Differentiation of the anomeric configuration and ring form of glucosyl-glycolaldehyde anions in the gas phase by mass spectrometry: isomeric discrimination between m/z 221 anions derived from disaccharides and chemical synthesis of m/z 221 standards

Article abstract of DOI:10.1016/j.carres.2006.11.021

Mass spectrometry of disaccharides in the negative-ion mode frequently generates product anions of m/z 221. With glucose-containing disaccharides, dissociation of isolated m/z 221 product ions in a Paul trap yielded mass spectra that easily differentiated between both anomeric configurations and ring forms of the ions. These ions were shown to be glucosyl-glycolaldehydes through chemical synthesis of their standards. By labeling the reducing carbonyl oxygen of disaccharides with ¹⁸O to mass discriminate between monosaccharides, it was established that the m/z 221 ions are comprised solely of an intact nonreducing sugar with a two-carbon aglycon derived from the reducing sugar, regardless of the disaccharide linkage position. This enabled the anomeric configuration and ring form of the ion to be assigned and the location of the ion to the nonreducing side of a glycosidic linkage to be ascertained. Detailed studies of experimental factors necessary for reproducibility in a Paul trap demonstrated that the unique dissociation patterns that discriminate between the isomeric m/z 221 ions could be obtained from month-to-month in conjunction with an internal energy-input calibrant ion that ensures reproducible energy deposition into isolated m/z 221 ions. In addition, MS/MS fragmentation patterns of disaccharide m/z 341 anions in a Paul trap enabled linkage positions to be assigned, as has been previously reported with other types of mass spectrometers.

Full text of DOI:10.1016/j.carres.2006.11.021

Carbohydrate Research 342 (2007) 217–235
Differentiation of the anomeric configuration and
ring form of glucosyl-glycolaldehyde anions in the
gas phase by mass spectrometry: isomeric discrimination
between m/z 221 anions derived from disaccharides
and chemical synthesis of m/z 221 standards
Tammy T. Fang,a Joseph Zirrollib and Brad Bendiaka,*
aDepartment of Cellular and Developmental Biology and Biomolecular Structure Program, University of Colorado
at Denver and Health Sciences Center, Aurora, CO 80045, USA
bSchool of Pharmacy, University of Colorado at Denver and Health Sciences Center, Aurora, CO 80045, USA
Received 8 August 2006; received in revised form 14 November 2006; accepted 20 November 2006
Available online 23 November 2006
Abstract—Mass spectrometry of disaccharides in the negative-ion mode frequently generates product anions of m/z 221. With glu-
cose-containing disaccharides, dissociation of isolated m/z 221 product ions in a Paul trap yielded mass spectra that easily differen-
tiated between both anomeric configurations and ring forms of the ions. These ions were shown to be glucosyl-glycolaldehydes
through chemical synthesis of their standards. By labeling the reducing carbonyl oxygen of disaccharides with 18O to mass discrim-
inate between monosaccharides, it was established that the m/z 221 ions are comprised solely of an intact nonreducing sugar with a
two-carbon aglycon derived from the reducing sugar, regardless of the disaccharide linkage position. This enabled the anomeric con-
figuration and ring form of the ion to be assigned and the location of the ion to the nonreducing side of a glycosidic linkage to be
ascertained. Detailed studies of experimental factors necessary for reproducibility in a Paul trap demonstrated that the unique dis-
sociation patterns that discriminate between the isomeric m/z 221 ions could be obtained from month-to-month in conjunction with
an internal energy-input calibrant ion that ensures reproducible energy deposition into isolated m/z 221 ions. In addition, MS/MS
fragmentation patterns of disaccharide m/z 341 anions in a Paul trap enabled linkage positions to be assigned, as has been previously
reported with other types of mass spectrometers.
ꢀ 2006 Elsevier Ltd. All rights reserved.
Keywords: Carbohydrate mass spectrometry; Discrimination of disaccharide product anions; Assignment of anomeric configuration; Differentiation
of ring form; Synthetic product ion standards
1. Introduction
form ion cyclotron resonance (FTICR) instruments,
now permits many oligosaccharides to be broken down
stepwise in the gas phase to isolable substructures. We
sought to address the question as to whether reducing
disaccharides, the smallest substructures derived from
oligosaccharides that still contain a glycosidic linkage
between two monosaccharides, can be effectively differ-
entiated and structurally elucidated as gas-phase species.
Disaccharide fragmentation patterns have been the topic
of some important fundamental investigations, both in
the positive-4–9 and negative-ion10–16 modes. Alkyl and
acyl derivatives have also proven valuable.3,17–19
Mass spectrometry (MS) is a valuable tool for analysis of
oligosaccharides. A number of different techniques have
been advocated as they often contribute different infor-
mation and provide overlapping coverage for structural
determination.1–3 The power of multiple isolation/disso-
ciation steps, carried out in ion traps or Fourier trans-
*
Corresponding author. Tel.: +1 303 724 3453; e-mail: Brad.
Bendiak@uchsc.edu
0008-6215/$ - see front matter ꢀ 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carres.2006.11.021

218
T. T. Fang et al. / Carbohydrate Research 342 (2007) 217–235
In the negative-ion mode, previous studies of disac-
phase. We have found no related studies that discrimi-
nate between anomeric configuration and ring forms
of sugars as isolated product ions, which may in part
be due to a lack of commercially available disaccharides
having nonreducing aldofuranosides. Nevertheless, fur-
anosides in general and those of glucose are found in
nature.27–30
charide fragmentation have been carried out with major
emphasis on the assignment of linkage positions, with
characteristic fragmentation patterns in the gas phase
clearly enabling the substitution position on the reduc-
ing sugar to be identified.10–15 Although these studies
were carried out under a wide range of dissociation con-
ditions, some consensus was reached concerning frag-
mentation products. The cross-ring cleavages of the
precursor ion m/z 341 [MꢀH]ꢀ generated product ions
having higher masses than a monosaccharide of m/z
323, 311, 281, 263, 251, and 221. These correspond to
losses of water, one to four –CHOH-backbone units,
and, in the case of the m/z 263 anion, two –CHOH-units
and one water. The presence and abundance of product
ions was dependent on linkage position and apparent
energy deposition into the ions using different methods.
Mechanisms have been proposed that the reducing
monosaccharide may initially rearrange to an open-
chain form, whereupon dissociation occurs highly pref-
erentially on the reducing sugar.11,12,16 Similar proposals
were made in the positive-ion mode with metal-ion
adduction.4–6,8 These results were interesting at the time
and warrant further detailed investigation now that MSn
has become widely available and controlled disassembly
of larger oligosaccharides to disaccharide substructures
is possible.
Fewer studies have been carried out with the intent to
discriminate anomeric configuration and ring forms of
monosaccharides within disaccharides where the disac-
charides have first been isolated as gas-phase ions. As
disaccharides comprise a very large set of isomers, estab-
lishing their structures by MS entirely in the gas phase is
not a trivial matter, without hydrolysis and GC–MS of
individual monomers,20,21 or using permethylation fol-
lowed by hydrolysis or reductive cleavage in solu-
tion.22,23 Discrimination between the a and b anomeric
configurations of the nonreducing sugar of reducing
disaccharides in the gas phase has been possible through
direct in-source dissociation using electrospray-ioniza-
tion,14,15 using mass-analyzed ion kinetic energy
(MIKE) spectrometry and kinetic energy release stud-
ies,24 upon collision-induced dissociation (CID) of iso-
lated disaccharide ions in the negative-ion mode as
alkoxy anions12 and chloride adducts,25 and recently
using two-dimensional variable-wavelength infrared
multiple photon dissociation/mass spectrometry.9 Also,
differentiation of anomeric configuration has been
observed in some cases following derivatization, using
alkyl derivatives of increasing length,17 upon examining
a synthetic set of galactose–fucose isomers having a benz-
yl group at the reducing end,26 and more recently, using
a synthetic set of methylated disaccharide standards
having pyranosyl-1-ene sugars at the reducing end.3 As
yet, though, no general method is available to determine
the configuration of a glycosidic linkage in the gas
Studies carried out with disaccharides in the negative-
ion mode10–16 prompted four questions concerning their
dissociation and the potential of garnering additional
information about their structures: (1) Do any of the
product ions of disaccharide dissociation, upon isolation
and further fragmentation, permit differentiation of ano-
meric configuration and/or ring form? Of the above
studies in the negative-ion mode,10–15 only one further
dissociated the ions derived by cross-ring cleavage of
the disaccharides.12 Of the ions investigated, one ion,
the m/z 221 anion, was reported to show some potential
to differentiate anomers. However, there was enough
reported variability in the spectral data for this ion from
disaccharides having different linkages to question
whether it was solely derived by dissociation events
occurring on the reducing sugar.12 (2) Do the dissocia-
tion spectra of product ions match those of synthetic
compounds? Although product ions derived from disac-
charides are often considered transitory intermediates,
some, for example, have been hypothesized to be anions
of a hexopyranoside linked to a pentose, tetrose, glycer-
aldehyde, or glycolaldehyde.11,12,16 If they are indeed
identical, their spectra should presumably match. More-
over, such synthetic compounds could provide useful
reference spectra provided they are the same molecules
as isolated gas-phase species. (3) Are the mass spectra
of the precursor and product ions reproducible enough
to enable their identification from month-to-month in
the same instrument, and even between instruments of
the same general type (such as Paul traps)? (4) Are the
product ions that have a larger m/z than a monosaccha-
ride (m/z 221 or greater) actually mixtures of isomers
themselves? Hypothesized mechanisms have proposed
that these ‘cross-ring cleavages’ occur on the open-chain
form of the reducing sugar.11,12,16 In the positive-ion
mode, substitution of hexose-containing disaccharides
at the carbonyl position with 18O has been performed
to selectively mass label ions derived from the reducing
monosaccharide.5,6,8 However, in the negative-ion
mode, to our knowledge, just one report of lactose
(b-D-Galp-(1!4)-D-Glc) 13C-labeled at the glucose C-1
position13 has been performed to effectively mass
discriminate between the two monosaccharides. Their
results were inconclusive in answering the above hypoth-
esis that cross-ring cleavages solely occur on the reduc-
ing sugar. For example, these authors observed m/z
281 and 282 ions in a ratio of 70:30. The m/z 282 anion
could have either resulted from cleavage of the C-5 to C-
6 portion of the reducing sugar or from a two-carbon

