Synthetic design for combinatorial chemistry. Solution and polymer- supported synthesis of polycyclic lactams by intramolecular cyclization of azomethine ylides

Article abstract of DOI:10.1021/ja9621051

The rapidly expanding field of combinatorial chemistry has stimulated the development of new methods and synthetic strategies for assembly of compound libraries. We propose four criteria that are desirable for synthetic routes to such libraries: (1) the s

Full text of DOI:10.1021/ja9621051

J. Am. Chem. Soc. 1997, 119, 6153-6167
6153
Synthetic Design for Combinatorial Chemistry. Solution and
Polymer-Supported Synthesis of Polycyclic Lactams by
Intramolecular Cyclization of Azomethine Ylides
Matthew A. Marx, Anne-Laure Grillot, Christopher T. Louer,
Krista A. Beaver, and Paul A. Bartlett*
Contribution from the Department of Chemistry, UniVersity of California,
Berkeley, California 94720-1460
ReceiVed June 24, 1996X
Abstract: The rapidly expanding field of combinatorial chemistry has stimulated the development of new methods
and synthetic strategies for assembly of compound libraries. We propose four criteria that are desirable for synthetic
routes to such libraries: (1) the sequence involves a small number of steps; (2) no more than one variable is introduced
in any step; (3) starting materials are readily obtained with a diverse selection of substituents, and (4) cyclic,
nonoligomeric structures represent the most interesting targets. Guided by these criteria, we have explored an
intramolecular version of the azomethine ylide cycloaddition reaction which utilizes readily available amino acids,
aldehydes, and 2° amines as inputs. We prepared a number of cycloadducts in solution to optimize conditions and
examine the scope of the process and to identify a synthetic strategy that would be amenable to solid-phase synthesis
of these compounds. Transfer of the sequence to solid phase was demonstrated by the synthesis of a number of
representative compounds, indicating that the chemistry is suitable for construction of a combinatorial library.
The utility of combinatorial chemistry for the production of
organic compound libraries is widely recognized.1-9 The advent
of techniques for the parallel and combinatorial assembly10 of
large compound libraries is revolutionizing synthesis design,
just as it is altering how compounds are screened for biological
activity. Library candidates are chosen for their ready synthesis
on polymer beads or by automated methods, and “combinatorial
synthetic design” is governed less by retrosynthetic analysis and
more by considerations of what input materials are available
and what can be made from them.8,27,28 Some traditional
strategies for target-directed synthesis are not applicable in this
new context; for example, convergence in a synthetic scheme
cannot be implemented readily when assembly proceeds on a
solid support. This milieu also has profound effects on
fundamental experimental issues such as reagent stoichiometry,
reaction conditions, and removal of side products. The input
materials should embody relatively simple functional group
classes that can be obtained with a broad range of substituents,
and they should not themselves require multistep preparation.
Synthetic libraries were first developed with peptides11,12,29,30
and the related “peptoids”31-33 for just these reasons. However,
the library concept has been extended to hetero- and polycyclic
structures, and it has stimulated the adaptation to solid support
of many procedures for carbon-carbon and carbon-heteroatom
bond formation (inter alia:34-38).
Within this context, three criteria for synthetic design
appropriate for assembly of a combinatorial library by the mix-
and-split strategy can be offered.
X Abstract published in AdVance ACS Abstracts, May 15, 1997.
(1) Moos, W. H.; Green, G. D.; Pavia, M. R. In Annual Reports in
Medicinal Chemistry; Bristol, J. A., Ed.; Academic Press, Inc.: San Diego,
CA, 1993; Vol. 28, pp 315-324.
(1) The sequence inVolVes a small number of steps which
are amenable to solid phase. The synthesis of a large number
of compounds simultaneously is impractical if conventional
purification steps are required in individual reaction workups.
As a result, provision must be made for convenient removal of
reagents and byproducts, and syntheses must be relatively short.
While the syntheses of oligopeptides and oligonucleotides have
been optimized to the point that multistep sequences can be
undertaken effectively, relatively few transformations approach
the efficiency demanded by a procedure that does not allow
purification of the intermediates. Several protocols have been
described for preparation of libraries in solution by short
sequences;8 however, most multistep combinatorial and parallel
synthesis strategies are carried out in a polymer-supported
format. Solid-phase synthesis avoids many common limitations
of solution-phase chemistry: excess reagent(s) can be used
because the product is isolated by simple filtration, and side
reactions can be tolerated if the byproducts are not bound to
the solid support.
(2) Gallop, M. A.; Barrett, R. W.; Dower, W. J.; Fodor, S. P. A.; Gordon,
E. M. J. Med. Chem. 1994, 37, 1233-1251.
(3) Gordon, E. M.; Barrett, R. W.; Dower, W. J.; Fodor, S. P. A.; Gallop,
M. A. J. Med. Chem. 1994, 37, 1385-1401.
(4) Lowe, G. Chem. Soc. ReV. 1995, 309-317.
(5) Terrett, N. K.; Gardner, M.; Gordon, D. W.; Kobylecki, R. J.; Steele,
J. Tetrahedron 1995, 51, 8135-8173.
(6) Thompson, L. A.; Ellman, J. A. Chem. ReV. 1996, 96, 555-600.
(7) Czarnik, A. W. Acc. Chem. Res. 1996, 29, 112-113.
(8) Armstrong, R. W.; Combs, A. P.; Tempest, P. A.; Brown, S. D.;
Keating, T. A. Acc. Chem. Res. 1996, 29, 123-131.
(9) Balkenhohl, F.; von dem Bussche-Hu¨nnefeld, C.; Lansky, A.; Zechel,
C. Angew. Chem., Int. Ed. Engl. 1996, 35, 2288-2337.
(10) We differentiate (1) “combinatorial synthesis”, which generates a
library containing all combinations of the components but in discrete fashion,
for example, on solid support beads, (2) “mixture synthesis”, in which
compounds are produced as molecular mixtures by including multiple
variants of the starting components in each reaction, and (3) “parallel
synthesis”, in which individual compounds are prepared in isolated reaction
vessels or wells. Combinatorial synthesis is exemplified by the Furka “mix-
and-split” strategy,11,12 mixture synthesis as described by Carell et al.,13
and parallel synthesis by the “Diversomer” technology of DeWitt et al.;14
many additional examples can also be found. A large library can be
assembled with fewer individual reactions by the combinatorial and mixture
strategies, in contrast to the parallel approach, but the former methods require
decoding,15-20 indexing,21,22 deconvolution,23-25 or affinity selection26 steps
to identify active compounds.
(11) Furka, A.; Sebestye´n, F.; Asgedom, M.; Dibo´, G. Int. J. Peptide
Protein Res. 1991, 37, 487-493.
(12) Lam, K. S.; Salmon, S.E.; Hersh, E. M.; Hruby, V. J.; Kazmierski,
W. M.; Knapp, R. J. Nature 1991, 354, 82-84 (Corrections Nature 1992,
360, 768).
S0002-7863(96)02105-1 CCC: $14.00 © 1997 American Chemical Society

6154 J. Am. Chem. Soc., Vol. 119, No. 26, 1997
Marx et al.
(2) No more than one Variable is introduced in any step. If
several variables are introduced in one step, then separate
reactions for each combination are required if all are to be
generated without producing molecular mixtures. This criterion
is particularly important for combinatorial syntheses that rely
on encoding strategies for compound identification.15,17-19
(3) Starting materials are readily obtained with a diVerse
selection of substituents. Exotic functional group classes for
which few examples are available, or input materials that must
be synthesized individually, are clearly less practical.
To these three, self-evident criteria, we would add a fourth,
more subjective one.
Scheme 1
(4) Cyclic, nonoligomeric structures represent the most
interesting targets. At the current stage of development in this
field, the assembly of polycyclic molecules on solid-phase poses
a greater challenge than that of acyclic or oligomeric com-
pounds; such targets therefore provide more opportunity for the
development of synthetic strategies and methods. Furthermore,
conformationally constrained molecules are generally more
attractive as lead structures from a biological assay, since they
provide more information about the three-dimensional require-
ments for ligand binding.
In this report, we describe methods for the synthesis of
polycyclic compounds that were devised specifically for ap-
plication to combinatorial chemistry.
The Azomethine Ylide Cycloaddition. The [3 + 2]
cycloaddition of stabilized azomethine ylides to alkenes and
alkynes (Scheme 1)39,40 meets all of the criteria offered above:
it provides a highly substituted, cyclic product under mild
conditions from readily available starting materials such as
aldehydes, amino acids, and alkenes. The mechanism of the
transformation has been studied extensively, and many variations
have been explored in synthesis. Indeed, the use of this
cycloaddition reaction in an intermolecular format has been
described for the parallel synthesis of a number of pyrrolidine
libraries.41-44
Although there are several methods for generating azomethine
ylides, the strategy pioneered by Grigg45 and Tsuge46 is
particularly attractive (Scheme 1). Condensation of an amino
acid with an aldehyde is followed by 1,2-prototropy, in the case
of primary amines, or loss of a proton from the iminium ion
intermediate, when a secondary amine is involved.47 Proton
loss from the tautomeric iminium ion, formed on condensation
of an amine with an R-dicarbonyl compound, is also a logical
route to the stabilized ylide, although few examples of this
process have been described.48 A third route to these ylides
via ring opening of the isomeric aziridines is well-known.49-52
When the ylide is generated by aldehyde-amine condensation
in the presence of the dipolarophile, three components are
brought together in a single reaction. As a result, in a
combinatorial synthesis where one of the components is tethered
to a solid support, the other components cannot be varied
independently, unless the sequence is limited to stabilized imines
that can be carried through the mix-and-split steps. This
limitation is removed if the [3 + 2] cycloaddition is carried out
in an intramolecular format, in which the dipolarophile is
attached to one of the other components (Scheme 2). The
intramolecular reaction has several other attractive features:
significant stereocontrol is anticipated, and the polycyclic
products are capable of a high degree of functionalization and
skeletal variation. Moreover, while the intermolecular reaction
of stabilized ylides is limited to electron-deficient dipolaro-
philes,45 it appears that the intramolecular reaction is feasible
with a wide range of unactivated alkenes and alkynes as the
olefinic partner.40
(13) Carell, T.; Wintner, E. A.; Sutherland, A. J.; Rebek, J., Jr.;
Dunayevskiy, Y. M.; Vouros, P. Chem. Biol. 1995, 2, 171-183.
(14) Hobbs DeWitt, S.; Czarnik, A. W. Acc. Chem. Res. 1996, 29, 114-
122.
(15) Ohlmeyer, M. H. J.; Swanson, R. N.; Dillard, L. W.; Reader, J. C.;
Asouline, G.; Kobayashi, R.; Wigler, M.; Still, W. C. Proc. Natl. Acad.
Sci. U.S.A. 1993, 90, 10922-10926.
(16) Nikolaiev, V.; Stierandova´, A.; Krchna´k, V.; Seligmann, B.; Lam,
K. S.; Salmon, S. E.; Lebl, M. Peptide Res. 1993, 6, 161-170.
(17) Kerr, J. M.; Banville, S. C.; Zuckermann, R. N. J. Am. Chem. Soc.
1993, 115, 2529-2531.
(18) Brenner, S.; Lerner, R. A. Proc. Natl. Acad. Sci. U.S.A. 1992, 89,
5381-5383.
(19) Nestler, H. P.; Bartlett, P. A.; Still, W. C. J. Org. Chem. 1994, 59,
4723-4724.
(20) Geysen, H. M.; Wagner, C. D.; Bodnar, W. M.; Markworth, C. J.;
Parke, G. J.; Schoenen, F. J.; Wagner, D. S.; Kinder, D. S. Chem. Biol.
1996, 3, 679-688.
(21) Pirrung, M. C.; Chen, J. J. Am. Chem. Soc. 1995, 117, 1240-1245.
(22) Depre´z, B.; Williard, X.; Bourel, L.; Herve´, C.; Hyafil, F.; Tartar,
A. J. Am. Chem. Soc. 1995, 117, 5405-5406.
(23) Kurth, M. J.; Ahlberg Randall, L. A.; Chen, C.; Melander, C.; Miller,
R. B. J. Org. Chem. 1994, 59, 5862-5864.
(24) Erb, E.; Janda, K. D.; Brenner, S. Proc. Natl. Acad. Sci. U.S.A.
1994, 91, 11422-11426.
(25) Freier, S. M.; Konings, D. A. M.; Wyatt, J. R.; Ecker, D. J. J. Med.
Chem. 1995, 38, 344-352.
(26) Zuckermann, R. N.; Kerr, J. M.; Siani, M. A.; Banville, S. C.; Santi,
D. V. Proc. Natl. Acad. Sci. U.S.A. 1992, 89, 4505-4509.
(27) Chen, C.; Ahlberg-Randall, L. A.; Miller, R. B.; Jones, A. D.; Kurth,
M. J. J. Am. Chem. Soc. 1994, 116, 2661-2662.
(37) Fru¨chtel, J. S.; Jung, G. Angew. Chem., Int. Ed. Engl. 1996, 35,
17-42.
(28) Ley, S. V.; Mynett, D. M.; Koot, W.-J. Synlett. 1995, 1017-1020.
(29) Geysen, H. M.; Melven, R. H.; Barteling, S. J. Proc. Natl. Acad.
Sci. U.S.A. 1984, 81, 3998-4002.
(38) Hermkens, P. H. H.; Ottenheijm, H. C. J.; Rees, D. Tetrahedron
1996, 52, 4527-4554.
(39) Padwa, A. In ComprehensiVe Organic Synthesis; Semmelhack, M.
F., Fleming, I., Trost, B. W., Eds.; Pergamon Press: Oxford, 1991; Vol.
IV, pp 1069-1109.
(30) Houghten, R. A. Proc. Nat. Acad. Sci. U.S.A. 1985, 82, 5131-
5135.
(31) Simon, R. J.; Kania, R. S.; Zuckermann, R. N.; Huebner, V. D.;
Jewell, D. A.; Banville, S.; Ng, S.; Wang, L.; Rosenberg, S.; Marlowe, C.
K.; Spellmeyer, D. C.; Tan, R.; Frankel, A. D.; Santi, D. V.; Cohen, F. E.;
Bartlett, P. A. Proc. Natl. Acad. Sci. U.S.A. 1992, 89, 9367-9371.
(32) Zuckermann, R. N.; Kerr, J. M.; Kent, S. B. H.; Moos, W. H. J.
Am. Chem. Soc. 1992, 114, 10646-10647.
(33) Zuckermann, R. N.; Martin, E. J.; Spellmeyer, D. C.; Stauber, G.
B.; Shoemaker, K. R.; Kerr, J. M.; Figliozzi, G. M.; Goff, D. A.; Siani, M.
A.; Simon, R. J.; Banville, S. C.; Brown, E. G.; Wang, L.; Richter, L. S.;
Moos, W. H. J. Med. Chem. 1994, 37, 2678-2685.
(34) Chiron Mimotopes Pty. Ltd., Clayton, Victoria, Australia.
(35) Calbiochem-Novabiochem AG, La¨ufelfingen, Switzerland.
(36) Ellman, J. A. Acc. Chem. Res. 1996, 29, 132-143.
(40) Wade, P. A. In ComprehensiVe Organic Synthesis; Semmelhack,
M. F., Fleming, I., Trost, B. W., Eds.; Pergamon Press: Oxford, 1991;
Vol. IV, pp 1111-1168.
(41) Murphy, M. M.; Schullek, J. R.; Gordon, E. M.; Gallop, M. A. J.
Am. Chem. Soc. 1995, 117, 7029-7030.
(42) Obst, U.; Gramlich, V.; Diederich, F.; Weber, L.; Banner, D. W.
Angew. Chem., Int. Ed. Engl. 1995, 34, 1739-1742.
(43) Hamper, B. C.; Dukesherer, D. R.; South, M. S. Tetrahedron Lett.
1996, 37, 3671-3674.
(44) Bicknell, A. J.; Hird, N. W. Bioorg. Med. Chem. Lett. 1996, 6,
2441-2444.
(45) Grigg, R.; Sridharan, V. In AdVances in Cycloaddition; Curran, D.
P., Ed.; JAI Press, Inc.: Greenwich, CT, 1993; Vol. 3, pp 161-204.

