• 2018-07
  • 2019-04
  • 2019-05
  • 2019-06
  • 2019-07
  • br O GlcNAcase Human OGA is a multidomain protein with


    O-GlcNAcase Human OGA is a multidomain protein with an N-terminal domain similar to glycoside hydrolase family 84 (GH84) enzymes, a stalk domain, a C-terminal pseudo histone acetyltransferase (HAT) domain, and several low-complexity regions (Figure 1f) [43]. A splice variant that lacks the HAT domain is less abundant and less active than OGA [44]. Although human OGA was first cloned and biochemically characterized in 2001 [43], it did not yield to structural studies until 2017. In the meantime, studies of bacterial homologs possessing a GH84 domain provided valuable insights into the hydrolytic mechanism and allowed the rational design of potent and specific OGA inhibitors, including GlcNAcstatin and thiamet-G [45, 46, 47, 48, 49, 50, 51]. In 2017, three independent research groups reported structures of human OGA, in apo form and in complex with small molecule inhibitors (Figure 1g) [,,]. One group also reported a series of structures of OGA complexed with different glycopeptide substrates [,], and these have begun to reveal general principles for substrate recognition. Key to obtaining the structures was the identification of stable OGA constructs for protein crystallization. Different approaches were used to generate stable OGA constructs (Figure 1f), but all maintained activity comparable to full-length OGA [,,]. The structures showed that the catalytic domain has the classic (β/α)8-barrel conformation characteristic of GH84 hydrolases, and also revealed an unusual dimeric structure not observed for previously studied bacterial OGA homologs (Figure 4a and b). The stalk domain forms a helical bundle comprising four main α-helices, three from the same OGA monomer and the fourth (labeled ‘arm’ in Figure 4a) contributed by the sister monomer via a flexible loop. The dimer is stabilized by burial of a large surface area with formation of an extensive network of hydrogen bonds, salt bridges, and 5,7-Dichlorokynurenic acid interactions. Residues on the dimerization interface are evolutionarily conserved in eukaryotes, and biophysical studies have confirmed that mammalian OGA exists as a stable dimer in solution [,]. The three structures of OGA-thiamet-G complexes are superimposable and reveal active site residues that engage in tight interactions with the inhibitor which is a transition state mimic (Figure 4c left panel) [,,]. An ordered water molecule positioned near the carbon corresponding to the anomeric position of the sugar reveals the trajectory of nucleophilic attack for glycoside hydrolysis. The structures provide direct structural evidence to support the proposed substrate-assisted mechanism for OGA [56] and unveil a set of ancillary residues contributing to the hydrolytic mechanism. A hydrophobic pocket that affords favorable contacts to the N-ethyl substituent of thiamet-G provides a structure-based rationale for selectivity against other β-hexosaminidases. PUGNAc, a potent but more promiscuous inhibitor, cannot leverage this deep pocket to achieve specific targeting of OGA (Figure 4c middle panel) []. VV347, an especially potent OGA inhibitor, makes favorably contacts with OGA surface residues (Figure 4c right panel) []. Taken together, the structures of OGA complexed with these compounds provide valuable knowledge to guide future efforts in inhibitor design, both for O-GlcNAc functional investigation, and potential therapeutic applications such as in Alzheimer’s disease [10]. The OGA-glycopeptide structures provide critical insight into how OGA recognizes diverse O-GlcNAcylated proteins. Structures of a catalytically impaired OGA mutant (D175N) complexed to five distinct glycopeptide substrates have been reported (Figure 1g) and show that all of these glycopeptides are bound in the substrate-binding cleft created by the dimerization of OGA (Figure 4d) [,]. Regardless of the sequence flanking the glycosylated residue or whether that residue is Ser or Thr, the GlcNAc moieties overlap perfectly and engage in significant interactions with active site residues. This sugar binding mode explains the absolute selectivity of OGA for hydrolyzing β-O-GlcNAc substrates over either α-O-GlcNAc or β-O-GalNAc substrates. More strikingly, although all of the peptides adopt a similar ‘V’-shaped binding conformation near the glycosylation site, the peptide backbones can be oriented in opposite directions (Figure 4d middle panel) []. Some of the binding conformations are further stabilized by intramolecular peptide interactions, and some peptide side chain-specific interactions are observed with the OGA cleft, such as the W(-3) subsite of the p53 glycopeptide with F223 and W679 (Figure 4d right panel). Mutation of these contact residues substantially impaired the binding of the p53 glycopeptide for OGA. Notably, different cleft surface residues of OGA participate in interactions with distinct glycopeptides []. These five glycopeptide structures, thus, illuminate a general mechanism for how human OGA recognizes and processes diverse protein substrates. The conserved strong interactions between the GlcNAc moiety and OGA active site residues are critical for recognizing and anchoring glycopeptides in the substrate-binding cleft, but contacts beyond the immediate catalytic pocket can enhance the binding affinity for individual substrates and may affect hydrolysis rates. In support of this, recent proteomic analysis revealed that protein O-GlcNAcylation turnover rates varied 5,7-Dichlorokynurenic acid dramatically in cells, depending on both the protein and the residue modified [57]. These results, in agreement with the structural discoveries, suggest that OGA favorably removes O-GlcNAc from certain substrates during dynamic O-GlcNAc regulation.