A biogenetically inspired heterodimerization approach to the synthesis of the core structure of the alkaloid fissoldhimine

Article abstract of DOI:10.1016/j.tetlet.2007.01.019

A biogenetically inspired heterodimerization reaction of N-substituted 2-pyrroline equivalents leads to the tricyclic core of the alkaloid fissoldhimine. Thus, pyrrolidin-2-ol derivatives, in which the nitrogen atom is substituted either with urea or thio

Full text of DOI:10.1016/j.tetlet.2007.01.019

Tetrahedron Letters 48 (2007) 1841–1844
A biogenetically inspired heterodimerization approach to the
synthesis of the core structure of the alkaloid fissoldhimine
Heather Twin, Wendy W.-H. Wen, David A. Powell, Alan J. Lough and Robert A. Batey*
Department of Chemistry, University of Toronto, 80, St. George St., Toronto, Ont., Canada M5S 3H6
Received 1 November 2006; revised 21 December 2006; accepted 3 January 2007
Available online 7 January 2007
Abstract—A biogenetically inspired heterodimerization reaction of N-substituted 2-pyrroline equivalents leads to the tricyclic core
of the alkaloid fissoldhimine. Thus, pyrrolidin-2-ol derivatives, in which the nitrogen atom is substituted either with urea or thiourea
functionality, serve as equivalents of the corresponding N-substituted 2-pyrrolines. Reaction of these compounds under Lewis acidic
(e.g., lanthanide triflate) or Brønsted acid conditions leads to a diastereomeric form of the tricyclic core of fissoldhimine. The reac-
tion can be envisaged to occur either via an asynchronous intermolecular inverse electron demand hetero-Diels–Alder reaction, or
through a tandem Mannich/ring-closure reaction.
Ó 2007 Elsevier Ltd. All rights reserved.
Fissoldhimine 1 was first isolated in 1994 by Wu et al.
from the perennial shrub Fissistigma oldhamii (Hemsl.)
Merr. (Annonaceae) (Fig. 1).1 This shrub is distributed
mainly in Southern China and Taiwan, and has been
used as a folk medicine to treat conditions such as sciat-
ica and arthritis. It is also known for its anti-inflamma-
tory and anti-tumor properties.2 The unique four-ring
fused structure of 1 was confirmed by X-ray analysis.
To date there have not been any reported approaches
to the synthesis of this intriguing alkaloid. We now
report our preliminary observations on the synthesis of
the core structure, inspired by a plausible hypothesis
of the biogenesis of fissoldhimine.
with butanal. It is also quite likely that fissoldhimine is
an artifact of the isolation process, and that instead it
is derived from tricyclic urea 3. Thus, 1 may be derived
from an aminoacetalization of 3 with butanal, present in
trace amounts with the butanol used in the extraction
process.1 In turn tricyclic compound 3 would be formed
by a heterodimerization of 2 or a synthetic equivalent.
The heterodimerization involves C–C bond formation
between C1 of 2 and C2 of a second equivalent of 2,
to form the bond C1–C20 in 1 (Fig. 1). Wu has proposed
a more detailed biogenetic hypothesis for fissoldhimine
that involves the coupling of the tautomeric isomers
1-pyrroline 4 and 2-pyrroline 5 (Fig. 2). Heterodimeriza-
tion of 4 with 5 via a Mannich reaction then gives
6. Sequential N-carbamoyl transfer from carbamoyl
phosphate can then lead to 3, a putative precursor to
Fissoldhimine can be considered to be formally derived
from a 2:1 or AA0B coupling3 of 2-pyrroline derivative 2
H PO CONH
2
2
4
H H
NH
O
O
H H
O
NH
2
4
H
H
N
N
O
N
N
O
N
1'
N
N
N
N
3
N
N
H
N
H
N
1
4
3
1
2
H
2'
3'
H N
2
HN
2
6
2
4
5
H
H
O
H
H
4'
1
2
3
H PO CONH
2
2
4
Figure 1. The alkaloid fissoldhimine 1 is formally derived from the
heterodimerization of 2-pyrroline derivative 2. Note: the numbering on
1 relates to its putative biogenesis from 2.
H
H
PrCHO
3
1
N
N
O
N
H
H
*
Corresponding author. Tel./fax: +1 416 978 5059; e-mail: rbatey@
chem.utoronto.ca
Figure 2. Biosynthetic hypothesis for fissoldhimine formation.
0040-4039/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetlet.2007.01.019

