mechanisms in protein o-glycan biosynthesis and clinical and

Mechanisms in Protein O-Glycan Biosynthesis andClinical and Molecular Aspects of ProteinO-Glycan Biosynthesis Defects: A ReviewSuzan Wopereis,1 Dirk J. Lefeber,1 Eva Morava,2 and Ron A. Wevers1*

Background: Genetic diseases that affect the biosynthe-sis of protein O-glycans are a rapidly growing group ofdisorders. Because this group of disorders does not havea collective name, it is difficult to get an overview ofO-glycosylation in relation to human health and dis-ease. Many patients with an unsolved defect in N-glycosylation are found to have an abnormal O-glyco-sylation as well. It is becoming increasingly evident thatthe primary defect of these disorders is not necessarilylocalized in one of the glycan-specific transferases, butcan likewise be found in the biosynthesis of nucleotidesugars, their transport to the endoplasmic reticulum(ER)/Golgi, and in Golgi trafficking. Already, disordersin O-glycan biosynthesis form a substantial group ofgenetic diseases. In view of the number of genes in-volved in O-glycosylation processes and the increasingscientific interest in congenital disorders of glycosyla-tion, it is expected that the number of identified dis-eases in this group will grow rapidly over the comingyears.Content: We first discuss the biosynthesis of proteinO-glycans from their building blocks to their secretionfrom the Golgi. Subsequently, we review 24 differentgenetic disorders in O-glycosylation and 10 different ge-netic disorders that affect both N- and O-glycosylation.The key clinical, metabolic, chemical, diagnostic, and ge-netic features are described. Additionally, we describemethods that can be used in clinical laboratory screeningfor protein O-glycosylation biosynthesis defects and their

pitfalls. Finally, we introduce existing methods that might beuseful for unraveling O-glycosylation defects in the future.© 2006 American Association for Clinical Chemistry

The human proteome, originating from expression of theprotein-coding genes of the genome, comprises �30 000proteins (1 ), a surprisingly low number considering thatthe genome of the nematode Caenorhabditis elegans com-prises 20 000 genes (2 ). However, a higher order ofcomplexity of protein products in humans arises frompretranslational events, such as alternative splicing, andposttranslational modifications, such as phosphorylationand glycosylation. Glycosylation, the enzymatic additionof carbohydrates to proteins or lipids, is the most commonand most complex form of posttranslational modification.This is illustrated by the estimation that 1% of human

1 Laboratory of Pediatrics and Neurology and 2 Department of Pediatrics,Radboud University Nijmegen Medical Center, Nijmegen, The Netherlands.

* Address correspondence to this author at: Laboratory of Pediatrics andNeurology (830), Institute of Neurology, Radboud University Nijmegen Med-ical Center, Geert Grooteplein 10, 6525 GA Nijmegen, The Netherlands. Fax31-24-3540297; e-mail [email protected].

Received November 2, 2005; accepted January 24, 2006.Previously published online at DOI: 10.1373/clinchem.2005.063040

3 Nonstandard abbreviations: hLys, hydroxylysine; CDG, congenital dis-orders of glycosylation; GalNAc, N-acetylgalactosamine; NeuAc, N-acetyl-neuraminic acid (sialic acid); GlcNAc, N-acetylglucosamine; sLex, sialyl Lewisx

antigen; GAG, glycosaminoglycan; GlcA, glucuronic acid (or glucuronate);EGF, epidermal growth factor; TSR, thrombospondin type-1 repeat; ER,endoplasmic reticulum; GNE/MNK, UDP-GlcNAc 2 epimerase/N-acetylman-nosamine kinase; Dol-P, dolichol phosphate; NST, nucleotide sugar trans-porter; CHO, Chinese hamster ovary; FUCT, GDP-Fuc transporter; �3-Gal-T,�3-galactosyltransferase; Cosmc, core 1 �3-Gal-T-specific molecular chaper-one; pp-GalNAc-T, polypeptide N-acetylgalactosaminyltransferase; EXTL ex-ostoses-like; COP, coatomer protein; ERGIC, endoplasmic reticulum-Golgiintermediate compartment; SNARE, soluble N-ethylmaleimide-sensitive fu-sion attachment protein receptor; COG, conserved oligomeric Golgi complex;GalNT, N-acetylgalactosyltransferase; FTC, familial tumoral calcinosis;B4GalT, �-1,4-galactosyltransferase; HME, hereditary multiple exostoses;MCD, macular corneal dystrophy; SED, spondyloepiphyseal dysplasia;DTDST, diastrophic dysplasia sulfate transporter; DTD, diastrophic dysplasia;ACGB1, achondrogenesis type 1B; AO-II, atelosteogenesis type II; EDM4,multiple epiphyseal dysplasia 4; PAPSS2, 3�-phosphoadenosine 5�-phospho-sulfate synthase 2; APS, adenosine 5�-phosphosulfate; PAPS, 3�-phosphoad-enosine 5�-phosphosulfate; WWS, Walker–Warburg syndrome; LGMD2, limb-girdle muscular dystrophy type 2; MEB, muscle–eye–brain disease; FCMD,Fukuyama-type congenital muscular dystrophy; FKRP, fukutin-related pro-tein; MDC, congenital muscular dystrophy; LARGE, N-acetylglucosaminyl-like protein; hIBM, hereditary inclusion body myopathy; DMRV, distal myop-athy with rimmed vacuoles; FUT, fucosyltransferase; IEF, isoelectric focusing;apoC-III, apolipoprotein C-III; and CMRD, chylomicron retention disease.

Clinical Chemistry 52:4574–600 (2006) Reviews

574

genes are required for this specific process (3 ). Further-more, more than one half of all proteins are glycosylated,according to estimates based on the SwissProt database(4 ). In humans, protein-linked glycans can be divided into3 categories: N-linked (linkage to the amide group ofAsn), O-linked [linkage to the hydroxyl group of Ser, Thr,or hydroxylysine (hLys)3], and C-linked (linkage to acarboxyl group of Trp) (5 ).

Initially, the study of glycoproteins and their role inhuman congenital diseases focused on N-linked glycans.The diseases in this pathway have collectively been re-ferred to as congenital disorders of glycosylation (CDG).N-Glycans share a common protein–glycan linkage andhave a common biosynthetic pathway that diverges onlyin the late Golgi stage. Endoglycosidases are available thatcan cleave intact N-glycans from the protein backbone,making it relatively easy to study alterations of N-glyco-sylation in health and disease. In contrast, O-glycans arebuilt on different protein glycan linkages and have ex-tremely diverse structures; in addition, there is no en-doglycosidase available for the release of intact O-glycans.However, methods for the chemical release of O-glycanshave been developed and have enabled the generation ofstructural information for O-glycans, making it morefeasible to study alterations in O-glycosylation in relationto health and disease. This review focuses on the biosyn-thesis of O-glycans and the human congenital disorders ofO-glycosylation and their screening.

Structures of O-Linked GlycansThe O-glycosylation process produces an immense mul-tiplicity of chemical structures. Each monosaccharide has3 or 4 attachment sites for linkage of other sugar residuesand can form a glycosidic linkage in an � or � configura-tion, allowing glycan structures to form branches. Glycanstherefore have a larger structural diversity in contrast toother cellular macromolecules such as proteins, DNA, andRNA, which form only linear chains. Theoretically, the 9common monosaccharides found in humans could beassembled into more than 15 million possible tetrasaccha-rides, all of which would be considered relatively simpleglycans (6 ).

The 7 different types of O-linked glycans found inhumans are summarized in Table 1. O-Linked glycans areclassified on the basis of the first sugar attached to a Ser,

Thr, or hLys residue of a protein. The mucin-type O-glycan, with N-acetylgalactosamine (GalNAc) at the re-ducing end, is the most common form in humans. In total,8 mucin-type core structures can be distinguished, depend-ing on the second sugar and its sugar linkage, of which cores1–6 and core 8 have been described in humans (summarizedin Table 2) (7). In addition to the 7 core structures, the Tn(GalNAc�1-Ser/Thr) and sialyl Tn [NeuAc�2–6GalNAc�1-Ser/Thr; where NeuAc is N-acetylneuraminic acid (sialicacid)] epitopes can be distinguished. The core structurescan be further modified; for example, by the addition ofan N-acetyllactosamine unit (Gal�1–4GlcNAc; whereGlcNAc is N-acetylglucosamine), also seen on N-glycans.The N-acetyllactosamine unit may be branched by aGlcNAc�1–6 residue or form repeating N-acetyllactosamineunits, called poly N-acetyllactosamine extensions. It canalso attach to the blood group determinants (A, B, and H)and the type 2 Lewis determinants [Lex, sialyl Lewisx

(sLex), and Ley]. N-Acetyllactosamine elongations areseen mainly on core 2 O-glycans. Sugars occurring at thenonreducing termini include NeuAc, Fuc, GlcNAc, andGalNAc. GlcNAc and Gal residues can be modified atposition 6 or at positions 3 and/or 6, respectively, bysulfation (8 ), and NeuAc residues can be further modifiedat positions 4, 7, 8, and 9 with O-acetyl ester groups (9 ).This gives rise to several hundreds of different mucin-type O-glycan structures, of which core 1 and 2 are mostabundant (7 ).

Another common type of O-glycosylation with largestructural diversity involves the glycosaminoglycans(GAGs). Proteoglycans are proteins containing GAGchains. GAGs are attached to a Ser residue of a protein viathe linker tetrasaccharide GlcA�1–3Gal�1–3Gal�1–4Xyl,except for keratan sulfate, which is linked to proteinseither through N- or core 1 O-glycans. GAGs are long,unbranched polysaccharides containing a disacchariderepeat that consists of either a GalNAc or GlcNAc residuecombined with a glucuronic acid (GlcA) or a Gal residue.Three different types of GAGs can be distinguished on thebasis of the composition of the disaccharide repeat: (a)dermatan sulfate and chondroitin sulfate (GlcA � GalNAc);(b) heparin/heparan sulfate (GlcA � GlcNAc); and (c) kera-tan sulfate (Gal � GlcNAc). GlcA in dermatan sulfate andheparin/heparan sulfate can be epimerized to iduronate.The heterogeneity of GAGs results from variable O-

Table 1. Different types of O-linked glycans in humans.Type of O-linked glycan Structure and peptide linkage Glycoprotein Reference(s)

Mucin-type (R)-GalNAc�1-Ser/Thr Secreted � plasma membrane (8 )GAG (R)-GlcA�1–3Gal�1–3Gal�1–4Xyl�1-Ser Proteoglycans (216, 217)O-linked GlcNAc GlcNAc�1-Ser/Thr Nuclear and cytoplasmic (218)O-linked Gal Glc�1–2�Gal�1-O-Lys Collagens (219)O-linked Man NeuAc�2–3Gal�1–4GlcNAc�1–2Man�1-Ser/Thr �-Dystroglycan (16)O-linked Glc Xyl�1–3Xyl�1–3�Glc�1-Ser EGF protein domains (220)O-linked Fuc NeuAc�2–6Gal�1–4GlcNAc�1–3�Fuc�1-Ser/Thr EGF protein domains (220)

Glc�1–3Fuc�1-Ser/Thr TSR repeats (24)

Clinical Chemistry 52, No. 4, 2006 575

sulfation at defined locations (10 ). An extra modificationstep occurs in heparin and heparan sulfate by the deacety-lation and N-sulfation of GlcNAc residues. Regions inwhich the hexosamine units are acetylated remain (al-most) unmodified and consist of disaccharide repeatswith GlcA, whereas regions with deacetylated hex-osamine units become highly sulfated and exist as disac-charide repeats with iduronate. Heparin is a highly anduniformly sulfated GAG, whereas heparan sulfate ishighly sulfated only in defined blocks (11 ).

The structures of the other 5 O-glycan types seem toshow less variability, and they occur mostly in oneconformation. A frequently occurring O-linked glycan isthe single GlcNAc linked to nuclear and cytosolic pro-teins. This posttranslational modification is more analo-gous to phosphorylation than to classical complex O-glycosylation because it is a reversible process catalyzedby the enzymes O-GlcNAc transferase and O-GlcNAcase,respectively (12 ), and the “normal glycosylation machin-ery” is not implicated (12, 13).

O-Galactosyl glycans have been found only on colla-gen domains. Gal or Glc�1–2Gal residues are covalentlylinked to hLys residues found in collagens, but not allhLys residues become glycosylated. The collagen 3-di-mensional structure depends on the extent of this post-translational modification. The quantities and types ofO-galactosyl glycans vary considerably not only amongthe different types of collagen, but also among the samecollagen type from different tissues and even the samecollagen type from different areas of the same type oftissue (14, 15).

O-Mannosyl glycans are a less common type of proteinmodification, present on a limited number of glycopro-teins in the brain, nerves, and skeletal muscle. The bestknown O-mannosylglycosylated protein is �-dystrogly-can, which is a skeletal muscle extracellular matrix protein(16). To date, only the NeuAc�2–3Gal�1–4GlcNAc�1–2Manstructure has been found in humans. �-Dystroglycancontaining Gal�1–4(Fuc�1–3)GlcNAc�1–2Man has beenfound in sheep brain (17, 18), and the O-mannosyl glycanHSO3-3GlcA�1–3Gal�1–4GlcNAc�1–2Man has been de-tected in rat brain (18, 19). Studies have also shown thatmammalian N-acetylglucosaminyltransferase IX acts onthe GlcNAc�1,2-Man�1-Ser/Thr moiety, suggesting that2,6-branched O-mannosyl glycan structures are formed in

the brain (20 ). It is therefore likely that structural diversityof O-mannosyl glycans will also be present in humans.

O-Glucosyl and O-fucosyl glycans are also rare typesof protein glycosylations that have been found in theepidermal growth factor homology regions (EGF mod-ules) of some human proteins. An EGF module is acommon structural motif found in several secreted andcell-surface proteins that is often involved in mediatingprotein–protein interactions. The EGF repeat is typically30–40 amino acids long and is characterized by 6 con-served Cys residues participating in 3 disulfide bridges.Glc is linked to the Ser residue in proteins in the putativeconsensus sequence C1XSXPC2 (where C1 and C2 are thefirst and second conserved cysteines of the EGF module,S is the modified Ser residue, and X can be any aminoacid) (21 ). O-Linked Glc can be further elongated with 1or 2 �1–3 linked xyloses and is found on proteins such ashuman factor VII, factor IX, and protein Z (22, 23). AllO-fucosylated glycoproteins are modified with a singleO-linked Fuc residue (e.g., urinary-type plasminogen ac-tivator, tissue-type plasminogen activator, and coagula-tion factors VII and XII) except for coagulation factor IX,which contains O-linked Fuc that is elongated to thetetrasaccharide NeuAc�2–6Gal�1–4GlcNAc�1–3Fuc�1-Ser/Thr. Most O-Fuc modifications on EGF repeats arefound on the consensus site C2X3–5S/TC3 (where C2 andC3 are the second and third conserved cysteines of theEGF repeat, S/T is the modified Ser/Thr residue, and Xcan be any residue) (22 ). A second type of O-fucosylationhas been identified. On thrombospondin type 1 repeats(TSRs), a disaccharide form of O-fucosyl glycans (Glc�1–3Fuc�1-Ser/Thr) is found on the human extracellularmatrix protein “thrombospondin-1” (24 ). TSRs are foundin many extracellular proteins. A single TSR is �60 aminoacids long and is characterized by conserved Cys, Trp,Ser, and Arg residues. The putative consensus sequencesite for this modification is WX5CX2/3S/TCX2G (22 ).

o-glycan consensus sitesFor most O-glycosylation types, a recognition consensussequence for the attachment of the first sugar residueremains unknown. The exceptions are the O-Glc andO-Fuc modifications, for which putative consensus siteshave been described [see above and Refs. (21, 22)]. Thelack of a consensus sequence can arise from the coexist-

Table 2. Diversity of mucin-type O-linked glycans.Core Structure Human tissue Reference(s)

1 Gal�1–3GalNAc Most cells and secreted proteins (7 )2 Gal�1–3 (GlcNAc�1–6)GalNAc All blood cells (221)3 GlcNAc�1–3GalNAc Colon and saliva (222, 223)4 GlcNAc�1–3 (GlcNAc�1–6)GalNAc Mucin-secreting cell types (221)5 GalNAc�1–3GalNAc Meconium (224)6 GlcNAc�1–6GalNAc Ovarian tissue (225)7 GlcNAc�1–6GalNAc8 Gal�1–3GalNAc Bronchia (226)

576 Wopereis et al.: Defects in Protein O-Glycan Biosynthesis

ence of multiple transferases with overlapping but differ-ent substrate specificities, as seen, e.g., in mucin-typeO-glycosylation, or is the result of a nonlimited consensussequence, as seen, e.g., in O-GlcNAc modifications. Sta-tistical studies yielded some general rules for mucin-typeO-glycans and O-GlcNAc modifications, leading to thedevelopment of algorithms for the prediction of these 2O-glycan types. These O-glycosylation prediction sites areavailable on the Internet. The NetOglyc 3.1 predictionserver correctly predicts 76% of the glycosylated residuesand 93% of the nonglycosylated residues in any protein(25 ).

Biosynthesis of O-GlycansThe main pathway for the biosynthesis of complex N- andO-linked glycans is located in the endoplasmic reticulum(ER) and Golgi compartments, the so-called secretorypathway. Glycosylation is restricted mainly to proteinsthat are synthesized and sorted in this secretory pathway,which includes ER, Golgi, lysosomal, plasma membrane,and secretory proteins. There is one exception; nuclearand cytosolic proteins can be modified with a singleO-linked GlcNAc (12 ). Proteins synthesized by ribosomesand sorted in the secretory pathway are directed to therough ER by an ER signal sequence in the NH2 terminus(26, 27). After protein folding is completed in the ER,these proteins move via transport vesicles to the Golgicomplex. The biosynthesis of O-glycans is initiated afterthe folding and oligomerization of proteins either in thelate ER or in one of the Golgi compartments (28–31).Intriguingly, for the biosynthesis of glycans, no templateis involved; whereas DNA forms the template for thesequence of amino acids in a protein, there is no suchequivalent for the design of glycans. The biosynthesis ofglycans can be divided into 3 stages. In the first stage,nucleotide sugars are synthesized in the cytoplasm. In thesecond stage, these nucleotide sugars are transported intothe ER or the Golgi. In the third stage, specific glycosyl-transferases attach the sugars to a protein or to a glycan inthe ER or Golgi. An additional prerequisite for properglycosylation is Golgi trafficking. Recently, it was discov-ered that a defect in a protein involved in Golgi trafficsecondarily caused abnormal N- and O-glycans in 2patients with CDG-IIe. For this reason, Golgi traffic willbe discussed briefly in this section.

biosynthesis of nucleotide sugarsMonosaccharides used for the biosynthesis of nucleotidesugars derive from dietary sources and salvage pathways.Glucose (Glc) and fructose (Fru) are the major carbonsources in humans from which all other monosaccharidescan be synthesized (Fig. 1). Series of phosphorylation,epimerization, and acetylation reactions convert theminto various high-energy nucleotide sugar donors (seeFig. 1). Nucleotide sugar biosynthesis takes place in thecytosol, except for CMP-NeuAc, which is synthesized inthe nucleus (32 ).

As observed in patients with CDG-Ia and CDG-Ib,aberrant glycosylation results from insufficient availabil-ity of GDP-Man. The availability of nucleotide sugars istightly regulated. UDP-GlcNAc, for example, inhibitsglutamine-fructose-6P-transaminase, which catalyzes thefirst step in the biosynthetic pathway of UDP-GlcNAc(33 ), and CMP-NeuAc inhibits UDP-GlcNAc-2-epimer-ase/N-acetylmannosamine kinase (GNE/MNK), whichcatalyzes the first 2 biosynthetic steps of CMP-NeuAc(34 ). Although much is known about the nucleotide sugarbiosynthesis pathways and their feedback regulators, theactual cytosol and Golgi steady-state concentrations ofmost nucleotide sugars are unknown at present. Further-more, because of the interconnected pathways of nucleo-tide sugar metabolism, the results of an individual en-zyme deficiency are difficult to predict.

