Immunoglobulin G (IgG) Fab glycosylation analysis using a new mass spectrometric high-throughput profiling method reveals pregnancy-associated changes.

The N-linked glycosylation of the constant fragment (Fc) of immunoglobulin G has been shown to change during pathological and physiological events and to strongly influence antibody inflammatory properties. In contrast, little is known about Fab-linked N-glycosylation, carried by ∼ 20% of IgG. Here we present a high-throughput workflow to analyze Fab and Fc glycosylation of polyclonal IgG purified from 5 µl of serum. We were able to detect and quantify 37 different N-glycans by means of MALDI-TOF-MS analysis in reflectron positive mode using a novel linkage-specific derivatization of sialic acid. This method was applied to 174 samples of a pregnancy cohort to reveal Fab glycosylation features and their change with pregnancy. Data analysis revealed marked differences between Fab and Fc glycosylation, especially in the levels of galactosylation and sialylation, incidence of bisecting GlcNAc, and presence of high mannose structures, which were all higher in the Fab portion than the Fc, whereas Fc showed higher levels of fucosylation. Additionally, we observed several changes during pregnancy and after delivery. Fab N-glycan sialylation was increased and bisection was decreased relative to postpartum time points, and nearly complete galactosylation of Fab glycans was observed throughout. Fc glycosylation changes were similar to results described before, with increased galactosylation and sialylation and decreased bisection during pregnancy. We expect that the parallel analysis of IgG Fab and Fc, as set up in this paper, will be important for unraveling roles of these glycans in (auto)immunity, which may be mediated via recognition by human lectins or modulation of antigen binding.

Comparative performance of four methods for high-throughput glycosylation analysis of immunoglobulin G in genetic and epidemiological research.

The biological and clinical relevance of glycosylation is becoming increasingly recognized, leading to a growing interest in large-scale clinical and population-based studies. In the past few years, several methods for high-throughput analysis of glycans have been developed, but thorough validation and standardization of these methods is required before significant resources are invested in large-scale studies. In this study, we compared liquid chromatography, capillary gel electrophoresis, and two MS methods for quantitative profiling of N-glycosylation of IgG in the same data set of 1201 individuals. To evaluate the accuracy of the four methods we then performed analysis of association with genetic polymorphisms and age. Chromatographic methods with either fluorescent or MS-detection yielded slightly stronger associations than MS-only and multiplexed capillary gel electrophoresis, but at the expense of lower levels of throughput. Advantages and disadvantages of each method were identified, which should inform the selection of the most appropriate method in future studies.

Skewed Fc glycosylation profiles of anti-proteinase 3 immunoglobulin G1 autoantibodies from granulomatosis with polyangiitis patients show low levels of bisection, galactosylation, and sialylation.

Granulomatosis with polyangiitis (GPA) is associated with circulating immunoglobulin (Ig) G anti-proteinase 3 specific (anti-PR3) anti-neutrophil cytoplasm antibodies (ANCA), which activate cytokine primed neutrophils via Fcgamma receptors. ANCA are class switched IgG antibodies implying T cell help in their production. Glycosylation of IgG Fc, under the control of T cell cytokines, determines the interaction between IgG and its receptors. Previous studies have reported aberrant glycosylation of Ig Fc in GPA patients. We investigated whether aberrant Fc glycosylation was present on anti-PR3 ANCA as well as whole IgG subclass preparations compared to healthy controls and whether this correlated with Birmingham vasculitis activity scores (BVAS), serum cytokines, and time to remission. Here, IgG Fc glycosylation of GPA patients and controls and anti-PR3 ANCA Fc glycosylation were determined by mass spectrometry of glycopeptides. IgG1 and IgG2 subclasses from GPA patients showed reduced galactosylation, sialylation, and bisection compared to healthy controls. Anti-PR3 IgG1 ANCA Fc galactosylation, sialylation, and bisection were reduced compared to total IgG1 in GPA. Galactosylation of anti-PR3 ANCA Fc correlated with inflammatory cytokines and time to remission but not BVAS. Bisection of anti-PR3 ANCA Fc correlated with BVAS. Total IgG1 and anti-PR3 IgG1 Fc galactosylation were weakly correlated, while bisection of IgG1 and anti-PR3 showed no correlation. Our data indicate that aberrant ANCA galactosylation may be driven in an antigen-specific manner.

