Deciphering the Functions of Protein O-GlcNAcylation with Chemistry

Abstract

O-GlcNAcylation is the modification of serine and threonine residues with β-N-acetylglucosamine (O-GlcNAc) on intracellular proteins. This dynamic modification is attached by O-GlcNAc transferase (OGT) and removed by O-GlcNAcase (OGA) and is a critical regulator of various cellular processes. Furthermore, O-GlcNAcylation is dysregulated in many diseases, such as diabetes, cancer, and Alzheimer’s disease. However, the precise role of this modification and its cycling enzymes (OGT and OGA) in normal and disease states remains elusive. This is partially due to the difficulty in studying O-GlcNAcylation with traditional genetic and biochemical techniques. In this review, we will summarize recent progress in chemical approaches to overcome these obstacles. We will cover new inhibitors of OGT and OGA, advances in metabolic labeling and cellular imaging, synthetic approaches to access homogeneous O-GlcNAcylated proteins, and cross-linking methods to identify O-GlcNAc-protein interactions. We will also discuss remaining gaps in our toolbox for studying O-GlcNAcylation and questions of high interest that are yet to be answered.

Graphical Abstract

O-GlcNAcylation is the attachment of O-linked β-N-acetyl glucosamine (O-GlcNAc) onto serine or threonine residues of numerous intracellular proteins (Figure 1). This essential post-translational modification (PTM) was an unexpected discovery by the Hart lab in 1984¹ and was later found ubiquitously in all eukaryotes except yeast.^2,3 Distinct from the cell surface glycans that mainly exist as oligosaccharides or polysaccharides, O-GlcNAcylation is a unique monosaccharide modification in the nucleus and cytoplasm.⁴ To date, over 4000 O-GlcNAcylated proteins have been reported.⁵ The glycosylation of these proteins is all catalyzed by a single pair of opposing enzymes: O-GlcNAc transferase (OGT) transfers O-GlcNAc from the sugar donor uridine diphosphate N-acetylglucosamine (UDP-GlcNAc) to various protein substrates,⁶ while O-GlcNAcase (OGA) removes this modification⁷ (Figure 1). This stands in great contrast to phosphorylation, which is controlled by hundreds of kinases and phosphatases.⁸ Although the molecular mechanisms regulating the O-GlcNAc cycling enzymes remain a challenging ongoing question, emerging evidence points to modulation of OGT by PTMs^9,10 and protein–protein interactions.^11,12 OGT’s activity is also highly regulated by cellular concentrations of UDP-GlcNAc,¹³ which is derived from glucose and incorporates fatty acid, nucleotide, and amino acid metabolites via the hexosamine biosynthetic pathway (HBP; Figure 1).¹⁴ Consequently, O-GlcNAcylation level changes in rapid response to the fluctuation of the nutrient and metabolic status of the cell to modulate proteins’ stability, activity, and localization. Taken together, O-GlcNAcylation serves as a nutrient sensor that dynamically regulates protein functions.

Figure 1.

O-GlcNAcylation cycle and the biological processes regulated by dynamic O-GlcNAcylation in cells. HBP: hexosamine biosynthetic pathway. Pol II: RNA polymerase II. SNAP-29: synaptosomal-associated protein 29. CaMKII: Ca²⁺/calmodulin-dependent protein kinase II.

