An ugi reaction in the total synthesis of (-)-dysibetaine

Article abstract of DOI:10.1002/anie.200805709

(-)-Dysibetaine has been synthesized in 11 steps from readily available L-malic acid (see scheme). The key step is a unique Ugi 4-center-3-component cyclization reaction, where an ester group acts as the carboxylic acid component. The use of 1,1,1,3,3,3-h

Full text of DOI:10.1002/anie.200805709

Angewandte
Chemie
DOI: 10.1002/anie.200805709
Natural Product Synthesis
An Ugi Reaction in the Total Synthesis of (À)-Dysibetaine**
Jerry Isaacson and Yoshihisa Kobayashi*
The Ugi reaction has been extensively used by chemists
because of its unique ability to generate complex products
from diverse starting materials in a single step. The four-
component condensation brings together an amine, a carbox-
ylic acid, a ketone or aldehyde, and an isocyanide function-
ality to yield an a-acylaminoamide.[1] The wide variability of
the components makes the Ugi reaction an attractive tool for
both diversity-oriented and target-oriented synthesis.[2] When
two of the components, for example a ketone and a carboxylic
acid functionality, are incorporated into a single molecule,
cyclic products are formed.[3,1a] We previously showed that
incorporation of convertible isocyanide 1 into this type of Ugi
Snider and Gu,[8] who completed the first total synthesis and
established the absolute stereochemistry of the molecule.
Scheme 1 outlines the retrosynthetic plan that we initially
envisioned for the synthesis of 2. The synthesis can be broken
down into three distinct phases, the first of which is the
Scheme 1. Retrosynthetic analysis of (À)-dysibetaine (2).
4-center-3-component cyclization reaction (U4C-3CR) is an
efficient method for the synthesis of pyroglutamic acid
analogues[4] and therefore sought to apply this reaction to
the stereocontrolled total synthesis of the marine natural
product (À)-dysibetaine (2). In the course of synthesizing 2
we discovered a unique and unusual Ugi reaction in which an
ester functions as the carboxylic acid component in the
reaction.
(À)-Dysibetaine (2) is a member of a group of natural
products that were isolated by Sakai et al. in the late 1990s
from the Micronesian sponge Dysidea herbacea.[5] We were
initially drawn to 2 by its potential neuroexcitotoxin activity
and because it is an attractive target for taking full advantage
of the U4C-3CR. We recently reported a racemic synthesis of
the molecule,[6] and it has been synthesized or studied
synthetically by various other research groups,[7] notably by
enantioselective synthesis of the necessary g-ketoacid A, with
the masked amine already installed. l-Malic acid is the
natural choice for the starting material because it contains the
necessary stereocenter and the carbon framework is suffi-
ciently oxidized. The second phase is the stereocontrolled
formation of B through a multiple functional group con-
densation/cyclization reaction, specifically the U4C-3CR. The
final challenges in our synthesis will be selective cleavage of
the amide bond and the unveiling of the betaine moiety.
In complex syntheses, a key factor underlying the utility of
the Ugi reaction is the ability to make further, successful
modifications to the products derived from the reaction. To
reach 2 it is necessary to selectively cleave one of the two
amide groups in intermediate B, both formed during the U4C-
3CR (Scheme 1). Compound 1 was therefore chosen as the
isocyanide because after the Ugi reaction the amide group
resulting from 1 is readily converted into N-acylindole, which
is hydrolyzed with mild reaction conditions.[9]
The first problem to overcome in this synthesis was the
elaboration of l-malic acid to the necessary g-ketoacid A. In
the past, we have had success on similar molecules using
(trimethylsily)diazomethane to elaborate a succinic acid
derivative to the corresponding ketone moiety.[4a] However,
attempts to apply the same chemistry to various protected l-
malic acid monoester derivatives failed, possibly as a result of
the intramolecular participation of the alcohol (protected or
unprotected) with the intermediate diazoketone[10]—although
no such products were recovered. Fortunately, we found that
this problem had been solved by using the strongly electron-
[*] J. Isaacson, Prof. Dr. Y. Kobayashi
Department of Chemistry and Biochemistry
University of California, San Diego
9500 Gilman Drive Mail Code 0343, La Jolla, CA 92093-0343 (USA)
Fax: (+1)858-822-5870
E-mail: ykoba@chem.ucsd.edu
Homepage:
http://www-chem.ucsd.edu/research/profile.cfm?cid=C04667
[**] We thank the University of California for financial support as well as
the UCSD Mass Spectrometry Facility and UCSD Small Molecule X-
Ray Facility.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.200805709.
Angew. Chem. Int. Ed. 2009, 48, 1845 –1848
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1845

