Analysis of Mucin-Type
O
-Linked Oligosaccharides 191
191
16
Structural Analysis
of Mucin-Type
O
-Linked Oligosaccharides
André Klein, Gérard Strecker, Geneviève Lamblin,
and Philippe Roussel
1. Introduction
The carbohydrate moiety of mucin is characterized by the presence of oligosaccha-
rides linked to the peptide backbone by an O-glycosidic linkage between an N-
acetylgalactosamine residue and a hydroxylated amino acid (serine or threonine).
These linkages are alkali labile and the carbohydrate chains can be released as oli-
gosaccharide-alditols by a β-elimination, with NaOH in the presence of NaBH
4
. The
structures of carbohydrate chains found in mucins can be as simple as the disaccharide
NeuAc α2→6GalNAc in ovine submaxillary mucin and as complex as the ones found
in human respiratory or salivary mucins, in which several hundred different carbohy-
drate chains exist (1,2). This diversity is generated (1) by the different monosaccha-
rides constituting the glycans, generally fucose, galactose, N-acetylgalactosamine,
N-acetylglucosamine, and N-acetyl neuraminic acid, but also other monosaccharides
such as ketodesoxynonulosonic acid or N-glycolylneuraminic acid, and, finally the
occurance of sulfation of galactose and N-acetylglucosamine (3,4); and (2) by the dif-
ference in length, in branching, and by the occurrence of all the different possible
linkages between the constituting monosaccharides. The diversity of mucin-type oli-
gosaccharides can be extreme. For example, 88 oligosaccharides have been isolated
from the respiratory mucins of a single individual (5–10) and more than 150 have been
isolated from the jelly coat from eggs of different species of amphibians (11–15).
This chapter gives an overall idea of the strategy of elucidation of primary structure
of glycans, the most current nuclear magnetic resonance (NMR) techniques used in
structure determination, and some of the mass spectrometry (MS) techniques avail-
able for the glycobiologist.
1.1. Strategy
The amount of pure oligosaccharide (or of a mixture of two compounds, or three at
the most) and the facilities that are available in the laboratory environment will define
From:
Methods in Molecular Biology, Vol. 125: Glycoprotein Methods and Protocols: The Mucins
Edited by: A. Corfield © Humana Press Inc., Totowa, NJ
192 Klein et al.
the strategy. Two types of techniques are used: a destructive technique, MS and a
nondestructive technique, NMR. Figure 1 summarizes the sequence of techniques.
When the amount of oligosaccharide permits the use of NMR first, especially if two-
dimentional (2D) NMR can be performed, MS is only used to confirm the structure if
it is a novel one; in most cases, it is not needed. MS is quite useful when there is not
enough material for NMR or when the mixture studied is too complex.
1.2. Nuclear Magnetic Resonance
NMR spectroscopy constitutes the most suitable method for the structure determi-
nation of carbohydrate chains. This method was introduced in the 1970s and rapidly
received a large application for analyzing the sequence of N-acetyllactosamine- and
oligomannosidic-type glycans.
Originally, and before the development of 2D NMR spectroscopy, the method was
limited to one-dimentional
1
H-NMR spectroscopy, which has led to the concept of
“structural-reporter groups.” In these conditions, depending on the field of spectrom-
eter (600–300 MHz), 20–100 nmol will constitute a sufficient amount of material to
apply a “finger-print” method. Nevertheless, the method is restricted to compounds
which are members of a series of closely related sequences, as it is happily the case for
most of O-glycans.
When the material is available in the range of 0.1–5 µmol, the de novo structural
elucidation of the sequences can be easily deduced from the compilation of data fur-
nished by various homo and heteronuclear 2D NMR methods.
