The stereochemical dependence of unimolecular dissociation of monosaccharide-glycolaldehyde anions in the gas phase: A basis for assignment of the stereochemistry and anomeric configuration of monosaccharides in oligosaccharides by mass spectrometry via a key discriminatory product ion of disaccharide fragmentation, m/z 221

Article abstract of DOI:10.1021/ja0717313

Mass spectrometry of hexose-containing disaccharides often yields product ions of m/z 221 in the negative ion mode. Using a Paul trap, isolation and collision-induced dissociation of the m/z 221 anions yielded mass spectra that easily differentiated their stereochemistry and anomeric configuration, for all 16 stereochemical variants. The ions were shown to be glycopyranosyl- glycolaldehydes through chemical synthesis of their standards. The stereochemistry dramatically affected fragmentation which was dependent on four relative stereochemical arrangements: (1) the relationship between the hydroxyl group at position 2 and the anomeric configuration, (2) a cis relationship of the anomeric position and positions 2 and 3 (1,2,3-cis), (3) a 1,2 trans-2,3 cis relationship, and (4) the relationship between the hydroxyl group at position 4 and the anomeric configuration. After labeling the reducing carbonyl oxygen of a series of disaccharides with ¹⁸O to mass-discriminate between their monosaccharide components, it was demonstrated that m/z 221 anions are comprised of an intact nonreducing sugar glycosidically linked to a 2-carbon aglycon derived from the reducing sugar, irrespective of the linkage position between monosaccharides. This enabled the location of the intact sugar to be assigned to the nonreducing side of a glycosidic linkage. Detailed studies of experimental factors necessary for reproducibility demonstrated that the unique mass spectrum for each m/z 221 anion could be obtained from month-to-month through the use of an internal energy-input calibrant ion that ensured reproducible energy deposition into the ions. The counterparts to these ions for the 2-acetamido-2-deoxyhexoses were m/z 262 anions, and the anomeric configuration and stereochemistry of these anions could also be reproducibly discriminated for N-acetylglucosamine and N-acetylgalactosamine. The fragmentation patterns of m/z 221 anions provide a firm reproducible basis for assignment of sugar stereochemistries in the gas phase.

Full text of DOI:10.1021/ja0717313

Published on Web 07/13/2007
The Stereochemical Dependence of Unimolecular
Dissociation of Monosaccharide-Glycolaldehyde Anions in the
Gas Phase: A Basis for Assignment of the Stereochemistry
and Anomeric Configuration of Monosaccharides in
Oligosaccharides by Mass Spectrometry via a Key
Discriminatory Product Ion of Disaccharide Fragmentation,
m/z 221
Tammy T. Fang and Brad Bendiak*
Contribution from the Department of Cellular and DeVelopmental Biology, and Biomolecular
Structure Program, UniVersity of Colorado at DenVer and Health Sciences Center, Aurora,
Colorado 80045
Received March 12, 2007; E-mail: brad.bendiak@uchsc.edu
Abstract: Mass spectrometry of hexose-containing disaccharides often yields product ions of m/z 221 in
the negative ion mode. Using a Paul trap, isolation and collision-induced dissociation of the m/z 221 anions
yielded mass spectra that easily differentiated their stereochemistry and anomeric configuration, for all 16
stereochemical variants. The ions were shown to be glycopyranosyl-glycolaldehydes through chemical
synthesis of their standards. The stereochemistry dramatically affected fragmentation which was dependent
on four relative stereochemical arrangements: (1) the relationship between the hydroxyl group at position
2 and the anomeric configuration, (2) a cis relationship of the anomeric position and positions 2 and 3
(1,2,3-cis), (3) a 1,2 trans-2,3 cis relationship, and (4) the relationship between the hydroxyl group at position
4 and the anomeric configuration. After labeling the reducing carbonyl oxygen of a series of disaccharides
with 18O to mass-discriminate between their monosaccharide components, it was demonstrated that m/z
221 anions are comprised of an intact nonreducing sugar glycosidically linked to a 2-carbon aglycon derived
from the reducing sugar, irrespective of the linkage position between monosaccharides. This enabled the
location of the intact sugar to be assigned to the nonreducing side of a glycosidic linkage. Detailed studies
of experimental factors necessary for reproducibility demonstrated that the unique mass spectrum for each
m/z 221 anion could be obtained from month-to-month through the use of an internal energy-input calibrant
ion that ensured reproducible energy deposition into the ions. The counterparts to these ions for the
2-acetamido-2-deoxyhexoses were m/z 262 anions, and the anomeric configuration and stereochemistry
of these anions could also be reproducibly discriminated for N-acetylglucosamine and N-acetylgalactosamine.
The fragmentation patterns of m/z 221 anions provide a firm reproducible basis for assignment of sugar
stereochemistries in the gas phase.
Introduction
stereochemistries can be extracted using higher-dimensional
NMR experiments, especially effectively after derivatization
Complex carbohydrates are important in a wide variety of
biological processes and display remarkable structural diversity.1-3
Knowledge of their primary structures is essential for under-
standing their biological functions. NMR spectroscopy has long
been a mainstay in structural carbohydrate chemistry because
the dihedral angle dependence of 3-bond J-couplings of protons
around sugar rings has provided a firm theoretical and practical
foundation to assign their stereochemistry and anomeric con-
figuration.4,5 Even within larger molecules, monosaccharide
with isotags.6-8 Mass spectrometry also has a long history in
analysis of complex carbohydrates.9-12 A strong argument can
be made to spotlight new MS approaches because it furnishes
structural information at 3 to 4 orders of magnitude greater
sensitivity than NMR. However, no firm basis yet exists in mass
spectrometry to enable the stereochemistry and anomeric
(6) Bush, C. A.; Martin-Pastor, M.; Imberty, A. Annu. ReV. Biophys. Biomol.
Struct. 1999, 28, 6500-6512.
(7) Bendiak, B.; Fang, T. T.; Jones, D. N. M. Can. J. Chem. 2002, 80, 1032-
1050.
(1) Kamerling, J. P.; Vliegenthart, J. F. G. Biol. Magn. Reson. 1993, 10, 1-194.
(2) Ovodov, Y. S. Biochemistry (Moscow) 2006, 71, 937-954.
(3) Heredia, A.; Guillen, R.; Jimenez, A.; Fernandez-Bolanos, J. ReV. Esp.
Cienc. Tecnol. Aliment. 1993, 33, 113-131.
(8) Armstrong, G. S.; Mandelshtam, V. A.; Shaka, A. J.; Bendiak, B. J. Magn.
Reson. 2005, 173, 160-168.
(9) Reinhold, V. N.; Carr, S. A. Mass Spectrom. ReV. 1983, 2, 153-221.
(10) Dell, A. AdV. Carbohydr. Chem. Biochem. 1987, 45, 19-72.
(11) Zaia, J. Mass Spectrom. ReV. 2004, 23, 161-227.
(12) Park, Y. M.; Lebrilla, C. B. Mass Spectrom. ReV. 2005, 24, 232-
264.
(4) Lemieux, R. U.; Kullnig, R. K.; Bernstein, H. J.; Schneider, W. G. J. Am.
Chem. Soc. 1957, 79, 1005-1006.
(5) Karplus, M. J. Am. Chem. Soc. 1963, 85, 2870-2871.
9
10.1021/ja0717313 CCC: $37.00 © 2007 American Chemical Society
J. AM. CHEM. SOC. 2007, 129, 9721-9736
9721

