Sequential Analysis of N- and O-Linked Glycosylation of 2D-PAGE
Separated Glycoproteins
Nicole L. Wilson,* Benjamin L. Schulz, Niclas G. Karlsson, and Nicolle H. Packer
Proteome Systems Ltd. 1/35-41 Waterloo Road, North Ryde, Sydney, NSW 1670, Australia
Received June 13, 2002
A robust method has been developed that allows analysis of both N- and O-linked oligosaccharides
released from glycoproteins separated using 2D-PAGE and then electroblotted to PVDF membrane.
This analysis provides efficient oligosaccharide profiling applicable to glycoproteomic analysis. The
method involves the enzymatic release of N-linked oligosaccharides using PNGase F followed by the
chemical release of O-linked oligosaccharides using reductive β-elimination and analysis using LCESI-MS. Oligosaccharides from the major plasma glycoproteins with a pI between 4 and 7 were
characterized from the glycoforms of haptoglobin, R2-HS-glycoprotein, serotransferrin, R1-antitrypsin,
and R1-antichymotrypsin. It was shown that the separation of protein glycoforms evident in 2D-PAGE
is partially due to the combined sialylation of the O-linked and N-linked oligosaccharides. Bi-, tri- and
tetra-antennary N-linked structures, which had differing levels of sialylation and fucosylation, were
found to be present on the glycoproteins analyzed, together with O-linked oligosaccharides such as
mono-, and disialylated T-antigen and a disialylated core type 2 hexasaccharide. In addition, N-linked
site-specific information was obtained by MALDI-MS analysis using tryptic digestion after PNGase F
release of the oligosaccharides.
Keywords: sialylation • glycoproteins • β-elimination • PNGase F • plasma
Introduction
The analysis of oligosaccharides on proteins, bound either
to an asparagine residue (N-linked) or the serine and threonine
residues (O-linked), has long been of interest to scientists,
especially in the study of the post-translational role of oligosaccharides in the biological function of proteins. Perhaps most
interesting is the alteration of these oligosaccharides in relation
to diseases and disease progression. Into this category, one can
include alterations due to inherited diseases (reviewed in ref
1) and progression of cancer (reviewed in refs 2 and 3). Hence,
a high-throughput sensitive method for the analysis of both
N-linked and O-linked oligosaccharides would prove very useful
to gain information on global glycosylation and individual
variation. This information is vital in order to be able to address
the question of disease related glycosylation.
The word “proteomics” has become established in biotechnology,4 and has become the accepted term for identifying as
many proteins from a particular organism, tissue, or cell as
possible. However, there is a second wave of technologies being
developed to extend the amount of information produced in
proteomics research by going beyond protein identification.
Hence, the name “glycoproteomics” has been established to
include glycosylation information in a proteomic context.
Two-dimensional polyacrylamide gel electrophoresis (2DPAGE) separates proteins according to charge and size.5 The
* To whom correspondence should be addressed. Locked bag 2073, North
Ryde, Sydney NSW 1670, Australia. Phone +61 2 98891830. Fax +61 2
98891805. E-mail: [email protected].
10.1021/pr025538d CCC: $22.00
 2002 American Chemical Society
technique concentrates and purifies protein samples for analysis. It is common to see glycoforms of glycoproteins separate
horizontally in a 2D gel. This phenomenon is caused by a
difference in the pI of the glycoproteins6 and is commonly
believed to be caused by a variability of sialylation of different
protein glycoforms. The expression of these different glycoforms may be useful as a diagnostic marker.7
Mass spectrometry has now reached a level in which the
study of glycoproteins and the attached oligosaccharide structures can be undertaken on a protein separated by 1D- and
2D-PAGE.8,9 There are two main methods of ionization that
have revolutionized protein mass spectrometry, matrix-assisted
laser desorption ionization (MALDI) and electrospray ionization
(ESI). Their adaptation for oligosaccharide analysis has been
successful, but analyses of both neutral and acidic oligosaccharides using the same general method have proven difficult.
Using MALDI-MS for oligosaccharide analysis, acidic and
neutral oligosaccharides usually need to be prepared in different matrixes and detected in negative or positive mode to
achieve optimal detection.10-13 Alternatively, an elegant method
has been described in which the sialic acid residues are
converted into methyl esters, allowing simultaneous detection
in the positive ion mode of both neutral and sialylated
oligosaccharides.14 Derivatization of the reducing end using
reductive amination or oxime formation has also been used in
order to introduce moieties to improve the ionization of
oligosaccharides, both for MALDI and ESI mass spectrometry
(reviewed in ref 15).
Journal of Proteome Research 2002, 1, 521-529
521
Published on Web 09/07/2002
research articles
Wilson et al.
