Structural Characterization of Carbohydrates by Fourier Transform Tandem Mass Spectrometry

Abstract

Fourier transform tandem mass spectrometry (MS/MS) provides high mass accuracy, high sensitivity, and analytical versatility and has therefore emerged as an indispensable tool for structural elucidation of biomolecules. Glycosylation is one of the most common posttranslational modifications, occurring in ~50% of proteins. However, due to the structural diversity of carbohydrates, arising from non-template driven biosynthesis, achievement of detailed structural insight is highly challenging. This review briefly discusses carbohydrate sample preparation and ionization methods, and highlights recent developments in alternative high-resolution MS/MS strategies, including infrared multiphoton dissociation (IRMPD), electron capture dissociation (ECD), and electron detachment dissociation (EDD), for carbohydrates with a focus on glycans and proteoglycans from mammalian glycoproteins.

INTRODUCTION

Glycosylation is one of the most prevalent posttranslational modifications (PTMs) and a variety of different types of glycans are also frequently attached to lipids. Glycans participate in many important biological processes, including protein folding and assembly, cell adhesion, cell signaling, immune response, molecular trafficking, and cancer metastasis [1–5]. Although the biological significance of glycosylation is well established, the structural diversity arising from complex nontemplate-based biosynthesis, makes glycan structural characterization a highly challenging task. Glycosylation encompasses, e.g., N-linked glycans, O-linked glycans, and the glycosaminoglycan (GAG) family of polysaccharides. As shown in Fig. (1) [6], N-glycans are branched carbohydrates containing a pentasaccharide core, normally attached to asparagine residues of proteins in an Asn-Xxx-Ser (or Thr) amino acid consensus sequence with Xxx ≠ proline. O-glycans are branched and linked with either serine or threonine but without a consensus amino acid sequence. GAGs are also attached to serine or threonine, but they are linear carbohydrates and often highly sulfated [7–8].

Fig. (1).

Representative structures of N-linked glycans, O-linked glycans, and glycosaminoglycans [6].

In order to fully understand the key roles glycans play in various cellular processes, knowledge of only the sequence of these compounds is far from sufficient. The linkage, degree of branching, and stereochemistry should also be determined. Traditionally, exoglycosidase digestion, nuclear magnetic resonance (NMR) spectroscopy, X-ray crystallography, and gas chromatography/mass spectrometry (GC/MS) have all been used for structural characterization of carbohydrates [9]. However, such methods do not provide the required sensitivity for glycoproteomics. Modern mass spectrometry is an important tool for structural analysis of carbohydrates, offering high sensitivity, high precision, and analytical versatility. Accurate masses obtained from high-resolution tandem mass spectrometry (MS/MS or MSⁿ) provide reproducible and reliable information for glycan structural characterization. There are two types of bond cleavages in MS/MS, including glycosidic cleavages (B-, C-, Y-, Z-type) and cross-ring cleavages (A- and X-type) as shown in Fig. (2) [10]. Glycosidic cleavages involve bond rupture between monosaccharides and provide information regarding monosaccharide composition. Cross-ring cleavages, which occur across carbohydrate rings, are particularly helpful in determining linkage type.

Fig. (2).

Nomenclature for tandem mass spectrometric product ions of glycans [10].

This review begins with a brief description of ionization techniques and sample preparation methods, and then highlights recent developments in Fourier transform tandem mass spectrometry for carbohydrate analysis with an emphasis on glycans and proteoglycans from mammalian glycoproteins. MSⁿ in low resolution ion traps, including collision activated dissociation (CAD) [11–13] and negative electron transfer dissociation (NETD) [14–15], and ultraviolet photodissociation (UVPD) in both ion traps and time-of-flight (TOF) instruments [16–18] are also powerful current strategies for oligosaccharide characterization. The resolution and mass accuracy of TOF analyzers are rapidly improving, however, Fourier transform mass analyzers are still superior in performance. Due to our focus on Fourier transform tandem mass spectrometric methods, other techniques are not covered. For more comprehensive reviews of carbohydrate analysis by MS we refer to previous literature [7–9, 19–22].