A R T I C L E S
Scheme 1
Fang and Bendiak
configuration of monosaccharides to be confidently assigned
when they are derived in the gas phase through successive
isolation/dissociation events from a larger oligosaccharide
molecule. If monosaccharides could be identified with certainty
based on some unique physical property in the gas phase, then
mass spectrometry could be employed with confidence on par
with NMR spectroscopy, at least for establishing the nature of
the stereochemistry of monomers derived from larger structures.
While the stereochemistry of individual monosaccharides can
be determined after hydrolysis/solvolysis through chromato-
graphic means (such as GC-MS),13,14 this does not permit the
order of individual monosaccharides having different stereo-
chemistries to be assigned within a larger molecule (i.e., which
specific monosaccharides are linked to which). The same
argument fundamentally applies to classical “permethylation
analysis”,15-17 where neither the anomeric configuration nor the
order of monosaccharides can be determined following hy-
drolysis or reductive cleavage of an oligosaccharide to its
monomeric constituents.
To establish such an order requires (1) a controlled disas-
sembly of a larger molecule in the gas phase yielding substruc-
tures from different regions of the molecule that are isolable
by MS methods and (2) a means to differentiate substructures
so that the stereochemistry of individual monosaccharides as
gas-phase species can be determined without ambiguity. As part
of the stereochemistry of a monosaccharide is inherent in the
asymmetry at its anomeric carbon, establishing the anomeric
configuration of each sugar is part and parcel of its structural
proof.
Fortunately, ion traps and Fourier transform ion cyclotron
resonance (FTICR) instruments enable multiple isolation/
dissociation steps to be performed18-21 so that issue 1, above,
can be reasonably effectively addressed. At some point,
however, differentiation of these substructures is required in a
way that can confidently establish the stereochemistry and
location of monosaccharides within a larger molecule. The best
ions to divulge this information have not been conspicuous as
yet.
We sought to determine whether disaccharides, the smallest
isolable substructures derived from oligosaccharides that still
contain a glycosidic linkage between two monosaccharides,
would dissociate to yield product ions from which detailed
stereochemical information might be obtained about either of
its monosaccharides in the gas phase. As disaccharides them-
selves represent a large number of isomeric molecules, dif-
ferentiating their structures in the gas phase has not been
generally possible. Studies in the positive22-25 and negative26-32
ion modes have primarily focused on determination of linkage
position in disaccharides; similar information about linkage sites
is possible with pyranosyl-1-enes derived from permethylated
oligosaccharides33,34 or from peracetylated disaccharides.35,36
Fewer studies have been carried out to address the identifica-
tion of anomeric configuration in disaccharides.31,37-40 In the
studies cited, emphasis was placed on mass spectral differentia-
tion between two disaccharides having identical sugars and
linkage positions, varying only in the anomeric configuration
of the nonreducing sugar. Identifying the stereochemistry of one
or the other sugars solely by dissociation of a disaccharide in
the gas phase, without knowing the monosaccharide composition
in adVance, is a rather more challenging endeavor. It not only
implies (1) that key product ions derived from disaccharides
must dissociate in a way unique to their stereochemistry but
also implies (2) that the origins of individual disaccharide
fragments derived by cleavage on either side of the glycosidic
linkage be known and (3) that product ions are not isobaric
mixtures themselves, containing fragments of sugars originating
from either side of the glycosidic linkage.
Bearing these issues in mind, we recently reported that the
m/z 221 anion derived from glucose-containing disaccharides
in the negative ion mode was in fact a glucosyl-glycolaldehyde
anion (Scheme 1) based on its comparison to synthetic R- and
â-glucopyranosyl- and R- and â-glucofuranosyl-glycolalde-
hydes.41 Both the anomeric configuration and ring forms of the
m/z 221anions could be clearly assigned by their MS dissociation
patterns. It was also demonstrated that the intact monosaccharide
(22) Zhou, Z.; Ogden, S.; Leary, J. A. J. Org. Chem. 1990, 55, 5444-5446.
(23) Hofmeister, G. E.; Zhou, Z.; Leary, J. A. J. Am. Chem. Soc. 1991, 113,
5964-5970.
(24) Dongre´, A. R.; Wysocki, V. H. Org. Mass Spectrom. 1994, 29, 700-702.
(25) Asam, M. R.; Glish, G. L. J. Am. Soc. Mass Spectrom. 1997, 8, 987-995.
(26) Ballistreri, A.; Montaudo, G.; Garozzo, D.; Giuffrida, M.; Impallomeni,
G. Rapid Commun. Mass Spectrom. 1989, 3, 302-304.
(27) Garozzo, D.; Giuffrida, M.; Impallomeni, G.; Ballistreri, A.; Montaudo,
G. Anal. Chem. 1990, 62, 279-286.
(28) Dallinga, J. W.; Heerma, W. Biol. Mass Spectrom. 1991, 20, 215-231.
(29) Carroll, J. A.; Ngoka, L.; Beggs, C. G.; Lebrilla, C. B. Anal. Chem. 1993,
65, 1582-1587.
(30) Garozzo, D.; Impallomeni, G.; Spina, E.; Green, B. N.; Hutton, T.
Carbohydr. Res. 1991, 221, 253-257.
(31) Mulroney, B.; Traeger, J. C.; Stone, B. A. J. Mass Spectrom. 1995, 30,
1277-1283.
(32) Mulroney, B.; Peel, J. B.; Traeger, J. C. J. Mass Spectrom. 1999, 34, 856-
871.
(33) Ashline, D.; Singh, S.; Hanneman, A.; Reinhold, V. Anal. Chem. 2005,
77, 6250-6252.
(34) Zhang, H.; Singh, S.; Reinhold, V. N Anal. Chem. 2005, 77, 6363-6270.
(35) Domon, B.; Mu¨ller, D. R.; Richter, W. J. Int. J. Mass Spectrom. Ion
Processes 1990, 100, 301-311.
(13) Seymour, F. Methods Carbohydr. Chem. 1993, 9, 59-85.
(14) Bendiak, B.; Fang, T. T. Carbohydr. Res. 2000, 327, 463-481.
(15) Lindberg, B.; Lo¨nngren, J. Methods Enzymol. 1978, 50, 3-33.
(16) Hakomori, S. J. Biochem. 1964, 55, 205-208.
(17) Gray, G. R. Methods Enzymol. 1990, 193, 573-586.
(18) Paul, W.; Reinhard, H. P.; von Zahn, U. Z. Phys. 1958, 152, 143-182.
(19) Hao, C.; March, R. E. Int. J. Mass Spectrom. 2001, 212, 337-357.
(20) Marshall, A. G. Int. J. Mass Spectrom. 2000, 200, 331-356.
(21) Blaum, K. Phys. Rep. 2006, 425, 1-78.
(36) Peltier, J. M.; MacLean, D. B.; Szarek, W. A. Rapid Commun. Mass
Spectrom. 1991, 5, 446-449.
(37) Smith, G.; Leary, J. A. J. Am. Soc. Mass Spectrom. 1996, 7, 953-957.
(38) Jiang, Y. J.; Cole, R. B. J. Am. Soc. Mass Spectrom. 2005, 16, 60-70.
(39) 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.
(40) Polfer, N. C.; Valle, J. J.; Moore, D. T.; Oomens, J.; Eyler, J. R.; Bendiak,
B. Anal. Chem. 2006, 78, 670-679.
(41) Fang, T. T.; Zirrolli, J.; Bendiak, B. Carbohydr. Res. 2007, 342, 217-235.
9
9722 J. AM. CHEM. SOC. VOL. 129, NO. 31, 2007

