Emerging Principles for the Therapeutic Exploitation of Glycosylation

See allHide authors and affiliations

Science  03 Jan 2014:
Vol. 343, Issue 6166, 1235681
DOI: 10.1126/science.1235681

Structured Abstract


Glycoproteins and glycolipids exist as an ensemble of glycosylated variants, or glycoforms. Specific glycoforms are directly modulated by microenvironmental cues and play a key role in a wide spectrum of biological processes. Consistent with this, certain glycoforms are also prominent in various pathological conditions. These structures are either targeted by exogenous pathogens or associated with specific disease stages or, in some cases, their aberrant expression acts as a trigger to a particular disorder. An increased molecular and structural understanding of the mechanistic role that specific glycoforms play in these pathological processes has driven the development of therapeutics and illuminated novel targets for drug design.

Embedded Image

Antibody glycosylation determines Fc functions. An example is the removal of an antibody’s Fc glycans (red, green, and blue) by a bacterial immune evasion factor, endoglycosidase S, which impedes Fc engagement with cellular receptors (orange) and therefore immunological effector cells.


Intervening in cellular glycosylation pathways provides a route to the alleviation of many of the symptoms of congenital metabolic disorders. Some of these same drugs also affect glycan-mediated virion assembly and offer an exciting prospect for the development of broad-spectrum antivirals against enveloped viruses. Further stages of the viral replicative cycle can be disrupted by considering their dependence on glycosylation, and this currently forms the basis of anti-influenza drugs and potential new classes of anti-inflammatories. The development of therapeutic glycoproteins has been greatly stimulated by the advances in recombinant cellular biosynthetic technologies that can produce defined glycoforms. A prominent example of this approach is the development of monoclonal antibodies with engineered glycosylation, which display enhanced in vivo properties. Furthermore, antibody glycosylation can also be directly modulated in vivo. Serum antibodies involved in autoimmunity can be inactivated by removal of their glycans by bacterial immune evasion factors, and this technology has shown great promise in preclinical studies. Glycopeptides offer intriguing possibilities in the development of anticancer vaccines given their ability to stimulate both humoral and cellular immunity. Additionally, the HIV glycan shield is proving to be an effective target for antibody neutralization and emerging targets for vaccine design and control of infection.


Antiviral therapy looks set to have a strong glycan component in the near future. Viral protein-folding inhibition by monosaccharide analogs and glycan-epitope–dependent antibody neutralization are both promising areas. Although a successful glycan-based vaccine to cancer or HIV has yet to be realized, recent advances in both glycopeptide immunization and elucidation of the unusual features of broadly neutralizing antibodies have provided fresh impetus to these goals. Glycan engineering will continue to deliver enhanced therapeutic glycoproteins, such as antibodies, with enhanced disease modifying properties. Last, the application of bacterial enzymes that cleave antibody glycans may offer new therapeutic opportunities.

Understanding Glycosylation

Glycosylation—the covalent addition of carbohydrates to proteins—is central to many biological processes. Recent advances in understanding the roles of glycans—for example, in protein folding and immune regulation—have revealed that glycans are also involved in many disease conditions, from cancer to microbial infection. Dalziel et al. (p. 10.1126/science.1235681) review the current knowledge of glycans in pathogen invasion, cancer, autoimmunity, and congenital diseases.


Glycosylation plays a key role in a wide range of biological processes. Specific modification to a glycan’s structure can directly modulate its biological function. Glycans are not only essential to glycoprotein folding, cellular homeostasis, and immune regulation but are involved in multiple disease conditions. An increased molecular and structural understanding of the mechanistic role that glycans play in these pathological processes has driven the development of therapeutics and illuminated novel targets for drug design. This knowledge has enabled the treatment of metabolic disorders and the development of antivirals and shaped cancer and viral vaccine strategies. Furthermore, an understanding of glycosylation has led to the development of specific drug glycoforms, for example, monoclonal antibodies, with enhanced potency.

Since the term glycobiology was introduced (1), it has come to be appreciated that glycosylation is fundamental to a wide spectrum of biological processes. Advances in glycan sequencing technology have shown that glycoproteins and glycolipids exist in many glycosylated variants, or glycoforms. These can differ substantially in their biochemical properties and functions. Biosynthetic control of individual glycan structures allows both temporal and developmental regulation of glycosylation according to the functional requirements of the cell. Glycans are structurally diverse, incorporating a wide range of monosaccharide residues and glycosidic linkages. Glycans are synthesized by a large array of sequentially and competitively acting biosynthetic enzymes located throughout the endoplasmic reticulum (ER) and Golgi apparatus (Fig. 1). Mouse knockouts of these enzymes are often embryonically lethal. Knockouts can also cause selective disruptions to specific biological processes. For example, deletion of the α2,6-sialyltransferase results in immune system dysfunction while having little effect elsewhere, whereas ablation of the Golgi α-mannosidase II chronically activates the innate immune system (2). These observations indicate that glycans are often functionally redundant outside particular biological contexts (3).