O-GlcNAcylation plays significant roles in a broad range of cellular processes, and here we highlight a few of them from recent publications (Figure 1). Previous studies have uncovered the functional impacts of O-GlcNAcylation on a number of transcription factors including p53,¹⁵ FoxO1,¹⁶ and Sp1.¹⁷ In contrast, the investigation of O-GlcNAcylation on the central transcriptional machinery RNA polymerase II (Pol II) remains a big challenge, as numerous glycosylation sites are located on its highly repetitive C-terminal domain (CTD)^18,19 and no specific O-GlcNAcylated CTD antibody is currently available. Despite this, in vitro evidence has suggested that O-GlcNAcylated Pol II facilitates the assembly of transcription preinitiation complex (PIC) onto DNA promoters.²⁰ In addition, OGT was found in the protein complex with Pol II and enriched at promoter regions in cells. More recently, OGA was reported as an important component of the transcription elongation complex.²¹ O-GlcNAcylation also complements phosphorylation as a modulator of signal transduction. In an interesting example, high glucose conditions induced O-GlcNAcylation of Ca²⁺/calmodulin-dependent protein kinase II (CaMKII). This activates the kinase to regulate downstream signaling, indicating that O-GlcNAc directly links intracellular glucose level to signal transduction (Figure 1).²² Additional examples have been extensively reviewed.^23,24 O-GlcNAcylation also influences cytoskeletal motor proteins such as mitochondrial motor-adaptor protein milton.²⁵ This occurs in response to high glucose stimulation and tunes the motility of mitochondria in neurons (Figure 1). O-GlcNAcylation is likewise involved in autophagy, as O-GlcNAcylation of synaptosomal-associated protein 29 (SNAP-29) regulates the maturation of autophagosome in accordance with nutrient levels (Figure 1).²⁶ Moreover, cellular feeding is in part governed by O-GlcNAcylation, particularly through insulin production and signaling.^27,28 Surprisingly, OGT also appears to have proteolytic activity, as it was found to cleave host cell factor 1 (HCF-1) in the same active site that catalyzes O-GlcNAcylation.²⁹ This activates HCF-1 to become a key regulator of the cell cycle.³⁰ A more recent study reported that the proteolysis of HCF-1 involves unprecedented O-GlcNAcylation on the cleaving glutamate residue.³¹ Overall, the discoveries highlighted here represent only a handful of the cellular processes that O-GlcNAcylation and its cycling enzymes regulate, underscoring their broad impacts in cell physiology.

Growing evidence supports an essential role of O-GlcNAcylation and its cycling enzymes in human diseases. Increased OGT and O-GlcNAcylation levels occur in various types of cancer and have been found to promote cancer growth and progression.³² In diabetic cells, a number of insulin signaling proteins and mitochondrial proteins are also hyper O-GlcNAcylated.^22,33 Alzheimer’s disease is thought to arise from tau protein oligomerization, which is promoted by its hyperphosphorylation.^34,35 O-GlcNAcylation and phosphorylation of tau have been shown to be mostly reciprocal in vitro, and tau O-GlcNAcylation appears to reduce its oligomerization in vivo.³⁶ Despite the significance of O-GlcNAcylation in normal and disease states, our mechanistic understanding on many of these processes remains elusive, such as the regulation of OGT substrate specificity and the functional impacts of its dysregulation in pathogenesis. Limiting our ability to answer these questions is the difficulty of studying O-GlcNAcylation in a specific manner using traditional genetic and biochemical approaches, such as gene deletion or routine profiling of the proteome. In this review, we will highlight recently developed chemical approaches to detect O-GlcNAcylation and investigate its functions, serving as new avenues to overcome the inherent challenges of this modification.

O-GLCNAC CYCLING ENZYMES

OGT and Its Inhibitors.

OGT is a multidomain protein belonging to the 41 family of glycosyltransferase in the CAZy database (Figure 2a). Kinetic and structural studies have established that OGT glycosylates substrates in an ordered bi-bi mechanism.^37–39 OGT binds UDP-GlcNAc first, followed by the peptide substrate. After sugar transfer, glycosylated peptide leaves prior to UDP. The binding site for UDP-GlcNAc is primarily comprised of two lobes of the catalytic domain (N-Cat and C-Cat in Figure 2a), separated by the intervening domain (Int-D), which is conserved in metazoans with its function remaining to be characterized.³⁷ A unique structural feature of OGT is the N-terminal tetratricopeptide repeat (TPR) domain, which forms an extended α-helical tunnel and has been suggested to participate in protein substrate binding.⁴⁰ The length of the TPR domain distinguishes the three splicing isoforms of OGT (ncOGT, mOGT, and sOGT) found in humans (Figure 2a), although differences in their functions are mostly unclear. The substrate recognition of OGT remains an open question as no consensus sequence motif has been identified in substrate proteins. To address this, efforts have focused on the crystallization of OGT-substrate complexes. The first crystal structure of human OGT was obtained with a truncated construct containing 4.5 repeats of the 13.5 TPRs of ncOGT (called OGT_4.5) in complex with UDP and CKII peptide substrate.³⁷ Following that, additional complex structures of OGT_4.5 with TAB1, HCF-1, RBL2, Ret, keratin-7, and lamin B1 peptides have been solved.^29,39,41 These structures have revealed that short peptides are mainly anchored by UDP-GlcNAc bound in the OGT active site, as well as backbone interactions with adjacent TPR residues. Limited interactions with peptide side-chains may explain the tolerance of OGT for a broad range of substrates. The structures have also assisted in designing OGT inhibitors, which are discussed below. While OGT_4.5 is fully capable of glycosylating a number of peptide substrates, it is incompetent in glycosylating protein substrates, indicating that the extended TPR domain is crucial for protein substrate binding. Because of the flexible nature of TPR, no crystal structure of full-length ncOGT is currently available. In addition, the substrate-binding mode of OGT beyond the immediate surroundings of the active site remains elusive.