Several steps in the biosynthesis of nucleotide sugarsrequire ATP; therefore, the metabolic state of the cellinfluences the availability of the nucleotide sugars. Thetight regulation of the biosynthesis of nucleotide sugarsmeans that alterations in a single nucleotide sugar cansignificantly impair glycosylation.

transport processes to generatemonosaccharide donors in the er/golgiThe nucleotide sugars are biosynthesized in the cytosol,and their monosaccharides must be translocated into thelumen of the ER and/or Golgi before they can be used forthe glycosylation process. Because nucleotide sugars can-not cross the membrane lipid bilayer, specific transportmechanisms are responsible for their translocation. Two

Fig. 1. Schematic overview of the biosynthesis of nucleotide sugars.Not all intermediate steps are shown. Circled monosaccharides are obtainedfrom dietary sources and/or salvage pathways. Gray stars indicate reactions thatrequire ATP. NeuAc-9P, N-acetylneuraminic acid 9-phosphate; ManNAc-6P, N-acetylmannosamine 6-phosphate; ManNAc, N-acetylmannosamine; GalNAc-1P,N-acetylgalactosamine 1-phosphate; GlcNAc-1P and GlcNAc-6P, N-acetylglu-cosamine 1-phosphate and 6-phosphate, respectively; GlcN-6P, glucosamine6-phosphate; Fru-6P, fructose 6-phosphate; Man-6P and Man-1P, mannose6-phosphate and 1-phosphate, respectively; Fuc-1P, fucose 1-phosphate; Glc-6Pand Glc-1P, glucose 6-phosphate and 1-phosphate, respectively; Gal-1P, galac-tose 1-phosphate.


transport mechanisms for the generation of monosaccha-ride donors in the ER/Golgi can be distinguished (Fig. 2).The first mechanism is the entrance of Man and Glcthrough binding to the lipid carrier dolichol phosphate(Dol-P). To date, this transport system has been describedonly in the ER. Cytosolic Dol-P-Man and Dol-P-Glc syn-thases link GDP-Man and UDP-Glc to the cytosolic site ofDol-P by cleaving off the nucleotide moiety. A hypothet-ical “flippase” then mediates the turnover of the Dol-Pmonosaccharide from the cytoplasmic leaflet to the lume-nal leaflet of the ER. Subsequently, the monosaccharidescan be used by ER-located glycosyltransferases (Fig. 2A)(35 ). As observed in patients with CDG-Ie who aredeficient for Dol-P-Man synthase and in patients withCDG-If who have mutations in the MPDU1 gene, knownto be required for efficient use of Dol-P-Man and Dol-P-Glc as donor substrates, abnormal glycosylation resultsfrom diminished Dol-P-monosaccharide transport (36, 37).In CDG-If patients, it was observed that the mannosyla-tion of N-glycans, glycosylphosphatidylinotisol anchors,and C-mannosyl glycans was defective. Although O-mannosylation was not studied, it is likely that this is alsoaberrant in these patients.

The second mechanism is the transport of nucleotidesugars through specific nucleotide sugar transporters(NSTs). NSTs belong to solute carrier family 35 and residein the Golgi and/or ER membranes with their C- andN-terminal regions exposed to the cytosol. These NSTs areantiporters in which nucleotide sugar entry into theER/Golgi is coupled to equimolar exit of the correspond-ing nucleoside monophosphate from the ER/Golgi lumen(38 ). The nucleotide moiety of the nucleotide sugar is the

recognition feature required for initial binding to the NST,whereas the attached monosaccharide finally determineswhether the entire nucleotide sugar is translocated. Afterentrance of the nucleotide sugar into the ER/Golgi lumen,a glycosyltransferase will transfer the monosaccharide toa glycan by cleaving off the nucleotide part. The nucleo-side diphosphates are converted to dianionic nucleosidemonophosphates (used for the antiporter) and inorganicphosphate by a nucleoside diphosphatase. It is postulatedthat inorganic phosphate exits the ER/Golgi lumen via aspecific transporter (Fig. 2B) (38 ). Nucleoside di- andmonophosphates can inhibit the nucleotide sugar trans-port process and the activity of glycosyltransferases.

Some NSTs transport more than one substrate: forexample, the UDP-Gal/UDP-GalNAc transporter (hereaf-ter referred to as UDP-Gal transporter) (39 ), the UDP-GlcA/UDP-GalNAc/UDP-GlcNAc transporter (hereafterreferred to as UDP-GlcA transporter) (40, 41), and therecently described UDP-Xyl/UDP-GlcNAc transporter(hereafter referred to as UDP-Xyl transporter) (42 ). Incontrast, the CMP-NeuAc (43 ), GDP-Fuc (44 ), and UDP-GlcNAc transporters (45 ) are monospecific.

In general, the transport of a nucleotide sugar occurs inthe organelle in which the corresponding glycosyltrans-ferase is localized. Some nucleotide sugars enter only thelumen of Golgi vesicles, others enter the lumen of ER-derived vesicles, and a few enter both. It has been shownthat the CMP-NeuAc, GDP-Fuc, UDP-GlcNAc, and UDP-Xyl transporters have a strict Golgi membrane localization(38, 42, 45), whereas the UDP-GlcA transporter is local-ized in the ER membrane (40 ). Experiments investigatingthe intraorganelle availability of nucleotide sugars haveshown that UDP-Xyl and UDP-Glc can also be found inthe ER, whereas UDP-GlcA and UDP-Glc can be found inthe Golgi (38 ), suggesting that the corresponding NSTsare yet to be identified.

Galactosylceramide is galactosylated by a galactosyl-transferase (UDP-galactose:ceramide galactosyltransferase)found exclusively in the ER, whereas the UDP-Gal trans-porter has mainly a Golgi localization. This galactosyl-transferase is produced only in specialized cells, suchas myelinating cells, spermatogonia, and in various epi-thelial cell types. The question of how an ER-residentglycosyltransferase could function without a source ofsubstrate was answered by showing that the galactosyl-transferase forms a complex with the UDP-Gal trans-porter (46 ). This led to the presence of a fraction of theUDP-Gal transporter in the ER. It is not clear whether thisis attributable to the retention of the UDP-Gal transporterby the galactosyltransferase or to recycling of the UDP-Gal transporter through the cis-Golgi. In this way, abiosynthetic pathway can be established only when re-quired (46 ). Recently, a second active mechanism hasbeen found for the ER localization of the UDP-Gal trans-porter. The UDP-Gal transporter is produced in 2 spliceforms, UGT1 and UGT2. UGT1 has a strict Golgi localiza-tion, whereas UGT2 shows dual localization in both the

Fig. 2. Schematic overview of the NST mechanisms.(A), Man (M) is transferred from GDP-Man to Dol-P. The Dol-P-Man “flips” over theER membrane, where Man is attached to the glycan by a specific mannosyltrans-ferase (ManT). (B), UDP-GalNAc is transported into the lumen of the ER/Golgiwith equimolar exit of a dianionic nucleoside monophosphate (UMP2�). GalNAcis attached to the protein by a specific pp-GalNAc-T, thereby simultaneouslycleaving off the nucleoside diphosphate. This is converted to UMP2� andinorganic phosphate (Pi) by a lumenal nucleoside diphosphatase. Inorganicphosphate is postulated to exit the ER lumen via a specific transporter.


ER and Golgi caused by a dilysine motif (KVKGS) in itsCOOH terminus (47 ).

As observed in the case of patients with CDG-IIc whohave a deficient GDP-Fuc transporter (FUCT) (48 ) and ina patient with CDG-IIf who has a deficient CMP-NeuActransporter (49 ), abnormal glycosylation results from di-minished NST function. In addition, in Chinese hamsterovary (CHO) lec8 and lec2 cells defective in UDP-Gal andCMP-NeuAc transport, respectively, 70%–90% of the gly-cans lacked that particular monosaccharide (38 ). It wasalso shown that the nucleotide sugar transport processdepends on the continuous production of nucleosidemonophosphates. Abeijon et al. (50 ) showed that in vitrotransport of GDP-Man into the Golgi is severely de-creased in a Saccharomyces cerevisiae guanosine diphos-phatase-null mutant. All glycoproteins and glycolipidsshowed impaired mannosylation (50 ). This results indi-cates that NSTs are critical components of glycosylationpathways.

transfer of nucleotide sugars to the glycanO-Glycans are assembled by the sequential action ofseveral specific, membrane-bound glycosyl-, O-acetyl-,and sulfotransferases in a highly controlled fashion (8 ).The pathways of O-glycosylation are determined by thedistinct substrate specificities of glycosyltransferases, sul-fotransferases, and O-acetyltransferases. Transferases in-volved in O-glycan biosynthesis are localized mainly inthe Golgi. Although many of these enzymes catalyzesimilar reactions, there is a surprisingly limited sequencehomology among different classes. The Golgi glycosyl-transferases described to date are all type II transmem-brane proteins, with a short N-terminal cytoplasmic do-main, a single hydrophobic membrane-spanning domain,and a large C-terminal catalytic domain localized in thelumen of the Golgi.

The activity of glycosyltransferases can be influencedby different factors. It is known, for example, that some ofthe glycosyltransferases require divalent cations, such asMn2� and/or Mg2�, for optimal action. In contrast to thereactions involving UDP- and GDP-nucleotide sugars, thebiosynthetic steps involving CMP-NeuAc do not requirethese cations (51 ). Petrova et al. (52 ) showed that divalentcations react strongly with nucleotide sugars in solution,thus altering their conformation.

Furthermore, it was recently discovered that humancore 1 �3-galactosyltransferase (core 1 �3-Gal-T), which isinvolved in the formation of core 1 (and core 2) mucin-type O-glycans, requires a molecular chaperone for itsfunctioning. This molecular chaperone is called core 1�3-Gal-T-specific molecular chaperone (Cosmc) and is anER-localized type II transmembrane protein that appearsto be required for the proper folding of the core 1�3-Gal-T enzyme. In the absence of functional Cosmc,core 1 �3-Gal-T is degraded in the proteosome (53 ). Thisraises the question of whether additional chaperonesspecific for other glycosyltransferases exist.

A third factor that might influence glycosyltransferaseactivity is the structure of the protein substrate. It isthought that the protein structure contains informationfor the action of specific transferases. This is seen, forexample, in proteoglycans, in which the core proteindictates whether it will receive a heparan sulfate or achondroitin sulfate chain (54 ), or in lysosomal enzymes,in which GlcNAc-phosphotransferase recognizes subtlemotifs in the secondary structure and selectively phos-phorylates the N-glycans on proteins that should reachthe lysosome (55, 56). However, how proteins are recog-nized by glycosyltransferases remains largely unknown.Finally, glycosyltransferase activity can be dependent onheterocomplex formation. O-Mannosyltransferase activ-ity, for example, is generated only when the genesPOMT1 and POMT2 (both encoding mannosyltrans-ferases) are coexpressed (57 ).

Golgi transferases can recognize a single sugar residue,a sugar sequence, or a peptide moiety, leading to variablespecificity. With very few exceptions, each type of trans-ferase is regio- and stereospecific. Glycosyltransferasesinvolved in the linkage of monosaccharides to the proteinbackbone and those involved in the core processing ofmucin-type O-glycans are specific and not involved inother classes of glycoconjugates, whereas most glycosyl-transferases involved in the elongation, branching, andtermination of glycans are not specific for one glycocon-jugate class. For example, the ubiquitous �2,6-sialyltrans-ferase ST6Gal I recognizes the N-acetyllactosamine unitand catalyzes the formation of an �2,6 linkage to terminalN-acetyllactosamine structures found on N-glycans, O-glycans, and glycosphingolipids, whereas the �1,4-galac-tosyltransferase (Gal-T1) galactosylates any terminal Glc-NAc residue.

The attachment of UDP-GalNAc in an � linkage to thehydroxyl residue of Ser or Thr in mucin-type O-glycans isa complex and as yet not fully understood process. Thistransfer is catalyzed by specific UDP-GalNAc: polypep-tide N-acetylgalactosaminyltransferases (EC 2.4.1.41).The mammalian family of UDP-GalNAc:polypeptideN-acetylgalactosaminyltransferases (pp-GalNAc-Ts) com-prises 15 members, the 15th being discovered only re-cently (58, 59). It is estimated that at least 24 uniquehuman pp-GalNAc-Ts exist on the basis of sequencehomology (59 ). The different pp-GalNAc-Ts have over-lapping but different specificities and are tissue specific(8, 59). It seems that mucin-type O-glycosylation proceedsin a hierarchical manner, because some of the character-ized pp-GalNAc-Ts glycosylate only peptides that arealready partly glycosylated (59 ). Currently, no consensussequence has been formulated because every pp-GalNAc-Thas its own specific attachment site. Only Ser and Thrresidues that are exposed on the protein surface will beglycosylated, as O-glycosylation is a postfolding event.Therefore, O-glycosylation takes place mainly in coil,turn, and linker regions. Furthermore, all attachment siteshave high Ser, Thr, and Pro content (25 ).


The biosynthesis of GAG structures differs from the“small O-linked glycans” in 2 significant ways: (a) thetransferases required are all specific and not involved inother glycoconjugate classes, with the exception of chon-droitin 6-sulfotransferase, keratan sulfate Gal-6-sulfo-transferase, and the GlcNAc 6-O-sulfotransferase that alsosulfates N-acetyllactosamine extensions (60 ); and (b) themechanism of GAG chain elongation is different. Chon-droitin/dermatan sulfate and heparin/heparan sulfateare synthesized on the common tetrasaccharide linker(GlcA�1–3Gal�1–3Gal�1–4Xyl). Chondroitin/dermatansulfate is synthesized when GalNAc is transferred to thelinkage region, whereas heparin/heparan sulfate is syn-thesized if GlcNAc is added first. It has been demon-strated that the human exostoses-like family (EXTL1, -2,and -3) is responsible for the heparin/heparan sulfatechain initiation with the attachment of the first and secondGlcNAc residues and that the exostoses enzymes exto-sin-1 and -2 are the copolymerases that elongate the GAGchain with (GlcA�1–4GlcNAc�1–4)n (10 ). Recently, chon-droitin GalNAc transferases I and II and chondroitinsynthetase were discovered (61–63). Chondroitin GalNActransferases I and II are responsible for the initiation of thechondroitin/dermatan sulfate GAG chain with the attach-ment of the first few GalNAc residues to the linker region,whereas chondroitin synthetase acts as a copolymeraseand is responsible for the elongation of chondroitin/dermatan sulfate with (GalNAc�1–4GlcA�1–3)n.

golgi trafficThe Golgi apparatus consists of several cisternae, startingfrom the nucleus with the cis-Golgi network, through thecis-, medial-, and trans-Golgi compartments, and endingwith the trans-Golgi network, which are organized in theform of a stack. The Golgi position and organization

within a cell are sustained largely through the combinedefforts of a complex cytoskeletal matrix composed ofmicrotubules, an actin–spectrin network, and intermedi-ate filaments. The interaction between these filamentsystems and Golgi membranes is mediated by mechano-chemical enzymes such as dyneins, kinesins, myosins,and dynamin and different structural proteins (64 ).

A schematic overview of the transport route from ERthrough the different Golgi compartments is shown inFig. 3 [for reviews, see Refs. (65, 66)]. The journey ofproteins between these compartments starts with the exitfrom 100–200 export sites on the ER in COPII-coatedvesicles. Coatamer proteins (COPs) recognize transportsignals present in the cytoplasmic tail of cargo membraneproteins for their incorporation in COPII vesicles. Threeclasses of ER export signals have been described to date.Most type I membrane proteins have a diacidic or dihy-drophobic motif, and the type II glycosyltransferases havea [RK(X)RK] motif proximal to the transmembrane do-main (67 ). Signals that direct soluble cargo into ER-derived vesicles are less well defined. It is thought thatsoluble proteins are exported from the ER in 2 ways: (a)through a passive bulk flow process; and (b) through anactive receptor-mediated process relying on receptor-likeproteins that attach proteins to the inner membrane of thecoated vesicle (68 ).

Different COPII-coated vesicles fuse to form the ER-Golgi intermediate compartment (ERGIC). From here,escaped ER proteins or misfolded proteins are trans-ported back to the ER via COPI-coated vesicles. TheERGIC elements are transported to and fused with thecis-Golgi network. From here, both anterograde and ret-rograde transport is mediated via COPI-coated vesicles.Three major protein families regulate vesicle transport.The ARF and Sar1 family GTPases are involved in COPI

Fig. 3. Schematic overview of Golgi trafficmechanisms.(1), cargo proteins and v-SNAREs are incorporatedinto a COPII-coated vesicle. COPII coat assembly ismediated by Sar1-GTP. The coated COPII vesiclesare subsequently budded from the ER membrane.(2), the COPII vesicle becomes uncoated and teth-ers to the ERGIC via a Rab protein and a tetheringfactor. The v- and t-SNAREs assemble into a 4-helixbundle. This trans-SNARE complex promotes fusionof the vesicle with ERGIC, where cargo is trans-ferred to (3), in which ER proteins and misfoldedproteins are transported back to the ER via COPI-coated vesicles. (4), cargo remains in a Golgicompartment for further processing. Glycosyltrans-ferases are transported via COPI-coated vesicles totheir specific Golgi compartments. CGN, Cis-Golginetwork; TGN, trans-Golgi network.


and COPII vesicle formation, which starts with the acti-vation of ARF and Sar1 by a nucleotide exchange factorinto ARF-GTP and Sar1-GTP. ARF-GTP and Sar1-GTPrecruit many additional components for the synthesis ofthe vesicle coat. Subsequently, the Rab family GTPasesmediate vesicle targeting. The mammalian Rab proteinfamily includes at least 63 isoforms. All cytosolic Rabproteins form a complex with the Rab guanine nucleotidedissociation inhibitor chaperone, which transports theRab proteins to the membrane of specific Golgi compart-ments, where they become activated to the GTP state.Activated Rabs mediate vesicle motility and the tetheringof transport intermediates to their target membranes. Thethird family consists of SNARE proteins, which directvesicle fusion. Each type of transport vesicle carries aspecific vesicle-SNARE (v-SNARE), which binds to atethering-SNARE (t-SNARE) on the target membrane,producing the trans-SNARE complex. After fusion, thecargo is transferred to that specific compartment (65, 66).

The Golgi is a very dynamic organelle; it has thecapacity to transform in response to specific stimuli orcellular changes. For example, the Golgi or any Golgi-likestructures fragment into numerous tubular and vesicularstructures when cells undergo mitosis: the ER export sitesdisappear, the Golgi integral membrane proteins aretrapped in the ER, and Golgi peripheral proteins areretargeted to the ER or cytoplasm. After mitosis is com-plete, the Golgi is readily re-formed by outgrowth fromthe ER and fusion of the tubular and vesicular structures(69 ).

Until recently, the Golgi was seen as a static organelle.In this model, Golgi enzymes are retained within oneGolgi cisterna, and cargo (the proteins that are trans-ported to and processed in the Golgi) are transportedthrough the different Golgi compartments in the antero-grade direction via COPI vesicles. However, at present,the cisternal maturation model is favored. It is nowbelieved that cargo remains in one cisterna and that thisGolgi compartment traffics in the anterograde direction,whereas the Golgi enzymes traffic backward by COPIvesicles. This model is based on the experimental obser-vations that cargo indeed remains in a specific cisternaand that COPI vesicles are enriched with Golgi enzymes(70, 71).

