Quantitative Glycomics of Human Whole Serum Glycoproteins Based on the Standardized Protocol for Liberating N-Glycans *S

Global glycomics of human whole serum glycoproteins appears to be an innovative and comprehensive approach to identify surrogate non-invasive biomarkers for various diseases. Despite the fact that quantitative glycomics is premised on highly efficient and reproducible oligosaccharide liberation from human serum glycoproteins, it should be noted that there is no validated protocol for which deglycosylation efficiency is proven to be quantitative. To establish a standard procedure to evaluate N-glycan release from whole human serum glycoproteins by peptide-N-glycosidase F (PNGase F) treatment, we determined the efficiencies of major N-glycan liberation from serum glycoproteins in the presence of reducing agents, surfactants, protease treatment, or combinations of pretreatments prior to PNGase F digestion. We show that de-N-glycosylation efficiency differed significantly depending on the condition used, indicative of the importance of a standardized protocol for the accumulation and comparison of glycomics data. Maximal de-N-glycosylation was achieved when serum was subjected to reductive alkylation in the presence of 2-hydroxyl-3-sulfopropyl dodecanoate, a surfactant used for solubilizing proteins, or related analogues, followed by tryptic digestion prior to PNGase F treatment. An optimized de-N-glycosylation protocol permitted relative and absolute quantitation of up to 34 major N-glycans present in serum glycoproteins of normal subjects for the first time. Moreover PNGase F-catalyzed de-N-glycosylation of whole serum glycoproteins was characterized kinetically, allowing accurate simulation of PNGase F-catalyzed de-N-glycosylation required for clinical glycomics using human serum samples. The results of the current study may provide a firm basis to identify new diagnostic markers based on serum glycomics analysis.

Sequencing the human genome and that of various pathogens has opened the door for proteomics, which has dramat-ically facilitated the search for diagnostic biomarkers. Proteomics approaches have emerged as indispensable tools to identify new disease markers from clinical specimens (1,2). In particular, human serum/plasma proteomes are considered the most informative proteomes from a medical/clinical point of view because they likely contain most human proteins as well as proteins derived from some pathogens (3).
An additional approach currently in development focuses on serum protein glycomics, the qualitative and quantitative characterization of the gross glycans present in serum (4). Indeed glycosylation is the most common posttranslational modification of cell surface and extracellular matrix proteins, and most plasma proteins are also thought to be heavily glycosylated. Changes in abundance and alterations in glycan profiles of serum and cell surface proteins have been shown to correlate with progression of cancer and other disease states (5)(6)(7). Recently Callewaert et al. (8) reported the glycomics analysis of 106 patients with chronic liver disorders at various stages of severity and revealed significant alterations in specific N-glycans depending on the presence of cirrhosis. Although prostate-specific antigen tests often suffer from lack of specificity in distinguishing benign prostate hyperplasia from prostate cancer, recent studies indicate that N-glycans of prostate-specific antigen found in prostate cancer differ significantly from those seen in benign prostate hyperplasia and therefore could be a potential indicator leading to improved sensitivity in diagnosing prostate cancer (9,10).
For glycomics analysis, glycans are often released from protein backbones. Asn-linked type glycans can be cleaved enzymatically by peptide-N-glycosidase F (PNGase F) 1 (peptide-N 4 -(N-acetyl-␤-glucosaminyl)asparagine amidase, EC 3.5.1.52; previously assigned as EC 3.2.2.18) (11) or glycoamidase A (EC 3.5.1.52) (12) and chemically by hydrazinolysis (13,14). The former is more preferably used because it yields intact oligosaccharides regardless of size or structure of the substrate carbohydrate moiety and a slightly modified protein in which Asn residues at the site of de-N-glycosylation are converted to Asp, whereas hydrazinolysis causes chemical modification including N-deacetylation of sialic acids and Nacetyl-D-hexosamines such as GlcNAc and GalNAc residues as well as extensive cleavage of polypeptide backbones. However, glycoproteins widely differ in susceptibility to enzymatic digestion because glycosylated sites are often obstructed by secondary and tertiary protein structure. To optimize efficiency of enzymatic release of N-glycans from individual/target glycoproteins, several conditions have been utilized using reducing agents, surfactants, protease treatment, or a combination of pretreatments prior to PNGase F digestion to make glycosylation sites more accessible. Although these procedures are often used to obtain qualitative information on N-glycan structures of specific glycoproteins, there are no standardized conditions allowing highly efficient and reproducible liberation of N-glycans from serum whole glycoprotein. It should be noted that quantitative glycomics is premised on non-biased, highly efficient, and reproducible oligosaccharide liberation from human serum glycoproteins. Therefore, our attention must be directed to establish a standardized procedure for liberating major N-glycans from human whole serum glycoproteins. Using an optimized protocol for quantitative glycomics, we revealed for the first time the absolute concentrations of major N-glycans occurring in human serum whole glycoproteins.

