The ring expansions of glyceraldehyde-derived aziridine-2-carboxylates to oxazolines take place with an uncommon regiochemistry

Article abstract of DOI:10.1002/1099-0690(200109)2001:18<3545::AID-EJOC3545>3.0.CO;2-K

The ring expansion of the D-glyceraldehyde-derived ethyl trans-N-benzoylaziridine-2-carboxylate 9 occurs with retention of configuration, giving the ethyl trans-oxazoline-5-carboxylate 10 as the only product. The observed regioselectivity is rather unusual, since aziridine-2-carboxylates generally rearrange to give oxazoline-4-carboxylates. Conversely, the ring expansion of the cis-N-benzoyl compound 11 under the same reaction conditions is much slower and less stereoselective, giving a mixture of cis- and trans-oxazolines, but with the same regioselectivity. The hydrolysis of 10 under mild conditions permits the synthesis of the 3-amino-3-deoxy-D-xylonic acid derivative 13.

Full text of DOI:10.1002/1099-0690(200109)2001:18<3545::AID-EJOC3545>3.0.CO;2-K

FULL PAPER
The Ring Expansions of Glyceraldehyde-Derived Aziridine-2-carboxylates to
Oxazolines Take Place with an Uncommon Regiochemistry
Giuliana Cardillo,*[a] Luca Gentilucci,*[a] and Galla Pericot Mohr[a]
Keywords: Nitrogen heterocycles / Lewis acids / Neighbouring-group effects / Ring expansion / Rearrangements
The ring expansion of the
D
-glyceraldehyde-derived ethyl
ring expansion of the cis-N-benzoyl compound 11 under the
trans-N-benzoylaziridine-2-carboxylate 9 occurs with reten- same reaction conditions is much slower and less stereoselec-
tion of configuration, giving the ethyl trans-oxazoline-5-carb- tive, giving a mixture of cis- and trans-oxazolines, but with
oxylate 10 as the only product. The observed regioselectivity
is rather unusual, since aziridine-2-carboxylates generally
the same regioselectivity. The hydrolysis of 10 under mild
conditions permits the synthesis of the 3-amino-3-deoxy-
D-
rearrange to give oxazoline-4-carboxylates. Conversely, the xylonic acid derivative 13.
Introduction
reasons, we decided to look for new ways to synthesise po-
lyhydroxylated amino acids from aziridine starting mat-
erials, as only a few examples from this field have been re-
ported in the literature.[13]
Aziridine-2-carboxylates are well known as useful re-
agents for the synthesis of many categories of nitrogen-con-
taining compounds, in particular α- or β-amino acids, by
way of nucleophilic ring opening with reversal of configura-
tion.[1,2] Among the synthetic applications of these com- Results and Discussion
pounds, the rearrangement of N-activated aziridines has be-
come the object of attention only very recently.[3] The ring
expansion of N-acylaziridines to oxazolines can be pro-
moted by thermal, acidic or nucleophilic conditions,[3] and
generally takes place with good regiocontrol and with reten-
tion of the pre-existing configuration. By means of this re-
action, we were able to obtain unusual hydroxy amino acid
derivatives in optically pure form.[4Ϫ6] It was suggested that
the Lewis acid promoted ring expansion of N-acylaziridines
occurs through an SN1Ј mechanism,[7] consisting of an ini-
tial CϪN bond rupture resulting in a carbocationic-like
transition state or a carbocationic intermediate,[8] followed
by ring closure to the oxazoline (Scheme 1).
In the course of our studies concerning Lewis acid pro-
moted ring expansion of optically pure trans-aziridine-2-
carboxylates or -2-imides, we decided to experiment with
the use of trans-N-acylaziridine-2-carboxylates derived from
-glyceraldehyde as polyhydroxylated aminopentanoic acid
precursors. For this reason we prepared the aziridines 1 and
2[14] by a procedure similar to, although slightly modified
from, that reported by Jähnisch.[14Ϫ16] Of the various aziri-
dine-2-carboxylate syntheses,[1,2] the diastereoselective
GabrielϪCromwell reaction, consisting of the addition of
ammonia to an α,β-dibromo compound, is a very simple
and practical one. The α,β-dibromo precursors can easily
be prepared by bromination of the corresponding α,β-un-
saturated compounds. The diastereoselectivity can be con-
trolled either by means of a chiral auxiliary substitu-
ent,[17,18] or by starting from an α,β-unsaturated compound
derived from a naturally occurring optically active com-
pound.[1] Ethyl (E)-4,5-isopropylidenedioxy-2-pentenoate
was prepared according to the literature.[19] As direct treat-
ment with Br2 is known to cause diol deprotection,[20] the
Scheme 1
Polyhydroxylated amino acids are compounds of particu-
lar interest thanks to their utility as alkaloid and azasugar
precursors.[9] An example is polyoxamic acid,[10,11] a five-
carbon α-amino acid present in some members of the po-
tent antifungal polyoxins. Polyhydroxylated β-amino acids
are much less well known, and to the best of our knowledge
only 3-amino-3-deoxy--arabinonic acid has so far been
synthesized from the five-carbon series.[12] For these
[a]
`
Dipartimento di Chimica ‘‘G. Ciamician’’, Universita degli
Studi di Bologna
Via Selmi 2, 40126 Bologna, Italy
Fax: (internat.) ϩ 39-051/209-9456
E-mail: gentil@ciam.unibo.it
Scheme 2
Eur. J. Org. Chem. 2001, 3545Ϫ3551
WILEY-VCH Verlag GmbH, 69451 Weinheim, 2001
1434Ϫ193X/01/0918Ϫ3545 $ 17.50ϩ.50/0
3545

G. Cardillo, L. Gentilucci, G. P. Mohr
FULL PAPER
protected pentenoate was instead treated with pyridinium
The relative and absolute configurations of 7 and 8 were
tribromide in water,[21] to give the corresponding dibromo determined by comparison of their 1H and 13C NMR spec-
derivative. GabrielϪCromwell addition of NH3 in DMSO tra with those of 1a and 2a respectively. The trans-aziridine
at 0 °C furnished a mixture of trans-1a and cis-2a azirid- 7 was derivatized with benzoyl chloride and the resulting
ines,[14] easily separated by flash chromatography, in a 62:38 N-benzoylaziridine 9 was subjected to ring expansion in the
ratio. Finally, 1a and 2a were derivatized with benzoyl presence of a Lewis acid (Scheme 4 and Table 1).
chloride to produce trans-1b and cis-2b (Scheme 2).
