Development of an Antibody Based Platform for the Analysis of Immune Cell Specific N-linked Glycosylation

Abstract

N-linked glycosylation plays an important role in both the innate and adaptive immune response through the modulation of cell surface receptors as well as general cell to cell interactions. The study of immune cell N-glycosylation is gaining interest but hindered by the complexity of cell type specific N-glycan analysis. Analytical techniques such as chromatography, LC-MS/MS and the use of lectins are all currently used to analyze cellular glycosylation. Issues with these analytical techniques include poor throughput, which is often limited to a single sample at a time, lack of structural information, the need for a large amount of starting material and the requirement for cell purification, thereby reducing their feasibility for N-glycan study. Here, we report development of a rapid antibody array-based approach for the capture of specific nonadherent immune cells coupled with MALDI-IMS to analyze cellular N-glycosylation. This workflow is adaptable to multiple N-glycan imaging approaches such as the removal or stabilization and derivatization of terminal sialic acid residues providing unique avenues of analysis that have otherwise not been explored in immune cell populations. The reproducibility, sensitivity and versatility of this assay provides an invaluable tool for researchers and clinical applications, significantly expanding the field of glyco-immunology.

Graphical Abstract

INTRODUCTION

N-glycosylation is a posttranslational modification that can be found on more than 50% of all proteins¹. This modification occurs at Asn-X-Ser/Thr amino acid sequences where a polysaccharide chain of sugar molecules is covalently linked to an asparagine residue via a N-glycosidic bond². N-glycan synthesis is a dynamic and multistep process that is dictated by the expression and availability of glycosidases, glycosyltransferases and sugar nucleotide donors. This multistep pathway results in an array of differentially structured N-glycans that decorate glycoproteins and influence protein folding, molecular trafficking, and signal transduction³. When these glycoproteins are trafficked to the cell membrane of immune cells, such as T cells, they can influence homing, signaling, development, differentiation, activation, and proliferation^4,5. Virtually all cell surface receptors used to characterize immune cells are glycoproteins themselves and glycosylation has been shown to influence receptor function and expression^6,7. Differential states of inflammation and immunosuppression influence immune cell surface N-glycosylation, subsequently altering function and warrant further investigation^8–10.

Immune cell glycosylation is an under studied field, due to the inherent difficulty of analyzing cell specific glycosylation. It is not possible to predict glycosylation patterns based off transcriptomic or proteomic analysis, as glycan additions are untemplated to sites that can hold a glycan and glycans can be modified once glycoproteins are expressed.¹¹ Current analytical techniques used to study immune cell glycosylation are either laborious and time consuming or only provide partial structural data¹². When analyzing specific immune cell populations, oftentimes a homogenous sample of cells is required, further complicating workflows¹³.

Our group has developed slide based matrix assisted laser desorption/ionization imaging mass spectrometry (MALDI-IMS) analytical techniques that allow for the N-glycan analyses of multiple sample types which include tissue, biofluids, targeted protein capture and cultured cells^14–19. One missing approach is the ability to analyze the N-glycan profiles of cells that grow in suspension culture, a common trait of most immune cells. Here we describe an antibody array based approach that captures cells based off cell surface receptor expression (e.g., CD4 or CD8), allowing for captured cells to be taken through N-glycan MALDI-IMS analyses. This assay streamlines workflows, allowing for the specific capture of immune cell sub-types from mixed populations, eliminating the need for negative or positive selection. By leveraging MALDI-IMS, immune cell glycan profiling can be conducted quickly and at scale while simultaneously providing unprecedented amounts of N-glycan structural data, giving further insight into the immunological glycome.

