Computationally Guided Catalyst Design in the Type i Dynamic Kinetic Asymmetric Pauson-Khand Reaction of Allenyl Acetates

Article abstract of DOI:10.1021/jacs.7b07121

The Rh(I)-catalyzed allenic Pauson-Khand reaction (APKR) is an efficient, redox-neutral method of synthesizing α-acyloxy cyclopentenones. An enantioselective APKR could provide access to chiral, nonracemic α-acyloxy and α-hydroxy cyclopentenones and their corresponding redox derivatives, such as thapsigargin, a cytotoxic natural product with potent antitumor activity. Rapid scrambling of axial chirality of allenyl acetates in the presence of Rh(I) catalysts enables the conversion of racemic allene to enantiopure cyclopentenone product in a dynamic kinetic asymmetric transformation (DyKAT). A combined experimental and computational approach was taken to develop an effective catalytic system to achieve the asymmetric transformation. The optimization of the denticity, and steric and electronic properties of the ancillary ligand (initially (S)-MonoPhos, 58:42 er), afforded a hemilabile bidentate (S)-MonoPhos-alkene-Rh(I) catalyst that provided α-acyloxy cyclopentenone product in up to 14:86 er. Enantioselectivity of the Rh(I)-(S)-MonoPhos-alkene catalyst was rationalized using ligand-substrate steric interactions and distortion energies in the computed transition states. This asymmetric APKR of allenyl acetates is a rare example of a Type I DyKAT reaction of an allene, the first example of DyKAT in a cyclocarbonylation reaction, and the first catalyst-controlled enantioselective APKR.

Full text of DOI:10.1021/jacs.7b07121

Article
pubs.acs.org/JACS
Cite This: J. Am. Chem. Soc. XXXX, XXX, XXX-XXX
Computationally Guided Catalyst Design in the Type I Dynamic
Kinetic Asymmetric Pauson−Khand Reaction of Allenyl Acetates
Lauren C. Burrows, Luke T. Jesikiewicz, Gang Lu, Steven J. Geib, Peng Liu,*
and Kay M. Brummond*
Department of Chemistry, University of Pittsburgh, Pittsburgh, Pennsylvania 15260, United States
S
* Supporting Information
ABSTRACT: The Rh(I)-catalyzed allenic Pauson−Khand
reaction (APKR) is an efficient, redox-neutral method of
synthesizing α-acyloxy cyclopentenones. An enantioselective
APKR could provide access to chiral, nonracemic α-acyloxy and
α-hydroxy cyclopentenones and their corresponding redox
derivatives, such as thapsigargin, a cytotoxic natural product
with potent antitumor activity. Rapid scrambling of axial
chirality of allenyl acetates in the presence of Rh(I) catalysts
enables the conversion of racemic allene to enantiopure
cyclopentenone product in a dynamic kinetic asymmetric
transformation (DyKAT). A combined experimental and
computational approach was taken to develop an effective
catalytic system to achieve the asymmetric transformation. The
optimization of the denticity, and steric and electronic properties of the ancillary ligand (initially (S)-MonoPhos, 58:42 er),
afforded a hemilabile bidentate (S)-MonoPhos-alkene-Rh(I) catalyst that provided α-acyloxy cyclopentenone product in up to
14:86 er. Enantioselectivity of the Rh(I)-(S)-MonoPhos-alkene catalyst was rationalized using ligand-substrate steric interactions
and distortion energies in the computed transition states. This asymmetric APKR of allenyl acetates is a rare example of a Type I
DyKAT reaction of an allene, the first example of DyKAT in a cyclocarbonylation reaction, and the first catalyst-controlled
enantioselective APKR.
acetates, we observed the formation of 2 in 76% yield.7
Surprisingly, the diastereoselectivity for the formation of 2 was
INTRODUCTION
■
An ultimate challenge in synthesis is the development of
2:1, even though the allene-yne 1 was a 1:1 mixture of
methods to prepare enantioenriched compounds. α-Acyloxy1
diastereomers. Subjecting a 5:1 dr of product 2 to the APKR
and α-hydroxy2 cyclopentenones, and their corresponding
reaction conditions showed the same dr after 24 h. Subjecting a
redox derivatives,3 appear in numerous natural products, but
3:1 mixture of allenyl acetates to the APKR reaction conditions
enantioselective access to these functional group arrays is
for 40 min at 90 °C, or at rt for 7 h, resulted in a 1:1 ratio of
limited. α-Oxygenated ketones are typically prepared via the
allene diastereomers with no evidence of APKR product 2.
Finally, no scrambling of axial chirality was observed in the
installation of a hydroxyl group adjacent to an existing
carbonyl.4 However, there are only a few methods for the
absence of the Rh(I) catalyst. These observations suggested the
asymmetric preparation of these groups using this strategy, and
potential for a Type I dynamic kinetic asymmetric trans-
stereoselectivity in these cases is substrate- or reagent-
controlled.5 An enantioselective method where either enan-
tiomer of an α-acyloxy ketone could be accessed with catalyst-
formation (DyKAT) where racemic allenyl acetate 4 could be
converted to a single enantiomer of the product 5 via the rapid
scrambling of the axial chirality of the allene (k2, k−2) and the
controlled selectivity would be a valuable addition to the
toolbox of asymmetric synthesis.6
selective APKR of one allene enantiomer (k1 > k3, or k3 > k1)
when using a chiral Rh(I) catalyst (Figure 1C).13
The allenic Pauson−Khand reaction (APKR) of allenyl
acetates, catalyzed by Rh biscarbonyl chloride dimer ([Rh-
Impressive advancements have been made in the enantiose-
lective Rh-catalyzed Pauson−Khand reactions (PKR) of
(CO)2Cl]2), is an efficient method of synthesizing α-acyloxy
enynes, including development of room temperature con-
cyclopentenones, and its utility for the preparation of highly
ditions,14 using aldehydes as a carbon monoxide replacement,15
and catalyst immobilization.16 However, the scope of the
enantioselective Rh-catalyzed PKR has been limited to only 1,6-
oxygenated ring systems has been demonstrated (Figure 1A).7
In addition, the APKR offers a redox economical approach to
bicyclo[5.3.0]decane ring structures prevalent in 6,12-guaiano-
lides8 such as thapsigargin (Tg, 3a, Figure 1B)9a−d,10,11 and
Mipsagargin (3b),9e and other polycyclic natural products.12
During studies to explore the scope of the APKR of allenyl
Received: July 8, 2017
© XXXX American Chemical Society
A
DOI: 10.1021/jacs.7b07121
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 2. A synergistic strategy for catalyst design.
reactions. We hypothesized that the efficacy and reliability of
the computational predictions could be improved by using the
three-step approach described in the present study. Using this
strategy, a DyKAT-based asymmetric APKR was efficiently
developed using a practical balance of computational and
experimental resources.
