Integrating glycomics and genomics uncovers SLC10A7 as essential factor for bone mineralization by regulating post-Golgi protein transport and glycosylation

Integrating glycomics and genomics uncovers SLC10A7 as essential factor for bone mineralization... Abstract Genomics methodologies have significantly improved elucidation of Mendelian disorders. The combination with high-throughput functional–omics technologies potentiates the identification and confirmation of causative genetic variants, especially in singleton families of recessive inheritance. In a cohort of 99 individuals with abnormal Golgi glycosylation, 47 of which being unsolved, glycomics profiling was performed of total plasma glycoproteins. Combination with whole-exome sequencing in 31 cases revealed a known genetic defect in 15 individuals. To identify additional genetic factors, hierarchical clustering of the plasma glycomics data was done, which indicated a subgroup of four patients that shared a unique glycomics signature of hybrid type N-glycans. In two siblings, compound heterozygous mutations were found in SLC10A7, a gene of unknown function in human. These included a missense mutation that disrupted transmembrane domain 4 and a mutation in a splice acceptor site resulting in skipping of exon 9. The two other individuals showed a complete loss of SLC10A7 mRNA. The patients’ phenotype consisted of amelogenesis imperfecta, skeletal dysplasia, and decreased bone mineral density compatible with osteoporosis. The patients’ phenotype was mirrored in SLC10A7 deficient zebrafish. Furthermore, alizarin red staining of calcium deposits in zebrafish morphants showed a strong reduction in bone mineralization. Cell biology studies in fibroblasts of affected individuals showed intracellular mislocalization of glycoproteins and a defect in post-Golgi transport of glycoproteins to the cell membrane. In contrast to yeast, human SLC10A7 localized to the Golgi. Our combined data indicate an important role for SLC10A7 in bone mineralization and transport of glycoproteins to the extracellular matrix. Introduction Protein glycosylation is an important biological process that is located in the secretory pathway of the cell. The majority of human diseases has been associated with abnormal protein glycosylation via glycomics profiling in plasma, however, the underlying mechanisms remain poorly understood. Congenital Disorders of Glycosylation (CDG) form an emerging group of >100 genetic disorders with a highly heterogeneous clinical spectrum. CDG with abnormal protein N-glycosylation can be identified via analysis of transferrin glycosylation in plasma, covering >50 subtypes. Subtypes with abnormal N-glycan structures are caused by deficient Golgi N-glycosylation, such as defects in glycosyltransferases (e.g. B4GALT1-CDG, MAN1B1-CDG) and in the supply of nucleotide-sugars (e.g. SLC35A2-CDG). Analysis of glycan structures in such defects is directly related to the underlying gene defect. Another major subgroup comprises genetic defects that disrupt Golgi homeostasis, for example defects in ion and pH maintenance and vesicular transport. The large number of genes involved in Golgi homeostasis, estimated about 5–10% of our genome, and their unknown influence on protein glycosylation complicate the identification of disease genes associated with abnormal glycosylation. Strategies for disease gene identification have evolved in recent years by introduction of high-throughput sequencing technologies as whole exome (1) and whole genome (2) sequencing. Success rates vary from roughly ∼20–50%, depending on the approach and clinical subgroup studied (3,4). Identification of recessive traits can be quite challenging, especially in isolated cases. Detailed phenotyping and transcriptome profiling (5–7) have been added to improve disease gene identification. The combination of genomics with high-throughput mass spectrometry-based omics methodologies can potentially further increase the diagnostic yield and, even more importantly, will contribute to the functional annotation of unknown genes. Since glycomics reflects the status of protein glycosylation in the secretory pathway of the cell (8), integration of glycomics and genomics is a promising strategy to unravel novel genetic factors for Golgi homeostasis (9). We here performed comparative glycomics profiling in a cohort of 99 patients with abnormal Golgi glycosylation, 47 with unsolved defect. Combination with exome sequencing in 31 cases resolved 15 patients with a known gene defect and led to identification of SLC10A7 as a novel genetic factor for Golgi homeostasis and glycoprotein trafficking. Molecular studies in cells and zebrafish indicate an important role for SLC10A7 in bone mineralization by influencing glycosylation and transport of proteoglycans and glycoproteins to the extracellular matrix. Results Integrating glycomics and whole-exome sequencing To explore the contribution of glycomics for gene identification, a cohort was selected of 99 patients with CDG, characterized by abnormal N-glycosylation of the plasma protein transferrin. A causative gene defect was found before in 52 patients, while 47 patients remained undiagnosed at the start of this study. The cohort was divided in four subgroups on basis of additional analysis of mucin type O-glycosylation of plasma apolipoprotein CIII (10) (Fig. 1A). Detailed information on protein-linked N-glycans can be obtained by glycomics, i.e. the mass spectrometry (MS)-based analysis of N-glycans that are released from proteins by the enzyme peptide N-glycosidase F (PNGaseF). Commonly, this is performed on total blood plasma proteins, while targeted analysis of individual glycoproteins can be more specific such as shown for transferrin (11,12) and immunoglobulins (13). We performed glycomics analysis of total plasma N-glycans (14) and of plasma purified intact transferrin (15) in the entire cohort of patients (Fig. 1B and Supplementary Material, Tables S1–S3). The aim was to identify glycosylation signatures that allow subgrouping of patients for targeted gene or whole-exome sequencing. Hierarchical clustering (HCL) of total plasma N-glycans (Fig. 1B) showed three major clusters of (i) B4GALT1, (ii) MGAT2 and (iii) MAN1B1. HCL of the intact transferrin glycomics data (Supplementary Material, Fig. S1) exhibited additional groups (iv) TMEM165 and SLC39A8 (15,16) and (v) SLC35A2 of known defects. Targeted sequencing of unsolved cases within these groups resulted in a novel mutation in B4GALT1 (P71), and in SLC39A8 (P44), and a heterozygous mutation in X-linked SLC35A2 (17) (female patient P62). No DNA was available of P45 for confirmation. Sufficient DNA was available of 31 patients for exome sequencing, 29 of which being singleton families. Filtering was done on basis of suspected recessive inheritance for known CDG genes. A mutation was found in TMEM165 in P69, which did not cluster in group 4 (TMEM165), likely because of the very mild glycosylation abnormalities. Manual inspection of the transferrin data for patient 63 revealed a very mild but identical profile as found in group 5 (SLC35A2), subsequently resulting in the identification of a mosaic mutation in SLC35A2 in the exome data (17). In addition, nine patients were identified with a defect in the Conserved-Oligomeric Golgi (COG) complex, and a novel case with mutations in SLC35A3 that clustered in the group of MGAT2-CDG (overview of all identified mutations in Table 1). All COG patients showed a characteristic Golgi homeostasis defect with loss of galactose and sialic acid and all but P80 were in the ApoCIII-0 subgroup (Fig. 1A). Together, these data indicate that glycomics profiles can be diagnostic for defects in glycosyltransferases and can be helpful in guiding the diagnostic process for other CDG subtypes. Table 1. Overview of identified mutations in known CDG disease genes Patient Gene GRCh37 Mutations SIFTa Align-GVGDb Mutation tester PolyPhen-2 P71 B4GALT1 Chr9 c.784T>C; p.(Cys262Arg) Deleterious Class C65 Disease causing Probably damaging P44 SLC39A8 Chr4 c.239T>C; p.(Ile80Thr) Tolerated Class C0 Disease causing Probably damaging P69 TMEM165 Chr4 c.376C>T; p.(Arg126Cys) Deleterious Class C65 Disease causing Probably damaging P19 SLC35A3 Chr1 c.200G>T; p.(Arg67Leu) Deleterious Class C0 Disease causing Probably damaging P27 COG1 Chr17 c.1070 + 5G>A; p.(Met357Ilefs*1) — — — — P73 COG1 Chr17 c.1070 + 5G>A; p.(Met357Ilefs*1) — — — — P78 COG2 Chr1 c.380del; p.(Lys127Argfs*5) — — — — c.389T>A; p.(Val130Glu) Deleterious Class C35 Disease causing Probably damaging P14 COG5 Chr7 c.2197del; p.(Val733Trpfs*11) — — — — P80 COG5 Chr7 c.2171_2173del; p.(Ala724del) — — — — P85 COG5 Chr7 c.704_706delinsTAG; — — — — p.(Ala235_Arg236delinsValGly) c.2338_2340dup; p.(Ile780dup) — — — — P30 COG5 Chr7 c.706C>T; p.(Arg236*) — — — — c.2324C>T; p.(Pro775Leu) Deleterious Class C0 Disease causing Probably damaging P38 COG6 Chr13 c.1075-9T>G; p.(Arg360_Val429del) — — — — P70 COG6 Chr13 c.1167-24A>G; p.Gly390Phefs*6 — — — — P62 SLC35A2 ChrX c.262G>C; p.(Ala88Pro) Tolerated Class C0 Disease causing Probably damaging P63 SLC35A2 ChrX c.818G>A; p.(Gly273Asp) Tolerated Class C0 Disease causing Probably damaging Patient Gene GRCh37 Mutations SIFTa Align-GVGDb Mutation tester PolyPhen-2 P71 B4GALT1 Chr9 c.784T>C; p.(Cys262Arg) Deleterious Class C65 Disease causing Probably damaging P44 SLC39A8 Chr4 c.239T>C; p.(Ile80Thr) Tolerated Class C0 Disease causing Probably damaging P69 TMEM165 Chr4 c.376C>T; p.(Arg126Cys) Deleterious Class C65 Disease causing Probably damaging P19 SLC35A3 Chr1 c.200G>T; p.(Arg67Leu) Deleterious Class C0 Disease causing Probably damaging P27 COG1 Chr17 c.1070 + 5G>A; p.(Met357Ilefs*1) — — — — P73 COG1 Chr17 c.1070 + 5G>A; p.(Met357Ilefs*1) — — — — P78 COG2 Chr1 c.380del; p.(Lys127Argfs*5) — — — — c.389T>A; p.(Val130Glu) Deleterious Class C35 Disease causing Probably damaging P14 COG5 Chr7 c.2197del; p.(Val733Trpfs*11) — — — — P80 COG5 Chr7 c.2171_2173del; p.(Ala724del) — — — — P85 COG5 Chr7 c.704_706delinsTAG; — — — — p.(Ala235_Arg236delinsValGly) c.2338_2340dup; p.(Ile780dup) — — — — P30 COG5 Chr7 c.706C>T; p.(Arg236*) — — — — c.2324C>T; p.(Pro775Leu) Deleterious Class C0 Disease causing Probably damaging P38 COG6 Chr13 c.1075-9T>G; p.(Arg360_Val429del) — — — — P70 COG6 Chr13 c.1167-24A>G; p.Gly390Phefs*6 — — — — P62 SLC35A2 ChrX c.262G>C; p.(Ala88Pro) Tolerated Class C0 Disease causing Probably damaging P63 SLC35A2 ChrX c.818G>A; p.(Gly273Asp) Tolerated Class C0 Disease causing Probably damaging a SIFT: Sorting Intolerant From Tolerant. b GVGD: Grantham Variation Grantham Deviation; classified from most likely to interfere with function (C65) to least likely (C0). Table 1. Overview of identified mutations in known CDG disease genes Patient Gene GRCh37 Mutations SIFTa Align-GVGDb Mutation tester PolyPhen-2 P71 B4GALT1 Chr9 c.784T>C; p.(Cys262Arg) Deleterious Class C65 Disease causing Probably damaging P44 SLC39A8 Chr4 c.239T>C; p.(Ile80Thr) Tolerated Class C0 Disease causing Probably damaging P69 TMEM165 Chr4 c.376C>T; p.(Arg126Cys) Deleterious Class C65 Disease causing Probably damaging P19 SLC35A3 Chr1 c.200G>T; p.(Arg67Leu) Deleterious Class C0 Disease causing Probably damaging P27 COG1 Chr17 c.1070 + 5G>A; p.(Met357Ilefs*1) — — — — P73 COG1 Chr17 c.1070 + 5G>A; p.(Met357Ilefs*1) — — — — P78 COG2 Chr1 c.380del; p.(Lys127Argfs*5) — — — — c.389T>A; p.(Val130Glu) Deleterious Class C35 Disease causing Probably damaging P14 COG5 Chr7 c.2197del; p.(Val733Trpfs*11) — — — — P80 COG5 Chr7 c.2171_2173del; p.(Ala724del) — — — — P85 COG5 Chr7 c.704_706delinsTAG; — — — — p.(Ala235_Arg236delinsValGly) c.2338_2340dup; p.(Ile780dup) — — — — P30 COG5 Chr7 c.706C>T; p.(Arg236*) — — — — c.2324C>T; p.(Pro775Leu) Deleterious Class C0 Disease causing Probably damaging P38 COG6 Chr13 c.1075-9T>G; p.(Arg360_Val429del) — — — — P70 COG6 Chr13 c.1167-24A>G; p.Gly390Phefs*6 — — — — P62 SLC35A2 ChrX c.262G>C; p.(Ala88Pro) Tolerated Class C0 Disease causing Probably damaging P63 SLC35A2 ChrX c.818G>A; p.(Gly273Asp) Tolerated Class C0 Disease causing Probably damaging Patient Gene GRCh37 Mutations SIFTa Align-GVGDb Mutation tester PolyPhen-2 P71 B4GALT1 Chr9 c.784T>C; p.(Cys262Arg) Deleterious Class C65 Disease causing Probably damaging P44 SLC39A8 Chr4 c.239T>C; p.(Ile80Thr) Tolerated Class C0 Disease causing Probably damaging P69 TMEM165 Chr4 c.376C>T; p.(Arg126Cys) Deleterious Class C65 Disease causing Probably damaging P19 SLC35A3 Chr1 c.200G>T; p.(Arg67Leu) Deleterious Class C0 Disease causing Probably damaging P27 COG1 Chr17 c.1070 + 5G>A; p.(Met357Ilefs*1) — — — — P73 COG1 Chr17 c.1070 + 5G>A; p.(Met357Ilefs*1) — — — — P78 COG2 Chr1 c.380del; p.(Lys127Argfs*5) — — — — c.389T>A; p.(Val130Glu) Deleterious Class C35 Disease causing Probably damaging P14 COG5 Chr7 c.2197del; p.(Val733Trpfs*11) — — — — P80 COG5 Chr7 c.2171_2173del; p.(Ala724del) — — — — P85 COG5 Chr7 c.704_706delinsTAG; — — — — p.(Ala235_Arg236delinsValGly) c.2338_2340dup; p.(Ile780dup) — — — — P30 COG5 Chr7 c.706C>T; p.(Arg236*) — — — — c.2324C>T; p.(Pro775Leu) Deleterious Class C0 Disease causing Probably damaging P38 COG6 Chr13 c.1075-9T>G; p.(Arg360_Val429del) — — — — P70 COG6 Chr13 c.1167-24A>G; p.Gly390Phefs*6 — — — — P62 SLC35A2 ChrX c.262G>C; p.(Ala88Pro) Tolerated Class C0 Disease causing Probably damaging P63 SLC35A2 ChrX c.818G>A; p.(Gly273Asp) Tolerated Class C0 Disease causing Probably damaging a SIFT: Sorting Intolerant From Tolerant. b GVGD: Grantham Variation Grantham Deviation; classified from most likely to interfere with function (C65) to least likely (C0). Figure 1. View largeDownload slide Integrating glycomics and genomics for gene identification. (A) Representation of the patient cohort and work flow. Following identification of abnormal glycosylation during classical CDG screening, 99 patients were subgrouped on basis of apoCIII isofocusing for analysis of mucin type O-glycosylation defects. (B) HCL based on Pearson’s correlations and heatmap of glycomics changes in 99 patients as compared to 40 controls. Figure 1. View largeDownload slide Integrating glycomics and genomics for gene identification. (A) Representation of the patient cohort and work flow. Following identification of abnormal glycosylation during classical CDG screening, 99 patients were subgrouped on basis of apoCIII isofocusing for analysis of mucin type O-glycosylation defects. (B) HCL based on Pearson’s correlations and heatmap of glycomics changes in 99 patients as compared to 40 controls. For patient 19, SLC35A3 appeared in the candidate gene list after exome sequencing. A previously reported SLC35A3 deficient patient with autism spectrum disorder showed normal plasma glycosylation and a reduction in tetra-antennary N-glycans in fibroblasts (18), while glycomics of total plasma N-glycans was altered in one patient with epilepsy and skeletal dysplasia (19) and was not tested in another case (20). CDG screening was not performed. The transferrin data of P19 showed a defect in the addition of N-acetyl-glucosamine (GlcNAc) to protein N-glycans (Supplementary Material, Fig. S2), in agreement with SLC35A3 being the Golgi transporter of Uridine 5′-diphosphate (UDP)-GlcNAc. UDP-GlcNAc is required for the formation of complex type N-glycans and additional N-glycan branching by addition of GlcNAc monosaccharides. Total plasma glycomics confirmed the defect in N-glycan branching with a major increase in bi-antennary glycans lacking one GlcNAc residue (glycans 4–3-0–1 and 4–3-1–1) and high mannose glycans (glycan 3–2-0–0, 4–2-0–0 and 5–2-0–0), as well as a major reduction of multi-antennary glycans (Supplementary Material, Table S4). In summary, 15 cases were solved with a known gene defect. Identification of SLC10A7 as novel genetic factor for abnormal Golgi glycosylation In the remaining group of 19 unsolved cases, only one sib pair (P17, P33) was available for combined interpretation of the exome data. Filtering of exome data was performed as described before in (21) and further candidate gene selection was done on basis of a model for autosomal recessive inheritance. Common and non-causative SNPs (frequency > 0.05%), deep intronic and synonymous variants were excluded. Using strict filtering criteria, six genes in P17 (three homozygous and three compound heterozygous) and four genes in P33 (two homozygous and two compound heterozygous) were selected respectively. Only two candidate genes remained that were present in both patients. These included a compound heterozygous mutation in SHROOM3 and a compound heterozygous mutation in SLC10A7. Segregation analysis showed that both variants in SHROOM3 (c.3649G>A and c.2429G>A), Supplementary Material, Table S5) resided on the same allele, thereby rejecting this variant as causative. SLC10A7 contained one genetic variant that was predicted to affect a splice acceptor site and was of paternal origin (Chr4[GRCh37]: g.147214148T>C; NM_001029998.5[SLC10A7]: c.722–16A>G). In addition, a missense mutation was identified of maternal origin (Chr4[GRCh37]: g.147425062C>T; NM_001029998.5[SLC10A7]: c.335G>A; p.[Gly112Asp]). To confirm the effect on splicing, mRNA was isolated from patient fibroblasts and analyzed by RT-PCR and Sanger sequencing, showing skipping of exon 9, which results in a frameshift mutation at position 241 and a premature stop codon (p.Ile241Argfs29*) (Fig. 2A). The missense mutation resulted in a substitution of a highly conserved apolar amino acid, glycine in human to alanine in yeast (Supplementary Material, Fig. S3), for a charged amino acid, aspartic acid, at position 112. SLC10A7 is a membrane protein with 10 predicted transmembrane domains (TMD). Using transmembrane prediction software (TMHMM), it was predicted that the missense substitution (p.Gly118Asp) in TMD4 altered the hydrophobicity profile and completely skipped out TMD4 ((Supplementary Material, Fig. S4). Figure 2. View largeDownload slide Identification of SLC10A7 as novel disease gene. (A) HCL within the ApoCIII-0 subgroup (Supplementary Material, Fig. S6) indicates four patients with similar glycomics signature, two of which with mutations in SLC10A7. Genetic studies uncover genetic deficiency of SLC10A7 in all patients. (B, C) Linear regression analysis to identify characteristic glycan abnormalities for SLC10A7-CDG. (D) Heatmap of characteristic N-glycans in SLC10A7-CDG as compared with the 11 patients in the linear regression model in Figure 2C. Figure 2. View largeDownload slide Identification of SLC10A7 as novel disease gene. (A) HCL within the ApoCIII-0 subgroup (Supplementary Material, Fig. S6) indicates four patients with similar glycomics signature, two of which with mutations in SLC10A7. Genetic studies uncover genetic deficiency of SLC10A7 in all patients. (B, C) Linear regression analysis to identify characteristic glycan abnormalities for SLC10A7-CDG. (D) Heatmap of characteristic N-glycans in SLC10A7-CDG as compared with the 11 patients in the linear regression model in Figure 2C. To identify additional patients with potential causative mutations in SLC10A7, the raw exome data of all unsolved patients were inspected, however, without success. We then attempted to identify patients with overlapping glycomics signatures. Principal component analysis (PCA) allowed to discriminate only the most clear glycosylation defects like MAN1B1- and B4GALT1-CDG (Supplementary Material, Fig. S5). HCL was then performed of the glycomics data of total plasma proteins in the full patient cohort using both Spearman and Pearson correlations, and the methods were validated by proper clustering of MAN1B1- and B4GALT1-CDG patients (Fig. 1B). Pearson correlation turned out to be the preferred method, which was also supported by the results from PCA clustering (Supplementary Material, Fig. S5). Subsequently, HCL of total plasma glycomics was done within the four ApoCIII subgroups, as defined in Figure 1. Within the ApoCIII-0 group of P17 and P33, two other patients (P32 and P39) were clustered in the same group (Fig. 2A and Supplementary Material, Fig. S6). All four patients showed an increase in glycans 4–3-0–1 (P = 2.7 × 10−25) and 4–3-1–1 (P = 6.7 × 10−14), which were the two most significantly different glycans as compared with controls. Sanger sequencing of all coding regions of SLC10A7, its promoter sequence and potential intragenic regulatory elements (see Supplementary Material, Methods) did not reveal any genetic variant of possible pathogenicity in patients 32 and 39. Subsequently, multiplex ligation-dependent probe analysis (MLPA) was performed to find evidence of a deletion in this gene; however, no large deletions or insertions were found in the patients on the tested positions of the probes (Supplementary Material, Tables S6–S8). In view of the strong glycomics similarity between these four patients, mRNA was isolated from patient fibroblasts and SLC10A7 cDNA was sequenced (Fig. 2A). For patients P17 and P33, besides exon 9, a signal was detected in the cDNA electropherograph in accordance with the sequence of exon 10. However, there seemed to be more c.335G>A containing mRNA than mRNA with the c.722–16A>G variant, which could be explained by nonsense-mediated mRNA decay of the mRNA harboring the exon 9 skip. The cDNA primers for exon 7–12 showed two bands in the