Resolving Heparan Sulfate Oligosaccharide Positional Isomers Using Hydrophilic Interaction Liquid Chromatography-Cyclic Ion Mobility Mass Spectrometry

Abstract

Heparan sulfate (HS) is a linear polysaccharide covalently attached to proteoglycans on cell surfaces and within extracellular matrices in all animal tissues. Many biological processes are triggered by the interactions among HS binding proteins and short structural motifs in HS chains. The determination of HS oligosaccharide structures using liquid chromatography-mass spectrometry (LC-MS) is made challenging by the existence of positional sulfation and acetylation isomers. The determination of uronic acid epimer positions is even more challenging. While hydrophilic interaction liquid chromatography (HILIC) separates HS saccharides based on their composition, there is a very limited resolution of positional isomers. This lack of resolution places a burden on the tandem mass spectrometry step for assigning saccharide isomers. In this work, we explored the use of the ion mobility dimension to separate HS saccharide isomers based on molecular shape in the gas phase. We showed that the combination of HILIC and cyclic ion mobility mass spectrometry (cIM-MS) was extremely useful for resolving HS positional isomers including uronic acid epimers and sulfate positions. Furthermore, HILIC-cIM-MS differentiated multicomponent HS isomeric saccharide mixtures. In summary, HILIC-cIM-MS provided high-quality data for analysis of HS oligosaccharide isomeric mixtures that may prove useful in the discovery of new structural motifs for HS binding proteins and for the targeted quality control analysis of commercial HS products.

Graphical Abstract

INTRODUCTION

Heparan sulfate (HS) is a linear polysaccharide linked to proteoglycans ubiquitously expressed on cell surfaces and in extracellular matrices in animal tissues.¹ A highly organized set of biosynthetic enzymes, including glycosyltransferases, deacetylases, epimerase, and sulfotransferases, biosynthesize HS chains in a spatially and temporally regulated manner, leading to heterogeneous sequence motifs that mediate interactions with proteins that underpin many well-described biological processes.^2,3

Mass spectrometry (MS) hyphenated techniques have played an important role in HS analysis, due to their high sensitivity, specificity for discerning subtle differences in structure, capability to examine complex mixtures, and ability to identify components in biologically active samples.^4,5 Prior to MS analysis, HS oligosaccharides may be separated using capillary electrophoresis^6,7 (CE) and chromatography modes including reversed phase-ion pairing (RP-IP),^8,9 hydrophilic interaction liquid chromatography (HILIC),^10-12 strong anion exchange (SAX),¹³ and porous graphitized carbon (PGC).¹⁴ By contrast, reversed-phase liquid chromatography (LC) has been used for the separation of permethylated, deuteroacetylated HS saccharides.¹⁵

Several dissociation techniques have been applied to the problem of HS structural elucidation by tandem mass spectrometry.^16-20 However, the complete sequence and modifications assignments for HS oligosaccharides become increasingly challenging for tandem MS as mixture complexity increases. In this context, ion mobility mass spectrometry (IM-MS) offers a gas-phase separation dimension to increase the resolution of highly structurally related HS oligosaccharides.^21-27 Moreover, IM lowers the need for long LC separations prior to MS analysis, reducing analysis time and increasing throughput.^28,29 IM has received limited attention for HS structural analysis,^{21,22,27,30,31} mostly due to the comparatively low resolving power for linear IM, in the range of 50–60 for most commercial instruments. Recently, the resolving power for IM-MS analysis has been improved by the introduction of the SELECT SERIES cyclic IM (cIM), in which a cyclic TWIMS cell is configured orthogonally to the axis of the mass spectrometer,²⁶ allowing the distinctive possibility of performing IMⁿ, reaching a resolving power higher than 900 on oligosaccharides,³² and increasing the possibility that it may solve on-going challenges observed for HS oligosaccharide analysis. As a separation technique, cIM-MS allows users to define the drift path length necessary to resolve structural isomers that are difficult to differentiate using tandem MS.

