or even to the bulky cyclo-
pentyl carbocycle (34) had no adverse affect on the chemistry.
However, when the branched substitution was shifted to the
ꢀ-carbon of the ester, this component preferentially homodimer-
ized, leading to substantially lower yield of heterocoupled
product 40 (albeit with a bulky oxazolidinone component).
Successful coupling of alkene 36 was notable, as many potential
oxidative side reactions involving the R-olefin could be envi-
sioned. The reaction also proceeded with the inclusion of ethers
(37) or trifluoromethyl groups (38, 39) in the corresponding
monomers. Electron-rich (46-48), -neutral (41-44), and -de-
ficient aromatic units (49, 50) are tolerated as well, with the
latter demonstrating that changing positions of the same
functionality about an aromatic ring is acceptable. Methyl esters
perform equally well as tert-butyl esters (compare 43 and 42),
Both of the methods outlined above proceed in relatively low
diastereoselectivities at the ꢀ-carbon in couplings involving
oxazolidinones. This problem can be circumvented when this
chemistry is strategically implemented in a target-oriented
(36) (a) Xue, C.-B.; et al. J. Med. Chem. 2001, 44, 2636–2660. (b) Kottirsch,
G.; Koch, G.; Feifel, R.; Neumann, U. J. Med. Chem. 2002, 45, 2289–
2293.
(37) Bordwell, F. G.; Harrelson, J. A., Jr. Can. J. Chem. 1990, 68, 1714–
(38) Joule, J. A.; Mills, K. Heterocyclic Chemistry, 4th ed.; Blackwell
1718.
Publishing: Malden, MA, 2000.
9
11552 J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008

Intermolecular Enolate Heterocoupling
A R T I C L E S
Table 3. Scope of 2,3-Dialkylsuccinate Couplings
a Isolated yield. b Diastereomeric ratios determined by 1H NMR integration.
Scheme 3. Diastereoselectivity Discrepancya
Scheme 4. Metal-Based Change in Diastereoselectivitya
a Reagents and conditions: (a) LDA (1.1 equiv), THF, -78 °C (30 min)
to 25 °C (5 min); 1 N HCl, 100%.
a Reagents and conditions: (a) LDA (2.1 equiv), THF, -78 °C (30 min)
to 25 °C (5 min); Fe(acac)3, 16-67%; (b) LDA (2.1 equiv), THF, -78 °C
(30 min) to 0 °C (5 min); Cu(2-ethylhexanoate)2, 25 °C, 9-68%.
synthesis (Vide infra, bursehernin synthesis). The use of alternate
auxiliaries (sultam- or menthol-based) led to inefficient cou-
plings (mostly homodimer of the auxiliary containing monomer)
and did not ameliorate the ꢀ-selectivity. This issue remains a
limitation of the enolate heterocoupling methodology.
ing was sought. A series of observations have been made and
experiments completed toward that end, leading to the most
complete understanding of the oxidative coupling reaction to
date.
As seen in Scheme 4, a subtle but significant contradiction
was noted in the coupling of N-(phenylacetyl)oxazolidinones
51 with propiophenones 52. The diastereomeric ratios of the
methyl-bearing carbon in 53, while low for both metals, were
Mechanism
With the initial investigations, optimization, and scope
exploration complete, a comprehensive mechanistic understand-
9
J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008 11553

A R T I C L E S
DeMartino et al.
Scheme 5. Homodimerization Control Experimentsa
Scheme 6. Single-Electron-Transfer-Induced Intramolecular
Cyclizationsa
a Reagents and conditions: (a) LDA (1.1 equiv), THF, -78 °C (10 min)
to 0 °C (10 min) to -78 °C (10 min) to 25 °C (5 min); Fe(acac)3, 27%; (b)
LDA (2.1 equiv), THF, -78 °C (10 min) to 0 °C (10 min) to -78 °C (10
min) to 0 °C (5 min); Cu(2-ethylhexanoate)2, 25 °C, 28%.
a Reagents and conditions: (a) LDA (1.1 equiv), THF, -78 °C (30 min)
to 25 °C (5 min); Fe(acac)3; (b) LDA (2.1 equiv), THF, -78 °C (30 min)
to 0 °C (5 min); Cu(2-ethylhexanoate)2, 25 °C.
