Stable annelated chiral NADH models with a rigidified amide part in the quinoline series: Synthesis, reactivity and grafting on a Merrifield resin

Article abstract of DOI:10.1016/S0040-4020(01)00166-1

The synthesis of new chiral nicotinamide adenine dinucleotide hydrogenated models derived from quinoline is described. Using a biomimetic approach, the out-of-plane positioning of the amide carbonyl was obtained by involving the chiral auxiliary in a lactam structure. It is shown that electron-donating groups on the benzene ring of the quinoline structure are necessary to obtain high chemical yields during the reduction of methyl benzoylformate. An interesting variation of the enantioselectivity as a function of magnesium ion concentration has been observed. Under the best conditions, methyl mandelate was obtained in up to 95% ee (R). To facilitate the recycling of these models, grafting of reagent 4 on a Merrifield resin has been developed. The resulting polymer-supported reagent 4 was tested in the asymmetric reduction of methyl benzoylformate.

Full text of DOI:10.1016/S0040-4020(01)00166-1

TETRAHEDRON
Pergamon
Tetrahedron 57 52001) 3087±3098
Stable annelated chiral NADH models with a rigidi®ed amide part
in the quinoline series: synthesis, reactivity and grafting on a
Merri®eld resin
p
Christiane Vitry, Jean-Luc Vasse, Georges Dupas, Vincent Levacher, Guy Queguiner and
Jean Bourguignon
Â
  Â
Laboratoire de Chimie Organique Fine et Heterocyclique associe au CNRS, INSA-IRCOF, B.P. 08,
F-76131 Mont Saint Aignan Cedex, France
Received 20 November 2000; accepted 8 February 2001
AbstractÐThe synthesis of new chiral nicotinamide adenine dinucleotide hydrogenated models derived from quinoline is described. Using
a biomimetic approach, the out-of-plane positioning of the amide carbonyl was obtained by involving the chiral auxiliary in a lactam
structure. It is shown that electron-donating groups on the benzene ring of the quinoline structure are necessary to obtain high chemical yields
during the reduction of methyl benzoylformate. An interesting variation of the enantioselectivity as a function of magnesium ion concentra-
tion has been observed. Under the best conditions, methyl mandelate was obtained in up to 95% ee 5R). To facilitate the recycling of these
models, grafting of reagent 4 on a Merri®eld resin has been developed. The resulting polymer-supported reagent 4 was tested in the
asymmetric reduction of methyl benzoylformate. q 2001 Elsevier Science Ltd. All rights reserved.
1. Introduction
features of the coenzyme seem to play a crucial role in the
stereochemical outcome of hydride transfer. The amide car-
bonyl of the 1,4-dihydronicotinamide part would be cis with
respect to the hydrogen atoms at C4.3 Moreover, this amide
carbonyl would be out of the plane of the dihydropyridine
structure and syn-oriented with respect to the transferred
hydrogen.4 We synthesised chiral models bearing an
aminoalcohol as chiral auxiliary where the amide function
is involved in a cyclic structure to mimic these conforma-
tional features of the coenzyme 5Fig. 1).5
For several decades now, nicotinamide adenine dinucleotide
hydrogenated 5NADH) models have been the focus of
considerable interest. These biomimetic models were ®rst
studied to throw light on the biochemical behaviour of the
coenzyme. Later models were also used for synthetic
purposes in chemoselective reactions. A further step
concerned the elaboration of chiral models, which could
be used in asymmetric reductions.1 Our group is involved
in these directions.
Following this biomimetic approach, it emerged that model
1 in the naphthyridine series5 and model 2 in the pyrido[3,2-
c]azepin series6 afforded the best results in terms of enantio-
selectivity during the reduction of methyl benzoylformate
5Scheme 1).
A family of ef®cient chiral reagents2 has been designed on
the basis of the following consideration. Two structural key
However, NADH models are subjected to pernicious side
reactions essentially caused by the effect of trace water on
Figure 1. 5a) Conformational analysis of the amide group in the coenzyme;
5b) biomimetic approach in the design of new chiral NADH models.
Keywords: biomimetic NADH model; asymmetric reduction; grafting;
Merri®eld resins.
p
Corresponding author. Tel.: 133-2-35-52-24-85; fax: 133-2-35-52-29-
62; e-mail: vlevache@ircof.insa-rouen.fr
Scheme 1. 5i) 51) Mg5ClO4)2/CH3CN/rt, 52) H2O.
0040±4020/01/$ - see front matter q 2001 Elsevier Science Ltd. All rights reserved.
PII: S0040-4020501)00166-1

3088
C. Vitry et al. / Tetrahedron 57 12001) 3087±3098
Scheme 3. Asymmetric reduction of methyl benzoylformate with model 3.
lactam 8 in an overall yield of 52%. The later was ®nally
converted into the desired target model 3 in two steps with
an overall yield of 70% 5Scheme 2).
Reduction of methylbenzoylformate with model 3 gave the
5R)-methyl mandelate with 84% ee and only 50% chemical
yield 5instead of 88 and 95%, respectively, with model 1
under the same conditions). As can be seen, annelation of
model 1 does not disturb the degree of enantioselectivity of
the hydride equivalent transfer. However, the chemical
yield was very much lowered. Loss of reactivity may be
ascribed to the lack of electron donating groups on the
annelated benzene ring, which would favour the hydride
equivalent departure 5Scheme 3).
Figure 2. Design of a new class of chiral NADH models.
the fragile 5,6 double bond.7 To tackle this problem, our
group has developed a new generation of models where
this bond is protected by annelation with an aromatic
ring.8 We wish to report in the present paper the synthesis
of annelated reagents 3 and 4 which are expected to
combine good stability and high degree of stereoselectivity.
With intend to facilitate the recycling of these models, we
developed the grafting of reagent 4 on a Merri®eld resin
5Fig. 2).
One may expect that increasing the electron donating effect
of the benzene ring would improve the reducing properties
of the model. To pursue this idea, the synthesis of a similar
model with a benzene ring bearing methoxy groups was
conceptualised using the same synthetic strategy. Unfortu-
nately, this approach failed at the addition reaction step of
the amino alcohol to the ethynyl derivative 6. In view of this
failure, we turned our interest to the preparation of model 4
designed so as to maintain the structural features de®ned for
the chiral auxiliary and to exhibit greater reactivity owing to
the presence of methoxy groups on the annelated benzene.
2. Result and discussion
2.1. Preparation of reagent 3: asymmetric reduction of
methyl benzoylformate
2.2. Preparation of reagent 4: asymmetric reduction of
methylbenzoylformate
Reagent 3 was synthesised using a route similar to that
employed for the synthesis of model 1.5 Cross-coupling of
quinoline 5 with trimethylsilylacetylene catalysed by
Pd25PPh3)2 and subsequent cleavage of the trimethylsilyl
group afforded the ethynyl derivative 6 in 69% yield. Addi-
tion of 5S)-phenylalaninol followed by reduction of the
enamine intermediate gave amine 7, which on treatment
with EtOH/H2O at re¯ux undergoes cyclization to produce
Our retrosynthetic analysis is based on a Friedlander type
condensation9 of 2-amino-4,5-dimethoxybenzaldehyde 910
with an appropriate b-ketoester 10. The crux of this reaction
scheme is the realisation of b-ketoester 10. Lactam 17 can
be accessed fromketal 12 by reductive amination of the
corresponding deprotected aldehyde followed by cycliza-
tion of the amino ester intermediate 5Scheme 4).
Among the numerous methods available11 for the synthesis
of b-ketoester derivatives, our choice was the De Milo
synthesis11e relying upon condensation of the dianion of
ethyl acetoacetate with the iodoketal 11 5Scheme 5).
Attempts to use the more available bromo derivative in
this condensation were unsuccessful leading in all cases to
the starting materials. This probably is due to the decreased
leaving ability of the bromine atom.12 Despite our efforts to
get high and reproducible yields of the b-ketoester 10, we
were only able to realise modest yields, 20±62% range
under the conditions described in Scheme 5. It was found
that the reproducibility of this reaction was highly depen-
dent on the quality of HMPA.
Thereafter, we also explored the Friedlander condensation
with the isolated amine 9 in the presence of Na2CO3 in
ethanol which afforded the desired quinoline in only 40%
Scheme 2. 5i) 51) Trimethylsilylacetylene/CuI/NEt3/PdCl25PPh3)2, 52)
OH2; 5ii) 51) 5S)-phenylalaninol, 52) NaBH4; 5iii) EtOH/H2O 595:5); 5iv)
51) MeI, 52) Na2S2O4.