control, belonging to the two transcripts NM_001029998.5 and NM_001200842.2, which are both expressed in fibroblasts. In addition, a third vague band was detected, according to the heterozygous skipping of exon 9. Interestingly, the cDNA of SLC10A7 was completely absent for P32 and P39 in multiple individual experiments (Fig. 2A). Subsequently, quantitative PCR (qPCR) analysis was performed, confirming the absence of cDNA. Culturing of skin fibroblasts in the presence of cycloheximide to inhibit nonsense-mediated mRNA decay showed additional bands for patients 17 and 33, indicating the presence of multiple unstable mRNA variants. For patients 32 and 39, no bands were detected even after incubation with cycloheximide (Supplementary Material, Fig. S7). Although several genetic mechanisms could still explain the absence of SLC10A7 cDNA, such as deep-intronic mutations or distant regulatory elements, these data clearly show a complete loss of function of the SLC10A7 protein. Diagnostic glycomics profile for SLC10A7-CDG In view of the importance of specific plasma glycomics signatures for facile diagnostics of future patients, we examined the possibility to identify a characteristic glycan pattern for classification of SLC10A7-CDG within the entire cohort of patients. The 4–3-0–1 and 4–3-1–1 glycans were able to discriminate the four SLC10A7-CDG patients from controls (Fig. 2B). Six patients with a known defect [TMEM165 (P69, P94); SLC39A8 (P44), ATP6V1A1 (P68), ATPV0A2 (P57), and MAN1B1 (P87)] and five patients with unsolved defect (P20, P34, P45, P66 and P72) showed similar values (95% CI). Comparative analysis of SLC10A7-CDG with this group of 11 patients showed that addition of glycans 5–4-0–1 (monosialylated N-glycan) and 9–2-0–0 (high-mannose glycan) allowed complete discrimination of SLC10A7-CDG (Fig. 2C). In Figure 2D, a heatmap of selected N-glycans is shown throughout the Golgi glycosylation pathway, indicating that SLC10A7 mutations affect different stages of protein N-glycosylation. In summary (Supplementary Material, Table S9), SLC10A7-CDG is characterized by decreased sialylation (49%, controls 69–71%), an increase of high-mannose glycans (14%, controls 6–7%) and a characteristic increase in glycans lacking GlcNAc (2.3%, controls 0.7–0.8%). A closer inspection of the transferrin glycoforms also revealed the presence of minor isoforms with GlcNAc-lacking N-glycans, in addition to the reduction of sialic acid (Supplementary Material, Fig. S8). As a first indication that the identified glycosylation abnormalities can be useful for diagnostics of future patients, we studied the plasma sample of a novel patient, recently identified via diagnostic exome-sequencing with a splice mutation in SLC10A7 (Chr4[GRCh37]: g.147431202C>T, NM_001029998.5[SLC10A7]: c.184–1G>A). Analysis of ApoCIII mucin type O-glycosylation showed an ApoCIII-0 profile while N-glycosylation of transferrin and total plasma N-glycans showed the characteristic abnormalities for SLC10A7-CDG (see Supplementary Material, Table S9 and Supplementary Material, Fig. S8). SLC10A7 is essential for teeth and skeletal development Patients with SLC10A7 deficiency share an overlapping clinical phenotype, characterized by short stature, defective enamel formation (amelogenesis imperfecta), skeletal dysplasia, facial dysmorphism, moderate hearing impairment and mildly impaired intellectual development (Fig. 3, Supplementary Material, Table S10). Figure 3. View largeDownload slide Clinical and radiological images of SLC10A7 deficient patients. Patient images (upper panel). Clinical photograph of male patient P17 (A, B) and of the teeth (C, D) at the age of 31 years. Note the left sided ptosis, prognathism, severe chest deformity, scoliosis and short stature, genua valga and abnormally shaped teeth with marked enamel hypoplasia and dental caries. In contrast to her brother, the lateral picture of the face of patient 33 showed mandibular hypoplasia (micrognathia) (E, F). Images of patients P32 and P39 are shown in (G–I), respectively. Radiographs (lower panel). Radiograph of the wrists and hands of P17 (J) showing brachydactyly with short and robust carpal and metacarpal bones and phalanges. Note the prolonged and mildly abnormal shaped distal ulnar end in the left hand X-ray. Radiograph of the lumbar spine, pelvis and hip of the female patient 33 (aged 33 years, K) and the male patient 17 (aged 31 years, L) with severe convexity lumbar scoliosis with hypoplastic and abnormally shaped pelvis (dominantly inferior pubic rami hypoplasia). Note also the narrow hip joint space, flatten femoral heads and short femoral necks. X-rays of P17 of the AP chest (M), lumbar (N) and lateral (O) spine with severe right convexity thoraco-lumbar scoliosis. There is no evidence of vertebral body deformity or fracture. Figure 3. View largeDownload slide Clinical and radiological images of SLC10A7 deficient patients. Patient images (upper panel). Clinical photograph of male patient P17 (A, B) and of the teeth (C, D) at the age of 31 years. Note the left sided ptosis, prognathism, severe chest deformity, scoliosis and short stature, genua valga and abnormally shaped teeth with marked enamel hypoplasia and dental caries. In contrast to her brother, the lateral picture of the face of patient 33 showed mandibular hypoplasia (micrognathia) (E, F). Images of patients P32 and P39 are shown in (G–I), respectively. Radiographs (lower panel). Radiograph of the wrists and hands of P17 (J) showing brachydactyly with short and robust carpal and metacarpal bones and phalanges. Note the prolonged and mildly abnormal shaped distal ulnar end in the left hand X-ray. Radiograph of the lumbar spine, pelvis and hip of the female patient 33 (aged 33 years, K) and the male patient 17 (aged 31 years, L) with severe convexity lumbar scoliosis with hypoplastic and abnormally shaped pelvis (dominantly inferior pubic rami hypoplasia). Note also the narrow hip joint space, flatten femoral heads and short femoral necks. X-rays of P17 of the AP chest (M), lumbar (N) and lateral (O) spine with severe right convexity thoraco-lumbar scoliosis. There is no evidence of vertebral body deformity or fracture. Patient P33 is a 33.5-year-old woman with short stature (−5.3 SD), amelogenesis imperfecta, skeletal dysplasia, severe scoliosis, genua valga, pedes planovalgi, mandibular hypoplasia, inguinal hernia, moderate hearing impairment (bilateral hypacusis) and subaverage intellectual functioning. Bone mineral density measurement in the area of the proximal femur and femoral neck clearly revealed decreased bone mineral mass, compatible with osteoporosis (Z-scores −2.9 and −3.6, respectively). Her brother (P17), 31-year-old, has a short stature (−7.0 SD), amelogenesis imperfecta, skeletal dysplasia, prognathism, severe scoliosis, genua valga, pedes planovalgi, submucous cleft palate, left sided myopia gravis, strabismus convergens, unilateral left sided ptosis, moderate hearing impairment (bilateral hypacusis) and subaverage intellectual functioning (IQ 70–75). Patient 39 is an adopted male patient of 8 years with short stature (−4.0 SD), dental decay with discolored enamel and dental crowding, failure to thrive, developmental delay, and mild conductive hearing loss on the right. A skeletal survey at age 6 showed mild skeletal anomalies with irregularity of the anterosuperior endplates of a few vertebral bodies near the thoracolumbar junction, mild bilateral coxa valga, and a hypoplastic right first rib. Dysmorphic features included: frontal bossing, dolichocephaly, flat and broad nasal bridge, long and relatively smooth philtrum, and mildly thin upper lip. The patient is prone to frequent, severe respiratory infections, one of which led to sepsis, and also suffers from chronic malabsorption. He continues to demonstrate poor growth despite being fed via G-tube. P32 is a 24-year male patient with short stature (<−3.0 SD), amelogenesis imperfecta with lack of enamel, skeletal dysplasia, inguinal hernia, global developmental delay, joint hypermobility, clubfeet, phimosis, shawl scrotum, and glaucoma. Facial dysmorphism consisted of dry brittle hair, long philtrum, a webbed neck and thin upper lip. Radiological analysis revealed skeletal dysplasia with cervical stenosis, kyphoscoliosis and pectus excavatum. The newly identified patient is a 12-month-old boy with a disproportionate short stature (−5, 4 SD) with short limbs, head circumference at −2, 1 SD, small thorax, feeding difficulties, inguinal hernia, hypermobile short joints and dysmorphic features (short neck, round face, micrognathia, hypertelorism, prominent eyes and long philtrum). He was born at term with a birth weight of 3032 g (p30), length at 41 cm (≪p3), and a head circumference of 33, 8 cm (p30). A skeletal survey one day after birth showed short tubular bones and irregular endplates with central indentations and (the suggestion of) coronal clefts of several lumbar vertebral bodies. Ophthalmologic examination and newborn hearing screening were normal. There is some gross motor delay. Cognitive development so far is normal. A zebrafish model recapitulates the human skeletal defect To further study the function of SLC10A7 in vivo, a zebrafish model was generated by morpholino knockdown of slc10a7. We designed a sp morpholino (sp MO) (slc10a7-e2i1) targeting the sp donor site of exon 2. The slc10A7-sp MO was injected into one-cell stage zebrafish embryos at two different concentrations 8 and 12 ng/nl. To visualize the skeletal development, Alcian blue staining was performed, which corresponds with the expression of glycosaminoglycans. In morphants injected with 8ng/nl of slc10a7-sp, Alcian blue cartilage staining revealed a relatively normal number of cerathobranchials (c1–c5) and a normal fin bud cartilage. However, while the teeth cartilage is straight in embryos injected with a control MO, it appears to be bent downwards in morphants. Injection of 12 ng/nl results in a severe phenotype with edema in the whole body, reduced head, eyes and curled body. Alcian blue staining revealed a reduced Meckel’s cartilage (m) and absence of cerathobranchial number four (c4) (Fig. 4). Figure 4. View largeDownload slide Skeletal phenotype in slc10a7 deficient zebrafish with Alcian blue staining (cartilage) on 6 dpf embryos. slc10A7 sp MO or a control morpholino (Cont MO) were injected at 8 and 12 ng/nl. (A–C) Lateral view of whole embryo. (A’–C’) close-up of head lateral. (A”–C”) close-up of head anterior view. (A”’–C”’) close-up of head dorsal view of bright field image showing cartilage staining. Figure 4. View largeDownload slide Skeletal phenotype in slc10a7 deficient zebrafish with Alcian blue staining (cartilage) on 6 dpf embryos. slc10A7 sp MO or a control morpholino (Cont MO) were injected at 8 and 12 ng/nl. (A–C) Lateral view of whole embryo. (A’–C’) close-up of head lateral. (A”–C”) close-up of head anterior view. (A”’–C”’) close-up of head dorsal view of bright field image showing cartilage staining. Slc10a7 zebrafish morphants display a defect in bone mineralization In view of the teeth decay and skeletal abnormalities in SLC10A7 patients, we further studied skeletal mineralization, by staining with alizarin red which reacts with calcium deposits in tissues (22). Mineralized bone was stained by alizarin red on 6 dpf embryos injected with slc10A7-sp at 8 ng/nl and 12 ng/nl. Morphants injected at 8 ng/nl had a shortened or absent palate cartilage (pc), absent operculum, widened palatal skeleton (ps), reduced cleithrum (c), notochord (n), fifth cerathobranchial (c5), and entopterygoid (en), as compared with control MO-injected embryos. Importantly, injection of higher dose slc10A7-sp MO at 12 ng/nl concentration caused a severe phenotype, without any detectable bone mineralization (Fig. 5). These results support an essential function for SLC10A7 in cartilage development and bone mineralization. Figure 5. View largeDownload slide Bone mineralization in slc10a7 deficient zebrafish by Alizarin red staining on 6 dpf embryos. slc10A7 sp MO, or a Cont MO were injected at 8 and 12 ng/nl. (A–E) Lateral and anterior views of confocal image stacks showing mineralized bone in red in the context of the embryo. (A’–E’) Stack of serial confocal images showing mineralized bone. (A”–D”) 3D renderings from serial confocal images showing mineralized bone. pc, palate cartilage; ps, palatal skeleton; c, cleithrum; n, notochord (n); c5, fifth cerathobranchial; and en, entopterygoid. Video’s are shown in Supplementary Material, Figure S9. Figure 5. View largeDownload slide Bone mineralization in slc10a7 deficient zebrafish by Alizarin red staining on 6 dpf embryos. slc10A7 sp MO, or a Cont MO were injected at 8 and 12 ng/nl. (A–E) Lateral and anterior views of confocal image stacks showing mineralized bone in red in the context of the embryo. (A’–E’) Stack of serial confocal images showing mineralized bone. (A”–D”) 3D renderings from serial confocal images showing mineralized bone. pc, palate cartilage; ps, palatal skeleton; c, cleithrum; n, notochord (n); c5, fifth cerathobranchial; and en, entopterygoid. Video’s are shown in Supplementary Material, Figure S9. SLC10A7 is localized to the secretory pathway in human cells The phenotypic presentations in human and zebrafish SLC10A7 deficiency indicate a defect in extracellular matrix mineralization. Since this process is localized to the secretory pathway of the cell, which also houses the glycosylation machinery, we first studied the localization of SLC10A7. Previous reports proposed a localization of FLAG-tagged SLC10A7 to the cell membrane in Xenopus laevis oocytes and ER in HEK293 cells (23), while a later study indicated intracellular localization of V5-tagged SLC10A7 in U2OS cells (24). In Protein Atlas, SLC10A7 was proposed in nucleoli. Recently, SLC10A7 in Saccharomyces cerevisiae and Candida albicans was found to be present in the plasma membrane (25). Using commercial antibodies, nucleoli were stained in control and patient fibroblasts. Since the SLC10A7 transcript is completely absent in two of the patients, and since western blot revealed multiple aspecific bands for both controls and patients, these results were interpreted as non-specific binding of the commercial antibodies tested (data not shown). We therefore generated a C-terminally tagged V5-SLC10A7 construct to study subcellular localization in Hela cells. After transient transfection, co-staining was performed with several organelle markers of the secretory pathway, ranging from endoplasmic reticulum to early endosomes. Co-localization was mainly observed with markers for the cis-, medial- and trans-Golgi network (Fig. 6A). To confirm similar localization in a different human cell line, fibroblasts were transfected with a lentiviral V5-SLC10A7 construct, which also showed Golgi localization (Fig. 6B). Cellular glycomics profiling of patient fibroblasts showed elevation of truncated glycans lacking GlcNAc, characteristic for SLC10A7 in plasma, in three out of the four cell lines (data not shown), indicating fibroblasts as a suitable model to study the cell biological consequences of SLC10A7 deficiency. Figure 6. View largeDownload slide Subcellular localization of SLC10A7. (A) Colocalization of V5-SLC10A7 (green) by transient transfection in Hela cells and co-staining with the indicated organelle markers (r2 values of 0.25, 0.49, 0.58 and 0.38 for Calnexin, ERGIC-53, Giantin and TGOLN2, respectively). (B) Colocalization of V5-SLC10A7 (green) by lentiviral transfection of fibroblasts and co-staining with the indicated organelle markers (r2 values of 0.11, 0.28, 0.44 and 0.47 for Calnexin, ERGIC-53, Giantin and TGOLN2, respectively). Markers (in purple) used include Calnexin (ER), ERGIC-53(ER-Golgi-Intermediate-Compartment), Giantin (cis- and medial-Golgi) and TGOLN2 (trans-Golgi network). (C) Staining of patient fibroblasts with Golgi marker TGOLN2 showed a more pronounced dilatation of the Golgi apparatus in the four SLC10A7-CDG patients compared to healthy control (Scale bar, 10 μm). Figure 6. View largeDownload slide Subcellular localization of SLC10A7. (A) Colocalization of V5-SLC10A7 (green) by transient transfection in Hela cells and co-staining with the indicated organelle markers (r2 values of 0.25, 0.49, 0.58 and 0.38 for Calnexin, ERGIC-53, Giantin and TGOLN2, respectively). (B) Colocalization of V5-SLC10A7 (green) by lentiviral transfection of fibroblasts and co-staining with the indicated organelle markers (r2 values of 0.11, 0.28, 0.44 and 0.47 for Calnexin, ERGIC-53, Giantin and TGOLN2, respectively). Markers (in purple) used include Calnexin (ER), ERGIC-53(ER-Golgi-Intermediate-Compartment), Giantin (cis- and medial-Golgi) and TGOLN2 (trans-Golgi network). (C) Staining of patient fibroblasts with Golgi marker TGOLN2 showed a more pronounced dilatation of the Golgi apparatus in the four SLC10A7-CDG patients compared to healthy control (Scale bar, 10 μm). SLC10A7 executes its effect on skeletal and bone development via impacting post-Golgi vesicular transport Staining of patient fibroblasts with organelle markers showed a dilation of the Golgi apparatus in ∼20% of the cells (Fig. 6C). In order to study the Golgi dynamics in more detail in connection with protein glycosylation, metabolic oligosaccharide engineering was performed as described in (26,27). Control and SLC10A7 deficient cells were metabolically labeled with SiaNAl for 7 h (Fig. 7A). The main pool of sialylated glycoconjugates was localized to the Golgi apparatus for all fibroblasts lines, in accordance with previous results. No obvious differences in the staining intensity were observed in SLC10A7 deficient cells when compared to control cells suggesting no clear impairment of the Golgi sialylation efficiency. In addition, a vesicular staining pattern of small vesicles was seen throughout the cell for all four SLC10A7 deficient patient fibroblasts, while this was barely visible in control cells. In order to get information on the nature of the observed vesicles, co-localization studies were performed with the early endosomal marker EEA1, the Golgi marker TMEM165 and the lysosomal marker LAMP2. Although no colocalization was observed with LAMP2 (data not shown) and TMEM165, some of the intracellular vesicular staining colocalized with EEA1 (Supplementary Material, Fig. S10), thus demonstrating that sialylated glycoconjugates accumulate at least partly in early endosomes in SLC10A7 deficient cells. Figure 7. View largeDownload slide Post-Golgi trafficking in SLC10A7 deficient fibroblasts. Fibroblasts from healthy individuals and SLC10A7 deficient patients were metabolically labeled with 500 µm of SiaNAl for 7 h (A) and chased for 48 h (B). The sialylated glycoconjugates were stained with azido-545 fluorescent probes and visualized by confocal microscopy. Scale bar, 30 µm in the upper panels and 10 µm in the zoomed panels. Figure 7. View largeDownload slide Post-Golgi trafficking in SLC10A7 deficient fibroblasts. Fibroblasts from healthy individuals and SLC10A7 deficient patients were metabolically labeled with 500 µm of SiaNAl for 7 h (A) and chased for 48 h (B). The sialylated glycoconjugates were stained with azido-545 fluorescent probes and visualized by confocal microscopy. Scale bar, 30 µm in the upper panels and 10 µm in the zoomed panels. As a further step to study glycoprotein trafficking, we designed a pulse-chase labeling experiment to study the fate of labeled glycoproteins. After a 7h pulse with SiaNAl metabolic labeling, the culture medium was replaced, followed by a chase of 48 h (Fig. 7B). In control cells, a perinuclear Golgi staining was mainly present at t = 0, while at 48 h, the fluorescence intensity was mainly detected at the plasma membrane. In contrast, SLC10A7 deficient fibroblasts hardly showed any staining of the plasma membrane and instead showed intracellular staining of vesicles and Golgi. These results indicate a post-Golgi transport defect of glycoproteins through the secretory pathway due to dysfunction of SLC10A7. In summary, our data indicate that SLC10A7 is important to transport glycoproteins and proteoglycans to produce a proper functioning extracellular matrix and for its mineralization. Discussion In this study, we showed the potential of combining high-throughput functional–omics methods with genomics for gene identification by uncovering SLC10A7 as a novel genetic factor for bone mineralization and protein sorting. Introduction of next-generation sequencing in patient diagnostics has significantly improved the diagnostic yield in genetic disease diagnostics (2). Still, interpretation of the causative nature of genetic variants in individual cases is an important bottleneck. High-throughput functional methodologies hold the potential to provide unique signatures for confirmation of genetic variants and at the same time provide a first clue for annotation of gene function and understanding disease mechanisms. Similarly, technological advances in MS have resulted in holistic methods to analyze metabolites (28,29), i.e. metabolomics, which is being implemented in rare disease diagnostics and has led to the identification of diagnostic markers for known metabolic diseases. Complementary to metabolomics, glycomics monitors the process of protein glycosylation, which