Glycobiology Advance Access originally published online on May 28, 2003
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Glycobiology, 2003, Vol. 13, No. 9 601-622
© 2003 Oxford University Press
Detailed glycan analysis of serum glycoproteins of patients with congenital disorders of glycosylation indicates the specific defective glycan processing step and provides an insight into pathogenesis
2 The Glycobiology Institute, Department of Biochemistry, Oxford University, South Parks Road, Oxford, OX1 3QU, U.K.
3 Medical Genetics Institute, Clinical Biology Department, Praça Pedro Nunes, 88, 4050 Porto, Portugal
4 Centre for Metabolic Disease, University of Leuven, Leuven, Belgium
5 Hospital S. João, Pediatrics Department, Porto, Portugal
6 Center for Human Genetics, University Hospital Gasthuisberg, Herestraat 49, B-3000 Leuven, Belgium
Received on January 15, 2003; revised on April 2, 2003; accepted on May 1, 2003
 |
Abstract |
|---|
 Top Abstract IntroductionResultsDiscussionMaterials and methodsReferences |
|---|
The fundamental importance of correct protein glycosylation is abundantly clear in a group of diseases known as congenital disorders of glycosylation (CDGs). In these diseases, many biological functions are compromised, giving rise to a wide range of severe clinical conditions. By performing detailed analyses of the total serum glycoproteins as well as isolated transferrin and IgG, we have directly correlated aberrant glycosylation with a faulty glycosylation processing step. In one patient the complete absence of complex type sugars was consistent with ablation of GlcNAcTase II activity. In another CDG type II patient, the identification of specific hybrid sugars suggested that the defective processing step was cell type–specific and involved the mannosidase III pathway. In each case, complementary serum proteome analyses revealed significant changes in some 31 glycoproteins, including components of the complement system. This biochemical approach to charting diseases that involve alterations in glycan processing provides a rapid indicator of the nature, severity, and cell type specificity of the suboptimal glycan processing steps; allows links to genetic mutations; indicates the expression levels of proteins; and gives insight into the pathways affected in the disease process.
Key words: CDGs / glycosylation / IgG / proteome / transferrin
|
Introduction |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and methodsReferences |
|---|
Congenital disorders of glycosylation (CDGs) are a group of genetic disorders characterized by low activity of one of the many enzymes, transporters, or other functional proteins required in the glycosylation processing pathway (Keir et al., 1999
; Yarema and Bertozzi, 2001). Diagnosis of these disorders is complicated by the fact that, although the clinical manifestations are varied and severe, the pathogenesis alone is insufficient to determine and classify the underlying cause of the disease. There is an urgent need for new strategies to address this problem in CDGs and in other pathologies for which similar symptoms may reflect different disease processes. Previously, in a group of autoimmune diseases, alterations in glycosylation have been shown to provide specific biochemical disease markers (Parekh et al., 1985; Watson et al., 1999) and also suggest underlying causes of pathogenesis. In a patient with a novel form of CDG, we demonstrate a strategy that can be used to identify the compromised enzymatic step, point to cell type–specific glycan processing pathways, allow links to genetic mutations, and indicate proteins and pathways affected in the disease process.
CDGs can be divided into two groups based on the region of the glycosylation pathway affected (Freeze, 2001). Group I disorders result from defects in the assembly of the dolichol-phosphate-linked oligosaccharide precursor (Carchon et al., 1999) in the cytosol and the endoplasmic reticulum (ER) (Jaeken et al., 1997, 1998; Imbach et al., 1999; Korner et al., 1999; Schachter and Jaeken, 1999; Dupre et al., 2000). CDG-II disorders arise from defects in the processing of the glycans in the ER and the Golgi (Schachter and Jaeken, 1999). They result in apparently normal levels of glycan site occupancy; however, the glycoproteins contain aberrant glycan structures. Genetic deficiencies in O-glycosylation have been reported and in some cases associated with congenital muscular dystrophy (Quintin et al., 1990).
As a rule, in CDGs neurological symptoms, such as psychomotor retardation and hypotonia, are common (Schachter et al., 1998; de Lonlay et al., 2001), strongly suggesting that normal complex-type glycans are essential for proper neurological development. The scarcity of individuals identified with these disorders (some 280 known as of 2001) probably indicates underdiagnosis as well as a low survival rate of affected embryos (Jaeken and Matthijs, 2001). Indeed, mice in which the Mgat1 gene was deleted died after 9 days in embryo due to defects in vascularization and neural tube closure (Ioffe and Stanley, 1994). Moreover, deletion of the Mgat2 gene, coding for GlcNAc transferase-II (GlcNAcT-II), caused frequent postnatal lethality, and 99% of the mice died within the first week after birth (Wang et al., 2000). The Mgat2 (coding for GlcNAcT-II)-null mice have similar phenotypes to CDG-IIa patients and therefore provide an insight into the disease pathogenesis. However, some of the clinical findings in mice are not observed in humans, for example, anemia, thrombocytopenia, glomerulonephritis, reticulocytosis, and kidney dysfunction. With age the mice develop signs of autoimmune disease, such as increasing amounts of autoantibodies in the serum. These phenotypic differences could be due to species-specific variations or may be associated with aging, in which case these symptoms may develop later in the CDG-IIa patients (Wang et al., 2001).
In general, CDGs are diagnosed from the analysis of a marker protein, usually serum transferrin (Tf) (Yamashita et al., 1993). A characteristic isoelectric focusing (IEF) banding pattern of normal Tf indicates that the predominant glycoprotein structure contains four sialic acid residues (S4) attached to two biantennary complex glycans on each of the two N-glycan sites (Asn413 and Asn611). A shift of this band pattern toward the cathode is indicative of a CDG and results from a decrease in sialic acid. In CDG-I, this reduction correlates with reduced glycan site occupancy and an increased proportion of Tf molecules with only one N-glycan or none at all. In CDG-II, aberrant glycan processing gives rise to a predominant S2 band. Although valuable as a diagnostic, Tf contains a very restricted set of biantennary glycans, and IEF provides only limited information about the deficient enzyme activity. In addition, neither deglycosylation nor altered glycan processing of Tf has been shown to have any effect on iron binding, recognition by reticulocytes, or iron transport into the cell (Spik et al., 1988; Hershberger et al., 1991; Mason et al., 1993) and, to date, has not been linked directly with any disease pathogenesis.