C. Vitry et al. / Tetrahedron 57 12001) 3087±3098
3089
Scheme 4. Retrosynthetic analysis of model 4.
579%) was condensed with b-ketoester 10 to produce the
required quinoline 12a in high yield 581%) 5Scheme 6).
The last part of the synthesis concerns the construction of
the lactam moiety 5Scheme 7). The chemoselective depro-
tection of the ketal was realised with formic acid16 followed
by reductive amination.17 The resulting crude amine 16a
was utilised in the lactamization step without further puri-
®cation. When cyclization was carried out under basic
conditions, lactam 17a was obtained in poor yields 5entries
1 and 2). Alternatively, the use of a large excess of Me3Al18
promoted the formation of the corresponding aluminium
amide which was subsequently cyclized to give lactam
17a in greater than 95% yield 5entry 3).
Scheme 5. 5i) 51) NaH 51 equiv.)/n-BuLi, 52) HMPA; 5ii) ICH2CH5OEt)2
511).
yield. The amine 9 was previously obtained with a modest
yield 535%) by reduction of the corresponding nitro deriva-
10
tive with sodiumdithionite.
Attempt to use a one pot
procedure fromthe nitroaldehyde did not improve the over-
all yield of this sequence. The poor stability of ortho-amino-
benzaldehydes owing to the occurrence of autocondensation
reactions13 prompted us to test the Borsche modi®cation.14
In a ®rst step, the nitroaldehyde was converted into a nitro-
imine 13a 581%) and subsequently reduced by Na2S15 to
furnish imine 14a 579%). The stable amino imine 14a
The remaining steps for the transformation of 17a to the
target reagent 4 are shown in Scheme 8. Although, quater-
nization of quinoline 8 proceeded smoothly with methyl
iodide in acetonitrile, under the same conditions, quinoline
Scheme 6. 5i) Na2S2O4/Na2CO3; 5ii) Na2CO3/10/EtOH±H2O 51:1); 5iii) p-toluidine/EtOH/re¯ux; 5iv) Na2S/EtOH; 5v) 10/piperidine/EtOH/re¯ux.
Scheme 7. 5i) HCOOH/rt/2 h; 5ii) 51) 5S)-pheynylalaninol/EtOH, 52) NaBH4/rt; 5iii) see Table 1.

3090
C. Vitry et al. / Tetrahedron 57 12001) 3087±3098
Table 1. Asymmetric reduction of methyl benzoylformate with model 4
Furthermore, we studied the potential of model 4 as an
asymmetric reducing agent. Reduction with NADH models
are generally performed in the presence of magnesium ions,
which play a fundamental role in the hydride equivalent
transfer.1 A ternary complex would be established between
the model, Mg21 ions and the substrate 5generally methyl
benzoylformate). Previously, we have suggested the ternary
complex depicted in Fig. 3 to account for the enantioselec-
tion observed during reduction of methyl benzoylformate.
In this complex, Mg21 ions would be highly coordinated
with the amide and alcohol functions of the model and
with the ketone and ester functions of the substrate to give
rise to 5R)-methyl mandelate.5
Entry
[Mg21
]
Conversion %a
ee%b
1
2
3
4
5
6
7
8
0
0
67
±
0.25
0.5
0.75
1
2
3
92 5R)
95 5R)
89 5R)
86 5R)
73 5R)
66 5R)
60 5R)
100
100
100
100
100
100
4
With respect to model 4, the presence of methoxy groups on
the annelated benzene ring offer additional sites for chela-
tion, which can compete with the amide and alcohol func-
tions in coordination with Mg21. Consequently, it was of
interest to test model 4 in the presence of various amounts of
Mg21 ions, the results are summarized in Table 1. The best
degree of stereoselectivity 5entry 3) is as high as that
obtained with model 2. As expected, addition of two
electron-donating groups improved the reactivity of the
model. The best enantioselectivity was observed at
0.5 equiv. of magnesium ions 5entry 3) and decreased to
60% ee with 4 equiv. of magnesium ions 5entry 8). There
has been very little mentioned in literature on variation of
the enantioselectivity as a function of magnesium ion
concentration.19 The phenomenon was not observed with
model 2 and strongly suggests an important contribution
of the methoxy groups in the stereochemical outcome of
the reduction at high Mg21 concentration. This phenomenon
may be ascribed to a competitive complexation of Mg21
with the methoxy groups leading to a poor enantioselective
ternary complex at higher Mg21 concentration. Interest-
ingly, the quinoliniumsalt 18a formed during the reaction
could be recovered in good yield. After reduction with
a
Calculated from 1H NMR spectrum.
Determined by HPLC with a Daicel Chiralcel OD column.
b
Scheme 8. 5i) MeOTf/CH2Cl2/rt/1 h; 5ii) Na2S2O4/Na2CO3/CHCl3/H2O.
sodiumdithionite, the recycled model
4 exhibited the
same performance in terms of reactivity and stereo-
selectivity during the reduction of methyl benzoylformate.
The high stability of model 4 makes this reagent a good
candidate to develop the preparation of polymer-supported
NADH model with intent to facilitate their recycle. In a ®rst
approach, we have focused on the grafting of these stable
NADH models on insoluble Merri®eld resins.
Figure 3. Proposed ternary complex 5model/Mg21/substrate) of models
bearing aminoalcohol as chiral auxiliary.
17a afforded quinolinium 18a along with tarry material.
Various attempts to isolate 18a fromthe crude product
were unsuccessful. The reaction of 17a with other methyl-
ating agents was examined. Similarly, after two weeks in
re¯uxing acetonitrile in the presense of methyl p-toluene-
sulfonate as methylating agent, the 1H NMR spectrumof the
crude product revealed the presence of quinoliniumsalt 18a
which was then isolated after ¯ash chromatography in 25%
yield together with some tarry material. However, we were
able to improve both the reaction time and the yield using
the strong methylating agent methyl tri¯uoromethanesulfo-
nate, which afforded quinolinium 18a in 42% yield after 1 h
at roomtemperature. The quantitative and regioselective
reduction of 18a was realised by sequential addition of a
large excess of sodiumdithionite and sodiumcarbonate
leading to model 4 in almost quantitative yield 5Scheme 8).
2.3. Preparation of polymer-supported reagent 4
Polymer-supported NADH models have been previously
reported by our laboratory20 and others.21 The methodol-
ogies described in the literature are mostly based on quater-
nization of the pyridine nitrogen using chloromethylated
polystyrenes 5Scheme 9, route a). Due to steric hindrance
at the nitrogen in compound 17a, signi®cant decrease effect
in reactivity towards the quaternization step was observed.
The poor reactivity of 17a led us to develop a new route to
attach reagent 4 on a Merri®eld resin. Given that phenol
derivatives have been shown to react readily with chloro-
methylated polystyrenes,22 an alternative strategy would
therefore consist in reacting the phenolic compound 19
with Merri®eld resins prior to quaternization and reduction
steps 5Scheme 9, route b).

C. Vitry et al. / Tetrahedron 57 12001) 3087±3098
3091
Scheme 9.
Consequently, we undertook the synthesis of 19 following
the same reaction sequence as that employed for the
preparation of compound 17a 5Scheme 10). The imine
13b was obtained in 84% yield by reacting the correspond-
ing ortho-nitrobenzaldehyde with p-toluidine. The phenol
function, having been protected as its benzyl ether, is
expected to be stable in the reaction sequence delineated.
Reduction of 13b was accomplished in 98% yield.
Friedlander condensation of the resulting amine 14b with
b-ketoester 10 occurred smoothly under mild conditions
leading to the quinoline 12b in 77% yield, which was subse-
quently treated with formic acid to generate the aldehyde
15b in almost quantitative yield. Reductive amination of
15b with 5S)-phenylalaninol afforded 16b in 95% overall
yield. Cyclization of 16b afforded the desired lactam 17b
in 72% yield. Hydrogenolytic cleavage of the O-benzyl
moiety furnished 19 in 95% yield.
Compound 19 was reacted with Merri®eld resin 51% DVB,
foˆ2±2.5 mmol) in DMF for 4 days at room temperature
mediated by K2CO3 to yield the functionalised resin 20 with
a loading of 0.74 mmol of compound 19/g 5estimated by
elementary analysis) 5Scheme 11).
To obtain further structural elucidation, a 13C gel-phase
NMR spectrumof the resin 20 was determined. The
spectrumwas compared to that of compound
clearly indicated that the grafting was successful at the
17b, which
phenol function 5Fig. 4).
The functionalised resin 20 was then treated with methyl
tri¯uoromethanesulfonate in CH2Cl2 for 8 h at room
temperature. Although the 13C gel-phase NMR experiments
Scheme 10. 5i) p-toluidine/EtOH/re¯ux; 5ii) Na2S/EtOH; 5iii) 10/piperi-
dine/EtOH/re¯ux; 5iv) HCOOH/rt/2 h; 5v) 51) 5S)-pheynylalaninol/EtOH,
52) NaBH4/rt; 5vi) NaOEt/EtOH/re¯ux/24 h; 5vii) H2/Pd±C/MeOH/24 h/rt.
Scheme 11. 5i) Merri®eld resin 51% DVB; 2±2.5 mmol of chloromethyl
groups/g of resin)/DMF/K2CO3/4 days.

