Nonenzymic polycyclisation of analogues of oxidosqualene with a preformed C-ring

Article abstract of DOI:10.1039/b801670d

Some nonenzymic epoxide-initiated polyolefin cyclisations are reported. The presented molecules are partially constrained analogues of (3S)-oxidosqualene, the natural substrate to many important cyclase enzymes. These model compounds feature a preformed C

Full text of DOI:10.1039/b801670d

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/obc | Organic & Biomolecular Chemistry
Nonenzymic polycyclisation of analogues of oxidosqualene with a preformed
C-ring†
Johan M. Winne,a Pierre J. De Clercq,*a Marco Milanesio,b Philip Pattisonc,d and Davide Viterbob
Received 30th January 2008, Accepted 5th March 2008
First published as an Advance Article on the web 8th April 2008
DOI: 10.1039/b801670d
Some nonenzymic epoxide-initiated polyolefin cyclisations are reported. The presented molecules are
partially constrained analogues of (3S)-oxidosqualene, the natural substrate to many important cyclase
enzymes. These model compounds feature a preformed C-ring with built-in stereochemical
information. The experimental results allow for an instructive comparison with the enzymic processes,
particularly those of the cyclases in steroid biosynthesis (i.e. lanosterol synthase).
Introduction
From a synthetic point of view, the natural polycyclisation
reactions of (3S)-2,3-oxidosqualene (OS) are arguably the most
efficient single-step transformations known to man. The enzymes
involved, which give rise to a broad range of complex tetra-
and pentacyclic triterpenoids, have been shown to cyclise their
linear substrate into either of the shown possible tetracyclic
cationic intermediates (dammarenyl or protosteryl cation, see
graphical overview in Scheme 2).1 Next to the remarkable product
specificity and selectivity of these carbocationic processes, their
most striking feature is the six-membered C-ring. All known
nonenzymic cationic OS cyclisations afford complex mixtures of
structures with at most three carbocyclic rings and, invariably, a
five-membered C-ring (Scheme 1, top).2 Furthermore, the stereo-
chemical discrepancies between the protosteryl and dammarenyl
cation are completely localised in the C-ring area (inversion of
every stereocenter) and are explained by different folding patterns
of the polyene chain inside the enzyme cavity. A chair–chair–
chair conformation is required to reach the dammarenyl cation,
whereas the protosteryl cation is the result of an alternative chair–
boat–chair fold. This prerequisite, first postulated in the Zu¨rich
proposal,3 has received convincing experimental support from
various areas.1a,3b
Biomimetic cationic olefin polycylisations, as pioneered by
Stork,4,5 offer a good opportunity to expose the fundamental
issues an OS cyclase has to address. In particular, several synthetic
groups have aimed to (partly) emulate these intriguing bio-
genetic reactions by designing polyenes with incorporated cation-
Scheme 1
stabilizing (C-S) groups and/or conformational restrictions.5
These modifications can be regarded as mimics of the role of the
cyclase enzymes, which is still unclear. Most noteworthy are the
achievements of William S. Johnson’s group in efficient cationic
tetra- and pentacyclisations by using initiating groups other than
epoxides and various C-S substituents on the reacting alkenes,
which also ensure six-membered C-ring formation.6 Our group on
the other hand has reported a few examples of nonenzymic polyene
cyclisations leading to the familiar ‘natural’ 6 : 5 CD-ring system.7
Interestingly, this did not require the use of any C-S groups, but
merely resulted from restraining the possible conformations of the
central part of the polyene chain (see for example aldehyde 17a in
Scheme 1).
aGhent University, Department of Organic Chemistry Krijgslaan 281 (S4),
B-9000 Gent, Belgium. E-mail: pierre.declercq@ugent.be; Fax: +32 9264
4998; Tel: +32 9264 4461
bDipartimento di Scienze e Tecnologie Avanzate and Nano-SiSTeMI In-
terdisciplinary Centre, Universita` del Piemonte Orientale “A. Avogadro” -
Via V. Bellini 25/G, I-15100 Alessandria, Italy
cSwiss-Norwegian Beamline, ESRF, BP220, F-38043 Grenoble Cedex,
France
dLaboratory of Crystallography, EPFL, CH-1015 Lausanne, Switzerland
† Electronic supplementary information (ESI) available: Experimental
procedures and spectroscopic data, 2D NMR data for compounds 20, 21a
and 22 and structural elucidation thereof, copies of selected NMR spectra
and table and figures on X-ray diffraction and computational analysis
of compound 21a. CCDC reference number 675998. For crystallographic
data in CIF or other electronic format and ESI see DOI: 10.1039/b801670d
1918 | Org. Biomol. Chem., 2008, 6, 1918–1925
This journal is
The Royal Society of Chemistry 2008
©

View Article Online
The cyclohexane ring (designated as Cꢀ-ring) grafted on the
polyene backbone of the model aldehyde 1 was designed to enforce
a chair-like fold of the pre-C-ring 1,5-diene.7b Regrettably, these
studies had to be limited to polyenes with a simplified initiating
part, as the extraneous Cꢀ-ring seemed to interfere with the early
cyclisation stages in the case of a full OS structure.8 The influence
of the single localised conformational restriction (imposed torsion
around the C11,C12-bond, using classical steroid numbering)
on the course of BCD-cyclisation was, however, striking (see
polycyclised structures 2 and 3). We felt compelled to further
evaluate the possible effect of this conformational restriction on
the full course of OS cyclisation (epoxide initiated ABCD(E)-
cyclisation). It should be clear that, at least in a concerted pathway,
the preorganisation of the pre-C-ring polyene part in either of the
possible chair-like folds (defining a helical constraint relative to the
epoxide, see Scheme 2) could be sufficient to direct the early AB-
ring system formation (chair–chair or chair–boat cyclisation). This
would mean that the entire stereochemical course of the enzymic
OS tetracyclisation is essentially caused by a relatively minor
and localised conformational restriction (such as the C11,C12
torsional angle).