Fig. 1 Mammalian glycan biosynthetic pathways.

Schematic depicting (A) the various classes of glycans (CS, chondroitin sulphate; HA, hyaluronan; HAS, hyaluronan synthase; HS, heparin and heparin sulphate; KS, keratin sulphate; GM and GD, mono- and disialylated glycosphingolipids, respectively; OM, oligomannose) and (B) their associated biosynthetic pathways in the ER and Golgi apparatus (6). Monosaccharide abbreviations are as follows: Gal, galactose; Glu; glucose; GlcA, glucuronic acid; GlcN, glucosamine; Fuc, fucose; SA, sialic acid; Xyl, xylose.

Glycans exert their biological influence in three ways. First, through their physicochemical properties that range from stabilizing protein folds to forming intrinsic components of the extracellular matrix (4, 5). Second, they are targets for recognition by glycan-binding proteins (GBPs) (6). Third, glycans can modulate the properties of the protein or lipid to which they are attached (6). Recognition of glycans by GBPs plays a central role in cellular communication and cell trafficking (7). These interactions pervade every multicellular system and have been much explored in areas such as embryo development (e.g., Notch/Fringe) and immunology (e.g., siglecs and selectins). Within the cell itself, GBPs are used in various regulatory pathways, as exemplified by the role of calnexin and calreticulin in the ubiquitous glycoprotein-folding quality-control mechanism within the ER (8).

The functional impact of glycosylation of individual molecules is well illustrated by its effect on serum glycoproteins, where glycans can influence a glycoprotein’s circulatory half-life and its functional interactions with other proteins. For example, antibody glycosylation influences antibody half-life and interaction with cellular immune receptors (911). Similarly, glycoforms of the red blood cell modulator erythropoietin substantially affect in vivo efficacy through their ability to directly modulate its clearance from the bloodstream (12).

In pathology, the many different roles of glycans reflect their multiple functions in both tissue homeostasis and host-microbe interactions. Their involvement either is that of targets for exogenous pathogens or involves aberrant expression that triggers or exacerbates an endogenous disease. Many pathogenic microbes make use of glycans during critical early steps in their invasion of host tissue. They achieve this through the use of various GBPs, such as adhesins, that facilitate their initial attachment to the mucosal surface (13). Indeed, terminal host glycan residues are the focal point of various invasive strategies by pathogens. For example, influenza viral particles bind directly to terminal sialic acid residues of respiratory epithelial cells (14, 15). On the other hand, some pathogens synthesize sialidases to remove sialic acid that obscures glycan motifs that the pathogen requires for mucosal attachment (16). In addition to the physical tissue invasion by bacterial pathogens, many symptoms of infection are the result of toxins produced by the bacteria themselves. Several of the most lethal forms of bacterial toxins target glycans; for example, the cholera toxin binds a sialylated glycolipid on the surface of the host’s intestinal epithelial surface (17).

Pathogens have also evolved several immune evasion strategies that exploit glycans. For example, the parasite Trypanosoma cruzi uses a trans-sialidase to transfer sialic acid from serum glycoproteins to its own cell surface as a means of both immunological masking of underlying epitopes and immunosuppression (18). However, a more widespread immune-evasion strategy is the direct synthesis by the pathogen of a limited repertoire of terminal glycan residues mimicking the host “self” structures. Examples include Neisseria meningitidis synthesis of polysialic acid, lewis antigens by Helicobacter pylori, and the multiple glycolipid epitope expression by Haemophilus influenza. This process of mimicry is the result of convergent evolutionary pressure on the pathogens. Similarly, certain enveloped viruses surround themselves with host-derived glycans when they exit an infected cell, contributing to viral evasion of antibody-mediated neutralization (19, 20).

In chronic disease, such as cancer, autoimmunity, and inherited disorders, aberrant glycosylation can be an effective diagnostic and prognostic marker (21). However, the precise contribution of glycan changes to disease progression or even genesis in many of these pathologies remains unclear. Nonetheless, a definitive example of the role of glycans in chronic disease can be found in congenital disorders of glycosylation (CDGs). These compose a group of rare multisymptom inherited disorders caused by defective genes directly involved in the enzymes for glycan processing or in supporting systems, such as protein trafficking. Identification of the exact pathway defects have revealed the physiological importance of glycosylation to specific processes (22). Although the incidence of CDGs is low, the recent identification of dysfunctional glycan biosynthetic enzymes in forms of congenital muscular dystrophies (23) indicates that such conditions may be more widespread than previously realized.

