Synthesis of highly enantioenriched 3,4-dihydroquinolin-2-ones by 6-Exo-trig radical cyclizations of axially chiral α-halo-ortho-alkenyl anilides

Article abstract of DOI:10.1021/ja9066282

Radical cyclizations (Bu₃SnH, Et₃B/air, rt) of racemic R-halo-ortho-alkenyl anilides provide 3,4-dihydroquinolin-2-ones in high yield. Cyclizations of enantioenriched precursors occur in similarly high yields and with transfer of axi

Full text of DOI:10.1021/ja9066282

Published on Web 10/02/2009
Synthesis of Highly Enantioenriched
3,4-Dihydroquinolin-2-ones by 6-Exo-trig Radical Cyclizations
of Axially Chiral r-Halo-ortho-alkenyl Anilides
David B. Guthrie, Steven J. Geib, and Dennis P. Curran*
Department of Chemistry, UniVersity of Pittsburgh, Pittsburgh, PennsylVania 15260
Received August 5, 2009; E-mail: curran@pitt.edu
Abstract: Radical cyclizations (Bu3SnH, Et3B/air, rt) of racemic R-halo-ortho-alkenyl anilides provide 3,4-
dihydroquinolin-2-ones in high yield. Cyclizations of enantioenriched precursors occur in similarly high yields
and with transfer of axial chirality to the new stereocenter of the products with exceptionally high fidelity
(often >95%). Single and tandem cyclizations of R-halo-ortho-alkenyl anilides bearing an additional
substituent on the R-carbon occur with high chirality transfer and high diastereoselectivity. Straightforward
models are proposed to interpret both the chirality transfer and diastereoselectivity aspects. These first
examples of an approach for axial chiral transfer from a reactive species in the amide to an acceptor suggest
broad potential for extension both within and beyond radical reactions.
Introduction
The dihydroquinolin-2-one ring is an important heterocycle
that is featured in both natural products and medicinally active
compounds.1 Dihydroquinolin-2-ones can exist as isolated ring
systems, as in the penigquinolones,2 pinolinone,3 R2ꢀ3 integrin
antagonists4 (see 1, Figure 1), and a class of HIV reverse
transcriptase inhibitors.5 Or, they can be conjoined with other
rings, as in scandine and the related meloscine alkaloids (see
2, Figure 1).6,7 Many methods exist to make racemic dihydro-
quinolin-2-ones, but only a few routes toward single enantiomers
have been reported.8 These methods suffer from various
drawbacks (low ee, limited generality), so new methods are
needed.
Figure 1. Representative 3,4-dihydroquinolin-2-ones with ring system
highlighted in red.
Many of the most important dihydroquinolin-2-ones have
alkyl or other substituents at the C-4 position, and convenient
ways to access such compounds in racemic form involve 6-exo-
trig cyclizations. Figure 2 shows examples of both radical9 and
ionic approaches. Jones reported that reduction of 3 with
Bu3SnH provided target dihydroquinolin-2-one 4 along with
comparable amounts of the product of 1,5-hydrogen transfer 5
in 51% combined yield.10 In this method (Approach 1), a radical
generated on the aryl ring cyclizes to an acceptor on the anilide
carbonyl group. Ikeda11 and Parsons12 took the reverse strategy
(Approach 2), generating a radical adjacent to the amide
(1) For a more complete collection of references to biologically active
3,4-dihydroquinolin-2-ones and methods of synthesis, see: Zhou, W.;
Zhang, L.; Jiao, N. Tetrahedron 2009, 65, 1982–1987.
(2) Ito, C.; Itoigawa, M.; Otsuka, T.; Tokuda, H.; Nishino, H.; Furukawa,
H. J. Nat. Prod. 2000, 63, 1344–1348.
(3) Kimura, Y.; Kusano, M.; Koshino, H.; Uzawa, J.; Fujioka, S.; Tani,
K. Tetrahedron Lett. 1996, 37, 4961–4964.
(4) (a) Ellis, D.; Kuhen, K. L.; Anaclerio, B.; Wu, B.; Wolff, K.; Yin,
H.; Bursulaya, B.; Caldwell, J.; Karanewsky, D.; He, Y. Bioorg. Med.
Chem. Lett. 2006, 16, 4246–4251. (b) Patel, M.; McHugh, R. J., Jr.;
Cordova, B. C.; Klabe, R. M.; Bacheler, L. T.; Erickson-Viitanen, S.;
Rodgers, J. D. Bioorg. Med. Chem. Lett. 2001, 11, 1943–1945.
(5) Seitz, W.; Geneste, H.; Backfisch, G.; Delzer, J.; Graef, C.; Hornberger,
W.; Kling, A.; Subkowskic, T.; Norbert, Z. Bioorg. Med. Chem. Lett.
2008, 18, 527–531.
(8) (a) El Ali, B.; Okuro, K.; Vasapollo, G.; Alper, H. J. Am. Chem. Soc.
1996, 118, 4264–4270. (b) Dong, C.; Alper, H. Tetrahedron: Asym-
metry 2004, 15, 35–40. (c) Okuro, K.; Kai, H.; Alper, H. Tetrahedron:
Asymmetry 1997, 8, 2307–2309. (d) Blay, G.; Cardona, L.; Torres,
L.; Pedro, J. R. Synthesis 2007, 108–112. (e) Bolm, C.; Hildebrand,
J. P. Tetrahedron Lett. 1998, 39, 5731–5734.
(6) Isolation: (a) Bernauer, K.; Englert, G.; Vetter, W. Experientia 1965,
21, 374–375. (b) Bernauer, K.; Englert, G.; Vetter, W.; Weiss, E. HelV.
Chim. Acta 1969, 52, 1886–1905. (c) Oberha¨nsli, W. E. HelV. Chim.
Acta 1969, 52, 1905–1911. (d) Plat, M.; Hachem-Mehri, M.; Koch,
M.; Scheidegger, U.; Potier, P. Tetrahedron Lett. 1970, 39, 3395–
3398.