In this article we describe a strategy by which the detailed analysis of the oligosaccharides attached to Tf, IgG, and total serum glycoproteins allows the faulty step in the glycosylation processing pathway to be identified. Coupled with proteomics, which showed that a wide range of proteins were affected, this strategy provides a basis for diagnosis, investigating the cause of the aberrant glycosylation, and identifying potential disease targets, ultimately opening a way to establishing a link with pathogenesis and a deeper insight into roles for glycosylation. Proof of principle was first demonstrated for a patient with a known deficiency in the Mgat2 gene, and the strategy was then applied to a patient with liver disease of unknown cause with a type 2 Tf IEF pattern.
Background to specific CDG patients
Patient HE is a girl with a typical CDG-Ia presentation. Symptoms include liver dysfunction, neurological disease, susceptibility to infection, and dysmorphic features. There is a confirmed deficiency in the phosphomannomutase enzyme that converts Man-6-phosphate to Man-1-phosphate in a reaction essential for the synthesis of the oligosaccharide precursor (Glc3Man9GlcNAc2) prior to its transfer to the protein.
Patient VT was diagnosed with CDG-IIa at 9.5 years (Jaeken et al., 1994). This syndrome is characterized by growth retardation, psychomotor retardation, and facial dysmorphism. However, unlike patients with CDG-Ia he had no cerebellar hypoplasia and no peripheral neuropathy. He suffered from recurrent infections and, regrettably, died in 2002 from a severe pulmonary infection. The affected enzyme is UDP-N-acetylglucosaminyl:
6-D-mannoside-ß-1,2-N-acetylglucosaminyl-transferase-II (GlcNAcT-II) that catalyzes an essential step in the conversion of an oligomannose to a complex type N-glycan. The human gene for GlcNAcT-II (Mgat2) has been cloned and localized to chromosome 14q21 (D'Agostaro et al., 1995; Tan et al., 1995). Southern blot analysis shows that there is only one copy of Mgat2 in the human genome and the entire coding region is a single exon with an open reading frame of 1341 bp (Tan et al., 1996). VT is homozygous for the A1467G point mutation in the Mgat2 gene resulting in a H262R point mutation in the GlcNAcT-II protein.
Patient AC has severe liver disease without neurological involvement. Tf IEF was performed because an enterohepatic presentation is associated with CDG-Ib (phosphomannoisomerase deficiency). In fact, she showed a type II sialotransferrin pattern, but the affected enzyme was not identified.
|
Results |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and methodsReferences |
|---|
Tf and IgG were prepared as described in the Materials and methods section. The concentration of IgG in serum from patient VT (6 mg/ml) was considerably lower than that of AC (10 mg/ml), although both were within the normal range for children between 6 and 15 years of age (5.4–16.1 mg/ml; KIDS Foundation of New Zealand, available at www.pidsnz.co.nz/home.htm). Sugars were released directly from the proteins in sodium dodecyl sulfate (SDS)–polyacrylamide gel electrophoresis (PAGE) gel bands, labeled with 2-aminobenzamide (2-AB, see Materials and methods) and analyzed as described.
Glycan profiles of serum Tf
Figure 1 shows the normal phase (NP) high-performance liquid chromatography (HPLC) glycan profiles of Tf derived from four serum samples. The two predominant peaks (peaks 28 and 33, Figure 1 and Table I; Figure 2 explains the annotations) in the control were monosialylated (A2G2S1) and disialylated (A2G2S2) biantennary complex structures. The corresponding IEF pattern of Tf (inset) indicated a predominant S4 band consistent with a fully glycosylated Tf molecule containing A2G2S2 glycan structures. As expected, the glycan profile of the CDG-I Tf (patient HE) was not significantly different from the control, despite a major cathodic shift in the IEF bands of the protein. This is consistent with the established finding that CDG-I affects the number of glycans attached to the protein but does not affect the processing of the glycans once they are attached.
|
|
|
Altered glycan processing has also been reported for CDG-Ia patients (Mills et al., 2001
); however,this was not observed from our data of patient HE. In the CDG-IIa glycan profile (patient VT) there were no detectable peaks corresponding to those in the control sample. Matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry (MS) and exoglycosidase digestions coupled with HPLC indicated that the single peak (peak 13, Figure 1 and Table I) with a glucose unit (GU) of 6.79 had a composition of (Hex)4(GlcNAc)3(Neu5Ac)1 consistent with a truncated monosialylated structure. This consists of a chitobiose core with a GlcNAcGalNeu5Ac antenna on the 1,3 arm and a single mannose on the 1,6 arm, (A2G1S1; data not shown). The corresponding IEF pattern (Figure 1, VT) showed a predominant S2 band, consistent with the presence of Tf containing the truncated glycan at both glycosylation sites.
The fourth glycan profile (patient AC; Figure 1) is of a Tf sample from a patient with a hitherto undesignated CDG-II. There was a prominent peak (35, GU = 8.41) containing the hybrid structure GlcNAc2Man5GlcNAcGalNeu5Ac in the HPLC profile and peaks in the MALDI profile from this compound both with and without some residual peptide as the result of incomplete hydrazinolysis. In addition, both HPLC and MALDI profiles contained other hybrid and complex structures (Table I). The MALDI MS spectra of both the hybrid glycan and the glycopeptide contained ions corresponding to the presence of a free acid group and its sodium salt. The MALDI tandem MS (MS/MS) fragmentation spectrum of the free acid contained a prominent ion produced by loss of sialic acid (291 mass units). This hybrid structure was also found without a sialic acid (Neu5Ac) terminating the 3-antenna (GU = 7.46). Its structure was confirmed by enzyme array analysis; in addition, the presence of the unusual Man5 hybrid moiety was confirmed by MALDI MS using a tandem quadrupole time-of-flight (Q-TOF) instrument to obtain fragmentation (Figure 3). The spectrum contained an ion at m/z 833.3 ([Hex]5), consistent with a hybrid structure, and a prominent peak at 365 mass units less than the molecular ion indicating that the 1,3 arm contains HexNAc and hexose, which were subsequently identified by exoglycosidase digestion as GlcNAc and galactose.
