DDOST Mutations Identified by Whole-Exome Sequencing Are Implicated in Congenital Disorders of Glycosylation

Abstract

Congenital disorders of glycosylation (CDG) are inherited autosomal-recessive diseases that impair N-glycosylation. Approximately 20% of patients do not survive beyond the age of 5 years old as a result of widespread organ dysfunction. Although most patients receive a CDG diagnosis based on abnormal glycosylation of transferrin, this test cannot provide a genetic diagnosis; indeed, many patients with abnormal transferrin do not have mutations in any known CDG genes. Here, we combined biochemical analysis with whole-exome sequencing (WES) to identify the genetic defect in an untyped CDG patient, and we found a 22 bp deletion and a missense mutation in DDOST, whose product is a component of the oligosaccharyltransferase complex that transfers the glycan chain from a lipid carrier to nascent proteins in the endoplasmic reticulum lumen. Biochemical analysis with three biomarkers revealed that N-glycosylation was decreased in the patient's fibroblasts. Complementation with wild-type-DDOST cDNA in patient fibroblasts restored glycosylation, indicating that the mutations were pathological. Our results highlight the power of combining WES and biochemical studies, including a glyco-complementation system, for identifying and confirming the defective gene in an untyped CDG patient. This approach will be very useful for uncovering other types of CDG as well.

Main Text

Glycosylation is the biosynthetic process of adding glycans to proteins and lipids and is an important modification of secretory and membrane-bound proteins. About 1%–2% of the genome is believed to be involved in glycosylation.¹ Defects within the N-glycosylation biosynthesis pathway that alter the assembly or processing of glycans result in congenital disorders of glycosylation (CDG).^2,3 Symptoms of CDG often begin in infancy, and nearly all organs, particularly the nervous system, can be affected.^4,5 The 80% of patients who survive the first 5 years often have medical problems for the rest of their lives.^6,7 The spectrum and severity of organ dysfunction is very broad, making the clinical diagnosis of this group of disorders especially challenging. There is currently an effective treatment for only one CDG subtype, namely, MPI-CDG (CDG-Ib [MIM 602579]), via oral mannose supplementation.² Although some types of CDG can be diagnosed via a blood test for the glycosylation status of serum transferrin, lipid-linked oligosaccharide analysis in fibroblasts, enzymatic tests, and mutation analysis, such strategies leave a significant number of CDG patients without a molecular diagnosis. Here, we have combined whole-exome sequencing (WES) with biochemical and functional analysis to identify the defective gene in an untyped CDG patient.

Clinical data, serum, and DNA samples were collected under research projects approved by the institutional review board of the Sanford-Burnham Medical Research Institute, and informed consent was obtained from the patient's family. At 6 months of age, the patient (CDG-0235), a male of northern European descent born in 2004, presented with failure to thrive, gastresophageal reflux disease, developmental delay, ear infections, and oromotor dysfunction; hypotonia, external strabismus, and mild to moderate liver dysfunction were noted at 1 year of age. The patient did not walk until 3 years of age. At 5 years old, his fine motor skills were at the 26 month level, and his gross motor skills were at the 16 month level. He experienced frequent constipation and a deficiency of factor XI, antithrombin-III, protein C, and protein S at 6 years of age. Brain magnetic resonance imaging suggested disordered myelination. The patient also had slightly advanced bone age and mild osteopenia. When he was 7 years old, he had two episodes of tremors that affected the extremities and that were not clearly associated with loss of consciousness. Upon his last evaluation at 7.5 years of age, his weight, height, and head circumference were in the 52^nd, 35^th, and below the 50^th percentiles, respectively. He is currently without language and functions best in a social setting. We analyzed his serum transferrin by using ESI-MS (electrospray ionization mass spectrometry) and MALDI-TOF (matrix-assisted laser desorption/ionization–time of flight), revealing a type 1 pattern in which both monoglycosylated and aglycosylated transferrin were markedly elevated. MALDI-TOF analysis of both N-linked and O-linked glycans from whole-serum proteins showed normal glycan structures. Common secondary glycosylation defects were also excluded.