T. T. Fang et al. / Carbohydrate Research 342 (2007) 217–235
219
fragment loss from the nonreducing sugar. Nonetheless,
such studies are, in general, very valuable as they enable
the origins of many ions to be clearly identified.
product ions are shown in Scheme 1. In one report12 it
was proposed that 6-linked glucose-containing disaccha-
rides may give rise to the aldehydo structure 1, whereas
2-linked disaccharides may give rise to the enol structure
3. As all the different tautomers/positional charge vari-
ants shown in Scheme 1 have the same m/z, it does
not appear straightforward to be able to discriminate
among them by mass spectrometry, and such ionic vari-
ants might rapidly equilibrate in the gas phase through
proton transfer. Experiments were performed to deter-
mine whether m/z 221 anions, derived in the gas phase
from related disaccharides having different linkage posi-
tions, were in fact the same ion or potentially different
isomers, where the two-carbon fragment might originate
from either monosaccharide of the disaccharides.
Two disaccharides varying solely in their anomeric
configuration (isomaltose, a-D-Glcp-(1!6)-D-Glc, and
gentiobiose, b-D-Glcp-(1!6)-D-Glc, precursor [MꢀH]ꢀ
m/z 341) were dissociated in a Paul trap in the nega-
tive-ion mode, and the resultant m/z 221 product ions
were isolated and further dissociated by MS3. The spec-
tra are shown in Figure 1, panels A and B. They differed
strikingly in the abundance of m/z 161, 113, and 101 ions,
under identical conditions in the Paul trap. Similarly, dis-
sociation of the m/z 221 ions derived from the 2-linked
disaccharides kojibiose, a-D-Glcp-(1!2)-D-Glc and sop-
horose, b-D-Glcp-(1!2)-D-Glc is shown in Figure 1,
Bearing the above questions in mind, experiments
were initiated, in a limited fashion with certain linkages
and monosaccharides, to address them. It is demon-
strated, for simple glucose-containing disaccharides hav-
ing 2-, 4-, and 6-linkages, that glucosyl-glycolaldehyde
product ions ([MꢀH]ꢀ 221) are derived solely from the
nonreducing sugar of these disaccharides, bearing an
additional two-carbon fragment derived from the reduc-
ing sugar. Synthesis of these (MW 222) structures dem-
onstrates that as [MꢀH]ꢀ ions they dissociate identically
to the m/z 221 ions derived from disaccharides and frag-
ment in a way that enables their anomeric configuration
and ring form to be easily and reproducibly assigned.
2. Results and discussion
2.1. Fragmentation patterns of m/z 221 anions isolated
from disaccharide precursors in the negative-ion mode are
identical to synthetic glucopyranosyl-glycolaldehydes
Dissociation of disaccharides in the negative-ion mode
has been reported in the literature,10–15 and four differ-
ent structures that have been proposed for the m/z 221
1
-
OH
-H
O
H
H
H
HO
HO
O
OH
O
O
or
or
OH
2
O
H
HO
HO
O
-
OH
-
O
H
H
OH
In-source
decay or
CID
HO
HO
O
H
O
-H
H,OH
-
OH
HO
OH
OH
-H
HO
H
3
O
OH
HO
O
O
m/z 341
HO
OH
OH
H
H
or
4
H
O
HO
HO
-
OH
m/z 221
O
Scheme 1.

220
T. T. Fang et al. / Carbohydrate Research 342 (2007) 217–235
H
H
H
-
H
-
OH
H
- H
H
OH
H
- H
H
H
O
O
H
H
H
HO
HO
O
H
HO
H
B
A
H
HO
OH
O
H
OH
O
H
H
O
H
H
131
161
100
95
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
100
95
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
131
101
.
113
99
161
.
221
.
221
.
159
159
87 101
0
0
100
150
200
100
150
200
m/z
m/z
H
H
-
H
H
H
-
OH
H
- H
OH
H
- H
H
H
H
O
O
H
H
H
H
D
HO
C
HO
HO
O
H
H
HO
OH
O
OH
O
H
H
O
H
H
131
161
100
95
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
100
95
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
101
131
161
113
221
159
221
99
159
87 101
0
0
100
150
200
100
150
200
m/z
m/z
Figure 1. Dissociation of m/z 221 anions selected for MS3 after disaccharide fragmentation in an LCQ Paul trap in the negative-ion mode.
Disaccharide precursor ions [MꢀH]ꢀ, m/z 341, were isolated and following dissociation, yielded m/z 221 anions that were isolated and subjected to a
further round of dissociation at controlled collision energies. Mass spectra were recorded from the m/z 221 ion isolated from: (A) isomaltose, a-D-
Glcp-(1!6)-D-Glc; (B) gentiobiose, b-D-Glcp-(1!6)-D-Glc; (C) kojibiose, a-D-Glcp-(1!2)-D-Glc; (D) sophorose, b-D-Glcp-(1!2)-D-Glc. Proposed
structures of the anions are shown above each spectrum. They are based on similarity of mass spectra to the synthetic glucosyl-glycolaldehydes
(Fig. 2) and data from 18O labeling of disaccharides at the reducing carbonyl oxygen (Figs. 3–6). Error bars in panel A represent the standard
deviation of the indicated ions calculated using the intensity of m/z 131 as exactly 100%, for 20 spectra accumulated over a 2-month period.

T. T. Fang et al. / Carbohydrate Research 342 (2007) 217–235
221
panels C and D. Note that in both cases where the non-
OH
5. R1 = H, R2 = -OCH2CH=CH2
O
reducing sugars were linked in a configuration, isolated
m/z 221 product ions dissociated to give essentially iden-
tical mass spectra (Fig. 1, panels A and C). Likewise,
when sugars were linked in b configuration, the m/z
221 product ions fragmented identically (panels B and
D), but very differently from their a-linked counterparts.
Proposed structures of the m/z 221 ions derived from the
disaccharides are shown with each mass spectrum. It is
important to point out that solely based on the informa-
tion given in Figure 1A–D, these proposed structures
could have alternatives, such as the reducing sugar
attached to a two-carbon piece from the nonreducing
sugar. However, the structures proposed are based on
additional evidence provided by the synthetic glucosyl-
glycolaldehydes and studies carried out with 18O-labeled
disaccharides, described later in this report.
The m/z 221 ions were also observed in low abun-
dance in fragmentation of the 4-linked disaccharides
maltose, a-D-Glcp-(1!4)-D-Glc, and cellobiose, b-D-
Glcp-(1!4)-D-Glc, in the Paul trap. With the 4-linked
disaccharides, these anions were furnished in greater
abundance directly from in-source fragmentation (as re-
ported in Ref. 15). The spectrum of the isolated m/z 221
product ion (not shown) from maltose was identical to
the a-linked ion (Fig. 1A and C), and the spectrum of
the m/z 221 product ion from cellobiose (not shown)
was identical to the b-linked ion (Fig. 1B and D). Con-
sequently, the m/z 221 anion appears to be identical
whether it originates from a 2-, 4-, or 6-linked disaccha-
ride, and it dissociates markedly differently when
derived from an a-linked disaccharide as compared to
one that is b-linked. It is worth noting that the m/z
221 product ion has been generated in much higher
abundance for 4-linked disaccharides under high energy
CID conditions in a sector instrument12 as compared to
CID conditions available in the LCQ Paul trap. We also
observed more abundant m/z 221 product ions using
collision conditions available with a triple quadrupole
instrument, described later.
The m/z 221 anion was observed in dissociation of
3-linked disaccharides in the Paul trap, but only in trace
quantities (<1% of the base product ion peak) in
spectra obtained with an adequate number of scans to
obtain very high signal-to-noise ratios. It was therefore
impractical to isolate and perform MS3 studies on the
ion derived from 3-linked disaccharides in the Paul
trap. However, as much as a 12% abundance as com-
pared to the base product peak has been noted using
in-source fragmentation in a triple quadrupole instru-
ment,15 and a small abundance has been seen in
higher-energy sector instruments using CID.12 We have
also noticed small quantities of the m/z 221 ion from
nigerose (a-D-Glcp-(1!3)-D-Glc) and laminaribiose
(b-D-Glcp-(1!3)-D-Glc) using MS/MS in Q2 of a triple
quadrupole instrument (1–5% of the base product ion
6. R1 = H, R2 = -OCH2CH=O
7. R1 = -OCH2CH=CH2, R2 = H
8. R1 = -OCH2CH=O, R2 = H
R1
HO
HO
OH
R2
HO
H
HO
R1
R2
O
9. R1 = H, R2 = -OCH2CH=CH2
10. R1 = H, R2 = -OCH2CH=O
11. R1 = -OCH2CH=CH2, R2 = H
12. R1 = -OCH2CH=O, R2 = H
OH
H
H
H
OH
Scheme 2.
peak). It appears that isolation of appreciable m/z 221
from 3-linked disaccharides will require higher energy
excitation conditions and perhaps finer control of
energy input into their m/z 341 anions than is available
in a Paul trap.
Mass spectra of the synthetic glucopyranosyl-glycol-
aldehydes in a Paul trap (compounds 6 and 8, Scheme
2) showed m/z 221 anions (MꢀH)ꢀ and small amounts
(<20%) of m/z 239 anions (the hydrates). In Figure 2,
panels A and B are shown the MS/MS spectra of the
synthetic a- and b-D-glucopyranosyl-glycolaldehydes
(m/z 221 anions), respectively. Dissociation patterns of
the synthetic compounds having defined anomeric con-
figurations and ring forms were essentially identical to
those of the respective m/z 221 product ions derived
from disaccharides having the same anomeric configura-
tions and ring forms. They were easily discriminated by
their fragmentation patterns. The data strongly indicate
that the synthetic glucopyranosyl-glycolaldehydes are
the same as the m/z 221 anions isolated by fragmenta-
tion of the 2-, 4-, and 6-linked disaccharides, and that
they readily discriminate anomeric configuration. A
potential product ion having an intact reducing sugar
with a two-carbon fragment derived from the nonreduc-
ing sugar does not appear to be a dissociation pathway
of significant abundance based on this data, which was
supported by 18O-labeling studies of disaccharides
described later.
2.2. Differentiation among the four isomeric D-glucosyl-
glycolaldehydes (a- and b-pyranosides and furanosides) in
the negative-ion mode
To further substantiate the nature of the m/z 221 prod-
uct ions derived from disaccharides, the a- and b-D-
glucofuranosyl-glycolaldehydes (10 and 12, Scheme 2)
were prepared. Their dissociation spectra, under identi-
cal conditions described for their pyranosyl counter-
parts, are shown in Figure 2, panels C and D. Clearly,
these molecules dissociate markedly differently than