CARBOHYDRATE IONIZATION METHODS

Matrix-assisted laser desorption/ionization (MALDI) and electrospray ionization (ESI) are the two major modern ionization methods for carbohydrates. MALDI involves co-crystallization of analytes and an organic matrix which acts as a chromophore for laser irradiation. The MALDI mechanism is still under debate, but it has been demonstrated that the choice of matrix plays a crucial role in obtaining strong signals from carbohydrates [19]. MALDI demonstrates high sensitivity, relatively high tolerance to contamination, and predominantly produces singly charged ions, such as (M + H)⁺ and (M + Na)⁺ ions in positive-ion mode. However, acidic (sialylated, sulfated, or phosphorylated) carbohydrates may undergo fragmentation in positive-ion mode during ionization, both in-source and post-source [23–24]. In negative-ion mode, MALDI mostly produces (M - H)⁻ ions.

ESI is a soft ionization method and particularly suitable to use with high-performance liquid chromatography (HPLC). Depending on sample composition, ESI produces both singly and multiply charged ions, which makes ESI more compatible with various MS/MS strategies than MALDI. It is believed that the sensitivity of ESI for carbohydrates is limited by their hydrophilicity [25]: there are three major steps in the formation of gas-phase ions during the ESI process: (a) formation of charged analyte-containing droplets; (b) disruption of charged droplets due to solvent evaporation and Coulomb repulsion; (c) formation of gas-phase ions from secondary droplets [26]. When the initial droplets are formed, hydrophobic analytes (such as most proteins and peptides) are distributed preferentially on the droplet surfaces, whereas hydrophilic analytes (such as oligosaccharides or heavily glycosylated proteins and peptides) are located inside the droplets. Consequently, ion signal from hydrophilic compounds is suppressed in the ESI process, resulting in low sensitivity [25]. Permethylated carbohydrates and carbohydrates derivatized with hydrophobic tags thus demonstrate enhanced sensitivity, up to 1,000 fold [21, 27–28]. Nano electrospray ionization (nano-ESI) shows significantly improved sensitivity and higher tolerance to salts and other contaminants compared to conventional ESI due to decreased initial droplet sizes [25, 29]. Nano-ESI also facilitates glycan structural analysis with extremely low sample consumption and has been widely utilized with nano-scale LC [30–33].

SAMPLE PREPARATION METHODS

a) Enzymatic and Chemical Release Methods

For enzymatic release of N-linked glycans, peptide-N-glycosidase F (PNGase F) is most commonly used. This enzyme cleaves the amide bond between asparagine residues and the glycan. PNGase F releases most N-linked glycans, except for those containing an α1→3 fucose linkage at the reducing end N-acetyl glucosamine (GlcNAc) [34]. In such cases, peptide-N-glycosidase A (PNGase A) can be used as an alternative. Endoglycosidase-H (Endo-H) is another popular enzyme for N-glycan release. This enzyme cleaves the glycan between the two GlcNAcs at the chitobiose core of high-mannose and hybrid glycans [35]. However, Endo-H derived N-glycans do not provide information regarding core fucosylation, because the reducing end GlcNAc remains bound to the protein. There is currently no universal enzyme for releasing O-glycans. Endo-α-N-acetylgalactosaminidase (O-glycanase) cleaves exclusively at serine/threonine-glycan bonds, but it is only active for core-1 type O-glycans [36]. Therefore, chemical release methods are more commonly used for releasing O-glycans.

