Cross-dehydrogenative couplings between indoles and β-keto esters: Ligand-assisted ligand tautomerization and dehydrogenation via a proton-assisted electron transfer to Pd(II)

Article abstract of DOI:10.1021/ja501681y

Cross-dehydrogenative coupling reactions between β-ketoesters and electron-rich arenes, such as indoles, proceed with high regiochemical fidelity with a range of β-ketoesters and indoles. The mechanism of the reaction between a prototypical β-ketoester, ethyl 2-oxocyclopentanonecarboxylate, and N-methylindole has been studied experimentally by monitoring the temporal course of the reaction by ¹H NMR, kinetic isotope effect studies, and control experiments. DFT calculations have been carried out using a dispersion-corrected range-separated hybrid functional (?B97X-D) to explore the basic elementary steps of the catalytic cycle. The experimental results indicate that the reaction proceeds via two catalytic cycles. Cycle A, the dehydrogenation cycle, produces an enone intermediate. The dehydrogenation is assisted by N-methylindole, which acts as a ligand for Pd(II). The computational studies agree with this conclusion, and identify the turnover-limiting step of the dehydrogenation step, which involves a change in the coordination mode of the β-keto ester ligand from an O,O′-chelate to an α-C-bound Pd enolate. This ligand tautomerization event is assisted by the p-bound indole ligand. Subsequent scission of the β′-C-H bond takes place via a proton-assisted electron transfer mechanism, where Pd(II) acts as an electron sink and the trifluoroacetate ligand acts as a proton acceptor, to produce the Pd(0) complex of the enone intermediate. The coupling is completed in cycle B, where the enone is coupled with indole. Pd(TFA)₂ and TFA-catalyzed pathways were examined experimentally and computationally for this cycle, and both were found to be viable routes for the coupling step.

Full text of DOI:10.1021/ja501681y

Article
pubs.acs.org/JACS
Cross-Dehydrogenative Couplings between Indoles and β‑Keto
Esters: Ligand-Assisted Ligand Tautomerization and
Dehydrogenation via a Proton-Assisted Electron Transfer to Pd(II)
†
‡
†
†
,‡
†
́
Mikko V. Leskinen, Adam Madarasz, Kai-Tai Yip, Aini Vuorinen, Imre Papai,* Antti J. Neuvonen,
́
́
́
and Petri M. Pihko*,†
†Department of Chemistry and NanoScience Center, University of Jyvaskyla, FI-40014 Jyvaskyla, Finland
̈
̈
̈
̈
‡Research Center for Natural Sciences, Hungarian Academy of Sciences, Magyar Tudosok Korutja 2, H-1117, Budapest, Hungary
S
* Supporting Information
ABSTRACT: Cross-dehydrogenative coupling reactions be-
tween β-ketoesters and electron-rich arenes, such as indoles,
proceed with high regiochemical fidelity with a range of β-
ketoesters and indoles. The mechanism of the reaction between a
prototypical β-ketoester, ethyl 2-oxocyclopentanonecarboxylate,
and N-methylindole has been studied experimentally by
1
monitoring the temporal course of the reaction by H NMR,
kinetic isotope effect studies, and control experiments. DFT
calculations have been carried out using a dispersion-corrected
range-separated hybrid functional (ωB97X-D) to explore the
basic elementary steps of the catalytic cycle. The experimental
results indicate that the reaction proceeds via two catalytic cycles.
Cycle A, the dehydrogenation cycle, produces an enone intermediate. The dehydrogenation is assisted by N-methylindole, which
acts as a ligand for Pd(II). The computational studies agree with this conclusion, and identify the turnover-limiting step of the
dehydrogenation step, which involves a change in the coordination mode of the β-keto ester ligand from an O,O′-chelate to an α-
C-bound Pd enolate. This ligand tautomerization event is assisted by the π-bound indole ligand. Subsequent scission of the β′-
C−H bond takes place via a proton-assisted electron transfer mechanism, where Pd(II) acts as an electron sink and the
trifluoroacetate ligand acts as a proton acceptor, to produce the Pd(0) complex of the enone intermediate. The coupling is
completed in cycle B, where the enone is coupled with indole. Pd(TFA)2 and TFA-catalyzed pathways were examined
experimentally and computationally for this cycle, and both were found to be viable routes for the coupling step.
INTRODUCTION
Scheme 1. Development of Dehydrogenative β′−C(sp3)−H
■
C(sp2)−H Coupling Reaction
Dehydrogenative cross-couplings, or cross-dehydrogenative
couplings between two partners with C−H bonds, constitute
an attractive strategy in chemical synthesis.1 In particular, when
the reaction partners include sp3 C−H bonds, the reactions can
be used to generate molecular complexity in three dimensions,
and at the same time allow functionalization in remote
positions. Although there have been significant advances in
this field in recent years,2,3 dehydrogenative functionalization
reactions involving remote sp3 C−H groups are still rare.4 In
part, this might be due to the fact that the mechanisms of
dehydrogenative cross-couplings are only partially understood.
Herein, we present a full account on the mechanistic
investigation and the scope of the selective Pd(II)-catalyzed
dehydrogenative cross-coupling reaction between indoles and
β-keto esters.5 This reaction is an example of a cross-
dehydrogenative coupling between sp3 and sp2 C−H bonds.