EXPERIMENTAL SECTION

Materials

Amine reactive slides (Nexterion Slide H Hydrogel Coated Glass Slides) were obtained from Applied Microarrays (Tempe, AZ). Well modules were obtained from Grace Bio-Labs (Bend, OR). The rotary tool is a Dremel 200 series. Ammonium bicarbonate, Corning cell strainer 40 μm, Sodium Azide, trifluoroacetic acid, 2F-Peracetyl-Fucose and α-cyano-4-hydroxycinnamic acid and octyl-β-D-glucopyranoside were obtained from Sigma Aldrich (St. Louis, MO). 0.5M EDTA pH 8 was from VWR (Radnor, PA). Peptide-N-glycosidase F Prime TM (PNGase F) and Sialidase Prime TM were obtained from N-Zyme Scientifics (Doylestown, PA). VWR Analog 3500 (89032-092) Orbital shaker 25-500 RPM. Human anti-CD7 (Clone 124-1D1), mouse anti-cd4 (clone GK1.5), mouse anti-cd8 (clone 53-6.7), Streptavidin-Cy5 (Cat. 434316) and ACK Lysing buffer was obtained from Thermo Fisher Scientific (Waltham, MA). Biotinylated mouse anti-cd4 (clone GK1.5) and mouse anti-cd8 (clone 53-6.7) were obtained from BioLegend (San Diego, CA). VECTASHIELD antifade mounting medium with DAPI was obtained from Vector Laboratories (Burlingame, CA). Jurkat cells (Clone E6-1) were obtained from ATCC (Manassas, VA).

Antibody array preparation

Amine reactive slides 25 mm X 75.1 mm) were initially filed to allow slides to fit in the Bruker slide holder plate. Wells were created by attaching a 24-well module overtop of the slide. Antibodies were diluted in PBS and spotted manually in each well at 200 ng per 1.5 μl spot. Spots were left to bind to the slide at room temperature for 1 hr in a preheated humidity chamber made from a culture dish with a Wypall X 60 paper towel and 2 rolled KimWipes saturated with distilled water. During binding, the slide was placed flat in the dish with the amine reactive side facing up, positioned between the two KimWipes. Slides were placed in a dissector until the antibody spots were completely dried. Antibody spots were washed with 0.1% octyl-β-D-glucopyranoside in 1X PBS (PBS-OGS) for 1 min, followed by blocking in 100mM ammonium bicarbonate solution pH 8 for 30 min. Antibodies were then deglycosylated by adding 100μ1 of 10 μg/ml PNGase F PRIME^™ diluted in HPLC water into each well and placed back into the humidity chamber and incubated at 37 °C for 2 hours. Following deglycosylation, antibody arrays were washed with PBS-OGS (3min X 3) with gentle shaking followed by PBS washes (3x) and a water wash (1 min). The deglycosylated antibody array was dried in a dissector prior to cell capture.

Cell Culture

Jurkat cells were cultured in RPMI 1640 medium, containing 10% (v/v) fetal bovine serum, and 1% (v/v) penicillin and streptomycin. Cells were grown at 37 °C in a 5% CO₂ humidified atmosphere.

2FF Treatment

Jurkat cells were plated in a 12 well cell culture plate at a concentration of 5 x 10⁵ cells/ml in a total volume of 2 mls in triplicate. Cells were treated with either water, 50, 100 or 200 μM/ml of 2FF dissolved in water for 3 days before cell capture and N-glycan analyses.

Splenocyte Isolation

C57BL/6 mice were sacrificed, and spleens were retrieved. Spleens were homogenized over a 40 μM strainer into a 50 ml conical. The strainer was washed with 5 ml of cell culture media (2x), allowing for the splenocytes to pass through the strainer and into the 50 ml conical. Cells were pelleted, media was removed and red blood cells were lysed with 1 ml of ACK lysing buffer for 1 min. The ACK reaction was neutralized with 6 ml of cell culture medium. The remaining cells were resuspended in FACS buffer and diluted to a concentration of 1 x 10⁷ cells/ml.