For the present study, which was focused on the AB-ring system
formation stage, new model compounds 4 and 5 were designed
(Scheme 2). They feature a preformed C-ring that should secure
the desired preorganisation of the central part of the OS polyene
chain, by controlling the configurations of the C-ring stereocenters
relative to the epoxide function. A conjugated diene in the
preformedringwas considered to be the terminatingnucleophile of
choice. It should make AB-ring cyclisation effectively irreversible,
assuring a true kinetic polycyclisation result. Furthermore, this
diene can act as a relay for the cationic reaction center to-
wards the D-ring cyclising stage, allowing for the formation of
interesting dammarane or protolanostane skeletons. The diene
system should also increase overall polycyclisation efficiency as
an internal C-S group. Finally, these models have the advantage
that no unnatural—possibly sterically hindering—substituents are
present.
Results and discussion
Synthesis of model compounds
For the convergent synthesis of the target epoxypolyenes 4 and
5 (Scheme 3) we envisioned symmetrical disconnections. These
allowed for a general strategy to introduce the required side chain
fragments on the more challenging 1,6-difunctionalised central
part, which was obtained using a reconnection strategy via the
synthetically more attractive trans-fused decalin systems.
The easily accessible (R)-octalone 6 was recognized as a
convenient enantiomerically enriched starting point.9 Following
Ireland’s strategy for introducing an angular methyl group in
the trans-position (Nagata hydrocyanation followed by nitrile
reduction),10 octalone 6 was converted to the ketal 7. Prior removal
of the ketone function via Shapiro reaction was essential for the
success of the nitrile reduction. The non-nucleophilic base lithium
hydride gave the best results for this highly regioselective reaction.
Ozonolysis of the double bond followed by reductive work-up
and silylation then gave the 1,6-bissilyl ether 8 as an advanced
intermediate. Finally, the required cyclohexadiene building block
9 was obtained via Shapiro reaction giving the corresponding
alkene which was then converted to the conjugated diene via a
bromination–elimination sequence. As no conclusive evidence for
the trans-fusion in these intermediates was obtained so far, we were
pleased to observe a large optical rotation for the C2-symmetrical
compound 9 ([a]2D0 +110 (c 1.36 in CHCl3)), which excludes the
possibility of the achiral S1-symmetrical cis-isomer.
After some experimentation, we developed a general and direct
strategy to introduce the initiating and terminating OS side chain
fragments, based on the B-alkyl Suzuki–Miyaura cross-coupling,
as has been reported in a preliminary communication.11 The
alkylborane reactants were prepared in situ via hydroboration of
the terminal alkene 10, easily obtained from 9 via monodesilylation
and Grieco–Sharpless elimination. Consecutive coupling of the
appropriate iodo alkene and (S)-epoxy iodo alkene in this manner,
gave the target epoxy polyene 5. Polyene 12 was coupled with
Scheme 2 Enzymic polycyclisation of (3S)-oxidosqualene and proposed model compounds: pathway (a) is followed in plants for biogenesis of many
tetra- and pentacyclic triterpenes, pathway (b) is followed for formation of lanosterol, cycloartenol or parkeol as steroid precursors.
