Table 1.
Serine hydrolases considered in this work with details on enzyme classes.
Figure 1.
Superimposition of the 42 ser-hydrolases considered and listed in Table 1.
All catalytic serines appear superposed at the center of the picture. Arrows highlight the oxyanion hole and the residue responsible for acid-base catalysis.
Figure 2.
Schematic illustration of the generation of BioGPS molecular descriptors.
(a) Starting from the GRID mapping of enzyme active site the BioGPS algorithm identifies points used for generating quadruplets and a Common Reference Framework. (b) In order to compare two cavities (active sites), the algorithm searches for similar quadruplets and then overlaps the corresponding 3D structures (all against all approach). At the end a series of probe scores is generated.
Figure 3.
Unsupervised Pattern Cognition Analysis (UPCA) and clustering of Ser hydrolases on the basis of BioGPS descriptors (global score).
The enzymes are labelled according to their PDB code. Lipases are indicated in blue, esterases in green, amidases in red and proteases in cyan.
Table 2.
CaLB mutants used for the validation of the BioGPS-UPCA model and taken from ref 20 and ref 39.
Figure 4.
Projection of CaLB mutants in the BioGPS-UPCA model (global score).
Improved mutants are highlighted as black triangles whereas poor mutants are in pink. WT CaLB (1TCA) is indicated by the blue arrow. The different classes of Ser hydrolases are reported in different colors, as also described in Figure 3.
Figure 5.
Unsupervised Pattern Cognition Analysis (UPCA) of BioGPS descriptors generated by H probe (shape).
The enzymes are labelled according to their PDB code and colored as in figure 3. Improved mutants are highlighted in black triangles and poor mutants are in pink triangles.
Figure 6.
Comparison of 1GVK (protease) and 2W22 (lipase) active site shape.
1GVK and 2W22 are represented as green and magenta cartoon respectively. Active site shapes are represented as wireframes: 1GVK active site shape in green while the active site shape of 2W22 is in magenta.
Figure 7.
Unsupervised Pattern Cognition Analysis (UPCA) of BioGPS descriptors generated by O probe (H-bond donor capacity).
The analyzed enzymes are labelled according to their PDB code and colored as in figure 3. Improved mutants are highlighted in black triangles whereas poor mutants are in pink triangles.
Figure 8.
Comparison of 1GVK (protease) and 2W22 (lipase) active site H-bond donor pseudo-MIFs.
The structures of 1GVK and 2W22 are represented in green and magenta cartoons respectively. 1GVK pseudo-MIFs are represented as green surfaces. 2W22 pseudo-MIFs are represented as magenta surfaces.
Figure 9.
Unsupervised Pattern Cognition Analysis (UPCA) of BioGPS descriptors generated by N1 probe (H-bond acceptor capabilities).
The analyzed enzymes are labelled according to their PDB code and colored as in figure 3. Improved mutants are highlighted as black triangles and poor mutants are pink triangles.
Figure 10.
Comparison of 1GVK (protease) and 2W22 (lipase) active site H-bond acceptor pseudo-MIFs.
1GVK and 2W22 are represented as green and magenta cartoon respectively. 1GVK pseudo-MIFs are represented as green surfaces. 2W22 pseudo-MIFs are represented as magenta surfaces.
Figure 11.
Unsupervised Pattern Cognition Analysis (UPCA) of BioGPS descriptors generated by DRY probe (hydrophobicity).
The analyzed enzymes are labelled according to their PDB code and colored as in figure 3. Improved mutants are highlighted in black triangles and poor mutants are in pink triangles.
Figure 12.
Comparison of 1GVK (protease) and 2W22 (lipase) active site hydrophobic pseudo-MIFs.
1GVK and 2W22 are represented as green and magenta cartoon respectively. 1GVK pseudo-MIFs are represented as green surfaces. 2W22 pseudo-MIFs are represented in magenta.