Sialic Acid in the Regulation of Blood Cell Production, Differentiation, and Turnover

Abstract

Sialic acid is a unique sugar moiety that resides in the distal and most accessible position of the glycans on mammalian cell surface and extracellular glycoproteins and glycolipids. The potential for sialic acid to obscure underlying structures has long been postulated, but the means by which such structural changes directly affect biological processes continues to be elucidated. Here, we appraise the growing body of literature detailing the importance of sialic acid for the generation, differentiation, function, and death of hematopoietic cells. We conclude that sialylation is a critical post-translational modification utilized in hematopoiesis to meet the dynamic needs of the organism by enforcing rapid changes in availability of lineage-specific cell types. Though long thought to be generated only cell-autonomously within the intracellular ER-Golgi secretory apparatus, emerging data also demonstrate previously unexpected diversity in the mechanisms of sialylation. Emphasis is afforded to the mechanism of extrinsic sialylation, whereby extracellular enzymes remodel cell surface and extracellular glycans, supported by charged sugar donor molecules from activated platelets.

Graphical Abstract

Sialylation is a post-translational glycosylation modification utilized in hematopoiesis to meet the dynamic needs of the organism by enforcing rapid changes in availability of lineage-specific cell types. Though long thought to be generated only cell-autonomously within the intracellular ER-Golgi secretory apparatus, emerging data also demonstrate previously unexpected diversity in the mechanisms of sialylation. Emphasis is afforded to the mechanism of extrinsic sialylation, whereby extracellular enzymes remodel cell surface and extracellular glycans, supported by charged sugar donor molecules from activated platelets.

Introduction

Maintaining effective levels of functional blood cells of all lineages is essential throughout the lifetime of an organism. These lineages include not only red cells, platelets, granulocytes, monocytes, and macrophages, but also adaptive immune cells such as T and B cells, to name a few. Although cell-intrinsic mechanisms execute the transitions required for normal development and turnover, these mechanisms need to be responsive to systemic cues conveying the immediate need for specific cell types. Glycans of cell surfaces and the extracellular milieu represent a critical layer of the cell-niche interface through which such cues are conveyed.

One glycan motif that profoundly impacts blood cell biology is sialic acid, a 9-carbon negatively charged monosaccharide occupying the outermost (and arguably the most exposed) positions of glycan chains in mammals and other deuterostomes. At least 3 important classes of mammalian lectins, or carbohydrate recognizing receptors, mediate the biological repercussions of the presence or absence of sialic acid: 1) Siglecs (sialic-acid-binding immunoglobulin-like lectins), which bind sialic acid and often contain intracellular cell signaling domains^1–3, 2) Galectins, a large family of proteins recognizing the underlying β-galactoside moiety uncovered by the removal of sialic acid^2,4, and 3) Selectins, which recognize sialylated tetra-saccharide structures such as sialyl Lewis X (SLe^x)^5,6. The latter, which exists in three varieties (E, L, and P-selectin), constitutes a well-studied class of cell adhesion molecules that mediate tethering interactions under shear stress between leukocytes and endothelial cells. Twenty distinct sialyltansferases mediate the attachment of sialic acid in α2,3, α2,6, or α2,8 linkage to underlying glycans, and four mammalian sialidases catalyze their removal^7,8. The emerging field of how sialylation contributes to production, maintenance, and function of blood cells is herein appraised.

Modes of Sialylation

Investigations into one particular sialyltransferase, ST6GAL1, which mediates the transfer of Sia(α2,6) to Gal(β1,4)GlcNAc-R termini of glycans, are notable for uncovering unexpected diversity in the modes of sialylation in biological systems. ST6GAL1 is associated with a diverse array of clinical conditions including stress, inflammation, atherosclerosis, alcoholism, as well as malignancies – particularly colon, breast and pediatric acute leukemias^9–16. The case for its role in oncogenic pathways is particularly strong, especially for those signaling pathways affecting behaviors such as sustained proliferation, stemness, chemo-/radiation-resistance, epithelial-to-mesenchymal transition, and invasion^17–22. ST6GAL1 is also implicated in suppressing differentiation of human pluripotent stem cells¹⁸.