Synthetic Design for Combinatorial Chemistry
J. Am. Chem. Soc., Vol. 119, No. 26, 1997 6155
Scheme 2
Scheme 3
a number of strategies can be devised for construction of
secondary allylic and homoallylic amines on solid phase by
alkylation or reductive amination; however, to drive the reaction
to completion without polyalkylation requires that the electro-
phile be attached to the support and the amine be the soluble
component used in excess. The scope of the intramolecular
cycloaddition reaction was explored in solution using a number
of secondary allylic amines that are commercially available or
were synthesized by conventional methods. These substrates
allowed us to evaluate aryl and primary and secondary alkyl
substituents at R1 and a number of allylic and homoallylic groups
for the unsaturated component.
Intramolecular cycloadditions that represent all of the ways
in which the dipolarophile can be connected to the ylide have
been described (Scheme 2).45,50,53,54 As specific examples in
which the olefin has been tethered via the stabilizing moiety
(connection d), the Weinreb51 and Heathcock52,55 groups have
shown that amide-linked ylides, generated by thermolysis of
the aziridine precursors, cyclize to the bicyclic lactams. Padwa50
and DeShong49 and their co-workers have reported a similar
reaction of aziridine esters to give bicyclic lactones. While
aziridine synthesis and pyrolysis may not be an attractive
sequence for solid support, the success of the cycloadditions
themselves suggested that an alternative method for generation
of the ylides would make this reaction applicable to combina-
torial synthesis. This report describes the development of
sequences A and B depicted in Scheme 3 for this purpose, with
an exploration of the various possibilities for substitution of
the bicyclic nucleus and adaptation of the chemistry to solid
support. The two routes are complementary in providing routes
to cycloadducts with different substitution patterns.
The third variable, the secondary amino acid, can be
incorporated as a single unit (e.g., sarcosine and other N-
alkylamino acids, or a cyclic amino acid such as proline) or
itself assembled on solid phase (e.g., by amine displacement of
an R-bromoacylamide). A number of cyclic and acyclic amino
acids were evaluated in the present study. The final variable
component is a nonenolizable carbonyl derivative; aromatic
aldehydes provide the richest source, but potentially reactive
ketones such as isatin and other stabilizing aldehydes such as
glyoxylate offer additional possibilities that were explored.
Design of a Combinatorial Library. To assemble a
combinatorial library on solid support, with introduction of no
more than one variable in each step, requires that one of the
variable components be linked to the solid phase before the next
variable is introduced. For the sequence of Scheme 3, path A,
Both paths A and B lead to the bicyclic pyrrolidino-lactam
skeleton and allow similar variations in the unsaturated second-
ary amine component. Where these two routes differ most
significantly is in the tri- and tetracyclic skeleta that can be
formed. In path A, cyclic amino acyl components (R2 and R3
linked) afford structures with the 3,8-diazatricyclo[6.3.0.01,5]-
2-undecanone backbone (and higher homologs), while cyclic
amines in sequence B (R3 and R4 linked) lead to 1,4-
diazatricyclo[6.3.0.02,6]-3-undecanone as the simplest tricyclic
skeleton.
(46) Kanemasa, K.; Tsuge, O. In AdVances in Cycloaddition; Curran,
D. P., Ed.; JAI Press, Inc.: Greenwich, CT, 1988; pp 99-159.
(47) Confalone, P. N.; Huie, E. M. J. Am. Chem. Soc. 1984, 106, 7175-
7178. Anslow, A. M.; Harwood, L. M.; Phillips, H.; Watkin, D. Tetrahedron
Asymmetry 1991, 2, 169-172.
(48) Ardill, H.; Grigg, R.; Sridharan, V.; Surendrakumar, S.; Thianpa-
tanagul, S.; Kanajun, S. J. Chem. Soc., Chem. Commun. 1986, 602-604.
(49) DeShong, P.; Kell, D. A.; Sidler, D. R. J. Org. Chem. 1985, 50,
2309-2315.
(50) Padwa, A.; Ku, H. J. Org. Chem. 1979, 44, 255-261.
(51) Sisko, J.; Weinreb, S. M. J. Org. Chem. 1991, 56, 3210-3211.
(52) Henke, B. R.; Kouklis, A. J.; Heathcock, C. H. J. Org. Chem. 1992,
57, 7056-7066.
(53) Harwood, L. M.; Lilley, I. A. Tetrahedron Lett. 1993, 34, 539-
540.
The potential size of libraries constructed according to these
schemes with full variation of all the components is vast, hence
an important element in the work we report here was to explore
the limits of this variability. We first examined the scope and
limitations of the intramolecular cycloadditions of Scheme 3,
paths A and B, in solution. Having demonstrated their
feasibility, we then carried out a number of examples on solid-
phase to demonstrate the utility of these sequences for combi-
natorial chemistry.
(54) Martin, S. F.; Cheavens, T. H. Tetrahedron Lett. 1989, 30, 7017-
7020.
(55) Heathcock, C. H.; Clasby, M.; Griffith, D. A.; Henke, B. R.; Sharpe,
M. J. Synlett. 1995, 467-474.

6156 J. Am. Chem. Soc., Vol. 119, No. 26, 1997
Marx et al.
Scheme 4
Figure 1. Crystal structures of bi- and tricyclic products.
Results and Discussion
Solution-Phase Chemistry: Demonstration of the Cycload-
dition via Path A. The feasibility of the cycloaddition was
first demonstrated by combining three commercially available
components: N-allylaniline, N-Boc-sarcosine, and benzaldehyde
(Scheme 4). An equimolar solution of amine 3s and benzal-
dehyde in toluene was heated at reflux under a Dean-Stark trap
to provide the bicyclic lactam 4s in 35% yield after 20 h. The
sequence with proline proceeded similarly to give the tricyclic
lactam 4p in 74% yield. Small-scale reactions (ca. 0.5 mmol
of amine) in the presence of molecular sieves or without any
explicit water removal showed no effect on the yield; therefore,
subsequent reactions were run without rigorously anhydrous
reaction conditions. The choice of solvent, however, had a
dramatic effect on the rate of reaction. Reactions carried out
in refluxing benzene proceeded much more slowly than those
in toluene, and attempted reactions in acetonitrile or ethylene
glycol dimethyl ether returned only starting materials. Further-
more, the addition of amine bases such as triethylamine or DBU
did not affect the reaction outcome. However, since an added
amine base was found to be crucial for success of the
glyoxamide strategy (Scheme 3, path B, described below), an
equivalent of triethylamine was generally included to maintain
complementarity between the two protocols.
A byproduct of the cyclization of 3s with benzaldehyde is
the 2:1 adduct 5. With a 2:1 ratio of benzaldehyde to amine
and a concentration of 0.5 M amine in toluene, the 2:1 adduct
is produced in approximately equivalent amount to the desired
bicyclic product. Submission of oxazolidine 5 and bicycle 4p
individually to refluxing toluene, with or without catalytic tosic
acid, showed that their formation is irreversible under the
Figure 2. Possible Ylide and Transition State Geometries.
stereochemistry is consistent with two possibilities for the
transition state structure of the sarcosine substrate: the Z,E-
and E,Z-ylide geometries, with endo and exo orientations of
the allyl group, respectively (Figure 2). However, only the E,Z-
exo geometry is possible for the proline substrate or, in the
alternative glyoxamide format, for cyclic amines like tetrahy-
droisoquinoline (see below). Moreover, if the endo transition
state were possible for the acyclic substrates, one might expect
some of the other C-phenyl stereoisomer from the E,E-ylide,
although that isomer is less reactive.39
1
reaction conditions. Analysis of the H NMR spectra arising
from a study of the concentration dependence of this reaction
revealed that a 2:1 molar mixture of aldehyde to amine and a
reaction concentration of 0.5-0.6 M (of amine in toluene)
provided the most consistent results, with the 2:1 and the 1:1
adduct being produced in approximately equimolar amounts and
with total mass balances of 60-70%.
The cycloadducts 4s and 4p are formed as single diastereo-
mers, with the cis ring fusion and the exo configuration of the
C-phenyl substituent according to 1D 1H NOE difference
spectroscopy. These assignments were confirmed with crystal
structures of the analogs 6 and 7 (Figure 1). The relative
Variation of the Amine Substituent. Using sarcosine as
the secondary amino acid and a simple allyl group as the
unsaturated component, we found that the reaction tolerates
primary and secondary alkyl substituents on the amide nitrogen,
in addition to aryl (entries 1-3, Table 1), but not hydrogen

