Optimizing Glycans of Therapeutic Glycoproteins for Prolonged Serum Half-life

Erythropoietin is perhaps the most successful biotechnology product to date. It is a circulating cytokine that binds to the erythropoietin receptor, inducing proliferation and differentiation of erythroid progenitors in the bone marrow. Thus, it has much value in treating anemias caused by lack of erythropoietin (e.g., renal failure) or by bone marrow suppression (e.g., after chemotherapy). Natural and recombinant forms of erythropoietin carry three sialylated complex N-glycans and one sialylated O-glycan. Although in vitro the activity of deglycosylated erythropoietin is comparable to that of the fully glycosylated molecule, its activity in vivo is reduced by about 90%, because poorly glycosylated erythropoietin is rapidly cleared by filtration in the kidney. Undersialylated erythropoietin is also rapidly cleared by galactose receptors in hepatocytes and macrophages (see Chapter 31). These problems can be reduced by having fully sialylated chains and by increasing the amount of tetra-antennary branching. This, in turn, increases its activity in vivo nearly tenfold. Another approach is to deliberately add an N-glycosylation site to increase half-life and activity in vivo. Covalently linking polyethylene glycol to the protein also reduces clearance by the kidney.

Erythropoietin is somewhat unusual because it is small enough to be cleared by the kidney if it is underglycosylated. For most glycoprotein therapeutics, a more important consideration is minimizing clearance by galactose-binding hepatic receptors by ensuring full sialylation of glycans. Because glycans can have such dramatic effects on the properties of these drugs, it is important to ensure that glycosylation is controlled during their production, as well as to satisfy regulatory requirements for batch-to-batch product consistency. Changes in culture pH, the availability of precursors and nutrients, and the presence or absence of various growth factors and hormones can each affect the extent of glycosylation, the degree of branching, and the completeness of sialylation. Sialidases and other glycosidases that are either secreted or released by dead cells can also cause degradation of the previously intact product in the culture medium. Thus, during the development of a glycoprotein drug, considerable effort is devoted to defining appropriate conditions for reproducibly producing relatively homogeneous and complete glycosylation.

Impact of Glycosylation on Licensing and Patentability of Biotherapeutic Agents

Modern patenting of new therapeutics typically requires definition of the composition of matter in the claimed molecule. This is relatively straightforward for small molecules of defined structure and for nonglycosylated proteins. However, with regard to glycoproteins, especially those with multiple glycosylation sites, it is virtually impossible to obtain a single preparation that contains only a single glycoform. Thus, most biotherapeutic agents that are glycoproteins consist of a mixture of glycoforms. Licensing bodies such as the U.S. Food and Drug Administration have come to recognize this and will allow for a certain range of variation in glycoforms and the complexity of the mixture. However, the manufacturer and the agency must agree on the extent such variation is acceptable for a given drug formulation. Biopharmaceutical companies therefore spend considerable effort in assuring that their products fall within these defined ranges, once these are approved by licensing bodies. The inherent difficulty in reproducing complex glycoform mixtures also complicates efforts to make generic forms of recombinant glycoprotein drugs. Given the complexities of producing glycotherapeutic agents in mammalian cells, even the smallest changes in growth conditions (temperature, pH, media, etc.) can have significant effects on the range of glycoforms found in any given batch of the product. In fact, the licensing agencies also use consistency in glycoform composition as an indirect measure of the quality of process control in production. Differences in glycosylation can even have implications for the patentability of agents in which the underlying polypeptide remains constant. Marked differences in glycosylation have been used to define agents as being uniquely different. However, it is usually necessary to show that the differences in glycosylation being claimed also have a significant effect in changing the functionality of the drug in question. The associated pharmaceutical licensing and legal issues are rapidly evolving to keep pace with scientific advances in this area.