The ring expansion of 1b and 2b to oxazolines was at-
tempted in the presence of several Lewis acids and in differ-
ent solvents. However, the acetonide protecting group was
partially or totally removed in all cases. For this reason we
decided to prepare aziridine-2-carboxylates derived from -
glyceraldehyde protected with the cyclohexylidene diol
group.
Scheme 4
The cyclohexylidene-protected -glyceraldehyde was pre-
pared by oxidation of di-O-cyclohexylidene--mannitol,[22]
and a Wittig olefination reaction with triethyl phos-
The ring expansion of trans-9 in the presence of 1 equiv.
phonoacetate allowed us to obtain the α,β-unsaturated ethyl of BF3·Et2O complex was performed at room temperature
(E)-4,5-O,O-cyclohexylidenedioxy-2-pentenoate 3.[23] Treat- in CHCl3. After 18 h, the reagent had almost disappeared,
ment of 3 with pyridinium tribromide[21] in water gave the as indicated by 1H NMR and GC-MS analysis of the crude
dibromo derivative 4 as a mixture of diastereoisomers, reaction mixture, showing the presence of trans-oxazoline
which were used without separation. Ammonia was 10 (Table 1, Entry 1), accompanied by a complex mixture
bubbled through a solution of these in DMSO at 0 °C, giv- of deprotected compounds. The reaction was repeated at
ing trans- and cis-aziridine 7 and 8 in a 65:35 diastereomeric reflux, and was judged complete after 2 h. Again, a mixture
ratio and good yield after separation by flash chromato- of deprotected compounds was found to be present in the
graphy. The other two isomers represented less than 10% reaction mixture (Entry 2).
altogether, as determined by GC-MS analysis of the crude
reaction mixture (Scheme 3).
Much better results were obtained on treatment of 9 with
1 equiv. of MgBr2·Et2O complex in THF at reflux, which
The reaction goes through the α,β-unsaturated α-bromo in 2 h gave the trans-oxazoline-5-carboxylate 10 in excellent
intermediate 5, the preferred conformation of which, shown yield (Table 1, Entry 3). No traces of any product resulting
in Scheme 3, was determined by means of semiempirical from bromine ring opening were detected.[26,27]
PM3 computations,[24] performed on geometries generated
The aziridine 9 was found to be rather reactive towards
by a Monte Carlo procedure.[25] The ammonia attack is ring expansion, even in the absence of a Lewis acid. Indeed,
likely to occur from the less hindered face, which provides a slow, spontaneous conversion was observed when it was
an explanation for the high diastereoselectivity observed ex- left to stand neat at room temperature, and after 4 d a cer-
perimentally in the formation of the C-3ϪN bond. Con- tain quantity of trans-oxazoline 10 was present (Entry 4),
versely, subsequent protonation of the resulting enolate, giv- as shown by 1H NMR, the rest being unchanged 9. In addi-
ing intermediate 6, is responsible for the formation of the tion, trans-9 exclusively underwent ring expansion, even at
cis/trans mixture.
reflux for 2 h in acetic acid, without any trace of products
originating from nucleophilic ring opening (Entry 5).[28]
These data indicate that the ring expansion of trans-9
takes place with complete control over the regioselectivity,
giving a single stereoisomer and retention of absolute con-
figuration. The trans relative stereochemistry of 10 was de-
1
termined on the basis of the 4-HϪ5-H H NMR coupling
constants, and the absolute stereochemistry could be defin-
itively assigned as (4S,5R) since only the aziridine C-2ϪN
bond had been broken during the ring-expansion reaction.
The cis-aziridine 8 was derivatized with benzoyl chloride
and the resulting cis-N-benzoylaziridine 11 was subjected to
ring-expansion conditions in the presence of a Lewis acid
(Scheme 5). The reaction was observed to be much slower
and less straightforward than the ring expansion of trans-9
(Table 1). Indeed, after 2 h at reflux in THF in the presence
of 1 equiv. of MgBr2, the degree of conversion into cis-
1
oxazoline 12 was around 15%, as determined by H NMR
and GC-MS analysis of the reaction mixture, the rest still
Scheme 3
3546
being 11 (Table 1, Entry 6). After 4 h, the degree of conver-
Eur. J. Org. Chem. 2001, 3545Ϫ3551

The Ring Expansions of Glyceraldehyde-Derived Aziridine-2-carboxylates to Oxazolines
FULL PAPER
Table 1. Ring expansions of trans-9 and cis-11 in the presence of 1 equiv. of Lewis acid
Entry
Aziridine
Lewis acid (1 equiv.)
Solvent
Temp. [°C]
Time [h]
10 (%)
12 (%)
1
2
3
4
5
6
7
8
9
9
BF3
BF3
MgBr2
Ϫ
Ϫ
MgBr2
MgBr2
BF3
CHCl3
CHCl3
THF
Ϫ
AcOH
THF
THF
CHCl3
room temp.
reflux
reflux
room temp.
reflux
reflux
18
2
2
96
2
2
40[a]
60[a]
85[b]
25[c]
55[c]
Ϫ
Ϫ
Ϫ
Ϫ
9
9
9
11
11
11
15[a]
reflux
room temp.
