Selective clearance of monoclonal antibodies via the mannose receptor is dependent on glycan pairing

Abstract

The pharmacokinetics (PK) of therapeutic monoclonal antibodies (mAbs) are influenced by N-glycosylation, a critical quality attribute (CQA) that affects serum half-life and receptor interactions. High-mannose N-glycans are known to accelerate mAb clearance, likely via the mannose receptor (MR). However, the impact of high-mannose glycan pairing - whether symmetrical or asymmetrical - on this process remains poorly understood. MAbs enriched in high-mannose N-glycans were fractionated using mannose receptor—affinity chromatography to isolate symmetrical and asymmetrical high-mannose glyco-pairs. These fractions were characterized for physicochemical properties and labeled for a cell-based internalization assay using MR-expressing SUP-B15 cells to quantify internalization rates via flow cytometry. A PK study in rats was conducted using a high-mannose enriched mAb drug product, and glyco-pair-specific clearance was analyzed by mass spectrometry following immunocapture. Symmetrical high-mannose glyco-pair exhibited the highest internalization rate in vitro, followed by asymmetrical high-mannose and symmetrical complex glyco-pairs. In vivo, symmetrical high-mannose glyco-pair showed the fastest clearance, with a half-life of 2.4 days, compared to 7.2 days for asymmetrical high-mannose and 17.4 days for symmetrical complex glyco-pairs. The area under the curve was reduced to 73% and 38% for asymmetrical and symmetrical high-mannose glyco-pairs, respectively, relative to the symmetrical complex glyco-pair. These differences were attributed solely to glycan pairing, as other physicochemical properties remained consistent across fractions. Competitive inhibition with mannan confirmed MR-mediated uptake in vitro. The findings also suggest potential implications for immunogenicity, as increased internalization by antigen-presenting cells may enhance antigen presentation and anti-drug antibody formation. This study demonstrates that glycan pairing significantly influences the PK of mAbs by correlating with differential MR interactions, providing evidence for MR-mediated faster clearance of high-mannose containing mAbs. Symmetrical high-mannose glyco-pair are cleared more rapidly than asymmetrical counterparts, underscoring the need to consider glycan pairing as a distinct CQA. Current control strategies based solely on released N-glycans may overlook clinically relevant heterogeneity. Incorporating high-mannose glyco-pair analysis into manufacturing and quality control processes could improve therapeutic consistency and reduce immunogenicity risks.

Introduction

Monoclonal antibodies (mAbs) are an important and growing class of drugs with more than 200 marketed mAbs and more than 1000 in clinical development¹. The effectiveness of therapeutic mAbs is significantly determined by their pharmacokinetic (PK) profiles with half-life of up to three weeks². These profiles are influenced by a variety of factors. On the one hand, mechanisms such as target mediated drug disposition, anti-drug antibodies (ADAs), and off-target binding play a role in shaping the PK. On the other hand, the structure of mAbs, their overall surface charge distribution, and post-translational modifications can also affect PK³. MAbs are mainly distributed through convective transport, with their large size limiting elimination through the kidneys. Instead, mAb clearance primarily occurs through intracellular catabolism via lysosomal degradation⁴. After intracellular uptake by non-specific pinocytosis, the mAbs can be salvaged by the neonatal Fc receptor (FcRn) pathway extending their serum half-life⁵. However, interactions with other receptors or lectins, such as the mannose receptor (MR, CD206) are discussed to influence the PK. The MR binds mAbs dependent on their N-glycosylation^6,7. N-glycosylation is a complex and heterogeneous post-translational modification of therapeutic mAbs. The N-glycans of recombinant Immunoglobulin G (IgG) produced in chinese hamster ovary (CHO) cell culture are usually separated into two main types: the complex N-glycans, specifically biantennary fucosylated neutral glycans with varying proportions of terminal galactose (A2G0F, A2G1F, A2G2F) and the high-mannose structures (M5-M9). The sialic acid content and hybrid type N-glycans are negligible as well as larger tri- and tetra-antennary variants due to steric limitations of the IgG Fc part^8,9. Each heavy chain of the Fc part contains a single N-glycosylation site, leading to a pair of N-glycans in the intact IgG, referred to as a glyco-pair, that can be either asymmetrical (two different N-glycan types) or symmetrical (two identical N-glycan types)^10,11. To date there is limited knowledge available in literature about N-glycan pairing. It was demonstrated that the combination of glycosylated heavy chains is not the result of a statistical distribution⁶. In average in mAbs high-mannose N-glycans are 1–3%, with contents reaching 25–35%, influenced by factors such as the cell line, process conditions, and media composition^8,12,13,14. This influences also the occurrence of asymmetrical or symmetrical high-mannose glyco-pairs, which was previously shown¹⁵.

N-glycans of mAbs are typically analyzed after enzymatic release and fluorescence labeling, or as glycopeptides, without determining their pairing. Intact mass spectrometry (MS) analysis provides a partial overview, as it primarily identifies only the most prevalent glycoforms, and the information on specific glycan pairings is constrained by the presence of isobaric masses. Certain types of N-glycans can influence the PK, biological activity, safety and immunogenicity of mAbs^16,17 and are considered as critical quality attributes (CQAs) as defined in the ICH Q8(R2) guideline (International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use considerations (ICH) guideline Q8 (R2) on pharmaceutical development). Manufacturing parameters such as cultivation time, pH, temperature, or nutrient feeds during cell culture can greatly alter N-glycan composition¹⁸. This necessitates thorough control strategies during production of therapeutic mAbs and makes the understanding of their impact crucial.

High-mannose N-glycans are cleared faster in various organisms, with M6-M9 transforming enzymatically into M5^{6,19,20,21,22,23,24}. The M5 structures are then probably cleared by a MR-mediated internalization and lysosomal degradation mechanism^6,25. The MR is a key component in the immune system, functioning as an endocytic and phagocytic pattern recognition receptor and is primarily found on macrophages, dendritic cells, and certain endothelial cells^26,27. MR-mAb interactions were systematically investigated by surface plasmon resonance (SPR) and mannose receptor—affinity chromatography (MR-AC) in a previous work²⁸. These investigations, along with published glycan PK information, supports the hypothesis that MR is responsible for the increased clearance of mAbs carrying high-mannose N-glycans. It was observed that glycan pairing influenced the MR interaction. MAbs with symmetrical high-mannose glyco-pairs showed stronger binding to the MR than asymmetrical high-mannose glyco-pairs, potentially leading to different clearance rates²⁸. To date, only limited glycoform-resolved PK studies are available considering glycan pairing, or they are missing glycan pairing effects as the analysis of the serum samples is based on released N-glycans or on glyco-peptides^6,10.

This study shows the variations in the clearance rates of different high-mannose glyco-pairs, focusing on the influence of the avidity effect in MR interaction in an in vitro system and an in vivo study. For this, a cell-based internalization assay was developed using a MR-expressing cell line and fluorescently labeled mAbs with specific high-mannose glyco-pairs. Internalization rates into the cells were followed over time using fluorescence-activated cell sorting (FACS). Evident differences between the uptake rates of glyco-pairs were found and finally confirmed in a rat PK study using a mAb drug product with defined glyco-pairs and MS-based analytics to obtain individual PK profiles.

