The nonglycosylated variant in therapeutic monoclonal antibodies preferentially forms large aggregates under typical thermal stresses used in forced degradation studies

ABSTRACT

Monoclonal antibodies (mAbs) feature a conserved N-linked glycosylation site in the CH2 domain, which exhibits heterogeneities in both occupancy and glycan structures. Previous studies have suggested that the unoccupied (nonglycosylated) variant exhibits decreased thermal stability, potentially impacting the overall stability of mAb products. This hypothesis, however, has remained largely unconfirmed, due to the low abundance of nonglycosylated variants in typical mAb products and the lack of effective analytical tools for detailed characterization of large aggregates with glycoform-specific information. Here, we used a postcolumn denaturation-assisted size exclusion chromatography mass spectrometry technique (SEC-PCD-MS) to reevaluate the effects of the nonglycosylated mAb variant on the thermal stability of mAb drugs during forced degradation studies. Our findings confirmed the compromised thermal stability of the nonglycosylated variant and its increased propensity to form large aggregates at elevated temperatures relevant to mAb-forced degradation studies. We also showed that this thermal stress-induced, nonglycosylation-mediated aggregation pathway could be widely observed in a diverse group of mAb molecules with varying properties. This study offers valuable insights into the rationale of selecting the appropriate temperature for mAb-forced degradation studies and highlights key considerations for data interpretation.

Introduction

Over the past decades, monoclonal antibody (mAb)-based therapeutics have revolutionized the treatment landscape for a variety of human diseases, ranging from viral infections, cardiovascular diseases, and autoimmune diseases to many types of cancer. Therapeutic mAbs are inherently complex molecules that can host a wide variety of post-translational modifications (PTMs), which may arise during their production and storage. Therefore, unlike small molecule counterparts, therapeutic mAbs are essentially a mixture of closely related proteoforms that collectively contribute to their overall properties and therapeutic functions. Among various PTMs, the conserved N-linked glycosylation, located within the CH2 domain of each heavy chain in all IgG subclasses, is considered a critical quality attribute that can exhibit a large degree of macro- and micro-heterogeneities during production. The presence or absence of this N-glycan, along with specific N-glycan structures, is known to affect the function and stability of the molecule.^1–8 For example, IgG1 molecules lacking core fucosylation are well recognized for their improved binding to Fc gamma receptor IIIa, making them highly attractive in therapeutic applications where an enhanced effector function is desired.^9,10 On the other hand, mAb molecules with high mannose N-glycans have been shown to exhibit suboptimal pharmacokinetic properties compared to those with complex N-glycans, showing increased clearance from the circulation and reduced half-life.^11,12

In addition to therapeutic functions, the Fc N-glycosylation status can also have a significant impact on a mAb molecule’s stability.¹³ In particular, nonglycosylated mAbs (i.e., the two N-glycosylation sites are both unoccupied) have demonstrated decreased thermal stability compared to their fully glycosylated counterparts (i.e., both N-glycosylation sites are occupied), as evidenced by reduced melting temperature reported by differential scanning calorimetry (DSC) experiments.¹⁴ These findings also correlated well with structural studies, such as those using hydrogen-deuterium exchange mass spectrometry (HDX-MS), suggesting that the absence of Fc N-glycan results in an increased protein backbone flexibility in the CH2 domain.^15,16 While prior reports have clearly demonstrated the compromised thermal stability of nonglycosylated mAb molecules,^13,17,18 an improved understanding is still needed regarding how these nonglycosylated variants, often present in relatively low abundances (e.g., <20%), could influence the overall stability of mAb products under thermal stress conditions relevant to their development, such as during forced degradation studies.