Figure 2.

O-GlcNAc cycling enzymes and their inhibitors. (a) Schematic representation of the three isoforms of human OGT (ncOGT, mOGT, and sOGT) and the model of full-length ncOGT built from the crystal structures of OGT_4.5 (PDB: 3PE4)³⁷ and TPR domain (PDB: 1W3B).⁴⁰ TPR: tetratricopeptide repeat domain. N-Cat: N-terminal catalytic domain. C-Cat: C-terminal catalytic domain. Int-D: intervening domain. (b) Inhibitors of OGT. (c) Schematic representation of the two isoforms of human OGA (lOGA and sOGA) and the crystal structure of a bacterial OGA, OgOGA (PDB: 2XSB).⁵⁵ Catalytic: catalytic domain. Stalk: stalk domain. HAT: pseudo-histone acetyl transferase domain. (d) Inhibitors of OGA.

Potent and specific inhibitors would be powerful tools to investigate the biological functions of OGT. Several promising inhibitors have been developed recently employing either substrate analogues or new chemical scaffolds identified from high-throughput screens (HTS; Figure 2b). UDP-5SGlcNAc, which mimics the sugar donor UDP-GlcNAc, is one of the most potent OGT inhibitors in vitro (K_i of 8 μM).⁴² Its cell permeable precursor Ac₄5SGlcNAc can be metabolically converted to UDP-5SGlcNAc through the GlcNAc salvage pathway to inhibit endogenous OGT. Another example is UDP-peptide conjugate Goblin 1, which is a bisubstrate-linked inhibitor with an IC₅₀ of 8 μM in vitro.⁴³ However, this conjugate is not cell permeable due to the negatively charged diphosphate group. Meanwhile, HTS led to the discovery of a quinolinone-6-sulfonamide scaffold.⁴⁴ Structure optimization has identified OSMI-1, which demonstrated improved potency (IC₅₀ of 2.7 μM) and selectivity over substrate analogues to block OGT’s activity in vitro, although its specificity in cells requires further elucidation.⁴⁵ Intriguingly, OSMI-1 has been applied to demonstrate that inhibition of OGT’s activity attenuates the replication of herpes simplex virus (HSV), similarly as genetic siRNA silencing of OGT.⁴⁶ Another interesting scaffold discovered from this HTS is benzoxazoli-none (BZX).⁴⁴ Optimization of this scaffold gave BZX2, which covalently inhibits OGT with a novel double displacement mechanism and cross-links two important residues (Lys842 and Cys917) in the enzyme active site.⁴⁷ While this inhibitor is highly effective in vitro, its specificity is low and thus not suitable for cellular studies. These new inhibitors possess significantly improved properties; however, none of them is potent or specific enough for in vivo applications. Furthermore, new strategies to develop substrate-specific inhibitors that allow targeting of O-GlcNAcylation on an individual or small family of proteins are a desired but unmet need.

OGA and Its Inhibitors.

