Community evaluation of glycoproteomics informatics solutions reveals high-performance search strategies for serum glycopeptide analysis.

Glycoproteomics is a powerful yet analytically challenging research tool. Software packages aiding the interpretation of complex glycopeptide tandem mass spectra have appeared, but their relative performance remains untested. Conducted through the HUPO Human Glycoproteomics Initiative, this community study, comprising both developers and users of glycoproteomics software, evaluates solutions for system-wide glycopeptide analysis. The same mass spectrometrybased glycoproteomics datasets from human serum were shared with participants and the relative team performance for N- and O-glycopeptide data analysis was comprehensively established by orthogonal performance tests. Although the results were variable, several high-performance glycoproteomics informatics strategies were identified. Deep analysis of the data revealed key performance-associated search parameters and led to recommendations for improved 'high-coverage' and 'high-accuracy' glycoproteomics search solutions. This study concludes that diverse software packages for comprehensive glycopeptide data analysis exist, points to several high-performance search strategies and specifies key variables that will guide future software developments and assist informatics decision-making in glycoproteomics.

Exploiting the Disialyl Galactose Activity of α2,6-Sialyltransferase from Photobacterium damselae To Generate a Highly Sialylated Recombinant α-1-Antitrypsin.

Sialic acids are sugars present in many animal glycoproteins and are of particular interest in biopharmaceuticals, where a lack of sialylation can reduce bioactivity. Here, we describe how α-2,6-sialyltransferase from Photobacterium damselae can be used to markedly increase the level of sialylation of CHO-produced α-1-antitrypsin. Detailed analysis of the sialylation products showed that in addition to the expected α-2,6-sialylation of galactose, a second disialyl galactose motif Neu5Ac-α2,3(Neu5Ac-α2,6)Gal was produced, which, to our knowledge, had never been detected on a mammalian glycoprotein. We exploited this disialyl galactose activity of the P. damselae in a multienzyme reaction to produce a highly sialylated α-1-antitrypsin. The influence of this unique disialylation on the in vitro activity of α-1-antitrypsin was studied, and a toolkit of mass spectrometry methods for identifying this new disialyl galactose motif in complex mixtures was developed.

Utility of Ion-Mobility Spectrometry for Deducing Branching of Multiply Charged Glycans and Glycopeptides in a High-Throughput Positive ion LC-FLR-IMS-MS Workflow.

High-throughput glycan analysis has become an important part of biopharmaceutical production and quality control. However, it is still a significant challenge in the field of glycomics to easily deduce isomeric glycan structures, especially in a high-throughput manner. Ion mobility spectrometry (IMS) is an excellent tool for differentiating isomeric glycan structures. However, demonstrations of the utility of IMS in high-throughput workflows such as liquid chromatography-fluorescence-mass spectrometry (LC-FLR-MS) workflows have been limited with only a small amount of collision cross section (CCS) data available. In particular, IMS data of glycan fragments obtained in positive ion mode are limited in comparison to those obtained in negative ion mode despite positive ion mode being widely used for glycomics. Here, we describe IMS TWCCSN2 data obtained from a high-throughput LC-FLR-IMS-MS workflow in positive ion mode. We obtained IMS data from a selection of RapiFluor-MS (RFMS) labeled N-glycans and also glycopeptides. We describe how IMS is able to distinguish isomeric N-glycans and glycopeptides using both intact IMS and fragment-based IMS glycan sequencing experiments in positive ion mode, without significantly altering the high-throughput nature of the analysis. For the first time, we were able to successfully use IMS in positive ion mode to determine the branching of isomeric glycopeptides and RFMS labeled glycans. Further, we highlight that IMS glycan sequencing of fragments obtained from RFMS labeled glycans was similar to that of glycopeptides. Finally, we show that the IMS glycan sequencing approach can highlight shared structural features of nonisomeric glycans in a high-throughput LC-FLR-IMS-MS workflow.

Cysteine Aminoethylation Enables the Site-Specific Glycosylation Analysis of Recombinant Human Erythropoietin using Trypsin.

