Highly regioselective nickel-catalyzed cross-coupling of N -tosylaziridines and alkylzinc reagents

Article abstract of DOI:10.1021/ja505823s

Herein, we report the first ligand-controlled, nickel-catalyzed cross-coupling of aliphatic N-tosylaziridines with aliphatic organozinc reagents. The reaction protocol displays complete regioselectivity for reaction at the less hindered C-N bond, and the products are furnished in good to excellent yield for a broad selection of substrates. Moreover, we have developed an air-stable nickel(II) chloride/ligand precatalyst that can be handled and stored outside a glovebox. In addition to increasing the activity of this catalyst system, this also greatly improves the practicality of this reaction, as the use of the very air-sensitive Ni(cod)₂ is avoided. Finally, mechanistic investigations, including deuterium-labeling studies, show that the reaction proceeds with overall inversion of configuration at the terminal position of the aziridine by way of aziridine ring opening by Ni (inversion), transmetalation (retention), and reductive elimination (retention).

Full text of DOI:10.1021/ja505823s

Article
pubs.acs.org/JACS
Highly Regioselective Nickel-Catalyzed Cross-Coupling of
N‑Tosylaziridines and Alkylzinc Reagents
Kim L. Jensen, Eric A. Standley, and Timothy F. Jamison*
Department of Chemistry, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States
S
* Supporting Information
ABSTRACT: Herein, we report the first ligand-controlled,
nickel-catalyzed cross-coupling of aliphatic N-tosylaziridines
with aliphatic organozinc reagents. The reaction protocol
displays complete regioselectivity for reaction at the less hindered
C−N bond, and the products are furnished in good to excellent
yield for a broad selection of substrates. Moreover, we have
developed an air-stable nickel(II) chloride/ligand precatalyst that
can be handled and stored outside a glovebox. In addition to increasing the activity of this catalyst system, this also greatly improves
the practicality of this reaction, as the use of the very air-sensitive Ni(cod)2 is avoided. Finally, mechanistic investigations, including
deuterium-labeling studies, show that the reaction proceeds with overall inversion of configuration at the terminal position of the
aziridine by way of aziridine ring opening by Ni (inversion), transmetalation (retention), and reductive elimination (retention).
A few years later, the Wolfe group reported the stoichiometric
insertion of palladium into the terminal position of aliphatic
aziridines containing a tethered olefin.10 Based on the relation-
ship between tether length and reactivity, it was postulated that
the alkene was functioning as a directing group, the presence of
which was critical for successful insertion. Only after changing
the electronic nature of the aziridine, by employing the more
strongly activating N-nosyl substituent, did it in one instance
prove possible to promote the oxidative insertion without the use
of the alkene directing group.
INTRODUCTION
■
The transition-metal-catalyzed activation of alkyl halides has
attracted enormous attention over the past decade, wherein
palladium, nickel, and copper in particular have proven to be
exceptionally effective catalysts for many diverse transforma-
tions,1 including C(sp3)−C(sp3) bond-forming reactions.2
Nevertheless, the vast majority of cross-coupling reactions utilize
carbon−halogen or carbon−OSO2R electrophiles; those involv-
ing oxidative addition into a C−N bond of the electrophile have
received much less attention.3
Although these pioneering contributions showed great
promise for the development of catalytic activation of aziridines
in direct cross-coupling reactions, the first actual coupling
reaction remained elusive for another decade. In 2012, the
first direct cross-coupling reaction was reported by the group
of Doyle. In this report, it was shown that styrene-derived
N-tosylaziridines can undergo facile C−C bond formation with
alkylzinc reagents promoted by nickel catalysis (Scheme 1b).11
The reaction proceeded with complete regioselectivity for the
weaker benzylic C−N bond.12 Notably, it was found necessary
to use the electron-deficient alkene ligand, dimethyl fumarate,
to induce reductive elimination and prevent β-hydride abstrac-
tion.13 Shortly thereafter, the Doyle group demonstrated that
the reaction sequence could be extended to include aliphatic
aziridines (Scheme 1c).14 Interestingly, this type of substrate had
proved unreactive under the previously developed conditions.
The activation was elegantly achieved by changing the N-sulfonyl
group from tosyl to cinsyl (Cn), which contains an electron-
deficient alkene in the 2-position of the arylsulfonyl group. By
using this protecting group, the cross-coupling products were
obtained in moderate to good yields as a mixture of inseparable
The aziridine functionality represents a versatile building block
in contemporary synthetic chemistry. Like other three-membered
heterocycles such as epoxides, aziridines possess significant ring
strain that renders them susceptible to nucleophilic ring opening.
In many cases, these reactions proceed with high regioselectivity,
and various carbon- and heteroatom-based nucleophiles have
been investigated over the past several years. In particular, the
addition of organocuprates has proven to be an efficient strategy
for C−C bond-forming reactions with aziridines.4
The oxidative addition of transition metals into aziridines to
form azametallacycle intermediates can also be effected under
mild conditions.5 For instance, arylaziridines have been shown
to undergo facile ring-expanding carbonylation reactions to
form β-lactams under rhodium catalysis.6 This transformation
proceeded with high selectivity for insertion into the benzylic
C−N bond. Notably, simple alkylaziridines proved unreactive
under these conditions.7
In 2002, Hillhouse demonstrated that aliphatic N-sulfonyl-
aziridines could undergo oxidative insertion of nickel to form
stable, isolable azanickelacyclobutane complexes (Scheme 1a).8,9
Interestingly, the oxidative insertion occurred with complete
regioselectivity for insertion into the less substituted aziridine
C−N bond. Studies of a deuterium-labeled aziridine provided
clear evidence for an SN2-type mechanism.