reflects the secretory pathway of the cell, requiring 5–10% of our genes for processes such as vesicular transport and ion homeostasis. As such, complementary functional–omics methodologies will be required for functional analysis of our genome. Application of glycomics in our selected cohort of patients supported the genetic diagnosis of known disease genes in 15 cases. The data generated from glycomics experiments on rare genetic disorders exhibit high dimensionality with numerous variables on limited samples. PCA would be an attractive way to project the multivariate data into lower dimensions for visual interpretation. However, PCA was only successful to discriminate defects in glycosyltransferases like MAN1B1 and B4GALT1, while more subtle abnormalities were difficult to classify, such as for patients 63 (SLC35A2) and 69 (TMEM165). Moreover, information on individual glycan levels between patients and controls is lost. HCL of specific subsets of patients allowed to classify patients with more subtle glycomics changes such as SLC10A7-CDG. A few additional subgroups of unsolved patients could be identified by HCL, however, no progress in gene identification could be made due to a lack of available DNA. Nevertheless, our integrated glycomics-genomics approach shows the potential to identify additional gene defects. To determine the diagnostic value of the glycomics profiles for future patients, it will be essential to establish the specificity and sensitivity in larger patient cohorts. As individual CDG are very rare, this will require a world-wide collaborative effort and building of a CDG patient registry. This will also allow to determine the prevalence of CDG subtypes, which is ultimately required to determine the positive predictive value of the glycomics profiles as identified in our study. Since many common disorders have already been associated with abnormal plasma glycomics profiles (30), it is essential to unravel genes that are directly or indirectly involved in protein glycosylation. The plasma glycomics abnormalities in this cohort of monogenetic disorders reflect the wide range of biological mechanisms that influence protein glycosylation. The link of SLC10A7 with protein glycosylation is unexpected, since the SLC10 family has thus far been associated with sodium-bile acid transporters (31). Members within this family are involved in the transport of taurocholate (SLC10A1), bile (SLC10A2) and sulfated steroids (SLC10A6). SLC10A4 is potentially involved neurotransmitter and mastocyte mediator secretion (32,33). SLC10A3, SLC10A5 and SLC10A7 remain functionally uncharacterized (31). Although SLC10A7 shares homology with other members of the family, this membrane protein has distinctive features. In contrast to all other members of the SLC10 protein family that contain eight predicted TMD and are present in vertebrates only, SLC10A7 contains 10 hydrophobic segments and homologs were identified in bacteria (23), yeast and plants. A phylogenetic tree of sodium bile symporters shows clustering of the SLC10A1-A6 homologs, while SLC10A7 clusters with orthologues that are present in eukaryotes. Human SLC10A7 shares 19.4 and 33.7% amino acid sequence identity with S. cerevisiae ScRch1 (34) and C. albicans CaRCH1 (25) proteins, respectively. Both transporters reside in the plasma membrane and are involved in regulation of cytosolic calcium homeostasis. The phenotype of amelogenesis imperfecta and skeletal dysplasia in SLC10A7 deficiency seems mainly linked to a defect in the generation and/or mineralization of the extracellular matrix. Previously, slc10a7 knock-out mice were identified with moderate skeletal dysplasia in a screen for skeletal phenotypes (35). Several types of skeletal dysplasia have been associated with a defect in the biosynthesis of proteoglycans (36). The defect in vesicular transport as we observed in patient fibroblasts could indicate a defect in proper deposition of proteoglycans in the extracellular matrix. Amelogenesis imperfecta is not a common feature of defects in proteoglycan biosynthesis. Amelogenesis imperfecta is characterized by defective enamel formation, a process known as amelogenesis (37,38). In a secretion stage, ameloblasts secrete a protein-containing matrix together with hydroxyapatite crystals. In a maturation stage, ameloblasts are re-oriented to secrete protein-degrading enzymes and calcium ions for mineralization, resulting in the formation of the hard enamel tissue, almost completely mineralized with calcium hydroxyapatite. Amelogenesis imperfecta can be classified roughly in hypoplastic and hypomineralized forms, associated with these two respective stages of amelogenesis. The discolored teeth in some SLC10A7 patients could indicate hypomineralized amelogenesis imperfect (37). This is in line with the reduced staining of alizarin red in slc10a7 morphant zebrafish, which stains skeletal calcium deposits (22). Several possible mechanisms could position SLC10A7 in the biological pathways of enamel mineralization. In view of the proposed role of SLC10A7 homologs in S. cerevisiae and C. albicans in regulating calcium homeostasis, it is tempting to speculate that human SLC10A7 plays an important role in calcium homeostasis during enamel mineralization. In addition, calcium might play a role in the calcium-dependent regulated vesicular transport during exocytosis (39), supported by the vesicular transport defect in fibroblasts. Finally, it can’t be excluded that altered protein glycosylation in ameloblasts contributes to the hypomineralized amelogenesis imperfecta phenotype (40). In summary, we have shown that integration of glycomics and genomics facilitates identification of novel genetic factors for Golgi homeostasis. Furthermore, our data indicate an important role for SLC10A7 in teeth and skeletal mineralization via an influence on protein glycosylation and the transport of proteoglycans and glycoproteins to the extracellular matrix. Materials and Methods Materials of participating individuals Blood and, if obtained, fibroblasts of individuals were sent to the Radboud University Medical Center, Translational Metabolic Laboratory, for diagnostics of CDG. This was based on clinical suspicion for an inborn error of metabolism. All participating individuals or their legal representatives gave informed consent for exome sequencing. Tissue and samples were obtained in accordance with the Declaration of Helsinki. For publication of facial images, written informed consent was obtained from the parents. The diagnosis of all patients with known CDG gene defect was genetically and biochemically confirmed in previous studies (Supplementary Material, Tables S1–S3). For establishment of reference intervals for total plasma N-glycans, plasma samples of 40 healthy controls were received from the Sanquin Blood Bank (Nijmegen) according to their protocols of informed consent. Screening for CDG Routine CDG screening tests for protein N-glycosylation by transferrin IEF as well as for protein mucin type O-glycosylation by ApoC-III IEF were performed as described previously in (21). In brief for transferrin IEF, 10 μl of plasma was incubated with ferric citrate buffer before the diluted sample was applied to a hydrated Immobiline dry gel (Servalyt pH 5–7; GE Healthcare) and run on a Phast System (GE Healthcare). Transferrin immunoprecipitation was performed with 60 μl of polyclonal rabbit anti-human transferrin antibody (8.5 g/l; Dako) followed by washing, fixation, staining and destaining steps to visualize the transferrin isoforms. The relative amounts of transferrin isoforms were determined by densitometry (Image Scanner Amersham/Biosciences; Lab Scan 6.0 and IQTL software) and compared with established reference intervals. In short for ApoC-III IEF, 2 μl of plasma was 15 times diluted with saline solution. Before electrophoresis, the gel was rehydrated in a solution containing 8 m urea. After blotting on a nitrocellulose membrane, the blot was washed and blocked before incubation with anti-ApoC-III antibody (1:2000, Rockland, no. 600–101-114). After incubation with the secondary anti-goat-HRP antibody (1:5000, Thermo Scientific, no. 31402), the blot was visualized by chemoluminescence (ECL reagent (Pierce) on a LAS3000 imaging system (Fujifilm). The relative amounts of ApoC-III isoforms were determined by densitometry (IQTL software) and compared with established reference intervals. Whole-exome sequencing Next generation sequencing and analysis was performed as described earlier in (41). The SureSelect Human All Exon 50Mb Kit (v4, Agilent) was used for exome enrichment, covering ∼21 000 genes. The exome library was sequenced on a SOLiD 5500xl sequencer (Life Technologies). Color space reads were iteratively mapped to the hg19 reference genome with the SOLiD LifeScope software version 2.1. We used our in-house annotation pipeline for annotation of called variants and indels (42). Variants were excluded based on a frequency of >0.2% in our in-house database of >1300 exomes. Also, synonymous variants, deep intronic variants and variants in untranslated regions were excluded. Quality criteria were applied and included variants called more than five times and variation of more than 20% for heterozygous variants and 80% for homozygous variants. Sanger sequencing of gDNA and cDNA In order to confirm the SLC10A7 variants identified by next-generation sequencing and to identify potential mutations in SLC10A7 in P32 and P39, bi-directional direct Sanger sequencing was performed using specific oligonucleotide primers flanking the exons. Total DNA was extracted using the QIAamp DNA kit (Qiagen, Venlo, The Netherlands). Total RNA from cultured fibroblasts was extracted using RNAbee (AMS Biotechnology, Abingdon, UK) and transcribed into cDNA using Superscript II and random primers (Invitrogen, Breda, The Netherlands). All coding regions of SLC10A7 (NM_001029998.5 and NM_001200842.2), its promoter sequence and potential intragenic regulatory elements were amplified using AmpliTaq Gold 360 Master Mix (Fisher Scientific, Landsmeer, The Netherlands) and the primers listed in Supplementary Material, Table S6. The PCR fragments were sequenced using the BigDye Terminator Kit v1.1 (Fisher Scientific, Landsmeer, The Netherlands) on a 3130xL Genetic Analyzer with M13 primers. Multiplex ligation-dependent probe amplification MLPA was performed on the extracted DNA following a procedure from MRC-Holland (Amsterdam, The Netherlands, www.mlpa.com). Fragments were separated with GeneScan 500LIZ dye size Standard (Fisher Scientific, Landsmeer, The Netherlands) on a 3130xL Genetic Analyzer. The data was analyzed by using Coffalyser.Net software (MRC-Holland, Amsterdam, The Netherlands). High-resolution QTOF MS of plasma transferrin Transferrin was immunopurified from control and patient plasma as previously described in (15) and analyzed by MS on a microfluidics-based platform (Agilent Technologies) consisting of an Agilent 1260 nanoLC-HPLC-chip system using a C8 protein chip coupled to an Agilent 6540 QTOF LC/MS system. Data analysis was performed using Agilent Mass Hunter Qualitative Analysis Software B.05. The distribution of raw charge data was deconvoluted to reconstructed mass data using Agilent BioConfirm Software (version 5). A set of 30 transferrin glycoforms (Supplementary Material, Table S2) was selected for relative quantitation, calculated to the total abundance of the 30 selected glycoforms. High-resolution QTOF MS of total plasma N-glycans Analysis of plasma N-glycans was performed based on Kronewitter et al. (43) with minor modifications. 10 μl of plasma were mixed in equal parts with an aqueous solution of 200 mm ammonium bicarbonate and 10 mm dithiothreitol. Protein denaturation was performed in mild conditions by alternating between boiling temperature and room temperature in a water bath for six cycles of five seconds each. For enzymatic release of N-glycans, 1 μl of PNGaseF (New England Biolabs, catalog no. P0704L) was added and the mixture was incubated for 22 h at 37°C. Ethanol precipitation was performed with 80% (v/v) ethanol to remove proteins from the glycans by adding 80 μl of ethanol and the mixture was frozen at −80°C for 45 min. The mixture was centrifuged at 14 000 rpm (Eppendorf) for 20 min and the supernatant was dried in vacuo (Thermo RVT4104 Refrigerated Vapor Trap). Further purification and enrichment of glycans was performed using a graphitized carbon cartridge (Grace Davison, catalog no.G4240–64010, 150 mg, 4.0 ml), with elution of N-glycans using 2.0 ml of 40% acetonitrile and 0.05% trifluoroacetic acid (v/v) in water. MS was performed on the microfluidics-based platform as above using a porous graphitized carbon chip. Dried N-glycans were reconstituted in 50 µl pure water (per 200 nl of serum) and 1 μl sample was loaded onto the enrichment column and analyzed using the chromatographic conditions as described in (14). MS spectra were acquired in the positive ion mode over a mass range of m/z 600–2000 with an acquisition time of 1.5 s per spectrum. Mass correction was enabled using Agilent Calibrant Mix G1969–85000 with reference masses of m/z 622.029, 922.010, 1221.991 and 1521.971. Raw LC-MS data were analyzed using the Molecular Feature Extraction algorithm (Mass Hunter Qualitative Analysis Software B.05). MS peaks were filtered with a signal-to-noise ratio of 5.0 and parsed into individual ion species. All individual ion species associated with single compounds (e.g. doubly and triply protonated ions, and all associated isotopologues) were summed to create extracted ion chromatograms (ECCs) based on expected isotopic distribution and charge state information. Using a mass tolerance of 20 ppm, deconvoluted masses of each ECC peak were compared against a theoretical glycan mass library (in-house) consisting of all possible complex, hybrid and high mannose type N-glycans that have been reported in human plasma. Hence, only glycan compositions containing hexose (Hex), N-acetylhexosamine (HexNAc), deoxyhexose (dHex) and N-acetylneuraminic acid (Neu5Ac) were included. Glycan structures are indicated with the number of Hex-HexNAc-dHex-Neu5Ac residues, respectively. Relative abundances of each glycan were obtained through normalization to the total volume of all detected glycan compounds. Plasma samples of 40 healthy controls were used to establish control ranges. These samples were also analyzed by transferrin QTOF MS to confirm normal transferrin glycosylation. Statistical analysis Data were analyzed using GraphPad Prism (version 5.03) and IBM SPSS (version 22.0) software. Mean and 95% CI for total plasma N-glycans were used to express the reference intervals in 40 normal controls. For determination of P-values, student’s t-test was used for the comparison of the glycans’ relative abundance between control and other groups (e.g. SLC10A7-CDG). Regression analysis using SPSS (method = forward stepwise [conditional]) was performed on all glycans to determine linear functions and to identify a characteristic glycan profile for SLC10A7 deficiency. HCL and bioinformatics 2D HCL was performed on the glycomics data of 99 CDG-II patients and 40 controls, consisting of the relative abundances of the transferrin glycoforms and the total plasma N-glycan structures. Genesis version 1.7.7 software was used (44) based on average linkage clustering and Pearson correlations. In order to verify if Pearson’s correlation was suitable for clustering of glycomics data, the same data was also analyzed by using Spearman’s correlation (showing many overlap between patients and control, data not shown), and PCA was performed. PCA was performed on the log 10 transformed glycomics data using Canoco version 5.04 (45). Zebrafish model for slc10a7 Zebrafish husbandry. Zebrafish were maintained at 28.5 C in a 10/14-h dark/light cycle. Protocols for experimental procedures were approved by the Ethics Board of St Michaels Hospital, Toronto, Canada (Protocol ACC660). MO knockdown. For knockdown, we used a sp MO oligonucleotide for slc10A7 (slc10A7sp). A standard control MO (ctrl MO) was also injected. The MO sequences are as follows: slc10A7 sp MO5’- AGGACCTGAAAGAAAGCACACTTAT-3’ and standard control MO5’-CCTCTTACCTCAGTTACAATTTATA-3’. MOs were designed by Gene Tools, LLC. Both MOs were injected individually and in combination into 1 cell stage zebrafish embryos. We injected MOs individually, slc10A7 sp at 0.4 mm, and standard control at 0.4 mm. Alcian blue staining in zebrafish for skeletal visualization Alcian blue (Sigma, St Louis, MO) was dissolved in 70% ethanol and 1% hydrochloric acid. Zebrafish embryos (6 dpf) were fixed in 4% paraformaldehyde (PAF) overnight at 4°C, and maintained in 100% methanol at −20°C until processing. The embryos were washed with phosphate-buffered saline with 0.1% Tween-20 (PBST). The embryos were bleached in 30% hydrogen peroxide for 2 h, washed with PBST and transferred into Alcian blue solution. Embryos were stained overnight at room temperature. The embryos were rinsed four times with acidified ethanol (HCl–EtOH): 5% hydrochloric acid, and 70% ethanol. Embryos were rinsed for 20 min in HCl-EtOH and re-hydrated by washing 10 min in a HCl-EtOH/H2Od series (75, 25, 50, 50, 25, 75 and 100%). Embryos were stored in 1 ml of glycerol-KOH at 4°C. For microscopy, embryos were imaged in bright field mode using a dissection microscope (Leica M205 FA) Alizarin red staining in zebrafish to visualize extracellular calcium deposition Alizarin red S (C.I. 74240, Sigma) was prepared with 0.5% alizarin red S powder in water. The zebrafish embryos (6 dpf) were fixed in 4% PAF overnight at 4°C. The embryos were washed with 1 ml 50% ethanol, with rocking, at room temperature for 10 min. After removing the 50% ethanol, 1 ml of 0.5% alizarin red stain solution was added to the embryos and rocked at room temperature overnight. Bleach solution was prepared fresh by mixing equal volumes of 3% H2O2 and 2% KOH for final concentration of 1.5% H2O2 and 1% KOH, 1 ml was added to the embryos, which were incubated for 10 min. For clearing, embryos were rocked at room temperature for 30 min in 1 ml of a solution of 20% glycerol and 0.25% KOH, and then for 2 h in 1 ml of 50% glycerol and 0.25% KOH. Embryos were stored in 50% glycerol and 0.1% KOH at 4°C. For microscopy, embryos were embedded in 1% low-melting agarose (BioShop), and imaged using a Zeiss laser-scanning confocal microscope (Zeiss LSM 700). Cell biology studies Cell culture conditions. Primary patient fibroblasts and SLC10A7-V5 complemented fibroblasts were cultured in M199 medium (PAN Biotech, P04–07050) supplemented with 10% fetal calf serum and 1% penicillin/streptomycin at 37°C in humidity saturated 5% CO2 atmosphere. HeLa cell cultures were maintained in high glucose Dulbecco’s modified Eagle’s medium (DMEM) with Glutamax and sodium pyruvate (Gibco 31966021), supplemented with 10% fetal calf serum (GE Healthcare Life Sciences Hyclone SV30160) and 1% antibiotic-antimycotic (Gibco 15240062) at 37°C and 5% CO2. Prior to transfection, cells were dissociated with 2 mm EDTA in PBS. Cells were transfected using the Neon Transfection System (ThermoFisher) with 1 µg plasmid DNA per 1 × 105 cells, using the following settings: 1005 V pulse voltage, 35 ms pulse width, 2 pulses. Following electroporation, cells were transferred to 24-well plates containing pre-warmed Opti-MEM without phenol red (Gibco 11058021) for regeneration. After 4 h, Opti-MEM was replaced for regular HeLa medium as described earlier. Construction of plasmids. A SLC10A7 cDNA without a stopcodon (Genbank Accession EU831744, Refseq NM_001029998) cloned in the vector pDONR221 (plasmid ID HsCD00295767) was obtained from the PlasmID Repository at Harvard Medical School (plasmid.med.harvard.edu). The cDNA was cloned into the lentiviral expression vector pLenti6.2/V5-DEST (Invitrogen) by using Gateway technology (Invitrogen), creating a SLC10A7 open reading frame with a C-terminal V5-tag. This construct was used for both stable transfection of fibroblast cells by lentiviral transduction, and for transient transfection experiments of HeLa cells. The production of lentiviral particles using HEK 293FT cells, and subsequent infections of fibroblast cells followed by selection of transduced cells with blasticidin (InvivoGen) was performed as described before in (46). Subcellular localization of SLC10A7 SLC10A7-V5 complemented fibroblasts and SLC10A7-V5 transfected Hela cells were seeded on 12 mm-diameter coverslips. When ∼50% confluent, cells were washed three times with PBS and fixed with 4% PAF in PBS for 10 min. Subsequently, the cells were washed three times with PBS and stored in a parafilm-sealed 24-well plate at 4°C. Cells were permeabilized and blocked in 2.5% bovine serum albumin (BSA), 0.1% Triton X-100 in PBS for 15 min. Coverslips were incubated with primary antibodies against V5 (mouse anti-V5, Thermo Fisher, R960–25) and a specific marker diluted in 2.5% BSA, 0.1% Triton X-100 in PBS for 1 h at room temperature (rabbit anti-Calnexin 1:400, Abcam, ab22595; rabbit anti-ERGIC-53 1:100, Sigma, E1031; rabbit anti-β-COP 1:2000, Abcam, ab2899; rabbit anti-SEC31A 1:200, Sigma, HPA005457; rabbit anti-Giantin 1:1000, BioLegend, PRB-114C; rabbit anti-TGOLN2 1:500, Sigma, HPA012723). After