Isobaric Multiplex Labeling Reagents for Carbonyl-Containing Compound (SUGAR) Tags: A Probe for Quantitative Glycomic Analysis

Abstract

Glycans are highly complex entities with multiple building units and different degrees of branched polymerization. Intensive research efforts have been directed to mass spectrometry (MS)-based qualitative and quantitative glycomic analysis due to the important functions of glycans. Among various strategies, isobaric labeling has become popular because of its higher multiplexing capacity. Over the past few years, several isobaric chemical tags have been developed for quantitative glycomics. However, caveats also exist for these tags, such as relatively low reporter ion yield for aminoxyTMT-labeled complex glycans. To overcome the limitations of existing isobaric chemical tags, we designed a class of novel isobaric multiplex reagents for carbonyl-containing compound (SUGAR) tags that can be used to label glycans for quantitative glycomic analysis. The quantitative performance including labeling efficiency, quantification accuracy, and dynamic range of these SUGAR tags has been evaluated, showing promising results. Finally, the 4-plex SUGAR tags have been utilized to investigate N-glycan changes of B-cell acute lymphoblastic leukemia (ALL) pediatric patients before and after chemotherapy.

GRAPHICAL ABSTRACT

Macromolecules are highly complex entities with different degrees of polymerization and a plethora of building blocks. Precise quantification of biomolecular changes between distinct physiological states is essential for understanding complex biological processes. Over the past decades, mass spectrometry (MS)-based qualitative and quantitative approaches have become one of the most popular means to reveal relative and absolute abundances of various biomolecules. The techniques are well established for both label-free and label-based strategies. Label-free strategies quantify abundances by signal intensities or spectral counting while label-based strategies also utilize signal intensities but offer the advantage of sample multiplexing.

There are three major types of isotope label-based strategies: isotopic, mass defect, and isobaric.¹ Isotopic and mass defect labeling strategies rely on either precursor ion intensity or area under the curve in MS¹ spectra for relative quantification. However, multiplexing capacity is typically limited to triplex comparisons due to increased spectral complexity for isotopic labeling and the requirement of ultrahigh resolving power for mass defect labeling. Due to higher multiplexing capacity, isobaric labeling has become a more widely implemented strategy for high-throughput quantitative analysis. Isobaric tags mainly consist of a reporter, a balancer, and a reactive group. Accurate relative quantification with isobaric tags relies heavily on the yield of reporter ions upon fragmentation. Furthermore, metallic element chelated tag labeling was developed recently for quantification analysis on MALDI instruments.²

Glycosylation is a prevalent post-translational modification that plays essential roles in biological processes, such as cell signaling and immunity response.³ Abnormal glycosylation is relevant to diseases including cancer, cardiovascular problems, and immunological disorders.⁴ However, native glycan analysis is challenging due to lack of chromophore for optical detection as well as poor ionization for MS detection. Several strategies have been developed to overcome the inherent limitations. The most widely used fluorescence reagents, such as 2-aminobenzoic acid (2-AA) and 2-aminobenzamide (2-AB),⁵ enable optical detection. With the development of label-based MS relative quantification, glycan structural characterization as well as relative quantification can be performed in a high-throughput manner. A few isobaric tags, including amino-xyTMT, iART, and QUANTITY,^6–8 have been developed in the past decade and relative quantification capabilities have been reported. However, caveats also exist for these tags, such as relatively low reporter ion yield for aminoxyTMT-labeled complex glycans. Although an additional multinotch MS³ scan has been reported to address this issue,⁹ the increased cycle time coupled with specially required instrumentation limits the utility and broad applicability of this approach. To overcome these limitations, here we developed a class of novel isobaric multiplex reagents for carbonyl-containing compound (SUGAR) tags for glycomics study.

EXPERIMENTAL SECTION

Materials and Reagents.