The use of graphitized carbon as an LC chromatographic
medium combined with MS has shown that chromatographic
separation of oligosaccharides can be achieved at high pH.16
This approach reduces the amount of sample and preparation
time needed because both acidic and neutral oligosaccharides
could be detected sensitively without further derivatization. The
same method of analysis can be used for both PNGase F
released N-linked and reductively β-eliminated O-linked oligosaccharides, with minimal handling of the sample. In addition,
the development of nanoflow liquid chromatographic separation and nanospray ESI has allowed the sensitivity of detection
to be increased to approximate that of MALDI-MS.17 These
combined technologies are now available to enable the highthroughput analysis of the glycosylation of small amounts of
separated glycoproteins.
The aim of this study was to develop a robust method for
oligosaccharide analysis of 2D-PAGE separated glycoproteins,
and in particular, to create a map of the glycosylation of some
of the human plasma proteins in the pI region between 4 and
7. This map provides a baseline of human plasma glycoprotein
glycosylation for future comparison of the glycosylation of
proteins in disease.
Materials and Methods
Materials. N-Glycosidase F (PNGase F, recombinant cloned
from Flavobacterium meningosepticum and expressed in E. coli)
lyophilized in 10 mM sodium phosphate buffer, pH 7.2, was
purchased from Sigma-Aldrich Corp. (St. Louis, MO) or in 100
mM sodium phosphate buffer, 25 mM EDTA, pH 7.2, from
Roche Molecular Biochemicals (Mannheim, Germany). The
human plasma sample was obtained from a healthy individual.
Sequence grade modified trypsin was obtained from Promega
(Madison, WI). All other chemicals were purchased from SigmaAldrich.
Method
Sample Preparation and 2D-PAGE. Human plasma samples
were prepared for 2D-PAGE by depletion of fibrinogen, using
a venom cross-linking method,18 immunoglobulin type G (IgG)
with immuno-affinity chromatography using immobilized protein G sepharose beads (Amersham Pharmacia, Uppsala,
Sweden), and human serum albumin by ethanol fractionation.19,20 2D-PAGE of depleted plasma was focused on an 11
cm, pI 4-7 IPG (Amersham Pharmacia), after precipitation of
the protein with ethanol, solubilization, and reduction and
alkylation.21 Human plasma proteins separated by 2D-PAGE
were electroblotted22 from 135 × 80 mm GelChip gels (Proteome Systems Inc, Boston, MA) to Immobilon-PSQ PVDF
membrane (Millipore, Bedford, MA). Blotted proteins were
stained either with periodic acid/Schiff base (PAS) for glycoproteins23 or Direct Blue (DB71) as a general protein stain.
Potential glycoproteins were identified by PAS, but as this stain
oxidizes the oligosaccharides, the corresponding Direct Blue
stained spots were identified and used for glycoprotein analysis.
PNGase F Enzymatic Cleavage and Purification of N-Linked
Oligosaccharides. Protein spots were cut from the PVDF
membrane and placed in separate wells of a 96-well microtiter
plate. The spots were then covered with 100 µL of 1% (w/v)
poly(vinylpyrrolidone) 40 000 in 50% methanol, agitated for 20
min, and transferred to new wells after thoroughly washing with
water. N-Linked oligosaccharides were released from the
protein by incubation with 5 µL of PNGase F (0.5 units/µL)
522
Journal of Proteome Research • Vol. 1, No. 6, 2002
Figure 1. Flowchart for the strategy employed for analysis of
2D-PAGE separated glycoproteins.
overnight at 37 °C. During the incubation, the sample wells
were surrounded by wells containing water to prevent evaporation, and the plate was sealed with amplification tape. To
recover the oligosaccharides, the samples were sonicated (in
the 96-well plate) for approximately 5 min, and the released
N-linked oligosaccharides were transferred from the wells onto
graphitized carbon microcolumns. The carbon microcolumns
were made by suspending the graphitized carbon from an
Alltech (Flemington, NJ) Carbograph Extract-Clean Tube in
methanol, and transferring 10 µl of the packing on to the top
of a reverse phase µC8 ZipTip (Millipore). The microcolumns
were washed with 3 × 10 µL of 90% (v/v) MeCN/0.5% formic
acid before being equilibrated with 0.5% formic acid. The
N-linked oligosaccharides were trapped on the column by
repeated filling and dispensing of the sample using a micropipet. The columns were washed using 3 × 10 µL of 0.5% formic
acid, and the N-linked oligosaccharides were eluted into
another 96-well plate using 2 × 10 µL of 25% (v/v) MeCN/0.5%
formic acid and dried under vacuum. Before analysis with LCESI-MS, the oligosaccharides were redissolved in water.