Sialyltransferases classically reside within the intracellular ER-Golgi secretory network, where they modify glycoconjugates in biosynthetic transit. Full-length ST6GAL1 has the hallmark of an archetypical glycosyltransferase: a type-2 transmembrane protein with a cytosolic N-terminal domain, a transmembrane domain, a stem region, and a globular C-terminal catalytic domain^23,24. However, ST6GAL1 is also abundant in extracellular compartments²⁵. The beta-site amyloid precursor protein-cleaving enzyme 1 (BACE1) was identified early as a critical enzyme that proteolytically liberates the C-terminal catalytic domain of ST6GAL1 from its transmembrane domain, generating a soluble yet catalytically active smaller form amenable to secretion^26–29. Hepatocytes are the principal source of soluble ST6GAL1 in the circulation³⁰, and their secretion of ST6GAL1 is a part of the acute phase response^31,32. Hepatic expression of ST6GAL1 is under the control of glucocorticoids³³ and cytokines that remain incompletely defined^33,34. Many other cells and tissues also release ST6GAL1 into their environment, including lactating mammary glands³⁵, neonatal intestinal epithelium^36,37, and mature B cells³⁸. Cancer cells, most notably lung, colon, and breast express and release copious quantities of ST6GAL1^39–43. Activated human platelets are also a rich source of circulating ST6GAL1, as well as a number of other extracellular glycosyltransferases that can be detected in plasma²⁵.

The hypothesis of ‘Extrinsic Sialylation’ posits that extracellular sialyltransferases, rather than simply being the byproducts of intracellular sialyltransferase expression, remain enzymatically active and directly sialylate secreted or cell-surface glycoproteins by a Golgi-independent pathway²⁵. Originally proposed by our group as an explanation for the discrepancies between intracellular sialyltransferase expression and cell surface glycoprotein sialylation, extrinsic sialylation predicts that cellular sialylation can be controlled either cell-autonomously or non-cell-autonomously, depending on the origin of the sialyltransferase. We propose three subtypes of extrinsic sialylation – endocrine, paracrine, and autocrine – underscoring the biological relevance of the physical relationship between secreting and target cells. Further, this hypothesis allows for rapid local and systemic modification of sialic acid-dependent processes by alterations in abundance of secreted sialyltransferase in a given context. In line with the observations that blood sialyltransferase levels vary in states of stress or disease, extrinsic sialylation provides a paradigm by which these changes may produce biologically relevant outcomes. ‘Intrinsic Sialylation’, in contrast, describes the classically accepted sialylation mechanism, mediated by cell-autonomous sialyltransferases residing within the intracellular ER-Golgi network. This is schematically depicted in Fig 1.

Fig 1. Modes of sialylation.

Sialic acid is synthesized in the cytosol from glycolytic precursors, then conjugated to cytidine triphosphate (CTP) to form CMP-Sialic acid, the charged sugar donor substrate that participates in sialylation reactions. The secretory pathway involves progressive remodeling of glycans attached to glycoproteins or glycolipids as they transit through the endoplasmic reticulum and Golgi apparatus. Intrinsic sialylation of a terminally galactosylated structure occurs in the trans-Golgi, after which glycoconjugates are typically transported to be embedded in the membrane or exported. Extrinsic sialylation differs in that extracellular sialyltransferases catalyze sialylation of secreted or cell surface glycans, independently of the ER-Golgi pathway. The relationship between the cell producing the sialyltransferase and the target cell can vary, as illustrated by autocrine (self-sialylation by secreted enzyme), paracrine (sialylation of a tissue by localized enzyme secretion), or endocrine (sialylation of distant targets after transport in the blood). Emerging evidence suggests that extracellular sialyltransferase can also be imported into recipient cells and utilized in intracellular reactions.

A conceptual stumbling block to the plausibility of extrinsic sialylation has been the contention that the necessary sugar donor substrate, CMP-Sia, is absent in the extracellular milieu. This notion was challenged initially by Wandall et al⁴⁴, who demonstrated that platelets store such sugar donor molecules within granules and release them upon activation. Later, the ability of activating platelets to supply sialic acid to extracellular sialyltransferases was demonstrated unequivocally in vitro and in vivo^45,46 and led to the notion that platelet degranulation is a central event regulating extrinsic sialylation²⁵. Extracellular vesicles, released from both living and dying cells, are also hypothesized to contribute to the extracellular supply of CMP-Sia.

Extrinsic sialylation may also operate by direct exchange of sialyltransferases between cells. We have observed that the larger, full-length ST6GAL1 isoform, which presumably has not been subject to BACE1 cleavage, also exists extracellularly as part of the secretome of breast cancer and multiple myeloma cells^47,48. This secreted larger ST6GAL1, owing to retention of the transmembrane domain, may be embedded in membrane-containing extracellular vesicles^48,49. There is emerging evidence that ST6GAL1 is found in small, non-membranous extracellular nanoparticles called exomeres, which can be internalized by recipient cells to remodel cell-surface glycans⁴⁹. Extracellular vesicle mediated internalization of another exo-glycosyltransferase, N-acetylglucosaminyltransferase-V (GnT-V), was also reported recently⁵⁰. Whether extracellular sialyltransferases universally exist within nanoparticles and the extent to which intercellular sialyltransferase exchange mediates sialylation remain unknown.