In cancer, a near-universal feature of tumor cells is altered glycosylation relative to the normal tissue from which they derive (24). Many observations are correlative and must be treated with caution with respect to causation. Nonetheless, some elements of tumor biology, such as metastasis and cell survival, have been linked to the presence of certain cellular glycoforms. In effect, many tumor cells make use of existing glycan-mediated mechanisms regulating cell motility and survival. These mechanisms involve expression of discrete glycan-GBP partnerships that directly influence critical parameters of tumor cell survival, such as apoptosis (25), or allow multiple cellular-extracellular matrix interactions and thus tissue migration and invasion (26). It remains unclear at what point in the tumorigenic process aberrant glycoforms are acquired. Whether glycans are involved in the neoplastic transformation, are a consequence of it, or appear as a result of a tumor’s microenvironment is still debated. However, direct oncogene and tumor suppressor signaling to key glycan biosynthetic genes suggests a relatively early appearance in at least some cancers (24).

In autoimmunity, aberrant recognition of glycans by both B and T cells can drive pathology. An example of an exogenous trigger of such pathology is the induction of antibodies against glycolipids in Guillain-Barré and Miller-Fisher syndromes after intestinal Campylobacter jejuni infection (27). The C. jejuni lipopolysaccharide contains terminal glycan motifs that resemble self terminal glycan structures. Antibodies produced by the immune system in response to infection not only bind to the lipopolysaccharide structures but can also cross-react with self glycolipids containing these motifs. This cross-reactivity can cause damage to the peripheral nervous tissue where these glycolipids are found. Antibodies against glycolipids have also been identified in a wide range of autoimmune conditions, including multiple sclerosis and systemic lupus erythematosus.

Cellular presentation of glycopeptide and glycolipid antigens can be potent modulators of T cells and thereby contribute to autoimmune pathologies (28). In multiple sclerosis, autoreactive T cells have been identified that are restricted to brain sulfatides presented by CD1a proteins of dendritic cells (29). The CD1 family is expressed by antigen-presenting cells and displays self and exogenous glycolipid and lipopeptide antigens to a range of T cells. Glycopeptide antigens can be presented by the major histocompatibility complex (MHC) system and contribute to autoimmunity. Changes in glycosylation can influence tolerance either by constraining glycopeptide presentation in the MHC system or by creating neo-antigens (28). An example of the latter is the induction of rheumatoid arthritis by a glycopeptide fragment of type II collagen (30) in which an O-galactose at a hydroxyl-lysine is critical to T cell recognition (31).

Glycosylation can also influence autoimmunity by modulating the activity of key regulatory components. For example, immunoglobulin G (IgG) antibody–mediated inflammation is associated with specific antibody glycoforms (32). These autoantibodies form immune complexes (IC) with self-antigens that in turn drive inflammation via recruitment of complement and effector cells, subsequently leading to localized tissue damage. Specific IgG Fc glycoforms (Fig. 2) are associated with IC formation, binding of activatory IgG Fc γ-receptors (FcγR), complement activation, and severity of inflammatory response in various autoimmune diseases, including rheumatoid arthritis. Conversely, fully sialyated IgG Fc seems to promote an anti-inflammatory response (9).

Fig. 2 Antibody glycan engineering.

(A) Structure of the Fc-FcγRIIIa complex and that of the deglycosylated Fc after cleavage by EndoS (90, 97). Deglycosylation reduces receptor binding and is being investigated for the treatment of autoimmune diseases (32, 92, 93) and the production of therapeutic deglycosylated antibodies (95, 96). (B) Crystal structures of different glycoforms exhibiting a range of effector functions (89, 98100).

In summary, the above insights, highlighting the pivotal role played by glycans in various human pathologies, are now driving efforts toward their medical exploitation (Fig. 3).

Fig. 3 Schematic representation of the major points of therapeutic intervention of glycosylation.

ER/Golgi/plasma membrane (PM) glycosylation processing is shown with major intervention strategies indicated. Specific structures indicated are as follows: category 1, UV-4 (N-9-methoxynonyl DNJ); 2, kifunensine; 3, IgG; 5a, carbohydrate mimetic selectin inhibitor; 5b, Relenza; 6, M6P; 7, NB-DNJ (miglustat).

Glycosylation in Vaccine Design

Effective vaccines should elicit a robust IgG response with associated memory B cells. Although pathogen-induced antibodies against glycans (anti-glycans) are primarily low-affinity IgM, potent IgG responses can be achieved. This is typified by our high-affinity IgG response to the nonhuman α-galactose epitope, which is a major obstacle to xenotransplantation. Indeed, this antibody-mediated tissue rejection initially gave impetus in the development of antiglycan vaccination strategies against tumors (33).

To circumvent the issue of poor immunogenicity, glycan-based vaccines need to be conjugated to helper T cell epitopes such as keyhole limpet hemocyanin (KLH). Synthetic chemistry has greatly aided this process by the use of specific linkers that do not disturb the immunogenicity of the glycan being conjugated. Coupling synthetic conjugation with synthetic carbohydrate synthesis using techniques such as one-pot and automated oligosaccharide synthesis has greatly facilitated both vaccine design and testing (34). New conjugation technology aimed at boosting immunological efficacy is also in development. Examples include novel T cell epitopes, Toll-like receptors, and tetanus toxoids (35). One fascinating variant is the direct conjugation of target antigen to the α-galactose epitope that exploits our zoonotic blockade of this motif (36). Last, the discovery that glycopeptides can be presented by the MHC system has also opened up the possibility of generating cytotoxic T cell responses toward tumors. Vaccination that induces the effector functions of both humoral and cellular immune systems are currently being investigated (34).

