Isolation and quantification of N-glycans from immunoglobulin G antibodies for quantitative glycosylation analysis

N-glycosylation is one of the critical quality attributes for the therapeutic monoclonal antibodies. Characterization of N-glycans of monoclonal antibodies provides valuable information about its therapeutic efficacy. We present a non-invasive method of isolating N-glycans from Immunoglobulin G based antibodies for glycosylation analysis. The method consists of purification of antibodies from biological solution and release of N-glycans with peptide-N-glycosidase F in a single consolidated process using a mini affinity ligand column (e.g. protein-A column). The method is highly reproducible with average coefficient of variation of 0.012 in the glycoform percentage distributions between the replicates. The method provides quantification of the molar yield of glycans as a function of molar concentration of antibody in a single analysis. To our knowledge, this is the first time this approach was used to detect and quantify any macro-heterogeneity of N-glycosylation in monoclonal antibody samples. This fairly rapid and very cost-efficient method would be of great interest for academic labs and biopharmaceutical industries.


BACKGROUND
Monoclonal antibodies (mAbs) represent the major fraction of approved bio-therapeutics for the treatment of cancer, autoimmune disorders and inflammatory diseases [1].N-linked glycosylation of immunoglobulin G (IgG) antibodies is an important biomarker that defines the therapeutic efficacy of mAbs or immune state of the human body.The structure and the terminal sugar type of the N-glycans play a critical role in modulating the pharmacokinetic properties and effector functions of antibodies.Appropriate glycosylation at a conserved site in the C H 2 domain of Fc region is essential for effector functions, including antibody-dependent cell-mediated cytotoxicity (ADCC) and complement-dependent cytotoxicity (CDC).For example, the absence of core fucose was found to enhance ADCC activity of mAb dramatically [2].The presence of galactose as terminal sugar in N-glycans of IgGs enhances their CDC activity [3,4], binding affinity to FcγRIIIa receptors [5] and ADCC activity [6].On the other hand, antibodies with terminal sialic acid in their N-glycans have reduced ADCC activity [7] and become anti-inflammatory [8].In addition, mAbs with the presence of a particular sugar type or linkage (e.g.N-glycolylneuraminic acid, galactose-α(1,3)-galactose epitope) in their glycans can trigger immunogenic responses in humans [9,10].Deglycosylated [11] or hemi-glycosylated [12] mAbs exert no effector functions, less thermally stable and prone to aggregation.
N-glycan distribution (micro-heterogeneity) in the mAbs is influenced by various bioprocess parameters such as the choice of production cell line, operating culture conditions and availability of nutrients [13].Either the depletion of glucose [14] or the replacement of glucose with alternate carbon sources [15] could influence the site occupancy (macro-heterogeneity) of glycosylation in mAb.Characterization of both micro-and macro-heterogeneity of N-glycans of mAbs is very important to understand their therapeutic efficacy and to optimize bioprocess parameters to produce effective mAbs.Glycosylation characterization of antibodies can be achieved either by mass spectrometry analysis of peptides bearing N-glycans or by the analysis of released glycans [16].The later method requires isolating N-glycans from glycoproteins using PNGase F digestion [17], fluorescent labelling of glycans and analysis by high performance liquid chromatography (HPLC) or capillary electrophoresis (CE).Hydrophilic interaction liquid chromatography (HILIC) based HPLC separation of labeled glycans is the most prominent method of analysing free glycans [18].Some advantages include separation of glycans with the same mass but with different structures (e.g.isomers with galactose on either [3] or [6] mannose arm of bi-antennary structure, α(2-3) or α(2-6) linkage of sialic acid).The development of ultra-performance liquid chromatography (UPLC) systems and efficient columns made this method very reliable in separation of glycans with high resolution.
The traditional way of isolating N-glycans from mAb is by PNGase