washing the cells three times with 2.5% BSA, 0.1% Triton X-100 in PBS, cells were incubated with secondary antibodies goat–anti-rabbit Alexa Fluor 568 (1:2000, Invitrogen, A11011) and rabbit-anti-mouse Alexa Fluor 488 (1:2000, Thermo Fisher, A21204). Coverslips were washed three times with PBS and mounted on glass slides with ProLong Diamond Antifade Mountant with DAPI (Invitrogen, P36962) and stored at 4°C. Cells were imaged using a Zeiss LSM880 confocal microscope with a Plan-Apochromat 63x/1.4 Oil DIC M27 objective. Image analysis was performed with Fiji (ImageJ 2.0.0-rc-61/1.51n) and ColocalizeR v0.8b (http://colocalizer.iple.be). Analysis of Golgi morphology in SLC10A7 deficient fibroblasts Primary patient fibroblasts were seeded on 12-mm diameter coverslips. When ∼50% confluent, cells were washed three times with PBS and fixed with 4% PAF in PBS for 10 min. Subsequently, the cells were washed three times with PBS and stored in a parafilm-sealed 24-well plate at 4°C. Cells were permeabilized in 0.5% saponin in PBS for 10 min. Cells were subsequently washed three times with PBS-T and blocked in 3% BSA in PBS-T for 30 min. Coverslips were incubated with primary antibodies against GM130 (mouse anti-GM130 1:250, BD Biosciences, 610823) and TGOLN2 (rabbit anti-TGOLN2 1:500, Sigma, HPA012723) diluted in 3% BSA in PBS-T. After washing the cells three times with PBS-T, cells were incubated with secondary antibodies goat–anti-rabbit Alexa Fluor 568 (1:2000, Invitrogen, A11011) and rabbit-anti-mouse Alexa Fluor 488 (1:2000, Thermo Fisher, A21204) diluted in 3% BSA in PBS-T. Coverslips were washed three times with PBS-T and mounted on glass slides with ProLong Diamond Antifade Mountant with DAPI (Invitrogen, P36962) and stored at 4°C. Metabolic glycoprotein labeling in SLC10A7 deficient fibroblasts Primary skin fibroblasts were maintained in DMEM supplemented with 10% fetal bovine serum (Dutscher), at 37°C in humidity saturated 5% CO2 atmosphere. Fibroblasts from both healthy and SLC10A7 deficient patients were grown overnight on glass coverslips (12-mm diameter). Medium was then changed with pre-warmed medium containing 500 µm of N-(4-pentynoyl)neuraminic acid (SiaNAl). Labeling lasted 7 h. The labeling was stopped by fixing the cells with 4% PAF. Cells were then permeabilized in 0.5% Triton X-100 for 10 min. After washes, permeabilized cells were incubated with 100 µl/coverslip of a freshly prepared ‘click solution’ (K2HPO4, 100 mm; Sodium ascorbate, 2.5 mm; CuSO4, 150 µm; BTTAA, 300 µm; Azide-Fluor 545 [Sigma-Aldrich no. 760757], 10 µm) (47). The bioconjugation reaction was performed during 45 min in the dark, at room temperature with gentle shaking. The pool of fluorescent glycoconjugates was visualized through an inverted Zeiss-LSM780 confocal microscope. Pictures were taken using Zen Imaging software. For comparison purposes, each picture was taken under the same settings. Colocalization of mislocalized glycoproteins with early endosomes Anti-EEA1 antibody was from BD Biosciences (no. 610456, Europe). Anti-TMEM165 antibody was from Sigma-Aldrich (no. HPA038299, Europe). Polyclonal goat anti-rabbit or goat anti-mouse conjugated with Alexa Fluor were purchased from Invitrogen Molecular Probes (respectively no.A21038 and no. A11001, Europe). After the click chemistry reaction described before, it is also possible to start an immunolabeling. Fixed cells were first incubated in blocking buffer (2% fetal bovine serum, 2% BSA [Roche no. 10735086001], 0.2% Gelatin [Sigma-Aldrich no. G-8150] in PBS 1×) for 1 h at room temperature in the dark, then incubated for 1 h with primary polyclonal antibody against TMEM165 and primary monoclonal antibody against EEA1 diluted respectively at 1:300 and 1:100 in blocking buffer. After washes with PBS, cells were incubated for 1 h with secondary antibodies anti-rabbit Alexa Fluor 700 and anti-mouse Alexa Fluor 488 diluted at 1:600 in blocking buffer. Pulse chase labeling of glycoproteins in SLC10A7 deficient fibroblasts Fibroblasts from both healthy and SLC10A7 deficient patients were cultured in presence of 500 µm of our alkyne tagged sugar SiaNAl in DMEM during 7 h (pulse). Medium containing SiaNAl was then replaced by regular DMEM and the fibroblasts were grown for 48 h (chase). Then cells were fixed, permeabilized, and reacted with the fluorescent probe Azide-Fluor 545 in presence of CuSO4 (150 µm) and of the ligand BTTAA (300 µm). Fluorescence was detected through an inverted Zeiss-LSM780 confocal microscope. Pictures were taken using Zen Imaging software. For comparison purposes, each picture was taken under the same settings. Acknowledgements We would like to thank the EURO-CDG2 consortium for helpful discussions and collaborations, Frans van den Brandt, Marion Ybema-Antoine, Marit Pullen and Christina Hahnen for technical assistance, and the Genome technology Center of the Radboudumc for support in exome sequencing. C.B., D.V. and F.F. are indebted to the Research Federation FRABio (Univ.Lille, CNRS, FR 3688, FRABio, Biochimie Structurale et Fonctionnelle des assemblages Biomoléculaires) for providing the scientific and technical environment conducive to achieving this work. Conflict of Interest statement. None declared. Funding This work was supported by grants from the Dutch Organization for Scientific Research, ZONMW [Medium Investment Grant 40–00506-98–9001 and VIDI Grant 91713359 to D.J.L, VENI grant 722015012 to M.v.S.]; as well as funding support from the Natural Sciences and Engineering Research Council of Canada [grant RGPIN 05389–14 to X.Y.W.]; Brain Canada Foundation and Health Canada [grant PSG14–3505 to X.Y.W.];, Canada Foundation for Innovation [grant number 26233 to X.Y.W.] and Ministry of Health of Malaysia [grant number R02087 to N.A.B.]. This work was further supported by the European Union’s Horizon 2020 research and innovation program under the ERA-NET Cofund action N 643578 (EUROCDG-2). T.H., N.O. and H.H. were supported by grants MZ CR AZV 16–31932A and ProgresQ26/LF. Author contributions K.R., L.H., T.H., H.H., N.O., M.S., and F.V.S. recruited the patients, reviewed the clinical and radiographic features and obtained biologic materials from patients. CDG group provided the patient materials for the Table in Figure 1. A.A. performed exome sequencing, database studies, mRNA and biochemical studies. N.A.B., G.R.P.O., S.v.H., K.H. and R.B. performed glycomics experiments and bio-informatics analyses. M.N., P.T.A.L., G.v.d.B. performed localization studies. X.Y.W. and K.B.A. designed and performed zebrafish modeling studies. D.V., C.B. and F.F. performed chemical glycobiology experiments. M.K., R.S., S.T., R.R. and L.V.D.H. performed DNA sequencing, cloning, and MLPA analysis. D.J.L. conceived and coordinated the study. A.A., N.A.B., M.V.S. and D.J.L. wrote the article with contributions from all coauthors. All coauthors edited, reviewed and approved the final article. Web Resources Protein atlas: https://www.proteinatlas.org/ CADD: http://cadd.gs.washington.edu/ ExAC: http://exac.broadinstitute.org/ Transmembrane prediction: http://www.cbs.dtu.dk/services/TMHMM/ MLPA: MRC-Holland: http://www.mlpa.com ColocalizeR v0.8b: http://colocalizer.iple.be SIFT: http://sift.bii.a-star.edu.sg/ Align-GVGD: http://agvgd.hci.utah.edu/ PolyPhen-2 - Polymorphism Phenotyping version 2: http://genetics.bwh.harvard.edu/pph2/ Mutation Taster: http://www.mutationtaster.org/ References 1 de Ligt J. , Willemsen M.H. , van Bon B.W. , Kleefstra T. , Yntema H.G. , Kroes T. , Vulto-van Silfhout A.T. , Koolen D.A. , de Vries P. , Gilissen C. et al. ( 2012 ) Diagnostic exome sequencing in persons with severe intellectual disability . N. Engl. J. Med ., 367 , 1921 – 1929 . Google Scholar CrossRef Search ADS PubMed 2 Gilissen C. , Hehir-Kwa J.Y. , Thung D.T. , van de Vorst M. , van Bon B.W. , Willemsen M.H. , Kwint M. , Janssen I.M. , Hoischen A. , Schenck A. et al. ( 2014 ) Genome sequencing identifies major causes of severe intellectual disability . Nature , 511 , 344 – 347 . Google Scholar CrossRef Search ADS PubMed 3 Vissers L. , van Nimwegen K.J.M. , Schieving J.H. , Kamsteeg E.J. , Kleefstra T. , Yntema H.G. , Pfundt R. , van der Wilt G.J. , Krabbenborg L. , Brunner H.G. et al. ( 2017 ) A clinical utility study of exome sequencing versus conventional genetic testing in pediatric neurology . Genet. Med ., 19 , 1055 – 1063 . Google Scholar CrossRef Search ADS PubMed 4 Stark Z. , Tan T.Y. , Chong B. , Brett G.R. , Yap P. , Walsh M. , Yeung A. , Peters H. , Mordaunt D. , Cowie S. et al. ( 2016 ) A prospective evaluation of whole-exome sequencing as a first-tier molecular test in infants with suspected monogenic disorders . Genet. Med ., 18 , 1090 – 1096 . Google Scholar CrossRef Search ADS PubMed 5 Akawi N. , McRae J. , Ansari M. , Balasubramanian M. , Blyth M. , Brady A.F. , Clayton S. , Cole T. , Deshpande C. , Fitzgerald T.W. et al. ( 2015 ) Discovery of four recessive developmental disorders using probabilistic genotype and phenotype matching among 4, 125 families . Nat. Genet ., 47 , 1363 – 1369 . Google Scholar CrossRef Search ADS PubMed 6 Cummings B.B. , Marshall J.L. , Tukiainen T. , Lek M. , Donkervoort S. , Foley A.R. , Bolduc V. , Waddell L.B. , Sandaradura S.A. , O'Grady G.L. et al. ( 2017 ) Improving genetic diagnosis in Mendelian disease with transcriptome sequencing . Sci. Transl. Med ., 9 , 1 – 11 . Google Scholar CrossRef Search ADS 7 Tarailo-Graovac M. , Shyr C. , Ross C.J. , Horvath G.A. , Salvarinova R. , Ye X.C. , Zhang L.H. , Bhavsar A.P. , Lee J.J. , Drogemoller B.I. et al. ( 2016 ) Exome sequencing and the management of neurometabolic disorders . N. Engl. J. Med ., 374 , 2246 – 2255 . Google Scholar CrossRef Search ADS PubMed 8 Moremen K.W. , Tiemeyer M. , Nairn A.V. ( 2012 ) Vertebrate protein glycosylation: diversity, synthesis and function . Nat. Rev. Mol. Cell Biol ., 13 , 448 – 462 . Google Scholar CrossRef Search ADS PubMed 9 De Matteis M.A. , Luini A. ( 2011 ) Mendelian disorders of membrane trafficking . N. Engl. J. Med ., 365 , 927 – 938 . Google Scholar CrossRef Search ADS PubMed 10 Wopereis S. , Grunewald S. , Huijben K.M. , Morava E. , Mollicone R. , van Engelen B.G. , Lefeber D.J. , Wevers R.A. ( 2006 ) Transferrin and apolipoprotein C-III isofocusing are complementary in the diagnosis of N- and O-glycan biosynthesis defects . Clin. Chem ., 53 , 180 – 187 . Google Scholar CrossRef Search ADS PubMed 11 Tegtmeyer L.C. , Rust S. , van Scherpenzeel M. , Ng B.G. , Losfeld M.E. , Timal S. , Raymond K. , He P. , Ichikawa M. , Veltman J. et al. ( 2014 ) Multiple phenotypes in phosphoglucomutase 1 deficiency . N. Engl. J. Med ., 370 , 533 – 542 . Google Scholar CrossRef Search ADS PubMed 12 Van Scherpenzeel M. , Timal S. , Rymen D. , Hoischen A. , Wuhrer M. , Hipgrave-Ederveen A. , Grunewald S. , Peanne R. , Saada A. , Edvardson S. et al. ( 2014 ) Diagnostic serum glycosylation profile in patients with intellectual disability as a result of MAN1B1 deficiency . Brain J. Neurol ., 137 , 1030 – 1038 . Google Scholar CrossRef Search ADS 13 Sadat M.A. , Moir S. , Chun T.W. , Lusso P. , Kaplan G. , Wolfe L. , Memoli M.J. , He M. , Vega H. , Kim L.J.Y. et al. ( 2014 ) Glycosylation, hypogammaglobulinemia, and resistance to viral infections . N. Engl. J. Med ., 370 , 1615 – 1625 . Google Scholar CrossRef Search ADS PubMed 14 Hua S. , Williams C.C. , Dimapasoc L.M. , Ro G.S. , Ozcan S. , Miyamoto S. , Lebrilla C.B. , An H.J. , Leiserowitz G.S. ( 2013 ) Isomer-specific chromatographic profiling yields highly sensitive and specific potential N-glycan biomarkers for epithelial ovarian cancer . J. Chromatogr. A , 1279 , 58 – 67 . Google Scholar CrossRef Search ADS PubMed 15 van Scherpenzeel M. , Steenbergen G. , Morava E. , Wevers R.A. , Lefeber D.J. ( 2015 ) High-resolution mass spectrometry glycoprofiling of intact transferrin for diagnosis and subtype identification in the congenital disorders of glycosylation . Transl. Res. J. Lab. Clin. Med ., 166 , 639 – 649.e631 . Google Scholar CrossRef Search ADS 16 Park J.H. , Hogrebe M. , Gruneberg M. , DuChesne I. , von der Heiden A.L. , Reunert J. , Schlingmann K.P. , Boycott K.M. , Beaulieu C.L. , Mhanni A.A. et al. ( 2015 ) SLC39A8 Deficiency: a Disorder of Manganese Transport and Glycosylation . Am. J. Hum. Genet ., 97 , 894 – 903 . Google Scholar CrossRef Search ADS PubMed 17 Ng B.G. , Buckingham K.J. , Raymond K. , Kircher M. , Turner E.H. , He M. , Smith J.D. , Eroshkin A. , Szybowska M. , Losfeld M.E. et al. ( 2013 ) Mosaicism of the UDP-galactose transporter SLC35A2 causes a congenital disorder of glycosylation . Am. J. Hum. Genet ., 92 , 632 – 636 . Google Scholar CrossRef Search ADS PubMed 18 Edvardson S. , Ashikov A. , Jalas C. , Sturiale L. , Shaag A. , Fedick A. , Treff N.R. , Garozzo D. , Gerardy-Schahn R. , Elpeleg O. ( 2013 ) Mutations in SLC35A3 cause autism spectrum disorder, epilepsy and arthrogryposis . J. Med. Genet ., 50 , 733 – 739 . Google Scholar CrossRef Search ADS PubMed 19 Edmondson A.C. , Bedoukian E.C. , Deardorff M.A. , McDonald-McGinn D.M. , Li X. , He M. , Zackai E.H. ( 2017 ) A human case of SLC35A3-related skeletal dysplasia . Am. J. Med. Genet. A , 173 , 2758 – 2762 . Google Scholar CrossRef Search ADS PubMed 20 Marini C. , Hardies K. , Pisano T. , May P. , Weckhuysen S. , Cellini E. , Suls A. , Mei D. , Balling R. , Jonghe P.D. et al. ( 2017 ) Recessive mutations in SLC35A3 cause early onset epileptic encephalopathy with skeletal defects . Am. J. Med. Genet. A , 173 , 1119 – 1123 . Google Scholar CrossRef Search ADS PubMed 21 Jansen J.C. , Cirak S. , van Scherpenzeel M. , Timal S. , Reunert J. , Rust S. , Perez B. , Vicogne D. , Krawitz P. , Wada Y. et al. ( 2016 ) CCDC115 deficiency causes a disorder of Golgi homeostasis with abnormal protein glycosylation . Am. J. Hum. Genet ., 98 , 310 – 321 . Google Scholar CrossRef Search ADS PubMed 22 Bensimon-Brito A. , Cardeira J. , Dionisio G. , Huysseune A. , Cancela M.L. , Witten P.E. ( 2016 ) Revisiting in vivo staining with alizarin red S–a valuable approach to analyse zebrafish skeletal mineralization during development and regeneration . BMC Dev. Biol ., 16 , 2. Google Scholar CrossRef Search ADS PubMed 23 Godoy J.R. , Fernandes C. , Doring B. , Beuerlein K. , Petzinger E. , Geyer J. ( 2007 ) Molecular and phylogenetic characterization of a novel putative membrane transporter (SLC10A7), conserved in vertebrates and bacteria . Eur. J. Cell Biol ., 86 , 445 – 460 . Google Scholar CrossRef Search ADS PubMed 24 Bijsmans I.T. , Bouwmeester R.A. , Geyer J. , Faber K.N. , van de Graaf S.F. ( 2012 ) Homo- and hetero-dimeric architecture of the human liver Na(+)-dependent taurocholate co-transporting protein . Biochem. J ., 441 , 1007 – 1015 . Google Scholar CrossRef Search ADS PubMed 25 Jiang L. , Alber J. , Wang J. , Du W. , Yang X. , Li X. , Sanglard D. , Geyer J. ( 2012 ) The Candida albicans plasma membrane protein Rch1p, a member of the vertebrate SLC10 carrier family, is a novel regulator of cytosolic Ca2+ homoeostasis . Biochem. J ., 444 , 497 – 502 . Google Scholar CrossRef Search ADS PubMed 26 Peanne R. , Vanbeselaere J. , Vicogne D. , Mir A.M. , Biot C. , Matthijs G. , Guerardel Y. , Foulquier F. ( 2013 ) Assessing ER and Golgi N-glycosylation process using metabolic labeling in mammalian cultured cells . Methods in Cell Biology , 118 , 157 – 176 . Google Scholar CrossRef Search ADS PubMed 27 Gilormini P.A. , Lion C. , Vicogne D. , Levade T. , Potelle S. , Mariller C. , Guerardel Y. , Biot C. , Foulquier F. ( 2016 ) A sequential bioorthogonal dual strategy: manNAl and SiaNAl as distinct tools to unravel sialic acid metabolic pathways . Chem. Commun ., 52 , 2318 – 2321 . Google Scholar CrossRef Search ADS 28 Zamboni N. , Saghatelian A. , Patti G.J. ( 2015 ) Defining the metabolome: size, flux, and regulation . Mol. Cell , 58 , 699 – 706 . Google Scholar CrossRef Search ADS PubMed 29 Argmann C.A. , Houten S.M. , Zhu J. , Schadt E.E. ( 2016 ) A next generation multiscale view of inborn errors of metabolism . Cell Metab ., 23 , 13 – 26 . Google Scholar CrossRef Search ADS PubMed 30 Lauc G. , Essafi A. , Huffman J.E. , Hayward C. , Knezevic A. , Kattla J.J. , Polasek O. , Gornik O. , Vitart V. , Abrahams J.L. et al. ( 2010 ) Genomics meets glycomics-the first GWAS study of human N-Glycome identifies HNF1alpha as a master regulator of plasma protein fucosylation . PLoS Genet ., 6 , e1001256. Google Scholar CrossRef Search ADS PubMed 31 Claro da Silva T. , Polli J.E. , Swaan P.W. ( 2013 ) The solute carrier family 10 (SLC10): beyond bile acid transport . Mol. Aspects Med ., 34 , 252 – 269 . Google Scholar CrossRef Search ADS PubMed 32 Pettersson H. , Zarnegar B. , Westin A. , Persson V. , Peuckert C. , Jonsson J. , Hallgren J. , Kullander K. ( 2017 ) SLC10A4 regulates IgE-mediated mast cell degranulation in vitro and mast cell-mediated reactions in vivo . Sci. Rep ., 7 , 1085. Google Scholar CrossRef Search ADS PubMed 33 Larhammar M. , Patra K. , Blunder M. , Emilsson L. , Peuckert C. , Arvidsson E. , Ronnlund D. , Preobraschenski J. , Birgner C. , Limbach C. et al. ( 2015 ) SLC10A4 is a vesicular amine-associated transporter modulating dopamine homeostasis . Biol. Psychiatry , 77 , 526 – 536 . Google Scholar CrossRef Search ADS PubMed 34 Zhao Y. , Yan H. , Happeck R. , Peiter-Volk T. , Xu H. , Zhang Y. , Peiter E. , van Oostende Triplet C. , Whiteway M. , Jiang L. ( 2016 ) The plasma membrane protein Rch1 is a negative regulator of cytosolic calcium homeostasis and positively regulated by the calcium/calcineurin signaling pathway in budding yeast . Eur. J. Cell Biol ., 95 , 164 – 174 . Google Scholar CrossRef Search ADS PubMed 35 Brommage R. , Liu J. , Hansen G.M. , Kirkpatrick L.L. , Potter D.G. , Sands A.T. , Zambrowicz B. , Powell D.R. , Vogel P. ( 2014 ) High-throughput screening of mouse gene knockouts identifies established and novel skeletal phenotypes . Bone Res ., 2 , 14034. Google Scholar CrossRef Search ADS PubMed 36 Sasarman F. , Maftei C. , Campeau P.M. , Brunel-Guitton C. , Mitchell G.A. , Allard P. ( 2016 ) Biosynthesis of glycosaminoglycans: associated disorders and biochemical tests . J. Inherit. Metab. Dis ., 39 , 173 – 188 . Google Scholar CrossRef Search ADS PubMed 37 Smith C.E.L. , Poulter J.A. , Antanaviciute A. , Kirkham J. , Brookes S.J. , Inglehearn C.F. , Mighell A.J. ( 2017 ) Amelogenesis imperfecta; genes, proteins, and pathways . Front. Physiol ., 8 , 435. Google Scholar CrossRef Search ADS PubMed 38 Wright J.T. , Carrion I.A. , Morris C. ( 2015 ) The molecular basis of hereditary enamel defects in humans . J. Dent. Res ., 94 , 52 – 61 . Google Scholar CrossRef Search ADS PubMed 39 Zhang X. , Jiang S. , Mitok K.A. , Li L. , Attie A.D. , Martin T.F.J. ( 2017 ) BAIAP3, a C2 domain-containing Munc13 protein, controls the fate of dense-core vesicles in neuroendocrine cells . J. Cell Biol ., 216 , 2151 – 2166 . Google Scholar CrossRef Search ADS PubMed 40 Akita H. , Fukae M. , Shimoda S. , Aoba T. ( 1992 ) Localization of glycosylated matrix proteins in secretory porcine enamel and their possible functional roles in enamel mineralization . Arch. Oral Biol ., 37 , 953 – 962 . Google Scholar CrossRef Search ADS PubMed 41 Stránecký V. , Hoischen A. , Hartmannová H. , Zaki M.S. , Chaudhary A. , Zudaire E. , Nosková L. , Barešová V. , Přistoupilová A. , Hodaňová K. et al. ( 2013 ) Mutations in ANTXR1 cause GAPO syndrome . Am. J. Hum. Genet ., 92 , 792 – 799 . Google Scholar CrossRef Search ADS PubMed 42 Vissers L.E. , de Ligt J. , Gilissen C. , Janssen I. , Steehouwer M. , de Vries P. , van Lier B. , Arts P. , Wieskamp N. , del Rosario M. et al. ( 2010 ) A de novo paradigm for mental retardation . Nat. Genet ., 42 , 1109 – 1112 . Google Scholar CrossRef Search ADS PubMed 43 Kronewitter S.R. , de Leoz M.L. , Peacock K.S. , McBride K.R. , An H.J. , Miyamoto S. , Leiserowitz G.S. , Lebrilla C.B. ( 2010 ) Human serum processing and analysis methods for rapid and reproducible N-glycan mass profiling . J. Proteome Res ., 9 , 4952 – 4959 . Google Scholar CrossRef Search ADS PubMed 44 Sturn A. , Quackenbush J. , Trajanoski Z. ( 2002 ) Genesis: cluster analysis of microarray data . Bioinformatics , 18 , 207 – 208 . Google Scholar CrossRef Search ADS PubMed 45 Braak CJFT. , Smilauer P. ( 2012 ) Canoco Reference Manual and User's Guide: Software for Ordination, Version 5.0 . Microcomputer Power , Ithaca USA . 46 Renkema G.H. , Wortmann S.B. , Smeets R.J. , Venselaar H. , Antoine M. , Visser G. , Ben-Omran T. , van den Heuvel L.P. , Timmers H.J. , Smeitink J.A. et al. ( 2015 ) SDHA mutations causing a multisystem mitochondrial disease: novel mutations and genetic overlap with hereditary tumors . Eur. J. Hum. Genet ., 23 , 202 – 209 . Google Scholar CrossRef Search ADS PubMed 47 Besanceney-Webler C. , Jiang H. , Zheng T. , Feng L. , Soriano del Amo D. , Wang W. , Klivansky L.M. , Marlow F.L. , Liu Y. , Wu P. ( 2011 ) Increasing the efficacy of bioorthogonal click reactions for bioconjugation: a comparative study . Angewandte Chem ., 50 , 8051 – 8056 . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Human Molecular Genetics Oxford University Press

Integrating glycomics and genomics uncovers SLC10A7 as essential factor for bone mineralization by regulating post-Golgi protein transport and glycosylation

Loading next page...
