Automated solid-phase synthesis of hyaluronan oligosaccharides

Article abstract of DOI:10.1021/ol301666n

Well-defined fragments of hyaluronic acid (HA) have been obtained through a fully automated solid-phase oligosaccharide synthesis. Disaccharide building blocks, featuring a disarmed glucuronic acid donor moiety and a di-tert-butylsilylidene-protected gluc

Full text of DOI:10.1021/ol301666n

ORGANIC
LETTERS
XXXX
Vol. XX, No. XX
000–000
Automated Solid-Phase Synthesis
of Hyaluronan Oligosaccharides
Marthe T. C. Walvoort,† Anne Geert Volbeda,† Niels R. M. Reintjens,†
Hans van den Elst,† Obadiah J. Plante,‡ Herman S. Overkleeft,†
,†
Gijsbert A. van der Marel,* and Jeroen D. C. Codee*
,†
ꢀ
Leiden Institute of Chemistry, Leiden University, P.O. Box 9502, 2300 RA Leiden,
The Netherlands, and Ancora Pharmaceuticals, 1B Gill Street, Woburn,
Massachusetts 01801, United States
marel_g@chem.leidenuniv.nl; jcodee@chem.leidenuniv.nl
Received June 18, 2012
ABSTRACT
Well-defined fragments of hyaluronic acid (HA) have been obtained through a fully automated solid-phase oligosaccharide synthesis.
Disaccharide building blocks, featuring a disarmed glucuronic acid donor moiety and a di-tert-butylsilylidene-protected glucosamine part,
were used in the rapid and efficient assembly of HA fragments up to the pentadecamer level, equipped with a conjugation-ready anomeric allyl
function.
Hyaluronan, or hyaluronic acid (HA), a member of
the glycosaminoglycan (GAG) family, is composed of
[f4)-β-D-GlcpA-(1f3)-β-D-GlcpNAc-(1f] tandem re-
peats reaching up to 104 disaccharides (∼3.7 Â 106 Da)
in length.1 HA is a major component of the extracellular
matrix, connective tissue, and synovial fluid in mammals
and also occurs in capsules of certain bacteria. Besides
functioning as a molecular lubricant, HA plays an
important role in many biological processes, including
inflammatory response, cellular proliferation, cellÀcell
recognition, cell migration, and cell adhesion.1À3 These
processes depend on interactions with a variety of HA
binding proteins on the cell surface or in the extracellular
fluid. One of the most prominent examples of HA-binding
proteins is CD44, a transmembrane receptor present
on leukocytes, fibroblasts, endothelial cells, and epithelial
cells, which is involved in many processes including
lymphocyte recruitment, T-cell signaling, apoptosis, and
tumor metastasis.4 The mode of action of HA fragments
has been shown to depend on its length. For example,
whereas long HA chains are immunosuppressive, small
HA fragments are immunostimulatory and function as an
endogenous danger signal.1,5,6 In the same vein, long HA
stretches are required for the formation of CD44 signaling
(4) Teriete, P.; Banerji, S.; Noble, M.; Blundell, C. D.; Wright, A. J.;
pickford, A. R.; Lowe, E.; Mahoney, D. J.; Tammi, M. I.; Kahmann,
J. D.; Campbell, I. D.; Day, A. J.; Jackson, D. G. Mol. Cell 2004, 13,
483–496.
(5) Yawamaki, H.; Hirohata, S.; Miyoshi, T.; Takahashi, K.; Ogawa,
H.; Shinohata, R.; Demircan, K.; Kusachi, S.; Yamamoto, K.; Ninomiya,
Y. Glycobiol. 2009, 19, 83–92.
(6) Termeer, C.; Benedix, F.; Sleeman, J.; Fieber, C.; Voith, U.;
Ahrens, T.; Miyake, K.; Freudenberg, M.; Galanos, C.; Simon, J. C.
J. Exp. Med. 2002, 195, 99–111.
† Leiden University.
‡ Ancora Pharmaceuticals.
(1) Esko, J. D.; Kimata, K.; Lindahl, U. In Essentials of Glycobiol-
ogy; Varki, A., Cummings, R. D., Esko, J. D., Freeze, H. H., Stanley, P.,
Bertozzi, C. R., Hart, G. W., Etzler, M. E., Eds.; Cold Spring Harbor
Laboratory Press: New York, 2009; pp 229À248.