Besides indoles, the reaction also accepts electron-rich
aromatics and phenols as the coupling partner6 and also allows
for a three-component coupling between arylboronates,
Received: February 21, 2014
Published: March 31, 2014
indoles, and β-keto esters (Scheme 1).7
© 2014 American Chemical Society
6453
dx.doi.org/10.1021/ja501681y | J. Am. Chem. Soc. 2014, 136, 6453−6462

Journal of the American Chemical Society
Article
RESULTS AND DISCUSSION
In our initial communication, we presented two possible
mechanistic scenarios for this reaction (Scheme 2). The first, a
■
Scheme 2. Palladium-Catalyzed Dehydrogenative β′-
Functionalization of β-Keto Ester with Indole and Originally
Proposed Reaction Mechanism
1
Figure 1. Monitoring of the temporal progress of the coupling by H
NMR spectroscopy. Reaction conditions: [1a]0 = 0.476 M, [2a]0 =
0.318 M, [tBuOOBz]0 = 0.413 M, 10 mol % Pd(TFA)2, 4:1 [D8]-
dioxane/AcOH, 300 K. The reported rates are averages of three
experiments. Rate0 and rate53 refer to the initial rate and the rate at t =
53 min, respectively.
“late indole”, scenario involves a Saegusa oxidation8 of 1a to
enone intermediate A followed by a Friedel−Crafts-type Pd-
catalyzed conjugate addition of indole 2a. The second, an “early
indole”, scenario starts with the well-established C3-palladation
of indole9 in which a C3-palladated indole species B is involved
in the dehydrogenation step, followed by reductive elimination.
Our early mechanistic investigations5 could not distinguish
between these two mechanistic possibilities. The key initial
observations were: (1) Isolated enone 4a also afforded the
coupling product with indole 2a, at a rate that was comparable
to the overall reaction rate, and (2) without indole 2a, only very
slow formation of enone 4a was observed. These observations
suggested that if enone 4a was an intermediate, its formation
might be dependent on the assistance of indole.
Using separately prepared enone 4a, the reaction between
enone 4a and indole 2a was investigated under two sets of
conditions. The initial concentration of 4a was set to 0.12 M,
close to the peak concentration of 4a obtained under the
standard oxidative coupling conditions (Figure 1). With
Pd(TFA)2 as the catalyst, the reaction between 2a and 4a
(Scheme 3) progresses at a rate comparable to the peak rate
Scheme 3. Reactions of Indole 2a and Enone 4a
In our early studies, the progress of the reaction was
monitored by withdrawing aliquots from the reaction mixture.5
In this work, we envisioned that the use of online NMR
methods to monitor the temporal progress of the reaction
would be most beneficial to reveal any fleeting intermediates
and to allow the simultaneous monitoring of several species,
including the oxidant.10
Kinetic Studies. Initially, the standard reaction of β-keto
1
ester 1a with 1-methylindole (2a) was monitored by H NMR
spectroscopy (Figure 1). The results show that [4a] builds up
and decays during the initial stage of the reaction, and product
formation ([3a]) follows a sigmoidal curve. The consumption
of indole 2a also plots a reverse sigmoidal curve. These results
strongly suggested that enone 4a is an intermediate, and
indeed, the rate of formation of 3a peaks close to the
concentration peak of 4a.11 The sigmoidal shape of the curve
for [4a] is characteristic of a delay caused by the buildup of the
intermediate in a consecutive reaction. The initial rate for the
consumption of β-keto ester 1a (−5.7 mM min−1) is also close
to the initial rate of the consumption of the oxidizer tBuOOBz
(−5.1 mM min−1).
obtained under the standard conditions (1.7 mM min−1 for
both cases). In contrast, TFA alone as the catalyst allowed the
reaction to proceed, but at a significantly slower rate (1.7 mM
min−1 with Pd(TFA)2 vs 0.26 min−1 with TFA, see Scheme 3).
These results indicated that Pd(II) also plays a role in the
second coupling step, with a possible acid-catalyzed background
reaction.
Interestingly, although TFA alone was an inefficient catalyst,
the effect of additional TFA to the overall reaction rate was
beneficial (Table 1). With added TFA, the reaction proceeded
at a reasonable rate even when using Pd(OAc)2 as the Pd(II)
source.
6454
dx.doi.org/10.1021/ja501681y | J. Am. Chem. Soc. 2014, 136, 6453−6462

Journal of the American Chemical Society
Article
a
Table 1. Effect of TFA on the Reaction Rate
b
cd
rate 4a (mM min−1
)
,
entry
TFA
rate 3a (mM min−1
)
1
2
3
4
0 mol %
0.1
1.7
2.3
2.6
0.1
2.9
3.6
3.3
10 mol %
20 mol %
40 mol %
a
The rates were obtained by monitoring the temporal progress of the
coupling by 1H NMR spectroscopy. Reaction conditions: 0−40 mol %
TFA, [1a]0 = 0.476 M, [2a]0 = 0.318 M, [tBuOOBz]0 = 0.413 M, 10
b
mol % Pd(OAc)2, 4:1 [D8]-Dioxane/AcOH, 300 K. Maximum rate
c
d
Initial rate. In control experiments without Pd(OAc)2 or with
Co(OAc)2 (20 mol %) or Fe(OAc)3 (20 mol %), no 4a or 3a was
produced.
Kinetic Isotopic Effects and Deuterium Labeling.
Although the above experiments established that enone 4a is
indeed a viable intermediate for the reaction, control
experiments without indole 2a clearly demonstrated that
enone formation is very slow in the absence of indole.5 The
reaction progress method allowed us to obtain kinetic isotope
effect (KIE) data for all key reaction components using
deuterium-labeled starting materials.12
The first KIE experiment involved a comparison of C2-H vs
C2-D-labeled indole (Figure 2a).13 All reaction components
displayed significant inverse KIEs in this experiment (Figure 2).
These results indicate that D-2a accelerates the formation of
the enone 4a. The increased rate of product formation (3a)
might be due to increased rate of the formation of the
intermediate.
A second set of KIE experiments were conducted with
deuterium-labeled β-keto ester D6-1a (Figure 2b). In this case,
unfortunately, the formation of the corresponding enone could
not be reliably monitored. Although the rate of consumption of
D6-1a appears to display an inverse KIE, possible initial H/D
exchange or differences in the rates of the formation of the
Pd(II) complexes of 1a could also account for this observation.
