Allylic stereocontrol of the intramolecular diels-alder reaction

Article abstract of DOI:10.1002/chem.200401215

The stereochemical outcome of the intramolecular Diels-Alder reaction of ester-linked 1,3,8-nonatrienes can be controlled by substituants about a stereogenic center attached to Cl. The scope and limitations of this approach have been investigated, with variation in substrate structure about the allylic stereocenter and the dieno phile. The stereochemical outcomes of these reactions are explained by reference to B3LYP/6-31G(d) transition structures. New insights into the conformational preferences of allylic alcohol derivatives are reported, results which allow an explanation of the differing levels of π-diastereofacial selectivity and cis/trans (i.e. endo/exo) selectivity from the reaction.

Full text of DOI:10.1002/chem.200401215

FULL PAPER
Allylic Stereocontrol of the Intramolecular Diels–Alder Reaction
Michael J. Lilly,[a] Natalie A. Miller,[b] Alison J. Edwards,[b] Anthony C. Willis,[b]
Peter Turner,[c] Michael N. Paddon-Row,*[d] and Michael S. Sherburn*[b]
Abstract: The stereochemical outcome
of the intramolecular Diels–Alder reac-
tion of ester-linked 1,3,8-nonatrienes
can be controlled by substituents about
a stereogenic center attached to C1.
The scope and limitations of this ap-
proach have been investigated, with
variation in substrate structure about
the allylic stereocenter and the dieno-
phile. The stereochemical outcomes of
these reactions are explained by refer-
ence to B3LYP/6–31G(d) transition
structures. New insights into the con-
formational preferences of allylic alco-
hol derivatives are reported, results
which allow an explanation of the dif-
fering levels of p-diastereofacial selec-
tivity and cis/trans (i.e. endo/exo) selec-
tivity from the reaction.
Keywords: chiral pool
·
cyclo-
addition density functional
·
calculations · diastereoselectivity ·
synthetic methods
Introduction
the generation of four stereoisomeric products that are in-
terrelated in cis/trans (exo/endo)[3] and p-facial senses. In
recent times, transition structures (TSs) for certain IMDA
reactions have been located by using density functional
theory (DFT). With reference to these TS geometries, cis/
trans and p-facial selectivities of IMDA reactions have been
discussed in detail.[4–10] This approach promises to greatly
simplify the prediction of IMDA stereoselectivities. Herein
we describe a detailed synthetic investigation into a new
stereocontrol method for IMDA reactions and the develop-
ment of theoretical models to rationalize these experimental
findings.[11]
There is continuing interest in the intramolecular Diels–
Alder (IMDA) reaction in both the realms of synthesis[1]
and biosynthesis.[2] An understanding of the stereochemical
preferences of IMDA reactions is pivotal to its successful
synthetic application and biosynthetic rationalization. The
stereochemical complexity of this reaction stems from the
four contiguous stereocenters that are created. The concert-
ed, suprafacial nature of the bond-forming process allows
[a] Dr. M. J. Lilly
Reported methods for obtaining enantiomerically pure
products from IMDA reactions (Scheme 1) involve the
placement of suitable substituents on the linking chain be-
Institute of Fundamental Sciences, Massey University Private Bag
11222, Palmerston North (New Zealand)
[b] N. A. Miller, Dr. A. J. Edwards, Dr. A. C. Willis, Dr. M. S. Sherburn
Research School of Chemistry, Australian National University
Canberra, ACT 0200 (Australia)
Fax : (+61)2-6125-8114
E-mail: sherburn@rsc.anu.edu.au
[c] Dr. P. Turner
School of Chemistry, University of Sydney
Sydney, NSW 2006 (Australia)
[d] Prof. M. N. Paddon-Row
School of Chemistry, University of New South Wales
Sydney, NSW 2052 (Australia)
Fax : (+61)2-9385-6141
E-mail: m.paddonrow@unsw.edu.au
Supporting information for this article (Cartesian coordinates and
energies of B3LYP/6–31G(d) optimized TS geometries, synthetic
details, compound characterization, crystallographic data and selected
1H and 13C NMR spectra) is available on the WWW under
http://www.chemeurj.org/ or from the author.
Scheme 1. The three reported methods for achieving stereocontrol in
IMDA reactions and the diene controller approach under scrutiny.
Chem. Eur. J. 2005, 11, 2525 – 2536
DOI: 10.1002/chem.200401215
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2525

tween the diene and dienophile,[12] by attaching an auxiliary
to the dienophile terminus,[13] and through enantioselective
cycloaddition of achiral trienes with chiral Lewis acids.[14]
We were intrigued by the possibility of controlling the ster-
eochemical outcome of IMDA reactions by constructing
chiral substrates with a stereocontrolling element at the ter-
minus of an acyclic diene. Previous reports of IMDA reac-
tions of substrates with diene auxiliaries employ either con-
formationally restricted chiral semicyclic dienes,[15] or are
complicated by the presence of additional tether substitu-
ents and dienophile auxiliaries.[16]
stereomeric with 1, was prepared to investigate the depen-
dence of stereoselectivity upon the relative configurations at
the allylic and homoallylic stereocenters. Chiral trienes 3
and 4 would uncouple the effects of the allylic and homoal-
lylic stereocenters, allowing a more general assessment of
the applicability of this approach. To shed light upon stereo-
selectivities, these reactions are examined computationally
by using the hybrid B3LYP functional together with the 6–
31G(d) basis set. It is generally accepted that B3LYP/6–
31G(d) models give reliable geometries of TSs and activa-
tion energies for pericyclic reactions.[4–10,18]
Herein, we explore the scope and limitations of the allylic
stereocontrol method for IMDA reactions. IMDA reactions
of ester-linked 1,3,8-nonatriene[17] 1 are reported, along with
cycloaddition reactions of three additional groups of sub-
strates; namely, triene esters 2–4 (Scheme 2). Series 2, dia-
Results
Synthesis of the chiral dienols
Chiral dienols 7, 10, 13, and 15 were prepared from the inex-
pensive and readily available chiral precursors l-ascorbic
acid, d-isoascorbic acid, l-malic acid, and l-lactic acid re-
spectively, as shown in Scheme 3. The syntheses of these di-
enols follow the same general pathway, involving the chain
extension of a chiral ester into the corresponding conjugated
diene.
Thus, ascorbic and isoascorbic acids were smoothly con-
verted into monoacetonides[19] before oxidative cleavage
with potassium carbonate and hydrogen peroxide in water.
The resulting potassium carboxylates were treated with
ethyl iodide to give a-hydroxy esters 5 and 8.[20] Regioselec-
tive reduction of diethyl malate with borane/dimethyl sul-
Scheme 2. Chiral trienes under investigation. The effect of the groups P,
E, and Z upon IMDA stereoselectivity is examined.
Scheme 3. Synthesis of chiral dienols 7, 10, 13, and 15: a) TBSCl, imidazole, DMF, RT, ascorbate 68%; isoascorbate 91%. b) DIBALH, CH2Cl2, ꢀ1008C
to ꢀ788C, ascorbate 86%; isoascorbate 58%. c) Ph3P=CHCH=CHCO2Et (16), CH2Cl2, reflux, then PhSH, AIBN, PhH, reflux; ascorbate 78%; malate
27%. d) DIBALH, CH2Cl2, ꢀ788C, ascorbate 74%; malate 93%. e) LiN(Si(CH3)3)2, (CH3O)2P(=O)CH2CH=CHCO2Me (17), THF, ꢀ788C, isoascorbate
70%; lactate 42%. f) DIBALH, Et2O, 0 8C, isoascorbate 80%; lactate 88%. g) LiAlH4, THF, 08C, 92%. h) Dess–Martin periodinane, CH2Cl2, 78%.
TBSCl=tert-butyldimethylsilyl chloride, DIBALH= diisopropylaluminum hydride, AIBN=2,2’-azobisisobutyronitrile.
2526
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 2525 – 2536

