Efficient and stereoselective dimerization of pyrroloindolizine derivatives inspired by a hypothesis for the biosynthesis of complex myrmicarin alkaloids

Article abstract of DOI:10.1021/jo701981q

(Chemical Equation Presented) Pyrroloindolizine derivatives participate in efficient and stereoselective homo- and heterodimerization reactions upon treatment with Bronsted or Lewis acids. The distinctive ability of pyrroloindolizines to act as azafulvenium ion precursors provides direct access to both heptacyclic and hexacyclic dimeric products. The inherent reactivity of these structures suggests a concise synthesis of complex myrmicarin alkaloids, via dimerization of pyrroloindolizines, and may have implications for the biosynthesis of these intriguing alkaloids.

Full text of DOI:10.1021/jo701981q

Efficient and Stereoselective Dimerization of Pyrroloindolizine
Derivatives Inspired by a Hypothesis for the Biosynthesis of
Complex Myrmicarin Alkaloids
Mohammad Movassaghi,* Alison E. Ondrus, and Bin Chen
Department of Chemistry, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139
moVassag@mit.edu
ReceiVed September 9, 2007
Pyrroloindolizine derivatives participate in efficient and stereoselective homo- and heterodimerization
reactions upon treatment with Brønsted or Lewis acids. The distinctive ability of pyrroloindolizines to
act as azafulvenium ion precursors provides direct access to both heptacyclic and hexacyclic dimeric
products. The inherent reactivity of these structures suggests a concise synthesis of complex myrmicarin
alkaloids, via dimerization of pyrroloindolizines, and may have implications for the biosynthesis of these
intriguing alkaloids.
Introduction
The complex myrmicarins are a family of structurally
fascinating and air-sensitive alkaloids isolated from the poison
gland of the African ant species Myrmicaria opaciVentris (Figure
1).1 Detailed spectroscopic studies have been used to assign the
relative stereochemistries of the complex myrmicarins 430A
(4)1c and 663 (5).1d While myrmicarin 663 (5) was isolated and
fully characterized, the extreme sensitivity of myrmicarin 430A
(4) required its structural assignment as a crude isolation mixture
using phase-sensitive 2D NMR techniques. Likewise, the
fragility and limited quantities of isolated myrmicarin 645 (6)
precluded its relative stereochemical assignment. An isomeric
myrmicarin 430B was also identified,1c but no structural
information on this compound has been reported.
The compelling architectures of these toxic alkaloids and the
challenges associated with their sensitivity inspired us to initiate
a program directed at their study and total synthesis.2 Our
(1) (a) Francke, W.; Schro¨der, F.; Walter, F.; Sinnwell, V.; Baumann,H.;
Kaib, M. Liebigs Ann. Chem. 1995, 965-977. (b) Schro¨der, F.; Franke, S.;
Francke, W. Tetrahedron 1996, 52, 13539-13546. (c) Schro¨der, F.;
Sinnwell, V.; Baumann, H.; Kaib, M. Chem. Commun. 1996, 2139-2140.
(d) Schro¨der, F.; Sinnwell, V.; Baumann, H.; Kaib, M.; Francke, W. Angew.
Chem., Int. Ed. Engl. 1997, 36, 77-80. (e) Schro¨der, F.; Francke, W.
Tetrahedron 1998, 54, 5259-5264.
FIGURE 1. Members of the myrmicarin alkaloids.
synthetic approach is based on our proposal that these complex
molecules could be accessed from activated pyrroloindolizine
derivatives (Scheme 1) through a potentially biomimetic dimer-
ization event.2a We have considered three possible pathways
10.1021/jo701981q CCC: $37.00 © 2007 American Chemical Society
Published on Web 11/17/2007
J. Org. Chem. 2007, 72, 10065-10074
10065

Movassaghi et al.
SCHEME 1. Our Proposed Biomimetic Dimerization of Pyrroloindolizines for the Synthesis of Myrmicarin 430A (4)
for the dimerization of myrmicarin 215 (2) to provide the
heptacyclic structure of myrmicarin 430A (4). In the first
scenario, we envisioned that protonation of (+)-myrmicarin
215B (2) at C9 could initiate a sequence of bond-forming events
mediated by highly electrophilic azafulvenium ion intermediates
(i.e., 8, Scheme 1A).2a In this sequence, cyclopentannulation
would occur through intermolecular trapping of the tricyclic
azafulvenium ion 8 by (+)-myrmicarin 215B (2), followed by
an intramolecular Friedel-Crafts trapping at C8b of intermediate
9 (Scheme 1A). Alternatively, the five contiguous stereocenters
in myrmicarin 430A (4) may be generated in a concerted
cycloaddition process involving neutral (+)-myrmicarin 215B
(2) and the tricyclic azafulvenium ion 8 (Scheme 1B). The
reactive intermediate 8 can be generated either by C9-protona-
tion of 2 or by expulsion of a leaving group at C8 of
pyrroloindolizine derivative 7. Consistent with FMO analysis,
a [6πa + 2πs] cycloaddition3 would afford the observed
stereochemistry at each of the stereocenters of myrmicarin 430A
(4). Another possible dimerization manifold involves activation
of (+)-myrmicarin 215B (2) or a tricyclic derivative 7 as the
stabilized radical intermediate 12, which could undergo inter-
molecular radical addition to another (+)-myrmicarin 215B (2)
to give 13 (Scheme 1C). Subsequent 5-exo-trig radical cycliza-
tion of 13 (C1-C8b bond formation) would provide the
heptacycle 14, which would lead to formation of myrmicarin
430A (4) via oxidation and tautomerization reactions. Herein
we report our studies on the reactivity of functional pyrroloin-
dolizines pertinent to our hypothesis regarding the biosynthesis
of these alkaloids. Additionally, relying on the unique reactivity
of pyrroloindolizines, we discuss the development of a conver-
gent strategy for the assembly of functional intermediates for
the synthesis of complex myrmicarins.
While our initial biosynthetic hypotheses (Scheme 1A)
suggested a succinct means of generating these complex
alkaloids, the proposed cyclopentannulation reactions required
an unprecedented mode of reactivity for vinyl pyrroles.4 Earlier
we reported the direct and highly diastereoselective homodimer-
ization of (+)-myrmicarin 215B (2), leading to generation of a
single heptacyclic product and introduction of four contiguous
stereocenters.2a In situ 1H NMR monitoring revealed that
(2) (a) Ondrus, A. E.; Movassaghi, M. Tetrahedron 2006, 62, 5287-
5297. (b) Movassaghi, M.; Ondrus, A. E. Org. Lett. 2005, 7, 4423-4426.