Hematopoiesis in the Bone Marrow

In adults, the marrow is the principal site for the production of blood cells of all lineages from hematopoietic stem and progenitor cells (HSPC). Blood cell production in secondary sites, or extramedullary hematopoiesis, also occurs, especially in response to extreme stress caused by diseases such as severe infections or blood cancers, or during fetal development. Two principal niches have been proposed as functionally distinct microenvironments within the bone marrow – the endosteal niche which lies adjacent to trabecular bone and houses quiescent HSPC, and the sinusoidal niche, which is associated with active HSPC, along with various cell types ingressing and egressing from the marrow⁵¹. The function of sialic acid in the earliest hematopoietic progenitors remains poorly understood. However, homing of HSPC into the marrow is dependent on tethering interactions between α2,3-sialic acid containing tetrasaccharide structures and E-selectin, which is constitutively expressed on marrow endothelial cells ^52,53. This is in contrast to endothelium of other tissues, in which E-selectin only becomes present on the cell surface hours after transcriptional upregulation by pro-inflammatory IL-1, TNF-α, or LPS⁵⁴. On human HSPC, a specialized Sle^x-containing CD44 glycoform called hematopoietic cell E/L-selectin ligand (HCELL) is a high affinity E-selectin ligand that is necessary for marrow homing^55,56. When this ligand is artificially synthesized by FTVI exofucosylation of CD44 on mesenchymal stem cells (MSCs), they gain the ability to migrate into the bone marrow despite lacking the chemokine receptor CXCR4⁵⁷. In regards to α2,6-sialylation, it is noteworthy that lineage-negative hematopoietic progenitors in the mouse marrow are virtually absent of native ST6GAL1 expression. Yet, these hematopoietic progenitors are richly endowed with cell surface α2,6-sialic acids due to the action of hepatocyte-derived ST6GAL1, as demonstrated in conditional knockout and bone marrow transplantation experiments⁵⁸. As HSPCs differentiate into the early, lineage-committed cells of the bone marrow, the functional importance of sialic acid becomes increasingly clear, as detailed below.

In the platelet lineage, productive thrombopoiesis requires the timely expression of both galactosylated and sialylated glycans. Medullary TPO and CXCL12 stimulate expression of B4GALT1 galactosyltransferase, which galactosylates β1 integrin to inhibit ligand binding and signaling ⁵⁹. This inhibition is needed to form the demarcation system (DMS), by which a single megakaryocyte is partitioned into thousands of proplatelets⁵⁹. Expression of ST3GAL1 and ST3GAL2 are also necessary for proplatelet formation⁶⁰. Sialylation of megakaryocytes by ST3GAL1 sialyltransferase, interestingly, obscures the Thomsen-Friedenreich (TF) autoantigen, preventing their recognition and untimely destruction by type I interferon-producing plasmacytoid dendritic cells⁶¹.

Early red blood cell precursors are nucleated in marrow erythroblastic islands, each consisting a single Siglec-1+ (CD169) macrophage that provisions iron to nearby erythroblasts⁶². This macrophage-erythroblast interaction is mediated by VCAM-1 - which binds erythroblast integrins - and Siglec-1, which binds a sialylated erythroblast ligand, likely MUC1 or CD43^63–65. Depletion of medullary Siglec-1+ macrophages compromises the dynamic reserve to support erythropoiesis under stressful conditions, such as anemia and thalassemia⁶².

In the monocyte lineage, sialic acid supports early maturation but resists later differentiation into macrophages (Fig 2a). Extracellular ST6GAL1 is rapidly sequestered on the surface of monocytes to extrinsically sialylate the M-CSF receptor, activating NF-kB, ERK, and AKT signaling and upregulating the key myeloid transcription factor PU.1⁶⁶. Further sialylation by ST6GAL1, ST3GAL1, and ST3GAL4 accompanies differentiation into dendritic cells⁶⁷. When monocytes prepare for extravasation, two parallel mechanisms – upregulation of surface neuraminidases and cleavage and release of sialyltransferases by BACE-1 – enable the membrane desialylation needed to trigger β1 integrin clustering and binding to VCAM-1, allowing for tissue infiltration and macrophage differentiation^27,68.

Fig 2. Diverse functions of sialic acid in hematopoietic cells.