3092
C. Vitry et al. / Tetrahedron 57 12001) 3087±3098
Figure 4. 13C gel-phase NMR spectra of resin 20 in comparison with 13C NMR spectra of 17b in CDCl3.
with the resulting resin 21 gave poor quality spectra, none-
theless some useful information were clearly distinctive. It
was observed that the carbonyl lactamchemical shift in
resin 21 5166.5 ppm) is much more shielded than that of
resin 20 5172 ppm). Given that the same shielding is
observed in compound 18a 5166.7 ppm) during the quater-
nization of compound 17a 5171.6 ppm), it can then be
concluded that under those conditions, quaternization
proceeded with resin 20. Supporting evidence for quater-
nization may be deduced from the 19F gel-phase NMR
spectrumof resin 21 which features a single signal at
278.5 ppmcorresponding to quinoliniumtri¯ate anion.
We then went to examine the reduction step by reacting
resin 21 with N-benzyl-1,4-dihydronicotinamide according
to a literature procedure.21 Having obtained resin 22, asym-
metric reduction of methyl benzoylformate was carried out
in the presence of Mg5ClO4)2 in a mixture of acetonitrile±
benzene 51:2). Under the conditions speci®ed in Scheme 12,
methyl mandelate was obtained in 50% yield and 72% ee
5R). Resin 22 afforded modest results in terms of chemical
yield and stereoselectivity in comparison with those
obtained with reagent 4 in homogenous conditions. The
low chemical yield obtained suggests either that the
Scheme 12. 5i) TfOMe/CH2Cl2/rt/8 h; 5ii) N-benzyl-1,4-dihydronicotin-
amide/C6H6/CH3CN 52:1)/rt/48 h; 5iii) Methyl benzoylformate/C6H6/
CH3CN 52:1)/rt/48 h.
reduction of polymer-supported quinolinium salt 18a was
incomplete or that a large number of functional groups are

C. Vitry et al. / Tetrahedron 57 12001) 3087±3098
3093
inaccessible to methyl benzoylformate. Accordingly, these
primary results might be improved by the use of a spacer
between the polymeric matrix and the reagent to increase
the accessibility of the model 5Scheme 12).
H, 3.5; N, 16.3. To the above crude N-oxide 51 g) was intro-
duced freshly distilled phosphorus oxychloride 55 mL). The
mixture was heated to re¯ux for 12 h, and the excess of
reagent was removed under vacuum. Ice was cautiously
added and the resulting solution was neutralized with
potassiumcarbonate. The resulting precipitate was ®ltered,
dried and puri®ed by sublimation at 908C under 1 Torr. The
overall yield for the two steps was 55%. White solid. Mp
3. Conclusion
1
1608C 5lit.25). IR 5KBr): 2231 cm21. H NMR 5CDCl3) d
The rational design of new NADH models based on a bio-
mimetic approach led us to prepare reagents 3 and 4, both of
which showed high degree of enantioselectivity in the
course of the reduction of methyl benzoylformate. The
presence of methoxy groups in reagent 4 improved the
reactivity of these annelated models and is probably respon-
sible for the high dependence of the stereoselectivity on
magnesium ion concentration. The second aspect of this
work aimed to obtain polymer-supported NADH models.
In this respect, the good stability of annelated models and
the possibility to recover the corresponding quinolinium
salt, makes this class of reagents of particular interest. To
this end, the grafting of model 4 to a Merri®eld resin was
successfully achieved by an original approach fromphenol
derivative 19. The resulting resin 22 was involved with
modest success in the reduction of methyl mandelate.
8.59 5s, 1H); 8.10 5d, Jˆ8.4 Hz, 1H); 7.99±7.68 5m, 3H).
Anal. calcd for C10H5ClN2: C, 63.68; H, 2.65; N, 14.85.
Found: C, 63.7; H, 2.6; N, 14.8.
4.1.2. 2-Ethynylquinoline-3-carbonitrile 66). Cuprous
iodide 50.16 g, 0.84 mmol) was dissolved in triethylamine
54 mL) at 408C in a ¯ask ¯ushed with argon. Trimethyl-
silylacetylene 50.17 mL, 1.2 mmol) was added at room
temperature. The mixture was stirred for 10 min and bis5tri-
phenylphosphine)palladium5II) chloride 50.3 g, 0.42 mmol)
was added. After 10 min at room temperature, 2-chloro-
quinoline-3-carbonitrile freshly puri®ed by sublimation
51.7 g, 9 mmol), triethylamine 510 mL) and trimethylsilyl-
acetylene 51.7 mL, 12 mmol) were added. The mixture was
stirred at 1008C for 12 h. After cooling, volatile products
were eliminated under reduced pressure. The residue was
extracted with Et2O 54£30 mL). The ethereal extract was
concentrated under reduced pressure. The resulting dark
residue was dissolved in THF 550 mL) and a solution of
0.1 M aqueous NaOH 545 mL) was added. After 2 h stirring,
evaporation of THF formed a precipitate, which was
®ltrated. The aqueous phase was extracted with Et2O
53£30 mL). The organic phase was dried 5MgSO4) and
concentrated under reduced pressure, giving a residue,
which was combined with the above precipitate. After puri-
®cation by chromatography on silica gel 5eluent: CH2Cl2/
EtOH, 95:5), compound 6 was obtained in 69% yield from
4. Experimental
4.1. General
The infrared spectra were recorded on a Beckmann IR 4250
1
spectrometer. The H and 13C NMR spectra were recorded
on a 200 MHz Bruker apparatus. Spectra were recorded in
deuteriochloroformor in hexadeuteriodimethylsulfoxide
5DMSO-d6). Chemical shift are given in ppm with TMS
or HMDS as internal reference. Chemicals were purchased
fromAldrich Co. and Janssen Co. and, unless otherwise
stated, were used without further puri®cation. Flash chro-
matographies were performed with silica 60 570±230 mesh
from Merck) and monitored by thin layer chromatography
5TLC) with silica plates 5Merck, Kieselgel 60 F254). The
following compounds were prepared by literature methods:
4-benzyloxy-3-methoxybenzaldehyde,23 4,5-dimethoxy-2-
nitrobenzaldehyde.24
1
5. White solid. Mp 2028C. IR 5KBr) 2231 cm21. H NMR
5CDCl3) d 8.59 5s, 1H); 8.19 5d, Jˆ8.5 Hz, 1H); 7.99±7.90
5m, 2H); 7.74 5t, Jˆ7.1 Hz, 1H); 3.60 5s, 1H). Anal. calcd
for C12H6N2: C, 80.88; H, 3.39; N, 15.72. Found: C, 80.7; H,
3.3; N, 15.6.
4.1.3. 2-[6S)-2-Benzyl-1-hydroxyethyl]-1-oxo-1,2,3,4-tetra-
hydrobenzo[1,3-b]naphthyridine 68).
A solution of
compound 6 51.8 g, 10 mmol) and 5S)-phenylalaninol
51.66 g, 11 mmol) in THF 535 mL) was heated to re¯ux
for 24 h. After evaporation of the solvent, the residue was
dissolved in methanol 550 mL) and treated with sodium
cyanoborohydride 51.3 g, 20 mmol) and zinc chloride
51.4 g, 10 mmol). After stirring at room temperature for
1 h, sodium cyanoborohydride 50.65 g, 10 mmol) and zinc
chloride 50.7 g, 5 mmol) were added. The resulting solution
was stirred for a further 1 h at this temperature. After addi-
tion of sodiumhydroxide 550 mL, 0.1 M aqueous solution),
the aqueous phase was extracted with dichloromethane
53£30 mL). The organic phase was dried 5MgSO4) and
evaporated to afford compound 7 as oil, which was used
in the next step without further puri®cation. Compound 7.
1H NMR 5CDCl3) d 8.47 5s, 1H); 8.05±7.40 5m, 4H); 7.23±
7.02 5m, 5H); 3.81±3.62 5m, 2H); 3.55±3.00 5m, 7H); 2.88±
2.70 5m, 2H). The oily residue was re¯uxed for 48 h in a
95:5 mixture of ethanol/water 520 mL). After evaporation of
ethanol, the obtained solid was puri®ed by chromatography
4.1.1. 2-Chloroquinoline-3-carbonitrile 65). Quinoline-3-
carbonitrile 53.0 g, 19 mmol) was dissolved in dichloro-
methane 520 mL). A solution of m-chloroperbenzoic acid
53.6 g, 21 mmol) in dichloromethane 535 mL) was then
added dropwise. The mixture was stirred for 30 min at
408C and a solution of m-chloroperbenzoic acid 51.95 g,
11 mmol) in dichloromethane 520 mL) was then added. Stir-
ring was maintained 24 h at 408C. After cooling at room
temperature, water 550 mL) was added, the aqueous phase
was basi®ed with sodiumcarbonate, then extracted with
dichloromethane. The combined organic phases were
dried 5MgSO4), the solvent was evaporated leading to the
crude N-oxide which was used without further puri®cation.
A sample of the N-oxide was recrystallized fromethanol to
1
afford a brown solid. Mp 1068C. IR 5KBr) 2235 cm21. H
NMR 5CDCl3) d 8.77 5d, Jˆ8.5 Hz, 1H); 8.62 5d, Jˆ1.3 Hz,
1H); 8.07 5s, 1H); 7.99±7.76 5m, 3H). Anal. calcd for
C10H6N2O: C, 70.58; H, 3.52; N, 16.45. Found: C, 70.2;