β-Elimination Release and Purification of O-Linked Oligosaccharides. After the removal of the N-linked oligosaccharides, each protein spot remaining on the membrane was rewet with approximately 2 µL of methanol. A solution of 50 mM
KOH and 0.5 M NaBH4 (20 µL) was applied, and the spots were
incubated for 16 h at 50 °C. The samples were neutralized by
adding 1 µL of glacial acetic acid, and the released and reduced
O-linked oligosaccharides were transferred onto homemade
cation exchange columns comprising 20 µl of AG50W-X8
cation-exchange resin (H+ - form) (BioRad, Hercules, CA)
N- and O-Linked Glycosylation Analysis
research articles
Figure 2. 2D-PAGE separated human plasma proteins electroblotted to PVDF membrane. A. Direct Blue 71 stained proteins. B. PAS
(periodic acid/Schiff) stained glycoproteins.
packed onto the top of µC8 ZipTips.24 Oligosaccharides were
eluted with 2 × 60 µL of water and dried under vacuum. The
remaining borate was removed by the addition of 50 µL of 1%
acetic acid in methanol and drying under vacuum five times.
The oligosaccharides were re-suspended in 10 µL of water for
LC-ESI-MS analysis.
LC-ESI-MS of Oligosaccharides. LC-ESI-MS was used for
the analysis of both the N- and O-linked oligosaccharides. The
samples were applied to a Hypercarb porous graphitized
carbon column (5 µm Hypercarb, 0.32 × 150 mm, Thermo
Hypersil, Runcorn UK) using a Surveyor autosampler (Thermo
Finnigan, San Jose, CA). A linear water-acetonitrile gradient
(0-25% acetonitrile in 30 min, followed by a 3 min wash with
90% acetonitrile) containing 10 mM NH4HCO3 was used to
separate the oligosaccharides. The flow rate was set at 200 µl/
min, and a split ratio of 1/30 gave a final flow rate of
approximately 6 µl/min through the column. ESI-MS was
performed in negative ion mode with three scan events: full
scan with mass range m/z 320-2000, dependent zoom scan,
and dependent MS/MS scan after collision induced fragmentation. Collision conditions used were normalized collision energy
of 40% and an activation time of 30 ms. Dynamic exclusion of
ions for zoom scan for 30 s was introduced after three selections
within 30 s.
Tryptic Digestion and MALDI-MS of Peptides. Tryptic
digestion was performed on protein spots on the membrane
before or after PNGase F enzymatic release of N-linked oligosaccharides as described above. The protein spots were incubated in 10 µl of 0.02 µg/µL trypsin in 25 mM NH4HCO3 for 3
h at 37 °C. Then, 15 µL of 50% (v/v) acetonitrile, 0.1% (v/v)
Journal of Proteome Research • Vol. 1, No. 6, 2002 523
research articles
Wilson et al.
Figure 3. LC-ESI-MS of N-linked oligosaccharides released from R2-HS-glycoprotein. The spectra were collected between 21 and 34
min of a 37 min gradient eluted from a porous graphitized carbon column. A. Base peak chromatogram and spectrum from R2-HSglycoprotein isoform spot with pI ≈ 4.21. B. Base peak chromatogram and spectrum from R2-HS-glycoprotein isoform spot with pI ≈
4.50. 9 ) GlcNAc, O ) Man, ] ) Gal, 4 ) Fuc, f ) NeuAc, and 0 ) GalNAc. Structures are based on previously reported human
plasma oligosaccharides (GlycoSuite DB).
trifluroacetic acid (TFA) was added to each well, and the
samples were sonicated for 10 min and diluted by the addition
of 10 µL of Milli-Q water to each well. The sample was placed
at 37 °C for an additional 20 min to reduce the concentration
of acetonitrile. µC8 ZipTip columns were pre-washed with 90%
acetonitrile/0.1% TFA (v/v) and then equilibrated with 0.1%
(v/v) TFA, and the samples were loaded onto the column and
desalted using a 0.1% TFA solution. Peptides were eluted and
524
Journal of Proteome Research • Vol. 1, No. 6, 2002
directly spotted onto the MALDI plate directly from the
columns using 1 µL of a matrix (10 mg/mL R-cyano-4hydroxycinnamic acid in 70% (v/v) acetonitrile, 0.0 5% (v/v)
TFA). MALDI-MS was performed on an Axima CFR instrument
(Shimadzu Kratos, Manchester, UK), equipped with an N2 laser
(337 nm, 10 Hz repetition rate). All of the mass spectra were
acquired in the positive reflectron mode with delayed pulse
extraction. Internal mass calibration was performed using the
N- and O-Linked Glycosylation Analysis
research articles
Table 1. N-Linked Oligosaccharide Composition and Average Sialylation of Human Plasma Glycoprotein Glycoforms
* Approximate pI based on linear pI gradient measurement. ** Structures based on assumption of similarities with reported human plasma oligosaccharides
(GlycoSuite DB). *** Results calculated from LC-ESI-MS intensities.
