Terminal Epitope-Dependent Branch Preference of Siglecs Toward N-Glycans

Siglecs are sialic acid–binding immunoglobulin-like lectins that play vital roles in immune cell signaling. Siglecs help the immune system distinguish between self and nonself through the recognition of glycan ligands. While the primary binding specificities of Siglecs are known to be divergent, their specificities for complex glycans remain unclear. Herein, we determined N-glycan binding profiles of a set of Siglecs by using a complex asymmetric N-glycan microarray. Our results showed that Siglecs had unique terminal epitope-dependent branch preference when recognizing asymmetric N-glycans. Specifically, human Siglec-3, -9, and -10 prefer the α1-3 branch when Siaα2-6Galβ1-4GlcNAc terminal epitope serves as the binding ligand but prefer the opposite α1-6 branch when Siaα2-3Galβ1-4GlcNAc epitope serves as the ligand. Interestingly, Siglec-10 exhibited dramatic binding divergence toward a pair of Neu5Ac-containing asymmetric N-glycan isomers, as well as their Neu5Gc-containing counterparts. This new information on complex glycan recognition by Siglecs provides insights into their biological roles and applications.


INTRODUCTION
Sialic acid-binding immunoglobulin-like lectins (Siglecs) are cell-surface transmembrane receptors that are differentially expressed on immune cells (Läubli and Varki, 2020). They play critical roles in immune cell signaling and help the immune system to distinguish self and nonself (Macauley et al., 2014). Most Siglecs, with the only exception being sialoadhesin/Siglec-1, have C-terminal regulatory motifs in their cytoplasmic domains that participate in the regulation of immune systems. On the Nterminal, each Siglec has a V-set immunoglobulin (Ig) domain that recognizes sialic acid-containing glycans (Duan and Paulson, 2020). There are 15 human Siglecs and 9 murine Siglecs. Among those, four are conserved across mammals 2,4,and 15). All remaining Siglecs are named CD33related Siglecs as they contain less conserved structure between humans and other vertebrates, but all have high homologies to CD33.
Siglecs are immune-modulatory receptors within the mammalian immune system. Most Siglecs have intracellular immunoreceptor tyrosine inhibitory motifs (ITIMs) that can, in principle, participate in inhibitory or activating signals. The binding of anti-Siglec antibodies or multivalent trans-ligand with inhibitory Siglecs can activate/ phosphorylate the ITIMs and produce negative signals (Duan and Paulson, 2020). Additionally, some Siglecs are specifically expressed on certain types of immune cells and presented as endocytic receptors. Hence, they were utilized as the desired target for drug development. For example, Siglec-3, also called CD33, is an inhibitory receptor that is relatively specifically expressed on myeloid lineage and endocytosed upon antibody binding, thus serving as a specific target for developing therapeutic antibodies. Gemtuzumab ozogamicin is the first approved CD33-targeting antibody-drug conjugate (ADC) and was used for induction therapy of acute myeloid leukemia (AML) (Laszlo et al., 2014).
Despite the diverse roles that Siglecs play in immune cell regulation and disease processes, their natural ligands, especially the fine binding specificity, toward complex glycans are relatively underinvestigated. Glycan microarray was developed for identifying interactions between glycans and glycan-binding proteins (GBPs) 2 decades ago (Fukui et al., 2002;Palma and Chai, 2019). It enabled simultaneous binding analysis of GBPs to hundreds of glycan structures and had become a major tool to unveil glycan-protein interactions (Gao et al., 2019b). Various versions of glycan microarray were used to investigate interactions between glycans and Siglecs (Blixt et al., 2003;Bochner et al., 2005;Campanero-Rhodes et al., 2006;Rillahan et al., 2012;Rillahan et al., 2013;Gao et al., 2019a). However, the fine specificity details of Siglecs toward natural complex glycans remain largely unknown.
