Stereodivergent Rhodium(III)-Catalyzed cis-Cyclopropanation Enabled by Multivariate Optimization

Article abstract of DOI:10.1021/jacs.8b04243

The design of stereodivergent transformations is of great interest to the synthetic community as it allows funneling of a given reaction pathway toward one stereochemical outcome or another by only minor adjustments of the reaction setup. Herein, we present a physical organic approach to invert the sense of induction in diastereoselective cyclopropanation of alkenes with N-enoxyphthalimides through rhodium(III) catalysis. Careful parametrization of catalyst-substrate molecular determinants allowed us to interrogate linear-free energy relationships and establish an intuitive and robust statistical model that correlates an extensive number of data points in high accuracy. Our multivariate correlations-steered mechanistic investigation culminated with a robust and general diastereodivergent cyclopropanation tool where the switch from trans- to cis-diastereoinduction is attributed to a mechanistic dichotomy. Selectivity might be determined by the flexibility of rhodacyclic intermediates derived from ring-opened versus -unopened phthalimides, induced by both their respective ring size and the Sterimol B₁ parameter of the Cp^X ligand on rhodium.

Full text of DOI:10.1021/jacs.8b04243

Article
pubs.acs.org/JACS
Cite This: J. Am. Chem. Soc. XXXX, XXX, XXX−XXX
Stereodivergent Rhodium(III)-Catalyzed cis-Cyclopropanation
Enabled by Multivariate Optimization
Tiffany Piou,†,§ Fedor Romanov-Michailidis,†,§ Melissa A. Ashley,†,⊥ Maria Romanova-Michaelides,‡,⊥
and Tomislav Rovis*,†
†Department of Chemistry, Columbia University, New York, New York, United States 10027
‡Department of Biochemistry, University of Geneva, 1211 Geneva 4, Switzerland
S
* Supporting Information
ABSTRACT: The design of stereodivergent transformations
is of great interest to the synthetic community as it allows
funneling of a given reaction pathway toward one stereo-
chemical outcome or another by only minor adjustments of
the reaction setup. Herein, we present a physical organic
approach to invert the sense of induction in diastereoselective
cyclopropanation of alkenes with N-enoxyphthalimides
through rhodium(III) catalysis. Careful parametrization of
catalyst−substrate molecular determinants allowed us to
interrogate linear-free energy relationships and establish an
intuitive and robust statistical model that correlates an
extensive number of data points in high accuracy. Our
multivariate correlations-steered mechanistic investigation
culminated with a robust and general diastereodivergent cyclopropanation tool where the switch from trans- to cis-
diastereoinduction is attributed to a mechanistic dichotomy. Selectivity might be determined by the flexibility of rhodacyclic
intermediates derived from ring-opened versus -unopened phthalimides, induced by both their respective ring size and the
Sterimol B1 parameter of the CpX ligand on rhodium.
Scheme 1. Diastereodivergent Rh(III)-Catalyzed
Cyclopropanation Enabled by C−H Bond Activation
INTRODUCTION
■
From the perspective of reaction development and process
efficiency, modulating the stereoselectivity outcome of
synthetic transformations by fine-tuning of the physicochem-
ical properties of the catalyst is highly desirable. Achieving
stereodivergency of a given reaction is a daunting task and is
typically the result of ad hoc adjustments of substrate or
catalyst structure, and/or reaction conditions. Hence, strategies
developed to this end cannot readily be applied to the design
of other stereodivergent reactions.1 Traditionally, substrate
modifications have been widely utilized to drive the reaction to
the desired stereochemical pathway, but their applications tend
to be limited to specifically engineered precursors.2−4 In this
context, catalyst design is more challenging but ideal for
developing versatile transformations.5−8 Thus, broadly appli-
cable transformations that display catalyst-controllable stereo-
divergency are in much demand, particularly as a means to
accelerate the screening process and structure−activity
relationship studies in drug discovery.9,10
lyst (Cp*Rh(III)) delivers the cyclopropane in a modest 2:1
ratio favoring the trans-diastereoisomer. On the other hand,
the use of a previously unreported monoisopropyl cyclo-
pentadienyl ligand (Cpi‑Pr) allows highly trans-diastereoselec-
tive cyclopropanation of activated alkenes.13 Cyclopropanes
are reasonably common in natural products,14−16 but their
stereoselective synthesis remains challenging. Although the
In the context of our program in rhodium(III) catalysis of
C−H activation,11,12 we have recently discovered an
unexpected cyclopropanation reaction of electron-deficient
alkenes using N-enoxyphthalimides (Scheme 1).13 Key to
achieving high trans-stereoselectivity is the design of the η5-
cyclopentadienyl ligand (CpX) on rhodium. For example, the
prototypical pentamethylcyclopentadienylrhodium(III) cata-
Received: May 7, 2018
© XXXX American Chemical Society
A
DOI: 10.1021/jacs.8b04243
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 1. Optimization of the reaction conditions. (A) Evaluation of Rh catalysts with N-enoxyphthalimide 1Aa. (B) Probing the impact of N-
enoxyphthalimide electronics with catalyst Rh1.
