Hexose rearrangements upon fragmentation of N-glycopeptides and reductively aminated N-glycans

Article abstract of DOI:10.1021/ac900278q

Tandem mass spectrometry of glycans and glycoconjugates in protonated form is known to result in rearrangement reactions leading to internal residue loss. Here we studied the occurrence of hexose rearrangements in tandem mass spectrometry of N-glycopeptid

Full text of DOI:10.1021/ac900278q

Anal. Chem. 2009, 81, 4422–4432
Hexose Rearrangements upon Fragmentation of
N-Glycopeptides and Reductively Aminated
N-Glycans
Manfred Wuhrer,* Carolien A. M. Koeleman, and Andre´ M. Deelder
Leiden University Medical Center, Biomolecular Mass Spectrometry Unit, Department of Parasitology,
P.O. Box 9600, 2300 RC Leiden, The Netherlands
Tandem mass spectrometry of glycans and glycoconju-
gates in protonated form is known to result in rearrange-
ment reactions leading to internal residue loss. Here we
studied the occurrence of hexose rearrangements in
tandem mass spectrometry of N-glycopeptides and reduc-
tively aminated N-glycans by MALDI-TOF/TOF-MS/MS
and ESI-ion trap-MS/MS. Fragmentation of proton ad-
ducts of oligomannosidic N-glycans of ribonuclease B that
were labeled with 2-aminobenzamide and 2-aminobenzoic
acid resulted in transfer of one to five hexose residues to
the fluorescently tagged innermost N-acetylglucosamine.
Glycopeptides from various biological sources with oli-
gomannosidic glycans were likewise shown to undergo
hexose rearrangement reactions, resulting in chitobiose
cleavage products that have acquired one or two hexose
moieties. Tryptic immunoglobulin G Fc-glycopeptides with
biantennary N-glycans likewise showed hexose rearrange-
ments resulting in hexose transfer to the peptide moiety
retaining the innermost N-acetylglucosamine. Thus, as a
general phenomenon, tandem mass spectrometry of re-
ductively aminated glycans as well as glycopeptides may
result in hexose rearrangements. This characteristic of
glycopeptide MS/MS has to be considered when develop-
ing tools for de novo glycopeptide structural analysis.
mentation of deprotonated, native species generally provides
very detailed structural information.8-10 (Tandem) mass spec-
trometry of tryptic glycopeptides is increasingly gaining popu-
larity as this technique is part of proteomics work-flows aiming
at protein identification, quantification, and characterization of
various post-translational modifications.11
A significant portion of tandem mass spectrometric approaches
in structural analysis of glycoconjugates deals with fragmentation
of protonated species. These ions are easily generated from
(tryptic) glycopeptides as well as reductively aminated glycans.
Protonated glycans and glycoconjugates are typically less stable
than the corresponding sodium adducts; therefore, dissociation
may be induced after mild collisional activation. Fragmentation
is characterized by the cleavage of (multiple) glycosidic bonds,
resulting in B-type and Y-type ions.11
In addition to glycosidic-bond cleavages, rearrangements
resulting in internal residue loss have been described in the
collisionally induced dissociation (CID) of protonated oligosac-
charides and glycoconjugates. These rearrangements have been
reported to result in the loss of internal monosaccharide12-16 or
oligosaccharide units14,17-19 which have been named “false” sugar
sequence ions.17 Next to proton adducts, ammonium adducts have
been reported to result in such rearrangement products.17
Recently, the MS/MS analysis of oligosaccharides derivatized by
reductive amination with benzylamine and N,N-dimethylation of
the secondary amine group revealed the migration of fucose to
Mass spectrometry is a key method for the glycosylation
analysis of biological samples. Several approaches are often
combined to analyze protein glycosylation at the levels of intact
glycoproteins, released glycans, or glycopeptides.1 Glycans may
be characterized in detail after enzymatic release by combining
high-resolution separation techniques with tandem mass
spectrometry.2-7 Linkage information is provided by cross-ring
cleavages, which are particularly prominent in high-energy
collision-induced dissociation of sodium-adducts of native or
permethylated glycans.2 Moreover, negative-ion mode frag-
(8) Harvey, D. J. J. Am. Soc. Mass Spectrom. 2005, 16, 647–659
(9) Harvey, D. J. J. Am. Soc. Mass Spectrom. 2005, 16, 631–646
(10) Wuhrer, M.; Deelder, A. M. Anal. Chem. 2005, 77, 6954–6959
.
.
.
(11) Wuhrer, M.; Catalina, M. I.; Deelder, A. M.; Hokke, C. H. J. Chromatogr.,
B 2007, 849, 115–128
(12) Ma, Y. L.; Vedernikova, I.; Van den, Heuvel, H.; Claeys, M. J. Am. Soc.
Mass Spectrom. 2000, 11, 136–144
(13) Warrack, B. M.; Hail, M. E.; Triolo, A.; Animati, F.; Seraglia, R.; Traldi, P.
J. Am. Soc. Mass Spectrom. 1998, 9, 710–715
(14) Harvey, D. J.; Mattu, T. S.; Wormald, M. R.; Royle, L.; Dwek, R. A.; Rudd,
P. M. Anal. Chem. 2002, 74, 734–740
(15) Tadano-Aritomi, K.; Hikita, T.; Kubota, M.; Kasama, T.; Toma, K.; Hakomori,
S. I.; Ishizuka, I. J. Mass Spectrom. 2003, 38, 715–722
(16) Dal Piaz, F.; De Leo, M.; Braca, A.; De Simone, F.; De Tommasi, N. Rapid
Commun. Mass Spectrom. 2007, 21, 286–296
(17) Ernst, B.; Mu¨ller, D. R.; Richter, W. J. Int. J. Mass Spectrom. 1997, 160,
283–290
.
