An official website of the United States government
NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Varki A, Cummings RD, Esko JD, et al., editors. Essentials of Glycobiology [Internet]. 4th edition. Cold Spring Harbor (NY): Cold Spring Harbor Laboratory Press; 2022. doi: 10.1101/glycobiology.4e.8
A multitude of glycosyltransferases (GTs), glycoside hydrolases (GHs), other enzymes, and nucleotide sugar transporters are required to synthesize and metabolize glycans. Also, many genes encode glycan-binding proteins (GBPs), which recognize specific glycan structures. This chapter provides a genomic perspective of the genes that code for GTs, GHs, and GBPs.
The glycome comprises all the glycan structures synthesized by an organism. It is analogous to the genome, the transcriptome, and/or the proteome but even more dynamic, and it has higher structural complexity that has yet to be fully defined. Cells of different types synthesize a subset of the glycome based on their differentiation state and physiological environment. The human and mouse glycomes have many glycan structures in common, but a few are unique or have divergent functional properties. For example, unlike humans, rodents synthesize cytidine monophospho-N-glycolylneuraminic acid (CMP-Neu5Gc), for the transfer of Neu5Gc to N- and O-glycans (Chapter 15). Similarly, the gene encoding α-1,3-galactosyltransferase (A3GALT2) is functional in the mouse but not in human (Chapter 20). The human and fly genomes include orthologous genes encoding GTs that catalyze the same reaction, but they also have GTs that are unique. Thus, protein O-fucosyltransferase 1 (POFUT1) in mammals and Ofut1 in flies transfer fucose to Notch receptors and are examples of an evolutionarily conserved GT. In contrast, flies do not make complex N-glycans with four branches, which are common in mammalian glycoproteins (Chapters 9 and 20). Additionally, flies make unique glycolipids absent from mammals that are important for conserved signaling pathways mediated by the epidermal growth factor (EGF) receptor or Notch receptors (Chapter 26).
The genome encodes all of the enzymes, transporters, and other activities necessary to construct and regulate the glycome of an organism. There were few complete genomes available in 1999 for the first edition of this book, approximately 650 genomes by the second edition (2009), 25,000 by the third edition (2015), and in August 2021 there were 259,000 permanent draft genomes in the Genomes OnLine Database (GOLD) of which more than 22,000 finished genomes are included in the manually curated Carbohydrate-Active enZYmes (CAZy) database. Likewise, the number of GTs with a known three-dimensional structure has grown from one in 1999 to 158 in 2015 and 282 in 2020. The CAZy database is dedicated to (i) coping with the staggering increases in sequences released in GenBank by the NCBI (National Center for Biotechnology Information), (ii) creating new families based on the literature, and (iii) reporting new functions/substrate specificities and 3D structures within existing families.
In the pre-genomic era, the glycobiology of mammals, invertebrates, plants, bacteria, and viruses did not overlap extensively, and progress in one domain did not immediately benefit others. With several genomes being released each day, the evolutionary history of GT and GH sequences has emerged. We now know that GTs from various organisms display the same basic structural folds, and we can harness the relationships between the sequence and the specificity of a GT (Chapter 6). The content of genomes can also be examined from a glycobiology perspective (e.g., by listing candidate GHs or GTs in a genome) and compared across genomes to see which families have expanded or disappeared during evolution. Examination of completely sequenced genomes suggests that a few percent of any genome encodes GTs and GHs. The number of GT genes in different organisms is variable, but varies much less within a taxonomic clade. The number of GTs tends to be greater than the number of GHs, except for organisms that forage on complex glycans as a carbon source. Thus, the genomes of saprophytic fungi and bacteria of the intestinal flora can encode several hundred enzymes for the breakdown of glycans (i.e., five to 10 times more GHs than the number of GTs). The number of genes that encode GBPs is more difficult to define because many have other functional domains and have not been annotated as GBPs. A conservative estimate is that mammalian genomes encode 100 to 200 GBPs.