222
T. T. Fang et al. / Carbohydrate Research 342 (2007) 217–235
H
H
H
-
H
H
- H
OH
H
-
H
- H
H
OH
H
O
H
O
H
HO
HO
H
H
H
A
B
H
H
HO
HO
O
OH
O
O
OH
O
H
H
H
H
161
131
100
95
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
100
95
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
131
101
161
159
113
221
221
159
99
87 101
0
0
100
150
200
100
150
200
m/z
m/z
-
-
OH
HO
OH
- H
O
H
H
H
HO
O
H
O
H
O
O
H
H
H
OH
OH
C
D
H
O
H
- H
H
H
H
H
H
H
OH
OH
101
161
100
100
95
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
95
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
99
161
159
221
173
221
101
113
177
99
0
0
100
150
200
100
150
200
m/z
m/z
Figure 2. Differentiation of synthetic glucosyl-glycolaldehyde isomers having different anomeric configurations and ring forms by MS/MS in the
negative-ion mode. The [MꢀH]ꢀ, m/z 221, precursor anions were isolated and MS/MS spectra recorded following dissociation in an LCQ Paul trap.
The spectra were recorded from: (A) a-D-glucopyranosyl-2-glycolaldehyde; (B) b-D-glucopyranosyl-2-glycolaldehyde; (C) a-D-glucofuranosyl-2-
glycolaldehyde; (D) b-D-glucofuranosyl-2-glycolaldehyde. Structures of the synthetic compounds are shown above each spectrum.
the glucopyranosyl-glycolaldehydes. Each dissociation
spectrum is unique and enables, for all four isomeric
variants, both the anomeric configuration and ring form
of the ions to be assigned.

T. T. Fang et al. / Carbohydrate Research 342 (2007) 217–235
223
2.3. The monosaccharide origin of the m/z 221 anion
derived from 2-, 4-, and 6-linked disaccharides based on
18O-labeling the carbonyl oxygen of the reducing glucose
having 4-linkages yielded relatively low quantities of the
m/z 221 ions in the Paul trap, although they were clearly
observable (1–5% of the base product ion peak); more
m/z 221 product ion was generated with in-source frag-
mentation using the trap. As with the 6-linked disaccha-
rides, the characteristic product ion of m/z 221 was
observed for the 18O-labeled 4-linked disaccharides, with
no observable m/z 223 product anions (Fig. 4,C and D).
This indicates that with 4-linked disaccharides, the m/z
221 anion was essentially entirely derived from the intact
nonreducing sugar, with the two-carbon fragment orig-
inating from the reducing monosaccharide. Possible
fragmentation locations are indicated in the structural
diagrams accompanying Figure 4.
The 2-linked disaccharides kojibiose and sophorose
were also 18O-labeled at the reducing carbonyl group,
and dissociation of the resultant m/z 343 anions in the
Paul trap was compared to the unlabeled compounds
(Fig. 5). MS/MS spectra of the 18O-labeled, m/z 343 pre-
cursor ions revealed m/z 223 product anions instead of
the m/z 221 anions that were observed for the unlabeled
compounds (Fig. 5, panels A–D). In the case of 2-linked
disaccharides, it appeared feasible that the two-carbon
fragment originated solely from carbons 1 and 2 of the
reducing sugar, as postulated in the fragmentation dia-
grams in Figure 5.
The purpose of these labeling studies was to experimen-
tally determine, in the negative-ion mode, whether m/z
221 anions originated by dissociation of either the
reducing or nonreducing sugar of disaccharide precursor
ions or whether the two-carbon fragment arose only
from one of them.
Previously, disaccharides exchanged at the carbonyl
oxygen of the reducing sugar with 18O5,8 have been
investigated as single Li+ adducts in the positive-ion
mode. In those cases the equivalent ion to the m/z 221
anion, in the positive-ion mode, was an m/z 229 ion
where the intact sugar residue was solely derived from
the nonreducing monosaccharide in the case of 6- or
4-linked disaccharides. We followed the same labeling
protocol that worked well and effectively exchanged
the reducing sugar carbonyl oxygen for the two 6-linked
disaccharides isomaltose and gentiobiose, the 4-linked
disaccharides maltose and cellobiose, and the 2-linked
disaccharides sophorose and kojibiose. It should be
noted that labeling the reducing carbonyl oxygen with
18O to mass-discriminate ions derived from disaccha-
rides in the negative-ion mode has not previously been
reported.
The MS/MS spectra of the unlabeled 6-linked disac-
charides in the negative-ion mode (isolated precursor
ions of m/z 341) are shown in Figure 3, panels A and
B, and spectra of their 18O-labeled isotopomers (precur-
sor ions of m/z 343) are shown in panels C and D. Nota-
bly, m/z 221 product ions were found in both cases, with
no observable m/z 223 product ions derived from the
18O-labeled molecules. This indicates that for 6-linked
disaccharides, the m/z 221 anions essentially solely con-
tain an intact nonreducing sugar with an additional two-
carbon aglycon derived from positions 5 and 6 of the
reducing sugar. This is consistent with the results of
fragmentation of the m/z 221 anions isolated from these
disaccharides in MS3 studies, where dissociation pat-
terns matched those of synthetic glucosyl-glycolalde-
hyde ions. It is also worthy of note that ions of m/z
281 and 251, previously hypothesized to arise by two-
and three-carbon losses from the reducing end of the
reducing sugar,10–15 were unchanged in m/z for the
18O-labeled molecules, supporting previous proposals.
The MS/MS spectra of the disaccharides maltose and
cellobiose, having linkages at the 4-position, are shown
in Figure 4, panels A and B, and spectra of their isotopo-
mers 18O-labeled at the carbonyl position of the reduc-
ing sugar are shown in Figure 4, panels C and D.
These experiments were carried out under higher energy
collision conditions using a triple-quadrupole instru-
ment with unit mass resolution for selection of the
respective m/z 341 or 343 precursor ions. Disaccharides
To check whether it was possible that the m/z 223
anion might comprise the intact reducing sugar, with a
two-carbon fragment derived from the nonreducing
sugar, it was isolated and further dissociated an addi-
tional step (MS3). Note that the fragmentation patterns
of the m/z 223 ions derived from either a- or b-linked
disaccharides having
a linkage at the 2-position
(Fig. 6A and B) were essentially the same as their respec-
tive 221 counterparts (Fig. 1C and D), with the sole
difference being that the m/z 131 product ion was con-
verted to an m/z of 133. However, to further verify that
the two-carbon fragment was actually derived from the
reducing portion of the 2-linked disaccharides, the syn-
thetic a- and b-glucopyranosyl-glycolaldehydes were
also exchanged at the carbonyl position with 18O. These
synthetic (m/z 223) compounds fragmented identically
(Fig. 6C and D) to the ions isolated from their respective
18O-labeled a- or b-linked disaccharides (Fig. 6A and B).
Compiling all the 18O-labeling studies, the key observa-
tions were (1) that the two-carbon fragment of m/z 221
(or 223) anions was always derived from the reducing
sugar of disaccharides during dissociation in the nega-
tive-ion mode, and (2) that the ring form and anomeric
configuration of the nonreducing sugar was maintained
intact for m/z 221 anions. In the nomenclature of
Domon and Costello,31 reducing disaccharides frag-
mented to give m/z 221 cross-ring product anions solely
of the A type, not the X type. It is worthy of note that
for nonreducing disaccharides having an alkyl aglycon
(methyl, ethyl, and butyl glycosides of lactose), X-type

