Understanding the Glycosylation Pathways Involved in the Biosynthesis of the Sulfated Glycan Ligands for Siglecs

Abstract

Carbohydrate sulfation plays a pivotal role in modulating the strength of Siglec–glycan interactions. Recently, new aspects of Siglec binding to sulfated cell surface carbohydrates have been discovered, but the class of glycan presenting these sulfated Siglec ligands has not been fully elucidated. In this study, the contribution of different classes of glycans to cis and trans Siglec ligands was investigated within cells expressing the carbohydrate sulfotransferase 1 (CHST1) or CHST2. For some Siglecs, the glycan class mediating binding was clear, such as O-glycans for Siglec-7 and N-glycans for Siglec-2 and Siglec-9. Both N-glycans and mucin-type O-glycans contributed to ligands for Siglec-3, -5, -8, and -15. However, significant levels of Siglec-3 and -8 ligands remained in CHST1-expressing cells lacking complex N-glycans and mucin-type O-glycans. A combination of genetic, pharmacological, and enzymatic treatment strategies ruled out heparan sulfates and glycoRNA as contributors, although Siglec-8 did exhibit some binding to glycolipids. Genetic disruption of O-mannose glycans within CHST1-expressing cells had a small but significant impact on Siglec-3 and -8 binding, demonstrating that this class of glycans can present sulfated Siglec ligands. We also investigated the ability of sulfated cis ligands to mask Siglec-3 and Siglec-7. For Siglec-7, cis ligands were again found to be mucin-type O-glycans. While N-glycans were the major sulfated trans ligands for Siglec-3, disruption of complex mucin-type O-glycans had the largest impact on Siglec-3 masking. Overall, this study enhances our knowledge of the types of sulfated glycans that can serve as Siglec ligands.

Graphical Abstract

1. INTRODUCTION

The immune system of animals plays a critical role in protecting against harmful pathogens.¹ To successfully combat these foreign invaders, striking a balance between activation and inhibition is essential to target and eliminate the specific pathogenic molecules in the body. As a breach of immune tolerance can result in autoimmune diseases where immune cells start attacking their own tissues or cells, cells or tissues have the ability to engage immune inhibitory receptors expressed on immune cells to help differentiate “self” from “non-self”.² Of the many mechanisms at play in performing this important function of self-recognition, engagement of cellular glycans with immunomodulatory receptors is one of the key mechanisms to regulate these receptors as glycans are abundant on the cell surface and have unique biochemical structures compared to pathogens.³ Unlike most bacteria and fungi, which generally do not express sialic acid (Sia), vertebrate cells commonly have many of their glycan conjugates capped with Sia.⁴ In mammals, sialic acids are recognized by sialic acid-binding immunoglobulin-type lectins (Siglecs) to help recognize self.⁵ Siglecs in humans are part of the immunoglobulin superfamily and consist of 15 members in humans and 9 in mice, mainly expressed on various types of white blood cells.⁶ Most Siglecs contain immunoreceptor tyrosine-based inhibitory motifs (ITIMs) or immunoreceptor tyrosine-based switch motifs (ITSMs) in their cytosolic tails that allow them to antagonize immune cell signaling pathways. Due to their ability to regulate immune cells, Siglecs are now considered glyco-immune checkpoints.^3,7

Abnormal Siglec function has been linked to various diseases.³ For example, cancer cells exploit the Siglec-sialic acid axis by presenting high levels of sialic acids on their surface.^8,9 Carbohydrate sulfation, alongside sialylation, has been identified as a vital monosaccharide modification for fine-tuning the interactions between glycans and several Siglecs, such as Siglec-2 (CD22), Siglec-7, Siglec-8, and Siglec-9. Notably, high carbohydrate sulfotransferase 1 (CHST1) expression has been linked to lower survival rates in various types of cancer.^10–13 Recently, Siglec-3 (CD33), Siglec-5/14, and Siglec-15 have also been found to interact with sulfated sialoglycans.^13,14 Sulfate modifications relevant to Siglecs typically occur at the 6-O position of either galactose (Gal) or N-acetylglucosamine (GlcNAc).^13,14 When sulfate groups are positioned near the sialic acid within two monosaccharides, they significantly enhance Siglec binding affinity to glycan ligands by stabilizing the interaction.^13,14 The strongest binding affinity for Siglec-3 and -8 has been observed with 6-O-disulfation at both Gal and GlcNAc on an α2–3-linked sialyl-LacNAc backbone.^11,13 In addition, it has been discovered that the brain-resident ligands for Siglec-3 and Siglec-8 are receptor protein tyrosine phosphatase zeta (RPTPζ) carrying sialylated keratan sulfates (KS), a type of glycosaminoglycan consisting of repeated LacNAc moieties with a high degree of 6-O-Gal and 6-O-GlcNAc sulfation.^15–18

CHSTs are an enzyme family involved in sulfation on multiple carbohydrate substrates.^13,14 Carbohydrate sulfation has significance as it can mediate multiple biological processes such as anticoagulation, cell-to-cell communication, and regulation of growth factor signaling.¹⁹ Notably, carbohydrate sulfation is a crucial modification for generating glycosaminoglycans like chondroitin sulfates, heparan sulfates, and keratan sulfates, which are made up of repeating disaccharide units with a high degree of sulfation.^20–22 Carbohydrate sulfation catalyzed by CHST1 or CHST2 has also been studied for its ability to strengthen the interactions between Siglecs and their sialoglycan ligands.^10,23,24 CHST1 is a transmembrane protein localized in the Golgi apparatus that transfers an activated form of sulfate, 3′-phosphoadenosine-5′-phosphosulfate (PAPS), to specific hydroxyl groups on carbohydrates.^25,26 Although the binding of Siglecs to sulfated glycans has been analyzed by a few biochemical approaches such as microarray with synthetic sulfated glycans and gene knockout/in or pharmaceutical perturbation of specific glycosyl transferases,^{13,14,17,27,28} information regarding the types of glycosylation and sialyltransferases that produce CHST1-dependent Siglec ligands remains limited. For example, it was shown that the interactions between Siglec-7 and CHST1 overexpressing (CHST1^O/E) HEK293 cells were lost in the absence of COSMC (C1GALT1-specific chaperone) expression while Siglec-3 and -8 still interacted with the CHST1^O/E HEK293 cells lacking N- or mucin-type O-glycosylation.¹⁴ In our previous study, we also investigated the effects of pharmaceutical inhibitors blocking N-glycosylation with kifunensine and mucin-type O-glycosylation with benzyl-α-GalNAc on Siglec-3. Consistent with the above observation in genetic knockouts (KO), we observed that both kifunensine and benzyl-α-GalNAc reduced Siglec-3 binding to CHST1 overexpressing U937 cells, suggesting that both N- and mucin-type O-glycosylation contribute to the biosynthesis of sulfated ligands for Siglec-3 in the presence of CHST1.¹³ Moreover, knockout of a gene encoding beta-galactoside alpha-2,3-sialyltransferase 4 (ST3Gal4) in the CHST1^O/E HEK293 cells reduced the binding of both Siglec-3 and -8, although marginal but significant residual binding for both Siglecs persisted, suggesting the involvement of other sialyltransferases in the sulfated glycan ligand biosynthesis for these Siglecs.