formation of tri- and tetra-substituted cyclopropanes can be
achieved routinely with very high diastereoselectivities using
metal-catalyzed insertion of diazo reagents,17 accessing 1,2-
disubstituted cyclopropanes is less straightforward. Indeed, the
majority of the developed methods afford their trans-
diastereomers; access to the corresponding cis-congeners has
remained elusive and relies on only a few available synthetic
approaches.18 Consequently, the finding of cis-diastereoselec-
tive reaction conditions would balance the existing method-
ologies and render the overall rhodium(III)-catalyzed cyclo-
propanation technology stereodivergent. Herein, we describe
the implementation of this objective.
highly correlative in a number of multidimensional regression
models. We envisioned that this approach would translate well
to the development of a diastereodivergent cyclopropanation
reaction, revealing the underlying physical organic parameters
responsible for cis-selectivity, which may in turn be exploited to
deliver a synthetically useful method.
With this idea in mind, we performed Sigman’s multidimen-
sional correlation analysis to gain additional information about
the features of the catalyst that contribute to stereoselectivity
(Figure 2).20 A set of parameters for several CpX-ligated
rhodium(III) complexes (Rh) was computed using the
molecular models in Figure 2B. This parameter set includes
IR vibrational frequencies (νS, an electronic descriptor), as well
as Tolman cone angles (ΘS) and Sterimol parameters (B1 and
B5, steric descriptors) that we previously reported for a large
number of piano-stool CpXRh(III) complexes (see the
Supporting Information). Given the preliminary results, N-
enoxyphthalimides provided an additional leverage for
perturbation. In order to describe the stereoelectronic features
of these substrates, methoxyphthalimides were used as
surrogates and their structures were computationally optimized
at the B3LYP/6-311+G** level of theory21,22 and the PBF
solvation model for methanol.23 From the optimized geo-
metries, a number of electronic descriptors such as electro-
philicity indicies, NBO charges,24 IR frequencies, dipole
moments, and polarizabilities25 were enumerated (see Figure
2B and the Supporting Information).
RESULTS AND DISCUSSION
■
Our efforts toward developing a cis-selective cyclopropanation
reaction began by evaluating the influence of cyclopentadienyl
ligands and N-enoxyphthalimides independently (Figure 1).
First, 15 CpX-ligated rhodium(III) complexes (Rh1−15) were
tested with the unsubstituted N-enoxyphthalimide 1Aa as
substrate and ethyl acrylate 2a as its coupling partner (Figure
1A). Diastereoselectivity inversion occurs with increasingly
encumbered CpX ligands, plateauing at a rather disappointing
64:36 diastereomer ratio with the bis-phenyl-Cp-ligated
catalyst Rh8 (Figure 1A, entry 8). In parallel, we also
examined the impact of N-enoxyphthalimide electronics
reasoning that while this change involves substrate engineering,
in situ cleavage of the N−O bond renders it a traceless
controller for the reaction. In the event, diastereoselectivity
was modest among the various phthalimide groups tested, all
of which favor the trans-cyclopropane diastereomer when used
with the prototypical Cp*Rh(III) catalyst (Rh1), save the
most electron-deficient 1F which delivers the cis-adduct with a
small preference of 51:49 dr (Figure 1B, entry 6).