Communications
Scheme 2. Synthesis of (À)-dysibetaine (2). a) NaN3 (2 equiv), NaI (1 equiv), acetonitrile, RT, 1.5 h (99%); b) 1 (1 equiv), 4 (1.5 equiv) HMDS
(6 equiv), HFIP, RT, 18 h (85%); c) Ac2O (12 equiv), pyridine (10 equiv), CH2Cl2, RT, 24 h (92%); d) CSA (0.2 equiv), benzene, 608C, 1.5 h (7a:
29% and diastereomer 7b: 50%); e) 1 m aq NaOH (0.03 equiv), MeOH, RT, 36 h (62%); f) 37% aq CH2O (10 equiv), H2, Pd/C, 1 m aq HCl
(1 equiv), MeOH, RT, 18 h; then NaHCO3(s) (99%); g) MeI (40 equiv), THF, RT, 24 h (84%); h) Dowex 550A, MeOH (100%). CSA=(Æ)-
camphorsulfonic acid, THF=tetrahydrofuran.
withdrawing chloral-derived protecting group.[11] The result-
ing a-chloroketone 3 was transformed into 4 in good yield
using the reaction conditions shown in Scheme 2.
Indole formation provided the activation needed for the
selective cleavage of the amide bond, and treatment of 7a
with a catalytic amount of NaOH (aq) in methanol afforded
methyl ester 8 with concomitant removal of the acetate ester.
As a testament to the ease of amide cleavage, TLC analysis
indicated that cleavage of the amide bond was very fast
(taking only a few minutes), meanwhile the acetate ester
group was removed much more slowly.
With 4 in hand the task before us now was to successfully
elaborate that molecule into a suitable g-ketoacid, such as A
(Scheme 1). However, various problems presented them-
selves which made it difficult to incorporate 4 into a successful
synthesis of 2 when beginning with the removal of the chloral
group. Although the chloral group is a common protecting
group for diols,[12] there is little precedent for using it to
protect an b-hydroxy acid such as in 4. The common method
for removal of a chloral protecting group involves reduction
We were now ready to take 8 through the final steps
needed to reach 2. Previous syntheses suggested a stepwise
reduction of the azide in 8 and subsequent Clarke–Eschweil-
ler methylation of the resulting amine group under high
pressure.[7b,8] We found that these steps were easily completed
in one pot under H2 at atmospheric pressure to give 9 in good
yield. The final steps followed existing procedures.[7b,8]
Quaternization of the amine group using methyl iodide gave
10, and subsequent hydrolysis with basic resin gave the
natural product (À)-dysibetaine (2), the spectral properties of
which matched with the natural product and our own previous
work.[5a,6] The synthesis was completed in 12 steps in the
longest linear sequence and in just 11 steps from l-malic acid.
The key to the success of this synthesis was the unique Ugi
reaction in which an ester functions as the carboxylic acid
component. This is particularly surprising because the acid
component is believed to act as a Brønsted acid by activating
the Schiff base to react with the isocyanide.[1a] This result is
not completely without precedence, however, as amides have
been known to participate in similar intramolecular multi-
component reactions (MCRs).[15] We initially chose ammoni-
um acetate as the amine source and 1 as the isocyanide, with
2,2,2-trifluoroethanol (TFE) as the solvent. The desired Ugi
product did form, albeit in very low yield. The major product
recovered from the reaction was the Passerini adduct of
chloral 11, acetic acid, and 1 (Scheme 3). To eliminate the
possibility of 11 forming we chose to use 1,1,1,3,3,3-hexam-
ethyldisilazane (HMDS), which has been used previously as
an ammonia equivalent in MCRs.[4a,16] Compound 4 was then
converted into the desired Ugi product 5 as a mixture of
diastereomers, although with unsatisfactory yield.
with zinc to remove chlorine before deprotection occurs[12]
—
reaction conditions that are incompatible with both 3 and 4.
We could convert 4 into the corresponding free-alcohol/
methyl ester (not shown) by treating it with a variety of bases
in methanol, but yields varied widely and were not reprodu-
cible.
A key to the success of our overall strategy is the ability to
reach the Ugi reaction in a reasonable number of steps.
Numerous other problems were encountered while trying to
elaborate 4 into a suitable g-ketoacid, they include an
excessive number of steps and low yields. Finally, we decided
to submit 4 directly to the conditions of the Ugi 4-center-3-
component cyclization reaction. Pleasingly, 5 did form,
revealing that the Ugi reaction can take place directly from
the ester 4. Reaction conditions were eventually developed to
make this a synthetically viable step. This unique Ugi reaction
is discussed in more detail below.
With a straightforward pathway to the Ugi product
available, we could now focus on the elaboration of 5 into
the natural product. The free alcohol in 5 was acetylated in
high yield to afford 6 (Scheme 2). Upon treatment with
catalytic acid, 6 cleanly converted into the corresponding N-
acylindole. At this point the two diastereomers could be
separated to give 7a and 7b (not shown).[13] X-Ray analysis of
the major diastereomer (7b) revealed that the minor diaste-
reomer (7a) is the one required to reach 2.[14] It was possible
to form the indole directly from 7; however acetylation of the
free alcohol was necessary to separate the diastereomers.
A screen of typical organic solvents as well as a range of
alcohols, acetic acid, and water revealed that the only other
1846
www.angewandte.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2009, 48, 1845 –1848

Angewandte
Chemie
Scheme 3. Reagents and conditions: a) 1 (1 equiv), NH4OAc (2 equiv),
TFE, RT, 16 h (11: 80% and 5: 18%). Ar=2-(2,2-dimethoxyethyl)-
phenyl.
solvent conducive to this reaction was 1,1,1,3,3,3-hexafluor-
oisopropanol (HFIP), which was found to be the best choice.
While attempting to improve the yield through modified
reaction conditions, we noticed that additional equivalents of
HMDS lowered the yield when TFE was the solvent, but
improved the yield when HFIP was used. The results of these
experiments are summarized in Table 1.
Scheme 4. Proposed mechanism of the U4C-3CR between 4 and 1.
conditions screened yielded 6 with better than 3:2 diastereo-
selectivity. While a diastereoselective Ugi reaction continues
to be an underlying goal for us, a general solution to this
problem remains elusive.
Table 1: Effect of changing the amount of the amine component
Our research group has previously reported a racemic, 16-
step synthesis of (À)-dysibetaine by using the U4C-3CR.[6]
The current stereoselective synthesis allowed us to access the
natural product in just 11 steps and in 11% overall yield from
l-malic acid. Employing 1 as the isocyanide and HMDS as the
amine in the cyclization reaction allowed the Ugi product to
be elaborated to (À)-dysibetaine using mild reaction con-
ditions. A key to the success of this synthesis was the
discovery that the activated ester 4, containing the necessary
carbon and heteroatom framework, could successfully partic-
ipate in the Ugi 4-center-3-component cyclization reaction.
(HMDS) with varying the solvent between TFE and HFIP.
Solvent
HMDS [equiv]
Yield of 5 [%]
1
2
3
4
1
2
3
4
33
29
21
17
25
40
52
59
Received: November 21, 2008
Published online: January 27, 2009
Keywords: betaines · isocyanides · natural products ·
.
total synthesis · Ugi reaction
This difference between solvents gives us insight into the
mechanism. In the case of TFE, 4 is decomposed by the
solvent and this effect is increased when more HMDS is
added. HFIP, however, is a non-nucleophilic solvent.[17] As the
solvent cannot react with the chloral-protected ester in 4, we
can rule out any mechanism involving such interactions.
We therefore propose a mechanism in which the newly
formed keto imine reacts with isocyanide 1 to give an
activated intermediate that can then react directly with the
activated ester group. Mumm rearrangement[1a] and subse-
quent solvolysis of the chloral protecting group then leads
directly to 6 (Scheme 4). The chloral protecting group was
originally chosen to inhibit intramolecular participation of the
thusly protected alcohol with a diazoketone intermediate. It is
therefore remarkable that the same protecting group allows
the intramolecular participation of the ester in the Ugi
reaction.
[1] a) A. Dꢀmling, I. Ugi, Angew. Chem. 2000, 112, 3300 – 3344;
Angew. Chem. Int. Ed. 2000, 39, 3168 – 3210; b) A. Dꢀmling,
Chem. Rev. 2006, 106, 17 – 89.
[2] a) D. L. Wright, C. V. Robotham, K. Aboud, Tetrahedron Lett.
2002, 43, 943 – 946; b) A. Dꢀmling, B. Beck, U. Eichelberger, S.
Sakamuri, S. Menon, Q.-Z. Chen, Y. Lu, L. A. Wessjohann,
Angew. Chem. 2006, 118, 7393 – 7397; Angew. Chem. Int. Ed.
2006, 45, 7235 – 7239.
[3] a) K. M. Short, A. M. M. Mjalli, Tetrahedron Lett. 1997, 38, 359 –
362; b) G. C. B. Harriman, Tetrahedron Lett. 1997, 38, 5591 –
5594; c) C. Hanusch-Kompa, I. Ugi, Tetrahedron Lett. 1998, 39,
2725 – 2728.
[4] a) J. Isaacson, C. B. Gilley, Y. Kobayashi, J. Org. Chem. 2007, 72,
3913 – 3916; b) M. Vamos, K. Ozboya, Y. Kobayashi, Synlett
2007, 1595 – 1599; c) C. B. Gilley, M. J. Buller, Y. Kobayashi,
Org. Lett. 2007, 9, 3631 – 3634; d) O. Kreye, B. Westermann,
L. A. Wessjohann, Synlett 2007, 3188 – 3192.
[5] a) R. Sakai, C. Oiwa, K. Takaishi, H. Kamiya, M. Tagawa,
Tetrahedron Lett. 1999, 40, 6941 – 6944; b) R. Sakai, K. Suzuki,
K. Shimamoto, H. Kamiya, J. Org. Chem. 2004, 69, 1180 – 1185.
[6] J. Isaacson, M. Loo, Y. Kobayashi, Org. Lett. 2008, 10, 1461 –
1463.
After screening a variety of reaction conditions, it was
determined that the best yield of 6 was obtained by using the
following conditions: 1 (1 equiv), 4 (1.5 equiv), and HMDS
(6 equiv) in HFIP. Unfortunately, none of the reaction
Angew. Chem. Int. Ed. 2009, 48, 1845 –1848
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1847

Communications
[7] a) D. J. Wardrop, M. S. Burge, Chem. Commun. 2004, 1230 –
[13] See the Supporting Information for full data and experimental
details.
[14] CCDC 714174 (7b) contains the supplementary crystallographic
data for this paper. These data can be obtained free of charge
from The Cambridge Crystallographic Data Centre via www.
ccdc.cam.ac.uk/data_request/cif.
[15] a) D. Behnke, R. Taube, K. Illgen, S. Nerdinger, E. Herdtweck,
Synlett 2004, 688 – 692; b) A. Shaabani, A. Maleki, H. Mofak-
ham, H. R. Khavasi, J. Comb. Chem. 2008, 10, 883 – 885.
[16] Q-Y. Song, B-L. Yang, S-K. Tian, J. Org. Chem. 2007, 72, 5407 –
5410.
1231; b) N. Langlois, B. K. Le Nguyen, J. Org. Chem. 2004, 69,
7558 – 7564; c) M. Katoh, C. Hisa, T. Honda, Tetrahedron Lett.
2007, 48, 4691 – 4694.
[8] B. B. Snider, Y. Gu, Org. Lett. 2001, 3, 1761 – 1763.
[9] a) E. Arai, H. Tokuyama, M. S. Linsell, T. Fukuyama, Tetrahe-
dron Lett. 1998, 39, 71 – 74; b) T. W. Greene, P. G. M. Wuts in
Protective Groups in Organic Synthesis, 4th ed., Wiley, New
York, 1999, p. 640.
[10] T. Ye, M. A. McKervey, Chem. Rev. 1994, 94, 1091 – 1160.
[11] Y. Takahashi, J. Kohno, T. Tsuchiya, Carbohydr. Res. 1998, 306,
349 – 360.
[17] I. A. Shuklov, N. V. Dubrovina, A. Bꢀrner, Synthesis 2007, 2925 –
2943.
[12] T. W. Greene, P. G. M. Wuts in Protective Groups in Organic
Synthesis, 4th ed., Wiley, New York, 1999, p. 305.
1848
www.angewandte.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2009, 48, 1845 –1848