Cell Capture

To reduce the potential for background signal, glycoprotein containing cell culture media is decanted and cells are resuspended in FACS buffer (PBS, 1% BSA, 0.1% EDTA, 0.05% sodium azide) further aiding in capture efficiency. Cells were washed (3x) in FACS buffer and resuspended in FACs buffer. 100 μl of cell suspension was added to each well and a plate sealer cut to size was used to seal the top of the antibody array. The slide was tapped horizontal to a VWR Analog 3500 Orbital shaker. Cell capture was performed at 4 °C, shaking at 250 rpms for 1 hr. Shaking conditions influence capture efficiency and plate shaker models have varying parameters. Shaking conditions should be optimized to the specific shaker used, proper agitation is important for cell capture, especially in a heterogenous sample.. Slides were then washed (2x) with 200 μl of PBS. The 24-well module was then removed and the slide was placed in a slide mailer containing 10% neutral buffered formalin for 20 min. After 20 min the slide was removed and placed in PBS at RT for up to 1 week before further processing.

AAXL and N-glycan Release

Sialic acid stabilization and derivatization was performed via a slide based sequential amidation-amidation reaction with dimethylamine and propargylamine, termed AAXL (Amidation-Alkyne Xtra Linker). This was conducted per our groups previously developed method.²⁰ If slides did not undergo AAXL prior to N-glycan release, fixed cells were delipidated by washing steps of 100% ethanol washes (2 min x 2), Carnoys solution (10% glacial acetic acid, 30% chloroform, and 60% 200 proof ethanol) (10 min x 2), 100% ethanol washes (2 min x 2) and HPLC water washes (3 min x 2). Slides were then dried in a desiccator.

To release N-glycans from captured cells, PNGase F Prime^™ (0.1 μg/μL in HPLC water) was sprayed onto the slide using a M5 TM-Sprayer (HTX Technologies). Spraying parameters consisted of 10 passes at 25 μL/min, 1200 mm/min, 45 C, 3 mm spacing between passes with 10 psi nitrogen gas. Slides were incubated for 2 hr at 37 °C in a preheated humidity chamber. MALDI matrix α-cyano-4-hydroxycinamic acid (CHCA, 7 mg/ml in 50% acetonitrile/0.1% trifluoracetic acid) was sprayed using the same M5 TM-sprayer with 10 passes at 70 μl/min, 1300 mm/min, 79 °C, 2.5 mm spacing between passes with 10 psi nitrogen gas. Two passes of ammonium phosphate monobasic (5mM) was sprayed across the slide to reduce matrix clustering and improve signal. Spraying parameters were performed at 70 μl/min, 1300 mm/min, 60 °C, 3 mm spacing between passes with 10 psi nitrogen gas. For sialidase treated slides, a mixture of Sialidase PRIME and PNGase F PRIME (0.1 mg/ml each) was sprayed using the same parameters as described for PNGase F alone.

MALDI Imaging Mass Spectrometry

N-glycan imaging was conducted using a timsTOF flex MALDI-QTOF mass spectrometer (Bruker) operated in positive ion mode at a m/z range of 700-4,000. Images were collected using a SmartBeam 3D laser that operated at 10,000 Hz using the M5 small smart beam setting at a laser spot size of 100μm run at a raster of 150 μm. 600 laser shots per pixel were collected with an ion transfer time of 120 μs, a prepulse storage of 25 μs, a collision radio frequency of 4000 Vpp, a multipole radio frequency of 500 Vpp and a collision cell energy if 25 eV.

Data Analysis

Data was imported into SCiLS Lab 2022a (Bruker) and normalized to total ion count. Manual N-glycan peak selection was conducted using theoretical mass values based on hexose composition. Putative structures shown in supplemental table S1, S2 and S3 are N-glycoforms built with GlycoWorkBench. Each spot was assigned its own individual region by manually circling the corresponding spot in the SCiLS software. Subsequently, the area under the peak (AUP) was exported from each region. Relative intensity values were calculated by dividing the AUP values of individual N-glycans by the summed AUP of all N-glycans detected. Each slide was spotted with antibody only spots in triplicate, which were deglycosylated and processed in the same way as spots with captured cells. The average signal from deglycosylated antibody only spots were used to subtract background signal from all cell capture spots. After antibody subtraction, if the AUP of an individual N-glycan was less than 2 the value was converted to 1 to avoid 0 or negative values. Data was further processed using Graphpad Prism 9.0.