Results

Characterization of asymmetrical and symmetrical high-mannose glyco-pairs

Generation of mAb1 and mAb2 with enriched asymmetrical and symmetrical high-mannose glyco-pairs encompasses Concanavalin A (ConA) and MR-AC (Figure S3 A, B, overview on N-glycans and generation of differently glycosylated mAbs in Figure S1, Figure S2). As previously shown, the ConA product pool of the mAbs was enriched in high-mannose N-glycans, mainly M5, and the MR-AC separates the different high-mannose glyco-pairs into the flow-through and elution peak fractions²⁸. To determine the percentages of the symmetrical and asymmetrical high-mannose glyco-pairs in the MR-AC fractions of mAb1 and mAb2, the fractions were digested by Endo F3 followed by reverse phase chromatography (RPC) – MS analysis (Table 1) as previously described¹⁵. This enzyme cleaves all complex N-glycans with core fucosylation at the initial N-acetylglucosamine residue leaving the high-mannose N-glycans intact. Fractions one and two (F1/2) of both mAbs, contain primarily the symmetrical complex glyco-pair. Fraction three (F3) contains predominantly the asymmetrical high-mannose glyco-pair, while fraction four (F4) contains the symmetrical high-mannose glyco-pair. The mass spectra of these fractions after Endo F3 digestion are presented for mAb1 in Figure S4 and for mAb2 in Figure S5. As already described, there are no differences in the mAb-MR interactions depending on the mAb format or sequence²⁸. Therefore, for further detailed characterization, only the different glyco-pairs of mAb2 were used. The purity of the glyco-pair fractions of mAb2 was verified by reinjection to the MR-AC column (Figure S3 C). The reinjection of F1/2 showed a flow-through peak, whereas F3 eluted with an offset to F1/2 and a strong peak tailing indicating slight interaction of the asymmetrical high-mannose glyco-pair with the MR. The reinjection of F4 displayed one peak at the expected retention time indicating MR binding of the symmetrical high-mannose glyco-pair. The fractions of mAb2 were characterized with respect to their aggregates, fragments, charge heterogeneity, and FcRn affinity (Table 2, Figure S6). The distinct glyco-pairs differ hardly in any attribute. The high molecular weight species (HMWs) were slightly elevated in the elution peak (F4), which likely can be attributed to higher MR avidity due to the presence of aggregates consisting of multiple asymmetrical or symmetrical high-mannose glyco-pair Fc parts. Additionally, low molecular weight species (LMWs) were marginally increased in the flow-through peak F1/2, likely due to the presence of binding incapable fragments in the source material. Charge heterogeneity of the fractions is similar, with a decreasing acidic peak group (APG) from F1/2 to F3 and F4. Reducing capillary gel electrophoresis (CGE) revealed similar results for each fraction. The FcRn affinity chromatography showed comparable retention times, with slightly decreasing elution times from F1/2 to F3 and F4. As previously described, the symmetrical high-mannose glyco-pair (M5/M5) relating to F4 showed slightly lower FcRn binding, but this is unlikely to result in a significant difference in PK in vivo^29,30,31. To ensure that different glyco-pairs did not impact the C_H2 domain stability containing the N-glycans, the thermostability of all fractions was assessed, confirming comparable stability across fractions (Figure S7). Hence, it can be concluded that the fractionated variants (symmetrical complex glyco-pair (F1/2), asymmetrical (F3) and symmetrical (F4) high-mannose glyco-pairs) can be regarded as identical with differences only in their N-glycosylation.

Table 1 Distribution of the asymmetrical and symmetrical high-mannose glyco-pairs in the MR-AC fractions determined by RPC-MS after Endo F3 digestion.

Table 2 Characterization of the symmetrical complex glyco-pair and the asymmetrical and symmetrical high-mannose glyco-pairs of mAb2.

Cell-based internalization assay

Given the varying MR binding affinities of asymmetrical and symmetrical high-mannose glyco-pairs of mAbs, this may result in different internalization rates into MR expressing cells. To investigate whether the internalization rates of the high-mannose glyco-pairs differ from those of the symmetrical complex glyco-pair, a cell-based internalization assay was developed. The human B lymphoblast SUP-B15 expressing the MR cell line was selected based on the results from Streit et al.³². The selection criteria for the cell line included a high MR gene expression, no expression of the mAb’s target, and selective internalization by MR, which means no expression of other Fcγ receptors. The cell-line SUP-B15 was cultivated and evaluated for its MR expression level using cell surface staining and FACS. The SUP-B15 cell line demonstrated positive MR staining compared to the isotype control, confirming MR expression (Figure S8).

To compare the internalization rate into SUP-B15 cells of the asymmetrical and symmetrical high-mannose glyco-pairs, MR-AC fractions of mAb1 were used reflecting the symmetrical complex glyco-pair (F1/2), and the asymmetrical (F3) and symmetrical high-mannose glyco-pairs (F4) (Table 1). These fractions were labeled with Alexa Fluor™ 647 (AF647). The degree of labeling was verified using a Nano-Drop spectrophotometer and RPC-MS after deglycosylation confirming a comparable labeling of the fractions (~ 7–8 mol AF647 per mol mAb) (Figure S9). Absence of remaining free label was confirmed by size exclusion chromatography (SEC) with fluorescence detection (Figure S10).

SUP-B15 cells were separately incubated in eight replicates with the labeled fractions and without mAbs (negative control) at 37 °C to enable binding to the MR on the cell surface and resulting internalization. Cells were harvested at defined timepoints over a total period of 70 h. The cells were washed to remove supernatant containing residual free labeled mAbs and subsequently measured by FACS for their AF647 signal from cell surface-bound or internalized labeled mAbs at distinct time points (Fig. 1A). The negative control showed a constant baseline fluorescence throughout the 70 h period. The symmetrical complex glyco-pair (F1/2) showed a linear uptake over time, attributed to approximately equal portions to unspecific cellular uptake and to residual levels (29%) of asymmetrical high-mannose glyco-pair in the fraction (Table 1). Both high-mannose glyco-pairs (F3, F4) exhibited a linear increase in cellular fluorescence over time. The symmetrical high-mannose glyco-pair (F4) had the steepest slope with a rate 2.9-fold higher than the slope of the symmetrical complex glyco-pair (F1/2). The asymmetrical high-mannose glyco-pair (F3) exhibited a slightly steeper, yet significant slope (1.4-fold), than the symmetrical complex glyco-pair (F1/2).