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 1918–1925 | 1919
©

View Article Online
Scheme 3 Reagents and conditions: i, 3 equiv. Et2AlCN, THF–PhMe, −5 ◦C to rt, 80%; ii, (a) TsN2H3, TsOH, mol. siev., THF, D, (b) 5 equiv. LiH,
THF–PhMe, D, 80–85%; iii, i-Bu2AlH, PhMe, 0 ◦C to rt, quant.; iv, N2H4·xH2O, N2H4·HCl, TEG, 135 ◦C, then excess KOH, rt to 210 ◦C (distill.), 95%;
v, O3, CH2Cl2, MeOH, −78 ◦C, then NaBH4, −78 ◦C to −15 ◦C; vi, 4 equiv. t-BuPh2SiCl, imidazole, DMAP, DMF, rt, 80% (v + vi), vii, TsOH, H2O,
acetone, D, quant.; viii, (a) NBS, bpo, CCl4,. D, (b) KOt-Bu, THF, D, (c) SnBu3H, AIBN, THF, D, 70% (a + b + c); ix, TBAF, Me4NOH, THF, 75% (from
9), 80–90% (from 11 or 11ꢀ), 65% (from 13 or 15); x, (a) 2 equiv. o-NO2PhSeCN, 2 equiv. PMe3, THF, (b) 2 equiv. H2O2, THF, 70–80%; xi, (a) 1.5 equiv.
Cy2BH, THF, 0 ◦C to rt, then H2O, (b) Cs2CO3, Ph3As, PdCl2(dppf), 0.96 equiv. (E)-1-iodo-2,6-dimethyl-hepta-1,5-diene or 1-bromo-2-methyl-propene
or (3S)- or (3R)-(4-iodo-3-methyl-but-3-enyl)-2,2-dimethyl-oxirane, DMF, rt, 70–80% (from 10), 50% (from 12 or 12ꢀ).
the enantiomeric epoxide-bearing side chain fragment to afford
epoxypolyene ent-4 which is the mirror image of the originally
proposed target structure 4 and was subsequently used in its stead.
Early cyclisation experiments with these models quickly showed
it would be desirable to have analogues with simplified terminating
chains. The Suzuki–Miyaura coupling based strategy allowed for
a straightforward synthesis of the truncated analogues ent-4ꢀ and
5ꢀ using 1-bromo-2-methylpropene as the terminating side chain
fragment (Scheme 3). Furthermore, by coupling silyloxypolyene10
directly with the initiating side chains and simply deprotecting the
silyl ether, the epoxy polyenols 14 and 16 were obtained. In these
cases, the hydroxyl group was expected to function as a terminating
nucleophile for the polycyclisation, affording tetracycles with a
tetrahydrofuran type D-ring.
Scheme 4
Cyclisation experiments
Following the Stork–Eschenmoser postulate,12 which advocates
stereocontrolled antiperiplanar additions on both the epoxide and
reacting alkenes, and reasonably assuming a chair-like fold for
formation of the A-ring, only a limited number of stereochemical
outcomes are expected for the AB-ring system upon polycyclisa-
tion of our model compounds (Scheme 4). The two alternative
AB-cyclisation results for each of the examined diastereomers are
distinguished by either a 9,10-anti or a 9,10-syn stereochemical
relation (classical steroid numbering is used), which would reflect
a B-chair or B-boat fold of the polyene, respectively. A priori,
it was expected that a relaxed B-chair fold would only be
possible in the models corresponding to the dammarenyl-type
configuration (14, ent-4, ent-4ꢀ), because a B-chair conformation in
the diastereomeric protosteryl-type models (16, 5, 5ꢀ) would suffer
from a severe syn-diaxial relation of the C-10 methyl substituent
and the very bulky C-14 quaternary center (see Fig. 1). It was
hoped that in this case B-boat cyclisation and thus a 9,10-syn
stereochemistry would result, a pathway which only very rarely
has been observed in nonenzymic reactions.13
The easily accessible models 14 and 16 were used to screen
the most commonly used conditions to achieve cationic polycy-
clisation of epoxy polyenes. The results for the protosteryl type
1920 | Org. Biomol. Chem., 2008, 6, 1918–1925
This journal is
The Royal Society of Chemistry 2008
©

View Article Online
Table 1 Cyclisation of protosteryl type model compounds 16 and 5ꢀ
Reaction conditions
Yield of 17
Yield of 18
Yield of 19
(i) 16, 1.0 equiv. picric acid, CH3NO2, rt
—
—
—
10%
25%
40%
15%
30%
—
(80%)a
30%b
10%b
3%b
(ii) 16, 1.5 equiv. MeAlCl2, CH2Cl2, −78 to −15 ◦
C
35%
30%
17%
24%
(iii) 16, 1.0 equiv. SnCl4, CH2Cl2, −78 to −25 ◦
C
(iv) 16, 0.25 equiv. SnCl4, CH2Cl2, 10 ◦
C
9%c
(v) 5ꢀ, 0.25 equiv. SnCl4, CH2Cl2, 10 ◦C
21%d
12%b
a Isolated yield of a complex fraction which contains diol 19a as a major isomer (NMR). b Isolated fraction contains minor amounts (10–20%) of
regioisomeric alkenes and HCl addition products thereof (NMR). c Isolated yield of tetracyclic alcohol 20 (Scheme 5). d Isolated yield of a 17 : 3 mixture
of protolanostane-type alcohols 21a and 21b (Scheme 5).
