Terminal Neu5Ac decoration of tumor-cell glycans engages inhibitory Siglec receptors on NK cells and macrophages, converting a "glycan shield" into an active immune checkpoint
Malignant transformation reprograms the Golgi glycosylation apparatus long before any protein-coding driver mutation becomes clinically evident. Sialyltransferases — the terminal-capping enzymes of the N- and O-glycan and glycosphingolipid biosynthetic pathways — are transcriptionally amplified by the same oncogenic circuitry that drives proliferation, producing a systemic shift toward hypersialylated surface glycoconjugates that will later be exploited as an immune checkpoint.
Sialic acid (N-acetylneuraminic acid, Neu5Ac) is synthesized in the cytosol from UDP-GlcNAc via the bifunctional enzyme GNE/MNK (UDP-GlcNAc 2-epimerase/ManNAc kinase), condensed to Neu5Ac-9-phosphate by NANS, dephosphorylated by NANP, then transported into the nucleus where CMP-sialic acid synthetase (CMAS) activates it to CMP-Neu5Ac — the obligate high-energy donor substrate for every sialyltransferase reaction. CMP-Neu5Ac is exported to the Golgi lumen by the antiporter SLC35A1 in exchange for CMP, where it becomes available to the resident sialyltransferase family.
Twenty distinct human sialyltransferase genes, organized into four subfamilies by linkage specificity, populate the trans-Golgi:
• ST3GAL1/2/3/4/6 — transfer Neu5Ac in α2,3 linkage to galactose, generating the core disaccharide motif Neu5Acα2,3Galβ1,3/4GlcNAc found on O-glycans (ST3GAL1 on core-1 mucin-type O-glycans) and N-glycan antennae. • ST6GAL1/2 — transfer Neu5Ac in α2,6 linkage to galactose on N-glycans; ST6GAL1 is the single most consistently overexpressed sialyltransferase across carcinomas and directly caps the Fc and Fab glycans of IgG-like domains, growth factor receptors (EGFR, TGF-β receptor), and death receptors (Fas/CD95), blunting apoptotic signaling. • ST6GALNAc1-6 — cap GalNAc-type O-glycans, generating the sialyl-Tn (STn) antigen, a classical pan-carcinoma marker recognized by antibody theranostics. • ST8SIA1-6 — add Neu5Ac in α2,8 linkage to a pre-existing sialic acid, building polysialic acid (polySia, up to >100 residues) on NCAM in neuroblastoma and small-cell lung cancer, and capping ganglioside GD3 to GD2 and GD1 series in melanoma and neuroblastoma.
Transcriptional control converges on the same oncogenic signaling nodes that drive tumor proliferation. Activated KRAS/BRAF-MAPK signaling increases ST6GAL1 promoter activity through ETS-family transcription factors; hypoxic tumor cores stabilize HIF-1α, which directly transactivates ST3GAL1 and ST3GAL4; and NF-κB, chronically active in inflamed tumor microenvironments, upregulates ST6GALNAc2. The net biochemical consequence is a several-fold increase in flux through the Golgi sialylation pathway well before any single glycoprotein's expression level changes — hypersialylation is a metabolic and enzymatic phenotype layered on top of the existing proteome, and mass-spectrometry glycomics of tumor versus adjacent normal tissue consistently shows a 2–8 fold enrichment of α2,6-sialylated N-glycan structures.
Once CMP-Neu5Ac flux and sialyltransferase expression are elevated, terminal capping proceeds across essentially every class of surface glycoconjugate simultaneously — N-glycans, mucin-type O-glycans, and glycosphingolipids — producing a physically thickened, negatively charged glycocalyx that both masks underlying antigenic epitopes and mechanically resists immune cell approach.
Sialic acid occupies the non-reducing terminus of essentially every mature complex-type glycan, and because it carries a carboxylate group with a pKa near 2.6, every added Neu5Ac residue contributes a fixed negative charge at physiological pH. Dense hypersialylation therefore does two things mechanically: it caps and masks the penultimate galactose and GalNAc residues that would otherwise be recognized by galectins, selectins in their unliganded state, and antigen-presenting receptors; and it builds a highly hydrated, electrostatically repulsive polyanionic brush that extends the effective glycocalyx thickness.
Atomic force microscopy measurements of the pericellular glycocalyx (Paszek/Weaver-style methodology, cell-surface AFM with functionalized cantilevers) show that many carcinoma cell lines present a glycocalyx 5–20 times thicker than matched non-transformed epithelial controls, driven substantially by hypersialylated, hyperextended mucins such as MUC1 and the sialomucin CD43. These bottlebrush glycoproteins have a proteinaceous core densely O-glycosylated at every second or third residue; when those O-glycans are further capped with sialic acid, the mucin domain becomes maximally extended by charge repulsion between adjacent sugar side chains, functioning as a physical spacer that keeps immune effector cells and complement components at a distance exceeding the reach of many activating receptor–ligand pairs (typical immunological synapse gaps of 13–15 nm).