Allylic Stereocontrol of the Intramolecular Diels–Alder Reaction
FULL PAPER
fide complex, followed by treatment of the resulting diol
with 2,2-dimethoxypropane[21] produced acetonide 11 in
70% yield after distillation. Silylation of the a-hydroxy
esters followed by reduction with diisobutylaluminum hy-
dride (DIBALH) at low temperature produced aldehydes 6,
9, and 14[22] cleanly. In the ascorbate series, the diene was in-
stalled through treatment of aldehyde 6 with stabilized ylide
16[23] in refluxing dichloromethane. The dienoate ester was
thereby produced as a mixture of (Z)- and (E)-stereoisom-
ers about the newly formed alkene (Z:E=79:21), which was
equilibrated to the (E,E)-diene under radical conditions.[24]
In the isoascorbate and lactate series, chain extensions were
carried out by Horner–Wadsworth–Emmons reaction with
phosphonate 17.[25] The diene esters were reduced cleanly to
the dienols with DIBALH. In the malate series, a two-step
reduction–oxidation[26] sequence was necessary for conver-
sion of ester 11 into aldehyde 12. A Wittig reaction between
sensitive aldehyde 12 and stabilized ylide 16 produced the
diene ester as a 50:50 mixture of (Z)- and (E)-stereoisomers
in a disappointing 35% yield after chromatography.[27] Fol-
lowing isomerization to the (E,E)-diene, DIBALH reduction
gave dienol 13 in 93% yield as a colorless oil. Yield optimi-
zation of this final sequence was postponed, pending the re-
sults of IMDA reactions in this series.
Reaction of dienol 7 with maleic anhydride furnished a
near quantitative yield of carboxylic acid 18, which was
treated with diazomethane to give diester 1c. Desilylation
of diester 1c gave an unacceptably low yield (26%) of sec-
ondary alcohol 1a. A much better result (85% yield) was,
however, obtained by deprotection of acid 18. The resulting
hydroxy acid 19 was converted to methyl ester 1a, which
was used to prepare trimethylsilyl derivative 1b and triiso-
propylsilyl derivative 1d by exposure to the appropriate tri-
alkylsilyl triflate. Reaction of alcohol 1a with para-nitroben-
zoyl chloride gave ester 1e. Fumarate ester 1 f and propy-
noate ester 1g were prepared by Steglich esterification of
dienol 7 with the requisite carboxylic acid.
Precursors 1a–e underwent IMDA reactions at high dilu-
tion in boiling toluene to produce mixtures of the four possi-
ble cycloadducts (Table 1). In all five cases, two of the four
adducts were formed in significantly larger amounts than
the other two.
Table 1. IMDA reactions of ascorbic acid derived trienes 1a–f.
Triene synthesis and IMDA reactions
The ascorbate series: Ascorbic acid derived dienol 7 was
converted into precursors for IMDA reaction differing in
both the stereocenter oxygen substituent P and the nature
of the dienophile. Seven triene derivatives were prepared
(Scheme 4, 1a–g).
Triene
substrate
1
P/E/Z
Solvent/
Reaction
time[a] [h]
Isolated
Adduct
ratio[c,d,e]
20:21:22:23
yield[b] [%]
a
b
c
d
e
f
H/H/CO2Me
TMS/H/CO2Me
TBS/H/CO2Me
TIPS/H/CO2Me
PNB/H/CO2Me
TBS/CO2Me/H
PhMe/5
PhMe/12
PhMe/15
PhMe/18
PhMe/12
PhCl/53
86
67
80
68
95
62
56:32:8:4
80:14:4:2
86:9:4:1
92:7:1:0
49:39:6:6
12:3:82:3
[a] Time required for >95% conversion, as judged by 1H NMR spectro-
scopy. [b] Combined isolated yield for the four adducts after chromatog-
raphy. [c] Determined from 1H NMR spectra and HPLC analyses of
crude reaction mixtures. Differences +/ꢀ3%. [d] Kinetic product ratios
are reported: control experiments confirmed that all cycloadducts were
stable to the reaction conditions. [e] The two minor adducts were not
fully characterized.
The stereochemistry of each the two major cycloadducts
from the three silyl ether substrates 1b–d was elucidated by
using 2D NMR experiments. The large coupling constant
between the axial hydrogen atoms at the ring junction (3J=
13.6–13.8 Hz) indicates the less thermodynamically stable
trans-ring fusion. The trans adducts 20 and 21 exhibited a
clear conformational preference in solution, which allowed
the new and existing stereochemistry to be correlated
through NOE measurements.[28] The identities of the alcohol
cycloadducts 20a and 21a were secured by conversion into
Scheme 4. Synthesis of triene derivatives 1a–g: a) maleic anhydride,
Et3N, DMAP, CH2Cl2, RT, 99%. b) CH2N2, Et2O, 18!1c, 74%; 19!1a,
80%. c) nBu4NF, THF, RT, 85%. d) TMSOTf, Et3N, DMAP, CH2Cl2, RT,
51%. e) TIPSOTf, 2,6-lutidine, CH2Cl2, RT, 91%. f) p-NO2C6H4COCl,
pyridine, DMAP, 258C, 82%; g) (E)-HO2CCH=CHCO2Me, DCC,
DMAP, Et2O, RT, 96 %. h) HO2CCꢁCH, DCC, DMAP, Et2O, RT, 65 %.
DMAP=4-dimethylaminopyridine, TIPS=triisopropylsilyl, TMS=trime-
thylsilyl, DCC=N,N’-dicyclohexylcarbodiimide.
Chem. Eur. J. 2005, 11, 2525 – 2536
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2527

M. N. Paddon-Row, M. S. Sherburn et al.
the corresponding TMS derivatives 20b and 21b. The major
and minor isomers obtained from this silylation reaction
were identical in every respect to those produced in the
IMDA reaction of TMS ether precursor 1b. Finally, the
identities of cycloadducts 20e and 21e were confirmed by
esterification of the alcohols 20a and 21a with para-nitro-
benzoyl chloride in the usual manner. Insufficient quantities
of the two minor cycloadducts were obtained from these ex-
periments for full characterization.
of this starting material? 2) Is the allylic stereocenter neces-
sary for stereoselectivity? To answer the first question, iso-
ascorbic acid derived dienol 10 was converted into maleate
and fumarate derivatives 2a and 2b in the same manner as
described previously (Scheme 6). To address the second
issue, malic acid derived dienol 13 was transformed into cor-
responding trienes 3x and 3y.
The stereoselectivities observed for maleate derivatives
1a–e (Table 1) are intriguing in that the level of p-diastereo-
facial selectivity varies considerably with group P. The alco-
hol substrate 1a exhibits only modest unlike[29] p-diastereo-
facial preference between the two trans-cycloadducts but
this improves considerably after the installation of common
silyl protecting groups.[30] A progressive improvement in
unlike selectivity is observed as the size of the silyl group is
increased (cf. 1b!1c!1d). The triisopropylsilyl (TIPS) de-
rivative 1d, the most sterically demanding protecting group
examined, instigates a remarkably high level of stereocon-
trol. Conversely—and unexpectedly—the para-nitroben-
zoate derivative 1e exhibits the lowest p-diastereofacial se-
lectivity of the five substrates.
As expected, the fumarate precursor 1 f was significantly
less reactive than the corresponding maleate 1c.[31] The
same isolated yields and stereochemical outcomes[28] were
obtained when 1 f was heated in toluene or chlorobenzene,
although the former required prolonged reaction times for
complete conversion. Fumarate 1 f underwent cycloaddition
with virtually complete unlike-selectivity and high cis-selec-
tivity, the latter feature being in stark contrast to previous
findings with pentadienyl fumarates, which generally furnish
mildly trans-selective IMDA reactions.[4,7,10,31,32]
Scheme 6. Conversion of 10 and 13 into 2a,b and 3x,y, respectively: a)
maleic anhydride, Et3N, DMAP, CH2Cl2, RT, then CH2N2, Et2O, 0 8C,
10!2a, 74%; 13!3x, 56%. b) (E)-HO2CCH=CHCO2Me, DCC,
DMAP, Et2O, RT, 10!2b, 78%; 1!3y, 74%.
The isoascorbic acid derived precursors underwent IMDA
reactions with approximately the same ease as the ascorbic
acid series, whereas the malic acid derived precursors cy-
clized somewhat more rapidly (Table 2). Product stereo-
chemistries in the isoascorbate series were determined in
the same manner as the ascorbate compounds. The stereo-
chemical identity of the major product from fumarate pre-
cursor 2b was established by selective acetal deprotection/
lactonization to 30 (Scheme 7). In the malic acid products,
the ring junction stereochemistry was clearly evident from
NMR coupling constants, as discussed previously.
Table 2. IMDA reactions of isoascorbic acid and malic acid derived pre-
The last precursor derived from ascorbic acid, the alkynic
dienophile 1g, underwent a smooth IMDA reaction at
1108C (Scheme 5).[31] Lacking questions of cis/trans-stereose-
lectivity, this substrate perhaps best exemplifies the unlike
p-diastereofacial preference found in this series.[33]
cursors 2a,b and 3x,y.
Isoascorbate and malate series: The results obtained thus
far raised two important questions. 1) Are the outcomes of
these reactions dependent upon the relative stereochemistry
Triene
substrate
X/E/Z
Solvent/
Reaction
time[a] [h]
Isolated
Adduct
ratio[c,d,e]
26:27:28:29
yield[b] [%]
2a
2b
3x
3y
OTBS/H/CO2Me
OTBS/CO2Me/H
H/H/CO2Me
PhMe/9.5
PhCl/72
PhMe/3.5
PhCl/43
86
89
89
89
1
78:16:4:2
9:6:66:19
41:39:11:9
32:33:18:17
H/CO2Me/H
[a] Time required for >95% conversion, as judged by H NMR spectros-
copy. [b] Combined isolated yield for the four adducts after chromatog-
raphy. [c] Determined from 1H NMR spectra and HPLC analyses of
crude reaction mixtures. [d] Kinetic product ratios are reported: control
experiments confirmed that all cycloadducts were stable to the reaction
conditions. [e] The stereochemical identities of the two minor adducts
were not elucidated.
Scheme 5. IMDA reaction of ascorbic acid derived alkynoate 1g. TBS=
tert-butyldimethylsilyl.
2528
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 2525 – 2536

