Total Synthesis of Hamigerans and Analogues Thereof. Photochemical Generation and Diels-Alder Trapping of Hydroxy-o-quinodimethanes

Article abstract of DOI:10.1021/ja030498f

A number of naturally occurring substances, including hamigerans, contain ring systems which are fused to an aromatic nucleus. A general and streamlined method for the construction of such benzannulated bi- and polycyclic carbon frameworks has been develo

Full text of DOI:10.1021/ja030498f

Published on Web 12/17/2003
Total Synthesis of Hamigerans and Analogues Thereof.
Photochemical Generation and Diels-Alder Trapping of
Hydroxy-o-quinodimethanes
K. C. Nicolaou,* David L. F. Gray, and Jinsung Tae
Contribution from the Department of Chemistry and The Skaggs Institute for Chemical Biology,
The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, California 92037, and
Department of Chemistry and Biochemistry, UniVersity of California, San Diego,
9500 Gilman DriVe, La Jolla, California 92093
Received August 18, 2003; E-mail: kcn@scripps.edu
Abstract: A number of naturally occurring substances, including hamigerans, contain ring systems which
are fused to an aromatic nucleus. A general and streamlined method for the construction of such
benzannulated bi- and polycyclic carbon frameworks has been developed, and its scope and limitations
were explored. On the basis of the photoenolization of substituted benzaldehydes and subsequent Diels-
Alder (PEDA) trapping of the generated hydroxy-o-quinodimethanes, this method was optimized to set the
stage for the total synthesis of several naturally occurring members of the hamigeran class. Specifically,
the developed synthetic technology served as the centerpiece process for the successful asymmetric
synthesis of hamigerans A (2), B (3), and E (7). In addition to the PEDA reactions, several other novel
reaction processes are described, including a high-yielding decarbonylative ring contraction and an oxidative
decarboxylation of a hydroxyl â-keto ester to afford an R-diketone. A number of analogues of these
biologically active natural products were also prepared by application of the developed technology.
Introduction
In the preceding paper in this issue,1 we described a total
synthesis of hybocarpone in which a photoenolization/Diels-
Alder (PEDA) cascade sequence played a pivotal role. Specif-
ically, this efficient process provided access to a benzocyclo-
hexane derivative which served as a precursor to hybocarpone’s
monomeric unit. The abundance of such structural motifs in
naturally occurring substances2 and the need to synthesize them
in the laboratory gifted us with an opportunity to explore this
reaction in detail and to apply it, in its intramolecular version,
to the total synthesis of hamigerans A, B, and E (1-4 and 7;
Figure 1). Isolated from the poecliosclerid sponge Hamigera
tarangaensis Bergquist and Fromont (family Achinoidae, syn.
Phorbasidae) collected at a depth of 30 m near the Hen and
Chicken Islands off the coast of New Zealand, the hamigerans
contain within their molecular architectures a substituted ben-
zenoid nucleus which is either fused onto a [4.3.0] or a [5.3.0]
bicyclic system featuring a cis junction, three or four contiguous
stereogenic centers, and an isopropyl group (1-5, hamigerans
A, B, and C; Figure 1) or appended to a cyclopentane moiety
carrying three stereogenic centers, a carboxylic acid group, and
Figure 1. Molecular structures of selected hamigerans (1-7).1,2
an isopropyl residue (6, 7, hamigerans E; Figure 1).3,4 It should
be noted that the absolute stereochemistry of hamigeran A (and
by extension B and C) has been proposed as shown by structure
2 by X-ray crystallographic analysis.3,4 The biological properties
of these compounds range from moderate cytotoxicity against
(1) Nicolaou, K. C.; Gray, D. L. F. J. Am. Chem. Soc. 2004, 126, 607-612.
(2) (a) Kupchan, S.; Morris, S.; Karim, A.; Marcks, C. J. Am. Chem. Soc.
1968, 90, 5923-5924. (b) Moir, L.; Thompson, R. J. Chem. Soc., Perkin
Trans. 1 1973, 1556-1559. (c) Pessoa, L.; De Lemos, T.; De Carvalho,
M.; Braz-Filho, R. Phytochemistry 1995, 40, 1777. (d) Takahashi, M.;
Sachihiko, I.; Takahashi, Y. Bull. Chem. Soc. Jpn. 1987 60, 2435-2443.
(e) Nagahama, S.; Iwaoka, T.; Ashitani, T. Mokuzai Gakkaishi 2000 46,
225-230. (f) Jianjun, G.; Guiqui, H.; Phytochemistry 1997, 44, 759-765.
(g) Parish, E. J.; Miles, H. D. J. Pharm. Sci. 1984, 73, 694-700.
(3) Cambie, R.; Lai, A.; Kernan, M. J. Nat. Prod. 1993, 58, 940-942.
(4) (a) Wellington, K.; Cambie, R.; Rutledge, P.; Bergquist, P. J. Nat. Prod.
2000, 63, 79-85. (b) Cambie, R. C.; Richard, C. E. F.; Rutledge, P. S.;
Wellington, K. D. Acta Crystallogr. 2001, C57, 958-960.
9
10.1021/ja030498f CCC: $27.50 © 2004 American Chemical Society
J. AM. CHEM. SOC. 2004, 126, 613-627
613