(A) Hematopoietic cells accrue sialic acid during early development, but rapidly shed their sialic acid when activated or in response to inflammatory stimuli. This desialylation rapidly generates pro-inflammatory granulocytes and macrophages, prepares dendritic cells for antigen presentation, and accelerates the senescence and turnover of platelets and erythrocytes. (B) Exposure to LPS and TNF-α in macrophages prompts rapid desialylation of the crucial cell surface receptors TLR4 and TNFR, activating a positive feedback loop of inflammation that promotes M1 differentiation but ultimately limits macrophage lifespan. (C) Macrophages and granulocytes express cell surface Siglecs to sense environmental sialic acid. Growth factors M-CSF and GM-CSF stimulate monocyte Siglec-7 and Siglec-9 expression, which sense tissue sialic acid ligands to promote PD-L1 and IL-10 expression, M2 differentiation, and cell death. (D) On neutrophils, Siglec-9 engages erythrocyte sialic acids to inhibit ROS production, NETosis, antimicrobial functions, and promote inflammatory cell death. (E) Lymphocytes possess abundant sialic acid, which in B cells plays a critical role in meeting developmental milestones via engagement of Siglec-2/CD22. Mature B and T cells rapidly lose sialic acid upon activation, exposing underlying galactose for galectin activation. In B cells, desialylation limits BCR signaling and allows for galectin engagement, which directs differentiation into memory or plasma cell phenotypes. In T cells, desialylation allows for robust cytotoxic cell activation but also promotes cell death. Galectin engagement further influences differentiation into specific helper T cell subtypes.

In the granulocytic lineage, secreted ST6GAL1 sialylates the earliest FcγRII/III+ granulocyte/monocyte progenitor (GMP) and granulocyte progenitor (GP) populations, blocking G-CSF/STAT3 signaling and expression of myeloperoxidase (MPO) and C/EBP-alpha (Fig 2a)⁶⁹. Mice lacking hepatic-derived circulatory ST6GAL1 are predisposed to neutrophilia, whereas treatment with exogenous ST6GAL1 tempers systemic neutrophilic inflammation^58,70–72. ST6GAL1 similarly blocks the bone marrow development of eosinophils, reducing airway eosinophilia by an IL-5 dependent pathway⁷³.

Among lymphocytes, the bone marrow is the site of B cell maturation, wherein galactose and sialic acid expression heralds specific stages of development. Assembly of the ‘pre-BCR’ sustains temporary signaling to support development of pre-B cells. Galactosylation of surface ligands enables direct contact with galectin-1 expressing stromal cells, trapping the pre-BCR in a galectin-1 lattice that also immobilizes galactosylated α4β1, α4β7, and α5β1 integrins on B cells. Formation of this synapse potently stimulates the pre-BCR’s intracellular tyrosine kinase activity^74,75. At the immature stage, a complete BCR is assembled, accompanied by increased sialylation of CD22 and CD45 by ST6GAL1, triggering robust BCR activation, the primary cue that positively selects cells at this stage for further maturation. Although most B cell sialylation is driven by endogenous ST6GAL1, immature B cells are also susceptible to extrinsic sialylation by bloodborne ST6GAL1. Extrinsic sialylation by hepatic ST6GAL1 licenses B cells for differentiation and survival, enhancing BCR and BAFF-mediated NF-kB, PI3K, and MAPK signaling⁷⁶. Even after differentiation into mature stages in the periphery, extrinsically sialylated cells continue to exhibit improved BCR signals, proliferation, and IgG production by a CD22-dependent mechanism⁷⁷.

Erythrocytes and Platelets

In the periphery, the function of sialic acid is most straightforward in the erythrocyte and platelet lineages. In both, sialic acid is a marker of youth and its loss signals senescence and cell death (Fig 2a). Such a response to desialylation may have developed as an adaptation to sepsis. Both platelets and erythrocytes lose sialic acid moieties during sepsis, attributed to the presence of microbial sialidases in the bloodstream. Erythrocyte Lu/BCAM, normally tethered in cis to sialylated glycophorin-C on the cell surface, is liberated upon desialylation to find new ligands such as laminins of the vasculature, an interaction associated with erythrocyte stasis and vessel occlusion in stroke and myocardial infarction^78–80. Conversely, mice lacking the receptor to recycle desialylated platelets, when infected with Streptococcus pneumoniae, develop disseminated intravascular coagulation (DIC), a deadly condition characterized by widespread thrombosis⁸¹. The ‘permissive thrombocytopenia’ triggered by platelet desialylation is thought to prevent this process by reducing the availability of platelets for maladaptive clots.