|
Glycan profiles of serum IgG
In normal serum IgG, approximately 5% of light chains are glycosylated (Youings et al., unpublished data); however, no light chain glycosylation was detected in either patient. Figure 4 shows the serum IgG profiles for patients VT and AC and for a pooled control serum preparation. In contrast to control IgG, in which 19 glycans were clearly identified, consistent with previous studies (Jefferis et al., 1990
; Wormald et al., 1997) (Table II), IgG from VT contains an extremely restricted set of glycan structures, none of which are present in control IgG. More than 70% of the heavy chain glycan pool consists of a single core fucosylated monoantennary sialylated glycan structure (FcA1G1S1; peak 17 in Table II). The absence of complex type glycosylation is consistent with the ablation of the activity of GlcNAcT-II, and the data indicate that normal processing of IgG has terminated in the Golgi following the removal of the mannose residues to generate the trimannosyl core and the processing of the 1,3 arm by GlcNAcT-I. This deficiency would be expected to decrease the stability of IgG toward proteases, (Leatherbarrow and Dwek, 1984) to decrease the affinity for the Fc
receptor on monocytes (Leatherbarrow et al., 1985), and to provide a potential epitope for the mannose receptor and, if multiply presented, for mannose-binding lectin (Malhotra et al., 1995). The predominance of a single IgG glycoform is in strong contrast with control IgG, where many differently glycosylated variants are present (Figure 4, control).
|
|
IgG heavy chains from patient AC were also aberrantly glycosylated. Compared with the control, there was an overall increase in galactosylation leading to decreased amounts of nongalactosylated (G0) and monogalactosylated (G1) biantennary-type glycans and an increase in sialylation (Figure 4). Seventy percent of glycans in the control samples were of the G0 and G1 type, compared to 56% in patient AC. Sialylated structures account for 14% in the control and 26% in patient AC.
Glycan profiles of whole serum
Figure 5 (top) shows the NP HPLC profiles of the 2-AB-labeled N-glycans released from serum samples from VT, AC, and a control preparation by optimized hydrazinolysis. Hydrazinolysis was used because in control experiments when glycans were released by peptide N-glycosidase F (PNGase F), it was clear that not all sugars were released by this enzyme even under reducing conditions (data not shown). In a typical control serum taken from healthy individuals, 29 peaks with GU values from 5.5 to 10.6 were assigned N-glycan structures, and no significant differences in glycosylation were observed. (4 healthy individuals and 3 pools of serum from 30 healthy individuals were analyzed). The structural assignment of these peaks was based on their elution positions measured in GU values (Figure 5, top; Table III) compared with standard glycans and confirmed by systematic digestion of the glycans by exoglycosidase arrays (Figure 5, bottom) (Guile et al., 1996), MALDI TOF, and electrospray (ESI) MS (Table III). Seventy-four percent of the structures were biantennary, and 26% were triantennary. Fluorescent labeling of the glycan pool with 2-AB is nonselective, and integration of the peaks in the HPLC data gives quantitative data from which the relative proportions of the glycan structures can be determined. The major biantennary structure was nonfucosylated and disialylated (A2G2S2, peak 33 in Table III) whereas the major triantennary structure was nonfucosylated and disialylated (A3G2S2, peak 36 in Table III). There was a significant proportion (64%) of core fucosylated structures.
|
|
The CDG-II serum (VT) contained none of the peaks identified in the control sample (Figure 5; Table III). The predominant structures (peaks 13 and 17) were the fucosylated and nonfucosylated sialylated hybrid glycans (FcA1G1S1 and A1G1S1) derived from the trimannosylchitobiose core (M3N2) that is the natural substrate for GlcNAcT-II. HPLC and MALDI MS indicated the presence of some digalactosylated (peaks 27, 30) and disialylated glycans (peaks 32, 37). These additional galactose and sialic acid residues may be attached to the bisecting GlcNAc, which is not normally found on glycans derived from serum proteins. This type of structure was identified (Wang et al., 2001
) in tissue samples from Mgat2-null mice. The elongation on bisecting GlcNAc in the absence of the 6-arm branch suggests that there is usually restricted assess of galactosyl transferase to the bisecting GlcNAc residue. In contrast to VT, and predictably because only site occupancy is disturbed in CDG-I, the serum glycans from patient HE were not significantly different from the control (data not shown).
The serum profile from patient AC is shown in Figure 5 (top). Of particular significance is the presence of oligomannose-type glycans, such as M5N2 at GU 6.2 (peak 7, Tables I and III). The Tf glycan profile shows a 10-fold increase in the level of the oligomannose glycan M5N2 (peak 7, GU 6.2) and also the presence of hybrid structures (peaks 23 and 35 GU 7.46 and 8.41, Table I). Although the hybrid structures may be masked by other large peaks at higher GU values in the serum glycan profile, their presence is confirmed by corresponding peaks from MS and enzyme array analysis. In patient AC, IgG secreted from B cells contains the same glycans as normal IgG; however, the relative proportions of the sugars are different (Figure 4, Table II). These data indicate that the altered glycosylation is not confined to serum glycoproteins predominantly synthesized in the liver. In addition, it is clear that B cells contain some compensatory pathway because no oligomannose or hybrid-type glycans were identified on IgG.
The glycan pools from serum from patients VT and AC and from healthy controls were resolved on the basis of charge using weak anion exchange chromatography (Figure 6, top). The extent of sialylation was determined by comparison with standard fetuin sugars. The variously charged fractions from patient AC were resolved by NP HPLC to confirm the assignment of the sialylated structures in Table III (Figure 6, bottom). Consistent with the complete ablation of GlcNAcT-II activity and the NP HPLC data, which indicated the absence of complex-type glycans, the glycoproteins from the VT serum contained only neutral and monosialylated glycans. AC contained significantly more neutral structures, consistent with the NP HPLC data, which suggested that this patient has a mannosidase deficiency.
|
A summary of the characteristics of each glycan profile derived from these three patients and a control sample along with the confirmed or suggested enzyme defects and related clinical diagnosis is presented in Table IV.
|
Differential proteome analysis of the serum proteins from VT, AC, and control
Further insight into the changes in protein glycosylation and protein expression in patients VT and AC was achieved by differential proteome analysis of the serum proteins by two-dimensional gel electrophoresis (2-DE). Representative 2-DE images of proteins separated on 3–10 nonlinear pH gradients are shown in Figure 7 for the two patients and a serum control pool prepared from 28 healthy normal individuals (blood group AB).
