FKBP results

Table 1 lists the free energies of binding of several compounds which were found to bind in the primary site during the first screening of the FKBP NMR experiment. These compounds were docked to FKBP in our computational approach and were found to prefer the primary binding site. Our predicted free energies of binding are within 1-1.7 kcal/mol of the reported experimental values. More importantly, our method for estimating affinities from the docked structures is able to distinguish between low- and high-affinity ligands as shown by its ability to predict the pipecolinic acid derivative as the tightest binding ligand from the compounds in Table 1.

Figure 4a shows the position of the lowest scoring docked structure of the pipecolinic acid derivative. It forms very similar interactions with the protein as described in the experimental NMR paper [1] for this system. Residues that exhibited NOEs with the bound inhibitors are labeled [1].

Docking compounds F3-F9 independently to FKBP, the results show these compounds to bind to the primary site. The benzanilide derivatives scored

Figure 4. Structure of the lowest energy configuration of (a) the pipecolinic acid derivative, compound F2, docked to FKBP; (b) compound F9 docked to FKBP in the presence of the lowest energy structure of the pipecolinic acid derivative; and (c) compound F14 docked to FKBP.

Table 1. Comparison of FKBP simulation results with experimental results: docking into primary site

Table 1. Comparison of FKBP simulation results with experimental results: docking into primary site

* All experimental AG values were calculated from reported Kd or IC50 values for the complexes.

between -5 and -3 kcal/mol (100 p M -10 mM) with HTS when docked in this primary site. The results of docking compounds F3-F9 in the presence of the lowest scoring configuration of F2 identified a binding pocket proximal to the primary pocket in agreement with the Shuker study. Table 2 contains the predicted free energies of binding of the docked compounds along with the experimental values. The predicted free energies of binding are within 0.51.7 kcal/mol ofthe experimental values. Although our computational method cannot clearly distinguish the rank ordering of the benzanilide compounds,

Table 2. Comparison of FKBP simulation results with experimental results: docking in the presence of a pipecolinic acid derivative

Compound R1 R2 R3 R4 AGpredicted AGexpl*

* All experimental AG values were calculated from reported Kd or IC50 values for the complexes.

it does predict compounds F7 and F9 to have the highest binding affinities among the benzanilides that were docked; this is in agreement with experiment. Although not done in this study, once the secondary pocket is identified, the docking could be directed at this particular pocket to reduce both the search space volume and the computational cost.

Figure 4b shows the lowest scoring docked position of compound F9 in the presence of the pipecolinic acid derivative, F2. This molecule adopts a slightly different conformation in the docked FKBP structure than in the NMR structure, due to the positioning of arginine 57. The arginine side chain partially obstructs the benzanilide binding pocket. Because this benzanilide is a low-affinity ligand, slight changes in the conformation of the protein near this secondary binding pocket appear to have little effect on the free energy of binding of these compounds.

Because the composite compounds F10-F14 have 16 to 20 rotatable bonds to sample during flexible docking of the ligand, the conformational space is just too large to be directly searched, even with our efficient docking method. Therefore we employed 'partially fixed docking' [ 18] studies of these larger compounds. Here a significant portion of the moiety that represents compound F2 within these ligands was fixed at the position optimized in the

Table 3. Comparison of FKBP simulation results with experimental results: docking of linked inhibitors

Compound

n

AG predicted

AG exp1

(kcal/mol)

(kcal/mol)*

F10

3

-11.4

-10.5

F11

4

-11.2

-10.2

F12

5

-10.6

-10.0

F13

6

-10.7

-9.0

F14

-

-10.4

-10.0

* All experimental AG values were calculated from reported Kd or IC50 values for the complexes.

* All experimental AG values were calculated from reported Kd or IC50 values for the complexes.

initial docking simulation of F2, while the rest of the molecule was free to rotate. Table 3 contains the results for these inhibitors. Comparing the predicted free energies to the experimental free energies of binding, we find that the predicted values are approximately within 0.5-1.7 kcal/mol of the experimental values. Compounds F10 and F11 are predicted to be the tightest binding inhibitors. This prediction is in agreement with the experimental results presented in Table 3.

Figure 4c shows the lowest energy configuration of compound F14 docked to FKBP. Comparing the position of the benzanilide group of this compound to the position in Figure 4b, we find that they occupy similar binding sites although in slightly different orientations.

These FKBP results demonstrate an important caveat for both the experimental and computational approach. Linking small molecules together to form one inhibitor can create tighter binding inhibitors, but not necessarily selective inhibitors or drug-like molecules. FKBP has a well-defined site that binds the pipecolinic core. However, the benzanilide site is near the surface and not well defined. The increased affinity of the composite ligand could be the result of the increase in the size of the inhibitor and not the selectivity of affinity of the benzanilide for its surface site. Increasing the size of the pipecolinic inhibitor would potentially increase the number of van der Waals interactions with the protein as well as the hydrophobicity of the inhibitor, which would make it more favorable to be bound to the protein rather than solvated in water.

Figure 5. Structure ofthe lowest energy configuration of (a) the acetohydroxamic acid docked to stromelysin; (b) compound S8 docked to stromelysin in the presence of acetohydroxamic acid and (c) compound S50 docked to stromelysin.
Table 4. Predicted stromelysin binding affinities

Compound

AG predicted (kcal/mol)

AG exp2 (kcal/mol)*

S1

-1.2

-2.4

S2**

-6.0

-4.8 ± 0.3

S4**

-6.2

> -2.7

S6**

-5.7

-5.1 ± 0.1

S7**

-5.2

-3.8 ± 0.1

S8**

-6.5

-6.4 ± 0.3

S11**

-6.0

-4.5 ± 0.4

S26**

-7.2

-6.4 ± 0.3

S49

-8.7

-9.0

S50

-9.1

-10.4

S51

-9.5

-7.5

S52

-10.0

-7.4

S53

-9.2

-10.7

* All experimental AG values were calculated from reported Kd or IC50 values for the complexes.

** Indicates docked in the presence of compound S1, acetohydrox-amic acid.

* All experimental AG values were calculated from reported Kd or IC50 values for the complexes.

** Indicates docked in the presence of compound S1, acetohydrox-amic acid.

0 0

Post a comment