Monosaccharide Stereochemistry by Mass Spectrometry
A R T I C L E S
of these ions was solely derived from the nonreducing sugar of
reducing disaccharides based on 18O-labeling of the carbonyl
group of the reducing sugar to mass-discriminate between the
origins of product ions derived from either side of the glycosidic
linkage. Here we report that gas-phase dissociation patterns can
be used to discriminate between all 16 stereochemical and
anomeric variants of D-aldohexopyranosyl-glycolaldehydes as
m/z 221 anions. The synthetic compounds (222 Da, [M - H]-
m/z 221) dissociate identically in the gas phase to the m/z 221
ions isolated from various disaccharides upon fragmentation.
Two common 2-acetamido-2-deoxy sugars, N-acetylglucosamine
(GlcNAc) and N-acetylgalactosamine (GalNAc), were also
differentiated in the gas phase as their R- and â-D-glycopy-
ranosyl-glycolaldehyde anions, enabling their anomeric config-
uration and stereochemistry to be assigned.
tion eigenstates42,43 (Figure 1E). Either the open-chain or cyclic
forms as shown may also exist in different charge variants due
to proton exchange among hydroxyl groups. Sugar ring opening
may be a fundamental limitation in identifying the stereochem-
istries and anomeric configurations of monosaccharides as (m/z
179) anions, at least using collision-induced dissociation (CID).
Similarly, we have been unable to differentiate monosaccharide
anomeric configurations and stereochemistries in the positive
ion mode as adducts of either sodium or lithium ions in the gas
phase, where the monosaccharide product ions were derived
by dissociation of disaccharides (data not shown).
Dissociation of Glycopyranosyl-Glycoaldehydes (m/z 221
Anions) in the Negative Ion Mode Discriminates Both
Anomeric Configuration and Stereochemistry of the Ions
for All 16 Possible Structural Variants. Another key ion
frequently observed during dissociation of disaccharides in the
negative ion mode is an m/z 221 anion. We demonstrated earlier
that these ions, derived from 2,4-, or 6-linked disaccharides
containing only glucose, were comprised of the intact nonre-
ducing sugar glycosidically linked to a glycolaldehyde molecule.
Dissociation of these ions in a Paul trap could be used to
distinguish between the anomeric configurations and ring forms
of glucosyl-glycoaldehydes.41 All 16 stereochemical variants of
glycopyranosyl-glycoaldehydes were synthesized as standards
in order to determine their dissociation patterns in the negative
ion mode (Chart 1). In Figure 2, the dissociation patterns of
the R- and â-D-glucopyranosyl- and galactopyranosyl-glycolal-
dehydes under identical dissociation conditions are shown
(compounds 1-4, [M - H]- precursor anion m/z 221, abbrevi-
ated as Hex-GA anions). The mass spectra yielded patterns that
clearly differentiated both the anomeric configuration and
stereochemistry of the anions. It is worth noting that the ratios
of product ions and even the presence of some ions were
markedly different between the stereoisomers. For example, in
comparing the R- and â-D-glucopyranosyl-glycolaldehyde an-
ions, (Figure 2, part A to part B) there was a major decrease in
the intensity of the m/z 101 product ion and a major increase in
the abundance of the m/z product 161 ion which was solely
dependent on the stereochemistry at the anomeric carbon.
Likewise, there were major fragmentation differences in com-
paring the R- and â-D-galactopyranosyl-glycolaldehyde anions
(Figure 2, parts C and D). Similarly, only by changing the
stereochemistry at the 4-position (Figure 2, proceeding from
panel B to panel D, for example), the dissociation was markedly
affected. Product ions at m/z 161, 101, and 113 were signifi-
cantly diminished in abundance, m/z 131 becoming the sole
dominant product ion with â-D-galactopyranosyl-glycolaldehyde
(Figure 2D). Reproducibility of the spectrum as shown in Figure
2A was similar to other glycosylglycolaldehydes and is indicated
with error bars that represent the standard deviation of 20
experiments performed over a 2 month period. The reproduc-
ibility has been the topic of detailed experimental studies
previously described41 and is briefly discussed later in this
article.
Results and Discussion
Monosaccharide Differentiation in the Gas Phase: The
Problem. A classic problem in mass spectrometry of oligosac-
charides is identification of the stereochemistry and anomeric
configuration of its monosaccharide constituents, when these
are derived by dissociation of a larger precursor ion in the gas
phase. To illustrate this general issue, we studied the dissociation
of disaccharides specifically 18O-labeled on the carbonyl group
of the reducing sugar ([M - H]- m/z 343) so that individual
monosaccharide product ions could be mass-discriminated.
Disaccharides which contained either a nonreducing glucose (R-
D-Glcp-(1-6)-D-Glc and â-D-Glcp-(1-6)-D-Glc) or a nonre-
ducing galactose (R-D-Galp-(1-6)-D-Glc and â-D-Galp-(1-6)-
D-Gal) in both anomeric configurations were 18O-labeled on the
carbonyl position of the reducing sugar. In a Paul trap, an ion
derived from disaccharides in the negative ion mode having an
m/z of 179 [M - H]- was invariably observed, representing
the nonreducing monosaccharide containing the original gly-
cosidic oxygen. Interestingly, anions having an m/z 181 were
not observed, which indicated a highly selective cleavage on
one side of the glycosidic linkage, at least for these reducing
disaccharides. Dissociation spectra of the m/z 179 product ions
isolated from the above disaccharides are shown in Figure 1,
where these represent MS3 spectra. Note that monosaccharide
anions (m/z 179) derived from disaccharides having the same
nonreducing monosaccharide in either an R or â linkage yielded
essentially identical mass spectra (compare, for example, both
Figure 1, part A to part B and part C to part D). Similarly, the
product ion ratios were only slightly different in comparing the
glucose and galactose anions (compare Figure 1, part A to part
C or part B to part D). On the basis of the fragmentation of
these m/z 179 anions, it can be concluded that (1) it is impossible
to assign the anomeric configuration of the nonreducing
monosaccharide using CID in a Paul trap and (2) it is also rather
difficult to assign the stereochemistry of either of these sugars.
While there was a minor difference observed in product ion
ratios between the gluco or galacto product ions in the Paul
trap, should either monosaccharide be derived from a completely
unknown disaccharide, one would be very hard pressed to assign
their stereochemistry with confidence. These studies support the
hypothesis that monosaccharide m/z 179 ions may interconvert
from cyclic structures to acyclic variants during acquisition of
enough internal vibrational energy to reach threshold dissocia-
In a similar vein, the negative ion spectra of the 12 other
aldohexopyranosyl-glycolaldehydes 5-16 (Chart 1) showed
unique and reproducible differences in ion ratios easily permit-
(42) Dallinga, J. W.; Heerma, W. Biomed. EnViron. Mass Spectrom. 1989, 18,
363-372.
(43) Carroll, J. A.; Willard, D.; Lebrilla, C. B. Anal. Chim. Acta 1995, 307,
431-447.
9
J. AM. CHEM. SOC. VOL. 129, NO. 31, 2007 9723

A R T I C L E S
Fang and Bendiak
Figure 1. Mass spectrometry of monosaccharide product ions derived from disaccharides 18O-labeled on the carbonyl group of the reducing sugar, illustrating
the difficulties in differentiating anomeric configuration and stereochemistry in gas-phase analysis of monosaccharide product anions. A series of four
disaccharides were 18O-labeled on the carbonyl group of the reducing sugar, isolated in the gas phase as [M - H]- m/z 343 precursor anions, dissociated
in the Paul trap, and then the m/z 179 product ion derived from the nonreducing monosaccharide was isolated and subjected to another dissociation (MS3).
Shown are the MS3 spectra of m/z 179 product anions derived from (A) R-D-Glcp-(1-6)-D-Glc, (B) â-D-Glcp-(1-6)-D-Glc, (C) R-D-Galp-(1-6)-D-Glc, and
(D) â-D-Galp-(1-6)-D-Gal. Structures of the m/z 343 precursor ions and m/z 179 product ions for each gas-phase dissociation sequence are shown above
individual panels. In panel E is shown the hypothetical sugar ring opening that may occur during acquisition of enough internal vibrational energy needed
for gas-phase dissociation.
ting their anomeric configurations and stereochemistries to be
assigned (Figures 3-5). A detailed examination of their
fragmentation patterns has enabled some empirical rules about
their dissociation to be extracted: (1) The abundance of the
m/z 101 anion was strongly dependent on the stereochemical
relationship between positions 1 and 2. If these positions
contained the aglycon and the 2-hydroxyl group in a cis
relationship, the m/z 101 anion was favored. Hence, the
R-gluco-, galacto-, allo-, and gulopyranosides strongly favored
the 101 anion over their respective â anomers, but the â-manno-,
altro-, talo-, and idopyranosides favored the 101 anion formation
over their respective R anomers.44 (2) Similarly, the m/z 113
9
9724 J. AM. CHEM. SOC. VOL. 129, NO. 31, 2007

