Component mapping of GAGs by RP–HPLC and ESI–MS/MS / H. Zhu et al. / Anal. Biochem. 465 (2014) 63–69
65
Desalting of PMP-labeled sugars after RP–HPLC separation
in Fig. S1 of the online supplementary material. They were eluted
faster at higher pH values; the elution times went from 14 min
at pH 6.0 to 3.5 min at pH 8.0. The resolutions of GlcN–PMP and
ManN–PMP at pH values of 6.0 and 7.2 were above 1.0. As shown
in Figs. S2 and S3, the higher the percentage of CH3CN, the worse
the resolution of GlcN–PMP and ManN–PMP. The resolutions of
GlcN–PMP and ManN–PMP at 15% CH3CN and 14% CH3CN were
both above 1.5, but the derivatives were eluted faster at 15%
CH3CN. Therefore, the optimal mobile phase was determined to
be 15% (v/v) CH3CN in 0.1 M PBS (pH 6.0). The separation of
GlcA–PMP and IdoA–PMP was optimized using five elution condi-
tions (see Table S1 in supplementary material), as shown in Fig. S5,
and the resolution of GlcA–PMP and IdoA–PMP under the second
set of elution conditions was the best.
A Sep-Park classic C18 column (WAT051910, Waters, Milford,
MA, USA) was pretreated as described in the product directions
and equilibrated with 1% aqueous methanol before use. The
PMP-labeled sugars collected from RP–HPLC were dried and redis-
solved in 100 ll of 1% methanol. The sample was loaded onto the
column and eluted with 10 column volumes of 1% methanol to
remove salts, and then it was washed with 6 column volumes of
100% methanol to obtain the desalted sugar derivative. The
desalted sugar derivative was vacuum-dried and preserved at
À20 °C.
ESI–MS analysis
Based on the above conditions, the following two-step isocratic
gradient was selected to separate 10 PMP-labeled monosaccha-
rides: 15% CH3CN in 0.1 M PBS (pH 6.0) to separate PMP-labeled
Man, GlcN, and ManN and 17% CH3CN in 0.1 M PBS (pH 6.0) to sep-
arate PMP-labeled GalN, GlcA, IdoA, GalA, Glc, Gal, and Xyl subse-
quently. These monosaccharide derivatives were well separated in
60 min (Fig. 2A).
Experiments were performed on a Thermo Scientific LTQ Orbi-
trap XL hybrid Fourier transform mass spectrometer (FTMS)
equipped with an ESI source. The sheath gas flow rate was set at
8 arb. The capillary temperature was 275 °C. The spray voltage,
capillary voltage, and tube lens were maintained at 3 kV, 17 V,
and 80 V, respectively. The mobile phase was 50% CH3CN at a flow
rate of 8 ll/min. All of the samples were dissolved in mobile phase
before injection. For CID–MS/MS analysis, the collision energy was
Mapping of PMP-labeled acidolysis components of GAGs
adjusted to 20 eV.