(3) For FMO analysis of cycloadditions involving fulvenes, see: (a)
Houk, K. N.; George, J. K.; Duke, R. E., Jr. Tetrahedron 1974, 30, 523-
533. (b) Houk, K. N. Acc. Chem. Res. 1975, 8, 361-369. For computational
analysis of fulvene FMOs, see: (c) Scott, A. P.; Agranat, I.; Biedermann,
U. P.; Riggs, N. V.; Radom, L. J. Org. Chem. 1997, 62, 2026-2038. (d)
Havenith, R. W. A.; Fowler, P. W.; Steiner, E. J. Chem. Soc., Perkin Trans.
2 2002, 2, 502-507. For formal [6 + 2] cycloadditions involving fulvene
species, see: (e) Hong, B.-C.; Shr, Y.-J.; Wu, J.-L.; Gupta, A. K.; Lin,
K.-J. Org. Lett. 2002, 4, 2249-2252. (f) Hafner, K.; Suda, M. Angew.
Chem., Int. Ed. Engl. 1976, 15, 314-315.
(4) For a review on the chemistry of C-vinyl pyrrole derivatives, see:
(a) Trofimov, B. A.; Sobenina, L. N.; Demenev, A. P.; Mikhaleva, A. I.
Chem. ReV. 2004, 104, 2481-2506. For general reviews on the chemistry
of pyrroles, see: (b) Pelkey, E. T. In FiVe-Membered Ring Systems:
Pyrroles and Benzo DeriVatiVes; Gribble, G., Joule, J., Eds.; Progress in
Heterocyclic Chemistry, Vol. 17; Elsevier Ltd.: Oxford, UK, 2005; pp 109-
141. (c) Jones, A., Ed. Pyrroles; Wiley: New York, 1990. (d) Alan, J. R.;
Bean, G. P. The Chemistry of Pyrroles; Academic Press: London 1977.
(e) Baltazzi, E.; Krimen, L. I. Chem. Rev. 1963, 63, 511-556.
10066 J. Org. Chem., Vol. 72, No. 26, 2007

Dimerization of Pyrroloindolizine DeriVatiVes
SCHEME 2. Our Synthesis of Isomyrmicarin 430A (17) by
SCHEME 3. Bond Formation Leading to Myrmicarin 430A
(4, Path A) and Isomyrmicarin 430A (17, Path B)
Direct Dimerization of (+)-Myrmicarin 215B (2)2a
TABLE 1. Treatment of (+)-Myrmicarin 215B (2) and Alcohol 15
with Brønsted Acids
Brønsted acid (trifluoracetic acid, TFA) treatment of (+)-
myrmicarin 215B (2) in benzene-d6 (0.050 M) resulted in
complete conversion to a single new dimeric compound, namely
the heptacyclic iminium ion 16 (Scheme 2). Deprotonation of
16 gave the air-sensitive isomyrmicarin 430A (17). Although
the direct dimerization product 16 was unstable to isolation,
chemical modification of this compound provided stable deriva-
tives that were amenable to full structural characterization,
enabling rigorous assignment of their connectivity and stereo-
chemistry through a combination of high-field 2D NMR
techniques (Scheme 2).2a Our analysis revealed that the obtained
dimer was regioisomeric at one bond and epimeric at one
stereocenter to myrmicarin 430A (4). Significantly, this structure
validated our mechanistic hypothesis that acid-promoted activa-
tion of (+)-myrmicarin 215B (2) can provide C8 electrophilic
derivatives susceptible to a highly diastereoselective attack by
the vinyl group of a second equivalent of (+)-myrmicarin 215B
(2).
According to this stepwise mechanism, the regiochemical
difference between the myrmicarin 430A (4) and isomyrmicarin
430A (17) structures arises from the proposed Friedel-Crafts
trapping of the hexacyclic iminium ion 19 (Scheme 3), where
the observed heptacycle 20 possesses a C1-C3b bond (Path
B) instead of the C1-C8b bond (Path A) found in the
framework of the natural alkaloid myrmicarin 430A (4).
Additionally, the isomyrmicarin 430A stereochemistry at C3 is
opposite to that found in myrmicarin 430A (4).
entry substrate solvent
acid (equiv)
additive
none
none
none
none
none
LiCl
(sat.)
LiCl
(sat.)
LiClO4
(sat.)
LiClO4
(sat.)
LiCIO4
(sat.)
product(s)a
1
2
3
4
5
6
2
2
15
15
15
15
C6D12
C6D12
THF-d8 TFA (1.10)
CD3CN TFA (1.10)
CD3OD TFA (1.10)
THF-d8 TFA (1.10)
TFA (<1.00)
TFA (>1.00)
20
24
20
20
21,20
2,20
7
8
15
2
THF-d8 AcOH (0.20)
THF-d8 TFA (1.10)
THF-d8 HCIO4 (1.10)
CD3OD HCIO4 (1.10)
CD3OD HCIO4 (<1.10)
CD3OD TFA (1.10)
2
24
9
15
15
15
15
15
24
10
11
12
13
21b
21,20
LiCIO4
(1.6 M)
p-MePhSH
(0.60 equiv)
Results and Discussion
22 + 20
(80:20)
23,24
Brønsted Acid-Promoted Dimerization Conditions. The
homodimerization of (+)-myrmicarin 215B (2) provided the
basis for a potentially direct formation of myrmicarin 430A (4).
The nature of the putative azafulvenium ion intermediates in
the dimerization suggested that the rate or reversibility of each
bond-forming event may be influenced by solvent and counte-
rion effects. In order to address the regio- and stereochemical
differences between the heptacycle we had obtained (20, Scheme
3) and the desired structure (10, Scheme 3), we examined the
DCO2D DCO2D (solvent) none
a Reactions monitored in situ by 1H NMR. b Pyrrole-ring protonated
forms.
acid promoted dimerization of (+)-myrmicarin 215B (2) under
different reaction conditions (Table 1). Due to the known
instability of myrmicarin 430A (4) and the observed sensitivity
of heptacyclic derivatives such as 16, these experiments were
J. Org. Chem, Vol. 72, No. 26, 2007 10067

Movassaghi et al.
monitored by in situ 1H NMR and the structure of air-sensitive
final products confirmed by conversion to isolable derivatives.
Unless strictly anhydrous conditions were required, the alcohol
15, which also undergoes trifluoroacetic acid (TFA)-induced
dimerization2a to isomyrmicarin 430A (17) via (+)-myrmicarin
215B (2), was employed in place of 2 due to its enhanced
stability.
of ring-protonated forms of the C8 methanol adduct 21, which
did not undergo dimerization (Table 1, entry 10). By contrast
in an unsaturated methanolic solution of LiClO4 (1.6 M),
portionwise addition of HClO4 (0.10 equiv/portion) to a solution
of 15 (0.026 M) effected complete formation of 21, followed
by approximately proportional conversion to 20 (Table 1, entry
11).
