Article Comprehensive Profiling of Glycosphingolipid using a Novel Broad Specificity Endoglycoceramidase in a High-Throughput Workflow Simone Albrecht, Saulius Vainauskas, Henning Stoeckmann, Ciara McManus, Christopher H. Taron, and Pauline Mary Rudd Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.6b00259 • Publication Date (Web): 01 Apr 2016 Downloaded from http://pubs.acs.org on April 6, 2016

Just Accepted

“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties. Page 1 of 22 Analytical Chemistry

1 2 3 4 Comprehensive Profiling of Glycosphingolipid Glycans using a Novel Broad Specificity 5 6 Endoglycoceramidase in a High-Throughput Workflow 7 8 9 Simone Albrecht 1, Saulius Vainauskas 2, Henning Stöckmann 1, 10 11 Ciara McManus 1, Christopher H. Taron 2Δ, Pauline M. Rudd 1Δ* 12 13 14 1NIBRT GlycoScience Group, National Institute for Bioprocessing, Research and Training, Fosters 15 16 Avenue, Mount Merrion, Blackrock, Dublin 4, Ireland 17 18 19 2New England Biolabs, Ipswich, MA, USA 20 21 22 23 24 ΔChristopher Taron and Pauline Rudd share senior authorship 25 26 27 28 *To whom correspondence should be addressed: Pauline M. Rudd, NIBRT GlycoScience Group, 29 National Institute for Bioprocessing, Research and Training, Fosters Avenue, Mount Merrion, 30 31 Blackrock, Dublin 4, Ireland. Tel.: +353 12158 142; Fax: +353 12158 116; E-mail: 32 33 [email protected] 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 1

ACS Paragon Plus Environment Analytical Chemistry Page 2 of 22

1 2 3 Abstract 4 5 The biological function of glycosphingolipids (GSLs) is largely determined by their head 6 7 group moiety. This has placed a renewed emphasis on detailed GSL head group structural analysis. 8 Comprehensive profiling of GSL head groups in biological samples requires the use of 9 10 endoglycoceramidases with broad substrate specificity and a robust workflow that enables their 11 12 high-throughput analysis. We present here the first high-throughput glyco-analytical platform for 13 GSL head group profiling. The workflow features enzymatic release of GSL glycans with a novel 14 15 broad-specificity endoglycoceramidase I (EGCase I) from Rhodococcus triatomea , selective glycan 16 17 capture on hydrazide beads on a robotics platform, 2AB-fluorescent glycan labelling and analysis by 18 UPLC-HILIC-FLD. R. triatomea EGCase I displayed a wider specificity than known EGCases and was 19 20 able to efficiently hydrolyze , , (n)Lc-type GSLs and . Our 21 workflow was validated on purified GSL standard and was applied to the characterization of 22 23 GSLs extracted from several mammalian cell lines and human serum. This study should facilitate 24 25 the analytical workflow in functional glycomics studies and biomarker discovery. 26

27 28 Keywords: , ultra-performance hydrophilic interaction liquid chromatography (UPLC-HILIC), 29 30 endoglycoceramidase, glycan profiling, glycosphingolipid, glycomics, high-throughput 31 32 33 34 35 Abbreviations: GSL, glycosphingolipid; MODY, maturity-onset diabetes of the young; rEGCase, 36 recombinant endoglycoceramidase; EGALC, endogalactosylceramidase; Gal, galactose; Cer, 37 38 ; 2AB, 2-aminobenzamide; UPLC-HILIC-FLD, ultra performance liquid chromatography- 39 40 hydrophilic interaction-fluorescence detection; IPTG, isopropyl-β-thiogalactopyranoside; ACN, 41 acetonitrile; MeOH, methanol; GU, glucose units; QTOF, quadruploe time-of-flight; WAX, weak anion 42 43 exchange; ABS, A.ureafaciens α(2-3/6/8)-sialidase; NAN1, S. pneumoniae α(2-3)-sialidase; BKF, 44 45 bovine kidney α(1-2/4)-fucosidase; AMF, almond meal α(1-3/4)-fucosidase; BTG, bovine testes 46 β(1-3/4)-galactosidase; SPG, S. pneumoniae β(1-4)-galactosidase; CBG, coffee bean α(1-3/4)- 47 48 galactosidase; JBH, jack bean β(1-2/3/4/6) -N-acetylhexosaminidase; CV, coefficient of variance; 49 50 SSEA, stage specific embryonic antigen; HLB, hydrophilic-lipophilic balance; Neu5Ac/S, N- 51 acetylneuraminic acid; Neu5Gc/Sg, N-glycolylneuraminic acid; Fuc/F, fucose; Hex, hexose; HexNAc, 52 53 N-acetylhexosamine; GlcNAc, N-acetylglucosylamine. 54 55 56 57 58 59 60 2

ACS Paragon Plus Environment Page 3 of 22 Analytical Chemistry

1 2 3 Introduction 4 5 Many mammalian secretory proteins and lipids contain covalently linked (glycans). 6 7 For these molecules, the structure and composition of their appended glycans plays a significant 8 role in their function, distribution and physical properties 1. As glycan biosynthesis is not directly a 9 10 template-driven process, glycan structures on secreted glycoproteins and cell surface lipids are 11 12 typically heterogeneous and complex. Additionally, glycans are subject to structural alterations 13 over their lifetime due to both normal and pathological physiological processes, a feature that 14 15 makes them attractive as potential biomarkers of disease. 16 17 In the past few years, there have been major advances in high-throughput glycomics 18 technologies for glycoprotein analysis. A low-cost, high-throughput, automated N-glycan sample 19 20 preparation platform for glycoprofiling of immunoglobulins (IgG), antibodies and glycoproteins 21 isolated from serum was recently described by our laboratory 2-5. To date the analysis of N-glycans 22 23 has largely been based on chromatographic profiling using ultra-performance hydrophilic 24 25 interaction liquid chromatography with fluorescence detection (UPLC-HILIC-FLD). This technique 26 permits rapid and semi-quantitative comparison of N-glycan structures across many samples. 27 28 Serum N-glycan profiling has been used to identify glycan biomarkers of various diseases such as 29 30 cancer (ovarian, prostate, breast, lung, pancreatic, stomach), mature onset diabetes of the young 31 (MODY) , as well as biomarkers associated with normal physiological processes like aging 6-8. While 32 33 this young field has made significant advances, it is desirable to expand the breadth of these 34 35 profiling studies to other families of glycoproteins and other , for example, 36 . 37 38 One attractive class of molecule for glycoprofiling studies are glycosphingolipids (GSLs), 39 40 lipids that possess a head group consisting of mono- or oligosaccharides attached to 41 the lipids or ceramide. More than 500 structural species that differ in their head group 42 43 glycan or fatty acid composition are known 9. GSL glycan head groups are associated with many 44 45 cellular processes such as cell differentiation, signalling and receptor functions for viruses, 46 antibodies or lectins 10 . GSLs are ubiquitous on cell membranes and also circulate in serum where 47 48 they are present in a free form or in complex with proteins 11 . Aberrant GSL glycosylation has been 49 50 repeatedly reported for different types of cancer including lung, breast, prostate and ovarian cancer 51 as well as brain tumors, multiple sclerosis, rheumatoid arthritis and lysosomal storage diseases 52 53 such as Gaucher’s and Fabry disease 12-15 . 54 55 In contrast to N-glycan profiling, high-throughput analysis of GSLs is still in its infancy. 56 Although methods have been described for the automated extraction of GSL from their matrix and 57 58 their subsequent characterization en masse using shotgun lipidomics 16 no such automated methods 59 60 3