With these descriptors in hand, we turned our attention to
designing a data set in which both the Rh catalyst and
phthalimide substrate are perturbed. To this end, 8 N-
enoxyphthalimides 1A−H and 12 rhodium(III) catalysts
Rh1−12 were selected and all possible permutations were
evaluated under the reaction conditions. The resultant cis/
trans-diastereomer ratios are presented in a color-coded matrix
in Figure 2A. When taking a closer look at the Rh1 column
We have recently reported a set of parameters for CpX
ligands on rhodium(III) where the intricacies of steric and
electronic effects are reduced to a collection of easily measured
molecular descriptors.19 These parameters have proven to be
(prototypical Cp*Rh(III) catalyst), a slight increase in dr
Phth
1/2
values is observed when moving from electron-rich (1D, E
B
DOI: 10.1021/jacs.8b04243
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
selectivity is located around the Rh2/1F and Rh2/1G
coordinates (75:25 and 70:30 dr). Of note, the cyclohexyl-
Cp-ligated complex (Rh2) and Rh8 share comparably large
cone angles (ΘS = 187 and 191°, respectively), a feature that
we anticipate to be important for high cis-diastereoinduction
(vide infra).
The high variability of dr values across catalyst/substrate
pairs is indicative of the magnitude of steric and electronic
coupling between catalyst and substrate in the selectivity-
imparting transition state (TS). One way to pinpoint such
interactions is by performing a multivariate correlation analysis
of our system.26 Accordingly, when the acquired parameters
are correlated with the measured dr values, expressed as ΔΔG‡,
the multidimensional model in Figure 2C is obtained that
encompasses the entire catalyst−substrate data set (96
experimental outcomes). This model presents a good
correlation (R2 = 0.93, intercept = −0.02) and is statistically
validated by the leave-one-out cross validation (L1O = 0.92).
Four parameters appear in the equation: one phthalimide-
Phth
1/2
based electronic descriptor, the redox potential (E ), and
three Rh catalyst-bound parameters, namely, the CO
symmetric stretching frequency (νS), Sterimol B1, and Sterimol
cone angle (ΘS) values.27 As these are normalized models, the
correlation coefficients denote the relative magnitude of each
effect. While the Sterimol cone angle (ΘS, normalized
coefficient: +0.17) describes the overall isotropic volume of
the CpX ligand, the larger coefficient (+0.78) of the Sterimol
B1 parameter underlines the importance of orientational
effects. Additionally, an interaction cross-term appears between
Phth
parameters E
and B1. Although the coefficient is lower
1/2
(+0.05), its inclusion improves the overall quality of the fit.28
On the basis of some literature precedence that crossed terms
signal direct noncovalent catalyst-substrate interaction,25 we
hypothesize that the electronic susceptibility of the phthali-
Phth
mide core to nucleophilic ring-opening (represented by E
)
1/2
and the accessible orientations of the CpX ring on the Rh
catalyst (represented by B1 and ΘS) dictate the geometry of
catalyst−substrate interaction at the selectivity-imparting TS.
Highest selectivity sits on a saddle point between a sterically
encumbered catalyst like Rh8 (B1 = 3.98 Å, ΘS = 191°) or
Rh2 (B1 = 3.96 Å, ΘS = 187°) and an electron-deficient
Phth
Phth
1/2
substrate like 1F (E
= −1.14 V) or 1G (E
= −1.52 V).
1/2
With the hypothesis that phthalimide ring-opening is
required, we further examined the role of Lewis acidic
additives on diastereoselectivity in the form of carboxylate
base counterions. Several alkali metal acetates were surveyed;29
the most drastic effect is seen when switching from CsOAc to
NaOAc with a more than 2-fold increase in cis-diastereose-
lectivity (from 31:69 to 67:33 dr) for the prototypical Rh1/1A
catalyst−substrate combination. Importantly, dr values rise
significantly for most other catalyst/substrate pairs as well
(color-coded matrix in Figure 3A). In search of a rationale of
this unexpected counterion effect, we turned our efforts to
establishing multivariate correlations for our 96-membered
catalyst-substrate panel using NaOAc in place of CsOAc
(Figure 3C).
