acyl substrate binding sites of CmaA1 during the cyclopropanation process was
obtained by analysing the MD simulations trajectories. The apo-state of CmaA1
was observed to have a closed conformation where the cofactor binding site is
inaccessible. Upon cofactor binding, H-bond between Pro202 of loop10 (L10) and
Asn11 of N-terminal a1 helix disrupts making the cofactor binding pocket accessible. Upon cofactor binding, the non-polar side chains of the substrate binding site
position towards the inner side of the pocket forming a hydrophobic environment
for the substrate. In order to exchange the methyl group from the cofactor to the
substrate, both the ligands tend to come close to each other facilitated by the
upliftment of loop10. These observations prompted to think that the protein can
remain in diverse conformations at different stages of its catalytic function and
considering only one conformation for drug design would not be sufficient. So
multiple structures obtained from the MD trajectories were used to generate, validate and use structure and ligand-based pharmacophore models.
7.1.2 Generation of Dynamic Structure-Based Pharmacophore
Models
The molecular dynamics simulations on CmaA1 revealed that the binding sites of
the enzyme exhibit huge conformational diversity, when bound to different ligands
at various stages of its function. To use this conformational diversity of the binding
sites in structure-based drug design, representative structures (snapshots) were
extracted from all the five MD trajectories at a regular interval of 5 ns, thus
obtaining a total of forty conformations of CmaA1 bound to different ligands in the
two binding sites. The crystal structure of CmaA1 reported in PDB was also added
to this pool. Now these 41 protein–ligand complexes were used to obtain
e-pharmacophore models as described in Sect. 5.3.2. The first step used was
evaluating the Glide energy terms. Active site of each CmaA1 structure was defined
as a cubical box of 12 * 12 * 12 Å
3 dimension, and the Glide [89] energy grids
were generated. Glide scores with XP descriptor information were obtained for the
already bound ligands keeping their original conformations unchanged (unlike a
typical docking where protein is held rigid while ligands are kept flexible). This
exercise calculated all the interaction energy components between the receptor–
ligand complexes, which were then submitted to the Phase module of Schrodinger
to develop energy-based e-pharmacophore [88, 95] models. Figure 3 depicts the
steps of the e-pharmacophore model generation and selection of best ones as virtual
screening filters.
7.1.3 Pharmacophore Model Validation
To examine the capabilities of the dynamics-based e-pharmacophore models to
successfully distinguish inhibitors and non-inhibitors of CmaA1, a set of 23
reported CmaA1 inhibitors (MIC:0.0125–12.5 lg/mL) [96] were used as a positive
Pharmacophore Modelling and Screening: Concepts, Recent …
41
obtained by analysing the MD simulations trajectories. The apo-state of CmaA1
was observed to have a closed conformation where the cofactor binding site is
inaccessible. Upon cofactor binding, H-bond between Pro202 of loop10 (L10) and
Asn11 of N-terminal a1 helix disrupts making the cofactor binding pocket accessible. Upon cofactor binding, the non-polar side chains of the substrate binding site
position towards the inner side of the pocket forming a hydrophobic environment
for the substrate. In order to exchange the methyl group from the cofactor to the
substrate, both the ligands tend to come close to each other facilitated by the
upliftment of loop10. These observations prompted to think that the protein can
remain in diverse conformations at different stages of its catalytic function and
considering only one conformation for drug design would not be sufficient. So
multiple structures obtained from the MD trajectories were used to generate, validate and use structure and ligand-based pharmacophore models.
7.1.2 Generation of Dynamic Structure-Based Pharmacophore
Models
The molecular dynamics simulations on CmaA1 revealed that the binding sites of
the enzyme exhibit huge conformational diversity, when bound to different ligands
at various stages of its function. To use this conformational diversity of the binding
sites in structure-based drug design, representative structures (snapshots) were
extracted from all the five MD trajectories at a regular interval of 5 ns, thus
obtaining a total of forty conformations of CmaA1 bound to different ligands in the
two binding sites. The crystal structure of CmaA1 reported in PDB was also added
to this pool. Now these 41 protein–ligand complexes were used to obtain
e-pharmacophore models as described in Sect. 5.3.2. The first step used was
evaluating the Glide energy terms. Active site of each CmaA1 structure was defined
as a cubical box of 12 * 12 * 12 Å
3 dimension, and the Glide [89] energy grids
were generated. Glide scores with XP descriptor information were obtained for the
already bound ligands keeping their original conformations unchanged (unlike a
typical docking where protein is held rigid while ligands are kept flexible). This
exercise calculated all the interaction energy components between the receptor–
ligand complexes, which were then submitted to the Phase module of Schrodinger
to develop energy-based e-pharmacophore [88, 95] models. Figure 3 depicts the
steps of the e-pharmacophore model generation and selection of best ones as virtual
screening filters.
7.1.3 Pharmacophore Model Validation
To examine the capabilities of the dynamics-based e-pharmacophore models to
successfully distinguish inhibitors and non-inhibitors of CmaA1, a set of 23
reported CmaA1 inhibitors (MIC:0.0125–12.5 lg/mL) [96] were used as a positive
Pharmacophore Modelling and Screening: Concepts, Recent …
41