POL Scientific
F digestion of either the heavy chain bands following SDS-PAGE gel separation of reduced mAb samples [19] or by direct in-solution reduction and digestion of purified mAb.The immobilization of either the enzyme or the mAb on to a solid support provides the advantage of increased efficiency of digestion and easy handling of reagents.Based on this approach, high throughput robotic platform methods have been developed using PNGase F immobilized micro fluidic chips [20,21] or immobilizing mAb on to multi-well plates [22,23].However, the affordability of high-throughput robotic platform based analysis of N-glycans in academic laboratories and small scale biopharmaceutical industries is limited.In addition, quantifying the glycan concentration in combination with the mAb concentration for the analysis of both micro-and macro-heterogeneity of glycosylation has not been addressed previously.
We present a relatively rapid, highly reproducible and cost-efficient method of isolating N-glycans of mAbs directly from cell culture supernatant using readily available mini affinity chromatography (e.g.Protein-A) columns (Fig. 1).The method involves purification of mAbs and PNGase F digestion of their N-glycans in a single consolidated process.The optimization and efficiency of the method was demonstrated by 2-aminobenzamide (2-AB) labelling and HILIC-based HPLC analysis of released N-glycans.This high yield and non-invasive method provides complete recovery of N-glycans thus allowing quantification of both antibody and glycan concentrations in a single analysis.We have devised a strategy to detect and quantify any macro-heterogeneity present in mAb product and demonstrated using the mAb samples produced under glucose-starvation conditions.

Phosphate buffer
To make 500 ml phosphate buffer, add 1.14 g Na 2 HPO 4 into 400 ml of ultrapure H 2 O. Add 0.24 g NaH 2 PO 4 into 100 ml of ultrapure H 2 O. Mix both solutions.pH should be 7.4 ± 0.05 without adjustment.
CAUTION: NaBH 3 CN should be handled in a fume hood.Less toxic 2-picoline borane with same molar concentration can be used as alternative to NaBH 3 CN.

Elution buffer (100 mM glycine-HCl)
To make 100 ml Elution buffer, add 0.75 g glycine into 90 ml of ultrapure H 2 O. Adjust pH to 2.7 with 1 N HCl and final volume to 100 ml with ultrapure water.

Neutralizing buffer (1 M TRIS-HCl)
To make 100 ml (1 M) neutralizing buffer, add 12.12 g TRIS into 90 ml of ultrapure H 2 O. Adjust pH to 9.0 with 1 N HCl and final volume to 100 ml with ultrapure water.

Buffer A (50 mM ammonium formate solution)
To make 2 L buffer A, add 3.9 ml formic acid into 1.9 L ultrapure H 1.2.Wash the column by adding 600 µl of phosphate buffer and centrifuge to remove the buffer.
1.3.Plug the bottom of the column and add antibody solution.
1.4.Incubate at room temperature for 15 min on a rocker or orbital shaker.
1.5.Remove the plug and centrifuge to remove the solution.
1.9.Incubate the columns at 37°C on a rocker or rotary shaker for 24 h.

HINTS: (i)
Columns can be filled with only 600 µl of solution at a time.Antibody solution can be concentrated with 30K filters to avoid multiple loading steps to reach desired antibody concentration.(ii) 100 pM antibody is sufficient for decent detection of glycans on HPLC.mAb concentration should be at least 100 µg/ml for accurate quantification of eluted antibody using Nanodrop TM spectrophotometer.For the cases of quantifying antibody, capture at least 50 µg and elute with total of 400 µl elution buffer.(iii) More PNGase F can be added to reduce the incubation time.One unit of Promega's PNGase F will catalyze the deglycosylation of 1 nmol of denatured Ribonuclease B in one min at 37°°C.When using PNGase F from other sources, amount should match these units.

Glycans collection
2.1.Add 1 or 2 nmol of DP4 standard to the protein-A column containing released glycans and mix gently (optional).