Received: June 10, 2014
Published: July 23, 2014
© 2014 American Chemical Society
11145
dx.doi.org/10.1021/ja505823s | J. Am. Chem. Soc. 2014, 136, 11145−11152

Journal of the American Chemical Society
Article
with excellent stereospecificity and stereochemical inversion.
Interestingly, the enantiopurity is maintained through the
reaction, enabling the synthesis of tertiary stereogenic centers
(Scheme 1e).
Scheme 1. Previously Reported Activation Studies and
Cross-Couplings Reactions of Aziridines
While all of the above-mentioned protocols provide access
to a wide variety of sulfonamide products arising from different
combinations of aziridines and coupling partners (aryl/alkyl,
alkyl/aryl, alkyl/alkyl, aryl/aryl), no C−C bond-forming reaction,
with complete regioselectivity, of unactivated aliphatic aziridines
and aliphatic coupling partners has been reported to date.
Therefore, we decided to study the nickel-catalyzed cross-
coupling of aliphatic N-tosylaziridines with aliphatic organozinc
reagents (Scheme 1f).17 The reaction protocol developed
through these studies displays complete regioselectivity for the
terminal position, and the products are furnished in good to
excellent yield for a broad selection of substrates. Moreover, we
have developed an air-stable nickel(II) chloride/ligand pre-
catalyst that can be handled and stored outside a glovebox. This
greatly improves the practicality of this reaction, as the use of the
very air-sensitive Ni(cod)2 can be avoided. Finally, mechanistic
studies, including deuterium-labeling studies to determine the
stereochemical course of the reaction, were performed.
RESULTS AND DISCUSSION
■
We initiated our studies by examining the reaction of methyl-
substituted aziridine 1a using the bipyridine Ni(0) complex
employed in the stoichiometric studies by Hillhouse.8,18 In terms
of the aliphatic coupling partner, organozinc reagents were chosen
as they generally display high stability and improved functional
group compatibility.19 Furthermore, a number of organozinc
reagents are commercially available as THF solutions, and we
therefore began our studies using these reagents. Employing
20 mol % of the complex provided the cross-coupled product 2 as
a single regioisomer in 60% isolated yield (Scheme 2). Analysis of
regioisomers, with C−C bond formation on the least substituted
side being favored, though in only moderate regioselectivity
(<5:1). The majority of the scope focused on arylzinc reagents;
however, a few aliphatic examples were also given. The newly
designed Cn-sulfonyl group is likely to serve a dual purpose in
this transformation by acting as a directing group for the
oxidative insertion into the aziridine and by inducing reductive
elimination in analogy to the dimethyl fumarate employed
in the protocol for styrenylaziridines. Interestingly, in contrast
to the stoichiometric studies by Hillhouse8 and Wolfe,10a when
a deuterium-labeled aziridine was subjected to the reaction
conditions, a mixture of diastereoisomers, arising from epimeri-
zation at the terminal position, was obtained. This is indicative
of a single-electron transfer (SET) mechanism or an SN2-type
activation followed by Ni−C bond homolysis.
Soon after, the group of Michael reported a highly efficient,
palladium-catalyzed cross-coupling of aliphatic N-nosylaziridines
with arylboronic acids (Scheme 1d).15 The transformation
proceeds with complete regioselectivity for the terminal
position, and the products are obtained in good yields. The
application of a phenol additive was found to be critical in order
to promote transmetalation and to prevent a detrimental
β-hydride elimination pathway. Through a deuterium-labeling
study, the reaction was shown to proceed with complete
inversion, which is consistent with the observations previously
made on the stoichiometric systems. More recently, Minakata
and co-workers demonstrated a stereo- and regiospecific
palladium-catalyzed coupling of arylaziridines and arylboronic
acids.16 Through the use of a Pd/NHC catalyst system, the
C−N bond activation occurs at the more substituted position
Scheme 2. Initial Conditions
the crude reaction mixture after workup showed that the prod-
uct was formed in a 2.2:1 ratio to TsNH2. This byproduct pre-
sumably arises from β-hydride abstraction and later hydrolysis of
the resulting imine.20 Reducing the catalyst loading to 10 mol %
resulted in similar product distribution and yield; however, the
reaction time required for full consumption of the aziridine was
significantly increased (44 h).
Consequently, to improve the yield and efficiency of the
reaction, a series of bidentate pyridine-derived ligands were
screened in combination with Ni(cod)2 (Table 1). Changing
the 2,2′-bipyridine (bpy) ligand to the more electron-rich
4,7-(MeO)2(bpy) provided an increase in product ratio (3:1).