1.3. Mass Spectrometry
MS has become an indispensable tool for the determination of carbohydrate struc-
tures. The information provided by this methodology ranges from the accurate mo-
lecular weight determination to the complete primary structure with a sensitivity such
that only picomoles of oligosaccharides are necessary. These remarkable advances
have been made possible with the appearence of novel methods of ionization such as
fast atom bombardment ionization (FAB), electrospray ionization (ESI), and matrix-
assisted laser desorption ionization (MALDI).
2. Methods
2.1. Nuclear Magetic Resonance
2.1.1. Proton-NMR as a Fingerprinting Method
The proton-NMR method was developed by Vliegenthart and colleagues during the
1970s and essentially applied to the structure determination of N-glycans of the N-
acetyllactosamine and oligomannoside type (16). More recently, a similar procedure
was summarized for the primary structural analysis of oligosaccharide-alditol released
from mucin-type O-glycosylproteins (17). This method is based on the recognition of
some atom resonances that constitute probes for representative structural elements.
These structural-reporter groups resonate outside the bulk constituted by the
nonanomeric protons.
1
H-NMR structural-reporter group signals correspond to the
following atom resonances: anomeric protons; GalNAc-ol H-2, H-4, H-5 and H-6'
Fig. 1
Analysis of Mucin-Type
O
-Linked Oligosaccharides 193
atoms; Gal H-3 and H-4 atoms; Fuc H-5 and H-6 atoms; NeuAc H-3ax and H-3eq
atoms; and CH
3
of the acetamido groups.
The first step of spectrum analysis consists of the identification of the core region
(Table 1), based on the characteristic chemical shifts of the H-2 and H-5 atom reso-
nances of the GalNAc-ol unit. Moreover, the quadruplet of the H-6' signal of GalNAc-
ol is upfield shifted out of the bulk at δ ~ 3.50 ppm in the case of an O-6 substitution
with sialic acid. The presence of α-2,3- or α-2,6-linked sialic acid is clearly shown by
the respective chemical shift of the H-3ax and H-3eq signals of the monosaccharides.
The H-3ax and H-3eq resonances of the α-2,3-linked NeuAc are systematically
downfield shifted, compared to the corresponding signals of α-2,6-linked NeuAc.
The attachment of NeuAc at O-3 of a Gal unit causes downfield shifts of the Gal
structural-reporter groups, as clearly indicated in Fig. 2, in which the NMR spectra of
asialo and sialo glycans are compared (compounds N-1 and A-1).
Fig. 1. Strategy of elucidation of oligosaccharide primary structure.
194Klein et al.
194
Table 1
Chemical Shifts of GalNAc-ol Residues Characteristic of the Nature of Oligosaccharide-Alditol Cores
Gal(β1–3)GalNAc-ol GlcNAc(β1–3)GalNAc-ol Gal(β1–3)[GlcNAc(β1–6)]GalNAc-ol NeuAc(α2–6)GalNAc-ol
H-2 4.393 4.286 4.391 4.246
H-3 4.063 3.995 4.069 3.846
H-4 3.506 3.546 3.468 3.411
H-5 4.193 4.141 4.277 4.020
H-6' 3.628 ND ND 3.532
Gal(β1–3)[NeuAc(α2–6)]GalNAc-ol GlcNAc(β1–3)[NeuAc(α2–6)]GalNAc-ol GlcNAc(β1–6)GalNAc-ol
H-2 4.378 4.260 4.242
H-3 4.055 3.984 3.841
H-4 3.534 ND 3.379
H-5 4.244 4.185 4.021
H-6' 3.486 3.490 3.933
Analysis of Mucin-Type
O
-Linked Oligosaccharides 195
Fig. 2.
1
H-NMR spectra of three oligosaccharide-alditols. N-1, basic structure devoid of
fucose and sialic acid residue; N-2, downfield shift of Gal
3
H-1 owing to the α-1,2 fucose; A-
1, downfield shift of Gal
3
H-1 and H-3 owing to the α-2,3 sialylation. The first superscript after
the abbreviated name of a monosaccharide residue indicates to which position of the adjacent
monosaccharide it is glycosidically linked (e.g., Gal
4
in the case of Galβ1→4 GlcNAcβ1→).
