br O GlcNAc transferase OGT belongs to
O-GlcNAc transferase OGT belongs to the metal-independent GT-B superfamily of glycosyltransferases, which has been well-reviewed previously [14,15]. OGT is an essential gene encoded on the X-chromosome, and it has two main regions: a long N-terminal tetratricopeptide repeat (TPR) region and a C-terminal catalytic region (Figure 1c) [16, 17, 18, 19, 20]. The two Rossmann-folded catalytic lobes that are characteristic of GT-B superfamily members are separated in primary sequence by a unique intervening domain of unknown function. Although there are two shorter splice variants, the TPR region of the primary, full-length OGT contains 13.5 TPRs. These are 34 amino luciferin definition helix-turn-helix motifs that typically mediate protein–protein interactions . Consistent with this, removing TPRs from OGT abolishes protein glycosylation even when the active site is still functional; moreover, some cellular proteins have been shown to form stable interactions with the TPR region [18,22,23,24,25]. A crystal structure of the TPR region of OGT obtained in 2004 (PDB 1W3B; Figures 1c and 2d) shows an elongated, right-handed superhelix . More recently, a structure of the TPR region containing a single point mutation distorts the superhelix, and it is speculated that this distortion alters the O-GlcNAc proteome or the OGT interactome, affecting brain development . The first structure of the catalytic region of human OGT was reported in 2011 (PDB 3PE4; Figure 2a) . This structure, a complex with UDP and a peptide substrate (CKII), contained 4.5 TPRs. The TPRs in this structure partially overlap with the previously crystallized TPR region, and a full-length model of OGT can be generated by superimposing these structures (Figure 2b, middle panel). All OGT structures obtained subsequently have used the 4.5 TPR construct as it crystallizes readily, and multiple structures of OGT complexed with substrates, substrate analogs, and products have been reported since 2011 (Figure 1d) [29, 30, 31]. Many of these are below 2.5 Å and include bound waters. To obtain ternary complexes that reflect how the substrates bind (i.e. pseudo-Michaelis complexes), it was necessary to use a hydrolysis-resistant analog of UDP-GlcNAc in which the ring ether oxygen is replaced with sulfur (UDP-5SGlcNAc) and/or to substitute the reactive serine or threonine in the peptide substrate with alanine or aminoalanine . Ternary glycopeptide product complexes were obtained simply by setting up crystals with UDP-GlcNAc or UDP-5SGlcNAc and a peptide substrate. Sulfur substitution does not alter the conformation of the substrates or products. Overall, the OGT substrate and product complexes have provided useful information about substrate preferences and reaction mechanisms. The following picture has emerged for substrate recognition. OGT first binds UDP-GlcNAc, and then the peptide substrate binds over it, making contact with the nucleotide-sugar for almost its entire length. The peptide backbone is anchored by polar contacts to OGT side chains, and there is also a polar contact between the α-phosphate of UDP-GlcNAc and the amide of the residue that becomes glycosylated, but there are almost no contacts from OGT to substrate side chains. This seemingly strange state of affairs is consistent with what is known about the lack of a strong consensus sequence for glycosylation. Some sequence preferences, such as prolines and β-branched residues in positions flanking the glycosite, have been identified from comprehensive glycosite mapping, but are insufficient to determine whether a site is glycosylated; these preferences can largely be explained by the requirement for the peptide to adopt an extended conformation over several residues in the active site [28,31,]. OGT complexes obtained with peptide substrates that bind in the TPR lumen of the 4.5 construct have provided important insight into how OGT substrates are selected. The first such complexes contained peptides derived from the HCF-1 cleavage region (PDBs 4N3A, 4N3B, 4N3C) (Figure 2b left panel) . OGT cleaves HCF-1 within one of several, centrally located, 26-amino acid repeats that contain a cleavage motif that binds in the active site and a threonine-rich region that binds in the TPR domain . The threonine-rich region is anchored in the TPR lumen by several asparagines that make bidentate hydrogen bonds to the peptide backbone (Figure 2c). Cleavage was abrogated when these asparagines were mutated to alanine . These structures suggested that some glycosylation substrates may use a similar binding mode. Indeed, the structure of an OGT construct fused to a peptide derived from the glycosylation substrate TAB1 shows that the TAB1 peptide binds identically to the HCF-1 threonine-rich region in the TPR lumen (Figure 2c) . Moreover, a global functional analysis has shown that a substantial fraction of OGT glycosylation substrates recognize the asparagine ladder . The evidence thus indicates that many OGT substrates are selected through extensive interactions both in the active site and to the contiguous TPR lumen. An unknown fraction of OGT substrates are thought to be directed to the active site through interactions with adaptor proteins that interaction with the TPR domain [23, 24, 25], and these adaptor proteins may also exploit the asparagine ladder for recognition. In this regard, it should be noted that the asparagine ladder lines the full length of the TPR lumen (Figure 2d) . Other binding modes to OGT likely exist for both substrates and adapter proteins, and further structural and biochemical studies will be required to define these interactions.