A R T I C L E S
Nicolaou et al.
which took special meaning in the context of hybocarpone
(intermolecular Diels-Alder trapping) and the hamigerans
(intramolecular Diels-Alder trapping). Encouraged by the
success of the hitherto unknown PEDA reaction with a 1,1-
disubstituted olefin (see preceding paper in this issue),1 we
proceeded to explore the reactivity of a series of activated
dienophiles toward benzaldehydes under photolytic conditions
as shown in Table 1. While the reaction appeared to be quite
general with a variety of substituted benzaldehydes, cyclizing
with olefins such as vinyl ketones, acrylate esters, acrylonitriles,
and acroleins to afford benzannulated systems, it was noticed
that certain dienophiles tended to polymerize rapidly under the
photolytic conditions employed. A systematic investigation of
reaction variables led to the identification of satisfactory
conditions for optimum results. As can be gleaned from Table
2, the effect of common reaction variables on reaction efficiency
was quite pronounced, and yields were improved substantially
from initial attempts (e.g., entries 1 and 2, Table 2). Thus, by
employing an excess of the polymerizable dienophiles, perform-
ing the reactions in dilute toluene solution, and using ordinary
Pyrex vessels (as opposed to quartz glass), these reactions
proceed efficiently as shown in Table 1. It was also found that
in going from the parent 2-methylbenzaldehyde (12; entry 1,
Table 1) to more substituted and electron-rich substrates, the
efficiency of the reaction improved substantially (entries 2 and
3, Table 1). This trend appears to be general across a range of
dienophiles and benzaldehydes as demonstrated by the more
electron-rich substrates corresponding to entries 3, 5-9, and
11-15 (Table 1). Aldehyde 16 also carries a methoxy substitu-
ent ortho to the aldehyde which might stabilize the dienol
species (e.g., 10; Figure 2) and increase the efficiency with
which the fleeting hydroxy-o-quinodimethane intermediate is
captured.12d We were pleased to find that commercially available
3-fluoro-2-methylbenzaldehyde (entry 10, Table 1) reacted with
methyl vinyl ketone to afford the corresponding bicyclic fluoride
in high yield under the developed photolytic conditions. An
example of a double PEDA reaction is also included in Table
1 (entry 17) in which a pentacyclic benzenoid system was
produced, albeit in low yield. It should also be noted that several
of the reported PEDA reactions represent new ground since they
involve hitherto unutilized-in-this-process dienophiles such as
cyclopentenone, acrylonitriles, and 1,1-disubstituted olefins,
leading to polycycles, nitrile products, and quaternary centers,
respectively (entries 4 and 17, 7 and 12, and 6, 8, and 14, Table
1).
Figure 2. General scheme for the synthesis of diverse bicycles 11 from
benzaldehydes 8 employed in the PEDA cascade reaction via hydroxy-o-
quinodimethanes.
the P-388 leukemia cells [e.g., 4-bromohamigeran B (4), IC50
) 13.5 µM] to strong antiviral activity against herpes and polio
viruses [e.g., hamigeran B (3), 100% inhibition at 132 µg/disk].4
Given their scarcity5 and biological actions, the laboratory
synthesis of these molecules was deemed important. In this paper
we describe the total synthesis of hamigerans A (1 and 2), B (3
and 4), and E (7), and a number of their analogues, as well as
an investigation of the generality and scope of the photoeno-
lization/Diels-Alder reaction, which facilitated the construction
of the hybocarpone and hamigeran structural motifs.6
Results and Discussion
Development of the PEDA Process. Despite the early
discovery of the photoenolization and subsequent Diels-Alder
trapping of o-methylbenzophenone by Yang and Rivas,7 the
potential of this process in total synthesis remained, for the most
part, relatively unexplored. The works of Charlton, Kraus, and
Quinkert are notable exceptions, but do not represent a
systematic study since they pertain to special circumstances.8-10
In contrast, the majority of the studies around this so-called
photoenolization process deal with its mechanism.11 Studies had
shown that dienol 10 (Figure 2) is a short-lived intermediate
which can either relax back to its ground state (i.e., 8) or be
trapped in a Diels-Alder fashion with electron-deficient di-
enophiles, leading to synthetically useful building blocks.12-15
It was with this background that we initiated this investigation,
(5) We gratefully acknowledge Professor R. Cambie for communications
pertaining to the hamigerans.
(6) Nicolaou, K. C.; Gray, D.; Tae, J. Angew Chem., Int. Ed. 2001, 40, 3675-
3678; Nicolaou, K. C.; Gray, D.; Tae, J. Angew Chem., Int. Ed. 2001, 40,
3679-3683. For a recent elegant total synthesis of hamigeran B, see: Clive,
D. L. J.; Wang, J. Angew Chem., Int. Ed. 2003, 42, 3406-3409.
(7) Yang, N.; Rivas, C. J. Am. Chem. Soc. 1961, 83, 2213.
(8) Charlton, J.; Koh, K.; Plourde, G. Tetrahedron Lett. 1989, 30, 3279-3282.
(9) (a) Kraus, G.; Zhao, G. J. Org. Chem. 1996, 61, 2770-2773. (b) Kraus,
G.; Chen, L.; Jacobson, R. Synth. Commun. 1993, 23, 2041-2049.
(10) Quinkert, G.; Stark, H. Angew. Chem., Int. Ed. Engl. 1983, 22, 647-655.
(11) For an overview of this photoenolization process, see: (a) Sammes, P.
Tetrahedron 1976, 32, 405 and references therein. (b) Weedon, A. C. In
The Chemistry of Enols; Rappoport, Z., Ed.; Wiley: Chichester, U.K., 1990;
pp 591-638 and references therein. (c) Charlton, J.; Alauddin, M.
Tetrahedron 1987, 43, 2873-2889 and references therein.
(12) For early examples of the photoenolization/Diels-Alder reactions with
benzylic ketones, see: (a) Haag, R.; Wirz, J.; Wagner, P. HelV. Chim. Acta
1977, 60, 2595-2607. (b) Pfau, M.; Combrisson, S.; Rowe, J.; Heindel,
N. Tetrahedron 1978, 34, 3459-3468. (c) Nerdel, F.; Brodowski, W. Chem.
Ber. 1968, 101, 1398-1406. (d) Wallace, T. W. Ph.D. Thesis, University
of London, 1974.
In further investigations, attempts were made to employ the
16
Narasaka-Mikami catalyst [(R)-BINOL)]TiCl2 to obtain
PEDA products enriched in one enantiomer. This effort proved
largely unsuccessful, although at relatively high catalyst loading,
measurable ee’s could be obtained (e.g., entry 1, Table 3). Given
the fleeting nature of the hydroxy-o-quinodimethane intermedi-
ate, the lack of profound enantioselection in this process is not
surprising.11a Furthermore, the diastereoselectivity observed in
these PEDA reactions is not spectacular either, with the ratio
between endo and exo Diels-Alder products ranging between
1.5:1 and 8:1 (see Table 1).
(13) For an early example of photoenolization/Diels-Alder reactions with
benzaldehydes, see: Arnold, B.; Mellows, S.; Sammes, P.; Wallace, T. J.
Chem. Soc., Perkin Trans. 1 1974, 3, 401-409.
(14) (a) Charlton, J.; Plourde, G.; Penner, G. Can. J. Chem. 1989, 67, 1010-
1014. (b) Connolly, T.; Durst, T. Tetrahedron 1997, 53, 15959-15982.
(15) For a recent review of the Diels-Alder reaction in total synthesis, see:
Nicolaou, K. C.; Snyder, S.; Montagnon, T.; Vassilikogiannakis, G. Angew.
Chem., Int. Ed. 2002, 41, 1668-1698.
The PEDA method is not without limitation. Thus, electron-
withdrawing groups on the aromatic nucleus block the produc-
(16) Mikami, K.; Motoyama, Y.; Terada, M. J. Am. Chem. Soc. 1994, 116,
2812-2820.
9
614 J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004

Synthesis of Hamigerans and Analogues Thereof
A R T I C L E S
Table 1. Benzannulation by Intermolecular Diels-Alder Trapping of Hydroxy-o-quinodimethanes Generated via Photoenolization (see
Figure 2)a
a o-Alkylbenzaldehyde (0.5-2 mmol) and olefin (4-20 equiv) were dissolved in deoxygenated toluene (0.03 M) in a Pyrex flask and irradiated at ambient
temperature (reactions warmed on irradiation) with a 450 W Hanovia lamp at a distance of 5 cm. Product ratios (endo:exo) as follows by entry number: (1)
ca. 4:1, (2) ca. 4:1, (3) ca. 3:1, (4) ca. 8:1, (5) ca. 12:4:1, (6) ca. 6:1, (7) ca. 4:1, (8) ca. 2:1, (9) ca. 9:3:1, (10) ca. 8:1, (11) ca. 6:1, (12) ca. 6:3:1, (13) ca.
2:1, (14) ca. 2:1, (15) ca. 2:1, (16) ca. 1.5:1, (17) ca. 2:1. b Product ratio determined by NMR spectroscopy.
tive pathway to annulated systems (an exception to this general
trend is the moderate success of entry 17, Table 1). Halogen
and nitro substituents (other than fluorine) were not tolerated
in this reaction. A photocyclization employing a pyridylben-
zaldehyde was likewise unsuccessful. While most of the
activated olefins tested as competent Diels-Alder partners, vinyl
9
J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004 615

A R T I C L E S
Nicolaou et al.
Table 2. Optimization of the Intermolecular Photochemically
Induced Diels-Alder Reactiona
c
amt of A
(equiv)
concn
(M)
λ
(nm)
tempb
(°C)
yield
(%)
entry
solvent
benzene
benzene
toluene
toluene
toluene
degassed toluene
1
2
3
4
5
6
2.0
2.0
2.0
6.0
6.0
6.0
0.2
0.2
0.2
0.02
0.02
0.02
260
300
300
300
300
300
75
75
75
75
45
45
15
28
38
63
70
83
Figure 3. General scheme for the synthesis of tricycles 36 from substrates
34 via the IMPEDA cascade via hydroxy-o-quinodimethanes 35.
a All reactions were performed by irradiation with a 450 W Hanovia
lamp on a 1.0 mmol scale. b Internal reaction temperature. The lower
temperature was obtained by increasing the distance of the reaction flask
from the lamp (3-8 cm). Combined yield of pure compounds. Products
were racemic and obtained as a separable mixture (ca. 2:1) of diastereo-
isomers.
chemical correlation was carried out to confirm this assumption,
particularly because NMR (coupling constant analysis and NOE
data) was inconclusive for several pairs of isomeric products.
The â-hydroxy ester 31 (entry 16, Table 1) was prepared by
microbial reduction of the corresponding â-keto ester using the
fungus Mucor racemosus, which was reported to be syn-selective
at a level of >99:1.19-21 From this whole-cell reduction, a single
product (31) was isolated whose optical rotation matched with
the literature value.20 This compound was, therefore, taken to
be the syn isomer. The product 31 obtained by the PEDA
reaction (entry 16, Table 1) proved identical to the microbial
reduction product by 1H NMR and 13C NMR spectroscopy (OH
and electron-withdrawing group syn to each other). As can be
seen from Figure 2, the syn isomer is the one which arises from
endo approach of the dienophile onto the hydroxy-o-quin-
odimethane. On the basis of this result and literature reports,11
the major reaction products were assigned as the syn diastereo-
isomers.
c
Table 3. Effect of (R)-BINOL-TiCl2 on the Diels-Alder Trapping
of a Photochemically Generated Hydroxy-o-quinodimethanea
amt of
(R)-BINOL−TiCl2
temp
(°C)
eeb
(%)
time
(h)
yield
(%)
entry
(equiv)
1
2
3
0.75
0.05
0.20
-40
-40
25
25
0
<5
12
12
6
30c
20d
45
a A solution of aldehyde 14 (0.2 mmol), methyl vinyl ketone (4.0 equiv),
and (R)-BINOL-TiCl2 were cooled to the indicated temperature and
irradiated (450 W Hanovia lamp) in deoxygentated toluene. Workup and
chromatography gave pure 15 along with significant amounts of recovered
starting matieral. b The enantiomeric excess was measured by chiral HPLC.
The absolute configuration was not determined. c 30% recovered starting
material. d 50% recovered starting material.
Intramolecular Trapping of Photochemically Generated
Hydroxy-o-quinodimethanes. As seen in Figure 3, a powerful
extension of this benzannulation methodology would be its
intramolecular variant (IMPEDA). No example of IMPEDA
reaction involving an all-carbon tether between the reacting
functionalities (e.g., 34; Figure 3) had been reported at the outset
of this work, and it was clear that the successful implementation
of such synthetic technology would provide a useful entry into
complex polycycles of the general form 36.22 To ascertain the
viability of this proposal, we proceeded to access the requisite
photocyclization precursors 46a-f, 47a-f, 48a-f, and 52-54
via a general and efficient route (Schemes 1 and 2). Thus, the
hydroxy aldehydes23 37-39 (Scheme 1) were protected as their
dithiane derivatives (BF‚OEt2, 1,3-propanedithiol, 90-95%
yield) and subsequently oxidized with IBX to deliver the
aldehydes 40-42 in high yields. The aldehydes so obtained
(40-42) were then reacted with various commercially available
phosphonates (A-D; Scheme 1), methylenetriphenylphospho-
sulfones and vinyl phosphates failed to give more than trace
amounts of PEDA products. Interestingly, while methyl acrylate
(entry 13, Table 1), methyl methacrylate (entry 14, Table 1),
and 3-penten-2-one (entry 5, Table 1) all reacted with substituted
benzaldehyde 16 in high yield to give the expected products,
the reaction proved very poor with methyl crotonate (entry 15,
Table 1).17
It has been determined that the (Z)-dienol 9 (Figure 2) is not
important in the Diels-Alder chemistry because it can rearrange
directly back to the starting material 8 by a 1,5-H shift (the
(E)-dienol has some small barrier to this rearrangement).11a
Given that the dienols which are trapped in the PEDA reaction
are exclusively of the (E)-dienol type and that the reaction
proceeds in a concerted fashion, the stereochemistry-determining
facet of this process is the endo or exo approach of the
dienophile to the dienol in the transition state (i.e., 10, Figure
2).18 While one might assume endo selectivity, a simple
(19) Buisson, D.; Cecchi, R.; Laffitte, J.-A.; Guzzi, U.; Azerad, R. Tetrahedron
Lett. 1994, 35, 3091-3094.
(20) Abalain, C.; Buisson, D.; Azerad, R. Tetrahedron: Asymmetry 1996, 7,
2983-2996.
(21) We gratefully acknowledge Professor D. Buisson for his generous and
timely gift of several fungal and yeast strains used in these experiments.
(22) For a report of an intramolecular photocyclization of a benzaldehyde moiety
with a heteroatom tether, see: Oppolzer, W.; Keller, K. Angew. Chem.,
Int. Ed. Engl. 1972, 11, 728-730.
(17) We suspect that a trace contaminant in the commercial formulation of
methyl crotonate might be catalyzing rapid decomposition of the dienophile
upon photochemical irradiation in this reaction.
(18) Charlton, J.; Plourde, G.; Koh, K. Can. J. Chem. 1989, 67, 574-579.
(23) Huang, Q.; Hunter, J.; Larock, R. J. Org. Chem. 2002, 67, 3437-3444.
9
616 J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004