(9) Other radical approaches include bimolecular additions, see ref 1, and
cyclizations to isocyanates, see: Minin, P. L.; Walton, J. C. J. Org.
Chem. 2003, 68, 2960–2963.
(10) (a) Clark, A. J.; Jones, K.; McCarthy, C.; Storey, J. M. D. Tetrahedron
Lett. 1991, 32, 2829–2832. (b) Clark, A. J.; Jones, K. Tetrahedron
1992, 48, 6875–6882. Related additions to aryl rings: (c) Lopez de
Turiso, F. G.; Curran, D. P. Org. Lett. 2005, 7, 151–154.
(11) (a) Sato, T.; Ishida, S.; Ishibashi, H.; Ikeda, M. J. Chem. Soc., Perk.
Trans. 1 1991, 353–359. (b) Ishibashi, H.; Sato, T.; Ikeda, M. Synthesis
2002, 6, 695–713.
(7) Synthesis: (a) Overman, L. E.; Robertson, G. M.; Robichaud, A. J.
J. Am. Chem. Soc. 1991, 113, 2598–2610. (b) Denmark, S. E.; Cottell,
J. J. AdV. Synth. Catal. 2006, 348, 2397–2042. (c) Selig, P.; Bach, T.
Angew. Chem., Int. Ed. 2008, 47, 5082–5084. (d) Selig, P.; Herdtweck,
E.; Bach, T. Chem.sEur. J. 2009, 15, 3509–3525.
(12) Allan, G. M.; Parsons, A. F.; Pons, J.-F. Synlett 2002, 1431–1434.
9
15492 J. AM. CHEM. SOC. 2009, 131, 15492–15500
10.1021/ja9066282 CCC: $40.75  2009 American Chemical Society

Radical Cyclizations of R-Halo-ortho-alkenyl Anilides
A R T I C L E S
Figure 2. Two 6-exo-trig cyclization approaches to racemic dihydroquino-
lin-2-ones.
carbonyl group and allowing it to cyclize onto an alkene at the
ortho position of the aryl ring. For example, cyclization of 6
under syringe pump conditions provided 7 in 79% isolated yield.
Cyclization strategies toward these targets are not limited to
radical approaches. Procter and co-workers have developed a
general approach to racemic dihydroquinolin-2-ones by using
intramolecular conjugation addition of R-arene sulfonyl anil-
ides.13 For example, base-promoted cyclization of 8 followed
by reductive desulfonylation with samarium diiodide provided
9 in 57% yield. R-Alkylation can be conducted between the
cyclization and the desulfonylation steps to provide products
with substitution at C-3. The sulfone can also be grafted to a
polymer bead for solid phase synthesis.
It is now widely recognized that ortho-substituted anilides
like 3, 6 and 8 are not planar, as implied by the two-dimensional
drawings in Figure 2. Instead, the planes of the amide group
and the aryl ring are roughly orthogonal, as shown by the
structures in Figure 3.14 Accordingly, the molecules are axially
chiral, and stable atropisomers result with suitable substitution
patterns.15 With such molecules, cyclizations (radical or oth-
erwise) might occur with transfer of axial chirality to give a
product whose configuration is predetermined by that of the
starting material.16 For example, tin hydride cyclizations of
o-iodoacrylanilides 10 provide indol-2-ones 11 in high yields
with excellent levels of chirality transfer. The scope of this
radical reaction is excellent, and anionic17 and organometallic18
variants also exist. Because this produces an indol-2-one rather
Figure 3. (top) Established approach to indol-2-ones compared to (bottom)
two potential approaches to dihydroquinolin-2-ones.
than a dihydroquinolin-2-one, this reaction can be viewed as a
lower homologue of Approach 1 in Figure 2.
However, in assessing prospects for making dihydroquinolin-
2-ones by axial chirality transfer, we felt that Approach 2 was
considerably more interesting than Approach 1 for three reasons.
First, the scope and generality of Approach 1 may be limited;
ꢀ,γ-unsaturated amides like 12 are not as easy to make or handle
as R,ꢀ-unsaturated analogs, and the competing 1,5-hydrogen
transfer reaction that these substrates permit (see 13) could be
a fatal design flaw in many cases.19 Second, Approach 2
represents a potentially new class of chirality transfer reaction
for axially chiral anilides. In all previous reactions, a reactive
intermediate (radical, anion, metal) on the aryl ring is generated
first and reacts in turn with an acceptor in a pendant side chain
on the amide. Is it possible to reverse the roles of the two
partners without sacrificing the high chirality transfer?
The third reason has to do with N-Ar bond rotation, which
must be avoided because it results in racemization. Consider
the generation of radical 13 from 12 in Approach 1. This event
(13) McAllister, L. A.; Turner, K. L.; Brand, S.; Stefaniak, M.; Procter,
D. J. J. Org. Chem. 2006, 71, 6497–6507.
(16) (a) Curran, D. P.; Qi, H.; Geib, S. J.; DeMello, N. C. J. Am. Chem.
Soc. 1994, 116, 3131–3132. (b) Curran, D. P.; Liu, W. D.; Chen, C. H.-
T. J. Am. Chem. Soc. 1999, 121, 11012–11013. (c) Ates, A.; Curran,
D. P. J. Am. Chem. Soc. 2001, 123, 5130–5131. (d) Curran, D. P.;
Chen, C. H. T.; Geib, S. J.; Lapierre, A. J. B. Tetrahedron 2004, 60,
4413–4424. (e) Petit, M.; Geib, S. J.; Curran, D. P. Tetrahedron 2004,
60, 7543–7552.
(14) (a) Oki, M. In Topics in Stereochemistry; Allinger, N. L., Eliel, E.,
Wilen, S. H., Eds.; Wiley InterScience: New York, 1983; Vol. 14, pp
1-81. (b) Oki, M. The Chemistry of Rotational Isomers; Springer-
Verlag: New York, 1993. (c) Wolf, C. Dynamic Stereochemistry of
Organic Compounds; RSC Publishing: Cambridge, UK, 2008.
