Student Maxim Treep
Examiner 1 Prof. Dr. G.W. Somsen Examiner 2 Dr. R. Haselberg Course Literature study
Date 11/08/2020
LC-MS & IM-MS in the analysis
of
MSc Chemistry
Analytical Sciences
Literature Thesis
LC-MS & IM-MS
in the analysis of released
N-glycans
by
Maxim Treep
12328936
June 2020
12 EC
April 2020 – August 2020
Supervisor/Examiner:
Examiner:
Prof. dr. G.W. Somsen
Dr. R. Haselberg
BioAnalytical Chemistry / Vrij Universiteit
Amsterdam
Abstract
Glycans are critical modulators of protein function, which includes processes such as regulation of cell-cell recognition, cell adhesion, immune response, fertilization, trafficking, and intra- and extracellular signaling events. With such a breadth in influence on protein function, reports increasingly showcase the considerable role of glycoforms in both health and disease. An important aspect in understanding the function of glycosylation is the characterization of glycoproteins and the study of glycoforms through the use of analytical methods. In this thesis the focus is on “released glycan analysis”, specifically N-glycans. The released N-glycan approach is mainly utilized to elucidate/quantitate glycan structures, however, may have potential in biomarker studies as well. Accordingly, the aim of this literature study is to explore the vast field of glycomics, through a focused view on the analysis of released N-glycans. The performance of emerging ion mobility mass spectrometry (IM-MS) instruments is evaluated in comparison to the well-established liquid chromatography mass spectrometry (LC-MS) approach. The diversity and concurrently the similarity of N-glycans has demanded the use of state-of-the-art equipment and oftentimes requires additional alteration of instruments to facilitate distinction. While the well-established LC-MS methods frequently demonstrate capability to separate an abundance of diverse glycan structures, the main challenge remains resolving isomeric structures. In this predicament, IM-MS instruments have demonstrated their potential in establishing good resolution between stereoisomers in a high-throughput manner, thereupon simplifying MS/MS data and reducing spurious glycan characterization. Adversely, standard IMS devices suffer from an inherent lack of resolving power, constraining IM-MS approaches to the analysis of low complexity glycan mixtures. Recently, a variety of applications have reported a significant increase in resolving power through novel instrument design and sophisticated parameter optimization. Nevertheless, IM-MS in glycomics is still in its infancy, meaning that these approaches are currently fixated on the development of IM-MS in technological aspect rather than glycan discovery and profiling.Once IM-MS is further matured in the analysis of N-glycans, it can be employed for large-case cohorts in search of biomarkers. Until then, the current standard IMS instruments remain employed in structure elucidative studies, although the coalescence of LC, IMS, and MS/MS bears great merit for the future of released N-glycan analysis.
Abbreviations
2-AB 2-aminobenzamide
2D-LC Two-dimensional liquid chromatography
AcN Acetonitrile
AP 2-aminopyridine
Asn Asparagine
CCS Collision cross-section
CID Collision induced dissociation
CV Compensation voltage
DTIMS Drift tube ion mobility spectrometry
ECD Electron capture dissociation
EED Electronic excitation dissociation
ESI Electrospray ionization
ETD Electron transfer dissociation
FAIMS Field asymmetric ion mobility spectrometry
FT-ICR Fourier-transform ion cyclotron resonance
GalNAc N-acetylgalactosamine
GlcNAc N-acetylglucosamine
GRIL glycan reductive isotope labelling
GU Glucose units
HILIC Hydrophilic interaction liquid chromatography
IEC Ion exchange chromatography
IM2 Tandem ion mobility
IMS Ion mobility spectrometry
IP Ion pair
IRMPD Infrared multiphoton dissociation
LC Liquid chromatography
LNH Lacto-N-hexaose
LNnH Lacto-N-neohexaose
LNT Lacto-N-tetraose
LNnT Lacto-N-neotetraose
MALDI Matrix-assisted laser desorption/ionization
MS Mass spectrometry
PCA Principle component analysis
PCMF Post-column make-up flow
PDAC Pancreatic ductal adenocarcinoma (PDAC)
PGC Porous graphitic carbon
PNGase F/A Peptide-N-glycosidase F/A
QqQ Triple quadrupole mass spectrometer
RPLC Reversed-phase liquid chromatography
RT Retention time
SA Selected accumulation
SLIM structures for lossless ion manipulations
SUPER Serpentine ultralong path with extended routing
TFA Trifluoroacetic acid
TIMS Trapped ion mobility spectrometry
(Q)TOF (Quadrupole) Time-of-flight
TWIMS Travelling wave ion mobility spectrometry
UV Ultraviolet
Table of contents
1 Introduction ... 1
2 Theoretical background ... 3
2.1 Glycobiology – N-glycan structure ... 3
2.2 Sample preparation ... 4 2.2.1 Glycan release ... 4 2.2.2 Glycan derivatization ... 4 2.3 Separation methods ... 5 2.3.1 Liquid chromatography ... 5 2.3.2 Ion mobility ... 7 2.4 Glycan ionization... 11 2.5 Glycan dissociation ... 11 2.6 Mass analyzers ... 13 2.6.1 Low-resolution ... 13 2.6.2 High-resolution ... 13 3 Glycan profiling ... 15
3.1 Liquid chromatography-mass spectrometry ... 15
3.2 Ion mobility mass spectrometry... 21
4 The utility of IMS in released N-glycan analysis ... 32
5 Concluding remarks and future outlook ... 34
1
1 Introduction
The variety of proteins is not solely determined by a high number of distinct genes (~20,300), instead their diversity is primarily governed by intricate alterations, comprised of allelic DNA variations, differential splicing of RNA transcripts and post-translational modifications (PTMs) [1]. Collectively, these alterations generate a wide array of proteoforms for the protein product of a single gene. Herein, PTMs are essential in assigning proteins to a specific role through extending their functions and regulating their activities [2]. Major modifications include, phosphorylation glycosylation and acetylation, of which glycosylation is the most abundant PTM giving rise to proteins in many different glycoforms [3]. These glycomodifications modulate the function of proteins, which includes regulation of cell-cell recognition, cell adhesion, immune response, fertilization, trafficking, and intra- and extracellular signaling events [4]. With such a breadth in influence on protein function, reports increasingly showcase the considerable role of glycoforms in both health and disease [2,5]. Several disease states governed by reduced expression of certain glycans and aberrant glycosylation include hereditary disorders, acute pancreatitis, immune and cardiovascular deficiencies, Alzheimer’s disease and muscular dystrophy, as well as various cancers [4]. Therefore, further understanding in both physiological and pathological processes mediated by the glycome is important in the development of new diagnostic strategies.