Cell Count Normalization

Post MALDI imaging, captured cells underwent hematoxylin and eosin (H&E) staining. A cover slip was added and cells were imaged via a Nano-zoomer 2.0-RS high resolution slide scanner (Hamamatsu) at 40x magnification. High resolution images were uploaded to Image J software allowing for cells within each capture spot to be counted. Potential staining artifacts were excluded from analysis by halving the average cell size from a represented spot to use as a lower limit of detection and doubling the average cell size to use as a upper limit of detection. Cell count normalization was conducted by dividing the total number of cells by the total signal intensity observed in each capture spot to acquire AUP per cell.

Immunocytochemistry staining

Fixed captured CD4 and CD8 mouse T cells were washed with PBS 2X and blocked with 3%BSA in 0.05% PBST for 1 hour. Cells were stained with biotinylated anti-CD8 and anti-CD4 antibodies (1/50), suspended in blocking buffer and incubated at 4 C overnight. Slides were subsequently washed 3X with PBST for 5 min under gentle shaking conditions. Streptavidin-Cy5 was suspended in blocking buffer (10μg/ml) and incubated at RT for 1 hr. Slides were washed 2X in PBS-Tween 20 (PBST) for 5 min followed by a 1X PBS wash. DAPI containing mounting medium was used to mount a cover slip. Negative controls consisted of anti-CD4 antibodies staining CD8 captured cells and anti-CD8 antibodies staining CD4 captured cells. Slides were imaged using a EVOS imaging system (Fisher Scientific).

RESULTS AND DISCUSSION

Given the importance of N-linked glycosylation in immune cell function, we wanted to develop a method that would allow for high throughput cell specific glycan analysis of non-adherent immune cells (Figure 1). Previous workflows used for the analyses of glycoproteins captured via antibody arrays were tried for the characterization of immune cell N-glycosylation, but failed due to poor capture efficiency, incompatible slide-based chemistries, and lack of slide transparency. Here we describe the key factors involved in developing and testing the methods required in capturing and imaging one specific type of immune cells, specifically T cells.

Figure 1.

Workflow of antibody array based N-glycan imaging of captured immune cells. A) Antibodies are spotted onto amine reactive slides, blocked and deglycosylated. Cell suspension is added to each well allowing for the antibody to bind to the cell surface receptor, immobilizing the cells B) Captured cells are fixed with 10% neutral buffered formalin and delipidated with Carnoys solution C) Terminal sialic acid residues are either stabilized and derivatized or enzymatically removed. D) PNGase F is sprayed alone or in combination with sialidase to release the cell surface N-glycans, CHCA matrix is applied and 5 mM ammonium phosphate is applied to remove matrix clustering. E) Slides are imaged by a MALDI QTOF timsTOF-fleX ( Bruker), obtaining m/z values associated to each spot that correspond to individual N-glycans. Created with BioRender.com

Figure 1A highlights the general mechanism of the assay utilizing an antibody array that captures cells through cell receptor/antibody immobilization. Our previous antibody array methods utilized nitrocellulose slides, however, these were associated with high background and loss of specific signal (data not shown). This required the utilization of another slide chemistry. Amine reactive slides were found to have the required binding and affinity for this method. Another required step was the fixing and delipidation of cells (Figure 1 B), which was found to be essential for cells retention via antibody capture.

The analysis of sialic acid is always a challenge of MALDI based methods due to potential lability from laser and we incorporate a method of sialic acid derivatization and stabilization to allow for their analysis (Figure 1C). Briefly, terminal sialic acids can either be attached via an α2,3 or an α2,6 linkage, which have been shown to have differential biological implications^20,21. Being isomers, these two linkages cannot be differentiated by mass alone. Sialic acid stabilization and derivatization was performed via a slide based amidation-amidation reaction (AAXL) directly onto the captured cells. AAXL induces a mass shift of 27 daltons for α2,6 linked sialic acids and 37 daltons for α2,3 linked sialic acids, providing clear resolution between the two isomers.²² Following traditional imaging workflows, PNGase F and CHCA matrix are sprayed onto slides using an automated sprayer followed by spraying of an ammonium phosphate buffer to reduce matrix cluster formation and increase assay sensitivity (Figure 1D). Imaging of glycans was performed using a Bruker tims TOF fleX MALDI mass spectrometer (Figure 1E) operated in QTOF mode.