Fig. 1

A: Internalization rates of the labeled glyco-pairs of mAb2 into MR-expressing cells. The AF647 labeled symmetrical complex glyco-pair (F1/2, green), the asymmetrical (F3, orange) and symmetrical (F4, blue) high-mannose glyco-pairs of mAb2 were incubated with SUP-B15 cells and SUP-B15 cells without mAbs as control (grey). The AF647 signal of the mAbs due to binding to the cells or internalization into the cells were measured by FACS over time. Data is presented as mean ± SD (n = 8) and fitted with a linear fit with R² > 0.94 to determine the slope. B: Inhibition of the internalization of the symmetrical high-mannose glyco-pair using a competitive cell-based internalization assay. A fixed concentration of the AF647-labeled symmetrical high-mannose glyco-pair (F4) of mAb1 was co-incubated with sequential dilutions of unlabeled mAb3 or unlabeled mannan for 46 h at 37 °C in SUP-B15 cells. After washing the cells, the AF647 signal was measured by FACS. Data is presented as mean ± SD (n = 3) and fitted with a four-parameter non-linear curve. Pointed lines show the upper and lower level of fluorescence. The lower limit is the fluorescence of the cells, and the upper limit is the fluorescence level of the cells with labeled F4 of mAb2, referring to the symmetrical high-mannose glyco-pair. Data is obtained from the mean (n = 6).

Two control experiments were designed to confirm the involvement of the MR in the uptake. First, to specifically inhibit the internalization of symmetrical high-mannose glyco-pair (F4), SUP-B15 cells were co-incubated with a fixed concentration of AF647-labeled symmetrical high-mannose glyco-pair (F4) in the presence of sequential dilutions of yeast mannan, a protein-free high affinity natural ligand of MR³³. Second, to demonstrate specific internalization via the MR, SUP-B15 cells were co-incubated with a fixed concentration of AF647-labeled symmetrical high-mannose glyco-pair (F4) and sequential dilutions of unlabeled complex glycosylated mAb3 lacking high-mannose N-glycans. The N-glycan profile of mAb3 was determined by glycan map showing trace levels of high-mannose N-glycans. The high-mannose glyco-pair distribution by RPC-MS after Endo F3 digestion was determined showing only symmetrical complex glyco-pairs (Figure S11). Controls included only cells and cells with the fixed concentration of AF647-labeled symmetrical high-mannose glyco-pair, which were used to measure the cell-dependent fluorescence level and the maximum fluorescence level resulting from the internalization of the symmetrical high-mannose glyco-pair (Fig. 1B). The AF467 signal was measured using FACS after 46 h of co-incubation. This time point was chosen based on the uptake experiment showing an ongoing internalization at 46 h. Mannan showed a concentration-dependent decrease in the fluorescence signal. MAb3 also exhibited a slight concentration-dependent decrease, starting at elevated concentrations. These findings suggest that mannan specifically inhibits internalization via MR, while mAb3 shows unspecific competition with the labeled symmetrical high-mannose glyco-pair (F4) at relatively high concentrations which agrees with the previously observed partially unspecific uptake of the symmetrical complex glyco-pair fraction (F1/2). As a result, the dose-response curves indicated that mannan has an IC50 value (0.24 mg/mL) that is 18-fold lower compared to the value of mAb3 (4.44 mg/mL).

High-mannose glyco-pair resolved PK study in rats

To further confirm the differential impact of asymmetrical and symmetrical high-mannose glyco-pairs of mAbs in vivo, a rat PK study with defined mAb2 material was designed (Fig. 2). It included the generation of the mAb2 material with similar levels of symmetrical and asymmetrical high-mannose glyco-pairs and symmetrical complex glyco-pair, a single intravenous (IV) administration of the high-mannose enriched mAb2 drug product to rats, and subsequent analysis of the individual glycan variants by MS. The enrichment of the high-mannose content was accomplished by using a preparative ConA affinity chromatography. The ConA product pool was buffer exchanged, formulated, and sterile filtrated. The product quality of the resulting high-mannose enriched mAb2 drug product was assessed (Table S1 and Figure S12). The mAb2 drug product contained a total level of 49.1% high-mannose N-glycans. Glyco-pairs consisted of 26% asymmetrical high-mannose glyco-pair and 42% symmetrical high-mannose glyco-pair, as well as 32% symmetrical complex glyco-pair. The purity by SEC and non-reducing CGE was > 95.0% monomer.

Fig. 2

High-mannose glyco-pair resolved PK study overview. The high-mannose content of mAb2 was enriched by ConA chromatography. The mAb2 drug product was prepared for the single IV injection to rats. Serum was taken at different time points. After immunocapture purification, mAb2 samples were analyzed for the total mAb concentration by LC-MS/MS after trypsin digestion, and for the glyco-pair distribution by RPC-MS after Endo F3 digestion.

To ensure mAb2 drug product’s stability throughout the PK study, a stability study was conducted. The mAb2 drug product was stored at 5 °C and 25 °C, with samples taken after one and two weeks to cover the duration of the PK study. These samples were analyzed and compared to the initial material (Table S2). All analyzed parameters were constant at intended storage (5 °C) and only slight increases of some parameters were observed at accelerated storage (25 °C).

A single intravenous 100 mg/kg bolus dose of the mAb2 drug product was administered to six male rats. Serum samples were taken at eight time points between 2 h and 336 h post-administration and subjected to immunocapture purification prior to liquid chromatography (LC) -MS analysis of the total mAb2 concentration and glyco-pair distribution. The total mAb2 concentration was determined using a LC-MS/MS assay based on a surrogate peptide from the Fc after trypsin digestion (Fig. 3A). Data from one rat was excluded for further analysis due to an abnormal PK profile (Figure S13). Glyco-pair analysis of mAb2 was performed by RPC-MS after Endo F3 digestion to determine the distribution of the glyco-pairs at each time point. The glyco-pair distribution of the first sampling point (2 h post administration) corresponded to the starting material as expected. By combining the total mAb concentration and the ratios of the high-mannose glyco-pairs at each time point, the individual PK profiles and PK parameters of the glyco-pairs could be calculated (Figure S14). To enable direct comparison of the PK profiles, despite the uneven distribution of glyco-pairs in the dose at the time of administration, their concentrations at each time point were calculated relative to the concentration at 2 h, the initial sampling time point (Fig. 3B). Over the 336 h period, the concentration of each glyco-pair decreased, with the high-mannose glyco-pairs decreasing much faster than the symmetrical complex glyco-pair. The symmetrical high-mannose glyco-pair also decreased faster than the asymmetrical high-mannose glyco-pair. The symmetrical high-mannose glyco-pair was almost completely cleared after 336 h. This effect was reflected in the PK parameters half-life (t_1/2), dose normalized area under the curve (AUC), and clearance (CL) (Fig. 3C). The t_1/2 was 17.4, 7.2, and 2.4 days for the symmetrical complex, asymmetrical and symmetrical high-mannose glyco-pair, respectively. This represented a 2.4- and 7.3-fold reduction in the t _1/2 of the high-mannose glyco-pairs compared to the symmetrical complex glyco-pair, respectively. Correspondingly, exposure (AUC) was reduced to 73% for the asymmetrical high-mannose glyco-pair and 38% for the symmetrical high-mannose glyco-pair in comparison to the complex glycosylated mAb variant. Similarly, the clearance increased with the presence of high-mannose N-glycans. A single high-mannose N-glycan at the Fc increased clearance approximately 2-fold, while two high-mannose N-glycans corresponded to an approximate 4-fold increase. These in vivo findings are well in agreement with the in vitro cell-based internalization data.