Recombinant human erythropoietin (rhEPO) is an important biopharmaceutical for which glycosylation is a critical quality attribute. Therefore, robust analytical methods are needed for the in-depth characterization of rhEPO glycosylation. Currently, the protease GluC is widely established for the site-specific glycosylation analysis of rhEPO. However, this enzyme shows disadvantages, such as its specificity and the characteristics of the resulting (glyco)peptides. The use of trypsin, the gold standard protease in proteomics, as the sole protease for rhEPO is compromised, as no natural tryptic cleavage site is located between the glycosylation sites Asn24 and Asn38. Here, cysteine aminoethylation using 2-bromoethylamine was applied as an alternative alkylation strategy to introduce artificial tryptic cleavage sites at Cys29 and Cys33 in rhEPO. The (glyco)peptides resulting from a subsequent digestion using trypsin were analyzed by reverse-phase liquid chromatography-mass spectrometry. The new trypsin-based workflow was easily implemented by adapting the alkylation step in a conventional workflow and was directly compared to an established approach using GluC. The new method shows an improved specificity, a significantly reduced chromatogram complexity, allows for shorter analysis times, and simplifies data evaluation. Furthermore, the method allows for the monitoring of additional attributes, such as oxidation and deamidation at specific sites in parallel to the site-specific glycosylation analysis of rhEPO.

GlycanAnalyzer: software for automated interpretation of N-glycan profiles after exoglycosidase digestions.

SUMMARY: Many eukaryotic proteins are modified by N-glycans. Liquid chromatography (ultra-performance -UPLC and high-performance-HPLC) coupled with mass spectrometry (MS) is conventionally used to characterize N-glycan structures. Software can automatically assign glycan structures by matching their observed retention times and masses with standardized values in reference databases. However, more precise confirmation of N-glycan structures can be derived using exoglycosidases, enzymes that remove specific monosaccharides from glycans. Exoglycosidase removal of monosaccharides results in signature peak shifts, in both UPLC and MS1, yielding an effective way to verify N-glycan structure with high detail (down to the position and isomeric linkage of each monosaccharide). Because manual interpretation of exoglycosidase data is complex and time consuming, we developed GlycanAnalyzer, a web application that pattern matches N-glycan peak shifts following exoglycosidase digestion and automates structure assignments. GlycanAnalyzer significantly improves assignment accuracy over other auto-assignment methods on tests with a monoclonal antibody and four glycan standards (100% versus 82% for the next best software). By automating data interpretation, GlycanAnalyzer enables the easier use of exoglycosidases to precisely define N-glycan structure. AVAILABILITY AND IMPLEMENTATION: http://glycananalyzer.neb.com. Datasets available online. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Clustering and curation of electropherograms: an efficient method for analyzing large cohorts of capillary electrophoresis glycomic profiles for bioprocessing operations.

The accurate assessment of antibody glycosylation during bioprocessing requires the high-throughput generation of large amounts of glycomics data. This allows bioprocess engineers to identify critical process parameters that control the glycosylation critical quality attributes. The advances made in protocols for capillary electrophoresis-laser-induced fluorescence (CE-LIF) measurements of antibody N-glycans have increased the potential for generating large datasets of N-glycosylation values for assessment. With large cohorts of CE-LIF data, peak picking and peak area calculations still remain a problem for fast and accurate quantitation, despite the presence of internal and external standards to reduce misalignment for the qualitative analysis. The peak picking and area calculation problems are often due to fluctuations introduced by varying process conditions resulting in heterogeneous peak shapes. Additionally, peaks with co-eluting glycans can produce peaks of a non-Gaussian nature in some process conditions and not in others. Here, we describe an approach to quantitatively and qualitatively curate large cohort CE-LIF glycomics data. For glycan identification, a previously reported method based on internal triple standards is used. For determining the glycan relative quantities our method uses a clustering algorithm to 'divide and conquer' highly heterogeneous electropherograms into similar groups, making it easier to define peaks manually. Open-source software is then used to determine peak areas of the manually defined peaks. We successfully applied this semi-automated method to a dataset (containing 391 glycoprofiles) of monoclonal antibody biosimilars from a bioreactor optimization study. The key advantage of this computational approach is that all runs can be analyzed simultaneously with high accuracy in glycan identification and quantitation and there is no theoretical limit to the scale of this method.