An important aspect in understanding the function of glycosylation is the characterization of glycoproteins through the agency of analytical methods. In the appliance of analytical methods, two distinct approaches are recognized, namely glycoproteomics and glycomics, which is the analysis of the intact glycoproteins/glycopeptides and released glycans, respectively. Each of these approaches have a specific advantage in terms of the type of information they procure, therefore, only with the coalescence of these approaches can the fine-mapping of the vast glycoproteome be achieved. A shared hurdle for these approaches is that structures of glycans, unlike proteins, cannot be directly traced back to the genome, and are dependent on the local milieu of enzymes, saccharides and the cell as a whole [6]. As a result, there is no “completed” structure, thus highlighting the challenge of qualifying glycan sequences, and defining their specific function and potential malignancy [7]. As the focus of this thesis, released glycan analysis is mainly utilized to elucidate/quantitate glycan structures. In addition, certain research groups postulate that specific glycans and their structural change can be used as a biomarker to monitor pathological onset so as to realize early diagnoses.However, there is no complete consensus on the reliability of this approach, therewith its clinical relevance, since site-specific and protein-specific information is lost upon glycan release [8]. In contrast, pharmaceutical industries are highly interested in novel released glycan methods as the monitoring of glycan structures and their levels is vital in ensuring the safety and efficacy of biopharmaceuticals.
The majority of glycans can be divided in classes based on their binding to a specific site on a polypeptide backbone, which results in the differentiation between N-linked glycans and O-linked glycans [6]. Even in the analysis of a certain glycan class, the characterization of differential glycan structures remains a challenging task, as will be discussed for released N-glycans in this thesis. These glycans are commonly analyzed by chromatographic, electromigratory, or mass spectrometric systems, being most often combined for the highest performance [9]. Hence, the following well-established instruments are acknowledged: capillary gel electrophoresis and hydrophilic liquid chromatography, both installed with laser induced fluorescence detection, as well as liquid chromatography in hyphenation with mass spectrometry (LC-MS). The latter is
2
discussed in this thesis, wherein LC is employed in separation modes such as reversed-phase liquid chromatography (RPLC), hydrophilic liquid chromatography (HILIC), and porous graphitized carbon chromatography (PGC). This approach excels in its sensitivity and selectivity, however the latter, to a certain extent, as structural isomers (a quite common occurrence for glycoconjugates) cannot be distinguished by their mass. Consequently, state of the art methods need to be employed to separate these minutely different molecules prior to mass detection. This has been achieved with LC and capillary electrophoresis; however, measurements can be lengthy and/or involve complex set ups (e.g., two-dimensional LC). Therefore, the interest in ion mobility spectrometry (IMS) has increased, the past decade, as its separation of ions is based on charge, shape, and size, thus allowing for conformational separation, while it operates in millisecond timeframes [10]. Thus, IMS in combination with MS (IM-MS) may be a high-throughput alternative for confident glycan structure assignment or glycomarker discovery studies. Accordingly, the aim of this literature study is to explore the vast field of glycomics, through a focused view on the analysis of released N-glycans. Wherein, the performance of the emerging IM-MS instruments is evaluated in comparison to the well-established LC-MS approach. For that reason, this study reviews the performance of IM-MS in its many forms and aims to evaluate its utility within the field of glycomics, while elucidating novel findings achieved with both approaches.3
2 Theoretical background
2.1
Glycobiology –
N
-glycan structure
Glycobiology is a particularly broad subject about the study of the structure, biosynthesis, biology and evolution of saccharides (i.e., carbohydrates or glycans) that are present in a wide array of organisms [6]. Therefore, this paragraph narrows its view to the part of glycobiology that is relevant within this study, namely N-glycans.
N-glycans are covalently attached to proteins via the amino acid asparagine (Asn) and are
produced in the endoplasmic reticulum and finalized in the Golgi. Glycoconjugates in mammalian species are commonly made up of around 17 monosaccharides, among those most notable for glycans are: glucose, mannose, galactose (Gal), acetylglucosamine (GlcNAc), N-acetylgalactosamine (GalNAc), fucose, and sialic acid [5]. An N-glycan always has a GlcNAc manose core (a remainder from the dolichol coupled precursor structure) to which a variable amount of the abovementioned monosaccharides can be coupled. The number of monosaccharides and the manner in which these are linked, branched, or form antennae, give rise to the distinction of three main classes, being high-mannose, hybrid, or complex (figure 1).
Figure 1 | Representative examples of N-glycans increasing in complexity from left to right (base,
high-mannose, hybrid and complex). Reprinted and modified from Lyons et al. [11].
The blueprint for these glycan structures, however complex, is not encoded directly in the genome, but instead the structures are produced by a set of competing enzymes [6,12].
4
As a result, an extensive variety of glycan structures that are attached to the protein arise, that diverge in linkages, length, number of antennae, and composition. Moreover, each anomeric carbon is a chiral center, upon which a glycosidic linkage can be formed in either an α- or ß-configuration [6]. These disparities effectuate the occurrence of a single type of protein in various glycoforms; commonly referred to as microheterogeneity. Although the amount of possible glycan structures, in theory, are difficult to grasp (~1012), fortunate for the analytical chemist, in naturethe amount of distinct structures is significantly less since the glycan biosynthetic pathway are heavily regulated. Nonetheless, it is the small variations between glycans that contribute predominantly to the difficulty in which the structure of glycans is determined by means of analytical methods.
2.2
Sample preparation
Analysis of solely the glycan structure requires the release from proteins (or other large biomolecules). Because N-glycans and O-glycans generally demand different release approaches, this preparation step indirectly acts as an enrichment. A subsequent step is glycan derivatization, often proving to be crucial in ensuring glycan structure stability, retentiveness, and ionization efficiency. As touched upon earlier, structures of glycoconjugates are not directly coded into the genome, therewith obstructing amplified production through genetically altered cell lines. Consequently, low abundance glycans are particularly challenging to analyze. Alternatively, enzymatically and chemically synthesized glycans are available, however, exceedingly expensive, thus, underlining the importance of efficient and high recovery sample preparations [13].
2.2.1
Glycan release
Broadly speaking, two approaches are employed to release glycan conjugates from large biomolecules, being enzymatic and chemical release. For N-Glycan release distinctly, enzymatic digestion using endoglycosidase or glycoamidase provides specific and complete sugar removal [12]. Most common for enzymatic N-glycan release is peptide-N-glycosidase F (PNGase F) and PNGase A [13]. The main impediment in this approach is that upscaling is hindered by the high cost of these enzymes. Additionally, in most common approaches glycans are released overnight through 37 °C incubation, hampering the development of high-throughput methods. Fortunately, publications by Lill et al. and Ren et al. demonstrate the possibility of rapid glycan release in 10 and 2 min, through microwave-assisted and immobilized PNGase F sonication methods, respectively [14,15].
The alternative is principally an economical choice. Chemical release is mainly employed in the form of hydrazinolysis or ammonium-based alkali-catalyzed ß-elimination [12]. However, these inexpensive chemicals are at the cost of long reaction times, large amounts of excess reagents and most detrimental for native glycan analysis is the manifestation of peeling reactions (loss of saccharides from the reducing terminus) [13]. In a way to reduce the apparent disadvantages of chemical release, Wang et al. proposes a novel method that utilizes aqueous ammonia to catalyze
N-glycan release reactions [16]. Even so, with a 16-hour reaction, high-throughput remains
unfeasible, its cost-effectiveness, simplicity, and relatively mild reaction conditions makes it an attractive release approach.