Allylic Stereocontrol of the Intramolecular Diels–Alder Reaction
FULL PAPER
Scheme 7. Acetonide deprotection/lactonization of 28b.
The major diastereomer obtained from intramolecular cy-
cloadditions of isoascorbic acid derived substrates 2a and
2b (Table 2) correlates with those obtained in the ascorbic
acid series 1c and 2 f (Table 1). Thus, a (Z)-dienophile pre-
cursor (i.e. 2a or 1c) gives the trans, unlike adduct, whereas
the (E)-dienophile precusor (i.e. 2b or 1 f) gives the cis,
unlike adduct. The adduct ratios are similar but not identi-
cal, indicating a contribution to the overall stereochemical
outcome by the configuration at the homoallylic site. No p-
diastereofacial selectivity was witnessed in cyclization of
either maleate 3x or fumarate 3y, however, indicating that a
lone homoallylic stereocenter is insufficient to invoke ster-
eoselection. It is noteworthy that the (E)-dienophile precur-
sor 3x undergoes a trans-selective IMDA reaction, a result
which is consistent with previous findings[4,7,10,31,32] but in
contrast to those of the other three fumarate precursors in
the present study.
Scheme 8. Conversion of dienol 15 into maleate esters 4a–d: a) maleic
anhydride, Et3N, DMAP, CH2Cl2, RT, 75%. b) nBu4NF, THF, RT, 44%.
c) CH2N2, Et2O, 0 8C, 31!4c; 87%, 31–4a, 44%. d) TMSOTf, Et3N,
DMAP, CH2Cl2, RT, 88%. e) TIPSOTf, Et3N, CH2Cl2, RT, 78%. f) (E)-
HO2CCH=CHCO2Me, DCC, DMAP, Et2O, RT, 89 %.
Table 3. IMDA reactions of lactic acid derived trienes 4a–e.
Lactate series: In the research described so far, it has been
demonstrated that stereogenic elements at the diene termi-
nus can induce high levels of stereoselectivity upon IMDA
reactions. Ascorbate- and isoascorbate-derived systems 1
and 2 are complicated by the presence of two stereocenters
and an acetal functional group. Would precursors lacking
these structural features also give rise to stereoselective
IMDA reactions? To answer this question and, furthermore,
to minimize the assumptions made in the computational
modeling of this reaction, we elected to carry out a synthetic
study on dienol 15. Thus, lactic acid derived dienol 15 was
converted into the four maleate esters 4a–d, and the fuma-
rate ester 4e using similar procedures to those described
previously (Scheme 8).
The five triene precursors underwent IMDA reactions in
refluxing toluene (Table 3) over slightly shorter reaction
times than those of the corresponding ascorbic and isoascor-
bic acid derived precursors (Table 1 and Table 2). Further-
more, mixtures of the four possible cycloadducts were pro-
duced in all cases.
Triene
substrate
4
P/E/Z
Solvent/
Reaction
time[a] [h]
Isolated
Adduct
ratio[c,d]
33:34:35:36
41:36:5:18[e]
56:30:6:8
61:27:5:7
63:26:5:6
8:28:40:24
yield[b] [%]
a
b
c
d
e
H/H/CO2Me
TMS/H/CO2Me
TBS/H/CO2Me
TIPS/H/CO2Me
TBS/CO2Me/H
PhMe/4.5
PhMe/6.5
PhMe/7.25
PhMe/8.25
PhCl/43
88
83
96
85
73
[a] Time required for >95% conversion, as judged by 1H NMR spectro-
scopy. [b] Combined isolated yield for the four adducts after chromatog-
raphy. [c] Determined from 1H NMR spectra and HPLC analyses of
crude reaction mixtures. Differences +/ꢀ3%. [d] Kinetic product ratios
are reported: control experiments confirmed that all cycloadducts were
stable to the reaction conditions. [e] Tricyclic bis-lactones 37 and 38 were
isolated instead of cis-adducts 35a and 36a in this series. See main text
for an explanation.
confirm the stereochemical identities of all 20 cycloadducts
in the lactate series.[28]
The main features of the lactate series can be summarized
as follows. The (Z)-dienophile precursors 4a–d, once again,
undergo strongly trans-selective IMDA reactions. The alco-
hol substrate 4a exhibits a very small p-diastereofacial selec-
tivity during cycloaddition. The corresponding silyl ethers
undergo more stereoslective IMDA reactions, with higher
levels of p-diastereofacial selectivity being obtained with
larger silyl groups. Nevertheless, the highest level of p-facial
selectivity delivered by a lactate-derived precursor (4d, un-
like:like=68:32) is considerably lower than that witnessed
in the ascorbate series (1d, unlike:like=93:7). The (E)-dien-
ophile precursor 4e undergoes a reaction with considerably
The four cycloadducts obtained from the IMDA reaction
of the maleate TBS ether 4c were separated by HPLC. The
ring junction geometry of cycloadducts 33c and 34c was evi-
dent from the large trans-diaxial coupling constants (3J=
13.5 Hz) between the ring junction hydrogen atoms. On the
basis of 2D NMR experiments it was possible only to eluci-
date the relative stereochemistry at the four new stereocen-
ꢀ
ters; the lack of conformational preference about the C1’
C5 bond precluded the correlation of new with existing ster-
eochemistry in this set of cycloadducts. Thus, five single-
crystal X-ray analyses were carried out on derivatives to
Chem. Eur. J. 2005, 11, 2525 – 2536
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2529