Mature erythrocyte surfaces are coated in heavily glycosylated glycophorins, the sialylation of which is necessary for entry by Plasmodium organisms, the causative agents of malaria^82–86. The evolutionary pressure to preserve a nonsense mutation in the cytidine monophosphate N-acetylneuraminic acid hydroxylase (CMAH) gene in great apes 3.2 million years ago, is thought to have interfered with Plasmodium infection by depleting the glycolyl subtype of sialic acid on erythrocytes^87,88. Several mechanisms also couple glycophorin sialylation to cellular half-life. First, loss of sialic acid alters membrane electrostatic charge to disrupt erythrocyte morphology, creating a more spheroid, less deformable shape that is recognized by phagocytes of the reticuloendothelial system⁸⁹. Second, exposure of underlying galactose also promotes recognition by macrophage galectin receptors, prompting phagocytosis and destruction⁹⁰. Third, sialidases directly induce erythrocyte cell death, or erypoptosis, by increasing phosphatidylserine presentation and release of intracellular calcium stores, an effect that can be recapitulated by exogenous sialidases and prevented by exogenous sialyltransferases^89,91–93.

Glycans play a profound role in platelet half-life, as uncovered by studies of platelet storage. Chilled platelets are rendered obsolete upon transfusion owing to two mechanisms. First, briefly chilled platelets aggregate Gp1ba, facilitating their recognition and uptake by hepatic macrophages via complement receptor 3 (CR3/aMB2 integrin)^94,95. This interaction depends on the exposure of terminal N-acetylglucosamine of Gp1ba and can be reversed by the exogenous provision of galactose or UDP-Gal^96,97. In contrast, platelets that have undergone long-term refrigeration can no longer be rescued by galactosylation, as they release endogenous sialidases to completely desialylate Gp1ba, naturally exposing underlying galactose that is vulnerable to cleavage and inactivation by the matrix metalloproteinase ADAM17⁹⁸. Desialylated Gp1ba also tends to cluster and is recognized by the galectin Ashwell-Morell receptor expressed on macrophages and hepatocytes, leading to platelet destruction and thrombocytopenia, as observed in ST3GAL4-deficient mice⁹⁹. Notably, platelet phagocytosis in hepatocytes activates JAK2-STAT3 signaling, leading to the release of TPO to stimulate restorative bone marrow thrombopoiesis¹⁰⁰.

Monocytes, Macrophages, Dendritic Cells, and Granulocytes

Professional phagocytes of the myeloid lineage, including monocyte-derived macrophages and dendritic cells, as well as granulocytic neutrophils and eosinophils, are poised for activation in response to infectious and inflammatory triggers. A general trend in phagocytes is an inflammatory activity that inversely correlates to their degree of sialylation. Macrophages are generally poorly sialylated in comparison to monocytes and dendritic cells, and further lose sialic acid in response to two pro-inflammatory molecules - LPS and TNF-α (Fig 2b)¹⁰¹. LPS triggers trafficking of NEU1 to the cell membrane in complex with cathepsin A, where it desialylates the LPS receptor, TLR4¹⁰². Desialylated TLR4 dissociates from Siglec-E, increasing its cell surface half-life, signaling potential, and downstream expression of IL-6 and MCP-1^102,103. Ultimately, this promotes an M1 phenotype, with expression of CD80, TNF-α, and IL-1β, creating a positive feedback pro-inflammatory pathway in response to gram-negative infection¹⁰⁴.

In contrast, TNF-α upregulates BACE-1, liberating ST6GAL1 and reducing sialylation of β1 integrin, thereby promoting adhesion of macrophages to endothelial cells¹⁰⁵. ST6GAL1 release also depletes sialylation of the TNF receptor, which selectively inhibits the pro-survival effects of TNF-α ^106,107. Although short-term TNFα-induced NF-kB activation remains intact in ST6GAL1-deficient macrophages, sustained NF-kB activation, central to the cytokine’s pro-survival effect, is compromised, resulting in tissue-infiltrating but short-lived macrophages¹⁰⁷.

In dendritic cells, desialylation improves phagocytosis, production of IFN-g, TNF-α, IL-10, and IL-6, and antigen presentation to T cells¹⁰⁸. The effects of sialic acid on antigen presentation are likely secondary to the major histocompatibility complex. Desialylated MHC-I associates more efficiently with β2-microglobulin, binds peptides with higher affinity, and has a longer half-life¹⁰⁹. Loss of sialic acid, by either sialidase treatment or knockout of ST6GAL1 or ST3GAL1, also increases MHC-II expression and antigen presentation to favor a Th1 response¹¹⁰. Pro-inflammatory stimuli, such as LPS, induce desialylation via NEU1 and NEU3, whereas glucocorticoids stimulate an increase in sialylation and differentiation into a more tolerogenic phenotype^111,112.

