Glycosylation

Abstract

The “danger” model of immunity posits that the immune system is triggered by endogenous danger signals, rather than exogenous non-self signals per se. It has been proposed that danger signals may consist of both intracellular “pre-packed” molecules released from damaged cells and stress-induced proteins. Here we focus on glycosylation aberrancies as a unifying concept for danger signaling. According to this proposition glycosylation patterns reliably reflect cellular phenotypic state and appearance of altered carbohydrate structures may constitute a pivotal phenotypic alteration that alarms the immune system to danger and initiates immunity. Viewed from this vantage point, healthy cells avert immune recognition by virtue of their normal terminal glycosylation patterns. By contrast, abnormal cells display and release glycoproteins and glycolipids with aberrant terminal glycosylation trees, which in turn immunologically flag these cells. Diverse carbohydrate-binding receptors are expressed on immune cells and are used to detect these phenotypic changes. Thus, in addition to the “pre-packed” and stress-induced signals this glycosylation-based signal represents an endogenous signal reliably reflecting the cell phenotypic status, enabling the immune system to monitor the tissue/cell's physical condition and to respond accordingly.

Introduction

The “danger” model¹ has provided a compelling new vantage point from which to view immunity. This model, which contrasts with the traditional “self:nonself (SNS) discrimination” model, posits that the immune system is geared towards responding to danger signals, rather than towards non-self per se. This basic difference between these two models implies an intrinsic difference in the nature of the signals that initiate the immune system.

How does the immune system sense stress or injury, and what is the molecular identity of the danger signals? This pivotal issue remains unresolved, with clues pointing in a number of directions. One unifying concept is that danger signals consist of intracellular “pre-packed” molecules released upon necrosis (“bad death”), but not programmed cell death/apoptosis (“good death”). Additional danger signals that have been proposed consist of stress-induced proteins, for example, heat shock proteins.² Both of these categories of putative danger signals share in common two critical features: (1) lack of exposure/expression by healthy cells or cells undergoing the normal programmed cell death; and (2) recognition by receptors on resting antigen-presenting cells (APC). Thus, the essential controlling signals within the danger model are endogenous, not exogenous.²

Additional, rather ignored, fundamental difference in the nature of the signals that initiate the immune response is that whereas SNS discrimination looks to genotypic differences (in the form of extraneous protein sequences that connote foreignness), the danger model looks to phenotypic differences (in the form of intrinsic cellular components that are somehow altered and emanate from or are exposed on stressed or injured cells). In accord with this proposal, a reliable danger signal should be dependent on the cellular condition, ranging from perfect cellular health to necrotic death, and should reveal the phenotypic status of the cell to the immune system.

A Short Synopsis of the Proposed Model

Here, we propose the centrality of cellular glycosylation status as a critical barometer of cellular well being that is being deciphered by the immune system via carbohydrate receptors that are involved in regulation of effector cells. Hence, this proposal directly links glycosylation patterns with the cell physical condition. Briefly, a healthy, normal cell will have intact terminal glycosylation branches on its exposed glycoproteins and glycolipids, which will not trigger the immune system, and may even actively interfere with immune activation. By contrast, abnormal cells, stressed or damaged, display or release aberrant terminal glycosylation branches, which may signal to the immune system deleterious cellular change, or danger. Hence, appearance of altered carbohydrate structures may constitute a pivotal phenotypic alteration that alarms the immune system to danger and initiate repair and remodeling systems and, ultimately, immunity.

A danger model that is glycosylation-centric is appealing for several reasons: (1) The sensing mechanism is global, since a generic post-translational process, present within all eukaryotic cells, is being monitored; (2) Reliance on a readout (glycosylation) that is not encoded by a gene template and exquisitely dependent on environmental conditions provides for a general and reliable alarm system; (3) The high turnover rate of surface glycosyl structures makes the system highly responsive; and (4) Carbohydrate recognition provides multi-faceted links to different immune effectors, so that diverse immunological decisions can be orchestrated with a relatively simple, universal trigger.

The hypothesis provides a unifying paradigm explaining many intriguing observations, such as why lectins are so abundant in the immune system and why is it that lectins, which detect terminal sialic acids (siglecs) are for the most part inhibitory while mannose receptors are activating receptors. By varying its glycosylation pattern such as sialic acid output, a cell has a relatively simple trigger to communicate with a plethora of immune cell receptors finely-tune a number of different responses, ranging from antigen-presenting cells' maturation, NK activity to the regulation of the a-cellular complement cascade.