Significance of Nonhuman Glycosylation on Biotherapeutic Products

Certain types of nonhuman glycosylation, such as Galα1-3Gal (“α-Gal”) units and/or the sialic acid N-glycolylneuraminic acid (Neu5Gc), are found in terminal positions of glycans on some biotherapeutic glycoproteins. The addition of α-Gal residues can only occur if the therapeutic product is produced in a nonhuman cell line that expresses an α1-3Gal transferase not present in humans (see Chapter 13). The addition of Neu5Gc occurs when the nonhuman cell line used produces this sialic acid and/or because the animal-derived products used in the media provide a metabolic source of Neu5Gc, which can be taken up, processed, and made available for sialylation reactions (see Chapter 14). Other glycans unique to nonhuman expression systems may exist as well. Since humans have circulating antibodies against α-Gal termini, Neu5Gc, and presumably other nonhuman glycan determinants, a potential for antigen–antibody responses exists that could be deleterious and/or affect therapeutic efficiency. An optimal solution is to use cell lines that do not express α-Gal or Neu5Gc (e.g., CHO cells, which express no α-Gal and only low levels of Neu5Gc), to eliminate Neu5Gc from materials used in tissue culture, and to carefully analyze the glycans present in recombinant protein therapeutics. Further studies of the effects of these antigenic modifications on half-life and pharmacokinetics are needed.

Salvage versus De Novo Synthesis

All monosaccharides needed for cellular glycan synthesis can be obtained from glucose through metabolic interconversions (see Chapter 4). Alternatively, monosaccharides can be derived from the diet or salvaged from degraded glycans. The relative contributions of different sources can vary with the cell type. For instance, even though all mammalian cells use sialic acid, only some contain high amounts of UDP-GlcNAc epimerase/N-acetylmannosamine kinase (GNE), which is required for the de novo synthesis of CMP-sialic acid. But sialic acid salvage from degraded glycans is quite efficient, decreasing the demand on the de novo pathway. Similarly, galactose, fucose, mannose, N-acetylglucosamine, and N-acetylgalactosamine can come from the diet or be salvaged for glycan synthesis, whereas glucuronic acid, iduronic acid, and xylose cannot. All monosaccharides derived from the diet or degraded glycans can be catabolized for energy, and again, cells vary in their reliance on the different pathways.

The variable contributions of these pathways are important for therapy of some diseases. For instance, patients with congenital disorder of glycosylation type Ib (CDG-Ib), who are deficient in phosphomannose isomerase, benefit greatly from oral mannose supplementation to bypass the insufficient supply of glucose-derived mannose-6-phosphate. A few CDG-IIc patients have been treated with fucose to restore synthesis of sialyl Lewis^x on leukocytes (see Chapter 42). Some patients with Crohn’s disease show clinical improvement with oral N-acetylglucosamine supplementation, but the mechanism is unknown. Mice deficient in GNE activity have kidney failure, but providing N-acetylmannosamine in the diet prevents this outcome. Clinical trials using N-acetylmannosamine to treat GNE-deficient patients with hereditary inclusion body myopathy type II (HIBM-II) are proposed, although it is unclear whether sialic acid deficiency is responsible for the human muscle pathology in HIBM-II.

Special Diets

Some monosaccharides and disaccharides can be toxic to humans who lack specific enzymes. For example, people who lack fructoaldolase (aldolase B) accumulate fructose-1-phosphate, which ultimately causes ATP depletion and disrupts glycogen metabolism. Prolonged fructose exposure in these people can be fatal, and fructose-limited diets are critical. Deficiencies in the ability to metabolize galactose (see Chapter 4) are mostly due to a severe reduction in galactose-1-phosphate uridyl transferase activity and cause galactosemia. Although these patients are asymptomatic at birth, ingesting milk leads to vomiting and diarrhea, cataracts, hepatomegaly, and even neonatal death. Low-galactose or galactose-free diets can prevent these life-threatening symptoms. However, even these diets do not prevent unexplained long-term complications, which include speech and learning disabilities and ovarian failure in almost 85% of females with galactosemia.