In this study, we investigated the major classes of glycans responsible for biosynthesizing sulfated glycan ligands for Siglecs by disrupting the expression of key enzymes involved in N- or mucin-type O-glycosylation. Using gene knockout cell lines with overexpressed CHST enzymes, we analyzed both cis and trans ligands of Siglecs. For trans ligands, we identified the primary glycan types serving as Siglec ligands and, in several cases, found that sulfated glycan ligands likely extend beyond N- and mucin-type O-glycans. Surprisingly, the preferred trans ligands did not always correspond to the preferred cis ligands. These findings expand our understanding of the glycan classes that present sulfated Siglec ligands.

2. RESULTS

2.1. Generation of MGAT1^−/− or COSMC^−/− U937 Cell Lines Overexpressing CHST1.

First, we performed RNA-seq analysis on U937 cells to evaluate the expression of genes related to glycosylation, recognizing the necessity of integrating comprehensive genomic and glycomic analyses into our study (Figure S1). Subsequently, to determine which classes of glycans display sulfated glycan ligands for each Siglec, we initially generated three CRISPR/Cas9 KO U937 cell lines: MGAT1^−/−, COSMC^−/−, and MGAT1^−/−COSMC^−/− double knockout (DKO) cells (Figure 1A). To validate these KO cell lines, we utilized four lectins: SNA to detect α2–6-linked sialosides, and MAL-II, SLBR-N, and diCBM40 to detect α2–3-linked sialic acids, in conjunction with flow cytometry. The absence of MGAT1 leads to the biosynthesis of oligomannose forms of N-glycans.²⁹ As N-glycans are the primary source for the Neu5Acα2–6Galβ1–4GlcNAc moiety, SNA exhibited the expected loss in binding in the MGAT1^−/− cells.^30,31 However, the DKO cells showed restored binding to SNA compared to the MGAT1^−/− cells (Figure 1B), which may be attributed to enhanced usage of an alternative acceptor by ST6Gal1 in the absence of N- and mucin-type O-glycans. MAL-II and SLBR-N lost binding in the COSMC^−/− cells (Figure 1C,D), which is in line with recognition of mucin-type O-glycans by these lectins.^32,33 For diCMB40, which was previously shown to bind N-glycans containing α2–3 linked Sia, its binding decreased in both MGAT1^−/− and COSMC^−/− single KO cells, although the primary interactions appeared to be with mucin-type O-glycans (Figure 1E).

Figure 1.

Validation of MGAT1^−/−, COSMC^−/−, and DKO U937 cells. (A) MGAT1^−/−, COSMC^−/−, and MGAT1^−/−COSMC^−/− (DKO) U937 cells were generated using CRISPR/Cas9. (B–E) The MGAT1^−/−, COSMC^−/−, and MGAT1^−/−COSMC^−/− DKO U937 cells were validated with (B) SNA, (C) MAL-II, (D) SLBR-N, and (E) diCBM40 lectins by flow cytometry. The gray histograms represent unstained cells. Data are presented as mean ± SD. Statistical significance was calculated using one-way ANOVA with Tukey’s multiple comparisons test.

Next, we introduced CHST1 into MGAT1^−/−, COSMC^−/−, and MGAT1^−/−COSMC^−/− cell lines via lentiviral transduction. We focused on analyzing the sulfated mucin-type O-glycome in the COSMC^−/−CHST1^O/E U937 cells using mass spectrometry-based sulfoglycomics (Figure 2A). In a single sulfation mode, mono- or disialylated core-1 O-glycans with single sulfation at 6-O-Gal were abundant (Figures 2B and S2A). In contrast, core-2 O-glycans containing two sulfate groups at both sides of 6-O-Gal residues were enriched in disulfation mode (Figure 2D and S2C). Compared to the WT CHST1^O/E U937 cells, the COSMC^−/−CHST1^O/E cells exhibited a substantial reduced number of both mono- and disulfated O-glycans (Figures 2C,E and S2B,D). A small residual amount of monosulfated Galβ1–3GalNAc (core-1) O-glycans (approximately 1%) in the COSMC^−/−CHST1^O/E cells was still observed (Figure S3), while disulfated species were completely undetectable. Overall, knocking out COSMC under CHST1 expression greatly diminished the level of the branched mucin-type O-glycans carrying 6-O-Gal sulfation.

Figure 2.

Mass spectra of mucin-type O-glycans from WT and COSMC^−/− CHST1^O/E U937 cells. (A) O-glycans from the WT and COSMC^−/− U937 cells expressing CHST1 were isolated and analyzed using MALDI and LC–MS. Spectra of mucin-type O-glycans, in both mono- and disulfated forms, were obtained from WT CHST1^O/E U937 cells (B,D) and COSMC^−/−CHST1^O/E U937 cells (C,E). Each MS spectrum was normalized to its highest peak.

2.2. Probing the Dependence of Sulfated N-Glycans and Mucin-Type O-Glycans on Siglec Ligands.

In our previous study, we demonstrated that overexpression of CHST1 or CHST2 in U937 cells enhances Siglec interactions with cell surface glycans (Figure S4). To better understand the glycosylation pathways responsible for the biosynthesis of sulfated glycan ligands, we tested the ability of our previously developed Siglec-Fc proteins to bind a panel of the CHST1^O/E U937 cells by flow cytometry.³⁴ When either the complex N- or mucin-type O-glycosylation pathways were inhibited, both resulted in reduced Siglec-3 binding (Figure 3A). However, there was still significant binding of Siglec-3-Fc to the MGAT1^−/−COSMC^−/−CHST1^O/E cells, suggesting the potential existence of alternative glycosylation routes alongside the complex N- and mucin-type O-glycosylation. For Siglec-5, sulfated O-glycans appeared to contribute the most, although eliminating N-glycans still resulted in a small perturbation of Siglec-5 binding (Figure 3B). In line with prior reports,^13,14 the COSMC^−/−CHST1^O/E cells completely abolished the presentation of Siglec-7 ligands (Figure 3C). Here, N-glycans did not contribute and, in fact, the MGAT1^−/−CHST1^O/E cells showed higher levels of Siglec-7 ligands. For Siglec-8, the elimination of either MGAT1 or COSMC had a marked impact on Siglec-8 binding. Like Siglec-3, the MGAT1^−/−COSMC^−/−CHST1^O/E cells still displayed 10% of Siglec-8 ligands (Figure 3D). Siglec-15 showed perturbed binding in the absence of complex N- or mucin-type O-glycans, and 95% of the binding was abolished in the MGAT1^−/−COSMC^−/−CHST1^O/E cells (Figure 3E).