|
Extensive mapping of the serum proteome enabled the detailed analysis of the expression profiles of all protein isoforms of transferrin detectable by 2-DE in the molecular weight range of 10–140 kDa and 3–10 pI. These spots are indicated in Figure 8A–C. The majority of Tf molecules are present as differently charged isoforms of a 77-kDa protein. Lower-molecular-weight isoforms of 60 kDa, 41 kDa, 35 kDa, 21 kDa, and 15–17 kDa represent truncated molecules that can be present in normal serum, probably by proteolysis in vivo. Post–sample collection protein digestion is unlikely to account for these truncated molecules because proteases were very effectively controlled by serum protease inhibitors and sample storage time at low temperature was kept to a minimum. Analysis of the two patients' sera indicates the selective presence of some of these low- molecular-weight Tf molecules. The data are summarized schematically in Figure 8, B3. In the serum of patient VT, compound 11 is present in addition to the main 77-kDa Tf array (compounds 1–6), whereas in the serum of patient AC, compounds 12 and 16 are present (see Figure 7). The different abundance of these Tf fragments might well arise from different susceptibility to degradation of the protein isoforms due to changes in glycosylation of the proteins. Differences in protein isoform expression of the 77-kDa isoforms as resolved on the 2-DE are shown in Figure 8C. These differences correlate well with the IEF patterns for Tf shown in Figure 1 in that a shift of isoform S4 to S2 is clearly visible for Tf from patient VT and even more pronounced, alongside a general reduction of Tf levels, in patient AC.
|
Differential analysis of the total 2-DE protein spot patterns provides a more comprehensive insight into the changes in protein expression and modification induced by the genetic defect in VT and by the putatively low activity of mannosidase III in AC. These analyses reveal changes in protein expression in response to the mutation per se and/or the disease pathogenesis. Changes in glycosylation can also be revealed, provided they give rise to a change in the physicochemical characteristics of the proteins (molecular weight and isoelectric point) that can be spatially resolved by the pH and gel gradient.
The differences in serum protein expression and spot position detected for patients VT and AC versus control are listed in Table V. A range of proteins, including clusterin, apolipoprotein E and L, Tf, 1-antitrypsin, complement C3 and C4, antithrombin III, fetuin, and -1 acid glycoprotein, known to be affected in CDG-Ia patients, are missing or modified in their isoelectric point and molecular weight in the two patients (Yuasa et al., 1995; Henry et al., 1997, 1999). As in CDG-I, clusterin is absent in these CDG-II patients. Additional proteins have been identified by our analysis that to our knowledge have not been reported to be affected by CDG. These molecules are the inter–alpha trypsin inhibitor heavy chain 4, complement factors B, complement factor H–related protein I, plasma glutathione peroxidase, leucine-rich -2-glycoprotein, serum paraoxonase arylesterase 1, plasma retinol binding protein, tetranectin, and transthyretin. These molecules are all glycosylated, and many play important roles in inflammation.
|
Changes observed in the protein pattern only for VT are found in antithrombin III, plasma glutathione peroxidase, leucine-rich
-2-glycoprotein, serum paraoxonase arylesterase 1, plasma retinol binding protein, and tetranectin. Interestingly, posttranslational modification of the prominent -1 acid glycoprotein appears not to be affected in this patient. For patient AC, specific changes are observed for the proteins transthyretin, carboxypeptidase N, complement factor B, fetuin A, ß2-glycoprotein 1, complement C3, and serum amyloid P. In contrast to the reduced molecular weight of serum amyloid P found in CDG-I, the expression level of this protein appears to be down-regulated in this potentially new form of CDG-II.
Mutation analysis of ManII and ManIIx on AC
Man IIa and Man IIx were sequenced at the cDNA level for patient AC and her parents. No pathogenic mutations were found. Two silent polymorphisms were found in the coding sequence of ManIIa: the patient is heterozygous for a 1299 G>A transition and homozygous for a 1449 G>A transition. In addition, the patient is heterozygous for a sequence variant 1235 A>G that results in a conserved substitution at the protein level (Q412R) in ManIIx. This variation was found in the normal population with a frequency of 27%. Thus these genes are probably not the cause of the deficiency in this patient.
|
Discussion |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and methodsReferences |
|---|
Since the first identification of a CDG in 1980, there has been a growing list of cases that can be ascribed to various mutations of enzymes associated with protein glycosylation (Jaeken et al., 1980
). However, to date only 9 defective enzymes have been identified from the 50 or so enzymes involved in the synthetic and processing steps of N-glycosylation (Freeze, 2001; Jaeken and Matthijs, 2001). This suggests that mutations leading to decreased activity of many of these enzymes may be incompatible with embryonic development and survival.
Strategy for identifying glycosylation processing disorders
Until now, diagnosis of CDGs has been based on an altered IEF profile of Tf as a marker protein. However, this technique is inadequate for complete characterization of the aberrant glycan structures, let alone pinpointing the defective processing step or the glycoproteins that might be affected. The strategy described here combines proteomics, which suggests potential disease targets and links to pathogenesis, with the detailed glycan analysis of a pool of proteins, such as serum. Even the analysis of the glycan structures of a specific isolated protein, such as IgG or Tf, can give a good indication of the key abnormal glycans, as was evident in our patient samples. Glycan analysis is useful for identifying glycosylation disorders for three reasons. First, a comparison of the glycan profiles of glycoproteins from a patient with those from a healthy control can highlight unexpected glycan structures and lead to the identification of a defective enzyme or blocked processing step in the patient. Second, the extent of the glycosylation changes may provide an insight into the severity of the enzyme deficiency or block. Third, the glycan profile of proteins expressed in different cells, such as Tf and IgG expressed in hepatocytes and B cells, respectively, can indicate the cell specificity of the disorder.
Serum proteome of patients VT and AC suggests disease targets
The 2-DE of serum glycoproteins from patients VT and AC and from a healthy control were compared to obtain an overview of the range of proteins directly or indirectly affected by structural changes in the glycans. In both patients the comparison with control serum revealed that many proteins were unaffected; however, in addition to transferrin and immunoglobulin, a range of serum glycoproteins was identified which did have altered expression levels and physicochemical characteristics (Table V). Probably more proteins where aberrant glycosylation does not lead to a significant change in molecular size or charge are also affected. Other proteins, including many complement components, were not detected on the CDG proteographs. This suggests that their expression has been directly or indirectly down-regulated as a consequence of the glycosylation defect. It is also possible that in some cases terminal Man or exposed GlcNAc residues on unprocessed sugars may have led to rapid clearance of some glycoproteins, thus lowering their concentration in the serum.