Table 2. N-Linked Oligosaccharide Composition and Average Sialylation of Human Serotransferrin Glycoforms
* Approximate pI based on linear pI gradient measurement. ** Structures based on assumption of similarities with reported human plasma oligosaccharides
(GlycoSuite DB). *** Results calculated from LC-ESI-MS intensities.
trypsin autolysis peaks. IonIQ peptide mass fingerprinting
program (Proteome Systems, Sydney, Australia) was used to
identify the proteins using the SwissProt database, allowing an
error margin of 50 ppm and one missed cleavage. Identification
of N-glycosylation sites due to the mass conversion of a
glycosylated asparagine into aspartic acid could be determined
using PNGase F followed by trypsin digestion.
Results and Discussion
The general strategy that was used for this study is outlined
in Figure 1. The human plasma sample was depleted of human
serum albumin, IgG, and fibrinogen, and the remaining
proteins were separated by 2D-PAGE and electroblotted to
PVDF membrane. Identical blots were stained with Direct Blue
71 or periodic acid/Schiff-base (PAS) stain. PAS staining
oxidizes carbohydrates so that stained proteins cannot be
further analyzed for their glycosylation. Thus, corresponding
glycoproteins identified by PAS staining were excised from the
Direct Blue 71 stained membrane and treated with PNGase F.
The released N-linked oligosaccharides were aspirated from the
membrane, desalted using microcolumns of graphitized carbon, and analyzed using LC-ESI-MS. The remaining O-linked
oligosaccharides on the protein were released using reductive
alkaline β-elimination and analyzed by the same LC-ESI-MS.
Alternatively, proteins on the membrane were digested with
trypsin before and after PNGase F treatment, and the peptides
on the membrane were analyzed using MALDI-MS.
2D Gel Electrophoresis and Staining. The depletion of
highly abundant proteins from plasma allowed more of the
less-dominant proteins in plasma (which are predominantly
glycoproteins) to be loaded onto the gel. Figure 2A shows the
typical 2D-PAGE of a human plasma protein blot. The PAS
stained blot shown in Figure 2B was used to determine which
proteins were potentially glycosylated. With the sensitivity of
PAS staining (1-10 µg of a highly glycosylated protein can be
detected), the majority of proteins in plasma detected by Direct
Blue were found to be glycosylated. A number of the major
glycoproteins were chosen for analysis: R2-HS-glycoprotein (2
Journal of Proteome Research • Vol. 1, No. 6, 2002 525
research articles
× N-linked, 3 × O-linked sites), R1-antichymotrypsin (2 ×
N-linked sites), R1-antitrypsin (3 × N-linked sites), haptoglobin
(4 × N-linked sites), and serotransferrin (2 × N-linked sites).
These proteins were also identified in SwissProt as glycoproteins and represent a variety of molecular weights and pI values
in a 2D-PAGE separation of human plasma proteins. The
glycoproteins are seen as trains of spots of the same protein
separating at different pI and/or molecular mass. Glycosylation
differences have the potential to change both these parameters,
and the determination of the oligosaccharides on the protein
glycoforms will help explain the presence of these glycoforms,
as well as identifying potential markers of differentiation and
disease.
Enzymatic Release of N-Linked Oligosaccharides. PNGase
F cleaves the oligosaccharides from glycoproteins to give
aspartic acid (in the place of asparagine in the peptide chain)
and the released oligosaccharides as products. By mapping the
oligosaccharides present on glycoproteins, information on the
normal glycosylation of plasma glycoproteins can be established. The mass spectra of the released N-linked oligosaccharides show that the ratio of those structures present on human
plasma glycoproteins vary between different glycoforms. Figure
3 shows the mass spectra of the N-linked oligosaccharides of
two R2-HS-glycoprotein glycoforms: pI ) 4.21, the most acidic
protein isoform (A) and pI ) 4.50, the most basic isoform (B).
These spectra exemplify the change in percentage of the
N-linked oligosaccharide structures. The [M-2H]2- ion of m/z
1110.8 (disialylated biantennary N-linked oligosaccharide)
becomes the most dominant structure as the pI increases, while
the di- and triantennary sialylated components with [M-2H]2ions of m/z 1438.5 and m/z 1511.5, respectively, are decreasing
concomitantly.
