Angewandte
Chemie
cles need to be overcome: 1) The reaction at the more
hindered site of the ketone, as required by the intramolecular
reaction, is difficult owing to an unfavorable steric interaction
between the metal and the b-substituent (opposite selectivity
to the intermolecular version); 2) the reactivity of non-
ethylene olefins could be an issue as they have shown low
reactivity in intermolecular settings; 3) the control of the
linear/branched regioselectivity is another concern; and
4) finding a mild way to form enamines with ketones that is
compatible with the alkylation conditions is nontrivial.
Hence, an efficient catalytic system had to be developed for
the intramolecular cyclization.
Cyclic ketone 1a was employed as the model substrate.
Initially, under the conditions that worked best for the
intermolecular reaction[8] (with IMes and 7-azaindoline as
the ligands), only 20% of the cyclized product (4a) was
obtained with approximately 35% of the olefin-migration
side product 5a. Interestingly, after carefully evaluating
several cocatalysts, the less hindered simple 2-aminopyridine
(2) was found to be more effective for this intramolecular
alkylation (Table 1).[11] Inspired by Koolꢀs aniline-based
catalysts for hydrazone or oxime formation,[12] 5-methyl-2-
aminobenzoic acid (3) was used as the cocatalyst to facilitate
condensation of 2-aminopyridine with the ketone substrate.
After a survey of rhodium precatalysts, ligands, solvents, and
additives, fused ring 4a was obtained in 74% yield (7:1 d.r.;
conditions A).[13] One key feature is that the reaction
conditions are both pH- and redox-neutral. In contrast to
the intermolecular reaction, complete selectivity for the
branched (Markovnikov addition) product was observed.
À
Moreover, no competitive activation of the ketone a-C C
bond was observed.[14]
To gain a better understanding of the reaction conditions,
a set of control experiments were carried out. Formation of
the bicyclic product was not observed without the Rh catalyst
or 2-aminopyridine (Table 1, entries 2 and 3). In the absence
of 5-methyl-2-aminobenzoic acid (3), 4a was formed in only
8% yield (entry 4). The product yield was slightly reduced
under anhydrous conditions or when [{Rh(coe)2Cl}2] and
PMePh2 were substituted by Wilkinsonꢀs catalyst (entries 5
and 6), although the exact reason is unclear. meta-Xylene
proved to be a better solvent than toluene and 1,4-dioxane
(entries 7 and 8). Interestingly, replacement of 2-aminopyr-
idine (2) with aniline alone or both aniline and pyridine only
provided a small amount of the cyclization product, confirm-
ing the important role of 2-aminopyridine in this trans-
formation (entries 9 and 10). Gratifyingly, the cyclization also
proceeds at lower temperature, for example, 1108C, but it
then requires a longer reaction time (entry 11). The use of
a catalytic amount of 2 or 2.5 mol% of the rhodium dimer
proved less efficient (entries 12 and 13). During these studies,
a complementary set of reaction conditions (conditions B)
was also discovered; it involves the use of 2 (25 mol%),
TsOH·H2O (10 mol%), and Wilkinsonꢀs catalyst (10 mol%)
at 1508C and provided 4a as a single diastereomer in
a comparable yield (entry 14). Substrate 1a could also be
cyclized in a cationic gold catalyzed process,[7] but a different
diastereomer was obtained as the major product.
Table 1: Selected optimization studies.[a]
The substrate scope was initially explored with different
ketones (Table 2). Whereas the reaction of cyclopentanone
1b under conditions A gave the corresponding product in
only 41% yield, surprisingly, the use of bulkier 3-methyl-2-
aminopyridine provided 4b in 71% yield as a single diaste-
reomer (S,S,S), but the exact reason is unclear. Generally, for
ketones that are known to be less prone to enamine
formation, for example, linear, aryl, or medium-sized cyclic
ketones,[15] the conditions with TsOH·H2O (conditions B) led
to higher reactivity than conditions A. For example, whereas
cycloheptanone 1c gave product 4c in only 43% yield under
conditions A (even at 1508C), conditions B afforded the
product in 82% yield. Moreover, low conversion (< 20%)
was observed for acyclic and aryl ketone substrates under
conditions A; however, under modified conditions B, the
desired cyclization products (4d and 4e) could be isolated
in synthetically useful yields. Cyclooctanone 1 f, a much more
challenging substrate, gave considerable olefin isomerization
with low conversion into the desired product even under
conditions B. Gratifyingly, the use of an electron-deficient
Entry Variations from “conditions A”
4a
d.r.
(4)
5a
[%][b]
[%][b]
1
2
3
4
5
6
–
74 (58)[c] 7:1
–
–
8
61
2
–
12
–
–
7
without [{Rh(coe)2Cl}2]
without 2
without 3
without H2O
[RhCl(PPh3)3] instead of [{Rh(coe)2Cl}2] 56
and PMePh2
–
–
3:1
6:1
6:1
7
8
9
toluene instead of m-xylene
1,4-dioxane instead of m-xylene
aniline instead of 2
56
21
10
6
6:1
7
–
–
–
3
3
–
–
2.4:1
1:1.4
2:1
10
11
12
13
14
aniline and pyridine instead of 2
1108C for 5 days instead of 1308C
2 (25 mol%) and 3 (7 mol%)
[{Rh(coe)2Cl}2] (2.5 mol%)
[RhCl(PPh3)3] (10 mol%), TsOH·H2O
(10 mol%), 2 (25 mol%), m-xylene,
1508C, 0.1m (conditions B)
64 (56)[c] 7:1
38
26
66 (59)
3.3:1
ligand,
tris(3,5-di(trifluoromethyl)phenyl)phosphine
5:1
(30 mol%), along with [{Rh(coe)2Cl}2] (5 mol%) and AgPF6
(10 mol%), suppressed the olefin isomerization, and pro-
vided the desired 8,5-fused bicycle (4 f).[16]
[d]
–
The functional-group compatibility was first examined
with cyclohexanone-based substrates (Table 3). As expected,
owing to the pH/redox neutrality, a remarkable range of
functional groups, including benzyl ethers, esters, acetals, and
[a] All reactions were run on 0.1 mmol scale with 1.0 mL of the indicated
solvent. [b] Determined by 1H NMR spectroscopy using 1,2-tetra-
chloroethane as the internal standard. [c] Yield of the isolated major
diastereomer. [d] Single diastereomer. coe=cyclooctene.
Angew. Chem. Int. Ed. 2015, 54, 15294 –15298
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Lim, Hee Nam