The enzyme that specifically cleaves O-GlcNAcylation is OGA, belonging to the CAZy glycoside hydrolase family 84.⁷ There are two splicing isoforms of OGA in humans: long OGA (lOGA) and short OGA (sOGA) (Figure 2c). lOGA consists of an N-terminal catalytic domain and a C-terminal pseudo-histone acetyl transferase (HAT) domain, linked by a stalk domain between them. sOGA lacks the pseudo-HAT domain and displays lower catalytic activity.^48,49 Since human OGA (hOGA) has not been amenable for crystallization, the structures of four bacterial OGAs have been solved: Bacteroides thetaiotaomicron hexosaminidase (BtGH84),^36,50,51 Clostridium perfringens NagJ (CpOGA),^52–54 Oceanicola granulosus glycosidase (OgOGA),⁵⁵ and Thermobaculum terrenum glycoside hydrolase (TtOGA).⁵⁶ These bacterial homologues share significant sequence similarity with hOGA in the catalytic domain, but not in the stalk domain or the pseudo-HAT domain, which have been suggested to contribute to hOGA substrate recognition. Nevertheless, these crystal structures of bacterial homologues and the biochemical characterization of hOGA have provided significant insights into the catalytic mechanism of O-GlcNAc hydrolysis and guided the rational design of OGA inhibitors.^57,58

Several highly potent and selective OGA inhibitors have been reported, most of which are transition state mimics (Figure 2d). PUGNAc is one of the first potent OGA inhibitors with a K_i of 46 nM, but it also inhibits other glycoside hydrolases.^59,60 GlcNAcstatin G improved upon this with a K_i of 4.1 nM for hOGA.⁶¹ Although it possesses a remarkable 900000-fold selectivity over β-hexosaminidase (the most structurally related enzyme to hOGA) in vitro, it is difficult to synthesize and has low water solubility, thus limiting its application in vivo. NAG-thiazoline represents another OGA inhibitor scaffold.⁶² A slight modification of this compound produced the potent inhibitor thiamet-G (K_i = 21 nM against hOGA) with excellent selectivity (37000-fold) over β-hexosaminidase.³⁶ Moreover, the synthesis of this inhibitor is straightforward, and the compound can cross the blood—brain barrier for potential applications in neurodegenerative diseases. For instance, thiamet-G has been employed to raise O-GlcNAcylation level in mouse models of Alzheimer’s disease, which reduced protein aggregation and alleviated Alzheimer’s symptoms.^36,63 More recently, stereoisomeric iminocyclitols have also been reported to inhibit hOGA.⁶⁴ Different isomers demonstrated distinct inhibition effects, with several exhibiting low nanomolar activity against hOGA in vitro. One of the isomers (Figure 2d) showed good bioavailability in mice and could also pass the blood—brain barrier.⁶⁴ For more detailed reviews on OGA inhibitors, please refer to refs 65 and 66.

DETECTION OF O-GLCNACYLATION

Lectins and Small Molecule Probes.

One of the main challenges in the field is detecting O-GlcNAcylation with high sensitivity and specificity. This modification has been traditionally detected using lectins (e.g., WGA)⁶⁷ and antibodies (e.g., CTD110.6).⁶⁸ Despite their widespread use, they potentially recognize other types of sugars,⁶⁹ and very few site-specific O-GlcNAc antibodies are currently available. Recently, a “synthetic lectin” was reported with high selectivity and 25-fold improved affinity for β-O-GlcNAc versus WGA.⁷⁰ In another study, the recombinant fungal lectin rPVL was also shown to be more specific than WGA and about 10-fold more potent.⁷¹ Additionally, rPVL contains multiple O-GlcNAc binding sites, which can bind to multivalent O-GlcNAc with an affinity 1900-fold greater than to monovalent O-GlcNAc. A different approach has exploited the ability of bacterial CpOGA to recognize the O-GlcNAc proteome.⁷² Even though CpOGA can hydrolyze O-GlcNAc from human cell lysates, its physiological role and recognition specificity require further investigation as no dynamic O-GlcNAcylation was detected in bacteria. Besides detecting global O-GlcNAcylation, development of site-specific antibodies and lectins will be invaluable for future O-GlcNAc detection.