Tables 1 and 2 show the composition of the major Nlinked oligosaccharide structures present on different protein glycoforms as identified using LC-ESI-MS. The structures assigned to these compositions are based on previously reported structures in human plasma (GlycoSuite DB,
www.glycosuite.com). Some general trends can be seen across
the majority of the glycoproteins studied. Each glycoform is
glycosylated by the same repertoire of oligosaccharides but in
varying amounts. The major oligosaccharide present on all the
proteins has a composition corresponding to a biantennary,
disialylated structure. The percentage of each individual oligosaccharide composition was calculated from the ion intensities
of individual components and the sum of all oligosaccharide
ion intensities. The oligosaccharides seen in negative ion mode
were all detected as doubly charged m/z ions. As the pI
increases, all of the protein glycoforms show an increase in
the amount of the biantennary, di- and monosialylated structures, a decreased amount of triantennary, trisialylated and core
fucosylated structures, and thus, an overall decrease in negative
charge with increasing pI. The average contribution to the
charge by sialic acid is calculated from the mass ion intensity
of each oligosaccharide and shows that as a general rule, the
average sialylation increases with a decrease in pI of the
protein. Other minor oligosaccharides, which are defined as
structures with LC-ESI-MS intensities <3% of the total, were
observed (Table 3) but would make little contribution to the
net charge on the glycoform.
Serotransferrin, however, behaves somewhat differently from
the other glycoproteins, with the average sialylation increasing
with the increase in pI of the protein (Table 2). The reason for
this can only be speculative at this stage and could be due to
526
Journal of Proteome Research • Vol. 1, No. 6, 2002
Wilson et al.
Table 3. Minor N-Linked Oligosaccharide Structures Found on
Human Glycoproteins
* Structures with LC-ESI-MS intensities <3% of the total. ** Structures
based on assumption of similarities with reported human plasma oligosaccharides (GlycoSuite DB).
Table 4. O-Linked Oligosaccharides Released from Human
Plasma R2-HS-glycoprotein
* Approximate pI based on liner pI gradient measurement. ** Structures
based on assumption of similarities with reported human plasma oligosaccharides (Glycosuite DB). *** Results calculated from LC-ESI-MS intensities.
other post-translational modifications or oligosaccharide site
occupancy. Serotransferrin also contains a high proportion (416%) of neutral oligosaccharides compared to the other plasma
proteins, and it cannot be ruled out that they may affect the
isoelectric point of the protein. However, no current model for
how neutral oligosaccharides could effect the pI of the protein
has been proposed.
PNGase F followed by Chemical Release of O-Linked
Oligosaccharides. β-Elimination in a reducing environment is
widely used to liberate and study glycoprotein O-linked oligosaccharide structures.25 The reduction of the reducing terminus of the eliminated oligosaccharide is necessary to prevent
the released oligosaccharides from undergoing subsequent
“peeling”. This release method following PNGase F digestion
allowed analysis of both the N- and O-linked oligosaccharides
to be performed from a single protein spot blotted to PVDF
membrane after 2D-PAGE separation.
The majority of the glycoproteins in this study are modified
with only N-linked glycosylation; however, R2-HS-glycoprotein
N- and O-Linked Glycosylation Analysis
research articles
Figure 4. LC-ESI-MS of O-linked oligosaccharides released from R2-HS-glycoprotein using reductive β-elimination after PNGase F
digestion. A. Spectrum obtained from isoform with pI ≈ 4.21. B. Spectrum obtained from isoform with pI ≈ 4.50.
also has three sites of O-linked glycosylation.26 The data
obtained by LC-ESI-MS of the major N- and O-linked
structures released from R2-HS-glycoprotein by serial treatment
with PNGase F and reductive β-elimination are shown in Tables
1 and 4 (Table 4 also shows the average sialylation of the
O-linked oligosaccharides calculated for the different glycoforms of R2-HS-glycoprotein). The data show that changes in
sialylation of both types of N- and O-linked structures are
contributing to the change in pI of the glycoforms. In Figure
4, the O-linked oligosaccharides released from the protein
glycoforms focusing at pI 4.21 and pI 4.50 of R2-HS-glycoprotein
are shown as singly charged ions. There is evidence of polysaccharide glucose contamination at this level of detection.