ACS Paragon Plus Environment Analytical Chemistry Page 4 of 22

1 2 3 are currently available for the preparation and chromatographic analysis of GSL glycans. 4 5 Enzymatically released GSL head groups are commonly analyzed using MALDI-TOF MS 17,18 . 6 7 Enzymes often play a critical role in glycomics workflows such as the enzymatic release of GSL 8 glycans by endoglycoceramidases (EGCases) which is preferred over harsh chemical release 9 10 methods that can result in low glycan yields 19,20 . However, for enzymatic release of GSL head 11 12 groups to be viable in a glycan profiling scheme, an enzyme with broad substrate specificity is 13 needed. To date very few EGCases have been tested for their ability to cleave a wide range of GSL 14 15 classes, limiting their utility for GSL head group profiling. 16 2,5 17 In the present study, we adapted our existing robotized N-glycan analysis platform for the 18 quantitative high-throughput profiling of 2AB-labeled mammalian GSL head groups using UPLC- 19 20 HILIC-FLD. An enabling component of our GSL glycan workflow is our identification and 21 characterization of a novel recombinant EGCase I enzyme from Rhodocococcus triatomea that 22 23 exhibits a broad GSL specificity, including the release of globo-series GSLs and Gal(β1-1)Cer, 24 25 important classes of GSLs that are not efficiently released by known EGCases. Finally, we 26 demonstrate the ability to characterize GSL head groups from both mammalian cell surfaces and 27 28 small volumes of blood serum. This analytical workflow will permit further exploration of the GSL 29 30 head group repertoire of GSLs from a broad range of biological sources and will enable studies 31 aiming to identify cellular or serum GSL-glycan biomarkers of disease. 32 33 34 35 Experimental Section 36 Chemical reagents and solvents were from Sigma-Aldrich. Glycosphingolipid standards were from 37 38 Sigma-Aldrich (GD1a, GD1b, GM1a, GT1b, GalCer, , Psychosine), Avanti (GD3, GlcCer) and 39 40 Matreya (FGM1, GM3, Gb4, LacCer). Human serum was pooled from apparently healthy donors 41 (courtesy of the U.K. Blood Transfusion Service). Recombinant Rhodococcus equi 42 43 endoglycoceramidase (rEGCase II) was from Takara Bio Inc. 44 45 46 Cell culture 47 48 HeLa (ATCC # CCL-2), NIH/3T3 (ATCC # CRL-1658), and HL60 (ATCC # CCL-240) cell lines were 49 50 obtained from the American Type Culture Collection (ATCC). HeLa and NIH/3T3 cells were cultured 51 in DMEM (Thermo Scientific HyClone) containing 10% (v/v) fetal bovine serum (FBS), 2 mM L- 52 53 glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin for 2 days at 37°C, then trypsinized 54 55 and collected by centrifugation. Cells were washed with cold 1X PBS and frozen at –80°C. HL60 56 cells were cultured in Iscove's Modified Dulbecco's Medium (Thermo Scientific HyClone) containing 57 58 59 60 4

ACS Paragon Plus Environment Page 5 of 22 Analytical Chemistry

1 2 3 10% (v/v) FBS, 2 mM L-glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin for 3 days at 4 5 37°C in suspension. Collected cells were washed with cold 1X PBS buffer and frozen at –80°C. 6 7 8 Cloning, expression and purification of Rhodococcus triatomea EGCase I 9 10 A gene encoding putative EGCase I (GenBank accession: EME18930.1) was identified in the R. 11 21 12 triatomea genome database by BLASTP and TBLASTN using the amino acid sequences of EGCase I 13 (R. equi ), EGCase II ( R. equi ) or EGALC ( R. equi ) as queries. A codon optimized sequence encoding 14 15 the R. triatomea EGCase I ORF with a C-terminal His-tag was synthesized by Genscript. A version of 16 17 R. triatomea EGCase I (27-489 aa) lacking its signal peptide and having a C-terminal His-tag was 18 created by PCR and cloned into the NdeI and XbaI sites of pJS119 22 to produce pJS119/RhtrECI. 19 20 Recombinant protein expression was performed in E. coli NEB Express cells (New England Biolabs) 21 carrying pJS119/RhtrECI followed by purity and activity assay of the enzyme (see the Experimental 22 23 Section in Supporting Information). 24 25 26 Extraction and purification of GSLs from cell membranes 27 28 A method adapted from Smith et al .23 was used for extraction of GSLs from cell membranes. Lipids 29 30 were extracted from aliquots of 2 x 10 5 cells. A mix of chloroform/methanol/H 2O (1:2:0.75; v/v/v) 31 was added, sonicated for 10 min and centrifuged (14,000 x g, 2 min). The supernatant was collected 32 33 and the step was repeated once. Mixtures of chloroform/methanol (1:1; v/v) and 34 35 chloroform/methanol (2:1; v/v) were sequentially added and sonicated for 10 min, after which 36 samples were centrifuged and the supernatants collected and pooled after each step. Pooled 37 38 supernatants were dried using a SpeedVac (Thermo Scientific). The dried extract was subjected to 39 40 n-butanol/water partitioning according to Vidugiriene et al .24 . The crude extract was 41 resuspended in n-butanol/water (1:1; v/v), vortexed and centrifuged at 1,000 x g for 2 min. The 42 43 organic phase (upper) was collected and back-extracted with n-butanol-saturated H 2O and the 44 45 aqueous phase (lower) was re-extracted with H 20-saturated n-butanol followed by vortexing and 46 centrifugation at 1000 x g for 2 min. The pooled butanol-phases were dried in a SpeedVac. 47 48 49 50 Extraction and purification of GSLs from serum 51 GSLs from serum were extracted using a quick protein precipitation method adapted from Huang et 52 53 al 25 . Briefly, 180 μL of methanol were added to aliquots of 20 μL human serum followed by 54 55 vortexing and centrifugation at 14,000 x g for 30 min at 4°C. The supernatants were dried in a 56 SpeedVac and subjected to n-butanol partitioning as described above. 57 58 59 60 5