Methanol (MeOH), ethanol (EtOH), acetonitrile (ACN), dichloromethane (DCM), dimethyl sulfoxide (DMSO), acetic acid (AA), formic acid (FA), and water were purchased from Fisher Scientific (Pittsburgh, PA). Formaldehyde, sodium cyanoborohydride, N-(3-(Dimethylamino)propyl)-N′-ethylcarbodiimide hydro-chloride (EDCI), N-methylmorpholine (NMM), 1-hydroxybenzotriazole hydrate (HOBt), glycine methyl ester hydro-chloride, hydrazine, triethylammonium bicarbonate buffer (TEAB, 1.0 M), and tris(2-carboxyethyl)phosphine hydro-chloride (TCEP) were purchased from Sigma-Aldrich (St. Louis, MO). PNGase F was purchased from Promega (Madison, WI). Oasis HLB 1 cc³ (30 mg) extraction cartridges were purchased from Waters Corporation (Milford, MA). Bovine thyroglobulin (BTG) was provided by Thermo Fisher Scientific (Rockford, IL). Microcon-30 kDa centrifugal filters (30K MWCO) were purchased from Merck Millipore Ltd. (Darmstadt, Germany). PolyGLYCOPLEX A beads (3 μm) were purchased from PolyLC Inc. (Columbia, MD). Fused silica capillary tubing (inner diameter 75 μm, outer diameter 375 μm) was purchased from Polymicro Technologies (Phoenix, AZ). All reagents were used without additional purification.

Human Serum Protein Preparation.

The Health Sciences Institute Review Board of the University of Wisconsin-Madison granted the permission to perform this study (2015–0864). Treatment was conducted according to protocols AALL1131 or AALL0932. Serum samples from three B-cell acute lymphoblastic leukemia (ALL) pediatric patients were collected at the following time points: before chemo-therapy and at 1 month, 3 months, and 6 months after the first day of consolidation chemotherapy. The concentration of serum protein was determined by microBCA assay.

N-Glycan Release by Filter-Aided N-Glycan Separation (FANGS).

N-Glycans of protein samples were released using FANGS¹⁰ with minor modifications. Briefly, protein samples were dissolved at a concentration of 1 μg/μL in 0.5 M TEAB buffer. TCEP (0.5 M, 5 μL) was added to the solution, which was then heat-denatured by switching sample tubes between 95 °C and room temperature water baths four times (15 s each). A 30 K MWCO filter was used to exchange 200 μL of 0.5 M TEAB buffer three times. The prepared protein samples on the MWCO filter were then incubated with 4 μL of PNGase F and 96 μL of 0.5 M TEAB for 16 h at 37 °C water bath. The released glycosylamines were separated from the deglycosylated proteins by centrifuging: glycosylamines were collected into the bottom tube and the deglycosylated proteins remained above the filter. The filter was washed with 100 μL of 0.5 M TEAB buffer for reconstituting the deglycosylated proteins. Both fractions were dried in vacuo. To convert glycosylamine to glycan (with free reducing end), 200 μL of 1% AA was added to the glycosylamine fraction, incubated for 4 h, and dried in vacuo.

N-Glycan SUGAR Labeling and Cleaning Up.

The synthesis of the SUGAR tag is described in Supporting Information. The starting materials for the synthesis of 4-plex SUGAR tags and NMR data are available in Figures S1−S3. SUGAR labeling reactions were performed using a stepwise strategy. Released N-glycans were mixed with 1 mg of SUGAR tag in 100 μL of MeOH containing 1% FA. After 10 min incubation, the solvent was removed in vacuo. Then, 100 μL of 1 M NaBH₃CN in DMSO:AA (7:3 v/v) was added to N-glycans. The reduction was performed at 70 °C for 2 h. The labeling reaction was cooled before cleanup.

An Oasis HLB 1 cc cartridge was used to remove excess labels and purify the labeled N-glycans. The cartridge was conditioned with 1 mL of 95% ACN, 1 mL of water, and 1 mL of 95% ACN. The crude reaction mixture was quickly loaded to the conditioned cartridge which was prefilled with 1 mL of 95% ACN. The cartridge was then washed with 1 mL of 95% ACN three times, and the labeled N-glycans were eluted with 1 mL of 50% ACN and 1 mL of water. The eluted fractions were combined, dried in vacuo, reconstituted in 50 μL of 75% ACN, and analyzed by MALDI-MS or LC-MS/MS immediately.

MALDI-MS Analysis for Labeling Efficiency Calculation.

Samples were prepared by premixing 1 μL of SUGAR-labeled N-glycans with 1 μL of 2,5-dihydroxybenzoic acid matrix (150 mg/mL in 2% N,N-dimethylaniline, 49% MeOH, and 49% water), and 1 μL of each matrix/sample mixture was spotted onto the MALDI target plate. A MALDI-LTQ-Orbitrap XL mass spectrometer (Thermo Scientific, Bremen, Germany) was used for characterizing labeling efficiency. Ionization was performed using a laser energy of 15 μJ. Spectra were acquired in the Orbitrap mass analyzer within a mass range of m/z 1000−4000 at a mass resolution of 30 K (at m/z 400).

