Báo cáo khoa học: The N-linked oligosaccharides of aminopeptidase N from Manduca sexta pptx

The N-linked oligosaccharides of aminopeptidase N from
Manduca sexta
Site localization and identification of novel N-glycan structures
Elaine Stephens
1
, Jane Sugars
2
, Sarah L. Maslen
1
, Dudley H. Williams
1
, Len C. Packman
2
and David J. Ellar
2
1
Department of Chemistry and
2
Department of Biochemistry, University of Cambridge, UK
Mass spectrometric studies on the N-linked glycans of
aminopeptidase 1 from Manduca sexta have revealed
unusual s tructures not previously observed on any insect
glycoprotein. Structure elucidation of these oligosaccha-
rides was carried out by high-energy collision-induced
dissociation (CID) using a matrix-assisted laser desorp-
tion/ionization time-of-flight/time-of-flight (MALDI-TOF/
TOF) tandem mass spectrometer. These key experiments
revealed that three out of the four N-linked glycosylation
sites in this protein (Asn295, Asn623 and Asn752) are
occupied with highly fucosylated N-glycans that possess
unusual d ifucosylated cor es. Cross-ring fragment ions

and ÔinternalÕ fragment ions observed in the CID spectra,
showed that these fucoses are found at the 3-position of
proximal GlcNAc and at the 3-position of distal GlcNAc
in the chitobiose unit. The latter substitution has only
been previously observed in nematodes. In addition, these
core structures can be decorated with novel fucosylated
antennae composed of Fuca(1-3)GlcNAc. Key frag-
ment ions revealed that these antennae are predominantly
found on the upper 6-arm of the core mannose. The
paucimannosidic N-glycan (Man
3
GlcNAc
2
), commonly
found on other insect glycoproteins, is the predominant
oligosaccharide found at the remaining N-glycosylation
site (Asn609).
Keywords: aminopeptidase N; fucose; high-energy CID;
insect glycosylation; MALDI-TOF/TOF.
As in other eukaryotic cells, many of the proteins in insect
cells are covalently modified b y N-glycosylation. The
protein N-glycosylation pathway in insect cells is similar
tothatobservedinmammaliancells(reviewedin[1–3]).
Each begins with the c otranslational transfer of a dolichol-
linked precursor oligosaccharide to a specific recognition
sequence (Asn-X-Ser/Thr; where X can be any amino acid
except Pro) within newly synthesized glycoproteins. Exo-
glycosidases and glycosyltransferases i n the endoplasmic
reticulum and Golgi apparatus catalyze trimming and
elongation reactions to yield th e common intermediate,

GlcNAcMan
3
GlcNAc
2
-N-Asn. In mammalian cells, ter-
minal glycosyltransferases can elongate this common inter-
mediate into the more elaborate structures of hybrid and
complex-type N-glycans. In contrast, insect cells have only
extremely low levels of the terminal glycosyltransferase
activities and, in most cases, have a competing exoglyco-
sidase that can r emove the terminal N-acetylglucosamine
(GlcNAc) residue from the G lcNAcMan
3
GlcNAc
2
-N-Asn.
Hence, the major processed N-glycan produced by insect
cells is usually the paucimannosidic structure, Man
3
Glc-
NAc
2
-N-Asn. T his is the case for t he honeybee glycoprotein
phospholipase A
2
(PLA
2
), where the major N-glycan
structures are composed of the common paucimannosidic
structure Man

3
GlcNAc
2
-N-Asn an d t he truncated stru cture
Man
2
GlcNAc
2
-N-A sn. However, a sma ll proportion of th e
N-linked glycan population on this protein was shown to
be substituted with fucosylated lacdiNAc [GalNAcb1-
4(Fuca1-3)GlcNAc] a ntennae a nd difucosylated c ores
[4,5]. The discovery of these structures demonstrated that
insect cells are capable of producing more elaborate
N-glycan structures. However, this is one of the f ew
endogenous insect glycoproteins that has been rigorously
characterized and the full structural diversity of the
N-glycan core and terminal structures that exist on insect
glycoproteins remains to be established.
It is imp ortant to understand insect protein glycosylation
because insects occupy an important evolutionary position
among eukaryotic o rganisms. Additionally, i nsect cells play
an important biotechnological role as hosts for t he produc-
tion of recombinant mammalian glycoproteins [3,6]. In the
context of this application, it is crucial t o realize the
differences between insect and mammalian N-glycan pro-
cessing pathways because N-glycans are known to influence
many different protein properties and functions, which
include enzymatic activity, conformation and immuno-
genicity. For example, the Fuca(1-3)GlcNAc substitution,

Correspondence to E. Stephens, Department o f Chemistry, U niversity
of Cambridge, Lensfield Road, Cambridge, CB2 1EW, UK.
Fax: +44 1223 336913; Tel.: +44 1223 336688;
E-mail:
Abbreviations: APN, aminopeptidase N; CHCA, a-cyano-
4-hydroxycinnamic acid; CID, collision-induced dissociation; DHB,
2,5-dihydroxybenzoic acid; GPI, glycosyl phosphotidylinositol; PLA
2
,
phospholipase A
2
;PNGaseA,peptideN-glycosidase A; Hex, hex ose;
HexNAc, N-acetylhexosamine; Fuc, fucose; GlcNAc, N-acetylgluco-
samine.
Enzymes: trypsin (EC 3.4.21.4); PNGase A (EC 3.5.1.52); PNGase F
(EC 3.5.1.52); a-
L
-fucosidase (EC 3.2.1.51); b-N-acetylhexosamini-
dase (EC 3.2.1.30).
(Received 2 2 July 2004, revised 6 September 2004,
accepted 8 September 2004)
Eur. J. Biochem. 271, 4241–4258 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04364.x
found in a small proportion of N-glycans on honeybee
PLA
2
is the major allergenic determinant in this protein
[4,5,7], and individuals who are allergic to honeybee venom
produce high levels of antibodies to the 3-linked fucose
moiety [7,8].
In the present study, w e have characterized the N-glycans

attached to the insect protein aminopeptidase N (APN1)
from the lepidopteran Manduca sexta (tob acco hornworm)
by mass spectrometry. This protein, which is found in the
brush border m embrane of the larval midgut, is attached to
the cell m embrane via a glycosyl phosphotidylinositol (GPI)
anchor [9]. Amino a cid sequence a nalysis shows four possible
N-glycosylation sites in APN1 (at asparagines 295, 609, 623
and 752) as well as several likely sites for O-glycosylation,
with 10 of these being in the C-terminal region of the protein
[10]. Importantly, the N- and/or O-linked glycans on this
membrane-associated protein are thought to mediate the
binding of the delta-endotoxins (Cry toxins) produced
as crystalline inclusions by the Gram-positive bacterium
Bacillus thuringiensis (Bt) [9,11]. These Cry toxins are
environmentally friendly insecticides and currently provide
protection against a wide range of insects, from beetle and
caterpillar crop pests to mosquito larvae [9,12]. The
sequences of four distinct APN isoforms (APN1–4) from
M. sexta larval midguts are available. Whilst APN2 is
known also to bind Cry1 toxins [13,14], no binding partners
have yet been identified for either APN3 or APN4.
The total monosaccharide composition of APN (inclu-
ding the N-linked glycans, O-linked glycans and the GPI
anchor) has been shown to include large amounts of
GlcNAc and Man with lesser amounts of GalNAc and
Fuc [10]. In addition, APN1 appeared mainly resistant to
deglycosylation with peptide N-glycosidase F (PNGase F),
indicating that the majority of glycans are substituted with
a1-3 linked fucose at the Asn-linked GlcNAc residue [15].
Furthermore, lectin binding studies have suggested the

presence of fucosylated paucimannose-type N-glycans [10].
However, a detailed site-specific structural characterization
of the N-linked glycans on APN has not been carried out,
nor is i t known which part of the glycoprotein interacts with
the toxin.
The initial purpose of the present s tudy was to provide a
thorough site-specific identification of the N-glycan struc-
tures on APN1 in order to reveal possible b inding sites for
Cry 1Ac. However, these studies revealed the presence of
extremely unusual complex-type glycan structures that can
contain up to four antennae composed of Fuca(1-3)Glc-
NAc, which have not been observed previously on any
glycoprotein. These novel antennae decorate the triman-
nosyl core, which is itself unusually difucosylated, with one
fucose attached to each of the c hitobiose G lcNAc residues.
Mass spectrometric analysis of N-linked glycopeptides
showed that these novel glycans are the major structures
at three out of the four N-glycosylation sites on APN1.
Experimental procedures
Purification of aminopeptidase N
The 120 kDa APN membrane protein was purified from
M. sexta brush border membrane vesicles as described
previously [9].
Alkylation and tryptic digestion
Approximately 200 lg of purified APN and 1 mg o f
honeybee phospholipase A
2
(PLA
2
, Sigma) were reduced,