Herein, we investigated the binding specificity of Siglec-3, -9, -10, and -F using a unique glycan microarray containing 98 structurally well-defined complex glycans, revealing a unique terminal epitopedependent branch preference toward asymmetric N-glycans. Particularly, a dramatic binding divergence of Siglec-10 toward a pair of N-glycan isomers was observed and further confirmed by synthesized Neu5Gc-containing counterparts. Later, quantitative assay by biolayer interferometry analyses suggested a 67-fold avidity difference among the Neu5Gc-containing isomers.
Chemoenzymatic Synthesis of N-Glycans N-glycans 38 and 54 were prepared as previously reported (Li et al., 2015). For the α2-6sialylation of 38, 100, 54, and 104, reactions were carried out in reaction systems containing Tris-HCl (100 mM, pH 8.0), an acceptor glycan (10 mM), CTP (15 mM), N-acetylneuraminic acid (Neu5Ac) or N-glycolylneuraminic acid (Neu5Gc) (15 mM), MgCl 2 (10 mM), and appropriate amounts of NmCSS and Pd26ST. Reactions were incubated at 37°C for 3 h and monitored by HPLC. After over 95% acceptor was converted, reactions were quenched by the addition of equal volumes of ice-cold ethanol, concentrated, and subject to HPLC separation to afford compounds 99, 101, 103, and 105. Product-containing fractions were pooled and lyophilized for characterization and next step modular assembly. For the β1-4galactosylation of 99 and 103, reactions were performed in mixtures containing Tris-HCl (100 mM, pH 7.5), an acceptor glycan (10 mM), UDP-Gal (15 mM), MgCl 2 (10 mM), and an appropriate amount of NmLgtB. Reactions were incubated at 37°C overnight and monitored by HPLC. After over 95% acceptor was converted, reactions were quenched, concentrated, and subject to HPLC separation of compounds 100 and 104. Product-containing fractions were pooled and lyophilized for characterization and subsequent synthesis. The α2-3sialylation of 100 and 104 was carried out in reaction systems containing Tris-HCl (100 mM, pH 8.0), an acceptor glycan (10 mM), CTP (15 mM), Neu5Gc (15 mM), MgCl 2 (10 mM), and appropriate amounts of NmCSS and PmST1-M144D. PmST1-M144D-catalyzed reactions were incubated at 37°C for 3 h and monitored by HPLC. After over 90% acceptor was converted, the reaction was quenched, concentrated, and subject to HPLC separation to afford compounds 102 and 106. Product-containing fractions were then pooled and lyophilized for characterization.
Newly synthesized N-glycans were purified by HPLC using a Waters XBridge BEH amide column (130 Å, 5 μm, 10 mm × 250 mm) under a gradient running condition (solvent A: water or 100 mM ammonium formate; solvent B: acetonitrile; flow rate: 4.5 ml/min, B%: 65-50% in 30 min) and monitored by UV absorbance at 210 nm. MALDI-TOF MS analyses were performed on UltrafleXtreme MALDI TOF/TOF Mass Spectrometer (Bruker). Scan range of MS was set according to molecular weight, and reflector mode was used for analysis. Mass spectra were obtained in negative extraction mode with the following voltage settings: ion source 1 (19.0 kV), ion source 2 (15.9 kV), and lens (9.3 kV). The reflector voltage was set to 20 kV. The laser was pulsed at 7 Hz and the pulsed ion extraction time was set at 400 ns. The laser power was kept in the range of 40-90%. 1 H NMR spectra were recorded on a Bruker AVANCE 600 (600 MHz) spectrometer at 25°C. All 1 H Chemical shifts (in ppm) were assigned according to D 2 O (δ 4.79 ppm).

Neu5Gc-N-glycan Microarray Fabrication
The AEAB labeled-glycans were prepared at a concentration of 100 μM in the printing buffer (150 mM phosphate, pH 8.5), and printed on multivalent NHS-derivatized microscope-glass slides (Z Biotech, LLC), each for 400 pL in replicates of six, as described previously (Heimburg-Molinaro et al., 2011). Noncontact printing was performed at room temperature with a humidity of 60% by a sciFLEXARRAYER S3 spotter (Scienion) with two PDC 80 Piezo Dispense Capillary. After overnight dehumidification under room temperature, the slides were washed with MilliQ water and subsequently blocked with 50 mM ethanolamine in 100 mM Tris-HCl (pH 9.0) for 2 h. The blocked slides were then washed with MilliQ water twice, dried, and stored desiccated at −20°C until use.

