A New Strategy Coupling Ion-Mobility-Selective CID and Cryogenic IR Spectroscopy to Identify Glycan Anomers.

Determining the primary structure of glycans remains challenging due to their isomeric complexity. While high-resolution ion mobility spectrometry (IMS) has recently allowed distinguishing between many glycan isomers, the arrival-time distributions (ATDs) frequently exhibit multiple peaks, which can arise from positional isomers, reducing-end anomers, or different conformations. Here, we present the combination of ultrahigh-resolution ion mobility, collision-induced dissociation (CID), and cryogenic infrared (IR) spectroscopy as a systematic method to identify reducing-end anomers of glycans. Previous studies have suggested that high-resolution ion mobility of sodiated glycans is able to separate the two reducing-end anomers. In this case, Y-fragments generated from mobility-separated precursor species should also contain a single anomer at their reducing end. We confirm that this is the case by comparing the IR spectra of selected Y-fragments to those of anomerically pure mono- and disaccharides, allowing the assignment of the mobility-separated precursor and its IR spectrum to a single reducing-end anomer. The anomerically pure precursor glycans can henceforth be rapidly identified on the basis of their IR spectrum alone, allowing them to be distinguished from other isomeric forms.

Introduction

Glycans play a central role in many biological processes, including protein folding, cellular signaling, and recognition of pathogens by the immune system.[1−3] The SARS-CoV-2 spike protein, for example, has a number of N-linked and O-linked glycans that both contribute to its folding as well as help shield it from immune detection.[4−9] Despite the importance of glycan structure, the intrinsic isomeric complexity of glycans makes primary sequence determination challenging. Many of the monosaccharide building blocks are isomers, making them indistinguishable by their mass. Moreover, they can link together at different positions, resulting in regioisomers and branched structures. Finally, the glycosidic linkage that forms at the anomeric carbon can exist either in an α- or β-configuration, and glycans with a free reducing-end OH can undergo mutarotation at the anomeric carbon in solution.

Techniques for glycan analysis, although rapidly developing, do not yet provide a comprehensive solution for deciphering their structure. Tandem mass spectrometry (MS) is currently the most widely used approach for glycan sequencing and analysis due to its speed and sensitivity.[10−12] This technique has recently unraveled many structural details such as linkage position, branching, and in some cases the anomericity of small glycans.[13−16]

Combining other analytical techniques with MS greatly expands its capability to analyze glycan structure. For example, coupling MS with liquid chromatography gives rise to a powerful analytical tool, which when combined with enzymatic digestion can be used to fully sequence even large oligosaccharides.[17−19] Although powerful, these techniques often require chemical modification, long incubation times, and multiple chromatographic runs, making them complex and time-consuming.

Ion mobility spectrometry (IMS), a technique in which gas-phase ions are separated based on their collisional cross section with an inert buffer gas, has been combined with MS to provide a rapid and effective tool for glycan analysis.[20−22] Such studies in which fragments generated by collision-induced dissociation (CID) are characterized by IMS have led to the identification of regioisomers and anomers of the precursor glycans.[23,24] Ultrahigh-resolution IMS has been able to separate subtle substructures in small glycans,[25−30] and a cyclic IMS instrument has recently been used to separate and partially identify the isomers of a set of pentasaccharides based on their fragmentation pattern.[28,29] However, the ever-increasing resolution of IMS devices leads to the observation of multiple peaks in the arrival-time distribution (ATD). This complicates the interpretation of IMS data, as the separated reducing-end anomers, which are in equilibrium in solution, can then be confused with other structures such as positional isomers, regioisomers, or conformers. For this reason, IMS–MS studies alone have difficulties identifying all the mobility-separated isomers unambiguously.