Indeed, the overall rate of the reaction did not exhibit any KIE
(3a: kH/kD = 0.98), and the initial rate of the consumption of
the oxidant indicated a small normal KIE (kH/kD = 1.07).
Because the consumption of the oxidant is most likely
correlated with the concentration of the enone 4a, these
results suggest that the dehydrogenative reaction that produces
4a is unlikely to exhibit a KIE, although this step must involve
the breaking of the β′-H bond of 1a.
Competition Studies. To obtain further insight into the
reaction mechanism, the KIEs were also assayed via
competition studies.12 As shown in Scheme 4, an intermo-
lecular competition reaction between substrates 2a and D-2a,
starting with 1a, gives a product distribution PH/PD 1:1.22. If
2a is not dissociated from Pd after the first stage of the reaction
(formation of enone 4a), then the observed KIE could be
explained by the more rapid rate of enone formation with D-2a.
However, the fact that a significant concentration of free enone
can be observed during the reaction (Figure 1) suggests that
the catalytic cycle responsible for the coupling of enone 4a and
Figure 2. Temporal progress of the coupling by 1H NMR
spectroscopy with deuterated starting materials. Reaction conditions:
(a) [1a]0 = 0.476 M, [D-2a]0 = 0.318 M, (b) [D6-1a]0 = 0.476 M,
[2a]0 = 0.318 M. For both experiments: [tBuOOBz]0 = 0.413 M, 10
mol % Pd(TFA)2, 4:1 [D8]-Dioxane/AcOH, 300 K. Rates and kH/kD
are averages of three experiments (a) or two experiments (b). Rate0
and ratexx refer to the initial rate and the rate at t = xx min,
respectively.
6455
dx.doi.org/10.1021/ja501681y | J. Am. Chem. Soc. 2014, 136, 6453−6462

Journal of the American Chemical Society
Article
parallel experiment (Figure 2b), suggesting that D6-1a is
dehydrogenated at a slower rate than 1a. Interestingly, some
H/D exchange appears to take place in this experiment,
producing D4-3a (Scheme 5).
Scheme 4. Competition Studies between 2a and D-2a
Finally, in an intramolecular competition experiment, the
mono-β′-deuterated D1-1a gave rise to product 3a that
exhibited a 48.6:51.4 H/D ratio (Scheme 6). The absence of
a KIE in this experiment suggests that the β′-H bond cleavage is
not turnover-limiting for the dehydrogenation cycle (see below
for further discussion).
Scheme 6. Intramolecular Competition Studies with β′-
Monodeuterated D1-1a
The effect of the electron-withdrawing alkyl ester was studied
using β-keto ester 1b. Under the standard conditions, the
reaction between 1b and 2a was significantly slower than the
standard reaction between 1a and 2a (0.52 mM min−1 with 1b
vs 1.7 mM min−1 with 1a). This rate difference was also
confirmed by an intermolecular competition between 1a and
1b (P3a/P3b value of 3, see Scheme 7). These rates reflect the
measured rates of the dehydrogenation step (rate of formation
of enone: 1.6 mM min−1 for the enone derived from 1b, vs 4.4
mM min−1 for 4a).14
indole 2a is at least partially separated from the first
dehydrogenation cycle that produces enone 4a. Therefore,
the observed KIE in Scheme 4a could be related to the C−C
formation step. To assay this possibility, intermolecular
competition experiments with 2a and D-2a were conducted
using enone 4a as the substrate. These experiments gave a PH/
PD that was much closer to unity under both Pd(II) catalysis
and TFA catalysis (Scheme 4b), suggesting that the KIE
observed with 1a (Scheme 4a) originates from the dehydrogen-
ation step and that 4a may not be fully dissociated from the Pd
complex that eventually leads to the product 3a.
Scheme 7. Intermolecular Competition between 1a and 1b
in the Coupling Process
In the competition reaction between 1a and D6-1a, the
nondeuterated 1a reacts faster than the deuterated D6-1a
(Scheme 5). This result is in agreement with the small normal
KIE observed for oxidant consumption with D6-1a in the
Scheme 5. Intermolecular Competition between D6-1a and
1a in the Coupling Process
In summary, these kinetic experiments revealed that (1) the
reaction likely proceeds in two stages, via an enone
intermediate (4a), (2) the formation of enone is dependent
on indole, and (3) the C−C bond formation step can proceed
under both acid catalysis as well as under Pd(II) catalysis. In
parallel with these experiments, the intimate details of the
mechanism were subjected to a computational study.
Computational Studies and Revision of the Mecha-
nism. Based on our experimental findings, we envisioned that
the Pd-catalyzed dehydrogenative cross-coupling reaction
between β-ketoester 1a and indole 2a takes place via two
distinct catalytic cycles (Scheme 8) corresponding to the
formation of the enone intermediate (cycle A) followed by the
6456
dx.doi.org/10.1021/ja501681y | J. Am. Chem. Soc. 2014, 136, 6453−6462

Journal of the American Chemical Society
Article
reference level (at −3.4 kcal/mol), and it is characterized by
chelating coordination of the enolate ligand through the two
carbonyl moieties (η2(O,O′) complex).17
Scheme 8. Schematic View of Catalytic Cycles for the
Dehydrogenation and the C−C Bond Formation Steps
For the β-hydride elimination step, the β′-hydrogen should
be accessible by the metal center, therefore, ligand rearrange-
ment is expected as a next step along the reaction pathway. The
simplest transformation would be an internal rearrangement of
the bonding between Pd and enolate preserving the same
stoichiometry, but other pathways (including various dissoci-
ation/association steps) are also feasible. Our attempts to
explore these transformations pointed to several transition
states lying higher in free energy than TSdepr. For instance, the
transition state connecting int1 with an η3-Pd-enolate
intermediate (i.e., displacement of the ester group from the
coordination sphere, see TSrear and int2 in Figure 4) is
predicted to be at 16.9 kcal/mol, giving rise to a barrier of 20.3
kcal/mol (relative to int1).18
C−C bond formation process with indole (cycle B). As
indicated in Scheme 8, the Pd(TFA)2(2a) species may
represent a common intermediate of the two cycles. We
carried out DFT calculations with the main aim at identifying
and characterizing the key elementary steps of these cycles.15
Dehydrogenation: Cycle A. In accordance with previous
studies on Pd-catalyzed dehydrogenation reactions, we
considered a sequence of C−H-activation/β-hydride elimina-
tion steps in cycle A. The assistance of indole was clearly
demonstrated in our experiments; therefore, we assumed that
2a acts as a coligand along the entire reaction pathway.