Glycan and proteome analysis of serum glycoproteins in patient VT is consistent with ablation of GlcNAcT-II enzyme activity and recurrent infections
In the case of patient VT, a single mutation in the Mgat2 gene leads to a completely defective GlcNAcT-II enzyme. This is a stringent mutation that leads to a loss of >98% of the normal enzyme activity of GlcNAcT-II in fibroblasts and mononuclear blood cells isolated from patient VT (Jaeken et al., 1996). The glycan analysis of serum glycoproteins (Figure 5, Table III) was consistent with a metabolic block in glycan processing that leads to the accumulation of truncated glycan structures, which are the expected substrates of the GlcNAcT-II enzyme. The 2-DE analysis shows a decreased negative charge (more basic pI) for some multisialylated proteins that most probably reflects missing sialyl groups.
Not surprisingly, the serum proteome analysis does not identify a protein factor that could directly be associated with the clinical characteristics of the syndrome (growth retardation, mental retardation, and facial dysmorphism) because these may be expected to involve proteins expressed in tissues other than the liver and B cells from which serum proteins are mainly derived. However, the expression levels, serum concentrations, structure, and possibly the function of proteins that play a vital role in blood coagulation (antithrombin), metabolism (e.g., lipid transport and exchange), and immunity (complement factors) appears to be affected with the obvious consequences for the fitness of the patient.
The severe recurrent infections suffered by patient VT and eventual death by severe pulmonary infection may well have been the clinical manifestations of aberrant structures and levels of components of the immune system. The level of IgG in the serum (
6 mg/ml) was at the low end of the normal range and may have contributed to the disease pathogenesis. Over 70% of the heavy chain glycan pool contained a single core fucosylated monoantennary sialylated glycan structure (FcA1G1S1, peak 17) that may compromise both the stability and receptor-mediated functions of IgG (Jefferis et al., 1998; Ghirlando et al., 1999). In contrast to normal serum IgG (Youings et al., 1996), no light chain glycosylation was detected in the IgG of patient VT, suggesting either defective glycan attachment or that the IgG was derived from only a subset of B cells. Low levels of antibodies have been reported in the mouse model of CDG-IIa, in which a partial block in pre–B cell development leads to reduced levels of mature B cells (Wang et al., unpublished data). Additional galactose and sialic acid residues may be attached to the bisecting GlcNAc residue (peaks 27, 30, 32, 37; Figure 5) which is not normally elongated in the serum (Wang et al., 2001). This type of structure was identified in tissue samples from Mgat2-null mice and supports the use of the mouse model for studying the human disease.
Patient VT represents a clear example of how a single point mutation in the gene of a glycosylation enzyme (GlcNAcT-II) can result in structural changes to a range of glycoproteins that are expressed in various tissues, leading to severe multisystemic clinical symptoms. The severity of symptoms caused by a defect in this enzyme is also clear from the low survival rates in both human patients and animal models. Mgat2-null mice have a reported survival of only 1% into the second week of postnatal life (Wang et al., 2001).
Glycan analysis is consistent with a block in the -mannosidase III pathway in patient AC
In patient AC, a different set of aberrant glycans accumulate. These are pentamannosyl hybrid type glycans with varying lengths of 1-3-linked antenna. The data indicate that GlcNAcT-I can modify the 1,3 arm and, subsequently, GalTase and sialylTase can further process the GlcNAc residue, implying that the later part of the glycan processing pathway is unaffected. The defect appears to be the inability to cleave two terminal mannose residues with 1- and 1-6 linkages to the 1-6-linked antenna of the core structure.
There are two major classes of mannosidase enzymes involved in N-glycosylation processing (Moremen et al., 1994). Class 1 -mannosidases specifically cleave 1,2-linked terminal mannose residues and typically trim Man9- GlcNAc2 to a Man5GlcNAc2 in the ER and Golgi. Class 2 -mannosidases cleave terminal mannose residues from a wider range of linkages, including 1,3; 1,6; and 1,2. The class 2 mannosidases localized to the Golgi typically cleave 1,3 and 1,6 linked mannose residues, a prerequisite for processing the 1,6 antenna of the core N-glycan structure.
The glycan analysis of glycoproteins from patient AC is consistent with the low activity of a class 2 -mannosidase that cleaves mannose residues in 1,3 and 1,6 linkages (Figure 9).
|
The most well-characterized class 2 enzyme is the
3/6-mannosidase II (MII), which initiates the first committed step of the conversion of a hybrid to a complex glycan (Schachter, 2001). The enzyme requires the addition of a terminal GlcNAc residue prior to mannose cleavage (GlcNAcMan5GlcNAc2 to GlcNAcMan3GlcNAc2). Prior addition of the terminal GlcNAc is affected by the enzyme GlcNAcT-I. Evidence for the accumulation of pentamannosyl hybrid-type glycans following mutation of the mannosidase II gene has been reported for ricin-resistant baby hamster kidney cells in culture (Hughes and Feeney, 1986). The knockout mouse lacking a functional mannosidase II gene showed a phenotype in homozygotes of dyserythropoiesis similar to the pathology of human congenital dyserythropoietic anemia (HEMPAS) (Chui et al., 1997). The MII knockout mice also showed a late-onset autoimmune disease similar to human systemic lupus erythematosus (Chui et al., 1997). Although normal erythropoiesis was inhibited in these mice, the nonerythroid cells continued to produce normal complex N-glycans, suggesting the existence of alternative mannosidase activities.
Two alternative routes of mannose cleavage in glycan processing have been described and suggest a network of biosynthetic pathways associated with the mannosidase enzymes (Figure 9). The human mannosidase IIx (Mx) is an enzyme closely related to MII. The gene encoding Mx (Man22) has been identified in the human genome by cross-hybridization with the cDNA from human MII (Misago, 1995; Oh-Eda et al., 2001). Loss of activity of this enzyme in Mx knockout mice results in ineffective spermatogenesis and infertility of the males (Fukuda and Akama, 2002). Mannose cleavage can also be provided by 3/6-mannosidase III (MIII) that has a GlcNAcT-I-independent activity and converts Man5GlcNAc2 to Man3GlcNAc2. This is described as a cobalt-activated enzyme that has been detected in rat liver Golgi, in baby hamster kidney cells, and significantly in the nonerythroid cells of MII-null mice (Moremen, 2002).
The role of each of these three alternative class 2 mannosidase enzymes (MII, Mx, and MIII) in glycosylation processing is unclear. Although each can trim terminal mannose groups from a hybrid glycan, the substrate specificities differ. It is also unlikely that all the enzymes are expressed in a single cell type and tissue- and species-specific expression of distinct mannosidases has been shown (Moremen et al., 1994).
