Communications
doi.org/10.1002/ejoc.202100021
°
Results and Discussion
d5 (Figure 1).[23,24] When the temperature was raised to 80 C,
the spectral data clearly exhibited each peak on azepine ring.
In particular, the 6-methylene protons, which could not be
observed in CDCl3, appeared at δ 4.50 ppm, accompanied by
the methylene protons at δ 4.28 ppm and ethoxy groups at
9-position of azepino[1,2-a]indoles in pyridine-d5, because of
the acceleration of interconversion of some conformers. The
typical structure was supported by their spectral data and
elemental analysis.
1
2
3
4
5
6
7
8
9
We first examined a [6+1] annulation reaction of the parent
substrate, 2-cyano-1-propargylindole 1a with Reformatsky
reagent, generated from ethyl bromoacetate and activated
zinc, in the presence of hafnium triflate as a standard
condition of [6+1] annulation reaction of N-tethered
ynenitriles.[23] Both azepino[1,2-a]indole 2a and pyridoindole
3a were respectively obtained via either a 7-endo-mode or a
6-exo-mode annulation of the β-aminoacrylate intermediates.
We were pleased to see the preliminary results and further
examined the Lewis acid screening to lower the formation of
6-exo-mode cyclization products (Table 1). The reaction with-
out Lewis acids gave 2a (49%) and 3a (38%), respectively.
The most suitable reaction condition of the monocyclic
azepines using hafnium triflate gave an unsatisfactory result
(entry 2). Both the reactions using either indium triflate or
ytterbium triflate resulted in low yields of products; however,
the scandium triflate-catalyzed reaction exclusively under-
went a [6+1] annulation reaction to give 2a in 99% yield.
The product, 10-amino-6H-azepino[1,2-a]indole-9-carbox-
ylate (2a), did not give a complete set of proton signals in
the 1H NMR spectrum. For instance, we cannot clearly see
some protons of both the 6-methylene protons and/or the
methylene protons of ethoxy carbonyl group on the azepine
Under the optimized reaction conditions in hand, we next
researched the scope of the [6+1] annulation reactions of 2-
cyano-1-propargylindoles with some Reformatsky reagents
and the results are summarized in Table 2. The reaction of 2-
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
cyano-1-(phenylpropargyl)indole 1b with
a Reformatsky
reagent (BrZnCH2CO2Et) succeeded to give the azepine 2b in
75% yield. Interestingly, the methyl signal of 9-ethoxy
carbonyl group of 2b was significantly upfield-shifted at δ
0.63 ppm in the 1H NMR spectrum due to the anisotropic
effect of 8-phenyl group. The upfield shift of the signal due
to methyl group of esters was also observed in the other 8-
phenylated azepino indole derivatives. We next performed
the reactions of both 1a and 1b with methyl ester
(BrZnCH2CO2Me) afforded methyl 10-amino-8-phenyl-5b and
methyl 10-amino-8-ethyl-6H-azepino[1,2-a]indole-8-carboxy-
late (5a), respectively. In order to investigate the substituent
effect on the indole ring, we prepared a wide variety of
substrates by almost the same method (the results are
summarized in the SI). The Reformatsky reaction of 5-meth-
oxyindole gave the azepino[1,2-a]indole 2d, accompanied by
the pyrido[1,2-a]indole 3d; however, the reaction with meth-
yl ester selectively afforded the azepine analog 2f in 78%
yield. The reaction of 1e with the bulky t-butyl ester resulted
in the low yield of 7e. The reactions of 5,6-dimethoxyindoles
1f and 1 g exclusively afforded two kinds of azepines 2f and
2g, respectively; however, the reactions of 5-fluoro-, 4,6-
difluoro-, and 4,6-dichloroindoles 1h–j with a Reformatsky
reagent, obtained a significant amount of pyrido[1,2-a]
indoles in each case. We also investigated the substituent
effect of these annulation reactions using 2-cyano-1-prop-
argylindoles 1k–1q bearing some substituents at the alkyne
terminus. We selected indole 1r as a tentative substrate for a
large–scale scandium-catalyzed [6+1] annulation reaction.
The reaction of 1r was tolerated at 0.30 g scale (1.00 mmol
scale) and afforded 2r in 72% yield. The 1 g scale reaction of
1x successfully afforded 0.5 g of 2 x, however, a half amount
of complex mixture was obtained. The scalability of this
annulation needs further optimization processes.
°
ring at 25 C due to the broadening in the spectral data in
CDCl3. In order to solve this inconveniency, we performed the
temperature variable NMR measurements of 2a in pyridine-
Table 1. Optimization of the [6+1] cycloaddition.[a]
Entry
Lewis acids
Reaction time (h)
Yield of 2a[b]
Yield of 3a
1
2
3
4
5
–
5
1
2.5
2
49
38
20
0
24
–
Hf(OTf)4
In(OTf)3
Yb(OTf)3
Sc(OTf)3
46
37[c]
31
0.5
99
[a] Reaction condition: 1a (0.19 mmol), zinc (0.19 mmol), ethyl bromoace-
tate (0.95 mmol) and catalyst (0.027 mmol) in 1,4-dioxane (1.00 mL).
[b] Yields of isolated products. [c] 1a was recovered in 50% yield.
The annulations of 2-cyanoindoles bearing hetaryl groups
at the alkyne terminus afforded 8-(thiophen-2-yl)- and 8-
(indolyl-3-yl)-azepines 2α and 2β; however, the reaction of
pyridine-substituted indole 1γ did not give azepino[1,2-a]
indole 2γ, but pyrido[1,2-a]indole 3γ in good yield.
In light of these encouraged results, the [6+1] annulation
reactions of different types of substrates, 2-alkynyl-1-
cyanomethyl indoles 8 with similar Reformatsky reagent were
performed and the results are shown in Table 3. After few
screenings of Lewis acids, we chose hafnium triflate as suitable
Figure 1. Temperature variable NMR measurements of 2a in pyridine-d5.
Eur. J. Org. Chem. 2021, 1553–1558
1554
© 2021 Wiley-VCH GmbH