demonstrated to be sufficiently electrophilic to cyclize with
tethered electron-rich olefins.46 Therefore, the ability of Fe(III)
and Cu(II) to promote such a cyclization was investigated. When
the enolate of oxazolidinone 56 is treated with either Fe(III) or
Cu(II), both oxidants promote cyclization to cyclopentane 57
(Scheme 6).47 This evidence strongly suggests that both metal
oxidants indeed execute enolate couplings through two single
electron transfers rather than a concerted two-electron oxidation,
and implies that the copper(II) and iron(III) enolates lead to
electrophilic enolate carbon atoms that may be thought of as at
least reasonable surrogates of carbonyl R-radicals. This result
may seem in conflict with the apparent inability of Fe(III) to
oxidize oxazolidinone 2, but the oxidation potential of the
corresponding enolates is likely different, given the presence
of the phenyl group. Moreover, the homodimerization is a
bimolecular reaction, whereas the cyclization would be unimo-
lecular. As such, a reversible oxidation of the enolate would be
more likely to result in trapping by a tethered nucleophile.
uniformly reversed as a function of the oxidant employed. While
Fe(III) afforded selectivity for ꢀ-stereochemistry, this trend was
reversed when Cu(II) was used, exhibiting a preference for
R-stereochemistry. These data inspired the hypothesis that the
two metals were potentially operating through disparate mech-
anisms and guided experimentation aimed at validation of this
theory.
It seemed reasonable to begin this investigation by performing
control experiments to determine the fate of each carbonyl when
no cross-coupling is possible (Scheme 5). When oxazolidinone
2 was treated with LDA followed by Fe(III), no homodimer 54
was formed. However, Cu(II) was capable of promoting
dimerization, albeit in low yield. Interestingly, when the same
experiments were performed using propiophenone 3, both metals
promoted quantitative dimerization to the 1,4-diketone 55. This
suggests that the cross-couplings employing Fe(III) occur
through an initial oxidation of the ketone enolate, as oxidation
of the oxazolidinone either is not possible or occurs reversibly
at a rate such that none of the bimolecular process can occur.
No such conclusive statement can be made for the Cu(II)
mechanism, but it is certainly likely that a similar deduction is
true, given the yields of the corresponding control experiments.
Although strongly suggestive of initial ketone oxidation by
Cu(II), this experiment does not rule out the possibility of more
than one operable mechanism for this oxidant.
If the supposition that a SET-based mechanism is indeed
operable, the venerable radical-induced cyclopropylmethyl
fragmentation should relay valuable mechanistic information.48
The pioneering studies of Newcomb have shown that radicals
adjacent to the 2,2-diphenylcyclopropyl moiety fragment with
picosecond kinetics, faster than the limit of diffusion, and
therefore would out-compete even the fastest first-order and all
second-order processes.49 By flanking such functionality to the
R-carbons of either coupling partner, fragmentation should be
the dominant pathway (over any bimolecular coupling process).
When cyclopropanes 58 and 60 (synthesized as shown in
Scheme 7) were subjected to the cross-coupling conditions with
their corresponding coupling partners 3 (Scheme 8a) and 2
(Scheme 8b), dissimilar results were once again observed for
the two metals. As anticipated, no cyclopropane-containing
coupled products (homo- or heterodimers) were isolated in either
case. The Cu(II) oxidation of cyclopropane 58 led to appreciable
quantitiess20% combined yieldsof a 1:1 mixture of two ring-
opened products: alkene 59a and diene 59b. As Scheme 9
illustrates, both compounds presumably arise from a concerted
SET oxidative cyclopropyl ring opening to intermediate 69,
While an SET mechanism has been repeatedly alluded to in
the literature for both iron(III)26,27,39,40 and copper(II),16,28,41-44
little proof has been offered to this end.44 Radical trapping
experiments involving the cyclization of the purported R-radical
13
45
onto a pendant R-olefin in order to prove that I2 and TiCl4
operate through SET-based mechanisms have been unsuccessful,
although no such studies have been attempted with Fe(III) or
Cu(II). Discreet oxazolidinone R-radicals (formed through
carbon-halogen bond homolysis) have been independently
(39) (a) Ramig, K.; Kuzemko, M. A.; McNamara, K.; Cohen, T. J. Org.
Chem. 1992, 57, 1968–1969. (b) Cohen, T.; McNamara, K.; Kuzemko,
M. A.; Ramig, K.; Landi, J. J., Jr.; Dong, Y. Tetrahedron 1993, 49,
7931–7942.
(40) Schmittel, M.; Burghart, A.; Malisch, W.; Reising, J.; Soellner, R. J.
Org. Chem. 1998, 63, 396–400.
(41) Kobayashi, Y.; Taguchi, T.; Morikawa, T.; Tokuno, E.; Sekiguchi, S.
Chem. Pharm. Bull. 1980, 28, 262–267.