Given the sequential and competing nature of glyco-syltransferases, the precise localization of these enzymeswithin the Golgi is of great importance. It is thought thatglycosyltransferases are arranged in an assembly line inthe Golgi, whereas early-acting transferases are localizedin the cis-Golgi, intermediate-acting transferases in themedial-Golgi, and terminating transferases in the trans-Golgi. A signal targeting glycosyltransferases to a specificGolgi localization has not yet been described. Studieshave indicated that glycosyltransferases from a certainGolgi compartment form high–molecular-mass com-plexes (72 ). The presence of multienzyme complexes is

likely to be functionally relevant in the regulation ofglycosylation and contribute to the maintenance of thesteady-state localization of the Golgi glycosyltransferases(72 ). When Nilsson and Warren (73 ) re-directed a Golgiresident glycosyltransferase to the ER, another Golgienzyme also was retained in the ER. Not all glycosyltrans-ferases form complexes; in particular, those found in thetrans-Golgi network seem to be unbound. Another factorthat is likely to play a role in the targeting of glycosyl-transferases is the thickness of the lipid bilayer, whichincreases en route to the plasma membrane. The fact thatGolgi proteins have shorter transmembrane domains thando plasma membrane proteins suggests that cisternae of aspecific compartment can accommodate glycosyltrans-ferases with a transmembrane domain of matchinglength. However, it has been shown that some solubleforms of glycosyltransferases, which have lost their trans-membrane domain, are retained in the Golgi probably asa result of being associated in complexes (70, 71). It islikely that more independent signals act together tomediate efficient Golgi localization.

conserved oligomeric golgi complex and itsrole in golgi trafficRecently, 2 patients were identified to have a defect insubunit 7 of the conserved oligomeric Golgi complex(COG7); the patients were classified as CDG type IIe (74 ).The mammalian COG complex contains 8 subunits, ofwhich COG1 through -4 form lobe A and COG5 through-8 form lobe B with COG4 as the core component linkingthe 2 lobes (75 ).

Mutations in COG subunits (COG1 through -8) ofCHO, yeast, and Drosophila melanogaster sperm cells havebeen shown to affect the structure and function of theGolgi, producing defects in glycoconjugate biosynthesis,intracellular protein sorting, protein secretion, and insome cases, cell growth. In the recessive COG1- andCOG2-null CHO mutants, for example, the Golgi showedan abnormal morphology with dilated cisternae andpleiotropic defects in several medial- and trans-Golgi–associated reactions affecting N-linked, O-linked, andlipid-linked glycoconjugates (76 ). The COG complex thusseems to play a role in determining and maintaining Golgistructure and morphology. Furthermore, COG works inconcert with COPI. Their function is to retrograde trans-port several Golgi resident proteins to the appropriateGolgi compartment where they reside. Evidence camefrom the work of Oka et al. (77 ), who investigated theconsequences of the loss and overexpression of COG on aset of Golgi resident type II transmembrane proteins,including members of the SNARE, Rab, and golgin pro-tein families. The expression and localization of someproteins were COG-dependent, whereas for others thiswas not the case. The COG-sensitive proteins are referredto as “GEARs”.


Functions of O-Linked GlycansVarious functions have been described for O-linked gly-cans. Only the main roles are given here; for details, thereader is referred to other reviews (8, 78). In general,O-linked glycans have been found to function in proteinstructure and stability, immunity, receptor-mediated sig-naling, nonspecific protein interactions, modulation of theactivity of enzymes and signaling molecules, and proteinexpression and processing. The biological roles of oligo-saccharides appear to span the spectrum from those thatare trivial to those that are crucial for the development,growth, function, or survival of an organism. A particularglycan may mediate diverse functions at distinct locationsat specific times within a single organism (79 ).

Just like N-glycans, O-glycans can influence the sec-ondary protein structure: the glycan can break the �-he-licity of peptides (80 ); can have a role in the tertiaryprotein structure [seen, e.g., on the porcine filamentous-shaped submaxillary mucin, in which release of an O-glycan leads to a globular shape (81 )]; and in the quater-nary protein structure and protein aggregation [seen, e.g.,in ovine submaxillary mucin, which only forms aggre-gates when it is O-glycosylated (82 )]. Subsequently, O-linked glycans maintain protein stability, heat resistance,hydrophilicity, and protease resistance by steric hin-drance (8 ).

Additionally, mucin-type O-glycans are important forthe binding of water. Mucins are proteins that are heavilyglycosylated with mucin-type O-glycans and are oftenpresent at outer surfaces lacking an impermeable layer,such as the surfaces of the digestive, genital, and respira-tory system tracts. These mucins bear clusters of sialy-lated glycans, which produce regions with a strong neg-ative charge. This gives mucins the capacity to bind largeamounts of water and form mucus. The gels observed innasal secretions, for example, are formed by secretedMUC2 polypeptides linked together to form long, cross-linked polymers holding water. The primary function ofthese viscous mucin solutions and gels is to form aprotective coating with antibacterial properties (83 ). Likemucin-type O-glycans, GAGs bind large volumes of watervia the strong negative charge of the sulfate groups,providing resilience or resistance to compression ratherthan lubrication or reduction of friction. GAGs are foundin extracellular matrices. In structural tissues, such as thecartilage of joints, GAGs can act as shock breakers by theslow reduction of their water content under high pressure.

Another important function of O-linked sugars is tomediate recognition between proteins. Glycan structurescan be substrates for nonenzymatic sugar-binding pro-teins, known as lectins. By interacting with lectins, gly-cans influence the targeting of the proteins to which theyare attached. Examples of glycan-mediated recognition ofglycoproteins are ubiquitous. For example, selectins andgalectins, representing 2 classes of lectins located in theleukocyte-vascular system, bind to carbohydrate epitopesthat induce cellular signaling, which in turn influences

many crucial cellular processes, including cell growth,apoptosis, endocytosis, cell-cell interactions, cell–ma-trix interaction, matrix network assembly, and oocytefertilization (84 ). Additionally, sialylated O-mannosylglycan serves as binding ligand for laminin in thedystroglycan complex, which is important in muscleand brain development (18 ). Moreover, O-linked gly-cans are known to have an effect on immunologicrecognition; for example, the ABO blood group anti-gens and recognition of glycopeptides by the MHCcomplex or by antibodies (85 ).

Subsequently, it is known that GAGs have a role innonspecific protein interactions. Cell surface proteogly-cans, for example, adhere to soluble polypeptide growthfactors through electrostatic interactions mediated bytheir GAGs, preventing the growth factors from diffusing.GAG interactions increase and stabilize concentrationgradients of growth factors (78 ).

The effects of O-linked glycosylation on the bioactivityof many signaling molecules, particularly hormones andcytokines, and a relatively small number of enzymes,have been described. In most cases, the influence is notvery strong (a difference of 2- or 3-fold), but ratherprovides a fine regulation mechanism. Effects leading toboth an increase and a decrease in biological activity havebeen described. For example, a mucin-type O-linkedglycan decreases the biological activity of interleukin-5(86 ), whereas it induces a higher enzymatic activity ofhuman lactase phlorizin hydrolase (87 ). The influence ofunusual carbohydrate modifications on the activity ofsignaling molecules appears often to be crucial and spe-cific. For example, O-Fuc on urinary-type plasminogenactivator was shown to be required for activation of itsreceptor, and the presence of O-fucosyl glycans seems tobe required for proper Notch function (22 ). Anotherexample is the dynamic O-GlcNAc modification thatseems to have an important role in a variety of signalingpathways, such as transcriptional regulation, proteasome-mediated protein degradation, insulin, and cellular stresssignaling. Recently, it was found that O-GlcNAc modu-lates the activity of critical intermediates involved in theregulation of neutrophil motility (88 ). Some very specificGAG structures are known to act as co-receptors, allowingactivation of the primary receptor necessary for the acti-vation of growth factors. The fibroblast growth factor, forexample, must interact with the heparan sulfate chain ofthe proteoglycan syndecan to activate the primary fibro-blast growth factor receptor (89 ).

Finally, O-linked glycosylation is essential for the ex-pression and processing of particular proteins. Glyco-phorin A, for example, is a heavily glycosylated proteinpresent on the surface of human erythrocytes. It has beenshown that O-linked sugars are necessary for cell surfaceexpression of this glycoprotein (90 ). The influence ofO-glycans in the processing of proteins is, for example,seen in pro-insulin-like growth factor II, which is cleavedinto IGF-II only when Thr75 contains an O-linked sugar (91).


As O-glycans are involved in numerous processes, it isinevitable that defects in O-glycan biosynthesis mightlead to severe abnormalities for cellular functioning.

Congenital Disorders in the Biosynthesis of O-Glycans inHumans

CDG form a group of autosomal recessive metabolicdisorders caused by defects in the biosynthesis of protein-linked glycans. To date, mainly genetic defects in N-glycan biosynthesis have been classified as CDG. Thedivision of CDG into types I and II is based on the locationof the defect in the N-glycan biosynthetic pathway. CDG-Iincludes all defects in the early N-glycan pathway in thecytoplasm or the ER and covers all steps until the transferof the glycan to the protein. CDG-II includes all defectslocalized in the processing of N-glycans on the glycosy-lated protein. These are situated mainly in the Golgicompartment. At present, most defects in the biosynthesisof protein O-linked glycans are not included in this CDGclassification and still have “popular” names and/or“biochemical” names that are informative about the na-ture of the disease. Some O-glycosylation disorders affectonly a particular O-glycan type, certain disorders affectmore O-glycan types, and others also affect the biosyn-thesis of other glycoconjugates. It is becoming increas-ingly evident that the primary defect of these disorders isnot necessarily localized in one of the glycan-specifictransferases, but can likewise be found in the biosynthesisof nucleotide sugars, their transport to the ER/Golgi, andin Golgi trafficking. The clinical variations within a dis-order and among the different inborn errors of O-glycanmetabolism are enormous. Defects can lead to a severeautosomal recessive multisystem syndrome with neuro-logic involvement, whereas some defects, for example,those in persons with the Bombay blood group or theLewis-null blood group, do not produce a clinical pheno-type. As O-glycosylation biosynthesis is a very complexprocess with an enormous number of genes involved, it isobvious that the disorders described to date are just thetip of the iceberg. This novel area of inborn errors ofmetabolism still needs further exploration.

This section discusses the clinical, molecular genetic,laboratory, and biochemical aspects of the known congen-ital disorders in the biosynthesis of O-glycans. Thehuman congenital disorders that affect the biosynthesisof protein O-linked glycans are summarized in Table3A, whereas the human congenital disorders with adefect affecting the biosynthesis of both N- and O-glycans are summarized in Table 3B. Both parts ofTable 3 also list the group(s) who first discovered thegenetic defects in the disorders.

defects in mucin-type o-glycan biosynthesisUDP-GalNAc transferase 3 (polypeptide N-acetylgalactosami-nyltransferase 3) deficiencyThe GALNT3 gene encodes UDP-GalNAc transferase 3(GalNT3; EC 2.4.1.41), which transfers UDP-GalNAc to

Thr/Ser of a protein backbone. GALNT3 is expressed inorgans that contain secretory epithelial glands. It is highlyexpressed in human pancreas, skin, kidney, and testis andweakly expressed in prostate, ovary, intestine, and colon(92 ). Patients with familial tumoral calcinosis (FTC) canhave mutations in the GALNT3 gene.FTC. FTC is an autosomal recessive progressive metabolicdisorder that manifests with massive calcium deposits inthe skin and subcutaneous tissues and unresponsivenessto parathyroid hormone (93 ). At present, FTC is the onlysyndrome with an isolated defect in mucin-type O-glycanbiosynthesis (94 ). The syndrome can be treated withphosphate-binding antacids (aluminum hydroxide) and alow-phosphorus diet combined with calcium deprivation,which reduces and prevents the recurrence of calcificmasses (95 ).

Laboratory findings: Hyperphosphatemia has been de-scribed accompanied by inappropriately normal orincreased concentrations of parathyroid hormoneand 1,25-dihydroxyvitamin D3, two essential regula-tors of phosphate metabolism. Serum calcium iswithin reference values, and 25-hydroxycholecalcif-erol is decreased (94 ).

Diagnosis: Recently immunostaining with a monoclonalantibody against GalNT3 revealed that the proteinwas absent in a frozen skin biopsy from a patientwith FTC, whereas GalNT3 was strongly expressedin the epidermis of a healthy individual. This sug-gests that immunostaining of skin biopsy samples forGalNT3 might be a useful tool in the diagnosis of thisdisorder (96 ). The diagnosis can be confirmed at themolecular genetic level (94 ).

defects in gag biosynthesis�-1,4-Galactosyltransferase 7 deficiencyThe B4GALT7 gene encodes �-1,4-galactosyltransferase 7(B4GalT7; EC 2.4.1.133), which transfers Gal to the Xyl-Serlinkage in the linker region of proteoglycans (97 ).B4GALT7 is expressed in human heart, pancreas, liver,and to a lesser extent, in placenta, kidney, brain, skeletalmuscle, and lung. Patients with the autosomal recessiveprogeroid variant of Ehlers–Danlos syndrome have mu-tations in the B4GALT7 gene.Progeroid variant of Ehlers–Danlos syndrome. To date, 5patients have been described with the progeroid variantof Ehlers–Danlos. Characteristic clinical features are apremature aging phenotype with a loose, elastic skin,failure to thrive, joint laxity, psychomotor retardation,hypotonia, and macrocephaly. Because proteoglycans areimportant structural components of the extracellular ma-trix of connective tissue, these patients suffer from skin,cartilage, and bone problems.

Laboratory findings: Thyroid, kidney, and liver functiontest results are within reference values. Urine organicacids, amino acids, mucopolysaccharides, and oligo-saccharides were all within reference values. In ad-


dition, nitroprusside test results, chromosomal stud-ies, serum creatine kinase concentrations, and growthhormone concentrations were all normal (98 ).

Diagnosis: This disorder can be diagnosed at the enzymelevel by use of an assay for galactosyltransferase I inhuman fibroblasts (99 ). Mutations can be found inthe corresponding gene (97 ).

Deficiencies of extosin-1, -2, and -3The EXT1 and EXT2 genes encode the proteins extosin-1and -2, respectively, which oligomerize with the copoly-merases (EC 2.4.1.224 and 2.4.1.225) responsible for theelongation of the heparin and heparan sulfate chains. Theexact function of extosin-3 is not known. The EXT genesare ubiquitously expressed in human tissue. Patients with

Table 3. Congenital disorders of glycosylation.A. Human congenital disorders of O-glycosylation.

Name OMIM Gene O-Glycan type Reference(s)

Defects in mucin type O-glycan biosynthesisFTC 211900 GALNT3 Mucin-type (94)

Defects in GAG biosynthesisProgeroid variant of Ehlers–Danlos 130070 B4GALT7 GAG (97)HME type I 133700 EXT1 Heparan/heparan sulfate (102)HME type II 133701 EXT2 Heparan/heparan sulfate (103)HME type III 600209 EXT3 Heparan/heparan sulfate (100)

Defects in GAG sulfationMCD 217800 CHST6 Keratan sulfate (104)SED type Omani 608637 CHST3 Chondroitin sulfate (111)ACGB1B 600972 DTDST Sulfated GAGs (122)AO-II 256050 DTDST Sulfated GAGs (123)Diastrophic dysplasia 222600 DTDST Sulfated GAGs (124)EDM4 226900 DTDST Sulfated GAGs (125)SED Pakistany 603005 ATPSK2 Sulfated GAGs (126)

Defects in O-galactosyl glycan biosynthesisEhlers–Danlos syndrome type VI 225400 PLOD O-linked Gal (134)

Defects in O-mannosyl glycan biosynthesisWWS 236670 POMT1, POMT2,

FCMD, or FKRPO-linked Man (148, 158, 159, 169)

LGMD2K 609308 POMT1 O-linked Man (142)MEB 253280 POMGNT1 or FKRP O-linked Man (152, 169)

Putative defects in O-mannosyl glycanbiosynthesis

FCMD 253800 FCMD O-linked Man? (162)CMD1C 606612 FKRP O-linked Man? (163)LGMD2I 607155 FKRP O-linked Man? (167)CMD1D 608840 LARGE O-linked Man? (166)

Defects in O-glycan sialylationhIMB 600737 GNE Sialylated O-linked glycans (188)DMRV 605820 GNE Sialylated O-linked glycans (189)Sialuria 269921 GNE Sialylated O-linked glycans (190)CDG-IIf SLC35A1 Sialylated O-linked glycans (49)

B. Human combined congenital disorders of N- and O-glycosylation.

Bombay blood group 211100 FUT1 or FUT2 A, B, H blood group determinants (193)Para-Bombay blood group 211100 FUT1 A, B, H blood group determinants (191)Non-secretor blood group 182100 FUT2 A, B, H blood group determinants (192)Lewis-negative blood group 111100 FUT3 Lea and Leb determinants (194)CDG-IIc 266265 FUCT1 Fucosylated N- and mucin-type

O-glycans(44, 48)

CDG-IId 607091 B4GALT1 N-Acetyllactosamine unit (200)CDG-IIe 608779 COG7 N- � O-glycans (74)CMRD 246700 SARA2 N-glycans; O-glycans? (211)Anderson disease 607689 SARA2 N-glycans; O-glycans? (211)CMRD with Marinesco–Sjögren syndrome 607692 SARA2 N-glycans; O-glycans? (211)


hereditary multiple exostoses (HME) have a mutatedEXT1, EXT2, or EXT3 gene.HME (types I, II, and III). HME is a genetically hetero-geneous autosomal dominant disorder characterized bythe development of multiple cartilage-capped benignbone tumors (exostoses) located mainly on the long bones.This disorder is often accompanied by skeletal deformitiesand short stature. In many cases, the exostoses transformto malignant tumors. Mutations in EXT1 and EXT2 ac-count for 44%–66% and 30% of HME patients, respec-tively, whereas EXT3 appears to be the minor locus (100).For a review on hereditary multiple exostoses, please seethe article by Wicklund et al. (101).

Laboratory findings: No laboratory findings have beenreported that may aid in diagnosis.

Diagnosis: Genetic confirmation of the diagnosis can beobtained by mutation analysis of EXT1, EXT2, andEXT3 (100, 102, 103).

defects in gag sulfationN-Acetylglucosamine-6-O-sulfotransferase deficiencyThe CHST6 gene encodes human GlcNAc-6-O-sulfotrans-ferase (EC 2.8.2.-) that transfers sulfate to the 6-O positionof GlcNAc and Gal residues in the poly-N-acetyllac-tosamine extensions in keratan sulfate. GlcNAc-6-O-sul-fotransferase is produced in human cornea, brain, spinalcord, and trachea. Macular corneal dystrophy (MCD) iscaused by distinct mutations in the gene CHST6.MCD (types I and II). MCD is a progressive autosomalrecessive disease in which minute, gray, punctuate opac-ities in the cornea lead to bilateral loss of vision. Onset ofclinical signs occurs in the first decade of life. Mostpatients have painful attacks with photophobia, foreignbody sensations, and recurrent corneal erosions. MCD ischaracterized by nonsulfated (MCD type I) or low-sul-fated (MCD type II) keratan sulfate (104).

Laboratory findings: MCD patients have an accumulationof GAGs in corneal fibroblasts (105, 106). Keratansulfate in serum and cartilage is nonsulfated orlow-sulfated (107, 108). The defect is thus not re-stricted to cells in the cornea. No abnormalities havebeen described in the total amount of urinary GAGs orin GAG subfractions. In addition, the electrophoreticmigration of urinary GAG subfractions was normal.