Infrared (IR) spectroscopy, on the other hand, is a technique that is sensitive to the most minute structural details of a molecule. While infrared multiple-photon dissociation (IRMPD) at ambient temperature has been used to identify the epimers and anomers of small glycans and their fragments,[23,31,32] cryogenic IR spectroscopy has enabled the fingerprinting of glycans of increasing size and isomeric complexity,[33−38] where the low temperatures of the analyte ions greatly enhance spectroscopic resolution by eliminating thermal inhomogeneous broadening.[39] Combining such techniques with ultrahigh-resolution IMS can be used to identify minor structural differences including positional isomers, even in glycans containing up to 10 monosaccharides.[36−38,40,41] Nevertheless, the vast number of possible structures makes it challenging to interpret complex vibrational spectra of larger oligosaccharides and determine their specific isomeric forms. Moreover, the size of most glycans puts them out of reach of current theoretical methods that are sometimes used to associate vibrational bands to a certain structure.[42,43] This problem can be overcome by using a database of spectral fingerprints constructed using glycan standards,[36,40] although analytical standards of pure isomeric forms are often expensive to produce or simply not available.

Breaking down glycans into smaller fragments by CID prior to spectroscopic investigation can simplify the interpretation of the spectra. This approach relies on breaking the molecule of interest into C- or Y-type fragments,[44] which are themselves intact glycans, and then matching their IR spectra to a database of known standards. This technique has recently revealed that for alkali-metal complexed C-type fragments, the anomericity of the glycosidic linkage is retained.[23,31,45,46] However, many glycosidic linkages preferentially produce B- or Y-type fragments, and this is a limiting factor to using only C-type fragments to sequence glycan structure. We have previously shown how Y-fragments can help identify glycan positional isomers.[41] In this work, we explore how Y-fragments, which carry the anomeric OH, can be used to distinguish and assign the reducing-end anomers from a set of mobility-separated oligosaccharides, thus allowing them to be distinguished from other isomeric forms in a sample. We propose a systematic approach, combining mobility-selective CID and cryogenic IR spectroscopy, which can identify the anomers of oligosaccharides using only a small set of anomerically pure and commercially available mono- and disaccharide standards. The spectra of the precursor molecules can then be assigned to a single anomer and added to a database, providing a spectral library of complex molecules for which anomerically pure standards are not available.

Experimental Approach

A newly developed instrument that combines ultrahigh-resolution ion mobility with CID and cryogenic infrared (IR) spectroscopy was used to perform the experiments presented in this work, and its details and characterization have been described recently.[47] Briefly, ions are generated by nanoelectrospray ionization (nESI) and guided by ion funnels into an ion mobility module, which uses structures for lossless ion manipulations (SLIM), implemented by a pair of mirror-image printed circuit boards (PCBs) as originally developed by Smith and co-workers.[25,26,48,49] Ions are accumulated in the first part of the SLIM device and then released as pulses onto a serpentine mobility path. They are driven through the 10 m long separation region by a traveling wave (TW) and separated based on their collisional cross section (CCS) with nitrogen, the pressure of which is maintained at 2.2 mbar. This IMS device features a resolving power of approximately 200 after a single separation cycle and reaches a resolution of 1000 after approximately 20 separation cycles.[47] Mobility-separated ions leaving the IMS region are guided through differentially pumped stages to a cryogenic, planar, multipole trap maintained at 45 K and a pressure of approximately 10–6 mbar. Accelerating the ions into the trap by increasing the difference in DC bias potential between the ion guides and the trap results in fragmentation by CID, and the fragments as well as precursor ions are detected by a reflectron time-of-flight mass spectrometer (ToF-MS).

The different mobility-separated precursors as well as their respective fragments can be identified by IR messenger tagging spectroscopy, which is performed in the ion trap. During a trapping cycle, the ions are collisionally cooled by an 80/20 He/N2 gas mixture and tagged by weakly bound nitrogen molecules. A continuous-wave fiber-pumped IR laser (IPG, USA) operated at an output power of either 0.5 or 1 W irradiates the contents of the trap, and when the photon energy is in resonance with a vibrational transition in the molecular ion, the photon is absorbed and results in the loss of the nitrogen tag, which is detected as a decrease in the tagged ion signal and an increase in the untagged ion signal. A cryogenic infrared spectrum is obtained by plotting the ratio of the tagged ion signal to the sum of the tagged and untagged ion signal as a function of laser wavenumber.

Isotopic Labeling

The reducing-end OH of maltopentaose was isotopically labeled with 18O by dissolving it in H218O for 1 week. This allowed the distinction of otherwise isomeric reducing and non-reducing-end fragments, which were investigated using a commercial Waters QToF Premier at 40 eV collision energy.