.
.
.
* To whom correspondence should be addressed. E-mail: m.wuhrer@lumc.nl.
.
(1) Geyer, H.; Geyer, R. Biochim. Biophys. Acta 2006, 1764, 1853–1869
.
(2) Zaia, J. Mass Spectrom. Rev. 2004, 23, 161–227
(3) Zaia, J. Mass Spectrom. Rev. 2009, 28, 254–272
.
.
.
(4) Wuhrer, M.; de Boer, A. R.; Deelder, A. M. Mass Spectrom. Rev. 2009,
.
28, 192–206
(5) Wuhrer, M.; Deelder, A. M.; Hokke, C. H. J. Chromatogr., B 2005, 825,
124–133
(6) Mechref, Y.; Novotny, M. V. Chem. Rev. 2002, 102, 321–369
.
(18) Wuhrer, M.; Balog, C. I. A.; Catimel, B.; Jones, F. M.; Schramm, G.; Haas,
H.; Doenhoff, M. J.; Dunne, D. W.; Deelder, A. M.; Hokke, C. H. FEBS J.
.
2006, 273, 2276–2292
(19) Wuhrer, M.; Koeleman, C. A. M.; Hokke, C. H.; Deelder, A. M. Rapid
Commun. Mass Spectrom. 2006, 20, 1747–1754
.
.
(7) Mechref, Y.; Novotny, M. V. Mass Spectrom. Rev. 2009, 28, 207–222
.
.
4422 Analytical Chemistry, Vol. 81, No. 11, June 1, 2009
10.1021/ac900278q CCC: $40.75  2009 American Chemical Society
Published on Web 05/06/2009

the tagged, innermost monosaccharide.20 Notably, because of the
fixation of the charge on the quarternary nitrogen, these deriva-
tives are not supposed to have mobile protons20 and the mecha-
nism of these rearrangements is not clear.
MALDI-TOF(/TOF)-MS(/MS). Samples were spotted with
DHB (10 mg/mL in 30% acetonitrile) matrix and analyzed
in the positive and/or negative (reflectron) mode by MALDI-
TOF(/TOF)-MS(/MS) on a Ultraflex II mass spectrometer
(Bruker Daltonics) containing a Smartbeam Nd:YAG laser (266
nm). Fragment analysis was achieved by laser-induced decom-
position using the LIFT-TOF/TOF-MS/MS facility. Between 2 000
and 10 000 tandem mass spectra were acquired from different sites
on the sample spot and were accumulated to obtain a representa-
tive fragment spectrum. Fragment ions were assigned using
GlycoWorkbench.24
LC-MS(/MS). Reverse phase-nanoLC-ion trap (IT)-MS/MS
of 2AB-labeled glycans was performed on a PepMap column using
an Ultimate 3000 nanoLC system (Dionex) equipped with a guard
column. The system was equilibrated with eluent A (0.4% aceto-
nitrile, 0.1% formic acid in water) at a flow rate of 300 nL/min.
After injection of the sample, a linear gradient to 25% eluent B
(H2O/acetonitrile 5:95, v/v, containing 0.1% formic acid) in 15
min was applied, followed by a gradient to 70% eluent B in
another 10 min and a final wash with 70% B for 5 min. The
nanoLC system was directly coupled to an Esquire High
Capacity Trap (HCTultra) ESI-IT-MS (Bruker) equipped with
an online nanospray source operating in the positive-ion mode
at 900-1200 V. The solvent was evaporated at 170 °C with a
nitrogen stream of 6 L/min. Ions from m/z 400-2000 were
registered. Automatic fragment ion analysis was enabled,
resulting in MS/MS spectra of the most abundant peaks.
Fragment ions were assigned using GlycoWorkbench.24
Ions generated by internal residue loss, if mistakenly inter-
preted as conventional glycosidic bond cleavages, may lead to the
false postulation of structural motifs and suggest the presence of
a mixture of isomers. Because of the lack of basic rules for the
occurrence of rearrangement products associated with internal
residue loss, tandem mass spectrometry of proton adducts is of
rather limited value for the de novo structural elucidation of
glycoconjugates. Notably, fucosylated glycans and glycoconjugates
have repeatedly been shown to undergo rearrangements reactions
very efficiently,11,14 but rearrangements involving other monosac-
charide moieties have likewise been shown.12,14,15,21
Here, we describe for the first time hexose rearrangements
in tandem mass spectrometry of proton adducts of N-glycopeptides
and reductively aminated N-glycans. A profound knowledge of the
occurrence of such rearrangement reactions would appear to be
essential to establish reliable de novo sequencing algorithms for
structural glycomics at the glycopeptide level.
EXPERIMENTAL SECTION
Materials and Reagents. Peptide N-glycosidase F (PNGase
F) was purchased from Roche Diagnostics, Mannheim, Germany.
Bovine ribonuclease B (RNase B), 2-aminobenzamide (2AB),
2-aminobenzoic acid (anthranilic acid, AA), sodium cyanoboro-
hydride, formic acid, and ammonia (25% aqueous solution) were
from Sigma-Aldrich (Zwijndrecht, The Netherlands). The HILIC
HPLC column (Amide-80, 4.6 mm × 25 cm, particle size 5 µm)
was from Tosoh Bioscience, Stuttgart, Germany. 2,5-Dihydroxy-
benzoic acid (DHB) was obtained from Bruker Daltonics (Bremen,
Germany). The PepMap column (3 µm; 75 µm × 150 mm) and
guard column (300 µm × 5 mm) were from Dionex, Amsterdam,
The Netherlands. Platin coated fused-silica electrospray needles
360 µm o.d., 20 µm i.d. with 10 µm opening) were from New
Objective (Cambridge, MA).