GlycopeptideGraphMS: Improved Glycopeptide Detection and Identification by Exploiting Graph Theoretical Patterns in Mass and Retention Time.

The leading proteomic method for identifying N-glycosylated peptides is liquid chromatography coupled with tandem fragmentation mass spectrometry (LCMS/MS) followed by spectral matching of MS/MS fragment masses to a database of possible glycan and peptide combinations. Such database-dependent approaches come with challenges such as needing high-quality informative MS/MS spectra, ignoring unexpected glycan or peptide sequences, and making incorrect assignments because some glycan combinations are equivalent in mass to amino acids. To address these challenges, we present GlycopeptideGraphMS, a graph theoretical bioinformatic approach complementary to the database-dependent method. Using the AXL receptor tyrosine kinase (AXL) as a model glycoprotein with multiple N-glycosylation sites, we show that those LCMS features that could be grouped into graph networks on the basis of glycan mass and retention time differences were actually N-glycopeptides with the same peptide backbone but different N-glycan compositions. Conversely, unglycosylated peptides did not exhibit this grouping behavior. Furthermore, MS/MS sequencing of the glycan and peptide composition of just one N-glycopeptide in the graph was sufficient to identify the rest of the N-glycopeptides in the graph. By validating the identifications with exoglycosidase cocktails and MS/MS fragmentation, we determined the experimental false discovery rate of identifications to be 2.21%. GlycopeptideGraphMS detected more than 500 unique N-glycopeptides from AXL, triple the number found by a database search with Byonic software, and detected incorrect assignments due to a nonspecific protease cleavage. This method overcomes some limitations of the database approach and is a step closer to comprehensive automated glycoproteomics.

Post-Column Make-Up Flow (PCMF) Enhances the Performance of Capillary-Flow PGC-LC-MS/MS-Based Glycomics.

Deep characterization of biologically relevant glycans remains challenging. Porous graphitized carbon-liquid chromatography tandem mass spectrometry (PGC-LC-MS/MS) enables the quantitative elucidation of glycan fine structures. However, the early PGC-LC elution of smaller glycans (tri-, tetra-, and pentasaccharides) at low organic solvent content hampers their detection. In efforts to improve the glycan profiling sensitivity and accuracy, we present a new capillary-flow PGC-LC-MS/MS-based configuration comprising a post-column make-up flow (PCMF) that supplies an ion-promoting organic solvent to separated glycans prior to their detection by MS. The analytical performance of this setup was systematically evaluated against our existing capillary-flow PGC-LC-MS/MS platform (Jensen et al., Nat. Protoc. 2012, 7, 1299). Specifically, the ion intensities and signal-to-noise ratios of various classes of nonderivatized glycans from N- and O-glycoproteins and fructooligosaccharide mixtures were compared using methanol (MeOH)-, isopropanol (IPA)-, and acetonitrile (ACN)-based PCMF at various concentrations. In particular, ACN- and IPA-based PCMF dramatically increased the signal response across all glycan types (30- to 100-fold), improved the MS/MS spectral quality, and reduced the quantitative glycoprofile variation between replicates. In particular, the detection of the early eluting glycans benefitted from the PCMF. The highest sensitivity gains were achieved with the supplements of 100% ACN and IPA (equating to 57% (v/v) net concentration at the ion source) while neither compromising the favorable PGC-LC properties including the high peak capacity and glycan isomer separation nor changing the MS detection behavior. In conclusion, PCMF-based PGC-LC-MS/MS dramatically improves the glycomics sensitivity, coverage, and quantitative accuracy not least for the difficult-to-detect early eluting and low-abundance glycans detached from N- and O-glycoproteins.