carboxymethylated and digested with trypsin as described
in [16].
HPLC separation
The glycopeptide/peptide mixtures were purified by reverse-
phase HPLC on a Hewlett-Packard 1050 HPLC equipped
with a Jupiter C18 column (250 · 4.6 mm, Phenomenex,
Macclesfield, Cheshire, UK). Solvent A was 0.1% (v/v)
trifluoroacetic acid in H
2
OandsolventBwas90%
acetonitrile containing 9.90% H
2
O and 0.1% trifluoroacetic
acid (v/v/v). The column was equilibrated with 100% A,
and the gradient was initiated 10 min after injection and
increased linearly to 80% B ove r 120 min. The flow rate was
1mLÆmin
)1
, and the elution was monitored at 214 nm.
Fractions were collected at 1 min intervals and dried down.
These fractions were redissolved in 4 0 lL o f 5 0% ace tonit-
rile/H
2
O (v/v) containing 0.1% (v/v) trifluoroacetic acid,
and 1 lL aliquots were screen ed for glycopeptides and
peptides by MALDI-TOF MS.
Deglycosylation of glycopeptides
Pooled HPLC fractions containing purified glyco peptides
from PLA
2

were dige sted with PNGase F (Roche) in
ammonium bicarbonate buffer ( 50 m
M
,pH8.4)for24hat
37 °C using 3 U of enzyme. The released glycans w ere
separated from PNGase F-resistant glycopeptides by puri-
ficationusingaSep-PakC18(WatersLtd,Elstree,
Hertfordshire, UK). The PLA
2
glycopeptide-containing
fraction and purified glycopeptides from APN1 were
digested with peptide N-glycosidase A (PNGase A; Roche)
in ammonium acetate buffer (50 m
M
,pH5.0),for16hat
37 °C using 0.2 mU of enzyme. The p roducts were purified
using a Sep-Pak C18.
Exoglycosidase digestions
These were carried out on glycopeptides using the following
enzymes and conditions: a-
L
-fucosidase (from bovine kid-
ney; Sigma), 0.2 U in 100 lLof50m
M
ammonium acetate
buffer, pH 5; b-N-acetylhexosaminidase (from Jack b ean;
GlykoÒ, Prozyme, San Leandro, CA, USA), 60 mU in
50 m
M
sodium citrate phosphate buffer, pH 5. Both

digestions were incubated at 3 7 °C for 24 h, after which,
the reactions were terminated by boiling for 1 min. The
products were dried down a nd analyzed by nanoLC-
MALDI-TOF MS.
NanoLC-MALDI-TOF MS
NanoLC was carried out using an LC-Packings Ultimate
HPLC (Dionex, Leeds, UK), which was used to generate
the g radient that flowed at 150 nLÆmin
)1
. T he glycopeptide
containing fraction, produced from sequential exoglycosi-
dase digestions, was dried down, resuspended in 40 lLof
0.1% (v/v) trifluoroacetic acid in H
2
Oand5lLofthe
sample was loaded onto a nanoLC column (C18,
4242 E. Stephens et al.(Eur. J. Biochem. 271) Ó FEBS 2004
75 lm · 5 cm; LC-Packings, Dionex) and the glycopep-
tides were eluted with increasing organic concen tration.
Solvent A was 0.1% (v/v) trifluoroacetic acid in H
2
Oand
solvent B was 90% acetonitrile containing 9.90% H
2
Oand
0.1% trifluoroacetic acid (v/v/v). The column was equili-
brated with 100% A, and the gradient was increased linearly
to 80% B over 60 min. The column effluent passed through
a UV detector (monitoring at 214 nm) directly to a Probot
sample fraction system (Dionex). This system mixed the

LC effluent with matrix [a-cyano-4-hydroxycinnamic acid
(CHCA);Sigma;5mgÆmL
)1
in 50% acetonitrile/0.1%
trifluoroacetic acid (v/v)] prior to s potting onto a 4700
MALDIplateat30sintervals.Thematrixspotswere
allowed to air dry prior to MALDI-TOF MS.
Chemical derivatization
Permethylation of PNGase A released glycans was per-
formed using the sodium hydroxide procedure and subse-
quently purified by Sep-Pak using an acetonitrile gradient
as described in [17].
MALDI-TOF MS and MALDI-MS/MS
All mass spectra were recorded on a 4700 Proteomics
analyzer with TOF/TOF optics (Applied Biosystems, Foster
City, CA, USA). This MALDI tandem mass spectrometer
used a 200 Hz frequency-tr ipled N d:YAG l aser operating at
a wavelength of 355 nm. The average of 1000 laser shots
was used t o obtain the MS and MS/MS spectra using
CHCA matrix. The MS/MS spectra obtained using 2,5-
dihydroxybenzoic acid (DHB; Fluka, Dorset, UK) matrix
were acquired using 5000–10 000 laser shots. For protein
identification experiments, database searching was carried
out against the NCBI protein database using
GPS EXPLORER
(Applied Biosystems). The mass spectrometric data used for
protein identification was provided by automated MS and
MS/MS acquisitions of peptides in HPLC purified fractions.
For MS/MS experiments, the collision e nergy, which is
defined by the potential difference between the source

acceleration voltage (8 kV) and the floating collision cell
(7 kV), was set at 1 kV. Inside the collision cell the selected
precursor io ns w ere c ollided with argon or air at a pressure
of 2 · 10
)6
Torr. Glycopeptide and p eptide samples p rovi-
ded in HPLC fractions were mixed 1 : 1 (v/v) CHCA
solution [10 mgÆmL
)1
in 50% acetonitrile/H
2
O (v/v) con-
taining 0.1% (v/v) triflu oroacetic acid] a nd spotted directly
onto the target plate and allowed to air dry. Permethylated
glycans for MS/MS were analyzed using DHB matrix
[10 mgÆmL
)1
in 50% (v/v) methanol/H
2
O].
Results
General strategies for glycopeptide characterization
The general mass spectrometric strategy for the character-
ization of the glycans and the identification of all glycos-
ylated sites of APN1 is summarized in Fig. 1. Reduced and
alkylated A PN was digested with t rypsin and the proteolytic
peptides were separated by reverse-phase HPLC. A small
aliquot of each HPLC fraction was analyzed by MALDI-
TOF MS in linear and reflectron mode, a nd those f ractions
containing glycopeptides from APN1 were identified. A

distinct feature of g lycans is their h eterogeneity. As a result,
glycopeptides tend to yield a series of low intensity peaks
in a MALDI-TOF MS spectrum, with mass differences
corresponding to sugar residues. Where possible, these
putative glycopeptides w ere identified b y high-energy
collision-induced dissociation (CID) of selected signals
[M + H]
+
, using a MALDI-TOF/TOF tandem mass
spectrometer. The resulting CID mass spectra gave minor
signals for fragment ions derived from the peptide portion,
which facilitated the identification of the g lycosylated
peptide. Those fractions containing unusual glycopeptides
from APN1 were pooled and treated with PNGase A. The
released glycans and peptides were separated, and the
peptide-containing fraction was analyzed b y MALDI-TOF
MS to co nfirm t he presence of the deglycosylated p eptides.
The glycan-containing fraction was permethylated and
major oligosaccharide signals were subjected to MALDI-
TOF/TOF high-energy CID. Fragment ions derived
glycosidic and high-energy cross-ring cleavages revealed
carbohydrate sequence and linkage information.
Peptide mapping
The HPLC chromatogram o f the tryptic digestion of APN
exhibited a low UV absorbance at 214 nm indicating that
subpicomolar quantities of peptides were present (data not
shown). Despite this, a nalysis of each H PLC fraction of the
APN d igest b y M ALDI-TOF MS revealed tryptic peptides
that covered  70% of t he sequence of APN1. Among these
peptides, two gave very minor signals at m/z 2707.324 and

m/z 2240.107, corresponding to tryptic peptides spanning
potential N-g lycosylation sites at Asn295 and Asn752 (data
not shown). These data indicate that these glycosylation
sites are either not occupied or partially glycosylated.
Fig. 1. Procedure for the site-specific characterization of APN N-gly-
cosylation.
Ó FEBS 2004 Novel insect N-glycans from Manduca sexta (Eur. J. Biochem. 271) 4243
Signals f or peptides spanning the remaining N-glycosylation
sites (Asn609 and Asn623) were not observed, suggesting
that these sites are f ully glycosylated. In a ddition to signals
corresponding to peptides from APN1, many major signals
from the digest mixture could not be mapped onto the
APN1 amino acid sequence. Consequently, each HPLC
fraction was reanalyzed by M ALDI-TOF MS and auto-
mated MS/MS was carried out on the five most abundant
peaks in each HPLC f raction. These MS and MS/MS data
were submitted for database searching against the NCBI
protein database using
GPS EXPLORER
software. Among the
 100 peptides that were chosen for MS/MS, 38 peptides
were from APN1 (a ccession number 2499901). T he remain-
ing peptides w ere derived from aminopeptidas e 4 ( M. sexta;
accession number 20260704; 20 peptides), aminopeptidase 3
(M. sexta; accession number 20279109; 10 pept ides) a nd the
Bacillus thuringiensis Cry 1Ac protein (accession number
143126; 20 peptides). The latter protein (Cry 1Ac) is a likely
contaminant as this protein is used to purify the APN
protein by affinity chromatography. However, the presence
of the two M. sexta aminopeptidase proteins (APN3 and

