Enzymatic glycosylation of reducing oligosaccharides linked to a solid phase or a lipid via a cleavable squarate linker

Article abstract of DOI:10.1016/S0008-6215(99)00135-4

Reducing oligosaccharides were converted into their corresponding glycosylamines, and these were reacted with 3,4-diethoxy-3-cyclobuten-1,2-dione (squaric acid diethyl ester). The resulting derivatives could be linked to amino-functionalized lipids, solids, or proteins. Treatment of the obtained lipid or solid conjugates with aqueous bromine or, alternatively, with ammonia-ammonium borate cleaved the linkage and regenerated the oligosaccharide glycosylamines, which were in turn rapidly hydrolyzed to the reducing oligosaccharides. To demonstrate the usefulness of this linkage in enzymatic oligosaccharide synthesis, lactose was linked to a lipid or a solid phase, the obtained conjugates were then subjected to two enzymatic glycosylations (either consecutively or 'one-pot'). The resulting materials were then cleaved to give, in both cases, the expected reducing tetrasaccharide (lacto-N-neotetraose) in good yield. Copyright (C) 1999 Elsevier Science Ltd.

Full text of DOI:10.1016/S0008-6215(99)00135-4

www.elsevier.nl/locate/carres
Carbohydrate Research 319 (1999) 80–91
Enzymatic glycosylation of reducing oligosaccharides
linked to a solid phase or a lipid via a cleavable
squarate linker
O. Blixt, T. Norberg *
Department of Chemistry, Swedish Uni6ersity of Agricultural Sciences, PO Box 7015, S-750 07 Uppsala, Sweden
Received 5 March 1999; accepted 24 May 1999
Abstract
Reducing oligosaccharides were converted into their corresponding glycosylamines, and these were reacted with
3,4-diethoxy-3-cyclobuten-1,2-dione (squaric acid diethyl ester). The resulting derivatives could be linked to amino-
functionalized lipids, solids, or proteins. Treatment of the obtained lipid or solid conjugates with aqueous bromine
or, alternatively, with ammonia–ammonium borate cleaved the linkage and regenerated the oligosaccharide
glycosylamines, which were in turn rapidly hydrolyzed to the reducing oligosaccharides. To demonstrate the
usefulness of this linkage in enzymatic oligosaccharide synthesis, lactose was linked to a lipid or a solid phase, the
obtained conjugates were then subjected to two enzymatic glycosylations (either consecutively or ‘one-pot’). The
resulting materials were then cleaved to give, in both cases, the expected reducing tetrasaccharide (lacto-N-neote-
traose) in good yield. © 1999 Elsevier Science Ltd. All rights reserved.
Keywords: Glycosylamines; Lipid conjugates; Glycoproteins; Enzymatic synthesis
syltransferase-catalyzed glycosylations in com-
bination with solid-phase techniques offer a
particularly simple way to synthesize oligosac-
charides on a laboratory scale. We have previ-
ously reported [3] synthesis of a sialyl Lex
tetrasaccharide by carrying out three consecu-
tive, high-yield enzymatic glycosylations on a
starting monosaccharide (GlcNAc) attached
to Sepharose via a disulfide linkage. In that
case, chemical synthesis (six steps from N-
acetylglucosamine) was used to provide a Glc-
NAc derivative with a thiol functionality, and
this was then reacted with an activated thiol
on the solid phase to obtain a reversible solid-
phase disulfide linkage. In an effort to develop
a simpler procedure for reversible linkage of
oligosaccharides to a solid phase, we have
now investigated squaric acid chemistry for
the purpose. We found that reducing oligosac-
1. Introduction
Methods for the synthesis of oligosaccha-
rides have developed considerably during the
last 20 years, but a chemical oligosaccharide
synthesis is still a relatively complicated task,
involving many synthetic steps. An alternative
to chemical oligosaccharide syntheses is enzy-
matic synthesis, where nucleotide sugars are
the donors and glycosyltransferases are the
catalysts [1,2]. Glycosyltransferases give regio-
and stereospecific glycosylations, without the
complicated protecting group patterns that
have to be used in chemical synthesis. Glyco-
* Corresponding author. Tel.: +46-18-671578; fax: +46-
18-673477.
E-mail address: thomas.norberg@kemi.slu.se (T. Norberg)
0008-6215/99/$ - see front matter © 1999 Elsevier Science Ltd. All rights reserved.
PII: S0008-6215(99)00135-4

O. Blixt, T. Norberg / Carbohydrate Research 319 (1999) 80–91
81
Scheme 1.

82
O. Blixt, T. Norberg / Carbohydrate Research 319 (1999) 80–91
charides could be linked to and detached from
a solid phase using a very simple procedure,
not requiring extensive expertise in organic
synthesis. Similar linkage of oligosaccharides
to lipids or proteins was also possible. To
demonstrate the usefulness of the new linkage
type in enzymatic oligosaccharide synthesis,
lactose was linked to a lipid or a solid phase,
and the obtained conjugates were subjected to
two enzymatic glycosylations (either consecu-
tively or ‘one-pot’), the resulting materials
were cleaved to give, in both cases, the ex-
pected reducing tetrasaccharide (lacto-N-
neotetraose) in good yield.
(22, 21, and 9 mol/mol HSA, respectively)
were comparable to those obtained by existing
methods of conjugating reducing oligosaccha-
rides to proteins [6,8], and the present new
method is therefore a viable alternative for
making oligosaccharide protein conjugates
(Schemes 1–3).
Enzymatic glycosylation of glycosylamine–
lipid conjugates: solution-phase enzymatic
oligosaccharide synthesis.—The obtained new
carbohydrate–lipid conjugates had the
property of being easily separated (by C-18
silica solid-phase extraction) from complex
aqueous reaction mixtures, and were there-
fore evaluated as starting materials in enzy-
matic oligosaccharide synthesis. The lactose
glycolipid 7 was subjected to enzymatic
glycosylation, first with UDP-GlcNAc and
2. Results and discussion
Preparation of glycosylamine–lipid, glycosy-
lamine–Sepharose and glycosylamine–HSA
conjugates.—The reaction of 3,4-diethoxy-3-
cyclobuten-1,2-dione (squaric acid diethyl es-
ter) with molecules carrying the amine
functionality
was
first
reported
by
Glu¨senkamp and co-workers in 1991 [4]. We
used the originally reported reaction condi-
tions, squaric acid diethyl ester, and glycosyl-
amines 1, 2, and
3
to prepare the
oligosaccharide squaric acid conjugates 4, 5,
and 6 in 77, 83, and 84% yields, respectively
(the oligosaccharide glycosylamines were ob-
tained from the parent reducing oligosaccha-
rides lactose, 3-sialyllactose, and lacto-N-
tetraose by treatment [5,6] with ammonia/am-
monium bicarbonate). The conjugates 4, 5,
and 6 were reasonably stable and could be
isolated and stored as lyophilized powders.
However, they reacted smoothly with amines
in aqueous pH 9 buffers to form conjugates of
various types. For example, the lactose deriva-
tive 4 reacted with n-hexylamine to form the
glycolipid 7 in 90% yield, easily isolated from
the reaction mixture through solid-phase ex-
traction with C-18 silica [7]. Under similar
reaction conditions, 4 reacted with extended
linker amino-Sepharose [3] to form conjugate
14 in good yield. Finally, derivatives 4, 5, and
6 could be reacted with proteins bearing lysine
o-amino groups, such as HSA, to give neogly-
coproteins (16, 17, and 18, respectively). The
obtained neoglycoprotein substitution degrees
Scheme 2.

O. Blixt, T. Norberg / Carbohydrate Research 319 (1999) 80–91
83
Scheme 3.
b-(1“3)-N-acetylglucosaminyltransferase [9],
which gave the trisaccharide glycolipid 8 in
99% yield after C-18 silica solid-phase extrac-
tion purification. Further reaction of 8 with
UDP-Gal and b-(1“4)-galactosyltransferase
gave the tetrasaccharide glycolipid 11 in 46%
yield (from 8). The same tetrasaccharide gly-
colipid could also be obtained, in a similar
yield (66% from 7), by carrying out the enzy-
matic glycosylations in a one-pot reaction
mixture, without purification of the intermedi-
ate trisaccharide glycolipid. The enzymatic re-
actions were rapid (1–3 h), and the reaction
mixtures (according to TLC analysis) con-
tained very little by-product. The low yields of
11 (46 and 66%, respectively) were probably a
result of loss occurring during the last solid-
phase extraction procedure. Preliminary ex-
periments showed that although disaccharide
and trisaccharide glycolipids like 7 and 8 are
almost quantitatively extracted from an
aqueous buffer solution, a tetrasaccharide gly-
colipid like 11, being more polar, is not. A
longer lipid on 11 would probably have given
better solid-phase extraction yields, but, on
the other hand, that would have required a
more lipophilic starting disaccharide glycol-
ipid, which would have been less water-soluble
and more prone to form micelles (a complicat-
ing factor). For example, the benzhydrylamine
analog of 7 is only sparingly soluble in water.
Therefore, in any case where solid-phase ex-
tractions are used to purify glycolipid-type
products from aqueous enzymatic glycosyla-
tions mixtures, the choice of the lipid part has
to be made carefully, considering solubility
properties of both starting and product
glycolipid.
The trisaccharide glycolipid 8 was cleaved
by treatment with aqueous bromine, giving
first the trisaccharide glycosylamine 9. If the
reaction mixture was pH-adjusted to 5–6,
complete hydrolysis to the reducing trisacchar-
ide 10 was observed within a few hours (as
expected from what is known about the pH
dependence of glycosylamine hydrolysis rates
[10,11]). We found, however, that the glycosy-
lamine could be more conveniently hy-
drolyzed, and then even at above neutral pH,
by addition of sodium borate to the reaction
mixture. The scope and generality of this new
borate-catalyzed hydrolysis is currently under
further investigation. The obtained trisaccha-
ride 10 was purified by gel-filtration; the yield
was 91%.
In an analogous way, tetrasaccharide gly-
colipid 11 was cleaved, to give first the glyco-
sylamine 12, then, without isolation, the
tetrasaccharide 13 (lacto-N-neotetraose, 85%
yield).
An alternative method for cleavage of the
oligosaccharide–lipid conjugates was also in-
vestigated. It was found that treatment of the
conjugates at room temperature with concen-
trated aqueous ammonia containing ammo-
nium borate resulted in cleavage, giving first
the glycosylamines, then the parent free sugars
in good yield (reaction with only aqueous

84
O. Blixt, T. Norberg / Carbohydrate Research 319 (1999) 80–91
1
ample, the H spectrum of 7 at 27 °C (Fig.
1(a)) showed a broad signal in the anomeric
region that was assigned to H-1. The other
glucose proton and carbon signals were also,
to various degrees, broadened. The phe-
nomenon was ascribed to the presence of two
CꢀN rotamers of 4, interconverting at an in-
termediate rate (on the NMR time scale). As
ammonia gave a much slower cleavage rate).
For example, treatment of 7 with aqueous
ammonia containing ammonium borate gave
lactose in 88% yield.
The NMR spectra of the squaric-acid-
derived oligosaccharides all showed the char-
acteristics of compounds with restricted
rotation around a CꢀN bond [4,12]. For ex-
1
Fig. 1. 400 MHz H NMR spectra of compound 7 in D2O at (a) 27 °C; (b) 40 °C; (c) 60 °C; and (d) 80 °C.

O. Blixt, T. Norberg / Carbohydrate Research 319 (1999) 80–91
85
Fig. 1. (Continued)
expected in such a situation, heating of 7 to
tion steps were carried out in this case). As
before [3,13], in solid-phase enzymatic glyco-
sylations, longer reaction times (18 h as com-
pared to 3 h), slightly larger excesses of
nucleotide sugar (2.0 equivalents as compared
to 1.3–1.5 equivalents) and larger excesses of
enzyme (80–100 mU/mmol substrate as com-
pared to 17–30 mU) were used to drive the
reactions to completion, as compared to the
solution case with the glycolipids. This can be
seen as a disadvantage if the availability of
enzymes is limiting. However, on the impor-
tant advantage side, filtration purification af-
ter each reaction step was much faster, simpler
and more quantitative than the corresponding
1
80 °C gave a H spectrum (Fig. 1(d)) with
sharp signals for all glucose nuclei.
Enzymatic glycosylation of glycosylamine–
Sepharose conjugates: solid-phase enzymatic
oligosaccharide synthesis.—The lactose–Sep-
harose conjugate 14, having an extended
linker, was investigated in enzymatic synthesis
(the reported procedure [3] for preparation
and quantification of 14 was considerably im-
proved, and these details are therefore given in
Section 3). Lactose–Sepharose conjugate 14
was treated with the same two enzymes under
similar reaction conditions as with the glycol-
ipids described above (two separate glycosyla-

86
O. Blixt, T. Norberg / Carbohydrate Research 319 (1999) 80–91
glycolipid solid-phase extraction, and this is
an advantage that becomes increasingly im-
portant with increasing number of steps in the
synthesis. The obtained Sepharose-bound te-
trasaccharide derivative 15 was cleaved,
analogous to the above with the glycolipids,
by treatment with aqueous bromine, followed
by neutralization and addition of sodium bo-
rate to complete the hydrolysis of the formed
glycosylamine to the reducing oligosaccharide.
The yield of lacto-N-neotetraose 13 (after gel-
filtration purification to remove salts and
traces of shorter oligosaccharides) was 89%
(calculated from 14). The alternative cleavage
method using concentrated aqueous ammo-
nia/ammonium borate was also demonstrated
on the same material 15, giving a yield of
lacto-N-neotetraose 13 (after gel-filtration
purification to remove salts and traces of
shorter oligosaccharides) of 75% (calculated
from 14).
propriate eluants, spots were visualized by UV
light and/or by dipping in 5% sulfuric acid,
followed by charring. For solid-phase extrac-
tions, reversed-phase C-18 Silica Gel (Interna-
tional Sorbent Technology Ltd, UK) or
Isolute cartridges (International Sorbent Tech-
nology) were used. Water for all solutions was
from a MilliQ water purification system (Mil-
lipore Corp., Bedford, Mass., USA), and was
degassed by vacuum treatment before use.
Gel-filtrations were performed on Sephadex
columns using MilliQ-purified water with
added n-butanol (5%, to prevent microorgan-
ism contamination). The solid-phase enzy-
matic reactions were carried out in small,
silanized (by brief treatment with 5%
dichlorodimethylsilane in hexane) columns
(5–15 mL) with a fritted glass filter and a
Teflon stopcock at the bottom end. After
charging with the appropriate reagents, the
column was sealed at the top with a ground
glass stopper, the Teflon stopcock was closed,
and the column was slowly rotated (by attach-
ment to a rotavapor glass tube) at room tem-
perature (rt) for the specified time. The
quantitative ninhydrin determinations of
amino groups on Sepharose were carried out
essentially as described [14]. Briefly, a soln of
ninhydrin in Me2SO (0.5 M, 50 mL) and an aq
sodium acetate soln (0.5 M, pH 5.2, 50 mL)
were added to a weighed amount of drained
gel (40–50 mg, approx. 50 mL, resuspended in
50 mL of water), then the mixture was heated
to 100 °C for 10 min. The resulting mixture
was diluted to 50 mL with 1:1 EtOH–water,
and the absorbance of the supernatant liquid
was then measured at 570 nm. The original
Sepharose amino group content was calcu-
lated from this adsorbance value and the ex-
tinction coefficient (m570=8750) of the formed
purple dye (diketohydrindylidenediketohyd-
rindamine).
In conclusion, using squaric acid chemistry
for reversible coupling of oligosaccharides to
lipids or solids, simple protocols were devel-
oped, which should be highly suitable for use
in enzymatic oligosaccharide synthesis with
reducing oligosaccharides as starting mater-
ials.
3. Experimental
General methods.—Concentrations were
performed under reduced pressure at B40 °C
bath temperature. NMR spectra were
recorded, if not otherwise indicated, at 30 or
80 °C using Bruker 400 and 600 MHz instru-
ments (data recorded at 80 °C are marked
with an asterisk). The following reference sig-
nals were used: acetone l 2.225 (1H in D2O),
acetone l 30.7 (13C in D2O). FABMS were
recorded in the positive-ion mode with a Jeol
JMS-SX/SX-102A mass spectrometer using
glycerol as matrix. MALDI-TOF MS of
protein conjugates were recorded in the posi-
tive-ion mode with a LDI-1700-XP instru-
ment, using sinapic acid (3,5-dimethoxy-4-
hydroxycinnamic acid) as matrix. Thin-layer
chromatography (TLC) was performed on
Kieselgel 60 F254 Fertigplatten (E. Merck,
Darmstadt, Germany). After elution with ap-
Uridine 5%-diphosphogalactose (UDP-Gal),
uridine
5%-diphospho-N-acetylglucosamine
(UDP-GlcNAc), bovine milk b-(1“4)-galac-
tosyltransferase, BSA, HSA, and 1,2-O-iso-
propylidene-a- -glucofuranose were from
D
Sigma Chemical Co. (St. Louis, IL, USA),
3,4-diethoxy-3-cyclobuten-1,2-dione (squaric
acid diethyl ester) and 1,8-diamino-3,6-
dioxaoctane were from Acros Chimica (Geel,

O. Blixt, T. Norberg / Carbohydrate Research 319 (1999) 80–91
87
H, H-4), 3.93 (d, 1 H, H-4%), 4.47 (d, 1 H, J1,2
7.8 Hz, H-1%), 4.77 (m, 2 H, CH3ꢀCH2ꢀOꢀ),
4.90, 5.16 (2 broad s, 1 H, H-1), 4.97* (J1,2
8.8* Hz, H-1*); 13C, l 15.5 (CH3ꢀCH2ꢀOꢀ),
69.0 (C-4%), 71.5, (C-2%), 71.5 (CH3ꢀCH2ꢀOꢀ),
72.7 (C-2), 73.0 (C-3%), 75.8 (C-3), 76.8 (C-4),
83.9 (broad s, C-1), 103.4 (C-1%). FABMS:
Anal. Calcd for C18H28O13N [M+H]:
466.1561. Found: 466.1559.
Belgium), 3%-sialyllactose, lacto-N-tetraose,
and lacto-N-neotetraose were from IsoSep AB
(Tullinge, Sweden) and recombinant Neisseria
meningitidis b-(1“3)-N-acetylglucosaminyl-
transferase was obtained as previously de-
scribed [9]. One unit of this enzyme is defined
[9] as the amount required to glycosylate (with
UDP-GlcNAc) 1 mmol of lactose in 1 min.
Epoxy Sepharose 6B was from Amersham
Pharmacia Biotech AB (Uppsala, Sweden).
General procedure for preparation of
oligosaccharide–squaric acid conjugates.—A
soln of reducing oligosaccharide (0.2 M) and
ammonium bicarbonate (0.2 M) in concd aq
ammonia was stirred and heated to 35 °C for
24 h. The mixture was then concentrated and
co-concentrated twice with water, the residue
was dissolved in a 1:1 sodium bicarbonate–
EtOH (0.1 M, pH 9) mixture to a concn of
0.13 M. 3,4-Diethoxy-3-cyclobuten-1,2-dione
was added to a concn of 0.4 M, and the pH
was kept at 9–10 by dropwise addition of aq
NaOH. After 1 h at rt, the mixture was con-
centrated and purified by gel-filtration on a
Sephadex G-15 column. Fractions containing
(the UV-absorbing) product were pooled and
lyophilized.
Lactose–squaric acid conjugate 4.—This
derivative was prepared essentially as de-
scribed in the previous paragraph. Thus, lac-
tose (1.60 g, 4.68 mmol) was converted to the
glycosylamine 1 using ammonium bicarbonate
(0.5 g, 5.8 mmol) and concd aq ammonia (30
mL). An NMR analysis of the crude glycosy-
lamine intermediate revealed signals at 4.12/
85.3 ppm (H-1/C-1) from the glycosylamine 1.
From the 1H NMR spectrum, it was estimated
that 80% of the original amount of lactose
had been converted into 1. The crude, concd
glycosylamine was reacted with 3,4-diethoxy-
3-cyclobuten-1,2-dione (1.77 mL, 12 mmol) in
a mixture of aq sodium bicarbonate (0.1 M,
pH 9, 10 mL) and EtOH (10 mL). The pH
was adjusted to 9–10 with 1 M aq NaOH
soln. After 3 h, the mixture was concentrated
and purified by gel-filtration (Sephadex G-15,
5×100 cm column). Appropriate fractions
were pooled and lyophilized giving 4 (1.68 g,
77%). NMR data: 1H, l 1.45 (t, 2 H,
CH3ꢀCH2ꢀOꢀ) 3.54 (t, 1 H, H-2), 3.56 (t, 2 H,
H-2%), 3.67 (dd, H-3%), 3.73 (1 H, H-3), 3.78 (1
Sialyllactose–squaric acid conjugate 5.—3-
Sialyllactose (100 mg) was converted, essen-
tially as previously described, into 5 (100 mg,
83%) using first concd ammonia/ammonium
bicarbonate (0.8 mL/13.5 mg), then 3,4-di-
ethoxy-3-cyclobuten-1,2-dione (0.11 mL) in
1:1 EtOH–aq sodium bicarbonate (4 mL).
NMR data: 1H, l 1.45 (t, 2 H, CH3ꢀCH2ꢀOꢀ),
1.81, 2.88 (2 H, H-3%%), 2.03 (NAc), 3.54 (1 H,
H-2), 3.59 (t, 2 H, H-2%), 3.60 (1 H, H-7%%),
3.64 (1 H, H-9%%), 3.66 (1 H, H-6%%), 3.71 (1 H,
H-4%%), 3.74 (1 H, H-3), 3.78 (1 H, H-4), 3.85
(1 H, H-5%%), 3.91 (1 H, H-8%%), 3.96 (d, 1 H,
H-4%), 4.12 (dd, H-3%), 4.56 (d, 1 H, J1,2 7.8 Hz,
H-1%), 4.77 (m, 2 H, CH3ꢀCH2ꢀO), 4.91, 5.16
(2 broad s, H, H-1); 13C,
l
15.5
(CH3ꢀCH2ꢀOꢀ), 22.5 (NAc), 40.1 (C-3%%), 52.1
(C-5%%), 63.0 (C-9%%), 68.0 (C-4%), 68.5 (C-4%%),
68.7 (C-7%%), 69.8 (C-2%), 71.5 (CH3ꢀCH2ꢀOꢀ),
72.1 (C-8%%), 72.7 (C-2), 73.3 (C-6%%), 75.6 (C-3),
75.9 (C-3%), 76.9 (C-4), 83.9 (broad s, C-1),
103.4 (C-1%). FABMS: Anal. Calcd for
C29H43O21N2 (M+H): 755.2358. Found:
755.2390.
Lacto-N-tetraose–squaric acid conjugate
6.—Lacto-N-tetraose (150 mg) was con-
verted, essentially as previously described, into
6 (155 mg, 84%) using first concd ammonia/
ammonium bicarbonate (1.0 mL/17 mg), then
3,4-diethoxy-3-cyclobuten-1,2-dione (0.13 mL)
in 1:1 EtOH–aq sodium bicarbonate (5 mL).
1
NMR data: H, l 1.45 (t, 2 H, CH3ꢀCH2ꢀO),
4.44 (d, 1 H, J1,2 7.8 Hz, H-1%%%), 4.47 (d, 1 H,
J1,2 7.8 Hz, H-1%), 4.74 (d, 1 H, J1,2 8.4 Hz,
H-1%%), 4.77 (m, 2 H, CH3ꢀCH2ꢀOꢀ), 4.91, 5.16
(2 broad s, 1 H, H-1); 13C, l 15.5
(CH3ꢀCH2ꢀOꢀ), 71.5 (CH3ꢀCH2ꢀOꢀ), 83.9
(broad s, C-1), 103.0 (C-1%%%), 103.4 (C-1%),
103.9 (C-1%%). FABMS: Anal. Calcd for
C32H51O23N2 [M+H]: 831.2882. Found:
831.2910.

88
O. Blixt, T. Norberg / Carbohydrate Research 319 (1999) 80–91
Lactose–hexylamine conjugate 7.—A soln
1.30, 1.35, 1.62, 3.62 (CH3ꢀCH2ꢀCH2ꢀCH2ꢀ
CH3ꢀCH2ꢀNHꢀ), 2.04 (s, 3 H, NAc), 3.48 (1
H, H-2), 3.59 (t, 1 H, H-2%), 3.71 (dd, H-3%),
3.73 (1 H, H-3), 3.74 (1 H, H-2%%), 3.77 (1 H,
H-4), 3.88 (1 H, H-4%%), 3.90 (1 H, H-3%%), 4.15
(d, 1 H, H-4%), 4.45 (d, 1 H, J1,2 7.8 Hz, H-1%),
4.68 (1 H, J1,2 8.4 Hz, H-1%%), 5.08 (broad s, 1
H, H-1); 13C, d 13.6, 22.3, 31.1, 25.6, 30.5,
44.9 (CH3ꢀCH2ꢀCH2ꢀCH2ꢀCH3ꢀCH2ꢀNHꢀ),
22.7 (NAc), 59.8 (C-4%%), 60.6 (C-3%%), 61.1 (C-
2%%), 68.6 (C-4%), 70.2 (C-2%), 72.7 (C-2), 74.8
(C-3), 75.1 (C-3%), 76.4 (C-4), 83.5 (C-1), 103.0
(C-1%%), 103.1 (C-1%). FABMS: Anal. Calcd for
C30H50O17N3 [M+H]: 724.3140. Found:
724.3178.
of lactose squaric acid conjugate 4 (0.80 g,
1.72 mmol) in a mixture of aq sodium bicar-
bonate (5.0 mL, 0.1 M, pH 9.0), EtOH (5.0
mL) and 2-propanol (1.0 ml) was stirred while
n-hexylamine (0.5 mL, 3.8 mmol) was added.
The mixture was further stirred at rt for 12 h,
then evaporated to dryness and resuspended
in water (50 mL). The soln was filtered, and
the filtrate was slowly passed through a C-18
silica gel column (100 mL, previously washed
with first 300 mL of MeOH, then 2000 mL of
water). After washing of the column with
water (200 mL), the desired material was
eluted with MeOH (200 mL). Appropriate
fractions were pooled, concd and lyophilized
to give compound 7 (0.80 g, 1.55 mmol, 90%).
Enzymatic galactosylation of 8 to gi6e 11.—
Conjugate 8 (10 mg, 13.8 mmol) was added to
a 4 mL aq buffer (20 mM in mangane-
se(II)chloride and 0.5 M in sodium cacody-
late) soln containing UDP-Gal (11 mg, 18.0
mmol), and bovine milk b-(1“4)-galactosyl-
transferase (500 mU, 0.05 mL of a 50% glyc-
erol stock soln). The reaction mixture was
stirred at rt for 3 h, after which TLC ana-
lysis (6:3:3:2 EtOAc–AcOH–MeOH–water)
showed complete consumption of starting ma-
terial. The mixture was diluted with water to
20 mL, and was then slowly passed through a
C-18 Isolute cartridge (10 g, previously
washed with first 50 mL of MeOH, then 200
mL of water). After washing of the column
with water (50 mL), the desired material was
eluted with MeOH (20 mL). Appropriate frac-
tions were pooled and lyophilized to give com-
pound 11 (5.7 mg, 6.4 mmol, 46%). NMR
1
NMR data: H, l 0.86, 1.29, 1.30, 1.35, 1.62,
3.62 (CH3ꢀCH2ꢀCH2ꢀCH2ꢀCH3ꢀCH2ꢀNHꢀ),
3.50 (1 H, H-2), 3.56 (t, 1 H, H-2%), 3.67 (dd,
H-3%), 3.74 (1 H, H-3), 3.78 (1 H, H-4), 3.93
(d, 1 H, H-4%), 4.47 (d, 1 H, J1,2 7.8 Hz, H-1%),
5.08 (broad s, 1 H, H-1), 5.06* (J1,2 8.8* Hz,
H-1*); 13C, l 13.6, 22.3, 31.1, 25.6, 30.5, 44.9
(CH3ꢀCH2ꢀCH2ꢀCH2ꢀCH3ꢀCH2ꢀNH), 69.2
(C-4%), 71.6 (C-2%), 73.1 (C-2), 73.2 (C-3%), 75.2
(C-3), 76.1 (C-4), 84.1 (C-1), 103.4 (C-1%).
FABMS: Anal. Calcd for C22H37O12N2 (M+
H): 521.2346. Found: 521.2337.
Enzymatic solution glycosylation of 7 with
i-(1“3)-N-acetylglucosaminyltransferase to
gi6e 8.—Conjugate 7 (20 mg, 38.5 mmol) was
added to a 4 mL aq buffer (15 mM in man-
ganese(II)chloride and 0.5 M in sodium ca-
codylate) soln containing UDP-GlcNAc (38
mg, 58.4 mmol), BSA (20 mg), and recombi-
1
data: H, l 0.86, 1.29, 1.30, 1.35, 1.62, 3.62
nant
b-(1“3)-N-acetylglucosaminyltrans-
(CH3ꢀCH2ꢀCH2ꢀCH2ꢀCH3ꢀCH2ꢀNHꢀ), 2.04
(s, 3 H, NAc), 4.46 (d, 1 H, J1,2 7.8 Hz, H-1%),
4.48 (1 H, H-1%%%), 4.72 (1 H, J1,2 8.4 Hz, H-1%%),
5.08 (broad s, 1 H, H-1); 13C, l 13.6, 22.3,
31.1, 25.6, 30.5, 44.9 (CH3ꢀCH2ꢀCH2ꢀCH2ꢀ
CH3ꢀCH2ꢀNHꢀ), 22.7 (NAc), 83.7 (C-1),
103.0 (C-1%%%), 103.2 (C-1%, C-1%%). FABMS:
Anal. Calcd for C36H60O22N3 [M+H]:
886.3668. Found: 886.3669.
ferase (20 mU, 0.2 mL of a 50% glycerol stock
soln). The reaction mixture was stirred at rt
for 5 h, after which TLC analysis (6:3:3:2
EtOAc–AcOH–MeOH–water) showed com-
plete consumption of starting material. The
mixture was diluted with water to 20 mL, and
was then slowly passed through a C-18 Isolute
cartridge (10 g, previously washed with first 50
mL of MeOH, then 200 mL of water). After
washing of the column with water (50 mL),
the desired material was eluted with MeOH
(20 mL). Appropriate fractions were pooled
and lyophilized to give compound 8 (27 mg,
One-pot enzymatic glycosylation of 7 to gi6e
11.—Conjugate 7 (15 mg, 29.0 mmol) was
added to a 4 mL aq buffer (15 mM in man-
ganese(II)chloride and 0.5 M in sodium ca-
codylate, pH 7.2) soln containing UDP-Gal
(23 mg, 37.7 mmol), UDP-GlcNAc (25.0 mg,
1
38 mmol, 99%). NMR data: H, l 0.86, 1.29,

O. Blixt, T. Norberg / Carbohydrate Research 319 (1999) 80–91
89
38 mmol), BSA (20 mg), b-(1-4)-galactosyl-
hadex G-15, 1.8×80 cm). Appropriate frac-
tions were pooled and lyophilized, giving
trisaccharide 10 (6.6 mg, 0.0125 mmol, 91%).
transferase (500 mU, 0.05 mL of a 50% glyc-
erol
stock
soln),
and
recombinant
(60
1
NMR data: H, l 2.04 (s, 3 H, NAc), 3.28 (t,
b-(1-3)-N-acetylglycosamintransferase
1 H, H-2b), 3.58 (t, 1 H, H-2a), 3.64 (t, 1 H,
H-3b), 3.65 (t, 1 H, H-4), 3.73 (t, 1 H, H-3a),
3.73 (dd, 1 H, H-3%), 3.59 (t, 1 H, H-2%), 3.76
(1 H, H-2%%), 3.84 (1 H, H-4%%), 3.90 (1 H, H-3%%),
4.15 (d, 1 H, H-4%), 4.44 (d, 1 H, J1,2 7.8 Hz,
H-1%), 4.66 (1 H, J1,2 7.8 Hz, H-1b) 4.70 (1 H,
H-1, J1,2 8.1 Hz, H-1%%), 5.22 (1 H, J1,2 3.8 Hz,
H-1a); 13C, l 22.7 (NAc), 56.0 (C-2%%), 60.6
(C-4%%), 61.0 (C-3%%), 68.8 (C-4%), 70.2 (C-4), 70.5
(C-2%), 70.9 (C-2a), 74.3 (C-2b), 74.9 (C-3b),
75.3 (C-3a), 75.4 (C-3%), 92.2 (C-1a), 96.1 (C-
1b), 103.2 (C-1%%), 103.3 (C-1%). FABMS: Anal.
Calcd for C20H36O16N [M+H]: 546.2034.
Found: 546.2029.
Clea6age of compound 11 (method A) to gi6e
lacto-N-neotetraose (13).—Compound 11 (5.0
mg, 5.8 mmol) was cleaved with bromine as
described (method A) for compound 8 to give,
after purification by gel-filtration, material (4.0
mg, 5.7 mmol, 85%) whose NMR data were
identical to those of an authentic sample [15]
of lacto-N-neotetraose (13).
Clea6age of 7 (method B) to gi6e lactose.—
The lactose–lipid conjugate 7 (21 mg, 0.040
mmol) was cleaved with ammonia/ammonium
borate as described (method B) to give, after
purification by gel-filtration, a material (12
mg, 88%) whose NMR data were identical to
those of an authentic sample of lactose.
Simplified preparation of long-chain amino
Sepharose 6B
Initial amino deri6atization of Sepharose 6B.
Epoxy Sepharose 6B (5 g) was swollen in
water, washed (water, 100 mL) and mixed (in
drained form) with a soln of 1,8-diamino-3,6-
dioxaoctane (1.0 g, 6.8 mmol) in aq sodium
bicarbonate buffer (0.1 M, pH 9.0, 30 mL).
The mixture was shaken for 12 h at rt, then
filtered and washed with water (500 mL total
volume). The first filtrate was saved and used
later. An aliquot of the gel (50 mL drained) was
removed and analyzed with a quantitative nin-
hydrin test (see above). An amino group con-
tent of 55 mmol/mL drained gel was found.
First elongation cycle. The bulk of the gel (in
drained form) was mixed with a freshly pre-
pared soln of 3,4-diethoxy-3-cyclobutene-1,2-
mU, 0.3 mL of a 50% glycerol stock soln). The
reaction mixture was stirred at rt for 3 h, after
which TLC analysis (6:3:3:2 EtOAc–AcOH–
MeOH–water) showed complete consumption
of starting material. The mixture was diluted
with water to 20 mL, and was then slowly
passed through a C-18 Isolute cartridge (10 g,
previously washed with first 50 mL of MeOH,
then 200 mL of water). After washing of the
column with water (50 mL), the desired mate-
rial was eluted with MeOH (20 mL). Appro-
priate fractions were pooled and lyophilized to
give compound 11 (17 mg, 19 mmol, 66%
yield). NMR data were as previously shown.
General procedure for clea6age of oligosac-
charide–lipid conjugates to gi6e reducing sugars
Method A (using bromine). A freshly pre-
pared soln of bromine in water (100 mM, 1.0
ml) was added, at rt, to an aq soln of oligosac-
charide–lipid conjugate (40 mM, 1.0 mL). The
orange bromine color decreased rapidly on
mixing. If necessary, more bromine soln was
added until a slight yellow color persisted. The
reaction mixture was then made slightly basic
(pH 7–8) by the addition of an aq 1 M
ammonium bicarbonate soln. Aq sodium bo-
rate (0.2 M, 1.0 mL, pH 8.0) was then added,
and the mixture was stirred at rt until TLC
indicated complete conversion of the glycosy-
lamine to the reducing sugar (1 h). The result-
ing mixture was purified by gel-filtration.
Method B (using ammonia/ammonium bo-
rate). Oligosaccharide–lipid conjugate (0.040
mmol) was dissolved in a mixture of aq boric
acid (2.0 mL, 0.2 M) and concd aq ammonia
(2.0 mL). The mixture was stirred at 30 °C for
3 h, then the mixture was filtered, and the
filtrate was concentrated and purified by gel-
filtration.
Clea6age of compound 8 (method A) to gi6e
reducing trisaccharide 10.—A soln of com-
pound 8 (10 mg, 0.038 mmol) in water (1.0
mL) was treated, essentially as described above
(method A), with bromine (1 mL of a 100 mM
aq soln). Adjustment of pH and treatment
with sodium borate as described above gave a
mixture that was purified by gel-filtration (Sep-

90
O. Blixt, T. Norberg / Carbohydrate Research 319 (1999) 80–91
amino group content of 15 mmol/mL was now
found, which, assuming a clean derivatization
reaction, corresponds to a lactose incorpora-
tion of 15 mmol/mL. This value was verified
by cleaving off the bound lactose (ammonia/
ammonium borate method, for general condi-
tions, see below) and analyzing the contents of
the supernatant by quantitative 1H NMR
dione (1.0 g, 5.9 mmol) in EtOH (15 mL) and
bicarbonate buffer (0.1 M, pH 9.0, 15 mL).
The mixture was shaken for 2 h at rt, then
filtered and washed with 1:1 EtOH–water (50
mL total) and water (200 mL). An aliquot of
the gel (50 mL drained) was removed and
analyzed with a quantitative ninhydrin test
(see above). A negligible amino group content
was found at this stage. The bulk of the gel (in
drained form) was mixed with the saved 1,8-
diamino-3,6-dioxaoctane filtrate from above.
The mixture was shaken for 12 h at rt, then
filtered and washed with water (500 mL total).
The first filtrate was saved and used later.
Second elongation cycle. The elongation cy-
cle described above was repeated once more,
giving a final long-chain amino Sepharose (45
chain atoms) that, according to the ninhydrin
test, had an amino group content of 36 mmol/
mL.
analysis (1,2-O-isopropylidene-a- -glucofura-
D
nose was added as internal standard). This
gave a value of 13 mmol/mL for the lactose
incorporation.
Enzymatic glycosylation of lactose–Seph-
arose deri6ati6e 14 to gi6e 15.—The Sepharose
derivative 14 from above (0.9 mL, 12 mmol)
was resuspended in sodium cacodylate buffer
soln (2.0 mL, 0.5 M, pH 7.2) containing man-
ganese(II)chloride (15 mM), BSA (10 mg),
UDP-N-acetylglucosamine (15 mg, 23 mmol),
and
recombinant
b-(1“3)-N-acetylgly-
Coupling of lactose–squaric acid conjugate 4
to long-chain amino Sepharose to gi6e lactose–
Sepharose deri6ati6e 14.—A soln of lactose
conjugate 4 (11 mg, 23 mmol) in bicarbonate
buffer (2.0 mL, 0.1 M, pH 9.0) was rotated for
12 h at rt with long-chain amino Separose
prepared as described above (1.0 mL, 36 mmol
amino groups). The solid was filtered off and
washed with water (30 mL). An aliquot of the
gel (50 mL drained) was removed and analyzed
with a quantitative ninhydrin test (see above).
An amino group content of 23 mmol/mL was
now found, which, assuming a clean deriva-
tization reaction, corresponds to a lactose in-
corporation of 13 mmol/mL. This value was
verified by cleaving off the bound lactose
(bromine method, for general conditions, see
below) and analyzing the contents of the su-
cosamintransferase (100 mU, 0.5 mL of a 50%
glycerol stock soln). The reaction mixture was
rotated for 18 h at rt, filtered, and washed
with water (20 mL). The solid was resus-
pended in sodium cacodylate buffer soln (2.0
mL, 0.5 M, pH 7.2) containing mangane-
se(II)chloride (20 mM), UDP-galactose (15
mg, 25 mmol), and b-(1“4)-galactosyltrans-
ferase (1.0 U, 0.05 mL of a 50% glycerol stock
soln). The reaction mixture was rotated for 18
h at rt, then filtered and washed with water
(50 mL). The resulting gel 15 was used as such
in the next step.
Clea6age of gel 15 (with bromine) to gi6e
lacto-N-neotetraose.—Gel 15 (0.9 mL) was
resuspended in water (4.0 mL) containing
bromine (15 mL, 0.29 mmol), and the mixture
was rotated for 10 min at rt. The gel was then
filtered and washed with water (10 mL). The
combined filtrate and washings was adjusted
to pH 7.0–8.0 with aq ammonium bicarbon-
ate (1 M), then aq sodium borate (0.5 mL, 0.2
M, pH 8.0) was added. The mixture was
stirred at rt for 1 h. Analysis by TLC (4:3:3:2
EtOAc–AcOH–MeOH–water) of the reac-
tion mixture at this stage showed the presence
of a major compound (lacto-N-neotetraose)
and trace amounts of the corresponding di-
and trisaccharide. The mixture was concen-
trated and purified by gel-filtration (Sephadex
G-15, 1.8×80 cm column). Appropriate frac-
1
pernatant by quantitative H NMR analysis
(1,2-O-isopropylidene-a- -glucofuranose was
D
added as internal standard). This gave a value
of 12 mmol/mL for the lactose incorporation.
In another experiment, lactose conjugate 4
(15 mg, 32 mmol) was mixed with long-chain
amino Separose (1.0 mL, 30 mmol amino
groups, a different batch than above) in bicar-
bonate buffer (2.0 mL, 0.1 M, pH 9.0). After
2 h, the solid was filtered off and washed with
water (30 mL). An aliquot of the gel (50 mL
drained) was removed and analyzed with a
quantitative ninhydrin test (see above), An

O. Blixt, T. Norberg / Carbohydrate Research 319 (1999) 80–91
91
tions were pooled, mixed with 1,2-O-isopropy-
Acknowledgements
lidene-a- -glucofuranose (1.1 mg, 5.0 mmol, in-
D
ternal standard) and lyophilized. The residue,
according to 1H NMR analysis, contained pure
lacto-N-neotetraose (13, 12 mmol, 92%).
The authors thank the Swedish Natural Sci-
ence Research Council and the Swedish Re-
search Council for Engineering Sciences for
financial support. They also thank Amersham
Pharmacia Biotech AB (Uppsala, Sweden) for
supplying the Sepharose material used in this
work.
Clea6age of gel 15 (with ammonia/ammonium
borate) to gi6e lacto-N-neotetraose.—Gel 15
(0.80 mL, 11 mmol) was resuspended in aq
boric acid (0.2 M, 1.0 mL) and concd ammonia
(1.0 mL) was added. The mixture was rotated
overnight at rt. The gel was filtered and washed
with water (10 mL). The combined filtrate and
washings was concentrated and purified by gel-
filtration (Sephadex G-15, 1.8×80 cm
column). Appropriate fractions were pooled,
the obtained soln was mixed with 1,2-O-iso-
References
[1] C.H. Wong, R.L. Halcomb, Y. Ichikawa, T. Kajimoto,
Angew. Chem., Int. Ed. Engl., 34 (1995) 521–546.
[2] C.H. Wong, R.L. Halcomb, Y. Ichikawa, T. Kajimoto,
Angew. Chem., Int. Ed. Engl., 34 (1995) 412–432.
[3] O. Blixt, T. Norberg, J. Org. Chem., 63 (1998) 2705–
2710.
[4] L. Tietze, M. Arlt, M. Beller, K.H. Glusenkamp, E.
Jahde, M.F. Rajewsky, Chem. Ber., 124 (1991) 1215.
[5] A. Lubineau, J. Auge, B. Drouillat, Carbohydr. Res., 266
(1995) 211–219.
propylidene-a- -glucofuranose (0.44 mg, 2.0
D
mmol, internal standard) and lyophilized. The
residue, according to NMR analysis, contained
lacto-N-neotetraose (8.2 mmol, 75%).
Coupling of 4, 5 or 6 to HSA to gi6e oligosac-
charide–protein conjugates 14, 15, and 16.—
HSA (50 mg, 0.75 mmol, 1/33 equivalent) was
dissolved in carbonate buffer (7.0 mL, pH 9.0).
When all protein had dissolved, oligosaccha-
ride derivative (25 mmol, i.e., 11.7 mg of 4, 18.9
mg of 5, or 20.8 mg of 6) was added, and the
soln was slowly stirred at rt for 18 h. The reac-
tion mixture was ultrafiltrated down to 2 mL,
water (10 mL) was added, and ultrafiltration
was repeated. After four ultrafiltration cycles,
the retentate was lyophilized to give the glyco-
conjugates: 14 (52 mg), 15 (48 mg), and 16 (53
mg). These were analyzed by MALDI-TOF
mass spectroscopy. The substitution degrees
were 22, 21, and 9 mol/mol of HSA, respec-
tively, corresponding to coupling yields of 70,
68 and 25%, respectively.
[6] E. Kallin, H. Lo¨nn, T. Norberg, Glycoconjugate J., 3
(1986) 311–319.
[7] M.M. Palcic, L.D. Heerze, M. Pierce, O. Hindsgaul,
Glycoconjugate J., 5 (1988) 49.
[8] G. Magnusson, A.Y. Chernyak, J. Kihlberg, L.O.
Kononov, in Y.C. Lee, R.T. Lee (Eds.), Neoglycoconju-
gates: Preparation and Applications, Academic Press, San
Diego, 1994, pp. 53–143.
[9] O. Blixt, I. van Die, T. Norberg, D.H. van den Eijnden,
Glycobiology, (1999), in press.
[10] E. Kallin, H. Lo¨nn, T. Norberg, M. Elofsson, J. Carbo-
hydr. Chem., 8 (1989) 597–611.
[11] H. Paulsen, K.W. Pflughaupt, in W. Pigman, D. Horton
(Eds.), The Carbohydrates, Vol. 1B, Academic Press,
New York, 1980, pp. 881–927.
[12] L. Kenne, P. Unger, T. Wehler, J. Chem. Soc. Perkin
Trans. 1, (1988) 1183–6.
[13] O. Blixt, T. Norberg, J. Carbohydr. Chem., 16 (2) (1997)
143–154.
[14] J. Greenstein, M. Winitz, Chemistry of the Amino Acids,
Vol. 2, Wiley, New York, 1961, pp. 1301–1312.
[15] V.K. Dua, C.A. Bush, Anal. Biochem., 133 (1983) 1–8.
.