Metabolic labeling is another prevailing strategy to detect O-GlcNAcylation. Similar to Ac₄5SGlcNAc, other GlcNAc analogues with bioorthogonal handles (e.g., Ac₄GlcNAz,⁷³ Ac₃6AzGlcNAc,⁷⁴ and Ac₄GlcNAlk⁷⁵ in Figure 3a) can be metabolically converted to UDP sugars in cells for O-GlcNAc enrichment and detection. These probes have been reviewed extensively.^5,76 By exploiting the endogenous epimerases, GalNAc sugars can also be converted to UDP-GlcNAc analogues in cells to monitor O-GlcNAcylation. A recent application of Ac₄GalNAz (Figure 3a) has discovered that O-GlcNAcylation could be a cotranslational process and protects nascent peptides from ubiquitination and degradation (Figure 1).^77,78 A new GlcNAc analogue (Ac₃4dGlcNAz, Figure 3a) lacking the 4′-OH group was reported to reduce nonspecific incorporation into extracellular glycans and increase resistance to OGA hydrolysis, thus leading to elevated specificity and sensitivity for O-GlcNAc detection.⁷⁹

Figure 3.

Chemical methods to detect O-GlcNAcylation. (a) GlcNAc and GalNAc analogues containing different functional groups that have been applied in metabolic labeling of O-GlcNAcylation. (b) The principle of Glyco-seek for sensitive quantification of O-GlcNAcylation. (c) O-GlcNAc imaging is enabled by O-GlcNAc metabolic labeling and EGFP fusion. EGFP: enhanced green fluorescent protein. TAMRA: carboxytetramethylrhodamine.

Copper-catalyzed or copper-free (e.g., cyclooctyne) click chemistry has been widely employed to detect O-GlcNAcylation in combination with the above metabolic probes. Owing to the toxicity of copper to cells and nonspecific reactivity of strained alkynes, other metal free, bioorthogonal “click” reactions have been developed, such as the inverse electron demand Diels–Alder (DAinv) reaction between 1,2,4,5-tetrazines and an alkene dienophile. One such study demonstrated its potential by employing methylcyclopropene GlcNAc analogue Ac₄GlcNCyoc (Figure 3a) for intracellular protein labeling.⁸⁰ Notably, few extracellular glycans were detected. DAinv could potentially allow dual labeling of different targets in cells along with azide–alkyne click chemistry. Continued improvement in the specificity and incorporation efficiency of metabolic probes will advance their applications in O-GlcNAc biology.

Sensitivity is another huge challenge in O-GlcNAc detection since many O-GlcNAcylated proteins are present in low abundance and low O-GlcNAc stoichiometry. A method termed Glyco-seek was recently shown to increase the sensitivity of O-GlcNAc detection by several orders of magnitude (Figure 3b).⁸¹ In this study, O-GlcNAc was first labeled by a mutant galactosyltransferase (GalT-Y289L) with GalNAz using an established chemoenzymatic method.⁸² Following click chemistry, O-GlcNAc residues were conjugated to biotin epitopes. DNA-tagged antibodies specific for biotin and the target protein bring the attached DNA strands into proximity if the protein is O-GlcNAcylated, and this can be quantified by DNA ligation and quantitative PCR. This new technique can detect unprecedented attomole levels of glycosylated proteins, although it may not be that effective for discovering new OGT substrates. Further progress toward more sensitive and specific methods for O-GlcNAc detection will aid research into the functions of this dynamic modification.

Mass Spectrometry.

Mass spectrometry (MS) has become the primary technique for discovering new OGT substrates. As discussed above, sensitivity is often an issue in O-GlcNAc detection. The problem is even more pronounced in MS methods due to the profound ion suppression effect of O-GlcNAcylated peptides. Consequently, enrichment of O-GlcNAcylated proteins/peptides is often essential for adequate detection. Normally this involves either enrichment with WGA or metabolic labeling followed by conjugation to biotin with click chemistry as discussed above. The difficulty of eluting biotinylated proteins/peptides from streptavidin beads promoted the development of cleavable biotin linkers. A photocleavable linker has been successfully applied to enrich O-GlcNAc; however, it is difficult to cleave quantitatively.^83,84 Diazobenzene linkers that can be cleaved under mild conditions with dithionite have also been employed in O-GlcNAc studies⁷⁴ but suffer from a similar problem.^84,85 A silane group was later shown to be efficiently cleaved with dilute acid for the enrichment of O-GlcNAcylated proteins and exhibited greatly improved selectivity.^85,86 Recently, a linker containing the 1-(4,4-dimethyl-2,6-dioxocyclohex-1-ylidene)ethyl (Dde) group was shown to be quantitatively cleaved by hydrazine and, importantly, gave a positively charged amine for better MS ionization.⁸⁷ Using immunoblot, this probe displayed better efficiency of global labeling and cleavage than the commonly used photocleavable alkyne-biotin in a test of cell lysates. Additionally, it could easily detect α-crystallin added to the cell lysates using MS. However, it has yet to be employed in MS detection of O-GlcNAcylation on endogenous proteins.

