Three-step synthesis of sialic acids and derivatives

Article abstract of DOI:10.1002/anie.200601555

(Chemical Equation Presented) Flexible yet efficient: Sialic acids such as L-N-acetylneuraminic acid (see picture) can be synthesized in only three steps by 1) vinylation of an aldose through a modified Petasis coupling reaction, 2) 1,3-dipolar cycloaddition with a nitrone to construct an isoxazolidine ring, and 3) base-catalyzed β elimination/ring opening of the isoxazolidine to generate a γ-hydroxy-α-keto acid.

Full text of DOI:10.1002/anie.200601555

Angewandte
Chemie
recently to the utilization of sialic acids and analogues as
probes or inhibitors in biological and medical research.[3]
Some of the sialic acid analogues have even achieved
commercial success, such as Relenza (4) and Tamiflu (5) for
the treatment of influenza.[4] Further studies on the roles of
sialic acids in biology as well as medicine are highly
warranted. However, a challenge remaining in this area is
the development of an efficient method for the synthesis of
sialic acids and analogues with a high degree of flexibility in
structural and stereochemical alteration.[5] In this context
several de novo syntheses of sialic acids have been
reported.[6–8] Most of these, however, are not cost effective
because the synthetic routes are lengthy and the overall yields
are low. Herein we describe our own advance in this direction,
in which a novel and interesting route was successfully
developed to synthesize sialic acids and derivatives from
inexpensive starting materials with more satisfactory profi-
ciency.
Sialic Acid Synthesis
DOI: 10.1002/anie.200601555
Three-Step Synthesis of Sialic Acids and
Derivatives**
We began our study with the synthesis of l-N-acetylneur-
aminic acid (6, l-Neu5Ac), which is the enantiomer of natural
d-Neu5Ac (1). Apotential application of the synthesis was to
provide a sufficient quantity of l-Neu5Ac (6) for screening l
peptides (by phage display) or d nucleic acids (by aptamer
selection) that can bind l sugars.[9] To maximize the synthetic
efficiency, we designed the retrosynthetic route shown in
Scheme 1. The key steps in the synthesis include: 1) the
simultaneous introduction of an amino group and a vinyl
group to an aldose by means of the newly developed Petasis
coupling reaction[10] and 2) the concise conversion of the vinyl
group to a g-hydroxy-a-keto acid moiety by means of a
diastereoselective 1,3-dipolar cycloaddition reaction and
subsequent ring opening. The scope of the proposed method
was expected to be highly flexible, because by utilizing
various types of aldoses (or polyhydroxy aldehydes) we would
be able to obtain an unlimited number of sialic acid
derivatives. At the same time, the efficiency of the method
would be maximized if any protection and deprotection of the
hydroxy groups could be strategically avoided.
Zhangyong Hong, Lei Liu, Che-Chang Hsu, and Chi-
Huey Wong*
Sialic acids are a family of monosaccharides of which more
than 50 natural derivatives have been identified.[1] The most
well-known member of the family is N-acetylneuraminic acid
(Neu5Ac, 1). Other important members include the 5-
glycolylamido derivative Neu5Gc (2) and the non-aminated
derivative KDN (3). These compounds represent one of the
most important constituents of glycoconjugates in biological
systems. They are involved in cell–cell recognition, blood
coagulation, fertilization, and other biological events.[2] The
peripheral positioning of sialic acids also instigates their
participation in the pathogenesis of a variety of diseases
including inflammatory disease, cancer metastasis, and virus
infection. As a result, increasing attention has been paid
[*] Dr. Z. Hong, Dr. L. Liu, Dr. C.-C. Hsu, Prof. C.-H. Wong
Department of Chemistry and
The standard Petasis reaction is a three-component
condensation of an a-hydroxy aldehyde, a primary or
secondary amine, and a substituted vinyl boronic acid (or
aryl boronic acid). However, our initial implementation of
this reaction with an aldose was not successful until we
realized that the unsubstituted vinyl boronic acid is an
unstable compound that deteriorates rapidly.[11] In searching
for a solution to this problem, we discovered that the
commercially available dibutyl vinyl boronic acid ester
could also be used in the Petasis coupling. An interesting
The Skaggs Institute for Chemical Biology
The Scripps Research Institute
10550 North Torrey Pines Road, La Jolla, CA 92037 (USA)
Fax: (+1)858-784-2409
E-mail: wong@scripps.edu
[**] This work was supported by the National Institutes of Health. We
thank Dr. Lisa Whalen for proofreading the manuscript.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
Angew. Chem. Int. Ed. 2006, 45, 7417 –7421
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
7417

Communications
Scheme 1. Retrosynthesis of l-Neu5Ac (6).