APN4) in the digestion m ixture complicated the character-
ization of the N-glycans on the major APN1 protein, as
inspection of their amino acid sequences revealed seven
consensus sequences for N-glycosylation in APN3 and five
in APN4. Therefore, these proteins are likely to be heavily
glycosylated and careful analysis of any glycopeptides
detected was necessary to ensure the correct assignment
was given. The following sections describe the detailed mass
spectrometric characterization of glycopeptides from
APN1, which is the major protein in the digestion mixture.
Glycopeptide analysis of APN1
Residue N295. MALDI-TOF MS analysis of HPLC
fraction number 6 8 in linear m ode revealed a s eries of
signals (from m/z 3700 to m/z 4800) (Fig. 2A). The mass
differences between these peaks were 146 Da (deoxyhexose)
and 203 Da (N-acetylhexosamine; HexNAc) that are char-
acteristic of glycopeptides. Subtraction of the mass of the
peptide LLLAMENYTAIPYYTMAQNLDMK(289–311)
(m/z
avg
2709.24 [M + H]
+
) from the molecular masses of
the glycopeptides observed, provided the oligosaccharide
residue compositions and stoichiometry summarized in
Table 1. Previous experiments in our laboratories have
shown that fucose is the only deoxyhexose found on APN
[10]. Taking this into account, these data suggest that
the N-linked glycans at Asn295 have compositions consis-
tent with fucosylated paucimannosidic core structures

(Fuc
1)2
Hex
2)3
HexNAc
2
) and a difucosylated structure
bearing one short antenna composed of HexNAc
(Fuc
2
Hex
3
HexNAc
3
). These glycans have been observed
previously in insect glycoproteins [3,6,18]. However, abun-
dant oligosaccharide components consisting of Hex
3
Hex-
NAc
3
Fuc
3
(m/z 4243.82 [M + H]
+
), Hex
3
HexNAc
4
Fuc

3
(m/z 4447.14 [M + H]
+
)andHex
3
HexNAc
4
Fuc
4
(m/z
4593.33 [M + H]
+
) were surprising because they corres-
pond to no known N-glycan s tructures. These assignments
were corroborated by analysis of the released peptide after
treatment with PNGase A. After treatment with this
enzyme, the deglycosylated peptide is one mass unit larger
than the expected mass of its unglycosylated counterpart
because PNGase A converts Asn (114 Da) (to which the
oligosaccharides had been attached) to Asp (115 Da) [15].
The MALDI-TOF mass spectrum shown in Fig. 2B
exhibits signals for the released peptide spanning residues
L289–K311. This peptide contains three methionine resi-
dues, which have all partially oxidized during storage to give
four signals at m/z 2708.3313 [M + H]
+
, m/z 2724.3005
[M + H]
+
, m/z 2740.3098 [M + H]

+
and m/z 2756.3022
[M + H]
+
. The expected monoisotopic masses for the
Fig. 2. MALDI-TOF MS of HPLC fraction number 68. (A) Linear
MALDI mass spectrum of the glycopept ides spanning residu es
Leu289–Lys311 from APN1, which contain the consensus sequence
for N-linked glycosylation at Asn295. Mass differences corresponding
to fucose (146 Da) and HexNAc (203 Da) are observed. (B) Reflectron
MALDI mass spectrum of methionine oxidized peptides spanning
residues Leu289–Lys311 after tr eatment with PNGase A.
Table 1. Glycopeptides L289–K311 spanning Asn295 observed by
MALDI-TOF M S. Average masses are i ndicated a nd m ajor c ompo-
nents are highlighted in bold text.
Observed
m/z [M + H]
+
Measured
carbohydrate
residue
mass (Da)
Calculated
carbohydrate
residue
mass (Da)
Carbohydrate
assignment
3732.01 1022.77 1022.96 Hex
2

HexNAc
2
Fuc
2
3748.21 1038.97 1038.96 Hex
3
HexNAc
2
Fuc
3894.05 1184.81 1185.10 Hex
3
HexNAc
2
Fuc
2
4097.60 1388.36 1388.30 Hex
3
HexNAc
3
Fuc
2
4243.82 1534.58 1534.44 Hex
3
HexNAc
3
Fuc
3
4447.14 1737.90 1737.64 Hex
3
HexNAc

4
Fuc
3
4593.33 1884.09 1883.78 Hex
3
HexNAc
4
Fuc
4
4244 E. Stephens et al.(Eur. J. Biochem. 271) Ó FEBS 2004
native, m ono-, di- and tri-oxidized peptide (in which Asn295
has been converted to A sp) are m/z 2708.3025 [M + H]
+
,
m/z 2724.2974 [M + H]
+
, m/z 2740.2920 [M + H]
+
,and
m/z 2756.2872 [M + H]
+
, respectively, giving a mass
accuracy of less than 11 p.p.m. in each case. These
assignments were confirmed by MALDI-CID of the most
abundant species at m/z 2740.3098 (data not shown).
Residue N609. The MALDI-TOF ma ss spectrum of the
HPLC fraction number 54 exhibited two signals that
differed in mass corresponding to a hexose r esidue
(Fig. 3A). Subtraction of the calculated monoisotopic mass
of the peptide VNYDNTTWGLITR(605–617) (m/z

mono
1552.7760 [M + H]
+
) from the measured masses of the
two glycopeptides observed, provided the oligosaccha-
ride residue compositions Hex
2
HexNAc
2
(m/z 2283.06
[M + H]
+
)andHex
3
HexNAc
2
(m/z 2445.14 [M + H]
+
).
These compositions correspond to nonfucosylated pauci-
mannosidic structures that have frequently been observed
as major constituents on other insect glycoproteins
[3,18–20]. These putative assignments were corroborated
by fragmentation of the glycopeptide at m/z 2445.14
Fig. 3. MALDI-TOF MS of H PLC fraction number 54. (A) Reflectron MALDI mass spectrum of the glycopeptides spanning residues Val605–
Arg617 from APN1, which contain the consensus sequence for N-linked glycosylation at Asn609. A mass difference corresponding to hexose
(162 D a) is observed. The s ignal at m/z 2441.14 corresponds to a nonglycosylated tryptic peptide Asn182–Arg203 from APN3. (B) MALDI-TOF/
TOF fragment ion m ass spec trum of the major glycopeptide Hex
3
HexNAc

2
-(Val605–Lys617) at m/z 2445.1389 [M + H]
+
.Theformationofthe
0,2
X
0
fragment is shown in the insert (R ¼ He x
3
HexNAc).
Ó FEBS 2004 Novel insect N-glycans from Manduca sexta (Eur. J. Biochem. 271) 4245
(Hex
3
HexNAc
2
-[V605–R617]) by MALDI high-energy
CID (Fig. 3B). The resulting CID spectrum is dominated
by fragment ions at high mass, which a re derived from the
cleavage of glycosidic linkages in the oligosaccharide
portion. The major signal at m/z 1 635.92 corresponds to a
0,2
X
0
fragment (Fig. 3B, insert), which has been observed
previously in the high-energy CID spectra of Asn-linked
glycopeptides [21]. Other signals at low mass are derived
from cleavage of the peptide portion and reveal peptide
sequence information (V605–R617).
Residue N623. MALDI-TOF MS analysis of two HPLC
fractions (fractions 43 and 44) gave a set of signals, which

differ by masses corresponding to 203 Da (HexNAc) and
146 Da (Fuc) (Fig. 4). Many sig nals w ere c ommon t o e ach
HPLC fraction (e.g. m/z 2770, m/z 2916, m/z 3322, etc.),
which s uggested that one set of heterogeneous glycopeptides
had e luted i nto t wo HPLC fractions. Surprisingly, MALDI-
CID of the two most intense signals in each fraction
revealed the presence of glycopeptides consisting of two
different peptide moieties. These peptides differ by a mass
that corresponds to a HexNAc residue (203 Da), which
significantly complicated the glycopeptide assignments.
Consequently, the other signals in each fraction were
subjected to MALDI-CID in order to confirm the identity
of each glycopeptide constituent. The CID spectra of the
smaller glycopeptide signals in fraction 43 exhibited non-
glycosylated peptide fragment ions that indicated the
peptide portions consist to SANRTVIHELSR(621–632)
(Fig. 5A) from APN1. However, MALDI-C ID of the
larger glycopeptides in that fraction (m/z 3322.55
[M + H]
+
, Fig. 5B) revealed the presence of a glycopep-
tide consisting of AFRNNNTLVPVNAR(739–752) from a
contaminating isoform of aminopeptidase (APN3). Subse-
quent MALDI-CID of the glycopeptides in HPLC fraction
44 revealed that the majority of signals are due to the latter
glycopeptide from APN3 (A739–R752) (Fig. 5C). How-
ever, a small proportion of the smaller, less polar glyco-
peptides in fraction 44 were shown to be from APN1 (e.g.
m/z 2568.19, m/z 2770.3 and m/z 2916.35). Consequently,
subtraction of the mass of the tryptic peptides span-

