Ring opening reactions of quinoline substituted epoxides

Article abstract of DOI:10.1016/j.tetlet.2003.10.062

6-Oxiranyl- and 3-oxiranyl-2-phenylquinoline-4-carboxylic acid diisopropylamides react with secondary amines and lithium amides to give (aminohydroxyethyl)quinolines but with opposite regioselectivities. Upon epoxidation of 3-formylquinoline 2 a ~5:1 mixt

Full text of DOI:10.1016/j.tetlet.2003.10.062

TETRAHEDRON
LETTERS
Pergamon
Tetrahedron Letters 44 (2003) 9299–9302
Ring opening reactions of quinoline substituted epoxides
Andrew N. Boa,* Stephen Clark, Paul R. Hirst and Robert Westwood†
Department of Chemistry, University of Hull, Cottingham Road, Hull HU6 7RX, UK
Received 4 September 2003; revised 3 October 2003; accepted 10 October 2003
Abstract—6-Oxiranyl- and 3-oxiranyl-2-phenylquinoline-4-carboxylic acid diisopropylamides react with secondary amines and
lithium amides to give (aminohydroxyethyl)quinolines but with opposite regioselectivities. Upon epoxidation of 3-formylquinoline
2 a ꢀ5:1 mixture of atropisomers is formed. This ratio is maintained upon epoxide ring-opening with amines in ethanol at reflux,
but with lithium amides at room temperature a 1.3:1 ratio of isomers is obtained.
© 2003 Elsevier Ltd. All rights reserved.
Brequinar sodium is a quinoline-4-carboxylate deriva-
tive originally described1 by researchers at DuPont
(Fig. 1). This molecule has been shown to be an
inhibitor of human dihydroorotate dehydrogenase
(DHOdehase), the fourth enzyme involved in the
biosynthetic pathway to pyrimidine nucleotides in
prokaryotic and eukaryotic organisms.2 As such, selec-
tive inhibitors of this enzyme have a wide range of
potential therapeutic applications3 although develop-
ment of brequinar as an anticancer agent was halted
due lack of efficacy at doses where toxic effects began
to appear. We have however been preparing and
screening brequinar-related compounds to probe the
active site of DHOdehase from a variety of organisms.4
were chosen to serve as one of several versatile precur-
sors to a wide range of 3- and 6-substituted quinolines.
Conversion of the aldehydes 1 and 2 to their corre-
sponding epoxides 3 and 4 proceeded cleanly using
dimethylsulphonium methylide in tert-butanol at 30°C
1
(Scheme 1). H NMR showed epoxide 3 to be a 1:1
mixture of diastereoisomeric conformers, distinguish-
able on the NMR timescale due to the restricted rota-
tion about the CArꢀCꢁO bond. The 1H NMR of
epoxide 4 however showed a mixture of isomers rang-
ing from about 5 to 6:1. In 2-substituted 1-naphthoic
acid diisopropylamides the rotation about the CArꢀCꢁO
bond is heavily restricted,6 so the related diisopropyl-
amide 4 can be considered to be a mixture of
atropisomers7 (diastereoisomeric conformers with a
half-life of 1000 s or more).
Figure 1. Brequinar sodium.
During the early part of this work we prepared the
aldehydes 1 and 2 via free radical bromination followed
by oxidation5 of the corresponding methyl substituted
2-phenylquinoline-4-carboxamide. These aldehydes
Keywords: regioselective; epoxide; amino alcohol; atropisomers.
* Corresponding author. Tel.: +44-1482 465022; fax: +44-1482-
466410; e-mail: a.n.boa@hull.ac.uk
Scheme 1.
Clayden6 has invoked a non-chelation controlled path-
way to explain the reactivity of 2-formylnaphthalene-1-
† Current address: Cyclacel Limited, James Lindsay Place, Dundee
Technopole, Dundee DD1 5JJ, UK; bwestwood@cyclacel.com
0040-4039/$ - see front matter © 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetlet.2003.10.062