Indeed, forced degradation studies, particularly those performed at elevated temperatures, are integral for the development of therapeutic mAbs. Using a temperature that is notably higher than the long-term storage temperature (e.g., 4–8°C) or the manufacturing exposure temperature (e.g., 25°C), forced degradation studies can reveal the potential degradation pathways of the mAb molecule within a short period of time (e.g., within weeks¹⁹). Such studies can be performed for different purposes throughout the various stages of mAb development, including developability/manufacturability assessment for candidate selection, formulation development, evaluation and development of stability-indicating analytical methods, and comparability assessment after process changes.²⁰ According to a recent industry-wide survey, the majority of the respondents used a temperature range of 40°C to 50°C over a period of 1–4 weeks for thermal stress when designing forced degradation studies of mAbs.²¹ This temperature range is selected because it allows for sufficient degradation of mAbs within the desired timeframe without reaching their melting temperature (T_m), which would lead to non-representative unfolding/denaturation. While this temperature range is comfortably below the T_m of a fully glycosylated mAb molecule, it is notably closer to the T_m of its nonglycosylated variant.¹⁴ Given that protein aggregation is one of the major degradation pathways for mAbs under thermal stress,²² it is reasonable to speculate that the nonglycosylated variant, although present in low abundance in most mAb drug products, may more readily contribute to aggregate formation compared to its fully glycosylated counterpart during forced degradation studies. This hypothesis, however, remains largely unconfirmed, presumably due to the lack of effective analytical tools that enable detailed aggregate characterization with glycoform-specific information.

Commonly, size exclusion chromatography (SEC) with ultraviolet (UV) detection is the most frequently used technique to monitor the aggregate formation of mAbs under thermal stress.²³ While this technique is robust, sensitive, and stability-indicating, it does not provide detailed characterization of the aggregates. When SEC separation is coupled with multi-angle light scattering (MALS) or native mass spectrometry (MS), it can measure the molecular weights of the aggregates, enabling the determination of the oligomerization states (e.g., dimer, trimer, etc.^24,25). However, more detailed analysis often requires the fractionation of the mAb aggregates, followed by extended characterization using various biophysical tools. For instance, combining limited proteolysis and LC-MS, the dimerization interfaces involved in mAb aggregation can be revealed at the domain levels (e.g., Fab-Fab, Fab-Fc, and Fc-Fc^26,27). Moreover, MS-based bottom-up analysis has also been successfully used to identify the covalent crosslinks that contributed to the formation of covalent mAb aggregates.^28,29 While the fractionation-based strategy is highly effective, it is also very demanding in terms of material and time. Moreover, due to the additional stress posed during the fractionation process, it may also disrupt the existing aggregates or even introduce artifactual ones.³⁰ Recently, we reported a postcolumn denaturation-assisted SEC-MS method (SEC-PCD-MS) that allows for rapid and in-depth characterization of mAb aggregates directly from the unfractionated drug samples.²⁶ By breaking down the mAb aggregates, which are often too large and heterogeneous for meaningful MS measurement, a more detailed characterization can be achieved on the released monomeric species. Hence, we aimed to use this technique to reexamine the impact of nonglycosylated mAb variant on the overall stability of mAb drugs under thermal stress conditions encountered in forced degradation studies.

In this report, we first present a case study showing the formation of large aggregates of a mAb sample when subjected to 45°C thermal stress. Using SEC-PCD-MS analysis, it was discovered that, upon thermal stress, the nonglycosylated variant was preferentially depleted from the monomer SEC peak and significantly enriched in the large aggregate SEC peak. This observation was more pronounced at 45°C compared to 40°C, and it was negligible at lower temperatures. After screening 24 different mAb molecules, the preferential aggregation of nonglycosylated variant in mAb samples was found to be very common at 45°C, but less so at 40°C. Finally, by spiking an aglycosylated mAb into its glycosylated counterpart at different levels and subjecting them to 45°C thermal stress, we demonstrated that the reduction of nonglycosylated species in the monomer, the enrichment of them in large aggregates, and the increase of large aggregate SEC peak were all closely correlated. For the first time, we showed that the nonglycosylated mAb variant, although present in low abundance in most mAb drug products, preferentially contributed to large aggregate formation under thermal stress conditions commonly used in forced degradation studies.