Fig. 3

PK profiles and parameters of the high-mannose glyco-pair resolved PK study in rats. A: Total mAb2 concentration over time presented as GeoMean ± SD of five rats measured by peptide mapping. B: Relative concentration of the symmetrical complex glyco-pair (green dots), the asymmetrical high-mannose glyco-pair (orange squares) and the symmetrical high-mannose glyco-pair (blue triangles) of mAb2 over time. Concentration was calculated relative to the highest concentration (2 h) after administration with CV% as error bars. C: PK parameters of each individual glyco-pair obtained from PK profiles. The t_1/2, the AUC and the CL was determined from the average of five rats. AUC was dose normalized according to the actual doses of each glyco-pair based on the levels in the drug product to enable direct comparison assuming linear PK.

To exclude a potential impact from different complex N-glycan pairing partners of the asymmetrical high-mannose glyco-pair on the PK, an investigation was conducted on the mAb2 ConA product pool. The corresponding MR-AC fractions were subjected to Endo H digestion to specifically remove all high-mannose N-glycans as previously described¹⁵. The RPC-MS analysis demonstrated that the mAb2 asymmetrical high-mannose glyco-pair, primarily found in the flow-through peak tailing (F3), were mainly paired with truncated monoantennary complex N-glycans (Figure S15). Both the undigested and the Endo H digested asymmetrical high-mannose glyco-pairs were re-injected onto the MR column (Figure S16), each showing a single flow-through peak. The peak tailing, determined by the peak asymmetry at 30% peak height, observed in the flow-through peak was indicative of a slight interaction with the MR as previously shown²⁸. The peak asymmetry of the asymmetrical high-mannose glyco-pair was 3.22 indicating weak interaction with the immobilized MR, while the asymmetry reduced to 1.74 for the Endo H digested asymmetrical high-mannose glyco-pair suggesting no interaction with the immobilized MR, similar to the symmetrical complex glyco-pair peak asymmetry. This indicated that the MR interaction depended on the presence of the high-mannose N-glycan of the asymmetrical glyco-pair. Therefore, the truncated complex N-glycan pairing partner did not influence the MR interaction or the PK profile of the asymmetrical high-mannose glyco-pair.

Discussion

An avidity effect on MR binding of symmetrical high-mannose glyco-pair was previously demonstrated, while only a slight interaction was observed with the asymmetrical high-mannose glyco-pair using MR-AC and SPR²⁸. The high-mannose N-glycan specificity of IgGs to the MR was published previously²⁸. To investigate whether the varying MR binding properties of the different high-mannose glyco-pairs of mAbs correspond to different MR-mediated internalization rates into cells, an internalization assay with MR-expressing cells and labeled high-mannose glyco-pairs has been established, as well as a high-mannose glyco-pair resolved PK study has been performed to confirm the results in vivo.

All glyco-pairs exhibited internalization, as shown by the increased cellular fluorescence over time. The internalization rates of both the asymmetrical and symmetrical high-mannose glyco-pair were higher than that for the symmetrical complex glyco-pair, likely due to their higher MR binding affinities. The avidity MR binding effect of the symmetrical high-mannose glyco-pair resulted in even faster internalization. The internalization of the symmetrical complex glyco-pair (F1/2) could be attributed to a combination of the residual 29% asymmetrical high-mannose glyco-pair content in fraction F1/2 and a different less specific endocytic pathway such as pinocytosis. MR-mediated internalization was inhibited by mannan which was shown with increasing mannan concentrations preventing the internalization of the symmetrical high-mannose glyco-pair, underscoring the role of MR in the internalization process. At elevated concentrations the MR specific internalization was also inhibited with a complex glycosylated mAb.

The PK study revealed different clearance rates of the asymmetrical and symmetrical high-mannose glyco-pairs of mAbs, with both being faster compared to the symmetrical complex glyco-pair. The symmetrical high-mannose glyco-pair had the shortest half-life, followed by the asymmetrical high-mannose glyco-pair, while the symmetrical complex glyco-pair displaying a typical IgG half-life². The PK parameters such as AUCs of the different glyco-pairings determined in this study aligned well with a previous study by Yu et al., who investigated different glycosylated IgGs individually in mouse PK studies³⁴. Their fully high-mannose glycosylated mAb corresponding to the symmetrical glyco-pair had approximately 36% exposure of the complex glycosylated mAb, which was consistent with the 38% exposure of the symmetrical high-mannose glyco-pair compared to the complex glycosylated mAb variant determined in this study. The findings further correlated with the MR binding data and the cell-based internalization results. It was shown that, compared with the symmetrical complex glyco-pair, the asymmetrical and symmetrical high-mannose glyco-pairs differed only in their N-glycans and not in other characteristics, and this confirmed that the faster internalization and clearance could be only contributed to the N-glycans. Additionally, the different clearance rates of the asymmetrical and symmetrical high-mannose glyco-pairs explained a biphasic PK profile of M5 N-glycans and an incomplete elimination of M5 N-glycans as previously reported^19,22. The symmetrical high-mannose glyco-pair was nearly completely cleared by 336 h due to MR interaction, while the asymmetrical high-mannose glyco-pair, despite weaker MR interaction, also cleared faster than the symmetrical complex glyco-pair. This contradicts the previous assumption that the slight MR interaction of the asymmetrical high-mannose glyco-pair, as indicated by the flow-through peak tailing in MR-AC, would not result in faster clearance. However, attempts to improve peak tailing separation by adjusting the column dimensions through coupling two columns did not enhance resolution. A previously reported PK study by Goetze et al.⁶ theorized through calculations that a single M5 N-glycan per mAb was sufficient to increase clearance mediated by the MR, but was unable to calculate individual PK properties of the different high-mannose glyco-pairs. Another PK study reported by Liu and Flynn³⁵showed that clearance rates of mAbs containing high-mannose were comparable, regardless of asymmetrical or symmetrical pairing, by using a similar set up consisting of enzyme digestion to achieve the three main glyco-pairs and LC-MS analysis. However, the samples were digested to Fc parts under reducing conditions, which could cause heavy chain shuffling and shifting the distribution and therefore falsifying the results towards a random distribution at each time point.

The asymmetrical high-mannose glyco-pair, predominantly paired with truncated monoantennary complex N-glycans, as previously reported^15,28, exhibited minor MR interaction but corresponded to an accelerated clearance. This could explain the faster clearance of the truncated monoantennary N-glycans observed in a recent study by Falck et al.^20,21. After removing the high-mannose N-glycan of the asymmetrical high-mannose glyco-pair, reinjection to the MR-AC eliminated the flow-through peak asymmetry, indicating that the faster clearance was likely attributable to the weak MR interaction of the single high-mannose N-glycan, rather than the affinity of core mannoses or the N-acetylglucosamine for the MR. Given that the MR used in the presented study contained all extracellular domains, a contribution of the MR can be excluded. However, another lectin specific for truncated complex N-glycans might also mediate internalization and faster clearance.

