Hydrophilic Interaction Chromatography HRMS with Acrylamide Monolithic Columns: A Novel Approach for Intact Antibody Glycoform Characterization

Glycosylation plays a key role in the pharmacokinetics and efficacy of therapeutic monoclonal antibodies (mAbs). Accurate glycoform profiling is essential, yet conventional RPLC-MS lacks the selectivity to resolve intact glycoforms and often misses low-abundance species. Hydrophilic interaction chromatography (HILIC) offers strong resolving power, but has not been applied to intact mAbs until now.

This study presents acrylamide monoliths for HILIC-MS intact mAb analysis. By tuning porogen composition, column performance was optimized, achieving excellent glycoform selectivity and sensitivity. Applied to five reference mAbs, the method delivered baseline separations (e.g., Rs 3.62 for G0F vs G0F/G0F in trastuzumab) and detected low-abundance glycoforms such as single G0F and M5/M5. Compared with RPLC-MS, acrylamide-monolith HILIC-MS provides improved resolution and opens new opportunities for bioanalytical studies of IgG heterogeneity.

The original article

Hydrophilic Interaction Chromatography HRMS with Acrylamide Monolithic Columns: A Novel Approach for Intact Antibody Glycoform Characterization

Annika A. M. van der Zon*, LoÏs N. Hana, Huda Husein, Thomas Holmark, Ziran Zhai, and Andrea F. G. Gargano*

Anal. Chem. 2025, 97, 25, 13569–13576

https://doi.org/10.1021/acs.analchem.5c02033

licensed under CC-BY 4.0

Selected sections from the article follow. Formats and hyperlinks were adapted from the original.

Monoclonal antibodies (mAbs) typically contain N-linked oligosaccharides (N-glycans) attached to the Fc region of the heavy chain. Modifications in their glycosylation profiles can impact their pharmacokinetics and protein efficacy. (1) Thus, determining the glycoform profile of therapeutic mAbs is essential. Multilevel approaches, including released glycans, peptide, subunit, and intact levels, are therefore used for product characterization. (2,3) Among these, the characterization of glycoproteins at an intact level provides information about the specific glycoform combinations, some of which cannot be retrieved at other levels (e.g., single N-glycosylation). Moreover, intact protein characterization reduces the sample preparation steps and, therefore, the likelihood of artificial modifications.

However, current separation and mass spectrometry (MS) techniques face limitations in intact-level analysis, hindering glycoform profiling of mAbs, particularly for detailed glycoform characterization. A key challenge lies in effectively separating intact mAb glycoforms. (4) Reversed-phase liquid chromatography (RPLC) coupled with MS is the most commonly used method for analyzing mAbs. (5,6) However, RPLC relies on a hydrophobicity-based retention mechanism, which lacks sufficient glycoform selectivity, leading to coelution of the glycoforms and limiting the range of identifiable species. Other approaches, including ion-exchange chromatography (IEC) or capillary electrophoresis (CE), have limited application in resolving neutral glycoforms of mAbs. (7)

Hydrophilic interaction chromatography (HILIC) is a promising alternative for glycoform separations, allowing for more in-depth characterization of glycoproteins by LC-MS approaches. In glycoprotein HILIC separations, neutral stationary phases (amide chemical selectors) are used with mobile-phase gradients of acetonitrile (ACN) to water, in combination with mobile-phase additives, like trifluoroacetic acid (TFA), that decrease the pH of the eluent and allow ion-pair formation with basic protein residues. (8−10) Under these conditions, glycans significantly contribute to protein retention, enabling high-resolution separation of glycoforms. This makes HILIC an ideal approach for glycoprotein analysis at the intact level, (9,11) with recent applications reporting high-resolution separations, such as biopharmaceuticals (e.g., erythropoietin), (8,12−15) and the Fc portion of serum immunoglobulins. (16)

Our group recently described poly(acrylamide-co-N,N-methylene-bis(acrylamide)) monolithic stationary phases for HILIC of proteins. (17) Monolithic column synthesis conditions can be tuned to modify pore morphology and surface area to create materials specifically designed for macromolecular separations. (18) Compared with silica-based stationary phases, acrylamide monoliths can be used with reduced percentages of TFA (e.g., 0.005% (v/v)). However, the presence of TFA remains crucial for effective glycoprotein analysis, as demonstrated by Passamonti et al. (19)