Hydrazinolysis is a commonly used chemical method [37–39], which has the advantage of releasing both N- and O-linked oligosaccharides, but also frequently introduces side reactions, such as loss of N- and O-acetyl and glycolyl groups under harsh reaction conditions [20]. Reductive β-elimination with sodium hydroxide and borohydride is also a popular method to chemically release O-glycans. When using strong base such as sodium hydroxide, non-selective release of other Ser/Thr PTMs and glycan degradation (peeling) are often observed [40–41]. Therefore, usage of mild bases such as ammonia or dimethylamine [40] has been explored. Recently, a novel O-glycan release method combining complete enzyme degradation with chemical release during the process of solid-phase permethylation was introduced [42]. Glycans attached to serine and threonine are effectively cleaved, but those attached to arginine are not, rendering this strategy highly specific.

b) Derivatization Methods

Permethylation of glycans prior to mass spectrometric analysis is a widely used method. When glycans are permethylated, they become significantly more hydrophobic, which has been reported to increase glycan sensitivity in MS with both ESI and MALDI [19, 43]. Moreover, permethylation enables structural analysis of sialylated glycans in positive-ion mode by stabilizing the labile sialic acid in MALDI [44–45]. Traditionally, permethylation is based on the reaction of glycans with iodomethane and sodium hydroxide prepared in dimethyl sulfoxide (DMSO) [46]. More recently, Novotny, Mechref and co-workers have developed high-throughput solid-phase permethylation of glycans prior to MS [47–48] by utilizing fused-silica capillaries or spin columns packed with sodium hydroxide powders or beads and also demonstrated that sequential double-permethylation can be used for improved structural characterization of sulfated glycans, including localization of sulfate groups [49].

Reductive amination is another widely used derivatization method. 2-aminobenzoic acid (2-AA), 2-aminobenzamide (2-AB) and 2-aminopyridine (2-AP) are some of the commonly used reagents [18–19, 50–52]. Reductive amination serves multiple purposes: first, by introducing a chromophore to the reducing end of glycans, UV or fluorescence detection becomes possible in HPLC. Second, because the introduced chromophore is hydrophobic, the retention in reverse phase HPLC improves at the same time that MS sensitivity increases. However, in order to ensure high yield of labeled carbohydrates, excess of labeling reagent is used, which requires a clean-up step. More detailed information on carbohydrate derivatization can be found in previous reviews [41, 53–54].

In addition to protonated/deprotonated carbohydrates, alkali, alkaline earth, and transition metals are widely used for ionization of glycans prior to MS/MS analysis [55–61]. Metal cations can stabilize labile acidic groups, e.g. sulfate groups [62–63]. It has also been demonstrated that metal-adducted glycans result in more cross-ring cleavages than their protonated counterparts in CAD [55, 58, 61]. Another advantage of metal adduction is elimination of rearrangement reactions during fragmentation. Tandem mass spectrometry of reductively aminated glycans and glycopeptides in their protonated form is known to result in fucose and hexose rearrangements, but this phenomenon is not observed for their sodium-adducted counterparts [52, 64–67]. Therefore, one should be aware of the possibility of rearrangements when assigning MS/MS spectra of protonated glycans and glycoconjugates.

Compared to metal adducts, anionic adducts are not as frequently used in mass spectrometry. Harvey explored a variety of anions in negative-ion mode ESI and reported that nitrate adducts yielded the most satisfactory spectra of N-linked glycans with little in-source fragmentation and high signal abundance [68]. CAD of nitrate- and chloride-adducted N-glycans resulted in abundant A- and C-type product ions, whereas bromide and iodide adducts generated few fragments. Cole and co-workers examined chloride adducts in both negative-ion MALDI and ESI, and found unique structurally-informative product ions [69–71]. For example, the abundance ratio of Cl⁻-adducted/non-Cl⁻-adducted product ions in CAD spectra was utilized to differentiate anomeric configurations of disaccharides [70].