(2) (a) Toole, B. P. Nat. Rev. Cancer 2004, 4, 528–539. (b) Toole, B. P.
Clin. Can. Res 2009, 15, 7462–7468.
(3) Jiang, D.; Liang, J.; Noble, P. W. Physiol. Rev. 2011, 91, 221–264.
r
10.1021/ol301666n
XXXX American Chemical Society

complexes that are thought to govern cancer cell prolifera-
tion, where small HA fragments appear to act as CD44
antagonists.1,2,7 HA fragments with a minimum size of six
monosaccharides can bind to CD44, and a HA 10-mer
effectively competes for binding with full length HA.8
To study HAÀprotein interaction, the availability of
well-defined HA fragments is a prerequisite, and therefore,
the synthesis of HA has been actively pursued. Various
strategies have been developed toward its assembly,9
including enzymatic10 and chemical methods involving
both postglycosylation11 and preglycosylation oxidation
approaches,12 and one-pot procedures.13 Recently, the first
studies toward soluble polymer-supported syntheses have
been described.14 Chemical solution-phase synthesis has
provided access to well-defined HA fragments composed
of two to ten monosaccharide residues.12e Application of a
soluble polymer support has delivered an HA-dimer.14
The repetitive nature of the HA polymer invites the
assembly of well-defined oligomers by means of an auto-
mated solid-phase approach.15 However, and as opposed
to the automated synthesis of oligopeptides and oligonu-
cleotides, the automated solid-phase synthesis of carbohy-
drates is not yet a routine operation and is hampered by
the lack of a standard set of carbohydrate building blocks
and coupling chemistry. A major challenge in the assembly
of HA, and GAGs in general, is the low reactivity of
the building blocks required.16 Previous work on the
soluble polymer-supported assembly of HA and heparin
fragments has made it clear that the translation of a
solution-phase synthesis to the (solid) support is not
a trivial operation.17 Solid-phase synthesis approaches
(including automated protocols), however, have the
intrinsic advantage that coupling reactions can be forced
to completion by the use of excess reagents and repetitive
coupling cycles. Although repetitive cycles and excess
reagent indicate the need for significant amounts of
building blocks, the fact that the overall assembly can
be high-yielding allows one to start an assembly sequence
on a relatively small scale, which would make the process
in fact quite building block efficient.15c This holds true
especially for higher oligomers composed of repeating
(mono)saccharides, such as present in HA. We now de-
scribe the first automated synthesis of a set of HA oligo-
mers up to the pentadeca level. Our work entails the first
example of the construction of well-defined short- and
medium-sized GAG oligomers using an automated solid-
phase carbohydrate synthesis protocol.
Our approach is based on the use of Merrifield resin,
functionalized with a butenediol linker system15a in com-
bination with a monomeric glucosamine synthon (1),12d
and the repetitive use of an orthogonally protected
GlcNHAc-GlcA building block (5, Scheme 1). The bute-
nediol linker is inert to all reaction conditions used during
the assembly of the oligomers and can be cleaved through a
cross-metathesis reaction to deliver an anomeric O-allyl
functionality. Because of the protecting group scheme
devised, the anomeric allyl group can be retained until
the end of the synthesis and immediately serve as a ligation
handle.18 The glucosamine and dimer building blocks 1
and 5 feature a di-tert-butylsilylidene ketal to mask the
C4- and C6-hydroxyl functions. This protecting group has
been selected because of its excellent acid stability, which
is of prime importance given the fact that repetitive
glycosylations are performed using relatively large amount
of Lewis acid (with respect to conventional solution-phase
conditions). The influence of the amine-protecting group
in building block 1 was assessed in a series of model
glycosylation reactions (Supporting Information). From
the protecting groups scrutizined (trichloroacetyl, trifluoro-
acetyl, trichloroethoxycarbonyl, and benzyloxycarbonyl),
the trichloroacetyl-protected glucosamine 1 emerged as
the most productive donor of the series. With this donor,
the key disaccharide building block 5 was assembled
as depicted in Scheme 1. In a chemoselective glycosyla-
tion reaction, N-phenyl trifluoroacetimidate donor 1 was
condensed with S-phenylglucuronic acid 2 (obtained from
D-glucose in seven steps through solely crystalline inter-
mediates, see the Supporting Information) to furnish
dimer 3,12d which was transformed into imidate donor 5
through hydrolysis of the thioacetal and installation of
the N-phenyltrifluoroacetimidate function.19,20 Following
this approach, building block 5 was readily assembled on
a multigram scale.
