A missense mutation in the catalytic domain of O‐GlcNAc transferase links perturbations in protein O‐GlcNAcylation to X‐linked intellectual disability

X‐linked intellectual disabilities (XLID) are common developmental disorders. The enzyme O‐GlcNAc transferase encoded by OGT, a recently discovered XLID gene, attaches O‐GlcNAc to nuclear and cytoplasmic proteins. As few missense mutations have been described, it is unclear what the aetiology of the patient phenotypes is. Here, we report the discovery of a missense mutation in the catalytic domain of OGT in an XLID patient. X‐ray crystallography reveals that this variant leads to structural rearrangements in the catalytic domain. The mutation reduces in vitro OGT activity on substrate peptides/protein. Mouse embryonic stem cells carrying the mutation reveal reduced O‐GlcNAcase (OGA) and global O‐GlcNAc levels. These data suggest a direct link between changes in the O‐GlcNAcome and intellectual disability observed in patients carrying OGT mutations.

Although ID can be caused by environmental factors such as maternal alcohol abuse and birth complications [6,8,9], genetics plays a crucial role in its aetiology. Mutations, deletions or rearrangements affecting genes involved in development and neuronal function can have severe consequences for proper brain development or cognitive function [10].
To date, approximately 1000 genes have been shown to be involved in ID [11]. A disproportionate number (141) of these are located on the X-chromosome, leading to the coining of the term X-linked intellectual disability (XLID) [5]. In the last few years, next-generation sequencing (NGS) has led to enormous progress in deciphering monogenic forms of XLID [3][4][5]10]. This unbiased NGS approach has accelerated identification of de novo mutations in newly described XLID genes, including OGT, which encodes O-b-N-acetylglucosamine (GlcNAc) transferase (OGT).
OGT catalyses an essential post-transitional modification, the addition of a single GlcNAc onto serine or threonine of nucleocytoplasmic proteins, using UDP-GlcNAc as donor substrate [12]. OGT is also known to be involved in the processing of the host cell factor C1 (HCF1), the proteolytic products of which are involved in the activation of many genes involved in cell cycle progression [13][14][15]. The enzyme is divided into a glycosyltransferase catalytic domain and an Nterminal domain consisting of 13.5 tetratricopeptide repeats (TPRs) that have been shown to contribute to protein substrate binding [16][17][18]. Together with the hydrolase O-GlcNAcase (OGA), the homeostasis of dynamic protein O-GlcNAcylation is maintained for optimal cellular function [19]. Effects of protein O-GlcNAcylation have been linked to protein synthesis [20], stability [21] and turnover [22,23] and complex processes such as cell cycle progression [24], stress response [25,26] and transcription [27]. O-GlcNAcylation has also been implicated in pathology, including cancer [28], type II diabetes [29], cardiovascular disease [30] and neurodegeneration [31], and it is essential for proper vertebrate development [32][33][34][35]. In a range of model systems, O-GlcNAcylation has also been found to be critical for processes in early development. In mice, Ogt knockout causes lethality, with mouse embryos dying at the blastocyst stage [36].
O-GlcNAcylation also plays a key role in stem cell biology. Numerous pluripotency factors such as Oct4 and Sox2 have been shown to be O-GlcNAcylated [33,37]. Furthermore, mutations in OGT trigger alterations in stem cell differentiation and development, affecting neuronal lineages [37,38]. To date, five OGT mutations, all located to the N-terminal TPR domain, have been reported in patients with XLID [39][40][41][42]. These patients display developmental delay, facial dysmorphia, clinodactyly and microcephaly. To date, it is not clear whether the underpinning mechanisms leading to these phenotypes involve changes in the O-GlcNAc proteome.
Here, we report a patient with moderate ID, coarse facial features and a de novo mutation located in the catalytic domain of OGT. This mutation induces structural rearrangements in the catalytic domain, leading to reduced levels of protein O-GlcNAcylation in vitro and in mESCs, suggesting a direct link between perturbations in protein O-GlcNAcylation and XLID.