A powerful strategy to identify low abundance O-GlcNAcylation is quantitative proteomics on isotopically recoded glycopeptides. In a recent example, a biotin-dibromide cleavable tag was applied to detect O-GlcNAcylated proteins of significantly lower abundance than was previously achievable.⁸⁶ Another study fed ¹³C labeled glucose to living cells to monitor O-GlcNAc turnover and compare to other types of PTMs.⁸⁸ O-GlcNAcylation turnover rate was found to spread over a wide range (0.02 h⁻¹ to 1.6 h⁻¹) but was generally slower than phosphorylation and acetylation. Other isotopic labeling strategies such as stable isotope labeling with amino acids in cell culture (SILAC) and β-elimination Michael addition (BEMAD) with deuterated DTT (d6) have also been used to quantitatively investigate O-GlcNAc and have been covered in previous reviews.^5,89

Another challenge of MS detection of O-GlcNAcylation is its lability under collision-induced fragmentation methods (e.g., collision-induced dissociation (CID), higher-energy collisional dissociation (HCD)). This problem was partially addressed with the development of electron transfer dissociation (ETD)⁹⁰ and electron capture dissociation (ECD),⁹¹ which enable the fragmentation of glycopeptides exclusively along the backbone while leaving the glycosidic bond intact. These fragmentation methods are capable of site mapping for PTMs such as O-GlcNAcylation, though both suffer from lower sequence coverage. Several studies have overcome these problems by utilizing more than one type of fragmentation.⁵ An improved technique, called electron transfer and higher-energy collision dissociation (EThcD), offers ETD fragmentation prior to HCD for more complete fragmentation and more confident site mapping of O-GlcNAcylation.⁹² One study applied this method to human leukocyte antigen (HLA) class 1 peptides, which are recognized on the cell surface by the immune system to distinguish healthy versus infected or cancerous cells.⁹³ They discovered O-GlcNAcylation on HLA peptides and found it extended with up to four other monosaccharides. The authors proposed this as the first example of intracellular O-GlcNAc being used as the base of an extracellular glycan. This study demonstrates the value of improved fragmentation techniques toward studying O-GlcNAc biology. Further progress will allow more efficient O-GlcNAc detection with higher sequence coverage.

Determining the stoichiometry of O-GlcNAcylation is another area of interest in the field. The O-GlcNAc residue is neutral and too small (203 Da) to produce a noticeable shift in SDS-PAGE gel, making it difficult to monitor the O-GlcNAcylation level of a particular protein. A mass-tagging approach has been devised in which a large polyethylene glycol (PEG) tag can be ligated to an O-GlcNAc residue, which allows bands of O-GlcNAcylated proteins to be discerned on SDS-PAGE gel based on the number of sugars.⁹⁴ More recently, intact protein mass spectrometry has also been employed for this purpose. A recent study detected around 20 residues of O-GlcNAc modified on the CTD of Pol II.¹⁹ It also demonstrated that the addition and removal of O-GlcNAc from the CTD occurs via distributive mechanisms by OGT and OGA, respectively. Intact protein MS was particularly essential in this case due to the repetitive nature of the CTD and the highly heterogeneous O-GlcNAcylation states, which make O-GlcNAc site assignment extremely difficult. Intact protein MS also gave better resolution for quantifying O-GlcNAc stoichiometry than mass tags. This was the first application of mass spectrometry to study the mechanism of dynamic O-GlcNAcylation and should prove useful for other O-GlcNAcylated proteins.

Fluorescence Imaging.