Glycan Release and Labeling. N-Glycans were released from
RNase B (50 mg) by PNGase F as described previously.22 Briefly,
protein was dissolved in PBS containing SDS (1%) and 2-mercap-
toethanol (0.5%) and incubated for 10 min at 100 °C. Chaps was
added to a final concentration of 1%, followed by overnight PNGase
F treatment at 37 °C. Released glycans were sequentially purified
by using RP- and graphitized carbon cartridges.22 Glycans were
labeled with the fluorescent compounds 2AB and AA by reductive
amination with sodium cyanoborohydride.23 The labeled glycans
were purified by HILIC HPLC with fluorescence detection
(λex-λem 360-425 nm). Eluent A consisted of 50 mmol/L
ammonium formiate (pH 4.4), and eluent B consisted of eluent
A/acetonitrile 20:80. A linear gradient from 100% to 40% eluent
B was applied at a flow rate of 1 mL/min. Peak-fractions were
analyzed by MALDI-TOF-MS, and fractions containing isomers
were pooled.
RESULTS
N-Glycans of bovine ribonuclease B (RNase B) were
released using PNGase F, labeled with 2-aminobenzoic acid
(AA) and 2-aminobenzamide (2AB), purified by HILIC with
fluorescence detection, and analyzed by MALDI-TOF/TOF-MS/
MS. Fragmentation of the deprotonated, AA-labeled hexaman-
nosidic (Man6GlcNAc2-AA) N-glycan in negative-ion mode is
shown in Figure 1A. The tandem mass spectrum is character-
ized by cleavages of glycosidic bonds as well as some cross-
ring cleavages, which are in full compliance with the structure
of the RNase B Man6 N-glycan as described in the literature.25,26
Analysis of the HPLC-purified Man6GlcNAc2-AA by MALDI-
TOF-MS in positive-ion mode resulted in the registration of
sodium adducts ([M + Na]+ at m/z 1378), proton adducts
([M + H]+ at m/z 1356), and potassium adducts ([M + K]+,
at m/z 1394) at relative intensities of approximately 60:20:
20, respectively. Proton adducts of the RNase B Man6 in
AA-labeled form (Figure 1B) and 2AB-labeled form (Figure
1C) were analyzed by MALDI-TOF/TOF-MS/MS. The frag-
mentation of the [M + H]+ species did predominantly result
in cleavages of glycosidic bonds, and no cross-ring cleavages
were observed. The most intense fragment ions at m/z 1176
arose from chitobiose cleavage (Figure 1B,C). Moreover, a
combination of chitobiose cleavage with loss of mannoses
accounted for a series of ions at m/z 1014, m/z 852, m/z 690,
(20) Broberg, A. Carbohydr. Res. 2007, 342, 1462–1469
(21) Hsu, F. F.; Turk, J. J. Am. Soc. Mass Spectrom. 2004, 15, 536–546
(22) Wuhrer, M.; Koeleman, C. A. M.; Deelder, A. M.; Hokke, C. H. FEBS J.
2006, 273, 347–361
(23) Bigge, J. C.; Patel, T. P.; Bruce, J. A.; Goulding, P. N.; Charles, S. M.; Parekh,
R. B. Anal. Biochem. 1995, 230, 229–238
.
.
(24) Ceroni, A.; Maass, K.; Geyer, H.; Geyer, R.; Dell, A.; Haslam, S. M. J.
Proteome Res. 2008, 7, 1650–1659
(25) Fu, D.; Chen, L.; O’Neill, R. A. Carbohydr. Res. 2009, 261, 173–186
(26) Zhuang, Z.; Starkey, J. A.; Mechref, Y.; Novotny, M. V.; Jacobson, S. C.
Anal. Chem. 2007, 79, 7170–7175
.
.
.
.
.
Analytical Chemistry, Vol. 81, No. 11, June 1, 2009 4423

4424 Analytical Chemistry, Vol. 81, No. 11, June 1, 2009

Figure 1. MALDI-TOF/TOF-MS/MS analysis of a reductively aminated hexamannosidic N-glycan. The Man6GlcNAc2 N-glycan from RNase B
was labeled with 2-aminobenzoic acid (AA) and analyzed by MALDI-TOF/TOF-MS/MS in negative ion mode ((A) [M - H]-) and positive-ion
mode ((B) [M + H]+). Man6GlcNAc2 from RNase B was labeled with 2AB and analyzed by MALDI-TOF/TOF-MS/MS in positive-ion mode ((C)
[M + H]+). Schemes often represent one of several possible assignments. Blue square, GlcNAc; green circle, mannose; double-headed arrow,
hexose; *, poorly resolved metastable ion. Hexagons indicate A-type cross-ring cleavages, with the white segment corresponding to the carbon
atoms retained on the A-type ion.24
m/z 528, m/z 366, and m/z 204. All the intense fragment ions
included part of the chitobiose core, which clearly showed a
preferred association of the proton with this part of the
molecule. Ions of low intensity at m/z 163, m/z 325, and m/z
487 were oxonium ions and indicated that protons did also
associate with oligomannosidic motifs of the molecule. Strik-
ingly, the fragment ion of the AA-labeled glycan at m/z 343
(Figure 1B) arising from chitobiose cleavage (Y1-ion) was
accompanied by a series of related ions with added hexoses
at m/z 505 (1 hexose), m/z 667 (2 hexoses), m/z 829 (3
hexoses), and m/z 991 (4 hexoses). A similar series of
hexose additions was observed for the chitobiose cleavage
(m/z 342) product of Man6GlcNAc2-2AB resulting in signals
at m/z 504, m/z 666, m/z 828, and m/z 990 with one to four
hexoses added (Figure 1C). Notably, these Y1-ions “deco-
rated” with hexoses were only observed in MALDI-TOF/
TOF-MS/MS of proton adducts, and no corresponding
hexose additions to the chitobiose cleavage products had
been observed in negative-ion mode MALDI-TOF/TOF-MS/
MS (Figure 1A) and on fragmentation of sodium adducts (not
shown). Moreover, these fragment ions did not correspond to
known structural motifs on RNase B glycans and were,
therefore, interpreted as rearrangement products. In order to
confirm the composition of these rearrangement products, the
ion observed at m/z 828 (Figure 1C) was subjected to further
investigation. Fragmentation of the 2AB-labeled N-glycan was
achieved by in-source decay, and the product ion at m/z 828
was selected and further fragmented by MALDI-TOF/TOF-MS/
MS (Figure 2). The resulting fragment-ion spectrum clearly
identified this species as the chitobiose cleavage product
carrying three hexose moieties.