 
/lp/ou_press/integrating-glycomics-and-genomics-uncovers-slc10a7-as-essential-0rdzza0CHX
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please email: journals.permissions@oup.com
ISSN
0964-6906
eISSN
1460-2083
D.O.I.
10.1093/hmg/ddy213
Publisher site
See Article on Publisher Site

Abstract

Abstract Genomics methodologies have significantly improved elucidation of Mendelian disorders. The combination with high-throughput functional–omics technologies potentiates the identification and confirmation of causative genetic variants, especially in singleton families of recessive inheritance. In a cohort of 99 individuals with abnormal Golgi glycosylation, 47 of which being unsolved, glycomics profiling was performed of total plasma glycoproteins. Combination with whole-exome sequencing in 31 cases revealed a known genetic defect in 15 individuals. To identify additional genetic factors, hierarchical clustering of the plasma glycomics data was done, which indicated a subgroup of four patients that shared a unique glycomics signature of hybrid type N-glycans. In two siblings, compound heterozygous mutations were found in SLC10A7, a gene of unknown function in human. These included a missense mutation that disrupted transmembrane domain 4 and a mutation in a splice acceptor site resulting in skipping of exon 9. The two other individuals showed a complete loss of SLC10A7 mRNA. The patients’ phenotype consisted of amelogenesis imperfecta, skeletal dysplasia, and decreased bone mineral density compatible with osteoporosis. The patients’ phenotype was mirrored in SLC10A7 deficient zebrafish. Furthermore, alizarin red staining of calcium deposits in zebrafish morphants showed a strong reduction in bone mineralization. Cell biology studies in fibroblasts of affected individuals showed intracellular mislocalization of glycoproteins and a defect in post-Golgi transport of glycoproteins to the cell membrane. In contrast to yeast, human SLC10A7 localized to the Golgi. Our combined data indicate an important role for SLC10A7 in bone mineralization and transport of glycoproteins to the extracellular matrix. Introduction Protein glycosylation is an important biological process that is located in the secretory pathway of the cell. The majority of human diseases has been associated with abnormal protein glycosylation via glycomics profiling in plasma, however, the underlying mechanisms remain poorly understood. Congenital Disorders of Glycosylation (CDG) form an emerging group of >100 genetic disorders with a highly heterogeneous clinical spectrum. CDG with abnormal protein N-glycosylation can be identified via analysis of transferrin glycosylation in plasma, covering >50 subtypes. Subtypes with abnormal N-glycan structures are caused by deficient Golgi N-glycosylation, such as defects in glycosyltransferases (e.g. B4GALT1-CDG, MAN1B1-CDG) and in the supply of nucleotide-sugars (e.g. SLC35A2-CDG). Analysis of glycan structures in such defects is directly related to the underlying gene defect. Another major subgroup comprises genetic defects that disrupt Golgi homeostasis, for example defects in ion and pH maintenance and vesicular transport. The large number of genes involved in Golgi homeostasis, estimated about 5–10% of our genome, and their unknown influence on protein glycosylation complicate the identification of disease genes associated with abnormal glycosylation. Strategies for disease gene identification have evolved in recent years by introduction of high-throughput sequencing technologies as whole exome (1) and whole genome (2) sequencing. Success rates vary from roughly ∼20–50%, depending on the approach and clinical subgroup studied (3,4). Identification of recessive traits can be quite challenging, especially in isolated cases. Detailed phenotyping and transcriptome profiling (5–7) have been added to improve disease gene identification. The combination of genomics with high-throughput mass spectrometry-based omics methodologies can potentially further increase the diagnostic yield and, even more importantly, will contribute to the functional annotation of unknown genes. Since glycomics reflects the status of protein glycosylation in the secretory pathway of the cell (8), integration of glycomics and genomics is a promising strategy to unravel novel genetic factors for Golgi homeostasis (9). We here performed comparative glycomics profiling in a cohort of 99 patients with abnormal Golgi glycosylation, 47 with unsolved defect. Combination with exome sequencing in 31 cases resolved 15 patients with a known gene defect and led to identification of SLC10A7 as a novel genetic factor for Golgi homeostasis and glycoprotein trafficking. Molecular studies in cells and zebrafish indicate an important role for SLC10A7 in bone mineralization by influencing glycosylation and transport of proteoglycans and glycoproteins to the extracellular matrix. Results Integrating glycomics and whole-exome sequencing To explore the contribution of glycomics for gene identification, a cohort was selected of 99 patients with CDG, characterized by abnormal N-glycosylation of the plasma protein transferrin. A causative gene defect was found before in 52 patients, while 47 patients remained undiagnosed at the start of this study. The cohort was divided in four subgroups on basis of additional analysis of mucin type O-glycosylation of plasma apolipoprotein CIII (10) (Fig. 1A). Detailed information on protein-linked N-glycans can be obtained by glycomics, i.e. the mass spectrometry (MS)-based analysis of N-glycans that are released from proteins by the enzyme peptide N-glycosidase F (PNGaseF). Commonly, this is performed on total blood plasma proteins, while targeted analysis of individual glycoproteins can be more specific such as shown for transferrin (11,12) and immunoglobulins (13). We performed glycomics analysis of total plasma N-glycans (14) and of plasma purified intact transferrin (15) in the entire cohort of patients (Fig. 1B and Supplementary Material, Tables S1–S3). The aim was to identify glycosylation signatures that allow subgrouping of patients for targeted gene or whole-exome sequencing. Hierarchical clustering (HCL) of total plasma N-glycans (Fig. 1B) showed three major clusters of (i) B4GALT1, (ii) MGAT2 and (iii) MAN1B1. HCL of the intact transferrin glycomics data (Supplementary Material, Fig. S1) exhibited additional groups (iv) TMEM165 and SLC39A8 (15,16) and (v) SLC35A2 of known defects. Targeted sequencing of unsolved cases within these groups resulted in a novel mutation in B4GALT1 (P71), and in SLC39A8 (P44), and a heterozygous mutation in X-linked SLC35A2 (17) (female patient P62). No DNA was available of P45 for confirmation. Sufficient DNA was available of 31 patients for exome sequencing, 29 of which being singleton families. Filtering was done on basis of suspected recessive inheritance for known CDG genes. A mutation was found in TMEM165 in P69, which did not cluster in group 4 (TMEM165), likely because of the very mild glycosylation abnormalities. Manual inspection of the transferrin data for patient 63 revealed a very mild but identical profile as found in group 5 (SLC35A2), subsequently resulting in the identification of a mosaic mutation in SLC35A2 in the exome data (17). In addition, nine patients were identified with a defect in the Conserved-Oligomeric Golgi (COG) complex, and a novel case with mutations in SLC35A3 that clustered in the group of MGAT2-CDG (overview of all identified mutations in Table 1). All COG patients showed a characteristic Golgi homeostasis defect with loss of galactose and sialic acid and all but P80 were in the ApoCIII-0 subgroup (Fig. 1A). Together, these data indicate that glycomics profiles can be diagnostic for defects in glycosyltransferases and can be helpful in guiding the diagnostic process for other CDG subtypes. Table 1. Overview of identified mutations in known CDG disease genes Patient Gene GRCh37 Mutations SIFTa Align-GVGDb Mutation tester PolyPhen-2 P71 B4GALT1 Chr9 c.784T>C; p.(Cys262Arg) Deleterious Class C65 Disease causing Probably damaging P44 SLC39A8 Chr4 c.239T>C; p.(Ile80Thr) Tolerated Class C0 Disease causing Probably damaging P69 TMEM165 Chr4 c.376C>T; p.(Arg126Cys) Deleterious Class C65 Disease causing Probably damaging P19 SLC35A3 Chr1 c.200G>T; p.(Arg67Leu) Deleterious Class C0 Disease causing Probably damaging P27 COG1 Chr17 c.1070 + 5G>A; p.(Met357Ilefs*1) — — — — P73 COG1 Chr17 c.1070 + 5G>A; p.(Met357Ilefs*1) — — — — P78 COG2 Chr1 c.380del; p.(Lys127Argfs*5) — — — — c.389T>A; p.(Val130Glu) Deleterious Class C35 Disease causing Probably damaging P14 COG5 Chr7 c.2197del; p.(Val733Trpfs*11) — — — — P80 COG5 Chr7 c.2171_2173del; p.(Ala724del) — — — — P85 COG5 Chr7 c.704_706delinsTAG; — — — — p.(Ala235_Arg236delinsValGly) c.2338_2340dup; p.(Ile780dup) — — — — P30 COG5 Chr7 c.706C>T; p.(Arg236*) — — — — c.2324C>T; p.(Pro775Leu) Deleterious Class C0 Disease causing Probably damaging P38 COG6 Chr13 c.1075-9T>G; p.(Arg360_Val429del) — — — — P70 COG6 Chr13 c.1167-24A>G; p.Gly390Phefs*6 — — — — P62 SLC35A2 ChrX c.262G>C; p.(Ala88Pro) Tolerated Class C0 Disease causing Probably damaging P63 SLC35A2 ChrX c.818G>A; p.(Gly273Asp) Tolerated Class C0 Disease causing Probably damaging Patient Gene GRCh37 Mutations SIFTa Align-GVGDb Mutation tester PolyPhen-2 P71 B4GALT1 Chr9 c.784T>C; p.(Cys262Arg) Deleterious Class C65 Disease causing Probably damaging P44 SLC39A8 Chr4 c.239T>C; p.(Ile80Thr) Tolerated Class C0 Disease causing Probably damaging P69 TMEM165 Chr4 c.376C>T; p.(Arg126Cys) Deleterious Class C65 Disease causing Probably damaging P19 SLC35A3 Chr1 c.200G>T; p.(Arg67Leu) Deleterious Class C0 Disease causing Probably damaging P27 COG1 Chr17 c.1070 + 5G>A; p.(Met357Ilefs*1) — — — — P73 COG1 Chr17 c.1070 + 5G>A; p.(Met357Ilefs*1) — — — — P78 COG2 Chr1 c.380del; p.(Lys127Argfs*5) — — — — c.389T>A; p.(Val130Glu) Deleterious Class C35 Disease causing Probably damaging P14 COG5 Chr7 c.2197del; p.(Val733Trpfs*11) — — — — P80 COG5 Chr7 c.2171_2173del; p.(Ala724del) — — — — P85 COG5 Chr7 c.704_706delinsTAG; — — — — p.(Ala235_Arg236delinsValGly) c.2338_2340dup; p.(Ile780dup) — — — — P30 COG5 Chr7 c.706C>T; p.(Arg236*) — — — — c.2324C>T; p.(Pro775Leu) Deleterious Class C0 Disease causing Probably damaging P38 COG6 Chr13 c.1075-9T>G; p.(Arg360_Val429del) — — — — P70 COG6 Chr13 c.1167-24A>G; p.Gly390Phefs*6 — — — — P62 SLC35A2 ChrX c.262G>C; p.(Ala88Pro) Tolerated Class C0 Disease causing Probably damaging P63 SLC35A2 ChrX c.818G>A; p.(Gly273Asp) Tolerated Class C0 Disease causing Probably damaging a SIFT: Sorting Intolerant From Tolerant. b GVGD: Grantham Variation Grantham Deviation; classified from most likely to interfere with function (C65) to least likely (C0). Table 1. Overview of identified mutations in known CDG disease genes Patient Gene GRCh37 Mutations SIFTa Align-GVGDb Mutation tester PolyPhen-2 P71 B4GALT1 Chr9 c.784T>C; p.(Cys262Arg) Deleterious Class C65 Disease causing Probably damaging P44 SLC39A8 Chr4 c.239T>C; p.(Ile80Thr) Tolerated Class C0 Disease causing Probably damaging P69 TMEM165 Chr4 c.376C>T; p.(Arg126Cys) Deleterious Class C65 Disease causing Probably damaging P19 SLC35A3 Chr1 c.200G>T; p.(Arg67Leu) Deleterious Class C0 Disease causing Probably damaging P27 COG1 Chr17 c.1070 + 5G>A; p.(Met357Ilefs*1) — — — — P73 COG1 Chr17 c.1070 + 5G>A; p.(Met357Ilefs*1) — — — — P78 COG2 Chr1 c.380del; p.(Lys127Argfs*5) — — — — c.389T>A; p.(Val130Glu) Deleterious Class C35 Disease causing Probably damaging P14 COG5 Chr7 c.2197del; p.(Val733Trpfs*11) — — — — P80 COG5 Chr7 c.2171_2173del; p.(Ala724del) — — — — P85 COG5 Chr7 c.704_706delinsTAG; — — — — p.(Ala235_Arg236delinsValGly) c.2338_2340dup; p.(Ile780dup) — — — — P30 COG5 Chr7 c.706C>T; p.(Arg236*) — — — — c.2324C>T; p.(Pro775Leu) Deleterious Class C0 Disease causing Probably damaging P38 COG6 Chr13 c.1075-9T>G; p.(Arg360_Val429del) — — — — P70 COG6 Chr13 c.1167-24A>G; p.Gly390Phefs*6 — — — — P62 SLC35A2 ChrX c.262G>C; p.(Ala88Pro) Tolerated Class C0 Disease causing Probably damaging P63 SLC35A2 ChrX c.818G>A; p.(Gly273Asp) Tolerated Class C0 Disease causing Probably damaging Patient Gene GRCh37 Mutations SIFTa Align-GVGDb Mutation tester PolyPhen-2 P71 B4GALT1 Chr9 c.784T>C; p.(Cys262Arg) Deleterious Class C65 Disease causing Probably damaging P44 SLC39A8 Chr4 c.239T>C; p.(Ile80Thr) Tolerated Class C0 Disease causing Probably damaging P69 TMEM165 Chr4 c.376C>T; p.(Arg126Cys) Deleterious Class C65 Disease causing Probably damaging P19 SLC35A3 Chr1 c.200G>T; p.(Arg67Leu) Deleterious Class C0 Disease causing Probably damaging P27 COG1 Chr17 c.1070 + 5G>A; p.(Met357Ilefs*1) — — — — P73 COG1 Chr17 c.1070 + 5G>A; p.(Met357Ilefs*1) — — — — P78 COG2 Chr1 c.380del; p.(Lys127Argfs*5) — — — — c.389T>A; p.(Val130Glu) Deleterious Class C35 Disease causing Probably damaging P14 COG5 Chr7 c.2197del; p.(Val733Trpfs*11) — — — — P80 COG5 Chr7 c.2171_2173del; p.(Ala724del) — — — — P85 COG5 Chr7 c.704_706delinsTAG; — — — — p.(Ala235_Arg236delinsValGly) c.2338_2340dup; p.(Ile780dup) — — — — P30 COG5 Chr7 c.706C>T; p.(Arg236*) — — — — c.2324C>T; p.(Pro775Leu) Deleterious Class C0 Disease causing Probably damaging P38 COG6 Chr13 c.1075-9T>G; p.(Arg360_Val429del) — — — — P70 COG6 Chr13 c.1167-24A>G; p.Gly390Phefs*6 — — — — P62 SLC35A2 ChrX c.262G>C; p.(Ala88Pro) Tolerated Class C0 Disease causing Probably damaging P63 SLC35A2 ChrX c.818G>A; p.(Gly273Asp) Tolerated Class C0 Disease causing Probably damaging a SIFT: Sorting Intolerant From Tolerant. b GVGD: Grantham Variation Grantham Deviation; classified from most likely to interfere with function (C65) to least likely (C0). Figure 1. View largeDownload slide Integrating glycomics and genomics for gene identification. (A) Representation of the patient cohort and work flow. Following identification of abnormal glycosylation during classical CDG screening, 99 patients were subgrouped on basis of apoCIII isofocusing for analysis of mucin type O-glycosylation defects. (B) HCL based on Pearson’s correlations and heatmap of glycomics changes in 99 patients as compared to 40 controls. Figure 1. View largeDownload slide Integrating glycomics and genomics for gene identification. (A) Representation of the patient cohort and work flow. Following identification of abnormal glycosylation during classical CDG screening, 99 patients were subgrouped on basis of apoCIII isofocusing for analysis of mucin type O-glycosylation defects. (B) HCL based on Pearson’s correlations and heatmap of glycomics changes in 99 patients as compared to 40 controls. For patient 19, SLC35A3 appeared in the candidate gene list after exome sequencing. A previously reported SLC35A3 deficient patient with autism spectrum disorder showed normal plasma glycosylation and a reduction in tetra-antennary N-glycans in fibroblasts (18), while glycomics of total plasma N-glycans was altered in one patient with epilepsy and skeletal dysplasia (19) and was not tested in another case (20). CDG screening was not performed. The transferrin data of P19 showed a defect in the addition of N-acetyl-glucosamine (GlcNAc) to protein N-glycans (Supplementary Material, Fig. S2), in agreement with SLC35A3 being the Golgi transporter of Uridine 5′-diphosphate (UDP)-GlcNAc. UDP-GlcNAc is required for the formation of complex type N-glycans and additional N-glycan branching by addition of GlcNAc monosaccharides. Total plasma glycomics confirmed the defect in N-glycan branching with a major increase in bi-antennary glycans lacking one GlcNAc residue (glycans 4–3-0–1 and 4–3-1–1) and high mannose glycans (glycan 3–2-0–0, 4–2-0–0 and 5–2-0–0), as well as a major reduction of multi-antennary glycans (Supplementary Material, Table S4). In summary, 15 cases were solved with a known gene defect. Identification of SLC10A7 as novel genetic factor for abnormal Golgi glycosylation In the remaining group of 19 unsolved cases, only one sib pair (P17, P33) was available for combined interpretation of the exome data. Filtering of exome data was performed as described before in (21) and further candidate gene selection was done on basis of a model for autosomal recessive inheritance. Common and non-causative SNPs (frequency > 0.05%), deep intronic and synonymous variants were excluded. Using strict filtering criteria, six genes in P17 (three homozygous and three compound heterozygous) and four genes in P33 (two homozygous and two compound heterozygous) were selected respectively. Only two candidate genes remained that were present in both patients. These included a compound heterozygous mutation in SHROOM3 and a compound heterozygous mutation in SLC10A7. Segregation analysis showed that both variants in SHROOM3 (c.3649G>A and c.2429G>A), Supplementary Material, Table S5) resided on the same allele, thereby rejecting this variant as causative. SLC10A7 contained one genetic variant that was predicted to affect a splice acceptor site and was of paternal origin (Chr4[GRCh37]: g.147214148T>C; NM_001029998.5[SLC10A7]: c.722–16A>G). In addition, a missense mutation was identified of maternal origin (Chr4[GRCh37]: g.147425062C>T; NM_001029998.5[SLC10A7]: c.335G>A; p.[Gly112Asp]). To confirm the effect on splicing, mRNA was isolated from patient fibroblasts and analyzed by RT-PCR and Sanger sequencing, showing skipping of exon 9, which results in a frameshift mutation at position 241 and a premature stop codon (p.Ile241Argfs29*) (Fig. 2A). The missense mutation resulted in a substitution of a highly conserved apolar amino acid, glycine in human to alanine in yeast (Supplementary Material, Fig. S3), for a charged amino acid, aspartic acid, at position 112. SLC10A7 is a membrane protein with 10 predicted transmembrane domains (TMD). Using transmembrane prediction software (TMHMM), it was predicted that the missense substitution (p.Gly118Asp) in TMD4 altered the hydrophobicity profile and completely skipped out TMD4 ((Supplementary Material, Fig. S4). Figure 2. View largeDownload slide Identification of SLC10A7 as novel disease gene. (A) HCL within the ApoCIII-0 subgroup (Supplementary Material, Fig. S6) indicates four patients with similar glycomics signature, two of which with mutations in SLC10A7. Genetic studies uncover genetic deficiency of SLC10A7 in all patients. (B, C) Linear regression analysis to identify characteristic glycan abnormalities for SLC10A7-CDG. (D) Heatmap of characteristic N-glycans in SLC10A7-CDG as compared with the 11 patients in the linear regression model in Figure 2C. Figure 2. View largeDownload slide Identification of SLC10A7 as novel disease gene. (A) HCL within the ApoCIII-0 subgroup (Supplementary Material, Fig. S6) indicates four patients with similar glycomics signature, two of which with mutations in SLC10A7. Genetic studies uncover genetic deficiency of SLC10A7 in all patients. (B, C) Linear regression analysis to identify characteristic glycan abnormalities for SLC10A7-CDG. (D) Heatmap of characteristic N-glycans in SLC10A7-CDG as compared with the 11 patients in the linear regression model in Figure 2C. To identify additional patients with potential causative mutations in SLC10A7, the raw exome data of all unsolved patients were inspected, however, without success. We then attempted to identify patients with overlapping glycomics signatures. Principal component analysis (PCA) allowed to discriminate only the most clear glycosylation defects like MAN1B1- and B4GALT1-CDG (Supplementary Material, Fig. S5). HCL was then performed of the glycomics data of total plasma proteins in the full patient cohort using both Spearman and Pearson correlations, and the methods were validated by proper clustering of MAN1B1- and B4GALT1-CDG patients (Fig. 1B). Pearson correlation turned out to be the preferred method, which was also supported by the results from PCA clustering (Supplementary Material, Fig. S5). Subsequently, HCL of total plasma glycomics was done within the four ApoCIII subgroups, as defined in Figure 1. Within the ApoCIII-0 group of P17 and P33, two other patients (P32 and P39) were clustered in the same group (Fig. 2A and Supplementary Material, Fig. S6). All four patients showed an increase in glycans 4–3-0–1 (P = 2.7 × 10−25) and 4–3-1–1 (P = 6.7 × 10−14), which were the two most significantly different glycans as compared with controls. Sanger sequencing of all coding regions of SLC10A7, its promoter sequence and potential intragenic regulatory elements (see Supplementary Material, Methods) did not reveal any genetic variant of possible pathogenicity in patients 32 and 39. Subsequently, multiplex ligation-dependent probe analysis (MLPA) was performed to find evidence of a deletion in this gene; however, no large deletions or insertions were found in the patients on the tested positions of the probes (Supplementary Material, Tables S6–S8). In view of the strong glycomics similarity between these four patients, mRNA was isolated from patient fibroblasts and SLC10A7 cDNA was sequenced (Fig. 2A). For patients P17 and P33, besides exon 9, a signal was detected in the cDNA electropherograph in accordance with the sequence of exon 10. However, there seemed to be more c.335G>A containing mRNA than mRNA with the c.722–16A>G variant, which could be explained by nonsense-mediated mRNA decay of the mRNA harboring the exon 9 