Sialylation is also converted into an active adhesion signal for metastatic dissemination through the biosynthesis of sialyl-Lewis X (sLeX, Neu5Acα2,3Galβ1,4[Fucα1,3]GlcNAc) and sialyl-Lewis A (sLeA) — the canonical ligands for the selectin family (E-, P-, L-selectin) expressed on activated endothelium and platelets. Tumor cells displaying elevated sLeX/sLeA density adhere transiently to selectin-coated vasculature during hematogenous spread, a step that correlates clinically with the serum biomarkers CA19-9 (sLeA-based, pancreatic/colorectal) and CA15-3-associated epitopes. Underneath this shield, tumor-associated antigens, MHC class I peptide complexes, and stress ligands (MICA/MICB) remain present but sterically inaccessible — setting up the receptor engagement events of the next stage rather than eliminating recognition outright.
Sialic acid capping would be a purely mechanical decoy strategy were it not for a dedicated family of immune inhibitory receptors — the sialic-acid-binding immunoglobulin-type lectins, or Siglecs — expressed on NK cells, macrophages, dendritic cells, and subsets of T and B cells, which read terminal Neu5Ac directly as a "self" signal and actively suppress the effector cells that carry them.
Siglecs are type-I transmembrane proteins of the immunoglobulin superfamily, built from an N-terminal V-set Ig domain that directly contacts sialic acid, followed by a variable number of C2-set Ig domains, a single transmembrane helix, and — for the inhibitory subclass — a cytoplasmic tail bearing one canonical immunoreceptor tyrosine-based inhibitory motif (ITIM, consensus (I/V/L)xYxx(L/V)) plus a second ITIM-like motif. Fourteen Siglecs are expressed in humans; most carry inhibitory tails (Siglec-2/CD22, -3/CD33, -5, -6, -7, -8, -9, -10, -11), while a smaller activating subset (Siglec-14, -15, -16) pairs with the ITAM-bearing adaptor DAP12.
The biophysics of Siglec engagement is a textbook case of avidity rescuing a weak monovalent interaction. A single Siglec V-set domain binds free Neu5Ac or a short sialyl-lactosamine trisaccharide with a dissociation constant in the low millimolar range (Kd ≈ 1–3 mM) — far too weak to drive stable signaling on its own. On the intact cell surface, however, the glycocalyx presents Neu5Ac at extremely high local density (hundreds of thousands of terminal sialic acids per cell), and Siglecs themselves cluster within cholesterol-rich membrane microdomains and can pre-associate in cis with sialylated ligands on their own cell surface, held in an autoinhibited "masked" state until a higher-avidity trans ligand on an opposing tumor cell membrane displaces the cis interaction. This multivalent, cooperative binding regime converts the millimolar monovalent affinity into an effective avidity in the sub-micromolar to high-nanomolar range at the immunological synapse — strong enough to sustain a productive inhibitory signal for the duration of an NK-tumor or macrophage-tumor contact.
Ligand specificity varies meaningfully between family members and shapes which tumors most exploit which receptor: Siglec-7 preferentially recognizes α2,8-linked disialic and branched gangliosides (disialyl motifs on GD3/GD2, explaining its relevance in neuroblastoma and melanoma), Siglec-9 recognizes both α2,3- and α2,6-linked sialyl-lactosamine broadly displayed on mucins and is the dominant NK/macrophage checkpoint in epithelial carcinomas (lung, breast, colorectal), and Siglec-10 on macrophages and B cells recognizes CD24 — a heavily O-glycosylated, GPI-anchored sialoglycoprotein overexpressed on ovarian and breast tumor cells — constituting a distinct "don't eat me" axis parallel to, and independent of, the CD47–SIRPα checkpoint.
Siglec engagement is not a passive block on adhesion — it is an active signal-transduction event. Phosphorylation of the ITIM tyrosines recruits the tandem SH2-domain phosphatases SHP-1 and SHP-2, which physically dephosphorylate the very activating receptor cascades that NK cells and macrophages depend on to kill, engulf, or present antigen from a tumor target, producing a net immunosuppressive signal that dominates over concurrent activating input.
Upon Siglec clustering at the immune synapse, Src-family kinases resident in the same membrane microdomain phosphorylate the tyrosines within the cytoplasmic ITIM and ITIM-like motifs. Phosphotyrosine creates a docking site for the tandem SH2 domains of SHP-1 (encoded by PTPN6) and, for several Siglecs including Siglec-9, SHP-2 (PTPN11). Recruitment relieves the phosphatases' auto-inhibited conformation and positions their catalytic domains directly adjacent to the activating receptor signalosome assembled a few nanometers away in the same synapse.