Fucose units can be easily identified according to the presence of methyl reso-
nances at ~ 1.2 ppm. The α-1,2 (H), α-1,3 (Le
x
), and α-1,4 (Le
a
) linkages are deduced
from the position of H-1, H-5, and H-6 resonances. For instance, the Le
x
and Le
a
epitopes can be characterized on the basis of their H-1 (Le
x
: δ ~ 5.13–5.14; Le
a
: δ ~
5.02–5.05), and H5 (Le
x
: δ ~ 4.80–4.85; Le
a
: δ ~ 4.86–4.88) resonances, whereas the
196 Klein et al.
H-5 signal of α-1,2-linked Fuc is observed at δ ~ 4.25–4.34 ppm. These NMR data
imply, respectively, the presence of type 2 (Galβ1-4GlcNAc) and type 1 (Galβ1-
3GlcNAc) backbone structures.
The analysis of compounds with a higher complexity in backbone sequence cannot
be developed in some pages, and such analyses often call for additional chemical (me-
thylation analysis) or physical (MS, nuclear Overhauser effect [NOE] measurements)
operations.
Figures 3 and 4 give examples of complex backbones. Compounds were analyzed
by methylation analysis, which precises the location of the hydroxyl groups impli-
cated in the glycosidic linkages. The comparison of the spectra N-4B, N-4A, N-6B,
and N-8 clearly indicates the presence of the same sequence Gal(β1-4)GlcNAc(β1-6)
GalNAc-ol, as shown by the NMR parameters of Gal
4
and GlcNAc
6
. This comparison
between N-1, A-1, and A-3 (Fig. 2) also gives the chemical shift of the sialylated and
terminal Gal unit O-3 linked to GalNAc-ol. Consequently, the upper branch of com-
pound A-3 is composed of two Gal and two GlcNAc units. Other models (not shown
here) also possess the sequences Gal(β1-3)GlcNAc(β1-3)Gal(β1-4)GlcNAc or
Gal(β1-4)GlcNAc(β1-3)Gal(β1-4)GlcNAc, with additional Fuc or NeuAc attached to
the terminal galactose units. Since the presence of these peripheral monosaccharides
affects essentially the chemical shifts of the terminal galactose, the anomeric protons
can be assigned step-by-step.
The compilation of 168 NMR spectra which are included in the review of Kamerling
et al. (17) clearly shows that a spectrum is unique and can be used as an “identity
card.” This review presents the carbohydrate chains in a logical order and gives a
classification according to the nature of the core, the nature of the backbone, and the
peripherical monosaccharides. Actually, the nature of the backbone can be often di-
Fig. 2. Continued.
Analysis of Mucin-Type
O
-Linked Oligosaccharides 197
rectly deduced from the linkage of the fucose units, as discussed previously. When the
presence of these structural elements has been established, the number of possibilities
decreases deeply, and a survey of the corresponding class of oligosaccharide-alditols
furnishes rapidly the structure of the compound. Most of the O-glycans that constitute
the carbohydrate moiety of mucins isolated from human tissues have now been de-
scribed, but completely new structures can be isolated from other biological sources.
If a sufficient amount of material is available, de novo structural elucidation of glycan
sequences should be performed by 2D NMR spectroscopy.
Fig. 3.
1
H-NMR spectra of oligosaccharide-alditols with the common tetrasaccharide N-1
(see Fig. 2) variously substituted.
198 Klein et al.
2.1.2.
De novo
Structural Elucidation of Glycan Sequences
by NMR Spectroscopy
Homonuclear correlated spectroscopy (COSY) provides information on directly
coupled protons with regard to coupling constants. Starting from the anomeric proton,
the H-2, H-3, H-4, and so on, atom resonances, which were masked in the bulk, can be
assigned. Nevertheless, such an assignment is generally difficult when the cross peaks
Fig. 3. Continued.