In our initial analysis, we developed the method to capture and analyze the N-linked glycans on an immortalized line of human T lymphocyte cells (Jurkat Cells). The cell surface maker CD7, which is found on mature T-cells, was used as the capture antigen. Figure 2 shows glycan imaging data of a mouse monoclonal antibody to CD7 spotted onto an amine reactive slide which was either left untreated or incubated with Jurkat cells. Certain glycan are observed on both the antibody used for capture and on the antibody captured cells as well as unique glycans detected only on captured cells (Figure 2). The complete list of N-linked glycans found on Jurkat cells and anti-CD7 is provided as supplementary table S1 and S2. Overlapping glycans that can be found on antibodies as well as captured cells can muddle analyses when trying to discern what proportion of signal is originating from cells vs antibody. Our next step in development focused on removing antibody glycan while retaining specificity and signal intensity.

Figure 2.

Overlapping antibody and Jurkat cell N-glycans. N-glycan images of spotted anti-CD7 antibody alone and spotted anti-CD7 antibody with captured Jurkat cells. Depicting 3 overlapping N-glycans and 3 cell specific N-glycans. A) Overlapping 1485.535 m/z N-glycan. B) Overlapping 1647.592 m/z N-glycan. C) Overlapping 1809.643 m/z N-glycan. D) Cell specific 1905.643 m/z N-glycan. E) Cell specific 2127.792 m/z N-glycan. F) Cell specific 2858.064 m/z N-glycan. G) Spotted glycosylated antibody and H&E image of captured cells with glycosylated antibody.

The N-linked glycans on the Fc region of immunoglobulin can and often do vary ²³, so the inherent glycosylation of the capture antibody may have a negative impact on the ability to comprehensively analyze the glycans on captured cells. That is, as Figure 3A shows, antibodies are glycoproteins and contain two conserved sites of glycosylation at N297 on their Fc region²³. To reduce the background of the antibody N-glycan signal, spotted antibodies are deglycosylated with PNGase F prior to cell capture. Briefly, 200 ng of antibody specific to a cell surface protein is spotted and slides are subsequently blocked with 100mM ammonium bicarbonate. Post blocking, 1 μg of PNGase F diluted in 100 μl of HPLC water is added to each well chamber following static incubation for 2 hr at 37 °C. Figure 3B shows an imaging result highlighting the most abundant antibody N-glycan at 1647.592 m/z from spots treated with and without PNGase F. As Figure 3C shows, there is a >10 fold reduction in the signal intensity of the top 3 most abundant N-glycans found on the spotted anti-CD7 antibody when comparing glycosylated and deglycosylated spots.

Figure 3.

Antibody deglycosylation prior to cell capture. A) Spotted antibodies are deglycosyated by adding 1 ug of PNGase F diluted in 100 μl of HPLC water into each well and statically incubated for 2 h at 37 °C B) N-glycan image comparison of the most abundant N-glycan 1647.592 m/z on glycostylated vs deglycosylated spotted anti-CD7 antibody. C) Total area under the peak quantification of the 3 most abundant N-glycan species from triplicate glycosylated and deglycosyltaed spotted anti-CD7 antibodies D) N-glycan image comparison of the most abundant N-glycan on Jurkat cells, 1905.643 m/z from cells captured with glycosylated and deglycosylated antibodies E) Total glycan intensity of Jurkat cells captured with glycosylated and deglycosylated antibodies normalized to cell count in triplicate. Statistical test includes a One-Way ANOVA (NS= P=0.7122) Created with BioRender.com

Having shown that we can remove the glycans on the capture antibody, we proceeded to determine if we could capture cells and analyze their N-linked glycosylation. One key consideration was to ensure that antibody function was not altered following deglycosylation. To test this, we compared total signal intensity of cells captured with deglycosylated and glycosylated antibodies. As Figures 3C&D show, antibodies that have been left untreated or treated can capture cells and there is a sufficient amount of cells captured by glycosylated and deglycosylated antibodies, producing adequate signal. Figure 3E shows no significant difference in total glycan signal when normalized to cell count between cells captured with glycosylated and deglycosylated antibodies. This indicates that wash steps after PNGase F digestion were sufficient in removing the enzyme, reducing the potential for deglycosylation and loss of signal during capture steps.