Materials
1-Butanol, ammonium bicarbonate, and sodium phosphate buffer were purchased from Kanto Chemical Co., Inc. (Tokyo, Japan). 2-Hydroxyl-3-sulfopropyl dodecanoate (HSD), 3-methyl-1-p-tolyltriazene (MTT), and sodium cyanoborohydride were purchased from Aldrich. Trypsin was purchased from Sigma-Aldrich. PNGase F (recombinant) and Pronase were obtained from Hoffmann-La Roche and Calbiochem, respectively. Large scale preparation of PNGase F was carried out according to a previously reported method (14,15). Briefly Flavobacterium meningosepticum (ATCC33958) was cultured, and the medium was centrifuged and filtered. The extract was concentrated by ultrafiltration, and ammonium sulfate was added to 90%. After centrifugation, the precipitate was resuspended in 0.1 M sodium phosphate buffer (pH 7.0) containing 1 M ammonium sulfate and 1 mM EDTA and then centrifuged. The supernatant was applied to a TSKbutyl-Toyopearl 650 M column, and the PNGase F fraction was collected and lyophilized. Sephadex G-15 resin was obtained from Amersham Biosciences, Bio-Gel P-4 (200 -400 mesh) was from Bio-Rad, and a ShimPack HRC-ODS silica column (6.0-mm internal diameter ϫ 150 mm) was from Shimadzu Co. (Kyoto, Japan). 2,5-Dihydroxybenzoic acid, human angiotensin II, bombesin, and adrenocorticotropic hormone 18 -39 were from Bruker Daltonics (Bremen, Germany). 2-Aminopyridine, acetonitrile (HPLC/MS grade), methanol (HPLC/MS grade), acetic acid, ammonium acetate, and other reagents were from Wako Pure Chemical Industries, Ltd. (Osaka, Japan). Both forms of N ␣ -((aminooxy)acetyl)tryptophanylarginine methyl ester (aoWR), aoWR(H) and aoWR(D), were prepared as described previously (16). Human sera from two normal subjects (males, 73 and 69 years old, designated as normal sera A and B, respectively) were purchased from Genomics Collaborative Inc. (Cambridge, MA). Human serum was also purchased from Sigma-Aldrich and was designated as normal serum C. A serum sample of a rheumatoid arthritis (RA) patient (female, 52 years old) was obtained from the Department of Orthopaedic Surgery, Hokkaido University School of Medicine (Sapporo, Japan) under the permission of the Commission of Bioethics.