(7) Misra, S.; Heldin, P.; Hascall, V. C.; Karamanos, N. K.; Skandalis,
S. S.; Markwald, R. R.; Ghatak, S. FEBS J. 2011, 278, 1429–1443.
€
(8) Tammi, R.; MacCallum, D.; Hascall, V. C.; Pienimaki, J.-P.;
Hyttinen, M.; Tammi, M J. Biol. Chem. 1998, 273, 28878–28888.
(9) See for a review on GAG synthesis: (a) Yeung, B. K. S.; Chong,
P. Y. C.; Petillo, P. A. J. Carbohydr. Chem. 2002, 21, 799–865. (b) Karst,
N. A.; Linhardt, R. J. Curr. Med. Chem. 2003, 10, 1993–2031.
(10) DeAngelis, P. L.; Oatman, L. C.; Gay, D. F. J. Biol. Chem. 2003,
27, 35199–35203.
€
(11) See, for example: (a) Slaghek, T. M.; Hypponen, T. K.; Kruiskamp,
P. H.; Ogawa, T.; Kamerling, J. P.; Vliegenthart, J. F. G. Tetrahedron
Lett. 1993, 34, 7939–7942. (b) Adamski-Werner, S. L.; Yeung, B. K. S.;
Miller-Deist, L. A.; Petillo, P. A. Carbohydr. Res. 2004, 339, 1255–1262.
Huang, L.; Huang, X. Chem.;Eur. J. 2007, 13, 529–540.
(12) (a) Blatter, G.; Jacquinet, J.-C. Carbohydr. Res. 1996, 288,
109–125. (b) Iyer, S. S.; Rele, S. M.; Baskaran, S.; Chaikof, E. L.
ꢀ
Tetrahedron 2003, 59, 631–638. (c) Dinkelaar, J.; Codee, J. D. C.;
van den Bos, L. J.; Overkleeft, H. S.; van der Marel, G. A. J. Org. Chem.
ꢀ
2007, 72, 5737–5742. (d) Dinkelaar, J.; Gold, H.; Overkleeft, H. A.; Codee,
J. D. C.;vanderMarel, G. A. J. Org. Chem. 2009, 74, 4208–4216. (e) Lu, X.;
Kamat, M. N.; Huang, L.; Huang, X. J. Org. Chem. 2009, 74, 7608–7617.
(13) Huang, L.; Huang, X. Chem.;Eur. J. 2007, 13, 529–540.
(14) (a) de Paz, J. L.; Mar Kayser, M.; Macchione, G.; Nieto, P. M.
Carbohydr. Res. 2010, 345, 565–571. (b) Mar Kayser, M.; de Paz, J. L.;
Nieto, P. M. Eur. J. Org. Chem. 2010, 2138–2147.
(15) (a) PLante, O. J.; Palmacci, E. R.; Seeberger, P. H. Science 2001,
€
291, 1523–1527. (b) Krock, L.; Esposito, D.; Castagner, B.; Wang,
€
C.-C.; Bindschadler, P.; Seeberger, P. H. Chem. Sci. 2012, 3, 1617–1622.
€
(c) Walvoort, M. T. C.; van den Elst, H.; Plante, O. J.; Krock, L.;
Seeberger, P. H.; Overkleeft, H. S.; van der Marel, G. A.; Codee,
ꢀ
J. D. C. Angew. Chem., Int. Ed. 2012, 51, 4393–4396.
(16) (a) de Jong, A.-R.; Hagen, B.; van der Ark, V.; Overkleeft, H. S.;
ꢀ
Codee, J. D. C.; van der Marel, G. A. J. Org. Chem. 2012, 77, 108–125.
(b) Zeng, Y.; Wang, Z.; Whitfield, D.; Huang, X. J. Org. Chem. 2008, 73,
7952–7962.
(18) Dondoni, A.; Marra, A. Chem. Soc. Rev. 2012, 41, 573–586.
(19) Yu, B.; Tao, H. Tetrahedron Lett. 2001, 42, 2405–2407.
(20) Gold, H.; Munneke, S.; Dinkelaar, J.; Aerts, J. M. F. G.;
(17) (a) Ojeda, R.; de Paz, J. L.; Martı
2003, 39, 2486–2487. (b) Ojeda, R.; Terentı
Lomas, M. Glycoconjugate J. 2004, 21, 179–195. (c) Czechura, P.;
Guedes, N.; Koptzki, S.; Vazquez, N.; Martın-Lomas, M.; Reichardt,
´
n-Lomas, M. Chem. Commun.
´
, O.; de Paz, J. L.; Martın-
´
ꢀ
´
Overkleeft, H. S.; Codee, J. D. C.; van der Marel, G. A. Carbohydr.
N. C. Chem. Commun. 2011, 47, 2390–2392.
Res. 2011, 346, 1467–1478.
B
Org. Lett., Vol. XX, No. XX, XXXX

explored several possibilities to circumvent this side
reaction.22 The use of different metathesis catalysts was
to no avail, and we therefore switched to the use of a
“decoy” substrate in the cleavage reaction. The addition of
an excess trichloroacetamide to the metathesis reaction
proved effective and completely suppressed the side reac-
tion. With these optimized reaction conditions in hand
we set out to assemble hepta-, undeca-, and pentadecasac-
charidic fragments of hyaluronic acid, as depicted in
Scheme 2. The three syntheses all commenced with glyco-
sylation of resin 6 with glucosamine donor 1 and subse-
quent Lev-deprotection. Three coupling/deprotection
cycles with disaccharide donor 5 gave heptasaccharide 8,
five coupling/deprotection cycles with 5 gave undecasac-
charide 9, and seven coupling/deprotection cycles with 5
provided pentadecasaccharide 10.23
Scheme 1. Synthesis of Disaccharide Building Block 5
At the onset of the solid-phase assembly of the HA
oligomers, it became clear that the disaccharide synthon 5
could not be used as the first building block,21 and we
therefore started our automated syntheses with the cou-
pling of glucosamine 1 on the solid support (6). To probe
the two building blocks, the on-resin deprotection steps,
and the efficacy of the cross-metathesis cleavage reaction
we initially assembled trisaccharide 7 (Scheme 2). Treat-
ment of resin 6 with building block 1 (2.7 equiv donor,
0.33 equiv TfOH, DCM, 0 °C, repeated three times)
was followed by levulinoyl removal (H2NNH2.HOAc,
7.8 equiv, pyridine/AcOH, 40 °C, repeated twice) (see the
Supporting Information for detailed reaction conditions).
Next, a similar reaction sequence using building block5 led
to the resin-bound trisaccharide. Upon cleavage from the
resin the crude trisaccharide 7 was obtained in 90% yield.
LCÀMS and NMR analysis of this trimer revealed that
both glycosylation reactions had proceeded with excellent
stereoselectivity and that no deletion sequences or orthoe-
ster side products were formed. The only observable side
product produced in this sequence of reactions proved to
be a trisaccharide of which one of the TCA groups was
transformed into a dichloroacetyl group (∼5%). In the
projected automated synthesis of higher HA oligomers,
partial conversion of some of the TCA protecting groups
to a DCA would significantly hinder characterization and
purification of the products, and in addition, harsher
conditions would be required for the removal of the
DCA groups at the end of the synthesis. Importantly, these
problems would increase drastically with the growing
length of the desired oligosaccharides. Dehalogenation of
the TCA functionalities could be the result of the nucleo-
philic attack of a tricyclohexylphosphine ligand of the
Grubbs catalyst on one of the chlorine atoms, and we
Scheme 2. Automated Solid-Phase HA-Assembly
LCÀMS analysis of the crude heptamer 8 (Supporting
Information) indicated that the synthesis had proceeded
very efficiently and only a minor deletion sequence was
detected (ratio pentamer/heptamer ∼1:80). The undeca-
and pentadecamers 9 and 10 proved to be too lipophilic for
HPLC analysis, but MALDI mass spectroscopy con-
firmed the presence of the target compounds in the crude
reaction mixtures. Partial deprotection of the oligomers
by cleavage of the silylidene ketals led to more hydrophilic
compounds, which could be analyzed and readily purified
by HPLC as depicted in Figure 1. Although there is rela-
tively little experience in the HPLC purification of pro-
tected oligosaccharides, these results indicate that a strat-
egy that allows for the partial deprotection of the material
(21) The combination of a reactive primary alcohol acceptor and a
glucuronic acid donor bearing a C-2 acyl protecting group can give rise
to acyl migration in the course of the glycosylation reaction. For this
ꢀ
type of migration, see: Berces, A.; Whitfield, D. M.; Nukada, T.; do
Santos Z., I.; Obuchowska, A.; Krepinsky, J. J. Can. J. Chem. 2004, 82,
1157–1171. Also see ref 12a.
(22) Kanemitsu, T.; Seeberger, P. H. Org. Lett. 2003, 5, 4541–4544.
(23) During the pentadacamer assembly, the reaction mixtures and
subsequent DCM washes were collected, from which 37% of unreacted
donor 5 was recovered.
Org. Lett., Vol. XX, No. XX, XXXX
C

that is released from the resin, such as described here,
delivers intermediates of such a polarity that they are
readily amendable to routine HPLC purification. In this
way, heptasaccharide 11 was isolated in 26% over 10 steps
(∼87% per step), undecasaccharide 12 in 32% over 14
steps (∼92% per step), and pentadecasaccharide 13in 18%
over 18 steps (∼91% per step), starting from 45 μmol of
functionalized resin 6.
Figure 2. Fragments of the 1H NMR (top) and 13C NMR
(bottom) spectra of pentadecamer 16.
fully automated fashion. The synthesis strategy is based on
the combined use of mono- and dimeric building blocks
and proceeded both rapidly and efficiently. The disacchar-
ide building block, which was used for the repetitive
elongation cycles, represents a disarmed donor glycoside,
indicating that low reactivity of glycosyl donors presents
no obstacle for the automated solid-phase assembly plat-
form. The protecting group scheme devised for the synth-
esis not only led to an effective assembly process but also
allowed for the incorporation of a conjugation-ready allyl
functionality on the anomeric center through the use of a
butenediol linker system in combination with an optimized
cross-metathesis cleavage reaction. The reported solid-
phase assembly indicates that the automated solid-phase
synthesis of other members of the glycosaminoglycan
family is within reach. Because of the modular nature of
the GAG structures, automated synthesis is a very attrac-
tive technique to assemble (libraries of) well-defined GAG
fragments, making these important saccharides available
for biological studies.
Figure 1. LC-traces of heptamer 11, undecamer 12, and penta-
decamer 13 prior to (A, C, D) and after (B, D, F) HPLC
purification.
Global deprotection of the compounds was accom-
plished by saponification of the benzoate esters, methyl
esters, and trichloroacetamides by treatment with 0.5 M
KOH in a mixture of THF/H2O to provide the “zwitter-
ionic” hepta-, undeca-, and pentadecamers, which were
purified by gel permeation chromatography. The penta-
decamer proved to be relatively poorly soluble in water,
and the addition of a few drops of aqueous ammonia was
required to fully solubilize the compound. Finally, the
amine groups in the target compounds were acetylated to
providethe hyaluronic acid fragments14, 15, and 16. Asan
indication of the overall efficiency, the single solid-phase
run of the pentadecamer required 28 h for the assembly of
the fully protected intermediate and delivered 16 mg of the
pure, allyl-functionalized final compound, the integrity of
which was fully ascertained by 1H and 13C spectroscopy as
depicted in Figure 2. The relatively simple NMR spectra
indicate that the HA-pentadecamer takes up a very regular
structure.
Acknowledgment. This work was supported by The
Netherlands Organization of Scientific Research (NWO)
and Top Institute Pharma (TIP). We thank C. Erkelens,
F. Lefeber, and K. Babu for their assistance with the NMR
experiments.
Supporting Information Available. All experimental
procedures and 1H and 13C spectra of all new compounds.
This material is available free of charge via the Internet at
http://pubs.acs.org.
In summary, we have described the first automated
solid-phase synthesis of a set of hyaluronic acid oligomers,
representing the first GAG oligomers tobe synthesized in a
The authors declare no competing financial interest.
D
Org. Lett., Vol. XX, No. XX, XXXX