(15) (a) Curran, D. P.; Hale, G. R.; Geib, S. J.; Balog, A.; Cass, Q. B.;
Degani, A. L. G.; Hernandes, M. Z.; Freitas, L. C. G. Tetrahedron:
Asymmetry 1997, 8, 3955–3975. (b) Adler, T.; Bonjoch, J.; Clayden,
J.; Font-Bard’a, M.; Pickworth, M.; Solans, X.; Sole´, D.; Vallverdu´,
L. Org. Biomol. Chem. 2005, 3, 3173–3183. (c) Petit, M.; Lapierre,
A. J. B.; Curran, D. P. J. Am. Chem. Soc. 2005, 127, 14994–14995.
(17) Guthrie, D. B.; Curran, D. P. Org. Lett. 2009, 11, 249–251.
(18) Lapierre, A. J. B.; Geib, S. J.; Curran, D. P. J. Am. Chem. Soc. 2007,
129, 494–495.
(19) Curran, D. P.; Abraham, A. C.; Liu, H. T. J. Org. Chem. 1991, 56,
4335–4337.
9
J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009 15493

A R T I C L E S
Guthrie et al.
Table 1. Isolated Yields of Radical Cyclization Products 14a-i
from Racemic Precursors 15a-ia
Scheme 1. Typical Synthesis of a Radical Cyclization Precursor
Illustrated with 15a
entry
R-I
R1
R2
R3
R4
R5
% yieldb
1
2
3
4
5
6
7
8
9
15a
15b
15c
15d
15e
15f
15g
15h
15i
Me
Me
Me
OMe
Me
TMS
Br
Me
Me
Me
Me
Me
H
Me
Me
Me
Me
Me
Me
CO2t-Bu
CO2t-Bu
CO2t-Bu
CO2t-Bu
CO2t-Bu
CO2t-Bu
4-BrC6H4
CN
H
H
H
H
H
H
H
H
83
85
97
PMB
PMB
Ts
PMB
PMB
PMB
PMB
PMB
ndc d
,
93
61
ndc
<70d
51e
used for the Heck reaction,21 and the trisubstituted alkene of
15i was appended by Negishi coupling.22
Me
Me
a Conditions: Bu3SnH and Et3B in 20 mL PhH were added over 2 h
via syringe to a 10 mM PhH solution of 15. b Yield of racemic 14 after
isolation by column chromatography on 10% w/w KF/silica gel. c nd )
Not determined. d Fixed concentration of Bu3SnH at 5 mM. e Directly
Cyclizations of Racemic Precursors. Table 1 compiles the
key results of cyclization experiments conducted under a
standard set of conditions. In a typical experiment, Bu3SnH and
Et3B initiator23 were dissolved in degassed PhH, and this
solution was added dropwise by syringe pump over 2 h to a
nondegassed PhH solution of substrate 15a at room temperature.
TLC analysis showed that the reaction was complete at the end
of the addition period. The solvent was removed, and the crude
mixture was directly chromatographed on 10% w/w KF/silica
gel.24 This removed the tin byproducts and at the same time
purified the target product 14a, which was isolated in 83% yield.
A similar experiment at a fixed Bu3SnH concentration of 10
mM was complete in 5 min, but provided only 55% yield of a
sample of 14a that was not completely separable from the
directly reduced product 20a.
Substrate 15b, with an N-p-methoxybenzyl (PMB) substituent,
provided product 14b in 83% purified yield. The ortho sub-
stituent was also varied in this series to include methoxy (15c,
lacking the p-methyl group), trimethylsilyl (TMS, 15e) and
bromine (15f) while holding the N-PMB group and the acceptor
constant. The corresponding products 14c,e,f were isolated in
97, 93, and 61% yields. The TMS group of 14e survived
chromatography over KF-doped silica gel.
The substituents on the acceptor were also changed to
p-bromophenyl (15g), cyano (15h), and dimethyl (15i). Reduc-
tion of the p-bromophenyl substrate 15g provided a complex
product mixture that did not obviously contain the expected
product 14g or the reduced product 20g. The aryl bromide of
15g is probably not the culprit for this failure, because bromide
15f cyclized smoothly. Instead, we speculate that the benzyl
radical resulting from cyclization will not react quickly enough
with tin hydride under these conditions (low concentration and
temperature)25 and instead finds other pathways. Reduction of
nitrile 15h under the syringe pump conditions provided product
14h in about 70% yield but in unsatisfactory purity even after
chromatography. A similar experiment conducted at fixed 5 mM
1
reduced 20 was detected in crude reaction mixture by TLC or H NMR.
must cause a significant decrease in the N-Ar rotation barrier
because one of the ortho-substituents is now a radical, which
is much smaller than its precursor. In contrast, the N-Ar rotation
barriers of precursor 15 and radical 16 in Approach 2 are likely
comparable because their substitution patterns are very similar.
Even so, strict prevention of N-Ar bond rotation is a necessary
but not sufficient condition for axial chirality transfer. Addition
to the alkene must also be face selective, as must addition to
the radical (rho selectivity) if it has three different substituents
(this later selectivity is not illustrated in Figure 3 because the
radical has two hydrogen atoms).
Here we report details of an in-depth study of 6-exo-trig
radical cyclizations of both racemic and enantiopure axially
chiral R-halo-ortho-alkenyl anilides. Dihydroquinolin-2-ones are
formed in good yields and with exceptionally high fidelity in
transfer of axial chirality. As a bonus, the reactions of alkyl-
substituted amide radicals are also highly diastereoselective.
These first results suggest that there is excellent potential to
extend Approach 2 for axial chirality transfer both within and
beyond radical reactions.