Figure 2. Optimization of reaction conditions by simultaneous
variation of Rh catalyst and phthalimide structures with CsOAc base.
(A) Heat-map of cis/trans-diastereomer ratios. (B) Models used for
parametrization. (C) Correlation plot of normalized molecular
descriptors to diastereomer ratios.
Phth
= −1.80 V, 27:73 dr; 1H, E
= −1.86 V, 24:76 dr) to
Phth
1/2
Phth
electron-deficient (1F, E
= −1.14 V, 51:49 dr; 1G, E
=
1/2
1/2
−1.52 V, 44:56 dr) phthalimide cores. Contrariwise, highly
trans-diastereoselective reactions are seen (from 13:87 to 6:94
dr, blue color) when examining Rh10, 11, and 12 columns that
correspond to tetra-, tri-, and monosubstituted CpX rings,
respectively. More synthetically useful cis selectivities are
observed with other CpXRh(III)/phthalimide combinations
(yellow-orange color), the two most selective catalyst/
substrate pairs being the bis-phenyl-Cp complex Rh8 with
electron-deficient nitro- (1F) and dichloro-substituted (1G)
phthalimides, delivering product in 88:12 and 86:14 dr,
respectively. In addition to Rh8, a second region of high
It turned out that this new data set is sensibly less
correlatable than the previously described CsOAc set and
requires a minimum of five parameters for statistical robustness
(R2 = 0.90, L1O = 0.89). Phthalimides are now best described
Phth
M06‑2x
by μB3LYP (dipole moment) and α
(mean polarizability);
while Rh catalysts by νS, qRh (electronic) and B1, ΘS (steric)
(Figure 3B). Closer inspection of the resultant correlation plot
C
DOI: 10.1021/jacs.8b04243
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
subgroups, one for each cis- and trans-outcome, followed by
building local correlation models, which displayed improved
statistical performance (R2 = 0.96 and 0.98; Q2 = 0.93 and
0.95). Of note, while the trans local model is largely
Phth
M06‑2x
electronically driven (parameters α
and qRh), the cis
local model has a large steric contribution (coefficient +0.53 in
front of ΘS). The appearance of correlative subgroups can be
traced to the existence of two mechanistic regimes. While the
mechanistic change is gradual with CsOAc and can be
accounted for by the parameters space, this event is more
abrupt with NaOAc and leads to clustering of correlations. The
best-performing catalyst/substrate pairs are once again Rh2,
Rh8 in combination with 1F, 1G. Despite being slightly
inferior to the Rh8/1F pair in terms of dr value, the
cyclohexyl-Cp-ligated catalyst Rh2 together with dichloro-
substituted phthalimide 1G gives higher isolated yield.
Therefore, it was selected for further investigations.
The generality of our cis-cyclopropanation methodology was
evaluated next with catalyst Rh2, NaOAc base, various 4,5-
dichlorophthalimides 1G, and alkenes 2 (Scheme 2).
Scheme 2. Generality of the cis-Cyclopropanation Reaction:
(A) Optimized Conditions; (B) N-Enoxyphthalimide Scope;
(C) Alkene Scope
Figure 3. Optimization of reaction conditions by simultaneous
variation of Rh catalyst and phthalimide structures with NaOAc base.
(A) Heat-map of cis/trans-diastereomer ratios. (B) Models used for
parametrization. (C) Correlation plot of normalized molecular
descriptors to diastereoselectivity (in the form of ΔΔG‡ values).