LC-MS/MS Analysis.

A self-fabricated nano-HILIC column (15 cm, 75 μm i.d., 3 μm PolyGlycoPlex A HILIC beads) was used for glycan separation. A Dionex Ultimate 3000 nanoLC system was coupled to Q Exactive HF Hybrid Quadrupole-Orbitrap Mass Spectrometer (Thermo Scientific, Bremen, Germany) for all LC-MS/MS analyses. Mobile phase A was deionized water with 0.1% FA, and mobile phase B was ACN with 0.1% FA. The flow rate was set at 0.3 μL/min, and the injection volume was 2 μL. The following gradient was used (time, % mobile phase B) unless otherwise specified: (0 min, 75%), (18 min, 75%), (78 min, 15%), (78.1 min, 10%), (90 min, 10%), (90.1 min, 75%), (100 min, 75%).

The following MS parameters were used for all data acquisition. Samples were ionized in positive ion mode with a spray voltage of 2 kV. S-lens RF level was set to be 55, and capillary temperature was set to be 275 °C. Full MS scans were acquired at m/z 500−2000 with resolving power of 30 K (at m/z 200). Maximum injection time of 100 ms, automatic gain control (AGC) target value of 1e6, and 1 microscan were used for full MS scans. Top 20 data-dependent MS² analysis was performed at a resolving power of 15 K (at m/z 200) with collision-induced dissociation (CID) operating with a normalized collision energy of 30. The first mass was fixed at m/z 110, and dynamic exclusion of acquired precursors was set to 30 s with a ± 10 ppm tolerance.

N-Glycan Data Analysis.

SUGAR-labeled N-glycans were identified by accurate mass matching. A peak list was exported and compared against an in-house database including most possible combinations of N-glycan units (Hexose (H), HexNAc (N), Fucose (F), and NeuAc (S)) with a mass tolerance of 10 ppm. Reporter ion intensities in MS² were used for relative quantification. Microsoft Excel was used for calculations and statistical analyses.

RESULTS AND DISCUSSION

With the successful development of N,N-dimethyl leucine tags (DiLeu)^11–13 for quantitative proteomics by N-terminal labeling, the dimethylleucine structure has been shown to produce abundant backbone fragments with high-intensity reporter ions for quantitative analysis. To implement conjugation with reducing end of glycans, the structure of SUGAR tags consists of a hydrazide as the reactive group with glycine as a balancer. Using naturally occurring amino acids as building blocks with straightforward functional group transformations, SUGAR tags can be synthesized in three steps (Scheme 1) with 67% overall yield.

Scheme 1.

Synthesis of SUGAR Tag with Amino Acid Building Blocks

In addition, the amino acid building blocks offer various commercially available isotope configurations to enable multiplexing capacity. By incorporating different heavy isotopes, 4-plex SUGAR tags with reporter ion mass difference of 1 Da can be readily synthesized and are shown in Figure 1. The multiplexing capacity can be further increased to 12-plex by employing mass defect concept (Figure S4). However, the subtle mass differences between reporter ions require higher resolving power. Future efforts will demonstrate the utility of 12-plex tags for high throughput quantitative analysis.

Figure 1.

Structure and isotope configurations of 4-plex SUGAR tags. Purple dot: ¹³C, orange dot: ²H, red dot: ¹⁵N.

Hydrazide chemistry is a commonly used conjugation for glycan reducing end labeling which converts hydrazide to hydrazone group. However, the reversible nature of hydrazide chemistry makes complete conjugation arduous. Thus, an alternative irreversible reductive amination is used for glycan reducing end labeling. Despite the high reactivity of the hydrazide group, a high concentration of a reducing agent, such as 1 M NaBH₃CN, can reduce glycans prior to hydrazone formation and yield reduced glycans. To further improve labeling efficiency with minimal sample loss, a stepwise reductive amination has been developed. Maltooctaose standard was mixed with SUGAR tag in methanol with 1% formic acid for 10 min to complete hydrazone formation. Then reduction was performed with NaBH₃CN in 7:3 (v/v) dimethyl sulfoxide:acetic acid to achieve complete labeling (Scheme 2). The stepwise reductive amination improved labeling efficiency by preventing glycan reduction prior to hydrazone formation and reduced labeling time by half (Figure S5).