Results and Discussion
Synthesis of Radical Cyclization Precursors. An initial set
of nine radical cyclization precursors 15a-i (Table 1) was
prepared in racemic form to assess the efficiency of the radical
cyclization as a function of substituent pattern. The synthesis
of 15a is typical and is summarized in Scheme 1. Heck reaction
of 2-iodo-N,4,6-trimethylaniline 1716e with t-butyl acrylate under
ligand-free conditions20 provided 18a in 92% yield. Acylation
of 18a with chloroacetyl chloride gave R-chloroanilide 19a in
90% yield. This was converted to the R-iodoanilide 15a by the
Finkelstein reaction in 93% yield. The syntheses of all the
precursors 15a-i are described in detail in the Supporting
Information. Most pathways were similar to Scheme 1, with
suitable variations for substrate. Other protocols were sometimes
(21) Jeffery, T. Tetrahedron 1996, 52, 10113–10130.
(22) Campbell, J. B., Jr.; Firor, J. W.; Davenport, T. W. Synth. Commun.
1989, 19, 2265–2272.
(23) Yorimitsu, H.; Oshima, K. In Radicals in organic synthesis; Renaud,
P., Sibi, M. P., Eds.; Wiley-VCH: Weinheim, 2001; Vol. 1, pp 11-
27.
(24) Harrowven, D. C.; Guy, I. L. Chem. Commun. 2004, 1968–1969.
(25) Franz, J. A.; Suleman, N. K.; Alnajjar, M. S. J. Org. Chem. 1986, 51,
19–25.
(20) Opatz, T.; Ferenc, D. Org. Lett. 2006, 8, 4473–4475.
9
15494 J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009

Radical Cyclizations of R-Halo-ortho-alkenyl Anilides
A R T I C L E S
Table 2. Summary of Results of Competition Experiments with
15b
[Bu3SnH]a
% yld 14bb
% yld 20bb
total yld
red:cycc
14.9 mM
19.9 mM
27.0 mM
37.0 mM
49.2 mM
58.8 mM
73.2 mM
67
64
59
58
50
45
33
21
25
30
39
46
45
43
89
89
88
97
95
90
76
24:76
28:72
33:67
40:60
48:52
50:50
57:43
a Initial concentration. b GC yield, using eicosane as an internal
standard. Yields are an average of three trials. c Ratio of reduced 20b to
cyclized 14b.
Figure 4. Mechanistic framework for rate constant measurements.
concentration gave the same product in about the same yield,
but now the purity was considerably better (95% estimated).
The dimethyl substrate 15i provided pure 14i in 51% isolated
yield. Small amounts of the directly reduced product were also
produced, presumably because the alkene of 15i is not as good
of a radical acceptor as the other alkenes.
Finally, a tosyl group was also probed as the N-substituent,
but reduction of 15d gave a mixture of products whose major
component (∼80%) was the directly reduced product 20d. This
failure is surprising since N-tosyl groups have been used in the
past to facilitate cyclizations of R-amidoyl radicals.26
Rate Constant Measurements. The necessity of syringe pump
conditions to minimize premature reduction in reactions of
compounds 15 implies a relatively low rate constant for 6-exo-
trig cyclization. This is not surprising since 6-heptenyl radicals
typically cyclize more slowly than analogous 5-hexenyl ones.
All the atoms connecting the radical precursor and radical
acceptor are sp2-hybridized in 15, so estimating rate constants
by analogy to other hexenyl and heptenyl radicals is risky.
Therefore, we undertook kinetic competition experiments27 to
determine the cyclization rate constant kc for the reaction of a
representative compound 15b.
The mechanistic framework for the rate constant analysis is
shown in Figure 4. Radical 16b derived from 15b can be directly
reduced in a bimolecular reaction with tin hydride to give 20b.
Alternatively, it can undergo 6-exo cyclization to provide a new
radical 21b, which in turn reacts with tin hydride to give 14b.
Measurement of the ratio of cyclized 14b to reduced 20b
products as a function of tin hydride concentration provides the
rate constant kc in the usual way.27
The competition experiments were performed by mixing stock
benzene solutions (nondegassed) of 15b (0.025 mmol), Bu3SnH
(0.030 mmol), and eicosane (internal GC standard) in round-
bottom flasks open to air. A hexane solution of Et3B (0.010
mmol) was added all at once. The reaction mixtures were stirred
for 10 min and were then subjected to GC analysis without
workup to provide standardized yields of 14b and 20b. The
average results of three trials at each concentration are sum-
marized in Table 2, and the usual kinetic competition plot of
this data is shown in Figure 5. Reactions were complete,
Figure 5. Kinetic competition plot of the data in Table 2 for reactions of
radical 16b.
reproducible, high-yielding (76-95%), and clean (no significant
side peaks in GC analysis) in all cases.
By using the estimated value kH ) 2.3 × 106 M-1 s-1 for
radical 16b,28 we calculated kc at 25 °C as 8 × 104 s-1. This
value is about an order of magnitude higher than that for the
6-exo cyclization of the 6-heptenyl radical,27 presumably due
to the activating effect of the ester on the alkene. It is
coincidentally close to the rate constant for 5-exo cyclization
of the 5-hexenyl radical (1 × 105 s-1), a value that synthetic
chemists often use as a touchstone for radical cyclizations of
intermediate rate.
The intercept of a plot of data for an irreversible cyclization
should be zero, but the intercept in Figure 5 appears to be
slightly positive. This raises the possibility that the cyclization
of 16b may be reversible. From the value of the intercept, we
calculated that the ring-opening rate constant of k-c of 21b must
be at least 25 times slower than cyclization, so the back
cyclization can probably be neglected for most practical
purposes. Even so, most 6-exo cyclizations to activated alkenes
are irreversible.29 We speculate that the unfavorable 1,3-steric
interactions between the N-substituent and the substituent of
C-8 of the cyclized intermediate 21b (see Figure 4) may both
retard the cyclization and promote the reverse reaction.
Resolution and Measurements of Rotation Barriers. To
perform the chirality transfer experiments, we secured highly
enantioenriched samples of seven of the precursors by resolution
(28) (a) Chatgilialoglu, C.; Dickhaut, J.; Giese, B. J. Org. Chem. 1991,
56, 6399–6403. (b) Chatgilialoglu, C.; Ingold, K. U.; Scaiano, J. C.