ning residues S621–R632 (APN1, m/z
mono
1382.7504
[M + H]
+
) and A739–R752 (APN3, monoisotopic mass
1585.8563 [M + H]
+
) from the measured mass of each
designated glycopeptide yielded the glycan compositions
summarized in Tables 2 and 3 for APN1 and APN3,
respectively. These data indicate that the N-linked glycans
at Asn623 on APN1 are similar in structure to those
unusual, highly fucosylated oligosaccharides found at
Asn295. Interestingly, most compositions for the N- glycans
from APN3 (Table 3) are the same as those found on
APN1, suggesting that similar structures are present.
However, one major glycopeptide at m/z 3176.49 [Hex
3
Hex-
NAc
4
Fuc
2
-(A739–R752)] is composed of a difucosylated
glycan moiety that is not observed in APN1. It is interesting
to note that the major
0,2
X
0

fragment ion, which was
observed in the fragment ion spectrum of Hex
3
HexNAc
2
-
(Val605–Arg617) (Fig. 3B) is not detected in the high-
energy CID spectra of the fucosylated glycopeptides
(Fig. 5). This cross-ring cleavage may be prevented by the
presence of fucose on the Asn-linked GlcNAc residue,
whichisindicatedineachspectrumbymajorfragments
corresponding to the peptide bearing HexNAcFuc.
Residue N752. M ALDI-TOF MS analysis o f fraction
number 63 revealed a series of signals (from m/z 3400 to
m/z 4400) (Fig. 6A) that differ by masses corresponding
to sugar residues (146 Da and 203 Da). Subtraction of the
calculated monoisotopic mass of the peptide spanning res-
idues NGSFIPANMRPWVYCTGLR(752–770) (m/z
mono
Fig. 4. MALDI mass spectra of HPLC fractions 43 (A) and 44 (B) containing the g lycopeptides spanning residues S er621–Arg632 (from APN1) and
Ala739–Arg752 (from APN3), respectively. Mass differe nces correspon ding to fucose (146 Da) and H exNAc (203 Da) are observed.
4246 E. Stephens et al.(Eur. J. Biochem. 271) Ó FEBS 2004
Fig. 5. MALDI-TOF/TOF fragment i on mass s pe ctra of ( A) the APN1 glycopeptide in f raction 43 [ Hex
3
HexNAc
3
Fuc
3
-(Ser621–Arg632)] at m/z
2916.32 [M + H]

+
, (B) the APN3 glycopeptide [Hex
3
HexNAc
4
Fuc
3
-(Ala739–Arg752)] in fraction 43 at m/z 3322.52 [M + H]
+
and ( C) the A PN3
glycopeptide [Hex
3
HexNAc
3
Fuc
2
-(Ala739–Arg752)] in fraction 44 a t m/z 2973.39 [M + H ]
+
.
Ó FEBS 2004 Novel insect N-glycans from Manduca sexta (Eur. J. Biochem. 271) 4247
2240.0745 [M + H]
+
) revealed the glycan compositions
summarized in Table 4. These assignments were corro-
borated by MALDI-CID o f selecte d signals, where
fragment ions from the peptide backbone confirmed the
sequence of the peptide portion (Fig. 6B). These data
indicate that the N-linked glycans at Asn752 are similar to
those highly fucosylated oligosaccharides found at Asn295
and Asn623. Analysis of the released peptide fraction after

treatment with PNGase A r evealed minor signals for the
native and methionine oxidized deglycosylated peptides
N752–R770 with Asn752 converted to Asp. However,
major signals for its glycosylated counterparts were also
observed indicating that these glycopeptides are m ostly
resistant to glycopeptidase A digestion (data not shown).
This is probably due to the presence of the free amino
group at Asn752 [22,23].
Structure elucidation of released N-glycans
by MALDI-TOF/TOF tandem mass spectrometry
The PNGase A-released glycan fractions containing un-
usual, highly fucosylated oligosaccharides from Asn295
and Asn623 were pooled and analyzed by MALDI-TOF
MS after permethylation (Fig. 7 and Table 5). In addition
to the major glycan components (Fuc
2-4
Hex
3
HexNAc
2)5
),
these data reveal minor amounts of large, highly fucos-
ylated N-glycans (Fuc
5–6
Hex
3
HexNAc
5)6
), which were
not detected in the g lycopeptide data. Many of these

N-glycan compositions (Fuc
3-6
Hex
3
HexNAc
3)6
) indicate
that difucosylated trimannosyl core structures, decorated
with fucosylated antennae, are likely to be present on
APN1.Detailedstructuralcharacterization o f the novel
highly fucosylated N-glycans was achieved by high-energy
CID of the major oligosaccharide signals using a
MALDI-TOF/TOF tandem mass spectrometer. Previous
work on native and permethylated N-glycans have shown
that this instrument can reveal unambiguous oligosaccha-
ride sequence, as well as linkage and branching informa-
tion [24–26].
Characterization of core fucosylation
In order to locate the fucose substitutions on the chito-
biose core, MALDI high-energy CID was carried out
on the glycan at m/z 1519.8 [M + Na]
+
,whichis
consistent with a difucosylated paucimannosidic struc-
ture (Fuc
2
Hex
3
HexNAc
2

) (Fig. 8A). As depicted in
Scheme 1A, structurally informative fragment ions include
1,5
X
1a
and Y
1a
(Domon & Costello nomenclature [27]),
which show that only one fucose is attached to the
reducing-end GlcNAc in the chitobiose core. Previous
studies have shown t hat the N-glycans on APN are
resistant to PNGase F, suggesting that this fucose is likely
to be attached to the C-3 position of proximal GlcNAc.
However, the presence of glycans bearing 6-linked fucose
on Asn-linked GlcNAc cannot be ruled out. Importantly,
the cross-ring fragment ion,
1,5
X
2a
, indicates that one
other fucose is attached to distal GlcNAc of the
chitobiose unit. This assignment is verified by two majo r
ÔinternalÕ fragment ions at m/z 671.32 and m/z 880.38. The
former ion (m/z 671.32) is the likely product of a Z-type
cleavage (Z
2
) followed by the elimination of fucose from
C-3 of distal GlcNAc [24]. The latter ion (m/z 880.38) is
likely to be derived from a C-type cleavage (C
3

) followed
Table 2. Glycopeptides fr om APN1 (S621–R632) spanning Asn 623 observed by MALDI-TOF MS. Major components are noted in bold text and
monoisotopic masses are indicated.
m/z [M + H]
+
Measured carbohydrate
residue mass (Da)
Calculated carbohydrate
residue mass (Da)
Carbohydrate
assignment HPLC fraction
2567.191 1184.441 1184.443 Hex
3
HexNAc
2
Fuc
2
43
a
+44
2770.279 1387.529 1387.512 Hex
3
HexNAc
3
Fuc
2
43
a
+44
a

2916.323 1533.573 1533.570 Hex
3
HexNAc
3
Fuc
3
43
a
+44
a
3119.425 1736.675 1736.650 Hex
3
HexNAc
4
Fuc
3
43
a
3265.452 1882.702 1882.708 Hex
3
HexNAc
4
Fuc
4
43
a
a
Assignments corroborated by MALDI-CID of glycopeptides.
Table 3. Glycopeptides from APN3 (A739–R752) spanning Asn743 observed by MALDI-TOF MS. Monoisotopic ma sses are i ndic ated and m ajor
components are noted in bold t ext.

m/z [M + H]
+
Measured carbohydrate
residue mass (Da)
Calculated carbohydrate
residue mass (Da)
Carbohydrate
assignment HPLC fraction
2770.310 1184.450 1184.443 Hex
3
HexNAc
2
Fuc
2
44
a
2973.388 1387.530 1387.512 Hex
3
HexNAc
3
Fuc
2
44
a
3119.492 1533.64 1533.570 Hex
3
HexNAc
3
Fuc
3

44
a
3176.489 1590.630 1590.590 Hex
3
HexNAc
4
Fuc
2
43
a
+44
a
3322.550 1736.68 1736.650 Hex
3
HexNAc
4
Fuc
3
43
a
+44
a
3468.630 1882.77 1882.708 Hex
3
HexNAc
4
Fuc
4
43 + 44
a