skip. The cDNA primers for exon 7–12 showed two bands in the control, belonging to the two transcripts NM_001029998.5 and NM_001200842.2, which are both expressed in fibroblasts. In addition, a third vague band was detected, according to the heterozygous skipping of exon 9. Interestingly, the cDNA of SLC10A7 was completely absent for P32 and P39 in multiple individual experiments (Fig. 2A). Subsequently, quantitative PCR (qPCR) analysis was performed, confirming the absence of cDNA. Culturing of skin fibroblasts in the presence of cycloheximide to inhibit nonsense-mediated mRNA decay showed additional bands for patients 17 and 33, indicating the presence of multiple unstable mRNA variants. For patients 32 and 39, no bands were detected even after incubation with cycloheximide (Supplementary Material, Fig. S7). Although several genetic mechanisms could still explain the absence of SLC10A7 cDNA, such as deep-intronic mutations or distant regulatory elements, these data clearly show a complete loss of function of the SLC10A7 protein. Diagnostic glycomics profile for SLC10A7-CDG In view of the importance of specific plasma glycomics signatures for facile diagnostics of future patients, we examined the possibility to identify a characteristic glycan pattern for classification of SLC10A7-CDG within the entire cohort of patients. The 4–3-0–1 and 4–3-1–1 glycans were able to discriminate the four SLC10A7-CDG patients from controls (Fig. 2B). Six patients with a known defect [TMEM165 (P69, P94); SLC39A8 (P44), ATP6V1A1 (P68), ATPV0A2 (P57), and MAN1B1 (P87)] and five patients with unsolved defect (P20, P34, P45, P66 and P72) showed similar values (95% CI). Comparative analysis of SLC10A7-CDG with this group of 11 patients showed that addition of glycans 5–4-0–1 (monosialylated N-glycan) and 9–2-0–0 (high-mannose glycan) allowed complete discrimination of SLC10A7-CDG (Fig. 2C). In Figure 2D, a heatmap of selected N-glycans is shown throughout the Golgi glycosylation pathway, indicating that SLC10A7 mutations affect different stages of protein N-glycosylation. In summary (Supplementary Material, Table S9), SLC10A7-CDG is characterized by decreased sialylation (49%, controls 69–71%), an increase of high-mannose glycans (14%, controls 6–7%) and a characteristic increase in glycans lacking GlcNAc (2.3%, controls 0.7–0.8%). A closer inspection of the transferrin glycoforms also revealed the presence of minor isoforms with GlcNAc-lacking N-glycans, in addition to the reduction of sialic acid (Supplementary Material, Fig. S8). As a first indication that the identified glycosylation abnormalities can be useful for diagnostics of future patients, we studied the plasma sample of a novel patient, recently identified via diagnostic exome-sequencing with a splice mutation in SLC10A7 (Chr4[GRCh37]: g.147431202C>T, NM_001029998.5[SLC10A7]: c.184–1G>A). Analysis of ApoCIII mucin type O-glycosylation showed an ApoCIII-0 profile while N-glycosylation of transferrin and total plasma N-glycans showed the characteristic abnormalities for SLC10A7-CDG (see Supplementary Material, Table S9 and Supplementary Material, Fig. S8). SLC10A7 is essential for teeth and skeletal development Patients with SLC10A7 deficiency share an overlapping clinical phenotype, characterized by short stature, defective enamel formation (amelogenesis imperfecta), skeletal dysplasia, facial dysmorphism, moderate hearing impairment and mildly impaired intellectual development (Fig. 3, Supplementary Material, Table S10). Figure 3. View largeDownload slide Clinical and radiological images of SLC10A7 deficient patients. Patient images (upper panel). Clinical photograph of male patient P17 (A, B) and of the teeth (C, D) at the age of 31 years. Note the left sided ptosis, prognathism, severe chest deformity, scoliosis and short stature, genua valga and abnormally shaped teeth with marked enamel hypoplasia and dental caries. In contrast to her brother, the lateral picture of the face of patient 33 showed mandibular hypoplasia (micrognathia) (E, F). Images of patients P32 and P39 are shown in (G–I), respectively. Radiographs (lower panel). Radiograph of the wrists and hands of P17 (J) showing brachydactyly with short and robust carpal and metacarpal bones and phalanges. Note the prolonged and mildly abnormal shaped distal ulnar end in the left hand X-ray. Radiograph of the lumbar spine, pelvis and hip of the female patient 33 (aged 33 years, K) and the male patient 17 (aged 31 years, L) with severe convexity lumbar scoliosis with hypoplastic and abnormally shaped pelvis (dominantly inferior pubic rami hypoplasia). Note also the narrow hip joint space, flatten femoral heads and short femoral necks. X-rays of P17 of the AP chest (M), lumbar (N) and lateral (O) spine with severe right convexity thoraco-lumbar scoliosis. There is no evidence of vertebral body deformity or fracture. Figure 3. View largeDownload slide Clinical and radiological images of SLC10A7 deficient patients. Patient images (upper panel). Clinical photograph of male patient P17 (A, B) and of the teeth (C, D) at the age of 31 years. Note the left sided ptosis, prognathism, severe chest deformity, scoliosis and short stature, genua valga and abnormally shaped teeth with marked enamel hypoplasia and dental caries. In contrast to her brother, the lateral picture of the face of patient 33 showed mandibular hypoplasia (micrognathia) (E, F). Images of patients P32 and P39 are shown in (G–I), respectively. Radiographs (lower panel). Radiograph of the wrists and hands of P17 (J) showing brachydactyly with short and robust carpal and metacarpal bones and phalanges. Note the prolonged and mildly abnormal shaped distal ulnar end in the left hand X-ray. Radiograph of the lumbar spine, pelvis and hip of the female patient 33 (aged 33 years, K) and the male patient 17 (aged 31 years, L) with severe convexity lumbar scoliosis with hypoplastic and abnormally shaped pelvis (dominantly inferior pubic rami hypoplasia). Note also the narrow hip joint space, flatten femoral heads and short femoral necks. X-rays of P17 of the AP chest (M), lumbar (N) and lateral (O) spine with severe right convexity thoraco-lumbar scoliosis. There is no evidence of vertebral body deformity or fracture. Patient P33 is a 33.5-year-old woman with short stature (−5.3 SD), amelogenesis imperfecta, skeletal dysplasia, severe scoliosis, genua valga, pedes planovalgi, mandibular hypoplasia, inguinal hernia, moderate hearing impairment (bilateral hypacusis) and subaverage intellectual functioning. Bone mineral density measurement in the area of the proximal femur and femoral neck clearly revealed decreased bone mineral mass, compatible with osteoporosis (Z-scores −2.9 and −3.6, respectively). Her brother (P17), 31-year-old, has a short stature (−7.0 SD), amelogenesis imperfecta, skeletal dysplasia, prognathism, severe scoliosis, genua valga, pedes planovalgi, submucous cleft palate, left sided myopia gravis, strabismus convergens, unilateral left sided ptosis, moderate hearing impairment (bilateral hypacusis) and subaverage intellectual functioning (IQ 70–75). Patient 39 is an adopted male patient of 8 years with short stature (−4.0 SD), dental decay with discolored enamel and dental crowding, failure to thrive, developmental delay, and mild conductive hearing loss on the right. A skeletal survey at age 6 showed mild skeletal anomalies with irregularity of the anterosuperior endplates of a few vertebral bodies near the thoracolumbar junction, mild bilateral coxa valga, and a hypoplastic right first rib. Dysmorphic features included: frontal bossing, dolichocephaly, flat and broad nasal bridge, long and relatively smooth philtrum, and mildly thin upper lip. The patient is prone to frequent, severe respiratory infections, one of which led to sepsis, and also suffers from chronic malabsorption. He continues to demonstrate poor growth despite being fed via G-tube. P32 is a 24-year male patient with short stature (<−3.0 SD), amelogenesis imperfecta with lack of enamel, skeletal dysplasia, inguinal hernia, global developmental delay, joint hypermobility, clubfeet, phimosis, shawl scrotum, and glaucoma. Facial dysmorphism consisted of dry brittle hair, long philtrum, a webbed neck and thin upper lip. Radiological analysis revealed skeletal dysplasia with cervical stenosis, kyphoscoliosis and pectus excavatum. The newly identified patient is a 12-month-old boy with a disproportionate short stature (−5, 4 SD) with short limbs, head circumference at −2, 1 SD, small thorax, feeding difficulties, inguinal hernia, hypermobile short joints and dysmorphic features (short neck, round face, micrognathia, hypertelorism, prominent eyes and long philtrum). He was born at term with a birth weight of 3032 g (p30), length at 41 cm (≪p3), and a head circumference of 33, 8 cm (p30). A skeletal survey one day after birth showed short tubular bones and irregular endplates with central indentations and (the suggestion of) coronal clefts of several lumbar vertebral bodies. Ophthalmologic examination and newborn hearing screening were normal. There is some gross motor delay. Cognitive development so far is normal. A zebrafish model recapitulates the human skeletal defect To further study the function of SLC10A7 in vivo, a zebrafish model was generated by morpholino knockdown of slc10a7. We designed a sp morpholino (sp MO) (slc10a7-e2i1) targeting the sp donor site of exon 2. The slc10A7-sp MO was injected into one-cell stage zebrafish embryos at two different concentrations 8 and 12 ng/nl. To visualize the skeletal development, Alcian blue staining was performed, which corresponds with the expression of glycosaminoglycans. In morphants injected with 8ng/nl of slc10a7-sp, Alcian blue cartilage staining revealed a relatively normal number of cerathobranchials (c1–c5) and a normal fin bud cartilage. However, while the teeth cartilage is straight in embryos injected with a control MO, it appears to be bent downwards in morphants. Injection of 12 ng/nl results in a severe phenotype with edema in the whole body, reduced head, eyes and curled body. Alcian blue staining revealed a reduced Meckel’s cartilage (m) and absence of cerathobranchial number four (c4) (Fig. 4). Figure 4. View largeDownload slide Skeletal phenotype in slc10a7 deficient zebrafish with Alcian blue staining (cartilage) on 6 dpf embryos. slc10A7 sp MO or a control morpholino (Cont MO) were injected at 8 and 12 ng/nl. (A–C) Lateral view of whole embryo. (A’–C’) close-up of head lateral. (A”–C”) close-up of head anterior view. (A”’–C”’) close-up of head dorsal view of bright field image showing cartilage staining. Figure 4. View largeDownload slide Skeletal phenotype in slc10a7 deficient zebrafish with Alcian blue staining (cartilage) on 6 dpf embryos. slc10A7 sp MO or a control morpholino (Cont MO) were injected at 8 and 12 ng/nl. (A–C) Lateral view of whole embryo. (A’–C’) close-up of head lateral. (A”–C”) close-up of head anterior view. (A”’–C”’) close-up of head dorsal view of bright field image showing cartilage staining. Slc10a7 zebrafish morphants display a defect in bone mineralization In view of the teeth decay and skeletal abnormalities in SLC10A7 patients, we further studied skeletal mineralization, by staining with alizarin red which reacts with calcium deposits in tissues (22). Mineralized bone was stained by alizarin red on 6 dpf embryos injected with slc10A7-sp at 8 ng/nl and 12 ng/nl. Morphants injected at 8 ng/nl had a shortened or absent palate cartilage (pc), absent operculum, widened palatal skeleton (ps), reduced cleithrum (c), notochord (n), fifth cerathobranchial (c5), and entopterygoid (en), as compared with control MO-injected embryos. Importantly, injection of higher dose slc10A7-sp MO at 12 ng/nl concentration caused a severe phenotype, without any detectable bone mineralization (Fig. 5). These results support an essential function for SLC10A7 in cartilage development and bone mineralization. Figure 5. View largeDownload slide Bone mineralization in slc10a7 deficient zebrafish by Alizarin red staining on 6 dpf embryos. slc10A7 sp MO, or a Cont MO were injected at 8 and 12 ng/nl. (A–E) Lateral and anterior views of confocal image stacks showing mineralized bone in red in the context of the embryo. (A’–E’) Stack of serial confocal images showing mineralized bone. (A”–D”) 3D renderings from serial confocal images showing mineralized bone. pc, palate cartilage; ps, palatal skeleton; c, cleithrum; n, notochord (n); c5, fifth cerathobranchial; and en, entopterygoid. Video’s are shown in Supplementary Material, Figure S9. Figure 5. View largeDownload slide Bone mineralization in slc10a7 deficient zebrafish by Alizarin red staining on 6 dpf embryos. slc10A7 sp MO, or a Cont MO were injected at 8 and 12 ng/nl. (A–E) Lateral and anterior views of confocal image stacks showing mineralized bone in red in the context of the embryo. (A’–E’) Stack of serial confocal images showing mineralized bone. (A”–D”) 3D renderings from serial confocal images showing mineralized bone. pc, palate cartilage; ps, palatal skeleton; c, cleithrum; n, notochord (n); c5, fifth cerathobranchial; and en, entopterygoid. Video’s are shown in Supplementary Material, Figure S9. SLC10A7 is localized to the secretory pathway in human cells The phenotypic presentations in human and zebrafish SLC10A7 deficiency indicate a defect in extracellular matrix mineralization. Since this process is localized to the secretory pathway of the cell, which also houses the glycosylation machinery, we first studied the localization of SLC10A7. Previous reports proposed a localization of FLAG-tagged SLC10A7 to the cell membrane in Xenopus laevis oocytes and ER in HEK293 cells (23), while a later study indicated intracellular localization of V5-tagged SLC10A7 in U2OS cells (24). In Protein Atlas, SLC10A7 was proposed in nucleoli. Recently, SLC10A7 in Saccharomyces cerevisiae and Candida albicans was found to be present in the plasma membrane (25). Using commercial antibodies, nucleoli were stained in control and patient fibroblasts. Since the SLC10A7 transcript is completely absent in two of the patients, and since western blot revealed multiple aspecific bands for both controls and patients, these results were interpreted as non-specific binding of the commercial antibodies tested (data not shown). We therefore generated a C-terminally tagged V5-SLC10A7 construct to study subcellular localization in Hela cells. After transient transfection, co-staining was performed with several organelle markers of the secretory pathway, ranging from endoplasmic reticulum to early endosomes. Co-localization was mainly observed with markers for the cis-, medial- and trans-Golgi network (Fig. 6A). To confirm similar localization in a different human cell line, fibroblasts were transfected with a lentiviral V5-SLC10A7 construct, which also showed Golgi localization (Fig. 6B). Cellular glycomics profiling of patient fibroblasts showed elevation of truncated glycans lacking GlcNAc, characteristic for SLC10A7 in plasma, in three out of the four cell lines (data not shown), indicating fibroblasts as a suitable model to study the cell biological consequences of SLC10A7 deficiency. Figure 6. View largeDownload slide Subcellular localization of SLC10A7. (A) Colocalization of V5-SLC10A7 (green) by transient transfection in Hela cells and co-staining with the indicated organelle markers (r2 values of 0.25, 0.49, 0.58 and 0.38 for Calnexin, ERGIC-53, Giantin and TGOLN2, respectively). (B) Colocalization of V5-SLC10A7 (green) by lentiviral transfection of fibroblasts and co-staining with the indicated organelle markers (r2 values of 0.11, 0.28, 0.44 and 0.47 for Calnexin, ERGIC-53, Giantin and TGOLN2, respectively). Markers (in purple) used include Calnexin (ER), ERGIC-53(ER-Golgi-Intermediate-Compartment), Giantin (cis- and medial-Golgi) and TGOLN2 (trans-Golgi network). (C) Staining of patient fibroblasts with Golgi marker TGOLN2 showed a more pronounced dilatation of the Golgi apparatus in the four SLC10A7-CDG patients compared to healthy control (Scale bar, 10 μm). Figure 6. View largeDownload slide Subcellular localization of SLC10A7. (A) Colocalization of V5-SLC10A7 (green) by transient transfection in Hela cells and co-staining with the indicated organelle markers (r2 values of 0.25, 0.49, 0.58 and 0.38 for Calnexin, ERGIC-53, Giantin and TGOLN2, respectively). (B) Colocalization of V5-SLC10A7 (green) by lentiviral transfection of fibroblasts and co-staining with the indicated organelle markers (r2 values of 0.11, 0.28, 0.44 and 0.47 for Calnexin, ERGIC-53, Giantin and TGOLN2, respectively). Markers (in purple) used include Calnexin (ER), ERGIC-53(ER-Golgi-Intermediate-Compartment), Giantin (cis- and medial-Golgi) and TGOLN2 (trans-Golgi network). (C) Staining of patient fibroblasts with Golgi marker TGOLN2 showed a more pronounced dilatation of the Golgi apparatus in the four SLC10A7-CDG patients compared to healthy control (Scale bar, 10 μm). SLC10A7 executes its effect on skeletal and bone development via impacting post-Golgi vesicular transport Staining of patient fibroblasts with organelle markers showed a dilation of the Golgi apparatus in ∼20% of the cells (Fig. 6C). In order to study the Golgi dynamics in more detail in connection with protein glycosylation, metabolic oligosaccharide engineering was performed as described in (26,27). Control and SLC10A7 deficient cells were metabolically labeled with SiaNAl for 7 h (Fig. 7A). The main pool of sialylated glycoconjugates was localized to the Golgi apparatus for all fibroblasts lines, in accordance with previous results. No obvious differences in the staining intensity were observed in SLC10A7 deficient cells when compared to control cells suggesting no clear impairment of the Golgi sialylation efficiency. In addition, a vesicular staining pattern of small vesicles was seen throughout the cell for all four SLC10A7 deficient patient fibroblasts, while this was barely visible in control cells. In order to get information on the nature of the observed vesicles, co-localization studies were performed with the early endosomal marker EEA1, the Golgi marker TMEM165 and the lysosomal marker LAMP2. Although no colocalization was observed with LAMP2 (data not shown) and TMEM165, some of the intracellular vesicular staining colocalized with EEA1 (Supplementary Material, Fig. S10), thus demonstrating that sialylated glycoconjugates accumulate at least partly in early endosomes in SLC10A7 deficient cells. Figure 7. View largeDownload slide Post-Golgi trafficking in SLC10A7 deficient fibroblasts. Fibroblasts from healthy individuals and SLC10A7 deficient patients were metabolically labeled with 500 µm of SiaNAl for 7 h (A) and chased for 48 h (B). The sialylated glycoconjugates were stained with azido-545 fluorescent probes and visualized by confocal microscopy. Scale bar, 30 µm in the upper panels and 10 µm in the zoomed panels. Figure 7. View largeDownload slide Post-Golgi trafficking in SLC10A7 deficient fibroblasts. Fibroblasts from healthy individuals and SLC10A7 deficient patients were metabolically labeled with 500 µm of SiaNAl for 7 h (A) and chased for 48 h (B). The sialylated glycoconjugates were stained with azido-545 fluorescent probes and visualized by confocal microscopy. Scale bar, 30 µm in the upper panels and 10 µm in the zoomed panels. As a further step to study glycoprotein trafficking, we designed a pulse-chase labeling experiment to study the fate of labeled glycoproteins. After a 7h pulse with SiaNAl metabolic labeling, the culture medium was replaced, followed by a chase of 48 h (Fig. 7B). In control cells, a perinuclear Golgi staining was mainly present at t = 0, while at 48 h, the fluorescence intensity was mainly detected at the plasma membrane. In contrast, SLC10A7 deficient fibroblasts hardly showed any staining of the plasma membrane and instead showed intracellular staining of vesicles and Golgi. These results indicate a post-Golgi transport defect of glycoproteins through the secretory pathway due to dysfunction of SLC10A7. In summary, our data indicate that SLC10A7 is important to transport glycoproteins and proteoglycans to produce a proper functioning extracellular matrix and for its mineralization. Discussion In this study, we showed the potential of combining high-throughput functional–omics methods with genomics for gene identification by uncovering SLC10A7 as a novel genetic factor for bone mineralization and protein sorting. Introduction of next-generation sequencing in patient diagnostics has significantly improved the diagnostic yield in genetic disease diagnostics (2). Still, interpretation of the causative nature of genetic variants in individual cases is an important bottleneck. High-throughput functional methodologies hold the potential to provide unique signatures for confirmation of genetic variants and at the same time provide a first clue for annotation of gene function and understanding disease mechanisms. Similarly, technological advances in MS have resulted in holistic methods to analyze metabolites (28,29), i.e. metabolomics, which is being implemented in rare disease diagnostics and has led to the identification of diagnostic markers for known metabolic diseases. Complementary to