1842
H. Twin et al. / Tetrahedron Letters 48 (2007) 1841–1844
TfO–
fissoldhimine. Related heterodimerizations of the N-
methylpyrollinium ion have been reported to give the
alkaloid 1,10-dimethyl-[2,30]bipyrrolidinyl.4 Similar di-
meric products have also been reported to be formed
as by-products under other conditions, such as, for
example, during the reduction of N-methyl-2-pyrrolidi-
none,5 or from the mercuric acetate catalyzed oxidation
of various substituted N-methylpyrrolidines.6 Analo-
gous dimeric alkaloids, such as ammonodendrine and
orensine are also known to be formed through hetero-
dimerization of D1-piperideine.7
Bn
S
H
N
Bn
N
S
N
(i) Dy(OTf)3 (7 mol%)
MeCN, 4 ˚C, 3 d
H
N
S
H
OH
(ii) TfOH, r.t., 16 h
68%
N
H
Bn
N
H
H
10
Scheme 1.
11
An alternative pathway to the biogenesis of 1/3 could
involve an earlier stage N-carbamoyl transfer from 5
to give 2. Urea 2 can also be formed through oxidation
and cyclization of N-carbamoylputrescine. Protonation
of 2 can then give iminium ion 7 (Fig. 3). Heterodimer-
ization of 2 with 7 can be envisaged to occur either in a
Mannich type process, via intermediate 8 to give 3.
Alternatively, an inverse electron demand hetero
Diels–Alder reaction of 2 with 7 can give 9, which on
rearrangement leads to adduct 3.
Our initial aim was to determine whether such a hetero-
dimerization mechanism of 2 or some analog could be
achieved chemically. Initial attempts used the thiourea
precursor 10 which we had previously used in our stud-
ies on the synthesis of martinelline analogs.8 We have
demonstrated that under mildly Lewis acidic conditions,
pyrrolidinols (such as 10) serve as N-substituted pyrro-
line equivalents (i.e., a thiourea analog of 2). An expo-
sure of thiourea 10 to a lanthanide triflate catalyst
afforded the dimerized product 11 as a single diastereo-
isomer in a 68% yield (Scheme 1). The stereochemistry
of 11 was established by an X-ray crystallographic study
(Fig. 4).9 Compound 11 can be formed via either step-
wise or concerted cycloaddition mechanisms analogous
to those outlined for the conversion of 2 into 3
(Fig. 3). Thus, Lewis acid promoted ionization of 10
would give a thiourea analog of 7, which could then lose
a proton to give the corresponding thiourea protected
Figure 4. Solid-state structure of 11 as determined by X-ray crystallo-
graphic analysis. Displacement ellipsoids are drawn at the 30%
probability level and H atoms are shown as spheres of arbitrary radii.
2-pyrroline (i.e., the thiourea analog of 2). The stereo-
chemistry of 11 (all-cis) would arise from an endo attack
of the dienophile on the diene, assuming a hetero Diels–
Alder mechanism. Compound 11 differs in two notable
respects to compound 3. Firstly, the relative stereochem-
istry is not that found in the natural product 1 or com-
pound 3. Secondly, the central ring of 11 is formed via
attack of the sulfur atom, rather than the nitrogen atom
of the thiourea, onto the electrophilic C-10 atom. Con-
versely, the central ring of 1 and 3 are formed via attack
of the nitrogen atom, rather than the oxygen atom of the
urea. This difference provides some support for a hetero
Diels–Alder reaction occurring for the formation of 11,
analogous to that shown in the direct [4+2] cycloaddi-
tion of 2 and 7 to give 9. It may however, reflect the rel-
ative thermodynamic stability of 11 over the isomeric
compound that would be formed through C–N ring
closure of the central ring.
H
NH2
NH
[4+2]
Cycloaddition
O
N
O
O
O
N
H
N
– H
H2N
H2N
N
H
7
H
2
9
While the stereochemistry of 11 was not that required
for fissoldhimine, we were encouraged that heterodimer-
ization had occurred. We next set out to establish
whether a similar approach could be used in a dimeriza-
tion of urea 2 or an equivalent. Our initial approach to 2
utilized pyrrolidinone as the starting material (Scheme
2). The reaction of pyrrolidinone with chloroacetyl
isocyanate formed imide 12. Deprotection of the chloro-
acetyl group10 was achieved by treatment with triethyl-
amine in methanol to give 13. Finally, the reduction of
13 using DIBAL–H followed by the addition of trimeth-
ylorthoformate in methanol gave the urea 14 in a 62%
yield over the three steps. Unfortunately, all attempts
cyclization
– H
Mannich
O
NH2
H
H
N
O
N
N
O
O
N
cyclization
– H
H2N
H2N
N
H
H
8
H
H
3
Figure 3. Alternative biosynthetic hypothesis for fissoldhimine
formation.