Figure 4. Decoordination of the ester group in intermediate int1: (a)
transition state (b) product state of rearrangement.
The calculations indicate that the deprotonation of 1a at the
α-carbon atom can occur easily via a tetracoordinate Pd(II)
complex involving two TFA ligands and both substrates bound
to the metal center (intermediate Pd(TFA)2(1a)(2a)). This
complex is predicted to be at −0.1 kcal/mol with respect to
Pd(TFA)2(2a) + 1a.16 The transition state identified for the
deprotonation is depicted in Figure 3a and it represents only a
small activation barrier (9.9 kcal/mol). It is apparent that the
C−H bond of 1a is activated by a neighboring TFA ligand
resulting in a Pd-enolate intermediate, which is stabilized by the
dissociation of the TFAH molecule. This latter reaction
intermediate (int1 in Figure 3b) lies slightly below the
Additional ligand rearrangement can lead to an intermediate
involved directly in the β-hydride elimination step (see int3 in
Figure 5). In this species, the enolate is bound covalently to Pd
via the α-carbon atom (η1(C) complex) and it displays
characteristic β-agostic interactions with the metal center.
Figure 3. Transition state located for the substrate deprotonation step
of cycle A (a) and the corresponding product state intermediate (b).
Relative Gibbs free energies (in kcal/mol, with respect to Pd-
(TFA)2(2a) + 1a) are shown in parentheses. Metal−ligand bonds are
indicated by dashed lines. For clarity of figures, hydrogen atoms are
omitted, except those involved in dehydrogenation.
Figure 5. The proton-coupled electron transfer (PCET) transition
state of cycle A and the corresponding reactant and product state
intermediates.
6457
dx.doi.org/10.1021/ja501681y | J. Am. Chem. Soc. 2014, 136, 6453−6462

Journal of the American Chemical Society
Article
This complex is computed to be at +3.2 kcal/mol on the free
energy scale. Surprisingly, the Pd-mediated C−H bond cleavage
does not yield the expected palladium-hydride species because
the located transition state (TSPCET in Figure 5) describes a
direct hydrogen migration to the free oxygen of the TFA ligand
without the formation of a palladium-hydride intermediate
(PdH species could only be identified computationally as very
unstable structures).11
The population analysis carried out for the transition state
and the corresponding intermediates reveals that this
elementary step can be characterized as a concerted proton-
coupled electron transfer (PCET) process involving proton
migration to TFA occurring in concert with 2e− electron
transfer to the metal center.19,20 This reaction step results in a
Pd(0)(TFAH)(2a)(4a) complex as an intermediate (int4 in
Figure 5) lying at 6.4 kcal/mol. The computed activation
barrier of the PCET step is 16.9 kcal/mol with respect to the
low-lying int1 intermediate, that is, much lower than that of the
ligand rearrangement step (20.3 kcal/mol, TSrear with respect
to int1). After this step, the catalytic cycle involves the oxidation
of Pd(0) and the elimination of the enone molecule. These
transformations were not examined computationally in the
present work.
To rationalize the results of KIE experiments with 2a, we
calculated the KIE values based on the relative barriers with the
different isotopomers of 2a. With D-2a, the calculated KIE is
0.85 considering that TSrear is the turnover-determining
transition state.12,21 The calculated value is very close to the
experimental KIE (0.77−0.85, see Figure 2). The magnitude of
the KIE is similar to cases where hybridization changes from
C(sp2) to C(sp3),22 suggesting that the steric environment of
C2 of 2a becomes significantly more crowded in the turnover-
determining transition state TSrear compared to the turnover-
determining intermediate. This result suggests that 2a assists
the tautomerization step by coordinating more tightly
(primarily via its 2,3-π-bond) to Pd(II) in the transition state
TSrear. Indeed, computations indicate that the bond distance
between the C2 atom of the coordinated indole and the Pd
atom is considerably shortened in the TSrear compared to the
related reactant state (int1) (the computed Pd−C2 bond
lengths are 2.28 and 2.34 Å, respectively). The need for indole
assistance in this step may result from increased electron
deficiency of the Pd(II) center in TSrear due to decoordination
of 1a.
To test this hypothesis, we also experimentally explored
other electron-rich ligands which would not react with 4a but
would nevertheless be able to withstand the oxidative
conditions. In addition to sulfoxides (DMSO and PhSOMe),23
1,2,3-trimethylindole and 1,3-dimethylindole were able to
significantly accelerate the dehydrogenation step (see the
Supporting Information).11
The Gibbs free energy diagram of the reaction route explored
for the dehydrogenation process is depicted in Figure 6. These
If the PCET step, where the β′-H bond is cleaved, would be
the turnover-determining step in cycle A, a much larger KIE
would be expected than that observed experimentally with β′-
deuterated 1a variants (the experimental KIE is close to 1, see
Figure 3 and Scheme 6). The calculated KIE for the PCET step
is quite large (7.06), and it is in sharp contrast with the value
obtained from the intramolecular competition experiment
(Scheme 6), which is expected to be most sensitive to any
KIE in the product-determining step. The fact that no primary
KIE is observed even under these conditions can be rationalized
by considering that the different hydrogen isotopes are not in
an equal environment after the turnover-limiting step (i.e.,
TSrear).24 The choice between β′-H vs β′-D abstraction has
already been made in the turnover-determining ligand
rearrangement step which leads to the formation of C-bound
enolate int3, The effect of deuterium substitution in the β′
position of the ligand on the ligand rearrangement step is
expected to be small, resulting in a negligible primary KIE.