M. N. Paddon-Row, M. S. Sherburn et al.
lower stereoselectivity than the corresponding ascorbate and
isoascorbate trienes, 1 f and 2b, respectively. Once again,
however, the cis,unlike diastereomer is the major adduct.
Whilst the stereoselectivity of IMDA reactions of precur-
sors 1–4 varies considerably, the following generalizations
can be made:
tional preferences of the C* substituents for the in, an, and
ou sites in the TS. What are these preferences? This ques-
tion was answered from model computational studies.
Computational Results
1) Maleate esters of dienols 7, 10, 13, and 15 undergo trans-
selective IMDA reactions, whereas the corresponding fu-
marate esters are generally cis-selective. The exception
is fumarate 3y, which undergoes a trans-selective IMDA
reaction. These selectivities are more pronounced the
larger the substituents attached to the allylic stereocen-
ter.
B3LYP/6–31G(d) density functional theory calculations
were conducted on maleate and fumarate IMDA precursors
4b, 4 f and 4g (Scheme 9).[36] A preliminary account of this
2) Chiral allylic alcohols (1a and 4a) give poor to moderate
levels of p-diastereofacial selectivity in IMDA reactions.
Silyl ether derivatives of these allylic alcohol precursors
give moderate to high levels of p-facial stereoselectivity,
with an increase in the size of the silyl protecting group
leading to an increase in stereoselection. An allylic ben-
zoyloxy derivative gave essentially no p-facial discrimi-
nation in these reactions.
Scheme 9. Structures of the three precursors under scrutiny by DFT and
the 18 possible diastereomeric TSs for the trans, like IMDA cycloaddition
mode of 4b.
3) An increase in the size of the alkyl group leads to an in-
crease in p-facial stereoselectivity (compare results of
substrates 1 and 2 with those of 4).
4) The two diastereomeric dienols 7 and 10 gave similar
but non-identical stereoisomer ratios, a result which
demonstrates that the allylic stereocenter controls the
outcome of these reactions to a large extent.
5) An allylic stereocenter appears to be necessary for p-
facial stereoselectivity, since substrates carrying an allylic
methylene and homoallylic stereocenter (3x,y) gave no
stereocontrol.
work has appeared, describing a model which was based
only on the computed trans IMDA TSs of the Z ester,
OSiH3 structure 4 f.[5] The computational modeling of these
reactions is complicated by the large number of possible re-
active orientations of each precursor. Thus, in addition to
ꢀ
the three conformations about the C* C1 bond, three con-
ꢀ
formations about the allylic C* O bond and s-cis/s-trans ori-
ꢀ
entations of the C9 CO2CH3 bond are feasible (Scheme 9).
These experimental results clearly demonstrate that the
presence of an allylic stereocenter can indeed give rise to
high levels of stereoselectivity in IMDA reactions. Inspec-
tion of molecular models and literature searches gave no
clear reasons for the observed experimental stereoselectivi-
ties. The C* substituents would be expected to adopt an ap-
proximately staggered arrangement with respect to the de-
In principle, therefore, a total of 72 diastereomeric TSs can
be envisioned for each IMDA precursor, with 18 diastereo-
meric TSs leading to each of the four cycloadducts, 33–36
(see Table 3 for these structures). It was impractical to cal-
culate all 72 TSs at this level of theory for each IMDA pre-
cusor and, consequently, relaxed potential energy scans
were carried out on model TSs to locate the lowest energy
[34,35]
ꢀ
ꢀ
veloping C1 C9 bond,
with allylic alkyl, silyloxy, and
conformation about the allylic C* O bond. Although we
hydrogen substituents distributed among the inside (in), anti
(an), and outside (ou) positions in TSs, as depicted by 41
(Figure 1). The facial selectivity is determined by the posi-
are confident that we have located the lowest energy confor-
mation in each case we cannot guarantee it to be the case.[37]
IMDA TSs for 4 f in which the ester adopts the s-cis confor-
mation were found to be slightly (2.5–6.0 kJmolꢀ1) lower in
energy than the corresponding s-trans TSs. In all other re-
spects, the two sets of TSs, one with s-cis and the other with
s-trans ester conformations showed very similar trends, par-
ticularly with respect to facial selectivity; hence discussion
here will be limited to the s-cis TSs.
These considerations led to a more manageable three TSs
leading to each of the four diastereomeric adducts. These
three TSs, corresponding to three low-energy conformations
ꢀ
about the C* C1 bond (see Figure 1), were located for each
of the four cycloaddition modes (cis/trans; like/unlike combi-
Figure 1. A representation of one of the trans-TSs. The Newman projec-
ꢀ
tion formula on the right depicts a view along the C* C1 bond of the
structure on the left.
nations) and their relative energies (including zero-point en-
2530
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 2525 – 2536

Allylic Stereocontrol of the Intramolecular Diels–Alder Reaction
FULL PAPER
ergies) were used to construct a predicted Boltzmann distri-
bution of stereoisomeric cycloadducts. The structures and
relative energies of 12 diastereomeric TSs for the IMDA re-
action of C*-OSiH3 Z-dienophile precursor 4 f are depicted
in Figure 2.
which are expected to be small, considering the low polarity
of the toluene or chlorobenzene solvents used in the experi-
ments—the close correlation between theory and experi-
ment is remarkable, providing compelling evidence in sup-
port of the reliability of our theoretical model.
For the (E)-dienophile series, the predicted IMDA adduct
ratio from the reaction of OSiH3 substrate 4g is
33:34:35:36=25:20:50:5 at 405 K in the gas phase. Experi-
mentally, in refluxing chlorobenzene solution (405 K), the
ratio of products for the -OTBS substrate 4e is 8:28:40:24.
Thus, theory correctly predicts: 1) a less stereoselective
IMDA reaction for a fumarate ester than the corresponding
maleate ester; 2) the identity of the major product, the cis,
unlike diastereoisomer, 3) the cis/trans-selectivity of the re-
action, and 4) in a qualitative manner, the p-diastereofacial
selectivity within the cis-adduct manifold. There is a discrep-
ancy between theory and experiment, however, in the p-dia-
stereofacial selectivity of the trans-adducts (exptl: unlike:-
like=22:78; theory unlike:like=57:43) Nevertheless, once
again, overall there is a very good correlation between ex-
periment and theory. Extensive CPU times were necessary
to carry out these calculations and it was not feasible to
carry out a similar computational analysis on TBS fumarate
precursor 4e.
The very good correlation between theory and experiment
once again demonstrates the validity of the computational
model.
Discussion
Following a brief discussion of general features of the TSs
located, the computational results will be interpreted to
answer the following three questions: 1) What is the origin
of the unlike p-diastereofacial selectivities of these reac-
tions? 2) Why does the magnitude of p-diastereofacial selec-
tivity change with different RO- substituents? 3) Why do fu-
marates 1 f, 2b and 4e undergo IMDA reactions with anom-
alously high cis-selectivities?
Figure 2. TSs and calculated Boltzmann populations (italic for individual
TSs, the underlined number is the total for each diastereomeric adduct)
at 383 K for the IMDA reaction of 4 f. The twelve TSs are viewed down
ꢀ
the C1–C* axis and the developing C1 C9 bond is depicted as a vertical
gray line. C*-H and C*-CH3 are labeled throughout and some atoms are
omitted for clarity. Structures a)–c) lead to the trans, unlike adduct 33 f;
d)–f) lead to trans, like adduct 34 f; g)–i) lead to cis, unlike adduct 35 f;
and j)–l) lead to the cis, like adduct 36 f. Structures on the left have the
-OSiMe3 group inside, those in the center column have the -H inside, and
those on the right have the -CH3 inside. To facilitate comparisons be-
tween TSs, dienophile approach from below is depicted throughout,
hence the configuration at the allylic stereocenter is inverted in the
unlike TSs.
TS geometries and relative energies: For all three trienes ex-
amined computationally, the lowest energy IMDA TSs have
unlike stereochemistry and exhibit a conformation in which
the C*-silyloxy group adopts the inside position, the C*-
methyl group is in the anti position, and the C9-CO2Me
group adopts the exo orientation. The preferences of the
methyl and silyloxy groups for the anti and the inside posi-
tions, respectively, in the lowest energy TSs are a conse-
quence of their respective innate tendencies to adopt these
positions in both the transition state and the ground state, ir-
respective of the presence of the other substituent (vide
infra). This conclusion is borne out by the calculations on
model IMDA precursors 42 and 43 (Scheme 10), each of
which contains only one of these substituents. As shown in
Table 4, the most favorable trans IMDA TS for 42 is 42a,
with a methyl group in the anti position, and that for 43 is
43a, with a silyloxy group in the inside position. The same
For both C*-OSiH3 triene 4 f and C*-OSiMe3 triene 4b,
theory correctly predicts the major–minor product se-
quence: the predicted Boltzmann populations of IMDA ad-
ducts from the reaction of 4b at 383 K in the gas phase
(33b:34b:35b:36b=81.9:15.3:1.2:1.6[28]) compares favorably
with the experimentally determined ratio in refluxing tolu-
ene (Table 3, entry 2; 33b:34b:35b:36b=56:30:6:8). Con-
sidering the complexity and multiplicity of TSs examined
and the omission of solvent effects in the calculations—
Chem. Eur. J. 2005, 11, 2525 – 2536
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2531