In NK cells, the physiological targets are the phosphotyrosine residues generated downstream of activating receptors NKG2D, DNAM-1 (CD226), and the DAP12/DAP10-coupled natural cytotoxicity receptors (NKp30, NKp46). SHP-1/2 dephosphorylate Vav1, PLCγ2, and components of the SLP-76/LAT-like scaffold that normally couple activating receptor ligation to actin remodeling, MTOC (microtubule-organizing center) polarization toward the target, and lytic granule (perforin/granzyme B) release. Flow-cytometric CD107a (LAMP-1) surface mobilization assays — the standard functional readout of NK degranulation — show reductions of 70–80% when Siglec-7 or Siglec-9 are co-engaged alongside an otherwise activating tumor target, even when NKG2D ligands (MICA/MICB) remain fully expressed; the glycan checkpoint dominates over the activating signal rather than merely diluting it.
In macrophages, the parallel consequence is suppression of Fcγ-receptor- and complement-receptor-mediated phagocytosis, together with reduced pro-inflammatory cytokine output (TNF-α, IL-12) and a shift toward an M2-like, tissue-remodeling transcriptional program. Because Siglec-10/CD24 operates through the same SHP-1/2 mechanism used by the CD47–SIRPα "don't eat me" checkpoint but engages an entirely distinct ligand-receptor pair, tumors frequently upregulate both axes in parallel, and dual blockade produces synergistic phagocytic activation in preclinical xenograft models beyond either antibody alone.
A 2019 study from the Bertozzi and Weissman laboratories (Barkal et al., Nature Immunology) showed that CD24 functions as a dominant "don't eat me" signal on ovarian and triple-negative breast cancer cells acting through macrophage Siglec-10, independent of CD47. Combined anti-CD24 antibody blockade with macrophage-activating stimuli restored phagocytic clearance of tumor cells in vitro and reduced tumor burden in syngeneic mouse models — establishing sialoglycan-Siglec engagement as a therapeutically actionable checkpoint distinct from the CD47 axis already in clinical development.
Because hypersialylation is enzymatically installed and the Siglec checkpoint depends on avidity built from dense terminal Neu5Ac, the glycan shield is pharmacologically reversible from three independent angles: physically stripping the sugar with a targeted sialidase, blocking the Siglec receptor itself with an antagonist antibody, or starving the biosynthetic pathway with a sialyltransferase inhibitor — each now in active preclinical or early clinical development.
Targeted sialidase conjugates fuse a catalytically active bacterial or viral neuraminidase domain to a tumor-targeting antibody Fab or scFv, restricting desialylation to antigen-positive cells and sparing systemic glycoproteins from indiscriminate cleavage — an important design constraint, since untargeted sialidase would strip sialic acid from erythrocytes, platelets, and circulating IgG with severe on-target/off-tumor toxicity. Palleon Pharmaceuticals' lead molecule E-602 (also referenced as PALG-2201) links a sialidase catalytic domain to a HER2- or tumor-antigen-targeting antibody scaffold and entered Phase 1 dose-escalation trials in advanced solid tumors in the 2023–2024 window, with correlative biomarkers tracking on-tumor Neu5Ac loss (lectin histochemistry, SNA/MAL-II staining) and restoration of NK/macrophage infiltrate activation markers.
Anti-Siglec blocking antibodies take the second route: rather than removing the ligand, they sterically occlude the Siglec V-set domain's sialic-acid-binding pocket or block clustering required for ITIM phosphorylation. Antagonist antibodies against Siglec-9 and Siglec-7 restore CD107a degranulation and cytotoxic killing of tumor targets in ex vivo NK/macrophage co-culture assays across multiple carcinoma cell lines, functioning analogously to PD-1/PD-L1 checkpoint blockade but targeting a glycan-recognition rather than a peptide-recognition checkpoint; anti-Siglec-10 and anti-CD24 antibodies pursue the same logic on the macrophage phagocytic side.
Sialyltransferase inhibition attacks the pathway further upstream. P-3Fax-Neu5Ac (a peracetylated, fluorinated CMP-Neu5Ac mimetic first developed by the Paulson laboratory) is taken up by cells, metabolized to the active CMP-3Fax-Neu5Ac species, and acts as a broad-spectrum competitive inhibitor across the sialyltransferase family, depleting newly synthesized surface glycans of terminal Neu5Ac over 48–72 hours of treatment without directly cleaving existing mature glycoconjugates. Because it inhibits biosynthesis rather than removing pre-existing sugar, its kinetics and tissue-selectivity profile differ from sialidase conjugates, and it is currently used primarily as a mechanistic and preclinical tool compound while more selective, tumor-targeted small-molecule and antibody-drug-conjugate variants are in development. Across all three strategies, the shared endpoint is the same: collapse the millimolar-to-nanomolar avidity gain that the intact glycocalyx provides, drop Siglec occupancy below the threshold needed for sustained ITIM phosphorylation, and let the intrinsic activating receptor signal on NK cells and macrophages dominate again.