Structure of tagatose-1,6-bisphosphate aldolase

ion of a proton from DHAP C1 would be feasible. Glu of FBPA corresponds to Glu in TBPA, which is also positioned on a glycine-rich loop. Due to the lack of reliable electron density, this residue has not been included in the TBPA model and it is presumed to be flexible and disordered. The sequence conservation of this loop in TBPA and FBPA is high (Fig. 2), which implies that Glu could adopt a similar position and contribute the same function as Glu in FBPA. We would therefore predict that mutagenesis of Glu in TBPA would have the same effect as seen for Glu in FBPA. An alternative hypothesis, which would still involve Glu/ Glu, to explain proton abstraction from DHAP C1 involves the use of activated water. In both the FBPA and TBPA PGH complex structures there is a well-defined water molecule within hydrogen bonding distance of the PGH N2 and correctly positioned to interact with the 1-proS -H of these class II aldolase DHAP complexes (Fig. 4). This water therefore represents a potential base for the proton abstraction. In FBPA and TBPA the water is hydrogen bonded to other solvents in the active site and in the former a hydrogen bond network extends up to Glu (8). In the absence of PGH the catalytic Zn of FBPA is five-coordinate, using three histidines and two water molecules (11). The water ligands are replaced by PGH and, by implication, DHAP during catalysis (8). The replacement of such cation-binding solvent by hydroxamates is common to other zinc enzyme inhibitor complexes (31) and a possibility is that the Zn contributes to activating a water. We note that details of a similar proton abstraction step in class I aldolases has proven difficult to explain. However, the recent work of Heine et al. (32) has indicated that for D-2-deoxyribose-5-phosphate aldolase a proton relay system activates an active site water, which then mediates proton transfer. Our analysis does not unambiguously explain the details of proton abstraction in the class II aldolase system but does suggest that the potential role of activated water should be investigated. The Structural Basis of Chiral Discrimination—The major differences between the TBPA and FBPA active sites are localized to that region where G3P has to bind, and, although the main chain of the enzymes overlay well, we note two sequence changes that alter side chain contributions to the active sites (Fig. 5). Gln and Asp of FBPA correspond to Ala and Ala, respectively, in TBPA. This results in more space on one side of the TBPA active site and influences the side-chain orientation of a conserved asparagine (Asn in FBPA; Asn in TBPA). The asparagine N 2 groups are structurally conserved when the TBPA and FBPA models are overlaid (not shown) as is the hydrogen bond formed with another conserved residue: Asp in FBPA and Asp in TBPA. In FBPA, Asn O 1 is held in place by accepting a hydrogen bond from Gln N 2 whereas Asn N 2 donates a hydrogen bond to Asp. In TBPA, however, the loss of functional groups and side-chain atoms results in more space in the TBPA active site. The side chain of TBPA Asn adopts a different conformer, and the O 1 group alters position with respect to that seen in FBPA. The result is that, although N 2 of Asn FBPA and Asn TBPA occupy the same relative position, the O 1 groups occupy different locations in the active site. The complexes with PGH mimic the ene-diolate enzyme structures (stages I through II in Fig. 1) formed from DHAP. We previously modeled the other reactant, G3P into the active site of FBPA on the basis of simple graphical considerations (stage III, 8) but have now used computational chemistry methods to position FBP and TBP in the active sites of their cognate enzymes, in effect to model stage IV (Fig. 1) for each aldolase. Although crude, these models serve to identify possible interactions of relevance to enzyme mechanism and specificity. The lowest energy FBP models in the active site of FBPA, of which one is shown in Fig. 4, placed the C6 phosphate to interact with Arg of the partner subunit and Ser. These residues are conserved in the class II FBP-aldolases, and the model is consistent with the kinetic analysis of mutant enzymes in which these two residues have been altered (9, 12). The C4 hydroxyl of FBP is positioned about 4 Å from Asn and 3 Å from the carboxylate side chain of Asp. Mutagenesis of the asparagine in FBPA to alanine produced an enzyme with 1.5% of the activity of wild-type protein (9), and the FBPA FBP model suggests a role in binding the C4 OH. Asp has previously been shown to be responsible for the protonation of the incoming C4 carbonyl group (Fig. 1, stage III, 10). The docking of TBP into the active site of TBPA did not produce a set of closely clustered models as observed for the FBPA FBP combination but indicated that a range of conformations of the G3P component are accessible. This suggests that the TBPA active site allows the substrate more conformational freedom than is the case for FBPA. We present one of the TBP TBPA models in Fig. 5, which, like the others, suggests that the G3P phosphate interacts with a basic patch, including His, Lys, and from the partner subunit Arg, which corresponds to Arg in FBPA discussed above. In addition the phosphate is placed near Thr, which corresponds to Ser in FBPA. His and Lys are altered in FBPA to Val and Gln, respectively (Fig. 5). The C4 OH group of TBP is directed into the space between the side chains of Ala and Asp and in position to interact with Asn. Asp corresponds to Asp in FBPA and, by analogy with FBPA, we judge it likely that this residue is responsible for the protonation of the C4 carbonyl of during the aldol condensation to produce TBP. An overlap of both enzyme active sites and substrate models (not shown) suggests that there would be a significant steric clash between the G3P of TBP with Asp of FBPA in addition Structure of Class II Tagatose-1,6-bisphosphate Aldolase 22023 by gest on Jne 4, 2017 hp://w w w .jb.org/ D ow nladed from to the electrostatic repulsion of anionic groups. Such interactions are likely a contributory discriminating factor in the specificity of FBPA. Superposition of FBP models in the active site of TBPA does not predict any major steric clash. In contrast to FBPA, TBPA displays poor stereochemical control with nearly 10% of the L-erythro configuration being observed and about 90% of the D-threo. In addition, TBPA can utilize glycoaldehyde, L-glyceraldehyde, acetaldehyde, and isobutyraldehyde as the DHAP partner (16). This lack of specificity can be explained by the availability of space in that part of the active site where the aldehyde binds. It appears that FBPA has an active site that is complementary to and highly specific for its cognate substrates. TBPA on the other hand is less specific due to the conformational freedom afforded its range of substrates. We conclude that the overall discrimination between the two systems is dominated by the strict specificity of FBPA, and, therefore, TBPA might be the better choice of enzyme to serve as the framework onto which new catalytic activities could be constructed. Acknowledgments—We thank European Synchrotron Radiation Facility for access and our colleagues for encouragement and excellent support in particular Drs. Graeme Thomson and Shaza Zgiby for help in the preparation of enzyme.