ACS Paragon Plus Environment Analytical Chemistry Page 6 of 22

1 2 3 Endoglycoceramidase digestion 4 5 Dried GSLs, whether commercial standards or extracted and purified from cells or serum, were re- 6 7 suspended in sodium actete buffer (pH 5.2) containing 1 mg/mL Triton-X-100 and incubated at 8 37°C for 16 h using different concentrations (2 mU – 120 mU) of R. triatomea rEGCase I or R. equi 2 9 10 mU of rEGCase II in a total volume of 20-200 μL. 11 12 13 14 15 Automated hydrazide-mediated glycan head group cleanup 16 17 Enzymatically released glycan head groups were captured and cleaned using a Hamilton Robotics 18 StarLet liquid-handling platform in an automated method adapted from Stöckmann et al.2. (see the 19 20 Experimental Section in Supporting Information). 21

22 23 2AB labeling and cleanup of excess label 24 25 Fluorescent labeling mix (5 μL; 350 mM 2-aminobenzamide (2AB), 1 M sodium cyanoborohydride 26 in acetic acid/dimethyl sulfoxide (30:70)) was dispensed into each sample plate well, and the plate 27 28 was incubated at 65°C with agitation at 700 rpm for 120 min. Excess label was removed by paper 29 30 chromatography in a 96-well plate according to Royle et al .26 . 31 32 33 UPLC-HILIC-FLD and UPLC-HILIC-FLD-MS 34 35 2AB-labeled glycans were analyzed by UPLC-HILIC-FLD/MS using a method adapted from Albrecht 36 et al .27 (see the Experimental Section in Supporting Information) 37 38 39 40 Safety Considerations 41 Chloroform is toxic and needs to be handled only by experienced and well-trained personal using 42 43 all safety laboratory measures possible (gloves, glasses, and fume hood). 44 45 46 Results and Discussion 47 48 49 50 Semi-automated GSL head group sample preparation and analysis 51 A semi-automated sample preparation workflow was established for GSL glycan head group 52 53 analysis (Figure 1). Our approach was inspired by a robust IgG N-glycan preparation workflow that 54 2 55 was recently reported by our laboratory . The adapted workflow for GSL head group analysis 56 utilizes endoglycoceramidase release of GSL head group glycans followed by selective glycan 57 58 capture on solid-supported hydrazide beads, high-throughput 2AB glycan labeling and cleanup by 59 60 6

ACS Paragon Plus Environment Page 7 of 22 Analytical Chemistry

1 2 3 HyperSep Diol SPE. Simultaneous preparation of 96 samples can be achieved using a liquid- 4 5 handling robotic workstation ahead of GSL head group glycan analysis by UPLC-HILIC-FLD or 6 7 UPLC-HILIC-FLD-MS. 8 The amphipathic nature of GSLs requires the use of an enzyme reaction buffer containing detergent 9 10 to permit the optimal release of head group glycans by endoglycoceramidase digestion. This 11 12 detergent needed to be removed prior to subsequent sample processing steps and glycan analysis. 13 Prior studies reported the use of “glycoblotting” 28 , a method where released glycans are transiently 14 15 bound to hydrazide beads, to isolate GSL head groups after endoglycoceramidase digestion 17,18 . 16 17 Therefore, we adapted an automated high-recovery version of glycoblotting used in our IgG N- 18 glycan preparation workflow 2 for use in our GSL glycan profiling workflow. The glycan capturing 19 20 step was performed using low-cost UltraLink hydrazide resin and was conducted on a temperature- 21 controlled robotic heater-shaker with vigorous agitation. 22 23 Additional considerations were made for labeling GSL head group glycans with a fluorescent dye. 24 25 We used the dye 2-aminobenzamide (2-AB) which is widely used for mole-based glycan 26 derivatization allowing for sensitive and quantitative analysis of reducing glycans by HILIC-FLD 27 28 chromatography. Although 2-AB labeling is broadly applicable to any reducing glycan species, 29 30 different approaches are required for the removal of excess labeling dye that largely depend on the 31 nature of the glycans being labeled. Compared to N-glycans that have a common pentasaccharide 32 33 core (i.e. the chitobiose core extended by three mannoses), GSL head groups are structurally more 34 35 diverse and range from monosaccharides to large sialylated oligosaccharides. 2-AB cleanup using 36 normal-phase SPE, which is routinely used for N-glycans, resulted in insufficient recovery of small 37 38 glycan structures. Therefore a more universal and gentle cleanup method based on paper 39 40 chromatography was used instead. This method was previously adapted for processing in a 96-well 41 plate format to ensure efficient sample-throughput 26 . 42 43 44 45 High-throughput characterization of on defined GSL substrates 46 Using our GSL glycan analysis workflow the substrate specificity of a novel recombinant R. 47 48 triatomea EGCase I (see the Results Section and Figure S-1 in the Supporting Information) was 49 50 evaluated on isolated GSL standards and experimentally compared to that of a recombinant 51 commercial EGCase II from R. equi (Table 1). A theoretical comparison with literature findings for 52 53 the specificity of the more closely related R.equi rEGCase I was included 9. Aliquots of 2 nmol 54 55 standard GSL substrates including cerebrosides (glucosylceramide [GlcCer], 56 [GalCer]), [LacCer], gangliosides (GM3, GM1a, FGM1, GD1a, GD1b, GD3 and GT1b), 57 58 globosides (Gb4), sulfatide and psychosine were incubated with either 2 mU EGCase I or EGCase II 59 60 7