Biolayer Interferometry Receptor Binding Assay and Data Analysis
The AEAB-labeled glycan 102 and 106 were labeled with Biotin by using the reagent EZ-Link ™ NHS-Biotin (Thermo Fisher). In detail, 1 mM AEAB-labeled glycan was incubated with 10 mM NHS-Biotin at room temperature for 10 min. Then, labeled glycans were purified by HPLC to homogeneity using an The purified Biotin-labeled glycans were quantified by HPLC as described above. Avidities were measured by biolayer interferometry using an Octet RED instrument (Pall FortéBio, Fremont, CA, United States). The prepared biotinylated glycans were preloaded onto streptavidin-coated biosensors at up to 100 nM for 3 min in 1× kinetic buffer (Pall FortéBio, Menlo Park, CA, United States). Siglec-10 was diluted to concentrations of 1 μM, 500 nM, and 250 nM with 1× kinetic buffer, respectively. The glycan-loaded biosensors were submerged in wells containing different concentrations of Siglec-10 for 5 min followed by 15 min of dissociation in 1× kinetic buffer at 25°C with the orbital shake speed of 1000 rpm. As a reference control for subtraction, glycan-loaded biosensors were also dipped in wells containing 1× kinetic buffer. The binding kinetics data were processed by the ForteBio data analysis software (version 11.1). The association and dissociation curves were fitted, and the avidity values were calculated by using a heterogeneous ligand (2:1) model.

Chemoenzymatic Synthesis of Neu5Gc-Containing N-Glycans
One interesting observation is that Siglec-10 showed high binding to an asymmetric N-glycan 47 but no binding to its positional isomer 63 (Figure 1). Such a dramatic binding divergence can be Frontiers in Molecular Biosciences | www.frontiersin.org April 2021 | Volume 8 | Article 645999 explained by its terminal epitope-dependent branch preference, as both ligands (Ac3LN and Ac6LN) on 47 are located on the terminal of favored branches, whereas both ligands are located on the unfavored branches of 63. Because Siglec-10 strongly prefers Neu5Gc-containing N-glycans, we speculate that a Neu5Gcmodified counterpart of 47 ( Figure 2A, compound 102) may be of higher affinity and a most favorable N-glycan ligand of Siglec-10. To test this hypothesis and to further validate the terminal epitope-dependent branch preference of Siglec-10, we enzymatically synthesized eight Neu5Gc-containing N-glycans ( Figure 2A). In detail, compounds 99 to 102 were assembled starting from previously prepared glycan 38 (Li et al., 2015). First, α2-6Neu5Gc was installed onto the α1-3 branch to achieve 99 by Pd26ST-catalyzed α2-6sialylation in the presence of cytidine-5′triphosphate (CTP), Neu5Gc, and NmCSS for the in situ generation of the sugar donor CMP-Neu5Gc. Then, β1-4Gal was installed onto the α1-6 branch by NmLgtB-catalyzed reaction in the presence of UDP-Gal to provide 100. The addition of α2-6Neu5Ac to the α1-6 branch of 100 by Pd26ST then provided 101. On the other hand, the addition of α2-3Neu5Gc to this branch by PmST1-M144D-catalyzed α2-3sialylation gave the desired asymmetric N-glycan 102. In the same synthetic manner, another four asymmetric N-glycans 103, 104, 105, and 106 were assembled starting from N-glycan 54. All compounds were purified and characterized by HPLC ( Figure 2B), mass spectrometry, and NMR (supporting information).
Avidity of Siglec-10 to N-Glycans 102 and 106 As shown in Figure 3, the binding signals of Siglec-10 to 102 is around 5-fold stronger than to its positional isomer 106 and 20-fold stronger than to linear glycans 97, suggesting compound 102 as a potential high-affinity ligand of Siglec-10. The avidity of Siglec-10 toward 102 and 106 was thus measured by biolayer interferometry (BLI). AEAB labeled 102 and 106 were further conjugated with NHS-Biotin and purified with HPLC, and then immobilized onto streptavidin-coated biosensors for BLI assay (Figure 4). The association and dissociation curves were fitted, and the avidity values were calculated with the consideration of the bivalency of the Siglec-10-Fc chimera protein.
The avidity values of Siglec-10 toward 102 and 106 were 0.11 μM and 7.34 μM, respectively, indicating a 67-fold higher avidity of 102 than 106. The result further confirmed the terminal epitopedependent branch preference and revealed a high avidity glycanbinding partner (102) of human Siglec-10.