Visualization of specific O-GlcNAcylated proteins in living cells will enable studying the spatial and temporal distribution of O-GlcNAcylated proteins in physiological and disease states. A Förster resonance energy transfer (FRET) method was introduced a decade ago for this purpose.⁹⁵ However, it requires the substrate to be genetically fused to two fluorescent proteins and a lectin, potentially affecting its endogenous O-GlcNAcylation and localization. A newer study has fused only enhanced green fluorescent protein (EGFP) to tau (Figure 3c),⁹⁶ a known OGT substrate and a critical protein in Alzheimer’s disease. First, the authors metabolically labeled O-GlcNAcylated proteins with Ac₄GalNAz. Following click chemistry with alkyne-TAMRA and imaging with fluorescence lifetime imaging microscopy (FLIM-FRET), they were able to monitor O-GlcNAcylated tau and β-catenin in living cells. Since FLIM-FRET detects donor fluorescence lifetime rather than acceptor fluorescence intensity, it reduced the background and increased the imaging sensitivity and resolution. A similar strategy utilizing methylcyclopropene-tetrazine click chemistry has imaged several O-GlcNAcylated proteins including OGT and p53 in cells.⁹⁷ This may prove to be useful considering the previously mentioned drawbacks of alkyne–azide click chemistry. Improved imaging techniques will facilitate characterizing the localization of O-GlcNAcylation in diseases and identifying any cellular heterogeneity among them.

FUNCTIONAL CHARACTERIZATION

Synthetic Proteins.

A lack of access to homogeneous samples of O-GlcNAcylated proteins has significantly hampered our understanding on the properties and functions of this modification in a site-specific manner. This challenge is being overcome with two general synthetic strategies. One employs native chemical ligation (NCL) and expressed protein ligation (EPL) to accomplish total protein synthesis. The first report on semisynthesis of a GlcNAcylated protein was accomplished with the preparation of S-GlcNAcylated CK2α by EPL. This synthetic glycoprotein was successfully applied to establish that S-GlcNAcylation on CK2α modulates its phosphorylation and substrate selectivity.⁹⁸ More recently, NCL and EPL were applied in combination to generate O-GlcNAcylated-Thr72 of α-synuclein (Figure 4a), a protein implicated in neurodegenerative diseases such as Parkinson’s disease.⁹⁹ They were able to demonstrate that O-GlcNAcylation on Thr72 reduces aggregation and toxicity of α-synuclein. Another group used a similar approach to produce O-GlcNAcylated-Ser400 of the tau protein for the first time, which will be valuable considering the lack of knowledge on the role of site-specific tau O-GlcNAcylation.¹⁰⁰

Figure 4.

Synthetic protein methods facilitating the functional investigation on O-GlcNAcylated proteins. (a) Native chemical ligation and expressed protein ligation methods for the synthesis of O-GlcNAcylated α-synuclein protein. (b) In a different strategy, GlcNAc-thiol or halogenated O-GlcNAc can be attached to dehydroalanine (Dha) through Michael addition or radical reaction, respectively. The S-GlcNAc on Cys112 of histone H2B promotes binding with the FACT protein complex, as well as histone ubiquitination by BRE1A enzyme. U: ubiquitination. FACT: facilitates chromatin transcript.

An alternative strategy exploits the unnatural amino acid dehydroalanine (Dha), which can be incorporated into a protein with several methods.^101,102 The Michael acceptor of Dha can react with a thiol nucleophile such as GlcNAc-thiol (Figure 4b) to generate specifically S-GlcNAcylated protein without the need for ligating fragments. This approach has been implemented in a couple of studies.^103,104 In particular, production of histone H2B S-GlcNAcylation on Cys112 using this method has found that S-GlcNAcylation increased binding with several proteins including the facilitates chromatin transcript (FACT) complex, which led to a new proposed mechanism for FACT mediated H2B ubiquitination by BRE1A (Figure 4b).¹⁰⁴ S-GlcNAcylated proteins are resistant to OGA hydrolysis, which is advantageous for O-GlcNAc detection but may impede experiments aimed at O-GlcNAc dynamics. A novel approach applying free radical chemistry to attach PTMs to Dha through a carbon–carbon bond was recently published by two independent groups (Figure 4b).^105,106 One of the studies found that a protein O-GlcNAcylated with this method was still a substrate for OGA, even though it contains extra carbon atoms between the α-carbon and the sugar.¹⁰⁵ Although protein (re)folding may limit the utility of these synthetic approaches for larger proteins, they present vast opportunities for understanding the properties and functions of O-GlcNAcylation in a site-specific manner.