Glucose contamination is a common problem in carbohydrate
analysis. The O-linked oligosaccharides show a similar trend
to that of the N-linked oligosaccharides in that the monosialylated structure decreases while the total disialylated structures increase with the increasing acidity of the isoform (Table
4). In the most acidic isoform, the disialylated hexasaccharide
accounts for greater than 50% of the O-linked oligosaccharides
present on the protein. As the pI of the isoform becomes
increasingly basic, the abundance of this structure decreases
until it is not visible in the spectrum (Figure 4B), while the
proportion of a monosialylated trisaccharide structure increases
accordingly. This monosialylated trisaccharide composition has
previously been characterized as the NeuAcR2-3Galβ1-3GalNAc
in human R2-HS-glycoprotein. Structures corresponding to the
other O-linked compositions have not previously been described on human R2-HS-glycoprotein, although the NeuAcR23Galβ1-3(NeuAcR2-3Galβ1-4GlcNAc)GalNAc and NeuAcR23Galβ1-3(NeuAcR1-6)GalNAc have been found in the bovine
analogue and could account for the compositions (GlycosuiteDB).
PNGase F followed by Tryptic Digestion and Identification
of N-Glycosylation Sites. During the PNGase F release of the
N-linked oligosaccharides, the protein backbone stays intact
and bound to the PVDF membrane; hence, it is possible to
digest the protein after N-linked de-glycosylation. This often
improves the peptide coverage of the protein because glycopeptides are seldom detected by the standard procedures for
detecting peptides by MALDI-MS. The method also has the
potential for detecting the sites of N-glycosylation because the
use of PNGase F results in the glycosylated asparagine being
deamidated to form aspartic acid. This increases the mass of
the peptide after de-glycosylation by 0.9848 Da from that
predicted by the gene-derived amino acid sequence. Figure 5A
shows the tryptic peptide mass spectrum of an R1-antitrypsin
separated glycoform with a pI of 5.02. Eighteen peptides of R1antitrypsin were identified in the spectrum, resulting in 49%
coverage of the amino acid sequence of the protein. None of
the masses of predicted N-linked glycopeptides were present.
However, when this protein glycoform was de-glycosylated
using PNGase F prior to the tryptic digestion, the peptide mass
spectrum identified 23 peptides, giving a protein coverage of
61% (Figure 5B). Two of the three predicted de-glycosylated
peptides appeared in the spectrum at masses of 0.98 Da greater
than predicted, at 1756.9 and 3692.8 Da, corresponding to
potential N-linked sites at N107 and N271. The de-glycosylated
tryptic peptide (N70) that was not seen in the MALDI-MS
spectrum after PNGase F de-glycosylation has an expected
deglycosylated tryptic mass of 3198.64 Da. This high molecular
weight peptide has a terminal lysine that may depress its
ionization in MALDI-MS.27 This third potential glycosylated site
may be able to be detected by the use of a different protease
or by the use of a different matrix to improve its ionization.
Alternatively, LC-ESI-MS could be used because this usually
gives better coverage of tryptic peptides from a single protein.
The data above illustrate that information of site specific
glycosylation could be obtained by mass spectrometry from
samples after PNGase F release of oligosaccharides, but also
Journal of Proteome Research • Vol. 1, No. 6, 2002 527
research articles
Wilson et al.
Figure 5. MALDI-MS of trypsin peptides of R1-antitrypsin from PVDF membrane. A. Tryptic digested peptides without PNGase F. B.
Tryptic digestion after PNGaseF digestion followed by tryptic digestion. Peptides containing a glycosylation site which are present in
the tryptic digest only after PNGase F digestion are included. The masses of the deglycosylated peptides correspond to sequences
where the glycosylated asparagine has been converted to aspartic acid following PNGase F treatment. All masses shown correspond
to tryptic peptide masses.
suggests that further optimization may be needed to obtain
specific information on particular glycosylation sites. The
approach to obtain this information may turn out to be sample
dependent, and a standard protocol that enables detection of
all potential glycosylation sites may be difficult to develop.
The capacity to analyze both the structures of the N-linked
oligosaccharides and the sites of N-linked glycosylation on the
protein glycoforms separated by 2D-PAGE allows the analysis
of glycoproteins to be integrated with other high-throughput
techniques in proteomics. Both N- and O-linked oligosaccharides can be analyzed from the same protein glycoform
separated by 2D-PAGE and electroblotted to PVDF membrane.
As a general rule of thumb, these analyses can be performed
on glycoproteins able to be visualized by Coomassie Blue stains.
By depleting the major proteins from human plasma prior to
2D-PAGE, it was possible to load sufficient quantities of the
less abundant glycoproteins to allow for their analysis. The
loading of higher concentrations of the lower abundant proteins makes it possible to obtain accurate identification of these
proteins which usually pose a problem for MALDI-MS.