We have successfully applied polyacrylamide monoliths to the HILIC-MS characterization of glycoforms from glycoproteins up to 40 kDa, including N- and O-glycans from the SARS-CoV-2 spike receptor-binding domain (∼32 kDa), horseradish peroxidase (∼40 kDa), as well as other glycoproteins. (20,21) This study aims to extend their application to the separation of glycoforms of immunoglobulin G (IgG) mAbs (∼150 kDa). Separating intact mAb glycoforms is challenging due to the combination of slow mass transfer caused by the large size of the antibodies and the small mass difference of the glycans relative to the overall mass of the glycoprotein (e.g., around 1.5% for a single Fc glycan in a mAb of about 150 kDa). (11) To optimize the separation performance of our monolithic materials, we tested several polymerization conditions to vary their pore properties. This was aimed at increasing the separation performance and reducing potential mAb carryover. A polymerization condition was selected and used for the HILIC-MS analysis of five IgG mAbs. Finally, we compared the masses observed in the analysis of a reference mAb using HILIC-MS with state-of-the-art RPLC-MS analysis.

Experimental Section

HILIC-UV

A Vanquish Neo nano-LC system (Thermo Fisher Scientific, Bremen, Germany) was used. We employed a trap-and-elute configuration to load samples onto the capillary column. (10) The weak solvent consisted of 2% (v/v) ACN in 98% water. The sample was first loaded onto a C4 trap column (0.3 mm i.d. × 5 mm, 5 μm, 300 Å, Thermo Fisher Scientific, Bremen, Germany) with a flow rate of 10 μL·min^–1. The mobile phases for the analytical pump were (A) 0.1% (v/v) TFA in 98:2 water:ACN and (B) 0.1% (v/v) TFA in 2:98 water:ACN. The gradient started at 95% B and decreased to 71% B in 3 min. Then, a gradient was programmed from 71 to 66% B in 20 min, followed by a reduction of mobile phase B to 50% in 5 min and subsequently to 20% B in 1 min. After 1 min, there were two washing cycles from 80 to 20% B for each 1 min. A re-equilibration step at 95% B took place for 9.9 min (factor of 2). For all the analyses, the flow rate was set to 1 μL·min^–1 and carried out at 50 °C. The injection volume was 2 μL (20 μL sample loop). Three blank injections of water were measured between the analyses of different mAb samples. For long-term storage, the columns were stored in 100% B, with both ends of the column inserted into LC vials containing this solvent (Figure S1).

HILIC-MS

A Q Exactive Plus MS instrument from Thermo Fisher Scientific (Bremen, Germany) was equipped with a Nanospray Flex ion source (Thermo Fisher Scientific, Bremen, Germany). The distance of the emitter from the inlet of the orifice was set to 2 mm. The instrument was operated at 2.00 kV (positive polarity) with a capillary temperature of 300 °C, an S-lens RF level of 100, and an HCD gas pressure of 1. The in-source CID was set to 80 eV for the mAbs. The scan range was set to 600–6,000 m/z (high mass range (HMR) mode), with 10 microscans and 200 ms as the maximum injection time. The resolution was set to 17,500 (200 m/z).

Data Processing

The UV data were processed using Chromeleon 7 (Thermo Fisher Scientific, Bremen, Germany). The raw MS data were visualized in Freestyle. The peak width of the mAbs was determined by using Genesis as the peak detection algorithm, with a peak height of 50%. For the deconvolution of the spectra and calculation of the average mass, Unidec (University of Arizona, Phoenix, USA) was used with the following parameters: sample mass rate of 0.1 Da, picking range of 1 Da, and picking threshold of 0.02. (23) The raw MS data of intact mAb can be found here: 10.5281/zenodo.15118514.