Notably, the reaction was less clean and a trace of halide-opened
aziridine was observed.21 Employing 1,10-phenanthroline
(phen) as the ligand did not improve the product distribution;
however, the reaction did turn out to be much faster and
cleaner. Further screening of phenanthroline derivatives
identified 3,4,7,8-tetramethyl-1,10-phenanthroline (Me4phen)
as the best ligand, improving the ratio to 4.5:1 and an isolated
11146
dx.doi.org/10.1021/ja505823s | J. Am. Chem. Soc. 2014, 136, 11145−11152

Journal of the American Chemical Society
Article
a
Table 1. Screening of Ligands
Table 2. Optimization of Ligand-to-Nickel Ratio
a
b
Me4phen/Ni
convn
ratio
yield (%)
1
2
3
4
5
1:1
1.25:1
1.50:1
1.75:1
2.0:1
>95
>95
>95
>95
>95
3.9:1 (80%)
4.6:1 (82%)
5.3:1 (84%)
6.0:1 (86%)
6.0:1 (86%)
77
79
80
nd
nd
a
Yield in parentheses refers to the maximum theoretical yield based on
b
the product distribution. All yields were determined after purification
by column chromatography on silica.
versus amount of ligand employed was provided by the
precatalyst containing 1.25 equiv of ligand to nickel, furnishing
the product in 79% yield.
a
Asterisk (∗) denotes the formation of halide-opened aziridine byproducts.
yield of 70%. Notably, low reactivity was observed for the ortho-
substituted 2,9-dimethyl-1,10-phenanthroline ligand.
With this precatalyst and these conditions in hand, we started
to evaluate different aziridines. However, the reaction of
aziridines with longer and more complex substituents turned
out to work less well, providing incomplete conversions and
various ratios of TsNH2. For instance, a chlorobutyl-substituted
aziridine was tested, and the product was obtained in a yield of
67% along with significant amounts of TsNH2. Consequently,
we decided to study the effect of solvent additives (Table 3).
In order to improve the practicality of the procedure and
avoid the use of Ni(cod)2, we decided to develop a precatalyst
system that can tolerate air and atmospheric moisture.22,23
Furthermore, studies by Hillhouse on the synthesis of azanickela-
cyclobutanes had demonstrated that a cod-free Ni(0) system
underwent insertion into the aziridine at a faster rate than
when cod was present. This improved reactivity has also been
demonstrated in a number of other instances. Consequently,
we desired to identify a cod-free nickel precatalyst system
for this transformation. Attempts to use NiCl2·DME as the
nickel source led to only low conversion, even after pro-
longed reaction times, and therefore, a different nickel source
was required.
a
Table 3. Optimization of Reaction Parameters
It is well-established that the combination of bpy- or phen-
derived ligands with nickel(II) salts in aqueous or ethanolic
solutions can yield nickel ligand adducts.24 However, the
identity of the counterion, temperature of synthesis, and solvent
system can cause significant differences in the composition
of the complex formed. For instance, the combination of
phenanthroline and NiCl2·6H2O in ethanol initially yields a
kinetic complex of the structure (phen)3NiCl2·xH2O, even when
the ligand and nickel sources are mixed in a 1:1 ratio.24a,b Upon
extended reaction time at ambient temperature or upon heating,
a complex of the nominal composition (phen)NiCl2·xH2O is
subsequently formed. We found that precatalyst mixtures
prepared at ambient and at elevated temperatures gave vastly
different outcomes for the catalytic coupling reaction. As
established by extensive elemental analyses, the root cause for
the difference in outcome was attributed to either an excess of
ligand, causing a beneficial outcome, or an excess of nickel, which
was significantly detrimental to the reaction. To neutralize these
differences, a different approach for the precatalyst synthesis was
devised. Rather than isolating the precatalyst as a solid by
filtration from the ethanol solution in which it is synthesized, the
solvent was removed by rotary evaporation, yielding a solid with
a precisely known ratio of ligand to nickel. A series of precatalysts
with varying ratios of ligand (Me4phen) to nickel were
synthesized and evaluated in the coupling reaction (Table 2).25
In general, all of the precatalysts behaved well, furnishing
the sulfonamide product 2 with high selectivity. The ratio of
product to TsNH2 generally improved with increasing amounts
of ligand. However, the best result in terms of isolated yield
R
mol %
solvent
concn (M) yield (%)
1
2
3
4
5
6
Cl(CH2)4
Cl(CH2)4
Cl(CH2)4
Cl(CH2)4
Cl(CH2)4
Me
10
10
10
10
5
THF
0.15
0.125
0.125
0.1
67
84
66
77
88
85
b
THF/DCM (3:1)
THF/DCE (3:1)
THF/DCE (3:2)
THF/DCE (3:2)
THF/DCE (3:2)
0.1
5
0.1
a
All yields were determined after purification by column chromatog-
raphy on silica. An inseparable byproduct derived from the
organozinc reagent, 3-methyl-1-butanol, was formed.