Monosaccharide Stereochemistry by Mass Spectrometry
A R T I C L E S
Chart 1
anion was always more abundant when the aglycon and
2-hydroxyl group were found in a cis relationship for all
anomeric pairs.44 In cases where the 3-hydroxyl group was also
cis to the 2-hydroxyl (1,2,3-cis), the 113 anion was even more
abundant. This was observed in comparing the 3-epimeric pairs
R-allo- and R-gluco-, R-gulo- and R-galacto-, â-talo- and â-ido-,
and to a lesser extent, â-manno- and â-altropyranosyl-glyco-
laldehydes. (3) The stereochemistry of the 4-position dramati-
cally affected the ratio of the m/z 131/161 ion abundances, which
was generally favored only for the â-anomers. Thus, in
proceeding from a 4-equatorial hydroxyl to the 4-axial epimer
for â-pyranosides, the ratio of 131/161 was markedly increased,
as seen in comparing the 4-epimeric pairs â-gluco- and
â-galacto-, â-allo- and â-gulo, â-manno- and â-talo-, and â-altro
and â-idopyranosyl-glycolaldehydes. For the R-pyranosides, this
was not observed. (4) A preponderance of the m/z 159 anion
relative to m/z 161 was only found in four cases, where the
stereochemistries of positions 1, 2, and 3 were uniquely related.
When positions 1 and 2 were trans but positions 2 and 3 were
cis, the m/z 159/161 ratio was dramatically increased. This
arrangement only occurs for the â-allo and â-gulo pair of
isomers and for the R-manno and R-talo pair.
Finally, some ions of moderate abundance appeared to be
nearly characteristic for one or at most a few stereochemical
variants. The m/z 203 anion, for example, was only found in
significant abundance for â-D-idopyranosyl-glycolaldehyde, and
the m/z 75 anion only with the R-talo variant. The m/z 143 anion
was similarly found above values of about 1% only in the R-Gul
and â-Ido variants, and the m/z 97 anion in the R-talo and to a
lesser extent, R-gulo and â-ido isomers. Thus, while some
dissociation channels were common to many glycopyranosyl-
glycolaldehydes, others were unique. It is worth noting that these
rules are only empirical propensities as for every stereochemistry
there were competing alternative dissociation pathways that gave
rise to other ions. We have considered potential mechanisms
regarding the stereochemical dependence of these dissociation
pathways but have specifically refrained from discussing them
in this article for three reasons. First, the true origin(s) of each
product ion are as yet uncertain and will require detailed studies
with isotopomers. Second, little is known and currently little is
possible to experimentally verify concerning the conformations
of high-energy eigenstates of these ions as they approach and/
or exceed their dissociation thresholds. Such measurements may
be possible in the future via ion spectroscopy in different
electromagnetic wavelength ranges.45,46 Third, some of these
molecules may exist as cyclic hemiacetal variants in the gas
phase, as described later in this report.
Dissociation of 2-Acetamido-2-deoxyglycopyranosyl-Gly-
coaldehydes (m/z 262 Anions) in the Negative Ion Mode
Discriminates Both Anomeric Configuration and Stereo-
chemistry of the Ions for All Four GlcNAc and GalNAc
Variants. In Figure 6 is shown the mass spectral differentiation
of the anomeric configuration and stereochemistry of two
HexNAc-glycolaldehydes for N-acetylhexosamines commonly
found in mammalian systems. Data are from synthetic molecules
with precursor ions ([M - H]- m/z 262) indicated above each
spectrum. Insets show an expansion of the fragment ion
intensities (to the multiplicative values indicated) for the m/z
region from 70 to 200, where the m/z 112 abundance is off the
vertical scale for three of the four insets. It is important to note
that the spectra shown in the insets were accumulated with
enough scans so that noise levels were less than 0.1% of the
base product ion peak. Under identical dissociation conditions,
(44) These same observations applied to idopyranosyl-glycolaldehydes (structures
15 and 16), even though the R form was found by NMR to be preferentially
in the 1C4 conformation and the â form was found in the 4C1 conformation.
(45) Duncan, M. A. Int. J. Mass Spectrom. 2000, 200, 545-569.
(46) Dunbar, R. C. Int. J. Mass Spectrom. 2000, 200, 571-589.
9
J. AM. CHEM. SOC. VOL. 129, NO. 31, 2007 9725

A R T I C L E S
Fang and Bendiak
Figure 2. Differentiation of synthetic glucopyranosyl- and galactopyranosyl-glycolaldehyde isomers having different anomeric configurations by MS/MS
in the negative ion mode. The [M - H]-, m/z 221 precursor anions were isolated, and MS/MS spectra were recorded following dissociation in an LCQ Paul
trap. Spectra were recorded from (A) R-D-glucopyranosyl-2-glycolaldehyde 1, (B) â-D-glucopyranosyl-2-glycolaldehyde 2, (C) R-D-galactopyranosyl-2-
glycolaldehyde 3, and (D) â-D-galactopyranosyl-2-glycolaldehyde 4. Error bars in panel A illustrate the standard deviation of the indicated major ions
calculated using the intensity of m/z 131 as exactly 100%, for 20 spectra accumulated over a 2 month period.
the HexNAc-glycolaldehydes all showed a base product ion at
m/z 202, and all showed an m/z 112 product ion of varying
relative abundance. Key observations were that the â-D-
GalpNAc- and â-D-GlcpNAc-glycolaldehydes (Figure 6, insets
to panels B and D) showed an m/z 100 ion of low abundance
that was essentially negligible in spectra of their respective
R-anomers (Figure 6, insets to panels A and C). Another
difference was the respective abundance of their m/z 140/142
product ions. In the R-linked HexNAc-glycolaldehydes, the m/z
140 ion was of greater abundance (insets to Figure 6, panels A
and C), whereas in the â-glycosides, the m/z 142 ion was of
much greater intensity (insets to Figure 6, panels B and D). In
addition the R-linked compounds showed m/z 128 anions that
were of negligible abundance in their respective â-anomers. Two
ions having m/z 184 and 196 were found in the GalpNAc-
glycolaldehydes that were of negligible abundance in GlcpNAc-
glycolaldehydes. The â-GalpNAc- and â-GlcpNAc-glycolalde-
hydes were also easily distinguished by a major difference in
their m/z 112/100 product ion ratios (compare Figure 6, panels
B and D), and the R-GalpNAc and R-GlcpNAc-glycolaldehydes
9
9726 J. AM. CHEM. SOC. VOL. 129, NO. 31, 2007

Monosaccharide Stereochemistry by Mass Spectrometry
A R T I C L E S
Figure 3. Discrimination of synthetic mannopyranosyl- and allopyranosyl-glycolaldehyde isomers having different anomeric configurations by MS/MS in
the negative ion mode. The [M - H]-, m/z 221 precursor anions were isolated, and MS/MS spectra were recorded in a Paul trap. Spectra were recorded from
(A) R-D-mannopyranosyl-2-glycolaldehyde 5, (B) â-D-mannopyranosyl-2-glycolaldehyde 6, (C) R-D-allopyranosyl-2-glycolaldehyde 7, and (D) â-D-
allopyranosyl-2-glycolaldehyde 8.
were easily discriminated based on a complete lack of the m/z
196 anion in the spectrum of the GlcpNAc form as well as a
difference in abundance of their m/z 112 ions (compare Figure
6 panels A and C). The relative abundances of these ions as
compared to the base product ion peak at 100% were repro-
ducibly within a few percent under defined conditions; therefore,
the spectra with even moderate signal/noise easily and repro-
ducibly differentiated between the four isomeric HexNAc-
glycolaldehydes in the gas phase.
etamido-2-deoxy-glycopyranosyl-Glycolaldehdyes Derived
from the Nonreducing Position of Disaccharides by Dis-
sociation in the Gas Phase. A true test as to whether the
dissociation spectra of synthetic glycosyl-glycolaldehyde anions
could serve as gas-phase standards to assign the stereochemistry
and anomeric configuration of monosaccharides derived from
larger carbohydrates was to compare their mass spectra to those
of m/z 221 anions derived by dissociation of disaccharides and
observe whether one stereochemistry could be clearly identified
from among all 16 possible stereochemical variants. Disaccha-
ride precursor ions were fragmented, and their respective m/z
Identification of the Stereochemistry and Anomeric Con-
figurations of Glycopyranosyl-Glycolaldehydes and 2-Ac-
9
J. AM. CHEM. SOC. VOL. 129, NO. 31, 2007 9727