Regarding the previously published characterization study of the MR-IgG interactions matching the results of this study’s cell-based internalization assay and glyco-pair resolved PK study, it is likely that the MR contributes to the faster clearance of high-mannose N-glycan containing mAbs. However, it cannot be excluded that other clearance mechanism could be involved e.g. other receptors and lectins with IgG binding and internalization ability.

The involvement of the MR in the uptake and increased clearance of high-mannose containing therapeutic proteins may further influence the immune response against mAbs. As a highly effective endocytic receptor, expressed on a variety of antigen presenting cells (e.g. dendritic cells), the MR is involved in the antigen presentation through major histocompatibility complexes (MHCI and MHCII)^26,36, and is further capable of sorting antigens during endocytosis into specific endosomal compartments for efficient cross-presentation³⁷. This increased internalization into dendritic cells could catalyze the formation of ADAs against therapeutic proteins with T-cell epitopes carrying high-mannose N-glycans. Consequently, the presence of high-mannose N-glycans on therapeutic proteins may become even more critical by impairing efficacy through increased clearance and ADA formation or due to higher risk of adverse events such as hypersensitivity reactions. The correlation of high-mannose N-glycans with antigen presentation and T-cell responses has also been reported in literature^38,39. Glycan pairing is also highly relevant in this immunogenicity aspect as symmetrical high-mannose glyco-pairs might facilitate antigen-presentation more than asymmetrical high-mannose glyco-pairs.

Conclusions

It has been demonstrated that the PK of mAbs are significantly influenced by the glycosylation profile, particularly the presence of high-mannose N-glycans. Through the administration of a single intravenous dose to male rats, it was observed that the symmetrical and asymmetrical high-mannose glyco-pairs exhibited faster clearance rates compared to the symmetrical complex glyco-pair. This was attributed to the high binding affinity of high-mannose glyco-pairs to the MR, which facilitated their internalization and subsequent degradation.

Considerable variations were observed in the PK parameters, including half-life and AUC, between the glyco-pairs. The shortest half-life and the highest clearance were shown by the symmetrical high-mannose glyco-pair, followed by the asymmetrical high-mannose glyco-pair and the symmetrical complex glyco-pairs. This gradient in clearance rates closely aligns with cell-based in vitro data on MR binding and internalization, confirming the role of MR in the differential clearance observed in vivo.

The negative impact of high-mannose glyco-pairs on PK highlights the importance of considering these glyco-pairs as CQAs, especially the symmetrical one. The variability in the presence and ratio of asymmetrical and symmetrical high-mannose glyco-pairs can arise from the manufacturing process. The implications of these findings are far-reaching for the development, manufacturing and control strategies of mAbs and other therapeutic glycoproteins. To avoid rapid clearance and potential immunogenicity issues, the presence of high-mannose glyco-pairs should be minimized. The necessity of incorporating glycan pairing analysis into the control strategy of N-glycans in therapeutic mAb production is highlighted by the study.

The understanding of glycan-mediated variations in mAb PK has been solidified by this comprehensive analysis, paving the way for more refined and effective therapeutic strategies.

Methods

Monoclonal antibodies (mAb)

MAb1 and mAb3 formatted as IgG1, and mAb2, a IgG1 - single chain variable fragment (scFv) meaning two scFvs were fused to the C_H3 domain of each heavy chain of an IgG1, were produced in-house by mammalian cell culture technology using a CHO cell line (American Type Culture Collection (ATCC)) in a fed-batch process and afterwards purified and formulated.

Preparation of mAbs with enriched high-mannose content by ConA chromatography

ConA chromatography was used to enrich the mAbs in high-mannose content leading to an increased proportion of asymmetrical and symmetrical high-mannose glyco-pairs ConA chromatography was used. The previously described method²⁸based on the manufacturers protocol was used for mAb1. The protocol was scaled-up to produce > 300 mg material of high-mannose enriched mAb2. For this, 200 mL of ConA Sepharose™ 4B Resin (Cytiva) were washed with Buffer A (20 mM Tris-HCl, 0.5 M NaCl, 1 mM MnCl₂, 1 mM CaCl₂, pH 7.4) and packed into a HiScale column (50/400, Cytiva). The resulting ConA column (186 mL, 9.5 cm height) was used on an Äkta™ Pure25 system (Cytiva). Data acquisition and instrument control were performed by Unicorn (version 7.5., Cytiva). The column was loaded with 800 mg mAb1 and eluted by a step gradient with 0.2 M methyl-α-D-glucopyranoside. For binding a flow-rate of 1 mL/min, for elution a flow-rate of 10 mL/min and for (re-) equilibration a flow-rate of 20 mL/min was used. The ConA product pool of different runs was collected and pooled. Ultra filtration and diafiltration (UF/DF, 30 kDa) was performed for concentration and buffer exchange to 0.002% NaCl. The material was formulated to 220 mM sucrose, 0.2 g/L PS20, pH 5.3–5.4 to a concentration of 32.8 mg/mL determined by SoloVPE (Repligen). By sterile filtration (0.22 µM PVDF filter) the material was transferred into sterile 15 mL falcons. The final material was analyzed by Glycan map, Endo F3 digestion followed by RPC-MS, SEC and non-reducing CGE.

Purification of glyco-pairs by mannose receptor – affinity chromatography (MR-AC)

Cloning, expression and purification of human MR and the following MR affinity column preparation was previously described²⁸. Two ~ 1 mL MR affinity columns (0.5 x ~ 5.0 cm) were coupled to increase resolution and load. The coupled columns were operated on a Äkta™ go system (Cytiva). Unicorn software (version 7.5., Cytiva) handled data collection and device control. After column equilibration with mobile phase A (12 mM Tris, 140 mM NaCl, 2.5 mM CaCl₂, pH 7.4) the sample (10–200 µg) was injected. This was followed by a 12 min gradient from 0 to 100% mobile phase B (12 mM Tris, 140 mM NaCl, 20 mM EDTA, pH 7.4) and 5 min at 100% mobile phase B. After each run, column re-equilibration was conducted with mobile phase A. A consistent flow rate of 0.50 mL/min and detection via UV at 280 nm was used. Fraction collection was performed with the Cytiva Fraction Collector F9-R. To gather enough protein, fractions from various chromatographic runs were collected, combined, concentrated, and formulated. The column was stored at 2–8 °C in 10 mM acetate, 140 mM NaCl, 0.01 mg/mL PS20, 0.05% ProClin3000, cOmplete (1 tablet/50 mL), pH 5.0.

Stability study

A volume of 1 mL mAb2 ConA product pool was filled into 1.5 mL reaction tubes (Eppendorf) and stored under defined conditions (5 °C, 25 °C) for 7 and 14 days. The samples were analyzed by pH, SEC, reducing CGE, CEX, FcRn affinity chromatography and Endo F3 digestion followed by RPC-MS.