In an effort to more efficiently trap possible electrophilic
monomeric or dimeric intermediates present in equilibrium with
the iminium ion 20, p-methylbenzenethiol (0.60 equiv) was
introduced to a methanol-d4 solution of 15 (0.020 M) prior to
portionwise substoichiometric addition of TFA. In this case,
introduction of TFA resulted in immediate conversion to the
C8-methanol adduct 21 followed by progressive formation of
the tricyclic C8-sulfide 22, completely consuming the thiol
additive (Table 1, entry 12). Concomitant dimerization of the
remaining C8-methanol adduct to 20 consumed the remainder
of the tricyclic substrate 15. Thiol adduct 22 was not subject to
dimerization and no other adducts were observed during the
reaction.5 An attempt to use a weaker organic acid as a proton
source and solvent by dissolving 15 in neat formic acid-d2
resulted in immediate conversion to the corresponding epimeric
formate esters 23, followed by slow conversion to a mixture of
ring-protonated tautomers 24, with approximately 5% (+)-
myrmicarin 215B (2) visible throughout (Table 1, entry 13).
In none of these experiments did we visualize any dimeric
species other than 20 (X- ) F3CCO2- or ClO4-, Table 1). While
the use of strongly acidic conditions favoring ring protonation
of (+)-myrmicarin 215B (2) prevented dimerization (Table 1,
entries 7, 8, and 12), conditions that would enable neutral (+)-
myrmicarin 215B (2) and an azafulvenium ion to coexist (Table
1, entries 2, 3, 4, 5, 6, 11, and 12) resulted in completely
selective and quantitative formation of 20. Tricyclic C8-adducts
were generated as equilibrating compounds in the presence of
nucleophilic solvent (Table 1, entries 5 and 11) or irreversibly
in the presence of stronger nucleophiles (Table 1, entry 12);
however, no nucleophilic adducts of the dimeric products were
observed. The substantially reduced rate of dimerization in the
presence of nucleophilic counterion under saturating conditions
(Table 1, entry 6) suggests possible reversible interception of
azafulvenium ion intermediates at either of the carbon-carbon
bond-forming steps (Scheme 1A). These observations col-
lectively suggest that if a dimeric azafulvenium ion existed (i.e.,
19, Scheme 3) it was a fleeting intermediate.
Possible [6πa + 2πs] Cycloaddition Pathway. The insensi-
tivity of the dimerization reaction described above to variations
in the reaction conditions prompted us to consider that the high
diastereoselectivity and efficiency of the process might be a
consequence of a concerted cycloaddition event instead of a
stepwise ionic sequence (Scheme 4). Frontier molecular orbital
(FMO) analysis of the proposed event revealed that a thermally
allowed [6πa + 2πs] cycloaddition between the E-azafulvenium
ion 25 and the E-alkene of (+)-myrmicarin 215B (2) could
provide the isomyrmicarin 430A (17) structure with the correct
stereochemistry in the cyclopentane ring (Scheme 4B).3 In this
sequence, association between the electron-poor azafulvenium
ion 25 and the electron-rich (+)-myrmicarin 215B (2) would
bring their convex faces together, whereupon a gearing effect
would move the vinyl group of (+)-myrmicarin 215B (2) to
the more sterically accessible side of azafulvenium ion 25.
The stoichiometric TFA-induced dimerization behavior of
(+)-myrmicarin 215B (2) was examined in methanol-d4, THF-
d8, acetonitrile-d3, and cyclohexane-d12. In cyclohexane-d12 the
TFA-promoted reactivity of (+)-myrmicarin 215B (2, 0.0050
M) was similar to that we had observed in benzene-d6.2a Namely,
introduction of substoichiometric TFA-promoted formation of
the dimer 20 (X- ) F3CCO2-) to approximately the same extent
as the added acid (Table 1, entry 1), while rapid addition of
superstoichiometric TFA yielded only pyrrole ring-protonated
products (Table 1, entry 2). Treatment of a THF-d8 solution of
alcohol 15 (0.026M) with trifluoroacetic acid (1.10 equiv)
resulted in immediate and quantitative conversion to (+)-
myrmicarin 215B (2), which was subsequently consumed to
form the dimer 20 (Table 1, entry 3). In this case, the
dimerization was substantially slower than the dimerization in
benzene-d6 and only reached 80% conversion to 20 within 24
h. In contrast, introduction of 1.10 equiv of TFA to a solution
of 15 in acetonitrile-d3 (0.025 M) effected complete conversion
to 20 within 30 s without a visible accumulation of (+)-
myrmicarin 215B (2, Table 1, entry 4). In methanol-d4, treatment
of a solution of 15 (0.026 M) with TFA (1.10 equiv) did not
afford (+)-myrmicarin 215B (2) but resulted in complete
formation of the C8 methanol adduct 21 within 30 s, which
underwent 95% conversion to 20 within 45 min (Table 1, entry
5) without a visible accumulation of (+)-myrmicarin 215B (2).
Significantly, no methanol adduct of dimeric products corre-
sponding to trapping of putative intermediates (i.e., 9, Scheme
1A) or a methanol adduct of 20 were observed.
To investigate the influence of the ionic strength of the
medium and the nature of the counterion on the acid-induced
chemistry of 2 we examined the effect of salt additives. While
addition of TFA (1.10 equiv) to a solution of 15 in THF-d8
(0.030 M) saturated with anhydrous lithium chloride (LiCl)
effected immediate conversion to (+)-myrmicarin 215B (2), it
substantially reduced the rate of the subsequent dimerization
to 20, yielding a 1:1 ratio of (+)-myrmicarin 215B (2) to 20
after a period of 190 min (Table 1, entry 6), whereas the same
ratio was obtained in 30 min in the absence of LiCl. Consistent
with our reported result for benzene-d6,2a the acetic acid (0.20
equiv) treatment of a THF-d8 solution of 15 (0.054 M) saturated
with LiCl effected complete conversion to (+)-myrmicarin 215B
(2) but did not promote dimerization (Table 1, entry 7). By
contrast, treatment of a THF-d8 solution of (+)-myrmicarin
215B (2, 0.0065 M) saturated with LiClO4 as the salt additive
with TFA (1.10 equiv) immediately generated a mixture of
pyrrole ring-protonated tricyclic species and produced none of
the heptacyclic iminium ion 20 (Table 1, entry 8).
To rule out possible influence of the trifluoroacetate anion
in the dimerization process, a THF-d8 solution of 15 (0.013 M)
saturated with LiClO4 was treated with HClO4 (1.10 equiv) in
place of TFA. Consistent with the observed TFA-induced
reactivity (Table 1, entry 8), introduction of HClO4 instantly
generated a mixture of ring protonated tautomers 24 (Table 1,
entry 9). Exposure of a methanol-d4 solution of 15 (0.013 M)
saturated with LiClO4 to HClO4 (1.10 equiv) afforded a mixture
(5) Heating a solution of this mixture of dimer 20 and sulfide 22
(55 °C) resulted only in decomposition of the dimer.