b
Adding CH2Cl2 to the reaction had a beneficial effect on the
outcome, providing the product in 84% yield. Unfortunately,
however, an inseparable byproduct (3-methyl-1-butanol) derived
from the organozinc reagent was also formed. The formation of
this byproduct was prevented by employing 1,2-dichloroethane
(DCE) as an additive, resulting in a yield of 77%.26 The last
improvement was achieved by lowering the catalyst loading to
5 mol %, providing the product in 88% yield. Gratifyingly, these
new conditions were also applicable to the initially tested
methylaziridine, affording 2 in 85% yield. The optimized conditions
were first evaluated by varying the aziridine component.27
As demonstrated in Table 4, the reaction worked well for
a wide variety of aziridines. Linear and branched aliphatic
substituents all provided the products 3−6 in good to excellent
yields (66−96%). The α-branched cyclohexyl substrate required
a slightly elevated temperature to go to completion. Substituents
containing additional functional groups were also well-tolerated,
11147
dx.doi.org/10.1021/ja505823s | J. Am. Chem. Soc. 2014, 136, 11145−11152

Journal of the American Chemical Society
Article
a
Table 4. Evaluation of Aziridines
Table 5. Evaluation of Commercially Available Organozinc
Reagents in THF
a
a
Reactions were performed on 0.50 mmol scale, and yields were
determined after purification by column chromatography on silica gel
b
(see Supporting Information for details). 1,10-Phenanthroline was
employed as ligand.
functional group that can serve as entry to ketones and amines.
Notably, the reaction did not proceed when cyanoethylzinc
bromide was used. Instead, polymerization of the reaction mixture
was observed, presumably due to the formation of acrylonitrile.
Finally, an acetal-containing organozinc reagent was evaluated.
The products were obtained with high yields (76 and 80%) for
both an aliphatic and benzylic aziridine (24 and 25).
As a final improvement of the developed cross-coupling
reaction, we decided to expand the reaction protocol to include
the use of more easily accessible organozinc reagents made in
DMA,29 without the use of highly activated (Rieke-type) zinc
(Table 6).30 Initial attempts to use these reagents showed an
a
Reactions were performed on 0.50 mmol scale, and yields were
determined after purification by column chromatography on silica gel
(see Supporting Information for details).
including an alkyl chloride (7), ester (8), and a phthalimide-
protected amine (9). Investigation of benzylic substituents was
of high priority, as the resulting products complement the
motif obtained by the previously published methods that use
arylzinc reagents or arylboronic acids as coupling partners.14,15
Fortuitously, the benzyl-substituted aziridines turned out to
work especially well, furnishing the products in excellent yields
(84−97%). In general, the outcome seemed to be unaffected
by both the electronic properties and the substitution pattern
of the aromatic ring. Both electron-withdrawing (11 and 12)
and electron-donating (13−16) substituents were tolerated,
and the substitution pattern did not affect the outcome. A
heterocyclic indole-based substrate also worked well, affording
the product 17 in 93% yield. This particular substrate was also
evaluated without the N-Boc-protecting group; however, this
did not lead to the formation of any of the desired coupling
product.
Table 6. Evaluation of Organozinc Reagents Synthesized in
a
DMA
Next, a number of organozinc reagents were evaluated.28
The reaction worked well for both alkene- and arene-containing
organozinc reagents, furnishing the products 19−21 in good yields
(73−85%) (Table 5). Interestingly, the use of 2-phenethylzinc
bromide led to significant formation of styrene when the
Me4phen ligand system was employed. By replacing Me4phen
with phenanthroline, the formation of styrene was significantly
suppressed and led to a considerably cleaner reaction outcome.
The more sterically hindered cyclohexylmethylzinc reagent
also underwent cross-coupling to furnish the product 22 in
72% yield. As with the cyclohexylaziridine, a slightly higher
temperature was required. Importantly, the reaction proved
tolerant of a nitrile-containing organozinc reagent, furnishing
the product 23 in 70% yield. The nitrile group is a very versatile
a
Reactions were performed on 0.50 mmol scale, and yields were
determined after purification by column chromatography on silica gel
(see Supporting Information for details). 3,4,7,8-Tetramethyl-1,10-
phenanthroline was employed as ligand. DME used in place of DCE.
b
c
almost complete lack of reactivity, suggesting that DMA could
be an unsuitable solvent for this transformation. However,
when the reaction of a commercially purchased organozinc
11148
dx.doi.org/10.1021/ja505823s | J. Am. Chem. Soc. 2014, 136, 11145−11152

Journal of the American Chemical Society
Article
reagent in THF was spiked with DMA, the outcome was not
significantly altered. It is well-known that lithium halides can
combine with organozinc reagents to form lithium organozincate
adducts of the structure RZnX2Li.31 These zincates generally
display improved nucleophilicity over standard organozinc
reagents of the type RZnX. This fact led us to suspect that the
commercially available alkylzinc halide reagents, made from
Rieke zinc, may contain trace amounts of lithium halide salts.
Indeed, after a short evaluation of reaction parameters, we found
that the addition of LiCl was critical to the reactivity of the
organozinc reagents in DMA. In the absence of LiCl, the
transmetalation did not occur at an adequate rate to allow the
coupling reaction to proceed.
formed. During the synthesis of these complexes, we observed a
significant difference in the rate of oxidative insertion depending
on the aziridine substituent. For instance, both the PMB- and
methylaziridines reacted much faster than the corresponding
hexyl-substituted aziridine (not shown). Whereas the insertion
occurred at ambient temperature for the two former, elevated
temperatures (60 °C) were found to be necessary for the
latter case.