A R T I C L E S
Fang and Bendiak
Figure 4. Differentiation of synthetic altropyranosyl- and gulopyranosyl-glycolaldehyde isomers having different anomeric configurations by MS/MS in the
negative ion mode. The [M - H]-, m/z 221 precursor anions were isolated, and MS/MS spectra were recorded in a Paul trap. Spectra were recorded from
(A) R-D-altropyranosyl-2-glycolaldehyde 9, (B) â-D-altropyranosyl-2-glycolaldehyde 10, (C) R-D-gulopyranosyl-2-glycolaldehyde 11, and (D) â-D-gulopyranosyl-
2-glycolaldehyde 12.
221 product ions were isolated and further dissociated in an
MS3 experiment (Figures 7 and 8). In the case of 4-linked
structures, the m/z 221 anions were obtained in higher abundance
through in-source fragmentation; hence, the spectra shown are
MS/MS spectra. Although the m/z 221 anions could be isolated
after MS/MS of 4-linked disaccharides in the gas phase and
further fragmented in MS3 experiments, the spectra required
many more scans to obtain good signal/noise as compared to
some of the spectra shown in Figures 7 and 8. This was simply
due to the low abundance (1-5%) of the m/z 221 ion generated
in the Paul trap from 4-linked disaccharides through resonantly
excited CID. In Figure 7, the spectra of m/z 221 anions derived
from four different disaccharides are shown. Disaccharides
comprised of two hexoses were 18O-labeled on the carbonyl
group of the reducing sugar so that product ions derived from
either side of the glycosidic linkage could be mass-discriminated.
In the case of structures having a hexose and a HexNAc, there
was no need to label them as components derived from either
side of the glycosidic linkage already have mass differences.
In Figure 7A, in-source fragmentation of a disaccharide
having a nonreducing â-D-mannopyranoside residue (â-D-Manp-
(1-4)-D-GlcNAc) yielded an m/z 221 anion that was isolated
9
9728 J. AM. CHEM. SOC. VOL. 129, NO. 31, 2007

Monosaccharide Stereochemistry by Mass Spectrometry
A R T I C L E S
Figure 5. Discrimination of synthetic talopyranosyl- and idopyranosyl-glycolaldehyde isomers having different anomeric configurations by MS/MS in the
negative ion mode. The [M - H]-, m/z 221 precursor anions were isolated, and MS/MS spectra were recorded in a Paul trap. Spectra were recorded from
(A) R-D-talopyranosyl-2-glycolaldehyde 13, (B) â-D-talopyranosyl-2-glycolaldehyde 14, (C) R-D-idopyranosyl-2-glycolaldehyde 15, and (D) â-D-idopyranosyl-
2-glycolaldehyde 16.
and further dissociated in the Paul trap to furnish a characteristic
mass spectrum. Comparison of this mass spectrum to all 16
variants of the synthetic glycopyranosyl-glycolaldehyde anions
shown in Figures 2-5 clearly indicated a match with just one
of them (Figure 3B), the â-D-mannopyranosyl-glycolaldehyde
6. In Figure 7B, the MS3 spectrum of an ion having m/z 221
derived from a disaccharide anion (â-D-Galp-(1-6)-D-GlcNAc
is shown. Comparison of this spectrum to all 16 variants of the
synthetic glycosyl-glycolaldehydes (Figures 2-5) again clearly
showed a match to just one of them (Figure 2D), the â-D-
galactopyranosyl-glycolaldehyde 4. The structures shown in
Figure 7, parts C and D, were disaccharides comprised of only
hexoses, which were 18O-labeled on the carbonyl of the reducing
sugar. For both structures, m/z 221 ions were isolated (no m/z
223 ions were observed).47 Again, the fragmentation patterns
of these ions matched their respective glycosyl-glycolaldehyde
anion spectra. For Figure 7C, the spectrum of the m/z 221 anion
derived from R-D-Manp-(1-6)-D-Man matched that of synthetic
(47) 18O-labeling was carried out with disaccharides solely comprised of hexoses
so that an anion comprised of a nonreducing sugar attached to a 2-carbon
fragment derived from the reducing sugar (m/z 221) could be differentiated
from a potential anion containing a reducing sugar attached to a 2-carbon
fragment derived from the nonreducing sugar (m/z 223).
9
J. AM. CHEM. SOC. VOL. 129, NO. 31, 2007 9729

A R T I C L E S
Fang and Bendiak
Figure 6. Differentiation of synthetic glycosyl-glycolaldehyde isomers of the N-acetylhexosamines GlcNAc and GalNAc, having different anomeric
configurations by MS/MS in the negative ion mode. The [M - H]-, m/z 262 precursor anions were isolated, and MS/MS spectra were recorded in a Paul
trap. Spectra were recorded from (A) (2-acetamido-2-deoxy-R-D-glucopyranosyl)-2-glycolaldehyde 17, (B) (2-acetamido-2-deoxy-â-D-glucopyranosyl)-2-
glycolaldehyde 18, (C) (2-acetamido-2-deoxy-R-D-galactopyranosyl)-2-glycolaldehyde 19, and (D) (2-acetamido-2-deoxy-â-D-galactopyranosyl)-2-glycola-
ldehyde 20. Insets illustrate vertical expansions of the m/z 70-200 regions, to the multiplicative extent indicated in each, so the m/z 112 product ion is
off-scale for panels A, C, and D. Shown in insets are multiple-scan spectra acquired with high signal/noise.
R-D-mannopyranosyl-glycolaldehyde 5 (Figure 3A), whereas in
Figure 7D, derived from R-D-Galp-(1-6)-D-Glc, the spectrum
of the m/z 221 anion matched that of synthetic R-D-galactopy-
ranosyl-glycolaldehyde 3 (Figure 2C). Similar results are shown
in Figure 8, parts A and B, where the mass spectra of m/z 221
anions derived from the 18O-labeled disaccharides â-D-Galp-
(1-6)-D-Gal and â-D-Galp-(1-4)-D-Glc both matched the
spectrum of the synthetic â-D-galactopyranosyl-glycolaldehyde
4 (Figure 2D). The spectra of the m/z 221 anions in Figure 7
and Figure 8, parts A and B, combined with 18O-labeling of
their precursor disaccharides, yielded results in full accordance
with data from a set of m/z 221 ions derived from six different
glucose-containing disaccharides previously reported.41 That is,
the m/z 221 anions could be used to assign the anomeric
9
9730 J. AM. CHEM. SOC. VOL. 129, NO. 31, 2007

Monosaccharide Stereochemistry by Mass Spectrometry
A R T I C L E S
Figure 7. Dissociation spectra of m/z 221 product ions derived from various disaccharide precursor ions in the negative ion mode. Shown are the dissociation
spectra of the m/z 221 product ion derived from (A) â-D-Manp-(1-4)-D-GlcNAc (m/z 382 f m/z 221f spectrum), (B) â-D-Galp-(1-6)-D-GlcNAc (m/z 382
f m/z 221f spectrum), (C) R-D-Manp-(1-6)-D-Man, 18O-labeled on the carbonyl of the reducing Man (m/z 343 f m/z 221 f spectrum), and (D) R-D-
Galp-(1-6)-D-Glc, 18O-labeled on the carbonyl of the reducing Glc (m/z 343 f m/z 221 f spectrum). Structures of the precursor and product ions are
indicated above each mass spectrum. The open-chain form of the reducing sugar was drawn solely to illustrate the fragmentation position; no implication
regarding the relative abundance of open-chain or cyclic form(s) of the reducing sugar in the gas phase is intended, nor are any inferences drawn concerning
dissociation mechanisms.
configuration and stereochemistry of the nonreducing aldohex-
opyranoside in these disaccharides, by comparison to the
complete series of aldohexopyranosyl-glycolaldehyde standards.
Disaccharides having a nonreducing HexNAc residue frag-
mented in the negative ion mode to yield m/z 262 product ions,
with no m/z 221 anions. Spectra of the m/z 262 anions derived
from these disaccharides also matched those of synthetic
standards. As illustrated in Figure 8, parts C and D, dissociation
spectra of the m/z 262 anions isolated in the gas phase from
â-D-GlcpNAc-(1-2)-D-Man and â-D-GalpNAc-(1-4)-D-Gal
matched the negative ion spectra of synthetic â-D-GlcpNAc-
glycolaldehyde 18 (Figure 6B) and â-D-GalpNAc-glycolalde-
hyde 20 (Figure 6D), respectively. Admittedly all 16 stereo-
chemical variants of the 2-acetamido-2-deoxyhexopyranosyl-
glycolaldehydes would need to be prepared to unambiguously
assign their stereochemistry, but it is relevant that the dissocia-
tion of each isolated m/z 262 anion matches those of the
appropriate synthetic standards and enables the anomeric
configuration and stereochemistry of these two 2-acetamido-2-
deoxyhexoses to be differentiated.
9
J. AM. CHEM. SOC. VOL. 129, NO. 31, 2007 9731