Synthetic Design for Combinatorial Chemistry
J. Am. Chem. Soc., Vol. 119, No. 26, 1997 6157
Table 1. Solution-Phase Cyclization of Unsaturated R-Aminoacylamides (Path A)
entry
R1
amino acid
Sar
Sar
Sar
Pro
Sar
Pro
Sar
Pro
n
R2
R3
yield (%)
(compd)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Ph
1
1
1
1
1
1
2
2
2
1
1
1
1
1
1
1
1
1
1
H
H
H
H
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
35
50
57
74
49
63
45
32
16
66a
60
47
30
27
75
26
30
21b
27
4s
benzyl
cyclohexyl
Ph
4p
7
Ph
Ph
phenyl
phenyl
H
benzyl
benzyl
Ph
benzyl
cyclohexyl
cyclohexyl
Ph
H
H
H
H
H
H
H
H
H
H
H
H
Sar
N-methylalanine
Sar
Sar
Sar
Sar
Pro
Sar
Sar
Sar
Sar
p-F-phenyl
p-MeO-phenyl
p-NO2-phenyl
2-naphthyl
p-MeO-phenyl
furyl
6
Ph
Ph
Ph
Ph
Ph
Ph
BuO2C
MeO2C, Me
2-oxoindolin-3-ylidene
18
19
a Reaction carried out in refluxing xylenes. b Product isolated as a 1.1:1 mixture of isomers.
itself. Although azomethine ylides derived from secondary
amides by aziridine thermolysis undergo intramolecular cy-
cloaddition,52 condensation of the sarcosyl or prolyl propargyl
amides with benzaldehyde simply leads to the imidazolidinones
8.
the propargyl and 3-butenyl groups were not successful in the
present context, leading to decomposition. The sarcosine analog
did afford a low yield of the unsaturated adduct 11, which
aromatized to the pyrrole 12 on standing, as DeShong et al.
observed for the related lactones.49 We also prepared the
sarcosyl amide of benzylmethylamine to investigate the aromatic
ring itself as the dipolarophile. Although we had seen no
evidence for cyclization onto the benzene ring in any of the
other benzylamine analogs we had studied, we were prompted
by the observations of Heathcock and co-workers52 of cycload-
dition of related ylides to the benzene ring. However, no cyclic
1
product was apparent in the H NMR spectrum of the crude
reaction mixture from sarcosyl N-benzyl-N-methylamide.
Variation of the Unsaturated Component. In addition to
the simple allyl group, several substituted allyl, homoallyl, and
propargyl derivatives were investigated as the unsaturated
component. Among the allyl analogs, phenyl substitution at
the terminal position is well tolerated (entries 5 and 6, Table
1); the ready availability of cinnamyl amines indicates that this
particular substitution pattern provides a rich opportunity for
diversity. Neither methyl nor methoxycarbonyl are tolerated
at the 2-position, the former because of apparent lack of
reactivity (no reaction observed in refluxing toluene) and the
latter because of intramolecular cycloaddition to give 9 on
deprotection of the secondary amine (84% yield). The 3-butenyl
group affords bicyclic lactams with the 5,6-ring system (entries
7 and 8), but when combined with the N-phenyl substituent and
sarcosine (entry 9), the cyclization proceeds in poor yield with
formation of considerable amounts of the 2:1 adduct. Cycload-
dition was not observed with the substituted homoallylic
derivative 10.
Variation of the Amino Acid. Aside from sarcosine and
proline, only modest variation in the amino acid component
appears to be tolerated. N-Allylaniline and benzaldehyde were
employed as the amine and carbonyl components, respectively.
The N-methylalanine derivative (Table 1, entry 10) required
higher temperatures than the sarcosine analog; however, the
reaction proceeded in better yield, and no evidence of an
oxazolidine byproduct was observed. Furthermore, the reaction
results in the stereospecific formation of a tetrasubstituted carbon
atom, as also occurs in cyclizations with proline (entries 4, 6,
and 8). Interestingly, pipecolinic N-allylanilide was unreactive
under the normal conditions, indicating that formation of the
iminium ion may be sterically disfavored in this case. Four
N-alkylglycines in addition to sarcosine were also evaluated,
Substituted acetylenes are also effective dipolarophiles for
intermolecular azomethine ylide cycloadditions;49,56 however,
(56) Grigg, R.; Gunaratne, H. Q. N.; Kemp, J. Tetrahedron 1990, 46,
6467-6482.

6158 J. Am. Chem. Soc., Vol. 119, No. 26, 1997
Marx et al.
above, this appears to be the wrong configuration for the desired
cyclization reaction. In an attempt to provide an alternative
means of stabilizing the iminium hydrogen, we investigated
salicylaldehyde as the carbonyl component; we reasoned that
the hydrogen-bonding scheme of 17 would allow the E,Z-ylide
to be formed. Unfortunately, all attempted cyclization reactions
involving imine 16 resulted in decomposition of the starting
material. It is clear that an alternative strategy will be required
to employ primary amino acids in this intramolecular cyclization
process.
Figure 3. Azomethine Ylide Formation by Prototropy or Metal-Cation
Variation of the Carbonyl Component. We examined the
cyclization with a number of different carbonyl components
using the sarcosine-derived N-phenyl (3s) and N-cyclohexylallyl
amides and in one case proline derivative 3p (Table 1, entries
11-19). A number of the products are particularly noteworthy.
The bicyclic adduct 6 crystallized from the reaction mixture,
and an X-ray structure verified the cis-relationship of the ring
junction protons and the aryl substituent (Figure 1).
Complexation.
Scheme 5
In addition to the aromatic aldehydes (Table 1, entries 1-16),
which were generally effective in the cyclization, three non-
aromatic carbonyl components, butyl glyoxylate, methyl pyru-
vate, and isatin (entries 17-19) also provided cyclic products.
In all of these compounds, the carbonyl group is activated by
conjugation to an aromatic ring or an R-carbonyl group; in our
hands, formaldehyde (as paraformaldehyde) or alkyl aldehydes
failed to afford cycloadducts. The reactions of 3s with butyl
glyoxylate and methyl pyruvate (entries 17 and 18) provide
points of cross-reference with the alternative strategy of path B
(Scheme 3), since the same ylides are available from condensa-
tion of the N-allylglyoxamide with sarcosine and N-methyl-
alanine esters (see below).
The two ketones, methyl pyruvate and isatin, afford cycload-
ducts with a quaternary center. In the reaction with methyl
pyruvate (entry 18), 18 is formed as a mixture of stereoisomers,
while isatin affords a single spirocycle 19 of undetermined
geometry (entry 19). Cyclization of the pyruvyl and isatin
derivatives prompted an investigation of the reactivity of a
number of other ketones, such as cyclohexanone and benzophe-
none; however, in each case, unreacted starting material was
recovered from the reaction mixture. Steric as well as electronic
factors presumably inhibited formation of the iminium ion
intermediate.
since they are readily accessible by the strategy employed by
the Chiron group for the “submonomer” assembly of peptoids.32
However, the N-phenyl, N-benzyl, N-cyclohexyl, and N-butyl
analogs all failed to give cyclic products under the standard
conditions. Complex product mixtures were observed, and no
starting material was recovered, which led us to believe that
the imine was being formed but that ylide formation or
cyclization was for some reason disfavored in these cases.
Attempted Reactions with Primary Amino Acids. The
abundance of commercially available primary amino acid
derivatives as well as the opportunity for further derivatization
that the secondary amine in the cycloadduct would afford
prompted us to try to incorporate these analogs in the synthetic
sequence. Grigg and co-workers have demonstrated ylide
formation from primary amino acid imines both by prototropy
and by metal complexation; in these cases, the imine is activated
by protonation or complexation in concert with the carbonyl
oxygen, as shown in Figure 3.45,57
Reaction of the glycine-derived amide 13 with benzaldehyde
under dehydrating conditions readily afforded imine 14 (Scheme
5); however, we were unable to induce this material to undergo
cyclization, either thermally, in the presence of silver acetate
or lithium bromide or with triethylamine as catalyst. Reaction
of 14 with benzaldehyde in the presence of silver acetate or
lithium bromide, as reported by Grigg for related intermolecular
reactions,45 resulted in decomposition at high temperatures and
recovery of starting material at lower temperatures. Combina-
tion of 14 directly with benzaldehyde in toluene at reflux
produced similar results, either with or without a catalyst.
Although an ylide is probably forming under these conditions,
it is likely to have the E,E-geometry depicted in 15; as discussed
Cycloaddition via Path B: Glyoxamide Synthesis, Azo-
methine Ylide Generation, and Cyclization. As described
above for path A, the feasibility of the condensation and
cyclization steps of path B (Scheme 3) were first verified in
solution phase experiments.58 The symmetrical tartramide 20
was prepared from diethyl tartrate as shown in Scheme 6 and
cleaved with periodic acid to afford the glyoxamide 21.59 This
oxidation is most conveniently carried out in 1:1 tetrahydrofuran/
ether from which the inorganic byproducts precipitate at the
end of the reaction; the crude glyoxamide is obtained in 80%
yield and can then be carried on without further purification.
Combination of this material with 1 equiv of 1,2,3,4-tetrahy-
(58) The phenylglyoxamide analog (PhCOCON(allyl)Ph) was studied
briefly as another readily accessible, nonenolizable R-ketoamide; however,
no reaction of this compound was observed even on prolonged heating with
the activated 2° amines that are successful with the simple glyoxamide
analogs.
(57) Grigg, R.; Gunaratne, H. Q. N.; Sridharan, V. Tetrahedron 1987,
43, 5887-5989.
(59) Stetter, H.; Skobel, H. Chem. Ber. 1987, 120, 643-645.

Synthetic Design for Combinatorial Chemistry
J. Am. Chem. Soc., Vol. 119, No. 26, 1997 6159
Scheme 6
Figure 4. NOE Interactions Observed for Cycloadducts.
Scheme 7
droisoquinoline in refluxing toluene for 16 hours afforded the
tetracyclic lactam 22 as a single diastereomer in 50% yield.
Addition of one equivalent of triethylamine to the reaction
mixture increased the yield to 72%; however, the yield was not
affected when molecular sieves were added; these yields are
based on the symmetrical tartramide and encompass the
oxidation, condensation, and cyclization steps.
The geometry of this alternative process was revealed by the
stereochemistry of two homoallyl and substituted allyl analogs
23 and 24 (Figure 4). NOE and coupling interactions estab-
lished the anti relationship between the ring fusions, which arises
from the same anti,syn-exo transition state demonstrated for the
related cycloadditions of Scheme 3, path A (see above). The
versatility of this sequence was determined by examining the
cycloaddition reaction with a variety of glyoxamides and amines.
Variation of the Amine Component. The variability
tolerated in the amine component was evaluated with N-allyl-
N-phenylglyoxamide 21 as the dicarbonyl partner. As illustrated
by Table 2 (entries 1-7), cyclic and acyclic R-amino acid esters
are effective in this cyclization procedure. The acyclic amino
esters afford the same ring system and some of the same
products that are available from the alternative sequence of path
A; indeed, the butyl ester of 25 (entry 2) is described above
(Table 1, entry 17). Steric hindrance plays an important role
in this condensation: while sarcosine ethyl ester affords the
bicyclic product 25 in 65% yield (as a 5:1 mixture of the exo:
endo isomers, Table 2, entry 2), the N-methylalanine ester gives
the quaternary product 26 in only 25% yield (entry 3). The
yield of the latter process is reduced by transamination of the
iminium intermediate to produce ethyl pyruvate and the sar-
cosine amide, which then condenses with starting glyoxamide
to give 27 (Scheme 7).
The condensation with methyl glycinate itself requires
refluxing xylene and only gives the bicyclic adduct in poor yield
(Table 2, entry 4). This result is similar to that obtained with
unsubstituted amino acids in the alternative path A, which we
attribute to preferential formation of the unproductive, anti-
anti isomer of the ylide (see above).39 In an attempt to expand
the scope of the methodology, we investigated the generation
of azomethine ylides via decarboxylation.60,61 We first at-
tempted to react glyoxamide 21 with sarcosine in the presence
of triethylamine, but the condensation was stymied because the
amino acid is insoluble. However, the reaction can be ac-
complished with the trimethylsilyl ester, as shown in entry 5;
condensation with bis(trimethylsilyl)sarcosine affords the bi-
cyclic pyrrolidine lactam 28 without any substituent at the
7-position (Scheme 8). This strategy appears to have limited
Scheme 8
generality, however, since no cycloadduct was obtained with
the N-methylalanine analog.
Perhaps the most interesting among the secondary amine
condensations are those involving cyclic derivatives, since these
components lead to novel ring systems that are not accessible
by path A (Table 2, entries 1, 6, and 7). We were not successful
in inducing unactivated amines such as pyrrolidine or piperidine
to react. In spite of the electron withdrawing effect of the amide
carbonyl, ylide formation in refluxing toluene requires an
additional activating substituent R to the nitrogen, such as the
carboxyl group of the amino esters or the aromatic ring of
tetrahydroisoquinoline. Moreover, cycloadducts were not ob-
tained from the acyclic analogs, benzyl amine, or methyl benzyl
amine.
Variation of the Amine Substituent. In addition to phenyl,
benzyl and cyclohexyl groups were evaluated as the nonpar-
ticipating amide substituent (Table 2, entries 8 and 9). Good
yields were obtained on condensation of the N-phenyl and
N-benzyl amides with tetrahydroisoquinoline, although the yield
of cycloadduct was significantly lower when the cyclohexyl
derivative was condensed with sarcosine ethyl ester.
Variation of the Unsaturated Component. The unsaturated
component proved to be a fruitful position for introducing
(60) Tsuge, O.; Kanemasa, S.; Ohe, M.; Takenaka, S. Chem. Lett. 1986,
973-976.
(61) Grigg, R.; Idle, J.; McMeekin, P.; Surendrakumar, S.; Vipond, D.
J. Chem. Soc., Perkin Trans. I 1988, 2703-2713.

