Divergent Enzymatic Assembly of a Comprehensive 64‐Membered IgG N‐Glycan Library for Functional Glycomics

Abstract N‐Glycosylation, a main post‐translational modification of Immunoglobulin G (IgG), plays a significant role in modulating the immune functions of IgG. However, the precise function elucidation of IgG N‐glycosylation remains impeded due to the obstacles in obtaining comprehensive and well‐defined N‐glycans. Here, an easy‐to‐implement divergent approach is described to synthesize a 64‐membered IgG N‐glycan library covering all possible biantennary and bisected N‐glycans by reprogramming biosynthetic assembly lines based on the inherent branch selectivity and substrate specificity of enzymes. The unique binding specificities of 64 N‐glycans with different proteins are deciphered by glycan microarray technology. This unprecedented collection of synthetic IgG N‐glycans can serve as standards for N‐glycan structure identification in complex biological samples and the microarray data enrich N‐glycan glycomics to facilitate biomedical applications.


Introduction
Immunoglobulin G, one of the most common antibodies presented primarily in the blood and tissues, plays a significant role in physiological and pathological activities.It has been discovered that IgG comprises a conserved N-glycosylation site at Asn297 in the Fc domain and 15-25% N-glycans in the Fab region, which can remarkably alter and modulate their biofunctions through the synergistic effects of glycan structures and intrinsic peptide DOI: 10.1002/advs.202303832sequence. [1]Recent studies have shown that alteration of IgG Fc glycosylation can modulate its effector functions and variations of IgG N-glycans are distinctly reflected in diseases. [2]Sialylation of Fc domain on IgG with immune complex can enhance the anti-inflammatory activity. [3]Removal of the core fucose on Fc domain can reinforce the antibody-dependent cell cytotoxicity (ADCC) by specifically increasing FcRIIIa binding, [4] and bisecting GlcNAc moiety can also generate improved ADCC due to the steric hindrance. [5]Through glycoengineering technologies, two FDAapproved defucosylated antibodies named Mogamulizumab and Obinutuzumab have been produced and clinically used.Disialylated glycans on IgG Fab domain are also found to influence the antigen binding process through dynamic simulation analysis on Fab crystal structure and the Fab glycosylation can affect the activation of B cells possibly due to the interactions of sialic acid-modified Fab glycans with immune-associated lectins like galectin-9 or siglec-10. [6]Meanwhile, an increased sialylation, bisection, and fucosylation of Fab N-glycans were observed in patients with multiple myeloma. [7]Consequently, structurally well-defined IgG N-glycans with high purity and sufficient quantity are urgently needed for elucidating the biofunctions of IgG N-glycosylation and developing diagnostic tools as well as glycoengineered antibody-based drugs.These compounds can be utilized as standards to conduct N-glycan identification in complex biological samples and as probes to investigate their functional mechanisms at molecular level, also as a precursor for preparations of homogenous glycoengineered antibody. [8]he structural complexity and heterogeneity of IgG N-glycans make it difficult to isolate adequate and structurally well-defined glycans from serum or tissue samples for biological function studies.Although great efforts have been devoted to preparing complex N-glycans with asymmetrical or symmetrical structures through chemical and chemoenzymatic methods, [9] the easy-to-implement systematic synthesis of comprehensive