Why Glycosylation?

Placing glycosylation at the center of the danger model has teleological appeal that is based on intrinsic properties of the glycosylation process. Glycosylation effects proteins folding and stability and influences their biological activity. Hence, glycosylation is an important post-translational modification that is widely distributed in living organisms, and most surface and secreted proteins (as well as lipids) are glycosylated. Glycosylation is a multistep process that is taking place in both the ER and the Golgi and involves multiple enzymes. The oligosaccharide decorating proteins are divided into O-linked and N-linked according to whether they are linked to the protein core via Ser or Asp residues, respectively. The complex oligosaccharides of N-linked can be divided into the “core” and “terminal” regions, which are composed of sugars added in the ER and Golgi apparatus, respectively. The inner core region is composed of N-acetylgluccosamine and three mannose residues, after the other mannose residues originally added in the ER were clipped. The terminal region consists of a variable number of N-acetylglucosamine-galactose-saccharide units linked to the core mannose residues. Most of the N-linked oligosaccharides on serum and cell surface glycoproteins are terminated in sialic acid. Hence, the highly mannosylated core structures are normally found only inside the cell as intermediates of glycoprotein synthesis and are not found on the finished glycoprotein where most of the residual core mannose residues are covered with terminal oligosaccharides. However, in contrast to protein synthesis where the order of the monomer units is determined by a gene template, no template is included in the biosynthetic machinery of the glycan chain. This template-independent process allows for considerable diversity in the glycan structure that is considered as an inherent characteristic of the glycan moieties and has an enormous potential for encoding biological information. This structural heterogeneity can result from two key features of the glycosylation process. One, different enzymes can compete with each other on the same substrate. Second, many of the reactions are often incomplete, leading to the secretion of a protein with a mixture of glycosylation patterns, or “glycoforms”. A frequent feature of glycoform heterogeneity is in the presence or absence of terminal sialic acid residues. Furthermore, the process of glycosylation demands a substantial investment of metabolic energy, and consequently, the content and exact structure of the glycan chains decorating glycoproteins is highly sensitive to intrinsic cellular conditions that in turn are sensitive to environmental conditions.

Therefore, we suggest that glycosylation constitutes a phenotypic indicator of cellular energy status and environmental conditions, and the immune system may have evolved to reliably detect this phenotypic status using lectin-like receptors. By relying on a post-translational modification (glycosylation aberrancies) that is highly energy-dependent, one is conveniently invoking a signal that is usable in the absence of molecular components that are extrinsic to the individual, that reliably reflects the cell phenotypic status, and is universal in nature (manifest on all cell types). As a general rule, glycoconjugate with exposed core or aberrant terminal glycosylation trees will indicate stress or damage and will immunologically flag cells carrying these abnormal glycoforms. By contrast, healthy cells avert immune recognition by virtue of their normal terminal glycosylation patterns. Because the glycosylation can vary over a wide range, they need not be limited to “on” or “off” signals but can act as intrinsic barometer of cellular well being, ranging from perfect cellular health (intact terminal glycosylation pattern such as sialic acid) to necrotic death (exposed core glycosylation pattern, i.e., loss of sialic acid and exposed mannose). Moreover, this type of “global” signal has the added advantage of having a relatively high turnover rate at the cell surface, allowing for rapid sensing of deleterious changes.

Support for this hypothesis, positing glycosylation as central to the sensing mechanisms of the innate immune system, comes from many directions in the glycosylation and innate immunity literatures. These relate to a number of aspects forming the basis for this hypothesis: (1) Glycosylation is dependent upon the health of cells and changes in glycosylation accompany and reliably reflect phenotypic changes in the cell. Thus, normal healthy cells display normal glycosylation and unhealthy cells display aberrant glycosylation; (2) Consequently, aberrant glycosylation at the cell surface reliably flags a cell under stress; (3) Immune cells have evolved molecular sensors (in the form of lectin-like receptors) for normal versus aberrant glycosylation, and respond accordingly; (4) Normal glycosylation (such as sialic acid at the terminal) can actively block immune attack, with certain glycoproteins functioning as key protective surface molecules that protect cells from immune attack; and (5) Abnormal glycosylation of these glycoproteins render cells susceptible to immune attack. Specifically, “core” oligosaccharides (such as exposed mannose usually present on nascent glycoproteins in the process of synthesis) exposed outside the cell can effectively trigger immune response.