Recent literature supports the idea that neutrophils share some of the pathways first described in macrophages and dendritic cells. For instance, activated neutrophils, including both those exposed to LPS and those isolated from individuals with COVID-19, exhibit high NEU1 activity, which improves TLR4 signaling, leading to exuberant ROS production and detrimental neutrophilic inflammation by a MMP-9 dependent pathway^113,114. NEU1 also improves neutrophil extravasation by desialylation of β2 integrin (CD11b/CD18), allowing for higher affinity interactions with endothelial ICAM-1/ICAM-2, especially in the lung^115–117.

A second trend in phagocytes is that they are potently suppressed by environmental sialic acid via Siglec engagement (Fig 2c). Siglec-7 and Siglec-9 are expressed in monocytes exposed to M-CSF and GM-CSF, and bind ST8SIA6-produced polysialic acid on ligands such as glycodelin-A, GD2 ganglioside, and MUC-1^118–122. Ligand engagement facilitates differentiation into an immunosuppressive M2 phenotype, with high PD-L1 and IL-10 expression in the context of the tumor microenvironment ^121,123,124. The ectodomain of Siglec-9 can also act in a paracrine fashion when proteolytically shed, likely binding to sialoglycans on CCR2 to alter MCP-1/CCR2 signaling, promoting the polarization of macrophages into an M2 phenotype ^125,126,127. Both Siglec-9 and its murine counterpart Siglec-E trigger macrophage apoptosis when engaged with aortic endothelial ligands, creating a negative feedback pathway to limit vascular inflammation^128,129. Siglec-7 can also trigger ROS-dependent, non-apoptotic cell death in macrophages, independent of intracellular phosphatase recruitment¹³⁰.

Neutrophils express Siglec-9, which binds α2,3-sialyl ligands on abundant mucosal glycoproteins and executes cell-fate decisions for the activated neutrophil via its intracellular ITIM domain (Fig 2d)^131–133. When engaged with environmental sialic acid, Siglec-9 limits tissue inflammation by blocking oxidative burst, neutrophil extracellular trap (NET) formation, and pathogen clearance^134,135. An important source of this environmental “self” signal is erythrocytes, which engage neutrophil Siglec-9 via glycophorin A to limit neutrophil activation in blood ¹³⁶. When activated neutrophils ligate Siglec-9, however, their endogenous upregulation of NADPH oxidase and ROS production triggers ROS-dependent cell death, rapidly suppressing inflammation¹³⁷. Interestingly, pro-inflammatory neutrophils have been observed to shed Siglec-9, shielding exposed sialic acid to avoid this inhibition of antimicrobial activity towards bacteria^138,139.

Finally, Siglec-8 (or murine Siglec-H) binds ST3GAL3-generated α2,3-sialyl ligands of the airway epithelium to suppress eosinophil and mast cell activation and survival^140–142. In eosinophils activated by IL-5 and IL-33, Siglec-8 ligation enables a unique β2 integrin-mediated cell death pathway, which signals via Akt, p38, and c-Jun to activate NADPH oxidase, generating excessive ROS, mitochondrial damage, and caspase activation^143–149.

B and T Lymphocytes

In contrast to phagocytes, lymphocytes are generally heavily sialylated and depend on sialic acid to meet milestones in their development. When activated, however, they rapidly shed sialic acid and become susceptible to the effects of galectins, receptors that specifically recognize the freshly exposed underlying galactose moieties, which direct their further differentiation (Fig 2e).

In B cells, sialic acid biology is shaped by the siglec CD22, which binds the α2,6-sialic acid motif constructed by ST6GAL1. Owing to the inability to form CD22 ligands, ST6GAL1-deficient mice suffer diminished BCR signaling, proliferation, and antibody responses to T-dependent and T-independent antigens^150–153. CD22 functions are context dependent. While its intracellular ITIM domain suppresses BCR signaling by SHP-1 phosphatase recruitment, its extracellular, sialic acid binding domain determines its proximity to the BCR in the plasma membrane¹⁵⁴. Therefore, though the effects of ST6GAL1 deficiency are due to CD22 disengagement, complete CD22 deficiency paradoxically improves BCR signaling^155–158. Major CD22 binding partners include CD45, IgM, and CD22 itself, the latter constituting homomultimeric microdomains that are regularly endocytosed and recycled via the endosomal pathway^159–161. Both CD22-CD22 and CD22-CD45 complexes sequester phosphatase activity away from the BCR, improving BCR signaling strength. Transient interactions with CD45 sustain CD22’s characteristic Brownian motion, which also enables ‘surveillance’ of BCR signaling by both phosphatase recruitment and endocytosis^162,163.