exposes clustering of catalyst−substrate combinations in two
regions according to their diastereoselectivity outcomes: (i)
penta-substituted CpX ligands Rh1−8 in combination with
more electron-deficient phthalimides like 1B, 1C, and 1E-G
tend to bias the system toward phthalimide ring-opening
resulting in higher cis-selectivity, whereas (ii) tetra-, tri-, and
monosubstituted ligands Rh10-12 lead to trans-selective
reactions when used with more electron-rich phthalimides
like 1A-E, 1H. The remaining permutations that include
electronically and sterically mismatched catalyst-substrate pairs
form a cluster displaying borderline selectivities that are poorly
predicted by the parameter space (Figure 3C). The clustering
behavior was scrutinized by splitting the entire data set in two
All reactions smoothly took place under mild conditions to
furnish cis-cyclopropanes 3 in good yields (52−88%) and high
diastereoselectivities (up to >20:1 dr). While electron-
withdrawing substituents on the N-enoxyphthalimide rendered
the cyclopropanation reaction more cis-selective (3ca, 3da, 3fa,
3ha, and 3ia), electron-donating groups provided slightly lower
D
DOI: 10.1021/jacs.8b04243
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
dr (3ba and 3ga). The position of the substituents (ortho,
meta, or para) is less critical for the efficiency of this
transformation (3ea, 3ha, and 3ca). The reaction was
compatible with versatile functional groups on the alkene
component as representatively demonstrated by ester (3ab,
3ac, 3ad, 3ae, and 3ag), ketone (3aj and 3ak), nitrile (3ah),
sulfone (3ai), and amide (3al). In addition, the reaction is
chemoselective and engages solely electron-deficient double
bonds (3af).
that the dr values are under kinetic control under both the
trans-13 and cis-selective reaction conditions.30
Mechanistically, the crucial step is cleavage of the alkenyl
C−H bond in 1 by in situ generated CpXRh(III) acetato-
complex I (Scheme 3C). To account for correlations and
isotope labeling studies, we propose the occurrence of two
parallel reaction routes: C−H activation can ensue either when
directed by the intact phthalimide moiety (Route a) such as
proposed for trans-cyclopropanation13,31 or through a solvent-
opened phthalimide ring II that acts as a bidentate ligand
(Route b).32 In the former case, C−H cleavage is rate-
determining and consistent with the presence of a KIE, while
in the latter case phthalimide ring-opening is presumably
turnover-limiting and no KIE is expected. Next, the resultant
seven- (III) or five-membered (VII) metalacycles undergo
alkene 2 migratory insertion leading to ring-expanded cyclic
intermediates IV or VIII, respectively, which are poised for a
second (intramolecular) migratory insertion into the enolic
carbon−carbon double bond that is selectivity-determining.
The mechanistic rationale behind this hypothesis evokes rapid
conformational equilibration of rhodacycles IV and VIII prior
to the second migratory insertion step. Equilibrium position
depends on the ring-size of the metalacyclic intermediate as
well as the steric and electronic features of the phthalimide
group and CpX ligand. In Route a, the small Sterimol B1 value
of Cpi‑Pr combined with conformationally flexible puckering of
a 9-membered ring lessens steric clash between the pseudoaxial
EWG group and Cp-ligand in conformer V, allowing formation
of thermodynamically more stable trans-cyclopropane. Contra-
riwise, in Route b, a merger of the large Sterimol B1 of
cyclohexyl-Cp with stiff puckering of a 7-membered ring leads
to prohibitive steric interactions with the EWG in IX. In this
case, the kinetic cis product is favored. The susceptibility of
phthalimide to ring-opening and thus to engage in Route b is
Next, a series of deuterium labeling experiments were
performed to interrogate the working model (Scheme 3).
Scheme 3. Proposed Mechanism: (A) KIE under trans-
Selective Conditions; (B) KIE under cis-Selective
Conditions; (C) Dichotomous Reaction Pathways
controlled by its electrophilicity, explaining the interplay
Phth
1/2
between parameters E
and B1 in the correlation equation.
Finally, the effect of base additive can be tentatively explained
by the higher Lewis acidity of Na+ versus Cs+ cations leading to
faster phthalimide ring-opening.