224
T. T. Fang et al. / Carbohydrate Research 342 (2007) 217–235
179
179
A100
B100
H
95
O
H
95
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
O
H
OH
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
O
H
H
OH
H
H
OH
H
H
HO
H
H
OH
HO
HO
H
H
O
HO
H
H
OH
OH
HO
H
H
OH
OH
H
OH
H
O
HO
H
OH
O
H
221
H
H
221
H
H
H
221
221
341
341
323
161
161
143
281
309
323
311
101
251
309
281
251
101
0
0
100
150
200
250
300
350
100
150
200
250
300
350
m/z
m/z
179
179
C100
100
95
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
D
18
O
H
18
95
H
O
H
OH
H
H
H
OH
H
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
H
O
H
OH
H
OH
HO
HO
H
HO
O
H
H
H
H
OH
OH
HO
HO
H
H
H
OH
OH
OH
H
O
HO
H
OH
H
H
221
O
H
H
H
221
H
221
221
343
343
325
281
311
325
323
163
215
163
311
215
281
101
145
251
251
101
0
0
100
150
200
250
300
350
100
150
200
250
300
350
m/z
m/z
Figure 3. Fragmentation patterns of 6-linked disaccharides and comparison to their 18O-labeled isotopomers by MS/MS in the negative-ion mode.
Precursor ions (m/z 341 or 343) were isolated and dissociated under the same conditions in a Paul trap. Shown are the MS/MS spectra of: (A)
isomaltose, a-D-Glcp-(1!6)-D-Glc; (B) gentiobiose, b-D-Glcp-(1!6)-D-Glc; (C) isomaltose 18O-labeled at the reducing carbonyl position; (D)
gentiobiose 18O-labeled at the reducing carbonyl position. Fragmentation diagrams that indicate the origin of the m/z 221 product ions are shown
above each spectrum. The open-chain form of the reducing sugar was drawn solely to indicate the fragmentation position; no implication regarding
the relative abundance of either the open-chain or cyclic form(s) of the reducing sugar in the gas phase is intended, nor are any inferences drawn
concerning dissociation mechanisms.

T. T. Fang et al. / Carbohydrate Research 342 (2007) 217–235
225
H
H
H
101
101
H
H
OH
OH
H
100
100
H
H
OH
O
H
OH
HO
H
O
H
HO
H
HO
H
H
H
221
221
HO
A
B
O
H
H
HO
H
OH
HO
221
221
H
OH
H
O
H
OH
H
HO
H
OH
H
H
H
89
HO
O
221
O
161
89
119
161
97
73
97
179
143
119
143
73
113
341
341
113
179
221
263
263
100
200
300
100
200
300
m/z
m/z
H
89
103
H
100
100
H
OH
H
H
H
OH
H
O
H
OH
H
OH
HO
H
O
H
HO
O
H
H
C
D
H
HO
H
HO
221
221
H
HO
H
OH
H
221
221
HO
O
H
H
OH
H
OH
103
H
OH
HO
H
18
O
H
H
HO
H
18
O
89
97
73
221
119
97
73
119
115
145
163
101
115
163
343
343
145
179
221
179
263
263
100
200
300
100
200
300
m/z
m/z
Figure 4. Dissociation of 4-linked disaccharides and comparison to their 18O-labeled isotopomers by MS/MS in the negative-ion mode. Precursor
ions (m/z 341 or 343) were selected and dissociated in the second quadrupole of a Sciex triple quadrupole instrument. Shown are the MS/MS spectra
of: (A) maltose, a-D-Glcp-(1!4)-D-Glc; (B) cellobiose, b-D-Glcp-(1!4)-D-Glc; (C) maltose 18O-labeled at the reducing carbonyl position; (D)
cellobiose 18O-labeled at the reducing carbonyl position. Fragmentation diagrams that indicate the possible origin(s) of the m/z 221 product ions are
shown above each spectrum. The open-chain form of the reducing sugar was drawn only to illustrate fragmentation position(s); no implication
regarding the relative abundance of either the open-chain or cyclic form(s) of the reducing sugar in the gas phase is intended, nor are any inferences
intended concerning dissociation mechanisms.

226
T. T. Fang et al. / Carbohydrate Research 342 (2007) 217–235
263
221
A100
100
95
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
B
95
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
H
H
OH
H
H
H
H
OH
OH
H
H
O
H
HO
H
HO
H
H
OH
HO
H
H
HO
H
H
H
O
HO
HO
H
H
HO
H
OH
OH
OH
H
O
HO
H
H
H
OH
O
221
221
H
H
O
O
263
221
323
341
323
341
309
179
179
309
0
0
100
150
200
250
300
350
100
150
200
250
300
350
m/z
m/z
265
223
C 100
100
95
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
D
95
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
H
H
OH
H
H
H
H
H
OH
OH
O
H
H
HO
H
HO
H
H
OH
HO
H
H
HO
H
H
H
O
HO
HO
H
H
HO
OH
OH
H
OH
H
H
O
HO
H
H
O
OH
223
18
O
223
18
O
H
H
265
223
325
343
325
343
311
179
179
311
0
0
100
150
200
250
300
350
100
150
200
250
300
350
m/z
m/z
Figure 5. Dissociation of 2-linked disaccharides and comparison to their 18O-labeled isotopomers by MS/MS in the negative-ion mode. Precursor
ions (m/z 341 or 343) were isolated and dissociated under identical CID conditions in a Paul trap. Shown are MS/MS spectra of: (A) kojibiose, a-D-
Glcp-(1!2)-D-Glc; (B) sophorose, b-D-Glcp-(1!2)-D-Glc; (C) kojibiose, 18O-labeled at the reducing carbonyl position; (D) sophorose, 18O-labeled at
the reducing carbonyl position. Fragmentation diagrams that indicate the origins of the m/z 221 or 223 anions are included with each spectrum. As
mentioned in Figs. 3 and 4, the open-chain form of the reducing sugar was drawn solely to illustrate the fragmentation position.

T. T. Fang et al. / Carbohydrate Research 342 (2007) 217–235
227
H
H
H
H
-
H
H
-
-
-
H
OH
H
OH
H
H
H
O
O
H
H
H
H
H
HO
HO
HO
HO
O
A
H
B
H
18
18
O
OH
OH
O
H
H
O
H
H
133
161
100
100
95
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
95
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
101
133
223
161
113
223
159
159
101
0
0
100
150
200
100
150
200
m/z
m/z
H
H
H
H
-
H
H
-
H
-
H
OH
H
-
OH
H
H
O
D
O
C
H
H
H
H
HO
HO
H
HO
HO
H
O
H
18
O
18
OH
H
OH
O
H
O
H
H
161
133
100
95
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
100
95
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
133
101
223
161
159
223
113
159
101
0
0
100
150
200
100
150
200
m/z
m/z
Figure 6. Comparison of dissociation patterns of 18O-labeled product ions having m/z 223 derived from 2-linked disaccharides with those of synthetic
glucosyl-glycolaldehydes 18O-labeled at the glycolaldehyde carbonyl oxygen. For panels (A) and (B) are shown MS3 spectra of the m/z 223
ions isolated from dissociation of: (A) kojibiose, a-D-Glcp-(1!2)-D-Glc, 18O-labeled at the reducing carbonyl oxygen and (B) sophorose,
b-D-Glcp-(1!2)-D-Glc, 18O-labeled at the reducing carbonyl oxygen. In panels (C) and (D) are shown MS/MS spectra of synthetic glucopyranosyl-
glycolaldehydes 18O-labeled at the glycolaldehyde carbonyl oxygen: (C) a-D-glucopyranosyl-2-glycolaldehyde; (D) b-D-glucopyranosyl-2-
glycolaldehyde.