With these complexes in hand, we decided to study their
ability to undergo transmetalation with the organozinc reagent
(Scheme 4a). At first, we subjected the methyl-substituted
Scheme 4. Studies of Azanickelacyclobutane Complexes
As before, a number of alkylzinc reagents worked well in
combination with a variety of aziridines. Both a linear aliphatic
and a fluoride-containing reagent furnished the products
(26−29) in good to excellent yields (84−93%). An ester
functionality was also tolerated, and the products were formed
in high yields (80−90%) for both an aliphatic and a benzylic
aziridine (30 and 31). The cyanopropyl reagent turned out
to work better in DMA than in THF, providing the products
(23 and 32) in high yields (79−82%). Lastly, a TBS-protected
alcohol was evaluated, furnishing the products (33 and 34) in
high yields (86−93%). It should be noted that the reactivity in
DMA is generally diminished compared to the THF system.
Consequently, the precatalyst derived from 1,10-phenanthroline
was employed for the majority of the reactions.
After having demonstrated that the protocol worked well
with a broad selection of substrates, we decided to study the
reaction in more detail in order to get a better understanding of
the mechanism. Doyle and co-workers have recently described
that the azametallacycle formed from styrenylaziridines and
(bpy)Ni(cod) did not work well in the coupling reaction.11
Only with the addition of external dimethyl fumarate ligand
were they able to obtain the coupling product, which was
formed in a low yield (18%). Consequently, we were interested
in elucidating whether the azanickelacyclobutane described
by Hillhouse is involved in the present reaction and, if so,
whether the oxidative insertion of nickel occurs with inversion.
Moreover, we wanted to determine the overall stereochemical
course of the terminal carbon (C1) in the cross-coupling reaction
sequence through deuterium-labeling studies. We commenced
our studies by synthesizing the expected azanickelacyclobutane
complexes. Following a similar protocol to that described by
Hillhouse,8 two complexes 35 and 36 were synthesized in
excellent yields (86−96%), starting from the Me4phen ligand and
Ni(cod)2 (Scheme 3). The complexes were poorly soluble in
complex 35 to the reaction conditions, mimicking the first
catalytic turnover (based on 10 mol % catalyst) using 30 equiv
of the organozinc reagent. Interestingly, no coupling product
was observed under these conditions. All of the metallacycle
had been consumed with concomitant formation of p-
toluenesulfonamide. The reaction was heterogeneous, and we
therefore hypothesized that the insolubility of the complex
could be the cause of this result. Interestingly, when DCE was
added to the mixture, the solution became much less
heterogeneous, and in this case, the cross-coupling product 2 was
formed in a 1.2:1 ratio to TsNH2 (80% combined NMR yield).
This experiment demonstrates that the azametallacycle is indeed a
competent reaction intermediate.32 Moreover, the results suggest
that the improved reaction outcome when using DCE as a
cosolvent is due to an increased solubility of the catalytic inter-
mediates. The azametallacycle 35 also worked well as a precatalyst
(5 mol %) in the coupling reaction between 1a and fluorobutylzinc
bromide, furnishing the product 37 in 88% yield (Scheme 4b). It
should be noted that the reaction did not work as well if the
additional 1.25 mol % of Me4phen ligand was not addedthe
reaction was slower and did not go to full conversion. This is in
accordance with the results of the optimization that an excess of
ligand relative to nickel is beneficial in this reaction.
Scheme 3. Synthesis of Azanickelacyclobutane Complexes
After having obtained results that support the intermediacy
of the azametallacycle in the reaction mechanism, we decided to
examine the stereochemical course of reaction. In order to do
so, a deuterium-labeled version of the PMB-substituted aziridine
38 was synthesized and subjected to the coupling conditions
(Scheme 5a). The acetal-containing organozinc reagent was
chosen for this study since the resulting coupling product easily
could be transformed in one step to a cyclic enamine structure
40.33 This structure would allow for the determination of stereo-
chemical configuration by NMR analysis. The cross-coupling
product 39 was formed in 75% yield, and upon treatment
with aqueous HCl in acetone, cyclization to the enamine 40
proceeded in 84% yield. Analysis by NMR spectroscopy showed
that the product was formed with essentially complete (>98%)
THF, allowing their isolation by precipitation from the reaction
mixture. In both cases, only the regioisomer derived from
oxidative insertion of nickel into the terminal C−N bond was
11149
dx.doi.org/10.1021/ja505823s | J. Am. Chem. Soc. 2014, 136, 11145−11152

Journal of the American Chemical Society
Article
Scheme 5. Cross-Coupling of Deuterium-Labeled Aziridine
and Synthesis of the Azametallacycle
Scheme 6. Proposed Mechanistic Cycle
inversion at the deuterium-labeled carbon atom.34 This result
is in accordance with the stoichiometric studies by Hillhouse
and in contrast to the studies by Doyle, where epimerization is
observed at C1 in the coupling reaction.14
To further ascertain in which step of the mechanism the
inversion occurs, the corresponding azametallacycle of the
deuterium-labeled aziridine 38 was investigated (Scheme 5b).
Under conditions similar to the cross-coupling reaction, the
azanickelacyclobutane 41 was successfully formed with 95%
inversion at C1, thereby confirming that the stereochemical
inversion occurs in the oxidative insertion step.
On the basis of these mechanistic studies, we propose a
catalytic cycle for the regioselective cross-coupling reaction as
outlined in Scheme 6.