3671.749 2085.890 2085.790 Hex
3
HexNAc
5
Fuc
4
43 + 44
a
a
Assignments corroborated by MALDI-CID.
4248 E. Stephens et al.(Eur. J. Biochem. 271) Ó FEBS 2004
by the elimination of fucose from C-3 (Scheme 2). These
mechanisms are probable b ecause previous mass spectro-
metric studies on native and permethylated glycans have
shown that there is a preferential loss of fucose from the
C-3 position of GlcNAc [17,24,25,28,29]. These assign-
ments were corroborated by comparing the fragment ions
observed to those formed (under the same conditions)
from a similar permethylated glycan purified from honey-
bee PLA
2
. This paucimannosidic glycan has the same
monosaccharide composition as the glycan from APN1 at
m/z 1519.8 (Hex
3
HexNAc
2
Fuc
2
). However, this structure,

which was previously characterized by NMR [30], has a
difucosylated core with the two fucose residues linked via
C-6 and C-3 to the Asn-linked GlcNAc. The resulting
high-energy CID spectrum (shown in Fig. 8B) exhibited
different fragment ions from those observed for the APN1
glycan. In particular, the signals for the monofucosylated
Y
1a
(m/z 474.23) and
1,5
X
1a
(m/z 502.22) fragments were
absent. Instead, major signals were observed at m/z 648.25
and m/z 676.26 for the difucosylated Y
1a
and
1,5
X
1a
fragments, respectively. Moreover, the B
3
ion shows t hat
the distal GlcNAc residue is not fucosylated (Scheme 1B)
and the major Ôinternal Õ ion at m/z 671.32 in Fig. 8A
(produced by elimination of fucose from the C-3 position
of the Z
2
ion) is absent. However, a major signal for the
ÔinternalÕ fragment at m/z 880.3 is still observed. This ion

Fig. 6. MALDI-TOF M S of HPLC fraction number 63. (A) MALDI-TOF m ass spectrum exhibiting signals fo r the t ryptic glycopeptides spanning
residues Asn752–Arg770 from APN1, which contain the consensus sequence for N-linked glycosylation at Asn752. (B) MALDI-TOF/TOF
fragment ion mass s pectrum of the major g lycopeptide Hex
3
HexNAc
3
Fuc
3
-(Asn752–Arg770) at m/z 3773.63 [M + H]
+
.
Ó FEBS 2004 Novel insect N-glycans from Manduca sexta (Eur. J. Biochem. 271) 4249
is likely to be formed by the same mechanism as shown in
Scheme 2 but, in this case, methanol is eliminated from
the C-3 position of GlcNAc instead of fucose. Together,
these data unequivocally establish the attachment of
fucose at the 3-position of distal G lcNAc in the difucos-
ylated paucimannosidic structure from APN1.
Characterization of fucosylated antennae
The monosaccharide compositions of the larger N-glycans
observed (Fuc
3)6
Hex
3
HexNAc
3)6
), suggested that these
structures are decorated with antennae consisting of Hex-
NAc and fucose. Previous studies on the glycans from
honeybee PLA

2
have shown that fucosylated lacdiNAc
[GalNAcb1-4(Fuca1-3)GlcNAc] antennae are present o n a
small population of N-glycans [4,5]. Antennae composed
solely of GlcNAc-Fuc were not observed. However, the
monosaccharide composition of the major N-glycan from
APN1 at m/z 1939.0 (Fuc
3
Hex
3
HexNAc
3
) indicates that
the latter novel antennae (GlcNAc-Fuc) is present on this
structure. Consequently, the existence of this antenna was
investigated by high-energy CID of this glycan and the
resulting fragment ion spectrum is shown in Fig. 9A. The
presence of two fucose residues on both GlcNAc residues
in the chitobiose unit (Scheme 3A) was demonstrated by
the characteristic fragment ions Y
1a
,
1,5
X
1a
,and
1,5
X
2a
.In

addition, this spectrum gave signals for
3,5
A
5
and
3,5
A
6
,
which show that both fucoses are linked via C-3 to each
GlcNAc residue in the core. Other important, structurally
informative fragment ions include the B
2a
ion (m/z 456.32)
and the cross-ring fragment a t m/z 1329.38 (
1,5
X
3a/b
). These
Table 4. Glycopeptides Asn752–Arg770 spanning Asn752 observed by
MALDI-TOF MS. Monoisotopic masses are indicated and major
constituents are noted in b old text.
m/z [M + H]
+
(monoisotopic)
Measured
carbohydrate
residue mass
(Da)
Calculated

carbohydrate
residue mass
(Da)
Carbohydrate
assignment
3424.5145 1184.440 1184.443 Hex
3
HexNAc
2
Fuc
2
3627.5753 1387.501 1387.512 Hex
3
HexNAc
3
Fuc
2
a
3773.6285 1533.554 1533.570 Hex
3
HexNAc
3
Fuc
3
a
3976.7195 1736.645 1736.650 Hex
3
HexNAc
4
Fuc

3
4122.7215 1882.647 1882.708 Hex
3
HexNAc
4
Fuc
4
4325.7823 2085.708 2085.790 Hex
3
HexNAc
5
Fuc
4
a
Assignments corroborated by MALDI-CID of glycopeptides.
Fig. 7. MALDI-TOF mass s pectra exhibiting major signals for permethylated glycans f rom Asn295 and Asn6 23 from APN1. N-glycans were r eleased
from APN1 tr yptic glycopeptid es by d ige stion with PNGase A. The released N -gly cans were permethylated and p urifi ed by Se p-Pa k. The 35%
(v/v) (A) and the 50% (v/v) (B ) a cetonitrile fractio ns w ere s creened b y M ALDI-TOF MS. T he insert shows an expanded r egion exhibiting m inor
signals for the high m ass components. Signals marked withanasteriskareglucosepolymercontaminants.
4250 E. Stephens et al.(Eur. J. Biochem. 271) Ó FEBS 2004
ions show that the glycan has o ne antenna composed of
Fuc-HexNAc. The absence of the ion at m/z 456.3 (B
2a
)in
the high-energy spectrum of the difucosylated pauciman-
nosidic structure (Fig. 8A) supports this assignment. Fur-
thermore, the signal at m/z 268.12 corresponds to the
elimination of fucose f rom C-3 of the C
2a
ion (by a s imilar

mechanism shown in Scheme 2), which indicates that this
fucose is linked via C-3 of HexNAc. Elimination of sugar
residues from the C-3 position of fragment ions containing
HexNAc is a common feature of fast-atom bombardment
(FAB) spectra [17,31]. More recently, analysis of both
permethylated [24] and native oligosaccharides [25] by high-
energy CID using a MALDI-TOF/TOF tandem mass
spectrometer have d emonstrated the presence o f major ions
derived from the preferential elimination of substituents
attached to the C-3 position. Furthermore, the latter study
[25] was a ble t o diffe rentiate between the LNFP II [Fuc a1-
Table 5. Assignments of molecular ions observed i n the MALDI-TOF
spectra o f permethylated N-glycans released from Asn295 and Asn62 3.
Major components are highlighted in bold t ext.
Observed
m/z [M + Na]
+
(monoisotopic)
Calculated
m/z [M + Na]
+
(monoisotopic)
Carbohydrate
assignment
1345.68 1345.67 Hex
3
HexNAc
2
Fuc
1519.80 1519.76 Hex

3
HexNAc
2
Fuc
2
1764.94 1764.89 Hex
3
HexNAc
3
Fuc
2
1939.02 1938.98 Hex
3
HexNAc
3
Fuc
3
2184.18 2184.10 Hex
3
HexNAc
4
Fuc
3
2358.24 2358.19 Hex
3
HexNAc
4
Fuc
4
2603.34 2603.32 Hex

3
HexNAc
5
Fuc
4
2777.45 2777.41 Hex
3
HexNAc
5
Fuc
5
3022.61 3022.53 Hex
3
HexNAc
6
Fuc
5
3196.71 3196.62 Hex
3
HexNAc
6
Fuc
6
Fig. 8. High-energy CID spectra of the [M + Na]
+
ions of permethylated glycans Hex
3
HexNAc
2
Fuc

2
from (A) APN1 and (B) honeybee
phospholipase A
2
. Sequence and linkage informative fragment ions for (A) a nd (B) are assigned in Scheme 1. Signals labeled with an asterisk are
derived from internal cleavages due to elimination of substituents from around the pyranose ring [24].
Ó FEBS 2004 Novel insect N-glycans from Manduca sexta (Eur. J. Biochem. 271) 4251
4(Gala1-3)GlcNA cb1-3Galb1-4Glc] and LNFP III [Gala1-
4(Fuca1-3)GlcNAcb1-3Galb1-4Glc] structures from hu-
man milk oligosaccharides, based on the elimination of
galactose and fucose, respectively, from the C-3 position of
GlcNAc. Therefore, the presence of the secondary fragment
ion at m/z 268.12 in Fig. 9A provides strong evidence f or a
Fuc(1-3)HexNAc linkage in the antenna. In addition, the
A-type cross-ring fragment ions,
0,4
A
4
and
3,5
A
4
, show that
this short fucosylated antenna is linked to the 6-arm of core
mannose. This assignment is confirmed by the presence of a
minor signal for
1,3
A
4
, w hich shows that only one mannose

residue is attached to C-3 of core b-mannose. Further
verification of t his substitution is provided by the dominant
signal at m/z 850.40 that is derived from the elimination of
mannose from the C-3 position of the C
4
ion, by a similar
mechanism shown in Scheme 2. Similar fragment ions
(nonreducing cross-ring fragments and i nternal ions), which
indicate th e presence of a structural isomer composed of one
antenna attached to the 3-arm of core mannose were not
observed, indicating that this structure is not present at a
detectable level.
The monosaccharide composition of another major
N-glycan on APN1 at m/z 2358.24 (Fuc
4
Hex
3
HexNAc
4
)
suggests the presence of difucosylated core structure decor-
ated with either: (a) one antenna composed of Hex-
NAc
2
Fuc
2
, (b) two antenn ae c omprised of HexNAcFuc
2
and HexNAc, or (c) two antennae consisting of HexNAc-
Fuc. The presence of a major fragment ion for B