Synthesis of Novel Detergents
1-Propanesulfonic Acid, 2-Hydroxy-3-lauramido (PHL)-3-Chloro-2-hydroxypropane sulfonic acid sodium salt (1.97 g, 10 mmol) was dissolved in 12.5% aqueous ammonia. The mixture was stirred for 24 h at room temperature, and the solvent was removed under reduced pressure to give 3-amino-2-hydroxypropane sulfonic acid sodium salt in quantitative yield. NaHCO 3 (84 mg, 1 mmol) and lauroyl chloride (231 l, 1 mmol) were added to a solution of amino sulfonic acid (177 mg, 1 mmol) in 10 ml of water, and the mixture was stirred for 2 h at room temperature. The reaction mixture was subjected to chromatography on a column of Wako gel 50C18 (2 cm 3 ) using MeOH as eluant and recrystallized to give PHL (positive ion mode MALDI-TOF m/z for C 15

Release of N-Glycans by PNGase F
Before PNGase F digestion, human serum (20 l) was pretreated using nine different conditions as follows. In Condition A, serum was mixed with 4 l of 100 mM ammonium bicarbonate (pH 7.8) and 26 l of H 2 O. In Condition B, serum was mixed with 4 l of 100 mM ammonium bicarbonate (pH 7.8), 21 l of H 2 O, and 5 l of trypsin (400 units) followed by incubation at 37°C for 1 h. Then trypsin was heat denatured at 80°C for 15 min. In Condition C, serum was mixed with 4 l of 100 mM ammonium bicarbonate (pH 7.8), 8 l of H 2 O, and 8 l of 50 mM DTT followed by incubation at 60°C for 30 min. 5 l of 135 mM iodoacetamide (IAA) in H 2 O was added, and the mixture was allowed to stand at room temperature for 1 h. Then 5 l of trypsin (400 units) was added and incubated at 37°C for 1 h followed by heat denaturation at 80°C for 15 min. In Condition D, serum was diluted with an equal volume of 50 mM Tris/HCl buffer (pH 7.8) containing 2% (w/v) SDS and 2% 2-mercaptoethanol and heated to 95°C for 5 min. Then an equal volume of buffer solution containing 8% (v/v) Triton X-100 was added. In Condition E, serum was mixed with 4 l of 100 mM ammonium bicarbonate (pH 7. was added, and the mixture was allowed to stand at room temperature for 45 min. Then 5 l of trypsin (400 units) was added and incubated at 37°C for 1 h followed by heat denaturation at 80°C for 15 min. Subsequently all the pretreated serum samples were treated with PNGase F (2 units) at 37°C for 24 h followed by heat denaturation at 90°C for 15 min. The final volume of all samples was adjusted to 200 l with 100 mM ammonium bicarbonate. All sample preparations were performed in triplicate except for Condition D, which was in duplicate.

Preparation of 2-Aminopyridine (PA)-oligosaccharides from Human Serum Glycoproteins
Following enzymatic release of serum N-glycans under the digestion conditions described above, an aliquot of each sample (50 l) was digested with 20 g of Pronase, and the mixture was purified by Bio-Gel P-4 column chromatography. Oligosaccharides obtained were reductively aminated with 1.7 M PA and 2.0 M sodium cyanoborohydride at 90°C for 1 h and then purified on a Sephadex G-15 column using 10 mM ammonium bicarbonate as eluant (17). After removing the solvent, the sample was dissolved in 500 l of water, and a 5-l aliquot was injected into the reversed-phase HPLC system.  (134)GlcNAc␤(134)GlcNAc) was prepared by PNGase F digestion of sialylglycopeptide, which was purified from hen egg yolk (18). Briefly fresh egg yolk was treated with phenol, and the supernatant was purified by gel filtration (Sephadex G-50 column and Sephadex G-25 column) and chromatographed on an anion exchange column (DEAE-Toyopearl 650 M) and then a cation exchange column (CM-Sephadex C-25). Purified sialylglycopeptide was digested with PNGase F, and then a standard PAoligosaccharide was prepared with the released sialyloligosaccharide by the procedure noted above.