Diagnosis: In earlier days, histopathologic diagnosis ofMCD was based on the fact that GAG deposits stainpositively with Hale’s colloidal iron, alcian blue,periodic acid–Schiff, and metachromic dyes (109).Currently, MCD can be diagnosed with an ELISAthat makes use of a monoclonal antibody specific fora sulfated epitope on keratan sulfate (110). Thesubdivision of MCD into types I and II is based on theresults of this ELISA. The diagnosis can be confirmedat the molecular genetic level (104).

Chondroitin 6-sulfotransferase 1 deficiencyThe CHST3 gene encodes chondroitin 6-sulfotransferase 1(EC 2.8.2.17), which catalyzes the sulfation of the 6-Oposition of GalNAc residues in chondroitin sulfate chains.CHST3 is widely expressed in adult tissues. It is expressedin the human heart, placenta, skeletal muscle, and pan-creas, but also in various immune tissues such as thespleen, lymph nodes, and thymus. Recently, it was foundthat mutations in the CHST3 gene cause autosomal reces-sive spondyloepiphyseal dysplasia (SED), Omani type(111).SED type Omani. Patients with SED Omani type arenormal in length at birth but show growth retardationlater, and are short in stature in adulthood (110–130 cm).Severe progressive kyphoscoliosis, severe arthriticchanges with joint dislocations, rhizomelic limbs, genuvalgum, cubitus valgus, mild brachydactyly, camptodac-tyly, and microdontia occur in this disease (112).

Laboratory findings: Laboratory investigations revealedhematologic indices and biochemistry results withinreference values, normal results for routine metabolicinvestigations, and normal karyotype. Urinary excre-tion of mucopolysaccharides was normal. Thyroidfunction was normal, and growth hormone, insulin-like growth factor 1, follicle-stimulating hormone,and prolactin concentrations were within referencevalues (112).

Diagnosis: Chondroitin 6-sulfotransferase 1 activity canbe measured in human fibroblasts (111). Mutationscan be found in the corresponding gene (111).

Diastrophic dysplasia sulfate transporter deficiencyThe DTDST gene (also called SLC26A2) encodes thediastrophic dysplasia sulfate transporter (DTDST), whichis a sulfate/chloride antiporter. The primary source ofsulfur for the sulfation pathway of proteoglycans is freeSO4

2�, which is transported to the cytoplasm mainly byDTDST. Mutations in the sulfate transporter lead toundersulfation of the GAGs. DTDST is ubiquitously ex-pressed. Mutations in the DTDST gene are the cause ofdiastrophic dysplasia (DTD), achondrogenesis type 1B(ACGB1), atelosteogenesis type II (AO-II), and multipleepiphyseal dysplasia 4 (EDM4). The clinical features inthe DTDST skeletal dysplasia family range from a rela-tively mild condition to severe conditions incompatiblewith life and are subdivided into the 4 syndromes listedabove. The disorders have autosomal recessive inheri-tance. The severity of the phenotype correlates with theunderlying DTDST mutation; mutations leading to stopcodons or transmembrane domain substitutions mostlylead to the most severe phenotype (ACGB1), whereasother structural or regulatory mutations usually lead toone of the less severe phenotypes (113). The classificationof DTD, AO-II, or EDM4, and thus of the severity of thedisease, depends on residual sulfate uptake capacity and


the extent of proteoglycan undersulfation (114). For areview, see the article by Rossi and Superti Furga (115).ACGB1. ACGB1 is among the most severe skeletal disor-ders in humans. The disease is characterized by severehypodysplasia of the spine, the rib cage, and the extrem-ities. ACGB1 is always lethal immediately after andsometimes even before birth (113).AO-II. AO-II is a lethal chondrodysplasia caused bycollapse of the airways, resulting from abnormalities inthe tracheal, laryncheal, and bronchial cartilage. Pheno-typically, AO-II is the severe variant of DTD. In additionto the clinical features described for DTD, AO-II is char-acterized by severe and progressive kyphosis, horizontalsacrum, and a gap between the first and the second toes(116).DTD. DTD is a skeletal dysplasia associated with shortstature [adult height, 100–140 cm (117)], joint contrac-tures, cleft palate, scoliosis, bilateral clubfeet, and charac-teristic clinical signs such as the so-called “hitchhikerthumb” and cystic swelling of external ears. Phenotypicvariability is wide (115, 118).EDM4. Patients with EDM4 have a condition with club-foot, scoliosis, mild finger deformity, and mildly short ornormal stature, but without palatal clefting, ear swelling,or thumb deviation (119).

Laboratory findings: Histochemical studies revealed thatcartilage proteoglycans of ACGB1 patients werequantitatively decreased and do not stain with tolu-idine blue. Impaired synthesis of sulfated proteogly-cans was observed in fibroblast cultures of an ACGB1patient (120). There have been no published studieson the quantitative excretion or the composition ofurinary GAGs, the catabolic products of proteogly-cans.

Diagnosis: Sulfate transport within cells can be mea-sured in human fibroblasts (121). The diagnosis ofDTDST deficiency can be confirmed at the moleculargenetic level (122–125).

3�-Phosphoadenosine 5�-phosphosulfate synthase 2 deficiencyThe ATPSK2 gene encodes the enzyme 3�-phosphoad-enosine 5�-phosphosulfate synthase 2 (PAPSS2; EC2.7.1.25/EC 2.7.7.4) (126). PAPSS2 is a bifunctional en-zyme that activates cytoplasmic SO4

2� into a high-energyform in 2 enzymatic steps: (a) its ATP-sulfurylase usesATP and SO4

2� to synthesize adenosine 5�-phosphosul-fate (APS) and (b) its APS kinase catalyzes the phosphor-ylation of APS to 3�-phosphoadenosine 5�-phosphosulfate(PAPS). PAPS is the universal sulfate donor for posttrans-lational protein sulfation. Defective PAPSS2 thus leads toundersulfation of GAGs. PAPSS2 is produced in humancartilage. In a large Pakistani family, comprising 8 gener-ations, a mutation was found in the ATPSK2 gene thatleads to SED, Pakistany type.SED, Pakistany type. The clinical features of SED, Paki-stany type, include short stature evident at birth; short,

bowed lower limbs; a mild, generalized brachydactyly,kyphoscoliosis, an abnormal gait, and early-onset degen-erative joint disease in the hands and knees. Radiographsshowed delayed epiphyseal ossification, especially of thehips and knees, and platyspondyly. Inheritance of thedisease is autosomal recessive (127).

Laboratory findings: No laboratory findings have beenreported that may aid in diagnosis (128).

Diagnosis: PAPSS2 activity can be measured in humanliver biopsy samples (128). Diagnosis can be con-firmed at the molecular genetic level (126).

defects in o-galactosyl glycan biosynthesisLysyl hydroxylase-1 deficiencyThe PLOD gene encodes lysyl hydroxylase-1 (EC1.14.11.4). Lysyl hydroxylase-1 catalyzes the formation ofhLys in collagens and other proteins with collagen-likeamino acid sequences by the hydroxylation of Lys resi-dues. hLys serves as attachment site for O-galactosylglycans and is essential for the formations of collagencross-links, contributing to collagen structure and stabil-ity. Lysyl hydroxylase-1 deficiency indirectly leads to anO-glycosylation defect. The function of the O-galactosylglycans is unclear, although it is suggested that they mayplay a role in recognizing and activating collagen recep-tors in the cell membrane (129). Subsequently, it has beenshown that there is a relationship between cross-linkcontent and the degree of collagen glycosylation (130).Lysyl hydroxylase is produced in human liver, heart,lung, skeletal muscle, brain, and placenta (131). Patientswith Ehlers–Danlos syndrome type VIa have mutations inthe PLOD gene.Ehlers–Danlos syndrome type VIa. Patients with Ehlers–Danlos type VIa are characterized clinically by neonatalkyphoscoliosis, generalized joint laxity, skin fragility, andsevere muscle hypotonia at birth. Arterial rupture hascaused death in some patients (132). The inheritance ofthe syndrome is autosomal recessive.

Laboratory findings: Urinary amino acids were withinreference values. Amino acid analysis of dermalcollagen showed a marked decrease in hLys content(133).

Diagnosis: The activity of lysyl-protocollagen hydroxy-lase can be measured in cultured skin fibroblasts(133). The diagnosis can be further confirmed bymutation analysis in the corresponding gene (134).

defects in o-mannosyl glycan biosynthesisO-Mannosyl glycan biosynthesis disorders are character-ized by an abnormal �-dystroglycan glycosylation. �-Dys-troglycan is an essential component of the dystropin–glycoprotein complex, which is produced in humantissues such as muscle, brain, nerve, and heart. Thedystrophin–glycoprotein complex is a multimeric trans-membrane complex, providing a tight connection be-tween the cytoskeleton and the extracellular matrix. Dys-


troglycan is generated from a single gene (DAG1) and issubsequently cleaved into 2 subunits: transmembrane�-dystroglycan and peripheral �-dystroglycan. In muscle,the intracellular side of transmembranic �-dystroglycanbinds to a variety of cytoplasmic molecules, such asdystropin, which in turn interacts with the cytoskeleton ofcells. The extracellular side of �-dystroglycan binds non-covalently to �-dystroglycan, which in turn binds toextracellular matrix proteins such as laminin (135, 136).Different mammalian glycan sequencing studies haverevealed that �-dystroglycan is heavily glycosylated withO-linked Man chains (�70%) and to a lesser extent withmucin-type O-glycans (�30%), which mediate protein-protein interactions (137, 138). For the screening of defectsin O-mannosyl glycan biosynthesis, the immunohisto-chemical staining of �-dystroglycan is used on musclebiopsies of patients. At present, �-dystroglycan is the onlyknown substrate for this type of glycosylation in mam-mals. Often the monoclonal antibodies VIA4-1 and IIH6,which recognize an unknown carbohydrate epitope in�-dystroglycan, are used. Antibodies against the corestructures of �- and �-dystroglycan serve as controls insuch experiments (139).

Protein O-mannosyltransferase deficiencyCoexpression of the POMT1 and the POMT2 genes isnecessary for the enzymatic activity of protein O-manno-syltransferase [EC 2.4.1.109 (57 )]. Mannosyltransferasecatalyzes the attachment of Man residues to Thr/Seramino acids of a protein. POMT1 is highly expressed inhuman testis, heart, and pancreas, whereas expression islower in kidney, skeletal muscle, brain, placenta, lung,and liver. Walker–Warburg syndrome (WWS) and limb-girdle muscular dystrophy type 2K (LGMD2K) can becaused by mutations in the POMT1 gene. POMT2 ishighly expressed in testis, with expression lower in mosttissues. Recently, it was discovered that mutations in thePOMT2 gene also cause WWS (140).WWS. In 20% of WWS patients, a mutation is found in thePOMT1 gene. The incidence of POMT2 mutations is in thesame range as that of POMT1 (140). The phenotype seenin the WWS patients with a POMT2 mutation is indistin-guishable from that of patients with POMT1 mutations.Patients with this rare autosomal recessive disorder havea life expectancy of �3 years (mean, 0.8 years). WWSpatients have malformations of the muscle, eye, and brain.Typical brain anomalies include hydrocephalus, cerebel-lar hypoplasia, absent corpus callosum and cerebellarvermis, cobblestone cortex, and fusion of the hemi-spheres. Additionally, WWS patients can have numerouseye anomalies, such as cataracts, microphthalmia, persis-tent hyperplastic primary vitreous, and Peters anomaly.WWS patients have little motor activity because of severemuscle dystrophy (141). For a review, see van Reeuwijk etal. (141).LGMD2K. Patients with LGMD2K have progressive mus-cle weakness involving the proximal muscles of the

shoulder and pelvic girdles. These patients also have aslow, progressive limb-girdle muscular dystrophy, a mildmicrocephaly, and severe mental retardation, but normalbrain imaging. Onset of the autosomal recessive disorderis in the first decade of life (142, 143).

Laboratory findings: In both WWS and LGMD2K pa-tients, serum creatine kinase was increased and stain-ing of �-dystroglycan by the VIA4-1 and/or IIH6monoclonal antibodies was abnormal (142, 144, 145).Decreased staining of the laminin �2-chain (merosin)has been observed in WWS, although patients havebeen reported with normal amounts of merosin(146, 147).

Diagnosis: The activity of protein O-mannosyltrans-ferase can be measured in human kidney cells (57 ).The diagnosis of POMT1 deficiency can be confirmedby mutation analysis of the gene (142, 148).

O-Mannosyl-�1,2-N-acetylglucosaminyltransferase-1 deficiencyThe POMGNT1 gene encodes the enzyme O-mannosyl-�1,2-N-acetylglucosaminyltransferase-1 (EC 2.4.1.101).This enzyme catalyzes the second step, the linkage of aGlcNAc residue to protein-bound Man, in the O-manno-syl glycan core structure. The enzyme O-mannosyl-�1,2-N-acetylglucosaminyltransferase-1 appears to be presentin all tissues. Muscle–eye–brain disease (MEB) is causedmainly by mutations in the POMGNT1 gene.MEB. MEB is a muscular dystrophy/neuronal migrationdisorder with a phenotype similar to, but less severe than,that of WWS patients. The life expectancy of MEB patientsis 10–30 years (144). Clinically, MEB is differentiated fromWWS mainly on the basis of the presence of a normal orthin corpus callosum and of pronounced cerebellar cysts,which are both absent in WWS patients (144). The inher-itance of the disorder is autosomal recessive. For a review,see Diesen et al. (149).

Laboratory findings: MEB patients have increased serumcreatine kinase and abnormal staining of �-dystrogly-can by VIA4-1 and/or IIH6 monoclonal antibodies(144, 150). Staining of the laminin �2-chain (merosin)is generally normal (150).

Diagnosis: The activity of O-mannosyl-�1,2-N-acetylglu-cosaminyltransferase-1 can be measured in humanmuscle biopsies (151). The diagnosis can be furtherconfirmed by molecular genetic analysis of the cor-responding gene (152).

putative defects in o-mannosyl glycanbiosynthesisFukutin deficiencyThe FCMD gene encodes the protein fukutin. The functionof fukutin is not known, but it is suggested to be aglycosyltransferase (153). Fukutin is produced in manyparts of the body, with the highest amounts in the brain,heart, pancreas, and skeletal muscle. Fukuyama-type con-genital muscular dystrophy (FCMD) and WWS can be


caused by mutations in the FCMD gene. Deficiencies infukutin are putative defects in O-mannosyl glycan bio-synthesis. These deficiencies might be reclassified de-pending on the function of fukutin.FCMD. FCMD is an autosomal recessive disorder inwhich patients manifest with generalized muscle weak-ness, severe hypotonia, mental retardation, and brainmalformations. Brain malformations are very similar tothose reported in WWS and MEB and include cerebraland cerebellar micropolygyria, hydrocephalus, fibroglialproliferation of the leptomeninges, focal interhemisphericfusion, and hypoplasia of the corticospinal tracts(154, 155). Compared with MEB and WWS patients, theeye involvement in patients with FCMD is more variable,ranging from myopia to retinal detachment, persistentvitreous bodies, persistent hyaloid artery, or microphthal-mia (156). For a review, see Toda et al. (157).WWS. Severe mutations in the FCMD gene lead to WWS(158, 159).

Laboratory findings: In both FCMD and WWS patients,serum creatine kinase was increased. Subsequently,abnormal staining of �-dystroglycan by the VIA4-1and/or IIH6 monoclonal antibodies and decreasedstaining of laminin �2-chain were observed in allcases investigated (146, 160, 161).

Diagnosis: As the function of fukutin is unknown atpresent, the diagnosis can be confirmed only at themolecular genetic level (158, 162).

Deficiencies in fukutin-related proteinThe FKRP gene encodes the fukutin-related protein(FKRP), the function of which is undefined at present.FKRP is expressed in skeletal muscle, placenta, and heartand weakly in brain, lung, liver, kidney, and pancreas.FKRP is predicted to be a tissue-specific glycosyltrans-ferase involved in the O-mannosylation of �-dystroglycan(163). Mutations in the FKRP gene with autosomal reces-sive inheritance have been found in congenital musculardystrophy type 1C (MDC1C), LGMD2I, WWS, and MEB,but also in asymptomatic cases (164). Deficiencies inFKRP are putative defects in O-mannosyl glycan biosyn-thesis. These deficiencies might be reclassified dependingon the function of FKRP.MDC1C. The onset of the characteristic clinical features ofMDC1C occurs at birth or within the first 6 months of life;these features include severe weakness and wasting of theshoulder-girdle muscles, hypertrophy and weakness ofthe leg muscles with an inability to walk, and a severerestrictive pulmonary disease leading to respiratory fail-ure in the second decade of life. Cardiomyopathy hasbeen observed in several patients (163). The brain is notalways involved in this disorder. Severe mutations canlead to structural cerebellar changes (165) or more exten-sive structural brain and eye involvement similar to thatseen in MEB and WWS (166).

LGMD2I. In LGMD2I, the age at onset of clinical signsranges from 6 months to 40 years. The disorder presentsas hypotonia, weakness in the hip and shoulder-girdlemuscles, and hypertrophy of the calf muscles. The spec-trum ranges from infants with early presentation and aDuchenne-like disease course, including cardiomyopathy,to milder phenotypes with a long-term outcome. More-over, these patients lack structural brain or eye involve-ment (167, 168).WWS and MEB. Recently, a patient diagnosed with WWSand a patient diagnosed with MEB were found to have amutated FKRP gene (169).Asymptomatic carriers for homozygous FRKP muta-tions. De Paula et al. (164) investigated 86 BrazilianLGMD genealogies and identified 4 persons with novelhomozygous FKRP gene mutations who were asymptom-atic.

Laboratory findings: In patients with the 4 disorders,serum creatine kinase was increased. In all MDC1Ccases investigated, staining �-dystroglycan byVIA4-1 and/or IIH6 monoclonal antibodies was ab-normal, and staining of laminin �2-chain was de-creased, whereas the LGMD2I cases studied showeda variable decreases in VIA4-1 and/or IIH6 stainingand often (but not always) a laminin �2-chain defi-ciency was found (163, 167, 170).

Diagnosis: As the function of FKRP is unknown atpresent, the diagnosis can be confirmed only at themolecular genetic level (163, 167, 169).

N-Acetylglucosaminyltransferase-like protein deficiencyThe LARGE gene encodes the N-acetylglucosaminyltrans-ferase-like protein (LARGE), a homolog of mammalian�1,3-N-acetylglucosaminyltransferase. The gene is ubiqui-tously expressed, with the highest expression in the heart,brain, and skeletal muscle (171). Mutations in the LARGEgene have been found in a patient with MDC1D. Barresi etal. (172) showed that gene transfer of LARGE not onlyrestores the �-dystroglycan function in a LARGE-deficientmouse and in an MDC1D patient, but also in patients withFCMD, MEB, WWS, LGMD2I, and other glycosyltrans-ferase-deficient muscular dystrophies. Kanagawa et al.(173) showed that the N-terminal domain of �-dystrogly-can serves as an intracellular recognition site for LARGE,which initiates subsequent functional glycosylation of�-dystroglycan. This glycosylation is essential for ligandbinding and cell surface laminin/perlecan organization.Thus, LARGE is the key determinant of the functionalexpression of �-dystroglycan (173). Recently, it wasshown that LARGE is localized in the Golgi, where itstimulates glycosylation. In a patient with MDC1D,LARGE was mislocalized and thus failed to have an effecton �-dystroglycan glycosylation (174). Inducing LARGEexpression or activity could be an attractive target for thedesign of therapeutics for glycosyltransferase-deficientcongenital muscular dystrophies (172). LARGE deficiency


is a putative defect in O-mannosyl glycan biosynthesis.This deficiency might be reclassified depending on thefunction of LARGE.MDC1D. One patient with MDC1D was identified with acompound heterozygous mutation in the LARGE gene(166). The LARGE-deficient patient presented with con-genital muscular dystrophy, profound mental retarda-tion, white matter changes, and subtle structural brainabnormalities (166).