The existence of multiple forms of the mannosidases allows the possibility that a compensatory mechanism can limit potential damage caused by a mutation in one individual enzyme. Also the clinical symptoms of any defect are likely to depend on the relative distribution of enzyme isoforms in different tissues (Moremen et al., 1994). The relatively mild condition of anemia that occurs in the MII-null mice can be explained by the processing of normal complex glycans in nonerythroid cells, through the apparent activity of MIII detected in these cells (Chui et al., 1997; Moremen, 2002). The enzymatic activity determined for Mx appears to be minimal and is less likely to account for a compensatory mechanism in these animals (Fukuda and Akama, 2002). The relatively mild pathological effects of the loss of human MII in HEMPAS could also be explained by a compensatory mechanism of activity offered by an alternative mannosidase isoform.
The unexpected hybrid glycans identified in the serum proteins of AC are the natural substrates for mannosidase II, IIx, or III, and their presence suggests the location of an enzyme defect in the processing pathway. However, the putative metabolic block is incomplete because there is still a significant proportion (50%) of normal complex glycans, both in the Tf sample and the serum sample (Tables I and III). These observations could be explained by at least one mutation that decreases the enzyme activity but does not ablate it. In both the Tf and whole-serum protein glycans, the 1,3 antenna of the core structure is processed by the addition of GlcNAc, galactose, and sialic acid, indicating that the later part of the glycan processing pathway is unaffected. The addition of a bisecting GlcNAc through the action of GlcNAcT-III can also block the processing of biantennary complex glycans, but an up-regulation of this enzyme is unlikely to be a cause of the hybrid glycans because no bisecting GlcNAc residues were detected in Tf (AC = 0%, control = 0%) and similar amounts in IgG (AC = 41%, control = 46%), and only low levels were present in the serum glycan profile (AC = 9%, control = 25%) (Figures 1, 4, 5 and Tables I, II, III).
The affected mannosidase gene in AC is unlikely to be MII because there is no evidence of the pathological symptoms characteristic of HEMPAS. Therefore, we proposed that a partial loss of mannosidase activity is a result of a defect in an alternative mannosidase isozyme, probably mannosidase III or mannosidase IIx. However, genetic sequence analysis of lymphocytes from AC showed that the mannosidase II and mannosidase IIx genes were apparently normal. This leads to the conclusion that the defect causing aberrant glycans in patient AC is likely to be in the enzyme mannosidase III, the gene for which has not yet been cloned.
A compensatory enzymatic mechanism may explain the severity of the disorder
Compensatory mechanisms have also been described for other mutations in the glycosylation pathway (De Praeter et al., 2000). Here a defect in glucosidase I leads to the up-regulation of a Golgi 1,2-endomannosidase (class I) that enables the normal requirement for glucose removal prior to the action of an 1,2-exomannosidase to be bypassed (Lubas and Spiro, 1987; Volker et al., 2002).
However, there is no evidence for a similar compensatory mechanism or bypass pathway for a defect in the GlcNAcT-II enzyme. We might speculate that the reason for this is that the gene is more recent in evolutionary development and consequently confined to mammalian cells that are able to process complex glycans. Significantly, there is only one copy of the Mgat2 gene, and it has an uninterrupted reading frame and no reported homologous sequences with other proteins. The consequence is that a single point mutation in this gene has serious physiological effects that give rise to a severe clinical condition.
According to the serum proteome, the proposed enzyme defect in patient AC appears to directly or indirectly affect the structure and/or serum concentrations of specific glycoproteins including transthyretin, carboxypeptidase N, complement factor B, fetuin A, ß2-glycoprotein 1, complement C3, and serum amyloid P, alongside the common CDG marker proteins of clusterin, Tf, and apolipoprotein E and L. Although it cannot be ruled out that the absence of these proteins is a secondary consequence of the diagnosed severe liver disease, at the least these data provide a link to pathogenesis.
In conclusion, we have demonstrated that glycan analysis of a pool or of specific proteins can allow the assignment of unexpected glycan structures. This information can lead to the identification of a defective glycosylation processing step. In addition, the glycan profile of proteins expressed in different cells can indicate the cell specificity of the disorder, whereas the extent of glycosylation changes provides an insight into the severity of the processing block. To complement this information, 2-DE of serum provides an overview of the range of proteins directly or indirectly affected by structural changes in the glycans, potentially providing insights into pathogenesis and disease targets.
|
Materials and methods |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and methodsReferences |
|---|
Extraction of Tf
Serum (15 ml) was loaded onto a column containing anti-human Tf IgG Sepharose 4B. After washing, the Tf was eluted with 0.2 M glycine/HCl pH 2.8 and saturated with iron using a ferric chloride solution. The sample was then concentrated on a dialysis membrane (Hershberger et al., 1991
).
Extraction of IgG
Serum samples were dialyzed against 0.1% trifluroacetic acid (TFA) overnight. IgG was isolated from the serum samples using a Protein G affinity column. (MabTrap kit, Amersham Biosciences, Uppsala, Sweden). Serum was applied at a flow rate of 1 ml/min to the column that had been equilibrated using the binding buffer supplied with the kit. The column was washed until the absorbance at 280 nm was zero. Elution buffer was added, and IgG was collected into tubes prepared with a neutralizing buffer. The isolated IgG was dialyzed against 0.1% TFA overnight, lyophilized for 24 h, and then dissolved in double-distilled water to a concentration of 1 mg/ml.
In-gel release of Tf glycans
Tf samples (5–10 µg) were applied to individual gel lanes for separation by SDS–PAGE with precast gels (NuPage 10% with 4-morpholine propane sulfonic acid, pH 7, as running buffer). Coomassie blue staining resulted in a distinct band at approximately 80 kDa, as determined from molecular weight markers. The protein bands from three identical lanes were removed from the stained gel by scalpel and treated with PNGase F to remove the attached glycans (Rudd et al., 2001).
In-gel release of IgG glycans
IgG samples (70 µg) were reduced using 0.5 M dithiothreitol (DTT), alkylated using 100 mM iodoacetamide, and then separated by SDS–PAGE (freshly made 12.5% gel). Coomassie blue staining showed distinct bands at approximately 50 kDa and 28 kDa, corresponding to IgG heavy and light chains, respectively. The gel bands were cut out and washed alternatively in sodium bicarbonate buffer (20 mM, pH 7) and acetonitrile. The glycans were released from the protein in the gel bands by PNGaseF digestion overnight at 37°C. The glycans were extracted from the gel pieces with water and sonication.