This oligosaccharide analysis technique is able to analyze
mixtures of neutral and acidic oligosaccharides without prefractionation or derivatization. This reduces the amount of
sample required and simplifies the automation of the analysis
protocol. Compared to other methods currently in use (refs 8,
10, and 14), sample handling in front of the mass spectrometric
analysis is kept to a minimum. Although the analysis time with
LC-ESI-MS is quite long (up to 45 min/sample), compared a
relatively quick MALDI-MS approach, the resolution of the
graphitized carbon column will provide information about
isomeric heterogeneity within the oligosaccharide sample. This
528
Journal of Proteome Research • Vol. 1, No. 6, 2002
information is not easily obtained with current methods for
analyzing oligosaccharides with MALDI-MS. The amount of
proteins that could be detected on a single 2D-PAGE gel, even
with as insensitive a stain as Commassie, could be substantial.
Even if the time could be reduced by a faster elution gradient
in LC-ESI-MS, the time that it would take to analyze all of
the proteins for glycosylation would be still be impractical
(weeks just for the elution time from the LC for analyzing both
N- and O-linked oligosaccharides from a gel with 200 proteins).
Hence, it is important to have specific stains for detection of
glycoproteins. In this report, the PAS-stain was used for
targeting glycoproteins for the analysis, but other more sensitive
glycoprotein stains could be used.28 The amount of analysis
could also be limited by intelligent preparation of samples
before 2D-PAGE separation, to target glycoproteins of interest.
In a comparative study, the interest will only be on proteins
that have been up or down regulated or have an altered
position on the gel. An efficient way to detect disease-related
glycosylation could be to specifically target those proteins for
analysis. For this kind of analysis, the described approach for
detecting glycosylation differences would be very well suited.
Acknowledgment. We thank Cameron Hill for his
technical expertise in running 2D PAGE and electroblotting and
Dr David Oxley for his technical support in mass spectrometry.
References
(1) Schachter, H. Congenital disorders involving defective N-glycosylation of proteins. Cell. Mol. Life Sci. 2001, 58, 1085-104.
(2) Dennis, J. W.; Granovsky, M.; Warren, C. E. Glycoprotein glycosylation and cancer progression. Biochim. Biophys. Acta 1999,
1473, 21-34.
research articles
N- and O-Linked Glycosylation Analysis
(3) Brockhausen, I. Biosynthesis and functions of O-glycans and
regulation of mucin antigen expression in cancer. Biochem. Soc.
Trans. 1997, 25, 871-874.
(4) Wilkins, M.; Williams, K. ProteomicssThe fundamentals of the
post-genome era. Drug Market Dev. 1998, 9, 150-154.
(5) Righetti, P. G.; Bassi, A. Isoelectric focusing of proteins and
peptides in gel slabs and in capillaries. Anal. Chim. Acta 1998,
372, 1-19.
(6) Packer, N. H.; Wilkins, M. R.; Golaz, O.; Lawson, M. A.; Gooley,
A. A.; Hochstrasser, D. F.; Redmond, J. W.; Williams, K. L.
Characterisation of human plasma glycoproteins separated by
two-dimensional gel electrophoresis. Biotechnology 1996, 14, 6670.
(7) Hoffmann, A.; Nimtz, M.; Conradt, H. S. Molecular characterisation of beta-trace protein in human serum and urinesA
potential diagnostic marker for renal diseases. Glycobiology 1997,
7, 499-506.
(8) Charlwood, J.; Skehal, M. J.; Camilleri, P. Analysis of N-linked
oligosaccharides released from glycoproteins separated by twodimensional gel electrophoresis. Anal. Biochem. 2000, 284, 4959.
(9) Mills, P. B.; Mills, K.; Lohnson, A. W.; Clayton, P. T.; Winchester,
B. G. Analysis by matrix assisted laser desorption/ionization-timeof-flight mass spectrometry of the post-translational modifications of alpha(1)-antitrypsin isoforms separated by two-dimensional polyacrylamide gel electrophoresis. Proteomics 2001, 1,
778-786.
(10) Küster, B.; Wheeler, S. F.; Hunter, A. P.; Dwek, R. A.; Harvey, D.
J. Sequencing of N-linked oligosaccharides directly from proteins
gels: In-gel deglycosylation followed by matrix-assisted laser
desorption/ionization mass spectrometry and normal phase highperformance liquid chromatography. Anal. Biochem. 1997, 250,
82-101.
(11) Papac, D. I.; Wong, A.; Jones, A. J. Analysis of acidic oligosaccharides and glycopeptides by matrix-assisted laser desorption/
ionization-time-of-flight-mass spectrometry. Anal. Chem. 1996,
68, 3215-3223.
(12) Mechref, Y.; Novotny, M. V. Matrix-assisted laser desorption
ionization mass spectrometry of acidic glycoconjugates facilitated
by the use of spermine as a co-matrix. J. Am. Soc. Mass Spectrom.
1998, 9, 1293-1302.
