Published: March 28, 2011
r 2011 American Chemical Society 5752 dx.doi.org/10.1021/ja201035b | J. Am. Chem. Soc. 2011, 133, 5752–5755
COMMUNICATION
pubs.acs.org/JACS
Why Do Some Fischer Indolizations Fail?
Nihan C-elebi-€Olc-€um, Ben W. Boal, Alexander D. Huters, Neil K. Garg,* and K. N. Houk*
Department of Chemistry and Biochemistry, University of California, Los Angeles, California 90095, United States
b
S Supporting Information
ABSTRACT: The mechanisms of the Fischer indole synthe-
sis and competing cleavage pathways were explored with
SCS-MP2/6-31G(d) and aqueous solvation calculations.
Electron-donating substituents divert the reaction pathway
to heterolytic N
N bond cleavage and preclude the acid-
promoted [3,3]-sigmatropic rearrangement.
Indole derivatives continue to receive substantial interestbecause of their wide range of biological activity.1
5 The
Fischer indole synthesis6 remains among the most widely used
approaches to indoles, with more than 700 reports over the
last 15 years.5 Despite the extensive application of the Fischer
indole sequence, certain substitution patterns cause the reac-
tion to fail.
A notable challenge for the Fischer indolization reaction is the
synthesis of C3 N-substituted indoles (1 f 3; Scheme 1).
Various 3-aminoindole derivatives display antimalarial, antimus-
carinic, antibacterial, antiviral, antiplasmoidal, and antihypergly-
cemic activities and are attractive pharmacological targets.7,8 How-
ever, to date there have been no examples of 3-aminoindole
synthesis by the Fischer method, and the corresponding prep-
arations of N-(indol-3-yl)amides9
12 and 3-pyrazolylindoles13,14
proceed poorly in the presence of protic acids. While the use
of Lewis acids (e.g., ZnCl2, ZnBr2) improves the efficiency of
these cyclizations,9 the question remains: Why do these Fischer
indolizations fail?
We have encountered similar difficulties in efforts to synthe-
size complex indoline-containing natural products using the
interrupted Fischer indolization cascade15 (4 þ 5f 6; Scheme 1).
Reactions between aryl hydrazines 4 and latent aldehydes 5
delivered 3-alkyl- and 3-aryl-substituted pyrrolidindolines 6a and
6b, respectively, in good yields,15 but the transformation failed en
route to 3-indolylpyrrolidindoline 6c, which was intended to be a
model study for the synthesis of psychotrimine16,17 (7) and
related alkaloids.
Table 1 shows a sampling of our unsuccessful attempts at
interrupted Fischer indolization of substrate 5c. Phenylhydrazine
was employed in initial experiments. Acetic acid-based condi-
tions, commonly used for the interrupted Fischer indolization
reaction, gave none of the desired indoline product (entries 1 and
2). Similarly, the use of stronger acids typically employed to
promote the Fischer indole synthesis was also unsuccessful
(entries 3
6).18 All of these experiments gave rise to two signi-
ficant byproducts: 3-methylindole and aniline. Comparable
results were obtained when arylhydrazone derivatives of 5c were
treated under acidic conditions.
Even though it is a widely utilized process, many essential
mechanistic details underlying the acid-promoted Fischer indo-
lization remain unclear. Previous computational investigations
on the mechanism of the Fischer indole reaction have been
limited to semiempirical methods,19 and the effect of substituents
on the possible competing pathways has not been addressed to
date. Here we report the first computational study of the
mechanism of the Fischer indole reaction using accurate quan-
tum-mechanical methods. We demonstrate that substituents on
the starting carbonyl compound play a pivotal role in the success
or failure of the Fischer indole synthesis. We also show that the
commonly used B3LYP method fails to reproduce the concerted
nature of the acid-promoted 3,4-diaza-Cope rearrangement.
We first studied the parent unsubstituted rearrangement using
different levels of theory, including CBS-QB3, B3LYP, SCS-
MP2, MP2, and M06-2X, as implemented in Gaussian 09.20
Solvation effects were taken into account in geometry optimiza-
tions and energy calculations using the SMD model.21 B3LYP
favors N
Nbond cleavage without C
Cbond formation for the
protonated species and failed to predict the concerted nature of
the sigmatropic rearrangement transition states upon substitu-
tion. These results and a detailed comparison of all of the methods
are given in the Supporting Information (SI). In the text, we
discuss results obtained at the SCS-MP2/6-31G(d)(water)//
MP2/6-31G(d)(water) level of theory, which provided the best
results in test calculations.
Figure 1 shows the free energies of the ene
hydrazine inter-
mediates and [3,3]-sigmatropic rearrangement transition states
relative to the phenylhydrazone for both the thermal and acid-
catalyzed (NR-protonated and Nβ-protonated) pathways. In the
thermal reaction, ene
hydrazine intermediate 9 lies 17.5 kcal/mol
higher in energy than phenylhydrazone 8. The rearrangement
transition state (t
TS) is concerted but asynchronous with
a very high activation barrier of 43.7 kcal/mol. Protonation of
either nitrogen gives earlier transition states, increased asynchro-
nicity, and a substantial decrease in the activation energy by
11
13 kcal/mol. Stabilization of the ene
hydrazine intermedi-
ates due to protonation (9a and 9b) is less significant (ΔΔG =
3.1 kcal/mol). Overall, the Nβ-protonated pathway is favored
by 1.5 kcal/mol and yields the rearranged product 10b in a con-
certed fashion. M06-2X significantly overestimated the barrier
of the [3,3]-sigmatropic rearrangement by 6
10 kcal/mol
(see the SI).