Results and Discussion

Separation of Glycoforms of Mabs at the Intact Level

ACN/water mobile phases with TFA (e.g., 0.1% (v/v)) are commonly used in HILIC, as demonstrated in several studies, (3,19,30−32) and are therefore applied in our study. While steep gradients (e.g., 90 to 50% ACN in 20 min) are used to separate mixtures of proteins with large differences in terms of amino acid sequence composition, (10) shallow gradients (e.g., 5–10% reduction of ACN over 30 min) are necessary to achieve highly efficient glycoform separations. (33)

Therefore, we began our investigation by studying the effect of reducing the ACN gradient window from 20% (85–65% (v/v) ACN) to 5% (71–66% (v/v) ACN) over 20 min (Figure 1). Trastuzumab was used as a model mAb for these experiments. Reducing the steepness of the mobile phase gradient resulted in a substantial improvement in glycoform resolution (Rs) (e.g., Rs between G0F/G0F and G0F/G1F of 0.45, 0.79, and 1.06 at 1%, 0.5%, and 0.2% B/min, respectively). This enabled the separation of up to six glycoform peaks at 0.2% B/min compared to two glycoform peaks at 1% B/min (Figure 1). HILIC-UV separations under these conditions were attempted, resulting in lower separation performance and only partial separations (Figure S5). We suggest that the separation loss is related to the large sample mass injection needed to obtain clear UV signals (0.5 μg using UV vs 0.1 μg in MS) and system dead volumes. The optimized conditions were tested with both O28/D72 and O26/D74 columns, confirming the observation from the analysis of the protein mixture, with O26/D74 offering better separation performance compared to O28/D72 (Figure S6).

Anal. Chem. 2025, 97, 25, 13569–13576: Figure 1. BPC of trastuzumab (0.1 mg·mL–1, 1 μL injection volume) showing the influence of the steepness of the ACN gradient on the glycoform separation. Top to bottom: gradient from 85 to 65% B (steepness 1.0%B·min–1), 75–65% B (steepness 0.5%B·min–1), and 71–66% B (steepness 0.2%B·min–1) in 20 min. Details of the glycans are given in Table S4. Isomeric structures such as G0F/G2F are also possible but not illustrated.

The optimized method conditions were used for the HILIC-MS analysis of five IgG mAbs, each injected at 100 ng per analysis: three IgG₁ mAbs (trastuzumab, NISTmAb, and ipilimumab) and two IgG₄ mAbs (nivolumab and pembrolizumab). All analyzed mAbs are Fc glycosylated and have high similarity in terms of Fc amino acid (AA) sequence (with respect to trastuzumab % identity of 100% for NISTmAb, missing one AA; ≈99% for ipilimumab, with two AAs different; and ≈95% for the IgG₄ models, with nine AAs different) but show significant differences in their Fab subunit (% identity between ≈85% and 76%, with up to 60 AA differences). Details of the structural characteristics of the analyzed model mAbs are reported in Section S4.

The base peak chromatograms (BPCs) and extracted ion chromatograms (EICs) from their LC-MS analysis are shown in Figures 2 and S7, and the glycoform structures are summarized in Table S4. The five mAbs eluted between 11 and 17 min. For the same glycoforms (e.g., G0F/G0F) across different mAbs, retention times varied within a 1- to 2-min window (e.g., around 12 min for single Fc glycosylation G0F). Peak widths differed between mAbs (e.g., narrower G0F/G1F peaks for NISTmAb compared to pembrolizumab, Table 2) and were generally broader for larger glycans. These differences did not appear to correlate with properties related to AA sequences, such as isoelectric point, sequence composition, or molecular weight (Table S5). Multiple distributions were observed in the EIC and were more pronounced in the case of larger glycans (e.g., G1F/G2F, Figure S7), suggesting the presence and separation of isomeric structures. While definitive identification of these species remains challenging, we propose several potential explanations: these could include deamidation variants (which HILIC can resolve at the peptide level), (34) glycated species (as suggested in Figure S8), resolution of isomeric glycans (e.g., G1F isomers observed at the released glycan level), (35) or even disulfide bond scrambling.

Anal. Chem. 2025, 97, 25, 13569–13576: Figure 2. (A) BPC with EIC traces of separated glycoforms of trastuzumab, (B) Deconvoluted spectra of trastuzumab at different elution times, expressed in min, as shown in each panel, (C) BPC with EIC traces of separated glycoforms of nivolumab. The EICs of single G0F, G1F, and deglycosylated peaks (dotted) are five times scaled. Details on the glycans and EIC traces are described in Tables S4 and S7. The glycoforms are illustrated according to ref (38). Isomeric structures such as G0F and G2F are also possible but not illustrated.