Figure 3.

Assessment of Siglec binding to the CHST1^O/E U937 cells lacking complex N- or mucin-type O-glycosylation pathways. CHST1 was transduced into the genetically modified cell lines: WT, MGAT1^−/−, COSMC^−/−, and DKO CHST1^O/E U937 cells. In turn, the binding of (A) Siglec-3, (B) Siglec-5, (C) Siglec-7, (D) Siglec-8, and (E) Siglec-15 was analyzed by flow cytometry. The gray histograms represent secondary staining only. Data are presented as mean ± SD. Statistical significance was calculated using one-way ANOVA with Tukey’s multiple comparisons test.

Siglec-2 and Siglec-9 prefer 6-O-sulfation on the underlying GlcNAc residue, which can be introduced by CHST2 overexpression (CHST2^O/E).¹³ Accordingly, we generated a parallel set of cells with the glycosylation mutants under CHST2^O/E. As anticipated from previous studies,^30,35 Siglec-2 binding was completely abolished in the MGAT1^−/−CHST2^O/E cells (Figure S5A), indicating that N-glycosylation is a major pathway for generating sulfated glycan ligands for Siglec-2. Likewise, robust N-glycan-dependent binding was observed in Siglec-9, while a 50% binding decrease occurred in the COSMC^−/−CHST2^O/E cells (Figure S5B). Mouse Siglec-E and Siglec-F were also tested against the CHST1^O/E cell lines with the glycosylation mutants. Like Siglec-7, the absence of COSMC led to the complete abrogation of Siglec-E binding (Figure S6A). As Siglec-E has been investigated extensively as a paralog of Siglec-9,³⁶ these results indicate distinct differences and demonstrate that the preference of mSiglec-E for sulfated glycans more closely matches that of Siglec-7. For mSiglec-F, a functional paralog of Siglec-8,³⁷ significant levels of mSiglec-F binding were maintained in the MGAT1^−/−COSMC^−/−CHST1^O/E cells (Figure S6B). Therefore, these results indicate that sulfated ligands of Siglec-3, Siglec-8, and mSiglec-F can be presented from a class of glycan that is neither N-glycans nor mucin-type O-glycans. A summary of the % binding data for N- or mucin-type O-glycosylation KO is presented in Table S1.

2.3. Residual Mucin-Type O-Glycans Do Not Contribute to Ligands for Siglec-3 and -8.

We hypothesized that the small fraction of remaining mucin-type O-glycans may maintain the binding to Siglec-3 and -8 as these glycans were not completely abolished in the COSMC^−/−CHST1^O/E cells (Figure 2C). To address this, we generated C1GALT1^−/− and MGAT1^−/−C1GALT1^−/− U937 cells because others have shown the complete abolishment of mucin-type O-glycans in the absence of C1GalT1.³⁸ After C1GalT1 was knocked out in the WT and MGAT1^−/− cells, CHST1 was virally transduced into the cells (Figure 4A). As expected, both C1GalT1^−/−CHST1^O/E and MGAT1^−/−C1GALT1^−/−CHST1^O/E cells resulted in the complete loss of Siglec-7 binding (Figure 4B). Next, we compared the binding of Siglec-3 and -8 to the COSMC^−/−CHST1^O/E, MGAT1^−/−COSMC^−/−CHST1^O/E, C1GALT1^−/−CHST1^O/E, and MGAT1^−/−C1GALT1^−/−CHST1^O/E cells (Figure 4C,D). Both MGAT1^−/−COSMC^−/−CHST1^O/E and MGAT1^−/−C1GALT1^−/−CHST1^O/E cells still presented sulfated glycan ligands for Siglec-3 and Siglec-8. These results confirm that there are other classes of glycans presenting sulfated ligands for Siglec-3 and Siglec-8 beyond N-glycans or mucin-type O-glycans.

Figure 4.

Assessment of COSMC^−/− and C1GALT1^−/− conditions in the presence or absence of MGAT1. (A) C1GALT1 in the WT and MGAT1^−/− U937 cells was knocked out using CRISPR/Cas9. Subsequently, CHST1 was transduced into the cells to assess the possibility of other glycosylation pathways generating sulfated glycan ligands for Siglec-3 and Siglec-8. These CHST1^O/E KO cells were screened with (B) Siglec-7, (C) Siglec-3, and (D) Siglec-8. The gray histograms represent secondary staining only. Data are presented as mean ± SD. Statistical significance was calculated using one-way ANOVA with Tukey’s multiple comparisons test.

2.4. KO of Key STs to Further Explore Sulfated Glycan Ligands of Siglecs.

Next, we investigated the requirement of three major sialyltransferases to generate Siglec ligands in U937 cell lines: ST6Gal1, ST3Gal1, and ST3Gal4.^14,15,39,40 To do so, we created the following three sets of cells: ST6GAL1^−/−, ST6GAL1^−/−ST3GAL1^−/− (ST DKO), and ST6GAL1^−/−ST3GAL1^−/−ST3GAL4^−/− (ST TKO) cells (Figure S7A). To validate these cell lines, we tested SNA, PNA, SLBR-N, and diCBM40. SNA showed binding dependent on the presence of ST6Gal1 (Figure S7B).⁴¹ The absence of ST3GAL1 expression contributed to the binding of PNA, and further increase in PNA binding was achieved in the additional ST3GAL4 knockout (KO) (Figure S7C). This might indicate ST3Gal4-mediated sialylation on O-glycans is unfavorable for PNA binding. The binding of SLBR-N and diCBM40 remained significant in the ST TKO cells compared to the COSMC^−/− cells (Figure S7D,E), suggesting that additional ST3Gal enzymes may be involved in the biosynthesis of sialylated mucin-type O-glycans for these two lectins.

