Catalyst-controlled aliphatic c-h oxidations with a predictive model for site-selectivity

Article abstract of DOI:10.1021/ja407388y

Selective aliphatic C-H bond oxidations may have a profound impact on synthesis because these bonds exist across all classes of organic molecules. Central to this goal are catalysts with broad substrate scope (small-molecule-like) that predictably enhance or overturn the substrate's inherent reactivity preference for oxidation (enzyme-like). We report a simple small-molecule, non-heme iron catalyst that achieves predictable catalyst-controlled site-selectivity in preparative yields over a range of topologically diverse substrates. A catalyst reactivity model quantitatively correlates the innate physical properties of the substrate to the site-selectivities observed as a function of the catalyst.

Full text of DOI:10.1021/ja407388y

Communication
pubs.acs.org/JACS
Catalyst-Controlled Aliphatic C−H Oxidations with a Predictive
Model for Site-Selectivity
Paul E. Gormisky and M. Christina White*
Department of Chemistry, University of Illinois Urbana−Champaign, Urbana, Illinois 61801, United States
S
* Supporting Information
ABSTRACT: Selective aliphatic C-H bond oxidations
may have a profound impact on synthesis because these
bonds exist across all classes of organic molecules. Central
to this goal are catalysts with broad substrate scope (small-
molecule-like) that predictably enhance or overturn the
substrate’s inherent reactivity preference for oxidation
(enzyme-like). We report a simple small-molecule, non-
heme iron catalyst that achieves predictable catalyst-
controlled site-selectivity in preparative yields over a
range of topologically diverse substrates. A catalyst
reactivity model quantitatively correlates the innate
physical properties of the substrate to the site-selectivities
observed as a function of the catalyst.
complete encapsulation of the active site to select based on
substrate topologythereby limiting the scope to one or a few
similar substrates. We describe a small-molecule catalyst that uses
a trajectory restriction strategy to achieve predictable, catalyst-
controlled site-selectivity while maintaining substrate generality.
In our design strategy, we endeavored to generate a small-
molecule catalyst that incorporates minimal steric blocking
elements10 to restrict the approach trajectories of certain C-H
bonds to the Fe-oxo (Figure 1). We hypothesized that such a
catalyst could alter intrinsic substrate bias by rendering catalyst/
substrate nonbonding interactions paramount while maintaining
structural flexibility such that substrates of diverse topologies are
accommodated. The 3D structure of (R,R)-Fe(PDP) (1) reveals
a wide 145° cone of possible approach trajectories of a substrate
to the putative Fe-oxo so that a combination of electronic and
steric/stereoelectronic factors influence site-selectivity variably
depending on the substrate. Modifications at the pyridine 6-
position of 1 suppressed reactivity, supporting reports that
catalysts with steric hindrance near the oxo exhibit greatly
diminished C-H oxidation reactivity.11 We thus synthesized a
catalyst with pendant aryl rings at the 5-position having o-CF3
groups. Ortho-CF3 disubstitution is ideal because its electron-
withdrawing properties deactivate the ligand toward oxidation
and its steric bulk (comparable to isopropyl, but rotationally
symmetric12) enforces a perpendicular biaryl alignment where
the CF3 groups extend toward the catalyst active site and narrow
the cone of possible approach trajectories to 76°.
mall-molecule catalysis has achieved predictable, substrate-
Scontrolled site-selective C-H oxidations with generality and
operational ease. We recently described a non-heme iron
hydroxylation catalyst, Fe(PDP) (1), that showed 3° (tertiary)
and 2° (secondary) aliphatic C-H bonds can be preparatively
differentiated based on electronic (favors electron-rich sites),
steric (favors unhindered sites), and stereoelectronic factors
(favors sites where strain relief is possible) that distinguish C-H
bonds from one another within a molecule.1 Catalyst 1 relies on
the constructive combination of these factors to favor a single site
of oxidation. While 1 provides good selectivity in many
molecules because of the pervasiveness of these inherent
reactivity differences among C-H bonds,2,3 the substrate
ultimately dictates site-selectivity. As a result, site-selectivity
suffers when individual factors diverge to favor distinct sites, and
modulating the magnitude of selectivity or achieving oxidation at
alternate sites is not possible without chemically changing the
substrate (e.g., directed oxidations).4
Catalyst-controlled selectivity that enhances or overturns the
substrate’s inherent selectivity preference is still at the forefront
in asymmetric catalysis5 and site-selective modification of
reactive functionality.6 Aliphatic C-H oxidation presents the
additional challenge of requiring a catalyst that is reactive enough
to oxidize very inert bonds, yet maintains the capacity for its
control elements to differentiate the subtle features of bonds
ubiquitous within molecules. Catalyst control is a hallmark of
enzymatic aliphatic C-H oxidations.7 However, despite signifi-
cant efforts to adopt the enzymatic strategies of utilizing shape8
and functional group recognition9 elements, efficient and general
small-molecule catalyst control has not been achieved.