Laboratory findings: The MDC1D patient has increasedserum creatine kinase. In addition, staining of �-dys-troglycan by the VIA4-1 and/or IIH6 monoclonalantibodies was abnormal, whereas staining of thelaminin �2-chain was normal (166).

Diagnosis: Because the function of LARGE is unknownat present, the diagnosis can be confirmed only at themolecular genetic level (166).

defects in o-glycan sialylationUDP-N-Acetylglucosamine 2 epimerase/N-acetylmannosaminekinase deficiencyThe GNE gene encodes the enzyme GNE/MNK (EC5.1.3.14/EC 2.7.1.60). GNE/MNK is a bifunctional en-zyme that catalyzes the first 2 steps of the biosynthesis ofthe nucleotide sugar CMP-NeuAc. GNE/MNK activity ishighest in the liver and placenta, and it s also found in theheart, brain, lung, kidney, skeletal muscle, and pancreas.Although CMP-NeuAc is incorporated in both N- andO-glycans, it seems that defects in GNE/MNK influenceonly the sialylation of O-linked glycans and not of N-glycans (175). Mutations in GNE/MNK have been de-scribed in hereditary inclusion body myopathy (hIBM), indistal myopathy with rimmed vacuoles (DMRV), and insialuria.hIBM and DMRV. Patients with hIBM and DMRV (alsoknown as Nonaka myopathy) have biallelic missensemutations in the epimerase and/or kinase domains of theGNE gene. hIBM and DMRV are autosomal recessiveforms of inclusion body myopathy that typically causeprogressive, severe, noninflammatory muscle disease,leading to myopathic weakness and atrophy of the limbmuscles. The quadriceps, however, is nearly alwaysspared. The syndromes manifest in early adult life. For areview, see Darvish (176).

Laboratory findings: Histologically, the muscle fibersdegenerate and form rimmed vacuoles in hIBM andDMRV, especially in the atrophic areas (177, 178).Creatine kinase was within reference values or mod-erately increased (179, 180). In most hIBM patientsinvestigated, hypoglycosylation of �-dystroglycanwas found with the VIA4-1 and/or IIH6 monoclonalantibodies, whereas staining of the laminin �2-chainwas normal (181, 182). Arachis hypogaea peanut ag-glutinin lectin, which recognizes unsialylated core 1O-glycans, reacts strongly with sarcolemmal glyco-proteins and with �-dystroglycan in DMRV patients,

but not in controls. The sialic acid content of O-glycans was decreased to 60%–80% of control valuesin DMRV patients (175).

Sialuria. Sialuria, formerly called French-type sialuria, isan autosomal dominant inborn error of metabolism inwhich the feedback control mechanism in the biosynthesisof CMP-NeuAc is lost. This is caused by mutations incodons 263 to 266 of GNE (183), which eliminates feed-back inhibition of this enzyme by CMP-NeuAc. To date, 7patients have been reported with sialuria; presentingclinical features included mild psychomotor delay, coarseface, recurrent upper respiratory tract infections, andhepatomegaly (183).

Laboratory findings: Highly increased concentrations ofurinary free sialic acid (NeuAc) and of fibroblast freesialic acid have been found in patients with sialuria(184, 185). Hypersialylated core 1 O-glycans wereobserved, whereas the sialylation of N-glycans wasnormal (Wopereis S, unpublished data).

Diagnosis: The epimerase and kinase activities can bothbe measured in human lymphoblastoid cells (186),and total GNE activity can be measured in humanleukocytes (187). The diagnosis can be confirmed bymolecular genetic approaches (188–191).

Golgi CMP-sialic acid transporter deficiencyThe SLC35A1 gene encodes the Golgi CMP-NeuAc trans-porter, which is responsible for the transport of cytoplas-mic CMP-NeuAc into the Golgi. The SLC35A1 gene islikely to be ubiquitously expressed. Recently, a patientclassified as having CDG-IIf was found to have mutationsin the SLC35A1 gene.CDG-IIf. The patient presented initially with spontane-ous massive bleeding of the posterior chamber of the righteye and cutaneous hemorrhages. The clinical phenotypeworsened to more severe hemorrhages, respiratory distresssyndrome, and opportunistic infections. Because of graft-vs-host disease after bone marrow transplantation, the patientdied at the age of 37 months (49).

Laboratory findings: The patient had macrothrombocyto-penia, neutropenia, and complete lack of sialyl Lex

antigen on polymorphonuclear cells. Severe throm-bocytopenia and abnormalities of the megakaryocytemorphology were found, including small mononu-clear or hyposegmented megakaryocytes, vacuolatedcells, and abnormal fragmentation of megakaryocytecytoplasm into large platelet masses. Coagulationand the enzyme activities of �1,3-fucosyl- and �2,3-sialyltransferase were all normal (49 ). Unsialylatedcore 1 O-glycans were detected with peanut aggluti-nin lectin staining. Plasma N-glycosylated proteinsappeared to have a normal sialylation pattern appar-ent from the normal serum transferrin isoform profile(49 ).


Diagnosis: The activity of the Golgi CMP-NeuAc trans-porter can be determined by cloning of the humancDNA alleles into a recombinant adeno-associatedvirus for gene complementation of lec2-deficientcells. Lec2-deficient cells are commercially availableCHO cells that have a deficient Golgi CMP-NeuActransporter (49 ). CMP-NeuAc transporter activitycan also be determined in human fibroblasts (74 ). Thediagnosis can be confirmed by mutation analysis ofthe SLC35A1 gene (49 ).

congenital disorders of glycosylationaffecting the biosynthesis of n- ando-glycansDeficiencies in fucosyltransferase 1 and 2The FUT1 and FUT2 genes encode fucosyltransferase 1(FUT1) and 2 (FUT2), respectively (EC 2.4.1.69). FUT1 and-2 catalyze the addition of Fuc to a Gal residue in an �1,2linkage, which is also known as the H determinant and isthe essential precursor for the A and B antigens. The A, B,and H blood group determinants are linked to the N-acetyllactosamine unit or to Gal�1–3GlcNAc structures ofN- and mucin-type O-glycans, found on erythrocytes oron secreted proteins in saliva and other secretions. It isthought that FUT1 and FUT2 are expressed in a tissue-specific manner, with the expression is restricted to cellsof mesodermal or endodermal origin. Human FUT2 isexpressed in the small intestine, colon, and lung, whereaswhich tissues express human FUT1 is unknown atpresent. Individuals with the Bombay, para-Bombay, andnon-secretor blood groups have mutations in the FUT1 orFUT2 gene.Bombay, para-Bombay, and non-secretor blood groups.The para-Bombay blood group is caused by a deficientFUT1 (191), whereas the non-secretor blood group iscaused by a deficient FUT2 (192). The Bombay bloodgroup can be caused by either a deficient FUT1 or adeficient FUT2 (193). Inheritance of the 3 conditions isautosomal recessive. Individuals with deficiencies in oneof the FUT enzymes have no clinical phenotype.

Laboratory findings: Individuals with the Bombay bloodgroup lack the H determinant in all tissues, andindividuals with the para-Bombay blood group syn-thesize H determinants in soluble form but not onerythrocytes. Individuals with the non-secretor bloodgroup lack the H determinants in soluble form, butstill have their erythrocyte antigens.

Diagnosis: The diagnosis of FUT1 or FUT2 deficiencycan be confirmed by mutation analysis of the genes(191–193).

Galactoside 3(4)-fucosyltransferase deficiencyThe FUT3 gene encodes a galactoside 3(4)-fucosyltrans-ferase (FUT3; EC 2.4.1.65). FUT3 can use both type 1 andtype 2 carbohydrate chains as substrate, producing eitheran �1,3 or �1,4 linkage of Fuc to the Gal residue. Type 1

Lewis determinants (Lea and Leb) are linked to Gal�1–3GlcNAc structures of N- and mucin-type O-glycans,whereas the type 2 Lewis determinants (Lex, sLex, andLey) are linked to the N-acetyllactosamine unit of N- andmucin-type O-glycans. FUT3 is highly expressed in thestomach, colon, small intestine, lung, and kidney and to alesser extent in the salivary glands, bladder, uterus, andliver. Individuals with a Lewis-negative blood group havemutations in the FUT3 gene.Lewis-null (Lea�/b�) blood group. Individuals with theLewis-null blood group have no clinical phenotype (194).Inheritance is autosomal recessive.

Laboratory findings: Persons with a Lewis-negativeblood group do not have the type 1 Lewis determi-nants. Because other fucosyltransferases (FUT4,FUT6, FUT7, or FUT9) catalyze the Fuc �1,3 and �1,4linkages to Gal residues of type 2 glycans, the Lex,sLex, and Ley determinants are expressed in personswith mutated FUT3.

Diagnosis: The diagnosis of FUT3 deficiency can beconfirmed by mutation analysis of the gene (194).

Golgi GDP fucose transporter deficiencyThe FUCT1 gene encodes a FUCT. The FUCT is likely tobe ubiquitous and has a strict Golgi localization. Fuc ispresent in both N-linked and mucin-type O-linked glyco-proteins, mainly as a constituent of N-glycans, the anti-genic determinants Lea, Leb, Lex, sLex, Ley, and the bloodgroup determinants A, B, and H linked to the N-acetyllac-tosamine unit. Additionally, Fuc is present in O-fucosylglycans. Patients with CDG-IIc, formerly called leukocyteadhesion deficiency type II, have mutations in the FUCT1gene.CDG-IIc. CDG-IIc is a rare autosomal recessive syndromecharacterized by recurrent infections, typical dysmorphicfeatures, and severe growth and psychomotor retarda-tion. In some cases, CDG-IIc can be treated with Fuc,depending on the nature of the mutation (195).

Laboratory findings: Patients with CDG-IIc have neutro-philia, the Bombay blood group, the Lewis-negativeblood group, and a lack of sLex on the polymorpho-nuclear cells. Cell binding to E- and P-selectin isseverely impaired (196). Mutations in FUCT lead to alack of Fuc residues in N- and mucin-type O-glycans,whereas the biosynthesis of O-fucosyl glycans isnormal (197). This is explained by the fact thatprotein O-fucosyltransferase 1, responsible for O-fucosylation of proteins, is localized in the ER (198).Serum of patients with CDG-IIc gives normal resultswith transferrin isoelectric focusing (IEF).

Diagnosis: The activity of Golgi FUCT can be deter-mined in human fibroblasts (199). The diagnosis canbe confirmed at the molecular genetic level in thecorresponding gene (44, 48).


B4GalT1 deficiencyThe B4GALT1 gene encodes B4GalT1 (EC 2.4.1.38), whichcatalyzes the binding of UDP-Gal in a �1,4 linkage toGlcNAc residues. B4GalT1 is ubiquitous, but it is foundonly at very low concentrations in the fetal and adultbrain. B4GalT1 is involved in the formation of the N-acetyllactosamine unit. CDG-IId is caused by mutations inthe B4GALT1 gene.CDG-IId. To date, only 1 patient with CDG-IId has beendescribed, and in that patient, only the N-glycans werestudied (200). However, in B4GALT1-knockout mice, bothN- and O-glycans from erythrocyte membrane glycopro-teins were found to have abnormal structures (201). Itmay be anticipated that O-glycan biosynthesis is alsoimpaired in patients with CDG-IId. The child with CDG-IId has mental retardation, macrocephaly attributable to aDandy–Walker malformation with progressive hydro-cephalus, myopathy, and blood clotting defects (202). Theinheritance is autosomal recessive.

Laboratory findings: The patient has increased serumcreatine kinase, prolonged activated partial pro-thrombin time, and an abnormal serum transferrinpattern by IEF (200). Apolipoprotein C-III (apoC-III)IEF results are normal in this patient, which is asexpected because B4GalT1 is not involved in thebiosynthesis of core 1 O-glycans (203, 204).

Diagnosis: B4GalT1 activity can be determined in hu-man fibroblasts (200). The diagnosis can be con-firmed by mutation analysis of the correspondinggene (200).

COG7 deficiencyThe COG7 gene encodes subunit 7 of the COG complex. Itis thought that the COG complex has a role in theregulation, compartmentalization, transport, and activityof several Golgi enzymes. It is not known in which humantissues COG7 is located, but it is likely to be a ubiquitousprotein. CDG-IIe is caused by mutations in the COG7gene.CDG-IIe. To date, only 2 siblings have been describedwith perinatal asphyxia and dysmorphia, including low-set dysplastic ears, micrognathia, a short neck, and loose,wrinkled skin. Generalized hypotonia, hepatosplenomeg-aly, and progressive jaundice developed shortly afterbirth. Both siblings developed severe epilepsy and diedfrom recurrent infections and cardiac insufficiency withinthe first 10 weeks of life (74 ).

Laboratory findings: Multiple lysosomal enzymes wereincreased and coagulation factors were decreased inthe serum of the 2 CDG-IIe patients. Transferrin andapoC-III both showed abnormal IEF patterns, andpeanut agglutinin lectin staining was increased in thefibroblasts. Increased amounts of CMP-NeuAc weredetected, whereas total amounts of serum NeuAcwere decreased. The transport rates of the CMP-NeuAc and UDP-Gal NSTs were reduced to 30% of

reference values. The activities of the core 1 galacto-syltransferase and the �2,3-sialyltransferase acting onmucin-type O-glycans were decreased. Peripheralblood indices, serum electrolytes, urea, and creati-nine were within reference values, whereas the liverenzymes and bilirubin were increased in serum.Metabolic investigations of urine from these patientsshowed increased amounts of galactitol, Gal, and Tyrmetabolites, but no succinylacetone in one of thepatients. Urinary organic acids, oligosaccharides, andmucopolysaccharides were within reference values(74, 205).

Diagnosis: The COG subunits can be identified in hu-man fibroblasts with rabbit polyclonal antibodiesagainst the different COG subunits (74, 76). Thediagnosis can be confirmed at the molecular geneticlevel in the corresponding gene (74 ).

GTP-binding protein (SAR1b) deficiencyThe SARA2 gene encodes the GTP-binding protein SAR1b(SAR1b-GTPase; EC 3.6.5.2). This enzyme is involved inthe ER-to-Golgi transport of proteins, where it has afunction in protein cargo selection and in the assembly ofthe coat of COPII vesicles. Sar1b is present in manytissues, including the small intestine, liver, muscle, andbrain. Mutations in the SARA2 gene lead to chylomicronretention disease (CMRD), Anderson disease, and CMRDwith Marinesco–Sjogren syndrome. CMRD, Andersondisease, and CMRD with Marinesco–Sjogren syndromeare all autosomal recessive disorders.CMRD and Anderson disease. Phenotypically, CMRDand Anderson disease present as a malabsorption syn-drome with severe diarrhea with steatorrhea, failure tothrive, and growth retardation. Mild neurologic distur-bances occur, including mental deficiency, loss of deeptendon reflexes, decreased vibratory sensation, axonalneuropathy, mild deficits of color perception, nystagmus,and action tremor. There is little acanthocytosis (206, 207).The distinction between CMRD and Anderson disease hasbeen made on the basis of the apparent differencesbetween the partitioning of chylomicrons and of lipiddroplets between membrane-bound and cytosolic com-partments.CMRD with Marinesco–Sjogren syndrome. Two broth-ers have been diagnosed with the Marinesco–Sjogrensyndrome combined with CMRD. Marinesco–Sjogrensyndrome is a rare form of cerebellar ataxia associatedwith congenital cataracts, mental deficiency, brisk tendonreflexes, skeletal anomalies, and cerebellar atrophy (208).

Laboratory findings: Apoprotein B is absent in the intes-tine and liver of patients with CMRD. Chylomicronscannot be synthesized, and VLDL and LDL areundetectable in the plasma. Patients have low bloodcholesterol and a deficiency in fat-soluble vitamins(209). [14C]Mannose incorporation is evidently de-creased in the total protein fraction of chylomicrons


in CMRD patients compared with controls, pointingtoward deficient N-glycan biosynthesis (210). Unfor-tunately, the effect of Sar1b-GTPase deficiency onO-glycosylation has not been studied, but it is likelyto be abnormal as well.

Diagnosis: The diagnosis of this disorder can be con-firmed by mutation analysis of the SARA2 gene (211).

clinical summary of congenital defects ino-glycosylationThe first individuals who were found to have a defect inthe biosynthesis of O-linked glycans were diagnosed withknown clinical syndromes, such as multiple exostoses, theprogeroid form of Ehlers–Danlos syndrome, and WWS. Inthese patients, the gene defect was discovered earlier thanthe underlying abnormality in the biochemical pathway.In most of these syndromes, there is genetic heterogene-ity, and not all forms are familial. For example, only 10%of the patients with multiple exostoses have the heredi-tary form of the syndrome, and only 20% of the patientswith WWS have a defect in the POMT1 gene.

Defects in the biosynthesis of protein-linked O-glycanslead to a highly heterogeneous group of diseases. Themajority of patients with “classical CDG” have a defect inN-glycan biosynthesis. They have common symptomssuch as muscle hypotonia, central nervous system abnor-malities, growth delay, feeding problems, coagulationdefects, and liver disease, and frequently show specificsigns such as abnormal fat distribution and invertednipples, which help with the early clinical diagnosis. Incontrast, patients with O-glycosylation disorders com-monly have involvement of only one organ or one organsystem and do not have the general symptoms that aresuggestive for an inborn error of metabolism. For exam-ple, patients with a defect in the biosynthesis of GAGsoften have cartilage problems leading to skeletal malfor-mations, whereas patients with a defect in the biosynthe-sis of O-mannosyl glycans present with abnormalities inthe musculo-cerebral system. Most of the disorders ofO-glycan biosynthesis seem to have very specific tissueexpression, whereas N-glycans are expressed ubiqui-tously. Another remarkable difference between N- andO-glycan deficiencies is that N-glycan deficiencies gener-ally have recessive inheritance, whereas in some of theO-glycan biosynthesis diseases, such as sialuria and HME,inheritance is autosomal dominant.

Mucin-type O-glycans are more or less an exceptionamong the O-glycans because they are expressed ubiqui-tously. Intriguingly, GalNTs, which are responsible forthe attachment of the first GalNAc residue to the protein,are tissue specific. This becomes obvious in patients withFTC (GalNT3 deficiency), who present only with massivecalcium deposits in the skin and subcutaneous tissues.These are the tissues in which the activity of this specifictransferase is particularly high. Most of the other trans-ferases involved in the biosynthesis of mucin-type O-glycans, such as galactosyl- and sialyltransferases, have a

broad tissue distribution. It is therefore to be expected thatdefects in mucin-type O-glycosylation that are not local-ized in one of the tissue-specific GalNTs will produce aphenotype in which more than one organ/organ system isinvolved.

Patients with combined defects in protein N- andO-glycosylation often have a phenotype that is a mixtureof the features of inborn errors in combination withcongenital malformations. For example, patients diag-nosed with CDG-IIe (COG7 deficiency) have a phenotypewith central nervous system involvement, hypotonia, andhepatopathy combined with severe congenital heart mal-formations, limb malformations, and skin abnormalities.This new group of metabolic diseases, presenting withcombined defects in N- and mucin-type O-glycosylation,is growing continuously. In most cases described to date,the genetic background is not yet known. Wopereis et al.(204) showed in a recent investigation that 75% of the 12CDG-IIx patients examined also had abnormal biosynthe-sis of core 1 O-glycans. Clinically, these patients presentedwith widely variable clinical features ranging from apatient with only hepatic dysfunction to patients having aunique phenotype with congenital cutis laxa and congen-ital brain malformations in association with skeletalanomalies (204).