We assume that the coordinative interaction of the opened
phthalimide ester group with rhodium(III) in intermediates IV
and VIII is weakened due to presence of electron-withdrawing
substituents on phthalimide that reduce the overall Lewis
basicity of the carbonyl oxygens. While for IV this renders the
9-membered rhodacycle conformationally flexible and thus less
sensitive to steric interaction between CpX and EWG, for VIII
the remaining X-type ligation from the nitrogen atom of the
opened phthalimide suffers less from the presence of electron-
withdrawing substituents and maintains the stiffness of the 7-
membered rhodacycle. This might explain the higher cis-
selectivities that we observe with electron-deficient N-enoxy-
phthalimides. Additionally, the weakly coordinating ester side
arm is outcompeted by the enolic carbon−carbon double bond
that binds tightly to rhodium(III) in VIII favoring cyclo-
propanation over carboamination.
While a significant kinetic isotope effect (KIE) of 2.0 is seen
under trans-selective cyclopropanation conditions (Scheme
3A),13 no KIE is observed for the cis-cyclopropanation reaction
(Scheme 3B), signaling that C−H activation is most probably
not the rate-determining step for the latter. Taken together,
these experimental results highlight the existence of a
mechanistic dichotomy in pathways leading to cis- or trans-
adducts, with a more electron-deficient phthalimide suggestive
of an in situ ring-opening impacting diastereoselectivity.
Additionally, cis−trans-isomerization experiments have shown
CONCLUSIONS
■
In summary, we have demonstrated that diastereoselectivity in
CpXRh(III)-catalyzed cyclopropanation of alkenes using N-
enoxyphthalimides can be switched from trans to cis by careful
modulation of the stereoelectronic properties of substrate and
catalyst. Guided by a 96-membered data set that was designed
according to geometric, steric, and electronic criteria and used
E
DOI: 10.1021/jacs.8b04243
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Chem. - Eur. J. 2010, 16, 11212. (c) Patureau, F. W.; Glorius, F.
Angew. Chem., Int. Ed. 2011, 50, 1977. (d) Patureau, F. W.; Wencel-
Delord, J.; Glorius, F. Aldrichimica Acta 2012, 45, 31. (e) Song, G.;
Wang, F.; Li, X. Chem. Soc. Rev. 2012, 41, 3651. (f) Chiba, S. Chem.
Lett. 2012, 41, 1554. (g) Piou, T.; Rovis, T. Acc. Chem. Res. 2018, 51,
170.
(12) For pioneering work, see: (a) Ueura, K.; Satoh, T.; Miura, M.
Org. Lett. 2007, 9, 1407. (b) Umeda, N.; Tsurugi, H.; Satoh, T.;
Miura, M. Angew. Chem., Int. Ed. 2008, 47, 4019. (c) Stuart, D. R.;
Burgess, K. M. N.; Fagnou, K.; Bertrand-Laperle, M. J. Am. Chem. Soc.
2008, 130, 16474. (d) Fukutani, T.; Umeda, N.; Hirano, K.; Satoh,
T.; Miura, M. Chem. Commun. 2009, 5141. (e) Morimoto, K.; Hirano,
K.; Satoh, T.; Miura, M. Org. Lett. 2010, 12, 2068. (f) Mochida, S.;
Umeda, N.; Hirano, K.; Satoh, T.; Miura, M. Chem. Lett. 2010, 39,
744. (g) Hyster, T. K.; Rovis, T. J. Am. Chem. Soc. 2010, 132, 10565.
(h) Rakshit, S.; Grohmann, C.; Besset, T.; Glorius, F. J. Am. Chem.
Soc. 2011, 133, 2350.
(13) Piou, T.; Rovis, T. J. Am. Chem. Soc. 2014, 136, 11292.
(14) Ebner, C.; Carreira, E. M. Chem. Rev. 2017, 117, 11651.
(15) Evans, D. A.; Woerpel, K. A.; Hinman, M. M.; Faul, M. M. J.
Am. Chem. Soc. 1991, 113, 726.
(16) Chen, Y.; Ruppel, J. V.; Zhang, X. P. J. Am. Chem. Soc. 2007,
129, 12074.
(17) (a) Chanthamath, S.; Takaki, S.; Shibatomi, K.; Iwasa, S.