Fig. 4.
1
H-NMR spectrum of oligosaccharide-alditols with the common tetrasaccharide N-1
(see Fig. 2) variously substituted.
Analysis of Mucin-Type
O
-Linked Oligosaccharides 199
occur around the diagonal. Two-dimensional total correlation spectroscopy (2D
1
H-
TOCSY) can be used to characterize all the proton resonances. A transfer of magneti-
zation from H-1 to H-6 is observed in the case of the α− and β-gluco configuration
(J
1,2
–J
4,5
~ 8 Hz), wheareas the small J
4,5
that characterizes the galacto configuration
interrupt the assignment at the H-4 resonance. Relayed COSY spectra have the advan-
tage of successively assigning the H-2 (COSY), H-3 (one-step-relayed COSY), and H-
4 (two-steps-relayed COSY) atom resonance of the carbohydrate units.
The example given in Fig. 5 clearly shows the presence of two β-Gal, one α-Gal
and one GlcNAc units. For N-acetylglucosamine, H-2, H-3, and H-4 signals are trip-
Fig. 4. Continued.
200Klein et al.
Fig. 5. COSY (left) and HMQC (right) NMR spectra of an heptasaccharide-alditol isolated from the oviducal mucin of Xenopus laevis. 2
II
,
proton H-2 of monosaccharide unit II. For the COSY spectrum, starting from the anomeric proton (i.e., 1
III
) the protons 2
III
, 3
III
, and 4
III
are
Analysis of Mucin-Type
O
-Linked Oligosaccharides 201
lets (transaxial hydrogens), whereas the same resonances of β-galactose are triplet,
pseudo-doublet, and pseudo-singlet, respectively. Fucose, which possesses an α-
galacto configuration, is easily distinguished from α-galactose owing to the overlap
of the correlations H-6→H-5→H-4 and H-1→H-2→H-3→H-4.
The heteronuclear multiple quantum-coherence spectroscopy (HMQC) relies on
the
1
H and
13
C, which are directly attached. The position of the glycosidic linkage is
clearly observed owing to a strong downfield shift (4–10 ppm) that affect the substi-
tuted carbon. In the example depicted in Fig. 5, the crosspeak 4
II
observed at ~ 70 ppm
corresponds to an unsubstituted carbon, in opposite to the signal 4
III
at 80 ppm.
The most elegant method for establishing the exact sequence of the oligosaccharide
is indisputably the heteronuclear multiple-bond correlation spectroscopy (HMBC)
which relies the
1
H and
13
C via their
3
J
H,C
coupling. Unfortunately, the method is too
insensitive for being used systematically. In the example given in Fig. 5 these ex-
pected connectivities should be observed: 1
IV
→3
IV
, 5
IV
, 4
III
; 1
III
→3
III
, 5
III
, 3
III
; 1
II
→3
II
,
5
II
, 3
I
, and so on.
2D Nuclear Overhauser effect spectroscopy (NOESY) or rotating-frame NOE spec-
troscopy (ROESY) is generally used for establishing the sequence of carbohydrate
chains. Since the strongest NOE is not always between the protons connected to the
linkage, the method may fail to establish the position of the glycosidic substitution.
Figure 6 describes such an assignment. A preliminary methylation analysis has shown
the presence of one terminal Gal, two O-3-substituted GalNAc, two O-4-substituted
Gal, and one O-3-substituted GalNAc-ol. The COSY experiment indicates that Gal
and GalNAc have β and α configuration, respectively. The exact assignment of the
five anomeric protons resulted from the connectivities 1
IV
/3
V
, 1
V
/4
IV
, 1
IV
/3
III
, 1
III
/4
II
,
and 1
II
/3
I
.