Synthesis of Hamigerans and Analogues Thereof
A R T I C L E S
Scheme 1. Synthesis of Substituted Benzaldehydes 46a-f,
47a-f, and 48a-f for Intramolecular Photoenolization/Diels-Alder
Reactionsa
aldehydes were subjected to fast chromatography, affording
substrates 46a-f, 47a-f, and 48a-f (see Scheme 1).
Cyclization substrates 52-54 were prepared according to
Scheme 2. Thus, diol 49 was protected as a bis(silyl ether)
[(TBS)Cl, Et3N, 94%] and then subjected to Wacker oxidation
[Pd(OAc)2, Cu(OAc)2, H2O, O2] to furnish ketone 50 (88%
yield). Horner-Wadsworth-Emmons homologation of this
ketone (50) with the anion derived from ester phosphonoacetate
A proceeded smoothly to afford ester 51 as a 3.5:1 mixture of
E/Z geometrical isomers. The protecting groups were removed
from this mixture (51) by the action of HF‚py in THF to afford
a diol which was selectively oxidized at the benzylic position
with activated MnO2 to give the aldehyde cyclization precursor
52. The remaining free hydroxyl group was protected with
acetate (Ac2O, collidine) and silyl protecting (TBSCl, imid)
groups to deliver substrates 53 and 54, respectively.
With a collection of suitably substituted benzaldehydes in
hand, their photochemical reactions were next investigated.
Thus, with the substrates bearing activated olefins connected
to the aromatic nucleus with a five- or a six-carbon tether (entries
1-13, Table 4), ultraviolet irradiation (450 W Hanovia lamp,
Pyrex filter) at ambient temperature led to a remarkably fast,
stereoselective, and efficient ring closure, furnishing the indi-
cated tricycles 56-68 (Table 4) in high yields.
The examples of Table 4 demonstrate the power of this
method in constructing both [4.4.0] (entries 1-4) and [4.3.0]
(entries 5-13) bicyclic systems which are fused onto aromatic
nuclei. Characteristically, the fusion between the two newly
formed rings is exclusively trans. Furthermore, the substituents
on the three contiguous centers C-9, C-10, and C-11 are all syn
to each other when disubstituted E olefins are employed (entries
1-3 and 5-7, Table 4). The results also demonstrate the
formation of quaternary centers (entries 4, 8, and 13, Table 4).
Limitations of the method include the inability to generate seven-
membered rings (e.g., entry 14, Table 4) even after prolonged
irradiation, and the failure to accommodate unactivated olefins
or alkynes as dienophiles (e.g., entries 15 and 16, Table 4). On
the other hand, the reaction performs admirably well with
substrate 55 (entry 13, Table 4) carrying a tetrasubstituted
activated olefin to afford a product (68) with two adjacent
quaternary centers, one of which carries a methoxy group. One
should also note the tolerance of an acetate or a TBS group
within the substrate (entries 10 and 11, Table 4).
a Reagents and conditions: (a) BF3‚OEt2 (1.0 equiv), CH2(CH2SH)2 (1.2
equiv), CH2Cl2, 0 °C, 1 h, 90-95%; (b) IBX (1.5 equiv), THF-DMSO
(1:1), 2 h, 90-95%; (c) NaH (3.0 equiv), phosphonate reagent A-D (3.0
equiv), THF, 2 h; then 40, 41, or 42 in THF, 25 °C, 2 h, 80-95%; (d)
CH3PPh3Br (3.0 equiv), nBuLi (3.0 equiv), THF, -20 °C, 1 h; then 40, 41,
or 42 in THF, 25 °C, 2 h, 86-93%; (e) K2CO3 (3.0 equiv), phosphonate F
(1.5 equiv), add 40, 41, or 42, MeOH, -30 to 0 °C, 1 h, 90-93%; (f)
PhI(OCOCF3)2 (2.0 equiv) in degassed MeCN-H2O (4:1), 0 °C, 0.5 h,
75-90%. IBX ) 1-hydroxy-1,2-benziodoxol-3(1H)-one 1-oxide.
Scheme 2. Synthesis of Functionalized Benzaldehydes 52-54 for
Photocyclizationa
Scheme 3 shows a possible transition state for the IMPEDA
reaction of 52. Among the successful IMPEDA examples,
entries 9 and 12 (Table 4) shed light on the stereochemical
aspects of this process and the possible transition state which
leads to the observed outcome. Thus, when (E)-52 (E olefin)
was irradiated, the all-syn product 64 (at C-9, C-10, and C-11;
see Scheme 3) was isolated as the major product, with its C-10
epimer [in which the ester group was anti to the neighboring
two substituents (67)] being formed in very small amounts
(>25:1 ratio). An endo approach of the dienophile onto the
reactive diene explains this stereochemical result (transition state,
X ) CO2Et, Y ) H; Scheme 3). To our surprise, when pure Z
olefin (Z)-52 was irradiated under identical conditions, a mixture
of the same products (67:64 ratio ca. 3:1) was isolated, with
stereoisomer 67 (64 inverted at C-10) predominating. Assuming
a concerted mechanism, this observation requires an exo
a Reagents and conditions: (a) (TBS)Cl (3.0 equiv), Et3N (4.5 equiv),
CH2Cl2, 12 h, 94%; (b) Pd(OAc)2 (0.15 equiv), Cu(OAc)2 (2.0 equiv),
DMA-H2O (10:1), O2 (1 atm), 16 h, 88%; (c) NaH (3.0 equiv), A (3.0
equiv), THF, 0-25 °C; then add 50 in THF, 55 °C, 3 h, 89% (3.5:1 E/Z);
(d) HF‚py (6.0 equiv), THF, 25 °C, 40 min, 90%; (e) MnO2 (20 equiv),
CH2Cl2, 2 h, 87%; (f) Ac2O, collidine, 6 h, 74%; (g) (TBS)Cl, imidazole,
CH2Cl2, DMAP(cat.), 12 h, 77%. DMA ) N,N-dimethylacetamide.
rane, or the Bestmann reagent24 (F; Scheme 1) to furnish the
desired olefinic products 43a-f, 44a-f, and 45a-f in good to
excellent yields. The dithiane protecting group was then removed
[PhI(OCOCF3)2, 0 °C, 75-90% yield],25 and the resulting
(24) Mueller, S.; Liepold, B.; Roth, G.; Bestmann, H.-J. Synlett 1996, 6, 521-
522.
(25) Stork, G.; Zhao, K. Tetrahedron Lett. 1989, 30, 287-290.
9
J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004 617