The finding that the turnover is determined by the ligand
tautomerization step (from a O,O′-chelate int1 to C-bound
int3) is interesting. α-Substituted β-dicarbonyl compounds are
known to be unproductive in α-arylation reactions,25 and our
results suggest that such low reactivity might have a kinetic
origin. Specifically, if the barrier for the formation of the C-
bound enolate from the O,O′-chelate intermediate is too high,
this would prevent both dehydrogenation and α-arylation
reactions. In a control experiment, methyl dimethylmalonate
(5), a very sluggish substrate for α-arylation,25a failed to give
any coupling products with 2a under the standard reaction
conditions (eq 1).11
Figure 6. Gibbs free energy diagram computed for the dehydrogen-
ation process.
results point toward a reasonable mechanism for the
dehydrogenation process, however, they indicate that the
η2(O,O′) to η1(C) rearrangement of the enolate ligand, and not
the PCET step, might be rate-determining in cycle A.
Indole-Assisted Dehydrogenation. To assess the role of
indole in the dehydrogenation cycle, we examined analogous
reaction pathways using a model with a TFAH coligand
replacing the indole. We find that the activation barriers are
notably higher than those presented above. The largest
difference was obtained for the ester decoordination step (the
barrier increased to 24.2 kcal/mol), indicating that indole
coordination is clearly beneficial in terms of the reaction rate.
Interestingly, the PCET mechanism of C−H bond cleavage is
maintained in the absence of indole as well, although the barrier
of this step is predicted to be slightly higher in this case (15.0
kcal/mol).
This result could be rationalized by the higher Lewis basicity
of the ester oxygens in 5, which might result in a O,O′-chelated
intermediate that would be too stable to undergo the
tautomerization to the C-bound enolate. Because ligands such
as indole are able to assist the tautomerization step, we can
6458
dx.doi.org/10.1021/ja501681y | J. Am. Chem. Soc. 2014, 136, 6453−6462

Journal of the American Chemical Society
Article
Protonation of the carbonyl oxygen followed by the
dissociation of the enolic form of the product is found to be
a reasonable scenario for the completion of the cycle, but other
product elimination routes may exist too.
It should be emphasized that the alternative C−C bond
formation mechanisms cannot be entirely ruled out either. For
example, for a Brønsted acid catalyzed pathway (TFA as the
catalyst), calculations predict a barrier very similar to that
obtained for the C-palladation pathway (19.9 kcal/mol).
Furthermore, a Lewis acid catalyzed pathway, involving Pd2+,
was found to have a slightly higher barrier (22.3 kcal/mol).11
Although there are close literature precedents involving Pd2+ as
a Lewis acid in Friedel−Crafts reactions between indoles and
α,β-unsaturated carbonyl compounds,27 it should be noted that
the reaction conditions (aprotic solvents, such as CH2Cl2 and
speculate that perhaps more efficient π-donor ligands might
overcome these limitations.
The C−C Coupling Process. According to the exper-
imental evidence, the C−C coupling process (cycle B) could
proceed either via acid catalysis or Pd(II) catalysis under acidic
conditions. The reaction between 2a and 4a could conceivably
take place via either Lewis acid catalysis (Pd2+), C-palladation of
2a, or Brønsted acid catalysis, among other possibilities. We
therefore examined these reaction routes computationally, and
plausible pathways were identified for these scenarios. For
brevity, herein we discuss only the C-palladation mechanism in
detail, which is predicted to have the lowest activation barrier
among the investigated pathways. The alternative pathways are
presented in the Supporting Information.11
The C-palladation pathway begins with the formation of a
Pd(TFA)2(2a)(4a) complex, which involves both coupling
partners (indole 2a and enone 4a). Although this complex lies
fairly high in free energy (11.8 kcal/mol with respect to
Pd(TFA)2(2a) + 4a), it can undergo C−H activation
(concerted metalation-deprotonation)26 via transition state
TSCH to yield intermediate intCH (Figure 7). The located
−
noncoordinating counterions, such as SbF6 ) are quite different
to those studied here (trifluoroacetate counterions, polar/protic
solvent).
Resting State of the Catalyst. Although the computations
give reasonable barriers for the C−C bond forming step
starting from Pd(TFA)2, the stability of the product complex
intCC suggests that recycling Pd(II) requires further assistance,
for example, from the acidic solvent or β-ketoesters. We were
able to crystallographically characterize a dinuclear
[Pd2(TFA)2(3a)2] complex from a reaction conducted with
100 mol % of Pd(TFA)2.11,28 In this complex, 3a is a bidentate
ligand for Pd(II), with a binding geometry that is strikingly
similar to intCC. This complex could also be characterized
computationally, and the experimentally and computationally
derived structures are presented in Figure 9. The isolability of
this complex suggests that Pd(II)-3a complexes are stable
intermediates, and their decomplexation might even limit the
turnover of the reaction.
Figure 7. C−H activation of indole 1a.
transition state is computed to be at 19.7 kcal/mol, and the
resulting intermediate is predicted to be at 8.9 kcal/mol. In
intCH, the indole is covalently bound to Pd, whereas enone 4a is
π-coordinated.