RESULTS AND DISCUSSION
■
Preparation of Racemic Allenyl Acetates 9a−f. Methyl
ketones 6a and 6b were prepared in three steps from 5-hexen-
1-ol (see Supporting Information, SI). Trimethylsilyl (TMS)-
alkyne 6c was synthesized in four steps from 4-pentyn-1-ol, as
previously reported.7 Propargyl acetates 7a−c were synthesized
in one step from methyl ketones 6a−c by addition of
ethynylmagnesium bromide, followed by acetyl chloride
(Conditions A, Scheme 1A). Propargyl pivalate 7d, octanoate
7e, benzoate 7f, and p-nitrobenzoate 7g were prepared from
TMS-methyl ketone 6c in two steps. Addition of ethynylmag-
nesium bromide followed by aqueous workup (Conditions B)
provided propargyl alcohol 8c in 88% yield. Propargyl pivalate
7d was synthesized by reacting propargyl alcohol 8c and pivalic
anhydride with catalytic scandium(III) trifluoromethanesulfo-
nate in 86% yield (Conditions C). Propargyl octanoate 7e was
prepared in 45% yield by reacting alcohol 8c with octanoic acid,
N,N′-dicyclohexylcarbodiimide (DCC) and 4-dimethyl amino-
pyridine (DMAP, Conditions D). Propargyl benzoate 7f was
prepared in 55% yield from propargyl alcohol 8c using benzoic
anhydride, triethylamine, and DMAP (Conditions E). P-
nitrobenzoyl chloride, and DMAP were reacted with propargyl
alcohol 8c to afford 7g in 84% yield (Conditions F). Propargyl
acetates 7a−g were reacted with Rh(II) bistrifluoroacetate
dimer ([Rh(OCOCF3)2]2) to afford the allenes 9a−g in high
yields (77−94%).
Identification of Catalyst and Substrate for Asym-
metric APKR Studies. To identify a preliminary reaction
model system for DyKAT optimization, three allene-ynes 9a−c,
with different substituents at the alkyne terminus were reacted
under neutral and cationic Rh(I) conditions (Conditions A and
B, respectively, Scheme 1B). The reaction of phenyl-substituted
alkyne 9a gave 10a in 30% and 61% yield, the triisopropylsilyl
(TIPS)-alkyne 9b afforded 10b in 50 and 16% yield, and the
trimethylsilyl (TMS)-substituted alkyne 9c afforded 10c in 47
and 70% yield, respectively. Because allene-yne 9c gave the
highest yield under the cationic Rh(I) conditions in the
presence of a phosphine ligand at lower temperature (50 versus
110 °C), we focused on the APKR of 9c using cationic Rh
catalysts in the development of the asymmetric catalytic system.
Identification of an Effective Ligand Class for
Asymmetric APKR. Several types of chiral ligands have been
Figure 1. APKR of allenyl acetates to access chiral nonracemic α-
acyloxy cyclopentenones.
enyne substrates.17 If successful, this work would represent a
rare example of a DyKAT reaction of an allene,18 the first
example of DyKAT in a cyclocarbonylation reaction, and the
first catalyst-controlled enantioselective APKR. Thus, we set
out to test the feasibility of the asymmetric APKR to prepare
enantioenriched α-acyloxy cyclopentenones.
A combination of theory and experiment was used to identify
a chiral Rh(I) catalyst to effect this asymmetric transformation
from racemic allenes. We utilized a rational catalyst design
process that relied on computational insights with experimental
validation, rather than experimentally screening a large number
of chiral ligands.19 This process comprised three steps: (1)
experimentally identifying a suitable ligand class affording the
APKR product in high yield with some enantioselectivity, (2)
elucidating the APKR mechanism involving that ligand class
using density functional theory (DFT) calculations to establish
the enantioselectivity-determining step, and (3) using an
iterative feedback loop involving rational selection of a ligand
based on DFT calculations and experimental observation, and
testing the performance of the ligand both experimentally and
computationally (Figure 2). Although computation is often
used to study reaction mechanisms and explain experimental
results, development of the tools required to design catalysts in
silico is still in its infancy.20 Computational predictions
involving transition metal catalysts for asymmetric trans-
formations are especially challenging because a number of
factors need to be considered, such as reactivity, regio- and
stereoselectivity, catalyst stability, and side- and background-
B
DOI: 10.1021/jacs.7b07121
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Scheme 1. Synthesis of Allenyl Acetates and Racemic APKR
Table 1. Experimental Screening of Ligand Classes in the
APKR of 9c
a
Enantiomeric ratios were determined by HPLC using a chiral
b
stationary phase. [Rh(CO)2Cl]2 (10 mol %), L1 (22 mol %), AgBF4
(20 mol %), DCE, CO (1 atm). [Rh(CO)2Cl]2 (10 mol %), L2 (22
mol %), AgBF4 (20 mol %), DCE, CO (1 atm). [Rh(IMes
c
d
L3)(cod)(Cl)] (10 mol %), AgBF4 (10 mol %) DCE, CO (1 atm).
e
f
Recovered starting material. Rh(cod)2BF4 (10 mol %), L4 (15 mol
%), DCE, CO (1 atm).
Rh(I)-(S)-MonoPhos (L4) catalyst. With this information, we
could evaluate how the structural modification of (S)-
MonoPhos (L4) would improve enantioselectivity while
maintaining good catalytic activity. Several previous computa-
tional studies of Co-catalyzed PKRs have been reported,25 as
well as a DFT study of a diastereoselective Rh-catalyzed PKR of
an enyne.26 In collaboration with Jordan, one of our groups has
reported a rationalization for the metal-dependent regioselec-
tivity of Rh(I)- and Mo(0)-catalyzed APKRs using DFT
calculations.27 However, all of the previous computational
studies on the Rh-catalyzed PKR reactions were performed
using Rh(CO)nCl or Rh(CO)n+ as the active catalyst (n = 1 or
2). The mechanism of PKR using other Rh(I) catalysts and the
effects of ligands on reactivity and enantioselectivity have not
been investigated computationally.
The accepted mechanism of the Rh-catalyzed APKR is
illustrated in Scheme 2.27 Coordination of allene-yne II to the
catalyst I affords complex III, which undergoes oxidative
cyclization to afford Rh(III) intermediate IV. Subsequent CO
insertion, followed by reductive elimination and product
dissociation provides cyclopentenone product VII.
The computed reaction energy profiles of the cationic Rh-
(S)-MonoPhos (L4)-catalyzed APKR with (R)-9c are shown in
Scheme 3. We considered two different oxidative cyclization
pathways with active Rh catalysts bearing either one or two CO
ligands. In each pathway, only the energies of the most stable
isomers are shown. The less stable isomers of the oxidative
cyclization transition states and the different reactivities of (R)-
and (S)-9c are discussed in detail later. The calculations were
performed using B3LYP/6-31G(d)−LANL2DZ(Rh) for geom-
etry optimizations and M06/6-311+G(d,p)−SDD(Rh) /SMD
(DCE) for single point energy calculations (see SI for
computational details). In both pathways, the resting state of
the catalyst is the square-planar complex 12. Coordination of
successfully used in the asymmetric PKR of enynes.21 We
evaluated the effectiveness of select ligands of these types in the
reaction of allene-yne 9c (Table 1). Initial studies with chiral
bidentate phosphines gave low yields and no enantioselectivity.