J. Am. Chem. Soc. 1981, 103, 7739–7742.
(26) Stork, G.; Mah, R. Heterocycles 1989, 28, 723–727.
(27) (a) Newcomb, M. Tetrahedron 1993, 49, 1151–1176. (b) Newcomb,
M. In Radicals in Organic Synthesis, 1st ed.; Renard, P., Sibi, M.,
Eds.; Wiley-VCH: Weinheim, 2001; Vol. 1, pp 317-336.
(29) Hanessian, S.; Dhanoa, D. S.; Beaulieu, P. L. Can. J. Chem. 1987,
65, 1859.
9
J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009 15495

A R T I C L E S
Guthrie et al.
Figure 7. Measured rotation barriers for amides 15 and 19 along with
literature benchmarks 22 and 23.
Figure 6. Rotational dynamics of N-methyl anilide 15a.
>98/2), and their resolutions were straightforward. Like 15a,
these samples probably have a substantial barrier to E/Z rotation,
but the population of the Z rotamer is so low at equilibrium
that it can be neglected entirely.
Small samples of four substrates bearing N-PMB groups were
then racemized to measure their respective barriers to N-aryl
bond rotation. The samples were dissolved in 90:10 hexane/i-
PrOH and heated in a sealed tube at an appropriate temperature.
Racemization was monitored by periodically measuring the er
of aliquots of the solution by analytical chiral HPLC. The
experimental data and details of the standard analysis are in
the Supporting Information.
on semipreparative chiral HPLC. Typically, samples of 15 were
dissolved in i-PrOH and injected onto either a Chiralcel OD or
(S,S)-Whelk-O 1 column (25.0 cm × 21.1 mm ID) and eluted
with a hexane:i-PrOH solvent mixture. Depending on the ease
of separation, samples ranging in size from 40-120 mg were
resolved per injection. After fraction collection and solvent
removal, the enantiomeric ratios of the resolved compounds
were measured by analytical chiral HPLC. Six of the compounds
exhibited the expected two peaks for the individual enantiomers,
which were isolated in >98% ee.
The HPLC behavior of N-methyl congener 15a was an
interesting exception. This exists as a 94/6 ratio E/Z of amide
The newly measured rotation barriers are summarized in
Figure 7 along with benchmarks 22 and 23 from previous work
in our group.16d,e The barrier to rotation of o-methyl substrate
1
rotamers as assessed by H NMR analysis (see Figure 6). The
amide rotamers are visible by silica TLC analysis but not
isolable under ambient laboratory conditions. Analytical HPLC
(Whelk, 80:20 hexanes:i-PrOH) of the sample showed a small
peak at 10.5 min, followed by two large peaks at 14.5 and 18.5
min. Analysis of the resolved fractions allowed interpretation
of this chromatogram. The peak at 10.5 min is the minor
Z-amide rotamer of (-)-15a, while the major E amide rotamer
of (+)-15a elutes at 14.5 min. The minor Z amide rotamer of
(+)-15a and the major E amide rotamer of (-)-15a overlap at
18.1 min. A plateau between the minor and major amide rotamer
peaks of each enantiomer shows partial interconversion on the
time scale of the HPLC separation. Based on this behavior, we
estimate that the barrier for E/Z rotation is in the vicinity of
∆GqE/Z ≈ 22 kcal/mol. This barrier is considerably higher than
that for typical amides;30 however, E/Z rotamers of anilides with
even larger ortho substituents have even higher barriers and can
be separated and handled at ambient temperature.31 This effect
complicates the measurement of the enantiomer ratio (er) of
resolved 15a, but the relatively small proportion of the minor
Z rotamer (6%) minimizes any error.
15b was ∆Gq ) 34.3 kcal/mol, while the o-Br substrate 15f
rot
had a similar barrier of 34.2 kcal/mol. The magnitudes of these
barriers are surprisingly high when contrasted against com-
pounds like 22, which has the alkenyl group on the amide rather
than the aryl ring; its barrier to rotation is about 3 kcal/mol
lower. Even 23, which contains a primary group on nitrogen
(allyl) rather than Me, has a barrier 1.5 kcal/mol lower than
that for 15b. With estimated half-lives >20 000 years, atropi-
somers like 15b,f are essentially indefinitely stable at ambient
temperature.
The barrier to rotation of o-methoxy substrate 15c was much
lower, ∆Gq ) 26.3 kcal/mol, and the R-chloro analog 19c
rot
was a little lower still at 25.3 kcal/mol. The 8.1 kcal/mol barrier
difference between 15c and 15b is significantly larger than
expected from related axially chiral systems. For example,
Sternhell’s model for effects of ortho-substituents in biphenyl
rotation32 predicts only a 2-4 kcal/mol difference. The half-
lives of the o-methoxy analogs are on the order of days at
ambient temperature. So rapid room temperature reactions can
be conducted with these substrates, but the resolved samples
must be stored in the cold.
The other samples with N-PMB groups instead of the N-Me
group did not have significant amounts of the Z rotamer at
equilibrium according to analysis of their 1H NMR spectra (E/Z
Chirality Transfer in 6-Exo-trig Cyclizations. Enantioenriched
samples of 15a-c,e-f,h-i were subjected to optimized reaction
conditions for reductive cyclization. The cyclized products were
(30) Stewart, W. E.; Siddall, T. H. Chem. ReV. 1970, 70, 517–551.
(31) (a) Chupp, J. P.; Olin, J. F. J. Org. Chem. 1967, 32, 2297–2303. (b)
Ototake, N.; Taguchi, T.; Kitagawa, O. Tetrahedron Lett. 2008, 49,
5458–5460. (c) Nobutaka, O.; Masashi, N.; Yasuo, D.; Haruhiko, F.;
Osamu, K. Chem.sEur. J. 2009, 15, 5090–5095.
(32) Bott, G.; Field, L. D.; Sternhell, S. J. Am. Chem. Soc. 1980, 102,
5618–5626.