The nickel(II) precatalyst is activated by two successive trans-
metalations of the organozinc reagent followed by reductive
elimination.35 This releases the active nickel(0) catalyst 42,
which undergoes oxidative insertion into the least hindered
C−N bond of the aziridine 1. This step proceeds via an SN2-
type mechanism (inversion) involving attack of nickel at C1
followed by C−C bond rotation (46) and ring closure to
form azanickelacyclobutane 43.36 Subsequent transmetalation32
(by way of retention) with the organozinc reagent forms
intermediate 44, which undergoes reductive elimination (with
retention) to release the product 45 (ZnBr adduct) and a
nickel(0) species which re-enters the catalytic cycle.
molecules.38 However, this stability is also the reason why this
moiety is sometimes avoided, as it can be challenging to cleave.
For that reason, we decided to demonstrate that the tosyl-
protecting group can be reductively cleaved to the free amine if
desired. Following an adapted literature procedure, the
sulfonamide 18 was successfully cleaved using magnesium as
a reductant in the presence of titanium tetraisopropoxide.39
The resulting amine was isolated (after acetylation to form 48)
in 67% yield over two steps (Scheme 7).
Scheme 7. Deprotection of the Sulfonamide
On the basis of our findings, we hypothesize that the active
transmetallating agent is a lithium organozincate, which is
known to be more nucleophilic and generally undergoes faster
transmetalation than unmodified organozinc reagents. For the
reactions performed with DMA as a cosolvent, addition of LiCl
was found to be necessary for product formation. As previously
discussed, the outcome of the reaction is improved when an
excess of ligand-to-nickel is used. Since the oxidative addition
proceeds smoothly in the absence of excess ligand, we hypo-
thesize that the additional ligand may be involved in the
reductive elimination step.
The sulfonamide moiety can be found in a great number of
biologically active molecules, including the APIs of numerous
drugs, such as sildenafil (erectile dysfunction), sultiame
(epilepsy), and brinzolamide (glaucoma).37 It is also considered
a carboxylic acid isostere and has been shown to increase
metabolic stability when introduced into certain peptide-like
As a final extension, we decided to examine if enantioenriched
aziridines could undergo cross-coupling reaction without loss or
erosion of enantiopurity. On the basis of the hypothesized
mechanism, we anticipated that no erosion should take place,
and indeed, when enantiopure benzylaziridine (S)-1i was
subjected to the coupling conditions, the sulfonamide product
(R)-10 was formed in excellent yield (96%) and with preserved
stereochemistry (>99% ee) (Scheme 8).
Scheme 8. Synthesis of Enantiopure Sulfonamide Product
11150
dx.doi.org/10.1021/ja505823s | J. Am. Chem. Soc. 2014, 136, 11145−11152

Journal of the American Chemical Society
Article
D’hooghe, M.; Catak, S.; Eum, H.; Waroquier, M.; Van Speybroeck,
V.; De Kimpe, N.; Ha, H.-J. Chem. Soc. Rev. 2012, 41, 643.
(5) For a recently published review on this topic, see: Huang, C. Y.;
Doyle, A. G. Chem. Rev. DOI: 10.1021/cr500036t.
CONCLUSIONS
■
In summary, we have developed the first ligand-controlled,
nickel-catalyzed cross-coupling of unactivated aliphatic
N-tosylaziridines with aliphatic organozinc reagents. The
reaction protocol displays complete regioselectivity for reaction
at the less hindered C−N bond, and the products are furnished
in good to excellent yield for a broad selection of substrates.
Moreover, the use of the very air-sensitive Ni(cod)2 was
avoided by the development of an air-stable nickel(II) chloride/
ligand precatalyst that can be handled and stored outside a
glovebox. Besides improving the practicality of the protocol, the
exclusion of 1,5-cyclooctadiene from the system also improved
the reactivity of the catalyst. Finally, mechanistic investigations,
including deuterium-labeling studies, show that the reaction
proceeds with overall inversion of configuration at the terminal
position of the aziridine by way of aziridine ring opening
(SN2-type) by Ni (inversion), transmetalation (retention), and
reductive elimination (retention). These results are in contrast
to the previously reported nickel-catalyzed reactions, in which
scrambling of the stereoconfiguration at the terminal carbon
atom is observed.
(6) (a) Alper, H.; Uros, F.; Smith, D. J. H. J. Am. Chem. Soc. 1983,
105, 6737. (b) Calet, S.; Urso, F.; Alper, H. J. Am. Chem. Soc. 1989,
111, 931. (c) Lu, S.-M.; Alper, H. J. Org. Chem. 2004, 69, 3558.
(7) For a stoichiometric Ni(CO)4/LiI-mediated aziridine carbon-
ylation, see: (a) Chamchaang, W.; Pinhas, A. R. J. Org. Chem. 1990, 55,
2943. For cobalt-catalyzed carbonylation reactions of aziridines, see:
(b) Piotti, M. E.; Alper, H. J. Am. Chem. Soc. 1996, 118, 111.
(c) Davoli, P.; Moretti, I.; Prati, F.; Alper, H. J. Org. Chem. 1999, 64,
518. (d) Mahadevan, V.; Getzler, Y. D. Y. L.; Coates, G. W. Angew.
Chem., Int. Ed. 2002, 41, 2781.
(8) Lin, B. L.; Clough, C. R.; Hillhouse, G. L. J. Am. Chem. Soc. 2002,
124, 2890.
(9) For a review on azametallacyclobutanes, see: Dauth, A.; Love, J.