Infants hydrolyze lactose (Galβ1-4Glc) quite well, but the level of intestinal lactase can be much lower or absent in adults because of down-regulation of lactase gene expression. About two thirds of the human population has lactase nonpersistence, making milk products a dietary annoyance. This is because unabsorbed lactose provides an osmotic load and is metabolized by colonic bacteria, causing diarrhea, abdominal bloating and pain, flatulence, and nausea. Lactase persistence has evolved in certain pastoral populations from northwestern Europe, India, and Africa, allowing milk consumption in adult life. However, many adults either avoid lactose-containing foods or use lactase tablets to improve lactose digestion.

Substrate Reduction Therapy

The failure to turn over glycans by lysosomal degradation causes serious problems for patients with lysosomal storage disorders. Deficiencies in individual lysosomal enzymes lead to pathological accumulation of their substrates in inclusion bodies inside the cells (see Chapter 41). One approach to treating these disorders is to inhibit initial glycan synthesis, a strategy termed substrate reduction therapy (SRT). Reduced synthesis of the initial compound decreases the load on the impaired enzyme, and some patients show significant clinical improvement. One example of a drug used for SRT is N-butyldeoxynojirimycin (or N-butyl-DNJ) (Miglustat, Zavesca^®), which has shown some efficacy for treating Gaucher’s disease (glucocerebrosidase deficiency).

Lysosomal Enzyme Replacement Therapy

Another approach for treating lysosomal storage disorders is enzyme replacement therapy. Unlike most therapeutic glycoproteins that interact with target receptors on the surface of cells, lysosomal enzymes developed for replacement therapy must be delivered intracellularly to lysosomes, their site of action. During the normal biosynthesis of lysosomal enzymes, their N-glycans become modified with mannose-6-phosphate (Man-6-P) residues, which target them to lysosomes using Man-6-P receptors (see Chapter 30). The challenge for enzyme replacement therapy is to get the enzymes targeted properly to lysosomes, where they can degrade accumulated substrate. A special case is enzyme replacement therapy for Gaucher’s disease. In this case, the enzyme has been successfully targeted to lysosomes of macrophages through their cell-surface mannose receptor (see Chapter 31). This approach required modification of the complex N-glycans of the enzyme isolated from placenta to expose terminal mannose residues, a process that involved digestion with various glycosidases (sialidase, galactosidase, and hexosaminidase). Nowadays, recombinant enzyme is produced in Lec1 mutant CHO cells (see Chapter 46), in which N-glycans with terminal mannose residues are present at all N-glycan sites.

The success of glucocerebrosidase treatment stimulated the development of lysosomal enzymes for treatment of other lysosomal storage diseases. Clinical trials have demonstrated clinical benefit in Fabry’s disease, mucopolysaccharidoses type I, II, and VI, and Pompe’s disease. However, the usefulness of replacement therapy is limited by the fact that injected enzymes do not have beneficial effects on all aspects of these diseases, and existing tissue damage is usually not reversible. Moreover, as Man-6-P receptors do not function in all cell types, N-glycan termini are needed to target all the affected cell types. In all instances, an additional challenge is to deliver enzymes efficiently to cells of the nervous system past the blood-brain barrier.

Chaperone Therapy

A third approach for treating lysosomal storage disorders takes advantage of the fact that some genetic defects lead to misfolding of the encoded enzyme in the endoplasmic reticulum (ER). There is evidence that low-molecular-weight molecules that are competitive inhibitors of some of these enzymes can act as “chaperones,” stabilizing the folded enzyme in the ER and effectively rescuing the mutation. The result is a higher steady-state concentration of active enzyme in the lysosome. Of course, the dose of such an inhibitor must be carefully adjusted to ensure that the inhibitory effects on enzyme function do not overshadow beneficial effects on folding. Fortunately, only a low level of enzyme restoration is needed to significantly reduce the accumulation of undigested glycan substrates, indicating that lysosomal hydrolases are normally present in large catalytic excess.