9
15496 J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009

Radical Cyclizations of R-Halo-ortho-alkenyl Anilides
A R T I C L E S
Table 3. Yields and Chirality Transfer Levels in Cyclizations of Enantiomeric Precursors 15 to Give 14a
entry
precursor
R1
R2
R3
R4
R5
er
% yield 14b
er
% cte
1
2
3
4
5
6
7
8
(+)-15a
(-)-15a
(+)-15b
(-)-15b
(+)-15c
(-)-15c
(+)-15e
(-)-15e
(P)-15f
(M)-15f
(+)-15h
(-)-15h
(+)-15i
(-)-15i
Me
Me
Me
Me
Me
Me
OMe
OMe
TMS
TMS
Br
Br
Me
Me
Me
Me
Me
Me
Me
H
CO2t-Bu
CO2t-Bu
CO2t-Bu
CO2t-Bu
CO2t-Bu
CO2t-Bu
CO2t-Bu
CO2t-Bu
CO2t-Bu
CO2t-Bu
CN
H
H
H
H
H
H
H
H
H
H
H
H
Me
Me
98.5/1.5
85/15
100/0
100/0
99/1
79
81
87
85
94
96
95
93
63
66
71
68
57
55
91/9
84/16
95/5
96/4
93/7
93/7
95/5
94/6
95/5
94/6
93/7
96/4
80/20
80/20
92
99
95
96
94
94
95
96
96
94
93e
97e
81
80
PMB
PMB
PMB
PMB
PMB
PMB
PMB
PMB
PMB
PMB
PMB
PMB
H
99/1
Me
Me
Me
Me
Me
Me
Me
Me
100/0
100/0
99/1
0/100
99.5/0.5
99/1
9
10
11
12
13
14
CN
Me
Me
99/1
0/100
Me
a Conditions: Bu3SnH and Et3B in 20 mL PhH were added over 2 h via syringe pump to a stirred 10 mM PhH solution of 15. b Yield of 14 after
isolation by column chromatography on 10% w/w KF/silica gel. c Absolute configuration determined by X-ray crystallography. d Percent chirality
transfer. e Neat Bu3SnH and Et3B were added sequentially in one portion to a stirred PhH solution of iodide with [Bu3SnH]i ) 5 mM.
isolated by column chromatography over KF-doped silica, and
enantiopurities were determined by chiral HPLC with reference
to racemic standards of the product 3,4-dihydroquinolin-2-ones.
The results of this series of experiments are shown in Table 3.
Both enantiomers of each substrate were tested, and provided
identical results within experimental error. Isolated yields of
each product were comparable to yields of reactions with
racemic substrate (Table 1). The precursor and product enan-
tiomers are correlated in Table 3 by sign of optical rotation,
except for 15f/14f, whose absolute configurations are known
(see below). Correlation by elution order in the chiral HPLC is
provided in the Supporting Information. Unfortunately, there
was no obvious correlation between the optical rotation or HPLC
elution order of substrates and the respective products of
cyclization, so assigning absolute configurations by analogy is
not straightforward.
Briefly, the levels of chirality transfer were outstanding in
almost all cases, ranging from 92 up to 99%. These high levels
were not changed by altering the N-substituent (Me or PMB)
or the ortho substituent (Me, Br, TMS). Even the samples of
15c with the o-methoxy group cyclized with excellent chirality
transfer (94%). The only exception was 15i, which gave 14i
with a chirality transfer of 80-81% (Entries 13-14). This
substrate differs from the others because it lacks an activating
group on the alkene and because its R4 group is Me, not H.
Either difference could be responsible for the erosion, as
discussed below. These values of chirality transfer in the first
examples of “Approach 2” cyclizations are comparable to and
usually better than those observed in “Approach 1” room
Figure 8. X-ray crystal structures of (P)-24 (top) and (S)-14f (bottom,
with front and side views).
temperature cyclizations of N-allyl-o-iodoacetamides (74-97%)
and N-acryloyl-o-iodoanilides (49-94%).
The substrate/product pair 15f/14f incorporating an ortho-
bromine atom was designed to allow assignment of absolute
configuration by X-ray crystallography. Unfortunately 15f did
not cooperate. With the goal of making crystalline salts, a sample
of (+)-15f (99/1 er) was treated with TFA in CH2Cl2 producing
R,ꢀ-unsaturated acid 24 in 71% yield (Figure 8). Salt formation
became unnecessary when that acid itself crystallized by slow
vapor diffusion method.
Acid 24 crystallized with four unique molecules in the unit
cell, as two pairs of dimers displaying hydrogen bonding
between the carboxylic acid groups. The differences in geom-
9
J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009 15497

A R T I C L E S
Guthrie et al.
Scheme 2. Synthesis and Separation of Two Diastereomers of
Precursor 25
Figure 9. Model for chirality transfer illustrated with 15f.
Likewise, the key radical SOMO and the alkene LUMO orbitals
are also nearly orthogonal in this ground state of the radical.
Coordinated rotation of the N-aryl bond, the aryl-alkenyl bond
and the CO-CH2• bond brings the radical to a twisted transition
state TS-16f. This in turn progresses to the cyclized radical (not
shown), and then the cyclized product (S)-14f after Bu3SnH
reduction.
Production of the enantiomeric product (R)-14f from radical
GS-16f requires either that the N-Ar or Ar-alkenyl bond rotates
the other way (or alternatively, rotates on past the TS). The
former rotation seems unlikely due to the very high barrier. The
latter is disfavored because the alkene CdC bond begins to
eclipse the N-Ar bond.
In the model shown in Figure 9, the twisting about the NC(O)-
CH2• bond has no stereochemical consequence because a new
stereocenter is not formed at the radical carbon atom. That
changes if one of the two hydrogen atoms, labeled He and Hz
in Figure 9, is another group. R-Amidoyl radicals of the general
structure NC(O)-CHR• prefer the s-cis orientation rather than
s-trans.35 In other words, an R group that is appended to radical
GS-16f alpha to the carbonyl group will take the position of
He, not Hz. The model TS-16f then predicts that a diastereose-
lective cyclization should occur to give a trans-3,4-disubstituted-
3,4-dihydroquinolin-2-one.