9300
A. N. Boa et al. / Tetrahedron Letters 44 (2003) 9299–9302
ratio of diastereoisomeric conformers. Reaction of 3
with lithium amides at room temperature however gave
exclusively the regioisomer 6. Unusually the reaction of
3 with lithium diethylamide was unsuccessful, and start-
ing material was recovered in good yields, even after 24
h. Structural assignments of 5 and 6 were confirmed by
correlating literature 1H NMR data of aryl b-amino
alcohols.9 Additionally 6b was oxidised slowly (PCC,
DCM, rt) to give an unstable aldehyde (HCꢁO, l 10.15
in CDCl3) as opposed to the ketone that would be
expected from 5b.
carboxamides with organometallic reagents under non-
chelation control, with attack of the nucleophile on the
aldehyde in either the s-cis or s-trans conformation but
always anti to the isopropyl groups. The absence of
chelating metals in our epoxidation reaction would
suggest that ours is a comparable reaction. Molecular
modelling8 indicated that the two conformations of the
aldehyde, s-trans-2 and s-cis-2, were of very similar
energy (Fig. 2). Assuming the sulphur ylid attacks
either conformation of the aldehyde equally well (on
the face opposite to the diisopropyl group), it is not
immediately clear therefore why a ꢀ5:1 ratio of prod-
ucts is obtained. In order for a >1:1 isomer ratio to be
obtained we presume the initial addition of the sulphur
ylid must be reversible allowing the betaines to revert to
the aldehyde and ylid. The steric hindrance from the
diisopropyl and phenyl substituents would suggest that
one of the betaine intermediates requires less energy to
adopt the anti-conformation needed for epoxide forma-
tion and so eliminates dimethyl sulphide more quickly
than the other resulting in the ꢀ5:1 epoxide ratio
observed.
A report from Singaram and co-workers described
reactions of styrene oxide with lithium amides which
gave only the 2-dialkylamino-1-phenylethanols,10 rather
than the opposite regiosiomer that we observed. A
mechanism involving attack of the nucleophile via ini-
tial addition to the carbonyl group then transfer to the
epoxide might explain this difference, though we saw
no evidence of products in which the diisopropyl amide
had been replaced, for example, by a morpholinyl
amide. Variable trace lithium halides present in the
butyl lithium used to generate the lithium amide could
be the source of this dramatic reversal, though Harris
and co-workers made and used lithium amides in situ
from butyl lithium as well. Chini and co-workers11
described that heating styrene oxide with diethylamine
in ethanol gave a similar regiochemical result to our
reactions under the same conditions, and that addition
of various soluble lithium salts to the reaction can
begin to reverse the regiochemical trend, though not as
dramatically as in our case. The participation of the
lithium ion coordinating the epoxide oxygen and an
incipient benzylic carbocation was used to explain this
trend. This same reason can be invoked to explain our
result, though the degree of regioselectivity (regiospe-
cificity) is still rather surprising and remains, as yet,
unexplained. The presence of the carboxamide function
is clearly a main difference in the substrates and so
further investigations will be necessary to make firmer
conclusions on the generality of these results.
Figure 2. Conformations of aryl aldehyde 2.
Our reason for preparing the epoxides 3 and 4 was for
them to serve as precursors to hydroxyethylquinolines
through nucleophilic ring opening of the epoxides. In
the case of epoxide 3, reaction with secondary amines
in ethanol gave the expected 6-(2%-amino-1%-hydroxy-
ethyl)quinolines 5 along with the opposite regioisomer 6
as a minor product (ꢀ10:1) (Scheme 2). The regioiso-
mers 5 and 6 were separated easily by column chro-
1
matography, and in each case H NMR showed a 1:1
In transferring the reactions from 3 to 4 further obser-
vations concerning the reactivity of these epoxides
towards amines and lithium amides were made (Scheme
3). Taking a 5:1 mixture of isomeric epoxides 4, the
Scheme 2. Reagents and conditions: Pathway A: R2NH,
EtOH, heat, 2 h: 5a:6a=12:1; 5b:6b 10:1; 5c:6c 8:1. Pathway
B: R2NLi, THF, rt, 0.5 h: 5:6=0:1 in all cases.
Scheme 3. Reagents and conditions: Pathway A: R2NH,
EtOH, heat, 4–16 h: 7:8=1:0 in all cases. Pathway B: R2NLi,
THF, rt, 1 h: 7:8=0:1 in all cases.