Scheme 2.

SUGAR Tag Stepwise Labeling and Reporter Ion Generation

SUGAR tag with stepwise reductive amination was further applied to the labeling of bovine thyroglobulin (BTG) standard. The workflow is shown in Figure 2. N-Glycans released from BTG by PNGase F were labeled with SUGAR tags. Under optimized labeling conditions, SUGAR tags offer nearly complete labeling for different types of glycans including high mannose, complex/neutral, fucosylated, and acidic glycans. Figure 3 highlights several N-glycan labeling comparisons. The labeling efficiencies for other N-glycans are expected to be similar due to same shared core structure of N-glycans. Labeling efficiencies shown in Figure S6 are calculated with following equation: labeling efficiency = labeled peak intensity/(labeled peak intensity + unlabeled peak intensity) × 100%.

Figure 2.

Workflow of quantitative glycomics analysis by 4-plex SUGAR.

Figure 3.

MALDI-MS spectra of N-glycans (Na⁺ adduct) released from BTG: unlabeled (A) and SUGAR-labeled (B). Peaks associated with the same glycans are highlighted in the same shade.

To maximize the fragmentation performance of SUGAR-labeled N-glycans, higher-energy collisional dissociation (HCD) performed with different normalized collision energies (NCE) was examined. Most backbone fragments were observed with NCE of 20−25. The reporter ion intensities were elevated with minimal loss of backbone fragment ions by increasing NCE to 30. At even higher NCE, the reporter ions became base peak with a loss of the majority of backbone fragments. Thus, NCE of 30 was chosen to yield high intensities of reporter ions along with abundant backbone fragments. The fragmentation behavior comparisons against aminoxyTMT-labeled glycans are shown in Figure 4. SUGAR-labeled N-glycans tend to produce higher intensities of reporter ions and preserve more backbone fragments. Acidic N-glycans play important biological functions such as stability, degradation, and antigenicity,¹⁴ thus attracting a great deal of research interests. However, acidic N-glycans often produce fewer reporter ions than that of neutral N-glycans. With optimized NCE, SUGAR-labeled acidic N-glycans produced more than a 4-fold increase of reporter ion intensities upon fragmentation, demonstrating the SUGAR tag’s suitability for acidic N-glycan quantitative analysis.

Figure 4.

ESI-MS/MS fragmentation comparison of aminoxyTMT-labeled and SUGAR-labeled N-glycans. AminoxyTMT-labeled H₈N₂ ([aminoxyTMT − H₈N₂ + 2H]²⁺) at NCE 25 (A) and 30 (C), SUGAR-labeled H₈N₂ ([SUGAR − H₈N₂ + 2H]²⁺) at NCE 25 (B) and 30 (D), AminoxyTMT-labeled H₆N₄FS ([aminoxyTMT − H₆N₄FS + 2H]²⁺) at NCE 25 (E), and SUGAR-labeled H₆N₄FS ([SUGAR − H₆N₄FS + 2H]²⁺) at NCE 25 (F).

Quantification performance of the 4-plex SUGAR tags was evaluated by labeling N-glycans at known ratios. N-Glycans released from BTG were aliquoted into four portions with known ratios at 1:1:1:1, 1:1:5:10, and 10:5:1:1 in triplicate and then labeled with 4-plex SUGAR tags. The intensities of reporter ions in MS² spectra for each glycan were used to calculate the experimental ratios. In Figure 5A, experimental 4-plex SUGAR-labeled ratios for N-glycans are plotted against theoretical ratios 1:1:5:10. Representative MS² reporter ions are shown in Figure 5B,C for two SUGAR-labeled N-glycans. The representative MS² reporter ions for ratios of 1:1:1:1 and 10:5:1:1 are shown in Figure S7. For all three known ratios, less than 15% relative errors are observed with standard deviations of 0.2, 0.18, and 0.22, demonstrating that the SUGAR quantification approach offers accurate quantitative results. Moreover, no retention time shift is observed for 4-plex SUGAR tag labeled glycans with HILIC column in Figure S8.

Figure 5.

