A chemoenzymatic total synthesis of the neurogenic starfish ganglioside LLG-3 using an engineered and evolved synthase

Article abstract of DOI:10.1002/anie.201204578

An LLG-3 oligosaccharide-fluoride (see scheme) can be assembled chemoenzymatically and readily coupled with various sphingosines by an engineered endoglycoceramidase glycosynthase. The lyso?ganglioside products are acylated to generate the individual isomers identified in the heterogeneous natural isolates, as well as modified glycosphingolipids.

Full text of DOI:10.1002/anie.201204578

Angewandte
Chemie
DOI: 10.1002/anie.201204578
Carbohydrates
A Chemoenzymatic Total Synthesis of the Neurogenic Starfish
Ganglioside LLG-3 Using an Engineered and Evolved Synthase
Jamie R. Rich* and Stephen G. Withers
Sialic acid containing glycosphingolipids (GSLs), or ganglio-
sides, are common to all vertebrate cells and play important
roles in the pathology of a variety of human ailments,
including Alzheimers disease, lysosomal storage disorders,
and kidney disease, as well as in the function of the
mammalian nervous system.[1] Accordingly, their synthesis
has been the focus of considerable efforts.[2] Although the
successful synthesis of gangliosides is still very much the
domain of the specialist laboratory, the state of the art has
advanced to the point where preparation of molecules
appropriate for investigating the influence of subtle structural
alterations on the function of GSLs is a reality.
Gangliosides are largely absent from the invertebrate
lineage with the exception of the echinoderms, from which
numerous molecules with rare or unique structural features
have been identified in relative abundance.[3] Examples
include GSLs containing modified sialic acids that are
frequently attached to other sialic acid residues by either
a(2–4) or a(2–11) linkages. Echinodermatous gangliosides
are also characterized by highly heterogeneous sphingolipids,
with structures uncommon in vertebrate systems predom-
inating. Several gangliosides from starfish, including the
tetrasaccharide LLG-3 from Linckia laevigata (1,
Scheme 1),[4] show promising evidence of neuritogenic activ-
ity in cell culture and are thus of pharmacological interest.[5]
Echinodermatous gangliosides have attracted attention
from synthetic chemists, in particular from Kiso and co-
workers, who recently reported an impressive total synthesis
of LLG-3.[6] We have been eager to develop an enzymatic
synthesis of 1 wherein the attendant benefits of biocatalysis
could be demonstrated, including high-yielding regio- and
stereo-selective coupling reactions and a minimum of syn-
thetic steps. However, the enzymes involved in the biosyn-
thesis of LLG-3 are uncharacterized and, most importantly,
no sialyltransferase (ST) has been found to catalyze the
formation of the relatively uncommon a(2–11) sialosyl link-
age. Herein we report a highly flexible chemoenzymatic
synthesis of the ganglioside LLG-3, thus highlighting the
suitability of an engineered and evolved endoglycoceramida-
se II glycosynthase for the synthesis of complex glycosphin-
golipids that feature natural heterogeneity in the sphingoli-
pid, rare sphingolipids, and “unusual” structural modifica-
tions of the oligosaccharide.
Endoglycoceramidase II (EGCase) from Rhodococcus sp.
strain M-777 is a retaining glycoside hydrolase (GH) from
CAZy family GH5 that catalyzes cleavage of glycosyl
ceramides to liberate an intact reducing oligosaccharide and
a ceramide.[7] This enzyme has been subjected to two rounds
of protein engineering. Firstly, a glycosynthase (GS) mutant
(E351S) was generated, giving glycosyl sphingosines through
coupling of an oligosaccharide fluoride with
a sphingosine.[8] Subsequently, the capacity
of this GS to utilize diverse sphingolipid
substrates, including d-ribo-phytosphingo-
sine, was enhanced many thousand fold
using directed evolution, resulting in the
mutant E351S D314Y.[9] In addition to an
engineered broad lipid specificity, EGCase
GS has been demonstrated to utilize a range
Scheme 1. The predominant structural isomer of LLG-3 from Linckia laevigata.
of structurally diverse glycosyl fluoride
substrates for the synthesis of vertebrate
LLG-3 from natural sources exists as a heterogeneous
mixture of ceramides with variability in both the d-ribo-
phytosphingosine and a-hydroxy acid components. Under-
standing how the structural features of LLG-3 contribute to
neurotrophic activity demands access to structurally well-
defined homogeneous material most readily provided by
synthesis.
