Mutations in N-acetylglucosamine (O-GlcNAc) transferase in patients with X-linked intellectual disability

Abstract

N-Acetylglucosamine (O-GlcNAc) transferase (OGT) regulates protein O-GlcNAcylation, an essential and dynamic post-translational modification. The O-GlcNAc modification is present on numerous nuclear and cytosolic proteins and has been implicated in essential cellular functions such as signaling and gene expression. Accordingly, altered levels of protein O-GlcNAcylation have been associated with developmental defects and neurodegeneration. However, mutations in the OGT gene have not yet been functionally confirmed in humans. Here, we report on two hemizygous mutations in OGT in individuals with X-linked intellectual disability (XLID) and dysmorphic features: one missense mutation (p.Arg284Pro) and one mutation leading to a splicing defect (c.463-6T>G). Both mutations reside in the tetratricopeptide repeats of OGT that are essential for substrate recognition. We observed slightly reduced levels of OGT protein and reduced levels of its opposing enzyme O-GlcNAcase in both patient-derived fibroblasts, but global O-GlcNAc levels appeared to be unaffected. Our data suggest that mutant cells attempt to maintain global O-GlcNAcylation by down-regulating O-GlcNAcase expression. We also found that the c.463-6T>G mutation leads to aberrant mRNA splicing, but no stable truncated protein was detected in the corresponding patient-derived fibroblasts. Recombinant OGT bearing the p.Arg284Pro mutation was prone to unfolding and exhibited reduced glycosylation activity against a complex array of glycosylation substrates and proteolytic processing of the transcription factor host cell factor 1, which is also encoded by an XLID-associated gene. We conclude that defects in O-GlcNAc homeostasis and host cell factor 1 proteolysis may play roles in mediation of XLID in individuals with OGT mutations.

Keywords: Congenital Disorders of Glycosylation; Host Cell Factor 1 (HCF-1); O-GlcNAcylation; O-linked N-acetylglucosamine (O-GlcNAc) transferase (OGT); X-linked Intellectual Disability; glycobiology; glycosyltransferase; metabolic disease.

Figure 1.

Clinical images from patients with hemizygous OGT mutations. A, picture of patient 1, showing dysmorphic features. B, pictures of patient 2, showing dysmorphic features. C, MRI image of patient 1, showing mild abnormalities pointing toward periventricular leukomalacia, indicated by red arrowheads. D, MRI image of patient 2, showing no brain abnormalities.

Figure 2.

Genetic analysis of hemizygous OGT mutations detected in patients with intellectual disability syndrome. A, location of known and newly discovered genetic variants in the OGT gene and their presence in families 1 and 2. B, RT-PCR in skin-derived fibroblasts from patient 2 bearing the c.463–6T>G mutation confirmed by sequence analysis. C, domain organization in human O-GlcNAc transferase. TLR, tetratricopeptide-like repeat. D, model for the full-length human O-GlcNAc transferase generated by superposition of crystallographic models for the human O-GlcNAc transferase catalytic domain (Protein Data Bank code 5V1D) and tetratricopeptide repeat domain (Protein Data Bank code 1W3B). Blue represents tetratricopeptide repeats, yellow represents tetratricopeptide-like repeat, and pink represents catalytic core.

Figure 3.

Characterization of global O-GlcNAc levels in patient-derived skin fibroblasts. A, immunoblot showing global O-GlcNAc levels in patient-derived fibroblasts detected using RL2 anti O-GlcNAc antibody. C, control; P, patient. B, the scatter plot indicates fractional ratio of any given actin-normalized signal averaged from three biological replicates, with the error bars representing the standard deviation. p values (Mann–Whitney U test): C1–P1 = 0.4; C1–P2 = 0.1; C2–P1 = 0.4; C2–P2 = 0.9. Achieved power (1 − β error probability) = 0.1. C, immunoblot showing global O-GlcNAc levels in patient-derived fibroblasts detected using the far Western method. The far Western method relies on detection of O-GlcNAcylated protein using a point mutant of a Halo-tagged bacterial O-GlcNAcase homolog, CpOGA^D298N, as a probe (32). Halo antibody was used to detect the bands where the probe was localized. D, the scatter plot indicates the fractional ratio of any given actin-normalized signal averaged from three biological replicates, with the error bars representing the standard deviation. p values (Mann–Whitney U test): C1–P1 = 0.9; C1–P2 = 0.7; C2–P1 = 0.7; C2–P2 = 0.9. Achieved power (1 − β error probability) = 0.1.

Figure 4.

Characterization of OGT and OGA levels in patient-derived skin fibroblasts. A, gene expression of OGT and OGA in patient fibroblasts measured by qPCR. Scatter plots represent mean data from two independent experiments. Gene expression was calculated according to ΔΔC(t) method. Tubulin expression was used as reference, and the data were normalized against control 1. B, immunoblot showing O-GlcNAc transferase (OGT) levels in patient-derived fibroblasts. C, control; P, patient. C, the scatter plot indicates fractional ratio of any given tubulin-normalized signal averaged from three biological replicates, with the error bars representing the standard deviation. p values (Mann–Whitney U test): C1–P1 = 0.9; C1–P2 = 0.1; C2–P1 = 0.9; C2–P2 = 0.7. Achieved power (1 − β error probability) = 0.1. D, immunoblot showing O-GlcNAcase (OGA) levels in patient-derived fibroblasts. E, the scatter plot indicates fractional ratio of any given tubulin-normalized signal averaged from three biological replicates, with the error bars representing the standard deviation. p values (Mann–Whitney U test): C1–P1 = 0.2; C1–P2 = 0.2; C2–P1 = 0.2; C2–P2 = 0.2. Achieved power (1 − β error probability) = 0.1. F, immunoblot comparing OGT protein molecular weight, as present in fibroblasts of patient 2 (Δ155–177) and of recombinantly expressed wild-type OGT (W) and Δ155–177-OGT (Δ).

Figure 5.

In vitro analysis of recombinantly expressed mutant OGT forms. A, thermal denaturing curve showing fraction of unfolded wild-type (TPR-WT), R284P (TPR-R284P), and Δ155–177 (TPR-Δ155–177) OGT TPR domains as a function of time. The data were fitted to Boltzmann sigmoidal curve equation. Error bars represent standard error of mean of seven replicates, and the plot is representative of three biological replicates. B, immunoblot showing the relative O-GlcNAcylation activities of wild-type OGT (W) and R284P-OGT (R) against de-O-GlcNAcylated HEK-293 lysate. C, domain organization of human HCF1 compared with its truncated form fused to N-terminal GST tag and C-terminal His tag (HCF1-rep1), which was used as a substrate for OGT proteolytic activity. D, immunoblot showing the relative proteolytic activities of wild-type OGT (W) and R284P-OGT (R) against HCF1-rep1. The samples were treated with an O-GlcNAcase homolog from C. perfringens (CpOGA) either toward the end of the reaction (+) or from the beginning (++). Closed circle, unprocessed HCF-1rep1; closed square, HCF1-rep1 cleavage product. E, kinetic assay of the proteolytic activity of wild-type OGT and R284P-OGT against HCF1-rep1 plotted as amount of product formed as a function of time. Error bars represent standard error of mean of four replicates. The data were fitted to hyperbolic curve equation.