A R T I C L E S
Nicolaou et al.
Table 4. Synthesis of Tricycles 36 from Substrates 34 through Intramolecular Photoenolization/Diels-Alder Cascade via
Hydroxy-o-quinodimethanes 35 (See Figure 3)a
a o-Alkylbenzaldehydes (0.1-0.5 mmol) were dissolved in deoxygenated toluene (0.05 M) in a Pyrex flask and irradiated at ambient temperature (reactions
warmed slightly upon irradiation) with a 450 W Hanovia lamp at a distance of 5 cm. Products were obtained as separable mixtures of isomers, the ratio of
which was related to be the E:Z ratios of the starting olefins. All products were racemic. Starting aldehydes and products shown are the major isomers.
b Starting olefin, E:Z > 25:1; product, C-10 epimers >25:1. c Starting olefin, E:Z > 20:1; product, C-10 epimers ca. 3:1. d Starting olefin, E:Z 1.2:1; product,
C-10 epimers 2.5:1.
approach (transition state, X ) H, Y ) CO2Et; Scheme 3) of
the incoming dienophile onto the hydroxy-o-quinodimethane to
explain the formation of the major product 67 from (Z)-52, and
an isomerization of olefin geometry under the irradiation
conditions to account for the formation of the all-syn isomer
64. In support of this notion is the fact that the two reaction
products 64 and 67 do not interconvert under the reaction
conditions. Similarly, irradiation of both (Z)-51 and (Z)-52 in
d6-benzene followed by H NMR spectroscopic analysis dem-
onstrated that olefin isomerization within the starting materials
1
9
618 J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004

Synthesis of Hamigerans and Analogues Thereof
A R T I C L E S
Scheme 3. Stereochemical Course of the Intramolecular
Photocyclization with a Substituent at C-6 and the Observed
cis/trans Isomerization of Reaction Substrate (Z)-52
Figure 4. Retrosynthetic analysis of the hamigerans 2 and 3 based on the
intramolecular trapping of a hydroxy-o-quinodimethane (69 or 70).
to speculate on the possibility that the entire family of these
natural products might arise biosynthetically via interconversion
of a few members. The required cyclization precursors 71 and
72 (Figure 4) were to be constructed using standard C-C bond
forming reactions. It was anticipated that much of the chemistry
employed for the synthesis of the IMPEDA methodology
substrates would be directly applicable to the hamigeran
synthesis.
a Irradiation of (E)-52 Leads Predominantly to 64 in 94% yield (64:67
> 25:1). Irradiation of (Z)-52 Leads Predominantly to 67 in 89% yield (67:
64 ) 3:1).
does occur to a significant extent on the time scale of these
processes. Many of the substrates used in these IMPEDA
reactions were irradiated as mixtures of E and Z geometrical
isomers, and in most cases the ratios of the two cyclization
products (C-10 epimers) were similar to the ratios of isomers
in the starting materials. In a few cases [importantly those which
took longer to react, including those of entries 12 and 13 (Table
4)], the products (as determined by 1H NMR spectroscopy) were
slightly, but clearly, enriched in the isomer corresponding to
the endo cyclization of the E olefin.
Manual inspection of molecular models confirmed the high
strain associated with both the exo approach for the E isomer
[(E)-52] and the endo arrangement for the Z isomer [(Z)-52],
precluding formation of tricycles with the C6-C10 syn junction,
which would require such modes of reaction. A final element
of stereocontrol is exerted by whatever substituent is placed at
the homobenzylic position (C-6) along the tether (transition state,
Scheme 3). The products 64 and 67 were obtained as single
diastereoisomers with the methyl (C-9) and hydroxymethyl (C-
6) groups anti to one another. The minimization of steric
repulsion between these two groups in the transition state must
be responsible for the formation of a single product. This
reaction feature raised the possibility of placing a stereocenter
at this site (C-6) and using it to template the formation of the
remaining four stereocenters in an IMPEDA reaction which
would lead to enantiomerically enriched products.
An additional cyclization substrate (75; Scheme 4) was
prepared as a model system to explore the IMPEDA approach
to hamigerans. This was to be a demanding test of the IMPEDA
process (vicinal quaternary centers) and an important model
study which would shape the hamigeran synthetic route. The
requisite substrate 75 was prepared as outlined in Scheme 4.
Thus, the previously described ketone 50 (Scheme 2) was added
to a solution of the lithium enolate of methyl R-methoxyacetate
at low temperature,26 leading to the tertiary alcohol 73, obtained
as a mixture of diastereoisomers. This crude mixture was treated
with SOCl2 in pyridine followed by exposure to NaOMe (to
isomerize nonconjugated elimination products), this sequence
affording the tetrasubstituted olefin 74 as an inseparable mixture
(ca. 1.5:1) of isomers. Exposure of the latter (74) to HF‚py
afforded a mixture of diols (not shown, high yield) which was
subjected to the action of activated MnO2 as before to access
the targeted intermediate 75 (a ca. 1.5:1 mixture of E and Z
isomers, 79% for two steps, not separated). Irradiation of this
mixture provided a ca. 1.5:1 mixture of racemic photocyclization
products 76 and 68 (favoring 68) in 76% yield.
At first glance, this was an excellent result, but, upon closer
inspection of the reaction products, a problem was uncovered.
The surprising outcome of this study was the finding that the
major IMPEDA product was 68 with the incorrect stereochem-
istry at C-10 (and also at C-5 as expected) irrespective of the
geometrical ratio of the starting materials (i.e., 75). Despite
numerous attempts to separate the E and Z isomers of the starting
aldehydes and irradiate them separately, the 1.5:1 ratio between
incorrect and correct C-10 isomers in the product was relatively
constant even when the cyclization precursor 75 was enriched
Initial Forays toward the Total Synthesis of the Hamig-
erans. In contemplating potential synthetic routes to the
hamigerans, we focused on utilizing the intramolecular trapping
of photochemically generated hydroxy-o-quinodimethane spe-
cies as shown retrosynthetically in Figure 4. The methodological
developments described above set a strong foundation for this
retrosynthetic blueprint. Thus, the hamigeran B series was
envisioned to derive from hamigeran A by a decarboxylation/
oxidation process (2 f 3, Figure 4). Incidentally, it is interesting
(26) Hiersemann, M.; Lauterbach, C.; Pollex, A. Eur. J. Org. Chem. 1999, 11,
2713-2724.
9
J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004 619

A R T I C L E S
Nicolaou et al.
Scheme 4. Synthesis of Substituted Benzaldehydes 75 (E and Z
Isomers) and Their Photocyclization to Tricycles 68 and 76a
the expected incorrect stereochemistry at the benzylic position
(C-5) was considered to be correctable via epimerization due
to its potential to be activated under basic, radical, or acidic
conditions, coupled with the usual preference for hydrindanes
to adopt a cis ring junction. The designed substrates 71 and 72
(Figure 4) were destined to be derived from a common
intermediate (i.e., 80) as shown in Scheme 5. For the isopropyl
substrate (i.e., 71), the commercially available racemic epoxide
(()-78 was utilized as a suitable precursor, whereas, for the
protected alcohol substrate (i.e., 72), enantiomerically enriched
(>99% ee) epoxide (S)-78 (obtained via the Jacobsen hydrolytic
kinetic resolution method employing the (S,S)-cobalt catalyst)27
was required (vide infra). Thus, to a solution of tert-butyl amide
77 (Scheme 5) was added 2 equiv of tBuLi (-78 f -20 °C)
to form the deep-red dianion which opened terminal epoxide
(()-78 or (S)-78 at -78 °C, leading to a secondary alcohol in
69% yield (Scheme 5).28 Following crude chromatographic
purification, this product was heated in benzene in the presence
of p-TsOH, causing expulsion of the amide nitrogen moiety by
the free secondary hydroxyl group and closure to the δ-lactone
(()-79 or 79 in 91% yield. The latter compound [(()-79 or
79] was then reduced with LAH, leading to the corresponding
diol, which was selectively monosilylated [(TBS)Cl-Et3N, 89%
yield] to afford the desired common intermediate (()-80 or (-)-
80. For the series bearing the isopropyl group prior to photo-
cyclization, the more accessible racemic compound was ex-
ploited since the integrity of the stereocenter would be lost in
the oxidation step. Thus, the free alcohol within (()-80 was
oxidized (SO3‚py-DMSO, 94% yield) to afford ketone 81,
which served as a substrate for the installment of the isopropyl
group. The addition of iPrMgCl to ketone 81 did not occur
without prior addition of CeCl3 to the Grignard reagent to
generate the less basic cerium species,29 which smoothly reacted
with the substrate to afford, in 94% yield, tertiary alcohol 82.
Elimination of the tertiary hydroxyl group from the latter
compound (82) was accomplished in a regioselective fashion
by the action of SOCl2-py at -50 °C, leading primarily to a
single diene (not shown, 80% yield, >10:1 ratio with the
alternate geometrical olefin isomer), which was subjected to
Wacker oxidation to afford the expected methyl ketone 83 in
81% yield. Hydrogenation (H2, 10% Pd/C) of the remaining
double bond in the latter compound proceeded smoothly (>95%
yield), and the material thus obtained was homologated employ-
ing standard conditions [(MeO)2P(O)CH2CO2Me-NaH]. This
treatment afforded ester 84 as a ca. 3.5:1 mixture of olefinic
isomers (94% combined yield). Desilylation of this substrate
(84) was accomplished by careful treatment with HF‚py in THF
(25 °C, 40 min). The final operation in accessing the targeted
photocyclization precursor 71 (Scheme 5) was a facile benzylic
oxidation initiated by exposure of the latter compound (85) to
SO3‚py-DMSO (high yield). Pleasantly, the planned photocy-
clization occurred as expected, and the tricyclic system 86 was
obtained in 91% yield when benzaldehyde 71 (Scheme 5) was
irradiated in deoxygenated benzene solution (450 W Hanovia
lamp, Pyrex filter). All stereocenters were formed as single
a Reagents and conditions: (a) MeOCH2CO2Me (9.0 equiv), THF, -78
°C; then add LDA (8.0 equiv) over 2 h; then add 50 in THF over 2 h, -78
to 0 °C, 2 h, 90%, mixture of isomers; (b) CH2Cl2-py (3:1), -50 °C; then
add SOCl2 (10.0 equiv), 15 min, -50 to -20 °C, 2 h, 90%, mixture of
isomers; (c) NaOMe (2.0 equiv) in MeOH, 50 °C, 4 h, 85% (1:1.5 E/Z
mixture, unassigned); (d) HF‚py (2.0 equiv), THF, 25 °C, 40 min.; (e) MnO2
(20 equiv), CH2Cl2, 25 °C, 2 h, 79% for two steps; (f) hν, 450 W Hanovia
lamp, Pyrex flask, toluene, ambient temperature, 1 h, 76%. LDA ) lithium
diisopropylamide.
in one or the other geometrical isomer. The outcome of these
experiments might be explained by E/Z isomerization of the
starting material in a fashion similar to what was observed for
(Z)-52 (Scheme 3). This proposal, however, remains unsub-
stantiated because we were unable to cleanly separate the
isomers of 75 (Scheme 4), and because an additional pathway
for C-10 epimerization was subsequently discovered (vide infra).
Despite the failure of the first model study to deliver the
desired product, the attractiveness of the IMPEDA methodology
was too good to abandon, particularly as it appeared ideal for
the benzannulated carbocyclic framework found within the
hamigeran structures (e.g., 1-4; Figure 1). The looming obstacle
to reaching the hamigerans appeared to be the correct setting
of the relative stereochemistry at the contiguous stereocenters
situated at C-6, C-5, C-9, and C-10. In a new attempt to solve
this problem, two cyclization precursors were designed, one in
which the isopropyl group at the C-6 position of the targeted
structure 71 (Figure 4) was already installed, and another in
which this position was occupied by a protected oxygen
functionality (72; Figure 4) which could serve as a handle to
introduce the obligatory isopropyl group after cyclization, the
latter functionality providing the flexibility for epimerization
at C-5 at some later stage. Priority was given to the C-5 position
in light of the results in Table 2 (all compounds have the
undesired trans C5-C9 junction). Establishing early on the trans
relationship between the C-6 isopropyl group and the C-9 methyl
substituent was a necessity because these two positions appeared
to be inert (toward a possible subsequent correction). In contrast,
(27) O’Neil, I.; Cleator, E.; Southern, J.; Hone, N.; Tapolczay, D. Synlett 2000,
5, 695-697.
(28) Watanabe, M.; Sahara, M.; Furukawa, S.; Bilodeau, R.; Snieckus, V.
Tetrahedron Lett. 1982, 23, 1647-1650.
(29) (a) Imamoto, T.; Takiyama, N.; Nakamura, K.; Hatajima, T.; Kamiya, Y.
J. Am. Chem. Soc. 1989, 111, 4392-4398. (b) Takeda, N.; Imamoto, T.
Org. Synth. 1999, 76, 228-238.
9
620 J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004

Synthesis of Hamigerans and Analogues Thereof
A R T I C L E S
Scheme 6. Synthesis of the 5-epi-Hamigerans 88-93a
Scheme 5. Synthesis of Racemic Aldehyde 71 and Its
Photocyclization to 86a
a Reagents and conditions: (a) 1% HCl in MeOH, 25 °C, 0.5 h, 90%;
(b) OsO4 (0.1 equiv), NMO (3.0 equiv), THF-tBuOH-H2O-py (20:20:
4:1), 12 h (a ca. 12:1 mixture of isomers), 92%; (c) SO3‚py (3.0 equiv),
Et3N (6.0 equiv), DMSO-CH2Cl2 (1:1), 0 °C, 2 h, 88%; (d) BBr3 (10.0
equiv), CH2Cl2, -78 °C, 3 h, 96%; (e) NBS (1.05 equiv), iPr2NH (0.1
equiv), CH2Cl2, 0 °C, 3 h, 90%; (f) (i) KOH, MeOH, 70 °C, 2 h; (ii)
nBu4NIO4 (2.0 equiv), dioxane, 100 °C, 1 h, 65%; (g) NBS (3.0 equiv),
DMF, 25 °C, 3 h, 92%. NMO ) 4-methylmorpholine N-oxide, and NBS
) N-bromosuccinimide.
operation was the acid-induced (HCl-MeOH) elimination of
water from 86 (Scheme 6), leading to the corresponding
conjugated olefin (95% yield), which was then subjected to
dihydroxylation with NMO-OsO4(cat.) in the presence of
pyridine to afford, stereoselectively, diol 87 in 92% yield (ca.
12:1 diastereoselectivity). This sequence makes it obvious that
the stereochemistries at C-10 and C-11 in intermediate 86 were
inconsequential since both diastereoisomers led smoothly to the
same intermediate olefin. At this stage the benzylic position
within 87 was oxidized with SO3‚py-DMSO, leading to the
expected hydroxy keto ester 88 (88% yield) whose methoxy
group was smoothly cleaved through the action of BBr3 at -78
°C to afford, in 96% yield, 5-epi-debromohamigeran 89.
Attempts to epimerize compound 89 at C-5 as planned under a
variety of conditions failed, however, as did experiments
intended to bring about the same result with some of its
precursors (i.e., the olefinic substrate between 86 and 87 and
the keto methoxy derivative 88). With an efficient route to these
compounds, and with an eye for future chemical biology studies,
we proceeded to synthesize a number of hamigeran derivatives,
albeit epimeric at C-5 (Scheme 6). Thus, selective monobro-
mination ortho to the phenolic group in 89 was smoothly
effected by following the conditions described by Krohn,30
which entail the utilization of stoichiometric amounts of NBS
and catalytic quantities of iPr2NH in CH2Cl2 at 0 °C. This highly
regioselective bromination furnished 5-epi-hamigeran A (90)
in 90% yield. Remarkably, omission of the base (iPr2NH, 5-10
mol %) in this procedure led to a completely random bromi-
nation of the aryl nucleus in 89. Additional members of the
hamigeran family were obtained upon exposure of 88 to base
a Reagents and conditions: (a) tBuLi (2.2 equiv), TMEDA (2.0 equiv),
-78 to -20 °C; then (()-75 or (S)-75 (1.0 equiv), THF, -78 to 0 °C, 2 h,
69%; (b) p-TsOH (2.0 equiv), toluene, reflux, 2 h, 91%; (c) LiAlH4 (2.0
equiv), THF, 25 °C, 0.5 h, 91%; (d) (TBS)Cl (1.1 equiv), Et3N (2.0 equiv),
12 h, 89%; (e) SO3‚py (3.0 equiv), Et3N (6.0 equiv), DMSO-CH2Cl2 (1:
1), 0 °C, 2 h, 94%; (f) iPrMgCl (2.0 equiv), CeCl3 (2.0 equiv), -78 to 0
°C, 1 h, 94%; (g) CH2Cl2-py (3:1), -50 °C; then add SOCl2 (10.0 equiv),
-50 to -20 °C, 0.5 h, 80%; (h) Pd(OAc)2 (0.1 equiv), Cu(OAc)2 (2.0 equiv),
DMA-H2O (10:1), O2 (1 atm), 16 h, 81%; (i) 10% Pd/C, H2 (1 atm),
NaHCO3 (solid, 5.0 equiv), EtOAc, 2 h, 95%; (j) (MeO)2P(O)CH2CO2Me
(3.0 equiv), NaH (3.0 equiv), THF, 60 °C, 3 h, 94% (mixture of E/Z isomers,
ca. 3.5:1); (k) HF‚py (2.0 equiv), THF, 25 °C, 10 min, 91%; (l) SO3‚py
(3.0 equiv), Et3N (6.0 equiv), DMSO-CH2Cl2 (1:1), 0 °C, 2 h, 88%; (m)
hν, 450 W Hanovia lamp, Pyrex filter, benzene, 20 min, 91%. MOM )
methoxymethyl, and H-W-E ) Horner-Wadsworth-Emmons reaction.
isomers except for the expected mixture at C-10 (ca. 3.5:1 ratio),
which arises as a consequence of the E/Z mixture in the starting
material 71.
With the entire hamigeran ring framework in place in
compound 86, we were now in a position to address the issue
of C-5 stereochemistry and the final functional group manipula-
tions required to arrive at the targeted structures. The next
(30) Krohn, K.; Bernhard, S.; Floerke, U.; Hayat, N. J. Org. Chem. 2000, 65,
3218-3222.
9
J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004 621

A R T I C L E S
Nicolaou et al.
Scheme 7. Postulated Mechanism for the Photochemically
Induced Isomerization (Inversion at C-10) of Compounds 88 and
95
Figure 5. Relative strain energies of 6,9-cis and 6,9-trans hamigeran-type
structures 88 and 96-98. See ref 34 for computational parameters.
positioned very near the aromatic ring in computationally
minimized structures. This alignment is supported by 1H NMR
spectroscopic data recorded for 88 in which the two methyl
groups are in significantly different chemical environments
(chemical shifts at 1.02 and 0.74 ppm). The close interactions
of the isopropyl group with the tricyclic portion of the molecule
are relieved in the 5-epi compounds, which may account for
the trans-hydrindane being uncharacteristically more stable (by
computation) than the natural cis isomer (88 vs 96; Figure 5).
We noted that replacing the isopropyl-group-bearing sp3 carbon
with an sp2 center (such as a carbonyl, 97 and 98; Figure 5) in
the modeling studies had a rather drastic effect on the energy
difference between the cis and trans isomers, with the cis isomer
now being highly favored (97 vs 98; Figure 5). In light of these
computational studies and experimental results, we embarked
on a second approach to the hamigerans which would proceed
through key intermediate 72 (Scheme 8).
The synthesis of the required compound 72 commenced from
the enantiomerically enriched alcohol (-)-80 (ee > 99%) as
shown in Scheme 8. Thus, protection of (-)-80 as its MOM
ether [(MOM)Cl, EtNiPr2, 83% yield] led to derivative 99,
whose Wacker oxidation [Pd(OAc)2, Cu(OAc)2, H2O, O2]
furnished methyl ketone 100 as the major product (81% yield)
together with small amounts of the corresponding aldehyde
(9%). Extension of the chain within 100 was carried out as
before (see Scheme 5), furnishing the E-R,â-unsaturated ester
101 as the major product, together with its Z geometrical isomer
in a ca. 3.5:1 ratio and in 94% combined yield. Desilylation of
101 proved problematic at first, since the olefinic bond was
found to rapidly deconjugate to its â,γ-unsaturated isomer 103
following removal of the protecting group. The nonconjugated
olefin would not return to conjugation under any of the several
basic or acidic conditions screened. This side reaction, which
was observed to a lesser extent in the isopropyl counterpart of
101, was finally overcome by employing HF‚py in THF at
ambient temperature (20 min) for the desilylation step, leading
to the desired benzylic alcohol 102 in 91% yield. Oxidation of
102 was cleanly accomplished by exposure to the SO3‚py-
DMSO protocol (92% yield), marking our arrival at the second
photocyclization precursor 72. And, gratifyingly, irradiation of
this substrate under the developed conditions gave cleanly the
expected mixture of C-10 tricyclic epimers in high yield. The
crude product so obtained was then treated with 1% methanolic
(KOH-MeOH) followed by treatment with nBu4NIO4 in
dioxane at reflux,31 a sequence that led to diketone 91 in 65%
overall yield from 88. Cleavage of the methyl ether (BBr3, -78
°C) afforded 92, which was subjected to aromatic bisbromination
by employing an excess of NBS in DMF to provide 5-epi-4-
bromohamigeran B (93) in 94% yield.
Among conditions which were employed in our attempts to
invert the C-5 stereochemistry, the irradiation of hydroxy keto
ester 88 is of special interest. Thus, upon exposure to light from
a UV Hanovia lamp in benzene solution, 88 was converted to
an equilibrium mixture of C-10 epimers (88:95 ratio ca. 1:3;
see Scheme 7). A likely reaction course for this equilibration
may involve a Norrish type I homolysis32,33 of the C10-C11 bond
to form a fleeting diradical species (94) which can either reclose
back to 88 or invert its stereochemistry at the radical center
prior to recombination, which would lead to the formation of
the C-10 epimer 95 (Scheme 7). Notably, no epimerization was
observed upon irradiation of compound 89 (Scheme 6), which
has a free phenol. If the proposed mechanism is correct, the
ketone chromophore plays a central role in this rather unusual
but potentially useful transformation.
At this stage, we performed molecular modeling and com-
putational studies which indicated that epimerization of com-
pounds such as 88 (Figure 5) would not be productive.34 At the
same time, these studies suggested that our alternate strategy
(delay of isopropyl group installation) could be successful. Thus,
it appears that the isopropyl moiety resides in an extremely
hindered position. In particular, one of its methyl groups is
(31) Santaniello, E.; Ponti, F.; Manzocchi, A. Tetrahedron Lett. 1980, 21, 2655-
2656.
(32) Barltrop, J.; Coyle, J. J. Chem. Soc. D 1969, 19, 1081-1082.
(33) For recent computational studies on the Norrish type I reaction and further
references, see: (a) Diau, E.; Kotting, C.; Zewail, A. ChemPhysChem 2001,
2, 273-293. (b) Diau, E.; Kotting, C.; Zewail, A. ChemPhysChem 2001,
2, 294-309.
(34) The hamigeran diastereoisomeric pairs were modeled with Accelrys Insight
II, 1998. Molecular dynamics calculations were performed with Accelrys
Discover v. 2.98 on a cluster of SGI Origin servers running Irix 6.5. The
relative energies of the diastereoisomeric were computed with the class II
pairwise force field cff91 v. 2.0 (see: Dinur, U.; Hagler A. In ReViews in
Computational Chemistry; Lipkowitz, K., Boyd, D., Eds.; VCH: New York,
1991; Vol. 2, p 527), with library parameters for bond, valence, torsion,
and out-of-plane terms. A total of 1000 structures were generated within
300 ps at 800 K. They were annealed to 300 K in 5 ps. The structures
were further minimized to convergence using
a conjugate gradient
algorithm, and the lowest energy structure cluster was selected as the
preferred structure. We thank Dr. C. N. C. Boddy for performing these
calculations.
9
622 J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004

Synthesis of Hamigerans and Analogues Thereof
A R T I C L E S
Scheme 9. Correction of the Stereochemistry at C-5 via
Scheme 8. Synthesis and Photocyclization of Precursor 72a
Base-Induced Epimerization and Elaboration toward the
Hamigerans 1-4a
a Reagents and conditions: (a) (MOM)Cl (2.0 equiv), iPr2NEt (6.0 equiv),
CH2Cl2, 25 °C, 12 h, 83%; (b) Pd(OAc)2 (0.1 equiv), Cu(OAc)2 (2.0 equiv),
DMA-H2O (10:1), O2 (1 atm), 16 h, 81%; (c) (MeO)2P(o)CH2CO2Me (3.0
equiv), NaH (3.0 equiv), THF, 60 °C, 3 h, 94% (mixture of E/Z isomers,
ca. 3.5:1); (d) HF‚py (2.0 equiv), THF, 25 °C, 20 min, 91%; (e) SO3‚py
(3.0 equiv), Et3N (6.0 equiv), DMSO-CH2Cl2 (1:1), 0 °C, 2 h, 92%; (f)
hν, 450 W Hanovia lamp, Pyrex filter, benzene, 25 min; then 1% anhydrous
HCl in MeOH, 60 °C, 1 h, 85%.
HCl at 60 °C to furnish the desired hydroxyolefin 104 by
concomitant elimination of water and cleavage of the MOM
group, in 85% overall yield from 72.
Our next attempt at the total synthesis of the hamigerans
utilized the alternate tricycle 104 (Scheme 9), which incorpo-
rated an oxygen functionality at C-6, in the hope that this
position could be manipulated to give the necessary activation
for the required C-5 epimerization. Although less direct, this
strategy had the potential to allow for an enantioselective
synthesis of the targeted compounds since an enantiopure
starting material was employed to synthesize 104. The efficiency
of the synthetic sequence leading to compound 104 facilitated
its multigram production with over 99% enantiomeric purity
as determined by chiral HPLC. Borrowing from the successful
advancement of the isopropyl substrate (see Scheme 6), hy-
droxyolefin 104 was dihydroxylated with NMO and catalytic
OsO4, giving triol 105 (94% yield) as shown in Scheme 9.
Again, the facial selectivity in this reaction was excellent (ca.
12:1 in favor of 105), presumably controlled by an approach in
which minimum interactions between the reagent and the
angular methyl group at the ring junction were encountered.
Selective protection of the vicinal hydroxyl groups in 105 was
achieved with the combination of 2-methoxypropene and
catalytic amounts of PPTS, a reaction that was followed by
exposure of the crude product to p-TsOH in MeOH (to cleave
the somewhat stable hemiketal formed at the C-6 hydroxyl
a Reagents and conditions: (a) OsO4 (0.1 equiv), NMO (3.0 equiv),
THF-tBuOH-H2O-py (20:20:4:1), 12 h, 94% (ca. 12:1 diastereoselec-
tivity); (b) (i) 2-methoxypropene (20 equiv), PPTS (0.3 equiv), CH2Cl2, 0
°C, 0.5 h; (ii) p-TsOH (1.0 equiv), MeOH, 0 °C, 0.5 h, 93%; (c) (i) DMP
(1.7 equiv), CH2Cl2, 0 °C, 1 h; (ii) DBU (0.5 equiv), CH2Cl2, 0 °C, 10
min, 93% for two steps; (d) iPrMgCl (2.0 equiv), CeCl3 (2.0 equiv), -78
to 0 °C, 1 h, 95%; (e) Et3SiH (50 equiv), TFA (20 equiv), CH2Cl2, 25 °C,
1 h, 65%; (f) SOCl2 (6.0 equiv) py-lutidine-CH2Cl2 (1:5:5), -50 to -20
°C, 2 h, 94%; (g) H2, catalyst; see Table 5 for conditions, yields, and product
distributions. DMP ) Dess-Martin periodinane, PPTS ) pyridinium
p-toluenesulfonic acid, and DBU ) 1,8-diazabicyclo[5.4.0]undec-7-ene.
1
b Product distributions determined by H NMR spectroscopy.
group), leading to acetonide 106 (93% yield for two steps).
Dess-Martin oxidation of the remaining hydroxy group in 106
then gave the corresponding ketone acetonide 98 (see Figure
5) in high yield. With the ketone at the homobenzylic position
(C-6), base-induced epimerization at C-5 became facile, requir-
9
J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004 623

A R T I C L E S
Nicolaou et al.
Table 5. Attempted exo Reduction of Trisubstituted Olefins 109-111
product distribution by 1H NMR
spectroscopic analysis (%)
entry
conditions
110
111
endo-112
exo-113
1
2
3
4
5
6
PtO2, 3 atm of H2, EtOAc, 6 h
Pd(OH)2, 3 atm of H2, EtOH, 12 h
10% Pd/C, 50 atm of H2, EtOAc, 6 h
Rh-Al2O3, 50 atm of H2, EtOH, 48 h
Rh black, 20 atm of H2, EtOH, 48 h
IrP(cyhex)3(COD)(py)PF6, 10 atm of H2, CH2Cl2, 48 h
6
24
40
7
3
10
7
71
50
33
6
20
15
19
21
3
negligible conversion to products
negligible conversion to products
a A mixture of olefins (109-111, 0.05 mmol) was dissolved in the indicated solvent and stirred under H2 pressure (Parr bomb) for the indicated times,
1
at which time the catalyst was removed by filtration and product ratios were determined by H NMR spectroscopic analysis.
ing only a brief exposure to DBU at 0° C to afford, in 93%
yield for two steps, the desired product 97 with the cis
[C5-C9] junction. Addition of the reagent formed by mixing
iPrMgCl and CeCl3 to ketone 97 proceeded smoothly and
stereoselectively to afford tertiary alcohol 107 as a single product
(95% yield) through convex (exo) face attack.
of the bicycle is suprising. The entries in Table 5 reveal the
unusual reluctance toward hydrogenation that these molecules
display, resisting even elevated pressures of hydrogen and highly
active catalyst systems. These observations, when taken together
with the completely exo-selective alkylation of ketone 97 with
the bulky isopropyl nucleophile, point to the isopropyl group
itself as the dominant stereochemistry-directing force around
the C-6 atom, overriding any effects inherent within the fused
ring system. The undesired product 112 was isolated together
with unreactive (and inseparable) tetrasubstituted olefinic
isomers 110 and 111 as well as small amounts of the desired
reduction product 113. The observed distribution of unreacted
olefins 110 and 111 was, in some experiments (entries 2 and 3,
Table 5), higher than in the starting mixture, suggesting that
olefin isomerization was a preferred reaction course under some
of the conditions employed. Having reached that far into the
sequence, it was decided to complete it with the C-6 epimeric
(major isomer) intermediate 112, with the aim of validating the
proposed route to the hamigerans and to obtain analogues of it.
Scheme 10 summarizes these efforts, which culminated in
the synthesis of four 6-epi-hamigeran analogues, 117-120. A
mixture of 110, 111, 112, and 113 (ca. 6:3:71:20) obtained from
reduction of 109-111 according to entry 1 (Table 5; PtO2, H2)
was subjected to the action of 3 M aqueous HCl in THF (1:1)
at 80 °C, conditions which not only cleaved the acetonide group
but also randomized the benzylic hydroxyl group, complicating
the mixture even further. Oxidation of this mixture (SO3‚py-
DMSO), however, resulted in a new mixture from which ketone
116 was chromatographically separated in 45% overall yield
from the mixture of 109-111. Unfortunately, we were unable
to isolate any material possessing the correct C-6 stereochemistry
at this stage, despite the presence of that isomer (113) after the
initial hydrogenation step. Demethylation of the phenolic group
within 116 with BBr3 at -78 °C proceeded smoothly, forming
Having installed the entire carbon skeleton of the hamigerans
within 107 (see Scheme 9), ways were then sought to excise
the extra hydroxyl group at C-6. An attempt to reductively
remove the free hydroxyl from 107 by reaction with TFA-Et3-
SiH failed, leading instead to the novel polycyclic ether 108
(65% yield), presumably via acetonide collapse, followed by
concomitant trapping of the incipient tertiary carbonium ion by
the resulting proximal C-10 hydroxyl group and reductive
cleavage of the benzylic C-O bond. The steric hindrance around
C-6 is probably responsible for the recalcitrance at this position
toward reduction. Nevertheless, conditions were eventually
found for the regioselective elimination of this hydroxyl group.
Thus, exposure of 107 to SOCl2-lutidine-pyridine in CH2Cl2
at -50 to -20 °C resulted in the formation of the trisubstituted
olefin 109 as the major product (77% yield) together with small
amounts of the tetrasubstituted olefins 110 (11%) and 111 (6%).
However, the intended reduction of this olefinic mixture to
furnish the desired hamigeran stereochemistry at C-6 as expected
from an exo face attack [cf. the addition of an isopropyl group
to the carbonyl group at C-6 in 97 (97 f 107)] proved elusive.
Thus, upon hydrogenation, the major product obtained (112)
exhibited consistently the wrong orientation of the isopropyl
moiety, despite the many catalysts and conditions employed (see
Table 5). Examination of manual and computational models of
109 raises the possibility that one of the methyl groups from
the isopropyl moiety is responsible for this unexpected result
as demonstrated by structure 114 (Scheme 9). Nevertheles, the
strong preference for hydrogen to be delivered to the endo face
9
624 J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004

Synthesis of Hamigerans and Analogues Thereof
A R T I C L E S
Scheme 10. Completion of the 6-epi-Hamigerans 116-120a
Scheme 11. Total Synthesis of 2 and 3a
a Reagents and conditions: (a) 3 M aqueous HCl-THF (1:1), 80 °C, 4
h, 70%; (b) SO3‚py (3.0 equiv), Et3N (6.0 equiv), DMSO-CH2Cl2 (1:1), 0
°C, 2 h, 93%; (c) BBr3 (10.0 equiv), CH2Cl2, -78 °C, 3 h, 93%; (d) NBS
(1.05 equiv), iPr2NH (0.1 equiv), CH2Cl2, 0 °C, 3 h, 94%; (e) (i) aqueous
KOH, MeOH, 70 °C, 2 h; (ii) nBu4NIO4 (2.0 equiv), dioxane, 100 °C, 1 h,
65%; (f) NBS (3.0 equiv), DMF, 25 °C, 1 h, 93%.
117, whose spectroscopic analysis (NOEs) established its
identity as 6-epi-debromohamigeran A. This compound was
converted to 6-epi-hamigeran A (118; 94%) by exposure to
stoichiometric amounts of NBS in the presence of catalytic
quantities of iPr2NH. As with the 5-epi-hamigeran series, 6-epi-
4-bromohamigeran B (120) was accessed by ester hydrolysis,
oxidative decarboxylation, and finally dibromination (NBS) of
119 in 60% overall yield as shown in Scheme 10.
The Final Drive to the Hamigerans. The failure of the last
plan to reach the hamigerans sent us on a new search for a viable
sequence toward their total synthesis. Our new strategy began
with intermediate 109 and proceeded along the lines delineated
in Scheme 11. The decisive observation was made when olefin
109 was subjected to hydroboration with BH3‚Me2S under
sonication conditions, leading, upon oxidative workup, to the
desired 6(S),7(R)-alcohol 121 as the major product (45% yield)
arising from exo addition, together with its 6(R),7(S)-stereo-
isomer (23% yield, endo addition). The reasoning behind this
attempt was predicated on the expectation that hydroboration
would not be overly sensitive to the steric shielding of the
isopropyl group at C-6, and that, rather, diastereoselection would
be based upon the steric environment around C-7. The chro-
matographically separated isomer 121 was then expeditiously
converted to the desired compound 113 by sequential reaction
with PhOC(S)Cl-py (85% yield) and nBu3SnH-AIBN (75%
yield) via the intermediacy of the corresponding thionocarbonate.
The arrival at the latter intermediate (113) paved the way for
the completion of the synthesis of all four targeted natural
products 1-4. Thus, removal of the acetonide group from 113
was accomplished by heating at 80 °C with 1 M aqueous HCl
in THF (1:1), affording the corresponding diol 128 (Scheme
14, 88% yield), which was oxidized with PDC to furnish the
desired hydroxy ketone 96 in 83% yield. It is interesting to note
a Reagents and conditions: (a) BH3‚Me2S (40 equiv), THF, sonication,
40 °C, 8 h, 68% (a ca. 2:1 mixture of two isomers favoring 121); (b) (i)
PhOC(S)Cl (2.0 equiv), py, 25 °C, 2 h; (ii) nBu3SnH (8.0 equiv), AIBN
(0.2 equiv), benzene, reflux, 2 h, 64% (two steps); (c) 1 M aqueous HCl-
THF (1:1), 80 °C, 1 h, 88%; (d) PDC (2.5 equiv), 4 Å molecular sieves,
CH2Cl2, 3 h, 83%; (e) BBr3 (10.0 equiv), CH2Cl2, -78 °C, 3 h, 94%; (f)
NBS (1.05 equiv), iPr2NH (0.1 equiv), CH2Cl2, 0 °C, 3 h, 95%; (g) Ba(OH)2
(15.0 equiv), MeOH-H2O (2:1), air, 25 °C, 2 h, 87%; (h) NBS (3.0 equiv),
DMF, 25 °C, 1 h, 94%. AIBN ) azobisisobutyronitrile, and PDC )
pyridinium dichromate.
that the use of SO3‚py-DMSO (as in the case of 116, Scheme
10) in this last step failed to deliver the desired ketone, an
observation providing a possible explanation as to why we did
not detect any of the correct isomer in the conversion of the
mixture containing 112 to 116 (see Scheme 10). Compound 96
served admirably to deliver first (-)-debromohamigeran A (1)
(upon exposure to BBr3, 94% yield) and (-)-hamigeran A (2)
(upon bromination of 1 with NBS-iPr2NH, 95% yield).
Although we had previously employed a successful sequence
to convert members of the hamigeran A series to their hamigeran
B counterparts (see 88 f 91, Scheme 6), for the natural series
we devised a more efficient approach based on a cascade
reaction initiated by Ba(OH)2 in MeOH-H2O (2:1) under
9
J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004 625

A R T I C L E S
Nicolaou et al.
Scheme 12. Oxidative Cleavage of Hydroxy Ketone 96 to
Diketone 122 and Ring-Contracted Ketone 123a
Scheme 13. Proposed Mechanism for the Decarbonylative Ring
Contraction of 122 to 123
Scheme 14. Synthesis of 7 and Related Analogues 129 and 132a
a Reagents and conditions: (a) aqueous KOH, MeOH, 70 °C, 2 h; (b)
nBu4N+IO4 (2.0 equiv), dioxane, 100 °C, 1 h, varied yields (122, 10-
-
50%, + 123, 10-40%); (c) BBr3 (10.0 equiv), CH2Cl2, -78 °C, 3 h, 86%.
aerobic conditions. In this novel cascade, initial saponification
of the methyl ester (the generated carboxylic acid can be
detected by TLC) is rapidly followed by a decarboxylation event
and, finally, autoxidation, with the net result of the keto hydroxy
ester of 2 being converted to the diketone functionality of 3 in
87% overall yield. Careful investigation of this reaction revealed
the following aspects: (a) oxygen gas is necessary for good
conversion; (b) the free phenolic group is essential for clean,
room-temperature reaction; (c) the nature of the base is important
to some extent. Finally, 4 was generated from 3 by NBS-
facilitated introduction of the required second bromine atom
(94% yield). The spectroscopic data for all four synthetic
hamigerans 1-4 matched those reported for the natural sub-
stances, and so did their optical rotations, confirming both their
absolute structures and enantiopurities.4
In our initial attack on the hamigeran B series, we utilized
the periodate-based oxidative cleavage protocol, which gave
considerable amounts of a side product which was characterized
as the intriguing ring-contracted hamigeran analogue 123
(Scheme 12). Careful monitoring of the reactions involved
revealed that compound 122 was formed initially and then
transformed slowly into 123, although the yields were low and
rather erratic. It was, however, soon realized that this ring
contraction was a photo-induced process, proceeding under UV
irradiation consistently and in high yield (122 f 123, 87% yield)
as shown mechanistically in Scheme 13. This process is
presumed to proceed via a Norrish type I fragmentation to afford
diradical 126, which, upon expulsion of a molecule of carbon
monoxide followed by intramolecular recombination, furnished
ring-contracted product 123 via diradical 127. Interestingly,
debromohamigeran B (124), in which the methyl protecting
group on the phenol has been removed, fails to undergo this
fragmentation reaction, being stable under the irradiation
conditions. Hydrogen bonding between the phenolic OH and
the proximal carbonyl group may be responsible for the
a Reagents and conditions: (a) MnO2 (20 equiv), CH2Cl2, 25 °C, 2 h,
90%; (b) (MOM)Cl (2.5 equiv), iPr2NEt (3.5 equiv), CH2Cl2, 0 °C, 0.5 h,
76%, (c) 30% aqueous H2O2-dioxane-2 M aqueous NaOH (1:8:2), 0 °C,
10 min, 70%; (d) 3 M aqueous HCl-THF (1:1), 25 °C, 3 h, 70%.
deactivation of the 1,2-diketone system toward this photolytic
rupture.
With the hamigerans A (1 and 2) and B (3 and 4) at hand,
we then turned our attention to other members of the hamigeran
family (Scheme 14). Thus, when vicinal diol 128 was exposed
to activated MnO2, a clean oxidative cleavage was observed,
leading to keto aldehyde 129 (90% yield). This intermediate
may lead to hamigeran C (5; see Figure 1), although this goal
was not pursued at this juncture. Instead, we targeted 7 as shown
in Scheme 14. Thus, MOM protection of 3 with (MOM)Cl and
EtNiPr2 led to diketone 130 in 76% yield. The latter compound
(130) was subjected to oxidative cleavage with basic hydrogen
peroxide in a biphasic medium (2 M aqueous NaOH-H2O2 in
9
626 J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004

Synthesis of Hamigerans and Analogues Thereof
A R T I C L E S
dioxane, 10 min, 0 °C), and the crude reaction mixture was
acidified (3 M aqueous HCl) prior to chromatographic purifica-
tion on acid-washed silica gel, furnishing pure 7 via its MOM
derivative 131 in 50% overall yield from 130. The same
sequence of reactions was applied in the conversion of 93 to
5-epi-4-bromohamigeran E (132) in similar overall yield. The
1H NMR signals in the spectra of both compounds 7 and 132
were broad, presumably due to restricted rotation around the
bond linking the two rings, a phenomenon that was somewhat
suppressed by using wet CD3CN as solvent and running the
spectra at 70 °C. Even under those conditions, however, the
signals were still noticeably broad, and several of the quaternary
carbons were not observable in the 13C NMR spectrum. Despite
this, 7 and 132 were fully characterized as pure compounds,
confirming their structures. It should be noted that, even though
compound 7 was identified as a natural product,4a this is the
first time it was isolated in pure form.
and demonstrated convincingly, while its applicability to natural
product synthesis has been illustrated in a compelling manner.
Stereochemical features of the IMPEDA process were character-
ized and parlayed into a successful asymmetric synthesis of
several naturally occurring hamigerans (1-4) and analogues
thereof. Further studies employing the developed chemistry
toward compound library construction and chemical biology
studies are both warranted and viable.
Acknowledgment. This work was financially supported by
the National Institutes of Health, The Skaggs Institute for
Chemical Biology, a Louis R. Jabinson fellowship (to D.L.F.G.),
and grants from Amgen, Merck, and Pfizer.
Note Added after ASAP Posting. There was an error in the
Scheme 4 reagents and conditions in the version posted on
12/17/03; the corrected version was posted on 1/6/04.
Supporting Information Available: Experimental procedures
and compound characterization. This material is available free
of charge via the Internet at http://pubs.acs.org.
Conclusion
The described PEDA and IMPEDA chemistry opens new
avenues to complex molecular frameworks via photo-induced
cascade reactions. Its generality and scope has been explored
JA030498F
9
J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004 627