Fig. 1
model 16 can be viewed in Table 1. Sharpless’ protic conditions
(entry (i))14 using picric acid in nitromethane gave a very complex
mixture of mostly dihydroxylic material with an intact conjugated
diene system. The mild methylaluminium dichloride Lewis acid
introduced by Corey as an efficient epoxypolyene cyclisation
promoter (entry (ii))15 gave two major monocyclised products 17a
and 18a next to some more polar impure fractions of the type
19a. Use of stannic chloride (entry (iii)), as originally introduced
by Stork,4 gave similar results. Sharpless and van Tamelen have
Scheme 5
optimised the nonenzymic polycyclisation of OS (Scheme 1) and
found that maximal polycyclised (tricyclic) material was obtained
using substoichiometric amounts of stannic chloride at moderate
reaction temperatures.2 In our case (entry (iv)), this procedure gave
a quite complex mixture of rearranged or monocyclised material
(as judged by the presence of an intact conjugated diene system
in NMR spectra) of which only the main compounds 17a and
18a could be obtained in a pure form. However, this procedure
uniquely yielded an easily separable, less polar fraction of solid ma-
terial of the expected polycyclised kind. Furthermore, this fraction
turned out to be a single diastereomer (the trans–anti–cis–anti–cis-
tetracycle 20, see Scheme 5). This configuration was unequivocally
determined by detailed analyses of NMR spectra (APT, HSQC,
HMBC, COSY and NOESY). As discussed earlier, the 9,10-anti
relation should indicate a chair-like fold for B-ring formation
(Scheme 4). Interestingly, the observed NOE relations in this trans–
anti–cis-tetracycle 20 did not support a chair-like conformation of
the B-ring (which, as discussed, would be a very strained system,
cf. Fig. 1), but instead revealed a (twist-)boat-like conformation, as
can be reached by a cis-decalin type ring flip of the BC-ring system
(see structure with some diagnostic NOE relations in Fig. 2). Also
the C-ring clearly adopts an unexpected conformation, with both
methyl substituents in a pseudo-equatorial position. Regrettably,
Fig. 2
our attempts to grow a suitable single crystal of compound 20
failed, so this interesting conformation could not be confirmed by
X-ray diffraction analysis.
When conditions from entries (i)–(iii) were applied to the
diastereomeric model compound 14 (corresponding to a mirrored
dammarenyl type stereochemistry), roughly the same results were
obtained and only monocyclised material of the type 17a, 18a and
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 1918–1925 | 1921
©

View Article Online
19a was observed. Using the Sharpless–van Tamelen conditions
(iv), no polycyclised compounds could be isolated.
Therefore the crystals were analysed exploiting the high brilliance
of a synchrotron radiation source and the collected data allowed
solving and refining of the structure. The results obtained, together
with 2D-NMR analyses, allowed the full characterisation of the
configurational and conformational features of compound 21a.
Because of the employed wavelength and of the presence of light
atoms only, the absolute structure could not be determined by
diffraction only. However, among the two possible enantiomorph
molecules compatible with the X-ray data, the configuration
shown in Fig. 4 was chosen because it is the one in agreement with
the adopted synthetic strategy. In fact, it shows the correct absolute
configuration of the quaternary centres in the C-ring, which are
preformed by enantioselective synthesis and can be unequivocally
“back-tracked” to the original chiral starting material of known
absolute configuration (compound 6). Furthermore, the absolute
configuration of the initiating (S)-epoxide is expected to give a
hydroxyl group with the configuration as shown in Fig. 4.
Next, we applied the at least partially successful conditions
(cf. entry (iv)) to the ‘full’ models ent-4 and 5 with natural OS
side chains. Again, mostly compounds of the previously obtained
monocyclised type were isolated (cf. 17, 18 and 19). In both cases, a
fraction was isolated which indicated the presence of polycyclised
material (as judged by the absence of signals for the conjugated
diene in NMR spectra). However, in the case of the dammarenyl
type model ent-4 this was only a very small (2%) and extremely
complex mixture. In the case of the protosteryl-type model 5, this
fraction (12%) showed a major tetracyclic component but could
not be sufficiently purified to be identified. Careful analysis of
the NMR spectra did reveal some striking similarities with those
obtained for tetracycle 20. The signals that would correspond to
the D-ring region, however, were complex and showed splitting
indicative of a 1 : 1 mixture of isomers.
Fortunately, the product diversity could be greatly reduced by
truncating the terminating polyene chain by one isoprenic unit
(substrates ent-4ꢀ and 5ꢀ). These models were synthesised and
cyclised on a relatively large scale (80 and 120 mg respectively).