A R T I C L E S
Fang and Bendiak
Figure 8. Dissociation spectra of m/z 221 and m/z 262 product ions derived from various disaccharide precursor ions in the negative ion mode. Shown are
the dissociation spectra of the m/z 221 or m/z 262 product ion derived from (A) â-D-Galp-(1-6)-D-Gal, 18O-labeled on the carbonyl of the reducing Gal (m/z
343 f m/z 221f spectrum), (B) â-D-Galp-(1-4)-D-Glc 18O-labeled on the carbonyl of the reducing Glc (m/z 343 f m/z 221f spectrum), (C) â-D-GlcpNAc-
(1-2)-D-Man (m/z 382 f m/z 262 f spectrum), and (D) â-D-GalpNAc-(1-4)-D-Gal, (m/z 382 f m/z 262 f spectrum). Structures of the precursor and
product ions are indicated above each mass spectrum. The open-chain form of the reducing sugar was drawn solely to indicate the fragmentation position(s);
no implication regarding the relative abundance of open-chain or cyclic form(s) of the reducing sugar in the gas phase is intended, nor are any inferences
drawn concerning dissociation mechanisms.
It is also important to emphasize that these ions were always
comprised of an intact nonreducing sugar with the 2-carbon
aglycon derived from the reducing sugar, as previously estab-
lished using glucose-containing disaccharides.41 In considering
the potential of isolating various disaccharides through multistep
disassembly from larger oligosaccharides in the gas phase, it is
highly relevant to also be able to assign the location of an
experimentally determined monosaccharide stereochemistry to
one specific side of the glycosidic linkage in disaccharides.
Assignment of the Linkage Position of Disaccharides
Based on Dissociation Patterns in the Negative Ion Mode.
It has been possible to assign the linkage position of disaccha-
rides in the negative ion mode based on previous studies carried
out using sector instruments,26-28 FTICR instruments,29 in-
9
9732 J. AM. CHEM. SOC. VOL. 129, NO. 31, 2007

Monosaccharide Stereochemistry by Mass Spectrometry
A R T I C L E S
source decomposition,30,31 ion traps,41 and triple-quadrupole
instruments.41 Our results with the hexose-containing disaccha-
rides reported in Figures 7 and 8 showed the same fragmentation
products for different linkages in approximately the ratios
previously reported41 and enabled the linkage position between
two sugars in disaccharides to be assigned (data not shown).
With disaccharides having a nonreducing HexNAc and a
reducing hexose, dissociation of the hexose occurred to give
fragmentation patterns similar to hexose-containing disaccha-
rides with product ions having a mass increment of +41 due to
the mass difference of the acetamido group as compared to a
hydroxyl group for the nonreducing sugar. With disaccharides
having the reducing HexNAc, linkages could also be ascertained,
but some of the product ions contained the +41 mass increment
due to the presence of the acetamido group. This data has not
been presented in detail herein as it was confirmatory to other
reports and the key focus of this investigation was the dissocia-
tion properties of the glycosyl-glycoaldehyde anions. It is
important to point out that the m/z 221 anions (for hexose-
containing disaccharides) were observed as product ions from
fragmentation of 2-, 3-, 4-, and 6-linked disaccharides, although
for 3-linked species, their abundance was less than 1% of the
base product ion peak when dissociated by resonant excitation
in a Paul trap. They were obtained in higher abundance for
3-linked disaccharides (1-2%) using in-source fragmentation
but were still too low in abundance for isolation and subsequent
dissociation of the m/z 221 anion. They have been obtained in
higher abundance using in-source fragmentation with triple-
quadrupole instruments and under higher-energy collision
conditions.31,41
Mass Spectral Reproducibility. In a previous report,41 issues
relevant to reproducible acquisition of mass spectra in a Paul
trap were described in detail, so they will only be briefly
described here. Spectral reproducibility is crucial for identifica-
tion of ions as “spectral fingerprints” and is particularly relevant
for discrimination between isobaric variants having identical
elemental compositions. Shown in Figure 2A is the variability
of the spectrum of R-D-glucopyranosyl-glycolaldehyde 1, where
error bars show a standard deviation of peak heights over the
collection of 20 independent spectra obtained over a 2 month
period. Spectral reproducibility could be maintained from day-
to-day and over several months by controlling four key
parameters: (1) Isolation width needed to be kept constant and
of moderate width (3 Finnigan “amu” units) purposefully to
improve reproducibility. (2) Frequent calibration of the reso-
nance excitation frequency was essential. Normally, after
calibration of the instrument, this frequency drifts and can do
so in the LCQ at a rate different than that of the mass
measurement frequencies. If this drifts by 0.2-0.5 amu, it will
slightly affect the ratios of product ions and either requires
recalibration or can be finely adjusted by slightly shifting the
excitation resonance frequency to manually optimize the total
product ion current. (3) The most important issue involves
defining the energy input into an ion, where the ratio of the
most abundant product ion (base) peak to the remaining
precursor ion was chosen as the most reliable criterion for
reproducible energy deposition into ions. We selected synthetic
R-D-glucopyranosyl-glycolaldehyde as an “internal energy-input
tuning ion” and routine standard, in a similar fashion to Bristow
et al.,48 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. Thus, over a period of months, the
collision energy needs to be slightly adjusted to attain the same
ratio of base product ion/precursor ion. (4) The number of ions
in the trap needed to be approximately the same for comparison
of a standard to an unknown (within an order of magnitude) to
have best reproducibility. This was tested systematically from
5 × 103 to 3 × 106 ions in the trap; the overall number of ions
in the trap slightly affected dissociation patterns, regardless of
other parameter settings. It is worth noting that for these 16
isomeric variants, the differences in dissociation were marked
enough that this was not a significant factor in their discrimina-
tion. Issues 2 and 3, above, were most essential to control on
an ongoing basis to yield reproducible spectra as illustrated in
Figure 2A.
Possible Cyclization of m/z 221 Anions in the Gas Phase:
Evidence for Cyclic Structures and Internal Hemiacetal
Formation of Glycosyl-Glycolaldehydes Based on NMR
Spectroscopy in Solution. Although NMR spectroscopy was
not the primary focus of this investigation, during the synthesis
of the glycopyranosyl-glycolaldehydes, 20 allyl-glycopyrano-
sides (Chart 2) were necessarily prepared. The allyl glycosides
all showed NMR spectra characteristic of one allyl signal set,
one anomeric proton signal, and one set of sugar ring protons,
which were completely assigned (see the Supporting Informa-
tion). However, NMR spectra of synthetic glycosyl-glycolal-
dehydes prepared from the allyl glycopyranosides indicated that
several of them exist in cyclic forms through formation of a
hemiacetal between the glycolaldehyde carbonyl group and the
sugar 2-hydroxyl group. In Figure 9 is shown the downfield
region of the NMR spectrum of â-D-galactopyranosyl-glycol-
aldehyde. The anomeric protons of the â-D-galactopyranosyl
moiety were observed as three sets of doublets in varying
3
proportions (with J1,2 ) 8 Hz for each of them). In addition
three sets of the glycolaldehyde H-1′, H-2′a, and H-2′b signals
were observed, the H-1′ signals being found downfield (Figure
9). Above the spectrum is shown the ratios of the cyclic and
open-chain isomers that contribute in solution to the overall
NMR spectrum. Form A was the open-chain hydrate, which
showed similar couplings of the glycolaldehyde H-1′ to both
H-2′a and H-2′b. Form B had one large coupling of the
glycolaldehyde H-1′ to one H-2′ signal (an antiperiplanar
arrangement) and one small coupling to another H-2′ signal (a
gauche arrangement). Form C had two small couplings of H-1′
to both H-2′a and H-2′b (both gauche). The NMR spectra of
all the glycopyranosyl-glycolaldehydes in D2O (Supporting
Information) showed that some of them existed in cyclic forms
in equilibrium with the open-chain hydrate (like the â-Gal
version shown in Figure 9), whereas some existed solely, or at
least highly predominantly, as the open-chain hydrate of the
glycoaldehyde. The 2-acetamido-2-deoxyhexopyranosyl-glycol-
aldehydes only existed as open-chain hydrates in solution.
Although the open-chain hydrates of the synthetic glycosyl-
glycolaldehydes exist in D2O solution as evaluated by NMR,
the hydrate ions (m/z 239) were not abundant species found in
the gas phase in the negative ion mode when nanospray from
(48) Bristow, A. W. T.; Nichols, W. F.; Webb, K. S.; Conway, B. Rapid
Commun. Mass Spectrom. 2002, 16, 2374-2386.
9
J. AM. CHEM. SOC. VOL. 129, NO. 31, 2007 9733