Combining Glucose Units, m/z, and Collision Cross Section Values: Multiattribute Data for Increased Accuracy in Automated Glycosphingolipid Glycan Identifications and Its Application in Triple Negative Breast Cancer.

Glycan head-groups attached to glycosphingolipids (GSLs) found in the cell membrane bilayer can alter in response to external stimuli and disease, making them potential markers and/or targets for cellular disease states. To identify such markers, comprehensive analyses of glycan structures must be undertaken. Conventional analyses of fluorescently labeled glycans using hydrophilic interaction high-performance liquid chromatography (HILIC) coupled with mass spectrometry (MS) provides relative quantitation and has the ability to perform automated glycan assignments using glucose unit (GU) and mass matching. The use of ion mobility (IM) as an additional level of separation can aid the characterization of closely related or isomeric structures through the generation of glycan collision cross section (CCS) identifiers. Here, we present a workflow for the analysis of procainamide-labeled GSL glycans using HILIC-IM-MS and a new, automated glycan identification strategy whereby multiple glycan attributes are combined to increase accuracy in automated structural assignments. For glycan matching and identification, an experimental reference database of GSL glycans containing GU, mass, and CCS values for each glycan was created. To assess the accuracy of glycan assignments, a distance-based confidence metric was used. The assignment accuracy was significantly better compared to conventional HILIC-MS approaches (using mass and GU only). This workflow was applied to the study of two Triple Negative Breast Cancer (TNBC) cell lines and revealed potential GSL glycosylation signatures characteristic of different TNBC subtypes.

GlycoStore: a database of retention properties for glycan analysis.

Summary: GlycoStore is a curated chromatographic, electrophoretic and mass-spectrometry composition database of N-, O-, glycosphingolipid (GSL) glycans and free oligosaccharides associated with a range of glycoproteins, glycolipids and biotherapeutics. The database is built on publicly available experimental datasets from GlycoBase developed in the Oxford Glycobiology Institute and then the National Institute for Bioprocessing Research and Training (NIBRT). It has now been extended to include recently published and in-house data collections from the Bioprocessing Technology Institute (BTI) A*STAR, Macquarie University and Ludger Ltd. GlycoStore provides access to approximately 850 unique glycan structure entries supported by over 8500 retention positions determined by: (i) hydrophilic interaction chromatography (HILIC) ultra-high performance liquid chromatography (U/HPLC) and reversed phase (RP)-U/HPLC with fluorescent detection; (ii) porous graphitized carbon (PGC) chromatography in combination with ESI-MS/MS detection; and (iii) capillary electrophoresis with laser induced fluorescence detection (CE-LIF). GlycoStore enhances many features previously available in GlycoBase while addressing the limitations of the data collections and model of this popular resource. GlycoStore aims to support detailed glycan analysis by providing a resource that underpins current workflows. It will be regularly updated by expert annotation of published data and data obtained from the project partners. Availability and implementation: http://www.glycostore.org. Supplementary information: Supplementary data are available at Bioinformatics online.

Improvement of electrospray stability in negative ion mode for nano-PGC-LC-MS glycoanalysis via post-column make-up flow.

Analysis of glycans via a porous graphitized carbon liquid chromatography (PGC-LC) coupled with electrospray ionization (tandem) mass spectrometry (ESI-MS(/MS)) is a powerful analytical method in the field of glycomics. Isobaric glycan structures can be identified reliably with the help of PGC-LC separation and subsequent identification by ESI-MS(/MS) in negative ion mode. In an effort to adapt PGC-LC-ESI-MS(/MS) to the nano-scale operation, spray instability along the nano-PGC-LC gradient was repeatedly observed on an LTQ Orbitrap Elite mass spectrometer equipped with a standard nano-electrospray ionization source. A stable electrospray was achieved with the implementation of a post-column make-up flow (PCMF). Thereby, acetonitrile was used to supplement the eluate from the nano-PGC-LC column. The improved spray stability enhanced detection and resolution of glycans during the analysis. This was in particular the case for smaller O-glycans which elute early in the high aqueous content regime of the nano-PGC-LC elution gradient. This study introduces PCMF as an easy-to-use instrumental adaptation to significantly improve spray stability in negative ion mode nano-PGC-LC-ESI-MS(/MS)-based analysis of glycans.