Challenges associated with creating a discrete match between
catalyst and substrate have led to extreme catalyst designse.g.,
We first examined the ability of catalyst Fe(CF3-PDP) (2) to
alter the intrinsic site-selectivities of oxidation with Fe(PDP) 1
over a topologically diverse selection of substrates (Table 1).
Oxidizing linear ester (+)-3 and trans-1,2-dimethylcyclohexane
Received: July 22, 2013
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja407388y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
2°:3°, entry 7), whereas (R,R)-2 leads to a turnover of site-
selectivity affording the methylene oxidation product, γ-ketone
(+)-14, in a preparatively useful 56% yield (4:1 2°:3°, entry 8).
Catalyst-controlled reactivity can further be applied in a more
complex dipeptide setting. While (R,R)-1 affords no selectivity
for oxidizing (+)-16 due to competing electronic and steric
effects (1:1 2°:3°, entry 9), (R,R)-2 provides 51% yield of
norvaline oxidation with excellent 9:1 2°:3° selectivity (entry
10). In contrast to 1 whose selectivities are dictated by the
interplay of electronic and sterics/stereoelectronics within the
substrate, 2 relies primarily on nonbonding interactions between
the catalyst and the substrate to control site-selectivities.
Significantly, 2 affects changes in site-selectivity relative to 1
under a uniform set of simple reaction conditions (room
temperature, open to air, 0.16 M) in preparatively useful yields
(ca. 54% isolated yield, mono-oxidized product).
Table 1. Catalyst-Controlled Oxidation of Simple Cyclic and
Acyclic Molecules
To broadly impact synthetic strategy, catalysts that exert
control on site-selectivities of oxidation must do so in a
predictable way on a diverse range of complex molecules.2 We
thus developed structure-based catalyst reactivity models that
enable systematic identification of the most likely sites of
oxidation on a molecule and then quantitative description and
prediction of the site-selectivity afforded by each catalyst. To
simplify the analysis of complex molecules with many potential
sites of oxidation, we developed a site filter that identifies likely
sites of oxidation based on parametrization of electronic [E =
NPA charges] and steric/stereoelectronics [S, assigned based on
Winstein−Holness values (“A values”),14 Figure 2].15,16 These
values were systematically categorized across all substrates; only
sites with either two red (highly reactive) or one red and one
purple (moderately reactive) parameter are considered suscep-
tible to oxidation.
We next sought to develop a model that mathematically relates
each catalyst’s site-selectivities to the properties of the substrate.