A. N. Boa et al. / Tetrahedron Letters 44 (2003) 9299–9302
9301
reaction with all three secondary amines in ethanol
gave the amino alcohols 7a–c, this time exclusively as
might be expected due to the increased steric hindrance
at the reaction centre. The 5:1 diastereoisomeric ratio
appeared to be retained in the product amino alcohols
as shown by 1H NMR of the crude mixtures. The major
isomer of amino alcohol 7b was purified from its
atropisomer by fractional crystallisation and single
crystal X-ray diffraction12 proved the regiochemical
path of ring opening. However, in one sample of epox-
ide 4 where the product was initially isolated as a 5.9:1
mixture of atropisomers, heating in toluene at reflux in
the absence of a secondary amine for 30 min reduced
this ratio to 2:1. Taking this 2:1 isomeric mixture
through to the amino alcohol 7b by treatment with
is highly unlikely that the atropisomers of amino alco-
hol 8 interconvert under these conditions. This result
therefore lends credence to the idea that a benzylic
carbocation is involved in the reaction under these
conditions. In this scenario the lithium ion can coordi-
nate both epoxide and carbonyl oxygen and this would
explain both the regiospecificity of the transformation
and the reduction in atropisomeric ratio of the product.
Conclusion
In conclusion, we report some interesting stereoselective
reactions of atropisomeric epoxides in which the regio-
chemistry of ring-opening is determined by selection of
either an amine or lithium amide. These reactions
should be readily extended to aldehydes and ketones in
both benzamide and naphthamide series, where the
conformational preferences may be much more dis-
tinct.14 Given that aryl amino alcohols and related
systems are found often as ligands in metal catalyst
systems and in biologically active molecules, the chem-
istry reported herein could ultimately provide useful
methodology for the stereoselective preparation of such
molecules.
1
morpholine under the conditions described above, H
NMR now indicated that the 2:1 diastereoisomeric
ratio of the starting material was converted back to a
ꢀ5:1 ratio in the product. When the purified major
atropisomer of 7b from the crystallisation study was
heated at reflux in ethanol for 6 h, subsequent evapora-
tion of the solvent and analysis of the resulting material
indicated that no isomerism had taken place. This
indicates that the 5:1 ratio observed in the product
amino alcohol must represent the kinetic atropisomeric
product ratio of the amino alcohols 7.
From the X-ray crystal structure of 7b the relative
stereochemistry of the stereogenic axis and centre (Fig.
3) of the major atropisomer was also identified. How-
ever, due to the fact that isomerisation of the epoxide
takes place so readily in toluene at reflux, this crystal
structure sheds no further light on the stereoselectivity
of the epoxidation reaction.
Acknowledgements
We thank the Engineering and Physical Sciences
Research Council (UK) for financial support of this
work by providing funds for the purchase of a new
X-ray diffractometer and a studentship (P.R.H.).
References
1. Dexter, D. L.; Hesson, D. P.; Ardecky, R. J.; Rao, G. V.;
Tippett, D. L.; Dusak, B. A.; Paull, K. D.; Plowman, J.;
DeLarco, B. M.; Narayanan, V. L.; Forbes, M. Cancer
Res. 1985, 45, 5563.
2. Jones, M. E. Annu. Rev. Biochem. 1980, 49, 253.
3. Batt, D. G. Exp. Opin. Ther. Patents 1999, 9, 41.
4. Boa, A. N.; Hirst, P. R.; Fishwick, C. W. G.; McConkey,
G. A.; Ramsey, C. G. C.; Stead, A. M. W., manuscript in
preparation.
5. Hass, H. B.; Bender, M. L. J. Am. Chem. Soc. 1949, 71,
1767.
6. Clayden, J.; McCarthy, C.; Westlund, N.; Frampton, C.
S. J. Chem. Soc., Perkin Trans. 1 2000, 1363 and refer-
ences cited therein.
7. Oki, M. Topics Stereochem. 1983, 14, 1.
8. Molecular mechanics (MM2) modelling studies were per-
formed with CS Chem3D Pro v 5.0 CambridgeSoft
Corporation).
Figure 3. ORTEP view13 of the amino alcohol 7b with a
water of crystallisation (50% probability ellipsoids and
selected hydrogen atoms omitted for clarity).
Reaction of 4 (5:1 atropisomer ratio) with the lithium
amides again gave the regioisomer 8 exclusively, as
might have been expected based on the precedence of
epoxide 3. However, the atropisomeric mixture in this
case was only 1.3:1. Given that these reactions were
conducted at room temperature over a short period, it
9. The chemical shift of the benzylic proton in 5 and 7 occur
in the range l 4.81–5.00, whereas in 6 and 8 it was l
3.91–4.16 (in CDCl3). These values are consistent with
data from numerous examples of phenylethyl amino alco-
hols in the literature, for example those that can be found
in Refs. 10 and 11.

9302
A. N. Boa et al. / Tetrahedron Letters 44 (2003) 9299–9302
10. Harris, C.; Fisher, G. B.; Beardsley, D.; Lee, L.; Goral-
ski, C. T.; Nicholson, L. W.; Singaram, B. J. Org. Chem.
1994, 59, 7746.
11. Chini, M.; Crotti, P.; Macchia, F. J. Org. Chem. 1991,
56, 5939.
12. Crystal data for 7b: Colourless crystal (0.26×0.25×0.12
mm) from methanol–water, coated with perfluoro-
polyether oil mounted on a glass fibre, C28H37N3O4,
M=479.61, monoclinic, space group P21/c, a=
Dc=1.217 g cm−3, v(Mo-Ka)=0.082 mm−1, 24450 reflec-
tions measured, 7383 unique (Rint=0.0912) which were
used in all calculations. R1=0.0383 on F>4|(F), wR2=
0.0837 on all data. Crystallographic data (excluding
structure factors) have been deposited with the Cam-
bridge Crystallographic Data Centre, reference CCDC
220980.
13. ORTEP3 for Windows: Farrugia, L. J. J. Appl. Crystal-
logr. 1997, 30, 565.
14. Clayden, J.; Westlund, N.; Beddoes, R. L.; Helliwell, M.
,
9.6304(14), b=25.463(4), c=11.1929(17) A, i=
3
J. Chem. Soc., Perkin Trans. 1 2000, 1351.
,
107.481(11)°, V=2618.0(7) A , T=150(2) K, Z=4,