In summary, the phenotypes of patients with a congen-ital defect in O-glycosylation are a continuum. This rangesfrom patients who have a defect in the biosynthesis ofO-glycans affecting only a few proteins and thereforehave only one or two tissues involved (for example,patients with a defect in O-mannosyl glycan biosynthe-sis), to patients who have a defect in the biosynthesis ofO-glycans disturbing many proteins and thus have morethan one organ/organ system involved (probably defectsin the biosynthesis of mucin-type O-glycans not localizedin a GalNT), to patients who have combined defects in thebiosynthesis of N- and O-glycans, who have a typicalmultisystem disease (for example patients with CDG-IIe).

Screening Methods for Unraveling Defects inO-Glycosylation Biosynthesis

Most of the congenital defects in the biosynthesis ofO-glycans have been found by genetic approaches. O-Glycans are very complex and heterogeneous structuresbecause of their variable composition, linkage, andbranching. In view of the estimated number of genesinvolved in O-glycosylation processes, it is likely that thecurrently described congenital disorders of O-glycosyla-tion form only a small part of a much larger group ofdefects. This area of inborn errors of metabolism stillneeds further research, beginning with the developmentof screening techniques to identify defects in the biosyn-thesis of the various types of O-glycans.

N-Linked glycans have a common protein–glycan link-age and a common biosynthetic pathway that divergesonly in the late stages. There thus is limited variability inthe design of plasma protein N-glycans. This explains the


success of transferrin IEF as a screening method to iden-tify defects in N-glycan biosynthesis. It will, however, beimpossible to screen for all defects in O-glycan biosynthe-sis with just one assay. Recently, a first approach to screenfor defects in the biosynthesis of the abundant mucin-typecore 1 O-glycan has been developed. An IEF assay ofapoC-III was used for this purpose (Fig. 4) (203). ApoC-IIIis a plasma protein with a single core 1 O-glycan atposition Thr94. Three isoforms of apoC-III (apoC-III0,apoC-III1, and apoC-III2) can be distinguished. They differin the number of NeuAc residues attached to the O-glycancore. Because NeuAc has a negative charge, it is possibleto separate the 3 isoforms by IEF. In normal humanplasma, apoC-III1 is the most abundant form, accountingfor �50%, followed by apoC-III2 with �45%, whereasapoC-III0 accounts for �5% of total apoC-III. The ratio ofthe 3 isoforms in normal human plasma varies with age asthe degree of apoC-III sialylation decreases. When thebiosynthesis of core 1 O-glycans is disturbed, an abnormalapoC-III isoform ratio is found (203). In combination withtransferrin IEF, this technique has been useful for thedetection of combined defects in N- and O-glycosylation(204, 212). In addition, apoC-III IEF profiling should becapable of identifying most diseases in which mucin-typecore 1 O-glycan but not N-glycan biosynthesis is affected.Familial tumoral calcinosis, the only genetically defineddefect in the biosynthesis of mucin-type O-glycans atpresent, may turn out to be an exception. ApoC-III IEFprofiling has not yet been performed in this disease. It isexpected to give a normal result because GalNT3, thedefective enzyme in FTC, is not expressed in liver tissue,where apoC-III is synthesized (94 ).

For the screening of defects in O-mannosyl glycanbiosynthesis, immunohistochemical staining of �-dystro-glycan with the monoclonal antibodies VIA4-1 and IIH6 isused on muscle biopsy specimens from patients. VIA4-1and IIH6 recognize a carbohydrate epitope on �-dystro-glycan. At present, �-dystroglycan is the only knownsubstrate for this type of glycosylation in mammals (139).Some pitfalls of this technique are that it is not clear whichcarbohydrate epitope(s) is recognized by the antibodiesand that batches of the same antibody differ in quality. In

the report by Huizing et al. (181), for example, batches ofVIA4-1 and IIH6 antibodies were found to give variableresults. It is known that �-dystroglycan carries O-manno-syl glycans as well as mucin-type O-glycans and N-glycans (137, 138). Thus, abnormal staining of �-dystro-glycan may occur in defects of N- or mucin-type O-glycanbiosynthesis. For future diagnostics, the challenge is tofind a secreted protein with 1 or 2 O-mannosyl glycansattached to it. This would allow the development of a lessinvasive screening technique using plasma samples todetect defects in O-mannosyl glycan biosynthesis.

In general, it would be advantageous to have a tech-nique that can give an overview of all O-glycans presentin a sample. The development of such a holistic approachis hampered by the fact that it is difficult to remove allO-glycans from their protein backbones. A general en-doglycosidase for release of all O-glycans remains to bediscovered. Alternatively, a chemical cleavage method,such as hydrazinolysis or (non)reductive �-eliminationmay be useful. O-Glycan profiles have been published forthe human glycoproteins glycophorin A, serum and se-cretory IgA, and neutrophil gelatinase B (213). A futurechallenge will be to test such methods on human bloodsamples, isolated cells, or biopsy materials. A first studyon profiling of total serum O-glycans described alter-ations in the glycans of patients with sialuria (Wopereis S,unpublished data).

Similarly, no screening method is available to identifydefects in GAG biosynthesis. Substantial amounts ofGAGs occur physiologically in urine. These GAGs derivemainly from limited proteolysis of proteoglycans from theglomerulus, the renal tubule, and the urinary tract. Theyare released into the extracellular environment withoutpassing the lysosomes and end up in urine having thesame sugar structure as in vivo. The largest proportion ofurinary GAG is chondroitin sulfate (62%–77%), whereasheparan sulfate accounts for �25% and dermatan sulfatefor only �5% (214). The mucopolysaccharidoses, a groupof inborn errors in GAG catabolism, can be diagnosed bymeasuring increased urinary GAGs. Reliable quantitativemethods for urinary GAGs, such as the dimethyl-methyl-ene blue assay, are available (215). Theoretically, patientswith defects in the GAG biosynthesis pathway would beexpected to have decreased GAG excretion in the urine.As assays for urinary GAG measurement have beendeveloped for diagnosing mucopolysaccharidoses, mostpublished studies have concentrated on the upper refer-ence limit. Most laboratories using this assay therefore donot have a lower reference limit and would disregard anyabnormally low value. Another limitation of this ap-proach to finding GAG biosynthesis disorders is the agedependency of urinary GAG excretion. Patients olderthan 15 years of age excrete only limited amounts ofGAGs in the urine. The lower reference limit would thusbe close to zero. This approach would therefore allow theidentification of only patients younger than 15 years withGAG biosynthesis defects.

Fig. 4. plasma apoC-III isoforms (apoC-III0, apoC-III1 and apoC-III2) andtheir mucin-type core 1 O-glycan structures after IEF.Lane 1, control; lanes 2 and 3, examples of patients with an abnormalbiosynthesis in mucin-type core 1 O-glycans.


Because chondroitin sulfate is the main GAG constitu-ent in the urine, defects in chondroitin sulfate biosynthe-sis may be found by measuring urinary GAGs. It mayrequire dedicated methods to measure urinary GAGsubspecies to detect defects of heparan sulfate and der-matan sulfate biosynthesis. New ways may be found bystudying the GAG composition in tissue samples. Thismay be accomplished by releasing the GAGs from theproteoglycans or by applying mass spectrometric tech-niques to peptide digests.

We thank Jack Fransen for assistance in the sections on theGolgi and Kristopher Clark for improving the English ofthis review. This work was supported by the EuropeanCommission [contract QLG-CT2000-0047 (Euroglycan)and contract 512131 (Euroglycanet)].

References1. Venter JC, Adams MD, Myers EW, Li PW, Mural RJ, Sutton GG, et

al. The sequence of the human genome. Science 2001;291:1304–51.

2. The C. elegans Sequencing Consortium. Genome sequence ofthe nematode C. elegans: a platform for investigating biology.Science 1998;282:2012–8.

3. Lowe JB, Marth JD. A genetic approach to mammalian glycanfunction. Annu Rev Biochem 2003;72:643–91.

4. Apweiler R, Hermjakob H, Sharon N. On the frequency of proteinglycosylation, as deduced from analysis of the SWISS-PROTdatabase. Biochim Biophys Acta 1999;1473:4–8.

5. Hofsteenge J, Muller DR, de Beer T, Loffler A, Richter WJ,Vliegenthart JF. New type of linkage between a carbohydrate anda protein: C-glycosylation of a specific tryptophan residue inhuman RNase Us. Biochemistry 1994;33:13524–30.

6. Dove A. The bittersweet promise of glycobiology. Nat Biotechnol2001;19:913–7.

7. Brockhausen I. Pathways of O-glycan biosynthesis in cancercells. Biochim Biophys Acta 1999;1473:67–95.

8. Van den Steen P, Rudd PM, Dwek RA, Opdenakker G. Conceptsand principles of O-linked glycosylation. Crit Rev Biochem MolBiol 1998;33:151–208.

9. Varki A. Diversity in the sialic acids. Glycobiology 1992;2:25–40.10. Sugahara K, Kitagawa H. Recent advances in the study of the

biosynthesis and functions of sulfated glycosaminoglycans. CurrOpin Struct Biol 2000;10:518–27.

11. Gallagher JT. Heparan sulfate: growth control with a restrictedsequence menu. J Clin Invest 2001;108:357–61.

12. Wells L, Hart GW. O-GlcNAc turns twenty: functional implicationsfor post-translational modification of nuclear and cytosolic pro-teins with a sugar. FEBS Lett 2003;546:154–8.

13. Love DC, Kochan J, Cathey RL, Shin SH, Hanover JA. Mitochon-drial and nucleocytoplasmic targeting of O-linked GlcNAc trans-ferase. J Cell Sci 2003;116:647–54.

14. Pinnell SR, Fox R, Krane SM. Human collagens: differences inglycosylated hydroxylysines in skin and bone. Biochim BiophysActa 1971;229:119–22.

15. Kivirikko KI, Myllyla R. Post-translational enzymes in the biosyn-thesis of collagen: intracellular enzymes. Methods Enzymol1982;82(Pt A):245–304.

16. Endo T. O-Mannosyl glycans in mammals. Biochim Biophys Acta1999;1473:237–46.

17. Smalheiser NR, Haslam SM, Sutton Smith M, Morris HR, Dell A.

Structural analysis of sequences O-linked to mannose reveals anovel Lewis X structure in cranin (dystroglycan) purified fromsheep brain. J Biol Chem 1998;273:23698–703.

18. Endo T. Structure, function, and pathology of O-mannosyl gly-cans. Glycoconj J 2004;21:3–7.

19. Yuen CT, Chai W, Loveless RW, Lawson AM, Margolis RU, Feizi T.Brain contains HNK-1 immunoreactive O-glycans of the sulfoglu-curonyl lactosamine series that terminate in 2-linked or 2,6-linked hexose (mannose). J Biol Chem 1997;272:8924–31.

20. Inamori K, Endo T, Gu J, Matsuo I, Ito Y, Fujii S, et al.N-Acetylglucosaminyltransferase IX acts on the GlcNAc � 1,2-Man � 1-Ser/Thr moiety, forming a 2,6-branched structure inbrain O-mannosyl glycan. J Biol Chem 2004;279:2337–40.

21. Shao L, Luo Y, Moloney DJ, Haltiwanger R. O-Glycosylation of EGFrepeats: identification and initial characterization of a UDP-glucose: protein O-glucosyltransferase. Glycobiology 2002;12:763–70.

22. Shao L, Haltiwanger RS. O-Fucose modifications of epidermalgrowth factor-like repeats and thrombospondin type 1 repeats:unusual modifications in unusual places. Cell Mol Life Sci2003;60:241–50.

23. Nishimura H, Kawabata S, Kisiel W, Hase S, Ikenaka T, Takao T,et al. Identification of a disaccharide (Xyl-Glc) and a trisaccharide(Xyl2-Glc) O-glycosidically linked to a serine residue in the firstepidermal growth factor-like domain of human factors VII and IXand protein Z and bovine protein Z. J Biol Chem 1989;264:20320–5.

24. Hofsteenge J, Huwiler KG, Macek B, Hess D, Lawler J, MosherDF, et al. C-Mannosylation and O-fucosylation of the throm-bospondin type 1 module. J Biol Chem 2001;276:6485–98.

25. Julenius K, Molgaard A, Gupta R, Brunak S. Prediction, conser-vation analysis, and structural characterization of mammalianmucin-type O-glycosylation sites. Glycobiology 2005;15:153–64.

26. Matlack KE, Mothes W, Rapoport TA. Protein translocation:tunnel vision. Cell 1998;92:381–90.

27. Rapoport TA, Jungnickel B, Kutay U. Protein transport across theeukaryotic endoplasmic reticulum and bacterial inner mem-branes. Annu Rev Biochem 1996;65:271–303.

28. Spiro RG. Protein glycosylation: nature, distribution, enzymaticformation, and disease implications of glycopeptide bonds.Glycobiology 2002;12:43R–56R.

29. Rottger S, White J, Wandall HH, Olivo JC, Stark A, Bennett EP, etal. Localization of three human polypeptide GalNAc-transferasesin HeLa cells suggests initiation of O-linked glycosylationthroughout the Golgi apparatus. J Cell Sci 1998;111:45–60.

30. Vertel BM, Walters LM, Flay N, Kearns AE, Schwartz NB. Xylosy-lation is an endoplasmic reticulum to Golgi event. J Biol Chem1993;268:11105–12.

31. Peters BP, Krzesicki RF, Perini F, Ruddon RW. O-Glycosylation ofthe �-subunit does not limit the assembly of chorionic gonado-tropin �� dimer in human malignant and nonmalignant tropho-blast cells. Endocrinology 1989;124:1602–12.

32. Kean EL. Nuclear cytidine 5�-monophosphosialic acid syn-thetase. J Biol Chem 1970;245:2301–8.

33. Kornfeld S, Kornfeld R, Neufeld EF, O’Brien PJ. The feedbackcontrol of sugar nucleotide biosynthesis in liver. Proc Natl AcadSci U S A 1964;52:371–9.

34. Lucka L, Krause M, Danker K, Reutter W, Horstkorte R. Primarystructure and expression analysis of human UDP-N-acetyl-glu-cosamine-2-epimerase/N-acetylmannosamine kinase, the bi-functional enzyme in neuraminic acid biosynthesis. FEBS Lett1999;454:341–4.

35. Schenk B, Fernandez F, Waechter CJ. The ins(ide) and out(side)of dolichyl phosphate biosynthesis and recycling in the endoplas-mic reticulum. Glycobiology 2001;11:61R–70R.


36. Kim S, Westphal V, Srikrishna G, Mehta DP, Peterson S, FilianoJ, et al. Dolichol phosphate mannose synthase (DPM1) muta-tions define congenital disorder of glycosylation Ie (CDG-Ie).J Clin Invest 2000;105:191–8.

37. Schenk B, Imbach T, Frank CG, Grubenmann CE, Raymond GV,Hurvitz H, et al. MPDU1 mutations underlie a novel humancongenital disorder of glycosylation, designated type If. J ClinInvest 2001;108:1687–95.

38. Hirschberg CB, Robbins PW, Abeijon C. Transporters of nucleo-tide sugars, ATP, and nucleotide sulfate in the endoplasmicreticulum and Golgi apparatus. Annu Rev Biochem 1998;67:49–69.

39. Segawa H, Kawakita M, Ishida N. Human and Drosophila UDP-galactose transporters transport UDP-N-acetylgalactosamine inaddition to UDP-galactose. Eur J Biochem 2002;269:128–38.

40. Muraoka M, Kawakita M, Ishida N. Molecular characterization ofhuman UDP-glucuronic acid/UDP-N-acetylgalactosamine trans-porter, a novel nucleotide sugar transporter with dual substratespecificity. FEBS Lett 2001;495:87–93.

41. Ishida N, Kawakita M. Molecular physiology and pathology of thenucleotide sugar transporter family (SLC35). Pflugers Arch 2004;447:768–75.

42. Ashikov A, Routier F, Fuhlrott J, Helmus Y, Wild M, GerardySchahn R, et al. The human solute carrier gene SLC35B4encodes a bifunctional nucleotide sugar transporter with speci-ficity for UDP-xylose and UDP-N-acetylglucosamine. J Biol Chem2005;280:27230–5.

43. Eckhardt M, Muhlenhoff M, Bethe A, Gerardy Schahn R. Expres-sion cloning of the Golgi CMP-sialic acid transporter. Proc NatlAcad Sci U S A 1996;93:7572–6.

44. Luhn K, Wild MK, Eckhardt M, Gerardy Schahn R, Vestweber D.The gene defective in leukocyte adhesion deficiency II encodes aputative GDP-fucose transporter. Nat Genet 2001;28:69–72.

45. Guillen E, Abeijon C, Hirschberg CB. Mammalian Golgi apparatusUDP-N-acetylglucosamine transporter: molecular cloning by phe-notypic correction of a yeast mutant. Proc Natl Acad Sci U S A1998;95:7888–92.

46. Sprong H, Degroote S, Nilsson T, Kawakita M, Ishida N, van derSluijs P, et al. Association of the Golgi UDP-galactose transporterwith UDP-galactose:ceramide galactosyltransferase allows UDP-galactose import in the endoplasmic reticulum. Mol Biol Cell2003;14:3482–93.

47. Kabuss R, Ashikov A, Oelmann S, Gerardy Schahn R, Bakker H.Endoplasmic reticulum retention of the large splice variant of theUDP-galactose transporter is caused by a dilysine motif. Glyco-biology 2005;15:905–11.

48. Lubke T, Marquardt T, Etzioni A, Hartmann E, von Figura K,Korner C. Complementation cloning identifies CDG-IIc, a newtype of congenital disorders of glycosylation, as a GDP-fucosetransporter deficiency. Nat Genet 2001;28:73–6.

49. Martinez Duncker I, Dupre T, Piller V, Piller F, Candelier JJ, TrichetC, et al. Genetic complementation reveals a novel human congen-ital disorder of glycosylation of type II, due to inactivation of theGolgi CMP-sialic acid transporter. Blood 2005;105:2671–6.

50. Abeijon C, Yanagisawa K, Mandon EC, Hausler A, Moremen K,Hirschberg CB, et al. Guanosine diphosphatase is required forprotein and sphingolipid glycosylation in the Golgi lumen ofSaccharomyces cerevisiae. J Cell Biol 1993;122:307–23.

51. Varki A. Factors controlling the glycosylation potential of the Golgiapparatus. Trends Cell Biol 1998;8:34–40.

52. Petrova P, Koca J, Imberty A. Molecular dynamics simulations ofsolvated UDP-glucose in interaction with Mg2� cations. EurJ Biochem 2001;268:5365–74.

53. Ju T, Cummings RD. A unique molecular chaperone Cosmc

required for activity of the mammalian core 1 �3-galactosyltrans-ferase. Proc Natl Acad Sci U S A 2002;99:16613–8.

54. Esko JD, Zhang L. Influence of core protein sequence onglycosaminoglycan assembly. Curr Opin Struct Biol 1996;6:663–70.

55. Dustin ML, Baranski TJ, Sampath D, Kornfeld S. A novel mu-tagenesis strategy identifies distantly spaced amino acid se-quences that are required for the phosphorylation of both theoligosaccharides of procathepsin D by N-acetylglucosamine1-phosphotransferase. J Biol Chem 1995;270:170–9.

56. Kornfeld S. Lysosomal enzyme targeting. Biochem Soc Trans1990;18:367–74.

57. Manya H, Chiba A, Yoshida A, Wang X, Chiba Y, Jigami Y, et al.Demonstration of mammalian protein O-mannosyltransferaseactivity: coexpression of POMT1 and POMT2 required for enzy-matic activity. Proc Natl Acad Sci U S A 2004;101:500–5.

58. Cheng L, Tachibana K, Iwasaki H, Kameyama A, Zhang Y, KubotaT, et al. Characterization of a novel human UDP-GalNAc trans-ferase, pp-GalNAc-T15. FEBS Lett 2004;566:17–24.