ACS Paragon Plus Environment Analytical Chemistry Page 8 of 22

1 2 3 at 37°C for 16 h in 50 mM sodium acetate buffer containing 0.1% Triton-X. The reaction was 4 5 performed directly on a 96-well sample plate that was subsequently used for automated cleanup, 6 7 minimizing the number of sample transfer steps. An internal standard (ISTD) was added for 8 relative UPLC-HILIC-FLD glycan quantitation and multiple replicates (n = 2-4) were performed. 9 10 Chromatographic peak retention times of the GSL head groups were converted into glucose unit 11 12 (GU) values using a 2AB-labeled dextran ladder that enables the independent comparison of 13 chromatographic profiles26 . Completeness of digestion was tested by complementary TLC assay 14 15 (see the Experimental Section in Supporting Information and Table S-2). 16 17 The coefficients of variance (CV) for the relative GSL head group quantitation after 18 endoglycoceramidase release and 2AB-labeling were ≤ 10% for each GSL standard/enzyme pair, 19 20 highlighting the reproducibility of the workflow as well as the defined substrate-specificity of the 21 endoglycoceramidases tested. Overall, R. triatomea rEGCase I showed very broad and robust GSL 22 23 hydrolyzing activity. The release efficiency of rEGCase I was clearly superior to rEGCase II for the 24 25 fucosylated FGM1 (1:0.4; rEGCase I : rEGCase II), the tri-sialylated ganglioside GT1b (1 : 26 0.3; rEGCase I : rEGCase II), the Gb4 (1 : 0.1; rEGCase I : rEGCase II) and cerebrosides. The 27 28 substrate specificity of our R.triatomea rEGCase I was comparable to literature findings for R.equi 29 30 rEGCase I except for Gb4, LacCer and cerebrosides 9. It is noteworthy that certain globosides are not 31 efficiently released by known EGCases 9,29 . Thus, the efficient release of the head group from Gb4 by 32 33 R. triatomea rEGCase I was compelling and suggested that this enzyme might be useful in the 34 35 profiling of globosides, which comprise many antigens (e.g. P-, P k-, Forssman-, SSEA-3- and SSEA-4- 36 antigen) in biological samples. Furthermore, R. triatomea rEGCase I efficiently hydrolyzed GlcCer 37 38 and to a lesser extent GalCer further indicating specificity differences from other Rhodococcal 39 40 EGCases that show little or no hydrolysis of cerebrosides 9,29,30 . 41 The GSL psychosine that has galactose linked to a sphingosine backbone (instead of ceramide) was 42 43 not released by rEGCase I. Psychosine is a component of brain GSLs. For example, it is of importance 44 45 in where it is thought to interfere with protein kinase C and accumulates in 46 microdomains of the brain which might result in disruption of lipid raft architecture 31 . For samples 47 48 of brain origin additional digestion with rEGCase II ( R. equi ) or EGALC ( R.equi ) that hydrolyze 49 50 psychosine to about 14% 29 is recommended. Finally, none of the known EGCases has shown the 51 ability to hydrolyze sulfatide 30 , a sulfated form of galactocerebroside, that is a highly enriched 52 53 component in the central and peripheral nervous system and plays an important role in the biology 54 32 55 of myelin-forming cells. Application of the workflow for profiling cellular GSL head groups 56 The suitability of our workflow for profiling cellular GSL glycans was assessed using mammalian 57 58 cell lines. In this experiment, R. triatomea rEGCase I was used to liberate glycan head groups from 59 60 8

ACS Paragon Plus Environment Page 9 of 22 Analytical Chemistry

1 2 3 GSLs extracted from mammalian NIH/3T3, HeLa and HL60 cell lines. In our study, only a small 4 5 number of samples needed to be processed, so GSL extraction was performed manually. However, 6 7 for larger numbers of samples, a robotic method for the automated extraction of GSLs as reported 8 by Stahlman et al 16 could be easily combined with our platform to permit the automated extraction 9 10 of GSLs from a variety of biological matrices including tissues, cells and biofluids 11 5 12 Well-resolved chromatographic profiles were obtained for aliquots of 2 x 10 cells after extraction 13 of GSLs according to Smith 23 , separation of polar impurities by n-butanol partitioning 24 , on-plate 14 15 treatment with 40 mU R. triatomea rEGCase I, followed by high-throughput sample preparation, 16 17 inclusion of an ISTD for relative peak quantification and analysis by UPLC-HILIC-FLD (Figure 2, 18 Table 2). The enzyme concentration (40 mU) was empirically chosen and was largely based on a 19 20 prior report of optimal cellular GSL release using 25-50 mU R. equi EGCase I 17 . Peak assignments 21 were confirmed by exoglycosidase treatment that resulted in characteristic shifts in GU values 22 23 depending on the presence of terminal sugars (data not shown) 33 . 24 25 High quality GSL glycan profiles were obtained for each cell line (Figure 2). Furthermore, the 26 glycans observed for each profile matched GSL class biases previously reported for NIH/3T3, HeLa 27 28 and HL60 cells that are rich in gangliosides, globosides and nLc-type GSLs, respectively 17 . These 29 30 data also further illustrate the ability of R. triatomea rEGCase I to hydrolyze globosides (see Hela 31 cell profile). Similarly, a broad range of nLc-type GSLs that included a N-glycolylneuraminic acid 32 33 (Neu5Gc)-containing species were released from HL60 extracts. Members of this class of GSLs are 34 35 not commercially available as standards and could not be evaluated in our experiments testing 36 EGCase release from defined GSL substrates. However, these data illustrate that rEGCase I is able to 37 38 liberate glycans from these lipids. Considered together, these data illustrate that our high- 39 40 throughput workflow can be used for efficient profiling of cellular GSL glycans. 41 42 43 Application of the workflow for profiling GSLs from human serum 44 45 Serum glycan profiling has become a useful tool for the monitoring of diseases and biomarker 46 discovery. We have recently introduced an automated IgG glycoprofiling platform 2 for the high- 47 48 throughput serum profiling of N-glycans derived from glycoproteins 5. Here we aimed to further 49 50 adapt the platform for the profiling of GSL glycans in human serum. 51 In contrast to cells or tissue, serum GSLs are not bound to membranes but are instead present in 52 53 their free form or embedded in protein-lipid complexes 11 . As such, we used a simplified extraction 54 55 method that included protein precipitation by methanol and centrifugation at low temperature 56 (Figure 1) as adapted from Huang et al 25 . A small volume (as low as 20 μL) of serum was sufficient 57 58 to obtain a high-quality fluorescence profiles within a total chromatographic run time of 40 minutes 59 60 9