Targeted next-generation sequencing analysis of 25 genes associated with CDG failed to uncover any mutations.⁸ To identify the defective gene in this patient, we performed WES targeting ∼22,000 genes by using the Roche NimbleGen SeqCap EZ whole-exome kit (NimbleGen, Madison WI) and the Illumina HiSeq platform to generate single-end 50 bp reads (San Diego, CA). NextGENe software mapped the reads to the hg19 reference genome. Reads were aligned and filtered as described previously.⁸ We obtained 7.71 Gb of mappable sequence data for this patient. Out of a total of 80,048,259 bases, 79,012,293 bases were covered, and there was an average coverage of 25×. More than 84% of the targeted regions had greater than 10× coverage. From an internally generated list of ∼250 genes involved in glycosylation, a total of 311 variants were detected in the coding regions and dinucleotide splice sites of these genes, and 213 of those were nonsynonymous. A total of 92 variants were detected in the noncoding regions up to +20 or −20 bases into the intron.

WES detected two mutations in DDOST (MIM 602202), which encodes a 48 kDa subunit of the N-glycosylation oligosaccharyltransferase (OST) complex (RefSeq accession number NM_005216.4). Using Sanger sequencing, we detected one copy of a 22 bp deletion in exon 11 (c.1265_1286del22) and one copy of a missense mutation (c.650G>A) that was predicted to change glycine to aspartic acid at codon 217 (p.Gly217Asp) in exon 6 (Figure 1). This particular glycine residue is highly conserved, and PolyPhen, SIFT, and PANTHER all predicted the mutation to be damaging.⁹ The deletion, on the other hand, creates a stop codon 7 amino acids downstream (p.Ile422Thrfs^∗7) and truncates 29 C-terminal amino acids. The mutant-allele percentages of the deletion and missense mutation were 41% and 64%, respectively, and their respective coverages were 44× and 36×. Neither mutation was present in the 1000 Genomes Project, in the dbSNP database, or in 100 ethnically matched controls (data not shown). Family studies confirmed that the missense mutation was inherited from the mother and that the deletion was inherited from the father (Figure 1).

Figure 1.

NGS Detection, Sanger-Sequencing Confirmation of Mutations in DDOST, and Pedigree for Patient CDG-0235

(A) NGS detection (labeled by arrow) of c.650G>A in exon 6.

(B) NGS detection (labeled by arrow) of c.1265_1286 del22 in exon 11.

(C) Sanger-sequencing confirmation of c.650G>A (labeled by arrow).

(D) Sanger-sequencing confirmation of c.1265_1286 del22 (labeled by arrow).

(E) Family pedigree showing that patient CDG-0235 inherited the missense mutation from his mother and the 22 bp deletion from his father. One sibling inherited the missense mutation, and the other sibling did not inherit either mutation.

The missense mutation detected in exon 6 of DDOST is present in the luminal domain of the endoplasmic reticulum (ER) and substitutes a highly conserved hydrophobic amino acid with a negatively charged hydrophilic amino acid that most likely impairs DDOST function. The deletion in exon 11 is adjacent to the transmembrane domain in the lumen of the ER and includes a small portion of the transmembrane domain. This deletion produces a protein with a 29 amino acid C-terminal truncation, which is presumably nonfunctional. We hypothesized that the mutations detected in patient CDG-0235 would impair DDOST function, most likely resulting in protein hypoglycosylation.

We then investigated whether these mutations impaired glycosylation in the patient. Using a polyclonal DDOST antibody (D6820 Sigma), we performed immunoblot analysis on patient fibroblasts and found that there was a ∼50% reduction in protein expression (Figure S1, available online). We saw no smaller bands, suggesting that the truncated protein was probably unstable and degraded. We used three markers to assess the glycosylation status of patient fibroblasts. One marker is an engineered DNase-I with a single glycosylation site that is required for normal secretion into the media when expressed in human fibroblasts.^10,11 DNase-I experiments were performed as described elsewhere.¹² Secreted DNase-I in fibroblasts of the patient and other CDG patients was strongly reduced compared to that in control fibroblasts (Figure 2A).^13,14 ICAM-1 is another marker that normally localizes in the plasma membrane, but it is degraded and rendered unstable by hypoglycosylation. We analyzed ICAM-1 expression in the patient fibroblasts by using flow cytometry with fluorescein isothiocyanate (FITC) anti-human ICAM-1 antibody (BioLegend, San Diego, CA). We also used immunofluorescence (IF) of ICAM-1 and immunoblot analysis with a monoclonal ICAM-1 antibody (R&D Systems, Minneapolis, MN) to assess ICAM-1 levels in patient fibroblasts. Previously, ICAM-1 expression in membrane fractions from CDG patients' fibroblasts was found to be absent or significantly reduced (He and Freeze, unpublished results; He et al. in preparation). ICAM-1 was also absent from the membranes of patient fibroblasts (Figure 2B–2D).