DISCUSSION AND CONCLUSION
Siglecs are attractive therapeutic targets and several related antibody-based therapies had been developed for the treatment of immune-related diseases. In certain applications, glycan ligands have an advantage over antibodies, such as their ability to dissociate from their target once endocytosed. However, glycan-based therapeutic strategies for cargo delivery and immunomodulation are underinvestigated due to the lack of suitable ligands (Angata et al., 2015). A comprehensive understanding of glycan recognition details by Siglecs is essential toward the discovery and designing of efficient ligands. In fact, recent advances in glycobiology have prompted such applications. For example, high specific efficient N-glycan ligands with chemical modifications toward Siglec-2 were reported (Peng and Paulson, 2017). Conjugates of toxins with this novel ligand could be efficiently internalized via Siglec-2, resulting in the killing of B-cell lymphoma cells.
In this study, we screened binding profiles of Siglec-3, -9, -10, and -F against a comprehensive N-glycan microarray to reveal glycan recognition details of Siglecs ( Table 1). The results showed a surprising terminal epitope-dependent branch preference toward N-glycans by Siglec-3, -9, and -10. These Siglecs prefer the α1-3 branch of N-glycans when α2-6sialylated epitopes serve as binding ligands, while they have an opposite preference to the α1-6 branch when α2-3sialylated epitopes serve as ligands. Such a feature could assist in designing high-affinity binding partners of Siglecs. For example, we designed and synthesized an asymmetric N-glycan (102) with much higher avidity than its positional isomer toward Siglec-10. Note that recombinant Siglec-Fc chimera proteins in the form of disulfidelinked homodimer were used in this study instead of native Siglecs. Even though such chimera proteins were widely used to reveal the glycan recognition of Siglecs and other human GBPs (Blixt et al., 2003;Bochner et al., 2005;Campanero-Rhodes et al., 2006;Rillahan et al., 2012;Rillahan et al., 2013;Gao et al., 2019b;Rodrigues et al., 2020), the nonnatural bivalent form could possibly influence their fine specificity toward glycan-binding partners.
High-avidity binding partners of Siglecs could lead to extensive academic and clinical implementations. For example, tumor cells can escape the surveillance of the immune system via inhibition of immune cells through immune checkpoints and their ligands. A promising therapeutic approach for cancer is to block these immune checkpoints, for example, the programmed cell death ligand 1 (PD-L1) and cytotoxic T lymphocyte-associated protein 4 (CTLA-4) (Leach et al., 1996;Topalian et al., 2012). A recent report showed that CD24-Siglec-10 interaction is an innate immune checkpoint that is essential for mediating antitumor immunity and can promote tumor immune escape. The modulation of this interaction is expected to become a new target for tumor therapy (Barkal et al., 2019). The  Additionally, CD24-Siglec-10 interaction could suppress the immune response to the danger-associated molecular pattern (DAMP) (Cai et al., 2009;Rillahan et al., 2012). It is thus tempting to speculate that the strong Siglec-10 binding partner 102, with or without further modification, may serve as an invaluable reagent to block this immune checkpoint.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

AUTHOR CONTRIBUTIONS
LL conceived and designed the project. SW,CC and LL performed glycan synthesis and glycan microarray assays, MG and DL performed kinetic assay, SW wrote the manuscript, and LL and XW revised the manuscript, which was edited and approved by all authors.

FUNDING
This work was supported by the National Institutes of Health (Grant No. U54HL142019 and U01GM125288 to LL). XW and MG were partially supported by the National Institutes of Health (Grant No. R21AI144433 to XW).