In this work, we describe a HILIC-ion suppressor (IS)-cIM-MS method to differentiate a mixture of synthetic HS oligosaccharide isomers with permutations on the stereochemistry of hexuronic acid and the number and the connectivity of sulfate units neighboring the central glucosamine residue. The isomers tested included the rare 3-O sulfation that mediates numerous HS–protein binding interactions.³³ We demonstrated the first use of cIM-MS for glycosaminoglycan analysis and the capacity to resolve the high level of isomerism derived from HS oligosaccharide structures.

EXPERIMENTAL SECTION

Materials.

Heparan sulfate synthetic oligosaccharides synthesized as previously described³⁴ were donated by Prof. Geert-Jan Boons at the University of Georgia. The reducing ends of synthetics HS compounds were aminopentoxyl (for compounds 1–18) and pNitroPhenyl (compound Gt20). A detailed description of the synthetic HS library is shown in Figure S1. Ultrapure water was produced using a Milli-Q system (Millipore, Burlington, MA). LC-MS grade acetonitrile was purchased from Fischer (Waltham, MA). All other chemicals were purchased from Sigma-Aldrich (St. Louis, MO).

HILIC-IS-cIM-MS.

HILIC-IS-cIM-MS analyses were performed using an ACQUITY UPLC H-Class chromatograph (Waters Corporation) coupled to a SELECT SERIES cyclic IM-MS (Waters Corporation) interfaced with a Dionex ACRS 500—2 mm ion suppressor system (Thermo Scientific, San Jose, CA). An Accucore 150 Amide HILIC column (2.1 mm × 250 mm, Thermo Fisher Scientific, San Jose, CA) was used with mobile phase A (50 mM ammonium formate, pH 4.4) and mobile phase B 100% acetonitrile. The elution program consisted of a gradient of 45–70% A developed in 45 min with 15 min equilibration. The flow rate was set to 95 μL/min. The ion suppressor regeneration solution consisted in 100 mM sulfuric acid pumped using a 200 μL/min flow rate achieved by gravity and no need of an extra pump. Sample volume injection is 5 μL (2–5 pmol/μL). The cIM-MS instrument was equipped with a tool-free probe electrospray ion source with the capillary voltage set to 2.2 kV in a negative ion mode. Prior to use, the system was mass-calibrated using sodium formate. The cone voltage was set to 20, source offset to 10 V, source temperature to 150 °C, desolvation temperature to 250 °C, and desolvation gas flow to 800 L/h. The cyclic cell device used nitrogen as the drift gas at a pressure of 1.74 mbar, a traveling wave height of 15 V, velocity of 375 m/s, injection time of 5 ms, and separation time of 20 ms. Dissociation was induced postmobility separation. The collision voltage was set to a 24–26 V ramp at the transfer cell. HDMS^E was used for data acquisition.

Infusion Experiments.

For MS method optimization, synthetic HS oligosaccharides at 10 pmol/μL were infused with a syringe pump (Cole-Parmer, IL) at 5 μL/min flow and interfaced with the ion suppressor device, before electrospray ionization (ESI), as described above.

Data Processing and Analysis.

The cIM-MS instrument was controlled using MassLynx 4.2 software. The data were processed using MassLynx 4.2, DriftScope 2.9, and MS^EViewer (all from Waters, Wilmslow, U.K.). Arrival time distributions (ATDs) were smoothed in MassLynx 4.2 using the Savitzky–Golay algorithm (two iterations over three bins). Data interpretation was assisted by GlycoWorkbench³⁵ and GAG-finder.³⁶ Fragments were annotated according to the Domon and Costello³⁷ nomenclature.

RESULTS AND DISCUSSION

Optimization of cIM-MS for Heparan Sulfate Oligosaccharide Analysis.