Pro-inflammatory pattern of IgG1 Fc glycosylation in multiple sclerosis cerebrospinal fluid.

BACKGROUND: Immunoglobulin G (IgG) effector functions are regulated by the composition of glycans attached to a conserved N-glycosylation site in the Fc part. Intrathecal production of IgG, especially IgG1, is a hallmark of multiple sclerosis (MS), but nothing is known about IgG Fc glycosylation in MS and in cerebrospinal fluid (CSF) in general. METHODS: We applied mass spectrometry of tryptic Fc glycopeptides to analyze IgG Fc glycosylation (sialylation, galactosylation, fucosylation, and bisecting N-acetylglucosamine (GlcNAc)) in 48 paired CSF and serum samples from adult patients with MS or a first demyelinating event highly suggestive of MS (designated as MS cases), and from healthy volunteers and patients with other non-inflammatory diseases (control group). p values were adjusted for multiple testing. RESULTS: Our experiments revealed four main results. First, IgG1 glycosylation patterns were different in CSF vs. serum, in the MS group and even in control donors without intrathecal IgG synthesis. Second, in MS patients vs. controls, IgG1 glycosylation patterns were altered in CSF, but not in serum. Specifically, in CSF from the MS group, bisecting GlcNAc were elevated, and afucosylation and galactosylation were reduced. Elevated bisecting GlcNAc and reduced galactosylation are known to enhance IgG effector functions. Third, hypothesis-free regression analysis revealed that alterations of afucosylation and bisecting GlcNAc in CSF from MS cases peaked 2-3 months after the last relapse. Fourth, CSF IgG1 glycosylation correlated with the degree of intrathecal IgG synthesis and CSF cell count. CONCLUSIONS: The CNS compartment as well as the inflammatory milieu in MS affect IgG1 Fc glycosylation. In MS, the CSF IgG1 glycosylation has features that enhance Fc effector functions.

Anti-citrullinated protein antibodies acquire a pro-inflammatory Fc glycosylation phenotype prior to the onset of rheumatoid arthritis.

OBJECTIVE: To determine whether anticitrullinated protein antibodies (ACPA) exhibit specific changes in Fc glycosylation prior to the onset of arthritis. METHODS: Serum samples of patients with ACPA-positive arthralgia (n=183) were collected at baseline and at various time points of follow-up. 105 patients developed arthritis after a median of 12 months (IQR 6-24) and were classified as having either rheumatoid arthritis (RA, n=48) or undifferentiated arthritis (UA, n=57) based on the 1987 American College of Rheumatology (ACR) criteria. ACPA and total serum IgG were isolated by affinity purification and cleaved by trypsin. ACPA-IgG1 Fc-glycopeptides were subsequently analysed by nano-liquid chromatography mass spectrometry and compared to those of total IgG1. RESULTS: At baseline, ACPA-IgG1 and total IgG1 from arthralgia patients displayed similar Fc glycosylation patterns. By contrast, at the onset of arthritis, ACPA exhibited a decrease in galactose residues in RA patients, but not in UA patients. This decrease occurred around 3 months prior to diagnosis and was paralleled by an increase in systemic inflammation (erythrocyte sedimentation rate). Galactosylation of total IgG1 was also decreased in RA, but this did not precede the onset of arthritis. Interestingly, we additionally noted a higher degree of ACPA-IgG1 Fc core fucosylation at baseline as compared with total IgG1, which further increased prior to diagnosis. CONCLUSIONS: ACPA display significant changes in Fc galactosylation and fucosylation prior to the onset of RA. These changes towards a more pro-inflammatory phenotype could be involved in driving the disease process.

