Automated Enzymatic Deglycosylation of Monoclonal Antibody with the SampleStream Platform

Abstract

Engineered monoclonal antibodies have proven themselves as invaluable biotherapeutics used in clinics worldwide. In discovery, development, and production, it is essential that they are accurately validated to ensure their homogeneity, consistency, safety, and effectiveness, which are commonly conducted via mass spectrometric analysis. However, validation processes can be manual, time-consuming, and costly. One hindrance in the analytical workflow is the relatively long time required to deglycosylate native mAbs with PNGase F on the benchtop, which is usually done to reduce the spectral complexity, increase the ionization efficiency, and facilitate the identification of glycosylation sites. To circumvent this obstacle, a workflow on the SampleStream Platform was developed and optimized to automate in-channel PNGase F-mediated deglycosylation of native monoclonal antibody at unprecedented speed, accomplishing 86% deglycosylation in 3 min. The presented workflow offers a promising strategy to reduce discovery and development costs and streamline the characterization of antibody-based biopharmaceuticals.

Graphical Abstract

INTRODUCTION

Engineered monoclonal antibodies (mAbs) are a rapidly growing class of biologics with a demonstrated therapeutic potential. Since 2021, the United States Food and Drug Administration has approved more than 100 mAbs for clinical use, and more than 160 have been approved by regulatory agencies worldwide.¹ To produce mAbs, they may be expressed recombinantly via Human Embryonic Kidney 293 or Chinese Hamster Ovary cells among others as an expression system.² Additionally, mouse hybridoma, phage display, or harvested single B cells may also be utilized.³ It is crucial that therapeutic mAbs be assessed on a batch-to-batch basis to ensure their homogeneity, consistency, safety, and effectiveness. Producing mAbs via these techniques carries inherent risks that can alter the identity of the mAb; therefore, fast and global analysis is essential to minimize risks and streamline characterization, as the market for therapeutic mAbs continues to expand.

The enzyme PNGase F is commonly used to cleave N-glycans which can reduce spectral complexity, improve sequence coverage of tandem-mass spectrometry (MS) analysis to facilitate identification of glycosylation sites, and increase ionization efficiency.⁴ It is preferred because it cleaves all types of N-glycans aside from those containing α−1–3-fucosylated biochinose cores (common in plants).^5–7 However, the length of incubation required to completely deglycosylate glycoproteins under native conditions typically ranges from 2 h to days.^8–10 Faster deglycosylation is critical to improving analytical throughput.

Recently, some progress has been made to speed up deglycosylation. An alternative enzyme, Endo-β-N-acetylglucosaminidase 1, as well as pressure cycling and microwave reactor technologies, modestly increased activity on native glycoproteins but still required at least 30 min for complete deglycosylation.^11–13 New England Biolabs’ Rapid PNGase F completely deglycosylates denatured, but not native, glycoproteins within 10 min.¹⁴ Analogous to filter-aided sample preparation used to digest proteins above a molecular weight cutoff (MWCO) membrane,¹⁵ we hypothesized that the Sample Stream Platform (SSP)¹⁶ could perform deglycosylation of native mAb within its channel. Herein is a workflow for the fast, automated in-channel PNGase F-mediated deglycosylation of native mAb with the SSP coupled to native intact heated electrospray ionization (HESI) MS analysis.

METHODS

Sample Preparation.

Humanized IgG1κ mAb (RM8671) was purchased from the National Institute of Standards and Technology (NIST). PNGase F (V483A) was purchased from Promega. Ammonium acetate (A11450), water (W64), and 6 N hydrochloric acid (SA56) for pH adjustment were purchased from Fisher Scientific. Urea (U0631) and N-lauroylsarcosine (L9150) were purchased from Sigma-Aldrich. NIST mAb and PNGase F were diluted in native solution (20 mM ammonium acetate, ~pH 6.8) to 100 ng/μL (w/v) and 0.1 U/μL, respectively. For the denatured sample, NIST mAb was diluted in a native solution with 6 M urea. For the enhanced enzyme sample, PNGase F was diluted in native solution with 0.5% N-lauroylsarcosine (NLS). All samples were prepared in glass autosampler vials (MicroSolv, 9512S-1MP-RS) and placed on the SSP at room temperature.

SampleStream Platform (SSP) and Native Intact Mass Spectrometry.