Methyl Esterification of Sialic Acid Residues and Labeling with aoWR
Following enzymatic release of serum N-glycans under Condition G, an aliquot of the sample (equivalent to 2 l of serum) was digested with 2 g of Pronase, and the mixture was subjected to purification on a Bio-Gel P-4 column. Whole N-glycans obtained were subjected to methyl esterification by treatment with MTT in DMSO-acetonitrile according to previously reported conditions with slight modification (19). Briefly lyophilized material (N-glycans) was dissolved in 20 l of 100 mM HCl, and 480 l of acetonitrile was added. The sample was then applied onto ϳ20 mg of Iatrobeads silica gel (Iatron Laboratories, Inc., Tokyo, Japan) packed in a disposable filter column, Mobicol polypropylene column (1 ml, MoBiTec, Gö ttingen, Germany), which had been preequilibrated with 1 M acetic acid and acetonitrile. The column was washed with acetonitrile by centrifugation. With the bottom cap in place, 100 l of 100 mM MTT in a 1:1 mixture of acetonitrile and DMSO was added, and the column was incubated for 1 h at 60°C. With the bottom cap still in place, 500 l of acetonitrile was added to the column and briefly mixed. Next the bottom cap was removed. The column was washed with acetonitrile, 2% acetic acid in acetonitrile, and 96% acetonitrile in water, successively. The methyl esterified free oligosaccharides were eluted from the silica gel by 50% aqueous acetonitrile. The recovered oligosaccharides were labeled with aoWR according to a method described previously (16). Briefly an aliquot (50 l) of the eluate (200 l) was directly mixed with 5 l of 2 mM aoWR and 50 l of 2% acetic acid in acetonitrile, and the mixture was heated to 60°C until the solvent evaporated (ϳ1 h).

MALDI-TOF Mass Spectrometry
Methyl-protected and aoWR-labeled N-glycans were dissolved in 10 l of water, the mixture was directly mixed with 2,5-dihydroxybenzoic acid (10 mg/ml in 30% acetonitrile) at a 1:10 dilution, and an aliquot (1 l) was deposited on a stainless steel target plate. MALDI-TOF data were obtained using an Ultraflex time-of-flight mass spectrometer (Bruker Daltonics, Bremen, Germany) equipped with a LIFT-TOF/TOF facility controlled by FlexControl 2.0 software according to the general procedure reported previously (20,21). All spectra were obtained using a reflectron mode with an acceleration voltage of 25 kV, a reflector voltage of 26.3 kV, and a pulsed ion extraction of 160 ns in the positive ion mode. These spectra were the sum of 1,000 laser shots. All peaks were picked by FlexAnalysis 2.0 using the Sophisticated Numerical Annotation Procedure (SNAP) algorithm that fits isotopic patterns to the matching experimental data. The algorithm provides the monoisotopic mass, the intensity and area under the envelope of the isotopic cluster, and the resolution of the peaks in the cluster. Estimation of N-linked type oligosaccharide structures was obtained by input of peak masses into the GlycoMod Tool (Swiss Institute of Bioinformatics) and GlycoSuite (Proteome Systems).

Enzyme Kinetics Study
Kinetics analysis of PNGase F was carried out at 37°C in 100 mM ammonium bicarbonate (200 l), and the reaction was terminated by heating in boiling water. Each velocity was determined at a 30 -150 M N-glycan concentration using serum denatured under Condition G. The amount of released oligosaccharides was determined by reductive amination with PA and HPLC analysis as described above. Initial rates were defined as the amount of product formed after incubation for 60 min, and K m app and V max values were determined by Sigma Plot, Enzyme Kinetics Module (SYSTAT Software Inc., Chicago, IL). Deglycosylation using PNGase F was characterized by Michaelis-Menten kinetics.

Effect of PNGase F Digestion Conditions on Efficiency of
Deglycosylation of Serum Whole Glycoproteins-Glycoproteins differ widely in susceptibility to PNGase F deglycosylation such that they often require denaturation prior to enzymatic treatment. Glycoproteins are typically denatured by heating in an appropriate detergent (e.g. SDS (22,23) or ALS (24)) or by protease (e.g. trypsin or chymotrypsin) pretreatment (17,25,26) with or without reductive alkylation. ALS, sodium-[(2-methyl-2-undecyl-1,3-dioxolan-4-yl) methoxyl]-1propanesulfonate, is designed to degrade at low pH condition to eliminate surfactant-caused interference with analysis (24). However, one of the decomposition products contains a ketone group; hence it can seriously interfere with labeling of oligosaccharides required to improve detection sensitivity via reductive amination or hydrazone/oxime formation toward the hemiacetal-reducing terminus. To overcome this difficulty, we sought alternatives with chemical properties similar to ALS and chose to evaluate HSD as a designated surfactant (Fig. 1). A quantitative comparison among different digestion conditions was evaluated by an established reversed-phase HPLC method following pyridylamination of released oligosaccharides. The areas of 14 major peaks in Fig. 2 (major oligosaccharide(s) present in each peak (17,27) are shown in Table I) were used for quantitative analysis.