The three ST KO U937 cell lines were transduced with a lentivirus carrying CHST1 or CHST2 to evaluate the effect of these sialyltransferases on Siglec binding. For CHST1^O/E cells, we observed that ST3Gal4 is a major sialyltransferase for generating sulfated glycan ligands for Siglec-3.^14,15 However, we also found that additional sialyltransferases can contribute to the biosynthesis of sulfated Siglec-3 ligands as the ST TKO condition still retained 10% of its binding (Figure 5A). Consistent with mucin-type O-glycosylation being the major glycosylation pathway for generating sulfated Siglec-5 ligands (Figure 3B), Siglec-5 ligands were completely abolished by the deletion of ST3GAL1 (Figure 5B). Siglec-7, a well-known O-glycan binder, retained 10% of its binding to the ST DKO CHST1^O/E cells,^14,40,42 but was further reduced in the cells lacking ST3GAL4 expression, suggesting that while ST3Gal1 is the primary sialyltransferase responsible for generating sulfated glycan ligands for Siglec-7, ST3Gal4 also plays a contributing role (Figure 5C). Consistent with a previous finding,¹⁵ ST3Gal4 generated most sulfated glycan ligands for Siglec-8. However, like Siglec-3, we observed the involvement of other sialyltransferases in the biosynthesis of sulfated glycan ligands for Siglec-8, contributing to the remaining 10% binding observed in the ST TKO condition (Figure 5D). Finally, Siglec-15 exhibited ST3Gal4-dependent binding (Figure 5E), which is reasonable given that ST3Gal4 acts on both N- and mucin-type O-glycan substrates,^14,43 and the data above indicate that both sulfated N-glycans and mucin-type O-glycans serve as ligands for Siglec-15 (Figure 3E).

Figure 5.

Evaluation of sialyltransferases involved in the biosynthesis of sulfated Siglec ligands.

Three panels of the ST6GAL1^−/−, ST6GAL1^−/−ST3GAL1^−/− (ST DKO), and ST6GAL1^−/−ST3GAL1^−/−ST3GAL4^−/− (ST TKO) U937 cells overexpressing CHST1 were generated. Subsequently, the effect of the absence of sialyltransferases was assessed using Siglec-Fc. The binding of (A) Siglec-3, (B) Siglec-5, (C) Siglec-7, (D) Siglec-8, and (E) Siglec-15 was analyzed by flow cytometry. The gray histograms represent secondary staining only. Data are presented as mean ± SD. Statistical significance was calculated using one-way ANOVA with Tukey’s multiple comparisons test.

For the CHST2^O/E cells, we tested the effects of these ST KO conditions on Siglec-2 and Siglec-9 binding. The absence of ST6GAL1 expression fully abrogated Siglec-2 interactions with the cells (Figure S8A), which is consistent with its enzymatic substrates being N-glycans.⁴⁴ In line with the previous findings demonstrating that ST3Gal4 is a major ST involved in the biosynthesis of Siglec-9 ligands,⁴⁰ Siglec-9 binding to the CHST2^O/E cells was greatly affected in the ST TKO CHST2^O/E cells (Figure S8B). For mSiglec-E, 95% binding loss was caused by the ST3GAL1 KO in the CHST1^O/E U937 cells, and the additional ST3GAL4 KO led to further loss similar to Siglec-7 (Figure S9A). Lastly, Siglec-F binding in CHST1^O/E cells was significantly reduced under ST6GAL1^−/− and ST DKO conditions, with the greatest loss observed in ST3GAL4^−/− cells. Of note, 10% of the binding remained under the additional ST3GAL4 KO similar to Siglec-3 or Siglec-8, which showed dramatic binding loss. This indicates that other sialyltransferases are involved in the sulfated ligand biosynthesis for Siglec-F (Figure S9B). Table S2 summarizes the percentage binding data for each Siglec under ST KO conditions.

2.5. Investigating Other Classes of Glycans that Can Present Sulfated Siglec Ligands.

Given that the binding of Siglec-3, Siglec-8, and mSiglec-F was mostly abrogated in DKO cells (Figure S10), this strongly supports the idea that carbohydrate sulfation on a class of glycan other than N-glycans or mucin-type O-glycans can serve as Siglec ligands. To investigate other classes of glycans involved in the biosynthesis of sulfated glycan ligands for Siglec-3 and Siglec-8, we explored the potential involvement of O-mannosylation, glycolipids, glycoRNA, and heparan sulfates. We knocked out UGCG and POMT1 in MGAT1^−/−COSMC^−/−U937 cells and introduced CHST1 to the resulting cell lines. In lectin assays with these newly created KO cell lines, we observed that diCBM40 and MAL-I showed a glycolipid-dependent binding loss (Figure S11). Siglec-3 exhibited a glycolipid-dependent increase in binding rather than a loss, whereas the binding of Siglec-8 and mSiglec-F was significantly reduced under the additional UGCG KO (Figures 6A,B and S12, Table S3). Consistent with these observations, pharmaceutical inhibition of glycolipid biosynthesis by Genz-123346 increased Siglec-3 binding but reduced Siglec-8 binding in the MGAT1^−/−COSMC^−/−CHST1^O/E cells (Figure S13). On the other hand, Siglec-8 binding was unaffected by Genz-123346 in WT CHST1^O/E cells, presumably because abundant sulfated N-glycans and mucin-type O-glycans dominated. These results support the idea that Siglec-8 and mSiglec-F may interact with glycolipids when CHST1 is present in the triple KO of MGAT1, COSMC, and UGCG.

Figure 6.

Investigation of multiple glycosylation pathways to evaluate their roles in Siglec ligand biosynthesis. (A,B) WT, MGAT1^−/−COSMC^−/− (DKO), DKO UGCG^−/−, and DKO UGCG^−/−POMT1^−/− cell lines overexpressing CHST1 were analyzed with (A) Siglec-3 and (B) Siglec-8. (C–F) The effect of RNase A on WT CHST1^O/E andDKO CHST1^O/E cells was assessed with (C) Siglec-3 and (D) Siglec-8. Binding in experimental groups was normalized to the binding level of the control, set to 100%. The gray histograms represent secondary staining only. Data are presented as mean ± SD. Statistical significance was calculated using Brown-Forsythe and Welch ANOVA tests (A,B) or using Student’s t-test (C,D).

Additionally, knocking out POMT1 on top of the MGAT1^−/−COSMC^−/−UGCG^−/− background marginally but significantly decreased Siglec-3, Siglec-8, and mSiglec-F binding in the presence of CHST1 expression (Figures 6A,B and S12, Table S3). These results suggest that sulfated O-mannose glycans can serve as ligands for Siglec-3, Siglec-8, and mSiglec-F. As there are two protein O-mannosyltransferases (POMT) enzymes (POMT1 and POMT2) that work together in the biosynthesis of O-mannosylation,⁴⁵ we used a complementary approach of disrupting the elongating enzyme, POMGNT1. Accordingly, we generated three MGAT1^−/−COSMC^−/−POMGNT1^−/−CHST1^O/E U937 clones. Although branched and sulfated O-mannose glycans were unexpectedly not completely abrogated in this triple KO (Figure S14), Siglec-3 and Siglec-8 binding levels were two-fold lower in these clones compared to the corresponding MGAT1^−/−COSMC^−/−CHST1^O/E cells (Figure S15). Additionally, we investigated CHO cells expressing CHST1, CHST6, and B3GNT7, with or without endogenous POMT1 expression (Figure S16). Consistent with sulfated O-mannosylation serving as Siglec ligands, Siglec-3 and Siglec-8 ligands were significantly higher in cells expressing POMT1. These findings suggest that O-mannose glycans can be involved in the biosynthesis of sulfated glycan ligands for Siglec-3, Siglec-8, and mSiglec-F.