2a/b
at m/z
456.3 (Fuc-HexNAc) a nd its c oncomitant secondary ion at
m/z 268.1 (elimination of fucose from C-3 of HexNAc) in
the high-energy CID spectru m of this glycan (Fig. 9B)
indicates that the latter structural possibility (c) is likely. In
addition, the absence of sign als for Fuc
2
HexNAc
2
(m/z 875;
B ion) and Fuc
2
HexNAc (m/z 630; B ion) support this
assignment. To corroborate this structural identification,
the fragmentation detected in Fig. 9B was c ompared to that
observed in a high-energy CID spectrum of a lacdiNAc
containing N-glycan from honeybee PLA
2
, acquired under
the same conditions (data not shown). In this case, signals
for the B
2
ion at m/z 456 and m/z 268 were a bsent. Instead,
major signals at m/z 701.20 (B
2
)forGalNAc1-4(Fuc1-
3)GlcNAc and m/z 513.11 (elimination of fucose from the
C-3 position of GlcNAc) w ere detected. The presence of
both these ions unequivocally show that p ermethylated

N-glycans fragment predictably under high-energy CID
conditions using a MALDI-TOF/TOF tandem mass spec-
trometer and the assignments given for the ions observed in
Fig. 9B are correct. Other key fragment ions in the CID
spectrum of the APN1 glycan are depicted in Scheme 3B.
Structurally informative fragment ions include
1,5
X
3a
,and
Y
3a
, which indicate that only one arm of the trimannosyl
core is substituted with the HexNAcFuc antennae. Import-
antly, the m inor signal for
3,5
A
4
shows that t hese two
Scheme 1. Glycosidic and cross-ring cleavages found in the high-energy
CID spectra of (A) Fuc
2
Hex
3
HexNAc
2
from APN1 (M. sexta) and (B)
Fuc
2
Hex

3
HexNAc
2
from PLA
2
(honeybee). High-energy fragment ions
are labeled using the nomenclature of Domon & Costello [27].
Scheme 2. Scheme showing the proposed route for the f ormation of the
major i nternal f ragment ion at m/z 880.3 i n the high-energy C ID spec-
trum of Fuc
2
Hex
3
HexNAc
2
from APN1.
4252 E. Stephens et al.(Eur. J. Biochem. 271) Ó FEBS 2004
antennae are linked to the upper m annose residue in the
core. This assignment is supported by the presence of a
major signal at m/z 1269.65 that is derived from a C
4
cleavage followed by the elimination of one mannose
residue from C-3 of core b-mannose. Inter estingly, the other
fragment ions in the spectrum show that a second, less
abundant isoform of this glycan exists, which is depicted in
Scheme 3C. The key fragment ions at m/z 1749.76 and
m/z 850.39 indicate that one antenna (consisting of
Fuc-HexNAc) c an be linked to t he 3-arm of core m annose.
The latter i on (m/z 850.39) is an ÔinternalÕ ion derived from a
C

4
cleavage followed by th e elimination of F ucGlcNAcMan
from the C-3 position of core mannose.
Together, these high-energy CID data indicate that the
major N-glycan structures on APN1 possess unusual
difucosylated core structures decorated with novel fucosyl-
ated antennae composed of Fuc(1-3)HexNAc. This Hex-
NAc residue is likely to be GlcNAc as GalNAc has never
been observed attached directly to the trimannosyl core of
any N-glycan structure. Other larger structures gave signals
that were too s mall to b e subjected to high-energy CID
(Hex
3
HexNAc
5)6
Fuc
5)6
). However, their compositions
suggest tri- and tetra-antennary structures substituted with
Fuc-GlcNAc antennae.
Characterization of fucosylated antennae
by sequential exoglycosidase digestions
The anomeric configurations of the fucose residues were
investigated by beef kidney a-fucosidase d igestion of the
Fig. 9. High-energy CID spectra of the [M + Na]
+
ions of the permethylated glycans (A) Fuc
3
Hex
3

HexNAc
3
and (B) Fuc
4
Hex
3
HexNAc
4
from
APN1. Sequence and linkage informative fragment ions f or (A) a nd (B) are assigned in Scheme 3. Signals labeled with an asterisk ar e derived from
internal cl eavages due to elimination of substituents from around the pyranose ring [24].
Ó FEBS 2004 Novel insect N-glycans from Manduca sexta (Eur. J. Biochem. 271) 4253
Scheme 3. Glycosidic ands cross-ring cleavages found in the high-energy CID spectra of (A) F uc
3
Hex
3
HexNAc
3
and (B and C) Fuc
4
Hex
3
HexNAc
4
from APN1. High-energy f ragme nt ions are labeled using th e nomenclature of Domon & Costello [27].
4254 E. Stephens et al.(Eur. J. Biochem. 271) Ó FEBS 2004
glycopeptides spanning Asn752 (Asn752–Arg770) that were
partially resistant to PNGase A digestion (Table 4;
Fig. 6A). The products were purified and analyzed by
nanoLC-MALDI-TOF MS. Comparison of the MALDI-

TOF mass spectrum (Fig. 10) with that shown in Fig. 6A
revealed that molecular ion signals for tri- and tetra-
fucosylated glycopeptides were lost. However, t he signals
for difucosylated structures re mained the major species.
Only minor quantities of mono- and nonfucosylated
glycopeptides were observed. These data indicate that the
difucosylated cores are difficult to digest for steric reasons
[31] and the fucose residues attached to the antennae have
been removed . This proposal was verified by subsequent
digestion of the antennae by b-N-acetylhexosaminidase
where all mole cular ion signals, apart from Hex
3
Hex-
NAc
2
Fuc
0-2
-(Asn752–Arg770), were diminished (data not
shown). Taken together, these exoglycosidase data establish
that the fucose residues on the antennae are a-linked to
GlcNAc.
Discussion
In the present study, the site-specific characterization of the
N-glycans on aminopeptidase 1 from Manduca sexta was
achieved using a MALDI-TOF/TOF tandem mass spec-
trometer. These MALDI-TOF/TOF data showed that the
glycoprotein preparation, purified by Cry 1Ac affinity
chromatography, c ontained three aminopeptidase p roteins:
APN1 (111 kDa), APN3 (108 kDa) and A PN4 ( 114 kDa),
with the APN1 protein being the most abundant. Inspec-

tion of the amino acid sequence of each protein revealed
four potential N-glycosylation sites in APN1, seven in
APN3 and five in APN4. Thus, the peptide digest mixture
could c ontain up t o 16 sets of g lycopeptides. In light of this
discovery, careful identification of any glycopeptides
detected was essential for the unambiguous assignment of
any glycosylation site. Consequently, a combination of
four criteria were employed in identifying each glycopep-
tide: (a) the presence of peaks in the MALDI mass
spectrum of an HPLC fraction with mass differences
corresponding to the masses of m onosaccharides. (b)
MALDI-TOF/TOF tandem MS/MS of the m ajority of
putative g lycopeptides observed. The resulting CID spectra
gave fragment ions that were derived from the pep tide
backbone, which mostly consisted of a series of C-terminal
y ions that led up to the glycosylated Asn residue. These
MS/MS data unambiguously identified each glycopeptide
constituent and proved essential for the assignment of
two sets of coeluting glycopep tides, which were com-
posed of two different peptide moieties derived from
APN1 [Asn623; SANRTVIHELSR(621–632)] and APN3
[Asn743; AFRNNNTLVPVNAR(739–752)]. Surprisingly,
these peptides differed by a mass corresponding to a
HexNAc residue (203 Da) and the amino acid sequence
information provided by the MALDI-TOF/TOF experi-
ments facilitated conclusive differentiation of each glyco-
peptide. (c) These MS/MS data were corroborated by
analysis of the deglycosylated peptide after treatment with
PNGase A. (d) Conversion of Asn to Asp by the
PNGase A enzyme was confirmed by MS/MS of the

released peptide.
The glycosylation status of whole APN has been
characterized previously in our laborato ries by monosac-
charide composition analysis and lectin binding studies
[10]. Purified APN was determined to contain GalNAc/
GlcNAc/Man/Fuc in the molar ratio of  6:10:7:3.It
was concluded that GalNAc was most likely to be derived
from the putative O-linked structures, which are thought
to form a C-terminal O-g lycosylated ÔstalkÕ structure in the
APN molecule. These O-linked structures were not
detected in the digestion mixture analyzed in the p resent
study, due to the lack of lysine and arginine residues
between this O-glycosylated region and the C-terminal
GPI anchor. This monosaccharide compositional data,
along with lectin binding studies and published i nforma-
tion on glycan biosynthesis in insects, led to the conclu-
sion that paucimannosidic structures with a1-3 core fucose
are likely to be the major structures on APN. However,
the amount of GlcNAc in APN was considered unusually
high for paucimannosidic N-glycans, and was attributed
to the O-linked glycans. Inspection of t he glycopeptide
compositions revealed in the present study, show that
some of this extra GlcNAc can be derived from the
N-glycans from APN1, as some HexNAc (GlcNAc)
residues are not fucosylated (e.g. Hex
3
HexNAc
3
Fuc
2