In the case of the protosteryl type model 5ꢀ, three major fractions
were isolated (entry (v), Table 1), one of which contained only
polycyclised material. The major constituent of this tetracyclic
fraction (21a, 85%(NMR)) was shown to possess the (9a,13b)-
D17,20-protolanostane skeleton (Scheme 5) via detailed NMR
analyses (APT, HSQC, HMBC, COSY, NOESY). The inseparable
minor isomer (21b, 15%(NMR)) was tentatively assigned a
(9a,13b)-D16,17-protolanostane structure based on the resolved
signals (mainly D-ring region). Both other major fractions con-
tained a single isomer of the previously obtained abortive A-
ring monocyclisation type (cyclic ether 17b and ketone 18b).
Furthermore, five minor fractions were isolated: a complex apolar
hydrocarbon fraction (6%), two fractions of the cyclohexenolic
A-ring monocyclised type (8 and 4%, structures 19b) and two very
minor polycyclised fractions, one complex (2.5%) and one which
could be assigned the unusual spirocyclic structure 22 (2%, Fig. 3)
via detailed analysis of NMR spectra. In total, 97% of the mass
balance was isolated.
Fig. 4
The crystal structure of 21a contains two molecules adopting
very similar conformations, with just small differences, as shown
in the superposition in Fig. SM1 in the supplementary material.†
As illustrated in Fig. 4, the A-ring adopts a regular chair confor-
mation, while the B-ring has a distorted twist-boat conformation
(endocyclic torsional angles in the two conformers, starting from
the C8 to C9 bond and then further around the ring in that
direction: −23.6, 64.4, −37.8, −28.9, 66.2, −39.9 and −2.7, 44.8,
−29.8, −30.6, 71.1, −53.4). The C-ring adopts a flattened half-
chair (or “sofa”) conformation, as expected by the presence of the
endocyclic double bond (black arrow in Fig. 4). Finally, another
interesting feature is the almost planar envelope conformation of
the D-ring (with four sp3 and one sp2 carbon atoms), despite the
absence of an endocyclic double bond.
The quantum mechanic (QM) geometry optimisations were
carried out to demonstrate that these peculiar conformational
features are due to the electronic structure of the molecule rather
than the unusual crystal packing (two almost identical conformers
packed along a long axis) and that these features are retained
also for the isolated molecule. After geometry optimisation at the
B3LYP/6-31G(d,p) level, the overall conformations of the two
molecules in the asymmetric unit converged to the same. The
(distorted) twist-boat conformation of the B-ring and the almost
planar conformation of the D-ring are retained, thus confirming
that the crystal structure is a minimum energy conformation
also as isolated molecule. The distortion of the B-ring can be
ascribed to the double bond in the C-ring (indicated by the black
arrow in Fig. 4). In other words, a perfect boat or twist-boat
conformation of the B-ring would distort the C-ring too much.
Fig. 3
Although obtaining sufficient 2D-NMR correlations for tetra-
cycle 21a was complicated by various overlaps and by the presence
of the minor regioisomer 21b, the data clearly showed the
conformation of this protolanostane skeleton to be essentially the
same as that established for tetracycle 20 (Fig. 2). Attempts at
growing single crystals only afforded very thin (<10 lm) plate-
like crystals not suitable for single crystal X-ray diffraction on
conventional diffractometers. This crystal morphology is probably
˚
due to the difficulty of growing a crystal along a long axis (c > 50 A,
see Experimental section), where the packing forces are very weak.
1922 | Org. Biomol. Chem., 2008, 6, 1918–1925
This journal is
The Royal Society of Chemistry 2008
©

View Article Online
The almost planar envelope conformation of the D-ring is imposed
by its exocyclic double bond and again the constraints due to
the double bond in the C-ring. To verify the relative stability
of the crystal structure conformation with its unusual B-ring
twist-boat, other conformations of compound 21a adopting more
conventional chair or boat conformations were generated by the
MOLDRAW software and optimised at the B3LYP/6-31G(d,p)
level. The hypothetical conformation with the B-ring in a standard
boat conformation did not represent an energy minimum. The
conformation with the B-ring assuming a chair conformation (see
Fig. SM2 in the supplementary material) resulted in a less stable
energy minimum (+8.2 kJ mol−1), and this instability explains the
preference for the (distorted) twist-boat B-ring conformation in
the crystal structure.