HIGH RESOLUTION MASS ANALYZERS

The tremendous complexity of biological samples continues to push the technical limitations of analytical instrumentation. Fourier transform ion cyclotron resonance (FT-ICR) [72] and orbitrap mass analyzers [73–75] provide the highest resolving power of current mass spectrometers and such high-end instruments are utilized in a number of areas, including biomolecular identification, characterization of PTMs, quantification in proteomics, glycomics, and metabolomics, as well as petroleum characterization. In addition to the superior resolution, FT-ICR and orbitrap instruments also provide ultrahigh mass accuracy [76–77]. Makarov et al. used automatic gain control (AGC) in an LTQ orbitrap mass spectrometer to control the number of ions in each fill over several orders of magnitude of analyte concentration [78]. With this technique, mass accuracy of <5 ppm was achieved with >95% probability at a dynamic range >5000, representing at least an order of magnitude higher than typical values for TOF instruments. Both FT-ICR and orbitrap mass analyzers are able to achieve <5 ppm mass error for externally calibrated spectra and <2 ppm for internally calibrated spectra [76].

Unlike mass analyzers that measure ion deflection (electric/magnetic sectors), stability of ion trajectories (quadrupole mass analyzer, quadrupole ion trap), or time of ion transit (TOF), FT-ICR and orbitrap mass analyzers are based on detection of induced image current in the time domain followed by Fourier transformation to yield a frequency domain spectrum, which is then converted to an m/z spectrum through the ion cyclotron equation (equation (1)) for FT-ICR mass analyzers [72] and through equation (2) for orbitrap mass analyzers [74]:

$${\omega }_c=\frac{\mathit{zeB}}{m}$$ (1)

in which ω_c is the ion cyclotron frequency, z is an integer, e is the elementary charge, B is the magnetic field strength, and m is the ion mass.

$${\omega }_a=\sqrt{\frac{\mathit{zek}}{m}}$$ (2)

in which ω_a is the ion axial oscillation frequency inside the orbitrap and k is the field curvature of the applied electric field. This principle inherently allows high resolution because resolution is proportional to the ion observation time and this time can be rather long (several seconds) because FT-ICR and orbitrap mass analyzers operate at ultrahigh vacuum [76]. A major difference between these two high resolution mass analyzers is that the FT-ICR requires a highly homogenous magnetic field of several Tesla whereas the orbitrap is purely electrostatic.

Sufficiently accurate mass measurements allow unique identification of biomolecular elemental composition. For example, the mass difference between potassium and sodium is 16 Da, which is also the mass difference between hexose (Hex, e.g., mannose or glucose) and deoxyhexose (dHex, e.g., fucose or rhamnose), and between N-glycolyl neuraminic acid (NeuGc) and N-acetyl neuraminic acid (NeuAc). With a conventional ion trap mass analyzer, it is impossible to distinguish, e.g., (Hex + Na − H) (mass 184.03029) and (dHex + K − H) (mass 184.00931) [79], whereas high resolution mass analyzers demonstrate outstanding capability for such applications [80]. In addition, high resolution mass spectrometers combined with isotopic labeling of glycans is a powerful tool for glycan quantification. For example, Orlando and co-workers reported a novel permethylation method using ¹³CH₃I and ¹²CH₂DI for comparative glycomics, introducing a mass difference of 0.002922 Da at each permethylation site.

Gas-phase ion-electron reactions, such as electron capture dissociation (ECD) [81], electron detachment dissociation (EDD) [82], and electron transfer dissociation (ETD) [83] have received increasing attention as alternative MS/MS strategies to CAD and infrared multiphoton dissociation (IRMPD), particularly for structural characterization of PTMs (see discussion below). ECD and EDD are best implemented in FT-ICR mass spectrometers, although ECD is also available in three-dimensional quadrupole ion traps [84–86] and digital ion traps [87–88]. The superiority of the FT-ICR for ECD and EDD is due to the required trapping of both electrons and polycations or anions, respectively, which is challenging in ion traps due to their low m/z cut-off [73]. In addition, ECD and EDD often generate complex MS/MS spectra, including mixtures of radical and even-electron species. Therefore, high resolution mass analysis is essential for unambiguously assigning product ions. ETD was first introduced in a radio-frequency linear quadrupole ion trap instrument [83], however, analogous to ECD and similar to EDD, ETD generates complex product ion spectra that are challenging to interpret without high resolving power and high mass accuracy [89–91]. ETD is commercially available on both FT-ICR and orbitrap instruments.