8
18
45[a]
Ϫ
30[a]
traces
[a]
1
[b]
Calculated on the basis of H NMR and GC-MS analysis of the crude reaction mixture. Ϫ
Calculated after purification by flash
[c]
1
chromatography, the rest being 9. Ϫ Calculated on the basis of H NMR and GC-MS analysis of the crude reaction mixture, the rest
being 9.
sion was 60%, but by then we were able to detect a mixture this reason, it was to have been anticipated that the ring
of cis-oxazoline 12 and trans-oxazoline 10 in a 65:35 ratio, expansion of both of the aziridines was likely to take place
the latter probably deriving from cis/trans isomerization. in favour of the C-3 positions, which seem more capable
After prolonged reaction times, the degree of conversion of stabilizing the incipient positive charge (Scheme 1). This
reached 75%, with a final 12/10 ratio of 40:60 (Entry 7), observed regioselectivity might be caused by the operation
and a significant presence of bromo derivatives, probably of a neighbouring-group effect,[6] as suggested by a study
originating from nucleophilic bromide ring opening.[25,26] of the Lewis acid·9 complex that we carried out using semi-
The same bromo derivatives were also obtained when the empirical computations.
reaction was performed in toluene, both at room temper-
To calculate the more stable MgBr2·9 structure, we con-
ature and at reflux. When the reaction was run for 18 h in sidered MgBr2 chelation at different positions, to simulate
the presence of BF3·Et2O in CHCl3 at room temperature, the existence of tautomeric complexes in equilibrium. In-
we were able to detect only traces of 12, together with de- deed, the aziridine N atom is pyramidalized by ring strain,
protected compounds (Entry 8).
making it more prone to complexation than the nitrogen
atom of planar amides.[7,8,29] We performed PM3 semiem-
pirical energy minimization[24] of a conformation set gener-
ated by a Monte Carlo procedure.[25] In a similar way, we
calculated the more stable PM3 conformation of uncom-
plexed 9. In the more stable MgBr2·9 structure, Mg coord-
inates both the benzoyl and the ester carbonyl O (Figure 1).
Scheme 5
The final ring opening of 10 was performed under mild
acidic conditions, giving the β-amino acid 13 (Scheme 6).
The regiochemistry of the ring expansion of 9 to 10 was
1
definitively assigned by H NMR decoupling experiments
performed on 13.
Figure 1. MgBr2·9 and BH3·9 structures calculated by semi-
empirical energy minimization
Scheme 6
The regioselectivity of the ring expansions of aziridine 9
and 11 in favour of the C-2 positions, and the high reactiv-
ity of 9, were rather unexpected, and contrasted with the
In this complex, however, there are only small or even no
experimentally observed behaviour of activated aziridine-2- changes in the aziridine interatomic distances relative to
carboxylates.[3] Moreover, it is to be assumed that the re- those in 9.[30] This situation reflects only slight aziridine ac-
gioselectivity is driven by the stability of the carbocationic tivation towards ring opening, and in addition the com-
intermediate or carbocationic-like transition state.[7,8] For plexation reduces the benzoyl O nucleophilicity,[7,8,29] and
Eur. J. Org. Chem. 2001, 3545Ϫ3551
3547

G. Cardillo, L. Gentilucci, G. P. Mohr
FULL PAPER
so is generally considered not to be productive in the re-
arrangement to oxazoline.[7,8]
The next most stable MgBr2·9 structure, 9.3 kcal/mol
higher in energy, shows Mg chelating both N and dioxol-
anic O-4Ј (Scheme 7). The aziridine interatomic distances
are lengthened with respect to 9, and this bond weakening
can activate ring expansion.[7,8] In this structural and con-
formational situation, the dioxolanic O-1Ј is forced to point
towards C-2, allowing the possibility that in the course of
ring expansion to trans-oxazoline-5-carboxylate 10, the C-2
carbocationic transition state might be stabilized by an O-
1Ј lone pair (Scheme 7). This stabilization does not appear
to be possible for the alternative (not observed experiment-
ally) ring expansion to the oxazoline-4-carboxylate by way
of an aziridine C-3 carbocation. All the other complexes
resulting from different modes of magnesium chelation were
found to be higher in energy.
Figure 2. BH3·9, MgBr2·11, and BH3·11 structures calculated by
semiempirical energy minimization
first being almost planar and the second slightly pyram-
idalized, probably due to the more weakly electron-with-
drawing nature of BH3 relative to Hϩ. The interatomic dis-
tances are in good agreement. Moreover, while O-coordin-
ated BH3·9 is found to be 1.1 kcal/mol higher in energy
than N-coordinated BH3·9, O-coordinated Hϩ-9 proves to
be 4.2 kcal/mol lower in energy than N-coordinated Hϩ-9,
confirming the trend observed for MgBr2·9. These compu-
tations moved us to believe that the substitution of BF3
for BH3 may confidently be accepted in the N-coordinated
species, and should offer good indications of the preferred
conformations assumed by the BF3·aziridine complex, from
the similar dimensions of the two boron derivatives.
Scheme 7
Repeating the PM3 computations for MgBr2·11 and the
AM1 computations for BH3·11 similarly, we again consid-
ered coordination at different positions and performed cal-
culations for different complexes. For the same structural
reasons as discussed above for the trans isomer, we consid-
ered that Lewis acid O-coordination was unlikely to be pro-
ductive.[7,8,29] The more stable N-coordinated MgBr2·11
structure, featuring NϪ and ester carbonyl OϪMg chela-
tion, and the more stable N-coordinated BH3·11, (Figure 2)
may be regarded as the active structures as far as aziridine
ring expansion is concerned.[7,8] Apparently, the cis disposi-
tions of the aziridine substituents turn the dioxolane groups
away from the aziridine moieties in both complexes. Con-
formations showing O-1Ј directed towards C-2 or -3 can
still be calculated, but are higher in energy. This renders the
activation rather more difficult, and might be responsible
for the slower ring-expansion reaction times observed with
the cis-aziridines.