2.2.2
Glycan derivatization
The most vital part about derivatization is to ensure molecule stability, considering that the measured molecule needs to be representative of its native state, or at least, the alterations need
5
to be known. In a way derivatization serves as a precaution to prevent unknown changes to the molecule. Furthermore, derivatization can be utilized to enable certain detection methods. For instance, a chromophore label enhances fluorescence and ultraviolet (UV) detection, while isotopic, hydrophobic, and linkage specific tags increase differentiation and sensitivity in certain MS-based assays. Naturally, the incentive is to choose a label that has a multi-facetted function. The most common derivatization strategy is through the reductive amination of the reducing end, forming a secondary amine [17]. Subsequently, popular labels such as 2-aminobenzamide (2-AB), 2-aminopyridine (AP), and more recently introduced RapiFluor-MS are coupled to the molecule of interest [18]. Additionally, glycans can be permethylated, which is the conversion of polar groups (i.e., hydroxyl, amine and carboxyl) to nonpolar methyl groups [12]. Even though permethylation is occasionally afflicted by differential reactivities of monosaccharides and peeling reactions, it ensures stability (notably for sialic acids), may improve ionization efficiency, and surprisingly, with the right separation set up allows for the separation between structural isomers [17].On a critical note, glycan release and subsequent derivatization remain a lengthy, laborious, and at times erroneous process. Developments in reducing these implications are ongoing, however moderate, because oftentimes this segment of the workflow receives the least attention [15,19].
2.3
Separation methods
Oligosaccharides and therewith glycans are composed of monosaccharide units and their composition can be determined by the exact mass obtained with mass spectrometry (MS) [12]. However, quite commonly occurring structural isomers have exactly the same mass, and are solely differential in conformation, hence, this requires a principle that separates based on that dimensionality beforehand. Moreover, separation prior to MS is important to reduce signal suppression of concurrent ions. This matter is amplified for complicated biological samples as it can contain thousands of distinctive glycan moieties. LC has been the methodology of choice up until now, for the following reasons: (1) Ample ways to customize the system to suit the application. (2) Well-established hyphenation with MS. (3) Allows for miniaturized set ups, such as capillary LC and nano LC for potential increase of sensitivity. Although, LC is very effective in minimizing ionization suppression by concurrent ions, the main challenge is the separation of structural isomers. This requires specialized separation modes and often times long gradients. One of the alternatives is the fast developing IMS. This technique is able to resolve conformational isomers, since the ion mobility principle is based on size, shape and charge of the analyte [10]. Furthermore, measurements are significantly quicker (milliseconds opposed to minutes), although at a cost of its resolving power. The addition of IMS to the prevalent LC-MS setup represents a powerful tool in the analysis of complex samples, but how does it perform in a standalone fashion or in a complementary form? This chapter explains the principles of the two separation techniques: LC and IMS. Their recent applications, and combined strengths in the field of glycomics are discussed in chapter 3 and 4, respectively.
2.3.1
Liquid chromatography
In the analysis of released glycans, liquid chromatography is the most commonly applied separation technique. Particularly, the separation modes RPLC, HILIC and PGC have proven to be most suitable [12,20]. Although one might argue that ion exchange chromatography (IEC; notably high-performance anion-exchange chromatography) is an equally valuable addition to the glycan analysis toolkit, since glycans are intrinsically acidic, thus often (multiply) charged under alkaline
6
conditions [21]. However, as of yet, IEC-MS is underrepresented in literature because the high salinity of the mobile phase impairs the coupling to MS. Therefore, IEC will not be discussed in this study in a standalone fashion, but important to note is its promising venue in emerging two-dimensional LC (2D-LC) set ups [22].This paragraph briefly introduces s RPLC, HILIC, and PGC and sheds light on their capabilities in glycan analysis.Reversed-phase chromatography
This separation mode is known for its versatility, robustness, and compatibility to most detection techniques, hence its wide application. The separation is based on the interaction between the nonpolar moieties of the analyte and the stationary phase. Because the mobile phase in RPLC is relatively polar in comparison to the stationary phase, very polar molecules interact almost solely with the mobile phase and leave the column first [23a]. In contrast, more non-polar compounds prefer interaction with the stationary phase. As a result, molecules of similar size elute in order of decreasing polarity. Through varying the mobile phase composition, the interaction between the analyte and the mobile phase is altered, facilitating more retention the more polar the composition (“weaker solvent strength”) and less retention, the less polar the composition (“stronger solvent strength”). .. Due to the hydrophilic nature of glycans in their native form, retention in reversed-phase is close to absent. Therefore, released glycans are first derivatized, according to what is described in paragraph 2.2.2., before they are subjected to separation. On some occasions, permethylation of the glycan is sufficient. In the analysis of glycans, RPLC suffers from low peak capacities (i.e., clustered peaks), since structural differences between glycans are minute, while the added label predominantly determines the retentiveness of the glycan. Novel solutions that can significantly improve separation efficiency are for instance two-dimensional LC or a combination between capillary electrophoresis and LC.
Hydrophilic interaction chromatography
In contrast to RPLC, with HILIC there is no necessity in derivatizing highly polar analytes in order to achieve sufficient retention as HILIC’s separation principle is primarily based on hydrophilicity; normal-phase chromatography with an aqueous-organic mobile phase [23b]. In HILIC one of the analyte retention mechanisms is believed to be caused by partitioning of the analyte between a water-enriched layer of stagnant eluent on a hydrophilic stationary phase and a relatively hydrophobic bulk eluent (mainly composed of 5-40% water in acetonitrile) [24]. Unlike RPLC, analyte retention is not increased with a more water-rich mobile phase, instead water acts like the “stronger” solvent, decreasing analyte retention [23b]. Importantly, at >40% water composition the water-enriched layer disappears and thus a major part of the HILIC retention mechanisms. In addition to the partitioning mechanisms, interactions of the analyte with the silica surface and the attached modifications are important to consider. Especially if analytes are charged their retention behavior can be drastically changed as they may be involved in electrostatic interactions with the stationary phase. Accordingly, sufficient retention and selectivity is achieved for glycans, considering their polarity and charged functional groups in certain conditions [25]. In addition, the high content of acetonitrile (AcN) in the mobile phase promotes the hyphenation with MS and translates to a relatively low backpressure, which allows the use of longer analytical columns. The shortcoming of HILIC, however, remains its inability to be a high-throughput, robust and reproducible platform for glycan analysis, limiting its use to qualitative approaches. Low throughput is a result of long equilibration times after an already lengthy measurement, while robustness and reproducibility are constrained by developments in
7
column technology, incomplete definition of the separation mechanism, as well as the susceptibility to small changes in the method. Although, in recent years major steps forward have been made in demonstrating the feasibility of a robust and reproducible HILIC-MS method, creating a platform for targeted released glycan quantification. Publications by Mancera-Arteu etal. and Shang et al. exemplify that this is achievable through various means, including thorough
method optimization and isotopic labeling [26,27].