(46) Yang, D.; Zheng, B.-F.; Gu, S.; Chan, P. W. H.; Zhu, N.-Y.
Tetrahedron: Asymmetry 2003, 14, 2927–2937.
(42) Kawabata, T.; Sumi, K.; Hiyama, T. J. Am. Chem. Soc. 1989, 111,
6843–6845.
(47) For precedent of proton abstraction from carbon radicals, see Brown,
H. C.; Midland, M. M. Angew. Chem., Int. Ed. Engl. 1972, 11, 692–
700.
(43) Porter, N. A.; Su, Q.; Harp, J. J.; Rosenstein, I. J.; McPhail, A. T.
Tetrahedron Lett. 1993, 34, 4457–4460.
(48) (a) Maillard, B.; Forrest, D.; Ingold, K. U. J. Am. Chem. Soc. 1976,
98, 7024–7026. (b) Mathew, L.; Warkentin, J. J. Am. Chem. Soc. 1986,
108, 7981–7984.
(44) Quermann, R.; Maletz, R.; Schaefer, H. J. Liebigs Ann. Chem. 1993,
1219–1223.
(45) Ojima, I.; Brandstadter, S. M.; Donovan, R. J. Chem. Lett. 1992, 1591–
1594.
(49) Newcomb, M.; Johnson, C. C.; Manek, M. B.; Varick, T. R. J. Am.
Chem. Soc. 1992, 114, 10915–10921.
9
11554 J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008

Intermolecular Enolate Heterocoupling
A R T I C L E S
Scheme 7. Cyclopropane Synthesesa
were detected, in spite of the known propensity of iron(III) salts
to be single-electron oxidants.40,51 This suggests that, while both
metal enolates are electrophilic at the R-carbon, the reactivity
of copper(II) enolates more closely resembles that of discreet
carbonyl R-radicals than that of iron(III) enolates. That being
said, these data, in combination with the intramolecular cy-
clization results discussed above, suggest that both of these
metals lead to enolates electrophilic at the R-carbon atom and,
more importantly, operate through SET-type mechanisms.
Having sufficiently demonstrated that an SET-based mech-
anism is likely operable, a clearer picture of the bond-forming
process was desired. As discussed earlier, the oxidant-induced
change in diastereoselectivity in the coupling of oxazolidinones
51 and propiophenones 52 is strongly suggestive of disparate
mechanisms (Scheme 4). Two distinct mechanistic possibilities
include biradical (or non-lithium enolate) coupling and a
metal-chelated mechanism. Enolate homodimerizations using
iron(III)39 and copper(II)16,28,42 have both been explained by the
former, while the latter has been cursorily mentioned for
Cu(II)43,52 and was also found to be operable for indole-enolate
couplings.15 When the yields for the couplings between oxazo-
lidinone 2 and propiophenone 3 (as oxidant stoichiometry is
altered) are adjusted to reflect oxidant efficiency (based on an
SET mechanism)53 and the data are again plotted as a function
of oxidant stoichiometry, another distinction between the two
oxidants is unearthed (Chart 2): Cu(II) is an exceedingly efficient
oxidant for this reaction at low oxidant stoichiometry, but this
effectiveness decreases linearly as the Cu(II) stiochiometry is
increased. This trend is consistent with a metal-templating effect
in the bond formation. At low oxidant concentration, there is a
much higher probability that the two enolate coupling partners
would adhere to the same copper atom. At higher oxidant
concentration, there is increased likelihood of non-productive
Cu(II)-enolate interactions. There is no such trend for the
Fe(III)-based couplings: the oxidant efficiency is essentially
consistent through changes in oxidant stoichiometry, suggesting
that the bond-forming process is not aided by a metal chelate.
a Reagents and conditions: (a) Et2Zn (5 equiv), CH2I2 (10 equiv), CH2Cl2,
0 to 25 °C, 15 h, 78%; (b) (COCl)2 (2.5 equiv), DMF (2 drops), 0 to 25 °C,
2.5 h; (c) n-BuLi (1.01 equiv), 65 (1.0 equiv), THF; 64 (1.0 equiv), 80%;
(d) Fe(acac)3 (0.03 equiv), PhMgBr (1.3 equiv), -78 °C, THF, 66%. DMF
) dimethylformamide.