AA-labeled and 2AB-labeled RNase B N-glycans with 5, 7, 8,
and 9 mannose residues were similarly analyzed by negative-ion
mode ([M - H]-) and positive-ion mode ([M + H]+) MALDI-
TOF/TOF-MS/MS (mass spectra not shown). Consistently,
mannose additions to the chitobiose cleavage products were
observed upon fragmentation of the protonated species, while
all negative mode fragment ions were in accordance with the
known RNase B N-glycans structures and the rules established
-10
for negative mode tandem mass spectrometry of glycans.8
In order to visualize the efficacy of the rearrangement events, the
intensities of the mannose-containing chitobiose cleavage products
were normalized to the intensity of the related ion lacking the
mannose (Y1-ion at m/z 343 and m/z 342 for AA-labeled and
2AB-labeled glycans, arising from chitobiose cleavage) and are
displayed in histograms (Figure 3B,C). The rearrangement ion
comprising one mannose had an intensity of between 6% and 11%
Analytical Chemistry, Vol. 81, No. 11, June 1, 2009 4425

mannose that results in the fragment ions of composition H1N1-
AA and H1N1-AB.
Notably, rearrangement products were not restricted to tandem
mass spectrometry by MALDI-TOF/TOF-MS/MS: analysis of the
RNase B Man6GlcNAc2-2AB N-glycan ([M + 2H]2+ at m/z 759)
by reverse phase-nanoLC-ESI-IT-MS/MS with collisional
activation likewise resulted in a hexose rearrangement product
at m/z 504 (Figure 4). The other 2AB-labeled oligomannosidic
N-glycans of RNase B were also analyzed which similarly revealed
the hexose rearrangement product at m/z 504. Rearrangement
products with a higher number of hexoses were also visible in
some cases, however, at very low intensities (not shown).
In the second part of the study, we addressed the question
whether similar hexose rearrangement events occur on tandem
mass spectrometry of glycopeptides, as these glycoconjugates are
predominantly analyzed in protonated form, similar to the oligo-
mannosidic glycans analyzed above. We started by doing an
extensive literature search of glycopeptide tandem mass spectra
which showed that similar fragment ions pointing to hexose
rearrangement reactions could indeed be found in two tandem
mass spectrometric analyses of glycopeptides with oligomanno-
sidic glycans.27,28 First, a tryptic glycopeptide from RNase B was
analyzed by Zhang and Chelius using ion trap multistage tandem
mass spectrometry.28 Using this approach, they fragmented a
glycopeptide with a pentamannosidic N-glycan. From the observed
fragments, they isolated the fragment resulting from the loss of
a single mannose. This ion was further fragmented, resulting
among others in an ion retaining three mannose residues. Isolation
and fragmentation of this ion provided the peptide moiety with a
dimannosidic structure, which was subjected to fragmentation,
followed by registration of the fragment ions (Figure 5A). Again,
while most ions were assigned to glycosidic bond cleavages, two
ions at m/z 840 and m/z 1002 remained unassigned in the original
report.28 On the basis of the above-mentioned indications of
hexose rearrangements, these ions may be assigned to [peptide
+ HexNAc + Hex + H]+ and [peptide + HexNAc + 2Hex +
H]+, respectively (Figure 5A). Notably, the latter ion corresponds
to the loss of an internal N-acetylglucosamine, with the retention
of the two mannose residues.
Figure 2. MALDI-TOF/TOF-MS/MS analysis of a hexose rearrange-
ment product. The fragment ion at m/z 828, which was observed in
MALDI-TOF/TOF-MS (postsource decay; Figure 1C) was selected
as an in-source decay ion for further fragment analysis (postsource
decay) by MALDI-TOF/TOF-MS/MS. Annotation was performed
similar to Figure 1.
of the intensity of the Y1-ion (Figure 3B,C). This ion had the
highest intensity of all mannose rearrangement products in all
10 fragment ion spectra of protonated oligomannosidic N-glycans.
In general terms, the more mannoses had to be transferred during
rearrangement, the lower the relative intensity of the product.