Demonstrating that we could capture cells, we next wanted to determine the ability to analyze cell specific N-glycosylation without the need to negatively or positively isolate cells prior to analyses. In figure 4A and 4B we showed that cell capture is dependent on the spotted antibodies specificity to cell surface markers expressed on the targeted cell. Antibodies specific to CD7 and CD8 as well as PBS were spotted in triplicate on the same slide. The slide was blocked, antibodies were deglycosylated and Jurkat cells were incubated over each spot. As shown in Figure 4, wells spotted with antibodies specific to CD7 captured the Jurkat cells while wells with CD8 and PBS only spots showed no capture and no subsequent N-glycan signal.

Figure 4.

Capture specificity and sensitivity. A) 200 ng of Anti-CD7, Anti-CD8 and PBS were spotted onto amine reactive slides. 5 x 10⁵ Jurkat cells were incubated over each spot resulting in cell capture occurring over the spotted anti-CD7 antibody only. Images include H&E stained captured Jurkat cells and its corresponding N-glycan image of 1905.633 m/z. No capture or N-glycan signal is observed on spotted PBS or anti-CD8 antibody. B) Mass spectra of 1905.633 m/z the most abundant Jurkat N-glycan. N-glycan signal observed on cells captured by anti-cd7 antibody. C) . Jurkat cells incubated in a dilution series of 5 x 10⁵, 3 x 10⁵, 2 x 10⁵, 1 x 10⁵ and 5 x 10⁴ cells over 200 ng of spotted anti-CD7 antibody. H&E and N-glycan image of captured Jurkat cells. D) Number of cells captured per spot plotted against the total N-glycan signal of the corresponding spot.

Assay sensitivity is dependent on antibody affinity as well as the expression level of the targeted receptor. The number of cells captured will influence the number of detectable N-glycans. In the context of this experiment, we targeted CD7 which is a pan T cell surface marker that is abundantly expressed on Jurkat cells²⁴. In Figure 4C, a dilution series of Jurkat cells ranging from 5 x 10⁵ to 5 x 10⁴ cells was incubated in each well containing a spotted anti-CD7 antibody in triplicate. Figure 4C displays the H&E image, as well as the N-glycan image of 1905.630 m/z, the most abundant N-glycan species, man 9. Lowering the amount of cells plated per well reduces the number of cells captured per spot which corresponds to a reduction in N-glycan signal. In Figure 4D, the number of cells captured per spot was plotted against the total N-glycan signal of the corresponding spot and a roughly linear trendline occurs, resulting in an R² value of 0.9352. Linearity of the N-glycan signal to number of cells captured is dependent on raster and resulting field range of laser shots. Within the Bruker timsControl (client version 4.0.4) we used a M5 small smart beam setting with a scan range of 100 μm which corresponded with a resulting field range of 139 μm. When run at a raster of 150 μm the space between laser shots is roughly 11 μm, resulting in nearly full coverage of the targeted area. Ablating the entire capture spot is key for linearity, allowing for the contribution of all captured cells to the spots N-glycan signal. High intra-slide linearity allows for cell count normalization of total N-glycan signal, permitting quantification of total glycosylation to be calculated and compared across experimental groups run on the same slide or within batched slides.