M. N. Paddon-Row, M. S. Sherburn et al.
Scheme 10. Model IMDA precursors employed to investigate the innate
conformational preferences of allylic silyloxy and allylic methyl groups.
Table 4. B3LYP/6–31G(d) relative energies [kJmolꢀ1].[a]
Figure 3. Most stable TSs for each of the three series of precursors. a)
trans, unlike-4 fTS-(a) (Figure 5a); b) trans, unlike-4bTS-(a) (see Figure
S1a[28] in the Supporting Information); c) cis, unlike-4gTS-(g) (see Fig-
ure S2g[28] in the Supporting Information). Top: front views (forming
ꢀ
bond lengths in ꢁ). Bottom: Newman projections down the O C* bond.
Most H atoms and C3–C8 have been omitted for clarity.
Entry
Structure
in
an
ou
Erel
0.0 (0.0)
1
2
3
4
5
6
7
8
42a[b]
42b[b]
42c[b]
43a[b]
43b[b]
43c[b]
44a
44b
44c
45a
45b
H
H
CH3
OSiH3
H
H
H
H
CH3
H
H
H
OSiH3
H
CH3
H
H
H
OSiH3
H
H
CH3
H
H
H
OSiH3
H
CH3
H
H
H
formations in these TSs are essentially the same as those for
isolated allylic ethers, with one substituent eclipsing the
double bond.[38] This is hardly a surprising finding, given the
highly extended length of the developing peripheral
bond[4,7,10] (the forming peripheral bond length reaches a
peak at 2.943 ꢁ for SiMe3-substituted TS 4bTS-(d), see
Figure S1d[28] in the Supporting Information) and the conse-
quent near planarity of the C1 carbon of the diene in these
TSs (the sum of the bond angles about this centre is general-
ly greater than 3588).
5.4 (5.5)
2.3 (2.0)
0.0 (0.0)
5.9 (6.8)
12.6 (12.7)
0.0 (0.0)
4.5 (3.9)
4.3 (4.1)
0.0 (0.0)
2.2 (3.15)
2.1 (1.75)
9
CH3
OSiH3
H
10
11
12
45c
H
OSiH3
Model B3LYP/6–31G(d) calculations on 1-ethyl-s-cis-bu-
tadiene (46), in which the diene moiety is constrained to co-
planarity, show that the staggered conformation of the C-
Me group is 2.7 kJmolꢀ1 more stable than the conformation
with C-Me eclipsing the C=C double bond (Table 4; entry 7
versus entry 8). This energy difference is similar in magni-
tude to the energetic preference of the anti conformation
over the inside conformation in the TSs for both 42 and 44,
thereby revealing the reactant-like geometry around C1 in
these IMDA TSs.
[a] Relative energies corrected for zero point energy in parentheses. All
TSs have trans-stereochemistry. Similar results were found for the cis-
TSs. [b] s-trans ester conformation.
trend is found for 44 and 45 in which the C9-ester substitu-
ent is absent, thereby demonstrating that the presence of
this group is not required for inducing the conformational
preferences of the methyl and silyloxy groups in these sys-
tems, although its presence does strengthen the preference
of the silyloxy group for the inside position.
ꢀ
For all TSs located, the forming internal (C4 C8) bond is
Our model for the favored TS for IMDA reactions, in
which the silyloxy group adopts the inside position and the
alkyl group the anti position, is reminiscent of that proposed
for both intermolecular 1,3-dipolar cycloadditions of nitrile
oxides to chiral allylic ethers,[35,39] and Craigꢂs silylacetal
tethered IMDA reactions.[40]
ꢀ
considerably shorter than the developing peripheral (C1
C9) bond, giving rise to significant bond-forming asynchro-
nicity (i.e. a difference between the two forming bond
lengths). This is exemplified in Figure 3 with the most stable
TSs for each of the three series of precursors, 4 f, 4b and 4g.
The two 9-Z-CO2Me TSs, trans, unlike-4 fTS-(a) and trans,
unlike-4bTS-(a) have large bond-forming asynchronicities
of 0.75 and 0.77 ꢁ, respectively, whereas the 9-E-CO2Me TS
cis, unlike-4gTS-(g) exhibits slightly lower levels of bond-
forming asynchronicity of 0.56 ꢁ. Similar bond-forming
asynchronicities and trends are found in the other TSs.
These geometric features are consistent with findings on re-
lated systems.[4,7,10] Inspection of the profiles of the 36
IMDA TSs located for trienes 4b, 4 f and 4g (Figure 2, and
Figures S1 and S2[28] in the Supporting Information) reveals
that the allylic substituents show little staggering about the
forming peripheral bond (r2) that is normally characteristic
of additions to allylic systems.[34,35] Indeed, the C1–C* con-
Origin of unlike p-diastereofacial selectivities: In the pre-
liminary study we suggested that the inside preference for
the -OSiR3 group might arise from an attractive electrostatic
interaction between the oxygen atom and the hydrogen
atom at C2.[5] Other workers have suggested that this con-
formational preference is due to the small lone pair size of
the silyl ether oxygen atom.[38] The considerably lower p-
facial selectivities observed experimentally in IMDA reac-
tions of alcohol substrates 1a and 4a and the p-nitroben-
zoate ester 1e versus the corresponding silyl ethers (Table 1
and Table 3) are consistent with our electrostatic argument,
but not with the oxygen lone pair size proposal. To shed
2532
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 2525 – 2536

Allylic Stereocontrol of the Intramolecular Diels–Alder Reaction
FULL PAPER
Table 5. Conformational preferences of allylic alcohol derivatives.[a,b]
light on this issue, the excess charge on the allylic oxygen
atom was calculated for model systems using natural popula-
tion analysis (NPA) using the natural bond orbital scheme
of Weinhold et al.[41] The results are as follows (in units of
electron charge): MeOSiMe3 ꢀ0.890; MeOSiH3 ꢀ0.845;
MeOH ꢀ0.740; MeOMe ꢀ0.558; methyl-p-nitrobenzoate
(alkoxy oxygen) ꢀ0.540. Thus, the charge on the oxygen
atom is significantly higher for silyl alkyl ethers than for al-
cohols, dialkyl ethers, or p-nitrobenzoate esters, which sup-
ports our electrostatic argument. Further evidence for the
effect being electrostatic in nature was obtained by examin-
ing the preferred conformations of allylic alcohol deriva-
tives. The energy differences between the conformation in
R
DE [kJmolꢀ1
]
-OpNB
-OMe
-OSiH3
-OSiMe3
-Oꢀ
3.0
0.36
0.57
ꢀ2.0
ꢀ5.1
[a] All geometries were optimized fully. [b] A positive value for DE
means that the out-of-plane conformation is preferred.
ꢀ
which the alkene p bond and the C O bond are roughly in-
plane and the out-of-plane conformation, for allyl alcohol
and pertinent derivatives are listed in Table 5. As the elec-
tron density about the oxygen atom increases, so does the
preference for the in-plane conformation. Indeed, the in-
plane conformation is preferred only with the -OSiMe3 and
-Oꢀ groups, all others preferring the out-of-plane conforma-
tion.[42] This trend is consistent with the aforementioned in-
creased electrostatic stabilization of the in-plane conformer.
Increased stabilization of the -OR inside conformation in
IMDA reactions should lead to even higher levels of facial
stereocontrol than those witnessed in Tables 1—3. This
might be achieved by using the alkoxide anion group -Oꢀ
(see Table 5). Attempts to examine the IMDA reactions of
metal alkoxides derived from alcohol 4a were thwarted by
the instability of the ester groups in the presence of the alk-
oxide. Investigations with more robust tethers are continu-
ing.
Table 6. Relative energies, Erel [kJmolꢀ1], of trans, unlike and trans, like
TSs for the IMDA reactions of 4b and 4 f.
Erel
Erel
trans-4bTS
trans-4 fTS
unlike, OSi in
unlike, H in
unlike, CH3 in
like, OSi in
like, H in
0.0
6.4
18.0
6.3
10.2
11.0
0.0
4.2
14.6
4.6
5.4
11.3
like, CH3 in
Magnitude of p-diastereofacial selectivities: Theory predicts
an IMDA reaction for the maleate SiMe3 substrate which is
considerably more p-diastereofacially selective than that of
the SiH3 substrate (unlike:like=5.9:1 (SiMe3); 2.8:1 (SiH3)).
This result is in qualitative agreement with the experimental
findings, which demonstrate that the larger the silyl protect-
ing group, the higher the p-diastereofacial selectivity. The
origin of this increase in selectivity with increasing size of
the SiR3 group is apparent from inspection of the relative
energies, Erel, of the like and unlike trans-TSs for the IMDA
reactions of 4 f and 4b, presented in Table 6. These data
reveal that the major difference in the Erel values for the
two systems is between the like TSs with H occupying the
inside position: trans, like-4 fTS-(e) and trans, like-4bTS-
(e) (see Figure 4). The Erel value increases significantly, from
5.4 kJmolꢀ1, for trans, like-4 fTS-(e), to 10.2 kJmolꢀ1 for
trans, like-4bTS-(e). This increase has the consequence of
bolstering the importance of the trans, unlike-4bTS-(a).
The cause of the destabilization of trans, like-4bTS-(e)
can be understood by examining the differences between
the SiMe3 and SiH3 TSs (Figure 4). IMDA TSs (e) for the
SiH3 and SiMe3 substrates are remarkably similar, with one
notable exception: the Si-O-C*-C1 dihedral angle is signifi-
cantly larger in the SiMe3 TS (ꢀ1308), compared to that in
the SiH3 TS (ꢀ1098). This difference is probably due to ad-
Figure 4. Representations of TSs (e) for the SiH3 (left hand side) and
ꢀ
SiMe3-containing (Z)-ester substrate 4. Top: views down O C* bond;
bottom: front views. Arrows show the direction of rotation about O-Si-
C*-C1.
verse steric interactions between the TMS group and the
methyl group of the CO2Me group in trans, like-4bTS-(e).
ꢀ
This increased twist about the O C* bond brings a TMS
ꢀ
methyl group into fairly close proximity with C* H (2.55 ꢁ;
see Figure 4); apparently little relief of strain is achieved by
the aforementioned twisting mode, consistent with the
higher Erel value for trans, like-4bTS-(e), compared to
trans, like-4 fTS-(e). This destabilization of TS(e) would be
enhanced by the presence of larger silyl substituents (i.e.
Chem. Eur. J. 2005, 11, 2525 – 2536
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2533