Scheme 8. Single-Electron-Transfer-Induced
2,2-Diphenylcyclopropylmethyl Openinga
The experiments detailed above were designed in order to
gain insight into the mechanisms of Fe(III)- and Cu(II)-promoted
oxidative heterocouplings, and the evidence accrued suggests
that the metals are operating by different mechanisms. In the
case of Fe(III), the following mechanism is proposed (Scheme
10). The ketone lithium enolate 71 is transmetalated to the
iron(III) enolate 74, an entity that can be thought of as an
oxidized enolate, where the unpaired electron is delocalized over
the four-atom system, as in 73. Once this transmetalation occurs,
the electronics of the enolate are altered such that the R-carbon
is now electrophilic and therefore susceptible to attack by the
lithium enolate of the oxazolidinone in the presumed bimolecular
rate-determining step. The consequence of this polarity reversal
is explained in Scheme 11. The approach of the coupling
partners can be depicted in a Newman projection in which there
are six possible staggered projections that can be drawn (two
diastereotopic sets of three). The two that likely dominate the
reaction would align the lithium enolate dipole and the umpol-
ung ferric enolate dipole in the same direction while positioning
a Reagents and conditions: (a) LDA (2.1 equiv), THF, -78 °C (10 min)
to 0 °C (10 min) to -78 °C (10 min) to 25 °C (5 min); Fe(acac)3; (b) LDA
(2.1 equiv), THF, -78 °C (10 min) to 0 °C (10 min) to -78 °C (10 min)
to 0 °C (5 min); Cu(2-ethylhexanoate)2, 25 °C.
Scheme 9. Cyclopropane Fragmentation
followed by either proton abstraction (59a)50 or the removal of
a second electron (59b). Cyclopropane 60 was similarly opened
and oxidized with Cu(II), affording diene 61 in 14% yield (only
trace amounts of the corresponding proton extraction product,
not pictured, were observed in this case). Interestingly, when
the same experiments were performed with Fe(III) as the
oxidant, only trace quantities (<5%) of ring-opened products
(51) Lide, D. R.; Frederikse, H. P. R. CRC Handbook of Chemistry and
Physics: A Ready-Reference Book of Chemical and Physical Data;
CRC: Boca Raton, FL, 1995.
(52) Chung, S. K.; Dunn, L. B., Jr. J. Org. Chem. 1983, 48, 1125–1127.
(53) Oxidant efficiency, or adjusted yield, is defined as the yield of oxidative
coupling divided by the maximum theoretical yield based on an SET
mechanism (one electron transferred per molecule of metal).
(50) For precedent of proton abstraction from carbon radicals, see: Brown,
H. C.; Midland, M. M. Angew. Chem., Int. Ed. Engl. 1972, 11, 692–
700.
9
J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008 11555

A R T I C L E S
DeMartino et al.
Chart 2. Adjusted Coupling Yield (2 and 3) as a Function of Oxidant Stoichiometry
Scheme 10. Proposed Fe(acac)3 Mechanism
Scheme 11. Newman Projection Alignment
approach would favor formation of the anti-diastereomer 4a,
as observed, and is consistent with the stereochemical result of
the analogous homodimerization of two “discreet radicals”.55
Conversely, a potential steric clash between the iron ligand
sphere and the proximal phenyl ring would likely raise the
relative energy of 76 and lead to the observed minor syn-
diastereomer 4b. The electrophilic ferric enolates are then
attacked by the lithium oxazolidinone enolates to afford radical
anions 77 and 78, from which one additional electron is removed
by another molecule of Fe(III) to provide 4a and 4b.
the opposite charge-bearing carbons in close, bond-forming
proximity.54 This rationale narrows the possible Newman
projections to 75 and 76.55 The low diastereoselectivities
obtained at the methyl-bearing carbon are indicative of the
admittedly subtle effects differentiating the two modes of
approach. In the favored transition state 75, the presumably
bulky iron ligand sphere is as far from any other steric element
as possible, and a potentially beneficial π-stacking interaction
between the two aromatic rings is geometrically feasible
(depending on the electronics of the aryl rings). This type of
The proposed mechanistic picture changes significantly when
Cu(II) is instead employed as the oxidant. In this case, although
it is not clear which enolate is transmetalated (as Cu(II) promotes
the homodimerization of both monomers, see Scheme 12), it
seems more likely that the ketone enolate is exchanged, based
on the corresponding efficiencies of the control experiments.
This turns out to be inconsequential, as both enolates are
appended to a single copper atom in the presumed rate-limiting
step: the second-order complexation of the copper enolate with
the remaining lithium enolate to yield diastereotopic transition
(54) For a discussion on intermolecular dipole interactions, see: Atkins,
P. W. Physical Chemistry; Oxford University Press: Oxford, UK, 1995.
(55) Matsumura, Y.; Nishimura, M.; Hiu, H.; Watanabe, M.; Kise, N. J.
Org. Chem. 1996, 61, 2809–2812.