Upon attempted cyclisation of the dammarenyl-type model ent-
4ꢀ, major amounts of the monocyclic product type (40% cyclic
ether and 32% ketonic material) were isolated. Unfortunately,
only minor and very complex (possibly) tetracyclic fractions
(6% total yield) were obtained next to some small hydrocarbon
and monocyclised fractions. Here, 98% of the mass balance was
isolated.
sponding to that assumed for the enzymic OS cyclisation (cbc)
is clearly not followed, as no tetracycles are observed with a
9,10-syn relation. Based on the observed conformations of the
tetracyclic products 20 and 21a (cf. Figs. 2 and 4), we propose
a polycyclisation process via the alternative cbcꢀ conformation,
in which the preformed C-ring adopts the flipped half-chair
conformation (following a 120◦ rotation around the C-8,C-14
bond). Alternatively, the observed trans–anti–cis stereochemistry
might be reached through an all-chair conformation (i.e. ccc). But,
as previously noted for the resulting polycyclic systems (Fig. 1)
and irrespective of the C-ring conformation, this is a very unlikely
polyene fold due to the pronounced syn-diaxial relation of the
methyl substituent at C-10 and the voluminous C-14 quaternary
group, which is unlikely to adopt an axial orientation.
The prevalence of the cbcꢀ fold over the envisioned (biomimetic)
cbc fold, that would afford the natural protosteroid stereochem-
istry, might be explained by the closer proximity of the reacting
p-systems (endocyclic diene and olefins in the pendant side chains)
in the case of axially oriented polyene chains (relative to the
preformed C-ring half-chair).
As for the D-ring formation stage, the second unnatural cis-
fusion is also explained by the inverted half-chair conformation
of the preformed C-ring. Unfortunately, the observed polycyclic
products do not allow us to distinguish between formation of
an initial tetracyclic cation with an a- or b-oriented side chain
(23i or 23ii, Scheme 6). In enzymic reactions (and even in the
absence of an enzyme)16 the protosteryl cation is known to undergo
an elaborate cascade of 1,2-hydride and -methyl shifts. Here this
cascade apparently stops after the first (also naturally occurring)
17–20 hydride shift (Scheme 6). The subsequent 13–17 hydride
shift could be prevented by a conformationally rigid p-cation
interaction with the C-ring alkene (structure 24), which would
also explain the driving force for the first shift.
Our analysis of the possible underlying cyclisation mechanisms
is presented in Fig. 5.
For the models 14, ent-4 and ent-4ꢀ, that were endowed with
the mirrored dammarenyl-type stereochemistry (for clarity, the
“natural” enantiomers are shown in Fig. 5), an explanation should
be found for their apparent inability to undergo a cyclisation
beyond the A-ring formation stage. Upon evaluating the required
all-chair polyene fold (ccc), a strong steric interaction of the
1,3-syn-diaxial type is obvious between methyl substituents at
C-8 and C-10 (classical steroid numbering). Because of the sp3
hybridisation at C-8, this effect is indeed expected to be much
more pronounced than in the natural OS dammarenyl-type fold.
This steric effect could thus prevent C-9 and C-10 from coming
within binding range. Alternatively, this might be achieved if the
B-ring part of the polyene chain would adopt a (twisted) boat-like
fold. However, this boat-type conformation seems unlikely because
it requires the sterically demanding quaternary substituent (C-
14) in an unaccommodating flagpole (boat) or axial (twist-boat)
position.
For the models 16, 5 and 5ꢀ, having the protosteryl-type
stereochemistry, at least one possible polyene fold should allow
for an extended polycyclisation reaction. The conformation corre-
Scheme 6
Fig. 5
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 1918–1925 | 1923
©

View Article Online
The unusual AB-spirofused tetracycle 22 can be regarded as a
polycyclised derivative of an intermediate cation en route to the
major ketonic rearrangement product 18b. This structure again
shows the preference of our model systems for the formation of
cis-fusions to the C-ring.
˚
˚
˚
cell parameters: a = 6.429(1) A, b = 13.631(2) A, c = 50.896(6) A.
3482 reflections were collected with R(int) = 0.0381 and the final
agreement factor was R = 0.20. Full crystal data and details of
data collections and refinements of compound 21a are given in
Table SM1 as supplementary material. Complete crystallographic
information on compound 21a has been deposited by the CCDC
with deposition number CCDC 675998.†
Conclusions
The crystal morphology (very thin plates with thickness <10
lm), surely related to the long c axis, prevented us from obtaining
good crystal and good agreement factors during the refinement.
Because of these problems, the resulting crystal structure showed
some disorder and a few restraints (see cif file for details) on
the distances involving the C(CH3)2 groups were necessary to
have a stable refinement, and none of the atoms could be refined
anisotropically. Indeed the space group attribution was correct
with the expected systematic absences, as indicated by the recon-
struction of the experimental reciprocal lattice (See Fig. SM3,
4 and 5 in the supplementary information). The conformational
features of the crystal structure from X-ray data were confirmed by
first principle QM calculations. Graphical manipulations to obtain
ordered model structures from the disordered crystal structures
were carried out employing the XP21 and MOLDRAW22 software.