Cancer vaccines based on tumor-associated glycans have received a significant amount of attention, and the epithelial mucin, MUC1, has been at the forefront of this work. MUC1 is both overproduced and aberrantly glycosylated in many forms of cancer (37). This abnormal mucin contains glycopeptide epitopes (e.g., O-glycan T, sialyl-Tn, and Tn) that can be immunogenic. Patients who mount a significant response to MUC1, particularly its glycan component, have a more favorable prognosis (38).

Attempts to elicit an effective anticancer immune response to MUC1 have exploited glycopeptides based on variable-number tandem-repeat peptides containing the O-glycan–rich region of MUC1. The MHC presents MUC1-derived glycopeptides containing cancer-associated Tn epitopes with greater efficiency than the unmodified peptide (39). Glycopeptides seem to elicit significant cellular responses and anti-glycan–mediated antibody-dependent cellular cytotoxicity (ADCC) (40). Multiple cancer vaccine trials are now in progress using specific synthetic MUC1 peptides and glycopeptides of varying peptide lengths and cancer-associated glycan structures (37, 40).

However, the expression of cancer-associated epitopes can vary dramatically in different patient groups. This may have contributed to the conclusion that a sialy-Tn/KLH vaccine did not impart significant survival benefit in a double-blind randomized phase III trial of 1028 women with metastatic breast cancer despite good seroconversion, specific IgG anti-sialyl-Tn titers, and promising phase II trials (41).

Sialylated glycolipids, termed gangliosides, have attracted considerable interest as potential targets of immunotherapy. However, vaccination strategies against gangliosides GD2 and GD3 have yet to yield positive clinical data (42). In contrast, the use of “nonself” tumor-associated gangliosides, such as N-glycolylneuraminic acid (Neu5Gc)–containing GM3, has yielded promising data (43). Neu5Gc is a modified sialic acid not made by humans because of a defective hydroxylase gene (44). Tumor-associated Neu5Gc is derived from the diet and is thought to promote localized weak chronic inflammation (45), which is now recognized as a necessary hallmark of cancer (46). There are currently two phase III vaccine trials taking place based on Neu5Gc including racotumomab (47), a murine monoclonal anti-idiotype.

Another prominent role for glycans in vaccine design is found in some immunotherapeutic approaches toward neutralizing enveloped viruses. HIV best exemplifies this field, where glycans constitute about half the mass of its attachment glycoprotein, gp120. Although the trimeric HIV gp120 attachment and the gp41 fusion spike are virally encoded, the glycan component is synthesized by the infected host cell. These glycans protect the virions from antibody neutralization. However, their very high density leads to impaired glycan processing and the generation of a population of oligomannose-type glycans, which are typically present at very low levels on cell-surface and secreted glycoproteins (48).

The viral glycan shield offers an attractive vaccine target because it is distinct from typical self-glycosylation. It is largely structurally conserved across clades relative to the underlying protein sequence (49), and broadly neutralizing antibodies (bnAbs) have been identified from some individuals that recognize the glycan shield (50). These features, together with the observation that passive transfer of these bnAbs offers significant protection to infection, suggest that a successful HIV vaccine may well have an anti-glycan component.

The first anti-glycan shield bnAb identified was IgG1 2G12, which is effective against a relatively broad range of antigenically unrelated HIV-1 isolates with the notable exception of the epidemiologically crucial clade C (51). Although 2G12 exhibits an unusual domain-exchanged structure (52), many other bnAbs (such as the PGT and the PG series) have now been isolated that conform to a more typical IgG structure and exhibit a wide neutralization breadth and potency (50). Unlike the epitope of 2G12, which is entirely oligomannose-type glycans, the PGT and PG series have a mixed glycan-peptide epitope (5355). In the case of PGT128, the breadth of the protein recognition is maintained by exploiting contacts with the peptide backbone.

The emergence of a large range of clonally unrelated bnAbs that recognize gp120 around the conserved N-glycan site at Asn322 (N332) raises the question of how different, unrelated antibodies can recognize the same site of vulnerability on the virion surface (55) The structures of 2G12, PGT128, and PGT135 in complex with outer domain fragments show highly distinct modes of binding, suggesting that there are multiple immunological solutions to the recognition of the glycan shield (Fig. 4). Furthermore, individual antibodies demonstrate substantial plasticity of recognition and tolerate some diversity in both the glycans recognized and shifts in the glycan location (56). One might argue that plasticity in recognition would lead to a greater likelihood of antibodies to this region being elicited by vaccination. However, an alternative view would argue that the lack of a single binding mode presents a challenge to rational vaccine design.