9
11556 J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008

Intermolecular Enolate Heterocoupling
A R T I C L E S
Scheme 12. Proposed Cu(2-ethylhexanoate)2 Mechanism
Scheme 13. Cu(II)-Chelated Transition States Leading to 35 and 35a
structures 81 and 82. Once again, subtle steric and electronic
effects can be invoked to explain the observed diastereoselection.
Complex 81 shows that the new carbon-carbon bond would
be formed through an energetically favorable chair-like transition
state, affording the observed major syn-diastereomer 4b, mini-
mizing the steric clash between the methyl group and the top
face of the oxazolidinone ring and potentially benefiting from
a favorable π-stacking interaction between the two aryl rings.
In contrast, complex 82 proceeds through a higher energy boat-
like transition state with considerable steric clash between the
aryl ring and the hydrogen atoms atop the auxiliary, leading to
the observed minor anti-diastereomer 4a. The exact nature of
the second oxidation is still unclear, with a minimum of three
possibilities. Radical anions 83 and 84 could be oxidized by
the pendant copper(I) species, expelling copper(0). This seems
an unlikely prospect, as no deposition of metallic copper(0) is
observed from the reaction mixture. In addition, while copper(I)
has been employed in carbanion and enolate homodimerization,
it is exercised in conjunction with a cooxidant such as O256 or
hexanoate ligands bridging the two bonded copper atoms.58
Moreover, the dimeric structure persists in a solution of
acetonitrile.59 With the presence of the copper-copper bond,
copper(II) oxidation of the copper(I) radical anions 83 and 84
seems to be the most likely conclusion of the mechanism.
If the copper-chelated mechanism is correct, then it should
be able to explain the diastereoselectivity observed in the
syntheses of unsymmetrical succinates (Table 3). Although the
reaction proceeded with minimal selectivity for linear esters,
the diastereomeric ratio was elevated (albeit still modest) for
lactone 35, at 2.4:1. Examining the potential transition states
reveals that analogous chair- and boat-like transition states (85
and 86, respectively) effectively predict the stereochemical
outcome of the reaction (Scheme 13). Once again, chair-like
transition state 85, while inherently lower in energy, also poses
no steric clash with the underside of the chiral auxiliary and
could potentially benefit from a favorable π-stacking interaction,
leading to the observed major diastereomer. The higher energy
boat-like transition state 86 includes an adverse steric relation-
ship with the dihydrocoumarin and bottom face of the oxazo-
lidinone, thus leading to the observed minor diastereomer, 35a.
57
I2 and has been shown to be incapable of acting as the lone
oxidant.52 These examples imply that copper in its 2+ oxidation
state is indeed the active oxidant and has thus been most
commonly utilized for enolate couplings. Alternatively, the
copper(I) radical anions could transmetalate to another molecule
of Cu(II), followed by the second SET process. This also seems
doubtful, as the relative kinetics for the mechanism of the
second-order transmetalation would be too slow to satisfy the
presumably high-energy radical anions. Lastly, and most
compelling, a second molecule of copper(II) could reoxidize
the pendant copper(I), which would then complete the net two-
electron oxidation. This process seems, a priori, bimolecular
as well, and therefore kinetically too slow for the removal of
the second electron. However, X-ray analysis of copper(II)
hexanoate, assumed to be structurally related, reveals a dimeric
paddlewheel structure in the crystalline state, with the four
Hammett analysis has historically been one of the most
valuable tools aiding mechanistic elucidation, often providing
clarity with regard to the rate-determining step and transition-
(56) (a) Kauffmann, T.; Beissner, G.; Berg, H.; Koeppelmann, E.; Legler,
J.; Schoenfelder, M. Angew. Chem., Int. Ed. 1968, 7, 540–541. (b)
Kauffmann, T.; Beissner, G.; Koeppelmann, E.; Kuhlmann, D.; Schott,
A.; Schrecken, H. Angew. Chem., Int. Ed. Engl. 1968, 7, 131–132.
(57) Hampton, K. G.; Christie, J. J. J. Org. Chem. 1975, 40, 3887–3889.
(58) (a) Doyle, A.; Felcman, J.; Gambardella, M. T. d. P.; Verani, C. N.;
Tristao, M. L. B. Polyhedron 2000, 19, 2621–2627. (b) Mishra, S.;
Daniele, S.; Hubert-Pfalzgraf, L. G. Chem. Soc. ReV. 2007, 36, 1770–
1787.
(59) Lah, N.; Giester, G.; Lah, J.; Segedin, P.; Leban, I. New J. Chem.
2001, 25, 753–759.
9
J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008 11557

A R T I C L E S
DeMartino et al.
