Imidazolium labelling permits the sensitive mass-spectrometric detection of N-glycosides directly from serum†

A novel imidazolium derivative (GITag) shows superior ionisation and consequently allows increased mass spectrometric detection capabilities of oligosaccharides and N-glycans. Here we demonstrate that human serum samples can be directly labelled by GITag on a MALDI target plate, abrogating prevalently required sample pretreatment or clean-up steps.


Materials and methods.
1-Hydroxylethyl-3-methylimidazolium tetrafluoroborate was obtained from Energy Chemicals Co. (Nanjing, China); N,N'dicyclohexylcarbodiimide (DCC) was obtained from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China); 4dimethylaminopyridine (DMAP) was obtained from J&K Chemicals Co. (Beijing, China); N-methylimidazolium, sodium cyanoborohydride (NaBH3CN), N-acetyl-D-glucosamine (GlcNAc) and Lactose monohydrate were supplied by Aladdin Chemicals Co. (Shanghai, China); Human serum was obtained from Nanjing General Hospital in accordance to the ethical provision number 2017NZKY-023-02. Recombinant PNGase F was expressed and purified as reported previously (Wang et al, Biosci. Rep. 2014, 34, e00149); Other bulk chemicals were obtained from commercial suppliers without further purification or modification. NMR spectra were registered on a Bruker AV-400 instrument or a Bruker AV-500 instrument using the residual solvent signal as the internal standard at 298 K. NMR data were processed using MestReNova (version 9.0.1). Chromatographic analyses were performed using a Nexera UPLC-FLD system coupled to an LCMS 8040 ESI mass spectrometer (both from Shimadzu Company, Kyoto, Japan). Array NMR experiments were recorded on a Varian 500 MHz spectrometer. High resolution Electrospray ionisation (ESI) mass spectra were recorded on a Micromass LCT mass spectrometer or a VG Quattro mass spectrometer.

Fluorimetric characterization of GITag-Lactose (11).
GITag-Lactose (11) was dissolved in water (final concentration 10 nM), and the excitation and emission spectra of 11 were recorded on a FluoroMax-4 fluorescence photometer (HORIBA, France). For the emission spectrum, the excitation wavelength was maintained at 350 nm and the emission wavelength was scanned from 325 to 500 nm. For the excitation spectrum, the emission wavelength was kept at 370 nm and the excitation wavelength was scanned from 200 to 350 nm ( Figure S1) Figure S1. Emission and excitation spectra of GITag-Lactose (11).

ESI-ToF-LC/MS and MALDI-ToF analysis and quantifications of 2AB and GITag labelled carbohydrates (8-11).
S3 intensities were measured for compounds 8 and 10 at Ex/Em wavelengths of 330/420 nm and for compounds 9 and 11 at 304/368 nm, respectively. Solvent A was 50 mM NH4COOH (pH 4.5) in water, and solvent B was acetonitrile. A linear gradient of 12−20% solvent B was applied from 0 to 3 min; then, solvent B was increased to 95% over 1 min and held at 95% for 2 min. Solvent B was then decreased to 12% in 1 min, and the column was equilibrated with the initial conditions for 3 min. The compounds 8, 9, 10 and 11 appeared at 6.9, 5.7, 5.6 and 4.7 min, respectively. Positive mass signals for 8 (

S6
Serial dilutions of compounds 8 -11 (1 μL) were spotted on a stain-less-steel MALDI plates and followed by adding 1 μL of 2,5dihydroxybenzoic acid (DHB) solution (20 mg/ml in 30:70 (v/v) acetonitrile:aqueous TFA (0.1%)). After the samples were dried, they were then analysed on a Bruker Autoflex Speed instrument (equipped with a 1000 Hz Smartbeam-II laser). Mass spectra were analysed by using Bruker Flexanalysis software version 3.3.80 (Supporting Figure S5). The correlation coefficient and linear range are shown in Table S1. The limit of detection (S/N=3) and quantification (S/N=10) of 8 -11 were determined using the signal intensity areas of the graphs shown in Supporting Figure S3, S4 and S6.