Single and Tandem Cyclizations of 2-Substituted-r-Amidoyl
Radicals. To test the prediction of diastereoselectivity, we
prepared precursor 25 by acylation of PMB-aniline 18c with
racemic 2-bromopropanoyl bromide, as shown in Scheme 2.
Precursors 25 have both a chiral axis and a stereocenter, so two
racemic diastereomers were produced. These proved to be easily
separable by column chromatography and were isolated in yields
of 33 and 65%. Each diastereomer was fully characterized, but
we could not unambiguously assign relative configurations.
Since both diastereomers gave the same products (see below),
this ambiguity did not prove to be important in the reaction
analysis.
etries between the four structures are small, so only one is shown
in Figure 8. In this structure, the plane of the aromatic ring is
almost completely orthogonal to the plane of the amide, with a
torsion angle of 83°. The alkene acceptor exists in the s-trans
conformation (an s-cis conformation would have significant
A-strain) and is twisted slightly out of planarity with the
aromatic ring (torsion angle ) 156°). Interestingly, the N-PMB
group is twisted to the same side of the molecule as the R,ꢀ-
unsaturated carboxylic acid, with their planes nearly parallel.
The 3.32-3.86 Å distance range between the two quasi-parallel
planes is close to ideal for π-π stacking.33 Notice that the PMB
group shields one face of the alkene. However, this shielding
cannot be important in the radical cyclization because the PMB
substituent is not essential for high chirality transfer. The
absolute configurations of all four molecules in the unit cell
were the same and were assigned by the anomalous scattering
method;34 24 and accordingly its precursor (+)-15f have the
(P) configuration.
The product 14f was more cooperative; both the racemate
and the pure enantiomers spontaneously solidified on standing.
A crystal of (+)-14fsthe product of cyclization of (+)-15fswas
grown by vapor phase diffusion and the structure was solved
as above; (+)-14f has the (S) configuration. Two views of this
structure are shown in Figure 8. The view on the left is oriented
to facilitate comparison with the structure of 15f. The view on
the right, looking into the side of the 3,4-dihydroquinolin-2-
one ring, shows how the serious 1,3-interaction between the
N-PMB group and the o-bromine atom is minimized. The angle
between these two groups cannot be 90° as in the precursor,
and it cannot be 0° as normally preferred for sp2 hybridized
atoms. Instead, a compromise of about 40° is reached because
the lactam ring adopts a boat-like conformation that twists the
N-Aryl bond. This twisting occurs without appreciable pyra-
t
midalization at N, and the CH2CO2 Bu substituent adopts the
less-crowded axial-like orientation.
Data for the reductive cyclizations of various isomers of 25
are summarized in Table 4. Racemic samples of 25 were reacted
with Bu3SnH and Et3B at room temperature according to the
usual syringe pump procedure. Both diastereomers gave the
single product trans-26 in excellent yield (95-97%). The trans
relative configuration was initially assigned based on analysis
These results validate a working model that is shown in Figure
9. Iodine abstraction by a tributyltin radical from (P)-15f
produces a ground state R-amidoyl radical GS-16f that has
roughly orthogonal amide and aryl planes as in the precursor.
(33) (a) Maddaluno, J. F.; Gresh, N.; Giessner-Prettre, C. J. Org. Chem.
1994, 59, 793–802. (b) Roesky, H. W.; Andruh, M. Coord. Chem.
ReV. 2003, 236, 91–119.
(35) Porter, N. A.; Giese, B.; Curran, D. P. Acc. Chem. Res. 1991, 24,
(34) Dunitz, J. D. Angew. Chem., Int. Ed. 2001, 40, 4167–4173.
296–304.
9
15498 J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009

Radical Cyclizations of R-Halo-ortho-alkenyl Anilides
A R T I C L E S
Scheme 3. Synthesis and Cyclization of Tandem Precursor 27
Table 4. Cyclizations of Secondary Radical Precursor 25 to 26
entry
substrate
er
product
% yielda
er
% chirality transfer
1
2
3
4
5
6
rac-25a
rac-25b
(+)-25a
(-)-25a
50/50 rac-26
50/50 rac-26
95
97
98
96
98
97
50/50
50/50
98/2
98/2
99/1
N/A
N/A
100
99
99
99.5
98/2
99/1
(+)-26
(-)-26
(-)-26
(+)-26
(+)-25b 100/0
(-)-25b 100/0
99.5/0.5
a Yield after isolation by column chromatography on 10% w/w KF/
silica gel.
of coupling constants (J3,4 ) 14 Hz), and this assignment was
later confirmed by crystallography (see below). The cis isomer
was not detected.
Next, both samples 25a,b were resolved by chiral HPLC to
high enantiopurity, and the four pure stereoisomers were
cyclized as usual (entries 3-6). Yields were reproducibly high
(96-98%), and chirality transfers were nearly perfect. The
values (99-100%) are even higher than those observed for the
primary R-amide radical precursor series. Each enantiomer in
a given diastereomeric pair gives a different enantiomer of the
product. Across the diastereomeric pairs, (+)-25a and (-)-25b
give the same enantiomer of the product (+)-26, so these
compounds must have the same configuration in the chiral axis
and opposite configurations at the stereocenter. Likewise for
the (-)-25a and (+)-25b pair; each gives (-)-26. These results
show that the stereoselectivity is dictated by the chiral axis and
not the stereocenter. In other words, the diastereomeric pairs
of bromides with the same chiral axis configuration provide the
same radical. This is sensible since the initially produced radical
has sp2 geometry.35,36
Radical cyclizations often proceed with high levels of
diastereoselectivity when the radical and alkene acceptor are
connected by a preformed ring, so we extended the design to
incorporate a second cyclization that incorporated new elements
of stereoselectivity.37 The synthesis and cyclizations of tandem
precursor 27a-b are shown in Scheme 3. Racemic 2-bromo-
4-pentenoic acid was converted to the acid chloride, which was
used to acylate 18c. This produced diastereomers 27a (35%
yield) and 27b (39% yield) that were separated by column
chromatography. Again, the relative configurations of 27a,b
were not assigned.
the experiments in Table 3, we cyclized one of the diastereomers
of 27 as a racemate, and resolved the other prior to cyclization.