GSLs, including GM3, GM1, and sialyl paragloboside.[10] The
merits of an enzyme with flexibility toward both donor and
acceptor substrates are obvious, and in this case access to such
an enzyme permits ready synthesis of GSL libraries. Subse-
quent chemical or enzymatic N-acylation of lyso-glycosyl
sphingosines, obtained by coupling of sphingosine(s) with
glycosyl fluorides, generates additional structural diversity
and permits the chemist to interrogate the importance of
individual lipid structures observed within heterogeneous
natural isolates.[1d] Our recent work has focused on establish-
ing enzymatic methods for the synthesis of complex oligo-
saccharide fluorides from simpler glycosyl fluorides, and the
use of these molecules with EGCase GS.
[*] Dr. J. R. Rich, Prof. S. G. Withers
Department of Chemistry, University of British Columbia
Vancouver, B.C., V6T 1Z1 (Canada)
E-mail: jrich@chem.ubc.ca
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201204578.
Angew. Chem. Int. Ed. 2012, 51, 1 – 5
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1
These are not the final page numbers!

.
Angewandte
Communications
In the absence of a sialyltransferase capable of sialyl-
transfer to the hydroxy group of the glycolyl amide in
Neu5Gc, enzymatic synthesis of this linkage remains imprac-
tical. Furthermore, methylation of the sialic acid is likely to
occur after the glycosylation step in nature, and neither the
requisite S-adenosyl-l-methionine:Neu5Ac 8-O-methyltrans-
ferase nor those membrane-bound enzymes observed in
related organisms have been cloned.[11] Fortunately, methods
for the chemical construction of the a(2–11) bond have been
reported.[12,6] We envisioned an approach to LLG-3 in which
chemoenzymatically generated Neua(2,3)Galb(1,4)Glca-F 2
could first be coupled chemically with 8-OMe Neu5Aca-Gc 3.
After deprotection, 4 would serve as a donor substrate of
EGCase GS for coupling to various lipids, potentially afford-
ing different glycosyl sphingosine variants of 1 in a remarkably
short reaction sequence (Scheme 2).
Scheme 3. Preparation of trisaccharide 2. a) Neu5Ac aldolase, pyruvic
acid; b) CMP-Neu5Ac synthetase, cytidine-5’-triphosphate, inorganic
pyrophosphatase; c) a-2,3-sialyltransferase, a-lactosyl-fluoride, alkaline
phosphatase, 70% for 7 (2 steps), 57–94% for 8 (3 steps); d) NaOMe,
MeOH, 44%; e) Pd/C, H2, MeOH, 94%.
conversion of NeuTFA 5 to trisaccharide 7. Unfortunately,
yields for deprotection of 7 were low (44%). However,
ManCbz 6 could be smoothly converted to the trisaccharide 8,
which was then isolated from a mixture of five enzymes and
associated substrates and by-products in yields of up to 94%
using solid-phase extraction (SPE). Hydrogenolysis of 8 over
Pd/C afforded the amine 2.
8-O-Methyl modification of sialic acids is relatively rare,
but nevertheless may be of considerable biological impor-
tance. For instance, LLG-3 contains 75 carbon atoms and
differs from ganglioside LMG-4 (Luidia maculata) only by 8-
O-methylation of Neu5Ac, yet LLG-3 shows significantly
more potent neuritogenic activity toward the rat pheochro-
mocytoma cell line PC-12.[5] This modification also clearly
prevents establishment of the a(2–8) linkage that is character-
istic of most polysialic acids. Both chemical and chemo-
enzymatic syntheses of 8-OMe sialic acid have been repor-
ted.[6,12e,15] Our synthesis of an 8-OMe sialic acid functional-
ized for coupling to 2 is partially illustrated in Scheme 4. Allyl
glycoside 9 underwent acid-catalyzed transacetalization to
afford 10 and 11. Despite an almost quantitative yield, the 2:3
equilibrium ratio of regioisomers was unfavourable. Selective
precipitation of 11 and subsequent re-equilibration of the
dioxolane 10 gave 11 in 65% yield. A sequence of O-
methylation, protecting group exchange, oxidation, and
partial deprotection afforded 3, the 8-OMe sialoside of
glycolic acid, in good yield. Protection of C-1 was maintained
to allow discrimination between the two carboxylic acids
present.
Scheme 2. Strategy for the construction of LLG-3 and structural
variants.
Methods for the chemoenzymatic synthesis of the trisac-
charide 2 were explored with the expectation that sialylation
of lactosyl fluoride using ST Cst-I (Campylobacter jejuni) and
a modified CMP-sialic acid donor would give an amine-
masked precursor of 2 (Scheme 3). We identified the azide,
trifluoroacetamide (TFA), and N-carboxybenzyl (Cbz) func-
tionalities as potentially suitable amine protecting groups.
Although 2-azido-2-deoxy-d-mannose could be transformed
into 5-azido-sialylactose through the actions of Neu5Ac
aldolase (Escherichia coli), CMP-sialic acid synthetase NSY-
05 (Neisseria meningitides 406Y),[13] and a(2,3)-sialyltransfer-
ase Cst-I,[14] yields were much inferior to those obtained for
Coupling of the amine 2 with acids 12 or 3 (Scheme 4)
could be achieved under common peptide coupling condi-
2
www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 1 – 5
These are not the final page numbers!

Angewandte
Chemie
the reaction of 4 with C18 d-ribo-phytosphingosine to afford
isolated methyl ester 15 in over 70% yield on a 20–40 mg
scale (Scheme 5). Cleavage of the ester could be achieved
with LiCl in pyridine to give the lyso-ganglioside 17.[12d]
A
fluorescent analogue of C18 LLG-3, 18, was prepared by
acylation of 15 with a BODIPY-FL-containing carboxylic
acid. To obtain the main target LLG-3 (1), the EGCase GS
product, C17-lyso-LLG-3 16, was acylated with an appropriate
a-hydroxy acid and deprotected by sequential cleavage of the
methyl ester and saponification of the remaining acetyl group.
The structural complexity observed in GSLs from echi-
noderms is remarkable, as is their potential for studying nerve
growth. We have developed a chemoenzymatic synthesis of
LLG-3 that capitalizes on the structurally permissive nature
of bacterial enzymes, namely Neu5Ac aldolase (E. coli),
CMP-sialic acid synthetase NSY-05 (N. meningitides 406Y),
and the a(2,3) sialyltransferase Cst-I (C. jejuni), for assembly
of highly unusual tetrasaccharide fluorides. Furthermore, our
synthesis exploits an endoglycoceramidase (Rhodococcus sp.)
for coupling of the oligosaccharide with various sphingosines.
This latter enzyme has been engineered to catalyze synthesis
instead of degradation, and then further modified to build on
its already liberal substrate specificity. Together, these
catalysts have permitted an efficient synthesis of the LLG-3
ganglioside. Most importantly, this approach allows ready
variation of each of the three principal constituents of
gangliosides: the oligosaccharide, sphingosine, and N-acyl
components. As a consequence, it is possible to produce
Scheme 4. Synthesis of 8-OMe-Neu5Aca(2-11)Neu5Gca(2–3)Galb(1-
4)GlcaF. a) PhCH(OMe)2, MeCN, camphorsulfonic acid, 64%; b) MeI,
NaH, DMF, À108C, 85%; c) 80% AcOH, 558C, 84%; d) Ac2O, CH2Cl2,
pyridine, 92%; e) NaIO4, RuCl3, CCl4, MeCN, H2O, 86%; f) NaOMe,
MeOH, 96%; g) PyBOP, DMF, 08C, then 2, 78% for 13, 81% for 4;
h) 30 mm 4, 100 mm LiOH, D2O, 75%. DMF=N,N-dimethylforma-
mide, PyBOP=1-benzotriazolyloxy-tris(pyrollidino)phosphonium hexa-
fluorophosphate.
tions, giving the tetrasaccharide-fluorides 13 and 4,
respectively. When PyBOP was added in greater
excess, one or more tetrasaccharide lactone by-
products were observed. The “preactivated” spi-
rolactone approach reported by Hindsgaul was not
pursued, owing to anticipated formation of regioi-
someric amides.[12c] Under standard conditions for
transesterification or saponification, both 13 and 4
underwent rapid intramolecular transamidation,
presumably via an intermediate cyclic imide.[12d]
Careful control of the pH value and duration of
the reaction allowed isolation of 14 with minimal
(< 10%) formation of the undesired amide.
Unfortunately, conditions that were designed to
prevent this side reaction altogether (LiCl, pyri-
dine, elevated temperature) resulted in poor yields
for the conversion of the glycosyl fluoride 4 to 14.
Scheme 5. Reaction conditions: a) EGCase glycosynthase, sphingosine, 71% for 15;
b) LiCl, pyridine, 1008C, 72%; c) (R)-2-O-acetyl-tricosanoic acid, PyBOP, DMF,
Et3N; d) LiCl, pyridine, 1008C; e) NaOH, 308C, 66% from 4; f) Bodipy-FL-OSu,
DMF, Et3N, 40%.
As characterization of a related GSL was report-
edly facilitated through esterification, we opted to
delay deprotection of the acid until after coupling
of 4 with the sphingosine.[16]
Test reactions indicated that EGCase mutant
E351S/D314Y catalyzed the coupling of 4 or 14 with d-
erythro-sphingosine and the two most abundant forms
identified in isolated LLG-3, the C17 and C18 d-ribo-phytos-
phingosines. In addition, the enzyme tolerated acetylated
donor 13, thus suggesting its utility in the preparation of the
widely distributed O-acetylated gangliosides, for which syn-
thetic methods are generally lacking. EGCase GS catalyzed
structurally defined homogeneous samples of individual
isoforms found among significantly heterogeneous natural
isolates, and furthermore, to create new ganglioside ana-
logues. Such molecules promise to be of use in delineating the
structural basis for the neuritogenic nature of some GSLs, and
could also assist in the identification of as-yet-unstudied
biochemical pathways in echinoderms.
Angew. Chem. Int. Ed. 2012, 51, 1 – 5
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3
These are not the final page numbers!

.
Angewandte
Communications
[9] S. M. Hancock, J. R. Rich, M. E. Caines, N. C. J. Strynadka, S. G.
Withers, Nat. Chem. Biol. 2009, 5, 508.
Received: June 12, 2012
Published online: && &&, &&&&
[10] J. R. Rich, A. Cunningham, M. Gilbert, S. G. Withers, Chem.
Commun. 2011, 47, 10806.
[11] a) A. A. Bergwerff, S. H. D. Hulleman, J. P. Kamerling, J. F. G.
Vliegenthart, L. Shaw, G. Reuter, R. Schauer, Biochimie 1992,
74, 25; b) A. Kelm, L. Shaw, R. Schauer, G. Reuter, Eur. J.
Biochem. 1998, 251, 874.
Keywords: carbohydrates · chemoenzymatic synthesis ·
gangliosides · glycosynthase · sialic acids
.
[1] a) X. Zhang, F. L. Kiechle, Ann. Clin. Lab. Sci. 2004, 34, 3; b) T.
Kolter, Chem. Phys. Lipids 2011, 164, 590; c) A. R. Mather, L. J.
Siskind, Adv. Exp. Med. Biol. 2011, 721, 121; d) C. A. Lingwood,
Cold Spring Harb. Perspect. Biol. 2011, DOI: 10.1101/cshper-
spect.a004788.
[2] Y. D. Vankar, R. R. Schmidt, Chem. Soc. Rev. 2000, 29, 201.
[3] R. Higuchi, M. Inagaki, K. Yamada, T. Miyamoto, J. Nat. Med.
2007, 61, 367.
[4] M. Inagaki, R. Isobe, R. Higuchi, Eur. J. Org. Chem. 1999, 771.
[5] M. Kaneko, K. Yamada, T. Miyamoto, M. Inagaki, R. Higuchi,
Chem. Pharm. Bull. 2007, 55, 462.
[6] H. Tamai, H. Ando, H.-N. Tanaka, T. Hosoda-Yabe, T. Yabe, H.
Ishida, M. Kiso, Angew. Chem. 2011, 123, 2378; Angew. Chem.
Int. Ed. 2011, 50, 2330.
[12] a) C.-T. Ren, C.-S. Chen, S.-H. Wu, J. Org. Chem. 2002, 67, 1376;
b) G.-T. Fan, C.-C. Lee, C. C. Lin, J.-M. Fang, J. Org. Chem. 2002,
67, 7565; c) J. C. McAuliffe, D. Rabuka, O. Hindsgaul, Org. Lett.
2002, 4, 3067; d) C.-T. Ren, C.-S. Chen, Y.-P. Yu, Y.-F. Tsai, P.-Y.
Lin, Y.-J. Chen, W. Zou, S.-H. Wu, Chem. Eur. J. 2003, 9, 1085;
e) S. Hanashima, D. Ishikawa, S. Akai, K.-i. Sato, Carbohydr.
Res. 2009, 344, 747.
[13] M.-F. Karwaski, W. W. Wakarchuk, M. Gilbert, Protein Expres-
sion Purif. 2002, 25, 237.
[14] C. P. Chiu, L. L. Lairson, M. Gilbert, W. W. Wakarchuk, S. G.
Withers, N. C. J. Strynadka, Biochemistry 2007, 46, 7196.
[15] a) G. Dufner, R. Schworer, B. Muller, R. R. Schmidt, Eur. J. Org.
Chem. 2000, 1467; b) T. J. Morley, S. G. Withers, J. Am. Chem.
Soc. 2010, 132, 9430; c) H. Yu, H. Cao, V. M. Tiwari, Y. Li, X.
Chen, Bioorg. Med. Chem. Lett. 2011, 21, 5037; d) A. Y. Khorlin,
I. M. Privalova, Carbohydr. Res. 1970, 13, 373.
[7] H. Izu, Y. Izumi, Y. Kurome, M. Sano, A. Kondo, I. Kato, M. Ito,
J. Biol. Chem. 1997, 272, 19846.
[8] M. D. Vaughan, K. Johnson, S. Defrees, X. Tang, R. A. J.
Warren, S. G. Withers, J. Am. Chem. Soc. 2006, 128, 6300.
[16] M. Inagaki, T. Miyamoto, R. Isobe, R. Higuchi, Chem. Pharm.
Bull. 2005, 53, 1551.
4
www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 1 – 5
These are not the final page numbers!

Angewandte
Chemie
Communications
Carbohydrates
J. R. Rich,* S. G. Withers
&&&& — &&&&
A Chemoenzymatic Total Synthesis of the
Neurogenic Starfish Ganglioside LLG-3
Using an Engineered and Evolved
Synthase
An LLG-3 oligosaccharide–fluoride (see
scheme) can be assembled chemoenzy-
matically and readily coupled with various
sphingosines by an engineered endogly-
coceramidase glycosynthase. The lyso
ganglioside products are acylated to
generate the individual isomers identified
in the heterogeneous natural isolates, as
well as modified glycosphingolipids.
Angew. Chem. Int. Ed. 2012, 51, 1 – 5
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5
These are not the final page numbers!