Fig. 5. (continued) successively assigned. The shape of the correlation peaks (triplet for H-2,
pseudo-doublet for H-3, pseudo-singlet for H-4) allows the estimatation of the coupling con-
stant: L (large) for J > 6 Hz; S (small) for J < 4 Hz. For the monosaccharide unit III, we observe
J
1,2
> 6 Hz; J
2,3
> 6 Hz; J
3,4
<4 Hz; J
4,5
< 1 Hz. These values demonstrate the following configu-
ration of the ring protons: H-1, axial; H-2, axial; H-3, axial; H-4, equatorial, which are charac-
teristic of a galactose residue in a β configuration. With the same demonstration, units II, II,
and IV possess, respectively the β-Gal, β-Glc, and α-Gal configuration. Hexose and N-
acetylhexosamine are discriminated according to their H-2 resonances, strongly deshielded in
the case of N-acetylhexosamine. For the HMQC spectrum, a comparison with the correspond-
ing COSY spectrum allows the assignment of the
13
C resonances (i.e. 4
III
represents the corre-
lation peak between Gal III
1
H-4/
13
C-4). For the β-Gal units, the
13
C-atom resonances 2
II
, 3
II
,
2
III
, and 4
III
are deshielded from 3 to 10 ppm, as compared to the standard values observed for
β-methylgalactoside. On the contrary, the signal 3
III
and 4
II
possess normal values (nonsubsti-
tuted hydroxyl group at position 3 and 4, respectively).
202Klein et al.
Fig. 6. COSY (left) and ROESY (right) NMR spectra of a hexasaccharide-alditol isolated from the oviducal mucin of Rana palustris. 2
II
,
proton H-2 of monosaccharide unit II. For the COSY spectrum, the carbohydrate units were identified on the basis of the set of
the vicinal
coupling constants as described in Fig. 5. For the ROESY spectrum, the correlations 1
VI
→ 3
V
, 1
V
→
4
IV
, 1
IV
→ 3
III,
1
III
→ 4
II
,
and 1
II
→ 3
I
clearly indicate the sequence of the oligosaccharide and confirm the assignment of each anomeric proton.
Analysis of Mucin-Type
O
-Linked Oligosaccharides 203
2.2. Mass Spectometry
2.2.1. FAB-Mass Spectometry
2.2.1.1. P
RINCIPLE
The sample containing 20–100 pmol of analyte is mixed with the matrix on the target.
Bombardment of this sample/matrix mixture in the ionization source with an energetic
beam of atoms such as argon or xenon forms ions that are extracted from the surface of
the target, accelerated, and then mass analyzed in the spectrometer. The FAB ionization
method can be applied to analysis of native or derivatized oligosaccharides.
2.2.1.2. M
ETHODS
Oligosaccharides must be prepared in a desalted form to remove all sodium or
other metal from the sample in order to enhance the production of the [M+H]
+
or [M-
H]
–
ion. For experiments in which the M+Na ion is desired, the appropriate salt can be
added to the sample. The matrix used is either glycerol for native oligosaccharides or
thioglycerol for permethylated samples (18). An alkaline matrix can be used such as
glycerol:water:triethanolamine (3:2:1, v/v/v) and will amplify the signal for sulfated
or sialylated compounds.
Native oligosaccharides exhibit poorer response than their derivatized forms. Car-
bohydrate permethylation and peracetylation are the most commonly used procedures.
The derivatization of the oligosaccharide presents many advantages : it increases the
sensitivity, the desalting step is rendered easier, and the fragmentation patterns of the
derivatives are well defined (18).
