A receptor dependent-4D QSAR approach to predict the activity of mutated enzymes

Abstract

Screening and selection tools to obtain focused libraries play a key role in successfully engineering enzymes of desired qualities. The quality of screening depends on efficient assays; however, a focused library generated with a priori information plays a major role in effectively identifying the right enzyme. As a proof of concept, for the first time, receptor dependent - 4D Quantitative Structure Activity Relationship (RD-4D-QSAR) has been implemented to predict kinetic properties of an enzyme. The novelty of this study is that the mutated enzymes also form a part of the training data set. The mutations were modeled in a serine protease and molecular dynamics simulations were conducted to derive enzyme-substrate (E-S) conformations. The E-S conformations were enclosed in a high resolution grid consisting of 156,250 grid points that stores interaction energies to generate QSAR models to predict the enzyme activity. The QSAR predictions showed similar results as reported in the kinetic studies with >80% specificity and >50% sensitivity revealing that the top ranked models unambiguously differentiated enzymes with high and low activity. The interaction energy descriptors of the best QSAR model were used to identify residues responsible for enzymatic activity and substrate specificity.

Figure 1

Flowchart of the overall RD-4D-QSAR process. (A) The chart explains the steps involved in the generation of interaction energy descriptors of the RD-4D-QSAR paradigm. (B) Schematic representation of the protocol that was used to generate different PLS models to derive models with maximum accuracy. Codes represent enzyme variants with different experimental K _cat values against two different substrates.

Figure 2

Analysis of the predicted activity values generated using PLS models derived from 1875 IEDs chosen based on the r value > 0.5. Graph shows the experimental vs. predicted activity values; training sets containing 7 codes (blue), test sets containing 3 codes (green); external validation set containing 2 codes (orange) and the respective experimental values (grey). The models sorted based on the least RMSD values of the training and the test set were used to predict the activity of the validation set (Table S5). The predictions on the validation set showed clear difference between the enzyme variants with high and low activity; however, the predicted K _cat values were only a little closer to the experimental values.

Figure 3

Analysis of the predicted activity values generated using PLS models derived from 6188 IEDs chosen based on the r value > 0.45. Graph shows the experimental vs. predicted activity values; training sets containing 7 codes (blue), test sets containing 3 codes (green); external validation set containing 2 codes (orange) and the respective experimental values (grey). The models sorted based on the least RMSD values of the training and the test set were used to predict the activity of the validation set (Table S6). The predicted K _cat values were very much closer to the experimental values.

Figure 4

Graph of experimental vs. predicted activity of the validation sets of the models derived from 6188 IEDs chosen based on the r value > 0.45.

Figure 5

Analysis of the predicted activity values generated using PLS models derived from 19764 IEDs chosen based on the r value > 0.4. Graph shows the experimental vs. predicted activity values; training sets containing 7 codes (blue), test sets containing 3 codes (green); external validation set containing 2 codes (orange) and the respective experimental values (grey). The models sorted based on the least RMSD values of the training and the test set were used to predict the activity of the validation set (Table S13). These models showed impressive q2 values but the RMSE values of the test set were relative higher than that of the previously obtained models.

Figure 6

The IEDs of the best QSAR model with negative rc that are crucial for enzyme activity were mapped on the active site of the enzyme to locate important E-S interactions. (A) Represents the LJ (green) and C (pink) IEDs on the CEPs of active site of the enzymes and (B) represents the same on the substrates. The 2D structures of the substrate S-2366 (left) and S-2288 (right) represent the binding conformations of pyroGlu-Pro and H-D-Ile-Pro of the substrates in the active site. This conformational change is one of the crucial differences observed in the active site that defines the specificity of the enzymes.

Figure 7

IEDs with positive and negative rc derived from the best QSAR model mapped on the substrates. Clear differences are seen in the arrangement of IEDs, precisely correlating with the enzymes showing high and low activity. The blue and red spheres represent IEDs with positive and negative rc respectively, within 2.5 Å radius to the substrate conformations. The alphabets represent the codes and the respective substrate conformations in the CEPs of a specific enzyme variant. The number represents the K _cat values. The regions encircled in blue and red over pyroGlu moiety are IEDs that differentiated enzymes with low and high activity respectively.