Materials

The oligosaccharides galacto-N-biose (Galβ1–3GalNAc) and maltopentaose were purchased from Carbosynth (UK). The tetrasaccharide lacto-N-neotetraose (LNnT) was purchased from Dextra (UK). The disaccharides methyl-α-lactoside, methyl-β-lactoside, methyl-α-maltosidase, and methyl-β-maltosidase were purchased from Synthose (Canada). Oxygen-18-labeled water, 97 atom % 18O was purchased from abcr, Germany. All samples were used without further purification and diluted in a 50:50 water–methanol solution to a concentration of 10–50 μM. Only the singly sodiated cations were investigated in this work. All gases were 99.9999% purity.

Results and Discussion

Identifying IMS-Separated Species of a Disaccharide

The singly sodiated disaccharide galacto-N-biose (Galβ1–3GalNAc), a core structure of mucin-type O-glycans,[50] showed only one major feature in its ATD, even after five cycles (∼50 m drift path) through the SLIM IMS device (Figure a). Two low-abundance structures appearing at 533 and 539 ms were also separated, but their IR spectra (Figure S1) show no resemblance to that of the main peak, suggesting that these are conformers that differ significantly in structure, possibly due to an open-ring form of the reducing-end GalNAc. However, further study of these species by fragmentation was not attempted due to their low abundance.

Figure 1

(a) Arrival-time distribution of the sodiated disaccharide Gal(β1–3)GalNAc after five separation cycles. (b) Comparison of the IR spectrum of the Y1-fragment generated from the major mobility peak to that of the pure α- and β-anomers of GalNAc, produced by C1 fragmentation from the disaccharides GalNAc(α-3)Gal and GalNAc(β-3)Gal. In this and subsequent figures, the spectrum of the α-anomer is in red, and that of the β-anomer is in blue.

The Y1-fragment from the major mobility peak was generated, and its vibrational spectrum was compared to those of anomerically pure standards (Figure b). The anomerically pure monosaccharides α-GalNAc and β-GalNAc were produced by C1 fragmentation of the disaccharides GalNAc-Gal and GalNAc(β-3)Gal, respectively, in which the anomericity of the glycosidic bond is retained.[46] The spectrum of the Y1-fragment matches that of the anomerically pure α-GalNAc, with all the bands appearing in the same positions with the same relative intensities. This identifies the major peak in the ATD of Figure a to be the α-reducing-end anomer of the precursor disaccharide Galβ1–3GalNAc. Given that mutarotation does not occur in the gas phase,[51] this suggests that the precursor disaccharide exists primarily in the α-configuration under the solution conditions studied. The IR fingerprint spectrum of the disaccharide was measured (Figure S1) and can be used to identify it in subsequent investigations.

Tetrasaccharide LNnT

In contrast to the example above, the human milk tetrasaccharide LNnT separates into two well-resolved peaks after two cycles on our ion mobility device (Figure a). Such bimodal distributions are commonly observed in ion mobility studies of reducing sugars, and it has been speculated, and in some cases confirmed, that they arise from a mixture of α- and β-reducing-end anomers.[27,28,37,51] To determine if the separated peaks in the ATD of Figure a indeed represent the reducing-end anomers and if so to assign them, we produce the Y2-fragments (corresponding to lactose) from the mobility-separated precursor ion and compare their spectra to those of anomerically pure methyl-α-lactoside and methyl-β-lactoside (Figure b), which are locked into a single anomer at the reducing end by O-methylation.[31,51]

Figure 2

(a) Mobility separation of the precursor [LNnT + Na]+ after two separation cycles (21 m drift path). (b) IR spectra of the Y2-fragment (lactose) generated from the first and second mobility-separated precursors (gray) compared to those of methyl-β-lactoside (blue) and methyl-α-lactoside (red).