Porous graphitic carbon chromatography
Initially, PGC was produced with the idea that it would be the ideal hydrophobic stationary phase, devoid of residual silanol groups often occurring in RPLC [28]. Remarkably, this phase has completely different “retentive properties” than classical non-polar phases and might even be considered a mixed-mode chromatography principle. Most notable retentive properties are: dispersive interactions, the polar retention effect, coulomb type interactions with acidic moieties, and the planar surface; mainly responsible for the resolution between closely related isomers [25]. Contrary to RPLC, PGC is suitable for the analysis of underivatized N-glycans. PGC is often acknowledged as a valid candidate for isomeric glycan analysis, nevertheless, akin to HILIC, validation of this approach is often obstructed by lack of robustness and the irreproducibility of the measurements. These implications predominantly occur due to aging of the column, demonstrating altered elution patterns. Additionally, the incomplete knowledge of the separation mechanism hampers standardization. However, a recent publication by Ashwood et al. reports on how to diminish these implications by utilizing an internal retention time standard, in the form of a dextran ladder [29]. Furthermore, close monitoring of the column performance notably reduces altered retention patterns.
Regardless of the separation mode, factors such as mobile phase composition/additives, the type of column, and the gradient have a considerable effect on the elution of glycans [20]. Furthermore, retention behavior of glycans is linkage specific, and the findings on the influence of certain monosaccharides on the retention of glycopeptides is contradictory to the findings on released glycans. Conclusively, retention behavior of glycans remains hard to predict and require further research.
2.3.2
Ion mobility
One of the first IMS instruments in combination with a single quadrupole was constructed in 1970, a drift tube ion mobility mass spectrometer (DTIMS), at that time known as “plasma chromatography”; conceptually regarded the simplest form of IMS [30,31].Initially, the technique has long been used for detection of explosives and illicit drugs at border crossings, and to gather evidence for the illegal use of chemical agents [32]. In the past decade there has been a surge in development for the combination between IMS and MS, utilized by many research areas, primarily due to the commercialization of the necessary instrumentation. This development has been extremely relevant for the analysis of glycans, in the sense that it offers a wealth of additional information and has the high-throughput capability that this research area so desperately needs [4]. This chapter elucidates the principles of various IMS strategies that are established over the years and are increasingly being employed in the analysis of released N-glycans.
Drift tube (DT) IMS
A drift tube is typically filled with helium (He)or in fewer instances, nitrogen (N2); these two gases
8
(CCS), or simply put, the area covered by a particle [31,33].This CCS can be determined for each ion with the measured drift time and the Mason-Schamp equation [34], shown below:𝛺 = 3𝑧𝑒 16𝑁( 2𝜋 µ𝑘𝐵𝑇 ) 1/2 1 𝐾0
Herein, K0 is the reduced mobility (measured mobility at ambient pressure/temperature), z is the
charge state of the ion, e is the elementary charge, N is the number density of the drift gas, µ is the reduced mass of the ion-neutral drift gas pair, kB as the Boltzmann constant, and T is the gas
temperature [32]. Importantly, the electric field has to be maintained sufficiently low to observe a good correlation between mobility (K0) and shape/size (Ω) [32,35].
The application of a static uniform electric field (5-100 V) propels ions; introduced into the drift tube, orthogonally to the gas flow [32,35]. Positive and negative ions can be measured at the same time because they travel in opposite directions and collected at opposing, segmented electrodes [35]. The time it takes for an ion to reach the end of the tube is determined by its CCS and charge. Regarding ions of the same charge: compact structures (e.g., branched glycans) travel faster than more elongated (e.g., linear glycans) ions, on account of fewer interactions with the drift gas. Other gases than He or N2, that are polarizable, can offer superior separation capabilities for isomeric
species, but this is highly dependent on the physical properties of a certain molecule [36]. First instruments generally suffered from an inherent lack of sensitivity, mainly due to diffusion issues and loss in duty cycle that was related to the pulsed analysis. In recent years, commercialized instruments, such as one by Agilent Technologies, offer increased sensitivity [33]. The utilization of a tandem ion funnel interface is particularly important, since it focuses the ions before and after the drift region [31].
Travelling wave (TW)IMS
Introduced by Kevin Giles et al (2004) and commercialized in 2006, by the release of the Synapt HDMS, this high performance instrument is especially known for its versatility [37,38]. It consists of a stacked-ring ion guide where positive and negative radio frequency (RF) voltages are periodically applied to axially propel the ions. Hence, the electric field forms a travelling wave, its magnitude and speed determine the ions’ separation. Ions as of sorts, surf on the travelling waves, wherein larger ions, deemed the “lesser surfers”, experience larger friction with the buffer gas and are forced behind the waves more frequently, thus increasing their drift time [31]. Mobility separation can be tuned by altering the wave height, velocity, or the pressure of the mobility cell [4]. Ion trajectories through the mobility cell have not been fully elucidated, consequently the CCS cannot be directly measured from the drift time of the analyte. As a result, a calibration is executed against a series of molecules with known CCSs and are ideally of similar ‘class’ to the molecule of interest. Unfortunately, earlier devices suffered from ion heating, which raised questions whether the native structure of analytes could be retained within the instrument. In more recent devices this seem to be solved and a study by Seo et al., on the conformational changes in heparin oligosaccharides confirms this [39].The resolving power of TWIMS (~ 40) is worse compared to DTIMS analyzers, however, its duty cycle and therewith sensitivity is extensively better. A welcome addition in most instruments is the presence of a collision cells before and after the mobility cell, which allows for the contemporaneous analysis of precursor and fragment ions [4].
9
(High-)field asymmetric (FA)IMS / Differential mobility spectrometry (DMS)
Unlike DTIMS and TWIMS, FAIMS, also known as DMS, operates at a high electric field (high E/N, > 30 Td) applied between two electrodes at reduced pressure, therefore, the CCS cannot be derived from the mobility data. In essence, the mobility of an analyte is not directly related to its structure. Accordingly, FAIMS is always used as a gas-phase separation technique at the front-end of the mass spectrometer because the separation of ions is based on the ratio of their mobility between low-field and high-field. This differential mobility is defined by the following properties: mass, shape, center of mass and dipole moment, as well as clustering effects between ions and neutral gas molecules [40]. Ordinarily, the carrier gas consists of nitrogen, however, researchers can adapt to the use of different gases (and/or dopants), better suited to their application [41]. The carrier gas is injected alongside the ions in the longitudinal direction, after which a compensation voltage is superimposed to the dispersion field. In the appliance of a compensation voltage, ions with the desired mobility are forced in a trajectory to vacate the cell, while other ions due to their different mobility disperse [4]. Therefore, FAIMS can operate as an ion mobility filter, similar to the role quadrupoles have, as an m/z filter. In that configuration the sensitivity is increased manifold, essentially due to a higher duty cycle, which translates into an increment in ion count. Interchangeably, by scanning the compensation voltage, multiple ions can be detected, but at the cost of sensitivity. Currently, FAIMS has the highest reported resolving powers of discussed IMS principles, with values as high as 500. However, due to the fact that FAIMS is a spatial mobility device rather than a time-based device like DTIMS or TWIMS constrains the comparison between these principles [42]. Furthermore, another major advantage of FAIMS is that wide array of configurations in hyphenation with different types of mass spectrometers are produced by many different manufacturers. An alternate spatial mobility device is the differential mobility analyzer (DMA). However, unlike FAIMS, ions are not separated by alternating high and low electric fields, rather a high-flow-rate sheath gas is combined with a perpendicular (low-field) electric field [32]. Publications that utilize the DMA devices are as of yet rare, mainly due to the lower achievable resolving power (R ≈ 50) and limited commercial availability [31].
