ORGANIC CHEMISTRY
Ligand-controlled divergent dehydrogenative
reactions of carboxylic acids via CÐH activation
Zhen Wang^1 †, Liang Hu^1 †, Nikita Chekshin^1 , Zhe Zhuang^1 , Shaoqun Qian^1 ,
Jennifer X. Qiao^2 , Jin-Quan Yu^1 *
Dehydrogenative transformations of alkyl chains to alkenes through methylene carbon-hydrogen (C–H)
activation remain a substantial challenge. We report two classes of pyridine-pyridone ligands that enable
divergent dehydrogenation reactions through palladium-catalyzedb-methylene C–H activation of
carboxylic acids, leading to the direct syntheses ofa,b-unsaturated carboxylic acids org-alkylidene
butenolides. The directed nature of this pair of reactions allows chemoselective dehydrogenation of
carboxylic acids in the presence of other enolizable functionalities such as ketones, providing
chemoselectivity that is not possible by means of existing carbonyl desaturation protocols. Product
inhibition is overcome through ligand-promoted preferential activation of C(sp^3 )–H bonds rather than
C(sp^2 )–H bonds or a sequence of dehydrogenation and vinyl C–H alkynylation. The dehydrogenation
reaction is compatible with molecular oxygen as the terminal oxidant.
D
ehydrogenation of aliphatic chains is an
important process in both bulk chemical
industry and fine chemical synthesis.
Various approaches have been developed
to meet the synthetic needs of desaturation
of carbonyl compounds ( 1 ). Exploration of
syn-eliminations of selenoxide and sulfoxide
intermediates has led to useful dehydrogenation
methods in synthesis ( 2 – 4 ). Other organic re-
agents, such asN-tert-butyl phenylsulfinimidoyl
chloride, hypervalent iodine, andN-oxoammonium
salts, have also been developed to prepare enones
( 5 – 7 ). Electrochemically driven desaturation of
carbonyl compounds was also recently achieved
( 8 ). An extensively researched catalytic pathway
for dehydrogenation is the formation of Pd(II)
enolates from ketones, with subsequentb-hydride
elimination affording enone products ( 9 – 13 ).
The Newhouse group has developed a method
to desaturate free carboxylic acids through pre-
formation of zinc enediolates, which are sub-
sequently oxidized toa,b-unsaturated acids by
Pd(II) complexes, with allyl acetate serving as
the stoichiometric oxidant (Fig. 1A) ( 14 ). How-
ever, synthetically usefula-branched carboxylic
acids were found to be challenging substrates
for the zinc enediolate method. Hence, it is
desirable to develop a complementary dehydro-
genation approach through methylene C–H
activation that can bypass the limitations of
enolate chemistry. However, the development of
synthetically useful dehydrogenation reactions
by means of C–H activation without installing
exogeneous directing groups faces two chal-
lenges: (i) the difficulty of activating methyl-
ene C–H bonds and (ii) product inhibition by,
or overreaction of, the olefin products (Fig. 1A).
In the past decade, we have focused on ligand-
enabled C–H activation reactions directed by
native functional groups such as free carboxylic
acids, free aliphatic amines, and native amides
( 15 ). Catalyticb-C–H activation of methyl C–H
bonds has led to the development of reactions
such as arylation, olefination, acetoxylation,
lactonization, and alkynylation. However,
methylene C–H activation reactions of acyclic
aliphatic substrates directed by innate func-
tionalities such as carboxylic acids remain a
substantial challenge (Fig. 1B) ( 16 – 18 ).
In 2006, Goldman, Brookhart, and colleagues
reported a landmark example of alkane de-
hydrogenation through methylene C–H acti-
vation ( 19 ). In their system, the challenge of
product inhibition was addressed by coupling
the dehydrogenation with a tandem metathe-
sis and hydrogenation reaction. Although this
tandem catalytic system provides a promising
strategy for converting low–molecular weight
alkanes to high–molecular weight alkanes, so
far it has proven difficult to use such a system
to develop dehydrogenative transformations
with synthetically relevant substrates as lim-
iting reagents for the synthesis of olefins. Re-
cently, dehydrogenation of cyclooctane by a
combination of cobalt and tungsten catalysts
via a radical pathway has also been observed,
albeit in modest yields ( 20 ). Our early efforts
to develop Pd(II)-catalyzed dehydrogenation
reactions with substrates as limiting reagents
through C–H activation involved the use of
oxazoline directing groups with aliphatic acids.