Strikingly, while the intensities of the rearrangement products with
two and three mannoses were still rather high for Man5 and Man6
N-glycans, rearrangement products with four or more mannoses
were of much lower intensity (Figure 3B,C). For Man7 and Man8,
a similarly sharp drop in intensities was observed: rearrangement
products with up to four mannoses were of high intensity, while
those with five and more mannoses were hardly detectable. For
Man9, in contrast, up to five mannoses were rather efficiently
transferred to fluorescently labeled innermost GlcNAc, while the
transfer of six or more mannoses was hardly observed. When
comparing the mannose transfer patterns (Figure 3B,C) with the
major structures of RNase B N-glycans (Figure 3A), we found
that the number of mannoses present on the R1-6-linked branch
of the trimannosyl core (boxed in red, Figure 3A) correlated with
the maximum number of mannoses that were still transferred
efficiently (red bars in Figure 3B,C). For the transfer of multiple
hexoses, this means that any mannose present on the six-branch
of the oligomannosidic N-glycan may be transferred efficiently to
the labeled innermost GlcNAc, compared to the other mannoses
present in the molecule. Whether this transfer occurs en bloc or
one by one may not be concluded from the present data. Notably,
our data give no indication as to the origin of single transferred
Second, the analysis by Nimtz et al.27 of a tryptic N-glycopep-
tide of a chicken eggshell glycoprotein carrying a Man7 N-glycan
structure by ESI-quadrupole TOF-MS/MS revealed similar rear-
rangement candidates: a major ion at m/z 920 ([peptide + HexNAc
+ H]+) arising from chitobiose cleavage (Figure 5B) was
accompanied by a signal at m/z 1082 which remained unassigned
in the original report.27 This ion may be explained as a hexose
rearrangement product ([peptide + HexNAc + Hex + H]+), as
indicated in Figure 5B.
Next to these two studies,27,28 in which putative rearrange-
ment products could be assigned with confidence as the
registered masses of the peaks were indicated, eight other
studies were found in the literature which showed candidate
hexose rearrangement peaks in 11 tandem mass spectra of
glycopeptides with oligomannosidic N-glycans (Table 1). The
registered mass of the putative rearrangement products was
(27) Nimtz, M.; Conradt, H. S.; Mann, K. Biochim. Biophys. Acta 2004, 1675,
71–80
.
(28) Zhang, S.; Chelius, D. J. Biomol. Tech. 2004, 15, 120–133
.
4426 Analytical Chemistry, Vol. 81, No. 11, June 1, 2009

Figure 3. Relative quantitation and schematic representation of hexose rearrangement products. (A) Schematic representation of the major
RNase B N-glycans. The R1-6-linked branch, which is postulated to undergo hexose transfer reactions particularly efficiently, is boxed in red.
Histograms of the relative intensities of hexose rearrangement products observed for AA-labeled (B) and 2AB-labeled (C) RNase B N-glycans
in MALDI-TOF/TOF-MS of proton adducts support the preferred rearrangement involving the R1-6-linked branch of the oligomannosidic N-glycans.
Intensities are given relative to the signal of the fluorescently labeled innermost GlcNAc (m/z 343 and m/z 342 in parts B and C, respectively).
not provided in these studies. Therefore, masses of the
candidate peaks were estimated using graphical interpolation,
and the resulting values are given in Table 1. In our literature
study, we encountered mannose rearrangement candidates for
glycopeptides from six different proteins from different organ-
isms, namely, rat, mouse, cattle, chicken, and cauliflower (Table
1). The study of Zhang and Chelius28 as well as four other
studies applied tandem mass spectrometry to RNase B tryptic
glycopeptides resulting in rearrangement candidates (Table
1).29-32 Two studies were analyzing RNase B tryptic glycopep-
tides without missed cleavage sites,28,30 while three studies
were looking at glycopeptides with a missed cleavage site.29,31,32
Analytical Chemistry, Vol. 81, No. 11, June 1, 2009 4427

Figure 4. ESI-IT-tandem mass spectrum of a diprotonated, 2AB-labeled oligomannosidic N-glycan. Man6GlcNAc2-2AB from RNase B was
analyzed by reverse phase-nanoLC-IT-MS/MS ([M + 2H]2+ at m/z 759). Next to ions arising from glycosidic bond cleavages, a hexose
rearrangement product was observed at m/z 504. Annotation was performed similar to Figure 1.
IgG Fc glycopeptides with complex-type N-glycans showed
hexose rearrangement products similar to those observed for
glycopeptides with oligomannosidic N-glycans. Figure 6 shows a
tandem mass spectrum (nanoLC-ESI-IT-MS/MS) of the tryptic
Fc glycopeptide of IgG1 purified from human plasma that carries
a neutral biantennary, digalactosylated N-glycan lacking core
fucose ([M + 2H]2+ at m/z 1406). While most ions could be
assigned to glycosidic bond cleavages resulting in B-ions and
Y-ions that may eliminate water and ammonia, a low-intensity
fragment ion at m/z 1554 corresponded to a theoretical
composition of a singly protonated peptide moiety carrying a
HexNAc and a hexose residue. Similar fragment ions corre-
sponding to [peptide + HexNAc + Hex + H]+ were observed
in fragment ion analysis of other tryptic Fc glycopeptides of
human IgG1, IgG2, and IgG4 (data not shown). While human
IgG Fc N-glycosylation is particularly well-characterized, hexose
attachment to the innermost GlcNAc has not been reported
yet. As such N-glycan modifications have to our knowledge
not been reported in mammals, we assume that the [peptide
+ HexNAc + Hex + H]+ fragment ion observed for human
IgG glycopeptides arose from hexose rearrangements, as
they have been described above for oligomannosidic glyco-
peptides. Moreover, a similar ion of theoretical composition
peptide + GlcNAc + Hex was observed on fragmentation
of a Fc glycopeptide with a core-R1-3-fucosylated, core-
xylosylated trimannosyl N-glycan exhibiting a bisecting
GlcNAc (Hex3HexNAc5dHex1Pent1) that was derived from
a glyco-engineered human monoclonal IgG2 produced in
genetically modified tobacco plants (signal at m/z 1522 in
Figure 7 of Rouwendal et al.;33 Table 1). These findings indicate
that hexose rearrangements may represent a general phenomenon
in collisional fragmentation of protonated IgG glycopeptides.