59. Ten Hagen KG, Fritz TA, Tabak LA. All in the family: the UDP-GalNAc:polypeptide N-acetylgalactosaminyltransferases. Glyco-biology 2003;13:1R–16R.

60. Habuchi O. Diversity and functions of glycosaminoglycan sulfo-transferases. Biochim Biophys Acta 2000;1474:115–27.

61. Uyama T, Kitagawa H, Tamura J, Sugahara K. Molecular cloningand expression of human chondroitin N-acetylgalactosaminyl-transferase: the key enzyme for chain initiation and elongation ofchondroitin/dermatan sulfate on the protein linkage region tet-rasaccharide shared by heparin/heparan sulfate. J Biol Chem2002;277:8841–6.

62. Uyama T, Kitagawa H, Tanaka J, Tamura J, Ogawa T, Sugahara K.Molecular cloning and expression of a second chondroitin N-acetylgalactosaminyltransferase involved in the initiation andelongation of chondroitin/dermatan sulfate. J Biol Chem 2003;278:3072–8.

63. Kitagawa H, Uyama T, Sugahara K. Molecular cloning andexpression of a human chondroitin synthase. J Biol Chem2001;276:38721–6.

64. Allan VJ, Thompson HM, McNiven MA. Motoring around the Golgi.Nat Cell Biol 2002;4:E236–42.

65. Beraud Dufour S, Balch W. A journey through the exocyticpathway. J Cell Sci 2002;115:1779–80.

66. Bonifacino JS, Glick BS. The mechanisms of vesicle budding andfusion. Cell 2004;116:153–66.

67. Giraudo CG, Maccioni HJ. Endoplasmic reticulum export ofglycosyltransferases depends on interaction of a cytoplasmicdibasic motif with Sar1. Mol Biol Cell 2003;14:3753–66.

68. Barlowe C. Signals for COPII-dependent export from the ER:what’s the ticket out? Trends Cell Biol 2003;13:295–300.

69. Altan Bonnet N, Sougrat R, Lippincott Schwartz J. Molecularbasis for Golgi maintenance and biogenesis. Curr Opin Cell Biol2004;16:364–72.

70. Young WWJ. Organization of Golgi glycosyltransferases in mem-branes: complexity via complexes. J Membr Biol 2004;198:1–13.

71. de Graffenried CL, Bertozzi CR. The roles of enzyme localizationand complex formation in glycan assembly within the Golgiapparatus. Curr Opin Cell Biol 2004;16:356–63.

72. Opat AS, van Vliet C, Gleeson PA. Trafficking and localization ofresident Golgi glycosylation enzymes. Biochimie 2001;83:763–73.

73. Nillson T, Warren G. Retention and retrieval in the endoplasmicreticulum and Golgi apparatus. Curr Opin Cell Biol 1994;6:517–21.

74. Wu X, Steet RA, Bohorov O, Bakker J, Newell J, Krieger M, et al.


Mutation of the COG complex subunit gene COG7 causes a lethalcongenital disorder. Nat Med 2004;10:518–23.

75. Loh E, Hong W. The binary interacting network of the conservedoligomeric Golgi tethering complex. J Biol Chem 2004;279:24640–8.

76. Ungar D, Oka T, Brittle EE, Vasile E, Lupashin VV, Chatterton JE,et al. Characterization of a mammalian Golgi-localized proteincomplex, COG, that is required for normal Golgi morphology andfunction. J Cell Biol 2002;157:405–15.

77. Oka T, Ungar D, Hughson FM, Krieger M. The COG and COPIcomplexes interact to control the abundance of GEARs, a subsetof Golgi integral membrane proteins. Mol Biol Cell 2004;15:2423–35.

78. Raman R, Sasisekharan V, Sasisekharan R. Structural insightsinto biological roles of protein-glycosaminoglycan interactions.Chem Biol 2005;12:267–77.

79. Varki A. Biological roles of oligosaccharides: all of the theoriesare correct. Glycobiology 1993;3:97–130.

80. Otvos LJ, Krivulka GR, Urge L, Szendrei GI, Nagy L, Xiang ZQ, etal. Comparison of the effects of amino acid substitutions and�-N- vs. �-O-glycosylation on the T-cell stimulatory activity andconformation of an epitope on the rabies virus glycoprotein.Biochim Biophys Acta 1995;1267:55–64.

81. Shogren R, Gerken TA, Jentoft N. Role of glycosylation on theconformation and chain dimensions of O-linked glycoproteins:light-scattering studies of ovine submaxillary mucin. Biochemis-try 1989;28:5525–36.

82. Rose MC, Voter WA, Sage H, Brown CF, Kaufman B. Effects ofdeglycosylation on the architecture of ovine submaxillary mucinglycoprotein. J Biol Chem 1984;259:3167–72.

83. Lagow E, DeSouza MM, Carson DD. Mammalian reproductivetract mucins. Hum Reprod Update 1999;5:280–92.

84. Reuter G, Gabius HJ. Eukaryotic glycosylation: whim of nature ormultipurpose tool? Cell Mol Life Sci 1999;55:368–422.

85. van den Steen P, Rudd P, Wormald M, Dwek R, Opdenakker G.O-Linked glycosylation in focus [Review]. Trends Glycosci Gly-cotechnol 2000;63:35–49.

86. Kodama S, Tsujimoto M, Tsuruoka N, Sugo T, Endo T, Kobata A.Role of sugar chains in the in-vitro activity of recombinant humaninterleukin 5. Eur J Biochem 1993;211:903–8.

87. Naim HY, Lentze MJ. Impact of O-glycosylation on the function ofhuman intestinal lactase-phlorizin hydrolase. Characterization ofglycoforms varying in enzyme activity and localization of O-glycoside addition. J Biol Chem 1992;267:25494–504.

88. Kneass ZT, Marchase RB. Protein O-GlcNAc modulates motility-associated signaling intermediates in neutrophils. J Biol Chem2005;280:14579–85.

89. Park PW, Reizes O, Bernfield M. Cell surface heparan sulfateproteoglycans: selective regulators of ligand-receptor encoun-ters. J Biol Chem 2000;275:29923–6.

90. Remaley AT, Ugorski M, Wu N, Litzky L, Burger SR, Moore JS, etal. Expression of human glycophorin A in wild type and glycosy-lation-deficient Chinese hamster ovary cells: role of N- andO-linked glycosylation in cell surface expression. J Biol Chem1991;266:24176–83.

91. Daughaday WH, Trivedi B, Baxter RC. Serum “big insulin-likegrowth factor II” from patients with tumor hypoglycemia lacksnormal E-domain O-linked glycosylation, a possible determinantof normal propeptide processing. Proc Natl Acad Sci U S A1993;90:5823–7.

92. Bennett EP, Hassan H, Clausen H. cDNA cloning and expressionof a novel human UDP-N-acetyl-�-D-galactosamine. PolypeptideN-acetylgalactosaminyltransferase, GalNAc-t3. J Biol Chem1996;271:17006–12.

93. Slavin RE, Wen J, Kumar D, Evans EB. Familial tumoral calcino-

sis: a clinical, histopathologic, and ultrastructural study with ananalysis of its calcifying process and pathogenesis. Am J SurgPathol 1993;17:788–802.

94. Topaz O, Shurman DL, Bergman R, Indelman M, Ratajczak P,Mizrachi M, et al. Mutations in GALNT3, encoding a proteininvolved in O-linked glycosylation, cause familial tumoral calcino-sis. Nat Genet 2004;36:579–81.

95. Martinez S. Tumoral calcinosis: 12 years later. Semin Musculo-skelet Radiol 2002;6:331–9.

96. Topaz O, Bergman R, Mandel U, Maor G, Goldberg R, Richard G,et al. Absence of intraepidermal glycosyltransferase ppGal-Nac-T3 expression in familial tumoral calcinosis. Am J Dermato-pathol 2005;27:211–5.

97. Okajima T, Fukumoto S, Furukawa K, Urano T. Molecular basisfor the progeroid variant of Ehlers-Danlos syndrome: identifica-tion and characterization of two mutations in galactosyltrans-ferase I gene. J Biol Chem 1999;274:28841–4.

98. Faiyaz Ul Haque M, Zaidi SH, Al Ali M, Al Mureikhi MS, KennedyS, et al. A novel missense mutation in the galactosyltransferase-I(B4GALT7) gene in a family exhibiting facioskeletal anomaliesand Ehlers-Danlos syndrome resembling the progeroid type. Am JMed Genet 2004;128A:39–45.

99. Quentin E, Gladen A, Roden L, Kresse H. A genetic defect in thebiosynthesis of dermatan sulfate proteoglycan: galactosyltrans-ferase I deficiency in fibroblasts from a patient with a progeroidsyndrome. Proc Natl Acad Sci U S A 1990;87:1342–6.

100. Francannet C, Cohen Tanugi A, Le Merrer M, Munnich A,Bonaventure J, Legeai-Mallet L. Genotype-phenotype correlationin hereditary multiple exostoses. J Med Genet 2001;38:430–4.

101. Wicklund CL, Pauli RM, Johnston D, Hecht JT. Natural historystudy of hereditary multiple exostoses. Am J Med Genet 1995;55:43–6.

102. Ahn J, Ludecke HJ, Lindow S, Horton WA, Lee B, Wagner MJ, etal. Cloning of the putative tumour suppressor gene for hereditarymultiple exostoses (EXT1). Nat Genet 1995;11:137–43.

103. Philippe C, Porter DE, Emerton ME, Wells DE, Simpson AH,Monaco AP. Mutation screening of the EXT1 and EXT2 genes inpatients with hereditary multiple exostoses. Am J Hum Genet1997;61:520–8.

104. Akama TO, Nishida K, Nakayama J, Watanabe H, Ozaki K,Nakamura T, et al. Macular corneal dystrophy type I and type IIare caused by distinct mutations in a new sulphotransferasegene. Nat Genet 2000;26:237–41.

105. Klintworth GK, Vogels FS. Macular corneal dystrophy: an inher-ited acid mucopolysaccharide storage disease of the cornealfibroblast. Am J Pathol 1964;45:565–86.

106. Morgan G. Macular dystrophy of the cornea. Br J Ophthalmol1966;50:57–67.

107. Thonar EJ, Meyer RF, Dennis RF, Lenz ME, Maldonado B, HassellJR, et al. Absence of normal keratan sulfate in the blood ofpatients with macular corneal dystrophy. Am J Ophthalmol1986;102:561–9.

108. Edward DP, Thonar EJ, Srinivasan M, Yue BJ, Tso MO. Maculardystrophy of the cornea: a systemic disorder of keratan sulfatemetabolism. Ophthalmology 1990;97:1194–200.

109. Garner A. Histochemistry of corneal macular dystrophy. InvestOphthalmol 1969;8:475–83.

110. Thonar EJ, Lenz ME, Klintworth GK, Caterson B, Pachman LM,Glickman P, et al. Quantification of keratan sulfate in blood as amarker of cartilage catabolism. Arthritis Rheum 1985;28:1367–76.

111. Thiele H, Sakano M, Kitagawa H, Sugahara K, Rajab A, Hohne W,et al. Loss of chondroitin 6-O-sulfotransferase-1 function resultsin severe human chondrodysplasia with progressive spinal in-volvement. Proc Natl Acad Sci U S A 2004;101:10155–60.


112. Rajab A, Kunze J, Mundlos S. Spondyloepiphyseal dysplasiaOmani type: a new recessive type of SED with progressive spinalinvolvement. Am J Med Genet 2004;126A:413–9.

113. Superti Furga A, Rossi A, Steinmann B, Gitzelmann R. A chon-drodysplasia family produced by mutations in the diastrophicdysplasia sulfate transporter gene: genotype/phenotype corre-lations. Am J Med Genet 1996;63:144–7.

114. Rossi A, Kaitila I, Wilcox WR, Rimoin DL, Steinmann B, Cetta G,et al. Proteoglycan sulfation in cartilage and cell cultures frompatients with sulfate transporter chondrodysplasias: relationshipto clinical severity and indications on the role of intracellularsulfate production. Matrix Biol 1998;17:361–9.

115. Rossi A, Superti Furga A. Mutations in the diastrophic dysplasiasulfate transporter (DTDST) gene (SLC26A2): 22 novel muta-tions, mutation review, associated skeletal phenotypes, anddiagnostic relevance. Hum Mutat 2001;17:159–71.

116. Rossi A, van der Harten HJ, Beemer FA, Kleijer WJ, Gitzelmann R,Steinmann B, et al. Phenotypic and genotypic overlap betweenatelosteogenesis type 2 and diastrophic dysplasia. Hum Genet1996;98:657–61.

117. Makitie O, Kaitila I. Growth in diastrophic dysplasia. J Pediatr1997;130:641–6.

118. Horton WA, Rimoin DL, Lachman RS, Skovby F, Hollister DW,Spranger J, et al. The phenotypic variability of diastrophicdysplasia. J Pediatr 1978;93:609–13.

119. Ballhausen D, Bonafe L, Terhal P, Unger SL, Bellus G, ClassenM, et al. Recessive multiple epiphyseal dysplasia (rMED): phe-notype delineation in eighteen homozygotes for DTDST mutationR279W. J Med Genet 2003;40:65–71.

120. Superti Furga A. A defect in the metabolic activation of sulfate ina patient with achondrogenesis type IB. Am J Hum Genet1994;55:1137–45.

121. Grayeli AB, Escoubet B, Bichara M, Julien N, Silve C, FriedlanderG, et al. Increased activity of the diastrophic dysplasia sulfatetransporter in otosclerosis and its inhibition by sodium fluoride.Otol Neurotol 2003;24:854–62.

122. Superti Furga A, Hastbacka J, Wilcox WR, Cohn DH, van derHarten HJ, Rossi A, et al. Achondrogenesis type IB is caused bymutations in the diastrophic dysplasia sulphate transportergene. Nat Genet 1996;12:100–2.

123. Hastbacka J, Superti Furga A, Wilcox WR, Rimoin DL, Cohn DH,Lander ES. Atelosteogenesis type II is caused by mutations inthe diastrophic dysplasia sulfate-transporter gene (DTDST): evi-dence for a phenotypic series involving three chondrodysplasias.Am J Hum Genet 1996;58:255–62.

124. Hastbacka J, de la Chapelle A, Mahtani MM, Clines G, ReeveDaly MP, Daly M, et al. The diastrophic dysplasia gene encodesa novel sulfate transporter: positional cloning by fine-structurelinkage disequilibrium mapping. Cell 1994;78:1073–87.

125. Superti Furga A, Neumann L, Riebel T, Eich G, Steinmann B,Spranger J, et al. Recessively inherited multiple epiphysealdysplasia with normal stature, club foot, and double layeredpatella caused by a DTDST mutation. J Med Genet 1999;36:621–4.

126. ul Haque MF, King LM, Krakow D, Cantor RM, Rusiniak ME,Swank RT, et al. Mutations in orthologous genes in humanspondyloepimetaphyseal dysplasia and the brachymorphicmouse. Nat Genet 1998;20:157–62.

127. Ahmad M, Haque MF, Ahmad W, Abbas H, Haque S, Krakow D,et al. Distinct, autosomal recessive form of spondyloepime-taphyseal dysplasia segregating in an inbred Pakistani kindred.Am J Med Genet 1998;78:468–73.

128. Xu Z, Wood TC, Adjei AA, Weinshilboum RM. Human 3�-phos-phoadenosine 5�-phosphosulfate synthetase: radiochemical en-

zymatic assay, biochemical properties, and hepatic variation.Drug Metab Dispos 2001;29:172–8.

129. Vogel W, Gish GD, Alves F, Pawson T. The discoidin domainreceptor tyrosine kinases are activated by collagen. Mol Cell1997;1:13–23.

130. Suarez KN, Romanello M, Bettica P, Moro L. Collagen type I of ratcortical and trabecular bone differs in the extent of posttransla-tional modifications. Calcif Tissue Int 1996;58:65–9.

131. Heikkinen J, Hautala T, Kivirikko KI, Myllyla R. Structure andexpression of the human lysyl hydroxylase gene (PLOD): introns9 and 16 contain Alu sequences at the sites of recombination inEhlers-Danlos syndrome type VI patients. Genomics 1994;24:464–71.

132. Yeowell HN, Walker LC. Mutations in the lysyl hydroxylase 1 genethat result in enzyme deficiency and the clinical phenotype ofEhlers-Danlos syndrome type VI. Mol Genet Metab 2000;71:212–24.

133. Krane SM, Pinnell SR, Erbe RW. Lysyl-protocollagen hydroxylasedeficiency in fibroblasts from siblings with hydroxylysine-defi-cient collagen. Proc Natl Acad Sci U S A 1972;69:2899–903.

134. Hyland J, Ala Kokko L, Royce P, Steinmann B, Kivirikko KI, MyllylaR. A homozygous stop codon in the lysyl hydroxylase gene in twosiblings with Ehlers-Danlos syndrome type VI. Nat Genet 1992;2:228–31.

135. Henry MD, Campbell KP. Dystroglycan inside and out. Curr OpinCell Biol 1999;11:602–7.

136. Winder SJ. The complexities of dystroglycan. Trends Biochem Sci2001;26:118–24.

137. Chiba A, Matsumura K, Yamada H, Inazu T, Shimizu T, KusunokiS, et al. Structures of sialylated O-linked oligosaccharides ofbovine peripheral nerve �-dystroglycan: the role of a novelO-mannosyl-type oligosaccharide in the binding of �-dystroglycanwith laminin. J Biol Chem 1997;272:2156–62.

138. Sasaki T, Yamada H, Matsumura K, Shimizu T, Kobata A, EndoT. Detection of O-mannosyl glycans in rabbit skeletal muscle�-dystroglycan. Biochim Biophys Acta 1998;1425:599–606.

139. Michele DE, Barresi R, Kanagawa M, Saito F, Cohn RD, Satz JS,et al. Post-translational disruption of dystroglycan-ligand interac-tions in congenital muscular dystrophies. Nature 2002;418:417–22.

140. van Reeuwijk J, Janssen M, van den Elzen C, Beltran Valero deBernabe D, Sabatelli P, Merlini L, et al. POMT2 mutations cause�-dystroglycan hypoglycosylation and Walker-Warburg syndrome.J Med Genet 2005;42:907–12.

141. van Reeuwijk J, Brunner HG, van Bokhoven H. Glyc-O-genetics ofWalker-Warburg syndrome. Clin Genet 2005;67:281–9.

142. Balci B, Uyanik G, Dincer P, Gross C, Willer T, Talim B, et al. Anautosomal recessive limb girdle muscular dystrophy (LGMD2)with mild mental retardation is allelic to Walker-Warburg syn-drome (WWS) caused by a mutation in the POMT1 gene.Neuromuscul Disord 2005;15:271–5.

143. Dincer P, Balci B, Yuva Y, Talim B, Brockington M, Dincel D, et al.A novel form of recessive limb girdle muscular dystrophy withmental retardation and abnormal expression of �-dystroglycan.Neuromuscul Disord 2003;13:771–8.

144. Cormand B, Pihko H, Bayes M, Valanne L, Santavuori P, Talim B,et al. Clinical and genetic distinction between Walker-Warburgsyndrome and muscle-eye-brain disease. Neurology 2001;56:1059–69.

145. Sabatelli P, Columbaro M, Mura I, Capanni C, Lattanzi G, MaraldiNM, et al. Extracellular matrix and nuclear abnormalities inskeletal muscle of a patient with Walker-Warburg syndromecaused by POMT1 mutation. Biochim Biophys Acta 2003;1638:57–62.

146. Voit T, Sewry CA, Meyer K, Hermann R, Straub V, Muntoni F, et


al. Preserved merosin M-chain (or laminin-� 2) expression inskeletal muscle distinguishes Walker-Warburg syndrome fromFukuyama muscular dystrophy and merosin-deficient congenitalmuscular dystrophy. Neuropediatrics 1995;26:148–55.