Given the importance of CD22 for BCR signaling, ST6GAL1 plays an outsized role in B cell development. Both CD22 expression and cell surface sialylation reach high levels in the late transitional B cell stages, with sialylation being sustained at high levels in the subsequent follicular but not marginal zone lineages⁷⁶. Early transitional B cells, in contrast to mature follicular B cells, have yet to assemble a functional Lyn-CD22-SHP-1 pathway to inhibit BCR signaling and promote tolerance and are implicated in autoimmunity¹⁶⁴. The poorly sialylated marginal zone B cells, a splenic population of IgM-producing, innate B cells, are abundant in ligand-free, or ‘unmasked’ CD22¹⁶⁵. Interestingly, this population is entirely compromised by the loss of intrinsic ST6GAL1 expression, possibly reflecting a state of high BCR suppression from the overwhelming engagement of CD22 ligands in trans^166,167. In contrast, other B cell populations compensate for ST6GAL1 and CD22 disruption by reliance on redundant pro-survival pathways, including those mediated by OCA-B or BLNK^168,169.

It is worth noting that the major secreted product of B cells, immunoglobulins, are also glycosylated, with approximately 10% of IgG molecules bearing biantennary terminal sialic acid on their N-terminal Fc domain. The importance of IgG Fc sialylation by ST6GAL1, which is critical to the anti-inflammatory effects of IVIG therapy, has been recognized for some time. More recently, it has come to light that IgG sialylation can occur independently of B cells^170–173. Studies suggest that IgG sialylation occurs after its secretion, but is also independent of hepatocytes and platelets, the major sources of blood ST6GAL1^174–176. Other post-secretion modifications of IgG also occur, such as the deacetylation of sialic acid in pregnant mice, which occurs independently of B cells and confers transplacental protection to newborns against Listeria monocytogenes¹⁷⁷. There is also evidence that sialylation of IgE and IgA similarly impact biologically relevant processes, which may occur by B cell-independent pathways^178,179. However, considerable discrepancy persists, with IgG being sialylated by traditional intracellular pathways in a number of other models, highlighting the complexity and context dependence of immunoglobulin sialylation.

In T cells, the majority of sialylation is associated with ST3GAL1 expression, which gradually increases during T cell maturation. Regulatory T cells upregulate ST6GAL1 and increase their α2,6-sialylation, though the significance of this remains unclear¹⁸⁰. In CD8+ T cells, ST3GAL1 reduces the strength of the MHCI-CD8b interaction by direct sialylation of CD8b, dampening sensitivity of T cells to low-affinity ligands¹⁸¹. CD8b sialylation also interferes with non-cognate (TCR-independent) interactions involved in intercellular adhesion^182,183. Accordingly, exogenous desialylation of a CD103+ subset of CD8+ T cells sensitizes them to tumor antigens on glioma cells¹⁸⁴. However, some researchers have cast doubt on this mechanism, as ST3GAL1-deficient cells have demonstrated preserved CD8-MHCI interactions in some models¹⁸⁵.

When activated, both mature B and T cells undergo a dramatic loss of sialic acid, a change that has opposing outcomes in these two cell types^186,187. In B cells, a loss of the highest affinity, sulfated CD22 ligands enables interactions with lower affinity ligands, including in trans with sialylated antigens presented by follicular dendritic cells (FDCs), which juxtapose CD22 with the BCR to block activation¹⁸⁸. Nevertheless, a modest restraint of BCR signaling by CD22 is necessary to prevent excess calcium influx and cell death, as CD22-deficient germinal center B cells fail to fully form into plasma and memory B cells^189–191. For T cells, reduced ST3GAL1 and increased NEU1 and NEU3 activity depletes sialic acid and replaces core 1 O-glycans with core 2 O-glycans^187,192,193. Desialylation improves T cell signaling and proliferation, priming them for the immune response¹⁹⁴. Desialylation of CD43, however, is a safeguard that triggers CD8+ T cell apoptosis if co-stimulation (of CD3 or by IL-2) does not occur^195,196. Desialylation by NEU1 is also critical for IL-4, IgG1 and IgE synthesis owing to desialylation of the ganglioside GM3, which typically attenuates CD3-induced calcium signaling^197,198. NEU1 enforces Th2 polarization by CD44 desialylation, enabling hyaluronic acid binding and asthmatic airway inflammation^199,200.