2.2.
Remove the plug and place the column immediately in fresh 2 ml tube.

2.3.
Centrifuge to collect the filtrate containing glycans.

2.4.
Plug the bottom of the column and add 250 µl of ultrapure water, place the cap and rinse the column by gently inverting the column with hand.

2.5.
Remove the plug and collect the filtrate into the same tube as in step 2.3.

2.6.
Plug the bottom of the column and add 600 µl of phosphate buffer and keep it aside for eluting antibody as described in section 3.

2.7.
Filter the glycans through 10K filter and collect the filtrate into 1.5 ml micro-centrifuge tube.
2.8.Dry the glycans in a vacuum concentrator.
HINT: Wash 10K filter at least once with 500 µl ultrapure water.

POL Scientific
3. Elution of antibody (optional) 3.1.Take the column from step 2.6 and wash it three times with phosphate buffer.

3.2.
Plug the bottom of column and add 300 µl of elution buffer.

3.3.
Gently mix by shaking with hand.

3.4.
Collect the filtrate into a 2 ml tube.

3.5.
Repeat steps 19 and 20 with 100 µl of elution buffer and collect the filtrate in the same tube as in step 3.4.

3.6.
Neutralize the eluted antibody with neutralizing buffer to pH ~7.2.

3.7.
Store it to quantify later by absorbance at 280 nm using UV-Vis spectrophotometer.
HINT: Eluted antibody can be stored at 4°C for short-term and -20°C for long term.
4.9.Dry the labeled glycans in a vacuum concentrator.
HINTS: (i) Dried glycans can be stored at -20°C.(ii) Glycan clean-up can be performed by placing HyperSep TM Diol cartridges on Eppendorf rack and applying positive pressure of air so that liquid flow rate is ≤ 2 ml/min.Glycans during elution can be collected in sample tubes placed on another Eppendorf rack stacked underneath.Alternatively, Vacuum Manifold designed for cartridges could be purchased.

NOTE:
The 2-AB labeled glycans can be analyzed using HPLC or UPLC system.The guidelines provided below are for the Water HPLC system equipped with 1525 µ Binary Pump, 2707 auto-sampler and 2475 fluorescent detector.The column heater housing Waters X-Bridge BEH amide column (3.5 µm, 4.6 × 250 mm, Part # 186004870) was set at 30°C.The data acquisition and analysis was done using Empower 2.0 software.The flow rate and gradient for mobile phase A (Buffer A) and mobile phase B (Acetonitrile) are presented in Table 1.The settings for fluorescent detector are presented in Table 2.

Analysis of glycans by HILIC-HPLC
5.1.Add 20 µl of ultrapure water to each glycan sample from step 4.9 and mix well.

5.2.
Prepare samples with HPLC vials by mixing 6 µl of sample with 24 µl of acetonitrile in HPLC auto-sampler tubes.

5.3.
Prepare 30 µl of dextran ladder standard in HPLC vial with 80% of acetonitrile.

5.4.
Place the vials and run the HPLC with the above mentioned settings.

5.5.
Integrate the dextran ladder standard chromatogram with empower software and obtain retention times for at least 12 glucose unit (GU) polymers.

5.6.
Fit the GU value as a function of retention time (t) of dextran standard data with a polynomial equation: GU = a × t + b × t 2 + c × t 3 .5.7.Integrate the chromatograms of samples and record the retention times and peak areas for each sample as outlined in Table 3.

POL Scientific
5.8.Calculate the GU values for each peak and assign structure based on reference GU values from Glycobase database (https://glycobase.nibrt.ie)5.9.
If needed, perform exo-glycosidase digestion of the labeled glycan sample and rerun on HPLC to confirm the structures assigned.
5.10.If internal standard is used, record the area of internal standard and sum of the areas of all N-glycans of a sample as outlined in Table 4 5.11.Perform the calculations for the samples as outlined in Table 4 to quantify the molar yield of glycan units per molar quantity of antibody sample.