MS/MS TECHNIQUES IN HIGH RESOLUTION MASS SPECTROMETRY

a) Collision Activated Dissociation (CAD)

CAD is the most widely utilized fragmentation technique for structural analysis of carbohydrates. Low-energy CAD, which involves multiple inelastic collisions with neutral molecules or atoms, typically produces glycosidic cleavages [21]. Cross-ring cleavages are less likely to be formed because higher energy is required to break two bonds. In positive-ion mode of neutral oligosaccharides, B- and Y-type ions are commonly observed, whereas in negative ion mode, cross-ring and abundant C-type fragments are often generated [68, 92–94]. For sialylated oligosaccharides, negative-ion CAD is dramatically different compared to non-sialylated glycans and higher energy is required for dissociation [95]. High-energy CAD provides more cross-ring cleavages, such as ^1,5X- and ^3,5A-type ions in both positive and negative ion mode [50, 96–98] compared to low-energy CAD. In addition to providing extensive structural information regarding glycan sequence and branching, high-energy CAD is capable of differentiating structural isomers [97, 99]. For example, the N-glycan Man7 released from ribonuclease B has three structural isomers. High-energy CAD of Man7 generated several characteristic product ions for each isomer, such as ^0,4A₃ and ^3,5A₃, and the intensity ratios of these species matched with the ratios determined by NMR spectroscopy [97]. However, high-energy CAD is currently mainly available in MALDI tandem time-of-flight (TOF/TOF) instruments, which limits its applicability. Further, high-energy CAD involves significant ion scattering which reduces sensitivity.

In Fourier transform mass spectrometry, low-energy CAD can be performed external to the mass analyzer, either in a linear quadrupole ion trap (for orbitrap and FT-ICR instruments), or in a hexapole collision cell (for FT-ICR). The FT-ICR mass analyzer also allows CAD inside the ICR cell by adding collision gas. In the latter case, two main variants of CAD can be performed: in on-resonance CAD, an excitation frequency equal to the ICR frequency of a precursor ion is applied while collision gas is pulsed into the ICR cell [22]. In this approach, product ions are formed off-axis, which limits the practicality of further fragmentation stages and also limits resolving power [100]. In sustained off-resonance irradiation (SORI)-CAD, the excitation frequency is slightly offset from the precursor ions’ cyclotron frequency. Consequently, ions are repeatedly excited to a relatively small cyclotron radius and then come back to the center of the ICR cell, thereby removing the shortcomings of on-resonance CAD. “In cell” CAD and CAD in an external linear ion trap both have the advantage of allowing multiple stages of tandem mass spectrometry, MSⁿ where n>2.

In SORI-CAD, precursor ions undergo multiple low-energy collisions with added background gas and are continuously vibrationally excited in a manner analogous to IRMPD [101] (see below). Solouki et al. reported that SORI-CAD MSⁿ (n>2) of permethylated oligosaccharides provided more extensive fragmentation compared to conventional CAD in a triple quadrupole mass spectrometer [102]. SORI-CAD has also been applied to protonated carrageenan sulfated oligosaccharides for determining the positions of sulfated residues [103], to deprotonated monosaccharides for determining phosphate locations [104], and to alkali metal-adducted disaccharides for studying the influence of metal ions on fragmentation behavior [105]. The latter work showed that, in SORI-CAD, the dissociation thresholds of oligosaccharide glycosidic cleavages depend on the size of alkali metals, while activation barriers for cross-ring cleavages are independent of alkali metal identity, but dependent on linkage type [105].