Application of GITag analysing N-glycans derived from human serum
Human serum was used without any pretreatment, 50 μL human serum, 60 μL sodium phosphate buffer (500 mM, pH 7.5), recombinant PNGase F (135 μg) and 50 μL GITag derivatisation solution (35 mM GITag and 0.1 M sodium cyanoborohydride in methanol/acetic acid solution (7:3, v/v)) were sequentially added directly on a MALDI-ToF sample carrier (brushed stainless steel) without the need for any sample transfer and centrifugation steps. The MALDI target with the sample mixture was put into a plastic petri dish and incubated at 37 °C for 12 h. Then the MALDI target was transferred to an incubator with the temperature of 55 °C until the sample was dried (4 h). The dried sample was resuspended by 200 μL deionized water, and two μL of the sample transferred on an unused sample spot on the MALDI-Target. After drying and overlaying the samples with 2 μL of DHB matrix (20 mg/ml in 30:70 (v/v) acetonitrile:aqueous TFA (0.1%)), the samples were directly subject to MALDI-ToF mass spectrometric analysis. All thirty-two major mass signals could be identified as either complex-(25 species), high-mannose-(5 species) or hybrid-type (2 species) N-glycans (Supporting Table S2).  In a first experiment, the labelling of the serum with 2AB and the GI-Tag was performed without any pretreatment ( Figure S7A).

S9
A second derivatisation experiment was not directly performed on a MALDI-ToF sample carrier but in a 1.5 mL sample vial; after the initial enzymatic N-glycan release by PNGase F (37 °C for 12 h), the insoluble part of the sample was removed by centrifugation (14000 g for 5 min). The clear supernatant was then incubated with 50 μL GITag or 2AB derivatisation solution (35 mM GITag or 2AB and 0.1 M sodium cyanoborohydride in methanol/acetic acid solution (7:3, v/v)) and at 55 °C (65 °C for 2AB derivatisation) for 4 h. Two μL of the sample were transferred onto the MALDI-Target, and after drying and overlaying the samples with 2 μL of DHB matrix (20 mg/ml in 30:70 (v/v) acetonitrile:aqueous TFA (0.1%)), the samples were subject to MALDI-ToF mass spectrometric analysis ( Figure S7B).
In a third experiment the human serum samples were subject to a solid-phase-extraction (SPE) cleanup-step after the initial

S11
enzymatic N-glycan release; This was achieved by isolating the released N-glycans using ENVI-Carb solid-phase extraction columns (500 mg bed volume, Supelco). These columns were pre-conditioned with 3 ml of deionized water, followed by 3 ml of 80% ACN containing 0.1% TFA (v/v) and finally re-equilibrated with 3 ml of deionized water. Samples containing enzymatically released N-glycans were loaded onto the cartridge and washed with 1.5 ml of water. The N-glycans were eluted using 1.5 ml of 40% acetonitrile containing 0.1% TFA (v/v). Then, samples were dried using centrifugal evaporation. The dried samples were then incubated with either 50 μL GITag or 2AB derivatisation solution (35 mM GITag or 2AB and 0.1 M sodium cyanoborohydride in methanol/acetic acid solution (7:3, v/v)) and at 55 °C (65 °C for 2AB derivatisation) for 4 h. Two μL of the sample were transferred onto the MALDI-Target, and after drying and overlaying the samples with 2 μL of DHB matrix (20 mg/ml in 30:70 (v/v) in aqueous acetonitrile: TFA (0.1%)), the samples were subject to MALDI-ToF mass spectrometric analysis ( Figure S7C).

Comparison of the detection efficiency of human serum N-glycans between carbohydrate tags via UPLC-FLD-MS analysis.
In a third experiment the human serum samples were subject to a solid-phase-extraction (SPE) cleanup-step after the initial enzymatic N-glycan release (Details described in Section 4.1.). The dried serum N-glycans samples were then incubated with either 50 μL GITag or 2AB derivatisation solution (35 mM GITag or 2AB and 0.1 M sodium cyanoborohydride in methanol/acetic acid solution (7:3, v/v)) and at 55 °C (65 °C for 2AB derivatisation) for 4 h. The derivatised N-glycan samples (10 µl) were mixed with 40 µl of acetonitrile. These mixtures were then injected into a UPLC-FLD-MS system (Nexera, Shimadzu Corporation, Kyoto, Japan) and profiled using a hydrophilic interaction liquid chromatography (HILIC) column for the separation of the analytes (Acquity BEH Glycan Column, 2.1×150 mm, 1.7 µm particle size; Waters, Ireland) at a column temperature of 60˚C.
The HPLC system consisted of an LC-30AD pump system, an RF-20Axs fluorescence detector set at excitation/emission wavelengths of 330/420 nm for 2AB-labelled N-glycans and 304/368 nm for GI-Tag-labelled N-glycans, respectively. Solvent A was 50 mM aqueous ammonium formate buffer (pH 4.5), and solvent B was acetonitrile. A linear gradient of 95-78% of B was applied from 0 to 6 min, and solvent B was then decreased to 55.9% over 38.5 min, with the flow rate set to 0.5 ml/min. The mass spectrometric analysis was performed using a 8040 ESI-ToF detector using positive SIM mode ( Figure S8A-D and Table   S3).