10068 J. Org. Chem., Vol. 72, No. 26, 2007

Dimerization of Pyrroloindolizine DeriVatiVes
SCHEME 4 a
a (A) FMO analysis of a concerted [6πa+2πs] cycloaddition between an azafulvenium ion and an alkene. (B) Possible cycloaddition involving the
E-azafulvenium ion 25 and (+)-myrmicarin 215b (2) leading to iminium ion 20 en route to isomyrmicarin 430A (17). (C) Potential cycloaddition leading
to iminium ion 10 en route to myrmicarin 430A (4).
SCHEME 5. Expected Heptacycles from a [6πa + 2πs]
Cycloaddition between 2 and Azafulvenium Ions 28 and 29
Interestingly, preliminary computations show that the Z-azaful-
venium ion 8 is favored over the E-isomer 25 by ∼1.3 kcal/
mol, consistent with predictions based on molecular models and
allylic strain considerations.6 This would suggest that the
observed dimerization may occur by equilibration of 8 and 25
accompanied by a faster cycloaddition involving the high-energy
isomer 25.
A persuasive factor in our consideration of this [6πa + 2πs]
cycloaddition was that an analogous process involving the
Z-azafulvenium ion 8 would provide precisely the connectivity
and stereochemistry found in myrmicarin 430A (4, Schemes
1B and 4C). Hence, mutual association of the convex faces of
(+)-myrmicarin 215B (2) and the Z-azafulvenium ion 8 would
enable the approach of the C8-C9 alkene of (+)-myrmicarin
215B (2) syn-coplanar to the C8 methine of Z-azafulvenium
ion 8, providing the iminium ion 10 via an antarafacial
cycloaddition en route to myrmicarin 430A (4, Scheme 4C).
Design and Synthesis of Conformationally Restricted
Azafulvenium Ion Precursors. While it may be speculated that
an enzyme-mediated protonation event could be responsible for
triggering a cycloaddition reaction and controlling the stereo-
chemistry of an azafulvenium ion in the biosynthesis of
myrmicarin 430A (4), we wished to examine this hypothesis
experimentally using conformationally restricted substrates. If
the dimerization was occurring through a concerted cycload-
dition mechanism and the regio- and stereochemical outcome
was dictated by the geometry of the azafulvenium ion, then a
structure locked in the Z-geometry should undergo a cycload-
dition with (+)-myrmicarin 215B (2) to provide the myrmicarin
430A (4) framework (Scheme 5). Therefore, we prepared the
alcohols 26-27, in which a siloxy tether favors a Z-azafulve-
nium ion geometry upon ionization of the C8 alcohol (Scheme
5). The Z-azafulvenium ion 28, which is expected to give the
ion 29. Additionally, sterically demanding alkyl groups on
desired myrmicarin 430A (4) carbon skeleton in a [6πa + 2πs]
silicon would further favor formation and cycloaddition via the
manifold, can better accommodate the azafulvenium ion in the
Z-azafulvenium ion 28.
eight-membered ring than the corresponding E-azafulvenium
The synthesis of the tethered azafulvenium ion precursor 40
commenced with oxidation of our previously reported (R)-
(6) DFT calculations performed using B3LYP/6-31G; gas phase, energy-
minimized structure of the free azafulvenium ions.
alcohol 322b to the aldehyde 33 using 2-iodoxybenzoic acid in
J. Org. Chem, Vol. 72, No. 26, 2007 10069

Movassaghi et al.
SCHEME 6. Synthesis of Azafulvenium Ion Precursor 40a
To investigate the heterodimerization event of interest
(Scheme 5) we required a means of generating the putative
azafulvenium ion from the conformationally restricted precursor
40 without activation of (+)-myrmicarin 215B (2). To avoid
rapid self-dimerization of (+)-myrmicarin 215B (2) in the
presence of Brønsted acid, we relied on Lewis acid activation
of the C8-alcohol 40. As validation of this strategy, under
optimal conditions, treatment of the tricyclic C8-alcohol 15 (1:1
mixture of C8 alcohols) with scandium trifluoromethane-
sulfonate (Sc(OTf)3, 0.20 equiv) in the presence of the E-
thiolketene acetal 41 (2.00 equiv, E(O):Z(O) ) 20:1)11 provided
the thiolester 42 in 85% yield (eq 1).12
Equation 1. Activation of 15 and π-nucleophilic trapping.
On the basis of these observations we sought to substitute
an electron-rich vinyl pyrroloindolizine derivative in place of
the thiolketene acetal 41 to probe the cycloaddition hypothesis
outlined in Scheme 5. In situ 1H NMR monitoring showed that
portionwise addition of 0.40 equiv of Sc(OTf)3 to a mixture of
(+)-myrmicarin 215B (2) and conformationally restricted
azafulvenium ion precursor 40 (2:40, 2:1) in acetonitrile-d3
selectively and completely produced a single new heterodimeric
species (g90% by 1H NMR)13 with no visible formation of any
isomyrmicarin 430A derivatives. This product proved to be
slightly more stable to isolation than isomyrmicarin 430A (17).
Isolation and structural analysis of this new air-sensitive
heterodimerization product by a combination of 2D NMR
techniques established the structure of the heterodimer 43 (eq
2), which possessed connectivity and stereochemistry consistent
with that of isomyrmicarin 430A (17, Scheme 2). Strong C1-
H/C4-H NOESY correlations confirmed the connectivity of
the cyclopentane ring, while C1-H/C10′-H and C2-H/C3-H
correlations secured the stereochemistry at C1, C2, and C3. A
C3-H/C9-H correlation confirmed that the C9 stereochemistry
of the alcohol 40 had been retained.
a Conditions: a) IBX, DMSO, 82%. b) TESOTf, 2,6-lutidine, PhH,
23 °C, 4:1 dr. c) TBAF, THF, 81% (2 steps). d) TIPSOTf, 2,6-lutidine,
CH2Cl2, -40 f 0 °C. 82%. e) KHMDS, Davis oxaziridine, THF -78 °C,
i
90%, 5:1 dr. f) TBAF, THF, 92%. g) Pr2SiCl2, DMAP, DMF, 77%. h)
LAH, Et2O, -78 f 0 °C, 77%, 6:1 dr.