Analytical methods

Glycan map: determination of 2-AA labeled N-glycans composition by RPC

A 50 µg mAb sample was buffer exchanged to 1X PBS, pH 7.4. By addition of 1.5 µL Rapid PNGase F (non-reducing format, New England Biolabs) and incubation for 30 min at 50 °C the N-glycans were released of the mAbs. Next, 40 µL of labeling solution (0.73 M 2-AA (2-anthranilic acid), 0.75 M 2-methypyridine borane complex in 85:15 ethanol: acid mix) were added, and the N-glycans were labeled over 2 h at 65 °C. The reaction mixture was then removed by evaporating the solution to dryness using a vacuum concentrator. The N-glycans were re-dissolved in 40 µL water. For purification, 35 µL were diluted with 400 µL acetonitrile and loaded onto pre-conditioned PhyTip^® HILIC columns. After washing three times with 96% (v/v) acetonitrile containing 2% (v/v) formic acid, the N-glycans were eluted with 200 µL water. To eliminate any leftover acetonitrile, the purified N-glycan solution was heated for additional 30 min at 50 °C. The N-glycans were analyzed using RPC, as outlined by Wilhelm et al.⁴⁰, on a Zorbax RRHD Bonus-RP column (2.1 × 150 mm, 1.8 μm, 80 Å, Agilent) with a 1290 Infinity II UHPLC system (Agilent). The flow rate was set to 0.25 mL/min and the column temperature of 70 °C was used. The mobile phases were composed of water with 0.5% acetic acid and 0.5% formic acid (A), and 20% acetonitrile, 5% 1-propanol, 5% 1-butanol, 0.5% acetic acid, and 0.5% formic acid (B). A linear gradient was applied as follows: 0.00:2, 33.00:35, 48.00:95, 53.00:95, 54.00:2, 59.00:2 (time in min: percentage of mobile phase B). Fluorescence detection of 2-AA-labeled N-glycans was performed at Ex 350 nm / Em 440 nm. Evaluation was based on an inhouse database of N-glycans. The N-gylcan idenficiation is based on LC-MS and exoglycosidases as well as the retention time order.

Characterization of glyco-pairs by RPC-MS

For 10 µg mAb, 1 µL (8 U) Endo F3 (New England Biolabs) was utilized in 1X GlycoBuffer 4 to a final volume of 20 µL, while 5 µL (25 000 U) Endo H (New England Biolabs) was used in 1X GlycoBuffer 3 to the same final volume. Both Endo F3 and Endo H digestions were incubated overnight at 37 °C and monitored by RPC-MS as previously published²⁸. The system used was a BioAccord LC-MS system (Waters), which included an ACQUITY UPLC I-Class PLUS system and an ACQUITY RDa detector based on electrospray ionization – time of flight (ESI-TOF), controlled by UNIFI Scientific Information Software (Waters). The MS data were collected in positive ion mode. RPC was executed on a BioResolve Reverse Phase mAb Polyphenyl column (450 Å, 2.7 µM, 2.1 × 50 mm, Waters) and at a flow rate of 0.25 mL/min with 80 °C column temperature. Mobile phase A consisted of 0.1% formic acid in water, while mobile phase B contained 0.09% formic acid dissolved in 2-propanol. After sample injection (2 µg) the following gradient was applied: 0.00:15, 1.00:15, 9.00:35, 9.20:95, 10.20:95, 10.40:15, 13.00:15 (time in min: percentage of mobile phase B). UV data were monitored at 214 nm and mass spectrometry data were collected in full scan mode within a mass range of 400–7000 m/z at positive polarity, using a 2 Hz scan rate, 140 V cone voltage, and standard lockmass correction mode. The capillary voltage was set to 1.5 kV, and the desolvation temperature was 550 °C. Data analysis was performed using the UNIFI Scientific Information System, employing the MaxEnt1 deconvolution algorithm.

Size exclusion chromatography (SEC)

SEC was performed using a Acquity UPLC BEH200 (4.6 × 300 mm, 1.7 μm, Waters) column on a 1290 Infinity II UHPLC system (Agilent). Analysis was carried out in an isocratic mode with one mobile phase (40 mM sodium phosphate, 400 mM sodium perchlorate, pH 6.8) at room temperature (RT). The mAb samples were diluted to a concentration of 1–5 mg/mL in mobile phase and 15–30 µg were injected. The runtime of 16 min was conducted by 0.3 mL/min. Elution was controlled at a wavelength of 280 nm.

Non-reducing and reducing capillary gel electrophoresis (CGE)

MAb samples were prepared to 0.5 mg/mL in 200 µL with matrix. For non-reducing CGE the matrix was: SDS sample buffer 50% (v/v, SCIEX), 5% of 0.25 mol/L iodoacetamide in water, and for reducing CGE: low pH SDS sample buffer 50% (v/v, SCIEX), 5% of DTE 1 mol/L in water. After heating for 10 min at 65 °C in a water bath (Julabo Pura 10), the samples were cooled down for 5 min at RT.

The CGE instrument Beckman Coulter ProteomeLab PA800 Plus equipped with a UV detector (SCIEX) was used. The peaks were evaluated by migration time. The relative percentage of the area under the curve with migration time correction (velocity corrected area, VCA) was determined for quantification. The capillary cartridge (bare fused silica, 50 μm ID, length 30.5 cm ± 0.2 cm, SCIEX) was prepared according to the instrumentation guideline from the manufacturer. Method parameters were set as follows: cartridge temperature 25 °C, buffer and sample tray temperature 15 °C. Separation was monitored by UV detection at 214 nm. One analysis cycle had the following steps: Basic wash for 3 min at 70 psi, acidic wash for 1 min at 70 psi, wash with water for 1 min at 70 psi, separation gel load for 10 min at 70 psi, followed by sample application with + 5.0 kV for 15 s and application of the separation voltage for non-reduced was + 15 kV (20 psi) for 20 min and for reduced was + 17 kV (20 psi) for 16 min.

Ion exchange chromatography (IEC)

IEC was performed on a 1260 Infinity II HPLC system (Agilent) with a BioPro IEX SF, S-5 μm, 100 × 4.6 mm (YMC) column. The following mobile phases were used: 20 mM MOPS, pH 7.1 (A) and 20 mM MOPS, 500 mM KCl, pH 7.1 (B). MAb samples were diluted to 1 mg/mL in water, and 15 µg were loaded. The 47 min run was performed with 0.8 mL/min at 35 °C column temperature and monitored by 280 nm. The following gradient was applied 0.00:8, 3.00:8, 35.00:21, 36.00:100, 41.00:100, 42.00:8, 47:8 (time in min: percentage of mobile phase B).

FcRn affinity chromatography

An Agilent 1260 Infinity II HPLC system were used with peek capillaries of 0.25 mm inner diameter with direct flow from the autosampler via the column to the detector. FcRn affinity column Gen2 (1 mL, Roche Diagnostics) was used and a pH gradient with mobile phase A (20 mM Bis-Tris, 140 mM NaCl, pH 5.8) to mobile phase B (20 mM Tris, 140 mM NaCl, pH 8.8) was set as follows 0.00:0; 1.00:0; 1.01:25; 31.00:100; 35.00:100; 35.01:0; 45.00:0 (time in min: percentage of mobile phase B). A flow-rate of 0.5 mL/min and 30 °C column temperature were applied. 50 µg mAb sample at 1 mg/mL was loaded and the elution was monitored by UV at 280 nm.