228
T. T. Fang et al. / Carbohydrate Research 342 (2007) 217–235
ions having a two-carbon fragment derived from the
nonreducing sugar have been reported in the negative-
ion mode.32 Consequently, the value of 18O-labeling
the carbonyl oxygen of reducing disaccharides in order
to mass discriminate between potential isomeric product
ions derived from portions of either monosaccharide
cannot be overemphasized.
ments, where both precursor and product ions are
subject to further fragmentation events. Therefore,
product ions directly originate from precursor ions in
a Paul trap rather than through successive intermediate
fragments. (2) Vibrational energy input into selected
ions in a Paul trap accrues through multiple successive
low-energy collisions. This is also fundamentally differ-
ent than the higher energy collisions that are typical of
triple quadrupole instruments, and especially in sector
instruments. For these reasons, the mass spectra of
disaccharides in a Paul trap would be expected to be
somewhat different than previous reports using the
above instruments, and differences in product ion ratios
were observed.
On a practical basis in the Paul trap, 3- and 6-linked
disaccharides could be differentiated from 2- and 4-
linked in the negative-ion mode because the m/z 263
ion was not present in mass spectra of 3- or 6-linked
structures, even in trace quantities. The 6-linked disac-
charides could be differentiated from the 3-linked ones
based on the much greater abundance of the m/z 221
product ion (in the 3-linked examples there was just a
trace), the fact that the m/z 251 ion was the most abun-
dant product ion above m/z 179 in 3-linked structures,
and the presence of a low abundance m/z 233 ion for
3-linked disaccharides that was not found, even in trace
quantities, with any other linkage (also reported using
CID in an FTICR instrument13). The 2- and 4-linked
disaccharides were easily discriminated based on the
much greater abundance of m/z 221 and 263 anions
for 2-linked disaccharides and on the fact that the m/z
281 anion was the most abundant product ion above
m/z 179 for 4-linked structures.
Regarding mechanistic aspects of disaccharide frag-
mentation, some previous hypotheses have suggested
that disaccharide precursor ions in the negative-ion
mode may fragment through successive losses of m/z
(ꢀ60) or C2H4O2 units.11,12 In the positive-ion mode,
Asam and Glish8 clearly demonstrated in a Paul trap
that a direct loss of C4H8O4 can occur from precursor
ions of 4- and 6-linked disaccharides, which they verified
using double resonance experiments. Our experiments in
the negative-ion mode also demonstrate that for 2-, 3-,
4-, and 6-linked disaccharides, a direct loss of C4H8O4
occurs to give the m/z 221 ion in the Paul trap. However,
we also sought to experimentally determine whether
alternative fragmentation pathways could give rise to
the m/z 221 ion under the fragmentation conditions
available in a triple quadrupole instrument, using the
precursor ion scan experiment (Table 2). In this experi-
ment, the origins of product ions derived by potential
multiple fragmentation events in Q2 of the triple quad-
rupole can be determined. Using 2-, 3-, 4-, and 6-linked
disaccharides, it was evident that in the triple quadru-
pole instrument, multiple mechanisms give rise to the
m/z 221 anion. For all the linkages, the direct loss of
2.4. Disaccharide linkage positions can be assigned based
on dissociation patterns in a Paul trap
As previously reported with sector instruments,10–12
FTICR,13 and in-source fragmentation during electro-
spray-ionization,14,15 dissociation of disaccharides in
the negative-ion mode yielded diagnostic fragmentation
patterns that enabled linkage positions to be assigned.
In the Paul trap, fragmentation patterns were also
observed that were characteristic of specific linkage
positions (Table 1). It is important to point out that
the data in Table 1 were accumulated with enough scans
to obtain high signal-to-noise ratios so that ions in the
abundance range of 0.1–1% of the base product ion peak
were clearly observable. Hence the criterion for the
complete absence of certain ions ((ꢀ) in Table 1) was
estimated as an intensity relative to the base product
peak no greater than 0.1%, or 1:1000 the intensity.
It is also worthy of note that true MS/MS is com-
pletely different than in-source fragmentation14,15 and
that MS/MS utilizing CID in a Paul trap is different
than MS/MS in sector, triple quadrupole, or FTICR
instruments for two reasons. (1) In a Paul trap, precur-
sor ions are excited at a unique resonance excitation fre-
quency to induce CID events only in these ions based on
their selective increase in axial velocity. Their product
ions are not subject to additional dissociation as little
or no additional vibrational energy is transferred to
product ions.8,33 This is fundamentally different than
CID in sector, triple quadrupole, or FTICR instru-
Table 1. Product ions generated in the LCQ Paul trap from disaccha-
ride precursor ions (m/z 341) having different linkagesa
Disaccharide
linkageb
Diagnostic product ion (m/z), relative
abundancec
323
281
263
251
233
221
1!2
1!3
1!4
1!6
+
LA
LA
+
+
LA
+
LA
LA
ꢀ
+
LA
LA
+
ꢀ
LA
ꢀ
TR
LA
+
LA
ꢀ
+
ꢀ
a Disaccharides were solely glucose-containing (see text), examined in
the negative-ion mode with nanospray infusion.
b Linkages of both anomeric configurations were examined.
c Diagnostic product ions were those of higher MW than a monosac-
charide (m/z 179), with multiple scans averaged so that noise levels
were <0.1% of the base product peak. As compared to the intensity
of the base product peak, (+) = abundance >5%; LA = low abun-
dance, 1–5%; TR = trace, 0.1–1.0% abundance; (ꢀ) = <0.1%
abundance.

T. T. Fang et al. / Carbohydrate Research 342 (2007) 217–235
229
Table 2. Disaccharide anions that give rise to the m/z 221 ion via multiple high-energy collision events in a Sciex triple quadrupole instrument:
precursor ion scansa
Disaccharide linkageb
Anions giving rise to m/z 221 anion, and approximate relative abundancec
1!2
1!3
1!4
1!6
341 (>95%); 311 (<5%); 281 (<5%)
341 (>50%); 311 (<50%)
341 (<50%); 281 (>50%)
341 (>80%); 323 (<5%); 311 (<10%); 281 (<20%)
a Precursor ion scans were carried out in the negative-ion mode on the m/z 221 product ion under conditions where the Sciex collision energy was set
to ꢀ15 V. Initial precursor ions were disaccharide anions (m/z 341). Disaccharides were solely glucose-containing (see text), introduced through
electrospray infusion.
b Linkages of both anomeric configurations were examined.
c Precursor ion scanning in triple quadrupole instruments characterizes multiple ions that can give rise to a given product ion by multiple stepwise
dissociation events in Q2. Abundances (> or < the values indicated) are presented as such because experiments were independently performed with
both a and b anomers, where the values represent mininum (>) or maximum (<) quantities of multiple-collision precursor species for both anomers.
C4H8O4 was in fact a facile dissociation process to di-
rectly generate the m/z 221 product ion from the m/z
341 precursor ion (Table 2). For 2-, 3-, and 6-linkages,
an initial loss of CH2O (ꢀ30) also occurred concurrently
to generate an m/z 311 intermediate, which then under-
went a second direct collisional loss of C3H6O3 (ꢀ90) to
produce the m/z 221 product ion. For the 2-, 4-, and
6-linkages, two successive losses of C2H4O2 (ꢀ60) also
occurred concurrently via the intermediacy of the m/z
281 product ion to generate the m/z 221 anion, this
being the most facile pathway only for 4-linked disac-
charides. Finally, for 6-linked disaccharides alone, a loss
of a water molecule (ꢀ18) occurred to generate an m/z
323 intermediate, which then underwent a collisional
loss of C4H6O3 (ꢀ102) to generate the m/z 221 product
ion via a fourth independent pathway. Conclusions
regarding disaccharide fragmentation both in the Paul
trap and triple quadrupole were (1) A direct loss of
C4H8O4 occurs in the Paul trap for disaccharides of all
linkages. (2) A direct loss of C4H8O4 occurs in a triple
quadrupole instrument for all linkages, but at least three
other mechanisms can give rise to multiple successive
fragmentation pathways, depending on the linkage posi-
tion. We have refrained from drawing mechanisms be-
cause first, they may be unique to disaccharides having
specific stereochemistries, linkages, and anomeric con-
figurations, second, they clearly are unique to specific
instruments, and third, isotopomers will probably be re-
quired to mass-discriminate ions so that the origins of
each product ion can be confidently determined.
over several months by keeping four parameters care-
fully controlled. The first was the isolation width, which
was kept constant for analysis of the m/z 221 anions (3
Finnigan ‘amu’) and moderately wide purposefully to
improve reproducibility. The second issue was the cali-
bration of the resonance excitation frequency. Normally
after calibration of the instrument, this frequency drifts
over time, and appears to drift at a different rate in the
LCQ than that of the mass measurement frequencies. If
this drifts by 0.2–0.5 amu, it will slightly affect the ratios
of resultant mass spectra, and either requires recalibra-
tion, or can be finely adjusted manually to optimize the
total product ion current by slightly shifting the excita-
tion resonance frequency.
The third issue involves carefully defining the energy
input into an ion, where the ratio of the most abundant
product ion (base) peak to the remaining precursor ion
peak was chosen as the most reliable criterion for repro-
ducible energy deposition into the ions. We selected syn-
thetic a-D-glucopyranosyl-glycolaldehyde as an internal
‘energy-input tuning ion’ and routine standard, in a sim-
ilar fashion to,34 and kept its base product ion:precursor
ion ratio (100:18) as constant as possible over the long
term to ensure consistent energy deposition into this
ion and other isomeric m/z 221 precursor ions. Hence,
over a period of months, the collision energy needs to
be slightly adjusted to attain the same ratio of base
product ion:precursor ion. The fourth issue was moni-
toring the number of ions in the trap. Variation of
dissociation behavior with numbers of ions was investi-
gated by varying in a stepwise manner the number of
ions in the trap (from 3 · 106 to 5 · 103) under otherwise
identical hardware settings and calibration conditions.
The overall number of ions in the trap was found to
slightly affect dissociation patterns, irrespective of other
parameter settings. Consequently, the levels of ions in
the trap need to be approximately the same for compar-
ison of a standard to an unknown (within an order of
magnitude) to have the best reproducibility. It is worth
noting that for these ions, the differences in dissociation
between the four isomers were so marked that this was
not a significant factor in their discrimination.
2.5. Mass spectral reproducibility
Spectral reproducibility is crucial for the identification of
ions as ‘spectral fingerprints’. This is especially germane
when isobaric variants need to be identified, particularly
from truly unknown molecules. Shown in Figure 1A is
the variability of the spectrum of a-D-glucopyranosyl-
glycolaldehyde. Error bars show standard deviation of
peak heights over the collection of 20 independent spec-
tra obtained over a two-month period. Spectral repro-
ducibility could be maintained from day-to-day and