For example, reaction with (R)-BINAP (L1) and AgBF4
afforded a 13% yield of 10c with a 50:50 enantiomeric ratio
(er) and (R)-MeO-BIPHEP (L2) and AgBF4 afforded a 21%
yield of 10c with a 51:49 er (Table 1, entries 1 and 2).22 Both
bisphosphine ligands L1 and L2 afforded aldehyde 11c as the
major product.23 Catalyst Rh(IMes L3)(cod)Cl (cod = 1,5-
cycloocta-diene) with the strongly electron donating N-
heterocyclic carbene (NHC) ligand was unreactive in the
APKR and afforded only unreacted starting material (Table 1,
entry 3).24 The poor yields observed in these initial
experiments prompted us to test less sterically demanding
and less σ-donating ligands. To this end, monodentate
phosphoramidite ligand (S)-MonoPhos (L4) provided 10c in
76% yield with a 58:42 er (Table 1, entry 4). Identification of
monodentate phosphoramidites as an effective ligand class for
the asymmetric APKR inspired a thorough examination of the
APKR mechanism in order to discover ligand structural
modifications to improve enantioselectivity.
Calculation of the APKR Mechanism and the
Enantioselectivity-Determining Transition State. After
the experimental identification of monodentate phosphorami-
dites as an effective ligand class, DFT calculations were
performed to determine the mechanism and the enantio-
determining transition state geometry of the APKR using the
C
DOI: 10.1021/jacs.7b07121
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
and C−C reductive elimination to form the cyclopentenone
product are both lower in energy than 16-TS (see SI for the
computed reaction energy profile of the complete catalytic
cycle). Thus, the oxidative cyclization is irreversible and
enantioselectivity-determining. These mechanistic insights
were leveraged computationally to understand the origin of
the enantioselectivity obtained using the monodentate
phosphoramidite ligand (S)-MonoPhos (L4).
Scheme 2. Accepted Mechanism of the Rh(I)-Catalyzed
APKR
To investigate the effects of (S)-MonoPhos (L4) on
enantioselectivity and to gain insights for the rational prediction
of a more selective ligand, we performed a thorough analysis of
the possible transition state isomers leading to both (R) and
(S)-cyclopentenone products. A systematic conformational
search of the oxidative cyclization transition states revealed
six transition state isomers in the reaction with (R)-9c to form
(R)-10c: the (S)-MonoPhos (L4) ligand may occupy any of the
three available coordination sites on the Rh and the ligand itself
may adopt two different conformations via rotation about the
P−Rh bond (Figure 3, 16-TS, 18-TS, 19-TS, 16-TSrot, 18-
TSrot, and 19-TSrot). Similarly, when the Rh catalyst reacts
with the other enantiomer of the allene-yne to form (S)-10c,
another six transition state isomers were located (20-TS, 21-
TS, 22-TS, 20-TSrot, 21-TSrot, and 22-TSrot). The lowest-
energy transition state structures leading to the (R)- and (S)-
products (16-TS and 21-TSrot, respectively) only differ in
energy by 1.0 kcal/mol. This small energy difference among
transition state isomers agrees with the poor enantioselectivity
that is observed experimentally in the APKR with this ligand.
These theoretical insights prompted us to explore different
strategies to reduce the conformational flexibility of the
oxidative cyclization transition states and optimize the
catalyst-substrate steric interactions to differentiate the (R)-
and (S)-selective transition states.
Scheme 3. Computed Reaction Energy Profile of the
Cationic Rh-(S)-MonoPhos (L4)-Catalyzed APKR
a
Improving Enantioselectivity by Modifying the
Phosphoramidite Scaffold. Our first strategy toward
improving the enantioselectivity of the APKR was to
experimentally test a ligand that is expected to have more
significant steric interactions with the substrate in the calculated
oxidative transition states shown in Figure 3. Spirocyclic ligand
(S)-SIPHOS (L5), which has previously been applied in the
asymmetric PKR of tethered enynes, has a wider “cone angle”
than (S)-MonoPhos (L4).28 This ligand provided a good yield
and, indeed, improved enantioselectivity (29:71 er, Table 2,
entry 2). This supports the hypothesis that enhanced ligand-
substrate steric interactions are beneficial for enantiocontrol.
However, efforts to further improve the selectivity using (S)-
SIPHOS (L5) were unsuccessful. For example, in order to
exaggerate the steric interactions between (S)-SIPHOS (L5)
and the acetate group of the substrate, allenyl pivalate 9d was
submitted to the APKR conditions (Table 2, entry 3). The
reaction was slowed and the enantioselectivity decreased to
39:61 er. These results imply that simply improving the steric
bulk of the monodentate phosphoramidite can only modestly
improve the enantioselectivity, as the (S)-SIPHOS (L5) ligand
may still occupy any of the three available binding sites on the
Rh center and the facile rearrangement of the ligand minimizes
the steric interactions between the ligand and the substrate. We
hypothesize that the decrease in enantioselectivity using (S)-
SIPHOS (L5) and pivalate 9d is due to the decreased binding
energy of the (S)-SIPHOS (L5) ligand that potentially leads to
the replacement of (S)-SIPHOS (L5) with CO in the oxidative
cyclization transition state. Taken together, the experimental
and computational results for (S)-MonoPhos (L4) and (S)-
a
Tetrafluoroborate counteranion is omitted in the calculations.
allene-yne 9c to replace two CO ligands in 12 requires 5.5 kcal/
mol to afford substrate complex 13, in which both the alkyne
and allene π bonds are coordinated to the Rh center.
Subsequent oxidative cyclization of complex 13 via 14-TS
requires an activation free energy of 21.9 kcal/mol with respect
to the resting state 12. In the alternative pathway shown in
black in Scheme 3, allene-yne 9c replaces one of the CO ligands
in 12 to form a square-based pyramidal complex 15, which is
7.7 kcal/mol less stable than the four-coordinated allene-yne-
Rh complex 13. Nonetheless, the 18-electron complex 15 more
readily undergoes oxidative cyclization than the 16-electron
complex 13. The five-coordinated square-based pyramidal
oxidative cyclization transition state 16-TS from 15 is
significantly more stable than the four-coordinated 14-TS.
The additional CO ligand stabilizes 16-TS and decreases the
overall activation energy to 17.6 kcal/mol with respect to 12.
The formation of rhodacycle intermediate 17 is highly
exothermic. The transition states of subsequent CO insertion
D
DOI: 10.1021/jacs.7b07121
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 3. APKR transition state isomers using (S)-MonoPhos (L4). Prot designates a different ligand conformation by rotation about the Rh−P
bond. Tetrafluoroborate counteranion is omitted in the calculations.