The observed intermolecular contacts involving the methanol
molecule were consistent with O–H · · · O hydrogen bonds and
indicated the presence of disorder, with both atoms in the solvent
moiety capable of acting as oxygen atoms. Therefore it was not
possible to establish, by geometric considerations alone, the atomic
species of the two solvent atomic positions. Nevertheless, the
values of the ADP parameters [0.088(6) and 0.18(2) for O3 and
C51 respectively] suggested the adopted assignment. It is worth
noting that this hydrogen bond network allowed the formation of
an hydrophilic layer, formed by OH groups of methanol and 21a,
and of an hydrophobic layer, constituted by the remaining part of
the 21a molecules.
Due to the complex nature of these cationic cyclisation ex-
periments, it is unwise to draw straightforward conclusions
with relevance to their biological counterparts. As expected, all
cyclisation experiments gave complex mixtures of compounds.
Furthermore, the polycyclised structures that were produced in
sufficient quantities to be isolated and characterised, all possessed
very unnatural cis-fusions to the C-ring. However, it should
be clear that (i) where extended polycyclisation occurs in our
model systems, the ABC-ring system stereochemistry is adequately
controlled by the combination of epoxide and (preformed) C-ring
chair configurations, (ii) a remarkable preference for a (distorted)
twist-boat conformation exists for the B-ring in the observed
tetracyclic lanostane-type compounds 20 and 21a, indicated by
crystal structure, 2D NMR analyses and DFT calculations, (iii)
this preference can be rationalised by severe intramolecular steric
compression in the case of a B-ring chair conformation and
similar effects are reasonably expected to control the polyene fold
during the nonenzymic polycyclisation of the examined model
compounds and (iv) this provides a model for how the elusive
B-ring (twist-)boat conformation might be achieved in steroid
biogenesis, in this case by preorganising the C-ring polyene part
in a specific chair-like fold. Such a conformational restriction has
been shown to effect biomimetic (six membered) C-ring formation
in a previous model study.7
Experimental
Synthesis and cyclisation of the model substrates
Computational details
Compounds 5, 6–12 and 15. Procedures and characterisation
as described previously in reference 11.
DFT molecular orbital calculations on the different conforma-
tions of compound 21a were performed using the Gaussian03
program23 on a 16-node linux cluster. The initial geometries where
those derived from X-ray crystal structure and were manipulated
by the MOLDRAW22 software and optimised employing the DFT
method based on Becke’s three parameters hybrid functional
and Lee–Yang–Parr’s gradient-corrected correlation functional
(B3LYP),24,25 using the 6-31 G(d,p) basis set26 as implemented in
the Gaussian03 software.
Compounds ent-4, ent-4ꢀ, 5ꢀ, 11ꢀ, 12ꢀ and 13. Synthetic pro-
cedures and spectral data for these compounds are provided as
electronic supplementary data.†
Cyclisation of compounds 16, ent-4ꢀ and 5ꢀ. Synthetic proce-
dures and characterisation data (including a detailed analysis of
2D NMR data for polycyclised compounds 20, 21a,b and 22) are
provided as electronic supplementary data.††
Acknowledgements
Crystal structure determination of compound 21a
We thank UGent for financial support. J.M.W. thanks the
F.W.O. Vlaanderen for a scholarship as aspirant.
Single crystal diffraction data on compound 21a were collected
at 100 K at the BM1a station (SNBL) of the ESRF synchrotron
facility,17 employing a monochromatised (k = 0.7114 A) radiation.
˚
Notes and References
Absorption correction was performed using SADABS.18 The
structure was solved by direct methods employing the SIR2002
software,19 and refined by full-matrix least-squares with SHELX-
97.20
1 For reviews, see: (a) K. U. Wendt, G. E. Schulz, E. J. Corey and D. R.
Liu, Angew. Chem., Int. Ed., 2000, 112, 2930; (b) I. Abe, M. Rohmer
and G. D. Prestwich, Chem. Rev., 1993, 93, 2189.
2 (a) K. B. Sharpless and E. E. van Tamelen, J. Am. Chem. Soc., 1969,
91, 1848; (b) E. E. van Tamelen, J. Willet, M. Schwartz and R. Nadeau,
J. Am. Chem. Soc., 1966, 88, 5937.
The data collection was carried out at 100 K on a crystal of
0.120 × 0.040 × 0.005 mm3. The space group was P212121 with unit
1924 | Org. Biomol. Chem., 2008, 6, 1918–1925
This journal is
The Royal Society of Chemistry 2008
©

View Article Online
3 (a) A. Eschenmoser, L. Ruzicka, O. Jeger and D. Arigoni, Helv. Chim.
Acta, 1955, 38, 1890; (b) A. Eschenmoser and D. Arigoni, Helv. Chim.