Glycoform separations were observed in all BPCs of both IgG₁ and IgG₄ mAbs, with three to six glycoform peaks (EIC Rs > 0.8) observed depending on the mAb (Figure S7 and Table 2). The BPC profiles generally reflected the abundance of the different glycoforms present. NISTmAb and trastuzumab showed four main glycoforms (90% of the observed signal, G0F/G0F to G1F/G2F), and nivolumab and ipilimumab showed two main glycoforms (95% of the signal, G0F/G0F and G0F/G1F). Similar to what was described in the HILIC analysis of Fc subunits of mAbs, (3) the elution order of the Fc glycoforms is related to the mass of the glycan (e.g., G0F/G1F elutes after G0F/G0F). mAb glycoforms with a single Fc glycan (the most abundant being G0F and G1F) were baseline separated from the main isoform having two Fc glycosylations (e.g., G0F vs G0F/G0F, Rs between 2.91 and 5.07, Table 2 and Figure S7). The resolution between two glycoforms was lower (e.g., G0F/G0F and G0F/G1F from 0.62 to 2.03). In addition to nivolumab, we observed a low-abundance deglycosylated peak, baseline resolved from the single Fc-glycosylated peak (Figure S7).

We attempted to replicate similar separations using a commercially available analytical-scale silica-based HILIC column (150 × 2.1 mm I.D., 1.7 μm, 300 Å glycoprotein amide; details and conditions summarized in Table S8) previously applied for glycoprotein analysis. (36,37) Using the same mobile phase and shallow gradients (2% B in 20 min), we observed coelution of the glycoforms of trastuzumab (Figure S9). We suggest that this difference between silica and acrylamide monolithic columns may be related to the material’s chemistry and/or morphology. Specifically, the silica column may have residual silanols, which act as acidic interaction sites, leading to secondary ion-exchange interactions with the mAb, which may negatively influence the separation of the glycoforms. Furthermore, the particle-bed morphology of the silica column, compared to the monolithic structure of our columns, may also contribute to the observed differences.

To investigate this further, we performed HILIC-UV experiments with both silica-based and acrylamide monolithic columns, varying the TFA concentration from 0.02 to 0.3% (v/v) and comparing trastuzumab elution under similar conditions (programmed gradient time/t0) (Figure S10). The silica column showed no glycoform separation at 0.3% (v/v) TFA, while partial separation was observed with the monolithic column (confirmed by HILIC-MS experiments in the case of 0.1% (v/v) TFA). At 0.02% (v/v) TFA, recoveries were low for both columns, indicating the importance of the concentration of TFA in the separation. In the case of acrylamide monoliths, the low recovery could also be related to interaction with silica (also present in the C4 trap column used for injection in HILIC experiments). Moreover, with the silica column, elution time remained essentially unchanged between 0.1% (v/v) and 0.02% (v/v) TFA, but the peak width significantly broadened, suggesting the presence of secondary ion-exchange interactions. Although more pronounced tailing was observed for the monolithic column at 0.02% (v/v) TFA, this column exhibited a consistent reduction in retention time as the TFA concentration decreased from 0.3 to 0.02% (v/v).

Conclusion

In conclusion, this study demonstrates the significant potential of acrylamide monolithic stationary phases for the high-resolution separation of mAb glycoforms. Optimization of the porogen composition, specifically increasing DMSO content, enhanced the stationary phase’s surface area and selectivity while reducing peak widths. The optimized acrylamide monolithic columns enabled, for the first time, detailed glycoform separations of intact mAbs (IgG₁ and IgG₄) using HILIC-MS. This approach offers a distinct advantage over conventional RPLC-MS, providing more comprehensive glycoform profiling and enabling the characterization of low-abundance species. The sensitivity of the method, thanks to the capillary column format, allows for reduced sample injection amounts (100 ng), making it promising for IgG studies in sample-limited bioanalytical applications.