Glycan involvement in the adhesion of Pseudomonas aeruginosa to tears.

The human eye is constantly bathed by tears, which protect the ocular surface via a variety of mechanisms. The O-linked glycans of tear mucins have long been considered to play a role in binding to pathogens and facilitating their removal in the tear flow. Other conjugated glycans in tears could similarly contribute to pathogen binding and removal but have received less attention. In the work presented here we assessed the contribution of glycan moieties, in particular the protein attached N-glycans, presented by the broad complement of tear proteins to the adhesion of the opportunistic pathogen Pseudomonas aeruginosa, a leading cause of microbial keratitis and ulceration of the cornea. Our adhesion assay involved immobilising the macromolecular components of tears into the wells of a polyvinyl difluoride (PVDF) microtitre filter plate and probing the binding of fluorescently labelled bacteria. Three P. aeruginosa strains were studied: a cytotoxic strain (6206) and an invasive strain (6294) from eye infections, and an invasive strain (320) from a urinary tract infection (UTI). The ocular isolates adhered two to three times more to human tears than to human saliva or porcine gastric mucin, suggesting ocular niche-specific adaptation. Support for the role of the N-glycans carried by human tear proteins in the binding and removal of P. aeruginosa from the eye was shown by: 1) pre-incubation of the bacteria with free component sugars, galactose, mannose, fucose and sialyl lactose (or combination thereof) inhibiting adhesion of all the P. aeruginosa strains to the immobilised tear proteins, with the greatest inhibition of binding of the ocular cytotoxic 6206 and least for the invasive 6294 strain; 2) pre-incubation of the bacteria with N-glycans released from the commercially available human milk lactoferrin, an abundant protein that carries N-linked glycans in tears, inhibiting the adhesion to tears of the ocular bacteria by up to 70%, which was significantly more binding inhibition than by the same amount of intact human lactoferrin or by the plant-derived N-glycans released from the rice recombinant lactoferrin; 3) pre-incubation of the bacteria with N-linked glycans released from human tear proteins inhibiting the adhesion of the ocular P. aeruginosa strains to immobilised tear proteins; 4) inhibition by the N-glycans from lactoferrin of the ability of an ocular strain of P. aeruginosa to invade corneal epithelial cells; 5) removal of terminal sialic acid and fucose moieties from the tear glycoproteins with α2-3,6,8 neuraminidase (sialidase) and α1-2,3,4 fucosidase resulting in a reduction in binding of the UTI P. aeruginosa isolate, but not the adhesion of the ocular cytotoxic (6206) or invasive (6294) isolates. Glycosidase activity was validated by mass spectrometry. In all cases, the magnitude of inhibition of bacterial adhesion by the N-glycans was consistently greater for the cytotoxic ocular strain than for the invasive ocular strain. Ocular P. aeruginosa isolates seems to exhibit different adhesion mechanism than previously known PAI and PAII lectin adhesion. The work may contribute towards the development of glycan-focused therapies to prevent P. aeruginosa infection of the eye.

Identification of pathogenic factors potentially involved in Staphylococcus aureus keratitis using proteomics.

Staphylococcus is a leading cause of microbial keratitis, characterized by destruction of the cornea by bacterial exoproteins and host-associated factors. The aim of this study was to compare extracellular and cell-associated proteins produced by two different isolates of S. aureus, a virulent clinical isolate (Staph 38) and a laboratory strain (Staphylococcus aureus 8325-4) of weaker virulence in the mouse keratitis model. Proteins were analyzed using 2D polyacrylamide gel electrophoresis and identified by subsequent mass spectrometry. Activity of staphylococcal adhesins was assessed by allowing strains to bind to various proteins adsorbed onto polymethylmethacrylate squares. Thirteen proteins in the extracellular fraction and eight proteins in the cell-associated fractions after bacterial growth were produced in increased amounts in the clinical isolate Staph 38. Four of these proteins were S. aureus virulence factor adhesins, fibronectin binding protein A, staphopain, glyceraldehyde-3-phosphate dehydrogenase 2 and extracellular adherence protein. The clinical isolate Staph 38 adhered to a greater extent to all mammalian proteins tested, indicating the potential of the adhesins to be active on its surface. Other proteins with increased expression in Staph 38 included potential moonlighting proteins and proteins involved in transcription or translation. This is the first demonstration of the proteome of S. aureus isolates from keratitis. These results indicate that the virulent clinical isolate produces more potentially important virulence factors compared to the less virulent laboratory strain and these may be associated with the ability of a S. aureus strain to cause more severe keratitis.