Glycoproteomic analysis of antibodies.

Antibody glycosylation has been shown to change with various processes. This review presents mass spectrometric approaches for antibody glycosylation analysis at the level of released glycans, glycopeptides, and intact protein. With regard to IgG fragment crystallizable glycosylation, mass spectrometry has shown its potential for subclass-specific, high-throughput analysis. In contrast, because of the vast heterogeneity of peptide moieties, fragment antigen binding glycosylation analysis of polyclonal IgG relies entirely on glycan release. Next to IgG, IgA has gained some attention, and studies of its O- and N-glycosylation have revealed disease-associated glycosylation changes. Glycoproteomic analyses of IgM and IgE are lagging behind but should complete our picture of glycosylation's influence on antibody function.

IgG Fc N-glycosylation in Guillain-Barré syndrome treated with immunoglobulins.

Intravenous immunoglobulin (IVIg) is the treatment of choice for Guillain-Barré syndrome (GBS), an immune-mediated peripheral neuropathy causing rapidly progressive limb weakness and respiratory failure. The working mechanism of IVIg in autoimmune diseases has not been elucidated, but previous studies indicate that some anti-inflammatory effects may be mediated by the N-glycosylation of the Fc-portion of IgG. GBS is a model disease to investigate these effects because GBS is an acute and monophasic disorder usually affecting healthy persons, which is treated with a standard course of IVIg, although the clinical response is highly variable. In the current study, the N-glycosylation of the Fc-portion of serum IgG was investigated in patients with GBS before and after treatment with IVIg in relation to clinical course and outcome. Glycoforms of serum IgG1 and IgG2 were determined separately by liquid chromatography mass spectrometry. These IgG subclasses were purified from the serum of 174 GBS patients before and in 150 patients 2 weeks after standard IVIg treatment regimen. Treatment-naive GBS patients compared with age- and sex-matched controls had lower levels of galactosylation of IgG1 and IgG2. IVIg preparations contained relatively high levels of galactosylated and sialylated IgG Fc glycoforms compared with serum IgG in patients. Treatment with IVIg resulted in an increase in serum of the Fc-galactosylation and -sialylation of both IgG1 and IgG2. The extent of normalization in serum IgG Fc glycosylation varied between patients. Multiple logistic regression analysis showed that patients with persistent low IgG galactosylation and sialylation despite IVIg treatment had the most severe forms of GBS and needed ventilator support more often. Kaplan-Meier analysis showed that these patients also needed more time to be able to walk again compared with patients with a normalized IgG Fc glycosylation profile. In conclusion, our results suggest that serum IgG Fc glycosylation in GBS is related to disease severity and clinical recovery after IVIg and may help to develop new measures to monitor the efficacy of treatment.

Changes in antigen-specific IgG1 Fc N-glycosylation upon influenza and tetanus vaccination.

Antibody effector functions have been shown to be influenced by the structure of the Fc N-glycans. Here we studied the changes in plasma or serum IgG Fc N-glycosylation upon vaccination of 10 Caucasian adults and 10 African children. Serum/plasma IgG was purified by affinity chromatography prior to and at two time points after vaccination. Fc N-glycosylation profiles of individual IgG subclasses were determined for both total IgG and affinity-purified anti-vaccine IgG using a recently developed fast nanoliquid chromatography-electrospray ionization MS (LC-ESI-MS) method. While vaccination had no effect on the glycosylation of total IgG, anti-vaccine IgG showed increased levels of galactosylation and sialylation upon active immunization. Interestingly, the number of sialic acids per galactose increased during the vaccination time course, suggesting a distinct regulation of galactosylation and sialylation. In addition we observed a decrease in the level of IgG1 bisecting N-acetylglucosamine whereas no significant changes were observed for the level of fucosylation. Our data indicate that dependent on the vaccination time point the infectious agent will encounter IgGs with different glycosylation profiles, which are expected to influence the antibody effector functions relevant in immunity.