To examine glycoRNA and GlcA-GlcNAc-containing heparan sulfates as possible carriers of sulfated glycan ligands for Siglec-3 and -8, we treated WT CHST1^O/E and MGAT1^−/−COSMC^−/−CHST1^O/E U937 cells with RNase A and heparin lyase, respectively. No significant decreases in the binding of Siglec-3 and -8 were observed upon treatment of cells with any of these enzymes (Figures 6C,D and S17). Taken together, these results rule out glycoRNA, and heparan sulfates as sulfated glycan ligands for Siglec-3 and -8 on U937 cells.

2.6. Sulfated Siglec Ligands Play a Role in cis Interactions.

U937 cells express endogenous Siglec-3, Siglec-7, Siglec-12, and Siglec-14 (Figure S18). To assess the class of glycan mediating cis interactions with Siglec-3 and Siglec-7 on U937 cells, we leveraged fluorescently labeled liposomes displaying a high-affinity and selective ligand specific for either Siglec-3 or Siglec-7.^46–48 Strong cis interactions are anticipated to “mask” the Siglec, leading to lower liposome binding (Figure 7A).^13,34 For Siglec-3, the MGAT1^−/−COSMC^−/−CHST1^O/E cells showed the highest liposome binding, suggesting that both sulfated N-glycans and mucin-type O-glycans mask Siglec-3 (Figure 7B). Of note, COSMC^−/− led to more liposome binding than MGAT1^−/−, indicating that sulfated O-glycans are major cis ligands for Siglec-3. CD33 expression levels were not significantly different between the cell lines (Figure 7C). For Siglec-7, unmasking was only observed in the COSMC^−/−CHST1^O/E cells (Figure 7D), despite equivalent expression levels (Figure 7E). These results solidify that sulfated O-glycans serve as the cis and trans ligands for Siglec-7.

Figure 7.

Liposome binding assay to assess the effect of MGAT1, COSMC, and ST KOs on cis interactions. (A) Two liposomes carrying high-affinity ligands for Siglec-3 or Siglec-7 were used to assess the effects of the glycosyltransferase or sialyltransferase KOs on the masked status of the endogenous Siglecs in CHST1^O/E cells. The liposome binding assay with the glycosylation pathway KO cells under CHST1 expression was measured on (B) Siglec-3, and (D) Siglec-7. Subsequently, the effects of the sialyl transferase KO CHST1^O/E cells on cis interactions were assessed on (F) Siglec-3, and (H) Siglec-7. The expression level of Siglecs per KO cell lines was measured by anti-Siglec-3 antibody and anti-Siglec-7 antibody (C,E,G,I). The gray histograms represent isotype controls. Data are presented as mean ± SD. Statistical significance was calculated using one-way ANOVA with Tukey’s multiple comparisons test.

In the ST KO cells, Siglec-3 became further masked in ST6Gal1^−/−CHST1^O/E cells, which could be due to enhanced α2–3 sialylation on N-glycans (Figure 7F). Notably, the additional ST3Gal1 KO resulted in an 8-fold unmasking of Siglec-3 and a further 3-fold increase in liposome binding was observed upon the additional deletion of ST3Gal4, without any changes in the expression level of Siglec-3 (Figure 7F,G). Given the strong impact of ST3Gal1 loss on the trans ligands of Siglec-7 (Figure 5C), it was surprising that Siglec-7 remained masked in the ST6GAL1^−/−ST3GAL1^−/− CHST1^O/E cells. However, the additional deletion of ST3GAL4 caused unmasking with no alteration in the expression level of Siglec-7 (Figure 7H,I). Since ST3Gal4 accounts for only 10% of the sulfated trans ligands for Siglec-7, this suggests that these remaining sulfated glycan ligands are sufficient to mask Siglec-7. Alternatively, ST3Gal4-mediated sulfated glycans (likely core 2 O-glycans) may serve as the major cis ligands for Siglec-7. Overall, these findings indicate that the major trans ligands of a Siglec do not necessarily predict its major cis ligands.

3. DISCUSSION

Recent chemoenzymatic synthesis of sulfated sialoglycans provides great tools to evaluate the interactions of Siglecs with sulfated ligands, which were previously studied using genetic approaches.^{10,12–14,17,49,50} These synthetic approaches complement genetic methods by allowing for precise control over glycan structures, enabling detailed analysis of branch- and glycan class-dependent binding to sulfated structures by certain Siglecs. While chemoenzymatic methods are advancing rapidly, assembling the full repertoire of possible sulfated glycans on the cell surface remains a complex challenge. Genetic manipulation of glycans on cells, as pioneered by Clausen and colleagues,⁵¹ has yielded valuable insights into Siglec-ligand interactions, but validation with synthetic ligands is essential to confirm that these interactions are functionally relevant. Combining glycosylation-deficient cells with the expression of CHSTs has revealed novel sulfate-dependent interactions in previous studies and in our current work,¹⁴ underscoring the importance of integrating both genetic and synthetic approaches to fully understand Siglec binding mechanisms. Figure 8 summarizes our findings for cis or trans ligands for Siglecs and will ultimately be tested once a larger repertoire of synthetic sulfated glycans is available. Notably, the glycosylation machinery can vary across cells and tissues, influencing the classes of sulfated glycans that are available to serve as Siglec ligands.^13,14

Figure 8.

Summary of classes of sulfated glycan ligands for Siglecs. Types of sulfated glycans for each Siglec on CHST1^O/E U937 cells are illustrated, along with information about the corresponding STs and the percentage of binding for each glycan type on U937 cells. A question mark indicates that further validation may be needed. The contribution of sialyltransferases to cis or trans interactions is noted. This figure does not provide evidence to determine whether sialyltransferases preferentially act on sulfated glycans or if sulfotransferases prefer sialylated substrates.