NaOH permethylation according to Ciucanu and Kerek (19) is a simple and effi-
cient procedure. Dried oligosaccharides are first dissolved in dimethyl sulfoxide. Then
finely powdered NaOH is added with methyliodine, the reaction mixture is placed in
an ultrasonic bath for 1 to 2 h at room temperature, and the reaction is stopped by
careful addition of water. The permethylated oligosaccharide is extracted by chloro-
form or by solid-phase extraction procedure over a Sep-Pak
®
cartridge (Waters,
Milford, MA). For neutral oligosaccharide-alditols, this is the method of choice. The
methylated oligosaccharide-alditols are dissolved in methanol containing sodium ac-
etate (0.1%) and loaded on the metal target with thioglycerol as a matrix (2). If suffi-
cient material is available, the methylated sample can be further analyzed after
methanolysis and acetylation to identify the methylated derivatives (20).
The NaOH methylation procedure is too strong for sulfated oligosaccharide. These
samples require a modified Hakomori procedure (21), which preserves the sulfate resi-
due. The oligosaccharides are dissolved in methyl sulfoxide to which is sequentially
added a solution of methylsulfinylmethanide (sodium hydride with methyl sulfoxide)
and methyliodine at room temperature. The reaction is stopped by water, and the me-
thylated oligosaccharides are purified over a C18, Sep-Pak cartridge (22).
2.2.1.3. R
ESULTS
Most informative signal are given by the pseudomolecular ions corresponding to
(M+H)
+
, (M+Na)
+
when recorded in the positive mode and (M-H)
–
in the negative
204 Klein et al.
mode. Molecular ions give information on the chemical composition of the oligosac-
charides if they are composed of the generally constituting monosaccharides (hexose,
N-acetylhexosamine, deoxyhexose, N-acetylneuraminic acid). The increments of
masses per residue are given in atomic mass units in Table 2, to this sum the mass
carresponding to the number of linkage has to be substracted.
When sufficient material is available, fragmentation of the oligosaccharide in the
source may occur. With underivatized oligosaccharides, the fragmentation patterns
are ambiguous, rendering the determination of the sequence uncertain. Permethylated
oligosaccharides have a pattern of breakdown that results from the cleavage of the
glycosidic bonds. Different pathways of fragmentation have been described, in which
the charge is retained either on the reducing or on the non-reducing end of the oli-
gosaccharide (18,23). When ring cleavages are observed, they are more difficult to
assign but give off the linkage position. Finally, secondary ions, resulting from the
preferential elimination of substituents on a certain position (i.e., 3-position of an N-
acetylhexosamine) complete the structural identification of the oligosaccharide.
2.2.2. MALDI-Mass Spectometry
2.2.2.1. P
RINCIPLE
The sample containing 1–100 pmol of analyte is mixed with the matrix and loaded on a
sample slide. The function of the matrix is to isolate the sample molecules from each other
and to absorb photons from an ultraviolet laser, usually a 337-nm N
2
laser. Absorption of
the laser energy by the matrix leads to a fast ejection of molecular material. Ionization
takes place by proton transfer from matrix ions to neutral analyte molecules. This mild
ionization occurs in a gas phase resulting from the energy of the laser at the surface of the
slide. MALDI is usually combined with a time-of-flight (TOF) mass spectrometer.
2.2.2.2. M
ETHODS
Oligosaccharide samples (10–50 pmol dissolved in 1 mL of water) and the matrix
solution (1 mL of a solution of 2, 5-dihydroxybenzoic acid [12 mg/mL] in meth-
Table 2
Calculation of Molecular Weight of Oligosaccharides
a
Mass incrementation in Mass incrementation in permethylated
Compounds native oligosaccharides oligosaccharides
Hexose 180 250
Desoxyhexose 164 220
N-acetylhexosamine 221 291
N-acetylhexosaminitol 223 307
N-acetylneuraminic acid 309 407
Glycosidic linkage –18 –46
aThe nominal value of m/z is given excluding the fractional mass increment of each atom; e.g., H is
counted as 1 instead of 1.008. The molecular weight is obtained by the sum of the respective molecular
weight of each constituting monosaccharide minus the weight corresponding to the number of glyco-
sidic linkages.
Không có nhận xét nào:
Đăng nhận xét