Salt-free fractionation of complex isomeric mixtures of glycosaminoglycan oligosaccharides compatible with ESI-MS and microarray analysis

Abstract

Heparin and heparan sulfate (Hp/HS) are linear complex glycosaminoglycans which are involved in diverse biological processes. The structural complexity brings difficulties in separation, making the study of structure-function relationships challenging. Here we present a separation method for Hp/HS oligosaccharide fractionation with cross-compatible solvent and conditions, combining size exclusion chromatography (SEC), ion-pair reversed phase chromatography (IPRP), and hydrophilic interaction chromatography (HILIC) as three orthogonal separation methods that do not require desalting or extensive sample handling. With this method, the final eluent is suitable for structure-function relationship studies, including tandem mass spectrometry and microarray printing. Our data indicate that high resolution is achieved on both IPRP and HILIC for Hp/HS isomers. In addition, the fractions co-eluted in IPRP could be further separated by HILIC, with both separation dimensions capable of resolving some isomeric oligosaccharides. We demonstrate this method using both unpurified reaction products from isomeric synthetic hexasaccharides and an octasaccharide fraction from enoxaparin, identifying isomers resolved by this multi-dimensional separation method. We demonstrate both structural analysis by MS, as well as functional analysis by microarray printing and screening using a prototypical Hp/HS binding protein: basic-fibroblast growth factor (FGF2). Collectively, this method provides a strategy for efficient Hp/HS structure-function characterization.

Figure 1

Multi-dimensional separation workflow. Data shown are taken from this study to represent the value of each step. (A) SEC separation of the partially depolymerized Hp/HS, enoxaparin sodium; (B) chemo-selective reductive amination of one SEC fraction (dp8) with AEAB; (C) IPRP separation of AEAB-labeled oligosaccharide SEC fraction (dp8); IPRP fractions could be either analyzed by (D) online HILIC LC/MS (IPRP fraction RT 35 to 36 min shown); or (E) separated by offline HILIC with fractions collected (IPRP fraction RT 36.8 to 38 min shown); (F) microarray immobilization of HILIC fractions and protein binding affinity assay (IPRP and HILIC fraction identities shown in Table S1).

Figure 2

Chemo-selectivity of reductive amination of Hp/HS with AEAB. (A) Scheme of AEAB reductive amination. The difference in pKa between the aromatic amine and the aliphatic amine achieves selective reactivity by controlling the reaction pH. (B) Extracted ion chromatogram of Hp disaccharide II-S: AEAB conjugate after reaction in different concentrations of acetic acid (v/v). The earlier eluting peak represents conjugation of AEAB through the aliphatic amine, with the desired conjugation product through the aromatic amine eluting later.

Figure 3

IPRP chromatogram of AEAB labeled synthetic hexasaccharide mixture with UV detection at (blue) 232 nm and (orange) 260 nm. After labeling with AEAB, these synthetic hexasaccharides were separated on IPRP. Peaks were initially assigned based on the elution time of the major peak in the individual synthetic reaction product (Supplementary Information Fig. S5). HILIC LC-MS data from fractions shown in black, red, and green are shown in Fig. 4.

Figure 4

The extracted ion chromatograms of [M -4H]⁴⁻ ion corresponding to AEAB-labeled [1, 2, 3, 8, 0, 1] from IPRP fractions. The EICs have shown that the derivatized hexasaccharide composition [1, 2, 3, 8, 0, 1] collected from IPRP in the 22 to 23 min fraction (top) eluted from the HILIC column around 27.1 min, while the same composition from the IPRP 25 to 26 min fraction (bottom) eluted around 28.0 min on HILIC. The IPRP 23 to 24 min fraction contains the shoulders of both isomer peaks in the IPRP chromatogram, and by HILIC (middle) both the 27.1 and the 28 min peak are partially resolved.

Figure 5

Separation of AEAB labeled enoxaparin octasaccharides using IPRP detected using 232 nm (blue trace) and 260 nm (orange trace) wavelengths. The heterogeneous enoxaparin octasaccharide fraction separated into many poorly-resolved peaks across the elution window. The four fractions highlighted in the chromatogram were collected for further HILIC separation and MS analysis as examples.

Figure 6

The HILIC EICs of [M -4H]⁴⁻ ion corresponding to [1, 3, 4, 8, 1, 1] from IPRP fractions. (A) In the IPRP 35 to 36 min fraction, this composition eluted at 27.8 min. (B) In the IPRP 40–41 min fraction, a different structure with this composition is eluted at 19.7 min. Oligosaccharide compositions are given as [∆HexA, HexA, GlcN, SO₃, Ac, AEAB].

Figure 7

The EICs of the top four octasaccharide compositions from the IPRP 35–36 min fraction. The octasaccharides, which are not separated by IPRP, are resolved by HILIC, including two clearly separated [1, 3, 4, 8, 2, 1] isomers, illustrating orthogonality between HILIC and IPRP chromatography. Oligosaccharide compositions are given as [∆HexA, HexA, GlcN, SO₃, Ac, AEAB].

Figure 8

FGF2-binding assay on Hp/HS octasaccharide-AEAB microarray. Each fraction is printed in replicates of 6 in a single block with droplet volume around 400 pL. Fractions from the multi-dimensional separation method are directly applicable to microarray study. Based on the fluorescent intensity of each spot, it indicates that each fraction has different binding affinities to FGF2. Fractions from No. 1 to No. 126 were eluents from the multi-dimensional separation method. Fractions No. 127 & 128 were unfractionated octasaccharide. Fraction No. 129 was synthetic octasaccharide as a positive control. An image of the microarray is in Supplementary Information Fig. S11. Detailed information of each fraction is listed in Table S1.

Figure 9

Correlation between IPRP retention time, HILIC retention time, and FGF2 binding response by microarray analysis at 1 µg/mL, represented by bubble size. A strong correlation between HILIC retention time and FGF2 binding is apparent, indicating that FGF2 is binding ligands that are more highly sulfated. No correlation between IPRP retention time and FGF2 binding is apparent.