Glycosylation as an Intrinsic Barometer of Cellular Well Being

The present model is based on the assumption that glycosylation is dependent upon the cellular health and changes in glycosylation accompany and reliably reflects phenotypic changes in the cell. However, the association between glycosylation patterns and cellular health was not tested directly and thus experimental data is missing. Nevertheless, evidence in the literature demonstrates that protein glycosylation is highly regulated and changes under different physiological-and cell culture-conditions and in disease.

The prediction emanating from our proposal is that environmental factors, because of their impact on the cell intrinsic condition, should influence the precise content and structure of oligosaccharide. Support for this association between the environment and the structure of oligosaccharide on glycoproteins comes from the field of recombinant protein biotechnology. Since many biopharmaceutical proteins are glycoproteins in their native state, and proper glycosylation can be critical for their activity, the production of properly glycosylated protein is essential. This has prompt studies directed toward understanding factors that influence glycosylation heterogeneity during the process of protein production in cultured mammalian cells. Overall these studies have demonstrated that culture conditions exert a major effect on the glycosylation pattern of recombinant proteins. For example, the expression of P-glycoprotein, encoded by the MDR gene, during glucose deprivation was associated with impaired glycosylation, resulting in reduction of the apparent molecular size from 170 to 150 kDa.³ Other factors such as the level of glucose, glucoseamine, ammonia, dissolved oxygen, temperature and the time of culture have all affected the glycosyaltion of either specific protein analyzed or the overall protein glycosylation.^4–10 Significantly, these changes were attributed to alternations in intracellular glycosyaltion rather than degradation of carbohydrate side chains after secretion of the protein to the media.^4–6,9 Mechanistically, these changes can be translated to changes in intracellular compartments that are responsible for glycosylation. For example, Acidic pH of the Golgi lumen is known to be crucial for correct glycosylation, and even minor changes in the PH can impair N-glycosylation.¹¹

Further support comes from diseases. Since as we suggest glycosylation is highly sensitive to the biochemical environment and glycoprotein patterns reflect the internal and external environment of the cells in which proteins are glycosylated they can be sensitive indicators of alternation in cell function brought about by disease or stress. Indeed, an increasing number of diseases are known to be associated with changes in the glycosylation profiles of specific glycoproteins, as determined using specific sugar binding properties of lectins comparing the glycosylation profiles of normal and diseased cells/tissues. These analyses have revealed that cells in diseased tissues carry oligosaccharides with structures that are markedly different from the same protein produced by a normal cell. These changes have diagnostic value and are being used as disease markers for a variety of disease types (reviewed in refs. ¹² and ¹³). Therefore, we suggest that when a cell is infected by a virus, becomes transformed into a malignant state, or is exposed to stressful environmental conditions it presents clues to its state on the outer surface of its membrane in the form of altered glycosylation patterns, which in turn can be detected by the immune system.

Immune Recognition is Often Based on Recognition of Carbohydrate Motifs—Lectins as Innate Sensors of Danger

Functionally, glycans play two major roles. One is to confer physiochemical properties on proteins (such as, size and charge). The other is to act as signals of cell surface recognition phenomena. The massages encoded in the structures of complex carbohydrates are deciphered through interactions with complementary sites on carbohydrate-binding proteins.¹⁴ Indeed, numerous carbohydrate-binding molecules (chiefly lectins) that participate in the immune response have been described, including such well-known lectins as mannose receptor, DC-SIGN (DC-specific ICAM-3-grabbing nonintegrin), Siglec family and various NK receptors, but there is no unifying picture of their primary role. Notwithstanding the role of glycan-lectin interaction in adhesion and migration, it has been appreciated for some time that glycosylation is linked to effective immunity and immune-regulation (reviewed in refs. ¹⁵ and ¹⁶), and glycosylation was even suggested to act as a “universal self” that in addition to protein sequences is being used for recognition of “self” versus “nonself”.¹⁷ However, in view of the danger model, with its emphasis on phenotypic (rather than genotypic) changes as key factor for initiating immunity, a different view is being proposed. Accordingly, stress or damage inflicted to the cell results in changes in glycosylation patterns on cell surface and secreted or released molecules. In turn, the immune cells have evolved molecular sensors with which they can detect normal glycosylation versus glycosylation abnormalities using receptors that recognize carbohydrate moieties. Consequently, immune cells that possess lectin-like molecules may detect changes in glycosylation states on cells they interact with, and respond accordingly. Specifically, the model suggests that glycoproteins with intact terminal motifs such as lactoseamine and sialic acid are a recognizable sign of healthy normal cell. On the other hand, cells undergoing necrotic death or damaged cells will release proteins in which glycosylation was terminated prematurely and those have the inner “core” region such as mannose residues at their terminals. Furthermore, it is likely that cells under stress, such as energy deprivation (hypoxia, glucose deprivation, etc.) will secrete or express on their surface proteins with permutations or modifications in their oligosaccharide sequence.