Among the available IM-MS configurations, the widely used TWIMS instrument utilizes a series of voltage pulses to store and propel ions through the buffer gas. However, these pulses induce vibrational excitation over the ions, leading to the possibility of unwanted fragmentation of labile compounds. Herein, cIM-MS voltage settings were optimized to achieve negligible sulfate dissociation. Based on the full MS spectra shown in Figure S2A-C, we found that the use of the lower static height traveling wave (SHTW) voltage of 15 V stands as an optimal value for labile compounds such as sulfated HS with similar performance to lower SHTW values (data not shown). Taken into account that low charge states result in more abundant sulfate losses and less abundant glycosidic bond and cross-ring product ions in both collision-induced dissociation (CID) and electron-based dissociation (ExD) techniques,^38,39 we optimized the spray ionization voltage to regulate the charge state of HS ions. As shown in Figure S3A-C, the variation of spray voltages affected the relative abundances of the synthetic saccharide charge states observed for infused synthetic HS oligosaccharide Gt20. The use of lower spray voltages increased the relative abundance of higher charge states compared to other charge states. Based on the spectra shown in Figure S3A-C, we concluded that voltages in the range of 2.2–2.4 are optimal for the analysis of HS oligosaccharides. We have also optimized CID collision energy to balance the intensities of glycosidic bond fragmentation (Figure S3D-G). The use of the ion suppressor improves the ionization process of HS oligosaccharides, increases absolute ion abundances, and reduces the level of cation adduction, the complexity of the spectra caused by multiplication of the signal, and the extent of overlap of isotopic distributions (see Figure S4). To illustrate measurement reproducibility, triplicate runs for compound #1 showed a retention time standard deviation of 0.03 min (Figure S5).

HILIC Performance for Related HS Structures.

To evaluate the performance of our HILIC-IS-cIM-MS system, we selected a set of oligosaccharides based on the same monosaccharide sequence substituted with a different number of sulfate groups concentrated in the central repeating unit of the synthetic saccharide. Consistent with our earlier publication,¹² the pure synthetic HS oligosaccharides produce distinct retention times according to the number of sulfate groups using HILIC (Figure 1). We also evaluated the selectivity of our LC system to distinguish HS isomers differing in the stereochemistry of hexuronic acid. As depicted in Figure S6, each pure synthetic HS separately analyzed produced a distinct retention time. We next assessed the HILIC retention times of a pair of pure HS isomers differing in the position of the sulfate unit substituting the iduronic acid residue neighboring the central glucosamine residue. As shown in Figure 2, HILIC produces distinct retention times with overlapping peak profiles for the selected HS isomers. However, this isomer pair displayed a greater degree of separation in the mobility dimension.

Figure 1.

HILIC-MS of pure synthetic HS saccharides substituted with the increasing number of sulfate groups concentrated in the central repeating unit of the oligosaccharide sequence, analyzed by separated LC injections of each compound.

Figure 2.

HILIC-MS of separately LC-injected synthetic HS saccharide sulfate positional HS isomers. (A) Extracted ion chromatogram showing HILIC resolution for HS #12 and HS #16. (B) HILIC-cIM-MS-extracted ion mobiligram showing mobility resolution for HS #12 and HS #16.

HILIC-cIM-MS HS Analysis Using Ion Suppression.

To demonstrate the HS ion behavior in the mobility space, we performed a set of experiments comparing HS analysis between ion suppressor activated and ion suppressor deactivated modes, using the pure synthetic oligosaccharide Gt20 (GlcNS-GlcA-GlcNS-GlcA-GlcNS-GlcA-GlcNS-GlcA-Pnp), by direct infusion. On the one hand, as depicted in Figure 3A,3C, the acquisition in ion-suppressor mode off resulted in a complex pattern due to the presence of sodium and ammonium adductions of HS Gt20. On the other, the same analysis was performed with the ion suppressor activated mode, resulted in a simpler spectrum in which the −6 charge state ion (Figure 3E), corresponding to the deprotonated Gt20, and −5 charge state ion (Figure 3G) were detected as the most abundant ions and cation-adducted species were at the noise level (Figure 3E,G respectively). The full mass spectrum for the analysis of Gt20 comparing the activated and deactivated modes of the ion suppressor is shown in Figure S4. Regarding the ion mobility dimension, as shown in Figure 3B, deactivated ion-suppressor mode analysis of HS Gt20, with its corresponding high level of adduction, resulted in an overlapped arrival time distribution spectrum that hindered the resolution of oligosaccharide isomers in the mobility dimension. However, activated ion-suppressor mode analysis of the HS Gt20 suppressor reduced the intensity of cation-adducted ionic species of Gt20, making the spectral pattern simpler in both the m/z dimension and in the mobility dimension (Figure 3H).

Figure 3.