We hypothesized that the difference in electronics (ΔEab = Eb −
Ea) and sterics/stereoelectronics (ΔSab = Sb − Sa), describing the
relative reactivity between the sites identified using the site filter
(a and b), could be proportional to the experimentally
determined site-selectivities (a:b)1 expressed as a difference in
transition-state energies (ΔΔG⧧ ≈ 1.36 log(a:b)). These data
(6) previously provided poor to moderate selectivity between
competing 2° (sterically more accessible) and 3° (more electron
rich) sites based on substrate control (entries 1 and 3). In
contrast, (S,S)-2 diverts reactivity toward the electronically
disfavored 2° sites by restricting access of the 3° sites to the
oxidant. The substantial improvement in site-selectivity with 2
(entries 2 and 4) affords useful levels of 2° oxidation products
(51% yield, 70% yield). In addition to enhancing selectivity in
previously poorly selective reactions, we questioned if 2 can also
completely overturn the substrate’s inherent selectivity to favor
an alternate site. Oxidizing trans-4-methylcyclohexyl acetate (10)
with (S,S)-1 provides selectivity for C4 oxidation based primarily
on electronics to afford alcohol 12 in 66% yield (entry 5). (S,S)-2
overturns this selectivity by exploiting a significant catalyst/
substrate repulsive nonbonding interaction with the C4 axial 3°
C-H bond and affords good yields (51%) of product at the
electronically deactivated C3/5 site (entry 6). Significantly, the
same effect is observed with a topologically distinct (acyclic) and
functionally dense isoleucine substrate, (+)-13. Oxidation with
(R,R)-1 affords 43% of alcohol (+)-15 as the major product (1:2
B
dx.doi.org/10.1021/ja407388y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
findings that oxidizing (−)-19 with (R,R)-1 furnishes (−)-6β-
hydroxy-triacetoxytricalysiolide B (20) in 26% yield and (−)-7-
oxo-triacetoxytricalysiolide B (21) in 18% yield with no site-
selectivity (1:1 C6:C7) and poor mass balance, suggesting
unselective oxidation at other activated sites (10% recovered
starting material). In contrast, 2’s reactivity model calculates an
11:1 C6:C7 ratio with higher mass balance due to 2’s ability to
were fit as a function of catalyst fcat(ΔEab,ΔSab) = ΔΔG⧧ to obtain
a 3D free energy relationship17 expressed by an equation for each
catalyst (Figure 3A,B).15 In examining the surface for oxidations
with 2, site-selectivity (i.e., ΔΔG⧧, Z-axis) correlates strongly
with ΔSab and is highest when there is a large difference in
sterics/stereoelectronics between two sites (ΔSab) in either
direction: the difference in electronics (ΔEab) can be negligible
or even large in the opposite direction. The correlations expressed
computationally are fully consistent with the empirical
observation that 2 induces catalyst-controlled changes in
ΔΔG⧧ as a result of nonbonding catalyst/substrate interactions.
In contrast, the surface for oxidations with 1 predicts that site-
selectivity is highest when electronic and steric/stereoelectronic
differences between two sites are large in the same direction. This
mathematically expresses the empirical observation that
oxidations with 1 are controlled by the confluence of favorable
steric/stereoelectronic and electronic properties within the
substrate. Comparing the calculated and experimental ΔΔG⧧
values for 2 and 1 for all substrates used to create the models
provides a good linear fit (Figure 3C). In addition to validating
our hypothesis1 that the basic physical organic parameters of
electronics and sterics/stereoelectronics of a substrate correlate
to site-selectivities in C-H oxidation, this finding demonstrates
for the first time that this relationship can be expressed
quantitatively and varied based on catalyst structure.
Next we evaluated the scope of 2’s ability to alter intrinsic site-
selectivities in complex molecule settings as well as the capacity
for the catalyst reactivity models to describe the divergent
selectivities. Applying the parametrized site filter to (−)-tri-
acetoxytricalysiolide B (19), a putative metabolite of the
diterpene cafestol found in coffee18 having eight potential sites
of oxidation, revealed four likely sites of oxidation: C6, C7, C11,
and C12 (Figures 2 and 4). Evaluating the electronic and steric
difference parameters between these sites indicates that the
selectivity factors are in opposition; there is a strong steric
preference for C6 and an electronic preference for C7 and C11
(Figure 4). Using C6 as our reference in 1’s reactivity model, we
calculate site-selectivity ratios of 1:1.1 (C6:C7), 1:1.4 (C6:C11),
and 4:1 (C6:C12), due to these divergent factors within the
substrate. The calculated values are consistent the experimental
respond to large steric/stereoelectronics (ΔS6,7 = 1.09, ΔS6,11
=
1.28, ΔS6,12 = 1.28). Experimentally, oxidizing (−)-19 with
(R,R)-2 affords (−)-20 in a 61% isolated yield with a significant
catalyst-dependent increase in site-selectivity of C6:C7 oxidation
from 1:1 to >10:1 (Figure 4). Interestingly, steric hindrance of
the axial hydrogen at C6 retards overoxidation of the alcohol to
the ketone by both 1 and 2. It is notable that excellent
enhancement of site-selectivity for C6 oxidation is observed with
2 despite opposing electronics favoring C7.
The greatest challenge for catalyst control is to override the
inherent site-selectivity of oxidation to favor an alternate site.