Nano differential scanning fluorimetry (nanoDSF)

Mab samples were analyzed in triplicates using a Prometheus Panta (Nano Temper) to monitor mAb unfolding by intrinsic fluorescence. Temperature was ramped from 20 to 90 °C with 1 °C/min. The ratio of fluorescence intensities at 350 nm to 330 nm was calculated and the melting temperature (T_m) of the mAb was determined by the maximum of the first derivative.

Cultivation of SUP-B15 cells and determination of receptor expression

The SUP-B15 cells (CRL-1929, ATCC) were cultivated in Iscove’s Modified Dulbecco’s Medium (IMDM, Gibco) + heat-inactivated 20% fetal calf serum (FCS, HyClone) and were sub-cultured every 3 to 4 days in fresh culture medium in T25/T75/T175 flasks. For this, the cells were centrifuged at 200 x g for 5 min at RT and the pellet was resuspended into culture medium. After counting by NucleoCounter^® (Chemometec) SUP-B15 cells were seeded to a concentration of 0.4 × 10⁶ cells/mL for three days and to 0.3 × 10⁶ cells/mL for four days. Cells were incubated at 37 °C and 5% CO₂ until next sub-culturing.

To determine the expression of MR in SUP-B15 cells, the cells were stained with an anti-MR mAb. 0.5 × 10⁶ SUP-B15 cells were transferred into 50 µL FACS buffer (1X PBS, pH 7.4 with 0.1% bovine serum albumin (BSA)) by centrifugation at 200 x g for 10 min at 4 °C. Afterwards 10 µL of Human Fc-Block reagent (BD Biosciences) were added and incubated for 5 min at RT. 3 µg of the allophycocyanin (APC) labeled anti-MR (CD206, clone 15 − 2) mAb (BioLegend) or the APC labeled isotype control (Mouse IgG1, k, clone MOPC-21, BD Biosciences) or the corresponding volume of FACS buffer were added and incubated for 30 min at 4 °C light protected. The cell suspensions were centrifuged at 200 x g for 10 min at 4 °C again for washing with FACS buffer before transferring into the measurement plate (96-well microplate, PP, U-bottom, Greiner Bio-One) for FACS analysis.

Labeling of mAbs

The Alexa Fluor™ 647 Microscale Protein Labeling Kit (Invitrogen) was used to label 100 µg mAb. The mAb was diluted to 1 mg/mL, 1/10 volume of 1 M sodium bicarbonate and 1.35 µL of the dye solution (7.94 nmol/µL stock solution) were added. After 15 min of incubation at RT the mAb was purified using the 0.5 mL Zeba™ Dye and Biotin Removal Spin Columns (ThermoFisher Scientific). The degree of labelling was determined by NanoDrop as described in the kit and controlled by RPC-MS after deglycosylation. For deglycosylation Rapid PNGase F (New England Biolabs) was used and the RPC-MS method mentioned above. The unbound Alexa Fluor™ 647 label were determined by SEC using the fluorescence detector (excitation 650 nm, emission 671 nm).

Cell-based internalization assay

Culture-derived SUP-B15 cells were transferred into the assay medium (IMDM + 0.1% BSA (Thermo Fisher Scientific)) by centrifugation at 200 x g for 5 min. For tracking the internalization over time, the cells were diluted to 1 × 10⁶ cells/mL and the Alexa Flour™ 647 labeled mAbs to 1 µg/mL with assay medium. The labeled mAbs were separately incubated with the cells in 96 well microplates for suspension cells (PS, Cellstar, U-bottom, Greiner Bio-One). For this, 100 µL of the cells (100,000 cells/well) were used with 50 µL labeled mAb (0.33 µg/mL per well). Incubation was light-protected at 37 °C and 5% CO₂ for different time points (0–70 h). For the competitive internalization, a fixed concentration of the labeled mAb was co-incubated with a sequential dilution of a competitive molecule (mannan (Sigma-Aldrich) or mAb3). The cells were diluted to 2 × 10⁶ cells/mL and the labeled mAb was diluted to 1 µg/mL with assay medium. The competitive molecule was diluted to 30 mg/mL and 1:3 sequential diluted (12 steps) with assay medium. For the co-incubation, 50 µL of the cells (100,000 cells/well), 50 µL of labeled mAb (0.33 µg/mL per well) and 50 µL of the dilution series of the competitive molecule (starting 10 mg/mL per well) were mixed in 96-well microplates for suspension cells (PS, Cellstar, U-bottom, Greiner Bio-One). Incubation was at 37 °C and 5% CO₂ for 46 h. After incubation, the assay medium was removed by centrifugation at 500 x g for 5 min and the cells were transferred into 150 µL ice cold FACS buffer (1X PBS, pH 7.4 + 0.1% BSA). This washing step was repeated once concentrating the cells into 100 µL FACS buffer and transferring the cells into the measurement plate (96-well microplate, PP, U-bottom, Greiner Bio-One) for FACS analysis.

FACS analysis

FACS analysis was performed by event measurements using the iQue Screener PLUS (Sartorius AG) and the corresponding software (iQue Forecyt^® Enterprise Server Edition 10.0 R2). The settings were a flow rate of 2 µL/sec, a sample amount of 20 µL, pre- and intermediate plate shakes for 5 s at 1000 rpm. The relative fluorescence signal (RFU) of the Alexa Fluor™ 647 or APC labeled mAbs were measured with the red laser via fluorescence activated cell scanning. A dot plot for all detected events (FSC-H vs. SSC-H; logarithmic scale) was used to separate viable cells from debris, air bubbles, unattached fluorophores, and possibly dead cells from viable cells. A further dot plot of the viable cells was created (FSC-H vs. FSC-A; logarithmic scale), gating only the single viable cells which are located near origin straight line. The detected emission values (RL-1-H) of the single viable cells was determined for each sample / well.

Ethics statement

The animal experiments were performed at Boehringer Ingelheim Pharmaceuticals, Ridgefield, CT, USA, in accordance with the Animal Welfare Act, Association for Assessment and Accreditation of Laboratory Animal Care International Guide (AALAC) and other applicable federal, state, and local laws or regulations, and the study was approved by the Institutional Animal Care and Use Committee (IACUC). 3R principles were applied and the study was reported in agreement with Animal Research: reporting of In Vivo Experiments (ARRIVE) guidelines.

Glyco-pair resolved rat PK study

Six healthy male rats (rat strain: Wistar Han obtained from Charles River Laboratories, Inc., Raleigh, NC) were administrated a single intravenous dose of 100 mg/kg of the mAb2 drug product enriched in high-mannose content. No exclusion criteria were defined, and no randomization or blinding was done. Blood samples were collected from each animal at pre-dose, 2, 8, 24, 48, 72, 96, 168 and 336 h post dose and processed for plasma into tubes containing K₃EDTA. Plasma samples were stored at -80 °C, protected from light. The study was a non-terminal study and all animals survived at the end of the study.