Facile C−C bond formation may take place from intCH
(Figure 8). The located transition state (TSCC) lies notably
lower in free energy than TSCH. The product state of the
coupling process (intCC) is a very stable species, wherein the
adduct 3a is bound by multiple bonds to the metal center.
Figure 9. Structure of the Pd2(TFA)2(3a)2] complex: (a) X-ray
structure, (b) overlay of X-ray (blue) and computed (red) structures.
As a test for this hypothesis, we found that addition of
dimethyl methylmalonate 5 (50 mol %) to the reaction mixture
results in a marked increase in the rate of formation of both 3a
and 4a (Scheme 9), pointing toward a possible assistance of 5
in releasing Pd(II).29 Alternatively, 5 could act as a ligand for
dehydrogenation, but we find this scenario less likely as β-
ketoester 1a alone cannot effectively promote the dehydrogen-
ation without the assistance from indole.5 We cannot, however,
rule out a third possibility that 5 assists in cycle B as a coligand.
Reaction Scope and Asymmetric Variants of the
Coupling Reaction. Although clarifying the mechanistic
picture of the coupling reaction was the main focus of this
study, we also present here the full scope of the transformation
and an asymmetric variant of the reaction. As described in our
original communication, the reaction readily tolerated electron-
rich, electron-poor, and sterically demanding indole-substrates,
and both free indole N−H as well as N-methyl and N-benzyl
Figure 8. Pd-catalyzed C−C coupling via a C3-palladated indole
species.
6459
dx.doi.org/10.1021/ja501681y | J. Am. Chem. Soc. 2014, 136, 6453−6462

Journal of the American Chemical Society
Article
Scheme 9. Effect of Malonate Ester 5
Scheme 11. Scope of the Diastereoselective Dehydrogenative
Coupling with 8-Phenylmenthyl β-Keto Esters
a
indoles are tolerated. However, sulfonyl or carbamate
protecting groups are not tolerated on the indole nitrogen.
Furthermore, a range of β-keto esters, including cyclic 5-, 6-,
and 7-membered β-keto esters, can be used. Additional
substrates that were not described in the initial communication
are shown in Scheme 10. The full scope is presented as a Chart
in the Supporting Information.
Scheme 10. Additional Substrates for the Indole−β-Keto
a
Ester Oxidative Coupling
a
Isolated yields of pure products are reported. Conditions, unless
otherwise indicated: 1 (1.5 equiv), 2 (0.4 mmol, 1.0 equiv), t-BuOOBz
(1.3 equiv), Pd(TFA)2 (0.1 equiv), i-PrOH/AcOH (4:1, 0.5 mL) at rt.
a
Isolated yields of pure products are reported. Conditions: 1 (1.5
equiv), 2 (0.4 mmol, 1.0 equiv), t-BuOOBz (1.3 equiv), Pd(TFA)2
(0.1 equiv), i-PrOH/AcOH (4:1, 0.5 mL) at 25 °C.
conditions with Pd(II) catalysis. With the combined
information obtained from online NMR monitoring experi-
ments, kinetic isotope effects, and computational studies, the
previously proposed reaction mechanism was revised. The
revised mechanism is presented in Scheme 12.
For the development of an asymmetric version of the
reaction, we have focused on using a chiral ester auxiliary.
Although the use of chiral acids or chiral anions might
conceivably induce enantioselectivity via either the Pd(II)- or
acid-catalyzed C−C bond formation pathways, our previous
experiments with chiral acids and chiral Pd phosphates were
not very encouraging.6 Instead, 8-phenylmenthyl esters30
exhibited useful levels of diastereoselectivity.31,32
Scheme 11 presents the scope of the transformation with
different 8-phenylmenthyl β-keto esters. In general, the
diastereoselectivities were good. The exception was 3i, which
afforded only moderate 2.1:1 dr. The reaction also tolerates
bromine substituents in the indole nucleus (3g−3i); this is a
useful feature for further functionalization of the products. The
observed sense of diastereoselection is in full accordance with
literature precedents30 and the model shown in Scheme 11
(inset).
The reaction involves indole already at the early stage of the
catalytic process as a π-bound ligand for Pd(II) that assists the
O−to−C tautomerization of β-keto ester 1a. This is the
turnover-determining step of the dehydrogenation cycle. The
assistance of indole ligand in the tautomerization step is evident
both from the secondary kinetic isotope effects observed for the
rate of the dehydrogenation with 2-deuterated N-methylindole
and from the computational studies. The dehydrogenation of
the β-keto ester is completed by a proton-assisted electron
transfer reaction where Pd(II) is simultaneously reduced to
Pd(0) and trifluoroacetate ligand accepts a proton from the β′
carbon. No Pd hydride intermediate could be characterized by
the computations. For the C−C bond forming step, three
plausible pathways involving either acid catalysis or Pd(II)
catalysis were identified by computations. Experimentally,
Pd(II) was found to accelerate the C−C bond formation, and
computationally the most feasible Pd(II)-catalyzed pathway
involves the palladation of indole at C3. However, Lewis acid
catalysis by Pd2+ cannot be excluded. Finally, the synthetic
utility of the protocol was expanded to include additional
Krapcho decarboxylation of 3aj provided the corresponding
ketone 6 in 62% yield (eq 1) and 98:2 er (eq 2).11
CONCLUSIONS
β-Keto esters and indoles can be dehydrogenatively cross-
coupled with a high regiochemical fidelity under very mild
■
6460
dx.doi.org/10.1021/ja501681y | J. Am. Chem. Soc. 2014, 136, 6453−6462

Journal of the American Chemical Society
Article
Scheme 12. Revised Catalytic Cycles
(3) For selected reviews of catalytic oxidative functionalizations of sp3
C−H bonds, see: (a) Jazzar, R.; Hitce, J.; Renaudat, A.; Sofack-
Kreutzer, J.; Baudoin, O. Chem.Eur. J. 2010, 16, 2654. (b) Li, H.; Li,
B.-J.; Shi, Z.-J. Catal. Sci. Technol. 2011, 1, 191. (c) Rouquet, G.;
Chatani, N. Angew. Chem., Int. Ed. 2013, 52, 2. (d) Girard, S. A.;
Knauber, T.; Li, C.-J. Angew. Chem., Int. Ed. 2014, 53, 74.
substrates, and an asymmetric version of the reaction could be
realized with 8-phenylmenthyl esters.