Figure 2.

Glycosylated Expression of DNase-I and ICAM-1 in Patient CDG-0235 is Reduced, Indicating a Glycosylation Defect

(A) DNase-I expression: the top band represents glycosylated DNase-I, and the bottom band represents nonglycosylated DNase-I. Expression of glycosylated DNase-I in patient CDG-0235 is reduced compared to that in both 12F and 50F normal controls.

(B) Flow cytometry detected a lower level of ICAM-1 expression at the cell surface in the patient compared to the control.

(C) Immunoblot analysis determined an absence of ICAM-1 expression in the patient fibroblasts compared to the control fibroblasts 50F and 42F and the additional Type 1a and Type Ib CDG patient fibroblasts.

(D) IF staining detected an absence of ICAM-1 expression in patient fibroblasts compared to control fibroblasts.

A third marker is an engineered green fluorescent protein (GFP) containing ER-targeting and ER-retention signals along with an N-glycosylation site (Glyc-ER-GFP). When the glycosylation site is occupied, fluorescence is lost as a result of steric hindrance by the N-glycan. However, fluorescence appears normal when the site is unoccupied. GFP fluoresces in fibroblasts from patients with several different CDG types, indicating a glycosylation defect (Losfeld et al. in preparation). We transfected two control fibroblast cell lines, 12F and 50F, and patient fibroblasts with ER-GFP, which is an ER-retained construct without the engineered glycosylation site as a control, and Glyc-ER-GFP to assess glycosylation status. We also incubated control fibroblasts with tunicamycin 24 hr after transfection to inhibit N-glycosylation. Control fibroblasts containing Glyc-ER-GFP lacked fluorescence (Figure 3A) because the N-glycosylation site was occupied, but nonglycosylated ER-GFP fluoresced. However, the fibroblasts transfected by Glyc-ER-GFP and treated with tunicamycin produced a fluorescence pattern indicating hypoglycosylation in these cells. In parallel, patient cells transfected by Glyc-ER-GFP showed a bright fluorescence without treatment, demonstrating an N-glycosylation deficiency (Figure 3A).

Figure 3.

GFP Expression in Patient CDG-0235 Indicates a Glycosylation Deficiency

(A) ER-GFP is not glycosylated and fluoresces in controls 12F and 50F and in patient CDG-0235. Controls 12F and 50F transfected by Glyc-ER-GFP and treated with the glycosylation inhibitor tunicamycin show bright fluorescence, indicating reduced glycosylation. Patient CDG-0235 cells transfected by Glyc-ER-GFP had bright fluorescence without treatment of tunicamycin compared to the controls cells, indicating a reticular N-glycosylation deficiency consistent with a CDG-type-I defect.

(B) CDG-0235 cells were complemented with either the mock vector or the wild-type DDOST. Complementation with wild-type DDOST in patient cells reduced the amount of GFP expression compared to patient cells complemented with the mock vector, indicating that the introduction of wild-type DDOST in patient cells restores glycosylation. The error bars represent standard deviations based on an average of 5 images from each group.

To confirm whether the DDOST mutations caused protein hypoglycosylation in vitro, we complemented cells with normal DDOST cDNA. This cDNA was cloned from a testis cDNA library and engineered to have an in-frame C-terminal dsRED fusion. The resulting dsRED-DDOST was then cloned into pBABE-Puro vector. 293T cells were transfected with 10 μg of pCL-Ampho (Imgenex Corp, San Diego CA) and 10 μg of pBABE-puro, dsRED-DDOST, or an empty vector. Viral supernatant was collected, patient fibroblasts were infected, and stable cell lines were created. Using the Glyc-ER-GFP marker, we established patient cell lines by stably expressing either the mock vector or the wild-type DDOST. The complemented cell lines were then transfected with ER-GFP and Glyc-ER-GFP, and the relative fluorescence was monitored. Maximum fluorescence decreased by ∼18% in the patient cells overexpressing wild-type DDOST compared to that in the control cells (Figure 3B) (Image J software). Additionally, we evaluated ICAM-1 expression. Patient fibroblasts with wild-type DDOST had increased ICAM-1 membrane expression compared to patient fibroblasts containing the empty vector (Figure 4A). Immunoblot analysis showed that ICAM-1 expression in patient fibroblasts infected with viruses expressing the wild-type-DDOST cDNA was more than three times greater than that of fibroblasts infected with the empty vector (Figure 4B). These studies demonstrate that introduction of wild-type DDOST restored glycosylation in our patient's fibroblasts, meaning that the mutations identified in DDOST are probably pathogenic.