3094
C. Vitry et al. / Tetrahedron 57 12001) 3087±3098
on silica gel 5eluent: CH2Cl2/EtOH, 95:5). Overall yield
from 6: 52%. Yellow solid. Mp 1578C. IR 5KBr):
remaining oil was distilled 5Kugelrohr). The fraction of
bp0.5ˆ1208C was ®nally puri®ed by chromatography on
silica gel 5eluent: petroleumether/ethyl acetate, 95:5) lead-
ing to 2.82 g of 10 5yellow oil). Yield 69%. IR: 1745,
1
1630 cm21. H NMR 5CDCl3) d 8.79 5s, 1H); 8.02±7.45
5m, 4H); 7.29±7.10 5m, 5H); 4.95±4.80 5m, 1H); 4.01±
3.90 5m, 2H); 3.65±3.53 5m, 2H); 3.20±3.00 5m, 4H).
Anal. calcd for C21H20N2O2: C, 72.32; H, 6.42; N, 9.92.
Found: C, 72.2; H, 6.4; N, 10.0.
1
1716 cm21. H NMR 5CDCl3) d 4.47 5t, Jˆ5.3 Hz, 1H);
4.17 5q, Jˆ7.1 Hz, 2H); 3.69±3.37 5m, 6H); 2.61 5t,
Jˆ7.1 Hz, 2H); 1.90 5dt, Jˆ7.1 and 5.3 Hz, 2H); 1.25 5t,
Jˆ7.1 Hz, 3H); 1.16 5t, Jˆ7.1 Hz, 6H). 13C NMR 5CDCl3) d
203.1; 167.9; 102.5; 62.4; 62.0; 50.0; 38.5; 28.3; 15.9;
14.75. Anal. calcd for C12H22O5: C, 58.52; H, 9.00.
Found: C, 58.3; H, 8.6.
4.1.4. 5-Methyl-2-[6S)-2-benzyl-1-hydroxyethyl]-1-oxo-
1,2,3,4,5,10-hexahydrobenzo[1,6-b]naphthyridine 63). A
solution of compound 8 51.32 g, 4 mmol) in acetonitrile
55 mL) and methyl iodide 55 mL, 80 mmol) was re¯uxed
for 12 h. After evaporation of the solvent, Et2O was
added. The resulting suspension was stirred for 2 h. The
precipitate was ®ltered, washed with Et2O and dried
4.1.7. 1,1-Diethoxy-2-iodoethane 611). To a solution of
1,1-diethoxy-2-bromoethane 520 g, 100 mmol) in acetone
5200 mL) was added sodium iodide 530.4 g, 200 mmol).
The solution was stirred at roomtemperature for 5 days.
The resulting precipitate was ®ltrated and washed with
acetone 5100 mL). Acetone was evaporated and the residue
was puri®ed by chromatography on silica gel 5eluent: petro-
leumether/ethyl acetate, 95:5) leading to 22.4 g of 11
5colourless oil). Yield 91%. IR: 2975, 2879, 1122, 1056,
5MgSO4) affording the quinoliniumsalt.
1H NMR
5DMSO-d6) d 9.59 5s, 1H); 8.59 5d, Jˆ9.0 Hz, 2H); 8.30
5t, Jˆ7.1 Hz, 1H); 8.02 5t, Jˆ7.1 Hz, 1H); 7.28±7.20 5m,
5H); 5.00±4.90 5m, 1H); 4.42 5s, 3H); 3.72±3.40 5m, 6H);
2.97±2.92 5m, 2H). The crude salt 50.330 g, 0.7 mmol) was
dissolved in a solution of degassed ethanol 54 mL) and water
510 mL). Sodium dithionite 50.685 g, 4 mmol) and sodium
carbonate 50.295 g, 2.8 mmol) were then added and the
mixture stirred under argon for 45 min. Addition of the
same amounts of sodium dithionite and sodium carbonate
were repeated twice at 30 min intervals. Finally, the
aqueous phase was extracted with dichloromethane
53£10 mL). After evaporation, the oily residue was kept in
the dark. Due to instability reagent 3 was not further
1
1005 cm21. H NMR 5CDCl3) d 4.55 5t, Jˆ5.5 Hz, 1H);
3.70±3.42 5m, 4H); 3.14 5d, Jˆ5.5 Hz, 2H); 1.17 5t,
Jˆ7.0 Hz, 6H). 13C NMR 5CDCl3) d 101.46; 61.93;
14.97; 5.44. Anal. calcd for C6H13IO2: C, 29.53; H, 5.37.
Found: C, 29.71; H, 5.24.
4.1.8. 6-6p-Tolylaminomethylidene)-3,4-dimethoxynitro-
benzene 613a). In a ¯ask were introduced 3,4-dimethoxy-6-
nitrobenzaldehyde 58.44 g, 40 mmol), p-toluidine 54.98 g,
40 mmol) and ethanol 5250 mL). The mixture was re¯uxed
for 2 h under stirring. After cooling at 08C, compound 13a
was isolated by ®ltration and dried under vacuum. Yield
9.75 g 581%). Yellow solid. Mp 1478C 5lit.26 1318C). IR
1
puri®ed. Yield 70%. H NMR 5CDCl3) d 7.22 5m, 7H);
6.98 5d, 1H); 6.87 5d,1H); 4.26 5m, 1H); 3.92±3.72 5m,
4H); 3.36±3.02 5m, 2H); 3.12 5s, 3H); 2.32 5t, 3H).
4.1.5. 3,4-Dimethoxy-6-aminobenzaldehyde 69). In a
solution of 3,4-dimethoxy-6-nitrobenzaldehyde10 50.422 g,
2 mmol) in methanol 525 mL) was introduced a solution of
sodium carbonate 50.424 g, 4 mmol) and sodium dithionite
50.696 g, 4 mmol) in water 515 mL). After 30 min, a solu-
tion of sodiumdithionite 50.0696 g, 4 mmol) in water
510 mL) was added. The solution became orange. After
1 h at roomtemperature, the solution was extracted with
dichloromethane 53£20 mL). The organic phases were
dried and concentrated under reduced pressure. The product
1
5KBr): 1567; 1519; 1329; 1286 cm21. H NMR 5CDCl3) d
9.04 5s, 1H); 7.78 5s, 1H); 7.62 5s, 1H); 7.22 5s, 4H); 4.08 5s,
3H); 4.01 5s, 3H); 2.39 5s, 3H). 13C NMR 5CDCl3) d 155.2;
153.1; 150.4; 148.6; 142.5; 136.7; 129.8; 126.0; 121.1;
109.85; 107.3; 56.6; 56.5; 21.0. Anal. calcd for
C16H16N2O4: C, 63.99; H, 5.37; N, 9.33. Found: C, 63.8;
H, 5.4; N, 9.5.
4.1.9. 6-6p-Tolylaminomethylidene)-3-benzyloxy-4-methoxy-
nitrobenzene 613b). Compound 13b was prepared in the
same manner as compound 13a, from4-benzyloxy-5-meth-
oxy-2-nitrobenzaldehyde 55.74 g, 20 mmol) and p-toluidine
52.14 g, 20 mmol) in ethanol 5300 mL). Yield 6.32 g 584%).
Yellow solid. Mp 1478C 5ethanol). IR 5KBr): 1570; 1517;
1327; 1283 cm21. 1H NMR 5CDCl3) d 9.04 5s, 1H); 7.79 5s,
1H); 7.69 5s, 1H); 7.47±7.39 5m, 5H); 7.23 5s, 4H); 5.26 5s,
2H); 4.08 5s, 3H); 2.40 5s, 3H). 13C NMR 5CDCl3) d 155.25;
153.62; 149.4; 148.6; 142.2; 136.7; 135.25; 129.8; 128.8;
128.4; 127.5; 126.2; 121.15; 110.1; 109.15; 71.3; 56.6; 21.0.
Anal. calcd for C22H20N2O4: C, 70.18; H, 5.35; N, 7.44.
Found: C, 70.2; H, 5.4; N, 7.4.
1
was not further puri®ed. Orange oil. Yield 35%. H NMR
5CDCl3) d 9.63 5s, 1H); 6.62 5s, 1H); 6.09 5brs, 3H); 3.83 5s,
3H), 3.80 5s, 3H).
4.1.6. Ethyl-6,6-diethoxy-3-oxohexanoic acid 610). In a
¯ask, ¯ushed with argon and cooled at 08C, was introduced
sodium hydride 50.850 g, 21 mmol, 60% dispersion in oil) in
anhydrous THF 5100 mL). Ethyl acetoacetate 52.7 mL,
21 mmol) was added dropwise, followed by a solution of
n-BuLi in hexane 52.5 M, 8.5 mL, 21 mmol). Stirring was
maintained for 15 min and hexamethylphosphoramide
511.1 mL, 63 mmol) was then added. After 20 min, 1,1-
diethoxy-2-iodoethane 11 55.185 g, 21 mmol) was added.
The mixture was stirred at room temperature for 12 h and
an aqueous saturated solution of ammonium chloride
520 mL) was slowly added. After evaporation of THF, the
residue was taken into water, then extracted with ethyl
acetate 54£100 mL). The combined organic phases were
washed with water and a saturated solution of sodium
chloride. After drying and evaporation to dryness, the
4.1.10. 6-6p-Tolylaminomethylidene)-3,4-dimethoxyani-
line 614a). A solution of compound 13a 56 g, 20 mmol) in
ethanol 5200 mL) was heated to re¯ux. To the hot solution,
sodium sul®de nonahydrate 510.57 g, 44 mmol) was added.
After a few minutes, a vigorous reaction occurred and the
heating was maintained for 10 min. After cooling at 08C for
4 h, the precipitate was ®ltered. The remaining solution was