Comparative cIM-MS analysis of synthetic oligosaccharide HS Gt20 using activated and deactivated modes of an ion suppressor. Selected m/z range for (A) −6 and (C) −5 charge states of deprotonated HS Gt20 without ion suppression. Selected ATD range for (B) −6 and (D) −5 charge states of deprotonated HS Gt20 without ion suppression. Selected m/z range for (E) −6 and (G) −5 charge states of deprotonated HS Gt20 with ion suppression. Selected ATD range for (F) −6 and (H) −5 charge states of deprotonated HS Gt20 without ion suppression. Asterisk refers to the number of ammonium adducts.

Similar results were observed for the −6 charge state of Gt20 at a higher arrival time distribution (A_t) range of the same mobiligram (Figure 3F). These results demonstrated the significant impact of cation adduction that caused overpopulation and overlapping of mobility traces corresponding to the different cation adduced species. This was a notable finding that will help to isolate ATD-resolved species and correctly sequence HS structures from the complex mixtures of HS oligosaccharides in subsequent tandem MS.

Furthermore, spatial conformations adopted by different HS ion charge states may differ regarding the charge localization within the molecule.^40,41 If so, this helps explain the observation that lower-charge-state ions displayed multimodal arrival time distributions (Figure 4A). A similar pattern was observed for the HS saccharide with an additional sulfate group (see Figure S7). We concluded that the use of the ion suppressor enhanced the abundances of higher charge states, in which charge saturation contributed to the unique spatial conformation of HS oligosaccharides that produced unimodal arrival time distributions (ATDs). As an example, the ATD for different deprotonated charge states of the synthetic HS #2 [M – 2H]²⁻ ion displayed drift behavior consistent with the presence of two conformations (Figure 4A). By contrast, the higher charge states presented unimodal distribution in their mobility dimension (Figure 4B,C). Interestingly, this effect was most prevalent for low charge states, for which charge has more freedom to be localized at different ionizable groups within the oligosaccharide structure. We therefore concluded that the lower charge state in multiple ionizable groups leads to a protomerism effect that give rises to multimodal ATDs. Multimodal distribution related to protomerism effects have been previously reported for small molecules under positive ionization mode; however, much work is required to comprehend this effect in larger molecules to reveal how strongly the arrival time distribution spectrum is affected during ESI in compounds bearing multiple ionizable groups as in highly sulfated HS.

Figure 4.

Arrival time distributions corresponding to different charge states of compound #2. (A) [M – 2H]²⁻, (B) [M – 3H]³⁻, and (C) [M – 4H]⁴⁻ at three passes (with 10 deprotonation sites based on carboxyl and sulfate units).

In summary, the use of ion suppression greatly enhances the ionization of highly sulfated HS oligosaccharides that facilitate HILIC-MS analysis. Figure S14 compares HILIC overlay chromatogram traces of compound #18 under ion suppressor mode ON versus ion suppressor mode OFF. Taken into account these findings, we next used the ion suppressor and the precise tuning of spray voltages to demonstrate clearly that deprotonated HS precursor ions produce tandem MS patterns that were easily assigned, due to the higher ion abundances that resulted from the use of ion suppression (Figures S8A,B, S14, and Table S1).

Analysis of Complex HS Oligosaccharide Mixtures.

HILIC is often employed in large glycomics studies because it is compatible with MS, derivatization of glycans is not required, and retention times are highly reproducible.⁴² In the present work, despite the high level of HS isomerism of the oligosaccharides tested, HILIC provided a degree of isomer resolution. To increase the resolution of coeluting HS oligosaccharides, we evaluated cIM arrival time distributions (Figure S9). For the selected HS oligosaccharides, we observed that despite the increase of separation time (20–40 ms) within the mobility cell, synthetic HS isomers #3 and #6 could not be baseline resolved. However, a complete resolution of the isomeric compounds was achieved in a combination of HILIC with cIM. Since compounds HS #3 and #6 were retention time-resolved, 20 ms was used for the analysis of these isomers. HILIC performance for resolving HS #1 and #7 is shown in Figure S13. This demonstrated the principle that the HILIC step provided separation based on saccharide structure that greatly complements the power of the mobility dimension.