,
Hex
3
HexNAc
4
Fuc
3
,Hex
3
HexNAc
5
Fuc
4
,etc.).Inaddi-
tion, the glycopeptide compositions from APN3 indicated
that this glycoprotein may contain less fucose than APN1
[e.g. Hex
3
HexNAc
4
Fuc
2
was a major component on
APN3 (Table 3), but was not detected on APN1].
Fig. 10 . MALDI-TOF spectrum showing the
molecular ion signals for the glycopeptides
NGSFIPANMRPWVYCTGLR(752–770)
after treatment with a-
L
-fucosidase. The signals

are 16 m ass units higher than exp ected for t he
native m/z [M + H]
+
due to complete
methionine oxidation.
Ó FEBS 2004 Novel insect N-glycans from Manduca sexta (Eur. J. Biochem. 271) 4255
Therefore, it is likely that a proportion of the extra
GlcNAc observed is derived from the other APN isoforms
present in the protein sample.
The glycopeptide data presented in the c urrent study
gave site-specific information, which showed that the
nonfucosylated paucimannosidic structure (Man
3
Glc-
NAc
2
) is the major component at Asn609 in APN1. This
glycan is the most common N-glycan found in insects or
cultured insect cells [3,19,20,32]. The other three sites on
APN1 (Asn295, Asn623 and Asn752) were shown to
contain very unusual, highly fucosylated N-glycans, with
the most abundant glycoform consisting of Hex
3
Hex-
NAc
3
Fuc
3
. These highly fucosylated glycans were charac-
terized by high-energy CID of the released permethylated

glycans using a MALDI-TOF/TOF tandem mass spec-
trometer. Careful comparison of the fragmentation
observed with that detected from the collision-induced
dissociation of known fucosylated N-glycans from honey-
bee PLA
2
and other permethylated [24] and native
oligosaccharides [25,26], revealed novel structures not
previously observed on any glycoprotein.
Important cross-ring and internal fragment ions in the
MALDI-CID spectra showed that fucose is attached to the
C-3 position of the Asn-linked GlcNAc and at the C-3
position of the distal GlcNAc of the chitobiose unit. These
unusual core structures have not previously been observed
in insect glycoproteins. However, glycoproteins from a
nematode (Haemonchus contortus) have been shown to
contain th is unusual core fucosylation w here the N-glycans
can carry up to three fucose residues in the chitobiose
moiety [31]. The fragment ion spectra also revealed the
existence o f novel fucosylated antennae, consisting of Fuc(1-
3)GlcNAc, which are attached to the difucosylated core
structures. Key fragment ions showed that these antennae
are predominantly linked to the 6-arm of core mannose,
although in biantennary glycans a small proportion of the
Fuc(1-3)GlcNAc moiety can be linked to the 3-arm of the
core mannose residue. Furthermore, molecular ion signals
observed f or the r eleased permethylated glycans s uggest that
putative tri- and even tetra-antennary glycans can exist.
However, the signals for t hese multiantennary components
were t oo s mall to gain fragment ion spectra and a lternative

assignments f or these minor structures cannot be completely
ruled out. Subsequent exoglycosidase digestions of intact
glycopeptides revealed the anomeric configurations of the
residues in the an tennae as Fuca(1-3)GlcNAc. Together,
these data show that the majority of the glycans on APN1
(75%) consist of novel, highly fucosylated structures that
aredepictedinFig.11.
The presence of the unusual fucosylated N-glycans in
M. sexta is of considerable interest. The glycosylation
analyses of endogenous insect glycoprotein s are generally
lacking, in contrast to those of recombinant mammalian
glycoproteins expressed in insect cells. Therefore, the
discovery of new structures widens our understanding of
the glycosylation of these lower animals. In addition, the
relatively large quantities of the fucosylated mono- and
bi-antennary N-glycans are unusual as compared to other
insect glycoproteins where paucimannosidic structures are
normally the major components [ 3,6,32]. Thus it appears
that M. sexta larvae (lep idopteran species) have a
significant ability to a-fu cosylate the GlcNAc antennae
and the distal GlcNAc of th e chitobiose un it in
N-glycans. Furthermore, the major N-glycan components
on APN1 carry antennae linked to the 6-arm of core
mannose. This is in accord with previous studies that
showed high levels of a membrane-bound b-N-acetylglu-
cosaminidase in lepidopteran insect cell lines, which
exclusively removes GlcNAc from the 3-arm of core
mannose [33].
It is interesting to speculate on a possible f unctional
role for these highly fucosylated N-glycans on APN. They

are unlikely to be involved in binding the Cry 1Ac toxin,
as previous studies have shown that GalNAc mediates
this binding event [9]. GalNAc has been found in the
terminal structures of lacdiNAc [GalNAcb1-4(Fuca1-
3)GlcNAc] containing N-glycans from honeybee PLA
2
,
but this terminal substitution was not observed on the
major N-glycans on APN1. Therefore, the C-terminal
O-glycosylated region of the APN molecule is the most
likely site for Cry 1Ac toxin binding. Studies are currently
under way to characterize this C-terminal region of APN.
However, fucosylated structures are involved in a number
of important intercellular recognition events in mamma-
lian cells, which include fertilization and development,
apoptosis and selectin cell adhesion [34]. In contrast, there
are only limited data on the role of fucose in nonmam-
malian organisms. However, the Fuca(1-3)GlcNAc moiety
has been shown to be a highly antigenic epitope of both
plant [35] and insect [36] glycoproteins and also accounts
for the immunogenic cross-reactivity between plant and
insect glycoproteins [36]. This substitution is also the
major allergenic determinant in honeybee PLA
2
[4,5,7].
Individuals who are allergic to honeybee ven om produce
appreciable levels of antibodies to the 3 -linked fucose of
the N-glycans in this protein [7,8]. The high abundance of
the nov el hig hly f ucosylated N-glycans at three N-glyco-
sylation sites on the APN protein indicates that they are a

prominent feature on the midgut membrane surface and
could plausibly play a role in mediating protein–protein
interactions.
Acknowledgements
The a uthors thank Dr Joe C arroll for helpful discussions, P rof. C. V.
Robinson for her guidance and support, and the BB SRC for funding
E. S. and J. S.
Fig. 11. Proposed structures of the highly fucosylated N-linked oligo-
saccharides at Asn295, Asn623 and Asn752 from APN1. Major con-
stituents are boxed. Monosaccharides, linkage and anomericity are
represented by shapes and line s as decribed in a p revious study [37].
4256 E. Stephens et al.(Eur. J. Biochem. 271) Ó FEBS 2004
References
1. Varki, A., Cummings, R., Esko, J., Freeze, H., Hart, G. & Marth,
J. (1999) Essentials of Glycobiology. Cold Spring Harbor Press,
Cold Spring Harbor, NY.
2. Ma
¨
rz, L., Altmann, F., Staudacher, E. & Kubelka, V. (1995)
Glycoproteins. Els evier, Amsterdam.
3. Altmann, F., Staudacher, E., Wilson, I.B.H. & Ma
¨
rz, L. (1999)
Insect cells as hosts for the expression of re co mbinant glycopro-
teins. Glycoconj. J . 16, 109–123.
4.Kubelka,V.,Altmann,F.&Ma
¨
rz, L. (1995) The asparagine-
linked carbohydrate of honeybee venom hyaluronidase. Glyco-
conj. J. 12 , 77–83.

5. Kubelka, V., Altma nn, F., Staudacher, E., Tretter, V., Ma
¨
rz, L.,
Hard, K., Kamerling, J.P. & Vliegenthart, J.F.G. ( 1993) Primary
structures of the N-linked carbohydrate chains from honeybee
venom phospholipase-A2. Eur. J. Bioc hem. 213, 1193–1204.
6. Jarvis, D.L., Kawar, Z.S. & Hollister, J.R. (1998) Engineering
N-glycosylation pathways in the baculovirus-insect cell system.
Curr. Op. Biotec hnol. 9, 528–533.
7. Weber, A., Schroder, H., Thalberg, K. & Ma
¨
rz, L. ( 1987) S pecific
interaction of I gE an tibodies with a carbohydrate epitope of
honey-bee venom phospholipase-A2. Allergy 42, 464–470.
8. Tretter,V.,Altmann,F.,Kubelka,V.,Ma
¨
rz,L.&Becker,W.M.
(1993) Fucose al pha-1,3-linked to the core re gion of glycoprotein
N-glycans creates an important epitope for IgE from honeybee
venom allergic i ndivid uals. Int. Arch. Allergy Immunol. 102,
259–266.
9. Knight, P .J.K., Crickmore, N. & Ellar, D .J. (1994) The receptor
for Bacillus thuringiensis CrylA(c) delta-endotoxin in the brush-
border membrane of the lepidopteran Manduca sexta is amino-
peptidase-N. Mol. Microbiol. 11, 429–436.
10. Knight, P.J.K., Carroll, J. & Ellar, D.J. (2004) Analysis of glycan
structures on the 120 kDa amin opeptidase N o f Manduca sexta
and their interactions with Bacillus thuringiensis Cry1Ac toxin.
Insect Biochem. Mol. Biol. 34, 101–112.
11. Bulla,L.A.,Kramer,K.J.,Cox,D.J.,Jones,B.L.,Davidson,L.I.