6160 J. Am. Chem. Soc., Vol. 119, No. 26, 1997
Marx et al.
Table 2. Solution-Phase Cyclization of Unsaturated Glyoxamides (Path B)
entry
R1
n
R2
R3
R4
amine
yield (%)
(compd)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Ph
Ph
Ph
Ph
Ph
Ph
Ph
1
1
1
1
1
1
1
1
1
2
2
2
1
1
1
1
1
1
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
THIQ
72
65a
25b
25c
40
61
40
72
37
58
64
66d
56
62
52
35
56
50
22
25
26
sarcosine Et ester
N-Me-Ala Et ester
Gly Me ester
N-TMS-sarcosine TMS ester
Pro Me ester
picolinic Me ester
THIQ
THIQ
THIQ
THIQ
sarcosine Et ester
THIQ
sarcosine Et ester
Pro Me ester
THIQ
Pro Me ester
THIQ
28
benzyl
cyclohexyl
Ph
benzyl
benzyl
benzyl
Ph
Ph
Ph
Ph
Ph
23
24
isopropyl
H
H
H
H
H
methyl
methyl
CO2Me
CO2Me
H
phenyl
a 5:1 exo:endo. b Accompanied by 25% of 27. c This reaction carried out in refluxing xylenes. d 2.6:1 exo:endo.
Table 3. Cyclization of Propargylic Amides
diversity into the target molecules, because considerable varia-
tion in this unit is allowed (Table 2, entries 10-18). The
homoallyl analogs (entries 10-12) cyclized with comparable
efficiency to the lower homologs (entries 1 and 2), affording
the 5,6-bicyclic lactams. On the allyl group itself, substitution
is tolerated at the allylic (entry 13), central (entries 14-17), and
terminal positions (entry 18). In the R-isopropyl analog (entry
13), the stereocenter directs the course of the cycloaddition so
that this group adopts the exo-configuration in the product, as
assigned from NOE difference experiments (Figure 4). Suc-
cessful cyclization of the carbomethoxy-substituted analog
(entries 16 and 17) contrasts with the results obtained in the
alternative route to ylide formation, in which intramolecular
Michael addition from the aminoacyl group intervened on
N-deprotection (see above).
Cyclization is also achieved with propargylic substituents
(Table 3), but a variety of side reactions lowered the isolated
yields of pyrroline product. Condensation of sarcosine ethyl
ester or THIQ with the simple propargyl and 2-butynyl
glyoxamides afforded low yields of pyrrolines, accompanied
by significant amounts of the pyrrole oxidation products. The
proline ester, in contrast, affords a pyrroline intermediate that
undergoes ring-opening to the azacyclooctadiene isomer 30
(entry 3), perhaps via electrocyclic ring-opening of ylide 29
(Scheme 9).
Alternative Method for Glyoxamide Preparation. Al-
though oxidative fragmentation of symmetrical tartramides is
an efficient source of glyoxamides for scouting solution phase
cycloadditions, it was not suitable for solid phase synthesis. An
effective alternative was Swern oxidation of the corresponding
glycolamides, which in turn are available from coupling the
unsaturated amine with acetoxyacetic acid, followed by metha-
nolysis of the ester (Scheme 10). The crude product, submitted
without purification to the condensation with THIQ, afforded
tetracyclic material in 64% yield, based on the glycolamide.
The slightly lower yield may be due to contamination of the
glyoxamide with triethylamine salts from the oxidation.
Comparison of the Two Methods of Ylide Generation. We
were curious whether the unsymmetrical, anti-syn ylide
required for cyclization was fully equilibrated prior to ring
formation or whether a kinetic preference is manifested on
condensation of equivalent components, e.g., a sarcosine amide
with a glyoxamide such that one carbonyl takes the syn position
and the other the anti. The latter possibility was ruled out by
two parallel experiments: condensation of sarcosine N-(allyl)-
aniline with N-(homoallyl)glyoxanilide versus condensation of
sarcosine N-(homoallyl)anilide with N-(allyl)glyoxanilide (Scheme
11). Both reactions gave the same 5,5-product in comparable
yields, thus showing that the ylides are fully equilibrated and
also that cyclization onto the allyl group occurs more rapidly
than onto the homoallyl moiety.
Solid-Phase Chemistry. The results of the solution-phase
experiments outlined above served to delineate the general steric
and electronic requirements of the cyclization reactions and
provided guidelines for choosing the unsaturated amines, amino
acids, and carbonyl components or secondary amines that could

Synthetic Design for Combinatorial Chemistry
J. Am. Chem. Soc., Vol. 119, No. 26, 1997 6161
Scheme 9
placement with an excess of allylamine afforded the secondary
amine 33, which was coupled to Fmoc-proline with DIC to give
the amide 34; small samples of this resin (ca. 10 mg) were
treated with piperidine to determine the Fmoc loading level by
colorimetric analysis.65 In addition, after each of the chemical
steps described above, small portions (ca. 50 mg) of function-
alized resin were treated with TFA and the material liberated
1
was analyzed by H NMR; at each point, the major product
visible in the spectrum of the crude material was the desired
compound.
The free aminoamide resin 35 was carried on to the
cycloaddition reaction. Use of either 2 or 10 equiv of
benzaldehyde in refluxing toluene did not change the amount
of tricyclic compound isolated upon cleavage. Moreover,
although the Kaiser test66,67 was not suitable for monitoring the
disappearance of the secondary amine, the chloranil test68,69
proved to be quite effective and allowed us to determine the
reaction progress qualitatively. Cleavage of the tricyclic product
from the resin afforded free acid 37 as a red oil; however,
analysis of the 1H NMR spectrum of the crude material clearly
indicated the presence of a single cycloadduct, with only minor
amounts of other products. To facilitate purification of the
zwitterionic product, the methyl ester was prepared using EDC/
DMAP/MeOH and purified by chromatography to give 38 as
an analytically pure solid in 60% overall yield from 34.
Scheme 10
go into a library. The next step was to demonstrate that the
general reaction schemes could be adapted to the solid-phase
format. This “transfer” involves two distinct stages: first,
devising a strategy for the assembly that would provide the
greatest flexibility (and maximize diversity) and that would take
advantage of the particular benefits of solid-phase format;
second, preparing individual compounds in a batchwise fashion
to find resins and conditions that would ultimately be suitable
for library construction.
A number of strategies are available for linking a synthetic
substrate to the solid phase. Two potential routes for construc-
tion of the secondary amines on the solid support are shown in
Scheme 12. While the commercial availability of a wide variety
of unsaturated bromides makes alkylation of a resin-bound
amine attractive, this sequence would not take advantage of one
of the primary advantages of the solid-phase format, namely
the ability to use an excess of reagent, since quaternization of
the resin-bound amines would present a problem. We therefore
pursued the alternative: displacement of a resin-bound halide.
Cycloadducts from path A (Scheme 3) were synthesized on
solid phase as outlined in Scheme 13. Initial experiments were
performed with the Rink amide derivatives of both conventional
polystyrene resin62 and the poly(ethyleneglycol)-derivatized
(Tentagel) resin.63 Although these resins were ultimately
abandoned in favor of the Wang resin,64 our initial studies led
to the formation of compounds 31s and 31p and indicated the
feasibility of the reaction scheme. The Wang resin, in which
the functionality is a p-alkoxybenzyl alcohol,64 is commercially
available and inexpensive, and it can be obtained with high
loading levels (1 mmol/g).
One practical aspect of these reactions is noteworthy. While
the coupling and deprotection steps are carried out in standard
peptide synthesis flasks and agitated with a wrist-action shaker,
the amine displacement and cycloaddition reactions are carried
out in conventional round-bottom flasks equipped with magnetic
stirrers and warmed with heating mantles. Although we were
concerned about the mechanical stability of these resins,
examination of an aliquot of resin beads after each of these steps
using a dissection microscope revealed that fewer than 1% of
the beads had fractured during the course of the reaction.
This synthesis indicated that the solution-phase methodology
could be transferred readily to the solid phase and afford
comparable yields of cycloadducts. Using the strategy outlined
in Scheme 14, we prepared a small library of compounds using
the [3 + 2] cycloaddition reaction. These compounds were
prepared in batch syntheses using a combination of two amines,
two amino acids, and two carbonyl components to give eight
separate compounds.
A single batch of Wang resin was functionalized with
chloromethylbenzoic acid, split in half, and allowed to react
with either allyl- or homoallylamine to give 33 (allyl) and 33
(homoallyl). Although this step was not monitored directly,
prolonged reaction times (>12 h) and vigorous stirring proved
to be important in ensuring that this step proceeded in >90%
yield, as determined by Fmoc quantitation after the subsequent
aminoacylation step. Each of these resin portions was split and
coupled with either Fmoc-Sar-OH or Fmoc-Pro-OH, affording
the four cyclization precursors 34. Fmoc quantitation of samples
from each of the four resin batches at this point indicated that
the loading levels were about half of the nominal loading. Each
batch of the free amines 35 was then split and reacted with
2-furaldehyde or benzaldehyde to give 36, which was cleaved
from the resin and treated with MeOH/EDC/DMAP or with
diazomethane to form the methyl esters 38. The yields for the
eight cycloadducts shown in Table 4 represent the yield of
purified material, based on the resin loading at the Fmoc-amino
acid step. Although the yields for the esters isolated from the
(62) Rink, H. Tetrahedron Lett. 1987, 28, 3787-3790.
(63) Bayer, E. Angew. Chem., Int. Ed. Engl. 1991, 30, 113-129.
(64) Lu, G.-S.; Mosjov, S.; Tam, J. P.; Merrifield, R. B. J. Org. Chem.
1981, 46, 3433-3436.
Ester 32 was prepared by esterification of Wang resin with
p-chloromethylbenzoic acid and diisopropylcarbodiimide (DIC)
in the presence of 4-(dimethylamino)pyridine (DMAP). Dis-