Results and Discussion
A patient with XLID possesses a missense mutation in the OGT catalytic domain The patient, E.T., was born at term with normal birth weight: 3354 g; length: 50 cm; head circumference: 34.5 cm; and an Apgar score of 9/9. He is the second child of an Estonian nonconsanguineous family. His elder brother is healthy. Facial asymmetry was noticed soon after birth (Fig. 1A). His toenails were also very small and with soft structure at birth. His development was evaluated as normal during the first months of life; he started to hold his head at 3 m and to turn at 6 m of age. Initially, he also had good weight gain, but after some months gastric reflux and gasses became very problematic.
Recurrent otitis started at 4 m, and after narcosis (the shunting of middle ears), his development arrested. His muscular tone became very flaccid. He started to turn at 12 months, to crawl at 14 months, to sit at 22 months and to walk at 3 years of age. First teeth came very late and slowly at 12 months. At the age of 16 month, alimentary vitamin B12 deficiency was diagnosed and treated with intramuscular injections. After that, his development improved significantly. However, at the age of 19 months persistent otitis recurred and his development arrested.
He was subjected to extensive clinical investigation at 19 months due to his developmental delay. He had coarse facial features with open mouth and drooling. He suffered from hypotonia with brisk tendon reflexes and truncal ataxia. There was a suspicion of epilepsy, but Electroencephalogram (EEG) was normal. Brain MRI was normal for that age. On cardiac evaluation, only very mild pericardial effusion was noticed, which did not affect cardiac function. Griffith scale evaluation suggested that his development corresponded to 8 months of age. He had serious problems with obstipation, but no obvious aetiology was detected. Extensive metabolic investigations were performed, which showed normal results (urinary organic acids, glycosaminoglycans, oligosaccharides, sialic acid, creatine/ guanidinoacetate and purine/pyrimidines; transferrin isoelectric focusing (TIEF) and acylcarnitine in serum; and neurotransmitters in cerebrospinal fluid). Chromosomal microarray analysis showed no abnormal copy number variations.
At the age of 7 years 8 months, his height was 127 cm (+1 SD), weight 23.4 kg (0 SD) and head circumference 52 cm (À0.5 SD). He is moderately intellectually disabled. He had coarse facial features, convergent strabismus, large ears and intense and constant drooling (Fig. 1B,C). He had no speech and moderately hyperactive behaviour, frequent body jerks, increased sensibility to light and sounds, and fears. In addition, he had hyperelastic connective tissue, which manifested as frequent joint dislocations and scoliosis. He presented mild T2-3 syndactyly, inverted nipples, abnormally pale skin and body temperature fluctuations in addition to his frequent infections. X-ray investigation showed cone-shape epiphyses of T2-T5 (Fig. 1D), fragmentation of the 1st rib and synostosis of 1-2 ribs. Brain MRI investigation at 5 years showed brain atrophy and mega cisterna magna (Fig. 1C). Ophthalmological investigation revealed astigmatism and myopia.
The N648Y mutation leads to structural changes in the catalytic domain Asn648 is highly conserved in OGT from C. elegans to H. sapiens (Fig. S1). Inspection of the human OGT crystal structure reveals that Asn648 maps to the interface of the OGT TPRs with the catalytic domain. The Asn648 side chain forms van der Waals interactions with that of Tyr642, while the loop between these two interacting residues (hereafter 642-648 loop) forms part of the composite OGT acceptor substrate binding cleft. It is thus possible that this mutation could affect the TPR-catalytic domain interface and lead to changes in the stability of the protein. We first analysed the effect of the mutation on the folding stability using differential scanning fluorimetry. Using an E. coli expression system, we purified a recombinant truncated form of OGT version containing the catalytic domain and 3.5 TPR repeats, for both wild-type (OGT WT ) and mutant (OGT N648Y ) OGT. Sigmoidal temperature-induced unfolding curves were obtained for both OGT WT and OGT N648Y with inflection points (T m ) of 44°C, indicative of identical unfolding temperatures (Fig. S2). Given that the mutation has the potential to alter the 642-648 loop, which forms an important part of the catalytic pocket, we next investigated possible structural changes by X-ray crystallography. Recombinant OGT N648Y was crystallised in the presence of the donor substrate analogue UDP-5S-GlcNAc and an acceptor peptide derived from the well-characterised OGT substrate TGF-betaactivated kinase 1-binding protein 1 (TAB1). Diffraction data were collected to 2.2 A, and initial refinement started from the published OGT WT ternary structure [43]. Mutation of Asn648 to the bulkier tyrosine appears to be accommodated without disruption of the domain organisation observed in the wild-type structure (RMSD of 0.4 A on all backbone Ca atoms) (Fig. 2B). However, the aromatic Tyr648 side chain forms a p-p stacking interaction with Tyr642 causing positional shifts of up to 0.7 A in the 642-648 loop. Strikingly, despite the presence of an acceptor peptide in the crystallisation condition, no corresponding electron density was observed in the OGT active site. This could be because the changes in 642-648 loop caused by the mutation affect the ability of OGT to bind to acceptor substrates (Fig. 2C). To explore this possibility, we used a recently established fluorescence polarimetry assay (FP) where increasing concentrations of OGT are titrated against a fixed concentration of a fluorescent probe that incorporated elements of both peptide and nucleotide substrates [44]. The direct binding affinity of the probe to OGT WT yielded a Kd value of 1.6 µM as previously reported, whereas the Kd value for the OGT N648Y was 7 times higher (11 µM), suggesting reduced substrate binding in agreement with the structural observations (Fig. 2D). In summary, the N648Y mutation may lead to changes in the OGT acceptor substrate binding cleft and affect substrate binding.
OGT N648Y does not glycosylate the model acceptor substrate TAB1 Given the absence of effects on stability and the apparent effect of the N648Y mutation in OGT substrate binding, it is possible that catalytic activity is affected. To investigate this, we used an enzyme assay where we incubated OGT WT or OGT N648Y with the well-characterised substrate TAB1. This substrate is efficiently monoglycosylated on Ser395 by wild-type OGT [45]. Western blotting analysis using a specific antibody which binds the O-GlcNAcylated form of TAB1 [45] showed that the OGT N648Y variant was unable to glycosylate the substrate (Fig. 2E), in contrast to the efficient glycosylation observed by OGT WT . Thus, OGT N648Y does not glycosylate the model acceptor substrate TAB1.