ACS Paragon Plus Environment Analytical Chemistry Page 10 of 22

1 2 3 (Figure 3). To determine the optimum enzyme concentration for GSL head group release, serum 4 5 GSLs were incubated with increasing concentrations of EGCase I (6 mU, 30 mU, 60 mU, 120 mU) 6 7 (Figure S-2). The best glycan yields were obtained for 60 mU EGCase I. 8 9 10 Our data show that human serum contains a complex mixture of GSL head groups that includes 11 12 gangliosides, globosides, nLc- and fucosylated Lc-type GSLs (Figure 3). Serum also contains an 13 abundance of free monosaccharides, however monosaccharides derived from GSLs are relatively 14 15 low in healthy human serum 34 . Due to their excessive presence, monosaccharides were disregarded 16 17 in our study although their concentration was considerably reduced by butanol-partitioning during 18 sample preparation (Figure 1 and Supporting Information Figure S-3). GSL-derived 19 20 monosaccharides are of special interest in certain diseases such as Gaucher’s disease for which 21 increased glucosylcerebroside levels are observed 34 . To analyze GSL monosaccharides using our 22 23 workflow, additional purification, for example, using glucose oxidation 18 or hydrophilic-lipophilic 24 25 balance (HLB, Oasis ®) solid phase extraction 35 would be needed. The integration of HLB for our 26 GSL-serum application is shown in Supporting Information Figure S-3) and the use of commercially 27 28 available 96-well HLB plates would permit automation of the cleanup step for high-throughput 29 30 applications. 31 32 33 Our structural assignments of serum-derived GSL glycans were based on exoglycosidase glycan 34 35 sequencing and MS, that were each performed after weak anion exchange (WAX) fractionation of 36 serum GSL head groups into pools of neutral and charged structures (see Supporting Information 37 38 for supporting data in Tables S-3 and S-4, and Figures S-4-10). Eighteen GSL head group structures 39 40 with a relative abundance of ≥ 0.1% were identified in human serum in this study and quantified 41 relative to an internal standard. Additional structures of minor relative abundance (< 0.1%) may be 42 43 present but were not further considered. Prior studies on human serum reported high variability in 44 45 total serum GSL concentration but low inter-individual variability in GSL composition and 46 proportioning 11,18 . Thus, with a view toward discriminating between healthy and diseased serum 47 48 GSL glycan profiles, our findings support the notion that relative glycan quantification using UPLC- 49 50 HILIC-FLD is sufficient over absolute glycan quantification 13 . 51 52 53 We observed GM3 and LacCer as the predominant serum GSL species (34% each), followed by Gb3 54 55 (14%), Gb4 (7%), nLc4 (3%), (α2-3)-sialylated nLc4 (S(3)-nLc4, 2% ), (α1-2/α1-4)-di-fucosylated 56 Lc4 (diF(2,4)-Lc4,1.5%), GM2 (1%) and several structures of minor abundance (< 1%) including 57 58 GA2, GM1a, GD1a, GD3, nLc6, S(3)-nLc4, S(6)-nLc6, F(2)-Lc4, F(4)-Lc4 and Hex 2HexNAc 2 (Table 3). 59 60 10

ACS Paragon Plus Environment Page 11 of 22 Analytical Chemistry

1 2 3 4 5 Several other studies have also sought to identify GSL glycan structures present in serum using 6 7 classical methods like high performance thin layer chromatography (HPTLC). However, due to 8 limits in separation capacity, the methods have mainly been used to analyze only selected 9 10 ganglioside or globoside structures 11,13,15,36,37 . A first approach to serum GSL head group profiling 11 12 using HILIC-HPLC after 2AB-labeling of enzymatically released glycans was performed by Wing in 13 2001 35 . However, GSL head group profiling by HPLC is time-consuming (~180 minutes per run) and 14 15 the separate analysis of neutral and charged structures was required due to the GSL extraction and 16 17 glycan purification methods used. Furthermore, the ceramide glycanase (from Macrobdella decora ) 18 used in that study showed poor hydrolysis of neutral GSLs. Finally, Furukawa et al . recently 19 20 reported on the quantitative glycosphingolipid-glycome analysis in the serum of 10 healthy human 21 subjects using MALDI-TOF MS 18 . Of the 42 MALDI-TOF MS signals related to potential GSL glycan 22 23 structures an average of 19-20 signals had an abundance of ≥ 0.1% of the total glycan pool. 18 24 25 Similar cohorts and quantities of serum-GSL glycan structures were detected when compared to 26 our study (Table 3). Although MALDI-TOF MS allows for the detection of additional trace signals 27 28 from GSL-glycans, quantification by MALDI-TOF MS is difficult and correction factors may have to 29 30 be introduced. Additionally, signals from components such as media, N-glycans or peptides can also 31 complicate glycan identification and contrary to chromatographic methods it is not possible to 32 33 separate glycoforms. Thus, the described UPLC-HILIC-FLD method offers an important alternative 34 35 approach for profiling serum GSLs that increases speed and throughput without compromising 36 detection and determination of relative abundance of a large repertoire of GSL glycan classes. 37 38 39 40 Conclusion 41 In this report, we present the first high-throughputworkflow for the profiling of GSL glycan head 42 43 groups and their subsequent analysis by UPLC-HILIC-FLD using a robotic platform including 44 45 selective glycan capture and release on hydrazide beads in solution. The workflow features a novel 46 R. triatomea rEGCase I that shows a broad specificity for globosides, fucosylated GSLs and 47 48 cerebrosides compared to other EGCases. The workflow was successfully applied to perform 49 50 systematic GSL head group profiling of human serum and will enable future profiling of GSL head 51 groups in clinical disease research. 52 53 54 55 Conflict of interest 56 The authors declare that they have no conflicts of interest with the contents of this article. 57 58 59 60 11

