SUPPLEMENTARY MATERIAL

Step 1. Data collection and preparation

Table 1. Summary statistics of protein kinases features in the 300 X-ray structure extracted from the Protein Data Bank considered in this study.

Kinases features of the PDB structure / Number / Percentage
DFG-in / 274 / 91.3%
DFG-out / 17 / 5.7%
Non-canonical DFG (e.g. unconventional phe position; not resolved residues of X-ray structure) / 9 / 3.0%
Phosphorilated / 76 / 25.3%
Covalently inactivated by the ligand / 4 / 1.3%
Missing Activation loop / 87 / 29.0%
Missing Glycine-rich loop / 15 / 5.0%
Missing loop between C-helix and 3 (see Fig. 1 of the main text) / 95 / 31.7%

Step 2. Scaffold selection

Computational details on BEAR procedure

Input file preparation is achieved for each complex by merging the ligand coordinates prepared in step 1 with those of the receptor template. The topology and coordinate files are created for each complex and are subsequently used for the structural refinement that is divided in three subtasks:

i)Structure minimization

ii)MD simulation

iii)Structure re-minimization.

The combination of these subtasks has already proved to be a reliable tool for the refinement of ligand conformations. The last step of BEAR procedure involves the evaluation of the free energies of binding with MM-PBSA and MM-GBSA by using the single complex structure obtained after the structural refinement. Other applications using MM-PBSA and MM-GBSA in combination with a single ligand-receptor structure were successfully obtainedfor estimating binding free energies if compared, for instance, to molecular docking scoring functions [1-7].

Details of different ranking procedures and consensus criteria

Table 2. Details of the three BEAR ranking procedures used.

Temperature / Constraints
Rank 1a / Fast heating to 300K / Only residues within 5Å from the ligand allowed to move
Rank 2a / gradual heating from 0 to 300K in 20ps / Only residues within 5Å from the ligand allowed to move
Rank 3a / gradual heating from 0 to 300K in 20ps / Only residues within 8Å from the ligand allowed to move

a) Settings common to all ranks: i) Minimization and re-minimization performed with 4000 steps of conjugate gradient energy minimization performed with distance-dependent dielectric constantand a 12Å cut-off for nonbonded interactions. ii) A length of 200ps for the MD step iii) Binding free energy evaluations performed with grid spacing of 0.5Å, internal and external dielectric constant of 1 and 80 respectively, solvent probe radius of 1.4 Å and SASA calculation performed with molsurf with =0.0072 kcal mol-1 Å-2 and b=0 kcal mol-1.

In silico screening results

Table 3. Ranked position of the first hits compounds (ID codes) obtained after each ranking procedure and consensus rank.

Rank 1 / Rank 2 / Rank 3 / Consensus rank
db+1_0076 / db+1_0076 / db+1_0076 / db+1_0076
db0_0140a / db+1_0059 / db+1_0024 / db+1_0071
db+1_0023 / db+1_0020 / db+1_0118 / db0_0152
db0_0060 / db+1_0102 / db+1_0102 / db+1_0023
db0_0065 / db0_0103 / db+1_0018 / db0_0130
db0_0036 / db+1_0071 / db+1_0068 / db0_0068
db0_0028 / db+1_0089 / db0_0024 / db0_0134
db+1_0014 / db0_0152 / db+1_0003 / db0_0153
db+1_0099 / db0_0140a / db+1_0061 / db0_0135
db+1_0118 / db0_0028 / db+1_0038 / db0_0076
db0_0064 / db0_0130 / db+1_0070 / db+1_0077
db+1_0102 / db0_0036 / db+1_0037 / db+1_0118
db+1_0055 / db+1_0023 / db+1_0013 / db+1_0074
db0_0134 / db+1_0055 / db+1_0099 / db0_0117
db+1_0001 / db+1_0074 / db+1_0054 / db0_0091
db+1_0071 / db+1_0092 / db+1_0052 / db+1_0089
db0_0130 / db+1_0090 / db+1_0027 / db0_0103
db+1_0043 / db0_0037 / db+1_0007 / db+1_0005
db+1_0020 / db+1_0000 / db0_0140a / db0_0034
db0_0037 / db+1_0017 / db0_0021 / db0_0077
db+1_0038 / db+1_0052 / db0_0037 / db+1_0001
db+1_0009 / db0_0134 / db0_0068 / db0_0021
db0_0033 / db0_0068 / db+1_0089 / db0_0102
db+1_0054 / db0_0023 / db+1_0001 / db0_0028
db+1_0016 / db+1_0024 / db0_0033 / db0_0023
db+1_0011 / db0_0091 / db+1_0079 / db+1_0069
db0_0030 / db+1_0018 / db+1_0092 / db0_0089
db+1_0018 / db0_0064 / db+1_0077 / db0_0025
db0_0117 / db+1_0093 / db+1_0008 / db+1_0029
db0_0068 / db0_0135 / db+1_0067 / db+1_0099

a) db0_0140 is an example of type-II inhibitor that binds to a protein kinase with DFG-out conformation. The inhibitor was ranked in the top position with the fast heating MD scheme (Rank 1, see Table 2 supplementary material) while it was penalized in terms of MM-PBSA score when slow heating ramp was used (Ranks 2 and 3, Table 2 supplementary material).

REFERENCES

[1] A.M. Ferrari, G. Degliesposti, M. Sgobba and G. Rastelli, Bioorg. Med. Chem., 15 (2007) 7865.

[2] G. Rastelli, G. Degliesposti, A. Del Rio and M. Sgobba, Chem. Biol. Drug Des., 73 (2009) 283.

[3] G. Rastelli, A. Del Rio, G. Degliesposti and M. Sgobba, J. Comput. Chem. (2009).

[4] G. Rastelli, R. Rosenfeld, R. Reid and D.V. Santi, J. Struct. Biol., 164 (2008) 18.

[5] M.K. Gilson and H.X. Zhou, Annu Rev Bioph Biom, 36 (2007) 21.

[6] I. Stoica, S.K. Sadiq and P.V. Coveney, J. Am. Chem. Soc., 130 (2008) 2639.

[7] B. Kuhn, P. Gerber, T. Schulz-Gasch and M. Stahl, J. Med. Chem., 48 (2005) 4040.