DISCUSSION
Hexose rearrangements were observed in tandem mass
spectrometry of proton adducts of both reductively aminated
N-glycans and tryptic N-glycopeptides. Strikingly, the transfer of
higher numbers of hexoses to the conjugated innermost GlcNAc
of reductively aminated glycans was occurring rather efficiently
in MALDI-TOF/TOF-MS. This may be due to the open-ring form
of the innermost GlcNAc, resulting in enhanced flexibility of this
residue and may point to a specific role of the protonated
secondary amine group of the AA- and 2AB-labels in catalyzing a
fragmentation/rearrangement reaction, as postulated by Harvey
et al.14 Mo et al. have not observed hexose rearrangement
products on the MALDI postsource decay MS of an RNase B
Man8GlcNAc2 N-glycan labeled with 4-aminobenzoic acid 2(di-
ethylamino)ethyl ester, which points to a role of the label in
(31) Henning, S.; Peter-Katalinic, J.; Pohlentz, G. J. Mass Spectrom. 2007, 42,
1415–1421
(32) Temporini, C.; Perani, E.; Calleri, E.; Dolcini, L.; Lubda, D.; Caccialanza,
G.; Massolini, G. Anal. Chem. 2007, 79, 355–363
.
(29) Liu, T.; Li, J.-D.; Zeng, R.; Shao, X.-X.; Wang, K.-Y.; Xia, Q.-C. Anal. Chem.
.
2001, 73, 5875–5885
(30) Pitchayawasin, S.; Isobe, M. Biosci. Biotechnol. Biochem. 2004, 68, 1424–
1433
.
(33) Rouwendal, G. J.; Wuhrer, M.; Florack, D. E.; Koeleman, C. A.; Deelder,
A. M.; Bakker, H.; Stoopen, G. M.; van, Die, I.; Helsper, J. P.; Hokke, C. H.;
Bosch, D. Glycobiology 2007, 17, 334–344.
.
4428 Analytical Chemistry, Vol. 81, No. 11, June 1, 2009

Figure 5. Examples for glycopeptide tandem mass spectra from literature that show indications for hexose rearrangements. (A) Nano-
ESI-MS5 spectrum of the tryptic glycopeptide from bovine RNase B.28 The double-charged glycopeptide carrying a pentamannosidic N-glycan
was selected as a precursor. In a series of four ion selection/fragmentation cycles, a dimannosidic glycopeptide was generated and
fragmented, as shown in the box. (B) NanoLC-ESI-quadrupole-TOF-MS/MS of a tryptic glycopeptide from chicken egg shell glycoprotein
carrying a heptamannosidic N-glycan.27 Part A was reprinted with permission from Zhang and Chelius (2004).28 Copyright 2004 The
Association of Biomolecular Resource Facilities. Part B was reprinted with permission from Nimtz et al. (2004).27 Copyright 2004 Elsevier.
Blue square, N-acetylglucosamine; green circle, mannose; yellow circle, galactose; white circle, hexose; pep, peptide moiety; #, loss of
water and/or ammonia.
Analytical Chemistry, Vol. 81, No. 11, June 1, 2009 4429

Table 1. Putative Hexose Rearrangement Products Observed Tandem Mass Spectra of Glycopeptides from
Literaturea
glycopeptide species
potential hexose rearrangement peak
proposed assignment
[pep + HexNAc1Hex1 + H]+
ref
tryptic glycopeptide from
chicken eggshell
glycoprotein with a
peak at m/z 1082 in deconvoluted
tandem mass spectrum obtained by
LC-ESI-quadrupole TOF-MS/MS
Nimtz et al.,27 Figure 1A; see
Figure 5B
Hex7HexNAc2 N-glycan
tryptic glycopeptide from
bovine RNase B with a
Hex5HexNAc2 N-glycan
peaks at m/z 840 and m/z 1002 in
MS5 experiment performed on a
[pep + HexNAc2Hex2 + 2H]2+
using chip-based
[pep + HexNAc1Hex1 + H]+ and Zhang and Chelius,28 Figure 4C; see
[pep + HexNAc1Hex2 + H]+
Figure 5A
infusion-ESI-IT-MS/MS
tryptic glycopeptides from
bovine RNase B with one
missed cleavage site
containing Hex5HexNAc2
and Hex6HexNAc2
peaks at approximately m/z 1083 in [pep + HexNAc1Hex1 + H]+
tandem mass spectra obtained by
capillary
Liu et al.,29 Figure 4A,B
electrophoresis-ESI-IT-MS/MS
N-glycans
tryptic glycopeptide from
bovine RNase B with one
missed cleavage site
containing a Hex5HexNAc2
N-glycan
peaks at approximately m/z 1083 in [pep + HexNAc1Hex1 + H]+
tandem mass spectra obtained by
nano-ESI-quadrupole-TOF-MS/MS
Henning et al.,31 Figure 2
tryptic glycopeptide from
bovine RNase B containing a tandem mass spectrum obtained by
peak at approximately m/z 840 in
[pep + HexNAc1Hex1 + H]+
Pitchayawasin and Isobe,30 Figure 4
Temporini et al.,32 Figure 6A
Hex6HexNAc2 N-glycan
tryptic glycopeptide from
bovine RNase B with one
missed cleavage site
containing a Hex5HexNAc2
N-glycan
LC-ESI-quadrupole-TOF-MS/MS
peaks at approximately m/z 1111 in [pep + HexNAc1Hex1 + H]+