Chart 3
Chart 4
state structure.60 These linear free energy relationships between
constant is consistent with the proposed mechanism, wherein
the rate-determining step is the bond-forming process between
the metal enolate 73 and oxazolidinone lithium enolate 72,
yielding radical anions 77 and 78. The relative magnitudes of
the F-values (Fox ≈ 2.6Fprop) reveal deeper insight into the nature
of the transition-state structure. These reaction constants suggest
a late transition state wherein the oxazolidinone enolate 72 has
lost more electron density through nucleophilic attack on the
electrophilic ferric enolate 74 than has the ferric enolate through
FeIII-to-FeII oxidation. This could suggest that the ground-state
iron enolate is “partially oxidized” in a fast equilibrium between
rate constant (krxn/kref) and substituent parameter (σp )
+ 61 allow
analysis of the change in charge distribution as the reaction
proceeds from the ground state to the transition state. Hammett
plots were constructed using the data collected in Table 2 for
the cross-couplings of 4′-substituted N-(phenylacetyl)oxazoli-
dinones with propiophenone 3 (Chart 3) and N-(phenylacetyl)-
oxazolidinone 2 with 4′-substituted propiophenones (Chart 4)
for both oxidants. In the case of Fe(III), linear relationships were
observed when varying substitution on either partner. Altering
the substitution on the oxazolidinone partner revealed a linear
correlation with a F-value (Fox) equal to -0.63 (R2 ) 0.996).62
A similarly linear correlation was observed upon varying the
substitution on the propiophenone coupling partner, providing
a F-value (Fprop) equal to -0.25 (R2 ) 0.934).63 These negative
reaction constants are indicative of a net loss of charge on the
enolate carbon on going from the ground state to the transition
state, as explained by the conversion of the full anion enolates
to the corresponding formal radicals. The small magnitudes are
indicative of one-electron chemistry; a considerably larger
absolute value would be expected if the rate-determining step
proceeds through a two-electron mechanism.60 This reaction
(62) kref is the rate constant for coupling of oxazolidinone 2 and propiophe-
none 3; krxn is the rate constant for coupling of substituted oxazoli-
dinones 51 with propiophenone 3. For the determination of Fox, the
following expression was used: krxn/kref ) log[1- xp/xr]/log[1- yp/
yr], where Fox is the reaction constant, xp is the millimoles of coupled
product formed from substituted oxazolidinone, xr is the initial
millilmoles of propiophenone, yp is the millimoles of product formed
from unsubstituted oxazolidinone, and yr is the initial millimoles of
propiophenone.
(63) kref is the rate constant for coupling of oxazolidinone 2 and propiophe-
none 3; krxn is the rate constant for coupling of oxazolidinone 2 with
substituted propiophenones 52. For the determination of Fprop, the
following expression was used: krxn/kref ) log[1- xp/xr]/log[1- yp/
yr], where Fprop is the reaction constant, xp is the millimoles of coupled
product formed from substituted propiophenone, xr is the initial
millimoles of oxazolidinone, yp is the millimoles of product formed
from unsubstituted propiophenone, and yr is the initial millimoles of
oxazolidinone.
(60) Anslyn, E. V.; Dougherty, D. A. Modern Physical Organic Chemistry;
University Science: Sausalito, CA, 2006.
(61) Swain, C. G.; Lupton, E. C., Jr J. Am. Chem. Soc. 1968, 90, 4328–
4337.
9
11558 J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008

Intermolecular Enolate Heterocoupling
A R T I C L E S
Scheme 14. (a) Lignan Lactone Retrosynthesis and (b) Total
73 and 74 (Scheme 10) and can most accurately be described
as residing within the continuum between the full anion ferric
and formal R-radical ferrous enolates.
Synthesis of (-)-Bursehernina
Interestingly, no such linear correlations were observed for
the corresponding experiments using Cu(II) as the oxidant
(Charts 3 and 4). Given the suggested mechanism (Scheme 12),
such proportionality would not be expected. The proposed rate-
determining step is the complexation of a lithium enolate and
a copper enolate, the end result of which, complexes 81 and
82, are isoelectronic with the ground-state reacting partners at
the carbon atom of the corresponding enolates. Instead, non-
linear plots were observed in both cases, which is generally
indicative of a change in mechanism or rate-determining step
as the substituents are altered.64 As discussed earlier, the control
experiments (Scheme 5) show that Cu(II) is clearly capable of
binding to and oxidizing both partners. As such, one might
anticipate two complementary transmetalation/chelation se-
quences involving 79 and 80 (thus a change in mechanism and
rate-determining step), the diversion of which might be governed
by differing electronics of the corresponding lithium enolates.