Because of the impossibility of obtaining energy minima
with BF3 we substituted it with BH3 and performed AM1
semiempirical computations for the complex BH3·9. We
considered both N- and O-boron coordination,[7,8,29] min-
imizing geometries furnished by the Monte Carlo proced-
ure.[24,25] We also calculated the preferred AM1 conforma-
tion of uncomplexed 9.[31] The N-boron-coordinated BH3·9
structure, computed to be more stable (Figure 2), shows in-
teratomic distances longer than those in 9, indicating strong
activation towards ring expansion.[7,8] The dioxolane O-1Ј
is directed towards C-2, and so the carbocationic transition-
state stabilization at C-2 by an O-1Ј lone pair, as described
above, still seems possible. The complex with B coordinated
to the benzoyl O is only 1.1 kcal/mol higher in energy than
the N-boron complex (Figure 1), but the interatomic dis-
tances are scarcely modified from those of 9, and this struc-
tural feature provides grounds for considering this situation
unreactive towards ring expansion.
The decision to exchange BF3 with BH3 for semiempir-
ical computations has some consequences. Indeed, it seems
extremely probable that the weaker BH3 should influence
the aziridine electron-density distribution in a less marked
Conclusion
We have studied the acid-promoted ring expansion of -
way than BF3. It is generally accepted that BF3 may be glyceraldehyde-derived N-benzoylaziridine-2-carboxylates
substituted by Hϩ to simplify computations,[7,29] and so we trans-9 and cis-11 to oxazolines, as potential polyhydroxyl-
also performed AM1 computations on protonated azirid- ated amino acid precursors. For trans-9 we observed a fast
ines. A comparison between N-coordinated Hϩ-9 and N- and regioselective rearrangement to the trans-oxazoline-5-
coordinated BH3·9 structures found excellent agreement, in carboxylate 10 with retention of configuration, while for
terms both of interatomic distances and of interatomic cis-11 we observed slow but still regioselective formation of
angles. Comparison between O-coordinated Hϩ-9 and O- both cis- and trans-oxazoline-5-carboxylates. The regioch-
coordinated BH3·9 found a difference in N geometry, the emistry of the expansion displayed by trans-9, as well as
3548
Eur. J. Org. Chem. 2001, 3545Ϫ3551

The Ring Expansions of Glyceraldehyde-Derived Aziridine-2-carboxylates to Oxazolines
FULL PAPER
Ethyl (2ЈS)-3-(1Ј,4Ј-Dioxaspiro[4,5]dec-2Ј-yl)aziridine-2-carboxylate
(7, 8): Ammonia was bubbled at 0 °C through a stirred solution of
crude 4 (0.49 g, 1.2 mmol) in DMSO (8 mL) for 25 min. After a
further 1 h, the solution was diluted with EtOAc (50 mL) and the
solution was washed three times with water (5 mL). The organic
layer was dried with Na2SO4 and the solvent was evaporated under
reduced pressure. The residue was purified by flash chromato-
graphy (EtOAc/cyclohexane, 50:50) to give trans-7 (0.16 g, 52%)
its high reaction rate, can be explained by allowing for the
existence of a neighbouring group participation effect, as
suggested by semiempirical computations. The trans-oxazo-
line-5-carboxylate 10 was hydrolysed under mild conditions
to the α-hydroxy β-amino acid 13, which is a protected form
of the unnatural 3-amino-3-deoxy--xylonic acid in an opti-
cally pure state.
˜
and cis-8 (0.087 g, 28%) as waxy solids. Ϫ (2R,3R)-7: IR: ν ϭ 1740,
1640, 1438, 1234, 1160, 1110, 1043 cmϪ1. Ϫ 1H NMR (CDCl3):
δ ϭ 1.29 (t, J ϭ 7.1 Hz, 3 H, CH3), 1.40Ϫ1.50 (m, 2 H, CH2),
1.50Ϫ1.75 (m, 9 H, NH ϩ CH2), 2.36Ϫ2.57 (m, 2 H, CHN ϩ
CHCϭO), 3.78 (dd, J ϭ 6.6, 8.3 Hz, 1 H, OCH2), 3.85Ϫ4.00 (m,
1 H, OCH2), 4.00Ϫ4.20 (m, 1 H, OCH), 4.22 (q, J ϭ 7.1 Hz, 2 H,
CH2CH3). Ϫ 13C NMR (CDCl3): δ ϭ 14.0, 23.9, 25.0, 29.7, 32.0,
35.1, 36.0, 39.7, 61.8, 67.5, 73.2, 110.6, 169.0. Ϫ GC-MS: m/z
(%) ϭ 255 (25) [Mϩ], 212 (80), 140 (75), 112 (90), 68 (100). Ϫ
[α]2D0 ϭ Ϫ37.9 (c ϭ 0.3, CHCl3). Ϫ C13H21NO4 (255.31): calcd. C
Experimental Section
General Remarks: Unless stated otherwise, chemicals were obtained
from commercial sources and used without further purification.
CH2Cl2 was distilled from P2O5. Toluene was distilled from mo-
lecular sieves. THF was distilled from sodium benzophenone ketyl.