Trapped IMS (TIMS)
Unlike the other principles, separation is not based on the differential dispersion of ions through the gas-phase, rather ions are captured in specific zones through a constant gas flow and an electric field gradient, followed by gradual release [43]. Making, in a way, TIMS the gas-phase equivalent of isoelectric focusing, whereas the other principles are analogous to capillary electrophoresis. Ions are pulsed into an ion funnel in a constant flow of drift gas, where they are radially focused before entering the tube in which ions are trapped by an electric field gradient [44]. In contrast to aforementioned principles, TIMS resolution is not limited by the radial diffusion of the ion swarm during transit through the drift tube.Ions enter the TIMS region, which is built of many consecutive plates and segmented into quadrants by the applied drift gas (exclusively nitrogen). Trapping of ions in different regions within the “trapping cell” is achieved through the administration of an axial electric field gradient [4]. Concurrently, a quadrupole field perpendicular to the electric field is created to confine ions in the center of the cell. Through steady decline of the axial electric field strength, ions from large (high CCS) to small (low CCS) size-tot-charge ratios are released from entrapment selectively. In correspondence to FAIMS, TIMS can be utilized as a mobility filter. Since TIMS operates at the low-field limit, CCS values can be extracted from the data, granted calibration of ions with known CCSs is performed. In its recent commercialization TIMS has already surpassed resolving powers of standard DTIMS and TWIMS
10
devices (~ 300), although demonstration of this feat is lacking due to the scarce amount of publications available of this approach, largely due to it being a relatively young technique compared to the other principles [45]. The described IMS strategies are visualized in figure 2, wherein it is important to notice the differences in the applied electric fields.Figure 2 | Representation of the four primary types of IMS systems. Each general IMS configuration is illustrated
as well as corresponding electric field plot. Reprinted from Morrison et al. [46].
Computational methodologies
The comparison between the CCS of computationally derived structures and the experimental values are paramount in the elucidation of structural information [4]. These comparative studies can offer information about the relative size and conformational dynamics of a certain species of molecules and create a better understanding of the mobility separation principles in general [47]. Modeling of glycans, as relatively small molecules might seem easily achievable, yet the complexity arising from the diversity and flexibility of this species make it much more challenging [4]. As many models exist, one chooses a method in consideration of the tradeoff between approximation accuracy and computational capacity [48]. Additionally, selection of the right algorithm is pivotal for the correct interpretation of experimental data. The most popular models available from low- to high computational stress are: projection approximation [49], exact hard-sphere scattering [50] and the trajectory method [51]. A contemporary publication by Bleiholder
et al. introduces the projection superposition approximation approach, that takes into account
size and shape effects, and jointly lowers computational stress [48]. “Despite the latest developments, carbohydrates are still a long way from their protein counterparts in terms of structural methodology”, as stated by Agirre et al. [52]. With the increasing complexity of glycans being measured in their respective matrix, the developments in computational methodologies may prove to be the crux that halts the unequivocal determination of structural features and conformational dynamics of these glycans.
11
2.4
Glycan ionization
An effective, yet structurally conservative ionization is the basis for a sensitive and elucidative glycan MS-based assay. In an effort to maintain the native glycan structure, soft-ionization techniques are employed, however, the low ionization efficiency of oligosaccharides makes development of a sensitive method a challenging task [12]. In addition, monosaccharides such as fucose and sialic acid are labile enough to be fragmented during the ionization process. Frequently used techniques are electron spray ionization (ESI) and matrix-assisted laser desorption/ionization (MALDI). Importantly, these techniques do not guarantee structure conservation. Initially, MALDI encountered in-source fragmentation and one had to resort to different matrices, alongside permethylation or derivatization of the analytes to maintain the initial structure [53]. There are alternatives that omit the need for derivatization, but the efficiency of these approaches is application specific. Once again, the importance of derivatization is emphasized as it not only ensures stability but can also be used to increase measurement sensitivity. Hence, labels that are easily charged or add to the hydrophobicity of the analyte in LC-ESI-MS approaches have a notable impact on the ionization efficiency [54].In present-day, MALDI-MS is extensively utilized in biomarker discovery and imaging approaches [55], though it often lacks the resolving power for profiling studies, unless it is combined with IMS or LC [56]. An excellent alternative is ESI, the most prevalent ionization technique in bioanalytical studies and often applied for the hyphenation of LC and MS. In addition, ESI offers higher sensitivity and imparts less internal energy on the ion compared to MALDI, which opens up the opportunity to study underivatized native glycans [12]. Lastly, ESI can either benefit or be significantly impaired by constituents of the mobile phase, dimensions of the chromatographic system (e.g., nano LC vs conventional LC) and the number of charge states for an analyte [57]. These parameters are thus vital in developing a sensitive MS measurement and require careful consideration in the method development process.
2.5
Glycan dissociation
The formation of fragments offers the opportunity for higher selectiveness in targeted MS-based assays (e.g., quantitative QqQ approaches) and is valuable in the elucidation of glycoconjugate structures in characterization studies. Many fragmentation approaches exist, from low energy dissociation methods such as collision induced dissociation (CID) and infrared multiphoton dissociation (IRMPD), to electron-based dissociation methods, for instance, electron transfer dissociation (ETD), electron capture dissociation (ECD), and more [58]. Carbohydrate fragmentation recognizes two common pathways, namely, cleavage of the bonds between the sugar rings (glycosidic cleavages) and cleavage of the rings themselves (crossring cleavages) [59]. Glycosidic cleavages provide details on the sequence, monosaccharide compositions, and branching, whereas crossring cleavages locate modifications and pinpoint the position of a glycosidic linkage within individual residues [58]. These fragmentations have a specific nomenclature, coined by Domon and Costello in 1988 [60]. In essence, the distinction is made between A (crossring), B and C (glycosidic) ions, that retain the charge on the nonreducing terminus, and X (crossring), Y and Z (glycosidic) ions, that retain the charge on the reducing terminus (figure 3) [59].
12
Figure 3 | Nomenclature introduced by Domon and Costello for describing the fragmentation of carbohydrates.
Reprinted from Harvey et al. [59].
To effectuate these fragmentations, low-energy CID fragmentation is the most commonly employed dissociation method for the fragmentation of positively charged ions. Specifically for native N-linked glycans, glycosidic fragmentation is the favored pathway and may induce the appearance of rearrangement ions, especially for glycans derivatized at the reducing terminus [58,59]. In an attempt to prevent formation of rearrangement ions, metal adduction and derivatization of the reducing terminus by reductive amination have been successfully employed [61]. Interestingly, negative and positive mode in MS/MS provide different fragmentation patterns for oligosaccharides and may prove advantageous in structure interpretation, especially when sialic acids are present. Despite its exceptional performance for many applications, collisional activation affords insufficient energy deposition for the formation of certain types of fragment ions in structure specific applications [62].