C. Vitry et al. / Tetrahedron 57 12001) 3087±3098
3095
concentrated under reduced pressure. Water 5100 mL) was
added to the residue and a new crop of precipitate was
obtained. Yield 79%. Mp 1238C 5ethanol) 5lit.26 1158C).
3-carboxylate 615a). A solution of 12a 55.87 g, 15 mmol)
in formic acid 5100 mL) was stirred 2 h at room tempera-
ture. The mixture was evaporated to dryness and dichloro-
methane 550 mL) and water 550 mL) were added to the
resulting residue. The aqueous phase was neutralized with
sodiumhydrogenocarbonate. The organic phase was sepa-
rated and the aqueous phase was extracted with dichloro-
methane 52£50 mL). The combined organic phases were
dried and concentrated leading to crude 15a, which was
not further puri®ed. Yield 100%. Yellow solid. Mp 988C.
1
IR 5KBr): 3346; 1626; 1557; 1507; 1272 cm21. H NMR
5CDCl3) d 8.44 5s, 1H); 7.14 5m, 4H); 6.81 5s, 1H); 6.43
5brs, 2H); 6.24 5s, 1H); 3.90 5s, 3H); 3.85 5s, 3H); 2.37 5s,
3H). 13C NMR 5CDCl3) d 161.05; 152.7; 149.5; 144.9;
140.5; 134.7; 129.6; 120.6; 116.2; 109.8; 99.0; 56.55;
55.7; 20.8. Anal. calcd for C16H18N2O2: C, 71.09; H, 6.71;
N, 10.36. Found: C, 71.1; H, 6.7; N, 10.4.
IR 5KBr): 1712 cm21 1H NMR 5CDCl3) d 9.97 5t,
.
4.1.11. 6-6p-Tolylaminomethylidene)-3-benzyloxy-4-meth-
oxyaniline 614b). Compound 14b was prepared in the
Jˆ1.5 Hz, 1H); 8.61 5s, 1H); 7.31 5s, 1H); 7.08 5s, 1H);
4.42 5q, Jˆ7.1 Hz, 2H); 4.05 5s, 3H); 4.01 5s, 3H); 3.69 5t,
Jˆ7.1 Hz, 2H); 2.97 5td, Jˆ7.1 and 1.5 Hz, 2H); 1.44 5t,
Jˆ7.1 Hz, 3H). 13C NMR 5CDCl3) d 202.1; 166.4; 157.1;
154.3; 149.9; 145.85; 138.2; 121.1; 107.3; 105.4; 61.2; 56.2;
56.0; 41.5; 30.3; 14.3. Anal. calcd for C17H19NO5: C, 64.34;
H, 6.03; N, 4.41. Found: C, 64.1; H, 6.0; N, 4.4.
same manner as compound 14a, fromcompound
13b
53.76 g, 10 mmol) and sodium sul®de nonahydrate 55.28 g,
22 mmol) in ethanol 5250 mL). Yield 98%. Mp 1418C
5ethanol). IR 5KBr): 3428; 1623; 1598; 1547; 1507 cm21
.
1H NMR 5CDCl3) d 8.44 5s, 1H); 7.46±7.33 5m, 5H); 7.22±
7.08 5m, 4H); 6.86 5s, 1H); 6.37 5brs, 2H); 6.26 5s, 1H); 5.19
5s, 2H); 3.87 5s, 3H); 2.37 5s, 3H). 13C NMR 5CDCl3) d
161.1; 152.1; 149.5; 144.6; 141.0; 136.5; 134.7; 129.6;
128.5; 127.9; 127.1; 120.7; 117.3; 110.2; 100.95; 70.42;
57.0; 20.9. Anal. calcd for C22H22N2O2: C, 76.28; H, 6.40;
N, 8.09. Found: C, 76.0; H, 6.4; N, 8.03.
4.1.15. Ethyl-2-62-formylethyl)-6-methoxy-7-benzyloxy-
quinoline-3-carboxylate 615b). Compound 15b was
prepared in the same manner as compound 15a, from 12b
54.67 g, 10 mmol) and formic acid 5150 mL). Yield 95%.
1
Yellow solid. H NMR 5CDCl3) d 9.97 5t, Jˆ1.5 Hz, 1H);
8.61 5s, 1H); 7.53±7.34 5m, 6H); 7.10 5s, 1H); 5.29 5s, 2H);
4.42 5q, Jˆ7.1 Hz, 2H); 4.00 5s, 3H); 3.69 5t, Jˆ7.1 Hz, 2H);
2.96 5td, Jˆ7.1 and 1.5 Hz, 2H); 1.44 5t, Jˆ7.1 Hz, 3H). 13C
NMR 5CDCl3) d 198.4; 166.3; 157.1; 153.7; 150.35; 145.5;
138.4; 135.8; 128.65; 128.2; 127.5; 121.3; 121.2; 108.4;
105.7; 70.85; 61.25; 56.1; 41.3; 30.4; 14.3.
4.1.12. Ethyl-2,2-63,3-diethoxypropyl)-6,7-dimethoxy-
quinoline-3-carboxylate 612a). To a solution of compound
14a 55.94 g, 20 mmol) and compound 10 54.92 g, 20 mmol)
in ethanol 5200 mL) was added a few drops of piperidine.
The resulting solution was re¯uxed for 24 h. After cooling
and evaporation of the solvent, the residue was puri®ed by
chromatography on neutral alumina 5eluent: petroleum
ether/ethyl acetate, 90:10) affording 6.30 g 581%) of
compound 12a. White powder. Mp 888C 5cyclohexane).
4.1.16. Ethyl 2-{[N66S)-1-hydroxy-3-phenyl)prop-2-yl]-3-
aminopropyl}-6,7-dimethoxyquinoline-3-carboxylate
616a). A solution of aldehyde 15a 53.17 g, 10 mmol) and
5S)-phenylalaninol 51.51 g, 10 mmol) in ethanol 575 mL)
was stirred at room temperature for 45 min. After cooling
at 08C, sodium borohydride 50.416 g, 10 mmol) was added.
The solution was then stirred at roomtemperature for a
further 45 min. A mixture of water 550 mL) and dichloro-
methane 550 mL) was then added, and the aqueous phase
was neutralized by adding 1N aqueous HCl. After phase
separation, the aqueous phase was extracted with dichloro-
methane 52£50 mL). The combined organic phases were
dried, evaporated to dryness leading to 4.75 g of crude
compound 16a, which was used in the next step without
1
IR 5KBr): 1717 cm21. H NMR 5CDCl3) d 8.57 5s, 1H);
7.38 5s, 1H); 7.08 5s, 1H); 4.69 5t, Jˆ7.1 Hz, 1H); 4.43 5q,
Jˆ7.1 Hz, 2H); 4.06 5s, 3H); 4.02 5s, 3H); 3.78±3.32 5m,
6H); 2.20±2.09 5m, 2H); 1.45 5t, Jˆ7.1 Hz, 3H); 1.21 5t,
Jˆ7.1 Hz, 6H). 13C NMR 5CDCl3) d 166.8; 159.25; 154.1;
149.7; 146.1; 138.0; 121.8; 121.05; 107.4; 105.4; 102.7;
61.15; 60.8; 56.2; 56.0; 33.3; 32.9; 15.3; 14.3. Anal. calcd
for C21H29NO6: C, 64.43; H, 7.47; N, 3.58. Found: C, 64.6;
H, 7.5; N, 3.5.
4.1.13. Ethyl-2,2-63,3-diethoxypropyl)-6-methoxy-7-benzyl-
oxyquinoline-3-carboxylate 612b). Compound 12b was
prepared in the same manner as compound 12a, from 14b
53.46 g, 10 mmol), 10 52.71 g, 11 mmol) and a few drops of
piperidine in ethanol 5200 mL). Chromatography on neutral
alumina 5eluent: petroleum ether/ethyl acetate, 75:25)
afforded 3.6 g 577%) of compound 12b. Beige powder.
further puri®cation. IR 5KBr): 1719 cm21
.
1H NMR
5CDCl3 200 MHz) d 8.53 5s, 1H); 7.55 5s, 1H); 7.17 5m,
5H); 7.02 5s, 1H); 4.40 5q, Jˆ7.1 Hz, 2H); 4.01 5s, 6H);
3.80 5dd, Jˆ12.5 and 5.2 Hz, 1H); 3.39 5m, 2H); 3.12 5m,
3H); 2.45 5m, 2H); 1.44 5t, Jˆ7.1 Hz, 3H). 13C NMR
5CDCl3) d 166.6; 159.6; 154.4; 149.85; 146.0; 138.6;
138.3; 129.1; 128.45; 126.25; 121.5; 121.2; 107.15; 105.4;
62.0; 61.2; 60.1; 56.3; 56.1; 46.0; 38.0; 34.8; 30.2; 14.3.
1
Mp 938C 5cyclohexane). IR 5KBr): 1714 cm21. H NMR
5CDCl3) d 8.56 5s, 1H); 7.52±7.29 5m, 6H); 7.10 5s, 1H);
5.31 5s, 2H); 4.67 5t, Jˆ7.1 Hz, 1H); 4.42 5q, Jˆ7.1 Hz, 2H);
4.00 5s, 3H); 3.78±3.33 5m, 6H); 2.18±2.08 5m, 2H); 1.44 5t,
Jˆ7.1 Hz, 3H); 1.20 5t, Jˆ7.1 Hz, 6H). 13C NMR 5CDCl3) d
166.85; 159.20; 153.3; 150.1; 146.0; 138.0; 135.95; 128.65;
128.15; 127.45; 121.9; 121.2; 108.8; 105.6; 102.8; 70.75;
61.2; 60.9; 56.1; 33.3; 32.9; 15.35; 14.3. Anal. calcd for
C27H33NO6: C, 69.36; H, 7.11; N, 3.00. Found: C, 69.2; H,
7.1; N, 3.2.
4.1.17. Ethyl 2-{[N66S)-1-hydroxy-3-phenyl)prop-2-yl]-3-
aminopropyl}-6-methoxy-7-benzyloxyquinoline-3-car-
boxylate 616b). Compound 16b was prepared in the same
manner as compound 16a, from 15b 5393 mg, 1 mmol), 5S)-
phenylalaninol 5151 mg, 1 mmol) and sodium borohydride
538 mg, 1 mmol). The crude product 5528 mg) was used in
the next step without further puri®cation. IR 5KBr):
1
1716 cm21. H NMR 5CDCl3 200 MHz) d 8.59 5s, 1H);
4.1.14. Ethyl-2-62-formylethyl)-6,7-dimethoxyquinoline-
7.52±7.09 5m, 12H); 5.30 5s, 2H); 4.40 5q, Jˆ7.1 Hz, 2H);