The genome of each new organism is now routinely searched for genes that show similarity to known GTs, GHs, or GBPs. To annotate a new gene, its sequence is compared to previously annotated genes (pairwise alignments) or gene families (model alignments). Sequence similarity is used as a proxy for homology, to infer the biochemical activity of the predicted protein based on its distance from previously characterized proteins. GTs can be highly versatile, and GHs, despite being more specific in terms of the substrates they recognize, can form large families with various activities. Thus, although it can be difficult to predict the precise reaction catalyzed by a GT or GH on the basis of its sequence relatedness to biochemically characterized enzymes, it is often possible to predict the anomeric linkage of the sugar transferred or hydrolyzed, respectively, or the broad glycan category targeted. Prokaryotic GHs have similarities to those of eukaryotes, and with some exceptions, prokaryotic and eukaryotic GTs have one of three folds, GT-A GT-B or GT-C, indicating three ancestors for essentially all GTs (Chapter 6). Additional information can also be obtained from the sequence of a new gene by searching for conserved motifs such as signal peptides, transmembrane domains, glycosylphosphatidylinositol (GPI) anchors (Chapter 12), or carboxy-terminal retrieval sequences.
Carbohydrate-active enzymes can be classified according to several criteria. Substrate specificity forms the simplest basis to assign an Enzyme Commission (EC) number by the International Union of Biochemistry and Molecular Biology (IUBMB). GHs are given the code EC 3.2.1.x, where x represents substrate specificity. Similarly, GTs are described as EC 2.4.y.z, where y defines the sugar transferred and z describes the precise donor and acceptor. The EC number is given only after experimental determination of enzymatic specificity. Currently there are 462 EC numbers describing the known activities of GTs and 213 for GHs.
The intrinsic problem with a classification system based on substrate (or product) specificity is that it does not appropriately accommodate enzymes that act on several substrates. It also fails to reflect the sequence or the structural features of these enzymes in relation to evolution. To circumvent these problems, a novel system was introduced for classifying GHs and GTs based on the relationship between amino acid sequence and folding similarities. Regardless of activity and substrate specificity, sequences that display similarity are grouped in the same family. For GTs, such classification was initiated in 1997 (approximately 500 sequences; 27 families). This classification is continuously updated in the CAZy database, which listed approximately 850,000 GT sequences in approximately 110 families in August 2021. CAZy gives access to the various families of GTs and GHs, but also polysaccharide lyases and their carbohydrate-binding modules, as well as several families of auxiliary activities, such as the recently described lytic polysaccharide monooxygenases. Each CAZy family is annotated with known enzyme activities and often includes catalytic and structural features. This summary is followed by a list of the proteins and open reading frames (ORFs) belonging to the family, with links to sequence and structural information in public databases. CAZy also features summary pages for approximately 22,000 publicly available genomes (∼95% of bacterial origin) in August 2021.
The earliest observation from the sequence-based families was that many are “polyspecific” and contain enzymes of different substrate specificity. Polyspecific families indicate that (1) the acquisition of new specificities by GHs and GTs is a common evolutionary event, (2) their substrate specificities can be engineered for experimental or applied purposes, and (3) their substrate (or product) specificities are governed by fine details of three-dimensional structure, not by global fold. Human GTs with experimentally determined activities are compiled in several excellent resources at Kyoto Encyclopedia of Genes and Genomes (KEGG), UniProt, and the GlycoGene Database (GGDB) in ACGG-DB (Chapter 52). In contrast, assignments for other organisms, like microbes, are often erroneous, because of misassignment of EC numbers based on too-distant sequence relatedness. The challenge of the postgenomic era is to characterize this ever-growing list of ORFs whose encoded proteins are candidate GTs of unknown donor, acceptor, and product specificities.
The information presented by the wide variety of glycans on glycoconjugates is deciphered by an equally versatile number of GBPs that recognize specific sugars, glycans, or glycopeptides (Chapters 28–38). To understand the biology behind protein–glycan interactions, it is imperative to identify all GBPs and their glycan ligands.
GBPs were identified in the past by systematic biochemical studies that determined their glycan-recognition properties. However, the recent explosion of sequenced genomes makes it possible to identify genes likely to encode GBPs by sequence similarity. For example, the mannose-binding lectins (MBLs) can easily be identified because they show motifs found in C-type lectins (Chapter 34), as well as a collagen-like domain that promotes their oligomerization and is necessary for host defense through complement activation. A variant allele with changes in both the promoter and structural regions of the human MBL2 gene influences the stability and serum concentration of the protein. Epidemiological studies have suggested that genetically determined variations in MBL serum concentration influence the susceptibility to, and the course of, different types of infections, autoimmune reactions, and metabolic and cardiovascular diseases. The fact that genetic variations in MBL are frequent indicates a dual role for MBL in host defense and highlights the power of genomics to aid our understanding of human disease.