Ϫ Flash chromatography was performed on Merck silica gel 60
(230Ϫ400 mesh), and solvents were simply distilled. Ϫ TLC was
performed on Merck fluorescent silica gel plates. Ϫ NMR spectra
were recorded with a Gemini Varian spectrometer at 300 (1H
NMR) and 75 (13C NMR) MHz. Chemical shifts are reported as
δ values relative to the solvent peak of CDCl3, defined at δ ϭ 7.27
(1H NMR) or δ ϭ 77.0 (13C NMR). Ϫ Infrared spectra were re-
corded with an FT-IR Nicolet 210 spectrometer. Ϫ Optical activity
measurements were performed with a PerkinϪElmer 343 polari-
meter. Ϫ GC-MS analysis was performed with a HP-5890 GC
coupled with a HP-5971 MS.
61.16, H 8.29, N 5.49; found C 61.20, H 8.31, N 5.50. Ϫ (2S,3R)-
1
8: IR: ν ϭ 1743, 1645, 1445, 1254, 1162, 1110, 1040 cmϪ1. Ϫ H
˜
NMR (CDCl3): δ ϭ 1.30 (t, J ϭ 7.4 Hz, 3 H, CH3), 1.37Ϫ1.50 (m,
2 H, CH2), 1.50Ϫ1.75 (m, 9 H, NH ϩ CH2), 2.34 (br. dd, 1 H,
CHN), 2.69 (br. d, 1 H, CHCϭO), 3.67Ϫ3.81 (m, 1 H, OCH2),
3.88Ϫ4.08 (m, 2 H, OCH2 ϩ OCH), 4.23 (q, J ϭ 7.4 Hz, 2 H,
CH2CH3). Ϫ 13C NMR (CDCl3): δ ϭ 14.0, 24.0, 25.0, 30.0, 33.3,
35.0, 36.5, 40.0, 61.9, 66.4, 74.3, 110.9, 168.8. Ϫ GC-MS: m/z
(%) ϭ 255 (25) [Mϩ], 212 (80), 140 (75), 112 (90), 68 (100). Ϫ
[α]2D0 ϭ ϩ18.0 (c ϭ 0.2, CHCl3). Ϫ C13H21NO4 (255.31): calcd. C
61.16, H 8.29, N 5.49; found C 61.21, H 8.27, N 5.47.
Computational Methods: For MgBr2·9 and MgBr2·11 we assumed
a number of alternative chelation possibilities: N and O-1Ј; N and
O-4Ј; N and ester carbonyl O; benzoyl O and O-1Ј; benzoyl O and
O-4Ј; benzoyl O and ester carbonyl O. We disregarded coordina-
tions not involving the acyl aziridine portion. For each complex we
generated a set of 200 random conformations by use of a Monte
Carlo procedure,[25] and each conformation energy was minimized
by means of semiempirical PM3 computations.[24] To perform semi-
empirical computations for BF3·9 and BF3·11, we substituted BF3
with BH3, due to the impossibility of obtaining energy minima. We
assumed benzoyl O or N coordination. We generated two sets of
Monte Carlo[25] conformations and minimized energies by means
of AM1 computations.[24] We calculated Hϩ·9 in a similar way, tak-
ing account both of benzoyl O and N protonation. In the Monte
Carlo procedure, we fixed a range for acyclic torsion variation of
45Ϫ180° and a range for ring torsion flexing of 15Ϫ90°. All the
torsion angles were simultaneously varied in a random manner. For
energy minimization we used PM3 or AM1 RHF calculations, fix-
ing the PolakϪRibiere algorithm and a gradient of less than 0.001
Ethyl
(2R,3R)-1-Benzoyl-3-{(2ЈS)-1Ј,4Ј-dioxaspiro[4,5]dec-2Ј-yl}-
aziridine-2-carboxylate (9): A solution of 7 (0.16 g, 0.63 mmol) in
CH2Cl2 (8 mL) was treated at room temperature with triethylamine
(0.10 mL, 0.75 mmol) and benzoyl chloride (0.065 mL, 0.75 mmol),
and the mixture was stirred for 2 h under an inert gas. Water was
then added and the mixture was extracted three times with CH2Cl2
(15 mL). The organic layers were collected and dried with Na2SO4,
and the solvent was evaporated under reduced pressure. Compound
9 was obtained pure after flash chromatography (EtOAc/cyclohex-
ane, 20:80) in quantitative yield, as an oil that slowly tended to
solidify (0.22 g, 97%). Ϫ IR: ν˜ ϭ 1735, 1650, 1438, 1229, 1160,
1
1111, 1040 cmϪ1. Ϫ H NMR (CDCl3): δ ϭ 1.05 (t, J ϭ 7.6 Hz, 3
H, CH3), 1.19Ϫ1.50 (m, 4 H, CH2), 1.50Ϫ1.82 (m, 6 H, CH2), 3.19
(dd, J ϭ 2.4, 4.0 Hz, 1 H, CHN), 3.43 (d, J ϭ 2.4 Hz, 1 H, CHCϭ
O), 3.97 (q, J ϭ 7.6 Hz, 2 H, CH2CH3), 3.98Ϫ4.10 (m, 1 H, OCH2),
4.20 (dd, J ϭ 6.6, 8.4 Hz, 1 H, OCH2), 4.26Ϫ4.37 (m, 1 H, OCH),
7.38Ϫ7.59 (m, 3 H, ArH), 7.92Ϫ8.10 (m, 2 H, ArH). Ϫ 13C NMR
(CDCl3): δ ϭ 13.8, 23.9, 25.1, 29.7, 35.1, 35.9, 39.7, 43.5, 61.8,
67.6, 73.4, 110.7, 128.3, 128.5, 130.5, 132.6, 167.1, 175.4. Ϫ GC-
MS: m/z (%) ϭ 359 (5) [Mϩ], 330 (4), 316 (8), 244 (6), 216 (3), 188
(7), 172 (4), 140 (8), 105 (100). Ϫ [α]2D0 ϭ ϩ14.2 (c ϭ 0.6, CHCl3).