aspect of this new finding was that water is necessary in the
coupling of the boronic acid ester. This indicates that an in
situ equilibrium between the boronic acid ester and boronic
acid might occur under the reaction conditions (see
Scheme 2). Owing to this exchange process, the coupling of
dibutyl vinyl boronic acid ester had to be performed at a
slightly higher reaction temperature (508C) and for a longer
reaction time (72 h) in EtOH/H2O (4:1) as compared to the
standard conditions for the Petasis reaction (258C, 24 h, in
pure EtOH).
Having solved the problem of incorporation of an
unsubstituted vinyl group, we next successfully accomplished
a one-pot synthesis of the key intermediate 9 (Scheme 2). In
designing such a one-pot reaction, we confirmed through the
NMR analysis of the crude reaction mixture that intermediate
11 was obtained as a nearly single diastereomer (> 99% de).
Rather than isolating compound 11, we treated the reaction
mixture with a catalytic amount of trifluoroacetic acid (TFA)
and stirred the reaction overnight. After the bis(4-methoxy-
phenyl)methyl group had been successfully removed from the
amino group, selective acetylation of the amino group was
accomplished by treatment with Ac2O/MeOH. At this stage
we performed flash column chromatography to isolate com-
pound 9, which was obtained as a single diastereomer
1
(> 99% de as determined by H and 13C NMR analysis) in
about 55% yield as calculated from the starting material l-
arabinose.[12]
With intermediate 9 in hand, we turned our attention to
the next challenge—the conversion of the vinyl group in 9 into
a g-hydroxy-a-keto acid moiety. Our proposal was to
accomplish this conversion by means of a tandem 1,3-dipolar
cycloaddition and ring-opening reaction without any protec-
tion of the hydroxy groups. In our first attempt, we utilized a
nitrile oxide ((E)-N-(2-ethoxy-2-oxoethylidene)methan-
amine oxide) as the dipole.[13] Unfortunately, although the
1,3-dipolar cycloaddition between this nitrile oxide and
compound 9 proceeded smoothly to provide compound 7,
À
we could not selectively cleave the N O bond in the
=
dihydroisoxazole ring because of the presence of the C N
bond. Having failed with the nitrile oxide, we next tried to use
Scheme 2. One-pot synthesis of intermediate 9 and the possible mechanism of the boronic ester version of Petasis reaction. R= bis(4-
methoxyphenyl)methyl.
7418
www.angewandte.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 7417 –7421

Angewandte
Chemie
a nitrone in the 1,3-dipolar cycloaddition.[14] It is noteworthy
that in the proposed product derived from a nitrone, the
carbon atom at the a position relative to the acid group would
not have the desired oxidation state (see Scheme 3). Without
much literature precedence, we hypothesized that a base
could be used to deprotonate the a-carbon and this might
prompt the b elimination of oxygen from the nitrogen atom,
affording the desired a-keto group. If this hypothesis were
valid, we would obtain the g-hydroxy-a-keto acid moiety from
a vinyl group in an unprecedented way. To test our hypothesis,
we carried out the 1,3-dipolar cycloaddition between com-
pound 9 and the N-methyl nitrone 12 (Scheme 3). Without
much difficulty, the desired product was successfully obtained
in fairly good yields (> 80%). Nevertheless, we were dis-
appointed to observe a complex mixture of diastereomers in
the product. Considerable attempts were carried out to
improve the diastereoselectivity by changing the solvent
(dioxane, MeOH, and EtOH), reaction temperature (from
room temperature to 808C), and the reactant concentration
(from neat to 0.05m). Unfortunately none of the conditions
were satisfactory.
At this stage we went on to test the deprotonation/b-
elimination hypothesis, because at least one of the chiral
centers (i.e. the chiral center at the a position relative to the
ester group) would disappear anyhow. Therefore, we treated
the product mixture with NaOMe (0.05m in MeOH) followed
by an aqueous workup. The hypothesized reaction proceeded
smoothly, affording l-Neu5Ac (and its 4-epimer) as the final
products in 60% yield. (Note: the ester group was also
hydrolyzed under the same reaction conditions.) 1H and
13C NMR analyses showed that the ratio between l-Neu5Ac
and its 4-epimer was about 3:1, which resulted from the
relatively poor diastereoselectivity in the 1,3-dipolar cyclo-
addition. To improve the diastereoselectivity, we tested
different nitrones in the 1,3-dipolar addition. After some
literature searching,[15] it came to our attention that the N-
methyl nitrone 12 (Scheme 3) should exist as a mixture of cis
and trans isomers. In contrast, when a bulky tert-butyl group is
attached to the nitrogen atom, the resulting nitrone 15 should
stay predominantly in the cis form. Taking these facts into
consideration, we expected that improved diastereoselectivity
could be attained by using the N-tert-butyl nitrone 15 in the
1,3-dipolar addition reaction.
To our satisfaction this hypothesis turned out to be
correct. Thus, when the N-tert-butyl nitrone 15 was utilized,
only two of the four possible isomers were produced in about
90% yield from the 1,3-dipolar cycloaddition. More impor-
tantly, these two isomers were found to have markedly
different Rf values by TLC analysis (Rf(major) = 0.56 vs.
Rf(minor) = 0.61 in 1:4 MeOH/CH2Cl2). Consequently it was
fairly easy to separate the two isomers by standard flash
column chromatography. NOE analyses of the purified
products indicate that the major isomer is compound 16.[16]
Solvent, temperature, and reactant concentration were found
to exert profound influences on the product ratio 16/17.
Under the optimal conditions (dioxane, 308C, reactant
concentration 0.05m), the product ration was eventually
optimized to be 10:1. Next, the purified compound 16 was
subjected to the base-catalyzed b elimination, which success-
1
fully afforded l-Neu5Ac (6) in 60% yield. H and 13C NMR
spectra of the synthetic l-Neu5Ac (6) were identical to the
previously reported data for d-Neu5Ac.[17]
Thus, at this point we have successfully synthesized l-
Neu5Ac from readily available l-arabinose. Starting from d-
arabinose we also successfully synthesized d-Neu5Ac by the
same three steps with an overall yield of 28%. It is important
Scheme 3. Synthesis of l-Neu5Ac (6).
Angew. Chem. Int. Ed. 2006, 45, 7417 –7421
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
7419

Communications
to emphasize that the newly developed synthetic method is
primates may provide immunity to HIV infection in such
species. Our synthesis of d-Neu5Gc began from commercially
available d-arabinose. Using the aforementioned one-pot
Petasis coupling we obtained compound 18 in 50% yield
(> 99% de), which was N-acylated by the activated ester of 2-
acetoxyacetic acid (Scheme 4). Next, compound 18 was
coupled with the N-tert-butyl nitrone 15 to afford compound
19 in 80% yield after separation from the undesired isomer
(d.r. 12:1) by column chromatography. Finally, base-catalyzed
b elimination and ester hydrolysis of 20 gave the target
product in 55% yield. Thus, d-Neu5Gc was successfully
highly flexible for the synthesis of sialic acid analogues,
because various aldose sugars (or polyhydroxy aldehydes) can
be employed as the starting material. To demonstrate this, we
describe further syntheses of some interesting sialic acid
derivatives.
d-N-Glycolylneuraminic acid (20, d-Neu5Gc; Scheme 4):
d-Neu5Gc is the hydroxylated derivative of d-Neu5Ac. It is
produced in many non-human mammals, yet it typically is not
detectable in humans.[18] It has been proposed that the
presence of d-Neu5Gc on the cell surface in non-human
Scheme 4. Synthesis of d-Neu5Gc (20).
Scheme 5. Synthesis of truncated and elongated sialic acids.
7420 www.angewandte.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 7417 –7421

Angewandte
Chemie
R. J. Thompson, Top. Curr. Chem. 1997, 186, 119 – 170; d) M. P.
DeNinno, Synthesis 1991, 583 – 593.
synthesized from readily available d-arabinose in three steps
with an overall yield of 22%.
[6] a) S. J. Danishefsky, M. P. DeNinno, S. H. Chen, J. Am. Chem.
Soc. 1988, 110, 3929 – 3940; b) S. J. Danishefsky, W. H. Pearson,
B. E. Segmuller, J. Am. Chem. Soc. 1985, 107, 1280 – 1285;
c) S. H. Kang, H. W. Choi, J. S. Kim, J. H. Youn, Chem.
Commun. 2000, 227; d) E. A. Voight, C. Rein, S. D. Burke, J.
Org. Chem. 2002, 67, 8489 – 8499; e) Dondoni, A. Marra, P.
Merino, J. Am. Chem. Soc. 1994, 116, 3324 – 3336; f) A. Dondoni,
A. Marra, A. Boscarato, Chem. Eur. J. 1999, 5, 3562 – 3572; g) T.
Takahashi, H. Tsukamoto, M. Kurosaki, H. Yamada, Synlett
1997, 1065 – 1066; h) M. Banwell, C. Desavi, K. Watson, J. Chem.
Soc. Perkin Trans. 1 1998, 2251 – 2252; i) L. S. Li, Y. L. Li, Y. Wu,
Org. Lett. 2000, 2, 891 – 894; j) E. A. Voight, C. R. Rein, S. D.
Burke, Tetrahedron Lett. 2001, 42, 8747 – 8749; k) K. G. Liu, S.
Yan, Y. L. Wu, Z. J. Yao, J. Org. Chem. 2002, 67, 6758 – 6763.
[7] a) W. Fitz, M. J. Kim, W. J. Hennen, H. M. Sweers, C.-H. Wong,
J. Am. Chem. Soc. 1988, 110, 6481 – 6486; b) C. Auge, S. David,
C. Gautheron, Tetrahedron Lett. 1984, 25, 4663 – 4664; c) S.
David, C. Auge, Pure Appl. Chem. 1987, 59, 1501 – 1508.
[8] a) T. H. Chan, M. C. Lee, J. Org. Chem. 1995, 60, 4228 – 4232;
b) D. M. Gordon, G. M. Whitesides, J. Org. Chem. 1993, 58,
7937 – 7938; c) M. Warwel, W. Fessner, Synlett 2000, 865 – 867;
d) T. H. Chan, C. J. Li, J. Chem. Soc. Chem. Commun. 1992,
747 – 748.
[9] a) I. A. Kozlov, S. Mao, Y. Xu, X. Huang, L. Lee, P. S. Sears, C.
Gao, A. R. Coyle, K. D. Janda, C.-H. Wong, ChemBioChem
2001, 2, 741 – 746; b) C.-C. Hsu, Z.-Y. Hong, M. Wada, D.
Franke, C.-H. Wong, Proc. Natl. Acad. Sci. USA 2005, 102, 9122.
[10] N. A. Petasis, I. A. Zavialov, J. Am. Chem. Soc. 1998, 120,
11798 – 11799.
[11] a) C. Morrill, R. H. Grubbs, J. Org. Chem. 2003, 68, 6031 – 6034;
b) D. S. Matteson, J. Am. Chem. Soc. 1960, 82, 4228 – 4233.
[12] To further confirm its identity, we performed ozonolysis of
compound 9 and successfully obtained l-ManNAc.
[13] a) K. V. Gothelf, K. A. Jorgensen, Chem. Rev. 1998, 98, 863 –
909; b) V. Jaeger, R. Schohe, Tetrahedron 1984, 40, 2199 – 2210;
c) V. Jaeger, I. Mueller, Tetrahedron 1985, 41, 3519 – 3528.
[14] a) A. P. Kozikowski, Acc. Chem. Res. 1984, 17, 410 – 416; b) D. P.
Curran, J. Am. Chem. Soc. 1983, 105, 5826 – 5833; c) P. Andreas,
A. Vasella, Helv. Chim. Acta 1999, 82, 1044 – 1065; d) R. Huber,
A. Vasella, Helv. Chim. Acta 1987, 70, 1461 – 1476; e) B. Bernet,
E. Krawczyk, A. Vasella, Helv. Chim. Acta 1985, 68, 2299 – 2311.
[15] Y. Inouye, K. Takaya, H. Kakisawa, Bull. Chem. Soc. Jpn. 1983,
56, 3541 – 3542.
[16] In the ROESY NMR spectrum of compound 16, there are strong
H3a–H2 and H3a–H4 NOE correlations. Arelatively weaker
NOE correlation is also observed between H2 and H4. However,
there is no NOE correlation between H3b and H2 (or between
H3b and H4). These data suggest that compound 16 should be
the 2S,4R isomer.
[17] a) E. S. Simon, M. D. Bednarski, G. M. Whitesides, J. Am. Chem.
Soc. 1988, 110, 7159 – 7163; b) The optical rotation value for the
synthetic l-Neu5Ac is [a]2D5 = + 34.38 (c = 0.5 in H2O); the
literature value for d-Neu5Ac (Sigma) is [a]2D5 = À32 Æ 28.
[18] a) E. A. Muchmore, Immun. Rev. 2001, 183, 86; b) Y. N. Malykh,
R. Schauer, L. Shaw, Biochimie 2001, 83, 623; c) A. Varki,
Biochimie 2001, 83, 615.
Truncated and elongated sialic acids: Some synthetic
methods have been developed in the past for the preparation
of 7-carbon analogues of sialic acids.[19] Herein, our own
synthesis of the seven-carbon sialic acid analogue 23 was
started from d-glyceraldehyde. Through the aforementioned
“one-pot” Petasis coupling (60% yield, > 99% de), 1,3-
dipolar cycloaddition (80% yield, d.r. 8:1), and base-cata-
lyzed b elimination and ester hydrolysis (60% yield), the
target compound (27) was successfully obtained with an
overall yield of 29% (Scheme 5). Similarly, we also synthe-
sized the first ten-carbon analogue of sialic acid (26) in three
steps from d-galactose with an overall yield of 22%.
In summary, in the present study we have developed a
novel, efficient method for the synthesis of sialic acids and
derivatives in only three steps from readily available starting
materials. Aboronic ester version of the Petasis coupling was
developed, which facilitated the coupling of an unsubstituted
vinyl group to an unprotected aldose. Next a highly diaste-
reoselective 1,3-dipolar cycloaddition reaction was estab-
lished that enabled the facile conversion of a vinyl group to an
isoxazolidine ring carrying an ester group. Subsequently we
discovered an unprecedented base-catalyzed ring-opening
reaction for the direct conversion of the isoxazolidine to a g-
hydroxy-a-keto acid. All three reactions were completely
compatible with the unprotected hydroxyl groups in carbo-
hydrates. Armed with this novel approach, we successfully
prepared several important, yet very expensive, sialic acid
derivatives, including l-N-acetylneuraminic acid, d-N-glyco-
lylneuraminic acid, a seven-carbon truncated analoguie, and a
ten-carbon elongated analogue of sialic acid, in an econom-
ically competitive fashion.
Received: April 20, 2006
Revised: July 31, 2006
Published online: October 10, 2006
Keywords: cycloaddition · sialic acids · total synthesis
.
[1] a) T. Angata, A. Varki, Chem. Rev. 2002, 102, 439 – 469; b) A.
Rosenberg, Biology of Sialic Acids, Plenum, New York, London,
1995.
[2] Essentials of Glycobiology (Ed.: A. Varki, R. Cummings, J. Esko,
H. Freeze, G. Hart, J. Marth), Cold Spring Harbor Laboratory
Press, New York, 1999.
[3] “N-Acetylneuraminic acid derivatives and mimetics as anti-
influenza agents”: R. Thomson, M. von Itzstein in Carbohy-
drate-Based Drug Discovery (Ed.: C.-H. Wong), Wiley-VCH,
Weinheim, 2003.
[4] a) M. von Itzstein, W. Y. Wu, G. B. Kok, M. S. Pegg, J. C.
Dyason, B. Jin, P. T. Van, M. L. Smythe, H. F. White, S. W.
Oliver, Nature 1993, 363, 418 – 423; b) C. U. Kim, W. Lew, M. A.
Williams, H. Liu, L. Zhang, S. Swaminathan, N. Bischofberger,
M. S. Chen, D. B. Mendel, C. Y. Tai, W. G. Laver, R. C. Stevens,
J. Am. Chem. Soc. 1997, 119, 681 – 690; c) A. Moscona, N. Engl.
J. Med. 2005, 353, 1363 – 1373.
[19] a) M. Warwel, W. D. Fessner, Synlett 2002, 2104; b) M. Banwell,
C. D. Savi, D. Hockless, K. Watson, Chem. Commun. 1998, 645 –
646.
[5] For recent reviews on the syntheses of sialic acids: a) L. S. Li, Y.
L Wu, Curr. Org. Chem. 2003, 7, 447 – 475; b) M. J. Kiefel, M.
von Itzstein, Chem. Rev. 2002, 102, 471 – 490; c) M. von Itzstein,
Angew. Chem. Int. Ed. 2006, 45, 7417 –7421
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
7421