H. Twin et al. / Tetrahedron Letters 48 (2007) 1841–1844
1843
O
H
O
O
N
NH
2
H
N
Cl
(a)
(b)
N
N
O
O
O
12
13
O
O
NH
NH
2
2
(c)
N
N
OMe
14
2
Scheme 2. Reagents and conditions: (a) chloroacetylisocyanate
(1 equiv), benzene; (b) Et3N (2 equiv), MeOH; and (c) DIBAL
(1.5 equiv), CH2Cl2, ꢀ78 °C, then MeOH, methyl orthoformate, PPTS
(cat.), rt.
to convert, either 14 or the initial reduction product
from 13, unambiguously, into 2 or heterodimerized
adducts such as 3 or 9 were unsuccessful. An adduct
formed from the reaction of 14 with Dy(OTf)3 did show
a molecular ion, M+ = 224, consistent with dimerized
Figure 5. Solid-state structure of 16a-endo (R = Bn) as determined by
X-ray crystallographic analysis. Displacement ellipsoids are drawn at
the 30% probability level and H atoms are shown as spheres of
arbitrary radii.
1
adducts, but the severely broadened signals in the H
and 13C precluded a definitive structural assignment.
Attempts to deprotect the benzyl group of 16a-endo,
under a variety of conditions, including hydrogenation
with Pearlman’s catalyst using a Parr hydrogenation
apparatus, were unsuccessful. Consequently, we
switched to the PMB (para-methoxybenzyl) protecting
group, since it is generally more easily removed. The
same two step strategy was used to form 15b
(R = PMB) in an 83% yield. Dimerization of 15b
(R = PMB) in the presence of 10 mol % of Dy(OTf)3,
gave 16b-endo (R = PMB) in an 84% yield (Scheme 3).
Once again the undesired endo diastereomer was
obtained, as revealed by a comparison of the 1H
NMR spectrum with that of the N-benzyl analog.
The use of a protecting group strategy could conceivably
overcome these problems, as we had shown in the reac-
tion of the analogous N-benzyl protected thiourea 11.
An N-benzyl protected urea 15a (R = Bn) was chosen,
due to the availability of the starting isocyanate, and
the similarity with 10. Hydroxyurea 15a was obtained
in two steps, in a 74% overall yield, from pyrrolidinone
via reaction with benzylisocyanate followed by reduc-
tion using DIBAL–H (Scheme 3).10 The reaction of
15a with 10 mol % of Dy(OTf)3 in acetonitrile at room
temperature led to the formation of 16a-endo (R = Bn)
in a 73% yield. The X-ray crystallographic analysis of
16a-endo revealed that the desired tricyclic core had
been formed; however, as observed with adduct 11, the
product was obtained as the endo diastereoisomer
(Fig. 5). Dimerization to 16a-endo occurs through C–C
and C–N bond formation, to give the same structural
core observed with fissoldhimine. This contrasts with
the formation of 11, in which the dimerization occurred
through C–C and C–S bond formation.
Our previous studies on the synthesis of martinelline,
which is also formed from a 2-pyrroline, had revealed
that subtle changes in reaction conditions, particularly
the choice of Lewis or protic acid and the solvent em-
ployed, can lead to substantial changes in diastereoselec-
tivity.11 Screening the reaction of 15b under a variety of
conditions led to the product as a mixture of the unde-
sired endo and desired exo diastereomers, 16b-endo and
16b-exo (Table 1). In every case 16b-endo was the pre-
dominant product. The best conditions found for the
formation of the desired diastereomer 16b-exo were
the use of trifluoroacetic anhydride (1.1 equiv) in
THF, which gave a 2:1 (endo:exo) ratio as judged by
R
O
N
H
H
(b)
(a)
N
N
O
R
O
1
crude H NMR analysis. However, we were unable to
establish chromatographic conditions to separate the
two diastereomers.
O
O
N
R
N
N
H
(c)
N
O
H
We therefore elected to investigate further the feasibility
of converting 16b-endo, which can be stereoselectively
formed, into epi-fissoldhimine. Optimal conditions for
the formation of 16b-endo, employed the reaction of
15b using 1.1 equiv of trifluoroacetic anhydride in tolu-
ene, to give the product in an 88% yield as the TFA salt,
with an endo:exo selectivity of >95:5 (endo:exo). Several
methods were attempted to remove the PMB protecting
OH
N
H
R
N
H
H
15a (R = Bn)
15b (R = PMB)
16a-endo (R = Bn)
16b-endo (R = PMB)
Scheme 3. Reagents and conditions: (a) RN@C@O (1.1 equiv),
toluene, reflux; (b) DIBAL–H (2 equiv), CH2Cl2, ꢀ78 °C; and (c)
Dy(OTf)3 (10 mol %), MeCN.

1844
H. Twin et al. / Tetrahedron Letters 48 (2007) 1841–1844
Table 1.
endo:exo ratio of the diasteroemeric tricyclic adducts
depended upon the reaction conditions. However, in
all cases the reactions led preferentially to the formation
of the adducts having an endo relative stereochemistry.
This contrasts with the relative stereochemistry of the
natural product fissoldhimine which has an exo stereo-
chemistry for the corresponding tricyclic core. Neverthe-
less, these results provide a laboratory analog to the key
step of the proposed biogenesis of fissoldhimine. Further
studies on synthetic approaches to this structurally
unique alkaloid will be reported in due course.
PMB
O
N
O
O
N
PMB
PMB
(c)
H
N
N
N
N
O
O
H
H
OH
N
N
H
H
PMB
PMB
N
H
N
H
H
H
15b
16b-exo
16b-endo
Reagents
CSA (5 mol %), THF
HCl (10% soln.), MeOH
Sc(OTf)3 (10 mol %), THF
BF3ÆOEt2, THF
TFAA (1.1 equiv), MeCN
TFAA (1.1 equiv), THF
TFAA (1.1 equiv), toluene
Diastereoselectivity
of 16b (endo:exo)a
90:10
90:10
85:15
>95:5
>95:5
65:35
>95:5
Acknowledgements
This work was supported by the Natural Science and
Engineering Research Council (NSERC) of Canada.
We thank Dr. A. B. Young for mass spectroscopic
analyses.
a Ratio determined by the analysis of the crude 1H NMR spectra.
O
O
References and notes
PMB
PMB
DDQ (5 equiv.)
N
N
N
N
O
O
H
1. Wu, J.-B.; Cheng, Y.-D.; Kuo, S.-C.; Wu, T.-S.; Iitaka,
Y.; Ebizuka, Y.; Sankowa, U. Chem. Pharm. Bull. 1994,
42, 2202–2204.
H
CH Cl :H O (20:1)
2
2
2
N
N
H
H
PMB
N
H N
2
H
H
72%
H
2. Su, J. Encyclopedia Chinese Materia Medica; New Medical
College, Shanghai Science and Technology, 1977; p 758.
3. Powell, D. A.; Batey, R. A. Tetrahedron Lett. 2003, 44,
7569–7573.
17
16b-endo
Scheme 4.
4. Wei, X.; Sumithran, S. P.; Deaciuc, A. G.; Burton, Harold
R.; Bush, L. P.; Dwoskin, L. P.; Crooks, P. A. Life Sci.
2005, 78, 495–505.
5. Swan, G. A.; Wilcock, J. D. J. Chem. Soc., Perkin Trans 1
1974, 885–891.
6. Leonard, N. J.; Cook, A. G. J. Am. Chem. Soc. 1959, 81,
5627–5631.
7. See, for example: Schoepf, C.; Braun, F.; Kreibich, K. Ann
1964, 674, 87–92.
groups of 16b-endo. The use of Pd/C under H2 or AlCl3
in anisole, TFA, lithium powder in naphthalene or p-
toluenesulfonic acid resulted in either no reaction or
decomposition. However, the use of DDQ resulted in
removal of just one of the PMB groups to give 17
(Scheme 4). Unfortunately, further attempts to depro-
tect 17 under a variety of conditions resulted in the
decomposition.
8. Batey, R. A.; Powell, D. A. Chem. Commun. 2001, 2362–
2363.
9. Crystallographic data (excluding structure factors) for the
structures in this Letter have been deposited with the
Cambridge Crystallographic Data Centre as supplemen-
tary publication numbers CCDC 626182 and 626183 for
compounds 11 and 16a-endo.
10. Katagiri, N.; Muto, M.; Nomura, M.; Higashikawa, T.;
Kaneko, C. Chem. Pharm. Bull. 1991, 39, 1112–1122.
11. Powell, D. A.; Batey, R. A. Org. Lett. 2002, 4, 2913–
2916.
In conclusion, fissoldhimine is an example of a hetero-
dimerized alkaloid. A plausible biogenetic route to
fissoldhimine involves the heterodimerization of pyrro-
line derivatives, via either a stepwise Mannich type path-
way, or an intermolecular aza-Diels–Alder reaction. The
reaction of both urea and thiourea pyrrolidinol deriva-
tives using Lewis acidic catalysis results in the formation
of tricyclic adducts analogous to fissoldhimine. The