Materials and methods

Materials

Deionized water was provided by a Milli-Q integral water purification system installed with a MilliPak Express 20 filter (Millipore Sigma, Burlington, MA). Invitrogen UltraPure 1 M Tris-HCl buffer, pH 7.5, and acetonitrile (ACN) (LC-MS grade) were purchased from Thermo Fisher Scientific (Waltham, MA). Formic acid (FA) (Fluka) was purchased from Honeywell (Morris Plains, NJ). Ammonium acetate (LC-MS grade) and 2-propanol (IPA; HPLC grade) were purchased from Sigma Aldrich (St. Louis, MO). All therapeutic mAbs used in this study were produced by Chinese hamster ovary (CHO) cells in-house at Regeneron (Tarrytown, NY). Strict sample handling protocols were implemented to minimize artificially induced aggregation during analysis. In particular, mAb samples were promptly stored at −80°C following thermal stress and were subjected to only a single freeze-thaw cycle before analysis. Additionally, the mAb samples were maintained in the autosampler at 5°C for a maximum duration of 24 hours prior to LC-MS analysis.

SEC-PCD-MS analysis

PCD-assisted SEC-MS was performed as described previously.^26,31 Briefly, native size exclusion chromatography (nSEC) was performed on a UltiMate 3000 UHPLC System (Thermo Fisher Scientific, Bremen, Germany) using an Acquity BEH200 SEC column (4.6 × 150 mm, 200 Å, 1.7 μm, Waters, Milford, MA) with a column temperature of 30°C. The antibody size variants were eluted with an isocratic flow of 150 mM ammonium acetate at 0.2 mL/min. For post-column denaturation (PCD), a denaturing solution consisting of 60% ACN, 36% water, and 4% FA was delivered using a secondary pump at a flow rate of 0.2 mL/min and mixed with the SEC eluent (1:1 ratio) using a T-mixer before MS detection. To enable online native MS analysis, the combined analytical flow (0.4 mL/min) was split into a microflow (<10 μL/min) for nanoelectrospray ionization (NSI)-MS detection and the remaining high flow for UV detection. A Thermo Q Exactive UHMR (Thermo Fisher Scientific, Bremen, Germany) equipped with a microflow-nanospray electrospray ionization (MnESI) source and a microfabricated monolithic multinozzle (M3) emitter (Newomics, Berkeley, CA) was used for native MS analysis.

Data analysis

Intact mass spectra from nSEC-MS analysis under native or PCD conditions were deconvoluted using Intact Mass software from Protein Metrics (Cupertino, CA). Extracted ion chromatograms were generated using XCalibur from Thermo Fisher Scientific (Bremen, Germany).

Results and discussion

Enrichment of nonglycosylated variant in large aggregates in mAb-A forced degradation study

During the development of mAb-A, a bispecific IgG4 subclass antibody, a forced degradation study was conducted by incubating the mAb-A formulated drug substance (FDS, 50 mg/mL) sample at 45°C for up to 28 days. Subsequent SEC-UV/MS analysis revealed notable increases of multiple size variant peaks compared to the T0 control sample, including the low molecular weight (LMW) peak, high molecular weight (HMW) peak, and the large aggregate (i.e., very high molecular weight, vHMW) peak (Figure S1a–e). Intriguingly, the vHMW peak showed the most significant increase compared to the other size variant peaks and was only observed upon thermal stress. Using concurrent native MS analysis, the LMW peak and HMW peak were identified as truncated mAb and mAb dimer species, respectively (Figure S1b,c). The mass spectra from the vHMW peak, on the other hand, were too heterogeneous to be deconvoluted (Figure S1e). After comparing the m/z range from the native mass spectrum of each species within the vHMW peak with those of known mAb aggregates, it was proposed that the vHMW peak was attributed to higher order oligomers of mAb-A, such as tetramer, pentamer, hexamer, and potentially even larger species (Figure S2a,b). However, due to the large size and heterogeneity, direct MS analysis of this peak did not provide meaningful insights into its root cause.