Table 2. Computationally Guided Ligand Screening in the Asymmetric APKR
a
c
c
c
entry
ligand
R
CO (atm)
T (°C)
time (h)
yield (%)
er (S:R)b
ΔG‡(S)
ΔG‡(R)
ΔΔG‡
d
d
1
L4
L5
L5
Ac 9c
Ac 9c
Piv 9d
Ac 9c
Ac 9c
Ac 9c
Ac 9c
Piv 9d
1.0
1.0
1.0
1.0
1.0
0.1
0.1
0.1
0.1
0.1
0.1
50
50
50
50
70
70
70
70
70
70
70
5
10c 76
10c 67
10d 60
10c 18
10c 76
10c 71
58:42
29:71
39:61
56:44
23:77
18:82
18:82
14:86
20:80
15:85
33:67
18.6
17.6
−1.0
−
−
+4.4
−4.4
−
−
−
−
−
2
10
12
15
15
15
20
15
17
20
15
−
−
26.3
−
−
30.7
3
e
f
f
4
L6
f
f
5
L7
L7
23.2
18.8
6
−
−
−
−
−
−
−
−
−
−
−
−
g
h
7
L7
10c 79
8
L7
L7
L7
L8
10d 50
10e 70
10f 77
10c 90
9
CO(CH2)6CH3 9e
Bz 9f
10
11
Ac 9c
−
a
1
Reactions were performed on a 0.05 mmol (13−17 mg) scale. Yields were determined by H NMR integration versus mesitylene standard.
Enantiomeric ratios were determined by HPLC using a chiral stationary phase. Absolute configurations are assigned by analogy to crystal structure
of benzoate 10f (Figure 6). All energies are in kcal/mol. See Figure 3 for activation energies of all computed transition state isomers. Rh(cod)2BF4
b
c
d
e
f
g
(5 mol %), L6 (6 mol %). See SI for activation energies of all computed transition state isomers. Performed on 0.19 mmol (50 mg) scale with 5.0
h
equiv mesitylene. Isolated yield is reported.
SIPHOS (L5) indicate that although monodentate phosphor-
amidite ligands provide desired reactivity in the APKR,
modification of the steric environment afforded only marginal
increases in the enantioselectivity. A strategy to minimize the
conformational flexibility of the phosphoramidite ligand is
needed.
Our second strategy toward improving selectivity was
incorporation of chelating ligands to prevent catalyst rearrange-
ment and enhance stereocontrol. In this regard, hybrid bidentate
phosphine-phosphoramidite ligands have proven advantageous
in Rh-catalyzed asymmetric hydroformylation reactions.29
Therefore, allene-yne 9c was experimentally tested in the
APKR using the bidentate (S)-MeAnilaPhos ligand (L6).30
This reaction afforded cyclopentenone product 10c in only
18% yield and 56:44 er (Table 2, entry 4). Byproduct α,β-
unsaturated aldehyde 11c (Table 1) was obtained in 31% yield.
Despite the low yield of 10c with this ligand, important
information was extracted regarding ligand effects on catalyst
reactivity by computationally exploring the reaction energy
profile using DFT calculations (Scheme 4, see SI for all
transition state isomers in reactions with both enantiomers of
9c). Similar to the reaction with monodentate phosphoramidite
ligands, the resting state of the catalyst bearing the bidentate
(S)-MeAnilaPhos ligand (L6) is the four-coordinated square-
planar complex 23. The oxidative cyclization pathways
involving both four- and five-coordinated Rh complexes were
E
DOI: 10.1021/jacs.7b07121
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
ilaPhos (L6) are not productive due to the unfavorable
substrate binding.
Scheme 4. Computed Reaction Energy Profile of the
Cationic Rh-(S)-MeAnilaPhos (L6)-Catalyzed APKR
a
Balancing Reactivity and Enantioselectivity Using
Hemilabile Bidentate Ligand (S)-MonoPhos-alkene
(L7). In order to increase both the enantioselectivity and
reactivity of the APKR, we applied a hemilabile bidentate ligand
(S)-MonoPhos-alkene (L7).31 (S)-MonoPhos-alkene (L7) was
synthesized (see SI) and applied in the APKR to afford product
10c in 76% yield and 23:77 er (Table 2, entry 5). This high
yield combined with improved enantioselectivity inspired
further reaction optimization. The reaction proved sensitive
to solvent, CO pressure, temperature, additives and scale. For
example, when the CO concentration was lowered to 0.1 atm
(balloon of 10% gas mixture of CO/Ar), the selectivity
improved to 18:82 er (Table 2, entry 6). Incorporation of
mesitylene as an additive was necessary for good yields in the
APKR. For example, the yield of the APKR of allenyl acetate 9c
improved from 49 to 76% when 1.0 equiv mesitylene was
added (see SI). Further details regarding optimization studies
can be found in the SI.
The yield of the asymmetric APKR was sensitive to scale. For
example, when the reaction was performed using 13 mg of
precursor 9c in an 8 mL test tube, product 10c was obtained in
71% yield (Table 2, entry 6). However, when 50 mg of the
same precursor was used under identical conditions, 10c was
obtained in 41% yield (see SI). By changing a variety of
reaction conditions (CO pressure, reaction vessel, and
mesitylene equivalents), we discovered that increasing the
equivalents of mesitylene from 1.0 to 5.0 gave a yield for the
larger scale reaction that was consistent with that of the smaller
scale reaction (79%, 18:82 er, entry 7). We attribute this result
to the ability of the mesitylene additive to stabilize the cationic
Rh catalyst.32
Several substrates differing by their acyloxy substituents were
submitted to the optimized reaction conditions. Incorporation
of a sterically bulky pivaloyloxy group in allene 9d resulted in
an increased er (14:86), but afforded a lower yielding APKR
(50%, compare entries 6 and 8). Allenyl octanoate 9e reacted in
similar yield (70%) and enantioselectivity (20:80 er) as 9c
(compare entries 6 and 9). Reaction of allenyl benzoate 9f
afforded the APKR product 10f in 77% yield and 15:85 er
(entry 10). Substrate 9g, functionalized with an electron-
withdrawing p-nitrobenzoyl group, afforded the APKR product
10g in 45% yield and 23:77 er (see SI).
Experimental Evidence for Type I DyKAT Mechanism.
Experiments were conducted to test whether the observed
enantioselectivity was arising from a dynamic kinetic resolution
(DKR), Type I DyKAT, or Type II DyKAT mechanism.13 Ph-
substituted alkyne 9a was chosen as a substrate for these
experiments because its enantiomers were readily resolved by
HPLC using a chiral stationary phase. Single enantiomers of
allenyl acetate 9a were isolated and subjected to the APKR
reaction conditions (Scheme 5, reaction 1). After 30 min, the
allenyl acetates had completely racemized, and no evidence of
APKR product 10a was observed. Therefore, allenyl acetate
racemization is faster than the APKR. Regardless of which
allene enantiomer was reacted, the same product (R)-10a was
obtained after 20 h. Thus, enantioselectivity is catalyst-
controlled, and is established during the APKR. The low yields
of these reactions are attributed to the small scale of the
experiments (3.5 mg each). In an additional experiment, a
single enantiomer of (−)-9a was subjected to the reaction
conditions without the Rh catalyst. The allene was recovered
a
Tetrafluoroborate counteranion is omitted in the calculations.