6162 J. Am. Chem. Soc., Vol. 119, No. 26, 1997
Marx et al.
Scheme 11
Scheme 12
Scheme 14
Scheme 13
above sequence are variable, they are comparable to or slightly
better than those obtained in solution. Some of this variability
may result from the esterification and isolation steps after
cleavage from the resin.
For translation of the cycloadditions via path B to solid phase,
we chose polystyrene resin with the Rink linker, an acid-labile
trialkoxybenzhydryl amine, as the tether point. The possibility
of an ester linkage to a Wang resin (p-alkoxybenzyl alcohol as
the tether point)64 was abandoned when it proved to be labile
to the conditions for methanolysis of the acetate prior to
oxidation. The sequence for solid-phase assembly was therefore
demonstrated as illustrated in Scheme 15, which involved the
same steps for formation of the secondary allylic amine as
employed for path A.
p-(Chloromethyl)benzoic acid was coupled to the Rink
linker62 with DIC and DMAP, and the halide was displaced
with excess allylamine with iodide catalysis. Acetoxyacetic acid
was then coupled to the resin-bound allylic amine 40 with DIC
and DMAP. Acetate hydrolysis was accomplished with potas-
sium carbonate in methanol and DMF, with the latter solvent
required to swell the polystyrene resin. The resin-bound
hydroxyacet amide 42 was oxidized cleanly to the glyoxamide
43 using the conditions developed by Swern.70 The dimethyl
sulfide and amine salt byproducts of this oxidation were
removed by washing the resin with several portions of dichlo-
romethane. Treatment of the resulting glyoxamide-derivatized
resin with THIQ and triethylamine in refluxing toluene for 18
(67) Kaiser, E.; Colescott, R. L.; Bossinger, C. D.; Cook, P. I. Anal.
Biochem. 1970, 34, 595-598.
(68) Kirchner, J. G. In Techniques of Chemistry, 2nd ed.; Wiley-
Interscience: 1978; Vol. XIV, p 209.
(69) Vojkovsky, T. Pept. Res. 1995, 8, 236-237.
(70) Mancuso, A. J.; Swern, D. Synthesis 1981, 165.
(65) Atherton, E.; Sheppard, R. C. In The Peptides; Udenfried, S.,
Meienhofer, J., Eds.; Academic Press: Orlando, FL, 1987; Vol. 9, pp 1-38.
(66) Sarin, V. K.; Kent, S. B. H.; Tam, J. P.; Merrifield, R. B. Anal.
Biochem. 1981, 117, 147-157.

Synthetic Design for Combinatorial Chemistry
J. Am. Chem. Soc., Vol. 119, No. 26, 1997 6163
Table 4. Yields of Cycloadducts from Solid-Phase Synthesis,
Path A
Scheme 15
entry
amino acid
n
R3
yielda (%)
1
2
3
4
5
6
7
8
Sar
Sar
Pro
Pro
Sar
Sar
Pro
Pro
1
2
1
2
1
2
1
2
phenyl
phenyl
phenyl
phenyl
furyl
furyl
furyl
32
29
63
35
75
39
49
52
furyl
a Yield of purified product, based on resin loading of Fmoc-amino
acid; post cleavage esterification with diazomethane except for entries
5, 7, and 8 (MeOH/EDC/DMAP).
Table 5. Cycloadducts from Solid-Phase Synthesis, Path B
unsaturated amine
A: allyl B: homoallyl C: cinnamyl
yielda (%)
secondary amine
(n ) 1,
(n ) 2,
(n ) 1,
R4 ) H)
R4 ) H)
R4 ) phenyl)
1: tetrahydroisoquinoline
2: sarcosine Et ester
3: proline Me ester
45
48b
34
30
30c
38
18
25
20
a Yields reported for material purified by column chromatography
and based on the nominal loading on the commercial resin. b 12:1 exo:
endo. c 3:1 exo:endo.
h, followed by trifluoroacetic acid-mediated release of the
products from the resin afforded bicyclic lactam 45 in 45%,
based on the nominal loading of the commercial resin, after
purification by column chromatography. This yield translates
into an average 89% per step, demonstrating that the sequence
illustrated in Scheme 8 was readily adapted to solid-phase
synthesis. The potential utility of the reaction sequence for
construction of a library of bicyclic lactams was evaluated by
synthesizing nine analogs in parallel fashion, as illustrated in
Scheme 16.
A sample of the p-(chloromethyl)benzoyl resin was divided
in three portions, and each was treated with one of three
unsaturated primary amines: allylamine, homoallylamine, and
cinnamylamine. The resin samples were coupled with R-ac-
etoxyacetic acid, deacetylated to afford the glycolamide resins,
and then further subdivided into three portions. The nine resin
batches were oxidized separately and condensed with one of
three secondary amines: THIQ, sarcosine ethyl ester, and proline
methyl ester. The final products were isolated by cleavage from
the resin with 95% TFA-water to give the bicyclic lactams
shown in Table 5.
Conclusions
The implementations of the intramolecular [3 + 2] azo-
methine ylide cycloaddition described above afford a variety
of functionalized, polycyclic lactams in solution and on solid
phase. The starting components are simple materials that are
readily available, the cyclization process tolerates considerable
variability among the substituents, and the two routes provide
a number of ring systems and substitution patterns on the
bicyclic pyrrolidine skeleton. Ten different ring systems with
the pyrrolidinolactam core have been prepared by these methods,
of which to our knowledge only the two bicyclic examples51,52
are represented in compounds previously described (Figure 5).
The two assembly sequences satisfy the criteria proposed for
an effective combinatorial synthesis and show considerable
1
The crude products were analyzed by H NMR, which, for
the allyl- and homoallylamine-derived products (Table 5,
columns 1 and 2), showed the exclusive presence of the
cycloaddition products. The yields for these cases were
determined after column chromatography and correspond to 80-
90% per step for the six reactions from commercial resin to
TFA-cleavage. In contrast, the yields were lower for the
cinnamylamine-based products (column 3). In these cases, 1H
NMR analysis of the crude products revealed the presence of
significant amounts of uncyclized materials, suggesting that
conversion of the substituted alkene requires a longer reaction
time than the unsubstituted derivatives.

6164 J. Am. Chem. Soc., Vol. 119, No. 26, 1997
Marx et al.
Scheme 16
Rf ) 0.27 (1:1 hexanes/EtOAc); IR 1691, 1496, 1402, 1304 cm-1; 1H
NMR δ 7.76 (m, 2), 7.55-7.11 (m, 8), 4.38 (dd, 1, J ) 5.1, 11.5),
4.09 (m, 1), 3.50 (dd, 1, J ) 4.1, 10.1), 2.75 (m, 1), 2.66 (m, 1), 2.53
(m, 1), 2.45 (m, 1), 2.34 (m, 1), 2.04 (m, 2), 1.75 (m, 2); 13C NMR δ
175.9, 139.4, 138.6, 128.7, 128.5, 128.0, 127.2, 124.5, 119.7, 79.8,
64.4, 51.8, 49.1, 40.1, 35.9, 35.5, 26.4; HRMS (FAB) calcd for
C21H23N2O (MH+): 319.1810. Found 319.1802.
N-Allyl-N-phenyl 3-Methyl-2,5-diphenyl-4-oxazolidinecarboxa-
mide (5). A mixture of amine 3s (83 mg, 0.40 mmol; see Supporting
Information), benzaldehyde (84 mg, 0.80 mmol), and Et3N (55 µL,
0.40 mmol) in toluene (3 mL) was treated according to the general
cyclization procedure to yield an oil which was purified by chroma-
tography (3:2 hexanes/EtOAc v:v) to give two products. The slower-
eluting fraction (21 mg, 25%) was bicycle 4s; the more quickly eluting
fraction (39 mg, 46%), 2:1 adduct, 5, was obtained as an inseparable
mixture of isomers: TLC Rf ) 0.63 (3:2 hexanes/EtOAc v:v); IR 1663,
1493 cm-1; 1H NMR δ 7.60 (d, J ) 6.5) and 7.57-7.17 (m) (15 total);
6.03-5.77 (m, 1), 5.42 (d, J ) 8.5) and 5.35 (d, J ) 5.0) (one total);
5.15-4.69 (m, 2); 4.34 (d, J ) 6.2), 4.32 (dd, J ) 6.4, 15.6) and 4.22
(dd, J ) 6.5, 14.4) (two total); 3.28 (d, 0.5, J ) 8.5), 2.32 (s), 2.18 (s),
2.16 (s), and 2.13 (s) (three total); 13C NMR δ 167.7, 140.9, 138.9,
138.8, 134.2, 132.4, 129.6, 129.3, 129.1, 128.9, 128.8, 128.7, 128.5,
128.3, 128.2, 127.8, 127.7, 127.0, 126.6, 118.3, 108.7, 99.9, 97.1, 83.3,
79.0, 72.8, 52.9, 52.5, 36.3, 26.0; HRMS (FAB) calcd for C26H27N2O2
(MH+) 399.2072, found 399.2064.
promise for the preparation of libraries of interesting and
previously unknown structures.
Experimental Section71
General Procedures. Standard Procedure for Cycloadditions of
Aminoacyl Amides in Solution. The amine (1 equiv), carbonyl
derivative (2 equiv), and Et3N (1 equiv) are dissolved in toluene to
give a final concentration of approximately 0.1 M in amine. The
resulting solution is heated at reflux until no starting amine remains
by TLC analysis (typically 16 h). The solution is transferred to a
separatory funnel using CH2Cl2, diluted with CH2Cl2, and washed with
1 N NaOH, and the aqueous layer is extracted with additional CH2Cl2.
The combined organic layers are washed with brine, dried, and
concentrated, and the products are purified by chromatography (EtOAc/
hexanes) or recrystallization.
Representative Syntheses. [3aR*-(3aσ,5σ,9aR*)]-Hexahydro-2,5-
diphenyl-2H,7H-pyrrolo[3,4-d]pyrrolizin-1-one (4p). A solution of
amine 3p (76 mg, 0.33 mmol; see Supporting Information) and
benzaldehyde (70 mg, 0.66 mmol) in toluene (10 mL) was heated at
reflux for 16 h and then concentrated to an oil which was purified by
chromatography (1:1 hexanes/EtOAc) to give tricycle 4p as a solid
(74 mg, 74%) after evaporation of the solvent: mp 131-132 °C; TLC
[2R*-(2r,3ar,6ar)]-5-Cyclohexylhexahydro-2-(4-methoxyphenyl)-
1-methyl-1H-pyrrolo[3,4-b]pyrrol-6-one (6). A solution of sarcosine
N-cyclohexyl-N-2-propenyl amide (210 mg, 1.00 mmol) and p-
anisaldehyde (272 mg, 243 µL, 2.00 mmol) in toluene (10 mL) was
heated at reflux for 3 h. Crystals formed on cooling to room
temperature; these crystals were collected by filtration, and the
remaining oil was purified by chromatography (4:1 EtOAc/hexanes)
to yield a white solid which was recrystallized from EtOAc to give
(71) General. Secondary amines were prepared, when necessary, by
the method of McCann and Overman.72 Unless otherwise noted: reaction
workups culminated in drying the solution over MgSO4 and evaporating
the solvent under reduced pressure; flash chromatography was performed
by the method of Still, Kahn, Mitra,73 using 60-mesh silica gel; NMR spectra
were recorded in CDCl3 and reported as chemical shift (multiplicity, number
of hydrogens, coupling constants in Hz) referenced to tetramethylsilane or
to the residual solvent peak (4.63 for HOD, 3.30 for CD2HOD, 7.24 for
CHCl3, 49.0 ppm for CD3OD, or 77.0 ppm for CDCl3).
(72) McCann, S.F.; Overman, L. E. J. Am. Chem. Soc. 1987, 109, 6107-
6144.
(73) Still, W. C.; Kahn, M.; Mitra, A. J. J. Org. Chem. 1978, 43, 2923-
2925.
Figure 5. Polycyclic pyrrolidino-lactam ring sytems prepared by
(74) Klein, U.; Mohrs, K.; Wild, H.; Steglich, W. Liebigs Ann. Chem.
1987, 485-489.
intramolecular azomethine ylide cycloaddition.

Synthetic Design for Combinatorial Chemistry
J. Am. Chem. Soc., Vol. 119, No. 26, 1997 6165
bicycle 6 as X-ray quality crystals (153 mg, 47% overall yield): mp
138-140 °C; IR 1677 cm-1; 1H NMR δ 7.20 (d, 2, J ) 8.7), 6.86 (d,
2, J ) 8.7), 3.98 (m, 1), 3.92 (d, 1, J ) 7.8), 3.80 (s, 3), 3.54 (dd, 1,
J ) 8.2, 10.0), 3.46 (dd, 1, J ) 6.3, 9.8), 3.08 (dd, 1, J ) 2.2, 10.0),
2.95 (m, 1), 2.43 (s, 3), 2.14 (m, 1), 1.91 (ddd, 1, J ) 2.5, 6.3, 13.0),
1.81-1.70 (m, 5), 1.47-1.32 (m, 4), 1.12 (m, 1); 13C NMR δ 171.2,
158.9, 133.6, 128.5, 113.8, 68.1, 66.6, 55.3, 50.5, 48.8, 44.2, 35.1, 32.5,
30.8, 29.9, 25.5, 25.4; MS (EI) 312.8 (100, M+); Anal. Calcd for
C20H28N2O2: C, 73.14; H, 8.59; N, 8.53. Found: C, 73.27; H, 8.50;
N, 8.55.
N-Allyl-N-phenylglyoxamide (21). To a solution of 190 mg (0.5
mmol) of diol 20 in 1 mL of diethyl ether and 0.5 mL of THF stirred
at 0 °C was added 114 mg (0.5 mmol) of periodic acid dihydrate in
two portions over 30 min. Stirring at 0 °C was continued for an
additional 45 min after which time a white solid had precipitated. The
supernatant was decanted, dried over 4Å molecular sieves, and
concentrated in vacuo to afford 164 mg (80%) of the glyoxamide 21
and its hydrate as a pale yellow oil: 1H NMR δ 4.40 (m, 2), 5.12-
5.24 (m, 2), 5.81-5.91 (m, 1), 7.15-7.45 (m, 5), 9.36 (s, 1). This
material was not purified but carried on directly into subsequent
reactions.
[3aR*-(3aσ,4r,5σ,9aR*)]-2,4,5-Triphenylhexahydro-2H,7H-pyr-
rolo[3,4-d]pyrrolizin-1-one (7). A solution of proline N-phenyl-N-
(E)-3-phenyl-2-propenyl amide (195 mg, 0.64 mmol), benzaldehyde
(135 mg, 1.30 mmol), and Et3N (89 µL, 0.64 mmol) in toluene (5 mL)
was reacted according to the general cyclization procedure. Chroma-
tography (4:1 hexanes/EtOAc) gave tricycle 7 as a white solid (160
mg, 63%). A small sample was recrystallized from EtOAc/hexanes to
provide an X-ray quality analytical sample: mp 203-204 °C; TLC Rf
[7aR*-(7ar,10ar,11aâ)]-5,6,7a,9,10,10a,11,11a-Octahydro-9-phenyl-
8H-pyrrolo[3′,4′:4,5]pyrrolo[2,1-a]isoquinolin-8-one (22). A solution
of 133 mg (1 mmol) of 1,2,3,4-tetrahydroisoquinoline, 101 mg (1 mmol)
of Et3N, and the crude preparation of glyoxamide 21 and its hydrate
(0.8 mmol) in 2 mL of toluene was stirred under nitrogen at 110 °C
for 20 h. After cooling, the solution was diluted with 30 mL of CH2-
Cl2 and washed with 15 mL of saturated NaHCO3. The aqueous wash
was extracted with two 20-mL portions of CH2Cl2, and the combined
organic extracts were dried (MgSO4) and concentrated in vacuo. The
residue was purified by chromatography (EtOAc/hexanes, 2:1) to afford
174 mg (72%) of tetracycle 22 as a white solid after evaporation of
the solvent: mp 165-166 °C; 1H NMR δ 2.24-2.38 (m, 2), 2.87 (dt,
1, J ) 15.6, 4.4), 3.02-3.08 (m, 2), 3.24-3.30 (m, 1), 3.44 (dt, 1, J )
17.2, 4.9), 3.66 (dd, 1, J ) 10.1, 2.3), 3.98 (d, 1, J ) 8.5), 4.10 (dd,
1, J ) 10.1, 8.1), 4.26 (t, 1, J ) 7.0), 7.04-7.07 (m, 1), 7.11-7.19
(m, 4), 7.36-7.40 (m, 2), 7.67-7.69 (m, 2); 13C NMR δ 26.9, 32.5,
39.8, 45.7, 53.2, 61.2, 68.2, 120.1, 124.9, 126.0, 126.1, 126.4, 128.86,
128.88, 134.4, 137.1, 139.1, 172.2. Anal. Calcd for C20H20N2O: C,
78.92; H, 6.62; N, 9.20. Found: C, 78.66; H, 6.72; N, 9.12.
1
) 0.20 (4:1 hexanes/EtOAc); IR 1700, 1494 cm-1; H NMR δ 7.72
(d, 2, J ) 8.0), 7.34 (dd, 2, J ) 7.8, 7.8), 7.29-6.99 (m, 7), 6.92 (d,
2, J ) 7.1), 6.88 (d, 2, J ) 7.2), 4.76 (d, 1, J ) 7.5), 4.13-4.08 (m,
2), 3.47 (d, 1, J ) 10.0), 3.29 (dd, 1, J ) 6.1, 11.9), 3.10 (m, 1), 2.88
(m, 1), 2.52 (m, 1), 2.24 (m, 1), 2.11-2.03 (m, 2); 13C NMR δ 177.9,
139.7, 137.5, 137.0, 129.6, 128.7, 128.1, 127.9, 127.8, 127.0, 126.5,
124.4, 119.5, 79.7, 72.3, 58.1, 47.9, 47.6, 41.9, 34.1, 28.3; MS (FAB)
m/z 395 (100, MH+); Anal. Calcd for C27H26N2O: C, 82.20; H, 6.64;
N, 7.10. Found: C, 81.83; H, 6.71; N, 6.95.
N,N′-Diallyl-N,N′-diphenyltartramide (20). To a solution of 3.00
g (15.8 mmol) of tartaric acid acetonide74 and 4.21 g (31.6 mmol) of
N-allylaniline in 160 mL of THF stirred at 0 °C was added successively
0.434 g (31.6 mmol) of 1-hydroxybenzotriazole and 6.52 g (31.6 mmol)
of dicyclohexylcarbodiimide. The mixture was stirred at room tem-
perature for 20 h, filtered and concentrated in vacuo. The crude yellow
oil was purified by column chromatography over 150 g of silica gel
(EtOAc/hexanes, 1:5) to afford 5.71 g (86%) of the diamide as a white
solid after evaporation of the solvent: mp 73-74 °C; 1H NMR δ 1.20
(s, 6), 4.19 (dd, 2, J ) 14.6, 6.2), 4.29 (dd, 2, J ) 14.6, 6.0), 4.70 (s,
2), 5.03-5.11 (m, 4), 5.75-5.85 (m, 2), 7.18-7.20 (m, 4), 7.33-7.40
(m, 6); 13C NMR δ 26.2, 52.9, 75.9, 112.1, 118.2, 128.1, 128.3, 129.3,
132.4, 141.0, 167.6. Anal. Calcd for C25H28N2O4: C, 71.41; H, 6.71;
N, 6.66. Found: C, 71.64; H, 6.81; N, 6.68.
N-(3-Butenyl)-N-phenyl[2R*-(2r,3ar,6ar)]-Hexahydro-1-methyl-
6-oxo-5-phenyl-1H-pyrrolo[3,4-b]pyrrole-2-carboxamide (Scheme
11). To a solution of 102 mg (0.25 mmol) of the di(N-homoallyl-N-
phenyl)tartramide in 0.5 mL of THF and 1 mL of diethyl ether stirred
in an ice bath was added 57 mg (0.25 mmol) of periodic acid dihydrate
in two portions over 30 min. Stirring was continued for an additional
45 min, at which time a white solid had deposited on the bottom of
the flask. The clear supernatant was decanted, dried (4Å molecular
sieves), concentrated in vacuo, diluted with 1 mL of toluene, and added
dropwise to a solution of 96 mg (0.47 mmol) of sarcosine N-allylanilide
and 66 µL (0.47 mmol) of Et3N in 1 mL of toluene. The mixture was
heated at 110 °C for 5 h, diluted with 40 mL of CH2Cl2, and washed
with 20 mL of saturated NaHCO3. The NaHCO3 wash was extracted
with two 20-mL portions of CH2Cl2. The combined organic extracts
were dried (MgSO4) and concentrated in vacuo. The residue was
purified by column chromatography over 30 g of silica gel (hexanes/
EtOAc, 1:1 then 1:2) to afford 128 mg (72%) of the 5,5-bicyclic adduct
as a pale yellow oil: 1H NMR δ 1.73-1.80 (m, 1), 2.23-2.33 (m, 3),
2.69 (s, 3), 3.34-3.39 (m, 2), 3.71-3.75 (m, 1), 3.77-3.85 (m, 2),
3.92 (m, 1, J ) 8.9), 3.98 (dd, 1, J ) 10.3, 8.7), 5.04-5.11 (m, 2),
5.73-5.83 (m, 1), 7.09-7.13 (m, 3), 7.29-7.36 (m, 3), 7.39-7.43 (m,
2), 7.53-7.55 (m, 2); 13C NMR δ 32.2, 33.7, 36.9, 37.7, 48.3, 52.7,
64.2, 69.5, 116.8, 120.2, 124.7, 128.2, 128.4, 128.7, 129.9, 135.2, 139.3,
141.8, 172.1, 176.0. Anal. Calcd for C24H27N3O2: C, 74.01; H, 6.99;
N, 10.79. Found: C, 74.21; H, 7.10; N, 10.65. The same compound
was isolated in 62% yield on combination of sarcosine N-homoallyl-
anilide and N-allylglyoxanilide.
To a solution of 5.50 g (12.6 mmol) of the tartramide acetonide in
60 mL of THF was added 60 mL of 1 M HCl. The mixture was heated
at reflux for 24 h, allowed to cool to room temperature, neutralized by
addition of saturated NaHCO3, and extracted with four 80-mL portions
of EtOAc. The combined organic extracts were dried (MgSO4) and
concentrated in vacuo. The crude brown solid was recrystallized from
EtOAc/hexanes to afford 3.37 g (71%) of tartramide 20 as a white
1
solid after evaporation of the solvent: mp 126-127 °C; H NMR δ
3.26 (br s, 2), 3.65 (s, 2), 4.10 (dd, 2, J ) 14.7, 6.0), 4.35 (dd, 2, J )
14.7, 5.6), 5.03-5.09 (m, 4), 5.73-5.80 (m, 2), 6.90 (br s, 4), 7.15-
7.27 (m, 6); 13C NMR δ 53.8, 69.7, 118.4, 128.0, 128.4, 129.7, 132.1,
139.8, 169.7. Anal. Calcd for C22H24N2O4: C, 69.46; H, 6.36; N,
7.36. Found: C, 69.47; H, 6.47; N, 7.36.
General Procedure for Formation of Glyoxamides and Conden-
sation with Amine Components. A solution of the symmetrical
tartramide precursor (0.25-0.5 mmol) in 1 mL of diethyl ether and
0.5 mL of THF is stirred at 0 °C and 1 equiv of periodic acid dihydrate
is added in two portions over 30 min. Stirring at 0 °C is continued for
an additional 45 min during which time a white solid precipitates. The
supernatant is decanted, dried over 4Å molecular sieves, and concen-
trated in vacuo to afford the glyoxamide and its hydrate as a pale yellow
oil. This material is not purified but carried on directly into subsequent
reactions.
[3aR*-(3ar,6ar)]-Hexahydro-1-methyl-5-phenyl-1H-pyrrolo[3,4-
b]pyrrol-6-one (28, Table 2, Entry 5). A mixture of 45 mg (0.5 mmol)
of sarcosine and 272 µL (1.1 mmol) of bis-trimethylsilylacetamide in
0.5 mL of acetonitrile was stirred at room temperature for 1 h and at
40 °C for 30 min and then combined with a solution of glyoxamide 2
derived from 90 mg (0.25 mmol) of the tartramide precursor 2 in 1
mL of toluene. The resulting mixture was heated at reflux for 18 h
and worked up as described for the general procedure to give 40 mg
(40%) of the bicyclic product as a pale yellow oil: 1H NMR δ 1.74-
1.83 (m, 1), 2.19-2.26 (m, 1), 2.49 (dt, 1, J ) 9.4, 6.4), 2.65 (s, 3),
2.93-3.02 (m, 1), 3.06-3.10 (m, 1), 3.19 (d, 1, J ) 9.1), 3.59 (dd, 1,
J ) 9.9, 3.9), 4.00 (dd, 1, J ) 9.9, 8.6), 7.11-7.15 (m, 1), 7.32-7.37
(m, 2), 7.61-7.64 (m, 2); 13C NMR δ 32.5, 34.5, 41.3, 53.8, 57.3,
85.0, 120.2, 124.7, 128.8, 139.3, 172.8; HRMS calcd for C13H17N2O
(MH+) m/z 217.1341, found m/z 217.1346.
A solution of the crude glyoxamide and 1-1.25 equiv of the amine
in toluene with 1-2 equiv of Et3N is heated at reflux for 18 h, cooled,
diluted with CH2Cl2, and washed with saturated NaHCO3. The aqueous
wash is back-extracted with CH2Cl2, and the combined organic extracts
are dried (MgSO4) and concentrated on a rotary evaporator. The residue
is purified by chromatography (EtOAc/hexanes) to afford the cycload-
duct. The following compounds were prepared in this fashion, with
the amount and yield of the purified product indicated.

6166 J. Am. Chem. Soc., Vol. 119, No. 26, 1997
Marx et al.
Methyl [3aR*-(3ar,7aâ,8ar)]-octahydro-3-oxo-2-phenylpyrrolo-
[3,4-b]pyrrolizine-7a-carboxylate (Table 2, Entry 6): 148 mg (61%)
as a pale yellow oil: 1H NMR δ 1.61 (dd, 1, J ) 13.1, 10.4), 1.77 (m,
1), 1.82-1.95 (m, 2), 2.21 (dt, 1, J ) 12.6, 7.0), 2.78 (dd, 1, J ) 13.1,
7.9), 3.06 (dt, 1, J ) 11.0, 6.8), 3.11-3.19 (m, 1), 3.48 (dd, 1, J )
10.1, 0.8), 3.72 (s, 3), 3.91 (dd, 1, J ) 10.1, 6.6), 4.01 (m, 1), 4.08 (d,
1, J ) 7.8), 7.12-7.16 (m, 1), 7.32-7.37 (m, 2), 7.54-7.56 (m, 2);
13C NMR δ 27.1, 36.1, 37.5, 41.5, 47.9, 49.8, 52.4, 66.3, 77.5, 120.1,
124.9, 128.8, 139.0, 171.9, 176.3; HRMS calcd for C17H21N2O3 (MH+)
m/z 301.1552, found m/z 301.1548.
(30 mL), NaI (ca. 100 mg), and allylamine (1.32 mL, 17.8 mmol) were
added, and the resulting mixture was heated at 60 °C for 4 h and cooled
to room temperature, and the resin was collected by filtration. This
amination reaction was repeated. The resin was washed with CH2Cl2
(2 × 50 mL), Et3N (50 mL), CH2Cl2 (3 × 50 mL), and Et2O (50 mL)
to give 6.02 g of allyl-functionalized resin.
Preparation of Homoallyl-Substituted Wang Resin. Wang resin
(7.11 g, 0.71 mmol/g, 5.04 mmol) was converted in the same fashion
to 8.67 g of the homoallyl functionalized resin.
Preparation of Sarcosyl-Allyl Wang Resin. Allyl-substituted resin
(2.89 g, e1.70 mmol, based on nominal resin loading), Fmoc-Sar-OH
(1.58 g, 5.10 mmol), and DMAP (ca. 50 mg) were placed in a solid-
phase synthesis flask, and CH2Cl2 (20 mL) was added. DIC (797 µL,
5.10 mmol) was added to this suspension, and the resulting mixture
was shaken for 12 h. The resin was collected, washed with DMF (50
mL), CH2Cl2 (10 × 50 mL), and Et2O (50 mL), giving 2.69 g of Fmoc-
protected sarcosyl resin with a loading of 0.32 mmol/g (by Fmoc
quantitation). This resin was shaken in 20 mL of a 20% solution of
piperidine in DMF (v:v) for 5 min. The resin was collected, and the
reaction was repeated for 30 min. The collected resin was washed
with CH2Cl2 (10 × 50 mL) and Et2O (50 mL) to afford 2.51 g of
sarcosyl-allyl resin (0.86 mmol based on the Fmoc quantitation).
Preparation of Sarcosyl-Homoallyl, Prolyl-Allyl, and Prolyl-
Homoallyl Wang Resins. In a similar fashion, allyl- and homoallyl-
substituted resins were converted to the sarcosyl-homoallyl (3.88 g,
2.05 mmol by Fmoc quantitation), prolyl-allyl (2.33 g, 0.74 mmol),
and prolyl-homoallyl (4.03 g, 1.91 mmol) resins.
Cleavage of Cycloadducts from Wang Resin. A sample of
functionalized resin is stirred in a 95:5 solution (v:v) of TFA/H2O (10
mL/g of functionalized resin) for 1 h at room temperature. The resin
is then collected by filtration and washed with 50 mL of CH2Cl2.
Concentration of the combined filtrates affords the crude cycloadduct.
Esterification of Cycloadducts from Solid Phase. EDC/MeOH.
The crude material resulting from cleavage from the resin is dissolved
in CH2Cl2 (0.1-0.2 M) at room temperature, and EDC (3 equiv), MeOH
(3 equiv), and a catalytic amount of DMAP are added. The resulting
mixture is allowed to stir for 18 h, diluted with CH2Cl2, and washed
with 1 N NaOH. The aqueous layer is further extracted with CH2Cl2,
and the combined organic layers are washed with brine, dried, and
concentrated to afford an oil which is purified by chromatography using
EtOAc as eluent.
[7aR*-(7ar,10r,10ar,11aâ)]-5,6,7a,9,10,10a,11,11a-Octahydro-
10-(1-methylethyl)-9-(phenylmethyl)-8H-pyrrolo[3′,4′:4,5]pyrrolo-
[2,1-a]isoquinolin-8-one (Table 2, Entry 13). To a solution of 165
µL (1.86 mmol) of oxalyl chloride in 4 mL of CH2Cl2 was added a
solution of 268 µL (3.72 mmol) of dimethyl sulfoxide in 4 mL of CH2-
Cl2. The mixture was stirred at -78 °C for 20 min, and a solution of
115 mg (0.47 mmol) of the glycolamide precursor in 4 mL of CH2Cl2
was added. After 30 min of stirring, 0.66 mL (4.70 mmol) of Et3N
was added, and the mixture was stirred at room temperature for 1 h.
The solution was then concentrated in vacuo to afford the desired
glyoxamide, which was diluted with 1 mL of toluene and added
dropwise to a solution of 58 µL (0.47 mmol) of 1,2,3,4-tetrahydroiso-
quinoline and 157 µL (1.13 mmol) of Et3N in 1 mL of toluene stirred
at room temperature under nitrogen. The mixture was heated at 110
°C for 18 h, cooled to room temperature, diluted with 30 mL of CH2-
Cl2, and washed with 15 mL of saturated NaHCO3. The aqueous wash
was extracted with two 20-mL portions of CH2Cl2. The combined
organic extracts were dried (MgSO4) and concentrated in vacuo. The
residue was purified by chromatography (EtOAc/hexanes, 2:1) to afford
95 mg (56%) of the cycloadduct as a colorless oil: 1H NMR δ 0.72
(d, 3, J ) 6.9), 0.82 (d, 3, J ) 6.9), 1.86-1.92 (m, 1), 2.05-2.17 (m,
2), 2.51-2.56 (m, 1), 2.81-2.88 (dt, 1, J ) 16.5, 4.6), 2.96-3.04 (m,
1), 3.09 (m, 1), 3.19-3.25 (m, 1), 3.37-3.43 (m, 1), 3.78 (d, 1, J )
8.2), 3.79 (d, 1, J ) 15.0), 4.02 (t, 1, J ) 6.9), 5.15 (d, 1, J ) 15.0),
6.96-6.98 (m, 1), 7.09-7.14 (m, 3), 7.26-7.34 (m, 5); 13C NMR δ
14.1, 14.5, 18.3, 27.1, 27.8, 33.8, 40.0, 44.0, 45.8, 60.9, 67.2, 125.8,
125.9, 126.1, 127.5, 128.1, 128.6, 128.8, 134.6, 136.3, 137.5, 173.3;
HRMS calcd for C25H28N2O (MH+) m/z 361.2280, found m/z 361.2272.
5,6,9,10-Tetrahydro-9-(phenylmethyl)-8H-pyrrolo[3′,4′:4,5]pyr-
rolo[2,1-a]isoquinolin-8-one (Table 3, entry 1): 80 mg (25%) of the
pyrrole as a beige solid after evaporation of the solvent: mp 145-146
1
°C; H NMR δ 3.10 (t, 2, J ) 6.7), 4.09 (s, 2), 4.41 (t, 2, J ) 6.7),
Diazomethane. The crude material resulting from cleavage from
the resin is taken up in THF and treated with an excess of diazomethane
at 0 °C. The solution is stirred overnight at room temperature to allow
evaporation of the excess diazomethane and then concentrated to afford
an oil which is purified by chromatography using EtOAc as eluent.
Representative Syntheses, Path A. Methyl 4-[[[3aR*-(3aσ,5σ,-
9aR*)]-Hexahydro-1-oxo-5-phenyl-2H,7H-pyrrolo[3,4-d]pyrrolizin-
2-yl]methyl]benzoate (Table 4, Entry 3). Prolyl-allyl Wang resin
(1.27 g, 0.66 mmol based on Fmoc-quantitation) was placed in a 25-
mL round-bottomed flask, and benzaldehyde (695 mg, 6.6 mmol), Et3N
(92 µL, 0.66 mmol), and toluene (17 mL) were added. The suspension
was heated at reflux for 24 h, and the resin was collected and washed
with CH2Cl2 (50 mL) and Et2O. Cleavage of the acid from the resin
followed by esterification with diazomethane according to the general
procedures provided 161 mg of the methyl ester (63% based on the
loading of the Fmoc-prolyl functionalized resin) as an oil: IR 1718,
4.71 (s, 2), 6.39 (s, 1), 7.17-7.36 (m, 8), 7.53 (m, 1); 13C NMR δ
28.9, 41.3, 46.0, 46.7, 97.7, 123.3, 127.2, 127.3, 127.4, 127.9, 128.3,
128.7, 128.9, 131.3, 133.1, 137.3, 138.0, 162.3. Anal. Calcd for
C21H18N2O: C, 80.23; H, 5.77; N, 8.91. Found C, 80.32; H, 5.90; N,
8.92.
Ethyl [2R*-(2r,6ar)]-2,4,5,6a-tetrahydro-1-methyl-6-oxo-5-(phen-
ylmethyl)-1H-pyrrolo[3,4-b]pyrrole-2-carboxylate (Table 3, entry
2): 74 mg (35%) of the pyrroline as a pale yellow oil: 1H NMR δ
1.27 (t, 3, J ) 7.1), 2.77 (s, 3), 3.69 (d, 1, J ) 12.7), 3.85 (dq, 1, J )
12.7, 1.8), 4.12-4.20 (m, 2), 4.43 (m, 3), 4.50 (m, 1), 5.75 (m, 1),
7.21-7.33 (m, 5); 13C NMR δ 14.4, 36.2, 45.5, 46.6, 60.8, 71.9, 75.8,
122.4, 127.7, 128.0, 128.8, 135.8, 141.7, 171.0, 171.5; HRMS calcd
for C17H21N2O3 (MH+) m/z 301.1552, found m/z 301.1554.
Methyl 2,3,4,7,8,9-hexahydro-9-oxo-8-(phenylmethyl)-1H-pyr-
rolo[3,4-b]azocine-5-carboxylate (30, Table 3, entry 3): 67 mg (20%)
of the ring-expanded product 30 as a bright yellow solid after
1
1680, 1280 cm-1; H NMR (CD3OD) δ 7.97 (d, 2, J ) 8.1), 7.40-
1
evaporation of the solvent: mp 155-156 °C; H NMR δ 1.39-1.46
7.22 (m, 7), 4.60 (d, 1, A of AB, J ) 15.0), 4.49 (d, 1, B of AB, J )
15.0), 4.17 (dd, 1, J ) 4.7, 12.1), 3.85 (s, 3), 3.54 (dd, 1, J ) 10.1,
10.1), 3.02 (dd, 1, J ) 4.2, 16.5), 2.52 (m, 1), 2.45-2.30 (m, 4), 1.93-
1.77 (m, 4); 13C NMR (CD3OD) δ 176.5, 166.6, 141.6, 137.7, 129.5,
129.3, 128.2, 127.8, 127.6, 127.1, 78.9, 64.4, 51.0, 48.7, 46.1, 41.1,
40.7, 35.0, 33.9, 25.7; HRMS (FAB) calcd for C24H27N2O3 (MH+)
391.2021, found: 391.2020.
Methyl 4-[[[2R*-(2r,3ar,6ar)]-2-(2-Furanyl)hexahydro-1-methyl-
6-oxo-1H-pyrrolo[3,4-b]pyrrol-5-yl]methyl]benzoate (Table 4, Entry
5). Sarcosyl-allyl Wang resin (800 mg, 0.28 mmol based on Fmoc-
quantitation) was suspended in toluene (10 mL) in a round-bottomed
flask. Furaldehyde (46 µL, 0.56 mmol) and Et3N (39 µL, 0.28 mmol)
were added to this solution, and the resulting mixture was heated at
reflux for 18 h. The mixture was cooled to room temperature, and the
resin was collected and washed with CH2Cl2 (10 × 5 mL). The
(m, 2), 2.86-2.89 (m, 2), 3.52-3.55 (m, 2), 3.64 (s, 2), 3.72 (s, 3),
4.63 (s, 2), 4.70 (br s, 1), 7.09 (s, 1), 7.23-7.34 (m, 5); 13C NMR δ
17.3, 25.2, 41.8, 46.7, 50.2, 51.7, 106.0, 126.3, 127.7, 128.1, 128.8,
134.8, 136.6, 140.1, 166.8, 167.8; HRMS calcd for C18H20N2O3 m/z
312.1474, found m/z 312.1476.
Syntheses on Solid Support. Preparation of p-(Allylamino-
methyl)benzoyl Wang Resin (Allyl-Substituted Wang Resin). Wang
resin (5.0 g, 0.71 mmol/g, 3.55 mmol), DMAP (ca. 100 mg), and
4-(chloromethyl)benzoic acid (1.2 g, 7.1 mmol) were suspended in CH2-
Cl2 (40 mL) in a solid-phase synthesis vessel. To this suspension was
added DIC (2.2 mL, 14 mmol), and the resulting mixture was shaken
for 24 h at room temperature. The resin was collected by filtration,
washed with DMF (50 mL) and CH2Cl2 (10 × 50 mL), and transferred
to a round-bottomed flask equipped with a magnetic stir bar. DMF

Synthetic Design for Combinatorial Chemistry
J. Am. Chem. Soc., Vol. 119, No. 26, 1997 6167
products were cleaved from the resin and esterified (EDC/MeOH)
according to the general procedures to give 74 mg of the methyl ester
(75% based on the loading of Fmoc-sarcosyl resin) as an oil: TLC Rf
mL (2.74 mmol) of Et3N was added. The mixture was stirred at room
temperature for 2 h. The resin was collected on a Buchner funnel,
washed with five 10-mL portions of CH2Cl2, and placed in a round-
bottomed flask containing 5 mL of toluene. After addition of 76 µL
(0.55 mmol) of Et3N and 68 µL (0.55 mmol) of 1,2,3,4-tetrahydroiso-
quinoline, the mixture was heated at 110 °C for 18 h. After cooling to
room temperature, the resin was collected on a Buchner funnel and
washed with two 10-mL portions of toluene and five 10-mL portions
of CH2Cl2. The resin was then placed in 10 mL of 95% TFA/H2O.
The red mixture was stirred at room temperature for 1 h and filtered.
The filtrate was neutralized by addition of saturated NaHCO3 and
extracted with two 50-mL portions of CH2Cl2 and one 50-mL portion
of EtOAc. The combined organic extracts were dried (MgSO4) and
concentrated in vacuo. 1H NMR analysis of the crude product indicated
the presence of one major product. Purification by chromatography
(CH2Cl2/methanol, 20:1 then 15:1) afforded 45 mg (45% based on
nominal loading) of tetracycle 45 as a yellow oil: 1H NMR δ 2.02-
2.11 (m, 1), 2.16-2.25 (m, 1), 2.84-2.89 (m, 2), 2.95-2.99 (m, 1),
3.04 (dd, 1, J ) 10.3, 2.3), 3.18-3.24 (m, 1), 3.33-3.38 (m, 1), 3.46
(dd, 1, J ) 10.2, 8.2), 3.89 (d, 1, J ) 8.6), 4.13 (t, 1, J ) 6.9), 4.35 (d,
1, J ) 14.8), 4.61 (d, 1, J ) 14.8), 6.05 (br s, 1), 6.60 (br s, 1), 6.99-
7.02 (m, 1), 7.08-7.14 (m, 3), 7.29-7.35 (m, 2), 7.77-7.82 (m, 2);
13C NMR δ 27.0, 32.9, 39.7, 45.9, 46.5, 51.7, 61.3, 67.3, 126.1, 126.5,
128.0, 128.3, 128.9, 132.9, 134.2, 140.3, 169.0, 172.5; HRMS calcd
for C22H23N3O2 (MH+), m/z 362.1869, found m/z 362.1872.
General Procedure for the Synthesis of Compounds of Table 5.
A mixture of oxalyl chloride (4 equiv) and dimethyl sulfoxide (8 equiv)
in CH2Cl2 (at 0.25 M) was stirred at -78 °C for 30 min, and the glycol-
amide Rink resin in CH2Cl2 (0.11 g/mL) was added. The suspension
was stirred at -78 °C for 1 h, Et3N (10 equiv) was added, and the
mixture was stirred at room temperature for 2 h. The resin was
collected on a fritted funnel, washed with five 20-mL portions of CH2-
Cl2, and dried briefly. To the resin in toluene (0.11 g/mL) was added
the amine (1.5 equiv) and Et3N (to neutralize amine-hydrochloride if
necessary, plus 1.5 equiv), and the mixture was heated at reflux for
18-20 h. The resin was collected on a fritted funnel, washed with
five 20-mL portions of CH2Cl2, dried, then suspended in 95%
trifluoroacetic acid in water, and stirred at room temperature for 1 h.
The mixture was filtered, and the filtrate was neutralized with saturated
NaHCO3 and extracted with two 80-mL portions of CH2Cl2 and one
80-mL portion of EtOAc. The combined organic extracts were dried
(MgSO4) and concentrated in vacuo, and the residue was purified by
chromatography on silica (eluted with CH2Cl2-methanol, 20-30:1)
to afford the indicated amount of product. The reported yields are
based on the nominal loading of the commercial Rink amide resin.
1
) 0.30 (EtOAc), IR 1722, 1677, 1277 cm-1; H NMR δ 8.00 (d, 2, J
) 8.3), 7.35-7.32 (m, 1), 7.31 (d, 2, J ) 8.3), 6.30 (dd, 1, J ) 1.9,
3.2), 6.18 (dd, 1, J ) 0.6, 3.1), 4.52 (d, 1, J ) 14.9), 4.45 (d, 1, J )
14.9), 3.93-3.89 (m, 1), 3.90 (s, 3), 3.70 (d, 1, J ) 8.5), 3.45 (dd, 1,
J ) 8.3, 10.0), 3.13 (m, 1), 2.99 (dd, 1, J ) 2.7, 10.0), 2.50 (s, 3), 2.36
(ddd, 1, J ) 5.8, 9.3, 12.8), 1.89 (ddd, 1, J ) 5.5, 5.5, 12.8); 13C NMR
δ 172.8, 166.6, 153.9, 141.9, 141.4, 129.9, 129.5, 127.9, 109.8, 107.8,
67.2, 61.3, 52.0, 52.0, 46.2, 38.7, 36.0, 33.3; HRMS (FAB) calcd for
C20H23N2O4 (MH+) 355.1657, found 355.1654.
Representative Syntheses: Path B. p-(Chloromethyl)benzoyl
Rink Resin (39). A sample of 4.00 g (2.20 mmol) of Rink amide
resin (0.55 mmol/g nominal loading) was swelled in 86 mL of DMF.
The solvent was drained and 86 mL of a 20% solution of piperidine in
DMF was added to the resin. The mixture was shaken for 5 min, the
solvent was drained, and another 86 mL of a 20% solution of piperidine
in DMF was added. The mixture was shaken for 20 min at room
temperature and drained. The resin was washed with seven 80-mL
portions of DMF and dried in vacuo. To the resin in 68 mL of CH2-
Cl2 were added successively 788 mg (4.40 mmol) of 4-chloromethyl-
benzoic acid, 2.054 g (4.40 mmol) of PyBroP, and 765 µL (4.40 mmol)
of diisopropylethylamine. The mixture was shaken at room temperature
for 2 h, and the solution was drained. The procedure was repeated
once. The resin was washed with six 40-mL portions of CH2Cl2 and
dried in vacuo. A ninhydrin test performed on a sample of the dried
beads was negative (no blue color observed), indicating that acylation
had proceeded to completion.
p-(N-Glycolyl)-N-(2-Propenyl)aminomethyl)benzoyl Rink Resin
(42). To a suspension of 1.79 g (0.98 mmol) of acylated resin 39 in
10 mL of DMF in a round-bottomed flask were added 0.74 mL (9.8
mmol) of allylamine and a catalytic amount of sodium iodide. The
mixture was heated at 75 °C for 3 h, and the resin was collected by
filtration. The procedure was repeated a second time, and heating was
prolonged overnight. The resin was collected on a fritted glass, washed
with five 20-mL portions of CH2Cl2, and dried. A chloranil test
performed on a sample of the dried beads was positive (blue color
observed), indicating the presence of a secondary amine.
To a suspension of 1.63 g (0.89 mmol) of aminated resin 40 in 15
mL of CH2Cl2 were added successively 210 mg (1.78 mmol) of
acetoxyacetic acid, 0.42 mL (2.67 mmol) of DIC, and a catalytic amount
of DMAP. The mixture was shaken at room temperature for 3 h, and
the resin was collected by filtration. The acylation reaction was
repeated once with shaking overnight. The resin was collected on a
fritted funnel, washed with five 20-mL portions of CH2Cl2, and dried.
A chloranil test performed on a sample of the dried beads was negative
(no blue color observed), indicating that acylation had proceeded to
completion.
Acknowledgment. This work was supported by a grant from
the National Institutes of Health. We also thank Miles, Inc.
for a graduate fellowship to M.A.M., Prof. Albert Padwa for
helpful comments, Samuel D. Gillett for his contributions to
the project, and Dr. Frederick Hollander of the Chexray Facility,
College of Chemistry, University of California, Berkeley, for
determination of the crystal structures of 6 and 7.
A suspension of 1.53 g (0.84 mmol) of acylated resin 41 and 1.16
g (8.4 mmol) of K2CO3 in 15 mL of methanol and 30 mL of DMF was
shaken at room temperature for 18 h. The resin was collected, washed
successively with water, methanol, and CH2Cl2, and dried in vacuo.
p-(N-Glycolyl)-N-(3-butenyl)- and -(cinnamyl)aminoethyl)benzoyl
Rink Resins. In a similar fashion, resin 39 was substituted with
3-butenylamine or cinnamylamine, acetoxyacetylated, and hydrolyzed
to give the homoallyl- and cinnamyl-glycolamide resins.
4-[[[7arR*,10ar,11aâ]-[5,6,7a,9,10,10a,11,11a-Octahydro-8-oxo-
8H-pyrrolo[3′,4′:4,5]pyrrolo[2,1-a]isoquinolin-9-yl]methyl]benza-
mide (45). A solution of 96 µL (1.10 mmol) of oxalyl chloride and
157 µL of dimethyl sulfoxide in 5 mL of CH2Cl2 was stirred at -78
°C for 30 min, and 498 mg (0.33 mmol) of allyl-glycolamide Rink
resin 42 in 5 mL of CH2Cl2 was added. The mixture was stirred at
-78 °C for 1 h and allowed to warm to room temperature, and 0.38
Supporting Information Available: Experimental Proce-
dures and characterization of synthetic intermediates and final
products not described above; 1H NMR spectra of crude products
off resin (Table 4); and products for which melting points, but
not recrystallization solvents, are reported (40 pages). See any
current masthead page for ordering and Internet access
instructions.
JA9621051