To investigate the underlying mechanism responsible for the large aggregate formation, SEC-PCD-MS analysis was performed on both the stressed sample and the T0 control sample. Interestingly, under the PCD conditions (i.e., 75 mM ammonium acetate, 2% FA, and 30% ACN), the large aggregates in the vHMW peak were effectively dissociated into monomers, which could be readily analyzed by MS, producing high-quality MS spectra (Figure S2c). After deconvolution, it is clear that the mAb-A species released from the large aggregates exhibited an unusually high level of nonglycosylated variant (30.4%) (Figure 1b). Given that the majority of aggregates in the vHMW peak were dissociated under PCD-MS conditions, as evidenced by the minimal remaining oligomer signals (Figure S2c), it is reasonable to infer that the measured nonglycosylation level from the dissociated monomer is representative of the large aggregate species. The deconvoluted mass spectra of the monomer SEC peak also showed a notable decrease in the nonglycosylated variant from the thermally stressed sample (0.6%) compared to the T0 control sample (1.7%).

Figure 1.

SEC-PCD-MS analyses of unstressed and 45°C, 4 weeks stressed mAb-A DS samples. (a) SEC-UV trace; (b) deconvoluted mass spectra of vHMW and monomer regions analyzed under PCD conditions; (c) XICs of nonglycosylated (NG), partially glycosylated (PG), and fully glycosylated (FG) mAb-A population from SEC-PCD-MS analysis of 45°C, 4 weeks stressed sample.

Together, these observations suggest that the nonglycosylated variant was preferentially transitioned from the monomeric state into the large oligomeric states under 45°C thermal stress. This behavior can also be readily visualized using the extracted ion chromatograms (XICs), generated from the SEC-PCD-MS analysis. Compared to the fully glycosylated main species (Figure 1c, green trace), which almost entirely remained in the monomer peak, the nonglycosylated variant (Figure 1c, blue trace) showed a significant enrichment in the vHMW peak and, to a much lesser extent, in the dimer peak. Interestingly, the partially glycosylated variant (Figure 1c, cyan trace), which has N-glycosylation on one of its two heavy chains, also showed minor enrichment in the vHMW peak. This suggests that the partially glycosylated variant also has a greater tendency to form large aggregates under thermal stress compared to the fully glycosylated main species, albeit not as much as the nonglycosylated variant. Finally, the same XIC analyses were also performed for specific glycoforms (e.g., G0F/G0F, G0F/G1F, and G1F/G1F) of the fully glycosylated species, which revealed highly similar profiles (Figure S3). This observation suggests that these fully glycosylated glycoforms exhibited comparable stability under the applied thermal stress conditions. Given that the nonglycosylated antibodies are well known for their reduced thermal stability in the CH2 domain,¹⁷ it is reasonable to hypothesize that under 45°C thermal stress, a fraction of the nonglycosylated variant may gradually unfold, exposing hydrophobic patches. These unfolding events could progressively lead to the formation of large aggregates through inter-molecular hydrophobic interactions. Additionally, as highlighted in previous works, several aggregation-prone motifs³² and hydrophobic residues¹⁸ near the glycosylation site may undergo localized conformational changes or become exposed after the removal of N-glycan. These structural characteristics within the Fc region of the nonglycosylated variant might contribute to its increased aggregation propensity. Interestingly, a recent report demonstrated that the thermal destabilization effect resulting from the removal of N-glycans in nonglycosylated antibodies can be partially compensated by introducing small-molecule drugs through conjugation approaches.⁷ It is hypothesized that by occupying the interstitial space between the two CH2 domains, these small-molecule drugs provide stabilizing interactions to the CH2 domain. These findings align with our observations, where the partially glycosylated variant exhibited lower aggregation propensity compared to the nonglycosylated variant. This difference is likely attributed to its ability to occupy a greater interstitial space, which improves its resistance to thermal unfolding. It is important to point out that after thermal stress, the fully glycosylated variant still constituted a significant portion of the observed vHMW species (Figure 1b). This observation suggests that the unfolded, nonglycosylated CH2 domain may also increasingly interact with other fully and partially glycosylated species, presumably serving as a nucleation core for large aggregate formation. Indeed, recent findings did support theories that electrostatic interactions, originating from both the Fab and Fc regions, play a significant role in mAb self-interaction, which is influenced by the presence or absence of Fc N-glycosylation.³³