The composition profiles of HP, HS, and ES (low-molecular-
weight heparin prepared by a b-elimination method) after hydroly-
sis at 110 °C for 3 h are shown in Fig. 2B–D, and GlcN, GlcA, and IdoA
were observed as expected. The GalN was from DS, which was a
common contaminant in HP/HS [26], and Gal and a little Xyl were
from the tetrasaccharide linker [27,28]. Both the repeating disac-
charide units of HP/HS and DS contain IdoA, so the IdoA in
Fig. 2B–C may be derived from HP/HS or DS. In addition, ManN
was found in ES, which was derived from GlcN by base-catalyzed
C-2 epimerization [29] during the preparation process, making it
distinguishable from HP/HS. The peak areas of the monosaccharides
were calibrated using linear equations (Fig. S6). The monosaccha-
ride compositions were calculated and are listed in Table 2. On
the whole, the ratio of uronic acids to GlcN was not detected as
1:1, which may be due to the degradation of uronic acids and also
the incomplete hydrolysis of these GAGs. Two unknown peaks in
HP, HS and ES (a and b in Fig. 2B–D), were also observed between
28 and 30 min. The peak areas of a and b varied with the hydrolysis
time, with component a exhibiting a maximum peak area at 8 h and
component b exhibiting its maximum response at 3 h (Fig. 3A),
indicating that peaks a and b corresponded to acidolysis-resistant
components derived from HP/HS. To elucidate their structures,
MS analysis was performed after desalting PMP-labeled compo-
nents a and b with a C18 SPE column. The molecular mass of com-
ponent b was 685 Da, validated by a mono-charged molecular ion in
the positive (m/z 686) and negative (m/z 684) ESI–MS modes, allow-
ing the conclusion that the structure of component b was a PMP-
labeled disaccharide likely composed of one GlcN, one GlcA/IdoA,
and two PMP according to the repeating unit of HP/HS. The positive
ESI–CID–MS/MS spectrum (Fig. 4B) of the parent molecular ion (m/z
686) illustrated that the main product ions at m/z 668, 512, and 525
and 507 were assigned as the losses of H2O, PMP, and glycosidic
bond cleavages, respectively, showing the existence of GlcN at the
nonreducing terminus. Component a also had a molecular mass of
685 Da and a similar structure to that of component b, as illustrated
in Fig. 4A. However, the structure of the uronic acid at the reducing
end of a and b could not be determined. A deamination reaction
with nitrous acid [23] was a method used to aid in the hydrolysis
of the disaccharide GlcN-IdoA/GlcA of HP/HS into free 2,5-anhydro
mannose and IdoA/GlcA (Fig. S7), which can be separated and
determined by HPAEC [15]. The peak areas of GlcA and IdoA in
the HP hydrolysate after the deamination reaction increased by
Deamination reaction
The hydrolysate (5
and mixed with 5 l of H2O. Then, 5
of 5.5 M NaNO2 were added. After 0.5 min, the reaction was
stopped with 10 l of 1.5 M Na2CO3 [23]. The mixture was filtered
with a 0.22- m centrifuge tube filter, and the filtrate was pre-
l
l) was transferred to a 1.5-ml glass tube
l
l
l of 1.5 M H2SO4 and 5
ll
l
l
served at À20 °C.
HPAEC conditions
The HPAEC analysis was performed as described previously by
Zhang and coworkers [15] with some modifications. Experiments
were performed on a Dionex ICS 3000 system (Dionex, Sunnyvale,
CA, USA) with a Carbopac PA 20 guard column (3 Â 30 mm,
Dionex) and a Carbopac PA 20 analytical column (3 Â 150 mm,
Dionex). The temperature and flow rate were set at 30 °C and
0.4 ml/min, respectively. A two-step isocratic elution process was
used, including 15 mM NaOH for 10 min and then 15 mM
NaOH–150 mM NaOAc for another 10 min. The entire process
was monitored by PAD (Au working electrode and Ag/AgCl refer-
ence electrode) using the Dionex default program of carbohydrate
quadruple waveform.
Results and discussion
Optimization of RP–HPLC conditions
The PMP labeling technique, reported by Pitt and Gorman [24],
has been widely used for carbohydrate analysis due to its mild
reaction conditions and simple cleanup procedure. PMP-labeled
monosaccharides were satisfactorily separated by 17% CH3CN in
0.1 M PBS (pH 6.7) in our laboratory’s previous work [20,25]. How-
ever, ManN–PMP and IdoA–PMP had identical retention times with
GlcN–PMP and GlcA–PMP, respectively, under the same chromato-
graphic conditions. Hence, several experiments were performed to
optimize the separation conditions of PMP-labeled Gal, Glc, Man,
Xyl, GalN, GlcN, ManN, GalA, GlcA, and IdoA.
The separation conditions of GlcN–PMP and ManN–PMP were
first optimized by 0.1 M PBS with different pH values, as shown
Zhu, He