Comparison of the derivatisation efficiency between carbohydrate tags via 1 H NMR analysis.
In order to compare the derivatisation efficiency of 2AB and the GITag with carbohydrates, array NMR experiments were performed following the evolution of the reaction over time in a Varian 500 MHz spectrometer, all samples were prepared at the same concentration, using the same batch solvents and conditions. The spectra were acquired at 65 °C with 8 scans at intervals of 30 minutes using the same gain level. The time delay from the additions of reagents into NMR tubes involving sample manipulation and NMR tuning and shimming were minimized as much as possible and the first 10 minutes from each manipulation (addition of tag or addition of NaBH3CN) were not considered.
The composition of the NMR tube used for each experiment is resumed in table S3. µmol) The reductive amination between GlcNAc 2 or Lac 3 with 2AB or GITag was carried out at 65 °C in a deuterated solvent system composed by DMSO-d6 and AcOH-d3 in a 7:3 ratio. The reaction was carried out in a close NMR tube where the chosen sugar at a fixed concentration (Table S3) was allowed to equilibrate with the amine tag to form the corresponding imine before reduction with NaBH3CN.
Conversion % was calculated using DMSO-d5 residual signal as internal standard. For each experiment the mmol of residual DMSO-d5 were calculated as a function of the known sugar amount added in each NMR tube (considered as cumulative integration of the α and β H-1 signals) as: Where mmolDMSO are the mmoles of residual DMSO-d5, mmolSug are the mmoles of the corresponding sugar present in the NMR tube, IDMSO is the integral of the residual DMSO-d5 signal, Iα is the integral of the H-1 α anomeric signal and Iβ is the integral of the H-1 β anomeric signal. The integration ranges are listed in table S4.

S15
mmolDMSO was normalized to the initial volume of deuterated DMSO used: In each NMR spectrum the conversion (χ %) to the product is expressed as function of IDMSO used as internal standard taking into account the addition volumes of the tag and NaBH3CN dissolved in DMSO-d6 as:

%%&' !"#
Where IP is the integral of the H-2 signal of the product (listed in table S5) and mLDMSO is the total volume of deuterated DMSO in the sample after tag and NaBH3CN addition. 350 µL of the DMSO-d6 solution of sugar were mixed with 150 µL of AcOH-d3, the 1 H spectrum was recorded, the T was raised to 65 °C and the tube was allowed to equilibrate for 500 seconds spinning at 20 Hz before acquisition of a new 1H spectrum.
The tube was quickly removed, 1 equivalent of tag was added, the tube was placed again in the spectrometer at a constant T of 65°C and an array of 1H spectra was recorded at intervals of 30 minutes. After 6 hours the tube was quickly removed, 3 equivalents of NaBH3CN were added the tube was placed again in the spectrometer at a constant T of 65°C and an array of 1H spectra was recorded at intervals of 30 minutes. Integral values (arbitrary unit) are listed below in table S6-9. 6. Preparative scale synthesis and characterization.

Synthesis of GITag (1)
DCC (619.0 mg, 3.0 mmol) was added to a reaction mixture of DMAP (36.7 mg, 0.3 mmol), and 1-hydroxylethyl-3-methyl imidazolium tetrafluoroborate (667.6 mg, 3.12 mmol) in acetonitrile (25 mL) and stirred at room temperature for 10 minutes. 4-(boc-amino)benzoic acid (740,3 mg, 3.12 mmol) was added and the solution was stirred at room temperature for 24 h. The white precipitate of dicyclohexylurea was removed by filtration and the clear filtrate solution was concentrated under reduced pressure by rotary evaporation. The residue was triturated with diethyl ether several times until TLC analysis shows complete removal of the starting materials. The residue was dissolved in 5 ml of trifluoroacetic acid/dichloromethane 1:1 (v/v) solution and stirred at room temperature for 20 min. The reaction mixture was concentrated under reduced pressure by rotary evaporation and the liquid residue was co-evaporated three times with 5 ml methanol to remove trifluoroacetic acid residues furnishing 1 (769.0 mg, 74 %) as a yellow oil.  152.8, 136.4, 131.7, 123.7, 122.7, 117.5, 114.7, 62.6, 48.5, 48.5, 35.7, 35.7. ESI-HRMS m/z: calculated for C13H16N3O2 + (M) + : 246.1237, found 246.1235.

General reductive amination procedure.
In a 50 mL sealable round-bottom glass flask a solution of glycoside 2 or 3 (0.5 mmol) in 5 mL of dimethyl sulfoxide/acetic acid (7:3; v:v) was treated with 2AB or GITag (0.5 mmol) and stirred at 65 °C for 3 h. Then, NaBH3CN (88.0 mg 1.4 mmol) was added at 65 °C and the mixture was stirred for 16 h at 65 °C. The solution was cooled down to room temperature and the solvent was removed via lyophilization.