Cubic section model

Figure 7. Diamond lattice and Jones cubic section model.

In 1982 Jones and Jacovac mechanically constructed a cubic section model using Framework Molecular Models." The cubic section model was conceived because... 

The goal of the present study was to develop a computer-based cubic section model of the substrate binding domain of HLADH. It was considered that the Jones cubic section model could be refined by use of computer assisted substrate overlay in combination with kinetic data on a wide variety of substrates. As in the Jones approach we used the alcohol products as the surrogate substrate structures. Thus, we determined the low energy conformation of alcohols produced from ketones that have been reported to be reduced by HLADH and for which comparative kinetic data vs cyclohexanol could be calculated. As well, we determined the preferred conformations of all alcohols that would have been produced from ketones subjected to but failing to undergo HLADH reduction. These calculations utilised molecular mechanics (MACROMODEL) and yielded accurate co-ordinates for ali atoms in each alcohol. Where enantiomeric or stereoisomeric alcohols were produced or capable of production, the co-ordinates of each were calculated. 

As in the Jones protocol the cubic section model of the substrate binding domain of HLADH were constructed using structures of alcohol products rather than ketone substrates. The alcohol products were originally chosen by the Jones group because the transition state the geometry for the reduction was considered to resemble that of the alcohol rather than that of the ketone. The relative rate of reduction of substrate vs cyclohexanone for each ketone was required to be known. Furthermore, configurations of alcohol products, enantiomeric excess values, yields and % conversion of substrate required for calculation of the priority number for each enantiomer of product should be measured under comparable conditions (i.e. pH, temperature, concentration of enzyme, coenzyme and substrate, etc.). According to Alderweireldt et al. (1988) HLADH models are valid only for reaction conditions used in the reactions from which the models are constructed. Furthermore, the model is oniy reliable if the reactions have been conducted under kinetic control. 

Structurally rigid substrate surrogates were added to the cubic section model before more flexible molecules as the following order signifies pentacyclic, tetracyclic, tricyclic, bicyclic ketones, trans/cis-decalones, methyl cyclohexanones and alkyl cyclohexanones. The hydroxyls of cyclohexanols were oriented axially with respect to cyclohexyl rings consistent with the Jones protocol and with the obsen/ation that... 

A goal of this work was to refine the cubic section model developed by Jones using computer modelling. Thus, equivalent orientations of compounds, cube size, I.SA. and origin location were used to construct the cubic section models of HLADH. Because the jacks and plastic tubing of the mechanical model occupy space, Jones changed... 

Some structures were examined which were not included in the cubic section model used for testing. For example, linear and branched 3-alkyl cyclohexanols can change some forbidden cubes in col 3 to allowed. Also 4-alkyl cyclohexanols were found to extend the allowed area in layers 2, 3, and 4. These substrates were not used in the construction because they were studied under different reaction conditions or no values of enantiomeric excess were available. Addition of substrates containing hetereocyclic bicyclic rings (70-73) would change cubes OaO, laN, laP, laQ, and 3E to allowed (Figure 16). 

Figure 17. Cubic section model (cel3) of the substrate binding domain of HLADH.

All twelve cubic section models listed Table 6 were tested to determine which constituted the best model for prediction of substrate reactions with HLADH. The effects of the origin placement, cube size and structure of substrates were tested. In some analyses compounds that had been previously used in construction of the models were used. If this were the case the entries relating to these products were individually removed from the model before a prediction was executed. For new products the minimization, orientation and alignment processes were conducted before execution of a prediction. 

Figure 19 shows an example of a prediction table and a display of positions of the atoms of a test molecule in the cubic section model (col 3). The test molecule is known not to be formed by HLADH reduction (the priority number is zero). Although no forbidden cubes are occupied one occupied cube has a very low priority value, 0.24. This suggests that the test molecule would not be expected to be produced by HLADH. The cubic section map for this molecule further showed that some carbons are near to forbidden cubes. It is usually better to check all visualisation methods when using ENZYME for prediction of reactivity. In practice it is found that the full prediction table gives the most accurate prediction. 

The cubic section model, col 3 (Table 6) was used for prediction of the relative rates of formation for all 73 products in their different conformations listed in Table 3. Of these. 

Table 9. Prediction of acceptance by HLADH using cubic section models with 1. SA or 0.65A cubes with origin at the corner of center cube, with and without added...

The X-ray crystallographic structure of the ternary complex, HLADH-NADH-dimethyl sulfoxide, DMSO, has been determined to 2.9 The crystallographic co-ordinates available from the Brookhaven protein data bank were entered into MACROMODEL and the amino acid residues surrounding the active site identified. Using MACROMODEL and ENZYME the substrate binding domain of HLADH was mapped as a cubic section model. The model obtained from this approach were compared with model col 3 obtained by substrate surrogate overlay. 

Model col 3 and the model constructed with all hydrogens in substrate surrogates (co13h) were compared with the cubic section model derived from display of the cubes occupied by residues of HLADH lining the substrate binding domain. Using... 

The cubic section models, co13 and co065, proved to be accurate in predicting acceptance of substrates and the stereochemical course of HLADH mediated reductions. The average priority value for all cubes, number of forbidden sites and number of low priority cubes occupied by atoms in a possible alcohol are used to estimate the acceptability of the corresponding ketone by HLADH. The prediction method developed can be applied to other enzymes (e.g. other dehydrogenases, esterases, or lipases) if kinetic and specificity data is available. 

Comparison of the X-ray based cubic section model with the substrate surrogate derived cubic section model gave a view of the amino acid residues around the substrate binding domain (Figure 21). Forbidden and boundary regions defined by the substrate surrogate overlay method are near the side chains amino acid residues. Substrate interactions with surface of the amino acid residues could be estimated by molecular mechanics (MACROMODEL). 

---