The N648Y mutation leads to hypoglycosylation in mES cells
Maintenance of O-GlcNAc homeostasis is essential for optimal cellular function, and perturbations in protein O-GlcNAcylation have been implicated in pathogenesis. We next investigated the effects of reduced OGT N648Y activity on O-GlcNAc homeostasis and levels of OGA/ OGT. We used CRISPR/Cas9 genome editing to introduce the N648Y mutation in the endogenous ogt locus carrying an additional N-terminal triple HA (3HA) tag (as described in detail in Materials and methods). Two independent clones each were obtained for mESCs carrying 3HA-tagged wild-type and 3HA-tagged mutant OGT, as verified by diagnostic restriction digest and sequencing. Western blotting analyses of 3HA-N648Y mESC cells were carried out to analyse OGT, OGA and protein O-GlcNAcylation levels (Fig. 3A). In agreement with loss of in vitro activity towards the TAB1 substrate (Fig. 2E) and the potential effect of the loop 642-648 on OGT substrate binding (Fig. 2D), this experiment revealed significantly reduced O-GlcNAc levels in 3HA-N648Y mESCs compared to the controls. This is evidence for a possible link between perturbation in protein O-GlcNAcylation and XLID. No reduction in OGT protein levels was detected. However, OGA levels appeared reduced in 3HA-N648Y mESCs (Fig. 3B). This hints at the existence of a compensatory mechanism that allows cells to partially maintain O-GlcNAc homeostasis by reducing OGA levels-as also observed in some of the other OGT XLID mutations in the TPR domain [39,46]. Thus, the N648Y mutation leads to hypoglycosylation in mES cells.