Angew. Chem., Int. Ed. 2013, 52, 5818. (b) Lindsay, V. N. G.; Nicolas,
C.; Charette, A. B. J. Am. Chem. Soc. 2011, 133, 8972. (c) Rosenberg,
to construct quantitative structure−selectivity relationships, we
promptly recognized that phthalimide ring-opening in
conjunction with appropriate sterics of the CpX ligand on
rhodium(III) play critical and synergistic roles in the
diastereoselectivity-determining TS. Selectivity might be
determined by the flexibility of rhodacycles VIII and IV
derived from ring-opened versus -unopened phthalimides,
respectively, induced by both their respective ring size (7- or 9-
membered ring) and the Sterimol B1 parameter of the CpX
ligand. The insights gained in the course of this study can be
generalized to countless situations of catalyst-controlled
selectivity, where multivariate regression analysis could supply
a guided walk toward reaction optimization.
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/jacs.8b04243.
■
S
Supplementary text, additional Figures S1−S46, Tables
S1 and S2, NMR spectra, computational details,
statistical model development, and additional references
1−18 (PDF)
Parameters of complexes (XLSX)
Additional experimental data (XLSX)
̌
́
M. L.; Vlasana; Gupta, N. S.; Wragg, D.; Tilset, M. J. Org. Chem.
2011, 76, 2465. (d) Lindsay, V. N. G.; Lin, W.; Charette, A. B. J. Am.
Chem. Soc. 2009, 131, 16383. (e) Nagashima, T.; Davies, H. M. L.
Org. Lett. 2002, 4, 1989.
AUTHOR INFORMATION
Corresponding Author
*tr2504@columbia.edu
■
(18) (a) Suematsu, H.; Kanchiku, S.; Uchida, T.; Katsuki, T. J. Am.
Chem. Soc. 2008, 130, 10327−10337. (b) Rosenberg, M. L.;
Krivokapic, A.; Tilset, M. Org. Lett. 2009, 11, 547−550.
(c) Brandenberg, O. F.; Prier, C. K.; Chen, K.; Knight, A. M.; Wu,
Z.; Arnold, F. H. ACS Catal. 2018, 8, 2629. (d) Bachmann, S.; Furler,
M.; Mezzetti, A. Organometallics 2001, 20, 2102. (e) Johansson, M. J.;
Gorin, D. J.; Staben, S. T.; Toste, F. D. J. Am. Chem. Soc. 2005, 127,
18002. (f) Hu, W.; Timmons, D. J.; Doyle, M. P. Org. Lett. 2002, 4,
901.
(19) Piou, T.; Romanov-Michailidis, F.; Romanova-Michaelides, M.;
Jackson, K. E.; Semakul, N.; Taggart, T. D.; Newell, B. S.; Rithner, C.
D.; Paton, R. S.; Rovis, T. J. Am. Chem. Soc. 2017, 139, 1296.
(20) For a review, see Sigman, M. S.; Harper, K. C.; Bess, E. N.;
Milo, A. Acc. Chem. Res. 2016, 49, 1292.
ORCID
Fedor Romanov-Michailidis: 0000-0002-0997-478X
Tomislav Rovis: 0000-0001-6287-8669
Author Contributions
§T.P. and F.R.-M. contributed equally.
Author Contributions
⊥M.R.-M. and M.A.A. contributed equally.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
We gratefully acknowledge NIGMS (GM80442) for funding
support.
■
(21) (a) Stephens, P. J.; Devlin, F. J.; Chabalowski, C. F.; Frisch, M.
J. J. Phys. Chem. 1994, 98, 11623. (b) Becke, A. D. J. Chem. Phys.
1993, 98, 5648. (c) Becke, A. D. J. Chem. Phys. 1993, 98, 1372.
(d) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B: Condens. Matter Mater.
Phys. 1988, 37, 785.
REFERENCES
■
(22) Krishnan, R.; Binkley, J. S.; Seeger, R.; Pople, J. A. J. Chem.
Phys. 1980, 72, 650.
(1) (a) Bihani, M.; Zhao, J. C.-G. Adv. Synth. Catal. 2017, 359, 534−
556. (b) Oliveira, M. T.; Luparia, M.; Audisio, D.; Maulide, N. Angew.
Chem., Int. Ed. 2013, 52, 13149.