With the orbitrap mass analyzer, in addition to CAD in the linear ion trap (ion trap-type CAD, which favors low energy fragmentation pathways [106–107]), CAD can be accomplished within the C-trap used to focus ions prior to injection into the orbitrap. This mode of CAD, termed high-energy C-trap dissociation (HCD) [108] yields similar MS/MS spectra to “beam-type” CAD in an external multi-pole. For example, for peptides, immonium ions are observed in addition to b- and y-type ions from backbone cleavage. Such ions, which cannot be observed in ion trap-type CAD due to the low m/z cut-off of such devices, can aid the identification of modified amino acids. The absence of a low m/z cut-off in HCD also allows quantification of, e.g., phosphopeptides with iTRAQ labels [109–110]. Such labels have low molecular weight and are therefore difficult to observe in ion trap-type CAD. HCD in orbitrap mass spectrometers thus shows great potential for both qualitative and quantitative analysis with excellent mass accuracy, resolution, and sensitivity. The absence of a low m/z cut-off is also beneficial in carbohydrate analysis because it allows detection of monosaccharide “marker” ions.

b) Infrared Multiphoton Dissociation (IRMPD)

IRMPD is a valuable tool in FT-ICR and other ion trap instruments because of its ability to readily yield secondary fragmentation and thereby provide higher fragmentation efficiency compared to CAD [22]. IRMPD has been applied to various kinds of biomolecules, such as proteins, peptides, carbohydrates, and oligonucleotides [111]. Typically, a 10.6 μm CO₂ laser is used for infrared irradiation, but tunable CO₂ laser IRMPD was recently implemented [112–114]. The latter approach was able to differentiate anomers of lithium-adducted methyl-glucopyranoside by comparing relative product ion abundances as function of laser wavelength [113]. Differentiation of lithium-tagged glucose-containing disaccharides was previously reported by Polfer et al. with 7–11 μm photons from a free electron laser [115]. That work was able to assign both linkage position and sugar anomeric configuration.

During IR activation, multiple photons are absorbed and the corresponding energy is redistributed over all precursor ion vibrational modes [113]. Therefore, multiple fragmentation events may take place. One of the major advantages of IRMPD in FT-ICR MS is that no collision gas needs to be introduced to the ICR cell [116]. In addition, IRMPD provides on-axis fragmentation, thus there is no loss of resolution.

Lebrilla and co-workers have compared IRMPD fragmentation patterns to those from CAD of model oligosaccharides, N-linked glycans, and O-linked glycans ionized with alkali metals in FT-ICR MS [24, 59, 117–118]. Similar to CAD, IRMPD favors low energy dissociation pathways, thus mostly glycosidic cleavages but also some cross-ring cleavages are observed in IRMPD spectra. However, contrary to CAD, IRMPD fragmentation efficiency increases with increasing glycan size: for large O-glycans, single-stage IRMPD resulted in cleavage of all glycosidic bonds [117]. Moreover, IRMPD of alkali metal-adducted oligosaccharides generates different fragmentation patterns compared to those from SORI-CAD, e.g., IRMPD readily yields monosaccharide residues, whereas multiple CAD MS/MS stages are required to achieve similar results [24, 59]. Therefore, IRMPD is not only a fast and efficient MS/MS method but it can also potentially serve as a complementary technique to CAD for structural characterization of glycans.

c) Electron Capture Dissociation (ECD)

ECD is performed by irradiating at least doubly positively charged precursor ions with low energy (<1 eV) electrons, generating charge-reduced radical species from electron capture and product ions from radical-driven fragmentation [81, 119–121]. Since its first introduction in 1998 [81], the application of ECD for biomolecular structural analysis has been rapidly expanding due to its complementary nature compared to CAD and IRMPD [122]. For example, ECD shows unique analytical utility in PTM characterization because labile PTMs are retained and can thus be localized within a molecule. ECD has been successfully utilized in this manner for characterizing, e.g., protein N-glycosylation [123–124], O-glycosylation [125], and phosphorylation [126–127].