metabolomics, glycomics monitors the process of protein glycosylation, which reflects the secretory pathway of the cell, requiring 5–10% of our genes for processes such as vesicular transport and ion homeostasis. As such, complementary functional–omics methodologies will be required for functional analysis of our genome. Application of glycomics in our selected cohort of patients supported the genetic diagnosis of known disease genes in 15 cases. The data generated from glycomics experiments on rare genetic disorders exhibit high dimensionality with numerous variables on limited samples. PCA would be an attractive way to project the multivariate data into lower dimensions for visual interpretation. However, PCA was only successful to discriminate defects in glycosyltransferases like MAN1B1 and B4GALT1, while more subtle abnormalities were difficult to classify, such as for patients 63 (SLC35A2) and 69 (TMEM165). Moreover, information on individual glycan levels between patients and controls is lost. HCL of specific subsets of patients allowed to classify patients with more subtle glycomics changes such as SLC10A7-CDG. A few additional subgroups of unsolved patients could be identified by HCL, however, no progress in gene identification could be made due to a lack of available DNA. Nevertheless, our integrated glycomics-genomics approach shows the potential to identify additional gene defects. To determine the diagnostic value of the glycomics profiles for future patients, it will be essential to establish the specificity and sensitivity in larger patient cohorts. As individual CDG are very rare, this will require a world-wide collaborative effort and building of a CDG patient registry. This will also allow to determine the prevalence of CDG subtypes, which is ultimately required to determine the positive predictive value of the glycomics profiles as identified in our study. Since many common disorders have already been associated with abnormal plasma glycomics profiles (30), it is essential to unravel genes that are directly or indirectly involved in protein glycosylation. The plasma glycomics abnormalities in this cohort of monogenetic disorders reflect the wide range of biological mechanisms that influence protein glycosylation. The link of SLC10A7 with protein glycosylation is unexpected, since the SLC10 family has thus far been associated with sodium-bile acid transporters (31). Members within this family are involved in the transport of taurocholate (SLC10A1), bile (SLC10A2) and sulfated steroids (SLC10A6). SLC10A4 is potentially involved neurotransmitter and mastocyte mediator secretion (32,33). SLC10A3, SLC10A5 and SLC10A7 remain functionally uncharacterized (31). Although SLC10A7 shares homology with other members of the family, this membrane protein has distinctive features. In contrast to all other members of the SLC10 protein family that contain eight predicted TMD and are present in vertebrates only, SLC10A7 contains 10 hydrophobic segments and homologs were identified in bacteria (23), yeast and plants. A phylogenetic tree of sodium bile symporters shows clustering of the SLC10A1-A6 homologs, while SLC10A7 clusters with orthologues that are present in eukaryotes. Human SLC10A7 shares 19.4 and 33.7% amino acid sequence identity with S. cerevisiae ScRch1 (34) and C. albicans CaRCH1 (25) proteins, respectively. Both transporters reside in the plasma membrane and are involved in regulation of cytosolic calcium homeostasis. The phenotype of amelogenesis imperfecta and skeletal dysplasia in SLC10A7 deficiency seems mainly linked to a defect in the generation and/or mineralization of the extracellular matrix. Previously, slc10a7 knock-out mice were identified with moderate skeletal dysplasia in a screen for skeletal phenotypes (35). Several types of skeletal dysplasia have been associated with a defect in the biosynthesis of proteoglycans (36). The defect in vesicular transport as we observed in patient fibroblasts could indicate a defect in proper deposition of proteoglycans in the extracellular matrix. Amelogenesis imperfecta is not a common feature of defects in proteoglycan biosynthesis. Amelogenesis imperfecta is characterized by defective enamel formation, a process known as amelogenesis (37,38). In a secretion stage, ameloblasts secrete a protein-containing matrix together with hydroxyapatite crystals. In a maturation stage, ameloblasts are re-oriented to secrete protein-degrading enzymes and calcium ions for mineralization, resulting in the formation of the hard enamel tissue, almost completely mineralized with calcium hydroxyapatite. Amelogenesis imperfecta can be classified roughly in hypoplastic and hypomineralized forms, associated with these two respective stages of amelogenesis. The discolored teeth in some SLC10A7 patients could indicate hypomineralized amelogenesis imperfect (37). This is in line with the reduced staining of alizarin red in slc10a7 morphant zebrafish, which stains skeletal calcium deposits (22). Several possible mechanisms could position SLC10A7 in the biological pathways of enamel mineralization. In view of the proposed role of SLC10A7 homologs in S. cerevisiae and C. albicans in regulating calcium homeostasis, it is tempting to speculate that human SLC10A7 plays an important role in calcium homeostasis during enamel mineralization. In addition, calcium might play a role in the calcium-dependent regulated vesicular transport during exocytosis (39), supported by the vesicular transport defect in fibroblasts. Finally, it can’t be excluded that altered protein glycosylation in ameloblasts contributes to the hypomineralized amelogenesis imperfecta phenotype (40). In summary, we have shown that integration of glycomics and genomics facilitates identification of novel genetic factors for Golgi homeostasis. Furthermore, our data indicate an important role for SLC10A7 in teeth and skeletal mineralization via an influence on protein glycosylation and the transport of proteoglycans and glycoproteins to the extracellular matrix. Materials and Methods Materials of participating individuals Blood and, if obtained, fibroblasts of individuals were sent to the Radboud University Medical Center, Translational Metabolic Laboratory, for diagnostics of CDG. This was based on clinical suspicion for an inborn error of metabolism. All participating individuals or their legal representatives gave informed consent for exome sequencing. Tissue and samples were obtained in accordance with the Declaration of Helsinki. For publication of facial images, written informed consent was obtained from the parents. The diagnosis of all patients with known CDG gene defect was genetically and biochemically confirmed in previous studies (Supplementary Material, Tables S1–S3). For establishment of reference intervals for total plasma N-glycans, plasma samples of 40 healthy controls were received from the Sanquin Blood Bank (Nijmegen) according to their protocols of informed consent. Screening for CDG Routine CDG screening tests for protein N-glycosylation by transferrin IEF as well as for protein mucin type O-glycosylation by ApoC-III IEF were performed as described previously in (21). In brief for transferrin IEF, 10 μl of plasma was incubated with ferric citrate buffer before the diluted sample was applied to a hydrated Immobiline dry gel (Servalyt pH 5–7; GE Healthcare) and run on a Phast System (GE Healthcare). Transferrin immunoprecipitation was performed with 60 μl of polyclonal rabbit anti-human transferrin antibody (8.5 g/l; Dako) followed by washing, fixation, staining and destaining steps to visualize the transferrin isoforms. The relative amounts of transferrin isoforms were determined by densitometry (Image Scanner Amersham/Biosciences; Lab Scan 6.0 and IQTL software) and compared with established reference intervals. In short for ApoC-III IEF, 2 μl of plasma was 15 times diluted with saline solution. Before electrophoresis, the gel was rehydrated in a solution containing 8 m urea. After blotting on a nitrocellulose membrane, the blot was washed and blocked before incubation with anti-ApoC-III antibody (1:2000, Rockland, no. 600–101-114). After incubation with the secondary anti-goat-HRP antibody (1:5000, Thermo Scientific, no. 31402), the blot was visualized by chemoluminescence (ECL reagent (Pierce) on a LAS3000 imaging system (Fujifilm). The relative amounts of ApoC-III isoforms were determined by densitometry (IQTL software) and compared with established reference intervals. Whole-exome sequencing Next generation sequencing and analysis was performed as described earlier in (41). The SureSelect Human All Exon 50Mb Kit (v4, Agilent) was used for exome enrichment, covering ∼21 000 genes. The exome library was sequenced on a SOLiD 5500xl sequencer (Life Technologies). Color space reads were iteratively mapped to the hg19 reference genome with the SOLiD LifeScope software version 2.1. We used our in-house annotation pipeline for annotation of called variants and indels (42). Variants were excluded based on a frequency of >0.2% in our in-house database of >1300 exomes. Also, synonymous variants, deep intronic variants and variants in untranslated regions were excluded. Quality criteria were applied and included variants called more than five times and variation of more than 20% for heterozygous variants and 80% for homozygous variants. Sanger sequencing of gDNA and cDNA In order to confirm the SLC10A7 variants identified by next-generation sequencing and to identify potential mutations in SLC10A7 in P32 and P39, bi-directional direct Sanger sequencing was performed using specific oligonucleotide primers flanking the exons. Total DNA was extracted using the QIAamp DNA kit (Qiagen, Venlo, The Netherlands). Total RNA from cultured fibroblasts was extracted using RNAbee (AMS Biotechnology, Abingdon, UK) and transcribed into cDNA using Superscript II and random primers (Invitrogen, Breda, The Netherlands). All coding regions of SLC10A7 (NM_001029998.5 and NM_001200842.2), its promoter sequence and potential intragenic regulatory elements were amplified using AmpliTaq Gold 360 Master Mix (Fisher Scientific, Landsmeer, The Netherlands) and the primers listed in Supplementary Material, Table S6. The PCR fragments were sequenced using the BigDye Terminator Kit v1.1 (Fisher Scientific, Landsmeer, The Netherlands) on a 3130xL Genetic Analyzer with M13 primers. Multiplex ligation-dependent probe amplification MLPA was performed on the extracted DNA following a procedure from MRC-Holland (Amsterdam, The Netherlands, www.mlpa.com). Fragments were separated with GeneScan 500LIZ dye size Standard (Fisher Scientific, Landsmeer, The Netherlands) on a 3130xL Genetic Analyzer. The data was analyzed by using Coffalyser.Net software (MRC-Holland, Amsterdam, The Netherlands). High-resolution QTOF MS of plasma transferrin Transferrin was immunopurified from control and patient plasma as previously described in (15) and analyzed by MS on a microfluidics-based platform (Agilent Technologies) consisting of an Agilent 1260 nanoLC-HPLC-chip system using a C8 protein chip coupled to an Agilent 6540 QTOF LC/MS system. Data analysis was performed using Agilent Mass Hunter Qualitative Analysis Software B.05. The distribution of raw charge data was deconvoluted to reconstructed mass data using Agilent BioConfirm Software (version 5). A set of 30 transferrin glycoforms (Supplementary Material, Table S2) was selected for relative quantitation, calculated to the total abundance of the 30 selected glycoforms. High-resolution QTOF MS of total plasma N-glycans Analysis of plasma N-glycans was performed based on Kronewitter et al. (43) with minor modifications. 10 μl of plasma were mixed in equal parts with an aqueous solution of 200 mm ammonium bicarbonate and 10 mm dithiothreitol. Protein denaturation was performed in mild conditions by alternating between boiling temperature and room temperature in a water bath for six cycles of five seconds each. For enzymatic release of N-glycans, 1 μl of PNGaseF (New England Biolabs, catalog no. P0704L) was added and the mixture was incubated for 22 h at 37°C. Ethanol precipitation was performed with 80% (v/v) ethanol to remove proteins from the glycans by adding 80 μl of ethanol and the mixture was frozen at −80°C for 45 min. The mixture was centrifuged at 14 000 rpm (Eppendorf) for 20 min and the supernatant was dried in vacuo (Thermo RVT4104 Refrigerated Vapor Trap). Further purification and enrichment of glycans was performed using a graphitized carbon cartridge (Grace Davison, catalog no.G4240–64010, 150 mg, 4.0 ml), with elution of N-glycans using 2.0 ml of 40% acetonitrile and 0.05% trifluoroacetic acid (v/v) in water. MS was performed on the microfluidics-based platform as above using a porous graphitized carbon chip. Dried N-glycans were reconstituted in 50 µl pure water (per 200 nl of serum) and 1 μl sample was loaded onto the enrichment column and analyzed using the chromatographic conditions as described in (14). MS spectra were acquired in the positive ion mode over a mass range of m/z 600–2000 with an acquisition time of 1.5 s per spectrum. Mass correction was enabled using Agilent Calibrant Mix G1969–85000 with reference masses of m/z 622.029, 922.010, 1221.991 and 1521.971. Raw LC-MS data were analyzed using the Molecular Feature Extraction algorithm (Mass Hunter Qualitative Analysis Software B.05). MS peaks were filtered with a signal-to-noise ratio of 5.0 and parsed into individual ion species. All individual ion species associated with single compounds (e.g. doubly and triply protonated ions, and all associated isotopologues) were summed to create extracted ion chromatograms (ECCs) based on expected isotopic distribution and charge state information. Using a mass tolerance of 20 ppm, deconvoluted masses of each ECC peak were compared against a theoretical glycan mass library (in-house) consisting of all possible complex, hybrid and high mannose type N-glycans that have been reported in human plasma. Hence, only glycan compositions containing hexose (Hex), N-acetylhexosamine (HexNAc), deoxyhexose (dHex) and N-acetylneuraminic acid (Neu5Ac) were included. Glycan structures are indicated with the number of Hex-HexNAc-dHex-Neu5Ac residues, respectively. Relative abundances of each glycan were obtained through normalization to the total volume of all detected glycan compounds. Plasma samples of 40 healthy controls were used to establish control ranges. These samples were also analyzed by transferrin QTOF MS to confirm normal transferrin glycosylation. Statistical analysis Data were analyzed using GraphPad Prism (version 5.03) and IBM SPSS (version 22.0) software. Mean and 95% CI for total plasma N-glycans were used to express the reference intervals in 40 normal controls. For determination of P-values, student’s t-test was used for the comparison of the glycans’ relative abundance between control and other groups (e.g. SLC10A7-CDG). Regression analysis using SPSS (method = forward stepwise [conditional]) was performed on all glycans to determine linear functions and to identify a characteristic glycan profile for SLC10A7 deficiency. HCL and bioinformatics 2D HCL was performed on the glycomics data of 99 CDG-II patients and 40 controls, consisting of the relative abundances of the transferrin glycoforms and the total plasma N-glycan structures. Genesis version 1.7.7 software was used (44) based on average linkage clustering and Pearson correlations. In order to verify if Pearson’s correlation was suitable for clustering of glycomics data, the same data was also analyzed by using Spearman’s correlation (showing many overlap between patients and control, data not shown), and PCA was performed. PCA was performed on the log 10 transformed glycomics data using Canoco version 5.04 (45). Zebrafish model for slc10a7 Zebrafish husbandry. Zebrafish were maintained at 28.5 C in a 10/14-h dark/light cycle. Protocols for experimental procedures were approved by the Ethics Board of St Michaels Hospital, Toronto, Canada (Protocol ACC660). MO knockdown. For knockdown, we used a sp MO oligonucleotide for slc10A7 (slc10A7sp). A standard control MO (ctrl MO) was also injected. The MO sequences are as follows: slc10A7 sp MO5’- AGGACCTGAAAGAAAGCACACTTAT-3’ and standard control MO5’-CCTCTTACCTCAGTTACAATTTATA-3’. MOs were designed by Gene Tools, LLC. Both MOs were injected individually and in combination into 1 cell stage zebrafish embryos. We injected MOs individually, slc10A7 sp at 0.4 mm, and standard control at 0.4 mm. Alcian blue staining in zebrafish for skeletal visualization Alcian blue (Sigma, St Louis, MO) was dissolved in 70% ethanol and 1% hydrochloric acid. Zebrafish embryos (6 dpf) were fixed in 4% paraformaldehyde (PAF) overnight at 4°C, and maintained in 100% methanol at −20°C until processing. The embryos were washed with phosphate-buffered saline with 0.1% Tween-20 (PBST). The embryos were bleached in 30% hydrogen peroxide for 2 h, washed with PBST and transferred into Alcian blue solution. Embryos were stained overnight at room temperature. The embryos were rinsed four times with acidified ethanol (HCl–EtOH): 5% hydrochloric acid, and 70% ethanol. Embryos were rinsed for 20 min in HCl-EtOH and re-hydrated by washing 10 min in a HCl-EtOH/H2Od series (75, 25, 50, 50, 25, 75 and 100%). Embryos were stored in 1 ml of glycerol-KOH at 4°C. For microscopy, embryos were imaged in bright field mode using a dissection microscope (Leica M205 FA) Alizarin red staining in zebrafish to visualize extracellular calcium deposition Alizarin red S (C.I. 74240, Sigma) was prepared with 0.5% alizarin red S powder in water. The zebrafish embryos (6 dpf) were fixed in 4% PAF overnight at 4°C. The embryos were washed with 1 ml 50% ethanol, with rocking, at room temperature for 10 min. After removing the 50% ethanol, 1 ml of 0.5% alizarin red stain solution was added to the embryos and rocked at room temperature overnight. Bleach solution was prepared fresh by mixing equal volumes of 3% H2O2 and 2% KOH for final concentration of 1.5% H2O2 and 1% KOH, 1 ml was added to the embryos, which were incubated for 10 min. For clearing, embryos were rocked at room temperature for 30 min in 1 ml of a solution of 20% glycerol and 0.25% KOH, and then for 2 h in 1 ml of 50% glycerol and 0.25% KOH. Embryos were stored in 50% glycerol and 0.1% KOH at 4°C. For microscopy, embryos were embedded in 1% low-melting agarose (BioShop), and imaged using a Zeiss laser-scanning confocal microscope (Zeiss LSM 700). Cell biology studies Cell culture conditions. Primary patient fibroblasts and SLC10A7-V5 complemented fibroblasts were cultured in M199 medium (PAN Biotech, P04–07050) supplemented with 10% fetal calf serum and 1% penicillin/streptomycin at 37°C in humidity saturated 5% CO2 atmosphere. HeLa cell cultures were maintained in high glucose Dulbecco’s modified Eagle’s medium (DMEM) with Glutamax and sodium pyruvate (Gibco 31966021), supplemented with 10% fetal calf serum (GE Healthcare Life Sciences Hyclone SV30160) and 1% antibiotic-antimycotic (Gibco 15240062) at 37°C and 5% CO2. Prior to transfection, cells were dissociated with 2 mm EDTA in PBS. Cells were transfected using the Neon Transfection System (ThermoFisher) with 1 µg plasmid DNA per 1 × 105 cells, using the following settings: 1005 V pulse voltage, 35 ms pulse width, 2 pulses. Following electroporation, cells were transferred to 24-well plates containing pre-warmed Opti-MEM without phenol red (Gibco 11058021) for regeneration. After 4 h, Opti-MEM was replaced for regular HeLa medium as described earlier. Construction of plasmids. A SLC10A7 cDNA without a stopcodon (Genbank Accession EU831744, Refseq NM_001029998) cloned in the vector pDONR221 (plasmid ID HsCD00295767) was obtained from the PlasmID Repository at Harvard Medical School (plasmid.med.harvard.edu). The cDNA was cloned into the lentiviral expression vector pLenti6.2/V5-DEST (Invitrogen) by using Gateway technology (Invitrogen), creating a SLC10A7 open reading frame with a C-terminal V5-tag. This construct was used for both stable transfection of fibroblast cells by lentiviral transduction, and for transient transfection experiments of HeLa cells. The production of lentiviral particles using HEK 293FT cells, and subsequent infections of fibroblast cells followed by selection of transduced cells with blasticidin (InvivoGen) was performed as described before in (46). Subcellular localization of SLC10A7 SLC10A7-V5 complemented fibroblasts and SLC10A7-V5 transfected Hela cells were seeded on 12 mm-diameter coverslips. When ∼50% confluent, cells were washed three times with PBS and fixed with 4% PAF in PBS for 10 min. Subsequently, the cells were washed three times with PBS and stored in a parafilm-sealed 24-well plate at 4°C. Cells were permeabilized and blocked in 2.5% bovine serum albumin (BSA), 0.1% Triton X-100 in PBS for 15 min. Coverslips were incubated with primary antibodies against V5 (mouse anti-V5, Thermo Fisher, R960–25) and a specific marker diluted in 2.5% BSA, 0.1% Triton X-100 in PBS for 1 h at room temperature (rabbit anti-Calnexin 1:400, Abcam, ab22595; rabbit anti-ERGIC-53 1:100, Sigma, E1031; rabbit anti-β-COP 1:2000, Abcam, ab2899; rabbit anti-SEC31A 1:200, Sigma, HPA005457; rabbit