High-throughput IgG Fc N-glycosylation profiling by mass spectrometry of glycopeptides.

Age and sex dependence of subclass specific immunoglobulin G (IgG) Fc N-glycosylation was evaluated for 1709 individuals from two isolated human populations. IgGs were obtained from plasma by affinity purification using 96-well protein G monolithic plates and digested with trypsin. Fc N-glycopeptides were purified and analyzed by negative-mode MALDI-TOF-MS with 4-chloro-α-cyanocinnamic acid (Cl-CCA) matrix. Age-associated glycosylation changes were more pronounced in younger individuals (<57 years) than in older individuals (>57 years) and in females than in males. Galactosylation and sialylation decreased with increasing age and showed significant sex dependence. Interestingly, the most prominent drop in the levels of galactosylated and sialylated glycoforms in females was observed around the age of 45 to 60 years when females usually enter menopause. The incidence of bisecting N-acetylglucosamine increased in younger individuals and reached a plateau at older age. Furthermore, we compared the results to the total IgG N-glycosylation of the same populations recently analyzed by hydrophilic interaction liquid chromatography (HILIC). Significant differences were observed in the levels of galactosylation, bisecting N-acetylglucosamine and particularly sialylation, which were shown to be higher in HILIC analysis. Age and sex association of glycosylation features was, to a large extent, comparable between MALDI-TOF-MS and HILIC IgG glycosylation profiling.

Association between galactosylation of immunoglobulin G and improvement of rheumatoid arthritis during pregnancy is independent of sialylation.

Rheumatoid arthritis (RA) is known to improve during pregnancy and to flare after delivery. Changes in the glycosylation of immunoglobulin G (IgG)'s fragment crystallizable (Fc) have been suggested to play a role herein. Recent animal studies indicate that not galactosylation but mainly sialylation is important in this respect. We aim to find new associations between IgG-Fc N-glycosylation and improvement of RA during pregnancy and the flare after delivery. Sera of RA patients (n = 251 pregnancies) and healthy controls (n = 32), all participating in a prospective cohort study on RA and pregnancy (PARA study), were collected before conception, during pregnancy, and after delivery. Using a recently developed fast and robust nanoRP-HPLC-sheath-flow-ESI-MS method the glycosylation of IgG Fc-glycopeptides was measured in a subclass specific manner, with relative standard deviations of <4% for the 8 most abundant IgG Fc glycopeptides during the entire measurement period of over 3 weeks. In patients and controls, several glycosylation changes were observed during pregnancy. In depth analysis of the association of these glycosylation changes with disease activity revealed that galactosylation, independent of sialylation, is associated with improvement of RA during pregnancy. Functional studies in human cell systems should be performed to obtain more insight into this matter.

Coupling porous sheathless interface MS with transient-ITP in neutral capillaries for improved sensitivity in glycopeptide analysis.

IgG antibodies are modulated in their function by the specific structure of the N-glycans attached to their Fc (fragment crystallizable) portions. However, the glycosylation analysis of antigen-specific IgGs is a challenging task as antibody levels to a given antigen only represent a fraction of the total IgG levels. Here, we investigated the use of a transient-ITP (t-ITP)--MS method for highly sensitive IgG1 glycosylation profiling as a complementary method to a high-throughput nano-RPLC-MS method. It was found that t-ITP-CZE using neutrally coated separation capillaries with a large volume injection (37% of capillary volume) and interfaced to MS with a sheathless porous sprayer yielded a 40-fold increase in sensitivity for IgG1 Fc glycopeptide analysis when compared to the conventional strategy. Furthermore, the glycoform profiles found with the t-ITP-CZE strategy were comparable to those from nano-RPLC-MS. In conclusion, the use of the highly sensitive t-ITP-CZE-MS method will provide information on IgG Fc glycosylation for those samples with IgG1 concentrations below the LODs of the conventional method.

High-throughput work flow for IgG Fc-glycosylation analysis of biotechnological samples.