(13) Karas, M.; Hillenkamp, D. J.; Ehring, H.; Stahl, B.; Strupat, K.;
Nordhoff, E. Matrix-assisted laser desorption/ionization mass
spectrometry with additives to 2,5-dihydroxybenzoic acid. Org.
Mass. Spec. 1998, 28, 1476-1481.
(14) Powell, A. K.; Harvey, D. J. Stabilization of sialic acids in N-linked
oligosaccharides and gangliosides for analysis by positive ion
matrix-assisted laser desorption/ionization mass spectrometry.
Rapid Commun. Mass Spectrom. 1996, 10, 1027-1032.
(15) Harvey, D. J. Matrix-assisted laser desorption/ionization mass
spectrometry of carbohydrates. Mass. Spec. Rev. 1999, 18, 349450.
(16) Kawasaki, N.; Ohta, M.; Hyuga, S.; Hyuga, M.; Hayakawa, T.
Application of liquid chromatography with tandem mass spec-
(17)
(18)
(19)
(20)
(21)
(22)
(23)
(24)
(25)
(26)
(27)
(28)
trometry to the analysis of the site-specific carbohydrate heterogeneity in erythropoietin. Anal. Biochem. 2000, 285, 82-91.
Okafo, G.; Burrow, L.; Carr, S. A.; Roberts, G. D.; Johnson, W.;
Camilleri, P. A coordinated high-performance liquid chromatographic, capillary electrophoretic, and mass spectrometric approach for the analysis of oligosaccharide mixtures derivatized
with 2-aminoacridone. Anal. Chem. 1996, 68, 4424-4430.
Hatton, M. W. Studies on the coagulant enzyme from Agkistrodon
rhodostoma venom isolation and some properties. Biochem. J.
1973, 131, 799-807.
Cohn, E. J.; Strong, L. E.; Hughes, W. L., Jr.; Mulford, D. J.;
Ashworth, J. N.; Melin, M.; Taylor, H. L. Preparation and properties of serum and plasma proteins. IV. A system for the separation
into fractions of the protein lipoproteins components of biological
tissues and fluids. J. Am. Chem. Soc. 1946, 68, 459-475.
Oncley, J. L.; Melin, M.; Richert, D. A.; Cameron, J. W.; Gross, P.
M., Jr. The separation of the antibodies, isoagglutinins, prothrombin, plasminogen and β1-lipoprotein into sub-fractions of
human plasma. J. Am. Chem. Soc. 1949, 71, 541-550.
Herbert, B.; Galvani, M.; Hamdan, M.; Olivieri, E.; McCarthy, J.;
Pedersen, S.; Righetti, P. G. Reduction and alkylation of proteins
of two-dimensional map analysis: Why, when and how? Electrophoresis 2001, 22, 2046-2057.
Kyhse-Anderson, J. Electroblotting of multiple gels: A simple
apparatus without buffer tank for rapid transfer of proteins from
polyacrylamide to nitrocellulose. J. Biochem. Biophys. Methods
1984, 10, 203-209.
Packer, N. H.; Ball, M. S.; Devine, P. L.; Patton, W. F Detection of
glycoproteins in gels and blots. In The Protein Protocols Handbook, 2nd ed.; Walker, J., Ed; Humana Press. Inc.: Totowa, NJ,
2002; Chapter 104, pp 761-772.
Schulz, B. L.; Packer, N. H.; Oxley, D.; Karlsson, N. G. Identification of a human tear fluid DMBT1 variant: A major high
molecular weight glycoprotein in human tears. Biochem. J. 2002,
in press.
Carlson, D. M. Oligosaccharide isolated from pig submaxillary
mucin. J. Biol. Chem. 1966, 241, 2984-2986.
Watzlawick, H.; Walsh, M. T.; Yoshioka, Y.; Schmid, K.; Brossmer,
R. Structure of the N- and O-glycans of the A-chain of human
plasma alpha 2HS-glycoprotein as deduced from the chemical
composition of the derivatives prepared by stepwise degradation
with endoglycosidases. Biochemistry 1992, 31, 12 198-12 203.
Brancia, F. L.; Oliver, S. G.; Gaskell, S. J. Improved matrix-assisted
laser desorption/ionization mass spectrometric analysis of tryptic
hydrolysates of proteins following guanidination of lysinecontaining peptides. Rapid Commun. Mass Spectrom. 2000, 14,
2070-2073.
Steinberg, T. H.; Top, K. P. O.; Berggren, K. N.; Kemper, C.; Jones,
L.; Diwu, Z. J.; Haugland, R. P.; Patton, W. F. Rapid and simple
single nanogram detection of glycoproteins in polyacrylamide
gels and on electroblots. Proteomics 2001, 1, 841-855.
PR025538D
Journal of Proteome Research • Vol. 1, No. 6, 2002 529