3096
C. Vitry et al. / Tetrahedron 57 12001) 3087±3098
4.00 5s, 3H); 3.81±3.29 5m, 5H); 3.00±2.50 5m, 6H); 1.95
5m, 2H); 1.43 5t, Jˆ7.1 Hz, 3H). 13C NMR 5CDCl3) d 166.6;
159.4; 153.5; 150.2; 145.85; 138.6; 138.3; 135.8; 129.1;
128.6; 128.4; 128.1; 127.4; 126.2; 121.5; 121.2; 108.5;
105.6; 70.7; 61.4; 60.1; 56.05; 45.8; 37.9; 34.6; 30.0; 14.3.
5CDCl3) d 171.6; 154.45; 152.4; 149.0; 144.28; 137.8;
136.4; 128.8; 128.6; 127.2; 126.7; 121.5; 107.8; 105.4;
64.0; 55.6; 35.1; 32.0; 29.65; 20.2.
4.1.21. 2-[6S)-1-Hydroxy-3-phenylprop-2-yl]-2,3,4,5-tetra-
hydro-8,9-dimethoxy-6-methyl-1-oxo-azepino[4,3-b]-
quinolinium tri¯uoromethanesulfonate 618a). Compound
17a 50.406 g, 1 mmol) and freshly distilled acetonitrile
55 mL) were introduced in a ¯ask ¯ushed with argon.
Methyl tri¯uoromethanesulfonate 50.136 mL, 1.2 mmol)
was added dropwise via a syringe and the mixture was stir-
red for 1 h at roomtemperature. After evaporation of aceto-
nitrile under reduced pressure, the residue was puri®ed by
chromatography on neutral alumina 5eluent: ethyl acetate/
ethanol, 90:10) affording 0.251 g 542%) of compound 18a.
4.1.18. -[6S)-1-Hydroxy-3-phenylprop-2-yl]-2,3,4,5-tetra-
hydro-8,9-dimethoxyazepino[4,3-b]quinolin-1-one 617a).
In a ¯ask ¯ushed with nitrogen, was introduced compound
16a 52.26 g, 5 mmol) dissolved in anhydrous THF 525 mL).
After cooling at 08C, a solution of trimethylaluminium in
heptane 52 M, 25 mL, 50 mmol) was added and the mixture
was re¯uxed for 48 h. After cooling, water was added drop-
wise, the resulting precipitate was ®ltered and washed with a
mixture of dichloromethane and water 51:1). The combined
organic phases were dried and concentrated under vacuum.
The residue was puri®ed by ¯ash chromatography 5neutral
alumina/eluent: dichloromethane/ethanol, 95:5) leading to
1.83 g of compound 17a. Yield 95%. Yellow solid. IR
1
Yellow solid. IR 5KBr): 1637 cm21. H NMR 5CDCl3) d
8.65 5s, 1H); 7.54 5s, 1H); 7.35 5s, 1H); 7.24 5m, 5H); 4.81
5m, 1H); 4.49 5s, 3H); 4.19 5s, 3H); 4.05 5s, 3H); 3.79 5m,
2H); 3.25±2.92 5m, 7H); 2.15 5m, 2H). 13C NMR 5CDCl3) d
166.7; 158.14; 154.3; 151.5; 142.5; 138.45; 137.6; 130.3;
128.8; 128.5; 126.7; 124.4; 123.6; 117.3; 107.7; 99.0; 62.3;
59.8; 57.6; 56.8; 42.4; 40.4; 34.9; 28.7; 28.4. Anal. calcd
for: C26H29F3N2O7S: C, 54.73; H, 5.12; N, 4.91. Found: C,
54.9; H, 5.1; N, 5.0.
1
5KBr): 1622 cm21. H NMR 5CDCl3) d 8.28 5s, 1H); 7.36
5s, 1H); 7.29 5m, 5H); 7.08 5s, 1H); 4.69 5m, 1H); 4.03 5s,
3H); 4.01 5s, 3H); 3.91 5d, Jˆ5.9 Hz, 2H); 3.36 5brs, 1H);
3.23±2.86 5m, 3H); 2.00 5m, 2H). 13C NMR 5CDCl3) d
171.6; 154.8; 153.7; 149.8; 145.5; 137.95; 136.0; 128.9;
128.6; 128.05; 126.7; 122.2; 107.1; 105.55; 64.05; 61.1;
56.2; 56.1; 44.8; 35.1; 33.05; 29.1. Anal. calcd for
C24H26N2O4: C, 70.92; H, 6.45; N, 6.89. Found: C, 70.9;
H, 6.4; N, 6.7.
4.1.22. 2-[6S)-1-Hydroxy-3-phenylprop-2-yl]-2,3,4,5,6,11-
hexahydro-8,9-dimethoxy-6-methylazepino[4,3-b]quino-
lin-1-one 64). Compound 18a 50.571 g, 10 mmol), degassed
water 518 mL) and degassed chloroform 52 mL) were
introduced in a ¯ask ¯ushed with argon, in the dark. The
following procedure was repeated once: sodiumdithionite
50.870 g, 5 mmol) and sodium carbonate 50.318 g, 3 mmol)
were added. After 90 min, a solution of sodium dithionite
51.740 g, 10 mmol) and sodium carbonate 50.318 g,
3 mmol) in degassed water 518 mL) was added. The reac-
tion mixture was stirred for 14 h. After phase separation, the
aqueous layer was extracted with chloroform53 £50 mL).
The combined organic layers were dried and concentrated
to dryness to give 4 50.422 g). Yield 95%. Yellow solid,
which was not further puri®ed before use in the reduction
4.1.19. 2-[6S)-1-Hydroxy-3-phenylprop-2-yl]-2,3,4,5-tetra-
hydro-8-benzyloxy-9-methoxyazepino[4,3-b]quinolin-1-
one 617b). To a solution of 16b 51.06 g, 2 mmol) in EtOH
550 mL) was added portion wise sodium 5138 mg, 6 mmol).
The solution was stirred at re¯ux for 24 h. After cooling, the
solution was neutralised 5pHˆ6±7) with a solution of 10%
aqueous HCl. After addition of water 550 mL) and CH2Cl2
550 mL), the organic layer was dried 5MgSO4). The aqueous
phase was extracted twice with CH2Cl2 52£25 mL). The
combined organic phases were dried and the solvent evapo-
rated under vacuum. The residue was chromatographed
5neutral alumina/eluent: dichloromethane/ethanol, 95:5)
affording 695 mg 572%) of lactam 17b as a beige solid.
1
of a substrate. H NMR 5CDCl3) d 7.25 5m, 5H); 6.60 5s,
1H); 6.40 5s, 1H); 4.51 5m, 1H); 4.12 5brs, 1H); 3.87 5s, 3H);
3.84 5s, 3H); 3.53 5s, 2H); 3.21±2.93 5m, 7H); 2.16 5m, 2H);
1.73 5m, 2H). 13C NMR 5CDCl3) d 175.5; 147.4; 146.5;
144.5; 138.5; 135.8; 129.0; 128.5; 126.5; 116.1; 111.8;
102.05; 98.9; 64.9; 61.1; 56.3; 45.05; 35.3; 34.5; 30.4;
27.7; 24.7.
1
Mp 938C. IR 5KBr): 1620 cm21. H NMR 5CDCl3) d 8.25
5s, 1H); 7.50±7.27 5m, 11H); 7.07 5s, 1H); 5.29 5s, 2H); 4.72
5m, 1H); 3.99 5s, 3H); 3.89 5m, 2H); 3.13±3.05 5m, 4H);
2.83 5t, Jˆ7.1 Hz, 2H); 1.98 5q, Jˆ7.1 Hz, 2H). 13C NMR
5CDCl3) d 171.6; 154.8; 152.7; 150.05; 145.6; 137.95;
135.9; 135.7; 128.9; 128.6; 128.5; 128.05; 127.3; 126.6;
122.2; 108.7; 105.8; 70.6; 63.8; 60.75; 56.1; 44.55; 35.1;
33.05; 29.1. Anal. calcd for C30H30N2O4: C, 74.67; H, 6.27;
N, 5.80. Found: C, 74.5; H, 6.3; N, 5.75.
4.1.23. Polymer-supported compound 19 6resin 20). To a
solution of compound 19 5350 mg, 0.9 mmol) and
potassium carbonate 5494 mg, 0.9 mmol) in dry DMF
510 mL) was added the chloromethylated polystyrene 5300
mg, 2±2.5 mmol of chloromethyl groups/g of resin, 0.6±
0.75 mmol of chloromethylated groups). The suspension
was stirred at roomtemperature for 4 days under N 2. The
resin was removed by ®ltration and washed successively
with water/THF 51:1, 3£50 mL), CH2Cl2 53£50 mL),
methanol 53£50 mL) and CH2Cl2 53£50 mL). Resin 20
was dried at roomtemperature under vacuumfor at least
12 h. Weight gain: 110 mg corresponding to 0.75 mmol of
compound 19/g of resin 5%Nˆ2.07 corresponding to
0.74 mmol of compound 19/g of resin). 13C gel-phase
NMR 575 MHz, CDCl3) d 172.0; 155.1; 153.2; 150.4;
4.1.20. 2-[6S)-1-Hydroxy-3-phenylprop-2-yl]-2,3,4,5-tetra-
hydro-8-hydroxy-9-methoxyazepino[4,3-b]quinolin-1-one
619). To a solution of 17b 5804 mg, 1.7 mmol) in methanol
550 mL) was added 10% Pd/C 5400 mg). The solution was
stirred for 24 h at room temperature under a hydrogen atmo-
sphere. The catalyst was ®ltered through a plug of cotton
and washed with methanol. Evaporation of the solvent
1
afforded 621 mg 595%) of 19 as a yellow solid. H NMR
5CDCl3) d 8.25 5s, 1H); 7.27 5m, 5H); 7.14 5s, 1H); 6.78 5s,
1H); 4.82 5m, 1H); 3.92 5m, 2H); 3.81 5s, 3H); 3.37 5m, 2H);
3.11 5m, 2H); 2.64 5m, 2H); 1.86 5m, 2H). 13C NMR