Most of the studies on GBPs have been restricted to mammalian proteins (e.g., C-type lectins, galectins, and Siglecs). Their counterparts in plants and other “lower” organisms are underexplored. An extended classification of GBPs is proposed in Chapter 28, where they are contrasted with binding proteins that recognize sulfated glycosaminoglycans (which seem to have emerged independently of each other, by convergent evolution).
Knowledge of the ligand specificity of a GBP is required to assign in vivo functions. In cases in which there is a dearth of information, rational predictions based on the framework and sequence of existing carbohydrate-recognition domains (CRDs) have been found useful. Legume lectins represent a class of GBPs identified decades ago that continues to provide perhaps the best model for protein–glycan recognition. Moreover, discovery of the legume lectin fold (jelly-roll motif) in mammalian lectins, such as galectins (Chapter 36), calnexin, and calreticulin (Chapter 34), highlights the preeminence of this fold in carbohydrate recognition across phylogeny. Earlier work in the identification of monosaccharide-binding specificities of legume lectins provided the framework for finding their relatives in all forms of life (Chapter 32). This approach led to the assignment of glycan specificities for proteins involved in the sorting of vesicular compartments and in glycoprotein folding in the endoplasmic reticulum (ER) and Golgi compartments of mammalian cells (Chapter 39). Of similar importance is the discovery of new galectins in the galectin-10 family and galectin-like proteins in genome databases using similarity searches (Chapter 36). Likewise, Siglecs, a family of sialic acid–binding I-type lectins involved in regulating multiple biological responses (Chapter 35), show signature sequence motifs, and 17 members have been identified in primates to date. It is important to note that the mere presence of a CRD does not necessarily translate into functional glycan-recognizing activity. This is because sequence motifs used to identify CRDs are often found in a functionally inactive, lectin-like, CRD fold (Chapter 34).
Glycan microarrays provide a high-throughput means of detecting the interactions of GBPs with the diverse oligosaccharide sequences of glycoproteins, glycolipids, and polysaccharides (Chapter 30). The use of glass slides, microarray printing technology, and surface patterning of engineered glycophages displaying unique carbohydrate epitopes allows the production of glycan microarrays with the potential to examine binding of all types of GBPs (lectins, antiglycan monoclonal or serum antibodies, and glycan-binding cytokines or chemokines) to several thousand unique glycans, simultaneously. Binding is assessed by fluorescent or spectrometric techniques. Glycan microarray data are provided by resources such as the Consortium for Functional Glycomics (http://www.functionalglycomics.org/glycomics/publicdata/primaryscreen.jsp) and the Imperial College Microarray Data Online Portal (https://glycosciences.med.ic.ac.uk/data.html), and several analysis software packages, including GLAD (https://glycotoolkit.com/GLAD/), MotifFinder (https://haablab.vai.org/tools/), MCAW (https://mcawdb.glycoinfo.org/) and CCARL (https://github.com/andrewguy/CCARL). Note that arrays that use different linkers and/or different attachment chemistries can give quite variable results, and the results need to be evaluated in the context of natural binding phenomena.
It has long been known that many viruses use host glycans as specific binding receptors for entering the cell (Chapter 37). Similarly, several viruses encode lytic enzymes that break down host cell surface glycans to release viral particles after viral replication. Genome sequencing reveals that many double-stranded DNA viruses also take advantage by adding sugars to host glycoproteins through the use of viral GTs (Chapter 42). Although biological roles of viral GTs are poorly understood, some functions have been identified. For example, the T4 bacteriophage encodes nucleases that degrade host cell DNA. To protect its own genome, the phage modifies its DNA by replacing cytosine with 5-hydroxymethylcytosine and subsequently transferring glucose (Glc) to the 5-hydroxymethylcytosine using a specific UDP-Glc:DNA Glc-transferase. The baculovirus enzyme ecdysteroid glucosyltransferase (EGT) disrupts the hormonal balance of the insect host by catalyzing the conjugation of ecdysteroid hormones with Glc or galactose (Gal). Expression of the EGT gene allows the virus to block molting and pupation of infected insect larvae. Similarly, Chloroviruses have enzymes in CAZy family GT4 for the glycosylation of their structural proteins. Serotype conversion in Shigella flexneri is mediated by temperate bacteriophages, which encode GTs that mediate O-antigen conversion by the addition of Glc to O-antigen units. Finally, giant viruses such as Acanthamoeba polyphaga mimivirus encode 12 putative GTs for the synthesis of complex O-glycans.