IgG N-glycans has not yet been reported.The exact structural elucidations and precise functional investigations of IgG N-glycosylation are still hampered due to the absence of an extensive and systematic N-glycan library with well-defined structures.To address this problem, we here report an easy-to-implement divergent approach to enzymatically assemble a comprehensive and unprecedented 64-membered IgG N-glycan library containing all possible biantennary and bisected N-glycans with or without core fucosylation from a readily accessible precursor (Figure 1).The key feature of this approach is that various asymmetrical N-glycans were enzymatically prepared by reprogramming biosynthetic assembly lines based on the inherent branch selectivity and substrate specificity of enzymes.More importantly, regioselective sialylation of N-glycans was achieved by differentiating the branches with specially designed assembly sequences to generate all terminal sialic acid linkage isomers of N-glycans on IgG.In addition, the synthetic 64 N-glycans were subjected to a systematic exploration into the binding specificities with plant lectins and immune-associated lectins using glycan microarray technology, thereby providing abundant information for the development of diagnostic reagents and antibody-based drugs.
With glycan 2 in hand, we turned our attention to divergent enzymatic synthesis of asymmetrical N-glycans by exploring inherent branch selectivity of ST6Gal1 and E. coli galactosidase as well as differentiating the branches with specifically designed glycosylation sequences for regioselective sialylation.First, selective 2,6-sialylation of 2 with CMP-Neu5Ac (1 equiv) and ST6Gal1 that preferred the bottom antenna provided asymmetrical glycan 9. [11] And selective removal of the galactose residue at bottom antenna of 2 was achieved by galactosidase from E. coli to afford Scheme 1. Divergent enzymatic synthesis of core-fucosylated biantennary N-glycans 1-16.asymmetrical glycan 3. [12] After the rapid purification using Bio-Gel P-4 size-exclusion chromatography, the desired glycans 9 and 3 were obtained with minor contamination of unreacted starting materials and byproducts that can be easily removed by semipreparative high performance liquid chromatography (HPLC) with a hydrophilic interaction chromatography (HILIC) column (Table S2, Supporting Information) to afford highly pure compounds.Given the fact that 2,6-sialoside can block further enzymatic modifications, treatment of 9 with ST3Gal4 and galactosidase from A. niger for corresponding 2,3-sialylation and removal of terminal Gal residue at upper antennae provided glycans 7 and 13, respectively.Compound 13 was further treated with 2-3,6,8neuraminidase for efficient cleavage of sialoside to give glycan 4 as the isomer of 3. Glycan 4 was characterized with terminal Glc-NAc and LacNAc moiety at upper antenna and bottom antenna, respectively.Many glycosyltransferases modify LacNAc but not GlcNAc moiety, [13] thus 2,3-sialylation of 4 at bottom antenna with ST3Gal4 and CMP-Neu5Ac afforded glycan 14 as the isomer of 13.Further galactosylation of 14 with B4GalT1 and UDP-Gal generated glycan 10 as the isomer of 9. Next, following a similar enzymatic glycosylation sequence, glycan 3 was successively treated with ST3Gal4 and B4GalT1 to give glycans 16 and 12.In parallel, 2,6-sialylation of compound 3 with ST6Gal1 and CMP-Neu5Ac afforded glycan 15, which was further galactosylated by B4GalT1 and UDP-Gal to provide glycan 11.Subsequently, 2,3sialylation of LacNAc moiety at the bottom antenna with ST3Gal4 and CMP-Neu5Ac afforded disialylated glycan 8 as the isomers of 5-7.