Siglec-2 was one of the first Siglecs identified to prefer sulfated glycans, showing enhanced binding to sulfation at the 6-O-GlcNAc position within the 6′SLN moiety.^12,28,52 Siglec-2 recognizes α2–6-linked sialic acid presented from N-glycans, a finding supported by loss of Siglec-2 ligands on cells lacking complex N-glycan biosynthesis or ST6Gal1.^14,53,54 Our current study, as well as that of Narimatsu and colleagues,¹⁴ clearly demonstrated that Siglec-2 retains specificity for N-glycan, even in the presence of abundant 6-O-GlcNAc sulfation. Other sulfotransferases that install a 6-O-GlcNAc sulfate group (e.g., CHST5 and CHST6) may also contribute to producing sulfated N-glycan ligands for Siglec-2.³⁰

Previous studies demonstrated that Siglec-3 binds N-glycans in a ST3Gal4-dependent manner in CHST1^O/E HEK293 cells.¹⁴ Our study, using U937 cells, agrees with these previous findings that sulfated N-glycans are a major trans ligand for Siglec-3. Interestingly, we found that sulfated mucin-type O-glycans act as the dominant cis ligands. This is further supported by the observation that ST3Gal1^−/− resulted in an unmasking effect on Siglec-3. In addition, sulfated O-mannose glycans were identified as trans ligands of Siglec-3, which is in line with a previous glycan microarray study showing that O-mannose glycans, even in the absence of sulfation, mediate binding to Siglec-3.⁵⁰ Given the abundance of O-mannosylated glycans and CHST1 in the brain, along with the genetic link between Siglec-3 and Alzheimer’s disease,^55–57 further exploration into the role of these ligands in microglia function is warranted. Interestingly, even in cells lacking most glycan types, residual Siglec-3 binding was observed, suggesting that other glycan types may also serve as ligands. One possibility could be O-fucosylated glycans with sialylation, as the mRNA expression of POFUT1 in U937 cells is higher than that of POMT1 or POMT2 based on our RNA-seq data.⁵⁸ Since the ST6Gal1^−/−ST3Gal1^−/−ST3Gal4^−/−CHST1^O/E cells still retain 10% of Siglec-3 binding and ST3Gal6 is highly expressed at the mRNA level, it is worth investigating the role of ST3Gal6 in the biosynthesis of sulfated glycan ligands for Siglec-3. Previously, we demonstrated that 6,6′-S,S-3′SLN is a more potent ligand for Siglec-3 than its mono- or nonsulfated counterparts.¹³ Ma et al. also reported that 6,6′-S,S-3′SLN yields the highest signal in a glycan microarray.²⁸ Consistent with previous findings,¹³ heparan sulfates, which lack sialic acids, do not function as Siglec-3 ligands because Siglec binding was entirely sialic acid-dependent.^13,14 Additionally, keratan sulfate (KS) glycosaminoglycans were identified as very good ligands for Siglec-3 and Siglec-8 by Schnaar and coworkers.^15,17,18,59 KS consists of repeated LacNAc units that are highly sulfated and contain terminal sialylation, extended from N-glycans (KS-I), mucin-type O-glycans (KS-II), and O-mannose glycan (KS-III). Further studies on sialylated KS could offer new insights into Siglec functions.

Siglec-5 and Siglec-14 are evolutionarily paired receptors that share identical glycan-binding domains and, thus, have similar ligand profiles.⁶⁰ Our previous research demonstrated that CHST1-mediated sulfation on cell surface glycans can enhance the interactions between Siglec-5 and its glycan ligands.¹³ Our current work demonstrates that Siglec-5 primarily recognizes sulfated mucin-type O-glycans. Moreover, ST3Gal1 was identified as a crucial sialyltransferase involved in the biosynthesis of sulfated glycan ligands for both Siglec-5 and Siglec-14, acting specifically on O-glycans. Further studies are needed to determine if Siglec-5 and Siglec-14 recognize sulfated glycans on mucins in primary cells. A successful approach in probing the mucin-type O-glycan specificity of Siglec-7 and -9 has been the use of mucin-specific proteases,⁶¹ which also could be used to address Siglec-5.

COSMC was identified as an essential gene for Siglec-7 ligand biosynthesis.⁴² This was demonstrated by genome-wide CRISPR and further confirmed in COSMC^−/−CHST1^O/E HEK293 cells, where Siglec-7 binding was completely abolished.^14,42 Siglec-7 is reported to have specificity for disialyl T antigen-containing mucins. Key sialyltransferases for its ligands include ST3Gal1/2 and ST6GalNAc2/3/4.^14,40,62,63 In our study, deletion of ST3Gal1 had the most dramatic impact on Siglec-7 trans ligands. Additionally, we found that ST3Gal4 contributes to approximately 10% of the sulfated trans ligands for Siglec-7. This suggests that Siglec-7 may recognize sulfated core-2 mucin-type O-glycans.⁶⁴ In this context, it is important to note that our MS analysis revealed that sulfated core-1 structures are abundant on U937 cells, although there are also core-2 glycans present in notable abundance. In contrast to the situation for trans ligands, Siglec-7 remained masked in the ST6Gal1^−/−ST3Gal1^−/− cells but was unmasked in the ST6Gal1^−/−ST3Gal1^−/−ST3Gal4^−/− cells. This suggests two possibilities: either ST3Gal4 complements ST3Gal1 in generating sulfated core-1 O-glycans as cis ligands for Siglec-7, or a mucin-type O-glycan other than a sialylated core-1 structure is a key ST3Gal3-dependent cis ligand for Siglec-7. This highlights that cis ligands for Siglec-7 likely differ from its trans ligands as observed for Siglec-3. In a glycan microarray, Ma et al. reported that Neu5Acα2–8Neu5Acα2–3(6S)Galβ1–4GlcNAc ranked highest.²⁸ This indicates that both α2–8 sialylation and 6-O-Gal sulfation have an additive effect on Siglec-7 and glycan interaction. Sulfated sialyl T antigen also ranked highly as a Siglec-7 ligand, though to a lesser extent than other ligands containing both sulfation and α 2–8 sialylation, it was stronger than its nonsulfated counterparts. Sulfated ligands will be crucial for further delineation of Siglec-7 ligand specificity. However, it is important to note that glycan microarrays do not replicate the natural high density of ligands presented by mucins. Thus, soluble mucin scaffolds will be an essential tool in fully defining Siglec-7 ligand preferences.⁶⁵

We found that Siglec-8 primarily recognizes both sulfated N- and mucin-type O-glycans, with additional contributions from O-mannose glycans as observed for Siglec-3. Notably, genetic and pharmacological approaches also demonstrated that Siglec-8 and mSiglec-F bind to glycolipids. The murine paralog of Siglec-8, Siglec-F, exhibits similar binding characteristics, including some glycolipid-dependent binding. It is notable that Siglec-F was shown to bind synthetically prepared disulfated GM1b, suggesting that sulfated glycolipids may also serve as ligands for Siglec-8 and Siglec-F.^27,66–68 Earlier studies identified sulfated glycolipids in human and rat kidneys, further supporting a potential biological significance, although this requires further exploration.^69,70 The major differences between Siglec-8 and mSiglec-F is that mSiglec-F recognizes nonsulfated glycan ligands better than Siglec-8.⁷¹ In a glycan microarray, Ma et al. discovered that Siglec-8 favored 6,6′-S,S-3′SLN, indicating additive effects of 6-O-Gal and 6-O-GlcNAc sulfation.²⁸ This finding aligns with earlier biochemical analyses of Siglec-8.¹¹