Bacterial GTs play a major role in their symbiosis and virulence. Some bacteria such as Campylobacter are able to N-glycosylate their proteins, but the most universal role for bacterial glycosylation is in the synthesis of cell-wall peptidoglycan, simple glycolipids, lipopolysaccharides, and complex exopolysaccharides (Chapter 21). The GTs involved in peptidoglycan biosynthesis are GT28 MurG, which adds N-acetylglucosamine (GlcNAc) to undecaprenyl diphospho-N-MurNAc, and GT51 MtgA, which polymerizes undecaprenyl diphospho-MurNAc-GlcNAc. Mycobacterium tuberculosis produces an extremely complex envelope that includes all of the above. In bacteria, the role of these glycans is to provide a barrier that affords mechanical, chemical, and biological protection to the cell. Some pathogenic or commensal bacteria produce an outer glycan layer that mimics that of their hosts, in order to evade host immune surveillance (Chapters 15 and 42). Pasteurella multocida produces a thick hyaluronan capsule. Oral streptococci produce two GTs for adhesion and virulence, and the EPax GT enables colonization of the gut by Enterococcus faecalis. Other pathogens such as Escherichia coli K1 and Neisseria meningitidis produce a poly α2–8 sialic acid capsule. Mammalian gut bacteria produce capsular polysaccharides thought to help the maturation of the host immune system.
Archaea devote ∼1% of their genes to GTs but, on average, devote only 0.25% to GHs, and there is almost no correlation with the number of GH genes and the overall number of genes. Surprisingly, ∼20% of sequenced archaeal genomes appear to be completely devoid of GHs. The most striking example is Methanosphaera stadtmanae, whose genome encodes at least 43 GTs but apparently no GHs. This is not due to sequence divergence, because GHs are readily detected in some Archaea. These observations suggest that (1) horizontal transfer is likely the determining factor behind archaeal GH repertoires, and (2) the Archaea in question do not recycle glycosidic bonds elaborated by their own GTs. Although they do not make peptidoglycans like bacteria, Archaea use nucleotide-activated oligosaccharides to produce a variety of extracellular polysaccharides such as the heteropolysaccharide “methanochondroitin” made by Methanosarcina barkeri, which resembles eukaryotic chondroitin sulfate (Chapter 17). Archaea also make glycophospholipids and one relevant GT is GDP-Glc:glucosyl-3-phosphoglycerate synthase from family GT81 in Methanococcoides burtonii. In CAZy family GT55, there are several archaeal GDP-Man:mannosyl-3-phosphoglycerate synthases. A number of Archaea have GTs related to bacterial and eukaryotic oligosaccharyltransferases (CAZy family GT66), which is consistent with the fact that Archaea use oligosaccharyldiphospholipids as sugar donors. It has been shown that the archaeon Methanococcus voltae uses this strategy to transfer N-glycans to flagellin and S-layer proteins. Like bacteria and eukaryotes, evolution toward an obligate symbiont lifestyle was also accompanied by gene loss in Archaea. For example, the tiny genome of Nanoarchaeum equitans appears to encode only three GTs and no GH.
With their large genomes and complex body plans that require regulated gene expression in different tissues and/or at different developmental stages, genomes of eukaryotes encode many more GTs and GHs than those of individual bacteria and Archaeal species. But, overall, prokaryotes appear to use a greater diversity of the monosaccharides that exist in nature (Chapters 20–23). Several eukaryotes have also undergone genome reduction and lost most of their GT genes. Thus, Plasmodium falciparum and Encephalitozoon cuniculi have only nine and eight GTs, respectively. Overall, the abundance of GTs in free-living eukaryotes correlates with evolution to multicellularity. Free-living fungi and the unicellular marine green alga Ostreococcus tauri have a number of GTs similar to certain bacteria.