A. Dalton Trans. 2012, 41, 7782.
(10) (a) Ney, J. E.; Wolfe, J. P. J. Am. Chem. Soc. 2006, 128, 15415.
For a related study on the Pd-catalyzed isomerization of N-
tosylaziridines to N-tosylimines, see: (b) Wolfe, J. P.; Ney, J. E. Org.
Lett. 2003, 5, 4607.
(11) Huang, C.-Y.; Doyle, A. G. J. Am. Chem. Soc. 2012, 134, 9541.
(12) For a Rh(III)-catalyzed C−C coupling between arenes and
arylaziridines by C−H activation, see: Li, X. W.; Yu, S. J.; Wang, F.;
Wan, B. S.; Yu, X. Z. Angew. Chem., Int. Ed. 2013, 52, 2577.
(13) For a review on the effect of olefins on transition-metal-
catalyzed cross-coupling reactions, see: Johnson, J. B.; Rovis, T. Angew.
Chem., Int. Ed. 2008, 47, 840.
ASSOCIATED CONTENT
■
S
* Supporting Information
Experimental procedures and spectral data (1H, 13C, 19F as
applicable) for all new compounds. This material is available
free of charge via the Internet at http://pubs.acs.org.
(14) Nielsen, D. K.; Huang, C.-Y.; Doyle, A. G. J. Am. Chem. Soc.
2013, 135, 13605.
(15) Duda, M. L.; Michael, F. E. J. Am. Chem. Soc. 2013, 135, 18347.
(16) Takeda, Y.; Ikeda, Y.; Kuroda, A.; Tanaka, S.; Minakata, S. J. Am.
Chem. Soc. 2014, 136, 8544.
(17) For a review on alkyl-organometallic reagents in cross-coupling
reactions, see: Jana, R.; Pathak, T. P.; Sigman, M. S. Chem. Rev. 2011,
111, 1417.
AUTHOR INFORMATION
■
Corresponding Author
tfj@mit.edu.
Notes
(18) For the synthesis of (bpy)Ni(cod) complex, see: Schwalbe, M.;
The authors declare no competing financial interest.
Walther, D.; Schreer, H.; Langer, J.; Gorls, H. J. Organomet. Chem.
̈
2006, 691, 4868.
ACKNOWLEDGMENTS
■
(19) (a) Organozinc Reagents, A Practical Approach; Knochel, P.,
Jones, P., Eds.; Oxford: New York, 1999. (b) Knochel, P.; Singer, R. D.
Chem. Rev. 1993, 93, 2117.
(20) Other mechanisms such as retro-[2 + 2] to form a nickel-nitrene
and an alkene, radical cleavage, or direct deptronotion by the
organozinc reagent cannot be excluded.
(21) Both the Cl- and Br-opened aziridine byproducts were observed.
It is anticipated that the source of the chloride could be chloride salts
(LiCl) in the organozinc reagents, from the precatalyst, or from DCE
when present.
This work is dedicated to the memory of Gregory L. Hillhouse.
K.L.J. and E.A.S. would like to thank the Lundbeck Foundation
for a postdoctoral fellowship and the NSF for a Graduate
Research Fellowship, respectively. Li Li (MIT) is acknowledged
for HRMS data, which were obtained on an instrument
purchased with the assistance of NSF Grant CHE-0234877.
NMR spectroscopy was carried out on instruments purchased
in part with funds provided by the NSF (CHE-9808061 and
CHE-8915028).
REFERENCES
■
(22) For an overview of the development of nickel(II) precatalysts,
see: Rosen, B. M.; Quasdorf, K. W.; Wilson, D. A.; Zhang, N.;
Resmerita, A.-M.; Garg, N. K.; Percec, V. Chem. Rev. 2011, 111, 1346.
(23) For representative examples of the syntheses and applications of
Ni(II) precatalysts, see: (a) Wolfe, J. P.; Buchwald, S. L. J. Am. Chem.
Soc. 1997, 119, 6054. (b) Ge, S.; Hartwig, J. F. Angew. Chem., Int. Ed.
2012, 51, 12837. (c) Standley, E. A.; Jamison, T. F. J. Am. Chem. Soc.
2013, 135, 1585. (d) Park, N. H.; Teverovskiy, G.; Buchwald, S. L.
(1) Tasker, S. Z.; Standley, E. A.; Jamison, T. F. Nature 2014, 509,
299.
(2) (a) Netherton, M. R.; Fu, G. C. Adv. Synth. Catal. 2004, 346,
1525. (b) Hu, X. Chem. Sci. 2011, 2, 1867.
(3) For representative examples of nickel-mediated C−N bond
activation, see: (a) Wenkert, E.; Han, A.-L.; Jenny, C-.H. J. Chem. Soc.,
Chem. Commun. 1988, 975. (b) Trost, B. M.; Spagnol, M. D. J. Chem.
Soc., Perkin Trans. 1 1995, 2083. (c) Blakey, S. B.; MacMillan, D. W. C.
J. Am. Chem. Soc. 2003, 125, 6046. (d) Maity, P.; Shacklady-McAtee,
D. M.; Yap, G. P. A.; Sirianni, E. R.; Watson, M. P. J. Am. Chem. Soc.
2013, 135, 280. (e) Tobisu, M.; Nakamura, K.; Chatani, N. J. Am.