ACS Paragon Plus Environment Analytical Chemistry Page 12 of 22

1 2 3 Author contributions 4 5 SA and SV designed, performed and analyzed the experiments of this study. HS provided technical 6 7 assistance. SA, SV and CHT wrote the paper. CMM, CHT and PMR coordinated the study. All authors 8 reviewed the results and approved the final version of the manuscript. 9 10 11 12 Supporting Information 13 This document file contains SupportingExperimental Section, Supporting Results, Supplementary 14 15 Figures S1-S9 and Tables S1-S4 and Supporting References. 16 17 18 References 19 20 (1) Marino, K.; Bones, J.; Kattla, J. J.; Rudd, P. M. Nat. Chem Biol. 2010 , 6, 713-723. 21 (2) Stockmann, H.; Adamczyk, B.; Hayes, J.; Rudd, P. M. Anal. Chem. 2013 , 85 , 8841-8849. 22 (3) AgilentTechnologies. http://www.chem.agilent.com/Library/flyers/Public/5991- 23 24 1140EN.pdf 2012 . 25 (4) Reusch, D.; Haberger, M.; Selman, M. H.; Bulau, P.; Deelder, A. M.; Wuhrer, M.; Engler, N. 26 Anal. Biochem. 2013 , 432 , 82-89. 27 (5) Stockmann, H.; O'Flaherty, R.; Adamczyk, B.; Saldova, R.; Rudd, P. M. Integrative biology : 28 quantitative biosciences from nano to macro 2015 , 7, 1026-1032. 29 30 (6) Adamczyk, B.; Tharmalingam, T.; Rudd, P. M. Biochim. Biophys. Acta 2012 , 1820 , 1347- 31 1353. 32 (7) Knezevic, A.; Gornik, O.; Polasek, O.; Pucic, M.; Redzic, I.; Novokmet, M.; Rudd, P. M.; 33 Wright, A. F.; Campbell, H.; Rudan, I.; Lauc, G. Glycobiology 2010 , 20 , 959-969. 34 (8) Thanabalasingham, G.; Huffman, J. E.; Kattla, J. J.; Novokmet, M.; Rudan, I.; Gloyn, A. L.; 35 36 Hayward, C.; Adamczyk, B.; Reynolds, R. M.; Muzinic, A.; Hassanali, N.; Pucic, M.; Bennett, A. 37 J.; Essafi, A.; Polasek, O.; Mughal, S. A.; Redzic, I.; Primorac, D.; Zgaga, L.; Kolcic, I.; Hansen, T.; 38 Gasperikova, D.; Tjora, E.; Strachan, M. W. J.; Nielsen, T.; Stanik, J.; Klimes, I.; Pedersen, O. B.; 39 Njølstad, P. R.; Wild, S. H.; Gyllensten, U.; Gornik, O.; Wilson, J. F.; Hastie, N. D.; Campbell, H.; 40 McCarthy, M. I.; Rudd, P. M.; Owen, K. R.; Lauc, G.; Wright, A. F. Diabetes 2013 , 62 , 1329- 41 42 1337. 43 (9) Ishibashi, Y.; Kobayashi, U.; Hijikata, A.; Sakaguchi, K.; Goda, H. M.; Tamura, T.; Okino, N.; 44 Ito, M. J. Lipid Res. 2012 , 53 , 2242-2251. 45 (10) Schnaar RL; Suzuki A; P., S. In Essentials of Glycobiology , Varki A; Cummings RD; JD, E., 46 Eds.; Cold Spring Harbor Laboratory Press New York Chapter 10, 2009. 47 48 (11) Senn, H.-J.; Orth, M.; Fitzke, E.; Wieland, H.; Gerok, W. Eur. J. Biochem. 1989 , 181 , 657- 49 662. 50 (12) Daniotti, J. L.; Vilcaes, A. A.; Torres Demichelis, V.; Ruggiero, F. M.; Rodriguez-Walker, M. 51 Front Oncol. 2013 , 3, 306. 52 (13) Zaprianova, E.; Deleva, D.; Ilinov, P.; Sultanov, E.; Filchev, A.; Christova, L.; Sultanov, B. 53 54 Neurochem. Res. 2001 , 26 , 95-100. 55 (14) Tsukuda, Y.; Iwasaki, N.; Seito, N.; Kanayama, M.; Fujitani, N.; Shinohara, Y.; Kasahara, 56 Y.; Onodera, T.; Suzuki, K.; Asano, T.; Minami, A.; Yamashita, T. PLoS One 2012 , 7, e40136. 57 (15) Ullman, M. D.; McCluer, R. H. J. Lipid Res. 1977 , 18 , 371-378. 58 59 60 12

ACS Paragon Plus Environment Page 13 of 22 Analytical Chemistry

1 2 3 (16) Ståhlman, M.; Ejsing, C. S.; Tarasov, K.; Perman, J.; Borén, J.; Ekroos, K. J. Chromatogr. B 4 5 2009 , 877 , 2664-2672. 6 (17) Fujitani, N.; Takegawa, Y.; Ishibashi, Y.; Araki, K.; Furukawa, J.; Mitsutake, S.; Igarashi, 7 Y.; Ito, M.; Shinohara, Y. J. Biol. Chem. 2011 , 286 , 41669-41679. 8 (18) Furukawa, J.-i.; Sakai, S.; Yokota, I.; Okada, K.; Hanamatsu, H.; Kobayashi, T.; Yoshida, Y.; 9 Higashino, K.; Tamura, T.; Igarashi, Y.; Shinohara, Y. J. Lipid Res. 2015 . 10 11 (19) Hakomori, S. I. J. Lipid Res. 1966 , 7, 789-792. 12 (20) Yowler, B. C.; Stoehr, S. A.; Schengrund, C. L. J. Lipid Res. 2001 , 42 , 659-662. 13 (21) Kumar, S.; Bala, M.; Raghava, G. P.; Mayilraj, S. Genome Announc 2013 , 1, e0012913. 14 (22) Furste, J. P.; Pansegrau, W.; Frank, R.; Blocker, H.; Scholz, P.; Bagdasarian, M.; Lanka, E. 15 Gene 1986 , 48 , 119-131. 16 17 (23) Smith, D. F.; Prieto, P. A. In Current Protocols in Molecular Biology ; John Wiley & Sons, 18 Inc. : Cold Spring Harbor (NY) Unit 17.3, 2001. 19 (24) Vidugiriene, J.; Menon, A. K. Methods Enzymol. 1995 , 250 , 513-535. 20 (25) Huang, Q.; Zhou, X.; Liu, D.; Xin, B.; Cechner, K.; Wang, H.; Zhou, A. Anal. Biochem. 2014 , 21 455 , 26-34. 22 23 (26) Royle, L.; Campbell, M. P.; Radcliffe, C. M.; White, D. M.; Harvey, D. J.; Abrahams, J. L.; 24 Kim, Y. G.; Henry, G. W.; Shadick, N. A.; Weinblatt, M. E.; Lee, D. M.; Rudd, P. M.; Dwek, R. A. 25 Anal. Biochem. 2008 , 376 , 1-12. 26 (27) Albrecht, S.; Lane, J. A.; Marino, K.; Al Busadah, K. A.; Carrington, S. D.; Hickey, R. M.; 27 Rudd, P. M. Br. J. Nutr 2014 , 111 , 1313-1328. 28 29 (28) Miura, Y.; Hato, M.; Shinohara, Y.; Kuramoto, H.; Furukawa, J.; Kurogochi, M.; Shimaoka, 30 H.; Tada, M.; Nakanishi, K.; Ozaki, M.; Todo, S.; Nishimura, S. Mol. Cell. Proteomics. 2008 , 7, 31 370-377. 32 (29) Ishibashi, Y.; Nakasone, T.; Kiyohara, M.; Horibata, Y.; Sakaguchi, K.; Hijikata, A.; 33 34 Ichinose, S.; Omori, A.; Yasui, Y.; Imamura, A.; Ishida, H.; Kiso, M.; Okino, N.; Ito, M. J. Biol. 35 Chem. 2007 , 282 , 11386-11396. 36 (30) Ito, M.; Yamagata, T. J. Biol. Chem. 1989 , 264 , 9510-9519. 37 (31) White, A. B.; Givogri, M. I.; Lopez-Rosas, A.; Cao, H.; van Breemen, R.; Thinakaran, G.; 38 Bongarzone, E. R. J Neurosci. 2009 , 29 , 6068-6077. 39 40 (32) Grassi, S.; Prioni, S.; Cabitta, L.; Aureli, M.; Sonnino, S.; Prinetti, A. Neurochem Res 2016 , 41 41 , 130-143. 42 (33) Royle, L.; Radcliffe, C. M.; Dwek, R. A.; Rudd, P. M. Methods Mol Biol. 2006 , 347 , 125-143. 43 (34) Muller, M. V. G.; Petry, A.; Vianna, L. P.; Breier, A. C.; Michelin-Tirelli, K.; Pires, R. F.; 44 Trindade, V. M. T.; Coelho, J. C. Braz. J. Pharm. Sci 2010 , 46 , 643-649. 45 46 (35) Wing, D. R.; Garner, B.; Hunnam, V.; Reinkensmeier, G.; Andersson, U.; Harvey, D. J.; 47 Dwek, R. A.; Platt, F. M.; Butters, T. D. Anal. Biochem. 2001 , 298 , 207-217. 48 (36) Gornati, R.; Bembi, B.; Tong, X.; Boscolo, R.; Bruno, B. Clin. Chim. Acta 1998 , 271 , 151- 49 161. 50 (37) Kundu, S. K.; Diego, I.; Osovitz, S.; Marcus, D. M. Arch Biochem Biophys 1985 , 238 , 388- 51 52 400. 53 54 55 56 57 58 59 60 13

ACS Paragon Plus Environment Analytical Chemistry Page 14 of 22

1 2 3 4 5

6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26

27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47

48 49 50 51 52 53 54 55 56 57 58 59 60 14

ACS Paragon Plus Environment Page 15 of 22 Analytical Chemistry

1 2 3 Tables 4 5 6 Table 1. Specificity of recombinant endoglycoceramidases using purified GSL substrates. 7 8 Comparison of the relative abundance of glycans released from standards using 2 mU R. 9 triatomea rEGCase I or R. equi rEGCase II as assessed by UPLC-HILIC-FLD and comparison to 10 11 literature results for R.equi rEGCase I 9. For a relative comparison of the release efficiency of the two 12 13 experimentally tested EGCases, the highest abundance of each glycolipid was normalized to “1”. 14 Unless otherwise stated (*) this abundance also represented 100% head group release by a parallel 15 16 TLC analysis (see Supplementary Table S-2). The coefficient of variance (CV) was < 10% for all 17 18 substrate/enzyme pairs tested. 19 20 average rel. glycan quantities Hydrolysis % 21 (n = 24) 22 R.triatomea R. equi R.equi 23 9 24 Substrate glycolipid structure rEGCase I rEGCase II GU rEGCase I 25 26 Gangliosides 27 GM3 Neu5Ac(α23)Gal(β14)Glc(β11)’Cer 1.0 1.0 3.15 28 100 29 GM1a Gal(β13)GalNAc(β14)[Neu5Ac(α23)] 1.0 0.9 4.39 30 Gal(β14)Glc(β11)’Cer 100 31 FGM1 Fuc(α12)Gal(β13)GalNAc(β14) 1.0 0.4 4.69 32 [Neu5Ac(α23)]Gal(β14)Glc(β11)’Cer 100 33 GD3 Neu5Ac(α28)Neu5Ac(α23)Gal(β14) 1.0 1.0 4.62 34 Glc(β11)’Cer 100 35 GD1a Neu5Ac(α23)Gal(β13)GalNAc(β14) 1.0 0.8 5.44 36 [Neu5Ac(α23)]Gal(β14)Glc(β11)’Cer 100 37 GD1b Gal(β13)GalNAc(β14)[Neu5Ac(α28) 1.0 0.8 6.03 38 Neu5Ac(α23)]Gal(β14)Glc(β11)’Cer 100 39 GT1b Neu5Ac(α23)Gal(β13)GalNAc(β14) 1.0 0.3 7.02 40 [Neu5Ac(α28)Neu5Ac(α23)]Gal(β1 n.d. 41 4)Glc(β11)’Cer 42 43 Globosides 44 Gb4 GalNAc(β13)Gal(α14)Gal(β14)Glc(β1 1.0 0.1 3.41 45 1)’Cer 33.9 (100) + 46 47 Cerebrosides 48 GlcCer Glc(β11)’Cer 1.0 0.2 0.90 49 6.7 (20.7) + 50 GalCer* Gal(β11)’Cer 0.05 0.1* 0 0.90 ‡ 51 Sulfatide HSO 3Gal(β11)’Cer 0 0 0 52 3 Psychosine Gal(β11)’sphingosine 0 1.0 0.90 53 n.d. 54 55 LacCer Gal(β14)Glc(β11)’Cer 1.0 0.9 2.00 56 29.9 (65.9) + 57 *According to a TLC-based assay, the total release efficiency on GalCer was approx. 5-10% for R. triatomea rEGCase I. 58 59 60 15

ACS Paragon Plus Environment Analytical Chemistry Page 16 of 22

1 2 3 +GSLs (2nmol) were incubated at 37°C for 12h with 1mU (or 10mU) of recombinant R.equi EGCase I in 20µL of 50mM sodium acetate 4 buffer, pH 5.5, containing 0.1% Triton X-100. 9 5 ‡GSLs possessing the the β-galactosyl-Cer linkage (e.g. trigalactosylCer) were completely resistant to the hydrolysis by R.equi rEGCase I 9n.d. not determined 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Table 2. Use of the workflow for profiling mammalian cellular GSLs . Shown are relative quantities 20 21 of glycans released from GSLs extracted from murine NIH/3T3 and human HeLa and HL60 cell lines 22 23 using 40 mU of R. triatomea rEGCase I in the high-throughput glycolipid workflow. The 24 quantification was performed relative to an internal standard for which the area was set to “1”. 25 26 27 rEGCase I (40 mU) 28 GSL GU av* NIH/3T3 HeLa HL60 29 30 LacCer 1.96 1.6 6.8 18.8 31 32 GA2 2.59 0.2 -- -- 33 GM3 3.11 5.0 3.3 0.6 34 GM2 3.51 4.0 5.5 -- 35 GM1a 4.37 0.5 -- -- 36 GD1a 5.42 0.1 -- -- 37 total 9.8 8.8 0.6 38 gangliosides 39 40 Gb3 2.71 -- 35.7 -- 41 Gb4 3.35 -- 6.4 -- 42 total globosides -- 42.1 -- 43 44 nLc3 2.72 0.5 1.1 45 nLc4 3.55 -- 0.8 3.2 46 S(3)-nLc4 4.60 -- 0.5 0.5 47 Sg(3)-nLc4 4.97 -- -- 0.4 48 nLc6 5.15 -- -- 0.1 49 S(3)-nLc6 6.06 -- -- 0.2 50 total nLc -- 1.8 5.5 51 52 TOTAL 11.4 59.5 24.9 53 *GU av : average Glucose Units. S: Neu5Ac. Sg: Neu5Gc. 54 55 56 57 58 59 60 16

ACS Paragon Plus Environment Page 17 of 22 Analytical Chemistry

1 2 3 Table 3: Relative abundance of GSL-glycans in human serum released by R. triatomea 4 5 rEGCase I (60 mU) and analysed by UPLC-HILIC-FLD (Figure 3), compared to their relative 6 18 7 abundances as determined by MALDI-TOF MS . GU and relative abundance are average (av) 8 values from four independent measurements (average CV ≤ 10%). Structural assignments were 9 10 confirmed by exoglycosidase sequencing and mass spectrometry. 11 12 MALDI* 18 13 relative relative 14 GU GSL GSL head group GSL class abundance av av abundance 15 (%) av 16 (%) 17 1.95 LacCer Gal(β14)Glc 34.2 18.9 a 18 2.60 GA2 GalNAc(β14)Gal(β14)Glc ganglioside 0.2 0.1 19 2.73 Gb3 Gal(α14)Gal(β14)Glc globoside 13.8 14.7 20 21 3.13 GM3 Neu5Ac(α23)Gal(β14)Glc ganglioside 33.8 50.3 a 22 3.41 Gb4 GalNAc(β13)Gal(α14)Gal(β14)Glc globoside 7.1 7.3 b 23 GalNAc(β14)[Neu5Ac(α23)]Gal(β1 3.55 GM2 ganglioside 1.0 0.2 24 4)Glc 25 3.60 nLc4 Gal(β14)GlcNAc(β13)Gal(β14)Glc neoLc 3.3 7.3 b 26 Fuc(α12)Gal(β13)GlcNAc(β1 4.05 F(2)Lc4 Lc 0.3 0.6 c 27 3)Gal(β14)Glc 28 4.17 Hex HexNAc nd Lc 0.2 < 0.1 29 2 2 30 4.38 F(4)Lc4 Gal(β13)[Fuc(α14)]GlcNAc(β13)Gal Lc 0.4 0.6 c 31 Gal(β13)GalNAc(β14)[NeuAc(α2 4.42 GM1a ganglioside 0.6 3.1 d 32 3)]Gal(β14)Glc Neu5Ac(α28)NeuAc(α23)Gal(β1 33 4.53 GD3 ganglioside 0.1 < 0.1 34 4)Glc Neu5Ac(α23)Gal(β14)GlcNAc(β1 35 4.63 S(3)nLc4 nLc 2.4 3.1 d 36 3)Gal(β14)Glc Neu5Ac(α26)Gal(β14)GlcNAc(β1 37 4.99 S(6)nLc4 nLc 0.8 3.1 d 38 3)Gal(β14)Glc Fuc(α12)Gal(β13)[Fuc(α1 39 5.12 diF(2,4)Lc4 Lc 1.5 0.7 40 4)]GlcNAc(β13)Gal(β14)Glc Gal(β14)GlcNAc(β13)Gal(β1 41 5.16 nLc6 nLc 0.1 nd 42 4)GlcNAc(β13)Gal(β14)Glc NeuAc(α23)Gal(β13)GalNAc(β14) 43 5.44 Gd1a ganglioside 0.1 0.3 44 [NeuAc(α23)]Gal(β14)Glc Neu5Ac(α23)Gal(β14)GlcNAc(β1 45 6.09 S(3)nLc6 nLc 0.2 0.2 e 46 3)Gal(β14)GlcNAc(β13)Gal(β14)Glc 47 nd: structural details not defined 48 *relative abundances were calculated from average absolute GSL glycan quantities (pmol/100μL) in10 humn serum samples as determined byMALDI-TOF MS by Furukawa et al 18 (Supplementary Table 4, Furukawa et al) 18 49 adeviations in relative abundances between the two studies for LacCer and GM3 might be due to the use of correction 50 factors for the absolute quantitation of LacCer and GM3 by MALDI-TOF MS 18 51 bno structural discrimination was made between Gb3 and (n)Lc4 by MALDI-TOF MS due to same m/z 18 52 cno structural discrimination was made between F-(n)Lc4 isomers by MALDI-TOF MS due to same m/z 18 53 dno structural discrimination was made between GM1 and S-(n)Lc4 by MALDI-TOF MS due to same m/z 18 54 em/z which represents (Hex4)(HexNAc2)(Neu5Ac1) was not further specified by Furukawa et al 18 but corresponds to 55 S(3)-nLc6 as identified in our study 56 57 58 59 60 17

ACS Paragon Plus Environment Analytical Chemistry Page 18 of 22

1 2 3 Figure legends 4 5 6 7 8 Figure 1. Schematic of the high-throughput GSL glycan preparation and analysis workflow . 9 The workflow includes on-plate endoglycoceramidase incubation of extracted GSLs from different 10 11 matrices to which an internal standard has been added, automated hydrazide bead cleanup of 12 13 released glycans, high-throughput 2AB-glycan labeling and cleanup of excess label followed by 14 UPLC-HILIC-FLD analysis. 15 16 17 18 Figure 2. UPLC-HILIC profiles of mammalian cellular GSL-glycans . The glycans were released by 19 R. triatomea rEGCase I (40 mU) (A) NIH/3T3-cells (B) HeLa-cells (C) HL60-cells. See Table 2 for 20 21 relative peak quantification. *Internal Standard. 22 23 24 Figure 3. Representative UPLC-HILIC-FLD profile and peak assignments of GSL-glycans from 25 26 human serum. Glycans were released by R. triatomea rEGCase I (60 mU). See Table 3 for relative 27 28 peak quantification. GU: glucose units. 29 30 31 32 33 34 35 36 37 38 39 40 41 42

43 44 45 46 47

48 49 50 51 52 53 54 55 56 57 58 59 60 18

ACS Paragon Plus Environment Page 19 of 22 Analytical Chemistry

1 2 3 Figure 1 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 19

ACS Paragon Plus Environment Analytical Chemistry Page 20 of 22

1 2 3 Figure 2 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 20

ACS Paragon Plus Environment Page 21 of 22 Analytical Chemistry

1 2 3 Figure 3 4 5

6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

49 50 51 52 53 54 55 56 57 58 59 60 21

ACS Paragon Plus Environment Analytical Chemistry Page 22 of 22

1 2 3 4 For TOC only 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 22

ACS Paragon Plus Environment