Concluding remarks
Protein O-GlcNAcylation has emerged as an essential PTM involved in disease and more recently in neuronal function and development [19,32,33,41]. Recent NGS studies have led to the identification of mutations in OGT that give rise to XLID, suggesting a direct link between O-GlcNAc and regulation of brain development. However, experimental evidence for a link between significant loss of OGT activity and changes in the O-GlcNAc proteome has remained elusive, presumably because of the location of these mutations in the N-terminal TPR substrate binding domain of OGT. Here, we report a patient with XLID attributable to an OGT missense mutation (Asp648Tyr), which maps to the OGT catalytic domain (Fig. 2A). In agreement with the five previously reported patients, the patient shows reduced IQ, limited speech, developmental delay, facial dysmorphia and hypotonia.
We initially delineated the effects of the mutation on the stability, structure and activity of the enzyme using in vitro approaches. Unlike the previously reported OGT XLID mutations, we were not able to detect changes in folding stability induced by the mutation (Fig. S2). Using X-ray crystallography, we revealed that the mutation did not induce large global conformational changes. However, the observed local conformational changes in the OGT acceptor substrate binding cleft could affect activity. Indeed, using in vitro analyses, the mutant led to significant reduction in the binding affinity of a bisubstrate fluorescence polarimetry probe and defects in the glycosylation of TAB1, suggesting a direct effect of the mutation on enzymatic activity (Fig. 2D,E).
To extend our investigations, we introduced the patient mutation into a male mouse embryonic stem cell line. In previous studies, examination of the steadystate global OGT, OGA and O-GlcNAc levels in different cell lines showed no significant changes in protein O-GlcNAc and OGT, while OGA levels appeared reduced in most of the analysed cell lines, suggesting a compensatory mechanism between OGT and OGA to maintain O-GlcNAc homeostasis (Fig. 3). To our surprise, the 3HA-N648Y mESC lines showed not only reduced levels of OGA, but also reduced protein O-GlcNAcylation levels. This suggests that perturbations in O-GlcNAc cycling may contribute to the patient ID phenotype. This now enables the search for specific proteins and O-GlcNAc sites, whose altered O-GlcNAcylation levels can be linked to cellular processes that could be causative for the observed phenotypes.
There is currently only a very limited number (6) of XLID patients identified that carry different mutations in OGT. It is thus not yet feasible to consider possible treatments. An interesting clue to a possible treatment is the observed changes in OGA levels. This is a mechanism that may compensate for loss in OGT activity or protein levels. This would lead to the hypothesis that (further) inhibition of OGA is a possible route towards re-establishing full O-GlcNAc homeostasis and possible alleviation of the patient phenotypes. Another route would be to increase cellular production of UDP-GlcNAc, which has been shown to directly lead to increased levels of protein O-GlcNAcylation. It has been demonstrated that this is possible to achieve by feeding cells with the UDP-GlcNAc precursor glucosamine, which enters the hexosamine biosynthetic pathway as glucosamine-6-phosphate and leads to increased UDP-GlcNAc levels [47][48][49][50]. Glucosamine, in the form of glucosamine sulfate, is a food supplement frequently taken by patients suffering from osteoarthritis, albeit with limited clinical evidence for significant benefits [51,52]. In an initial experiment, at the age of 8 years of our patient we started the treatment with glucosamine sulfate (400 mg three times a day). This treatment has been performed over a period of 13 months without any adverse effects (unpublished data). Parents have noticed that the patient is more active; he has more energy and has increased plasticity in movements. He started to use syllables, and teeth development progressed. However, there are no changes in clinical biochemical analyses, and in the absence of a proper randomised trial, it is not possible to attribute these improvements to the treatments. Nevertheless, this is an approach that could be evaluated once sufficient patient numbers are available.

Protein expression and purification
Truncated (323-1044) OGT constructs were expressed in E. coli BL21(DE3) RIPL cells as N-terminal GST fusion protein, as described previously [43]. Briefly, transformed E. coli cells were grown in LB broth at 37°C with agitation until OD 600 reached 0.8, at which point the temperature was lowered to 18°C and expression was induced by addition of 100 µM IPTG. Cells were lysed using French Press in base buffer (0.1 M Tris/HCl, pH 7.5, 0.15 M NaCl, 0.5 mM TCEP) supplemented with 25 mM imidazole, 0.1 mgÁmL À1 DNase I and protease inhibitor cocktail (1 mM benzamidine, 0.2 mM PMSF, 5 mM leupeptin). Then, recombinant proteins were affinity-purified following manufacturers' guidelines. After cleavage of affinity tags, dialysed protein was loaded onto 5 mL HiTrap Q Sepharose FF anion exchange resin (GE Healthcare) and eluted with a linear gradient up to 60% of buffer B (0.1 M Tris/HCl, pH 8.5, 500 mM NaCl). Peak fractions were pooled, concentrated and further purified via size exclusion chromatography using a 300 mL Superdex TM 200 column (GE Healthcare) equilibrated with base buffer. The peak fractions were concentrated to 10 mgÁmL À1 . For crystallisation, truncated mutant OGT was used fresh at 7 mgÁmL À1 concentration. For all other purposes, proteins were concentrated to 10 mgÁmL À1 , mixed 1 : 1 with 50% glycerol, snap-frozen and stored at À80°C until use.>

Structure solution
Crystallisation of truncated OGT N648Y (residues 323-1044) was performed as described previously [43]. Briefly, experiments were performed at 22°C using 24-well hanging drop crystallisation plates, by combining 1 µL drops containing 7 mgÁmL À1 OGT N648Y (in base buffer), 3 mM UDP-5S-GlcNAc and 3 mM acceptor peptide derived from TAB1 sequence (PVSVPYSSAQSTS) with 2 µL of reservoir solution (1.45 M K 2 HPO 4 , 8 mM EDTA and 1% xylitol). Large, clear-faced crystals appeared overnight. Prior to diffraction experiments, individual crystals were cryoprotected in reservoir solution supplemented with 3.5 M maleic acid, 3 mM UDP-5S-GlcNAc and 3 mM acceptor peptide and flash-frozen in liquid nitrogen. Diffraction data were collected at the European Synchrotron Radiation Facility beamline ID23. Data were processed with XDS [53] and scaled to 2.2 A using Scala [54]. The structure was solved by molecular replacement using the structure for OGT WT (PDB: 5C1D; [43]) as the search model. The resulting model was manually refined using Coot [54] and REFMAC [55], respectively. Although the mutant protein was crystallised in the presence of donor and acceptor, there was no evidence for the latter in the electron density maps and was therefore excluded from the model. Coordinates and structure factors were deposited in the PDB (PDB: 6Q4M). Scaling and model building statistics can be seen in Table 1.