Thus, the proposed model predicts that the immune system will respond differently to “core” vs. “terminal” oligosaccharide motifs. In general, this underlying principle can explain the particular features of groups of lectin-receptors playing a role in immune regulation. Two major lectin-receptor families best exemplify this underlying principle. The first is the sialic acid binding Ig-like lectins (Siglecs) that provide inhibitory signals to various subsets of cells due to the presence of conserved motifs similar to immunoreceptor tyrosine-based inhibitory motifs (ITIMs).^18,19 The second is the mannose receptor that has lectin activity for sugars terminated in mannose, fucose or N-acetyl-glucosamine as well as other “mannose-specific” lectins. In contrast to Siglecs, the binding of mannose receptors to exposed mannose residues results in cell activation and enhanced antigen presentation.²⁰ Glycosylation can, therefore, be a general mechanism for generating immune response while utilizing a limited set of receptor-ligand interactions.

Healthy Cells Avert Immune Recognition by Virtue of their Normal Terminal Glycosylation Patterns—The Unique Case of Sialic Acid

Sialic acid is of special interest because it is usually occurring as a terminal saccharide unit and thus exposed to the surroundings, it has large variety of naturally occurring variants and it is the only sugar residue of glycoprotein that bears a net negative electrical charge. This negative charge significantly contributes to the overall negative charge of the glycoprotein itself and the cell surface, and may cause repulsion of cells. Due to its negative charge sialic acid masks the underlying glycan and specifically the terminal galactose.

Because of its exposed position and the presence of sialidases sialic acid has a short half-life. Thus, sustained presence of sialic acid on the surface glycoprotein will probably require constant synthesis of new glycoconjugates or cycling of glycoproteins from the cell surface through the Golgi²¹ where it will be resialylated and then transfered back to the cell surface. Thus, the maintenance of the level of sialic acid at the surface requires constant investment of metabolic energy.

These properties, specifically the latter, makes sialic acid a sensitive indicator for changes in the cell's energetic status that are almost immediately reflected by the level of sialic acid on the cell's surface. According to this proposal, cells under stress or energy deprived will probably find it difficult to maintain normal levels of sialic acid on its surface. Considering the exposed position of sialic acid on cell surfaces, it seems plausible that this residue has evolved as recognition markers, the presence or absence of which can be detected by receptors that selectively bind to sialylated or desialylated glycans. While the binding site for some lectins—such as certain galectin—are spoiled by sialylation^18,22 siglecs, require specific sialic acid linkage for recognition, and as mentioned above induce a negative signal. Hence, healthy cells with an intact carbohydrate-tree with sialic acids at their terminals will avoid immune attack.

In support of this notion, a number of observations suggest that removal of sialic acid residues rendered target cells more vulnerable to NK cell attack.^23–25 Moreover, mice deficient in the enzyme α-mannosidase-II and hence lacking the N-glycan branching developed autoimmune disease, suggesting that N-glycan branching protects against sterile inflammatory response of the innate immune system directed against the body's own cells.²⁶

Summary

The proposed hypothesis provides an alternative and unifying explanation for many intriguing (and sometimes disparate) observations that have been made over the years regarding the role of carbohydrates in immune recognition and activation and the intriguing abundance and important roles of lectins in immunity. Specifically, we suggest that glycosylation decorating cell surfaces is being monitored by the immune system as an indicator for cellular well-being. Accordingly, changes in glycosylation in the form of truncated branches with exposed “core” structures is a sign of stress and trigger activation (as expected) while normally glycosylated proteins with intact “terminal” structures is a sign of normal and healthy cells and hence do not elicit a response and even inhibits it.