For any analytical method, variability is always an issue. Thus, we wanted to determine the day-to-day variability of the method; to that end cell capture followed by AAXL and N-glycan imaging was performed on Jurkat cells. Each prep was staggered by one day (Day 1-3) over the course of three total days of preparation. Jurkat cells were plated at 5 x 10⁶ cells/ml and collected from the same culture flask which was passaged after each day, maintaining similar confluency. Figure 5A shows the captured and H&E stained Jurkat cells from all 3 days. Intraday capture variability was low and ranged from 5%-14% coefficient of variation (CV) while day to day variability was higher at 26% CV. The average number of cells captured per spot across all days was 16,586. Figure 5B displays the relative intensity values of the 20 most abundant N-glycans observed on Jurkat cells. Variability was low across the top 20 species with an average CV of 13%. In Figure 5D, N-glycans were grouped based off structural features and the variability reduced to an average of 5% CV across all subclasses.

Figure 5.

Reproducibility. Jurkat cells were captured in triplicate over 3 separate days. Terminal sialic acids were stabilized and derivatized followed by N-glycan imaging. A) H&E image of captured Jurkat cells from each day B) N-glycan images of 1905.643 m/z, Man 9 C) Graph displays the relative intensity values of the top 20 most abundant N-glycans from each day of capture. D) Graph displays the relative intensity values of grouped N-glycans based of structural features.

As we had successfully shown the ability to capture specific cells and perform quantitative N-linked glycan analysis, we wanted to see if we could use the method to detect changes in N-linked glycosylation. To that end, we treated Jurkat cells with 2F-Peracetyl-Fucose (2FF), an inhibitor of fucosylation²⁵. Jurkat cells were treated with either drug vehicle (water), or 50, 100 or 200 μm of 2FF for 3 days. Cell viability was relatively unaffected by treatment and remained between 80% and 95% at the time of collection (Figure S1). Sialidase in combination with PNGase F was used to image the N-glycans.²⁶ Sialidase cleaves the terminal sialic acid residues, collapsing the differentially sialylated species into one peak, increasing unsialylated N-glycan peak intensities. This imaging technique increases the assays sensitivity to detect changes in fucosylation. Figure 6A-B shows representative images of 2320.840 m/z which is a triantennary doubly fucosylated N-glycan that exhibits a dose response reduction in signal intensity as 2FF is increased. The adjacent image depicts 2028.719 m/z, the unfucosylated N-glycan species of 2320.841 m/z which exhibits a dose response increase in signal intensity as 2FF is increased. In Figure 6C, percent relative intensity trends of grouped N-glycans based on structural features define significant changes within the surface N-glycosylation profiles of the Jurkat cells in response to 2FF treatment. Total fucosylation exhibited a significant decrease as 2FF concentration increased. This decrease in fucosylaton was further accentuated in multi-fucosylated species and inversely non-fucosylated N-glycan species significantly increase in relative intensity as 2FF was increased, further highlighting 2FF’s inhibition of fucosylation. High mannose N-glycan species were unaffected by 2FF, exhibiting relative intensity values unchanged across all treatment groups, highlighting 2FF’s specificity towards fucose inhibition and the assays ability to detect changes in expression levels of one monosaccharide.

Figure 6.

Jurkat N-glycan modulation. Jurkat cells were treated with water, 50 μm, 100 μm and 200 μm of 2FF for 3 days in biological triplicate and captured in technical triplicate for N-glycan analysis. Sialidase and PNGase F were sprayed to increase the assays sensitivity to changes in fucosylation. A) N-glycan image of 2320.841 m/z, a doubly fucosylated triantemiary N-glycan B) N-glycan image of 2028.719 m/z, a unfuscylated triantennary N-glycan C) Graph of relative intensity values of total fucosylated, multi-fucosylated, unfucosylated and high mannose N-glycans across all treatment groups. Statistical test includes a One-Way ANOVA (*= P< 0.01, **= P< 0.001, ***= P< 0.0001, ****= P<0.00001, NS= P>0.1858)

To further challenge the assay, we wanted to capture specific cell types from a mixed population. We isolated splenocytes from C57BL/6 mice which consist of an assortment of different cell types that generally include B cells, T cells, macrophages, NK cells and dendritic cells. In this experiment we wanted to specifically capture T helper (CD4) and cytotoxic T cells (CD8) using spotted antibodies specific to CD4 and CD8. Cell suspension was added to wells containing either CD4 or CD8 antibodies which subsequently captured its corresponding cell type. Cell capture validation was conducted by performing immunofluorescent staining on captured cells to confirm specific capture. In Figure 7A, captured cells were stained for with their corresponding cell surface marker for the positive control or the opposing cell surface marker in the negative control. The positive control resulted in cell surface staining while the negative control resulted in no staining, confirming specific capture.