Fig. 4 Antibody recognition of the glycan shield of HIV.

Schematic representation of the viral spike of HIV based on cryogenic electron microscopy (beige) with the glycan shield (green) (56). The position of the glycan at N322 is shown together with the crystal structures of the Fabs of PGT128 (red), PGT135 (gray), and 2G12 (blue) (52, 54, 55). These antibodies recognize more than one glycan and, with the exception of 2G12, also bind the underlying protein. Further information on the relationship between gp120 glycans and 2G12 recognition is available in video format in the supplementary materials (movie S1).

Can bnAbs against the glycan shield serve as the basis for a rational vaccine design in HIV? Glycan-binding antibodies can be elicited in experimental models such as macaques infected with a pathogenic R5 strain of simian SIVADA8 (57) and yeast mannan immunization studies (58). Therefore, although many hurdles remain, it is certainly conceivable that an anti-glycan immune response will be an important component of a successful vaccine.

Therapeutic Disruption of Glycan-Protein Interactions

There is a wide spectrum of therapeutic opportunity that has arisen from the detailed knowledge of glycan-protein interactions in both the endogenous signaling systems and the life cycles of pathogens. This therapeutic area draws extensively on synthetic chemistry and emerging technologies, such as phage display, for the design and construction of high-affinity inhibitors (59, 60).

Many pathogens rely on specific glycan sequences as part of their infection strategy, and this has led to the concept of using these or their derivatives to impede infection. Small-scale trials have been reported that used free glycans to inhibit a range of microbes as well as several pathogenic viruses (17). Such receptor mimicry strategies often need to overcome the large avidity of the natural receptor generated by the sum of multiple presented low-affinity glycan binding sites. An example of a successful strategy that takes avidity considerations into account is the inhibition of shiga-like toxin of the Escherichia coli strain O157:H7 using a multimeric compound that has numerous glycan termini (17). Low-affinity glycan-protein interactions are also being targeted in development of anti-inflammatories, which antagonize selectins that mediate leukocyte extravasion through the recognition glycan epitopes such as sialyl-Lex. Although directly mimicking sialyl-Lex results in poor inhibition because of low affinity, this structure has been used as a lead candidate in the synthesis of higher affinity mimetics (59).

Synthetic mimics of glycans are also used as antivirals. Two of the four approved drugs against the influenza virus, zanamivir (Relenza, GlaxoSmithKline) (15) and oseltamivir (Tamiflu, Genentech) (14), are directed against the viral envelope glycoprotein neuraminidase, an enzyme that enables the release of viral particles from infected cells by cleaving sialic acid on the cell surface. Knowledge of its crystal structure led to the development of the drug Relenza, which is specific against influenza and does not inhibit mammalian neuraminidases (61). Tamiflu was developed as an orally available neuraminidase inhibitor, based on Relenza. Although some chemical features of Relenza were maintained, a range of modifications aimed at formulating it as a pill reduced its structural similarity to the natural substrate, a factor that may have contributed to the comparably faster emergence of viral escape mutants against this drug.

Faced with high mutation rates of viral target proteins, direct acting antivirals generally present low genetic barriers to resistance. However, targeting host cell pathways to raise the genetic barrier for viral escape mutants would circumvent this problem and potentially provide broader acting antivirals. Viruses rely on host cell processes, such as glycosylation, to enable cellular uptake, replication, secretion, and spread of infectious particles. For rapidly mutating RNA viruses like hepatitis C virus (HCV), HIV, influenza, and dengue, the concept of a generic therapy capable of dealing with current master species, as well as existing and potential mutants and reassortants, is appealing. Antivirals that could target several different virus species simultaneously could provide better treatment options for the increasing numbers of coinfected patients.

The eukaryotic ER has a protein quality-control machinery that monitors and enables the correct folding of nascent glycoproteins (8). After the sequential trimming of terminal glucoses attached to N-glycans by ER α-glucosidases I and II, the interaction of monoglucosylated glycoproteins with calnexin and/or calreticulin prevents folding intermediates from premature export, aggregation, or destruction. Once ER α-glucosidase II removes the final glucose residue, glycoproteins are liberated from the calnexin/calreticulin cycle. If the protein has failed to reach its native conformation at this point, it will be recognized by uridine diphosphate (UDP)–glucose glucosyltransferase, which transfers one glucose residue from UDP-glucose to regenerate Glc1Man9GlcNAc2, (where Glc is glucose; Man, mannose; and GlcNAc, N-acetylglucosamine) restoring recognition by calreticulin and calnexin.