230
T. T. Fang et al. / Carbohydrate Research 342 (2007) 217–235
2.6. Properties of the m/z 221 anions in the gas phase:
evidence for possible open-chain and cyclic structures
based on 1H NMR studies of the glucosyl-glycolaldehydes
in solution
tion of the NMR spectra (Schemes 3 and 4, Table 3).
These cyclic forms have significantly different J-cou-
plings between the glycolaldehyde H-10 and its own H-
20a and H-20b, as opposed to the open-chain hydrate.
Also, differences were seen in the J-couplings of the a-
glucofuranoside ring protons for cyclic forms 10b and
10c (Scheme 4, Table 3), whereas the open-chain hydrate
10a had J-couplings similar to that of a-allylglucofurano-
side (described later in this report) or a-methylglucofur-
anoside.35 This was not observed for the cyclic forms of
the b-glucopyranoside-glycolaldehyde 8b and 8c (Scheme
3, Table 3), indicating that for the a-glucofuranosides,
some minor change in the five-membered ring conforma-
tion takes place when hemiacetal formation occurs be-
tween the glycolaldehyde carbonyl and the sugar OH-2.
As indicated in Scheme 1, four potential classes of gas-
phase anions for individual glucosyl-glycolaldehydes
have previously been postulated, with charge located
potentially on the two-carbon fragment of the former
reducing sugar or on alkoxy anions of hydroxyl groups
of the hexose portion of the ion, where the reducing frag-
ment may be either an aldehyde or enol tautomer. Even
though we have synthesized the glucosyl-glycolaldehydes
and demonstrated their mass spectra to be the same as
those derived from disaccharide dissociation, transient
proton transfers over rapid time scales could occur, espe-
cially under resonance excitation conditions in the trap
when selected ions build up high internal energy as they
approach their dissociation threshold. Consequently,
none of the potential structures shown in Scheme 1 can
be ruled out in equilibria using MS methods alone. Their
gas-phase structures may be even more complicated,
because we have observed three forms for two of the
glucosyl-glycolaldehydes (the b-glucopyranoside and
4
Interestingly, a weak JH,H (<1.0 Hz) was also observed
between the glycolaldehyde H-1 proton and the sugar
ring H-2 proton for structure 8c, a coherent through-
bond coupling similarly sometimes observed between
protons across glycosidic linkages.36 The open-chain hy-
drate could not, of course, contribute to the m/z 221 ion
unless it loses water, but the cyclic forms are isobaric with
the open-chain aldehydo species and any of its potential
tautomers shown in Scheme 1. It is worth noting that
some of the hydrate (m/z 239, ꢁ20% as abundant as
m/z 221) was observed in the gas phase under our ioniza-
tion conditions (nanospray from methanol). These data
indicate that the m/z 221 anions in the gas phase might
exist in additional cyclic forms, via hemiacetal formation
as is typical of reducing sugars. Cyclization may endow
the ions with unique gas-phase equilibria that in part
contribute to the spectral differences observed between
the different anomers and ring forms.
1
the a-glucofuranoside, 8 and 10) by H NMR spectros-
copy in D2O solution (Schemes 3 and 4, and compound
characterization data under the synthesis section). The
other two glucosyl-glycolaldehydes 6 and 12 were only
found in solution as their hydrates. Compounds 8 and
10 exist in an open-chain form as a hydrate, and as two
cyclic, six-membered ring forms through hemiacetal for-
mation via ring closure of the aldehyde with OH-2 of the
sugar, in different proportions as determined by integra-
8a (61%)
OH
H
O
H
H
H
HO
HO
O
OH
OH
OH
H2O
8b (17%)
O
8
OH
OH
O
H
H
H
H
H
HO
HO
O
HO
HO
O
H
OH
O
O
OH
H
H
8c (22%)
OH
H
O
OH
HO
HO
O
H
O
H
H
H
Scheme 3.

T. T. Fang et al. / Carbohydrate Research 342 (2007) 217–235
231
HO
HO
H
10a (11%)
H
O
OH
OH
H
H
OH
O
H
H
H2O
H
H
OH
HO
HO
HO
HO
10
O
H
H
10b (33%)
O
H
H
H
H
OH
H
OH
O
O
H
H
H
H
O
H
H
OH
O
H
H
OH
HO
HO
H
H
10c (56%)
O
H
H
OH
H
H
O
O
H
H
H
HO
Scheme 4.
Table 3. 1H NMR data (500 MHz) for a- and b-D-glucopyranosyl- and glucofuranosyl-glycolaldehyde hydrates and their cyclic forms 6, 8, 10, and
12
Proton signal
6
8a
8b
8c
10a
10b
10c
12
Chemical shifta
H-1
H-2
H-3
H-4
H-5
H-6a
H-6b
H-10
H-20a
H-20b
4.957 d
3.554 dd
3.744 t
4.499 d
3.316 dd
3.497 t
4.554 d
3.658 dd
3.633 t
4.507 d
3.316 dd
3.69c
5.217 d
4.184 t
4.34 t
4.147 dd
3.88 ddd
3.807 dd
3.66 dd
5.190 t
5.415 d
4.246 dd
4.335 d
4.407 dd
3.882 ddd
3.833 dd
3.679 dd
4.976 dd
3.667 dd
3.560 dd
5.491 d
4.253 dd
4.304 d
4.335 dd
3.884 ddd
3.837 dd
3.677 dd
5.121 br sd
4.004 dd
3.517 dd
5.034 s
4.218 br s
4.258 d
4.195 dd
3.990 ddd
3.857 dd
3.702 dd
5.164 t
3.411 t
3.393 t
3.76c
3.472 t
3.72c ddd
3.858 dd
3.758 dd
5.220 t
3.461 ddd
3.915 dd
3.725 dd
5.203 t
3.861 dd
3.661 dd
3.86c
3.589 ddd
3.91c dd
3.76c dd
5.145 br sd
3.984 d
4.231 dd
3.819 dd
5.064 dd
4.051 dd
3.819 dd
3.703 dd
3.511 dd
3.742 dd
3.575 dd
3.677 dd
3.489 dd
3.948 dd
Coupling constantb
J1,2
J2,3
J3,4
J4,5
J5,6a
J5,6b
3.8
9.8
9.3
9.6
1.9
5.4
7.9
9.3
9.1
9.6
2.3
5.8
7.7
9.8
9.6
n.d.e
n.d.e
5.4
7.8
9.8
n.d.e
9.9
2.1
5.4
4.2
3.4
4.8
7.9
3.0
n.d.e
2.5
1.2
3.1
9.0
2.3
6.0
2.7
1.1
3.1
8.8
2.4
6.1
61.0
61.0
4.6
9.1
2.7
6.0
J6a,6b
ꢀ12.0
ꢀ12.4
ꢀ12.5
ꢀ13.3
ꢀ12.0
ꢀ12.0
ꢀ12.0
ꢀ12.1
0
J1 ,2 a
0
4.5
5.2
4.6
5.2
2.7
9.1
61.0
1.8
4.4
5.1
2.4
8.8
2.1
1.5
4.5
0
J1 ,2 b
0
5.1
0
J2 a,2 b
0
ꢀ10.6
ꢀ10.7
ꢀ11.8
ꢀ12.6
ꢀ10.8
ꢀ11.2
ꢀ12.3
ꢀ10.7
a Chemical shifts are in parts per million with spectra recorded at 25 ꢁC in D2O relative to internal acetone at d = 2.225 ppm.
b In Hertz.
c Chemical shifts could only be determined to two decimal places due to spectral overlap.
d br s = broad singlet.
e n.d. = coupling constant not determined due to spectral overlap.

232
T. T. Fang et al. / Carbohydrate Research 342 (2007) 217–235
2.7. Preparation of glucosyl-glycolaldehydes
tion, ring form, and location of the m/z 221 anion to
the nonreducing position, in addition to the information
about linkage position that could be assigned previously
based on disaccharide ion (m/z 341) fragmentation in
the negative-ion mode.10–15 While it is premature to gen-
eralize these observations to other disaccharides with
different monosaccharide stereochemistries, it is worthy
of note that m/z 221 anions have been observed as prod-
ucts from other 2-, 4-, and 6-linked disaccharides.10–15
For 3-linked disaccharide precursor ions, only trace
amounts of the m/z 221 product ions were observed
following dissociation in a Paul trap, in quantities too
small to practically isolate and generate MS3 spectra.
The glucopyranosyl- and glucofuranosyl-glycolalde-
hydes were prepared according to Scheme 5, with details
presented in Section 3. Glycosidation to generate the
allyl glycosides was essentially complete, but yielded
all four allyl glycoside products that required separation
by HPLC. Ozonolysis was clean and furnished the
glycolaldehyde products essentially quantitatively as
judged by NMR, bubbling O3-rich oxygen into a sample
in D2O over a 6–8 min time period. The use of D2O
made quantitative reaction times straightforward to
optimize as an aliquot of the sample could be rapidly
analyzed nondestructively by NMR over different reac-
tion times. Characterization of compounds by NMR
and MS/MS is either tabulated in Table 3 or presented
along with the title compounds.
3. Experimental
3.1. Mass spectrometry
2.8. Concluding remarks
Mass spectrometry was performed to MS2 or MS3 using
either a Thermoquest (Finnigan) classic LCQ instru-
ment of Paul trap design equipped with a Finnigan
nanospray source or MS2 using a Sciex API 3000
equipped with an electrospray-ionization source in the
negative-ion mode. The LCQ was operated using auto-
matic gain control, with a typical needle voltage of
2.8 kV and a heated capillary temperature of 200 ꢁC,
with sample introduced at a flow rate of 0.4 lL/min,
except in the analysis of sugar-glycolaldehydes where
the temperature was lowered to 120 ꢁC. The MSn exper-
iments were carried out with a relative collision energy
that ranged from 27% to 34% and an activation q value
of 0.25. The activation time was set at 30 ms. A total of
20–100 scans were accumulated for each spectral acqui-
sition. The triple quadrupole instrument was operated at
collision gas levels in the second quadrupole of 10 CAD
units (Sciex), collision energies ranging from ꢀ5 to
ꢀ50 V (data in Fig. 4 at ꢀ30 V, and precursor ion scans
in Table 2 at ꢀ15 V), a needle voltage of ꢀ3500 V,
temperature of the countercurrent gas at 250 ꢁC, declus-
tering potential of ꢀ15 V, focusing potential of ꢀ75 V,
with an infusion flow of 70 lL/min. Product-ion scans
were taken from m/z 50–400 over 3 s. Spectra were
The m/z 221 anions derived from glucose-containing
disaccharides were demonstrated, by comparison of
their mass spectra to chemically synthesized standards,
to be glucosyl-glycolaldehydes. Dissociation spectra of
all four isomeric variants revealed two important fea-
tures: (1) The anomeric configuration of the ions can
be easily differentiated. (2) The ring forms (pyranoside
or furanoside) are easily discriminated. Cleavage of
2-, 4-, and 6-linked glucose-containing disaccharides
occurred to give m/z 221 anions that were comprised
solely of the intact nonreducing sugar and a two-carbon
fragment derived from the reducing sugar, clearly estab-
lished by 18O labeling of the carbonyl oxygen of their
reducing glucose residues. This has been hypothe-
sized11,12,16 but not previously firmly established in the
negative-ion mode by isotopically labeling one specific
sugar of disaccharides to mass discriminate between
their monosaccharides after dissociation. This structural
information should be valuable in the studies of larger
oligosaccharides where first disaccharide anions of m/z
341 and second their m/z 221 product ions may be
selected after multiple dissociation/isolation events. It
is of value to be able to assign the anomeric configura-
OH
O
OH
O
H+
+
HO
HO
HO
HO
O
H,OH
OH
OH
O3
OH
(Plus other cyclic forms and anomers)
OH
O
HO
HO
O
O
OH
Scheme 5.