Previously, 6-S-SLe^X was identified as a potent epitope for Siglec-9.^10,27 In HEK293 cells, ST3Gal4 and, to a lesser extent, ST3Gal6 were found to contribute to the biosynthesis of Siglec-9 ligands.¹⁴ In our previous study, we observed enhanced Siglec-9 ligands upon CHST2 or CHST4 expression in U937 cells.¹³ Interestingly, while a marginal reduction in Siglec-9 ligand expression was observed in COSMC^−/−CHST2^O/E cells, knocking out MGAT1 had a more pronounced impact on Siglec-9 binding. This observation aligns with Siglec-9’s preference for the 6-S-3′SLN moiety on the α1–6 branch of N-glycans with repeated LacNAc units, while showing minimal recognition of O-glycans.¹⁷ Ma et al. reported that sialyl T antigen with 6-O-GalNAc sulfation ranked highest in a glycan microarray, indicating that sulfated O-glycans can serve as Siglec-9 ligands. These findings suggest that Siglec-9 interacts with sulfated N- and O-glycans, with the dominant glycan class being depending on the specific cell or tissue type. Further analysis revealed that Siglec-9 favors 6-S-3′ SLN moieties.²⁸ In the ST6Gal1^−/−ST3Gal1^−/−ST3Gal4^−/−CHST2^O/E cells, we observed that 30% of Siglec-9 binding was retained. Based on our RNA-seq data, this residual binding may be attributed to highly expressed ST3Gal6, which has been demonstrated to play a role in Siglec-9 ligand biosynthesis.¹⁴

Siglec-15 is highly conserved between humans and mice, with both species exhibiting nearly identical ligand-binding profiles. ST6GalNAc1 was shown to be important for biosynthesis of Siglec-15 ligands.¹⁴ While ST6GalNAc1 can create thesialyl Tn antigen and the disaccharide was originally proposed to be a preferred ligand for Siglec-15, several recent studies have not reached the same conclusion. Instead, these studies found that Siglec-15 has specificity for a Neu5Ac α2–6 linked to an underlying GlcNAc.^72,73 It was previously shown that COSMC^−/−CHST1^O/E HEK293 cells completely lose their ability to express Siglec-15 ligands, while only minor reductions in binding are observed in MGAT1^−/− and GCNT1^−/−CHST1^O/E HEK293 cells.¹⁴ In U937 cells, both sulfated N-glycans and mucin-type O-glycans contribute equally to Siglec-15 binding. Complete binding loss occurs in MGAT1^−/−COSMC^−/−CHST1^O/E cells. ST3Gal4 was identified as a key enzyme in the creation of sulfated N- and O-glycan ligands for Siglec-15. Given that Siglec-15 binding was reduced by 20–30% in GCNT1^−/−CHST1^O/E HEK293 cells,¹⁴ further investigation using synthetic sulfated glycans is needed to determine if Siglec-15 prefers sulfated core 2 O-glycans.

In conclusion, our study highlights how carbohydrate sulfation enhances the potency of glycan ligands for Siglecs, likely expanding the range of glycans that are otherwise inactive without sulfation. CHST1, expressed in specific regions such as the brain, cartilage, and cornea—areas where inflammation is tightly regulated—appears to play a role in modulating immune responses. Studies in CHST1^−/− mice showed that the absence of CHST1 leads to an increased accumulation of airway macrophages and lymphocytes, suggesting that CHST1 may regulate immune cell functions through glycan modification.⁷¹ Future research should focus on how carbohydrate sulfation on sialoglycans influences Siglec function.

4. METHODS

4.1. CRISPR/Cas9.

CRISPR/Cas9-mediated deletion of MGAT1, COSMC, C1GalT1, POMT1, POMGNT1, and UGCG genes in U937 cells was carried out as described previously except that crRNA with the following sequences were used: MGAT1 (CCGGACCACCGCAAGTTCCA), COSMC (ATGCTAGGACACATTAGGAT), C1GALT1 (ATCCTATTGCTGATCCACAG), POMT1 (CCTGTAGTGGTGACGGCTGACAT), POMGNT1 (TTGAATACACCTCTACGTCC), and UGCG (CCGATTACACCTCAACAAGA).⁷⁴ After 2 weeks, the single-cell sorted cells became confluent, and we selected the CRISPR/Cas9 KO clones using Siglec-1-Fc for UGCG^−/−, Siglec-2-Fc for MGAT^−/−, and Siglec-7-Fc for COSMC^−/− or C1GALT1^−/−.^35,42 MGAT1^−/−COSMC^−/− cells were obtained by knocking out COSMC on top of MGAT1^−/−. After screening, clones that lost Siglec-1-Fc, Siglec-2-Fc or Siglec-7-Fc binding were further sequenced and validated through Sanger sequencing. For POMGNT1^−/−, clones were selected based on sequencing results.

4.2. Viral Transduction.

cDNA encoding CHST1 and CHST2 was inserted into RP172 or RP173 lentiviral vector as described previously.¹³ Briefly, the mixture containing RP172, RP18, RP19, and TransIT-LT1 Transfection Reagent (Mirus Bio) was transfected into HEK293T cells and cultured for 3 days. The cultured medium was then mixed with Lenti-X Concentrator (Takara Bio) in a 3:1 ratio and incubated for 10 min at 4 °C. After centrifugation at 1500 rcf at 4 °C for 45 min, pellets containing viruses were resuspended in PBS and aliquoted into 20 μL for transduction and stored at −80 °C until use. To generate CHST1 or CHST2 overexpressing cell lines, viruses carrying these genes were added to 200,000 U937 cells and cultured for 3 days. After 3 days, the transduced cells were cultured with RPMI medium containing 1% penicillin/streptomycin, 10% fetal bovine serum, and 3.33 μg/mL Zeocin for selection.

4.3. Staining Methods for Flow Cytometry.

U937 cells in a U-bottom 96-well plate were washed once with PBS. Siglec-Fc was used at approximately 12 μg/mL, and Goat anti-hIgG with AF647 (polyclonal, 0.4 μg/mL, ThermoFisher) was mixed in a 1:1 volume ratio immediately before staining. Then, 100 μL of the mixture was used for staining each well. After incubation on ice for 15 min, the stained cells were washed once with PBS, resuspended in PBS, and analyzed using a Fortessa X-20 flow cytometer (BD Biosciences). Flow cytometric data were analyzed using FlowJo v10 software.