Therefore, as an alternative methodology, electron aided dissociation (ExD) — a variety of
dissociation methods based on electron energies — is gaining momentum in glycopeptide analysis
but has shown to be suitable for fragmentation of released glycans as well. “These methods entail the transfer of an electron to or from a selected multiply charged molecular ion, yielding a radical ion which then undergoes fragmentation” [58]. A typical electron aided dissociation fragmentation spectrum includes an abundance of crosslink cleavages. In case either dissociation method individually fails to provide sufficient information, a combination between these methods may provide more comprehensive results. For example, the conjoined CID and ETD form the dissociation method EThcD.
Lastly, ultraviolet photodissociation (UVPD) is a relatively young technique, with alluring attributes for glycan structure elucidation. In contrast to its photo dissociative counterpart IRMPD, where cleavages are formed through low energy dissociation pathways, UVPD produces higher energy photons (Eultraviolet > Einfrared) that promote differential fragmentation through
pathways with significantly higher activation energies [62]. Moreover, UVPD is well-equipped to produce more A and X-type ions (i.e. crossring cleavages) which are highly useful in studying branching patterns in glycans. Since, the dissociation process is independent of the charge polarity, both cations and anions can be measured without the known implications for CID. Further contrast to CID is the disparity in the modulation of the fragmentation process.
13
While the translational excitation of ions in CID is principally based on the collision energy, the net energy deposition from UVPD is altered through laser parameters: wavelength, photon flux and exposure time, and thus arguably more easily tunable. Despite its flexibility and unique capabilities, UV lamps/lasers are still costly, and commercial UVPD-integrated systems are limited, but certainly holds merit for prospective studies.2.6
Mass analyzers
The mass analyzer is situated at the end of the analytical workflow and its performance is therefore highly dependent on preceding principles, as illuminated in the paragraphs above. The combination of its sensitivity, selectivity, and to a certain extent its structure elucidative capability ascribe it as the unprecedented detection technique in glycomics [63]. Generally, a distinction can be made between low-resolution and high-resolution mass spectrometers (HRMS). Essentially, these types of instruments are not distinguished by their performance, but purposely constrained to certain approaches at which they thrive, which is construed in paragraph 2.6.1. and 2.6.2.
2.6.1
Low-resolution
Triple quadruples (QqQ) and ion traps are regarded the low-resolution devices among all mass spectrometers. In this thesis the focus lies on QqQ instruments considering that ion traps are utilized on mere rare occasions in the analysis of released glycans, although ion traps hold merit on occasions where extensive fragmentation and MSn analysis are needed for structure
elucidation. Even though low-resolution has a negative tone to it, the appeal to QqQ instruments are attributes such as high sensitivity and reproducibility/robustness. For this reason, QqQ instruments are predominantly employed for quantitative- and selective screening assays [64]. To design highly selective and sensitive approaches an instrument is ordinarily run in multiple reaction monitoring (MRM) mode. This type of measurement entails the selection of a specific precursor ion (Q1), subsequently fragmentation of the respective precursor ion (q2), and finally a selective scan of carefully chosen fragments representing the precursor (Q3). Despite needing to scan multiple fragments in the last quadrupole, a QqQ is able to retain high sensitivity (high duty cycle) owing to its fast scanning speed, unmatched by high-resolution instruments.
2.6.2
High-resolution
Among the high-resolution instruments, time-of-flight (TOF), Fourier-transform ion cyclotron resonance (FT-ICR), and orbitrap are primarily exploited. The power of these instruments is their capability to determine the exact mass with very high accuracy (down to 0.0001 Da), and respectively allows for the determination of the elemental composition of small molecules [65]. Unfortunately, the bottleneck regarding frequently occurring structural isomers in glycoconjugates cannot be resolved by high-resolution instruments either, because that would require the differentiation of ~1 eV in the heat of formation, equivalent to the exact mass determination of ~10-9 Da [65]. TOF-analyzers are mostly first choice in high-throughput and
untargeted screening approaches, due to their fast acquisition rate relative to orbitrap and especially FT-ICR. Moreover, TOF is the least costly of these three types of high-resolution instruments. The rapid measurements of TOF instruments are particularly useful in screening approaches in search of relevant glycosidic markers as nominal equivalent masses from different molecule species are easily differentiated by means of accurate mass [12]. For instances requiring the highest resolution obtainable, FT-ICR and orbitrap are preferred [65].
14
Especially in the analysis of highly complex samples, developing a structure elucidative approach is challenging, essentially both types of instruments are required. This explains the emergence of hybrid-devices (e.g., Q-TOF) that are equipped to have the capability of fragmentation whilst maintaining its accurate mass measurement. This can offer higher selectivity and/or allow for more confident structure assignment.15
3 Glycan profiling
Before exploring the plethora of LC-MS and IM-MS-based applications in the analysis of the glycome it is important to reiterate the aim of a released glycan approach, opposed to glycopeptide and whole glycoprotein (intact measurements) analysis. In order to understand the physiological and pathophysiological role of glycans, by utilizing a released glycan approach, the macro- and microheterogeneity of glycosylation is determined through the elucidation of glycoconjugate monosaccharide composition and linkage [63]. Typically, omics, and therewith released glycan analysis, recognize two different means of studying differential glycan structures, namely, discovery (e.g., biomarker screening) and targeted analysis (e.g., clinical assays). Released glycan analysis may serve as a diagnostic tool in targeted assays, whereas glycomic discovery oftentimes serves in complement to glycoproteomics in the understanding of glycobiology [6]. Modern separation systems hyphenated to mass analyzers provide very high resolving powers. However, generally, comprehensive structural determination requires iterative fragmentations and complementary techniques such as NMR [63]. Nonetheless, the analysis of the vast isomeric and isobaric heterogeneity of glycoconjugates demands the use of the highest performing equipment. Thus, the idea of an individual method that can resolve at single glycan entity level is unfeasible, especially if multiplex studies are concerned [66]. As an answer to this impediment, in recent research the aim is to add findings to large databases in an attempt to gradually map the glycoproteome. As a result, the role of glycoinformatics is becoming progressively more important in qualitative and quantitative profiling of glycan structures, as well as, storage of immense amounts of data and the development of prediction algorithms [67].
3.1
Liquid chromatography-mass spectrometry
As delineated in chapter 2.3, there are a few separation modes that are employed commonly in the analysis of glycans, including RPLC, HILIC and PGC. Accordingly, in this chapter their capability in glycan profiling is discussed through the review of recent applications.
In a study by Melmer et al., (ion pair) RPLC, HILIC and PGC were statistically compared by the means of multiple linear regression [25]. This model correlated the retention properties of the stationary phases with the monosaccharide composition regarding protein N-glycans. These glycans were released and 2-AB labeled from proteins RNase-B, fetuin and IgG. The advantage of this statistical approach is that it quantitatively assesses the contributions of the different monosaccharides to the retention in each investigated chromatography mode. Expectedly, the highest selectivity was achieved with PGC and HILIC, however, the addition of an ion pair (IP) to the RPLC mode greatly increased separation and peak shapes for isomeric sialylated glycans in comparison to generic RPLC. This was substantiated by the relatively high correlation between IP-RPLC and PGC (0.69), because simplistically put, PGC is a hydrophobic phase with ionic (and some other) properties. An overall insightful overview of monosaccharide selectivity’s and correlation between separation modes is illustrated by figure 4, wherein the separation modes are plotted against each other, and the specific glycan types are annotated.
16
Figure 4 | Retention times of high-mannose ( ), fucosylated complex ( ), non-fucosylated ( ) and complex sialo
glycans ( ) in HILIC, IR-RPC and PGC plotted against one another, respectively. Light grey arrows indicate the effect of the addition of a core-fucose. The black arrows indicate the effect of the addition of two sialic acids. Isomers are marked by dotted ellipses. Reprinted from Melmer et al. [25].
From figure 4 the high orthogonality between HILIC and IP-RPLC is directly apparent, which is underlined by a correlation coefficient of 0.17. Furthermore, HILIC and PGC are significantly more correlated for ionic species (0.67) than for neutral molecules (0.38). Essentially, the data indicates the high resolving power that can be achieved with a 2D-LC set up, especially if separation modes are highly orthogonal. Nevertheless, this is constrained for IP-RPLC and HILIC by the incompatibility of the mobile phases and would require novel modulations strategies. Additionally, both separation modes would perform suboptimally in the 2nd dimension due to
their required equilibration times. In contrast, PGC functions well with the IP-RPLC mobile phase and may offer the additional resolving power needed in certain applications. The fact that these separation modes demonstrate low correlation may be considered a double-edged-sword, because even though moderate correlation between separation modes serves a 2D-LC setup well, diversely, low correlation means incomparable retention times (RTs) between separation modes, hence a universal glycan database cannot be realized.
Recent glycomic research trends have exhibited an increase in the analysis of isomeric glycans, since studies have shown that these minutely dissimilar molecules have a considerable influence on biological processes [68,69]. Ultimately, the quantification of certain linkage isomers could facilitate the discovery of diagnostic biomarkers [26]. Indeed, the obvious choice for analyzing these isomers would be HILIC or PGC, however that questions the relevance of RPLC in this specific area of glycomics. The inherent problem of poor resolution between isomeric species stems from the principle of RPLC, as these isomers solely differ in structure and not hydrophobicity, meaning an indistinguishable difference in interaction with the stationary phase. In order to improve resolution between isomers the most common measures are: highly specific derivatization (linkage specific sialic acid derivatization [70]), long linear gradients (i.e. long analysis time), miniature systems (e.g. capillary LC), 2D-LC set ups and increased column temperature[20,21,71]. A method developed by Zhou and colleagues demonstrates that with a relatively generic approach and increased column temperatures isomeric separation can be achieved [72]. The upside of RPLC is its entrenchment in bioanalysis, i.e., the equipment is most developed, widely available, robust, and offers ample opportunity to customize to a specific application. Nevertheless, some alterations (e.g. derivatization) aiming to improve resolution between isomers can be detrimental to reproducibility and robustness.
17
Conclusively, two options are available: (1) design of RPLC methods that enhances isomeric separation while maintaining reproducibility and robustness, (2) development of novel strategies that aid in accomplishing robust and reproducible HILIC and PGC assays. The latter is presently best represented in literature and is therefore the focal point of this chapter.In contrast to RPLC, HILIC is a separation mode catered to highly hydrophilic (charged) species and hereby able to retain and selectively separate native glycans, although these are usually derivatized on the reducing end to improve ionization. Moreover, the highly organic mobile phase increases ionization efficiency and therefore promotes the measurement of low abundant glycan species. In HILIC, and glycomics as a whole, there is a conundrum about the balance between high sensitivity and the reproducibility of the method. A publication by Shang et al. discusses this matter after a thorough method development for the analysis of N-glycosylation patterns in mAb therapeutics with HILIC-ESI-QTOF [27]. Herein, the primary requirement for the ideal N-glycan profiling method would be the ability to simultaneously detect and identify the major-, minor, and trace-level (<1%) glycan species in a rapid and definitive manner. Therefore, chromatography and MS systems should need to be robust, be designed for optimal sensitivity and resolution, and the acquired data should be reproducible. In an attempt to achieve the “ideal method” Shang et al. proposed the performance attributes shown in table 1.
Table 1 | Attributes and Design Goals of an Optimal LC/MS Method for Characterization of Released N-Glycans
from Biotherapeutic mAbs. Reprinted from Shang et al. [27]
Desirable Attributes of an Optimal LC/MS N-Glycan Method
Chromatography Separation and detection of all commonly observed mAb N-glycans, including the
less abundant afucosylated, sialylated, high mannose, and hybrid structures Robust and reproducible chromatographic profiles (i.e., retention time, signal intensities, and relative peak areas), especially for low abundance N-glycans
LC/MS interface Easy method setup for routine operation Minimal column shedding (low m/z noise)
MS compatibility for minimal ion source fouling and robust method performance Total ion current (TIC) chromatogram compares well to the fluorescence profile Stable MS signal during course of a run
MS data quality Optimal sensitivity to facilitate analysis of low-level species Minimal signal splitting due to multiple charge states Reduced level of analyte adduction (e.g., sodium, TFA, etc.)
Minimal artifact peaks from gas phase fragmentation and electrochemical degradation
Accurate mass measurements for N-glycan identification
In earlier method designs the mobile phase was devoid of additives in order to omit the formation of adducts during ionization and consequently improve the sensitivity of the method. Adduct formation and multiply charged ions for higher glycans systematically decreases signal intensities and are the primary reason that make direct quantitation incredibly challenging [73]. Although, the method succeeded in sensitively measuring all glycan levels, reproducibility of the measurements was an issue.
18
In the venture of tackling this impediment, trifluoroacetic acid (TFA) seemed to be the key in ensuring reproducible separations, through what is believed the protonation of the carboxylic groups of the sialic acids, which suppressed the charge repulsion with the charges on the stationary phase and thus improved retention of these glycan species.Regardless of the many studies that have notified the detrimental effect of TFA on the ionization of analytes, the researchers valued the robustness and reproducibility gained above the highest achievable sensitivity [74]. That underlines the necessity of understanding the aim of an application and striking the right balance between the many important performance attributes that make the method fit for purpose.
Concerning isomeric separation, zic-HILIC phases are commonly employed. The appended selectivity is facilitated by what is believed the electrostatic interactions between the glycans and the sulfobetaine stationary phase [21]. A novel µZIC-HILIC-MS method developed by Mancera-Arteu et al. aimed to separate N-glycan isomers released from human alpha-1-acid-glycoprotein [26]. More specifically, the goal was to distinguish between specific linkage-type isomers of sialic acid and fucose. A pivotal step in ensuring confident structure assignment was the inclusion of glycan reductive isotope labelling (GRIL). Glycans were isotope-coded with [12C6]-aniline ([12C6]
AN) and [13C6]-aniline ([13C6]AN) after a certain exoglycosidase digestion (dependent on the
measurement). This enabled the comparison of two samples in the same run, which corrects for the lack of RT reproducibility between runs and avoids many artifacts that add to mass spectrometer variations (e.g., ion suppression). Moreover, GRIL allows for the precise comparison of glycan fingerprints before and after digestion with a specific exoglycosidase, which is illustrated in figure 5.
Figure 5 | Experimental workflow followed in this work to characterize sialic acid (A) and fucose linkage-type
19
The authors succeeded in assigning most of the glycan isomers, the remaining unassigned glycans are the least abundant ones wherefore the MS signal was too low for reliable characterization. The main drawback of this approach is the difference in ionization between intact and desialylated1 glycans. Regardless, the GRIL approach allowed for the characterization of many differential triantennary and tetraantennary glycans, a feat accomplished by few. The success of this method substantiates the authors long term goal of employing this methodology for robust and reliable identification of new biomarkers in patho-glycomic studies. In a successive, preliminary study, utilizing a similar analytical method, Mancera-Arteu et al. made a step in achieving this long term goal, through the use of multivariate data analysis of aberrant glycosylation in pancreatic ductal adenocarcinoma (PDAC) [75]. Presently, multivariate data analysis is utilized in most of the omics fields, although only sparsely employed in glycomics. However, in this study the researchers proved the strength of multivariate analysis and found a potential biomarker that allows for the differentiation between PDAC and chronic pancreatis patients.Indeed, dedicated HILIC applications are able to resolve isomeric glycans, however the most prominent chromatography technique in that respect is PGC; principally due to its shape selectivity, analogues to IMS. The main drawback that has halted PGC from broader application is an incomplete definition of the separation mechanism. Consequently, glycan structure RTs cannot be accurately predicted using in silico models [29]. In addition, an eminent issue is low retention reproducibility, predominantly caused by the column’s high susceptibility to contamination [73]. Moreover, aged columns showcase altered elution characteristics impeding the development of standardized assays [76]. Fortunately, in recent research a direction is taken toward resolving said obstacles in quite the number of unique approaches, including the GRIL approach, which has proven its utility in HILIC, also showcases its value in isomeric PGC analysis [77].
A bottleneck, apparent in earlier discussed publications, especially for low abundant glycans, is poor measurement reproducibility, which is also an issue for the first eluting glycans in PGC separations, due to the low organic modifier composition at the start of the gradient. Hinneburg
et al. demonstrates a relatively simple way to improve reproducibility of mentioned glycans,
namely by the addition of post-column make-up flow (PCMF) to an established capillary-flow PGC-LC-MS/MS method [78]. The idea of PCMF is to increase the organic modifier composition between column and ionization source. Hereby glycan separation is uncompromised while sensitivity is improved. In this approach a 3 µL/min column flow was supplemented with a 4 µL/min PCMF (43 and 57% v/v, respectively). In parable to the original method, to which this “modified method” is compared to, glycans were released from bovine fetuin, ribonuclease B and human immunoglobulin G. For the PCMF, three organic modifiers were tested (i.e. methanol, AcN and isopropanol), with a composition between 70 and 100% (v/v, organic/water), of which, quite expectedly, 100% organic composition induced the highest glycan sensitivity for all modifiers. The authors report a great increase in signal response across all glycan types (30- to 100-fold), particularly for AcN and isopropanol the highest analyte signals were measured, of which AcN is practically most appealing. Important to note is that the added PCMF not only improves sensitivity but also ensures the spray stability [79]. Conclusively, alongside increased sensitivity, the PCMF module improves coverage and quantitative accuracy of early eluting and low-abundance glycans in an easy to implement manner [78].
1Desialyation: removal of the terminal sialic acid residues on glycans (desialylated, not to be confused with di-sialylated)
20
Nevertheless, the application of Hinneburg et al. does not focus on the PGC column itself, which is arguably the most significant source of variability in a PGC-based method. Therefore, in the pursuit of diminishing variability and developing robust analytical methods a good understanding of the separation principle is paramount. A study by Pabst et al. investigates some of the primary factors that influence separation, such as electrosorption, temperature, and ion polarity, which unsurprisingly (considering PGC as mixed-mode), provided some counterintuitive results [76]. Of course, simply each additional monosaccharide unit and in turn every type of linkage induces a change in separation. In that respect is the systematic mapping of RT and elution order of biologically relevant glycans, even though cumbersome, quite effective for assessing the quality of separation and confident structure assignment for future studies [80]. Accordingly, Abrahams etal. and collaborators have taken the initiative to achieve this through the establishment of a PGC
RT library as part of the elution centric database GlycoStore (www.glycostore.org). The publication by Abrahams et al. reports an addition to the library of 100+ glycan structures, which were released from purified standard glycoproteins (e.g., IgG, IgA, fetuin, lactoferrin, ovalbumin, RNase B, and alpha 1-acid glycoprotein). Data for each glycan structure includes the relative RT and the respective MS/MS spectrum. Similarly to the method by Mancera-Arteu et al., exoglycosidase was utilized for the in depth characterization of glycan structures [26]. Certainly, a comprehensive RT database can be decisive for the reproducibility and robustness of PGC chromatography methods in routine glycan analysis, granted vigilance is maintained for inter lab variance and performance of aged columns. In terms of highly complex sample analysis, considerable amounts of additional data are still required, which can be realized with consistent inclusion of data over the coming years. Ultimately, the ideal would be the accurate automated assignment of glycan structures to aid in glycomarker discovery.
Lastly, Ashwood et al., alternatively, introduces a PGC based method that reduces system-dependent variation in RT and peak area, in furtherance of establishing system-insystem-dependent retention values [29]. In order to achieve this, Ashwood et al. developed a method that makes use of a so-called dextran ladder, serving as an internal RT standard. In previous studies a dextran ladder has been successfully implemented for the determination of system-independent retention constants measured in glycose units (GU), on a HILIC system with fluorescence detection [81]. However, Ashwood et al. are the first to develop a dextran ladder approach that employs a PGC-ESI-MS/MS system, potentially improving isomer separation (PGC vs HILIC) and structure assignment (MS vs fluorescence). To commence the method development, the elution behavior of a reduced dextran ladder was investigated, and parameters that are important for the use of the dextran ladder as an internal standard were ascertained. Resultantly, the measurements indicated that the elution of the dextran ladder was best fitted with a logarithmic elution pattern. After the establishment of the method, it was tested with analysis of N-glycans released and reduced from complex protein mixtures, inclusive of secreted protein mixtures, cell lysates and tissue lysates. RT variation was normalized to a GU index and peak areas were normalized by quantifying glycans based on their extracted ion chromatogram peak area as a ratio to the dextran ladder internal standards. In figure 6 the normalization process is visualized.