Preventing the formation of mature, infectious virus by inhibiting the glycoprotein-processing pathway provides an attractive target for the development of broad-spectrum antivirals. Any virus that depends on calnexin-mediated folding of their envelope glycoproteins is vulnerable to this approach, and inhibitors of ER α-glucosidases, such as the iminosugars, have demonstrated in vitro antiviral efficacy against representative targets from nine viral families: Herpes-, Hepadna-, Flavi-, Toga-, Rhabdo-, Arena-, Orthomyxo-, Paramyxo-, and Retroviridae. Subsequent in vivo trials in small-animal disease models have shown higher than expected efficacy, notably for those viruses causing acute disease and featuring a pronounced cytokine involvement in disease progression, like influenza and dengue virus treated with N-9-methoxynonyl DNJ (UV-4) (62). A direct effect of ER α-glucosidase inhibition on (glycosylated) cytokines or their receptors needs to be evaluated. For viruses causing chronic disease, prevention of emergence of viral escape mutants when treating for many weeks or months is an anticipated and desired aspect of developing host cell–based therapeutic strategies. For HIV and HCV, such resistance-proof treatment has been reported in vitro (63).

Modulation of Endogenous Glycan Metabolism

Congenital defects in glycoprotein and glycolipid metabolism are increasingly treatable by circumventing the metabolic blockade using dietary supplementation of required monosaccharides, administration of required enzymes, specific enzyme inhibition, or chaperone rescue.

Of all the CDGs described to date (22), only two are currently treatable to some extent by dietary supplementation (substrate replacement therapy): Mannose supplementation relieves some symptoms in mannose-6 phosphate isomerase-CDG, which is caused by mutations in phosphomannose isomerase (6466). In another CDG, a guanosine diphosphate (GDP)–fucose transporter deficiency strongly reduced fucosylation of cell surface glycoproteins (64). This prevents selectins involved in leucocyte extravasation from binding, resulting in leukocytosis and increased sensitivity to infections. An intriguing finding is that the defective Golgi GDP-fucose transporter is not required for the therapeutic effect of exogenous l-fucose. The identity of an alternative transport system that would explain this conundrum remains elusive (64).

Enzyme replacement therapy (ERT) has emerged as an effect treatment for lysosomal storage diseases (LSDs). LSDs are caused by genetic mutations that result in defective enzymes affecting the activity or trafficking of lysosomal hydrolases. Gaucher patients suffer from a deficiency of the enzyme glucocerebrosidase, and, for severely affected patients, replacement of the deficient enzyme by intravenous infusion with recombinant analogs imiglucerase (Cerezyme, Genzyme) and velaglucerase (GA-GCB and VPRIV, Shire Human Genetic Therapies) is an effective treatment.

The generality of ERT is also shown by the use of α-galactosidase derivatives for treatment of Fabry disease (6769) and the use of α-l-iduronidase (laronidase, Aldurazyme, Genzyme) (70), iduronate-2-sulfatase (idursulfase, Elaprase, Shire Human Genetic Therapies), and glycosaminoglycan N-acetylgalactosamine 4-sulfatase (acrylsulfatase B, Naglazyme, BioMarin Pharmaceutical) to overcome enzyme deficiencies in various mucopolysaccharidoses storage diseases (71).

Monosaccharide analogs that partially inhibit specific biosynthetic enzymes are also the basis of several clinical treatments. The glucose analog N-butyldeoxynojirimycin (NB-DNJ) is an iminosugar that reversibly inhibits ceramide glucosyltransferase. This latter activity is the basis for its approval as an alternative approach to ERT in type I Gaucher’s disease, sometimes referred to as substrate reduction therapy. The ability of this iminosugar to cross the blood-brain barrier enables the treatment of LSD with neurological manifestations and is approved for the management of neurogenerative Niemann-Pick disease type C (72).

Inhibitors οf α-glucosidase are also being evaluated as an alternative therapeutic approach in the treatment of noninsulin-dependent diabetes mellitus (73). The iminosugar miglitol is a clinically approved treatment that works by inhibiting the α-glucoside hydrolase enzyme sucrase-isomaltase located in the brush border of the small intestine. This impairs carbohydrate digestion and slows glucose absorption into the blood stream, reducing postprandial hyperglycemia. Other structures under evaluation include triterpenes hybrids, pseudo-carbohydrates (acarbose and voglibose), and noncarbohydrate mimetics (73).

A seemingly paradoxical role of glucocerebrosidase inhibitors such as iminosugars is that of pharmacological chaperone therapy in the treatment of LSDs. Low doses of iminosugars such as isofagomine (Plicera, Amicus Therapeutics) partially rescue lysosomal ceramide glucosyltransferase activity in type I Gaucher’s disease (74). The most accepted explanation for this mechanism is that the reversible binding of the inhibitor to the misfolded enzyme in the ER enables sufficient folding around its active site to rescue it somewhat from degradation and allow it to traffic to lysosomes.

Remodeling Specific Glycoprotein Glycans for Therapy

Many therapeutic proteins are glycoproteins, and, although some are purified from natural sources, the majority are recombinantly expressed. The choice of expression system heavily influences the glycosylation. There have been notable efforts in controlling the glycosylation of glycoprotein production systems motivated by the potential immunological properties of nonself glycans, the impact of glycosylation pharmacokinetics, and the in vivo functionality (75).