Fluorescence polarimetry assay
To evaluate the expected binding differences between the OGT WT and the OGT N648Y , we titrated a previously published fluorescently labelled UDP-peptide bisubstrate conjugate [56] with increasing concentrations of either OGT WT or the OGT N648Y in 0.1 M Tris/HCl pH 7.5, 0.15 M NaCl, 0.5 mM TCEP and 5% DMSO. Each sample, containing 25 lL, was incubated in the dark for 30 min before the read-out with a PHERAstar plate reader (BMG LAB-TECH). Subsequent data analysis was performed using GRAPHPAD PRISM 6 as outlined previously [56].

Activity assays
O-GlcNAcylation assays were performed on TAB1 protein (residues 7-420). TAB1 (1 µM) was combined with fulllength OGT WT or OGT N648Y (0.1 µM) in the presence of 100 µM UDP-GlcNAc. The reaction mixtures were incubated at 37°C for 2-50 min and subsequently stopped by addition of LDS loading buffer (4x, Thermo Fisher Scientific). Proteins were resolved by SDS/PAGE (4-12% acrylamide [Life Technologies]) and transferred onto nitrocellulose membranes (GE Healthcare). After antibody treatment, progress of the reaction was visualised using LI-COR Odyssey Scanner and associated quantification software.
Generation of 3HA-tag N567K mES cell line For the generation of the 3HA mES cell line, we transfected wild-type male E14-TG2a.IV AW2 mES cells with pBABED puro U6 and pX335 (Cas9 D10A) vectors containing the gRNA sequences selected using WTSI Genome Editing [58]. Silent mutations were designed in addition to the intended mutations, removing PstI and BfmI restriction sites (Fig. S2). A gene block containing these changes was obtained from IDT (International DNA technologies). The gene block was introduced into the cloned 2 kb region by restriction-free cloning [59] and then confirmed by DNA sequencing [59]. For transfection, 2 9 10 5 cells were seeded onto gelatine-coated 24-well plates and transfected using 0.33 µg of each vector and Lipofectamine 3000 according to manufacturer's instructions. After 24 h, media was replaced and puromycin (3 µgÁmL À1 ) was added to the cells for selection until control untransfected cells were all dead (48 h). Then, the cells were replated into 10-cm plates for recovery. After that, single-cell selection was carried out using limited dilution according to 0.3 cell per well in 96well plates. For confirmation of mutations, diagnostic restriction digests and genomic DNA sequencing analysis were carried out. For the restriction fragment length polymorphism assay, 3HADiag_F and 3HADiag_R primers were used to amplify by PCR the mutated site and the silent mutation which eliminates a PstI restriction site (Table S1). The size of the PCR product (546 bp vs 450 bp for wild-type) was used to screen for successful integration. To further confirm, the PCR product was then digested using PstI or BfmI. Clones negative for the restriction enzyme assay were then sequenced to confirm the presence of the modification.
For generation of the 3HA-OGT N648Y mESCs, we transfected the previously generated 3HA mESCs with pBABED puro U6 and pX335 (Cas9 D10A) vectors containing the gRNA sequences using the same procedure we previously used for the insertion of the 3HA-tag (Fig. S3). Restriction enzymes and genomic DNA sequencing analysis were used for mutation identification. For the restriction fragment length polymorphism assay, MouseEstDiag_F and Mou-seEstDiag_R primers were used to amplify by PCR the mutated site and the silent mutation which eliminates an Eco130I restriction site (Table S2). The size of the bands following digestion (167 bp and 433 bp for wild-type versus 600 bp for mutant) was used to screen for successful integration. Clones negative for cutting in the restriction

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Fig. S1. Sequence alignment shows the highly conserved Asparagine 648 from H. sapiens to C. elegans. Fig. S2. Effects of the N648Y mutation on unfolding temperature. Fig. S3. Gene-editing of mouse ES cells to introduce 3HA-tag into the endogenous OGT gene. Fig. S4. Gene-editing of mouse ES cells to introduce N648Y mutation into endogenous OGT gene. Table S1. Primers and geneblock used for introducing 3HA-tag to OGT gene and genotyping candidate 3HA-OGT WT mES cell line. Table S2. Primers and geneblock used for introducing N648Ymutation to OGT gene and genotyping candidate 3HA-OGT N648Y mES cell line.