considered computationally. Coordination of allene-yne 9c to
replace two CO ligands in 23 requires 20.3 kcal/mol to afford
the four-coordinated substrate complex 24. Oxidative cycliza-
tion of 24 via 25-TS requires a very high barrier of 36.0 kcal/
mol with respect to 23. Alternatively, exchanging one of the
CO ligands in 23 with allene-yne 9c yields a five-coordinated
square-based pyramidal complex 26 with one CO ligand in the
apical position. The unfavorable substrate binding energy of
22.4 kcal/mol indicates that the sterically demanding bidentate
ligand prohibits the allene-yne from binding to Rh in a
bidentate fashion to effect oxidative cyclization. Similar to the
reaction with the monodentate (S)-MonoPhos (L4) ligand, the
oxidative cyclization of the 18-electron complex 26 (via 27-TS)
is more facile than the oxidative cyclization of the 16-electron
complex 24. Nonetheless, the most favorable oxidative
cyclization pathway with the (S)-MeAnilaPhos (L6) ligand
still requires a high activation free energy of 26.3 kcal/mol with
respect to the catalyst resting state 23. This is significantly less
favorable than the oxidative cyclization of the same substrate
with the (S)-MonoPhos (L4)-based Rh catalyst (ΔG‡ = 17.6
kcal/mol, see Scheme 3). These results suggest that undesired
side reactions may compete with the oxidative cyclization
process with the Rh-(S)-MeAnilaPhos (L6) catalyst and thus
lead to the low yield of the APKR products. Isolation of
aldehyde 11c as a major product of this reaction suggests that
only the allene moiety is coordinating to the Rh catalyst. Taken
together, the above experimental and computational inves-
tigations clearly revealed the important role of phosphoramidite
ligands on the reactivity and enantioselectivitycationic Rh(I)
catalysts bearing monodentate ligands (S)-MonoPhos (L4) and
(S)-SIPHOS (L5) are reactive but poorly enantioselective due
to the conformational flexibility of the oxidative cyclization
transition states with these ligands. In contrast, reactions with
the bidentate phosphine-phosphoramidite ligand (S)-MeAn-
F
DOI: 10.1021/jacs.7b07121
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
unbound to the Rh are possible. To elucidate the most
preferred pathway in the enantio-determining oxidative
cyclization step, both alkene bound and unbound pathways
were investigated using DFT calculations (Scheme 6). In both
pathways, the resting state of the catalyst is the square planar
complex 30 that exists in equilibrium with alkene-bound
complex 29, which is 8.7 kcal/mol higher in energy. The
hemilabile nature of the (S)-MonoPhos-alkene (L7) ligand in
29 and 30 indicates a Curtin-Hammett system in which the
exact mechanism of oxidative cyclization is determined by the
energy difference of the oxidative cyclization transition states.
In the alkene unbound pathway (Scheme 6, left), coordination
of the allene-yne 9c affords five-coordinated substrate complex
31, in which the Rh is coordinated with two CO ligands and the
hemilabile (S)-MonoPhos-alkene (L7) in a monodentate
fashion. Formation of the diastereomeric complexes (R)- and
(S)-31, in which the Rh binds to (R)- and (S)-9c, respectively,
require substrate binding energies of 12.0 and 18.4 kcal/mol
with respect to 30. In this alkene unbound APKR pathway, the
oxidative cyclization of (R)-9c via (R)-32-TS requires a 4.4
kcal/mol lower activation free energy than the reaction with
(S)-9c via (S)-32-TS. This energy difference predicts the
enantioselective formation of the APKR product (R)-10c.
In the alternative alkene bound pathway (Scheme 6, right),
the oxidative cyclization occurs via five-coordinated square-
based pyramidal transition states (R)- and (S)-35-TS bearing
only one CO ligand and the hemilabile (S)-MonoPhos-alkene
(L7) ligand bound in a bidentate fashion. Interestingly, the
substrate complexes (R)- and (S)-34 in this one-CO pathway
remain four-coordinated with the alkene unbound from the Rh
center. This result is consistent with the reaction profiles of
mono- and bidentate ligands L4 and L6 that the 16-electron
four-coordinated Rh(I) reactant complexes are more stable
Scheme 5. Experiments Supporting Type I DyKAT
Mechanism
ab
,
a
Reactions were performed using 3.5 mg each of allenyl acetate
b
starting material. Enantiomeric ratios were determined by HPLC
c
using a chiral stationary phase. Complete racemization (50:50 er)
d
observed after 30 min. Yields were determined by 1H NMR
integration versus mesitylene standard.
(96%) and no scrambling of axial chirality was observed,
supporting the hypothesis that allene isomerization is catalyzed
by Rh (Scheme 5, reaction 2). Taken together, the results of
these mechanistic studies support catalyst-controlled enantio-
selectivity enabled by rapid, Rh-catalyzed allenyl acetate
isomerization. Therefore, the enantioselective APKR of allenyl
acetates is classified as Type I DyKAT.
Computational Investigation of the Mechanism of the
APKR with (S)-MonoPhos-alkene (L7). Although the Rh is
expected to adopt a five-coordinated square-based pyramidal
geometry in the oxidative cyclization, two different pathways
involving the alkene group of the ligand L7 either bound or
a
Scheme 6. Computed Reaction Energy Profile of the Cationic Rh-(S)-MonoPhos-alkene (L7)-Catalyzed APKR
a
Tetrafluoroborate counteranion is omitted in the calculations.
G
DOI: 10.1021/jacs.7b07121
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
than the 18-electron complexes (see Schemes 3 and 4). The
formation of the four-coordinated reactant complexes (R)- and
(S)-34 are also facile with low substrate binding energies.
However, these 16-electron species are less reactive than the
18-electron complexes (R)- and (S)-31 in the oxidative
cyclization. The alkene bound, one-CO pathway has a higher
overall activation energy (ΔG‡ = 23.6 kcal/mol, (S)-35-TS)
than the alkene unbound pathway (ΔG‡ = 18.8 kcal/mol, (R)-
32-TS). In summary, DFT calculations predict that the APKR
with (S)-MonoPhos-alkene (L7) occurs via the alkene unbound
pathway and affords (R)-10c.
Experimental Validation of Alkene Unbound Pathway
and Predicted Absolute Configuration of the Major
Product. To experimentally test whether the reaction proceeds
via the alkene unbound pathway, saturated ligand L8 was applied
in the APKR. The reaction of 9c using (S)-MonoPhos-sat’d
(L8) ligand provided 10c in 90% yield and 33:67 er, with the
same major enantiomeric product as that obtained with L7
(Table 2, compare entries 11 and 7). This result suggests that
the alkene in L7 is unbound in the oxidative cyclization
transition state, as predicted by DFT calculations (Scheme 6).
The higher enantioselectivity with phosphoramidite-alkene
ligand L7 suggests that the hemilability of the ligand plays a
key role in achieving a high level of enantiocontrol by
preventing rearrangement or dissociation of the ligand.
Calculations with (S)-MonoPhos-alkene (L7) predict that
the (R) absolute configuration of APKR product 10c is
preferred (Scheme 6). To test this computational prediction
experimentally, the APKR was performed on allene-yne 9f to
give benzoate 10f as an amorphous white solid (77% yield,
15:85 er, Table 2, entry 10). After two recrystallizations in
pentane, the enantiopurity of 10f was enhanced to 2:98 er. An
X-ray quality crystal was grown by slow cooling in hexanes. The
absolute configuration of benzoate 10f was unambiguously
assigned as (R) by X-ray crystallography (Figure 4). Therefore,
possible in the reaction with each enantiomer of allene-yne 9c.