The SSP was operated with native solution, 5 kDa MWCO membrane, 50 μL sample push at 175 μL/min, various focus parameters, and 200 μL elution volume at 100 μL/min elution flow rate. A 10 μL volume of water was injected by a CTC Pal 3 robot for blanks with a 125 μL focus volume at 175 μL/min focus flow rate.¹⁶ Thermo Fisher Scientific Q Exactive ultrahigh mass range (UHMR) MS HESI source parameters were 3.1 kV spray voltage, 20 au sheath gas, 5 au auxiliary gas, 320 °C inlet capillary temperature, and 200 au RF level. Scans were acquired for 2 min with ultrahigh mass range mode ON, 30.0 eV in-source CID, 2.0 trapping gas pressure, 1e6 AGC target, 200 ms maximum injection time, 1 μscan, 2,000 to 10,000 m/z scan range, and 17,500 resolution (at m/z 200).

Data Analysis.

PNGase F and glycosylated, half-deglycosylated, and completely deglycosylated NIST mAb proteoforms were quantified with ProSight Native software (Proteinaceous, Inc.).¹⁷ Spectra were charge-state deconvoluted by kDecon in the Batch Deconvolution workflow with overlapping sliding windows, and the detected mass components were searched against theoretical masses-of-interest (Supplementary Table 1) with a wide 200 ppm mass tolerance to ensure annotation of all present proteoforms. For each annotated proteoform, selected ion chromatogram area under the curve (AUC) was calculated from the top 5 charge states with a 1.0 m/z tolerance. Statistics and data visualization were performed in R 4.4.2 and Adobe Illustrator 2024. Percent deglycosylation was calculated with eq 1:

$$[\text{Degly}\text{AUC}+(\text{Half}-\text{degly}\text{AUC}/2\left)\right]/\text{Total}\text{AUC}\times 100$$ (1)

RESULTS AND DISCUSSION

The workflow for automated enzymatic deglycosylation of native mAb with the SSP coupled to a Q Exactive UHMR is outlined in Figure 1. The SSP has been described in detail previously;¹⁶ briefly, the robot sequentially aspirates PNGase F and native mAb, injects them into a circular channel, focuses them to the center via liquid pumps while simultaneously performing buffer exchange and deglycosylation, and then directly elutes the species to the MS for analysis.

Figure 1.

Workflow schematic for automated enzymatic deglycosylation of native mAb coupled to native intact HESI-MS.

To optimize native in-channel deglycosylation with the SSP, variables, including temperature, PNGase F amount, focus flow rate, and additives, were investigated. Across all experiments, 300 ng of NIST mAb was injected for each deglycosylation reaction, whereas other variables were limited to remaining sensitive to changes in PNGase F activity. Additionally, the SSP was sufficiently blanked after each injection to eliminate carry over activity. Less than 0.2% carry over of PNGase F was achieved with 5, 3, 3, and 2 blanks at 37, 45, 50, and 55 °C, respectively; 2 blanks were also used at 60 and 68 °C (Supplementary Figure 1).

Temperature was investigated first because it significantly affects enzymatic activity.¹⁸ Percent deglycosylation was calculated in quintet at each temperature, as displayed in Figure 2A. PNGase F, focus flow rate, and sample focusing time in the SSP channel were controlled at 0.3 U, 175 μL/min, and 2.86 min, respectively. The optimal temperature range for PNGase F activity fell between 45 to 55 °C; 50 °C was used in further experimentation to balance PNGase F activity without observing its extensive loss, which may be attributed to thermal denaturation shifting its charge states below the lower end of the measured scan range.

Figure 2.

Effect of (A) temperature, (B) PNGase F concentration, and (C) focus flow rate on percent deglycosylation and PNGase F AUC (n = 5 for panel A and n = 3 for all other panels). (D) Effect of 6 M urea and 0.5% NLS on percent deglycosylation and PNGase F AUC. Error bars indicate standard deviation.