M. N. Paddon-Row, M. S. Sherburn et al.
TIPS), which would result in a higher selectivity for the
trans, unlike adduct 33.
Closing Remarks
The synthetic scope and limitations of the IMDA allylic ster-
eocontrol method have been investigated with respect to al-
lylic substituents and dienophile geometry, and a theoretical
model has been devised to explain these findings. The com-
putational results offer unprecedented insights into the
subtle stereocontrolling influences at play in these syntheti-
cally important reactions, and point the way forward for
future investigations into additions to chiral allylic systems.
The unlike p-diastereofacial selectivity observed in these re-
actions comes about through the preference for a reactive
conformation about the bond connecting the diene and the
stereocenter in which the silyloxy group adopts an inside
orientation, as summarized in Figure 6. The dienophile
Anomalous fumarate cis-stereoselectivities: In the maleate
series, for both SiH3 (Figure 2) and SiMe3 (see Figure S1[28]
in the Supporting Information) substrates, TSs leading to
the trans-fused bicyclic products are considerably more
stable than those leading to the cis-fused adducts. These re-
sults are consistent with the experimental findings, and are
in agreement with computational and synthetic investiga-
tions of smaller achiral pentadienyl maleate substrates.[4]
In the case of fumarate systems, TSs leading to trans- and
cis-fused adducts are generally quite similar in energy, with
a slight trans IMDA selectivity generally witnessed in the
laboratory with small achiral substrates.[4,7,10,31,32] In stark
contrast, substrates 1 f, 2b, and 4e undergo IMDA reactions
with anomalously high (up to 85:15) cis-selectivities. The
origin of this unprecedented preference for the cis-cycload-
duct can be traced, once again, to preferred inside location
of the C*-OSiR3 substituent (Figure 5). In the fumarate
trans-TSs, the E-CO2Me group must reside directly beneath
the inside silyloxy group, which necessitates a destabilizing
gauche conformation between the SiH3 and C*-CH3 groups
ꢀ
about the C* O bond (highlighted in the dashed box in
Figure 5). In the corresponding cis-TS, the E-CO2Me group
is located in the exo position, hence the inside silyl group is
Figure 6. Theoretical model for p-diastereofacial selectivity in addition
reactions to chiral allylic silyloxy systems.
ꢀ
free to adopt a lower energy conformation about the C* O
bond in which it avoids the C*-methyl group (note the posi-
tioning of the SiH3 group in the endo region in cis, unlike-
TS-(g)). The destabilization of trans, unlike-4gTS-(a), rela-
tive to cis, unlike-4gTS-(g), will be amplified by the pres-
ence of larger silyl groups, which will result in a higher se-
lectivity for the cis-fused adduct.
reacts at the more accessible p-face of the diene; in the ab-
sence of overriding electronic effects, this will be on the side
of the diene in which the smaller substituent resides. This
model, which also explains the outcomes of intermolecular
Diels–Alder reactions of acyclic chiral dienols and deriva-
tives,[29b,c] is being further explored and tested.
A particularly important finding from both experimental
and computational studies is the requirement for the C*-
oxygen atom to be electron rich, thereby strengthening its
preference for the inside position by stabilizing electrostatic
interactions with the diene C2-H group. Thus, the silyloxy
group is much superior to hydroxy, alkoxy, and ester groups
in its stereocontrolling ability.
The enantiomerically pure cycloadducts formed in these
reactions are rich in functionality, which invites further
transformation into structures common with natural prod-
ucts. For example, the mixture of ascorbic acid derived sec-
ondary alcohols 20a and 21a underwent smooth transacetal-
ization[43] to a separable mixture of primary alcohols 48 and
49 (Scheme 11). The major diastereoisomer was converted
to the primary iodide[44] and the acetal group was removed.
The resulting diol underwent radical cyclization with tris-
(trimethylsilyl)silane[45] to give tricycle 52, with the expected
cis-indane geometry.[46] The tricylic framework of 52 is
found naturally in marasmane sesquiterpenes.[47]
Figure 5. The two most stable (-OSiH3 inside, -CH3 anti) fumarate TSs
leading to the trans, unlike (left) and cis, unlike (right) adducts. TS (g) is
energetically favored over TS (a). Top: views down the C9–C8 axis (C3–
ꢀ
C8 are omitted for clarity); bottom: views down the C* O bond. The
gauche interaction between the SiH3 and CH3 groups is highlighted in the
dashed box. This conformation is brought about by the steric (and possi-
bly electronic) destabilization in the endo region in TS (a) indicated in
the top left depiction.
2534
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 2525 – 2536