Racemic 27a was treated with Bu3SnH under syringe pump
conditions at room temperature (Scheme 3). After column
chromatography, tricyclic product rac-28 was isolated in 51%
yield as a single diastereomer. Other fractions of side products
contained, among other unidentified compounds, two minor
diastereomers of the 6-endo-trig/5-exo-trig sequence, as evi-
denced by two distinct doublets in the upfield region denoting
the respective methyl signals of the diastereomers. Enantiopure
samples of 27b were accessed by resolution with chiral HPLC,
and were also subjected to the reaction conditions. These
reactions provided the same diastereomer but opposite enanti-
omers of 28 in similar yields and excellent chirality transfer
(98-99%).
Because the minor products and various impurities were
inseparable from each other, it was not possible to characterize
any of the secondary products or to provide an accurate measure
of the diastereoselectivity in the second cyclization. However,
we conservatively estimate that none of the minor products
exceeded 10% of the mixture, so the second cyclization must
also be reasonably diastereoselective (g4/1).
The structure of rac-28 was established by X-ray crystal-
lography, and a diagram is shown in Figure 10. The ring fusion
is trans,38 and J3,4 is again 14.1 Hz with a dihedral angle H3-
C3-C4-H4 of 176° in the crystal. This confirms the assignment
of trans relative configuration of 26 above.
Tandem cyclization of a single enantiomer of 27 can give as
many as 16 stereoisomers of 28; however, we already know
that racemization will not occur and that the first cyclization
should be highly diastereoselective for the trans isomer. This
reduces to four the number of likely product isomers. Since we
already understood the effect of the relative stereochemistry from
Models for the first and second cyclizations of the radicals
generated from 27 are shown in Figure 11. The model for the
first cyclization, which also serves for the radical derived from
25, follows directly from the model for the primary radical
(36) Interestingly, however, the rate of rotation about the NC(O)-CHMe(•)
may be similar to or possibly even slower than the rate of radical
cyclization. If that is the case, then both diastereomers of precursor
must produce the same s-cis radical. For rotation rate information,
see: (a) Strub, W.; Roduner, E.; Fischer, H. J. Phys. Chem. 1987, 91,
4379–4383. (b) Musa, O. M.; Choi, S. Y.; Horner, J. H.; Newcomb,
M. J. Org. Chem. 1998, 63, 786–793.
(37) Curran, D. P.; Porter, N. A.; Giese, B. Stereochemistry of Radical
Reactions: Concepts, Guidelines, and Synthetic Applications; VCH
Publishers, Inc.: New York, 1996; see especially pp 30-82.
Figure 10. X-ray crystal structure of tricycle 28.
9
J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009 15499

A R T I C L E S
Guthrie et al.
2-ones in high yield. The precursors are axially chiral (stable
atropisomers) but the products are not. Cyclizations of enan-
tioenriched precursors occur with transfer of axial chirality to
the new stereocenter of the products with exceptionally high
fidelity (often >95%). Single and tandem cyclizations of R-halo-
ortho-alkenyl anilides bearing an additional substituent on the
R-carbon occur with both high chirality transfer and high
diastereoselectivity. Overall, the new method is an attractive
route for stereoselective synthesis of diverse 3,4-dihydroquino-
lin-2-ones.
Crystal structures have been used to rigorously assign
configurations of one precursor and two products. Straightfor-
ward models interpret both the chirality transfer and diastereo-
selectivity aspects, reinforcing the notion that predictions of
stereochemical outcomes of reactions of axially chiral amides
are both easy and reliable.
Figure 11. Models of first and second cyclizations of radicals derived
from 27.
Prior reactions with axial chirality transfer in the anilide series
have featured reactive intermediates (radicals, anions, transition
metals) on the aryl ring of the anilide and acceptors in the amide.
This first-in-class example of the reverse approach suggests
broad potential for axial chirality transfer from assorted reactive
intermediates in the amide to acceptors on the aryl ring.
(Figure 9). The alkyl group on the radical center (R ) Me or
CH2CHdCH2) orients itself s-cis to the anilide carbonyl group.
Then, twisting as indicated in TS-29 directs the first cyclization
with control of both relative and absolute configuration. The
structure of the major isomer 28 formed in the second cyclization
follows directly from the Beckwith-Houk model39 of a chairlike
transition state TS-30 with equatorial-like substituents on the
forming ring. The minor products can arise from a chairlike
transition state with the radical substituent in a quasi-axial
orientation, or boat-like transition states with either axial or
equatorial orientations. In flexible systems, the chair and boat
can flip to provide additional possible geometries, but the rigidity
of the ring fusion to the 3,4-dihydroquinolone ring prevents that
flip in this case.
Acknowledgment. We thank the National Science Foundation
for funding this work and the National Institutes of Health for a
grant to purchase a NMR spectrometer. We dedicate this paper to
Dr. William V. Curran on the occasion of his 80th birthday.
Supporting Information Available: Contains procedures of
synthesis and characterization data of all new compounds and
copies of NMR spectra. This material is available free of charge
via the Internet at http://pubs.acs.org.
Conclusions
JA9066282
Radical cyclizations (Bu3SnH, Et3B/air, rt) of an assortment
of R-halo-ortho-alkenyl anilides provide 3,4-dihydroquinolin-
(39) Spellmeyer, D. C.; Houk, K. N. J. Org. Chem. 1987, 52, 959–974.
(a) Beckwith, A. L. J.; Schiesser, C. H. Tetrahedron Lett. 1985, 26,
373–376.
(38) Akritopoulou-Zanze, I.; Whitehead, A.; Waters, J. E.; Henry, R. F.;
Djuric, S. W. Tetrahedron Lett. 2007, 48, 3549–3552.
9
15500 J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009