While 45°C thermal stress is commonly applied in forced degradation studies of mAbs, other temperatures are also frequently used in various stability studies, such as long-term storage stability and accelerated stability. To examine whether the nonglycosylated variant also contributes to large aggregate formation under milder thermal stress conditions, additional thermal stability studies of mAb-A were conducted. These studies included testing at 5°C for 6 months, 25°C for 3 months, 30°C for 3 months, 35°C for 1 month, and 40°C for 1 month. Due to the material availability, a mAb-A FDS sample formulated at a higher concentration (150 mg/mL) was used. Under the conditions tested, a similar vHMW peak was only observed in the sample incubated at 40°C for 1 month (Figure S4a). Subsequent SEC-PCD-MS analysis of this sample also confirmed the reduction of nonglycosylated variant in the monomer peak and its corresponding enrichment in the vHMW peak (Figure S4b). These findings indicate that a thermal stress temperature greater than 40°C, which is close to the melting onset temperature of nonglycosylated antibodies, is likely necessary to trigger this nonglycosylation-mediated aggregation pathway. While this unique aggregation pathway is unlikely to occur under typical mAb long-term storage (4–8 °C) or accelerated stability conditions (25°C to 37 °C), and thus, low levels of nonglycosylated variant may not pose a significant concern for an mAb product from a stability perspective, it highlights important considerations for the use of high-temperature stability data (e.g., >40 °C) in predicting low-temperature stability through modeling approaches.

Nonglycosylation-mediated aggregation under thermal stress in diverse mAb molecules

We next sought to explore if this aggregation pathway, driven by nonglycosylation under thermal stress, is present in a broad range of mAb molecules. A total of 24 additional mAb molecules, encompassing both IgG1 and IgG4 subclasses, as well as monospecific (msAb) and bispecific antibody (bsAb) formats, were subjected to thermal stress under both 40°C and 45°C for up to 4 weeks. These molecules exhibited a wide range of nonglycosylation levels, spanning from 0.1% to 5.4%, as determined by deconvoluted mass spectra (Table S1). Each molecule was examined using its respective FDS sample, with concentrations ranging from 37.1 to 217.6 mg/mL and formulation pH values spanning from 5.0 to 6.0 (Table S1). Common formulation excipients found in mAb products, such as sucrose and polysorbates (PS20 and PS80), were included in most of the formulations. Therefore, this list should effectively represent a diverse range of mAb FDS samples that may be encountered during development (Table S1).

As shown in Figure 2a, a majority of mAb molecules experienced a notable growth of large aggregates under 45°C thermal stress, with a median increase of ~7% in vHMW peak abundance after 4 weeks. This trend and the magnitude of increase were much lower for 40°C thermal stress. Concomitantly, a larger decrease in the relative abundance of the nonglycosylated variant in the monomer SEC peak was also observed for most mAb molecules under 45°C stress compared to those under 40°C stress (Figure 2b). Meanwhile, most mAb molecules exhibited an anticipated enrichment of nonglycosylated variant in large aggregates under 45°C stress, with only a few exceptions (Table S1, likely due to formation of larger aggregates unrecoverable by SEC). Three examples were highlighted: an IgG1 msAb (Figure 2c), an IgG4 msAb (Figure 2d), and an IgG4 bsAb (Figure 2e), each demonstrating the simultaneous formation of a vHMW peak, a reduction in the nonglycosylated variant within the monomer SEC peak, and its enrichment in the vHMW SEC peak. These results demonstrated that this nonglycosylation-mediated aggregation pathway can occur in both IgG1 and IgG4 subclasses, in either monospecific or bispecific formats. Notably, mAb-2 (Figure 2c) and mAb-20 (Figure 2e) FDS samples only contained 0.4% and 1.3% of the nonglycosylated variant, respectively. Despite this, the formation of noticeable large aggregates was still observed under 45°C stress.