Application
The ultimate validation for the development of any new
method is its application to relevant target-oriented synthesis.
Oxidative coupling of two different types of enolates has been
sparingly employed in complex natural product synthesis,23,25,65
likely as a consequence of the absence of an intermolecular
heterocoupling method. By way of demonstration, two simple
case studies were chosen in order to highlight the utility of the
oxidative heterocoupling of enolates.
a Reagents and conditions: (a) 87 (1.0 equiv), LDA (1.15 equiv), LiCl
(5.0 equiv), PhMe, -78 °C (10 min) to 0 °C (10 min) to -78 °C (10 min),
88 (1.75 equiv), PhMe, LDA (1.85 equiv), -78 °C, 30 min, then Cu(2-
ethylhexanoate)2, -78 to 25 °C, 20 min; (b) LiBH4 (10 equiv), MeOH (5.0
equiv), THF, -78 to -10 °C, 1.5 h; (c) DBU (10 equiv), PhMe, 110 °C,
24 h, 41% overall.
Bursehernin (89, Scheme 14b) is a lignan lactone isolated
by McDoniel and Cole in 1972.7 89 was isolated from the
organic extracts of Bursera schlechtendalii, a Mexican plant,
and was found to exhibit antitumor activity against the cancer
test system, 9KB (adenocarcinoma of nasal pharynx). This
lactone is a member of an extremely large class of natural
products known as lignans. Lignans are biosynthetically derived
from the oxidative dimerization of cinnamic alcohols through
single-electron oxidation of phenols.10 Much more complex
lignans are derived from these simple dimers through a vast
array of oxidative transformations, including selective dif-
ferential oxidation of the two aromatic rings. From this
biosynthetic pathway arises symmetric and unsymmetric diben-
zyl lignan lactones, the latter being the category to which
bursehernin belongs. Retrosynthetically, it seemed that an
intuitive method for constructing any lignan lactone would be
through the oxidative coupling of enolates (Scheme 14a). Many
groups have reported the synthesis of symmetric dibenzyl
lactones through use of oxidative enolate homocoupling.27,66
However, enolate heterocoupling provided an avenue to access
unsymmetric dibenzyl lactones by this retrosynthetic strategy
for the first time.
removal of the chiral auxiliary with lithium borohydride afforded
a primary alcohol (not pictured), which was treated with 1,8-
diazabicyclo[5.4.0]undec-7-ene (DBU) in refluxing toluene to
cyclize to the lactone. The basic conditions of this final step
also served to epimerize the R-carbon (C-3), efficiently funneling
to a single diastereomer favoring the R,ꢀ-trans-dialkyl lactone.
The result of this sequence was a three-step, enantio- and
diastereoselective total synthesis of (-)-bursehernin proceeding
in 41% overall yield. It also marked a significant advance in
the synthetic efficiency (both yield and number of steps) of
unsymmetric dibenzyl lignan lactones,67a,b adding to Sibi’s
elegant approach to lignans, with which these lactones might
also be rapidly accessed.67c It is noteworthy that lignan lactones
have been synthetically elaborated to many lignan subclasses,10
and thus, this advance represents a general entry into this family
of natural products.
Matrix metalloproteinases (MMPs) are a family of zinc-
containing enzymes known to aid the degradation and recon-
struction of connective tissue.8 As such, MMPs have been active
targets of pharmaceutical research toward the cures for onco-
logical, cardiovascular, and inflammatory diseases. Discovered
22 years ago, 2,3-dialkylsuccinylhydroxamic acid derivatives
91 (Scheme 15a) have proven one of the most broadly potent
class of MMP inhibitors and have dominated the therapeutic
strategies ever since.8 Succinic half-esters/amides 90 are
As shown in Scheme 14b, the oxidative union of oxazolidi-
none 87 and tert-butyl ester 88 proceeded smoothly, furnishing
47 in 58% yield, as an inconsequential (Vide infra) 1.6:1 mixture
of diastereomers (see Table 3). Chemoselective reductive
(64) Carpenter, B. K. Determination of Organic Reaction Mechanisms;
Wiley: New York, 1984.
(65) Martin, C. L.; Overman, L. E.; Rohde, J. M. J. Am. Chem. Soc. 2008,
130, 7568–7569.
(67) (a) Honda, T.; Kimura, N.; Sato, S.; Kato, D.; Tominaga, H. J. Chem.