Validation of the curation pipeline of UniCarb-DB: building a global glycan reference MS/MS repository.

The UniCarb-DB database is an emerging public glycomics data repository, containing over 500 tandem mass spectra (as of March 2013) of glycans released from glycoproteins. A major challenge in glycomics research is to provide and maintain high-quality datasets that will offer the necessary diversity to support the development of accurate bioinformatics tools for data deposition and analysis. The role of UniCarb-DB, as an archival database, is to provide the glycomics community with open-access to a comprehensive LC MS/MS library of N- and O- linked glycans released from glycoproteins that have been annotated with glycosidic and cross-ring fragmentation ions, retention times, and associated experimental metadata descriptions. Here, we introduce the UniCarb-DB data submission pipeline and its practical application to construct a library of LC-MS/MS glycan standards that forms part of this database. In this context, an independent consortium of three laboratories was established to analyze the same 23 commercially available oligosaccharide standards, all by using graphitized carbon-liquid chromatography (LC) electrospray ionization (ESI) ion trap mass spectrometry in the negative ion mode. A dot product score was calculated for each spectrum in the three sets of data as a measure of the comparability that is necessary for use of such a collection in library-based spectral matching and glycan structural identification. The effects of charge state, de-isotoping and threshold levels on the quality of the input data are shown. The provision of well-characterized oligosaccharide fragmentation data provides the opportunity to identify determinants of specific glycan structures, and will contribute to the confidence level of algorithms that assign glycan structures to experimental MS/MS spectra. This article is part of a Special Issue entitled: Computational Proteomics in the Post-Identification Era. Guest Editors: Martin Eisenacher and Christian Stephan.

Glycomic characterization of basal tears and changes with diabetes and diabetic retinopathy.

As a secreted fluid, the state of tear glycosylation is particularly important in the role of immunity of the ocular surface. Tears are a valuable source of non-invasive biomarkers for disease and there are continued efforts to characterize their components thoroughly. In this study, a small volume of basal tears (5 µL) was collected from healthy controls, patients with diabetes without retinopathy and patients with diabetes and retinopathy. The detailed N- and O-linked tear protein glycome was characterized and the relative abundance of each structure determined. Of the 50 N-linked glycans found, 89% were complex with 50% containing a bisecting N-acetylglucosamine, 65% containing a core fucose whilst 33% were sialylated. Of the 8 O-linked glycans detected, 3 were of cores 1 and 5 of core 2 type, with a majority of them being sialylated (90%). Additionally, these glycan structures were profiled across the three diabetic disease groups. Whilst the higher abundant structures did not alter across the three groups, only five low abundance N-linked glycans and 1 O-linked glycan did alter with the onset of diabetes mellitus and diabetic retinopathy (DR). These results suggest the conservation of glycan types on basal tear proteins between individuals and point to only small changes in glycan expression on the proteins in tears with the development of diabetes and DR.

Antibody glycan quality predicted from CHO cell culture media markers and machine learning.

N-glycosylation can have a profound effect on the quality of mAb therapeutics. In biomanufacturing, one of the ways to influence N-glycosylation patterns is by altering the media used to grow mAb cell expression systems. Here, we explore the potential of machine learning (ML) to forecast the abundances of N-glycan types based on variables related to the growth media. The ML models exploit a dataset consisting of detailed glycomic characterisation of Anti-HER fed-batch bioreactor cell cultures measured daily under 12 different culture conditions, such as changes in levels of dissolved oxygen, pH, temperature, and the use of two different commercially available media. By performing spent media quantitation and subsequent calculation of pseudo cell consumption rates (termed media markers) as inputs to the ML model, we were able to demonstrate a small subset of media markers (18 selected out of 167 mass spectrometry peaks) in a Chinese Hamster Ovary (CHO) cell cultures are important to model N-glycan relative abundances (Regression - correlations between 0.80-0.92; Classification - AUC between 75.0-97.2). The performances suggest the ML models can infer N-glycan critical quality attributes from extracellular media as a proxy. Given its accuracy, we envisage its potential applications in biomaufactucuring, especially in areas of process development, downstream and upstream bioprocessing.

Micro- and macroheterogeneity of N-glycosylation yields size and charge isoforms of human sex hormone binding globulin circulating in serum.

Human sex hormone binding globulin (hSHBG) is a serum glycoprotein central to the transport and targeted delivery of sex hormones to steroid-sensitive tissues. Several molecular mechanisms of action of hSHBG, including the function of its attached glycans remain unknown. Here, we perform a detailed site-specific characterization of the N- and O-linked glycosylation of serum-derived hSHBG. MS-driven glycoproteomics and glycomics combined with exoglycosidase treatment were used in a bottom-up and top-down manner to determine glycosylation sites, site-specific occupancies and monosaccharide compositions, detailed glycan structures, and the higher level arrangement of glycans on intact hSHBG. It was found that serum-derived hSHBG is N-glycosylated at Asn(351) and Asn(367) with average molar occupancies of 85.1 and 95.3%, respectively. Both sites are occupied by the same six sialylated and partly core fucosylated bi- and triantennary N-Glycoforms with lactosamine-type antennas of the form (±NeuAcα6)Galß4GlcNAc. N-Glycoforms of Asn(367) were slightly more branched and core fucosylated than Asn(351) N-glycoforms due probably to a more surface-exposed glycosylation site. The N-terminal Thr(7) was fully occupied by the two O-linked glycans NeuAcα3Galß3(NeuAcα6)GalNAc (where NeuAc is N-acetylneuraminic acid and GalNAc is N-acetylgalactosamine) and NeuAcα3Galß3GalNAc in a 1:6 molar ratio. Electrophoretic analysis of intact hSHBG revealed size and charge heterogeneity of the isoforms circulating in blood serum. Interestingly, the size and charge heterogeneity were shown to originate predominantly from differential Asn(351) glycan occupancies and N-glycan sialylation that may modulate the hSHBG activity. To date, this work represents the most detailed structural map of the heterogeneous hSHBG glycosylation, which is a prerequisite for investigating the functional aspects of the hSHBG glycans.

Harnessing the potential of machine learning for advancing "Quality by Design" in biomanufacturing.

Ensuring consistent high yields and product quality are key challenges in biomanufacturing. Even minor deviations in critical process parameters (CPPs) such as media and feed compositions can significantly affect product critical quality attributes (CQAs). To identify CPPs and their interdependencies with product yield and CQAs, design of experiments, and multivariate statistical approaches are typically used in industry. Although these models can predict the effect of CPPs on product yield, there is room to improve CQA prediction performance by capturing the complex relationships in high-dimensional data. In this regard, machine learning (ML) approaches offer immense potential in handling non-linear datasets and thus are able to identify new CPPs that could effectively predict the CQAs. ML techniques can also be synergized with mechanistic models as a 'hybrid ML' or 'white box ML' to identify how CPPs affect the product yield and quality mechanistically, thus enabling rational design and control of the bioprocess. In this review, we describe the role of statistical modeling in Quality by Design (QbD) for biomanufacturing, and provide a generic outline on how relevant ML can be used to meaningfully analyze bioprocessing datasets. We then offer our perspectives on how relevant use of ML can accelerate the implementation of systematic QbD within the biopharma 4.0 paradigm.