Glucosamine and Chondroitin Sulfate

Glucosamine (often mixed with chondroitin sulfate) has been promoted to relieve symptoms of osteoarthritis, which involves the age-dependent erosion of articular cartilage. Cartilage provides a cushion between the bones to minimize mechanical damage, and a net loss of cartilage occurs when the degradation rate exceeds the synthetic rate. A number of clinical trials report that glucosamine improves osteoarthritis symptoms, and some claim to restore partially the structure of the eroded cushion, in particular in the knees. Superficially, this would seem to make sense, because primary glycans of cartilage include hyaluronan (see Chapter 15) and chondroitin sulfate, both of which contain hexosamines within their structure (see Chapter 16). However, there are conflicting reports, and the outcome may depend on study design and the type and source of material (e.g., chloride vs. sulfate salts and the overall level of purity). Nevertheless, veterinarians have treated animals with glucosamine for over two decades with apparently positive results. Furthermore, double-blind, placebo-controlled studies in humans have shown a decreased rate of joint space narrowing. It should be noted that glucosamine might also alter UDP-GlcNAc and potentially UDP-GalNAc levels, thus affecting cellular responses involving any major class of glycans.

Positive effects of chondroitin sulfate on osteoarthritis are less well-documented. Indeed, how the acidic chondroitin sulfate polymer can be absorbed and delivered to its proposed site of action remains an open question. Further studies are needed to determine how dosage affects the circulating levels of these supplements, whether they are absorbed by the target tissue, and if they actually lead to changes in cartilage metabolism.

Xylitol and Sorbitol in Chewing Gum

Many studies suggest that chewing gum containing sugar alditols, specifically xylitol and sorbitol, can help control the development of dental caries. Mothers who chew xylitol-sweetened gum may even block transmission of caries-causing bacteria to their children. The benefit of these reduced sugars seems to be based on stimulation of salivary flow, but an antimicrobial effect is also possible. Xylitol also inhibits the expression and secretion of proinflammatory cytokines from macrophages and inhibits the growth of Porphyromonas gingivalis, one of the suspected causes of periodontal disease. Children who drank xylitol solutions also had a lower occurrence of otitis media.

Milk Oligosaccharides

Human milk contains about 70 g/liter of lactose and 5–10 g/liter of free oligosaccharides. More than 130 different glycan species have been identified with lactose at the reducing end, including poly-N-acetyllactosamine units. Some glycans are α2-3- and/or α2-6-sialylated and/or fucosylated in α1-2, α1-3, and/or α1-4 linkages. In contrast, bovine milk, the typical mainstay in human infant formulas, contains much smaller amounts of these glycans. These differences may account for some of the physiological advantages seen for breast-fed versus formula-fed infants. The glycans may also favor growth of a nonpathogenic bifidogenic microflora and/or block pathogen adhesion that causes infections and diarrhea. Surprisingly, a substantial number of human milk oligosaccharides remain almost undigested in the infant’s intestine and are excreted intact into the urine. Whether supplementing infant formula with specific, biologically active free glycans enhances infant health is unknown.

Chitosan

Chitosan (partially de-N-acetylated chitin, a polymer of β-linked N-acetylglucosamine residues) is a mixture of different-sized water-soluble polymers and has been reported to have hypocholesterolemic, antimicrobial, and weight-reduction effects. Combined analyses from various studies show a slight benefit in body weight reduction, but this benefit is less pronounced in more rigorous evaluations. The effects are likely to be of limited clinical significance.

Microbial Vaccines

Vaccines consisting solely of glycan components typically elicit poor immunity, especially in infants. The primary limitation is that glycans are T-cell-independent antigens and therefore do not effectively stimulate T-helper-dependent activation and class switching of B-cell-mediated immunity. Conjugate vaccines with oligosaccharides coupled to carrier proteins have proven to be highly effective. For example, Haemophilus influenzae type b (Hib) causes an acute lower respiratory infection among young children. A conjugated form of an Hib-derived oligosaccharide coupled to a protein carrier is now routinely given to infants in developed countries and has resulted in a more than 95% decrease in incidence of infections in vaccinated populations. New vaccines are being developed using conjugated components of other bacterial capsular polysaccharides, and the studies appear promising.