anti-Giantin 1:1000, BioLegend, PRB-114C; rabbit anti-TGOLN2 1:500, Sigma, HPA012723). After washing the cells three times with 2.5% BSA, 0.1% Triton X-100 in PBS, cells were incubated with secondary antibodies goat–anti-rabbit Alexa Fluor 568 (1:2000, Invitrogen, A11011) and rabbit-anti-mouse Alexa Fluor 488 (1:2000, Thermo Fisher, A21204). Coverslips were washed three times with PBS and mounted on glass slides with ProLong Diamond Antifade Mountant with DAPI (Invitrogen, P36962) and stored at 4°C. Cells were imaged using a Zeiss LSM880 confocal microscope with a Plan-Apochromat 63x/1.4 Oil DIC M27 objective. Image analysis was performed with Fiji (ImageJ 2.0.0-rc-61/1.51n) and ColocalizeR v0.8b (http://colocalizer.iple.be). Analysis of Golgi morphology in SLC10A7 deficient fibroblasts Primary patient fibroblasts were seeded on 12-mm diameter coverslips. When ∼50% confluent, cells were washed three times with PBS and fixed with 4% PAF in PBS for 10 min. Subsequently, the cells were washed three times with PBS and stored in a parafilm-sealed 24-well plate at 4°C. Cells were permeabilized in 0.5% saponin in PBS for 10 min. Cells were subsequently washed three times with PBS-T and blocked in 3% BSA in PBS-T for 30 min. Coverslips were incubated with primary antibodies against GM130 (mouse anti-GM130 1:250, BD Biosciences, 610823) and TGOLN2 (rabbit anti-TGOLN2 1:500, Sigma, HPA012723) diluted in 3% BSA in PBS-T. After washing the cells three times with PBS-T, cells were incubated with secondary antibodies goat–anti-rabbit Alexa Fluor 568 (1:2000, Invitrogen, A11011) and rabbit-anti-mouse Alexa Fluor 488 (1:2000, Thermo Fisher, A21204) diluted in 3% BSA in PBS-T. Coverslips were washed three times with PBS-T and mounted on glass slides with ProLong Diamond Antifade Mountant with DAPI (Invitrogen, P36962) and stored at 4°C. Metabolic glycoprotein labeling in SLC10A7 deficient fibroblasts Primary skin fibroblasts were maintained in DMEM supplemented with 10% fetal bovine serum (Dutscher), at 37°C in humidity saturated 5% CO2 atmosphere. Fibroblasts from both healthy and SLC10A7 deficient patients were grown overnight on glass coverslips (12-mm diameter). Medium was then changed with pre-warmed medium containing 500 µm of N-(4-pentynoyl)neuraminic acid (SiaNAl). Labeling lasted 7 h. The labeling was stopped by fixing the cells with 4% PAF. Cells were then permeabilized in 0.5% Triton X-100 for 10 min. After washes, permeabilized cells were incubated with 100 µl/coverslip of a freshly prepared ‘click solution’ (K2HPO4, 100 mm; Sodium ascorbate, 2.5 mm; CuSO4, 150 µm; BTTAA, 300 µm; Azide-Fluor 545 [Sigma-Aldrich no. 760757], 10 µm) (47). The bioconjugation reaction was performed during 45 min in the dark, at room temperature with gentle shaking. The pool of fluorescent glycoconjugates was visualized through an inverted Zeiss-LSM780 confocal microscope. Pictures were taken using Zen Imaging software. For comparison purposes, each picture was taken under the same settings. Colocalization of mislocalized glycoproteins with early endosomes Anti-EEA1 antibody was from BD Biosciences (no. 610456, Europe). Anti-TMEM165 antibody was from Sigma-Aldrich (no. HPA038299, Europe). Polyclonal goat anti-rabbit or goat anti-mouse conjugated with Alexa Fluor were purchased from Invitrogen Molecular Probes (respectively no.A21038 and no. A11001, Europe). After the click chemistry reaction described before, it is also possible to start an immunolabeling. Fixed cells were first incubated in blocking buffer (2% fetal bovine serum, 2% BSA [Roche no. 10735086001], 0.2% Gelatin [Sigma-Aldrich no. G-8150] in PBS 1×) for 1 h at room temperature in the dark, then incubated for 1 h with primary polyclonal antibody against TMEM165 and primary monoclonal antibody against EEA1 diluted respectively at 1:300 and 1:100 in blocking buffer. After washes with PBS, cells were incubated for 1 h with secondary antibodies anti-rabbit Alexa Fluor 700 and anti-mouse Alexa Fluor 488 diluted at 1:600 in blocking buffer. Pulse chase labeling of glycoproteins in SLC10A7 deficient fibroblasts Fibroblasts from both healthy and SLC10A7 deficient patients were cultured in presence of 500 µm of our alkyne tagged sugar SiaNAl in DMEM during 7 h (pulse). Medium containing SiaNAl was then replaced by regular DMEM and the fibroblasts were grown for 48 h (chase). Then cells were fixed, permeabilized, and reacted with the fluorescent probe Azide-Fluor 545 in presence of CuSO4 (150 µm) and of the ligand BTTAA (300 µm). Fluorescence was detected through an inverted Zeiss-LSM780 confocal microscope. Pictures were taken using Zen Imaging software. For comparison purposes, each picture was taken under the same settings. Acknowledgements We would like to thank the EURO-CDG2 consortium for helpful discussions and collaborations, Frans van den Brandt, Marion Ybema-Antoine, Marit Pullen and Christina Hahnen for technical assistance, and the Genome technology Center of the Radboudumc for support in exome sequencing. C.B., D.V. and F.F. are indebted to the Research Federation FRABio (Univ.Lille, CNRS, FR 3688, FRABio, Biochimie Structurale et Fonctionnelle des assemblages Biomoléculaires) for providing the scientific and technical environment conducive to achieving this work. Conflict of Interest statement. None declared. Funding This work was supported by grants from the Dutch Organization for Scientific Research, ZONMW [Medium Investment Grant 40–00506-98–9001 and VIDI Grant 91713359 to D.J.L, VENI grant 722015012 to M.v.S.]; as well as funding support from the Natural Sciences and Engineering Research Council of Canada [grant RGPIN 05389–14 to X.Y.W.]; Brain Canada Foundation and Health Canada [grant PSG14–3505 to X.Y.W.];, Canada Foundation for Innovation [grant number 26233 to X.Y.W.] and Ministry of Health of Malaysia [grant number R02087 to N.A.B.]. This work was further supported by the European Union’s Horizon 2020 research and innovation program under the ERA-NET Cofund action N 643578 (EUROCDG-2). T.H., N.O. and H.H. were supported by grants MZ CR AZV 16–31932A and ProgresQ26/LF. Author contributions K.R., L.H., T.H., H.H., N.O., M.S., and F.V.S. recruited the patients, reviewed the clinical and radiographic features and obtained biologic materials from patients. CDG group provided the patient materials for the Table in Figure 1. A.A. performed exome sequencing, database studies, mRNA and biochemical studies. N.A.B., G.R.P.O., S.v.H., K.H. and R.B. performed glycomics experiments and bio-informatics analyses. M.N., P.T.A.L., G.v.d.B. performed localization studies. X.Y.W. and K.B.A. designed and performed zebrafish modeling studies. D.V., C.B. and F.F. performed chemical glycobiology experiments. M.K., R.S., S.T., R.R. and L.V.D.H. performed DNA sequencing, cloning, and MLPA analysis. D.J.L. conceived and coordinated the study. A.A., N.A.B., M.V.S. and D.J.L. wrote the article with contributions from all coauthors. All coauthors edited, reviewed and approved the final article. Web Resources Protein atlas: https://www.proteinatlas.org/ CADD: http://cadd.gs.washington.edu/ ExAC: http://exac.broadinstitute.org/ Transmembrane prediction: http://www.cbs.dtu.dk/services/TMHMM/ MLPA: MRC-Holland: http://www.mlpa.com ColocalizeR v0.8b: http://colocalizer.iple.be SIFT: http://sift.bii.a-star.edu.sg/ Align-GVGD: http://agvgd.hci.utah.edu/ PolyPhen-2 - Polymorphism Phenotyping version 2: http://genetics.bwh.harvard.edu/pph2/ Mutation Taster: http://www.mutationtaster.org/ References 1 de Ligt J. , Willemsen M.H. , van Bon B.W. , Kleefstra T. , Yntema H.G. , Kroes T. , Vulto-van Silfhout A.T. , Koolen D.A. , de Vries P. , Gilissen C. et al. ( 2012 ) Diagnostic exome sequencing in persons with severe intellectual disability . N. Engl. J. Med ., 367 , 1921 – 1929 . Google Scholar CrossRef Search ADS PubMed 2 Gilissen C. , Hehir-Kwa J.Y. , Thung D.T. , van de Vorst M. , van Bon B.W. , Willemsen M.H. , Kwint M. , Janssen I.M. , Hoischen A. , Schenck A. et al. ( 2014 ) Genome sequencing identifies major causes of severe intellectual disability . Nature , 511 , 344 – 347 . Google Scholar CrossRef Search ADS PubMed 3 Vissers L. , van Nimwegen K.J.M. , Schieving J.H. , Kamsteeg E.J. , Kleefstra T. , Yntema H.G. , Pfundt R. , van der Wilt G.J. , Krabbenborg L. , Brunner H.G. et al. ( 2017 ) A clinical utility study of exome sequencing versus conventional genetic testing in pediatric neurology . Genet. Med ., 19 , 1055 – 1063 . Google Scholar CrossRef Search ADS PubMed 4 Stark Z. , Tan T.Y. , Chong B. , Brett G.R. , Yap P. , Walsh M. , Yeung A. , Peters H. , Mordaunt D. , Cowie S. et al. ( 2016 ) A prospective evaluation of whole-exome sequencing as a first-tier molecular test in infants with suspected monogenic disorders . Genet. Med ., 18 , 1090 – 1096 . Google Scholar CrossRef Search ADS PubMed 5 Akawi N. , McRae J. , Ansari M. , Balasubramanian M. , Blyth M. , Brady A.F. , Clayton S. , Cole T. , Deshpande C. , Fitzgerald T.W. et al. ( 2015 ) Discovery of four recessive developmental disorders using probabilistic genotype and phenotype matching among 4, 125 families . Nat. Genet ., 47 , 1363 – 1369 . Google Scholar CrossRef Search ADS PubMed 6 Cummings B.B. , Marshall J.L. , Tukiainen T. , Lek M. , Donkervoort S. , Foley A.R. , Bolduc V. , Waddell L.B. , Sandaradura S.A. , O'Grady G.L. et al. ( 2017 ) Improving genetic diagnosis in Mendelian disease with transcriptome sequencing . Sci. Transl. Med ., 9 , 1 – 11 . Google Scholar CrossRef Search ADS 7 Tarailo-Graovac M. , Shyr C. , Ross C.J. , Horvath G.A. , Salvarinova R. , Ye X.C. , Zhang L.H. , Bhavsar A.P. , Lee J.J. , Drogemoller B.I. et al. ( 2016 ) Exome sequencing and the management of neurometabolic disorders . N. Engl. J. Med ., 374 , 2246 – 2255 . Google Scholar CrossRef Search ADS PubMed 8 Moremen K.W. , Tiemeyer M. , Nairn A.V. ( 2012 ) Vertebrate protein glycosylation: diversity, synthesis and function . Nat. Rev. Mol. Cell Biol ., 13 , 448 – 462 . Google Scholar CrossRef Search ADS PubMed 9 De Matteis M.A. , Luini A. ( 2011 ) Mendelian disorders of membrane trafficking . N. Engl. J. Med ., 365 , 927 – 938 . Google Scholar CrossRef Search ADS PubMed 10 Wopereis S. , Grunewald S. , Huijben K.M. , Morava E. , Mollicone R. , van Engelen B.G. , Lefeber D.J. , Wevers R.A. ( 2006 ) Transferrin and apolipoprotein C-III isofocusing are complementary in the diagnosis of N- and O-glycan biosynthesis defects . Clin. Chem ., 53 , 180 – 187 . Google Scholar CrossRef Search ADS PubMed 11 Tegtmeyer L.C. , Rust S. , van Scherpenzeel M. , Ng B.G. , Losfeld M.E. , Timal S. , Raymond K. , He P. , Ichikawa M. , Veltman J. et al. ( 2014 ) Multiple phenotypes in phosphoglucomutase 1 deficiency . N. Engl. J. Med ., 370 , 533 – 542 . Google Scholar CrossRef Search ADS PubMed 12 Van Scherpenzeel M. , Timal S. , Rymen D. , Hoischen A. , Wuhrer M. , Hipgrave-Ederveen A. , Grunewald S. , Peanne R. , Saada A. , Edvardson S. et al. ( 2014 ) Diagnostic serum glycosylation profile in patients with intellectual disability as a result of MAN1B1 deficiency . Brain J. Neurol ., 137 , 1030 – 1038 . Google Scholar CrossRef Search ADS 13 Sadat M.A. , Moir S. , Chun T.W. , Lusso P. , Kaplan G. , Wolfe L. , Memoli M.J. , He M. , Vega H. , Kim L.J.Y. et al. ( 2014 ) Glycosylation, hypogammaglobulinemia, and resistance to viral infections . N. Engl. J. Med ., 370 , 1615 – 1625 . Google Scholar CrossRef Search ADS PubMed 14 Hua S. , Williams C.C. , Dimapasoc L.M. , Ro G.S. , Ozcan S. , Miyamoto S. , Lebrilla C.B. , An H.J. , Leiserowitz G.S. ( 2013 ) Isomer-specific chromatographic profiling yields highly sensitive and specific potential N-glycan biomarkers for epithelial ovarian cancer . J. Chromatogr. A , 1279 , 58 – 67 . Google Scholar CrossRef Search ADS PubMed 15 van Scherpenzeel M. , Steenbergen G. , Morava E. , Wevers R.A. , Lefeber D.J. ( 2015 ) High-resolution mass spectrometry glycoprofiling of intact transferrin for diagnosis and subtype identification in the congenital disorders of glycosylation . Transl. Res. J. Lab. Clin. Med ., 166 , 639 – 649.e631 . Google Scholar CrossRef Search ADS 16 Park J.H. , Hogrebe M. , Gruneberg M. , DuChesne I. , von der Heiden A.L. , Reunert J. , Schlingmann K.P. , Boycott K.M. , Beaulieu C.L. , Mhanni A.A. et al. ( 2015 ) SLC39A8 Deficiency: a Disorder of Manganese Transport and Glycosylation . Am. J. Hum. Genet ., 97 , 894 – 903 . Google Scholar CrossRef Search ADS PubMed 17 Ng B.G. , Buckingham K.J. , Raymond K. , Kircher M. , Turner E.H. , He M. , Smith J.D. , Eroshkin A. , Szybowska M. , Losfeld M.E. et al. ( 2013 ) Mosaicism of the UDP-galactose transporter SLC35A2 causes a congenital disorder of glycosylation . Am. J. Hum. Genet ., 92 , 632 – 636 . Google Scholar CrossRef Search ADS PubMed 18 Edvardson S. , Ashikov A. , Jalas C. , Sturiale L. , Shaag A. , Fedick A. , Treff N.R. , Garozzo D. , Gerardy-Schahn R. , Elpeleg O. ( 2013 ) Mutations in SLC35A3 cause autism spectrum disorder, epilepsy and arthrogryposis . J. Med. Genet ., 50 , 733 – 739 . Google Scholar CrossRef Search ADS PubMed 19 Edmondson A.C. , Bedoukian E.C. , Deardorff M.A. , McDonald-McGinn D.M. , Li X. , He M. , Zackai E.H. ( 2017 ) A human case of SLC35A3-related skeletal dysplasia . Am. J. Med. Genet. A , 173 , 2758 – 2762 . Google Scholar CrossRef Search ADS PubMed 20 Marini C. , Hardies K. , Pisano T. , May P. , Weckhuysen S. , Cellini E. , Suls A. , Mei D. , Balling R. , Jonghe P.D. et al. ( 2017 ) Recessive mutations in SLC35A3 cause early onset epileptic encephalopathy with skeletal defects . Am. J. Med. Genet. A , 173 , 1119 – 1123 . Google Scholar CrossRef Search ADS PubMed 21 Jansen J.C. , Cirak S. , van Scherpenzeel M. , Timal S. , Reunert J. , Rust S. , Perez B. , Vicogne D. , Krawitz P. , Wada Y. et al. ( 2016 ) CCDC115 deficiency causes a disorder of Golgi homeostasis with abnormal protein glycosylation . Am. J. Hum. Genet ., 98 , 310 – 321 . Google Scholar CrossRef Search ADS PubMed 22 Bensimon-Brito A. , Cardeira J. , Dionisio G. , Huysseune A. , Cancela M.L. , Witten P.E. ( 2016 ) Revisiting in vivo staining with alizarin red S–a valuable approach to analyse zebrafish skeletal mineralization during development and regeneration . BMC Dev. Biol ., 16 , 2. Google Scholar CrossRef Search ADS PubMed 23 Godoy J.R. , Fernandes C. , Doring B. , Beuerlein K. , Petzinger E. , Geyer J. ( 2007 ) Molecular and phylogenetic characterization of a novel putative membrane transporter (SLC10A7), conserved in vertebrates and bacteria . Eur. J. Cell Biol ., 86 , 445 – 460 . Google Scholar CrossRef Search ADS PubMed 24 Bijsmans I.T. , Bouwmeester R.A. , Geyer J. , Faber K.N. , van de Graaf S.F. ( 2012 ) Homo- and hetero-dimeric architecture of the human liver Na(+)-dependent taurocholate co-transporting protein . Biochem. J ., 441 , 1007 – 1015 . Google Scholar CrossRef Search ADS PubMed 25 Jiang L. , Alber J. , Wang J. , Du W. , Yang X. , Li X. , Sanglard D. , Geyer J. ( 2012 ) The Candida albicans plasma membrane protein Rch1p, a member of the vertebrate SLC10 carrier family, is a novel regulator of cytosolic Ca2+ homoeostasis . Biochem. J ., 444 , 497 – 502 . Google Scholar CrossRef Search ADS PubMed 26 Peanne R. , Vanbeselaere J. , Vicogne D. , Mir A.M. , Biot C. , Matthijs G. , Guerardel Y. , Foulquier F. ( 2013 ) Assessing ER and Golgi N-glycosylation process using metabolic labeling in mammalian cultured cells . Methods in Cell Biology , 118 , 157 – 176 . Google Scholar CrossRef Search ADS PubMed 27 Gilormini P.A. , Lion C. , Vicogne D. , Levade T. , Potelle S. , Mariller C. , Guerardel Y. , Biot C. , Foulquier F. ( 2016 ) A sequential bioorthogonal dual strategy: manNAl and SiaNAl as distinct tools to unravel sialic acid metabolic pathways . Chem. Commun ., 52 , 2318 – 2321 . Google Scholar CrossRef Search ADS 28 Zamboni N. , Saghatelian A. , Patti G.J. ( 2015 ) Defining the metabolome: size, flux, and regulation . Mol. Cell , 58 , 699 – 706 . Google Scholar CrossRef Search ADS PubMed 29 Argmann C.A. , Houten S.M. , Zhu J. , Schadt E.E. ( 2016 ) A next generation multiscale view of inborn errors of metabolism . Cell Metab ., 23 , 13 – 26 . Google Scholar CrossRef Search ADS PubMed 30 Lauc G. , Essafi A. , Huffman J.E. , Hayward C. , Knezevic A. , Kattla J.J. , Polasek O. , Gornik O. , Vitart V. , Abrahams J.L. et al. ( 2010 ) Genomics meets glycomics-the first GWAS study of human N-Glycome identifies HNF1alpha as a master regulator of plasma protein fucosylation . PLoS Genet ., 6 , e1001256. Google Scholar CrossRef Search ADS PubMed 31 Claro da Silva T. , Polli J.E. , Swaan P.W. ( 2013 ) The solute carrier family 10 (SLC10): beyond bile acid transport . Mol. Aspects Med ., 34 , 252 – 269 . Google Scholar CrossRef Search ADS PubMed 32 Pettersson H. , Zarnegar B. , Westin A. , Persson V. , Peuckert C. , Jonsson J. , Hallgren J. , Kullander K. ( 2017 ) SLC10A4 regulates IgE-mediated mast cell degranulation in vitro and mast cell-mediated reactions in vivo . Sci. Rep ., 7 , 1085. Google Scholar CrossRef Search ADS PubMed 33 Larhammar M. , Patra K. , Blunder M. , Emilsson L. , Peuckert C. , Arvidsson E. , Ronnlund D. , Preobraschenski J. , Birgner C. , Limbach C. et al. ( 2015 ) SLC10A4 is a vesicular amine-associated transporter modulating dopamine homeostasis . Biol. Psychiatry , 77 , 526 – 536 . Google Scholar CrossRef Search ADS PubMed 34 Zhao Y. , Yan H. , Happeck R. , Peiter-Volk T. , Xu H. , Zhang Y. , Peiter E. , van Oostende Triplet C. , Whiteway M. , Jiang L. ( 2016 ) The plasma membrane protein Rch1 is a negative regulator of cytosolic calcium homeostasis and positively regulated by the calcium/calcineurin signaling pathway in budding yeast . Eur. J. Cell Biol ., 95 , 164 – 174 . Google Scholar CrossRef Search ADS PubMed 35 Brommage R. , Liu J. , Hansen G.M. , Kirkpatrick L.L. , Potter D.G. , Sands A.T. , Zambrowicz B. , Powell D.R. , Vogel P. ( 2014 ) High-throughput screening of mouse gene knockouts identifies established and novel skeletal phenotypes . Bone Res ., 2 , 14034. Google Scholar CrossRef Search ADS PubMed 36 Sasarman F. , Maftei C. , Campeau P.M. , Brunel-Guitton C. , Mitchell G.A. , Allard P. ( 2016 ) Biosynthesis of glycosaminoglycans: associated disorders and biochemical tests . J. Inherit. Metab. Dis ., 39 , 173 – 188 . Google Scholar CrossRef Search ADS PubMed 37 Smith C.E.L. , Poulter J.A. , Antanaviciute A. , Kirkham J. , Brookes S.J. , Inglehearn C.F. , Mighell A.J. ( 2017 ) Amelogenesis imperfecta; genes, proteins, and pathways . Front. Physiol ., 8 , 435. Google Scholar CrossRef Search ADS PubMed 38 Wright J.T. , Carrion I.A. , Morris C. ( 2015 ) The molecular basis of hereditary enamel defects in humans . J. Dent. Res ., 94 , 52 – 61 . Google Scholar CrossRef Search ADS PubMed 39 Zhang X. , Jiang S. , Mitok K.A. , Li L. , Attie A.D. , Martin T.F.J. ( 2017 ) BAIAP3, a C2 domain-containing Munc13 protein, controls the fate of dense-core vesicles in neuroendocrine cells . J. Cell Biol ., 216 , 2151 – 2166 . Google Scholar CrossRef Search ADS PubMed 40 Akita H. , Fukae M. , Shimoda S. , Aoba T. ( 1992 ) Localization of glycosylated matrix proteins in secretory porcine enamel and their possible functional roles in enamel mineralization . Arch. Oral Biol ., 37 , 953 – 962 . Google Scholar CrossRef Search ADS PubMed 41 Stránecký V. , Hoischen A. , Hartmannová H. , Zaki M.S. , Chaudhary A. , Zudaire E. , Nosková L. , Barešová V. , Přistoupilová A. , Hodaňová K. et al. ( 2013 ) Mutations in ANTXR1 cause GAPO syndrome . Am. J. Hum. Genet ., 92 , 792 – 799 . Google Scholar CrossRef Search ADS PubMed 42 Vissers L.E. , de Ligt J. , Gilissen C. , Janssen I. , Steehouwer M. , de Vries P. , van Lier B. , Arts P. , Wieskamp N. , del Rosario M. et al. ( 2010 ) A de novo paradigm for mental retardation . Nat. Genet ., 42 , 1109 – 1112 . Google Scholar CrossRef Search ADS PubMed 43 Kronewitter S.R. , de Leoz M.L. , Peacock K.S. , McBride K.R. , An H.J. , Miyamoto S. , Leiserowitz G.S. , Lebrilla C.B. ( 2010 ) Human serum processing and analysis methods for rapid and reproducible N-glycan mass profiling . J. Proteome Res ., 9 , 4952 – 4959 . Google Scholar CrossRef Search ADS PubMed 44 Sturn A. , Quackenbush J. , Trajanoski Z. ( 2002 ) Genesis: cluster analysis of microarray data . Bioinformatics , 18 , 207 – 208 . Google Scholar CrossRef Search ADS PubMed 45 Braak CJFT. , Smilauer P. ( 2012 ) Canoco Reference Manual and User's Guide: Software for Ordination, Version 5.0 . Microcomputer Power , Ithaca USA . 46 Renkema G.H. , Wortmann S.B. , Smeets R.J. , Venselaar H. , Antoine M. , Visser G. , Ben-Omran T. , van den Heuvel L.P. , Timmers H.J. , Smeitink J.A. et al. ( 2015 ) SDHA mutations causing a multisystem mitochondrial disease: novel mutations and genetic overlap with hereditary tumors . Eur. J. Hum. Genet ., 23 , 202 – 209 . Google Scholar CrossRef Search ADS PubMed 47 Besanceney-Webler C. , Jiang H. , Zheng T. , Feng L. , Soriano del Amo D. , Wang W. , Klivansky L.M. , Marlow F.L. , Liu Y. , Wu P. ( 2011 ) Increasing the efficacy of bioorthogonal click reactions for bioconjugation: a comparative study . Angewandte Chem ., 50 , 8051 – 8056 . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Journal

Human Molecular GeneticsOxford University Press

Published: Sep 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off