C. Vitry et al. / Tetrahedron 57 12001) 3087±3098
3097
138.2; 137.1; 122.4; 108.8; 106.0; 70.95; 64.0; 61.0; 56.2;
35.4; 33.4; 29.4.
described above. Yield 67±100%. Enantiomeric excess:
95% 5R).
4.1.24. Polymer-supported quinolinium 18 6resin 21). To
a suspension of resin 20 5350 mg, 0.75 mmol of compound
19/g of resin, 0.26 mmol of 19) in CH2Cl2 510 mL) was
References
added methyl tri¯uoromethanesulfonate 595 mL,
2
1. 5a) Inouye, Y.; Oda, J.; Baba, N. In Reductions with Chiral
Dihydropyridine Reagents in Asymmetric Synthesis, Morrison,
J. D., Ed.; Academic: New York, 1983; vol. 2, pp 92±124.
5b) Ohno, A.; Ushida, S. Mechanistic Models of Asymmetric
Reductions. In Lecture Notes in Bio-Organic Chemistry,
Baulieu, E., Jaenicke, L., Massey, V., Williams, R. J. P.,
Winnacker, E. L., Zerner, B., Eds.; Springer Verlag: Berlin,
1986; pp 1±105. 5c) Burgess, V. A.; Davies, S. G.; Skerj, R. T.
Tetrahedron: Asymmetry 1991, 2, 299±328.
mequiv.). The resin was stirred at room temperature for
4 h. The polymer was ®ltrated and washed successively
with water/THF 51:1, 3£50 mL), CH2Cl2 53£50 mL),
methanol 53£50 mL) and CH2Cl2 53£50 mL). Resin 21
5371 mg) was dried at room temperature under vacuum
for at least 12 h. 13C gel-phase NMR 575 MHz, CDCl3) d
166.5 ppm. 19F gel-phase NMR 5188 MHz, CDCl3) d
278.5 ppm. Anal. found %Nˆ1.71 corresponding to
0.61 mmol of quinolinium 18/g of resin.
2. See for example: 5a) Dupas, G.; Levacher, V.; Bourguignon,
J.; Queguiner, G. Heterocycles 1994, 39, 405±429.
5b) Levacher, V.; Dupas, G.; Queguiner, G.; Bourguignon, J.
Trends Heterocycl. Chem. 1995, 4, 293±302.
4.1.25. Polymer-supported model 4 6resin 22). To a solu-
tion of benzene 54 mL) and acetonitrile 52 mL), ¯ushed with
N2, was added resin 21 51 g, 0.61 mmol of quinolinium 18/g
of resin, 0.61 mmol) and N-benzyl-1,4-dihydronicotinamide
5261 mg, 1.2 mmol). The suspension was stirred at room
temperature for 48 h in the dark. The polymer was removed
by ®ltration and washed successively with water/THF 51:1,
3£50 mL), CH2Cl2 53£50 mL), methanol 53£50 mL) and
CH2Cl2 53£50 mL). Resin 22 51 g) was dried at room
temperature under vacuum for at least 12 h. Anal. found
%Nˆ1.80 corresponding to 0.65 mmol of model 4/g of
resin.
3. Li, H.; Goldstein, B. J. Med. Chem. 1992, 35, 3560±3567.
4. 5a) Ecklund, H.; Samama, J. P.; Jones, T. A. Biochemistry
1984, 23, 5982. 5b) Skarzynski, T.; Moody, P. C. E.;
Wonacott, A. J. J. Mol. Biol. 1987, 193, 171.
Â
Â
5. Combret, Y.; Torche, J. J.; Ple, N.; Du¯os, J.; Dupas, G.;
Bourguignon, J.; Queguiner, G. Tetrahedron 1991, 47,
9369±9382.
Â
6. Bedat, J.; Levacher, V.; Dupas, G.; Queguiner, G.;
Bourguignon, J. Chem. Lett. 1996, 359±360.
7. 5a) Johnson, C. C.; Gardner, J. L.; Snelter, C. H.; Metzler, D. E.
Biochemistry 1963, 2, 689. 5b) Youn Kim, C. S.; Chaykin, S.
Biochemistry 1968, 7, 2339. 5c) Van Eikeren, P.; Kenny, P.;
Tokmakian, R. J. Am. Chem. Soc. 1979, 101, 7406.
4.1.26. Reduction of methyl benzoylformate with resin
22. In a ¯ask, ¯ushed with argon, were introduced resin 22
51 g, 0.65 mmol of model 4/g of resin, 0.65 mmol), aceto-
nitrile 53 mL), methyl benzoylformate 592 mL, 0.65 mmol)
and magnesium perchlorate 5145 mg, 0.65 mmol). The solu-
tion was stirred at roomtemperature for 24 h in the dark.
The resin was removed by ®ltration and successively
washed with acetonitrile 53£10 mL), water/THF 51:1)
52£10 mL). Organic solvents were evaporated and the
resulting aqueous phase was extracted twice with CH2Cl2
510 mL). After drying 5MgSO4) and evaporation of the
solvent, the residue was chromatographed on silica gel
5eluent: Et2O/cyclohexane, 2:1). Yield 50%. Enantiomeric
excesses were determined by HPLC analysis using a
Chiracel OD column 5250£4.6 mm; 10 mm). Chromato-
graphic conditions: injection: 20 ml 50.5 mg of methyl
mandelate in 10 mL of hexane). Eluent: hexane/2-propanol,
90:10. Flow rate: 1 mL/min. Pressure: 300 psi. Tempera-
ture: 228C. UV detection: lˆ235 nm. Retention time:
9.2 min [5S)-enantiomer] and 14.8 min [5R)-enantiomer].
Enantiomeric excess: 72% 5R).
8. 5a) Cazin, J.; Dupas, G.; Bourguignon, J.; Queguiner, G.
Tetrahedron Lett. 1986, 27, 2375±2378. 5b) Levacher, V.;
Boussad, N.; Dupas, G.; Bourguignon, J.; Queguiner, G.
Tetrahedron 1992, 48, 831±840. 5c) Charpentier, P.;
Â
Lobregat, V.; Levacher, V.; Dupas, G.; Queguiner, G.;
Bourguignon, J. Tetrahedron Lett. 1998, 39, 4013±4016.
9. Cheng, C. C.; Yan, S. J. In The Friedlander synthesis of
quinolines organic reactions, Wiley: New York, 1982; vol.
28, pp 37±202.
10. 5a) Rao, K. V.; Kuo, H. S. J. Heterocycl. Chem. 1979, 16,
1241. 5b) Sha®ee, A.; Parang, K.; Khazan, M.; Ghasemian,
F. J. Heterocycl. Chem. 1992, 29, 1859.
11. 5a) Oikawa, Y.; Sugano, K.; Yonemitsu, O. J. Org. Chem.
1978, 43, 2087±2090. 5b) Huckin, S. N.; Weiler, L. J. Am.
Chem. Soc. 1974, 96, 1082±1087. 5c) Ohta, S.; Shimabayashi,
A.; Hayakawa, S.; Sumino, N.; Okamoto, M. Synthesis 1985,
45±48. 5d) Langlois, M.; Yang, D.; Soulier, J. L.; Florac, C.
Synth. Commun. 1992, 22, 3115±3127. 5e) Lee, C. J.; De Milo,
A. B.; Waters, R. M. Synth. Commun. 1996, 26, 153±160.
12. 5a) Hampton, K. G.; Flannery, R. E. J. Chem. Soc., Perkin
Trans. 1 1973, 2308. 5b) Hampton, K. G.; Flannery, R. E.
J. Org. Chem. 1972, 37, 2806±2810.
4.1.27. Typical procedure for the reduction of methyl
benzoylformate with the free models 3 and 4. In a ¯ask,
¯ushed with argon, were introduced model 4 50.422 g,
1 mmol), acetonitrile 53 mL), methyl benzoylformate
5142 mL, 1 mmol) and magnesium perchlorate 5110 mg,
0.5 mmol). The resulting solution was stirred at room
temperature for 24 h in the dark. After addition of water
510 mL), the organic solvent was evaporated under reduced
pressure and the resulting aqueous phase was extracted with
CH2Cl2 53£10 mL). After drying 5MgSO4) and evaporation
of CH2Cl2, the residue was puri®ed and analysed as
È
13. 5a) Friedlander, P.; Henriques, S. Chem. Ber. 1882, 15, 2572.
5b) McGeachin, S. G. Can. J. Chem. 1966, 44, 2323. 5c) Abert,
A.; Yamamoto, H. J. Chem. Soc. B 1966, 956±963.
14. 5a) Borsche, W.; Ried, W. Liebigs Ann. Chem. 1943, 554, 269.
5b) Borsche, W.; Barthenheier, J. Liebigs Ann. Chem. 1941,
548, 50.
15. Porter, H. K. The Zinin reduction of nitroarenes. Org. React.
1973, 20, 455±481.
16. Gorgues, A. Bull. Soc. Chim. Fr. 1974, 529±530.