Kinetics of glycan release
The kinetics of PNGase F digestion of N-glycans from an antibody molecule provides valuable information for optimization of time and enzyme concentration.The PNGase F enzyme mediated kinetics of digestion of N-glycans from two monoclonal antibodies of different sizes and structure, EG2-hFc or aIL8-hFc, were investigated.EG2-hFc is a chimeric heavy chain only monoclonal antibody of 80 KDa [24], whereas aIL8-hFc is an IgG-1 type humanized monoclonal antibody of 146 KDa [25].The details on antibodies and their production are provided in Supplementary Materials and Methods.Similar molar quantities (0.8 nmol) of both antibodies were used.The rate of digestion of the N-glycans by PNGaseF from EG2-hFc and aIL8-hFc antibodies immobilized to protein-A are presented in Figure 2A and 2B, respectively.The antibody samples eluted from protein-A columns after PNGase F digestion were run in SDS-PAGE to visualize the extent of deglycosylation as a function of time and PNGase F concentration.The corresponding SDS-PAGE images of EG2-hFc and aIL8-hFc are presented in Figure 2C and 2D, respectively.The molecular weight of N-glycan unit is typically range from 1.7 to 2.6 KDa depending on the extent of galactosylation and sialylation.At a PNGase F concentration of 50 units/ml, the rate of digestion of N-glycans from EG2-hFc is at least 4-fold higher than of aIL8-hFc.Though both antibodies share similar Fc fragments from human IgG-1, the presence of light chains in aIL8-hFc might makes it difficult for PNGase F to access and cleave the N-glycans from this antibody.A 5-fold increase in PNGase F concentration enhanced the rate of digestion of N-glycans 2-fold in the case of aIL8-hFc.The majority of mAbs and human antibodies are of similar size as aIL8-hFc and would be expected to show similar kinetics.Though higher enzyme concentration should rapidly release the glycans and reduce the isolation time, the PNGase F concentration and incubation time should be chosen based on the antibody type and the balanced choice between time and cost.Higher incubation temperature may also be used to enhance the rate of isolation of N-glycans by PNGase F.

Glycan recovery efficiency and reproducibility
The efficiency of recovering N-glycans and reproducibility of the glycoform distributions by this method was investigated.The glycans were extracted from 0.8 nmol of EG2-hFc and 0.4 nmol of aIL8-hFc.In both cases, 67 units/ml of PNGase F was used and incubated at 37°C for 24 h.The profiles of N-glycans extracted from EG2-hFc and aIL8-hFc are presented in Figure 3A and 3B, respectively.The total yield of glycans recovered is presented as an inset bar graph within each figure as the average molar units of N-glycans per mole of antibody.Each antibody molecule consists of 2 units of N-glycans and hence the total yield has a maximum theoretical value of 2. Each N-glycan unit is labeled with one 2-AB molecule at the reducing end of GlcNAc released from N-glycosylation site.Hence, the number of N-glycans is proportional to the total area of labeled N-glycan units.The calculated yields of glycans from EG2-hFc and aIl8-hFc were 97% and 94%, respectively.These results along with SDS-PAGE analysis (Fig. 2A-2D) demonstrate that complete recovery of N-glycans is obtained with the current method without the need of reducing the mAb.EG2-hFc and aIl8-hFc have two distinct glycosylation profiles.EG2-hFc has a glycosylation profile with highly galactosylated and sialylated N-glycans.The reproducibility of the glycan distribution was also very high with average coefficient of variation of 0.012 in the glycoform percentage distributions between the replicates.
Mini protein-G columns can also be used as an alternative to protein-A columns (Table S1).The method was also tested to isolated N-glycans from polyclonal IgGs in human serum (Table S2).