Figure 2.

SEC-PCD-UV/MS analyses of 24 mAbs stressed under 40 and 45°C for 4 weeks. (a) vHMW growth measured by SEC-UV. (b) Decrease of relative NG population in the monomer species. (c-e) Three examples of mAbs stressed under 45°C for 4 weeks as analyzed by SEC-PCD-MS showing the redistribution of NG in vHMW peak.

This study demonstrated that the nonglycosylation-mediated aggregation under thermal stress might be prevalent and could play an important role in the degradation pathways of a diverse range of mAb molecules in forced degradation studies during their development. As the unfolding of the CH2 domain under thermal stress was hypothesized to initiate this aggregation process, the thermodynamic properties of the mAb molecules, such as the onset and melting temperatures, may play an important role in modulating the initial aggregation. Additionally, the colloidal stability of mAb molecules, which is influenced by both the molecular surface characteristics (e.g., charge patches and hydrophobic regions) and formulation conditions, is likely to interact with these thermodynamic factors to determine the aggregation behavior. While it is compelling to further explore how specific molecular features (e.g., T_m, charge, hydrophobicity) or formulation conditions might differentially contribute to this interesting aggregation pathway, such exploration requires separate and systematic investigations (e.g., studying mAb characteristics at identical concentrations and formulation conditions), which are beyond the scope of this study. Finally, as both 40°C and 45°C are widely used throughout the industry for forced degradation studies of mAbs,^34,35 this study offers valuable insights into the rationale of selecting the appropriate temperature for such studies and highlights important considerations for data interpretation.

The levels of nonglycosylation correlate with large aggregation formation under thermal stress

While different mAb molecules might exhibit different tendencies to form large aggregates via the nonglycosylation-mediated pathway, it is reasonable to assume that, for the same mAb molecule, the degree of nonglycosylation should correlate with large aggregate formation. To verify this, a spike-in experiment was performed by mixing the mAb-B sample with its nonglycosylated counterpart, creating a series of samples with increasing levels of nonglycosylation. The nonglycosylated (or aglycosylated) mAb-B was produced by substituting the N-glycosite Asn297 with Gln, resulting in a homogeneous pool of mAb molecules lacking any Fc N-glycans. This substitution was selected for its minimal structural impact, introducing only a single ‘CH2’ group to the side chain and maintaining an unchanged charge profile (compared to PNGase F-mediated deglycosylation reaction). Samples with four spike-in levels, 0%, 2%, 5%, and 20%, were prepared and subsequently subjected to thermal stress at 45°C for 2 weeks. After thermal stress, SEC-UV analysis revealed a significant increase in the relative abundances of large aggregates (i.e., vHMW), from 9.5% in the “0%” spike-in sample to 37.9% in the “5%” spike-in sample, suggesting a positive correlation between the nonglycosylation level and large aggregate formation (Figure 3a). Interestingly, this trend was reversed for the “20%” spike-in sample, which showed a slight decrease in vHMW peak abundance compared to the “2%” and “5%” spike-in samples. It is hypothesized that the large aggregates formed in this sample may grow to a size that can no longer be recovered by SEC analysis. By examining the deconvoluted mass spectra of the monomer and vHMW SEC peaks obtained from SEC-PCD-MS analysis, it is evident that the nonglycosylated variant was preferentially transitioned from the monomeric state into large aggregates under 45°C stress (Figure 3b). This is particularly apparent for the “20%” spike-in sample, where the highly abundant nonglycosylated variant within the monomer peak of the T0 sample dramatically reduced after 2 weeks. Concomitantly, it became the predominant component of the newly formed large aggregates. This phenomenon is consistent across all spike-in samples and is more clearly illustrated by the bar graph, summarizing the abundance changes of the nonglycosylated variant in both monomer and vHMW peaks, before and after thermal stress (Figure 3c). Together, this study demonstrated that the level of nonglycosylation in an mAb sample correlated with the formation of large aggregates under 45°C thermal stress, which was mediated by the nonglycosylated variant.