As shown in Fig. 3a, the total amount of deglycosylated glycans differs significantly depending on the conditions used. When enzyme digestion was performed without denaturation pretreatment (Condition A), the releasing efficiency was significantly low, supporting previous findings that denaturation of substrate before deglycosylation is indispensable for high efficiency release of glycans. Tryptic digestion prior to PNGase F digestion (Condition B) and in combination with reductive alkylation (Condition C) improved efficiencies by ϳ88 and ϳ127%, respectively, over Condition A. Likewise deglycosylation efficiency was improved following treatment with solubilizing agents combined with reductive alkylation, although improvement differed depending on which surfactants were used. SDS, ALS, or HSD improved efficiency by ϳ63, ϳ75, or ϳ104%, respectively. Without reductive alkylation, improvement by addition of ALS and HSD remained ϳ19 and ϳ52%, respectively (data not shown), indicating that using reductive alkylation is quite effective in improving deglycosylation efficiency not only for protease-assisted but also for surfactant-assisted deglycosylation. Combining HSD, reductive alkylation, and trypsin digestion (Condition G) improved deglycosylation efficiency by ϳ134% compared with Condition A, an improvement nearly the same as that of Condition C. However, considering that only 22 of the most abundant human serum proteins constitute 99% of the total protein mass (28) and that low abundance glycoproteins highly resistant to PNGase F may occur in the remaining 1%, the most rigorous condition (Condition G) would be recommended to ensure maximal deglycosylation efficiency. Note that HSD was proven useful as a protein solubilizer for the first time, although its efficacy as a cleaning agent has been reported (29). Because we confirmed the utility of HSD for PNGase F-catalyzed deglycosylation, we synthesized HSD analogues (PHL and PHM; Fig. 1) and evaluated their activities as protein solubilizer(s). When serum was reductively alkylated in the presence of PHL (Condition H) or PHM (Condition I) rather than HSD (Condition G) followed by trypsin treatment prior to PNGase F digestion, we observed that deglycosylation efficiencies were almost identical to that achieved by Condition G (Fig. 3b).
The relative quantitative profiles of released N-glycans were compared among the three most efficient digestion conditions (Conditions C, F, and G). As shown in Fig. 3c, the profile obtained under Condition F differed somewhat from those obtained under Conditions C and G possibly due to the presence of glycosylation sites that are less susceptible to PN-Gase F in the absence of trypsin digestion. The observation that additional tryptic digestion substantially improved deglycosylation efficiency appeared to be reasonable because it was indicated earlier that hydrolysis rates of PNGase F may be primarily determined by peptide length (30). However, tryptic digestion further complicates biological material and therefore makes the glycomics analysis even more challenging. The reproducibility of deglycosylation efficiency was fairly good (coefficient of variation Ͻ8%) throughout Conditions A-G when sample preparations from serum and HPLC analyses were performed in triplicate (except for Condition D, which was performed in duplicate). The relative quantity of each peak, calculated by the relative peak area ratios, was also reproducible (i.e. coefficient of variation of relative quantity was less than 14% for those peaks whose relative quantities are more than 1%) throughout Conditions A-G. In this regard, tryptic digestion may be omitted to simplify the analysis depending on the purpose of the study. Our finding, however, indicates that the quantitative serum glycomic profile could be severely affected by the digestion condition used, and therefore care should be taken when handling the quantitative glycomics data. As an efficient tool to rapidly purify oligosaccharides from highly complicated biological matrices, we recently developed a chemoselective glycoblotting platform utilizing synthetic polymers displaying