Catalyst 2 achieved this in oxidizing simple substrates 10 and
(+)-13, and we sought to further challenge 2 in a complex
molecule setting. Applying our site filter to (+)-artemisinin
(22)having nine potential sites of oxidationeliminates all
but C10 and C9 based on very unfavorable electronics and/or
sterics at alternate sites. Catalyst 1 is calculated to give a 1.3:1
C10:C9 ratio because it responds to the divergent substrate
biasing factors: a strong electronic preference for 3° oxidation at
C10 (ΔE10,9 = 1.48) and an opposing steric preference for 2°
oxidation at C9 (ΔS10,9 = −1.70).15 Consistent with this,
oxidizing (+)-22 with (S,S)-1 afforded 54% yield of (+)-10β-
hydroxy-artemisinin (24) with 23% yield of (+)-9-oxo-
artemisinin (23) in a useful 2:1 C10:C9 selectivity1a (Figure
5). Despite the substrate’s strong electronic bias favoring C10,
the reactivity model for 2 calculates a 17:1 ratio favoring C9
oxidation based on the large steric difference parameter. This
may be understood based on 2’s ability to exploit nonbonding
interactions between its biaryl ligand and the substrate’s rigid
lactone ring system to restrict approach trajectories of the
electron-rich C10 C-H bond to the Fe-oxo. Gratifyingly, (S,S)-2
dramatically turns over the substrate controlled selectivity seen
with (S,S)-1, oxidizing at the C9 site in an 11:1 C9:C10 ratio and
furnishing 52% yield of (+)-23. Catalyst 2’s ability to override
strong electronic substrate bias is analogous to what was
observed with 19, but on a topologically distinct structure.
Notably, previous to this work, only P-450 enzymes evolved in
the laboratory specifically for the oxidation of (+)-22 provided
comparable levels of selectivity for C9,19 highlighting the power
C
dx.doi.org/10.1021/ja407388y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Communication
ASSOCIATED CONTENT
* Supporting Information
Experimental procedures and characterization data. This material
is available free of charge via the Internet at http://pubs.acs.org.
■
S
AUTHOR INFORMATION
Corresponding Author
white@scs.uiuc.edu
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank Dr. D. Rogness and G. Snapper for initial amino acid
and peptide oxidation studies, Dr. D. Gray for crystallographic
analysis of (R,R)-2 and (+)-25, Prof. A. Pfaltz for a gift of Ir
catalysts, and Starbucks for coffee grounds. We are grateful to
Bristol-Meyers Squibb, Amgen, and the University of Illinois for
financial support. P.E.G. is a NSF Graduate Research Fellow.
REFERENCES
■
(1) (a) Chen, M. S.; White, M. C. Science 2007, 318, 783. (b) Chen, M.
S.; White, M. C. Science 2010, 327, 566.
(2) White, M. C. Science 2012, 335, 807.
(3) (a) Litvinas, N. D.; Brodsky, B. H.; Du Bois, J. Angew. Chem., Int.
Ed. 2009, 48, 4513. (b) Liu, W.; Groves, J. T. J. Am. Chem. Soc. 2010,
132, 12847. (c) Paradine, S. P.; White, M. C. J. Am. Chem. Soc. 2012, 134,
2036.
of 2 to access new sites of reactivity without the need for
substrate specificity.
(4) (a) Bigi, M. A.; Reed, S. A.; White, M. C. J. Am. Chem. Soc. 2012,
134, 9721. (b) Desai, L. V.; Hull, K. L.; Sanford, M. S. J. Am. Chem. Soc.
2004, 126, 9542. (c) Giri, R.; Chen, X.; Yu, J.-Q. Angew. Chem., Int. Ed.
2005, 44, 2112. (d) Simmons, E. M.; Hartwig, J. F. Nature 2012, 483, 70.
(5) Balskus, E. P.; Jacobsen, E. N. Science 2007, 317, 1736.
(6) Lewis, C. A.; Miller, S. J. Angew. Chem., Int. Ed. 2006, 45, 5616.
(7) (a) Lewis, J. C.; Coelho, P. S.; Arnold, F. H. Chem. Soc. Rev. 2011,
40, 2003. (b) Kille, S.; Zilly, F. E.; Acevedo, J. P.; Reetz, M. T. Nature
Chem. 2011, 3, 738.
(8) (a) Groves, J. T.; Neumann, R. J. Am. Chem. Soc. 1989, 111, 2900.