Relative quantification performance of 4-plex SUGAR-labeled N-glycans released from BTG. Labeled N-glycans were mixed at ratios of 1:1:5:10 and analyzed in triplicate. The reporter ion intensity ratio results of 116/115, 117/115, and 118/115 are plotted at the log scale. Box plots show the median (line), the 25th and 75th percentile (box), and the 5th and 95th percentile (whiskers) (A). Representative MS spectra reporter ion range for SUGAR-labeled N-glycans H₄N₃FS (B) and H₆N₄FS (C) are shown.

As dynamic expression of N-glycans is highly relevant to biological processes, we employed the SUGAR tags to quantify N-glycans extracted from a complex biological system at different biological states. Acute lymphoblastic leukemia (ALL) is one of the most predominant cancers in children, accounting for 26.8% childhood cancer diagnoses worldwide.¹⁵ The use of central nervous system directed chemotherapy has increased the 5-year-event-free survival rate to around 80% in standard-risk ALL.¹⁶ However, it has been reported that childhood cancer survivors suffer from neurobehavioral morbidity including diverse aspects of cognitive function, attention, processing speed, memory, academic achievement, and emotional health, all of which negatively impact their quality of life.¹⁷ A previous study¹⁸ indicated that the neurotoxicity of chemotherapy could be assessed by studying alterations of protein expression levels. For example, the Tau protein level in cerebrospinal fluid (CSF) serves as a biomarker of neuronal loss during active treatment for ALL. Limited work¹⁹ has been done on quantitative glycomics during chemotherapy. It is important to investigate glycan level changes during the treatment to potentially facilitate biomarker discovery and lead to elucidation of pathogenesis mechanisms and evaluation of treatment outcomes.

N-Glycan changes in human serum proteins of three B-cell ALL pediatric patients were compared before induction and one month, three months, and six months after the chemotherapy. Four-plex SUGAR tags were used to label N-glycans released from equal amounts of human serum proteins at different time points. Quantitative glycomic analysis was carried out by following the workflow in Figure 2.

Quantitative analyses of selected N-glycans are summarized in Figure 6 with various types of N-glycans. Majority of quantified N-glycans reveal a trend of down-regulation after induction chemotherapy. A full list of N-glycans with fold changes are reported in Table S1. In total, 145 N-glycans were identified and quantified with SUGAR labeling approach. Of these, 68 N-glycans were quantified across all three patients. The observed down-regulated N-glycan expression could be explained by elimination of blasts after chemotherapy. As cancer cell metastatic growth can increase branching, fucosylation, and sialylation of N-glycans,²⁰ chemotherapy could decrease such N-glycans by reverting this process. Indeed, the fucosylated and sialylated N-glycans show significant down-regulation in patients after chemotherapy. The proof-of-principle study with SUGAR approach reveals the macroscopic relationship between chemotherapy and N-glycan expression. Further extensive and more in-depth investigations are needed to explore pathogenesis mechanisms and treatment outcomes.

Figure 6.

Selected N-glycan relative quantification of equal amounts of human serum protein from ALL patients before (SUGAR-115) and 1 month (SUGAR-116), 3 months (SUGAR-117), and 6 months (SUGAR-118) after induction chemotherapy. Ratios represent intensities of reporter ions for SUGAR-labeled N-glycans. Error bars represent the standard deviation of three biological replicates.

CONCLUSIONS

In summary, new isobaric SUGAR tags with amino acid building blocks were developed in this study with improvements in the following aspects: low cost, high yield, complete labeling, high reporter ion yield, accurate and precise quantification, and applicability for complex samples. A hydrazide reactive group enabled glycan reducing end conjugation while stepwise reductive amination strategy was developed to achieve complete glycan labeling. The fragmentation of SUGAR-labeled glycans preserve more backbone fragments with higher reporter ion intensities for qualitative and accurate quantitative glycomics. SUGAR tags also show accurate quantification with broad dynamic range. Furthermore, the 4-plex SUGAR tag quantitation approach was applied to a complex biological system to investigate N-glycan changes of B-cell ALL pediatric patients prior to and after chemotherapy. It was found that most N-glycans were down-regulated after chemotherapy, possibly due to cancer cell reduction. The successful development of SUGAR tags offers a useful chemical tool for implementation in many biological and clinical applications. The simple building blocks, complete conjugation, desired fragmentation, and accurate quantification make it a precise contrivance for quantitative glycomics study. In conclusion, we anticipate that the novel SUGAR tagging approach can be widely applied in multiple areas of biomedical research.