(23) (a) Tannor, D. J.; Marten, B.; Murphy, R.; Friesner, R. A.;
Sitkoff, D.; Nicholls, A.; Ringnalda, M.; Goddard, W. A., III; Honig, B.
J. Am. Chem. Soc. 1994, 116, 11875. (b) Marten, B.; Kim, K.; Cortis,
C.; Friesner, R. A.; Murphy, R. B.; Ringnalda, M. N.; Sitkoff, D.;
Honig, B. J. Phys. Chem. 1996, 100, 11775.
(24) Glendening, E. D.; Badenhoop, J. K.; Reed, A. E.; Carpenter, J.
E.; Bohmann, J. A.; Morales, C. M.; Landis, C. R.; Weinhold, F. NBO
6.0; Theoretical Chemistry Institute, University of Wisconsin,
(2) Aggarwal, V. K.; Alonso, E.; Fang, G.; Ferrara, M.; Hynd, G.;
Porcelloni, M. Angew. Chem., Int. Ed. 2001, 40, 1433.
(3) Deng, X.-M.; Cai, P.; Ye, S.; Sun, X.-L.; Liao, W.-W.; Li, K.;
Tang, Y.; Wu, Y.-D.; Dai, L.-X. J. Am. Chem. Soc. 2006, 128, 9730.
(4) Johansson, C. C.; Bremeyer, N.; Ley, S. V.; Owen, D. R.; Smith,
S. C.; Gaunt, M. J. Angew. Chem., Int. Ed. 2006, 45, 6024.
(5) Krautwald, S.; Carreira, E. M. J. Am. Chem. Soc. 2017, 139, 5627.
(6) Conway, J. H., Jr.; Rovis, T. J. Am. Chem. Soc. 2018, 140, 135.
(7) Wodrich, M. D.; Ye, B.; Gonthier, J. F.; Corminboeuf, C.;
Cramer, N. Chem. - Eur. J. 2014, 20, 15409.
(8) Hong, S. Y.; Jeong, J.; Chang, S. Angew. Chem., Int. Ed. 2017, 56,
2408−2412.
(9) Ganesan, A. Drug Discovery Today 2002, 7, 47−55.
(10) Cernak, T.; Dykstra, K. D.; Tyagarajan, S.; Vachal, P.; Krska, S.
W. Chem. Soc. Rev. 2016, 45, 546.
Madison, WI, 2013.
(25) Polarizability (α
Phth
M06‑2x
) single-point energy calculations were
determined at the M06-2x/cc-PVTZ++ level of theory.
(26) Neel, A. J.; Milo, A.; Sigman, M. S.; Toste, F. D. J. Am. Chem.
Soc. 2016, 138, 3863.
(27) In total, two series of regression equations were constructed,
but only one is shown here because it benefits from better
convergence and includes a phthalimide-bound parameter in addition
to the catalyst ones. For a more detailed discussion on the building of
correlations, see the Supporting Information.
(11) For recent reviews, see: (a) Colby, D. A.; Bergman, R. G.;
Ellman, J. A. Chem. Rev. 2010, 110, 624. (b) Satoh, T.; Miura, M.
F
DOI: 10.1021/jacs.8b04243
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(28) Regression equations from the second series were also
improved after inclusion of an interaction cross-term; see the
Supporting Information.
(29) For KOAc, 51:49 dr is observed. For LiOAc, only trace
amounts of the desired cyclopropane product are seen, presumably
due to sparing solubility of LiOAc in TFE.
(30) An isomerization control experiment has been performed with
two different catalysts, Rh2 and Rh8: the dr values do not change as a
function of reaction conversion, and the subjection of either
diastereomer to the reaction conditions does not lead to any
detectable isomerization; see the Supporting Information.
(31) For a discussion of alternative mechanistic interpretations in
rhodium(III)-catalysis involving oxidizing directing groups, see
Vasquez-Cespedes, S.; Wang, X.; Glorius, F. ACS Catal. 2018, 8, 242.
(32) Piou, T.; Rovis, T. Nature 2015, 527, 86.
G
DOI: 10.1021/jacs.8b04243
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX