Total synthesis of 5-N-acetylardeemin and amauromine: Practical routes to potential MDR reversal agents

Article abstract of DOI:10.1021/ja991558d

The total synthesis of the title compounds has been accomplished. The key step involves the kinetic stereoselective conversion of 19 → 20. The synthesis of 20 represents for the first time a direct method for constructing exo-pyrroloindoles from protected

Full text of DOI:10.1021/ja991558d

J. Am. Chem. Soc. 1999, 121, 11953-11963
11953
Total Synthesis of 5-N-Acetylardeemin and Amauromine: Practical
Routes to Potential MDR Reversal Agents
Kristopher M. Depew,a,c,d Stephen P. Marsden,c,e Danuta Zatorska,b Andrzej Zatorski,a
William G. Bornmann,b and Samuel J. Danishefsky*,a,b
Contribution from the Laboratory for Bioorganic Chemistry and the PreparatiVe Synthesis Core Facility,
Memorial Sloan-Kettering Cancer Center, 1275 York AVenue, Box 106, New York, New York 10021, and
Department of Chemistry, Columbia UniVersity, HaVemeyer Hall, New York, New York 10027
ReceiVed May 10, 1999
Abstract: The total synthesis of the title compounds has been accomplished. The key step involves the kinetic
stereoselective conversion of 19 f 20. The synthesis of 20 represents for the first time a direct method for
constructing exo-pyrroloindoles from protected tryptophans in a highly diastereoselective manner. This step
was followed by reverse prenylation (see conversion of 20 f 27). Using the methodology worked out for the
titled compounds, a practical synthesis of several promising MDR reversal agents was possible. Biological
data that provided the basis for selection of candidates for advanced study are presented. Preliminary profiling
of the zones of the molecules that are responsive to changes while still retaining MDR reversal ability are
described. On the basis of these findings, compounds 2, 50, and 51 were selected for more extensive biological
follow-up.
Introduction
insensitive (by selection) to this and other cytotoxic agents. Cells
manifesting “operational resistance” (i.e., regardless of mech-
anism) to cytotoxic agents are said to be multidrug resistant
(MDR).3 Thus, 5-N-acetylardeemin would be termed a naturally
occurring MDR reversal agent.4 One mechanism widely as-
sociated with MDR apparently arises from overexpression of
an efflux glycoprotein, Pgp-170.5 However, MDR is a multi-
factorial phenomenon.6 Inhibition of Pgp-170 alone may not
suffice to accomplish reversal because cells employ diverse
strategies to thwart the action of drugs.
In vitro studies indicated that a 10 µM concentration of 2
enhanced vinblastine sensitivity to the resistant cell line (KBV-
1), which dramatically lowered the resistance from 1600- to
6-fold.1a Put another way, in this cell line, compound 2 allows
for maintenance of vinblastine cytotoxicity, but at a 250-fold
reduced dosage. Were this type of capability generalizable and
translatable to clinical reality, it could have major ramifications
for chemotherapy. In separate experiments, 2 was 10-fold more
effective than verapamil in chemosensitizing KBV-1 to vin-
blastine. These findings were potentially impressive in that
increased potency reduces the likelihood of toxic complications
from the MDR reversal agent itself. Interestingly, while
compound 3 was also found to be active, ardeemin (1) was,
apparently, rather less potent as an MDR reversal agent in its
initial screen against KBV-1 cells.1a The presence of the N-5
acetate group in 2 versus 1 and the resulting impact on MDR
activity is intriguing.
In 1993, as part of a screening program for biologically active
metabolites, McAlpine and co-workers found that extracts of
the fungus Aspergillus fischerii (var. brasiliensis) demonstrated
the ability to restore vinblastine sensitivity to a tumor cell line
that was otherwise insensitive.1a,b Isolation of the active
components from the fermentation mixture led to the charac-
terization of three structurally related agents, which were called
the ardeemins for their ability to reverse drug insensitivity (vide
infra). The major and most active constituent was named 5-N-
acetylardeemin (Figure 1, 2). Two other constituents isolated
from the product mixture were termed ardeemin (1) and 15bâ-
hydroxy-5-N-acetylardeemin (3). Structurally, the ardeemins
belong to an interesting class of natural products, which we
have termed the “reverse prenyl” hexahydropyrrolo[2,3-b]indole
alkaloids. Another reverse prenyl alkaloid is the potent vasodi-
lator amauromine (4), isolated by Takase and co-workers from
the Amauroascus sp. 6237.2
Our interest in the ardeemins was influenced to no small
extent by the method that guided their isolation and purification.
The bioassay logic involved the ability of fermentation isolates
to restore vinblastine sensitivity to cell lines that had become
a Bioorganic Chemistry, Sloan-Kettering Institute for Cancer Research.
b Preparative Synthesis Core Facility, Sloan-Kettering Institute for Cancer
Research.
c Columbia University.
d Current address: Department of Chemistry, Harvard University,
Cambridge, MA, 02138.
e Current address: Department of Chemistry, Imperial College of Science,
Technology and Medicine, London, SW7 2AY, UK.
(3) For a review of this area see: Molecular and Cellular Biology of
Multidrug Resistance in Tumor Cells; Robinson, I. B., Ed.; Plenum Press:
New York, 1991.
(4) For another instance of a naturally occurring MDR reversal agent
see: Stratman, K.; Burgoyne, D.; Moore, R. E.; Patterson, G. M. L.; Smith,
C. D. J. Org. Chem. 1994, 59, 7219.
(5) (a) Georges, E.; Sharom, F. J.; Ling, V. AdV. Pharmacol. 1990, 21,
185. (b) Pastan, I. H.; Gottesman, M. M. Important AdV. Oncol. 1988, 1,
13.
(6) Drug Resistance; Hait, W. N., Ed.; Kluwer Academic Publishers:
Boston, 1996.
(1) (a) Biological activity: Karwowski, J. P.; Jackson, M.; Rasmussen,
R. R.; Humphrey, P. E.; Poddig, J. B.; Kohl, W. L.; Scherr, M. H.; Kadam,
S.; McAlpine, J. B. J. Antibiot. 1993, 46, 374. (b) Isolation and structure:
Hochlowski, J. E.; Mullally, M. M.; Spanton, S. G.; Whittern, D. N.; Hill,
P.; McAlpine, J. B. J. Antibiot. 1993, 46, 380.
(2) (a) Isolation: Takase, S.; Iwami, M.; Ando, T.; Okamoto, M.;
Yoshida, K.; Horiai, H.; Kohsaka, M.; Aoki, H.; Imanaka, H. J. Antibiot.
1984, 37, 1320. (b) Structure: Takase, S.; Kawai, Y.; Uchida, I.; Tanaka,
H.; Aoki, H. Tetrahedron 1985, 41, 3037.
10.1021/ja991558d CCC: $18.00 © 1999 American Chemical Society
Published on Web 12/09/1999

11954 J. Am. Chem. Soc., Vol. 121, No. 51, 1999
Depew et al.
Figure 1.
Inspection of the ardeemin architecture reveals an alkylatively
cyclized L-tryptophan residue, linked in a diketopiperazine
arrangement to a D-serine residue. Furthermore, the mixed
diketopiperazine arrangement in the ardeemin series is fused,
through a benzopyrazinone motif, to an anthranilic acid. The
related and simplified structure of amauromine is comprised of
a C2-symmetric arrangement, wherein two identical reverse
prenylated cyclic tryptophan moieties are united as a diketopip-
erazine. Not surprisingly, acidic hydrolysis of amauromine leads
to cleavage of the linking ring and loss of the angular reverse
prenyl group to yield L-tryptophan. By way of comparison we
include the structure of the calcium channel blocker verapamil,
which is a well-known MDR reversal agent. Unfortunately, the
clinical efficacy of verapamil is sharply compromised by
attendant cardiotoxicity (vide infra).7 At this time, there is no
information on the question of the degree of mechanistic
homology, if any, between verapamil and 5-N-acetylardeemin.
Figure 2.
and physostigmine.14 It is conceivable that in those alkaloids
that contain the reverse prenyl group at C3 this function is
introduced biosynthetically by alkylation, in an SN2′ sense, via
a prenylated cysteine of a prenyl transferase. The latter may, in
turn, be generated via the action of prenyl pyrophosphate on a
cysteine residue.15 Subsequent or concurrent 5-exo cyclization16
of the tryptophyl R-amino group onto the intermediate indole-
nine would complete construction of the pyrroloindole. For
thermodynamic and, possibly, kinetic reasons, the fusion of the
pyrroloindole moiety is established in a cis sense. In the
biosynthesis of the ardeemins, the reverse prenyl group emerges
syn to the tryptophan “carboxyl” function and on the exo face
of the pendant pyrroloindole system. However, in other related
systems, both the syn and anti (exo) dispositions are encountered
(Figure 2). The stereochemical relationship of the pyrroloindole
junction and the tryptophyl-derived acyl group will be discussed
further as the account unfolds.
A common feature of the ardeemins and amauromine is the
presence of a “reverse prenyl” (R,R-dimethallyl) group at the
junction of rings B and C of the cyclized tryptophan. Numerous
examples of cyclized tryptophan alkaloids, differing in the nature
of the substitution at the BC junction, are known, such as in
gypsetin,8 brevianamide E,9 asperlicin E,10 himastatin,11 and the
flustramines.12 Other patterns of cyclized tryptophans that lack
the angular prenyl or reverse prenyl feature are hodgkinsine13
The ardeemins and amauromine are available from fermenta-
tion only with considerable difficulty and were not at all
available to our laboratory. Thus, total chemical synthesis would
be the only recourse for our laboratory to gain access to
significant quantities of these compounds. With synthetically
derived material, we would hope to study 5-N-acetylardeemin
itself, in detail, as to its in vivo and in vitro performances. With
synthetic intermediates as well as the natural product at hand,
we would also be in a position to generate analogue structures
in the context of a SAR evaluation.
(7) Ozols, R. F.; Cunnion, R. F.; Klecker, R. W.; Hamiltion, T. C.;
Ostchega Y.; Parillo, J. E.; Young, R. C. J. Clin. Oncol. 1987 5, 641.
(8) a) Isolation: Shinohara, C.; Hasumi, K.; Takei, Y.; Endo, A. J.
Antibiot. 1994, 47, 163. (b) Nuber, B. Hansske, F.; Shinohara, C.; Miura,
S.; Hasumi, K.; Endo, A. J. Antibiot. 1994, 47, 168. (c) Synthesis:
Schkeryantz, J. M.; Woo, J. C. G.; Danishefsky, S. J. J. Am. Chem. Soc.
1995, 117, 7025.
(9) (a) Isolation: Birch, A. J.; Wright, J. J. Tetrahedron 1970, 26, 2329.
(b) Synthesis: Kametani, T.; Kanaya, N.; Ihara, M. J. Am. Chem. Soc. 1980,
102, 3974.
(10) Houck, D. R.; Ondeyka, J.; Zink, D. L.; Inamine, E.; Goetz, M. A.;
Hensens, O. D. J. Antibiot. 1988, 41, 882.
(11) Kamenecka, T. M.; Danishefsky, S. J. Angew. Chem., Int. Ed. Engl.
1998, 37, 2995.
In addition to the strong inducements at the biological level
for a total synthesis venture, the chemical issues requiring
(12) Flustramine A: (a) Carle´, J. S.; Christophersen, C. J. Am. Chem.
Soc. 1979, 101, 4012. (b) Carle´, J. S.; Christophersen, C. J. Org. Chem.
1980, 45, 1586. Flustramine C: (c) Carle´, J. S.; Christophersen, C. J. Org.
Chem. 1981, 46, 3440. Flustramine D: (d) Wright, J. L. C. J. Nat. Prod.
1984, 47, 893. (e) Laycock, M. V.; Wright, J. L. C.; Findlay, J. A.; Patil,
P. D. Can. J. Chem. 1986, 64, 1312.
(14) Takano, S.; Ogasawara, K. In The Alkaloids; Brossi, A., Ed.;
Academic Press: San Diego, 1989; Vol. 36, p 225.
(15) For conjectures concerning the biosynthesis of such natural products
see: (a) Bycroft B. W.; Landon, W. J. Chem. Soc., Chem. Commun. 1970
168. (b) Bycroft, B. W.; Landon, W. J. Chem. Soc., Chem. Commun. 1970,
967.
(13) Fridrichsons, J.; Mackay, M. F.; Mathieson, A. M. Tetrahedron
1974, 30, 85, and references therein.
(16) Baldwin, J. E. J. Chem. Soc., Chem. Commun. 1976, 734.

5-N-Acetylardeemin and Amauromine
J. Am. Chem. Soc., Vol. 121, No. 51, 1999 11955
Scheme 1
attention in such a project were of inherent interest. Given the
range of structures bearing angularly functionalized pyrroloin-
dole systems (vide supra), it seemed probable that solutions to
the ardeemin or amauromine assembly problems would find
broader applicability. Herein, we present our successful total
syntheses of 5-N-acetylardeemin (2) and amauromine (4).17 We
do this in the context of a more general route to the introduction
of allyl and reverse prenyl functionality in the angular position
of the hexahydropyrroloindole alkaloids. We also describe some
potentially informative SAR work, which was enabled by the
synthetic advances.
While all reasonable possibilities for achieving such a direct
alkylative cyclization of a suitable tryptophan derivative had
certainly not been exhausted,18-20 another as yet totally untested
and, therefore, more interesting possibility presented itself. In
this view of the problem, a cis-fused pyrrolindole (cf. 7) would
be fashioned by a more precedented oxidative cyclization21 of
6 via the generalized oxidant “Ox+”. It was further supposed
that following appropriate activation, homolysis or heterolysis
of the angular leaving group could be induced. Regarding the
latter possibility, electronic accession from the indolic nitrogen
could greatly facilitate expulsion of the angular hetero group,
resulting in cation 8. It was further supposed that a suitable
prenyl-based nucleophile (cf. 9) would react with 8 via allylic
transposition to generate 10. Given the high preference for cis
versus trans fusion, it seemed likely that the formation of 10
from 7 (via cationoid system 8) would occur with overall
retention of configuration at the junction center corresponding
to C16a in ardeemin numbering. An alternate option, involving
homolysis of the leaving group and recombination with a
functional 1,1-dimethallyl radical, was examined and will be
discussed (vide infra).
Viewed in these terms, the critical stereochemical eVent in
the proposed oxidatiVe cyclization step is the establishment of
the relationship of C5a and C15b (ardeemin numbering) in
system 7. In the ensuing alkylation event (7 f 8 f 10), the
overall stereochemical outcome at C16a (retention) is, in
essence, controlled by the chirality at C5a, given the high
predilection for consolidation of the pyrroloindole moiety in a
cis junction. We assumed that this preference would pertain in
either the homolytic or heterolytic alkylation options.
Synthetic Planning
For obvious reasons, we hoped to take advantage of the
commercial availability (of L-tryptophan) and its various deriva-
tives (see generalized structure with P and P′ unspecified). Our
first subgoal would be pyrroloindole 10, bearing the angular
reverse prenyl group in the required syn relationship to the
tryptophan carboxyl (Scheme 1). An idealized solution to the
problem is seen in the hypothetical construction 6 f 10. In
this perception, reverse prenylation of 6 at position 3 of the
indole would set the stage for cyclization via the side chain
NH function of a suitably protected tryptophan. This general
concept, in various modalities, had been inherent in several
previous approaches toward the construction of the pyrrolo-
indoles.18-20 Its success would depend, in the first instance, on
introduction of a functional carbon electrophile, capable in some
way of progression to a reverse prenyl group. Included under
this formalism would be the special case where the electrophile
(E+) corresponded to the intact reverse prenyl group. Moreover,
successful application brings with it the proviso that the chiral
center of the tryptophan substrate shall order the sense of
pyrroloindole formation such that the acyl group is presented
on the convex face of the cup-shaped product.
The literature precedents for achieving high degrees of
stereoselection in the desired sense and in high yield for the
oxidative cyclization reaction of 6 f 7 were not fully encourag-
ing. Detailed studies of Hino22 and Crich,23,24 in the protonically
induced version of this reaction (Ox ) H+), demonstrated that
the product with the endo carboxyl is thermodynamically more
stable and may also be competitive at the kinetic level. Of
course, to reach the ardeemins and amauromine, access to the
exo (kinetic) product 10 would be necessary. This system
(17) For a preliminary account of this work, see: Marsden, S. P.; Depew,
K. M.; Danishefsky, S. J. J. Am. Chem. Soc. 1994, 116, 11143.
(18) (a) Jackson, A. H.; Smith, A. E. Tetrahedron 1965, 21, 989. (b)
Bocchi, V.; Casnati, G.; Marchelli, R. Tetrahedron 1978, 34, 929. (c)
Muthusubramanian, P.; Carle, J. S.; Christophersen, C. Acta Chem. Scand.
1983, B37, 803. For an example in which a protected tryptophan gave C3
and C1 bis(prenylated) products as 1:1 junction isomers in 29 and 22%
yield, see: (d) Takase, S.; Kawai, Y.; Uchida, I. Tetrahedron Lett. 1984,
25, 4673.
(21) Ohno, M.; Spande, T. F.; Witkop, B. J. Am. Chem. Soc. 1968, 90,
6521. For a recent application of this oxidative closure logic see:
Kamenecka, T. M.; Danishefsky, S. J. Angew. Chem., Int. Ed. Engl. 1998,
37, 2993.
(19) Wenkert, E.; Sliwa, H. Bioorg. Chem. 1977, 6, 443.
(22) Taniguchi, M.; Hino, T. Tetrahedron 1981, 37, 1487.
(23) Crich, D.; Davies, J. W. J. Chem. Soc., Chem. Commun. 1989, 1418.
(24) Bourne, G. T.; Crich, D.; Davies, J. W.; Horwell, D. C. J. Chem.
Soc., Perkin Trans. 1 1991, 1693.
(20) Cf.: (a) Hino, T.; Hasumi, K.; Yamaguchi, H.; Taniguchi, M.;
Nakagawa, M. Chem. Pharm. Bull. 1985, 33, 5202. (b) Nakagawa, M.;
Ma, J.; Hino, T. Heterocycles 1990, 30, 451.

11956 J. Am. Chem. Soc., Vol. 121, No. 51, 1999
Depew et al.
Scheme 2
Scheme 3a
a Reaction conditions: (a) NaOH, Cbz-Cl, cat. (n-Bu)4NHSO4, CH2Cl2, 83%; (b) N-PSP, cat. p-TSA, CH2Cl2, 84%; (c) methyl acrylate, (n-
Bu)3SnH, AlBN, TolH, reflux, 55%; (d) allyl-Sn(n-Bu)3, AlBN, toluene, reflux, 56%.
corresponds to the exo configuration for the tryptophyl-derived
acyl function with respect to the cup-shaped pyrroloindole
system. Given the logic of the projected route discussed above,
which anticipated oxidative cyclization followed by alkylation,
the former step must be such as to present the acyl function
exo at the stage of 7. The work of Crich23,24 cited above and
that of Hino22 in the context of variously related cyclizations
suggested the likelihood of mixtures at the kinetic level if the
reaction is driven to completion prior to equilibration. Given
the studies of the Crich group, thermodynamic equilibration
would unlikely afford the desired exo product.25 In principle, a
solution to our problem might be achieved by kinetic protonation
of a C15a enolate (cf. 13), possibly derived from the endo series
system 12 (Scheme 2). Such a “kinetic” protonation might
provide access to the desired 7. Clearly, if we were to follow a
course based on such a prospectus, it would be necessary to
start with D-tryptophan (11) en route to endo 12. The latter
would contain the pyrroloindole version of 7, but would be
epimeric at C15a. The success of such an enterprise could also
not be anticipated with confidence.
Discussion and Results
Since a great deal of research had already been conducted
by previous workers21 into oxidative cyclizations of tryptophan
derivatives using oxygen-based electrophiles, we decided on a
newer type of venture. Inspired by an earlier success in our lab
with selenocyclization as a means of fashioning a pyrrolo ring
(via concurrent formation of carbon-selenium and nitrogen-
carbon bonds),26,27 we turned to this approach for the problem
at hand.
In our first attempt at the reduction of Scheme 1 to practice,
we protected L-tryptophan in the form of phenylhydantoin 14
(Scheme 3). A few attempts at selenocyclization of 14 were
unsuccessful. Before exhausting a variety of possibilities for
(26) Cf.: (a) Nicolaou, K. C.; Lysenko, Z. J. Am. Chem. Soc. 1977, 99,
3185. (b) Clive, D. L. J. Tetrahedron 1978, 34, 1049.
(27) Danishefsky, S. J.; Berman, E. M.; Ciufolini, M.; Etheredge, S. J.;
Segmuller, B. E. J. Am. Chem. Soc. 1985, 107, 3891.
(25) For a detailed analysis of the issues inherent in this puzzling result
see: Crich, D.; Bruncko, M.; Natarajan, S.; Teo, B. K.; Tocheri, D.-A.
Tetrahedron 1995, 51, 2215.

5-N-Acetylardeemin and Amauromine
J. Am. Chem. Soc., Vol. 121, No. 51, 1999 11957
Scheme 4. Proposed Transition Ensembles for the Selenocyclization Reaction
using 14, we backed off the idealized version of the synthesis
and investigated the concept in the tryptamine rather than
tryptophan series. Specifically, we prepared the bis(Cbz) deriva-
tive 15. Indeed, in this case, treatment with N-phenylselenoph-
thalimide (N-PSP)28 in methylene chloride in the presence of
catalytic p-toluenesulfonic acid (p-TSA) afforded 16 in 84%
yield. Thus, an important element of the basic plan for the total
synthesis had been reduced to practice.
Fresh from this success, we investigated the feasibility of
angular allylation, prenylation, and reverse prenylation at C3
(indole numbering). Our first initiative involved attempted
applications of the elegant free radical chemistry pioneered by
Keck29 (and subsequently practiced in our own lab)30 for the
introduction of an allylic functionality via carbon-bound halogen.
Treatment of selenide 16 with tri(n-butyl)tin hydride in the
presence of allyltri(n-butyl)tin led to reduction rather than
reductive alkylation. Apparently the tertiary pyrroloindolyl free
radical, at least under these conditions, is more adept at hydrogen
atom abstraction than at de-stannylative allylation (see formation
of 17). Success was realized, albeit at a modest level, through
the reaction of 16 under photolytic conditions with hexa(n-
butyl)distannane31 in the presence of allyltri(n-butyl)stannane
to afford 18 in 56% yield.
stereochemical connectivity, even in the absence of fully
supportive precedents (vide supra). We first turned our attentions
to a feasible protection of L-tryptophan, which still provides an
opportunity for achieving the gross cyclization goal. In the event,
L-tryptophan methyl ester was converted to its bis(Boc) deriva-
tive 19 using Boc anhydride under phase transfer conditions,
as shown (Scheme 4). Reaction of compound 19 with N-PSP
in the presence of p-TSA occurred smoothly. Early in our
studies,17 when reaction was carried out in the presence of excess
sodium sulfate and p-TSA as catalyst (10 mol %), wherein we
were monitoring the progress of the reaction by NMR analysis,
the appearance of two isomers was indicated. The rough ratio
of these products seemed to be ca. 1:1. It further appeared that
as the reaction progressed, the ratio of the two compounds
changed to the point where there was a clear major and a minor
product (ca. 6-9:1). Since the NMR signals due to the two
compounds were similar and since the concentration of one
seemed to grow at the expense of the other, we assumed that
the two products corresponded to the endo versus exo acyl
systems.
As we continued to examine the selenocyclization reaction,
we could not repeat this curious observation pertaining to a
change in ratio as the reaction progresses. In these follow-up
studies, using again p-TSA, there was produced from the outset
a mixture of cyclization products in a ratio of ca. 8-10:1.
Indeed, we found that this apparently kinetic ratio could be
improved still further by modification of the cyclization method.
The heterogeneous conditions associated with the use of p-TSA
in our first attempts were not very well suited for scale-up. We
used sodium sulfate to dry the hydrated form of p-TSA, which
by itself is a poor catalyst for this particular substrate. Also,
under these conditions, our exo:endo product ratios varied from
9:1 to 16:1. To correct both problems, we turned to anhydrous
Encouraged by these successes, we returned to the tryptophan
series. At this stage we were particularly interested in pursuing
a kinetic solution to the problem of the C5a-C15b-C16a
(28) Nicolaou, K. C.; Clareman, D. A.; Barnette, W. E.; Seitz, S. P. J.
Am. Chem. Soc. 1979, 101, 3704.
(29) Keck, G. E.; Yates, J. B. J. Am. Chem. Soc. 1982, 104, 5829.
Coincident with our report17 Crich reported related chemistry with a 3a-
bromopyrroloindole (see: Bruncko, M.; Crich D.; Samy, R. J. Org. Chem.
1994, 59, 5543.
(30) Webb, R.; Danishefsky, S. J. Tetrahedron Lett. 1983, 24, 1357.
(31) cf. Keck, G. E.; Enholm. E. J.; Yates, J. B.; Wiley, M. R. Tetrahdron
1985, 41, 4079.

11958 J. Am. Chem. Soc., Vol. 121, No. 51, 1999
Depew et al.
pyridinium p-toluenesulfonate (PPTS). In the presence of an
equimolar amount of PPTS, reaction of 19 with N-PSP occurred
remarkably smoothly to deliver the selenocyclization product
in 93% yield as an 18:1 mixture of 20:21. Under these
conditions, the ratio of the two stereoisomers 20(exo):21(endo)
as judged by the integrated levels of the methyl ester resonance
Moreover, the assignment of stereochemistry at C5a and C16a
of the major product could not be established in a convincing
way. Nonetheless, we proceeded to address the all-critical
introduction of a reverse prenyl function. On the basis of our
experiences in the tryptamine series described above, we began
with an experiment directed to free radical allylation.
1
in H NMR spectra did not materially change throughout the
Photolysis of the 20:21 mixture, enriched in the former, with
allyltri(n-butyl)tin in the presence of hexa(n-butyl)distannane
gave rise, in 95% yield, to a mixture, which by NMR anaylsis
was comprised of a major and a minor allylation product (see
structures 22 and 23) in much the same ratio as was operative
in the precursor phenylseleno series (20 and 21). With these
compounds in hand, the stereochemistry at C5a could be
surmised. One-dimensional NOE experiments on the major
product revealed that irradiation of the allylic protons caused
an enhancement of one of the H16 protons. Irradiation of this
proton enhanced the other H16 proton and H15b. Irradiation of
the H15b proton strongly enhanced the latter H16 proton and
weakly enhanced the former. Hence, the allyl group seemed to
be on the opposite plane, i.e., trans to H15b. If the assignment
is correct, compound 22 corresponds to the desired diastereomer.
Accordingly, by the logic discussed above, 20 is in the required
exo series. However, because we had noted some H15b
enhancement by both H16 protons, the decision was not clear-
cut. A total synthesis of amauromine or one of the ardeemins
would hopefully confirm the assignment.
Attention was now directed toward installation of the reverse
prenyl group via photolysis. Irradiation of the 20/21 mixture
(rich in the former) in the presence of prenyltri(n-butyl)tin and
hexa(n-butyl)distannane31 led to complete reductive cleavage
of the phenylseleno function (see formation of 24). Irradiation
of the mixture in the presence of other potential radical acceptors
(e.g., methyl acrylate and methyl 2-butynoate) resulted in either
no reaction or reductive cleavage. In one case, however, we
could install an acrylate group to give 25 by reaction with ethyl
â-tri(n-butyl)stannylacrylate,34 albeit in 21% yield. In this
product mixture, we saw evidence of a dimeric product (HRMS),
which could only have resulted from the coupling of two tertiary
radicals. Photolysis of the 20/21 mixture with hexa(n-butyl)-
distannane produced this dimer (26) in 10-15% yield. Com-
pound 26 represents a potential solution toward natural products
of the dimeric tryptamine and tryptophan class of indole
alkaloids.35 In principle, photolysis of tryptamine 16 could allow
direct access to the Calycanthaceous class of plant natural
products.
course of the reaction. Given these findings and given the fact
that the apparent equilibration of 20 in favor of 21 could not
be duplicated, we now believe that the high preference in the
reaction in favor of 21 is a kinetic effect.
We also note that the initial report, which implied that the
endo product was being equilibrated in favor of the exo product,
was evaluated by Crich and co-workers.32 Their research
demonstrated that, as is usual in related systems, the endo isomer
(cf. 21) is actually more stable than the exo isomer (20).
Although the Crich conditions for base-induced equilibration
were radically different from ours, we can only assume that
the thermodynamic relationship of 20 and 21, even in the setting
of our experiment, is not materially different.
Attempts to account for the high degree of stereoselection in
the intramolecular selenoamination reaction are complicated by
the fact that, at present, there is no fully understandable theory
as to the central observation from earlier workers25 that the endo
product is thermodynamically more stable than the exo product
in related systems. The extent of the kinetic level of stereocontrol
in the phenylselenocyclization26-28 may well rest on the degree
to which cyclization corresponds to anti-addition to the 2,3-
indolic double bond by the electrophilic oxidant and the
nucleophilic tryptophan nitrogen (Scheme 4, see structures 20p
and 21p). This “SN2-like” mode of cyclization should be
contrasted with another cyclization pathway, which progresses
through a discreet iminium species en route to pyrroloindole
formation. In the latter case, the stereochemical outcome had,
in substance, been determined by the sense of attack at C3 of
the indole. Thus, it seems unlikely that the formation of the
indolenine cyclization intermediate could manifest a high degree
of kinetic stereoselection. By contrast, in the more “SN2-like”
variation, kinetic stereoselection seems more probable. Thus it
is possible that formation of the selenonium species 20p and
21p en route to the SN2-type of cyclization is reversible and
that the kinetic ratio of 20:21 may reflect the relative ease of
cyclization of the pre-exo and pre-endo intermediates 20p and
21p, respectively. In each instance, we position the car-
bomethoxy function antiperiplanar to the N-Boc function as the
urethane nitrogen prepares to displace the carbon-selenonium
bond with inversion. While the final endo product 21 is
apparently more stable than the exo product 20, judging by the
base-induced equilibration reaction described by Crich,32 the
pre-exo cycling ensemble 20p could well be more stable than
the pre-endo ensemble 21p. In the latter situation, the car-
bomethoxy group is possibly drawn more deeply into the
concave pocket of the pyrroloindole than is the case in the final
product 20. In summary, the stability order of the two precy-
clization ensembles may be quite different from that of the
energy-minimized Version of the products 20 and 21.33
We next attempted to activate the 20/21 mixture (highly
enriched in favor of the former) toward reverse prenylation in
a cationic sense. Exposure of the selenides to Lewis acids such
as TiCl4, BF3‚OEt2, and AgOTf in the presence of prenyltri(n-
butyl)tin brought forth a complex mixture of products, probably
associated with cleavage of the phenyl selenide function.
(34) (a) Russell, G. A.; Tashtoush, H.; Ngoviwatchai, P. J. Am. Chem.
Soc. 1984, 106, 4622. (b) Baldwin, J. E.; Kelly, D. R. J. Chem. Soc., Chem.
Commun. 1985, 682. (c) Barrett, A. G. M.; Pilipauskas, D. J. Org. Chem.
1991, 56, 2787.
(35) We attribute the formation of 26 to result from the simple coupling
of two free tertiary radicals. NMR experiments of 26 at room temperature
gave broad peaks, which sharpened at 55 °C to one set of pyrroloindole
peaks, which suggests a dimeric structure of C2-symmetry. This product
potentially represents a very direct solution to the problem of constructing
fungal natural products of the C3a-bis(pyrroloindole) alkaloid class such
as (a) the leptosins: Takahashi, C.; Takai, Y.; Kimura, Y.; Numata, A.;
Shigematsu, N.; Tanaka, H. Phytochemistry 1995, 38, 155. (b) Chaetocin:
Weber, H. P. Acta Crystallogr. B 1972, 28, 2945. (c) Ditryptophenaline:
Springer, J. P.; Buchi, G.; Clardy, J. Tetrahedron Lett. 1977, 18, 2403. (d)
WIN 64821: Barrow, C. J.; Cai, P.; Snyder, J. K.; Sedlock, D. M.; Sun, H.
H.; Cooper, R. J. Org. Chem. 1993, 58, 6016. See Supporting Information
for details.
Returning to the syntheses, separation at the stage of the
methyl esters 20 and 21 could not be accomplished in our hands.
(32) Crich, D.; Huang, X.; Manuscript submitted. That the ratio of 20:
21 was not changing during the course of the reaction of 19 with N-PSP
had already been described in the dissertation of Kristopher Depew,
Columbia University, 1998 (vide infra ref 42).
(33) The extent to which the critical carbomethoxy is drawn into the
cavity is a function of the precise bond angles of the selenonium species
and the requirement of collinearity in the displacement process.

5-N-Acetylardeemin and Amauromine
J. Am. Chem. Soc., Vol. 121, No. 51, 1999 11959
Scheme 5a
a Reaction conditions: (a) NaOH, cat. (n-Bu)4NHSO4, BOC2O, CH2Cl2, 91%; (b) N-PSP, CH2Cl2, PPTS, 93%; (c) allyl-Sn(n-Bu)3, (n-Bu)6Sn2,
toluene, hν, 23 °C, 94%; (d) (n-Bu)3SnH, AlBN, methyl acrylate, toluene, reflux, 55%; (e) (n-Bu)6Sn2, toluene, hν, 23 °C, 10-15%; (f) (n-Bu)6Sn2,
ethyl â-tri(n-butyl)stannylacrylate, toluene, hν, 23 °C, 21%.
Scheme 6a
a Reaction conditions: (a) MeOTf, 2,6-di-(tert-butyl)pyridine, prenyltri(n-butyl)stannane, CH2Cl2, -78 °C to reflux, 60% (9:1 27:28); (b) NaOH,
THF/MeOH/H2O, reflux, 98%.
However, activation of a 9:1 mixture of 20/21 with methyl
trifluoromethanesulfonate (MeOTf) in the presence of 2,6-di-
(tert-butyl)pyridine and prenyltri(n-butyl)tin, beginning from
-78 °C and proceeding to reflux in CH2Cl2 under argon, led to
a 60% yield of the desired angular reverse prenyl products 27
and 28 (not separated, Scheme 5). The ratio of these products
was essentially the same as that of the starting mixture. In
addition, in some instances, up to 15% of a monomethyl
carbamate byproduct 29 was produced.36 The overall yield of
reverse prenylation was 75%. Again, we assumed the alkylation
reaction would produce the cis-fused 5,5-ring junction, even
though this course would have involved net retention in the
displacement step.
us to avoid laborious chromatography of acids 30 and 31, which
we used earlier to enable full purification of the major
diastereomer 30. Happily, simple crystallization from hexane
now delivered the major diastereomer methyl ester 27 in ca.
57% overall yield from 20/21. We note that this key intermediate
for the total synthesis of the ardeemins is available in three steps
from 19. In principle, it can service syntheses of the ardeemins,
of amauromine, or many of the other reverse prenyl hexahy-
dropyrroloindole alkaloids.
We first turned our attention to the amauromine problem.
Cleavage of the two Boc functions of 27, using iodotrimethyl-
silane (TMSI), gave rise to the doubly deprotected ester, 32
(Scheme 6). Coupling of 32 with acid 30 under mediation by
BOP chloride gave the amide 33. There seemed to be no
complication from competitive acylation at the anilino nitrogen.
Treatment of product 33 with TMSI produced amauromine.
Undoubtedly, this welcome result had come about by cleavage
of the two Boc groups accompanied by cyclization of the
carbomethoxy function onto the proximal NH group. That the
total synthesis of amauromine (4) had, in fact, been ac-
The methyl esters were cleanly hydrolyzed, as shown, to
provide the tricyclic acids 30 and 31 in quantitative yield, with
no apparent loss of stereochemical integrity. Having in hand a
highly enriched (18:1) mixture of methyl esters 27 and 28 in
essentially the same ratio as the starting 20/21 mixture allowed
(36) Presumably, 29 arises by methylation of the carbonyl group of the
more exposed t-Boc group accompanied by de-tert-butylation.

11960 J. Am. Chem. Soc., Vol. 121, No. 51, 1999
Depew et al.
Scheme 7a
a Reaction conditions: (a) TMSl, MeCN, 0 °C, 83%; (b) 30, BOP-Cl, Et3N, CH2Cl2, 58%; (c) TMSl, MeCN, 0 °C, 58%.
Scheme 8
complished was Vouchsafed by full spectral, chromatographic,
and optical rotation comparisons of the fully synthetic material
with an authentic sample proVided by the Fujisawa pharma-
ceutical company.
With the total synthesis of amauromine achieved, we next
focused on the synthesis of the ardeemins. Building blocks 27,
30, and 32 had already been prepared in homogeneous form,
from the amauromine program. The next subgoal for reaching
the ardeemins was that of amide bond formation between
modified amino acid 30 and the amino group of D-alanine
methyl ester (34).
Surprisingly, during this attempted coupling, under conditions
very similar to those used to prepare 33, substantial difficulties
were encountered. In the event, a variety of coupling agents
(DCC, DMAP; triethylamine, DCC, HOBT; and BOPCl) all
led to the production of protected dipeptide 35 (Scheme 7).
However, in each case, NMR analysis revealed the production
of two very closely related compounds. We took the formation
of this mixture to suggest the partial racemization of one of the
amino acid components in the synthesis of 35 (see structure
36). Actually, it was not clear whether this epimerization had
been sustained during the carboxyl activation process en route
to coupling or after the coupling. While we did not rigorously
establish whether the racemization had occurred in the tryp-
tophan- or the alanine-derived moieties, however, it can be
argued, based on the lack of a corresponding difficulty in the
virtual homo coupling leading to 33 en route to amauromine,
that the stereochemical vulnerability in the pre-ardeemin case
was lodged in the alanine.
acid derived carboxyl groups can be converted to acid fluorides
without epimerization under mild, nonacidic conditions. Fur-
thermore, Carpino demonstrated that such acid fluorides couple
smoothly to amines to generate peptide bonds, again without
racemization. In the event, the action of compound 30 with
cyanuric fluoride and pyridine in the presence of methylene
chloride at -15 °C generated the acid fluoride 37 (Scheme 8).
Reaction of this substance under Schotten-Baumann-like
conditions with D-alanine methyl ester (34) resulted in amide
bond formation yielding the protected peptide 35 in 71% overall
yield. Only a single product seemed to have been produced.
The partial racemization problem had thus been overcome. In
like fashion, glycine methyl ester and D-phenylalanine methyl
ester coupled to 37 to give peptides 38 and 39, respectively.
Proceeding toward our ardeemin goal, the two Boc groups
of compounds 35 were cleaved with TMSI in freshly distilled
acetonitrile at 0 °C. There was thus generated the diamine 40
in 86% yield. When this compound was subjected to the action
of methanolic ammonia, containing catalytic quantities of
DMAP, the diketopiperazine 42 (R ) Me) was obtained in 70%
yield. Under these conditions the seco-amide 41 was generated
in approximately 10-15% yield. However, this compound could
easily be removed by chromatography. Unlike the seco-
amauromine case (cf. 33 f 4), the diamine 40 did not
spontaneously cyclize. Smooth cyclization to 42 required the
presence of N,N-dimethylaminopyridine (DMAP), without
which reaction was slow. In a similar way, diketopiperazines
43 and 44 were generated from seco systems 38 and 39,
respectively.
Fortunately this problem lent itself to a solution. The key to
(37) Carpino, L. A.; Mansous, E.-S. M. E.; Sadat-Aalee, D. J. Org. Chem.
1991, 56, 2611.
progress lay in the work of Carpino37 which indicated that amino

5-N-Acetylardeemin and Amauromine
J. Am. Chem. Soc., Vol. 121, No. 51, 1999 11961
Scheme 9a
a Reaction conditions: (a) cyanuric fluoride, pyridine, CH2Cl2, -15 °C, 93%; (b) D-Ala-OMe‚HCl, NaHCO3, H2O, CH2Cl2; (c) Gly-OMe, NaHCO3,
H2O, CH2Cl2, 71% from 30; (d) D-Phe-OMe, NaHCO3, H2O, CH2Cl2; (e) TMSl, MeCN, 0 °C, 86%; (f) NH3 sat. in MeOH, cat. DMAP, 86%.
The stage for completing the total synthesis of 5-N-acetyl-
ardeemin was now well set. Our strategy for accomplishing
fusion of an anthranilic acid onto ring D was based on work by
Eguchi,38 who developed an elegant synthesis of benzo-
quinazolinones utilizing an aza-Wittig-like reaction. To ben-
zoylate 42 with o-azidobenzoyl chloride (45), we required a
base stronger than Et3N, which had been used by Eguchi
(Scheme 9). We found that KHMDS worked well to give imide
46 in 50-80% yield. The yields were variable, and we found
it hard to completely convert all of 42 to 46. On larger scales,
we preferred to use o-azidobenzoic anhydride (47) as the
acylating agent for technical reasons. With this reagent,
compound 48 was reliably delivered in high yield. For example,
during our analogue synthesis, we found that n-BuLi (diluted
with THF) as a base was superior to KHMDS. When the lithium
anion of the 8-demethyl analogue 43 was quenched with a THF
solution of 47, compound 48 was obtained in 96% yield.
When compound 46 was stirred with P(n-Bu)3 in benzene,
ardeemin (1) was isolated in 90% yield.39 In a similar way,
treatment of 48 with P(n-Bu)3 gives 49 (8-desmethylardeemin)
in 86% yield. Finally, acetylation of 1 was effected with acetyl
chloride under the conditions shown to give 5-N-acetylardeemin
(2) in 82% yield. For larger scale work, we have found that
simply warming ardeemin in a 9:1 solution of acetic anhydride/
Hu¨nig’s base for 36 h at 60 °C causes precipitation of 2 during
the reaction. Filtration and chromatography of the mother liquor
leads to an overall yield of 85% for 2. The fully synthetic 5-N-
acetylardeemin (mp 229-231 °C) was identical in all respects
to naturally deriVed material furnished by the Abbott Pharma-
ceutical Company. In a similar way, compound 49 was
converted to the nor analogue of 2 (50) and by trifluoroacety-
lation to 51.
culture. One important question was the importance of the
N-acetyl group at the indoline nitrogen. As noted above,
ardeemin (1) itself was reported to be “inactive” as a reversal
agent with respect to KBV-1 cells. Was there, indeed, some
kind of profound electronic or steric property that the acetyl
group imparted to 2 to influence its activity relative to 1? What
effect would there be in deleting rings E and F, which are
nonexistent in most of the members of this natural product class?
If such compounds were still active, the synthesis of a viable
reversal agent would be substantially simplified. We also
wondered about the effect of modifying the character of C8
and the consequences of modifying the angular substituent at
C16a.
Our analogues were prepared along the following lines. The
allyl analogue (52) of 5-N-acetylardeemin (2) was synthesized
in seven steps from 22 following the same set of conditions as
described for 2 (Scheme 10). Truncated versions of the
ardeemins addressed to test the requirement of the quinazolinone
portion of 2 were made via synthetic intermediate 42. Trifluo-
roacetylation of 42 under the conditions shown gives a bis-
(trifluoroacetylated) product. The latter via de-trifluoroacety-
lation at N9, as shown, leads to 53. Attempted monoacetylation
of 43 leads to acetylation of the alanyl NH group of the
diketopiperazine ring (see compound 54).
The synthesis of the C8 benzyl compound 55 was ac-
complished in an interesting way. In principle, it would have
been possible to reach this goal via the previously described
diketopiperazine 44 following the three-step sequence practiced
in the synthesis of 2 (from 42) and 49 (from 43). A more
pleasing possibility started with the previously described nor
compound 50. The benzyl group was introduced in 43% yield
by deprotonation at C8 (following a Seebach type of protocol)40
followed by reaction with benzyl bromide.
With the total synthesis accomplished, we undertook a brief
survey of the regions of the ardeemins, which could be
structurally modified in a straightforward way. The activities
of such accessible analogues would be investigated first in cell
We are confident that the stereochemistry is indeed as shown
in 55. Thus, no NOE was observed between H8 and H15b,
which indicated a trans relationship between these two hydro-
gens. Also, H15b was shifted far upfield (2.53 vs 4.41 ppm for
2) due to shielding by the benzyl group. Late stage alkylation
of C8 streamlines the route to analogues of the central amino
(38) Takeuchi, H.; Hagiwara, S.; Eguchi, S. Tetrahedron 1989, 45, 6375.
(39) On occasion, this reaction is accompanied by the formation of 42
(in ca. 15% yield). Conceivably, this compound arises from attack of the
phosphine-imine nucleophile on the adjacent imide carbonyl group. This
would lead, formally, to a benzoazetidene en route to the observed 42. More
likely, an adventitious external nucleophile cleaves the labile exocyclic
imide.
(40) Seebach, D.; Bossler, H.; Gru¨ndler, H.; Shoda, S.-I. HelV. Chim.
Acta 1991, 74, 197.

11962 J. Am. Chem. Soc., Vol. 121, No. 51, 1999
Scheme 10. Synthesis of 1, 2, and Analoguesa
Depew et al.
a Reaction conditions: (a) KHMDS, THF, -78 °C, 45, 80%; (b) n-BuLi, THF, -78 °C, 47, 96%; (c) P(n-Bu)3, benzene, 72%; (d) Hu¨nig’s base,
acetic anhydride, 60 °C, 36 h, 85%; (e) trifluoroacetic anhydride, pyridine, 60 °C.
Scheme 11. Synthesis of 16a-Allyl-5-N-acetylardeemin, 8R-Benzylardeemin, and Two Truncated Analoguesa
a Reaction conditions: (a) trifluoroacetic anhydride, pyridine, 60 °C; (b) K2CO3, MeOH; (c) acetic anhydride, pyridine; (d) LDA, n-BuLi, THF,
-78 °C; benzyl bromide, 33%.
acid, bypassing the need to synthesize a fresh diketopiperazine
at the stage of precursor 30 for each projected C8 derivative
(Scheme 11).41
VBL100) to develop a basic sense of each compound’s relative
toxicity to the sensitive wild type. From there, we examined
the effects of these agents in conjunction with vinblastine in a
cell line, which was 760-fold resistant to vinblastine. The results
have been published elsewhere42,43 and are detailed in the
Supporting Information. We summarize here the most salient
findings.
Ardeemin (1) and 5-N-acetylardeemin (2) display comparable
cytotoxicity, while the 16a-allyl analogue (52) is 2- to 3-fold
less cytotoxic than 2 against these cell lines. As is the case with
1 and 2, 52 is more cytotoxic to resistant lines than to their
respective parent cell lines (collaterally sensitive). The excep-
tions were the 5-N-trifluoroacetyl analogue 51 and the 8R-benzyl
analogue (55), which were less cytotoxic to the resistant cell
line than to the parent (CCRF-CEM). Addressing deletion of
the acetyl group, we found that 1 was somewhat more cytotoxic
than 2 in the above sensitive cell lines. However, the reversal
power of 1 and its 8-desmethyl analogue (49) were slightly less
potent than 2, when studied in conjunction with vinblastine.
Compounds 42, 50, 51, 52, 53,and 55 were examined,
particularly in comparison with 1, 2, and verapamil, as the latter
is a widely discussed reversal agent. Since we did not prepare
each of these materials in amounts required for in vivo
evaluation, we used in vitro screens to orient our efforts. We
hoped that these initial experiments would identify the conse-
quences of our structural modifications relative to the parent 2.
We would then advance the most effective compound(s) to in
vivo studies. General cytotoxicity screens were conducted
against two wild-type tumor cell lines (DC-3F, CCRF-CEM)
and their MDR counterparts (DC-3F/ADII, CCRF-CEM/
(41) For some additional studies in the ardeemin area see: (a) Martin-
Santamaria, S.; Espada, M.; Avendan˜o, C. Tetrahedron 1997, 53, 16795.
(b) Bartolome, M. T.; Buenadicha, F. L.; Avendan˜o, C.; Sollhuber, M.
Tetrahedron-Asym. 1998, 9, 249. (c) Buenadicha, F. L.; Bartolome, M. T.;
Aguirre, M. J.; Avendan˜o, C.; Sollhuber, M. Tetrahedron-Asym. 1998, 9,
483. (d) Madrigal, A.; Grande, M.; Avendan˜o, C. J. Org. Chem. 1998, 63,
2724. (e) Fernandez, M.; Heredia, M. L.; de la Cuesta, E.; Avendan˜o, C.
Tetrahedron 1998, 54, 2777. (f) Sanchez, J. D.; Ramos, M. T.; Avendan˜o,
C. Tetrahedron 1998, 54, 969. (g) Caballero, E.; Avendano, C.; Menendez,
J. C. Tetrahedron Asymmetry 1998, 9, 3025.
(42) Chou, T. C.; Depew, K. M.; Zheng, Y.-H.; Safer, M. L.; Chan, D.;
Helfrich, B. Zatorska, D.; Zatorski, A.; Bornmann, W.; Danishefsky, S. J.
Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 8369.
(43) Depew, K. M., Ph.D. Dissertation, Columbia University, 1998.

5-N-Acetylardeemin and Amauromine
J. Am. Chem. Soc., Vol. 121, No. 51, 1999 11963
Thus, while the 5-N-acetyl group is not essential for reversal
power, it seems to help. Hence, the nonacylated 1 and 49 were
not advanced further. The trifluoroacetyl derivative 51 did seem
to have some advantage over 2 in the vinblastine combination
study. This compound is being considered for future develop-
ment. We note that the reversal power of verapamil could equal
that of 2 only when administered at 5 times the concentration
of 2. Given the peripheral toxicity problems of verapamil, such
an increase could not even be considered.
We also found that the hexacyclic compounds (2, 50, and
51) were much more efficacious than their tetracyclic analogues
(42, 53). For example, the latter two performed poorly in a
vinblastine combination study by being >400-fold weaker than
2 at re-sensitizing the CCRF-CEM/VBL100 cells. Thus, the
truncation of rings E and F vastly reduces reversal potential.
Hence, neither compound was advanced.
Following a preliminary round of in vitro screening, we
selected our parent structure 2 and the 8-desmethyl analogue
50 for more in-depth analyses involving other chemotherapeutic
agents (vinblastine, doxorubicin, Taxol) and several in vivo
studies using two separate mouse lines (B6D2F1 and Swiss
nude). The results of this study were also reported earlier.42,43
Both of these ardeemins (i) reversed drug resistance to vinblas-
tine and Taxol by >700-fold in the CCRF-CEM/VBL100 cell
line, (ii) killed MDR cells more efficaciously than wild-type
cells, and (iii) exhibited strong synergistic effects with doxo-
rubicin and vinblastine against the growth of MDR neoplastic
cells.
In vivo studies indicated that nontoxic doses of doxorubicin
and 2 significantly increased the life span of B6D2F1 mice
inoculated with wild-type and doxorubicin-resistant P388 tumor
cells. In separate experiments, using nude mice bearing human
MX-1 mammary carcinoma xenografts, one particular dose
combination regimen of doxorubicin with 2 produced two
tumor-free mice (out of eight in the group) by the 49th day
after the tumor was introduced.37 The ardeemins also demon-
strated remarkably low toxicity in mice. Quantities of 2 as high
as 300 mg/kg per day for 3 days or 150 mg/kg per day for 8
days were tolerated without noticeable weight loss. This is in
sharp contrast to Verapamil, which is far more toxic. Presently
then, we are at the stage of a detailed assessment of the relative
advantages of 5-N-acetylardeemin relative to several of the most
promising analogues with respect to further development.
context of a stereoselective solution to the construction of
reverse-prenylated hexahydropyrroloindole alkaloids. A key step
is the N-PSP-induced phenyl selenocyclization of a suitable
tryptophan derivative.28 Remarkably, this reaction gives rise to
very high stereoselection favoring the exo product 20. This result
is, apparently, not the result of equilibration. Rather, it must
reflect a largely kinetic selection. The phenylseleno function is
then replaced with retention of configuration by methyl triflate-
induced alkylation via prenyl tri(n-butyl)stannane. The remain-
ing steps require the formation of three amide bonds, an aza-
Wittig reaction to complete the ardeemin backbone, and the
insertion of an acetate group onto the hindered N5 nitrogen.
Using the strategy outlined for 2, several analogues containing
singular modifications all around the backbone of 2 were
synthesized. This chemistry enabled the production of multigram
quantities of the more promising compounds for further biologi-
cal analysis. We also note that this technology was practiced
recently by Joullie´ and co-workers in a total synthesis of
roquefortine.44 Studies related to the bioefficacy of the ardeemins
will be described in due course.
Acknowledgment. This work was supported by the National
Institutes of Health (Grant Nos. HL25848 (S.J.D.) and CA08748
(Sloan-Kettering Institute). We also acknowledge the Engineer-
ing and Physical Science Research Council (U.K.) for a NATO
Postdoctoral Fellowship for S.P.M. and the Pfizer Corp. for a
Pfizer Predoctoral Fellowship to K.M.D. We thank the Fujisawa
Pharmaceutical Co., Japan, for an authentic sample of amau-
romine and Abbott Laboratories, Abbott Park, IL, for authentic
samples of ardeemin and 5-N-acetylardeemin. We also thank
Professor Crich for communicating his findings to us in advance
of submission. Last, we thank Vinka Parmakovich and Barbara
Sporer for obtaining mass spectral data (Columbia University)
and Dr. George Sukenick for providing spectral and NMR
analyses (S.K.I.).
Supporting Information Available: Experimental proce-
dures and spectral data for compounds 1, 2, 4, 16-23, 25-27,
30, 32, 33, 35, 37, 40, 42-44, 46-50, and 52 as well as a table
listing cytotoxicity and reversal data for 2 and its analogues
(PDF). This information is available free of charge via the
Internet at http://pubs.acs.org.
Summary
JA991558D
We have presented an efficient total synthesis of 5-N-acetyl-
ardeemin (2) (nine steps in 12.5% overall yield from 19) in the
(44) Chen, W.-C.; Joullie´, M. M. Tettrahedron Lett. 1998, 39, 8401.