Glycan removal from serum proteins
Hydrazinolysis. Each dialyzed serum sample (50 µl) was subjected to hydrazinolysis to remove the N-glycans. After extensive lyophilization, the samples were sealed in glass tubes with 0.1 ml anhydrous hydrazine under an atmosphere of argon. The tubes were heated to 85°C at a rate of 10°C per h and held at this temperature for 12 h. The tubes were then cooled, and the hydrazine was allowed to evaporate. Toluene (250 µl) was added and evaporated (x5).
Reacetylation
Each sample was then dissolved in 200 µl 0.2 M sodium acetate pH 8 and reacetylated for 60 min at room temperature by the addition of 50 µl acetic anhydride. The sample was then desalted by passage through a 0.5 ml Dowex AG50 X12 column and elution with 4 x 0.5 ml water. The sample was dried and redissolved in 100 µl distilled water.
Glycan purification
The reacetylated glycans were purified by applying the aqueous solution to 1-ml columns of processed glass beads (Oxford GlycoSciences, Oxford, U.K.) held in BioRad (Hercules, CA) disposable columns. The applied samples were washed successively with 5 x 5 ml double-distilled water, 2 x 5 ml ethanol, 6 x 5 ml butanol, and 8:2:1 (v/v/v) butanol/ethanol/water. The glycans were finally eluted into 4 x 0.5 ml water. The samples were dried and redissolved in 100 µl double-distilled water in preparation for labeling with 2-AB.
Fluorescent labeling with 2-AB
The glycans were fluorescently labeled with 2-AB by reductive amination according to the method described by Bigge et al. (1995) using the Oxford GlycoSciences Signal labeling kit (OGS, Abingdon, Oxon, U.K.).
Glycan analysis
HPLC analysis was by normal phase (Glycosep N). Structural assignment of the peaks arising from this separation was made following subsequent treatment with multienzyme arrays of exoglycosidases (Guile et al., 1996). Confirmation of structures was obtained by MS (MALDI and ESI).
HPLC
NP HPLC was performed according to the low salt buffer system as previously described (Guile et al., 1996) using a 4.6 x 250 mm Glycosep-N column (OGS). The system was calibrated using an external standard of hydrolyzed and 2-AB-labeled glucose oligomers to create a dextran ladder. Weak anion exchange (WAX) HPLC (Guile et al., 1994) was performed using a Vydac 301VHP575 7.5 x 50 mm column (Anachem, Bedfordshire, U.K.) according to the modified methodology (Zamze et al., 1998).
MALDI MS
Positive-ion MALDI TOF mass spectra were recorded with a Micromass TofSpec 2E reflectron-TOF mass spectrometer (Micromass, Manchester, U.K.) fitted with delayed extraction and a nitrogen laser (337 nm). The acceleration voltage was 20 kV; the pulse voltage was 3200 V; and the delay for the delayed extraction ion source was 500 ns. Samples were prepared by adding 0.5 µl of an aqueous solution of the sample to the matrix solution (0.3 µl of a saturated solution of 2,5-dihydroxybenzoic acid [DHB] in acetonitrile) on the stainless steel target plate and allowing it to dry at room temperature. The sample/matrix mixture was then recrystallized from ethanol (Harvey, 1993). All measured monoisotopic masses were within 0.2 Da from the calculated values.
MALDI MS/MS data
MALDI MS/MS data were recorded with a Micromass Q-TOF mass spectrometer fitted with an experimental MALDI ion source (Harvey et al., 2000). The sample, in 1 µl water, was mixed with 2,5-DHB (0.9 µl of a saturated solution in acetonitrile) on the MALDI probe and allowed to dry. The probe was introduced into the ion source of the mass spectrometer via a vacuum lock, and the laser was fired at 10 Hz. Signals were accumulated for 5 s for each spectrum, and positive ion spectra were accumulated until a satisfactory signal:noise ratio was obtained. For MS/MS (collision-induced dissociation) spectra, the precursor ion ([M + Na]+) was selected with a mass window of about 3 Da, argon was used as the collision gas, and the collision energy was set to record fragments across the entire mass range.
HPLC LC-ESI-MS
ESI–liquid chromatography/MS (LC/MS) data were obtained with a Waters CapLC HPLC system interfaced with a Micromass Q-TOF mass spectrometer fitted with a Z-spray electrospray ion source and operated in positive ion mode. A 1 x 150 mm microbore NP HPLC column was packed with stationary phase material from a Glycosep N column (Oxford GlycoSciences). The operating conditions for the mass spectrometer were: source temperature 100°C; desolvation temperature 120°C; desolvation gas flow 200 L/h; capillary voltage 3000 V; cone voltage 30 V; TOF survey scan time 1 s, mass range 50–2300; TOF MS/MS scan time 1 s, survey scan 950–1600 with detection mass range 50–3500; and mass selection resolution about 3 Da. The MS to MS/MS automatic switching was initiated when the nominated peak intensity rose above 4. Switching back to MS mode occurred after 30 s or when the peak intensity fell below 1. The mass of this peak was then ignored by the automatic switching routine for 30 s. An automatic collision energy (CE) profile was used with CEs of 14–32 V. Results from the different CEs were combined for data evaluation.
2-DE
Two hundred micrograms of the respective human serum were dissolved in 375 µl of sample buffer (5 M urea, 2 M thiourea, 0.002 M tributyl-phosphine, 0.0065 M DTT, 0.065 M 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate, 0.15 M NDSB-256) and were subjected to 2-DE. Ampholytes were added to the sample at 0.9% Servalyte 3–10, 0.45% Servalyte 2–4 and 9–11. Immobilized pH gradient gels (Immobiline DryStrip 3–10 NL) were rehydrated in the sample and IEF was carried out at 70 kVh at 17°C according to the method described by Sanchez et al. (1997). Following focusing, the immobilized pH gradient gel strips were immediately equilibrated for 15 min in 6 M urea/2% (w/v) SDS, 2% (w/v) DTT, 50 mM Tris, pH 6.8, 30% glycerol. Proteins were separated in the second dimension at 30 mA per gel and 20°C on 9–16% T, 2.7% C SDS–PAGE gels chemically bound to the rear glass plate in an electrophoresis tank similar to that described by Amess and Tolkovsky (1995). Following electrophoresis, the gels were fixed in 40% (v/v) ethanol:10% (v/v) acetic acid and stained with the fluorescent dye OGT MP17 according to the method previously described (Hassner et al., 1984). Sixteen-bit monochrome fluorescence images at 200 mm resolution were obtained by scanning the gels with an Apollo II linear fluorescence scanner (OGS).