The IR spectra of the Y2-fragments generated from each mobility peak are clearly distinguishable from each other, and each resembles one of the O-methylated standards, which allows us to identify the reducing anomers. The absence of the anomeric OH in the standards accounts for the slightly different absorption patterns observed in the 3630 to 3670 cm–1 region compared to the Y2-fragments. Moreover, the slight shifts in the band positions can be attributed to differences in the hydrogen bonding network resulting from methylating the OH at the anomeric carbon. Nevertheless, the comparison between the spectra of the Y2-fragments and the individual anomers of O-methylated lactose is compelling and allows us to assign peaks 1 and 2 in the ATD of LNnT (Figure a) to the β- and α-anomers, respectively. We can then add the anomer-specific IR spectra of both lactose and LNnT (Figure S2) to a spectroscopic database for analyzing human milk oligosaccharides.

Maltopentaose

To investigate how this procedure can be used to analyze a reducing sugar that features a more complex ATD, we investigated the different drift peaks of maltopentaose, which is commonly used in the preparation of dextran ladders for capillary electrophoreses.[52] The ATD obtained after three IMS cycles is displayed in Figure a and contains four well-separated peaks. Similar results were observed in previous high-resolution IMS studies, and it has been speculated that the two major peaks originate from the reducing-end anomers, while the other structures may represent open-ring configurations.[29,53]

Figure 3

(a) Arrival-time distribution of sodiated maltopentaose, which separates into four species after three IM separation cycles (31 m drift path). (b) Comparison of the cryogenic vibrational spectra of the Y2-fragments from the second and third mobility peaks with those of methyl-α-maltosidase and methyl-β-maltosidase.

To investigate the identity of the separated species, we produced the Y2-fragment (corresponding to maltose) from each peak in the ATD and compared the spectra of these fragments to those of the anomerically pure methyl-α-maltosidase and methyl-β-maltosidase. Since the Y2-fragment is isomeric to the C2-fragment in this molecule, a separate experiment was performed on a Waters QToF Premier, in which we labeled the reducing end with 18O in order to distinguish between these two fragment types. The tandem mass spectra (Figure S3) showed that the Y2-fragment is preferentially formed over the C2-fragment. Assuming similar dissociation propensities during the CID process on our spectroscopy instrument, we expect to observe predominantly the Y2-fragment species there as well. Upon comparing the spectra of the fragments from the two major drift peaks to the anomerically pure methylated standards (Figure b), we find that a transition at 3655 cm–1 is missing in the spectra of both O-methylated standards, which can be assigned to the anomeric OH stretch vibration. Other small differences between the spectra of the Y2-fragments and the O-methylated standards, such as the shape of the low-frequency band around 3510 cm–1, can be attributed to a change in the hydrogen-bonding network upon removal of one OH oscillator. Despite these slight differences, we can match the fragment from peak 2 to the methylated α-anomer and the fragment from peak 3 to that of the methylated β-anomer. This further illustrates the sensitivity of IR spectroscopy to the slightest change in molecular structure.

We can thus assign the two major peaks in the ATD of maltopentaose to the reducing-end anomers. The spectra of the Y2-fragments from peaks 1 and 4 resemble each other (Figure S4); however, they do not show any resemblance to the O-methylated standards and must therefore correspond to some other conformations at the reducing end, such as an open-ring form, as proposed previously,[53] or a furanose ring.[54]

Conclusions

In this work, we have demonstrated how the combination of ion-mobility-selective CID with cryogenic IR spectroscopy can be used to identify the reducing-end anomers of biologically relevant oligosaccharides. By exploiting the fact that Y-fragments retain the anomeric information on their precursor structures, we are able to identify the mobility-separated anomers of larger glycans by spectroscopically matching their fragments to anomerically pure, commercially available mono- and disaccharides or those selectively formed by C-fragments that retain the anomericity of the glycosidic bond.

This method also allows us to obtain the anomerically pure IR fingerprints of large molecules for which methylated standards are often not obtainable. These fingerprints can in turn be used as standards to assign structures of still larger, more complex oligosaccharides as we have demonstrated in a recent publication.[41] Ultimately, a sufficiently large database can be created, allowing for rapid structural identification of large, unknown oligosaccharides.

More broadly, the general approach employed here can be used to identify the other types of CID fragments, which will help in piecing together the primary structure of isomerically complex glycans.