Allylic Stereocontrol of the Intramolecular Diels–Alder Reaction
FULL PAPER
[3] We use the terms cis and trans in place of endo and exo to identify
TSs and products. endo and exo are ambiguous terms when describ-
ing TSs and products of (E)-1,2-disubstituted dienophiles.
[4] M. J. Lilly, M. N. Paddon-Row, M. S. Sherburn, C. I. Turner, Chem.
Commun. 2000, 2213–2214.
[5] M. N. Paddon-Row, M. S. Sherburn, Chem. Commun. 2000, 2215–
2216.
[6] D. J. Tantillo, K. N. Houk, M. E. Jung, J. Org. Chem. 2001, 66, 1938–
1940.
[7] C. I. Turner, R. M. Williamson, M. N. Paddon-Row, M. S. Sherburn,
J. Org. Chem. 2001, 66, 3963–3969.
[8] T. N. Cayzer, L. S.-M. Wong, P. Turner, M. N. Paddon-Row, M. S.
Sherburn, Chem. Eur. J. 2002, 8, 739–750.
[9] G. A. Jones, M. N. Paddon-Row, M. S. Sherburn, C. I. Turner, Org.
Lett. 2002, 4, 3789–3792.
[10] T. N. Cayzer, M. N. Paddon-Row, M. S. Sherburn, Eur. J. Org. Chem.
2003, 4059–4068.
[11] For preliminary communications, see: M. J. Lilly, M. S. Sherburn,
Chem. Commun. 1997, 967–968 and reference [5].
[12] For seminal contributions to tether substituent-based stereocontrol,
see: a) T. Kametani, H. Matsumoto, H. Nemoto, K. Fukumoto, J.
Am. Chem. Soc. 1978, 100, 6218–6220; b) W. R. Roush, J. Org.
Chem. 1979, 44, 4008–4010; c) T. Mukaiyama, N. Iwasawa, Chem.
Lett. 1981, 29–32; d) M. P. Edwards, S. V. Ley, S. G. Lister, Tetrahe-
dron Lett. 1981, 22, 361–364; e) K. C. Nicolaou, R. L. Magolda, J.
Org. Chem. 1981, 46, 1506–1508; f) W. R. Roush, A. G. Myers, J.
Org. Chem. 1981, 46, 1509–1511; g) K. A. Parker, T. Iqbal, J. Org.
Chem. 1982, 47, 337–342; h) R. K. Boeckman, T. E. Barta, J. Org.
Chem. 1985, 50, 3421–3433; i) W. R. Roush, M. Kageyama, Tetrahe-
dron Lett. 1985, 26, 4327–4330.
[13] For seminal contributions to dienophile auxiliary-based stereocon-
trol, see: a) W. R. Roush, H. R. Gillis, A. I. Ko, J. Am. Chem. Soc.
1982, 104, 2269–2283; b) D. A. Evans, K. T. Chapman, J. Bisaha,
Tetrahedron Lett. 1984, 25, 4071–4074; c) W. Oppolzer, D. Dupuis,
Tetrahedron Lett. 1985, 26, 5437–5440.
Scheme 11. Synthesis of 52: a) Amberlyst resin, Me2CO, RT, 21 h, 96%;
b) imidazole (1.9 equiv), Ph3P (1.6 equiv), I2 (1.5 equiv), CHCl3, RT, 20 h,
67%; c) Amberlyst resin, MeOH:H2O (5:1), reflux, 18 h, 82%; d) (Me3-
Si)3SiH (1.1 equiv), AIBN, PhH, reflux, 45 min, 64%.
[14] a) N. Iwasawa, J. Sugimori, Y. Kawase, K. Narasaka, Chem. Lett.
1989, 1947–1950; b) K. Furuta, A. Kanematsu, H. Yamamoto, S. Ta-
kaoka, Tetrahedron Lett. 1989, 30, 7231–7232; c) K. Narasaka, M.
Saitou, N. Iwasawa, Tetrahedron: Asymmetry 1991, 2, 1305–1318;
d) K. Ishihara, H. Kurihara, H. Yamamoto, J. Am. Chem. Soc. 1996,
118, 3049–3050; e) D. A. Evans, J. S. Johnson, J. Org. Chem. 1997,
62, 786–787; f) K. Ishihara, H. Kurihara, M. Matsumoto, H. Yama-
moto, J. Am. Chem. Soc. 1998, 120, 6920–6930; g) D. A. Evans,
D. M. Barnes, J. S. Johnson, R. Lectka, P. von Matt, S. J. Miller, J. A.
Murry, R. D. Norcross, E. A. Shaughnessy, K. R. Campos, J. Am.
Chem. Soc. 1999, 121, 7582–7594; h) G. Zhou, Q. Y. Hu, E. Corey,
J. Org. Lett. 2003, 5, 3979–3982.
Acknowledgements
Funding from the Australian Research Council is gratefully acknowl-
edged, as are computing time allocations from the Australian Partnership
for Advanced Computing and the Australian Centre for Advanced Com-
puting and Communications. We also thank Dr. G. A. Jones for assis-
tance.
[15] a) B. Fraser-Reid, Z. Benko, R. Guiliano, K. M. Sun, N. Taylor, J.
Chem. Soc. Chem. Commun. 1984, 1029–1030; b) J. E. Baldwin, R.
Chesworth, J. S. Parker, A. T. Russell, Tetrahedron Lett. 1995, 36,
9551–9554; c) D. J. Hart, J. Li, W. L. Wu, A. P. Kozikowski, J. Org.
Chem. 1997, 62, 5023–5033; d) S. Hofman, G. Debaecke, B. Kenda,
P. J. Declercq, Synthesis 1998, 479–489.
[16] a) A. Ichihara, S. Miki, H. Kawagishi, S. Sakamura, Tetrahedron
Lett. 1989, 30, 4551–4554; b) S. Miki, Y. Sato, H. Tabuchi, H.
Oikawa, A. Ichihara, S. Sakamura, J. Chem. Soc. Perkin Trans. 1
1990, 1228–1229; c) D. A. Evans, W. C. Black, J. Am. Chem. Soc.
1993, 115, 4497–4513.
[1] a) D. F. Taber, Intramolecular Diels-Alder and Alder Ene Reactions,
Springer, Berlin, 1984; b) A. G. Fallis, Can. J. Chem. 1984, 62, 183–
234; c) E. Ciganek, Org. React. 1984, 32, 1–374; d) D. Craig, Chem.
Soc. Rev. 1987, 16, 187–238; e) W. Roush in Advances in Cycloaddi-
tion, Vol. 2 (Ed.: D. P. Curran), JAI, Greenwich, CT, 1990, pp. 91–
146; f) W. R. Roush in Comprehensive Organic Synthesis, Vol. 5
(Eds.: B. M. Trost, I. Fleming, L. A. Paquette), Pergamon, Oxford,
1991, pp. 513–550; g) T. Oh, M. Reilly, Org. Prep. Proc. Int. 1994,
26, 129–158; h) D. Craig in Stereoselective Synthesis, 4th ed. (Eds.:
G. Helmchen, R. W. Hoffmann, J. Mulzer, E. Schumann) Methods
of Organic Chemistry (Houben-Weyl) No. E21c, Thieme, Stuttgart,
1995, pp. 2872–2904; i) A. G. Fallis, Acc. Chem. Res. 1999, 32, 464–
474; j) B. R. Bear, S. M. Sparks, K. J. Shea, Angew. Chem. 2001, 113,
864–894; Angew. Chem. Int. Ed. 2001, 40, 820–849.
[2] a) S. Laschat, Angew. Chem. 1996, 108, 313–315; Angew. Chem. Int.
Ed. Engl. 1996, 35, 289–291; b) A. Ichihara, H. Oikawa, Curr. Org.
Chem. 1998, 2, 365–394; c) G. Pohnert, ChemBiochem 2001, 2, 873–
875; d) E. M. Stocking, R. M. Williams, Angew. Chem. 2003, 115,
3186–3223; Angew. Chem. Int. Ed. 2003, 42, 3078–3115; e) M. C.
de la Torre, M. A. Sierra, Angew. Chem. 2004, 116, 162–184;
Angew. Chem. Int. Ed. 2004, 43, 160–181.
[17] 1,3,8-Nonatriene numbering is retained for the esters described in
this paper to facilitate comparisons with other IMDA precursors of
the same general class.
[18] a) O. Wiest, D. C. Montiel, K. N. Houk, J. Phys. Chem. A 1997, 101,
8378–8388; b) O. Wiest, K. N. Houk in Topics in Current Chemistry,
Vol. 183, (Ed.: R. F. Nalewajski), Springer, Berlin, 1996, pp. 1–24.
[19] M. E. Jung, T. J. Shaw, J. Am. Chem. Soc. 1980, 102, 6304–6311.
[20] E. Abushanab, P. Vemishetti, R. W. Leiby, H. K. Singh, A. B. Mikki-
lineni, D. C.-J. Wu, R. Saibaba, R. P. Panzica, J. Org. Chem. 1988,
53, 2598–2602. This paper provides a detailed experimental proce-
dure for preparing a diastereoisomer of 7 starting from d-isoascorbic
Chem. Eur. J. 2005, 11, 2525 – 2536
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2535