T. T. Fang et al. / Carbohydrate Research 342 (2007) 217–235
233
acquired with unit mass resolution in Q1 and Q3. Sam-
ples were introduced into the source of either instrument
in MeOH or MeOH–water mixtures having up to 5%
water or 18O water. High-resolution MS measurements
were carried out in the negative-ion mode on a Thermo-
quest 7 T FTICR instrument interfaced to a front end
linear ion trap equipped with a nanospray source (Finn-
igan LTQ-FTICR), located in the research resources
center at the University of Illinois at Chicago medical
center.
219 [MꢀH]ꢀ, m/z, relative abundance as % of base peak
bracketed: for 5, 71 (13%), 89 (14%), 101 (54%), 113
(21%), 143 (4%) 159 (10%), 161 (100%, base), 219
(10%, precursor); for 7, 71 (4%), 101 (18%), 113 (8%),
143 (2%), 159 (7%), 161 (100%, base), 219 (9%, precur-
sor). HRMS: [MꢀH]ꢀ, for 5, m/z 219.087 for
C9H15O6; Calcd 219.086; for 7, m/z 219.087 for
C9H15O6; Calcd 219.086. The allyl glucofuranosides
have not been previously prepared. NMR assignments
of a and b anomeric configuration were based on
similarity of the sugar ring J-couplings to the previously
assigned methyl glucofuranosides.35 For 9, allyl a-D-
3.2. Nuclear magnetic resonance (NMR) spectroscopy
1
glucofuranoside, H NMR (D2O): allyl group, d 4.274
0
0
0
0
0
0
0
0
NMR spectroscopy was carried out at a field strength of
500 MHz (1H) on a Varian Inova instrument. Acquisi-
tions were performed on synthetic compounds in D2O
at 25 ꢁC with a trace of acetone as an internal chemical
shift standard at d = 2.225 ppm, and are reproducible to
within 0.002 ppm. 1H–1H J-couplings are accurate to
within 0.2 Hz. Assignments were made either through
decoupling in 1D experiments or through 2D gCOSY37
correlations. In some cases, a presaturation pulse was
applied to diminish the HOD peak intensity.
(dddd, 1H, J1 a,1 b ꢀ12.9, J1 a,2 5.6, J1 a,3 a 1.3, J1 a,3 b
1.3 Hz, H-10a), 4.149 (dddd, 1H, J1 b,2 6.2, J1 b,3 a 1.2,
J1 b,3 b 1.2 Hz, H-10b), 5.969 (dddd, 1H, J2 ,3 a 17.3,
0 0 0 0
0
0
0
0
J2 ,3 b 10.5 Hz, H-20), 5.350 (dddd, 1H, J3 a,3 b ꢀ1.5 Hz,
H-30a), 5.262 (dddd, 1H, H-30b), glucoside, 5.234 (d,
1H, J1,2 4.3 Hz, H-1), 4.169 (dd, 1H, J2,3 3.4 Hz, H-2),
4.316 (dd, 1H, J3,4 4.6 Hz, H-3), 4.126 (dd, 1H, J4,5
8.0 Hz, H-4), 3.884 (ddd, 1H, J5,6a 3.0, J5,6b 6.4 Hz, H-
5), 3.807 (dd, 1H, J6a,6b ꢀ12.0 Hz, H-6a), 3.657 (dd,
1H, H-6b). MS/MS: negative-ion, Paul trap, precursor
m/z 219 [MꢀH]ꢀ; m/z, relative abundance as % base
peak bracketed: 85 (1.3%), 101 (91%), 113 (4%), 161
(100%, base), 219 (6%, precursor). HRMS: [MꢀH]ꢀ,
for 9, m/z 219.087 for C9H15O6, Calcd 219.086. For
11, allyl b-D-glucofuranoside, 1H NMR (D2O): allyl
0
0
0
0
3.3. Synthesis of allyl glucosides
The allyl glucosides 5, 7, 9, and 11 (Scheme 2) were
prepared by Fischer-type glycosidation. D-Glucose
(100 mg) was dissolved in 10 mL of allyl alcohol by heat-
ing to about 90 ꢁC. The sample was cooled to room
temperature, trifluoroacetic acid (Aldrich, 0.385 mL)
was added to 1 M, and the mixture was heated in a
Teflon-capped tube under argon to 105 ꢁC for 26 h.
The sample was concentrated under high vacuum, and
dissolved in 0.5 mL water-followed by 4.5 mL aceto-
nitrile. The resultant allyl glycosides of all four configu-
rations and ring forms were separated by HPLC, first
in about 10 batches on a column (2.15 · 60 cm) of
glycopak N (Waters) eluted with 95:5 (v/v) acetonitrile
(Mallinckrodt, ChromAR grade)–water at 5 mL/min,
with detection by UV at 200 nm using a Waters 486
variable-wavelength detector. This purified both glucose
allyl pyranosides to homogeneity and separated them
from the furanosides. The furanosides were completely
separated (after concentration) on a column of Shodex
DC-613 (6 · 150 mm) converted to the Cs form,38 eluted
with 95:5 (v/v) acetonitrile–water at 0.6 mL/min, detect-
ing at 200 nm. The allyl a- and b-D-glucopyranosides 5
and 7 have been previously prepared and assigned using
1H MR at 270 MHz.39 Our data were essentially identi-
cal to that reported, except that at 500 MHz, small (1.1–
1.3 Hz) couplings were also observed between the allyl
H-10 and H-30 sets of protons, and approximately a
ꢀ1.5 Hz coupling was observed between the H-30a and
H-30b protons themselves. MS/MS spectra of com-
pounds 5 and 7, negative-ion, Paul trap, precursor m/z
0
0
0
0
group, d 4.209 (dddd, 1H, J1 a,1 b ꢀ12.8, J1 a,2 5.6,
J1 a,3 a 1.1, J1 a,3 b 1.1 Hz, H-10a), 4.080 (dddd, 1H,
0
0
0
0
J1 b,2 6.3, J1 b,3 a 1.0, J1 b,3 b 1.0 Hz, H-10b), 5.947 (dddd,
0
0
0
0
0
0
1H, J2 ,3 a 17.3, J2 ,3 b 10.4 Hz, H-20), 5.344 (dddd, 1H,
0
0
0
0
J3 a, 3 b ꢀ1.5 Hz, H-30a), 5.271 (dddd, 1H, H-30b), gluco-
side, 5.031 (s, 1H, J1,2 1.0 Hz, H-1), 4.176 (br s, 1H, J2,3
1.0 Hz, H-2), 4.262 (d, 1H, J3,4 4.6 Hz, H-3), 4.176
(overlapped with H-2, dd, 1H, J4,5 8.9 Hz, H-4), 3.990
(ddd, 1H, J5,6a 2.7, J5,6b 6.0 Hz, H-5), 3.859 (dd, 1H,
J6a,6b ꢀ12.0 Hz, H-6a), 3.696 (dd, 1H, H-6b). MS/MS:
negative-ion, Paul trap, precursor m/z 219 [MꢀH]ꢀ,
m/z, relative abundance as % base peak bracketed: 85
(1.4%), 101 (100%, base), 113 (7%), 161 (75%), 219
(7%, precursor). HRMS: [MꢀH]ꢀ, for 11, m/z 219.086
for C9H15O6; Calcd 219.086.
0
0
3.4. Synthesis of glucosyl-glycolaldehydes
The glucopyranoside-glycolaldehyde products 6 and 8
(Scheme 2) have been reported,39–42 but in only one
paper was the NMR spectrum of the a-D-glucopyrano-
syl-glycolaldehyde published but not fully assigned (in
Supplementary data of Ref. 40). In the other papers,
neither NMR nor MS data were reported as the sugar-
glycolaldehydes were directly converted to Schiff base
products.39,41,42 The glucosyl-glycolaldehydes 6, 8, 10,
and 12 were prepared directly from the purified allyl
glycosides. Allyl glycosides 5, 7, 9, or 11 (10 mg) were