tandem mass spectra obtained by
nano-ESI-quadrupole-TOF-MS/MS
tryptic glycopeptide from
cauliflower xyloglycan
endotransglycosidase 16A
with a Hex6HexNAc2
N-glycan
peak at approximately m/z 2765 in
deconvoluted tandem mass
spectrum obtained by
[pep + HexNAc1Hex1 + H]+
Hendriksson et al.,37 Figure 5
Itoh et al.,38 Figure 3A
LC-ESI-quadrupole-TOF-MS/MS
tryptic glycopeptide from rat peaks at approximately m/z 1472 and [pep + HexNAc1Hex1 + H]+;
brain Thy-1 with a
Hex5HexNAc4dHex2
N-glycan
m/z 1634 in tandem mass spectra
obtained by LC-ESI-IT-MS/MS
[pep + HexNAc1Hex2 + H]+
[pep + HexNAc1Hex1 + H]+
[pep + HexNAc1Hex1 + H]+
tryptic glycopeptide from rat peak at approximately m/z 980 in
Itoh et al.,38 Figure 5A
brain Thy-1 with a
Hex5HexNAc2 N-glycan
tryptic glycopeptide from
murine cyclooxygenase 2
with a Hex9HexNAc2
N-glycan
tandem mass spectrum obtained by
LC-ESI-IT-MS/MS
peak at approximately m/z 1988 in
tandem mass spectra obtained by
nano-ESI-quadrupole-TOF-MS/MS
Nemeth et al.,39 Figure 3B
trytpic glycopeptide from rat peak at approximately m/z 1123 in
[pep + HexNAc1Hex1 + H]+
Itoh et al.,40 Figure 5A1
brain LAMP glycoprotein
with a Hex5HexNAc2
N-glycan
tandem mass spectrum obtained by
LC-ESI-IT-MS/MS
tryptic glycopeptide from rat peak at approximately m/z 1081 in
brain Kilon glycoprotein with tandem mass spectrum obtained by
[pep + HexNAc1Hex1 + H]+
[pep + HexNAc1Hex1 + H]+
Itoh et al.,40 Figure 8A1
a Hex5HexNAc2 N-glycan
tryptic glycopeptide from
glycoengineered human
IgG1 expressed in tobacco
with a
Hex3HexNAc5dHex1Pent1
N-glycan
LC-ESI-IT-MS/MS
peak at m/z 1522 in tandem mass
spectrum obtained by
Rouwendal et al.,33 Figure 7
LC-ESI-IT-MS/MS
a For hexose rearrangement candidates which were not labeled with a registered mass, the mass was determined by graphical interpolation and
showed good agreement with the theoretical mass of the hexose rearrangement products (deviation below 2 Da).
inducing hexose rearrangement reactions.34 Notably, the rear-
rangement reactions for AA-labeled and 2AB-labeled oligoman-
nosidic N-glycans were less pronounced in ESI-IT-MS/MS of the
diprotonated forms (Figure 4) than in MALDI-TOF/TOF-MS of
the single protonated species (Figure 1C), and rearrangements
with transfer of multiple hexoses were restricted to MALDI-TOF/
TOF-MS, which may indicate a role of the charge state in
determining rearrangement events.
The ions at m/z 505, m/z 667, m/z 829, and m/z 991, that are
interpreted as rearrangement products, show a mass difference
of 41 Da when compared to the series of GlcNAc2-AA (m/z 546),
Man1GlcNAc2-AA (m/z 708), Man2GlcNAc2-AA (m/z 870), and
Man3GlcNAc2AA (m/z 1032), respectively (Figure 1C). However,
no conclusive explanation for such a loss of a 41 Da functional
(34) Mo, W.; Takao, T.; Sakamoto, H.; Shimonishi, Y. Anal. Chem. 1998, 70,
4520–4526
.
4430 Analytical Chemistry, Vol. 81, No. 11, June 1, 2009

Figure 6. NanoLC-ESI-IT-MS/MS of a tryptic Fc glycopeptide of IgG1 from human plasma (precursor [M + 2H]2+ at m/z 1406). The attached
oligosaccharide is a nonsialylated biantennary, noncorefucosylated N-glycan. Annotation was performed similar to Figure 5.
group could be found: it may theoretically correspond to a
C2N1H3 unit (even electron) or a C2O1H1 unit (odd electron).
We observed a correlation between the size of the 6-branch of
the RNase B oligomannosidic N-glycans and the number of
hexoses which is still transferred efficiently to the fluorescently
labeled GlcNAc. This suggests that the R1-6-linked mannose of
the trimannosyl core, together with its substituents, may easily
act as a donor in the transfer of multiple hexoses to the secondary
amine group of the tag, probably due to the pronounced flexibility
of the 6-branch. We would assume that this transfer of multiple
mannoses occurs en bloc and is part of a rearrangement reaction
that leads to the cleavage of the chitobiose core. However, transfer
one by one of the mannoses cannot be ruled out at the current
stage. Moreover, while the preferred transfer of the 6-antenna
substituents seems to result in multiply mannosylated rearrange-
ment products, the current data do not allow us to specify the
origin of the single mannose included in the H1N1-AA and H1N1-
AB fragment ions.