Immunoglobulin G (IgG) fragment crystallizable (Fc) glycosylation is crucial for antibody effector functions such as antibody-dependent cellular cytotoxicity and complement-dependent cytotoxicity. To monitor IgG Fc glycosylation, high-throughput techniques for glycosylation analysis are needed in the biotechnology industry. Here we describe the development of a fully automated high-throughput method based on glycopeptide analysis. Samples are prepared in 96-well plates. The IgG's are purified directly from fermentation broths by means of immobilized protein A followed by trypsin digestion. Glycopeptides are purified by hydrophilic interaction solid-phase extraction and analyzed by electrospray mass spectrometry in the positive-ion mode. Data are automatically processed and relative intensities of the various IgG glycopeptides are obtained. The intermediate precision of the method is below 5% for the five major glycoforms of an IgG1 antibody. The newly developed method is suitable for glycosylation profiling of IgG's from fermentation broths. We compared the developed method to other glycoanalytical methods and successfully applied it to analyze the fermentation time course of two different clones of the same therapeutic antibody.

Comparison of the Fc glycosylation of fetal and maternal immunoglobulin G.

Human immunoglobulin G (IgG) molecules are composed of two Fab portions and one Fc portion. The glycans attached to the Fc portions of IgG are known to modulate its biological activity as they influence interaction with both complement and various cellular Fc receptors. IgG glycosylation changes significantly with pregnancy, showing a vast increase in galactosylation and sialylation and a concomitant decrease in the incidence of bisecting GlcNAc. Maternal IgGs are actively transported to the fetus by the neonatal Fc receptor (FcRn) expressed in syncytiotrophoblasts in the placenta, providing the fetus and newborn with immunological protection. Two earlier reports described significant differences in total glycosylation between fetal and maternal IgG, suggesting a possible glycosylation-selective transport via the placenta. These results might suggest an alternative maternal transport pathway, since FcRn binding to IgG does not depend on Fc-glycosylation. These early studies were performed by releasing N-glycans from total IgG. Here, we chose for an alternative approach analyzing IgG Fc glycosylation at the glycopeptide level in an Fc-specific manner, providing glycosylation profiles for IgG1 and IgG4 as well as combined Fc glycosylation profiles of IgG2 and 3. The analysis of ten pairs of fetal and maternal IgG samples revealed largely comparable Fc glycosylation for all the analyzed subclasses. Average levels of galactosylation, sialylation, bisecting GlcNAc and fucosylation were very similar for the fetal and maternal IgGs. Our data suggest that the placental IgG transport is not Fc glycosylation selective.

Fc-glycosylation of IgG1 is modulated by B-cell stimuli.

We have recently shown that IgG1 directed against antigens thought to be involved in the pathogenesis of rheumatoid arthritis harbor different glycan moieties on their Fc-tail, as compared with total sera IgG1. Given the crucial roles of Fc-linked N-glycans for the structure and biological activity of IgG, Fc-glycosylation of antibodies is receiving considerable interest. However, so far little is known about the signals and factors that could influence the composition of these carbohydrate structures on secreted IgG produced by B lymphocytes. Here we show that both "environmental" factors, such as all-trans retinoic acid (a natural metabolite of vitamin A), as well as factors stimulating the innate immune system (i.e. CpG oligodeoxynucleotide, a ligand for toll-like receptor 9) or coming from the adaptive immune system (i.e. interleukin-21, a T-cell derived cytokine) can modulate IgG1 Fc-glycosylation. These factors affect Fc-glycan profiles in different ways. CpG oligodeoxynucleotide and interleukin-21 increase Fc-linked galactosylation and reduce bisecting N-acetylglucosamine levels, whereas all-trans retinoic acid significantly decreases galactosylation and sialylation levels. Moreover, these effects appeared to be stable and specific for secreted IgG1 as no parallel changes of the corresponding glycans in the cellular glycan pool were observed. Interestingly, several other cytokines and molecules known to affect B-cell biology and antibody production did not have an impact on IgG1 Fc-coupled glycan profiles. Together, these data indicate that different stimuli received by B cells during their activation and differentiation can modulate the Fc-linked glycosylation of secreted IgG1 without affecting the general cellular glycosylation machinery. Our study, therefore, furthers our understanding of the regulation of IgG1 glycosylation at the cellular level.