The use of nonhuman mammalian cell lines has the potential to introduce immunogenic epitopes such as α-galactose and Neu5Gc (76), although Neu5Gc can arise in any mammalian system when Neu5Gc is present in the media. In plants, xylose and core N-glycan α1,3-fucose epitopes can arise, which are known immunogenic epitopes. Insect cell lines can also exhibit α1,3-fucose and naturally exhibit extensive mannose-terminating structures. Last, yeast N-glycans are typically extended immunogenic mannans. These features are important consideration in the production of biologicals. For example, the presence of α-galactose epitopes on cetuximab, a monoclonal from a murine cell line for use in metastatic colorectal cancer, has led to anaphylaxis in some patients (77).

Despite these observations, nonhuman cell lines such as Chinese hamster ovary (CHO) cells (which remain the industry standard), baby hamster kidney cells, murine myeloma, and murine hybridoma cell lines are prevalent production systems (76). However, the desire to control glycosylation and use a wider spectrum of nonhuman production systems has motivated important advances in glycoprotein engineering. Beyond the large panel of CHO cell glycosylation mutants that has been developed by mutagenesis and lectin selection, transgenically altered CHO cells enable both the elimination of potentially hazardous glycosyltransferases and the fine-tuning of glycosylation to manufacture desired glycoforms.

A range of methodologies has been established to remove the fucose of anticancer therapeutic monoclonal antibodies (mAbs) (78, 79), including use of stable cell lines deficient in the α1,6-fucosyltransferase; cell lines deficient in GlcNAc transferase-I, which stalls the biosynthesis of the glycans at Man5GlcNAc2; and inhibitors such as kifunensine, which can trap the N-glycans in afucosylated oligomannose glycoforms (11). Furthermore, consideration of the effect of sialylation on pharmacokinetics has led to optimization of the sialylation machinery for the production of highly sialylated glycoproteins (e.g., erythropoietin).

Plants display a remarkable tolerance toward manipulation of their intrinsic glycan biosynthetic pathways (80). This has allowed the removal of potentially immunogenic residues, and the addition of appropriate extension enzymes (e.g., β1,4-galactosyltransferases), alongside the more challenging terminators such as α2,6-sialyltransferse, has allowed a fine-tuning of the desired glycan population (80).

Yeast glycosylation machinery has been extensively remodeled, exemplified by work on Pichia pastoris, yielding a panel of cell lines that offer the capacity to generate different glycoforms (81). Interestingly, because this system often has significantly elevated glycosyltransferase activity, it has been possible to manufacture antibodies with a different range of complex-type glycans beyond the natural protein-imposed biantennary type, giving access to a portfolio of effector functionality beyond that which can be achieved by mutation alone (82).

Another approach uses chemoenzymatic methods to remodel the protein glycosylation after expression (75). This approach enables chemical control of glycosylation. The control potentially afforded by this approach could have an important impact, for example, in the fine-tuning of antibody effector functions (83).

Glycan engineering has also lead to advancements in enzyme replacement therapies. Substantial mechanistic challenges in ERT treatment of LSD are targeting the affected cells or tissue and delivering the enzyme intracellularly to the lysosomes. Recombinant enzymes can be produced in CHO cells and their glycans enzymatically remodeled to include mannose-6-phosphate (M6P), critical for their targeting to lysosomes via the M6P receptor (84). A mannose-terminating glycoform also helps other issues such as macrophage targeting (e.g., in Gaucher’s disease). Uptake is mediated by mannose receptors, and transport across the blood-brain barrier is boosted through interaction with M6P/insulin-like growth factor 2 (85). Other expression systems are currently being explored, including human fibrosarcoma cells, egg whites, transgenic animals, and plant cells (86).

mAbs offer great promise in the treatment of disease by their ability to specifically bind targets through the Fab domains while also possessing the ability to recruit the cellular and humoral immune system through the Fc domain. The glycosylation of IgG, the predominant class of antibody used therapeutically, modulates the binding to the FcγR family of cellular immune receptors (87). Although Fc glycans are critical to many antibody effector functions, receptor binding can be restored by the compensatory addition of Fc amino acid substitutions (88). However, to date the majority of monoclonals under clinical trial consideration are glycosylated.

In the case of cellular targets such as cancerous cells, the immobilization of antibody across a cell surface flags the cell for immune recognition, even in cases where the primary mode of action could be ascribed to the physical mAb binding, such as the blocking of a growth factor receptor. In addition, even mAbs that target soluble ligands, such as the growth factors, can rely on their Fc domains because it is through FcγR recognition that the growth factors are cleared for destruction. Given the importance of Fc-FcγR interactions, augmenting the binding of the Fc domain to the cellular FcγRs can enhance the efficacy of many proinflammatory antibodies.