Chem. Soc. 2014, 136, 5587.
Org. Lett. 2014, 16, 220. (e) Standley, E. A.; Smith, S. J.; Muller, P.;
̈
Jamison, T. F. Organometallics 2014, 33, 2012.
(24) (a) Pfeiffer, P.; Tappermann, F. Z. Anorg. Allg. Chem. 1933, 215,
273. (b) Jaeger, F. M.; Dijk, J. A. Z. Anorg. Allg. Chem. 1936, 227, 273.
(c) Fontaine, F. G. Acta Crystallogr., Sect. E: Struct. Rep. Online 2001,
57, m270. (d) Ikotun, O. F.; Ouellette, W.; Lloret, F.; Julve, M.; Doyle,
R. P. Eur. J. Inorg. Chem. 2007, 2083. (e) Awad, D. J.; Conrad, F.;
(4) (a) Aziridines and Epoxides in Organic Synthesis; Yudin, A., Ed.;
Wiley-VCH: Weinheim, Germany, 2006. (b) Tanner, D. Angew.
Chem., Int. Ed. Engl. 1994, 33, 599. (c) Hu, X. E. Tetrahedron 2004, 60,
́
2701. (d) Lu, P. Tetrahedron 2010, 66, 2549. (e) Stankovic, S.;
11151
dx.doi.org/10.1021/ja505823s | J. Am. Chem. Soc. 2014, 136, 11145−11152

Journal of the American Chemical Society
Article
Koch, A.; Schilde, U.; Poppl, A.; Strauch, P. Inorg. Chim. Acta 2010,
̈
363, 1488.
(25) See Supporting Information for details on the precatalyst
syntheses.
(26) It should be noted that the DCE is able to act as an oxidant by
reaction with the nickel(0) catalyst. However, the rate with which this
reaction proceeds is lower than the rate of the aziridine activation.
When full conversion of the aziridine is obtained, the reaction will
eventually discolor due to this reaction. For a reference on this type of
reaction, see: (a) Takahashi, S.; Suzuki, Y.; Hagihara, N. Chem. Lett.
1974, 3, 1363. For a related reaction with DCM, see: (b) Csok, Z.;
Vechorkin, O.; Harkins, S. B.; Scopelliti, R.; Hu, X. L. J. Am. Chem. Soc.
2008, 130, 8156.
(27) It should be noted that these reaction conditions were not
applicable to 2-aryl-N-tosylaziridines.
(28) Arylzinc reagents did not provide satisfactory results under these
conditions, providing the coupling products in low yields.
(29) For the synthesis of alkylzinc reagents in DMA, see Supporting
Information and Huo, S. Org. Lett. 2003, 5, 423.
(30) (a) Rieke, R. D. Science 1989, 246, 1260. (b) Zhu, L.;
Wehmeyer, R. M.; Rieke, R. D. J. Org. Chem. 1991, 56, 1445.
(31) For representative examples, see: (a) Fleckenstein, J. E.;
Koszinowski, K. Organometallics 2011, 30, 5018. (b) Achonduh, G. T.;
Hadie, N.; Valente, C.; Avola, S.; O’Brien, C. J.; Organ, M. G. Chem.
Commun. 2010, 46, 4109. (c) Hunter, H. N.; Hadei, N.; Blagojevic, V.;
Patschinski, P.; Achonduh, G. T.; Avola, S.; Bohme, D. K.; Organ, M.
G. Chem.Eur. J. 2011, 17, 7845.
(32) Although the stoichiometric experiments in Scheme 4a indicate
that transmetalation with azanickelacyclobutane 43 does proceed, the
outcome (product ratios) of those experiments is significantly different
from the results obtained in the coupling reaction. Since the relative
rates of azametallacycle formation (ring closure) from 46 and
transmetalation is unknown, the possibility of transmetalation
occurring from the open structure 46 in the catalytic reaction cannot
be excluded.
(33) For a study on the synthesis of this type of structures, see:
Pattenden, L. C.; Adams, H.; Smith, S. A.; Harrity, J. P. A. Tetrahedron
2008, 64, 2951.
(34) For a related example, in which the oxidative insertion of nickel
into the C−O of an epoxide happens with inversion of configuration,
see: Beaver, M. G.; Jamison, T. F. Org. Lett. 2011, 13, 4140.
(35) For a study on the preparation and reactions of a similar
complex (bpy)Ni(Et)2, see: Saito, T.; Uchida, Y.; Misono, A.;
Yamamoto, A.; Morifuji, K.; Ikeda, S. J. Am. Chem. Soc. 1966, 88, 5198.
(36) This mechanism is consistent with the observation of 1,2-
disubstituted aziridines being unsuccessful in the coupling reaction.
(37) Dauban, P.; Dodd, R. H. Org. Lett. 2000, 2, 2327 and references
cited therein.
(38) De Bont, D. B. A.; Sliedregt-Bol, K. M.; Hofmeyer, L. J. F.;
Liskamp, R. M. J. Bioorg. Med. Chem. 1999, 7, 1043.
(39) Shohji, N.; Kawaji, T.; Okamoto, S. Org. Lett. 2011, 13, 2626.
11152
dx.doi.org/10.1021/ja505823s | J. Am. Chem. Soc. 2014, 136, 11145−11152