Figure 3.

SEC-PCD-MS analyses of mAb-B samples spiked with 0, 2, 5, and 20% of nonglycosylated mAb-B and stressed under 45°C for 2 weeks. (a) SEC-UV traces of both T0 and 45°C, 2 weeks stressed samples. (b) Deconvoluted mass spectra and (c) determined relative abundances of NG species from the monomer regions of unstressed mAb-B samples, and monomer and vHMW regions of 45°C, 2 weeks stressed mAb-B samples.

Conclusion

While it is well known that nonglycosylated mAb molecules exhibit reduced thermal stability, it is still unclear how these variants, which are often present in relatively low abundances (e.g., less than 20%), might affect the overall stability of mAb products when subjected to thermal stress conditions in forced degradation studies. In addition to the low abundance of variants, the lack of effective analytical tools for detailed characterization of large mAb aggregates also poses a challenge. In this study, a recently developed SEC-PCD-MS method was applied to reexamine the impact of nonglycosylation on mAb aggregation under thermal stress conditions. For the first time, we showed that nonglycosylated variants directly contributed to the formation of large aggregates under thermal stresses of 45°C and 40°C, conditions commonly used in forced degradation studies of mAbs. By breaking down the large mAb aggregates into monomers, the SEC-PCD-MS method clearly revealed that the nonglycosylated variant was preferentially depleted from the monomeric states and was enriched within the large aggregates. This unique mechanism was absent under milder thermal stress conditions, indicating that a temperature near the melting point of nonglycosylated mAb molecules is necessary. By screening a diverse set of mAb molecules with varying properties and formulations, we demonstrated that this nonglycosylation-mediated aggregation pathway under thermal stress may be widely present in mAb molecules. In addition, the occurrence and extent of this aggregation pathway were significantly more pronounced at 45°C compared to 40°C. Finally, using spike-in samples, we demonstrated that the levels of nonglycosylation directly correlated with the formation of large aggregates under 45°C stress, which was mediated by nonglycosylated variant. This work not only highlighted the complexity of the degradation pathways for mAb molecules under thermal stress but also offered valuable insights for selecting appropriate temperature for performing forced degradation studies. While 45°C thermal stress is more likely to reveal this nonglycosylation-mediated aggregation, it may also obscure other degradation pathways that are more subtle. Furthermore, this finding also highlights important considerations when using high-temperature stability data (e.g., >40 °C) to predict low-temperature stability through modeling approaches, given that distinct degradation mechanisms may be involved. Finally, the precise mechanisms underlying this aggregation pathway, as well as its modulation by specific molecular features and formulation conditions, remain unexplored in the current study, highlighting the need for future research.

Funding Statement

This work was supported by the Regeneron Pharmaceuticals.

Abbreviations

mAb

monoclonal antibody

SEC

size exclusion chromatography

PCD

postcolumn denaturation

MS

mass spectrometry

PTM

post-translational modification

DSC

differential scanning calorimetry

HDX-MS

hydrogen-deuterium exchange mass spectrometry

UV

ultraviolet

MALS

multi-angle light scattering

ACN

acetonitrile

FA

formic acid

IPA

2-propanol

NSI

nanospray electrospray ionization

FDS

formulated drug substance

LMW

low molecular weight

HMW

high molecular weight

vHMW

very high molecular weight

XIC

extracted ion chromatogram

msAb

monospecific antibody

bsAb

bispecific antibody

NG

nonglycosylated

PG

partially glycosylated

FG

fully glycosylated

Disclosure statement

The authors declare the following competing financial interest(s): J.K.K., S.T., Y.Y., M. H., J. M., D. E. K., K. S. G., M. S., S.W., and N.L. are full-time employees and shareholders of Regeneron Pharmaceuticals Inc.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/19420862.2025.2543768