dimethylsulfoxide (Scheme 6).7 Treatment of the aldehyde 33
with excess triisopropylsilyl trifluoromethanesulfonate in ben-
zene at 50 °C for 4 h effected C8-silyl enol ether formation
and Friedel-Crafts cyclization to provide a C3-silyloxy pyr-
roloindolizine as a 6:1 C3-epimeric mixture. The acidic workup
of this mixture and careful chromatographic separation yielded
the diastereomerically pure ketone 36 in 81% overall yield. For
large-scale preparation, an alternative sequence was employed,
which involved subjection of aldehyde 33 to triethylsilyl
trifluoromethanesulfonate at 23 °C for 25 min, exhaustive
desilylation of the crude mixture, facile chromatographic
separation of the C3 epimeric alcohols, and silylation of the
major C3 epimer 35 (Scheme 6). In this series, the major C3S-
hydroxy pyrroloindolizine and related derivatives were found
to be far more stable than the minor C3R-hydroxy series and
were thus employed in subsequent steps. Hydroxylation of the
lithium enolate of 36 with Davis oxaziridine8 generated the
R-keto alcohol 37 in 85% yield as a 4:1 mixture of chromato-
graphically separable C8-epimers.9 Unveiling10 the C3-alcohol
allowed the introduction of the desired [1,3,2]-dioxasilocine
substructure of ketone 39. Reduction of the C8 ketone provided
the C8-alcohol 40 (10:1 dr, major shown, Scheme 6), which
was obtained in diastereomerically pure form after flash column
chromatography.
Equation 2. Diastereoselective condensation of (+)-myrmi-
carin 215B (2) and alcohol 40.
While molecular model analysis of the E-azafulvenium ion
29 (Scheme 5) en route to the isomyrmicarin 430A framework
indicated that this compound was significantly more strained
(7) Frigerio, M.; Santagostino, M. Tetrahedron Lett. 1994, 35, 8019-
8022.
(8) Davis, F. A.; Vishwakarma, L. C.; Billmers, J. G.; Finn, J. J. Org.
Chem. 1984, 49, 3241-3243.
(9) Attempts to introduce the C9-alcohol via the C8-C9 silyl enol ether
by dihydroxylation or epoxidation failed, likely as a result of the sensitivity
of the vinyl pyrrole nucleus toward oxidation.
(11) For the synthesis of E(O)- and Z(O)-thiolketene acetal 41, see:
Evans, D. A.; Burgey, C. S.; Kozlowski, M. C.; Tregay, S. W. J. Am. Chem.
Soc. 1999, 121, 686-699 and DeRoy, P. L.; Charette, A. B. Org. Lett.
2003, 5, 4163-4165, respectively.
(10) The presence of a free C3-hydroxyl group significantly complicated
the introduction of the C9-hydoxyl group.
10070 J. Org. Chem., Vol. 72, No. 26, 2007

Dimerization of Pyrroloindolizine DeriVatiVes
SCHEME 7. Synthesis of (9R)-Alcohol 47a
SCHEME 8. Diastereoselective Condensation of
(+)-Myrmicarin 215B (2) and Alcohol 47
a Conditions: a) PPh3, DEAD, p-NO2C6H4CO2H, THF, 61% (72% brsm).
i
b) LiOH, THF-H2O, 50 °C, 100%. c) TBAF, THF, 92%. d) Pr2SiCl2,
DMAP, DMF, 29%. e) LAH, Et2O, -78 f 0 °C, 16% (24% brsm).
than the corresponding Z-azafulvenium ion 28, it remained
possible that 29 could be accessed from (9S)-alcohol 40 (eq 2)
as a high energy intermediate. By contrast, a similar analysis
of the C9-epimer of 40 (i.e., 27, Scheme 5) clearly showed that
the corresponding E-azafulvenium ion derivative would suffer
a significant allylic strain that was absent in the desired
Z-azafulvenium ion. On the basis of the expectation that 27
(Scheme 5) would provide greater preference for formation of
28 over 29, we embarked on the preparation of the C9-epimer
of 40 (Scheme 7). To garner significant quantities of the required
C9-epimer, the C9-alcohol 37 was subjected to Mitsunobu
inversion14 with p-nitrobenzoic acid (Scheme 7) followed by
hydrolysis and desilylation to provide the diol 45. While
introduction of the tether had proceeded efficiently with the C9
epimer 38 (Scheme 6), exposure of diol 45 to the same
conditions provided only 29% of the desired product 46 along
with a mixture of C9- and C3-mono- and bis-silylated products
(Scheme 7), suggesting a more challenging formation of the
epimeric [1,3,2]-dioxasilocine substructure. Separation of 46
from the mixture and desilylation of the undesired products
allowed recycling to access the target compound 46. As another
testament to the conformational and reactivity differences
between the C9-epimers of these pyrroloindolizines, lithium
aluminum hydride reduction of the ketone 46 afforded the
unstable C3,C8,C9-triol derivative as the major product, along
with the desired alcohol 47 and recovered starting material.
Despite these challenges, the desired azafulvenium ion precursor
47 was obtained in low yield as a single diastereomer (Scheme
7) in this sequence, allowing its examination as an azafulvenium
ion precursor.
g90% yield by 1H NMR.15 Reciprocal C4/C1-H and C1/C4-H
HMBC correlations revealed that the dimer 49 again had the
same connectivity as isomyrmicarin 430A (Scheme 8). C1-H/
C10′-H and C2-H/C3-H NOESY correlations revealed the
stereochemistry at C1, C2, and C3 was identical to that of 43
and isomyrmicarin 430A (17). Finally, C1-H/C9-H and C9-
H/C10′-H correlations confirmed that the heterodimers 49 and
43 differed only in their C9-stereochemistry. It should be noted
that the prohibitively strained E-azafulvenium ion 48b is an
unlikely intermediate in this dimerization reaction. Additionally,
an antarafacial [6πa + 2πs] cycloaddition involving a Z-
azafulvenium ion and (+)-myrmicarin 215B (2) would furnish
a product with the stereochemistry at C3 opposite to that found
in 49. These factors taken together suggest that the observed
heterodimerization does not proceed via a concerted cycload-
dition. Instead, the structure of condensation product 49 is more
consistent with a stepwise addition of (+)-myrmicarin 215B
(2) to the azafulvenium ion 48a, followed by attack of the
pyrrole at to the resulting electrophilic C1 (Scheme 8). While
these results did not rule out a concerted [6πa + 2πs] pathway
for the formation of either isomyrmicarin 430A (17) or
myrmicarin 430A (4), they did implicate significant obstacles
in implementation of a single-step cyclopentannulation employ-
ing these azafulvenium ions toward the synthesis of the complex
myrmicarins.
A Directed Heterodimerization Approach to Functional
and Dimeric Pyrroloindolizines. To overcome the inherent
propensity to form the isomyrmicarin substructure in the
observed homo- and heterodimerization reactions we required
a means of decoupling the two bond-forming events of the
cyclopentannulation. Hence, we sought the selective electrophilic
activation of a pyrroloindolizine derivative 50 (Scheme 9)
followed by nucleophilic trapping by 51 to provide the C2-C3
bond in the heterodimer 53. We reasoned that the introduction
of a “blocking” functional group (Y) at C8 of the pyrroloin-
dolizine 51 (Scheme 9) should either prevent an intramolecular
In the key heterodimerization event, exposure of an equal
mixture of (+)-myrmicarin 215B (2) and alcohol 47 to the
conditions described previously (eq 2) completely and exclu-
sively afforded a new heterodimeric product (Scheme 8) in
(12) Under the conditions described above, thiolester 42 was obtained
as an inseparable mixture of diastereomers (dr, 1:17:2.5:13). Use of the
corresponding Z-thiolketeneacetal (2.00 equiv, E(O):Z(O) ) 1:12) gave 42
with a diminished level of diastereoselection (dr, 1:2.1:1.4:2.3).
1
(13) In situ H NMR monitoring of the homogeneous reaction mixture
showed complete consumption of 40 with concomitant formation of 43 and
no other visible compounds except remaining excess 2.
(14) For a review see: Hughes, D. L. Org. Prep. Proced. Int. 1996, 28,
127-164.
1
(15) In situ H NMR monitoring of the homogeneous reaction mixture
showed complete consumption of 47 and 2 with concomitant clean and
exclusive formation of 49.
J. Org. Chem, Vol. 72, No. 26, 2007 10071

Movassaghi et al.
SCHEME 9. Directed Heterodimerization of Functional
Pyrroloindolizines
SCHEME 10. Synthesis of the Hexacycle 61 by Directed
Dimerization of Functional Pyrroloindolizinesa
Friedel-Crafts addition to directly give the desired hexacycle
55 or promote cleavage of the C1-C3b bond in an isomyrmi-
carin derivative 54 (Scheme 9). Introduction of a functional
handle at C9 of the electrophilic precursor would enable
adjustment of the stereochemistry at C3 after dimerization. This
strategy could also benefit from the use of tethered azafulvenium
ions (i.e., as 40 and 47) to secure dimeric products with a high
level of diastereoselection. Significantly, this approach would
provide access to functional hexacyclic derivatives pertinent to
alternative late-stage cyclopentannulation chemistries, such as
free-radical-mediated cyclization as suggested by our proposed
radical dimerization of pyrroloindolizines (Scheme 1C). With
these considerations in mind, we investigated the directed
heterodimerization of pyrroloindolizine derivatives.
a Conditions: a) TIPSOTf, Et3N, CH2Cl2, -40 °C, 100%. b) KHMDS;
Davis’ oxaziridine, THF, -78 °C, 86%, 5:2 dr. c) NaH, MeI, DMF, 23 °C,
94%. d) Tf2O, 2,6-di-tert-butyl-4-methylpyridine, CH2Cl2, -78 °C, 84%.
e) PPTS, H2O, MeCN, C6H6, 23 °C, 70%.
of 60 to the diketone 61 was achieved by exposure to pyridinium
p-toluenesulfonate in benzene-acetonitrile-water at 23 °C
under strictly oxygen-free conditions.16 Importantly, this two-
step sequence provides a directed condensation between two
pyrroloindolizine fragments (i.e., 57 and 59), while at the same
time securing the C1-vinylogous amide needed for activation
and cyclopentannulation, in addition to the C9-ketone for control
of C3-stereochemistry.
In parallel to these efforts, we investigated the use of
conformationally restricted pyrroloindolizine electrophiles to
gain stereochemical control in the heterodimerization reaction.
For example, when the azafulvenium precursor 40 is treated
with catalytic scandium trifluromethanesulfonate in the presence
of the silyl enol ether 57 (1.56 equiv), heterodimer 62 is obtained
in 73% yield as a 5:1 mixture of two diastereomers (eq 3).
Key C2-H/C4-H and C9-H/C3-H NOE correlations clearly
secure the structure of the major diastereomer as shown in eq
3.
The (R)-ketone 562b served as a versatile building block for
the preparation of the necessary functional pyrroloindolizines
(Scheme 10). The silyl enol ether 57 was prepared quantitatively
and exclusively as the Z-isomer by treatment of the ketone 56
with triisopropyl trifluoromethanesulfonate in the presence of
triethylamine at low temperature. Hydroxylation of the lithium
enolate of 56 followed by O-methylation with methyl iodide
and sodium hydride provided the C9-methoxy ketopyrroloin-
dolizine 59 as a mixture of C9-epimers (5:2). As in the synthesis
of ketone 37, strict control over the stoichiometry of the oxidant
during the hydroxylation step was necessary to avoid further
oxidation of the product 58. Inspired by our earlier studies, we
explored the direct electrophilic activation of the vinylogous
amide 59 in the presence of silyl enol ether 57. Despite
pronounced Brønsted acid sensitivity, the silyl enol ether 57
served as a superb nucleophile in trapping the activated
derivative of vinylogous amide 59. In this event, treatment of
an equal mixture of the R-methoxy ketone 59 and the silyl enol
ether 57 with trifluoromethanesulfonic anhydride in the presence
of 2,6-di-tert-butyl-4-methylpyridine provided the methyl vinyl
ether 60 in 70% yield as a mixture of diastereomers. Hydrolysis
10072 J. Org. Chem., Vol. 72, No. 26, 2007

Dimerization of Pyrroloindolizine DeriVatiVes
diameter: 3.0 cm, height: 10 cm) to afford the R-methoxy ketone
59 (173 mg, 94%, mixture of C9 epimers, 9:5 dr) as a colorless
oil. H NMR (500 MHz, C6D6, 20 °C, 9:5 dr) δ 4.30 (q, J ) 6.8
Equation 3. Diastereoselective synthesis of hexacycle 62 from
silyl enol ether 57 and the conformationally restricted pyrro-
loindolizine 40.
1
Hz, 1H), 4.28 (q, J ) 6.7 Hz, 1H), 2.94-3.16 (m, 6H), 2.71 (dd,
J ) 15.8, 8.1 Hz, 1H), 2.70 (dd, J ) 15.7, 8.1 Hz, 1H), 2.50 (ddd,
J ) 16.9, 11.8, 6.7 Hz, 1H), 2.70 (dd, J ) 15.7, 8.1 Hz, 1H), 2.50
(ddd, J ) 16.0, 10.5, 5.8 Hz, 1H), 2.50 (ddd, J ) 16.0, 10.7, 5.2
Hz, 1H), 2.38 (ddd, J ) 16.5, 6.4, 0.9 Hz, 2H), 2.13 (ddd, J )
16.9, 11.8, 6.7 Hz, 1H), 2.12 (ddd, J ) 16.4, 12.1, 6.3 Hz, 1H),
1.73-1.79 (m, 2H), 1.47-1.55 (m, 2H), 1.45-1.49 (m, 12H),
1.29-1.42 (m, 4H), 0.99-1.20 (m, 2H), 0.62-0.72 (m, 2H). 13C
NMR (125.8 MHz, C6D6, 20 °C, 9:5 dr) δ 195.5, 195.4, 137.5,
137.5, 126.8, 126.8, 121.1, 121.0, 115.3, 115.1, 81.3, 81.1, 56.6,
56.6, 56.1, 56.1, 36.2, 36.2, 29.7, 29.6, 29.3, 29.1, 19.0, 18.5, 16.4,
16.3, 22.6, 22.5, 20.3, 20.3, 19.9, 19.9. FTIR (neat) cm-1: 2929
(s), 1651 (s, CdO), 1494 (s), 1320 (m), 1037 (w). HRMS-EI (m/
z): calcd for C16H23NNaO2 [M + Na]+: 284.1621, found:
284.1626. TLC (2.5% Et3N, 17.5% EtOAc, 20% CH2Cl2-hexanes),
Rf: 0.26 (UV, CAM).
With these promising initial results, our current efforts are
focused on merging of the fragment assembly strategy illustrated
in Scheme 10 with the use of conformationally biased azaful-
venium ion precursors (i.e., 40 and 47) to access functional and
dimeric pyrroloindolizines with enhanced levels of diastereo-
selection. Additionally, the C1-carbonyl of dimeric pyrroloin-
dolizines allows investigation of free-radical- or azafulvenium
ion-mediated pathways to introduce the C1-C8b bond found
in complex myrmicarins guided by the hypotheses outlined in
Scheme 1.
Conclusions
Vinyl pyrroloindolizines possess a unique structure that
predisposes them to highly efficient activation at C8. This mode
of activation enables a variety of activated pyrroloindolizine
derivatives to undergo reaction with neutral pyrroloindolizines
to afford the corresponding dimeric structures. In an ionic
manifold, Brønsted acid activation of the natural alkaloid (+)-
myrmicarin 215B (2) leads to highly efficient and stereoselective
homodimerization to form the heptacyclic isomyrmicarin 430A
(17) under a variety of reaction conditions. Likewise, Lewis
acid activation of the conformationally restricted azafulvenium
precursors 40 or 47 in the presence of (+)-myrmicarin 215B
(2) affords heptacyclic derivatives 43 and 49, respectivelys
structures that are consistent with a highly efficient nonconcerted
ionic heterodimerization process. Through strategic design of
the pyrroloindolizine species involved in the dimerization (e.g.,
57 and 59), the two bond-forming events of the cyclopentan-
nulation can be decoupled, providing hexacyclic dimers en route
to the complex myrmicarin alkaloid structures. Importantly,
dimeric products such as 61 and 62 contain functional groups
needed to investigate alternative radical or ionic pathways to
control the regio- and stereochemistry of the cyclopentannulation
process. While representing a concise route to the complex
myrmicarins, the dimerization of these pyrroloindolizines offers
insight with potential relevance to their biosynthesis. Such
considerations continue to guide our study of these structurally
fascinating alkaloids.
Z-4-Ethyl-3-[1-(triisopropyl-silanyloxy)propenyl]-1,2,5,6,7,7a-
hexahydro-pyrrolo[2,1,5-cd]indolizine (57). Triisopropylsilyl tri-
fluoromethanesulfonate (24.2 µL, 89.9 µmol, 1.05 equiv) was added
dropwise to an anhydrous solution of ketone 56 (19.8 mg, 85.6
µmol, 1 equiv) and triethylamine (59.7 µL, 428 µmol, 5.00 equiv)
in dichloromethane (1.70 mL) at -45 °C such that the intense
yellow color produced upon adding each drop had completely
disappeared before the next drop was added. After complete addition
of triisopropylsilyl trifluoromethanesulfonate, an ice-cold solution
of saturated aqueous sodium bicarbonate solution (3.5 mL) was
added. The reaction flask was immediately removed from the
cooling bath, and the mixture was diluted with an additional portion
of ice-cold saturated aqueous sodium bicarbonate solution (1.5 mL)
and diethyl ether (7.5 mL). The aqueous layer was separated and
extracted with diethyl ether (2 × 6 mL). The combined organic
phases were washed with ice-cold brine (3.5 mL), were dried over
anhydrous sodium sulfate, were filtered, and were concentrated
under reduced pressure to afford the pure Z-triisopropylsilyl enol
1
ether 57 (33.2 mg, 100%) as a colorless oil. H NMR (500 MHz,
C6D6, 20 °C) δ 4.91 (q, 1H, J ) 6.7 Hz), 3.31 (tdd, 1H, J ) 10.7,
5.5, 3.8 Hz), 2.65-2.88 (m, 4H), 2.57 (ddd, 1H, J ) 16.3, 6.0, 0.7
Hz), 2.37 (ddd, 1H, J ) 16.3, 11.8, 6.9 Hz), 1.99 (dt, J ) 11.6, 5.5
Hz, 1H), 1.92 (d, 3H, J ) 6.7 Hz), 1.66 (ddddd, 1H, J ) 13.7, 6.7,
4.1, 2.7, 0.7 Hz), 1.49-1.61 (m, 2H), 1.29-1.39 (m, 1H), 1.37 (t,
3H, J ) 7.5 Hz), 0.82 (tdd, 1H, J ) 12.7, 10.7, 2.7 Hz). 13C NMR
(125.8 MHz, C6D6, 20 °C) δ 148.2, 129.3, 121.7, 119.1, 116.7,
104.6, 55.6, 37.5, 30.4, 25.9, 23.0, 21.1, 19.9, 18.7, 16.2, 14.3, 12.3.
FTIR (neat): 2943 (s), 2865 (s), 1661 (s, CdC), 1463 (s), 1060
(s). HRMS-EI (m/z): calcd for C24H42NOSi [M + H]+: 388.3030,
found: 388.3046. TLC (alumina gel, 10% EtOAc-hexanes), Rf:
0.72 (UV, CAM).
1,3-Bis-(1-ethyl-3,4,4a,5,6,7-hexahydro-pyrrolo[2,1,5-cd]in-
dolizin-2-yl)-4-methoxy-2-methyl-pent-3-en-1-one (60). A solu-
tion of 2,6-di-tert-butyl-4-methylpyridine (377 mg, 1.84 mmol, 5.00
equiv) in dichloromethane (500 µL + 250 µL rinse) was added to
an anhydrous solution of triisopropylsilyl enol ether 57 (160 mg,
413 µmol, 1.13 equiv) and the R-methoxyketone 59 (96.0 mg, 367
µmol, 1 equiv, mixture of C9 epimers, 5:2 dr) in dichloromethane
(2.50 mL) at 23 °C, and the resulting solution was cooled to
-78 °C. Trifluoromethanesulfonic anhydride (31.0 µL, 184 µmol,
0.500 equiv) was added dropwise via syringe, causing the solution
to become opaque and deep burgundy within 90 s. After 30 min,
three additional portions of trifluoromethanesulfonic anhydride (31.0
µL, 184 µmol, 0.500 equiv) were added in 30 min intervals. When
40 min had elapsed after the last addition, saturated aqueous sodium
bicarbonate solution (1.5 mL) was added, and the aqueous phase
was allowed to freeze (<5 s), and the reaction flask was removed
from the cooling bath. After approximately 5 min at 23 °C the
biphasic mixture was diluted with ethyl acetate (15 mL) and an
additional portion of saturated aqueous sodium bicarbonate solution
Experimental Section
1-(1-Ethyl-3,4,4a,5,6,7-hexahydro-pyrrolo[2,1,5-cd]indolizin-
2-yl)-2-methoxy-propan-1-one (59). Sodium hydride (60% disper-
sion in mineral oil, 56.0 mg, 1.40 mmol, 2.00 equiv) was added in
a single portion to an anhydrous solution of the R-hydroxy ketone
58 (173 mg, 700 µmol, 1 equiv) and methyl iodide (131 µL, 2.10
mmol, 3.00 equiv) in dimethylformamide (4.75 mL) at 23 °C. The
resulting suspension gradually became a clear and coloress solution
within 5 min. After 1 h, the solution was slowly poured into a
mixture of water and saturated aqueous ammonium chloride solution
(1:1, 20 mL) causing vigorous gas evolution, and this mixture was
diluted with a solution of ethyl acetate and hexanes (3:2, 25 mL).
The aqueous phase was separated and extracted with ethyl acetate-
hexanes (3:2, 4 × 20 mL). The combined organic phases were
washed with brine (15 mL), were dried over anhydrous sodium
sulfate, were filtered, and were concentrated under reduced pressure.
The residue was purified by flash column chromatography on silica
gel (eluent: 2.5% triethylamine and 37.5% ethyl acetate in hexanes,
(16) The oxidation of the more electron-rich pyrroloindolozine substruc-
ture is greatly accelerated in the presence of acid.
J. Org. Chem, Vol. 72, No. 26, 2007 10073

Movassaghi et al.
(5 mL) and water (1.5 mL). The aqueous layer was separated and
extracted with ethyl acetate (3 × 7.5 mL). The combined aqueous
phases were washed with brine (5 mL), were dried over anhydrous
sodium sulfate, and were filtered. Benzene (5 mL) was added to
the organic phase, and it was concentrated to a deep burgundy
solution (final volume approximately 250 µL). This sample was
directly subjected to flash column chromatography on silica gel
(eluent: 2% triethylamine, 2% ethyl acetate, and 5% dichlo-
romethane in hexanes, then flushed with 5% triethylamine in ethyl
acetate to recover any remaining R-methoxy ketone, diameter: 3.0
cm, height: 25 cm) to afford the desired heterodimer 60 (142 mg,
23.4, 23.3, 22.9, 22.9, 22.8, 21.9, 21.8, 21.8, 20.1, 20.1, 20.0, 20.0,
20.0, 20.0, 20.0, 20.0, 19.9, 17.9, 17.6, 17.6, 16.5, 16.1, 16.0, 16.0,
15.8, 15.8, 15.6, 15.4, 15.2. FTIR (neat) cm-1: 2928 (s), 1643 (s,
CdO), 1496 (s), 1427 (s), 1321 (m). HRMS-EI (m/z): calcd for
C31H43N2O2 [M + H]+: 475.3319, found: 475.3316. TLC (Et3N-
pretreated silica gel, 1% Et3N, 9% EtOAc-hexanes), Rf: 0.31 (UV,
CAM).
Acknowledgment. M.M. is a Firmenich Assistant Professor
of Chemistry and a Beckman Young Investigator. A.E.O.
acknowledges a Novartis Graduate Fellowship. We thank
Professor Robert G. Griffin and Dr. Tony Bielecki for use of a
high-field instrument at the MIT-Harvard Center for Magnetic
Resonance (EB-002026). We thank Dr. Li Li for obtaining mass
spectrometric data at the Department of Chemistry’s Instru-
mentation Facility, Massachusetts Institute of Technology. We
thank Dr. Timothy E. Barder in the Buchwald group and Mr.
Omar K. Ahmad for their assistance with DFT calculations. We
are grateful for financial support by NIH-NIGMS (GM074825).
1
84%, mixture of stereoisomers, 3:2:1 dr) as a pale orange oil. H
1
NMR H NMR (500 MHz, C6D6, 20 °C, 3:2:1 dr) δ 4.06 (q, J )
7.0 Hz, 1H), 4.04 (q, J ) 7.0 Hz, 1H), 4.04 (q, J ) 7.0 Hz, 1H),
3.24 (s, 3H), 3.24 (s, 3H), 3.24 (s, 3H), 3.09-3.46 (m, 6H), 2.30-
3.09 (m, 36H), 2.14-2.25 (m, 6H), 2.09 (s, 3H), 2.07 (s, 3H), 2.03
(s, 3H), 1.75 (d, J ) 7.0 Hz, 3H), 1.74 (d, J ) 7.0 Hz, 3H), 1.72
(d, J ) 7.0 Hz, 3H), 1.33-1.97 (m, 24H), 1.15-1.30 (m, 18H),
0.73-1.02 (m, 6H). 13C NMR (125.8 MHz, C6D6, 20 °C, 3:2:1 dr)
δ 196.0, 195.6, 195.9, 150.7, 150.6, 150.5, 136.5, 136.3, 136.3,
130.6, 130.3, 129.6, 126.8, 126.6, 126.4, 126.4, 126.4, 126.4, 123.0,
122.9, 122.8, 120.3, 120.3, 120.2, 118.2, 118.2, 118.1, 117.9, 117.8,
117.7, 113.2, 113.1, 112.4, 56.6, 56.6, 56.5, 56.1, 56.1, 55.9, 55.8,
55.7, 55.7, 48.7, 48.6, 47.9, 37.5, 37.4, 37.4, 36.6, 36.5, 36.3, 30.6,
30.4, 30.3, 30.1, 30.0, 30.0, 28.1, 28.0, 27.6, 26.7, 26.5, 26.3, 23.4,
Supporting Information Available: Experimental procedures
and spectroscopic data for selected new compounds. This material
is available free of charge via the Internet at http://pubs.acs.org.
JO701981Q
10074 J. Org. Chem., Vol. 72, No. 26, 2007