We calculated the activation energies of all 12 transition state
isomers using the (S)-MonoPhos-alkene (L7) ligand to
determine the origin of enantioselectivity. The two lowest-
energy transition states in the reaction with each enantiomer
are shown in Figure 5 (see SI for other higher-energy transition
state isomers). The two lowest-energy transition states in the
reaction with (R)-9c that lead to (R)-10c (ΔG‡ = 18.8 and 19.8
kcal/mol) are both much more stable than the lowest-energy
transition states in the reaction with (S)-9c that forms (S)-10c
(ΔG‡ = 23.2 and 24.6 kcal/mol). Here, the DFT-calculations
predicted much greater enantioselectivity (ΔΔG‡ = 4.4 kcal/
mol) than that observed experimentally (18:82 er). Our
benchmark calculations suggest the choice of density functional
and basis set has minimal effects on the calculated ΔΔG‡ value
(see SI for details). Thus, we hypothesize that the lower
enantioselectivity observed experimentally may be attributed to
experimental factors such as CO concentration, and counter-
anion and additive effects, which were not considered in the
DFT calculations.
To explore the origin of the enantioselectivity, we calculated
the distortion energies of the ligand and the substrate in the
oxidative cyclization transition states.33 The distortion energy
represents the total energy required for the (S)-MonoPhos-
alkene ligand (L7) and the allene-yne substrate to adopt their
respective transition state geometries from the undistorted
ground state conformation (see SI for details). The relative
distortion energies ΔΔEdist with respect to the least-distorted
transition state (R)-32-TS are shown in Figure 5. The two
lowest-energy transition states leading to (R)-10c (Figure 5A
and B) have significantly lower total distortion energies than
the two lowest-energy transition states leading to (S)-10c
(Figure 5C and D). Other higher-energy transition state
isomers suffer even greater distortion energies (See SI for
details). These results indicate that the higher barriers in the
pathways to form the minor enantiomer (S)-10c are due to the
unfavorable steric repulsions between the substrate and ligand
L7. In (S)-32-TS, one steric interaction that is easily identified
is between the binaphthyl backbone of the ligand and the TMS
group of the substrate, which is evidenced by the short H···H
distance (2.17 Å) between the TMS and the binaphthyl group.
In comparison, in the lowest-energy transition state leading to
(R)-10c (Figure 5A), the distance between these two groups is
much longer (2.34 Å). The second lowest-energy transition
state isomer that leads to (S)-10c (Figure 5D) is destabilized by
an unfavorable steric repulsion between the forming seven-
membered ring and the binaphthyl moiety on the ligand.
Summary of Ligand Effects on Reactivity and
Enantioselectivity. The computed activation free energies
in the enantioselectivity-determining oxidative cyclization step
of the APKR of 9c with three different ligands (L4, L6, and L7)
are summarized in Figure 6. For each reaction, multiple
transition state isomers are located within a few kcal/mol,
indicating the importance of the systematic conformational
search to locate all possible transition state isomers in the
computational prediction of enantioselectivity. The DFT
calculations revealed significant effects of the phosphoramidite
ligands on both reactivity and selectivity. While most of the
transition state isomers with (S)-MonoPhos (L4) have very
similar activation free energies in the (R)- and (S)-selective
pathways, the (R)-selective transition state isomers with the
(S)-MonoPhos-alkene (L7) ligand are noticeably more stable
than the (S)-selective transition states. Both the (R)- and (S)-
Figure 4. Crystal structure of (R)-10f.
we assign the absolute configuration of acetate product 10c as
(R), based on analogous HPLC retention times of the major
and minor enantiomers. This stereochemical assignment of
(R)-10c matches that predicted by computation (Scheme 6).
Origin of Enantioselectivity of (S)-MonoPhos-alkene
Ligand (L7). In the favorable alkene unbound APKR pathway,
six different oxidative cyclization transition state isomers are
H
DOI: 10.1021/jacs.7b07121
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 5. Lowest-energy oxidative cyclization transition state isomers of APKR with (S)-MonoPhos-alkene ligand (L7). The total distortion energies
of the ligand and the substrate in the transition states are with respect to that of the lowest-energy transition state (R)-32-TS′. The allene-yne
substrate is highlighted in green. Rh is in blue and phosphorus is in orange.
CONCLUSIONS
■
The asymmetric APKR of allenyl acetates represents the first
example of a DyKAT carbonylation reaction, the first catalyst-
controlled enantioselective PKR of an allene, and a rare
example of a DyKAT of an allene. A reactive and
enantioselective catalyst was designed through a unique
combination of DFT studies and laboratory experiments.
Once phosphoramidites were identified as a reactive ligand
class with low enantioselectivity (L4, 58:42 er), mechanistic
information provided by computation enabled us to achieve
good enantioselectivity (L7, 14:86 er) after experimentation
with only four different phosphoramidite ligands, demonstrat-
ing the power of computation in streamlining the selection of
chiral ligands. The computational study addressed a series of
mechanistic questions that are critical for the proper prediction
of ligand effects on reactivity and enantioselectivity of APKR,
Figure 6. Summary of computed activation energies of the oxidative
cyclization transition state isomers in the Rh-catalyzed APKR reactions
including facile catalyst rearrangement, the number of CO
ligands bound to the Rh center, and the binding mode of a
of allene-yne (R)- and (S)-9c with different ligands. (R)-selective
hemilabile phosphoramidite ligand. After obtaining these
transition state isomers are in the reaction with (R)-9c and form (R)-
mechanistic insights, the calculated reactivity and selectivity
10c. (S)-selective transition state isomers are in the reaction with (S)-
(ΔG‡ and ΔΔG‡) were used as a guide to understand the
9c and form (S)-10c.
experimental yield and er and to make rational prediction of a
more selective and reactive ligand. The enantioselectivity of
hemilabile (S)-MonoPhos-alkene (L7) was accurately predicted
by computation and the absolute configuration of the major
enantiomer was corroborated by X-ray crystallography.
Computational tools are rapidly developing toward being
selective pathways with the bidentate (S)-MeAnilaPhos (L6)
ligand require high activation energy, indicating the low APKR
reactivity of this class of ligand.
able to predict enantioselective catalysts in silico. However, the
I
DOI: 10.1021/jacs.7b07121
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
A. G. J. Am. Chem. Soc. 2007, 129, 5381−5383. (c) Moser, B. R. J. Nat.
Prod. 2008, 71, 487−491. (d) Murakami, N.; Ye, Y.; Kawanishi, M.;
Aoki, S.; Kudo, N.; Yoshida, M.; Nakayama, E. E.; Shioda, T.;
Kobayashi, M. Bioorg. Med. Chem. Lett. 2002, 12, 2807−2810. (e) Abd
El-Gaber, M. K.; Yasuda, S.; Iida, E.; Mukai, C. Org. Lett. 2017, 19,
320−323.
present study accentuated the challenges in the practical
application of computational ligand design. A simple “brute-
force”-type screening of the relative energies of enantioselec-
tivity-determining transition states is often not sufficient, as the
modification of ligand denticity and structure may lead to
significant change in the reaction mechanism. The present
study offers an informed approach through close collaboration
of experimental and computational chemists to achieve rational
ligand design with greater efficiency and reliability. This
iterative experimental and computational approach to catalyst
design enables real-time updates and validation to the
computational predictions, which in turn can be immediately
used to inform the next generation of experiments.