Ϫ C20H25NO5 (359.42): calcd. C 66.84, H 7.01, N 3.90; found C
66.80, H 6.99, N 3.92.
˚
kcal/A mol as the termination condition.
Ethyl (2R)-2,3-Dibromo-3-(1Ј,4Ј-dioxaspiro[4,5]dec-2Ј-yl)propanoate
(4): The α,β-unsaturated ester 3 (0.29 g, 1.2 mmol) was mechanic-
ally stirred at room temperature in water (5 mL) with pyridinium
tribromide (90%, 0.51 g, 1.4 mmol), with exclusion of light. After
1 h, a saturated solution of Na2SO3 was added until the brown
colour disappeared. The reaction mixture was extracted three times
with Et2O (10 mL) and the collected organic layers were dried with
Na2SO4. The solvent was evaporated under reduced pressure and
the residue was used without further purification (0.49 g, 100%,
Lewis Acid Promoted Ring Expansion of trans-9
Ethyl (4S,5R)-4-{(2ЈS)-1Ј,4Ј-Dioxaspiro[4,5]dec-2Ј-yl}-2-phenyl-4,5-
dihydro-1,3-oxazole-5-carboxylate (10): MgBr2·Et2O (0.16 g,
0.61 mmol) was added under an inert gas to a solution of 9 (0.22 g,
two diastereoisomers according to analysis of the crude reaction
1
mixture). Ϫ H NMR (CDCl3): δ ϭ 1.30 (t, 3 H, CH3), 1.30Ϫ1.40 0.61 mmol) in THF (5 mL), and the mixture was refluxed for 2 h.
(m, 4 H, CH2), 1.40Ϫ1.80 (m, 6 H, CH2), 3.70 ϩ 3.95 (dd, 1 H,
OCH2), 3.83 ϩ 4.12 (dd, 1 H, OCH2), 4.26 (q, 2 H, CH2CH3), 4.36
The solvent was removed under reduced pressure, a saturated solu-
tion of NaHCO3 (5 mL) was then added, and the mixture was ex-
ϩ 4.52 (d, 1 H, CHBr), 4.52 ϩ 4.75 (m, 2 H, OCH ϩ CHBr). Ϫ tracted three times with CH2Cl2 (15 mL). The collected organic
GC-MS: m/z (%) ϭ 400 (5) [Mϩ], 357 (55), 257 (20), 177 (13), 141 layers were dried with Na2SO4, and the solvent was evaporated
(13), 125 (38), 97 (39), 55 (100).
under reduced pressure. The mixture was purified by flash chroma-
3549
Eur. J. Org. Chem. 2001, 3545Ϫ3551

G. Cardillo, L. Gentilucci, G. P. Mohr
FULL PAPER
tography on silica gel (cyclohexane/EtOAc, 80:20) to afford 10
(0.19 g, 85%) and unchanged 9 (0.028 g, 15%). Ϫ Compound 10:
pressure. The residue was extracted three times with CH2Cl2
(5 mL), and the collected organic layers were dried with Na2SO4.
IR: ν ϭ 1744, 1650, 1438, 1229, 1166, 1110, 1043 cmϪ1. Ϫ 1H The solvent was evaporated under reduced pressure and the residue
˜
NMR (CDCl3): δ ϭ 1.25 (t, J ϭ 7.2 Hz, 3 H, CH3), 1.30Ϫ1.42 (m,
was purified by flash chromatography (cyclohexane/EtOAc, 40:60)
˜
2 H, CH2), 1.42Ϫ1.78 (m, 8 H, CH2), 4.00 (dd, J ϭ 6.3, 7.3 Hz, 1 to give 13 as a viscous oil (0.18 g, 89%). Ϫ IR: ν ϭ 3416, 3065,
1
H, OCH2), 4.10 (dd, J ϭ 6.3, 6.9 Hz, 1 H, OCH2), 4.27 (q, J ϭ 1733, 1719, 1653, 1646, 1527, 1281, 1096, 1017 cmϪ1. Ϫ H NMR
7.2 Hz, 2 H, CH2CH3), 4.42Ϫ4.59 (m, 2 H, OCH ϩ CHN), 5.03 (CDCl3): δ ϭ 1.30 (t, J ϭ 7.2 Hz, 3 H, CH3), 1.30Ϫ1.45 (m, 2 H,
(d, J ϭ 5.7 Hz, 1 H, CHCϭO), 7.40Ϫ7.58 (m, 3 H, ArH),
CH2), 1.45Ϫ1.65 (m, 9 H, OH ϩ CH2), 3.77 (dd, J ϭ 6.2, 8.4 Hz,
1 H, OCH2), 4.17 (dd, J ϭ 6.6, 8.4 Hz, 1 H, OCH2), 4.26 (q, J ϭ
7.97Ϫ8.10 (m, 2 H, ArH). Ϫ 13C NMR (CDCl3): δ ϭ 14.2, 23.8,
25.2, 29.7, 34.8, 35.6, 61.7, 65.0, 73.0, 76.2, 77.7, 108.6, 128.4, 7.2 Hz, 2 H, CH2CH3), 4.45 (d, J ϭ 2.7 Hz, 1 H, CHOH), 4.54 (dt,
131.7, 169.3, 180.7. Ϫ GC-MS: m/z (%) ϭ 359 (6) [Mϩ], 330 (5), J ϭ 2.7, 6.9 Hz, 1 H, OCH), 4.66 (dt, J ϭ 3.3, 9.3 Hz, 1 H, CHN),
316 (10), 262 (3), 244 (7), 216 (11), 200 (8), 172 (12), 141 (100), 105
6.85 (br. d, J ϭ 9.3 Hz, 1 H, NH), 7.35Ϫ7.60 (m, 3 H, ArH),
(36). Ϫ [α]2D0 ϭ ϩ13.7 (c ϭ 1.0, CHCl3). Ϫ C20H25NO5 (359.42): 7.75Ϫ7.84 (m, 2 H, ArH). Ϫ 13C NMR (CDCl3): δ ϭ 14.1, 23.7,
calcd. C 66.84, H 7.01, N 3.90; found C 66.82, H 7.00, N 3.92.
Ethyl (2S,3R)-1-Benzoyl-3-{(2ЈS)-1Ј,4Ј-dioxaspiro[4,5]dec-2Ј-yl}-
aziridine-2-carboxylate(11): According to the same procedure as de-
scribed for the synthesis of 9, a solution of 8 (0.087 g, 0.34 mmol)
in CH2Cl2 (4 mL) was treated with triethylamine (0.056 mL,
0.41 mmol) and benzoyl chloride (0.036 mL, 0.41 mmol). After the
usual workup, 11 was obtained pure after flash chromatography
(cyclohexane/EtOAc, 90:10) in quantitative yield as a oil that slowly
24.0, 25.1, 34.5, 36.0, 51.6, 62.4, 66.1, 71.1, 74.3, 110.3, 126.9,
127.2, 128.4, 128.6, 133.5, 167.3, 172.5. Ϫ GC-MS: m/z (%) ϭ 377
(9) [Mϩ], 334 (28), 274 (7), 218 (6), 176 (21), 141 (11), 105 (100).
Ϫ [α]2D0 ϭ Ϫ3.5 (c ϭ 0.3, CHCl3). Ϫ C20H27NO6 (377.44): calcd. C
63.64, H 7.21, N 3.71; found C 63.59, H 7.19, N 3.75.
Acknowledgments
We thank the M.U.R.S.T. (60% and Cofin ’98) and Bologna Uni-
versity (Funds for Selected Topics) for financial support of this re-
search.
˜
tended to solidify (0.12 g, 99%). Ϫ IR: ν ϭ 1748, 1653, 1441, 1229,
1166, 1110, 1033 cmϪ1. Ϫ 1H NMR (CDCl3): δ ϭ 1.30 (t, J ϭ
7.2 Hz, 3 H, CH3), 1.56Ϫ1.83 (m, 10 H, CH2), 2.92 (dd, J ϭ 6.7,
8.2 Hz, 1 H, CHN), 3.68 (d, J ϭ 6.7 Hz, 1 H, CHCϭO), 3.80 (dd,
J ϭ 6.2, 8.4 Hz, 1 H, OCH2), 4.05 (dd, J ϭ 6.6, 8.4 Hz, 1 H,
OCH2), 4.28 (q, J ϭ 7.2 Hz, 2 H, CH2CH3), 4.25Ϫ4.45 (m, 1 H,
OCH), 7.39Ϫ7.67 (m, 3 H, ArH), 8.20Ϫ8.40 (m, 2 H, ArH). Ϫ 13C
NMR (CDCl3): δ ϭ 14.2, 24.0, 25.1, 29.7, 34.9, 36.5, 37.7, 45.0,
61.9, 66.4, 74.3, 110.9, 128.2, 130.0, 133.3, 167.2, 176.1. Ϫ GC-MS:
m/z (%) ϭ 359 (3) [Mϩ], 330 (2), 316 (10), 286 (2), 262 (5), 216 (4),
188 (7), 172 (4), 140 (6), 105 (100). Ϫ [α]2D0 ϭ Ϫ33.2 (c ϭ 1.2,
CHCl3). Ϫ C20H25NO5 (359.42): calcd. C 66.84, H 7.01, N 3.90;
found C 66.82, H 7.03, N 3.88.
[1]
D. Tanner, Angew. Chem. Int. Ed. Engl. 1994, 33, 599Ϫ619.
[2]
H. M. I. Osborn, J. Sweeney, Tetrahedron: Asymmetry 1997,
8, 1693Ϫ1715.
[3]
W. McCoull, F. A. Davis, Synthesis 2000, 10, 1347Ϫ1365.
G. Cardillo, L. Gentilucci, A. Tolomelli, C. Tomasini, Tetrahed-
[4]
ron Lett. 1997, 38, 6953Ϫ6956.
G. Cardillo, L. Gentilucci, A. Tolomelli, Tetrahedron Lett.
[5]
1999, 40, 8261Ϫ8264.
S. Armaroli, G. Cardillo, L. Gentilucci, M. Gianotti, A. Tolo-
[6]
Lewis Acid Promoted Ring Expansion of cis-11
melli, Org. Lett. 2000, 2, 1105Ϫ1107 and references therein.
[7]
K. Hori, T. Nishiguchi, A. Nabeja, J. Org. Chem. 1997, 62,
Ethyl (4S,5S)-4-{(2ЈS)-1Ј,4Ј-Dioxaspiro[4,5]dec-2Ј-yl}-2-phenyl-4,5-
dihydro-1,3-oxazole-5-carboxylate (12): MgBr2·Et2O (0.086 g,
0.33 mmol) was added under an inert gas to a solution of 11
(0.12 g, 0.33 mmol) in THF (5 mL), and the mixture was refluxed
for 8 h. The solvent was removed under reduced pressure, a satur-
ated solution of NaHCO3 (5 mL) was added, and the mixture was
extracted three times with CH2Cl2 (5 mL). The collected organic
layers were dried with Na2SO4, and the solvent was evaporated
under reduced pressure. The mixture was purified by flash chroma-
tography on silica gel (cyclohexane/EtOAc, 80:20) to afford 10
(0.054 g, 45%) and an inseparable mixture of bromo derivatives and
12 (30%, according to the reaction mixture analysis). Ϫ Compound
12: 1H NMR (CDCl3): δ ϭ 1.25 (t, J ϭ 7.1 Hz, 3 H, CH3),
1.30Ϫ1.75 (m, 10 H, CH2), 4.10 (dd, J ϭ 6.3, 7.3 Hz, 1 H, OCH2),
4.20 (dd, J ϭ 6.3, 7.0 Hz, 1 H, OCH2), 4.24 (q, J ϭ 7.1 Hz, 2 H,
CH2CH3), 4.42Ϫ4.57 (m, 1 H, OCH), 4.61 (dd, J ϭ 3.3, 10.8 Hz,
1 H, CHN), 5.15 (d, J ϭ 10.8 Hz, 1 H, CHCϭO), 7.40Ϫ7.58 (m,
3 H, ArH), 7.97Ϫ8.10 (m, 2 H, ArH). Ϫ 13C NMR (CDCl3): δ ϭ
14.0, 24.0, 25.5, 29.7, 34.5, 35.6, 62.0, 65.0, 71.5.0, 76.0, 77.7, 108.6,
128.7, 130.8, 169.3, 180.4. Ϫ GC-MS: m/z (%) ϭ 359 (5) [Mϩ], 330
(2), 316 (8), 244 (5), 216 (6), 172 (78), 141 (100), 117 (6), 105 (24).