Divergent Enzymatic Synthesis of Bisected N-Glycans and Core-Unmodified N-Glycans
First, attention was focused on the enzyme-mediated preparation of various bisected N-glycans with core fucose.1,4-Mannosylglycoprotein 1,4-N-acetylglucosaminyltransferase (MGAT3) is responsible for catalyzing the addition of bisecting GlcNAc onto N-glycans. [14]Having glycans 1-4 in hand, we investigated the substrate specificity of MGAT3 (Scheme 2A), which indicated that galactosylation of the GlcNAc residue at the bottom antenna (glycans 2 and 4) prevented the activity of MGAT3.

Structural Characterization of the Representative Bisected N-Glycans
The structural identifications of all synthesized N-glycans were confirmed by detailed analysis of NMR and ESI-MS spectroscopy.Herein, the representative bisected complex N-glycan 24, which contained core fucose and different sialic acid linkages (2,6-Neu5Ac and 2,3-Neu5Ac) at the terminal, was fully characterized (Page S37, Supporting Information) by 1D and 2D NMR spectra shown in Figure 2A-C.In 1 H NMR spectra, anomeric proton signals resonate from 4.45 to 5.18 ppm.Despite the highly overlapped signals, these signals can be successfully assigned by 2D 1 H-13 C HSQC experiment (Figure 2C).The anomeric signal of the bisecting GlcNAc-5 labeled in red was unique, its chemical shift ( = 4.46 ppm) is obviously smaller than anomeric chemical shifts ( = 4.56-5.18ppm) of other GlcNAc residues.The characteristic signals at 1.21 ppm corresponded to methyl of core fucose residue.Interestingly, the terminal 2,3and 2,6-linked Neu5Ac residues can be readily distinguished by the chemical shifts of H-3.The H-3 signals ( H3ax = 1.84 and  H3eq = 2.75 ppm) of 2,3linked Neu5Ac displayed the lower field values than ones ( H3ax = 1.72 and  H3eq = 2.68 ppm) of 2,6-linked Neu5Ac, respectively.Additionally, four disialylated isomeric N-glycans 21-24 exhibited different retention times in HILIC column by LC-MS analysis, demonstrating that all these synthesized N-glycans could be utilized as standards to conduct N-glycan identification in complex biological samples.

Glycan Microarray Analysis
Glycan microarray technologies are widely used for the rapid analysis of carbohydrate-protein interactions, which makes it possible to probe the binding specificities of N-glycans in a highthroughput manner.Recently, Paulson and coworkers chemoenzymatically synthesized various N/O-glycans with extended Lac-NAc repeats for glycan microarray profiles of receptor specificities of influenza virus and middle east respiratory syndrome coronavirus. [15]9s] Additionally, a neoglycoprotein array enabling a high-affinity multivalent screening was developed by Gildersleeve and coworkers to decipher the recognitions between N-glycans and proteins such as plant lectin, bacterial-related lectin, and immune-associated lectin. [16]o better understand the binding specificities of diverse Nglycans on IgG, the well-characterized compounds 1-64 in this study were attached with 2-[(methylamino)oxy]ethanamine spacer, [17] and the resulting derivatives were printed on Nhydroxysuccinimide (NHS)-activated glass slides for deciphering the interactions with glycan-binding proteins using glycan microarray technology. [18]

Glycan-Binding Specificities of Immune-Associated Lectins
The bindings of five representative immune-associated lectins with N-glycans were investigated.Dendritic cell-specific ICAM-3-grabbing nonintegrin (DC-SIGN) and liver/lymph nodespecific ICAM-3 grabbing non-integrin (L-SIGN), are two transmembrane C-type lectin receptors respectively expressed by dendritic cells and type II alveolar cells of human lungs, which mediate immune responses and participate in viral recognition and microbial pathogens presentation. [20]It is reported that DC-SIGN binds high-mannose and branched-fucosylated structures, [21] while the binding specificity of L-SIGN with Nglycans has not yet been systematically explored by glycoarray technology.In our arrays, DC-SIGN and L-SIGN both presented preference toward the N-glycan core heptasaccharide structure with or without fucosylation (Figure 4A,B, e.g., compounds 1 and 33).It is worth noting that core fucosylated N-glycan were preferably recognized by DC-SIGN, whereas L-SIGN mainly bound defucosylated glycans.Human dendritic cell immunoreceptor (DCIR), an inhibitory receptor that is expressed on antigen-presenting cells including B cells, monocytes, and myeloid dendritic cells (DCs), involves in vital activities in immune responses and autoimmune diseases. [22]DCIR was reported to bind fucosylated and mannose-modified oligosaccharides, [22c] and has been proven to recognize pathogenic organisms and participate in the attachment for human immunodeficiency virus (HIV) in DCs, [22b] while the precise structures of N-glycan ligands are still ambiguous.We tested the recognition specificities of hDCIR on our  S1 (Supporting Information).
To better elucidate the N-glycosylation functions of IgG, we tested the bindings of four type I Fc receptors, and two lectins (Siglec-10 and Galectin-9) that mediate the activation of B cells possibly due to the interactions with Fab glycans. [6]Unfortunately, all synthetic N-glycans displayed no binding to four type I Fc receptors, indicating that the protein segment is necessary for IgG bindings to Fc receptors.Siglec-10 is acknowledged to be more inclined to bind glycans containing 2,6-sialylated LacNAc motifs. [24]The binding profiles of asymmetric N-glycans and isomers are sophisticated, of which N-glycans with 2,6-sialylated LacNAc on 1,3-Man-branch and 2,3-sialylated LacNAc on 1,6-Man branch represented a relative stronger recognition specificity (Figure 4D, e.g., compounds 23 and 7).Galectin-9 is a galactoside-binding lectin that plays essential roles in mediat-ing B cell activation and altering the progression of infectious diseases. [25]Both galactose and LacNAc moieties are the common domains that interact with Galectin-9 and other galectins.In our tests, a broader tolerance to LacNAc-containing N-glycans of Galetin-9 was discovered, and nearly all LacNAc-containing glycans were bound, of which N-glycans with 2,3-Neu5Ac were preferred over 2,6-sialylated ones (Figure 4E, e.g., compound 5 vs 6, compounds 21 vs 22).
All the results are shown as average relative fluorescence unit (RFU) that was calculated for four independent replicates on the glycoarrays after removing the highest and lowest signals.The error bars represent the standard deviation (SD) among the values of four replicate spots.The RFU and SD of data are provided in Table S1 (Supporting Information).