The genomes of “higher” plants encode more GTs than any other organism, with approximately 560 in Arabidopsis, approximately 800 in poplar, and approximately 1200 in Arachis hypogaea! “Higher” plants have huge genomes resulting from several rounds of complete genome duplication. The massive number of GTs in plants is due to the expansion of several extremely populated GT families. For example, Arabidopsis, poplar, and A. hypogaea have about 120, 280, and 400 GT1 genes, respectively. “Higher” plants are characterized by extremely complex cell walls made of various polysaccharides that can be rather simple like cellulose, more complex as in hemicelluloses (e.g., xylans, glucuronoxylans, galactomannans, xyloglucans), or extremely complex like the “hairy” regions of pectin (Chapter 24). Biosynthesis of pectin alone requires the action of dozens of GTs. Differential expression in various tissues is probably one of the driving forces behind the accumulation of hundreds of genes encoding GTs in plants. Likewise, a diverse array of GHs is involved in the remodeling of the plant cell wall during plant growth. Thus, Arabidopsis, poplar, and A. hypogaea genomes encode about 420, 620, and 950 GHs, respectively.
Vertebrates are characterized by a large diversity of GT genes. Human GTs fall into 46 CAZy families, a number similar to that of plants. Families present only in vertebrates are GT6, GT12, and most GT29 family members. Vertebrates usually have many different GT29 sialyltransferases belonging to several distinct subfamilies whereas invertebrates have only one member of a particular sialyltransferase subfamily. However, there are no GT families that are unique to humans or primates. The completion of the first animal genomes also revealed a relative paucity in the number of encoded GHs. Thus, the human genome codes for only 93 GHs, with a dozen devoted to the digestion of only three glycans: sucrose, lactose, and a portion of starch. The digestion of the immense majority of the plant cell-wall polysaccharides in the diet is “outsourced” to the multitude of different microorganisms that colonize the human gut. The genetic material of this flora, the “microbiome,” greatly enlarges our limited genome. For instance, a single species of our gut bacteria such as Bacteroides cellulosilyticus WH2 encodes four times more GHs (408) than our own genome.
One of the initial surprises that came with the completion of the first genomes was that the human genome encodes fewer GTs (242) than that of the nematode Caenorhabditis elegans (273). Interestingly, Drosophila melanogaster has only 155 GT genes. These gross numbers, however, mask important biological differences. The comparative abundance of GTs in C. elegans compared with humans is essentially due to four GT families more highly represented in the nematode: GT1 glucuronyltransferases (79 in C. elegans, 35 in human), GT11 fucosyltransferases (26 in C. elegans, three in human), GT14 β-xylosyltransferases and β1-6 GlcNAc-transferases (20 in C. elegans, 11 in human), and GT92 galactosyltransferases (27 in C. elegans, none in human). For most other GT families, C. elegans appears to have the same number, or fewer, GT genes than humans. With more than 415 GT genes, the bdelloid rotifer Adineta vaga is the animal with the largest known number of GTs. This large number is probably due to the ameiotic reproduction mode of this animal, which is accompanied by a large rate of horizontal gene acquisition from other organisms. C. elegans has 114 GHs, whereas D. melanogaster has 104 and humans have 93. GH18 chitinase is highly represented in C. elegans and D. melanogaster with 43 and 22 members, respectively.
In addition to catalytic specificity, the amino acid sequence of some GTs and GHs can also contain one or more additional domains that modulate the activity of the GT or GH. The most striking example is the two-domain mammalian heparan synthases that have evolved for the addition of alternating sugars to form a polysaccharide (Chapters 16 and 17). The amino-terminal domain, which adds β1-4 glucuronic acid (GlcA) residues, belongs to GT47, whereas the carboxy-terminal domain, which adds α1-4 GlcNAc residues, belongs to GT64. Some strains of bacteria have a heparan synthase, which also consists of two catalytic modules from families GT2 and GT45 (Figure 8.1), thereby providing a beautiful example of convergent evolution. A similar example of convergent evolution is found among chondroitin synthases, in which human enzymes are made of GT31 and GT7 catalytic domains, whereas the bacterial equivalent is made of tandem GT2 catalytic domains. Human LARGE is another bifunctional glycosyltransferase made of two domains. The amino-terminal domain, which adds α1-3 xylose (Xyl) residues, belongs to GT8, whereas the carboxy-terminal domain, which adds β1-3 GlcA residues, belongs to GT49.