Isotope labeled standards may provide a key in future experi-
ments to establish preferred migration paths in hexose rearrange-
ments. Isotope labeling approaches may likewise be necessary
to elucidate the transfer paths in fucose rearrangements.19 Other
derivatization approaches like permethylation, in contrast, are
expected to suppress most of the rearrangement reactions, as has
been shown for fucose rearrangements,19 and may therefore only
provide a limited contribution to the elucidation of rearrangement
paths and mechanisms.
CID and ETD on ion traps, quadrupole fragmentation applying a
range of different energies, MALDI-TOF/TOF-MS with postsource
decay and high-energy fragmentation, as well as infrared mul-
tiphoton fragmentation (IRMPD) and 157 nm photodissociation
should provide deeper insights in rearrangements occurring in
glycopeptide fragmentation.11,35,36
One may speculate as to the acceptor site of the transferred
hexoses in tandem mass spectrometry of glycopeptides: the
innermost GlcNAc as well as various functional groups of the pep-
tide moiety may possibly serve as acceptor structures. The
contribution of these different candidate acceptor sites is expected
to depend on the sequence and structure of the individual peptide
moiety as well as on the structure of the glycan part. Strikingly,
while many tandem mass spectra of glycopeptides with complex-
type N-glycans are reported in literature, products of rearrange-
ment reactions of complex-type N-glycans were only found for IgG
tryptic Fc glycopeptides. This indicates that particular peptide
sequences, including the IgG tryptic Fc glycopeptide, possibly
together with the charge state and charge distribution of the
glycopeptide, may stimulate rearrangement reactions by providing
mobile protons and hexose acceptor groups.
(35) Seipert, R. R.; Dodds, E. D.; Clowers, B. H.; Beecroft, S. M.; German, J. B.;
Lebrilla, C. B. Anal. Chem. 2008, 80, 3684–3692
(36) Zhang, L.; Reilly, J. P. J. Proteome Res. 2009, 8, 734–742
(37) Henriksson, H.; Denman, S. E.; Campuzano, I. D.; Ademark, P.; Master,
E. R.; Teeri, T. T.; Brumer, H., III Biochem. J. 2003, 375, 61–73
(38) Itoh, S.; Kawasaki, N.; Harazono, A.; Hashii, N.; Matsuishi, Y.; Kawanishi,
T.; Hayakawa, T. J. Chromatogr., A 2005, 1094, 105–117
(39) Nemeth, J. F.; Hochgesang, G. P., Jr.; Marnett, L. J.; Caprioli, R. M.
Biochemistry 2001, 40, 3109–3116
(40) Itoh, S.; Hachisuka, A.; Kawasaki, N.; Hashii, N.; Teshima, R.; Hayakawa,
T.; Kawanishi, T.; Yamaguchi, T. Biochemistry 2008, 47, 10132–10154
.
.
.
.
N-Glycopeptide fragmentation is known to depend largely on
the type of tandem mass spectrometer and fragmentation mode
applied. A comparative study on glycopeptide fragmentation using
.
.
Analytical Chemistry, Vol. 81, No. 11, June 1, 2009 4431

Monosaccharide rearrangements seem to be closely associated
with the fragmentation event. In this study we observed rear-
rangement reactions occurring at very different time scales and
charge states (MALDI-TOF/TOF-MS versus ESI-IT-MS/MS),
which is in line with an earlier report.19 Hence, we do not expect
that the choice of specific mass spectrometric instruments and
settings will allow the suppression of these rearrangement
reactions in tandem mass spectrometry using collision-induced
dissociation of protonated glycoconjugates.
Ions resulting from internal residue loss are generally identified
on the basis of their unique composition which may not be explained
by glycosidic bond cleavages alone. Hexose rearrangements may,
however, result in ions of nonspectacular composition and may,
therefore, be mistaken as a conventional fragment arising from
glycosidic bond cleavage. Hence, we predict that the actual preva-
lence of rearrangement reactions with internal residue loss may be
much higher than currently indicated. In the case of multistage
tandem mass spectrometry, as applied by Zhang and Chelius,28 the
accumulation of “hidden” rearrangement products of nonspectacular
composition in the course of multiple fragmentation events may
actually lead to an accumulation of rearrangement products in the
multistage fragment spectrum (see Figure 5A).
most cases not labeled with the registered mass. This makes
reinvestigation and interpretation of tandem mass spectra par-
ticularly difficult. There is clearly a custom of not giving masses
of peaks that could not be assigned by the investigators. We would
like to stimulate the use of mass labels for assigned peaks as well
as nonassigned peaks.
Awareness of the possibility of hexose rearrangements in
fragmentation of glycopeptides is expected to be important for de
novo glycopeptide structural analysis, which may be performed
by manual interpretation or with the aid of software tools. Notably,
software for the interpretation of glycopeptide MS/MS data should
be able to deal with the fact that some fragments might not directly
represent structural motifs of the intact glycopeptide but rather
correspond to rearrangement products. These signals should,
therefore, not be mistaken as indicator of the presence of minor
isomers in glycopeptide mixtures.
ACKNOWLEDGMENT
We thank L. Renee Ruhaak for critically reading the manu-
script. We are grateful to Cornelis H. Smit for assistance with the
analysis of the LC-MS/MS data of oligomannosidic glycans.
We studied a range of tandem mass spectra of N-glycopeptides
from the literature for putative hexose rearrangement products
(Table 1). Unfortunately, the candidate rearrangement peaks,
which we saw in many literature tandem mass spectra, were in
Received for review February 5, 2009. Accepted April 7,
2009.
AC900278Q
4432 Analytical Chemistry, Vol. 81, No. 11, June 1, 2009