234
T. T. Fang et al. / Carbohydrate Research 342 (2007) 217–235
dissolved in 2 mL of D2O; O3, produced from pure O2
gas using a commercial ozone generator, was bubbled
through the sample for 6–8 min. The samples were
immediately analyzed by NMR spectroscopy to check
for reaction completion, freeze-dried to remove formal-
dehyde, and then stored in aqueous solution. Ozonolysis
was efficient and clean, with essentially quantitative
reaction yields during this (optimized) time period. Sim-
ilar high reaction yields (albeit under different condi-
tions) have been previously reported.39 Aliquots were
removed for NMR spectroscopy and MS analyses.
NMR spectroscopy demonstrated that the glucosyl-gly-
colaldehyde products were either virtually entirely in the
open-chain form as the hydrate (6 and 12), or were pres-
ent in three forms having different proportions. For 8
and 10, two additional six-membered cyclic forms were
present (8a–c and 10a–c, Schemes 3 and 4), through
hemiacetal formation between the sugar OH-2 and the
glycolaldehyde carbonyl. For 6, a-D-glucopyranosyl-
2-glycolaldehyde, NMR, hydrate: Table 3. MS/MS,
Figure 2A, negative-ion, Paul trap, precursor m/z 221,
[MꢀH]ꢀ, m/z, relative abundance as % base peak brack-
eted: 161 (19%), 131, (100%, base), 113 (18%), 101
(87%), 221 (precursor, 16%) other ions less than 3%:
159, 129, 111, 99. HRMS: [MꢀH]ꢀ, for 6, m/z 221.067
for C8H13O7; Calcd 221.066. For 8, b-D-glucopyran-
osyl-2-glycolaldehyde, three forms; % abundance of
8a:8b:8c 61:17:22, based on integration of H-10 signals.
NMR: Table 3. MS/MS, Figure 2B, negative-ion, Paul
trap, precursor m/z 221, [MꢀH]ꢀ, m/z, relative abun-
dance as % base peak bracketed: 161 (100%, base),
159 (5%), 131 (89%), 101 (6%), 87 (3%), 221 (precursor,
19%). HRMS: [MꢀH]ꢀ, for 8, m/z 221.067 for C8H13O7;
Calcd 221.066. For 10, a-D-glucofuranosyl-2-glycolalde-
solved in 100 lL of H218O (Isotec). The sample was
kept in a sealed vial for 3 weeks. It was lyophilized
and redissolved in 100 lL of H218O for use as a stock
solution, stored frozen, and aliquots were diluted at least
100-fold in MeOH prior to analysis. The 18O-labeled
glucopyranosyl-glycolaldehydes 6 and 8 were prepared
by directly dissolving them in 100 lL of H218O without
the use of acetyl chloride, whereupon the carbonyl
group was replaced with 18O after 24 h. The 18O-labeled
precursor ions were isolated in the gas phase free from
any residual unlabeled precursor ions having a 2 Da
mass difference prior to dissociation.
Acknowledgments
We thank Dr. Robert Murphy for use of his ozone gen-
erator and Dr. Yan Wang for running high-resolution
MS at the University of Illinois at Chicago. This work
was supported in part by a grant from the National Sci-
ence Foundation (CHE-0137986). Funds from the NSF,
NIH, Howard Hughes medical foundation, and the
Keck foundation supported instrumentation used in this
research (NMR and mass spectrometers).
References
1. Zaia, J. Mass Spectrom. Rev. 2004, 23, 161–227.
2. Park, Y. M.; Lebrilla, C. B. Mass Spectrom. Rev. 2005, 24,
232–264.
3. Ashline, D.; Singh, S.; Hanneman, A.; Reinhold, V. Anal.
Chem. 2005, 77, 6250–6262.
4. Zhou, Z.; Ogden, S.; Leary, J. A. J. Org. Chem. 1990, 55,
5444–5446.
hyde, three forms;
% abundance of 10a:10b:10c
11:33:56, based on integration of H-10 signals. NMR:
Table 3. MS/MS, Figure 2C, negative-ion, Paul trap,
precursor m/z 221, [MꢀH]ꢀ, m/z, relative abundance
as % base peak bracketed: 177 (3%), 173 (7%), 161
(81%), 159 (73%), 113 (11%), 101 (100%, base), 99
(81%), 221 (precursor, 15%). HRMS: [MꢀH]ꢀ, for 10,
m/z 221.066 for C8H13O7; Calcd 221.066. For 12, b-D-
glucofuranosyl-2-glycolaldehyde, NMR, hydrate: Table
3. MS/MS, Figure 2D, negative-ion, Paul trap, precur-
sor m/z 221, [MꢀH]ꢀ, m/z, relative abundance as % base
peak bracketed: 161 (100%, base), 101 (10%), 99 (3%),
221 (precursor, 13%). HRMS: [MꢀH]ꢀ, for 12, m/z
221.066 for C8H13O7; Calcd 221.066.
5. Hofmeister, G. E.; Zhou, Z.; Leary, J. A. J. Am. Chem.
Soc. 1991, 113, 5964–5970.
6. Staempfli, A.; Zhou, Z.; Leary, J. A. J. Org. Chem. 1992,
57, 3590–3594.
7. Dongre, A. R.; Wysocki, V. H. Org. Mass Spectrom. 1994,
29, 700–702.
8. Asam, M. R.; Glish, G. L. J. Am. Soc. Mass Spectrom.
1997, 8, 987–995.
9. Polfer, N. C.; Valle, J. J.; Moore, D. T.; Oomens, J.; Eyler,
J. R.; Bendiak, B. Anal. Chem. 2006, 78, 670–679.
10. Ballistreri, A.; Montaudo, G.; Garozzo, D.; Giuffrida, M.;
Impallomeni, G. Rapid Commun. Mass Spectrom. 1989, 3,
302–304.
11. Garozzo, D.; Giuffrida, M.; Impallomeni, G.; Ballistreri,
A.; Montaudo, G. Anal. Chem. 1990, 62, 279–286.
12. Dallinga, J. W.; Heerma, W. Biol. Mass Spectrom. 1991,
20, 215–231.
3.5. 18O-Labeling of the carbonyl of reducing disaccha-
rides and glucosyl-glycolaldehydes
13. Carroll, J. A.; Ngoka, L.; Beggs, C. G.; Lebrilla, C. B.
Anal. Chem. 1993, 65, 1582–1587.
14. Garozzo, D.; Impallomeni, G.; Spina, E.; Green, B. N.;
Hutton, T. Carbohydr. Res. 1991, 221, 253–257.
15. Mulroney, B.; Traeger, J. C.; Stone, B. A. J. Mass
Spectrom. 1995, 30, 1277–1283.
16. Mulroney, B.; Peel, J. B.; Traeger, J. C. J. Mass Spectrom.
1999, 34, 856–871.
Labeling of the carbonyl oxygen of the reducing sugar of
disaccharides was carried out essentially as described.5
Briefly, the disaccharide was labeled with 18O by adding
0.5 lL of acetyl chloride to 1 mg of the disaccharide dis-

T. T. Fang et al. / Carbohydrate Research 342 (2007) 217–235
235
17. Mendonca, S.; Cole, R. B.; Zhu, J.; Cai, Y.; French, A. D.;
Johnson, G. P.; Laine, R. A. J. Am. Soc. Mass Spectrom.
2005, 14, 63–78.
18. Domon, B.; Muller, D. R.; Richter, W. J. Int. J. Mass
Spectrom. Ion Processes 1990, 100, 301–311.
19. Peltier, J. M.; MacLean, D. B.; Szarek, W. A. Rapid
Commun. Mass Spectrom. 1991, 3, 446–449.
20. Seymour, F. R. Meth. Carbohydr. Chem. 1993, 9, 59–85.
21. Bendiak, B.; Fang, T. T. Carbohydr. Res. 2000, 327, 463–
481.
30. Rendina, M. A. C.; Giuffra, S. E.; Galeffi, C.; Monache, E.
M. D. Atti Della Accademia Nazionale Dei Lincei Rendi-
conti-Classe D, Scienze Fisiche-Matematiche e Naturali
1971, 50, 29–32.
31. Domon, B.; Costello, C. E. Glycoconjugate J. 1988, 5, 397–
409.
32. Carroll, J. A.; Willard, D.; Lebrilla, C. B. Anal. Chim.
Acta 1995, 307, 431–447.
33. McClellan, J. E.; Quarmby, S. T.; Yost, R. A. Anal. Chem.
2002, 74, 5799–5806.
22. Lindberg, B.; Lo¨nngren, J. Methods Enzymol. 1978, 50, 3–
33.
23. Gray, G. R. Methods Enzymol. 1990, 193, 573–586.
24. Smith, G.; Leary, J. A. J. Am. Soc. Mass Spectrom. 1996,
7, 953–957.
25. Jiang, Y.; Cole, R. B. J. Am. Soc. Mass Spectrom. 2005,
16, 60–70.
26. Xue, J.; Song, L.; Khaja, S. D.; Locke, R. D.; West, C. M.;
Laine, R. A.; Matta, K. L. Rapid Commun. Mass
Spectrom. 2004, 18, 1947–1955.
27. Ray, T. C.; Smith, A. R. W.; Wait, R.; Hignett, R. C. Eur.
J. Biochem. 1987, 170, 357–361.
28. Freire, C. S. R.; Silvestre, A. J. D.; Silva, A. M. S.; Neto,
C. P.; Domingues, P. Holzforschung 2004, 58, 501–503.
29. Mukhamed’yarova, M. M.; Chumbalov, T. K. Chem. Nat.
Compd. 1977, 13, 243.
34. Bristow, A. W. T.; Nichols, W. F.; Webb, K. S.; Conway,
B. Rapid Commun. Mass Spectrom. 2002, 16, 2374–
2386.
35. Angyal, S. J. Carbohydr. Res. 1979, 77, 37–50.
36. Otter, A.; Bundle, D. R. J. Magn. Reson. Ser. B 1995, 109,
194–201.
37. Hurd, R. E. J. Magn. Reson. 1990, 87, 422–428.
38. Martensson, S.; Levery, S. B.; Fang, T. T.; Bendiak, B.
Eur. J. Biochem. 1998, 258, 603–622.
39. Holme, K. R.; Hall, L. D. Carbohydr. Res. 1992, 225, 291–
306.
40. Bendiak, B.; Salyan, M. E.; Pantoja, M. J. Org. Chem.
1995, 60, 8245–8256.
41. Bernstein, M. A.; Hall, L. D. Carbohydr. Res. 1980, 78,
C1–C-3.
42. Weingarten, S.; Thiem, J. Synlett 2003, 7, 1052–1054.