M. N. Paddon-Row, M. S. Sherburn et al.
acid. An earlier reference (E. Abushanab, M. Bessodes, K. Antona-
kis, Tetrahedron Lett. 1984, 25, 3841–3844) outlines the synthesis of
the corresponding methyl ester of 7. Although both of these papers
refer to the preparation of compound 7, neither of them report its
physical and spectroscopic properties.
[36] All TSs were fully optimized and characterized by harmonic fre-
quency calculations. The Gaussian 98 package was used for all calcu-
lations: M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria,
M. A. Robb, J. R. Cheeseman, V. G. Zakrzewski, J. A. Montgomer-
y Jr., R. E. Stratmann, J. C. Burant, S. Dapprich, J. M. Millam, A. D.
Daniels, K. N. Kudin, M. C. Strain, O. Farkas, J. Tomasi, V. Barone,
M. Cossi, R. Cammi, B. Mennucci, C. Pomelli, C. Adamo, S. Clif-
ford, J. Ochterski, G. A. Petersson, P. Y. Ayala, Q. Cui, K. Moroku-
ma, P. Salvador, J. J. Dannenberg, D. K. Malick, A. D. Rabuck, K.
Raghavachari, J. B. Foresman, J. Cioslowski, J. V. Ortiz, A. G.
Baboul, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaro-
mi, R. Gomperts, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham,
C. Y. Peng, A. Nanayakkara, M. Challacombe, P. M. W. Gill, B.
Johnson, W. Chen, M. W. Wong, J. L. Andres, C. Gonzalez, M.
Head-Gordon, E. S. Replogle, J. A. Pople, in Gaussian 98, Revision
A.11, Gaussian, Inc., Pittsburgh PA, 2001.
[21] S. Saito, T. Ishikawa, A. Kuroda, K. Koga, T. Moriwake, Tetrahedron
1992, 48, 4067–4086.
[22] N. Smith, P. Kocienski, S. Street, Synthesis 1996, 652.
[23] a) M. W. Bredenkamp, J. S. Lesch, J. S. Malherbe, E. M. Molnar,
D. F. Schneider, Tetrahedron Lett. 1980, 21, 4199–4202; b) R. W.
Lang, H.-J. Hansen, Org. Synth. 1984, 62, 202–209.
[24] R. Annunziata, M. Cinquini, F. Cozzi, C. Gennari, L. Raimondi, J.
Org. Chem. 1987, 52, 4674–4681.
[25] K. Sato, S. Mizuno, M. Hirayama, J. Org. Chem. 1967, 32, 177.
[26] D. B. Dess, J. C. Martin, J. Org. Chem. 1983, 48, 4155–4156.
[27] Wadsworth–Horner–Emmons reaction between aldehyde 12 and the
lithium salt of phosphonate 17 gave the (E,E)-dienoate ester in a
similarly disappointing yield.
[28] CCDC-243538 (34a), CCDC-243536 (37), CCDC-244532 (38),
CCDC-243539 (39), and CCDC-243537 (40) contain the supplemen-
tary crystallographic data for this paper. These data can be obtained
free of charge from The Cambridge Crystallographic Data Centre
via www.ccdc.cam.ac.uk/data_request/cif.
[29] We use the Seebach–Prelog descriptor like to describe a cycloadduct
resulting from the approach of the dienophile to the re face of the
diene with an allylic stereocenter of R configuration. The term
unlike refers to si/R and re/S combinations, see: a) D. Seebach, V.
Prelog, Angew. Chem. 1982, 94, 696; Angew. Chem. Int. Ed. Engl.
1982, 21, 654–660. For consistent use of this convention, the priori-
ties of the groups about the allylic stereocenter are assigned such
that the sp2 carbon atom has a higher priority than the sp3 carbon
atom, see: b) R. Tripathy, R. W. Franck, K. D. Onan, J. Am. Chem.
Soc. 1988, 110, 3257–3262; c) W. Adam, J. Glaser, K. Peters, M.
Prein, J. Am. Chem. Soc. 1995, 117, 9190–9193.
[37] Each TS was fully optimized—that is, no geometrical constraints
were imposed—and characterized by carrying out a harmonic vibra-
tional frequency calculation. In principle, each TS corresponding to
ꢀ
a particular conformation about the C* C1 bond (torsional angle
ꢀ
y) is capable of exhibiting conformational isomerism about the C*
O bond (torsional angle q), there being three such isomers. These
three C*–O conformational isomeric TSs were not fully optimized
for reasons of computational economy. Instead, for each TS which
was optimized with respect to y, a potential energy curve was calcu-
lated with respect to q. For each value of q of this scan, the geome-
try of the molecule was fully optimized with the exception that the
forming bonds between the diene and dienophile were frozen at the
values corresponding to those typically found for this type of IMDA
TS. Such a potential energy scan resulted in the location of three
energy minima corresponding to three different values of q. The ge-
ometry corresponding to the lowest energy minimum was then fully
optimized—with no geometrical constraints imposed—to give the
TS corresponding to that value of q. Although the TSs correspond-
ing to the other two values of q were neither optimized, nor further
considered, our procedure almost certainly guarantees that we have
located the lowest energy TS conformation for that particular value
of y. The fact that we obtain such good agreement between the ex-
perimental and computational selectivities strongly suggests that our
procedure is a valid one.
[30] For a review on stereochemical control using silicon-containing com-
pounds, see: I. Fleming, A. Barbero, D. Walter, Chem. Rev. 1997, 97,
2063–2192.
[31] For seminal investigations into IMDA reactions of this general type,
see: J. D. White, B. G. Sheldon, J. Org. Chem. 1981, 46, 2273–2280.
[32] For synthetic studies on IMDA reactions of pentadienyl maleates
and fumarates, see: a) S. D. Burke, S. M. Smith Strickland, T. H.
Powner, J. Org. Chem. 1983, 48, 454–459; b) P. R. Jenkins, K. A.
Menear, P. Barraclough, M. S. Nobbs, J. Chem. Soc. Chem.
Commun. 1984, 1423–1424; c) P. Magnus, C. Walker, P. R. Jenkins,
K. A. Menear, Tetrahedron Lett. 1986, 27, 651–654; d) M. K. Eberle,
H. P. Weber, J. Org. Chem. 1988, 53, 231–235; e) J. F. He, Y. L. Wu,
Tetrahedron 1988, 44, 1933–1940; f) M. J. Batchelor, J. M. Mellor, J.
Chem. Soc. Perkin Trans. 1 1989, 985–995; g) J. Becher, H. C. Niel-
sen, J. P. Jacobsen, O. Simonsen, H. Clausen, J. Org. Chem. 1988, 53,
1862–1871; h) M. Toyota, Y. Wada, K. Fukumoto, Heterocycles
1993, 35, 111–114; i) L. Berthon, A. Tahri, D. Uguen, Tetrahedron
Lett. 1994, 35, 3937–3940; j) K. Takatori, K. Hasegawa, S. Narai, M.
Kajiwar, Heterocycles 1996, 42, 525–528; k) S. Arseniyadis, R.
Brondi-Alves, D. V. Yashunsky, P. Potier, L. Toupet, Tetrahedron
1997, 53, 1003–1014; l) T. N. Cayzer, M. N. Paddon-Row, D. Moran,
A. D. Payne, M. S. Sherburn, P. Turner, J. Org. Chem. submitted.
[33] The IMDA reaction of the corresponding 2-butynoate ester was
thwarted by cycloadduct aromatization under the reaction condi-
tions.
[38] a) S. L. Shambayati, J. F. Blake, S. G. Wierschke, W. Jorgenson, S. L.
Schreiber, J. Am. Chem. Soc. 1990, 112, 697–703; b) B. W. Gung,
J. P. Melnick, M. A. Wolf, A. King, J. Org. Chem. 1995, 60, 1947–
1951.
[39] K. N. Houk, S. R. Moses, Y.-D. Wu, N. G. Rondan, V. Jꢃger, R.
Schohe, F. R. Fronczek, J. Am. Chem. Soc. 1984, 106, 3880–3882.
[40] P. J. Ainsworth, D. Craig, J. C. Reader, A. M. Z. Slawin, A. J. P.
White, D. J. Williams, Tetrahedron 1995, 51, 11601–11622.
[41] A. E. Reed, L. A. Curtiss, F. Weinhold, Chem. Rev. 1988, 88, 899–
926.
ꢀ
[42] The out-of-plane C O/C=C conformation is preferred for allylic al-
cohol benzoates, see: N. C. Gonnella, K. Nakanishi, V. S. Martin,
K. B. Sharpless, J. Am. Chem. Soc. 1982, 104, 3775–3776.
[43] J. W. Coe, W. R. Roush, J. W. Coe, W. R. Roush, J. Org. Chem. 1989,
54, 915–930.
[44] G. L. Lange, C. Gottardo, Synth. Commun. 1990, 20, 1473–1479.
[45] C. Chatgilialoglu, Acc. Chem. Res. 1992, 25, 188–194.
[46] a) D. L. J. Clive, D. R. Cheshire, L. Set, J. Chem. Soc. Chem.
Commun. 1987, 353–355; b) D. L. J. Clive, H. W. Manning, T. L. B.
Boivin, M. H. D. Postema, J. Org. Chem. 1993, 58, 6857–6873.
[47] W. A. Ayer, L. M. Browne, Tetrahedron 1981, 37, 2199–2248.
[34] M. N. Paddon-Row, N. G. Rondan, K. N. Houk, J. Am. Chem. Soc.
1982, 104, 7162–7166.
[35] K. N. Houk, M. N. Paddon-Row, N. G. Rondan, Y. D. Wu, F. K.
Brown, D. C. Spellmeyer, J. T. Metz, Y. Li, R. J. Loncharich, Science
1986, 231, 1108–1117.
Received: November 28, 2004
Published online: February 24, 2005
2536
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 2525 – 2536