(4) Chen, B.-C.; Zhou, P.; Davis, F. A.; Ciganek, E. Organic Reactions;
John Wiley & Sons, Inc., 2004.
(5) (a) Simeonov, S. P.; Nunes, J. P. M.; Guerra, K.; Kurteva, V. B.;
Afonso, C. A. M. Chem. Rev. 2016, 116, 5744−5893. (b) Marín-
Barrios, R.; García-Cabeza, A. L.; Moreno-Dorado, F. J.; Guerra, F. M.;
Massanet, G. M. J. Org. Chem. 2014, 79, 6501−6509. (c) Smithen, D.
A.; Mathews, C. J.; Tomkinson, N. C. O. Org. Biomol. Chem. 2012, 10,
3756−3762.
(6) Krautwald, S.; Carreira, E. M. J. Am. Chem. Soc. 2017, 139, 5627−
5639.
(7) Brummond, K. M.; Davis, M. M.; Huang, C. J. Org. Chem. 2009,
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/jacs.7b07121.
■
S
74, 8314−8320.
(8) (a) Grillet, F.; Huang, C.; Brummond, K. M. Org. Lett. 2011, 13,
6304−6307. (b) Tap, A.; Jouanneau, M.; Galvani, G.; Sorin, G.;
Detailed experimental procedures and characterization
data for all new compounds, HPLC traces of compounds
10a−g (PDF)
Computational details and Cartesian coordinates of
optimized geometries (PDF)
Lannou, M.-I.; Fer
8140.
́ ́
ezou, J.-P.; Ardisson, J. Org. Biomol. Chem. 2012, 10,
(9) (a) Doan, N. T. Q.; Christensen, S. B. Curr. Pharm. Des. 2015, 21,
5501−5517. (b) Doan, N. T.; Paulsen, E. S.; Sehgal, P.; Møller, J. V.;
Nissen, P.; Denmeade, S. R.; Isaacs, J. T.; Dionne, C. A.; Christensen,
S. B. Steroids 2015, 97, 2−7. (c) Doan, N. T. Q.; Crestey, F.; Olsen, C.
E.; Christensen, S. B. J. Nat. Prod. 2015, 78, 1406−1414. (d) Winther,
A.-M. L.; Liu, H.; Sonntag, Y.; Olesen, C.; le Maire, M.; Soehoel, H.;
Olsen, C.-E.; Christensen, S. B.; Nissen, P.; Møller, J. V. J. Biol. Chem.
2010, 285, 28883−28892. (e) Mahalingam, D.; Wilding, G.;
Denmeade, S.; Sarantopoulas, J.; Cosgrove, D.; Cetnar, J.; Azad, N.;
Bruce, J.; Kurman, M.; Allgood, V. E.; Carducci, M. Br. J. Cancer 2016,
114, 986−994.
X-ray crystallographic data for compound 10f (CIF)
AUTHOR INFORMATION
■
Corresponding Authors
*pengliu@pitt.edu
*kbrummon@pitt.edu
ORCID
(10) (a) Chu, H.; Smith, J. M.; Felding, J.; Baran, P. S. ACS Cent. Sci.
2017, 3, 47−51. (b) Chen, D.; Evans, P. A. J. Am. Chem. Soc. 2017,
139, 6046−6049.
Gang Lu: 0000-0002-7319-4315
Peng Liu: 0000-0002-8188-632X
Kay M. Brummond: 0000-0003-3595-6806
(11) We expect that an asymmetric APKR will ultimately enable
stereoselective access to C-2 analogues of the bicyclo[5.3.0]decane
ring system. For example, despite recent syntheses of Tg by Baran and
Evans,10 to date only four C-2 Tg analogues have been prepared, and
when compared to Tg, their activities ranged from 10 times more
potent to 40 times less potent.9a Moreover, naturally occurring Tg
analogues which lack a C-2 ester show decreased SERCA binding
affinity.9d
Notes
The authors declare no competing financial interest.
The crystallographic data is deposited with the Cambridge
Structural Database under Reference No. CCDC 1547487.
ACKNOWLEDGMENTS
■
We thank the NIH (GM-54161) and the NSF (CHE-1654122)
for funding. Calculations were performed at the Center for
Research Computing at the University of Pittsburgh and the
Extreme Science and Engineering Discovery Environment
(XSEDE) supported by NSF.
(12) Jorgensen, L.; McKerrall, S. J.; Kuttruff, C. A.; Ungeheuer, F.;
Felding, J.; Baran, P. S. Science 2013, 341, 878−882.
(13) Bhat, V.; Welin, E. R.; Guo, X.; Stoltz, B. M. Chem. Rev. 2017,
117, 4528−4561.
̂
(14) Kim, D. E.; Kim, I. S.; Ratovelomanana-Vidal, V.; Genet, J.-P.;
Jeong, N. J. Org. Chem. 2008, 73, 7985−7989.
(15) (a) Furusawa, T.; Morimoto, T.; Ikeda, K.; Tanimoto, H.;
Nishiyama, Y.; Kakiuchi, K.; Jeong, N. Tetrahedron 2015, 71, 875−881.
(b) Kwong, F. Y.; Lee, H. W.; Qiu, L.; Lam, W. H.; Li, Y.-M.; Kwong,
H. L.; Chan, A. S. C. Adv. Synth. Catal. 2005, 347, 1750−1754.
(16) Sawano, T.; Thacker, N. C.; Lin, Z. K.; McIsaac, A. R.; Lin, W.
B. J. Am. Chem. Soc. 2015, 137, 12241−12248.
REFERENCES
■
(1) (a) Brady, S. F.; Singh, M. P.; Janso, J. E.; Clardy, J. J. Am. Chem.
Soc. 2000, 122, 2116−2117. (b) Zhao, Y.-M.; Gu, P.; Tu, Y.-Q.; Fan,
C.-A.; Zhang, Q. Org. Lett. 2008, 10, 1763−1766. (c) Castro, V.;
Murillo, R.; Klaas, C. A.; Meunier, C.; Mora, G.; Pahl, H. L.; Merfort, I.
Planta Med. 2000, 66, 591−595. (d) Wang, Q.; Chen, T.-H.; Bastow,
K. F.; Morris-Natschke, S. L.; Lee, K.-H.; Chen, D.-F. J. Nat. Prod.
2013, 76, 305−310.
(17) (a) Ricker, J. D.; Geary, L. M. Top. Catal. 2017, 60, 609−619.
(b) Quintero-Duque, S.; Dyballa, K. M.; Fleischer, I. Tetrahedron Lett.