Schematic examples of modular glycosyltransferases. GT family modules are shown in red and blue. Other modules in various colors are CBM13, ricin-like carbohydrate-binding module; SH3, src homology domain 3; X84, putative glycan-binding module; PBP, penicillin-binding (more...)
Other modular GTs can feature an appended GBP domain. The best-known examples are polypeptide N-acetylgalactosaminyltransferases (ppGalNAcTs; GALNT) that transfer N-acetylgalactosamine (GalNAc) to Ser or Thr residues (Chapters 6 and 10). In these enzymes, a GT27 catalytic domain is linked to a GBP domain related to ricin and classified as CBM13 in the CAZy database. The GBP domain binds to the GalNAc residue transferred to protein by the GT27 catalytic domain and tethers the enzyme to the substrate. Another example is mouse polypeptide β-xylosyltransferase 2, in which a GT14 catalytic domain is linked to a carboxy-terminal domain that is thought to act as a GBP.
The GHs can also be modular, with the catalytic domain appended to one or more other modules whose role is to bind polysaccharides. Although human GHs are infrequently modular, those of microbes involved in plant cell-wall degradation can have more than five different modules assembled in a single polypeptide. Human acidic chitinase is an example of a mammalian modular GH having a CBM14 domain appended at the carboxyl terminus of the GH18 catalytic domain. The most intricate architecture of GHs is found in certain bacteria, such as Clostridium thermocellum, which elaborate a macromolecular complex called a “cellulosome” in which a large variety of modular plant cell-wall hydrolases are assembled together on a scaffolding protein. This strategy enables the assembly of dozens of catalytic modules simultaneously targeting the various polysaccharides that make up the plant cell wall.
In summary, the genome, comprising the DNA of an organism, includes all the genes that produce the glycome, which comprises all of the glycans made by an organism. Although, within an organism, almost every cell that contains a nucleus and mitochondria has an identical genome; cells typically differ in the portion of the genome and, therefore the glycome, they express. Thus, the glycan complement of a cell depends on which genes are actively transcribed and which transcripts are translated and stably expressed. Transcription, splicing, translation, and posttranslational processing may vary depending on the state of differentiation and the physiological environment of a cell. Therefore, during development and differentiation, and under different environmental conditions, the glycan repertoire of a cell represents a subset of all the glycans that an organism is capable of making. To describe this variation, it is common to qualify the term glycome when referring to the glycans made by a particular tissue or cell type (e.g., T-cell glycome, hepatocyte glycome, or serum glycome), and to note the particular stage of development (e.g., fetal liver glycome, breast cancer serum glycome).
The glycome of a given cell in an organism can undergo substantial changes in response to environmental stimuli ranging from pH and ionic strength to hormonal stimulation or inflammation. Combined with the “assembly-line” nature of the Golgi apparatus (Chapter 4) and potential remodeling by GHs, full knowledge of the GT and GH transcriptome is only a rough predictor of the actual glycome of a given cell type, albeit a very helpful one. Apart from transcriptional control, the glycome is regulated by posttranscriptional control by microRNAs (miRs). For example, GALNT7 is a target of miR-30d, a microRNA that is known to promote melanoma metastasis in patients and mouse models. The down-regulation of GALNT7 phenocopied the expression of miR-30d. Subsequently, miRNAs have emerged as key regulators of the glycome owing to their ability to regulate multiple GT and GH mRNA targets. Nearly 80 GT and GH genes have been identified as targets of miRNAs to date.
The authors appreciate helpful comments and suggestions from Kelley Moremen and Tadashi Suzuki.
The content of this book is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 Unported license. To view the terms and conditions of this license, visit https://creativecommons.org/licenses/by-nc-nd/4.0/
Your browsing activity is empty.
Activity recording is turned off.
See more...
