Organic Letters
Letter
outcome of these reactions has not been predictable.20 Endo
additions, onto O, are particularly of interest as they lead to
oxygen-containing heterocycles.21
Scheme 3. Possible Pathways for Olefin/Carbonyl
Cyclization
We have generated our carbon-centered radicals by HAT to
N-vinylindoles 1. Padwa has shown that such enamines are
useful for the synthesis of alkaloids.12,22 The reduction of the
carbonyl of I and generation of a xanthate from the resulting
alcohol led to a radical that adds to the 2-position of the indole
and affords the tetracyclic indole derivatives II (Scheme 2).12
Scheme 2. N-Vinylindole as a Useful Intermediate in
Alkaloid Synthesis
As shown by the first three entries in Table 1, changing the
axial ligands in our cobaloxime catalyst,9b,c from THF in C-1 to
MeOH in C-2 and H2O in C-3, decreases the turnover
number. (Catalysts C-2 and C-3 are known to activate H2 less
effectively than C-1.9a−c) C-1 and C-3 gave roughly the same
product ratios; C-2 gave a somewhat greater ratio of
cycloisomerization to isomerization with low conversion.
Changes in the size of the cycloalkanone ring, however,
significantly affect the 2/3 ratio. A five-membered ring favors
isomerization (entry 4), whereas medium-sized rings (seven-
or eight-membered, entries 1 and 6) favor cycloisomerization;
the larger 12-membered ring (entry 7) shows little selectivity.
The use of CpCr(CO)3H26 as a catalyst (entry 8) induces only
isomerization, with low conversion. Increased C-1 loading
(entry 9) or H2 pressure (entry 10) slightly improves the
selectivity. Control experiments (entries 11 and 12) confirm
that both cobalt and H2 are indispensable to the reaction,
supporting a HAT mechanism. No hydrogenation of cyclized
or uncyclized products has been detected.
We believed that useful radicals might also be generated by
H• transfer onto the vinyl groups of 1. When we tried our
cobaloxime/H2 system9 on an N-vinylindole with a tethered
cyclohexanone (1a), we obtained the cycloisomerized product
2a along with the isomerization product 3a (Scheme 1C and
Table 1, entry 2). It seemed likely that both 2a and 3a had
Table 1. Initial Optimization Studies in the Cyclization of
a
N-Vinylindoles with Cobaloxime/H2
b
b
entry
substrate
catalyst
C-1
C-2
C-3
C-1
C-1
C-1
C-1
CpCr(CO)3H
C-1
conversion (%) ratio (2:3)
1
2
3
4
5
6
7
8
1d, n = 8
1d, n = 8
1d, n = 8
1b, n = 5
1a, n = 6
1c, n = 7
1e, n = 12
1d, n = 8
1d, n = 8
1d, n = 8
1d, n = 8
1d, n = 8
86
45
47
82
88
84
87
29
93
95
<5
<5
55:31
18:27
31:16
14:68
40:48
48:36
49:38
<1:27
61:32
69:26
f
We then considered the Co−salen complexes10a,27 and
silane reductants28,29 that Shenvi had used in the cyclo-
isomerization of dienes.10 We selected N-vinylindole with a
tethered cyclooctanone (1d) as a model substrate for our
cyclizations onto CO and were encouraged when we
obtained 10% 2d along with a trace of the isomerization
product 3d at room temperature(see the SI).
c
Optimization revealed that the best condition (with 12 mol
% catalyst B and silane, at 60 °C for 16 h) had the substrate 0.1
M in benzene. The results of catalyst optimizations are
summarized in Scheme 4. The nature (steric and electronic) of
the salen ligand has a large effect,27,30 with catalysts A, B, C,
and I all being effective, whereas catalysts DG showed lower
reactivity. Catalyst H gave the best selectivity for the
cycloisomerized product 2d, but with only moderate
conversion (see the SI). The auxiliary ligand (Cl, Br, or I) of
the Co(III) salen catalyst affects the selectivity (probably
because it affects the rate at which “Co−H” is generated).The
nature of the silane reductant is crucial:29 tertiary silanes are
less reactive, presumably for steric reasons, whereas Ph2SiH2
shows decent reactivity along with good cycloisomerization
selectivity (up to 9.2:1). We saw the same ring size effects as
with the cobaloxime/H2 system, albeit with higher selectivity
for 2 and higher conversion; again, the cyclopentanone 1b gave
3b as the main product, with only a trace (<1%) of the
cycloisomerization product 2b.
9
d
10
C-1
no [Co]
C-1
11
12
e
f
a
Conditions: 1 (0.2 mmol), 7.0 mol % catalyst, 4.8 atm H2, benzene
b
(0.05M), 50 °C, 3 days. Determined by 1H NMR with internal
c
d
e
standard. 20 mol % C-1. Under 6.1 atm H2. Under Ar without H2.
f
Starting material recovered.
arisen from the α-aminyl radicals23,24 formed by H• transfer
(Scheme 3, pathway B), but they could also have come from
the acid-promoted olefin/carbonyl cyclization (Scheme 3,
pathway A).25 The latter possibility is, however, unlikely in
view of the hydrolytic instability of enamines like 1a in the
presence of acid12 (attempts to take the NMR of 1a in CDCl3
just showed indole and the methyl ketone (eq 1), whereas 1a is
stable in acetone-d6). We believe that both 2a and 3a are the
result of HAT (Scheme 3, pathway B).9a,10a
B
Org. Lett. XXXX, XXX, XXX−XXX