However, the synthetic utility of these trans-
formations is limited by the extra steps required
to install and remove the directing group ( 21 );
moreover, the reaction is limited to a single
cyclopentanecarboxylic acid substrate. We re-
port a pair of ligand-enabled Pd(II)-catalyzed
divergent dehydrogenation reactions. A wide
range of aliphatic carboxylic acids can be di-
rectly converted intoa,b-unsaturated acids org-alkylidene butenolides through activation
of either methylene or methyl C–H bonds.
Product inhibition was overcome by the de-
sign of ligands with different bite angles that
either prevented vinyl C–H activation of the
typically more reactivea,b-unsaturated acids
or enabled a tandem vinyl C–H activation and
alkynyl bromide coupling, leading to buteno-
lides that mask the carboxylic acid moiety from
further reactions (Fig. 1C). Computational stud-
ies of the Pd-catalyzed C–H activation step with
density functional theory (DFT) showed that
the ligand forming a five-membered chelate
with Pd is less competent in C(sp^2 )–H cleav-
age compared with the ligand that forms a
six-membered chelate, consistent with our
experimental findings.
Our efforts toward the development of de-
hydrogenation reactions began with the prepa-
ration of a series of bidentate pyridine-pyridone
ligands. On the basis of previous experimental
and theoretical studies, 2-pyridones have been
established as a powerful class of ligands for
promoting Pd(II)-catalyzed C–H activation
through ligand participation ( 22 – 24 ). Compu-
tational studies suggest that 2-pyridone is likely
toserveasanX-typeligandinamannersim-
ilar to that of NH(acetyl) moiety, acting as an
internal base directly involved in the C–H bond
cleavage step ( 23 ). Prompted by the success of
the acetyl-protected amino quinoline (APAQ)
class of ligands for methylene C–H activation
directed by the Wasa amide auxiliary ( 25 ), we
begantodesignaclassofbidentateligandsby
substitution of the NH(acetyl) moiety with
2-pyridone and investigated their reactivity
for methylene C–H activation directed by in-
nate functionalities. Bidentate pyridone lig-
andsL5toL7posited to coordinate to Pd
(II) as six-membered chelates were prepared
first since they were shown to be superior in
the APAQ ligands (table S1). With hexanoic
acid as the model substrate and 1,4-dioxane
as the solvent, we observed the formation of
2-hexenoic acid in 15% yield with ligandL5.
However, attempts at improving the reaction
yield using ligandsL5toL7under various
conditions proved unsuccessful. In light of
these results, we hypothesized that the low
yield of the reaction might have resulted from
inhibition of the catalyst by the olefin prod-
uct. Indeed, it was observed that the addition
of 15 mol % of the olefin product completely
halted the reaction. We surmised that a po-
tential pathway for product inhibition could
be the unproductive activation of theb-vinyl
C(sp^2 )–H bond of 2-hexenoic acid. Indeed, in
a hydrogen-deuterium exchange experiment,
deuterium incorporation was observed at the
vinyl position of the olefin product in the
presence of AcOH-d 4 (see supplementary ma-
terials for details). A related study indicated
that bidentate pyridone ligands forming five-
membered chelates with Pd(II) are not reactiveSCIENCEscience.org 3 DECEMBER 2021¥VOL 374 ISSUE 6572 1281
(^1) The Scripps Research Institute, 10550 N. Torrey Pines Road,
La Jolla, CA 92037, USA.^2 Discovery Chemistry, Bristol-Myers
Squibb, P.O. Box 4000, Princeton, NJ 08543, USA.
*Corresponding author. Email: [email protected]
These authors contributed equally to this work.
RESEARCH | REPORTS