4.4. Liposome Binding Assay.

After counting the cells, 200,000 cells were added to each well of a U-bottom 96-well plate and washed once with PBS. Subsequently, naked liposomes and liposomes carrying synthetic ligands for Siglec-3 and Siglec-7 were resuspended in PBS containing 10% FBS (v/v) to create a 50 μM reaction solution. The cells in each well were incubated with 50 μL of the liposome solution at 4 °C for 30 min. After incubation, the cells were washed once with PBS and analyzed using flow cytometry. Signals from the liposomes carrying Siglec ligands were subtracted from the signals of the naked liposomes.

4.5. Siglec-Fc Expression and Purification.

All Siglec-Fc proteins used in this study were expressed in WT CHO cells, as previously described.^34,75

4.6. Pharmacological Manipulation of Glycolipids.

Genz-123346 (Toronto Research Chemicals) were stored at −20 °C in cell-grade DMSO. Genz-123346 was added to complete media (RPMI supplemented with 10% FBS (v/v), 100 U/mL penicillin, and 100 μg/mL streptomycin) to a final concentration of 5 μM. Media containing the inhibitor was then added to cells (4.5 × 10⁵) in a six-well dish for 72 h. The cells were subsequently prepared for Siglec-Fc staining as described elsewhere.

4.7. RNaseA Digestion.

RNase A solution (Thermofisher Scientific) was diluted to 2 μM in PBS. The diluted RNase A solution was then added to cells and the mixture was incubated at RT for 5 min⁷⁶ After treatment, the cells were washed with PBS and subsequently prepared for Siglec-Fc staining as described elsewhere.

4.8. Statistical Analysis.

All statistical analyses in this study were performed using GraphPad Prism version 7 software. For experiments comparing two groups, an unpaired two-tailed Student’s t-test was used to assess statistical significance. For data sets with more than two groups, one-way ANOVA followed by Tukey’s post hoc test, as well as Brown-Forsythe and Welch ANOVA tests, were conducted.

4.9. Glycan Release and Permethylation.

Proteins were extracted from harvested U937 cells by sonication in 20 mM Tris–HCl buffer containing 0.1% Triton X-100 and 150 mM NaCl, reduced and alkylated by 10 mM of dithiothreitol and 50 mM of iodoacetamide, respectively, for 1 h at 37 °C, and then precipitated in 10% trichloroacetic acid for 30 min at −20 °C. Protein pellets were washed in cold acetone to remove detergent and then reconstituted in 50 mM ammonium bicarbonate buffer for overnight trypsin digestion at 37 °C, followed by 6 h of chymotrypsin digestion at 37 °C. O-glycans were released from de-N-glycosylated glycopeptides by reductive elimination in 1 M NaBH₄/100 mM NaOH at 37 °C for 3 days. Reaction mixtures were neutralized and loaded onto a C18 Sep-Pak cartridge (Waters Corporation, Milford, MA) prepacked with 1 mL of activated Dowex 50 W-X8 resin, and the reduced O-glycans were eluted in 5% acetic acid. After removal of borates by repeated coevaporation with 10% acetic acid in methanol, the O-glycan samples were permethylated and then fractionated using a Waters OASIS MAX cartridge into neutral, monosulfated, and multiply sulfated glycan pools, as described previously.⁷⁷

4.10. MS Data Acquisition and Analysis.

Aliquots of the permethylated sulfated glycan samples were mixed 1:1 with 10 mg mL⁻¹ 3,4-diaminobenzophenone matrix in 50% acetonitrile and spotted onto the MALDI plate for data acquisition on an AB SCIEX MALDI TOF/TOF 5800 system in negative mode. For nanoLC-MS/MS analysis, samples were dissolved in 10% acetonitrile, applied to an Ultimate 3000 RSLC system connected to an Orbitrap Fusion Tribrid Mass Spectrometer (Thermo Fisher Scientific, Waltham, MA) via a PicoView nanosprayer (New Objective, Woburn, MA), and separated using a constant flow rate of 300 nL/min at 50 °C on a ReproSil-Pur 120 C18-AQ column (120 Å, 1.9 μm, 200 × 0.075 mm; Dr. Maisch HPLC GmbH, Ammerbuch, Germany). The solvent system used was buffer A (0.1% formic acid in 100% H₂O) and buffer B (0.1% formic acid in 100% acetonitrile), with a 70 min linear gradient of 30–80% B. Data-dependent acquisition in negative mode was performed with full MS and HCD MS² (stepped collision energy at 50 ± 10% or 55 ± 10%) in the Orbitrap at 120,000 and 30,000 resolutions, respectively. All MS and MS/MS data were manually assigned, and overlaid extracted ion chromatograms of select O-glycans of interest were plotted at 5 ppm at a fixed intensity scale for an approximate visualization of their relative abundance.

4.11. U937 Cell Line RNA Isolation and Bulk RNA-Sequencing.

Human monocytic cell line, U937 (CRL-1593.2; ATCC), was cultured in RPMI 1640 medium (Gibco) supplemented with 10% heat-inactivated fetal bovine serum (Gibco) and 1× penicillin/streptomycin (Gibco) at maintained at 37 °C in 5% CO₂ incubator. At confluency, U937 cells were centrifuged at 300g for 5 min and washed with 1× sterile DPBS. Total RNA was extracted from 1 million cells per replicate using Purelink RNA Mini Kit (Invitrogen), following the manufacturer’s protocol. Isolated RNA was stored in −80 °C freezer prior to use. cDNA libraries were prepared using Illumina TruSeq RNA Library Prep Kit v2. Concentration and quality of cDNA libraries were then measured using Qubit Flex (Invitrogen) and 2100 Bioanalyzer (Agilent), respectively. Paired-end 150 bp sequencing data were generated using an Illumina HiSeq platform (Novogene). For transcriptome analysis, the quality of demultiplexed raw sequencing data were validated using FastQC v0.11.2 followed by removal of Illumina adapter using CutAdapt v3.3. Adapter-trimmed sequencing reads were then aligned to the reference GRCh38/hg38 human genome assembly using STAR v2.7.0. Mapped sequencing reads were counted and assigned to gene transcripts using HT-seq count and Ensembl version 81 annotation file. The transcript per kilobase million (TPM) values were obtained for all replicates, and principal splice transcript isoforms (APPRIS annotation) were selected for plotting mRNA expression levels of glycosyltransferases.

Supplementary Material

ASSOCIATED CONTENT

Supporting Information

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acschembio.4c00677.

Figures, Tables, MALDI-MS Data, and RNA-seq Data (PDF)

ABBREVIATIONS

CHST

carbohydrate sulfotransferase

Siglec

sialic acid-binding immunoglobulin-type lectins

ST

sialyltransferase

DKO

double knockout

TKO

triple knockout

CHST1 ^O/E

CHST1 overexpression

CRISPR

clustered regularly interspaced short palindromic repeats

SNA

Sambucus nigra lectin

MAL-II

Maackia Amurensis Lectin II

SLBR

Siglec-like binding region