O-GlcNAc-Protein Interactions.

Modulation of protein–protein interactions is believed to be an important function of O-GlcNAcylation. The interacting complex between O-GlcNAc monosaccharide modified protein and the binding protein, however, is generally too weak to be pulled down. To address this issue, an N-acyl diazirine functional group was introduced to GlcNAc (called Ac₄GlcNDAz, Figure 3a) to cross-link interacting proteins upon UV irradiation.¹⁰⁷ However, OGT prefers UDP-GlcNAc over UDP-GlcNDAz, which is problematic due to already low cross-linking efficiencies of diazirine. This issue was alleviated by the discovery of mutant OGT(C917A) that favors UDP-GlcNDAz over UDP-GlcNAc, leading to significantly improved O-GlcNDAz incorporation.¹⁰⁸ The ability to efficiently cross-link O-GlcNAc with interacting proteins will be highly valuable for mapping the interactome of O-GlcNAcylated proteins for functional characterization. Isotopic labeling methods such as SILAC can also help overcome sensitivity issues.

CONCLUSION AND FUTURE PERSPECTIVES

A number of new strategies for studying O-GlcNAcylation have been published in recent years. Improved OGT inhibitors have been introduced to complement OGA inhibitors already in use. However, off-target effects still plague these inhibitors; thus more specific ones are still needed. Metabolic labeling with novel GlcNAc analogues and advances in mass spectrometry have aided its detection on both known and new proteins. Methods to obtain specifically O-GlcNAcylated proteins are also proving effective for investigating the function of O-GlcNAcylation in a much more targeted way. Further application of these synthetically modified proteins will undoubtedly lead to new discoveries on the essential roles of this unique modification. One function of O-GlcNAcylation of particular interest is its ability to modulate protein–protein interactions, for which photoactivatable GlcNAc analogues will provide considerable insights. Strategies that afford more specific and higher yields of photo-cross-linking will enable the study of interactions that are weaker and involve less abundant proteins.

Many goals remain to be reached, including structures of OGT and OGA bound to protein substrates, improved MS detection, and site-specific O-GlcNAc antibodies. Ultimately, these advances will allow us to answer important questions such as the role of O-GlcNAcylation in transcriptional and epigenetic regulation. In addition, its function in neurobiology is of particular importance owing to its potential role in Alzheimer’s disease and other neurodegenerative diseases but has not yet reached a full understanding. Answers to these questions will lead to better knowledge of the pathogenesis of diseases and the development of new therapeutics.

KEYWORDS:

post-translational modification (PTM)

the covalent modification of a protein during or after translation, normally accomplished by enzymes

glycosylation

a post-translational modification where a carbohydrate is attached to a protein

O-GlcNAcylation

the modification of serine or threonine residues of intracellular proteins with O-linked β-N-acetylglucosamine

lectin

a highly specific carbohydrate-binding protein or macromolecule

click chemistry

ligation reactions that are modular, high yielding, wide in scope and highly exothermic so that they give only one reaction product; includes azide–alkyne cycloadditions as discussed here

EThcD

a mass spectrometry fragmentation technique that produces fragments from electron transfer dissociation followed by higher-energy collisional dissociation in a single MS/MS spectrum to provide more sequence information and to improve sequence coverage

Förster resonance energy transfer

distance-dependent nonradiative transfer of energy between two molecules

native chemical ligation

a conjugation reaction between a peptide-alpha-thioester and a peptide bearing an N-terminal cysteine used to construct a longer polypeptide chain while retaining the native peptide backbone

expressed protein ligation

a protein semisynthesis method that permits the in vitro ligation of a recombinant polypeptide C-terminal thioester with a synthetic peptide bearing an N-terminal cysteine; the C-terminal thioester is obtained using a mutant intein splicing protein

diazirine

a class of organic molecules consisting of a carbon bound to two nitrogen atoms, which are double-bonded to each other, forming a three-membered ring; upon UV irradiation, diazirines generate reactive carbenes that can cross-link with nearby molecules