Figure 4.

Complementation of Patient CDG-0235 Cells with Wild-Type DDOST Restores Cell-Surface ICAM-1 Expression

(A) Patient fibroblasts complemented with wild-type-DDOST cDNA tagged with dsRed were gated into OST-positive and OST-negative groups according to red-fluorescence intensity. Patient cells complemented with wild-type DDOST (CDG235+DDOST) had significantly increased ICAM-1 expression compared to patient cells with the empty vector (CDG235).

(B) Immunoblot analysis revealed that ICAM-1 expression in patient cells complemented with wild-type-DDOST cDNA (235+DDOST) was more than three times greater than that of empty-vector (235+v)-transduced patient fibroblasts.

Here, we report a CDG patient with mutations in DDOST. This gene is localized in chromosomal region 1p36.1, contains 11 coding exons, and spans 9 kb of sequence.⁹ The protein product is a component of the OST complex and was first characterized in yeast.¹⁵ The OST complex is specific to the N-glycosylation pathway and is responsible for transferring glycans from lipid-linked oligosaccharides to proteins.^16,17 The mammalian OST complex consists of seven subunits: RPN1, RPN2, DDOST, DAD1, STT3A or STT3B, TUSC3 or MAGT1, and OST4.¹⁸

The function of DDOST has yet to be fully elucidated. Yeast strains deficient in DDOST have reduced OST activity and a destabilized OST complex,^15,19 and impaired DDOST leads to underglycosylation of glycoproteins.¹⁵ DDOST is believed to recognize the dolichol-P-P-oligosaccharide and provide the oligosaccharide donor and acceptor binding sites that are critical for enzyme activity.^20,21

Four patients with nonsyndromic intellectual disability have been identified with mutations in the OST subunits TUSC3 (MIM 601385) and MAGT1 (MIM 300715).²² Interestingly, these patients did not show hypoglycosylation of transferrin, suggesting that either brain-specific glycoproteins are affected or the impact on transferrin was undetectable.²² In 50 other untyped CDG patients, we found no variants that were predicted to impair DDOST function. Given that mutations within the OST complex result in a fairly mild phenotype and can result in intellectual disability, this gene might be of interest in patients with intellectual disability.

Our WES-analysis pipeline first focuses on a library of 250 glycosylation genes. Analysis of the types of variants detected in the glycan-gene library and whether they are present in dbSNP or the 1000 Genomes Project can aid in the selection of candidate variants for Sanger-sequencing confirmation. The 22 bp deletion detected in DDOST along with the missense mutation led us to focus on this gene. Candidates identified with WES must be confirmed by functional biochemical tests or by surrogate markers for a glycosylation deficiency in patient cells. In this study, we used our recently developed glycosylation makers to show that complementing patient fibroblasts with wild-type-DDOST cDNA substantially restored glycosylation. These markers, ICAM-1 and Glyc-ER-GFP, can be used in general for the assessment of complementation in cells that lack full glycosylation site occupancy, regardless of the putative gene defect identified by WES.

More patients will probably be found with defects in the other OST subunits. It will be interesting to see whether these patients can be detected by the transferrin screening test, given that this test produced normal results for the previously reported OST-subunit defects (TUSC3-CDG and MAGT1-CDG).²² Finally, patients with OST-subunit defects can give us insight into the importance of each subunit to OST function on the basis of their associated phenotypes. In accordance with the recently introduced CDG nomenclature, we label this disease DDOST-CDG.²³

Web Resources

The URLs for data presented herein are as follows:

• Software PowerTools for Genetic Analysis, www.softgenetics.com

• UCSC Genome Browser, http://genome.ucsc.edu/