aminooxy functionality (31)(32)(33). The combination of the described standardized sugar-liberating protocol and glycoblotting technique for high throughput, large scale disease-related serum glycomics is currently in progress in our laboratory. man serum proteins, the major N-glycans present in human serum were quantified. Oligosaccharides of normal serum A released under Condition G were methyl esterified followed by labeling with aoWR(H) and subjected to MALDI-TOF analysis. It has been communicated that oligosaccharides with masses greater than about 1000 Da exhibited similar signal strengths, irrespective of structure, when examined on the MALDI-TOF MS system (35). In addition, methyl esterified N-glycans are proven to exhibit signal strength in positive ion MALDI-TOF MS comparable to neutral oligosaccharides (36); thus a positive ion MALDI mass spectrum of mixtures of neutral and methyl esterified sialic acid containing oligosaccharides should reflect the relative proportions of those glycans. As shown in Fig. 4, the analysis allowed detection of signals of up to 34 glycans. We confirmed that all these signals are aoWR derivatives by comparing m/z differences when label-  Table I. AU, arbitrary units. Fig. 2 ing was performed by aoWR(D), a deuterated (d 3 -methyl) analog of aoWR(H) (16). Mono-and disialylated biantennary glycans were the two most abundant N-glycans in human serum, reflecting that many high abundance glycoproteins (i.e. transferrin (37), fibrinogen (38), ␣ 1 -antitrypsin (39), haptoglobin (40) etc.) are glycosylated by these glycans at high levels. The next three most abundant glycans were core fucosylated neutral biantennary glycans, the major oligosaccharides of IgG, the most high abundance glycoprotein in human serum (41). In addition, oligosaccharides with higher branches (tri-and tetra-antennary) were detected at relatively low rates. Note that six of these low abundance N-glycans (oligosaccharides 9, 17, 21, 24, 28, and 31 in Table II), including sialylated tetra-antennary oligosaccharides, were detected for the first time when whole serum glycomics was analyzed. This analysis may be attributable to the improved deglycosylation efficiency seen under optimized digestion condition.

TABLE I Major oligosaccharides in each peak
Absolute quantitation of major N-glycans present in normal human sera was performed by first quantitating the major N-glycan (oligosaccharide 2 in Table I and Fig. 2) using HPLC and then analysis of the relative signal strength of each glycan obtained from MALDI-TOF MS analysis. Absolute quantitation of the major N-glycan was performed by an absolute calibration method following injection of known concentrations of the same oligosaccharide onto the HPLC system. The estimated absolute concentration of each major N-glycan present in normal sera A-C is summarized in Table II. To assess the reproducibility of the quantitation methods used, which include methyl esterification, aoWR derivatization, and quantitation by MALDI-TOF analysis, these procedures were performed in triplicate for the analysis of normal serum C, and mean values are shown with S.D. The reproducibility was found to be reasonably good because S.E. values were mostly within 20% for those signals whose absolute concentrations were higher than 6 M (which corresponds to ϳ1% of total N-glycan concentration). The total concentration of Nglycans was calculated to be 700 -850 M. To the best of our knowledge, the absolute concentration of N-glycans in human whole serum was determined for the first time with estimation of variability associated with several healthy controls. Given that a large proportion of core fucosylated neutral biantennary glycans (oligosaccharides 3, 6, and 11 in Table II) is derived primarily from IgG, which has two N-glycosylation sites per molecule, the calculated total concentration of those oligosaccharides (50 -75 M) is well within the range of serum IgG concentration (50 -100 M). To detect changes in glycan expression profiles of less abundant glycoproteins, further study to effectively prefractionate minor components of serum is in progress in our laboratory.
We also analyzed the serum N-glycomic profile of a patient with RA (Supplemental Fig. S1 and Supplemental Table S1). We observed that the concentration of one agalactosylated glycan (oligosaccharide 3 in Fig. 4 and Table II) was about 7-12 times higher in a patient with RA than those of normal subjects, most likely reflecting the well documented hypogalactosylation of serum IgG in RA (6,42). It was also FIG. 3. Comparison of PNGase F digestion conditions on N-glycan release efficiencies (a and b) and N-glycan profiles of human  serum glycoproteins (c). Mean Ϯ S.D. (n ϭ 3), 1 mean Ϯ range (n ϭ 2). 2 Only reduction with 2-mercaptoethanol was performed. Optimized concentrations employed for HSD, PHL, and PHM were 0.2, 0.02, and 0.002%, respectively; thus PHL and PHM appeared to be more effective than HSD in improving PNGase F-catalyzed deglycosylation. Red. Alkyl., reductive alkylation. The error bars represent the S.D. value.  Table II. observed that the calculated total concentration of core fucosylated neutral biantennary glycans (oligosaccharides 3, 6, and 11 in Table II), three of major N-glycans of IgG, was ϳ270 M, which is 3-6 times higher than that in normal subjects. This may be attributable to the previous findings that the mean IgG level was raised above normal values (43) and that the increase reflects unknown autoimmune reactions in the early stage of RA (44). Thus, the described absolute quantitation may lead to improved sensitivity in diagnosing RA because it allows estimating the concentration of oligosaccharides that properly reflects the relative abundance of glycoprotein(s) in a disease state.
Kinetics Analysis of PNGase F Using Human Serum as Substrate-Speed is required to apply glycomics analysis to clinical applications such as analysis of disease diagnosis and prognosis. To shorten the time required for deglycosylation of human serum glycoproteins, enzyme kinetics analysis was performed using whole serum glycoproteins as substrate following denaturation in Condition G. Although deglycosylation by PNGase F is known to follow Michaelis-Menten kinetics, the enzyme kinetics has not been elucidated when gross human serum glycoproteins were used as substrates. Based on the concentration (ϳ0.8 mM) of N-glycosylation sites present in whole serum glycoproteins estimated above, K m app and V max values were determined to be 2.40 ϫ 10 1 M and 7.12 ϫ 10 Ϫ2 M/min, respectively (Fig. 5a). These values are in good agreement with the range previously reported for several glycopeptides (30).
As shown in Fig. 5b, the reaction time course simulated from the estimated kinetics parameter showed fairly good agreement with experimental data. Both simulated and experimental data indicate that the amount of recovered N-glycans reached a plateau between 12 and 24 h under the conditions used. Fig. 5c shows the time course of recovery of each major individual N-glycan. Deglycosylation rates differ substantially among these glycoforms: those of sialylated glycans tended to be slower than those of neutral glycans. However, considering that it is understood that the structure of glycoforms has

Quantitative Glycomics of Human Serum Glycoproteins
little effect on PNGase F activity (30), the different deglycosylation rates likely reflect varying susceptibility of serum glycoproteins to PNGase F. This observation again promotes awareness that the quantitative serum glycomic profile can be significantly affected by digestion conditions. Based on kinetics parameters, enzymatic activity was simulated under various conditions (i.e. the amount of serum and enzyme or reaction time). Accordingly we experimentally confirmed that PNGase F-catalyzed deglycosylation times could be shortened substantially while maintaining recovery efficiency identical to that achieved by Condition G followed by 24 h of PNGase F digestion (data not shown). Under the optimized protocol, 5 l of human serum is pretreated according to Condition G (Ͻ3 h) followed by digestion with 10 units of PNGase F only for 2 h. We believe that the present protocol ensures maximal deglycosylation of general glycoproteins and should meet requirements of high throughput clinical glycomics.
In the present report, we optimized conditions of PNGase F-catalyzed deglycosylation suitable for high throughput human serum glycomics based on quantitative comparisons of typical digestion conditions and a detailed kinetics study. We show that deglycosylation efficiency can differ significantly depending on conditions used, thus indicating the importance of a standardized protocol for the accumulation and compar-ison of glycomics data. Furthermore relative and absolute quantitation of human whole serum glycomics was achieved for the first time, providing a firm basis to explore the clinical glycomics research. Glycomics and proteomics are two of the few options available for identifying serum biomarkers, because DNA-or RNA-based diagnostics are not applicable to serum in which no corresponding genome or transcriptome exists. In this regard, it is worthy to mention that serum glycomics analysis offers more than just a means to explore disease-related glycan markers because the discovery of a unique glycan expression profile promptly provides a valuable strategy in developing a particular glycoform-focused reverse genomics.