Many HS–protein binding interactions depend on the overall HS sulfation, the arrangements of sulfated residues, and the hexuronic acid stereochemistry.⁴³ In most cases, a structural motif consisting of 5–8 mers leads to the efficient binding with protein partners. In this context, we evaluated the performance for HILIC-IS-cIM-MS for the mixtures of HS hexasaccharide isomers, differing in the stereochemistry of the hexuronic acid neighboring the central glucosamine residue. For each of the following analysis, the samples of HS isomers were HILIC-injected as mixtures, postmobility dissociated. The two-dimensional (2-D) plots were extracted based on retention time and ATDs for [M – 3H]³⁻. To identify each of the oligosaccharides within the mixtures, a previous analysis was required to obtain the retention time and the ATD data for individual compounds under the same analysis conditions. We first analyzed a mixture of four isomeric 2-O-sulfated hexamers (Figure 5). We selected [M – 3H]³⁻ ions based on drift behavior that leads to an optimal separation of the ionic species and to lower complexity ATDs (ATD traces corresponding to each of the isomers are shown in Figures S10-S12, respectively, obtained from separate LC injections of each of the individual isomers). As shown in the two-dimensional plot based on HILIC retention time versus ATD, the HILIC-IS-cIM-MS systems resolved the complete HS mixture. This demonstrated a clear resolution of precursor ions that differed by the stereochemistry of hexuronic acid at a single residue within the oligosaccharide. This was a significant finding that will help reduce the burden on the tandem MS step to resolve such isomers.

Figure 5.

Analysis of an isomeric mixture of 2-O-sulfated HS [M – 3H]³⁻.

We next evaluated the ability of the HILIC-IS-cIM-MS system to resolve the mixtures of positional isomers containing the rare and biologically significant 3-O-sulfation and differing in the stereochemistry of the hexuronic acid residue neighboring the central 3-O-sulfated glucosamine unit (Figure 6). Again, HILIC-IS-cIM-MS resolved the isomeric mixture. We expect that this finding will be of particular relevance to the study of 3-O-sulfation from polysaccharide lyase digests of tissue-derived HS since isomeric mixtures will likely be present.

Figure 6.

Analysis of an isomeric mixture of 3-O-sulfated HS [M – 3H]³⁻.

Finally, we extended the analysis for a complex mixture of HS 2,3-O sulfated hexasaccharide isomers. As depicted in Figure 7, using 20 ms of cyclic separation time, we observed complete resolution of HS isomers differing in both the stereochemistry and the sulfate position of the iduronic acid residue that neighbors the central glucosamine unit. These results highlighted the usefulness of HILIC-IS-cIM-MS to differentiate a mixture with a high level of isomerism at the precursor ion level. In principle, this resolution of HS structural isomer mixtures using HILIC-IS-cIM-MS will complement the ability to interpret subsequent tandem mass spectra. The unprocessed 2-D plots of isomeric HS mixtures are shown in Figures S10-S12. The results obtained herein represent the first glycomic approach for HS analysis using high-resolution cyclic ion mobility.

Figure 7.

Analysis of an isomeric mixture of 2,3-O-sulfated HS [M – 3H]³⁻.

CONCLUSIONS

We demonstrated the capability of HILIC-IS-cIM-MS to resolve the mixtures of isomeric HS oligosaccharides. Specifically, the combination of HILIC and ion mobility provides isomeric separation of HS saccharides at the precursor level, which complement to use of collisional dissociation tandem MS for spectrum-centric determination of HS saccharide structures.

We showed that ion suppression was a crucial step for ion mobility-based HS analysis because it minimized the distribution of charge states and cation-adducted species. It is important to notice that cation adduction multiplied the number of precursor ions observed for a single oligosaccharide structure, decreased sensitivity, increased spectral complexity, and, most importantly, precluded the resolution of HS structures in the mobility dimension. This was likely because each ion had a unique spatial conformation and A_t value, resulting in an overpopulated and overlapped arrival time distribution dimension.

The results obtained herein suggested that HILIC-IS-cIM-MS will be useful to complement the ability of subsequent tandem MS for analysis of polysaccharide lyase-digested HS chains in studies aimed at the discovery of new structural motifs for HS binding proteins as well as for the targeted analysis and quality control of commercial HS products. Moreover, the representation of data in 2-D plots might easily find a use for fingerprint comparison during the analysis of alterations in HS structure that occur during disease processes.