Acta, 2005, 88, 3011.
4 G. Stork and A. W. Burgstahler, J. Am. Chem. Soc., 1955, 77, 5068.
5 For reviews, see: (a) P. A. Bartlett, in Asymmetric Synthesis, ed.
J. D. Morrison, Academic Press, New York, 1984, vol. 3, p. 341; (b) R. A.
Yoder and J. N. Johnston, Chem. Rev., 2005, 105, 4730.
6 See, for example: W. S. Johnson, S. J. Telfer, S. Cheng and U. Schubert,
J. Am. Chem. Soc., 1987, 109, 2517; P. V. Fish and W. S. Johnson, J. Org.
Chem., 1994, 59, 2324.
7 (a) S.-Z. Zhou, M. Sey, P. J. De Clercq, M. Milanesio and D. Viterbo,
Angew. Chem., Int. Ed., 2000, 39, 2861; (b) K. Berckmoes, P. J. De
Clercq and D. Viterbo, J. Am. Chem. Soc., 1995, 117, 5857.
8 K. Berckmoes, PhD Thesis, Ghent University, 1996.
9 M. Pfau, I. Jabin and G. Revial, J. Chem. Soc., Perkin Trans. 1, 1993,
1935.
10 R. E. Ireland, M. I. Dawson, S. C. Welch, A. Hagenbach, J. Bordner
and B. Trus, J. Am. Chem. Soc., 1973, 95, 7829.
11 J. M. Winne, B. Guang, J. D’Herde and P. J. De Clercq, Org. Lett., 2006,
8, 4815, and references therein.
12 W. S. Johnson, Acc. Chem. Res., 1968, 1, 1.
13 (a) E. J. Corey and H. B. Wood, J. Am. Chem. Soc., 1996, 118, 11982;
(b) E. J. Corey, J. Lee and D. R. Liu, Tetrahedron Lett., 1994, 35, 9149;
(c) E. E. van Tamelen and R. J. Anderson, J. Am. Chem. Soc., 1972, 94,
8225.
14 K. B. Sharpless, J. Am. Chem. Soc., 1970, 9, 6999.
15 E. J. Corey and M. Sodeoka, Tetrahedron Lett., 1991, 48, 7005.
16 E. J. Corey, S. C. Virgil and S. Sarshar, J. Am. Chem. Soc., 1991, 113,
8171.
19 M. C. Burla, M. Camalli, B. Carrozzini, G. L. Cascarano, C. Gia-
covazzo, G. Polidori and R. Spagna, J. Appl. Crystallogr., 2003, 36,
1103.
20 G. M. Sheldrick, SHELXL-97, University of Go¨ttingen, Germany,
1997; web site: http://shelx.uni-ac.gwdg.de/SHELX/.
ꢁ
C
21 SHELXTL -
1998. 2003. Madison Wisconsin 53719 USA, Bruker
AXS, Inc. 1998.
22 P. Ugliengo, D. Viterbo and G. Z. Chiari, Kristallografiya, 1993, 207,
9; web site: http://www.moldraw.unito.it.
23 M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A.
Robb, J. R. Cheeseman, J. A. Montgomery, Jr., T. Vreven, K. N.
Kudin, J. C. Burant, J. M. Millam, S. S. Iyengar, J. Tomasi, V. Barone,
B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G. A. Petersson, H.
Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa,
M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X.
Li, J. E. Knox, H. P. Hratchian, J. B. Cross, C. Adamo, J. Jaramillo,
R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi,
C. Pomelli, J. W. Ochterski, P. Y. Ayala, K. Morokuma, G. A. Voth,
P. Salvador, J. J. Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D.
Daniels, M. C. Strain, O. Farkas, D. K. Malick, A. D. Rabuck, K.
Raghavachari, J. B. Foresman, J. V. Ortiz, Q. Cui, A. G. Baboul, S.
Clifford, J. Cioslowski, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz,
I. Komaromi, 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, C. Gonzalez and J. A. Pople, Gaussian 03, Revision
B.05, Gaussian, Inc., Pittsburgh PA, 2003.
24 A. D. Becke, J. Chem. Phys., 1993, 98, 5648–5652.
25 C. Lee, W. Yang and R. G. Parr, Phys. Rev. B: Condens. Matter Mater.
Phys., 1988, 37, 785–789.
26 W. J. Hehre, L. Radom, P. v. R. Schleyer and J. A. Pople, Ab initio
Molecular Orbital Theory, Wiley, New York, 1986.
17 Web site: http://www.esrf.fr.
18 G. M. Sheldrick, SADABS, University of Go¨ttingen, Germany, 1996.
This journal is
The Royal Society of Chemistry 2008
Org. Biomol. Chem., 2008, 6, 1918–1925 | 1925
©