Soc., Perkin Trans. 1994, 1, 1043–1046. (b) Enders, D.; Lausberg,
V.; Del Signore, G.; Berner, O. M. Synthesis 2002, 515–522. (c) Sibi,
M. P.; Liu, P.; Ji, J.; Hajra, S.; Chen, J.-X. J. Org. Chem. 2002, 67,
1738–1745.
(66) (a) Belletire, J. L.; Fry, D. F. J. Org. Chem. 1987, 52, 2549–2555. (b)
Kise, N.; Ueda, T.; Kumada, K.; Terao, Y.; Ueda, N. J. Org. Chem.
2000, 65, 464–468.
9
J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008 11559

A R T I C L E S
DeMartino et al.
Scheme 15. (a) Hydroxamic Acid Synthesis and (b) Synthesis of
two-step sequence, proceeding in 56% overall yield, is a highly
efficient approach to the core of this inhibitor and further
demonstrates the power and utility of oxidative intermolecular
heterocoupling of enolates.
MMP Inhibitor Intermediatea
Conclusions
The power and potential utility of oxidative enolate coupling
was unveiled upon its initial discovery in 1935. This admittedly
non-intuitive chemistry provides a one-step means of forging a
carbon-carbon bond directly, with no need for superfluous
prefunctionalization of the monomeric starting materials. The
reaction was limited to homodimerization for 38 years, when
the only two studies involving the analogous cross-coupling
were reported. Surprisingly, these accounts were not expanded
upon until this research program was initiated in 2006, geared
at broadening the utility of the chemistry through an intimate
understanding of the reaction. Carefully designed exploratory
investigations have illuminated the strengths and limitations of
this method and have led to the most complete mechanistic
understanding of this type of chemistry to date. Exploiting the
innate oxidation state of the monomeric carbonyl species allows
for highly convergent access to complex molecules that would
be difficult to procure by other chemical methods. This
convergency has been demonstrated through streamlined access
to lignan natural products and medicinally relevant compounds.
While oxidative enolate heterocoupling is a field still in its
infancy, it has been demonstrated to greatly simplify complex
synthetic problems.25,31,33,65,68 It is anticipated that this area of
research will continue to be fruitful and will find wide
application in the synthetic community.
a Reagents and conditions: (a) H2O2 (10 equiv), LiOH (5 equiv), THF/
H2O (3:1), 0 to 25 °C, 36 h.
regularly employed as intermediates in the syntheses of these
drug candidates, but the enantioselective routes to such com-
pounds generally suffer from lengthy sequences and low overall
yields.9,36 This is a particularly debilitating problem in a
discovery setting, where an innumerable quantity of structurally
diverse analogues must be generated in order to appropriately
probe the structure-activity relationships between the inhibitors
and enzymes. For example, guanadinium hydroxamate 93 was
recently found to be a potent dual inhibitor of MMPs and tumor
necrosis factor R, converting enzymes with high water solubility
(Scheme 15b);9 the synthesis of this inhibitor is reported in 12
steps and 7.6% overall yield. Part of the inefficiency of this
sequence can be attributed to the synthesis of succinic half-
ester 92, a key common intermediate for the generation of all
of the reported analogues, which required six steps to procure
in less than 48% yield (yields for two steps not reported).
Oxidative enolate heterocoupling provides an alternate route to
92: coupling of the appropriate oxazolidinone and tert-butyl ester
(not pictured) provides oxazolidinone 41 (see Table 3), the
hydrolysis of which (under non-epimerizing conditions) pro-
ceeded smoothly, thus intercepting the published synthesis. This
Acknowledgment. We thank Drs. D. H. Huang and L. Paster-
nack for NMR spectroscopic assistance, Dr. G. Siuzdak for mass
spectrometric assistance, and Drs. A. Rhinegold and R. Chadha
for X-ray crystallographic assistance. Financial support for this work
was provided by The Scripps Research Institute, Amgen, Astra-
Zeneca, the Beckman Foundation, Bristol-Myers Squibb, DuPont,
Eli Lilly, GlaxoSmithKline, Pfizer, Roche, the Searle Scholarship
Fund, the Sloan Foundation, and the NIH (NIGMS).
Supporting Information Available: Full characterization,
including 1H and 13C NMR spectra, and experimental procedures
for selected compounds; complete refs 11e and 36a; X-ray
crystallographic data, in CIF format. This material is available
free of charge via the Internet at http://pubs.acs.org.
JA804159Y
(68) Baran, P. S.; Guerrero, C. A.; Hafensteiner, B. D.; Ambhaikar, N. B.
Angew. Chem., Int. Ed. 2005, 44, 3892–3895.
9
11560 J. AM. CHEM. SOC. VOL. 130, NO. 34, 2008