A R T I C L E S
Chart 2
Fang and Bendiak
methanol solution was carried out (a maximum of ∼20% of
the m/z 221 anion was ever observed).49 Taken together, the
data indicate that the m/z 221 anions may exist in multiple cyclic/
acyclic forms in the gas phase through internal hemiacetal
formation as is characteristic of reducing sugars. Cyclization
might confer upon these ions unique gas-phase equilibria that
in part contribute to the stereochemical dependence of their
spectral differences. This adds another layer of complexity in
considering possible mechanisms of their dissociation.
charides, where 18O-labeling of the reducing carbonyl group has
been invaluable for mass-discrimination of product ion fragments
originating from either side of the glycosidic linkage (ref 41
and herein). Many more isotopically labeled reducing disac-
charides will need to be investigated in order to draw any firm
general conclusions about the origins of the m/z 221 anions.
Moreover, at least in the Paul trap using CID, 3-linked
disaccharides give rise to m/z 221 anions in only trace
abundance, although higher energy dissociation conditions and
in-source fragmentation can give rise to somewhat larger
quantities, as high as 12% of the base product peak.28,31,41 (3)
Neither of the carbons of the glycolaldehyde portion of the m/z
221 ions are chiral. This limits the number of possible
stereochemical and/or positional variants of the monosaccharide-
glycoaldehydes. With the next largest fragment typically derived
from disaccharides in the negative ion mode (m/z 251), one more
chiral carbon and two linkage positions of the aglycon (the 2-
or 3-positions of D- or L-glyceraldehyde) contribute to 4 times
the number of overall stereochemical variants. (4) Our prelimi-
nary data with glucosyl-glycolaldehydes indicates that pyrano-
sides can also be differentiated from furanosides,41 at least for
this sugar.
Summary and Conclusions
The utility of repeated isolation/dissociation steps in a multi-
stage disassembly of oligosaccharides using ion traps or FTICR
instruments has now enabled substructures to be isolated on a
routine basis.18-21 The m/z 221 anion, isolated as a substructure
in the negative ion mode from disaccharides or potentially from
larger oligosaccharides using MSn, appears particularly useful
and worthy of further detailed study for four reasons: (1) The
stereochemistry and anomeric configuration of all 16 variants
of the aldohexopyranosyl-glycolaldehydes can be reproducibly
differentiated by their dissociation “signatures” in a Paul trap
in the negative ion mode. This has not been possible using the
m/z 179 anions of monosaccharides derived from disaccharide
m/z 341 precursors, where neither the anomeric configuration
nor the stereochemistry can typically be ascertained (Figure 1).
(2) The m/z 221 anions derived from reducing disaccharides
contain the intact nonreducing sugar and a 2-carbon aglycon
derived from the reducing sugar, irrespective of the anomeric
configuration or linkage position. This is based on our observa-
tions up to this point with 14 different 2-, 4-, and 6-linked disac-
The N-acetylhexosamines GalNAc and GlcNAc can be
discriminated as m/z 262 anions (HexNAc-glycolaldehydes) as
can their anomers. This should be useful in analysis of
glycoprotein and other HexNAc-containing oligosaccha-
rides.
Assignment of both the stereochemistry and anomeric con-
figuration of the nonreducing monosaccharide component of
reducing disaccharides was previously only possible using NMR
spectroscopy or, with some sugars, through glycosidase treat-
ments. Differences in fragmentation of all 16 glycopyranosyl-
glycolaldehyde variants provide a fundamental spectral basis
and spectral standards for the potential assignment of monosac-
(49) Another ion, m/z 237, became evident in small quantities and more abundant
in increasing quantities as the source temperature was increased from 120
to 250 °C. We have tentatively assigned this as an artifact resulting from
oxidation of the carbonyl to a carboxyl group that can occur in the source
if the temperature of the heated capillary is too high.
9
9734 J. AM. CHEM. SOC. VOL. 129, NO. 31, 2007

Monosaccharide Stereochemistry by Mass Spectrometry
A R T I C L E S
accumulated for each spectral acquisition. Samples were introduced in
methanol or methanol/water mixtures having up to 5% water or 18O
water. High-resolution MS measurements were carried out using the
same solvent mixture in the negative ion mode on a Thermoquest 7 T
FTICR instrument interfaced to a front end linear ion trap equipped
with a nanospray source (Finnigan LTQ-FTICR), located in the
Research Resources Center at the University of Illinois at Chicago
Medical Center.
18O-Labeling of Reducing Disaccharides. Disaccharides were
obtained from Sigma or V-labs. Initially, labeling of the carbonyl
oxygen of the reducing sugar of disaccharides was carried out essentially
as described,23,41 a procedure that utilized 0.5 µL of acetyl chloride
added to the disaccharide dissolved in 100 µL of H218O (Isotec).
However, a time-course was performed to evaluate labeling in the
absence of acetyl chloride. It was found that after 3 days in H218O
alone, labeling occurred essentially completely in the absence of the
acetyl chloride. The following method was therefore used: disaccharide
(1 mg) was dissolved in 100 µL of H218O (Isotec), and after 3 days the
solution was frozen. This stock could be stored essentially indefinitely
and was diluted just prior to use 100-1000-fold in methanol prior to
nanospray.
Nuclear Magnetic Resonance Spectroscopy. NMR spectroscopy
was carried out at a field strength of 500 MHz (1H frequency) on a
Varian Inova instrument. Acquisitions were performed on synthetic
compounds in D2O at 25 °C with a trace of acetone as an internal
chemical shift standard at δ ) 2.225 ppm and are reproducible to within
0.002 ppm. 1H-1H J-couplings are accurate to within 0.2 Hz.
Assignments were either made through decoupling experiments in 1D
experiments or through 2D gCOSY50 correlations. In some cases, a
presaturation pulse was applied to the HOD peak to diminish its
intensity.
Synthesis of Allyl Glycopyranosides. Monosaccharides were
obtained from Sigma or ICN Biomedicals. Allyl glycosides 21-40
(Chart 2) were prepared by basic Fischer-type glycosidation as shown
in the first step of Scheme 2. Monosaccharides (100 mg) were dissolved
in 10 mL of allyl alcohol by heating to about 90 °C for 5-15 min.
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 taken up in 0.5 mL of water
followed by 4.5 mL of acetonitrile. Resultant allyl glycosides of all
four configurations and ring forms were separated by normal phase
HPLC. First, separation was performed in about 10 batches on a column
(2.15 cm × 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 yielded from 1 to 3 pure allylglycosides, either pyrano-
sides or furanosides. When two isomeric compounds were not separated
by the above column, they were further fractionated by normal phase
separation on a Shodex DC-613 (6 mm × 150 mm) column either
converted to the Cs+ or Li+ form,51 eluted with 95/5, v/v, acetonitrile/
water at 0.6 mL/min, detecting at 200 nm. If required, isolation of
individual allyl-glycopyranosides on columns other than the glycopak
N is presented in the Supporting Information. Some of the allyl
glycopyranosides have been previously prepared with accompanying
NMR data (21-24, 37, 38,52,53 25,54 26,55 32,56 39, 4057). Our NMR
Figure 9. NMR spectroscopy (500 MHz, D2O) of â-D-galactopyranosyl-
2-glycolaldehyde 4, illustrating the equilibrium between the acyclic hydrate
form (A) and the two cyclic variants (B and C) as shown in the structural
diagram. Downfield regions of the NMR spectrum show the glycolaldehyde
H-1′ signals (from 5.05 to 5.22 ppm) and the â-Galp H-1 signals (from
4.41 to 4.51 ppm) for the three different species in equilibria, integrated to
give the percentages shown along with each structure. Not all sugar
stereochemistries showed cyclization; notably, the 2-acetamido-2-deoxy-
hexose-glycolaldehydes were incapable of cyclization at the 2-position and
showed no other variants but the open-chain hydrates.
charides within larger oligosaccharides through multistep disas-
sembly to disaccharide substructures. At this point in time,
additional methods other than CID to dissociate larger oligosac-
charides would also be desirable so that useful sets of disac-
charide substructures, especially overlapping disaccharides,
could be isolated. This would enable more effective use to be
made of the information that can be garnered from the glycosyl-
glycolaldehyde anions.
Experimental Section
Mass Spectrometry. Mass spectrometry was performed to MS2 or
MS3 using a Thermoquest (Finnigan) classic LCQ instrument of Paul
trap design equipped with a Finnigan nanospray source. The LCQ was
operated using automatic gain control, with a typical needle voltage of
2.8 kV and a heated capillary temperature of 120 °C for glycoside-
glycolaldehydes (200 °C for allyl glycosides) with the sample introduced
at 0.4 µL/min. The MSn experiments 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 30 ms. A total of 20-100 scans were
(50) Hurd, R. E. J. Magn. Reson. 1990, 87, 422-428.
(51) Mårtensson, S.; Levery, S. B.; Fang, T. T.; Bendiak, B. Eur. J. Biochem.
1998, 258, 603-622.
(52) Lee, R. T.; Lee, Y. C. Carbohydr. Res. 1974, 37, 193-201.
(53) Holme, K. R.; Hall, L. D. Carbohydr. Res. 1992, 225, 291-306.
(54) Winnik, F. M.; Carver, J. P.; Krepinsky, J. J. J. Org. Chem. 1982, 47,
2701-2707.
(55) Utille, J.-P.; Priem, B. Carbohydr. Res. 2000, 329, 431-439.
(56) Rich, J. R.; Wakarchuk, W. W.; Bundle, D. R. Chem. Eur. J. 2006, 12,
845-858.
(57) Wong, T. C.; Townsend, R. R.; Lee, Y. C. Carbohydr. Res. 1987, 170,
27-46.
9
J. AM. CHEM. SOC. VOL. 129, NO. 31, 2007 9735