The influence of various substituents was evaluated compu-
tationally (Table 2). A single methyl substituent (entry 2) leads
to additional stabilization of both the ene
hydrazine intermedi-
ates (ΔΔG≈ 4 kcal/mol) and [3,3]-sigmatropic rearrangement
Received: February 2, 2011

5753 dx.doi.org/10.1021/ja201035b |J. Am. Chem. Soc. 2011, 133, 5752–5755
Journal of the American Chemical Society COMMUNICATION
transition states (ΔΔG ≈ 6 kcal/mol) relative to the parent
reaction (entry 1). The energies of the protonated ene

hydrazines are essentially identical, and β
TS is still favored
over r
TS. The favorable [3,3]-sigmatropic rearrangement of
the monomethylated substrate (entry 2) is consistent with
experimental data for the Fischer indole synthesis.5
Condensation of phenylhydrazines with 3-substituted hemi-
aminals or lactols (the so-called interrupted Fischer indolization
strategy15) involves disubstituted phenylhydrazone intermedi-
ates. We found that the second substituent further stabilizes the
intermediates and transition states by 1
3 kcal/mol (entry 3)
relative to the monomethyl-substituted reaction. The NR- and
Nβ-protonated pathways have comparable energies (Figure 2).
The reaction profile obtained with the indolyl substituent, on
the other hand, is completely different (entry 4 and Figure 3).
r
TS-indolyl (ΔG = 18.0 kcal/mol) is noticeably lower in
energy than β
TS-indolyl (ΔG = 21.2 kcal/mol), but the
favored transition state is not that of a [3,3]-sigmatropic re-
arrangement (Figure 3). Instead, the intrinsic reaction coordi-
nate (IRC) gives the stable π complex 11. In solution, this com-
plex dissociates to form aniline and iminyl carbocation 12.
Therefore, for the indolyl-substituted reaction, the NR-protona-
ted pathway leads to dissociation rather than rearrangement in
solution. This suggests that the iminyl carbocation formed by the
heterolytic N
N bond cleavage is stabilized by the electron-
donating indolyl substituent and that this is responsible for the
failure of the Fischer indolization for this substitution pattern. In
place of an indolyl substituent, an acylated amine was evaluated
(entry 5). Similarly, heterolytic N
N bond cleavage was favored
over the [3,3]-sigmatropic rearrangement. This result explains
why the acid-catalyzed Fischer indolization of amide-containing
substrates has proved challenging.9
14
To better understand this behavior, we calculated the hetero-
lytic bond dissociation enthalpies (BDEs) of NR- and Nβ-proto-
nated ene
hydrazine intermediates (Table 3). As highlighted by
entries 1
5, substantial weakening of the N
N bond occurs in
9a having more electron-donating substituents on the terminal
alkene. The activation barriers of the NR-protonated species are
lowered, and the transition states became more dissociative. The
dissociative character of the weak N
N bond eventually pre-
cludes the [3,3]-sigmatropic rearrangement, and the ene
hyd-
razine intermediate collapses to aniline and a stabilized iminyl
carbocation. The heterolytic N
N bond cleavage leads to side
reactions rather than the Fischer indolization. These results are in
accord with our experimental observations and the experimental
findings by Mann and Cook.22 Previous studies of substituent
effects on the Cope rearrangement and related 3,3-sigmatropic
shifts indicated that substituents that stabilize either associative
or dissociative transition states accelerate the concerted rearran-
gement.23 Extreme stabilization of the dissociative transition
state can eventually lead to dissociation, as was noted previously
for amido-Cope rearrangements.24
In contrast to electron-donating substituents, electron-with-
drawing groups weaken the N
N bond in 9b and stabilize the
N
N bond in 9a (Table 3, entry 8 vs 6). This suggests that
changing the amino substituent to amido would somewhat
disfavor the competing dissociative pathway. Indeed, the N-acyl
group notably increases the strength of the N
N bond in the
NR-protonated ene
hydrazines relative to indolyl (Table 3,
entries 4 and 7). However, the BDE is still low (8.9 kcal/mol)
in comparison with the case having only alkyl substituents
Scheme 1. Classical and Interrupted Fischer Indolization
Sequences
Table 1. Interrupted Fischer Indolization Attempts with 5c
entry conditions
1 1:1 AcOH/H2O, 25 to 110
C
2 AcOH, 25 to 110
C
3 TFA, C2H4Cl2, 25 to 80
C
4 HCl(aq), CH3CN, 25 to 120
C
5 H2SO4(aq), CH3CN, 25 to 120
C
6 TsOH, t-BuOH, 25 to 80
C
Figure 1. Free energies (ΔG, in kcal/mol) for the transformation of
hydrazone to imine for the (top) thermal and (bottom) acid-promoted
reactions [SCS-MP2/6-31G(d)(water)//MP2
6-31G(d)(water)].