The reaction between indoles and β-ketoesters appears to be
possible only because indoles can serve the double role of a
substrate and a ligand in the ligand-assisted tautomerization
step, the turnover-determining step of the dehydrogenation
cycle. This finding should encourage researchers to look for
similar effects in other dehydrogenation and cross-dehydrogen-
ative coupling reactions.
(4) For intramolecular PdII-catalyzed dehydrogenative arylations of
sp3 C−H bonds, see: (a) Lieg
̀
ault, B.; Fagnou, K. Organometallics
2008, 27, 4841. (b) Pierre, C.; Baudoin, O. Tetrahedron 2013, 69,
4473.
(5) Leskinen, M. V.; Yip, K.-T.; Valkonen, A.; Pihko, P. M. J. Am.
Chem. Soc. 2012, 134, 5750.
ASSOCIATED CONTENT
■
(6) Yip, K.-T.; Nimje, R. Y.; Leskinen, M. V.; Pihko, P. M. Chem.
Eur. J. 2012, 18, 12590.
(7) Nimje, R. Y.; Leskinen, M. V.; Pihko, P. M. Angew. Chem., Int. Ed.
S
* Supporting Information
Experimental procedures, additional experiments pertaining to
the mechanism, characterization data, computational details,
and copies of NMR spectra and GC chromatograms.This
material is available free of charge via the Internet at http://
pubs.acs.org.
2013, 52, 4818.
(8) Ito, Y.; Suginome, M. In Handbook of Organopalladium Chemistry
for Organic Synthesis; Negishi, E.-I., Ed.; Wiley: New York, 2002; Vol.
2, p 2873.
(9) (a) Itahara, T.; Ikeda, M.; Sakakibara, T. J. Chem. Soc., Perkin
Trans. 1 1983, 1361. (b) Itahara, T.; Kawasaki, K.; Ouseto, F. Synthesis
1984, 236. (c) Yokoyama, Y.; Matsumoto, T.; Murakami, Y. J. Org.
Chem. 1995, 60, 1486. (d) Jia, C.; Lu, W.; Kitamura, T.; Fujiwara, Y.
Org. Lett. 1999, 1, 2097. (e) Grimster, N. P.; Gauntlett, C.; Godfrey, C.
R. A.; Gaunt, M. J. Angew. Chem., Int. Ed. 2005, 44, 3125. (f) Maehara,
A.; Tsurugi, H.; Satoh, T.; Miura, M. Org. Lett. 2008, 10, 1159.
(10) For examples of online NMR monitoring in detection of several
AUTHOR INFORMATION
Corresponding Author
papai.imre@ttk.mta.hu, petri.pihko@jyu.fi.
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
intermediates, see (a) Ref 7. (b) Sahoo, G.; Rahaman, H.; Madaras
́
z,
■
́
A.; Pap
́
ai, I.; Melarto, M.; Valkonen, A.; Pihko, P. M. Angew. Chem., Int.
We thank Dr. Elina Kalenius, Mr. Esa Haapaniemi, and
Academy Prof. Kari Rissanen for assistance with mass
spectrometry, NMR spectroscopy, and X-ray crystallography,
respectively. Financial support from Tekes, Academy of Finland
(project 138854), Hungarian Scientific Research Fund (OTKA,
grant K-81927), AB Enzymes, CABB, Fermion, Hormos,
Ed. 2012, 51, 13144.
(11) See the Supporting Information for details.
́
(12) (a) Gomez-Callego, M.; Sierra, M. A. Chem. Rev. 2011, 111,
4857. For an insightful essay, see (b) Simmons, E. M.; Hartwig, J. F.
Angew. Chem., Int. Ed. 2012, 51, 3066.
(13) C3-deuterated 2a rapidly exchanged the deuterium label with
AcOH, with and without Pd(TFA)2 catalyst. In contrast, the
deuterium label at C2 was preserved with D-2a under the reaction
conditions.
(14) Tanaka, D.; Romeril, S. P.; Myers, A. G. J. Am. Chem. Soc. 2005,
127, 10323.
(15) Most of the DFT calculations (geometry optimizations,
vibrational analysis, estimation of solvent effects) were carried out at
ωB97X-D/SDDP level of theory. For each located structure, we
carried out additional single-point energy calculations using the same
Orion, University of Jyvaskyla, and COST CM0905 is gratefully
̈
̈
acknowledged.
REFERENCES
■
(1) For review of catalytic dehydrogenative cross-couplings, see:
Yeung, C. S.; Dong, V. M. Chem. Rev. 2011, 111, 1215.
(2) For unselective examples of dehydrogenative arylations of sp3 C−
H bonds, see: (a) Deng, G.; Zhao, L.; Li, C.-J. Angew. Chem., Int. Ed.
2008, 47, 6278. (b) Guo, X.; Li, C.-J. Org. Lett. 2011, 13, 4977.
6461
dx.doi.org/10.1021/ja501681y | J. Am. Chem. Soc. 2014, 136, 6453−6462

Journal of the American Chemical Society
Article
functional but a larger basis set (supplemented by diffusion functions).
The reported energetics refers to relative solution-phase Gibbs free
energies. For further details, see Supporting Information.
(16) The Pd(TFA)2(2a) + 1a state was arbitrarily chosen as a
reference level for the estimation of relative Gibbs free energies
because the experimental evidence indicated involvement of 2a in the
dehydrogenation of 1a.
(17) For an early review on coordination chemistry of β-dicarbonyl
compounds, see Kawaguchi, S. Coord. Chem. Rev. 1986, 70, 51. For
experimental studies of Pd(II) enolates, see: (b) Culkin, D. A.;
Hartwig, J. F. Organometallics 2004, 23, 3398. (c) Wolkowski, J. P.;
Hartwig, J. F. Angew. Chem., Int. Ed. 2002, 41, 4289.
dinuclear Pd(II) species, see Sanhueza, I. A.; Wagner, A. M.; Sanford,
M. S.; Schoenebeck, F. Chem. Sci. 2013, 4, 2767.
(29) In addition to 5, addition of DMPU also increases the overall
reaction rate (ratemax = 2.0 mM min−1 for 3a with 50 mol% of DMPU
vs 1.7 mM min−1 under the standard conditions).
(30) Corey, E. J.; Ensley, H. E. J. Am. Chem. Soc. 1975, 97, 6908.
(31) The diastereoselective version also tolerates cyclic 6-membered
β-keto esters, but the products could not be obtained in pure form.
(32) The absolute stereochemistry of the products was determined
by X-ray analysis of 3aj. Other products were assigned by analogy. See
the Supporting Information for details.
(18) The computed barrier is consistent with those reported for
analogous O-bound enolate to C-bound enolate tautomerization
processes of Ni- and Pd-enolate complexes, although the structure of
the enolate ligand is different: (a) Cam
́
pora, J.; Maya, C. M.; Palma, P.;
Carmona, E.; Gutierrez, E.; Ruiz, C.; Graiff, C.; Tiripicchio, A.
́
Chem.Eur. J. 2005, 11, 6889. (b) Oertel, A. M.; Ritleng, V.; Busiah,
A.; Veiros, L. F.; Chetcuti, M. J. Organometallics 2011, 30, 6495.
(19) For recent comprehensive reviews on proton-coupled electron
transfer reactions, see (a) Huynh, M. H. V.; Meyer, T. J. Chem. Rev.
2007, 107, 5004. (b) Weinberg, D. R.; Gagliardi, C. J.; Hull, J. F.;
Murphy, C. F.; Kent, C. A.; Westlake, B. C.; Paul, A.; Ess, D. H.;
McCafferty, D. G.; Meyer, T. J. Chem. Rev. 2012, 112, 4016.
(20) For studies describing metal-catalyzed C−H bond cleavage
reactions in terms of the PCET mechanism, see (a) Seu, C. S.; Appel,
A. M.; Doud, M. D.; DuBois, D. L.; Kubiak, C. P. Energy Environ. Sci.
2012, 5, 6480. (b) Nielsen, R. J.; Goddard, W. A., III J. Am. Chem. Soc.
2006, 128, 9651. Note that in the latter work, this mechanism is
referred to as reductive β-hydride elimination. The computed charges
of the int3, TSPCET and int4 stationary points indicate that a proton
and not a hydridic or radical H is transferred in the present hydrogen
migration step, and therefore, we think the use of the PCET term is
justified here (for details of population analysis, see the Supporting
Information).
(21) The identity of the catalyst resting state (i.e., the turnover-
determining intermediatesee Kozuch, S.; Shaik, S. Acc. Chem. Res.
2011, 44, 101) is uncertain. In the KIE calculations, it was assumed
that the resting state does not involve 2a as the ligand. For more
details about the KIE calculations, see the Supporting Information.
(22) Anslyn, E. V.; Dougherty, D. A. Modern Physical Organic
Chemistry; University Science Books: Herndon, VA, 2006; pp 435−
437. ISBN 1-891389-31-9.
(23) (a) Diao, T.; Stahl, S. S. J. Am. Chem. Soc. 2011, 133, 14566.
(b) Diao, T.; Pun, D.; Stahl, S. S. J. Am. Chem. Soc. 2013, 135, 8205.
(24) The system studied herein is different to the scenarios discussed
in ref 12b. In the fifth scenario of this paper, the absence of KIE in an
intramolecular competition experiment is attributed to the reversibility
of the C−H bond cleavage step because in this particular example, the
C−H and C−D bonds would otherwise be equally accessible in the
product-determining step. This is not the case in this mechanistic
scenario.
(25) For accounts describing the scope of α-arylation with α-
dicarbonyl compounds, see (a) Kawatsura, M.; Hartwig, J. F. J. Am.
Chem. Soc. 1999, 121, 1473. (b) Fox, J. M.; Huang, X.; Chieffi, A.;
Buchwald, S. L. J. Am. Chem. Soc. 2000, 122, 1360 For a discussion of
ligand effects on α-arylation of β-dicarbonyl compounds, see: ref 17c.
The authors propose that bulky ligands assist in the reductive
elimination step, but no mention is made of the ligand effects on the
tautomerization step.
(26) For a review on base-induced concerted metalation-deprotona-
tion mechanism, see (a) Ackermann, L. Chem. Rev. 2011, 111, 1314.
See also (b) Gorelsky, S. I.; Lapointe, D.; Fagnou, K. J. Am. Chem. Soc.
2008, 130, 10848. (c) Biswas, B.; Sugimoto, M.; Sakaki, S.
Organometallics 2000, 19, 3895.
(27) Aikawa, K.; Honda, K.; Mimura, S.; Mikami, K. Tetrahedron Lett.
2011, 52, 6682.
(28) We cannot rule out the involvement of dinuclear Pd(II) species
in the C-C bond forming process. For an example of palladation in a
6462
dx.doi.org/10.1021/ja501681y | J. Am. Chem. Soc. 2014, 136, 6453−6462