(b) Cook, B. R.; Reinert, T. J.; Suslick, K. S. J. Am. Chem. Soc. 1986, 108,
7281. (c) Breslow, R.; Huang, Y.; Zhang, X.; Yang, Y. Proc. Natl. Acad.
Sci. U.S.A. 1997, 94, 11156.
(9) Das, S.; Incarvito, C. D.; Crabtree, R. H.; Brudvig, G. W. Science
2006, 312, 1941.
(10) Zhang, W.; Loebach, J. L.; Wilson, S. R.; Jacobsen, E. N. J. Am.
Chem. Soc. 1990, 112, 2801.
(11) Suzuki, K.; Oldenburg, P. D.; Que, L., Jr. Angew. Chem., Int. Ed.
2008, 47, 1887.
Models for 1 and 2 are supported by empirical data for
substrates incorporated into the original data sets. We next
sought to test the predictive power of these models for
(+)-nectaryl derivative 25, a synthetic terpene-like molecule
used in commercial fragrances that had not been included in the
data sets for either catalyst. Applying our site filter, many likely
sites of oxidation remained (C11, C10/12, C9/13, C8, C7 and
C3): the conformational flexibility of (+)-25 and electronic
similarity of its sites made selective oxidation with either catalyst
a challenging prospect. Applying the catalyst reactivity models,
oxidation of (+)-25 was predicted to modestly favor the more
electron rich, 3° C11 site (1.5:1) for 1 and the least sterically
encumbered C10/12 site (3:1) for 2 (Figure 6).15 Consistent
with these calculations, oxidizing (+)-25 with (S,S)-1 affords 29%
yield of C11 hydroxyl (+)-27 and 23% yield of the C10/12
ketones 26 with poor selectivity (1.3:1). In contrast, (S,S)-2 is
able to overcome the electronic substrate bias to furnish C10/12
oxidation products 26 in a 52% yield with good selectivity (6:1).
This example illustrates 2’s capacity to effect predictable control
on site-selectivity based on nonbonding interactions, even in
complex substrates with high degrees of conformational
flexibility. The site-selectivity models for 1 and 2 are validated
as predictive tools, particularly for substrates whose electronic,
steric, and stereoelectronic features are well represented by the
substrates incorporated into the original data sets.
We show that catalyst control of site-selectivity in aliphatic
C‑H oxidations is possibledespite the significant challenges
associated with controlling highly reactive intermediates
without needing a specific match between one catalyst and one
substrate. The development of quantitative structure-based
catalyst reactivity models will lead to more targeted application
of C-H oxidations at late stages of complex molecule synthesis20
and enable site-divergent diversification of bioactive molecules.
The discovery that site-selectivities of oxidation can be
mathematically correlated to substrate properties as a function
of the catalyst should inform and inspire future catalyst designs.
(12) Leroux, F. ChemBioChem 2004, 5, 644.
(13) For a catalyst with comparable methylene selectivities, but
diminished yields and mass balance relative to Fe(PDP): Prat, I.;
Gomez, L.; Canta, M.; Ribas, X.; Costas, M. Chem.Eur. J. 2013, 19,
1908.
(14) Winstein, S.; Holness, N. J. J. Am. Chem. Soc. 1955, 77, 5562.
(15) For complete details, see SI.
(16) (a) Molecular Operating Environment (MOE), 2011.10; Chemical
Computing Group Inc.: Montreal, 2011. (b) Gaussian 09, Revision B.01;
Gaussian, Inc.: Wallingford, CT, 2010.
(17) Harper, K. C.; Sigman, M. S. Science 2011, 333, 1875.
(18) Bigi, M. A.; Liu, P.; Zou, L.; Houk, K. N.; White, M. C. Synlett
2012, 23, 2768.
(19) Zhang, K.; Shafer, B. M.; Demars, M. D. M., II; Stern, H. A.; Fasan,
R. J. Am. Chem. Soc. 2012, 134, 18695.
(20) (a) Fraunhoffer, K. J.; Bachovchin, D. A.; White, M. C. Org. Lett.
2005, 7, 223. (b) Stang, E. M.; White, M. C. Nat. Chem. 2009, 1, 547.
(c) McMurray, L.; O’Hara, F.; Gaunt, M. J. Chem. Soc. Rev. 2011, 40,
1885.
D
dx.doi.org/10.1021/ja407388y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX