Mild tagging procedures for the structural analysis of glycans

Article abstract of DOI:10.1016/S0008-6215(01)00115-X

The reductive oxyamination of model glycan structures has been investigated as a mild, alternative tagging procedure to reductive amination using O-(4-nitrobenzyl)-hydroxylamine. Oxime formation was quantitative, but the reduction step did not always go t

Full text of DOI:10.1016/S0008-6215(01)00115-X

www.elsevier.nl/locate/carres
Carbohydrate Research 333 (2001) 59–71
Mild tagging procedures for the structural analysis of
glycans
Steven L. Ramsay,a Craig Freeman,b Philip B. Grace,a John W. Redmond,c
John K. MacLeoda,*
aResearch School of Chemistry, Australian National Uni6ersity, Canberra, ACT 0200, Australia
bThe John Curtin School of Medical Research, Australian National Uni6ersity, Canberra, ACT 0200, Australia
cResearch School of Biological Sciences, Australian National Uni6ersity, Canberra, ACT 0200, Australia
Received 22 November 2000; accepted 18 April 2001
Abstract
The reductive oxyamination of model glycan structures has been investigated as a mild, alternative tagging
procedure to reductive amination using O-(4-nitrobenzyl)-hydroxylamine. Oxime formation was quantitative, but the
reduction step did not always go to completion. Novel O- and N-substituted 7-hydroxycoumaryl- and 3-methoxyben-
zylhydroxylamines were synthesized and shown to couple quantitatively with model saccharides by oxime formation
and reductive hydroxyamination, respectively, under very mild, aqueous conditions. The fluorescent derivatives
produced show good chromatographic and mass spectrometric properties. Both procedures are suitable for the
labeling of carbohydrates and oligosaccharide fragments from glycosaminoglycan structures, such as heparin and
heparan sulfate. © 2001 Elsevier Science Ltd. All rights reserved.
Keywords: Glycosaminoglycan (GAG); Oxime formation; Reductive hydroxyamination; 3-Methoxy-benzylhydroxylamine deriva-
tives; 7-Hydroxycoumarylhydroxylamine derivatives; HPLC; Electrospray-ionization mass spectrometry (ESIMS)
1. Introduction
binding–recognition systems is often limited
to picomole quantities of material. In order to
elucidate the structure of these glycan se-
quences, it is essential to be able to separate
the glycan from any other components present
efficiently and to be able to detect the struc-
ture with high sensitivity. The hydrophilic na-
ture of carbohydrates, coupled with the fact
that they do not possess a suitable chro-
mophore, has meant that they are not readily
amenable to separation techniques such as
reversed-phase HPLC. Although refractive in-
dex and pulsed amperometric detection may
be used for detection of native carbohydrates,
they are not sufficiently sensitive to detect the
low concentrations of glycans found in many
biological systems.
Carbohydrates possess considerable struc-
tural diversity, and as such, the structural
elucidation of complex carbohydrates presents
a challenging problem to the analytical
chemist. The investigation of unusual struc-
tural sequences involved in biologically active
Abbre6iations: 2,5-An-Man6SO3, 6-sulfo-2,5-anhydro-
mannose; 2,5-An-Man, 2,5-anhydro- -mannose; ESIMS,
electrospray-ionization mass spectrometry; GAG, gly-
cosaminoglycan; Glc, -glucose; GlcNAc, N-acetyl- -glu-
cosamine; GlcNSO3, 2-sulfamido-2-deoxy-a- -glucopyranose;
HPLC, high-performance liquid chromatography; HS, hep-
aran sulfate; NBHA, O-(4-nitrobenzyl)hydroxylamine.
* Corresponding author. Tel.: +61-2-61253762; fax: +61-
2-61258114.
D-
D
D
D
D
E-mail address: macleod@rsc.anu.edu.au (J.K. MacLeod).
0008-6215/01/$ - see front matter © 2001 Elsevier Science Ltd. All rights reserved.
PII: S0008-6215(01)00115-X

60
S.L. Ramsay et al. / Carbohydrate Research 333 (2001) 59–71
use of a peptide/GAG fragment complex was
The glycosaminoglycans (GAGs), heparin
found to increase sensitivity, obviate the inter-
ference from inorganic cations, and reduce
desulfation of lower molecular weight
oligosaccharides. However, the degree of
desulfation was found to increase with in-
creasing numbers of sulfate groups present.
Measurement of the molecular weight may be
used to establish the number of monosaccha-
ride subunits, sulfate groups, and N-acetyl
groups present, but it does not provide further
structural information such as the location of
sulfate groups.
and heparan sulfate (HS), are known to play
crucial roles in many biological processes.1–5
HS is a complex sulfated polysaccharide, con-
sisting of chains of up to 400 modified sugar
residues. It is present in most multicellular
animals and has an ubiquitous distribution,
being expressed on the cell surface and in
extracellular matrices of most tissues. HS ex-
ists as a proteoglycan, and despite consider-
able progress in sequencing and cloning the
core polypeptides of the molecule, only limited
structural information has been derived from
the structurally diverse carbohydrate portion.
Although HS is initially biosynthesized as a
simple repeat of a glucuronic acid–N-acetyl-
glucosamine disaccharide unit, subsequent
modifications such as N-deacetylation, N-re-
The methods commonly employed for sepa-
ration of GAG fractions from either nitrous
acid or enzymatic cleavage are gel-permeation
chromatography (GPC), strong anion-
exchange (SAX) chromatography,14 ion-pair-
ing reversed-phase high-performance liquid
chromatography (IP-RP-HPLC),15 polyacry-
lamide gel electrophoresis (PAGE)16–18 and
capillary electrophoresis (CE).19–21 The major-
ity of these techniques requires buffers and/or
salt solutions for optimum separation. The
anionic nature of GAGs bestows a high
affinity for Na+, K+, Ca2+ and other cationic
species, which can complicate the isolation of
pure, salt-free compounds and render the sam-
ples unstable for long-term storage. These
counter-ions also pose a major problem for
analysis of GAG oligosaccharides by mass
spectrometry as they can adduct to the ioniz-
ing molecule in varying proportions, produc-
ing multiple molecular ions at different m/z
values. The desalting and replacement of these
cations with a single counter-ion usually re-
quires dialysis for the larger fragments or
exchange with a salt, such as NH4HCO3,
which is volatile enough for the excess to be
removed under vacuum.
A number of procedures have been devised
to tag glycans selectively at the reducing ter-
minus with a group that will enhance the
sensitivity of detection, many of which also
facilitate chromatographic separations. These
have been the subject of a recent review by
Hase.22 Reductive amination is the most com-
monly used method for the introduction of a
tag at the reducing terminus of the carbohy-
drate. It typically requires heating the sample
at elevated temperatures and low pH for ex-
sulfation, C-5 epimerization of
D
-glucuronic
acid residues to -iduronic acid, and various
L
O-sulfations of each of the residues results in
an extremely diverse structure. HS molecules
generally contain short stretches (4–16
residues) of highly sulfated monosaccharides
joined by relatively long stretches of non-
sulfated units. Heparin, which is present only
in mast cell granules, represents an extreme
form of HS where sulfation and epimerization
are at their most extensive.
In view of its inherent sensitivity and the
potential for the formation of structurally in-
formative fragment ions, mass spectrometry
has been investigated as a means of character-
ization of GAG fragments.6–13 Because of the
presence of multiple sulfate groups, the mass
spectrometric analysis of GAG fragments has
traditionally been carried out in the negative-
ion mode. When mass analyzed as free acids
or as ammonium salts, extensive gas-phase
desulfation occurs.11,13 This lowers the overall
sensitivity and tends to suppress the formation
of structurally informative fragment ions. Al-
though sodium salts are much more stable, the
formation of clustered sodiated ions compli-
cates the assignment of the molecular
mass.11,13
Heparin fragments have been analyzed in
the positive-ion mode as complexes with an
arginine-rich peptide or protein using matrix-
assisted laser desorption ionization time-of-
flight (MALDI-TOF) mass spectrometry.8 The

S.L. Ramsay et al. / Carbohydrate Research 333 (2001) 59–71
61
were also prepared and shown to form N,N-
disubstituted hydroxylamines with sugars un-
der extremely mild conditions.
tended periods with a large excess of an aryl-
amine reagent in the presence of the reducing
agent. Such harsh reaction conditions can lead
to the degradation of heat-sensitive carbohy-
drates and desulfation of GAG fragments.
The procedure also suffers from considerable
variations in yield, because of the competing
direct reduction of the carbohydrate itself,
which occurs at the same pH. One of the
favored reagents for reductive amination of
sugars, 2-aminopyridine, has found particular
application in the quantitative analysis by
HPLC of sialylated N-linked oligosaccharides
from glycoproteins,23,24 utilizing an improved
two-step procedure that overcomes the com-
petition between derivatization and reduc-
tion.25 This still requires high temperatures for
both Schiff’s base formation (90 °C for 60
min) and reduction (80 °C for 50 min) and
was considered unsuitable for sulfated sugars.
Investigations were therefore initiated into
developing a procedure for selective deriva-
tization at the reducing terminus under mild
conditions that would be suitable, not only for
GAG fragments, but also for a range of dif-
ferent sugars. Properties that were required of
the tag were that it should (1) quantitatively
and selectively couple to the reducing termi-
nus of the carbohydrate to form a stable
derivative under mild reaction conditions; (2)
be fluorescent in aqueous media; (3) be hydro-
phobic enough to facilitate reversed-phase
HPLC separations; (4) have a low molecular
weight so that the chromatographic property
of the derivatized oligosaccharide is primarily
determined by the structure of the carbohy-
drate moiety and not that of the tag; and (5)
introduce a site that enhances protonation
efficiency and increases mass spectrometric
sensitivity in the positive-ion mode.
2. Results and discussion
Several N- and O-substituted hydroxy-
lamine derivatives (1–4) were synthesized, and
their spectroscopic properties together with
that of NBHA are shown in Table 1. Reaction
conditions for the coupling of these tags to
oligosaccharides were optimized using model
monosaccharides.
D
-Glucose (Glc) and N-ace-
tyl- -glucosamine (GlcNAc) were used as
D
model hexoses. Nitrous acid degradation of
heparin and heparan sulfate generates prod-
ucts with 2,5-anhydro-
D
-mannose (5, 2,5-An-
-mannose
Man) and 2,5-anhydro-6-O-sulfo-
D
(6, 2,5-An-Man6SO3) residues at the reducing
terminus. These monosaccharides were there-
fore used as model compounds for the label-
ing of heparin and heparan sulfate fragments
arising from nitrous acid degradation.
The established conditions for oxime forma-
tion require heating the reducing oligosaccha-
ride in pyridine with an O-substituted
hydroxylamine,26,27 or mixing an aldehyde in
water and sodium hydroxide with addition of
hydroxylamine.28 However, alkaline condi-
tions have been shown to cause C-2 epimeriza-
tion of N-sulfo glucosamines29 and could also
lead to alkaline ‘peeling’, that is the stepwise
hydrolysis of residues from the reducing ter-
minus.30 As pyridine can be difficult to re-
Table 1
Spectroscopic properties of N- and O-substituted hydroxy-
lamine tags
Compound
UV max Fluorescence
A method that was expected to meet these
requirements was reductive oxyamination with
O-substituted hydroxylamine derivatives.
(nm)
(nm)
O-(4-Nitrobenzyl)-
284
Commercially
available
O-(4-nitroben-
hydroxylamine (NBHA)
O-(4-Methoxybenzyl)-
hydroxylamine (1)
zyl)hydroxylamine (NBHA)26,27 was used to
optimize the reaction conditions. Fluorescent
O-substituted methoxybenzyl- and 7-hydroxy-
coumaryl derivatives of hydroxylamine were
synthesized, and their reaction with model
glycans was investigated. Subsequently, the
corresponding N-substituted hydroxylamines
207, 275 307
8-Aminooxymethyl-7-hydroxy- 335
460
coumarin (2)
N-(3-Methoxybenzyl)-
hydroxylamine (3)
7-Hydroxy-8-hydroxyamino-
methylcoumarin (4)
207, 275 307
335 460

62
S.L. Ramsay et al. / Carbohydrate Research 333 (2001) 59–71
labeling. Instead, we explored the use of
aqueous conditions at mildly acidic pH values.
NBHA was introduced by Darvill and co-
workers as a reducing terminus tag as it of-
fered improved chromatographic separation
for neutral and acidic oligosaccharides, whilst
also providing
a
suitable UV chro-
mophore.26,27 In order to avoid the use of
pyridine and the rather laborious derivatiza-
tion procedure used by Darvill and co-work-
ers, we investigated the reaction of NBHA
with Glc and GlcNAc, using water alone as
the solvent. The formation of the oxime at pH
values from pH 3 to 6 was monitored by
reversed-phase HPLC using a diode array de-
tector. The reaction produced the oxime
product in quantitative yields at an optimum
value of pH 4, which was used for all subse-
quent oxime preparations. HPLC analysis did,
however, indicate formation of multiple prod-
ucts (Fig. 1(a)). The fact that HPLC separa-
tion had effectively desalted the derivatives
allowed the separated derivatives to be col-
lected and subjected to electrospray-ionization
mass spectrometric (ESIMS) analysis. The
mass spectrum of each of the products indi-
cated that they all had the same molecular
move, neither of these conditions for oxime
formation was suitable for GAG fragment
Fig. 1. HPLC chromatograms showing: (a) GlcNAc–NBHA oxime derivative and (b) reduced GlcNAc–NBHA derivative.

S.L. Ramsay et al. / Carbohydrate Research 333 (2001) 59–71
63
matogram complicates the analysis of un-
known structures and reduces the overall
sensitivity. Moreover, monitoring of the Glc-
oxime derivatives indicated that hydrolysis of
the purified oximes collected from HPLC oc-
curs at a rate of about 2% per day during
storage below 0 °C in 0.1% TFA. It was there-
fore decided to reduce the oxime derivative to
a single, more stable N,O-disubstituted hy-
droxylamine product to improve chromato-
graphic analysis, increase sensitivity, and
allow storage over longer periods. The forma-
tion of a secondary amine group would also
provide a suitable site for protonation during
positive ion ESIMS analysis.
weight, which suggested that they were iso-
meric, namely, the E- and Z-acyclic oximes,
and a- and b-cyclic products (Scheme 1). Iso-
lation of any one of the peaks, followed by
reinjection into the HPLC, produced a chro-
matogram that was similar to that of the
original mixture, showing that the products
were readily interconvertible. HPLC analysis
showed that, at room temperature and pH 4,
a quantitative yield of Glc–NBHA and Glc-
NAc–NBHA derivatives was achieved after
130 and 350 min, respectively. At 70 °C and
pH 4 the reaction with Glc and GlcNAc was
complete in 30 and 80 min, respectively. In
contrast, coupling of NBHA to the exposed
aldehyde group of 2,5-An-Man (5) and 2,5-
An-Man6SO3 (6) was complete at room tem-
perature and pH 4 in under 20 min.
Sodium cyanoborohydride (NaCNBH3) was
chosen as a reductant as it is a ‘soft’ reducing
agent and has a pH dependency, whereby
some selectivity can be gained.31 Oximes were
preformed at pH 4 with excess NBHA, and
The fact that several oxime products of Glc
and GlcNAc appear in the HPLC chro-
Scheme 1. Formation of various isomeric oxime derivatives with glucose.

64
S.L. Ramsay et al. / Carbohydrate Research 333 (2001) 59–71
the pH was then adjusted to pH 3 before the
addition of 2 equiv of NaCNBH3. The reac-
tion was monitored by reversed-phase HPLC.
The reduction of the Glc–NBHA and Glc-
NAc–NBHA oxime derivatives to their re-
spective N,O-disubstituted hydroxylamine
derivative both resulted in a single major
product (Fig. 1(b), Scheme 2). Whilst the re-
duction was slow (ꢀ5 h at room temperature
for Glc–NBHA and 20 h for GlcNAc–
NBHA) yields were consistently \90%, but
not quantitative due to the presence of traces
of unreduced oxime (Fig. 1(b)). The products
were stable for some months when stored at
−20 °C. The reason for the slow reduction of
these derivatives appears to lie in the sluggish
interconversion of the non-reducible cyclic to
acyclic isomers (Scheme 1). This was sup-
ported by the observation that reduction of
the acyclic 2,5-An-Man–NBHA oxime deriva-
tive proceeded rapidly and quantitatively. Re-
duction of the 2,5-An-Man6SO3 –NBHA
oxime, however, also did not go to comple-
tion, presumably due to the presence of the
sulfate group.
multiple isomeric oxime derivatives. The
oximes were produced in quantitative yield,
under mild reaction conditions (pH 4, rt).
Reduction of the oximes resulted in a single
major, stable, fluorescent N,O-disubstituted
hydroxylamine product together with variable
amounts of unreduced oxime, comparable to
the results for the NBHA-oximes. The identi-
ties of the oximes and their reduction products
were confirmed by ESIMS that revealed
molecular ions consistent with the proposed
structures.
The rationale behind the preparation of the
N-substituted hydroxylamines as tagging
reagents for the sugars resulted from the ob-
servation of an anomalous peak in the HPLC
trace during studies of the reductive oxyami-
nation reaction. While probing the kinetics of
the NaCNBH3 reductive oxyamination of Glc-
NAc and NBHA, we observed under certain
conditions (pH\4, rt, 10:1 GlcNAc–NBHA)
a new product at a much shorter HPLC reten-
tion time. This was isolated and identified by
ESIMS–MS as a reduced bis-sugar derivative
of NBHA with an MH+ at m/z 579. We
postulated that its formation must involve a
quaternary ammonium ion intermediate, re-
ducible by NaCNBH3 without the need for the
presence of acid. This unexpected result led us
to propose that the nitrone product formed
between an N-substituted hydroxylamine and
the aldehyde group of our model sugars,
which contains a formal positive charge on the
nitrogen atom, should likewise be reduced by
NaCNBH3 to the corresponding N,N-disubsti-
tuted hydroxylamine under near neutral pH
conditions.
Fluorescent reagents are required for chro-
matographic detection of these derivatives at
the sub-picomolar level. Since the UV chro-
mophore NBHA is not fluorescent and none
are available commercially, it was necessary to
synthesize low-molecular-weight fluorescent
reagents. The fluorescent O-substituted hy-
droxylamine
tags
O-(4-methoxybenzyl)-
hydroxylamine (1) and 8-aminooxymethyl-7-
hydroxycoumarin (2) were prepared and re-
acted with the four model glycans (Glc,
GlcNAc, 5 and 6), under the conditions used
for NBHA derivatization. As with NBHA,
HPLC analysis indicated that 1 and 2 formed
To test this proposal, the two fluorescent
N-substituted hydroxylamines, N-(3-methoxy-
Scheme 2. Reduction of E- and Z-oximes to a single N,O-substituted hydroxylamine.

S.L. Ramsay et al. / Carbohydrate Research 333 (2001) 59–71
65
benzyl)-hydroxylamine (3) and 7-hydroxy-8-
hydroxyaminomethylcoumarin (4) were syn-
thesized by initially forming an oxime, via
condensation of hydroxylamine with the rele-
vant aryl aldehyde, and then rapidly reducing
the oxime with NaCNBH3 at pH 3. The four
model glycans (Glc, GlcNAc, 5 and 6) were
reacted with 3 and 4 under aqueous conditions
at pH 6.5 and room temperature in the pres-
ence of NaCNBH3 in a one-pot reaction to
give the corresponding N,N-disubstituted hy-
droxylamines in quantitative yield. The reac-
tion was carried out at pH 6.5 as N,N-disub-
stituted hydroxylamines have been shown to
autoxidize at pH 7 and above,32,33 whilst be-
low pH 6 competitive direct reduction of the
aldose is possible. By performing the reaction
on Glc and GlcNAc without added
NaCNBH3, the nitrone intermediates (Scheme
3) were shown by HPLC to be present in the
aqueous solution at pH 6.5 at surprisingly
high equilibrium concentrations of 17 and
12%, respectively. Moreover, the nitrones
could be isolated from aqueous solution, and
identified by ESIMS. Nitrones are commonly
reduced to the corresponding substituted
amine with powerful reductants such as palla-
dium–H2 or lithium aluminum hydride,34,35
Scheme 3. Nitrone formation and reduction of 2,5-anhydro-
derivative.
D-mannose and D-glucose using an N-substituted hydroxylamine

66
S.L. Ramsay et al. / Carbohydrate Research 333 (2001) 59–71
but it appears that the nitrone is sufficiently
polarized for the milder NaCNBH3 to reduce
it. There are potentially several isomeric forms
for the intermediate nitrone derivatives
analogous to those of the oxime derivatives,
namely, acyclic E- and Z-nitrones (all model
glycans), a- and b-pyranose and a- and b-
furanose conformers (Glc and GlcNAc only).
The nitrone intermediates in the cyclic forms
are non-reducible, and their formation would
hinder the nitrone reduction process unless
interconversion between cyclic and acyclic
contributors is rapid. The Glc–nitrone inter-
mediate is reduced quickly and quantitatively
so it is presumably acyclic. The reduction of
the GlcNAc–nitrone is slow by comparison
even if carried out as a one-pot procedure,
possibly because it exists predominantly in
one or more of the cyclic conformers. This is
supported by the one-pot reductive hydrox-
yaminations of 2,5-An-Man (5) and 2,5-An-
Man6SO3 (6) at pH 6.5 and room
temperature, which were complete within 30
min (Scheme 3), since the aldehyde group of
2,5-An-Man cannot be tied up in a cyclic
structure.
series of labeled glucose polymers, Glc1–14+
differing in molecular weight (ESIMS) by 162
mass units. The obvious benefit of reversed-
phase HPLC analysis of the labeled product,
as opposed to traditional anion-exchange
chromatography, is that the fractions which
are eluted do not contain buffers such as
sodium hydroxide and sodium acetate, and so
are immediately amenable to mass spectro-
metric analysis.
Commercial corn syrup, also labeled with 4,
was subjected to HPLC analysis, and the col-
lected fractions were analyzed by ESIMS. Pos-
itive-ion ESIMS analysis of the labeled Glcn
polymers produced protonated molecular ions
for each fraction (Fig. 3). The series also
exhibited formation of Y-type glycosidic
cleavages,36 with charge retention on the basic
N,N-disubstituted hydroxylamine moiety.
In order to investigate the tagging of hep-
arin and heparan sulfate oligosaccharide frag-
ments which arise from enzymatic cleavage, a
monosaccharide that is frequently present at
the reducing end of these fragments, 2-deoxy-
2-sulfamido-a-
D
-glucopyranose
(GlcNSO3),
was subjected to both oxime formation and
reductive hydroxyamination with 1 and 3, re-
spectively. The oxime derivative of GlcNSO3
was formed quantitatively with 1 but consisted
of a mixture of isomers. As the reduction of
oxime derivatives was earlier shown to not be
quantitative for Glc, GlcNAc and 2,5-An-
Man6SO3, reduction of the oxime derivative
of GlcNSO3 was not attempted. The N,N-
To demonstrate the chromatographic and
mass spectrometric properties of the N-substi-
tuted tags, a dextran hydrolysate and corn
syrup were both labeled with 4, using the
reductive hydroxyamination procedure de-
scribed, and analyzed using reversed-phase
HPLC. The chromatogram of the labeled dex-
tran hydrosylate (Fig. 2) indicated a discrete
Fig. 2. Reversed-phase HPLC chromatogram of a dextran hydrolysate labeled with 4.

S.L. Ramsay et al. / Carbohydrate Research 333 (2001) 59–71
67
disubstituted hydroxylamine derivative of
GlcNSO3, from reductive hydroxyamination
with 3, although a single compound, was pro-
duced in a yield of only 49%. Oxime forma-
tion is therefore the preferred method for
derivatization of enzymatically cleaved hep-
arin and heparan sulfate fragments some of
which contain GlcNSO3 at the reducing termi-
nus. Although the number of isomers present
could complicate chromatographic separation
of the oximes, this problem may be minimized
when dealing with large oligosaccharides when
properties of the sulfated oligosaccharide, and
not the tag, can dominate the separation char-
acteristics of the derivative. This is especially
true when using PAGE gel chromatography,
for which compound 2 was specifically
prepared.
To summarize, the N-substituted hydroxy-
lamines (3 and 4) satisfy all of the previously
stated ideal properties required for tagging the
reducing end of labile sugars. The reductive
hydroxyamination proceeds in a one-pot reac-
tion selectively and quantitatively (with the
exception of GlcNSO3) to give a single deriva-
tive; conditions are extremely mild (aq solu-
tion, pH 6.5, rt); the products are fluorescent,
have good HPLC properties and ESIMS sen-
sitivity. The N,N-disubstituted hydroxyamine
products autoxidise above pH 7, and unlike
the oximes, are therefore not amenable to
standard PAGE gel chromatography per-
formed at pH\8.32,33 Although 3 and 4 are
not as yet commercially available, they can be
readily prepared using the procedure outlined
in Section 3.
The O-substituted hydroxylamines (1 and 2)
formed oximes quantitatively with all of the
model monosaccharides (Glc, GlcNAc,
GlcNSO3, 5 and 6) under mild conditions (aq
solution, pH 4, rt), but reduction to the corre-
sponding N,O-disubstituted oxyamine, al-
though consistently high, was less than
quantitative for Glc, GlcNAc and 6. It is
therefore recommended that derivatization
should be carried through only as far as the
oxime. Formation of multiple isomeric oximes
with Glc and GlcNAc could complicate
HPLC analysis, but they would be suitable for
PAGE gel-chromatography because of their
stability at basic pH values. Thus this is the
method of choice for the study of N-sulfated
oligosaccharides from enzymatic cleavage of
heparins.
Application of the above two complemen-
tary derivatization procedures to the struc-
Fig. 3. ESIMS analysis of collected corn syrup fractions (Glcn) labeled with 4, where n=2, 3, or 4.

68
S.L. Ramsay et al. / Carbohydrate Research 333 (2001) 59–71
tural analysis of heparins will be the subject of
future publications.
follows. Nitrous acid was perpared by adding
NaNO2 (100 mL, 1 M) to citrate buffer at pH
4 (400 mL, 1 M) and allowed to stand at rt for
10 min prior to use. Glucosamine (or glu-
cosamine 6-sulfate) (10 mL, 1 M) dissolved in
water (1 mL) was then added to the HNO2
solution and allowed to stand at rt for 30 min.
Excess HNO2 was eliminated by addition of
ammonium sulfamate until evolution of gas
3. Experimental
General methods.—Melting points were
measured on a Reichert melting point appara-
1
tus without correction. H NMR spectra were
recorded at 500 or 300 MHz with Varian
Inova or Gemini spectrometers, respectively.
Chemical shifts are given in l-units (ppm)
with reference to either Me4Si (l 0) as an
internal standard, or to water (l 4.78) when
D2O was used as the solvent. High-resolution
electron-ionization mass spectra (HREIMS)
were acquired on a Micromass AutoSpec dou-
ble-focussing mass spectrometer operating at a
resolution of 10,000 with PFK as reference.
Electrospray-ionization mass spectra (ESIMS)
were acquired using a Micromass Quattro II
triple quadrupole mass spectrometer, fitted
with an electrospray probe. The solvent com-
position was 1:1 water–MeCN, and the flow
rate into the electrospray probe was typically
10 mL/min. The cone voltage used was in the
range 40–90 V.
ceased. The solution of 2,5-anhydro-
D
-man-
nose (or 2,5-anhydro-6-O-sulfo-
D
-mannose)
was stored below 0 °C until needed.
Hydrolysis of dextran.—Dextran (100 mg)
was dissolved in HCl (1 mL, 1 M) and heated
at 100 °C for 2 h. The hydrosylate was
lyophilized prior to use.
General procedure for the coupling of model
saccharides to O-substituted hydroxylamine
deri6ati6es 6ia an oxime intermediate.—Sac-
charides were reacted with a tenfold excess of
O-substituted hydroxylamine derivative in cit-
rate or acetate buffer (100 mM, pH 4). When
the reaction was carried out at rt, HPLC
analysis revealed that the oxime was formed in
quantitative yield for Glc and GlcNAc after
130 and 350 min, respectively. The reaction
with 2,5-An-Man and 2,5-An-Man6SO3 was
complete in less than 20 min. At 70 °C the
reaction with Glc and GlcNAc was completed
in 30 and 80 min, respectively. The pH was
then adjusted to pH 3 using HCl (1 M) before
addition of 2 equiv of NaCNBH3. Reduced
products were purified by reversed-phase
HPLC.
Coupling of saccharides to N-substituted hy-
droxylamine deri6ati6es 6ia a nitrone intermedi-
ate.—Saccharides were reacted with a tenfold
excess of N-substituted hydroxylamine deriva-
tive in phosphate buffer (100 mM, pH 6.5),
with the typical final concentration of tag in
the range of 50–100 mM. NaCNBH3 was
then added to a concentration of 50 mM. The
reaction with 2,5-An-Man was complete in
less than 20 min, whereas the reaction with
Glc and GlcNAc was complete in 1 and 4 h,
respectively, at rt.
Re6ersed-phase HPLC analysis.—Separa-
tions were performed using a Hewlett–Pack-
ard HP1090M ternary solvent delivery system,
fitted with a diode-array detector and a
HP1046A programmable fluorescence detec-
tor. Alltech Alltima C18 columns (5 mm, 250×
4.6 mm) were used for all analyses. The
formation of tagged model glycans was moni-
tored using a binary solvent system consisting
of 0.1% aq TFA as solvent A, and 0.1% TFA
in MeCN as solvent B. A solvent gradient was
used with the concentration of solvent B rising
from an initial concentration of 5% to a final
concentration of 40% over 20 min. The flow
rate was 1 mL/min.
Preparation or purchase of model saccha-
rides.—Glucose, N-acetylglucosamine, 2-de-
oxy-2-sulfamido-a- -glucopyranose and all
D
other reagents were purchased from Sigma–
Aldrich unless stated otherwise. Commercial
corn syrup was obtained from a health food
O-(4-Methoxybenzyl)hydroxylamine (1).—
To a stirred solution of 4-methoxybenzyl alco-
hol (1.28 g, 10 mmol) in THF (50 mL), was
added N-hydroxyphthalimide (1.63 g, 10
mmol) and triphenylphosphine (2.62 g, 10
store. 2,5-Anhydro-
dro-6-O-sulfo- -mannose were prepared by
an adaptation of published procedures37 as
D
-mannose and 2,5-anhy-
D

S.L. Ramsay et al. / Carbohydrate Research 333 (2001) 59–71
69
tion mixture was allowed to warm to rt and
was stirred until the reaction was complete
(ꢀ2 h). The solvents were then removed un-
der vacuum, and CH2Cl2 (10 mL) was added.
Ice water (10 g) was added with stirring, and
the aqueous layer was extracted with CH2Cl2
(2×10 mL). The organic layers were com-
bined and dried (MgSO4), and the solvent was
removed under vacuum to give 8-formyl-7-
methoxymethyloxy coumarin as a mustard
solid (468 mg, 95%). To a stirred solution of
8-formyl-7-methoxymethyloxycoumarin (580
mg, 2.5 mmol) in EtOH (10 mL) and CH2Cl2
(10 mL) at rt, was added NaBH4 (500 mg,
13.2 mmol). The solution was stirred for 30
min until the dark-yellow color had faded.
Hydrochloric acid (10%) was added sparingly,
with stirring, so that only sufficient acid was
added to destroy the excess NaBH4 without
hydrolyzing the methoxymethyl group. This
was achieved by addition of acid until the
evolution of hydrogen ceased. The aqueous
layer was then extracted twice with CH2Cl2.
The organic layers were combined and dried
(MgSO4), and the solvent was removed under
vacuum to give 8-hydroxymethyl-7-methoxy-
methyloxycoumarin as a pale-yellow oily
residue (550 mg, 93%). To a stirred solution of
8-hydroxymethyl-7-methoxymethyloxycou-
marin (500 mg, 2.1 mmol) in THF (100 mL),
was added N-hydroxyphthalimide (350 mg,
2.1 mmol) and triphenylphosphine (550 mg,
2.1 mmol), and the mixture stirred at rt until
the reagents were dissolved. DEAD (0.5 mL,
3.2 mmol) was then added slowly to the solu-
tion over a period of several min. The color of
the solution remained red–yellow once the
DEAD was in excess. The reaction mixture
was stirred overnight at rt. The solvent was
reduced to one-third the volume under vac-
uum to give a pale-yellow precipitate that was
collected via filtration. Recrystallization of the
precipitate from EtOH–CH2Cl2 gave N-ph-
thaloyl - 8 - hydroxylaminemethylene - 7 - meth-
oxymethyloxycoumarin as white crystals (640
mg, 80%). The methoxymethyl protecting
group was removed by stirring N-phthaloyl
8-hydroxylaminemethylene-7-methoxymethy-
loxycoumarin (365 mg, 0.96 mmol) in TFA (5
mL) and HCl (5 mL, 10%) at rt until TLC
showed that deprotection was complete (ꢀ2
mmol) which was stirred at rt until dissolved.
Diethyl azodicarboxylate (DEAD) (1.74 mL,
11 mmol) or dipropyl azodicarboxylate
(DPAD) (2.17 mL, 11 mmol) was then added
slowly to the solution over a period of several
min. The color of the solution remained red–
yellow once the DEAD or DPAD was in
excess. Solvent was removed under vacuum,
and the product was dissolved in CH2Cl2 and
filtered. The product was recrystallized from
CH2Cl2–hexane several times to give white
crystals of N-phthaloyl 4-methoxybenzyl-O-
hydroxylamine (1.95 g, 69%). The N-ph-
thaloyl
4-methoxybenzyl-O-hydroxylamine
(850 mg, 3 mmol) was dissolved in MeOH (60
mL), hydrazine hydrate (0.46 mL, 9.5 mmol)
was added, and the solution stirred at rt for 3
h. A pH of 2 was maintained by addition of
HCl (2 M). The solution was cooled on ice,
and the resultant precipitate was removed by
vacuum filtration. Solvent was removed, and
the product was partitioned between CH2Cl2
and water (adjusted to pH 8 with NH4HCO3).
Addition of methanolic or ethereal HCl pro-
duced the hydrochloride salt of 1 as white
crystals. Addition of hexane to the solution
produced additional 1 (369 mg, 64%) as white
1
crystals: mp 145–146.5 °C; H NMR (D2O): l
7.32 (d, 2 H, J 8.6 Hz), 6.93 (d, 2 H, J 8.6 Hz),
4.89 (s, 2 H); 3.72 (s, 3 H); HREIMS: Anal.
Calcd for C8H11NO2, 153.0790; found: m/z
153.0794.
8 - Aminooxymethyl - 7 - hydroxycoumarin
(2).—To stirred TFA (10 mL) on ice was
added 7-acetylumbelliferone (1.5 g, 7.4 mmol)
and hexamethylenetetramine (1.5 g, 10.7
mmol). The solution was allowed to warm to
rt, and then it was refluxed for 8 h. The excess
TFA was removed under vacuum, 2 vol of
water were then added to the remaining solu-
tion, and the mixture was warmed at 60 °C
whilst stirring for 30 min. Upon cooling on ice
a pale-yellow solid precipitated from solution
and was collected by filtration and washed
several times with water to give 8-formyl-7-
hydroxycoumarin38 as a pale-yellow solid (757
mg, 54%). Methoxymethyl chloride (1.2 mL,
15.8 mmol) was slowly added to a stirred
solution of 8-formyl-7-hydroxycoumarin (400
mg, 2.1 mmol) in THF (20 mL) and triethy-
lamine (2 mL, 14.3 mmol) at 0 °C. The reac-

70
S.L. Ramsay et al. / Carbohydrate Research 333 (2001) 59–71
H), 4.32 (s, 2 H); 3.76 (s, 3 H); 13C NMR
(CDCl3): l 159.6, 135.0, 129.6, 121.9, 114.9,
114.0, 56.6, 55.1; HREIMS: Anal. Calcd for
C8H11NO2, 153.0790; found: m/z 153.0788.
7-Hydroxy-8-hydroxyaminomethylcoumarin
h). Solvent was then removed under vacuum,
and the residue was recrystallized from 1:1
MeOH–CH2Cl2 to give N-phthaloyl 8-hy-
droxylaminemethylene-7-hydroxycoumarin as
an off-white precipitate (285 mg, 88%). To a
solution of N-phthaloyl 8-hydroxylamine-
methylene-7-hydroxycoumarin (60 mg, 0.18
mmol), stirred in 1:1 EtOH–CH2Cl2 (20 mL)
was added hydrazine hydrate (60 mL, 1.9
mmol). The reaction was stirred at rt until
TLC showed that removal of the phthaloyl
group was complete (ꢀ2 h). The solution was
then made alkaline (pH 9–10) with NaHCO3
solution (1 M) and partitioned between
CH2Cl2 and water, making sure that all traces
of excess hydrazine were removed. The or-
ganic layer was dried (MgSO4) and then re-
acidified with aq methanolic HCl. The solvent
was then gradually removed under vacuum
until the hydrochloride salt of 8-
aminooxymethyl-7-hydroxycoumarin (2) (28
mg, 64%) precipitated out of solution as a
(4).—8-Formyl-7-hydroxycoumarin
was
formed as described for 2. Hydroxylamine
hydrochloride (400 mg, 5.8 mmol) was stirred
with 8-formyl-7-hydroxycoumarin (600 mg,
3.2 mmol) in 1:1 water–THF (20 mL), with
the pH adjusted to pH 4 using NaOH (2 M),
for 1 h at rt. 7-Hydroxy-8-[(hydroxyimino)-
methyl]coumarin (540 mg, 83%) was precipi-
tated by slowly removing volatile solvents un-
der vacuum. 7-Hydroxy-8-[(hydroxyimino)-
methyl]coumarin (500 mg, 2.4 mmol) was re-
duced by stirring in THF (10 mL) with
NaCNBH3 (900 mg, 14.3 mmol) whilst main-
taining a pH of 2–3 using HCl (2 M) at rt.
The solvent was slowly removed under vac-
uum until the product precipitated as the hy-
drochloride salt. The process was repeated
several times to give 7-hydroxy-8-hydrox-
yaminomethylcoumarin (4) (560 mg, 96%) as
a white solid: mp\310 °C; 1H NMR (MeOH-
d4): l 7.64 (d, 1 H, J 9.5 Hz), 7.22 (d, 1 H, J
8.6 Hz), 6.62 (d, 1 H, J 8.6 Hz), 5.99 (d, 1 H,
J 9.5 Hz), 4.16 (s, 2 H); 13C NMR (D2O): l
170.95, 166.91, 160.93, 153.33, 138.58, 120.11,
119.37, 118.37, 109.66, 50.64.
1
white solid: mp 168–171 °C; H NMR (D2O):
l 7.83 (d, 1 H, J 10.0 Hz), 7.49 (d, 1 H, J 9.0
Hz), 6.85 (d, 1 H, J 8.5 Hz); 6.20 (d, 1 H, J
10.5 Hz), 5.22 (s, 2 H); HREIMS: Anal. Calcd
for C10H9NO4, 207.0532; found: m/z 207.0531.
N-(3-Methoxybenzyl)hydroxylamine (3).—
3-Methoxybenzaldehyde (5 mL, 41.1 mmol)
was stirred with hydroxylamine hydrochloride
(5 g, 72.0 mmol) in EtOH (25 mL), with
NaOH (2 M) used to adjust the pH to 4. After
stirring for 30 min at rt, the solvent was
removed until a white precipitate formed, and
the precipitate was collected by vacuum filtra-
tion. The oxime was reduced without further
purification by stirring in MeOH (25 mL) with
NaCNBH3 (1.25 g, 19.9 mmol) whilst main-
taining pH 2–3 using HCl (2 M) at rt for 30
min. After removing the solvent under vac-
uum, the product was back-extracted with
CH2Cl2 and NaHCO3 solution (1 M, pH 9).
Acidification of the dried CH2Cl2 portion with
ethereal HCl caused some hydrochloride salt
of the product to precipitate from solution,
but an oil began to form. To avoid oil forma-
tion, hexane was used to crystallize out the
References
1. Gallagher, J. T. Biochem. Soc. Trans. 1997, 25, 1206–
1209.
2. Stringer, S. E.; Gallagher, J. T. Int. J. Cell Biol. 1997, 29,
709–714.
3. Bernfield, M.; Gotte, M.; Park, P. W.; Reizes, O.;
Fitzgerald, M. L.; Lincecum, J.; Zako, M. Ann. Re6.
Biochem. 1999, 68, 729–777.
4. Engelberg, H. Cancer 1999, 85, 257–272.
5. Tumova, S.; Woods, A.; Couchman, J. R. Int. J.
Biochem. Cell Biol. 2000, 32, 269–288.
6. Dell, A.; Rogers, M. E.; Thomas-Oates, J. E.; Huckerby,
T. N.; Sanderson, P. N.; Nieduszynski, I. A. Carbohydr.
Res. 1988, 179, 7–19.
7. Khoo, K. H.; Morris, H. R.; Mcdowell, R. A.; Dell, A.;
Maccarana, M.; Lindahl, U. Carbohydr. Res. 1993, 244,
205–223.
8. Juhasz, P.; Biemann, K. Carbohydr. Res. 1995, 270, 131–
147.
9. Ii, T.; Kubota, M.; Okuda, S.; Hirano, T.; Ohashi, M.
Eur. Mass Spectrom. 1995, 1, 11–22.
remaining
N-(3-methoxybenzyl)hydroxyl-
10. Ii, T.; Kubota, M.; Okuda, S.; Hirano, T.; Ohashi, M.
Glycoconjugate J. 1995, 12, 162–172.
11. Chai, W.; Luo, J.; Lim, C. K.; Lawson, A. M. Anal.
Chem. 1998, 70, 2060–2066.
amine hydrochloride (3) (2.33 g, 30%) from
1
solution as a white solid: mp 133–135 °C; H
NMR (D2O): l 7.34 (t, 1 H, J 8.7), 7.02 (m, 3

S.L. Ramsay et al. / Carbohydrate Research 333 (2001) 59–71
71
25. Kondo, A.; Suzuki, J.; Kuraya, N.; Hase, S.; Kato, I.;
Ikenaka, T. Agric. Biol. Chem. 1990, 54, 2169–2170.
26. von Deyn, W.; York, W. S.; Albersheim, P.; Darvill, A.
G. Carbohydr. Res. 1990, 201, 135–144.
12. Rhomberg, A. J.; Ernst, S.; Sasisekharan, R.; Biemann,
K. Proc. Natl. Acad. Sci. USA 1998, 95, 4176–4181.
13. Chai, W.; Hounsell, E. F.; Bauer, C. J.; Lawson, A. M.
Carbohydr. Res. 1995, 269, 139–156.
27. Pauly, M.; York, W. S.; Guillen, R.; Albersheim, P.;
Darvill, A. G. Carbohydr. Res. 1996, 282, 1–12.
28. Furniss, B. S.; Hannaford, A. J.; Rogers, V.; Smith, P.
W. G.; Tatchell, A. R. Vogel’s Textbook of Practical
Organic Chemistry, Including Qualitati6e Organic Analy-
sis, 4th ed.; Longman Group: London, 1978; pp. 809–
812.
29. Yamada, S.; Watanabe, M.; Sugahara, K. Carbohydr.
Res. 1998, 309, 261–268.
30. Whistler, R. L.; BeMiller, J. N. Ad6. Carbohydr. Chem.
1958, 13, 289–329.
31. Borch, R. F.; Bernstein, M. D.; Durst, H. D. J. Am.
Chem. Soc. 1971, 93, 2897–2904.
32. Johnson, D. H.; Rogers, M. A. T.; Trappe, G. J. Chem.
Soc. 1956, 1093–1103.
33. Rogers, M. A. T. J. Chem. Soc. 1956, 2784–2789.
34. Cusmano, G. Gazz. Chim. Ital. 1921, 51, 306–309.
35. Exner, O. Collect. Czech. Chem. Commun. 1955, 20,
202–208.
36. Domon, B.; Costello, C. E. Glycoconjugate J. 1988, 5,
397–409.
14. Linhardt, R. J.; Rice, K. G.; Kim, Y. S.; Lohse, D. L.;
Wang, H. M.; Loganathan, D. Biochem. J. 1988, 254,
781–787.
15. Karamanos, N. K.; Vanky, P.; Tzanakakis, G. N.; Tse-
genidis, T.; Hjerpe, A. J. Chromatogr. 1997, 765, 169–
179.
16. Rice, K. G.; Rottink, M. K.; Linhardt, R. J. Biochem. J.
1987, 244, 515–522.
17. Edens, R. E.; Al-Hakim, A.; Weiler, J. M.; Rethwisch, D.
G.; Fareed, J.; Linhardt, R. J. J. Pharm. Sci. 1992, 81,
823–827.
18. Turnbull, J. E.; Hopwood, J. J.; Gallagher, J. T. Proc.
Natl. Acad. Sci. USA 1999, 96, 2698–2703.
19. Desai, U. R.; Wang, H.; Ampofo, S. A.; Linhardt, R. J.
Anal. Biochem. 1993, 213, 120–127.
20. Grimshaw, J. Electrophoresis 1997, 18, 2408–2414.
21. Karamanos, N. K.; Hjerpe, A. Electrophoresis 1998, 19,
2561–2571.
22. Hase, S. J. Chromatogr. A 1996, 720, 173–182.
23. Lee, Y. C.; Lee, B. I.; Tomiya, N.; Takahashi, N. Anal.
Biochem. 1990, 188, 259–266.
24. Rice, K. G.; Takahashi, N.; Namiki, Y.; Tran, A. D.;
Lisi, P. J.; Lee, Y. C. Anal. Biochem. 1992, 206, 278–287.
37. Horton, D.; Philips, K. D. Methods Carbohydr. Chem.
1976, 7, 68–70.
38. Spa¨th, E.; Pailer, M. Chem. Ber. 1935, 68, 940–943.
.