Figure 7.

Murine CD4 and CD8 T cell capture specificity and N-glycan analyses A). Splenocytes were initially isolated from C57BL/6 mice in biological triplicate and captured in technical triplicate. Red blood cells were lysed and the remaining splenocytes were washed and resuspended in FACs buffer. Anti-CD4 and anti-CD8 antibodies were used in the cell capture workflow to capture T helper (CD4) and cytotoxic T cells (CD8) from the mixed population of cells. A) Capture specificity was confirmed via immunofluorescent staining using biotinylated CD4 and CD8 primary antibodies and cy5 conjugated streptavidin. B) H&E and N-glycan image of captured CD4 and CD8 T cells C) Graph of relative intensity values of N-glycans found on both CD4 and CD8 T cells D) Graphs comparing relative intensity values of High mannose, α2, 6 Neu5Gc and fucosylated N-glycans between CD4 and CD8 T cells. Statistical test includes a One-Way ANOVA (*= P< 0.01, **= P< 0.001, ***= P< 0.0001, ****= P<0.00001).

Splenocytes from three male C57BL/6 mice were isolated and used for capture. Both CD4 and CD8 antibody spots averaged 9,532 cells per spot and a total of 17 N-glycans were detected across both cell types, as seen in Figure 7C. There was no N-glycan species specific to either CD4 or CD8 T cells, instead there were differences in the relative intensity when grouping N-glycans into subclasses based off structural features. In Figure 7D, CD4 T cells exhibited a higher relative intensity of high mannose N-glycans while CD8 T cells had a higher relative intensity of α2,6 linked sialic acid residues as well as increased levels of fucosylation.

CONCLUSION

Described in this workflow is a new approach that utilizes an antibody array based MALDI-IMS platform to detect and analyze N-glycans from specific immune cell populations. The utilization of antibody arrays in combination with MALDI-IMS mitigates the need not only for positive or negative isolation of specific cell types but also the need for sample clean up prior to N-glycan MS detection and is amenable to automation. Typical immune cell based N-glycan analysis requires lengthy isolation and purification steps, hindering larger scale studies. This approach streamlines workflows allowing for direct N-glycan detection on specific immune cell populations from complex sample types such as splenocytes or PBMCs. Captured cells can be counted, and therefore released glycan signal intensities can be reported on a per cell basis. Lectin based flow cytometry is commonly used to profile N-glycosylation of specific immune cell types in mixed sample populations. However, lectins only bind to specific glycan domains, providing only partial structural data. With the utilization of an on slide sialic acid derivatization and PNGase F release, we can elucidate several hundred different individual N-glycans from captured cells. Further method optimization of this assay can increase its clinical applicability. Multiple cell types such as B cells, macrophages, monocytes and NK cells can be targeted for capture using well validated antibodies commonly used for immune cell characterization in flow cytometry. This will allow full N-glycan profiling of a patient’s immune system, enabling in depth biomarker analysis as well as characterizing immune response in accordance to therapies, vaccinations or other treatments. Besides circulating immune cells, targets such as circulating cancer cells and tumor infiltrating immune cells are targets of interest that will require further method optimization to analyze.

Supplementary Material

supplementary material

Table S1: All observed N-glycans on Jurkat cells (PDF)

Table S2: All observed N-glycans on anti-CD7 monoclonal antibody (PDF)

Table S3: All observed N-glycans on murine CD4 and CD8 T cells (PDF)

Figure S1: 2FF treated Jurkat viability (PDF)