́
2015, 56, 2634−2650. (c) Lledo, A.; Verdaguer, X.; Riera, A. The
(2) (a) Wender, P. A.; Jesudason, C. D.; Nakahira, H.; Tamura, N.;
Tebbe, A. L.; Ueno, Y. J. Am. Chem. Soc. 1997, 119, 12976−12977.
(b) Urabe, D.; Asaba, T.; Inoue, M. Bull. Chem. Soc. Jpn. 2016, 89,
1137−1144. (c) Hao, H.-D.; Trauner, D. J. Am. Chem. Soc. 2017, 139,
4117−4122. (d) Xu, G.; Hou, A.-J.; Zheng, Y.-T.; Zhao, Y.; Li, X.-L.;
Peng, L.-Y.; Zhao, Q.-S. Org. Lett. 2006, 9, 291−293. (e) Zhang, Y.; Di,
Y.; He, H.; Li, S.; Lu, Y.; Gong, N.; Hao, X. Eur. J. Org. Chem. 2011,
2011, 4103−4107.
Pauson−Khand Reaction; John Wiley & Sons, Ltd., 2012; pp 147−180.
(18) (a) Zhang, Z. B.; Bender, C. F.; Widenhoefer, R. A. J. Am. Chem.
Soc. 2007, 129, 14148−14149. (b) Osborne, J. D.; Randell-Sly, H. E.;
Currie, G. S.; Cowley, A. R.; Willis, M. C. J. Am. Chem. Soc. 2008, 130,
17232−17233. (c) Chen, M.; Roush, W. R. J. Am. Chem. Soc. 2011,
133, 5744−5747. (d) Tran, D. N.; Cramer, N. Angew. Chem., Int. Ed.
2013, 52, 10630−10634. (e) Trost, B. M.; Fandrick, D. R.; Dinh, D. C.
J. Am. Chem. Soc. 2005, 127, 14186−14187. (f) Kawai, J.; Chikkade, P.
K.; Shimizu, Y.; Kanai, M. Angew. Chem., Int. Ed. 2013, 52, 7177−7180.
(3) (a) Xu, C.; Han, A.; Virgil, S. C.; Reisman, S. E. ACS Cent. Sci.
2017, 3, 278−282. (b) Ren, F.; Hogan, P. C.; Anderson, A. J.; Myers,
J
DOI: 10.1021/jacs.7b07121
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(19) Commentaries and tutorial reviews regarding computational
catalyst design: (a) Poree, C.; Schoenebeck, F. Acc. Chem. Res. 2017,
50, 605−608. (b) Peng, Q.; Duarte, F.; Paton, R. S. Chem. Soc. Rev.
2016, 45, 6093−6107. (c) Sunoj, R. B. Acc. Chem. Res. 2016, 49,
1019−1028. (d) Houk, K. N.; Cheong, P. H.-Y. Nature 2008, 455,
309−313.
(20) Selected examples of transition metal catalysts identified by
computation: (a) Straker, R. N.; Peng, Q.; Mekareeya, A.; Paton, R. S.;
Anderson, E. A. Nat. Commun. 2016, 7, 10109. (b) Nielsen, M. C.;
Bonney, K. J.; Schoenebeck, F. Angew. Chem., Int. Ed. 2014, 53, 5903−
5906. (c) Doney, A. C.; Rooks, B. J.; Lu, T.; Wheeler, S. E. ACS Catal.
2016, 6, 7948−7955.
(21) (a) Jeong, N.; Sung, B. K.; Choi, Y. K. J. Am. Chem. Soc. 2000,
122, 6771−6772. (b) Kim, D. E.; Ratovelomanana-Vidal, V.; Jeong, N.
Adv. Synth. Catal. 2010, 352, 2032−2040.
(22) Gawley, R. E. J. Org. Chem. 2006, 71, 2411−2416.
(23) Scheffel, D. J.; Cole, A. R.; Jung, D. M.; Schiavelli, M. D. J. Am.
Chem. Soc. 1980, 102, 267−270.
(24) (a) Nelson, D. J.; Nolan, S. P. Chem. Soc. Rev. 2013, 42, 6723−
6753. (b) Voutchkova, A. M.; Appelhans, L. N.; Chianese, A. R.;
Crabtree, R. H. J. Am. Chem. Soc. 2005, 127, 17624−17625.
(25) (a) Rodriguez, A. M.; Prieto, P. Tetrahedron 2016, 72, 7443−
7448. (b) Fager-Jokela, E.; Muuronen, M.; Khaizourane, H.; Vazquez-
Romero, A.; Verdaguer, X.; Riera, A.; Helaja, J. J. Org. Chem. 2014, 79,
10999−11010. (c) Yamanaka, M.; Nakamura, E. J. Am. Chem. Soc.
2001, 123, 1703−1708.
(26) (a) Baik, M. H.; Mazumder, S.; Ricci, P.; Sawyer, J. R.; Song, Y.
G.; Wang, H.; Evans, P. A. J. Am. Chem. Soc. 2011, 133, 7621−7623.
(b) Wang, H.; Sawyer, J. R.; Evans, P. A.; Baik, M. H. Angew. Chem.,
Int. Ed. 2008, 47, 342−345.
(27) Bayden, A. S.; Brummond, K. M.; Jordan, K. D. Organometallics
2006, 25, 5204−5206.
(28) Fan, B. M.; Xie, J. H.; Li, S.; Tu, Y. Q.; Zhou, Q. L. Adv. Synth.
Catal. 2005, 347, 759−762.
(29) (a) Chikkali, S. H.; van der Vlugt, J. I.; Reek, J. N. H. Coord.
Chem. Rev. 2014, 262, 1−15. (b) Franke, R.; Selent, D.; Borner, A.
Chem. Rev. 2012, 112, 5675−5732. (c) Wassenaar, J.; Reek, J. N. H.
Org. Biomol. Chem. 2011, 9, 1704−1713.
(30) Vallianatou, K. A.; Kostas, I. D.; Holz, J.; Borner, A. Tetrahedron
Lett. 2006, 47, 7947−7950.
(31) (a) Rossler, S. L.; Krautwald, S.; Carreira, E. M. J. Am. Chem.
̈
Soc. 2017, 139, 3603−3606. (b) Defieber, C.; Ariger, M. A.; Moriel, P.;
Carreira, E. M. Angew. Chem., Int. Ed. 2007, 46, 3139−3143.
(c) Drinkel, E.; Briceno, A.; Dorta, R. Organometallics 2010, 29,
̃
2503−2514. (d) Bhaskararao, B.; Sunoj, R. B. J. Am. Chem. Soc. 2015,
137, 15712−15722.
(32) Ojima, I.; Vu, A. T.; Bonafoux, D. Science of Synthesis; Thieme
Chemistry, 2001; pp 531−532.
(33) (a) Bickelhaupt, F. M.; Houk, K. N. Angew. Chem., Int. Ed. 2017,
56, 10070−10086. (b) Fernandez, I.; Bickelhaupt, F. M. Chem. Soc.
Rev. 2014, 43, 4953−4967. (c) Ess, D. H.; Houk, K. N. J. Am. Chem.
Soc. 2008, 130, 10187−10198.
K
DOI: 10.1021/jacs.7b07121
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX