The importance of electrostatic interactions in the stereoselective 1,3-dipolar cycloadditions of nitrones to chiral allyl ethers: An experimental and force field approach

Article abstract of DOI:10.1002/(SICI)1099-0690(199809)1998:9<1823::AID-EJOC1823>3.0.CO;2-H

The 1,3-dipolar cycloaddition of formaldehyde N-benzylnitrone with β′-alkoxy- and γ-alkoxy-α,β-unsaturated esters was investigated. The stereochemical outcome of these reactions was nicely rationalized on the basis of an interpretation of the inside alkox

Full text of DOI:10.1002/(SICI)1099-0690(199809)1998:9<1823::AID-EJOC1823>3.0.CO;2-H

FULL PAPER
The Inside Alkoxy Effect Revisited
The Importance of Electrostatic Interactions in the Stereoselective 1,3-Dipolar
Cycloadditions of Nitrones to Chiral Allyl Ethers: An Experimental and Force
Field Approach
Rita Annunziata, Maurizio Benaglia, Mauro Cinquini, Franco Cozzi, and Laura Raimondi*
´
Centro CNR and Dipartimento di Chimica Organica e Industriale, Universita di Milano
via Golgi 19, I-20133 Milano, Italy
Received March 30, 1998
Keywords: Nitrones / 1,3-Dipolar cycloadditions / Chiral allyl ethers / Transition state modeling
The 1,3-dipolar cycloaddition of formaldehyde N-
electrostatic interactions in the reaction TS. The force field
benzylnitrone with βЈ-alkoxy- and γ-alkoxy-α,β-unsaturated approach previously developed for evaluating the
esters was investigated. The stereochemical outcome of
these reactions was nicely rationalized on the basis of an
interpretation of the inside alkoxy theory emphasizing the
stereoselection in nitrile oxide cycloadditions to chiral
alkenes was successfully extended to nitronesЈ reactions.
Introduction
Figure 2. The inside alkoxy theory: ab initio results. ∆E in kcal/mol
(ref.[2a] [a]
)
1,3-Dipolar cycloadditions to alkenes are reactions of
broad synthetic versatility, and have been extensively stud-
ied from both the experimental and the theoretical point of
view.[1] HoukЈs inside alkoxy theory[2] offers a powerful tool
for predicting the stereochemical outcome of the reaction:
the theory, originally proposed for nitrile oxide reactions
with terminal alkenes to give 5-substituted 4,5-dihydroisox-
azoles, predicts the predominant formation of the 5,5Ј-anti
stereoisomer (Figure 1).
Figure 1. Anti and syn cycloadducts formed in the reaction of a
nitrile oxide with a terminal alkene bearing a stereocenter in the
allylic position
[a]
Values in parentheses are from ref.[2c]
.
minus attacking the alkene carbon close to the stereocenter
in the transition state (Figure 3).[1c][4][5]
Figure 3. Atomic charges on the 1,3-dipolar unit in the
A rationale was found in the conformational preference
for the alkoxy residue at the stereocenter to occupy the in-
side position in the transition structure of the reaction (Fig-
ure 2). This preference was believed to originate from stere-
oelectronic factors.[2a][2b] A force field approach was suc-
cessfully developed from ab initio calculations to allow a
prediction of the reaction stereoselectivity for inter- and in-
tramolecular nitrile oxide plus chiral alkenes reactions.[2c][3]
After HoukЈs and JägerЈs studies on nitrile oxide cyclo-
additions, the reactions of chiral allyl ethers with others di-
RHF/3-21G TS (ref.[4]
)
Indeed, oxygen-containing 1,3-dipoles such as nitrile ox-
poles were investigated by othersЈ and our group.[1b] To ten- ides and nitrones exhibit the strongest partial negative
tatively explain the different levels of stereoselectivity exper- charge on the terminal oxygen atom and lead to the higher
imentally observed for these reactions we proposed a stereoselectivity. On the other hand, when the dipole ter-
rationale that, within the framework of the inside alkoxy minus is a nitrogen, both the negative charge and the stereo-
theory, correlates the stereoselection of the cycloaddition selectivity decrease, and the diastereoisomeric ratios mainly
with the different atomic charges featured by the dipole ter- depend on steric effects. Therefore, the origin of the stereo-
Eur. J. Org. Chem. 1998, 1823Ϫ1832
WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1998
1434Ϫ193X/98/0909Ϫ1823 $ 17.50ϩ.50/0
1823

R. Annunziata, M. Benaglia, M. Cinquini, F. Cuzzi, L. Raimondi
FULL PAPER
selectivity was mainly ascribed to electrostatic rather than dipole in the force field should allow for a reliable predictiv-
to orbital interactions, and was believed to be due to the ity in all different cases.
destabilizing repulsion generated between the negatively
charged terminus of the 1,3-dipole and the electron rich al-
lylic oxygen in the outside position (Figure 2).
Cycloadditions to βЈ-Alkoxy-α,β-Unsaturated Esters
Experimental support to this interpretation was collected
As mentioned above, βЈ-alkoxy-α,β-unsaturated esters
comparing the reactions of benzonitrile oxide and diazo- (the so-called Baylis-Hillman adducts), differently protected
methane with a series of βЈ-alkoxy-α,β-unsaturated esters at the allylic oxygen, were selected as substrates to compare
(see below).[5] Gratifyingly, this rationalization turned out the stereochemical outcome of the reactions of elec-
to be helpful also to explain the surprising stereochemical tronically different but sterically analogous (both are linear
outcome of the diazomethane cycloaddition to γ-alkoxy- dipoles) benzonitrile oxide and diazomethane.[1][4][5] The
α,β-unsaturated esters and diesters that mainly produces the experimental results showed that these cycloadditions pre-
syn adducts (Figure 4).[6] This result can nicely be explained dominantly afforded the syn isomer as predicted by the in-
by taking into account the stabilizing attractive interaction side alkoxy rationale.[9] However, the benzonitrile oxide re-
between the electron rich allylic oxygen in the outside posi- actions were more stereoselective and more sensitive to the
tion and the positively charged carbon atom of diazometh- nature of the allylic oxygen than the diazomethane ones.
ane (Figure 3). Thus, this new interpretation of the inside Thus, the preferential formation of the syn adducts and the
alkoxy theory seemed general enough to maintain its val- differences in the diastereoisomeric ratios were in full agree-
idity and predictive power also for reactions affording a ment with the rationale that relates the sense and the extent
non-conventional stereochemical result and even not fol- of the stereoselectivity to the different amount of negative
lowing the inside alkoxy model.
charge on the dipole terminus in the transition structure:
the higher the negative charge, the higher the inside alkoxy
stereoselection.[5]
Figure 4. Diazomethane cycloaddition to γ-alkoxy-α,β-unsaturated
esters (ref.[6]
)
To complete this study and to further validate the electro-
static interpretation of the inside alkoxy theory, in the pres-
ent study a series of nitrone cycloadditions on the same
Baylis-Hillman adducts were performed. In order to avoid
the formation of a new stereocenter at C-3 of the generated
isoxazolidine, the reactions were carried out with a non-
stereogenic nitrone. This was prepared by reaction of for-
maldehyde and benzyl hydroxylamine in the presence of 4
˚
A molecular sieves (r.t., overnight) and was not isolated but
directly reacted with alkenes 1؊12 (1Ϫ18 hours, 80°C).
Chromatographic purification of the crude products al-
lowed isolation of isoxazolidines 13a,bϪ24a,b; all results
are collected in Table 1.
The syn/anti ratios were determined by 1H- and 13C-
NMR spectroscopy, and the relative configurations were as-
To conclude our investigations on the origin of stereo- signed to the cycloadducts on the basis of NMR analysis
control in the 1,3-dipolar cycloadditions to chiral allyl and chemical correlation. All major isomers obtained from
ethers, we decided to extend our study to nitrone cycload- the n-propyl derivatives were chemically interconverted:
ditions. A nitrone is electronically very similar to a nitrile thus, from 15a (purified by flash chromatography) both
oxide (both dipoles contain the same CϪNϪO dipolar unit pure 19a and 23a were obtained (Scheme 1). The obser-
and have similar negative charges on the oxygen atom ), but vation of chemical shift trends for diagnostic protons (for
is sterically quite different , being a bent and not a linear instance, HC-5 always resonates at lower fields in the major
1,3-dipole. In some previous studies we already observed isomers of 13؊24; for a collection of relevant NMR data
some analogies between the sense of the stereoselectivity of see Table 4 in the Experimental Section), allowed us to as-
inter-[7] and intramolecular[8] nitrone and nitrile oxide[1b] sign the same relative configuration to all major adducts a.
cycloadditions to chiral allyl ethers. Our goal in the present The unambiguous establishment of the syn relative stereo-
study was to compare the stereoselectivity of the cycload- chemistry was obtained by converting pure 14a into aceton-
ditions of nitrones with βЈ-alkoxy- and γ-alkoxy-α,β-un- ide 26a via the diol 25a (Scheme 1).
saturated esters to those observed for nitrile oxides and di-
NMR analysis of this rigid, bicyclic compound revealed
azomethane reactions on similar substrates.[1b][5][6] In ad- strong NOEs between the two axial protons HC-5Ј and
dition, we wanted to extend the force field treatment of ni- HC-5ЈЈ, and between these and the same (pro-S) H at C-
trile oxide plus chiral alkene cycloadditions[2c][3] to other 4 of the isoxazoline ring. Molecular Mechanics (MM2*)
1,3-dipoles with the hope that, if indeed the different stereo- calculations performed on the diastereoisomeric acetonides
selectivities are due to the different atomic charges on the 26a and 26b confirmed our stereochemical assignment,
dipole in the TS, simply changing the parameters for the since only a cis relationship between these three hydrogens
1824
Eur. J. Org. Chem. 1998, 1823Ϫ1832

The Inside Alkoxy Effect Revisited
FULL PAPER
Table 1. Nitrone cycloadditions to Baylis-Hillman adducts
perature can account for this result, and indeed, when
cycloadded to alkenes at the same temperature (20°C), the
two dipoles exhibited similar stereoselections (Table 1, # 13,
14). The characteristic effect exerted on the stereoselection
by the different protective groups of the allylic oxygen in
the nitrile oxide reactions[5] was maintained also in nitrone
cycloadditions; thus the syn/anti ratios observed for the silyl
ethers 5؊8, featuring an electron rich, silicon protected al-
lylic oxygen, were higher than those observed for the corre-
sponding acetates 9؊12, featuring an electron poor allylic
oxygen (Table 1, # 5Ϫ8 vs 9Ϫ12). When the reaction tem-
perature was decreased from 80° down to 20°C, the allylic
oxygen effect on the stereoselectivity became more evident,
the syn stereoselectivity increasing more for the silyl than
for the acetyl derivatives (Table 1, # 13 vs 5 and 14 vs 11).
This dependence of the stereoselection on the nature of the
allylic oxygen nicely fits in our rationale based on electro-
static interaction: the more electron rich the allylic oxygen,
the stronger the electrostatic repulsion with the negatively
charged dipole oxygen when the allylic oxygen is in the out-
side position, the higher the inside alkoxy stereoselectivity.
In concluding the discussion on these cycloadditions, the
remarkably high level of syn stereoselectivity observed in
the reactions carried out on the unprotected alcohols 1؊4
must be commented. Generally,[1b] unprotected chiral al-
lylic alcohols undergo poorly stereoselective 1,3-dipolar
cycloadditions, the stereochemical outcome being the result
of the contrasting tendency of the OH group to occupy the
inside position for stereoelectronic reasons, or the outside
one to form a H bond with the negatively charged terminus
of the incoming dipole. In the case of substrates 1؊4, the
formation of a H bond between the OH group and the
nitrone oxygen would lead to the production of the anti
diastereoisomer, and thus can be ruled out on the basis of
the observed stereoselections. On the other hand, the cyclo-
addition can occur on a substrate conformation in which
the OH group is H-bonded to the carbonyl oxygen, but,
this being the case, the formation of the syn isomer would
arise from a sterically very unfavorable attack of the nitrone
oxygen antiperiplanar to the C-H bond at the stereocenter.
It is likely therefore that the reaction occurs on the inside
hydroxy conformation of alcohols 1-4, particularly stabil-
ized by the reduced steric requirements of the OH group,
that, being smaller than the OTBS or the OCOMe group of
5؊12, more easily occupy the inside position.[10] A similar
behaviour was observed also for benzonitrile oxide and di-
azomethane cycloadditions to alcohols 1Ϫ4.[5]
Entry Alkene
R
X
Product Yield% syn/anti
1
1
Me
Ph
n-Pr
i-Pr
Me
Ph
n-Pr
i-Pr
Me
Ph
n-Pr
i-Pr
Me
n-Pr
H
13a,b
14a,b
15a,b
16a,b
17a,b
18a,b
19a,b
20a,b
21a,b
22a,b
23a,b
24a,b
17a,b
23a,b
74
73
82
68
52
95
53
72
88
80
84
77
68
52
88:12
81:19
90:10[a]
80:20
63:37
68:32
74:26
83:17
60:40
52:48
67:33
65:35
83:17[c]
70:30[d]
2
2
H
3
3
H
H
4
4
5
5
TBS
TBS
TBS
TBS
Ac
6
6
7
7
8
8
9
9
10
11
12
13[b]
14[b]
10
11
12
5
Ac
Ac
Ac
TBS
Ac
11
Diastereoisomeric ratio determined via 13C-NMR spectroscopy.
[a]
Ϫ [b] reaction performed for 48 hours at 20°C. Ϫ [c] in the reaction
of 5 with benzonitrile oxide a syn/anti ratio of 86:14 was obtai-
[d]
ned.[b]
Ϫ
in the reaction of 11 with benzonitrile oxide a 77:23
syn/anti ratio was obtained.[b]
Scheme 1. Chemical correlation between compounds 15a, 19a, and
23a, and synthesis of compound 26a
Nitrone Cycloadditions to γ-Alkoxy-α,β-Unsaturated Esters
and Diesters
(as in 26a), calling for a syn relative configuration of 14a,
can account for the experimentally observed NOEs.
As mentioned in the introduction, diazomethane has
In commenting these results we can say that the sense of been found to react with γ-alkoxy-α,β-unsaturated esters
stereoselectivity of these nitrone cycloadditions to Baylis- and diesters with complete regiocontrol and good to excel-
Hillman adducts agrees with that of nitrile oxide and diazo- lent degrees of syn stereoselectivity (Figure 4).[6] In previous
methane reactions, predominantly producing the syn cyclo- studies on the cycloadditions of a nitrile oxide to these sub-
adduct.[5] However, the syn/anti ratios observed in the strates,[1c][11] the reaction was shown to be poorly regiose-
nitrone reactions were lower than those observed for the lective. The available data, however, indicate a moderate
benzonitrile oxide ones. Likely the different reaction tem- anti stereoselection for the regioisomer in which the dipole
Eur. J. Org. Chem. 1998, 1823Ϫ1832
1825

R. Annunziata, M. Benaglia, M. Cinquini, F. Cuzzi, L. Raimondi
Also in the case of the 5-methoxycarbonyl substituted
FULL PAPER
oxygen attacks the β-carbon of the unsaturated ester. On
the basis of the electronic similarity between nitrile oxides compounds 30c,d؊34c,d, the preferential formation of one
and nitrones, we anticipated a poor regiocontrol also for of the two isomers, namely d, was observed. Chemical shift
the nitrone cycloadditions. Nevertheless, we thought that trends strongly suggested that all the major isomers fea-
evaluation of syn/anti stereoselectivity in both regioisomers tured the same relative configuration at the C-4/C-4Ј stereo-
arising from the attack of differently charged dipole termini centers. Unambiguous configurational assignment required
to the alkene carbon close to the stereocenter could be a the transformation of one of the cycloadducts into a rigid
probing test for assessing the real importance of electro- derivative amenable to NMR analysis. Thus, isoxazolidine
static effects in the stereochemical outcome of these reac- 32d was converted (40% aqueous HF, THF, r.t., Scheme 3)
tions.
into the bicyclic lactone 35. The observation of NOEs be-
Thus, esters (E)- and (Z)-27, (Z)-28, and (E)- and (Z)-29 tween HC-5, HC-4, and HC-4Ј indicated that all these pro-
were prepared and reacted with formaldehyde NϪbenzylni- tons are located on the same side of the convex structure,
trone (1Ϫ4 h, 80°C) to afford isoxazolidines 30؊34 as mix- and allowed the attribution of the 4,5-cis-4,4Ј-syn configu-
tures of regio- and diastereoisomers (Scheme 2). Chro- ration to 32d.
matographic purification followed by 1H- and 13C-NMR
analysis (see Tables 5 and 6 in the Experimental Section)
allowed regioisomer identification and evaluation of dia-
stereoisomeric ratios.
Scheme 3. Synthesis of bicyclic lactone 35
Scheme 2. Nitrone cycloaddition to (E)- and (Z)-27Ϫ32
In commenting these data the following observations can
be made. In the case of the cycloadditions leading to a, b
adducts, the observed anti stereoselectivity is in agreement
with the sense of stereoselection observed in nitrile oxide
cycloadditions to similar substrates,[1c][11] and confirms the
tendency of nitrones to behave similarly to nitrile oxides
and to react with chiral allyl ethers as predicted by the in-
side alkoxy theory (anti stereoselection).[12] Since no quanti-
tative data were available for nitrile oxide cycloadditions,
comparison of the levels of stereocontrol for these reactions
was not possible.
The results of the nitrone cycloadditions leading to c, d
isoxazolidines are much more interesting, since their stereo-
selectivity can be compared qualitatively and quantitatively
with that of the diazomethane reactions with the same sub-
strates. First of all, it must be noted that both reactions are
always syn stereoselective.[6][13] As mentioned before, this
result can be rationalized on the basis of a favorable electro-
static interaction between the partially positively charged C
terminus of both dipoles and the allylic oxygen located in
the outside position. Secondly, when the reactions are indi-
vidually compared, the nitrone cycloadditions are con-
stantly slightly less stereoselective than the diazomethane
ones. Even if this comparison must be cautiously considered
(the two dipoles clearly have different steric requirements
As far as the regioselection is concerned, the 4-meth- and react at different temperatures), it is remarkable that
oxycarbonyl substituted derivatives 30a,b؊34a,b were al- the dipole featuring the more positively charged C atom (i.e.
ways obtained in minor amount. For these compounds the diazomethane) reacts more stereoselectively, in full agree-
stereoselectivity ranged from fair to excellent in favor of the ment with our rationalization of the inside alkoxy theory
a isomers, to which, on the basis of NMR chemical shift that relates the extent of stereoselection to the dipole/sub-
trends and of comparison of NMR data with those already strate electrostatic interaction in the transition state.
available for similar cycloadducts,[7][8] the 5,5Ј-anti configu-
As those to esters 27Ϫ29, also the nitrone cycloadditions
ration was assigned. The attribution of the stereochemistry to diesters 36؊39 were poorly regioselective, generally af-
at C-4 and C-5 was trivial, the relative configuration at fording a slight excess of the 5,5-dimethoxycarbonyl substi-
these stereocenters being pre-determined by the alkene tuted compounds, that, as expected, were obtained only as
structure.[1]
the syn isomers 40d؊43d (Scheme 4).
1826
Eur. J. Org. Chem. 1998, 1823Ϫ1832

The Inside Alkoxy Effect Revisited
FULL PAPER
Scheme 4. Nitrone cycloadditions to 36؊39
parameters to be included in commonly used force fields.
In this way, energy differences can be evaluated for dia-
stereoisomeric transition structures, as is usually done for
diastereoisomeric ground states. Because of its good per-
formances in the treatment of small organic molecules,
MM2 force field[16] was selected to be parameterized. The
“Transition State Modeling” has been applied successfully
to various organic reactions,[15] and in particular to nitrile
oxide cycloadditions to chiral alkenes.[2c][3] This set of pa-
rameters was subsequenty included as a substructure in the
MacroModel/Batchmin[17] MM2* force field, thus allowing
application of the usual conformational analysis techniques
(e.g. Monte Carlo procedures) to the nitrile oxide plus chiral
alkene transition structures.
It is well known in the literature that itЈs the nature of the
chiral alkene that determines the sense of stereoselectivity in
the 1,3-dipolar cycloaddition reactions. However, the role
of the 1,3-dipole was deeply investigated, and its effect on
the reaction stereoselectivity was correlated to its struc-
ture[1] as well as to the charge distribution on the dipolar
moiety in the reaction TS.[4][5][6] Thus, the stronger the
negative charge on the dipole terminus interacting with the
allylic substituent, the more destabilized the outside posi-
tion for an allylic oxygenated residue, the more reinforced
the inside alkoxy effect. On the other hand, when the dipole
terminus interacting with the allylic substituent is positively
charged, an opposite effect was observed: the outside con-
formation is electrostatically stabilized, resulting in an out-
side alkoxy effect.
We reasoned that the force field treatment of the nitrile
oxide plus alkenes could be extended to other 1,3-dipolar
cycloadditions. The parameters describing the alkene moi-
ety can be left unchanged, and different parameters can be
included to describe the different 1,3-dipoles; of course, also
atomic charges must be changed on changing the nature of
the 1,3-dipole. The original formulation of the parameter
set for the nitrile oxide plus alkene TS,[2c][3] however, treated
the electrostatic interactions by inclusion of bond dipoles,
while only atomic charges are easily obtained from ab initio
calculations. A preliminary step was thus necessary to allow
a charge-based electrostatic parameterization of the force
field. Thus the nitrile oxide plus alkenes substructure in-
cluded in MM2* force field (MacroModel/Batchmin 5.5
package)[17] was slightly modified to allow direct inclusion
of atomic charges instead of dipoles. The dipole and the
alkene non-hydrogen atoms were given charges evaluated
from fitting of the electrostatic potential at the 3-21G le-
vel,[4] with hydrogen contributions summed in the heavy
atoms to which they are bonded. The inside alkoxy param-
eters had to be slightly modified to account for the differ-
ences included by switching from empirically evaluated di-
poles to ab initio calculated atomic charges.
For one of these compounds, namely 42d, the configura-
tion was unambiguously established by NMR analysis of
the bicyclic lactone 44 obtained as described above for 35
(Scheme 5). In this case the strong NOE interaction be-
tween H-4 and H-4Ј showed the syn disposition of these
protons. The cycloadditions affording the 4,4-dimeth-
oxycarbonyl substituted regioisomers occurred with a much
lower degree of stereocontrol, generally producing a low ex-
cess of the expected anti a isomers over the syn b ones. At
present we are not able to explain this poor stereocontrol;
it must be noted, however, that the anti/syn ratios closely
resemble those observed for Michael-type reactions of the
same substrates;[14] this competitive reaction mechanism
can influence the stereoselection.
Scheme 5. Synthesis of bicyclic lactone 44
Molecular Mechanics Studies of 1,3-Dipolar Cycloadditions
When the inside alkoxy theory was proposed for the ni-
The performance of the force field was tested against pre-
trile oxide plus alkene reaction, a force field was param- vious calculations to verify that it was left unchanged from
eterized on the basis of ab initio data[2c][3] within the frame the original parameterization.[2c][3] Systematic pseudo-
of the “Transition State Modeling”.[15] Within this ap- Monte Carlo techniques,[18] together with subsequent en-
proach, a simple model of the reaction TS is studied with ab ergy minimization (MC/EM), were applied to the confor-
initio (or semiempirical) methods, thus obtaining suitable mational analysis of the transition structures for all isomers
Eur. J. Org. Chem. 1998, 1823Ϫ1832
1827

R. Annunziata, M. Benaglia, M. Cinquini, F. Cuzzi, L. Raimondi
FULL PAPER
(see the Experimental Section for details). Selected results Table 3. Calculated vs experimental anti/syn ratios for nitrone
cycloadditions to alkenes
are collected in Table 2.
Table 2. Calculated vs experimental anti/syn ratios for nitrile oxide
cycloadditions to alkenes
In all calculations, Bn was replaced with Me, and TBS with TMS.
Ϫ
[a] For previously calculated and experimental anti/syn ratios, see
[b] [c]
ref.[2c]. Ϫ
For experimental anti/syn ratios, see ref.[5]. Ϫ
For
experimental anti/syn ratios, see refs.[1c] and[11]
.
Gratifyingly, the behaviour of the force field was left un-
changed from the original parameterization in all test calcu-
lations (Table 2, # 1Ϫ4).[2c][3] We thus proceeded to evalu-
ate the diastereoisomeric ratios for benzonitrile oxide cyclo-
additions to Baylis-Hillman adducts, and were pleased to
observe also in this case a good agreement with the exper-
imental data (# 5, 6). Moreover, the performance of the
In all calculations, Bn was replaced with Me, and TBS with TMS.
[a]
[b]
Ϫ
For experimental anti/syn ratios, see ref.[8a]. Ϫ For experi-
mental anti/syn ratios, see ref.[8b]. Ϫ
For experimental anti/syn
[c]
ratios, see ref.[7]
.
force field remains reliable also when the regioisomeric man adducts (# 6Ϫ9), and for intermolecular reactions of
cycloadducts are considered: when the carbon atom of the γϪalkoxy-α,β-unsaturated esters (# 10Ϫ13). In this last
nitrile oxide is the terminus interacting with the allylic oxy- case, the different stereoselectivity obtained within the dif-
gen in the reaction TS, the cycloaddition is slightly syn- ferent regioisomers is well reproduced by the calculation:
selective (# 7, 8).[1c][11] Also in this case, the agreement within the 4-methoxycarbonyl substituted regioisomers the
within the experimental and the calculated data is com- 5,5Ј-anti cycloadduct is favored, deriving from an inside al-
pletely satisfying.
koxy conformation of the TS. The TS leading to the 5-
The subsequent parameterization of MM2* force field methoxycarbonyl substituted regioisomers, on the other
for nitrone plus alkene cycloadditions was almost trivial. To hand, prefers an outside alkoxy conformation. The differ-
make the matter simpler, a new substructure was included ence in the two regioisomeric TSs is mainly due to electro-
for this TS in MM2* force field, with geometric parameters static interactions within the oxygenated allylic substituent
derived from BernardiЈs RHF/4-31G transition structure[19] and the 1,3-dipolar moiety.
and atomic charges derived from 3-21G fitting of the elec-
A final observation can be derived from the calculated
trostatic potential.[4] All the parameters involving the al- TSs. In the cycloaddition to non-conjugated alkenes, the
kene portion were exactly the same included in the nitrile alkyl residue on the alkene bearing the allylic stereocenter
oxide plus alkene substructure. Conformational analysis of tends to occupy the exo region in the transition structure.
the transition structures (MC/EM procedure) were per- For intramolecular reactions on (E)-alkenes, this effect is
formed considering both endo and exo approaches between counteracted by the same configurational preference of the
the reactants. Some results (calculated vs experimental dia- connecting chain. When the reaction on α,β-unsaturated es-
stereoisomeric ratios) are collected in Table 3.
ters is examined, on the other hand, the endo region is al-
The calculated diastereoisomeric ratios are always in ways occupied by the activating ester group, both in inter-
agreement with the experimentally observed ones; this is and in intramolecular reactions. This fact seems to suggest
true for intramolecular reactions on electron rich as well as that the secondary orbital overlap, responsible for this ef-
electron poor alkenes (Table 3, # 1Ϫ5), for intermolecular fect, can be simulated within a force field treatment via elec-
cycloadditions on terminal alkenes Ϫ including Baylis-Hill- trostatic interactions.
1828
Eur. J. Org. Chem. 1998, 1823Ϫ1832

The Inside Alkoxy Effect Revisited
Table 4. Significant H- and 13C-NMR chemical shifts for isoxazolidines 13Ϫ26a, b
FULL PAPER
1
H-3
H-4
H-5Ј
H-5Љ
C-3
C-4
C-5
C-5Ј
13a
13b
14a
14b
15a
15b
16a
16b
17a
17b
18a
18b
19a
19b
20a
20b
21a
21b
22a
22b
23a
23b
24a
24b
25a
25b
26a
2.93
2.80; 2.35
2.80; 2.55
2.58
4.04
4.02
5.10
4.98
3.85
3.85
3.70
3.68
4.09
4.24
5.05
5.01
3.97
3.99
3.93
3.87
5.26
5.31
6.22
6.19
5.30
5.24
5.19
5.24
4.99
4.73
5.06
1.12
54.8
54.5
54.7
54.4
54.7
54.4
54.9
54.0
54.6
54.6
54.2
54.2
54.9
54.5
55.2
54.3
54.0
54.1
54.1
53.6
54.1
54.2
54.2
53.7
56.1
n.d.
55.4
31.7
35.4
32.2
35.6
33.8
35.9
32.6
37.8
32.5
31.8
32.2
31.8
30.8
34.2
30.9
35.2
34.1
34.4
34.0
34.7
33.8
34.8
33.7
36.6
30.2
n.d.
48.6
88.7
88.5
89.0
87.6
88.6
88.0
88.4
87.5
89.4
88.5
89.8
89.0
89.5
88.7
90.0
88.7
86.9
86.4
87.2
86.9
87.0
86.4
87.4
86.7
87.5
n.d.
94.0
69.5
70.7
75.9
76.5
73.0
74.4
77.8
78.5
69.8
70.4
76.8
75.7
72.6
75.4
76.0
75.5
71.4
71.3
75.4
76.8
73.8
74.2
77.3
78.1
77.2
n.d.
85.8
2.93
1.16
2.90
2.90
2.98; 2.91
2.98; 2.91
2.97; 2.90
2.91
-
2.58
-
2.68; 2.52
2.68; 2.52
2.72; 2.59
2.71; 2.57
2.77; 2.38
2.77; 2.38
2.60
2.40; 2.80
2.82; 2.41
2.71; 2.31
2.85; 2.52
2.70; 2.37
2.70; 2.33
2.70; 2.33
2.65; 2.40
2.79; 2.40
2.73; 2.32
2.67; 2.32
2.80; 2.42
2.57; 2.33
2.60; 1.83
2.50; 2.30
2.60; 1.73
-
-
-
-
3.02; 2.77
3.02; 2.77
2.93; 2.07
2.83; 2.07
3.08; 2.76
2.97; 2.83
3.12; 2.69
2.97; 2.79
3.01; 2.78
3.01; 2.78
2.96; 2.40
2.87; 2.40
3.00; 2.73
3.00; 2.73
3.01; 2.77
3.01; 2.77
2.86
1.10
1.19
-
-
-
-
-
-
1.22
1.24
-
-
-
-
-
-
3.53; 3.46
4.03; 3.90
3.47; 3.20
2.86
2.90; 2.73
[a]
In all calculations, Bn was replaced with Me, and TBS with TMS. Ϫ For previously calculated and experimental anti/syn ratios, see
[b]
[c]
ref.[2c]. Ϫ For experimental anti/syn ratios, see ref.[5]. Ϫ For experimental anti/syn ratios, see refs.[1c] and[11]
.
Conclusions
zene[20] was distilled from Na, Et2O from Na and subsequently
from LiAlH4, CH2Cl2 from CaH2; these solvents were stored on 4-
Atomic charges on the transition structure allow for a
rationale of the stereoselectivity observed in different 1,3-
dipolar cycloadditions. It is well known that the sense of
the stereoselectivity is determined by the nature of the al-
kene, while different stereoisomeric ratios are obtained by
varying the 1,3-dipole. This effect was shown to be depend-
ent on the differences in the charge distribution in the reac-
tion TS, while the TS geometry plays a minor role. In-
clusion of force field parameters for the different 1,3-di-
polar moieties of the TS, while leaving unchanged the al-
keneЈs parameterization, allows for a reliable evaluation of
the reaction stereoselectivity.
˚
A molecular sieves. Commercial DMF was used without additional
˚
purification and stored on 4-A molecular sieves; pyridine was
stored on KOH. All reactions employing dry solvents were per-
formed in a nitrogen atmosphere.
Alkenes 1Ϫ12,[5] 27Ϫ29,[6] are known compounds; 36Ϫ39 were
prepared from the appropriated aldehyde following the known pro-
cedure.[14a] All spectroscopic data were in agreement with the
chemical structure.
General Procedure for the 1,3-Dipolar Cycloaddition of Formal-
dehyde N-Benzylnitrone to Alkenes 1؊12, 27؊29, 36؊39: To a 0.1
 solution of benzyl hydroxylamine (1.5 eq) in dry benzene, pa-
˚
raformaldehyde (8 eq) and 4 A molecular sieves were added, and
Special thanks are due to Valentina Molteni, Valeria Affaticati,
Alberto Palaveri and Domenico Zanferrari for their helpful collab-
oration.
the mixture stirred overnight under nitrogen. A 0.1  solution of
the appropriate alkene (1 eq) in dry benzene was then added, and
the reaction temperature was raised to 80°C. The reaction was
monitored via TLC; after 1.5Ϫ8 hours, the reaction mixture was
cooled to 20°C and filtered. After removal of the solvent, the crude
was purified by flash chromatography (hexanes/Et2O as eluant). All
cycloadducts were obtained as oils. Chemical yields are reported in
Table 1 and Schemes 2 and 4; significant 1H- and 13C-NMR chemi-
cal shifts are collected in Tables 4Ϫ6.
Experimental Section
CHN Analyses: Perkin-Elmer 240 instrument. Ϫ 1H and 13C-
NMR: Bruker AM300 at 300.133 and 75.47 MHz, respectively;
CDCl3 as solvent; chemical shifts are reported in δ relative to TMS.
Ϫ All NMR spectra were recorded at 50°C.
Silica gel (230Ϫ400 mesh) was used for flash chromatography.
Organic extracts were dried with Na2SO4 and filtered before re-
13a, b: hexanes/Et2O, 2:8. C14H19NO4: calcd. C 63.38, H 7.22,
N 5.28; found C 63.42, H 7.15, N 5.33. Ϫ 14a, b: hexanes/Et2O,
moval of the solvent. Dry solvents were prepared as follows: ben- 3:7. C19H21NO4: calcd. C 69.71, H 6.47, N 4.28; found C 69.71, H
Eur. J. Org. Chem. 1998, 1823Ϫ1832 1829

R. Annunziata, M. Benaglia, M. Cinquini, F. Cuzzi, L. Raimondi
Table 5. Significant H- and 13C-NMR chemical shifts for cycloadducts 30Ϫ43a, b.
FULL PAPER
1
H-3
H-4
H-5
H-5Ј
C-3
C-4
C-5
C-5Ј
30a
30b
31a
32a
32b
33a
33b
34a
40a
40b
41a
41b
42a
42b
43a
43b
3.27
3.48
4.38
4.43
4.18
4.07
4.63
4.36
4.42
4.23
4.76
4.78
4.70
4.76
4.54
4.67
4.80
4.74
3.74
3.74
3.90
4.11
4.21
4.17
4.32
4.23
3.91
3.95
4.08
4.07
4.13
4.24
4.23
4.37
58.0
57.9
58.2
58.3
-
57.9
57.6
57.8
63.0
62.1
62.5
62.9
63.4
63.7
62.3
62.1
49.0
48.7
48.2
48.0
-
50.1
49.0
48.7
65.7
65.8
65.8
61.5
66.0
64.3
n.d.
65.8
83.9
83.2
82.1
83.6
-
80.9
79.9
80.0
84.6
84.7
84.6
84.7
85.7
86.2
81.8
81.0
75.2
74.2
73.6
67.4
-
76.2
75.5
73.7
73.6
73.0
72.7
72.6
67.9
67.6
74.0
74.2
3.20
3.32
3.27; 3.13
3.33; 3.13
3.27; 3.13
3.20
3.27
3.23
4.00; 3.92
4.10; 3.95
3.30; 3.20
3.23
4.03; 3.92
4.03; 3.90
3.33; 3.07
3.57
3.55
3.43
n.d.
3.45
3.37
3.60
-
-
-
-
-
-
-
-
1
Table 6. Significant H- and 13C-NMR chemical shifts for isoxazolidines 30Ϫ44c, d.
H-3
H-4
H-5
H-4Ј
C-3
C-4
C-5
C-4Ј
30c
30d
31c
31d
32d
33c
33d
34d
35
40d
41d
42d
43d
44
3.20; 2.67
3.17; 2.90
3.13; 2.90
3.33; 3.07
3.33; 3.00
n.d.
3.20; 2.93
3.30; 3.03
3.28; 2.58
3.33; 3.20
3.33; 3.20
3.37; 3.00
3.33; 3.07
3.20; 2.80
3.13
3.02
3.13
3.07
3.00
n.d.
3.00
3.03
3.33
3.62
3.59
3.53
3.60
3.45
4.47
4.33
4.70
4.42
4.53
4.50
4.27
4.55
4.70
-
3.64
3.74
3.76
3.73
3.80
n.d.
4.27
4.07
4.79
3.97
n.d.
3.96
4.16
4.85
57.0
56.9
56.7
55.3
57.6
56.8
57.6
56.6
55.2
55.2
55.4
58.6
57.0
61.8
52.1
52.1
53.2
53.0
55.2
50.0
51.0
50.3
55.2
54.9
54.0
56.8
52.1
55.2
77.5
77.3
77.9
76.9
77.1
77.2
77.4
76.4
77.8
88.0
85.4
85.4
87.0
86.1
74.9
74.4
74.1
72.0
67.5
75.5
76.2
73.4
75.2
71.4
72.3
67.8
74.1
74.8
-
-
-
-
6.49, N 4.25. Ϫ 15a, b: hexanes/Et2O, 3:7. C16H23NO4: calcd. C
N 4.20. Ϫ 24a, b: hexanes/Et2O, 9:1. C18H25NO5: calcd. C 64.46,
65.51, H 7.90, N 4.77; found C 65.46, H 7.86, N 4.80. Ϫ 16a, b: H 7.51, N 4.18; found C 64.54, H 7.55, N 4.14. Ϫ 30aϪd: hexanes/
hexanes/Et2O 3:7. C16H23NO4: calcd. C 65.51, H 7.90, N 4.77; Et2O, 7:3. C21H25NO4: calcd. C 70.96, H 7.09, N 3.94; found C
found C 65.55, H 7.87, N 4.81. Ϫ 17a, b: hexanes/Et2O, 8:2.
C20H33NO4Si: calcd. C 63.29, H 8.76, N 3.69; found C 63.24, H calcd. C 70.96, H:7.09, N 3.94; found C 70.89, H 7.10, N 3.99.
8.80, N 3.72. Ϫ 18a, b: hexanes/Et2O, 1:1. C25H35NO4Si: calcd. C 32aϪd: hexanes/Et2O, 7:3. C20H33NO4Si: calcd. C 62.57, H 8.67,
70.90, H 7.12, N 3.91. Ϫ 31aϪd: hexanes/Et2O, 7:3. C21H25NO4:
67.99, H 7.99, N 3.17; found C 67.89, H 8.03, N 3.15. Ϫ 19a, b: N 3.65; found C 62.50, H 8.72, N 3.70. Ϫ 33aϪd: hexanes/Et2O,
hexanes/Et2O, 1:1. C22H37NO4Si: calcd. C 64.82, H 9.15, N 3.44; 6:4. C20H27NO5: calcd. C 66.46, H 7.53, N 3.87; found C 66.54, H
found C 64.88, H 9.20, N 3.40. Ϫ 20a, b: hexanes/Et2O, 7:3.
C22H37NO4Si: calcd. C 64.82, H 9.15, N 3.44; found C 64.77, H 66.46, H 7.53, N 3.87; found C 66.41, H 7.51, N 3.82. Ϫ 40aϪd:
9.11, N 3.47. Ϫ 21a, b: hexanes/Et2O, 9:1. C16H21NO5: calcd. C hexanes/Et2O, 7:3. C23H27NO6: calcd. C 66.81, H 6.58, N 3.39;
62.53, H 6.89, N 4.56; found C 62.59, H 6.92, N 4.58. Ϫ 22a, b: found C 66.75, H 6.53, N 3.42. Ϫ 41aϪd: hexanes/Et2O, 6:4.
7.50, N 3.90. Ϫ 34aϪd: hexanes/Et2O, 6:4. C20H27NO5: calcd. C
hexanes/Et2O, 8:2. C21H23NO5: calcd. C 68.28, H 6.28, N 3.79;
C24H29NO7: calcd. C 63.64, H 6.59, N 3.16; found C 63.57, H 6.62,
found C 68.20, H 6.32, N 3.82. Ϫ 23a, b: hexanes/Et2O, 9:1.
N 3.18. Ϫ 42aϪd: hexanes/Et2O, 75:25. C22H35NO6Si: calcd. C
C18H25NO5: calcd. C 64.46, H 7.51, N 4.18; found C 64.36, H 7.53, 60.38, H 8.06, N 3.20; found C 60.31, H 8.04, N:3.16. Ϫ 43aϪd:
1830 Eur. J. Org. Chem. 1998, 1823Ϫ1832

The Inside Alkoxy Effect Revisited
FULL PAPER
hexanes/Et2O, 6:4. C19H25NO7: calcd. C 60.15, H 6.64, N 3.69;
found C 60.09, H 6.60, N 3.74.
alkene TS.[23] This substructure was obtained starting from the one
for nitrile oxide plus alkene, previously modified by inclusion of
atomic charges instead of dipoles.[23] The new parameters for the
nitrone moiety were obtained from the RHF/4-31G transition
structure located by Bernardi et al.;[19] for the electrostatic calcu-
lations, the RHF/3-21G CHELPG charges were used.[4]
Chemical Correlation Between 15a and 19a: To compound 15a (1
mmol), dissolved in dry DMF (0.3 ml), 2 mmols of TBSCl and 2.5
mmols of imidazole were added, and the mixture allowed to stir
overnight. Water was added, and the product was extracted with
ethyl ether and purified by flash chromatography. 19a was obtained
in 95% yield.
The conformational searches were performed for each dia-
stereoisomer (when the case, in both the endo and exo approach)
in vacuo, using the systematic pseudo-Monte Carlo/energy mini-
mization (MC/EM) procedure.[18] The extraannular bonds of the
isoxazolines and isoxazolidines which can undergo free rotation
were used as variables (in the case of intramolecular cycloadditions,
the conformational mobility of the fused ring was considered by
including ring bonds as variables in the Monte Carlo search). 250
Monte Carlo steps per torsion were shown to ensure convergence
of the conformational search (on test jobs, a higher number of step
per torsional variable did not locate any new minimum within 20
kJ/mol with respect to the first 250 steps). Standard convergence
Chemical Correlation Between 15a and 23a: to compound 15a (1
mmol), dissolved in dry CH2Cl2 (2.5 ml), 1.1 mmols of pyridine
and 1.1 mmols of AcCl were added at 0°C. After 1 h, the mixture
was treated with 3.7% HCl and extracted with CH2Cl2. The crude
was purified by flash chromatography. 23a was obtained in 84%
yield.
Synthesis of 26a from 14a: to a suspension of LiAlH4 (1 mmol)
in dry Et2O (10 ml) a solution of 14a (1 mmol) in dry Et2O (5 ml)
was slowly added at 0°C. After 1 h, the reaction was quenched at
0°C by a slow addition of H2O (0.184 ml), 10% NaOH (0.184 ml)
and H2O (0.368 ml), and then filtered. Flash chromatography
(Et2O) afforded pure 25a in 58% yield as an oil. 1H and 13C-NMR
significant data are collected in Table 4. C18H21NO3: calcd. C
72.22, H 7.07, N 4.68; found C 72.29, H 7.10, N 4.64.
criteria (0.05 kJ AϪ1 molϪ1) and the truncated Newton conjugate
˚
gradient procedure[23] were used for energy minimization. All un-
iques conformations within 20 kJ molϪ1 were stored, and the dia-
stereoisomeric ratios were evaluated in each case considering a
Boltzmann distribution at the suitable temperature (the tempera-
ture at which the corresponding experimental ratio was obtained).
25a (0.2 mmols) was dissolved in 2,2-dimethoxypropane (2 ml)
and a crystal of PTSA was added. After stirring overnight, solid
NaHCO3 was added, the mixture was filtered and the crude puri-
fied by flash chromatography on Florisil (200Ϫ300 mesh; hexanes/
Et2O, 6:4), to avoid decomposition of the product. 26a was reco-
[1]
[1a] A. Padwa (Ed.); 1,3-Dipolar Cycloaddition Chemistry, Wiley,
[1b]
New York, N.Y., 1984. Ϫ
M. Cinquini, F. Cozzi, “1,3-Di-
polar Cycloadditions” in Houben Weyl E21C Ϫ Stereoselective
[1c]
Synthesis. G. Thieme, Stuttgart, 1995, p. 2953. Ϫ
mondi, Gazz. Chim. Ital. 1997, 127, 167.
L. Rai-
1
vered as an oil in 60% yield. Ϫ H and 13C-NMR significant data
[2]
are collected in Table 4. Ϫ C21H25NO3: calcd. C 74.31, H 7.42, N
4.13; found C 74.40, H 7.45, N 4.10.
[2a] K. N. Houk, S. R. Moses, Y.-D. Wu, N. G. Rondan, V. Jäger,
R. Schohe, F. R. Fronczek, J. Am. Chem. Soc. 1984, 106, 3880.
[2b]
Ϫ
K. N. Houk, H.-Y. Duh, Y.-D. Wu, S. R. Moses, J. Am.
[2c]
Synthesis of 35 from 32d: To a solution of 32d (0.14 mmols) in
THF (4 ml), 40% HF (100 µl) was added, and the mixture was
stirred at RT for 2 hours. After addition of solid NaHCO3, ad-
ditional stirring (15Ј) and filtration, the crude was purified by flash
chromatography (hexanes/Et2O 6:4; 38%). Ϫ 35 was a pale oil;
C13H15NO3: calcd. C 66.94, H 6.48, N 6.00; found C 67.00, H
Chem. Soc. 1986, 108, 2754. Ϫ
L. Raimondi, Y.-D. Wu, F.
K. Brown, K. N. Houk, Tetrahedron Lett. 1992, 33, 4409.
F. K. Brown, L. Raimondi, Y.-D. Wu, K. N. Houk, Tetrahedron
Lett. 1992, 33, 4405.
[3]
[4]
[5]
[6]
R. Annunziata, M. Benaglia, M. Cinquini, L. Raimondi, Tetra-
hedron 1993, 49, 8629.
R. Annunziata, M. Benaglia, M. Cinquini, F. Cozzi, L. Rai-
mondi, J. Org. Chem. 1995, 60, 4697.
1
6.42, N 5.95. Ϫ Significant H- and 13C-NMR data are collected
R. Annunziata, M. Benaglia, M. Cinquini, F. Cozzi, L. Rai-
mondi, in Electronic Conference on Heterocyclic Chemistry
(ECHET96). (Eds.: H. R. Rzepa, J. Syndner), (CD-ROM);
Royal Society of Chemistry Publications, 1996 (see also:
http://www.ch.ic.ac.uk/ectoc/echet96/papers/026/index.htm).
CD-ROM included in J. Chem. Soc., Chem. Commun. 1997, no.
6. Very recently, the attribution of the syn 4,4Ј relative stereo-
chemistry was confirmed on similar compounds via X-Ray
in Table 6.
Synthesis of 44 from 42d: The same procedure described for the
synthesis of 35 was applied. Ϫ 44 was obtained as an oil in 41%
yield (hexanes/Et2O, 1:1). Ϫ C15H17NO5: calcd. C 61.85, H 5.88,
N 4.81; found C 61.84, H 5.93, N 4.78. Ϫ Significant 1H- and 13C-
NMR data are collected in Table 6.
´
`
´
analysis: M. Martın-Vila, N. Hanafi, J. M. Jimenez, A. Alvarez-
Lorena, J. F. Piniella, V. Branchadell, A. Oliva, R. M. Ortun˜o,
J. Org. Chem. 1998, 63, 3581.
2D NOESY Spectra: for all 2D NOESY NMR experiments the
samples were prepared by dissolving 5-6 mg of the required com-
pound in 0.75 ml of CDCl3. The temperature employed for running
the 2D experiment was 50°C and the solution was degassed. Pure
absorption 2D spectra were recorded using NOESY pulse se-
quence[21] 90° Ϫ t1 Ϫ 90° Ϫ τm Ϫ 90° Ϫ t2 and the method of
phase-cycling described by Wüthrich[22] with time-proportional
phase incrementation (TPPI).[22] The following parameter and pro-
cedures are commonly employed: spectral width of 2800 Hz, a 1024
ϫ 1024 data matrix, 256 time increments of 80 transients each;
Fourier transformation were carried out with zero-filling only in f1
using shifted sine-bell apodization function in both dimensions. A
mixing time of 1.5 s and a relaxation delay of 10.0 s were used for
the acetonide 26a . The mixing time and the relaxation delay values
employed were 3.0 and 15.0 s for the bicyclic lactones 35 and 44 .
[7]
R. Annunziata, M. Cinquini, F. Cozzi, P. Giaroni, L. Raimondi,
Tetrahedron Lett. 1991, 32, 1659.
[8] [8a]
R. Annunziata, M. Cinquini, F. Cozzi, L. Raimondi, Tetra-
[8b]
hedron 1987, 43, 4051. Ϫ
R. Annunziata, M. Cinquini, F.
Cozzi, L. Raimondi, J. Org. Chem. 1990, 55, 1901.
[9]
The inside alkoxy cycloadduct features a 5,5Ј-syn (not anti) rela-
tive configuration because of the different priorities of the sub-
stituents at the C-5.
[10]
All cycloadditions were performed in benzene as solvent (Ex-
perimental Section and note 20). Alkene 3 was reacted with
formaldehyde N-benzylnitrone in DMSO as solvent (80°C, 2 h,
50%). The diastereoisomeric ratio (15a/b Ն 85:15) was the same
observed for the reaction performed in benzene, thus ruling out
the presence of an hydrogen bond within the reactants in the
TS. The presence of a H bond between the OH and the π bond
in the ground state conformation of allylic alcohols was sug-
gested: S. D. Kahn, C. F. Pau, A. R. Chamberlin, W. J. Hehre,
J. Am. Chem. Soc. 1987, 109, 650.
Computational Procedures: All calculations were performed
[11] [11a] V. Jäger, I. Müller, R. Schohe, M. Frey, R. Ehrler, B. Häfele,
D. Schröter, Lect. Heterocycl. Chem. 1985, 8, 79. Ϫ [11b] Unpub-
lished results from our laboratories.
using the MacroModel/Batchmin 5.5 package[17] and the MM2*
[16]
force field, augmented with a substructure for the nitrone plus
Eur. J. Org. Chem. 1998, 1823Ϫ1832
1831

R. Annunziata, M. Benaglia, M. Cinquini, F. Cuzzi, L. Raimondi
FULL PAPER
[12]
´
For recent results in this field, see: F. Busque, P. de March,
M. Figueredo, J. Font, M. Monsalvatje, A. Virgili, A. Alvarez-
Larena, J. F. Piniella, J. Org. Chem. 1996, 61, 8578.
For related results, see: G. Galley, P. G. Jones, M. Pätzel, Tetra-
hedron: Asymmetry 1996, 7, 2073.
Lipton, C. Caufield, G. Chang, T. Hendrickson, W. C. Still, J.
Comput. Chem. 1990, 11, 440.
[18]
J. M.Goodman, W. C. Still, J. Comput. Chem. 1991, 12, 1110.
[13]
[14] [14a]
[19]
J. J. W.McDouall, M. A. Robb, U. Niazi, F. Bernardi, H. B.
Schlegel, J. Am. Chem. Soc. 1987, 109, 4642. The RHF/4-31G
TS was very similar to the RHF/3-21G one, on which the fitting
of the electrostatic potential was performed (see ref.[4]).
Y. Yamamoto, Y. Chounan, S. Nishii, T. Ibuka, H. Kita-
[14b]
hara, J. Am. Chem. Soc. 1992, 114, 7652. Ϫ
For a recent
[20]
study on Michael-type addition to (E)- and (Z)-γ-alkoxy-α,β-
unsaturated esters leading to a prevalence of syn adducts, see:Y.
Xiang, H.-J. Gi, D. Niu, R. F. Schinazi, K. Zhao, J. Org. Chem.
1997, 62, 7430.
Cycloaddition reactions were performed in benzene on a re-
duced scale. Special care is needed, since this solvent is known
to be highly toxic and carcinogenic.
[21]
R. R. Ernst, G. Bodenhausen, A. Wokaun, Principles of NMR
[15] [15a] K. N. Houk, M. N. Paddon-Row, N. G. Rondan, Y.-D. Wu,
in One and Two Dimensions. Clarendon Press, Oxford, 1987.
[22] [22a]
F. K. Brown, D. C. Spellmeyer, J. T. Metz, Y. Li, R. J. Loncha-
D. Marion, K. Wüthrich, Biochem. Bioph. Res. Commun.
[15b]
[22b]
rich, Science 1986, 231, 1108. Ϫ
Houk, Chem. Rev. 1993, 93, 2439.
J. E. Esterowicz, K. N.
1983, 113, 967. Ϫ
240.
K. Nagayama, J. Magn. Res. 1986, 66,
[16]
[17]
[23]
[24]
U. Burkert, N. L. Allinger, Molecular Mechanics. ACS Mono-
graph 177; American Chemical Society; Washington D.C.,
1982.
MacroModel substructures can be obtained directly from the
authors (e-mail: laura@iumchz.chimorg.unimi.it).
J. W. Ponder, F. M. Richards, J. Comput. Chem. 1987, 8, 1016.
[98133]
F. Mohamadi, N. G. J. Richards, W. C. Guida, R. Liskamp, M.
1832
Eur. J. Org. Chem. 1998, 1823Ϫ1832