MALDI-TOF-MS analysis of sialylated glycans and glycopeptides using 4-chloro-α-cyanocinnamic acid matrix.

For MALDI analysis of glycans and glycopeptides, the choice of matrix is crucial in minimizing desialylation by mass spectrometric in-source and metastable decay. Here, we evaluated the potential of 4-chloro-α-cyanocinnamic acid (Cl-CCA) for MALDI-TOF-MS analysis of labile sialylated tryptic N-glycopeptides and released N- and O-glycans. Similar to DHB, but in contrast to CHCA, the Cl-CCA matrix allowed the analysis of sialylated N-glycans and glycopeptides in negative ion mode MALDI-TOF-MS. Dried droplet preparations of Cl-CCA provided microcrystals with a homogeneous spatial distribution and high shot-to-shot repeatability similar to CHCA, which simplified the automatic measurement and improved the resolution and mass accuracy. Interestingly, reflectron-positive ion mode analysis of 1-phenyl-3-methyl-5-pyrazolone (PMP)-labeled O-glycans with Cl-CCA revealed more complete profiles than with DHB and CHCA. In conclusion, we clearly demonstrate the high potential of this rationally designed matrix for glycomics and glycoproteomics.

IgG fc N-glycosylation changes in Lambert-Eaton myasthenic syndrome and myasthenia gravis.

N-glycosylation of the immunoglobulin Fc moiety influences its biological activity by, for example, modulating the interaction with Fc receptors. Changes in IgG glycosylation have been found to be associated with various inflammatory diseases. Here we evaluated for the first time IgG Fc N-glycosylation changes in well-defined antibody-mediated autoimmune diseases, that is, the neurological disorders Lambert-Eaton myasthenic syndrome and myasthenia gravis, with antibodies to muscle nicotinic acetylcholine receptors or muscle-specific kinase. IgGs were purified from serum or plasma by protein A affinity chromatography and digested with trypsin. Glycopeptides were purified and analyzed by MALDI-FTICR-MS. Glycoform distributions of both IgG1 and IgG2 were determined for 229 patients and 56 controls. We observed an overall age and sex dependency of IgG Fc N-glycosylation, which was in accordance with literature. All three disease groups showed lower levels of IgG2 galactosylation compared to controls. In addition, LEMS patients showed lower IgG1 galactosylation. Notably, the galactosylation differences were not paralleled by a difference in IgG sialylation. Moreover, the level of IgG core-fucosylation and bisecting N-acetylglucosamine were evaluated. The control and disease groups revealed similar levels of IgG Fc core-fucosylation. Interestingly, LEMS patients below 50 years showed elevated levels of bisecting N-acetylglucosamine on IgG1 and IgG2, demonstrating for the first time the link of changes in the level of bisecting N-acetylglucosamine with disease.

Cotton HILIC SPE microtips for microscale purification and enrichment of glycans and glycopeptides.

Solid-phase extraction microtips are important devices in modern bioanalytics, as they allow miniaturized sample preparation for mass spectrometric analysis. Here we introduce the use of cotton wool for the preparation of filter-free HILIC SPE microtips. To this end, pieces of cotton wool pads (approximately 500 µg) were packed into 10 µL pipet tips. The performance of the tips was evaluated for microscale purification of tryptic IgG Fc N-glycopeptides. Cotton wool HILIC SPE microtips allowed the removal of salts, most nonglycosylated peptides, and detergents such as SDS from glycoconjugate samples. MALDI-TOF-MS glycopeptide profiles were very repeatable with different tips as well as reused tips, and very similar profiles were obtained with different brands of cotton wool pads. In addition, we used cotton HILIC microtips to purify N-glycans after N-glycosidase F treatment of IgG and transferrin followed by MALDI-TOF-MS detection. In conclusion, we establish cotton wool microtips for glycan and glycopeptide purification with subsequent mass spectrometric detection.

Immunoglobulin G glycopeptide profiling by matrix-assisted laser desorption ionization Fourier transform ion cyclotron resonance mass spectrometry.

Immunoglobulin G (IgG) fragment crystallizable (Fc) glycosylation is essential for Fc-receptor-mediated activities. Changes in IgG Fc glycosylation have been found to be associated with various diseases. Here we describe a high-throughput IgG glycosylation profiling method. Sample preparation is performed in 96-well plate format: IgGs are purified from 2 microL of human plasma using immobilized protein A. IgGs are cleaved with trypsin, and the resulting glycopeptides are purified by reversed-phase or hydrophilic interaction solid-phase extraction. Glycopeptides are analyzed by intermediate pressure matrix-assisted laser desorption ionization Fourier transform ion cyclotron resonance mass spectrometry (MALDI-FTICR-MS). Notably, both dihydroxybenzoic acid (DHB) and alpha-cyano-4-hydroxycinnamic acid (CHCA) matrixes allowed the registration of sialylated as well as nonsialylated glycopeptides. Data were automatically processed, and IgG isotype-specific Fc glycosylation profiles were obtained. The entire method showed an interday variation below 10% for the six major glycoforms of both IgG1 and IgG2. The method was found suitable for isotype-specific high-throughput IgG glycosylation profiling from human plasma. As an example we successfully applied the method to profile the IgG glycosylation of 62 human samples.

A strategy to incorporate prior knowledge into correlation network cutoff selection.

Correlation networks are frequently used to statistically extract biological interactions between omics markers. Network edge selection is typically based on the statistical significance of the correlation coefficients. This procedure, however, is not guaranteed to capture biological mechanisms. We here propose an alternative approach for network reconstruction: a cutoff selection algorithm that maximizes the overlap of the inferred network with available prior knowledge. We first evaluate the approach on IgG glycomics data, for which the biochemical pathway is known and well-characterized. Importantly, even in the case of incomplete or incorrect prior knowledge, the optimal network is close to the true optimum. We then demonstrate the generalizability of the approach with applications to untargeted metabolomics and transcriptomics data. For the transcriptomics case, we demonstrate that the optimized network is superior to statistical networks in systematically retrieving interactions that were not included in the biological reference used for optimization.

Systematic Evaluation of Normalization Methods for Glycomics Data Based on Performance of Network Inference.

Glycomics measurements, like all other high-throughput technologies, are subject to technical variation due to fluctuations in the experimental conditions. The removal of this non-biological signal from the data is referred to as normalization. Contrary to other omics data types, a systematic evaluation of normalization options for glycomics data has not been published so far. In this paper, we assess the quality of different normalization strategies for glycomics data with an innovative approach. It has been shown previously that Gaussian Graphical Models (GGMs) inferred from glycomics data are able to identify enzymatic steps in the glycan synthesis pathways in a data-driven fashion. Based on this finding, here, we quantify the quality of a given normalization method according to how well a GGM inferred from the respective normalized data reconstructs known synthesis reactions in the glycosylation pathway. The method therefore exploits a biological measure of goodness. We analyzed 23 different normalization combinations applied to six large-scale glycomics cohorts across three experimental platforms: Liquid Chromatography - ElectroSpray Ionization - Mass Spectrometry (LC-ESI-MS), Ultra High Performance Liquid Chromatography with Fluorescence Detection (UHPLC-FLD), and Matrix Assisted Laser Desorption Ionization - Furier Transform Ion Cyclotron Resonance - Mass Spectrometry (MALDI-FTICR-MS). Based on our results, we recommend normalizing glycan data using the 'Probabilistic Quotient' method followed by log-transformation, irrespective of the measurement platform. This recommendation is further supported by an additional analysis, where we ranked normalization methods based on their statistical associations with age, a factor known to associate with glycomics measurements.