The IgG Fc glycans are typically of the biantennary complex type, exhibiting high levels of fucosyslation of the core GlcNAc residue, partial galactosylation, and bisecting GlcNAc. Of these structures, about <20% are sialylated. The reason for the low levels of branching and terminal structures are the constraints on glycan processing imposed by the protein. The two opposing glycans within the homodimeric Fc lie along the surface of the Cγ2 domain, and they form extensive hydrophobic interactions with the apolar faces of the saccharide residues (Fig. 2). These interactions are maximal upon biosynthetic conversion from oligomannose and hybrid-type structures to the complex type (89). Importantly, the fucose residue is highly solvent exposed in the unliganded form but clashes with the glycan of one of the key receptors, FcγRIIIA, the main FcγR of natural killer cells. This clash reduces FcγRIIIA binding and limits ADCC of fucosylated antibody glycoforms (90).

Glycosylation can also contribute to the anti-inflammatory properties of antibodies. High-dose (1 to 2 g/l) intravenous administration of IgG, known as IVIg, is an effective therapeutic treatment for autoantibody-mediated inflammation (10). IVIg is prepared from plasma derived from large numbers of healthy donors. An anti-inflammatory component has been identified as a subset of this IgG population bearing α2,6-sialylated Fc glycoforms. Indeed, the efficacy of sialylated Fc provides a route to the clinic of recombinant immunoglobulin-based anti-inflammatories (9). The precise anti-inflammatory mechanism of sialylated Fc remain a subject of debate (91).

Last, enzymatic deglycosylation of IgG provides several therapeutic opportunities. Deglycosylation of IgG significantly decreases binding of antibodies to FcγRs. Although several endoglycosidases have been discovered capable of cleaving the Fc glycans (Fig. 2), only endoglycosidase S (EndoS) is uniquely specific to IgG (92). This allows the possibility of selective deglycosylation of antibodies in vivo to treat patients with antibody-mediated autoimmunity (93). In addition to specificity for IgG at the protein level, EndoS only cleaves biantennary glycans (83). EndoS is secreted by Streptococcus pyogenes as part of its immune evasion strategy. Perhaps as a consequence of its effect on antibodies, EndoS does not appear to be significantly immunogenic in animals. EndoS has shown promise in the treatment of variety of autoimmune conditions, including systemic lupus erythematosus (94).

EndoS also offers a route to improving the efficacy of therapeutic antibodies. EndoS can be used to generate deglycosylated antibodies, which exhibit immunosuppressive properties (95, 96). Serum IgG is present at concentrations of around 10 mg/ml and results in the almost-complete saturation of cellular FcγRs. By stripping the receptors of endogenous IgG with EndoS, the receptor binding activity of mAbs can be dramatically enhanced if the mAb is introduced either after EndoS activity has been cleared from the serum or by the expression of the mAb as a glycoform resistant to EndoS hydrolysis, such as oligomannose- or hybrid-type glycoforms (97).


A broad spectrum of opportunities for the medical exploitation of glycosylation has emerged. Importantly, several clinically effective therapeutics have already been developed. Intervening in cellular glycosylation pathways provides a route to the alleviation of many of the symptoms of congenital metabolic disorders. Some of these same drugs also have an impact on glycan-mediated virion assembly and offer an exciting prospect for the development of broad-spectrum antivirals against enveloped viruses. Further stages of the viral replicative cycle can be disrupted by considering their dependence on glycosylation, and this currently forms the basis of anti-influenza drugs and offers a route to the development of a new class of anti-inflammatory small-molecule drugs. The development of therapeutic glycoproteins has been greatly stimulated by the advances in recombinant cellular biosynthetic technologies that can produce defined glycoforms. mAbs with engineered glycosylation have emerged that display enhanced in vivo properties and have reached the clinic. Serum antibodies involved in autoimmunity can be inactivated by removal of their glycans by bacterial immune evasion factors, and this technology shows great promise in preclinical studies. Last, glycopeptides offer intriguing possibilities in the development of anticancer vaccines given their ability to stimulate both humoral and cellular immunity. In addition, the HIV glycan shield is proving to be an effective target for antibody neutralization. Design strategies that target conserved glycan clusters could contribute to the next generation of HIV vaccine candidates.

  • This article is dedicated to Chris Scanlan, who passed away after a short battle with cancer during the writing of this article.

References and Notes

  1. Acknowledgments: Chris Scanlan passed away after a short battle with cancer during the writing of this article. He was a valued and much admired colleague. M.D. and M.C. are supported by the International AIDS Initiative (IAVI), the Scripps Center for HIV/AIDS Vaccine Immunology and Discovery (CHAVI-ID), and the Medical Research Council. M.C. is a Fellow of Oriel College, Oxford. N.Z. is a Research Fellow of Merton College, Oxford, and is also supported by United Therapeutics Corporation, USA, and the Oxford Glycobiology Institute. We thank T. Bowden for helpful discussions and M. Wormald for the video of gp120. M.C. and C.N.S. have made a patent application titled “Combined therapeutic use of antibodies and endoglycosidases” (WO2013110946; international publication date 1 August 2013).
View Abstract

Navigate This Article