3081Ϫ3088.
[8]
F. Ferraris, W. J. Drury III, C. Cox, T. Lectka, J. Org. Chem.
1998, 63, 4568Ϫ4569 and references therein.
[9]
G. Casiraghi, F. Zanardi, Chem. Rev. 1995, 95, 1677Ϫ1716.
[10]
C. Palomo, M. Oirbiade, A. Esnal, A. Landa, J. I. Mirando,
A. Linden, J. Org. Chem. 1998, 63, 5838Ϫ5856.
[11]
A. K. Ghosh, Y. Wang, J. Org. Chem. 1999, 64, 2789Ϫ2795
and references therein.
[12]
M. Bols, I. Lundt, Acta Chem. Scand. 1991, 45, 280Ϫ284.
[13]
A. Dureault, F. Carreaux, J. C. Depezay, Synthesis 1991, 2,
150Ϫ155.
[14]
K. Jähnish, Liebigs Ann./Recueil 1997, 757Ϫ760.
H.-D. Ambrosi, W. Duczek, M. Ramm, E. Gründemann, B.
[15]
Schulz, K. Jähnish, Liebigs Ann. Chem. 1994, 1013Ϫ1018.
K. Jähnish, F. Tittelbach, E. Gründemann, M. Schneider, Eur.
[16]
J. Org. Chem. 2000, 3957Ϫ3960.
[17]
P. Garner, O. Dogan, S. Pillai, Tetrahedron Lett. 1994, 35,
1653Ϫ1656.
[18]
G. Cardillo, L. Gentilucci, C. Tomasini, M. P. V. Castejon-
Bordas, Tetrahedron: Asymmetry 1996, 7, 755Ϫ762.
[19]
J. A. Marshall, J. D. Trometer, D. G. Cleary, Tetrahedron 1989,
45, 391Ϫ402.
[20]
T. W. Greene, P. G. M. Wuts, in Protective Groups in Organic
Synthesis, Wiley-Interscience, New York, Chichester, Brisbane,
Toronto, Singapore, 1991, pp. 123Ϫ127.
Ethyl (2R,3S) 3-Benzoylamino-3-{(2ЈS)-1Ј,4Ј-dioxaspiro[4,5]dec-2Ј-
yl}-2-hydroxypropanoate: trans-Oxazoline 10 (0.19 g, 0.53 mmol)
was dissolved in THF (5 mL) at 0 °C and treated with 0.05  HCl
(3 mL). After this had stirred for 4 h, the mixture was neutralized
with sat. NaHCO3, and the THF was evaporated under reduced
[21]
P. G. Baraldi, R. Bazzani, S. Manfredini, D. Simoni, M. J.
Robins, Tetrahedron Lett. 1993, 34, 3177Ϫ3180.
T. Sugiyama, H. Sugawara, M. Watanabe, K. Yamashita,
Agric. Biol. Chem. 1984, 48, 1841Ϫ1844.
[22]
3550
Eur. J. Org. Chem. 2001, 3545Ϫ3551

The Ring Expansions of Glyceraldehyde-Derived Aziridine-2-carboxylates to Oxazolines
FULL PAPER
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
R. Annunziata, M. Cinquini, F. Cozzi, G. Dondio, L. Rai-
mondi, Tetrahedron 1987, 43, 2369Ϫ2380.
HYPERCHEM 5.1 Release Pro for Windows Molecular
Modeling System, Hypercube Inc, Copyright, 1999.
CHEMPLUSTM Release, Hypercube Inc, Copyright,
1993؊1997).
G. Righi, R. D’Achille, C. Bonini, Tetrahedron Lett. 1996, 37,
6893Ϫ6896.
D. Tanner, C. Birgersson, H. K. Dhaliwal, Tetrahedron Lett.
1990, 31, 1903Ϫ1906.
J. Legters, L. Thijs, B. Zwanenburg, Recl. Trav. Chim. Pais-Bas
1992, 111, 59Ϫ68.
L. Dubius, A. Meth, E. Tourette, R. H. Dodd, J. Org. Chem.
1994, 59, 434Ϫ441.
˚
˚
NϪC-2 and NϪC-3: 1.48 A; benzoyl CϪO: 1.21 A; benzoyl
˚
˚
CϪN: 1.44 A; C-2ϪC-3: 1.51 A.
˚ ˚
NϪC-2 and NϪC-3: 1.45 A; benzoyl CϪO: 1.27 A; benzoyl
˚
˚
CϪN: 1.40 A; C-2ϪC-3: 1.51 A.
Received February 20, 2001
[O01081]
Eur. J. Org. Chem. 2001, 3545Ϫ3551
3551