These samples were used for bioanalysis, including plasma immunocapture purification, LC-MS/MS quantitation of the total mAb concentration and RPC-MS glyco-pairing distribution determination. The lower level of quantification (LLOQ) was 10 µg/mL. The glyco-pair distribution was determined by an immunocapture, Endo F3 digestion and LC-MS. The glyco-pair distribution was determined by the deconvolution.

Total drug analysis by surrogate peptide LC-MS/MS

Plasma samples were analyzed for the total mAb2 drug product concentration using the immunocapture, trypsin digestion and LC-MS/MS method. Peptide VVSVLTVLHQDWLNGK was used as the surrogate for the quantification of mAb2. Peptide VVSVLTVLHQDWLNG [Lys(¹³C6,¹⁵N2)] was used as the internal standard for the LC-MS/MS analysis.

For the 2 h to 168 h plasma samples 5 µL were used and for the 336 h plasma sample 50 µL were used with 20 µL of 0.15 mg/mL biotinylated goat anti-human IgG (Jackson immuno research), 495 or 450 µL of Tris-buffered saline with 0.1% Tween-20 (TBS-T), and 25 µL of freshly prepared 10 mg/mL magnetic beads solution Streptavidin MagneSphere^® Paramagnetic Particles, Promega) were added to a 96-deepwell plate (PP, Thermo Fisher Scientific) and incubated for 2 h at RT. After magnetic separation of the beads, the samples were washed three times with 300 µL TBS-T and twice with 300 µL water, then eluted with 75 µL 25 mM HCl on a Kingfisher Flex magnetic bead handler. The eluent was immediately neutralized with 10 µL of 1 M Tris-HCl (pH 8.0).

To each immunocaptured sample 10 µL of 0.1% RapiGest (Waters) in 100 mM ammonium bicarbonate and 10 µL of 100 mM TCEP in 100 mM ammonium bicarbonate were added and incubated for 1 h at 60 °C. For further 30 min incubation at RT in the dark 10 µL of 200 mM iodoacetamide in 100 mM ammonium bicarbonate were added. Then, 5 µL of 100 ng/mL surrogate peptide in water/acetonitrile (50/50 v/v) and 5 µL of 5 µg/µL trypsin (sequencing grade, Promega) in 50 mM acetic acid were used for each sample. Following overnight digestion at 37 °C with gentle shaking, 10 µL of 10% formic acid in water was added and incubated at 37 °C for further 30 min. Centrifugation was performed before LC-MS/MS analysis. Samples were analyzed using a Water BEH C18 reversed phase column (300 μm x 50 mm, 300 Å, 1.7 μm) with a Waters Acquity UPLC M-Class system coupled to a Sciex 6500 + triple quadrupole mass spectrometer. Mobile phases were 0.1% formic acid in water (A) or in acetonitrile (B) running at a flow rate of 10 µL/min at 50 °C column temperature. After sample injection (10 µL) the following gradient was applied: 0.00:30, 1.00:30, 7.00:95, 8.50:95, 8.51:30, 10.00:30 (time in min: percentage of mobile phase B).

The mass spectrometry parameters were set as follows: positive electrospray ionization mode with + 4500 V electrospray voltage, 15 nebulizer gas units, 10 axillary gas units, 300 °C ion source temperature, 25 curtain gas units and unit resolution on both Q1 and Q3. Multiple-reaction monitoring with parent-to-product ion transitions of 603.7 to 805.4 for VVSVLTVLHQDWLNGK, and 606.4 to 809.4 for VVSVLTVLHQDWLNG [Lys(¹³C6,¹⁵N2)] were set.

The total drug immunocapture LC-MS/MS assay qualifications involved testing several key parameters, including accuracy, precision, specificity and carryover using calibration standards and quality control dilution samples prepared from the 32.8 mg/mL mAb2 drug product material and rat plasma. Sciex 6500 + with Analyst software automatically calculated and annotated the peak areas. A calibration curve for total drug was generated using peak area ratios of calibration standards, applying a linear, 1/x² weighted regression to calculate the total drug concentrations in rat plasma samples.

Glyco-pair analysis by Endo F3 digestion and LC-MS

Plasma samples were analyzed for glyco-pair investigation using an immunocapture, Endo F3 enzymatic digestion, LC-MS method. Intact analysis standards were prepared at 10, 50, and 200 µg/mL from the 32.8 mg/mL mAb2 drug product material and rat plasma.

50 µL of plasma sample were used with 16 µL of 1.5 mg/mL biotinylated goat anti-human IgG (Jackson immuno research), 450 µL TBS-T, and 25 µL of freshly prepared 10 mg/mL magnetic beads solution (Streptavidin MagneSphere^® Paramagnetic Particles, Promega) were added to a 96-deepwell plate (PP, Thermo Fisher Scientific) and incubated for 2 h at RT. After magnetic separation of the beads, the samples were washed three times with 300 µL TBS-T and twice with 300 µL water, then eluted with 75 µL 25 mM HCl on a Kingfisher Flex magnetic bead handler and immediately neutralized with 4 µL of 1 M Tris-HCl (pH 8.0) to pH 4.5. Then, 6 µL of 10X Glycobuffer 4 and 1 µL of Endo F3 (New England Biolabs) were added followed by incubation at 37 °C overnight. Following overnight digestion, a 10 µL of 10% formic acid in water was added to each sample and incubated at 37 °C for 30 min. Samples were centrifuged for LC-MS analysis.

A Waters BEH C4 reversed phase column (300 μm x 50 mm, 300 Å, 1.7 μm) using a Waters Acquity UPLC M-Class coupled to an AB Sciex 7600 zenoTOF mass spectrometer was used and operated at 65 °C with a flow-rate of 5 µL/min. Mobile phases consisted of 0.1% formic acid in water (A) and in 2-propanol (B). After sample injection (10 µL) the following gradient was applied: 0.00:15, 1.00:15, 4.00:90, 7.50:90, 7.51:15, 9.00:15 (time in min: percentage of mobile phase B).

The mass spectrometry was operated with positive mode and parameters were + 5000 V electrospray voltage, 15 nebulizer gas units, 15 axillary gas units, 450 °C ion source temperature, and 35 curtain gas units, mass range 1200–3500 m/z, 0.25 s accumulation time. Sciex 7600 with Sciex OS software were used for data acquisition and processing. The different glyco-pairs were distinguished based on their deconvoluted masses, and their respective peak intensities were utilized to determine their abundances.

Data processing for PK analysis

The ToxKin Verssion 4.1.4 (Entimo AG) was used for the non-compartmental PK analysis. The last three time points (96 h, 168 h and 336 h) were used to calculate the half-life (t_1/2) and the clearance (CL) for rat 1–4. Due to much lower concentration of the 336 h sample of rat 5, the time points 72 h, 96 h and 168 h were used. Data from one rat (rat 1) was excluded for further analysis due to an abnormal PK profile (Figure S 13).