147. Wewer UM, Durkin ME, Zhang X, Laursen H, Nielsen NH, TowfighiJ, et al. Laminin �2 chain and adhalin deficiency in the skeletalmuscle of Walker-Warburg syndrome (cerebro-ocular dysplasia-muscular dystrophy). Neurology 1995;45:2099–101.

148. Beltran Valero de Bernabe D, Currier S, Steinbrecher A, Celli J,van Beusekom E, van der Zwaag B, et al. Mutations in theO-mannosyltransferase gene POMT1 give rise to the severeneuronal migration disorder Walker-Warburg syndrome. Am JHum Genet 2002;71:1033–43.

149. Diesen C, Saarinen A, Pihko H, Rosenlew C, Cormand B, DobynsWB, et al. POMGnT1 mutation and phenotypic spectrum inmuscle-eye-brain disease. J Med Genet 2004;41:e115.

150. Kano H, Kobayashi K, Herrmann R, Tachikawa M, Manya H,Nishino I, et al. Deficiency of �-dystroglycan in muscle-eye-braindisease. Biochem Biophys Res Commun 2002;291:1283–6.

151. Zhang W, Vajsar J, Cao P, Breningstall G, Diesen C, Dobyns W,et al. Enzymatic diagnostic test for muscle-eye-brain type con-genital muscular dystrophy using commercially available re-agents. Clin Biochem 2003;36:339–44.

152. Yoshida A, Kobayashi K, Manya H, Taniguchi K, Kano H, MizunoM, et al. Muscular dystrophy and neuronal migration disordercaused by mutations in a glycosyltransferase, POMGnT1. DevCell 2001;1:717–24.

153. Aravind L, Koonin EV. The fukutin protein family—predictedenzymes modifying cell-surface molecules. Curr Biol 1999;9:R836–7.

154. Fukuyama Y, Osawa M, Suzuki H. Congenital progressive mus-cular dystrophy of the Fukuyama type—clinical, genetic andpathological considerations. Brain Dev 1981;3:1–29.

155. Yoshioka M, Kuroki S. Clinical spectrum and genetic studies ofFukuyama congenital muscular dystrophy. Am J Med Genet1994;53:245–50.

156. Hino N, Kobayashi M, Shibata N, Yamamoto T, Saito K, OsawaM. Clinicopathological study on eyes from cases of Fukuyamatype congenital muscular dystrophy. Brain Dev 2001;23:97–107.

157. Toda T, Kobayashi K, Takeda S, Sasaki J, Kurahashi H, Kano H,et al. Fukuyama-type congenital muscular dystrophy (FCMD) and�-dystroglycanopathy. Congenit Anom (Kyoto) 2003;43:97–104.

158. de Bernabe DB, van Bokhoven H, van Beusekom E, Van denAkker W, Kant S, Dobyns WB, et al. A homozygous nonsensemutation in the fukutin gene causes a Walker-Warburg syndromephenotype. J Med Genet 2003;40:845–8.

159. Silan F, Yoshioka M, Kobayashi K, Simsek E, Tunc M, Alper M, etal. A new mutation of the fukutin gene in a non-Japanese patient.Ann Neurol 2003;53:392–6.

160. Matsumoto H, Hayashi YK, Kim DS, Ogawa M, Murakami T,Noguchi S, et al. Congenital muscular dystrophy with glycosyla-tion defects of �-dystroglycan in Japan. Neuromuscul Disord2005;15:342–8.

161. Zanoteli E, Rocha JC, Narumia LK, Fireman MA, Moura LS,Oliveira AS, et al. Fukuyama-type congenital muscular dystrophy:a case report in the Japanese population living in Brazil. ActaNeurol Scand 2002;106:117–21.

162. Kobayashi K, Nakahori Y, Miyake M, Matsumura K, Kondo Iida E,Nomura Y, et al. An ancient retrotransposal insertion causesFukuyama-type congenital muscular dystrophy. Nature 1998;394:388–92.

163. Brockington M, Blake DJ, Prandini P, Brown SC, Torelli S, BensonMA, et al. Mutations in the fukutin-related protein gene (FKRP)cause a form of congenital muscular dystrophy with secondary

laminin �2 deficiency and abnormal glycosylation of �-dystrogly-can. Am J Hum Genet 2001;69:1198–209.

164. de Paula F, Vieira N, Starling A, Yamamoto LU, Lima B, CassiaPavanello R, et al. Asymptomatic carriers for homozygous novelmutations in the FKRP gene: the other end of the spectrum. EurJ Hum Genet 2003;11:923–30.

165. Topaloglu H, Brockington M, Yuva Y, Talim B, Haliloglu G, BlakeD, et al. FKRP gene mutations cause congenital musculardystrophy, mental retardation, and cerebellar cysts. Neurology2003;60:988–92.

166. Longman C, Brockington M, Torelli S, Jimenez Mallebrera C,Kennedy C, Khalil N, et al. Mutations in the human LARGE genecause MDC1D, a novel form of congenital muscular dystrophywith severe mental retardation and abnormal glycosylation of�-dystroglycan. Hum Mol Genet 2003;12:2853–61.

167. Brockington M, Yuva Y, Prandini P, Brown SC, Torelli S, BensonMA, et al. Mutations in the fukutin-related protein gene (FKRP)identify limb girdle muscular dystrophy 2I as a milder allelicvariant of congenital muscular dystrophy MDC1C. Hum MolGenet 2001;10:2851–9.

168. Mercuri E, Brockington M, Straub V, Quijano Roy S, Yuva Y,Herrmann R, et al. Phenotypic spectrum associated with muta-tions in the fukutin-related protein gene. Ann Neurol 2003;53:537–42.

169. Beltran Valero de Bernabe D, Voit T, Longman C, Steinbrecher A,Straub V, Yuva Y, et al. Mutations in the FKRP gene can causemuscle-eye-brain disease and Walker-Warburg syndrome. J MedGenet 2004;41:e61.

170. Schachter H, Vajsar J, Zhang W. The role of defective glycosyla-tion in congenital muscular dystrophy. Glycoconj J 2004;20:291–300.

171. Peyrard M, Seroussi E, Sandberg Nordqvist AC, Xie YG, Han FY,Fransson I, et al. The human LARGE gene from 22q12.3-q13.1 isa new, distinct member of the glycosyltransferase gene family.Proc Natl Acad Sci U S A 1999;96:598–603.

172. Barresi R, Michele DE, Kanagawa M, Harper HA, Dovico SA, SatzJS, et al. LARGE can functionally bypass �-dystroglycan glycosyl-ation defects in distinct congenital muscular dystrophies. NatMed 2004;10:696–703.

173. Kanagawa M, Saito F, Kunz S, Yoshida Moriguchi T, Barresi R,Kobayashi YM, et al. Molecular recognition by LARGE is essentialfor expression of functional dystroglycan. Cell 2004;117:953–64.

174. Brockington M, Torelli S, Prandini P, Boito C, Dolatshad NF,Longman C, et al. Localization and functional analysis of theLARGE family of glycosyltransferases: significance for musculardystrophy. Hum Mol Genet 2005;14:657–65.

175. Tajima Y, Uyama E, Go S, Sato C, Tao N, Kotani M, et al. Distalmyopathy with rimmed vacuoles: impaired o-glycan formation inmuscular glycoproteins. Am J Pathol 2005;166:1121–30.

176. Darvish D. Magnesium may help patients with recessive hered-itary inclusion body myopathy: a pathological review. Med Hy-potheses 2003;60:94–101.

177. Askanas V, Engel WK. Sporadic inclusion-body myositis andhereditary inclusion-body myopathies: current concepts of diag-nosis and pathogenesis. Curr Opin Rheumatol 1998;10:530–42.

178. Ceuterick C, Martin JJ. Sporadic early adult-onset distal myop-athy with rimmed vacuoles: immunohistochemistry and electronmicroscopy. J Neurol Sci 1996;139:190–6.

179. Zlotogora J. Hereditary disorders among Iranian Jews. Am J MedGenet 1995;58:32–7.

180. Nonaka I, Sunohara N, Ishiura S, Satoyoshi E. Familial distalmyopathy with rimmed vacuole and lamellar (myeloid) bodyformation. J Neurol Sci 1981;51:141–55.


181. Huizing M, Rakocevic G, Sparks SE, Mamali I, Shatunov A,Goldfarb L, et al. Hypoglycosylation of �-dystroglycan in patientswith hereditary IBM due to GNE mutations. Mol Genet Metab2004;81:196–202.

182. Broccolini A, Gliubizzi C, Pavoni E, Gidaro T, Morosetti R,Sciandra F, et al. �-Dystroglycan does not play a major patho-genic role in autosomal recessive hereditary inclusion-bodymyopathy. Neuromuscul Disord 2005;15:177–84.

183. Leroy JG, Seppala R, Huizing M, Dacremont G, De Simpel H, VanCoster RN, et al. Dominant inheritance of sialuria, an inborn errorof feedback inhibition. Am J Hum Genet 2001;68:1419–27.

184. Fontaine G, Biserte G, Montreuil J, Dupont A, Farriaux JP. Lasialurie: un trouble metabolique original. Helv Paediatr Acta1968;Suppl 17:1–32.

185. Thomas GH, Reynolds LW, Miller CS. Overproduction of N-acetylneuraminic acid (sialic acid) by sialuria fibroblasts. PediatrRes 1985;19:451–5.

186. Hinderlich S, Salama I, Eisenberg I, Potikha T, Mantey LR,Yarema KJ, et al. The homozygous M712T mutation of UDP-N-acetylglucosamine 2-epimerase/N-acetylmannosamine kinaseresults in reduced enzyme activities but not in altered overallcellular sialylation in hereditary inclusion body myopathy. FEBSLett 2004;566:105–9.

187. Nishino I, Noguchi S, Murayama K, Driss A, Sugie K, Oya Y, et al.Distal myopathy with rimmed vacuoles is allelic to hereditaryinclusion body myopathy. Neurology 2002;59:1689–93.

188. Eisenberg I, Avidan N, Potikha T, Hochner H, Chen M, Olender T,et al. The UDP-N-acetylglucosamine 2-epimerase/N-acetylman-nosamine kinase gene is mutated in recessive hereditary inclu-sion body myopathy. Nat Genet 2001;29:83–7.

189. Kayashima T, Matsuo H, Satoh A, Ohta T, Yoshiura K, Matsu-moto N, et al. Nonaka myopathy is caused by mutations in theUDP-N-acetylglucosamine-2-epimerase/N-acetylmannosaminekinase gene (GNE). J Hum Genet 2002;47:77–9.

190. Seppala R, Lehto VP, Gahl WA. Mutations in the human UDP-N-acetylglucosamine 2-epimerase gene define the disease sialuriaand the allosteric site of the enzyme. Am J Hum Genet 1999;64:1563–9.

191. Kelly RJ, Ernst LK, Larsen RD, Bryant JG, Robinson JS, Lowe JB.Molecular basis for H blood group deficiency in Bombay (Oh) andpara-Bombay individuals. Proc Natl Acad Sci U S A 1994;91:5843–7.

192. Kelly RJ, Rouquier S, Giorgi D, Lennon GG, Lowe JB. Sequenceand expression of a candidate for the human Secretor bloodgroup �(1,2)fucosyltransferase gene (FUT2): homozygosity foran enzyme-inactivating nonsense mutation commonly correlateswith the non-secretor phenotype. J Biol Chem 1995;270:4640–9.

193. Koda Y, Soejima M, Johnson PH, Smart E, Kimura H. Missensemutation of FUT1 and deletion of FUT2 are responsible for IndianBombay phenotype of ABO blood group system. Biochem Bio-phys Res Commun 1997;238:21–5.

194. Koda Y, Kimura H, Mekada E. Analysis of Lewis fucosyltrans-ferase genes from the human gastric mucosa of Lewis-positiveand -negative individuals. Blood 1993;82:2915–9.

195. Etzioni A, Sturla L, Antonellis A, Green ED, Gershoni Baruch R,Berninsone PM, et al. Leukocyte adhesion deficiency (LAD) typeII/carbohydrate deficient glycoprotein (CDG) IIc founder effectand genotype/phenotype correlation. Am J Med Genet 2002;110:131–5.

196. Frydman M, Etzioni A, Eidlitz Markus T, Avidor I, Varsano I,Shechter Y, et al. Rambam-Hasharon syndrome of psychomotorretardation, short stature, defective neutrophil motility, andBombay phenotype. Am J Med Genet 1992;44:297–302.

197. Sturla L, Rampal R, Haltiwanger RS, Fruscione F, Etzioni A,Tonetti M. Differential terminal fucosylation of N-linked glycans

versus protein O-fucosylation in leukocyte adhesion deficiencytype II (CDG IIc). J Biol Chem 2003;278:26727–33.

198. Luo Y, Haltiwanger RS. O-Fucosylation of notch occurs in theendoplasmic reticulum. J Biol Chem 2005;280:11289–94.

199. Lubke T, Marquardt T, von Figura K, Korner C. A new type ofcarbohydrate-deficient glycoprotein syndrome due to a de-creased import of GDP-fucose into the Golgi. J Biol Chem1999;274:25986–9.

200. Hansske B, Thiel C, Lubke T, Hasilik M, Honing S, Peters V, et al.Deficiency of UDP-galactose:N-acetylglucosamine �-1,4-galacto-syltransferase I causes the congenital disorder of glycosylationtype IId. J Clin Invest 2002;109:725–33.

201. Kotani N, Asano M, Iwakura Y, Takasaki S. Impaired galactosy-lation of core 2 O-glycans in erythrocytes of �-1,4-galactosyl-transferase knockout mice. Biochem Biophys Res Commun1999;260:94–8.

202. Peters V, Penzien JM, Reiter G, Korner C, Hackler R, Assmann B,et al. Congenital disorder of glycosylation IId (CDG-IId)—a newentity: clinical presentation with Dandy-Walker malformation andmyopathy. Neuropediatrics 2002;33:27–32.

203. Wopereis S, Grunewald S, Morava E, Penzien JM, Briones P,Garcia Silva MT, et al. Apolipoprotein C-III isofocusing in thediagnosis of genetic defects in O-glycan biosynthesis. Clin Chem2003;49:1839–45.

204. Wopereis S, Morava E, Grunewald S, Adamowicz M, Huijben K,Lefeber DJ, et al. Patients with unsolved congenital disorders ofglycosylation type II can be subdivided in six distinct biochemicalgroups. Glycobiology 2005;15:1312–9.

205. Spaapen LJ, Bakker JA, van der Meer SB, Sijstermans HJ, SteetRA, Wevers RA, et al. Clinical and biochemical presentation ofsiblings with COG-7 deficiency, a lethal multiple O- and N-glycosylation disorder. J Inherit Metab Dis 2005;28:707–14.

206. Roy CC, Levy E, Green PH, Sniderman A, Letarte J, Buts JP et al.Malabsorption, hypocholesterolemia, and fat-filled enterocyteswith increased intestinal apoprotein B: chylomicron retentiondisease. Gastroenterology 1987;92:390–9.

207. Gauthier S, Sniderman A. Action tremor as a manifestation ofchylomicron retention disease. Ann Neurol 1983;14:591.

208. Aguglia U, Annesi G, Pasquinelli G, Spadafora P, Gambardella A,Annesi F, et al. Vitamin E deficiency due to chylomicron retentiondisease in Marinesco-Sjogren syndrome. Ann Neurol 2000;47:260–4.

209. Malloy MJ, Kane JP. Hypolipidemia. Med Clin North Am 1982;66:469–84.

210. Levy E, Marcel Y, Deckelbaum RJ, Milne R, Lepage G, SeidmanE, et al. Intestinal apoB synthesis, lipids, and lipoproteins inchylomicron retention disease. J Lipid Res 1987;28:1263–74.

211. Jones B, Jones EL, Bonney SA, Patel HN, Mensenkamp AR,Eichenbaum Voline S, et al. Mutations in a Sar1 GTPase of COPIIvesicles are associated with lipid absorption disorders. NatGenet 2003;34:29–31.

212. Wopereis S, Morava E, Grunewald S, Mills PB, Winchester BG,Clayton P, et al. A combined defect in the biosynthesis of N- andO-glycans in patients with cutis laxa and neurological involve-ment: the biochemical characteristics. Biochim Biophys Acta2005;1741:156–64.

213. Royle L, Mattu TS, Hart E, Langridge JI, Merry AH, Murphy N, etal. An analytical and structural database provides a strategy forsequencing O-glycans from microgram quantities of glycopro-teins. Anal Biochem 2002;304:70–90.

214. Lee EY, Kim SH, Whang SK, Hwang KY, Yang JO, Hong SY.Isolation, identification, and quantitation of urinary glycosamino-glycans. Am J Nephrol 2003;23:152–7.

215. de Jong JG, Wevers RA, Liebrand van Sambeek R. Measuringurinary glycosaminoglycans in the presence of protein: an im-


proved screening procedure for mucopolysaccharidoses basedon dimethylmethylene blue. Clin Chem 1992;38:803–7.

216. Kresse H, Hausser H, Schonherr E, Bittner K. Biosynthesis andinteractions of small chondroitin/dermatan sulphate proteogly-cans. Eur J Clin Chem Clin Biochem 1994;32:259–64.

217. Lin X. Functions of heparan sulfate proteoglycans in cell signal-ing during development. Development 2004;131:6009–21.

218. Hart GW. Dynamic O-linked glycosylation of nuclear and cytoskel-etal proteins. Annu Rev Biochem 1997;66:315–35.

219. Seyer JM, Kang AH. Covalent structure of collagen: amino acidsequence of cyanogen bromide peptides from the amino-termi-nal segment of type III collagen of human liver. Biochemistry1977;16:1158–64.

220. Harris RJ, Spellman MW. O-Linked fucose and other post-translational modifications unique to EGF modules. Glycobiology1993;3:219–24.

221. Kuhns W, Rutz V, Paulsen H, Matta KL, Baker MA, Barner M, etal. Processing O-glycan core 1, Gal �1–3GalNAc �-R. Specifici-ties of core 2, UDP-GlcNAc:Gal �1–3 GalNAc-R(GlcNAc to Gal-NAc) �6-N-acetylglucosaminyltransferase and CMP-sialic acid:Gal �1–3GalNAc-R �3-sialyltransferase. Glycoconj J 1993;10:381–94.

222. Klein A, Carnoy C, Wieruszeski JM, Strecker G, Strang AM, vanHalbeek H, et al. The broad diversity of neutral and sialylatedoligosaccharides derived from human salivary mucins. Biochem-istry 1992;31:6152–65.

223. Podolsky DK. Oligosaccharide structures of human colonic mu-cin. J Biol Chem 1985;260:8262–71.

224. Hounsell EF, Lawson AM, Feeney J, Gooi HC, Pickering NJ, StollMS, et al. Structural analysis of the O-glycosidically linkedcore-region oligosaccharides of human meconium glycoproteinswhich express oncofoetal antigens. Eur J Biochem 1985;148:367–77.

225. Yazawa S, Abbas SA, Madiyalakan R, Barlow JJ, Matta KL.N-Acetyl-�-D-glucosaminyltransferases related to the synthesis ofmucin-type glycoproteins in human ovarian tissue. CarbohydrRes 1986;149:241–52.

226. van Halbeek H, Strang AM, Lhermitte M, Rahmoune H, LamblinG, Roussel P. Structures of monosialyl oligosaccharides isolatedfrom the respiratory mucins of a non-secretor (O, Lea�b�)patient suffering from chronic bronchitis: characterization of anovel type of mucin carbohydrate core structure. Glycobiology1994;4:203–19.


mechanisms in protein o-glycan biosynthesis and clinical and

Documents