A R T I C L E S
Scheme 2
Fang and Bendiak
results were consistent, although in many of these reports NMR spectra
were acquired at lower field. At 500 MHz, additional splittings of the
allyl group protons due to small couplings of the H-1′a to H-1′b and
both H-1′a and H-1′b to H-3′a and H-3′b were observed, and more
assignments and endocyclic J-couplings could be made. Also, in one
case (for 3256) the solvent used was CD3OD rather than D2O; thus,
chemical shifts were somewhat different, but endocyclic J-couplings
were essentially the same. Three further points were worthy of note
regarding NMR spectra of the allyl-glycopyranosides. First, for allyl-
R-D-idopyranoside 35 the predominant conformation was the 1C4 form
(it may also exist to an undefined extent in a skew form or interconvert
between the two forms), which was not the case for the other
allylglycosides (Chart 2). Second, characteristic for all the allyl
glycopyranosides was the upfield shift of the H-5 signal of the
â-glycopyranosides relative to the H-5 of their respective R-glycopy-
ranosides. Third, many of the allyl furanosides were encountered during
HPLC purification steps, and their NMR spectra clearly distinguished
them from pyranosides based on the previous assignment of methyl
glycofuranoside J-couplings by Angyal.58 In the interest of comparing
and consolidating the higher-field data for all the allyl glycopyranosides
in this report, NMR spectra and tabulated NMR data for 21-40 are
presented in the Supporting Information. No previous reports have
presented MS/MS spectra of allyl glycopyranosides in the negative ion
mode; for 21-40 these are presented in the Supporting Information
along with quantitative peak data using the Paul trap and high-resolution
MS data using the LTQ-FTICR.
commercial ozone generator, was bubbled through the sample for 6-8
min. Samples were immediately analyzed by NMR to check for reaction
completion, freeze-dried to remove formaldehyde, then stored frozen
in aqueous solution. Ozonolysis was a clean, essentially quantitative
reaction during this (optimized) reaction period, as judged by the
complete disappearance of starting compound. Similar high reaction
yields (albeit under different reaction conditions) have been previously
reported for preparation of some glycosyl-glycolaldehydes from their
respective allyl glycosides.53 Aliquots were removed for NMR and MS
analyses. NMR spectra were often more complex than expected, because
the glycosyl-glycolaldehydes were often present in three forms: the
open-chain hydrate and two additional six-membered cyclic forms, as
indicated for the â-D-galactopyranosyl-glycolaldehyde in Figure 9. Some
of the glycosyl-glycolaldehydes were only found in the open-chain
hydrate form; the HexNAc-glycolaldehydes, for example. Wherever
possible, complete sets of J-couplings and assignments for each form
are reported in tables in the Supporting Information. MS/MS spectra
of the glycopyranosyl-glycolaldehydes in the negative ion mode are
presented in Figures 2-6. Quantitative relative MS/MS ion abundances
from the Paul trap and high-resolution MS data using the LTQ-FTICR
is presented in the Supporting Information. Compounds 1-20 will be
available through the University of Colorado technology transfer office,
12635 East Montview Blvd., Suite 350, Aurora, Colorado 80045, U.S.A.
Acknowledgment. We thank Dr. Robert Murphy for use of his
ozone generator and Dr. Yan Wang for running high-resolution
MS at the University of Illinois at Chicago. This work was sup-
ported in part by a Grant from the National Science Foundation
(CHE-0137986). Funds from the NSF, NIH, Howard Hughes
Medical Foundation, and the Keck Foundation supported instru-
mentation used in this research (NMR and mass spectrometers).
Synthesis of Glycopyranosyl-Glycolaldehydes. A few glycopyra-
nosyl-glycolaldehydes have been previously prepared,41,53,59-61 but for
only two structures has NMR data been presented (for 1 and 2).41,59
MS data in the negative ion mode has only recently been presented for
1 and 2.41 In the other reports,53,60,61 products were directly converted
to Schiff base derivatives without characterization. The glycopyranosyl-
glycolaldehydes 1-20 (Chart 1) were prepared from the pure allyl
glycopyranosides 21-40 (Chart 2) by ozonolysis of the allyl double
bond as shown in step 2 of Scheme 2. Allyl glycosides (10 mg) were
dissolved in 2 mL of D2O. Ozone, produced from pure O2 gas using a
Supporting Information Available: NMR spectra of 1-40,
tabulated NMR assignments, J-couplings, and chemical shifts
for 1-40, mass spectra (negative ion mode) of 21-40, intensity
data for mass spectral peaks for 1-40, high-resolution MS data
for 1-40, additional HPLC columns, if necessary, and counter-
ions used for normal phase purification of 21-40. This material
is available free of charge via the Internet at http://pubs.acs.org.
(58) Angyal, S. J. Carbohydr. Res. 1979, 77, 37-50.
(59) Bendiak, B.; Salyan, M. E.; Pantoja, M. J. Org. Chem. 1995, 60, 8245-
8256.
(60) Bernstein, M. A.; Hall, L. D. Carbohydr. Res. 1980, 78, C1-C3.
(61) Weingarten, S.; Thiem, J. Synlett 2003, 7, 1052-1054.
JA0717313
9
9736 J. AM. CHEM. SOC. VOL. 129, NO. 31, 2007