Cancer Vaccines

Several carbohydrate-based vaccines are under development to treat cancer. Some of these vaccines are based on ganglioside immunogens present on certain types of cancer cells (e.g., gangliosides GM2 and GD2 in melanomas and globo H in breast cancer). In some cases, vaccines are composed of glycan-based haptens conjugated to a protein carrier. In one case, a breast cancer antigen sialyl-Tn (sialylα2-6GalNAcα-) is synthesized chemically and conjugated to a protein carrier (keyhole limpet hemocyanin). Sialyl-Tn is found on cancer mucins, and its expression correlates with progression to metastatic disease (see Chapter 44). A more direct strategy thus consists of attaching the sialyl-Tn units in their natural linkages to serine or threonine residues of the Muc-1 mucin polypeptide repeat. Although several cancer vaccine candidates are presently under investigation, they have yet to reach clinical utility.

Blocking Infection

As discussed in Chapter 34, many microbes and toxins bind to mammalian tissues by recognizing specific glycan ligands. Thus, small soluble glycans or glycan mimetics can be used to block the initial attachment of microbes and toxins to cell surfaces (or block their release), and thus prevent or suppress infection. Because many of these organisms naturally gain access through the airways or gut, the glycan-based drugs can be delivered directly without the requirement of being distributed systemically. Examples of such applications currently under study include milk oligosaccharides that are believed to be natural antagonists of intestinal infection in infants (see above) and polymers that will block the binding of viruses such as influenza. Although backed by a strong scientific rationale and robust in vitro studies, such “antiadhesive” therapies have not yet found much practical application.

Inhibition of Selectin-mediated Leukocyte Trafficking

If specific glycan-protein interactions in vivo are responsible for selective cell–cell interactions and a resulting pathology, then infusion of small-molecule analogs or mimetics of the natural ligand could theoretically be useful. The best-studied example is the selectin-mediated recruitment of neutrophils (and other leukocytes) into sites of inflammation or ischemia/reperfusion injury, which involves specific selectin–glycan interactions occurring in the vascular system (see Chapter 31). Initial approaches focused on the use of sialyl Lewis^x derivatives. However, the tetrasaccharide is a difficult synthetic target and has poor oral availability and a short serum half-life. To improve these shortcomings, attempts were made to design “glycomimetics,” which are compounds that preserve the essential functionality of the parent tetrasaccharide but eliminate unwanted polar functional groups and synthetically cumbersome glycan components. An example of the design of a monosaccharide glycomimetic starting from sialyl Lewis^x is shown in Figure 51.3. First, the sialic acid residue was replaced with a charged glycolic acid group, the N-acetylglucosamine residue was then replaced with an ethylene glycol linker, and finally the galactose residue was replaced with a linker moiety. The resulting glycomimetic had E-selectin binding affinity comparable to sialyl Lewis^x but with a simpler structure.

FIGURE 51.3

Glycomimetic E-selectin inhibitors based on sialyl Lewis^x.

Simple monovalent compounds such as sialyl Lewis^x and its glycomimetics have proven effective in various animal inflammatory models. This finding is somewhat surprising, because the binding constants for some of these monovalent ligands (with K_d values in the millimolar range) are much poorer than that for the physiological ligand. In fact, similar studies in humans did not show a clear benefit. Studies of the PSGL-1 (P-selectin glycoprotein ligand-1) glycosulfopeptide that has high affinity for P- and L-selectin are presently being pursued, as well as dendrimers consisting of multivalent forms of Lewis antigens. These conjugates have considerably longer serum half-lives than the monovalent glycans and glycomimetics. In addition, the inhibitory effects of heparin on P- and L-selectin are being considered. Finally, an alternate approach consists of finding specific inhibitors of the enzymes involved in forming endogenous ligand determinants in vivo. For example, the fucosyltransferase involved in sialyl Lewis^x biosynthesis (FUT7) is an attractive target.