Quantifying macro-heterogeneity
We have tested the ability of the current method to quantify glycosylation macro-heterogeneity of mAb samples produced by CHO-EG2 cells in limited glucose media.The methods (Supplementary Materials and Methods), cell culture performance and the antibody production profiles are provided in Figure S2. Figure 4A represents SDS-PAGE analysis of reduced EG2-hFc samples along with a control antibody sample produced under nutrient-rich onditions (exponential phase).It can be visualized that starting from 24 h there was an increased production of non-glycosylated mAb with time.The fraction of non-glycosylated mAb in the product was quantified with the current method (Fig. 4B).From this analysis, we can observe that only 9% of non-glycosylated antibody was produced in the first 24 h whereas 30% and 14% of non-glycosylated mAb was produced on day 2 and 3, respectively.The corresponding N-glycan profiles of these samples are shown in the Figure 4C.These profiles along with macro-heterogeneity levels (Fig. 4B) indicate that majority of glycosylated mAb was produced until the glucose is exhausted (within 24 h) with reduced galactosylation (compared to control) and later most of the mAb produced is non-glycosylated.Hence, this method would be useful to obtain quantitative information on the extent of both micro-and macro-heterogeneity of glycosylation together along with the kinetics of antibody production under nutrient-limited culture conditions.structures: Verify with a glycan analysis method involving reduction of antibody (in-solution or gel-based method) 2. Neither internal standard nor glycans peaks were detected.
2.1 Glycans were not labeled: Check the labeling solution.
2.2 Glycans were lost during processing: Check the glycan clean-up cartridges.

Figure 1 .
Figure 1.Schematic representation of the steps involved in the process of isolation and characterization of N-glycans from antibodies.The steps in the dashed-line boxes are optional and only needed to quantity the molar yield of glycans from antibodies.IS: internal standard; HPLC, high performance liquid chromatography; CE, capillary electrophoresis; MS, mass spectrometry, A280, absorbance at 280 nm.

4. 2 - 4 . 1 . 4 . 6 . 4 . 7 .
AB labeling and clean-up of glycans Add 5 µl of 2-AB solution to the dried glycans tube from step 2.8 and mix.4.2.Incubate at 65°C for 2 h.4.3.Wash the HyperSep TM Diol cartridge with 1 ml of ultrapure water.4.4.Condition the cartridge with 4 ml of acetonitrile (1 ml each time).4.5.Add 40 µl of acetonitrile to the labeled glycans, mix the solution and transfer the glycans on to the top of the cartridge.Let it sit for 15 min to allow the glycans to bind.Wash the cartridge with 6 ml of acetonitrile (1 ml at each time).

Figure 2 .
Figure 2. Kinetics of PNGase F digestion of N-glycans from monoclonal antibodies immobilized onto protein-A.A and B. Quantities of glycans released by the PNGase F digestion of immobilized EG2-hFc (A) and aIL8-hFc monoclonal antibodies (B), as a function of digestion time with two different PNGase F concentrations.C and D. SDS-PAGE analysis under reducing condition of corresponding EG2-hFc (C) and aIL8-hFc (D) samples eluted from Protein-A column after digestion with PNGase F for indicated hours (top line).L, Protein ladder, C, Control sample (t = 0).

Figure 3 .Figure 4 .
Figure 3. N-glycan profiles of EG2-hFc (A) and aIL8-hFc (B) obtained by Protein-A method.The glycans were labeled with 2-AB and analysed by HILIC-HPLC method.Each figure consists chromatograms of 6 replicates from three independent analyses, normalized with the average area of internal standard.The mole to mole ratio of recovered glycans is shown in the inset of each panel.The table represents corresponding percentage glycan distribution in each antibody.The error represents S.D. of the values from six replicates.

TROUBLESHOOTING 1 . 1 . 1 1 . 2 3
Samples contain only internal standard peak with no trace of glycans.No antibody captured: Check pH of mAb solution.Antibody captured: Check PNGase F activity or verify the reaction buffer pH.1.Antibody with uncommon Fc sequence or N-glycan