Galectins, the receptors recognizing the underlying galactose moieties exposed by the removal of sialic acids, have a critical role in directing further differentiation of activated lymphocytes in secondary lymphoid tissues. Galectin-9 has a regulatory effect by inhibiting GC B cells and promoting regulatory T cells. In activated B cells, it interferes with CD22 interactions, incorporating CD22 and CD45 into BCR signalosomes to suppress signaling^201,202. As a side note, GC B cells that construct I-branches on their N-glycans by expression of the glycosyltransferase GCNT2 can prevent galectin-9 binding, thus preserving their signaling²⁰³. In T cells, galectin-9, expressed under regulatory control of SMAD3, directly binds CD44 to assemble a CD44-TGF-β-TGF-βRI complex that stabilizes SMAD3 and FoxP3 activity, promoting the regulatory T cell phenotype²⁰⁴.

Galectin-3 promotes memory B cell production while suppressing T cell maturation. Expressed in naive B cells, it coordinates with IL-4 to promote survival and differentiation into memory B cells²⁰⁵. In T cells, galectin-3 binds CD45 and CD71 to induce cell death, particularly at the thymic double-negative stage²⁰⁶. Galectin-3 deficiency leads to exuberant IFN-g production and autoimmunity²⁰⁷.

Galectin-1 and galectin-8 are both expressed downstream of Blimp-1 in B cells and promote plasma cell differentiation^205,208,209. Galectin-1 in particular activates Syk, Btk, and PI3K independently of the BCR and can also be secreted in a paracrine manner by nearby monocytes^210,211. In T cells, galectin-1 binds glycans on CD7, CD43, and CD45 to promote phosphatase activity and death, an effect that can be inhibited by ST6GAL1-mediated sialylation, which obscures the galactose to block galectin binding²¹². Th1 and Th17 cells are most susceptible to galectin-1 mediated cell death, whereas Th2 cells are partially protected by their increased sialylation²¹³. The structurally related galectin-2 binds β1 integrin and activates intrinsic apoptosis in T cells, with cytochrome C release, caspase 9 and 3 activation, and DNA fragmentation²¹⁴. Galectin-8, on the other hand, is unique in binding α2,3-sialyl ligands synthesized by ST3GAL6 on FOXP3+ regulatory T cells, where it competes with macrophage siglec-1 to prevent cell death^215,216. In so doing, galectin-8 sustains regulatory T cell TGF-β and IL-2 signaling and expression of CTLA-4 and IL-10^217,218. Thus, the presence and balance of specific galectins drastically alters the differentiation of B and T cells to impact the overall immune response.

Conclusion

Sialylation has evolved as a means to rapidly and reversibly alter protein functions to effect broad cellular changes and dynamically meet the rapidly changing needs of an organism. The extensive body of work covered in this article illustrates a vast diversity of molecular pathways affected by sialylation in hematopoietic cells, regulating their generation, differentiation, function, and destruction. In the bone marrow, most developing cells gradually acquire sialic acid as both a mark of maturity and as a critical functional element that sustains survival and differentiation. In the periphery, desialylation is both a marker and a trigger of inflammation - activated cells of most lineages rapidly shed sialic acid, yet are resisted by non-hematopoietic sialoglycans, which maintain homeostasis by binding inhibitory siglecs. Finally, loss of sialic acid limits changes to cell fate, triggering death by a variety of mechanisms in myeloid and lymphoid lineages, or otherwise altering differentiation by galectin engagement. Recent findings have expanded the scope of how sialylation can occur to encompass non-traditional mechanisms such as extrinsic sialylation, wherein extracellular sialyltransferases catalyze sialylation outside the ER-Golgi complex. The current literature presents compelling data on how extrinsic sialylation by extracellular ST6GAL1 can convey critical endocrine, paracrine, and even autocrine signals by post-translational remodeling of glycans. Given that the current understanding of extrinsic sialylation is limited to a single sialyltransferase, it remains to be determined whether other glycosyltransferases can participate in similar extracellular reactions. Other glycosyltransferases including galactosyl-, fucosyl-, N-acetylglucosaminyl-, and other sialyl-transferases are also abundantly present in the extracellular milieu, in addition to the sialidases, galactosidases, and likely other glycosidases. Together, these extracellular anabolic and catabolic glycan modifying enzymes raise the tantalizing suggestion that wholesale glycan remodeling may occur by the extrinsic pathway. This knowledge continues to seek fruition in its application as novel therapeutic approaches for the endless spectrum of diseases involving the immune system and blood.