Next, the effect of PNGase F concentration on percent deglycosylation was examined with focus flow rate and focus time limited to 175 μL/min and 2.86 min, respectively. Percent deglycosylation was measured in triplicate with PNGase F injections of 0.06, 0.18, 0.3, 0.45, 0.6, and 0.9 U which corresponds to 0.2, 0.6, 1, 1.5, 2, and 3-fold the minimum recommended 1:1 PNGase F (U) to substrate (μg) ratio according to Promega’s protocol.⁹ The extent of deglycosylation plateaued at 0.9 U of PNGase F, so it was selected as most optimal (Figure 2B). Under native conditions, the need for excess enzyme was expected; still, assuming the minimum reaction volume on the benchtop is 10 μL, an approximately 10-fold less absolute amount of PNGase F is needed per reaction with the SSP at a 3:1 ratio. Under denaturing conditions at a 1:20 ratio, 0.5 U PNGase F would be used, which is in parity with the SSP at a 2:1 ratio. Lastly, leftover deglycosylated material is not produced or wasted, which is especially valuable in situations where source material is limited.

The opposing fluidic flow in SampleStream increases the local analyte concentration in the channel, thus potentially increasing the rate of reaction. Flow rates for the sample focusing step were varied to 20, 70, 120, 170, 220, and 270 μL/min for injections in triplicate as shown in Figure 2C, with PNGase F amount and focus time limited to 0.3 U and 3 min, respectively. Interestingly, the PNGase F AUC modestly increased with focus flow rate, which might be explained by faster flow rates replacing hot solution in the channel more quickly with room temperature solution, promoting retention of PNGase F in its native state. Percent deglycosylation plateaued at 270 μL/min, so it was selected as most optimal.

Then, the effect of additives on percent deglycosylation was assessed, aiming to increase the rate of reaction via reversible denaturation of mAb with 6 M urea or enhanced activity of PNGase F with 0.5% NLS.^19,20 Kinetic curves ranging from 1 to 5 min focus times were compared to a control curve (Figure 2D) with the PNGase F amount and focus flow rate at 0.3 U and 270 μL/min, respectively. Minimal increase in percent deglycosylation with additives compared to the control curve was observed. Urea apparently increased PNGase F AUC at shorter focus times, likely due to denatured mAb signal overlapping with native PNGase F (Supplementary Figure 2) until urea is sufficiently washed out of the channel by 3 min focus time. Denatured mAb signal may also overlap with native deglycosylated mAb, implicating an overestimation of percent deglycosylation at 1 and 2 min focus times.

Lastly, a kinetic curve as illustrated in Figure 3A demonstrates that the SSP can achieve 86% deglycosylation in 3 min with the selected conditions. PNGase F mean AUC increases and then decreases over time which may be attributed to suppressed ionization via insufficient removal of salts and detergents and thermal denaturation. A representative spectrum of deglycosylated NIST mAb after 3 min is shown in Figure 3B and its deconvolved spectrum with ProSight Native in Supplementary Figure 3. However, complete deglycosylation is often essential for regulatory or analytical purposes; further optimizations or additional strategies are necessary to implement this workflow in settings where complete deglycosylation is required.

Figure 3.

(A) Kinetic curve ranging from 1 to 5 min focus time with 50 °C channel temperature, 0.9 U PNGase F, and 270 μL/min focus flow rate (n = 3). (B) Spectrum of deglycosylated NIST mAb obtained with the selected conditions using a sample focusing time in the channel of 3 min.

CONCLUSIONS

The SSP can expedite and automate enzymatic deglycosylation of N-glycans from native mAb, offering an advantageous alternative to lengthy, manual, and often wasteful deglycosylation on the benchtop. This workflow can be optimized for other N-linked glycoproteins but may be limited by samples with inaccessible N-glycans shielded by tertiary structure. The concept of native in-channel digestion can be adapted to other digestion types such as reduction, conjugation, or proteolytic digestions and combined with tandem-MS for even deeper analyses in industrial, academic, or regulatory settings.

Supplementary Material

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jasms.5c00242.

Theoretical masses table for the proteoform search space, carry over analysis bar charts, spectrum of PNGase F plus denatured NIST mAb, and a deconvolved spectrum of deglycosylated native NIST mAb (PDF)

ABBREVIATIONS

mAb

monoclonal antibody

MS

mass spectrometry

MWCO

molecular weight cutoff

SSP

SampleStream Platform

HESI

heated electrospray ionization

NIST

National Institute of Standards and Technology

NLS

N-lauroylsarcosine

UHMR

ultrahigh mass range

AUC

area under the curve

U

units