3098
C. Vitry et al. / Tetrahedron 57 12001) 3087±3098
17. Saavedra, J. E. J. Org. Chem. 1985, 50, 2271±2273.
18. Basha, A.; Lipton, M.; Weinreb, S. M. Tetrahedron Lett. 1977,
48, 4171±4174.
22. 5a) Levacher, V.; Moberg, C. J. Org. Chem. 1995, 60, 1755±
1762. 5b) Levacher, V.; Moberg, C. React. Polym. 1995, 24,
183±193. 5c) Levacher, V.; Adolfsson, H.; Moberg, C. Acta
Chem. Scand. 1996, 50, 454±457.
19. 5a) Makino, T.; Numozawa, T.; Baba, N.; Oda, J.; Inouye, Y.
J. Chem. Soc., Perkin Trans. 1 1980, 7±10. 5b) Leroy, C.;
Levacher, V.; Dupas, G.; Queguiner, G.; Bourguignon, J.
Tetrahedron: Asymmetry 1997, 8, 3309±3318.
23. Castellan, A.; Grelier, S.; Kessab, L.; Nourmamode, A.;
Hannachi, Y. J. Chem. Soc., Perkin Trans. 2 1996, 1131±
1138.
24. Kumar, S.; Wachtel, E. J.; Keinan, E. J. Org. Chem. 1993, 58,
3821±3827.
Â
20. 5a) Losset, D.; Dupas, G.; Bourguignon, J.; Queguiner, G.
Polym. Bull. 1989, 649±654. 5b) Dupas, G.; Decormeille,
25. Nandeeshaiah, S. K.; Sarvottam, Y. Indian J. Chem., Sect. B
1994, 33, 375.
26. Rilliet, A. Helv. Chim. Acta 1922, 5, 547±552.
Â
A.; Bourguignon, J.; Queguiner, G. Tetrahedron 1989, 45,
2579±2590.
21. Obika, S.; Nishiyama, T.; Tatematsu, S.; Nishimoto, M.;
Miyashita, K.; Imanishi, T. Heterocycles 1998, 49, 261±267
5and references therein).