While ECD has been widely applied to peptides and proteins, its application towards carbohydrates has just begun to emerge. Due to the lack of basic groups in carbohydrates and their frequent relatively small size, it is often difficult to obtain multiply protonated precursor ions (multiple positive charges are required for ECD). The first application of ECD towards carbohydrate analysis circumvented this issue by involving aminoglycans, which can easily be multiply protonated. However, this work reported only glycosidic cleavages from ECD [128]. Our group utilized alkali, alkaline earth, and transition metals to generate multiply positively charged ions for model oligosaccharides and observed complementary fragmentation patterns from IRMPD and ECD of these species, as shown in Fig. (3) [60]. Here, IRMPD (Fig. (3a)) is shown to yield extensive glycosidic cleavages, as well as several less abundant cross-ring fragments. Following ECD (Fig. (3b)), fewer product ions are observed compared to IRMPD, however, cross-ring cleavage is the dominant fragmentation pathway, providing valuable information regarding linkage type. We also examined ECD and IRMPD of metal-adducted N-linked glycans released from a glycoprotein and found that ECD generates unique fragments, complementary to those from IRMPD [129]. ECD of different metal-adducted species showed different fragmentation patterns, but sulfate groups were retained in all ECD spectra, providing valuable sulfate location information. O’Connor and co-workers explored the application of CAD and hot ECD (i.e., ECD with ~10 eV electrons) towards permethylated oligosaccharides [130]. For linear oligosaccharides, CAD and hot ECD resulted in similar product ions. In contrast, CAD generated A-, B-, and Y-type ions for branched N-glycans, while hot ECD generated C-, Z- and complementary A- and X-type cross-ring fragment pairs. All these studies show that ECD has great potential to serve as a glycan structural analysis technique.

Fig. (3).

IRMPD (a) and ECD (b) spectra of Co²⁺-adducted N-linked glycan in FT-ICR MS. Complementary product ions are observed in IRMPD and ECD spectra [60]. Reprinted with permission from [60]. Copyright (2007) American Chemical Society.

d) Electron Detachment Dissociation (EDD)

EDD was first introduced for the characterization of peptide anions in 2001 [82]. In EDD, polyanions are irradiated with >10 eV electrons to yield electron detachment, forming radical anions which undergo further radical-driven fragmentation. EDD has been applied to peptides [82, 131–132], oligonucleotides [133–134], gangliosides [135], model oligosaccharides [136], and GAG-derived oligosaccharides [137–143]. Our work demonstrated that EDD of neutral and sialylated oligosaccharides results in fragments complementary to those from IRMPD and CAD, including extensive cross-ring product ions such as ^1,5A, ^3,5A, ^1,5X and ^3,5X ions [136]. When applying EDD to sialylated N-glycans released from glycoproteins, EDD preferentially yielded X-, Y- and Z-type ions while IRMPD resulted in mostly A-, B- and C-type ions. Cross-ring cleavages generated by EDD were completely complementary to IRMPD in each case studied [144]. The extensive work by Amster and co-workers on EDD of GAGs has shown that extensive glycosidic and cross-ring cleavages are generated, as exemplified in Fig. (4). In contrast, IRMPD and CAD of GAGs mainly generate glycosidic cleavages. EDD was also shown to be able to differentiate the isomers glucuronic acid and iduronic acid [138]. The complementarity of EDD and IRMPD/CAD makes them highly promising approaches for structural characterization of carbohydrates, particularly acidic carbohydrates in negative-ion mode.

Fig. (4).

Tandem mass spectra of [M - 2H]²⁻ precursor ions of a tetrasaccharide, obtained by (a) EDD, (b) IRMPD, and (c) CAD [137]. EDD of GAGs generates extensive glycosidic and cross-ring cleavages, including some unique cross-ring fragments compared to IRMPD and CAD. Reprinted with permission from [137]. Copyright (2007) Springer.

e) Electron Induced Dissociation (EID)

In electron induced dissociation (EID), precursor ions are irradiated with electrons of energy exceeding the ionization energy of the precursor neutral molecules, but below their second ionization energy. EID generally results in extensive fragmentation [128, 143] and has been utilized to characterize aminoglycans [128], GAG tetrasaccharides [143], phosphorylated metabolites [145], and fluorescently labeled oligosaccharides [146]. EID of GAG tetrasaccharides yielded EDD-like fragmentation, including cleavages of all glycosidic bonds, and abundant cross-ring product ions, highly different from the product ions resulting from IRMPD and CAD [143]. EID of singly protonated oligosaccharides provided complementary structural information compared to SORI-CAD, including B-, C-, Y-, Z-, and ^1,5X-type ions [146]. These results show great potential of EID as an alternative strategy to characterize carbohydrates. Particularly, EID is compatible with singly charged analytes.

CONCLUSIONS

In summary, tandem mass spectrometry is a highly valuable tool for structural determination of carbohydrates. The high resolution and high mass accuracy of Fourier transform instruments enable detailed carbohydrate structural analysis even from low abundance species. High mass accuracy allows unique identification of carbohydrate elemental composition and high resolution/high resolving power is essential for achieving accurate interpretation of MS/MS spectra. Permethylation, reductive amination, and metal adduction of carbohydrates serve to increase ionization responses and often aid in providing rich structural information following MS/MS fragmentation. Alternative high resolution MS/MS methods, such as IRMPD, ECD, and EDD, generate highly informative fragments, complementary to those from CAD. Thus, high resolution tandem mass spectrometry is continuing to evolve for improved carbohydrate structural analysis.

Table 1.

Tandem Mass Spectrometric Techniques for Glycans

MS/MS Techniques	Vibrational Activation				Ion-Electron Reactions
	CAD	IRMPD	SORI-CAD	HCD	ECD	EDD	EID
Activation energy source	Slow heating by inelastic collisions between reactant ions and neutral gas molecules	Absorption of multiple infra-red photons	Collisions between neutral gas and ions in cyclotron motion undergoing alternating radial acceleration and deceleration as a result of an off-resonance rf pulse	Same as CAD	Polycations capture low energy electrons (<1 eV)	Polyanions irradiated with >10 eV electrons to yield electron detachment	Cat- or anions are irradiated with > 10 eV electrons to cause excitation but not ionization
Precursor ion charge states	Any charge state	Any charge state	Any charge state	Any charge state	≥ 2+	≥ 2−	Any charge state
Glycan fragmentation features	Tend to generate glycosidic cleavages but 0,2A and 2,4A-type ions are also commonly observed.				Generate complementary product ions compared to vibrational activation techniques and often yield more cross-ring cleavages
Available mass analyzers	Most types	FT-ICR, ion trap	FT-ICR	Orbitrap	FT-ICR, ion trap	FT-ICR, ion trap	FT-ICR

ABBREVIATIONS

2-AA

2-aminobenzoic acid

2-AB

2-aminobenzamide

2-AP

2-aminopyridine

AGC

Automatic gain control

CAD

Collision activated dissociation

DMSO

Dimethyl sulfoxide

ECD

Electron capture dissociation

EDD

Electron detachment dissociation

EID

Electron induced dissociation

Endo-H

Endoglycosidase-H

ESI

Electrospray ionization

ETD

Electron transfer dissociation

FTICR-MS

Fourier transform ion cyclotron resonance mass spectrometry

GAG

Glycosaminoglycan

GC

Gas chromatography

GlcNAc

N-acetyl glucosamine

HCD

High-energy C-trap dissociation

HPLC

High-performance liquid chromatography

MALDI

Matrix-assisted laser desorption/ionization

MS

Mass spectrometry

MS/MS or MSⁿ

Tandem mass spectrometry

Nano ESI

Nano electrospray ionization

NETD

Negative electron transfer dissociation

NMR

Nuclear magnetic resonance

PNGase A

Peptide-N-glycosidase A

PNGase F

Peptide-N-glycosidase F

PTM

Posttranslational modification

SORI-CAD

Sustained off-resonance irradiation collision activated dissociation

TOF

Time of flight

UVPD

Ultraviolet photodissociation