Image analysis
Scanned images were processed with a custom version of MELANIE II. The resolved protein features were checked, and gel and staining artifacts were removed manually. From triplicates of 2-DE gel images obtained for each sample, master images were generated representing features that were present on at least two out of three gels. The 2-DE gel for the normal human serum had previously been calibrated for molecular weight and isoelectric points and mapped extensively by nanoelectrospray MS/MS following in-gel protein trypsinolysis (manuscript submitted for publication). This image served as a template for the differential image analysis and protein spot identification.
The resolved protein features were checked manually, quantified on the basis of fluorescence signal intensity by summing pixels within each feature boundary, and recorded as a percentage of the total feature intensity on the image.
Mutation analysis of ManII and ManIIx on AC
RNA was isolated from an EBV-transformed lymphoblast cell line from the patient and the parents with Trizol (Invitrogen, Merelbeke, Belgium) according to the manufacturer's protocol. cDNA was prepared with Superscript II Reverse Transcriptase (Invitrogen) and oligo(dT)12–18 priming. The complete coding region of the mannosidase IIA (GenBank accession number NM_002372) and mannosidase IIx (NM_006112) was sequenced on this cDNA in both directions. The primer sequences are available upon request.
|
Acknowledgements |
|---|
We thank R. Bateman and R. Tyldesley of Micromass for access to the MALDI-Q-TOF mass spectrometer and the Higher Education Funding Council for England and the Biotechnology and Biological Sciences Research Council for providing funds to purchase the other mass spectrometers. We thank David Chittenden and Chris Brock for running and curating the proteome gels. P.M.R. and R.A.D. thank H.C.L. for her constant inspiration.
1 To whom correspondence should be addressed; e-mail: pmr{at}glycob.ox.ac.uk 
|
Abbreviations |
|---|
2-AB, 2-aminobenzamide; CE, collision energy; CDG, congenital disorder of glycosylation; 2-DE, two-dimensional gel electrophoresis; DTT, dithiothreitol; DHB, dihydroxybenzoic acid; ER, endoplasmic reticulum; ESI, electrospray ionization; GU, glucose unit; HEMPAS, hereditary erythroblastic multinuclearity with a positive acidified serum lysis test (human congenital dyserythropoietic anemia type II); HPLC, high-performance liquid chromatography; IEF, isoelectric focusing; LC/MS, liquid chromatography/mass spectrometry; MALDI, matrix-assisted laser desorption/ionization; MS, mass spectrometry; MS/MS, tandem mass spectrometry; NP, normal phase; PAGE, polyacrylamide gel electrophoresis; PNGase F, peptide N-glycosidase F; Q, quadrupole; Tf, transferrin; SDS, sodium dodecyl sulfate; TFA, trifluoroacetic acid; TOF, time-of-flight; WAX, weak anion exchange
|
References |
|---|
|
TopAbstractIntroductionResultsDiscussionMaterials and methodsReferences |
|---|
Amess, B. and Tolkovsky, A.M. (1995) Programmed cell death in sympathetic neurons: a study by two-dimensional polyacrylamide gel electrophoresis using computer image analysis. Electrophoresis, 16, 1255–1267.[CrossRef][ISI][Medline]
Bigge, J.C., Patel, T.P., Bruce, J.A., Goulding, P.N., Charles, S.M., and Parekh, R.B. (1995) Nonselective and efficient fluorescent labeling of glycans using 2-amino benzamide and anthranilic acid. Anal. Biochem., 230, 229–238.[CrossRef][ISI][Medline]
Carchon, H., Van Schaftingen, E., Matthijs, G., and Jaeken, J. (1999) Carbohydrate-deficient glycoprotein syndrome type IA (phosphomannomutase-deficiency). Biochim. Biophys. Acta, 1455, 155–165.[Medline]
Chui, D., Oh-Eda, M., Liao, Y.F., Panneerselvam, K., Lal, A., Marek, K.W., Freeze, H.H., Moremen, K.W., Fukuda, M.N., and Marth, J.D. (1997) Alpha-mannosidase-II deficiency results in dyserythropoiesis and unveils an alternate pathway in oligosaccharide biosynthesis. Cell, 90, 157–167.[CrossRef][ISI][Medline]
D'Agostaro, G.A., Zingoni, A., Moritz, R.L., Simpson, R.J., Schachter, H., and Bendiak, B. (1995) Molecular cloning and expression of cDNA encoding the rat UDP-N-acetylglucosamine: alpha-6-D-mannoside beta-1,2-N-acetylglucosaminyltransferase II. J. Biol. Chem., 270, 15211–15221.
de Lonlay, P., Seta, N., Barrot, S., Chabrol, B., Drouin, V., Gabriel, B.M., Journel, H., Kretz, M., Laurent, J., Le Merrer, M., and others. (2001) A broad spectrum of clinical presentations in congenital disorders of glycosylation I: a series of 26 cases. J. Med. Genet., 38, 14–19.
De Praeter, C M., Gerwig, G.J., Bause, E., Nuytinck, L.K., Vliegenthart, J.F., Breuer, W., Kamerling, J.P., Espeel, M.F., Martin, J.J., De Paepe, A.M., and others. (2000) A novel disorder caused by defective biosynthesis of N-linked oligosaccharides due to glucosidase I deficiency. Am. J. Hum. Genet., 66, 1744–1756.[CrossRef][ISI][Medline]
Domon, B. and Costello, C.E. (1988) A systematic nomenclature for carbohydrate fragmentations in FAB-MS/MS spectra of glycoconjugates. Glycoconj. J., 5, 397–409.[CrossRef][ISI]
Dupre, T., Barnier, A., De Lonlay, P., Cormier Daire, V., Durand, G., Codogno, P., and Seta, N. (2000) Defect in N-glycosylation of proteins is tissue-dependent in congenital disorders of glycosylation Ia. Glycobiology, 10, 1277–1281.
Freeze, H. H. (2001) Update and perspectives on congenital disorders of glycosylation. Glycobiology, 11, 129R–143R.
Fukuda, M.N. and Akama, T.O. (2002) In vivo role of alpha-mannosidase IIx: ineffective spermatogenesis resulting from ta