Conclusion
In summary, we developed an easy-to-implement approach to efficiently synthesize a comprehensive and systemic IgG www.advancedscience.comN-glycan library containing all possible biantennary and bisected N-glycans for functional glycomics studies by glycan microarray technology.Key features include divergent enzymatic assembly of various N-glycans via reprogramming biosynthetic assembly lines based on the inherent branch selectivity and substrate specificity of enzymes, as well as regioselective sialylation of Nglycans by differentiating the branches with specially designed assembly sequences to generate all terminal sialic acid linkageisomers.The binding studies of 64 N-glycans with different proteins demonstrated that the architecture and topology of complex N-glycan affected recognition.This work enriches the synthesis space of N-glycan, provides invaluable standards for N-glycan structure identification in complex biological samples, and offers opportunities for precise functional investigations of IgG Nglycosylation to optimize the performances of glycoengineered therapeutic antibodies. [26]

24 .
Following similar enzymatic glycosylation conditions, compound 32 was successively converted into glycans 28 and 23.To diversity-oriented synthesize asymmetrical bisected N-glycans (Scheme 2D), selective 2,6-sialylation of the bottom antenna LacNAc moiety of 18 using ST6Gal1 and CMP-Neu5Ac (1 equiv) was achieved to afford glycan 25.Treatment of 25 with galactosidase from A. niger for the cleavage of terminal galactose residue at the upper antenna provided glycan 29, which was further treated by 2-3,6,8-neuraminidase to remove terminal Neu5Ac residue to generate glycan 20 as the isomer of 19.The resulting LacNAc moiety of 20 at bottom antenna was 2,3-sialylated by ST3Gal4 and CMP-Neu5Ac to afford glycan 30.The terminal GlcNAc residue of 30 at the upper antenna was selectively galactosylated by B4GalT1 and UDP-Gal (1 equiv) to give glycan 26 as the isomer of 25 without affecting bisecting GlcNAc.Additionally, to further diversify the IgG N-glycan library, the core-fucosylated www.advancedscience.comN-glycans1-32 were treated with a robust fucosidase FucA1 for the removal of fucoside to afford core-unmodified N-glycans 33-64, respectively.[9k]

Figure 2 .
Figure 2. NMR and LC-MS analysis of complex bisected N-glycans.A) Anomeric proton signals, unique signals of core fucose residue, and sialic acid residue in 1 H NMR spectra of 24.B) Chemical structure of 24.C) Anomeric signals in 1 H-13 C HSQC spectra of 24.D) LC-MS analysis of four disialylated N-glycans 21-24.