& Lookhart, G.L. (1981) Purification and characterization of the
entomocidal p rotoxin o f Bacillus thuringiensis. J. Biol. Chem. 256,
3000–3004.
12. Gill, M. & Ellar, D.J. (2002) Transgenic Drosophila reveals a
functional in vivo recepto r for t he Bacillus thuringiensis toxin
Cry1Ac1. Insect Mol. Biol. 11, 619–625.
13. Luo, K., Lu, Y. & Adang, M.J. ( 1996) A 106 kDa fo rm of ami-
nopeptidase is a receptor for Bacillus thuringiensis Cry1c delta-
endotoxin in the brush border membrane of Manduca sexta. Insect
Biochem. Mol. Biol. 26, 783–791.
14. Denolf, P., Hendrickx, K., Van Damme, J., Jansens, S ., Peferoen,
M., Degheele, D. & Van Rie, J. (1997) Cloning and character-
ization of Manduca sexta and Plutella xylostella midgut amino-
peptidase N enzymes related to Ba cillus thuringiensis toxin-binding
proteins. Eur. J. Bioc hem. 248, 748–761.
15. Altmann, F., Schweiszer, S. & W eber, C. (1995) K inetic
comparison of peptide N-glycosidase F and N-glyc osidase A re-
veals several differences in substrate specificity. Glycoconj. J. 12,
84–93.
16. Stimson, E., Hope, J., Chong, A. & Burlingame, A.L. (1999) Site-
specific c haract erization of the N-linked g lycans of murine prion
protein by high-perfo rmance liq uid chromatography electrospray
mass spectrometry and exoglycosidase digestions. Biochemist ry 38,
4885–4895.
17. Dell, A., Khoo, K-H., Panico, M., McDowell, R.A., Etienne,
A., Reason, A.J. & Morris, H.R. (1992) FAB-MS a nd ES-MS
of glycoproteins. In Glyc obiology. A P ractical Approach (Fuku da,
M. & Kobata, A., eds), pp. 187–222. Oxford University Press,
Oxford.
18. Rudd, P.M., Downing, K.A., Cadene, M., Harvey, D.J., Worm-

ald, M.R., Weir, I., Dwek, R.A., Rifkin, D.B. & Gleizes, P.E.
(2000) Hybrid and complex glycans are linked to the conserved
N-glycosylation site of t he thir d e ight-cysteine domain o f L TBP-1
in insect cell s. Biochemistry 39 , 1596–1603.
19. Lopez, M., Coddeville, B ., Langridge, J. , P lancke, Y., Chaabihi,
H., Chirat, F., Harduin-Lepers, A., Cerutti, M., Verbert, A. &
Delannoy, P. (1997) Microheterogen eity of the oligosaccharides
carried by the recombinant bovine lactoferrin expressed in
Mamestra brassicae cells. Glycobiology 7, 635–651.
20.Kubelka,V.,Altmann,F.,Kornfeld,G.&Ma
¨
rz, L. (1994)
Structures of the N-linked oligosaccharides of the membrane-
glycoproteins from 3 lepidopteran cell-lines (SF-21, IZD-
MB-0503, BM-N). Ar ch. Biochem. B iophys. 308, 1 48–157.
21. Ku
¨
ster, B., Naven, T.J.P. & Harvey, D.J. (1996) Effect of redu-
cing-terminal substituents on the high-energy collision-induced
dissociation matrix-assisted laser desorption/ionization mass
spectra of oligosaccharides. Rapid C ommun. Mass Spectrom. 10 ,
1645–1651.
22. Plummer, T.H. Jr & Tarentino, A.L. (1981) Facile cleavage of
complex oligosaccharides from glycopeptides by almond em ulsin
peptide-N-glycos idase. J. Biol. Chem. 256, 10243–10246.
23. Plummer, T.H. Jr, Phelan, A.W. & Tarentino, A.L. (1987)
Detection and quantification of peptide-N-4-(N-acetyl-beta-glu-
cosaminyl) asparagine amidases. Eur. J. B ioc hem. 163, 167–
173.
24. Stephens, E., Maslen, S.L., Green, L.G. & Williams, D.H. (2004)

Fragmentation characteristics of neutral N-linked glycans using a
MALDI-TOF/TOF tandem mass spectrometer. Anal. Chem. 76,
2343–2354.
25. Spina, E., Sturiale, L., Romeo, D ., Impallomeni, G., Garozzo, D.,
Waidelich, D. & Glueckmann, M. (2004) New fragmentation
mechanisms in matrix-assisted laser desorption/ionization time-
of-flight/time-of-flight tandem mass spectrometry. Rapid Com-
mun. Mass Spec trom. 18, 392–398.
26. Mechref, Y., Novotny, M.V. & Krishan, C. (2003) Structural
characterization of oligosaccharides using MALDI-TOF/TOF
tandem mass spectrometry. Anal. Chem. 75, 4895–4903.
27. Domon, B. & C ostello, C.E. (1988) A s ystematic nomenclature f or
carbohydrate fragmentations in FAB-MS/MS spectra of glyco-
conjugates. Glycoconjugate J . 5, 397–409.
28. Harvey, D.J., Bateman, R.H. & Green, M.R. (1997) High-energy
collision-induced f ragmentation of complex oligosaccharides
ionized by matrix-assisted laser desorption/ionization mass
spectrometry. J. Mass Spectrom. 32, 167–187.
29. Dell, A. & Thomas-Oates, J. ( 1989) Analysis of Carbohydrates by
GLC and MS . CRC Press, Boc a Raton, FL.
30. Staudacher, E ., A ltmann, F., Ma
¨
rz, L ., Hard, K., Kamerling, J.P.
& Vliegenthart, J.F. (1992) Alpha-1-6 (alpha-1-3)-difucosylation
of the asparagine-b ound N-acetylglucosamine in h oneybee venom
phospholipase-A2. Glycoconj. J. 9, 82–85.
31. Haslam, S .M., Coles, G.C., Munn, E.A., Smith, T.S., Smith, H.F.,
Morris, H.R. & Dell, A. (1996) Haemonchus contortus glycopro-
teins contain N-linked o ligosaccharides with novel highly fucosy-
lated core structures. J. Biol. Chem. 271, 30561–30570.

32. Kulakosky, P.C., Hughes, P.R. & Wood, H.A. (1998) N-linked
glycosylation of a baculovirus-expressed recombinant glycopro-
tein in insect larvae and tissue culture cells. Glycobiology 8, 741–
745.
33. Altmann, F., Schwihla, H., Staudacher, E., Glo
¨
ssl,J.&Ma
¨
rz, L.
(1995) Insect cells contain an unusual, membrane bound beta-
N-acetylglucosaminidase probably involved in the processing of
protein N-glycans. J. Biol. Chem. 270, 17344–17349.
34. Staudacher, E., Altmann, F., Wilson, I.B.H. & Ma
¨
rz, L. (1999)
Fucose in N -glycans: From plant to man. Biochem. Biophys. Acta
1473, 216–236.
Ó FEBS 2004 Novel insect N-glycans from Manduca sexta (Eur. J. Biochem. 271) 4257
35. Ramirez-Soto, D. & Poretz, R.D. ( 1991) The (1-3)-linked alpha-
1-fucosyl group of the N -glycans of the W istaria flor ibunda lectins
is recognized by a r abbit antiserum. Carbohydr. R es. 213, 27–36.
36. Prenner, C., Mach, L., Glo
¨
ssl, J. & Ma
¨
rz, L. (1992) The anti-
genicity of the carbohydrate moiety of an insect glycoprotein,
honeybee (Apis mellifera) venom phospholipase-A(2) – t he role of
alpha-1,3-fucosylation of the asparagine-bound N-acetylglucosa-
mine. Biochem. J . 284, 337–380.

37.Butler,M.,Quelhas,D.,Critchley,A.J.,Carchon,H.,
Hebestreit, H.F., Hibbert, R.G., Vilarinho, L., Teles, E.,
Matthijs, G., Schollen, E., Argibay, P., H arvey, D.J., Dwek,
R.A., Jaeken, J. & Rudd, P.M. (2003) Detailed glycan analysis
of serum glycoproteins of patients with congenital disorders of
glycosylation i ndicates the specific defective g lycan processing s tep
and provides an insight into pathogenesis. Glycobiology 13, 601–
622.
4258 E. Stephens et al.(Eur. J. Biochem. 271) Ó FEBS 2004